A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 8, 17:45 UTC
main
[Fix] Fix `logprobs=0` handling for `/inference/v1/generate` endpoint (#34010)
Sumanth R Hegde committed 9 months ago
[Bugfix] Fix no attribute error of SharedFusedMoE (DeepSeek-V3.1 as test model) (#33993)
xuebwang-amd committed 9 months ago
[Rocm][Bugfix] Fix dtype not same for gemm_a4w4 op (#33734)
Charlie Fu committed 9 months ago
[Refactor] Remove align block size logic in `moe_permute` (#33449)
Wentao Ye committed 9 months ago
[Model Runner V2] support apply penalty for spec decode (#33251)
zhrrr committed 9 months ago
[DOC] [ROCm] Update docker deployment doc (#33971)
vllmellm committed 9 months ago
[KV Connector] Add missing method overrides to MultiConnector (#33292)
Seiji Eicher committed 9 months ago
[Log] Optimize duplicate startup log (#33944)
Wentao Ye committed 9 months ago
Chauncey committed 9 months ago
[Docs] Update link to Benchmark CLI documentation (#33254)
Eldar Kurtić committed 9 months ago
[XPU][5/N] add wna16 xpu kernel (#33973)
zofia committed 9 months ago
[Refactor] Consolidate sequence normalization and enc-dec parsing (#33928)
Cyrus Leung committed 9 months ago
[Model] Support MiniCPM-o 4.5 (#33431)
tc-mb committed 9 months ago
[Docs] Add sections on process architecture and minimum CPU resources (#33940)
Michael Goin committed 9 months ago
[ROCm][AITER] Fix AITER import regression for explicit backend selection (#33749)
Andreas Karatzas committed 9 months ago
[FIX] guidance: use max(vocab_size, len(tokenizer)) for n_vocab (#33509)
FredericOdermatt committed 9 months ago
[Bugfix] Fix models and tests for transformers v5 (#33977)
Raushan Turganbay committed 9 months ago
Update `WeightTransferConfig` to be more standard like the others (#33989)
Harry Mellor committed 9 months ago
[Docs] Improve documentation (#33799)
SorenDreano committed 9 months ago
[Bugfix][Model] Support LoRA on Qwen3 Output Embedding (#29816)
Kurt Shuster committed 9 months ago