mk-roberta-base / trainer_state.json
Stojanco Tudzarski
back to original push
f8f498b
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 9.177373602536292,
"global_step": 110000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04,
"learning_rate": 4.9791423327215084e-05,
"loss": 8.0732,
"step": 500
},
{
"epoch": 0.08,
"learning_rate": 4.958284665443017e-05,
"loss": 7.3801,
"step": 1000
},
{
"epoch": 0.13,
"learning_rate": 4.9374269981645254e-05,
"loss": 7.0667,
"step": 1500
},
{
"epoch": 0.17,
"learning_rate": 4.9165693308860336e-05,
"loss": 6.906,
"step": 2000
},
{
"epoch": 0.21,
"learning_rate": 4.8957116636075424e-05,
"loss": 6.7114,
"step": 2500
},
{
"epoch": 0.25,
"learning_rate": 4.8748539963290505e-05,
"loss": 6.6333,
"step": 3000
},
{
"epoch": 0.29,
"learning_rate": 4.8539963290505594e-05,
"loss": 6.4948,
"step": 3500
},
{
"epoch": 0.33,
"learning_rate": 4.8331386617720675e-05,
"loss": 6.3157,
"step": 4000
},
{
"epoch": 0.38,
"learning_rate": 4.812280994493576e-05,
"loss": 6.1185,
"step": 4500
},
{
"epoch": 0.42,
"learning_rate": 4.7914233272150845e-05,
"loss": 5.9335,
"step": 5000
},
{
"epoch": 0.46,
"learning_rate": 4.770565659936593e-05,
"loss": 5.7535,
"step": 5500
},
{
"epoch": 0.5,
"learning_rate": 4.749707992658101e-05,
"loss": 5.5631,
"step": 6000
},
{
"epoch": 0.54,
"learning_rate": 4.72885032537961e-05,
"loss": 5.3707,
"step": 6500
},
{
"epoch": 0.58,
"learning_rate": 4.707992658101118e-05,
"loss": 5.2234,
"step": 7000
},
{
"epoch": 0.63,
"learning_rate": 4.687134990822627e-05,
"loss": 5.1324,
"step": 7500
},
{
"epoch": 0.67,
"learning_rate": 4.666277323544135e-05,
"loss": 5.023,
"step": 8000
},
{
"epoch": 0.71,
"learning_rate": 4.645419656265643e-05,
"loss": 4.9164,
"step": 8500
},
{
"epoch": 0.75,
"learning_rate": 4.624561988987152e-05,
"loss": 4.8293,
"step": 9000
},
{
"epoch": 0.79,
"learning_rate": 4.60370432170866e-05,
"loss": 4.7737,
"step": 9500
},
{
"epoch": 0.83,
"learning_rate": 4.582846654430168e-05,
"loss": 4.6915,
"step": 10000
},
{
"epoch": 0.88,
"learning_rate": 4.561988987151678e-05,
"loss": 4.6345,
"step": 10500
},
{
"epoch": 0.92,
"learning_rate": 4.541131319873186e-05,
"loss": 4.6052,
"step": 11000
},
{
"epoch": 0.96,
"learning_rate": 4.520273652594694e-05,
"loss": 4.5157,
"step": 11500
},
{
"epoch": 1.0,
"learning_rate": 4.499415985316203e-05,
"loss": 4.4535,
"step": 12000
},
{
"epoch": 1.04,
"learning_rate": 4.478558318037711e-05,
"loss": 4.4238,
"step": 12500
},
{
"epoch": 1.08,
"learning_rate": 4.45770065075922e-05,
"loss": 4.3574,
"step": 13000
},
{
"epoch": 1.13,
"learning_rate": 4.436842983480728e-05,
"loss": 4.3355,
"step": 13500
},
{
"epoch": 1.17,
"learning_rate": 4.415985316202236e-05,
"loss": 4.3007,
"step": 14000
},
{
"epoch": 1.21,
"learning_rate": 4.395127648923745e-05,
"loss": 4.2482,
"step": 14500
},
{
"epoch": 1.25,
"learning_rate": 4.374269981645253e-05,
"loss": 4.1904,
