A recent refactoring PR #67 introduced some bugs into eval_single_step.py, which evaded all of our CI and tests, and made the script no longer run. This PR is a quick (and somewhat dirty) fix to get the script to run again. Future PRs will refactor the source of the confusion that gave rise to these bugs, and also improve test coverage to avoid such issues in the future.
A recent refactoring PR #67 introduced some bugs into
eval_single_step.py
, which evaded all of our CI and tests, and made the script no longer run. This PR is a quick (and somewhat dirty) fix to get the script to run again. Future PRs will refactor the source of the confusion that gave rise to these bugs, and also improve test coverage to avoid such issues in the future.