import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
import os

def load_model_and_tokenizer(model_path):
    # First, try loading from the directory
    try:
        print(f"Attempting to load model from directory: {model_path}")
        model = AutoModelForCausalLM.from_pretrained(model_path)
    except Exception as e:
        print(f"Failed to load from directory. Error: {e}")
        # If that fails, try loading the specific .safetensors file
        safetensors_path = os.path.join(model_path, "model.safetensors")
        if os.path.exists(safetensors_path):
            print(f"Attempting to load model from file: {safetensors_path}")
            model = AutoModelForCausalLM.from_pretrained(safetensors_path)
        else:
            raise ValueError(f"Could not find model at {model_path} or {safetensors_path}")
    
    # Load the tokenizer from the original DistilGPT2 model
    tokenizer = AutoTokenizer.from_pretrained("distilgpt2")
    
    return model, tokenizer

def generate_text(model, tokenizer, prompt, max_length=125, num_return_sequences=1):
    input_ids = tokenizer.encode(prompt, return_tensors='pt')
    
    # Generate text
    output = model.generate(
        input_ids,
        max_length=max_length,
        num_return_sequences=num_return_sequences,
        no_repeat_ngram_size=6,
        top_k=25,
        top_p=0.99,
        temperature=0.34
    )
    
    return [tokenizer.decode(seq, skip_special_tokens=True) for seq in output]

def main():
    model_path = r"literalpathtothefoldernamed\checkpoint-4000" #change this to where you have the folder on your computer
    
    print(f"Attempting to load model...")
    model, tokenizer = load_model_and_tokenizer(model_path)
    
    print("Model loaded successfully. Enter prompts to generate text. Type 'quit' to exit.")
    
    while True:
        prompt = input("Enter a prompt: ")
        if prompt.lower() == 'quit':
            break
        
        generated_texts = generate_text(model, tokenizer, prompt)
        
        print("\nGenerated Text:")
        for i, text in enumerate(generated_texts, 1):
            print(f"{i}. {text}\n")

if __name__ == "__main__":
    main()