import gradio as gr
from transformers import AutoModelForSequenceClassification, AutoTokenizer

# Load the model and tokenizer
model_name = "Priyanshuchaudhary2425/EmotiNet"
model = AutoModelForSequenceClassification.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

# Class list
class_list = ['sadness', 'joy', 'love', 'anger', 'fear', 'surprise']

# Define the function to make predictions with your model
def predict_emotion(text):
    inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
    outputs = model(**inputs)
    probabilities = outputs.logits.softmax(dim=1).tolist()[0]
    return {class_list[label]: probability for label, probability in enumerate(probabilities)}

# Create a Gradio interface for your model
output_probabilities = gr.Label(num_top_classes=6)

interface = gr.Interface(
    fn=predict_emotion,
    inputs=gr.Textbox(lines=5, label="Enter your text here"),
    outputs=output_probabilities,
    title="Emotion Detection",
    description="This model predicts the probabilities of different emotions (sadness, joy, love, anger, fear, surprise) based on the input text.",
    examples=[
        ["In her warm embrace, I found solace, a refuge from the chaos of the world. Every beat of her heart echoed the melody of love, drawing me closer with each tender touch."],
        ["Fury surged through my veins, a tempest of resentment and indignation, fueled by the betrayal of trust. In that moment, every word spoken was a dagger, piercing through the facade of civility."],
        ["Tears silently traced their path down my cheeks, carrying the weight of unspoken sorrows, each drop a testament to the pain within. In the quiet of the night, I grappled with the emptiness that engulfed my soul, longing for the light of hope to pierce through the darkness."]
    ]
)

# Launch the Gradio interface with sharing enabled
interface.launch(share=True)