bukejiyu 37569cca86 [feat]add fast_weights_iterator (#3258) ...

* add fast_weights_iterator

* update

* update

2025-08-07 22:36:46 +08:00

..

attention

【Fix Bug】 修复 fa3 支持集中式bug (#3235)

2025-08-06 16:24:27 +08:00

backends

qwen3_moe (#3084)

2025-08-06 14:45:27 +08:00

moe

qwen3_moe (#3084)

2025-08-06 14:45:27 +08:00

quantization

[bugfix]fix blockwisefp8 and all_reduce (#3243)

2025-08-06 23:54:33 +08:00

sample

[Feature] support seed parameter (#3161)

2025-08-06 15:20:47 +08:00

__init__.py

[LLM] First commit the llm deployment code

2025-06-09 19:20:15 +08:00

activation.py

native top_p_sampling (#2901)

2025-07-22 14:09:59 +08:00

embeddings.py

[bugfix]fix blockwisefp8 and all_reduce (#3243)

2025-08-06 23:54:33 +08:00

linear.py

[feat]add fast_weights_iterator (#3258)

2025-08-07 22:36:46 +08:00

lm_head.py

[bugfix]fix blockwisefp8 and all_reduce (#3243)

2025-08-06 23:54:33 +08:00

mtp_linear.py

polish code with new pre-commit rule (#2923)

2025-07-19 23:19:27 +08:00

normalization.py

polish code with new pre-commit rule (#2923)

2025-07-19 23:19:27 +08:00

rotary_embedding.py

polish code with new pre-commit rule (#2923)

2025-07-19 23:19:27 +08:00

utils.py

w4a8 offline (#3074)

2025-07-30 16:33:30 +08:00