def fit(self, trn_data, dev_data, save_dir,
n_embed=100,
pretrained_embed=None,
embed_dropout=.33,
n_lstm_hidden=400,
n_lstm_layers=3,
lstm_dropout=.33,
n_mlp_arc=500,
n_mlp_rel=100,
mlp_dropout=.33,
optimizer='adam',
lr=2e-3,
mu=.9,
nu=.9,
epsilon=1e-12,
clip=5.0,
decay=.75,
decay_steps=5000,
patience=100,
arc_loss='sparse_categorical_crossentropy',
rel_loss='sparse_categorical_crossentropy',
metrics=('UAS', 'LAS'),
n_buckets=32,
batch_size=5000,
epochs=50000,
early_stopping_patience=100,
tree=False,
punct=False,
min_freq=2,
run_eagerly=False, logger=None, verbose=True,
**kwargs):
return super().fit(**merge_locals_kwargs(locals(), kwargs))
def fit(self, trn_data: Any, dev_data: Any, save_dir: str, word_embed: Union[str, int, dict] = 200,
ngram_embed: Union[str, int, dict] = 50, embedding_trainable=True, window_size=4, kernel_size=3,
filters=(200, 200, 200, 200, 200), dropout_embed=0.2, dropout_hidden=0.2, weight_norm=True,
loss: Union[tf.keras.losses.Loss, str] = None,
optimizer: Union[str, tf.keras.optimizers.Optimizer] = 'adam', metrics='f1', batch_size=100,
epochs=100, logger=None, verbose=True, **kwargs):
return super().fit(**merge_locals_kwargs(locals(), kwargs))
def fit(self,
trn_data: Any,
dev_data: Any,
save_dir: str,
word_embed: Union[str, int, dict] = 200,
ngram_embed: Union[str, int, dict] = 50,
embedding_trainable=True,
window_size=4,
kernel_size=3,
filters=(200, 200, 200, 200, 200),
dropout_embed=0.2,
dropout_hidden=0.2,
weight_norm=True,
loss: Union[tf.keras.losses.Loss, str] = None,
optimizer: Union[str, tf.keras.optimizers.Optimizer] = 'adam',
metrics='accuracy',
batch_size=100,
epochs=100,
logger=None,
verbose=True,
**kwargs):
assert kwargs.get('run_eagerly',
True), 'NgramConvTaggingModel can only run eagerly'
kwargs['run_eagerly'] = True
return super().fit(**merge_locals_kwargs(locals(), kwargs))
def __init__(self, config: SerializableDict = None, map_x=True, map_y=True, lower=True, n_buckets=32,
n_tokens_per_batch=5000, min_freq=2,
**kwargs) -> None:
super().__init__(**merge_locals_kwargs(locals(), kwargs))
self.form_vocab: Vocab = None
self.cpos_vocab: Vocab = None
self.rel_vocab: Vocab = None
self.puncts: tf.Tensor = None
def __init__(self,
config: SerializableDict = None,
map_x=True,
map_y=True,
use_char=False,
**kwargs) -> None:
super().__init__(**merge_locals_kwargs(locals(), kwargs))
self.word_vocab: Optional[Vocab] = None
self.tag_vocab: Optional[Vocab] = None
self.char_vocab: Optional[Vocab] = None
def fit(self,
trn_data: Any,
dev_data: Any,
save_dir: str,
transformer: str,
max_length: int = 128,
optimizer='adamw',
warmup_steps_ratio=0.1,
use_amp=False,
batch_size=32,
epochs=3,
logger=None,
verbose=1,
**kwargs):
return super().fit(**merge_locals_kwargs(locals(), kwargs))
def fit(self, trn_data, dev_data, save_dir,
transformer,
optimizer='adamw',
learning_rate=5e-5,
weight_decay_rate=0,
epsilon=1e-8,
clipnorm=1.0,
warmup_steps_ratio=0,
use_amp=False,
max_seq_length=128,
batch_size=32,
epochs=3,
metrics='accuracy',
run_eagerly=False,
logger=None,
verbose=True,
**kwargs):
return super().fit(**merge_locals_kwargs(locals(), kwargs))
def fit(self,
trn_data: str,
dev_data: str = None,
save_dir: str = None,
embeddings=100,
embedding_trainable=False,
rnn_input_dropout=0.2,
rnn_units=100,
rnn_output_dropout=0.2,
epochs=20,
lower=False,
logger=None,
loss: Union[tf.keras.losses.Loss, str] = None,
optimizer: Union[str, tf.keras.optimizers.Optimizer] = 'adam',
metrics='f1',
batch_size=32,
dev_batch_size=32,
lr_decay_per_epoch=None,
verbose=True,
**kwargs):
return super().fit(**merge_locals_kwargs(locals(), kwargs))
