app.py

import spaces
import os
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
from transformers import TextStreamer
import gradio as gr

text_generator = None
is_hugging_face = True
model_id = "google/gemma-2-9b-it"
huggingface_token = os.getenv("HUGGINGFACE_TOKEN")
device = "auto" # torch.device("cuda" if torch.cuda.is_available() else "cpu")
device = "cuda"
dtype = torch.bfloat16

if not huggingface_token:
    pass
    print("no HUGGINGFACE_TOKEN if you need set secret ")
    #raise ValueError("HUGGINGFACE_TOKEN environment variable is not set")
        

    
    
    



tokenizer = AutoTokenizer.from_pretrained(model_id, token=huggingface_token)

print(model_id,device,dtype)
histories = []
#model = None



if not is_hugging_face:
    model = AutoModelForCausalLM.from_pretrained(
        model_id, token=huggingface_token ,torch_dtype=dtype,device_map=device
    )
    text_generator = pipeline("text-generation", model=model, tokenizer=tokenizer,torch_dtype=dtype,device_map=device,stream=True ) #pipeline has not to(device)
    
    if next(model.parameters()).is_cuda:
        print("The model is on a GPU")
    else:
        print("The model is on a CPU")

    #print(f"text_generator.device='{text_generator.device}")
    if str(text_generator.device).strip() == 'cuda':
        print("The pipeline is using a GPU")
    else:
        print("The pipeline is using a CPU")

print("initialized")

@spaces.GPU(duration=30)
def generate_text(messages):
    if is_hugging_face:#need everytime initialize for ZeroGPU
        model = AutoModelForCausalLM.from_pretrained(
                model_id, token=huggingface_token ,torch_dtype=dtype,device_map=device
            )
    streamer = TextStreamer(tokenizer, skip_prompt=True)    
    text_generator = pipeline("text-generation", model=model, tokenizer=tokenizer,torch_dtype=dtype,device_map=device ,streamer=streamer) #pipeline has not to(device)
    result = text_generator(messages, max_new_tokens=256, do_sample=True, temperature=0.7)
    print(result)
    generated_output = ""
    for token in result:
        generated_output += token["generated_token"]
        yield generated_output


def call_generate_text(message, history):
   # history.append({"role": "user", "content": message})
    print(message)
    print(history)
   
    messages = history+[{"role":"user","content":message}]
    try:
        
        for text in generate_text(messages):
            yield text
    except RuntimeError  as e:
        print(f"An unexpected error occurred: {e}")
        yield ""

demo = gr.ChatInterface(call_generate_text,type="messages")

if __name__ == "__main__":
    demo.launch(share=True)