Merge branch 'panoptic-segmentation' into panoptic-segmentation

official/nlp/projects/teams/teams.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""TEAMS model configurations and instantiation methods."""
+import dataclasses
+
+import gin
+import tensorflow as tf
+
+from official.modeling import tf_utils
+from official.modeling.hyperparams import base_config
+from official.nlp.configs import encoders
+from official.nlp.modeling import layers
+from official.nlp.modeling import networks
+
+
+@dataclasses.dataclass
+class TeamsPretrainerConfig(base_config.Config):
+  """Teams pretrainer configuration."""
+  # Candidate size for multi-word selection task, including the correct word.
+  candidate_size: int = 5
+  # Weight for the generator masked language model task.
+  generator_loss_weight: float = 1.0
+  # Weight for the replaced token detection task.
+  discriminator_rtd_loss_weight: float = 5.0
+  # Weight for the multi-word selection task.
+  discriminator_mws_loss_weight: float = 2.0
+  # Whether share embedding network between generator and discriminator.
+  tie_embeddings: bool = True
+  # Number of bottom layers shared between generator and discriminator.
+  # Non-positive value implies no sharing.
+  num_shared_generator_hidden_layers: int = 3
+  # Number of bottom layers shared between different discriminator tasks.
+  num_discriminator_task_agnostic_layers: int = 11
+  generator: encoders.BertEncoderConfig = encoders.BertEncoderConfig()
+  discriminator: encoders.BertEncoderConfig = encoders.BertEncoderConfig()
+  # Used for compatibility with continuous finetuning where common BERT config
+  # is used.
+  encoder: encoders.EncoderConfig = encoders.EncoderConfig()
+
+
+@gin.configurable
+def get_encoder(bert_config,
+                embedding_network=None,
+                hidden_layers=layers.Transformer):
+  """Gets a 'EncoderScaffold' object.
+
+  Args:
+    bert_config: A 'modeling.BertConfig'.
+    embedding_network: Embedding network instance.
+    hidden_layers: List of hidden layer instances.
+
+  Returns:
+    A encoder object.
+  """
+  # embedding_size is required for PackedSequenceEmbedding.
+  if bert_config.embedding_size is None:
+    bert_config.embedding_size = bert_config.hidden_size
+  embedding_cfg = dict(
+      vocab_size=bert_config.vocab_size,
+      type_vocab_size=bert_config.type_vocab_size,
+      hidden_size=bert_config.hidden_size,
+      embedding_width=bert_config.embedding_size,
+      max_seq_length=bert_config.max_position_embeddings,
+      initializer=tf.keras.initializers.TruncatedNormal(
+          stddev=bert_config.initializer_range),
+      dropout_rate=bert_config.dropout_rate,
+  )
+  hidden_cfg = dict(
+      num_attention_heads=bert_config.num_attention_heads,
+      intermediate_size=bert_config.intermediate_size,
+      intermediate_activation=tf_utils.get_activation(
+          bert_config.hidden_activation),
+      dropout_rate=bert_config.dropout_rate,
+      attention_dropout_rate=bert_config.attention_dropout_rate,
+      kernel_initializer=tf.keras.initializers.TruncatedNormal(
+          stddev=bert_config.initializer_range),
+  )
+  if embedding_network is None:
+    embedding_network = networks.PackedSequenceEmbedding(**embedding_cfg)
+  kwargs = dict(
+      embedding_cfg=embedding_cfg,
+      embedding_cls=embedding_network,
+      hidden_cls=hidden_layers,
+      hidden_cfg=hidden_cfg,
+      num_hidden_instances=bert_config.num_layers,
+      pooled_output_dim=bert_config.hidden_size,
+      pooler_layer_initializer=tf.keras.initializers.TruncatedNormal(
+          stddev=bert_config.initializer_range),
+      dict_outputs=True)
+
+  # Relies on gin configuration to define the Transformer encoder arguments.
+  return networks.encoder_scaffold.EncoderScaffold(**kwargs)

official/nlp/projects/teams/teams_experiments.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Lint as: python3
+# pylint: disable=g-doc-return-or-yield,line-too-long
+"""TEAMS experiments."""
+import dataclasses
+from official.core import config_definitions as cfg
+from official.core import exp_factory
+from official.modeling import optimization
+from official.nlp.data import pretrain_dataloader
+from official.nlp.projects.teams import teams_task
+
+
+AdamWeightDecay = optimization.AdamWeightDecayConfig
+PolynomialLr = optimization.PolynomialLrConfig
+PolynomialWarmupConfig = optimization.PolynomialWarmupConfig
+
+
+@dataclasses.dataclass
+class TeamsOptimizationConfig(optimization.OptimizationConfig):
+  """TEAMS optimization config."""
+  optimizer: optimization.OptimizerConfig = optimization.OptimizerConfig(
+      type="adamw",
+      adamw=AdamWeightDecay(
+          weight_decay_rate=0.01,
+          exclude_from_weight_decay=["LayerNorm", "layer_norm", "bias"],
+          epsilon=1e-6))
+  learning_rate: optimization.LrConfig = optimization.LrConfig(
+      type="polynomial",
+      polynomial=PolynomialLr(
+          initial_learning_rate=1e-4,
+          decay_steps=1000000,
+          end_learning_rate=0.0))
+  warmup: optimization.WarmupConfig = optimization.WarmupConfig(
+      type="polynomial", polynomial=PolynomialWarmupConfig(warmup_steps=10000))
+
+
+@exp_factory.register_config_factory("teams/pretraining")
+def teams_pretrain() -> cfg.ExperimentConfig:
+  """TEAMS pretraining."""
+  config = cfg.ExperimentConfig(
+      task=teams_task.TeamsPretrainTaskConfig(
+          train_data=pretrain_dataloader.BertPretrainDataConfig(),
+          validation_data=pretrain_dataloader.BertPretrainDataConfig(
+              is_training=False)),
+      trainer=cfg.TrainerConfig(
+          optimizer_config=TeamsOptimizationConfig(), train_steps=1000000),
+      restrictions=[
+          "task.train_data.is_training != None",
+          "task.validation_data.is_training != None"
+      ])
+  return config

official/nlp/projects/teams/teams_experiments_test.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Lint as: python3
+"""Tests for teams_experiments."""
+
+from absl.testing import parameterized
+import tensorflow as tf
+
+# pylint: disable=unused-import
+from official.common import registry_imports
+# pylint: enable=unused-import
+from official.core import config_definitions as cfg
+from official.core import exp_factory
+
+
+class TeamsExperimentsTest(tf.test.TestCase, parameterized.TestCase):
+
+  @parameterized.parameters(('teams/pretraining',))
+  def test_teams_experiments(self, config_name):
+    config = exp_factory.get_exp_config(config_name)
+    self.assertIsInstance(config, cfg.ExperimentConfig)
+    self.assertIsInstance(config.task.train_data, cfg.DataConfig)
+
+
+if __name__ == '__main__':
+  tf.test.main()

