File size: 356 Bytes
5096607 |
1 2 3 4 5 |
we manually tuned the checkpoints here, and then noted the procedure approximately in training.py
the procedure in training.py reaches ~53.7% whereas the saved models are 54.7%. The difference is reproducible, yet we didn't want to spend time on it since this model will most probably only be reported as activity rather than result
not on deliverable v1 |