| from transformers import AutoTokenizer, AutoModelForCausalLM |
| from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training |
| import torch |
| |
|
|
| |
| MODEL = "Viet-Mistral/Vistral-7B-Chat" |
| device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') |
| print('device =', device) |
|
|
| |
| model = AutoModelForCausalLM.from_pretrained( |
| 'Viet-Mistral/Vistral-7B-Chat', |
| torch_dtype=torch.bfloat16, |
| device_map="auto", |
| use_cache=True, |
| cache_dir='./hf_cache' |
| ) |
|
|
| tokenizer = AutoTokenizer.from_pretrained(MODEL, cache_dir='./hf_cache') |
|
|
| lora_config = LoraConfig.from_pretrained( |
| "thviet79/model-QA-medical", |
| cache_dir='/workspace/thviet/hf_cache' |
| ) |
|
|
| |
| model = get_peft_model(model, lora_config) |
|
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
|
|
| def respond( |
| message, |
| history: list[tuple[str, str]], |
| system_message: str, |
| max_tokens, |
| temperature, |
| top_p, |
| ): |
| sys_prompt = "Bạn là một trợ lí ảo Tiếng Việt về lĩnh vực y tế." |
| conversation = [{"role": "system", "content": sys_prompt}] |
| |
|
|
| for val in history: |
| if val[0]: |
| conversation.append({"role": "user", "content": val[0]}) |
| if val[1]: |
| messages.append({"role": "assistant", "content": val[1]}) |
|
|
| conversation.append({"role": "user", "content": message}) |
|
|
| input_ids_list = tokenizer.apply_chat_template(conversation, return_tensors="pt").to(device) |
| response = "" |
|
|
| for message in tokenizer.batch_decode(model.generate( |
| input_ids=input_ids, |
| max_new_tokens=max_tokens, |
| do_sample=True, |
| top_p=top_p, |
| temperature=temperature, |
| )[:, input_ids_list.size(1):], skip_special_tokens=True): |
| token = message.strip() |
| response += token |
| yield response |
|
|
| demo = 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__": |
| demo.launch() |
|
|