ctrl_discovery_3 / trainer_state.json
prajjwal1
v1
804e6aa
raw history blame
No virus
67.6 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 273072,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.990844905372942e-05,
"loss": 1.2008,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.9816898107458844e-05,
"loss": 0.904,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 4.972534716118826e-05,
"loss": 0.9083,
"step": 1500
},
{
"epoch": 0.02,
"learning_rate": 4.963379621491768e-05,
"loss": 0.9092,
"step": 2000
},
{
"epoch": 0.03,
"learning_rate": 4.9542245268647095e-05,
"loss": 0.8994,
"step": 2500
},
{
"epoch": 0.03,
"learning_rate": 4.945069432237652e-05,
"loss": 0.9007,
"step": 3000
},
{
"epoch": 0.04,
"learning_rate": 4.9359143376105936e-05,
"loss": 0.906,
"step": 3500
},
{
"epoch": 0.04,
"learning_rate": 4.926759242983536e-05,
"loss": 0.9026,
"step": 4000
},
{
"epoch": 0.05,
"learning_rate": 4.917604148356478e-05,
"loss": 0.8885,
"step": 4500
},
{
"epoch": 0.05,
"learning_rate": 4.9084490537294195e-05,
"loss": 0.8877,
"step": 5000
},
{
"epoch": 0.06,
"learning_rate": 4.899293959102362e-05,
"loss": 0.8886,
"step": 5500
},
{
"epoch": 0.07,
"learning_rate": 4.890138864475303e-05,
"loss": 0.9007,
"step": 6000
},
{
"epoch": 0.07,
"learning_rate": 4.880983769848245e-05,
"loss": 0.8963,
"step": 6500
},
{
"epoch": 0.08,
"learning_rate": 4.871828675221187e-05,
"loss": 0.8941,
"step": 7000
},
{
"epoch": 0.08,
"learning_rate": 4.8626735805941294e-05,
"loss": 0.8834,
"step": 7500
},
{
"epoch": 0.09,
"learning_rate": 4.853518485967071e-05,
"loss": 0.8852,
"step": 8000
},
{
"epoch": 0.09,
"learning_rate": 4.8443633913400135e-05,
"loss": 0.8888,
"step": 8500
},
{
"epoch": 0.1,
"learning_rate": 4.835208296712955e-05,
"loss": 0.8922,
"step": 9000
},
{
"epoch": 0.1,
"learning_rate": 4.826053202085897e-05,
"loss": 0.8927,
"step": 9500
},
{
"epoch": 0.11,
"learning_rate": 4.816898107458839e-05,
"loss": 0.8933,
"step": 10000
},
{
"epoch": 0.12,
"learning_rate": 4.807743012831781e-05,
"loss": 0.8863,
"step": 10500
},
{
"epoch": 0.12,
"learning_rate": 4.798587918204723e-05,
"loss": 0.8867,
"step": 11000
},
{
"epoch": 0.13,
"learning_rate": 4.7894328235776645e-05,
"loss": 0.8905,
"step": 11500
},
{
"epoch": 0.13,
"learning_rate": 4.780277728950607e-05,
"loss": 0.8957,
"step": 12000
},
{
"epoch": 0.14,
"learning_rate": 4.7711226343235486e-05,
"loss": 0.8859,
"step": 12500
},
{
"epoch": 0.14,
"learning_rate": 4.761967539696491e-05,
"loss": 0.8905,
"step": 13000
},
{
"epoch": 0.15,
"learning_rate": 4.752812445069432e-05,
"loss": 0.8861,
"step": 13500
},
{
"epoch": 0.15,
"learning_rate": 4.7436573504423745e-05,
"loss": 0.8919,
"step": 14000
},
{
"epoch": 0.16,
"learning_rate": 4.734502255815316e-05,
"loss": 0.8914,
"step": 14500
},
{
"epoch": 0.16,
"learning_rate": 4.7253471611882586e-05,
"loss": 0.8973,
"step": 15000
},
{
"epoch": 0.17,
"learning_rate": 4.7161920665612e-05,
"loss": 0.893,
"step": 15500
},
{
"epoch": 0.18,
"learning_rate": 4.707036971934142e-05,
"loss": 0.8924,
"step": 16000
},
{
"epoch": 0.18,
"learning_rate": 4.6978818773070844e-05,
"loss": 0.8805,
"step": 16500
},
{
"epoch": 0.19,
"learning_rate": 4.6887267826800255e-05,
"loss": 0.8978,
"step": 17000
},
{
"epoch": 0.19,
"learning_rate": 4.679571688052968e-05,
"loss": 0.89,
"step": 17500
},
{
"epoch": 0.2,
"learning_rate": 4.6704165934259096e-05,
"loss": 0.8874,
"step": 18000
},
{
"epoch": 0.2,
"learning_rate": 4.661261498798852e-05,
"loss": 0.8893,
"step": 18500
},
{
"epoch": 0.21,
"learning_rate": 4.652106404171794e-05,
"loss": 0.8872,
"step": 19000
},
{
"epoch": 0.21,
"learning_rate": 4.642951309544736e-05,
"loss": 0.8905,
"step": 19500
},
{
"epoch": 0.22,
"learning_rate": 4.633796214917678e-05,
"loss": 0.8919,
"step": 20000
},
{
"epoch": 0.23,
"learning_rate": 4.6246411202906195e-05,
"loss": 0.8846,
"step": 20500
},
{
"epoch": 0.23,
"learning_rate": 4.615486025663561e-05,
"loss": 0.8914,
"step": 21000
},
{
"epoch": 0.24,
"learning_rate": 4.606330931036503e-05,
"loss": 0.8863,
"step": 21500
},
{
"epoch": 0.24,
"learning_rate": 4.5971758364094454e-05,
"loss": 0.8921,
"step": 22000
},
{
"epoch": 0.25,
"learning_rate": 4.588020741782387e-05,
"loss": 0.8903,
"step": 22500
},
{
"epoch": 0.25,
"learning_rate": 4.5788656471553295e-05,
"loss": 0.8882,
"step": 23000
},
{
"epoch": 0.26,
"learning_rate": 4.569710552528271e-05,
"loss": 0.8945,
"step": 23500
},
{
"epoch": 0.26,
"learning_rate": 4.5605554579012136e-05,
"loss": 0.8944,
"step": 24000
},
{
"epoch": 0.27,
"learning_rate": 4.5514003632741546e-05,
"loss": 0.8784,
"step": 24500
},
{
"epoch": 0.27,
"learning_rate": 4.542245268647097e-05,
"loss": 0.8843,
"step": 25000
},
{
"epoch": 0.28,
"learning_rate": 4.533090174020039e-05,
"loss": 0.8882,
"step": 25500
},
{
"epoch": 0.29,
"learning_rate": 4.5239350793929805e-05,
"loss": 0.8891,
"step": 26000
},
{
"epoch": 0.29,
"learning_rate": 4.514779984765923e-05,
"loss": 0.8896,
"step": 26500
},
{
"epoch": 0.3,
"learning_rate": 4.5056248901388646e-05,
"loss": 0.8924,
"step": 27000
},
{
"epoch": 0.3,
"learning_rate": 4.496469795511807e-05,
"loss": 0.8812,
"step": 27500
},
{
"epoch": 0.31,
"learning_rate": 4.487314700884749e-05,
"loss": 0.8935,
"step": 28000
},
{
"epoch": 0.31,
"learning_rate": 4.4781596062576904e-05,
"loss": 0.8907,
"step": 28500
},
{
"epoch": 0.32,
"learning_rate": 4.469004511630632e-05,
"loss": 0.8887,
"step": 29000
},
{
"epoch": 0.32,
"learning_rate": 4.4598494170035745e-05,
"loss": 0.8898,
"step": 29500
},
{
"epoch": 0.33,
"learning_rate": 4.450694322376516e-05,
"loss": 0.8846,
"step": 30000
},
{
"epoch": 0.34,
"learning_rate": 4.441539227749458e-05,
"loss": 0.8858,
"step": 30500
},
{
"epoch": 0.34,
"learning_rate": 4.4323841331224004e-05,
"loss": 0.8933,
"step": 31000
},
{
"epoch": 0.35,
"learning_rate": 4.423229038495342e-05,
"loss": 0.8854,
"step": 31500
},
{
"epoch": 0.35,
"learning_rate": 4.414073943868284e-05,
"loss": 0.8931,
"step": 32000
},
{
"epoch": 0.36,
"learning_rate": 4.4049188492412255e-05,
"loss": 0.8904,
"step": 32500
},
{
"epoch": 0.36,
"learning_rate": 4.395763754614168e-05,
"loss": 0.884,
"step": 33000
},
{
"epoch": 0.37,
"learning_rate": 4.3866086599871097e-05,
"loss": 0.8963,
"step": 33500
},
{
"epoch": 0.37,
"learning_rate": 4.377453565360052e-05,
