{"schema":"kbp-graph/v1","blueprint_id":"preview:5fb02e4b02b2dcdd2ca03830f16857c6b71cb83004661cca27a394ebba255208","lineage":"sha256:299ae55c78187cb32a172c02cab9e123e36af0cb9bbcb33503178621e78b9939","revision":0,"exported_by":"preview: tools/preview/kbp_preview.py over `met circuit plan` (not `met blueprint export`)","arm":{"mode":"decode","rows":1,"digest":"152bf34432db217a1dc2d69c3483e5625966448fd688c1cee8d51e91fbf4f278"},"layers":[{"index":0,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l0.gdn","l0.dense_ffn"]},{"index":1,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l1.gdn","l1.dense_ffn"]},{"index":2,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l2.gdn","l2.dense_ffn"]},{"index":3,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l3.attn","l3.dense_ffn"]},{"index":4,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l4.gdn","l4.dense_ffn"]},{"index":5,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l5.gdn","l5.dense_ffn"]},{"index":6,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l6.gdn","l6.dense_ffn"]},{"index":7,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l7.attn","l7.dense_ffn"]},{"index":8,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l8.gdn","l8.dense_ffn"]},{"index":9,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l9.gdn","l9.dense_ffn"]},{"index":10,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l10.gdn","l10.dense_ffn"]},{"index":11,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l11.attn","l11.dense_ffn"]},{"index":12,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l12.gdn","l12.dense_ffn"]},{"index":13,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l13.gdn","l13.dense_ffn"]},{"index":14,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l14.gdn","l14.dense_ffn"]},{"index":15,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l15.attn","l15.dense_ffn"]},{"index":16,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l16.gdn","l16.dense_ffn"]},{"index":17,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l17.gdn","l17.dense_ffn"]},{"index":18,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l18.gdn","l18.dense_ffn"]},{"index":19,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l19.attn","l19.dense_ffn"]},{"index":20,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l20.gdn","l20.dense_ffn"]},{"index":21,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l21.gdn","l21.dense_ffn"]},{"index":22,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l22.gdn","l22.dense_ffn"]},{"index":23,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l23.attn","l23.dense_ffn"]},{"index":24,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l24.gdn","l24.dense_ffn"]},{"index":25,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l25.gdn","l25.dense_ffn"]},{"index":26,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l26.gdn","l26.dense_ffn"]},{"index":27,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l27.attn","l27.dense_ffn"]},{"index":28,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l28.gdn","l28.dense_ffn"]},{"index":29,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l29.gdn","l29.dense_ffn"]},{"index":30,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l30.gdn","l30.dense_ffn"]},{"index":31,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l31.attn","l31.dense_ffn"]},{"index":32,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l32.gdn","l32.dense_ffn"]},{"index":33,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l33.gdn","l33.dense_ffn"]},{"index":34,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l34.gdn","l34.dense_ffn"]},{"index":35,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l35.attn","l35.dense_ffn"]},{"index":36,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l36.gdn","l36.dense_ffn"]},{"index":37,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l37.gdn","l37.dense_ffn"]},{"index":38,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l38.gdn","l38.dense_ffn"]},{"index":39,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l39.attn","l39.dense_ffn"]},{"index":40,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l40.gdn","l40.dense_ffn"]},{"index":41,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l41.gdn","l41.dense_ffn"]},{"index":42,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l42.gdn","l42.dense_ffn"]},{"index":43,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l43.attn","l43.dense_ffn"]},{"index":44,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l44.gdn","l44.dense_ffn"]},{"index":45,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l45.gdn","l45.dense_ffn"]},{"index":46,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l46.gdn","l46.dense_ffn"]},{"index":47,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l47.attn","l47.dense_ffn"]},{"index":48,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l48.gdn","l48.dense_ffn"]},{"index":49,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l49.gdn","l49.dense_ffn"]},{"index":50,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l50.gdn","l50.dense_ffn"]},{"index":51,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l51.attn","l51.dense_ffn"]},{"index":52,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l52.gdn","l52.dense_ffn"]},{"index":53,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l53.gdn","l53.dense_ffn"]},{"index":54,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l54.gdn","l54.dense_ffn"]},{"index":55,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l55.attn","l55.dense_ffn"]},{"index":56,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l56.gdn","l56.dense_ffn"]},{"index":57,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l57.gdn","l57.dense_ffn"]},{"index":58,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l58.gdn","l58.dense_ffn"]},{"index":59,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l59.attn","l59.dense_ffn"]},{"index":60,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l60.gdn","l60.dense_ffn"]},{"index":61,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l61.gdn","l61.dense_ffn"]},{"index":62,"kind":"linear_attention","template":"gdn+dense_ffn","blocks":["l62.gdn","l62.dense_ffn"]},{"index":63,"kind":"full_attention","template":"attn+dense_ffn","blocks":["l63.attn","l63.dense_ffn"]}],"nodes":[{"id":"embed.embed","order":0,"layer":null,"block":"embed","block_type":"embed","kind":null,"op":"embed","roles":[],"inserted":false,"precision":{"weight":null,"activation":null,"output":"bf16"},"steps":["gather bf16"],"group":0,"kernel":null,"family":null,"point":null,"compute":null,"numerics":"reference","where":"host","state":null},{"id":"l0.gdn.input_norm","order":1,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":1,"kernel":"norm::rms_norm_residual","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l0.gdn.qkvz","order":2,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":2,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l0.gdn.ba","order":3,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":3,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l0.gdn.gates","order":4,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":3,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l0.gdn.conv","order":5,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":4,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l0.gdn.conv_ckpt","order":6,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":4,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l0.gdn.l2","order":7,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":4,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l0.gdn.recur","order":8,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":5,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l0.gdn.out_norm","order":9,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":6,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l0.gdn.out","order":10,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":7,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l0.gdn.add","order":11,"layer":0,"block":"l0.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":8,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l0.dense_ffn.post_norm","order":12,"layer":0,"block":"l0.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":8,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l0.dense_ffn.gate_up","order":13,"layer":0,"block":"l0.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":9,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l0.dense_ffn.act","order":14,"layer":0,"block":"l0.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":10,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l0.dense_ffn.down","order":15,"layer":0,"block":"l0.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":11,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l0.dense_ffn.add","order":16,"layer":0,"block":"l0.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":12,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l1.gdn.input_norm","order":17,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":12,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l1.gdn.qkvz","order":18,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":13,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l1.gdn.ba","order":19,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":14,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l1.gdn.gates","order":20,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":14,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l1.gdn.conv","order":21,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":15,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l1.gdn.conv_ckpt","order":22,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":15,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l1.gdn.l2","order":23,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":15,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l1.gdn.recur","order":24,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":16,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l1.gdn.out_norm","order":25,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":17,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l1.gdn.out","order":26,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":18,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l1.gdn.add","order":27,"layer":1,"block":"l1.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":19,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l1.dense_ffn.post_norm","order":28,"layer":1,"block":"l1.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":19,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l1.dense_ffn.gate_up","order":29,"layer":1,"block":"l1.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":20,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l1.dense_ffn.act","order":30,"layer":1,"block":"l1.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":21,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l1.dense_ffn.down","order":31,"layer":1,"block":"l1.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":22,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l1.dense_ffn.add","order":32,"layer":1,"block":"l1.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":23,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l2.gdn.input_norm","order":33,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":23,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l2.gdn.qkvz","order":34,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":24,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l2.gdn.ba","order":35,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":25,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l2.gdn.gates","order":36,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":25,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l2.gdn.conv","order":37,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":26,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l2.gdn.conv_ckpt","order":38,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":26,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l2.gdn.l2","order":39,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":26,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l2.gdn.recur","order":40,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":27,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l2.gdn.out_norm","order":41,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":28,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l2.gdn.out","order":42,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":29,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l2.gdn.add","order":43,"layer":2,"block":"l2.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":30,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l2.dense_ffn.post_norm","order":44,"layer":2,"block":"l2.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":30,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l2.dense_ffn.gate_up","order":45,"layer":2,"block":"l2.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":31,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l2.dense_ffn.act","order":46,"layer":2,"block":"l2.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":32,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l2.dense_ffn.down","order":47,"layer":2,"block":"l2.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":33,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l2.dense_ffn.add","order":48,"layer":2,"block":"l2.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":34,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l3.attn.input_norm","order":49,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":34,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l3.attn.q","order":50,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":35,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l3.attn.q_split","order":51,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":35,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l3.attn.k","order":52,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":36,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l3.attn.v","order":53,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":36,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l3.attn.q_norm","order":54,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":37,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l3.attn.k_norm","order":55,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":38,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l3.attn.rope","order":56,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":39,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l3.attn.kv_write","order":57,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":40,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l3.attn.attend","order":58,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":41,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l3.attn.gate_mul","order":59,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":42,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l3.attn.o","order":60,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":43,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l3.attn.add","order":61,"layer":3,"block":"l3.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":44,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l3.dense_ffn.post_norm","order":62,"layer":3,"block":"l3.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":44,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l3.dense_ffn.gate_up","order":63,"layer":3,"block":"l3.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":45,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l3.dense_ffn.act","order":64,"layer":3,"block":"l3.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":46,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l3.dense_ffn.down","order":65,"layer":3,"block":"l3.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":47,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l3.dense_ffn.add","order":66,"layer":3,"block":"l3.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":48,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l4.gdn.input_norm","order":67,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":48,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l4.gdn.qkvz","order":68,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":49,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l4.gdn.ba","order":69,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":50,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l4.gdn.gates","order":70,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":50,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l4.gdn.conv","order":71,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":51,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l4.gdn.conv_ckpt","order":72,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":51,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l4.gdn.l2","order":73,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":51,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l4.gdn.recur","order":74,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":52,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l4.gdn.out_norm","order":75,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":53,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l4.gdn.out","order":76,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":54,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l4.gdn.add","order":77,"layer":4,"block":"l4.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":55,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l4.dense_ffn.post_norm","order":78,"layer":4,"block":"l4.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":55,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l4.dense_ffn.gate_up","order":79,"layer":4,"block":"l4.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":56,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l4.dense_ffn.act","order":80,"layer":4,"block":"l4.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":57,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l4.dense_ffn.down","order":81,"layer":4,"block":"l4.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":58,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l4.dense_ffn.add","order":82,"layer":4,"block":"l4.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":59,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l5.gdn.input_norm","order":83,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":59,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l5.gdn.qkvz","order":84,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":60,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l5.gdn.ba","order":85,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":61,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l5.gdn.gates","order":86,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":61,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l5.gdn.conv","order":87,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":62,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l5.gdn.conv_ckpt","order":88,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":62,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l5.gdn.l2","order":89,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":62,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l5.gdn.recur","order":90,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":63,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l5.gdn.out_norm","order":91,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":64,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l5.gdn.out","order":92,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":65,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l5.gdn.add","order":93,"layer":5,"block":"l5.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":66,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l5.dense_ffn.post_norm","order":94,"layer":5,"block":"l5.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":66,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l5.dense_ffn.gate_up","order":95,"layer":5,"block":"l5.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":67,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l5.dense_ffn.act","order":96,"layer":5,"block":"l5.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":68,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l5.dense_ffn.down","order":97,"layer":5,"block":"l5.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":69,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l5.dense_ffn.add","order":98,"layer":5,"block":"l5.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":70,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l6.gdn.input_norm","order":99,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":70,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l6.gdn.qkvz","order":100,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":71,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l6.gdn.ba","order":101,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":72,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l6.gdn.gates","order":102,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":72,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l6.gdn.conv","order":103,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":73,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l6.gdn.conv_ckpt","order":104,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":73,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l6.gdn.l2","order":105,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":73,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l6.gdn.recur","order":106,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":74,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l6.gdn.out_norm","order":107,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":75,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l6.gdn.out","order":108,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":76,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l6.gdn.add","order":109,"layer":6,"block":"l6.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":77,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l6.dense_ffn.post_norm","order":110,"layer":6,"block":"l6.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":77,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l6.dense_ffn.gate_up","order":111,"layer":6,"block":"l6.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":78,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l6.dense_ffn.act","order":112,"layer":6,"block":"l6.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":79,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l6.dense_ffn.down","order":113,"layer":6,"block":"l6.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":80,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l6.dense_ffn.add","order":114,"layer":6,"block":"l6.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":81,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l7.attn.input_norm","order":115,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":81,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l7.attn.q","order":116,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":82,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l7.attn.q_split","order":117,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":82,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l7.attn.k","order":118,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":83,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l7.attn.v","order":119,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":83,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l7.attn.q_norm","order":120,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":84,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l7.attn.k_norm","order":121,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":85,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l7.attn.rope","order":122,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":86,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l7.attn.kv_write","order":123,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":87,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l7.attn.attend","order":124,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":88,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l7.attn.gate_mul","order":125,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":89,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l7.attn.o","order":126,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":90,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l7.attn.add","order":127,"layer":7,"block":"l7.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":91,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l7.dense_ffn.post_norm","order":128,"layer":7,"block":"l7.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":91,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l7.dense_ffn.gate_up","order":129,"layer":7,"block":"l7.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":92,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l7.dense_ffn.act","order":130,"layer":7,"block":"l7.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":93,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l7.dense_ffn.down","order":131,"layer":7,"block":"l7.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":94,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l7.dense_ffn.add","order":132,"layer":7,"block":"l7.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":95,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l8.gdn.input_norm","order":133,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":95,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l8.gdn.qkvz","order":134,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":96,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l8.gdn.ba","order":135,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":97,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l8.gdn.gates","order":136,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":97,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l8.gdn.conv","order":137,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":98,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l8.gdn.conv_ckpt","order":138,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":98,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l8.gdn.l2","order":139,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":98,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l8.gdn.recur","order":140,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":99,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l8.gdn.out_norm","order":141,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":100,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l8.gdn.out","order":142,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":101,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l8.gdn.add","order":143,"layer":8,"block":"l8.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":102,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l8.dense_ffn.post_norm","order":144,"layer":8,"block":"l8.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":102,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l8.dense_ffn.gate_up","order":145,"layer":8,"block":"l8.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":103,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l8.dense_ffn.act","order":146,"layer":8,"block":"l8.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":104,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l8.dense_ffn.down","order":147,"layer":8,"block":"l8.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":105,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l8.dense_ffn.add","order":148,"layer":8,"block":"l8.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":106,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l9.gdn.input_norm","order":149,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":106,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l9.gdn.qkvz","order":150,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":107,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l9.gdn.ba","order":151,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":108,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l9.gdn.gates","order":152,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":108,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l9.gdn.conv","order":153,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":109,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l9.gdn.conv_ckpt","order":154,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":109,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l9.gdn.l2","order":155,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":109,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l9.gdn.recur","order":156,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":110,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l9.gdn.out_norm","order":157,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":111,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l9.gdn.out","order":158,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":112,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l9.gdn.add","order":159,"layer":9,"block":"l9.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":113,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l9.dense_ffn.post_norm","order":160,"layer":9,"block":"l9.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":113,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l9.dense_ffn.gate_up","order":161,"layer":9,"block":"l9.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":114,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l9.dense_ffn.act","order":162,"layer":9,"block":"l9.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":115,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l9.dense_ffn.down","order":163,"layer":9,"block":"l9.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":116,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l9.dense_ffn.add","order":164,"layer":9,"block":"l9.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":117,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l10.gdn.input_norm","order":165,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":117,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l10.gdn.qkvz","order":166,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":118,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l10.gdn.ba","order":167,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":119,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l10.gdn.gates","order":168,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":119,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l10.gdn.conv","order":169,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":120,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l10.gdn.conv_ckpt","order":170,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":120,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l10.gdn.l2","order":171,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":120,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l10.gdn.recur","order":172,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":121,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l10.gdn.out_norm","order":173,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":122,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l10.gdn.out","order":174,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":123,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l10.gdn.add","order":175,"layer":10,"block":"l10.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":124,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l10.dense_ffn.post_norm","order":176,"layer":10,"block":"l10.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":124,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l10.dense_ffn.gate_up","order":177,"layer":10,"block":"l10.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":125,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l10.dense_ffn.act","order":178,"layer":10,"block":"l10.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":126,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l10.dense_ffn.down","order":179,"layer":10,"block":"l10.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":127,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l10.dense_ffn.add","order":180,"layer":10,"block":"l10.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":128,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l11.attn.input_norm","order":181,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":128,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l11.attn.q","order":182,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":129,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l11.attn.q_split","order":183,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":129,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l11.attn.k","order":184,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":130,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l11.attn.v","order":185,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":130,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l11.attn.q_norm","order":186,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":131,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l11.attn.k_norm","order":187,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":132,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l11.attn.rope","order":188,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":133,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l11.attn.kv_write","order":189,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":134,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l11.attn.attend","order":190,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":135,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l11.attn.gate_mul","order":191,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":136,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l11.attn.o","order":192,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":137,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l11.attn.add","order":193,"layer":11,"block":"l11.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":138,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l11.dense_ffn.post_norm","order":194,"layer":11,"block":"l11.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":138,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l11.dense_ffn.gate_up","order":195,"layer":11,"block":"l11.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":139,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l11.dense_ffn.act","order":196,"layer":11,"block":"l11.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":140,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l11.dense_ffn.down","order":197,"layer":11,"block":"l11.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":141,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l11.dense_ffn.add","order":198,"layer":11,"block":"l11.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":142,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l12.gdn.input_norm","order":199,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":142,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l12.gdn.qkvz","order":200,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":143,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l12.gdn.ba","order":201,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":144,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l12.gdn.gates","order":202,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":144,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l12.gdn.conv","order":203,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":145,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l12.gdn.conv_ckpt","order":204,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":145,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l12.gdn.l2","order":205,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":145,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l12.gdn.recur","order":206,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":146,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l12.gdn.out_norm","order":207,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":147,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l12.gdn.out","order":208,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":148,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l12.gdn.add","order":209,"layer":12,"block":"l12.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":149,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l12.dense_ffn.post_norm","order":210,"layer":12,"block":"l12.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":149,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l12.dense_ffn.gate_up","order":211,"layer":12,"block":"l12.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":150,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l12.dense_ffn.act","order":212,"layer":12,"block":"l12.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":151,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l12.dense_ffn.down","order":213,"layer":12,"block":"l12.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":152,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l12.dense_ffn.add","order":214,"layer":12,"block":"l12.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":153,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l13.gdn.input_norm","order":215,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":153,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l13.gdn.qkvz","order":216,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":154,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l13.gdn.ba","order":217,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":155,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l13.gdn.gates","order":218,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":155,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l13.gdn.conv","order":219,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":156,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l13.gdn.conv_ckpt","order":220,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":156,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l13.gdn.l2","order":221,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":156,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l13.gdn.recur","order":222,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":157,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l13.gdn.out_norm","order":223,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":158,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l13.gdn.out","order":224,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":159,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l13.gdn.add","order":225,"layer":13,"block":"l13.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":160,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l13.dense_ffn.post_norm","order":226,"layer":13,"block":"l13.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":160,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l13.dense_ffn.gate_up","order":227,"layer":13,"block":"l13.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":161,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l13.dense_ffn.act","order":228,"layer":13,"block":"l13.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":162,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l13.dense_ffn.down","order":229,"layer":13,"block":"l13.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":163,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l13.dense_ffn.add","order":230,"layer":13,"block":"l13.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":164,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l14.gdn.input_norm","order":231,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":164,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l14.gdn.qkvz","order":232,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":165,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l14.gdn.ba","order":233,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":166,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l14.gdn.gates","order":234,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":166,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l14.gdn.conv","order":235,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":167,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l14.gdn.conv_ckpt","order":236,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":167,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l14.gdn.l2","order":237,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":167,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l14.gdn.recur","order":238,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":168,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l14.gdn.out_norm","order":239,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":169,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l14.gdn.out","order":240,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":170,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l14.gdn.add","order":241,"layer":14,"block":"l14.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":171,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l14.dense_ffn.post_norm","order":242,"layer":14,"block":"l14.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":171,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l14.dense_ffn.gate_up","order":243,"layer":14,"block":"l14.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":172,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l14.dense_ffn.act","order":244,"layer":14,"block":"l14.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":173,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l14.dense_ffn.down","order":245,"layer":14,"block":"l14.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":174,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l14.dense_ffn.add","order":246,"layer":14,"block":"l14.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":175,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l15.attn.input_norm","order":247,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":175,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l15.attn.q","order":248,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":176,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l15.attn.q_split","order":249,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":176,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l15.attn.k","order":250,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":177,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l15.attn.v","order":251,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":177,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l15.attn.q_norm","order":252,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":178,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l15.attn.k_norm","order":253,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":179,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l15.attn.rope","order":254,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":180,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l15.attn.kv_write","order":255,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":181,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l15.attn.attend","order":256,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":182,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l15.attn.gate_mul","order":257,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":183,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l15.attn.o","order":258,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":184,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l15.attn.add","order":259,"layer":15,"block":"l15.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":185,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l15.dense_ffn.post_norm","order":260,"layer":15,"block":"l15.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":185,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l15.dense_ffn.gate_up","order":261,"layer":15,"block":"l15.