Update README.md

README.md

@@ -248,4 +248,78 @@ The following hyperparameters were used during training:
 - Transformers 4.35.2
 - Pytorch 2.1.0+cu118
 - Datasets 2.15.0
-- Tokenizers 0.15.0
+- Tokenizers 0.15.0
+
+### Example of usage
+
+```python
+from datasets import load_dataset
+from transformers import TrainingArguments
+from transformers import CLIPProcessor, AutoModelForImageClassification
+
+processor = CLIPProcessor.from_pretrained("Andron00e/CLIPForImageClassification-v1")
+model = AutoModelForImageClassification.from_pretrained("Andron00e/CLIPForImageClassification-v1")
+
+dataset = load_dataset("Andron00e/CIFAR100-custom")
+dataset = dataset["train"].train_test_split(test_size=0.2)
+from datasets import DatasetDict
+
+val_test = dataset["test"].train_test_split(test_size=0.5)
+dataset = DatasetDict({
+    "train": dataset["train"],
+    "validation": val_test["train"],
+    "test": val_test["test"],
+})
+
+def transform(example_batch):
+    inputs = processor(text=[classes[x] for x in example_batch['labels']], images=[x for x in example_batch['image']], padding=True, return_tensors='pt')
+    inputs['labels'] = example_batch['labels']
+    return inputs
+
+def collate_fn(batch):
+    return {
+        'input_ids': torch.stack([x['input_ids'] for x in batch]),
+        'attention_mask': torch.stack([x['attention_mask'] for x in batch]),
+        'pixel_values': torch.stack([x['pixel_values'] for x in batch]),
+        'labels': torch.tensor([x['labels'] for x in batch])
+    }
+
+training_args = TrainingArguments(
+  output_dir="./outputs",
+  per_device_train_batch_size=16,
+  evaluation_strategy="steps",
+  num_train_epochs=4,
+  fp16=False,
+  save_steps=100,
+  eval_steps=100,
+  logging_steps=10,
+  learning_rate=2e-4,
+  save_total_limit=2,
+  remove_unused_columns=False,
+  push_to_hub=False,
+  report_to='tensorboard',
+  load_best_model_at_end=True,
+)
+
+from transformers import Trainer
+
+trainer = Trainer(
+    model=model,
+    args=training_args,
+    data_collator=collate_fn,
+    compute_metrics=compute_metrics,
+    train_dataset=dataset.with_transform(transform)["train"],
+    eval_dataset=dataset.with_transform(transform)["validation"],
+    tokenizer=model.processor,
+)
+
+train_results = trainer.train()
+trainer.save_model()
+trainer.log_metrics("train", train_results.metrics)
+trainer.save_metrics("train", train_results.metrics)
+trainer.save_state()
+
+metrics = trainer.evaluate(processed_dataset['test'])
+trainer.log_metrics("eval", metrics)
+trainer.save_metrics("eval", metrics)
+```