def predict(self, predict_data=None, return_logits=True):
if not self.model:
raise RuntimeError(
'You must compile your model before training/testing/predicting. Use `trainer.build()`.'
)
cache = self.cache
cfg = self.cfg.predict
cfg.return_logits = return_logits
if predict_data is None:
predict_data = np.arange(self.graph.num_nodes)
if not isinstance(predict_data, Sequence):
predict_data = gf.asarray(predict_data)
predict_data = self.predict_sequence(predict_data)
cache.predict_data = predict_data
logits = self.predict_step(predict_data)
if not return_logits:
logits = softmax(logits)
return logits.squeeze()
def predict(self, predict_data=None, return_prob=True):
"""
Predict the output probability for the input data.
Note:
----------
You must compile your model before training/testing/predicting.
Use `model.build()`.
Parameters:
----------
predict_data: Numpy 1D array, optional.
The indices of objects to predict.
if None, predict the all objects.
return_prob: bool.
whether to return the probability of prediction.
Return:
----------
The predicted probability of each class for each object,
for node classification task, it has shape
(num_nodes, num_node_classes).
"""
if not self.model:
raise RuntimeError(
'You must compile your model before training/testing/predicting. Use `model.build()`.'
)
if predict_data is None:
predict_data = np.arange(self.graph.num_nodes, dtype=gg.intx())
if not isinstance(predict_data, Sequence):
predict_data = self.predict_sequence(predict_data)
self.predict_data = predict_data
logit = self.predict_step(predict_data)
if return_prob:
logit = softmax(logit)
return logit
def predict(self, index=None, return_prob=True):
"""
Predict the output probability for the input node index.
Note:
----------
You must compile your model before training/testing/predicting.
Use `model.build()`.
Parameters:
----------
index: Numpy 1D array, optional.
The indices of nodes to predict.
if None, predict the all nodes.
return_prob: bool.
whether to return the probability of prediction.
Return:
----------
The predicted probability of each class for each node,
shape (n_nodes, n_classes).
"""
if not self.model:
raise RuntimeError(
'You must compile your model before training/testing/predicting. Use `model.build()`.'
)
if index is None:
index = np.arange(self.graph.n_nodes, dtype=intx())
else:
index = asintarr(index)
sequence = self.predict_sequence(index)
logit = self.predict_step(sequence)
if return_prob:
logit = softmax(logit)
return logit
