promptdojo_

Mission: baseline model showdown — step 7 of 7

Checkpoint

One last thing before we move on. Same surface as a write step — but the lesson doesn't complete until this passes.

Run the showdown's trained rung end to end. Fit: sweep thresholds 0-7 and pick the one maximizing accuracy on TRAIN (max with a key). Score: majority (always 0) and the learned threshold rule, both on the same test set. Print the two board lines, then the gap line for the learned rung: f"learned rung: train={...:.0%} test={...:.0%} gap={...:+.0%}".

full-screen editor opens — close anytime to keep reading.