| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.997580899624864, | |
| "global_step": 85500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0019883135247577978, | |
| "loss": 1.8357, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0019766270495155955, | |
| "loss": 0.5446, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0019649405742733933, | |
| "loss": 0.4222, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0019532540990311914, | |
| "loss": 0.3959, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.001941567623788989, | |
| "loss": 0.3926, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.001929881148546787, | |
| "loss": 0.3792, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0019181946733045848, | |
| "loss": 0.3692, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0019065081980623824, | |
| "loss": 0.3622, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0018948217228201803, | |
| "loss": 0.3643, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0018831352475779778, | |
| "loss": 0.3535, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0018714487723357758, | |
| "loss": 0.3642, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0018597622970935737, | |
| "loss": 0.3462, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0018480758218513713, | |
| "loss": 0.3508, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0018363893466091692, | |
| "loss": 0.3387, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0018247028713669672, | |
| "loss": 0.332, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0018130163961247647, | |
| "loss": 0.3429, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0018013299208825626, | |
| "loss": 0.3331, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0017896434456403606, | |
| "loss": 0.3313, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0017779569703981581, | |
| "loss": 0.3336, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.001766270495155956, | |
| "loss": 0.327, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.001754584019913754, | |
| "loss": 0.3274, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0017428975446715515, | |
| "loss": 0.3202, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0017312110694293495, | |
| "loss": 0.323, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0017195245941871475, | |
| "loss": 0.3146, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.001707838118944945, | |
| "loss": 0.3192, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.001696151643702743, | |
| "loss": 0.3235, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0016844651684605404, | |
| "loss": 0.3212, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0016727786932183384, | |
| "loss": 0.3048, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0016610922179761364, | |
| "loss": 0.3092, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0016494057427339339, | |
| "loss": 0.3054, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0016377192674917318, | |
| "loss": 0.3075, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0016260327922495298, | |
| "loss": 0.3143, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0016143463170073273, | |
| "loss": 0.2979, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0016026598417651253, | |
| "loss": 0.3036, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0015909733665229232, | |
| "loss": 0.2952, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0015792868912807207, | |
| "loss": 0.3052, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0015676004160385187, | |
| "loss": 0.3079, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0015559139407963166, | |
| "loss": 0.3024, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0015442274655541142, | |
| "loss": 0.3012, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.001532540990311912, | |
| "loss": 0.2966, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00152085451506971, | |
| "loss": 0.3007, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0015091680398275076, | |
| "loss": 0.2952, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0014974815645853055, | |
| "loss": 0.2933, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0014857950893431033, | |
| "loss": 0.2942, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.001474108614100901, | |
| "loss": 0.298, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.001462422138858699, | |
| "loss": 0.2969, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0014507356636164965, | |
| "loss": 0.2905, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0014390491883742944, | |
| "loss": 0.2926, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0014273627131320924, | |
| "loss": 0.2925, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00141567623788989, | |
| "loss": 0.2911, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0014039897626476879, | |
| "loss": 0.292, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0013923032874054858, | |
| "loss": 0.2962, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0013806168121632833, | |
| "loss": 0.289, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0013689303369210813, | |
| "loss": 0.2966, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0013572438616788792, | |
| "loss": 0.2871, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0013455573864366768, | |
| "loss": 0.2855, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0013338709111944747, | |
| "loss": 0.2823, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.2679596245288849, | |
| "eval_runtime": 1712.4301, | |
| "eval_samples_per_second": 148.053, | |
| "eval_steps_per_second": 6.169, | |
| "step": 28523 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0013221844359522727, | |
| "loss": 0.2827, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0013104979607100702, | |
| "loss": 0.2822, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0012988114854678681, | |
| "loss": 0.2789, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0012871250102256659, | |
| "loss": 0.274, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0012754385349834636, | |
| "loss": 0.2786, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0012637520597412616, | |
| "loss": 0.2728, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.001252065584499059, | |
| "loss": 0.2812, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.001240379109256857, | |
| "loss": 0.2763, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.001228692634014655, | |
| "loss": 0.2783, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0012170061587724525, | |
| "loss": 0.2767, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0012053196835302505, | |
| "loss": 0.2828, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0011936332082880484, | |
| "loss": 0.2765, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.001181946733045846, | |
| "loss": 0.2723, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.0011702602578036439, | |
| "loss": 0.2768, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0011585737825614418, | |
| "loss": 0.2729, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0011468873073192394, | |
| "loss": 0.2785, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0011352008320770373, | |
| "loss": 0.2765, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.001123514356834835, | |
| "loss": 0.2693, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.0011118278815926328, | |
| "loss": 0.266, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0011001414063504307, | |
| "loss": 0.2737, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0010884549311082285, | |
| "loss": 0.2774, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.0010767684558660262, | |
| "loss": 0.2702, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0010650819806238242, | |
| "loss": 0.2772, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.0010533955053816217, | |
| "loss": 0.2718, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0010417090301394196, | |
| "loss": 0.2777, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0010300225548972176, | |
| "loss": 0.2732, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.0010183360796550151, | |
| "loss": 0.2716, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.001006649604412813, | |
| "loss": 0.2728, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.0009949631291706108, | |
| "loss": 0.2792, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0009832766539284088, | |
