FastDeploy

mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 16:48:03 +08:00

Files

freeliuzc 2f473ba966 [Feature][MTP]Support MTP for rl-model (#4009 )

* qk norm for speculate decode C16

* support mtp in v1_scheduler mode

* support mtp rope_3d

* support mtp features

* add unit test && del some log

---------

Co-authored-by: yuanxiaolan <yuanxiaolan01@baidu.com>
Co-authored-by: xiaoxiaohehe001 <hiteezsf@163.com>

2025-09-10 13:34:37 +08:00

draft_model_postprocess.cu

[LLM] First commit the llm deployment code

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

draft_model_preprocess.cu

[Feature][MTP]Support MTP for rl-model (#4009 )

2025-09-10 13:34:37 +08:00

draft_model_set_value_by_flags.cu

[LLM] First commit the llm deployment code

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

draft_model_update.cu

[Feature][MTP]support new speculative decoding method named hybrid mtp with ngram (#3610 )