"loss": 0.89,
"step": 34000
},
{
"epoch": 0.38,
"learning_rate": 4.368298470732994e-05,
"loss": 0.8852,
"step": 34500
},
{
"epoch": 0.38,
"learning_rate": 4.3591433761059355e-05,
"loss": 0.8831,
"step": 35000
},
{
"epoch": 0.39,
"learning_rate": 4.349988281478878e-05,
"loss": 0.895,
"step": 35500
},
{
"epoch": 0.4,
"learning_rate": 4.3408331868518196e-05,
"loss": 0.8907,
"step": 36000
},
{
"epoch": 0.4,
"learning_rate": 4.331678092224761e-05,
"loss": 0.8956,
"step": 36500
},
{
"epoch": 0.41,
"learning_rate": 4.322522997597703e-05,
"loss": 0.8895,
"step": 37000
},
{
"epoch": 0.41,
"learning_rate": 4.3133679029706454e-05,
"loss": 0.8941,
"step": 37500
},
{
"epoch": 0.42,
"learning_rate": 4.304212808343587e-05,
"loss": 0.8919,
"step": 38000
},
{
"epoch": 0.42,
"learning_rate": 4.2950577137165296e-05,
"loss": 0.8986,
"step": 38500
},
{
"epoch": 0.43,
"learning_rate": 4.285902619089471e-05,
"loss": 0.8929,
"step": 39000
},
{
"epoch": 0.43,
"learning_rate": 4.276747524462413e-05,
"loss": 0.8906,
"step": 39500
},
{
"epoch": 0.44,
"learning_rate": 4.267592429835355e-05,
"loss": 0.8919,
"step": 40000
},
{
"epoch": 0.44,
"learning_rate": 4.258437335208297e-05,
"loss": 0.8983,
"step": 40500
},
{
"epoch": 0.45,
"learning_rate": 4.249282240581239e-05,
"loss": 0.8801,
"step": 41000
},
{
"epoch": 0.46,
"learning_rate": 4.2401271459541805e-05,
"loss": 0.8857,
"step": 41500
},
{
"epoch": 0.46,
"learning_rate": 4.230972051327123e-05,
"loss": 0.8778,
"step": 42000
},
{
"epoch": 0.47,
"learning_rate": 4.221816956700065e-05,
"loss": 0.8971,
"step": 42500
},
{
"epoch": 0.47,
"learning_rate": 4.212661862073007e-05,
"loss": 0.8845,
"step": 43000
},
{
"epoch": 0.48,
"learning_rate": 4.203506767445948e-05,
"loss": 0.8927,
"step": 43500
},
{
"epoch": 0.48,
"learning_rate": 4.1943516728188905e-05,
"loss": 0.8932,
"step": 44000
},
{
"epoch": 0.49,
"learning_rate": 4.185196578191832e-05,
"loss": 0.8867,
"step": 44500
},
{
"epoch": 0.49,
"learning_rate": 4.1760414835647746e-05,
"loss": 0.8911,
"step": 45000
},
{
"epoch": 0.5,
"learning_rate": 4.166886388937716e-05,
"loss": 0.8915,
"step": 45500
},
{
"epoch": 0.51,
"learning_rate": 4.157731294310658e-05,
"loss": 0.8925,
"step": 46000
},
{
"epoch": 0.51,
"learning_rate": 4.1485761996836004e-05,
"loss": 0.8828,
"step": 46500
},
{
"epoch": 0.52,
"learning_rate": 4.1394211050565415e-05,
"loss": 0.8966,
"step": 47000
},
{
"epoch": 0.52,
"learning_rate": 4.130266010429484e-05,
"loss": 0.8939,
"step": 47500
},
{
"epoch": 0.53,
"learning_rate": 4.1211109158024256e-05,
"loss": 0.8884,
"step": 48000
},
{
"epoch": 0.53,
"learning_rate": 4.111955821175368e-05,
"loss": 0.8925,
"step": 48500
},
{
"epoch": 0.54,
"learning_rate": 4.10280072654831e-05,
"loss": 0.8884,
"step": 49000
},
{
"epoch": 0.54,
"learning_rate": 4.093645631921252e-05,
"loss": 0.8912,
"step": 49500
},
{
"epoch": 0.55,
"learning_rate": 4.084490537294194e-05,
"loss": 0.8882,
"step": 50000
},
{
"epoch": 0.55,
"learning_rate": 4.0753354426671356e-05,
"loss": 0.8879,
"step": 50500
},
{
"epoch": 0.56,
"learning_rate": 4.066180348040077e-05,
"loss": 0.8872,
"step": 51000
},
{
"epoch": 0.57,
"learning_rate": 4.057025253413019e-05,
"loss": 0.9009,
"step": 51500
},
{
"epoch": 0.57,
"learning_rate": 4.0478701587859614e-05,
"loss": 0.8885,
"step": 52000
},
{
"epoch": 0.58,
"learning_rate": 4.038715064158903e-05,
"loss": 0.8896,
"step": 52500
},
{
"epoch": 0.58,
"learning_rate": 4.0295599695318455e-05,
"loss": 0.8895,
"step": 53000
},
{
"epoch": 0.59,
"learning_rate": 4.020404874904787e-05,
"loss": 0.8945,
"step": 53500
},
{
"epoch": 0.59,
"learning_rate": 4.0112497802777296e-05,
"loss": 0.8949,
"step": 54000
},
{
"epoch": 0.6,
"learning_rate": 4.002094685650671e-05,
"loss": 0.8928,
"step": 54500
},
{
"epoch": 0.6,
"learning_rate": 3.992939591023613e-05,
"loss": 0.9009,
"step": 55000
},
{
"epoch": 0.61,
"learning_rate": 3.983784496396555e-05,
"loss": 0.8918,
"step": 55500
},
{
"epoch": 0.62,
"learning_rate": 3.9746294017694965e-05,
"loss": 0.8809,
"step": 56000
},
{
"epoch": 0.62,
"learning_rate": 3.965474307142439e-05,
"loss": 0.8929,
"step": 56500
},
{
"epoch": 0.63,
"learning_rate": 3.9563192125153806e-05,
"loss": 0.8911,
"step": 57000
},
{
"epoch": 0.63,
"learning_rate": 3.947164117888323e-05,
"loss": 0.8933,
"step": 57500
},
{
"epoch": 0.64,
"learning_rate": 3.938009023261265e-05,
"loss": 0.8925,
"step": 58000
},
{
"epoch": 0.64,
"learning_rate": 3.9288539286342065e-05,
"loss": 0.8909,
"step": 58500
},
{
"epoch": 0.65,
"learning_rate": 3.919698834007148e-05,
"loss": 0.893,
"step": 59000
},
{
"epoch": 0.65,
"learning_rate": 3.9105437393800906e-05,
"loss": 0.8879,
"step": 59500
},
{
"epoch": 0.66,
"learning_rate": 3.901388644753032e-05,
"loss": 0.8947,
"step": 60000
},
{
"epoch": 0.66,
"learning_rate": 3.892233550125975e-05,
"loss": 0.8921,
"step": 60500
},
{
"epoch": 0.67,
"learning_rate": 3.8830784554989164e-05,
"loss": 0.8891,
"step": 61000
},
{
"epoch": 0.68,
"learning_rate": 3.873923360871858e-05,
"loss": 0.8894,
"step": 61500
},
{
"epoch": 0.68,
"learning_rate": 3.8647682662448e-05,
"loss": 0.8917,
"step": 62000
},
{
"epoch": 0.69,
"learning_rate": 3.8556131716177416e-05,
"loss": 0.8839,
"step": 62500
},
{
"epoch": 0.69,
"learning_rate": 3.846458076990684e-05,
"loss": 0.8944,
"step": 63000
},
{
"epoch": 0.7,
"learning_rate": 3.837302982363626e-05,
"loss": 0.8929,
"step": 63500
},
{
"epoch": 0.7,
"learning_rate": 3.828147887736568e-05,
"loss": 0.8811,
"step": 64000
},
{
"epoch": 0.71,
"learning_rate": 3.81899279310951e-05,
"loss": 0.89,
"step": 64500
},
{
"epoch": 0.71,
"learning_rate": 3.809837698482452e-05,
"loss": 0.8904,
"step": 65000
},
{
"epoch": 0.72,
"learning_rate": 3.800682603855394e-05,
"loss": 0.8901,
"step": 65500
},
{
"epoch": 0.73,
"learning_rate": 3.7915275092283356e-05,
"loss": 0.8976,
"step": 66000
},
{
"epoch": 0.73,
"learning_rate": 3.7823724146012773e-05,
"loss": 0.892,
"step": 66500
},
{
"epoch": 0.74,
"learning_rate": 3.773217319974219e-05,
"loss": 0.8906,
"step": 67000
},
{
"epoch": 0.74,
"learning_rate": 3.7640622253471615e-05,
"loss": 0.8846,
"step": 67500
},
{
"epoch": 0.75,
"learning_rate": 3.754907130720103e-05,
"loss": 0.8861,
"step": 68000
},
{
"epoch": 0.75,
"learning_rate": 3.7457520360930456e-05,
"loss": 0.8873,
"step": 68500
},
{