"step": 15000
},
{
"epoch": 1.29,
"learning_rate": 4.353412314366761e-05,
"loss": 4.1877,
"step": 15500
},
{
"epoch": 1.33,
"learning_rate": 4.33255464708827e-05,
"loss": 4.1422,
"step": 16000
},
{
"epoch": 1.38,
"learning_rate": 4.311696979809778e-05,
"loss": 4.0952,
"step": 16500
},
{
"epoch": 1.42,
"learning_rate": 4.2908393125312865e-05,
"loss": 4.0837,
"step": 17000
},
{
"epoch": 1.46,
"learning_rate": 4.269981645252795e-05,
"loss": 4.0552,
"step": 17500
},
{
"epoch": 1.5,
"learning_rate": 4.2491239779743035e-05,
"loss": 4.0141,
"step": 18000
},
{
"epoch": 1.54,
"learning_rate": 4.228266310695812e-05,
"loss": 3.9979,
"step": 18500
},
{
"epoch": 1.59,
"learning_rate": 4.2074086434173204e-05,
"loss": 3.9504,
"step": 19000
},
{
"epoch": 1.63,
"learning_rate": 4.1865509761388286e-05,
"loss": 3.9522,
"step": 19500
},
{
"epoch": 1.67,
"learning_rate": 4.1656933088603374e-05,
"loss": 3.9057,
"step": 20000
},
{
"epoch": 1.71,
"learning_rate": 4.1448356415818456e-05,
"loss": 3.8997,
"step": 20500
},
{
"epoch": 1.75,
"learning_rate": 4.123977974303354e-05,
"loss": 3.8594,
"step": 21000
},
{
"epoch": 1.79,
"learning_rate": 4.1031203070248626e-05,
"loss": 3.823,
"step": 21500
},
{
"epoch": 1.84,
"learning_rate": 4.082262639746371e-05,
"loss": 3.8168,
"step": 22000
},
{
"epoch": 1.88,
"learning_rate": 4.0614049724678796e-05,
"loss": 3.7902,
"step": 22500
},
{
"epoch": 1.92,
"learning_rate": 4.040547305189388e-05,
"loss": 3.7825,
"step": 23000
},
{
"epoch": 1.96,
"learning_rate": 4.019689637910896e-05,
"loss": 3.7529,
"step": 23500
},
{
"epoch": 2.0,
"learning_rate": 3.998831970632405e-05,
"loss": 3.7309,
"step": 24000
},
{
"epoch": 2.04,
"learning_rate": 3.977974303353913e-05,
"loss": 3.6979,
"step": 24500
},
{
"epoch": 2.09,
"learning_rate": 3.957116636075421e-05,
"loss": 3.6644,
"step": 25000
},
{
"epoch": 2.13,
"learning_rate": 3.93625896879693e-05,
"loss": 3.6569,
"step": 25500
},
{
"epoch": 2.17,
"learning_rate": 3.915401301518438e-05,
"loss": 3.6378,
"step": 26000
},
{
"epoch": 2.21,
"learning_rate": 3.894543634239947e-05,
"loss": 3.6143,
"step": 26500
},
{
"epoch": 2.25,
"learning_rate": 3.873685966961455e-05,
"loss": 3.5909,
"step": 27000
},
{
"epoch": 2.29,
"learning_rate": 3.852828299682963e-05,
"loss": 3.6046,
"step": 27500
},
{
"epoch": 2.34,
"learning_rate": 3.831970632404472e-05,
"loss": 3.577,
"step": 28000
},
{
"epoch": 2.38,
"learning_rate": 3.81111296512598e-05,
"loss": 3.5381,
"step": 28500
},
{
"epoch": 2.42,
"learning_rate": 3.7902552978474884e-05,
"loss": 3.5436,
"step": 29000
},
{
"epoch": 2.46,
"learning_rate": 3.769397630568997e-05,
"loss": 3.5292,
"step": 29500
},
{
"epoch": 2.5,
"learning_rate": 3.7485399632905054e-05,
"loss": 3.5224,
"step": 30000
},
{
"epoch": 2.54,
"learning_rate": 3.727682296012014e-05,
"loss": 3.5099,
"step": 30500
},
{
"epoch": 2.59,
"learning_rate": 3.706824628733523e-05,
"loss": 3.4796,
