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 11, 12:30 UTC
main
[CI] Whisper logprobs tests (#30504)
Nicolò Lucchesi committed 11 months ago
[CI] Update several models in registry that are available online now (#30514)
Michael Goin committed 11 months ago
[Docs] Remove references to `VLLM_ATTENTION_BACKEND` (#30564)
Matthew Bonanni committed 11 months ago
Add IBM and Red Hat to compute resources sponsors (#30581)
Michael Goin committed 11 months ago
[Bugfix] Pass FA version in `MultiHeadAttention` (#30575)
Matthew Bonanni committed 11 months ago
rasmith committed 11 months ago
[ci] Mark PrimeRL integration test as soft fail (#30578)
Kevin H. Luu committed 11 months ago
[Refactor] Reduce duplicate code in `per_token_group_quant` cuda kernels (#30496)
Wentao Ye committed 11 months ago
[MoE-FP8-modelopt] Add FlashInfer alignment padding for intermediate dimensions (#29748)
danielafrimi committed 11 months ago
[Fix]Load kv-cache dtype from hf_quant_config.json automatically (#29980)
danielafrimi committed 11 months ago
[CI/Build] Add x86 CPU wheel release pipeline (#28848)
Li, Jiang committed 11 months ago
[Perf] Set split_k to 1 for triton_kernels (#30528)
Xin Yang committed 11 months ago
[ROCm] Enable Triton ScaledMM fallback + kernel selection fix (#26668)
shivampr committed 11 months ago
[CI] Fix mypy for vllm/v1/executor (#30517)
Wentao Ye committed 11 months ago
[bugfix] fix bug when top_logprobs=0 with spec decoding (#30059)
realliujiaxu committed 11 months ago
[Frontend] Fixes anthropic streaming message_start usage nesting (#30266)
Benjamin Bartels committed 11 months ago
[Kernel] Support CUDA Graphs in 3D Triton Attention Kernel (#28306)
jvlunteren committed 11 months ago
[Bug] Fix attention_backend arg string parsing (#30534)
Michael Goin committed 11 months ago
fix(gguf): Disable bfloat16 for GGUF on blackwell device (#30408)
Christina Norman committed 11 months ago
[DeepSeek V3.2] Proper drop_thinking logic (#30490)
Vladislav Nosivskoy committed 11 months ago