LICENSE
README.md
pyproject.toml
licenses/Anthropic-Apache-2.0.txt
licenses/Anthropic-NOTICE
licenses/CUTLASS-LICENSE.txt
licenses/FlashAttention-LICENSE.txt
licenses/THIRD_PARTY_NOTICES.md
src/miniworld_engine/__init__.py
src/miniworld_engine/_atomic.py
src/miniworld_engine/_typecheck.py
src/miniworld_engine/cli.py
src/miniworld_engine/py.typed
src/miniworld_engine/settings.py
src/miniworld_engine.egg-info/PKG-INFO
src/miniworld_engine.egg-info/SOURCES.txt
src/miniworld_engine.egg-info/dependency_links.txt
src/miniworld_engine.egg-info/entry_points.txt
src/miniworld_engine.egg-info/requires.txt
src/miniworld_engine.egg-info/top_level.txt
src/miniworld_engine/architecture/__init__.py
src/miniworld_engine/architecture/_engine_impl.py
src/miniworld_engine/architecture/exceptions.py
src/miniworld_engine/architecture/primitives.py
src/miniworld_engine/architecture/tensor_layout.py
src/miniworld_engine/architecture/typing.py
src/miniworld_engine/architecture/blocks/__init__.py
src/miniworld_engine/architecture/blocks/_engine_impl.py
src/miniworld_engine/architecture/blocks/diffusion_transformer.py
src/miniworld_engine/architecture/blocks/msa_module.py
src/miniworld_engine/architecture/blocks/pair_bias_hoist.py
src/miniworld_engine/architecture/blocks/pairformer.py
src/miniworld_engine/architecture/blocks/rope_swa_af3_transformer.py
src/miniworld_engine/architecture/blocks/swa_atom_transformer.py
src/miniworld_engine/architecture/blocks/template.py
src/miniworld_engine/architecture/checkpoints/__init__.py
src/miniworld_engine/architecture/checkpoints/backend_attention.py
src/miniworld_engine/architecture/diffusion/__init__.py
src/miniworld_engine/architecture/diffusion/augmentation.py
src/miniworld_engine/architecture/diffusion/base/__init__.py
src/miniworld_engine/architecture/diffusion/base/diffuser.py
src/miniworld_engine/architecture/diffusion/base/scheduler.py
src/miniworld_engine/architecture/diffusion/base/solver.py
src/miniworld_engine/architecture/diffusion/decoupled_xpred/__init__.py
src/miniworld_engine/architecture/diffusion/decoupled_xpred/diffuser.py
src/miniworld_engine/architecture/diffusion/decoupled_xpred/scheduler.py
src/miniworld_engine/architecture/diffusion/decoupled_xpred/solver.py
src/miniworld_engine/architecture/diffusion/edm/__init__.py
src/miniworld_engine/architecture/diffusion/edm/diffuser.py
src/miniworld_engine/architecture/diffusion/edm/preconditioned.py
src/miniworld_engine/architecture/diffusion/edm/sampling.py
src/miniworld_engine/architecture/diffusion/edm/scheduler.py
src/miniworld_engine/architecture/diffusion/edm/schedules.py
src/miniworld_engine/architecture/diffusion/edm/solver.py
src/miniworld_engine/architecture/layers/__init__.py
src/miniworld_engine/architecture/layers/adaln.py
src/miniworld_engine/architecture/layers/attention_pair_bias.py
src/miniworld_engine/architecture/layers/augmented_attention.py
src/miniworld_engine/architecture/layers/conditioned_transition.py
src/miniworld_engine/architecture/layers/embeddings.py
src/miniworld_engine/architecture/layers/msa_pair_weighted_averaging.py
src/miniworld_engine/architecture/layers/ops.py
src/miniworld_engine/architecture/layers/outer_product.py
src/miniworld_engine/architecture/layers/swa_atom_attention.py
src/miniworld_engine/architecture/layers/transition.py
src/miniworld_engine/architecture/layers/triangle_updates.py
src/miniworld_engine/architecture/utils/__init__.py
src/miniworld_engine/architecture/utils/align.py
src/miniworld_engine/architecture/utils/se3.py
src/miniworld_engine/architecture/utils/transform.py
src/miniworld_engine/autotune/__init__.py
src/miniworld_engine/autotune/axes.csv
src/miniworld_engine/autotune/buckets.py
src/miniworld_engine/autotune/builder.py
src/miniworld_engine/autotune/cache.py
src/miniworld_engine/autotune/cache_backfill.py
src/miniworld_engine/autotune/cache_status.py
src/miniworld_engine/autotune/capture.py
src/miniworld_engine/autotune/checkpoint_cases.py
src/miniworld_engine/autotune/compile_budget.py
src/miniworld_engine/autotune/configs.py
src/miniworld_engine/autotune/derive.py
src/miniworld_engine/autotune/devices.py
src/miniworld_engine/autotune/fused_config.py
src/miniworld_engine/autotune/hopper_cuda_config.py
src/miniworld_engine/autotune/module_registry.py
src/miniworld_engine/autotune/native.py
src/miniworld_engine/autotune/native_compile.py
src/miniworld_engine/autotune/native_history.py
src/miniworld_engine/autotune/plan.py
src/miniworld_engine/autotune/policy.py
src/miniworld_engine/autotune/preflight.py
src/miniworld_engine/autotune/round_cache.py
src/miniworld_engine/autotune/run_all.py
src/miniworld_engine/autotune/shape_key.py
src/miniworld_engine/autotune/shard.py
src/miniworld_engine/autotune/smem_log.py
src/miniworld_engine/autotune/space_backfill.py
src/miniworld_engine/autotune/triton_cache.py
src/miniworld_engine/autotune/viability.py
src/miniworld_engine/autotune/width_evidence.py
src/miniworld_engine/autotune/configs/default/adaln_bwd_dx_dlnw_triton.csv
src/miniworld_engine/autotune/configs/default/adaln_bwd_pre_dx_triton.csv
src/miniworld_engine/autotune/configs/default/adaln_epilogue_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/adaln_epilogue_triton.csv
src/miniworld_engine/autotune/configs/default/adaln_fwd_gate_triton.csv
src/miniworld_engine/autotune/configs/default/adaln_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/adaln_gemm_gate_triton.csv
src/miniworld_engine/autotune/configs/default/augmented_attention_bwd_atomic_triton.csv
src/miniworld_engine/autotune/configs/default/augmented_attention_bwd_pre_triton.csv
src/miniworld_engine/autotune/configs/default/augmented_attention_bwd_reduce_triton.csv
src/miniworld_engine/autotune/configs/default/augmented_attention_bwd_split_triton.csv
src/miniworld_engine/autotune/configs/default/augmented_attention_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/bias_only_attention_bwd_pre_triton.csv
src/miniworld_engine/autotune/configs/default/bias_only_attention_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/bias_only_attention_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_bwd_gemm_swiglu_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_bwd_swiglu_flat_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_expand_swiglu_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_expand_swiglu_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_fwd_b2b_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_fwd_b2b_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_squeeze_gate_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_squeeze_gate_triton.csv
src/miniworld_engine/autotune/configs/default/cond_transition_swiglu_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_bwd_dx_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_bwd_gate_dropres_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_bwd_gate_flat_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_bwd_gate_recompute_flat_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_bwd_gate_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_gate_dropres_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_gate_flat_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_gate_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/gated_projection_gate_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_bwd_atomic_strided_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_bwd_atomic_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_bwd_foldstats_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_bwd_split_mmajor_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_bwd_split_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_fwd_mmajor_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_fwd_recompute_noaffine_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_fwd_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_fwd_saveact_strided_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_fwd_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_fwd_strided_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_linear_bwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_linear_fwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_linear_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/layernorm_stats_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_dropout_bwd_packed_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_dropout_fwd_packed_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_mlp_fwd_gemm_b2b_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_mlp_fwd_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_bwd_dx_gather_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_bwd_dx_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_bwd_dx_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_bwd_layernorm_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_bwd_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_fwd_gemm_b2b_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_fwd_gemm_gather_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_fwd_gemm_layernorm_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_fwd_gemm_layernorm_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_edge_tail_fwd_gemm_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_message_bwd_dx_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_message_bwd_reduce_dbias_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_message_fwd_gelu_reduce_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_message_fwd_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_node_message_bwd_dx_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_node_message_bwd_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_node_message_fwd_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/mpnn_relative_position_bwd_reduce_triton.csv
src/miniworld_engine/autotune/configs/default/outer_product_mean_bwd_epilogue_dw_triton.csv
src/miniworld_engine/autotune/configs/default/outer_product_mean_bwd_epilogue_dx_triton.csv
src/miniworld_engine/autotune/configs/default/outer_product_mean_bwd_layernorm_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/outer_product_mean_epilogue_triton.csv
src/miniworld_engine/autotune/configs/default/outer_product_mean_layernorm_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/pair_weighted_averaging_bwd_gate_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/pair_weighted_averaging_bwd_layernorm_gemm_dw_triton.csv
src/miniworld_engine/autotune/configs/default/pair_weighted_averaging_bwd_layernorm_gemm_dx_dlnw_triton.csv
src/miniworld_engine/autotune/configs/default/pair_weighted_averaging_bwd_layernorm_gemm_softmax_triton.csv
src/miniworld_engine/autotune/configs/default/pair_weighted_averaging_gate_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/pair_weighted_averaging_layernorm_gemm_softmax_triton.csv
src/miniworld_engine/autotune/configs/default/pair_weighted_averaging_layernorm_gemm_triton.csv
