Closed michaelbzms closed 11 months ago
The program has to be in 1024 tokens. Each weight in your network would occupy some tokens, so I think most people agree that it's not possible to fit a good NN into this limit.
I see. Yes, even for logistic regression, we would need at least 768 weights/tokens (768 being the input dimension).
Suppose I were to train a simple neural network using python to use in my AI. How would I be able to use it in this project?
I am not familiar with C#, so I am trying to see if such an approach would be viable. Is there a matrix multiplication library in C# that we are allowed to use? How do we set the weights of these matrices?