GPT2-News / trainer_state.json
TrLOX's picture
Updated
383f842
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 20.0,
"global_step": 274040,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04,
"learning_rate": 4.9908772441979274e-05,
"loss": 2.048,
"step": 500
},
{
"epoch": 0.07,
"learning_rate": 4.981754488395855e-05,
"loss": 2.048,
"step": 1000
},
{
"epoch": 0.11,
"learning_rate": 4.972631732593782e-05,
"loss": 2.048,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 4.9635089767917094e-05,
"loss": 2.048,
"step": 2000
},
{
"epoch": 0.18,
"learning_rate": 4.954386220989637e-05,
"loss": 2.048,
"step": 2500
},
{
"epoch": 0.22,
"learning_rate": 4.945263465187564e-05,
"loss": 2.048,
"step": 3000
},
{
"epoch": 0.26,
"learning_rate": 4.9361407093854914e-05,
"loss": 2.048,
"step": 3500
},
{
"epoch": 0.29,
"learning_rate": 4.9270179535834185e-05,
"loss": 2.048,
"step": 4000
},
{
"epoch": 0.33,
"learning_rate": 4.9178951977813456e-05,
"loss": 2.048,
"step": 4500
},
{
"epoch": 0.36,
"learning_rate": 4.9087724419792734e-05,
"loss": 2.048,
"step": 5000
},
{
"epoch": 0.4,
"learning_rate": 4.8996496861772005e-05,
"loss": 2.048,
"step": 5500
},
{
"epoch": 0.44,
"learning_rate": 4.8905269303751276e-05,
"loss": 2.048,
"step": 6000
},
{
"epoch": 0.47,
"learning_rate": 4.8814041745730554e-05,
"loss": 2.048,
"step": 6500
},
{
"epoch": 0.51,
"learning_rate": 4.8722814187709825e-05,
"loss": 2.048,
"step": 7000
},
{
"epoch": 0.55,
"learning_rate": 4.8631586629689096e-05,
"loss": 2.048,
"step": 7500
},
{
"epoch": 0.58,
"learning_rate": 4.8540359071668374e-05,
"loss": 2.048,
"step": 8000
},
{
"epoch": 0.62,
"learning_rate": 4.8449131513647645e-05,
"loss": 2.048,
"step": 8500
},
{
"epoch": 0.66,
"learning_rate": 4.8357903955626916e-05,
"loss": 2.048,
"step": 9000
},
{
"epoch": 0.69,
"learning_rate": 4.8266676397606194e-05,
"loss": 2.048,
"step": 9500
},
{
"epoch": 0.73,
"learning_rate": 4.8175448839585465e-05,
"loss": 2.048,
"step": 10000
},
{
"epoch": 0.77,
"learning_rate": 4.808422128156474e-05,
"loss": 2.048,
"step": 10500
},
{
"epoch": 0.8,
"learning_rate": 4.7992993723544014e-05,
"loss": 2.048,
"step": 11000
},
{
"epoch": 0.84,
"learning_rate": 4.7901766165523285e-05,
"loss": 2.048,
"step": 11500
},
{
"epoch": 0.88,
"learning_rate": 4.7810538607502556e-05,
"loss": 2.048,
"step": 12000
},
{
"epoch": 0.91,
"learning_rate": 4.771931104948183e-05,
"loss": 2.048,
"step": 12500
},
{
"epoch": 0.95,
"learning_rate": 4.76280834914611e-05,
"loss": 2.048,
"step": 13000
},
{
"epoch": 0.99,
"learning_rate": 4.7536855933440376e-05,
"loss": 2.048,
"step": 13500
},
{
"epoch": 1.02,
"learning_rate": 4.744562837541965e-05,
"loss": 2.048,
"step": 14000
},
{
"epoch": 1.06,
"learning_rate": 4.735440081739892e-05,
"loss": 2.048,
"step": 14500
},
{
"epoch": 1.09,
"learning_rate": 4.7263173259378197e-05,
"loss": 2.048,
"step": 15000
},
{
"epoch": 1.13,
"learning_rate": 4.717194570135747e-05,
"loss": 2.048,
"step": 15500
},
{
"epoch": 1.17,
"learning_rate": 4.7080718143336746e-05,
"loss": 2.048,
"step": 16000
},
{
"epoch": 1.2,
"learning_rate": 4.698949058531602e-05,
"loss": 2.048,
"step": 16500
},
{
"epoch": 1.24,
"learning_rate": 4.689826302729529e-05,
"loss": 2.048,
"step": 17000
},
{
"epoch": 1.28,
"learning_rate": 4.6807035469274566e-05,
"loss": 2.048,
"step": 17500
},
{
"epoch": 1.31,
"learning_rate": 4.671580791125384e-05,
"loss": 2.048,
"step": 18000
},
{
"epoch": 1.35,
"learning_rate": 4.662458035323311e-05,
"loss": 2.048,
"step": 18500
},
{
"epoch": 1.39,
"learning_rate": 4.653335279521238e-05,
"loss": 2.048,
"step": 19000
},
{
"epoch": 1.42,
"learning_rate": 4.644212523719165e-05,
"loss": 2.048,
"step": 19500
},
{
"epoch": 1.46,
"learning_rate": 4.635089767917092e-05,
"loss": 2.048,
"step": 20000
},
{
"epoch": 1.5,
"learning_rate": 4.62596701211502e-05,
"loss": 2.048,
"step": 20500
},
{
"epoch": 1.53,
"learning_rate": 4.616844256312947e-05,
"loss": 2.048,
"step": 21000
},
{
"epoch": 1.57,
"learning_rate": 4.607721500510875e-05,
"loss": 2.048,
"step": 21500
},
{
"epoch": 1.61,
"learning_rate": 4.598598744708802e-05,
"loss": 2.048,
"step": 22000
},
{
"epoch": 1.64,
"learning_rate": 4.589475988906729e-05,
"loss": 2.048,
"step": 22500
},
{
"epoch": 1.68,
"learning_rate": 4.580353233104657e-05,
"loss": 2.048,
"step": 23000
},
{
"epoch": 1.72,
"learning_rate": 4.571230477302584e-05,
"loss": 2.048,
"step": 23500
},
{
"epoch": 1.75,
"learning_rate": 4.562107721500511e-05,
"loss": 2.048,
"step": 24000
},
{
"epoch": 1.79,
"learning_rate": 4.552984965698439e-05,
"loss": 2.048,
"step": 24500
},
{
"epoch": 1.82,
"learning_rate": 4.543862209896366e-05,
"loss": 2.048,
"step": 25000
},
{
"epoch": 1.86,
"learning_rate": 4.534739454094293e-05,
"loss": 2.048,
"step": 25500
},
{
"epoch": 1.9,
"learning_rate": 4.52561669829222e-05,
"loss": 2.048,
"step": 26000
},
{
"epoch": 1.93,
"learning_rate": 4.516493942490147e-05,
"loss": 2.048,
"step": 26500
},
{
"epoch": 1.97,
"learning_rate": 4.5073711866880744e-05,
"loss": 2.048,
"step": 27000
},
{
"epoch": 2.01,
"learning_rate": 4.498248430886002e-05,
"loss": 2.048,
"step": 27500
},
{
"epoch": 2.04,
"learning_rate": 4.489125675083929e-05,
"loss": 2.048,
"step": 28000
},
{
"epoch": 2.08,
"learning_rate": 4.480002919281857e-05,
"loss": 2.048,
"step": 28500
},
{
"epoch": 2.12,
"learning_rate": 4.470880163479784e-05,
"loss": 2.048,
"step": 29000
},
{
"epoch": 2.15,
"learning_rate": 4.461757407677711e-05,
"loss": 2.048,
"step": 29500
},
{
"epoch": 2.19,
"learning_rate": 4.452634651875639e-05,
"loss": 2.048,
"step": 30000
},
{
"epoch": 2.23,
"learning_rate": 4.443511896073566e-05,
"loss": 2.048,
"step": 30500
},
{
"epoch": 2.26,
"learning_rate": 4.434389140271493e-05,
"loss": 2.048,
"step": 31000
},
{
"epoch": 2.3,
"learning_rate": 4.425266384469421e-05,
"loss": 2.048,
"step": 31500
},
{
"epoch": 2.34,
"learning_rate": 4.416143628667348e-05,
"loss": 2.048,
"step": 32000
},
{
"epoch": 2.37,
"learning_rate": 4.407020872865275e-05,
"loss": 2.048,
"step": 32500
},
{
"epoch": 2.41,
"learning_rate": 4.397898117063203e-05,
"loss": 2.048,
"step": 33000
},
{
"epoch": 2.44,
"learning_rate": 4.38877536126113e-05,
"loss": 2.048,
"step": 33500
},
{
"epoch": 2.48,
"learning_rate": 4.379652605459057e-05,
"loss": 2.048,
"step": 34000
},
{
"epoch": 2.52,