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":186,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l15.dense_ffn.act","order":262,"layer":15,"block":"l15.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":187,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l15.dense_ffn.down","order":263,"layer":15,"block":"l15.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":188,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l15.dense_ffn.add","order":264,"layer":15,"block":"l15.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":189,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l16.gdn.input_norm","order":265,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":189,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l16.gdn.qkvz","order":266,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":190,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l16.gdn.ba","order":267,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":191,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l16.gdn.gates","order":268,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":191,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l16.gdn.conv","order":269,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":192,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l16.gdn.conv_ckpt","order":270,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":192,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l16.gdn.l2","order":271,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":192,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l16.gdn.recur","order":272,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":193,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l16.gdn.out_norm","order":273,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":194,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l16.gdn.out","order":274,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":195,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l16.gdn.add","order":275,"layer":16,"block":"l16.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":196,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l16.dense_ffn.post_norm","order":276,"layer":16,"block":"l16.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":196,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l16.dense_ffn.gate_up","order":277,"layer":16,"block":"l16.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":197,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l16.dense_ffn.act","order":278,"layer":16,"block":"l16.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":198,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l16.dense_ffn.down","order":279,"layer":16,"block":"l16.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":199,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l16.dense_ffn.add","order":280,"layer":16,"block":"l16.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":200,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l17.gdn.input_norm","order":281,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":200,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l17.gdn.qkvz","order":282,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":201,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l17.gdn.ba","order":283,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":202,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l17.gdn.gates","order":284,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":202,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l17.gdn.conv","order":285,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":203,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l17.gdn.conv_ckpt","order":286,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":203,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l17.gdn.l2","order":287,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":203,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l17.gdn.recur","order":288,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":204,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l17.gdn.out_norm","order":289,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":205,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l17.gdn.out","order":290,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":206,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l17.gdn.add","order":291,"layer":17,"block":"l17.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":207,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l17.dense_ffn.post_norm","order":292,"layer":17,"block":"l17.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":207,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l17.dense_ffn.gate_up","order":293,"layer":17,"block":"l17.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":208,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l17.dense_ffn.act","order":294,"layer":17,"block":"l17.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":209,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l17.dense_ffn.down","order":295,"layer":17,"block":"l17.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":210,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l17.dense_ffn.add","order":296,"layer":17,"block":"l17.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":211,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l18.gdn.input_norm","order":297,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":211,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l18.gdn.qkvz","order":298,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":212,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l18.gdn.ba","order":299,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":213,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l18.gdn.gates","order":300,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":213,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l18.gdn.conv","order":301,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":214,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l18.gdn.conv_ckpt","order":302,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":214,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l18.gdn.l2","order":303,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":214,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l18.gdn.recur","order":304,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":215,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l18.gdn.out_norm","order":305,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":216,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l18.gdn.out","order":306,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":217,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l18.gdn.add","order":307,"layer":18,"block":"l18.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":218,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l18.dense_ffn.post_norm","order":308,"layer":18,"block":"l18.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":218,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l18.dense_ffn.gate_up","order":309,"layer":18,"block":"l18.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":219,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l18.dense_ffn.act","order":310,"layer":18,"block":"l18.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":220,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l18.dense_ffn.down","order":311,"layer":18,"block":"l18.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":221,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l18.dense_ffn.add","order":312,"layer":18,"block":"l18.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":222,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l19.attn.input_norm","order":313,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":222,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l19.attn.q","order":314,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":223,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l19.attn.q_split","order":315,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":223,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l19.attn.k","order":316,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":224,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l19.attn.v","order":317,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":224,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l19.attn.q_norm","order":318,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":225,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l19.attn.k_norm","order":319,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":226,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l19.attn.rope","order":320,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":227,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l19.attn.kv_write","order":321,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":228,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l19.attn.attend","order":322,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":229,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l19.attn.gate_mul","order":323,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":230,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l19.attn.o","order":324,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":231,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l19.attn.add","order":325,"layer":19,"block":"l19.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":232,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l19.dense_ffn.post_norm","order":326,"layer":19,"block":"l19.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":232,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l19.dense_ffn.gate_up","order":327,"layer":19,"block":"l19.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":233,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l19.dense_ffn.act","order":328,"layer":19,"block":"l19.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":234,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l19.dense_ffn.down","order":329,"layer":19,"block":"l19.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":235,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l19.dense_ffn.add","order":330,"layer":19,"block":"l19.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":236,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l20.gdn.input_norm","order":331,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":236,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l20.gdn.qkvz","order":332,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":237,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l20.gdn.ba","order":333,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":238,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l20.gdn.gates","order":334,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":238,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l20.gdn.conv","order":335,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":239,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l20.gdn.conv_ckpt","order":336,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":239,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l20.gdn.l2","order":337,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":239,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l20.gdn.recur","order":338,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":240,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l20.gdn.out_norm","order":339,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":241,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l20.gdn.out","order":340,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":242,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l20.gdn.add","order":341,"layer":20,"block":"l20.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":243,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l20.dense_ffn.post_norm","order":342,"layer":20,"block":"l20.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":243,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l20.dense_ffn.gate_up","order":343,"layer":20,"block":"l20.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":244,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l20.dense_ffn.act","order":344,"layer":20,"block":"l20.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":245,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l20.dense_ffn.down","order":345,"layer":20,"block":"l20.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":246,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l20.dense_ffn.add","order":346,"layer":20,"block":"l20.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":247,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l21.gdn.input_norm","order":347,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":247,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l21.gdn.qkvz","order":348,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":248,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l21.gdn.ba","order":349,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":249,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l21.gdn.gates","order":350,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":249,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l21.gdn.conv","order":351,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":250,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l21.gdn.conv_ckpt","order":352,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":250,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l21.gdn.l2","order":353,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":250,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l21.gdn.recur","order":354,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":251,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l21.gdn.out_norm","order":355,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":252,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l21.gdn.out","order":356,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":253,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l21.gdn.add","order":357,"layer":21,"block":"l21.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":254,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l21.dense_ffn.post_norm","order":358,"layer":21,"block":"l21.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":254,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l21.dense_ffn.gate_up","order":359,"layer":21,"block":"l21.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":255,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l21.dense_ffn.act","order":360,"layer":21,"block":"l21.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":256,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l21.dense_ffn.down","order":361,"layer":21,"block":"l21.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":257,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l21.dense_ffn.add","order":362,"layer":21,"block":"l21.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":258,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l22.gdn.input_norm","order":363,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":258,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l22.gdn.qkvz","order":364,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":259,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l22.gdn.ba","order":365,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":260,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l22.gdn.gates","order":366,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":260,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l22.gdn.conv","order":367,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":261,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l22.gdn.conv_ckpt","order":368,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":261,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l22.gdn.l2","order":369,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":261,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l22.gdn.recur","order":370,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":262,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l22.gdn.out_norm","order":371,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":263,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l22.gdn.out","order":372,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":264,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l22.gdn.add","order":373,"layer":22,"block":"l22.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":265,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l22.dense_ffn.post_norm","order":374,"layer":22,"block":"l22.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":265,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l22.dense_ffn.gate_up","order":375,"layer":22,"block":"l22.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":266,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l22.dense_ffn.act","order":376,"layer":22,"block":"l22.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":267,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l22.dense_ffn.down","order":377,"layer":22,"block":"l22.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":268,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l22.dense_ffn.add","order":378,"layer":22,"block":"l22.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":269,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l23.attn.input_norm","order":379,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":269,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l23.attn.q","order":380,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":270,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l23.attn.q_split","order":381,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":270,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l23.attn.k","order":382,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":271,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l23.attn.v","order":383,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":271,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l23.attn.q_norm","order":384,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":272,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l23.attn.k_norm","order":385,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":273,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l23.attn.rope","order":386,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":274,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l23.attn.kv_write","order":387,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":275,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l23.attn.attend","order":388,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":276,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l23.attn.gate_mul","order":389,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":277,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l23.attn.o","order":390,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":278,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l23.attn.add","order":391,"layer":23,"block":"l23.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":279,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l23.dense_ffn.post_norm","order":392,"layer":23,"block":"l23.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":279,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l23.dense_ffn.gate_up","order":393,"layer":23,"block":"l23.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":280,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l23.dense_ffn.act","order":394,"layer":23,"block":"l23.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":281,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l23.dense_ffn.down","order":395,"layer":23,"block":"l23.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":282,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l23.dense_ffn.add","order":396,"layer":23,"block":"l23.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":283,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l24.gdn.input_norm","order":397,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":283,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l24.gdn.qkvz","order":398,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":284,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l24.gdn.ba","order":399,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":285,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l24.gdn.gates","order":400,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":285,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l24.gdn.conv","order":401,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":286,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l24.gdn.conv_ckpt","order":402,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":286,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l24.gdn.l2","order":403,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":286,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l24.gdn.recur","order":404,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":287,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l24.gdn.out_norm","order":405,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":288,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l24.gdn.out","order":406,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":289,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l24.gdn.add","order":407,"layer":24,"block":"l24.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":290,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l24.dense_ffn.post_norm","order":408,"layer":24,"block":"l24.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":290,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l24.dense_ffn.gate_up","order":409,"layer":24,"block":"l24.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":291,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l24.dense_ffn.act","order":410,"layer":24,"block":"l24.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":292,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l24.dense_ffn.down","order":411,"layer":24,"block":"l24.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":293,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l24.dense_ffn.add","order":412,"layer":24,"block":"l24.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":294,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l25.gdn.input_norm","order":413,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":294,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l25.gdn.qkvz","order":414,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":295,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l25.gdn.ba","order":415,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":296,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l25.gdn.gates","order":416,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":296,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l25.gdn.conv","order":417,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":297,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l25.gdn.conv_ckpt","order":418,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":297,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l25.gdn.l2","order":419,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":297,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l25.gdn.recur","order":420,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":298,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l25.gdn.out_norm","order":421,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":299,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l25.gdn.out","order":422,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":300,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l25.gdn.add","order":423,"layer":25,"block":"l25.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":301,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l25.dense_ffn.post_norm","order":424,"layer":25,"block":"l25.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":301,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l25.dense_ffn.gate_up","order":425,"layer":25,"block":"l25.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":302,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l25.dense_ffn.act","order":426,"layer":25,"block":"l25.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":303,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l25.dense_ffn.down","order":427,"layer":25,"block":"l25.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":304,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l25.dense_ffn.add","order":428,"layer":25,"block":"l25.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":305,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l26.gdn.input_norm","order":429,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":305,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l26.gdn.qkvz","order":430,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":306,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l26.gdn.ba","order":431,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":307,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l26.gdn.gates","order":432,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":307,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l26.gdn.conv","order":433,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":308,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l26.gdn.conv_ckpt","order":434,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":308,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l26.gdn.l2","order":435,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":308,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l26.gdn.recur","order":436,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":309,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l26.gdn.out_norm","order":437,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":310,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l26.gdn.out","order":438,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":311,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l26.gdn.add","order":439,"layer":26,"block":"l26.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":312,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l26.dense_ffn.post_norm","order":440,"layer":26,"block":"l26.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":312,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l26.dense_ffn.gate_up","order":441,"layer":26,"block":"l26.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":313,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l26.dense_ffn.act","order":442,"layer":26,"block":"l26.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":314,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l26.dense_ffn.down","order":443,"layer":26,"block":"l26.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":315,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l26.dense_ffn.add","order":444,"layer":26,"block":"l26.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":316,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l27.attn.input_norm","order":445,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":316,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l27.attn.q","order":446,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":317,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l27.attn.q_split","order":447,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":317,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l27.attn.k","order":448,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":318,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l27.attn.v","order":449,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":318,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l27.attn.q_norm","order":450,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":319,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l27.attn.k_norm","order":451,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":320,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l27.attn.rope","order":452,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":321,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l27.attn.kv_write","order":453,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":322,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l27.attn.attend","order":454,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":323,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l27.attn.gate_mul","order":455,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":324,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l27.attn.o","order":456,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":325,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l27.attn.add","order":457,"layer":27,"block":"l27.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":326,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l27.dense_ffn.post_norm","order":458,"layer":27,"block":"l27.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":326,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l27.dense_ffn.gate_up","order":459,"layer":27,"block":"l27.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":327,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l27.dense_ffn.act","order":460,"layer":27,"block":"l27.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":328,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l27.dense_ffn.down","order":461,"layer":27,"block":"l27.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":329,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l27.dense_ffn.add","order":462,"layer":27,"block":"l27.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":330,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l28.gdn.input_norm","order":463,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":330,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l28.gdn.qkvz","order":464,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":331,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l28.gdn.ba","order":465,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":332,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l28.gdn.gates","order":466,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":332,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l28.gdn.conv","order":467,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":333,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l28.gdn.conv_ckpt","order":468,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":333,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l28.gdn.l2","order":469,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":333,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l28.gdn.recur","order":470,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":334,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l28.gdn.out_norm","order":471,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":335,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l28.gdn.out","order":472,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":336,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l28.gdn.add","order":473,"layer":28,"block":"l28.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":337,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l28.dense_ffn.post_norm","order":474,"layer":28,"block":"l28.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":337,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l28.dense_ffn.gate_up","order":475,"layer":28,"block":"l28.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":338,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l28.dense_ffn.act","order":476,"layer":28,"block":"l28.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":339,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l28.dense_ffn.down","order":477,"layer":28,"block":"l28.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":340,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l28.dense_ffn.add","order":478,"layer":28,"block":"l28.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":341,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l29.gdn.input_norm","order":479,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":341,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l29.gdn.qkvz","order":480,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":342,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l29.gdn.ba","order":481,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":343,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l29.gdn.gates","order":482,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":343,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l29.gdn.conv","order":483,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":344,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l29.gdn.conv_ckpt","order":484,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":344,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l29.gdn.l2","order":485,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":344,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l29.gdn.recur","order":486,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":345,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l29.gdn.out_norm","order":487,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":346,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l29.gdn.out","order":488,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":347,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l29.gdn.add","order":489,"layer":29,"block":"l29.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":348,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l29.dense_ffn.post_norm","order":490,"layer":29,"block":"l29.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":348,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l29.dense_ffn.gate_up","order":491,"layer":29,"block":"l29.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":349,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l29.dense_ffn.act","order":492,"layer":29,"block":"l29.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":350,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l29.dense_ffn.down","order":493,"layer":29,"block":"l29.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":351,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l29.dense_ffn.add","order":494,"layer":29,"block":"l29.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":352,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l30.gdn.input_norm","order":495,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":352,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l30.gdn.qkvz","order":496,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":353,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l30.gdn.ba","order":497,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":354,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l30.gdn.gates","order":498,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":354,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l30.gdn.conv","order":499,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":355,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l30.gdn.conv_ckpt","order":500,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":355,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l30.gdn.l2","order":501,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":355,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l30.gdn.recur","order":502,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":356,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l30.gdn.out_norm","order":503,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":357,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l30.gdn.out","order":504,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":358,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l30.gdn.add","order":505,"layer":30,"block":"l30.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":359,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l30.dense_ffn.post_norm","order":506,"layer":30,"block":"l30.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":359,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l30.dense_ffn.gate_up","order":507,"layer":30,"block":"l30.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":360,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l30.dense_ffn.act","order":508,"layer":30,"block":"l30.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":361,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l30.dense_ffn.down","order":509,"layer":30,"block":"l30.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":362,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l30.dense_ffn.add","order":510,"layer":30,"block":"l30.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":363,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l31.attn.input_norm","order":511,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":363,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l31.attn.q","order":512,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":364,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l31.attn.q_split","order":513,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":364,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l31.attn.k","order":514,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":365,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l31.attn.v","order":515,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":365,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l31.attn.q_norm","order":516,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":366,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l31.attn.k_norm","order":517,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":367,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l31.attn.rope","order":518,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":368,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l31.attn.kv_write","order":519,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":369,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l31.attn.attend","order":520,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":370,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l31.attn.gate_mul","order":521,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":371,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l31.attn.o","order":522,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":372,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l31.attn.add","order":523,"layer":31,"block":"l31.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":373,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l31.dense_ffn.post_norm","order":524,"layer":31,"block":"l31.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":373,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l31.dense_ffn.gate_up","order":525,"layer":31,"block":"l31.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":374,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l31.dense_ffn.act","order":526,"layer":31,"block":"l31.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":375,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l31.dense_ffn.down","order":527,"layer":31,"block":"l31.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":376,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l31.dense_ffn.add","order":528,"layer":31,"block":"l31.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":377,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l32.gdn.input_norm","order":529,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":377,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l32.gdn.qkvz","order":530,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":378,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l32.gdn.ba","order":531,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":379,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l32.gdn.gates","order":532,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":379,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l32.gdn.conv","order":533,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":380,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l32.gdn.conv_ckpt","order":534,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":380,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l32.gdn.l2","order":535,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":380,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l32.gdn.recur","order":536,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":381,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l32.gdn.out_norm","order":537,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":382,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l32.gdn.out","order":538,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":383,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l32.gdn.add","order":539,"layer":32,"block":"l32.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":384,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l32.dense_ffn.post_norm","order":540,"layer":32,"block":"l32.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":384,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l32.dense_ffn.gate_up","order":541,"layer":32,"block":"l32.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":385,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l32.dense_ffn.act","order":542,"layer":32,"block":"l32.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":386,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l32.dense_ffn.down","order":543,"layer":32,"block":"l32.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":387,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l32.dense_ffn.add","order":544,"layer":32,"block":"l32.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":388,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l33.gdn.input_norm","order":545,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":388,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l33.gdn.qkvz","order":546,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":389,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l33.gdn.ba","order":547,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":390,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l33.gdn.gates","order":548,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":390,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l33.gdn.conv","order":549,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":391,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l33.gdn.conv_ckpt","order":550,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":391,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l33.gdn.l2","order":551,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":391,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l33.gdn.recur","order":552,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":392,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l33.gdn.out_norm","order":553,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":393,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l33.gdn.out","order":554,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":394,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l33.gdn.add","order":555,"layer":33,"block":"l33.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":395,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l33.dense_ffn.post_norm","order":556,"layer":33,"block":"l33.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":395,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l33.dense_ffn.gate_up","order":557,"layer":33,"block":"l33.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":396,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l33.dense_ffn.act","order":558,"layer":33,"block":"l33.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":397,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l33.dense_ffn.down","order":559,"layer":33,"block":"l33.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":398,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l33.dense_ffn.add","order":560,"layer":33,"block":"l33.