Checkpoint
One last thing before we move on. Same surface as a write step — but the lesson doesn't complete until this passes.
Print the full labeled matrix for this eval — headers exactly as in the read step (" predicted 1 predicted 0" / "actual 1 TP=... FN=..." / "actual 0 FP=... TN=..."), then the accuracy line, then the makeup-detector line: "positive row mostly misses:" with whether FN > TP. An 80% accuracy whose actual-1 row is mostly FN is the majority-class baseline in disguise.