app.py

import gradio as gr
import os
import time
from huggingface_hub import InferenceClient
from huggingface_hub import hf_hub_download
import chatglm_cpp

pipeline = None

def load(repo_id, filename):
    global pipeline
    local_dir = f"./Models/{repo_id}"
    hf_hub_download(repo_id=repo_id, filename=filename, local_dir=local_dir)
    model = os.path.join(local_dir, filename)
    max_length = 8192
    pipeline = chatglm_cpp.Pipeline(model, max_length=max_length)
    return f"Model {filename} from {repo_id} loaded successfully."
load("None1145/ChatGLM3-6B-Theresa-GGML", "ChatGLM3-6B-Theresa-GGML-Q4_0.bin")

messages = []

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    global messages

    if pipeline is None:
        yield "Error: No model loaded. Please load a model first."
        return

    response = "..."
    for _ in range(0, 3):
        yield response
        time.sleep(1)
        response += " ..."
    
    generation_kwargs = dict(
        max_length=8192,
        max_context_length=max_tokens,
        do_sample=temperature > 0,
        top_k=0,
        top_p=top_p,
        temperature=temperature,
        repetition_penalty=1.0,
        stream=True,
    )

    if messages == []:
        messages = [chatglm_cpp.ChatMessage(role="system", content=system_message)]

    messages.append(chatglm_cpp.ChatMessage(role="user", content=message))

    response = ""
    for chunk in pipeline.chat(messages, **generation_kwargs):
        response += chunk.content
        yield response
        
    messages.append(chatglm_cpp.ChatMessage(role="assistant", content=response))

with gr.Blocks() as chat:
    with gr.Row():
        repo_id_input = gr.Textbox(label="Repo ID", value="None1145/ChatGLM3-6B-Theresa-GGML")
        filename_input = gr.Textbox(label="Filename", value="ChatGLM3-6B-Theresa-GGML-Q4_0.bin")
        load_button = gr.Button("Load Model")
        load_status = gr.Textbox(label="Load Status", interactive=False)
    load_button.click(load, inputs=[repo_id_input, filename_input], outputs=load_status)

    chat_interface = gr.ChatInterface(
        respond,
        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=0.7, step=0.1, label="Temperature"),
            gr.Slider(
                minimum=0.1,
                maximum=1.0,
                value=0.95,
                step=0.05,
                label="Top-p (nucleus sampling)",
            ),
        ],
    )

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