|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 127.03252032520325, |
|
"global_step": 187500, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 7.864000000000001e-06, |
|
"loss": 8.3573, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 1.5856e-05, |
|
"loss": 6.6355, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 2.3848e-05, |
|
"loss": 6.2822, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 2.71, |
|
"learning_rate": 3.184000000000001e-05, |
|
"loss": 6.0633, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 3.39, |
|
"learning_rate": 3.9832e-05, |
|
"loss": 5.9146, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 4.07, |
|
"learning_rate": 4.7824e-05, |
|
"loss": 5.8117, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 4.74, |
|
"learning_rate": 5.5816e-05, |
|
"loss": 5.7342, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 5.42, |
|
"learning_rate": 6.380800000000001e-05, |
|
"loss": 5.6725, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 6.1, |
|
"learning_rate": 7.18e-05, |
|
"loss": 5.6293, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 6.78, |
|
"learning_rate": 7.9792e-05, |
|
"loss": 5.594, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 7.45, |
|
"learning_rate": 7.95614647887324e-05, |
|
"loss": 5.5631, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 8.13, |
|
"learning_rate": 7.911121126760565e-05, |
|
"loss": 5.5393, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 8.81, |
|
"learning_rate": 7.866095774647888e-05, |
|
"loss": 5.5223, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 9.49, |
|
"learning_rate": 7.821070422535212e-05, |
|
"loss": 5.5064, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 10.16, |
|
"learning_rate": 7.776045070422535e-05, |
|
"loss": 5.4923, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 10.84, |
|
"learning_rate": 7.73101971830986e-05, |
|
"loss": 5.482, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 11.52, |
|
"learning_rate": 7.685994366197184e-05, |
|
"loss": 5.4725, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 12.2, |
|
"learning_rate": 7.640969014084507e-05, |
|
"loss": 5.4627, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 12.87, |
|
"learning_rate": 7.595943661971832e-05, |
|
"loss": 5.4567, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 13.55, |
|
"learning_rate": 7.550918309859156e-05, |
|
"loss": 5.4456, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 14.23, |
|
"learning_rate": 7.505892957746479e-05, |
|
"loss": 5.4407, |
|
"step": 21000 |
|
}, |
|
{ |
|
"epoch": 14.91, |
|
"learning_rate": 7.460867605633804e-05, |
|
"loss": 5.4347, |
|
"step": 22000 |
|
}, |
|
{ |
|
"epoch": 15.58, |
|
"learning_rate": 7.415842253521126e-05, |
|
"loss": 5.4284, |
|
"step": 23000 |
