Thread

I built non-autoregressive decision models with RL a year ago

1363 points · 319 comments · nandakishor_ml

krenerd · 1789942822 · focus · HN ↗
both RL part and Calibration honestly doesn't make sense to me.
why does this use rl at all? i don't see why just CE cant resolve it? and why is it calibrated?

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.