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":399,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l34.gdn.input_norm","order":561,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":399,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l34.gdn.qkvz","order":562,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":400,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l34.gdn.ba","order":563,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":401,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l34.gdn.gates","order":564,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":401,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l34.gdn.conv","order":565,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":402,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l34.gdn.conv_ckpt","order":566,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":402,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l34.gdn.l2","order":567,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":402,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l34.gdn.recur","order":568,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":403,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l34.gdn.out_norm","order":569,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":404,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l34.gdn.out","order":570,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":405,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l34.gdn.add","order":571,"layer":34,"block":"l34.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":406,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l34.dense_ffn.post_norm","order":572,"layer":34,"block":"l34.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":406,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l34.dense_ffn.gate_up","order":573,"layer":34,"block":"l34.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":407,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l34.dense_ffn.act","order":574,"layer":34,"block":"l34.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":408,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l34.dense_ffn.down","order":575,"layer":34,"block":"l34.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":409,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l34.dense_ffn.add","order":576,"layer":34,"block":"l34.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":410,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l35.attn.input_norm","order":577,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":410,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l35.attn.q","order":578,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":411,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l35.attn.q_split","order":579,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":411,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l35.attn.k","order":580,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":412,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l35.attn.v","order":581,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":412,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l35.attn.q_norm","order":582,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":413,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l35.attn.k_norm","order":583,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":414,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l35.attn.rope","order":584,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":415,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l35.attn.kv_write","order":585,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":416,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l35.attn.attend","order":586,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":417,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l35.attn.gate_mul","order":587,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":418,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l35.attn.o","order":588,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":419,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l35.attn.add","order":589,"layer":35,"block":"l35.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":420,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l35.dense_ffn.post_norm","order":590,"layer":35,"block":"l35.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":420,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l35.dense_ffn.gate_up","order":591,"layer":35,"block":"l35.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":421,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l35.dense_ffn.act","order":592,"layer":35,"block":"l35.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":422,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l35.dense_ffn.down","order":593,"layer":35,"block":"l35.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":423,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l35.dense_ffn.add","order":594,"layer":35,"block":"l35.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":424,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l36.gdn.input_norm","order":595,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":424,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l36.gdn.qkvz","order":596,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":425,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l36.gdn.ba","order":597,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":426,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l36.gdn.gates","order":598,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":426,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l36.gdn.conv","order":599,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":427,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l36.gdn.conv_ckpt","order":600,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":427,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l36.gdn.l2","order":601,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":427,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l36.gdn.recur","order":602,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":428,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l36.gdn.out_norm","order":603,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":429,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l36.gdn.out","order":604,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":430,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l36.gdn.add","order":605,"layer":36,"block":"l36.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":431,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l36.dense_ffn.post_norm","order":606,"layer":36,"block":"l36.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":431,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l36.dense_ffn.gate_up","order":607,"layer":36,"block":"l36.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":432,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l36.dense_ffn.act","order":608,"layer":36,"block":"l36.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":433,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l36.dense_ffn.down","order":609,"layer":36,"block":"l36.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":434,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l36.dense_ffn.add","order":610,"layer":36,"block":"l36.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":435,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l37.gdn.input_norm","order":611,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":435,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l37.gdn.qkvz","order":612,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":436,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l37.gdn.ba","order":613,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":437,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l37.gdn.gates","order":614,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":437,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l37.gdn.conv","order":615,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":438,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l37.gdn.conv_ckpt","order":616,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":438,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l37.gdn.l2","order":617,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":438,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l37.gdn.recur","order":618,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":439,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l37.gdn.out_norm","order":619,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":440,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l37.gdn.out","order":620,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":441,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l37.gdn.add","order":621,"layer":37,"block":"l37.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":442,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l37.dense_ffn.post_norm","order":622,"layer":37,"block":"l37.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":442,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l37.dense_ffn.gate_up","order":623,"layer":37,"block":"l37.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":443,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l37.dense_ffn.act","order":624,"layer":37,"block":"l37.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":444,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l37.dense_ffn.down","order":625,"layer":37,"block":"l37.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":445,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l37.dense_ffn.add","order":626,"layer":37,"block":"l37.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":446,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l38.gdn.input_norm","order":627,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":446,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l38.gdn.qkvz","order":628,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":447,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l38.gdn.ba","order":629,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":448,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l38.gdn.gates","order":630,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":448,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l38.gdn.conv","order":631,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":449,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l38.gdn.conv_ckpt","order":632,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":449,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l38.gdn.l2","order":633,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":449,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l38.gdn.recur","order":634,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":450,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l38.gdn.out_norm","order":635,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":451,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l38.gdn.out","order":636,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":452,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l38.gdn.add","order":637,"layer":38,"block":"l38.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":453,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l38.dense_ffn.post_norm","order":638,"layer":38,"block":"l38.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":453,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l38.dense_ffn.gate_up","order":639,"layer":38,"block":"l38.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":454,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l38.dense_ffn.act","order":640,"layer":38,"block":"l38.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":455,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l38.dense_ffn.down","order":641,"layer":38,"block":"l38.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":456,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l38.dense_ffn.add","order":642,"layer":38,"block":"l38.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":457,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l39.attn.input_norm","order":643,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":457,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l39.attn.q","order":644,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":458,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l39.attn.q_split","order":645,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":458,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l39.attn.k","order":646,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":459,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l39.attn.v","order":647,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":459,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l39.attn.q_norm","order":648,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":460,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l39.attn.k_norm","order":649,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":461,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l39.attn.rope","order":650,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":462,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l39.attn.kv_write","order":651,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":463,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l39.attn.attend","order":652,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":464,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l39.attn.gate_mul","order":653,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":465,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l39.attn.o","order":654,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":466,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l39.attn.add","order":655,"layer":39,"block":"l39.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":467,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l39.dense_ffn.post_norm","order":656,"layer":39,"block":"l39.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":467,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l39.dense_ffn.gate_up","order":657,"layer":39,"block":"l39.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":468,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l39.dense_ffn.act","order":658,"layer":39,"block":"l39.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":469,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l39.dense_ffn.down","order":659,"layer":39,"block":"l39.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":470,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l39.dense_ffn.add","order":660,"layer":39,"block":"l39.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":471,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l40.gdn.input_norm","order":661,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":471,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l40.gdn.qkvz","order":662,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":472,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l40.gdn.ba","order":663,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":473,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l40.gdn.gates","order":664,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":473,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l40.gdn.conv","order":665,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":474,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l40.gdn.conv_ckpt","order":666,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":474,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l40.gdn.l2","order":667,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":474,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l40.gdn.recur","order":668,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":475,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l40.gdn.out_norm","order":669,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":476,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l40.gdn.out","order":670,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":477,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l40.gdn.add","order":671,"layer":40,"block":"l40.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":478,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l40.dense_ffn.post_norm","order":672,"layer":40,"block":"l40.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":478,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l40.dense_ffn.gate_up","order":673,"layer":40,"block":"l40.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":479,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l40.dense_ffn.act","order":674,"layer":40,"block":"l40.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":480,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l40.dense_ffn.down","order":675,"layer":40,"block":"l40.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":481,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l40.dense_ffn.add","order":676,"layer":40,"block":"l40.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":482,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l41.gdn.input_norm","order":677,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":482,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l41.gdn.qkvz","order":678,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":483,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l41.gdn.ba","order":679,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":484,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l41.gdn.gates","order":680,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":484,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l41.gdn.conv","order":681,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":485,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l41.gdn.conv_ckpt","order":682,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":485,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l41.gdn.l2","order":683,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":485,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l41.gdn.recur","order":684,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":486,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l41.gdn.out_norm","order":685,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":487,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l41.gdn.out","order":686,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":488,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l41.gdn.add","order":687,"layer":41,"block":"l41.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":489,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l41.dense_ffn.post_norm","order":688,"layer":41,"block":"l41.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":489,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l41.dense_ffn.gate_up","order":689,"layer":41,"block":"l41.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":490,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l41.dense_ffn.act","order":690,"layer":41,"block":"l41.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":491,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l41.dense_ffn.down","order":691,"layer":41,"block":"l41.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":492,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l41.dense_ffn.add","order":692,"layer":41,"block":"l41.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":493,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l42.gdn.input_norm","order":693,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":493,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l42.gdn.qkvz","order":694,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":494,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l42.gdn.ba","order":695,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":495,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l42.gdn.gates","order":696,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":495,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l42.gdn.conv","order":697,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":496,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l42.gdn.conv_ckpt","order":698,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":496,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l42.gdn.l2","order":699,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":496,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l42.gdn.recur","order":700,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":497,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l42.gdn.out_norm","order":701,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":498,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l42.gdn.out","order":702,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":499,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l42.gdn.add","order":703,"layer":42,"block":"l42.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":500,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l42.dense_ffn.post_norm","order":704,"layer":42,"block":"l42.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":500,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l42.dense_ffn.gate_up","order":705,"layer":42,"block":"l42.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":501,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l42.dense_ffn.act","order":706,"layer":42,"block":"l42.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":502,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l42.dense_ffn.down","order":707,"layer":42,"block":"l42.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":503,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l42.dense_ffn.add","order":708,"layer":42,"block":"l42.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":504,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l43.attn.input_norm","order":709,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":504,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l43.attn.q","order":710,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":505,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l43.attn.q_split","order":711,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":505,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l43.attn.k","order":712,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":506,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l43.attn.v","order":713,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":506,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l43.attn.q_norm","order":714,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":507,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l43.attn.k_norm","order":715,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":508,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l43.attn.rope","order":716,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":509,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l43.attn.kv_write","order":717,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":510,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l43.attn.attend","order":718,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":511,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l43.attn.gate_mul","order":719,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":512,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l43.attn.o","order":720,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":513,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l43.attn.add","order":721,"layer":43,"block":"l43.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":514,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l43.dense_ffn.post_norm","order":722,"layer":43,"block":"l43.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":514,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l43.dense_ffn.gate_up","order":723,"layer":43,"block":"l43.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":515,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l43.dense_ffn.act","order":724,"layer":43,"block":"l43.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":516,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l43.dense_ffn.down","order":725,"layer":43,"block":"l43.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":517,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l43.dense_ffn.add","order":726,"layer":43,"block":"l43.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":518,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l44.gdn.input_norm","order":727,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":518,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l44.gdn.qkvz","order":728,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":519,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l44.gdn.ba","order":729,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":520,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l44.gdn.gates","order":730,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":520,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l44.gdn.conv","order":731,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":521,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l44.gdn.conv_ckpt","order":732,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":521,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l44.gdn.l2","order":733,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":521,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l44.gdn.recur","order":734,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":522,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l44.gdn.out_norm","order":735,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":523,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l44.gdn.out","order":736,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":524,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l44.gdn.add","order":737,"layer":44,"block":"l44.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":525,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l44.dense_ffn.post_norm","order":738,"layer":44,"block":"l44.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":525,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l44.dense_ffn.gate_up","order":739,"layer":44,"block":"l44.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":526,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l44.dense_ffn.act","order":740,"layer":44,"block":"l44.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":527,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l44.dense_ffn.down","order":741,"layer":44,"block":"l44.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":528,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l44.dense_ffn.add","order":742,"layer":44,"block":"l44.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":529,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l45.gdn.input_norm","order":743,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":529,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l45.gdn.qkvz","order":744,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":530,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l45.gdn.ba","order":745,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":531,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l45.gdn.gates","order":746,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":531,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l45.gdn.conv","order":747,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":532,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l45.gdn.conv_ckpt","order":748,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":532,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l45.gdn.l2","order":749,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":532,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l45.gdn.recur","order":750,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":533,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l45.gdn.out_norm","order":751,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":534,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l45.gdn.out","order":752,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":535,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l45.gdn.add","order":753,"layer":45,"block":"l45.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":536,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l45.dense_ffn.post_norm","order":754,"layer":45,"block":"l45.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":536,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l45.dense_ffn.gate_up","order":755,"layer":45,"block":"l45.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":537,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l45.dense_ffn.act","order":756,"layer":45,"block":"l45.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":538,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l45.dense_ffn.down","order":757,"layer":45,"block":"l45.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":539,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l45.dense_ffn.add","order":758,"layer":45,"block":"l45.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":540,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l46.gdn.input_norm","order":759,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":540,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l46.gdn.qkvz","order":760,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":541,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l46.gdn.ba","order":761,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":542,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l46.gdn.gates","order":762,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":542,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l46.gdn.conv","order":763,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":543,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l46.gdn.conv_ckpt","order":764,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":543,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l46.gdn.l2","order":765,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":543,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l46.gdn.recur","order":766,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":544,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l46.gdn.out_norm","order":767,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":545,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l46.gdn.out","order":768,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":546,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l46.gdn.add","order":769,"layer":46,"block":"l46.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":547,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l46.dense_ffn.post_norm","order":770,"layer":46,"block":"l46.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":547,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l46.dense_ffn.gate_up","order":771,"layer":46,"block":"l46.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":548,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l46.dense_ffn.act","order":772,"layer":46,"block":"l46.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":549,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l46.dense_ffn.down","order":773,"layer":46,"block":"l46.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":550,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l46.dense_ffn.add","order":774,"layer":46,"block":"l46.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":551,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l47.attn.input_norm","order":775,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":551,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l47.attn.q","order":776,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":552,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l47.attn.q_split","order":777,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":552,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l47.attn.k","order":778,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":553,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l47.attn.v","order":779,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":553,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l47.attn.q_norm","order":780,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":554,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l47.attn.k_norm","order":781,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":555,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l47.attn.rope","order":782,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":556,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l47.attn.kv_write","order":783,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":557,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l47.attn.attend","order":784,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":558,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l47.attn.gate_mul","order":785,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":559,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l47.attn.o","order":786,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":560,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l47.attn.add","order":787,"layer":47,"block":"l47.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":561,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l47.dense_ffn.post_norm","order":788,"layer":47,"block":"l47.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":561,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l47.dense_ffn.gate_up","order":789,"layer":47,"block":"l47.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":562,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l47.dense_ffn.act","order":790,"layer":47,"block":"l47.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":563,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l47.dense_ffn.down","order":791,"layer":47,"block":"l47.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":564,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l47.dense_ffn.add","order":792,"layer":47,"block":"l47.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":565,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l48.gdn.input_norm","order":793,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":565,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l48.gdn.qkvz","order":794,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":566,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l48.gdn.ba","order":795,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":567,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l48.gdn.gates","order":796,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":567,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l48.gdn.conv","order":797,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":568,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l48.gdn.conv_ckpt","order":798,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":568,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l48.gdn.l2","order":799,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":568,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l48.gdn.recur","order":800,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":569,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l48.gdn.out_norm","order":801,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":570,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l48.gdn.out","order":802,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":571,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l48.gdn.add","order":803,"layer":48,"block":"l48.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":572,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l48.dense_ffn.post_norm","order":804,"layer":48,"block":"l48.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":572,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l48.dense_ffn.gate_up","order":805,"layer":48,"block":"l48.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":573,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l48.dense_ffn.act","order":806,"layer":48,"block":"l48.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":574,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l48.dense_ffn.down","order":807,"layer":48,"block":"l48.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":575,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l48.dense_ffn.add","order":808,"layer":48,"block":"l48.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":576,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l49.gdn.input_norm","order":809,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":576,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l49.gdn.qkvz","order":810,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":577,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l49.gdn.ba","order":811,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":578,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l49.gdn.gates","order":812,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":578,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l49.gdn.conv","order":813,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":579,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l49.gdn.conv_ckpt","order":814,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":579,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l49.gdn.l2","order":815,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":579,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l49.gdn.recur","order":816,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":580,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l49.gdn.out_norm","order":817,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":581,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l49.gdn.out","order":818,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":582,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l49.gdn.add","order":819,"layer":49,"block":"l49.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":583,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l49.dense_ffn.post_norm","order":820,"layer":49,"block":"l49.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":583,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l49.dense_ffn.gate_up","order":821,"layer":49,"block":"l49.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":584,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l49.dense_ffn.act","order":822,"layer":49,"block":"l49.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":585,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l49.dense_ffn.down","order":823,"layer":49,"block":"l49.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":586,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l49.dense_ffn.add","order":824,"layer":49,"block":"l49.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":587,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l50.gdn.input_norm","order":825,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":587,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l50.gdn.qkvz","order":826,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":588,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l50.gdn.ba","order":827,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":589,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l50.gdn.gates","order":828,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":589,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l50.gdn.conv","order":829,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":590,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l50.gdn.conv_ckpt","order":830,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":590,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l50.gdn.l2","order":831,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":590,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l50.gdn.recur","order":832,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":591,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l50.gdn.out_norm","order":833,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":592,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l50.gdn.out","order":834,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":593,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l50.gdn.add","order":835,"layer":50,"block":"l50.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":594,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l50.dense_ffn.post_norm","order":836,"layer":50,"block":"l50.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":594,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l50.dense_ffn.gate_up","order":837,"layer":50,"block":"l50.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":595,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l50.dense_ffn.act","order":838,"layer":50,"block":"l50.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":596,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l50.dense_ffn.down","order":839,"layer":50,"block":"l50.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":597,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l50.dense_ffn.add","order":840,"layer":50,"block":"l50.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":598,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l51.attn.input_norm","order":841,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":598,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l51.attn.q","order":842,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":599,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l51.attn.q_split","order":843,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":599,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l51.attn.k","order":844,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":600,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l51.attn.v","order":845,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":600,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l51.attn.q_norm","order":846,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":601,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l51.attn.k_norm","order":847,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":602,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l51.attn.rope","order":848,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":603,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l51.attn.kv_write","order":849,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":604,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l51.attn.attend","order":850,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":605,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l51.attn.gate_mul","order":851,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":606,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l51.attn.o","order":852,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":607,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l51.attn.add","order":853,"layer":51,"block":"l51.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":608,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l51.dense_ffn.post_norm","order":854,"layer":51,"block":"l51.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":608,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l51.dense_ffn.gate_up","order":855,"layer":51,"block":"l51.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":609,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l51.dense_ffn.act","order":856,"layer":51,"block":"l51.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":610,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l51.dense_ffn.down","order":857,"layer":51,"block":"l51.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":611,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l51.dense_ffn.add","order":858,"layer":51,"block":"l51.