search

HumanCompatibleAI / evaluating-rewards

Library to compare and evaluate reward functions
https://arxiv.org/abs/2006.13900
Apache License 2.0
61 stars 7 forks source link

Support normalizing divergence heatmaps #17

Closed AdamGleave closed 4 years ago

AdamGleave commented 4 years ago

Support dividing by $D(Zero,R_T)$ to normalize $D(R_S,R_T)$ and make EPIC distance comparable between columns.

codecov[bot] commented 4 years ago

Codecov Report

Merging #17 into master will decrease coverage by 0.2%. The diff coverage is 47.61%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master      #17      +/-   ##
==========================================
- Coverage   84.54%   84.34%   -0.21%     
==========================================
  Files          45       45              
  Lines        2893     2906      +13     
==========================================
+ Hits         2446     2451       +5     
- Misses        447      455       +8
Impacted Files Coverage Δ
...c/evaluating_rewards/analysis/gridworld_rewards.py 100% <ø> (ø) :arrow_up:
src/evaluating_rewards/analysis/visualize.py 81.86% <43.75%> (-2.24%) :arrow_down:
...uating_rewards/analysis/plot_divergence_heatmap.py 66% <50%> (-0.33%) :arrow_down:
...ting_rewards/analysis/plot_gridworld_divergence.py 94.04% <66.66%> (-1.08%) :arrow_down:

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update faa59d0...dad5a36. Read the comment docs.