@citnaj Sorry for what is likely a naive question, but when you train this, you convert ImageNet images to grayscale to provide training examples?
-
If you don't mind just ONE more Q, does it depend on a pixel for pixel match? If instead of converting to grayscale, I transformed imagenet into ASCII art and then trained, is there any chance of making a model that would do something interesting when given ascii art as input?
So you're wanting to convert ascii art back into something that more closely resembles the original images? You could certainly do that and it should work at least to a certain extent. Might be interesting!
-
-
Awesome, thanks, I'll add it to my list. I eventually want to make full movies with GPT-2 text output, but in shorter term, maybe try rendering GPT-2 ascii art.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.