"step": 31000
},
{
"epoch": 2.63,
"learning_rate": 3.685966961455031e-05,
"loss": 3.475,
"step": 31500
},
{
"epoch": 2.67,
"learning_rate": 3.66510929417654e-05,
"loss": 3.4557,
"step": 32000
},
{
"epoch": 2.71,
"learning_rate": 3.644251626898048e-05,
"loss": 3.4315,
"step": 32500
},
{
"epoch": 2.75,
"learning_rate": 3.6233939596195564e-05,
"loss": 3.4356,
"step": 33000
},
{
"epoch": 2.79,
"learning_rate": 3.602536292341065e-05,
"loss": 3.4244,
"step": 33500
},
{
"epoch": 2.84,
"learning_rate": 3.5816786250625733e-05,
"loss": 3.4026,
"step": 34000
},
{
"epoch": 2.88,
"learning_rate": 3.5608209577840815e-05,
"loss": 3.42,
"step": 34500
},
{
"epoch": 2.92,
"learning_rate": 3.5399632905055903e-05,
"loss": 3.3832,
"step": 35000
},
{
"epoch": 2.96,
"learning_rate": 3.5191056232270985e-05,
"loss": 3.3545,
"step": 35500
},
{
"epoch": 3.0,
"learning_rate": 3.4982479559486067e-05,
"loss": 3.3327,
"step": 36000
},
{
"epoch": 3.05,
"learning_rate": 3.4773902886701155e-05,
"loss": 3.3129,
"step": 36500
},
{
"epoch": 3.09,
"learning_rate": 3.4565326213916237e-05,
"loss": 3.3115,
"step": 37000
},
{
"epoch": 3.13,
"learning_rate": 3.4356749541131325e-05,
"loss": 3.2831,
"step": 37500
},
{
"epoch": 3.17,
"learning_rate": 3.4148172868346407e-05,
"loss": 3.2743,
"step": 38000
},
{
"epoch": 3.21,
"learning_rate": 3.393959619556149e-05,
"loss": 3.2856,
"step": 38500
},
{
"epoch": 3.25,
"learning_rate": 3.3731019522776576e-05,
"loss": 3.267,
"step": 39000
},
{
"epoch": 3.3,
"learning_rate": 3.352244284999166e-05,
"loss": 3.264,
"step": 39500
},
{
"epoch": 3.34,
"learning_rate": 3.331386617720674e-05,
"loss": 3.2614,
"step": 40000
},
{
"epoch": 3.38,
"learning_rate": 3.310528950442183e-05,
"loss": 3.2554,
"step": 40500
},
{
"epoch": 3.42,
"learning_rate": 3.289671283163691e-05,
"loss": 3.2308,
"step": 41000
},
{
"epoch": 3.46,
"learning_rate": 3.2688136158852e-05,
"loss": 3.2329,
"step": 41500
},
{
"epoch": 3.5,
"learning_rate": 3.247955948606708e-05,
"loss": 3.2174,
"step": 42000
},
{
"epoch": 3.55,
"learning_rate": 3.227098281328216e-05,
"loss": 3.1932,
"step": 42500
},
{
"epoch": 3.59,
"learning_rate": 3.206240614049725e-05,
"loss": 3.1748,
"step": 43000
},
{
"epoch": 3.63,
"learning_rate": 3.185382946771233e-05,
"loss": 3.1664,
"step": 43500
},
{
"epoch": 3.67,
"learning_rate": 3.164525279492741e-05,
"loss": 3.1862,
"step": 44000
},
{
"epoch": 3.71,
"learning_rate": 3.14366761221425e-05,
"loss": 3.1649,
"step": 44500
},
{
"epoch": 3.75,
"learning_rate": 3.122809944935758e-05,
"loss": 3.1528,
"step": 45000
},
{
"epoch": 3.8,
"learning_rate": 3.101952277657267e-05,
"loss": 3.1512,
"step": 45500
},
{
"epoch": 3.84,
"learning_rate": 3.081094610378775e-05,
"loss": 3.1362,
"step": 46000
},
{
"epoch": 3.88,
"learning_rate": 3.0602369431002834e-05,
"loss": 3.1517,
"step": 46500
},
{
"epoch": 3.92,
"learning_rate": 3.039379275821792e-05,
"loss": 3.1104,
