# Why gradients are so small in deep learning?

The learning rate in my model is 0.00001 and the gradients of the model is within the distribution of [-0.0001, 0.0001]. Is it normal?

• Hi and welcome to AI SE. The gradient is usually a vector (if you have more than one parameter), so it is not a scalar. Please, clarify this. – nbro Feb 29 at 12:53
• @nbro, Yes, that true. The value distribution is within [-0.0001, 0.0001] – GoingMyWay Feb 29 at 14:23
• These values are not so small if you're at the end of the training. At which epoch do you have these values? Could you also provide more details about your architecture, loss function, etc.? – nbro Feb 29 at 16:24
• Seems kinda arbitrary, really. If anything, you might think about scaling up your loss function to make it more sensitive. The main concern would be reducing harmful rounding due to the limited precision of floating-points. – Nat Mar 2 at 13:44
• @nbro at the early stage of training. I am using Normal initializer with mean=0 and std=1. – GoingMyWay Mar 10 at 8:58