[XPU] Fix Eagle3 initialization on XPU (#43957)

Chaojun Zhang committed 4 months ago

[XPU] Fix CUDA API shims breaking Torch Dynamo during AOT compile (#43092)

Łukasz Ślusarczyk committed 4 months ago

[ROCm][Test] Fix test_per_token_group_quant_fp8 tolerance for 1-ULP FP8 rounding on gfx950 (#46944)

Spandan Tiwari committed 4 months ago

[MRV2] Enable mm prefix bidi attention support on MRV2 (#46942)

Isotr0py committed 4 months ago

[Bugfix][Voxtral Realtime] Fix token feedback timeout silent hang (#44461)

Ting SUN committed 4 months ago

[Bugfix][Model] Fix crash loading Mamba/Mamba2 checkpoints without an `architectures` field (#46037)

Ting SUN committed 4 months ago

[Feature] Support sequence parallel without the need for DP, 1.9%~5.0% E2E Throughput Improvement (#47070)

Wentao Ye committed 4 months ago

[Bugfix][TurboQuant] Preserve KV cache dtype in backend shape (#47609)

Lucas Wilkinson committed 4 months ago

[Bugfix][Pooling] Forward instruction to Jina reranker scoring prompts (#47590)

Ting SUN committed 4 months ago

[Bugfix][Gemma4] Fix FA4 mm_prefix mask: add sliding window and absolute q_idx (#47332)

Luciano Martins committed 4 months ago

[Bugfix][Model] Allow Run:ai memory_limit sentinel values (#47337)

Ting SUN committed 4 months ago

[ROCm][CI] Fix Kernels and Kernels attention test failures (#47519)

Carl Persson committed 4 months ago

[Bugfix][Frontend] Preserve default sampling params in batch chat (#47597)

Ting SUN committed 4 months ago

[Perf] Remove redundant op for GLM 5.2 (#47198)

Wentao Ye committed 4 months ago

[Bugfix][Frontend][gpt-oss] Recover raw tail when Harmony parser ends non-terminal (#47379)

yzong-rh committed 4 months ago

[CI] Bump `huggingface-hub` from `v1.10.2` to `v1.22.0` (#47551)

Harry Mellor committed 4 months ago

[Misc] Validate Pooling cache_salt Values (#46966)

Taneem Ibrahim committed 4 months ago

[Misc] Update request-extras parity for batch chat completion (#47333)

Taneem Ibrahim committed 4 months ago

[Misc] Forward request-level prompt extras for cross-encoder scoring (#46939)

Taneem Ibrahim committed 4 months ago

[Misc] Preserve cross-encoder pooling extra kwargs (#47082)

Taneem Ibrahim committed 4 months ago