vllm-project/vllm

Pull Requests Commits

ywang96 committed 1 year ago

e17250f0

[Quantization] Modify the logic of BNB double quantization (#19742)

jeejeelee committed 1 year ago

Verified 49599150

[Misc][ROCm] Enforce no unused variable in ROCm C++ files (#19796)

houseroad committed 1 year ago

Verified 8d1e89d9

Fix FA2 fallback for Blackwell V1 (#19781)

mgoin committed 1 year ago

Verified 36239f79

[Frontend] Expose custom args in OpenAI APIs (#16862)

afeldman-nm committed 1 year ago

Verified dfada85e

[BugFix] Fix use_cudagraph=False (#19612)

zou3519 committed 1 year ago

Verified ed333497

[Multimodal] Use fast processor for Qwen2/2.5-VL (#19789)

WoosukKwon committed 1 year ago

Verified d49adea1

[Core] More fixes to MultiModalEmbeddings type handling (#19715)

russellb committed 1 year ago

Verified 14fdd21d

[TPU] Update torch-xla version to include paged attention tuned block change (#19813)

QiliangCui committed 1 year ago

Verified 04fefe7c

[Core] Do not copy array during hashing (#19484)

lgeiger committed 1 year ago

Verified 3b523e38

Disable "Forbid direct 'import triton'" check for `vllm/triton_utils/importing.py` in an extensible way (#19783)

afeldman-nm committed 1 year ago

Verified 16c16301

docs: fix Slack bulletpoint in README (#19811)

nathan-weinberg committed 1 year ago

Verified 9206d0ff

[v1] Support mamba2 (#19327)

heheda12345 committed 1 year ago

Verified a89209b7

[Docs] Add Huzaifa Sidhpurwala to vuln mgmt team doc (#19808)

russellb committed 1 year ago

Verified ffacb222

[Bugfix] fix RAY_CGRAPH_get_timeout is not set successfully (#19725)

chaunceyjiang committed 1 year ago

Verified 12575cfa

[Hardware][AMD] integrate aiter chunked prefill into vllm (#18596)

Zzz9990 committed 1 year ago

Verified 8b6e1d63

[Qwen] Add tagging rule for Qwen related PRs (#19799)

houseroad committed 1 year ago

Verified 735a9de7

[Platform] Allow platform use V1 Engine by default (#19792)

wangxiyuan committed 1 year ago

Verified 257ab954

[doc] fix the incorrect label (#19787)

reidliu41 committed 1 year ago

Verified cca91a7a

[Minor] Zero-initialize attn output buffer (#19784)

WoosukKwon committed 1 year ago

Verified f04d6045

[V1] Decouple GPU and TPU `InputBatch` (#19778)

afeldman-nm committed 1 year ago

Verified 19a53b27

[V1][P/D] An native implementation of xPyD based on P2P NCCL (#18242)

Abatom committed 1 year ago

Verified eccdc831

[V1] Add API docs for EncoderCacheManager (#19294)

russellb committed 1 year ago

Verified 5f52a846

[Misc] Add __str__ for RequestStatus (#19780)

lk-chen committed 1 year ago

Verified d4629dc4

[MISC] correct DeviceConfig device field static type analysis (#19699)

andyxning committed 1 year ago

Verified 6e9cc73f

[MISC] correct copy_blocks src_to_dists param type (#19696)

andyxning committed 1 year ago

Verified c53711bd

[TPU] Update torch version to include paged attention kernel change (#19706)

Chenyaaang committed 1 year ago

Verified dac8cc49

[Feature][ROCm] Add full graph capture support for TritonAttentionBackend (#19158)

charlifu committed 1 year ago

Verified a44b1c95

[Bugfix] Fix faulty triton importing logic when using Ray for DP (#19734)

mgoin committed 1 year ago

Verified b447624e

[Misc] Update lmcache connector with the latest connector apis (#19441)

YaoJiayi committed 1 year ago

Verified cda92307

Older