"epoch": 0.76,
"learning_rate": 3.736596941465987e-05,
"loss": 0.8897,
"step": 69000
},
{
"epoch": 0.76,
"learning_rate": 3.727441846838929e-05,
"loss": 0.8891,
"step": 69500
},
{
"epoch": 0.77,
"learning_rate": 3.718286752211871e-05,
"loss": 0.8875,
"step": 70000
},
{
"epoch": 0.77,
"learning_rate": 3.709131657584813e-05,
"loss": 0.903,
"step": 70500
},
{
"epoch": 0.78,
"learning_rate": 3.699976562957755e-05,
"loss": 0.9041,
"step": 71000
},
{
"epoch": 0.79,
"learning_rate": 3.6908214683306966e-05,
"loss": 0.9015,
"step": 71500
},
{
"epoch": 0.79,
"learning_rate": 3.681666373703639e-05,
"loss": 0.8953,
"step": 72000
},
{
"epoch": 0.8,
"learning_rate": 3.672511279076581e-05,
"loss": 0.8998,
"step": 72500
},
{
"epoch": 0.8,
"learning_rate": 3.663356184449523e-05,
"loss": 0.8869,
"step": 73000
},
{
"epoch": 0.81,
"learning_rate": 3.654201089822464e-05,
"loss": 0.8871,
"step": 73500
},
{
"epoch": 0.81,
"learning_rate": 3.6450459951954065e-05,
"loss": 0.898,
"step": 74000
},
{
"epoch": 0.82,
"learning_rate": 3.635890900568348e-05,
"loss": 0.8971,
"step": 74500
},
{
"epoch": 0.82,
"learning_rate": 3.6267358059412906e-05,
"loss": 0.8986,
"step": 75000
},
{
"epoch": 0.83,
"learning_rate": 3.6175807113142324e-05,
"loss": 0.8942,
"step": 75500
},
{
"epoch": 0.83,
"learning_rate": 3.608425616687174e-05,
"loss": 0.8951,
"step": 76000
},
{
"epoch": 0.84,
"learning_rate": 3.5992705220601165e-05,
"loss": 0.8935,
"step": 76500
},
{
"epoch": 0.85,
"learning_rate": 3.5901154274330575e-05,
"loss": 0.8812,
"step": 77000
},
{
"epoch": 0.85,
"learning_rate": 3.580960332806e-05,
"loss": 0.8954,
"step": 77500
},
{
"epoch": 0.86,
"learning_rate": 3.5718052381789416e-05,
"loss": 0.8942,
"step": 78000
},
{
"epoch": 0.86,
"learning_rate": 3.562650143551884e-05,
"loss": 0.8932,
"step": 78500
},
{
"epoch": 0.87,
"learning_rate": 3.553495048924826e-05,
"loss": 0.8915,
"step": 79000
},
{
"epoch": 0.87,
"learning_rate": 3.544339954297768e-05,
"loss": 0.8794,
"step": 79500
},
{
"epoch": 0.88,
"learning_rate": 3.53518485967071e-05,
"loss": 0.8998,
"step": 80000
},
{
"epoch": 0.88,
"learning_rate": 3.5260297650436516e-05,
"loss": 0.8926,
"step": 80500
},
{
"epoch": 0.89,
"learning_rate": 3.516874670416593e-05,
"loss": 0.8897,
"step": 81000
},
{
"epoch": 0.9,
"learning_rate": 3.507719575789536e-05,
"loss": 0.8856,
"step": 81500
},
{
"epoch": 0.9,
"learning_rate": 3.4985644811624774e-05,
"loss": 0.8958,
"step": 82000
},
{
"epoch": 0.91,
"learning_rate": 3.489409386535419e-05,
"loss": 0.8901,
"step": 82500
},
{
"epoch": 0.91,
"learning_rate": 3.4802542919083615e-05,
"loss": 0.8843,
"step": 83000
},
{
"epoch": 0.92,
"learning_rate": 3.471099197281303e-05,
"loss": 0.8991,
"step": 83500
},
{
"epoch": 0.92,
"learning_rate": 3.4619441026542456e-05,
"loss": 0.8977,
"step": 84000
},
{
"epoch": 0.93,
"learning_rate": 3.4527890080271874e-05,
"loss": 0.8917,
"step": 84500
},
{
"epoch": 0.93,
"learning_rate": 3.443633913400129e-05,
"loss": 0.9006,
"step": 85000
},
{
"epoch": 0.94,
"learning_rate": 3.434478818773071e-05,
"loss": 0.8889,
"step": 85500
},
{
"epoch": 0.94,
"learning_rate": 3.425323724146013e-05,
"loss": 0.8955,
"step": 86000
},
{
"epoch": 0.95,
"learning_rate": 3.416168629518955e-05,
"loss": 0.898,
"step": 86500
},
{
"epoch": 0.96,
"learning_rate": 3.4070135348918966e-05,
"loss": 0.8845,
"step": 87000
},
{
"epoch": 0.96,
"learning_rate": 3.397858440264839e-05,
"loss": 0.8898,
"step": 87500
},
{
"epoch": 0.97,
"learning_rate": 3.388703345637781e-05,
"loss": 0.8933,
"step": 88000
},
{
"epoch": 0.97,
"learning_rate": 3.3795482510107225e-05,
"loss": 0.8903,
"step": 88500
},
{
"epoch": 0.98,
"learning_rate": 3.370393156383664e-05,
"loss": 0.8931,
"step": 89000
},
{
"epoch": 0.98,
"learning_rate": 3.3612380617566066e-05,
"loss": 0.9005,
"step": 89500
},
{
"epoch": 0.99,
"learning_rate": 3.352082967129548e-05,
"loss": 0.8839,
"step": 90000
},
{
"epoch": 0.99,
"learning_rate": 3.342927872502491e-05,
"loss": 0.898,
"step": 90500
},
{
"epoch": 1.0,
"learning_rate": 3.3337727778754324e-05,
"loss": 0.8996,
"step": 91000
},
{
"epoch": 1.01,
"learning_rate": 3.324617683248374e-05,
"loss": 0.7676,
"step": 91500
},
{
"epoch": 1.01,
"learning_rate": 3.3154625886213165e-05,
"loss": 0.7695,
"step": 92000
},
{
"epoch": 1.02,
"learning_rate": 3.3063074939942576e-05,
"loss": 0.7604,
"step": 92500
},
{
"epoch": 1.02,
"learning_rate": 3.2971523993672e-05,
"loss": 0.761,
"step": 93000
},
{
"epoch": 1.03,
"learning_rate": 3.287997304740142e-05,
"loss": 0.7576,
"step": 93500
},
{
"epoch": 1.03,
"learning_rate": 3.278842210113084e-05,
"loss": 0.7618,
"step": 94000
},
{
"epoch": 1.04,
"learning_rate": 3.269687115486026e-05,
"loss": 0.7585,
"step": 94500
},
{
"epoch": 1.04,
"learning_rate": 3.260532020858968e-05,
"loss": 0.7636,
"step": 95000
},
{
"epoch": 1.05,
"learning_rate": 3.25137692623191e-05,
"loss": 0.7609,
"step": 95500
},
{
"epoch": 1.05,
"learning_rate": 3.2422218316048517e-05,
"loss": 0.7605,
"step": 96000
},
{
"epoch": 1.06,
"learning_rate": 3.2330667369777934e-05,
"loss": 0.7686,
"step": 96500
},
{
"epoch": 1.07,
"learning_rate": 3.223911642350735e-05,
"loss": 0.7633,
"step": 97000
},
{
"epoch": 1.07,
"learning_rate": 3.2147565477236775e-05,
"loss": 0.7644,
"step": 97500
},
{
"epoch": 1.08,
"learning_rate": 3.205601453096619e-05,
"loss": 0.7687,
"step": 98000
},
{
"epoch": 1.08,
"learning_rate": 3.1964463584695616e-05,
"loss": 0.767,
"step": 98500
},
{
"epoch": 1.09,
"learning_rate": 3.187291263842503e-05,
"loss": 0.7663,
"step": 99000
},
{
"epoch": 1.09,
"learning_rate": 3.178136169215446e-05,
"loss": 0.7657,
"step": 99500
},
{
"epoch": 1.1,
"learning_rate": 3.168981074588387e-05,
"loss": 0.7682,
"step": 100000
},
{
"epoch": 1.1,
"learning_rate": 3.159825979961329e-05,
"loss": 0.7721,
"step": 100500
},
{
"epoch": 1.11,
"learning_rate": 3.150670885334271e-05,
"loss": 0.7673,
"step": 101000
},
{
"epoch": 1.12,
"learning_rate": 3.1415157907072126e-05,
"loss": 0.7649,
"step": 101500
},
{
"epoch": 1.12,
"learning_rate": 3.132360696080155e-05,
"loss": 0.7737,
"step": 102000
},
{
"epoch": 1.13,
"learning_rate": 3.123205601453097e-05,
"loss": 0.7688,
"step": 102500
},
{
"epoch": 1.13,
"learning_rate": 3.114050506826039e-05,
"loss": 0.7705,
"step": 103000
},
{
"epoch": 1.14,
