A recorded run

Wine

Asked: Predict cultivar (one of 3 kinds) from 13 columns of 178 rows.
What happened: 15 ideas tried, 2 kept. It stopped on its own when progress levelled off.
Result: 0.0230Log-loss on data it never saw0.0062 better than its own checks predicted, so it didn't fool itself.

Try your own data

Recorded without an LLM: ideas come from a built-in search.

Scroll to follow the climb, idea by idea

For the curious

The full record

Every number the run recorded, exactly as measured.