Massimo Bini 2813b9c4bf FEAT Add DeLoRA (#2780) ...

Implements DeLoRA: "Decoupling Angles and Strength in Low-rank
Adaptation" (https://huggingface.co/papers/2503.18225).

Similar to DoRA, DeLoRA decouples the angular learning from the
adaptation strength, but it also allows to limit the norm of the change.
This way, DeLoRA promises to reduce the risk of catastrophic forgetting
and to be more robust to hyper-parameter settings such as the learning
rate.

2025-10-17 16:24:46 +02:00

..

bnb

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

regression

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

__init__.py

TST: torch compile tests (#1725)

2024-05-17 18:03:27 +02:00

conftest.py

CI Testing transformers deprecations (#2817)

2025-10-13 16:53:35 +02:00

test_adaption_prompt.py

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

test_arrow.py

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

test_auto.py

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

test_boft.py

ENH BOFT don't save boft_P buffer (#2050)

2024-09-13 13:56:47 +02:00

test_bufferdict.py

Trainable Tokens: Support for Weight Tying (#2399)

2025-03-06 14:09:01 +01:00

test_common_gpu.py

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

test_config.py

ENH: Store PEFT version in PEFT config file (#2782)

2025-09-30 11:09:18 +02:00

test_cpt.py

FIX: CPT should not be tested with sequence classification (#2507)

2025-04-23 18:46:27 +02:00

test_custom_models.py

FEAT Add DeLoRA (#2780)

2025-10-17 16:24:46 +02:00

test_decoder_models.py

FEAT Add DeLoRA (#2780)

2025-10-17 16:24:46 +02:00

test_encoder_decoder_models.py

FEAT Add DeLoRA (#2780)

2025-10-17 16:24:46 +02:00

test_feature_extraction_models.py

FEAT Add DeLoRA (#2780)

2025-10-17 16:24:46 +02:00

test_gptqmodel.py

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

test_gpu_examples.py

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

test_helpers.py

MNT Use Python 3.9 as RUFF target version (#2483)

2025-04-08 16:44:55 +02:00

test_hub_features.py

Use hub_online_once in trainable token tests (#2701)

2025-08-05 12:58:55 +02:00

test_incremental_pca.py

TST: Add more HF Hub model caching (#2682)

2025-07-30 20:02:07 +02:00

test_initialization.py

FEAT Add DeLoRA (#2780)

2025-10-17 16:24:46 +02:00

test_integrations.py

ENH Add LoRA multihead attention module (#1324)

2025-01-08 17:35:43 +01:00

test_lora_megatron.py

TST Make tests pass on Cambricon MLUs (#1747)

2024-06-06 10:44:03 +02:00

test_lora_variants.py

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

test_lorafa.py

FEAT Add LoRA-FA to PEFT (#2468)

2025-04-10 10:53:19 +02:00

test_loraplus.py

TST Enable BNB tests on XPU (#2396)

2025-03-06 16:18:47 +01:00

test_low_level_api.py

Fix module target edge cases (#2773)

2025-09-30 11:09:44 +02:00

test_mapping.py

FIX #2295: Warn when user reloads modified model (#2306)

2025-01-07 18:10:07 +01:00

test_mixed.py

FIX: Multiple active adapters with auxiliary layers (#2758)

2025-08-29 17:54:19 +02:00

test_multitask_prompt_tuning.py

CHORE Replace deprecated torch_dtype with dtype (#2837)

2025-10-16 14:59:09 +02:00

test_other.py

FIX Correctly determine no_split_modules (#2570)

2025-06-16 17:21:06 +02:00

test_poly.py

TST Use plain asserts in tests (#1448)

2024-02-14 16:43:47 +01:00

test_randlora.py

align xpu behavior w/ cuda (#2551)

2025-06-02 17:23:42 +02:00

test_seq_classifier.py

FEAT Add DeLoRA (#2780)

2025-10-17 16:24:46 +02:00

test_shira.py

FEAT Add SHiRA Adapters (#2584)

2025-07-14 11:16:10 +02:00

test_stablediffusion.py

ENH Make OFT faster and more memory efficient (#2575)

2025-06-26 14:27:03 +02:00

test_target_parameters.py

FIX Multiple issues with target_parameters (#2710)

2025-08-12 13:59:29 +02:00

test_torch_compile.py

FEAT Add MiSS as a replacement for Bone. (#2604)

2025-08-01 18:37:20 +02:00

test_trainable_tokens.py

FIX Handle embed scale for trainable tokens, LoRA (#2825)

2025-10-14 12:35:31 +02:00

test_tuners_utils.py

ENH: Add set_requires_grad method (#2807)

2025-10-13 16:54:16 +02:00

test_vblora.py

align xpu behavior w/ cuda (#2551)

2025-06-02 17:23:42 +02:00

test_vera.py

align xpu behavior w/ cuda (#2551)

2025-06-02 17:23:42 +02:00

test_vision_models.py

ENH Make OFT faster and more memory efficient (#2575)

2025-06-26 14:27:03 +02:00

test_xlora.py

FIX X-LoRA embed_scale support #2830 (#2831)

2025-10-14 15:54:15 +02:00

testing_common.py

FEAT Add DeLoRA (#2780)

2025-10-17 16:24:46 +02:00

testing_utils.py

CHORE: Clean up config kwargs in custom model tests (#2736)

2025-08-19 11:55:25 +02:00