{ "artifacts": "/home/simon/Projects/mverify/artifacts", "eval": { "n": 500, "accuracy": 0.86, "pass": { "precision": 0.4848, "recall": 0.9697, "f1": 0.6465, "tp": 64, "fp": 68, "fn": 2 }, "fail": { "precision": 0.9946, "recall": 0.8433, "f1": 0.9127, "tp": 366, "fp": 2, "fn": 68 }, "fpr_good": 0.0303, "blatant_fail_recall": 0.9563, "blatant_n": 183, "per_category": { "pass": { "n": 66, "accuracy": 0.9697, "fail_recall": null }, "off_topic": { "n": 62, "accuracy": 1.0, "fail_recall": 1.0 }, "empty": { "n": 62, "accuracy": 1.0, "fail_recall": 1.0 }, "truncated": { "n": 63, "accuracy": 0.9524, "fail_recall": 0.9524 }, "ignored_constraint": { "n": 59, "accuracy": 0.8644, "fail_recall": 0.8644 }, "wrong_format": { "n": 62, "accuracy": 0.8387, "fail_recall": 0.8387 }, "different_question": { "n": 63, "accuracy": 0.254, "fail_recall": 0.254 }, "confident_refusal": { "n": 63, "accuracy": 1.0, "fail_recall": 1.0 } }, "latency_ms": { "mean": 0.127, "p95": 0.217, "max": 2.349 }, "targets": { "fpr_good_lt": 0.05, "blatant_fail_recall_gt": 0.95, "p95_ms_lt": 200.0 }, "pass_targets": true }, "eval_bayes": { "n": 500, "accuracy": 0.794, "pass": { "precision": 0.2319, "recall": 0.2424, "f1": 0.237, "tp": 16, "fp": 53, "fn": 50 }, "fail": { "precision": 0.884, "recall": 0.8779, "f1": 0.8809, "tp": 381, "fp": 50, "fn": 53 }, "fpr_good": 0.7576, "blatant_fail_recall": 0.918, "blatant_n": 183, "per_category": { "pass": { "n": 66, "accuracy": 0.2424, "fail_recall": null }, "off_topic": { "n": 62, "accuracy": 1.0, "fail_recall": 1.0 }, "empty": { "n": 62, "accuracy": 0.9355, "fail_recall": 0.9355 }, "truncated": { "n": 63, "accuracy": 0.8254, "fail_recall": 0.8254 }, "ignored_constraint": { "n": 59, "accuracy": 0.8136, "fail_recall": 0.8136 }, "wrong_format": { "n": 62, "accuracy": 0.8065, "fail_recall": 0.8065 }, "different_question": { "n": 63, "accuracy": 0.7619, "fail_recall": 0.7619 }, "confident_refusal": { "n": 63, "accuracy": 1.0, "fail_recall": 1.0 } }, "latency_ms": { "mean": 0.0, "p95": 0.0, "max": 0.0 }, "targets": { "fpr_good_lt": 0.05, "blatant_fail_recall_gt": 0.95, "p95_ms_lt": 200.0 }, "pass_targets": false }, "test": { "n": 151, "accuracy": 0.8874, "pass": { "precision": 0.5294, "recall": 0.9474, "f1": 0.6792, "tp": 18, "fp": 16, "fn": 1 }, "fail": { "precision": 0.9915, "recall": 0.8788, "f1": 0.9317, "tp": 116, "fp": 1, "fn": 16 }, "fpr_good": 0.0526, "blatant_fail_recall": 0.9643, "blatant_n": 56, "per_category": { "pass": { "n": 19, "accuracy": 0.9474, "fail_recall": null }, "off_topic": { "n": 19, "accuracy": 0.9474, "fail_recall": 0.9474 }, "empty": { "n": 19, "accuracy": 1.0, "fail_recall": 1.0 }, "truncated": { "n": 19, "accuracy": 0.9474, "fail_recall": 0.9474 }, "ignored_constraint": { "n": 18, "accuracy": 0.9444, "fail_recall": 0.9444 }, "wrong_format": { "n": 19, "accuracy": 0.8421, "fail_recall": 0.8421 }, "different_question": { "n": 19, "accuracy": 0.4737, "fail_recall": 0.4737 }, "confident_refusal": { "n": 19, "accuracy": 1.0, "fail_recall": 1.0 } }, "latency_ms": { "mean": 0.087, "p95": 0.195, "max": 0.235 }, "targets": { "fpr_good_lt": 0.05, "blatant_fail_recall_gt": 0.95, "p95_ms_lt": 200.0 }, "pass_targets": false }, "gold": { "n": 48, "accuracy": 0.8333, "pass": { "precision": 0.6667, "recall": 0.7692, "f1": 0.7143, "tp": 10, "fp": 5, "fn": 3 }, "fail": { "precision": 0.9091, "recall": 0.8571, "f1": 0.8824, "tp": 30, "fp": 3, "fn": 5 }, "fpr_good": 0.2308, "blatant_fail_recall": 0.875, "blatant_n": 16, "per_category": { "pass": { "n": 13, "accuracy": 0.7692, "fail_recall": null }, "off_topic": { "n": 5, "accuracy": 0.8, "fail_recall": 0.8 }, "empty": { "n": 4, "accuracy": 1.0, "fail_recall": 1.0 }, "truncated": { "n": 4, "accuracy": 1.0, "fail_recall": 1.0 }, "ignored_constraint": { "n": 7, "accuracy": 0.8571, "fail_recall": 0.8571 }, "wrong_format": { "n": 5, "accuracy": 0.8, "fail_recall": 0.8 }, "different_question": { "n": 4, "accuracy": 0.5, "fail_recall": 0.5 }, "confident_refusal": { "n": 6, "accuracy": 1.0, "fail_recall": 1.0 } }, "latency_ms": { "mean": 0.083, "p95": 0.155, "max": 0.196 }, "targets": { "fpr_good_lt": 0.05, "blatant_fail_recall_gt": 0.95, "p95_ms_lt": 200.0 }, "pass_targets": false } }