search

piEsposito / blitz-bayesian-deep-learning

A simple and extensible library to create Bayesian Neural Network layers on PyTorch.
GNU General Public License v3.0
918 stars 107 forks source link

Make the boston example converge in readme #88

Closed kirill-fedyanin closed 3 years ago

kirill-fedyanin commented 3 years ago

The code for the boston regression example in readme doesn't converge. It seems to differ from `blitz/examples/bayesian_regression_boston.py' a little bit. I changed an optimizer and added the activation function to match the code from examples and now the readme version converges as well.

There are other differences, like a number of epochs (100 vs 1000), cuda usage, and the complexity_cost_weight argument in elbo, but as the readme started to converge anyway, I've left out these changes to keep it simple.

piEsposito commented 3 years ago

Seems nice, thank you so much for contributing to BLiTZ. I'm just gonna let the tests run and then approve it.