vllm-project/vllm

Pull Requests Commits

[V1][Scheduler] Use dict for running queue

njhill committed 1 year ago

90eb28ca

[Misc] Clean up processor tests (#14771)

DarkLight1337 committed 1 year ago

Verified 8e9ffd37

[V1][Minor] Minor enhancements on scheduler (#14732)

WoosukKwon committed 1 year ago

Verified 01b3fd0a

[Bugfix] Fix prompt format of GLM4V (#14539)

DarkLight1337 committed 1 year ago

Verified f53a0586

[VLM] Support loading InternVideo2.5 models as original InternVLChatModel (#14738)

Isotr0py committed 1 year ago

Verified b1cc4dfe

[VLM] Support pan-and-scan for Gemma3 multi-modal processor (#14672)

DarkLight1337 committed 1 year ago

Verified 38240392

[Bugfix] fix benchmark moe (#14653)

jeejeelee committed 1 year ago

Verified a73122de

[CI/Build] Delete ultravox LoRA test (#14730)

jeejeelee committed 1 year ago

Verified bd44b812

[Bugfix] Fix chunked prefill for GGUF (#14666)

SzymonOzog committed 1 year ago

Verified 55211b01

[Quant] Bamba SupportsQuant (#14698)

kylesayrs committed 1 year ago

Verified 5d043c16

[Quant] BartModel SupportsQuant (#14699)

kylesayrs committed 1 year ago

Verified 36d1ccb2

[V1][TPU] Add assertion on multi-step-scheduler (#14707)

lsy323 committed 1 year ago

Verified 1bc3b739

[Config][Disaggregated] Add timeout configuration for the torch.store and add KVTransferConfig.kv_connector_extra_config (#14367)

hasB4K committed 1 year ago

Verified 1bd32bc8

[BugFix][TritonMLA] Process weights after model loading for GGUF (#14555)

tywuAMD committed 1 year ago

Verified 128bf752

[ROCm][FP8] Fix for adjustments needed only for fnuz (#14689)

gshtras committed 1 year ago

Verified a94a699c

Add ray[data] as tpu dependency (#14691)

richardsliu committed 1 year ago

Verified ab426ec9

[bugfix] fixup warning message for plugged schedulers for v1 (#14700)

joerunde committed 1 year ago

Verified 165290d3

[release] Add force remove for TPU logs (#14697)

khluu committed 1 year ago

Verified ce201246

[V1] Allow sliding window + prefix caching (#13069)

WoosukKwon committed 1 year ago

Verified 53be4a86

[BugFix][V1] Fix parallel sampling finishing/aborts (#14512)

njhill committed 1 year ago

Verified f5d3acd4

[FEAT] [ROCm] [Embedding] Add encoder-only model support into ROCm Flash Attention to enable embedding models. (#14664)

tjtanaa committed 1 year ago

Verified 916836bb

[ROCm] Enable chunked prefill/paged attention in MLA on ROCm (#14316)

SageMoore committed 1 year ago

Verified d9f83d62

[Bugfix] Missing thumbnail from NVLM-D processor (#14633)

ameyanjarlekar committed 1 year ago

Verified 4a754fcf

[Model] Add support for Gemma 3 (#14660)

WoosukKwon committed 1 year ago

Verified c0c25e25

[ROCm][Bugfix] Ensure that the moe_wna16_gemm kernel is not built on ROCm platforms. (#14629)

SageMoore committed 1 year ago

Verified 45f3f3f5

[CPU] Upgrade CPU backend to torch-2.6 (#13381)

bigPYJ1151 committed 1 year ago

Verified ff47aab0

[Kernel] Add ModelOpt FP4 Checkpoint Support (#12520)

pavanimajety committed 1 year ago

Verified debd6bbf

[V1][Bugfix][Spec Decode] Fix incorrect outputs in V1 speculative decoding due to batch indexing (#14645)

benchislett committed 1 year ago

Verified 5c538c37

[Kernel] GGUF MoE kernel (#14613)

SzymonOzog committed 1 year ago

Verified e22ee1e7

[Core] Refactor `QKVCrossParallelLinear` implementation to support BNB 4-bit quantization (#14545)

Isotr0py committed 1 year ago

Verified e392d858

Older