├── .gitignore ├── Easy-Tutorials ├── 01-esay-example.ipynb ├── 02-triton-max.ipynb ├── 03-martix-mul.ipynb ├── 04-fused-apply-rope.ipynb ├── 05-RMSNorm.ipynb ├── 06-fused-add-norm.ipynb ├── 07-fused-mul-silu效果不好.ipynb ├── 08-fused-silu.ipynb ├── 09-test-kernel-speed.ipynb ├── 10-softmax.ipynb ├── README.md ├── convert │ └── glm4_to_qwen2_format.py ├── imgs │ ├── example.png │ ├── llama3-1B.png │ ├── qwen2-0.5B.png │ └── speed.png ├── train.py ├── train.sh └── train_model │ ├── __pycache__ │ └── utils.cpython-310.pyc │ ├── ds_config.json │ ├── logs │ ├── llama3-1B-with-kernel-pretrain │ │ ├── log_history.json │ │ ├── log_history_22.json │ │ └── log_history_23.json │ ├── llama3-1B-with-kernel-unsloth-loss-pretrain │ │ ├── log_history.json │ │ ├── log_history_24.json │ │ └── log_history_25.json │ ├── llama3-1B-without-kernel-pretrain │ │ └── log_history.json │ ├── qwen2-0.5B-with-kernel-pretrain-bf16-loss │ │ └── log_history.json │ ├── qwen2-0.5B-with-kernel-sft │ │ └── log_history.json │ ├── qwen2-0.5B-with-kernel-unsloth-loss-pretrain │ │ ├── log_history.json │ │ └── log_history_24.json │ ├── qwen2-0.5B-with-kernel │ │ ├── log_history.json │ │ ├── log_history_22.json │ │ ├── log_history_23.json │ │ └── log_history_old.json │ ├── qwen2-0.5B-without-kernel-sft │ │ └── log_history.json │ ├── qwen2-0.5B-without-kernel │ │ ├── log_history.json │ │ └── log_history_old.json │ ├── qwen2-7B-with-kernel-sft │ │ └── log_history.json │ ├── qwen2-7B-with-kernel │ │ └── log_history.json │ ├── qwen2-7B-without-kernel-sft │ │ └── log_history.json │ └── qwen2-7B-without-kernel │ │ └── log_history.json │ ├── plot.ipynb │ ├── utils.py │ └── zero2.json ├── README.md ├── mdy_triton ├── __init__.py ├── __pycache__ │ ├── __init__.cpython-310.pyc │ └── __init__.cpython-312.pyc ├── core │ ├── __init__.py │ ├── __pycache__ │ │ ├── __init__.cpython-310.pyc │ │ ├── __init__.cpython-312.pyc │ │ ├── cross_entyopy_loss.cpython-310.pyc │ │ ├── cross_entyopy_loss.cpython-312.pyc │ │ ├── fused_add_norm.cpython-310.pyc │ │ ├── fused_add_norm.cpython-312.pyc │ │ ├── fused_apply_rope.cpython-310.pyc │ │ ├── fused_apply_rope.cpython-312.pyc │ │ ├── fused_silu.cpython-310.pyc │ │ ├── fused_silu.cpython-312.pyc │ │ ├── rmsnorm.cpython-310.pyc │ │ ├── rmsnorm.cpython-312.pyc │ │ ├── triton_maxmin.cpython-310.pyc │ │ └── triton_maxmin.cpython-312.pyc │ ├── cross_entyopy_loss.py │ ├── fused_add_norm.py │ ├── fused_apply_rope.py │ ├── fused_silu.py │ ├── rmsnorm.py │ └── triton_maxmin.py └── replace_kernel │ ├── __init__.py │ ├── __pycache__ │ ├── __init__.cpython-310.pyc │ ├── glm4.cpython-310.pyc │ ├── llama.cpython-310.pyc │ ├── llama3.cpython-310.pyc │ ├── mistral.cpython-310.pyc │ ├── qwen2.cpython-310.pyc │ └── qwen2_vl.cpython-310.pyc │ ├── glm4.py │ ├── llama.py │ ├── mistral.py │ ├── phi3.py │ ├── qwen2.py │ └── qwen2_vl.py ├── setup.py └── triton_kernels ├── __pycache__ ├── causal_conv1d.cpython-310.pyc ├── causal_conv1d_triton.cpython-310.pyc ├── replace_conv1d_cuda.cpython-310.pyc └── triton_causal_conv1d.cpython-310.pyc ├── causal_conv1d ├── __init__.py ├── __pycache__ │ ├── causal_conv1d.cpython-310.pyc │ └── triton_causal_conv1d.cpython-310.pyc ├── causal_conv1d.ipynb ├── replace_conv1d_cuda.py └── triton_causal_conv1d.py ├── cross_entropy ├── test_ce.py ├── triton_ce.py └── triton_cross_entropy.ipynb ├── deepseekv3 ├── README.md ├── TritonAdam.py ├── __pycache__ │ ├── TritonAdam.cpython-310.pyc │ ├── bf16mv_optimizer.cpython-310.pyc │ ├── fp8_kernel.cpython-310.pyc │ ├── fp8_module.cpython-310.pyc │ ├── kernel.cpython-310.pyc │ ├── kernel.cpython-312.pyc │ ├── new_fp8_module.cpython-310.pyc │ ├── new_fp8_module.cpython-312.pyc │ ├── offical_kernel.cpython-310.pyc │ ├── offical_matmul.cpython-310.pyc │ └── offical_matmul.cpython-312.pyc ├── fp8-benchmark.ipynb ├── fp8_kernel.py ├── imgs │ ├── image.png │ ├── image1.png │ └── image2.png ├── offical_kernel.py └── optmizer.ipynb ├── dpsk_fp8linear ├── __pycache__ │ └── fp8.cpython-310.pyc ├── fp8.py └── test_fp8.ipynb ├── dyt ├── __pycache__ │ ├── dyt.cpython-310.pyc │ ├── dyt.cpython-312.pyc │ ├── dyt2.cpython-312.pyc │ ├── rmsnorm.cpython-310.pyc │ └── rmsnorm.cpython-312.pyc ├── dyt.py ├── dyt2.py ├── rmsnorm.py └── test_dyt.ipynb ├── flash_attn_triton3.4 ├── test_hooper.ipynb └── triton_flash_attn.py ├── flash_muon ├── README.md ├── image.png ├── test.ipynb └── triton_flash_muon.py ├── grpo ├── README.md ├── __pycache__ │ └── replace_grpo_trainer.cpython-310.pyc ├── build │ └── lib │ │ └── triton_grpo_loss │ │ └── run_bs_one_by_one.py ├── grpo_loss.ipynb ├── grpo_loss_trl==0.16.ipynb ├── grpo_loss_trl==0.16.py ├── imgs │ ├── clamp.jpg │ ├── memory.jpg │ ├── reward.jpg │ └── triton_code.jpg ├── infos.jsonl ├── replace_grpo_trainer_trl==0.14.py ├── replace_grpo_trainer_trl==0.15.py ├── setup.py ├── test_memory.py └── triton_grpo_loss │ ├── __init__.py │ ├── __pycache__ │ ├── __init__.cpython-310.pyc │ ├── __init__.cpython-312.pyc │ ├── core.cpython-310.pyc │ └── core.cpython-312.pyc │ ├── core.py │ ├── decouple_logp_and_loss.py │ └── run_bs_one_by_one.py ├── lighting_attention ├── __init__.py ├── lighting_attention.ipynb └── lighting_attention.py ├── megatron_patchs ├── README.md ├── __init__.py ├── __pycache__ │ ├── __init__.cpython-310.pyc │ ├── __init__.cpython-312.pyc │ ├── add_kernel.cpython-310.pyc │ ├── add_kernel.cpython-312.pyc │ ├── bf16mv_adam.cpython-310.pyc │ ├── bf16mv_adam.cpython-312.pyc │ ├── comm.cpython-310.pyc │ ├── comm.cpython-312.pyc │ ├── cross_entropy_losss.cpython-310.pyc │ ├── cross_entropy_losss.cpython-312.pyc │ ├── dpsk_grouped_linear.cpython-310.pyc │ ├── dpsk_grouped_linear.cpython-312.pyc │ ├── dpsk_layernorm_linear.cpython-310.pyc │ ├── dpsk_layernorm_linear.cpython-312.pyc │ ├── dpsk_linear.cpython-310.pyc │ ├── dpsk_linear.cpython-312.pyc │ ├── fp8_matmul.cpython-310.pyc │ ├── fp8_matmul.cpython-312.pyc │ ├── manger.cpython-310.pyc │ ├── manger.cpython-312.pyc │ ├── swiglu.cpython-310.pyc │ ├── swiglu.cpython-312.pyc │ ├── zb2p.cpython-310.pyc │ └── zb2p.cpython-312.pyc ├── add_kernel.py ├── bf16mv_adam.py ├── comm.py ├── cross_entropy_losss.py ├── dpsk_grouped_linear.py ├── dpsk_layernorm_linear.py ├── dpsk_linear.py ├── fp8_matmul.py ├── image.png ├── manger.py ├── swiglu.py └── zb2p.py ├── mla_attention ├── __pycache__ │ └── mla.cpython-312.pyc ├── mla.ipynb ├── mla.py └── test.ipynb └── nsa_attention ├── __pycache__ ├── combine.cpython-312.pyc ├── compress_attn.cpython-312.pyc ├── compress_attn_v2.cpython-312.pyc ├── compresss_attn_v2.cpython-312.pyc ├── compute_attn.cpython-312.pyc ├── exp_family.cpython-312.pyc ├── nas_attn.cpython-312.pyc ├── nsa_attn.cpython-312.pyc ├── pku_nsa.cpython-312.pyc ├── select_attn.cpython-312.pyc ├── select_attn_v2.cpython-312.pyc ├── select_attn_v3.cpython-312.pyc ├── torch_code.cpython-312.pyc └── triton_flash_attn2.cpython-312.pyc ├── combine.py ├── compress_attn.py ├── compress_attn_v2.py ├── exp_family.py ├── nsa_attn.py ├── pku_nsa.py ├── select_attn.py ├── select_attn_v2.py ├── select_attn_v3.py ├── torch_code.py ├── triton_flash_attn2.py ├── 定向爆破.ipynb └── 精度和性能测试.ipynb /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/.gitignore -------------------------------------------------------------------------------- /Easy-Tutorials/01-esay-example.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/01-esay-example.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/02-triton-max.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/02-triton-max.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/03-martix-mul.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/03-martix-mul.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/04-fused-apply-rope.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/04-fused-apply-rope.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/05-RMSNorm.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/05-RMSNorm.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/06-fused-add-norm.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/06-fused-add-norm.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/07-fused-mul-silu效果不好.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/07-fused-mul-silu效果不好.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/08-fused-silu.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/08-fused-silu.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/09-test-kernel-speed.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/09-test-kernel-speed.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/10-softmax.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/10-softmax.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/README.md -------------------------------------------------------------------------------- /Easy-Tutorials/convert/glm4_to_qwen2_format.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/convert/glm4_to_qwen2_format.py -------------------------------------------------------------------------------- /Easy-Tutorials/imgs/example.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/imgs/example.png -------------------------------------------------------------------------------- /Easy-Tutorials/imgs/llama3-1B.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/imgs/llama3-1B.png -------------------------------------------------------------------------------- /Easy-Tutorials/imgs/qwen2-0.5B.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/imgs/qwen2-0.5B.png -------------------------------------------------------------------------------- /Easy-Tutorials/imgs/speed.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/imgs/speed.png -------------------------------------------------------------------------------- /Easy-Tutorials/train.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train.py -------------------------------------------------------------------------------- /Easy-Tutorials/train.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train.sh -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/__pycache__/utils.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/__pycache__/utils.cpython-310.pyc -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/ds_config.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/ds_config.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-pretrain/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-pretrain/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-pretrain/log_history_22.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-pretrain/log_history_22.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-pretrain/log_history_23.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-pretrain/log_history_23.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-unsloth-loss-pretrain/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-unsloth-loss-pretrain/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-unsloth-loss-pretrain/log_history_24.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-unsloth-loss-pretrain/log_history_24.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-unsloth-loss-pretrain/log_history_25.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/llama3-1B-with-kernel-unsloth-loss-pretrain/log_history_25.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/llama3-1B-without-kernel-pretrain/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/llama3-1B-without-kernel-pretrain/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel-pretrain-bf16-loss/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel-pretrain-bf16-loss/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel-sft/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel-sft/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel-unsloth-loss-pretrain/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel-unsloth-loss-pretrain/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel-unsloth-loss-pretrain/log_history_24.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel-unsloth-loss-pretrain/log_history_24.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel/log_history_22.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel/log_history_22.