src/miniworld_engine/autotune/configs/default/qk_norm_rope_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/qk_norm_rope_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/rmsnorm_adamod_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/rmsnorm_adamod_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/rmsnorm_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/rmsnorm_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/rope_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_inproj_bwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_inproj_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_inproj_fwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_inproj_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_output_bwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_output_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_output_swiglu_fwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_output_swiglu_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_softmax_bwd_dkdv_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_softmax_bwd_dq_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_softmax_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_swiglu_bwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_swiglu_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/swa_dit_swiglu_dw_triton.csv
src/miniworld_engine/autotune/configs/default/swa_gate_out_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/transition_bwd_swiglu_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/transition_expand_swiglu_triton.csv
src/miniworld_engine/autotune/configs/default/transition_fwd_b2b_ktiled_triton.csv
src/miniworld_engine/autotune/configs/default/transition_fwd_b2b_triton.csv
src/miniworld_engine/autotune/configs/default/transition_layernorm_expand_swiglu_triton.csv
src/miniworld_engine/autotune/configs/default/transition_squeeze_residual_triton.csv
src/miniworld_engine/autotune/configs/default/triangle_attention_bwd_atomic_triton.csv
src/miniworld_engine/autotune/configs/default/triangle_attention_bwd_dkdv_triton.csv
src/miniworld_engine/autotune/configs/default/triangle_attention_bwd_dq_triton.csv
src/miniworld_engine/autotune/configs/default/triangle_attention_bwd_pre_contig_triton.csv
src/miniworld_engine/autotune/configs/default/triangle_attention_bwd_pre_triton.csv
src/miniworld_engine/autotune/configs/default/triangle_attention_fwd_contig_triton.csv
src/miniworld_engine/autotune/configs/default/triangle_attention_fwd_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_bwd_gate_packed_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_bwd_gate_packed_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_bwd_gate_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_gemm_gate_mmajor_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_gemm_gate_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_gemm_gate_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_input_dual_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_input_ln_residual_bwd_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_outproj_bwd_gate_recompute_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_outproj_gemm_gate_saveact_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_outproj_gemm_gate_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_outproj_layernorm_gemm_gate_triton.csv
src/miniworld_engine/autotune/configs/default/trimul_output_f567_train_triton.csv
src/miniworld_engine/autotune/configs/grid/adaln_bwd_dx_dlnw_triton.csv
src/miniworld_engine/autotune/configs/grid/adaln_bwd_pre_dx_triton.csv
src/miniworld_engine/autotune/configs/grid/adaln_epilogue_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/adaln_epilogue_triton.csv
src/miniworld_engine/autotune/configs/grid/adaln_fwd_gate_triton.csv
src/miniworld_engine/autotune/configs/grid/adaln_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/adaln_gemm_gate_triton.csv
src/miniworld_engine/autotune/configs/grid/augmented_attention_bwd_atomic_triton.csv
src/miniworld_engine/autotune/configs/grid/augmented_attention_bwd_pre_triton.csv
src/miniworld_engine/autotune/configs/grid/augmented_attention_bwd_reduce_triton.csv
src/miniworld_engine/autotune/configs/grid/augmented_attention_bwd_split_triton.csv
src/miniworld_engine/autotune/configs/grid/augmented_attention_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/bias_only_attention_bwd_pre_triton.csv
src/miniworld_engine/autotune/configs/grid/bias_only_attention_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/bias_only_attention_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_bwd_gemm_swiglu_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_bwd_swiglu_flat_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_expand_swiglu_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_expand_swiglu_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_fwd_b2b_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_fwd_b2b_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_squeeze_gate_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_squeeze_gate_triton.csv
src/miniworld_engine/autotune/configs/grid/cond_transition_swiglu_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_bwd_dx_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_bwd_gate_dropres_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_bwd_gate_flat_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_bwd_gate_recompute_flat_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_bwd_gate_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_gate_dropres_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_gate_flat_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_gate_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/gated_projection_gate_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_bwd_atomic_strided_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_bwd_atomic_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_bwd_foldstats_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_bwd_split_mmajor_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_bwd_split_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_fwd_mmajor_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_fwd_recompute_noaffine_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_fwd_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_fwd_saveact_strided_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_fwd_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_fwd_strided_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_linear_bwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_linear_fwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_linear_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/layernorm_stats_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_dropout_bwd_packed_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_dropout_fwd_packed_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_mlp_fwd_gemm_b2b_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_mlp_fwd_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_bwd_dx_gather_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_bwd_dx_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_bwd_dx_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_bwd_layernorm_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_bwd_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_fwd_gemm_b2b_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_fwd_gemm_gather_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_fwd_gemm_layernorm_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_fwd_gemm_layernorm_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_edge_tail_fwd_gemm_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_message_bwd_dx_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_message_bwd_reduce_dbias_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_message_fwd_gelu_reduce_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_message_fwd_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_node_message_bwd_dx_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_node_message_bwd_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_node_message_fwd_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/mpnn_relative_position_bwd_reduce_triton.csv
src/miniworld_engine/autotune/configs/grid/outer_product_mean_bwd_epilogue_dw_triton.csv
src/miniworld_engine/autotune/configs/grid/outer_product_mean_bwd_epilogue_dx_triton.csv
src/miniworld_engine/autotune/configs/grid/outer_product_mean_bwd_layernorm_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/outer_product_mean_epilogue_triton.csv
src/miniworld_engine/autotune/configs/grid/outer_product_mean_layernorm_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/pair_weighted_averaging_bwd_gate_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/pair_weighted_averaging_bwd_layernorm_gemm_dw_triton.csv
src/miniworld_engine/autotune/configs/grid/pair_weighted_averaging_bwd_layernorm_gemm_dx_dlnw_triton.csv
src/miniworld_engine/autotune/configs/grid/pair_weighted_averaging_bwd_layernorm_gemm_softmax_triton.csv
src/miniworld_engine/autotune/configs/grid/pair_weighted_averaging_gate_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/pair_weighted_averaging_layernorm_gemm_softmax_triton.csv
src/miniworld_engine/autotune/configs/grid/pair_weighted_averaging_layernorm_gemm_triton.csv
src/miniworld_engine/autotune/configs/grid/qk_norm_rope_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/qk_norm_rope_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/rmsnorm_adamod_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/rmsnorm_adamod_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/rmsnorm_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/rmsnorm_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/rope_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_inproj_bwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_inproj_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_inproj_fwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_inproj_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_output_bwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_output_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_output_swiglu_fwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_output_swiglu_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_softmax_bwd_dkdv_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_softmax_bwd_dq_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_softmax_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_swiglu_bwd_fp32_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_swiglu_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_dit_swiglu_dw_triton.csv
src/miniworld_engine/autotune/configs/grid/swa_gate_out_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/transition_bwd_swiglu_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/transition_expand_swiglu_triton.csv
src/miniworld_engine/autotune/configs/grid/transition_fwd_b2b_ktiled_triton.csv
src/miniworld_engine/autotune/configs/grid/transition_fwd_b2b_triton.csv
src/miniworld_engine/autotune/configs/grid/transition_layernorm_expand_swiglu_triton.csv
src/miniworld_engine/autotune/configs/grid/transition_squeeze_residual_triton.csv
src/miniworld_engine/autotune/configs/grid/triangle_attention_bwd_atomic_triton.csv
src/miniworld_engine/autotune/configs/grid/triangle_attention_bwd_dkdv_triton.csv
src/miniworld_engine/autotune/configs/grid/triangle_attention_bwd_dq_triton.csv
src/miniworld_engine/autotune/configs/grid/triangle_attention_bwd_pre_contig_triton.csv
src/miniworld_engine/autotune/configs/grid/triangle_attention_bwd_pre_triton.csv
src/miniworld_engine/autotune/configs/grid/triangle_attention_fwd_contig_triton.csv
