def finetune(self, Xs, Y=None, batch_size=None):
"""
:param \*Xs: lists of text inputs, shape [batch, n_fields]
:param Y: floating point targets
:param batch_size: integer number of examples per batch. When N_GPUS > 1, this number
corresponds to the number of training examples provided to each GPU.
"""
return BaseModel.finetune(self, Xs, Y=Y, batch_size=batch_size)
def predict(self, Xs, context=None, **kwargs):
"""
Produces list of most likely class labels as determined by the fine-tuned model.
:param \*Xs: lists of text inputs, shape [batch, n_fields]
:returns: list of class labels.
"""
return BaseModel.predict(self, Xs, context=context, **kwargs)
def predict_proba(self, Xs, context=None, **kwargs):
"""
Produces probability distribution over classes for each example in X.
:param \*Xs: lists of text inputs, shape [batch, n_fields]
:returns: list of dictionaries. Each dictionary maps from X2 class label to its assigned class probability.
"""
return BaseModel.predict_proba(self, Xs, context=context, **kwargs)
def featurize(self, Xs, **kwargs):
"""
Embeds inputs in learned feature space. Can be called before or after calling :meth:`finetune`.
:param \*Xs: lists of text inputs, shape [batch, n_fields]
:returns: np.array of features of shape (n_examples, embedding_size).
"""
return BaseModel.featurize(self, Xs, **kwargs)
def finetune(self, Xs, Y=None, batch_size=None):
"""
:param \*Xs: lists of text inputs, shape [batch, n_fields]
:param Y: integer or string-valued class labels. It is necessary for the items of Y to be sortable.
:param batch_size: integer number of examples per batch. When N_GPUS > 1, this number
corresponds to the number of training examples provided to each GPU.
"""
return BaseModel.finetune(self, Xs, Y=Y, batch_size=batch_size)
def featurize(self, pairs):
"""
Embeds inputs in learned feature space. Can be called before or after calling :meth:`finetune`.
:param pairs: Array of text, shape [batch, 2]
:returns: np.array of features of shape (n_examples, embedding_size).
"""
return BaseModel.featurize(self, pairs)
def featurize(self, questions, answers):
"""
Embeds inputs in learned feature space. Can be called before or after calling :meth:`finetune`.
:param questions: List or array of text, shape [batch]
:param answers: List or array of text, shape [n_answers, batch]
:returns: np.array of features of shape (n_examples, embedding_size).
"""
return BaseModel.featurize(self, zip(questions, answers))
def predict(self, pairs):
"""
Produces a list of most likely class labels as determined by the fine-tuned model.
:param pairs: Array of text, shape [batch, 2]
:returns: list of class labels.
"""
return BaseModel.predict(self, pairs)
def predict_proba(self, pairs):
"""
Produces a probability distribution over classes for each example in X.
:param pairs: Array of text, shape [batch, 2]
:returns: list of dictionaries. Each dictionary maps from a class label to its assigned class probability.
"""
return BaseModel.predict_proba(self, pairs)
def predict(self, Xs, max_length=None):
"""
Produces list of most likely class labels as determined by the fine-tuned model.
:param \*Xs: lists of text inputs, shape [batch, n_fields]
:param max_length: the number of tokens to be included in the document representation.
Providing more than `max_length` tokens as input will result in truncation.
:returns: list of class labels.
"""
return BaseModel.predict(self, Xs, max_length=max_length)
def featurize(self, pairs, max_length=None):
"""
Embeds inputs in learned feature space. Can be called before or after calling :meth:`finetune`.
:param pairs: Array of text, shape [batch, 2]
:param max_length: the number of byte-pair encoded tokens to be included in the document representation.
Providing more than `max_length` tokens as input will result in truncation.
:returns: np.array of features of shape (n_examples, embedding_size).
"""
return BaseModel.featurize(self, pairs, max_length=max_length)
def predict_proba(self, Xs, max_length=None):
"""
Produces probability distribution over classes for each example in X.
:param \*Xs: lists of text inputs, shape [batch, n_fields]
:param max_length: the number of tokens to be included in the document representation.
Providing more than `max_length` tokens as input will result in truncation.
:returns: list of dictionaries. Each dictionary maps from X2 class label to its assigned class probability.
"""
return BaseModel.predict_proba(self, Xs, max_length=max_length)
def predict(self, pairs, max_length=None):
"""
Produces a list of most likely class labels as determined by the fine-tuned model.
:param pairs: Array of text, shape [batch, 2]
:param max_length: the number of byte-pair encoded tokens to be included in the document representation.
Providing more than `max_length` tokens as input will result in truncation.
:returns: list of class labels.
"""
return BaseModel.predict(self, pairs, max_length=max_length)
def predict_proba(self, pairs, max_length=None):
"""
Produces a probability distribution over classes for each example in X.
:param pairs: Array of text, shape [batch, 2]
:param max_length: the number of byte-pair encoded tokens to be included in the document representation.
Providing more than `max_length` tokens as input will result in truncation.
:returns: list of dictionaries. Each dictionary maps from a class label to its assigned class probability.
"""
return BaseModel.predict_proba(self, pairs, max_length=max_length)
def predict(self, questions, answers):
"""
Produces a list of most likely class labels as determined by the fine-tuned model.
:param question: List or array of text, shape [batch]
:param answers: List or array of text, shape [batch, n_answers]
:returns: list of class labels.
"""
raw_ids = BaseModel.predict(self, list(zip(questions, answers)))
return [ans[i] for ans, i in zip(answers, raw_ids)]
def predict_proba(self, X1, X2, max_length=None):
"""
Produces a probability distribution over classes for each example in X.
:param X1: List or array of text, shape [batch]
:param X2: List or array of text, shape [batch]
:param max_length: the number of byte-pair encoded tokens to be included in the document representation.
Providing more than `max_length` tokens as input will result in truncation.
:returns: list of dictionaries. Each dictionary maps from a class label to its assigned class probability.
"""
return BaseModel.predict_proba(self, X1, X2, max_length=max_length)
def predict(self, questions, answers, max_length=None):
"""
Produces a list of most likely class labels as determined by the fine-tuned model.
:param question: List or array of text, shape [batch]
:param answers: List or array of text, shape [batch, n_answers]
:param max_length: the number of byte-pair encoded tokens to be included in the document representation.
Providing more than `max_length` tokens as input will result in truncation.
:returns: list of class labels.
"""
raw_ids = BaseModel.predict(self,
list(zip(questions, answers)),
max_length=max_length)
return [ans[i] for ans, i in zip(zip(*answers), raw_ids)]
