m2m100_418M_nya_en_rel / trainer_state.json
Davlan's picture
Upload trainer_state.json
1f7d5c7
raw
history blame
No virus
57.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 232677,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.989255491518285e-05,
"loss": 2.956,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.97851098303657e-05,
"loss": 2.4622,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 4.967766474554855e-05,
"loss": 2.3321,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 4.9570219660731405e-05,
"loss": 2.2234,
"step": 2000
},
{
"epoch": 0.03,
"learning_rate": 4.946277457591425e-05,
"loss": 2.156,
"step": 2500
},
{
"epoch": 0.04,
"learning_rate": 4.93553294910971e-05,
"loss": 2.1099,
"step": 3000
},
{
"epoch": 0.05,
"learning_rate": 4.924788440627995e-05,
"loss": 2.0528,
"step": 3500
},
{
"epoch": 0.05,
"learning_rate": 4.91404393214628e-05,
"loss": 2.0336,
"step": 4000
},
{
"epoch": 0.06,
"learning_rate": 4.903299423664565e-05,
"loss": 1.9938,
"step": 4500
},
{
"epoch": 0.06,
"learning_rate": 4.8925549151828506e-05,
"loss": 1.9726,
"step": 5000
},
{
"epoch": 0.07,
"learning_rate": 4.8818104067011355e-05,
"loss": 1.9286,
"step": 5500
},
{
"epoch": 0.08,
"learning_rate": 4.8710658982194204e-05,
"loss": 1.9026,
"step": 6000
},
{
"epoch": 0.08,
"learning_rate": 4.860321389737705e-05,
"loss": 1.9147,
"step": 6500
},
{
"epoch": 0.09,
"learning_rate": 4.84957688125599e-05,
"loss": 1.8774,
"step": 7000
},
{
"epoch": 0.1,
"learning_rate": 4.838832372774275e-05,
"loss": 1.8614,
"step": 7500
},
{
"epoch": 0.1,
"learning_rate": 4.82808786429256e-05,
"loss": 1.8656,
"step": 8000
},
{
"epoch": 0.11,
"learning_rate": 4.8173433558108456e-05,
"loss": 1.8337,
"step": 8500
},
{
"epoch": 0.12,
"learning_rate": 4.80659884732913e-05,
"loss": 1.8322,
"step": 9000
},
{
"epoch": 0.12,
"learning_rate": 4.7958543388474154e-05,
"loss": 1.7918,
"step": 9500
},
{
"epoch": 0.13,
"learning_rate": 4.7851098303657e-05,
"loss": 1.7813,
"step": 10000
},
{
"epoch": 0.14,
"learning_rate": 4.774365321883985e-05,
"loss": 1.7649,
"step": 10500
},
{
"epoch": 0.14,
"learning_rate": 4.76362081340227e-05,
"loss": 1.7451,
"step": 11000
},
{
"epoch": 0.15,
"learning_rate": 4.752876304920556e-05,
"loss": 1.7723,
"step": 11500
},
{
"epoch": 0.15,
"learning_rate": 4.74213179643884e-05,
"loss": 1.7512,
"step": 12000
},
{
"epoch": 0.16,
"learning_rate": 4.7313872879571255e-05,
"loss": 1.728,
"step": 12500
},
{
"epoch": 0.17,
"learning_rate": 4.7206427794754104e-05,
"loss": 1.728,
"step": 13000
},
{
"epoch": 0.17,
"learning_rate": 4.709898270993695e-05,
"loss": 1.7277,
"step": 13500
},
{
"epoch": 0.18,
"learning_rate": 4.69915376251198e-05,
"loss": 1.7009,
"step": 14000
},
{
"epoch": 0.19,
"learning_rate": 4.688409254030265e-05,
"loss": 1.6931,
"step": 14500
},
{
"epoch": 0.19,
"learning_rate": 4.677664745548551e-05,
"loss": 1.734,
"step": 15000
},
{
"epoch": 0.2,
"learning_rate": 4.666920237066835e-05,
"loss": 1.6718,
"step": 15500
},
{
"epoch": 0.21,
"learning_rate": 4.6561757285851205e-05,
"loss": 1.6895,
"step": 16000
},
{
"epoch": 0.21,
"learning_rate": 4.6454312201034054e-05,
"loss": 1.7088,
"step": 16500
},
{
"epoch": 0.22,
"learning_rate": 4.63468671162169e-05,
"loss": 1.6694,
"step": 17000
},
{
"epoch": 0.23,
"learning_rate": 4.623942203139975e-05,
"loss": 1.6648,
"step": 17500
},
{
"epoch": 0.23,
"learning_rate": 4.613197694658261e-05,
"loss": 1.6718,
"step": 18000
},
{
"epoch": 0.24,
"learning_rate": 4.602453186176545e-05,
"loss": 1.6606,
"step": 18500
},
{
"epoch": 0.24,
"learning_rate": 4.5917086776948306e-05,
"loss": 1.6807,
"step": 19000
},
{
"epoch": 0.25,
"learning_rate": 4.5809641692131155e-05,
"loss": 1.6352,
"step": 19500
},
{
"epoch": 0.26,
"learning_rate": 4.5702196607314004e-05,
"loss": 1.6454,
"step": 20000
},
{
"epoch": 0.26,
"learning_rate": 4.559475152249685e-05,
"loss": 1.6509,
"step": 20500
},
{
"epoch": 0.27,
"learning_rate": 4.548730643767971e-05,
"loss": 1.6083,
"step": 21000
},
{
"epoch": 0.28,
"learning_rate": 4.537986135286255e-05,
"loss": 1.6177,
"step": 21500
},
{
"epoch": 0.28,
"learning_rate": 4.527241626804541e-05,
"loss": 1.6097,
"step": 22000
},
{
"epoch": 0.29,
"learning_rate": 4.5164971183228256e-05,
"loss": 1.6025,
"step": 22500
},
{
"epoch": 0.3,
"learning_rate": 4.5057526098411105e-05,
"loss": 1.6162,
"step": 23000
},
{
"epoch": 0.3,
"learning_rate": 4.4950081013593954e-05,
"loss": 1.6092,
"step": 23500
},
{
"epoch": 0.31,
"learning_rate": 4.48426359287768e-05,
"loss": 1.592,
"step": 24000
},
{
"epoch": 0.32,
"learning_rate": 4.473519084395966e-05,
"loss": 1.5961,
"step": 24500
},
{
"epoch": 0.32,
"learning_rate": 4.46277457591425e-05,
"loss": 1.5711,
"step": 25000
},
{
"epoch": 0.33,
"learning_rate": 4.4520300674325357e-05,
"loss": 1.5789,
"step": 25500
},
{
"epoch": 0.34,
"learning_rate": 4.4412855589508206e-05,
"loss": 1.5774,
"step": 26000
},
{
"epoch": 0.34,
"learning_rate": 4.4305410504691055e-05,
"loss": 1.5854,
"step": 26500
},
{
"epoch": 0.35,
"learning_rate": 4.4197965419873904e-05,
"loss": 1.5462,
"step": 27000
},
{
"epoch": 0.35,
"learning_rate": 4.409052033505676e-05,
"loss": 1.5591,
"step": 27500
},
{
"epoch": 0.36,
"learning_rate": 4.39830752502396e-05,
"loss": 1.5775,
"step": 28000
},
{
"epoch": 0.37,
"learning_rate": 4.387563016542246e-05,
"loss": 1.542,
"step": 28500
},
{
"epoch": 0.37,
