the general evaluator class More...
Inheritance diagram for nabu.neuralnetworks.evaluators.evaluator.Evaluator:
Public Member Functions | |
| def | __init__ (self, conf, dataconf, model) |
| Evaluator constructor. More... | |
| def | evaluate (self) |
| evaluate the performance of the model More... | |
| def | update_loss (self, loss, inputs, input_seq_length, targets, target_seq_length) |
| update the validation loss for a batch of data More... | |
Public Attributes | |
| conf | |
| model | |
| input_dataconfs | |
| target_dataconfs | |
Detailed Description
the general evaluator class
an evaluator is used to evaluate the performance of a model
Constructor & Destructor Documentation
§ __init__()
| def nabu.neuralnetworks.evaluators.evaluator.Evaluator.__init__ | ( | self, | |
| conf, | |||
| dataconf, | |||
| model | |||
| ) |
Evaluator constructor.
- Parameters
-
conf the evaluator configuration as a ConfigParser dataconf the database configuration model the model to be evaluated
Member Function Documentation
§ evaluate()
| def nabu.neuralnetworks.evaluators.evaluator.Evaluator.evaluate | ( | self | ) |
evaluate the performance of the model
- Returns
- the loss as a scalar tensor
- an operation to update the loss
- the number of batches in the validation set as an integer
§ update_loss()
| def nabu.neuralnetworks.evaluators.evaluator.Evaluator.update_loss | ( | self, | |
| loss, | |||
| inputs, | |||
| input_seq_length, | |||
| targets, | |||
| target_seq_length | |||
| ) |
update the validation loss for a batch of data
- Parameters
-
loss the current loss inputs the inputs to the neural network, this is a list of [batch_size x ...] tensors input_seq_length The sequence lengths of the input utterances, this is a list of [batch_size] vectors targets the targets to the neural network, this is a list of [batch_size x max_output_length] tensors. target_seq_length The sequence lengths of the target utterances, this is a list of [batch_size] vectors
- Returns
- an operation to update the loss
The documentation for this class was generated from the following file:
