Questions tagged [exploding-gradient-problem]

For questions related to the exploding gradient problem, which is the numerical problem associated with the significant increase (or explosion) of the numbers of the gradient vector of an objective function with respect to the parameters of a neural network, which is being trained with a gradient-based optimization algorithm and backpropagation. There is also the related vanishing gradient problem, which arises when the numbers become very small.

Filter by
Sorted by
Tagged with
0
votes
0answers
13 views

Cause of Exploding NLLLoss [migrated]

I have been trying to make Transformer based language model, for the loss function Negative Log-likelihood is implemented. For some reason, after a few iterations, there is a steep increase in ...
2
votes
2answers
67 views

What are the common pitfalls that we could face when training neural networks?

Apart from the vanishing or exploding gradient problems, what are other problems or pitfalls that we could face when training neural networks?
2
votes
1answer
52 views

Which work originally introduced gradient clipping?

The Deep Learning book mentions that it's been used for years but the oldest sources it mentions are from 2012: A simple type of solution has been in use by practitioners for many years: clipping ...
4
votes
1answer
70 views

What effect does batch norm have on the gradient?

Batch norm is a technique where they essentially standardize the activations at each layer, before passing it on to the next layer. Naturally, this will affect the gradient through the network. I have ...
5
votes
1answer
995 views

Why do ResNets avoid the vanishing gradient problem?

I read that, if we use the sigmoid or hyperbolic tangent activation functions in deep neural networks, we can have some problems with the vanishing of the gradient, and this is visible by the shapes ...
6
votes
2answers
273 views

How to deal with large (or NaN) neural network's weights?

My weights go from being between 0 and 1 at initialisation to exploding into the tens of thousands in the next iteration. In the 3rd iteration they become so large that only arrays of nan values are ...