[GLM5.3 Perf] Optimize glm 5.3 metadata op, 1.6~4.8x kernel level performance improvement (#58450)

Wentao Ye committed 27 days ago

[Bugfix][GLM-5.3-Flash] kpool corruption with speculative decoding (#58454)

Matt Mastracci committed 27 days ago

[Benchmark] Record model_id in bench latency/throughput --output-json (#58112)

Yashasvi Chaurasia committed 27 days ago

[Bugfix][GLM-5.3-Flash] SM90 sparse MLA: index_kpool mismatch leads to corruption via unread query token (#58704)

Matt Mastracci committed 27 days ago

[Bugfix][Frontend] Keep logprobs of parser-suppressed streaming chunks (#58583)

Denis Ermakov committed 27 days ago

[Bugfix] Stop allocator fragmentation from shrinking the KV cache during memory profiling (#58430)

Robert Shaw committed 27 days ago

[Core][Logging] Fix JSON logging process decoration (#57957)

Mark McLoughlin committed 27 days ago

[Perf][Kernel] Vectorized flat abs-max for dynamic per-tensor FP8 quantization (#58194)

Monishver committed 27 days ago

[Kernel][Perf] Register-resident path for per-token-group 8-bit quant (#55330)

chuan932 committed 27 days ago

[Pooling] Preserve reranker tokenization with document limits (#57666)

Taneem Ibrahim committed 27 days ago

[Bugfix] Fix generative scoring body cancellation (#57729)

Taneem Ibrahim committed 27 days ago

[MRV2] Minor model_runner.py code cleanup (#58610)

Nick Hill committed 27 days ago

[SpecDecode] Add LiLiCorr drafter (#57934)

Andrii Skliar committed 27 days ago

[ROCm][CI] Pass weight_shape in MXFP8 block32 linear tests (#58698)

djramic committed 27 days ago

[Core] Model console logging as CLI configuration (#57205)

Mark McLoughlin committed 27 days ago

[CI] Shard (H100) Helion Kernels five ways (#58645)

Thang Nguyen committed 27 days ago

[Docs] Add return annotation to `fused_mm_input_norm_triton` (#58687)

Harry Mellor committed 27 days ago

[MoE] Defer the TRTLLM-Gen top-k finalize on the modular path (#58635)

Yongye Zhu committed 27 days ago

[UX][Frontend] Introduce `vllm preload` cli for fast restart (#56680)

Isotr0py committed 27 days ago

[Perf][DSv4] Fuse inverse RoPE + FP8 quant into FlashInfer sparse MLA (#58621)

Yongye Zhu committed 27 days ago