|
}, |
|
{ |
|
"epoch": 16.26, |
|
"learning_rate": 7.370816901408451e-05, |
|
"loss": 5.424, |
|
"step": 24000 |
|
}, |
|
{ |
|
"epoch": 16.94, |
|
"learning_rate": 7.325791549295775e-05, |
|
"loss": 5.1332, |
|
"step": 25000 |
|
}, |
|
{ |
|
"epoch": 17.62, |
|
"learning_rate": 7.2807661971831e-05, |
|
"loss": 4.5278, |
|
"step": 26000 |
|
}, |
|
{ |
|
"epoch": 18.29, |
|
"learning_rate": 7.235740845070423e-05, |
|
"loss": 3.2777, |
|
"step": 27000 |
|
}, |
|
{ |
|
"epoch": 18.97, |
|
"learning_rate": 7.190715492957747e-05, |
|
"loss": 2.4646, |
|
"step": 28000 |
|
}, |
|
{ |
|
"epoch": 19.65, |
|
"learning_rate": 7.145690140845072e-05, |
|
"loss": 2.2162, |
|
"step": 29000 |
|
}, |
|
{ |
|
"epoch": 20.33, |
|
"learning_rate": 7.100664788732395e-05, |
|
"loss": 2.0715, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 21.0, |
|
"learning_rate": 7.055639436619719e-05, |
|
"loss": 1.9643, |
|
"step": 31000 |
|
}, |
|
{ |
|
"epoch": 21.68, |
|
"learning_rate": 7.010614084507043e-05, |
|
"loss": 1.8819, |
|
"step": 32000 |
|
}, |
|
{ |
|
"epoch": 22.36, |
|
"learning_rate": 6.965588732394366e-05, |
|
"loss": 1.8187, |
|
"step": 33000 |
|
}, |
|
{ |
|
"epoch": 23.04, |
|
"learning_rate": 6.920563380281691e-05, |
|
"loss": 1.7671, |
|
"step": 34000 |
|
}, |
|
{ |
|
"epoch": 23.71, |
|
"learning_rate": 6.875538028169015e-05, |
|
"loss": 1.7254, |
|
"step": 35000 |
|
}, |
|
{ |
|
"epoch": 24.39, |
|
"learning_rate": 6.830512676056338e-05, |
|
"loss": 1.6912, |
|
"step": 36000 |
|
}, |
|
{ |
|
"epoch": 25.07, |
|
"learning_rate": 6.785487323943663e-05, |
|
"loss": 1.6589, |
|
"step": 37000 |
|
}, |
|
{ |
|
"epoch": 25.75, |
|
"learning_rate": 6.740461971830987e-05, |
|
"loss": 1.631, |
|
"step": 38000 |
|
}, |
|
{ |
|
"epoch": 26.42, |
|
"learning_rate": 6.69543661971831e-05, |
|
"loss": 1.606, |
|
"step": 39000 |
|
}, |
|
{ |
|
"epoch": 27.1, |
|
"learning_rate": 6.650411267605634e-05, |
|
"loss": 1.584, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 27.78, |
|
"learning_rate": 6.605385915492959e-05, |
|
"loss": 1.5641, |
|
"step": 41000 |
|
}, |
|
{ |
|
"epoch": 28.46, |
|
"learning_rate": 6.560360563380282e-05, |
|
"loss": 1.5455, |
|
"step": 42000 |
|
}, |
|
{ |
|
"epoch": 29.13, |
|
"learning_rate": 6.515335211267606e-05, |
|
"loss": 1.528, |
|
"step": 43000 |
|
}, |
|
{ |
|
"epoch": 29.81, |
|
"learning_rate": 6.47030985915493e-05, |
|
"loss": 1.5117, |
|
"step": 44000 |
|
}, |
|
{ |
|
"epoch": 30.49, |
|
"learning_rate": 6.425284507042254e-05, |
|
"loss": 1.4979, |
|
"step": 45000 |
|
}, |
|
{ |
|
"epoch": 31.17, |
|
"learning_rate": 6.380259154929578e-05, |
|
"loss": 1.4843, |
|
"step": 46000 |
|
}, |
|
{ |
|
"epoch": 31.84, |
|
"learning_rate": 6.335233802816903e-05, |
|
"loss": 1.4728, |
|
"step": 47000 |
|
}, |
|