"learning_rate": 4.3705298496569844e-05,
"loss": 2.048,
"step": 34500
},
{
"epoch": 2.55,
"learning_rate": 4.3614070938549115e-05,
"loss": 2.048,
"step": 35000
},
{
"epoch": 2.59,
"learning_rate": 4.352284338052839e-05,
"loss": 2.048,
"step": 35500
},
{
"epoch": 2.63,
"learning_rate": 4.3431615822507664e-05,
"loss": 2.048,
"step": 36000
},
{
"epoch": 2.66,
"learning_rate": 4.3340388264486935e-05,
"loss": 2.048,
"step": 36500
},
{
"epoch": 2.7,
"learning_rate": 4.324916070646621e-05,
"loss": 2.048,
"step": 37000
},
{
"epoch": 2.74,
"learning_rate": 4.3157933148445484e-05,
"loss": 2.048,
"step": 37500
},
{
"epoch": 2.77,
"learning_rate": 4.3066705590424756e-05,
"loss": 2.048,
"step": 38000
},
{
"epoch": 2.81,
"learning_rate": 4.2975478032404033e-05,
"loss": 2.048,
"step": 38500
},
{
"epoch": 2.85,
"learning_rate": 4.2884250474383305e-05,
"loss": 2.048,
"step": 39000
},
{
"epoch": 2.88,
"learning_rate": 4.279302291636258e-05,
"loss": 2.048,
"step": 39500
},
{
"epoch": 2.92,
"learning_rate": 4.2701795358341853e-05,
"loss": 2.048,
"step": 40000
},
{
"epoch": 2.96,
"learning_rate": 4.2610567800321125e-05,
"loss": 2.048,
"step": 40500
},
{
"epoch": 2.99,
"learning_rate": 4.2519340242300396e-05,
"loss": 2.048,
"step": 41000
},
{
"epoch": 3.03,
"learning_rate": 4.242811268427967e-05,
"loss": 2.048,
"step": 41500
},
{
"epoch": 3.07,
"learning_rate": 4.233688512625894e-05,
"loss": 2.048,
"step": 42000
},
{
"epoch": 3.1,
"learning_rate": 4.2245657568238216e-05,
"loss": 2.048,
"step": 42500
},
{
"epoch": 3.14,
"learning_rate": 4.215443001021749e-05,
"loss": 2.048,
"step": 43000
},
{
"epoch": 3.17,
"learning_rate": 4.206320245219676e-05,
"loss": 2.048,
"step": 43500
},
{
"epoch": 3.21,
"learning_rate": 4.1971974894176036e-05,
"loss": 2.048,
"step": 44000
},
{
"epoch": 3.25,
"learning_rate": 4.188074733615531e-05,
"loss": 2.048,
"step": 44500
},
{
"epoch": 3.28,
"learning_rate": 4.178951977813458e-05,
"loss": 2.048,
"step": 45000
},
{
"epoch": 3.32,
"learning_rate": 4.1698292220113856e-05,
"loss": 2.048,
"step": 45500
},
{
"epoch": 3.36,
"learning_rate": 4.160706466209313e-05,
"loss": 2.048,
"step": 46000
},
{
"epoch": 3.39,
"learning_rate": 4.1515837104072405e-05,
"loss": 2.048,
"step": 46500
},
{
"epoch": 3.43,
"learning_rate": 4.1424609546051676e-05,
"loss": 2.048,
"step": 47000
},
{
"epoch": 3.47,
"learning_rate": 4.133338198803095e-05,
"loss": 2.048,
"step": 47500
},
{
"epoch": 3.5,
"learning_rate": 4.124215443001022e-05,
"loss": 2.048,
"step": 48000
},
{
"epoch": 3.54,
"learning_rate": 4.115092687198949e-05,
"loss": 2.048,
"step": 48500
},
{
"epoch": 3.58,
"learning_rate": 4.105969931396876e-05,
"loss": 2.048,
"step": 49000
},
{
"epoch": 3.61,
"learning_rate": 4.096847175594804e-05,
"loss": 2.048,
"step": 49500
},
{
"epoch": 3.65,
"learning_rate": 4.087724419792731e-05,
"loss": 2.048,
"step": 50000
},
{
"epoch": 3.69,
"learning_rate": 4.078601663990658e-05,
"loss": 2.048,
"step": 50500
},
{
"epoch": 3.72,
"learning_rate": 4.069478908188586e-05,
"loss": 2.048,
"step": 51000
},
{
"epoch": 3.76,
"learning_rate": 4.060356152386513e-05,
"loss": 2.048,
"step": 51500
},
{
"epoch": 3.8,
"learning_rate": 4.051233396584441e-05,
"loss": 2.048,
"step": 52000
},
{
"epoch": 3.83,
"learning_rate": 4.042110640782368e-05,
"loss": 2.048,
"step": 52500
},
{
"epoch": 3.87,
"learning_rate": 4.032987884980295e-05,
"loss": 2.048,
"step": 53000
},
{
"epoch": 3.9,
"learning_rate": 4.023865129178223e-05,
"loss": 2.048,
"step": 53500
},
{
"epoch": 3.94,
"learning_rate": 4.01474237337615e-05,
"loss": 2.048,
"step": 54000
},
{
"epoch": 3.98,
"learning_rate": 4.005619617574077e-05,
"loss": 2.048,
"step": 54500
},
{
"epoch": 4.01,
"learning_rate": 3.996496861772005e-05,
"loss": 2.048,
"step": 55000
},
{
"epoch": 4.05,
"learning_rate": 3.987374105969932e-05,
"loss": 2.048,
"step": 55500
},
{
"epoch": 4.09,
"learning_rate": 3.978251350167859e-05,
"loss": 2.048,
"step": 56000
},
{
"epoch": 4.12,
"learning_rate": 3.969128594365786e-05,
"loss": 2.048,
"step": 56500
},
{
"epoch": 4.16,
"learning_rate": 3.960005838563713e-05,
"loss": 2.048,
"step": 57000
},
{
"epoch": 4.2,
"learning_rate": 3.950883082761641e-05,
"loss": 2.048,
"step": 57500
},
{
"epoch": 4.23,
"learning_rate": 3.941760326959568e-05,
"loss": 2.048,
"step": 58000
},
{
"epoch": 4.27,
"learning_rate": 3.932637571157495e-05,
"loss": 2.048,
"step": 58500
},
{
"epoch": 4.31,
"learning_rate": 3.923514815355423e-05,
"loss": 2.048,
"step": 59000
},
{
"epoch": 4.34,
"learning_rate": 3.91439205955335e-05,
"loss": 2.048,
"step": 59500
},
{
"epoch": 4.38,
"learning_rate": 3.905269303751277e-05,
"loss": 2.048,
"step": 60000
},
{
"epoch": 4.42,
"learning_rate": 3.896146547949205e-05,
"loss": 2.048,
"step": 60500
},
{
"epoch": 4.45,
"learning_rate": 3.887023792147132e-05,
"loss": 2.048,
"step": 61000
},
{
"epoch": 4.49,
"learning_rate": 3.877901036345059e-05,
"loss": 2.048,
"step": 61500
},
{
"epoch": 4.52,
"learning_rate": 3.868778280542987e-05,
"loss": 2.048,
"step": 62000
},
{
"epoch": 4.56,
"learning_rate": 3.859655524740914e-05,
"loss": 2.048,
"step": 62500
},
{
"epoch": 4.6,
"learning_rate": 3.850532768938841e-05,
"loss": 2.048,
"step": 63000
},
{
"epoch": 4.63,
"learning_rate": 3.8414100131367684e-05,
"loss": 2.048,
"step": 63500
},
{
"epoch": 4.67,
"learning_rate": 3.8322872573346955e-05,
"loss": 2.048,
"step": 64000
},
{
"epoch": 4.71,
"learning_rate": 3.823164501532623e-05,
"loss": 2.048,
"step": 64500
},
{
"epoch": 4.74,
"learning_rate": 3.8140417457305504e-05,
"loss": 2.048,
"step": 65000
},
{
"epoch": 4.78,
"learning_rate": 3.8049189899284775e-05,
"loss": 2.048,
"step": 65500
},
{
"epoch": 4.82,
"learning_rate": 3.795796234126405e-05,
"loss": 2.048,
"step": 66000
},
{
"epoch": 4.85,
"learning_rate": 3.7866734783243324e-05,
"loss": 2.048,
"step": 66500
},
{
"epoch": 4.89,
"learning_rate": 3.7775507225222595e-05,
"loss": 2.048,
"step": 67000
},
{
"epoch": 4.93,
"learning_rate": 3.768427966720187e-05,
"loss": 2.048,
"step": 67500
},
{
"epoch": 4.96,
"learning_rate": 3.7593052109181144e-05,
"loss": 2.048,
"step": 68000
},
{
"epoch": 5.0,
"learning_rate": 3.7501824551160415e-05,
"loss": 2.048,
"step": 68500
},
{
"epoch": 5.04,
"learning_rate": 3.741059699313969e-05,
"loss": 2.048,
"step": 69000
},
{
"epoch": 5.07,
