init

VitTrainer.py

@@ -7,9 +7,7 @@ import numpy as np
 import torch
 from datasets import Dataset, load_metric, ClassLabel
 from datasets import load_metric
-from transformers import AutoConfig
-from transformers import TrainingArguments, Trainer
-from transformers import ViTForImageClassification
+from transformers import ViTConfig, TrainingArguments, Trainer, ViTForImageClassification
 from torch.utils.data import DataLoader
 import torchvision.transforms as transforms
 from transformers import ViTFeatureExtractor
@@ -126,11 +124,16 @@ class VitTrainer:
           if isinstance(v, torch.Tensor):
             print(k, v.shape)
 
+        """
         model = ViTForImageClassification.from_pretrained('google/vit-base-patch16-224-in21k',
-                                                          num_labels=self.num_labels,
-                                                          id2label=self.id2label,
-                                                          label2id=self.label2id)
-        model = ViTForImageClassification(model.config)
+                                                  num_labels=self.num_labels,
+                                                  id2label=self.id2label,
+                                                  label2id=self.label2id)
+        """
+        configuration = ViTConfig(num_labels=self.num_labels,
+                                                  id2label=self.id2label,
+                                                  label2id=self.label2id)
+        model = ViTForImageClassification(configuration)
 
         trainer = Trainer(
             model,