official/nlp/projects/teams/teams_pretrainer.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Trainer network for TEAMS models."""
+# pylint: disable=g-classes-have-attributes
+
+import tensorflow as tf
+
+from official.modeling import tf_utils
+from official.nlp.modeling import layers
+from official.nlp.modeling import models
+
+
+class ReplacedTokenDetectionHead(tf.keras.layers.Layer):
+  """Replaced token detection discriminator head.
+
+  Arguments:
+    encoder_cfg: Encoder config, used to create hidden layers and head.
+    num_task_agnostic_layers: Number of task agnostic layers in the
+      discriminator.
+    output: The output style for this network. Can be either 'logits' or
+      'predictions'.
+  """
+
+  def __init__(self,
+               encoder_cfg,
+               num_task_agnostic_layers,
+               output='logits',
+               name='rtd',
+               **kwargs):
+    super(ReplacedTokenDetectionHead, self).__init__(name=name, **kwargs)
+    self.num_task_agnostic_layers = num_task_agnostic_layers
+    self.hidden_size = encoder_cfg['embedding_cfg']['hidden_size']
+    self.num_hidden_instances = encoder_cfg['num_hidden_instances']
+    self.hidden_cfg = encoder_cfg['hidden_cfg']
+    self.activation = self.hidden_cfg['intermediate_activation']
+    self.initializer = self.hidden_cfg['kernel_initializer']
+
+    self.hidden_layers = []
+    for i in range(self.num_task_agnostic_layers, self.num_hidden_instances):
+      self.hidden_layers.append(
+          layers.Transformer(
+              num_attention_heads=self.hidden_cfg['num_attention_heads'],
+              intermediate_size=self.hidden_cfg['intermediate_size'],
+              intermediate_activation=self.activation,
+              dropout_rate=self.hidden_cfg['dropout_rate'],
+              attention_dropout_rate=self.hidden_cfg['attention_dropout_rate'],
+              kernel_initializer=self.initializer,
+              name='transformer/layer_%d_rtd' % i))
+    self.dense = tf.keras.layers.Dense(
+        self.hidden_size,
+        activation=self.activation,
+        kernel_initializer=self.initializer,
+        name='transform/rtd_dense')
+    self.rtd_head = tf.keras.layers.Dense(
+        units=1, kernel_initializer=self.initializer,
+        name='transform/rtd_head')
+
+    if output not in ('predictions', 'logits'):
+      raise ValueError(
+          ('Unknown `output` value "%s". `output` can be either "logits" or '
+           '"predictions"') % output)
+    self._output_type = output
+
+  def call(self, sequence_data, input_mask):
+    """Compute inner-products of hidden vectors with sampled element embeddings.
+
+    Args:
+      sequence_data: A [batch_size, seq_length, num_hidden] tensor.
+      input_mask: A [batch_size, seq_length] binary mask to separate the input
+        from the padding.
+
+    Returns:
+      A [batch_size, seq_length] tensor.
+    """
+    attention_mask = layers.SelfAttentionMask()([sequence_data, input_mask])
+    data = sequence_data
+    for hidden_layer in self.hidden_layers:
+      data = hidden_layer([sequence_data, attention_mask])
+    rtd_logits = self.rtd_head(self.dense(data))
+    return tf.squeeze(rtd_logits, axis=-1)
+
+
+class MultiWordSelectionHead(tf.keras.layers.Layer):
+  """Multi-word selection discriminator head.
+
+  Arguments:
+    embedding_table: The embedding table.
+    activation: The activation, if any, for the dense layer.
+    initializer: The intializer for the dense layer. Defaults to a Glorot
+      uniform initializer.
+    output: The output style for this network. Can be either 'logits' or
+      'predictions'.
+  """
+
+  def __init__(self,
+               embedding_table,
+               activation=None,
+               initializer='glorot_uniform',
+               output='logits',
+               name='mws',
+               **kwargs):
+    super(MultiWordSelectionHead, self).__init__(name=name, **kwargs)
+    self.embedding_table = embedding_table
+    self.activation = activation
+    self.initializer = tf.keras.initializers.get(initializer)
+
+    self._vocab_size, self.embed_size = self.embedding_table.shape
+    self.dense = tf.keras.layers.Dense(
+        self.embed_size,
+        activation=self.activation,
+        kernel_initializer=self.initializer,
+        name='transform/mws_dense')
+    self.layer_norm = tf.keras.layers.LayerNormalization(
+        axis=-1, epsilon=1e-12, name='transform/mws_layernorm')
+
+    if output not in ('predictions', 'logits'):
+      raise ValueError(
+          ('Unknown `output` value "%s". `output` can be either "logits" or '
+           '"predictions"') % output)
+    self._output_type = output
+
+  def call(self, sequence_data, masked_positions, candidate_sets):
+    """Compute inner-products of hidden vectors with sampled element embeddings.
+
+    Args:
+      sequence_data: A [batch_size, seq_length, num_hidden] tensor.
+      masked_positions: A [batch_size, num_prediction] tensor.
+      candidate_sets: A [batch_size, num_prediction, k] tensor.
+
+    Returns:
+      A [batch_size, num_prediction, k] tensor.
+    """
+    # Gets shapes for later usage
+    candidate_set_shape = tf_utils.get_shape_list(candidate_sets)
+    num_prediction = candidate_set_shape[1]
+
+    # Gathers hidden vectors -> (batch_size, num_prediction, 1, embed_size)
+    masked_lm_input = self._gather_indexes(sequence_data, masked_positions)
+    lm_data = self.dense(masked_lm_input)
+    lm_data = self.layer_norm(lm_data)
+    lm_data = tf.expand_dims(
+        tf.reshape(lm_data, [-1, num_prediction, self.embed_size]), 2)
+
+    # Gathers embeddings -> (batch_size, num_prediction, embed_size, k)
+    flat_candidate_sets = tf.reshape(candidate_sets, [-1])
+    candidate_embeddings = tf.gather(self.embedding_table, flat_candidate_sets)
+    candidate_embeddings = tf.reshape(
+        candidate_embeddings,
+        tf.concat([tf.shape(candidate_sets), [self.embed_size]], axis=0)
+    )
+    candidate_embeddings.set_shape(
+        candidate_sets.shape.as_list() + [self.embed_size])
+    candidate_embeddings = tf.transpose(candidate_embeddings, [0, 1, 3, 2])
+
+    # matrix multiplication + squeeze -> (batch_size, num_prediction, k)
+    logits = tf.matmul(lm_data, candidate_embeddings)
+    logits = tf.squeeze(logits, 2)
+
+    if self._output_type == 'logits':
+      return logits
+    return tf.nn.log_softmax(logits)
+
+  def _gather_indexes(self, sequence_tensor, positions):
+    """Gathers the vectors at the specific positions.
+
+    Args:
+        sequence_tensor: Sequence output of shape
+          (`batch_size`, `seq_length`, `num_hidden`) where `num_hidden` is
+          number of hidden units.
+        positions: Positions ids of tokens in batched sequences.
+
+    Returns:
+        Sequence tensor of shape (batch_size * num_predictions,
+        num_hidden).
+    """
+    sequence_shape = tf_utils.get_shape_list(
+        sequence_tensor, name='sequence_output_tensor')
+    batch_size, seq_length, width = sequence_shape
+
+    flat_offsets = tf.reshape(
+        tf.range(0, batch_size, dtype=tf.int32) * seq_length, [-1, 1])
+    flat_positions = tf.reshape(positions + flat_offsets, [-1])
+    flat_sequence_tensor = tf.reshape(sequence_tensor,
+                                      [batch_size * seq_length, width])
+    output_tensor = tf.gather(flat_sequence_tensor, flat_positions)
+
+    return output_tensor
+
+
+@tf.keras.utils.register_keras_serializable(package='Text')
+class TeamsPretrainer(tf.keras.Model):
+  """TEAMS network training model.
+
+  This is an implementation of the network structure described in "Training
+  ELECTRA Augmented with Multi-word Selection"
+  (https://arxiv.org/abs/2106.00139).
+
+  The TeamsPretrainer allows a user to pass in two transformer encoders, one
+  for generator, the other for discriminator (multi-word selection). The
+  pretrainer then instantiates the masked language model (at generator side) and
+  classification networks (including both multi-word selection head and replaced
+  token detection head) that are used to create the training objectives.
+
+  *Note* that the model is constructed by Keras Subclass API, where layers are
+  defined inside `__init__` and `call()` implements the computation.
+
+  Args:
+    generator_network: A transformer encoder for generator, this network should
+      output a sequence output.
+    discriminator_mws_network: A transformer encoder for multi-word selection
+      discriminator, this network should output a sequence output.
+    num_discriminator_task_agnostic_layers: Number of layers shared between
+      multi-word selection and random token detection discriminators.
+    vocab_size: Size of generator output vocabulary
+    candidate_size: Candidate size for multi-word selection task,
+      including the correct word.
+    mlm_activation: The activation (if any) to use in the masked LM and
+      classification networks. If None, no activation will be used.
+    mlm_initializer: The initializer (if any) to use in the masked LM and
+      classification networks. Defaults to a Glorot uniform initializer.
+    output_type: The output style for this network. Can be either `logits` or
+      `predictions`.
+  """
+
+  def __init__(self,
+               generator_network,
+               discriminator_mws_network,
+               num_discriminator_task_agnostic_layers,
+               vocab_size,
+               candidate_size=5,
+               mlm_activation=None,
+               mlm_initializer='glorot_uniform',
+               output_type='logits',
+               **kwargs):
+    super().__init__()
+    self._config = {
+        'generator_network':
+            generator_network,
+        'discriminator_mws_network':
+            discriminator_mws_network,
+        'num_discriminator_task_agnostic_layers':
+            num_discriminator_task_agnostic_layers,
+        'vocab_size':
+            vocab_size,
+        'candidate_size':
+            candidate_size,
+        'mlm_activation':
+            mlm_activation,
+        'mlm_initializer':
+            mlm_initializer,
+        'output_type':
+            output_type,
+    }
+    for k, v in kwargs.items():
+      self._config[k] = v
+
+    self.generator_network = generator_network
+    self.discriminator_mws_network = discriminator_mws_network
+    self.vocab_size = vocab_size
+    self.candidate_size = candidate_size
+    self.mlm_activation = mlm_activation
+    self.mlm_initializer = mlm_initializer
+    self.output_type = output_type
+    self.embedding_table = (
+        self.discriminator_mws_network.embedding_network.get_embedding_table())
+    self.masked_lm = layers.MaskedLM(
+        embedding_table=self.embedding_table,
+        activation=mlm_activation,
+        initializer=mlm_initializer,
+        output=output_type,
+        name='generator_masked_lm')
+    discriminator_cfg = self.discriminator_mws_network.get_config()
+    self.num_task_agnostic_layers = num_discriminator_task_agnostic_layers
+    self.discriminator_rtd_head = ReplacedTokenDetectionHead(
+        encoder_cfg=discriminator_cfg,
+        num_task_agnostic_layers=self.num_task_agnostic_layers,
+        output=output_type,
+        name='discriminator_rtd')
+    hidden_cfg = discriminator_cfg['hidden_cfg']
+    self.discriminator_mws_head = MultiWordSelectionHead(
+        embedding_table=self.embedding_table,
+        activation=hidden_cfg['intermediate_activation'],
+        initializer=hidden_cfg['kernel_initializer'],
+        output=output_type,
+        name='discriminator_mws')
+
+  def call(self, inputs):
+    """TEAMS forward pass.
+
+    Args:
+      inputs: A dict of all inputs, same as the standard BERT model.
+
+    Returns:
+      outputs: A dict of pretrainer model outputs, including
+        (1) lm_outputs: A `[batch_size, num_token_predictions, vocab_size]`
+        tensor indicating logits on masked positions.
+        (2) disc_rtd_logits: A `[batch_size, sequence_length]` tensor indicating
+        logits for discriminator replaced token detection task.
+        (3) disc_rtd_label: A `[batch_size, sequence_length]` tensor indicating
+        target labels for discriminator replaced token detection task.
+        (4) disc_mws_logits: A `[batch_size, num_token_predictions,
+        candidate_size]` tensor indicating logits for discriminator multi-word
+        selection task.
+        (5) disc_mws_labels: A `[batch_size, num_token_predictions]` tensor
+        indicating target labels for discriminator multi-word selection task.
+    """
+    input_word_ids = inputs['input_word_ids']
+    input_mask = inputs['input_mask']
+    input_type_ids = inputs['input_type_ids']
+    masked_lm_positions = inputs['masked_lm_positions']
+
+    # Runs generator.
+    sequence_output = self.generator_network(
+        [input_word_ids, input_mask, input_type_ids])['sequence_output']
+
+    lm_outputs = self.masked_lm(sequence_output, masked_lm_positions)
+
+    # Samples tokens from generator.
+    fake_data = self._get_fake_data(inputs, lm_outputs)
+
+    # Runs discriminator.
+    disc_input = fake_data['inputs']
+    disc_rtd_label = fake_data['is_fake_tokens']
+    disc_mws_candidates = fake_data['candidate_set']
+    mws_sequence_outputs = self.discriminator_mws_network([
+        disc_input['input_word_ids'], disc_input['input_mask'],
+        disc_input['input_type_ids']
+    ])['encoder_outputs']
+
+    # Applies replaced token detection with input selected based on
+    # self.num_discriminator_task_agnostic_layers
+    disc_rtd_logits = self.discriminator_rtd_head(
+        mws_sequence_outputs[self.num_task_agnostic_layers - 1], input_mask)
+
+    # Applies multi-word selection.
+    disc_mws_logits = self.discriminator_mws_head(mws_sequence_outputs[-1],
+                                                  masked_lm_positions,
+                                                  disc_mws_candidates)
+    disc_mws_label = tf.zeros_like(masked_lm_positions, dtype=tf.int32)
+
+    outputs = {
+        'lm_outputs': lm_outputs,
+        'disc_rtd_logits': disc_rtd_logits,
+        'disc_rtd_label': disc_rtd_label,
+        'disc_mws_logits': disc_mws_logits,
+        'disc_mws_label': disc_mws_label,
+    }
+
+    return outputs
+
+  def _get_fake_data(self, inputs, mlm_logits):
+    """Generate corrupted data for discriminator.
+
+    Note it is poosible for sampled token to be the same as the correct one.
+    Args:
+      inputs: A dict of all inputs, same as the input of `call()` function
+      mlm_logits: The generator's output logits
+
+    Returns:
+      A dict of generated fake data
+    """
+    inputs = models.electra_pretrainer.unmask(inputs, duplicate=True)
+
+    # Samples replaced token.
+    sampled_tokens = tf.stop_gradient(
+        models.electra_pretrainer.sample_from_softmax(
+            mlm_logits, disallow=None))
+    sampled_tokids = tf.argmax(sampled_tokens, axis=-1, output_type=tf.int32)
+
+    # Prepares input and label for replaced token detection task.
+    updated_input_ids, masked = models.electra_pretrainer.scatter_update(
+        inputs['input_word_ids'], sampled_tokids, inputs['masked_lm_positions'])
+    rtd_labels = masked * (1 - tf.cast(
+        tf.equal(updated_input_ids, inputs['input_word_ids']), tf.int32))
+    updated_inputs = models.electra_pretrainer.get_updated_inputs(
+        inputs, duplicate=True, input_word_ids=updated_input_ids)
+
+    # Samples (candidate_size-1) negatives and concat with true tokens
+    disallow = tf.one_hot(
+        inputs['masked_lm_ids'], depth=self.vocab_size, dtype=tf.float32)
+    sampled_candidates = tf.stop_gradient(
+        sample_k_from_softmax(mlm_logits, k=self.candidate_size-1,
+                              disallow=disallow))
+    true_token_id = tf.expand_dims(inputs['masked_lm_ids'], -1)
+    candidate_set = tf.concat([true_token_id, sampled_candidates], -1)
+
+    return {
+        'inputs': updated_inputs,
+        'is_fake_tokens': rtd_labels,
+        'sampled_tokens': sampled_tokens,
+        'candidate_set': candidate_set
+    }
+
+  @property
+  def checkpoint_items(self):
+    """Returns a dictionary of items to be additionally checkpointed."""
+    items = dict(encoder=self.discriminator_mws_network)
+    return items
+
+  def get_config(self):
+    return self._config
+
+  @classmethod
+  def from_config(cls, config, custom_objects=None):
+    return cls(**config)
+
+
+def sample_k_from_softmax(logits, k, disallow=None, use_topk=False):
+  """Implement softmax sampling using gumbel softmax trick to select k items.
+
+  Args:
+    logits: A [batch_size, num_token_predictions, vocab_size] tensor indicating
+      the generator output logits for each masked position.
+    k: Number of samples
+    disallow: If `None`, we directly sample tokens from the logits. Otherwise,
+      this is a tensor of size [batch_size, num_token_predictions, vocab_size]
+      indicating the true word id in each masked position.
+    use_topk: Whether to use tf.nn.top_k or using iterative approach where the
+      latter is empirically faster.
+
+  Returns:
+    sampled_tokens: A [batch_size, num_token_predictions, k] tensor indicating
+    the sampled word id in each masked position.
+  """
+  if use_topk:
+    if disallow is not None:
+      logits -= 10000.0 * disallow
+    uniform_noise = tf.random.uniform(
+        tf_utils.get_shape_list(logits), minval=0, maxval=1)
+    gumbel_noise = -tf.math.log(-tf.math.log(uniform_noise + 1e-9) + 1e-9)
+    _, sampled_tokens = tf.nn.top_k(logits + gumbel_noise, k=k, sorted=False)
+  else:
+    sampled_tokens_list = []
+    vocab_size = tf_utils.get_shape_list(logits)[-1]
+    if disallow is not None:
+      logits -= 10000.0 * disallow
+
+    uniform_noise = tf.random.uniform(
+        tf_utils.get_shape_list(logits), minval=0, maxval=1)
+    gumbel_noise = -tf.math.log(-tf.math.log(uniform_noise + 1e-9) + 1e-9)
+    logits += gumbel_noise
+    for _ in range(k):
+      token_ids = tf.argmax(logits, -1, output_type=tf.int32)
+      sampled_tokens_list.append(token_ids)
+      logits -= 10000.0 *  tf.one_hot(
+          token_ids, depth=vocab_size, dtype=tf.float32)
+    sampled_tokens = tf.stack(sampled_tokens_list, -1)
+  return sampled_tokens

