All pull requests

Pull requests

• TST Speed up the Mixtral weight conversion tests on CPU

  triaged
  #4023·peft-jambot opened Oct 20, 2026Contributor·

  ·
  7

• FIX DeFT, TinyLoRA and PSOFT second adapter overwriting fan_in_fan_out

  triaged
  #4022·Zhuoxi2000 opened Oct 20, 2026·

  ·
  1

• TST Raise bf16 tolerance in state dict regression tests

  triaged
  #4020·peft-jambot opened Oct 20, 2026Contributor·

  ·
  2

• docs: improve LayerNorm Tuning documentation

  triaged
  #4011·bappisinsane opened Oct 19, 2026·

  ·
  1

  7

• docs: clarify AdaLoRA importance measurement and training phases

  triaged
  #3984·bappisinsane opened Oct 15, 2026·2 tasks done·

  ·
  6

• FIX make adapter injection atomic on validation failure

  triaged
  #3981·eSVeeF opened Oct 14, 2026Contributor·

  ·
  1

  8

• Broadcast full adapter weights

  triaged
  #3965·michaelbenayoun opened Oct 13, 2026Member·

  ·
  11

• TST Add tests for merging LoRA with tied weights

  triaged
  #3954·BenjaminBossan opened Oct 13, 2026Member·

  ·
  1

• FIX Preserve CPT input token roles

  triaged
  #3953·aswanth-07 opened Oct 13, 2026Contributor·

  ·
  1

  18

• Add a TiedLinear wrapper/proxy for lm_head when using tied embeddings

  triaged
  #3948·winglian opened Oct 12, 2026Contributor·

  ·
  3

  

• FIX Healthcheck now also works for prompt learning

  triaged
  #3944·BenjaminBossan opened Oct 12, 2026Member·

  ·
  1

• FIX Support MoE expert target_parameters with TP/EP

  triaged
  #3942·DaoyuanLi2816 opened Oct 12, 2026Contributor·

  ·
  1

  4

• FIX Reshard on disable_adapter() exit only when needed

  triaged
  #3938·hivaze opened Oct 12, 2026Contributor·

  ·
  4

• FEAT Add PaCA (Partial Connection Adaptation, ICLR 2025)

  triaged
  #3927·WooSunghyeon opened Oct 10, 2026·9 tasks done·

  ·
  1

  60

• Fix PeftMixedModel.disable_adapter state restoration

  triaged
  #3910·YueshenZ opened Oct 8, 2026·

  ·
  1

• NoRA

  triaged
  #3908·Joluck opened Oct 8, 2026Contributor·

  ·
  22

  

• FIX Apply OFT module_dropout to the rotation blocks again

  triaged
  #3905·Nicholas022400701 opened Oct 8, 2026Contributor·

  ·
  1

  4

• FIX Validate the requested adapter configuration during hotswap

  triaged
  #3865·DaoyuanLi2816 opened Oct 2, 2026Contributor·

  ·
  1

  4

• FIX Preserve trainability when loading mixed adapters

  triaged
  #3862·aswanth-07 opened Oct 1, 2026Contributor·

  ·
  1

  20

• FIX Avoid making ranknum in AdaLoRA trainable by accident

  triaged
  #3859·BenjaminBossan opened Oct 1, 2026Member·

  ·
  1

• aLoRA: speed up forward pass

  triaged
  #3847·githubnemo opened Sep 29, 2026Collaborator·

  ·
  3

• FIX add_weighted_adapter with rank_pattern for element-wise combinations

  triaged
  #3841·amogh-nagri-11 opened Sep 29, 2026Contributor·

  ·
  1

  4

• FIX Adapter shape for FSDP2 layers on meshes not named "fsdp"

  triaged
  #3838·hivaze opened Sep 28, 2026Contributor·

  ·
  1

  36

• FIX Several fixes to Conv2d layers, extend test coverage

  triaged
  #3837·BenjaminBossan opened Sep 28, 2026Member·

  ·
  2

  1

• FIX strict_adapter_check for load_adapter and from_pretrained

  triaged
  #3832·rupeshpoojary9 opened Sep 28, 2026Contributor·

  ·
  1

  14