{ |
|
"epoch": 32.52, |
|
"learning_rate": 6.290208450704226e-05, |
|
"loss": 1.4601, |
|
"step": 48000 |
|
}, |
|
{ |
|
"epoch": 33.2, |
|
"learning_rate": 6.24518309859155e-05, |
|
"loss": 1.4495, |
|
"step": 49000 |
|
}, |
|
{ |
|
"epoch": 33.88, |
|
"learning_rate": 6.200157746478873e-05, |
|
"loss": 1.4395, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 34.55, |
|
"learning_rate": 6.155132394366198e-05, |
|
"loss": 1.43, |
|
"step": 51000 |
|
}, |
|
{ |
|
"epoch": 35.23, |
|
"learning_rate": 6.110107042253522e-05, |
|
"loss": 1.4211, |
|
"step": 52000 |
|
}, |
|
{ |
|
"epoch": 35.91, |
|
"learning_rate": 6.0650816901408453e-05, |
|
"loss": 1.4127, |
|
"step": 53000 |
|
}, |
|
{ |
|
"epoch": 36.59, |
|
"learning_rate": 6.0200563380281696e-05, |
|
"loss": 1.4031, |
|
"step": 54000 |
|
}, |
|
{ |
|
"epoch": 37.26, |
|
"learning_rate": 5.975030985915493e-05, |
|
"loss": 1.3951, |
|
"step": 55000 |
|
}, |
|
{ |
|
"epoch": 37.94, |
|
"learning_rate": 5.9300056338028174e-05, |
|
"loss": 1.3876, |
|
"step": 56000 |
|
}, |
|
{ |
|
"epoch": 38.62, |
|
"learning_rate": 5.8849802816901416e-05, |
|
"loss": 1.3801, |
|
"step": 57000 |
|
}, |
|
{ |
|
"epoch": 39.3, |
|
"learning_rate": 5.839954929577465e-05, |
|
"loss": 1.3744, |
|
"step": 58000 |
|
}, |
|
{ |
|
"epoch": 39.97, |
|
"learning_rate": 5.7949295774647894e-05, |
|
"loss": 1.3673, |
|
"step": 59000 |
|
}, |
|
{ |
|
"epoch": 40.65, |
|
"learning_rate": 5.7499042253521136e-05, |
|
"loss": 1.3608, |
|
"step": 60000 |
|
}, |
|
{ |
|
"epoch": 41.33, |
|
"learning_rate": 5.7048788732394365e-05, |
|
"loss": 1.3547, |
|
"step": 61000 |
|
}, |
|
{ |
|
"epoch": 42.01, |
|
"learning_rate": 5.659853521126761e-05, |
|
"loss": 1.3496, |
|
"step": 62000 |
|
}, |
|
{ |
|
"epoch": 42.68, |
|
"learning_rate": 5.614828169014085e-05, |
|
"loss": 1.3429, |
|
"step": 63000 |
|
}, |
|
{ |
|
"epoch": 43.36, |
|
"learning_rate": 5.569802816901409e-05, |
|
"loss": 1.3381, |
|
"step": 64000 |
|
}, |
|
{ |
|
"epoch": 44.04, |
|
"learning_rate": 5.524777464788733e-05, |
|
"loss": 1.3322, |
|
"step": 65000 |
|
}, |
|
{ |
|
"epoch": 44.72, |
|
"learning_rate": 5.479752112676057e-05, |
|
"loss": 1.328, |
|
"step": 66000 |
|
}, |
|
{ |
|
"epoch": 45.39, |
|
"learning_rate": 5.434726760563381e-05, |
|
"loss": 1.3233, |
|
"step": 67000 |
|
}, |
|
{ |
|
"epoch": 46.07, |
|
"learning_rate": 5.389701408450704e-05, |
|
"loss": 1.319, |
|
"step": 68000 |
|
}, |
|
{ |
|
"epoch": 46.75, |
|
"learning_rate": 5.3446760563380284e-05, |
|
"loss": 1.3135, |
|
"step": 69000 |
|
}, |
|
{ |
|
"epoch": 47.43, |
|
"learning_rate": 5.2996507042253526e-05, |
|
"loss": 1.3088, |
|
"step": 70000 |
|
}, |
|
{ |
|
"epoch": 48.1, |
|
"learning_rate": 5.254625352112676e-05, |
|
"loss": 1.305, |
|
"step": 71000 |
|
}, |
|
{ |
|
