Spaces:
Sleeping
Sleeping
| from .BaseLLM import BaseLLM | |
| from peft import PeftModel | |
| import os | |
| from transformers import AutoModelForCausalLM, AutoTokenizer | |
| class LocalModel(BaseLLM): | |
| def __init__(self, model, adapter_path = None): | |
| super(LocalModel, self).__init__() | |
| model_name = model | |
| self.model = AutoModelForCausalLM.from_pretrained( | |
| model_name, | |
| torch_dtype="auto", | |
| device_map="auto", | |
| ) | |
| if isinstance(adapter_path,str): | |
| self.model = PeftModel.from_pretrained(self.model, adapter_path) | |
| elif isinstance(adapter_path,list): | |
| for path in adapter_path: | |
| self.model = PeftModel.from_pretrained(self.model, path) | |
| self.tokenizer = AutoTokenizer.from_pretrained(model_name) | |
| self.model_name = model | |
| self.messages = [] | |
| def initialize_message(self): | |
| self.messages = [] | |
| def ai_message(self, payload): | |
| self.messages.append({"role": "ai", "content": payload}) | |
| def system_message(self, payload): | |
| self.messages.append({"role": "system", "content": payload}) | |
| def user_message(self, payload): | |
| self.messages.append({"role": "user", "content": payload}) | |
| def get_response(self,temperature = 0.8): | |
| text = self.tokenizer.apply_chat_template( | |
| self.messages, | |
| tokenize=False, | |
| add_generation_prompt=True | |
| ) | |
| model_inputs = self.tokenizer([text], return_tensors="pt").to(self.model.device) | |
| generated_ids = self.model.generate( | |
| **model_inputs, | |
| max_new_tokens=512 | |
| ) | |
| generated_ids = [ | |
| output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids) | |
| ] | |
| response = self.tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0] | |
| return response | |
| def chat(self,text,temperature = 0.8): | |
| self.initialize_message() | |
| self.user_message(text) | |
| response = self.get_response(temperature = temperature) | |
| return response | |
| def print_prompt(self): | |
| for message in self.messages: | |
| print(message) |