app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
from datasets import load_dataset
import pandas as pd

# Load model and tokenizer
model_name = "google/flan-t5-small"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

# Load CSV data (replace with your dataset logic)
def load_data():
    dataset = load_dataset("LR36/BehavioralAnalysis", split="train")  # Replace with your dataset name
    return dataset["train"]

# Generate text (inference)
def generate_text(prompt):
    inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=512)
    outputs = model.generate(**inputs, max_new_tokens=100)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# Fine-tune button (simplified example)
def fine_tune():
    dataset = load_data()
    # Add your training logic here (see previous examples)
    return "Fine-tuning complete! (Note: Models reset when Space stops.)"

# Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("# FLAN-T5 Demo")
    with gr.Tab("Generate Text"):
        prompt = gr.Textbox(label="Input Prompt")
        generate_btn = gr.Button("Generate")
        output = gr.Textbox(label="Output")
        generate_btn.click(fn=generate_text, inputs=prompt, outputs=output)
    
    with gr.Tab("Fine-Tune"):
        train_btn = gr.Button("Train on CSV Data")
        train_output = gr.Textbox(label="Training Status")
        train_btn.click(fn=fine_tune, outputs=train_output)

demo.launch()