Add PrefixRepetitionRandomDataset to `vllm bench serve` datasets (#20638)

Seiji Eicher committed about 1 year ago

[Fix] enable swap_ab for pplx problem size computation (#22991)

shixianc committed about 1 year ago

[CI] Speed up Whisper tests by reusing server (#22859)

Michael Goin committed about 1 year ago

Use regex in convert-results-json-to-markdown.py (#22989)

Michael Goin committed about 1 year ago

[Bugfix] Added more env vars to hash (#22449)

nvjullin committed about 1 year ago

[Core] direct indexing on self.block_table_np in compute_slot_mapping (#22940)

Zebing Lin committed about 1 year ago

[Log] Debug Once for Randomizing dummy data for DP Rank (#22860)

Wentao Ye committed about 1 year ago

[Model] Granite-4 support loading quantized checkpoint (#22925)

Chih-Chieh Yang committed about 1 year ago

[Kernels] Clean up FusedMoeMethodBase and modular kernel setup. Remove extra arguments from modular kernel methods. (#22035)

bnellnm committed about 1 year ago

[Structured Output] Make the output of structured output example more complete (#22481)

Shanshan Shen committed about 1 year ago

[Benchmarks] Include image data when ShareGPT4V dataset is used. (#22955)

Chenheli Hua committed about 1 year ago

[FIXBUG] Correctly Apply Grammar Bitmask in Mixed Batches (#22896)

JartX committed about 1 year ago

[BugFix] Skip the Q component for QKVParallelLinear in the case of QKVCrossParallelLinear since its width is 0 (#22369)

sstamenk committed about 1 year ago

[V0 Deprecation] Remove advance_step (#22969)

Woosuk Kwon committed about 1 year ago

[Core] Allow full cudagraph with separate attention routines and orthogonal to compilation, add support for FA2 and FlashInfer (#20059)

fhl2000 committed about 1 year ago

[Frontend] Expose do_log_stats interval to env (#22905)

Csrayz committed over 1 year ago

[CI] Remove duplicated docs build from buildkite (#22924)

Harry Mellor committed over 1 year ago

[Misc] Ignore ep_kernels_workspace (#22807)

Jee Jee Li committed over 1 year ago

[V1] [Hybrid] Support using float32 for state in Hybrid Models (Mamba2, Mamba1, Minimax) (#22928)

Thomas Parnell committed over 1 year ago

Improve multimodal hasher performance for re-used Image prompts (#22825)

Staszek Paśko committed over 1 year ago