|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 84.95695514272768, |
|
"global_step": 187500, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 7.864000000000001e-06, |
|
"loss": 8.2983, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 1.5856e-05, |
|
"loss": 6.5434, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 2.3848e-05, |
|
"loss": 6.1765, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 3.184000000000001e-05, |
|
"loss": 5.9601, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"learning_rate": 3.9832e-05, |
|
"loss": 5.8266, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 2.72, |
|
"learning_rate": 4.7824e-05, |
|
"loss": 5.7332, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 3.17, |
|
"learning_rate": 5.5816e-05, |
|
"loss": 5.6554, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 3.62, |
|
"learning_rate": 6.380800000000001e-05, |
|
"loss": 5.5975, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 4.08, |
|
"learning_rate": 7.18e-05, |
|
"loss": 5.5568, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 4.53, |
|
"learning_rate": 7.9792e-05, |
|
"loss": 5.5224, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 4.98, |
|
"learning_rate": 7.95614647887324e-05, |
|
"loss": 5.4944, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 5.44, |
|
"learning_rate": 7.911121126760565e-05, |
|
"loss": 5.473, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 5.89, |
|
"learning_rate": 7.866095774647888e-05, |
|
"loss": 5.4557, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 6.34, |
|
"learning_rate": 7.821070422535212e-05, |
|
"loss": 5.4392, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 6.8, |
|
"learning_rate": 7.776045070422535e-05, |
|
"loss": 5.4274, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 7.25, |
|
"learning_rate": 7.73101971830986e-05, |
|
"loss": 5.4168, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 7.7, |
|
"learning_rate": 7.685994366197184e-05, |
|
"loss": 5.4051, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 8.16, |
|
"learning_rate": 7.640969014084507e-05, |
|
"loss": 5.3963, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 8.61, |
|
"learning_rate": 7.595943661971832e-05, |
|
"loss": 5.3901, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 9.06, |
|
"learning_rate": 7.550918309859156e-05, |
|
"loss": 5.3848, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 9.52, |
|
"learning_rate": 7.505892957746479e-05, |
|
"loss": 5.2904, |
|
"step": 21000 |
|
}, |
|
{ |
|
"epoch": 9.97, |
|
"learning_rate": 7.460867605633804e-05, |
|
"loss": 4.6695, |
|
"step": 22000 |
|
}, |
|
{ |
|
"epoch": 10.42, |
|
"learning_rate": 7.415842253521126e-05, |
|
"loss": 3.6226, |
|
"step": 23000 |
|
}, |
|
{ |
|
"epoch": 10.87, |
