| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 15.829463908822287, | |
| "global_step": 187500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.864000000000001e-06, | |
| "loss": 8.3245, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.5856e-05, | |
| "loss": 6.5687, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.3848e-05, | |
| "loss": 6.2288, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.184000000000001e-05, | |
| "loss": 6.0385, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.9832e-05, | |
| "loss": 5.8994, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.7824e-05, | |
| "loss": 5.8037, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.5816e-05, | |
| "loss": 5.7343, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.380800000000001e-05, | |
| "loss": 5.6803, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.18e-05, | |
| "loss": 5.6338, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.9792e-05, | |
| "loss": 5.6008, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.95614647887324e-05, | |
| "loss": 5.5729, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.911121126760565e-05, | |
| "loss": 5.5499, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.866095774647888e-05, | |
| "loss": 5.5315, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.821070422535212e-05, | |
| "loss": 5.5169, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.776045070422535e-05, | |
| "loss": 5.5034, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 7.73101971830986e-05, | |
| "loss": 5.4946, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.685994366197184e-05, | |
| "loss": 5.4869, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.640969014084507e-05, | |
| "loss": 5.4772, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 7.595943661971832e-05, | |
| "loss": 5.4702, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 7.550918309859156e-05, | |
| "loss": 5.4652, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.505892957746479e-05, | |
| "loss": 5.4591, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 7.460867605633804e-05, | |
| "loss": 5.4519, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 7.415842253521126e-05, | |
| "loss": 5.4488, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.370816901408451e-05, | |
| "loss": 5.4432, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 7.325791549295775e-05, | |
| "loss": 5.4395, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 7.2807661971831e-05, | |
| "loss": 5.422, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 7.235740845070423e-05, | |
| "loss": 5.0814, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.190715492957747e-05, | |
| "loss": 4.1945, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 7.145690140845072e-05, | |
| "loss": 2.9624, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.100664788732395e-05, | |
| "loss": 2.5772, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 7.055639436619719e-05, | |
| "loss": 2.3932, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 7.010614084507043e-05, | |
| "loss": 2.2725, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 6.965588732394366e-05, | |
| "loss": 2.1769, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 6.920563380281691e-05, | |
| "loss": 2.1027, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 6.875538028169015e-05, | |
| "loss": 2.0415, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.830512676056338e-05, | |
| "loss": 1.9906, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.785487323943663e-05, | |
| "loss": 1.9475, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 6.740461971830987e-05, | |
| "loss": 1.9084, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 6.69543661971831e-05, | |
| "loss": 1.8699, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 6.650411267605634e-05, | |
| "loss": 1.839, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 6.605385915492959e-05, | |
| "loss": 1.812, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 6.560360563380282e-05, | |
| "loss": 1.7872, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 6.515335211267606e-05, | |
| "loss": 1.7652, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 6.47030985915493e-05, | |
| "loss": 1.7445, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 6.425284507042254e-05, | |
| "loss": 1.726, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 6.380259154929578e-05, | |
| "loss": 1.7086, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.335233802816903e-05, | |
| "loss": 1.6932, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.290208450704226e-05, | |
| "loss": 1.6773, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 6.24518309859155e-05, | |
| "loss": 1.6641, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 6.200157746478873e-05, | |
| "loss": 1.6506, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 6.155132394366198e-05, | |
| "loss": 1.6375, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 6.110107042253522e-05, | |
| "loss": 1.6268, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 6.0650816901408453e-05, | |
| "loss": 1.6168, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 6.0200563380281696e-05, | |
| "loss": 1.6061, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 5.975030985915493e-05, | |
| "loss": 1.5971, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 5.9300056338028174e-05, | |
| "loss": 1.5872, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 5.8849802816901416e-05, | |
| "loss": 1.5787, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 5.839954929577465e-05, | |
| "loss": 1.5703, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 5.7949295774647894e-05, | |
| "loss": 1.5626, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 5.7499042253521136e-05, | |
| "loss": 1.5536, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 5.7048788732394365e-05, | |
| "loss": 1.5468, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 5.659853521126761e-05, | |
| "loss": 1.5387, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 5.614828169014085e-05, | |
| "loss": 1.5322, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 5.569802816901409e-05, | |
| "loss": 1.5254, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 5.524777464788733e-05, | |
| "loss": 1.5206, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 5.479752112676057e-05, | |
| "loss": 1.513, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 5.434726760563381e-05, | |
| "loss": 1.5089, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 5.389701408450704e-05, | |
| "loss": 1.5033, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 5.3446760563380284e-05, | |
| "loss": 1.4973, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 5.2996507042253526e-05, | |
