[Bugfix] Return the tokenizer from maybe_make_thread_pool so it survives pickling (#45460)

WEI CHENG CHIU committed 5 months ago

[Render] Add `/derender` endpoints for disaggregated postprocessing (#43606)

Martin Hickey committed 5 months ago

[CI] Wait for SSL cert refresher events in the test (#45489)

Andreas Karatzas committed 5 months ago

[Bugfix] Replace deprecated Qwen2VLImageProcessorFast with Qwen2VLImageProcessor (#42700)

longguo committed 5 months ago

[Bugfix] Set type/role explicitly in streaming message_start event (#45376)

WEI CHENG CHIU committed 5 months ago

[BugFix] Avoid prematurely freeing cached mm encoder outputs (#45347)

Nick Hill committed 5 months ago

[Bugfix][CPU] Don't build triton-cpu on arm64 release image (#45401)

Kevin H. Luu committed 5 months ago

[ROCm][DSV4][Perf] Fuse inverse-RoPE and cache bf16 wo_a in o-projection (#45103)

Fangzhou Ai committed 5 months ago

Add the QuantizedActivation linear-kernel contract (#44260)

Michael Goin committed 5 months ago

[Kernel] Consolidate Marlin thread-tile padding across all dense Marlin paths (#45295)

Michael Goin committed 5 months ago

[Model Runner v2] Migration from v1 to v2, with Qwen and DSv2 MOE models [3/N] (#42667)

Wentao Ye committed 5 months ago

[Attention] Improve attention benchmarks: configs and profiling (#39336)

Matthew Bonanni committed 5 months ago

[ROCm] Bump Torch to 2.11 (#45362)

Micah Williamson committed 5 months ago

[Refactor] Deprecate ResponsesParser wrapper, inline parsing into ParsableContext (#45431)

Flora Feng committed 5 months ago

[Core][AMD] Propagate shutdown timeout to MultiprocExecutor (#43154)

Ryan Rock committed 5 months ago

[Perf] Use native DSA indexer decode path for next_n > 2 on SM100 (#45322)

qizixi committed 5 months ago

[Migration] Migrate GGUF quantization support to plugin (#39612)

Isotr0py committed 5 months ago

[Model][Dflash] Enable Dflash support for Qwen3NextForCausalLM targets (#45319)

Jonas I. Liechti committed 5 months ago

[Kernel][Helion][1/N] Add Helion kernel for dynamic_per_token_scaled_fp8_quant (#33790)

Xiaohong (Sean) Chen committed 5 months ago

[Model] Add encoder CUDA graph support to Lfm2VL (#44930)

Yi Zhong committed 5 months ago