[AMD][OCP MX][CI] Fix tests to not dispatch on `UNFUSED_TRITON` backend on MI300, improve w_mxfp4_a_fp8 emulation support (#46142)

fxmarty-amd committed 4 months ago

[ROCm][CI] Fix nixl tests (#45219)

Andreas Karatzas committed 4 months ago

[Minimax-M3] BF16/FP8 Indexer using MSA (#45892)

Yongye Zhu committed 4 months ago

[Bugfix]: Fix unquantized gpt-oss weight loading broken by FusedMoE r… (#45818)

Priyansh Jain committed 4 months ago

[CI][ROCm] Skip unsupported test cases on ROCm (#46160)

fxmarty-amd committed 4 months ago

[ROCm][CI] fix fp8 range in vit_fp8_quant (#46410)

Divakar Verma committed 4 months ago

[Docs] Add Qwen3 forced alignment online example (#46197)

Taneem Ibrahim committed 4 months ago

fix(moe_wna16): access tp_size via moe_config for RoutedExperts compatibility (#45404)

Willow Lopez committed 4 months ago

[ROCm][CI] pass merge-base to container for python-only wheel metadata (#45869)

Divakar Verma committed 4 months ago

[MyPy] Fix mypy for `vllm/benchmarks` (#39896)

Martin Hickey committed 4 months ago

Add weights padding for fp8 per-block online quantization (#44763)

Yan Ma committed 4 months ago

[ROCm][Test] Fix stale test_gfx950_moe MXFP4 oracle tests (#46260)

Spandan Tiwari committed 4 months ago

[ROCm][CI] Skip Quark mxfp4 tests unless Quark version is compatible with Torch version (#46431)

Micah Williamson committed 4 months ago

[CI/Build] Remove BaiChuanForCausalLM from the LoRA test (#46494)

Jee Jee Li committed 4 months ago

[LoRA] Gate all_gather on fully_sharded_loras inside _mcp_apply; rewrite regression test (#45715)

lcheng committed 4 months ago

[Bugfix] GPT-OSS Autodrop reasoning in Response API and cleanup (#45048)

yzong-rh committed 4 months ago

[HARDWARE][POWER] Enable fp16 support for PowerPC (#46135)

Rukhaiya2004 committed 4 months ago

Filter Pydantic-internal markers from validation error param (#46457)

Muhammad Fawaz committed 4 months ago

[Bugfix] Responses API assistant EasyInputMessageParam input (#44361)

yzong-rh committed 4 months ago

[Bugfix][KV Offload] Fix swap_blocks_batch on the default stream (#46379)

Itay Etelis committed 4 months ago