src/miniworld_engine/autotune/configs/grid/triangle_attention_fwd_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_bwd_gate_packed_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_bwd_gate_packed_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_bwd_gate_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_gemm_gate_mmajor_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_gemm_gate_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_gemm_gate_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_input_dual_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_input_ln_residual_bwd_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_outproj_bwd_gate_recompute_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_outproj_gemm_gate_saveact_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_outproj_gemm_gate_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_outproj_layernorm_gemm_gate_triton.csv
src/miniworld_engine/autotune/configs/grid/trimul_output_f567_train_triton.csv
src/miniworld_engine/autotune/data/adaln_bwd_dx_dlnw_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/adaln_bwd_pre_dx_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/adaln_bwd_pre_dx_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/adaln_bwd_pre_dx_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/adaln_bwd_pre_dx_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/adaln_epilogue_saveact_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/adaln_epilogue_saveact_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/adaln_epilogue_saveact_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/adaln_epilogue_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/adaln_epilogue_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/adaln_epilogue_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/adaln_epilogue_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/adaln_epilogue_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/adaln_fwd_gate_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/adaln_fwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/adaln_fwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/adaln_fwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/adaln_gemm_gate_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/adaln_gemm_gate_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_atomic_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_atomic_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_pre_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_pre_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_pre_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_pre_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_reduce_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_reduce_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_split_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/augmented_attention_bwd_split_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/augmented_attention_fwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/augmented_attention_fwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/augmented_attention_fwd_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/augmented_attention_fwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/bias_only_dispatch/NVIDIA_A100_80GB_PCIe_sm80_triton3.6.0.json
src/miniworld_engine/autotune/data/bias_only_dispatch/NVIDIA_B200_sm100_triton3.7.1.json
src/miniworld_engine/autotune/data/bias_only_dispatch/NVIDIA_RTX_A5000_sm86_triton3.6.0.json
src/miniworld_engine/autotune/data/bias_only_dispatch/NVIDIA_RTX_A6000_sm86_triton3.6.0.json
src/miniworld_engine/autotune/data/cond_transition_bwd_gemm_swiglu_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_bwd_gemm_swiglu_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_bwd_gemm_swiglu_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_bwd_gemm_swiglu_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_bwd_swiglu_flat_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_bwd_swiglu_flat_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_bwd_swiglu_flat_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_bwd_swiglu_flat_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_expand_swiglu_saveact_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_expand_swiglu_saveact_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_expand_swiglu_saveact_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_expand_swiglu_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_expand_swiglu_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_expand_swiglu_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_expand_swiglu_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_expand_swiglu_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_fwd_b2b_saveact_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_fwd_b2b_saveact_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_fwd_b2b_saveact_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_fwd_b2b_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_fwd_b2b_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_fwd_b2b_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_fwd_b2b_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_fwd_b2b_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_squeeze_gate_saveact_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_squeeze_gate_saveact_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_squeeze_gate_saveact_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_squeeze_gate_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_squeeze_gate_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_squeeze_gate_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_squeeze_gate_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_squeeze_gate_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_swiglu_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/cond_transition_swiglu_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/cond_transition_swiglu_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/cond_transition_swiglu_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_dx_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_bwd_dx_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/gated_projection_bwd_dx_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_dx_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_dropres_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_dropres_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_dropres_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_dropres_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_dropres_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_flat_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_flat_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_flat_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_flat_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_recompute_flat_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_recompute_flat_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_recompute_flat_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_bwd_gate_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_gate_dropres_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_gate_dropres_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/gated_projection_gate_dropres_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_gate_dropres_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_gate_flat_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_gate_flat_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/gated_projection_gate_flat_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_gate_flat_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_gate_gemm_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_gate_gemm_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/gated_projection_gate_gemm_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_gate_gemm_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_gate_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/gated_projection_gate_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/gated_projection_gate_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/gated_projection_gate_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_bwd_atomic_strided_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_bwd_atomic_strided_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/layernorm_bwd_atomic_strided_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_bwd_atomic_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_bwd_atomic_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_bwd_foldstats_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_mmajor_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_mmajor_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_mmajor_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_mmajor_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_mmajor_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_bwd_split_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_fwd_mmajor_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_fwd_mmajor_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_fwd_mmajor_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_fwd_mmajor_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_fwd_recompute_noaffine_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_fwd_recompute_noaffine_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_fwd_recompute_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_fwd_recompute_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_fwd_saveact_strided_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_fwd_saveact_strided_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/layernorm_fwd_saveact_strided_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_fwd_saveact_strided_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_fwd_saveact_strided_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_fwd_saveact_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_fwd_saveact_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/layernorm_fwd_saveact_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_fwd_strided_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_fwd_strided_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_fwd_strided_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_fwd_strided_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_linear_bwd_fp32_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_linear_bwd_fp32_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_linear_fwd_fp32_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_linear_fwd_fp32_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_linear_fwd_fp32_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_linear_fwd_fp32_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_linear_fwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_linear_fwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_linear_fwd_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_linear_fwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_stats_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/layernorm_stats_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/layernorm_stats_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/layernorm_stats_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/ln_bwd_dispatch/NVIDIA_A100_80GB_PCIe_sm80_triton3.6.0.json
src/miniworld_engine/autotune/data/ln_bwd_dispatch/NVIDIA_B200_sm100_triton3.7.1.json
