JohnsonXi

ValueError: invalid literal for int() with base 10: 'attention'

1

`if args.checkpoint: checkpoint = torch.load(args.checkpoint, map_location='cpu') state_dict = checkpoint['model'] # reshape positional embedding to accomodate for image resolution change pos_embed_reshaped = interpolate_pos_embed(state_dict['visual_encoder.pos_embed'],model.visual_encoder) state_dict['visual_encoder.pos_embed'] = pos_embed_reshaped if not args.evaluate: if config['distill']:...

windows系统推理生成的图片保存路径下有中文 ，cv.imwrite无法保存图片

2

cv2.imwrite(os.path.join(save_dir, 'single', str(num) + '.png'), img) save_dir路径中有中文 ，保存不了图片；但无报错或提示； for img in outputs: # count the number of images in the folder num = len(os.listdir(os.path.join(save_dir, 'single'))) cv2.imwrite(os.path.join(save_dir, 'single', str(num) + '.png'),...

Can not load in ComfyUI and --network_args "split_qkv=True" does not work

2

lora trained by sd-scipts can not be loaded in ComfyUI loadload node, so i try to "split_qkv" use order --network_args "split_qkv=True"  but seems to does not work.  Log...