Train a tensorflow model minimizing the loss of several batches - python

I would like to train the weights of a model based on the sum of the loss value of several batches. However it seems that once you run the graph for each of the individual batches, the object that is returned is just a regular numpy array. So when you try and use an optimizer like GradientDescentOptimizer, it no longer has information about the variables that were used to calculate the sum of the losses, so it can't find the gradients of the weights that what help minimize the loss. Here's an example tensorflow script to illustrate what I'm talking about:
weights = tf.Variable(tf.ones([num_feature_values], tf.float32))
feature_values = tf.placeholder(tf.int32, shape=[num_feature_values])
labels = tf.placeholder(tf.int32, shape=[1])
loss_op = some_loss_function(weights, feature_values, labels)
with tf.Session() as sess:
for batch in batches:
feed_dict = fill_feature_values_and_labels(batch)
#Calculates loss for one batch
loss = sess.run(loss_op, feed_dict=feed_dict)
#Adds it to total loss
total_loss += loss
# Want to train weights to minimize total_loss, however this
# doesn't work because the graph has already been run.
optimizer = tf.train.GradientDescentOptimizer(1.0).minimize(total_loss)
with tf.Session() as sess:
for step in xrange(num_steps):
sess.run(optimizer)
The total_loss is a numpy array and thus cannot be used in the optimizer. Does anyone know a way around the problem, where I want to use information across many batches but still need the graph intact in order to preserve the fact that the total_loss is a function of the weights?

The thing you optimize in any of the trainers must be a part of the graph, here what you train on is the actual realized result, so it won't work.
I think the way you should probably do this is to construct your input as a batch of batches e.g.
intput = tf.placeholder("float", (number_of_batches, batch_size, input_size)
Then have your target also be a 3d tensor which can be trained on.

Related

How should I keep track of total loss while training a network with a batched dataset?

I am attempting to train a discriminator network by applying gradients to its optimizer. However, when I use a tf.GradientTape to find the gradients of loss w.r.t training variables, None is returned. Here is the training loop:
def train_step():
#Generate noisy seeds
noise = tf.random.normal([BATCH_SIZE, noise_dim])
with tf.GradientTape() as disc_tape:
pattern = generator(noise)
pattern = tf.reshape(tensor=pattern, shape=(28,28,1))
dataset = get_data_set(pattern)
disc_loss = tf.Variable(shape=(1,2), initial_value=[[0,0]], dtype=tf.float32)
disc_tape.watch(disc_loss)
for batch in dataset:
disc_loss.assign_add(discriminator(batch, training=True))
disc_gradients = disc_tape.gradient(disc_loss, discriminator.trainable_variables)
Code Description
The generator network generates a 'pattern' from noise. I then generate a dataset from that pattern by applying various convolutions to the tensor. The dataset that is returned is batched, so I iterate through the dataset and keep track of the loss of my discriminator by adding the loss from this batch to the total loss.
What I do know
tf.GradientTape returns None when there is no graph connection between the two variables. But isn't there a graph connection between loss and trainable variables? I believe my mistake has something to do with how I keep track of loss in the disc_loss tf.Variable
My Question
How do I keep track of loss while iterating through a batched dataset so that I may use it later to calculate gradients?
The base answer here is that the assign_add function of tf.Variable is not differentiable, thus no gradient can be calculated between the variable disc_loss and the discriminator trainable variables.
In this very specific case, the answer was
disc_loss = disc_loss + discriminator(batch, training=True)
In future cases of similar problems, be sure to check that all operations used while being watched by the gradient tape are differentiable.
This link has a list of differentiable and non-differentiable tensorflow ops. I found it very useful.

How to inject data into a graph when using an input pipeline?