src/miniworld_engine/autotune/data/ln_bwd_dispatch/NVIDIA_H100_80GB_HBM3_sm90_triton3.6.0.json
src/miniworld_engine/autotune/data/ln_bwd_dispatch/NVIDIA_RTX_A5000_sm86_triton3.6.0.json
src/miniworld_engine/autotune/data/ln_bwd_dispatch/NVIDIA_RTX_A6000_sm86_triton3.6.0.json
src/miniworld_engine/autotune/data/mpnn_edge_tail_bwd_dx_gather_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_edge_tail_bwd_dx_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_edge_tail_bwd_layernorm_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_edge_tail_fwd_gemm_b2b_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_edge_tail_fwd_gemm_gather_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_edge_tail_fwd_gemm_layernorm_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_message_bwd_dx_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_message_bwd_reduce_dbias_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_message_fwd_gelu_reduce_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_message_fwd_gemm_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_node_message_bwd_dx_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_node_message_bwd_recompute_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_node_message_fwd_gemm_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/mpnn_relative_position_bwd_reduce_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/qk_norm_rope_bwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/qk_norm_rope_bwd_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/qk_norm_rope_bwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/qk_norm_rope_fwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/qk_norm_rope_fwd_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/qk_norm_rope_fwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/rmsnorm_adamod_bwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/rmsnorm_adamod_fwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/rmsnorm_bwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/rmsnorm_bwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/rmsnorm_fwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/rmsnorm_fwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/rmsnorm_fwd_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/rmsnorm_fwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/rope_fwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/rope_fwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/rope_fwd_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/rope_fwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/swa_gate_out_fwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/swa_gate_out_fwd_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/swa_gate_out_fwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/transition_bwd_swiglu_recompute_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/transition_expand_swiglu_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/transition_fwd_b2b_ktiled_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/transition_fwd_b2b_ktiled_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/transition_fwd_b2b_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/transition_fwd_b2b_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/transition_layernorm_expand_swiglu_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/transition_layernorm_expand_swiglu_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_atomic_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_atomic_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_dkdv_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_dkdv_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_dkdv_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_dkdv_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_dq_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_dq_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_dq_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_dq_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_pre_contig_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_pre_contig_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_pre_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_pre_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_pre_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_bwd_pre_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_fwd_contig_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/triangle_attention_fwd_contig_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_fwd_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/triangle_attention_fwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/triangle_attention_fwd_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/triangle_attention_fwd_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_packed_recompute_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_packed_recompute_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_packed_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_packed_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_packed_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_packed_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_packed_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_recompute_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_recompute_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_recompute_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/trimul_bwd_gate_recompute_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_gemm_gate_mmajor_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_gemm_gate_mmajor_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/trimul_gemm_gate_saveact_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_gemm_gate_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_gemm_gate_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_gemm_gate_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/trimul_gemm_gate_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/trimul_gemm_gate_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_input_dual_bwd_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/trimul_input_ln_residual_bwd_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/trimul_input_ln_residual_bwd_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/trimul_outproj_bwd_gate_recompute_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_outproj_bwd_gate_recompute_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/trimul_outproj_bwd_gate_recompute_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/trimul_outproj_bwd_gate_recompute_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_outproj_gemm_gate_saveact_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_outproj_gemm_gate_saveact_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_outproj_gemm_gate_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_outproj_gemm_gate_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/trimul_outproj_gemm_gate_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/trimul_outproj_gemm_gate_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_outproj_layernorm_gemm_gate_triton/NVIDIA A100 80GB PCIe (sm80).json
src/miniworld_engine/autotune/data/trimul_outproj_layernorm_gemm_gate_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/trimul_outproj_layernorm_gemm_gate_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/data/trimul_outproj_layernorm_gemm_gate_triton/NVIDIA RTX A5000 (sm86).json
src/miniworld_engine/autotune/data/trimul_outproj_layernorm_gemm_gate_triton/NVIDIA RTX A6000 (sm86).json
src/miniworld_engine/autotune/data/trimul_output_f567_train_triton/NVIDIA B200 (sm100).json
src/miniworld_engine/autotune/data/trimul_output_f567_train_triton/NVIDIA H100 80GB HBM3 (sm90).json
src/miniworld_engine/autotune/manifests/NVIDIA A100 80GB PCIe (sm80).csv
src/miniworld_engine/autotune/manifests/NVIDIA H100 80GB HBM3 (sm90).csv
src/miniworld_engine/autotune/manifests/NVIDIA RTX A5000 (sm86).csv
src/miniworld_engine/autotune/manifests/NVIDIA RTX A6000 (sm86).csv
src/miniworld_engine/build/__init__.py
src/miniworld_engine/build/audit.py
src/miniworld_engine/build/bundle_coverage.py
src/miniworld_engine/build/classify.py
src/miniworld_engine/build/kernel_bundle.py
src/miniworld_engine/build/key_gaps.py
src/miniworld_engine/build/key_gaps_allowed.csv
src/miniworld_engine/build/launch_bind.py
src/miniworld_engine/build/matrix.py
src/miniworld_engine/build/gpu_to_kernels/_common.csv
src/miniworld_engine/build/gpu_to_kernels/sm_100.csv
src/miniworld_engine/build/gpu_to_kernels/sm_80.csv
src/miniworld_engine/build/gpu_to_kernels/sm_86.csv
src/miniworld_engine/build/gpu_to_kernels/sm_90.csv
src/miniworld_engine/integrations/__init__.py
src/miniworld_engine/integrations/anthropic.py
src/miniworld_engine/integrations/anthropic_msa.py
src/miniworld_engine/integrations/anthropic_training.py
src/miniworld_engine/integrations/anthropic_trimul.py
src/miniworld_engine/integrations/attention_pair_bias_b200.py
src/miniworld_engine/integrations/bias_only_dit.py
src/miniworld_engine/integrations/bias_only_dit_hoist.py
src/miniworld_engine/integrations/bias_only_dit_train.py
src/miniworld_engine/integrations/h100_master.py
src/miniworld_engine/integrations/local_dit.py
src/miniworld_engine/integrations/local_dit_h100.py
src/miniworld_engine/integrations/opm_train.py
src/miniworld_engine/integrations/optimizer.py
src/miniworld_engine/integrations/pwa_train.py
src/miniworld_engine/integrations/pwa_wide.py
src/miniworld_engine/integrations/token_dit.py
src/miniworld_engine/integrations/token_dit_train.py
src/miniworld_engine/integrations/triattn_b200.py
src/miniworld_engine/integrations/triattn_h100.py
src/miniworld_engine/integrations/triattn_h100_sm90.py
src/miniworld_engine/integrations/trimul_b200.py
src/miniworld_engine/integrations/trimul_h100.py
src/miniworld_engine/integrations/csrc/dgv_bwd.cu
src/miniworld_engine/integrations/csrc/ln_vg.cu
src/miniworld_engine/integrations/csrc/opm_epilogue.cu
src/miniworld_engine/integrations/csrc/pair3.cu
src/miniworld_engine/integrations/csrc/pair_wide.cu
src/miniworld_engine/integrations/csrc/pwa_ctr.cu
src/miniworld_engine/integrations/csrc/pwa_fwd2.cu
src/miniworld_engine/integrations/csrc/pwa_fwd3.cu
src/miniworld_engine/integrations/csrc/pwa_gateout.cu
src/miniworld_engine/integrations/csrc/pwa_glue3.cu
src/miniworld_engine/integrations/csrc/pwa_wide.cu
src/miniworld_engine/integrations/csrc/wgmma_n64.cuh
src/miniworld_engine/integrations/csrc/sm100/opm_sm100.cu
src/miniworld_engine/integrations/csrc/sm100/pwa_sm100.cu
src/miniworld_engine/integrations/csrc/sm100/sm100.cuh
src/miniworld_engine/kernels/__init__.py
src/miniworld_engine/kernels/_capture.py
src/miniworld_engine/kernels/_codegraph.py
src/miniworld_engine/kernels/_compile.py
src/miniworld_engine/kernels/_nvcc.py
src/miniworld_engine/kernels/_tiles.py
src/miniworld_engine/kernels/prebuilt.py
src/miniworld_engine/kernels/adaln/__init__.py
src/miniworld_engine/kernels/adaln/interface.py
src/miniworld_engine/kernels/adaln/reference.py
src/miniworld_engine/kernels/adaln/triton/__init__.py
src/miniworld_engine/kernels/adaln/triton/inference.py
src/miniworld_engine/kernels/adaln/triton/ln_strided.py
src/miniworld_engine/kernels/adaln/triton/training.py
src/miniworld_engine/kernels/augmented_attention/__init__.py
src/miniworld_engine/kernels/augmented_attention/interface.py
src/miniworld_engine/kernels/augmented_attention/reference.py
