All pull requests

Pull requests

• Add Raon-1 mini inference support

  #2015·asynclee opened Oct 26, 2026·

  ·

• Fix speculative processor history

  #2014·pierreablin opened Oct 26, 2026·

  ·

• Add sparse attention optimization for DeepSeek-V3.2

  #2013·vinceclifford opened Oct 26, 2026Contributor·

  ·

• Fix MiniCPM RoPE theta default

  #2012·liuyunxian-yana opened Oct 26, 2026·

  ·
  1

• fix head dims rotated, only part gets rotated using partial_rotary_factor instead of all of them

  #2011·begcdn opened Oct 26, 2026·

  ·

• Fix shared-memory race in KL and JS divergence kernels

  #2010·JayeshSuryavanshi opened Oct 26, 2026·

  ·
  1

• fix(gpt_oss): detect transformers [E, H, 2F] layout in sanitize() (fixes #1994)

  #2009·kru2710shna opened Oct 25, 2026·1 task done·

  ·
  1

• Fix quantized KV cache packed storage width

  #2008·sehejjain opened Oct 25, 2026·

  ·

• apply the checkpoint's sliding_window setting when no layer_types list is present

  #2007·begcdn opened Oct 25, 2026·

  ·

• Skip fully masked key tiles in Flint sliding-window prefill

  #2006·tc3oliver opened Oct 24, 2026·

  ·

• Fix Jamba MoE routing weights to match reference

  #2004·Knightkolla opened Oct 23, 2026Contributor·

  ·
  1

• properly pass --clear-cache-threshold to LoRA trainer

  #2000·Lumonopolis opened Oct 22, 2026·

  ·

• fix(gemma3n): apply KV sharing without cache (#1674)

  #1999·Yu-Haikuo opened Oct 22, 2026·

  ·
  1

• fix: handle nested parameter tags in Qwen3 Coder parser

  #1996·rafiimam opened Oct 21, 2026·

  ·
  1

• Fix nested parameter tags in qwen3_coder tool parser

  #1993·axiom-of-choice opened Oct 19, 2026Contributor·

  ·
  1

  1

• Add NVIDIA Nemotron-Labs Dory model

  new_model
  #1988·pmichalas opened Oct 17, 2026·1 task done·

  ·
  2

• feat(qwen3_5): honour explicit layer_types over full_attention_interval

  #1987·hermitdave opened Oct 17, 2026·

  ·

• Add --loss-chunk-size to reduce LoRA training memory with large vocabularies

  #1983·spicysharkco opened Oct 16, 2026·

  ·

• nemotron_h: support Nemotron 3.5 Bolt (heterogeneous Puzzle expert counts, omni key layout)

  new_model
  #1982·vichzp opened Oct 16, 2026·

  ·

• Fix pipeline load when a shard holds another stage's scales

  #1980·ashokgelal opened Oct 15, 2026·

  ·

• Fix multi-token attention mask in RotatingKVCache

  #1978·robertcprice opened Oct 15, 2026·

  ·

• Fix silent drop of unparseable tool calls in server

  #1972·santanu20 opened Oct 14, 2026·

  ·
  1

• qwen3_moe honors rope_scaling; handles transformers >= 5

  #1968·hierophantos opened Oct 13, 2026·

  ·

• Qwen3-Coder parser loses precision when parsing large integer tool arguments

  #1967·anidotdev opened Oct 12, 2026·

  ·

• Fix /v1/models sending 200 before the cache scan can fail

  #1965·HamedShams opened Oct 12, 2026·

  ·
  1

  1