|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"global_step": 162036, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.98457132982794e-05, |
|
"loss": 3.1609, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9691426596558794e-05, |
|
"loss": 2.4105, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.953713989483818e-05, |
|
"loss": 2.2104, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.938285319311758e-05, |
|
"loss": 2.0895, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.9228566491396974e-05, |
|
"loss": 1.9949, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.907427978967637e-05, |
|
"loss": 1.936, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.8919993087955766e-05, |
|
"loss": 1.8854, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.876570638623516e-05, |
|
"loss": 1.8358, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 4.861141968451455e-05, |
|
"loss": 1.7885, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 4.845713298279395e-05, |
|
"loss": 1.738, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 4.830284628107334e-05, |
|
"loss": 1.6968, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 4.814855957935274e-05, |
|
"loss": 1.6704, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 4.7994272877632133e-05, |
|
"loss": 1.657, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 4.783998617591153e-05, |
|
"loss": 1.6384, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 4.768569947419092e-05, |
|
"loss": 1.6121, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 4.753141277247032e-05, |
|
"loss": 1.6024, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 4.737712607074971e-05, |
|
"loss": 1.6227, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 4.722283936902911e-05, |
|
"loss": 1.5951, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 4.70685526673085e-05, |
|
"loss": 1.6013, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 4.69142659655879e-05, |
|
"loss": 1.5513, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 4.675997926386729e-05, |
|
"loss": 1.5365, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 4.660569256214669e-05, |
|
"loss": 1.515, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.645140586042608e-05, |
|
"loss": 1.5089, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 4.629711915870548e-05, |
|
"loss": 1.5182, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 4.614283245698487e-05, |
|
"loss": 1.4778, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 4.5988545755264264e-05, |
|
"loss": 1.4955, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 4.583425905354366e-05, |
|
"loss": 1.4609, |
|
"step": 13500 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 4.5679972351823056e-05, |
|
"loss": 1.4568, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 4.552568565010245e-05, |
|
"loss": 1.4787, |
|
"step": 14500 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.537139894838184e-05, |
|
"loss": 1.4693, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 4.5217112246661236e-05, |
|
"loss": 1.4202, |
|
"step": 15500 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 4.506282554494063e-05, |
|
"loss": 1.4152, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 4.490853884322003e-05, |
|
"loss": 1.4079, |
|
"step": 16500 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 4.475425214149942e-05, |
|
"loss": 1.4345, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 4.459996543977882e-05, |
|
"loss": 1.4313, |
|
"step": 17500 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 4.444567873805821e-05, |
|
"loss": 1.4218, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 4.4291392036337604e-05, |
|
"loss": 1.4003, |
|
"step": 18500 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 4.4137105334617e-05, |
|
"loss": 1.3965, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.3982818632896395e-05, |
|
"loss": 1.405, |
|
"step": 19500 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 4.382853193117579e-05, |
|
"loss": 1.3694, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 4.367424522945519e-05, |
|
"loss": 1.4026, |
|
"step": 20500 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 4.3519958527734576e-05, |
|
"loss": 1.369, |
|
"step": 21000 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 4.336567182601398e-05, |
|
"loss": 1.3753, |
|
"step": 21500 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 4.321138512429337e-05, |
|
"loss": 1.3669, |
|
"step": 22000 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 4.305709842257276e-05, |
|
"loss": 1.3495, |
|
"step": 22500 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 4.290281172085216e-05, |
|
"loss": 1.368, |
|
"step": 23000 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.2748525019131555e-05, |
|
"loss": 1.3656, |
|
"step": 23500 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.259423831741095e-05, |
|
"loss": 1.3532, |
|
"step": 24000 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.2439951615690346e-05, |
|
"loss": 1.3494, |
|
"step": 24500 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.2285664913969735e-05, |
|
"loss": 1.3629, |
|
"step": 25000 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.213137821224913e-05, |
|
"loss": 1.3426, |
|
"step": 25500 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.1977091510528526e-05, |
|
"loss": 1.3425, |
|
"step": 26000 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.182280480880792e-05, |