"step": 47000
},
{
"epoch": 3.96,
"learning_rate": 3.0185216085433004e-05,
"loss": 3.1216,
"step": 47500
},
{
"epoch": 4.0,
"learning_rate": 2.997663941264809e-05,
"loss": 3.1177,
"step": 48000
},
{
"epoch": 4.05,
"learning_rate": 2.9768062739863174e-05,
"loss": 3.0815,
"step": 48500
},
{
"epoch": 4.09,
"learning_rate": 2.9559486067078256e-05,
"loss": 3.0839,
"step": 49000
},
{
"epoch": 4.13,
"learning_rate": 2.935090939429334e-05,
"loss": 3.0499,
"step": 49500
},
{
"epoch": 4.17,
"learning_rate": 2.9142332721508426e-05,
"loss": 3.0597,
"step": 50000
},
{
"epoch": 4.21,
"learning_rate": 2.8933756048723514e-05,
"loss": 3.0588,
"step": 50500
},
{
"epoch": 4.25,
"learning_rate": 2.87251793759386e-05,
"loss": 3.0393,
"step": 51000
},
{
"epoch": 4.3,
"learning_rate": 2.8516602703153684e-05,
"loss": 3.041,
"step": 51500
},
{
"epoch": 4.34,
"learning_rate": 2.8308026030368766e-05,
"loss": 3.0235,
"step": 52000
},
{
"epoch": 4.38,
"learning_rate": 2.809944935758385e-05,
"loss": 3.0257,
"step": 52500
},
{
"epoch": 4.42,
"learning_rate": 2.7890872684798936e-05,
"loss": 3.0206,
"step": 53000
},
{
"epoch": 4.46,
"learning_rate": 2.768229601201402e-05,
"loss": 3.0039,
"step": 53500
},
{
"epoch": 4.51,
"learning_rate": 2.7473719339229102e-05,
"loss": 2.9904,
"step": 54000
},
{
"epoch": 4.55,
"learning_rate": 2.7265142666444187e-05,
"loss": 2.9984,
"step": 54500
},
{
"epoch": 4.59,
"learning_rate": 2.7056565993659272e-05,
"loss": 2.9806,
"step": 55000
},
{
"epoch": 4.63,
"learning_rate": 2.6847989320874357e-05,
"loss": 2.9864,
"step": 55500
},
{
"epoch": 4.67,
"learning_rate": 2.663941264808944e-05,
"loss": 2.9916,
"step": 56000
},
{
"epoch": 4.71,
"learning_rate": 2.6430835975304524e-05,
"loss": 2.9799,
"step": 56500
},
{
"epoch": 4.76,
"learning_rate": 2.622225930251961e-05,
"loss": 2.9744,
"step": 57000
},
{
"epoch": 4.8,
"learning_rate": 2.6013682629734694e-05,
"loss": 2.9689,
"step": 57500
},
{
"epoch": 4.84,
"learning_rate": 2.5805105956949775e-05,
"loss": 2.9755,
"step": 58000
},
{
"epoch": 4.88,
"learning_rate": 2.559652928416486e-05,
"loss": 2.9355,
"step": 58500
},
{
"epoch": 4.92,
"learning_rate": 2.5387952611379945e-05,
"loss": 2.9549,
"step": 59000
},
{
"epoch": 4.96,
"learning_rate": 2.517937593859503e-05,
"loss": 2.9552,
"step": 59500
},
{
"epoch": 5.01,
"learning_rate": 2.497079926581011e-05,
"loss": 2.9608,
"step": 60000
},
{
"epoch": 5.05,
"learning_rate": 2.4762222593025197e-05,
"loss": 2.9073,
"step": 60500
},
{
"epoch": 5.09,
"learning_rate": 2.455364592024028e-05,
"loss": 2.9192,
"step": 61000
},
{
"epoch": 5.13,
"learning_rate": 2.4345069247455367e-05,
"loss": 2.922,
"step": 61500
},
{
"epoch": 5.17,
"learning_rate": 2.4136492574670448e-05,
"loss": 2.8877,
"step": 62000
},
{
"epoch": 5.21,
"learning_rate": 2.3927915901885533e-05,
"loss": 2.9045,
"step": 62500
},
{
"epoch": 5.26,
"learning_rate": 2.3719339229100618e-05,
"loss": 2.8876,