official/nlp/projects/teams/teams_pretrainer_test.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Tests for TEAMS pre trainer network."""
+
+import tensorflow as tf
+
+from tensorflow.python.keras import keras_parameterized  # pylint: disable=g-direct-tensorflow-import
+from official.modeling import activations
+from official.nlp.modeling.networks import encoder_scaffold
+from official.nlp.modeling.networks import packed_sequence_embedding
+from official.nlp.projects.teams import teams_pretrainer
+
+
+# This decorator runs the test in V1, V2-Eager, and V2-Functional mode. It
+# guarantees forward compatibility of this code for the V2 switchover.
+@keras_parameterized.run_all_keras_modes
+class TeamsPretrainerTest(keras_parameterized.TestCase):
+
+  # Build a transformer network to use within the TEAMS trainer.
+  def _get_network(self, vocab_size):
+    sequence_length = 512
+    hidden_size = 50
+    embedding_cfg = {
+        'vocab_size': vocab_size,
+        'type_vocab_size': 1,
+        'hidden_size': hidden_size,
+        'embedding_width': hidden_size,
+        'max_seq_length': sequence_length,
+        'initializer': tf.keras.initializers.TruncatedNormal(stddev=0.02),
+        'dropout_rate': 0.1,
+    }
+    embedding_inst = packed_sequence_embedding.PackedSequenceEmbedding(
+        **embedding_cfg)
+    hidden_cfg = {
+        'num_attention_heads':
+            2,
+        'intermediate_size':
+            3072,
+        'intermediate_activation':
+            activations.gelu,
+        'dropout_rate':
+            0.1,
+        'attention_dropout_rate':
+            0.1,
+        'kernel_initializer':
+            tf.keras.initializers.TruncatedNormal(stddev=0.02),
+    }
+    return encoder_scaffold.EncoderScaffold(
+        num_hidden_instances=2,
+        pooled_output_dim=hidden_size,
+        embedding_cfg=embedding_cfg,
+        embedding_cls=embedding_inst,
+        hidden_cfg=hidden_cfg,
+        dict_outputs=True)
+
+  def test_teams_pretrainer(self):
+    """Validate that the Keras object can be created."""
+    vocab_size = 100
+    test_generator_network = self._get_network(vocab_size)
+    test_discriminator_network = self._get_network(vocab_size)
+
+    # Create a TEAMS trainer with the created network.
+    candidate_size = 3
+    teams_trainer_model = teams_pretrainer.TeamsPretrainer(
+        generator_network=test_generator_network,
+        discriminator_mws_network=test_discriminator_network,
+        num_discriminator_task_agnostic_layers=1,
+        vocab_size=vocab_size,
+        candidate_size=candidate_size)
+
+    # Create a set of 2-dimensional inputs (the first dimension is implicit).
+    num_token_predictions = 2
+    sequence_length = 128
+    word_ids = tf.keras.Input(shape=(sequence_length,), dtype=tf.int32)
+    mask = tf.keras.Input(shape=(sequence_length,), dtype=tf.int32)
+    type_ids = tf.keras.Input(shape=(sequence_length,), dtype=tf.int32)
+    lm_positions = tf.keras.Input(
+        shape=(num_token_predictions,), dtype=tf.int32)
+    lm_ids = tf.keras.Input(shape=(num_token_predictions,), dtype=tf.int32)
+    inputs = {
+        'input_word_ids': word_ids,
+        'input_mask': mask,
+        'input_type_ids': type_ids,
+        'masked_lm_positions': lm_positions,
+        'masked_lm_ids': lm_ids
+    }
+
+    # Invoke the trainer model on the inputs. This causes the layer to be built.
+    outputs = teams_trainer_model(inputs)
+    lm_outs = outputs['lm_outputs']
+    disc_rtd_logits = outputs['disc_rtd_logits']
+    disc_rtd_label = outputs['disc_rtd_label']
+    disc_mws_logits = outputs['disc_mws_logits']
+    disc_mws_label = outputs['disc_mws_label']
+
+    # Validate that the outputs are of the expected shape.
+    expected_lm_shape = [None, num_token_predictions, vocab_size]
+    expected_disc_rtd_logits_shape = [None, sequence_length]
+    expected_disc_rtd_label_shape = [None, sequence_length]
+    expected_disc_disc_mws_logits_shape = [
+        None, num_token_predictions, candidate_size
+    ]
+    expected_disc_disc_mws_label_shape = [None, num_token_predictions]
+    self.assertAllEqual(expected_lm_shape, lm_outs.shape.as_list())
+    self.assertAllEqual(expected_disc_rtd_logits_shape,
+                        disc_rtd_logits.shape.as_list())
+    self.assertAllEqual(expected_disc_rtd_label_shape,
+                        disc_rtd_label.shape.as_list())
+    self.assertAllEqual(expected_disc_disc_mws_logits_shape,
+                        disc_mws_logits.shape.as_list())
+    self.assertAllEqual(expected_disc_disc_mws_label_shape,
+                        disc_mws_label.shape.as_list())
+
+  def test_teams_trainer_tensor_call(self):
+    """Validate that the Keras object can be invoked."""
+    vocab_size = 100
+    test_generator_network = self._get_network(vocab_size)
+    test_discriminator_network = self._get_network(vocab_size)
+
+    # Create a TEAMS trainer with the created network.
+    teams_trainer_model = teams_pretrainer.TeamsPretrainer(
+        generator_network=test_generator_network,
+        discriminator_mws_network=test_discriminator_network,
+        num_discriminator_task_agnostic_layers=2,
+        vocab_size=vocab_size,
+        candidate_size=2)
+
+    # Create a set of 2-dimensional data tensors to feed into the model.
+    word_ids = tf.constant([[1, 1, 1], [2, 2, 2]], dtype=tf.int32)
+    mask = tf.constant([[1, 1, 1], [1, 0, 0]], dtype=tf.int32)
+    type_ids = tf.constant([[1, 1, 1], [2, 2, 2]], dtype=tf.int32)
+    lm_positions = tf.constant([[0, 1], [0, 2]], dtype=tf.int32)
+    lm_ids = tf.constant([[10, 20], [20, 30]], dtype=tf.int32)
+    inputs = {
+        'input_word_ids': word_ids,
+        'input_mask': mask,
+        'input_type_ids': type_ids,
+        'masked_lm_positions': lm_positions,
+        'masked_lm_ids': lm_ids
+    }
+
+    # Invoke the trainer model on the tensors. In Eager mode, this does the
+    # actual calculation. (We can't validate the outputs, since the network is
+    # too complex: this simply ensures we're not hitting runtime errors.)
+    _ = teams_trainer_model(inputs)
+
+  def test_serialize_deserialize(self):
+    """Validate that the TEAMS trainer can be serialized and deserialized."""
+    vocab_size = 100
+    test_generator_network = self._get_network(vocab_size)
+    test_discriminator_network = self._get_network(vocab_size)
+
+    # Create a TEAMS trainer with the created network. (Note that all the args
+    # are different, so we can catch any serialization mismatches.)
+    teams_trainer_model = teams_pretrainer.TeamsPretrainer(
+        generator_network=test_generator_network,
+        discriminator_mws_network=test_discriminator_network,
+        num_discriminator_task_agnostic_layers=2,
+        vocab_size=vocab_size,
+        candidate_size=2)
+
+    # Create another TEAMS trainer via serialization and deserialization.
+    config = teams_trainer_model.get_config()
+    new_teams_trainer_model = teams_pretrainer.TeamsPretrainer.from_config(
+        config)
+
+    # Validate that the config can be forced to JSON.
+    _ = new_teams_trainer_model.to_json()
+
+    # If the serialization was successful, the new config should match the old.
+    self.assertAllEqual(teams_trainer_model.get_config(),
+                        new_teams_trainer_model.get_config())
+
+
+if __name__ == '__main__':
+  tf.test.main()

