logoalt Hacker News

stingraeyesterday at 6:13 PM0 repliesview on HN

the model is a set of weights, you can take a snapshot and test it. Reinforcement learning itself is largely testing and tuning.