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":612,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l52.gdn.input_norm","order":859,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":612,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l52.gdn.qkvz","order":860,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":613,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l52.gdn.ba","order":861,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":614,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l52.gdn.gates","order":862,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":614,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l52.gdn.conv","order":863,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":615,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l52.gdn.conv_ckpt","order":864,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":615,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l52.gdn.l2","order":865,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":615,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l52.gdn.recur","order":866,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":616,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l52.gdn.out_norm","order":867,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":617,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l52.gdn.out","order":868,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":618,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l52.gdn.add","order":869,"layer":52,"block":"l52.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":619,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l52.dense_ffn.post_norm","order":870,"layer":52,"block":"l52.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":619,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l52.dense_ffn.gate_up","order":871,"layer":52,"block":"l52.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":620,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l52.dense_ffn.act","order":872,"layer":52,"block":"l52.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":621,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l52.dense_ffn.down","order":873,"layer":52,"block":"l52.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":622,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l52.dense_ffn.add","order":874,"layer":52,"block":"l52.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":623,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l53.gdn.input_norm","order":875,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":623,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l53.gdn.qkvz","order":876,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":624,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l53.gdn.ba","order":877,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":625,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l53.gdn.gates","order":878,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":625,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l53.gdn.conv","order":879,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":626,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l53.gdn.conv_ckpt","order":880,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":626,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l53.gdn.l2","order":881,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":626,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l53.gdn.recur","order":882,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":627,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l53.gdn.out_norm","order":883,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":628,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l53.gdn.out","order":884,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":629,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l53.gdn.add","order":885,"layer":53,"block":"l53.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":630,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l53.dense_ffn.post_norm","order":886,"layer":53,"block":"l53.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":630,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l53.dense_ffn.gate_up","order":887,"layer":53,"block":"l53.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":631,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l53.dense_ffn.act","order":888,"layer":53,"block":"l53.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":632,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l53.dense_ffn.down","order":889,"layer":53,"block":"l53.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":633,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l53.dense_ffn.add","order":890,"layer":53,"block":"l53.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":634,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l54.gdn.input_norm","order":891,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":634,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l54.gdn.qkvz","order":892,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":635,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l54.gdn.ba","order":893,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":636,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l54.gdn.gates","order":894,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":636,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l54.gdn.conv","order":895,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":637,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l54.gdn.conv_ckpt","order":896,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":637,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l54.gdn.l2","order":897,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":637,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l54.gdn.recur","order":898,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":638,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l54.gdn.out_norm","order":899,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":639,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l54.gdn.out","order":900,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":640,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l54.gdn.add","order":901,"layer":54,"block":"l54.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":641,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l54.dense_ffn.post_norm","order":902,"layer":54,"block":"l54.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":641,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l54.dense_ffn.gate_up","order":903,"layer":54,"block":"l54.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":642,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l54.dense_ffn.act","order":904,"layer":54,"block":"l54.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":643,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l54.dense_ffn.down","order":905,"layer":54,"block":"l54.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":644,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l54.dense_ffn.add","order":906,"layer":54,"block":"l54.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":645,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l55.attn.input_norm","order":907,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":645,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l55.attn.q","order":908,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":646,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l55.attn.q_split","order":909,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":646,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l55.attn.k","order":910,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":647,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l55.attn.v","order":911,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":647,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l55.attn.q_norm","order":912,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":648,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l55.attn.k_norm","order":913,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":649,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l55.attn.rope","order":914,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":650,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l55.attn.kv_write","order":915,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":651,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l55.attn.attend","order":916,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":652,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l55.attn.gate_mul","order":917,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":653,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l55.attn.o","order":918,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":654,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l55.attn.add","order":919,"layer":55,"block":"l55.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":655,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l55.dense_ffn.post_norm","order":920,"layer":55,"block":"l55.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":655,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l55.dense_ffn.gate_up","order":921,"layer":55,"block":"l55.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":656,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l55.dense_ffn.act","order":922,"layer":55,"block":"l55.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":657,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l55.dense_ffn.down","order":923,"layer":55,"block":"l55.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":658,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l55.dense_ffn.add","order":924,"layer":55,"block":"l55.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":659,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l56.gdn.input_norm","order":925,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":659,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l56.gdn.qkvz","order":926,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":660,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l56.gdn.ba","order":927,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":661,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l56.gdn.gates","order":928,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":661,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l56.gdn.conv","order":929,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":662,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l56.gdn.conv_ckpt","order":930,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":662,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l56.gdn.l2","order":931,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":662,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l56.gdn.recur","order":932,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":663,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l56.gdn.out_norm","order":933,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":664,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l56.gdn.out","order":934,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":665,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l56.gdn.add","order":935,"layer":56,"block":"l56.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":666,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l56.dense_ffn.post_norm","order":936,"layer":56,"block":"l56.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":666,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l56.dense_ffn.gate_up","order":937,"layer":56,"block":"l56.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":667,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l56.dense_ffn.act","order":938,"layer":56,"block":"l56.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":668,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l56.dense_ffn.down","order":939,"layer":56,"block":"l56.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":669,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l56.dense_ffn.add","order":940,"layer":56,"block":"l56.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":670,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l57.gdn.input_norm","order":941,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":670,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l57.gdn.qkvz","order":942,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":671,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l57.gdn.ba","order":943,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":672,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l57.gdn.gates","order":944,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":672,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l57.gdn.conv","order":945,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":673,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l57.gdn.conv_ckpt","order":946,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":673,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l57.gdn.l2","order":947,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":673,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l57.gdn.recur","order":948,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":674,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l57.gdn.out_norm","order":949,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":675,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l57.gdn.out","order":950,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":676,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l57.gdn.add","order":951,"layer":57,"block":"l57.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":677,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l57.dense_ffn.post_norm","order":952,"layer":57,"block":"l57.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":677,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l57.dense_ffn.gate_up","order":953,"layer":57,"block":"l57.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":678,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l57.dense_ffn.act","order":954,"layer":57,"block":"l57.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":679,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l57.dense_ffn.down","order":955,"layer":57,"block":"l57.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":680,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l57.dense_ffn.add","order":956,"layer":57,"block":"l57.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":681,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l58.gdn.input_norm","order":957,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":681,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l58.gdn.qkvz","order":958,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":682,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l58.gdn.ba","order":959,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":683,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l58.gdn.gates","order":960,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":683,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l58.gdn.conv","order":961,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":684,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l58.gdn.conv_ckpt","order":962,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":684,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l58.gdn.l2","order":963,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":684,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l58.gdn.recur","order":964,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":685,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l58.gdn.out_norm","order":965,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":686,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l58.gdn.out","order":966,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":687,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l58.gdn.add","order":967,"layer":58,"block":"l58.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":688,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l58.dense_ffn.post_norm","order":968,"layer":58,"block":"l58.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":688,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l58.dense_ffn.gate_up","order":969,"layer":58,"block":"l58.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":689,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l58.dense_ffn.act","order":970,"layer":58,"block":"l58.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":690,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l58.dense_ffn.down","order":971,"layer":58,"block":"l58.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":691,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l58.dense_ffn.add","order":972,"layer":58,"block":"l58.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":692,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l59.attn.input_norm","order":973,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":692,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l59.attn.q","order":974,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":693,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l59.attn.q_split","order":975,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":693,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l59.attn.k","order":976,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":694,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l59.attn.v","order":977,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":694,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l59.attn.q_norm","order":978,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":695,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l59.attn.k_norm","order":979,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":696,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l59.attn.rope","order":980,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":697,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l59.attn.kv_write","order":981,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":698,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l59.attn.attend","order":982,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":699,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l59.attn.gate_mul","order":983,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":700,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l59.attn.o","order":984,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":701,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l59.attn.add","order":985,"layer":59,"block":"l59.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":702,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l59.dense_ffn.post_norm","order":986,"layer":59,"block":"l59.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":702,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l59.dense_ffn.gate_up","order":987,"layer":59,"block":"l59.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":703,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l59.dense_ffn.act","order":988,"layer":59,"block":"l59.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":704,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l59.dense_ffn.down","order":989,"layer":59,"block":"l59.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":705,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l59.dense_ffn.add","order":990,"layer":59,"block":"l59.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":706,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l60.gdn.input_norm","order":991,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":706,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l60.gdn.qkvz","order":992,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":707,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l60.gdn.ba","order":993,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":708,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l60.gdn.gates","order":994,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":708,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l60.gdn.conv","order":995,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":709,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l60.gdn.conv_ckpt","order":996,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":709,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l60.gdn.l2","order":997,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":709,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l60.gdn.recur","order":998,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":710,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l60.gdn.out_norm","order":999,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":711,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l60.gdn.out","order":1000,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":712,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l60.gdn.add","order":1001,"layer":60,"block":"l60.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":713,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l60.dense_ffn.post_norm","order":1002,"layer":60,"block":"l60.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":713,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l60.dense_ffn.gate_up","order":1003,"layer":60,"block":"l60.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":714,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l60.dense_ffn.act","order":1004,"layer":60,"block":"l60.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":715,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l60.dense_ffn.down","order":1005,"layer":60,"block":"l60.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":716,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l60.dense_ffn.add","order":1006,"layer":60,"block":"l60.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":717,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l61.gdn.input_norm","order":1007,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":717,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l61.gdn.qkvz","order":1008,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":718,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l61.gdn.ba","order":1009,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":719,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l61.gdn.gates","order":1010,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":719,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l61.gdn.conv","order":1011,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":720,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l61.gdn.conv_ckpt","order":1012,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":720,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l61.gdn.l2","order":1013,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":720,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l61.gdn.recur","order":1014,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":721,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l61.gdn.out_norm","order":1015,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":722,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l61.gdn.out","order":1016,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":723,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l61.gdn.add","order":1017,"layer":61,"block":"l61.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":724,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l61.dense_ffn.post_norm","order":1018,"layer":61,"block":"l61.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":724,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l61.dense_ffn.gate_up","order":1019,"layer":61,"block":"l61.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":725,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l61.dense_ffn.act","order":1020,"layer":61,"block":"l61.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":726,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l61.dense_ffn.down","order":1021,"layer":61,"block":"l61.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":727,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l61.dense_ffn.add","order":1022,"layer":61,"block":"l61.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":728,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l62.gdn.input_norm","order":1023,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":728,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l62.gdn.qkvz","order":1024,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:qkvz","roles":["qkvz"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":729,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l62.gdn.ba","order":1025,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:ba","roles":["ba"],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"f32"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":730,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l62.gdn.gates","order":1026,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_gates","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32,f32"},"steps":["compute f32"],"group":730,"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","point":{"gate":"softplus","decay":"head"},"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l62.gdn.conv","order":1027,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"conv1d_update","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"f32"},"steps":["state f32","compute f32"],"group":731,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:update"},{"id":"l62.gdn.conv_ckpt","order":1028,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"state_snapshot","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":null},"steps":["state f32"],"group":731,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"conv:snapshot"},{"id":"l62.gdn.l2","order":1029,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"l2_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["compute f32"],"group":731,"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l62.gdn.recur","order":1030,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"gdn_recurrence","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"f32"},"steps":["state f32","compute f32"],"group":732,"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","point":{"decay":"head","gate_act":"silu"},"compute":"cuda_core","numerics":"reference","where":"device","state":"h:update"},{"id":"l62.gdn.out_norm","order":1031,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"gated_rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":733,"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l62.gdn.out","order":1032,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"linear:gdn_out","roles":["gdn_out"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":734,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l62.gdn.add","order":1033,"layer":62,"block":"l62.gdn","block_type":"gdn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":735,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l62.dense_ffn.post_norm","order":1034,"layer":62,"block":"l62.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":735,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l62.dense_ffn.gate_up","order":1035,"layer":62,"block":"l62.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":736,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l62.dense_ffn.act","order":1036,"layer":62,"block":"l62.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":737,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l62.dense_ffn.down","order":1037,"layer":62,"block":"l62.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":738,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l62.dense_ffn.add","order":1038,"layer":62,"block":"l62.dense_ffn","block_type":"dense_ffn","kind":"linear_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":739,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l63.attn.input_norm","order":1039,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":739,"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","point":null,"compute":"memory","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","where":"device","state":null},{"id":"l63.attn.q","order":1040,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"linear:q","roles":["q"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":740,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l63.attn.q_split","order":1041,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"split","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["move bf16"],"group":740,"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l63.attn.k","order":1042,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"linear:k","roles":["k"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":741,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l63.attn.v","order":1043,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"linear:v","roles":["v"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":741,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l63.attn.q_norm","order":1044,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":742,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l63.attn.k_norm","order":1045,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"qk_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":743,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l63.attn.rope","order":1046,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"rope","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"steps":["compute f32"],"group":744,"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l63.attn.kv_write","order":1047,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"kv_write","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":null},"steps":["cache bf16"],"group":745,"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","point":null,"compute":"memory","numerics":"reference","where":"device","state":"k:write,v:write"},{"id":"l63.attn.attend","order":1048,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"paged_attention","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["cache bf16","scores f32","softmax f32","accumulate f32"],"group":746,"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":"k:read,v:read"},{"id":"l63.attn.gate_mul","order":1049,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"sigmoid_gate_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":747,"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l63.attn.o","order":1050,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"linear:o","roles":["o"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":748,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l63.attn.add","order":1051,"layer":63,"block":"l63.attn","block_type":"attn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":749,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l63.dense_ffn.post_norm","order":1052,"layer":63,"block":"l63.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"rms_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"f32","output":"bf16"},"steps":["compute f32"],"group":749,"kernel":"norm::residual_add_rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l63.dense_ffn.gate_up","order":1053,"layer":63,"block":"l63.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:gate_up","roles":["gate_up"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":750,"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l63.dense_ffn.act","order":1054,"layer":63,"block":"l63.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"silu_mul","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":751,"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"l63.dense_ffn.down","order":1055,"layer":63,"block":"l63.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"linear:down","roles":["down"],"inserted":false,"precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight nvfp4/g16->bf16","mma bf16*bf16","accumulate f32","scale f32"],"group":752,"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"l63.dense_ffn.add","order":1056,"layer":63,"block":"l63.dense_ffn","block_type":"dense_ffn","kind":"full_attention","op":"residual_add","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":753,"kernel":"residual_add::bf16_residual_add","family":"residual_add","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"head.final_norm","order":1057,"layer":null,"block":"head","block_type":"head","kind":null,"op":"final_norm","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"bf16"},"steps":["compute f32"],"group":754,"kernel":"norm::rms_norm","family":"rms_norm","point":null,"compute":"memory","numerics":"reference","where":"device","state":null},{"id":"head.lm_head","order":1058,"layer":null,"block":"head","block_type":"head","kind":null,"op":"lm_head","roles":[],"inserted":false,"precision":{"weight":"bf16","activation":"bf16","output":"bf16"},"steps":["act bf16","weight bf16->bf16","mma bf16*bf16","accumulate f32","scale none"],"group":755,"kernel":"gemv::dense_gemv_bf16","family":"dense_bf16","point":null,"compute":"cuda_core","numerics":"reference","where":"device","state":null},{"id":"head.argmax","order":1059,"layer":null,"block":"head","block_type":"head","kind":null,"op":"argmax","roles":[],"inserted":false,"precision":{"weight":null,"activation":"bf16","output":"i32"},"steps":["compare bf16"],"group":756,"kernel":null,"family":null,"point":null,"compute":null,"numerics":"reference","where":"host","state":null}],"edges":[{"id":"embed.h","from":0,"to":1,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"embed.h","from":0,"to":11,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.gdn.xn","from":1,"to":2,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.gdn.xn","from":1,"to":3,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.gdn.qkvz","from":2,"to":5,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l0.gdn.qkvz","from":2,"to":9,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l0.gdn.ba","from":3,"to":4,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l0.gdn.decay","from":4,"to":8,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l0.gdn.beta","from":4,"to":8,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l0.gdn.conv","from":5,"to":6,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l0.gdn.conv","from":5,"to":7,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l0.gdn.qkv","from":7,"to":8,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l0.gdn.core","from":8,"to":9,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l0.gdn.gated","from":9,"to":10,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l0.gdn.o","from":10,"to":11,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.gdn.y","from":11,"to":12,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.gdn.y","from":11,"to":16,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.dense_ffn.xn","from":12,"to":13,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.dense_ffn.gu","from":13,"to":14,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l0.dense_ffn.act","from":14,"to":15,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l0.dense_ffn.d","from":15,"to":16,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.dense_ffn.y","from":16,"to":17,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l0.dense_ffn.y","from":16,"to":27,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.gdn.xn","from":17,"to":18,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.gdn.xn","from":17,"to":19,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.gdn.qkvz","from":18,"to":21,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l1.gdn.qkvz","from":18,"to":25,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l1.gdn.ba","from":19,"to":20,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l1.gdn.decay","from":20,"to":24,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l1.gdn.beta","from":20,"to":24,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l1.gdn.conv","from":21,"to":22,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l1.gdn.conv","from":21,"to":23,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l1.gdn.qkv","from":23,"to":24,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l1.gdn.core","from":24,"to":25,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l1.gdn.gated","from":25,"to":26,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l1.gdn.o","from":26,"to":27,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.gdn.y","from":27,"to":28,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.gdn.y","from":27,"to":32,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.dense_ffn.xn","from":28,"to":29,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.dense_ffn.gu","from":29,"to":30,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l1.dense_ffn.act","from":30,"to":31,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l1.dense_ffn.d","from":31,"to":32,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.dense_ffn.y","from":32,"to":33,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l1.dense_ffn.y","from":32,"to":43,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.gdn.xn","from":33,"to":34,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.gdn.xn","from":33,"to":35,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.gdn.qkvz","from":34,"to":37,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l2.gdn.qkvz","from":34,"to":41,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l2.gdn.ba","from":35,"to":36,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l2.gdn.decay","from":36,"to":40,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l2.gdn.beta","from":36,"to":40,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l2.gdn.conv","from":37,"to":38,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l2.gdn.conv","from":37,"to":39,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l2.gdn.qkv","from":39,"to":40,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l2.gdn.core","from":40,"to":41,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l2.gdn.gated","from":41,"to":42,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l2.gdn.o","from":42,"to":43,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.gdn.y","from":43,"to":44,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.gdn.y","from":43,"to":48,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.dense_ffn.xn","from":44,"to":45,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.dense_ffn.gu","from":45,"to":46,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l2.dense_ffn.act","from":46,"to":47,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l2.dense_ffn.d","from":47,"to":48,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.dense_ffn.y","from":48,"to":49,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l2.dense_ffn.y","from":48,"to":61,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.attn.xn","from":49,"to":50,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.attn.xn","from":49,"to":52,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.attn.xn","from":49,"to":53,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.attn.qg","from":50,"to":51,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l3.attn.q","from":51,"to":54,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l3.attn.gate","from":51,"to":59,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l3.attn.k","from":52,"to":55,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l3.attn.v","from":53,"to":57,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l3.attn.v","from":53,"to":58,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l3.attn.qn","from":54,"to":56,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l3.attn.kn","from":55,"to":56,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l3.attn.qr","from":56,"to":58,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l3.attn.kr","from":56,"to":57,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l3.attn.kr","from":56,"to":58,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l3.attn.a","from":58,"to":59,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l3.attn.ag","from":59,"to":60,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l3.attn.o","from":60,"to":61,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.attn.y","from":61,"to":62,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.attn.y","from":61,"to":66,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.dense_ffn.xn","from":62,"to":63,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.dense_ffn.gu","from":63,"to":64,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l3.dense_ffn.act","from":64,"to":65,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l3.dense_ffn.d","from":65,"to":66,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.dense_ffn.y","from":66,"to":67,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l3.dense_ffn.y","from":66,"to":77,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.gdn.xn","from":67,"to":68,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.gdn.xn","from":67,"to":69,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.gdn.qkvz","from":68,"to":71,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l4.gdn.qkvz","from":68,"to":75,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l4.gdn.ba","from":69,"to":70,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l4.gdn.decay","from":70,"to":74,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l4.gdn.beta","from":70,"to":74,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l4.gdn.conv","from":71,"to":72,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l4.gdn.conv","from":71,"to":73,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l4.gdn.qkv","from":73,"to":74,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l4.gdn.core","from":74,"to":75,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l4.gdn.gated","from":75,"to":76,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l4.gdn.o","from":76,"to":77,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.gdn.y","from":77,"to":78,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.gdn.y","from":77,"to":82,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.dense_ffn.xn","from":78,"to":79,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.dense_ffn.gu","from":79,"to":80,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l4.dense_ffn.act","from":80,"to":81,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l4.dense_ffn.d","from":81,"to":82,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.dense_ffn.y","from":82,"to":83,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l4.dense_ffn.y","from":82,"to":93,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.gdn.xn","from":83,"to":84,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.gdn.xn","from":83,"to":85,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.gdn.qkvz","from":84,"to":87,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l5.gdn.qkvz","from":84,"to":91,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l5.gdn.ba","from":85,"to":86,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l5.gdn.decay","from":86,"to":90,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l5.gdn.beta","from":86,"to":90,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l5.gdn.conv","from":87,"to":88,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l5.gdn.conv","from":87,"to":89,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l5.gdn.qkv","from":89,"to":90,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l5.gdn.core","from":90,"to":91,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l5.gdn.gated","from":91,"to":92,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l5.gdn.o","from":92,"to":93,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.gdn.y","from":93,"to":94,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.gdn.y","from":93,"to":98,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.dense_ffn.xn","from":94,"to":95,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.dense_ffn.gu","from":95,"to":96,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l5.dense_ffn.act","from":96,"to":97,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l5.dense_ffn.d","from":97,"to":98,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.dense_ffn.y","from":98,"to":99,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l5.dense_ffn.y","from":98,"to":109,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.gdn.xn","from":99,"to":100,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.gdn.xn","from":99,"to":101,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.gdn.qkvz","from":100,"to":103,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l6.gdn.qkvz","from":100,"to":107,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l6.gdn.ba","from":101,"to":102,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l6.gdn.decay","from":102,"to":106,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l6.gdn.beta","from":102,"to":106,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l6.gdn.conv","from":103,"to":104,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l6.gdn.conv","from":103,"to":105,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l6.gdn.qkv","from":105,"to":106,