It would also be interesting to expand the benchmarking metrics to include other useful information, e.g. level of confidence on the predictions the agent is doing.
@evangriffiths @kongzii feel free to add more input here as to which metrics make the most sense.
It would also be interesting to expand the benchmarking metrics to include other useful information, e.g. level of confidence on the predictions the agent is doing.
@evangriffiths @kongzii feel free to add more input here as to which metrics make the most sense.