| from typing import Any, Dict, List | |
| import torch, re | |
| import transformers | |
| from transformers import AutoModelForCausalLM, AutoTokenizer | |
| dtype = torch.bfloat16 if torch.cuda.get_device_capability()[0] == 8 else torch.float16 | |
| class EndpointHandler: | |
| def __init__(self, path=""): | |
| tokenizer = AutoTokenizer.from_pretrained(path, trust_remote_code = True) | |
| model = AutoModelForCausalLM.from_pretrained( | |
| path, | |
| return_dict = True, | |
| device_map = "auto", | |
| load_in_8bit = True, | |
| torch_dtype = dtype, | |
| trust_remote_code = True, | |
| ) | |
| gen_config = model.generation_config | |
| gen_config.max_new_tokens = 100 | |
| gen_config.temperature = 0 | |
| gen_config.num_return_sequences = 1 | |
| gen_config.pad_token_id = tokenizer.eos_token_id | |
| gen_config.eos_token_id = tokenizer.eos_token_id | |
| self.generation_config = gen_config | |
| self.pipeline = transformers.pipeline( | |
| 'text-generation', model=model, tokenizer=tokenizer | |
| ) | |
| def __call__(self, data: Dict[dict, Any]) -> Dict[str, Any]: | |
| inputs = data.pop("inputs", data) | |
| question = data.pop("question", None) | |
| context = data.pop("context", None) | |
| temp = data.pop("temp", None) | |
| max_tokens = data.pop("max_tokens", None) | |
| bos_token = "<s>" | |
| original_system_message = "Below is an instruction that describes a task. Write a response that appropriately completes the request." | |
| system_message = "Use the provided context followed by a question to answer it." | |
| full_prompt = f"""<s>### Instruction: | |
| {system_message} | |
| ### Context: | |
| {context} | |
| ### Question: | |
| {question} | |
| ### Answer: | |
| """ | |
| full_prompt = " ".join(full_prompt.split()) | |
| self.generation_config.max_new_tokens = max_tokens | |
| self.generation_config.temperature = temp | |
| result = self.pipeline(full_prompt, generation_config = self.generation_config)[0]['generated_text'] | |
| match = re.search(r'### Answer:(.*?)###', result, re.DOTALL) | |
| if match: | |
| result = match.group(1).strip() | |
| pattern = r"### Answer:(.*)" | |
| match = re.search(pattern, result) | |
| if match: | |
| result = match.group(1).strip() | |
| return result |