Commits vllm-project/vllm

[ROCm][CI] Stabilize 400 error return code for invalid schema inputs (#43016)

AndreasKaratzas committed 123 days ago

Verified 59405908

[kv_offload]: Add DSv4 support (#43142)

orozery committed 123 days ago

Verified 357fddf6

[KV Connector] Keep MooncakeStore full hits block-aligned (#43494)

Dao007forever committed 123 days ago

Verified 0902d8e6

[Model Runner v2] Force v1 runner for tests (#43233)

yewentao256 committed 123 days ago

Verified 33d7cbe0

[Bugfix] Fix reasoning dropped on streaming boundary deltas (#42691)

sfeng33 committed 123 days ago

Verified b32fe416

Revert "[Misc] add humming to dependencies" (#43492)

mgoin committed 123 days ago

Verified 10d264a2

[ROCm][Critical] Fix the GDN import bug (#43486)

tjtanaa committed 123 days ago

Verified 46f95b2e

[Mooncake] Add metrics for MooncakeStoreConnector operations (#43392)

Dao007forever committed 123 days ago

Verified 819c610f

[MoE] Migrate W4A8 CT to oracle kernel setup (#42680)

bedeks committed 123 days ago

Verified 4438b6e7

[Docs] Fix stale version number in token_classify.md (#43489)

fuergaosi233 committed 123 days ago

Verified 8737e4a8

[Docs] Fix stale version number in token_embed.md (#43488)

fuergaosi233 committed 123 days ago

Verified 7c2ff1f8

[MM] Enable FlashInfer metadata support for Qwen2.5-VL vision attention (#42787)

huanghua1994 committed 123 days ago

Verified a0be71ee

[Bugfix][Frontend] Fix input_audio parsing when uuid is present (#43414)

ffggs committed 123 days ago

Verified d8b385b7

[ROCm][CI] Remove benchmarks test group and shard long test groups (#41669)

AndreasKaratzas committed 124 days ago

Verified 2a7d5b73

[Kernel] Batch invariant NVFP4 linear using cutlass (#39912)

jzakrzew committed 124 days ago

Verified 5bb8d276

fix(eagle3): read norm_before_fc from eagle_config for NVIDIA checkpoint (#42143)

popkart-EZ committed 124 days ago

Verified 3f3e8626

Keep scheduler alive for delayed KV connector frees (#43433)

lucifer1004 committed 124 days ago

Verified 82536acc

[ModelOpt] Support Qwen3.5/3.6 VLM quantized prefix mapping (#42546)

meenchen committed 124 days ago

Verified 09a219c0

[Bugfix] Fix native Triton top-k/top-p kernel assumes contiguous logi… (#42739)

zhougit86 committed 124 days ago

Verified d19db109

[Misc] Added missing return type annotations to improve mypy and IDE tooling (#43383)

taneem-ibrahim committed 124 days ago

Verified 3a1c0621

[7/n] Migrate pos_encoding and norm kernels to libtorch stable ABI (continued) (#43209)

cleonard530 committed 124 days ago

Verified a7be0f34

[XPU] reudce host overhead of XPU MOE (#42915)

mayuyuace committed 124 days ago

Verified 54d15363

[XPU]feat: enable FP8 block-scaled quantization on XPU (#42952)

majian4work committed 124 days ago

Verified a5bbd81e

[ROCm][CI] Fix ROCm LoRA Transformers fallback with full CUDA graphs (#41577)

AndreasKaratzas committed 124 days ago

Verified d28bdf93

[Bugfix] Auto-raise max_num_batched_tokens for prefix-LM multimodal models (#43051)

ashwing committed 124 days ago

Verified 84e35155

[ROCm][CI] Stabilize Granite tool-use and test URL construction (#43017)

AndreasKaratzas committed 124 days ago

Verified 76ea1d5d

[ROCm][CI] Stabilize runner teardown between sampler tests (#43023)

AndreasKaratzas committed 124 days ago

Verified 6a4723a2

[DSV4] More multi-stream enablement for c4a (#42925)

zyongye committed 124 days ago

Verified 367cb819

Add `model` to `WeightTransferEngine.__init__` (#42922)

SumanthRH committed 124 days ago

Verified 3cb83c95

[Attention] Add head_dim=512 support for FlashInfer trtllm attention backend (#38822)

djmmoss committed 124 days ago

Verified 552bbe6f