|
"learning_rate": 7.370816901408451e-05, |
|
"loss": 2.5281, |
|
"step": 24000 |
|
}, |
|
{ |
|
"epoch": 11.33, |
|
"learning_rate": 7.325791549295775e-05, |
|
"loss": 2.2386, |
|
"step": 25000 |
|
}, |
|
{ |
|
"epoch": 11.78, |
|
"learning_rate": 7.2807661971831e-05, |
|
"loss": 2.0823, |
|
"step": 26000 |
|
}, |
|
{ |
|
"epoch": 12.23, |
|
"learning_rate": 7.235740845070423e-05, |
|
"loss": 1.9728, |
|
"step": 27000 |
|
}, |
|
{ |
|
"epoch": 12.69, |
|
"learning_rate": 7.190715492957747e-05, |
|
"loss": 1.8923, |
|
"step": 28000 |
|
}, |
|
{ |
|
"epoch": 13.14, |
|
"learning_rate": 7.145690140845072e-05, |
|
"loss": 1.8241, |
|
"step": 29000 |
|
}, |
|
{ |
|
"epoch": 13.59, |
|
"learning_rate": 7.100664788732395e-05, |
|
"loss": 1.7682, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 14.05, |
|
"learning_rate": 7.055639436619719e-05, |
|
"loss": 1.7244, |
|
"step": 31000 |
|
}, |
|
{ |
|
"epoch": 14.5, |
|
"learning_rate": 7.010614084507043e-05, |
|
"loss": 1.6804, |
|
"step": 32000 |
|
}, |
|
{ |
|
"epoch": 14.95, |
|
"learning_rate": 6.965588732394366e-05, |
|
"loss": 1.6465, |
|
"step": 33000 |
|
}, |
|
{ |
|
"epoch": 15.41, |
|
"learning_rate": 6.920563380281691e-05, |
|
"loss": 1.6159, |
|
"step": 34000 |
|
}, |
|
{ |
|
"epoch": 15.86, |
|
"learning_rate": 6.875538028169015e-05, |
|
"loss": 1.5909, |
|
"step": 35000 |
|
}, |
|
{ |
|
"epoch": 16.31, |
|
"learning_rate": 6.830512676056338e-05, |
|
"loss": 1.5664, |
|
"step": 36000 |
|
}, |
|
{ |
|
"epoch": 16.76, |
|
"learning_rate": 6.785487323943663e-05, |
|
"loss": 1.5481, |
|
"step": 37000 |
|
}, |
|
{ |
|
"epoch": 17.22, |
|
"learning_rate": 6.740461971830987e-05, |
|
"loss": 1.5282, |
|
"step": 38000 |
|
}, |
|
{ |
|
"epoch": 17.67, |
|
"learning_rate": 6.69543661971831e-05, |
|
"loss": 1.5103, |
|
"step": 39000 |
|
}, |
|
{ |
|
"epoch": 18.12, |
|
"learning_rate": 6.650411267605634e-05, |
|
"loss": 1.4946, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 18.58, |
|
"learning_rate": 6.605385915492959e-05, |
|
"loss": 1.4793, |
|
"step": 41000 |
|
}, |
|
{ |
|
"epoch": 19.03, |
|
"learning_rate": 6.560360563380282e-05, |
|
"loss": 1.4647, |
|
"step": 42000 |
|
}, |
|
{ |
|
"epoch": 19.48, |
|
"learning_rate": 6.515335211267606e-05, |
|
"loss": 1.4525, |
|
"step": 43000 |
|
}, |
|
{ |
|
"epoch": 19.94, |
|
"learning_rate": 6.47030985915493e-05, |
|
"loss": 1.4408, |
|
"step": 44000 |
|
}, |
|
{ |
|
"epoch": 20.39, |
|
"learning_rate": 6.425284507042254e-05, |
|
"loss": 1.429, |
|
"step": 45000 |
|
}, |
|
{ |
|
"epoch": 20.84, |
|
"learning_rate": 6.380259154929578e-05, |
|
"loss": 1.4191, |
|
"step": 46000 |
|
}, |
|
{ |
|
"epoch": 21.3, |
|
"learning_rate": 6.335233802816903e-05, |
|
"loss": 1.4093, |
|
"step": 47000 |
|
}, |
|
{ |
|
"epoch": 21.75, |