"epoch": 48.78, |
|
"learning_rate": 5.2096000000000004e-05, |
|
"loss": 1.3006, |
|
"step": 72000 |
|
}, |
|
{ |
|
"epoch": 49.46, |
|
"learning_rate": 5.1645746478873246e-05, |
|
"loss": 1.2968, |
|
"step": 73000 |
|
}, |
|
{ |
|
"epoch": 50.14, |
|
"learning_rate": 5.119549295774648e-05, |
|
"loss": 1.2933, |
|
"step": 74000 |
|
}, |
|
{ |
|
"epoch": 50.81, |
|
"learning_rate": 5.0745239436619724e-05, |
|
"loss": 1.2884, |
|
"step": 75000 |
|
}, |
|
{ |
|
"epoch": 51.49, |
|
"learning_rate": 5.029498591549297e-05, |
|
"loss": 1.2853, |
|
"step": 76000 |
|
}, |
|
{ |
|
"epoch": 52.17, |
|
"learning_rate": 4.9844732394366195e-05, |
|
"loss": 1.2819, |
|
"step": 77000 |
|
}, |
|
{ |
|
"epoch": 52.85, |
|
"learning_rate": 4.939447887323944e-05, |
|
"loss": 1.2795, |
|
"step": 78000 |
|
}, |
|
{ |
|
"epoch": 53.52, |
|
"learning_rate": 4.894422535211268e-05, |
|
"loss": 1.2754, |
|
"step": 79000 |
|
}, |
|
{ |
|
"epoch": 54.2, |
|
"learning_rate": 4.8493971830985916e-05, |
|
"loss": 1.2715, |
|
"step": 80000 |
|
}, |
|
{ |
|
"epoch": 54.88, |
|
"learning_rate": 4.804371830985916e-05, |
|
"loss": 1.2691, |
|
"step": 81000 |
|
}, |
|
{ |
|
"epoch": 55.56, |
|
"learning_rate": 4.75934647887324e-05, |
|
"loss": 1.2652, |
|
"step": 82000 |
|
}, |
|
{ |
|
"epoch": 56.23, |
|
"learning_rate": 4.714321126760564e-05, |
|
"loss": 1.2631, |
|
"step": 83000 |
|
}, |
|
{ |
|
"epoch": 56.91, |
|
"learning_rate": 4.669295774647888e-05, |
|
"loss": 1.2598, |
|
"step": 84000 |
|
}, |
|
{ |
|
"epoch": 57.59, |
|
"learning_rate": 4.624270422535212e-05, |
|
"loss": 1.2575, |
|
"step": 85000 |
|
}, |
|
{ |
|
"epoch": 58.27, |
|
"learning_rate": 4.5792450704225356e-05, |
|
"loss": 1.2552, |
|
"step": 86000 |
|
}, |
|
{ |
|
"epoch": 58.94, |
|
"learning_rate": 4.534219718309859e-05, |
|
"loss": 1.2509, |
|
"step": 87000 |
|
}, |
|
{ |
|
"epoch": 59.62, |
|
"learning_rate": 4.4891943661971834e-05, |
|
"loss": 1.2492, |
|
"step": 88000 |
|
}, |
|
{ |
|
"epoch": 60.3, |
|
"learning_rate": 4.4441690140845077e-05, |
|
"loss": 1.2468, |
|
"step": 89000 |
|
}, |
|
{ |
|
"epoch": 60.98, |
|
"learning_rate": 4.399143661971831e-05, |
|
"loss": 1.2441, |
|
"step": 90000 |
|
}, |
|
{ |
|
"epoch": 61.65, |
|
"learning_rate": 4.3541183098591555e-05, |
|
"loss": 1.2413, |
|
"step": 91000 |
|
}, |
|
{ |
|
"epoch": 62.33, |
|
"learning_rate": 4.30909295774648e-05, |
|
"loss": 1.2402, |
|
"step": 92000 |
|
}, |
|
{ |
|
"epoch": 63.01, |
|
"learning_rate": 4.2640676056338026e-05, |
|
"loss": 1.237, |
|
"step": 93000 |
|
}, |
|
{ |
|
"epoch": 63.69, |
|
"learning_rate": 4.219042253521127e-05, |
|
"loss": 1.2347, |
|
"step": 94000 |
|
}, |
|
{ |
|
"epoch": 64.36, |
|
"learning_rate": 4.174016901408451e-05, |
|
"loss": 1.2326, |
|
"step": 95000 |
|
}, |
|
{ |
|
"epoch": 65.04, |