src/miniworld_engine/kernels/augmented_attention/whole_op.py
src/miniworld_engine/kernels/augmented_attention/cuda/__init__.py
src/miniworld_engine/kernels/augmented_attention/cuda/attn_dkv.cu
src/miniworld_engine/kernels/augmented_attention/cuda/attn_dq.cu
src/miniworld_engine/kernels/augmented_attention/cuda/attn_dqb.cu
src/miniworld_engine/kernels/augmented_attention/cuda/attn_fwd.cu
src/miniworld_engine/kernels/augmented_attention/cuda/attn_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/apb/__init__.py
src/miniworld_engine/kernels/augmented_attention/cuda/apb/apb_rows.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/__init__.py
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/attn_dkv.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/attn_dkv_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/attn_dqb.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/attn_dqb_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/attn_fwd2.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/attn_fwd_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/attn_inf.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/attn_inf_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/driver.py
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/glue.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100/sm100.cuh
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/__init__.py
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/cond_bwd.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/cond_fwd.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/dw_atom.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/fused_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/gemm_rg_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/gemm_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/post_bwd.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/post_fwd.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/pre_bwd.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/pre_fwd.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/rows_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/sm100.cuh
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom/tr_bwd_gate.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/__init__.py
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lattn90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lattn90_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lattn_dkv.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lattn_dkv_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lattn_dq.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lattn_dq_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lattn_fwd.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lattn_fwd_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lbias.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lbias_tf32.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/lcross.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm100_atom_local/local.cuh
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/__init__.py
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/cond90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/cond_bwd90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/cond_bwd_sh90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/finalize90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/pack90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/post90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/post_bwd90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/pre90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/pre_bwd90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/row90.cuh
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/tr_bwd90.cu
src/miniworld_engine/kernels/augmented_attention/cuda/sm90_atom/wg90.inc
src/miniworld_engine/kernels/augmented_attention/triton/__init__.py
src/miniworld_engine/kernels/augmented_attention/triton/main.py
src/miniworld_engine/kernels/augmented_attention/triton/memory_efficient.py
src/miniworld_engine/kernels/bias_only_attention/__init__.py
src/miniworld_engine/kernels/bias_only_attention/dispatch.py
src/miniworld_engine/kernels/bias_only_attention/interface.py
src/miniworld_engine/kernels/bias_only_attention/reference.py
src/miniworld_engine/kernels/bias_only_attention/triton/__init__.py
src/miniworld_engine/kernels/bias_only_attention/triton/gate_out.py
src/miniworld_engine/kernels/bias_only_attention/triton/main.py
src/miniworld_engine/kernels/bias_only_dit/__init__.py
src/miniworld_engine/kernels/bias_only_dit/interface.py
src/miniworld_engine/kernels/bias_only_dit/reference.py
src/miniworld_engine/kernels/bias_only_dit/cuda/__init__.py
src/miniworld_engine/kernels/bias_only_dit/cuda/bias_only_dit_f32_rows.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bias_only_dit_hoist_rows.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bias_only_dit_rows.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bias_only_dit_train_rows.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_bwd_mid.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_bwd_tail.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_front_bf16.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_front_tf32.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_gemm_sm90.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_hoist_bwd.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_inf3_sm90.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_psb.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_psb32.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_pvdpb.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_tail2_bf16.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_tail2_tf32.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_tail_bf16.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_tail_tf32.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bo_wgrad.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/bwd_fused.py
src/miniworld_engine/kernels/bias_only_dit/cuda/cond_tables_sm100.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/dpb32_sm100.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/dpb_sm100.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/dpbx2_sm100.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/gemm_glu_tf32.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/gemm_resln_sm100.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/hoist.py
src/miniworld_engine/kernels/bias_only_dit/cuda/inf3_bf16.py
src/miniworld_engine/kernels/bias_only_dit/cuda/inf3_h100.py
src/miniworld_engine/kernels/bias_only_dit/cuda/pv_gate_inf.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/pv_gate_tf32.cu
src/miniworld_engine/kernels/bias_only_dit/cuda/runner.py
src/miniworld_engine/kernels/bias_only_dit/cuda/tf32.py
src/miniworld_engine/kernels/bias_only_dit/cuda/train.py
src/miniworld_engine/kernels/checks/__init__.py
src/miniworld_engine/kernels/checks/adaln.py
src/miniworld_engine/kernels/checks/augmented_attention.py
src/miniworld_engine/kernels/checks/bias_only_attention.py
src/miniworld_engine/kernels/checks/conditioned_transition.py
src/miniworld_engine/kernels/checks/gated_projection.py
src/miniworld_engine/kernels/checks/layernorm.py
src/miniworld_engine/kernels/checks/layernorm_linear.py
src/miniworld_engine/kernels/checks/mpnn_edge_dropout.py
src/miniworld_engine/kernels/checks/mpnn_edge_mlp.py
src/miniworld_engine/kernels/checks/mpnn_edge_tail.py
src/miniworld_engine/kernels/checks/mpnn_message.py
src/miniworld_engine/kernels/checks/mpnn_node_message.py
src/miniworld_engine/kernels/checks/mpnn_relative_position.py
src/miniworld_engine/kernels/checks/outer_product_mean.py
src/miniworld_engine/kernels/checks/pair_weighted_averaging.py
src/miniworld_engine/kernels/checks/rmsnorm.py
src/miniworld_engine/kernels/checks/rmsnorm_adamod.py
src/miniworld_engine/kernels/checks/rope.py
src/miniworld_engine/kernels/checks/swa_dit.py
src/miniworld_engine/kernels/checks/tm1.py
src/miniworld_engine/kernels/checks/tm2.py
src/miniworld_engine/kernels/checks/transition.py
src/miniworld_engine/kernels/checks/triangle_attention.py
src/miniworld_engine/kernels/checks/triangle_multiplication.py
src/miniworld_engine/kernels/checks/trimul_inproj.py
src/miniworld_engine/kernels/conditioned_transition/__init__.py
src/miniworld_engine/kernels/conditioned_transition/interface.py
src/miniworld_engine/kernels/conditioned_transition/reference.py
src/miniworld_engine/kernels/conditioned_transition/whole_op.py
src/miniworld_engine/kernels/conditioned_transition/cuda/__init__.py
src/miniworld_engine/kernels/conditioned_transition/cuda/gemm_swiglu.py
src/miniworld_engine/kernels/conditioned_transition/cuda/gemm_swiglu2_sm100.cu
src/miniworld_engine/kernels/conditioned_transition/cuda/gemm_swiglu_sm100.cu
src/miniworld_engine/kernels/conditioned_transition/cuda/gemm_swiglu_sm90.cu
src/miniworld_engine/kernels/conditioned_transition/cuda/token_dit_common.cuh
src/miniworld_engine/kernels/conditioned_transition/cuda/token_dit_rows.cu
src/miniworld_engine/kernels/conditioned_transition/cuda/token_dit_train_rows.cu
src/miniworld_engine/kernels/conditioned_transition/cuda/train.py
src/miniworld_engine/kernels/conditioned_transition/triton/__init__.py
src/miniworld_engine/kernels/conditioned_transition/triton/composed.py
src/miniworld_engine/kernels/conditioned_transition/triton/dispatch.py
src/miniworld_engine/kernels/conditioned_transition/triton/fwd_saveact.py
src/miniworld_engine/kernels/conditioned_transition/triton/inference.py
src/miniworld_engine/kernels/conditioned_transition/triton/token_dit_attn.py
src/miniworld_engine/kernels/conditioned_transition/triton/token_dit_kernels.py
src/miniworld_engine/kernels/conditioned_transition/triton/token_dit_runner.py
src/miniworld_engine/kernels/conditioned_transition/triton/training.py
src/miniworld_engine/kernels/drivers/__init__.py
src/miniworld_engine/kernels/drivers/adaln.py
src/miniworld_engine/kernels/drivers/augmented_attention.py
src/miniworld_engine/kernels/drivers/bias_only_attention.py
src/miniworld_engine/kernels/drivers/conditioned_transition.py
src/miniworld_engine/kernels/drivers/gated_projection.py
src/miniworld_engine/kernels/drivers/hopper.py
src/miniworld_engine/kernels/drivers/layernorm.py
src/miniworld_engine/kernels/drivers/layernorm_linear.py
src/miniworld_engine/kernels/drivers/mpnn_edge_dropout.py
src/miniworld_engine/kernels/drivers/mpnn_edge_mlp.py
src/miniworld_engine/kernels/drivers/mpnn_edge_tail.py
src/miniworld_engine/kernels/drivers/mpnn_message.py
src/miniworld_engine/kernels/drivers/mpnn_node_message.py
src/miniworld_engine/kernels/drivers/mpnn_relative_position.py
src/miniworld_engine/kernels/drivers/outer_product_mean.py
src/miniworld_engine/kernels/drivers/pair_weighted_averaging.py
src/miniworld_engine/kernels/drivers/rmsnorm.py
src/miniworld_engine/kernels/drivers/rmsnorm_adamod.py
src/miniworld_engine/kernels/drivers/rope.py
src/miniworld_engine/kernels/drivers/swa_dit.py
src/miniworld_engine/kernels/drivers/tm1.py
src/miniworld_engine/kernels/drivers/tm2.py
src/miniworld_engine/kernels/drivers/transition.py
src/miniworld_engine/kernels/drivers/triangle_attention.py
src/miniworld_engine/kernels/drivers/triangle_multiplication.py
src/miniworld_engine/kernels/drivers/trimul_inproj.py
src/miniworld_engine/kernels/gated_projection/__init__.py
src/miniworld_engine/kernels/gated_projection/interface.py
