{ "eval_source": "corpus_v2/eval.jsonl", "eval_mode": "held_out", "is_held_out": true, "eval_items": 54, "note": "Metrics reflect independent held-out evaluation.", "before": { "items": 54, "loss": 1.8596, "accuracy": 0.2407, "mean_confidence": 0.3148, "ece": 0.1869, "brier": 0.7831, "brier_top1": 0.1308, "by_type": { "choice": { "items": 54, "loss": 1.8596, "accuracy": 0.2407, "mean_confidence": 0.3148, "ece": 0.1869, "brier": 0.7831, "brier_top1": 0.1308 } } }, "after": { "items": 54, "loss": 0.6661, "accuracy": 0.8704, "mean_confidence": 0.9589, "ece": 0.0886, "brier": 0.2397, "brier_top1": 0.1096, "by_type": { "choice": { "items": 54, "loss": 0.6661, "accuracy": 0.8704, "mean_confidence": 0.9589, "ece": 0.0886, "brier": 0.2397, "brier_top1": 0.1096 } } }, "comparison": { "delta_accuracy": 0.6297, "delta_loss": -1.1935, "delta_ece": -0.0983, "delta_brier": -0.5434, "delta_brier_top1": -0.0212, "delta_mean_confidence": 0.6441 }, "calibration": { "choice": { "items": 400, "temperature": 3.9046413898468018, "issues": [], "accuracy": 0.9575, "mean_confidence": 0.9617928987797445 }, "score": { "items": 0, "temperature": 1.0, "issues": [] }, "noul": { "items": 0, "temperature": 1.0, "issues": [] } }, "abstention_thresholds": { "choice:6-10": 0.5204 }, "training": { "epochs": 4, "micro_batch": 2, "grad_accum": 32, "encoder_lr": 2.5e-05, "head_lr": 0.0001, "min_lr": 1e-06, "weight_decay": 0.01, "grad_clip": 1.0, "loss": "rlcd", "rl_samples": 4, "sigma_start": 0.4, "sigma_end": 0.1, "w_sph": 0.75, "w_rps": 1.0, "shuffle_options": [ "choice" ], "option_layout": null, "freeze_encoder": false, "calib_max": 400, "calib_frac": 0.1, "calib_seed": 20260922, "seed": 0, "max_len": 256, "head_max_len": null, "amp": null, "gradient_checkpointing": null, "log_every": 100, "label_smoothing": 0.0, "text_column": "text", "label_column": "label", "question_id": "label", "instructions": null, "eval_data": "corpus_v2/eval.jsonl", "target_error": 0.1, "min_abstain_n": 10 } }