Model openai

Training model: gpt-4o-2024-11-20
Inference model: gpt-4o-mini

Investigations

Performance

DatasetValidation accuracyTest accuracy
espionage0.70.8
potions0.80.6
timetravel_insurance0.550.55
titanic0.68627450980392160.6078431372549019
wisconsin0.82539682539682540.7575757575757576
wisconsin0.8730158730158730.8636363636363636