|
"learning_rate": 6.290208450704226e-05, |
|
"loss": 1.399, |
|
"step": 48000 |
|
}, |
|
{ |
|
"epoch": 22.2, |
|
"learning_rate": 6.24518309859155e-05, |
|
"loss": 1.3902, |
|
"step": 49000 |
|
}, |
|
{ |
|
"epoch": 22.66, |
|
"learning_rate": 6.200157746478873e-05, |
|
"loss": 1.3811, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 23.11, |
|
"learning_rate": 6.155132394366198e-05, |
|
"loss": 1.3738, |
|
"step": 51000 |
|
}, |
|
{ |
|
"epoch": 23.56, |
|
"learning_rate": 6.110107042253522e-05, |
|
"loss": 1.3651, |
|
"step": 52000 |
|
}, |
|
{ |
|
"epoch": 24.01, |
|
"learning_rate": 6.0650816901408453e-05, |
|
"loss": 1.361, |
|
"step": 53000 |
|
}, |
|
{ |
|
"epoch": 24.47, |
|
"learning_rate": 6.0200563380281696e-05, |
|
"loss": 1.3519, |
|
"step": 54000 |
|
}, |
|
{ |
|
"epoch": 24.92, |
|
"learning_rate": 5.975030985915493e-05, |
|
"loss": 1.3456, |
|
"step": 55000 |
|
}, |
|
{ |
|
"epoch": 25.37, |
|
"learning_rate": 5.9300056338028174e-05, |
|
"loss": 1.3394, |
|
"step": 56000 |
|
}, |
|
{ |
|
"epoch": 25.83, |
|
"learning_rate": 5.8849802816901416e-05, |
|
"loss": 1.3332, |
|
"step": 57000 |
|
}, |
|
{ |
|
"epoch": 26.28, |
|
"learning_rate": 5.839954929577465e-05, |
|
"loss": 1.3273, |
|
"step": 58000 |
|
}, |
|
{ |
|
"epoch": 26.73, |
|
"learning_rate": 5.7949295774647894e-05, |
|
"loss": 1.3222, |
|
"step": 59000 |
|
}, |
|
{ |
|
"epoch": 27.19, |
|
"learning_rate": 5.7499042253521136e-05, |
|
"loss": 1.3159, |
|
"step": 60000 |
|
}, |
|
{ |
|
"epoch": 27.64, |
|
"learning_rate": 5.7048788732394365e-05, |
|
"loss": 1.3117, |
|
"step": 61000 |
|
}, |
|
{ |
|
"epoch": 28.09, |
|
"learning_rate": 5.659853521126761e-05, |
|
"loss": 1.3067, |
|
"step": 62000 |
|
}, |
|
{ |
|
"epoch": 28.55, |
|
"learning_rate": 5.614828169014085e-05, |
|
"loss": 1.3015, |
|
"step": 63000 |
|
}, |
|
{ |
|
"epoch": 29.0, |
|
"learning_rate": 5.569802816901409e-05, |
|
"loss": 1.2972, |
|
"step": 64000 |
|
}, |
|
{ |
|
"epoch": 29.45, |
|
"learning_rate": 5.524777464788733e-05, |
|
"loss": 1.291, |
|
"step": 65000 |
|
}, |
|
{ |
|
"epoch": 29.9, |
|
"learning_rate": 5.479752112676057e-05, |
|
"loss": 1.2877, |
|
"step": 66000 |
|
}, |
|
{ |
|
"epoch": 30.36, |
|
"learning_rate": 5.434726760563381e-05, |
|
"loss": 1.2836, |
|
"step": 67000 |
|
}, |
|
{ |
|
"epoch": 30.81, |
|
"learning_rate": 5.389701408450704e-05, |
|
"loss": 1.2798, |
|
"step": 68000 |
|
}, |
|
{ |
|
"epoch": 31.26, |
|
"learning_rate": 5.344721126760564e-05, |
|
"loss": 1.276, |
|
"step": 69000 |
|
}, |
|
{ |
|
"epoch": 31.72, |
|
"learning_rate": 5.2996507042253526e-05, |
|
"loss": 1.2723, |
|
"step": 70000 |
|
}, |
|
{ |
|
"epoch": 32.17, |
|
"learning_rate": 5.254670422535211e-05, |
|
"loss": 1.2676, |
|
"step": 71000 |
|
}, |
|
{ |
|
