Hacker News new | ask | show | jobs
by elandau25 1778 days ago
That's a good idea. It will likely be trickier to apply a method like this for text. Decomposition of the problem is less obvious than with vision tasks.