app.py

import spaces
import gradio as gr
import torch


# Use a pipeline as a high-level helper
from transformers import pipeline




@spaces.GPU(duration=120)
def llama3_1_8B(question):
    
    print("RUNNING PIPE")
    pipe = pipeline("text-generation", model="NousResearch/Hermes-3-Llama-3.1-8B", max_new_tokens=200, device=0)
    print("PIPE DONE") 
    messages = [
    {"role": "user", "content": question},
    ]
    if torch.cuda.is_available():
        num_devices = torch.cuda.device_count()
        print(f"Number of CUDA devices: {num_devices}")
        
        for i in range(num_devices):
            print(f"Device {i}: {torch.cuda.get_device_name(i)}")
                    
    else:
        print("CUDA is not available.")

    print("GATHERING RESPONSES")
    responses = pipe(messages)

    return str(responses)

def greet(name):
    return "Hello " + name + "!!???"

demo = gr.Interface(fn=llama3_1_8B, inputs="text", outputs="text")
demo.launch()