"epoch": 32.62, |
|
"learning_rate": 5.2096000000000004e-05, |
|
"loss": 1.2645, |
|
"step": 72000 |
|
}, |
|
{ |
|
"epoch": 33.08, |
|
"learning_rate": 5.1645746478873246e-05, |
|
"loss": 1.2604, |
|
"step": 73000 |
|
}, |
|
{ |
|
"epoch": 33.53, |
|
"learning_rate": 5.119549295774648e-05, |
|
"loss": 1.2577, |
|
"step": 74000 |
|
}, |
|
{ |
|
"epoch": 33.98, |
|
"learning_rate": 5.0745239436619724e-05, |
|
"loss": 1.2552, |
|
"step": 75000 |
|
}, |
|
{ |
|
"epoch": 34.44, |
|
"learning_rate": 5.029498591549297e-05, |
|
"loss": 1.2516, |
|
"step": 76000 |
|
}, |
|
{ |
|
"epoch": 34.89, |
|
"learning_rate": 4.9844732394366195e-05, |
|
"loss": 1.2488, |
|
"step": 77000 |
|
}, |
|
{ |
|
"epoch": 35.34, |
|
"learning_rate": 4.939447887323944e-05, |
|
"loss": 1.2458, |
|
"step": 78000 |
|
}, |
|
{ |
|
"epoch": 35.8, |
|
"learning_rate": 4.894467605633804e-05, |
|
"loss": 1.2443, |
|
"step": 79000 |
|
}, |
|
{ |
|
"epoch": 36.25, |
|
"learning_rate": 4.8493971830985916e-05, |
|
"loss": 1.2398, |
|
"step": 80000 |
|
}, |
|
{ |
|
"epoch": 36.7, |
|
"learning_rate": 4.804371830985916e-05, |
|
"loss": 1.2374, |
|
"step": 81000 |
|
}, |
|
{ |
|
"epoch": 37.15, |
|
"learning_rate": 4.75934647887324e-05, |
|
"loss": 1.2341, |
|
"step": 82000 |
|
}, |
|
{ |
|
"epoch": 37.61, |
|
"learning_rate": 4.714321126760564e-05, |
|
"loss": 1.2321, |
|
"step": 83000 |
|
}, |
|
{ |
|
"epoch": 38.06, |
|
"learning_rate": 4.669295774647888e-05, |
|
"loss": 1.23, |
|
"step": 84000 |
|
}, |
|
{ |
|
"epoch": 38.51, |
|
"learning_rate": 4.624270422535212e-05, |
|
"loss": 1.2265, |
|
"step": 85000 |
|
}, |
|
{ |
|
"epoch": 38.97, |
|
"learning_rate": 4.5792450704225356e-05, |
|
"loss": 1.2253, |
|
"step": 86000 |
|
}, |
|
{ |
|
"epoch": 39.42, |
|
"learning_rate": 4.534219718309859e-05, |
|
"loss": 1.2227, |
|
"step": 87000 |
|
}, |
|
{ |
|
"epoch": 39.87, |
|
"learning_rate": 4.4891943661971834e-05, |
|
"loss": 1.2204, |
|
"step": 88000 |
|
}, |
|
{ |
|
"epoch": 40.33, |
|
"learning_rate": 4.4441690140845077e-05, |
|
"loss": 1.2171, |
|
"step": 89000 |
|
}, |
|
{ |
|
"epoch": 40.78, |
|
"learning_rate": 4.399143661971831e-05, |
|
"loss": 1.2154, |
|
"step": 90000 |
|
}, |
|
{ |
|
"epoch": 41.23, |
|
"learning_rate": 4.3541183098591555e-05, |
|
"loss": 1.2132, |
|
"step": 91000 |
|
}, |
|
{ |
|
"epoch": 41.69, |
|
"learning_rate": 4.30909295774648e-05, |
|
"loss": 1.2118, |
|
"step": 92000 |
|
}, |
|
{ |
|
"epoch": 42.14, |
|
"learning_rate": 4.2640676056338026e-05, |
|
"loss": 1.2088, |
|
"step": 93000 |
|
}, |
|
{ |
|
"epoch": 42.59, |
|
"learning_rate": 4.2189971830985916e-05, |
|
"loss": 1.2068, |
|
"step": 94000 |
|
}, |
|
{ |
|
"epoch": 43.04, |
|
"learning_rate": 4.173971830985916e-05, |
|
"loss": 1.2063, |
|
"step": 95000 |
|
}, |
|
{ |
|
