Add TorchCodec as a video decoding backend (#46609)

Nicolas Hug committed 4 months ago

fix(distributed): propagate distributed_timeout_seconds to NCCL device groups (#45159)

jialoop-git committed 4 months ago

feat(cpu): add CPU support for Mamba ShortConv (#35059)

Rahul Vishwakarma committed 4 months ago

[Rust Frontend] Add DeepSeek V3.2 roundtrip fixture (#47619)

Reid committed 4 months ago

[Bugfix] [Quantization] Fix loading for CT DSV2 (#47780)

Kyle Sayers committed 4 months ago

[XPU] Fix PP accuracy on XPU device (#47253)

YiSheng5 committed 4 months ago

[Frontend] add per-request timing `metrics` field to response body of Chat/Completions APIs (#46768)

nv-nedelman-1 committed 4 months ago

[ROCm][CI][Bugfix] Fix flaky parallel tool-call streaming (test assertion + Mistral/Granite parsers) (#47550)

akii96 committed 4 months ago

[ROCm][CI] Adding Rust parity (#47478)

Andreas Karatzas committed 4 months ago

[Bugfix][ROCm] Fix memory access fault in AITER MLA backend for DPA+FP8 KV (#47276)

Simon Danielsson committed 4 months ago

[Bugfix] Fix dp mtp hang (#40589)

SherryC41 committed 4 months ago

[Quantization] add humming moe backend to all dense/moe oracles (#41652)

Jinzhen Lin committed 4 months ago

[Bugfix]Fix DeepSeek-V4 fp8_ds_mla KV cache reshape (#47716)

Zhiyi Hong committed 4 months ago

[CI] Fix some errors on `main` (#47726)

Harry Mellor committed 4 months ago

Enable B12x backend for non-gated MoEs (like Nemotron) (#43328)

Andrii Skliar committed 4 months ago

[Bugfix][Model] Add stability window to DiffusionGemma to match HF stability_threshold semantics (#45965)

Nathaniel McVicar committed 4 months ago

[Kernel][Helion][1/N] Add Helion kernel for silu_and_mul_per_block_quant (#43994)

Xiaohong (Sean) Chen committed 4 months ago

[Bugfix] Preserve FP8 indexer WK pairs across incremental load_weights (#46168)

lcheng committed 4 months ago

[macOS][CPU][Installation] Fix the broken installation of vllm 0.24.0 in macos + cpu (#47457)

Ranran committed 4 months ago

Make the Transformers modeling backend as fast as native vLLM (#47187)

Harry Mellor committed 4 months ago