[Bugfix] Allow 64-bit integer values for LoRA IDs to avoid overflow/truncation (#27876)

Madeesh Kannan committed about 1 year ago

[CI/Build] Add gpt-oss LoRA test (#27870)

Jee Jee Li committed about 1 year ago

[Perf] Decouple torch op from GDA to leverage torch.compile (#27871)

Jiangyun Zhu committed about 1 year ago

[BugFix] Don’t compute reorder threshold when there are no attention groups (#27861)

Huamin Li committed about 1 year ago

[Bugfix] Avoid too small block m/n for FlexAttention kernel option (#27853)

Isotr0py committed about 1 year ago

[Kimi-Linear] Correct prefixes and add compatibility to AWQ quants (#27834)

toncao committed about 1 year ago

[Hardware][Powerpc] Fix VLLM_CPU_OMP_THREADS_BIND="auto" low CPU utilization for Power (#27734)

Akash kaothalkar committed about 1 year ago

[benchmark] Make request IDs unique across clients by default (#27723)

Seiji Eicher committed about 1 year ago

[CI Test] Add Scheduled Integration Test (#27765)

Wentao Ye committed about 1 year ago

[CI/Build][Intel] Enable performance benchmarks for Intel Gaudi 3 (#26919)

Jakub Sochacki committed about 1 year ago

[Misc] Make all tool scripts executable (#27831)

Matthew Bonanni committed about 1 year ago

[BugFix] Fix broken import in initialize_ray_cluster() (#27838)

Nick Hill committed about 1 year ago

[Feature] Batch invariant torch.compile (#27660)

Paul Zhang committed about 1 year ago

[Core][Perf] Only invoke save_new_computed_blocks when computed blocks are not empty (#27799)

Jialin Ouyang committed about 1 year ago

[Refactor] Remove `VLLM_DEEPEP_LOW_LATENCY_ALLOW_NVLINK` (#27750)

Wentao Ye committed about 1 year ago

[Fix] Skip `record_sleep_state` logic in `PrometheusStatsLogger` if not in dev mode (#27789)

Sumanth R Hegde committed about 1 year ago

[EP/DP][API Server] Enable DP-aware routing in OpenAI API requests (#24945)

cong-meta committed about 1 year ago

[Core][Bookkeeping] Update cu_num_accepted_tokens for all req_index (#27629)

Jialin Ouyang committed about 1 year ago

[Bugfix] Fix 2 precommit issues - (mamba_block_size, kv_cache_config) (#27811)

Tyler Michael Smith committed about 1 year ago

[Model] Use the same fused_moe configs for all H200 devices (#23642)

Roger Meier committed about 1 year ago