official/nlp/projects/teams/teams_task.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""TEAMS pretraining task (Joint Masked LM, Replaced Token Detection and )."""
+
+import dataclasses
+import tensorflow as tf
+
+from official.core import base_task
+from official.core import config_definitions as cfg
+from official.core import task_factory
+from official.modeling import tf_utils
+from official.nlp.data import pretrain_dataloader
+from official.nlp.modeling import layers
+from official.nlp.projects.teams import teams
+from official.nlp.projects.teams import teams_pretrainer
+
+
+@dataclasses.dataclass
+class TeamsPretrainTaskConfig(cfg.TaskConfig):
+  """The model config."""
+  model: teams.TeamsPretrainerConfig = teams.TeamsPretrainerConfig()
+  train_data: cfg.DataConfig = cfg.DataConfig()
+  validation_data: cfg.DataConfig = cfg.DataConfig()
+
+
+def _get_generator_hidden_layers(discriminator_network, num_hidden_layers,
+                                 num_shared_layers):
+  if num_shared_layers <= 0:
+    num_shared_layers = 0
+    hidden_layers = []
+  else:
+    hidden_layers = discriminator_network.hidden_layers[:num_shared_layers]
+  for _ in range(num_shared_layers, num_hidden_layers):
+    hidden_layers.append(layers.Transformer)
+  return hidden_layers
+
+
+def _build_pretrainer(
+    config: teams.TeamsPretrainerConfig) -> teams_pretrainer.TeamsPretrainer:
+  """Instantiates TeamsPretrainer from the config."""
+  generator_encoder_cfg = config.generator
+  discriminator_encoder_cfg = config.discriminator
+  discriminator_network = teams.get_encoder(discriminator_encoder_cfg)
+  # Copy discriminator's embeddings to generator for easier model serialization.
+  hidden_layers = _get_generator_hidden_layers(
+      discriminator_network, generator_encoder_cfg.num_layers,
+      config.num_shared_generator_hidden_layers)
+  if config.tie_embeddings:
+    generator_network = teams.get_encoder(
+        generator_encoder_cfg,
+        embedding_network=discriminator_network.embedding_network,
+        hidden_layers=hidden_layers)
+  else:
+    generator_network = teams.get_encoder(
+        generator_encoder_cfg, hidden_layers=hidden_layers)
+
+  return teams_pretrainer.TeamsPretrainer(
+      generator_network=generator_network,
+      discriminator_mws_network=discriminator_network,
+      num_discriminator_task_agnostic_layers=config
+      .num_discriminator_task_agnostic_layers,
+      vocab_size=generator_encoder_cfg.vocab_size,
+      candidate_size=config.candidate_size,
+      mlm_activation=tf_utils.get_activation(
+          generator_encoder_cfg.hidden_activation),
+      mlm_initializer=tf.keras.initializers.TruncatedNormal(
+          stddev=generator_encoder_cfg.initializer_range))
+
+
+@task_factory.register_task_cls(TeamsPretrainTaskConfig)
+class TeamsPretrainTask(base_task.Task):
+  """TEAMS Pretrain Task (Masked LM + RTD + MWS)."""
+
+  def build_model(self):
+    return _build_pretrainer(self.task_config.model)
+
+  def build_losses(self,
+                   labels,
+                   model_outputs,
+                   metrics,
+                   aux_losses=None) -> tf.Tensor:
+    with tf.name_scope('TeamsPretrainTask/losses'):
+      metrics = dict([(metric.name, metric) for metric in metrics])
+
+      # Generator MLM loss.
+      lm_prediction_losses = tf.keras.losses.sparse_categorical_crossentropy(
+          labels['masked_lm_ids'],
+          tf.cast(model_outputs['lm_outputs'], tf.float32),
+          from_logits=True)
+      lm_label_weights = labels['masked_lm_weights']
+      lm_numerator_loss = tf.reduce_sum(lm_prediction_losses * lm_label_weights)
+      lm_denominator_loss = tf.reduce_sum(lm_label_weights)
+      mlm_loss = tf.math.divide_no_nan(lm_numerator_loss, lm_denominator_loss)
+      metrics['masked_lm_loss'].update_state(mlm_loss)
+      weight = self.task_config.model.generator_loss_weight
+      total_loss = weight * mlm_loss
+
+      # Discriminator RTD loss.
+      rtd_logits = model_outputs['disc_rtd_logits']
+      rtd_labels = tf.cast(model_outputs['disc_rtd_label'], tf.float32)
+      input_mask = tf.cast(labels['input_mask'], tf.float32)
+      rtd_ind_loss = tf.nn.sigmoid_cross_entropy_with_logits(
+          logits=rtd_logits, labels=rtd_labels)
+      rtd_numerator = tf.reduce_sum(input_mask * rtd_ind_loss)
+      rtd_denominator = tf.reduce_sum(input_mask)
+      rtd_loss = tf.math.divide_no_nan(rtd_numerator, rtd_denominator)
+      metrics['replaced_token_detection_loss'].update_state(rtd_loss)
+      weight = self.task_config.model.discriminator_rtd_loss_weight
+      total_loss = total_loss + weight * rtd_loss
+
+      # Discriminator MWS loss.
+      mws_logits = model_outputs['disc_mws_logits']
+      mws_labels = model_outputs['disc_mws_label']
+      mws_loss = tf.keras.losses.sparse_categorical_crossentropy(
+          mws_labels, mws_logits, from_logits=True)
+      mws_numerator_loss = tf.reduce_sum(mws_loss * lm_label_weights)
+      mws_denominator_loss = tf.reduce_sum(lm_label_weights)
+      mws_loss = tf.math.divide_no_nan(mws_numerator_loss, mws_denominator_loss)
+      metrics['multiword_selection_loss'].update_state(mws_loss)
+      weight = self.task_config.model.discriminator_mws_loss_weight
+      total_loss = total_loss + weight * mws_loss
+
+      if aux_losses:
+        total_loss += tf.add_n(aux_losses)
+
+      metrics['total_loss'].update_state(total_loss)
+      return total_loss
+
+  def build_inputs(self, params, input_context=None):
+    """Returns tf.data.Dataset for pretraining."""
+    if params.input_path == 'dummy':
+
+      def dummy_data(_):
+        dummy_ids = tf.zeros((1, params.seq_length), dtype=tf.int32)
+        dummy_lm = tf.zeros((1, params.max_predictions_per_seq), dtype=tf.int32)
+        return dict(
+            input_word_ids=dummy_ids,
+            input_mask=dummy_ids,
+            input_type_ids=dummy_ids,
+            masked_lm_positions=dummy_lm,
+            masked_lm_ids=dummy_lm,
+            masked_lm_weights=tf.cast(dummy_lm, dtype=tf.float32))
+
+      dataset = tf.data.Dataset.range(1)
+      dataset = dataset.repeat()
+      dataset = dataset.map(
+          dummy_data, num_parallel_calls=tf.data.experimental.AUTOTUNE)
+      return dataset
+
+    return pretrain_dataloader.BertPretrainDataLoader(params).load(
+        input_context)
+
+  def build_metrics(self, training=None):
+    del training
+    metrics = [
+        tf.keras.metrics.SparseCategoricalAccuracy(name='masked_lm_accuracy'),
+        tf.keras.metrics.Mean(name='masked_lm_loss'),
+        tf.keras.metrics.SparseCategoricalAccuracy(
+            name='replaced_token_detection_accuracy'),
+        tf.keras.metrics.Mean(name='replaced_token_detection_loss'),
+        tf.keras.metrics.SparseCategoricalAccuracy(
+            name='multiword_selection_accuracy'),
+        tf.keras.metrics.Mean(name='multiword_selection_loss'),
+        tf.keras.metrics.Mean(name='total_loss'),
+    ]
+    return metrics
+
+  def process_metrics(self, metrics, labels, model_outputs):
+    with tf.name_scope('TeamsPretrainTask/process_metrics'):
+      metrics = dict([(metric.name, metric) for metric in metrics])
+      if 'masked_lm_accuracy' in metrics:
+        metrics['masked_lm_accuracy'].update_state(labels['masked_lm_ids'],
+                                                   model_outputs['lm_outputs'],
+                                                   labels['masked_lm_weights'])
+
+      if 'replaced_token_detection_accuracy' in metrics:
+        rtd_logits_expanded = tf.expand_dims(model_outputs['disc_rtd_logits'],
+                                             -1)
+        rtd_full_logits = tf.concat(
+            [-1.0 * rtd_logits_expanded, rtd_logits_expanded], -1)
+        metrics['replaced_token_detection_accuracy'].update_state(
+            model_outputs['disc_rtd_label'], rtd_full_logits,
+            labels['input_mask'])
+
+      if 'multiword_selection_accuracy' in metrics:
+        metrics['multiword_selection_accuracy'].update_state(
+            model_outputs['disc_mws_label'], model_outputs['disc_mws_logits'],
+            labels['masked_lm_weights'])
+
+  def train_step(self, inputs, model: tf.keras.Model,
+                 optimizer: tf.keras.optimizers.Optimizer, metrics):
+    """Does forward and backward.
+
+    Args:
+      inputs: a dictionary of input tensors.
+      model: the model, forward pass definition.
+      optimizer: the optimizer for this training step.
+      metrics: a nested structure of metrics objects.
+
+    Returns:
+      A dictionary of logs.
+    """
+    with tf.GradientTape() as tape:
+      outputs = model(inputs, training=True)
+      # Computes per-replica loss.
+      loss = self.build_losses(
+          labels=inputs,
+          model_outputs=outputs,
+          metrics=metrics,
+          aux_losses=model.losses)
+      # Scales loss as the default gradients allreduce performs sum inside the
+      # optimizer.
+      scaled_loss = loss / tf.distribute.get_strategy().num_replicas_in_sync
+    tvars = model.trainable_variables
+    grads = tape.gradient(scaled_loss, tvars)
+    optimizer.apply_gradients(list(zip(grads, tvars)))
+    self.process_metrics(metrics, inputs, outputs)
+    return {self.loss: loss}
+
+  def validation_step(self, inputs, model: tf.keras.Model, metrics):
+    """Validatation step.
+
+    Args:
+      inputs: a dictionary of input tensors.
+      model: the keras.Model.
+      metrics: a nested structure of metrics objects.
+
+    Returns:
+      A dictionary of logs.
+    """
+    outputs = model(inputs, training=False)
+    loss = self.build_losses(
+        labels=inputs,
+        model_outputs=outputs,
+        metrics=metrics,
+        aux_losses=model.losses)
+    self.process_metrics(metrics, inputs, outputs)
+    return {self.loss: loss}

