Comment by ac29

5 hours ago

Doesnt seem suspect to me, training runs have checkpoints and there is no reason you cant release a checkpoint even if you are still training the model

0 comments