Closed barkincavdaroglu closed 1 year ago
Something isn't working properly. Although the problem of vanishing gradients is resolved, they diverge a lot after epoch >= 5. https://github.com/tomgoldstein/loss-landscape
This is no longer the case.
Something isn't working properly. Although the problem of vanishing gradients is resolved, they diverge a lot after epoch >= 5. https://github.com/tomgoldstein/loss-landscape