{ "model": "rnn", "seed": 0, "params": null, "val_selected": { "ckpt": "best.pt", "params": 36620, "val_token_acc": 0.7045454545454546, "val_exact_match": 0.36666666666666664, "note": "val-selected checkpoint (early stop / best-on-val per prereg \u2014 val EM here is selection-holed, not an untouched test estimate" }, "final": { "ckpt": "last.pt", "params": 36620, "val_token_acc": 0.5909090909090909, "val_exact_match": 0.16666666666666666, "note": "final checkpoint, no val selection", "probe": { "code": "P4", "acc": 0.01, "correct": 1, "total": 100, "errors": [ { "n": 101, "target": 103, "pred": 101 }, { "n": 102, "target": 103, "pred": 199 }, { "n": 103, "target": 107, "pred": 101 }, { "n": 104, "target": 107, "pred": 119 }, { "n": 105, "target": 107, "pred": 199 }, { "n": 106, "target": 107, "pred": 101 }, { "n": 107, "target": 109, "pred": 199 }, { "n": 109, "target": 113, "pred": 899 }, { "n": 110, "target": 113, "pred": 101 }, { "n": 111, "target": 113, "pred": 1 }, { "n": 112, "target": 113, "pred": 99 }, { "n": 113, "target": 127, "pred": 10 }, { "n": 114, "target": 127, "pred": 19 }, { "n": 115, "target": 127, "pred": 99 }, { "n": 116, "target": 127, "pred": 19 }, { "n": 117, "target": 127, "pred": 19 }, { "n": 118, "target": 127, "pred": 1 }, { "n": 119, "target": 127, "pred": 899 }, { "n": 120, "target": 127, "pred": 199 }, { "n": 121, "target": 127, "pred": 999 }, { "n": 122, "target": 127, "pred": 299 }, { "n": 123, "target": 127, "pred": 299 }, { "n": 124, "target": 127, "pred": 299 }, { "n": 125, "target": 127, "pred": 999 }, { "n": 126, "target": 127, "pred": 997 }, { "n": 127, "target": 131, "pred": 999 }, { "n": 128, "target": 131, "pred": 999 }, { "n": 129, "target": 131, "pred": 299 }, { "n": 130, "target": 131, "pred": 101 }, { "n": 131, "target": 137, "pred": 10 }, { "n": 132, "target": 137, "pred": 299 }, { "n": 133, "target": 137, "pred": 899 }, { "n": 134, "target": 137, "pred": 899 }, { "n": 135, "target": 137, "pred": 899 }, { "n": 136, "target": 137, "pred": 19 }, { "n": 137, "target": 139, "pred": 899 }, { "n": 138, "target": 139, "pred": 899 }, { "n": 139, "target": 149, "pred": 899 }, { "n": 140, "target": 149, "pred": 119 }, { "n": 141, "target": 149, "pred": 19 }, { "n": 142, "target": 149, "pred": 899 }, { "n": 143, "target": 149, "pred": 899 }, { "n": 144, "target": 149, "pred": 899 }, { "n": 145, "target": 149, "pred": 899 }, { "n": 146, "target": 149, "pred": 199 }, { "n": 147, "target": 149, "pred": 899 }, { "n": 148, "target": 149, "pred": 899 }, { "n": 149, "target": 151, "pred": 899 }, { "n": 150, "target": 151, "pred": 893 }, { "n": 151, "target": 157, "pred": 99 }, { "n": 152, "target": 157, "pred": 999 }, { "n": 153, "target": 157, "pred": 899 }, { "n": 154, "target": 157, "pred": 899 }, { "n": 155, "target": 157, "pred": 899 }, { "n": 156, "target": 157, "pred": 99 }, { "n": 157, "target": 163, "pred": 999 }, { "n": 158, "target": 163, "pred": 899 }, { "n": 159, "target": 163, "pred": 899 }, { "n": 160, "target": 163, "pred": 101 }, { "n": 161, "target": 163, "pred": 19 }, { "n": 162, "target": 163, "pred": 997 }, { "n": 163, "target": 167, "pred": 199 }, { "n": 164, "target": 167, "pred": 199 }, { "n": 165, "target": 167, "pred": 99 }, { "n": 166, "target": 167, "pred": 199 }, { "n": 167, "target": 173, "pred": 99 }, { "n": 168, "target": 173, "pred": 99 }, { "n": 169, "target": 173, "pred": 899 }, { "n": 170, "target": 173, "pred": 199 }, { "n": 171, "target": 173, "pred": 99 }, { "n": 172, "target": 173, "pred": 999 }, { "n": 173, "target": 179, "pred": 99 }, { "n": 174, "target": 179, "pred": 99 }, { "n": 175, "target": 179, "pred": 999 }, { "n": 176, "target": 179, "pred": 99 }, { "n": 177, "target": 179, "pred": 999 }, { "n": 178, "target": 179, "pred": 999 }, { "n": 179, "target": 181, "pred": 899 }, { "n": 180, "target": 181, "pred": 893 }, { "n": 181, "target": 191, "pred": 9 }, { "n": 182, "target": 191, "pred": 999 }, { "n": 183, "target": 191, "pred": 899 }, { "n": 184, "target": 191, "pred": 899 }, { "n": 185, "target": 191, "pred": 899 }, { "n": 186, "target": 191, "pred": 999 }, { "n": 187, "target": 191, "pred": 999 }, { "n": 188, "target": 191, "pred": 899 }, { "n": 189, "target": 191, "pred": 899 }, { "n": 190, "target": 191, "pred": 899 }, { "n": 191, "target": 193, "pred": 899 }, { "n": 192, "target": 193, "pred": 299 }, { "n": 193, "target": 197, "pred": 899 }, { "n": 194, "target": 197, "pred": 899 }, { "n": 195, "target": 197, "pred": 899 }, { "n": 196, "target": 197, "pred": 899 }, { "n": 197, "target": 199, "pred": 899 }, { "n": 198, "target": 199, "pred": 899 }, { "n": 199, "target": 211, "pred": 899 }, { "n": 200, "target": 211, "pred": 199 } ], "flagged_errors": [], "flagged_pred_fraction": 0.0, "distinct_flagged_preds": 0, "easy_total": 60, "easy_wrong": 59, "easy_err_rate": 0.9833333333333333 }, "halting": { "code": "H4", "mean_steps": 3.123047472250582, "corr_gap": 0.2214679834807299, "min_steps": 2, "max_steps": 20, "gaps": [ 1, 2, 1, 2, 1, 4, 3, 2, 1, 2, 1, 4, 3, 2, 1, 2, 1, 4, 3, 2, 1, 6, 5, 4, 3, 2, 1, 2, 1, 6, 5, 4, 3, 2, 1, 4, 3, 2, 1, 2, 1, 4, 3, 2, 1, 6, 5, 4, 3, 2, 1, 6, 5, 4, 3, 2, 1, 2, 1, 6, 5, 4, 3, 2, 1, 4, 3, 2, 1, 2, 1, 6, 5, 4, 3, 2, 1, 4, 3, 2, 1, 6, 5, 4, 3, 2, 1, 8, 7, 6, 5, 4, 3, 2, 1, 4, 3, 2, 1 ], "steps_by_n": [ 3.014195680618286, 3.135343313217163, 3.090144634246826, 3.1713850498199463, 2.900202989578247, 3.0106613636016846, 3.2104332447052, 3.325666666030884, 2.9300780296325684, 2.890700578689575, 2.9069631099700928, 2.9783923625946045, 2.9744765758514404, 3.008406162261963, 2.8595876693725586, 2.9244143962860107, 3.0445709228515625, 3.0422301292419434, 2.99932861328125, 2.9410030841827393, 3.0114541053771973, 3.107120990753174, 3.0779876708984375, 3.136584520339966, 2.921391010284424, 3.0053720474243164, 3.1654398441314697, 3.233349561691284, 3.1324732303619385, 3.0467514991760254, 3.1338253021240234, 3.2645983695983887, 3.2252910137176514, 3.281956672668457, 3.031636953353882, 3.1119115352630615, 3.2983484268188477, 3.3865015506744385, 3.0938613414764404, 