official/nlp/projects/teams/teams_task_test.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Tests for teams_task."""
+
+from absl.testing import parameterized
+import tensorflow as tf
+
+from official.nlp.configs import encoders
+from official.nlp.data import pretrain_dataloader
+from official.nlp.projects.teams import teams
+from official.nlp.projects.teams import teams_task
+
+
+class TeamsPretrainTaskTest(tf.test.TestCase, parameterized.TestCase):
+
+  @parameterized.parameters((1, 1), (0, 1), (0, 0), (1, 0))
+  def test_task(self, num_shared_hidden_layers,
+                num_task_agnostic_layers):
+    config = teams_task.TeamsPretrainTaskConfig(
+        model=teams.TeamsPretrainerConfig(
+            generator=encoders.BertEncoderConfig(
+                vocab_size=30522, num_layers=2),
+            discriminator=encoders.BertEncoderConfig(
+                vocab_size=30522, num_layers=2),
+            num_shared_generator_hidden_layers=num_shared_hidden_layers,
+            num_discriminator_task_agnostic_layers=num_task_agnostic_layers,
+        ),
+        train_data=pretrain_dataloader.BertPretrainDataConfig(
+            input_path="dummy",
+            max_predictions_per_seq=20,
+            seq_length=128,
+            global_batch_size=1))
+    task = teams_task.TeamsPretrainTask(config)
+    model = task.build_model()
+    metrics = task.build_metrics()
+    dataset = task.build_inputs(config.train_data)
+
+    iterator = iter(dataset)
+    optimizer = tf.keras.optimizers.SGD(lr=0.1)
+    task.train_step(next(iterator), model, optimizer, metrics=metrics)
+    task.validation_step(next(iterator), model, metrics=metrics)
+
+if __name__ == "__main__":
+  tf.test.main()

official/nlp/projects/triviaqa/inputs.py

@@ -48,15 +48,15 @@ def _flatten_dims(tensor: tf.Tensor,
     rank = tensor.shape.rank
     if rank is None:
       raise ValueError('Static rank of `tensor` must be known.')
-    if first_dim < 0:
+    if first_dim < 0:  # pytype: disable=unsupported-operands
       first_dim += rank
-    if first_dim < 0 or first_dim >= rank:
+    if first_dim < 0 or first_dim >= rank:  # pytype: disable=unsupported-operands
       raise ValueError('`first_dim` out of bounds for `tensor` rank.')
-    if last_dim < 0:
+    if last_dim < 0:  # pytype: disable=unsupported-operands
       last_dim += rank
-    if last_dim < 0 or last_dim >= rank:
+    if last_dim < 0 or last_dim >= rank:  # pytype: disable=unsupported-operands
       raise ValueError('`last_dim` out of bounds for `tensor` rank.')
-    if first_dim > last_dim:
+    if first_dim > last_dim:  # pytype: disable=unsupported-operands
       raise ValueError('`first_dim` must not be larger than `last_dim`.')
 
     # Try to calculate static flattened dim size if all input sizes to flatten

official/nlp/serving/export_savedmodel_util.py

@@ -13,12 +13,19 @@
 # limitations under the License.
 
