search

large-ocr-model / large-ocr-model.github.io

134 stars 5 forks source link

No configuration found for 'vl4str' #5

Closed lyb18758 closed 3 months ago

lyb18758 commented 4 months ago

下载并且载入权重了但是报了这个错误，请问是什么原因呢 Traceback (most recent call last): File "test.py", line 174, in main() File "D:\Anaconda\envs\lm\lib\site-packages\torch\autograd\grad_mode.py", line 28, in decorate_context return func(*args, kwargs) File "test.py", line 102, in main model = load_from_checkpoint(args.checkpoint, kwargs).eval().to(args.device) File "D:\pythonProject\lm\strhub\models\utils.py", line 112, in load_from_checkpoint raise InvalidModelError("No configuration found for '{}'".format(experiment)) from None strhub.models.utils.InvalidModelError: No configuration found for 'vl4str'

large-ocr-model commented 4 months ago

What is the naming of the weights you are loading? Please ensure that the loaded weight names contain the keyword 'clip4str'.

lyb18758 commented 4 months ago

What is the naming of the weights you are loading? Please ensure that the loaded weight names contain the keyword 'clip4str'.

载入的是clip4str_b_plus.ckpt，从inference中提供的百度网盘链接下载来的 全部报错信息如下 python test.py --checkpoint clip4str_b_plus.ckpt --data_root test --clip_model_path ViT-B-16.pt
Namespace(alpha=0.1, batch_size=128, cased=False, checkpoint='clip4str_b_plus.ckpt', clip_model_path='ViT-B-16.pt', clip_refine=False, data_root='test', device='cuda', new=False, num_workers=4, punctuation=False, rotation=0, sample_K=5, sample_K2=5, sample_prompt=None, sample_total=50) Additional keyword arguments: {'charset_test': '0123456789abcdefghijklmnopqrstuvwxyz'} load weights from checkpoint clip4str_b_plus.ckpt

config of VL4STR: image_freeze_nlayer: -1, text_freeze_nlayer: 6, freeze_language_backbone: False, freeze_image_backbone: False use_language_model: True, context_length: 16, cross_token_embeding: False, cross_loss_weight: 1.0 use_share_dim: True, image_detach: True cross_gt_context: True, cross_cloze_mask: False, cross_fast_decode: False

Traceback (most recent call last): File "test.py", line 174, in main() File "D:\Anaconda\envs\lm\lib\site-packages\torch\autograd\grad_mode.py", line 28, in decorate_context return func(*args, kwargs) File "test.py", line 102, in main model = load_from_checkpoint(args.checkpoint, kwargs).eval().to(args.device) File "D:\pythonProject\lm\strhub\models\utils.py", line 115, in load_from_checkpoint raise InvalidModelError("No configuration found for '{}'".format(experiment)) from None strhub.models.utils.InvalidModelError: No configuration found for 'vl4str'

large-ocr-model commented 4 months ago

"Please create a new folder named clip in the working directory and place the ViT-B-16.pt file inside the clip folder. Then execute the following command again: 

python test.py --checkpoint clip4str_b_plus.ckpt --data_root test --clip_model_path clip/ViT-B-16.pt
lyb18758 commented 4 months ago

"Please create a new folder named clip in the working directory and place the ViT-B-16.pt file inside the clip folder. Then execute the following command again:

python test.py --checkpoint clip4str_b_plus.ckpt --data_root test --clip_model_path clip/ViT-B-16.pt

ok 成功了 谢谢