Results
2
issues of
MK
find a bug when resume from checkpoint. In finetune.py, resume code as below: ```python if os.path.exists(checkpoint_name): print(f"Restarting from {checkpoint_name}") adapters_weights = torch.load(checkpoint_name) model = set_peft_model_state_dict(model, adapters_weights) else: print(f"Checkpoint {checkpoint_name} not...