Feature request add an option to pass activation function to TTA

Question

Feature request add an option to pass activation function to TTA

ternaus opened this issue 5 years ago · 1 comments

https://github.com/BloodAxe/pytorch-toolbelt/blob/develop/pytorch_toolbelt/inference/tta.py#L135

In many cases averaging logits works worse than averaging probabilities => would be nice to be able to pass user-defined activation function. For example softmax or sigmoid.

Answer 1 · 2019-09-27T22:04:55.000Z

A better option would be to apply desired functions on the model itself:

https://github.com/BloodAxe/Kaggle-2019-Blindness-Detection/blob/master/retinopathy/inference.py#L159