search

juliusfrost / cfrl-rllib

5 stars 0 forks source link

Gen new eval vids #77

Closed eweiner closed 3 years ago

eweiner commented 3 years ago

Code for new eval task. Change setting in explain.py to "eval_task": "guess_successful", or "original" to turn off

juliusfrost commented 3 years ago

Could you fill out the stubs created by this commit?

juliusfrost commented 3 years ago

Additionally, we need the eval policy order to not be random in the performance selection task. This is because the user always selects the policies in order in the user study.