app.py

import gradio as gr
import tensorflow as tf
import numpy as np

# Define a function to preprocess the text input
def preprocess(text):
    tokenizer = tf.keras.preprocessing.text.Tokenizer()
    tokenizer.fit_on_texts([text])
    text = tokenizer.texts_to_sequences([text])
    text = tf.keras.preprocessing.sequence.pad_sequences(text, maxlen=500, padding='post', truncating='post')
    return text

# Load the pre-trained model
model = tf.keras.models.load_model('sentimentality.h5')

# Define a function to make a prediction on the input text
def predict_sentiment(text):
    # Preprocess the text
    text = preprocess(text)
    # Make a prediction using the loaded model
    proba = model.predict(text)[0]
    # Normalize the probabilities
    proba /= proba.sum()
    # Return the probability distribution
    return {'Positive': float(proba[0]), 'Negative': float(proba[1]), 'Neutral': float(proba[2])}

# Define the possible classes
classes = ['Positive', 'Negative', 'Neutral']

# Define the Gradio interface
iface = gr.Interface(
    fn=predict_sentiment,
    inputs=gr.inputs.Textbox(label='Enter text here'),
    outputs=gr.outputs.Bar(label='Sentiment Distribution', 
                           colors=['#2a9d8f', '#e76f51', '#264653'], 
                           value=[0.3, 0.3, 0.3], 
                           min=0, 
                           max=1),
    title='SENTIMENT ANALYSIS'
)

# Set the class labels for the output
iface.outputs[0].type = 'bar'
iface.outputs[0].label = 'Sentiment Distribution'
iface.outputs[0].determine_labels_from_data = False
iface.outputs[0].labels = classes

# Launch the interface
iface.launch()