A high-throughput and memory-efficient inference and serving engine for LLMs
main
[Core][V1] Support trace_decode_token_ids for deterministic decode replay (#46701)
Zhixia Liu committed 2 months ago
[CI][XPU] Skip test_fused_shared_expert.py on XPU (#53035)
Qiming Zhang committed 2 months ago
[ROCm][Perf] Fuse DeepSeek-V4 mHC post/pre and RMSNorm with AITER (#52737)
Shanshan Shen committed 2 months ago
[ROCm] [Bugfix] Preserve CPU query offsets during capture (#51585)
akii96 committed 2 months ago
Amit Moryossef committed 2 months ago
[Attention] Avoid redundant mask compute in GDN metadata build (#52078)
Xin Yang committed 2 months ago
[CI][ROCm] Standardize AMD test job labels by device (#52976)
Andreas Karatzas committed 2 months ago
fix: report stop_sequence stop_reason in Anthropic Messages API (#45807)
Yufeng He committed 2 months ago
[Doc] Update Gaudi HPU committers (#52726)
pmanczak committed 2 months ago
[Bugfix][Core] Reserve the KV null block when validating max_model_len (#47272)
92hyungjun committed 2 months ago
[Model] Remove unused DeepseekV32Indexer forward (#53021)
Woosuk Kwon committed 2 months ago
[ROCm][CI] Speed up `test_rocm_aiter_qk_norm_rope_kvcache_fusion` (#53004)
Micah Williamson committed 2 months ago
[Distributed] Enable FlashInfer all-reduce by default (#52998)
Woosuk Kwon committed 2 months ago
[refactor] consolidate cp attn ops (#52839)
Summer Yang committed 2 months ago
[KV Connector] Add decode offloading to Mooncake Store consumers (#52466)
Guanyi Chen committed 2 months ago
[Model Runner V2][Spec Decode] Fix draft logits cache column stride in gumbel_sample (#53017)
Giancarlo Delfin committed 2 months ago
[kv_offload] fix(metrics): rename kv_offload_tiering_block_{queries,hits} → chunk (#52812)
Ronen Schaffer committed 2 months ago
Sage Moore committed 2 months ago
[ROCm]: Bump triton 3.7 commit (#52819)
Rohan Potdar committed 2 months ago
[CI] Fix and extend PR/issue auto-labeling (#51459)
jcotant-inferact committed 2 months ago