|
"loss": 1.3437, |
|
"step": 26500 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.166851810708732e-05, |
|
"loss": 1.3376, |
|
"step": 27000 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.1514231405366714e-05, |
|
"loss": 1.3369, |
|
"step": 27500 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.13599447036461e-05, |
|
"loss": 1.3048, |
|
"step": 28000 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 4.12056580019255e-05, |
|
"loss": 1.3037, |
|
"step": 28500 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.1051371300204894e-05, |
|
"loss": 1.3129, |
|
"step": 29000 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 4.089708459848429e-05, |
|
"loss": 1.3117, |
|
"step": 29500 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.0742797896763686e-05, |
|
"loss": 1.3195, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.0588511195043074e-05, |
|
"loss": 1.2952, |
|
"step": 30500 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 4.043422449332248e-05, |
|
"loss": 1.3208, |
|
"step": 31000 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 4.0279937791601866e-05, |
|
"loss": 1.2683, |
|
"step": 31500 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 4.012565108988126e-05, |
|
"loss": 1.3242, |
|
"step": 32000 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 3.997136438816066e-05, |
|
"loss": 1.2866, |
|
"step": 32500 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 3.981707768644005e-05, |
|
"loss": 1.2814, |
|
"step": 33000 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 3.966279098471944e-05, |
|
"loss": 1.267, |
|
"step": 33500 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 3.9508504282998845e-05, |
|
"loss": 1.2745, |
|
"step": 34000 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 3.9354217581278234e-05, |
|
"loss": 1.2754, |
|
"step": 34500 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 3.9199930879557636e-05, |
|
"loss": 1.2779, |
|
"step": 35000 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 3.9045644177837025e-05, |
|
"loss": 1.2782, |
|
"step": 35500 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 3.889135747611642e-05, |
|
"loss": 1.2749, |
|
"step": 36000 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 3.8737070774395817e-05, |
|
"loss": 1.2708, |
|
"step": 36500 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 3.858278407267521e-05, |
|
"loss": 1.2362, |
|
"step": 37000 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 3.84284973709546e-05, |
|
"loss": 1.2475, |
|
"step": 37500 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 3.8274210669234004e-05, |
|
"loss": 1.2623, |
|
"step": 38000 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 3.811992396751339e-05, |
|
"loss": 1.2401, |
|
"step": 38500 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 3.796563726579279e-05, |
|
"loss": 1.256, |
|
"step": 39000 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.7811350564072184e-05, |
|
"loss": 1.2678, |
|
"step": 39500 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.765706386235158e-05, |
|
"loss": 1.2604, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.7502777160630976e-05, |
|
"loss": 1.2583, |
|
"step": 40500 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.7348490458910365e-05, |
|
"loss": 1.2637, |
|
"step": 41000 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.719420375718976e-05, |
|
"loss": 1.2521, |
|
"step": 41500 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 3.7039917055469156e-05, |
|
"loss": 1.2563, |
|
"step": 42000 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 3.688563035374855e-05, |
|
"loss": 1.2083, |
|
"step": 42500 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 3.673134365202794e-05, |
|
"loss": 1.2254, |
|
"step": 43000 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 3.657705695030734e-05, |
|
"loss": 1.2491, |
|
"step": 43500 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 3.642277024858673e-05, |
|
"loss": 1.2368, |
|
"step": 44000 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 3.626848354686613e-05, |
|
"loss": 1.2189, |
|
"step": 44500 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 3.6114196845145524e-05, |
|
"loss": 1.237, |
|
"step": 45000 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 3.595991014342492e-05, |
|
"loss": 1.2249, |
|
"step": 45500 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 3.5805623441704315e-05, |
|
"loss": 1.2603, |
|
"step": 46000 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 3.565133673998371e-05, |
|
"loss": 1.2219, |
|
"step": 46500 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 3.54970500382631e-05, |
|
"loss": 1.2132, |
|
"step": 47000 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 3.53427633365425e-05, |
|
"loss": 1.2137, |
|
"step": 47500 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 3.518847663482189e-05, |
|
"loss": 1.208, |
|
"step": 48000 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 3.503418993310129e-05, |
|
"loss": 1.1826, |
|
"step": 48500 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 3.487990323138068e-05, |
|
"loss": 1.2155, |
|
"step": 49000 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.472561652966008e-05, |
|
"loss": 1.1884, |
|
"step": 49500 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 3.457132982793947e-05, |
|
"loss": 1.1927, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 3.441704312621887e-05, |