| "loss": 1.4928, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 5.254625352112676e-05, | |
| "loss": 1.4864, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 5.2096000000000004e-05, | |
| "loss": 1.4815, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 5.1645746478873246e-05, | |
| "loss": 1.478, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 5.119549295774648e-05, | |
| "loss": 1.4724, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 5.0745239436619724e-05, | |
| "loss": 1.4692, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 5.029498591549297e-05, | |
| "loss": 1.4632, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 4.9844732394366195e-05, | |
| "loss": 1.4605, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 4.939447887323944e-05, | |
| "loss": 1.4567, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 4.894422535211268e-05, | |
| "loss": 1.4521, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 4.8493971830985916e-05, | |
| "loss": 1.4484, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 4.804371830985916e-05, | |
| "loss": 1.4447, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 4.75934647887324e-05, | |
| "loss": 1.4404, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 4.714321126760564e-05, | |
| "loss": 1.4375, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 4.669295774647888e-05, | |
| "loss": 1.4346, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 4.624270422535212e-05, | |
| "loss": 1.4301, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 4.5792450704225356e-05, | |
| "loss": 1.428, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 4.534219718309859e-05, | |
| "loss": 1.4242, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 4.4891943661971834e-05, | |
| "loss": 1.4227, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 4.4441690140845077e-05, | |
| "loss": 1.4177, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 4.399143661971831e-05, | |
| "loss": 1.416, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 4.3541183098591555e-05, | |
| "loss": 1.4134, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 4.30909295774648e-05, | |
| "loss": 1.4099, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 4.2640676056338026e-05, | |
| "loss": 1.4077, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 4.219042253521127e-05, | |
| "loss": 1.4056, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 4.174016901408451e-05, | |
| "loss": 1.4028, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 4.1289915492957746e-05, | |
| "loss": 1.4, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 4.083966197183099e-05, | |
| "loss": 1.3985, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 4.038940845070423e-05, | |
| "loss": 1.3956, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 3.9939154929577466e-05, | |
| "loss": 1.3942, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 3.948890140845071e-05, | |
| "loss": 1.3915, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 8.53, | |
| "learning_rate": 3.903864788732395e-05, | |
| "loss": 1.3893, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 3.8588394366197187e-05, | |
| "loss": 1.3872, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 3.813814084507042e-05, | |
| "loss": 1.3847, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 8.78, | |
| "learning_rate": 3.7687887323943664e-05, | |
| "loss": 1.3825, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 3.723763380281691e-05, | |
| "loss": 1.3807, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 3.678738028169015e-05, | |
| "loss": 1.3793, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 3.6337126760563385e-05, | |
| "loss": 1.3765, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 3.588687323943662e-05, | |
| "loss": 1.3754, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 3.543661971830986e-05, | |
| "loss": 1.3738, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 3.49863661971831e-05, | |
| "loss": 1.3711, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 3.453611267605634e-05, | |
| "loss": 1.3698, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 3.408585915492958e-05, | |
| "loss": 1.3677, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 3.363560563380282e-05, | |
| "loss": 1.3668, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 3.3185352112676054e-05, | |
| "loss": 1.3649, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 3.2735098591549296e-05, | |
| "loss": 1.3639, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 3.228484507042254e-05, | |
| "loss": 1.3602, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 3.183459154929578e-05, | |
| "loss": 1.3609, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 3.138433802816902e-05, | |
| "loss": 1.3591, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "learning_rate": 3.093408450704225e-05, | |
| "loss": 1.3573, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 10.13, | |
| "learning_rate": 3.0483830985915498e-05, | |
| "loss": 1.3562, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 10.22, | |
| "learning_rate": 3.0033577464788734e-05, | |
| "loss": 1.3537, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 10.3, | |
| "learning_rate": 2.9583323943661973e-05, | |
| "loss": 1.3532, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 10.38, | |
| "learning_rate": 2.9133070422535215e-05, | |
| "loss": 1.351, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 10.47, | |
| "learning_rate": 2.8682816901408454e-05, | |
| "loss": 1.35, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 10.55, | |
| "learning_rate": 2.8232563380281693e-05, | |
| "loss": 1.3485, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 2.7782309859154932e-05, | |
| "loss": 1.3484, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 2.733205633802817e-05, | |
| "loss": 1.3466, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 2.6881802816901413e-05, | |
| "loss": 1.3473, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 10.89, | |
| "learning_rate": 2.643154929577465e-05, | |
| "loss": 1.3429, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 2.5981295774647888e-05, | |
| "loss": 1.343, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 11.06, | |
| "learning_rate": 2.553104225352113e-05, | |
| "loss": 1.3412, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 11.14, | |
| "learning_rate": 2.508078873239437e-05, | |
| "loss": 1.3409, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 11.23, | |
| "learning_rate": 2.4630535211267605e-05, | |
| "loss": 1.3381, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 11.31, | |
| "learning_rate": 2.4180281690140847e-05, | |
| "loss": 1.3384, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 2.3730028169014086e-05, | |