I am using an initializable iterator in my code. The iterator returns batches of size 100 from a csv dataset that has 20.000 entries. During training, however, I came across a problem. Consider this piece of code:
def get_dataset_iterator(batch_size):
# parametrized with batch_size
dataset = ...
return dataset.make_initializable_iterator()
## build a model and train it (x is the input of my model)
iterator = get_dataset_iterator(100)
x = iterator.get_next()
y = model(x)
## L1 norm as loss, this works because the model is an autoencoder
loss = tf.abs(x - y)
## training operator
train_op = tf.train.AdamOptimizer(0.01).minimize(loss)
with tf.Session() as sess:
for epoch in range(100):
sess.run(iterator.initializer)
# iterate through the whole dataset once during the epoch and
# do 200 mini batch updates
for _ in range(number_of_samples // batch_size):
sess.run(train_op)
print(f'Epoch {epoch} training done!')
# TODO: print loss after epoch here
I am interested in the training loss AFTER finishing the epoch. It makes most sense to me that I calculate the average loss over the whole training set (e.g. feeding all 20.000 samples through the network and averaging their loss). I could reuse the dataset iterator here with a batch size of 20.000, but I have declared x as the input.
So the questions are:
1.) Does the loss calculation over all 20.000 examples make sense? I have seen some people do the calculation with just a mini-batch (the last batch of the epoch).
2.) How can I calculate the loss over the whole training set with an input pipeline? I have to inject all of training data somehow, so that I can run sess.run(loss) without calculating it over only 100 samples (because x is declared as input).
EDIT FOR CLARIFICATION:
If I wrote my training loop the following way, there would be some things that bother me:
with tf.Session() as sess:
for epoch in range(100):
sess.run(iterator.initializer)
# iterate through the whole dataset once during the epoch and
# do 200 mini batch updates
for _ in range(number_of_samples // batch_size):
_, current_loss = sess.run([train_op, loss])
print(f'Epoch {epoch} training done!')
print(current_loss)
Firstly, loss would still be evaluated before doing the last weight update. That means whatever comes out is not the latest value. Secondly, I would not be able to access current_loss after exiting the for loop so I would not be able to print it.
1) Loss calculation over the whole training set (before updating weights) does make sense and is called batch gradient descent (despite using the whole training set and not a mini batch).
However, calculating a loss for your whole dataset before updating weights is slow (especially with large datasets) and training will take a long time to converge. As a result, using a mini batch of data to calculate loss and update weights is what is normally done instead. Although using a mini batch will produce a noisy estimate of the loss it is actually good enough estimate to train networks with enough training iterations.
EDIT:
I agree that the loss value you print will not be the latest loss with the latest updated weights. Probably for most cases it really doesn't make much different or change results so people just go with how you have wrote the code above. However, if you really want to obtain the true latest loss value after you have done training (to print out) then you will just have to run the loss op again after you have done a train op e.g.:
for _ in range(number_of_samples // batch_size):
sess.run([train_op])
current_loss = sess.run([loss])
This will get your true latest value. Of course this wont be on the whole dataset and will be just for a minibatch of 100. Again the value is likely a good enough estimate but if you wish to calculate exact loss for whole dataset you will have to run through your entire set e.g. another loop and then average the loss:
...
# Train loop
for _ in range(number_of_samples // batch_size):
_, current_loss = sess.run([train_op, loss])
print(f'Epoch {epoch} training done!')
# Calculate loss of whole train set after training an epoch.
sess.run(iterator.initializer)
current_loss_list = []
for _ in range(number_of_samples // batch_size):
_, current_loss = sess.run([loss])
current_loss_list.append(current_loss)
train_loss_whole_dataset = np.mean(current_loss_list)
print(train_loss_whole_dataset)
EDIT 2:
As pointed out doing the serial calls to train_op then loss will call the iterator twice and so things might not work out nicely (e.g. run out of data). Therefore my 2nd bit of code will be better to use.
I think the following code will answer your questions:
(A) how can you print the batch loss AFTER performing the train step? (B) how can you calculate the loss over the entire training set, even though the dataset iterator gives only a batch each time?
import tensorflow as tf
import numpy as np
dataset_size = 200
batch_size= 5
dimension = 4
# create some training dataset
dataset = tf.data.Dataset.\
from_tensor_slices(np.random.normal(2.0,size=(dataset_size,dimension)).
astype(np.float32))
dataset = dataset.batch(batch_size) # take batches
iterator = dataset.make_initializable_iterator()
x = tf.cast(iterator.get_next(),tf.float32)
w = tf.Variable(np.random.normal(size=(1,dimension)).astype(np.float32))
loss_func = lambda x,w: tf.reduce_mean(tf.square(x-w)) # notice that the loss function is a mean!
loss = loss_func(x,w) # this is the loss that will be minimized
train_op = tf.train.GradientDescentOptimizer(0.1).minimize(loss)
# we are going to use control_dependencies so that we know that we have a loss calculation AFTER the train step
with tf.control_dependencies([train_op]):
loss_after_train_op = loss_func(x,w) # this is an identical loss, but will only be calculated AFTER train_op has
# been performed
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
# train one epoch
sess.run(iterator.initializer)
for i in range(dataset_size//batch_size):
# the training step will update the weights based on ONE batch of examples each step
loss1,_,loss2 = sess.run([loss,train_op,loss_after_train_op])
print('train step {:d}. batch loss before step: {:f}. batch loss after step: {:f}'.format(i,loss1,loss2))
# evaluate loss on entire training set. Notice that this calculation assumes the the loss is of the form
# tf.reduce_mean(...)
sess.run(iterator.initializer)
epoch_loss = 0
for i in range(dataset_size // batch_size):
batch_loss = sess.run(loss)
epoch_loss += batch_loss*batch_size
epoch_loss = epoch_loss/dataset_size
print('loss over entire training dataset: {:f}'.format(epoch_loss))
As for your question whether it makes sense to calculate loss over the entire training set - yes, it makes sense, for evaluation purposes. It usually does not make sense to perform training steps which are based on all of the training set since this set is usually very large and you want to update your weights more often, without needing to go over the entire training set each time.

Why is this TensorFlow code so slow?