"epoch": 43.5, |
|
"learning_rate": 4.12894647887324e-05, |
|
"loss": 1.2025, |
|
"step": 96000 |
|
}, |
|
{ |
|
"epoch": 43.95, |
|
"learning_rate": 4.083921126760564e-05, |
|
"loss": 1.2013, |
|
"step": 97000 |
|
}, |
|
{ |
|
"epoch": 44.4, |
|
"learning_rate": 4.038895774647888e-05, |
|
"loss": 1.2002, |
|
"step": 98000 |
|
}, |
|
{ |
|
"epoch": 44.86, |
|
"learning_rate": 3.993825352112676e-05, |
|
"loss": 1.198, |
|
"step": 99000 |
|
}, |
|
{ |
|
"epoch": 45.31, |
|
"learning_rate": 3.9488e-05, |
|
"loss": 1.1962, |
|
"step": 100000 |
|
}, |
|
{ |
|
"epoch": 45.76, |
|
"learning_rate": 3.903774647887324e-05, |
|
"loss": 1.1953, |
|
"step": 101000 |
|
}, |
|
{ |
|
"epoch": 46.22, |
|
"learning_rate": 3.8587492957746483e-05, |
|
"loss": 1.1927, |
|
"step": 102000 |
|
}, |
|
{ |
|
"epoch": 46.67, |
|
"learning_rate": 3.813723943661972e-05, |
|
"loss": 1.1909, |
|
"step": 103000 |
|
}, |
|
{ |
|
"epoch": 47.12, |
|
"learning_rate": 3.768698591549296e-05, |
|
"loss": 1.1891, |
|
"step": 104000 |
|
}, |
|
{ |
|
"epoch": 47.58, |
|
"learning_rate": 3.72367323943662e-05, |
|
"loss": 1.1876, |
|
"step": 105000 |
|
}, |
|
{ |
|
"epoch": 48.03, |
|
"learning_rate": 3.678647887323944e-05, |
|
"loss": 1.1868, |
|
"step": 106000 |
|
}, |
|
{ |
|
"epoch": 48.48, |
|
"learning_rate": 3.633622535211268e-05, |
|
"loss": 1.1841, |
|
"step": 107000 |
|
}, |
|
{ |
|
"epoch": 48.94, |
|
"learning_rate": 3.588597183098592e-05, |
|
"loss": 1.1844, |
|
"step": 108000 |
|
}, |
|
{ |
|
"epoch": 49.39, |
|
"learning_rate": 3.543571830985916e-05, |
|
"loss": 1.1815, |
|
"step": 109000 |
|
}, |
|
{ |
|
"epoch": 49.84, |
|
"learning_rate": 3.4985464788732395e-05, |
|
"loss": 1.1809, |
|
"step": 110000 |
|
}, |
|
{ |
|
"epoch": 50.29, |
|
"learning_rate": 3.453521126760564e-05, |
|
"loss": 1.1784, |
|
"step": 111000 |
|
}, |
|
{ |
|
"epoch": 50.75, |
|
"learning_rate": 3.408495774647888e-05, |
|
"loss": 1.1774, |
|
"step": 112000 |
|
}, |
|
{ |
|
"epoch": 51.2, |
|
"learning_rate": 3.3634704225352115e-05, |
|
"loss": 1.1761, |
|
"step": 113000 |
|
}, |
|
{ |
|
"epoch": 51.65, |
|
"learning_rate": 3.318445070422535e-05, |
|
"loss": 1.175, |
|
"step": 114000 |
|
}, |
|
{ |
|
"epoch": 52.11, |
|
"learning_rate": 3.273419718309859e-05, |
|
"loss": 1.1742, |
|
"step": 115000 |
|
}, |
|
{ |
|
"epoch": 52.56, |
|
"learning_rate": 3.2283943661971836e-05, |
|
"loss": 1.1726, |
|
"step": 116000 |
|
}, |
|
{ |
|
"epoch": 53.01, |
|
"learning_rate": 3.183323943661972e-05, |
|
"loss": 1.1719, |
|
"step": 117000 |
|
}, |
|
{ |
|
"epoch": 53.47, |
|
"learning_rate": 3.1382985915492955e-05, |
|
"loss": 1.1699, |
|
"step": 118000 |
|
}, |
|
{ |
|
"epoch": 53.92, |
|
"learning_rate": 3.09327323943662e-05, |
|
"loss": 1.1684, |
|
"step": 119000 |
|
}, |
|
{ |