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l6.gdn.core","from":106,"to":107,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l6.gdn.gated","from":107,"to":108,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l6.gdn.o","from":108,"to":109,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.gdn.y","from":109,"to":110,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.gdn.y","from":109,"to":114,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.dense_ffn.xn","from":110,"to":111,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.dense_ffn.gu","from":111,"to":112,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l6.dense_ffn.act","from":112,"to":113,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l6.dense_ffn.d","from":113,"to":114,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.dense_ffn.y","from":114,"to":115,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l6.dense_ffn.y","from":114,"to":127,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.attn.xn","from":115,"to":116,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.attn.xn","from":115,"to":118,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.attn.xn","from":115,"to":119,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.attn.qg","from":116,"to":117,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l7.attn.q","from":117,"to":120,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l7.attn.gate","from":117,"to":125,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l7.attn.k","from":118,"to":121,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l7.attn.v","from":119,"to":123,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l7.attn.v","from":119,"to":124,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l7.attn.qn","from":120,"to":122,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l7.attn.kn","from":121,"to":122,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l7.attn.qr","from":122,"to":124,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l7.attn.kr","from":122,"to":123,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l7.attn.kr","from":122,"to":124,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l7.attn.a","from":124,"to":125,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l7.attn.ag","from":125,"to":126,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l7.attn.o","from":126,"to":127,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.attn.y","from":127,"to":128,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.attn.y","from":127,"to":132,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.dense_ffn.xn","from":128,"to":129,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.dense_ffn.gu","from":129,"to":130,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l7.dense_ffn.act","from":130,"to":131,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l7.dense_ffn.d","from":131,"to":132,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.dense_ffn.y","from":132,"to":133,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l7.dense_ffn.y","from":132,"to":143,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.gdn.xn","from":133,"to":134,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.gdn.xn","from":133,"to":135,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.gdn.qkvz","from":134,"to":137,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l8.gdn.qkvz","from":134,"to":141,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l8.gdn.ba","from":135,"to":136,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l8.gdn.decay","from":136,"to":140,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l8.gdn.beta","from":136,"to":140,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l8.gdn.conv","from":137,"to":138,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l8.gdn.conv","from":137,"to":139,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l8.gdn.qkv","from":139,"to":140,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l8.gdn.core","from":140,"to":141,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l8.gdn.gated","from":141,"to":142,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l8.gdn.o","from":142,"to":143,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.gdn.y","from":143,"to":144,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.gdn.y","from":143,"to":148,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.dense_ffn.xn","from":144,"to":145,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.dense_ffn.gu","from":145,"to":146,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l8.dense_ffn.act","from":146,"to":147,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l8.dense_ffn.d","from":147,"to":148,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.dense_ffn.y","from":148,"to":149,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l8.dense_ffn.y","from":148,"to":159,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.gdn.xn","from":149,"to":150,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.gdn.xn","from":149,"to":151,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.gdn.qkvz","from":150,"to":153,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l9.gdn.qkvz","from":150,"to":157,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l9.gdn.ba","from":151,"to":152,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l9.gdn.decay","from":152,"to":156,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l9.gdn.beta","from":152,"to":156,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l9.gdn.conv","from":153,"to":154,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l9.gdn.conv","from":153,"to":155,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l9.gdn.qkv","from":155,"to":156,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l9.gdn.core","from":156,"to":157,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l9.gdn.gated","from":157,"to":158,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l9.gdn.o","from":158,"to":159,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.gdn.y","from":159,"to":160,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.gdn.y","from":159,"to":164,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.dense_ffn.xn","from":160,"to":161,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.dense_ffn.gu","from":161,"to":162,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l9.dense_ffn.act","from":162,"to":163,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l9.dense_ffn.d","from":163,"to":164,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.dense_ffn.y","from":164,"to":165,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l9.dense_ffn.y","from":164,"to":175,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.gdn.xn","from":165,"to":166,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.gdn.xn","from":165,"to":167,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.gdn.qkvz","from":166,"to":169,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l10.gdn.qkvz","from":166,"to":173,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l10.gdn.ba","from":167,"to":168,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l10.gdn.decay","from":168,"to":172,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l10.gdn.beta","from":168,"to":172,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l10.gdn.conv","from":169,"to":170,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l10.gdn.conv","from":169,"to":171,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l10.gdn.qkv","from":171,"to":172,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l10.gdn.core","from":172,"to":173,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l10.gdn.gated","from":173,"to":174,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l10.gdn.o","from":174,"to":175,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.gdn.y","from":175,"to":176,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.gdn.y","from":175,"to":180,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.dense_ffn.xn","from":176,"to":177,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.dense_ffn.gu","from":177,"to":178,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l10.dense_ffn.act","from":178,"to":179,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l10.dense_ffn.d","from":179,"to":180,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.dense_ffn.y","from":180,"to":181,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l10.dense_ffn.y","from":180,"to":193,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.attn.xn","from":181,"to":182,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.attn.xn","from":181,"to":184,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.attn.xn","from":181,"to":185,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.attn.qg","from":182,"to":183,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l11.attn.q","from":183,"to":186,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l11.attn.gate","from":183,"to":191,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l11.attn.k","from":184,"to":187,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l11.attn.v","from":185,"to":189,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l11.attn.v","from":185,"to":190,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l11.attn.qn","from":186,"to":188,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l11.attn.kn","from":187,"to":188,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l11.attn.qr","from":188,"to":190,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l11.attn.kr","from":188,"to":189,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l11.attn.kr","from":188,"to":190,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l11.attn.a","from":190,"to":191,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l11.attn.ag","from":191,"to":192,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l11.attn.o","from":192,"to":193,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.attn.y","from":193,"to":194,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.attn.y","from":193,"to":198,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.dense_ffn.xn","from":194,"to":195,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.dense_ffn.gu","from":195,"to":196,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l11.dense_ffn.act","from":196,"to":197,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l11.dense_ffn.d","from":197,"to":198,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.dense_ffn.y","from":198,"to":199,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l11.dense_ffn.y","from":198,"to":209,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.gdn.xn","from":199,"to":200,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.gdn.xn","from":199,"to":201,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.gdn.qkvz","from":200,"to":203,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l12.gdn.qkvz","from":200,"to":207,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l12.gdn.ba","from":201,"to":202,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l12.gdn.decay","from":202,"to":206,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l12.gdn.beta","from":202,"to":206,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l12.gdn.conv","from":203,"to":204,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l12.gdn.conv","from":203,"to":205,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l12.gdn.qkv","from":205,"to":206,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l12.gdn.core","from":206,"to":207,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l12.gdn.gated","from":207,"to":208,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l12.gdn.o","from":208,"to":209,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.gdn.y","from":209,"to":210,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.gdn.y","from":209,"to":214,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.dense_ffn.xn","from":210,"to":211,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.dense_ffn.gu","from":211,"to":212,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l12.dense_ffn.act","from":212,"to":213,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l12.dense_ffn.d","from":213,"to":214,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.dense_ffn.y","from":214,"to":215,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l12.dense_ffn.y","from":214,"to":225,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.gdn.xn","from":215,"to":216,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.gdn.xn","from":215,"to":217,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.gdn.qkvz","from":216,"to":219,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l13.gdn.qkvz","from":216,"to":223,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l13.gdn.ba","from":217,"to":218,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l13.gdn.decay","from":218,"to":222,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l13.gdn.beta","from":218,"to":222,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l13.gdn.conv","from":219,"to":220,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l13.gdn.conv","from":219,"to":221,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l13.gdn.qkv","from":221,"to":222,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l13.gdn.core","from":222,"to":223,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l13.gdn.gated","from":223,"to":224,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l13.gdn.o","from":224,"to":225,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.gdn.y","from":225,"to":226,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.gdn.y","from":225,"to":230,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.dense_ffn.xn","from":226,"to":227,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.dense_ffn.gu","from":227,"to":228,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l13.dense_ffn.act","from":228,"to":229,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l13.dense_ffn.d","from":229,"to":230,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.dense_ffn.y","from":230,"to":231,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l13.dense_ffn.y","from":230,"to":241,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.gdn.xn","from":231,"to":232,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.gdn.xn","from":231,"to":233,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.gdn.qkvz","from":232,"to":235,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l14.gdn.qkvz","from":232,"to":239,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l14.gdn.ba","from":233,"to":234,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l14.gdn.decay","from":234,"to":238,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l14.gdn.beta","from":234,"to":238,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l14.gdn.conv","from":235,"to":236,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l14.gdn.conv","from":235,"to":237,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l14.gdn.qkv","from":237,"to":238,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l14.gdn.core","from":238,"to":239,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l14.gdn.gated","from":239,"to":240,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l14.gdn.o","from":240,"to":241,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.gdn.y","from":241,"to":242,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.gdn.y","from":241,"to":246,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.dense_ffn.xn","from":242,"to":243,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.dense_ffn.gu","from":243,"to":244,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l14.dense_ffn.act","from":244,"to":245,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l14.dense_ffn.d","from":245,"to":246,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.dense_ffn.y","from":246,"to":247,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l14.dense_ffn.y","from":246,"to":259,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.attn.xn","from":247,"to":248,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.attn.xn","from":247,"to":250,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.attn.xn","from":247,"to":251,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.attn.qg","from":248,"to":249,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l15.attn.q","from":249,"to":252,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l15.attn.gate","from":249,"to":257,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l15.attn.k","from":250,"to":253,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l15.attn.v","from":251,"to":255,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l15.attn.v","from":251,"to":256,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l15.attn.qn","from":252,"to":254,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l15.attn.kn","from":253,"to":254,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l15.attn.qr","from":254,"to":256,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l15.attn.kr","from":254,"to":255,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l15.attn.kr","from":254,"to":256,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l15.attn.a","from":256,"to":257,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l15.attn.ag","from":257,"to":258,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l15.attn.o","from":258,"to":259,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.attn.y","from":259,"to":260,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.attn.y","from":259,"to":264,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.dense_ffn.xn","from":260,"to":261,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.dense_ffn.gu","from":261,"to":262,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l15.dense_ffn.act","from":262,"to":263,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l15.dense_ffn.d","from":263,"to":264,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.dense_ffn.y","from":264,"to":265,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l15.dense_ffn.y","from":264,"to":275,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.gdn.xn","from":265,"to":266,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.gdn.xn","from":265,"to":267,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.gdn.qkvz","from":266,"to":269,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l16.gdn.qkvz","from":266,"to":273,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l16.gdn.ba","from":267,"to":268,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l16.gdn.decay","from":268,"to":272,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l16.gdn.beta","from":268,"to":272,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l16.gdn.conv","from":269,"to":270,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l16.gdn.conv","from":269,"to":271,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l16.gdn.qkv","from":271,"to":272,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l16.gdn.core","from":272,"to":273,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l16.gdn.gated","from":273,"to":274,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l16.gdn.o","from":274,"to":275,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.gdn.y","from":275,"to":276,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.gdn.y","from":275,"to":280,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.dense_ffn.xn","from":276,"to":277,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.dense_ffn.gu","from":277,"to":278,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l16.dense_ffn.act","from":278,"to":279,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l16.dense_ffn.d","from":279,"to":280,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.dense_ffn.y","from":280,"to":281,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l16.dense_ffn.y","from":280,"to":291,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.gdn.xn","from":281,"to":282,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.gdn.xn","from":281,"to":283,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.gdn.qkvz","from":282,"to":285,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l17.gdn.qkvz","from":282,"to":289,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l17.gdn.ba","from":283,"to":284,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l17.gdn.decay","from":284,"to":288,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l17.gdn.beta","from":284,"to":288,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l17.gdn.conv","from":285,"to":286,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l17.gdn.conv","from":285,"to":287,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l17.gdn.qkv","from":287,"to":288,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l17.gdn.core","from":288,"to":289,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l17.gdn.gated","from":289,"to":290,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l17.gdn.o","from":290,"to":291,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.gdn.y","from":291,"to":292,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.gdn.y","from":291,"to":296,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.dense_ffn.xn","from":292,"to":293,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.dense_ffn.gu","from":293,"to":294,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l17.dense_ffn.act","from":294,"to":295,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l17.dense_ffn.d","from":295,"to":296,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.dense_ffn.y","from":296,"to":297,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l17.dense_ffn.y","from":296,"to":307,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.gdn.xn","from":297,"to":298,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.gdn.xn","from":297,"to":299,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.gdn.qkvz","from":298,"to":301,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l18.gdn.qkvz","from":298,"to":305,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l18.gdn.ba","from":299,"to":300,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l18.gdn.decay","from":300,"to":304,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l18.gdn.beta","from":300,"to":304,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l18.gdn.conv","from":301,"to":302,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l18.gdn.conv","from":301,"to":303,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l18.gdn.qkv","from":303,"to":304,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l18.gdn.core","from":304,"to":305,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l18.gdn.gated","from":305,"to":306,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l18.gdn.o","from":306,"to":307,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.gdn.y","from":307,"to":308,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.gdn.y","from":307,"to":312,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.dense_ffn.xn","from":308,"to":309,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.dense_ffn.gu","from":309,"to":310,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l18.dense_ffn.act","from":310,"to":311,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l18.dense_ffn.d","from":311,"to":312,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.dense_ffn.y","from":312,"to":313,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l18.dense_ffn.y","from":312,"to":325,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.attn.xn","from":313,"to":314,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.attn.xn","from":313,"to":316,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.attn.xn","from":313,"to":317,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.attn.qg","from":314,"to":315,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l19.attn.q","from":315,"to":318,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l19.attn.gate","from":315,"to":323,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l19.attn.k","from":316,"to":319,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l19.attn.v","from":317,"to":321,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l19.attn.v","from":317,"to":322,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l19.attn.qn","from":318,"to":320,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l19.attn.kn","from":319,"to":320,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l19.attn.qr","from":320,"to":322,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l19.attn.kr","from":320,"to":321,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l19.attn.kr","from":320,"to":322,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l19.attn.a","from":322,"to":323,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l19.attn.ag","from":323,"to":324,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l19.attn.o","from":324,"to":325,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.attn.y","from":325,"to":326,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.attn.y","from":325,"to":330,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.dense_ffn.xn","from":326,"to":327,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.dense_ffn.gu","from":327,"to":328,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l19.dense_ffn.act","from":328,"to":329,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l19.dense_ffn.d","from":329,"to":330,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.dense_ffn.y","from":330,"to":331,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l19.dense_ffn.y","from":330,"to":341,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.gdn.xn","from":331,"to":332,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.gdn.xn","from":331,"to":333,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.gdn.qkvz","from":332,"to":335,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l20.gdn.qkvz","from":332,"to":339,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l20.gdn.ba","from":333,"to":334,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l20.gdn.decay","from":334,"to":338,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l20.gdn.beta","from":334,"to":338,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l20.gdn.conv","from":335,"to":336,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l20.gdn.conv","from":335,"to":337,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l20.gdn.qkv","from":337,"to":338,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l20.gdn.core","from":338,"to":339,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l20.gdn.gated","from":339,"to":340,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l20.gdn.o","from":340,"to":341,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.gdn.y","from":341,"to":342,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.gdn.y","from":341,"to":346,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.dense_ffn.xn","from":342,"to":343,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.dense_ffn.gu","from":343,"to":344,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l20.dense_ffn.act","from":344,"to":345,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l20.dense_ffn.d","from":345,"to":346,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.dense_ffn.y","from":346,"to":347,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l20.dense_ffn.y","from":346,"to":357,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.gdn.xn","from":347,"to":348,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.gdn.xn","from":347,"to":349,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.gdn.qkvz","from":348,"to":351,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l21.gdn.qkvz","from":348,"to":355,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l21.gdn.ba","from":349,"to":350,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l21.gdn.decay","from":350,"to":354,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l21.gdn.beta","from":350,"to":354,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l21.gdn.conv","from":351,"to":352,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l21.gdn.conv","from":351,"to":353,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l21.gdn.qkv","from":353,"to":354,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l21.gdn.core","from":354,"to":355,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l21.gdn.gated","from":355,"to":356,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l21.gdn.o","from":356,"to":357,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.gdn.y","from":357,"to":358,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.gdn.y","from":357,"to":362,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.dense_ffn.xn","from":358,"to":359,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.dense_ffn.gu","from":359,"to":360,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l21.dense_ffn.act","from":360,"to":361,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l21.dense_ffn.d","from":361,"to":362,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.dense_ffn.y","from":362,"to":363,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l21.dense_ffn.y","from":362,"to":373,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.gdn.xn","from":363,"to":364,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.gdn.xn","from":363,"to":365,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.gdn.qkvz","from":364,"to":367,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l22.gdn.qkvz","from":364,"to":371,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l22.gdn.ba","from":365,"to":366,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l22.gdn.decay","from":366,"to":370,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l22.gdn.beta","from":366,"to":370,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l22.gdn.conv","from":367,"to":368,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l22.gdn.conv","from":367,"to":369,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l22.gdn.qkv","from":369,"to":370,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l22.gdn.core","from":370,"to":371,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l22.gdn.gated","from":371,"to":372,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l22.gdn.o","from":372,"to":373,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.gdn.y","from":373,"to":374,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.gdn.y","from":373,"to":378,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.dense_ffn.xn","from":374,"to":375,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.dense_ffn.gu","from":375,"to":376,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l22.dense_ffn.act","from":376,"to":377,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l22.dense_ffn.d","from":377,"to":378,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.dense_ffn.y","from":378,"to":379,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l22.dense_ffn.y","from":378,"to":391,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.attn.xn","from":379,"to":380,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.attn.xn","from":379,"to":382,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.attn.xn","from":379,"to":383,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.attn.qg","from":380,"to":381,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l23.attn.q","from":381,"to":384,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l23.attn.gate","from":381,"to":389,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l23.attn.k","from":382,"to":385,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l23.attn.v","from":383,"to":387,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l23.attn.v","from":383,"to":388,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l23.attn.qn","from":384,"to":386,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l23.attn.kn","from":385,"to":386,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l23.attn.qr","from":386,"to":388,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l23.attn.kr","from":386,"to":387,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l23.attn.kr","from":386,"to":388,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l23.attn.a","from":388,"to":389,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l23.attn.ag","from":389,"to":390,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l23.attn.o","from":390,"to":391,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.attn.y","from":391,"to":392,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.attn.y","from":391,"to":396,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.dense_ffn.xn","from":392,"to":393,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.dense_ffn.gu","from":393,"to":394,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l23.dense_ffn.act","from":394,"to":395,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l23.dense_ffn.d","from":395,"to":396,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.dense_ffn.y","from":396,"to":397,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l23.dense_ffn.y","from":396,"to":407,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.gdn.xn","from":397,"to":398,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.gdn.xn","from":397,"to":399,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.gdn.qkvz","from":398,"to":401,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l24.gdn.qkvz","from":398,"to":405,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l24.gdn.ba","from":399,"to":400,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l24.gdn.decay","from":400,"to":404,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l24.gdn.beta","from":400,"to":404,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l24.gdn.conv","from":401,"to":402,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l24.gdn.conv","from":401,"to":403,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l24.gdn.qkv","from":403,"to":404,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l24.gdn.core","from":404,"to":405,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l24.gdn.gated","from":405,"to":406,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l24.gdn.o","from":406,"to":407,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.gdn.y","from":407,"to":408,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.gdn.y","from":407,"to":412,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.dense_ffn.xn","from":408,"to":409,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.dense_ffn.gu","from":409,"to":410,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l24.dense_ffn.act","from":410,"to":411,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l24.dense_ffn.d","from":411,"to":412,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.dense_ffn.y","from":412,"to":413,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l24.dense_ffn.y","from":412,"to":423,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.gdn.xn","from":413,"to":414,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.gdn.xn","from":413,"to":415,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.gdn.qkvz","from":414,"to":417,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l25.gdn.qkvz","from":414,"to":421,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l25.gdn.ba","from":415,"to":416,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l25.gdn.decay","from":416,"to":420,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l25.gdn.beta","from":416,"to":420,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l25.gdn.conv","from":417,"to":418,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l25.gdn.conv","from":417,"to":419,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l25.gdn.qkv","from":419,"to":420,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l25.gdn.core","from":420,"to":421,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l25.gdn.gated","from":421,"to":422,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l25.gdn.o","from":422,"to":423,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.gdn.y","from":423,"to":424,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.gdn.y","from":423,"to":428,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.dense_ffn.xn","from":424,"to":425,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.dense_ffn.gu","from":425,"to":426,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l25.dense_ffn.act","from":426,"to":427,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l25.dense_ffn.d","from":427,"to":428,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.dense_ffn.y","from":428,"to":429,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l25.dense_ffn.y","from":428,"to":439,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.gdn.xn","from":429,"to":430,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.gdn.xn","from":429,"to":431,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.gdn.qkvz","from":430,"to":433,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l26.gdn.qkvz","from":430,"to":437,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l26.gdn.ba","from":431,"to":432,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l26.gdn.decay","from":432,"to":436,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l26.gdn.beta","from":432,"to":436,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l26.gdn.conv","from":433,"to":434,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l26.gdn.conv","from":433,"to":435,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l26.gdn.qkv","from":435,"to":436,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l26.gdn.core","from":436,"to":437,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l26.gdn.gated","from":437,"to":438,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l26.gdn.o","from":438,"to":439,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.gdn.y","from":439,"to":440,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.gdn.y","from":439,"to":444,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.dense_ffn.xn","from":440,"to":441,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.dense_ffn.gu","from":441,"to":442,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l26.dense_ffn.act","from":442,"to":443,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l26.dense_ffn.d","from":443,"to":444,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.dense_ffn.y","from":444,"to":445,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l26.dense_ffn.y","from":444,"to":457,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.attn.xn","from":445,"to":446,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.attn.xn","from":445,"to":448,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.attn.xn","from":445,"to":449,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.attn.qg","from":446,"to":447,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l27.attn.q","from":447,"to":450,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l27.attn.gate","from":447,"to":455,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l27.attn.k","from":448,"to":451,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l27.attn.v","from":449,"to":453,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l27.attn.v","from":449,"to":454,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l27.attn.qn","from":450,"to":452,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l27.attn.kn","from":451,"to":452,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l27.attn.qr","from":452,"to":454,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l27.attn.kr","from":452,"to":453,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l27.attn.kr","from":452,"to":454,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l27.attn.a","from":454,"to":455,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l27.attn.ag","from":455,"to":456,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l27.attn.o","from":456,"to":457,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.attn.y","from":457,"to":458,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.attn.y","from":457,"to":462,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.dense_ffn.xn","from":458,"to":459,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.dense_ffn.gu","from":459,"to":460,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l27.dense_ffn.act","from":460,"to":461,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l27.dense_ffn.d","from":461,"to":462,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.dense_ffn.y","from":462,"to":463,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l27.dense_ffn.y","from":462,"to":473,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.gdn.xn","from":463,"to":464,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.gdn.xn","from":463,"to":465,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.gdn.qkvz","from":464,"to":467,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l28.gdn.qkvz","from":464,"to":471,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l28.gdn.ba","from":465,"to":466,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l28.gdn.decay","from":466,"to":470,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l28.gdn.beta","from":466,"to":470,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l28.gdn.conv","from":467,"to":468,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l28.gdn.conv","from":467,"to":469,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l28.gdn.qkv","from":469,"to":470,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l28.gdn.core","from":470,"to":471,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l28.gdn.gated","from":471,"to":472,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l28.gdn.o","from":472,"to":473,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.gdn.y","from":473,"to":474,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.gdn.y","from":473,"to":478,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.dense_ffn.xn","from":474,"to":475,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.dense_ffn.gu","from":475,"to":476,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l28.dense_ffn.act","from":476,"to":477,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l28.dense_ffn.d","from":477,"to":478,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.dense_ffn.y","from":478,"to":479,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l28.dense_ffn.y","from":478,"to":489,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.gdn.xn","from":479,"to":480,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.gdn.xn","from":479,"to":481,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.gdn.qkvz","from":480,"to":483,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l29.gdn.qkvz","from":480,"to":487,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l29.gdn.ba","from":481,"to":482,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l29.gdn.decay","from":482,"to":486,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l29.gdn.beta","from":482,"to":486,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l29.gdn.conv","from":483,"to":484,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l29.gdn.conv","from":483,"to":485,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l29.gdn.qkv","from":485,"to":486,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l29.gdn.core","from":486,"to":487,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l29.gdn.gated","from":487,"to":488,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l29.gdn.o","from":488,"to":489,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.gdn.y","from":489,"to":490,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.gdn.y","from":489,"to":494,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.dense_ffn.xn","from":490,"to":491,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.dense_ffn.gu","from":491,"to":492,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l29.dense_ffn.act","from":492,"to":493,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l29.dense_ffn.d","from":493,"to":494,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.dense_ffn.y","from":494,"to":495,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l29.dense_ffn.y","from":494,"to":505,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.gdn.xn","from":495,"to":496,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.gdn.xn","from":495,"to":497,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.gdn.qkvz","from":496,"to":499,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l30.gdn.qkvz","from":496,"to":503,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l30.gdn.ba","from":497,"to":498,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l30.gdn.decay","from":498,"to":502,