"learning_rate": 4.3768185080605306e-05,
"loss": 1.5408,
"step": 29000
},
{
"epoch": 0.38,
"learning_rate": 4.3660739995788155e-05,
"loss": 1.5484,
"step": 29500
},
{
"epoch": 0.39,
"learning_rate": 4.3553294910971004e-05,
"loss": 1.5625,
"step": 30000
},
{
"epoch": 0.39,
"learning_rate": 4.3445849826153853e-05,
"loss": 1.5276,
"step": 30500
},
{
"epoch": 0.4,
"learning_rate": 4.33384047413367e-05,
"loss": 1.5524,
"step": 31000
},
{
"epoch": 0.41,
"learning_rate": 4.323095965651955e-05,
"loss": 1.5458,
"step": 31500
},
{
"epoch": 0.41,
"learning_rate": 4.312351457170241e-05,
"loss": 1.541,
"step": 32000
},
{
"epoch": 0.42,
"learning_rate": 4.3016069486885256e-05,
"loss": 1.5313,
"step": 32500
},
{
"epoch": 0.43,
"learning_rate": 4.2908624402068105e-05,
"loss": 1.5507,
"step": 33000
},
{
"epoch": 0.43,
"learning_rate": 4.2801179317250954e-05,
"loss": 1.5189,
"step": 33500
},
{
"epoch": 0.44,
"learning_rate": 4.269373423243381e-05,
"loss": 1.5292,
"step": 34000
},
{
"epoch": 0.44,
"learning_rate": 4.258628914761665e-05,
"loss": 1.5109,
"step": 34500
},
{
"epoch": 0.45,
"learning_rate": 4.247884406279951e-05,
"loss": 1.4981,
"step": 35000
},
{
"epoch": 0.46,
"learning_rate": 4.237139897798236e-05,
"loss": 1.5229,
"step": 35500
},
{
"epoch": 0.46,
"learning_rate": 4.2263953893165206e-05,
"loss": 1.496,
"step": 36000
},
{
"epoch": 0.47,
"learning_rate": 4.2156508808348055e-05,
"loss": 1.4987,
"step": 36500
},
{
"epoch": 0.48,
"learning_rate": 4.2049063723530904e-05,
"loss": 1.5046,
"step": 37000
},
{
"epoch": 0.48,
"learning_rate": 4.194161863871375e-05,
"loss": 1.5233,
"step": 37500
},
{
"epoch": 0.49,
"learning_rate": 4.18341735538966e-05,
"loss": 1.4931,
"step": 38000
},
{
"epoch": 0.5,
"learning_rate": 4.172672846907946e-05,
"loss": 1.5,
"step": 38500
},
{
"epoch": 0.5,
"learning_rate": 4.16192833842623e-05,
"loss": 1.5132,
"step": 39000
},
{
"epoch": 0.51,
"learning_rate": 4.1511838299445156e-05,
"loss": 1.4723,
"step": 39500
},
{
"epoch": 0.52,
"learning_rate": 4.1404393214628005e-05,
"loss": 1.5036,
"step": 40000
},
{
"epoch": 0.52,
"learning_rate": 4.1296948129810854e-05,
"loss": 1.4907,
"step": 40500
},
{
"epoch": 0.53,
"learning_rate": 4.11895030449937e-05,
"loss": 1.4806,
"step": 41000
},
{
"epoch": 0.54,
"learning_rate": 4.108205796017656e-05,
"loss": 1.4775,
"step": 41500
},
{
"epoch": 0.54,
"learning_rate": 4.097461287535941e-05,
"loss": 1.466,
"step": 42000
},
{
"epoch": 0.55,
"learning_rate": 4.086716779054226e-05,
"loss": 1.4855,
"step": 42500
},
{
"epoch": 0.55,
"learning_rate": 4.0759722705725106e-05,
"loss": 1.4749,
"step": 43000
},
{
"epoch": 0.56,
"learning_rate": 4.0652277620907955e-05,
"loss": 1.4617,
"step": 43500
},
{
"epoch": 0.57,
"learning_rate": 4.0544832536090804e-05,
"loss": 1.4674,
"step": 44000
},
{
"epoch": 0.57,
"learning_rate": 4.043738745127365e-05,
"loss": 1.437,
"step": 44500
},
{
"epoch": 0.58,
"learning_rate": 4.032994236645651e-05,
"loss": 1.4632,
"step": 45000
},
{
"epoch": 0.59,
"learning_rate": 4.022249728163935e-05,
"loss": 1.4692,
"step": 45500
},
{
"epoch": 0.59,
"learning_rate": 4.011505219682221e-05,
"loss": 1.4655,
"step": 46000
},
{
"epoch": 0.6,
"learning_rate": 4.0007607112005056e-05,
"loss": 1.4469,
"step": 46500
},
{
"epoch": 0.61,
"learning_rate": 3.9900162027187905e-05,
"loss": 1.4606,
"step": 47000
},
{
"epoch": 0.61,
"learning_rate": 3.9792716942370754e-05,
"loss": 1.4746,
"step": 47500
},
{
"epoch": 0.62,
"learning_rate": 3.968527185755361e-05,
"loss": 1.4479,
"step": 48000
},
{
"epoch": 0.63,
"learning_rate": 3.957782677273645e-05,
"loss": 1.4187,
"step": 48500
},
{
"epoch": 0.63,
"learning_rate": 3.947038168791931e-05,
"loss": 1.4497,
"step": 49000
},
{
"epoch": 0.64,
"learning_rate": 3.936293660310216e-05,
"loss": 1.4174,
"step": 49500
},
{
"epoch": 0.64,
"learning_rate": 3.9255491518285006e-05,
"loss": 1.4391,
"step": 50000
},
{
"epoch": 0.65,
"learning_rate": 3.9148046433467855e-05,
"loss": 1.4355,
"step": 50500
},
{
"epoch": 0.66,
"learning_rate": 3.9040601348650704e-05,
"loss": 1.441,
"step": 51000
},
{
"epoch": 0.66,
"learning_rate": 3.893315626383356e-05,
"loss": 1.4265,
"step": 51500
},
{
"epoch": 0.67,
"learning_rate": 3.88257111790164e-05,
"loss": 1.4499,
"step": 52000
},
{
"epoch": 0.68,
"learning_rate": 3.871826609419926e-05,
"loss": 1.4564,
"step": 52500
},
{
"epoch": 0.68,
"learning_rate": 3.861082100938211e-05,
"loss": 1.4362,
"step": 53000
},
{
"epoch": 0.69,
"learning_rate": 3.8503375924564956e-05,
"loss": 1.4338,
"step": 53500
},
{
"epoch": 0.7,
"learning_rate": 3.8395930839747805e-05,
"loss": 1.4099,
"step": 54000
},
{
"epoch": 0.7,
"learning_rate": 3.828848575493066e-05,
"loss": 1.4213,
"step": 54500
},
{
"epoch": 0.71,
"learning_rate": 3.81810406701135e-05,
"loss": 1.4142,
"step": 55000
},
{
"epoch": 0.72,
"learning_rate": 3.807359558529636e-05,
"loss": 1.411,
"step": 55500
},
{
"epoch": 0.72,
"learning_rate": 3.796615050047921e-05,
"loss": 1.4071,
"step": 56000
},
{
"epoch": 0.73,
"learning_rate": 3.785870541566206e-05,
"loss": 1.4536,
"step": 56500
},
{
"epoch": 0.73,
"learning_rate": 3.7751260330844906e-05,
"loss": 1.4205,
"step": 57000
},
{
"epoch": 0.74,
"learning_rate": 3.764381524602776e-05,
"loss": 1.4275,
"step": 57500
},
{
"epoch": 0.75,
"learning_rate": 3.7536370161210604e-05,
"loss": 1.4142,
"step": 58000
},
{
"epoch": 0.75,
"learning_rate": 3.742892507639346e-05,
"loss": 1.4007,
"step": 58500
},
{
"epoch": 0.76,
"learning_rate": 3.732147999157631e-05,
"loss": 1.3993,
"step": 59000
},
{
"epoch": 0.77,
"learning_rate": 3.721403490675916e-05,
"loss": 1.3797,
"step": 59500
},
{
"epoch": 0.77,
"learning_rate": 3.7106589821942007e-05,
"loss": 1.4012,
"step": 60000
},
{
"epoch": 0.78,
"learning_rate": 3.6999144737124856e-05,
"loss": 1.4133,
"step": 60500
},
{
"epoch": 0.79,
"learning_rate": 3.689169965230771e-05,
"loss": 1.4276,
"step": 61000
},
{
"epoch": 0.79,
"learning_rate": 3.6784254567490554e-05,
"loss": 1.4062,
"step": 61500
},
{
"epoch": 0.8,
"learning_rate": 3.667680948267341e-05,
"loss": 1.3683,
"step": 62000
},
{
"epoch": 0.81,
"learning_rate": 3.656936439785626e-05,
"loss": 1.3942,
"step": 62500
},
{
"epoch": 0.81,
"learning_rate": 3.646191931303911e-05,
"loss": 1.3861,
"step": 63000
},
{
"epoch": 0.82,
"learning_rate": 3.6354474228221956e-05,
"loss": 1.4005,
"step": 63500
},
{
"epoch": 0.83,
"learning_rate": 3.624702914340481e-05,
"loss": 1.3732,
"step": 64000
},
{
"epoch": 0.83,
"learning_rate": 3.6139584058587654e-05,
"loss": 1.4009,
"step": 64500
},
{
"epoch": 0.84,
"learning_rate": 3.603213897377051e-05,
"loss": 1.3785,
"step": 65000
},
{
"epoch": 0.84,
"learning_rate": 3.592469388895336e-05,
"loss": 1.3883,
"step": 65500
},
{
"epoch": 0.85,
"learning_rate": 3.581724880413621e-05,
"loss": 1.351,
"step": 66000
},
{
"epoch": 0.86,
"learning_rate": 3.570980371931906e-05,
"loss": 1.3579,
"step": 66500
},
{
"epoch": 0.86,
"learning_rate": 3.5602358634501906e-05,
"loss": 1.4154,
"step": 67000
},
{
"epoch": 0.87,
"learning_rate": 3.5494913549684755e-05,
"loss": 1.3607,
"step": 67500
},
{
"epoch": 0.88,
"learning_rate": 3.5387468464867604e-05,
"loss": 1.3604,
"step": 68000
},
{
"epoch": 0.88,
"learning_rate": 3.528002338005046e-05,
"loss": 1.3624,
"step": 68500
},
{
"epoch": 0.89,
"learning_rate": 3.517257829523331e-05,
"loss": 1.3805,
"step": 69000
},
{
"epoch": 0.9,
"learning_rate": 3.506513321041616e-05,
"loss": 1.3659,
"step": 69500
},
{
"epoch": 0.9,
"learning_rate": 3.495768812559901e-05,
"loss": 1.3777,
"step": 70000
},
{
"epoch": 0.91,
"learning_rate": 3.485024304078186e-05,
"loss": 1.3516,
"step": 70500
},
{
"epoch": 0.92,
"learning_rate": 3.4742797955964705e-05,
"loss": 1.3718,
"step": 71000
},
{
"epoch": 0.92,
"learning_rate": 3.463535287114756e-05,
"loss": 1.37,
"step": 71500
},
{
"epoch": 0.93,
"learning_rate": 3.452790778633041e-05,
"loss": 1.3502,
"step": 72000
},
{
"epoch": 0.93,
"learning_rate": 3.442046270151326e-05,
"loss": 1.3591,
"step": 72500
},
{
"epoch": 0.94,
"learning_rate": 3.431301761669611e-05,
"loss": 1.3483,
"step": 73000
},
{
"epoch": 0.95,
"learning_rate": 3.420557253187896e-05,
"loss": 1.3578,
"step": 73500
},
{
"epoch": 0.95,
"learning_rate": 3.4098127447061806e-05,
"loss": 1.3465,
"step": 74000
},
{
"epoch": 0.96,
"learning_rate": 3.3990682362244655e-05,
"loss": 1.3694,
"step": 74500
},
{
"epoch": 0.97,
"learning_rate": 3.388323727742751e-05,
"loss": 1.3631,
"step": 75000
},
{
"epoch": 0.97,
"learning_rate": 3.377579219261035e-05,
"loss": 1.3675,
"step": 75500
},
{
"epoch": 0.98,
"learning_rate": 3.366834710779321e-05,
"loss": 1.3631,
"step": 76000
},
{
"epoch": 0.99,
"learning_rate": 3.356090202297606e-05,
"loss": 1.3432,
"step": 76500
},
{
"epoch": 0.99,
"learning_rate": 3.3453456938158914e-05,
"loss": 1.3441,
"step": 77000
},
{
"epoch": 1.0,
"learning_rate": 3.3346011853341756e-05,
"loss": 1.3543,
"step": 77500
},
{
"epoch": 1.01,
"learning_rate": 3.323856676852461e-05,
"loss": 1.2648,
"step": 78000
},
{
"epoch": 1.01,
"learning_rate": 3.313112168370746e-05,
"loss": 1.2416,
"step": 78500
},
{
"epoch": 1.02,
"learning_rate": 3.302367659889031e-05,
"loss": 1.2426,
"step": 79000
},
{
"epoch": 1.03,
"learning_rate": 3.291623151407316e-05,
"loss": 1.2268,
"step": 79500
},
{
"epoch": 1.03,
"learning_rate": 3.280878642925601e-05,
"loss": 1.2366,
"step": 80000
},
{
"epoch": 1.04,
"learning_rate": 3.270134134443886e-05,
"loss": 1.2198,
"step": 80500
},
{
"epoch": 1.04,
"learning_rate": 3.2593896259621706e-05,
"loss": 1.2172,
"step": 81000
},
{
"epoch": 1.05,
"learning_rate": 3.248645117480456e-05,
"loss": 1.226,
"step": 81500
},
{
"epoch": 1.06,
"learning_rate": 3.2379006089987404e-05,
"loss": 1.2192,
"step": 82000
},
{
"epoch": 1.06,
"learning_rate": 3.227156100517026e-05,
"loss": 1.2385,
"step": 82500
},
{
"epoch": 1.07,
"learning_rate": 3.216411592035311e-05,
"loss": 1.2127,
"step": 83000
},
{
"epoch": 1.08,
"learning_rate": 3.205667083553596e-05,
"loss": 1.234,
"step": 83500
},
{
"epoch": 1.08,
"learning_rate": 3.194922575071881e-05,
"loss": 1.2267,
"step": 84000
},
{
"epoch": 1.09,
"learning_rate": 3.184178066590166e-05,
"loss": 1.2302,
"step": 84500
},
{
"epoch": 1.1,
"learning_rate": 3.1734335581084505e-05,
"loss": 1.2387,
"step": 85000
},
{
"epoch": 1.1,
"learning_rate": 3.162689049626736e-05,
"loss": 1.243,
"step": 85500
},
{
"epoch": 1.11,
"learning_rate": 3.151944541145021e-05,
"loss": 1.2139,
"step": 86000
},
{
"epoch": 1.12,
"learning_rate": 3.141200032663306e-05,
"loss": 1.2217,
"step": 86500
},
{
"epoch": 1.12,
"learning_rate": 3.130455524181591e-05,
"loss": 1.2122,
"step": 87000
},
{
"epoch": 1.13,
"learning_rate": 3.119711015699876e-05,
"loss": 1.2253,
"step": 87500
},
{
"epoch": 1.13,
"learning_rate": 3.108966507218161e-05,
"loss": 1.2172,
"step": 88000
