{ "method": "single temperature minimizing pooled validation negative log-likelihood", "temperature": 12.595144782442853, "checkpoint_sha256": "e009fe101e9846279b26be6b292a33f5672894a1b319bd1210a6b71632ea5dee", "fit_questions": 832, "held_out_eval_questions": 832, "held_out_eval_raw": { "accuracy": 0.6538461538461539, "mean_chosen_probability": 0.9124184985473408, "nll": 6.075739293144633, "brier": 0.5495115861270571, "probability_ece": 0.2618336556889736 }, "held_out_eval_calibrated": { "accuracy": 0.6538461538461539, "mean_chosen_probability": 0.6008282562788685, "nll": 1.1272864350823428, "brier": 0.45237775773126754, "probability_ece": 0.15262696983052562 } }