huggingface/peft issues and pull requests

#262 - can i load two layer lora weights？

Issue - State: closed - Opened by songbaiTalk about 1 year ago - 3 comments

#259 - TorchScript Convert

Issue - State: closed - Opened by syngokhan about 1 year ago - 4 comments

#253 - So will the original larger model and lora model be loaded at the same time when inferring？

Issue - State: closed - Opened by BEILOP over 1 year ago - 2 comments

#251 - What to use for `model_id` when loading from a model trained locally and not published?

Issue - State: closed - Opened by philharmonikerzzy over 1 year ago - 6 comments

#246 - when I use it for chatglm ，error like this

Issue - State: closed - Opened by shyoulala over 1 year ago - 3 comments
Labels: solved

#245 - LoRA parameters in the model have `requires_grad=False`

Issue - State: closed - Opened by bhavnicksm over 1 year ago - 4 comments
Labels: solved

#244 - Implementation of adapter?

Issue - State: open - Opened by jzhang38 over 1 year ago - 5 comments
Labels: wip

#242 - Multi-gpu training still has issues

Issue - State: closed - Opened by macabdul9 over 1 year ago - 12 comments
Labels: solved

#242 - Multi-gpu training still has issues

Issue - State: closed - Opened by macabdul9 over 1 year ago - 11 comments
Labels: solved

#236 - loss 0 with load_in_8bit = True

Issue - State: open - Opened by chuckhope over 1 year ago - 8 comments

#235 - Adding Zero-Init Attention Adaptation

Issue - State: closed - Opened by shermansiu over 1 year ago - 3 comments
Labels: enhancement

#230 - Resume the train of a LoRA

Issue - State: open - Opened by milyiyo over 1 year ago - 2 comments
Labels: wip

#226 - Code for PEFT-LoRA tuning of bigscience/T0_3B on raft/twitter_complaints

Issue - State: closed - Opened by VukVukovic over 1 year ago - 4 comments

#221 - Instruction-FT chatGLM with LORA and Chinese self-instruct data

Issue - State: closed - Opened by vxfla over 1 year ago - 2 comments

#219 - Support for using LoRA on a PreTrainedModel with no task-specific head

Issue - State: closed - Opened by tgaddair over 1 year ago - 4 comments
Labels: solved

#217 - Very Slow Inference on PEFT-LORA fine-tunned FLAN-UL2

Issue - State: closed - Opened by TejaGollapudi over 1 year ago - 23 comments
Labels: solved

#216 - Example Finetune_opt_bnb_peft.ipynb runs in colab notebook but fails on my system with seemingly the same requirements installed: "expected scalar type Half but found Float"

Issue - State: closed - Opened by jahangmar over 1 year ago - 12 comments

#211 - How to use multiple lora at the same time for text generation?

Issue - State: closed - Opened by fangqiang817 over 1 year ago - 5 comments

#207 - how to use peft at torch <=1.12.1 and CUDA <= 10.2 envrionment

Issue - State: closed - Opened by cqfxchina over 1 year ago - 10 comments

#206 - Can't find 'word_embedding' using Prompt-tuning with Deepspeed

Issue - State: closed - Opened by KaiLv69 over 1 year ago - 7 comments

#205 - Int8 training example error with multi-gpu - flan-t5

Issue - State: closed - Opened by djaym7 over 1 year ago - 18 comments
Labels: solved

#194 - tuning BLOOMZ 176B

Issue - State: closed - Opened by alex-ht over 1 year ago - 13 comments

#189 - I'm using pytorch 2 for faster training, getting thiserror

Issue - State: closed - Opened by imrankh46 over 1 year ago - 4 comments
Labels: solved

#186 - TypeError: dispatch_model() got an unexpected keyword argument 'offload_index'

Issue - State: closed - Opened by imrankh46 over 1 year ago - 17 comments

#184 - How do I continue training with PEFT?

Issue - State: closed - Opened by therealadityashankar over 1 year ago - 9 comments

#184 - How do I continue training with PEFT?

Issue - State: closed - Opened by therealadityashankar over 1 year ago - 7 comments

#161 - GPT2 Training GPU Memory Increase with LoRA and Zero 3

Issue - State: closed - Opened by dumpmemory over 1 year ago - 10 comments

#156 - RuntimeError: self and mat2 must have the same dtype

Issue - State: closed - Opened by imrankh46 over 1 year ago - 16 comments
Labels: solved

#151 - [examples/conditional_generation/peft_prefix_tuning_seq2seq.ipynb] need to fix the code

Issue - State: closed - Opened by 126pikaqiu over 1 year ago - 13 comments
Labels: solved

#148 - Expected Memory reduction when using PEFT for flant5-XL

Issue - State: closed - Opened by abdulvirta over 1 year ago - 7 comments

#137 - RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn

Issue - State: closed - Opened by PetroMaslov over 1 year ago - 15 comments
Labels: solved

#135 - is there a reason why generate with Prompt Tuning would be different for Seq2Seq_LM?

Issue - State: closed - Opened by aldopareja over 1 year ago - 5 comments

#131 - Error training GPT-2 with LoRA INT8

Issue - State: closed - Opened by holmad over 1 year ago - 7 comments

#128 - [SpeechToText] Wav2vec2 Support

Issue - State: closed - Opened by omarsou over 1 year ago - 9 comments

#128 - [SpeechToText] Wav2vec2 Support

Issue - State: closed - Opened by omarsou over 1 year ago - 10 comments

#118 - example on converting PEFT+INT8 trained model to ONNX for faster inference

Pull Request - State: closed - Opened by pacman100 over 1 year ago - 8 comments

#116 - Any support for masked language modeling?

Issue - State: closed - Opened by ryan-caesar-ramos over 1 year ago - 6 comments

#114 - AttributeError: 'DataParallel' object has no attribute 'config' when call get_peft_model.

Issue - State: closed - Opened by hjsg1010 over 1 year ago - 3 comments

#108 - ImportError: cannot import name 'prepare_model_for_training' from 'peft'

Issue - State: closed - Opened by jordancole21 over 1 year ago - 18 comments

#96 - Incorrect Saving Peft Models using HuggingFace Trainer

Issue - State: closed - Opened by agemagician over 1 year ago - 7 comments
Labels: solved

#94 - [Feature] Add support for LongT5 and LongFormer

Issue - State: closed - Opened by jordancole21 over 1 year ago - 2 comments
Labels: PRs welcome to address this

#91 - Add Late Prompt Tuning

Issue - State: closed - Opened by jackapbutler over 1 year ago - 2 comments
Labels: PRs welcome to address this

#88 - RuntimeError: expected scalar type Float but found Half

Issue - State: closed - Opened by ScottishFold007 over 1 year ago - 1 comment

#70 - [Feature] Add support for Donut (Multimodal Model)

Issue - State: closed - Opened by WaterKnight1998 over 1 year ago - 1 comment
Labels: PRs welcome to address this

#42 - Add support for T-Few

Issue - State: closed - Opened by lewtun over 1 year ago - 4 comments
Labels: PRs welcome to address this

#40 - Better warning message or error when target module names are not available

Issue - State: closed - Opened by sayakpaul over 1 year ago - 7 comments

#40 - Better warning message or error when target module names are not available

Issue - State: closed - Opened by sayakpaul over 1 year ago - 6 comments

GitHub / huggingface/peft issues and pull requests