regretful-agent icon indicating copy to clipboard operation
regretful-agent copied to clipboard

Possible typo in code (or paper)

Open MrEarle opened this issue 3 years ago • 0 comments

Hi!

I was looking at the progress monitor, which the paper shows it's calculation as:

paper_equation

But this line in the code shows a slightly different equation (one closing parenthesis changed position): https://github.com/chihyaoma/regretful-agent/blob/5caf7b500667981bc7064e4d31b49e83db64c95a/tasks/R2R-pano/models/policy_model.py#L133

This would translate to (in the paper notation):

code_equation

The difference is that in the first, the tanh is included within the sigmoid, and, in the second equation, it's outside the sigmoid.

Is there any major difference between using these two equations?

MrEarle avatar Sep 23 '21 21:09 MrEarle