"learning_rate": 3.7319369435118964e-05,
"loss": 2.048,
"step": 69500
},
{
"epoch": 5.11,
"learning_rate": 3.7228141877098235e-05,
"loss": 2.048,
"step": 70000
},
{
"epoch": 5.15,
"learning_rate": 3.7136914319077506e-05,
"loss": 2.048,
"step": 70500
},
{
"epoch": 5.18,
"learning_rate": 3.704568676105678e-05,
"loss": 2.048,
"step": 71000
},
{
"epoch": 5.22,
"learning_rate": 3.6954459203036055e-05,
"loss": 2.048,
"step": 71500
},
{
"epoch": 5.25,
"learning_rate": 3.6863231645015326e-05,
"loss": 2.048,
"step": 72000
},
{
"epoch": 5.29,
"learning_rate": 3.67720040869946e-05,
"loss": 2.048,
"step": 72500
},
{
"epoch": 5.33,
"learning_rate": 3.6680776528973875e-05,
"loss": 2.048,
"step": 73000
},
{
"epoch": 5.36,
"learning_rate": 3.6589548970953146e-05,
"loss": 2.048,
"step": 73500
},
{
"epoch": 5.4,
"learning_rate": 3.649832141293242e-05,
"loss": 2.048,
"step": 74000
},
{
"epoch": 5.44,
"learning_rate": 3.6407093854911695e-05,
"loss": 2.048,
"step": 74500
},
{
"epoch": 5.47,
"learning_rate": 3.6315866296890966e-05,
"loss": 2.048,
"step": 75000
},
{
"epoch": 5.51,
"learning_rate": 3.6224638738870244e-05,
"loss": 2.048,
"step": 75500
},
{
"epoch": 5.55,
"learning_rate": 3.6133411180849515e-05,
"loss": 2.048,
"step": 76000
},
{
"epoch": 5.58,
"learning_rate": 3.6042183622828787e-05,
"loss": 2.048,
"step": 76500
},
{
"epoch": 5.62,
"learning_rate": 3.5950956064808064e-05,
"loss": 2.048,
"step": 77000
},
{
"epoch": 5.66,
"learning_rate": 3.5859728506787336e-05,
"loss": 2.048,
"step": 77500
},
{
"epoch": 5.69,
"learning_rate": 3.576850094876661e-05,
"loss": 2.048,
"step": 78000
},
{
"epoch": 5.73,
"learning_rate": 3.567727339074588e-05,
"loss": 2.048,
"step": 78500
},
{
"epoch": 5.77,
"learning_rate": 3.558604583272515e-05,
"loss": 2.048,
"step": 79000
},
{
"epoch": 5.8,
"learning_rate": 3.549481827470442e-05,
"loss": 2.048,
"step": 79500
},
{
"epoch": 5.84,
"learning_rate": 3.54035907166837e-05,
"loss": 2.048,
"step": 80000
},
{
"epoch": 5.88,
"learning_rate": 3.531236315866297e-05,
"loss": 2.048,
"step": 80500
},
{
"epoch": 5.91,
"learning_rate": 3.522113560064224e-05,
"loss": 2.048,
"step": 81000
},
{
"epoch": 5.95,
"learning_rate": 3.512990804262152e-05,
"loss": 2.048,
"step": 81500
},
{
"epoch": 5.98,
"learning_rate": 3.503868048460079e-05,
"loss": 2.048,
"step": 82000
},
{
"epoch": 6.02,
"learning_rate": 3.494745292658007e-05,
"loss": 2.048,
"step": 82500
},
{
"epoch": 6.06,
"learning_rate": 3.485622536855934e-05,
"loss": 2.048,
"step": 83000
},
{
"epoch": 6.09,
"learning_rate": 3.476499781053861e-05,
"loss": 2.048,
"step": 83500
},
{
"epoch": 6.13,
"learning_rate": 3.467377025251789e-05,
"loss": 2.048,
"step": 84000
},
{
"epoch": 6.17,
"learning_rate": 3.458254269449716e-05,
"loss": 2.048,
"step": 84500
},
{
"epoch": 6.2,
"learning_rate": 3.449131513647643e-05,
"loss": 2.048,
"step": 85000
},
{
"epoch": 6.24,
"learning_rate": 3.44000875784557e-05,
"loss": 2.048,
"step": 85500
},
{
"epoch": 6.28,
"learning_rate": 3.430886002043497e-05,
"loss": 2.048,
"step": 86000
},
{
"epoch": 6.31,
"learning_rate": 3.421763246241424e-05,
"loss": 2.048,
"step": 86500
},
{
"epoch": 6.35,
"learning_rate": 3.412640490439352e-05,
"loss": 2.048,
"step": 87000
},
{
"epoch": 6.39,
"learning_rate": 3.403517734637279e-05,
"loss": 2.048,
"step": 87500
},
{
"epoch": 6.42,
"learning_rate": 3.394394978835207e-05,
"loss": 2.048,
"step": 88000
},
{
"epoch": 6.46,
"learning_rate": 3.385272223033134e-05,
"loss": 2.048,
"step": 88500
},
{
"epoch": 6.5,
"learning_rate": 3.376149467231061e-05,
"loss": 2.048,
"step": 89000
},
{
"epoch": 6.53,
"learning_rate": 3.367026711428989e-05,
"loss": 2.048,
"step": 89500
},
{
"epoch": 6.57,
"learning_rate": 3.357903955626916e-05,
"loss": 2.048,
"step": 90000
},
{
"epoch": 6.6,
"learning_rate": 3.348781199824843e-05,
"loss": 2.048,
"step": 90500
},
{
"epoch": 6.64,
"learning_rate": 3.339658444022771e-05,
"loss": 2.048,
"step": 91000
},
{
"epoch": 6.68,
"learning_rate": 3.330535688220698e-05,
"loss": 2.048,
"step": 91500
},
{
"epoch": 6.71,
"learning_rate": 3.321412932418625e-05,
"loss": 2.048,
"step": 92000
},
{
"epoch": 6.75,
"learning_rate": 3.312290176616552e-05,
"loss": 2.048,
"step": 92500
},
{
"epoch": 6.79,
"learning_rate": 3.3031674208144794e-05,
"loss": 2.048,
"step": 93000
},
{
"epoch": 6.82,
"learning_rate": 3.294044665012407e-05,
"loss": 2.048,
"step": 93500
},
{
"epoch": 6.86,
"learning_rate": 3.284921909210334e-05,
"loss": 2.048,
"step": 94000
},
{
"epoch": 6.9,
"learning_rate": 3.2757991534082614e-05,
"loss": 2.048,
"step": 94500
},
{
"epoch": 6.93,
"learning_rate": 3.266676397606189e-05,
"loss": 2.048,
"step": 95000
},
{
"epoch": 6.97,
"learning_rate": 3.257553641804116e-05,
"loss": 2.048,
"step": 95500
},
{
"epoch": 7.01,
"learning_rate": 3.2484308860020434e-05,
"loss": 2.048,
"step": 96000
},
{
"epoch": 7.04,
"learning_rate": 3.239308130199971e-05,
"loss": 2.048,
"step": 96500
},
{
"epoch": 7.08,
"learning_rate": 3.230185374397898e-05,
"loss": 2.048,
"step": 97000
},
{
"epoch": 7.12,
"learning_rate": 3.2210626185958254e-05,
"loss": 2.048,
"step": 97500
},
{
"epoch": 7.15,
"learning_rate": 3.211939862793753e-05,
"loss": 2.048,
"step": 98000
},
{
"epoch": 7.19,
"learning_rate": 3.20281710699168e-05,
"loss": 2.048,
"step": 98500
},
{
"epoch": 7.23,
"learning_rate": 3.1936943511896074e-05,
"loss": 2.048,
"step": 99000
},
{
"epoch": 7.26,
"learning_rate": 3.184571595387535e-05,
"loss": 2.048,
"step": 99500
},
{
"epoch": 7.3,
"learning_rate": 3.175448839585462e-05,
"loss": 2.048,
"step": 100000
},
{
"epoch": 7.33,
"learning_rate": 3.1663260837833894e-05,
"loss": 2.048,
"step": 100500
},
{
"epoch": 7.37,
"learning_rate": 3.1572033279813166e-05,
"loss": 2.048,
"step": 101000
},
{
"epoch": 7.41,
"learning_rate": 3.148080572179244e-05,
"loss": 2.048,
"step": 101500
},
{
"epoch": 7.44,
"learning_rate": 3.1389578163771715e-05,
"loss": 2.048,
"step": 102000
},
{
"epoch": 7.48,
"learning_rate": 3.1298350605750986e-05,
"loss": 2.048,
"step": 102500
},
{
"epoch": 7.52,
"learning_rate": 3.120712304773026e-05,
"loss": 2.048,
"step": 103000
},
{
"epoch": 7.55,
"learning_rate": 3.1115895489709535e-05,
"loss": 2.048,
"step": 103500
},
{
"epoch": 7.59,
"learning_rate": 3.1024667931688806e-05,
"loss": 2.048,
"step": 104000
},
{
"epoch": 7.63,