| "loss": 0.2709, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0009715901786862065, | |
| "loss": 0.2664, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0009599037034440042, | |
| "loss": 0.2728, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.0009482172282018021, | |
| "loss": 0.2701, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.0009365307529595999, | |
| "loss": 0.2707, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0009248442777173977, | |
| "loss": 0.27, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.0009131578024751954, | |
| "loss": 0.2719, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0009014713272329933, | |
| "loss": 0.2682, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.0008897848519907911, | |
| "loss": 0.2675, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0008780983767485888, | |
| "loss": 0.2689, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.0008664119015063867, | |
| "loss": 0.2629, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0008547254262641845, | |
| "loss": 0.2636, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0008430389510219822, | |
| "loss": 0.2635, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.0008313524757797801, | |
| "loss": 0.2645, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.0008196660005375779, | |
| "loss": 0.2622, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0008079795252953757, | |
| "loss": 0.2627, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.0007962930500531734, | |
| "loss": 0.2576, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0007846065748109713, | |
| "loss": 0.2662, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.0007729200995687691, | |
| "loss": 0.2574, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.0007612336243265668, | |
| "loss": 0.2657, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0007495471490843647, | |
| "loss": 0.2597, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.0007378606738421625, | |
| "loss": 0.2529, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0007261741985999603, | |
| "loss": 0.2514, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.0007144877233577581, | |
| "loss": 0.2649, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.000702801248115556, | |
| "loss": 0.2621, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0006911147728733537, | |
| "loss": 0.2591, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.0006794282976311514, | |
| "loss": 0.2655, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0006677418223889493, | |
| "loss": 0.253, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.24398696422576904, | |
| "eval_runtime": 1707.3448, | |
| "eval_samples_per_second": 148.494, | |
| "eval_steps_per_second": 6.187, | |
| "step": 57046 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.0006560553471467471, | |
| "loss": 0.2498, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.0006443688719045449, | |
| "loss": 0.2548, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 0.0006326823966623427, | |
| "loss": 0.2552, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.0006209959214201405, | |
| "loss": 0.2537, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.0006093094461779383, | |
| "loss": 0.2466, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.000597622970935736, | |
| "loss": 0.2544, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0005859364956935339, | |
| "loss": 0.2527, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.0005742500204513317, | |
| "loss": 0.2471, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.0005625635452091294, | |
| "loss": 0.2549, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.0005508770699669273, | |
| "loss": 0.2455, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 0.0005391905947247251, | |
| "loss": 0.2487, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.0005275041194825229, | |
| "loss": 0.2438, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 0.0005158176442403207, | |
| "loss": 0.2501, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.0005041311689981185, | |
| "loss": 0.256, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.0004924446937559163, | |
| "loss": 0.2507, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 0.0004807582185137141, | |
| "loss": 0.2445, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.0004690717432715119, | |
| "loss": 0.2506, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.00045738526802930967, | |
| "loss": 0.2516, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.0004456987927871075, | |
| "loss": 0.2439, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 0.00043401231754490525, | |
| "loss": 0.2501, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.0004223258423027031, | |
| "loss": 0.2412, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 0.0004106393670605009, | |
| "loss": 0.2455, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 0.0003989528918182987, | |
| "loss": 0.2464, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 0.0003872664165760965, | |
| "loss": 0.2471, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.00037557994133389426, | |
| "loss": 0.2453, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 0.0003638934660916921, | |
| "loss": 0.2461, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 0.0003522069908494899, | |
| "loss": 0.2433, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 0.0003405205156072877, | |
| "loss": 0.253, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.00032883404036508553, | |
| "loss": 0.2444, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.00031714756512288327, | |
| "loss": 0.2446, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 0.0003054610898806811, | |
| "loss": 0.2454, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.0002937746146384789, | |
| "loss": 0.2487, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.0002820881393962767, | |
| "loss": 0.2407, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.00027040166415407454, | |
| "loss": 0.241, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.0002587151889118723, | |
| "loss": 0.2432, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 0.00024702871366967007, | |
| "loss": 0.2425, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 0.0002353422384274679, | |
| "loss": 0.2413, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0002236557631852657, | |
| "loss": 0.2468, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 0.0002119692879430635, | |
| "loss": 0.2391, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.00020028281270086128, | |
| "loss": 0.238, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 0.0001885963374586591, | |
| "loss": 0.2399, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.0001769098622164569, | |
| "loss": 0.2449, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.0001652233869742547, | |
| "loss": 0.2415, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.0001535369117320525, | |
| "loss": 0.2413, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 0.0001418504364898503, | |
| "loss": 0.2433, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.00013016396124764808, | |
| "loss": 0.2426, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 0.0001184774860054459, | |
| "loss": 0.2379, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 0.0001067910107632437, | |
| "loss": 0.2358, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.510453552104151e-05, | |
| "loss": 0.2455, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 8.34180602788393e-05, | |
| "loss": 0.2449, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.17315850366371e-05, | |
| "loss": 0.2399, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 6.00451097944349e-05, | |
| "loss": 0.2366, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.8358634552232706e-05, | |
| "loss": 0.2434, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.6672159310030504e-05, | |
| "loss": 0.2407, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.4985684067828304e-05, | |
| "loss": 0.2394, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.3299208825626102e-05, | |
| "loss": 0.2439, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.6127335834239036e-06, | |
| "loss": 0.2351, | |
| "step": 85500 | |
| } | |
| ], | |
| "max_steps": 85569, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.7771623363208806e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |