Failed to reserver PEFT model "PeftModelForCausalLM.__init__() missing 1 required positional argument: 'peft_config'"

[shaowei-su]

System Info

optimum==1.14.1
peft==0.6.1
pytorch-lightning==2.1.0
pytorch-pretrained-bert==0.6.2
torch==2.0.0+cu118
torchaudio==2.0.0+cu118
torchmetrics==1.2.0
torchvision==0.15.1+cu118

Who can help?

No response

Information

• [ ] The official example scripts
• [X] My own modified scripts

Tasks

• [ ] An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• [X] My own task or dataset (give details below)

Reproduction (minimal, reproducible, runnable)

Model initialization

model = AutoModelForCausalLM.from_pretrained(...)  # original HF model
model = model.to_bettertransformer()  # convert to use optimum library
model = get_peft_model(model, lora_config)  # convert to use LoRA
...

Model save

unwrapped_model = BetterTransformer.reverse(unwrapped_model)

Failure message

2023-11-14, 14:48:00.033		
TypeError: PeftModelForCausalLM.__init__() missing 1 required positional argument: 'peft_config'
2023-11-14, 14:48:00.033		
    reversed_model = bt_model.__class__(config)
2023-11-14, 14:48:00.033		
  File "/home/default_user/.conda/envs/user/lib/python3.10/site-packages/optimum/bettertransformer/transformation.py", line 340, in reverse
2023-11-14, 14:48:00.033		
    unwrapped_model = BetterTransformer.reverse(unwrapped_model)

Expected behavior

BetterTransformer should be able to restore the original PEFT model.