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 5, 18:30 UTC
main
[Bugfix] Use null block (0) for padded block table entries (#35431)
SandishKumarHN committed 7 months ago
[Bug fix][Quantization] Fix dummy weight loading (#38478)
Asaf Gardin committed 7 months ago
[ROCm][CI] Pin test_hybrid test to TRITON_ATTN on ROCm (#38381)
Micah Williamson committed 7 months ago
[EPLB] Optmize eplb mapping and record in router for prefill (#36261)
Ilya Markov committed 7 months ago
[Feat][Spec Decode] DFlash (#36847)
Benjamin Chislett committed 7 months ago
[3/n] Migrate cutlass/scaled_mm_entry.cu torch stable ABI (#37221)
mikaylagawarecki committed 7 months ago
[Refactor] Unify engine process monitoring in engine manager and add Ray backend support (#35862)
fangyuchu committed 7 months ago
[Bugfix][MLA] Change default SM100 MLA prefill backend back to TRT-LLM (#38562)
Matthew Bonanni committed 7 months ago
Chendi.Xue committed 7 months ago
[NVIDIA] Bugfix NVFP4 DGX Spark and RTX50 (#38423)
Johnny committed 7 months ago
[Mamba] Add stochastic rounding support (#35753)
roikoren755 committed 7 months ago
[ROCm][Bugfix] fix exception related to trust_remote_code for MiniMax-M2.1-MXFP4 (#37698)
Hongxia Yang committed 7 months ago
[Bugfix] Handle ParallelLMHead in compressed-tensors get_quant_method (#37291)
Matthias Gehre committed 7 months ago
[Misc] Add @tomeras91 as a maintainer of Nemotron related code + mamba block (#38547)
tomeras91 committed 7 months ago
[Bugfix] Remove false-positive format mismatch warnings in FLA ops (#38255)
Thomas Parnell committed 7 months ago
[Bugfix][CPU] Skip set_num_threads after thread binding (#38535)
Li, Jiang committed 7 months ago
[Core][CI] Add opt-in media URL caching via VLLM_MEDIA_CACHE (#37123)
Andreas Karatzas committed 7 months ago
Fix ambiguous num_blocks for hybrid attn mamba (#37236)
Collin McCarthy committed 7 months ago
[Bugfix][Frontend] Return 400 for corrupt/truncated image inputs instead of 500 (#38253)
aliialsaeedii committed 7 months ago
[Bugfix] Fix shared-object aliasing in n>1 streaming with tool calls (#38158)
yzong-rh committed 7 months ago