app.py

import torch
from transformers import DistilBertTokenizer, DistilBertForSequenceClassification

def load_model(model_path, device):
    model = DistilBertForSequenceClassification.from_pretrained(model_path)
    model.to(device)
    model.eval()
    return model

def run_inference(model, tokenizer, label_decoder, device, user_input):
    model.eval()  # Set the model to evaluation mode

    # user_input = input("Enter a text for prediction: ")

    # Tokenize user input
    input_ids = tokenizer.encode(user_input, return_tensors="pt").to(device)

    with torch.no_grad():
        outputs = model(input_ids)
        predicted_label = torch.argmax(outputs.logits, dim=1).tolist()

        # Extracting the text and predicted outcome
        input_text = tokenizer.decode(input_ids[0], skip_special_tokens=True)
        predicted_outcome = label_decoder[predicted_label[0]]

        # Display the results
        print(f"Text: {input_text}")
        print(f"Predicted Outcome: {predicted_outcome}")
        print()
        return predicted_outcome  # Add a new line for better readability

# Example usage
model_path = "/home/lwasinam/AI_Projects/hate_speech_detection/model6"  # Replace with the actual path to your model
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
tokenizer = DistilBertTokenizer.from_pretrained("distilbert-base-uncased")  # Replace with your desired tokenizer

# Load model
model = load_model(model_path, device)


label_decoder = {0: "Not Hate", 1: "Hate",}


# Assuming you have label_decoder defined


import streamlit as st

st.title("Hate Speech Detection")

user_input = st.text_input("Enter your text:")
if user_input:
    result = run_inference(model, tokenizer, label_decoder, device, user_input)
    st.write("Inference Result:", result)