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel/log_history_23.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel/log_history_23.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel/log_history_old.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-with-kernel/log_history_old.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-without-kernel-sft/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-without-kernel-sft/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-without-kernel/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-without-kernel/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-0.5B-without-kernel/log_history_old.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-0.5B-without-kernel/log_history_old.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-7B-with-kernel-sft/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-7B-with-kernel-sft/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-7B-with-kernel/log_history.json: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-7B-without-kernel-sft/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-7B-without-kernel-sft/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/logs/qwen2-7B-without-kernel/log_history.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/logs/qwen2-7B-without-kernel/log_history.json -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/plot.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/plot.ipynb -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/utils.py -------------------------------------------------------------------------------- /Easy-Tutorials/train_model/zero2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/Easy-Tutorials/train_model/zero2.json -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/README.md -------------------------------------------------------------------------------- /mdy_triton/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mdy_triton/__pycache__/__init__.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/__pycache__/__init__.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/__pycache__/__init__.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/__pycache__/__init__.cpython-312.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__init__.py -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/__init__.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/__init__.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/__init__.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/__init__.cpython-312.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/cross_entyopy_loss.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/cross_entyopy_loss.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/cross_entyopy_loss.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/cross_entyopy_loss.cpython-312.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/fused_add_norm.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/fused_add_norm.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/fused_add_norm.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/fused_add_norm.cpython-312.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/fused_apply_rope.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/fused_apply_rope.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/fused_apply_rope.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/fused_apply_rope.cpython-312.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/fused_silu.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/fused_silu.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/fused_silu.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/fused_silu.cpython-312.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/rmsnorm.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/rmsnorm.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/rmsnorm.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/rmsnorm.cpython-312.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/triton_maxmin.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/triton_maxmin.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/core/__pycache__/triton_maxmin.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/__pycache__/triton_maxmin.cpython-312.pyc -------------------------------------------------------------------------------- /mdy_triton/core/cross_entyopy_loss.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/cross_entyopy_loss.py -------------------------------------------------------------------------------- /mdy_triton/core/fused_add_norm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/fused_add_norm.py -------------------------------------------------------------------------------- /mdy_triton/core/fused_apply_rope.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/fused_apply_rope.py -------------------------------------------------------------------------------- /mdy_triton/core/fused_silu.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/fused_silu.py -------------------------------------------------------------------------------- /mdy_triton/core/rmsnorm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/rmsnorm.py -------------------------------------------------------------------------------- /mdy_triton/core/triton_maxmin.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/core/triton_maxmin.py -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/__init__.py -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/__pycache__/__init__.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/__pycache__/__init__.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/__pycache__/glm4.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/__pycache__/glm4.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/__pycache__/llama.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/__pycache__/llama.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/__pycache__/llama3.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/__pycache__/llama3.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/__pycache__/mistral.