INNER CODE UNIT · Python

learning_rate

shibing624/pytextclassifier · pytextclassifier/bert_classfication_utils.py:87

    learning_rate: float = 4e-5
    local_rank: int = -1
    logging_steps: int = 50
    loss_type: str = None
    loss_args: dict = field(default_factory=dict)
    manual_seed: int = None
    max_grad_norm: float = 1.0
    max_seq_length: int = 128
    model_name: str = None
    model_type: str = None
    multiprocessing_chunksize: int = -1
    n_gpu: int = 1
    no_cache: bool = True
    no_save: bool = False
    not_saved_args: list = field(default_factory=list)
    num_train_epochs: int = 1
    optimizer: str = "AdamW"
    output_dir: str = "outputs/"

View source record →