app.py

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
import streamlit as st

# Load the GPT-2 large model and tokenizer
model_name = "gpt2-large"
tokenizer = AutoTokenizer.from_pretrained(model_name)
# Add padding token to the tokenizer
tokenizer.pad_token = tokenizer.eos_token  # Set padding token to EOS token

model = AutoModelForCausalLM.from_pretrained(model_name)

# Function to generate a blog post based on a topic title
def generate_blog(topic_title, max_length=200):
    # Step 1: Encode the input
    inputs = tokenizer.encode_plus(topic_title, return_tensors='pt', padding=True)
    input_ids = inputs['input_ids']
    attention_mask = inputs['attention_mask']
    
    # Step 2: Generate model output
    output_ids = model.generate(input_ids, attention_mask=attention_mask, max_length=max_length, num_return_sequences=1, pad_token_id=tokenizer.eos_token_id)
    
    # Step 3: Decode the output
    blog_post = tokenizer.decode(output_ids[0], skip_special_tokens=True)
    
    return blog_post

# Example usage
#topic_title = input("Enter a topic title for the blog post: ")
#blog_post = generate_blog(topic_title)
#print("\nGenerated Blog Post:\n")
#print(blog_post)

title = st.text_area('Enter title')
if title:
    out = generate_blog(title)
    st.json(out)