| {"global_step": 524, "max_steps": 524, "reached_max_steps": true, "epoch": 2.0, "train_s": 2938.7, "best_step": 524, "best_dev": 0.7163519859313965, "dev_curve": [[224, 0.73056], [256, 0.72731], [261, 0.72756], [288, 0.73088], [320, 0.73078], [352, 0.73097], [384, 0.72852], [416, 0.72513], [448, 0.71996], [480, 0.71891], [500, 0.71863], [512, 0.71856], [522, 0.71843], [524, 0.71836], [524, 0.71836], [524, 0.71846], [524, 0.71856], [524, 0.71854], [524, 0.71878], [524, 0.71878], [524, 0.72028], [524, 0.71734], [524, 0.71871], [524, 0.71762], [524, 0.71805], [524, 0.71811], [524, 0.71635]], "packed": true, "length_grouped": false, "max_length": 2176, "per_device_bs": 41, "grad_accum": 1, "world_size": 1, "train_items": 10727, "devpass": true, "kept_prior": false, "pass_seed": "42", "one_epoch_pass": true, "group": {"arch": "dense", "params_b": 2.517, "length_class": "long", "completion_p50": 354, "seq_p50": 462, "seq_p99": 1935, "num_records": 38882, "total_tokens": 20908270, "hours": 1.25, "epochs_est": 1.49, "epochs_measured": 4.238, "aug_marks": {"matrices": 294, "zero_frac_max": 0.0, "zero_frac_mean": 0.0, "pruned_frac": 0.0, "rms_scaled_frac": 0.011, "rms_log2_dev_max": 0.966, "detector": "neighbour-v2"}, "aug_guess": "none-or-weak"}} |