app.py

#! /bin/bash
# -*- coding: utf-8 -*-
"""Gradio.ipynb

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1NhPAjcqhdmjOyMrg7j8IHqzGlJARGIjs
"""
import gradio as gr
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, StoppingCriteria, StoppingCriteriaList, TextIteratorStreamer
from threading import Thread

new_model="Mervyn999/mistral-7b-distilabel-mini-DPO"
model = AutoModelForCausalLM.from_pretrained(new_model,
                                            low_cpu_mem_usage=True,
                                            torch_dtype=torch.float16,
                                            load_in_4bit=True,
                                            device_map="cuda")
tokenizer = AutoTokenizer.from_pretrained(new_model)
# model = model.to('cuda:0')

class StopOnTokens(StoppingCriteria):
    def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:
        stop_ids = [29, 0]
        for stop_id in stop_ids:
            if input_ids[0][-1] == stop_id:
                return True
        return False

def predict(message, history):

    history_transformer_format = history + [[message, ""]]
    stop = StopOnTokens()

    #Wrap the prompt using the right chat template
    messages = "".join([f"### Instruction:\n{item[0]}\n\n### Response:\n{item[1]}" #curr_system_message +
                for item in history_transformer_format])

    model_inputs = tokenizer([messages], return_tensors="pt").to("cuda")
    streamer = TextIteratorStreamer(tokenizer, timeout=10., skip_prompt=True, skip_special_tokens=True)
    generate_kwargs = dict(
        model_inputs,
        streamer=streamer,
        max_new_tokens=1024,
        do_sample=True,
        top_p=0.95,
        top_k=1000,
        temperature=1.0,
        num_beams=1,
        stopping_criteria=StoppingCriteriaList([stop])
        )
    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()

    partial_message  = ""
    for new_token in streamer:
        if new_token != '<':
            partial_message += new_token
            yield partial_message



title = "Jaewon's finetuned LLM"

gr.close_all()
# gr.ChatInterface(predict).queue().launch(debug=True)
gr.ChatInterface(
    predict,
    chatbot=gr.Chatbot(height=300),
    textbox=gr.Textbox(placeholder="Send a message", container=False, scale=7),
    title="Chat with Mistral-7B DPO tuned",
    theme="soft",
    retry_btn=None,
    undo_btn="Delete Previous",
    clear_btn="Clear",
).queue().launch()