 """Common library to export a SavedModel from the export module."""
+import os
+import time
 from typing import Dict, List, Optional, Text, Union
 
+from absl import logging
 import tensorflow as tf
+
 from official.core import export_base
 
 
+MAX_DIRECTORY_CREATION_ATTEMPTS = 10
+
+
 def export(export_module: export_base.ExportModule,
            function_keys: Union[List[Text], Dict[Text, Text]],
            export_savedmodel_dir: Text,
@@ -39,7 +46,39 @@ def export(export_module: export_base.ExportModule,
     The savedmodel directory path.
   """
   save_options = tf.saved_model.SaveOptions(function_aliases={
-      "tpu_candidate": export_module.serve,
+      'tpu_candidate': export_module.serve,
   })
   return export_base.export(export_module, function_keys, export_savedmodel_dir,
                             checkpoint_path, timestamped, save_options)
+
+
+def get_timestamped_export_dir(export_dir_base):
+  """Builds a path to a new subdirectory within the base directory.
+
+  Args:
+    export_dir_base: A string containing a directory to write the exported graph
+      and checkpoints.
+
+  Returns:
+    The full path of the new subdirectory (which is not actually created yet).
+
+  Raises:
+    RuntimeError: if repeated attempts fail to obtain a unique timestamped
+      directory name.
+  """
+  attempts = 0
+  while attempts < MAX_DIRECTORY_CREATION_ATTEMPTS:
+    timestamp = int(time.time())
+
+    result_dir = os.path.join(export_dir_base, str(timestamp))
+    if not tf.io.gfile.exists(result_dir):
+      # Collisions are still possible (though extremely unlikely): this
+      # directory is not actually created yet, but it will be almost
+      # instantly on return from this function.
+      return result_dir
+    time.sleep(1)
+    attempts += 1
+    logging.warning('Directory %s already exists; retrying (attempt %s/%s)',
+                    str(result_dir), attempts, MAX_DIRECTORY_CREATION_ATTEMPTS)
+  raise RuntimeError('Failed to obtain a unique export directory name after '
+                     f'{MAX_DIRECTORY_CREATION_ATTEMPTS} attempts.')

official/nlp/serving/serving_modules.py

@@ -80,11 +80,10 @@ class SentencePrediction(export_base.ExportModule):
           lower_case=params.lower_case,
           preprocessing_hub_module_url=params.preprocessing_hub_module_url)
 
-  @tf.function
-  def serve(self,
+  def _serve_tokenized_input(self,
                              input_word_ids,
                              input_mask=None,
-            input_type_ids=None) -> Dict[str, tf.Tensor]:
+                             input_type_ids=None) -> tf.Tensor:
     if input_type_ids is None:
       # Requires CLS token is the first token of inputs.
       input_type_ids = tf.zeros_like(input_word_ids)
@@ -97,7 +96,26 @@ class SentencePrediction(export_base.ExportModule):
         input_word_ids=input_word_ids,
         input_mask=input_mask,
         input_type_ids=input_type_ids)
-    return dict(outputs=self.inference_step(inputs))
+    return self.inference_step(inputs)
+
+  @tf.function
+  def serve(self,
+            input_word_ids,
+            input_mask=None,
+            input_type_ids=None) -> Dict[str, tf.Tensor]:
+    return dict(
+        outputs=self._serve_tokenized_input(input_word_ids, input_mask,
+                                            input_type_ids))
+
+  @tf.function
+  def serve_probability(self,
+                        input_word_ids,
+                        input_mask=None,
+                        input_type_ids=None) -> Dict[str, tf.Tensor]:
+    return dict(
+        outputs=tf.nn.softmax(
+            self._serve_tokenized_input(input_word_ids, input_mask,
+                                        input_type_ids)))
 
   @tf.function
   def serve_examples(self, inputs) -> Dict[str, tf.Tensor]:

official/nlp/tasks/sentence_prediction.py

@@ -13,10 +13,10 @@
 # limitations under the License.
 
 """Sentence prediction (classification) task."""
+import dataclasses
 from typing import List, Union, Optional
 
 from absl import logging
-import dataclasses
 import numpy as np
 import orbit
 from scipy import stats
@@ -140,14 +140,25 @@ class SentencePredictionTask(base_task.Task):
     del training
     if self.task_config.model.num_classes == 1:
       metrics = [tf.keras.metrics.MeanSquaredError()]
+    elif self.task_config.model.num_classes == 2:
+      metrics = [
+          tf.keras.metrics.SparseCategoricalAccuracy(name='cls_accuracy'),
+          tf.keras.metrics.AUC(name='auc', curve='PR'),
+      ]
     else:
       metrics = [
-          tf.keras.metrics.SparseCategoricalAccuracy(name='cls_accuracy')
+          tf.keras.metrics.SparseCategoricalAccuracy(name='cls_accuracy'),
       ]
     return metrics
 
   def process_metrics(self, metrics, labels, model_outputs):
     for metric in metrics:
+      if metric.name == 'auc':
+        # Convert the logit to probability and extract the probability of True..
+        metric.update_state(
+            labels[self.label_field],
+            tf.expand_dims(tf.nn.softmax(model_outputs)[:, 1], axis=1))
+      if metric.name == 'cls_accuracy':
         metric.update_state(labels[self.label_field], model_outputs)
 
   def process_compiled_metrics(self, compiled_metrics, labels, model_outputs):

official/nlp/tasks/translation.py

@@ -13,11 +13,11 @@
 # limitations under the License.
 
 """Defines the translation task."""
+import dataclasses
 import os
 from typing import Optional
 
 from absl import logging
-import dataclasses
 import sacrebleu
 import tensorflow as tf
 import tensorflow_text as tftxt

official/nlp/tasks/translation_test.py

@@ -85,7 +85,8 @@ class TranslationTaskTest(tf.test.TestCase):
   def test_task(self):
     config = translation.TranslationConfig(
         model=translation.ModelConfig(
-            encoder=translation.EncDecoder(), decoder=translation.EncDecoder()),
+            encoder=translation.EncDecoder(num_layers=1),
+            decoder=translation.EncDecoder(num_layers=1)),
         train_data=wmt_dataloader.WMTDataConfig(
             input_path=self._record_input_path,
             src_lang="en", tgt_lang="reverse_en",
@@ -102,7 +103,8 @@ class TranslationTaskTest(tf.test.TestCase):
   def test_no_sentencepiece_path(self):
     config = translation.TranslationConfig(
         model=translation.ModelConfig(
-            encoder=translation.EncDecoder(), decoder=translation.EncDecoder()),
+            encoder=translation.EncDecoder(num_layers=1),
+            decoder=translation.EncDecoder(num_layers=1)),
         train_data=wmt_dataloader.WMTDataConfig(
             input_path=self._record_input_path,
             src_lang="en", tgt_lang="reverse_en",
@@ -122,7 +124,8 @@ class TranslationTaskTest(tf.test.TestCase):
         sentencepeice_model_prefix)
     config = translation.TranslationConfig(
         model=translation.ModelConfig(
-            encoder=translation.EncDecoder(), decoder=translation.EncDecoder()),
+            encoder=translation.EncDecoder(num_layers=1),
+            decoder=translation.EncDecoder(num_layers=1)),
         train_data=wmt_dataloader.WMTDataConfig(
             input_path=self._record_input_path,
             src_lang="en", tgt_lang="reverse_en",
@@ -137,7 +140,8 @@ class TranslationTaskTest(tf.test.TestCase):
   def test_evaluation(self):
     config = translation.TranslationConfig(
         model=translation.ModelConfig(
-            encoder=translation.EncDecoder(), decoder=translation.EncDecoder(),
+            encoder=translation.EncDecoder(num_layers=1),
+            decoder=translation.EncDecoder(num_layers=1),
             padded_decode=False,
             decode_max_length=64),
         validation_data=wmt_dataloader.WMTDataConfig(

official/nlp/train.py

@@ -27,9 +27,15 @@ from official.core import task_factory
 from official.core import train_lib
 from official.core import train_utils
 from official.modeling import performance
+from official.nlp import continuous_finetune_lib
 
 FLAGS = flags.FLAGS
 
+flags.DEFINE_integer(
+    'pretrain_steps',
+    default=None,
+    help='The number of total training steps for the pretraining job.')
+
 
 def main(_):
   gin.parse_config_files_and_bindings(FLAGS.gin_file, FLAGS.gin_params)
@@ -40,12 +46,18 @@ def main(_):
     # may race against the train job for writing the same file.
     train_utils.serialize_config(params, model_dir)
 
+  if FLAGS.mode == 'continuous_train_and_eval':
+    continuous_finetune_lib.run_continuous_finetune(
+        FLAGS.mode, params, model_dir, pretrain_steps=FLAGS.pretrain_steps)
+
+  else:
     # Sets mixed_precision policy. Using 'mixed_float16' or 'mixed_bfloat16'
-  # can have significant impact on model speeds by utilizing float16 in case of
-  # GPUs, and bfloat16 in the case of TPUs. loss_scale takes effect only when
-  # dtype is float16
+    # can have significant impact on model speeds by utilizing float16 in case
+    # of GPUs, and bfloat16 in the case of TPUs. loss_scale takes effect only
+    # when dtype is float16
     if params.runtime.mixed_precision_dtype:
-    performance.set_mixed_precision_policy(params.runtime.mixed_precision_dtype)
+      performance.set_mixed_precision_policy(
+          params.runtime.mixed_precision_dtype)
     distribution_strategy = distribute_utils.get_distribution_strategy(
         distribution_strategy=params.runtime.distribution_strategy,
         all_reduce_alg=params.runtime.all_reduce_alg,

official/pip_package/setup.py

@@ -81,6 +81,7 @@ setup(
         'official.pip_package*',
         'official.benchmark*',
         'official.colab*',
+        'official.recommendation.ranking.data.preprocessing*',
     ]),
     exclude_package_data={
         '': ['*_test.py',],

