app.py

import spaces
import os

from huggingface_hub import Repository
from huggingface_hub import login

login(token = os.environ['HUB_TOKEN'])

repo = Repository(
    local_dir="backend_fn",
    repo_type="dataset",
    clone_from=os.environ['DATASET'],
    token=True,
    git_email='[email protected]'
)
repo.git_pull()

import json
import uuid
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
from threading import Thread
from backend_fn.feedback import feedback
from gradio_modal import Modal

"""
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
"""
model_name = "Merdeka-LLM/merdeka-llm-3.2b-128k-instruct"

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_name)

streamer = TextIteratorStreamer(tokenizer, timeout=300, skip_prompt=True, skip_special_tokens=True)

histories = []
action = None

session_id = uuid.uuid1().__str__()

@spaces.GPU
def respond(
    message,
    history: list[tuple[str, str]],
    # system_message,
    max_tokens = 4096,
    temperature = 0.01,
    top_p = 0.95,
):
    messages = [
        {"role": "system", "content": "You are a professional lawyer who is familiar with Malaysia Law."}
    ]

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message})

    response = ""

    text = tokenizer.apply_chat_template(
        messages,
        tokenize=False,
        add_generation_prompt=True,
    )
    model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

    generate_kwargs = dict(
        model_inputs,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        streamer=streamer
    )
    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()
    for new_token in streamer:
      if new_token != '<':
          response += new_token
          yield response

"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""

def submit_feedback(value):
    feedback(session_id, json.dumps(histories), value, action)


with gr.Blocks() as demo:
    def vote(history,data: gr.LikeData):
        global histories
        global action
        histories = history
        action = data.liked

    with Modal(visible=False) as modal:
        textb = gr.Textbox(
            label='Actual response',
            info='Leave blank if the answer is good enough'
        )

        submit_btn = gr.Button(
            'Submit'
        )

        submit_btn.click(submit_feedback,textb)
        submit_btn.click(lambda: Modal(visible=False), None, modal)
        submit_btn.click(lambda x: gr.update(value=''), [],[textb])


    ci = gr.ChatInterface(
        respond,
        # fill_height=True
        # additional_inputs=[
        #     # gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
        #     gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        #     gr.Slider(minimum=0.1, maximum=4.0, value=0.1, step=0.1, label="Temperature"),
        #     gr.Slider(
        #         minimum=0.1,
        #         maximum=1.0,
        #         value=0.95,
        #         step=0.05,
        #         label="Top-p (nucleus sampling)",
        #     ),
        # ],
    )


    ci.chatbot.show_copy_button=True
    # ci.chatbot.value=[(None,"Hello! I'm here to assist you with understanding the laws and acts of Malaysia.")]
    # ci.chatbot.height=500

    ci.chatbot.like(vote, ci.chatbot, None).then(
        lambda: Modal(visible=True), None, modal
    )

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