Skip to content

Conversation

@DominikKuenkele
Copy link

Description

Fix the calculation of the accuracy when running the signal game with GumbelSoftmax.
Here, the mean of the accuracies for the batch was calculated already in the loss function, although it should be only calculated in the forward call of the game.

Related Issue (if any)

Issue #256

Motivation and Context

Signal game can be run using GumbelSoftmax (see Issue #256)

How Has This Been Tested?

Error is fixed and the agents are trained on the data, using GumbelSoftmax.

@facebook-github-bot facebook-github-bot added the CLA Signed Do not delete this pull request or issue due to inactivity. label Sep 1, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed Do not delete this pull request or issue due to inactivity.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants