Adrian Abeyta c42ff4f4fd [BugFix][torch.compile] KV scale calculation issues with FP8 quantization (#25513) ...

Signed-off-by: adabeyta <aabeyta@redhat.com>

2025-09-29 15:52:04 -04:00

..

attention

Kernel-override Determinism [1/n] (#25603)

2025-09-26 16:59:09 -07:00

core

[P/D] NIXL Updates (#25844)

2025-09-29 04:46:30 +00:00

engine

[Core] GC Debug callback (#24829)

2025-09-27 17:53:31 +00:00

executor

[Misc] Don't log shm dequeue delay warning on worker side (#25720)

2025-09-26 01:08:30 +00:00

kv_offload

[Docs] Enable fail_on_warning for the docs build in CI (#25580)

2025-09-24 19:30:33 +00:00

metrics

[Core] Don't count preempted tokens in prefix cache hit rate (#25787)

2025-09-27 00:16:40 +00:00

pool

[Performance] V1 Pooling Models E2E Performance Optimization (#23162)

2025-08-21 13:26:09 +00:00

sample

[Spec Decode] Add Batch Parallel Ngram. Upto 8x lower overhead. (#24986)

2025-09-25 15:22:03 -07:00

spec_decode

[XPU]Fix xpu spec decoding UTs, avoid using cuda graph (#25847)

2025-09-29 05:15:10 +00:00

structured_output

[Build] Update Xgrammar to 0.1.25 (#25467)

2025-09-23 21:25:46 +00:00

worker

[BugFix][torch.compile] KV scale calculation issues with FP8 quantization (#25513)

2025-09-29 15:52:04 -04:00

__init__.py

[V1] AsyncLLM Implementation (#9826)

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

cudagraph_dispatcher.py

Remove redundant cudagraph dispatcher warning (#25841)

2025-09-28 17:12:42 -04:00

kv_cache_interface.py

[Hybrid Allocator] Support full attention with different hidden size (#25101)

2025-09-19 23:43:59 -07:00

outputs.py

[Misc] Simplify PoolerOutput and move to v1/outputs (#25629)

2025-09-25 06:47:03 +00:00

request.py

[Core] Don't count preempted tokens in prefix cache hit rate (#25787)

2025-09-27 00:16:40 +00:00

serial_utils.py

fix: revert cast to cpu in MsgpackEncoder._encode_tensor to avoid hidden performance regressions (#25738)

2025-09-26 01:18:05 -07:00

utils.py

[Bugfix] Optimize CpuGpuBuffer initialization (#25447)

2025-09-27 08:07:36 +08:00