Matthias Gehre 047413375c [Attention][AMD] Make flash-attn optional (#30361) ...

Signed-off-by: Matthias Gehre <matthias.gehre@amd.com>

2026-01-15 17:18:24 +00:00

..

attention

[Attention][AMD] Make flash-attn optional (#30361)

2026-01-15 17:18:24 +00:00

core

[Feature] Support async scheduling + PP (#32359)

2026-01-15 12:06:23 -05:00

engine

[2/N] Move cache factories to MM registry (#32382)

2026-01-15 01:02:30 -08:00

executor

[Feature] Support async scheduling + PP (#32359)

2026-01-15 12:06:23 -05:00

kv_offload

OffloadingConnector: Add cpu_bytes_to_use configuration (#24498)

2026-01-12 15:00:43 +00:00

metrics

[Metrics] Model FLOPs Utilization estimation (#30738)

2025-12-18 01:40:51 +00:00

pool

[Refactor] Clean up pooler modules (#31897)

2026-01-08 00:07:43 +08:00

sample

[ROCm][Bugfix] Disable hip sampler to fix deepseek's accuracy issue on ROCm (#32413)

2026-01-15 16:35:47 +00:00

spec_decode

[Bugfix] Fix stale common_attn_metadata.max_seq_len in speculative decoding with Eagle (#32312)

2026-01-15 06:39:37 +00:00

structured_output

[Misc] Various code simplifications (#31666)

2026-01-04 18:35:56 -08:00

worker

[Model Runner V2] Support FlashInfer backend & Fix CUDA Graph bug [1/2] (#32348)

2026-01-15 08:59:23 -08:00

__init__.py

[V1] AsyncLLM Implementation (#9826)

2024-11-11 23:05:38 +00:00

cudagraph_dispatcher.py

[BugFix] Fix assert batch_descriptor.num_tokens == num_tokens_padded (#30173)

2025-12-09 10:36:12 -05:00

kv_cache_interface.py

Decouple page_size_bytes calculation in AttentionSpec for TPU/RPA Compatibility. (#31635)

2026-01-08 09:00:07 +00:00

outputs.py

[perf][async] support non cpu sync get logprob tensors for spec (#31336)

2026-01-09 21:24:51 +00:00

request.py

[BugFix] scheduler: Fix resuming of preempted requests after async load (#31583)

2026-01-10 12:39:25 -08:00

serial_utils.py

[Model] Move multimodal_cpu_fields definition to field config (#30181)

2025-12-06 13:40:02 +00:00

utils.py

[Feature] Add iteration level logging and enhance nvtx marker (#31193)

2026-01-09 00:13:39 +00:00