src/miniworld_engine/kernels/gated_projection/reference.py
src/miniworld_engine/kernels/gated_projection/whole_op.py
src/miniworld_engine/kernels/gated_projection/triton/__init__.py
src/miniworld_engine/kernels/gated_projection/triton/main.py
src/miniworld_engine/kernels/gated_projection/triton/residual.py
src/miniworld_engine/kernels/gated_projection/triton/swa.py
src/miniworld_engine/kernels/layernorm/__init__.py
src/miniworld_engine/kernels/layernorm/compile_native.py
src/miniworld_engine/kernels/layernorm/dispatch.py
src/miniworld_engine/kernels/layernorm/interface.py
src/miniworld_engine/kernels/layernorm/reference.py
src/miniworld_engine/kernels/layernorm/cuda/__init__.py
src/miniworld_engine/kernels/layernorm/cuda/layer_norm_cuda_kernel.cu
src/miniworld_engine/kernels/layernorm/cuda/setup.py
src/miniworld_engine/kernels/layernorm/cuda/general/__init__.py
src/miniworld_engine/kernels/layernorm/cuda/general/bind.cpp
src/miniworld_engine/kernels/layernorm/cuda/general/ln_bwd.cu
src/miniworld_engine/kernels/layernorm/cuda/general/ln_common.cuh
src/miniworld_engine/kernels/layernorm/cuda/general/ln_fwd.cu
src/miniworld_engine/kernels/layernorm/triton/__init__.py
src/miniworld_engine/kernels/layernorm/triton/lowreg.py
src/miniworld_engine/kernels/layernorm/triton/main.py
src/miniworld_engine/kernels/layernorm/triton/persistent.py
src/miniworld_engine/kernels/layernorm/triton/transpose.py
src/miniworld_engine/kernels/layernorm_linear/__init__.py
src/miniworld_engine/kernels/layernorm_linear/autograd.py
src/miniworld_engine/kernels/layernorm_linear/interface.py
src/miniworld_engine/kernels/layernorm_linear/reference.py
src/miniworld_engine/kernels/layernorm_linear/triton/__init__.py
src/miniworld_engine/kernels/layernorm_linear/triton/fused.py
src/miniworld_engine/kernels/layernorm_linear/triton/mmajor_bwd.py
src/miniworld_engine/kernels/layernorm_linear/triton/pair_bias.py
src/miniworld_engine/kernels/layernorm_linear/triton/recompute.py
src/miniworld_engine/kernels/layernorm_linear/triton/stats.py
src/miniworld_engine/kernels/layernorm_linear/triton/te_style.py
src/miniworld_engine/kernels/mpnn_edge_dropout/__init__.py
src/miniworld_engine/kernels/mpnn_edge_dropout/interface.py
src/miniworld_engine/kernels/mpnn_edge_dropout/reference.py
src/miniworld_engine/kernels/mpnn_edge_dropout/triton/__init__.py
src/miniworld_engine/kernels/mpnn_edge_dropout/triton/main.py
src/miniworld_engine/kernels/mpnn_edge_layernorm/__init__.py
src/miniworld_engine/kernels/mpnn_edge_layernorm/interface.py
src/miniworld_engine/kernels/mpnn_edge_layernorm/reference.py
src/miniworld_engine/kernels/mpnn_edge_layernorm/triton/__init__.py
src/miniworld_engine/kernels/mpnn_edge_layernorm/triton/main.py
src/miniworld_engine/kernels/mpnn_edge_mlp/__init__.py
src/miniworld_engine/kernels/mpnn_edge_mlp/interface.py
src/miniworld_engine/kernels/mpnn_edge_mlp/reference.py
src/miniworld_engine/kernels/mpnn_edge_mlp/triton/__init__.py
src/miniworld_engine/kernels/mpnn_edge_mlp/triton/compute.py
src/miniworld_engine/kernels/mpnn_edge_mlp/triton/main.py
src/miniworld_engine/kernels/mpnn_edge_tail/__init__.py
src/miniworld_engine/kernels/mpnn_edge_tail/interface.py
src/miniworld_engine/kernels/mpnn_edge_tail/reference.py
src/miniworld_engine/kernels/mpnn_edge_tail/triton/__init__.py
src/miniworld_engine/kernels/mpnn_edge_tail/triton/compute.py
src/miniworld_engine/kernels/mpnn_edge_tail/triton/main.py
src/miniworld_engine/kernels/mpnn_message/__init__.py
src/miniworld_engine/kernels/mpnn_message/interface.py
src/miniworld_engine/kernels/mpnn_message/reference.py
src/miniworld_engine/kernels/mpnn_message/triton/__init__.py
src/miniworld_engine/kernels/mpnn_message/triton/_policy.py
src/miniworld_engine/kernels/mpnn_message/triton/inference.py
src/miniworld_engine/kernels/mpnn_message/triton/main.py
src/miniworld_engine/kernels/mpnn_node_message/__init__.py
src/miniworld_engine/kernels/mpnn_node_message/interface.py
src/miniworld_engine/kernels/mpnn_node_message/reference.py
src/miniworld_engine/kernels/mpnn_node_message/triton/__init__.py
src/miniworld_engine/kernels/mpnn_node_message/triton/main.py
src/miniworld_engine/kernels/mpnn_relative_position/__init__.py
src/miniworld_engine/kernels/mpnn_relative_position/interface.py
src/miniworld_engine/kernels/mpnn_relative_position/reference.py
src/miniworld_engine/kernels/mpnn_relative_position/triton/__init__.py
src/miniworld_engine/kernels/mpnn_relative_position/triton/main.py
src/miniworld_engine/kernels/outer_product_mean/__init__.py
src/miniworld_engine/kernels/outer_product_mean/interface.py
src/miniworld_engine/kernels/outer_product_mean/reference.py
src/miniworld_engine/kernels/outer_product_mean/triton/__init__.py
src/miniworld_engine/kernels/outer_product_mean/triton/main.py
src/miniworld_engine/kernels/pair_weighted_averaging/__init__.py
src/miniworld_engine/kernels/pair_weighted_averaging/interface.py
src/miniworld_engine/kernels/pair_weighted_averaging/reference.py
src/miniworld_engine/kernels/pair_weighted_averaging/triton/__init__.py
src/miniworld_engine/kernels/pair_weighted_averaging/triton/main.py
src/miniworld_engine/kernels/registry/registry.csv
src/miniworld_engine/kernels/registry/registry_kernel.csv
src/miniworld_engine/kernels/registry/registry_kernel.units.json
src/miniworld_engine/kernels/registry/registry_module.csv
src/miniworld_engine/kernels/registry/tile_order_exempt.csv
src/miniworld_engine/kernels/registry/undeveloped.csv
src/miniworld_engine/kernels/registry/width_evidence.json
src/miniworld_engine/kernels/rmsnorm/__init__.py
src/miniworld_engine/kernels/rmsnorm/interface.py
src/miniworld_engine/kernels/rmsnorm/reference.py
src/miniworld_engine/kernels/rmsnorm/cuda/__init__.py
src/miniworld_engine/kernels/rmsnorm/triton/__init__.py
src/miniworld_engine/kernels/rmsnorm/triton/main.py
src/miniworld_engine/kernels/rope/__init__.py
src/miniworld_engine/kernels/rope/interface.py
src/miniworld_engine/kernels/rope/reference.py
src/miniworld_engine/kernels/rope/triton/__init__.py
src/miniworld_engine/kernels/rope/triton/main.py
src/miniworld_engine/kernels/rope/triton/qk_norm.py
src/miniworld_engine/kernels/smooth_lddt/__init__.py
src/miniworld_engine/kernels/smooth_lddt/interface.py
src/miniworld_engine/kernels/smooth_lddt/reference.py
src/miniworld_engine/kernels/smooth_lddt/cuda/__init__.py
src/miniworld_engine/kernels/smooth_lddt/cuda/sm100/__init__.py
src/miniworld_engine/kernels/smooth_lddt/cuda/sm100/smooth_lddt.cu
src/miniworld_engine/kernels/swa_dit/__init__.py
src/miniworld_engine/kernels/swa_dit/autograd.py
src/miniworld_engine/kernels/swa_dit/dispatch.py
src/miniworld_engine/kernels/swa_dit/interface.py
src/miniworld_engine/kernels/swa_dit/reference.py
src/miniworld_engine/kernels/swa_dit/cuda/__init__.py
src/miniworld_engine/kernels/swa_dit/cuda/loader.py
src/miniworld_engine/kernels/swa_dit/cuda/pdl.cuh
src/miniworld_engine/kernels/swa_dit/cuda/swa_attn.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_ffn_bwd.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_ffn_fwd.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_fp32.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_mod.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_mod_wg.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_pack.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_qkvg_fwd.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_slab.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_wgrad.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_wgrad_wg.cu
src/miniworld_engine/kernels/swa_dit/cuda/swa_wgrad_wg_f32.cu
src/miniworld_engine/kernels/swa_dit/cuda/gen/wgmma_bf16.inc
src/miniworld_engine/kernels/swa_dit/cuda/gen/wgmma_tf32.inc
src/miniworld_engine/kernels/swa_dit/cuda/sm100/__init__.py
src/miniworld_engine/kernels/swa_dit/cuda/sm100/attn_dkv.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/attn_dkvq.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/attn_dq.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/attn_fwd1h.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/attn_fwd3.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/attn_fwd_tf32.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/ffn_bwd_dy.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/ffn_bwd_gate.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/ffn_fwd2.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/ffn_fwd_tf32.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/mod_bwd.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/mod_fwd.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/mod_fwd_tf32.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/oproj_bwd.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/qkvg_bwd.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/qkvg_fwd.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/qkvg_fwd2.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/qkvg_fwd_tf32.cu
src/miniworld_engine/kernels/swa_dit/cuda/sm100/sm100.cuh
src/miniworld_engine/kernels/swa_dit/cuda/sm100/tf32_fwd.py
src/miniworld_engine/kernels/swa_dit/cuda/sm100/wprep_tf32.cu
src/miniworld_engine/kernels/swa_dit/triton/__init__.py
src/miniworld_engine/kernels/swa_dit/triton/backward.py
src/miniworld_engine/kernels/swa_dit/triton/backward_fp32.py
src/miniworld_engine/kernels/swa_dit/triton/forward.py
src/miniworld_engine/kernels/swa_dit/triton/forward_fp32.py
src/miniworld_engine/kernels/tm1/__init__.py
src/miniworld_engine/kernels/tm1/interface.py
src/miniworld_engine/kernels/tm1/reference.py
src/miniworld_engine/kernels/tm1/triton/__init__.py
src/miniworld_engine/kernels/tm1/triton/main.py
src/miniworld_engine/kernels/tm2/__init__.py
src/miniworld_engine/kernels/tm2/interface.py
src/miniworld_engine/kernels/tm2/reference.py
src/miniworld_engine/kernels/tm2/triton/__init__.py
src/miniworld_engine/kernels/tm2/triton/main.py
src/miniworld_engine/kernels/token_pair_init/__init__.py
src/miniworld_engine/kernels/token_pair_init/interface.py
src/miniworld_engine/kernels/token_pair_init/reference.py
src/miniworld_engine/kernels/token_pair_init/cuda/__init__.py
src/miniworld_engine/kernels/token_pair_init/cuda/sm100/__init__.py
src/miniworld_engine/kernels/token_pair_init/cuda/sm100/token_pair_init.cu
src/miniworld_engine/kernels/token_pair_init/cuda/sm90/__init__.py
src/miniworld_engine/kernels/token_pair_init/cuda/sm90/token_pair_init.cu
src/miniworld_engine/kernels/transition/__init__.py
src/miniworld_engine/kernels/transition/interface.py
src/miniworld_engine/kernels/transition/reference.py
src/miniworld_engine/kernels/transition/whole_op.py
src/miniworld_engine/kernels/transition/cuda/__init__.py
src/miniworld_engine/kernels/transition/cuda/_weights.py
src/miniworld_engine/kernels/transition/cuda/fused_generic_sm90a.py
src/miniworld_engine/kernels/transition/cuda/fused_sm100a.py
src/miniworld_engine/kernels/transition/cuda/fused_sm80.py
src/miniworld_engine/kernels/transition/cuda/fused_sm90a.py
src/miniworld_engine/kernels/transition/cuda/fused_wide_sm100a.py
src/miniworld_engine/kernels/transition/cuda/fused_wide_sm90a.py
src/miniworld_engine/kernels/transition/cuda/setup.py
src/miniworld_engine/kernels/transition/cuda/transition_b2b_kernel.cu
src/miniworld_engine/kernels/transition/cuda/transition_cuda.cpp
src/miniworld_engine/kernels/transition/cuda/transition_cuda_kernel.cu
src/miniworld_engine/kernels/transition/cuda/transition_expand_gate_kernel.cu
src/miniworld_engine/kernels/transition/cuda/transition_fused_bwd_sm90a_kernel.cu
