Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix: explore_eval don't learn if logged action not in predicted actions #4262

Merged
merged 7 commits into from
Nov 8, 2022

Conversation

olgavrou
Copy link
Collaborator

@olgavrou olgavrou commented Nov 2, 2022

No description provided.

@@ -164,6 +165,8 @@ void do_actual_learning(explore_eval& data, multi_learner& base, VW::multi_ex& e
if (data.known_cost.action == a_s[i].action) { action_probability = a_s[i].score; }
}

if (action_probability == 0) { return; }

float threshold = action_probability / data.known_cost.probability;

if (!data.fixed_multiplier) { data.multiplier = std::min(data.multiplier, 1 / threshold); }
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

now that probs can be zero, this 1 / threshold could cause division by zero

olgavrou and others added 2 commits November 2, 2022 09:29
@olgavrou olgavrou changed the title fix: explore_eval don't learn if prob is zero fix: explore_eval don't learn if logged action not in predicted actions Nov 2, 2022
@olgavrou olgavrou merged commit 0406c0f into VowpalWabbit:master Nov 8, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants