News for hackers. Stuff that matters.

langkilde

As you should! :) Kudos for creating something!
I am really freaked out by the conversational performance displayed by the Flamingo Visual Language model recently described by Google DeepMind. It feels like there is a major shift in what ML models can do underway. Am I exaggerating the significance of this model or not? I want to discard it as another ML toy, but it feels much more significant this time...
Nice work @eyvindn! A few questions:

1. Am I just imagining things, or do the results depend on the speed?

2. I'm able to erase the shape very easily at max speed, despite setting it to persist. Have you studied how much shape loss is required for the shape to vanish? (ex video https://youtu.be/zMQkTyzdphc)