tf_hyperparameters_tuning.py

# Hyperparameter tuning using Tensorflow and Tensor board

# Testing performance with varioud optimizers
HP_OPTIMIZER = hp.HParam('optimizer', hp.Discrete(['adam', 'sgd', 'rmsprop']))
METRIC_ACCURACY = 'accuracy'
with tf.summary.create_file_writer('logs/hparam_tuning').as_default():
  hp.hparams_config(
    hparams=[HP_OPTIMIZER],
    metrics=[hp.Metric(METRIC_ACCURACY, display_name='Accuracy')],
  )

# Record accuaracy for each combination of hyperparameters
def run(run_dir, hparams):
  with tf.summary.create_file_writer(run_dir).as_default():
    hp.hparams(hparams)  # record the values used in this trial
    accuracy = train_test_model(hparams)
    tf.summary.scalar(METRIC_ACCURACY, accuracy, step=1)

# Fit model with each combination of hyperparameters
def train_test_model(hparams):
  model = build_model()
  model.fit(train_dataset, epochs=EPOCHS,
            steps_per_epoch=STEPS_PER_EPOCH)
  _, accuracy = model.evaluate(val_dataset)
  return accuracy


# Run model with different combinations of hyperparameters
session_num = 0
for optimizer in HP_OPTIMIZER.domain.values:
  hparams = {
      HP_OPTIMIZER: optimizer,
  }
  run_name = "run-%d" % session_num
  print('--- Starting trial: %s' % run_name)
  print({h.name: hparams[h] for h in hparams})
  run('logs/hparam_tuning/' + run_name, hparams)
  session_num += 1