app.py

import streamlit as st
from transformers import GPT2Tokenizer, GPT2LMHeadModel

# Initialize the tokenizer and model
model_name = 'gpt2-large'
tokenizer = GPT2Tokenizer.from_pretrained(model_name)
model = GPT2LMHeadModel.from_pretrained(model_name)

# Set the title for the Streamlit app
st.title("GPT-2 Blog Post Generator")

# Text input for the user
text = st.text_area("Enter your Topic: ")

def generate_text(text):
    try:
        # Encode input text
        encoded_input = tokenizer(text, return_tensors='pt')

        # Generate text
        output = model.generate(
            input_ids=encoded_input['input_ids'],
            max_length=200,  # Specify the max length for the generated text
            num_return_sequences=1,  # Number of sequences to generate
            no_repeat_ngram_size=2,  # Avoid repeating n-grams of length 2
            top_k=50,  # Limits the sampling pool to top_k tokens
            top_p=0.95,  # Cumulative probability threshold for nucleus sampling
            temperature=0.7,  # Controls the randomness of predictions
            attention_mask=encoded_input['attention_mask'],  # Correct attention mask
            pad_token_id=tokenizer.eos_token_id  # Use the end-of-sequence token as padding
        )

        # Decode generated text
        generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
        return generated_text

    except Exception as e:
        st.error(f"An error occurred: {e}")
        return None

if st.button("Generate"):
    generated_text = generate_text(text)
    if generated_text:
        # Display the generated text
        st.subheader("Generated Blog Post")
        st.write(generated_text)