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 9, 18:15 UTC
main
[Voxtral] Add new streaming arch (#32861)
Patrick von Platen committed 10 months ago
[CI/Build][CPU] Fix failed pooling tests and macos smoke test (#32907)
Li, Jiang committed 10 months ago
[Misc] Add `get_name` to missing AttentionBackends (#32698)
Nicolò Lucchesi committed 10 months ago
[CI][Models] Add VLM Support for Sequence Classification Conversion (#32885)
Andreas Karatzas committed 10 months ago
[Bugfix] Fix _CPU_MOE_ACT AssertionError when vLLM config not set (#32777)
Karan Bansal committed 10 months ago
[CI] Fix mypy for `vllm/v1/structured_output` (#32722)
Wentao Ye committed 10 months ago
Luka Govedič committed 10 months ago
[BugFix] deepseek_v32_encoding: Replace asserts with proper exceptions (#32884)
Rishabh Saini committed 10 months ago
[Misc] Log vLLM logo when starting server (#32796)
Nick Hill committed 10 months ago
[MoE Refactor] Move `select_experts` from `FusedMoEQuantMethod` -> `FusedMoE` (#31996)
bnellnm committed 10 months ago
[BugFix] Fix invalid flashinfer_fused_moe_blockscale_fp8 op registration (#32855)
Fadi Arafeh committed 10 months ago
[Perf] Create TMA-aligned input scale tensor for DeepGemm on Hopper (#32619)
Xin Yang committed 10 months ago
[Refactor] Remove unused tpu files (#32610)
Wentao Ye committed 10 months ago
Add llmcompressor fp8 kv-cache quant (per-tensor and per-attn_head) (#30141)
Eldar Kurtić committed 10 months ago
[Bugfix][Attention] Explicitly report support for kv_cache_dtype bfloat16 (#32795)
Matthew Bonanni committed 10 months ago
Fadi Arafeh committed 10 months ago
[CI][Attention] Add more CI dependencies for attention tests (#32487)
Matthew Bonanni committed 10 months ago
[Feature] Add --ssl-ciphers CLI argument for TLS cipher control (#30937)
RickyChen / 陳昭儒 committed 10 months ago
Support custom URI schemes and trace handlers for profiler (#32393)
David Ramon Prados committed 10 months ago
[UX] Default api_server_count to dp_size if not specified (#32525)
Tyler Michael Smith committed 10 months ago