App.py

import gradio as gr
import tensorflow as tf
from transformers import BertTokenizer, TFBertModel
import numpy as np

# Load your model
model = tf.keras.models.load_model('models/model_files')

# Load tokenizer
tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')

def preprocess_text(text):
    inputs = tokenizer(text, return_tensors='tf', padding=True, truncation=True, max_length=512)
    return inputs

def predict(text, image, structured):
    text_inputs = preprocess_text(text)
    image = tf.image.resize(image, (224, 224))
    image = tf.keras.applications.resnet50.preprocess_input(image)
    structured = (structured - structured.mean()) / structured.std()
    prediction = model.predict([text_inputs['input_ids'], text_inputs['attention_mask'], image, structured])
    return prediction[0][0]

# Define the chat function
def chat_response(user_input):
    return f"Model response to: {user_input}"

# Define the code execution function
def execute_code(code):
    exec_globals = {}
    exec(code, exec_globals)
    return exec_globals.get("output", "No output")

with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            chat_input = gr.Textbox(lines=2, placeholder="Enter your message here...")
            chat_output = gr.Textbox(lines=5, placeholder="Model response will appear here...")
            chat_button = gr.Button("Send")
        with gr.Column():
            code_input = gr.Textbox(lines=10, placeholder="Enter your code here...")
            code_output = gr.Textbox(lines=5, placeholder="Code output will appear here...")
            code_button = gr.Button("Run Code")
    
    chat_button.click(chat_response, inputs=chat_input, outputs=chat_output)
    code_button.click(execute_code, inputs=code_input, outputs=code_output)

demo.launch()