Michael Lapshin d421a2e8fc [Perf] Remove CPU-GPU sync in heterogeneous vocabulary speculative decoding (#57396)

2026-09-22 18:34:19 -04:00

..

dynamic

…

__init__.py

…

custom_class_proposer.py

…

dflash.py

…

draft_model.py

…

eagle.py

…

extract_hidden_states.py

…

gemma4.py

…

llm_base_proposer.py

[Fast Start] Cache the MTP draft model in a separate daemon group (#57312)

2026-09-22 03:13:38 +00:00

medusa.py

…

metadata.py

…

metrics.py

…

ngram_proposer_gpu.py

…

ngram_proposer.py

…

step3p5.py

…

suffix_decoding.py

…

utils.py

…

vocab_mapping.py

[Perf] Remove CPU-GPU sync in heterogeneous vocabulary speculative decoding (#57396)

2026-09-22 18:34:19 -04:00