},
{
"epoch": 1.14,
"learning_rate": 3.0982219987364455e-05,
"loss": 1.2296,
"step": 88500
},
{
"epoch": 1.15,
"learning_rate": 3.087477490254731e-05,
"loss": 1.2216,
"step": 89000
},
{
"epoch": 1.15,
"learning_rate": 3.076732981773016e-05,
"loss": 1.2205,
"step": 89500
},
{
"epoch": 1.16,
"learning_rate": 3.065988473291301e-05,
"loss": 1.2127,
"step": 90000
},
{
"epoch": 1.17,
"learning_rate": 3.055243964809586e-05,
"loss": 1.2324,
"step": 90500
},
{
"epoch": 1.17,
"learning_rate": 3.044499456327871e-05,
"loss": 1.2124,
"step": 91000
},
{
"epoch": 1.18,
"learning_rate": 3.033754947846156e-05,
"loss": 1.2364,
"step": 91500
},
{
"epoch": 1.19,
"learning_rate": 3.0230104393644408e-05,
"loss": 1.2111,
"step": 92000
},
{
"epoch": 1.19,
"learning_rate": 3.012265930882726e-05,
"loss": 1.2082,
"step": 92500
},
{
"epoch": 1.2,
"learning_rate": 3.0015214224010106e-05,
"loss": 1.2292,
"step": 93000
},
{
"epoch": 1.21,
"learning_rate": 2.990776913919296e-05,
"loss": 1.2236,
"step": 93500
},
{
"epoch": 1.21,
"learning_rate": 2.980032405437581e-05,
"loss": 1.2178,
"step": 94000
},
{
"epoch": 1.22,
"learning_rate": 2.9692878969558657e-05,
"loss": 1.2165,
"step": 94500
},
{
"epoch": 1.22,
"learning_rate": 2.958543388474151e-05,
"loss": 1.21,
"step": 95000
},
{
"epoch": 1.23,
"learning_rate": 2.947798879992436e-05,
"loss": 1.2239,
"step": 95500
},
{
"epoch": 1.24,
"learning_rate": 2.9370543715107214e-05,
"loss": 1.2094,
"step": 96000
},
{
"epoch": 1.24,
"learning_rate": 2.926309863029006e-05,
"loss": 1.1993,
"step": 96500
},
{
"epoch": 1.25,
"learning_rate": 2.9155653545472912e-05,
"loss": 1.2143,
"step": 97000
},
{
"epoch": 1.26,
"learning_rate": 2.9048208460655764e-05,
"loss": 1.2092,
"step": 97500
},
{
"epoch": 1.26,
"learning_rate": 2.894076337583861e-05,
"loss": 1.2128,
"step": 98000
},
{
"epoch": 1.27,
"learning_rate": 2.8833318291021462e-05,
"loss": 1.2339,
"step": 98500
},
{
"epoch": 1.28,
"learning_rate": 2.872587320620431e-05,
"loss": 1.2002,
"step": 99000
},
{
"epoch": 1.28,
"learning_rate": 2.8618428121387157e-05,
"loss": 1.2068,
"step": 99500
},
{
"epoch": 1.29,
"learning_rate": 2.851098303657001e-05,
"loss": 1.2026,
"step": 100000
},
{
"epoch": 1.3,
"learning_rate": 2.840353795175286e-05,
"loss": 1.2186,
"step": 100500
},
{
"epoch": 1.3,
"learning_rate": 2.8296092866935707e-05,
"loss": 1.2251,
"step": 101000
},
{
"epoch": 1.31,
"learning_rate": 2.818864778211856e-05,
"loss": 1.2054,
"step": 101500
},
{
"epoch": 1.32,
"learning_rate": 2.8081202697301412e-05,
"loss": 1.2136,
"step": 102000
},
{
"epoch": 1.32,
"learning_rate": 2.7973757612484258e-05,
"loss": 1.2064,
"step": 102500
},
{
"epoch": 1.33,
"learning_rate": 2.786631252766711e-05,
"loss": 1.1948,
"step": 103000
},
{
"epoch": 1.33,
"learning_rate": 2.7758867442849963e-05,
"loss": 1.1918,
"step": 103500
},
{
"epoch": 1.34,
"learning_rate": 2.7651422358032808e-05,
"loss": 1.2137,
"step": 104000
},
{
"epoch": 1.35,
"learning_rate": 2.754397727321566e-05,
"loss": 1.1977,
"step": 104500
},
{
"epoch": 1.35,
"learning_rate": 2.7436532188398513e-05,
"loss": 1.2128,
"step": 105000
},
{
"epoch": 1.36,
"learning_rate": 2.7329087103581362e-05,
"loss": 1.1802,
"step": 105500
},
{
"epoch": 1.37,
"learning_rate": 2.722164201876421e-05,
"loss": 1.1854,
"step": 106000
},
{
"epoch": 1.37,
"learning_rate": 2.711419693394706e-05,
"loss": 1.1809,
"step": 106500
},
{
"epoch": 1.38,
"learning_rate": 2.7006751849129912e-05,
"loss": 1.1923,
"step": 107000
},
{
"epoch": 1.39,
"learning_rate": 2.6899306764312758e-05,
"loss": 1.1918,
"step": 107500
},
{
"epoch": 1.39,
"learning_rate": 2.679186167949561e-05,
"loss": 1.1987,
"step": 108000
},
{
"epoch": 1.4,
"learning_rate": 2.6684416594678463e-05,
"loss": 1.1786,
"step": 108500
},
{
"epoch": 1.41,
"learning_rate": 2.657697150986131e-05,
"loss": 1.179,
"step": 109000
},
{
"epoch": 1.41,
"learning_rate": 2.646952642504416e-05,
"loss": 1.194,
"step": 109500
},
{
"epoch": 1.42,
"learning_rate": 2.6362081340227013e-05,
"loss": 1.1954,
"step": 110000
},
{
"epoch": 1.42,
"learning_rate": 2.625463625540986e-05,
"loss": 1.1996,
"step": 110500
},
{
"epoch": 1.43,
"learning_rate": 2.614719117059271e-05,
"loss": 1.2026,
"step": 111000
},
{
"epoch": 1.44,
"learning_rate": 2.6039746085775564e-05,
"loss": 1.1892,
"step": 111500
},
{
"epoch": 1.44,
"learning_rate": 2.593230100095841e-05,
"loss": 1.1968,
"step": 112000
},
{
"epoch": 1.45,
"learning_rate": 2.5824855916141262e-05,
"loss": 1.1719,
"step": 112500
},
{
"epoch": 1.46,
"learning_rate": 2.571741083132411e-05,
"loss": 1.1855,
"step": 113000
},
{
"epoch": 1.46,
"learning_rate": 2.5609965746506963e-05,
"loss": 1.1762,
"step": 113500
},
{
"epoch": 1.47,
"learning_rate": 2.550252066168981e-05,
"loss": 1.187,
"step": 114000
},
{
"epoch": 1.48,
"learning_rate": 2.539507557687266e-05,
"loss": 1.1892,
"step": 114500
},
{
"epoch": 1.48,
"learning_rate": 2.5287630492055514e-05,
"loss": 1.2085,
"step": 115000
},
{
"epoch": 1.49,
"learning_rate": 2.518018540723836e-05,
"loss": 1.181,
"step": 115500
},
{
"epoch": 1.5,
"learning_rate": 2.5072740322421212e-05,
"loss": 1.1654,
"step": 116000
},
{
"epoch": 1.5,
"learning_rate": 2.496529523760406e-05,
"loss": 1.1677,
"step": 116500
},
{
"epoch": 1.51,
"learning_rate": 2.4857850152786913e-05,
"loss": 1.2037,
"step": 117000
},
{
"epoch": 1.51,
"learning_rate": 2.4750405067969762e-05,
"loss": 1.1787,
