import os
import gradio as gr
import matplotlib.pyplot as plt
from transformers import ViTForImageClassification, TrainingArguments, Trainer
from datasets import load_dataset

def finetune_model(epochs, save_at_num_epoch, learning_rate):
    # Load the dataset
    dataset = load_dataset("imagenet")

    # Initialize the model
    model = ViTForImageClassification.from_pretrained("google/vit-base-patch16-224")

    # Define the training arguments
    training_args = TrainingArguments(
        output_dir="vit-finetuned",
        num_train_epochs=epochs,
        save_strategy="steps",
        save_steps=save_at_num_epoch,
        learning_rate=learning_rate,
    )

    # Create the trainer and fine-tune the model
    trainer = Trainer(model=model, args=training_args, train_dataset=dataset["train"])
    train_metrics = trainer.train()

    # Save the fine-tuned model
    model.save_pretrained("vit-finetuned")

    # Plot the loss graph
    plt.figure(figsize=(8, 6))
    plt.plot(train_metrics.history["loss"])
    plt.title("Model Loss")
    plt.xlabel("Epoch")
    plt.ylabel("Loss")
    plt.savefig("loss_graph.png")

    return "Fine-tuning complete!"

# Create the Gradio interface
with gr.Blocks() as demo:
    gr.Markdown("# Fine-Tune a Model")

    with gr.Column():
        epochs = gr.Slider(label="Epochs", minimum=1, maximum=10, value=3)
        save_at_num_epoch = gr.Slider(label="Save Model Every X Epochs", minimum=1, maximum=epochs, value=1)
        learning_rate = gr.Slider(label="Learning Rate", minimum=1e-5, maximum=1e-3, value=2e-5)
        run_button = gr.Button("Fine-Tune Model")

    status = gr.Textbox(label="Fine-Tuning Status")
    loss_graph = gr.Image(label="Loss Graph")

    run_button.click(finetune_model, inputs=[epochs, save_at_num_epoch, learning_rate], outputs=[status, loss_graph])

if __name__ == "__main__":
    demo.launch()