View source on GitHub
|
Configurable model options for a BERT model.
mediapipe_model_maker.text_classifier.BertModelOptions(
seq_len: int = 128, do_fine_tuning: bool = True, dropout_rate: float = 0.1
)
See https://arxiv.org/abs/1810.04805 (BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding) for more details.
Attributes: seq_len: Length of the sequence to feed into the model. do_fine_tuning: If true, then the BERT model is not frozen for training. dropout_rate: The rate for dropout.
Attributes | |
|---|---|
seq_len
|
Dataclass field |
do_fine_tuning
|
Dataclass field |
dropout_rate
|
Dataclass field |
Methods
__eq__
__eq__(
other
)
Class Variables | |
|---|---|
| do_fine_tuning |
True
|
| dropout_rate |
0.1
|
| seq_len |
128
|