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l30.gdn.beta","from":498,"to":502,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l30.gdn.conv","from":499,"to":500,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l30.gdn.conv","from":499,"to":501,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l30.gdn.qkv","from":501,"to":502,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l30.gdn.core","from":502,"to":503,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l30.gdn.gated","from":503,"to":504,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l30.gdn.o","from":504,"to":505,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.gdn.y","from":505,"to":506,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.gdn.y","from":505,"to":510,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.dense_ffn.xn","from":506,"to":507,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.dense_ffn.gu","from":507,"to":508,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l30.dense_ffn.act","from":508,"to":509,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l30.dense_ffn.d","from":509,"to":510,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.dense_ffn.y","from":510,"to":511,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l30.dense_ffn.y","from":510,"to":523,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.attn.xn","from":511,"to":512,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.attn.xn","from":511,"to":514,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.attn.xn","from":511,"to":515,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.attn.qg","from":512,"to":513,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l31.attn.q","from":513,"to":516,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l31.attn.gate","from":513,"to":521,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l31.attn.k","from":514,"to":517,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l31.attn.v","from":515,"to":519,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l31.attn.v","from":515,"to":520,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l31.attn.qn","from":516,"to":518,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l31.attn.kn","from":517,"to":518,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l31.attn.qr","from":518,"to":520,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l31.attn.kr","from":518,"to":519,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l31.attn.kr","from":518,"to":520,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l31.attn.a","from":520,"to":521,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l31.attn.ag","from":521,"to":522,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l31.attn.o","from":522,"to":523,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.attn.y","from":523,"to":524,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.attn.y","from":523,"to":528,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.dense_ffn.xn","from":524,"to":525,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.dense_ffn.gu","from":525,"to":526,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l31.dense_ffn.act","from":526,"to":527,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l31.dense_ffn.d","from":527,"to":528,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.dense_ffn.y","from":528,"to":529,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l31.dense_ffn.y","from":528,"to":539,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.gdn.xn","from":529,"to":530,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.gdn.xn","from":529,"to":531,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.gdn.qkvz","from":530,"to":533,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l32.gdn.qkvz","from":530,"to":537,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l32.gdn.ba","from":531,"to":532,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l32.gdn.decay","from":532,"to":536,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l32.gdn.beta","from":532,"to":536,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l32.gdn.conv","from":533,"to":534,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l32.gdn.conv","from":533,"to":535,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l32.gdn.qkv","from":535,"to":536,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l32.gdn.core","from":536,"to":537,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l32.gdn.gated","from":537,"to":538,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l32.gdn.o","from":538,"to":539,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.gdn.y","from":539,"to":540,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.gdn.y","from":539,"to":544,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.dense_ffn.xn","from":540,"to":541,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.dense_ffn.gu","from":541,"to":542,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l32.dense_ffn.act","from":542,"to":543,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l32.dense_ffn.d","from":543,"to":544,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.dense_ffn.y","from":544,"to":545,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l32.dense_ffn.y","from":544,"to":555,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.gdn.xn","from":545,"to":546,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.gdn.xn","from":545,"to":547,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.gdn.qkvz","from":546,"to":549,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l33.gdn.qkvz","from":546,"to":553,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l33.gdn.ba","from":547,"to":548,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l33.gdn.decay","from":548,"to":552,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l33.gdn.beta","from":548,"to":552,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l33.gdn.conv","from":549,"to":550,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l33.gdn.conv","from":549,"to":551,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l33.gdn.qkv","from":551,"to":552,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l33.gdn.core","from":552,"to":553,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l33.gdn.gated","from":553,"to":554,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l33.gdn.o","from":554,"to":555,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.gdn.y","from":555,"to":556,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.gdn.y","from":555,"to":560,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.dense_ffn.xn","from":556,"to":557,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.dense_ffn.gu","from":557,"to":558,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l33.dense_ffn.act","from":558,"to":559,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l33.dense_ffn.d","from":559,"to":560,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.dense_ffn.y","from":560,"to":561,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l33.dense_ffn.y","from":560,"to":571,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.gdn.xn","from":561,"to":562,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.gdn.xn","from":561,"to":563,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.gdn.qkvz","from":562,"to":565,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l34.gdn.qkvz","from":562,"to":569,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l34.gdn.ba","from":563,"to":564,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l34.gdn.decay","from":564,"to":568,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l34.gdn.beta","from":564,"to":568,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l34.gdn.conv","from":565,"to":566,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l34.gdn.conv","from":565,"to":567,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l34.gdn.qkv","from":567,"to":568,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l34.gdn.core","from":568,"to":569,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l34.gdn.gated","from":569,"to":570,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l34.gdn.o","from":570,"to":571,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.gdn.y","from":571,"to":572,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.gdn.y","from":571,"to":576,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.dense_ffn.xn","from":572,"to":573,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.dense_ffn.gu","from":573,"to":574,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l34.dense_ffn.act","from":574,"to":575,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l34.dense_ffn.d","from":575,"to":576,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.dense_ffn.y","from":576,"to":577,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l34.dense_ffn.y","from":576,"to":589,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.attn.xn","from":577,"to":578,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.attn.xn","from":577,"to":580,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.attn.xn","from":577,"to":581,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.attn.qg","from":578,"to":579,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l35.attn.q","from":579,"to":582,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l35.attn.gate","from":579,"to":587,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l35.attn.k","from":580,"to":583,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l35.attn.v","from":581,"to":585,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l35.attn.v","from":581,"to":586,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l35.attn.qn","from":582,"to":584,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l35.attn.kn","from":583,"to":584,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l35.attn.qr","from":584,"to":586,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l35.attn.kr","from":584,"to":585,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l35.attn.kr","from":584,"to":586,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l35.attn.a","from":586,"to":587,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l35.attn.ag","from":587,"to":588,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l35.attn.o","from":588,"to":589,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.attn.y","from":589,"to":590,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.attn.y","from":589,"to":594,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.dense_ffn.xn","from":590,"to":591,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.dense_ffn.gu","from":591,"to":592,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l35.dense_ffn.act","from":592,"to":593,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l35.dense_ffn.d","from":593,"to":594,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.dense_ffn.y","from":594,"to":595,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l35.dense_ffn.y","from":594,"to":605,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.gdn.xn","from":595,"to":596,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.gdn.xn","from":595,"to":597,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.gdn.qkvz","from":596,"to":599,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l36.gdn.qkvz","from":596,"to":603,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l36.gdn.ba","from":597,"to":598,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l36.gdn.decay","from":598,"to":602,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l36.gdn.beta","from":598,"to":602,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l36.gdn.conv","from":599,"to":600,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l36.gdn.conv","from":599,"to":601,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l36.gdn.qkv","from":601,"to":602,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l36.gdn.core","from":602,"to":603,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l36.gdn.gated","from":603,"to":604,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l36.gdn.o","from":604,"to":605,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.gdn.y","from":605,"to":606,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.gdn.y","from":605,"to":610,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.dense_ffn.xn","from":606,"to":607,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.dense_ffn.gu","from":607,"to":608,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l36.dense_ffn.act","from":608,"to":609,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l36.dense_ffn.d","from":609,"to":610,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.dense_ffn.y","from":610,"to":611,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l36.dense_ffn.y","from":610,"to":621,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.gdn.xn","from":611,"to":612,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.gdn.xn","from":611,"to":613,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.gdn.qkvz","from":612,"to":615,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l37.gdn.qkvz","from":612,"to":619,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l37.gdn.ba","from":613,"to":614,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l37.gdn.decay","from":614,"to":618,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l37.gdn.beta","from":614,"to":618,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l37.gdn.conv","from":615,"to":616,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l37.gdn.conv","from":615,"to":617,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l37.gdn.qkv","from":617,"to":618,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l37.gdn.core","from":618,"to":619,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l37.gdn.gated","from":619,"to":620,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l37.gdn.o","from":620,"to":621,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.gdn.y","from":621,"to":622,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.gdn.y","from":621,"to":626,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.dense_ffn.xn","from":622,"to":623,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.dense_ffn.gu","from":623,"to":624,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l37.dense_ffn.act","from":624,"to":625,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l37.dense_ffn.d","from":625,"to":626,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.dense_ffn.y","from":626,"to":627,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l37.dense_ffn.y","from":626,"to":637,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.gdn.xn","from":627,"to":628,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.gdn.xn","from":627,"to":629,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.gdn.qkvz","from":628,"to":631,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l38.gdn.qkvz","from":628,"to":635,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l38.gdn.ba","from":629,"to":630,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l38.gdn.decay","from":630,"to":634,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l38.gdn.beta","from":630,"to":634,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l38.gdn.conv","from":631,"to":632,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l38.gdn.conv","from":631,"to":633,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l38.gdn.qkv","from":633,"to":634,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l38.gdn.core","from":634,"to":635,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l38.gdn.gated","from":635,"to":636,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l38.gdn.o","from":636,"to":637,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.gdn.y","from":637,"to":638,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.gdn.y","from":637,"to":642,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.dense_ffn.xn","from":638,"to":639,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.dense_ffn.gu","from":639,"to":640,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l38.dense_ffn.act","from":640,"to":641,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l38.dense_ffn.d","from":641,"to":642,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.dense_ffn.y","from":642,"to":643,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l38.dense_ffn.y","from":642,"to":655,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.attn.xn","from":643,"to":644,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.attn.xn","from":643,"to":646,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.attn.xn","from":643,"to":647,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.attn.qg","from":644,"to":645,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l39.attn.q","from":645,"to":648,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l39.attn.gate","from":645,"to":653,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l39.attn.k","from":646,"to":649,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l39.attn.v","from":647,"to":651,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l39.attn.v","from":647,"to":652,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l39.attn.qn","from":648,"to":650,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l39.attn.kn","from":649,"to":650,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l39.attn.qr","from":650,"to":652,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l39.attn.kr","from":650,"to":651,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l39.attn.kr","from":650,"to":652,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l39.attn.a","from":652,"to":653,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l39.attn.ag","from":653,"to":654,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l39.attn.o","from":654,"to":655,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.attn.y","from":655,"to":656,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.attn.y","from":655,"to":660,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.dense_ffn.xn","from":656,"to":657,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.dense_ffn.gu","from":657,"to":658,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l39.dense_ffn.act","from":658,"to":659,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l39.dense_ffn.d","from":659,"to":660,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.dense_ffn.y","from":660,"to":661,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l39.dense_ffn.y","from":660,"to":671,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.gdn.xn","from":661,"to":662,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.gdn.xn","from":661,"to":663,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.gdn.qkvz","from":662,"to":665,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l40.gdn.qkvz","from":662,"to":669,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l40.gdn.ba","from":663,"to":664,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l40.gdn.decay","from":664,"to":668,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l40.gdn.beta","from":664,"to":668,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l40.gdn.conv","from":665,"to":666,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l40.gdn.conv","from":665,"to":667,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l40.gdn.qkv","from":667,"to":668,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l40.gdn.core","from":668,"to":669,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l40.gdn.gated","from":669,"to":670,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l40.gdn.o","from":670,"to":671,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.gdn.y","from":671,"to":672,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.gdn.y","from":671,"to":676,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.dense_ffn.xn","from":672,"to":673,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.dense_ffn.gu","from":673,"to":674,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l40.dense_ffn.act","from":674,"to":675,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l40.dense_ffn.d","from":675,"to":676,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.dense_ffn.y","from":676,"to":677,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l40.dense_ffn.y","from":676,"to":687,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.gdn.xn","from":677,"to":678,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.gdn.xn","from":677,"to":679,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.gdn.qkvz","from":678,"to":681,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l41.gdn.qkvz","from":678,"to":685,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l41.gdn.ba","from":679,"to":680,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l41.gdn.decay","from":680,"to":684,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l41.gdn.beta","from":680,"to":684,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l41.gdn.conv","from":681,"to":682,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l41.gdn.conv","from":681,"to":683,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l41.gdn.qkv","from":683,"to":684,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l41.gdn.core","from":684,"to":685,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l41.gdn.gated","from":685,"to":686,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l41.gdn.o","from":686,"to":687,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.gdn.y","from":687,"to":688,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.gdn.y","from":687,"to":692,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.dense_ffn.xn","from":688,"to":689,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.dense_ffn.gu","from":689,"to":690,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l41.dense_ffn.act","from":690,"to":691,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l41.dense_ffn.d","from":691,"to":692,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.dense_ffn.y","from":692,"to":693,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l41.dense_ffn.y","from":692,"to":703,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.gdn.xn","from":693,"to":694,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.gdn.xn","from":693,"to":695,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.gdn.qkvz","from":694,"to":697,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l42.gdn.qkvz","from":694,"to":701,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l42.gdn.ba","from":695,"to":696,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l42.gdn.decay","from":696,"to":700,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l42.gdn.beta","from":696,"to":700,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l42.gdn.conv","from":697,"to":698,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l42.gdn.conv","from":697,"to":699,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l42.gdn.qkv","from":699,"to":700,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l42.gdn.core","from":700,"to":701,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l42.gdn.gated","from":701,"to":702,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l42.gdn.o","from":702,"to":703,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.gdn.y","from":703,"to":704,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.gdn.y","from":703,"to":708,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.dense_ffn.xn","from":704,"to":705,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.dense_ffn.gu","from":705,"to":706,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l42.dense_ffn.act","from":706,"to":707,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l42.dense_ffn.d","from":707,"to":708,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.dense_ffn.y","from":708,"to":709,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l42.dense_ffn.y","from":708,"to":721,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.attn.xn","from":709,"to":710,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.attn.xn","from":709,"to":712,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.attn.xn","from":709,"to":713,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.attn.qg","from":710,"to":711,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l43.attn.q","from":711,"to":714,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l43.attn.gate","from":711,"to":719,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l43.attn.k","from":712,"to":715,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l43.attn.v","from":713,"to":717,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l43.attn.v","from":713,"to":718,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l43.attn.qn","from":714,"to":716,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l43.attn.kn","from":715,"to":716,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l43.attn.qr","from":716,"to":718,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l43.attn.kr","from":716,"to":717,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l43.attn.kr","from":716,"to":718,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l43.attn.a","from":718,"to":719,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l43.attn.ag","from":719,"to":720,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l43.attn.o","from":720,"to":721,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.attn.y","from":721,"to":722,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.attn.y","from":721,"to":726,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.dense_ffn.xn","from":722,"to":723,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.dense_ffn.gu","from":723,"to":724,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l43.dense_ffn.act","from":724,"to":725,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l43.dense_ffn.d","from":725,"to":726,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.dense_ffn.y","from":726,"to":727,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l43.dense_ffn.y","from":726,"to":737,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.gdn.xn","from":727,"to":728,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.gdn.xn","from":727,"to":729,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.gdn.qkvz","from":728,"to":731,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l44.gdn.qkvz","from":728,"to":735,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l44.gdn.ba","from":729,"to":730,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l44.gdn.decay","from":730,"to":734,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l44.gdn.beta","from":730,"to":734,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l44.gdn.conv","from":731,"to":732,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l44.gdn.conv","from":731,"to":733,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l44.gdn.qkv","from":733,"to":734,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l44.gdn.core","from":734,"to":735,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l44.gdn.gated","from":735,"to":736,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l44.gdn.o","from":736,"to":737,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.gdn.y","from":737,"to":738,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.gdn.y","from":737,"to":742,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.dense_ffn.xn","from":738,"to":739,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.dense_ffn.gu","from":739,"to":740,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l44.dense_ffn.act","from":740,"to":741,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l44.dense_ffn.d","from":741,"to":742,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.dense_ffn.y","from":742,"to":743,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l44.dense_ffn.y","from":742,"to":753,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.gdn.xn","from":743,"to":744,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.gdn.xn","from":743,"to":745,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.gdn.qkvz","from":744,"to":747,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l45.gdn.qkvz","from":744,"to":751,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l45.gdn.ba","from":745,"to":746,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l45.gdn.decay","from":746,"to":750,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l45.gdn.beta","from":746,"to":750,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l45.gdn.conv","from":747,"to":748,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l45.gdn.conv","from":747,"to":749,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l45.gdn.qkv","from":749,"to":750,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l45.gdn.core","from":750,"to":751,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l45.gdn.gated","from":751,"to":752,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l45.gdn.o","from":752,"to":753,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.gdn.y","from":753,"to":754,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.gdn.y","from":753,"to":758,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.dense_ffn.xn","from":754,"to":755,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.dense_ffn.gu","from":755,"to":756,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l45.dense_ffn.act","from":756,"to":757,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l45.dense_ffn.d","from":757,"to":758,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.dense_ffn.y","from":758,"to":759,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l45.dense_ffn.y","from":758,"to":769,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.gdn.xn","from":759,"to":760,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.gdn.xn","from":759,"to":761,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.gdn.qkvz","from":760,"to":763,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l46.gdn.qkvz","from":760,"to":767,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l46.gdn.ba","from":761,"to":762,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l46.gdn.decay","from":762,"to":766,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l46.gdn.beta","from":762,"to":766,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l46.gdn.conv","from":763,"to":764,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l46.gdn.conv","from":763,"to":765,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l46.gdn.qkv","from":765,"to":766,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l46.gdn.core","from":766,"to":767,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l46.gdn.gated","from":767,"to":768,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l46.gdn.o","from":768,"to":769,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.gdn.y","from":769,"to":770,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.gdn.y","from":769,"to":774,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.dense_ffn.xn","from":770,"to":771,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.dense_ffn.gu","from":771,"to":772,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l46.dense_ffn.act","from":772,"to":773,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l46.dense_ffn.d","from":773,"to":774,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.dense_ffn.y","from":774,"to":775,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l46.dense_ffn.y","from":774,"to":787,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.attn.xn","from":775,"to":776,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.attn.xn","from":775,"to":778,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.attn.xn","from":775,"to":779,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.attn.qg","from":776,"to":777,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l47.attn.q","from":777,"to":780,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l47.attn.gate","from":777,"to":785,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l47.attn.k","from":778,"to":781,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l47.attn.v","from":779,"to":783,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l47.attn.v","from":779,"to":784,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l47.attn.qn","from":780,"to":782,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l47.attn.kn","from":781,"to":782,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l47.attn.qr","from":782,"to":784,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l47.attn.kr","from":782,"to":783,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l47.attn.kr","from":782,"to":784,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l47.attn.a","from":784,"to":785,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l47.attn.ag","from":785,"to":786,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l47.attn.o","from":786,"to":787,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.attn.y","from":787,"to":788,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.attn.y","from":787,"to":792,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.dense_ffn.xn","from":788,"to":789,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.dense_ffn.gu","from":789,"to":790,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l47.dense_ffn.act","from":790,"to":791,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l47.dense_ffn.d","from":791,"to":792,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.dense_ffn.y","from":792,"to":793,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l47.dense_ffn.y","from":792,"to":803,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.gdn.xn","from":793,"to":794,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.gdn.xn","from":793,"to":795,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.gdn.qkvz","from":794,"to":797,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l48.gdn.qkvz","from":794,"to":801,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l48.gdn.ba","from":795,"to":796,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l48.gdn.decay","from":796,"to":800,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l48.gdn.beta","from":796,"to":800,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l48.gdn.conv","from":797,"to":798,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l48.gdn.conv","from":797,"to":799,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l48.gdn.qkv","from":799,"to":800,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l48.gdn.core","from":800,"to":801,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l48.gdn.gated","from":801,"to":802,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l48.gdn.o","from":802,"to":803,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.gdn.y","from":803,"to":804,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.gdn.y","from":803,"to":808,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.dense_ffn.xn","from":804,"to":805,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.dense_ffn.gu","from":805,"to":806,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l48.dense_ffn.act","from":806,"to":807,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l48.dense_ffn.d","from":807,"to":808,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.dense_ffn.y","from":808,"to":809,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l48.dense_ffn.y","from":808,"to":819,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.gdn.xn","from":809,"to":810,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.gdn.xn","from":809,"to":811,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.gdn.qkvz","from":810,"to":813,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l49.gdn.qkvz","from":810,"to":817,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l49.gdn.ba","from":811,"to":812,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l49.gdn.decay","from":812,"to":816,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l49.gdn.beta","from":812,"to":816,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l49.gdn.conv","from":813,"to":814,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l49.gdn.conv","from":813,"to":815,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l49.gdn.qkv","from":815,"to":816,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l49.gdn.core","from":816,"to":817,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l49.gdn.gated","from":817,"to":818,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l49.gdn.o","from":818,"to":819,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.gdn.y","from":819,"to":820,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.gdn.y","from":819,"to":824,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.dense_ffn.xn","from":820,"to":821,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.dense_ffn.gu","from":821,"to":822,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l49.dense_ffn.act","from":822,"to":823,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l49.dense_ffn.d","from":823,"to":824,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.dense_ffn.y","from":824,"to":825,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l49.dense_ffn.y","from":824,"to":835,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.gdn.xn","from":825,"to":826,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.gdn.xn","from":825,"to":827,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.gdn.qkvz","from":826,"to":829,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l50.gdn.qkvz","from":826,"to":833,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l50.gdn.ba","from":827,"to":828,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l50.gdn.decay","from":828,"to":832,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l50.gdn.beta","from":828,"to":832,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l50.gdn.conv","from":829,"to":830,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l50.gdn.conv","from":829,"to":831,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l50.gdn.qkv","from":831,"to":832,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l50.gdn.core","from":832,"to":833,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l50.gdn.gated","from":833,"to":834,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l50.gdn.o","from":834,"to":835,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.gdn.y","from":835,"to":836,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.gdn.y","from":835,"to":840,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.dense_ffn.xn","from":836,"to":837,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.dense_ffn.gu","from":837,"to":838,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l50.dense_ffn.act","from":838,"to":839,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l50.dense_ffn.d","from":839,"to":840,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.dense_ffn.y","from":840,"to":841,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l50.dense_ffn.y","from":840,"to":853,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.attn.xn","from":841,"to":842,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.attn.xn","from":841,"to":844,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.attn.xn","from":841,"to":845,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.attn.qg","from":842,"to":843,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l51.attn.q","from":843,"to":846,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l51.attn.gate","from":843,"to":851,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l51.attn.k","from":844,"to":847,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l51.attn.v","from":845,"to":849,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l51.attn.v","from":845,"to":850,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l51.attn.qn","from":846,"to":848,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l51.attn.kn","from":847,"to":848,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l51.attn.qr","from":848,"to":850,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l51.attn.kr","from":848,"to":849,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l51.attn.kr","from":848,"to":850,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l51.attn.a","from":850,"to":851,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l51.attn.ag","from":851,"to":852,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l51.attn.o","from":852,"to":853,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.attn.y","from":853,"to":854,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.attn.y","from":853,"to":858,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.dense_ffn.xn","from":854,"to":855,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.dense_ffn.gu","from":855,"to":856,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l51.dense_ffn.act","from":856,"to":857,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l51.dense_ffn.d","from":857,"to":858,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.dense_ffn.y","from":858,"to":859,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l51.dense_ffn.y","from":858,"to":869,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.gdn.xn","from":859,"to":860,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.gdn.xn","from":859,"to":861,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.gdn.qkvz","from":860,"to":863,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l52.gdn.qkvz","from":860,"to":867,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l52.gdn.ba","from":861,"to":862,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l52.gdn.decay","from":862,"to":866,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l52.gdn.beta","from":862,"to":866,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l52.gdn.conv","from":863,"to":864,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l52.gdn.conv","from":863,"to":865,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l52.gdn.qkv","from":865,"to":866,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l52.gdn.core","from":866,"to":867,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l52.gdn.gated","from":867,"to":868,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l52.gdn.o","from":868,"to":869,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.gdn.y","from":869,"to":870,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.gdn.y","from":869,"to":874,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.dense_ffn.xn","from":870,"to":871,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.dense_ffn.gu","from":871,"to":872,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l52.dense_ffn.act","from":872,"to":873,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l52.dense_ffn.d","from":873,"to":874,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.dense_ffn.y","from":874,"to":875,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l52.dense_ffn.y","from":874,"to":885,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.gdn.xn","from":875,"to":876,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.gdn.xn","from":875,"to":877,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.gdn.qkvz","from":876,"to":879,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l53.gdn.qkvz","from":876,"to":883,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l53.gdn.ba","from":877,"to":878,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l53.gdn.decay","from":878,"to":882,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l53.gdn.beta","from":878,"to":882,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l53.gdn.conv","from":879,"to":880,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l53.gdn.conv","from":879,"to":881,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l53.gdn.qkv","from":881,"to":882,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l53.gdn.core","from":882,"to":883,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l53.gdn.gated","from":883,"to":884,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l53.gdn.o","from":884,"to":885,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.gdn.y","from":885,"to":886,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.gdn.y","from":885,"to":890,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.dense_ffn.xn","from":886,"to":887,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.dense_ffn.gu","from":887,"to":888,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l53.dense_ffn.act","from":888,"to":889,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l53.dense_ffn.d","from":889,"to":890,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.dense_ffn.y","from":890,"to":891,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l53.dense_ffn.y","from":890,"to":901,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.gdn.xn","from":891,"to":892,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.