vllm-project/vllm

Pull Requests Commits

[Tests] Standardize RNG seed utility across test files (#32982)

sjhddh committed 5 days ago

Verified 0ccecf88

[Tests] Clarify pytest skip reasons with actionable context (#32981)

sjhddh committed 5 days ago

Verified 0b9a735e

[Perf] Cache xpu_get_mem_info() result to avoid duplicate calls (#32983)

sjhddh committed 5 days ago

Verified 14d03b8d

[Dev UX] Add auto-detection for VLLM_PRECOMPILED_WHEEL_VARIANT during install (#32948)

mgoin committed 6 days ago

Verified d0cbac58

Auth_token added in documentation as it is required (#32988)

ruizcrp committed 6 days ago

Verified c0d82045

[ROCm][ViT] Enable Flash Attention Triton backend on RDNA3/RDNA4 (#32944)

monajafi-amd committed 6 days ago

Verified 97ef11dd

[Bugfix] Fix FusedMoE LoRA kernel offs_token out of bound value (#32279)

xyang16 committed 6 days ago

Verified ecc3dd66

[Core][Bugfix] allow graceful worker termination (#32965)

joerunde committed 6 days ago

Verified 7e1f10d5

[Performance] Split FlashAttn attention and cache update (#25954)

ElizaWszola committed 6 days ago

Verified a28b94e6

[fix] add VLLM_OBJECT_STORAGE_SHM_BUFFER_NAME to compile factors (#32912)

dolpm committed 6 days ago

Verified 0118cdcc

[CI] fix version comparsion and exclusion patterns in upload-release-wheels.sh (#32971)

Harry-Chen committed 6 days ago

Verified 136c499f

[Bugfix] Fix missing is_layer_skipped check for FusedMoE in AWQConfig (#32935)

joninco committed 6 days ago

Verified ebd0a17e

[Refactor] Clean up unused variables & func (#32692)

yewentao256 committed 6 days ago

Verified 37c9859f

[Refactor] Rename `gptq_marlin` to `marlin` to match MoE (#32952)

mgoin committed 6 days ago

Verified 4561f139

[CI][AMD][BugFix] Update wvSplitK (and other skinny_gemm wrappers) to ensure tensors passed will be made contiguous for the kernel (#32831)

rasmith committed 6 days ago

Verified 6cc6d92b

[Bug] Fix benchmark script `moe_permute_unpermute` (#32949)

yewentao256 committed 6 days ago

Verified dfab5f37

fix: Add glm4_moe_lite to MLA detection (#32614)

marksverdhei committed 6 days ago

Verified 586a57ad

[cudagraphs] Refactor cudagraph capture loop (#32946)

LucasWilkinson committed 6 days ago

Verified 3a414595

[Model Runner V2] Add KV Connector support (#32742)

njhill committed 6 days ago

Verified 8518b304

[Bugfix][CI] Fix pre-commit (#32956)

MatthewBonanni committed 6 days ago

Verified 2d6b5371

[CI][torch nightlies] Use main Dockerfile with flags for nightly torch tests (#30443)

orionr committed 6 days ago

Verified 68b0a6c1

[V1][Hybrid] Mamba Prefix Caching with align mode (#30877)

peakcrosser7 committed 6 days ago

Verified 5206e5e2

[Model] Enable LoRA support for internvl2 (#32397)

MatteoFari committed 6 days ago

Verified fec9da0a

[torch.compile][CI] Add back attn fusion on hopper/ada (#32940)

ProExpertProg committed 6 days ago

Verified bbbd696a

[Frontend] add logprob, compression_rate to 'verbose_json' features (#31059)

sangbumlikeagod committed 6 days ago

Verified 9b77bb79

[Hardware][AMD][CI][Bugfix] Fix Kernels Attention Cache test (#32904)

mawong-amd committed 6 days ago

Verified 305e53ad

[ROCm][PD] Remove unused moriio connector proxy code (#32939)

markmc committed 6 days ago

Verified 1cb4341f

[Bugfix] Fix FP8 MoE EP Weight Loading for ModelOpt Llama4 (#32886)

baonudesifeizhai committed 6 days ago

Verified 1fb648bf

[Misc] Postpone torch_profiler deprecation (#32867)

NickLucche committed 6 days ago

Verified 7e223097

[Bugfix] Disable tma_aligned_scales in test_fusions_e2e (#32916)

xyang16 committed 6 days ago

Verified 90c20079

Newer Older