3.022742986679077, 3.0877559185028076, 3.202849864959717, 3.175102472305298, 3.2327229976654053, 3.0024092197418213, 3.0831265449523926, 3.2592060565948486, 3.3100247383117676, 3.1473474502563477, 3.071608066558838, 3.163006544113159, 3.2788093090057373, 3.24808406829834, 3.296496868133545, 3.0659124851226807, 3.1418488025665283, 3.3155531883239746, 3.416996717453003, 2.898960828781128, 2.858566999435425, 2.882908344268799, 2.96085524559021, 2.9518916606903076, 2.9972848892211914, 2.83113169670105, 2.901516914367676, 3.0369534492492676, 3.0480051040649414, 2.9847333431243896, 2.941075325012207, 2.98722767829895, 3.0637786388397217, 3.0518579483032227, 3.094906806945801, 2.921523094177246, 2.9897327423095703, 3.1205570697784424, 3.164365768432617, 3.1843652725219727, 3.113823175430298, 3.196122884750366, 3.2959306240081787, 3.2752671241760254, 3.3164381980895996, 3.111222743988037, 3.1724257469177246, 3.317817449569702, 3.412120819091797, 3.267571210861206, 3.1642038822174072, 3.3226613998413086, 3.455141305923462, 3.400142192840576, 3.4917120933532715, 3.1787307262420654, 3.2746527194976807, 3.4824793338775635, 3.616750955581665, 2.89864182472229 ] }, "per_example": [ { "n": "3", "target": 5, "pred": 9, "ok": false }, { "n": "4", "target": 5, "pred": 7, "ok": false }, { "n": "8", "target": 11, "pred": 7, "ok": false }, { "n": "9", "target": 11, "pred": 1, "ok": false }, { "n": "10", "target": 11, "pred": 11, "ok": true }, { "n": "12", "target": 13, "pred": 23, "ok": false }, { "n": "13", "target": 17, "pred": 41, "ok": false }, { "n": "15", "target": 17, "pred": 37, "ok": false }, { "n": "18", "target": 19, "pred": 33, "ok": false }, { "n": "21", "target": 23, "pred": 23, "ok": true }, { "n": "24", "target": 29, "pred": 23, "ok": false }, { "n": "25", "target": 29, "pred": 23, "ok": false }, { "n": "26", "target": 29, "pred": 67, "ok": false }, { "n": "31", "target": 37, "pred": 41, "ok": false }, { "n": "32", "target": 37, "pred": 29, "ok": false }, { "n": "39", "target": 41, "pred": 97, "ok": false }, { "n": "43", "target": 47, "pred": 37, "ok": false }, { "n": "49", "target": 53, "pred": 79, "ok": false }, { "n": "50", "target": 53, "pred": 41, "ok": false }, { "n": "54", "target": 59, "pred": 53, "ok": false }, { "n": "56", "target": 59, "pred": 57, "ok": false }, { "n": "61", "target": 67, "pred": 67, "ok": true }, { "n": "65", "target": 67, "pred": 67, "ok": true }, { "n": "68", "target": 71, "pred": 79, "ok": false }, { "n": "78", "target": 79, "pred": 89, "ok": false }, { "n": "82", "target": 83, "pred": 29, "ok": false }, { "n": "84", "target": 89, "pred": 59, "ok": false }, { "n": "88", "target": 89, "pred": 89, "ok": true }, { "n": "92", "target": 97, "pred": 31, "ok": false }, { "n": "95", "target": 97, "pred": 61, "ok": false } ] }, "signature": { "code": "O4", "sat_start_eval": null, "val_hi_eval_idx": null, "transition_width_evals": null, "n_evals": 1000, "final_train_em": 0.7971014492753623, "final_val_em": 0.16666666666666666, "max_train_em": 1.0 } }