Pull Requests huggingface/peft

fix: layers_to_transform now correctly matches layer index on MoE models

#3028 opened 2026-02-07 19:56 by Mr-Neutr0n

fix: correct typo 'occurence' to 'occurrence'

#3027 opened 2026-02-07 12:57 by thecaptain789

[TinyLoRA] initial tinylora implementation

#3024 opened 2026-02-06 17:41 by kashif

Fix docker CPU build

#3023 opened 2026-02-05 11:00 by githubnemo

Add tooling to bump action version

#3022 opened 2026-02-04 12:44 by githubnemo

update get_error_factor to cache up with the latest transformers change

#3021 opened 2026-02-04 06:01 by jiqing-feng

Fix moe layers to transform

#3017 opened 2026-02-01 11:48 by MichalMraz

fix: eliminate cross-terms bug when combining adapters with different weights

#3013 opened 2026-01-30 02:14 by wingding12

Upgrade GitHub Actions for Node 24 compatibility

#3008 opened 2026-01-22 23:17 by salmanmkc

[feature] Tiny modification to enable OFT for finetuning embedding layers

#3005 opened 2026-01-21 17:23 by zqiu24

Fix error of PEFT with disable adapters and FSDP

#3001 opened 2026-01-19 10:48 by Isalia20

Add get base model state dict

#3000 opened 2026-01-16 11:23 by Isalia20

[WIP] Support transformers weight conversion

#2995 opened 2026-01-14 17:31 by BenjaminBossan

Add AdaMSS tuner with Adaptive Subspace Allocation (ASA)

#2987 opened 2026-01-10 07:12 by LonglongaaaGo

fix: auto-untie word embeddings on merge_and_unload when both are adapted

#2972 opened 2025-12-31 09:25 by www-spam

Implements EWoRA in PEFT

#2971 opened 2025-12-31 09:17 by HarshKohli

intorduce AdaDoRA implementation using both DoRA and AdaLoRa together

#2969 opened 2025-12-25 13:16 by OrMullerHahitti

Add UniLoRA tuner to PEFT

#2968 opened 2025-12-23 16:14 by KaiyangLi1992

Add DataLoader to MathQA benchmarking script

#2954 opened 2025-12-14 15:53 by ParagEkbote

Integration of PVeRA

#2952 opened 2025-12-11 18:54 by leofillioux

[FEAT] Add FeRA (Frequency-Energy Constrained Routing Adaptation) method

#2951 opened 2025-12-11 15:35 by YinBo0927

FIX: warmup_ratio deprecated (fixes #2949) wait-transformers-v5

#2950 opened 2025-12-11 15:10 by shantanugupta2004

Intergrate MonteCLoRA (TMLR 2025 accepted) into PEFT

#2943 opened 2025-12-09 08:07 by victor7246

Fix: Respect `inference_mode` when setting adapters with `modules_to_save` (Issue #2928)

#2931 opened 2025-11-29 13:09 by ada-ggf25

ENH: Tie weights for target_modules in Lora (#2864)

#2879 opened 2025-10-29 06:01 by romitjain

WoRA integration into PEFT

#2872 opened 2025-10-26 20:45 by sambhavnoobcoder

Integrate ABBA to LoHA

#2842 opened 2025-10-16 05:34 by monatis

[FEAT] Add LoReTTA

#2818 opened 2025-10-08 10:24 by mbaddar1

Add conv2d support for fourierft and other improvements

#2794 opened 2025-09-21 19:44 by frutiemax92

[WIP] Update `LoraConfig` for KaSA implementation

#2698 opened 2025-08-02 05:41 by iambogeumkim