DimaandDmytro Kuntso 128adabfe0 [Bugfix] Fix Gemma4 MTP block_table batch_size mismatch under concurrent load (#43982) ...

Signed-off-by: Dmytro Kuntso <dkuntso@amazon.co.uk>
Co-authored-by: Dmytro Kuntso <dkuntso@amazon.co.uk>

2026-06-03 17:11:10 -07:00

..

__init__.py

[V1][BugFix] Add __init__.py to v1/spec_decode/ (#13359)

2025-02-16 09:39:08 -08:00

custom_class_proposer.py

[Feature] Support custom callable proposer backend for speculative decoding (#39487)

2026-05-13 16:53:01 +00:00

dflash.py

[Spec Decode] Allow causal DFlash (#43445)

2026-05-28 21:18:44 +00:00

draft_model.py

[Spec Decode] Move SpecDecodeBaseProposer out of eagle.py (#40732)

2026-04-23 22:28:27 +00:00

eagle.py

[Spec Decode] Move SpecDecodeBaseProposer out of eagle.py (#40732)

2026-04-23 22:28:27 +00:00

extract_hidden_states.py

[Perf] Optimize hidden state extraction logic (#37374)

2026-05-22 18:23:08 -04:00

gemma4.py

[Bugfix] Fix Gemma4 MTP block_table batch_size mismatch under concurrent load (#43982)

2026-06-03 17:11:10 -07:00

llm_base_proposer.py

[Model] Add Gemma4 Unified (encoder-free) support (#44429)

2026-06-03 12:01:39 -07:00

medusa.py

[CI] Enable mypy import following for vllm/spec_decode (#33282)

2026-01-30 06:43:32 +00:00

metadata.py

[V1][spec decode] return logprobs for spec decoding (#26060)

2025-10-22 22:59:59 -07:00

metrics.py

[Metrics] Some small refactoring for better maintainability (#33898)

2026-03-20 16:11:34 +00:00

ngram_proposer_gpu.py

[Perf][1/n] Eliminate various GPU<->CPU syncs (#41429)

2026-05-11 20:36:03 +00:00

ngram_proposer.py

[Performance] Split FlashAttn attention and cache update (#25954)

2026-01-23 17:28:06 -08:00

step3p5.py

[Model]Support Step-3.7-Flash (#43859)

2026-05-28 17:01:48 -07:00

suffix_decoding.py

[CI] Enable mypy import following for vllm/spec_decode (#33282)

2026-01-30 06:43:32 +00:00

utils.py

[Feat] Unified Synthetic Acceptance Rate for V1 and V2 (#40662)

2026-04-24 00:48:08 +00:00