"learning_rate": 3.093344037366808e-05,
"loss": 2.048,
"step": 104500
},
{
"epoch": 7.66,
"learning_rate": 3.0842212815647355e-05,
"loss": 2.048,
"step": 105000
},
{
"epoch": 7.7,
"learning_rate": 3.0750985257626626e-05,
"loss": 2.048,
"step": 105500
},
{
"epoch": 7.74,
"learning_rate": 3.0659757699605904e-05,
"loss": 2.048,
"step": 106000
},
{
"epoch": 7.77,
"learning_rate": 3.0568530141585175e-05,
"loss": 2.048,
"step": 106500
},
{
"epoch": 7.81,
"learning_rate": 3.0477302583564443e-05,
"loss": 2.048,
"step": 107000
},
{
"epoch": 7.85,
"learning_rate": 3.038607502554372e-05,
"loss": 2.048,
"step": 107500
},
{
"epoch": 7.88,
"learning_rate": 3.029484746752299e-05,
"loss": 2.048,
"step": 108000
},
{
"epoch": 7.92,
"learning_rate": 3.0203619909502263e-05,
"loss": 2.048,
"step": 108500
},
{
"epoch": 7.96,
"learning_rate": 3.0112392351481537e-05,
"loss": 2.048,
"step": 109000
},
{
"epoch": 7.99,
"learning_rate": 3.0021164793460808e-05,
"loss": 2.048,
"step": 109500
},
{
"epoch": 8.03,
"learning_rate": 2.992993723544008e-05,
"loss": 2.048,
"step": 110000
},
{
"epoch": 8.06,
"learning_rate": 2.9838709677419357e-05,
"loss": 2.048,
"step": 110500
},
{
"epoch": 8.1,
"learning_rate": 2.974748211939863e-05,
"loss": 2.048,
"step": 111000
},
{
"epoch": 8.14,
"learning_rate": 2.96562545613779e-05,
"loss": 2.048,
"step": 111500
},
{
"epoch": 8.17,
"learning_rate": 2.9565027003357177e-05,
"loss": 2.048,
"step": 112000
},
{
"epoch": 8.21,
"learning_rate": 2.947379944533645e-05,
"loss": 2.048,
"step": 112500
},
{
"epoch": 8.25,
"learning_rate": 2.9382571887315723e-05,
"loss": 2.048,
"step": 113000
},
{
"epoch": 8.28,
"learning_rate": 2.9291344329294994e-05,
"loss": 2.048,
"step": 113500
},
{
"epoch": 8.32,
"learning_rate": 2.9200116771274265e-05,
"loss": 2.048,
"step": 114000
},
{
"epoch": 8.36,
"learning_rate": 2.9108889213253543e-05,
"loss": 2.048,
"step": 114500
},
{
"epoch": 8.39,
"learning_rate": 2.9017661655232814e-05,
"loss": 2.048,
"step": 115000
},
{
"epoch": 8.43,
"learning_rate": 2.8926434097212085e-05,
"loss": 2.048,
"step": 115500
},
{
"epoch": 8.47,
"learning_rate": 2.8835206539191363e-05,
"loss": 2.048,
"step": 116000
},
{
"epoch": 8.5,
"learning_rate": 2.8743978981170634e-05,
"loss": 2.048,
"step": 116500
},
{
"epoch": 8.54,
"learning_rate": 2.8652751423149905e-05,
"loss": 2.048,
"step": 117000
},
{
"epoch": 8.58,
"learning_rate": 2.856152386512918e-05,
"loss": 2.048,
"step": 117500
},
{
"epoch": 8.61,
"learning_rate": 2.847029630710845e-05,
"loss": 2.048,
"step": 118000
},
{
"epoch": 8.65,
"learning_rate": 2.837906874908773e-05,
"loss": 2.048,
"step": 118500
},
{
"epoch": 8.68,
"learning_rate": 2.8287841191067e-05,
"loss": 2.048,
"step": 119000
},
{
"epoch": 8.72,
"learning_rate": 2.819661363304627e-05,
"loss": 2.048,
"step": 119500
},
{
"epoch": 8.76,
"learning_rate": 2.8105386075025546e-05,
"loss": 2.048,
"step": 120000
},
{
"epoch": 8.79,
"learning_rate": 2.8014158517004817e-05,
"loss": 2.048,
"step": 120500
},
{
"epoch": 8.83,
"learning_rate": 2.7922930958984088e-05,
"loss": 2.048,
"step": 121000
},
{
"epoch": 8.87,
"learning_rate": 2.7831703400963366e-05,
"loss": 2.048,
"step": 121500
},
{
"epoch": 8.9,
"learning_rate": 2.7740475842942637e-05,
"loss": 2.048,
"step": 122000
},
{
"epoch": 8.94,
"learning_rate": 2.7649248284921908e-05,
"loss": 2.048,
"step": 122500
},
{
"epoch": 8.98,
"learning_rate": 2.7558020726901186e-05,
"loss": 2.048,
"step": 123000
},
{
"epoch": 9.01,
"learning_rate": 2.7466793168880457e-05,
"loss": 2.048,
"step": 123500
},
{
"epoch": 9.05,
"learning_rate": 2.737556561085973e-05,
"loss": 2.048,
"step": 124000
},
{
"epoch": 9.09,
"learning_rate": 2.7284338052839002e-05,
"loss": 2.048,
"step": 124500
},
{
"epoch": 9.12,
"learning_rate": 2.7193110494818274e-05,
"loss": 2.048,
"step": 125000
},
{
"epoch": 9.16,
"learning_rate": 2.710188293679755e-05,
"loss": 2.048,
"step": 125500
},
{
"epoch": 9.2,
"learning_rate": 2.7010655378776823e-05,
"loss": 2.048,
"step": 126000
},
{
"epoch": 9.23,
"learning_rate": 2.6919427820756094e-05,
"loss": 2.048,
"step": 126500
},
{
"epoch": 9.27,
"learning_rate": 2.682820026273537e-05,
"loss": 2.048,
"step": 127000
},
{
"epoch": 9.31,
"learning_rate": 2.6736972704714643e-05,
"loss": 2.048,
"step": 127500
},
{
"epoch": 9.34,
"learning_rate": 2.6645745146693914e-05,
"loss": 2.048,
"step": 128000
},
{
"epoch": 9.38,
"learning_rate": 2.6554517588673188e-05,
"loss": 2.048,
"step": 128500
},
{
"epoch": 9.41,
"learning_rate": 2.646329003065246e-05,
"loss": 2.048,
"step": 129000
},
{
"epoch": 9.45,
"learning_rate": 2.637206247263173e-05,
"loss": 2.048,
"step": 129500
},
{
"epoch": 9.49,
"learning_rate": 2.628083491461101e-05,
"loss": 2.048,
"step": 130000
},
{
"epoch": 9.52,
"learning_rate": 2.618960735659028e-05,
"loss": 2.048,
"step": 130500
},
{
"epoch": 9.56,
"learning_rate": 2.6098379798569554e-05,
"loss": 2.048,
"step": 131000
},
{
"epoch": 9.6,
"learning_rate": 2.6007152240548825e-05,
"loss": 2.048,
"step": 131500
},
{
"epoch": 9.63,
"learning_rate": 2.5915924682528096e-05,
"loss": 2.048,
"step": 132000
},
{
"epoch": 9.67,
"learning_rate": 2.5824697124507374e-05,
"loss": 2.048,
"step": 132500
},
{
"epoch": 9.71,
"learning_rate": 2.5733469566486645e-05,
"loss": 2.048,
"step": 133000
},
{
"epoch": 9.74,
"learning_rate": 2.5642242008465916e-05,
"loss": 2.048,
"step": 133500
},
{
"epoch": 9.78,
"learning_rate": 2.5551014450445194e-05,
"loss": 2.048,
"step": 134000
},
{
"epoch": 9.82,
"learning_rate": 2.5459786892424465e-05,
"loss": 2.048,
"step": 134500
},
{
"epoch": 9.85,
"learning_rate": 2.5368559334403736e-05,
"loss": 2.048,
"step": 135000
},
{
"epoch": 9.89,
"learning_rate": 2.527733177638301e-05,
"loss": 2.048,
"step": 135500
},
{
"epoch": 9.93,
"learning_rate": 2.5186104218362282e-05,
"loss": 2.048,
"step": 136000
},
{
"epoch": 9.96,
"learning_rate": 2.509487666034156e-05,
"loss": 2.048,
"step": 136500
},
{
"epoch": 10.0,
"learning_rate": 2.500364910232083e-05,
"loss": 2.048,
"step": 137000
},
{
"epoch": 10.04,
"learning_rate": 2.4912421544300105e-05,
"loss": 2.048,
"step": 137500
},
{
"epoch": 10.07,
"learning_rate": 2.4821193986279376e-05,
"loss": 2.048,
"step": 138000
},
{
"epoch": 10.11,
"learning_rate": 2.472996642825865e-05,
"loss": 2.048,
"step": 138500
},
{
"epoch": 10.14,