|
"epoch": 54.37, |
|
"learning_rate": 3.048247887323944e-05, |
|
"loss": 1.1676, |
|
"step": 120000 |
|
}, |
|
{ |
|
"epoch": 54.83, |
|
"learning_rate": 3.003222535211268e-05, |
|
"loss": 1.1662, |
|
"step": 121000 |
|
}, |
|
{ |
|
"epoch": 55.28, |
|
"learning_rate": 2.9581971830985918e-05, |
|
"loss": 1.1647, |
|
"step": 122000 |
|
}, |
|
{ |
|
"epoch": 55.73, |
|
"learning_rate": 2.9131718309859157e-05, |
|
"loss": 1.1643, |
|
"step": 123000 |
|
}, |
|
{ |
|
"epoch": 56.18, |
|
"learning_rate": 2.86814647887324e-05, |
|
"loss": 1.1639, |
|
"step": 124000 |
|
}, |
|
{ |
|
"epoch": 56.64, |
|
"learning_rate": 2.8231211267605635e-05, |
|
"loss": 1.1617, |
|
"step": 125000 |
|
}, |
|
{ |
|
"epoch": 57.09, |
|
"learning_rate": 2.7780957746478874e-05, |
|
"loss": 1.1606, |
|
"step": 126000 |
|
}, |
|
{ |
|
"epoch": 57.54, |
|
"learning_rate": 2.7330704225352116e-05, |
|
"loss": 1.1608, |
|
"step": 127000 |
|
}, |
|
{ |
|
"epoch": 58.0, |
|
"learning_rate": 2.6880450704225355e-05, |
|
"loss": 1.159, |
|
"step": 128000 |
|
}, |
|
{ |
|
"epoch": 58.45, |
|
"learning_rate": 2.643019718309859e-05, |
|
"loss": 1.1572, |
|
"step": 129000 |
|
}, |
|
{ |
|
"epoch": 58.9, |
|
"learning_rate": 2.5979943661971833e-05, |
|
"loss": 1.1572, |
|
"step": 130000 |
|
}, |
|
{ |
|
"epoch": 59.36, |
|
"learning_rate": 2.5529690140845072e-05, |
|
"loss": 1.1562, |
|
"step": 131000 |
|
}, |
|
{ |
|
"epoch": 59.81, |
|
"learning_rate": 2.5079436619718314e-05, |
|
"loss": 1.1551, |
|
"step": 132000 |
|
}, |
|
{ |
|
"epoch": 60.26, |
|
"learning_rate": 2.462918309859155e-05, |
|
"loss": 1.1547, |
|
"step": 133000 |
|
}, |
|
{ |
|
"epoch": 60.72, |
|
"learning_rate": 2.417892957746479e-05, |
|
"loss": 1.153, |
|
"step": 134000 |
|
}, |
|
{ |
|
"epoch": 61.17, |
|
"learning_rate": 2.372867605633803e-05, |
|
"loss": 1.1526, |
|
"step": 135000 |
|
}, |
|
{ |
|
"epoch": 61.62, |
|
"learning_rate": 2.327842253521127e-05, |
|
"loss": 1.1522, |
|
"step": 136000 |
|
}, |
|
{ |
|
"epoch": 62.08, |
|
"learning_rate": 2.2828169014084506e-05, |
|
"loss": 1.1505, |
|
"step": 137000 |
|
}, |
|
{ |
|
"epoch": 62.53, |
|
"learning_rate": 2.2377915492957748e-05, |
|
"loss": 1.1505, |
|
"step": 138000 |
|
}, |
|
{ |
|
"epoch": 62.98, |
|
"learning_rate": 2.1927661971830987e-05, |
|
"loss": 1.1485, |
|
"step": 139000 |
|
}, |
|
{ |
|
"epoch": 63.43, |
|
"learning_rate": 2.1477859154929578e-05, |
|
"loss": 1.1482, |
|
"step": 140000 |
|
}, |
|
{ |
|
"epoch": 63.89, |
|
"learning_rate": 2.102715492957747e-05, |
|
"loss": 1.1466, |
|
"step": 141000 |
|
}, |
|
{ |
|
"epoch": 64.34, |
|
"learning_rate": 2.0576901408450704e-05, |
|
"loss": 1.1467, |
|
"step": 142000 |
|
}, |
|
{ |
|
"epoch": 64.79, |
|
"learning_rate": 2.0126647887323946e-05, |
|
"loss": 1.146, |
|