"learning_rate": 3.10489541219898e-05,
"loss": 0.764,
"step": 103500
},
{
"epoch": 1.14,
"learning_rate": 3.0957403175719225e-05,
"loss": 0.7654,
"step": 104000
},
{
"epoch": 1.15,
"learning_rate": 3.086585222944864e-05,
"loss": 0.7762,
"step": 104500
},
{
"epoch": 1.15,
"learning_rate": 3.0774301283178067e-05,
"loss": 0.7708,
"step": 105000
},
{
"epoch": 1.16,
"learning_rate": 3.0682750336907484e-05,
"loss": 0.778,
"step": 105500
},
{
"epoch": 1.16,
"learning_rate": 3.05911993906369e-05,
"loss": 0.7741,
"step": 106000
},
{
"epoch": 1.17,
"learning_rate": 3.0499648444366325e-05,
"loss": 0.7714,
"step": 106500
},
{
"epoch": 1.18,
"learning_rate": 3.0408097498095746e-05,
"loss": 0.774,
"step": 107000
},
{
"epoch": 1.18,
"learning_rate": 3.031654655182516e-05,
"loss": 0.7704,
"step": 107500
},
{
"epoch": 1.19,
"learning_rate": 3.022499560555458e-05,
"loss": 0.7797,
"step": 108000
},
{
"epoch": 1.19,
"learning_rate": 3.0133444659284e-05,
"loss": 0.7647,
"step": 108500
},
{
"epoch": 1.2,
"learning_rate": 3.0041893713013418e-05,
"loss": 0.774,
"step": 109000
},
{
"epoch": 1.2,
"learning_rate": 2.9950342766742838e-05,
"loss": 0.7714,
"step": 109500
},
{
"epoch": 1.21,
"learning_rate": 2.985879182047226e-05,
"loss": 0.7638,
"step": 110000
},
{
"epoch": 1.21,
"learning_rate": 2.976724087420168e-05,
"loss": 0.7724,
"step": 110500
},
{
"epoch": 1.22,
"learning_rate": 2.9675689927931093e-05,
"loss": 0.7723,
"step": 111000
},
{
"epoch": 1.22,
"learning_rate": 2.9584138981660514e-05,
"loss": 0.767,
"step": 111500
},
{
"epoch": 1.23,
"learning_rate": 2.9492588035389934e-05,
"loss": 0.7745,
"step": 112000
},
{
"epoch": 1.24,
"learning_rate": 2.9401037089119355e-05,
"loss": 0.7657,
"step": 112500
},
{
"epoch": 1.24,
"learning_rate": 2.9309486142848776e-05,
"loss": 0.7706,
"step": 113000
},
{
"epoch": 1.25,
"learning_rate": 2.9217935196578196e-05,
"loss": 0.7767,
"step": 113500
},
{
"epoch": 1.25,
"learning_rate": 2.9126384250307613e-05,
"loss": 0.7777,
"step": 114000
},
{
"epoch": 1.26,
"learning_rate": 2.9034833304037034e-05,
"loss": 0.7705,
"step": 114500
},
{
"epoch": 1.26,
"learning_rate": 2.8943282357766448e-05,
"loss": 0.7829,
"step": 115000
},
{
"epoch": 1.27,
"learning_rate": 2.8851731411495868e-05,
"loss": 0.775,
"step": 115500
},
{
"epoch": 1.27,
"learning_rate": 2.876018046522529e-05,
"loss": 0.7815,
"step": 116000
},
{
"epoch": 1.28,
"learning_rate": 2.866862951895471e-05,
"loss": 0.7703,
"step": 116500
},
{
"epoch": 1.29,
"learning_rate": 2.857707857268413e-05,
"loss": 0.7736,
"step": 117000
},
{
"epoch": 1.29,
"learning_rate": 2.848552762641355e-05,
"loss": 0.7747,
"step": 117500
},
{
"epoch": 1.3,
"learning_rate": 2.839397668014297e-05,
"loss": 0.7707,
"step": 118000
},
{
"epoch": 1.3,
"learning_rate": 2.8302425733872385e-05,
"loss": 0.7666,
"step": 118500
},
{
"epoch": 1.31,
"learning_rate": 2.8210874787601806e-05,
"loss": 0.7867,
"step": 119000
},
{
"epoch": 1.31,
"learning_rate": 2.8119323841331223e-05,
"loss": 0.7759,
"step": 119500
},
{
"epoch": 1.32,
"learning_rate": 2.8027772895060643e-05,
"loss": 0.7743,
"step": 120000
},
{
"epoch": 1.32,
"learning_rate": 2.7936221948790064e-05,
"loss": 0.7708,
"step": 120500
},
{
"epoch": 1.33,
"learning_rate": 2.7844671002519484e-05,
"loss": 0.7725,
"step": 121000
},
{
"epoch": 1.33,
"learning_rate": 2.7753120056248905e-05,
"loss": 0.7687,
"step": 121500
},
{
"epoch": 1.34,
"learning_rate": 2.7661569109978326e-05,
"loss": 0.7779,
"step": 122000
},
{
"epoch": 1.35,
"learning_rate": 2.757001816370774e-05,
"loss": 0.774,
"step": 122500
},
{
"epoch": 1.35,
"learning_rate": 2.747846721743716e-05,
"loss": 0.7671,
"step": 123000
},
{
"epoch": 1.36,
"learning_rate": 2.738691627116658e-05,
"loss": 0.7721,
"step": 123500
},
{
"epoch": 1.36,
"learning_rate": 2.7295365324896e-05,
"loss": 0.7758,
"step": 124000
},
{
"epoch": 1.37,
"learning_rate": 2.720381437862542e-05,
"loss": 0.7689,
"step": 124500
},
{
"epoch": 1.37,
"learning_rate": 2.711226343235484e-05,
"loss": 0.773,
"step": 125000
},
{
"epoch": 1.38,
"learning_rate": 2.702071248608426e-05,
"loss": 0.7776,
"step": 125500
},
{
"epoch": 1.38,
"learning_rate": 2.6929161539813673e-05,
"loss": 0.7724,
"step": 126000
},
{
"epoch": 1.39,
"learning_rate": 2.6837610593543094e-05,
"loss": 0.7885,
"step": 126500
},
{
"epoch": 1.4,
"learning_rate": 2.6746059647272515e-05,
"loss": 0.7737,
"step": 127000
},
{
"epoch": 1.4,
"learning_rate": 2.6654508701001935e-05,
"loss": 0.7715,
"step": 127500
},
{
"epoch": 1.41,
"learning_rate": 2.6562957754731356e-05,
"loss": 0.7709,
"step": 128000
},
{
"epoch": 1.41,
"learning_rate": 2.6471406808460776e-05,
"loss": 0.7732,
"step": 128500
},
{
"epoch": 1.42,
"learning_rate": 2.6379855862190193e-05,
"loss": 0.7802,
"step": 129000
},
{
"epoch": 1.42,
"learning_rate": 2.6288304915919614e-05,
"loss": 0.7736,
"step": 129500
},
{
"epoch": 1.43,
"learning_rate": 2.6196753969649028e-05,
"loss": 0.782,
"step": 130000
},
{
"epoch": 1.43,
"learning_rate": 2.610520302337845e-05,
"loss": 0.7728,
"step": 130500
},
{
"epoch": 1.44,
"learning_rate": 2.601365207710787e-05,
"loss": 0.7773,
"step": 131000
},
{
"epoch": 1.44,
"learning_rate": 2.592210113083729e-05,
"loss": 0.7658,
"step": 131500
},
{
"epoch": 1.45,
"learning_rate": 2.583055018456671e-05,
"loss": 0.7782,
"step": 132000
},
{
"epoch": 1.46,
"learning_rate": 2.573899923829613e-05,
"loss": 0.7794,
"step": 132500
},
{
"epoch": 1.46,
"learning_rate": 2.564744829202555e-05,
"loss": 0.7806,
"step": 133000
},
{
"epoch": 1.47,
"learning_rate": 2.5555897345754965e-05,
"loss": 0.7708,
"step": 133500
},
{
"epoch": 1.47,
"learning_rate": 2.5464346399484386e-05,
"loss": 0.7712,
"step": 134000
},
{
"epoch": 1.48,
"learning_rate": 2.5372795453213806e-05,
"loss": 0.781,
"step": 134500
},
{
"epoch": 1.48,
"learning_rate": 2.5281244506943223e-05,
"loss": 0.7807,
"step": 135000
},
{
"epoch": 1.49,
"learning_rate": 2.5189693560672644e-05,
"loss": 0.7862,
"step": 135500
},
{
"epoch": 1.49,
"learning_rate": 2.5098142614402065e-05,
"loss": 0.7796,
"step": 136000
},
{
"epoch": 1.5,
"learning_rate": 2.5006591668131485e-05,
"loss": 0.7842,
"step": 136500
},
{
"epoch": 1.51,
"learning_rate": 2.4915040721860902e-05,
"loss": 0.782,
"step": 137000
},
{
"epoch": 1.51,
"learning_rate": 2.4823489775590323e-05,
"loss": 0.7808,
