4 ms·
I'm surprised that this few samples made such a big difference. Perhaps the learning rate is jacked up? Otherwise, the last couple of batches during regular tra
by Scene_Cast2 3y ago
I'm surprised that this few samples made such a big difference. Perhaps the learning rate is jacked up? Otherwise, the last couple of batches during regular training would set the overall model tone.
- 3abiton 3y agoOne thing they did not test was the complexity of the task. I doubt with "hundred" samples it would be possible to get it to be that accurately usable. But experiments are needed to back up that claim.