def prepare_config(self, teacher_block_num, student_block_num,
transfer_teacher_layers):
# using small model for testing
task_config = distillation.BertDistillationTaskConfig(
teacher_model=bert.PretrainerConfig(encoder=encoders.EncoderConfig(
type='mobilebert',
mobilebert=encoders.MobileBertEncoderConfig(
num_blocks=teacher_block_num)),
cls_heads=[
bert.ClsHeadConfig(
inner_dim=256,
num_classes=2,
dropout_rate=0.1,
name='next_sentence')
],
mlm_activation='gelu'),
student_model=bert.PretrainerConfig(encoder=encoders.EncoderConfig(
type='mobilebert',
mobilebert=encoders.MobileBertEncoderConfig(
num_blocks=student_block_num)),
cls_heads=[
bert.ClsHeadConfig(
inner_dim=256,
num_classes=2,
dropout_rate=0.1,
name='next_sentence')
],
mlm_activation='relu'),
train_data=pretrain_dataloader.BertPretrainDataConfig(
input_path='dummy',
max_predictions_per_seq=76,
seq_length=512,
global_batch_size=10),
validation_data=pretrain_dataloader.BertPretrainDataConfig(
input_path='dummy',
max_predictions_per_seq=76,
seq_length=512,
global_batch_size=10))
# set only 1 step for each stage
progressive_config = distillation.BertDistillationProgressiveConfig()
progressive_config.layer_wise_distill_config.transfer_teacher_layers = (
transfer_teacher_layers)
progressive_config.layer_wise_distill_config.num_steps = 1
progressive_config.pretrain_distill_config.num_steps = 1
optimization_config = optimization.OptimizationConfig(
optimizer=optimization.OptimizerConfig(
type='lamb',
lamb=optimization.LAMBConfig(weight_decay_rate=0.0001,
exclude_from_weight_decay=[
'LayerNorm', 'layer_norm',
'bias', 'no_norm'
])),
learning_rate=optimization.LrConfig(
type='polynomial',
polynomial=optimization.PolynomialLrConfig(
initial_learning_rate=1.5e-3,
decay_steps=10000,
end_learning_rate=1.5e-3)),
warmup=optimization.WarmupConfig(
type='linear',
linear=optimization.LinearWarmupConfig(
warmup_learning_rate=0)))
exp_config = cfg.ExperimentConfig(
task=task_config,
trainer=prog_trainer_lib.ProgressiveTrainerConfig(
progressive=progressive_config,
optimizer_config=optimization_config))
# Create a teacher model checkpoint.
teacher_encoder = encoders.build_encoder(
task_config.teacher_model.encoder)
pretrainer_config = task_config.teacher_model
if pretrainer_config.cls_heads:
teacher_cls_heads = [
layers.ClassificationHead(**cfg.as_dict())
for cfg in pretrainer_config.cls_heads
]
else:
teacher_cls_heads = []
masked_lm = layers.MobileBertMaskedLM(
embedding_table=teacher_encoder.get_embedding_table(),
activation=tf_utils.get_activation(
pretrainer_config.mlm_activation),
initializer=tf.keras.initializers.TruncatedNormal(
stddev=pretrainer_config.mlm_initializer_range),
name='cls/predictions')
teacher_pretrainer = models.BertPretrainerV2(
encoder_network=teacher_encoder,
classification_heads=teacher_cls_heads,
customized_masked_lm=masked_lm)
# The model variables will be created after the forward call.
_ = teacher_pretrainer(teacher_pretrainer.inputs)
teacher_pretrainer_ckpt = tf.train.Checkpoint(
**teacher_pretrainer.checkpoint_items)
teacher_ckpt_path = os.path.join(self.get_temp_dir(),
'teacher_model.ckpt')
teacher_pretrainer_ckpt.save(teacher_ckpt_path)
exp_config.task.teacher_model_init_checkpoint = self.get_temp_dir()
return exp_config
optimization_config = optimization.OptimizationConfig(
optimizer=optimization.OptimizerConfig(
type='lamb',
lamb=optimization.LAMBConfig(
weight_decay_rate=0.01,
exclude_from_weight_decay=['LayerNorm', 'bias', 'norm'],
clipnorm=1.0)),
learning_rate=optimization.LrConfig(
type='polynomial',
polynomial=optimization.PolynomialLrConfig(
initial_learning_rate=1.5e-3,
decay_steps=10000,
end_learning_rate=1.5e-3)),
warmup=optimization.WarmupConfig(
type='linear',
linear=optimization.LinearWarmupConfig(warmup_learning_rate=0)))
# copy from progressive/utils.py due to the private visibility issue.
def config_override(params, flags_obj):
"""Override ExperimentConfig according to flags."""
# Change runtime.tpu to the real tpu.
params.override({'runtime': {
'tpu': flags_obj.tpu,
}})
# Get the first level of override from `--config_file`.
# `--config_file` is typically used as a template that specifies the common
# override for a particular experiment.
for config_file in flags_obj.config_file or []:
params = hyperparams.override_params_dict(params,