"step": 117500
},
{
"epoch": 1.52,
"learning_rate": 2.464295998315261e-05,
"loss": 1.179,
"step": 118000
},
{
"epoch": 1.53,
"learning_rate": 2.4535514898335464e-05,
"loss": 1.2019,
"step": 118500
},
{
"epoch": 1.53,
"learning_rate": 2.4428069813518313e-05,
"loss": 1.1787,
"step": 119000
},
{
"epoch": 1.54,
"learning_rate": 2.432062472870116e-05,
"loss": 1.1677,
"step": 119500
},
{
"epoch": 1.55,
"learning_rate": 2.4213179643884014e-05,
"loss": 1.1919,
"step": 120000
},
{
"epoch": 1.55,
"learning_rate": 2.4105734559066863e-05,
"loss": 1.1859,
"step": 120500
},
{
"epoch": 1.56,
"learning_rate": 2.3998289474249712e-05,
"loss": 1.1597,
"step": 121000
},
{
"epoch": 1.57,
"learning_rate": 2.389084438943256e-05,
"loss": 1.1806,
"step": 121500
},
{
"epoch": 1.57,
"learning_rate": 2.378339930461541e-05,
"loss": 1.1694,
"step": 122000
},
{
"epoch": 1.58,
"learning_rate": 2.3675954219798263e-05,
"loss": 1.1896,
"step": 122500
},
{
"epoch": 1.59,
"learning_rate": 2.356850913498111e-05,
"loss": 1.1897,
"step": 123000
},
{
"epoch": 1.59,
"learning_rate": 2.346106405016396e-05,
"loss": 1.1796,
"step": 123500
},
{
"epoch": 1.6,
"learning_rate": 2.3353618965346813e-05,
"loss": 1.1745,
"step": 124000
},
{
"epoch": 1.61,
"learning_rate": 2.3246173880529662e-05,
"loss": 1.1828,
"step": 124500
},
{
"epoch": 1.61,
"learning_rate": 2.3138728795712514e-05,
"loss": 1.1875,
"step": 125000
},
{
"epoch": 1.62,
"learning_rate": 2.3031283710895363e-05,
"loss": 1.1716,
"step": 125500
},
{
"epoch": 1.62,
"learning_rate": 2.2923838626078212e-05,
"loss": 1.1768,
"step": 126000
},
{
"epoch": 1.63,
"learning_rate": 2.2816393541261065e-05,
"loss": 1.1821,
"step": 126500
},
{
"epoch": 1.64,
"learning_rate": 2.2708948456443914e-05,
"loss": 1.1672,
"step": 127000
},
{
"epoch": 1.64,
"learning_rate": 2.2601503371626763e-05,
"loss": 1.173,
"step": 127500
},
{
"epoch": 1.65,
"learning_rate": 2.2494058286809612e-05,
"loss": 1.1751,
"step": 128000
},
{
"epoch": 1.66,
"learning_rate": 2.238661320199246e-05,
"loss": 1.1967,
"step": 128500
},
{
"epoch": 1.66,
"learning_rate": 2.227916811717531e-05,
"loss": 1.1695,
"step": 129000
},
{
"epoch": 1.67,
"learning_rate": 2.2171723032358162e-05,
"loss": 1.179,
"step": 129500
},
{
"epoch": 1.68,
"learning_rate": 2.206427794754101e-05,
"loss": 1.1793,
"step": 130000
},
{
"epoch": 1.68,
"learning_rate": 2.1956832862723864e-05,
"loss": 1.1579,
"step": 130500
},
{
"epoch": 1.69,
"learning_rate": 2.1849387777906713e-05,
"loss": 1.1466,
"step": 131000
},
{
"epoch": 1.7,
"learning_rate": 2.1741942693089562e-05,
"loss": 1.208,
"step": 131500
},
{
"epoch": 1.7,
"learning_rate": 2.1634497608272414e-05,
"loss": 1.1592,
"step": 132000
},
{
"epoch": 1.71,
"learning_rate": 2.1527052523455263e-05,
"loss": 1.1558,
"step": 132500
},
{
"epoch": 1.71,
"learning_rate": 2.1419607438638112e-05,
"loss": 1.158,
"step": 133000
},
{
"epoch": 1.72,
"learning_rate": 2.1312162353820965e-05,
"loss": 1.1566,
"step": 133500
},
{
"epoch": 1.73,
"learning_rate": 2.1204717269003814e-05,
"loss": 1.1722,
"step": 134000
},
{
"epoch": 1.73,
"learning_rate": 2.1097272184186666e-05,
"loss": 1.1612,
"step": 134500
},
{
"epoch": 1.74,
"learning_rate": 2.0989827099369515e-05,
"loss": 1.1706,
"step": 135000
},
{
"epoch": 1.75,
"learning_rate": 2.0882382014552364e-05,
"loss": 1.1746,
"step": 135500
},
{
"epoch": 1.75,
"learning_rate": 2.0774936929735213e-05,
"loss": 1.1568,
"step": 136000
},
{
"epoch": 1.76,
"learning_rate": 2.0667491844918062e-05,
"loss": 1.1564,
"step": 136500
},
{
"epoch": 1.77,
"learning_rate": 2.056004676010091e-05,
"loss": 1.1566,
"step": 137000
},
{
"epoch": 1.77,
"learning_rate": 2.0452601675283764e-05,
"loss": 1.1684,
"step": 137500
},
{
"epoch": 1.78,
"learning_rate": 2.0345156590466613e-05,
"loss": 1.166,
"step": 138000
},
{
"epoch": 1.79,
"learning_rate": 2.0237711505649465e-05,
"loss": 1.1439,
"step": 138500
},
{
"epoch": 1.79,
"learning_rate": 2.0130266420832314e-05,
"loss": 1.1535,
"step": 139000
},
{
"epoch": 1.8,
"learning_rate": 2.0022821336015163e-05,
"loss": 1.1571,
"step": 139500
},
{
"epoch": 1.81,
"learning_rate": 1.9915376251198015e-05,
"loss": 1.1573,
"step": 140000
},
{
"epoch": 1.81,
"learning_rate": 1.9807931166380864e-05,
"loss": 1.1682,
"step": 140500
},
{
"epoch": 1.82,
"learning_rate": 1.9700486081563713e-05,
"loss": 1.1506,
"step": 141000
},
{
"epoch": 1.82,
"learning_rate": 1.9593040996746566e-05,
"loss": 1.1513,
"step": 141500
},
{
"epoch": 1.83,
"learning_rate": 1.9485595911929415e-05,
"loss": 1.17,
"step": 142000
},
{
"epoch": 1.84,
"learning_rate": 1.9378150827112264e-05,
"loss": 1.1538,
"step": 142500
},
{
"epoch": 1.84,
"learning_rate": 1.9270705742295113e-05,
"loss": 1.1461,
"step": 143000
},
{
"epoch": 1.85,
"learning_rate": 1.9163260657477962e-05,
"loss": 1.1353,
"step": 143500
},
{
"epoch": 1.86,
"learning_rate": 1.9055815572660814e-05,
"loss": 1.1523,
"step": 144000
},
{
"epoch": 1.86,
"learning_rate": 1.8948370487843663e-05,
"loss": 1.1547,
"step": 144500
},
{
"epoch": 1.87,
"learning_rate": 1.8840925403026512e-05,
"loss": 1.1092,
"step": 145000
},
{
"epoch": 1.88,
"learning_rate": 1.8733480318209365e-05,
"loss": 1.1662,
"step": 145500
},
{
"epoch": 1.88,
"learning_rate": 1.8626035233392214e-05,
"loss": 1.1539,
"step": 146000
},
{
"epoch": 1.89,
"learning_rate": 1.8518590148575063e-05,
"loss": 1.1571,
"step": 146500
},
{
