A high-throughput and memory-efficient inference and serving engine for LLMs
main
[Bugfix][KV Offload] Centralize shared mmap cleanup in CPU worker (#51622)
AlexHuang committed 3 months ago
[Bugfix][Core] Preserve Mamba running CoW after external hits (#51766)
Dao007forever committed 3 months ago
profiler: add PrivateUse1 activity support for custom backends (#50977)
David Holtz committed 3 months ago
[Perf] Avoid repeated multimodal prompt update scans (#51774)
Tianyu Guo committed 3 months ago
[XPU] [Linear] enable torch linear backend for blockwise gemm on xpu (#50826)
zofia committed 3 months ago
[Doc] Fix typos in speculative decoding docs (#51500)
Kyungmin Lee committed 3 months ago
[MRV2][Spec] Fuse AR speculator multi-step decodes back into one CUDA graph (#46849)
Yizhou committed 3 months ago
[ROCm][DSV4] Preserve native MXFP4 TP8 shard allocation (#51473)
Fangzhou Ai committed 3 months ago
Harry Mellor committed 3 months ago
[CPU][Zen] Route BF16 MoE inference through zentorch on AMD (#44201)
Priyansh Jain committed 3 months ago
[XPU] install xpu-manager for device monitor (#50831)
Yan Ma committed 3 months ago
[CI][AMD] Persist the openai-harmony tiktoken vocab cache across jobs (#51666)
stefankoncarevic committed 3 months ago
[Attention] Fix MLA prefill workspace allocation size (#51733)
Wei Zhao committed 3 months ago
[Kernel] Optimize long-context MLA cache gathers (#51739)
Yongye Zhu committed 3 months ago
[Bugfix][Frontend] Report Cohere stop sequences correctly (#51556)
cherry77-cloud committed 3 months ago
replace batch_norm to numerically identical without cudnn (#51734)
Khushali Desai committed 3 months ago
Preserve revision pins in secondary artifact loaders (#51446)
Clinton Thomas committed 3 months ago
[Misc] Use VLLMValidationError in scoring input validation (#51753)
Frank committed 3 months ago
[BugFix][Core] free_blocks: restore prepend (LIFO) reuse order when prefix caching is off (#51482)
Ming Huang committed 3 months ago
[Perf] Adaptive budget for spec scheduled input tokens, ~60% better Kimi K3 DSpark TTFT (#51725)
Wentao Ye committed 3 months ago