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/__pycache__/mistral.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/__pycache__/qwen2.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/__pycache__/qwen2.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/__pycache__/qwen2_vl.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/__pycache__/qwen2_vl.cpython-310.pyc -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/glm4.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/glm4.py -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/llama.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/llama.py -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/mistral.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/mistral.py -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/phi3.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/phi3.py -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/qwen2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/qwen2.py -------------------------------------------------------------------------------- /mdy_triton/replace_kernel/qwen2_vl.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/mdy_triton/replace_kernel/qwen2_vl.py -------------------------------------------------------------------------------- /setup.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/setup.py -------------------------------------------------------------------------------- /triton_kernels/__pycache__/causal_conv1d.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/__pycache__/causal_conv1d.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/__pycache__/causal_conv1d_triton.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/__pycache__/causal_conv1d_triton.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/__pycache__/replace_conv1d_cuda.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/__pycache__/replace_conv1d_cuda.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/__pycache__/triton_causal_conv1d.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/__pycache__/triton_causal_conv1d.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/causal_conv1d/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /triton_kernels/causal_conv1d/__pycache__/causal_conv1d.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/causal_conv1d/__pycache__/causal_conv1d.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/causal_conv1d/__pycache__/triton_causal_conv1d.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/causal_conv1d/__pycache__/triton_causal_conv1d.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/causal_conv1d/causal_conv1d.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/causal_conv1d/causal_conv1d.ipynb -------------------------------------------------------------------------------- /triton_kernels/causal_conv1d/replace_conv1d_cuda.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/causal_conv1d/replace_conv1d_cuda.py -------------------------------------------------------------------------------- /triton_kernels/causal_conv1d/triton_causal_conv1d.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/causal_conv1d/triton_causal_conv1d.py -------------------------------------------------------------------------------- /triton_kernels/cross_entropy/test_ce.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/cross_entropy/test_ce.py -------------------------------------------------------------------------------- /triton_kernels/cross_entropy/triton_ce.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/cross_entropy/triton_ce.py -------------------------------------------------------------------------------- /triton_kernels/cross_entropy/triton_cross_entropy.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/cross_entropy/triton_cross_entropy.ipynb -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/README.md -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/TritonAdam.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/TritonAdam.py -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/TritonAdam.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/TritonAdam.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/bf16mv_optimizer.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/bf16mv_optimizer.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/fp8_kernel.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/fp8_kernel.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/fp8_module.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/fp8_module.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/kernel.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/kernel.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/kernel.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/kernel.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/new_fp8_module.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/new_fp8_module.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/new_fp8_module.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/new_fp8_module.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/offical_kernel.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/offical_kernel.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/offical_matmul.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/offical_matmul.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/__pycache__/offical_matmul.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/__pycache__/offical_matmul.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/fp8-benchmark.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/fp8-benchmark.ipynb -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/fp8_kernel.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/fp8_kernel.py -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/imgs/image.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/imgs/image.png -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/imgs/image1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/imgs/image1.png -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/imgs/image2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/imgs/image2.png -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/offical_kernel.