I am training a deep neural network multi-class classifier using TensorFlow. The network outputs the linear values from the final layer, which the tf.nn.softmax_cross_entropy_with_logits cost function takes as input. However, I don't really care about that linear output per se - I want to know what it looks like when the softmax function is applied to it.
Below the relevant parts of my code:
def train_network(x, num_hidden_layers):
prediction = neural_network_model(x, num_hidden_layers)
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=prediction, labels=y))
optimizer = tf.train.AdamOptimizer(learning_rate=0.01).minimize(cost)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
# train the network
...
# get the network output; x_test is my test data (len=663)
output = sess.run(prediction,feed_dict={x: x_test})
# get softmax values of output
for i in range(len(x_test)):
softm = sess.run(tf.nn.softmax(output[i]))
pred_class = sess.run(tf.argmax(softm))
print(pred_class)
...
Now, that final for-loop in which I calculate the softmax values is extremely slow. Why is that, and how do I do this properly?

How to get class predictions from a trained Tensorflow classifier?

I have have trained binary classifier model. The model class contains self.cost, self.initial_state, self.final_state and self.logits params. It is saved simply with tf.train.Saver:
saver = tf.train.Saver(tf.global_variables(), max_to_keep=1)
saver.save(session, 'model.ckpt')
After the model was trained I load it as:
with tf.variable_scope("Model", reuse=False):
model = MODEL(config, is_training=False)
with tf.Session() as session:
saver = tf.train.Saver(tf.global_variables())
saver.restore(session, 'model.ckpt')
However, my model.run function returns cross-entropy loss which is the last op in the graph. I don't need loss, I need the model predictions for each batch element
logits = tf.sigmoid(tf.nn.xw_plus_b(last_layer, self.output_w, self.output_b))
where last_layer is a 800x1 matrix which I then later reshape into 32x25x1 (batch_size, sequence_length, 1) matrix. It is this matrix that contains the model prediction values in [0-1] range.
So, how can I use this model to make a prediction for single element matrix 1x1x1?
Add the OPs necessary to compute accuracy, something like what I have copied below (simply copied out of the closest model I had at hand).
self.logits_flat = tf.argmax(logits, axis=1, output_type=tf.int32)
labels_flat = tf.argmax(labels, axis=1, output_type=tf.int32)
accuracy = tf.cast(tf.equal(self.logits_flat, labels_flat), tf.float32, name='accuracy')
Now when you run the model (either during test or training time) add accuracy to the sess.run call as:
sess.run([train_op, accuracy], feed_dict=...)
or
sess.run([accuracy, logits], feed_dict=...)
All you're doing when you call sess.run is to tell tensorflow to compute the value of whatever you ask for. You need to pass it in any data it needs to perform those computations. Tensorflow is lazy, it won't perform any computations that aren't explicitly necessary to produce the results you request. E.g. if you run the second version of sess.run listed above the optimizer will not be run and hence your weights will not be updated.
Note that you can add the OPs after the network was trained because none of them actually add any variables so they won't affect the save/restore process any.

How does the tensorflow word2vec tutorial update embeddings?

This thread comes close: What is the purpose of weights and biases in tensorflow word2vec example?
But I am still missing something from my interpretation of this: https://github.com/tensorflow/tensorflow/blob/r1.2/tensorflow/examples/tutorials/word2vec/word2vec_basic.py
From what I understand, you feed the network the indices of target and context words from your dictionary.
_, loss_val = session.run([optimizer, loss], feed_dict=feed_dict)
average_loss += loss_val
The batch inputs are then looked up to return the vectors that are randomly generated at the beginning
embeddings = tf.Variable(
tf.random_uniform([vocabulary_size, embedding_size], -1.0, 1.0))
# Look up embeddings for inputs.
embed = tf.nn.embedding_lookup(embeddings, train_inputs)
Then an optimizer adjusts the weights and biases to best predict the label as opposed to num_sampled random alternatives
loss = tf.reduce_mean(
tf.nn.nce_loss(weights=nce_weights,
biases=nce_biases,
labels=train_labels,
inputs=embed,
num_sampled=num_sampled,
num_classes=vocabulary_size))
# Construct the SGD optimizer using a learning rate of 1.0.
optimizer = tf.train.GradientDescentOptimizer(1.0).minimize(loss)
My questions are as follows:
Where do the embeddings variable get updated?. It appears to me that I could get the final result by either running the index of a word through the neural network, or by just taking the final_embeddings vectors and using that. But I do not understand where embeddings is ever changed from its random initialization.
If I were to draw this computation graph, what would it look like (or better yet, what is the best way to actually do so)?
Is this running all of the context/target pairs in the batch at once? Or one by one?
Embeddings: Embeddings is a variable. It gets updated every time you do backprop (while running optimizer with loss)
Grpah: Did you try saving the graph and displaying it in tensorboard ? Is this what you're looking for ?
Batching: Atleast in the example you linked, he is doing batch processing using the function at line 96. https://github.com/tensorflow/tensorflow/blob/r1.2/tensorflow/examples/tutorials/word2vec/word2vec_basic.py#L96
Please correct me if I misunderstood your question.

Categories

Resources