This is my PR for multi-task eval (draft, not ready to be merged). Currently only the ReadMe is added, but I wanted to open this up for discussion with the rest of the team. I believe we need new functions for evaluating in multi-task and a different HDF5 structure that is centered around tasks. Feedback is welcome.
This is my PR for multi-task eval (draft, not ready to be merged). Currently only the ReadMe is added, but I wanted to open this up for discussion with the rest of the team. I believe we need new functions for evaluating in multi-task and a different HDF5 structure that is centered around tasks. Feedback is welcome.