gdn.xn","from":891,"to":893,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.gdn.qkvz","from":892,"to":895,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l54.gdn.qkvz","from":892,"to":899,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l54.gdn.ba","from":893,"to":894,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l54.gdn.decay","from":894,"to":898,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l54.gdn.beta","from":894,"to":898,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l54.gdn.conv","from":895,"to":896,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l54.gdn.conv","from":895,"to":897,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l54.gdn.qkv","from":897,"to":898,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l54.gdn.core","from":898,"to":899,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l54.gdn.gated","from":899,"to":900,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l54.gdn.o","from":900,"to":901,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.gdn.y","from":901,"to":902,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.gdn.y","from":901,"to":906,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.dense_ffn.xn","from":902,"to":903,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.dense_ffn.gu","from":903,"to":904,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l54.dense_ffn.act","from":904,"to":905,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l54.dense_ffn.d","from":905,"to":906,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.dense_ffn.y","from":906,"to":907,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l54.dense_ffn.y","from":906,"to":919,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.attn.xn","from":907,"to":908,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.attn.xn","from":907,"to":910,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.attn.xn","from":907,"to":911,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.attn.qg","from":908,"to":909,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l55.attn.q","from":909,"to":912,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l55.attn.gate","from":909,"to":917,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l55.attn.k","from":910,"to":913,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l55.attn.v","from":911,"to":915,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l55.attn.v","from":911,"to":916,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l55.attn.qn","from":912,"to":914,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l55.attn.kn","from":913,"to":914,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l55.attn.qr","from":914,"to":916,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l55.attn.kr","from":914,"to":915,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l55.attn.kr","from":914,"to":916,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l55.attn.a","from":916,"to":917,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l55.attn.ag","from":917,"to":918,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l55.attn.o","from":918,"to":919,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.attn.y","from":919,"to":920,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.attn.y","from":919,"to":924,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.dense_ffn.xn","from":920,"to":921,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.dense_ffn.gu","from":921,"to":922,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l55.dense_ffn.act","from":922,"to":923,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l55.dense_ffn.d","from":923,"to":924,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.dense_ffn.y","from":924,"to":925,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l55.dense_ffn.y","from":924,"to":935,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.gdn.xn","from":925,"to":926,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.gdn.xn","from":925,"to":927,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.gdn.qkvz","from":926,"to":929,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l56.gdn.qkvz","from":926,"to":933,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l56.gdn.ba","from":927,"to":928,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l56.gdn.decay","from":928,"to":932,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l56.gdn.beta","from":928,"to":932,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l56.gdn.conv","from":929,"to":930,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l56.gdn.conv","from":929,"to":931,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l56.gdn.qkv","from":931,"to":932,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l56.gdn.core","from":932,"to":933,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l56.gdn.gated","from":933,"to":934,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l56.gdn.o","from":934,"to":935,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.gdn.y","from":935,"to":936,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.gdn.y","from":935,"to":940,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.dense_ffn.xn","from":936,"to":937,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.dense_ffn.gu","from":937,"to":938,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l56.dense_ffn.act","from":938,"to":939,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l56.dense_ffn.d","from":939,"to":940,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.dense_ffn.y","from":940,"to":941,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l56.dense_ffn.y","from":940,"to":951,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.gdn.xn","from":941,"to":942,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.gdn.xn","from":941,"to":943,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.gdn.qkvz","from":942,"to":945,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l57.gdn.qkvz","from":942,"to":949,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l57.gdn.ba","from":943,"to":944,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l57.gdn.decay","from":944,"to":948,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l57.gdn.beta","from":944,"to":948,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l57.gdn.conv","from":945,"to":946,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l57.gdn.conv","from":945,"to":947,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l57.gdn.qkv","from":947,"to":948,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l57.gdn.core","from":948,"to":949,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l57.gdn.gated","from":949,"to":950,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l57.gdn.o","from":950,"to":951,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.gdn.y","from":951,"to":952,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.gdn.y","from":951,"to":956,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.dense_ffn.xn","from":952,"to":953,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.dense_ffn.gu","from":953,"to":954,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l57.dense_ffn.act","from":954,"to":955,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l57.dense_ffn.d","from":955,"to":956,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.dense_ffn.y","from":956,"to":957,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l57.dense_ffn.y","from":956,"to":967,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.gdn.xn","from":957,"to":958,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.gdn.xn","from":957,"to":959,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.gdn.qkvz","from":958,"to":961,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l58.gdn.qkvz","from":958,"to":965,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l58.gdn.ba","from":959,"to":960,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l58.gdn.decay","from":960,"to":964,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l58.gdn.beta","from":960,"to":964,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l58.gdn.conv","from":961,"to":962,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l58.gdn.conv","from":961,"to":963,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l58.gdn.qkv","from":963,"to":964,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l58.gdn.core","from":964,"to":965,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l58.gdn.gated","from":965,"to":966,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l58.gdn.o","from":966,"to":967,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.gdn.y","from":967,"to":968,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.gdn.y","from":967,"to":972,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.dense_ffn.xn","from":968,"to":969,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.dense_ffn.gu","from":969,"to":970,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l58.dense_ffn.act","from":970,"to":971,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l58.dense_ffn.d","from":971,"to":972,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.dense_ffn.y","from":972,"to":973,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l58.dense_ffn.y","from":972,"to":985,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.attn.xn","from":973,"to":974,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.attn.xn","from":973,"to":976,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.attn.xn","from":973,"to":977,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.attn.qg","from":974,"to":975,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l59.attn.q","from":975,"to":978,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l59.attn.gate","from":975,"to":983,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l59.attn.k","from":976,"to":979,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l59.attn.v","from":977,"to":981,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l59.attn.v","from":977,"to":982,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l59.attn.qn","from":978,"to":980,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l59.attn.kn","from":979,"to":980,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l59.attn.qr","from":980,"to":982,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l59.attn.kr","from":980,"to":981,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l59.attn.kr","from":980,"to":982,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l59.attn.a","from":982,"to":983,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l59.attn.ag","from":983,"to":984,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l59.attn.o","from":984,"to":985,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.attn.y","from":985,"to":986,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.attn.y","from":985,"to":990,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.dense_ffn.xn","from":986,"to":987,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.dense_ffn.gu","from":987,"to":988,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l59.dense_ffn.act","from":988,"to":989,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l59.dense_ffn.d","from":989,"to":990,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.dense_ffn.y","from":990,"to":991,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l59.dense_ffn.y","from":990,"to":1001,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.gdn.xn","from":991,"to":992,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.gdn.xn","from":991,"to":993,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.gdn.qkvz","from":992,"to":995,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l60.gdn.qkvz","from":992,"to":999,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l60.gdn.ba","from":993,"to":994,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l60.gdn.decay","from":994,"to":998,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l60.gdn.beta","from":994,"to":998,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l60.gdn.conv","from":995,"to":996,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l60.gdn.conv","from":995,"to":997,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l60.gdn.qkv","from":997,"to":998,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l60.gdn.core","from":998,"to":999,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l60.gdn.gated","from":999,"to":1000,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l60.gdn.o","from":1000,"to":1001,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.gdn.y","from":1001,"to":1002,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.gdn.y","from":1001,"to":1006,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.dense_ffn.xn","from":1002,"to":1003,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.dense_ffn.gu","from":1003,"to":1004,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l60.dense_ffn.act","from":1004,"to":1005,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l60.dense_ffn.d","from":1005,"to":1006,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.dense_ffn.y","from":1006,"to":1007,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l60.dense_ffn.y","from":1006,"to":1017,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.gdn.xn","from":1007,"to":1008,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.gdn.xn","from":1007,"to":1009,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.gdn.qkvz","from":1008,"to":1011,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l61.gdn.qkvz","from":1008,"to":1015,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l61.gdn.ba","from":1009,"to":1010,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l61.gdn.decay","from":1010,"to":1014,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l61.gdn.beta","from":1010,"to":1014,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l61.gdn.conv","from":1011,"to":1012,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l61.gdn.conv","from":1011,"to":1013,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l61.gdn.qkv","from":1013,"to":1014,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l61.gdn.core","from":1014,"to":1015,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l61.gdn.gated","from":1015,"to":1016,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l61.gdn.o","from":1016,"to":1017,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.gdn.y","from":1017,"to":1018,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.gdn.y","from":1017,"to":1022,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.dense_ffn.xn","from":1018,"to":1019,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.dense_ffn.gu","from":1019,"to":1020,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l61.dense_ffn.act","from":1020,"to":1021,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l61.dense_ffn.d","from":1021,"to":1022,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.dense_ffn.y","from":1022,"to":1023,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l61.dense_ffn.y","from":1022,"to":1033,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.gdn.xn","from":1023,"to":1024,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.gdn.xn","from":1023,"to":1025,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.gdn.qkvz","from":1024,"to":1027,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l62.gdn.qkvz","from":1024,"to":1031,"format":"bf16","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim*2","fused":false},{"id":"l62.gdn.ba","from":1025,"to":1026,"format":"f32","rows":"n","dim":"lin_v_heads*2","fused":true},{"id":"l62.gdn.decay","from":1026,"to":1030,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l62.gdn.beta","from":1026,"to":1030,"format":"f32","rows":"n","dim":"lin_v_heads","fused":false},{"id":"l62.gdn.conv","from":1027,"to":1028,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l62.gdn.conv","from":1027,"to":1029,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":true},{"id":"l62.gdn.qkv","from":1029,"to":1030,"format":"f32","rows":"n","dim":"lin_k_heads*lin_k_dim*2+lin_v_heads*lin_v_dim","fused":false},{"id":"l62.gdn.core","from":1030,"to":1031,"format":"f32","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l62.gdn.gated","from":1031,"to":1032,"format":"bf16","rows":"n","dim":"lin_v_heads*lin_v_dim","fused":false},{"id":"l62.gdn.o","from":1032,"to":1033,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.gdn.y","from":1033,"to":1034,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.gdn.y","from":1033,"to":1038,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.dense_ffn.xn","from":1034,"to":1035,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.dense_ffn.gu","from":1035,"to":1036,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l62.dense_ffn.act","from":1036,"to":1037,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l62.dense_ffn.d","from":1037,"to":1038,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.dense_ffn.y","from":1038,"to":1039,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l62.dense_ffn.y","from":1038,"to":1051,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.attn.xn","from":1039,"to":1040,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.attn.xn","from":1039,"to":1042,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.attn.xn","from":1039,"to":1043,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.attn.qg","from":1040,"to":1041,"format":"bf16","rows":"n","dim":"q_heads*head_dim*2","fused":true},{"id":"l63.attn.q","from":1041,"to":1044,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l63.attn.gate","from":1041,"to":1049,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l63.attn.k","from":1042,"to":1045,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l63.attn.v","from":1043,"to":1047,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l63.attn.v","from":1043,"to":1048,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l63.attn.qn","from":1044,"to":1046,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l63.attn.kn","from":1045,"to":1046,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l63.attn.qr","from":1046,"to":1048,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l63.attn.kr","from":1046,"to":1047,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l63.attn.kr","from":1046,"to":1048,"format":"bf16","rows":"n","dim":"kv_heads*head_dim","fused":false},{"id":"l63.attn.a","from":1048,"to":1049,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l63.attn.ag","from":1049,"to":1050,"format":"bf16","rows":"n","dim":"q_heads*head_dim","fused":false},{"id":"l63.attn.o","from":1050,"to":1051,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.attn.y","from":1051,"to":1052,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.attn.y","from":1051,"to":1056,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.dense_ffn.xn","from":1052,"to":1053,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.dense_ffn.gu","from":1053,"to":1054,"format":"bf16","rows":"n","dim":"inter*2","fused":false},{"id":"l63.dense_ffn.act","from":1054,"to":1055,"format":"bf16","rows":"n","dim":"inter","fused":false},{"id":"l63.dense_ffn.d","from":1055,"to":1056,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"l63.dense_ffn.y","from":1056,"to":1057,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"head.xn","from":1057,"to":1058,"format":"bf16","rows":"n","dim":"hidden","fused":false},{"id":"head.logits","from":1058,"to":1059,"format":"bf16","rows":"n","dim":"vocab","fused":false}],"groups":[{"id":0,"rule":"embed_row_copy","kernels":[],"where":"host","numerics":"reference","compute":null,"nodes":[0]},{"id":1,"rule":"input_norm_residual","kernels":["norm::rms_norm_residual"],"where":"device","numerics":"reference","compute":"memory","nodes":[1]},{"id":2,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[2]},{"id":3,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[3,4]},{"id":4,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[5,6,7]},{"id":5,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[8]},{"id":6,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[9]},{"id":7,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[10]},{"id":8,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[11,12]},{"id":9,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[13]},{"id":10,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[14]},{"id":11,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[15]},{"id":12,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[16,17]},{"id":13,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[18]},{"id":14,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[19,20]},{"id":15,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[21,22,23]},{"id":16,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[24]},{"id":17,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[25]},{"id":18,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[26]},{"id":19,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[27,28]},{"id":20,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[29]},{"id":21,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[30]},{"id":22,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[31]},{"id":23,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[32,33]},{"id":24,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[34]},{"id":25,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[35,36]},{"id":26,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[37,38,39]},{"id":27,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[40]},{"id":28,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[41]},{"id":29,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[42]},{"id":30,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[43,44]},{"id":31,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[45]},{"id":32,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[46]},{"id":33,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[47]},{"id":34,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[48,49]},{"id":35,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[50,51]},{"id":36,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[52,53]},{"id":37,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[54]},{"id":38,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[55]},{"id":39,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[56]},{"id":40,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[57]},{"id":41,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[58]},{"id":42,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[59]},{"id":43,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[60]},{"id":44,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[61,62]},{"id":45,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[63]},{"id":46,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[64]},{"id":47,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[65]},{"id":48,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[66,67]},{"id":49,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[68]},{"id":50,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[69,70]},{"id":51,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[71,72,73]},{"id":52,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[74]},{"id":53,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[75]},{"id":54,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[76]},{"id":55,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[77,78]},{"id":56,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[79]},{"id":57,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[80]},{"id":58,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[81]},{"id":59,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[82,83]},{"id":60,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[84]},{"id":61,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[85,86]},{"id":62,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[87,88,89]},{"id":63,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[90]},{"id":64,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[91]},{"id":65,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[92]},{"id":66,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[93,94]},{"id":67,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[95]},{"id":68,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[96]},{"id":69,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[97]},{"id":70,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[98,99]},{"id":71,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[100]},{"id":72,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[101,102]},{"id":73,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[103,104,105]},{"id":74,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[106]},{"id":75,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[107]},{"id":76,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[108]},{"id":77,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[109,110]},{"id":78,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[111]},{"id":79,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[112]},{"id":80,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[113]},{"id":81,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[114,115]},{"id":82,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[116,117]},{"id":83,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[118,119]},{"id":84,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[120]},{"id":85,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[121]},{"id":86,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[122]},{"id":87,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[123]},{"id":88,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[124]},{"id":89,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[125]},{"id":90,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[126]},{"id":91,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[127,128]},{"id":92,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[129]},{"id":93,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[130]},{"id":94,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[131]},{"id":95,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[132,133]},{"id":96,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[134]},{"id":97,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[135,136]},{"id":98,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[137,138,139]},{"id":99,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[140]},{"id":100,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[141]},{"id":101,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[142]},{"id":102,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[143,144]},{"id":103,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[145]},{"id":104,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[146]},{"id":105,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[147]},{"id":106,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[148,149]},{"id":107,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[150]},{"id":108,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[151,152]},{"id":109,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[153,154,155]},{"id":110,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[156]},{"id":111,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[157]},{"id":112,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[158]},{"id":113,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[159,160]},{"id":114,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[161]},{"id":115,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[162]},{"id":116,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[163]},{"id":117,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[164,165]},{"id":118,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[166]},{"id":119,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[167,168]},{"id":120,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[169,170,171]},{"id":121,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[172]},{"id":122,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[173]},{"id":123,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[174]},{"id":124,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[175,176]},{"id":125,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[177]},{"id":126,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[178]},{"id":127,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[179]},{"id":128,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[180,181]},{"id":129,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[182,183]},{"id":130,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[184,185]},{"id":131,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[186]},{"id":132,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[187]},{"id":133,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[188]},{"id":134,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[189]},{"id":135,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[190]},{"id":136,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[191]},{"id":137,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[192]},{"id":138,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[193,194]},{"id":139,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[195]},{"id":140,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[196]},{"id":141,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[197]},{"id":142,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[198,199]},{"id":143,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[200]},{"id":144,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[201,202]},{"id":145,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[203,204,205]},{"id":146,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[206]},{"id":147,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[207]},{"id":148,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[208]},{"id":149,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[209,210]},{"id":150,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[211]},{"id":151,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[212]},{"id":152,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[213]},{"id":153,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[214,215]},{"id":154,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[216]},{"id":155,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[217,218]},{"id":156,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[219,220,221]},{"id":157,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[222]},{"id":158,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[223]},{"id":159,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[224]},{"id":160,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[225,226]},{"id":161,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[227]},{"id":162,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[228]},{"id":163,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[229]},{"id":164,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[230,231]},{"id":165,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[232]},{"id":166,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[233,234]},{"id":167,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[235,236,237]},{"id":168,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[238]},{"id":169,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[239]},{"id":170,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[240]},{"id":171,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[241,242]},{"id":172,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[243]},{"id":173,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[244]},{"id":174,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[245]},{"id":175,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[246,247]},{"id":176,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[248,249]},{"id":177,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[250,251]},{"id":178,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[252]},{"id":179,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[253]},{"id":180,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[254]},{"id":181,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[255]},{"id":182,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[256]},{"id":183,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[257]},{"id":184,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[258]},{"id":185,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[259,260]},{"id":186,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[261]},{"id":187,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[262]},{"id":188,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[263]},{"id":189,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[264,265]},{"id":190,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[266]},{"id":191,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[267,268]},{"id":192,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[269,270,271]},{"id":193,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[272]},{"id":194,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[273]},{"id":195,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[274]},{"id":196,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[275,276]},{"id":197,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[277]},{"id":198,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[278]},{"id":199,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[279]},{"id":200,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[280,281]},{"id":201,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[282]},{"id":202,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[283,284]},{"id":203,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[285,286,287]},{"id":204,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[288]},{"id":205,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[289]},{"id":206,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[290]},{"id":207,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[291,292]},{"id":208,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[293]},{"id":209,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[294]},{"id":210,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[295]},{"id":211,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[296,297]},{"id":212,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[298]},{"id":213,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[299,300]},{"id":214,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[301,302,303]},{"id":215,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[304]},{"id":216,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[305]},{"id":217,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[306]},{"id":218,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[307,308]},{"id":219,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[309]},{"id":220,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[310]},{"id":221,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[311]},{"id":222,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[312,313]},{"id":223,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[314,315]},{"id":224,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[316,317]},{"id":225,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[318]},{"id":226,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[319]},{"id":227,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[320]},{"id":228,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[321]},{"id":229,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[322]},{"id":230,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[323]},{"id":231,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[324]},{"id":232,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[325,326]},{"id":233,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[327]},{"id":234,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[328]},{"id":235,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[329]},{"id":236,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[330,331]},{"id":237,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[332]},{"id":238,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[333,334]},{"id":239,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[335,336,337]},{"id":240,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[338]},{"id":241,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[339]},{"id":242,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[340]},{"id":243,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[341,342]},{"id":244,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[343]},{"id":245,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[344]},{"id":246,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[345]},{"id":247,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[346,347]},{"id":248,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[348]},{"id":249,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[349,350]},{"id":250,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[351,352,353]},{"id":251,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[354]},{"id":252,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[355]},{"id":253,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[356]},{"id":254,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[357,358]},{"id":255,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[359]},{"id":256,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[360]},{"id":257,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[361]},{"id":258,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[362,363]},{"id":259,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[364]},{"id":260,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[365,366]},{"id":261,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[367,368,369]},{"id":262,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[370]},{"id":263,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[371]},{"id":264,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[372]},{"id":265,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[373,374]},{"id":266,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[375]},{"id":267,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[376]},{"id":268,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[377]},{"id":269,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[378,379]},{"id":270,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[380,381]},{"id":271,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[382,383]},{"id":272,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[384]},{"id":273,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[385]},{"id":274,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[386]},{"id":275,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[387]},{"id":276,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[388]},{"id":277,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[389]},{"id":278,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[390]},{"id":279,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[391,392]},{"id":280,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[393]},{"id":281,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[394]},{"id":282,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[395]},{"id":283,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[396,397]},{"id":284,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[398]},{"id":285,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[399,400]},{"id":286,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[401,402,403]},{"id":287,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[404]},{"id":288,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[405]},{"id":289,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[406]},{"id":290,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[407,408]},{"id":291,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[409]},{"id":292,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[410]},{"id":293,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[411]},{"id":294,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[412,413]},{"id":295,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[414]},{"id":296,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[415,416]},{"id":297,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[417,418,419]},{"id":298,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[420]},{"id":299,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[421]},{"id":300,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[422]},{"id":301,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[423,424]},{"id":302,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[425]},{"id":303,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[426]},{"id":304,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[427]},{"id":305,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[428,429]},{"id":306,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[430]},{"id":307,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[431,432]},{"id":308,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[433,434,435]},{"id":309,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[436]},{"id":310,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[437]},{"id":311,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[438]},{"id":312,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[439,440]},{"id":313,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[441]},{"id":314,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[442]},{"id":315,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[443]},{"id":316,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[444,445]},{"id":317,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[446,447]},{"id":318,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[448,449]},{"id":319,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[450]},{"id":320,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[451]},{"id":321,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[452]},{"id":322,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[453]},{"id":323,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[454]},{"id":324,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[455]},{"id":325,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[456]},{"id":326,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[457,458]},{"id":327,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[459]},{"id":328,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[460]},{"id":329,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[461]},{"id":330,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[462,463]},{"id":331,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[464]},{"id":332,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[465,466]},{"id":333,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[467,468,469]},{"id":334,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[470]},{"id":335,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[471]},{"id":336,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[472]},{"id":337,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[473,474]},{"id":338,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[475]},{"id":339,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[476]},{"id":340,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[477]},{"id":341,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[478,479]},{"id":342,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[480]},{"id":343,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[481,482]},{"id":344,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[483,484,485]},{"id":345,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[486]},{"id":346,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[487]},{"id":347,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[488]},{"id":348,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[489,490]},{"id":349,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[491]},{"id":350,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[492]},{"id":351,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[493]},{"id":352,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[494,495]},{"id":353,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[496]},{"id":354,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[497,498]},{"id":355,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[499,500,501]},{"id":356,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[502]},{"id":357,