"epoch": 1.9,
"learning_rate": 1.8411145063757915e-05,
"loss": 1.1228,
"step": 147000
},
{
"epoch": 1.9,
"learning_rate": 1.8303699978940764e-05,
"loss": 1.1421,
"step": 147500
},
{
"epoch": 1.91,
"learning_rate": 1.8196254894123617e-05,
"loss": 1.1603,
"step": 148000
},
{
"epoch": 1.91,
"learning_rate": 1.8088809809306466e-05,
"loss": 1.1051,
"step": 148500
},
{
"epoch": 1.92,
"learning_rate": 1.7981364724489315e-05,
"loss": 1.1423,
"step": 149000
},
{
"epoch": 1.93,
"learning_rate": 1.7873919639672164e-05,
"loss": 1.1548,
"step": 149500
},
{
"epoch": 1.93,
"learning_rate": 1.7766474554855013e-05,
"loss": 1.1424,
"step": 150000
},
{
"epoch": 1.94,
"learning_rate": 1.7659029470037862e-05,
"loss": 1.1435,
"step": 150500
},
{
"epoch": 1.95,
"learning_rate": 1.7551584385220714e-05,
"loss": 1.1592,
"step": 151000
},
{
"epoch": 1.95,
"learning_rate": 1.7444139300403563e-05,
"loss": 1.1364,
"step": 151500
},
{
"epoch": 1.96,
"learning_rate": 1.7336694215586416e-05,
"loss": 1.1412,
"step": 152000
},
{
"epoch": 1.97,
"learning_rate": 1.7229249130769265e-05,
"loss": 1.1428,
"step": 152500
},
{
"epoch": 1.97,
"learning_rate": 1.7121804045952114e-05,
"loss": 1.131,
"step": 153000
},
{
"epoch": 1.98,
"learning_rate": 1.7014358961134966e-05,
"loss": 1.1438,
"step": 153500
},
{
"epoch": 1.99,
"learning_rate": 1.6906913876317815e-05,
"loss": 1.117,
"step": 154000
},
{
"epoch": 1.99,
"learning_rate": 1.6799468791500664e-05,
"loss": 1.1552,
"step": 154500
},
{
"epoch": 2.0,
"learning_rate": 1.6692023706683516e-05,
"loss": 1.1362,
"step": 155000
},
{
"epoch": 2.0,
"learning_rate": 1.6584578621866365e-05,
"loss": 1.0437,
"step": 155500
},
{
"epoch": 2.01,
"learning_rate": 1.6477133537049214e-05,
"loss": 0.9948,
"step": 156000
},
{
"epoch": 2.02,
"learning_rate": 1.6369688452232067e-05,
"loss": 1.0094,
"step": 156500
},
{
"epoch": 2.02,
"learning_rate": 1.6262243367414916e-05,
"loss": 1.0137,
"step": 157000
},
{
"epoch": 2.03,
"learning_rate": 1.6154798282597765e-05,
"loss": 1.0197,
"step": 157500
},
{
"epoch": 2.04,
"learning_rate": 1.6047353197780614e-05,
"loss": 1.0113,
"step": 158000
},
{
"epoch": 2.04,
"learning_rate": 1.5939908112963463e-05,
"loss": 1.0123,
"step": 158500
},
{
"epoch": 2.05,
"learning_rate": 1.5832463028146315e-05,
"loss": 1.018,
"step": 159000
},
{
"epoch": 2.06,
"learning_rate": 1.5725017943329164e-05,
"loss": 1.0256,
"step": 159500
},
{
"epoch": 2.06,
"learning_rate": 1.5617572858512013e-05,
"loss": 0.9943,
"step": 160000
},
{
"epoch": 2.07,
"learning_rate": 1.5510127773694866e-05,
"loss": 1.0117,
"step": 160500
},
{
"epoch": 2.08,
"learning_rate": 1.5402682688877715e-05,
"loss": 1.0127,
"step": 161000
},
{
"epoch": 2.08,
"learning_rate": 1.5295237604060567e-05,
"loss": 1.012,
"step": 161500
},
{
"epoch": 2.09,
"learning_rate": 1.5187792519243416e-05,
"loss": 0.9863,
"step": 162000
},
{
"epoch": 2.1,
"learning_rate": 1.5080347434426265e-05,
"loss": 1.0253,
"step": 162500
},
{
"epoch": 2.1,
"learning_rate": 1.4972902349609116e-05,
"loss": 1.0143,
"step": 163000
},
{
"epoch": 2.11,
"learning_rate": 1.4865457264791965e-05,
"loss": 0.9931,
"step": 163500
},
{
"epoch": 2.11,
"learning_rate": 1.4758012179974814e-05,
"loss": 1.0074,
"step": 164000
},
{
"epoch": 2.12,
"learning_rate": 1.4650567095157666e-05,
"loss": 0.9954,
"step": 164500
},
{
"epoch": 2.13,
"learning_rate": 1.4543122010340515e-05,
"loss": 1.0081,
"step": 165000
},
{
"epoch": 2.13,
"learning_rate": 1.4435676925523368e-05,
"loss": 1.0072,
"step": 165500
},
{
"epoch": 2.14,
"learning_rate": 1.4328231840706217e-05,
"loss": 1.0019,
"step": 166000
},
{
"epoch": 2.15,
"learning_rate": 1.4220786755889066e-05,
"loss": 1.0164,
"step": 166500
},
{
"epoch": 2.15,
"learning_rate": 1.4113341671071917e-05,
"loss": 1.0449,
"step": 167000
},
{
"epoch": 2.16,
"learning_rate": 1.4005896586254766e-05,
"loss": 0.9966,
"step": 167500
},
{
"epoch": 2.17,
"learning_rate": 1.3898451501437615e-05,
"loss": 1.0105,
"step": 168000
},
{
"epoch": 2.17,
"learning_rate": 1.3791006416620467e-05,
"loss": 1.0485,
"step": 168500
},
{
"epoch": 2.18,
"learning_rate": 1.3683561331803316e-05,
"loss": 1.0047,
"step": 169000
},
{
"epoch": 2.19,
"learning_rate": 1.3576116246986165e-05,
"loss": 1.0103,
"step": 169500
},
{
"epoch": 2.19,
"learning_rate": 1.3468671162169016e-05,
"loss": 1.0032,
"step": 170000
},
{
"epoch": 2.2,
"learning_rate": 1.3361226077351865e-05,
"loss": 1.0013,
"step": 170500
},
{
"epoch": 2.2,
"learning_rate": 1.3253780992534717e-05,
"loss": 1.0028,
"step": 171000
},
{
"epoch": 2.21,
"learning_rate": 1.3146335907717566e-05,
"loss": 1.0134,
"step": 171500
},
{
"epoch": 2.22,
"learning_rate": 1.3038890822900415e-05,
"loss": 1.036,
"step": 172000
},
{
"epoch": 2.22,
"learning_rate": 1.2931445738083268e-05,
"loss": 1.0139,
"step": 172500
},
{
"epoch": 2.23,
"learning_rate": 1.2824000653266117e-05,
"loss": 0.9908,
"step": 173000
},
{
"epoch": 2.24,
"learning_rate": 1.2716555568448966e-05,
"loss": 0.9896,
"step": 173500
},
{
"epoch": 2.24,
"learning_rate": 1.2609110483631816e-05,
"loss": 0.9982,
"step": 174000
},
{
"epoch": 2.25,
"learning_rate": 1.2501665398814665e-05,
"loss": 1.0115,
"step": 174500
},
{
"epoch": 2.26,
"learning_rate": 1.2394220313997516e-05,
"loss": 0.9939,
"step": 175000
},
{
"epoch": 2.26,
"learning_rate": 1.2286775229180367e-05,
"loss": 0.9957,
"step": 175500
},
{
"epoch": 2.27,