official/projects/basnet/README.md

+# BASNet: Boundary-Aware Salient Object Detection
+
+This repository is the unofficial implementation of the following paper. Please
+see the paper
+[BASNet: Boundary-Aware Salient Object Detection](https://openaccess.thecvf.com/content_CVPR_2019/html/Qin_BASNet_Boundary-Aware_Salient_Object_Detection_CVPR_2019_paper.html)
+for more details.
+
+## Requirements
+[![TensorFlow 2.4](https://img.shields.io/badge/TensorFlow-2.4-FF6F00?logo=tensorflow)](https://github.com/tensorflow/tensorflow/releases/tag/v2.4.0)
+[![Python 3.7](https://img.shields.io/badge/Python-3.7-3776AB)](https://www.python.org/downloads/release/python-379/)
+
+## Train
+```shell
+$ python3 train.py \
+  --experiment=basnet_duts \
+  --mode=train \
+  --model_dir=$MODEL_DIR \
+  --config_file=./configs/experiments/basnet_dut_gpu.yaml
+```
+
+## Test
+```shell
+$ python3 train.py \
+  --experiment=basnet_duts \
+  --mode=eval \
+  --model_dir=$MODEL_DIR \
+  --config_file=./configs/experiments/basnet_dut_gpu.yaml
+  --params_override='runtime.num_gpus=1, runtime.distribution_strategy=one_device, task.model.input_size=[256, 256, 3]'
+```
+
+## Results
+Dataset    | maxF<sub>β</sub> | relaxF<sub>β</sub>   | MAE
+:--------- | :--------------- | :------------------- | -------:
+DUTS-TE    | 0.865            | 0.793                | 0.046
+

official/projects/basnet/configs/basnet.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""BASNet configuration definition."""
+import dataclasses
+import os
+from typing import List, Optional, Union
+
+from official.core import exp_factory
+from official.modeling import hyperparams
+from official.modeling import optimization
+from official.modeling.hyperparams import config_definitions as cfg
+from official.vision.beta.configs import common
+
+
+@dataclasses.dataclass
+class DataConfig(cfg.DataConfig):
+  """Input config for training."""
+  output_size: List[int] = dataclasses.field(default_factory=list)
+  # If crop_size is specified, image will be resized first to
+  # output_size, then crop of size crop_size will be cropped.
+  crop_size: List[int] = dataclasses.field(default_factory=list)
+  input_path: str = ''
+  global_batch_size: int = 0
+  is_training: bool = True
+  dtype: str = 'float32'
+  shuffle_buffer_size: int = 1000
+  cycle_length: int = 10
+  resize_eval_groundtruth: bool = True
+  groundtruth_padded_size: List[int] = dataclasses.field(default_factory=list)
+  aug_rand_hflip: bool = True
+  file_type: str = 'tfrecord'
+
+
+@dataclasses.dataclass
+class BASNetModel(hyperparams.Config):
+  """BASNet model config."""
+  input_size: List[int] = dataclasses.field(default_factory=list)
+  use_bias: bool = False
+  norm_activation: common.NormActivation = common.NormActivation()
+
+
+@dataclasses.dataclass
+class Losses(hyperparams.Config):
+  label_smoothing: float = 0.1
+  ignore_label: int = 0  # will be treated as background
+  l2_weight_decay: float = 0.0
+  use_groundtruth_dimension: bool = True
+
+
+@dataclasses.dataclass
+class BASNetTask(cfg.TaskConfig):
+  """The model config."""
+  model: BASNetModel = BASNetModel()
+  train_data: DataConfig = DataConfig(is_training=True)
+  validation_data: DataConfig = DataConfig(is_training=False)
+  losses: Losses = Losses()
+  gradient_clip_norm: float = 0.0
+  init_checkpoint: Optional[str] = None
+  init_checkpoint_modules: Union[
+      str, List[str]] = 'backbone'  # all, backbone, and/or decoder
+
+
+@exp_factory.register_config_factory('basnet')
+def basnet() -> cfg.ExperimentConfig:
+  """BASNet general."""
+  return cfg.ExperimentConfig(
+      task=BASNetModel(),
+      trainer=cfg.TrainerConfig(),
+      restrictions=[
+          'task.train_data.is_training != None',
+          'task.validation_data.is_training != None'
+      ])
+
+
+# DUTS Dataset
+DUTS_TRAIN_EXAMPLES = 10553
+DUTS_VAL_EXAMPLES = 5019
+DUTS_INPUT_PATH_BASE_TR = 'DUTS_DATASET'
+DUTS_INPUT_PATH_BASE_VAL = 'DUTS_DATASET'
+
+
+@exp_factory.register_config_factory('basnet_duts')
+def basnet_duts() -> cfg.ExperimentConfig:
+  """Image segmentation on duts with basnet."""
+  train_batch_size = 64
+  eval_batch_size = 16
+  steps_per_epoch = DUTS_TRAIN_EXAMPLES // train_batch_size
+  config = cfg.ExperimentConfig(
+      task=BASNetTask(
+          model=BASNetModel(
+              input_size=[None, None, 3],
+              use_bias=True,
+              norm_activation=common.NormActivation(
+                  activation='relu',
+                  norm_momentum=0.99,
+                  norm_epsilon=1e-3,
+                  use_sync_bn=True)),
+          losses=Losses(l2_weight_decay=0),
+          train_data=DataConfig(
+              input_path=os.path.join(DUTS_INPUT_PATH_BASE_TR,
+                                      'tf_record_train'),
+              file_type='tfrecord',
+              crop_size=[224, 224],
+              output_size=[256, 256],
+              is_training=True,
+              global_batch_size=train_batch_size,
+          ),
+          validation_data=DataConfig(
+              input_path=os.path.join(DUTS_INPUT_PATH_BASE_VAL,
+                                      'tf_record_test'),
+              file_type='tfrecord',
+              output_size=[256, 256],
+              is_training=False,
+              global_batch_size=eval_batch_size,
+          ),
+          init_checkpoint='gs://cloud-basnet-checkpoints/basnet_encoder_imagenet/ckpt-340306',
+          init_checkpoint_modules='backbone'
+      ),
+      trainer=cfg.TrainerConfig(
+          steps_per_loop=steps_per_epoch,
+          summary_interval=steps_per_epoch,
+          checkpoint_interval=steps_per_epoch,
+          train_steps=300 * steps_per_epoch,
+          validation_steps=DUTS_VAL_EXAMPLES // eval_batch_size,
+          validation_interval=steps_per_epoch,
+          optimizer_config=optimization.OptimizationConfig({
+              'optimizer': {
+                  'type': 'adam',
+                  'adam': {
+                      'beta_1': 0.9,
+                      'beta_2': 0.999,
+                      'epsilon': 1e-8,
+                  }
+              },
+              'learning_rate': {
+                  'type': 'constant',
+                  'constant': {'learning_rate': 0.001}
+              }
+          })),
+      restrictions=[
+          'task.train_data.is_training != None',
+          'task.validation_data.is_training != None'
+      ])
+  return config

official/projects/basnet/configs/basnet_test.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Tests for basnet configs."""
+
+# pylint: disable=unused-import
+from absl.testing import parameterized
+import tensorflow as tf
+
+from official.core import exp_factory
+from official.modeling.hyperparams import config_definitions as cfg
+from official.projects.basnet.configs import basnet as exp_cfg
+
+
+class BASNetConfigTest(tf.test.TestCase, parameterized.TestCase):
+
+  @parameterized.parameters(('basnet_duts',))
+  def test_basnet_configs(self, config_name):
+    config = exp_factory.get_exp_config(config_name)
+    self.assertIsInstance(config, cfg.ExperimentConfig)
+    self.assertIsInstance(config.task, exp_cfg.BASNetTask)
+    self.assertIsInstance(config.task.model,
+                          exp_cfg.BASNetModel)
+    self.assertIsInstance(config.task.train_data, exp_cfg.DataConfig)
+    config.task.train_data.is_training = None
+    with self.assertRaises(KeyError):
+      config.validate()
+
+
+if __name__ == '__main__':
+  tf.test.main()

official/projects/basnet/configs/experiments/basnet_dut_gpu.yaml

+runtime:
+  distribution_strategy: 'mirrored'
+  mixed_precision_dtype: 'float32'
+  num_gpus: 8
+task:
+  train_data:
+    dtype: 'float32'
+  validation_data:
+    resize_eval_groundtruth: true
+    dtype: 'float32'