"step": 63000
},
{
"epoch": 5.3,
"learning_rate": 2.3510762556315703e-05,
"loss": 2.8986,
"step": 63500
},
{
"epoch": 5.34,
"learning_rate": 2.3302185883530785e-05,
"loss": 2.8924,
"step": 64000
},
{
"epoch": 5.38,
"learning_rate": 2.309360921074587e-05,
"loss": 2.8761,
"step": 64500
},
{
"epoch": 5.42,
"learning_rate": 2.2885032537960955e-05,
"loss": 2.8782,
"step": 65000
},
{
"epoch": 5.46,
"learning_rate": 2.267645586517604e-05,
"loss": 2.878,
"step": 65500
},
{
"epoch": 5.51,
"learning_rate": 2.2467879192391125e-05,
"loss": 2.8706,
"step": 66000
},
{
"epoch": 5.55,
"learning_rate": 2.225930251960621e-05,
"loss": 2.8647,
"step": 66500
},
{
"epoch": 5.59,
"learning_rate": 2.2050725846821295e-05,
"loss": 2.8407,
"step": 67000
},
{
"epoch": 5.63,
"learning_rate": 2.1842149174036376e-05,
"loss": 2.86,
"step": 67500
},
{
"epoch": 5.67,
"learning_rate": 2.163357250125146e-05,
"loss": 2.8585,
"step": 68000
},
{
"epoch": 5.72,
"learning_rate": 2.1424995828466546e-05,
"loss": 2.8472,
"step": 68500
},
{
"epoch": 5.76,
"learning_rate": 2.121641915568163e-05,
"loss": 2.8495,
"step": 69000
},
{
"epoch": 5.8,
"learning_rate": 2.1007842482896713e-05,
"loss": 2.8375,
"step": 69500
},
{
"epoch": 5.84,
"learning_rate": 2.0799265810111798e-05,
"loss": 2.8456,
"step": 70000
},
{
"epoch": 5.88,
"learning_rate": 2.0590689137326883e-05,
"loss": 2.8388,
"step": 70500
},
{
"epoch": 5.92,
"learning_rate": 2.0382112464541968e-05,
"loss": 2.8346,
"step": 71000
},
{
"epoch": 5.97,
"learning_rate": 2.017353579175705e-05,
"loss": 2.8259,
"step": 71500
},
{
"epoch": 6.01,
"learning_rate": 1.9964959118972134e-05,
"loss": 2.8211,
"step": 72000
},
{
"epoch": 6.05,
"learning_rate": 1.975638244618722e-05,
"loss": 2.7936,
"step": 72500
},
{
"epoch": 6.09,
"learning_rate": 1.9547805773402304e-05,
"loss": 2.7904,
"step": 73000
},
{
"epoch": 6.13,
"learning_rate": 1.9339229100617386e-05,
"loss": 2.7982,
"step": 73500
},
{
"epoch": 6.17,
"learning_rate": 1.913065242783247e-05,
"loss": 2.7948,
"step": 74000
},
{
"epoch": 6.22,
"learning_rate": 1.8922075755047556e-05,
"loss": 2.8152,
"step": 74500
},
{
"epoch": 6.26,
"learning_rate": 1.871349908226264e-05,
"loss": 2.7862,
"step": 75000
},
{
"epoch": 6.3,
"learning_rate": 1.8504922409477726e-05,
"loss": 2.7915,
"step": 75500
},
{
"epoch": 6.34,
"learning_rate": 1.829634573669281e-05,
"loss": 2.7796,
"step": 76000
},
{
"epoch": 6.38,
"learning_rate": 1.8087769063907896e-05,
"loss": 2.7892,
"step": 76500
},
{
"epoch": 6.42,
"learning_rate": 1.7879192391122977e-05,
"loss": 2.7827,
"step": 77000
},
{
"epoch": 6.47,
"learning_rate": 1.7670615718338062e-05,
"loss": 2.7857,
"step": 77500
},
{
"epoch": 6.51,
"learning_rate": 1.7462039045553147e-05,
"loss": 2.7889,
"step": 78000
},
{
"epoch": 6.55,
"learning_rate": 1.7253462372768232e-05,
"loss": 2.748,
"step": 78500
},
{
"epoch": 6.59,
"learning_rate": 1.7044885699983314e-05,
