{ "device": "cuda", "math_check": { "max_abs_derivative_error": 1.013873429656087e-09, "prior_sum": 0.9999999999999999, "prior_endpoint_mass": 0.33637619018554665, "prior": [ 0.16818809509277333, 0.08809852600097658, 0.0695514678955081, 0.06136894226074231, 0.057277679443359375, 0.05551528930664039, 0.05551528930664034, 0.057277679443359375, 0.06136894226074226, 0.06955146789550798, 0.08809852600097658, 0.16818809509277333 ] }, "baseline_continuous": { "accuracy": 0.8755555748939514, "sampled_count_accuracy": null, "test_I_nats": 0.9117828607559204, "test_KL_to_beta_binomial": 0.2944762110710144, "test_endpoint_mass": 0.6830476522445679, "train_ce_checkpoints": [ 0.7499760571858931, 0.34990795623751686 ], "activation_storage_bits_per_coordinate": 32.0 }, "capacity_shaped_binomial": { "accuracy": 0.897777795791626, "sampled_count_accuracy": 0.8955555558204651, "test_I_nats": 0.9857069849967957, "test_KL_to_beta_binomial": 0.09662292897701263, "test_endpoint_mass": 0.5673930644989014, "train_ce_checkpoints": [ 0.8077772284934674, 0.2767513006394584 ], "activation_storage_bits_per_coordinate": 3.169925001442312 } }