official/projects/basnet/evaluation/metrics.py

+# Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Evaluation metrics for BASNet.
+
+The MAE and maxFscore implementations are a modified version of
+https://github.com/xuebinqin/Binary-Segmentation-Evaluation-Tool
+
+"""
+import numpy as np
+import scipy.signal
+
+
+class MAE:
+  """Mean Absolute Error(MAE) metric for basnet."""
+
+  def __init__(self):
+    """Constructs MAE metric class."""
+    self.reset_states()
+
+  @property
+  def name(self):
+    return 'MAE'
+
+  def reset_states(self):
+    """Resets internal states for a fresh run."""
+    self._predictions = []
+    self._groundtruths = []
+
+  def result(self):
+    """Evaluates segmentation results, and reset_states."""
+    metric_result = self.evaluate()
+    # Cleans up the internal variables in order for a fresh eval next time.
+    self.reset_states()
+    return metric_result
+
+  def evaluate(self):
+    """Evaluates with masks from all images.
+
+    Returns:
+      average_mae: average MAE with float numpy.
+    """
+    mae_total = 0.0
+
+    for (true, pred) in zip(self._groundtruths, self._predictions):
+      # Computes MAE
+      mae = self._compute_mae(true, pred)
+      mae_total += mae
+
+    average_mae = mae_total / len(self._groundtruths)
+
+    return average_mae
+
+  def _mask_normalize(self, mask):
+    return mask/(np.amax(mask)+1e-8)
+
+  def _compute_mae(self, true, pred):
+    h, w = true.shape[0], true.shape[1]
+    mask1 = self._mask_normalize(true)
+    mask2 = self._mask_normalize(pred)
+    sum_error = np.sum(np.absolute((mask1.astype(float) - mask2.astype(float))))
+    mae_error = sum_error/(float(h)*float(w)+1e-8)
+
+    return mae_error
+
+  def _convert_to_numpy(self, groundtruths, predictions):
+    """Converts tesnors to numpy arrays."""
+    numpy_groundtruths = groundtruths.numpy()
+    numpy_predictions = predictions.numpy()
+
+    return numpy_groundtruths, numpy_predictions
+
+  def update_state(self, groundtruths, predictions):
+    """Update segmentation results and groundtruth data.
+
+    Args:
+      groundtruths : Tuple of single Tensor [batch, width, height, 1],
+                     groundtruth masks. range [0, 1]
+      predictions  : Tuple of single Tensor [batch, width, height, 1],
+                     predicted masks. range [0, 1]
+    """
+    groundtruths, predictions = self._convert_to_numpy(groundtruths[0],
+                                                       predictions[0])
+    for (true, pred) in zip(groundtruths, predictions):
+      self._groundtruths.append(true)
+      self._predictions.append(pred)
+
+
+class MaxFscore:
+  """Maximum F-score metric for basnet."""
+
+  def __init__(self):
+    """Constructs BASNet evaluation class."""
+    self.reset_states()
+
+  @property
+  def name(self):
+    return 'MaxFScore'
+
+  def reset_states(self):
+    """Resets internal states for a fresh run."""
+    self._predictions = []
+    self._groundtruths = []
+
+  def result(self):
+    """Evaluates segmentation results, and reset_states."""
+    metric_result = self.evaluate()
+    # Cleans up the internal variables in order for a fresh eval next time.
+    self.reset_states()
+    return metric_result
+
+  def evaluate(self):
+    """Evaluates with masks from all images.
+
+    Returns:
+      f_max: maximum F-score value.
+    """
+
+    mybins = np.arange(0, 256)
+    beta = 0.3
+    precisions = np.zeros((len(self._groundtruths), len(mybins)-1))
+    recalls = np.zeros((len(self._groundtruths), len(mybins)-1))
+
+    for i, (true, pred) in enumerate(zip(self._groundtruths,
+                                         self._predictions)):
+      # Compute F-score
+      true = self._mask_normalize(true) * 255.0
+      pred = self._mask_normalize(pred) * 255.0
+      pre, rec = self._compute_pre_rec(true, pred, mybins=np.arange(0, 256))
+
+      precisions[i, :] = pre
+      recalls[i, :] = rec
+
+    precisions = np.sum(precisions, 0) / (len(self._groundtruths) + 1e-8)
+    recalls = np.sum(recalls, 0) / (len(self._groundtruths) + 1e-8)
+    f = (1 + beta) * precisions * recalls / (beta * precisions + recalls + 1e-8)
+    f_max = np.max(f)
+    f_max = f_max.astype(np.float32)
+
+    return f_max
+
+  def _mask_normalize(self, mask):
+    return mask / (np.amax(mask) + 1e-8)
+
+  def _compute_pre_rec(self, true, pred, mybins=np.arange(0, 256)):
+    """Computes relaxed precision and recall."""
+    # pixel number of ground truth foreground regions
+    gt_num = true[true > 128].size
+
+    # mask predicted pixel values in the ground truth foreground region
+    pp = pred[true > 128]
+    # mask predicted pixel values in the ground truth bacground region
+    nn = pred[true <= 128]
+
+    pp_hist, _ = np.histogram(pp, bins=mybins)
+    nn_hist, _ = np.histogram(nn, bins=mybins)
+
+    pp_hist_flip = np.flipud(pp_hist)
+    nn_hist_flip = np.flipud(nn_hist)
+
+    pp_hist_flip_cum = np.cumsum(pp_hist_flip)
+    nn_hist_flip_cum = np.cumsum(nn_hist_flip)
+
+    precision = pp_hist_flip_cum / (pp_hist_flip_cum + nn_hist_flip_cum + 1e-8
+                                   )  # TP/(TP+FP)
+    recall = pp_hist_flip_cum / (gt_num + 1e-8)  # TP/(TP+FN)
+
+    precision[np.isnan(precision)] = 0.0
+    recall[np.isnan(recall)] = 0.0
+
+    pre_len = len(precision)
+    rec_len = len(recall)
+
+    return np.reshape(precision, (pre_len)), np.reshape(recall, (rec_len))
+
+  def _convert_to_numpy(self, groundtruths, predictions):
+    """Converts tesnors to numpy arrays."""
+    numpy_groundtruths = groundtruths.numpy()
+    numpy_predictions = predictions.numpy()
+
+    return numpy_groundtruths, numpy_predictions
+
+  def update_state(self, groundtruths, predictions):
+    """Update segmentation results and groundtruth data.
+
+    Args:
+      groundtruths : Tuple of single Tensor [batch, width, height, 1],
+                     groundtruth masks. range [0, 1]
+      predictions  : Tuple of signle Tensor [batch, width, height, 1],
+                     predicted masks. range [0, 1]
+    """
+    groundtruths, predictions = self._convert_to_numpy(groundtruths[0],
+                                                       predictions[0])
+    for (true, pred) in zip(groundtruths, predictions):
+      self._groundtruths.append(true)
+      self._predictions.append(pred)
+
+
+class RelaxedFscore:
+  """Relaxed F-score metric for basnet."""
+
+  def __init__(self):
+    """Constructs BASNet evaluation class."""
+    self.reset_states()
+
+  @property
+  def name(self):
+    return 'RelaxFScore'
+
+  def reset_states(self):
+    """Resets internal states for a fresh run."""
+    self._predictions = []
+    self._groundtruths = []
+
+  def result(self):
+    """Evaluates segmentation results, and reset_states."""
+    metric_result = self.evaluate()
+    # Cleans up the internal variables in order for a fresh eval next time.
+    self.reset_states()
+    return metric_result
+
+  def evaluate(self):
+    """Evaluates with masks from all images.
+
+    Returns:
+      relax_f: relaxed F-score value.
+    """
+
+    beta = 0.3
+    rho = 3
+    relax_fs = np.zeros(len(self._groundtruths))
+
+    erode_kernel = np.ones((3, 3))
+
+    for i, (true,
+            pred) in enumerate(zip(self._groundtruths, self._predictions)):
+      true = self._mask_normalize(true)
+      pred = self._mask_normalize(pred)
+
+      true = np.squeeze(true, axis=-1)
+      pred = np.squeeze(pred, axis=-1)
+      # binary saliency mask (S_bw), threshold 0.5
+      pred[pred >= 0.5] = 1
+      pred[pred < 0.5] = 0
+      # compute eroded binary mask (S_erd) of S_bw
+      pred_erd = self._compute_erosion(pred, erode_kernel)
+
+      pred_xor = np.logical_xor(pred_erd, pred)
+      # convert True/False to 1/0
+      pred_xor = pred_xor * 1
+
+      # same method for ground truth
+      true[true >= 0.5] = 1
+      true[true < 0.5] = 0
+      true_erd = self._compute_erosion(true, erode_kernel)
+      true_xor = np.logical_xor(true_erd, true)
+      true_xor = true_xor * 1
+
+      pre, rec = self._compute_relax_pre_rec(true_xor, pred_xor, rho)
+      relax_fs[i] = (1 + beta) * pre * rec / (beta * pre + rec + 1e-8)
+
+    relax_f = np.sum(relax_fs, 0) / (len(self._groundtruths) + 1e-8)
+    relax_f = relax_f.astype(np.float32)
+
+    return relax_f
+
+  def _mask_normalize(self, mask):
+    return mask/(np.amax(mask)+1e-8)
+
+  def _compute_erosion(self, mask, kernel):
+    kernel_full = np.sum(kernel)
+    mask_erd = scipy.signal.convolve2d(mask, kernel, mode='same')
+    mask_erd[mask_erd < kernel_full] = 0
+    mask_erd[mask_erd == kernel_full] = 1
+    return mask_erd
+
+  def _compute_relax_pre_rec(self, true, pred, rho):
+    """Computes relaxed precision and recall."""
+    kernel = np.ones((2 * rho - 1, 2 * rho - 1))
+    map_zeros = np.zeros_like(pred)
+    map_ones = np.ones_like(pred)
+
+    pred_filtered = scipy.signal.convolve2d(pred, kernel, mode='same')
+    # True positive for relaxed precision
+    relax_pre_tp = np.where((true == 1) & (pred_filtered > 0), map_ones,
+                            map_zeros)
+
+    true_filtered = scipy.signal.convolve2d(true, kernel, mode='same')
+    # True positive for relaxed recall
+    relax_rec_tp = np.where((pred == 1) & (true_filtered > 0), map_ones,
+                            map_zeros)
+
+    return np.sum(relax_pre_tp) / np.sum(pred), np.sum(relax_rec_tp) / np.sum(
+        true)
+
+  def _convert_to_numpy(self, groundtruths, predictions):
+    """Converts tesnors to numpy arrays."""
+    numpy_groundtruths = groundtruths.numpy()
+    numpy_predictions = predictions.numpy()
+
+    return numpy_groundtruths, numpy_predictions
+
+  def update_state(self, groundtruths, predictions):
+    """Update segmentation results and groundtruth data.
+
+    Args:
+      groundtruths : Tuple of single Tensor [batch, width, height, 1],
+                     groundtruth masks. range [0, 1]
+      predictions  : Tuple of single Tensor [batch, width, height, 1],
+                     predicted masks. range [0, 1]
+    """
+
+    groundtruths, predictions = self._convert_to_numpy(groundtruths[0],
+                                                       predictions[0])
+    for (true, pred) in zip(groundtruths, predictions):
+      self._groundtruths.append(true)
+      self._predictions.append(pred)
+