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/offical_kernel.py -------------------------------------------------------------------------------- /triton_kernels/deepseekv3/optmizer.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/deepseekv3/optmizer.ipynb -------------------------------------------------------------------------------- /triton_kernels/dpsk_fp8linear/__pycache__/fp8.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dpsk_fp8linear/__pycache__/fp8.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/dpsk_fp8linear/fp8.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dpsk_fp8linear/fp8.py -------------------------------------------------------------------------------- /triton_kernels/dpsk_fp8linear/test_fp8.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dpsk_fp8linear/test_fp8.ipynb -------------------------------------------------------------------------------- /triton_kernels/dyt/__pycache__/dyt.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/__pycache__/dyt.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/dyt/__pycache__/dyt.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/__pycache__/dyt.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/dyt/__pycache__/dyt2.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/__pycache__/dyt2.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/dyt/__pycache__/rmsnorm.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/__pycache__/rmsnorm.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/dyt/__pycache__/rmsnorm.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/__pycache__/rmsnorm.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/dyt/dyt.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/dyt.py -------------------------------------------------------------------------------- /triton_kernels/dyt/dyt2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/dyt2.py -------------------------------------------------------------------------------- /triton_kernels/dyt/rmsnorm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/rmsnorm.py -------------------------------------------------------------------------------- /triton_kernels/dyt/test_dyt.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/dyt/test_dyt.ipynb -------------------------------------------------------------------------------- /triton_kernels/flash_attn_triton3.4/test_hooper.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/flash_attn_triton3.4/test_hooper.ipynb -------------------------------------------------------------------------------- /triton_kernels/flash_attn_triton3.4/triton_flash_attn.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/flash_attn_triton3.4/triton_flash_attn.py -------------------------------------------------------------------------------- /triton_kernels/flash_muon/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/flash_muon/README.md -------------------------------------------------------------------------------- /triton_kernels/flash_muon/image.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/flash_muon/image.png -------------------------------------------------------------------------------- /triton_kernels/flash_muon/test.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/flash_muon/test.ipynb -------------------------------------------------------------------------------- /triton_kernels/flash_muon/triton_flash_muon.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/flash_muon/triton_flash_muon.py -------------------------------------------------------------------------------- /triton_kernels/grpo/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/README.md -------------------------------------------------------------------------------- /triton_kernels/grpo/__pycache__/replace_grpo_trainer.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/__pycache__/replace_grpo_trainer.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/grpo/build/lib/triton_grpo_loss/run_bs_one_by_one.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/build/lib/triton_grpo_loss/run_bs_one_by_one.py -------------------------------------------------------------------------------- /triton_kernels/grpo/grpo_loss.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/grpo_loss.ipynb -------------------------------------------------------------------------------- /triton_kernels/grpo/grpo_loss_trl==0.16.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/grpo_loss_trl==0.16.ipynb -------------------------------------------------------------------------------- /triton_kernels/grpo/grpo_loss_trl==0.16.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/grpo_loss_trl==0.16.py -------------------------------------------------------------------------------- /triton_kernels/grpo/imgs/clamp.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/imgs/clamp.jpg -------------------------------------------------------------------------------- /triton_kernels/grpo/imgs/memory.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/imgs/memory.jpg -------------------------------------------------------------------------------- /triton_kernels/grpo/imgs/reward.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/imgs/reward.jpg -------------------------------------------------------------------------------- /triton_kernels/grpo/imgs/triton_code.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/imgs/triton_code.jpg -------------------------------------------------------------------------------- /triton_kernels/grpo/infos.jsonl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/infos.jsonl -------------------------------------------------------------------------------- /triton_kernels/grpo/replace_grpo_trainer_trl==0.14.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/replace_grpo_trainer_trl==0.14.py -------------------------------------------------------------------------------- /triton_kernels/grpo/replace_grpo_trainer_trl==0.15.