[Scheduler] Tune --long-prefill-token-threshold adaptiveness (#58459)

Robert Shaw committed 28 days ago

[CI][ROCM] Add the Fusion E2E TP2 Quick group on MI355, and the AITER MLA fix it needs (#58369)

stefankoncarevic committed 28 days ago

[Test][Determinism] Cover chunked prefill in the batch-invariance suite (#55612)

Bob Ok committed 28 days ago

[Bugfix][DSA] Bound DeepSelect sentinel columns in the sparse top-k remap (#58215)

Juntian Liu committed 28 days ago

[CI] Share BF16 baselines across quantization comparison tests (#58469)

Aarushi Jain committed 28 days ago

[5/12][ci-selector][CI] Skip the Proton GPU test when another CUPTI tool is injected (#58455)

Kevin H. Luu committed 28 days ago

[Bugfix][MRV2] Align dummy idx_mapping dtype to avoid runtime jit (#58462)

Nick Hill committed 28 days ago

[CI] Disable JIT warmup by default in VllmRunner (#58452)

Misha Goin committed 28 days ago

[CI][Bugfix] Limit MRV2 sampler JIT warmup registration to ROCm (#58465)

Kevin H. Luu committed 28 days ago

[ROCm][Bugfix] Fix TileLang mHC fused RMSNorm on 64-wide wavefronts (#58419)

djramic committed 28 days ago

[ROCm][Test] Cover MoRI graph replay and output lifetime (#58093)

Andreas Karatzas committed 29 days ago

[Docs] Fix docstring typos (output_dytpe, kwrags, Abbrivations) (#55936)

陈志谦 committed 29 days ago

[Bugfix] Pick a KV block size supported by every attention backend (#49845)

Divy committed 29 days ago

[CI] Select one GPU for the H200 initialized snapshot E2E step (#58351)

Thang Nguyen committed 29 days ago

[Perf] Use breakable CUDA graphs (no torch.compile) by default under VLLM_BATCH_INVARIANT so the tuned matmul configs see the runtime M (#57586)

Lio Einaudi committed 29 days ago

Doc: add DiffusionGemma to supported models (#46466)

CBP committed 29 days ago

[Bugfix][ROCm] Fix startup OOM in AITER MLA FP8 prefill workspace sizing (#57923)

Simon Danielsson committed 29 days ago

[Bugfix][NIXL] Restore successful push completion reporting (#58188)

Dao007forever committed 29 days ago

[CI][ROCm] Add an MI355 Kimi-K3 unit test group (#58012)

Oxana Korzh committed 29 days ago

[ROCm][CI] Validate Mooncake and NIXL prefill/decode accuracy (#58095)

Andreas Karatzas committed 29 days ago