How to Add Tensorboard to the Tensorflow Evaluation Process

Question

I took the enclosed Abalone example and made sure that I understood it ... Well, I think I know. But since another evaluation project that I am working on is producing complete garbage - I tried to add a tensor board, so I can understand what is happening.

Base code https://www.tensorflow.org/extend/estimators

I added a session and a writer

    # Set model params
    model_params = {"learning_rate": 0.01}
    with  tf.Session ()   as  sess: 
        # Instantiate Estimator
        nn = tf.contrib.learn.Estimator(model_fn=model_fn, params=model_params)
        writer  =  tf.summary.FileWriter ( '/tmp/ab_tf' ,  sess.graph)
        nn.fit(x=training_set.data, y=training_set.target, steps=5000)   
        # Score accuracy
        ev = nn.evaluate(x=test_set.data, y=test_set.target, steps=1)


And added 1 line in the model_fn function so it looks like this...


def model_fn(features, targets, mode, params):
  """Model function for Estimator."""

  # Connect the first hidden layer to input layer
  # (features) with relu activation
  first_hidden_layer = tf.contrib.layers.relu(features, 49)

  # Connect the second hidden layer to first hidden layer with relu
  second_hidden_layer = tf.contrib.layers.relu(first_hidden_layer, 49)

  # Connect the output layer to second hidden layer (no activation fn)
  output_layer = tf.contrib.layers.linear(second_hidden_layer, 1)

  # Reshape output layer to 1-dim Tensor to return predictions
  predictions = tf.reshape(output_layer, [-1])
  predictions_dict = {"ages": predictions}

  # Calculate loss using mean squared error
  loss = tf.losses.mean_squared_error(targets, predictions)

  # Calculate root mean squared error as additional eval metric
  eval_metric_ops = {
      "rmse": tf.metrics.root_mean_squared_error(
          tf.cast(targets, tf.float64), predictions)
  }

  train_op = tf.contrib.layers.optimize_loss(
      loss=loss,
      global_step=tf.contrib.framework.get_global_step(),
      learning_rate=params["learning_rate"],
      optimizer="SGD")


  tf.summary.scalar('Loss',loss)

  return model_fn_lib.ModelFnOps(
      mode=mode,
      predictions=predictions_dict,
      loss=loss,
      train_op=train_op,
      eval_metric_ops=eval_metric_ops)

Finally added

writer.close()

When I run the code ... I get the data file in / tmp / ab _tf, this file is NOT null. But it is also only 139 bytes ... which means that nothing is being written ....

When I open it with a tensor board - no data.

What am I doing wrong?

Rate any input ...

Answer 1

. model_dir .

, model_dir './tmp/model', - logdir =./tmp/model

Answer 2

, . , , model_dir :

nn = tf.contrib.learn.Estimator(model_fn = model_fn, params = model_params, model_dir = FLAGS.log_dir)

TensorFlow API : https://www.tensorflow.org/api_docs/python/tf/contrib/learn/Estimator

Answer 3

. :

merged = tf.summary.merge_all()  # easiest way to do it when you have lots of summaries

merged op :

summary, _ = sess.run([merged, nn.train_op], feed_dict=...)

, , , , :

writer.add_summary(summary, tf.train.get_global_step())

I'm not sure how to pass additional operators to nn.fitand nn.evaluate, but there should be a way to do this. Otherwise, just call:

summary = sess.run(merged, feed_dict=...)

separately, and then write down the summary data.