加载预训练模型

桂萍

于 2024-08-29 22:21:31 发布

阅读量171

点赞数 1

分类专栏： 24年实验笔记文章标签：深度学习 python 神经网络

本文链接：https://blog.csdn.net/tfxzgp/article/details/141650506

版权

24年实验笔记专栏收录该内容

24 篇文章 0 订阅

订阅专栏

报错

/home/cszx/miniconda3/envs/zgp_3dads/lib/python3.7/site-packages/timm/models/_factory.py:121: UserWarning: Mapping deprecated model name vit_base_patch8_224_dino to current vit_base_patch8_224.dino.
  **kwargs,

Running on class bagel

Traceback (most recent call last):
  File "main.py", line 66, in <module>
    run_3d_ads()
  File "main.py", line 23, in run_3d_ads
    patchcore = PatchCore()
  File "/home/cszx/c1/zgp/3D-ADS-main/patchcore_runner.py", line 17, in __init__
    "RGB iNet": RGBInetFeatures(),
  File "/home/cszx/c1/zgp/3D-ADS-main/feature_extractors/features.py", line 21, in __init__
    self.deep_feature_extractor = Model(device=self.device)
  File "/home/cszx/c1/zgp/3D-ADS-main/feature_extractors/features.py", line 178, in __init__
    **kwargs)
  File "/home/cszx/miniconda3/envs/zgp_3dads/lib/python3.7/site-packages/timm/models/_factory.py", line 121, in create_model
    **kwargs,
  File "/home/cszx/miniconda3/envs/zgp_3dads/lib/python3.7/site-packages/timm/models/_registry.py", line 145, in _fn
    return current_fn(pretrained=pretrained, pretrained_cfg=pretrained_cfg or current_tag, **kwargs)
  File "/home/cszx/miniconda3/envs/zgp_3dads/lib/python3.7/site-packages/timm/models/vision_transformer.py", line 1883, in vit_base_patch8_224
    model = _create_vision_transformer('vit_base_patch8_224', pretrained=pretrained, **dict(model_args, **kwargs))
  File "/home/cszx/miniconda3/envs/zgp_3dads/lib/python3.7/site-packages/timm/models/vision_transformer.py", line 1750, in _create_vision_transformer
    raise RuntimeError('features_only not implemented for Vision Transformer models.')
RuntimeError: features_only not implemented for Vision Transformer models.

名字改回wide_resnet50_2就可以了

下载模型并加载

.pth格式

              self.backbone = timm.create_model(model_name=backbone_name, pretrained=True,
                                              pretrained_cfg_overlay=dict(
                                                  file='/home/cszx/c1/zgp/3D-ADS-main/checkpoint/dino_vitbase8_pretrain.pth'),
                                              **kwargs)

.safetensors格式

  self.backbone = timm.create_model(model_name=backbone_name, pretrained=True,
                                          pretrained_cfg_overlay=dict(
                                              file='/home/cszx/c1/zgp/3D-ADS-main/checkpoint/model.safetensors'),
                                           **kwargs)