I have problems figuring out the right parameters in order to get my neural network working. I want to implement digit recognition using the Mnist data set of 60.000 26x26 images. I also learned that it makes sense to split the data into multiple "chunks" but I have a couple more questions: How many images per chunk? How to set the learning rate? How many backpropagation cycles per chunk? Do you know helpful articles where I can get that kind of information?
First of all, if this is your first network I would suggest you using Keras - it is easy to use and get the feeling of a network. When you have understood the principles of that you could use pyTorch/Tensorflow (i personally use PyTorch).
Now to your question: I assume you are implementing a standard neural network (that is not a CNN).
Chunks: These "chunks" are normally called batches and it is a way to divide your data into smaller pieces - it has several advantages/disadvantages, but the important thing is that it reduces the memory used. Unless you have a lot(!!) of memory you cannot train your network on 50 000 images at a time, thus you might want to parse in the first 256 images, then the next 256 images and so on. The size of the batch is determined mostly by your memory - you should be okay with a batch size of 256 (you can try to increase it if you want).
Learning rate: The learning rate depends on your optimizer. Say you are using SGD (stochastic gradient descent) the standard in PyTorch is 0.1, which normally works very well. It is a parameter you can tune if you are getting some "odd" results e.g "nan" in the weights, or if you have a complicated problem, but I would suggest you to just stick with the standard and maybe try different values, when you know your network works well. If it keeps giving you "nan", try reduce it to 0.01 for instance.
Back prop: Backpropagation is when you update your weights, which you do after you have trained a batch, thus you will do one "back propagation" pr. batch.
I would suggest you looking at "threeblueonebrows" tutorial - https://www.youtube.com/watch?v=aircAruvnKk - which explains the concept of a network, also a kaggle guide using pytorch - https://www.kaggle.com/kanncaa1/pytorch-tutorial-for-deep-learning-lovers.