vllm-project/vllm

Pull Requests Commits

Disable the broken gemm_a8w8_blockscale

gshtras committed 5 days ago

981e1d83

Change AMDGCN_USE_BUFFER_OPS default

gshtras committed 6 days ago

23d8239c

Merge: ROCM:upstream_fp8_with_static_scales_gpt_oss

gshtras committed 6 days ago

7b4424b4

Add aiter fused_add_rmsnorm_pad to gpt-oss

Rohan138 committed 6 days ago

577128bd

Merge ROCm:rope_kvcache_fusion

gshtras committed 6 days ago

25b5317c

[ROCm] Serving Fails on Radeon Due to AITER Dtype Import (#30952)

vllmellm committed 6 days ago

cee749f6

RC related config changes

gshtras committed 6 days ago

dd0b749b

Check for truthy `rope_parameters` not the existence of it (#30983)

hmellor committed 10 days ago

72506c98

[Bugfix] Remove `tile_size=64` for mm_prefix triton attention (#30973)

Isotr0py committed 10 days ago

b2eb84de

adds jais 2 support (#30188)

sarathc-cerebras committed 10 days ago

ac43367c

[Fix][FlexAttention] return max logical block index to handle reused blocks (#30915)

ivanium committed 10 days ago

30fe765e

[BugFix] Partial revert of #29558 (DeepEP HT + PIECEWISE CG support) (#30910)

LucasWilkinson committed 11 days ago

2c0ee0fd

[v1] Add PrefixLM support to TritonAttention backend (#30386)

Isotr0py committed 11 days ago

55f1fc1b

[BugFix] Workspace allocation during profile run : DeepEPHighThroughput + DeepGEMM (#30899)

varun-sundar-rabindranath committed 11 days ago

17f39880

[CI][Bugfix] Fix flaky `tests/entrypoints/openai/test_audio.py::test_chat_streaming_audio` (#30878)

NickLucche committed 11 days ago

682c3858

[XPU] fix broken fp8 online quantization for XPU platform (#30831)

yma11 committed 12 days ago

f124b567

[Bugfix][CPU] Fix CPU backend ROPE dispatch for VL models (#30829)

bigPYJ1151 committed 12 days ago

d78e128b

[BugFix] Fix memory spike in workspace allocation (#30744)

LucasWilkinson committed 12 days ago

761b730d

[ROCm] [Bugfix] Fix torch sdpa hallucination (#30789)

tjtanaa committed 12 days ago

f34eca5f

[CI] Skip ci failure test (#30804)

yewentao256 committed 12 days ago

4cd332f3

[Core][MM] Optimize encoder cache manager by operating with embeddings only (#30475)

ywang96 committed 12 days ago

16484d39

[CI/Build] Skip broken ViT backend functionality test tempoarily (#30782)

Isotr0py committed 12 days ago

e397bd65

[Bugfix] Fix broken ViT attention selection for Blackwell device (#30731)

Isotr0py committed 12 days ago

6a88d590

[CustomOp] Extract ApplyRotaryEmb as CustomOp and unify the dispatch logic (#29873)

shen-shanshan committed 12 days ago

ad8c0731

[Benchmarks] `auto_tune.sh`: Use hostname variable for server requests (#30529)

KevinMusgrave committed 13 days ago

Verified c01d5898

Update batch invariant to use attention config (#30704)

MatthewBonanni committed 13 days ago

Verified 60dbf7d8

[Bugfix] Fail instead of ignoring when CompilationConfig gets invalid args (#30708)

mgoin committed 13 days ago

Verified a450c64a

[docs][fix] Update Arm CPU vLLM wheel installation docs (#30594)

fadara01 committed 13 days ago

Verified b2191abd

[Bugfix] Fix ViT with FlashAttention on ROCm (#30703)

MatthewBonanni committed 13 days ago

Verified 51e5b3e3

[Platform] Refactor Platform attention backend selection to avoid breakpoint for OOT platform (#30212)

Isotr0py committed 13 days ago

Verified ec154c36

Older