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/replace_grpo_trainer_trl==0.15.py -------------------------------------------------------------------------------- /triton_kernels/grpo/setup.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/setup.py -------------------------------------------------------------------------------- /triton_kernels/grpo/test_memory.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/test_memory.py -------------------------------------------------------------------------------- /triton_kernels/grpo/triton_grpo_loss/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/triton_grpo_loss/__init__.py -------------------------------------------------------------------------------- /triton_kernels/grpo/triton_grpo_loss/__pycache__/__init__.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/triton_grpo_loss/__pycache__/__init__.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/grpo/triton_grpo_loss/__pycache__/__init__.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/triton_grpo_loss/__pycache__/__init__.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/grpo/triton_grpo_loss/__pycache__/core.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/triton_grpo_loss/__pycache__/core.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/grpo/triton_grpo_loss/__pycache__/core.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/triton_grpo_loss/__pycache__/core.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/grpo/triton_grpo_loss/core.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/triton_grpo_loss/core.py -------------------------------------------------------------------------------- /triton_kernels/grpo/triton_grpo_loss/decouple_logp_and_loss.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/triton_grpo_loss/decouple_logp_and_loss.py -------------------------------------------------------------------------------- /triton_kernels/grpo/triton_grpo_loss/run_bs_one_by_one.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/grpo/triton_grpo_loss/run_bs_one_by_one.py -------------------------------------------------------------------------------- /triton_kernels/lighting_attention/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /triton_kernels/lighting_attention/lighting_attention.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/lighting_attention/lighting_attention.ipynb -------------------------------------------------------------------------------- /triton_kernels/lighting_attention/lighting_attention.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/lighting_attention/lighting_attention.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/README.md -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__init__.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/__init__.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/__init__.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/__init__.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/__init__.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/add_kernel.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/add_kernel.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/add_kernel.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/add_kernel.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/bf16mv_adam.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/bf16mv_adam.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/bf16mv_adam.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/bf16mv_adam.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/comm.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/comm.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/comm.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/comm.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/cross_entropy_losss.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/cross_entropy_losss.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/cross_entropy_losss.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/cross_entropy_losss.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/dpsk_grouped_linear.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/dpsk_grouped_linear.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/dpsk_grouped_linear.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/dpsk_grouped_linear.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/dpsk_layernorm_linear.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/dpsk_layernorm_linear.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/dpsk_layernorm_linear.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/dpsk_layernorm_linear.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/dpsk_linear.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/dpsk_linear.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/dpsk_linear.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/dpsk_linear.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/fp8_matmul.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/fp8_matmul.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/fp8_matmul.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/fp8_matmul.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/manger.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/manger.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/manger.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/manger.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/swiglu.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/swiglu.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/swiglu.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/swiglu.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/zb2p.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/zb2p.cpython-310.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/__pycache__/zb2p.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/__pycache__/zb2p.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/add_kernel.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/add_kernel.