"step": 143000 |
|
}, |
|
{ |
|
"epoch": 65.25, |
|
"learning_rate": 1.9676394366197185e-05, |
|
"loss": 1.1452, |
|
"step": 144000 |
|
}, |
|
{ |
|
"epoch": 65.7, |
|
"learning_rate": 1.9226140845070424e-05, |
|
"loss": 1.1447, |
|
"step": 145000 |
|
}, |
|
{ |
|
"epoch": 66.15, |
|
"learning_rate": 1.8775887323943663e-05, |
|
"loss": 1.1441, |
|
"step": 146000 |
|
}, |
|
{ |
|
"epoch": 66.61, |
|
"learning_rate": 1.8325633802816902e-05, |
|
"loss": 1.143, |
|
"step": 147000 |
|
}, |
|
{ |
|
"epoch": 67.06, |
|
"learning_rate": 1.787538028169014e-05, |
|
"loss": 1.1433, |
|
"step": 148000 |
|
}, |
|
{ |
|
"epoch": 67.51, |
|
"learning_rate": 1.7425126760563384e-05, |
|
"loss": 1.1414, |
|
"step": 149000 |
|
}, |
|
{ |
|
"epoch": 67.97, |
|
"learning_rate": 1.697487323943662e-05, |
|
"loss": 1.1413, |
|
"step": 150000 |
|
}, |
|
{ |
|
"epoch": 68.42, |
|
"learning_rate": 1.652461971830986e-05, |
|
"loss": 1.14, |
|
"step": 151000 |
|
}, |
|
{ |
|
"epoch": 68.87, |
|
"learning_rate": 1.60743661971831e-05, |
|
"loss": 1.14, |
|
"step": 152000 |
|
}, |
|
{ |
|
"epoch": 69.32, |
|
"learning_rate": 1.562411267605634e-05, |
|
"loss": 1.139, |
|
"step": 153000 |
|
}, |
|
{ |
|
"epoch": 69.78, |
|
"learning_rate": 1.5173859154929578e-05, |
|
"loss": 1.1385, |
|
"step": 154000 |
|
}, |
|
{ |
|
"epoch": 70.23, |
|
"learning_rate": 1.4723605633802817e-05, |
|
"loss": 1.1385, |
|
"step": 155000 |
|
}, |
|
{ |
|
"epoch": 70.68, |
|
"learning_rate": 1.4272901408450706e-05, |
|
"loss": 1.1374, |
|
"step": 156000 |
|
}, |
|
{ |
|
"epoch": 71.14, |
|
"learning_rate": 1.3822647887323945e-05, |
|
"loss": 1.1362, |
|
"step": 157000 |
|
}, |
|
{ |
|
"epoch": 71.59, |
|
"learning_rate": 1.3372394366197183e-05, |
|
"loss": 1.137, |
|
"step": 158000 |
|
}, |
|
{ |
|
"epoch": 72.04, |
|
"learning_rate": 1.2922140845070423e-05, |
|
"loss": 1.1355, |
|
"step": 159000 |
|
}, |
|
{ |
|
"epoch": 72.5, |
|
"learning_rate": 1.2471887323943664e-05, |
|
"loss": 1.1342, |
|
"step": 160000 |
|
}, |
|
{ |
|
"epoch": 72.95, |
|
"learning_rate": 1.2021633802816903e-05, |
|
"loss": 1.1347, |
|
"step": 161000 |
|
}, |
|
{ |
|
"epoch": 73.4, |
|
"learning_rate": 1.157138028169014e-05, |
|
"loss": 1.133, |
|
"step": 162000 |
|
}, |
|
{ |
|
"epoch": 73.86, |
|
"learning_rate": 1.1121126760563381e-05, |
|
"loss": 1.1339, |
|
"step": 163000 |
|
}, |
|
{ |
|
"epoch": 74.31, |
|
"learning_rate": 1.0670873239436622e-05, |
|
"loss": 1.1332, |
|
"step": 164000 |
|
}, |
|
{ |
|
"epoch": 74.76, |
|
"learning_rate": 1.022061971830986e-05, |
|
"loss": 1.1319, |
|
"step": 165000 |
|
}, |
|
{ |
|
"epoch": 75.22, |
|
"learning_rate": 9.7703661971831e-06, |
|
"loss": 1.1311, |
|
"step": 166000 |
|
}, |
|
{ |
|
"epoch": 75.67, |
|
"learning_rate": 9.32056338028169e-06, |
|
