How to interpret caffe log with debug_info?

by

At first glance you can see this log section divided into two: [Forward] and [Backward]. Recall that neural network training is done via forward-backward propagation:
A training example (batch) is fed to the net and a forward pass outputs the current prediction.
Based on this prediction a loss is computed.
The loss is then derived, and a gradient is estimated and propagated backward using the chain rule.

Caffe Blob data structure
Just a quick re-cap. Caffe uses Blob data structure to store data/weights/parameters etc. For this discussion it is important to note that Blob has two “parts”: data and diff. The values of the Blob are stored in the data part. The diff part is used to store element-wise gradients for the backpropagation step.

Forward pass

You will see all the layers from bottom to top listed in this part of the log. For each layer you’ll see:

I1109 ...]     [Forward] Layer conv1, top blob conv1 data: 0.0645037

I1109 ...]     [Forward] Layer conv1, param blob 0 data: 0.00899114
I1109 ...]     [Forward] Layer conv1, param blob 1 data: 0

Layer "conv1" is a convolution layer that has 2 param blobs: the filters and the bias. Consequently, the log has three lines. The filter blob (param blob 0) has data

 I1109 ...]     [Forward] Layer conv1, param blob 0 data: 0.00899114

That is the current L2 norm of the convolution filter weights is 0.00899.
The current bias (param blob 1):

 I1109 ...]     [Forward] Layer conv1, param blob 1 data: 0

meaning that currently the bias is set to 0.

Last but not least, "conv1" layer has an output, "top" named "conv1" (how original…). The L2 norm of the output is

 I1109 ...]     [Forward] Layer conv1, top blob conv1 data: 0.0645037

Note that all L2 values for the [Forward] pass are reported on the data part of the Blobs in question.

Loss and gradient
At the end of the [Forward] pass comes the loss layer:

I1109 ...]     [Forward] Layer loss, top blob loss data: 2031.85

I1109 ...]     [Backward] Layer loss, bottom blob fc1 diff: 0.124506

In this example the batch loss is 2031.85, the gradient of the loss w.r.t. fc1 is computed and passed to diff part of fc1 Blob. The L2 magnitude of the gradient is 0.1245.

Backward pass
All the rest of the layers are listed in this part top to bottom. You can see that the L2 magnitudes reported now are of the diff part of the Blobs (params and layers’ inputs).

Finally
The last log line of this iteration:

[Backward] All net params (data, diff): L1 norm = (2711.42, 7086.66); L2 norm = (6.11659, 4085.07)

reports the total L1 and L2 magnitudes of both data and gradients.

What should I look for?

  1. If you have nans in your loss, see at what point your data or diff turns into nan: at which layer? at which iteration?

  2. Look at the gradient magnitude, they should be reasonable. IF you are starting to see values with e+8 your data/gradients are starting to blow up. Decrease your learning rate!

  3. See that the diffs are not zero. Zero diffs mean no gradients = no updates = no learning. If you started from random weights, consider generating random weights with higher variance.

  4. Look for activations (rather than gradients) going to zero. If you are using "ReLU" this means your inputs/weights lead you to regions where the ReLU gates are “not active” leading to “dead neurons”. Consider normalizing your inputs to have zero mean, add "BatchNorm" layers, setting negative_slope in ReLU.

More Related Contents:

  1. Common causes of nans during training of neural networks
  2. What is `weight_decay` meta parameter in Caffe?
  3. Caffe: What can I do if only a small batch fits into memory?
  4. What is `lr_policy` in Caffe?
  5. Caffe | solver.prototxt values setting strategy
  6. How to calculate optimal batch size
  7. How to create caffe.deploy from train.prototxt
  8. How to reduce a fully-connected (`”InnerProduct”`) layer using truncated SVD
  9. Why binary_crossentropy and categorical_crossentropy give different performances for the same problem?
  10. What is a `”Python”` layer in caffe?
  11. Many to one and many to many LSTM examples in Keras
  12. Epoch vs Iteration when training neural networks [closed]
  13. Cost function training target versus accuracy desired goal
  14. What is the difference between loss function and metric in Keras? [duplicate]
  15. Keras Text Preprocessing – Saving Tokenizer object to file for scoring
  16. Does Caffe need data to be shuffled?
  17. Why should weights of Neural Networks be initialized to random numbers? [closed]
  18. Tackling Class Imbalance: scaling contribution to loss and sgd
  19. Is deep learning bad at fitting simple non linear functions outside training scope (extrapolating)?
  20. multi-layer perceptron (MLP) architecture: criteria for choosing number of hidden layers and size of the hidden layer? [closed]
  21. How to interpret loss and accuracy for a machine learning model [closed]
  22. Fine Tuning of GoogLeNet Model
  23. What is the difference between Keras model.evaluate() and model.predict()?
  24. Machine Learning Two class classification [closed]
  25. Artificial Neural Network Toplogy
  26. Clarification on a Neural Net that plays Snake
  27. Why do we need to call zero_grad() in PyTorch?
  28. ResNet: 100% accuracy during training, but 33% prediction accuracy with the same data
  29. Keras: model.evaluate vs model.predict accuracy difference in multi-class NLP task
  30. caffe: model definition: write same layer with different phase using caffe.NetSpec()

Leave a Comment