"loss": 2.7607,
"step": 79000
},
{
"epoch": 6.63,
"learning_rate": 1.68363090271984e-05,
"loss": 2.7597,
"step": 79500
},
{
"epoch": 6.67,
"learning_rate": 1.6627732354413484e-05,
"loss": 2.7591,
"step": 80000
},
{
"epoch": 6.72,
"learning_rate": 1.641915568162857e-05,
"loss": 2.759,
"step": 80500
},
{
"epoch": 6.76,
"learning_rate": 1.621057900884365e-05,
"loss": 2.7519,
"step": 81000
},
{
"epoch": 6.8,
"learning_rate": 1.6002002336058735e-05,
"loss": 2.745,
"step": 81500
},
{
"epoch": 6.84,
"learning_rate": 1.579342566327382e-05,
"loss": 2.7569,
"step": 82000
},
{
"epoch": 6.88,
"learning_rate": 1.5584848990488905e-05,
"loss": 2.7549,
"step": 82500
},
{
"epoch": 6.92,
"learning_rate": 1.5376272317703987e-05,
"loss": 2.748,
"step": 83000
},
{
"epoch": 6.97,
"learning_rate": 1.5167695644919072e-05,
"loss": 2.7653,
"step": 83500
},
{
"epoch": 7.01,
"learning_rate": 1.4959118972134157e-05,
"loss": 2.7258,
"step": 84000
},
{
"epoch": 7.05,
"learning_rate": 1.475054229934924e-05,
"loss": 2.729,
"step": 84500
},
{
"epoch": 7.09,
"learning_rate": 1.4541965626564325e-05,
"loss": 2.7198,
"step": 85000
},
{
"epoch": 7.13,
"learning_rate": 1.4333388953779412e-05,
"loss": 2.7184,
"step": 85500
},
{
"epoch": 7.18,
"learning_rate": 1.4124812280994495e-05,
"loss": 2.7079,
"step": 86000
},
{
"epoch": 7.22,
"learning_rate": 1.391623560820958e-05,
"loss": 2.7243,
"step": 86500
},
{
"epoch": 7.26,
"learning_rate": 1.3707658935424663e-05,
"loss": 2.6942,
"step": 87000
},
{
"epoch": 7.3,
"learning_rate": 1.3499082262639748e-05,
"loss": 2.7002,
"step": 87500
},
{
"epoch": 7.34,
"learning_rate": 1.3290505589854831e-05,
"loss": 2.7085,
"step": 88000
},
{
"epoch": 7.38,
"learning_rate": 1.3081928917069916e-05,
"loss": 2.6986,
"step": 88500
},
{
"epoch": 7.43,
"learning_rate": 1.2873352244285e-05,
"loss": 2.7054,
"step": 89000
},
{
"epoch": 7.47,
"learning_rate": 1.2664775571500085e-05,
"loss": 2.7214,
"step": 89500
},
{
"epoch": 7.51,
"learning_rate": 1.2456198898715168e-05,
"loss": 2.7054,
"step": 90000
},
{
"epoch": 7.55,
"learning_rate": 1.2247622225930253e-05,
"loss": 2.7054,
"step": 90500
},
{
"epoch": 7.59,
"learning_rate": 1.2039045553145336e-05,
"loss": 2.7097,
"step": 91000
},
{
"epoch": 7.63,
"learning_rate": 1.1830468880360421e-05,
"loss": 2.6911,
"step": 91500
},
{
"epoch": 7.68,
"learning_rate": 1.1621892207575505e-05,
"loss": 2.6768,
"step": 92000
},
{
"epoch": 7.72,
"learning_rate": 1.141331553479059e-05,
"loss": 2.6735,
"step": 92500
},
{
"epoch": 7.76,
"learning_rate": 1.1204738862005673e-05,
"loss": 2.703,
"step": 93000
},
{
"epoch": 7.8,
"learning_rate": 1.099616218922076e-05,
"loss": 2.6704,
"step": 93500
},
{
"epoch": 7.84,
"learning_rate": 1.0787585516435843e-05,
"loss": 2.6833,
"step": 94000
},
{
"epoch": 7.88,
"learning_rate": 1.0579008843650926e-05,
"loss": 2.6809,
"step": 94500
},
{
"epoch": 7.93,
"learning_rate": 1.0370432170866011e-05,