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[503]},{"id":358,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[504]},{"id":359,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[505,506]},{"id":360,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[507]},{"id":361,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[508]},{"id":362,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[509]},{"id":363,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[510,511]},{"id":364,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[512,513]},{"id":365,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[514,515]},{"id":366,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[516]},{"id":367,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[517]},{"id":368,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[518]},{"id":369,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[519]},{"id":370,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[520]},{"id":371,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[521]},{"id":372,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[522]},{"id":373,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[523,524]},{"id":374,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[525]},{"id":375,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[526]},{"id":376,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[527]},{"id":377,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[528,529]},{"id":378,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[530]},{"id":379,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[531,532]},{"id":380,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[533,534,535]},{"id":381,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[536]},{"id":382,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[537]},{"id":383,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[538]},{"id":384,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[539,540]},{"id":385,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[541]},{"id":386,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[542]},{"id":387,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[543]},{"id":388,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[544,545]},{"id":389,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[546]},{"id":390,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[547,548]},{"id":391,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[549,550,551]},{"id":392,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[552]},{"id":393,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[553]},{"id":394,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[554]},{"id":395,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[555,556]},{"id":396,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[557]},{"id":397,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[558]},{"id":398,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[559]},{"id":399,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[560,561]},{"id":400,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[562]},{"id":401,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[563,564]},{"id":402,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[565,566,567]},{"id":403,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[568]},{"id":404,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[569]},{"id":405,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[570]},{"id":406,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[571,572]},{"id":407,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[573]},{"id":408,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[574]},{"id":409,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[575]},{"id":410,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[576,577]},{"id":411,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[578,579]},{"id":412,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[580,581]},{"id":413,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[582]},{"id":414,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[583]},{"id":415,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[584]},{"id":416,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[585]},{"id":417,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[586]},{"id":418,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[587]},{"id":419,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[588]},{"id":420,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[589,590]},{"id":421,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[591]},{"id":422,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[592]},{"id":423,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[593]},{"id":424,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[594,595]},{"id":425,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[596]},{"id":426,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[597,598]},{"id":427,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[599,600,601]},{"id":428,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[602]},{"id":429,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[603]},{"id":430,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[604]},{"id":431,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[605,606]},{"id":432,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[607]},{"id":433,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[608]},{"id":434,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[609]},{"id":435,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[610,611]},{"id":436,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[612]},{"id":437,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[613,614]},{"id":438,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[615,616,617]},{"id":439,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[618]},{"id":440,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[619]},{"id":441,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[620]},{"id":442,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[621,622]},{"id":443,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[623]},{"id":444,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[624]},{"id":445,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[625]},{"id":446,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[626,627]},{"id":447,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[628]},{"id":448,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[629,630]},{"id":449,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[631,632,633]},{"id":450,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[634]},{"id":451,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[635]},{"id":452,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[636]},{"id":453,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[637,638]},{"id":454,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[639]},{"id":455,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[640]},{"id":456,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[641]},{"id":457,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[642,643]},{"id":458,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[644,645]},{"id":459,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[646,647]},{"id":460,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[648]},{"id":461,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[649]},{"id":462,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[650]},{"id":463,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[651]},{"id":464,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[652]},{"id":465,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[653]},{"id":466,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[654]},{"id":467,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[655,656]},{"id":468,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[657]},{"id":469,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[658]},{"id":470,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[659]},{"id":471,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[660,661]},{"id":472,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[662]},{"id":473,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[663,664]},{"id":474,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[665,666,667]},{"id":475,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[668]},{"id":476,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[669]},{"id":477,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[670]},{"id":478,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[671,672]},{"id":479,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[673]},{"id":480,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[674]},{"id":481,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[675]},{"id":482,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[676,677]},{"id":483,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[678]},{"id":484,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[679,680]},{"id":485,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[681,682,683]},{"id":486,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[684]},{"id":487,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[685]},{"id":488,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[686]},{"id":489,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[687,688]},{"id":490,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[689]},{"id":491,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[690]},{"id":492,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[691]},{"id":493,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[692,693]},{"id":494,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[694]},{"id":495,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[695,696]},{"id":496,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[697,698,699]},{"id":497,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[700]},{"id":498,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[701]},{"id":499,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[702]},{"id":500,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[703,704]},{"id":501,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[705]},{"id":502,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[706]},{"id":503,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[707]},{"id":504,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[708,709]},{"id":505,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[710,711]},{"id":506,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[712,713]},{"id":507,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[714]},{"id":508,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[715]},{"id":509,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[716]},{"id":510,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[717]},{"id":511,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[718]},{"id":512,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[719]},{"id":513,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[720]},{"id":514,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[721,722]},{"id":515,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[723]},{"id":516,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[724]},{"id":517,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[725]},{"id":518,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[726,727]},{"id":519,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[728]},{"id":520,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[729,730]},{"id":521,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[731,732,733]},{"id":522,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[734]},{"id":523,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[735]},{"id":524,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[736]},{"id":525,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[737,738]},{"id":526,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[739]},{"id":527,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[740]},{"id":528,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[741]},{"id":529,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[742,743]},{"id":530,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[744]},{"id":531,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[745,746]},{"id":532,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[747,748,749]},{"id":533,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[750]},{"id":534,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[751]},{"id":535,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[752]},{"id":536,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[753,754]},{"id":537,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[755]},{"id":538,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[756]},{"id":539,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[757]},{"id":540,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[758,759]},{"id":541,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[760]},{"id":542,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[761,762]},{"id":543,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[763,764,765]},{"id":544,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[766]},{"id":545,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[767]},{"id":546,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[768]},{"id":547,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[769,770]},{"id":548,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[771]},{"id":549,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[772]},{"id":550,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[773]},{"id":551,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[774,775]},{"id":552,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[776,777]},{"id":553,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[778,779]},{"id":554,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[780]},{"id":555,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[781]},{"id":556,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[782]},{"id":557,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[783]},{"id":558,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[784]},{"id":559,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[785]},{"id":560,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[786]},{"id":561,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[787,788]},{"id":562,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[789]},{"id":563,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[790]},{"id":564,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[791]},{"id":565,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[792,793]},{"id":566,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[794]},{"id":567,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[795,796]},{"id":568,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[797,798,799]},{"id":569,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[800]},{"id":570,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[801]},{"id":571,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[802]},{"id":572,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[803,804]},{"id":573,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[805]},{"id":574,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[806]},{"id":575,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[807]},{"id":576,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[808,809]},{"id":577,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[810]},{"id":578,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[811,812]},{"id":579,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[813,814,815]},{"id":580,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[816]},{"id":581,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[817]},{"id":582,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[818]},{"id":583,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[819,820]},{"id":584,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[821]},{"id":585,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[822]},{"id":586,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[823]},{"id":587,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[824,825]},{"id":588,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[826]},{"id":589,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[827,828]},{"id":590,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[829,830,831]},{"id":591,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[832]},{"id":592,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[833]},{"id":593,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[834]},{"id":594,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[835,836]},{"id":595,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[837]},{"id":596,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[838]},{"id":597,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[839]},{"id":598,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[840,841]},{"id":599,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[842,843]},{"id":600,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[844,845]},{"id":601,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[846]},{"id":602,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[847]},{"id":603,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[848]},{"id":604,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[849]},{"id":605,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[850]},{"id":606,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[851]},{"id":607,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[852]},{"id":608,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[853,854]},{"id":609,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[855]},{"id":610,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[856]},{"id":611,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[857]},{"id":612,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[858,859]},{"id":613,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[860]},{"id":614,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[861,862]},{"id":615,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[863,864,865]},{"id":616,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[866]},{"id":617,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[867]},{"id":618,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[868]},{"id":619,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[869,870]},{"id":620,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[871]},{"id":621,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[872]},{"id":622,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[873]},{"id":623,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[874,875]},{"id":624,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[876]},{"id":625,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[877,878]},{"id":626,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[879,880,881]},{"id":627,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[882]},{"id":628,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[883]},{"id":629,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[884]},{"id":630,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[885,886]},{"id":631,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[887]},{"id":632,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[888]},{"id":633,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[889]},{"id":634,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[890,891]},{"id":635,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[892]},{"id":636,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[893,894]},{"id":637,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[895,896,897]},{"id":638,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[898]},{"id":639,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[899]},{"id":640,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[900]},{"id":641,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[901,902]},{"id":642,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[903]},{"id":643,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[904]},{"id":644,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[905]},{"id":645,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[906,907]},{"id":646,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[908,909]},{"id":647,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[910,911]},{"id":648,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[912]},{"id":649,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[913]},{"id":650,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[914]},{"id":651,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[915]},{"id":652,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[916]},{"id":653,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[917]},{"id":654,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[918]},{"id":655,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[919,920]},{"id":656,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[921]},{"id":657,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[922]},{"id":658,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[923]},{"id":659,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[924,925]},{"id":660,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[926]},{"id":661,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[927,928]},{"id":662,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[929,930,931]},{"id":663,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[932]},{"id":664,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[933]},{"id":665,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[934]},{"id":666,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[935,936]},{"id":667,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[937]},{"id":668,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[938]},{"id":669,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[939]},{"id":670,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[940,941]},{"id":671,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[942]},{"id":672,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[943,944]},{"id":673,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[945,946,947]},{"id":674,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[948]},{"id":675,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[949]},{"id":676,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[950]},{"id":677,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[951,952]},{"id":678,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[953]},{"id":679,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[954]},{"id":680,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[955]},{"id":681,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[956,957]},{"id":682,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[958]},{"id":683,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[959,960]},{"id":684,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[961,962,963]},{"id":685,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[964]},{"id":686,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[965]},{"id":687,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[966]},{"id":688,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[967,968]},{"id":689,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[969]},{"id":690,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[970]},{"id":691,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[971]},{"id":692,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[972,973]},{"id":693,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[974,975]},{"id":694,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[976,977]},{"id":695,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[978]},{"id":696,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[979]},{"id":697,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[980]},{"id":698,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[981]},{"id":699,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[982]},{"id":700,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[983]},{"id":701,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[984]},{"id":702,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[985,986]},{"id":703,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[987]},{"id":704,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[988]},{"id":705,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[989]},{"id":706,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[990,991]},{"id":707,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[992]},{"id":708,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[993,994]},{"id":709,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[995,996,997]},{"id":710,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[998]},{"id":711,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[999]},{"id":712,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1000]},{"id":713,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[1001,1002]},{"id":714,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1003]},{"id":715,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[1004]},{"id":716,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1005]},{"id":717,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[1006,1007]},{"id":718,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1008]},{"id":719,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1009,1010]},{"id":720,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1011,1012,1013]},{"id":721,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1014]},{"id":722,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[1015]},{"id":723,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1016]},{"id":724,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[1017,1018]},{"id":725,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1019]},{"id":726,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[1020]},{"id":727,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1021]},{"id":728,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[1022,1023]},{"id":729,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1024]},{"id":730,"rule":"gdn_ba_gates_gemv_per_row","kernels":["ssm_preprocess::dense_gemv_ba_gates"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1025,1026]},{"id":731,"rule":"gdn_conv_l2_f32_per_row","kernels":["causal_conv1d::causal_conv1d_update_l2norm_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1027,1028,1029]},{"id":732,"rule":"gdn_recurrence_f32_per_row","kernels":["gated_delta_rule::gated_delta_rule_decode_f32"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1030]},{"id":733,"rule":"gdn_out_norm_f32_per_row","kernels":["norm::gated_rms_norm_f32_input"],"where":"device","numerics":"reference","compute":"memory","nodes":[1031]},{"id":734,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1032]},{"id":735,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[1033,1034]},{"id":736,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1035]},{"id":737,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[1036]},{"id":738,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1037]},{"id":739,"rule":"cross_layer_add_norm","kernels":["residual_add_rms_norm_exact::residual_add_rms_norm_exact"],"where":"device","numerics":"bit_identical(residual_add_rms_norm_exact_microtest)","compute":"memory","nodes":[1038,1039]},{"id":740,"rule":"w4a16_gemv_qg_1row","kernels":["w4a16_gemv::w4a16_gemv_qg"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1040,1041]},{"id":741,"rule":"w4a16_gemv_dual_kv_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1042,1043]},{"id":742,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[1044]},{"id":743,"rule":"qk_norm_rows","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[1045]},{"id":744,"rule":"rope_mrope_interleaved","kernels":["rope_mrope_interleaved::rope_forward_mrope_interleaved"],"where":"device","numerics":"reference","compute":"memory","nodes":[1046]},{"id":745,"rule":"kv_write_bf16","kernels":["reshape_and_cache::reshape_and_cache_flash"],"where":"device","numerics":"reference","compute":"memory","nodes":[1047]},{"id":746,"rule":"paged_attention_bf16","kernels":["paged_decode::paged_decode_attn"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1048]},{"id":747,"rule":"sigmoid_gate_mul","kernels":["residual_add::sigmoid_gate_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[1049]},{"id":748,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1050]},{"id":749,"rule":"residual_add_post_norm","kernels":["norm::residual_add_rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[1051,1052]},{"id":750,"rule":"w4a16_gemv_dual_sw_gate_up_1row","kernels":["w4a16_gemv_fused::w4a16_gemv_dual_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1053]},{"id":751,"rule":"silu_mul_split_1row","kernels":["moe_silu_mul::moe_silu_mul"],"where":"device","numerics":"reference","compute":"memory","nodes":[1054]},{"id":752,"rule":"w4a16_gemv_sw_1row","kernels":["w4a16_gemv::w4a16_gemv_sw"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1055]},{"id":753,"rule":"ffn_residual_add","kernels":["residual_add::bf16_residual_add"],"where":"device","numerics":"reference","compute":"memory","nodes":[1056]},{"id":754,"rule":"final_norm","kernels":["norm::rms_norm"],"where":"device","numerics":"reference","compute":"memory","nodes":[1057]},{"id":755,"rule":"lm_head_bf16_gemv","kernels":["gemv::dense_gemv_bf16"],"where":"device","numerics":"reference","compute":"cuda_core","nodes":[1058]},{"id":756,"rule":"argmax_host","kernels":[],"where":"host","numerics":"reference","compute":null,"nodes":[1059]}],"collapsed":{"prologue":["embed"],"layout":["gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","gdn+dense_ffn","attn+dense_ffn"],"epilogue":["head"],"templates":{"embed":{"blocks":["embed"],"nodes":[{"id":"embed.embed","op":"embed","kind":null,"block_type":"embed","precision":{"weight":null,"activation":null,"output":"bf16"},"kernel":null,"family":null,"compute":null,"where":"host","group":0,"inserted":false}],"edges":[],"ports":{"in":[],"out":[0]},"instances":[]},"gdn+dense_ffn":{"blocks":["gdn","dense_ffn"],"nodes":[{"id":"gdn.input_norm","op":"rms_norm","kind":"linear_attention","block_type":"gdn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"norm::rms_norm_residual","family":"rms_norm","compute":"memory","where":"device","group":0,"inserted":false},{"id":"gdn.qkvz","op":"linear:qkvz","kind":"linear_attention","block_type":"gdn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":1,"inserted":false},{"id":"gdn.ba","op":"linear:ba","kind":"linear_attention","block_type":"gdn","precision":{"weight":"bf16","activation":"bf16","output":"f32"},"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","compute":"cuda_core","where":"device","group":2,"inserted":false},{"id":"gdn.gates","op":"gdn_gates","kind":"linear_attention","block_type":"gdn","precision":{"weight":null,"activation":"f32","output":"f32,f32"},"kernel":"ssm_preprocess::dense_gemv_ba_gates","family":"gdn_ba_gates_gemv","compute":"cuda_core","where":"device","group":2,"inserted":false},{"id":"gdn.conv","op":"conv1d_update","kind":"linear_attention","block_type":"gdn","precision":{"weight":null,"activation":"bf16","output":"f32"},"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","compute":"cuda_core","where":"device","group":3,"inserted":false},{"id":"gdn.conv_ckpt","op":"state_snapshot","kind":"linear_attention","block_type":"gdn","precision":{"weight":null,"activation":"f32","output":null},"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","compute":"cuda_core","where":"device","group":3,"inserted":false},{"id":"gdn.l2","op":"l2_norm","kind":"linear_attention","block_type":"gdn","precision":{"weight":null,"activation":"f32","output":"f32"},"kernel":"causal_conv1d::causal_conv1d_update_l2norm_f32","family":"causal_conv1d_l2norm","compute":"cuda_core","where":"device","group":3,"inserted":false},{"id":"gdn.recur","op":"gdn_recurrence","kind":"linear_attention","block_type":"gdn","precision":{"weight":null,"activation":"f32","output":"f32"},"kernel":"gated_delta_rule::gated_delta_rule_decode_f32","family":"gdn_recurrence","compute":"cuda_core","where":"device","group":4,"inserted":false},{"id":"gdn.out_norm","op":"gated_rms_norm","kind":"linear_attention","block_type":"gdn","precision":{"weight":null,"activation":"f32","output":"bf16"},"kernel":"norm::gated_rms_norm_f32_input","family":"gated_rms_norm","compute":"memory","where":"device","group":5,"inserted":false},{"id":"gdn.out","op":"linear:gdn_out","kind":"linear_attention","block_type":"gdn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":6,"inserted":false},{"id":"gdn.add","op":"residual_add","kind":"linear_attention","block_type":"gdn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"norm::residual_add_rms_norm","family":"rms_norm","compute":"memory","where":"device","group":7,"inserted":false},{"id":"dense_ffn.post_norm","op":"rms_norm","kind":"linear_attention","block_type":"dense_ffn","precision":{"weight":null,"activation":"f32","output":"bf16"},"kernel":"norm::residual_add_rms_norm","family":"rms_norm","compute":"memory","where":"device","group":7,"inserted":false},{"id":"dense_ffn.gate_up","op":"linear:gate_up","kind":"linear_attention","block_type":"dense_ffn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":8,"inserted":false},{"id":"dense_ffn.act","op":"silu_mul","kind":"linear_attention","block_type":"dense_ffn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","compute":"memory","where":"device","group":9,"inserted":false},{"id":"dense_ffn.down","op":"linear:down","kind":"linear_attention","block_type":"dense_ffn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":10,"inserted":false},{"id":"dense_ffn.add","op":"residual_add","kind":"linear_attention","block_type":"dense_ffn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","compute":"memory","where":"device","group":11,"inserted":false}],"edges":[{"from":0,"to":1,"format":"bf16","fused":false},{"from":0,"to":2,"format":"bf16","fused":false},{"from":1,"to":4,"format":"bf16","fused":false},{"from":1,"to":8,"format":"bf16","fused":false},{"from":2,"to":3,"format":"f32","fused":true},{"from":3,"to":7,"format":"f32","fused":false},{"from":3,"to":7,"format":"f32","fused":false},{"from":4,"to":5,"format":"f32","fused":true},{"from":4,"to":6,"format":"f32","fused":true},{"from":6,"to":7,"format":"f32","fused":false},{"from":7,"to":8,"format":"f32","fused":false},{"from":8,"to":9,"format":"bf16","fused":false},{"from":9,"to":10,"format":"bf16","fused":false},{"from":10,"to":11,"format":"bf16","fused":false},{"from":10,"to":15,"format":"bf16","fused":false},{"from":11,"to":12,"format":"bf16","fused":false},{"from":12,"to":13,"format":"bf16","fused":false},{"from":13,"to":14,"format":"bf16","fused":false},{"from":14,"to":15,"format":"bf16","fused":false}],"ports":{"in":[0,10],"out":[15]},"instances":[0,1,2,4,5,6,8,9,10,12,13,14,16,17,18,20,21,22,24,25,26,28,29,30,32,33,34,36,37,38,40,41,42,44,45,46,48,49,50,52,53,54,56,57,58,60,61,62]},"attn+dense_ffn":{"blocks":["attn","dense_ffn"],"nodes":[{"id":"attn.input_norm","op":"rms_norm","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","compute":"memory","where":"device","group":0,"inserted":false},{"id":"attn.q","op":"linear:q","kind":"full_attention","block_type":"attn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":1,"inserted":false},{"id":"attn.q_split","op":"split","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"kernel":"w4a16_gemv::w4a16_gemv_qg","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":1,"inserted":false},{"id":"attn.k","op":"linear:k","kind":"full_attention","block_type":"attn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":2,"inserted":false},{"id":"attn.v","op":"linear:v","kind":"full_attention","block_type":"attn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv_fused::w4a16_gemv_dual","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":2,"inserted":false},{"id":"attn.q_norm","op":"qk_norm","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"norm::rms_norm","family":"rms_norm","compute":"memory","where":"device","group":3,"inserted":false},{"id":"attn.k_norm","op":"qk_norm","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"norm::rms_norm","family":"rms_norm","compute":"memory","where":"device","group":4,"inserted":false},{"id":"attn.rope","op":"rope","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":"bf16,bf16"},"kernel":"rope_mrope_interleaved::rope_forward_mrope_interleaved","family":"rope","compute":"memory","where":"device","group":5,"inserted":false},{"id":"attn.kv_write","op":"kv_write","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":null},"kernel":"reshape_and_cache::reshape_and_cache_flash","family":"kv_write","compute":"memory","where":"device","group":6,"inserted":false},{"id":"attn.attend","op":"paged_attention","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"paged_decode::paged_decode_attn","family":"paged_decode_attn","compute":"cuda_core","where":"device","group":7,"inserted":false},{"id":"attn.gate_mul","op":"sigmoid_gate_mul","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"residual_add::sigmoid_gate_mul","family":"sigmoid_gate_mul","compute":"memory","where":"device","group":8,"inserted":false},{"id":"attn.o","op":"linear:o","kind":"full_attention","block_type":"attn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":9,"inserted":false},{"id":"attn.add","op":"residual_add","kind":"full_attention","block_type":"attn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"norm::residual_add_rms_norm","family":"rms_norm","compute":"memory","where":"device","group":10,"inserted":false},{"id":"dense_ffn.post_norm","op":"rms_norm","kind":"full_attention","block_type":"dense_ffn","precision":{"weight":null,"activation":"f32","output":"bf16"},"kernel":"norm::residual_add_rms_norm","family":"rms_norm","compute":"memory","where":"device","group":10,"inserted":false},{"id":"dense_ffn.gate_up","op":"linear:gate_up","kind":"full_attention","block_type":"dense_ffn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv_fused::w4a16_gemv_dual_sw","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":11,"inserted":false},{"id":"dense_ffn.act","op":"silu_mul","kind":"full_attention","block_type":"dense_ffn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"moe_silu_mul::moe_silu_mul","family":"silu_mul","compute":"memory","where":"device","group":12,"inserted":false},{"id":"dense_ffn.down","op":"linear:down","kind":"full_attention","block_type":"dense_ffn","precision":{"weight":"nvfp4/g16","activation":"bf16","output":"bf16"},"kernel":"w4a16_gemv::w4a16_gemv_sw","family":"w4a16_gemv","compute":"cuda_core","where":"device","group":13,"inserted":false},{"id":"dense_ffn.add","op":"residual_add","kind":"full_attention","block_type":"dense_ffn","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"residual_add_rms_norm_exact::residual_add_rms_norm_exact","family":"rms_norm","compute":"memory","where":"device","group":14,"inserted":false}],"edges":[{"from":0,"to":1,"format":"bf16","fused":false},{"from":0,"to":3,"format":"bf16","fused":false},{"from":0,"to":4,"format":"bf16","fused":false},{"from":1,"to":2,"format":"bf16","fused":true},{"from":2,"to":5,"format":"bf16","fused":false},{"from":2,"to":10,"format":"bf16","fused":false},{"from":3,"to":6,"format":"bf16","fused":false},{"from":4,"to":8,"format":"bf16","fused":false},{"from":4,"to":9,"format":"bf16","fused":false},{"from":5,"to":7,"format":"bf16","fused":false},{"from":6,"to":7,"format":"bf16","fused":false},{"from":7,"to":9,"format":"bf16","fused":false},{"from":7,"to":8,"format":"bf16","fused":false},{"from":7,"to":9,"format":"bf16","fused":false},{"from":9,"to":10,"format":"bf16","fused":false},{"from":10,"to":11,"format":"bf16","fused":false},{"from":11,"to":12,"format":"bf16","fused":false},{"from":12,"to":13,"format":"bf16","fused":false},{"from":12,"to":17,"format":"bf16","fused":false},{"from":13,"to":14,"format":"bf16","fused":false},{"from":14,"to":15,"format":"bf16","fused":false},{"from":15,"to":16,"format":"bf16","fused":false},{"from":16,"to":17,"format":"bf16","fused":false}],"ports":{"in":[0,12],"out":[17]},"instances":[3,7,11,15,19,23,27,31,35,39,43,47,51,55,59,63]},"head":{"blocks":["head"],"nodes":[{"id":"head.final_norm","op":"final_norm","kind":null,"block_type":"head","precision":{"weight":null,"activation":"bf16","output":"bf16"},"kernel":"norm::rms_norm","family":"rms_norm","compute":"memory","where":"device","group":0,"inserted":false},{"id":"head.lm_head","op":"lm_head","kind":null,"block_type":"head","precision":{"weight":"bf16","activation":"bf16","output":"bf16"},"kernel":"gemv::dense_gemv_bf16","family":"dense_bf16","compute":"cuda_core","where":"device","group":1,"inserted":false},{"id":"head.argmax","op":"argmax","kind":null,"block_type":"head","precision":{"weight":null,"activation":"bf16","output":"i32"},"kernel":null,"family":null,"compute":null,"where":"host","group":2,"inserted":false}],"edges":[{"from":0,"to":1,"format":"bf16","fused":false},{"from":1,"to":2,"format":"bf16","fused":false}],"ports":{"in":[0],"out":[]},"instances":[]}}},"provenance":{"kind":"preview","engine_commit":"c5b9a80e635df7c1056a5b617272970f25daba64","planner_sha256":"f4735654ee1d5a9bf4f56e836969140d5bfb26231104bf13118103c69ee8b156","planner_built_from":"c5b9a80e635df7c1056a5b617272970f25daba64","planner_header":{"recipe":"qwen3.8/qwen3.8-27b-nvfp4-unsloth","checkpoint":"unsloth/Qwen3.8-27B-NVFP4","target":"gb10/qwen3.8-27b/nvfp4","settings":"activation_quantization=adaptive decode_split_silu=on gemv_sw=on kv_cache_dtype=bf16 lm_head_dtype=bf16 rms_norm_act_quant=off row_tiers=by_rows ssm_ba_gates_hopper=on ssm_batched_recurrent=on ssm_h_dtype=f32 w4a16_tc=on","opt-in levers":"none"},"commands":["met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode decode --rows 1","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 2","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 4","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 8","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 12","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 16","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 24","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 32","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 48","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 64","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 96","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode multi_seq --rows 128","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode verify --rows 2","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode verify --rows 3","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode verify --rows 4","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode draft --rows 1","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode draft --rows 2","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode draft --rows 5","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode draft --rows 8","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode draft --rows 9","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode draft --rows 16","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode draft --rows 17","met circuit plan --root <tree> --checkpoint qwen3.8/qwen3.8-27b-nvfp4-unsloth --hardware gb10 --precision recipe --format plan --mode draft --rows 32"],"files":[{"path":"kernels/circuits/INSTANCES.toml","sha256":"31418b33cf20e56d353d5246079b12caa97ed3bca56ec33b2b6182076d43f32c"},{"path":"recipes/qwen3.8/qwen3.8-27b-nvfp4-unsloth.yaml","sha256":"d740dae13f464b32f85f808427be20cc86a9990fc52921aff716ac6e65724e44"},{"path":"kernels/gb10/HARDWARE.toml","sha256":"24901961e415fe6a90afba954dfaa0f523f907351f42f85d9da4c19c8f7842db"},{"path":"kernels/DEVICES.toml","sha256":"256d85840f0e77ec84c5fc739f30aacff6bf24775cc8bef344e36647e51e9c25"},{"path":"kernels/circuits/qwen3_5.toml","sha256":"64a2d3106d56a0fb8137c25811e763623f91d15783b4708fa8874c63a9313814"},{"path":"kernels/circuits/blocks/qwen3_hybrid.toml","sha256":"cfe66fc69723840ffe0ff9c5c12b91bb51935aef45898d5ae56567b5f2d38ab6"},{"path":"kernels/gb10/common/KERNEL_FAMILIES.toml","sha256":"4b8e793753ee539110969ff27da74aaa7d4685a612f5ba7b4062a86dc92b6c51"},{"path":"kernels/circuits/checkpoints/unsloth--Qwen3.8-27B-NVFP4.toml","sha256":"ab75a565830ec530cf5adf3d62ebfa16b048b7c08ef6c6683b92119a10e4ad59"}]}}