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/bf16mv_adam.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/bf16mv_adam.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/comm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/comm.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/cross_entropy_losss.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/cross_entropy_losss.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/dpsk_grouped_linear.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/dpsk_grouped_linear.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/dpsk_layernorm_linear.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/dpsk_layernorm_linear.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/dpsk_linear.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/dpsk_linear.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/fp8_matmul.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/fp8_matmul.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/image.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/image.png -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/manger.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/manger.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/swiglu.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/swiglu.py -------------------------------------------------------------------------------- /triton_kernels/megatron_patchs/zb2p.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/megatron_patchs/zb2p.py -------------------------------------------------------------------------------- /triton_kernels/mla_attention/__pycache__/mla.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/mla_attention/__pycache__/mla.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/mla_attention/mla.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/mla_attention/mla.ipynb -------------------------------------------------------------------------------- /triton_kernels/mla_attention/mla.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/mla_attention/mla.py -------------------------------------------------------------------------------- /triton_kernels/mla_attention/test.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/mla_attention/test.ipynb -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/combine.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/combine.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/compress_attn.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/compress_attn.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/compress_attn_v2.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/compress_attn_v2.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/compresss_attn_v2.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/compresss_attn_v2.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/compute_attn.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/compute_attn.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/exp_family.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/exp_family.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/nas_attn.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/nas_attn.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/nsa_attn.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/nsa_attn.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/pku_nsa.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/pku_nsa.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/select_attn.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/select_attn.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/select_attn_v2.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/select_attn_v2.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/select_attn_v3.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/select_attn_v3.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/torch_code.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/torch_code.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/__pycache__/triton_flash_attn2.cpython-312.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/__pycache__/triton_flash_attn2.cpython-312.pyc -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/combine.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/combine.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/compress_attn.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/compress_attn.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/compress_attn_v2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/compress_attn_v2.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/exp_family.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/exp_family.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/nsa_attn.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/nsa_attn.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/pku_nsa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/pku_nsa.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/select_attn.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/select_attn.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/select_attn_v2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/select_attn_v2.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/select_attn_v3.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/select_attn_v3.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/torch_code.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/torch_code.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/triton_flash_attn2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/triton_flash_attn2.py -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/定向爆破.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/定向爆破.ipynb -------------------------------------------------------------------------------- /triton_kernels/nsa_attention/精度和性能测试.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/mdy666/mdy_triton/HEAD/triton_kernels/nsa_attention/精度和性能测试.ipynb --------------------------------------------------------------------------------