I wrote one of the most comprehensive deep learning tutorials for using PyTorch for Numer.AI stock market prediction.
It is a binary classification problem, and the tutorial includes Kaggle style ROC_AUC plots which are rarely seen in PyTorch.
Comments are welcomed, I am sure I have bugs and mistakes.
https://github.com/QuantScientist/Deep-Learning-Boot-Camp/blob/master/day02-PyTORCH-and-PyCUDA/PyTorch/18-PyTorch-NUMER.AI-Binary-Classification-BCELoss.ipynb
Best,
1 Like
BCELoss has weights parameter but you don’t mention how to use it 
1 Like
Since I have not used the weights /didnt need to use them. Once I do, I shall update the tutorial.
