vocabulary=vocabulary,
with_meta_optimizer=True,
restore_path=the_only_pupil_restore_path,
restore_optimizer_path=os.path.join(training_path, 'checkpoints', 'final'),
save_path=os.path.join(base, 'loss_best', 'test', 'pupil_training'),
allow_growth=True,
batch_size=BATCH_SIZE,
checkpoint_steps=None,
result_types=['perplexity', 'loss', 'bpc', 'accuracy'],
printed_result_types=['perplexity', 'loss', 'bpc', 'accuracy'],
stop=OPTIMIZER_RANGE,
# stop=4000,
train_dataset_text=train_text,
validation_dataset_texts=[valid_text],
results_collect_interval=1,
additions_to_feed_dict=opt_inf_add_feed,
validation_additions_to_feed_dict=valid_add_feed,
no_validation=False,
)
env.test(
restore_path=os.path.join(base, 'loss_best', 'test', 'pupil_training', 'checkpoints/final'),
save_path=os.path.join(base, 'loss_best', 'test', 'testing'),
additions_to_feed_dict=valid_add_feed,
validation_dataset_texts=[test_text],
valid_batch_kwargs=dict(
vocabulary=vocabulary
),
printed_result_types=['perplexity', 'loss', 'bpc', 'accuracy']
)
results_collect_interval=config['results_collect_interval'],
no_validation=False,
validation_batch_size=BATCH_SIZE,
valid_batch_kwargs=dict(num_unrollings=config['num_unrollings'], ),
log_launch=False,
)
if config['train']:
restore_path = os.path.join(os.path.expanduser(config['save_path']),
'checkpoints/all_vars/best')
if config['test']:
env.test(
restore_path=restore_path,
save_path=os.path.expanduser(config['save_path']) + '/testing',
vocabulary=vocabulary,
validation_datasets={'test': test_text},
printed_result_types=['perplexity', 'loss', 'bpc', 'accuracy'],
validation_batch_size=BATCH_SIZE,
valid_batch_kwargs=dict(num_unrollings=config['num_unrollings'], ),
log_launch=False,
)
if config['file_dialog']:
env.file_dialog(
restore_path=restore_path,
vocabulary=vocabulary,
input_replica_file=os.path.expanduser(config['input_replica_file']),
result_file=os.path.expanduser(config['file_dialog_answers_file']),
character_positions_in_vocabulary=cpiv,
batch_generator_class=BatchGenerator,
reset_state_after_model_answer=args.
reset_state_after_model_answer, # if True after bot answer hidden state is reset
answer_len_limit=500., # max number of characters in bot answer
# pupil_restore_paths=['debug_empty_meta_optimizer/not_learning_issue_es20_nn20/checkpoints/0'],
# stop=stop_specs,
restore_path=RESTORE_PATH,
save_path=training_path,
stop=1000,
results_collect_interval=500,
summary=False,
add_graph_to_summary=False,
train_dataset=dict(train='train'),
train_batch_kwargs=dict(data_dir=data_dir),
valid_batch_kwargs=dict(data_dir=data_dir),
# train_dataset_text='abc',
validation_datasets=dict(valid='validation'),
learning_rate=dict(
init=best_conf['learning_rate/init'],
decay=1.,
period=1e+6,
type='exponential_decay',
),
batch_size=BATCH_SIZE,
no_validation=True,
)
env.test(
restore_path=os.path.join(training_path, 'checkpoints/final'),
save_path=os.path.join(base, metric + '_best', 'test', 'testing'),
additions_to_feed_dict=valid_add_feed,
validation_datasets=dict(test='test'),
valid_batch_kwargs=dict(data_dir=data_dir),
printed_result_types=['perplexity', 'loss', 'bpc', 'accuracy'])
save_path=path,
# restore_path='lstm_sample_test/scipop3_1000_bs256_11.12/checkpoints/2000',
learning_rate=learning_rate,
batch_size=BATCH_SIZE,
checkpoint_steps=None,
result_types=['perplexity', 'loss', 'bpc', 'accuracy'],
printed_result_types=['perplexity', 'loss', 'bpc', 'accuracy'],
stop=stop_specs,
# stop=1000,
train_dataset=dict(train='train'),
train_batch_kwargs=dict(valid_size=VALID_SIZE),
valid_batch_kwargs=dict(valid_size=VALID_SIZE),
# train_dataset_text='abc',
validation_datasets=dict(valid='validation'),
results_collect_interval=100,
additions_to_feed_dict=train_add_feed,
validation_additions_to_feed_dict=valid_add_feed,
no_validation=False,
summary=False,
add_graph_to_summary=False,
)
env.test(restore_path=os.path.join(path, 'checkpoints/best'),
save_path=os.path.join(path, 'test'),
additions_to_feed_dict=valid_add_feed,
validation_datasets=dict(test='test'),
valid_batch_kwargs=dict(valid_size=VALID_SIZE),
printed_result_types=['perplexity', 'loss', 'bpc', 'accuracy'])
# log_device_placement=True)
valid_add_feed = [ # {'placeholder': 'sampling_prob', 'value': 1.},
{
'placeholder': 'dropout',
'value': 1.
}
]
add_metrics = ['bpc', 'perplexity', 'accuracy']
tf.set_random_seed(1)
env.build_pupil(batch_size=32,
num_layers=1,
num_nodes=[100],
num_output_layers=1,
num_output_nodes=[],
vocabulary_size=vocabulary_size,
embedding_size=150,
num_unrollings=4,
init_parameter=3.,
num_gpus=1,
regime='training_with_meta_optimizer',
additional_metrics=add_metrics,
going_to_limit_memory=True)
env.test(restore_path='lstm/text8_pretrain/checkpoints/200',
save_path='lstm/text8_pretrain/validation200',
vocabulary=vocabulary,
additions_to_feed_dict=valid_add_feed,
validation_dataset_texts=[valid_text],
printed_result_types=['perplexity', 'loss', 'bpc', 'accuracy'])
