transformers_local.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
from threading import Thread
import spaces

tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.3")
model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-Instruct-v0.3", device_map="auto")
terminators = [
    tokenizer.eos_token_id,
    tokenizer.convert_tokens_to_ids("<|eot_id|>")
]


@spaces.GPU(duration=120)
def chat_mistral7b_v0dot3(message: str, 
              history: list, 
              temperature: float, 
              max_new_tokens: int
             ) -> str:

    conversation = []
    for user, assistant in history:
        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
    conversation.append({"role": "user", "content": message})

    input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt").to(model.device)
    
    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)

    generate_kwargs = dict(
        input_ids= input_ids,
        streamer=streamer,
        max_new_tokens=max_new_tokens,
        do_sample=True,
        temperature=temperature,
        eos_token_id=terminators,
    )
    # This will enforce greedy generation (do_sample=False) when the temperature is passed 0, avoiding the crash.             
    if temperature == 0:
        generate_kwargs['do_sample'] = False
        
    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()

    outputs = []
    for text in streamer:
        outputs.append(text)
        #print(outputs)
        yield "".join(outputs)
        

# Gradio block
with gr.Blocks() as demo:
    gr.Markdown("# ChatInterface with Mistral and Transformers 🤗")
    gr.ChatInterface(
        fn=chat_mistral7b_v0dot3,
        fill_height=True,
        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
        additional_inputs=[
            gr.Slider(minimum=0,
                      maximum=1, 
                      step=0.1,
                      value=0.95, 
                      label="Temperature", 
                      render=False),
            gr.Slider(minimum=128, 
                      maximum=4096,
                      step=1,
                      value=512, 
                      label="Max new tokens", 
                      render=False ),
            ],
    )
    
  
if __name__ == "__main__":
    demo.launch()