"loss": 1.1303, |
|
"step": 167000 |
|
}, |
|
{ |
|
"epoch": 76.12, |
|
"learning_rate": 8.869859154929579e-06, |
|
"loss": 1.1318, |
|
"step": 168000 |
|
}, |
|
{ |
|
"epoch": 76.57, |
|
"learning_rate": 8.419605633802818e-06, |
|
"loss": 1.1304, |
|
"step": 169000 |
|
}, |
|
{ |
|
"epoch": 77.03, |
|
"learning_rate": 7.969352112676057e-06, |
|
"loss": 1.1297, |
|
"step": 170000 |
|
}, |
|
{ |
|
"epoch": 77.48, |
|
"learning_rate": 7.519098591549297e-06, |
|
"loss": 1.1295, |
|
"step": 171000 |
|
}, |
|
{ |
|
"epoch": 77.93, |
|
"learning_rate": 7.068845070422535e-06, |
|
"loss": 1.1291, |
|
"step": 172000 |
|
}, |
|
{ |
|
"epoch": 78.39, |
|
"learning_rate": 6.618591549295776e-06, |
|
"loss": 1.1287, |
|
"step": 173000 |
|
}, |
|
{ |
|
"epoch": 78.84, |
|
"learning_rate": 6.167887323943662e-06, |
|
"loss": 1.128, |
|
"step": 174000 |
|
}, |
|
{ |
|
"epoch": 79.29, |
|
"learning_rate": 5.717633802816902e-06, |
|
"loss": 1.1283, |
|
"step": 175000 |
|
}, |
|
{ |
|
"epoch": 79.75, |
|
"learning_rate": 5.267380281690141e-06, |
|
"loss": 1.1272, |
|
"step": 176000 |
|
}, |
|
{ |
|
"epoch": 80.2, |
|
"learning_rate": 4.817126760563381e-06, |
|
"loss": 1.1273, |
|
"step": 177000 |
|
}, |
|
{ |
|
"epoch": 80.65, |
|
"learning_rate": 4.366873239436621e-06, |
|
"loss": 1.1271, |
|
"step": 178000 |
|
}, |
|
{ |
|
"epoch": 81.11, |
|
"learning_rate": 3.9166197183098595e-06, |
|
"loss": 1.126, |
|
"step": 179000 |
|
}, |
|
{ |
|
"epoch": 81.56, |
|
"learning_rate": 3.4663661971830985e-06, |
|
"loss": 1.1263, |
|
"step": 180000 |
|
}, |
|
{ |
|
"epoch": 82.01, |
|
"learning_rate": 3.016112676056338e-06, |
|
"loss": 1.1261, |
|
"step": 181000 |
|
}, |
|
{ |
|
"epoch": 82.46, |
|
"learning_rate": 2.5658591549295773e-06, |
|
"loss": 1.1247, |
|
"step": 182000 |
|
}, |
|
{ |
|
"epoch": 82.92, |
|
"learning_rate": 2.115605633802817e-06, |
|
"loss": 1.1267, |
|
"step": 183000 |
|
}, |
|
{ |
|
"epoch": 83.37, |
|
"learning_rate": 1.6653521126760563e-06, |
|
"loss": 1.1246, |
|
"step": 184000 |
|
}, |
|
{ |
|
"epoch": 83.82, |
|
"learning_rate": 1.2150985915492959e-06, |
|
"loss": 1.1251, |
|
"step": 185000 |
|
}, |
|
{ |
|
"epoch": 84.28, |
|
"learning_rate": 7.648450704225354e-07, |
|
"loss": 1.1251, |
|
"step": 186000 |
|
}, |
|
{ |
|
"epoch": 84.73, |
|
"learning_rate": 3.145915492957747e-07, |
|
"loss": 1.1246, |
|
"step": 187000 |
|
}, |
|
{ |
|
"epoch": 84.96, |
|
"step": 187500, |
|
"total_flos": 1.0104234636302352e+20, |
|
"train_loss": 0.10515284895833334, |
|
"train_runtime": 4947.3494, |
|
"train_samples_per_second": 77617.32, |
|
"train_steps_per_second": 37.899 |
|
} |
|
], |
|
"max_steps": 187500, |
|
"num_train_epochs": 85, |
|
"total_flos": 1.0104234636302352e+20, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|