"loss": 2.6766,
"step": 95000
},
{
"epoch": 7.97,
"learning_rate": 1.0161855498081094e-05,
"loss": 2.6719,
"step": 95500
},
{
"epoch": 8.01,
"learning_rate": 9.95327882529618e-06,
"loss": 2.6801,
"step": 96000
},
{
"epoch": 8.05,
"learning_rate": 9.744702152511263e-06,
"loss": 2.6555,
"step": 96500
},
{
"epoch": 8.09,
"learning_rate": 9.536125479726348e-06,
"loss": 2.656,
"step": 97000
},
{
"epoch": 8.13,
"learning_rate": 9.32754880694143e-06,
"loss": 2.6629,
"step": 97500
},
{
"epoch": 8.18,
"learning_rate": 9.118972134156516e-06,
"loss": 2.6591,
"step": 98000
},
{
"epoch": 8.22,
"learning_rate": 8.9103954613716e-06,
"loss": 2.6521,
"step": 98500
},
{
"epoch": 8.26,
"learning_rate": 8.701818788586686e-06,
"loss": 2.6594,
"step": 99000
},
{
"epoch": 8.3,
"learning_rate": 8.493242115801769e-06,
"loss": 2.65,
"step": 99500
},
{
"epoch": 8.34,
"learning_rate": 8.284665443016854e-06,
"loss": 2.6492,
"step": 100000
},
{
"epoch": 8.38,
"learning_rate": 8.076088770231937e-06,
"loss": 2.6443,
"step": 100500
},
{
"epoch": 8.43,
"learning_rate": 7.867512097447022e-06,
"loss": 2.6567,
"step": 101000
},
{
"epoch": 8.47,
"learning_rate": 7.658935424662106e-06,
"loss": 2.6213,
"step": 101500
},
{
"epoch": 8.51,
"learning_rate": 7.45035875187719e-06,
"loss": 2.6659,
"step": 102000
},
{
"epoch": 8.55,
"learning_rate": 7.241782079092274e-06,
"loss": 2.6456,
"step": 102500
},
{
"epoch": 8.59,
"learning_rate": 7.033205406307358e-06,
"loss": 2.6431,
"step": 103000
},
{
"epoch": 8.64,
"learning_rate": 6.824628733522444e-06,
"loss": 2.6418,
"step": 103500
},
{
"epoch": 8.68,
"learning_rate": 6.616052060737528e-06,
"loss": 2.6437,
"step": 104000
},
{
"epoch": 8.72,
"learning_rate": 6.407475387952612e-06,
"loss": 2.6398,
"step": 104500
},
{
"epoch": 8.76,
"learning_rate": 6.198898715167696e-06,
"loss": 2.6565,
"step": 105000
},
{
"epoch": 8.8,
"learning_rate": 5.99032204238278e-06,
"loss": 2.6403,
"step": 105500
},
{
"epoch": 8.84,
"learning_rate": 5.7817453695978644e-06,
"loss": 2.6243,
"step": 106000
},
{
"epoch": 8.89,
"learning_rate": 5.5731686968129486e-06,
"loss": 2.6253,
"step": 106500
},
{
"epoch": 8.93,
"learning_rate": 5.3645920240280335e-06,
"loss": 2.6358,
"step": 107000
},
{
"epoch": 8.97,
"learning_rate": 5.156015351243118e-06,
"loss": 2.6267,
"step": 107500
},
{
"epoch": 9.01,
"learning_rate": 4.947438678458202e-06,
"loss": 2.6483,
"step": 108000
},
{
"epoch": 9.05,
"learning_rate": 4.738862005673286e-06,
"loss": 2.6325,
"step": 108500
},
{
"epoch": 9.09,
"learning_rate": 4.53028533288837e-06,
"loss": 2.6104,
"step": 109000
},
{
"epoch": 9.14,
"learning_rate": 4.321708660103454e-06,
"loss": 2.6263,
"step": 109500
},
{
"epoch": 9.18,
"learning_rate": 4.113131987318538e-06,
"loss": 2.6118,
"step": 110000
}
],
"max_steps": 119860,
"num_train_epochs": 10,
"total_flos": 4.514572271830057e+17,
"trial_name": null,
"trial_params": null
}