"learning_rate": 2.4638738870237922e-05,
"loss": 2.048,
"step": 139000
},
{
"epoch": 10.18,
"learning_rate": 2.4547511312217193e-05,
"loss": 2.048,
"step": 139500
},
{
"epoch": 10.22,
"learning_rate": 2.4456283754196468e-05,
"loss": 2.048,
"step": 140000
},
{
"epoch": 10.25,
"learning_rate": 2.4365056196175742e-05,
"loss": 2.048,
"step": 140500
},
{
"epoch": 10.29,
"learning_rate": 2.4273828638155017e-05,
"loss": 2.048,
"step": 141000
},
{
"epoch": 10.33,
"learning_rate": 2.4182601080134288e-05,
"loss": 2.048,
"step": 141500
},
{
"epoch": 10.36,
"learning_rate": 2.4091373522113562e-05,
"loss": 2.048,
"step": 142000
},
{
"epoch": 10.4,
"learning_rate": 2.4000145964092833e-05,
"loss": 2.048,
"step": 142500
},
{
"epoch": 10.44,
"learning_rate": 2.3908918406072104e-05,
"loss": 2.048,
"step": 143000
},
{
"epoch": 10.47,
"learning_rate": 2.381769084805138e-05,
"loss": 2.048,
"step": 143500
},
{
"epoch": 10.51,
"learning_rate": 2.3726463290030653e-05,
"loss": 2.048,
"step": 144000
},
{
"epoch": 10.55,
"learning_rate": 2.3635235732009928e-05,
"loss": 2.048,
"step": 144500
},
{
"epoch": 10.58,
"learning_rate": 2.35440081739892e-05,
"loss": 2.048,
"step": 145000
},
{
"epoch": 10.62,
"learning_rate": 2.3452780615968474e-05,
"loss": 2.048,
"step": 145500
},
{
"epoch": 10.66,
"learning_rate": 2.3361553057947748e-05,
"loss": 2.048,
"step": 146000
},
{
"epoch": 10.69,
"learning_rate": 2.327032549992702e-05,
"loss": 2.048,
"step": 146500
},
{
"epoch": 10.73,
"learning_rate": 2.317909794190629e-05,
"loss": 2.048,
"step": 147000
},
{
"epoch": 10.76,
"learning_rate": 2.3087870383885565e-05,
"loss": 2.048,
"step": 147500
},
{
"epoch": 10.8,
"learning_rate": 2.299664282586484e-05,
"loss": 2.048,
"step": 148000
},
{
"epoch": 10.84,
"learning_rate": 2.290541526784411e-05,
"loss": 2.048,
"step": 148500
},
{
"epoch": 10.87,
"learning_rate": 2.2814187709823385e-05,
"loss": 2.048,
"step": 149000
},
{
"epoch": 10.91,
"learning_rate": 2.272296015180266e-05,
"loss": 2.048,
"step": 149500
},
{
"epoch": 10.95,
"learning_rate": 2.263173259378193e-05,
"loss": 2.048,
"step": 150000
},
{
"epoch": 10.98,
"learning_rate": 2.25405050357612e-05,
"loss": 2.048,
"step": 150500
},
{
"epoch": 11.02,
"learning_rate": 2.2449277477740476e-05,
"loss": 2.048,
"step": 151000
},
{
"epoch": 11.06,
"learning_rate": 2.235804991971975e-05,
"loss": 2.048,
"step": 151500
},
{
"epoch": 11.09,
"learning_rate": 2.226682236169902e-05,
"loss": 2.048,
"step": 152000
},
{
"epoch": 11.13,
"learning_rate": 2.2175594803678296e-05,
"loss": 2.048,
"step": 152500
},
{
"epoch": 11.17,
"learning_rate": 2.208436724565757e-05,
"loss": 2.048,
"step": 153000
},
{
"epoch": 11.2,
"learning_rate": 2.1993139687636842e-05,
"loss": 2.048,
"step": 153500
},
{
"epoch": 11.24,
"learning_rate": 2.1901912129616113e-05,
"loss": 2.048,
"step": 154000
},
{
"epoch": 11.28,
"learning_rate": 2.1810684571595387e-05,
"loss": 2.048,
"step": 154500
},
{
"epoch": 11.31,
"learning_rate": 2.1719457013574662e-05,
"loss": 2.048,
"step": 155000
},
{
"epoch": 11.35,
"learning_rate": 2.1628229455553936e-05,
"loss": 2.048,
"step": 155500
},
{
"epoch": 11.39,
"learning_rate": 2.1537001897533207e-05,
"loss": 2.048,
"step": 156000
},
{
"epoch": 11.42,
"learning_rate": 2.1445774339512482e-05,
"loss": 2.048,
"step": 156500
},
{
"epoch": 11.46,
"learning_rate": 2.1354546781491756e-05,
"loss": 2.048,
"step": 157000
},
{
"epoch": 11.49,
"learning_rate": 2.1263319223471028e-05,
"loss": 2.048,
"step": 157500
},
{
"epoch": 11.53,
"learning_rate": 2.11720916654503e-05,
"loss": 2.048,
"step": 158000
},
{
"epoch": 11.57,
"learning_rate": 2.1080864107429573e-05,
"loss": 2.048,
"step": 158500
},
{
"epoch": 11.6,
"learning_rate": 2.0989636549408848e-05,
"loss": 2.048,
"step": 159000
},
{
"epoch": 11.64,
"learning_rate": 2.089840899138812e-05,
"loss": 2.048,
"step": 159500
},
{
"epoch": 11.68,
"learning_rate": 2.0807181433367393e-05,
"loss": 2.048,
"step": 160000
},
{
"epoch": 11.71,
"learning_rate": 2.0715953875346668e-05,
"loss": 2.048,
"step": 160500
},
{
"epoch": 11.75,
"learning_rate": 2.062472631732594e-05,
"loss": 2.048,
"step": 161000
},
{
"epoch": 11.79,
"learning_rate": 2.053349875930521e-05,
"loss": 2.048,
"step": 161500
},
{
"epoch": 11.82,
"learning_rate": 2.0442271201284484e-05,
"loss": 2.048,
"step": 162000
},
{
"epoch": 11.86,
"learning_rate": 2.035104364326376e-05,
"loss": 2.048,
"step": 162500
},
{
"epoch": 11.9,
"learning_rate": 2.025981608524303e-05,
"loss": 2.048,
"step": 163000
},
{
"epoch": 11.93,
"learning_rate": 2.0168588527222305e-05,
"loss": 2.048,
"step": 163500
},
{
"epoch": 11.97,
"learning_rate": 2.007736096920158e-05,
"loss": 2.048,
"step": 164000
},
{
"epoch": 12.01,
"learning_rate": 1.998613341118085e-05,
"loss": 2.048,
"step": 164500
},
{
"epoch": 12.04,
"learning_rate": 1.989490585316012e-05,
"loss": 2.048,
"step": 165000
},
{
"epoch": 12.08,
"learning_rate": 1.9803678295139396e-05,
"loss": 2.048,
"step": 165500
},
{
"epoch": 12.12,
"learning_rate": 1.971245073711867e-05,
"loss": 2.048,
"step": 166000
},
{
"epoch": 12.15,
"learning_rate": 1.962122317909794e-05,
"loss": 2.048,
"step": 166500
},
{
"epoch": 12.19,
"learning_rate": 1.9529995621077216e-05,
"loss": 2.048,
"step": 167000
},
{
"epoch": 12.22,
"learning_rate": 1.943876806305649e-05,
"loss": 2.048,
"step": 167500
},
{
"epoch": 12.26,
"learning_rate": 1.9347540505035765e-05,
"loss": 2.048,
"step": 168000
},
{
"epoch": 12.3,
"learning_rate": 1.9256312947015036e-05,
"loss": 2.048,
"step": 168500
},
{
"epoch": 12.33,
"learning_rate": 1.9165085388994307e-05,
"loss": 2.048,
"step": 169000
},
{
"epoch": 12.37,
"learning_rate": 1.907385783097358e-05,
"loss": 2.048,
"step": 169500
},
{
"epoch": 12.41,
"learning_rate": 1.8982630272952853e-05,
"loss": 2.048,
"step": 170000
},
{
"epoch": 12.44,
"learning_rate": 1.8891402714932127e-05,
"loss": 2.048,
"step": 170500
},
{
"epoch": 12.48,
"learning_rate": 1.88001751569114e-05,
"loss": 2.048,
"step": 171000
},
{
"epoch": 12.52,
"learning_rate": 1.8708947598890676e-05,
"loss": 2.048,
"step": 171500
},
{
"epoch": 12.55,
"learning_rate": 1.8617720040869947e-05,
"loss": 2.048,
"step": 172000
},
{
"epoch": 12.59,
"learning_rate": 1.852649248284922e-05,
"loss": 2.048,
"step": 172500
},
{
"epoch": 12.63,
"learning_rate": 1.8435264924828493e-05,