|
"learning_rate": 4.1289915492957746e-05, |
|
"loss": 1.231, |
|
"step": 96000 |
|
}, |
|
{ |
|
"epoch": 65.72, |
|
"learning_rate": 4.083966197183099e-05, |
|
"loss": 1.2281, |
|
"step": 97000 |
|
}, |
|
{ |
|
"epoch": 66.4, |
|
"learning_rate": 4.038940845070423e-05, |
|
"loss": 1.2273, |
|
"step": 98000 |
|
}, |
|
{ |
|
"epoch": 67.07, |
|
"learning_rate": 3.9939154929577466e-05, |
|
"loss": 1.2251, |
|
"step": 99000 |
|
}, |
|
{ |
|
"epoch": 67.75, |
|
"learning_rate": 3.948890140845071e-05, |
|
"loss": 1.2234, |
|
"step": 100000 |
|
}, |
|
{ |
|
"epoch": 68.43, |
|
"learning_rate": 3.903864788732395e-05, |
|
"loss": 1.2208, |
|
"step": 101000 |
|
}, |
|
{ |
|
"epoch": 69.11, |
|
"learning_rate": 3.8588394366197187e-05, |
|
"loss": 1.2195, |
|
"step": 102000 |
|
}, |
|
{ |
|
"epoch": 69.78, |
|
"learning_rate": 3.813814084507042e-05, |
|
"loss": 1.2184, |
|
"step": 103000 |
|
}, |
|
{ |
|
"epoch": 70.46, |
|
"learning_rate": 3.7687887323943664e-05, |
|
"loss": 1.2158, |
|
"step": 104000 |
|
}, |
|
{ |
|
"epoch": 71.14, |
|
"learning_rate": 3.723763380281691e-05, |
|
"loss": 1.2136, |
|
"step": 105000 |
|
}, |
|
{ |
|
"epoch": 71.82, |
|
"learning_rate": 3.678738028169015e-05, |
|
"loss": 1.2134, |
|
"step": 106000 |
|
}, |
|
{ |
|
"epoch": 72.49, |
|
"learning_rate": 3.6337126760563385e-05, |
|
"loss": 1.2108, |
|
"step": 107000 |
|
}, |
|
{ |
|
"epoch": 73.17, |
|
"learning_rate": 3.588687323943662e-05, |
|
"loss": 1.2091, |
|
"step": 108000 |
|
}, |
|
{ |
|
"epoch": 73.85, |
|
"learning_rate": 3.543661971830986e-05, |
|
"loss": 1.2071, |
|
"step": 109000 |
|
}, |
|
{ |
|
"epoch": 74.53, |
|
"learning_rate": 3.49863661971831e-05, |
|
"loss": 1.2061, |
|
"step": 110000 |
|
}, |
|
{ |
|
"epoch": 75.2, |
|
"learning_rate": 3.453611267605634e-05, |
|
"loss": 1.2041, |
|
"step": 111000 |
|
}, |
|
{ |
|
"epoch": 75.88, |
|
"learning_rate": 3.408585915492958e-05, |
|
"loss": 1.2019, |
|
"step": 112000 |
|
}, |
|
{ |
|
"epoch": 76.56, |
|
"learning_rate": 3.363560563380282e-05, |
|
"loss": 1.2012, |
|
"step": 113000 |
|
}, |
|
{ |
|
"epoch": 77.24, |
|
"learning_rate": 3.3185352112676054e-05, |
|
"loss": 1.1996, |
|
"step": 114000 |
|
}, |
|
{ |
|
"epoch": 77.91, |
|
"learning_rate": 3.2735098591549296e-05, |
|
"loss": 1.1986, |
|
"step": 115000 |
|
}, |
|
{ |
|
"epoch": 78.59, |
|
"learning_rate": 3.228484507042254e-05, |
|
"loss": 1.1973, |
|
"step": 116000 |
|
}, |
|
{ |
|
"epoch": 79.27, |
|
"learning_rate": 3.183459154929578e-05, |
|
"loss": 1.1962, |
|
"step": 117000 |
|
}, |
|
{ |
|
"epoch": 79.95, |
|
"learning_rate": 3.138433802816902e-05, |
|
"loss": 1.194, |
|
"step": 118000 |
|
}, |
|
{ |
|
"epoch": 80.62, |
|
"learning_rate": 3.093408450704225e-05, |
|
"loss": 1.1932, |
|
"step": 119000 |
|
}, |
|
{ |
|
