Wentao Ye d0948f754d [Bug] Honor spec decode draft model config (#61240) ...

Signed-off-by: yewentao256 <zhyanwentao@126.com>

2026-10-12 15:13:23 -04:00

..

dflash

[Spec Decode] Skip metadata build for dflash/dspark when all model backends support it (#60692)

2026-10-12 00:09:10 -07:00

dflash2

[SpecDecode] Add LiLiCorr drafter (#57934)

2026-09-25 10:08:28 -04:00

dspark

[EPLB] Add EPLB support to dspark draft models (DeepSeekV4) (#57549)

2026-10-12 10:03:09 -04:00

eagle

[Spec Decode] Rename AR speculators to StandaloneAR / TargetDependentAR (#60335)

2026-10-07 14:01:59 -04:00

gemma4

[Spec Decode] Rename AR speculators to StandaloneAR / TargetDependentAR (#60335)

2026-10-07 14:01:59 -04:00

lilicorr

[SpecDecode] Add LiLiCorr drafter (#57934)

2026-09-25 10:08:28 -04:00

mtp

[Spec Decode] Rename AR speculators to StandaloneAR / TargetDependentAR (#60335)

2026-10-07 14:01:59 -04:00

multi_module_mtp

[MRV2] Miscellaneous code structure updates (#60754)

2026-10-09 17:58:09 -07:00

ngram

[MRV2] Miscellaneous code structure updates (#60754)

2026-10-09 17:58:09 -07:00

standalone_ar

[Bug] Honor spec decode draft model config (#61240)

2026-10-12 15:13:23 -04:00

target_dependent_ar

[Spec Decode] Skip metadata build for dflash/dspark when all model backends support it (#60692)

2026-10-12 00:09:10 -07:00

__init__.py

[Spec Decode] Rename AR speculators to StandaloneAR / TargetDependentAR (#60335)

2026-10-07 14:01:59 -04:00

acceptance_estimator.py

[Perf][Spec Decode] Avoid triton recompiles in the acceptance estimator (#57107)

2026-09-28 13:46:25 -07:00

adaptive_verification.py

[Spec Decode] Enable Gemma4 DSpark adaptive verification with FlashInfer (#57263)

2026-09-26 17:01:17 +08:00

extract_hidden_states.py

[MRV2] Miscellaneous code structure updates (#60754)

2026-10-09 17:58:09 -07:00

rejection_sampler_utils.py

[Bugfix] Fix out-of-bounds lookahead in adaptive block verification (#60945)

2026-10-12 09:21:35 -04:00

rejection_sampler.py

[Feature][Spec Decode] Support sampling mask replay for MRV2 MTP (#59359)

2026-10-01 13:51:39 +08:00

speculator.py

[Spec Decode] Skip metadata build for dflash/dspark when all model backends support it (#60692)

2026-10-12 00:09:10 -07:00

utils.py

[MRV2] Miscellaneous code structure updates (#60754)

2026-10-09 17:58:09 -07:00