Hacker News new | ask | show | jobs
by readhistory 417 days ago
In ML, it's pretty classic actually. You train on one set, and evaluate on another set. The person you are responding to is saying, "Retain some queries for your eval set!"