Giancarlo Delfin dec13a33b7 [Model Runner V2][Spec Decode] Add multi-layer MTP speculator (#48892) ...

Signed-off-by: Giancarlo Delfin <gdelfin@inferact.ai>

2026-07-30 15:54:15 -07:00

..

autoregressive

[Tests][Spec Decode] Add gemma4 MTP acceptance rates test (#47920)

2026-07-27 16:15:54 -07:00

dflash

[Model Runner V2][Spec Decode] Add multi-layer MTP speculator (#48892)

2026-07-30 15:54:15 -07:00

dspark

Support loading sample_from_anchor flag from speculators config (#48639)

2026-07-20 14:52:43 -07:00

eagle

[Bugfix] DFlash fc sized wrong when num_target_layers != num_hidden_layers (#48524)

2026-07-21 14:42:52 -07:00

gemma4

[Model Runner V2][Spec Decode] Handle tuple hidden states from MTP draft models (#46786)

2026-06-27 18:38:07 -07:00

mtp

Revert "[Perf][GLM-5.2] Blackwell decode optimizations" (#49768)

2026-07-24 16:04:59 -07:00

multi_module_mtp

[Model Runner V2][Spec Decode] Add multi-layer MTP speculator (#48892)

2026-07-30 15:54:15 -07:00

__init__.py

[Spec Decode] DSpark (#46995)

2026-07-01 08:30:24 -07:00

rejection_sampler_utils.py

[MRV2][Spec Decode] Avoid rejection sampler OOM by chunking (#48630)

2026-07-23 18:13:13 +08:00

rejection_sampler.py

[MRV2][Spec Decode] Avoid rejection sampler OOM by chunking (#48630)

2026-07-23 18:13:13 +08:00

speculator.py

[Model Runner V2][Spec Decode] Add multi-layer MTP speculator (#48892)

2026-07-30 15:54:15 -07:00

utils.py

[Perf] Use blocking CUDA events to avoid busy polling cuda driver lock (#47081)

2026-07-07 09:36:10 +00:00