"learning_rate": 1.2179330144363218e-05,
"loss": 1.0009,
"step": 176000
},
{
"epoch": 2.28,
"learning_rate": 1.2071885059546067e-05,
"loss": 1.0102,
"step": 176500
},
{
"epoch": 2.28,
"learning_rate": 1.1964439974728917e-05,
"loss": 1.0058,
"step": 177000
},
{
"epoch": 2.29,
"learning_rate": 1.1856994889911766e-05,
"loss": 0.9999,
"step": 177500
},
{
"epoch": 2.3,
"learning_rate": 1.1749549805094617e-05,
"loss": 1.0141,
"step": 178000
},
{
"epoch": 2.3,
"learning_rate": 1.1642104720277466e-05,
"loss": 1.0004,
"step": 178500
},
{
"epoch": 2.31,
"learning_rate": 1.1534659635460317e-05,
"loss": 0.9727,
"step": 179000
},
{
"epoch": 2.31,
"learning_rate": 1.1427214550643167e-05,
"loss": 0.9989,
"step": 179500
},
{
"epoch": 2.32,
"learning_rate": 1.1319769465826018e-05,
"loss": 0.9968,
"step": 180000
},
{
"epoch": 2.33,
"learning_rate": 1.1212324381008867e-05,
"loss": 1.0138,
"step": 180500
},
{
"epoch": 2.33,
"learning_rate": 1.1104879296191716e-05,
"loss": 1.0079,
"step": 181000
},
{
"epoch": 2.34,
"learning_rate": 1.0997434211374567e-05,
"loss": 0.9866,
"step": 181500
},
{
"epoch": 2.35,
"learning_rate": 1.0889989126557416e-05,
"loss": 1.0007,
"step": 182000
},
{
"epoch": 2.35,
"learning_rate": 1.0782544041740267e-05,
"loss": 1.004,
"step": 182500
},
{
"epoch": 2.36,
"learning_rate": 1.0675098956923117e-05,
"loss": 0.9913,
"step": 183000
},
{
"epoch": 2.37,
"learning_rate": 1.0567653872105968e-05,
"loss": 1.0002,
"step": 183500
},
{
"epoch": 2.37,
"learning_rate": 1.0460208787288817e-05,
"loss": 1.0069,
"step": 184000
},
{
"epoch": 2.38,
"learning_rate": 1.0352763702471668e-05,
"loss": 0.9947,
"step": 184500
},
{
"epoch": 2.39,
"learning_rate": 1.0245318617654517e-05,
"loss": 0.9987,
"step": 185000
},
{
"epoch": 2.39,
"learning_rate": 1.0137873532837368e-05,
"loss": 0.9895,
"step": 185500
},
{
"epoch": 2.4,
"learning_rate": 1.0030428448020217e-05,
"loss": 0.9888,
"step": 186000
},
{
"epoch": 2.4,
"learning_rate": 9.922983363203067e-06,
"loss": 0.9905,
"step": 186500
},
{
"epoch": 2.41,
"learning_rate": 9.815538278385918e-06,
"loss": 0.9872,
"step": 187000
},
{
"epoch": 2.42,
"learning_rate": 9.708093193568769e-06,
"loss": 0.9855,
"step": 187500
},
{
"epoch": 2.42,
"learning_rate": 9.600648108751618e-06,
"loss": 1.0014,
"step": 188000
},
{
"epoch": 2.43,
"learning_rate": 9.493203023934467e-06,
"loss": 0.9987,
"step": 188500
},
{
"epoch": 2.44,
"learning_rate": 9.385757939117317e-06,
"loss": 0.9986,
"step": 189000
},
{
"epoch": 2.44,
"learning_rate": 9.278312854300168e-06,
"loss": 1.0028,
"step": 189500
},
{
"epoch": 2.45,
"learning_rate": 9.170867769483017e-06,
"loss": 0.9882,
"step": 190000
},
{
"epoch": 2.46,
"learning_rate": 9.063422684665868e-06,
"loss": 0.983,
"step": 190500
},
{
"epoch": 2.46,
"learning_rate": 8.955977599848719e-06,
"loss": 0.984,
"step": 191000
},
{
"epoch": 2.47,
"learning_rate": 8.848532515031568e-06,
"loss": 1.0114,
"step": 191500
},
{
"epoch": 2.48,
"learning_rate": 8.741087430214417e-06,
"loss": 0.9921,
"step": 192000
},
{
"epoch": 2.48,
"learning_rate": 8.633642345397267e-06,
"loss": 0.9994,
"step": 192500
},
{
"epoch": 2.49,
"learning_rate": 8.526197260580118e-06,
"loss": 0.9947,
"step": 193000
},
{
"epoch": 2.49,
"learning_rate": 8.418752175762967e-06,
"loss": 0.9933,
"step": 193500
},
{
"epoch": 2.5,
"learning_rate": 8.311307090945818e-06,
"loss": 0.985,
"step": 194000
},
{
"epoch": 2.51,
"learning_rate": 8.203862006128668e-06,
"loss": 0.9863,
"step": 194500
},
{
"epoch": 2.51,
"learning_rate": 8.09641692131152e-06,
"loss": 0.9924,
"step": 195000
},
{
"epoch": 2.52,
"learning_rate": 7.988971836494368e-06,
"loss": 0.9814,
"step": 195500
},
{
"epoch": 2.53,
"learning_rate": 7.881526751677217e-06,
"loss": 0.9837,
"step": 196000
},
{
"epoch": 2.53,
"learning_rate": 7.774081666860068e-06,
"loss": 0.9849,
"step": 196500
},
{
"epoch": 2.54,
"learning_rate": 7.666636582042919e-06,
"loss": 1.0009,
"step": 197000
},
{
"epoch": 2.55,
"learning_rate": 7.559191497225768e-06,
"loss": 1.0074,
"step": 197500
},
{
"epoch": 2.55,
"learning_rate": 7.451746412408618e-06,
"loss": 0.9983,
"step": 198000
},
{
"epoch": 2.56,
"learning_rate": 7.344301327591468e-06,
"loss": 1.0009,
"step": 198500
},
{
"epoch": 2.57,
"learning_rate": 7.236856242774319e-06,
"loss": 0.9858,
"step": 199000
},
{
"epoch": 2.57,
"learning_rate": 7.129411157957168e-06,
"loss": 0.9908,
"step": 199500
},
{
"epoch": 2.58,
"learning_rate": 7.021966073140019e-06,
"loss": 0.973,
"step": 200000
},
{
"epoch": 2.59,
"learning_rate": 6.9145209883228686e-06,
"loss": 0.98,
"step": 200500
},
{
"epoch": 2.59,
"learning_rate": 6.807075903505719e-06,
"loss": 0.9783,
"step": 201000
},
{
"epoch": 2.6,
"learning_rate": 6.699630818688568e-06,
"loss": 0.9697,
"step": 201500
},
{
"epoch": 2.6,
"learning_rate": 6.592185733871418e-06,
"loss": 0.9787,
"step": 202000
},
{
"epoch": 2.61,
"learning_rate": 6.484740649054269e-06,
"loss": 0.9739,
"step": 202500
},
{
"epoch": 2.62,
"learning_rate": 6.37729556423712e-06,
"loss": 0.9966,
"step": 203000
},
{
"epoch": 2.62,
"learning_rate": 6.269850479419969e-06,
"loss": 0.9894,
"step": 203500
},
{
"epoch": 2.63,
"learning_rate": 6.1624053946028185e-06,
"loss": 0.965,
"step": 204000
},
{
"epoch": 2.64,
"learning_rate": 6.054960309785669e-06,
"loss": 0.9836,
"step": 204500
},
{
"epoch": 2.64,
"learning_rate": 5.947515224968519e-06,
"loss": 0.9728,
"step": 205000