| "loss": 1.3371, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 11.48, | |
| "learning_rate": 2.327977464788733e-05, | |
| "loss": 1.3365, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 11.57, | |
| "learning_rate": 2.2829521126760564e-05, | |
| "loss": 1.3347, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 11.65, | |
| "learning_rate": 2.2379267605633803e-05, | |
| "loss": 1.3342, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 11.73, | |
| "learning_rate": 2.1929014084507045e-05, | |
| "loss": 1.3335, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 11.82, | |
| "learning_rate": 2.1478760563380284e-05, | |
| "loss": 1.3317, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 11.9, | |
| "learning_rate": 2.102850704225352e-05, | |
| "loss": 1.3307, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 11.99, | |
| "learning_rate": 2.0578253521126762e-05, | |
| "loss": 1.3302, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 12.07, | |
| "learning_rate": 2.0128e-05, | |
| "loss": 1.329, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "learning_rate": 1.967774647887324e-05, | |
| "loss": 1.3284, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 12.24, | |
| "learning_rate": 1.9227492957746482e-05, | |
| "loss": 1.3278, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 12.33, | |
| "learning_rate": 1.877723943661972e-05, | |
| "loss": 1.3257, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "learning_rate": 1.8326985915492957e-05, | |
| "loss": 1.3256, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 12.49, | |
| "learning_rate": 1.78767323943662e-05, | |
| "loss": 1.3255, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 12.58, | |
| "learning_rate": 1.7426478873239438e-05, | |
| "loss": 1.3241, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 12.66, | |
| "learning_rate": 1.6976225352112677e-05, | |
| "loss": 1.3241, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "learning_rate": 1.6525971830985916e-05, | |
| "loss": 1.3223, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 12.83, | |
| "learning_rate": 1.6075718309859155e-05, | |
| "loss": 1.3214, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 12.92, | |
| "learning_rate": 1.5625464788732398e-05, | |
| "loss": 1.3218, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 1.5175211267605635e-05, | |
| "loss": 1.3207, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 13.09, | |
| "learning_rate": 1.4724957746478874e-05, | |
| "loss": 1.3195, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 13.17, | |
| "learning_rate": 1.4274704225352114e-05, | |
| "loss": 1.32, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 13.25, | |
| "learning_rate": 1.3824450704225353e-05, | |
| "loss": 1.3186, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 13.34, | |
| "learning_rate": 1.3374197183098592e-05, | |
| "loss": 1.3161, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 13.42, | |
| "learning_rate": 1.2923943661971831e-05, | |
| "loss": 1.317, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 13.51, | |
| "learning_rate": 1.2473690140845072e-05, | |
| "loss": 1.3164, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 13.59, | |
| "learning_rate": 1.2023436619718311e-05, | |
| "loss": 1.3155, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 13.68, | |
| "learning_rate": 1.157318309859155e-05, | |
| "loss": 1.3142, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "learning_rate": 1.1122929577464789e-05, | |
| "loss": 1.3146, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 1.067267605633803e-05, | |
| "loss": 1.3138, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 1.022242253521127e-05, | |
| "loss": 1.3152, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 9.772169014084507e-06, | |
| "loss": 1.3123, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "learning_rate": 9.321915492957746e-06, | |
| "loss": 1.3113, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 14.18, | |
| "learning_rate": 8.871661971830987e-06, | |
| "loss": 1.3114, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 14.27, | |
| "learning_rate": 8.421408450704226e-06, | |
| "loss": 1.3103, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 14.35, | |
| "learning_rate": 7.971154929577467e-06, | |
| "loss": 1.3109, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 14.44, | |
| "learning_rate": 7.520901408450705e-06, | |
| "loss": 1.3107, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 14.52, | |
| "learning_rate": 7.070647887323944e-06, | |
| "loss": 1.3096, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 14.61, | |
| "learning_rate": 6.620394366197184e-06, | |
| "loss": 1.309, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 14.69, | |
| "learning_rate": 6.170140845070423e-06, | |
| "loss": 1.3094, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 14.77, | |
| "learning_rate": 5.719887323943662e-06, | |
| "loss": 1.3085, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 14.86, | |
| "learning_rate": 5.269633802816901e-06, | |
| "loss": 1.3076, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 4.819380281690141e-06, | |
| "loss": 1.3078, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 4.369126760563381e-06, | |
| "loss": 1.3066, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 15.11, | |
| "learning_rate": 3.91887323943662e-06, | |
| "loss": 1.3062, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 3.4686197183098598e-06, | |
| "loss": 1.3054, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 15.28, | |
| "learning_rate": 3.018366197183099e-06, | |
| "loss": 1.3051, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 15.37, | |
| "learning_rate": 2.568112676056338e-06, | |
| "loss": 1.3052, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "learning_rate": 2.1178591549295775e-06, | |
| "loss": 1.3048, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 15.53, | |
| "learning_rate": 1.6676056338028171e-06, | |
| "loss": 1.3056, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 15.62, | |
| "learning_rate": 1.2173521126760563e-06, | |
| "loss": 1.3057, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 15.7, | |
| "learning_rate": 7.670985915492958e-07, | |
| "loss": 1.3052, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 3.1684507042253523e-07, | |
| "loss": 1.3041, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 15.83, | |
| "step": 187500, | |
| "total_flos": 1.0106837180325298e+20, | |
| "train_loss": 0.12197015299479166, | |
| "train_runtime": 8235.0921, | |
| "train_samples_per_second": 46629.715, | |
| "train_steps_per_second": 22.768 | |
| } | |
| ], | |
| "max_steps": 187500, | |
| "num_train_epochs": 16, | |
| "total_flos": 1.0106837180325298e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |