[CI] Fix and wire encoder/manager cudagraph unit tests (#49339)

Nick Hill committed 3 months ago

[CI][Bugfix] Reduce max_model_len in OOT embedding test to fix KV-cache OOM on small GPUs (#49351)

Flora Feng committed 3 months ago

Fix Qwen3-VL M-RoPE on the Transformers modeling backend (grids + compile) (#49292)

Aritra Roy Gosthipaty committed 3 months ago

[Bugfix][Spec Decode] Restrict embedding-width share guard to EAGLE drafts (#47953)

Evan Takahashi committed 3 months ago

[Bugfix] Fix DSA crash under breakable piecewise cudagraphs (#49302)

Nick Hill committed 3 months ago

[ROCm][CI] Fix AITER MLA fp8 decode metadata regression test (#48845)

stefankoncarevic committed 3 months ago

[ROCm] Bump AITER to v0.1.16.post5 (#48683)

Fangzhou Ai committed 3 months ago

Ci/add laguna xs gsm8k (#49241)

Michael Goin committed 3 months ago

[CI] Wire tests/models/inkling into a B200 job (#49325)

Nick Hill committed 3 months ago

[Test] Add DeepSeek MTP parallel-load tests (#41653)

Stefano Castagnetta committed 3 months ago

[Bugfix] Prefix-cache metrics double-counted when a KV connector defers requests (#48860)

Seiji Eicher committed 3 months ago

[Bugfix] Handle MLA fallback during FA4 JIT warmup (#49306)

Roberto L. Castro committed 3 months ago

[XPU][DeepSeekV4]Add DeepSeek-V4 fuse_index_q SYCL kernel path (#45991)

Avery Miao committed 3 months ago

[BugFix] Set graph_pool_id before FULL CUDA graph capture in ModelRunner V2 (#48843)

Ilya Markov committed 3 months ago

[Misc] Use VLLMValidationError in chat completion tool and batch validators (#49214)

Umut Polat committed 3 months ago

[Bugfix][CPU] Fix Clang OpenMP build on macOS (#49021)

Sya committed 3 months ago

[bugfix] Fix Cosmos3 Edge checkpoint weights filtering, video loading, prompt expansion (#49190)

bastefaniak committed 3 months ago

[3/N][KV-Cache Layout Refactor] Standardize Mamba cache; drop `get_transfer_cache_regions` (#44456)

Lucas Wilkinson committed 3 months ago

[Rust][Benchmark] Use async HTTP clients (#49295)

Bugen Zhao committed 3 months ago

[Misc] Remove old now unsupported `max_num_partial_prefills` and `max_long_partial_prefills` (#49244)

Nicolò Lucchesi committed 3 months ago