"step": 137500
},
{
"epoch": 1.52,
"learning_rate": 2.4731938829319744e-05,
"loss": 0.7748,
"step": 138000
},
{
"epoch": 1.52,
"learning_rate": 2.464038788304916e-05,
"loss": 0.7783,
"step": 138500
},
{
"epoch": 1.53,
"learning_rate": 2.454883693677858e-05,
"loss": 0.7832,
"step": 139000
},
{
"epoch": 1.53,
"learning_rate": 2.4457285990508e-05,
"loss": 0.7795,
"step": 139500
},
{
"epoch": 1.54,
"learning_rate": 2.4365735044237416e-05,
"loss": 0.7836,
"step": 140000
},
{
"epoch": 1.54,
"learning_rate": 2.4274184097966836e-05,
"loss": 0.7779,
"step": 140500
},
{
"epoch": 1.55,
"learning_rate": 2.4182633151696257e-05,
"loss": 0.7813,
"step": 141000
},
{
"epoch": 1.55,
"learning_rate": 2.4091082205425677e-05,
"loss": 0.7818,
"step": 141500
},
{
"epoch": 1.56,
"learning_rate": 2.3999531259155095e-05,
"loss": 0.7813,
"step": 142000
},
{
"epoch": 1.57,
"learning_rate": 2.3907980312884515e-05,
"loss": 0.7787,
"step": 142500
},
{
"epoch": 1.57,
"learning_rate": 2.3816429366613936e-05,
"loss": 0.784,
"step": 143000
},
{
"epoch": 1.58,
"learning_rate": 2.3724878420343356e-05,
"loss": 0.7923,
"step": 143500
},
{
"epoch": 1.58,
"learning_rate": 2.3633327474072774e-05,
"loss": 0.7752,
"step": 144000
},
{
"epoch": 1.59,
"learning_rate": 2.354177652780219e-05,
"loss": 0.7755,
"step": 144500
},
{
"epoch": 1.59,
"learning_rate": 2.345022558153161e-05,
"loss": 0.7859,
"step": 145000
},
{
"epoch": 1.6,
"learning_rate": 2.3358674635261032e-05,
"loss": 0.7763,
"step": 145500
},
{
"epoch": 1.6,
"learning_rate": 2.326712368899045e-05,
"loss": 0.7812,
"step": 146000
},
{
"epoch": 1.61,
"learning_rate": 2.317557274271987e-05,
"loss": 0.7764,
"step": 146500
},
{
"epoch": 1.61,
"learning_rate": 2.308402179644929e-05,
"loss": 0.7743,
"step": 147000
},
{
"epoch": 1.62,
"learning_rate": 2.2992470850178707e-05,
"loss": 0.7861,
"step": 147500
},
{
"epoch": 1.63,
"learning_rate": 2.2900919903908128e-05,
"loss": 0.7746,
"step": 148000
},
{
"epoch": 1.63,
"learning_rate": 2.280936895763755e-05,
"loss": 0.7802,
"step": 148500
},
{
"epoch": 1.64,
"learning_rate": 2.271781801136697e-05,
"loss": 0.7859,
"step": 149000
},
{
"epoch": 1.64,
"learning_rate": 2.2626267065096386e-05,
"loss": 0.7774,
"step": 149500
},
{
"epoch": 1.65,
"learning_rate": 2.2534716118825804e-05,
"loss": 0.779,
"step": 150000
},
{
"epoch": 1.65,
"learning_rate": 2.2443165172555224e-05,
"loss": 0.7815,
"step": 150500
},
{
"epoch": 1.66,
"learning_rate": 2.2351614226284645e-05,
"loss": 0.7847,
"step": 151000
},
{
"epoch": 1.66,
"learning_rate": 2.2260063280014062e-05,
"loss": 0.7717,
"step": 151500
},
{
"epoch": 1.67,
"learning_rate": 2.2168512333743483e-05,
"loss": 0.7701,
"step": 152000
},
{
"epoch": 1.68,
"learning_rate": 2.2076961387472903e-05,
"loss": 0.7807,
"step": 152500
},
{
"epoch": 1.68,
"learning_rate": 2.1985410441202324e-05,
"loss": 0.7786,
"step": 153000
},
{
"epoch": 1.69,
"learning_rate": 2.189385949493174e-05,
"loss": 0.7894,
"step": 153500
},
{
"epoch": 1.69,
"learning_rate": 2.180230854866116e-05,
"loss": 0.7766,
"step": 154000
},
{
"epoch": 1.7,
"learning_rate": 2.171075760239058e-05,
"loss": 0.7792,
"step": 154500
},
{
"epoch": 1.7,
"learning_rate": 2.1619206656119996e-05,
"loss": 0.7785,
"step": 155000
},
{
"epoch": 1.71,
"learning_rate": 2.1527655709849416e-05,
"loss": 0.7795,
"step": 155500
},
{
"epoch": 1.71,
"learning_rate": 2.1436104763578837e-05,
"loss": 0.7835,
"step": 156000
},
{
"epoch": 1.72,
"learning_rate": 2.1344553817308258e-05,
"loss": 0.7737,
"step": 156500
},
{
"epoch": 1.72,
"learning_rate": 2.1253002871037675e-05,
"loss": 0.785,
"step": 157000
},
{
"epoch": 1.73,
"learning_rate": 2.1161451924767095e-05,
"loss": 0.7859,
"step": 157500
},
{
"epoch": 1.74,
"learning_rate": 2.1069900978496516e-05,
"loss": 0.7842,
"step": 158000
},
{
"epoch": 1.74,
"learning_rate": 2.0978350032225936e-05,
"loss": 0.7808,
"step": 158500
},
{
"epoch": 1.75,
"learning_rate": 2.0886799085955354e-05,
"loss": 0.7732,
"step": 159000
},
{
"epoch": 1.75,
"learning_rate": 2.0795248139684774e-05,
"loss": 0.7785,
"step": 159500
},
{
"epoch": 1.76,
"learning_rate": 2.070369719341419e-05,
"loss": 0.7794,
"step": 160000
},
{
"epoch": 1.76,
"learning_rate": 2.0612146247143612e-05,
"loss": 0.7893,
"step": 160500
},
{
"epoch": 1.77,
"learning_rate": 2.052059530087303e-05,
"loss": 0.779,
"step": 161000
},
{
"epoch": 1.77,
"learning_rate": 2.042904435460245e-05,
"loss": 0.7792,
"step": 161500
},
{
"epoch": 1.78,
"learning_rate": 2.033749340833187e-05,
"loss": 0.7902,
"step": 162000
},
{
"epoch": 1.79,
"learning_rate": 2.0245942462061288e-05,
"loss": 0.7835,
"step": 162500
},
{
"epoch": 1.79,
"learning_rate": 2.0154391515790708e-05,
"loss": 0.7831,
"step": 163000
},
{
"epoch": 1.8,
"learning_rate": 2.006284056952013e-05,
"loss": 0.7852,
"step": 163500
},
{
"epoch": 1.8,
"learning_rate": 1.997128962324955e-05,
"loss": 0.7855,
"step": 164000
},
{
"epoch": 1.81,
"learning_rate": 1.9879738676978966e-05,
"loss": 0.7821,
"step": 164500
},
{
"epoch": 1.81,
"learning_rate": 1.9788187730708384e-05,
"loss": 0.7762,
"step": 165000
},
{
"epoch": 1.82,
"learning_rate": 1.9696636784437804e-05,
"loss": 0.7852,
"step": 165500
},
{
"epoch": 1.82,
"learning_rate": 1.9605085838167225e-05,
"loss": 0.7857,
"step": 166000
},
{
"epoch": 1.83,
"learning_rate": 1.9513534891896642e-05,
"loss": 0.7872,
"step": 166500
},
{
"epoch": 1.83,
"learning_rate": 1.9421983945626063e-05,
"loss": 0.7749,
"step": 167000
},
{
"epoch": 1.84,
"learning_rate": 1.9330432999355483e-05,
"loss": 0.7867,
"step": 167500
},
{
"epoch": 1.85,
"learning_rate": 1.9238882053084904e-05,
"loss": 0.7801,
"step": 168000
},
{
"epoch": 1.85,
"learning_rate": 1.914733110681432e-05,
"loss": 0.7734,
"step": 168500
},
{
"epoch": 1.86,
"learning_rate": 1.905578016054374e-05,
"loss": 0.7798,
"step": 169000
},
{
"epoch": 1.86,
"learning_rate": 1.896422921427316e-05,
"loss": 0.7862,
"step": 169500
},
{
"epoch": 1.87,
"learning_rate": 1.887267826800258e-05,
"loss": 0.788,
"step": 170000
},
{
"epoch": 1.87,
"learning_rate": 1.8781127321731997e-05,
"loss": 0.7863,
"step": 170500
},
{
"epoch": 1.88,
"learning_rate": 1.8689576375461417e-05,
"loss": 0.7865,
"step": 171000
},
{
"epoch": 1.88,
"learning_rate": 1.8598025429190838e-05,
"loss": 0.79,
"step": 171500
},
{
"epoch": 1.89,
