import os import gradio as gr from transformers import AutoTokenizer from pymongo import MongoClient import openai DB_NAME = os.getenv("MONGO_DBNAME", "taiwan-llm") USER = os.getenv("MONGO_USER") PASSWORD = os.getenv("MONGO_PASSWORD") uri = f"mongodb+srv://{USER}:{PASSWORD}@{DB_NAME}.kvwjiok.mongodb.net/?retryWrites=true&w=majority" mongo_client = MongoClient(uri) db = mongo_client[DB_NAME] conversations_collection = db['conversations'] DESCRIPTION = """ # Language Models for Taiwanese Culture

✍️ Online Demo • 🤗 HF Repo • 🐦 Twitter • 📃 [Paper Coming Soon] • 👨️ Github Repo


# 🌟 Checkout New [Taiwan-LLM UI](http://www.twllm.com) 🌟 Taiwan-LLaMa is a fine-tuned model specifically designed for traditional mandarin applications. It is built upon the LLaMa 2 architecture and includes a pretraining phase with over 5 billion tokens and fine-tuning with over 490k multi-turn conversational data in Traditional Mandarin. ## Key Features 1. **Traditional Mandarin Support**: The model is fine-tuned to understand and generate text in Traditional Mandarin, making it suitable for Taiwanese culture and related applications. 2. **Instruction-Tuned**: Further fine-tuned on conversational data to offer context-aware and instruction-following responses. 3. **Performance on Vicuna Benchmark**: Taiwan-LLaMa's relative performance on Vicuna Benchmark is measured against models like GPT-4 and ChatGPT. It's particularly optimized for Taiwanese culture. 4. **Flexible Customization**: Advanced options for controlling the model's behavior like system prompt, temperature, top-p, and top-k are available in the demo. ## Model Versions Different versions of Taiwan-LLaMa are available: - **Taiwan-LLM v3.0 (This demo)** - **Taiwan-LLM v2.0** - **Taiwan-LLM v1.0** The models can be accessed from the provided links in the Hugging Face repository. Try out the demo to interact with Taiwan-LLaMa and experience its capabilities in handling Traditional Mandarin! """ LICENSE = """ ## Licenses - Code is licensed under Apache 2.0 License. - Models are licensed under the LLAMA Community License. - By using this model, you agree to the terms and conditions specified in the license. - By using this demo, you agree to share your input utterances with us to improve the model. ## Acknowledgements Taiwan-LLaMa project acknowledges the efforts of the [Meta LLaMa team](https://github.com/facebookresearch/llama) and [Vicuna team](https://github.com/lm-sys/FastChat) in democratizing large language models. """ DEFAULT_SYSTEM_PROMPT = "你是人工智慧助理，以下是用戶和人工智能助理之間的對話。你要對用戶的問題提供有用、安全、詳細和禮貌的回答。 您是由國立臺灣大學的林彥廷博士生為研究目的而建造的。" endpoint_url = os.environ.get("ENDPOINT_URL", "http://127.0.0.1:8080") MAX_MAX_NEW_TOKENS = 4096 DEFAULT_MAX_NEW_TOKENS = 1536 max_prompt_length = 8192 - MAX_MAX_NEW_TOKENS - 10 model_name = "yentinglin/Llama-3-Taiwan-8B-Instruct" tokenizer = AutoTokenizer.from_pretrained(model_name) with gr.Blocks() as demo: gr.Markdown(DESCRIPTION) chatbot = gr.Chatbot() with gr.Row(): msg = gr.Textbox( container=False, show_label=False, placeholder='Type a message...', scale=10, ) submit_button = gr.Button('Submit', variant='primary', scale=1, min_width=0) with gr.Row(): retry_button = gr.Button('🔄 Retry', variant='secondary') undo_button = gr.Button('↩️ Undo', variant='secondary') clear = gr.Button('🗑️ Clear', variant='secondary') saved_input = gr.State() with gr.Accordion(label='Advanced options', open=False): system_prompt = gr.Textbox(label='System prompt', value=DEFAULT_SYSTEM_PROMPT, lines=6) max_new_tokens = gr.Slider( label='Max new tokens', minimum=1, maximum=MAX_MAX_NEW_TOKENS, step=1, value=DEFAULT_MAX_NEW_TOKENS, ) temperature = gr.Slider( label='Temperature', minimum=0.1, maximum=1.0, step=0.1, value=0.3, ) top_p = gr.Slider( label='Top-p (nucleus sampling)', minimum=0.05, maximum=1.0, step=0.05, value=0.95, ) def user(user_message, history): return "", history + [[user_message, None]] def bot(history, max_new_tokens, temperature, top_p, system_prompt): messages = [{"role": "system", "content": system_prompt}] for user, bot in history: messages.append({"role": "user", "content": user}) messages.append({"role": "assistant", "content": bot}) history[-1][1] = "" response = openai.ChatCompletion.create( model=model_name, messages=messages, max_tokens=max_new_tokens, temperature=temperature, top_p=top_p, n=1, stream=True, ) for chunk in response: if 'choices' in chunk: delta = chunk['choices'][0]['delta'] if 'content' in delta: history[-1][1] += delta['content'] yield history conversation_document = { "model_name": model_name, "history": history, "system_prompt": system_prompt, "max_new_tokens": max_new_tokens, "temperature": temperature, "top_p": top_p, } conversations_collection.insert_one(conversation_document) msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then( fn=bot, inputs=[ chatbot, max_new_tokens, temperature, top_p, system_prompt, ], outputs=chatbot ) submit_button.click( user, [msg, chatbot], [msg, chatbot], queue=False ).then( fn=bot, inputs=[ chatbot, max_new_tokens, temperature, top_p, system_prompt, ], outputs=chatbot ) def delete_prev_fn( history: list[tuple[str, str]]) -> tuple[list[tuple[str, str]], str]: try: message, _ = history.pop() except IndexError: message = '' return history, message or '' def display_input(message: str, history: list[tuple[str, str]]) -> list[tuple[str, str]]: history.append((message, '')) return history retry_button.click( fn=delete_prev_fn, inputs=chatbot, outputs=[chatbot, saved_input], api_name=False, queue=False, ).then( fn=display_input, inputs=[saved_input, chatbot], outputs=chatbot, api_name=False, queue=False, ).then( fn=bot, inputs=[ chatbot, max_new_tokens, temperature, top_p, system_prompt, ], outputs=chatbot, ) undo_button.click( fn=delete_prev_fn, inputs=chatbot, outputs=[chatbot, saved_input], api_name=False, queue=False, ).then( fn=lambda x: x, inputs=[saved_input], outputs=msg, api_name=False, queue=False, ) clear.click(lambda: None, None, chatbot, queue=False) gr.Markdown(LICENSE) demo.queue(max_size=128) demo.launch(max_threads=10)