Unable to load fine tuned model from local directory

Hello, I'm trying to finetune the florence-2-base-ft model from huggingface, for that, I use the following train function and save my finetuned model with the .save_pretrained() method. Then, I would like to load this finetuned model with .from_pretrained() method. However, I got an error message.

Steps to reproduce

Here is the train_model() function that I use for finetuning the florence-2-base-ft model.

    def train_model(train_loader, val_loader, model, processor, device, epochs=10, lr=1e-6):
        
        optimizer = torch.optim.AdamW(model.parameters(), lr=lr)
        num_training_steps = epochs * len(train_loader)
        lr_scheduler = get_scheduler(
            name="linear",
            optimizer=optimizer,
            num_warmup_steps=0,
            num_training_steps=num_training_steps,
        )

        for epoch in range(epochs):
            model.train()
            train_loss = 0
            i = -1
            for batch in tqdm(train_loader, desc=f"Training Epoch {epoch + 1}/{epochs}"):
                i += 1
                inputs, answers = batch

                input_ids = inputs["input_ids"]
                pixel_values = inputs["pixel_values"]
                labels = processor.tokenizer(text=answers, return_tensors="pt", padding=True, return_token_type_ids=False).input_ids.to(device)

                outputs = model(input_ids=input_ids, pixel_values=pixel_values, labels=labels)
                loss = outputs.loss

                loss.backward()
                optimizer.step()
                lr_scheduler.step()
                optimizer.zero_grad()

                train_loss += loss.item()

            avg_train_loss = train_loss / len(train_loader)
            print(f"Average Training Loss: {avg_train_loss}")

            # Validation phase
            model.eval()
            val_loss = 0
            with torch.no_grad():
                for batch in tqdm(val_loader, desc=f"Validation Epoch {epoch + 1}/{epochs}"):
                    inputs, answers = batch

                    input_ids = inputs["input_ids"]
                    pixel_values = inputs["pixel_values"]
                    labels = processor.tokenizer(text=answers, return_tensors="pt", padding=True, return_token_type_ids=False).input_ids.to(device)

                    outputs = model(input_ids=input_ids, pixel_values=pixel_values, labels=labels)
                    loss = outputs.loss

                    val_loss += loss.item()

            avg_val_loss = val_loss / len(val_loader)
            print(f"Average Validation Loss: {avg_val_loss}")

            # Save model checkpoint
            output_dir = f"./model_checkpoints/epoch_{epoch+1}"
            os.makedirs(output_dir, exist_ok=True)
            model.save_pretrained(output_dir)
            processor.save_pretrained(output_dir)

        return model, processor

Then, to load my finetuned model, I do it like that :

    model_finetuned = AutoModelForCausalLM.from_pretrained("./model_checkpoints/epoch_2",  trust_remote_code=True)

Stacktrace

How can I solve this?
Did I do something wrong to save/load my model, or is it due to the fact that florence-2 was released recently?

Thanks in advance!