src/miniworld_engine/kernels/transition/cuda/transition_fused_fwd_sm90a_kernel.cu
src/miniworld_engine/kernels/transition/cuda/transition_fused_sm90a.cu
src/miniworld_engine/kernels/transition/cuda/transition_gatebwd_kernel.cu
src/miniworld_engine/kernels/transition/cuda/anthropic_v5/tmn_kernels.cuh
src/miniworld_engine/kernels/transition/cuda/anthropic_v5/tmn_ptx.cuh
src/miniworld_engine/kernels/transition/cuda/anthropic_v5/common/tmn_math.cuh
src/miniworld_engine/kernels/transition/cuda/generic_sm90a/bind.cu
src/miniworld_engine/kernels/transition/cuda/generic_sm90a/gen_wgmma.py
src/miniworld_engine/kernels/transition/cuda/generic_sm90a/tr90.cuh
src/miniworld_engine/kernels/transition/cuda/generic_sm90a/wgmma_ops.cuh
src/miniworld_engine/kernels/transition/cuda/sm100/sm100.cuh
src/miniworld_engine/kernels/transition/cuda/sm100/tr_bwd_sm100.cu
src/miniworld_engine/kernels/transition/cuda/sm100/tr_fwd_sm100.cu
src/miniworld_engine/kernels/transition/cuda/sm100/transition_sm100.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tbwd_d384.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tbwd_d64.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tfwd_d256.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tfwd_d384.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tfwd_d64.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tgate_ab.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tgate_w.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tgemm_nd.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tln_w.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tlnbwd_w.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tsmall_w.cu
src/miniworld_engine/kernels/transition/cuda/sm100/widths/tswiglu_w.cu
src/miniworld_engine/kernels/transition/cuda/sm80/sm80_common.cuh
src/miniworld_engine/kernels/transition/cuda/sm80/tr_bwd_dw_sm80.cuh
src/miniworld_engine/kernels/transition/cuda/sm80/tr_bwd_pw_sm80.cuh
src/miniworld_engine/kernels/transition/cuda/sm80/tr_bwd_sm80.cuh
src/miniworld_engine/kernels/transition/cuda/sm80/tr_bwd_w_sm80.cuh
src/miniworld_engine/kernels/transition/cuda/sm80/tr_fwd_sm80.cuh
src/miniworld_engine/kernels/transition/cuda/sm80/transition_sm80.cu
src/miniworld_engine/kernels/transition/cuda/wide/bind.cu
src/miniworld_engine/kernels/transition/cuda/wide/d256_fwd.cu
src/miniworld_engine/kernels/transition/cuda/wide/d64_bwd.cu
src/miniworld_engine/kernels/transition/cuda/wide/d64_fwd.cu
src/miniworld_engine/kernels/transition/cuda/wide/d64_fwd_h32x4.cu
src/miniworld_engine/kernels/transition/cuda/wide/d64_fwd_h32x5.cu
src/miniworld_engine/kernels/transition/cuda/wide/d64_fwd_small.cu
src/miniworld_engine/kernels/transition/cuda/wide/d64_fwd_small3.cu
src/miniworld_engine/kernels/transition/cuda/wide/d64_fwd_small_f2.cu
src/miniworld_engine/kernels/transition/cuda/wide/d64_fwd_small_f3.cu
src/miniworld_engine/kernels/transition/cuda/wide/dxln.cu
src/miniworld_engine/kernels/transition/cuda/wide/gate.cu
src/miniworld_engine/kernels/transition/cuda/wide/gate_noh.cu
src/miniworld_engine/kernels/transition/cuda/wide/launch_util.cuh
src/miniworld_engine/kernels/transition/cuda/wide/lnsg.cu
src/miniworld_engine/kernels/transition/cuda/wide/squeeze.cu
src/miniworld_engine/kernels/transition/cuda/wide/kernels/dxn_lnbwd.cu
src/miniworld_engine/kernels/transition/cuda/wide/kernels/gate_gemm2.cu
src/miniworld_engine/kernels/transition/cuda/wide/kernels/ln_swiglu_gemm.cu
src/miniworld_engine/kernels/transition/cuda/wide/kernels/squeeze_gemm.cu
src/miniworld_engine/kernels/transition/cuda/wide/kernels/transition_bwd_d64.cu
src/miniworld_engine/kernels/transition/cuda/wide/kernels/transition_fwd_d256.cu
src/miniworld_engine/kernels/transition/cuda/wide/kernels/transition_fwd_d64.cu
src/miniworld_engine/kernels/transition/cuda/wide/kernels/wgmma256rs.cuh
src/miniworld_engine/kernels/transition/cuda/wide/kernels/wgmma_n.cuh
src/miniworld_engine/kernels/transition/triton/__init__.py
src/miniworld_engine/kernels/transition/triton/fused.py
src/miniworld_engine/kernels/transition/triton/main.py
src/miniworld_engine/kernels/transition/triton/residual.py
src/miniworld_engine/kernels/triangle_attention/__init__.py
src/miniworld_engine/kernels/triangle_attention/interface.py
src/miniworld_engine/kernels/triangle_attention/reference.py
src/miniworld_engine/kernels/triangle_attention/whole_op.py
src/miniworld_engine/kernels/triangle_attention/cuda/__init__.py
src/miniworld_engine/kernels/triangle_attention/cuda/b200_triattn.py
src/miniworld_engine/kernels/triangle_attention/cuda/bias_backward.py
src/miniworld_engine/kernels/triangle_attention/cuda/bias_fusion.cu
src/miniworld_engine/kernels/triangle_attention/cuda/bias_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/build_bias.py
src/miniworld_engine/kernels/triangle_attention/cuda/build_dq.py
src/miniworld_engine/kernels/triangle_attention/cuda/build_gate.py
src/miniworld_engine/kernels/triangle_attention/cuda/build_ln.py
src/miniworld_engine/kernels/triangle_attention/cuda/build_native.py
src/miniworld_engine/kernels/triangle_attention/cuda/build_wgrad.py
src/miniworld_engine/kernels/triangle_attention/cuda/dq.cu
src/miniworld_engine/kernels/triangle_attention/cuda/dq_backward.py
src/miniworld_engine/kernels/triangle_attention/cuda/dq_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/fa3_utils.h
src/miniworld_engine/kernels/triangle_attention/cuda/fwd_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/gate_backward.py
src/miniworld_engine/kernels/triangle_attention/cuda/gate_delta.cu
src/miniworld_engine/kernels/triangle_attention/cuda/gate_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sm90.py
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sm90_generic.py
src/miniworld_engine/kernels/triangle_attention/cuda/h100_triattn.py
src/miniworld_engine/kernels/triangle_attention/cuda/ln_backward.py
src/miniworld_engine/kernels/triangle_attention/cuda/ln_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/ln_residual.cu
src/miniworld_engine/kernels/triangle_attention/cuda/manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/projection_dgrad.cu
src/miniworld_engine/kernels/triangle_attention/cuda/q_fwd_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/q_projection_attention.cu
src/miniworld_engine/kernels/triangle_attention/cuda/q_projection_attention.py
src/miniworld_engine/kernels/triangle_attention/cuda/qg_fwd_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/qg_projection_attention.cu
src/miniworld_engine/kernels/triangle_attention/cuda/qg_projection_attention.py
src/miniworld_engine/kernels/triangle_attention/cuda/qkv_fwd_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/qkv_projection_attention.cu
src/miniworld_engine/kernels/triangle_attention/cuda/qkv_projection_attention.py
src/miniworld_engine/kernels/triangle_attention/cuda/training_forward.cu
src/miniworld_engine/kernels/triangle_attention/cuda/training_forward.py
src/miniworld_engine/kernels/triangle_attention/cuda/wgrad.cu
src/miniworld_engine/kernels/triangle_attention/cuda/wgrad_backward.py
src/miniworld_engine/kernels/triangle_attention/cuda/wgrad_manifest.json
src/miniworld_engine/kernels/triangle_attention/cuda/b200_sources/sm100.cuh
src/miniworld_engine/kernels/triangle_attention/cuda/b200_sources/triattn_mod_sm100.cu
src/miniworld_engine/kernels/triangle_attention/cuda/b200_sources/triattn_sm100.cu
src/miniworld_engine/kernels/triangle_attention/cuda/b200_sources/triattn_wide_sm100.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/bias_fusion.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/dq.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/fa3_utils.h
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/front_fused2.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/front_fused3.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/front_fused3_sm90.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/front_grouped.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/module.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/pack.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/sm90.cuh
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/training_forward.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/triattn_bwd_sm90.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/triattn_generic_sm90.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/triattn_mod_sm90.cu
src/miniworld_engine/kernels/triangle_attention/cuda/h100_sources/triattn_sm90.cu
src/miniworld_engine/kernels/triangle_attention/triton/__init__.py
src/miniworld_engine/kernels/triangle_attention/triton/atomic.py
src/miniworld_engine/kernels/triangle_attention/triton/bidirectional.py
src/miniworld_engine/kernels/triangle_attention/triton/main.py
src/miniworld_engine/kernels/trimul_inproj/__init__.py
src/miniworld_engine/kernels/trimul_inproj/autograd.py
src/miniworld_engine/kernels/trimul_inproj/interface.py
src/miniworld_engine/kernels/trimul_inproj/reference.py
src/miniworld_engine/kernels/trimul_inproj/whole_op.py
src/miniworld_engine/kernels/trimul_inproj/cuda/__init__.py
src/miniworld_engine/kernels/trimul_inproj/cuda/_h100_bmm.py
src/miniworld_engine/kernels/trimul_inproj/cuda/_h100_driver.py
src/miniworld_engine/kernels/trimul_inproj/cuda/_h100_infer_kernel.py
src/miniworld_engine/kernels/trimul_inproj/cuda/_h100_launch.py
src/miniworld_engine/kernels/trimul_inproj/cuda/_h100_manifest.py
src/miniworld_engine/kernels/trimul_inproj/cuda/_h100_pack.py
src/miniworld_engine/kernels/trimul_inproj/cuda/_h100_runtime.py
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_infer.py
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_train.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b1.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_aux.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_d384_tmasave.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_direct.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_dpl_pipeline.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_emitxn_tma.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_gemm.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_k1_prefetch.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_reorder.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_scalar_tmasave.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_ss.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_ss_regtune.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_ss_saved.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_static.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b200style_tmasave.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_b7.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_d64_b200gate.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_inference.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_native.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_output.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_single.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_single_b7.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_single_output.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_training.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_uni_d64_b200gate.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_uni_wide_inference.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_uni_wide_training.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_wide_inference.py
