Remix.run Logo
kazinator 2 days ago

Ironically, you could use the sigmoid function instead of horses. The training stimulus slowly builds over multiple iteration and then suddenly, flip: the wrong prediction reverses.