I ran the pretraining script

CUDA_VISIBLE_DEVICES=1 accelerate launch pretrain_e4t.py --pretrained_model_name_or_path="CompVis/stable-diffusion-v1-4" --clip_model_name_or_path="ViT-H-14::laion2b_s32b_b79k" --domain_class_token="cat" --placeholder_token="*s" --prompt_template=normal --save_sample_prompt="a photo of the *s, a photo of the *s in monet style" --reg_lambda=0.01 --domain_embed_scale=0.1 --output_dir="pretrained-cat" --train_image_dataset="/dataset/e4t/dataset/cat/" --iterable_dataset --resolution=512 --train_batch_size=16 --learning_rate=1e-6 --scale_lr --checkpointing_steps=10000 --log_steps=1000 --max_train_steps=100000 --unfreeze_clip_vision --mixed_precision="fp16" --enable_xformers_memory_efficient_attention

and encountered the error
TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType

This bug is from the line 99 of "e4t-diffusion/e4t/utils.py", where ckpt is None for os.path.exists(ckpt_path).
When checking the related code, I notice that in line 238 and 249 of pretrain_u4t.py, the ckpt_path is specially set to a weight_offsets.pt and encoder.pt. However, there are no such files before pretraining in a base model like "CompVis/stable-diffusion-v1-4", so it seems rather weird. Please help me fix this issue, thank you!