"loss": 2.048,
"step": 173000
},
{
"epoch": 12.66,
"learning_rate": 1.8344037366807767e-05,
"loss": 2.048,
"step": 173500
},
{
"epoch": 12.7,
"learning_rate": 1.825280980878704e-05,
"loss": 2.048,
"step": 174000
},
{
"epoch": 12.74,
"learning_rate": 1.8161582250766313e-05,
"loss": 2.048,
"step": 174500
},
{
"epoch": 12.77,
"learning_rate": 1.8070354692745587e-05,
"loss": 2.048,
"step": 175000
},
{
"epoch": 12.81,
"learning_rate": 1.797912713472486e-05,
"loss": 2.048,
"step": 175500
},
{
"epoch": 12.84,
"learning_rate": 1.788789957670413e-05,
"loss": 2.048,
"step": 176000
},
{
"epoch": 12.88,
"learning_rate": 1.7796672018683404e-05,
"loss": 2.048,
"step": 176500
},
{
"epoch": 12.92,
"learning_rate": 1.770544446066268e-05,
"loss": 2.048,
"step": 177000
},
{
"epoch": 12.95,
"learning_rate": 1.761421690264195e-05,
"loss": 2.048,
"step": 177500
},
{
"epoch": 12.99,
"learning_rate": 1.7522989344621224e-05,
"loss": 2.048,
"step": 178000
},
{
"epoch": 13.03,
"learning_rate": 1.74317617866005e-05,
"loss": 2.048,
"step": 178500
},
{
"epoch": 13.06,
"learning_rate": 1.734053422857977e-05,
"loss": 2.048,
"step": 179000
},
{
"epoch": 13.1,
"learning_rate": 1.7249306670559044e-05,
"loss": 2.048,
"step": 179500
},
{
"epoch": 13.14,
"learning_rate": 1.7158079112538315e-05,
"loss": 2.048,
"step": 180000
},
{
"epoch": 13.17,
"learning_rate": 1.706685155451759e-05,
"loss": 2.048,
"step": 180500
},
{
"epoch": 13.21,
"learning_rate": 1.697562399649686e-05,
"loss": 2.048,
"step": 181000
},
{
"epoch": 13.25,
"learning_rate": 1.6884396438476136e-05,
"loss": 2.048,
"step": 181500
},
{
"epoch": 13.28,
"learning_rate": 1.679316888045541e-05,
"loss": 2.048,
"step": 182000
},
{
"epoch": 13.32,
"learning_rate": 1.6701941322434684e-05,
"loss": 2.048,
"step": 182500
},
{
"epoch": 13.36,
"learning_rate": 1.6610713764413956e-05,
"loss": 2.048,
"step": 183000
},
{
"epoch": 13.39,
"learning_rate": 1.6519486206393227e-05,
"loss": 2.048,
"step": 183500
},
{
"epoch": 13.43,
"learning_rate": 1.64282586483725e-05,
"loss": 2.048,
"step": 184000
},
{
"epoch": 13.47,
"learning_rate": 1.6337031090351772e-05,
"loss": 2.048,
"step": 184500
},
{
"epoch": 13.5,
"learning_rate": 1.6245803532331047e-05,
"loss": 2.048,
"step": 185000
},
{
"epoch": 13.54,
"learning_rate": 1.615457597431032e-05,
"loss": 2.048,
"step": 185500
},
{
"epoch": 13.57,
"learning_rate": 1.6063348416289596e-05,
"loss": 2.048,
"step": 186000
},
{
"epoch": 13.61,
"learning_rate": 1.5972120858268867e-05,
"loss": 2.048,
"step": 186500
},
{
"epoch": 13.65,
"learning_rate": 1.588089330024814e-05,
"loss": 2.048,
"step": 187000
},
{
"epoch": 13.68,
"learning_rate": 1.5789665742227412e-05,
"loss": 2.048,
"step": 187500
},
{
"epoch": 13.72,
"learning_rate": 1.5698438184206684e-05,
"loss": 2.048,
"step": 188000
},
{
"epoch": 13.76,
"learning_rate": 1.5607210626185958e-05,
"loss": 2.048,
"step": 188500
},
{
"epoch": 13.79,
"learning_rate": 1.5515983068165233e-05,
"loss": 2.048,
"step": 189000
},
{
"epoch": 13.83,
"learning_rate": 1.5424755510144507e-05,
"loss": 2.048,
"step": 189500
},
{
"epoch": 13.87,
"learning_rate": 1.5333527952123778e-05,
"loss": 2.048,
"step": 190000
},
{
"epoch": 13.9,
"learning_rate": 1.5242300394103051e-05,
"loss": 2.048,
"step": 190500
},
{
"epoch": 13.94,
"learning_rate": 1.5151072836082325e-05,
"loss": 2.048,
"step": 191000
},
{
"epoch": 13.98,
"learning_rate": 1.5059845278061598e-05,
"loss": 2.048,
"step": 191500
},
{
"epoch": 14.01,
"learning_rate": 1.496861772004087e-05,
"loss": 2.048,
"step": 192000
},
{
"epoch": 14.05,
"learning_rate": 1.4877390162020144e-05,
"loss": 2.048,
"step": 192500
},
{
"epoch": 14.09,
"learning_rate": 1.4786162603999418e-05,
"loss": 2.048,
"step": 193000
},
{
"epoch": 14.12,
"learning_rate": 1.469493504597869e-05,
"loss": 2.048,
"step": 193500
},
{
"epoch": 14.16,
"learning_rate": 1.4603707487957962e-05,
"loss": 2.048,
"step": 194000
},
{
"epoch": 14.2,
"learning_rate": 1.4512479929937237e-05,
"loss": 2.048,
"step": 194500
},
{
"epoch": 14.23,
"learning_rate": 1.4421252371916511e-05,
"loss": 2.048,
"step": 195000
},
{
"epoch": 14.27,
"learning_rate": 1.4330024813895782e-05,
"loss": 2.048,
"step": 195500
},
{
"epoch": 14.3,
"learning_rate": 1.4238797255875055e-05,
"loss": 2.048,
"step": 196000
},
{
"epoch": 14.34,
"learning_rate": 1.414756969785433e-05,
"loss": 2.048,
"step": 196500
},
{
"epoch": 14.38,
"learning_rate": 1.40563421398336e-05,
"loss": 2.048,
"step": 197000
},
{
"epoch": 14.41,
"learning_rate": 1.3965114581812874e-05,
"loss": 2.048,
"step": 197500
},
{
"epoch": 14.45,
"learning_rate": 1.3873887023792148e-05,
"loss": 2.048,
"step": 198000
},
{
"epoch": 14.49,
"learning_rate": 1.3782659465771423e-05,
"loss": 2.048,
"step": 198500
},
{
"epoch": 14.52,
"learning_rate": 1.3691431907750694e-05,
"loss": 2.048,
"step": 199000
},
{
"epoch": 14.56,
"learning_rate": 1.3600204349729966e-05,
"loss": 2.048,
"step": 199500
},
{
"epoch": 14.6,
"learning_rate": 1.3508976791709241e-05,
"loss": 2.048,
"step": 200000
},
{
"epoch": 14.63,
"learning_rate": 1.3417749233688512e-05,
"loss": 2.048,
"step": 200500
},
{
"epoch": 14.67,
"learning_rate": 1.3326521675667787e-05,
"loss": 2.048,
"step": 201000
},
{
"epoch": 14.71,
"learning_rate": 1.323529411764706e-05,
"loss": 2.048,
"step": 201500
},
{
"epoch": 14.74,
"learning_rate": 1.3144066559626334e-05,
"loss": 2.048,
"step": 202000
},
{
"epoch": 14.78,
"learning_rate": 1.3052839001605605e-05,
"loss": 2.048,
"step": 202500
},
{
"epoch": 14.82,
"learning_rate": 1.2961611443584878e-05,
"loss": 2.048,
"step": 203000
},
{
"epoch": 14.85,
"learning_rate": 1.2870383885564152e-05,
"loss": 2.048,
"step": 203500
},
{
"epoch": 14.89,
"learning_rate": 1.2779156327543427e-05,
"loss": 2.048,
"step": 204000
},
{
"epoch": 14.92,
"learning_rate": 1.2687928769522698e-05,
"loss": 2.048,
"step": 204500
},
{
"epoch": 14.96,
"learning_rate": 1.259670121150197e-05,
"loss": 2.048,
"step": 205000
},
{
"epoch": 15.0,
"learning_rate": 1.2505473653481245e-05,
"loss": 2.048,
"step": 205500
},
{
"epoch": 15.03,
"learning_rate": 1.2414246095460518e-05,
"loss": 2.048,
"step": 206000
},
{
"epoch": 15.07,
"learning_rate": 1.232301853743979e-05,
"loss": 2.048,
"step": 206500
},
{
"epoch": 15.11,
"learning_rate": 1.2231790979419064e-05,
"loss": 2.048,
