wojzaremba / algorithm-learning Goto Github PK

Hi,

In Section 5, it is said that :

For each task, we train a separate model, starting with sequences of complexity 6 and incrementing
by 4 once it achieves 100% accuracy on held-out examples of the current length. Training stops
once the model successfully generalizes to examples of complexity 1000.

From this part, I cannot understand whether you meant the 100% accuracy in training set or test set. Or is help-out examples totally different ? I tried to get the idea from your code but it looks like a bit complicated.

After training on 30 million characters, the model hasn't achieved 100% accuracy on 2,4, or 6 characters (it gets stuck at about 60% on a per-character basis). I've left all the parameters to their default values. Do I need to change something, perhaps the random seed?

Hello,

I have only recently started investigating NPI and your algorithm-learning neural network but you video looks very good and I was able to install everything to start the training as per your github post.

With this in mind, and still being very new to lua, I looked through the code but did not see any command-line flags or switches that allows me to save, load, or test a trained network.

If you could please help me with this functionality then I would greatly appreciate it as I need to be able train a network to add, for example, and then save that network for later use when I will load it and feed some numbers to it for addition.

I would like to do this with some of the other functionality as well.

Any help would be greatly appreciated.

Cheers,
Lonnie

What are these symbols stand for? I cannot understand them by just reading the code.

local chars = {"q", "e", "s", "p", "a", "c", "y", "k", "v"}