},
{
"epoch": 2.65,
"learning_rate": 5.840070140151369e-06,
"loss": 0.977,
"step": 205500
},
{
"epoch": 2.66,
"learning_rate": 5.732625055334219e-06,
"loss": 0.9655,
"step": 206000
},
{
"epoch": 2.66,
"learning_rate": 5.625179970517069e-06,
"loss": 0.9802,
"step": 206500
},
{
"epoch": 2.67,
"learning_rate": 5.517734885699919e-06,
"loss": 0.9992,
"step": 207000
},
{
"epoch": 2.68,
"learning_rate": 5.410289800882768e-06,
"loss": 0.9975,
"step": 207500
},
{
"epoch": 2.68,
"learning_rate": 5.302844716065619e-06,
"loss": 0.9721,
"step": 208000
},
{
"epoch": 2.69,
"learning_rate": 5.195399631248469e-06,
"loss": 0.9695,
"step": 208500
},
{
"epoch": 2.69,
"learning_rate": 5.08795454643132e-06,
"loss": 0.9832,
"step": 209000
},
{
"epoch": 2.7,
"learning_rate": 4.980509461614169e-06,
"loss": 0.9873,
"step": 209500
},
{
"epoch": 2.71,
"learning_rate": 4.873064376797019e-06,
"loss": 0.9886,
"step": 210000
},
{
"epoch": 2.71,
"learning_rate": 4.765619291979869e-06,
"loss": 0.9753,
"step": 210500
},
{
"epoch": 2.72,
"learning_rate": 4.658174207162719e-06,
"loss": 0.9785,
"step": 211000
},
{
"epoch": 2.73,
"learning_rate": 4.550729122345569e-06,
"loss": 0.9704,
"step": 211500
},
{
"epoch": 2.73,
"learning_rate": 4.44328403752842e-06,
"loss": 0.9855,
"step": 212000
},
{
"epoch": 2.74,
"learning_rate": 4.3358389527112696e-06,
"loss": 0.9773,
"step": 212500
},
{
"epoch": 2.75,
"learning_rate": 4.2283938678941194e-06,
"loss": 0.9894,
"step": 213000
},
{
"epoch": 2.75,
"learning_rate": 4.120948783076969e-06,
"loss": 0.9701,
"step": 213500
},
{
"epoch": 2.76,
"learning_rate": 4.01350369825982e-06,
"loss": 0.9865,
"step": 214000
},
{
"epoch": 2.77,
"learning_rate": 3.90605861344267e-06,
"loss": 0.9934,
"step": 214500
},
{
"epoch": 2.77,
"learning_rate": 3.7986135286255197e-06,
"loss": 0.9765,
"step": 215000
},
{
"epoch": 2.78,
"learning_rate": 3.6911684438083696e-06,
"loss": 0.9633,
"step": 215500
},
{
"epoch": 2.78,
"learning_rate": 3.58372335899122e-06,
"loss": 0.9625,
"step": 216000
},
{
"epoch": 2.79,
"learning_rate": 3.4762782741740698e-06,
"loss": 0.9676,
"step": 216500
},
{
"epoch": 2.8,
"learning_rate": 3.36883318935692e-06,
"loss": 0.9642,
"step": 217000
},
{
"epoch": 2.8,
"learning_rate": 3.26138810453977e-06,
"loss": 0.9923,
"step": 217500
},
{
"epoch": 2.81,
"learning_rate": 3.15394301972262e-06,
"loss": 0.9641,
"step": 218000
},
{
"epoch": 2.82,
"learning_rate": 3.04649793490547e-06,
"loss": 0.9727,
"step": 218500
},
{
"epoch": 2.82,
"learning_rate": 2.93905285008832e-06,
"loss": 0.9868,
"step": 219000
},
{
"epoch": 2.83,
"learning_rate": 2.8316077652711702e-06,
"loss": 0.9504,
"step": 219500
},
{
"epoch": 2.84,
"learning_rate": 2.72416268045402e-06,
"loss": 0.9811,
"step": 220000
},
{
"epoch": 2.84,
"learning_rate": 2.6167175956368704e-06,
"loss": 0.9813,
"step": 220500
},
{
"epoch": 2.85,
"learning_rate": 2.5092725108197202e-06,
"loss": 0.9687,
"step": 221000
},
{
"epoch": 2.86,
"learning_rate": 2.40182742600257e-06,
"loss": 0.9673,
"step": 221500
},
{
"epoch": 2.86,
"learning_rate": 2.2943823411854204e-06,
"loss": 0.9653,
"step": 222000
},
{
"epoch": 2.87,
"learning_rate": 2.1869372563682703e-06,
"loss": 0.9479,
"step": 222500
},
{
"epoch": 2.88,
"learning_rate": 2.0794921715511205e-06,
"loss": 0.9802,
"step": 223000
},
{
"epoch": 2.88,
"learning_rate": 1.9720470867339704e-06,
"loss": 0.9624,
"step": 223500
},
{
"epoch": 2.89,
"learning_rate": 1.8646020019168205e-06,
"loss": 0.9822,
"step": 224000
},
{
"epoch": 2.89,
"learning_rate": 1.7571569170996706e-06,
"loss": 0.9747,
"step": 224500
},
{
"epoch": 2.9,
"learning_rate": 1.6497118322825206e-06,
"loss": 0.9614,
"step": 225000
},
{
"epoch": 2.91,
"learning_rate": 1.5422667474653705e-06,
"loss": 0.9729,
"step": 225500
},
{
"epoch": 2.91,
"learning_rate": 1.4348216626482206e-06,
"loss": 0.9679,
"step": 226000
},
{
"epoch": 2.92,
"learning_rate": 1.3273765778310707e-06,
"loss": 0.9551,
"step": 226500
},
{
"epoch": 2.93,
"learning_rate": 1.2199314930139207e-06,
"loss": 0.9762,
"step": 227000
},
{
"epoch": 2.93,
"learning_rate": 1.1124864081967708e-06,
"loss": 0.9737,
"step": 227500
},
{
"epoch": 2.94,
"learning_rate": 1.0050413233796209e-06,
"loss": 0.9718,
"step": 228000
},
{
"epoch": 2.95,
"learning_rate": 8.975962385624709e-07,
"loss": 0.9666,
"step": 228500
},
{
"epoch": 2.95,
"learning_rate": 7.901511537453207e-07,
"loss": 0.9854,
"step": 229000
},
{
"epoch": 2.96,
"learning_rate": 6.827060689281709e-07,
"loss": 0.9535,
"step": 229500
},
{
"epoch": 2.97,
"learning_rate": 5.752609841110209e-07,
"loss": 0.9665,
"step": 230000
},
{
"epoch": 2.97,
"learning_rate": 4.678158992938709e-07,
"loss": 0.9463,
"step": 230500
},
{
"epoch": 2.98,
"learning_rate": 3.60370814476721e-07,
"loss": 0.9664,
"step": 231000
},
{
"epoch": 2.98,
"learning_rate": 2.52925729659571e-07,
"loss": 0.9573,
"step": 231500
},
{
"epoch": 2.99,
"learning_rate": 1.4548064484242105e-07,
"loss": 0.95,
"step": 232000
},
{
"epoch": 3.0,
"learning_rate": 3.8035560025271085e-08,
"loss": 0.9636,
"step": 232500
},
{
"epoch": 3.0,
"step": 232677,
"total_flos": 3.713762606980301e+17,
"train_loss": 1.2468314008385242,
"train_runtime": 62368.4442,
"train_samples_per_second": 37.307,
"train_steps_per_second": 3.731
}
],
"max_steps": 232677,
"num_train_epochs": 3,
"total_flos": 3.713762606980301e+17,
"trial_name": null,
"trial_params": null
}