"step": 207000
},
{
"epoch": 15.14,
"learning_rate": 1.2140563421398336e-05,
"loss": 2.048,
"step": 207500
},
{
"epoch": 15.18,
"learning_rate": 1.204933586337761e-05,
"loss": 2.048,
"step": 208000
},
{
"epoch": 15.22,
"learning_rate": 1.1958108305356882e-05,
"loss": 2.048,
"step": 208500
},
{
"epoch": 15.25,
"learning_rate": 1.1866880747336156e-05,
"loss": 2.048,
"step": 209000
},
{
"epoch": 15.29,
"learning_rate": 1.177565318931543e-05,
"loss": 2.048,
"step": 209500
},
{
"epoch": 15.33,
"learning_rate": 1.1684425631294702e-05,
"loss": 2.048,
"step": 210000
},
{
"epoch": 15.36,
"learning_rate": 1.1593198073273975e-05,
"loss": 2.048,
"step": 210500
},
{
"epoch": 15.4,
"learning_rate": 1.1501970515253248e-05,
"loss": 2.048,
"step": 211000
},
{
"epoch": 15.44,
"learning_rate": 1.1410742957232522e-05,
"loss": 2.048,
"step": 211500
},
{
"epoch": 15.47,
"learning_rate": 1.1319515399211795e-05,
"loss": 2.048,
"step": 212000
},
{
"epoch": 15.51,
"learning_rate": 1.1228287841191068e-05,
"loss": 2.048,
"step": 212500
},
{
"epoch": 15.55,
"learning_rate": 1.113706028317034e-05,
"loss": 2.048,
"step": 213000
},
{
"epoch": 15.58,
"learning_rate": 1.1045832725149615e-05,
"loss": 2.048,
"step": 213500
},
{
"epoch": 15.62,
"learning_rate": 1.0954605167128886e-05,
"loss": 2.048,
"step": 214000
},
{
"epoch": 15.65,
"learning_rate": 1.0863377609108159e-05,
"loss": 2.048,
"step": 214500
},
{
"epoch": 15.69,
"learning_rate": 1.0772150051087433e-05,
"loss": 2.048,
"step": 215000
},
{
"epoch": 15.73,
"learning_rate": 1.0680922493066706e-05,
"loss": 2.048,
"step": 215500
},
{
"epoch": 15.76,
"learning_rate": 1.0589694935045979e-05,
"loss": 2.048,
"step": 216000
},
{
"epoch": 15.8,
"learning_rate": 1.0498467377025252e-05,
"loss": 2.048,
"step": 216500
},
{
"epoch": 15.84,
"learning_rate": 1.0407239819004526e-05,
"loss": 2.048,
"step": 217000
},
{
"epoch": 15.87,
"learning_rate": 1.0316012260983799e-05,
"loss": 2.048,
"step": 217500
},
{
"epoch": 15.91,
"learning_rate": 1.0224784702963072e-05,
"loss": 2.048,
"step": 218000
},
{
"epoch": 15.95,
"learning_rate": 1.0133557144942345e-05,
"loss": 2.048,
"step": 218500
},
{
"epoch": 15.98,
"learning_rate": 1.0042329586921618e-05,
"loss": 2.048,
"step": 219000
},
{
"epoch": 16.02,
"learning_rate": 9.95110202890089e-06,
"loss": 2.048,
"step": 219500
},
{
"epoch": 16.06,
"learning_rate": 9.859874470880163e-06,
"loss": 2.048,
"step": 220000
},
{
"epoch": 16.09,
"learning_rate": 9.768646912859438e-06,
"loss": 2.048,
"step": 220500
},
{
"epoch": 16.13,
"learning_rate": 9.67741935483871e-06,
"loss": 2.048,
"step": 221000
},
{
"epoch": 16.17,
"learning_rate": 9.586191796817983e-06,
"loss": 2.048,
"step": 221500
},
{
"epoch": 16.2,
"learning_rate": 9.494964238797256e-06,
"loss": 2.048,
"step": 222000
},
{
"epoch": 16.24,
"learning_rate": 9.40373668077653e-06,
"loss": 2.048,
"step": 222500
},
{
"epoch": 16.27,
"learning_rate": 9.312509122755803e-06,
"loss": 2.048,
"step": 223000
},
{
"epoch": 16.31,
"learning_rate": 9.221281564735074e-06,
"loss": 2.048,
"step": 223500
},
{
"epoch": 16.35,
"learning_rate": 9.130054006714349e-06,
"loss": 2.048,
"step": 224000
},
{
"epoch": 16.38,
"learning_rate": 9.038826448693622e-06,
"loss": 2.048,
"step": 224500
},
{
"epoch": 16.42,
"learning_rate": 8.947598890672895e-06,
"loss": 2.048,
"step": 225000
},
{
"epoch": 16.46,
"learning_rate": 8.856371332652167e-06,
"loss": 2.048,
"step": 225500
},
{
"epoch": 16.49,
"learning_rate": 8.765143774631442e-06,
"loss": 2.048,
"step": 226000
},
{
"epoch": 16.53,
"learning_rate": 8.673916216610715e-06,
"loss": 2.048,
"step": 226500
},
{
"epoch": 16.57,
"learning_rate": 8.582688658589987e-06,
"loss": 2.048,
"step": 227000
},
{
"epoch": 16.6,
"learning_rate": 8.49146110056926e-06,
"loss": 2.048,
"step": 227500
},
{
"epoch": 16.64,
"learning_rate": 8.400233542548533e-06,
"loss": 2.048,
"step": 228000
},
{
"epoch": 16.68,
"learning_rate": 8.309005984527807e-06,
"loss": 2.048,
"step": 228500
},
{
"epoch": 16.71,
"learning_rate": 8.217778426507079e-06,
"loss": 2.048,
"step": 229000
},
{
"epoch": 16.75,
"learning_rate": 8.126550868486353e-06,
"loss": 2.048,
"step": 229500
},
{
"epoch": 16.79,
"learning_rate": 8.035323310465626e-06,
"loss": 2.048,
"step": 230000
},
{
"epoch": 16.82,
"learning_rate": 7.944095752444899e-06,
"loss": 2.048,
"step": 230500
},
{
"epoch": 16.86,
"learning_rate": 7.852868194424171e-06,
"loss": 2.048,
"step": 231000
},
{
"epoch": 16.9,
"learning_rate": 7.761640636403446e-06,
"loss": 2.048,
"step": 231500
},
{
"epoch": 16.93,
"learning_rate": 7.670413078382719e-06,
"loss": 2.048,
"step": 232000
},
{
"epoch": 16.97,
"learning_rate": 7.579185520361991e-06,
"loss": 2.048,
"step": 232500
},
{
"epoch": 17.0,
"learning_rate": 7.487957962341264e-06,
"loss": 2.048,
"step": 233000
},
{
"epoch": 17.04,
"learning_rate": 7.396730404320537e-06,
"loss": 2.048,
"step": 233500
},
{
"epoch": 17.08,
"learning_rate": 7.305502846299811e-06,
"loss": 2.048,
"step": 234000
},
{
"epoch": 17.11,
"learning_rate": 7.214275288279084e-06,
"loss": 2.048,
"step": 234500
},
{
"epoch": 17.15,
"learning_rate": 7.123047730258357e-06,
"loss": 2.048,
"step": 235000
},
{
"epoch": 17.19,
"learning_rate": 7.03182017223763e-06,
"loss": 2.048,
"step": 235500
},
{
"epoch": 17.22,
"learning_rate": 6.940592614216904e-06,
"loss": 2.048,
"step": 236000
},
{
"epoch": 17.26,
"learning_rate": 6.8493650561961765e-06,
"loss": 2.048,
"step": 236500
},
{
"epoch": 17.3,
"learning_rate": 6.7581374981754485e-06,
"loss": 2.048,
"step": 237000
},
{
"epoch": 17.33,
"learning_rate": 6.666909940154722e-06,
"loss": 2.048,
"step": 237500
},
{
"epoch": 17.37,
"learning_rate": 6.575682382133995e-06,
"loss": 2.048,
"step": 238000
},
{
"epoch": 17.41,
"learning_rate": 6.484454824113269e-06,
"loss": 2.048,
"step": 238500
},
{
"epoch": 17.44,
"learning_rate": 6.393227266092541e-06,
"loss": 2.048,
"step": 239000
},
{
"epoch": 17.48,
"learning_rate": 6.301999708071815e-06,
"loss": 2.048,
"step": 239500
},
{
"epoch": 17.52,
"learning_rate": 6.210772150051088e-06,
"loss": 2.048,
"step": 240000
},
{
"epoch": 17.55,
"learning_rate": 6.119544592030361e-06,
"loss": 2.048,
"step": 240500
},
{
"epoch": 17.59,
"learning_rate": 6.028317034009634e-06,
"loss": 2.048,
"step": 241000
},
{
"epoch": 17.63,
"learning_rate": 5.937089475988907e-06,