"epoch": 81.3, |
|
"learning_rate": 3.0483830985915498e-05, |
|
"loss": 1.1914, |
|
"step": 120000 |
|
}, |
|
{ |
|
"epoch": 81.98, |
|
"learning_rate": 3.0033577464788734e-05, |
|
"loss": 1.1906, |
|
"step": 121000 |
|
}, |
|
{ |
|
"epoch": 82.66, |
|
"learning_rate": 2.9583323943661973e-05, |
|
"loss": 1.1879, |
|
"step": 122000 |
|
}, |
|
{ |
|
"epoch": 83.33, |
|
"learning_rate": 2.9133070422535215e-05, |
|
"loss": 1.189, |
|
"step": 123000 |
|
}, |
|
{ |
|
"epoch": 84.01, |
|
"learning_rate": 2.8682816901408454e-05, |
|
"loss": 1.1872, |
|
"step": 124000 |
|
}, |
|
{ |
|
"epoch": 84.69, |
|
"learning_rate": 2.8232563380281693e-05, |
|
"loss": 1.186, |
|
"step": 125000 |
|
}, |
|
{ |
|
"epoch": 85.37, |
|
"learning_rate": 2.7782309859154932e-05, |
|
"loss": 1.1848, |
|
"step": 126000 |
|
}, |
|
{ |
|
"epoch": 86.04, |
|
"learning_rate": 2.733205633802817e-05, |
|
"loss": 1.1835, |
|
"step": 127000 |
|
}, |
|
{ |
|
"epoch": 86.72, |
|
"learning_rate": 2.6881802816901413e-05, |
|
"loss": 1.1831, |
|
"step": 128000 |
|
}, |
|
{ |
|
"epoch": 87.4, |
|
"learning_rate": 2.643154929577465e-05, |
|
"loss": 1.1811, |
|
"step": 129000 |
|
}, |
|
{ |
|
"epoch": 88.08, |
|
"learning_rate": 2.5981295774647888e-05, |
|
"loss": 1.1803, |
|
"step": 130000 |
|
}, |
|
{ |
|
"epoch": 88.75, |
|
"learning_rate": 2.553104225352113e-05, |
|
"loss": 1.1789, |
|
"step": 131000 |
|
}, |
|
{ |
|
"epoch": 89.43, |
|
"learning_rate": 2.508078873239437e-05, |
|
"loss": 1.1791, |
|
"step": 132000 |
|
}, |
|
{ |
|
"epoch": 90.11, |
|
"learning_rate": 2.4630535211267605e-05, |
|
"loss": 1.1769, |
|
"step": 133000 |
|
}, |
|
{ |
|
"epoch": 90.79, |
|
"learning_rate": 2.4180281690140847e-05, |
|
"loss": 1.1766, |
|
"step": 134000 |
|
}, |
|
{ |
|
"epoch": 91.46, |
|
"learning_rate": 2.3730028169014086e-05, |
|
"loss": 1.1757, |
|
"step": 135000 |
|
}, |
|
{ |
|
"epoch": 92.14, |
|
"learning_rate": 2.327977464788733e-05, |
|
"loss": 1.1741, |
|
"step": 136000 |
|
}, |
|
{ |
|
"epoch": 92.82, |
|
"learning_rate": 2.2829521126760564e-05, |
|
"loss": 1.1734, |
|
"step": 137000 |
|
}, |
|
{ |
|
"epoch": 93.5, |
|
"learning_rate": 2.2379267605633803e-05, |
|
"loss": 1.1733, |
|
"step": 138000 |
|
}, |
|
{ |
|
"epoch": 94.17, |
|
"learning_rate": 2.1929014084507045e-05, |
|
"loss": 1.1723, |
|
"step": 139000 |
|
}, |
|
{ |
|
"epoch": 94.85, |
|
"learning_rate": 2.1478760563380284e-05, |
|
"loss": 1.1715, |
|
"step": 140000 |
|
}, |
|
{ |
|
"epoch": 95.53, |
|
"learning_rate": 2.102850704225352e-05, |
|
"loss": 1.1699, |
|
"step": 141000 |
|
}, |
|
{ |
|
"epoch": 96.21, |
|
"learning_rate": 2.0578253521126762e-05, |
|
"loss": 1.1689, |
|
"step": 142000 |
|
}, |
|
{ |
|
"epoch": 96.88, |
|
"learning_rate": 2.0128e-05, |
|
"loss": 1.1691, |
|
"step": 143000 |
|
}, |
|