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_wide_training.py
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/b1g.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/b1s.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/b7g.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/b7m.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/bind_b200.cpp
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/k1w.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/k1wb.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/k3g.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/k3w.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/sm100.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/tmap.h
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/wide_aux.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/b200_sources/wide_bwd.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/PROVENANCE.json
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/b1_fused.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/b1_lowreg.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/b1_pipeline_math.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/b1_stream_ln.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/common_recompute.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/configs.json
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/ln_recompute.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/lowreg_stats.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b1/xhat_read.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k1ss.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k1ssregtune.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k1sssave.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k1wb_epilogue.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k3.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k3d384_tmasave.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k3reorder.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k3scalar_tmasave.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b200style/k3tmasave.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b7_384/joint.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b7_384/single_wg.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b7_768/joint.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/b7_768/single_wg.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/common/front_mn_primitives.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/common/front_primitives.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/common/warp_primitives.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/common_b7/common_recompute.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/common_b7/ln_recompute.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/common_b7/single_wg.inc
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/d64_train/d64_bwd_wg_b7gate.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/d64_train/d64_bwd_wg_gatepipe.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/d64_train/d64_wg.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/d64_train/save_k3_d64.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/d64_train/uni_bwd_wg_affinereg.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/d64_train/uni_bwd_wg_b7gate.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/d64_train/uni_bwd_wg_gatepipe.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/inference/defines.json
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/inference/tmn_k3_wide.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/inference/tmn_kernels.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/inference/tmn_ptx.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/inference/tmn_sm90_unit.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/inference/common/tmn_math.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/output/original_k1.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/output/save_k3.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/single_b7/joint.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/single_output/kernel.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/single_output/selection.json
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/uni_wide/front_shared.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/uni_wide/output.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/uni_wide/selection.json
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/upstream/csrc/tmn_kernels.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/upstream/csrc/tmn_ptx.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/upstream/csrc/common/tmn_math.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_forward/front_shared.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_forward/output.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_forward/output_stream.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_forward/selection.json
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_forward/wgmma_n.cuh
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/compact_projection_d512.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/compact_projection_fallback_d512.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/contract_d256_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/contract_gp_d384_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/contract_gp_d512_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/contract_gp_l768.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/delta_norm_d512.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/delta_norm_fallback_d512.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/forward_norm_d384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/front_d256.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/front_pre_d384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/front_pre_d512.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/front_pre_ss.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/gate.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/gate_affine.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/input_ln_d256_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/input_ln_d384_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/input_ln_d512_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/input_reduce_d512_l768.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/input_reduce_l768.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/ln_in_d512.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/lt_selection.json
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/output_dense.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/output_ln_d256_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/output_ln_d256_l768.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/output_ln_d384_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/output_ln_d384_l768.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/output_ln_d512_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/output_ln_d512_l768.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/source_d256_l384.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/source_d256_l768.cu
src/miniworld_engine/kernels/trimul_inproj/cuda/h100_sources/wide_train/weight_reduce_d512_l768.cu
src/miniworld_engine/kernels/trimul_inproj/triton/__init__.py
src/miniworld_engine/kernels/trimul_inproj/triton/_autotune.py
src/miniworld_engine/kernels/trimul_inproj/triton/back.py
src/miniworld_engine/kernels/trimul_inproj/triton/back_fused.py
src/miniworld_engine/kernels/trimul_inproj/triton/backward_fused.py
src/miniworld_engine/kernels/trimul_inproj/triton/bidirectional.py
src/miniworld_engine/kernels/trimul_inproj/triton/contract.py
src/miniworld_engine/kernels/trimul_inproj/triton/gate_elem.py
src/miniworld_engine/kernels/trimul_inproj/triton/output_fused.py
src/miniworld_engine/kernels/trimul_inproj/triton/unidirectional.py
src/miniworld_engine/modules/__init__.py
src/miniworld_engine/modules/dispatch.py
src/miniworld_engine/modules/exceptions.py
src/miniworld_engine/modules/functional.py
src/miniworld_engine/modules/primitives.py
src/miniworld_engine/modules/adaptive_layernorm/__init__.py
src/miniworld_engine/modules/adaptive_layernorm/module.py
src/miniworld_engine/modules/attention_pair_bias/__init__.py
src/miniworld_engine/modules/attention_pair_bias/module.py
src/miniworld_engine/modules/augmented_attention/__init__.py
src/miniworld_engine/modules/augmented_attention/module.py
src/miniworld_engine/modules/bias_only_dit/__init__.py
src/miniworld_engine/modules/bias_only_dit/hoist.py
src/miniworld_engine/modules/bias_only_dit/module.py
src/miniworld_engine/modules/conditioned_transition/__init__.py
src/miniworld_engine/modules/conditioned_transition/module.py
src/miniworld_engine/modules/dit/__init__.py
src/miniworld_engine/modules/dit/module.py
src/miniworld_engine/modules/input_feature_embedder/__init__.py
src/miniworld_engine/modules/input_feature_embedder/module.py
src/miniworld_engine/modules/local_dit/__init__.py
src/miniworld_engine/modules/local_dit/module.py
src/miniworld_engine/modules/mpnn/__init__.py
src/miniworld_engine/modules/mpnn/_functional.py
src/miniworld_engine/modules/mpnn/batch.py
src/miniworld_engine/modules/mpnn/conversion.py
src/miniworld_engine/modules/mpnn/data.py
src/miniworld_engine/modules/mpnn/dropout.py
src/miniworld_engine/modules/mpnn/features.py
src/miniworld_engine/modules/mpnn/layers.py
src/miniworld_engine/modules/mpnn/legacy.py
src/miniworld_engine/modules/mpnn/loss.py
src/miniworld_engine/modules/mpnn/masking.py
src/miniworld_engine/modules/mpnn/module.py
src/miniworld_engine/modules/mpnn/naive.py
src/miniworld_engine/modules/msa_pair_weighted_averaging/__init__.py
src/miniworld_engine/modules/msa_pair_weighted_averaging/module.py
src/miniworld_engine/modules/outer_product/__init__.py
src/miniworld_engine/modules/outer_product/module.py
src/miniworld_engine/modules/pairformer/__init__.py
src/miniworld_engine/modules/pairformer/graph_runner.py
src/miniworld_engine/modules/pairformer/module.py
src/miniworld_engine/modules/swa_atom_attention/__init__.py
src/miniworld_engine/modules/swa_atom_attention/module.py
src/miniworld_engine/modules/swa_dit/__init__.py
src/miniworld_engine/modules/swa_dit/module.py
src/miniworld_engine/modules/transition/__init__.py
src/miniworld_engine/modules/transition/module.py
src/miniworld_engine/modules/triangle_attention/__init__.py
src/miniworld_engine/modules/triangle_attention/bidirectional.py
src/miniworld_engine/modules/triangle_attention/module.py
src/miniworld_engine/modules/triangle_multiplication/__init__.py
src/miniworld_engine/modules/triangle_multiplication/baseline_dtv1.py
src/miniworld_engine/modules/triangle_multiplication/baseline_dtv1_bidir.py
src/miniworld_engine/modules/triangle_multiplication/bidirectional.py
src/miniworld_engine/modules/triangle_multiplication/module.py
src/miniworld_engine/modules/triangle_multiplication/reference.py
src/miniworld_engine/ops/__init__.py
src/miniworld_engine/tools/__init__.py
src/miniworld_engine/tools/audit_thresholds.py
src/miniworld_engine/tools/import_anthropic.py
src/miniworld_engine/tools/make_default_configs.py
src/miniworld_engine/viz/__init__.py
src/miniworld_engine/viz/kernel_flow.py
src/miniworld_engine/viz/measure_bars.py
src/miniworld_engine/viz/style.py
src/miniworld_engine/viz/sweep_page.py