"loss": 2.048,
"step": 241500
},
{
"epoch": 17.66,
"learning_rate": 5.845861917968181e-06,
"loss": 2.048,
"step": 242000
},
{
"epoch": 17.7,
"learning_rate": 5.7546343599474535e-06,
"loss": 2.048,
"step": 242500
},
{
"epoch": 17.73,
"learning_rate": 5.663406801926726e-06,
"loss": 2.048,
"step": 243000
},
{
"epoch": 17.77,
"learning_rate": 5.572179243905999e-06,
"loss": 2.048,
"step": 243500
},
{
"epoch": 17.81,
"learning_rate": 5.480951685885272e-06,
"loss": 2.048,
"step": 244000
},
{
"epoch": 17.84,
"learning_rate": 5.3897241278645456e-06,
"loss": 2.048,
"step": 244500
},
{
"epoch": 17.88,
"learning_rate": 5.298496569843818e-06,
"loss": 2.048,
"step": 245000
},
{
"epoch": 17.92,
"learning_rate": 5.207269011823092e-06,
"loss": 2.048,
"step": 245500
},
{
"epoch": 17.95,
"learning_rate": 5.116041453802365e-06,
"loss": 2.048,
"step": 246000
},
{
"epoch": 17.99,
"learning_rate": 5.0248138957816385e-06,
"loss": 2.048,
"step": 246500
},
{
"epoch": 18.03,
"learning_rate": 4.933586337760911e-06,
"loss": 2.048,
"step": 247000
},
{
"epoch": 18.06,
"learning_rate": 4.842358779740184e-06,
"loss": 2.048,
"step": 247500
},
{
"epoch": 18.1,
"learning_rate": 4.751131221719457e-06,
"loss": 2.048,
"step": 248000
},
{
"epoch": 18.14,
"learning_rate": 4.6599036636987305e-06,
"loss": 2.048,
"step": 248500
},
{
"epoch": 18.17,
"learning_rate": 4.568676105678003e-06,
"loss": 2.048,
"step": 249000
},
{
"epoch": 18.21,
"learning_rate": 4.477448547657277e-06,
"loss": 2.048,
"step": 249500
},
{
"epoch": 18.25,
"learning_rate": 4.38622098963655e-06,
"loss": 2.048,
"step": 250000
},
{
"epoch": 18.28,
"learning_rate": 4.2949934316158226e-06,
"loss": 2.048,
"step": 250500
},
{
"epoch": 18.32,
"learning_rate": 4.203765873595096e-06,
"loss": 2.048,
"step": 251000
},
{
"epoch": 18.35,
"learning_rate": 4.112538315574369e-06,
"loss": 2.048,
"step": 251500
},
{
"epoch": 18.39,
"learning_rate": 4.021310757553642e-06,
"loss": 2.048,
"step": 252000
},
{
"epoch": 18.43,
"learning_rate": 3.930083199532915e-06,
"loss": 2.048,
"step": 252500
},
{
"epoch": 18.46,
"learning_rate": 3.838855641512188e-06,
"loss": 2.048,
"step": 253000
},
{
"epoch": 18.5,
"learning_rate": 3.747628083491461e-06,
"loss": 2.048,
"step": 253500
},
{
"epoch": 18.54,
"learning_rate": 3.6564005254707343e-06,
"loss": 2.048,
"step": 254000
},
{
"epoch": 18.57,
"learning_rate": 3.5651729674500075e-06,
"loss": 2.048,
"step": 254500
},
{
"epoch": 18.61,
"learning_rate": 3.4739454094292807e-06,
"loss": 2.048,
"step": 255000
},
{
"epoch": 18.65,
"learning_rate": 3.382717851408554e-06,
"loss": 2.048,
"step": 255500
},
{
"epoch": 18.68,
"learning_rate": 3.291490293387827e-06,
"loss": 2.048,
"step": 256000
},
{
"epoch": 18.72,
"learning_rate": 3.2002627353670995e-06,
"loss": 2.048,
"step": 256500
},
{
"epoch": 18.76,
"learning_rate": 3.109035177346373e-06,
"loss": 2.048,
"step": 257000
},
{
"epoch": 18.79,
"learning_rate": 3.017807619325646e-06,
"loss": 2.048,
"step": 257500
},
{
"epoch": 18.83,
"learning_rate": 2.926580061304919e-06,
"loss": 2.048,
"step": 258000
},
{
"epoch": 18.87,
"learning_rate": 2.8353525032841924e-06,
"loss": 2.048,
"step": 258500
},
{
"epoch": 18.9,
"learning_rate": 2.7441249452634652e-06,
"loss": 2.048,
"step": 259000
},
{
"epoch": 18.94,
"learning_rate": 2.652897387242738e-06,
"loss": 2.048,
"step": 259500
},
{
"epoch": 18.98,
"learning_rate": 2.5616698292220113e-06,
"loss": 2.048,
"step": 260000
},
{
"epoch": 19.01,
"learning_rate": 2.4704422712012845e-06,
"loss": 2.048,
"step": 260500
},
{
"epoch": 19.05,
"learning_rate": 2.3792147131805577e-06,
"loss": 2.048,
"step": 261000
},
{
"epoch": 19.08,
"learning_rate": 2.287987155159831e-06,
"loss": 2.048,
"step": 261500
},
{
"epoch": 19.12,
"learning_rate": 2.1967595971391037e-06,
"loss": 2.048,
"step": 262000
},
{
"epoch": 19.16,
"learning_rate": 2.105532039118377e-06,
"loss": 2.048,
"step": 262500
},
{
"epoch": 19.19,
"learning_rate": 2.01430448109765e-06,
"loss": 2.048,
"step": 263000
},
{
"epoch": 19.23,
"learning_rate": 1.9230769230769234e-06,
"loss": 2.048,
"step": 263500
},
{
"epoch": 19.27,
"learning_rate": 1.8318493650561962e-06,
"loss": 2.048,
"step": 264000
},
{
"epoch": 19.3,
"learning_rate": 1.7406218070354692e-06,
"loss": 2.048,
"step": 264500
},
{
"epoch": 19.34,
"learning_rate": 1.6493942490147424e-06,
"loss": 2.048,
"step": 265000
},
{
"epoch": 19.38,
"learning_rate": 1.5581666909940157e-06,
"loss": 2.048,
"step": 265500
},
{
"epoch": 19.41,
"learning_rate": 1.4669391329732887e-06,
"loss": 2.048,
"step": 266000
},
{
"epoch": 19.45,
"learning_rate": 1.3757115749525619e-06,
"loss": 2.048,
"step": 266500
},
{
"epoch": 19.49,
"learning_rate": 1.2844840169318347e-06,
"loss": 2.048,
"step": 267000
},
{
"epoch": 19.52,
"learning_rate": 1.193256458911108e-06,
"loss": 2.048,
"step": 267500
},
{
"epoch": 19.56,
"learning_rate": 1.102028900890381e-06,
"loss": 2.048,
"step": 268000
},
{
"epoch": 19.6,
"learning_rate": 1.0108013428696541e-06,
"loss": 2.048,
"step": 268500
},
{
"epoch": 19.63,
"learning_rate": 9.195737848489272e-07,
"loss": 2.048,
"step": 269000
},
{
"epoch": 19.67,
"learning_rate": 8.283462268282003e-07,
"loss": 2.048,
"step": 269500
},
{
"epoch": 19.71,
"learning_rate": 7.371186688074734e-07,
"loss": 2.048,
"step": 270000
},
{
"epoch": 19.74,
"learning_rate": 6.458911107867465e-07,
"loss": 2.048,
"step": 270500
},
{
"epoch": 19.78,
"learning_rate": 5.546635527660195e-07,
"loss": 2.048,
"step": 271000
},
{
"epoch": 19.81,
"learning_rate": 4.6343599474529264e-07,
"loss": 2.048,
"step": 271500
},
{
"epoch": 19.85,
"learning_rate": 3.7220843672456576e-07,
"loss": 2.048,
"step": 272000
},
{
"epoch": 19.89,
"learning_rate": 2.809808787038389e-07,
"loss": 2.048,
"step": 272500
},
{
"epoch": 19.92,
"learning_rate": 1.8975332068311197e-07,
"loss": 2.048,
"step": 273000
},
{
"epoch": 19.96,
"learning_rate": 9.852576266238505e-08,
"loss": 2.048,
"step": 273500
},
{
"epoch": 20.0,
"learning_rate": 7.298204641658152e-09,
"loss": 2.048,
"step": 274000
},
{
"epoch": 20.0,
"step": 274040,
"total_flos": 2.7362600759486054e+18,
"train_runtime": 264850.0877,
"train_samples_per_second": 1.035
}
],
"max_steps": 274040,
"num_train_epochs": 20,
"total_flos": 2.7362600759486054e+18,
"trial_name": null,
"trial_params": null
}