{ |
|
"epoch": 97.56, |
|
"learning_rate": 1.967774647887324e-05, |
|
"loss": 1.1675, |
|
"step": 144000 |
|
}, |
|
{ |
|
"epoch": 98.24, |
|
"learning_rate": 1.9227492957746482e-05, |
|
"loss": 1.1671, |
|
"step": 145000 |
|
}, |
|
{ |
|
"epoch": 98.92, |
|
"learning_rate": 1.877723943661972e-05, |
|
"loss": 1.167, |
|
"step": 146000 |
|
}, |
|
{ |
|
"epoch": 99.59, |
|
"learning_rate": 1.8326985915492957e-05, |
|
"loss": 1.1659, |
|
"step": 147000 |
|
}, |
|
{ |
|
"epoch": 100.27, |
|
"learning_rate": 1.78767323943662e-05, |
|
"loss": 1.1653, |
|
"step": 148000 |
|
}, |
|
{ |
|
"epoch": 100.95, |
|
"learning_rate": 1.7426478873239438e-05, |
|
"loss": 1.1647, |
|
"step": 149000 |
|
}, |
|
{ |
|
"epoch": 101.63, |
|
"learning_rate": 1.6976225352112677e-05, |
|
"loss": 1.1639, |
|
"step": 150000 |
|
}, |
|
{ |
|
"epoch": 102.3, |
|
"learning_rate": 1.6525971830985916e-05, |
|
"loss": 1.163, |
|
"step": 151000 |
|
}, |
|
{ |
|
"epoch": 102.98, |
|
"learning_rate": 1.6075718309859155e-05, |
|
"loss": 1.1615, |
|
"step": 152000 |
|
}, |
|
{ |
|
"epoch": 103.66, |
|
"learning_rate": 1.5625464788732398e-05, |
|
"loss": 1.1615, |
|
"step": 153000 |
|
}, |
|
{ |
|
"epoch": 104.34, |
|
"learning_rate": 1.5175211267605635e-05, |
|
"loss": 1.1595, |
|
"step": 154000 |
|
}, |
|
{ |
|
"epoch": 105.01, |
|
"learning_rate": 1.4724957746478874e-05, |
|
"loss": 1.1603, |
|
"step": 155000 |
|
}, |
|
{ |
|
"epoch": 105.69, |
|
"learning_rate": 1.4274704225352114e-05, |
|
"loss": 1.1597, |
|
"step": 156000 |
|
}, |
|
{ |
|
"epoch": 106.37, |
|
"learning_rate": 1.3824450704225353e-05, |
|
"loss": 1.1591, |
|
"step": 157000 |
|
}, |
|
{ |
|
"epoch": 107.05, |
|
"learning_rate": 1.3374197183098592e-05, |
|
"loss": 1.1586, |
|
"step": 158000 |
|
}, |
|
{ |
|
"epoch": 107.72, |
|
"learning_rate": 1.2923943661971831e-05, |
|
"loss": 1.1576, |
|
"step": 159000 |
|
}, |
|
{ |
|
"epoch": 108.4, |
|
"learning_rate": 1.2473690140845072e-05, |
|
"loss": 1.1563, |
|
"step": 160000 |
|
}, |
|
{ |
|
"epoch": 109.08, |
|
"learning_rate": 1.2023436619718311e-05, |
|
"loss": 1.1559, |
|
"step": 161000 |
|
}, |
|
{ |
|
"epoch": 109.76, |
|
"learning_rate": 1.157318309859155e-05, |
|
"loss": 1.1559, |
|
"step": 162000 |
|
}, |
|
{ |
|
"epoch": 110.43, |
|
"learning_rate": 1.1122929577464789e-05, |
|
"loss": 1.1543, |
|
"step": 163000 |
|
}, |
|
{ |
|
"epoch": 111.11, |
|
"learning_rate": 1.067267605633803e-05, |
|
"loss": 1.1553, |
|
"step": 164000 |
|
}, |
|
{ |
|
"epoch": 111.79, |
|
"learning_rate": 1.022242253521127e-05, |
|
"loss": 1.154, |
|
"step": 165000 |
|
}, |
|
{ |
|
"epoch": 112.47, |
|
"learning_rate": 9.772169014084507e-06, |
|
"loss": 1.1535, |
|
"step": 166000 |
|
}, |
|
{ |
|
"epoch": 113.14, |
|
"learning_rate": 9.321915492957746e-06, |
|
"loss": 1.1536, |
|
"step": 167000 |
|
}, |
|
{ |
|
"epoch": 113.82, |
|
"learning_rate": 8.871661971830987e-06, |
|
"loss": 1.1525, |
|
"step": 168000 |
|
}, |
|
{ |
|
"epoch": 114.5, |
|
"learning_rate": 8.421408450704226e-06, |
|
"loss": 1.1524, |
|
"step": 169000 |
|
}, |
|
{ |
|
"epoch": 115.18, |
|
"learning_rate": 7.971154929577467e-06, |
|
"loss": 1.152, |
|
"step": 170000 |
|
}, |
|
{ |
|
"epoch": 115.85, |
|
"learning_rate": 7.520901408450705e-06, |
|
"loss": 1.1514, |
|
"step": 171000 |
|
}, |
|
{ |
|
"epoch": 116.53, |
|
"learning_rate": 7.070647887323944e-06, |
|
"loss": 1.151, |
|
"step": 172000 |
|
}, |
|
{ |
|
"epoch": 117.21, |
|
"learning_rate": 6.620394366197184e-06, |
|
"loss": 1.1504, |
|
"step": 173000 |
|
}, |
|
{ |
|
"epoch": 117.89, |
|
"learning_rate": 6.170140845070423e-06, |
|
"loss": 1.1495, |
|
"step": 174000 |
|
}, |
|
{ |
|
"epoch": 118.56, |
|
"learning_rate": 5.719887323943662e-06, |
|
"loss": 1.1494, |
|
"step": 175000 |
|
}, |
|
{ |
|
"epoch": 119.24, |
|
"learning_rate": 5.269633802816901e-06, |
|
"loss": 1.1493, |
|
"step": 176000 |
|
}, |
|
{ |
|
"epoch": 119.92, |
|
"learning_rate": 4.819380281690141e-06, |
|
"loss": 1.149, |
|
"step": 177000 |
|
}, |
|
{ |
|
"epoch": 120.6, |
|
"learning_rate": 4.369126760563381e-06, |
|
"loss": 1.1485, |
|
"step": 178000 |
|
}, |
|
{ |
|
"epoch": 121.27, |
|
"learning_rate": 3.91887323943662e-06, |
|
"loss": 1.1482, |
|
"step": 179000 |
|
}, |
|
{ |
|
"epoch": 121.95, |
|
"learning_rate": 3.4686197183098598e-06, |
|
"loss": 1.148, |
|
"step": 180000 |
|
}, |
|
{ |
|
"epoch": 122.63, |
|
"learning_rate": 3.018366197183099e-06, |
|
"loss": 1.1477, |
|
"step": 181000 |
|
}, |
|
{ |
|
"epoch": 123.31, |
|
"learning_rate": 2.568112676056338e-06, |
|
"loss": 1.147, |
|
"step": 182000 |
|
}, |
|
{ |
|
"epoch": 123.98, |
|
"learning_rate": 2.1178591549295775e-06, |
|
"loss": 1.1466, |
|
"step": 183000 |
|
}, |
|
{ |
|
"epoch": 124.66, |
|
"learning_rate": 1.6676056338028171e-06, |
|
"loss": 1.1475, |
|
"step": 184000 |
|
}, |
|
{ |
|
"epoch": 125.34, |
|
"learning_rate": 1.2173521126760563e-06, |
|
"loss": 1.1459, |
|
"step": 185000 |
|
}, |
|
{ |
|
"epoch": 126.02, |
|
"learning_rate": 7.670985915492958e-07, |
|
"loss": 1.1466, |
|
"step": 186000 |
|
}, |
|
{ |
|
"epoch": 126.69, |
|
"learning_rate": 3.1684507042253523e-07, |
|
"loss": 1.1462, |
|
"step": 187000 |
|
}, |
|
{ |
|
"epoch": 127.03, |
|
"step": 187500, |
|
"total_flos": 1.0102999800563001e+20, |
|
"train_loss": 0.015288788411458334, |
|
"train_runtime": 1005.2623, |
|
"train_samples_per_second": 381989.844, |
|
"train_steps_per_second": 186.518 |
|
} |
|
], |
|
"max_steps": 187500, |
|
"num_train_epochs": 128, |
|
"total_flos": 1.0102999800563001e+20, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|