[BugFix][Spec Decode] Use float64 for uniform_probs (#23803)

Woosuk Kwon committed about 1 year ago

[CI/Build][Bugfix] Fix Qwen VL tests on CPU (#23818)

Li, Jiang committed about 1 year ago

[Doc]: fix typos in .md files (including those of #23751) (#23825)

Didier Durand committed about 1 year ago

[FIXBUG] Add return_success parameter to moe_wna16_weight_loader function (#22797)

JartX committed about 1 year ago

[Model] Add PP support and VLM backbone compatability for GPT-OSS (#23680)

Isotr0py committed about 1 year ago

[New Model]: Support GteNewModelForSequenceClassification (#23524)

wang.yuqi committed about 1 year ago

[Kernel] cuda kernels for upcoming decode context parallel feature (#23791)

yzds committed about 1 year ago

[Bugfix] when set offline model running error (#23711)

rongfu.leng committed about 1 year ago

[CI] enable idefics3 and fuyu-8b test in multimodal test (#23790)

Jiangyun Zhu committed about 1 year ago

[Transform] [Quantization] Add transforms to compressed tensors (#22486)

Kyle Sayers committed about 1 year ago

Add deprecation warning for lora_extra_vocab_size (#23635)

Jinheng committed about 1 year ago

[CI] make all multi-gpu weight loading tests run nightly (#23792)

Alex committed about 1 year ago

Gracefully handle edge cases in harmony utils (#23155)

Jan Kessler committed about 1 year ago

[Perf] Tune configs for triton block fp8 gemm H100/H200 (#23748)

Michael Goin committed about 1 year ago

[Feature] models: pass layer prefix to replace_linear_class for per-layer quantization routing. Addresses #23239 (#23556)

Shrey Gupta committed about 1 year ago

Migrate Qwen inputs to TensorSchema (#23473)

Benji Beck committed about 1 year ago

[Bugfix] Fix for V1 priority scheduling crashes at preemption (#23713)

Hanchenli committed about 1 year ago

[Feature] Add `VLLM_DISABLE_PAD_FOR_CUDAGRAPH` to Avoid Hang Issue (#23595)

Wentao Ye committed about 1 year ago

[Bugfix] Fix Marlin NVFP4 for modelopt (#23659)

Michael Goin committed about 1 year ago

DP/EP Support for gpt-oss with deepep-ht comm kernel on SM100 (#23608)

Yongye Zhu committed about 1 year ago