DAINAPP
patreon
So you guys don't think i'm pulling your leg.
🕑 Added 2022-03-11 17:07:58 +0000 UTCThe devil really are in the details. Slowly been managing to improve the model, but each day there is something new to learn, how batch affect the model and what is the best values for the training.
The models that I have trained so far are pretty good, but it's seen that it only that the right value so they can become great. The test last night manage to make the model work a lot better for bigger motion, but it removed some of the details. So I'm trying something today that will keep the motion and details.
At least now I'm managing to train the RL model along the anime one. So both should be ready together.
After that (and the build release) I still need to make a new Dataset for RL, because the current datset is really limiting the RL model.
Comments
the AC
Indeed. Power on.
GRisk
At the same time it's great to learn how each change affect the model. This is stuff that I can use on all my other projects.
the AC
I've run into stuff like that. Make small improvements that fix the thing you're focused on but it breaks a bunch of others inadvertently. So you have to decide which fix is more important, if you can even reverse it.
