def metrics_builder():
"""Returns a `list` of `tf.keras.metric.Metric` objects."""
return [
keras_metrics.NumBatchesCounter(),
keras_metrics.NumExamplesCounter(),
keras_metrics.FlattenedNumExamplesCounter(name='num_tokens',
mask_zero=True),
keras_metrics.FlattenedCategoricalAccuracy(vocab_size=VOCAB_SIZE,
mask_zero=True),
]
def metrics_builder():
"""Returns a `list` of `tf.keras.metric.Metric` objects."""
pad_token, _, _, _ = shakespeare_dataset.get_special_tokens()
return [
keras_metrics.NumBatchesCounter(),
keras_metrics.NumExamplesCounter(),
keras_metrics.NumTokensCounter(masked_tokens=[pad_token]),
keras_metrics.MaskedCategoricalAccuracy(masked_tokens=[pad_token]),
]
def metrics_builder():
return [
keras_metrics.MaskedCategoricalAccuracy(name='accuracy_with_oov',
masked_tokens=[pad_token]),
keras_metrics.MaskedCategoricalAccuracy(
name='accuracy_no_oov', masked_tokens=[pad_token, oov_token]),
# Notice BOS never appears in ground truth.
keras_metrics.MaskedCategoricalAccuracy(
name='accuracy_no_oov_or_eos',
masked_tokens=[pad_token, oov_token, eos_token]),
keras_metrics.NumBatchesCounter(),
keras_metrics.NumTokensCounter(masked_tokens=[pad_token])
]
def metrics_builder():
return [
keras_metrics.FlattenedCategoricalAccuracy(
# Plus 4 for PAD, OOV, BOS and EOS.
vocab_size=FLAGS.vocab_size + 4,
name='accuracy_with_oov',
masked_tokens=pad_token),
keras_metrics.FlattenedCategoricalAccuracy(
vocab_size=FLAGS.vocab_size + 4,
name='accuracy_no_oov',
masked_tokens=[pad_token, oov_token]),
# Notice BOS never appears in ground truth.
keras_metrics.FlattenedCategoricalAccuracy(
vocab_size=FLAGS.vocab_size + 4,
name='accuracy_no_oov_or_eos',
masked_tokens=[pad_token, oov_token, eos_token]),
keras_metrics.NumBatchesCounter(),
keras_metrics.FlattenedNumExamplesCounter(name='num_tokens',
mask_zero=True),
]