"learning_rate": 1.8506474482920255e-05,
"loss": 0.7871,
"step": 172000
},
{
"epoch": 1.9,
"learning_rate": 1.8414923536649675e-05,
"loss": 0.7882,
"step": 172500
},
{
"epoch": 1.9,
"learning_rate": 1.8323372590379096e-05,
"loss": 0.7811,
"step": 173000
},
{
"epoch": 1.91,
"learning_rate": 1.8231821644108517e-05,
"loss": 0.7827,
"step": 173500
},
{
"epoch": 1.91,
"learning_rate": 1.8140270697837934e-05,
"loss": 0.7741,
"step": 174000
},
{
"epoch": 1.92,
"learning_rate": 1.8048719751567354e-05,
"loss": 0.7743,
"step": 174500
},
{
"epoch": 1.92,
"learning_rate": 1.795716880529677e-05,
"loss": 0.785,
"step": 175000
},
{
"epoch": 1.93,
"learning_rate": 1.7865617859026192e-05,
"loss": 0.7774,
"step": 175500
},
{
"epoch": 1.93,
"learning_rate": 1.777406691275561e-05,
"loss": 0.7921,
"step": 176000
},
{
"epoch": 1.94,
"learning_rate": 1.768251596648503e-05,
"loss": 0.7827,
"step": 176500
},
{
"epoch": 1.94,
"learning_rate": 1.759096502021445e-05,
"loss": 0.7891,
"step": 177000
},
{
"epoch": 1.95,
"learning_rate": 1.7499414073943868e-05,
"loss": 0.7873,
"step": 177500
},
{
"epoch": 1.96,
"learning_rate": 1.7407863127673288e-05,
"loss": 0.7817,
"step": 178000
},
{
"epoch": 1.96,
"learning_rate": 1.731631218140271e-05,
"loss": 0.7802,
"step": 178500
},
{
"epoch": 1.97,
"learning_rate": 1.722476123513213e-05,
"loss": 0.7864,
"step": 179000
},
{
"epoch": 1.97,
"learning_rate": 1.7133210288861547e-05,
"loss": 0.7872,
"step": 179500
},
{
"epoch": 1.98,
"learning_rate": 1.7041659342590964e-05,
"loss": 0.7952,
"step": 180000
},
{
"epoch": 1.98,
"learning_rate": 1.6950108396320384e-05,
"loss": 0.7861,
"step": 180500
},
{
"epoch": 1.99,
"learning_rate": 1.6858557450049805e-05,
"loss": 0.7835,
"step": 181000
},
{
"epoch": 1.99,
"learning_rate": 1.6767006503779222e-05,
"loss": 0.7922,
"step": 181500
},
{
"epoch": 2.0,
"learning_rate": 1.6675455557508643e-05,
"loss": 0.7884,
"step": 182000
},
{
"epoch": 2.0,
"learning_rate": 1.6583904611238063e-05,
"loss": 0.6872,
"step": 182500
},
{
"epoch": 2.01,
"learning_rate": 1.6492353664967484e-05,
"loss": 0.6674,
"step": 183000
},
{
"epoch": 2.02,
"learning_rate": 1.64008027186969e-05,
"loss": 0.6726,
"step": 183500
},
{
"epoch": 2.02,
"learning_rate": 1.630925177242632e-05,
"loss": 0.6642,
"step": 184000
},
{
"epoch": 2.03,
"learning_rate": 1.6217700826155742e-05,
"loss": 0.6699,
"step": 184500
},
{
"epoch": 2.03,
"learning_rate": 1.612614987988516e-05,
"loss": 0.674,
"step": 185000
},
{
"epoch": 2.04,
"learning_rate": 1.6034598933614577e-05,
"loss": 0.6681,
"step": 185500
},
{
"epoch": 2.04,
"learning_rate": 1.5943047987343997e-05,
"loss": 0.6729,
"step": 186000
},
{
"epoch": 2.05,
"learning_rate": 1.5851497041073418e-05,
"loss": 0.671,
"step": 186500
},
{
"epoch": 2.05,
"learning_rate": 1.5759946094802835e-05,
"loss": 0.6731,
"step": 187000
},
{
"epoch": 2.06,
"learning_rate": 1.5668395148532256e-05,
"loss": 0.6765,
"step": 187500
},
{
"epoch": 2.07,
"learning_rate": 1.5576844202261676e-05,
"loss": 0.6757,
"step": 188000
},
{
"epoch": 2.07,
"learning_rate": 1.5485293255991097e-05,
"loss": 0.6647,
"step": 188500
},
{
"epoch": 2.08,
"learning_rate": 1.5393742309720514e-05,
"loss": 0.6795,
"step": 189000
},
{
"epoch": 2.08,
"learning_rate": 1.5302191363449934e-05,
"loss": 0.6673,
"step": 189500
},
{
"epoch": 2.09,
"learning_rate": 1.5210640417179353e-05,
"loss": 0.6701,
"step": 190000
},
{
"epoch": 2.09,
"learning_rate": 1.5119089470908774e-05,
"loss": 0.6726,
"step": 190500
},
{
"epoch": 2.1,
"learning_rate": 1.5027538524638191e-05,
"loss": 0.6646,
"step": 191000
},
{
"epoch": 2.1,
"learning_rate": 1.493598757836761e-05,
"loss": 0.6707,
"step": 191500
},
{
"epoch": 2.11,
"learning_rate": 1.484443663209703e-05,
"loss": 0.6727,
"step": 192000
},
{
"epoch": 2.11,
"learning_rate": 1.4752885685826448e-05,
"loss": 0.6799,
"step": 192500
},
{
"epoch": 2.12,
"learning_rate": 1.4661334739555868e-05,
"loss": 0.6755,
"step": 193000
},
{
"epoch": 2.13,
"learning_rate": 1.4569783793285289e-05,
"loss": 0.6706,
"step": 193500
},
{
"epoch": 2.13,
"learning_rate": 1.4478232847014708e-05,
"loss": 0.6699,
"step": 194000
},
{
"epoch": 2.14,
"learning_rate": 1.4386681900744125e-05,
"loss": 0.6771,
"step": 194500
},
{
"epoch": 2.14,
"learning_rate": 1.4295130954473546e-05,
"loss": 0.6832,
"step": 195000
},
{
"epoch": 2.15,
"learning_rate": 1.4203580008202966e-05,
"loss": 0.6803,
"step": 195500
},
{
"epoch": 2.15,
"learning_rate": 1.4112029061932385e-05,
"loss": 0.6646,
"step": 196000
},
{
"epoch": 2.16,
"learning_rate": 1.4020478115661804e-05,
"loss": 0.6792,
"step": 196500
},
{
"epoch": 2.16,
"learning_rate": 1.3928927169391223e-05,
"loss": 0.6656,
"step": 197000
},
{
"epoch": 2.17,
"learning_rate": 1.3837376223120643e-05,
"loss": 0.6719,
"step": 197500
},
{
"epoch": 2.18,
"learning_rate": 1.3745825276850064e-05,
"loss": 0.6716,
"step": 198000
},
{
"epoch": 2.18,
"learning_rate": 1.3654274330579481e-05,
"loss": 0.6728,
"step": 198500
},
{
"epoch": 2.19,
"learning_rate": 1.35627233843089e-05,
"loss": 0.6708,
"step": 199000
},
{
"epoch": 2.19,
"learning_rate": 1.347117243803832e-05,
"loss": 0.6755,
"step": 199500
},
{
"epoch": 2.2,
"learning_rate": 1.3379621491767741e-05,
"loss": 0.6762,
"step": 200000
},
{
"epoch": 2.2,
"learning_rate": 1.3288070545497158e-05,
"loss": 0.6811,
"step": 200500
},
{
"epoch": 2.21,
"learning_rate": 1.3196519599226579e-05,
"loss": 0.676,
"step": 201000
},
{
"epoch": 2.21,
"learning_rate": 1.3104968652955998e-05,
"loss": 0.6763,
"step": 201500
},
{
"epoch": 2.22,
"learning_rate": 1.3013417706685415e-05,
"loss": 0.6701,
"step": 202000
},
{
"epoch": 2.22,
"learning_rate": 1.2921866760414836e-05,
"loss": 0.6742,
"step": 202500
},
{
"epoch": 2.23,
"learning_rate": 1.2830315814144256e-05,
"loss": 0.6745,
"step": 203000
},
{
"epoch": 2.24,
"learning_rate": 1.2738764867873675e-05,
"loss": 0.6716,
"step": 203500
},
{
"epoch": 2.24,
"learning_rate": 1.2647213921603094e-05,
"loss": 0.6737,
"step": 204000
},
{
"epoch": 2.25,
"learning_rate": 1.2555662975332513e-05,
"loss": 0.6726,
"step": 204500
},
{
"epoch": 2.25,
"learning_rate": 1.2464112029061933e-05,
"loss": 0.6713,
"step": 205000
},
{
"epoch": 2.26,
"learning_rate": 1.2372561082791352e-05,
"loss": 0.6828,
"step": 205500
},
{
"epoch": 2.26,
"learning_rate": 1.2281010136520773e-05,
