def post(self): parser = reqparse.RequestParser() parser.add_argument('question', required=True) args = parser.parse_args() # Configs logger = logging.getLogger(__name__) MODEL_CONFIG_CLASSES = list(MODEL_WITH_LM_HEAD_MAPPING.keys()) MODEL_TYPES = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES) # parser=argparse.ArgumentParser(description="talk to Digital Jesus") # parser.add_argument("text",type=str,help='message to send to digital Jesus') # args=parser.parse_args() # print(args.text) step = 0 new_user_input_ids = tokenizer.encode( (f">> User:{args.question}") + tokenizer.eos_token, return_tensors='pt') bot_input_ids = new_user_input_ids chat_history_ids = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id, top_p=0.92, top_k=50) answer = "Jesus: {}".format( tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True)) return {'answer': answer}, 200
DataCollatorForLanguageModeling, DataCollatorForPermutationLanguageModeling, HfArgumentParser, LineByLineTextDataset, PreTrainedTokenizer, TextDataset, Trainer, TrainingArguments, set_seed, ) logger = logging.getLogger(__name__) MODEL_CONFIG_CLASSES = list(MODEL_WITH_LM_HEAD_MAPPING.keys()) MODEL_TYPES = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES) @dataclass class ModelArguments: """ Arguments pertaining to which model/config/tokenizer we are going to fine-tune, or train from scratch. """ model_name_or_path: Optional[str] = field( default=None, metadata={ "help": "The model checkpoint for weights initialization. Leave None if you want to train a model from scratch." }, )
AutoTokenizer, # preprocess batches of tensors for MLM DataCollatorForLanguageModeling, HfArgumentParser, PreTrainedTokenizer, Trainer, TrainingArguments, set_seed, ) from loguru import logger from paccmann_proteomics.data.datasets.language_modeling import ( LineByLineTextDatasetCached, LineByLineTextDatasetChunksCached, LineByLineTextDataset) MODEL_CONFIG_CLASSES = list( MODEL_WITH_LM_HEAD_MAPPING.keys()) # [ModelConfig, RobertaConfig, ...] MODEL_TYPES = tuple( conf.model_type for conf in MODEL_CONFIG_CLASSES) # ('roberta', 'bert', ...) @dataclass class ModelArguments: """ Arguments pertaining to which model/config/tokenizer we are going to fine-tune, or train from scratch. """ model_name_or_path: Optional[str] = field( default=None, metadata={ 'help':