import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM

# Load the model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b")
model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b")

def generate(prompt):
    inputs = tokenizer(prompt, return_tensors="pt")
    outputs = model.generate(inputs['input_ids'], max_length=100000)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response

# Create a Gradio Interface
gr.Interface(fn=generate, inputs="text", outputs="text").launch()