"loss": 0.6741,
"step": 206000
},
{
"epoch": 2.27,
"learning_rate": 1.218945919025019e-05,
"loss": 0.6757,
"step": 206500
},
{
"epoch": 2.27,
"learning_rate": 1.209790824397961e-05,
"loss": 0.6788,
"step": 207000
},
{
"epoch": 2.28,
"learning_rate": 1.200635729770903e-05,
"loss": 0.6765,
"step": 207500
},
{
"epoch": 2.29,
"learning_rate": 1.1914806351438449e-05,
"loss": 0.6758,
"step": 208000
},
{
"epoch": 2.29,
"learning_rate": 1.1823255405167869e-05,
"loss": 0.6792,
"step": 208500
},
{
"epoch": 2.3,
"learning_rate": 1.1731704458897288e-05,
"loss": 0.6772,
"step": 209000
},
{
"epoch": 2.3,
"learning_rate": 1.1640153512626707e-05,
"loss": 0.6787,
"step": 209500
},
{
"epoch": 2.31,
"learning_rate": 1.1548602566356126e-05,
"loss": 0.6822,
"step": 210000
},
{
"epoch": 2.31,
"learning_rate": 1.1457051620085546e-05,
"loss": 0.6723,
"step": 210500
},
{
"epoch": 2.32,
"learning_rate": 1.1365500673814965e-05,
"loss": 0.6687,
"step": 211000
},
{
"epoch": 2.32,
"learning_rate": 1.1273949727544384e-05,
"loss": 0.6849,
"step": 211500
},
{
"epoch": 2.33,
"learning_rate": 1.1182398781273803e-05,
"loss": 0.6772,
"step": 212000
},
{
"epoch": 2.33,
"learning_rate": 1.1090847835003224e-05,
"loss": 0.6738,
"step": 212500
},
{
"epoch": 2.34,
"learning_rate": 1.0999296888732642e-05,
"loss": 0.6732,
"step": 213000
},
{
"epoch": 2.35,
"learning_rate": 1.0907745942462063e-05,
"loss": 0.6776,
"step": 213500
},
{
"epoch": 2.35,
"learning_rate": 1.081619499619148e-05,
"loss": 0.6744,
"step": 214000
},
{
"epoch": 2.36,
"learning_rate": 1.07246440499209e-05,
"loss": 0.6792,
"step": 214500
},
{
"epoch": 2.36,
"learning_rate": 1.063309310365032e-05,
"loss": 0.6808,
"step": 215000
},
{
"epoch": 2.37,
"learning_rate": 1.0541542157379739e-05,
"loss": 0.6777,
"step": 215500
},
{
"epoch": 2.37,
"learning_rate": 1.0449991211109159e-05,
"loss": 0.683,
"step": 216000
},
{
"epoch": 2.38,
"learning_rate": 1.0358440264838578e-05,
"loss": 0.6767,
"step": 216500
},
{
"epoch": 2.38,
"learning_rate": 1.0266889318567997e-05,
"loss": 0.6744,
"step": 217000
},
{
"epoch": 2.39,
"learning_rate": 1.0175338372297416e-05,
"loss": 0.6777,
"step": 217500
},
{
"epoch": 2.39,
"learning_rate": 1.0083787426026836e-05,
"loss": 0.6794,
"step": 218000
},
{
"epoch": 2.4,
"learning_rate": 9.992236479756255e-06,
"loss": 0.6705,
"step": 218500
},
{
"epoch": 2.41,
"learning_rate": 9.900685533485674e-06,
"loss": 0.6696,
"step": 219000
},
{
"epoch": 2.41,
"learning_rate": 9.809134587215093e-06,
"loss": 0.6829,
"step": 219500
},
{
"epoch": 2.42,
"learning_rate": 9.717583640944514e-06,
"loss": 0.6702,
"step": 220000
},
{
"epoch": 2.42,
"learning_rate": 9.626032694673932e-06,
"loss": 0.6759,
"step": 220500
},
{
"epoch": 2.43,
"learning_rate": 9.534481748403353e-06,
"loss": 0.6822,
"step": 221000
},
{
"epoch": 2.43,
"learning_rate": 9.442930802132772e-06,
"loss": 0.6771,
"step": 221500
},
{
"epoch": 2.44,
"learning_rate": 9.35137985586219e-06,
"loss": 0.6752,
"step": 222000
},
{
"epoch": 2.44,
"learning_rate": 9.25982890959161e-06,
"loss": 0.6752,
"step": 222500
},
{
"epoch": 2.45,
"learning_rate": 9.16827796332103e-06,
"loss": 0.6752,
"step": 223000
},
{
"epoch": 2.46,
"learning_rate": 9.07672701705045e-06,
"loss": 0.6829,
"step": 223500
},
{
"epoch": 2.46,
"learning_rate": 8.985176070779868e-06,
"loss": 0.6732,
"step": 224000
},
{
"epoch": 2.47,
"learning_rate": 8.893625124509287e-06,
"loss": 0.683,
"step": 224500
},
{
"epoch": 2.47,
"learning_rate": 8.802074178238706e-06,
"loss": 0.6717,
"step": 225000
},
{
"epoch": 2.48,
"learning_rate": 8.710523231968126e-06,
"loss": 0.678,
"step": 225500
},
{
"epoch": 2.48,
"learning_rate": 8.618972285697545e-06,
"loss": 0.6752,
"step": 226000
},
{
"epoch": 2.49,
"learning_rate": 8.527421339426964e-06,
"loss": 0.6805,
"step": 226500
},
{
"epoch": 2.49,
"learning_rate": 8.435870393156383e-06,
"loss": 0.6821,
"step": 227000
},
{
"epoch": 2.5,
"learning_rate": 8.344319446885804e-06,
"loss": 0.6802,
"step": 227500
},
{
"epoch": 2.5,
"learning_rate": 8.252768500615223e-06,
"loss": 0.6703,
"step": 228000
},
{
"epoch": 2.51,
"learning_rate": 8.161217554344643e-06,
"loss": 0.6724,
"step": 228500
},
{
"epoch": 2.52,
"learning_rate": 8.069666608074062e-06,
"loss": 0.6758,
"step": 229000
},
{
"epoch": 2.52,
"learning_rate": 7.978115661803481e-06,
"loss": 0.6798,
"step": 229500
},
{
"epoch": 2.53,
"learning_rate": 7.8865647155329e-06,
"loss": 0.6832,
"step": 230000
},
{
"epoch": 2.53,
"learning_rate": 7.79501376926232e-06,
"loss": 0.6691,
"step": 230500
},
{
"epoch": 2.54,
"learning_rate": 7.70346282299174e-06,
"loss": 0.6848,
"step": 231000
},
{
"epoch": 2.54,
"learning_rate": 7.611911876721157e-06,
"loss": 0.6699,
"step": 231500
},
{
"epoch": 2.55,
"learning_rate": 7.520360930450578e-06,
"loss": 0.682,
"step": 232000
},
{
"epoch": 2.55,
"learning_rate": 7.428809984179996e-06,
"loss": 0.6789,
"step": 232500
},
{
"epoch": 2.56,
"learning_rate": 7.3372590379094165e-06,
"loss": 0.6749,
"step": 233000
},
{
"epoch": 2.57,
"learning_rate": 7.245708091638835e-06,
"loss": 0.6749,
"step": 233500
},
{
"epoch": 2.57,
"learning_rate": 7.154157145368255e-06,
"loss": 0.6771,
"step": 234000
},
{
"epoch": 2.58,
"learning_rate": 7.062606199097674e-06,
"loss": 0.6848,
"step": 234500
},
{
"epoch": 2.58,
"learning_rate": 6.971055252827094e-06,
"loss": 0.677,
"step": 235000
},
{
"epoch": 2.59,
"learning_rate": 6.879504306556513e-06,
"loss": 0.6817,
"step": 235500
},
{
"epoch": 2.59,
"learning_rate": 6.787953360285932e-06,
"loss": 0.6776,
"step": 236000
},
{
"epoch": 2.6,
"learning_rate": 6.696402414015351e-06,
"loss": 0.6794,
"step": 236500
},
{
"epoch": 2.6,
"learning_rate": 6.604851467744772e-06,
"loss": 0.6825,
"step": 237000
},
{
"epoch": 2.61,
"learning_rate": 6.51330052147419e-06,
"loss": 0.6717,
"step": 237500
},
{
"epoch": 2.61,
"learning_rate": 6.4217495752036104e-06,
"loss": 0.6819,
"step": 238000
},
{
"epoch": 2.62,
"learning_rate": 6.330198628933029e-06,
"loss": 0.674,
"step": 238500
},
{
"epoch": 2.63,
"learning_rate": 6.238647682662448e-06,
"loss": 0.6757,
"step": 239000
},
{
"epoch": 2.63,
"learning_rate": 6.147096736391868e-06,
"loss": 0.6771,
"step": 239500
},
{
"epoch": 2.64,
"learning_rate": 6.055545790121287e-06,
"loss": 0.6848,
"step": 240000
},
{
"epoch": 2.64,
"learning_rate": 5.9639948438507066e-06,
"loss": 0.6724,
"step": 240500
},
{
"epoch": 2.65,
"learning_rate": 5.872443897580126e-06,
"loss": 0.6797,
"step": 241000
},
{
"epoch": 2.65,
"learning_rate": 5.780892951309545e-06,
"loss": 0.676,
"step": 241500
},
{
"epoch": 2.66,
"learning_rate": 5.689342005038965e-06,
"loss": 0.6777,
"step": 242000
},
{
"epoch": 2.66,
"learning_rate": 5.597791058768383e-06,
"loss": 0.6707,
"step": 242500
},
{
"epoch": 2.67,
"learning_rate": 5.506240112497803e-06,
"loss": 0.6837,
"step": 243000
},
{
"epoch": 2.68,
"learning_rate": 5.414689166227222e-06,
"loss": 0.6732,
"step": 243500
},
{
"epoch": 2.68,
"learning_rate": 5.323138219956641e-06,
"loss": 0.6745,
"step": 244000
},
{
"epoch": 2.69,
"learning_rate": 5.231587273686061e-06,
"loss": 0.6766,
"step": 244500
},
{
"epoch": 2.69,
"learning_rate": 5.14003632741548e-06,
"loss": 0.6726,
"step": 245000
},
{
"epoch": 2.7,
"learning_rate": 5.0484853811449e-06,
"loss": 0.6821,
"step": 245500
},
{
"epoch": 2.7,
"learning_rate": 4.956934434874319e-06,
"loss": 0.6708,
"step": 246000
},
{
"epoch": 2.71,
"learning_rate": 4.865383488603738e-06,
"loss": 0.6837,
"step": 246500
},
{
"epoch": 2.71,
"learning_rate": 4.773832542333158e-06,
"loss": 0.6752,
"step": 247000
},
{
"epoch": 2.72,
"learning_rate": 4.682281596062577e-06,
"loss": 0.6718,
"step": 247500
},
{
"epoch": 2.72,
"learning_rate": 4.590730649791997e-06,
"loss": 0.6812,
"step": 248000
},
{
"epoch": 2.73,
"learning_rate": 4.499179703521416e-06,
"loss": 0.6803,
"step": 248500
},
{
"epoch": 2.74,
"learning_rate": 4.407628757250835e-06,
"loss": 0.6803,
"step": 249000
},
{
"epoch": 2.74,
"learning_rate": 4.316077810980255e-06,
"loss": 0.6784,
"step": 249500
},
{
"epoch": 2.75,
"learning_rate": 4.224526864709674e-06,
"loss": 0.6765,
"step": 250000
},
{
"epoch": 2.75,
"learning_rate": 4.132975918439093e-06,
"loss": 0.6792,
"step": 250500
},
{
"epoch": 2.76,
"learning_rate": 4.0414249721685125e-06,
"loss": 0.6786,
"step": 251000
},
{
"epoch": 2.76,
"learning_rate": 3.949874025897931e-06,
"loss": 0.6765,
"step": 251500
},
{
"epoch": 2.77,
"learning_rate": 3.858323079627351e-06,
"loss": 0.673,
"step": 252000
},
{
"epoch": 2.77,
"learning_rate": 3.7667721333567704e-06,
"loss": 0.6727,
"step": 252500
},
{
"epoch": 2.78,
"learning_rate": 3.6752211870861897e-06,
"loss": 0.6751,
"step": 253000
},
{
"epoch": 2.78,
"learning_rate": 3.583670240815609e-06,
"loss": 0.6832,
"step": 253500
},
{
"epoch": 2.79,
"learning_rate": 3.4921192945450287e-06,
"loss": 0.6742,
"step": 254000
},
{
"epoch": 2.8,
"learning_rate": 3.400568348274448e-06,
"loss": 0.6738,
"step": 254500
},
{
"epoch": 2.8,
"learning_rate": 3.3090174020038674e-06,
"loss": 0.681,
"step": 255000
},
{
"epoch": 2.81,
"learning_rate": 3.2174664557332867e-06,
"loss": 0.6724,
"step": 255500
},
{
"epoch": 2.81,
"learning_rate": 3.125915509462706e-06,
"loss": 0.6734,
"step": 256000
},
{
"epoch": 2.82,
"learning_rate": 3.0343645631921253e-06,
"loss": 0.6821,
"step": 256500
},
{
"epoch": 2.82,
"learning_rate": 2.9428136169215446e-06,
"loss": 0.6754,
"step": 257000
},
{
"epoch": 2.83,
"learning_rate": 2.851262670650964e-06,
"loss": 0.6805,
"step": 257500
},
{
"epoch": 2.83,
"learning_rate": 2.7597117243803832e-06,
"loss": 0.6777,
"step": 258000
},
{
"epoch": 2.84,
"learning_rate": 2.668160778109803e-06,
"loss": 0.6756,
"step": 258500
},
{
"epoch": 2.85,
"learning_rate": 2.5766098318392223e-06,
"loss": 0.6835,
"step": 259000
},
{
"epoch": 2.85,
"learning_rate": 2.4850588855686416e-06,
"loss": 0.675,
"step": 259500
},
{
"epoch": 2.86,
"learning_rate": 2.3935079392980604e-06,
"loss": 0.6733,
"step": 260000
},
{
"epoch": 2.86,
"learning_rate": 2.3019569930274798e-06,
"loss": 0.6761,
"step": 260500
},
{
"epoch": 2.87,
"learning_rate": 2.2104060467568995e-06,
"loss": 0.6753,
"step": 261000
},
{
"epoch": 2.87,
"learning_rate": 2.118855100486319e-06,
"loss": 0.679,
"step": 261500
},
{
"epoch": 2.88,
"learning_rate": 2.027304154215738e-06,
"loss": 0.6707,
"step": 262000
},
{
"epoch": 2.88,
"learning_rate": 1.9357532079451574e-06,
"loss": 0.6744,
"step": 262500
},
{
"epoch": 2.89,
"learning_rate": 1.844202261674577e-06,
"loss": 0.678,
"step": 263000
},
{
"epoch": 2.89,
"learning_rate": 1.7526513154039962e-06,
"loss": 0.6655,
"step": 263500
},
{
"epoch": 2.9,
"learning_rate": 1.6611003691334153e-06,
"loss": 0.6812,
"step": 264000
},
{
"epoch": 2.91,
"learning_rate": 1.5695494228628347e-06,
"loss": 0.6797,
"step": 264500
},
{
"epoch": 2.91,
"learning_rate": 1.4779984765922542e-06,
"loss": 0.6727,
"step": 265000
},
{
"epoch": 2.92,
"learning_rate": 1.3864475303216735e-06,
"loss": 0.6738,
"step": 265500
},
{
"epoch": 2.92,
"learning_rate": 1.2948965840510928e-06,
"loss": 0.6799,
"step": 266000
},
{
"epoch": 2.93,
"learning_rate": 1.203345637780512e-06,
"loss": 0.6779,
"step": 266500
},
{
"epoch": 2.93,
"learning_rate": 1.1117946915099314e-06,
"loss": 0.68,
"step": 267000
},
{
"epoch": 2.94,
"learning_rate": 1.020243745239351e-06,
"loss": 0.6783,
"step": 267500
},
{
"epoch": 2.94,
"learning_rate": 9.286927989687702e-07,
"loss": 0.6851,
"step": 268000
},
{
"epoch": 2.95,
"learning_rate": 8.371418526981894e-07,
"loss": 0.6734,
"step": 268500
},
{
"epoch": 2.96,
"learning_rate": 7.455909064276089e-07,
"loss": 0.6796,
"step": 269000
},
{
"epoch": 2.96,
"learning_rate": 6.540399601570283e-07,
"loss": 0.6754,
"step": 269500
},
{
"epoch": 2.97,
"learning_rate": 5.624890138864476e-07,
"loss": 0.6813,
"step": 270000
},
{
"epoch": 2.97,
"learning_rate": 4.7093806761586695e-07,
"loss": 0.6758,
"step": 270500
},
{
"epoch": 2.98,
"learning_rate": 3.7938712134528625e-07,
"loss": 0.6788,
"step": 271000
},
{
"epoch": 2.98,
"learning_rate": 2.8783617507470556e-07,
"loss": 0.6753,
"step": 271500
},
{
"epoch": 2.99,
"learning_rate": 1.962852288041249e-07,
"loss": 0.6819,
"step": 272000
},
{
"epoch": 2.99,
"learning_rate": 1.0473428253354427e-07,
"loss": 0.675,
"step": 272500
},
{
"epoch": 3.0,
"learning_rate": 1.3183336262963613e-08,
"loss": 0.6794,
"step": 273000
},
{
"epoch": 3.0,
"step": 273072,
"total_flos": 4122145169772675840,
"train_runtime": 309389.2046,
"train_samples_per_second": 0.883
}
],
"max_steps": 273072,
"num_train_epochs": 3,
"total_flos": 4122145169772675840,
"trial_name": null,
"trial_params": null
}