|
"loss": 1.2108, |
|
"step": 50500 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 3.426275642449826e-05, |
|
"loss": 1.2166, |
|
"step": 51000 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.410846972277766e-05, |
|
"loss": 1.2063, |
|
"step": 51500 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 3.395418302105705e-05, |
|
"loss": 1.1726, |
|
"step": 52000 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 3.3799896319336446e-05, |
|
"loss": 1.1837, |
|
"step": 52500 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 3.364560961761584e-05, |
|
"loss": 1.17, |
|
"step": 53000 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 3.349132291589524e-05, |
|
"loss": 1.1594, |
|
"step": 53500 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 3.333703621417463e-05, |
|
"loss": 1.1853, |
|
"step": 54000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 3.318274951245402e-05, |
|
"loss": 1.1132, |
|
"step": 54500 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 3.302846281073342e-05, |
|
"loss": 1.0881, |
|
"step": 55000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 3.2874176109012814e-05, |
|
"loss": 1.0903, |
|
"step": 55500 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 3.271988940729221e-05, |
|
"loss": 1.0954, |
|
"step": 56000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 3.25656027055716e-05, |
|
"loss": 1.0853, |
|
"step": 56500 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 3.2411316003851e-05, |
|
"loss": 1.0738, |
|
"step": 57000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 3.225702930213039e-05, |
|
"loss": 1.1271, |
|
"step": 57500 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 3.2102742600409786e-05, |
|
"loss": 1.1006, |
|
"step": 58000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 3.194845589868918e-05, |
|
"loss": 1.0875, |
|
"step": 58500 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 3.179416919696858e-05, |
|
"loss": 1.0988, |
|
"step": 59000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 3.1639882495247966e-05, |
|
"loss": 1.0578, |
|
"step": 59500 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 3.148559579352737e-05, |
|
"loss": 1.0941, |
|
"step": 60000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 3.133130909180676e-05, |
|
"loss": 1.0985, |
|
"step": 60500 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 3.117702239008615e-05, |
|
"loss": 1.0863, |
|
"step": 61000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 3.102273568836555e-05, |
|
"loss": 1.0962, |
|
"step": 61500 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 3.0868448986644945e-05, |
|
"loss": 1.1075, |
|
"step": 62000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 3.071416228492434e-05, |
|
"loss": 1.0849, |
|
"step": 62500 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 3.0559875583203736e-05, |
|
"loss": 1.0525, |
|
"step": 63000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 3.0405588881483125e-05, |
|
"loss": 1.0866, |
|
"step": 63500 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 3.0251302179762524e-05, |
|
"loss": 1.0725, |
|
"step": 64000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 3.0097015478041917e-05, |
|
"loss": 1.1056, |
|
"step": 64500 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 2.994272877632131e-05, |
|
"loss": 1.0848, |
|
"step": 65000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 2.9788442074600708e-05, |
|
"loss": 1.068, |
|
"step": 65500 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 2.96341553728801e-05, |
|
"loss": 1.0791, |
|
"step": 66000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 2.94798686711595e-05, |
|
"loss": 1.0883, |
|
"step": 66500 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 2.9325581969438892e-05, |
|
"loss": 1.0806, |
|
"step": 67000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 2.9171295267718284e-05, |
|
"loss": 1.076, |
|
"step": 67500 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 2.9017008565997683e-05, |
|
"loss": 1.0845, |
|
"step": 68000 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 2.8862721864277076e-05, |
|
"loss": 1.0839, |
|
"step": 68500 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 2.8708435162556468e-05, |
|
"loss": 1.0899, |
|
"step": 69000 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 2.8554148460835867e-05, |
|
"loss": 1.0548, |
|
"step": 69500 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 2.839986175911526e-05, |
|
"loss": 1.0654, |
|
"step": 70000 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 2.8245575057394652e-05, |
|
"loss": 1.0835, |
|
"step": 70500 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 2.809128835567405e-05, |
|
"loss": 1.0763, |
|
"step": 71000 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 2.7937001653953443e-05, |
|
"loss": 1.0636, |
|
"step": 71500 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 2.7782714952232843e-05, |
|
"loss": 1.0585, |
|
"step": 72000 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 2.7628428250512235e-05, |
|
"loss": 1.0698, |
|
"step": 72500 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 2.7474141548791627e-05, |
|
"loss": 1.076, |
|
"step": 73000 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 2.7319854847071026e-05, |
|
"loss": 1.071, |
|
"step": 73500 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 2.716556814535042e-05, |
|
"loss": 1.0881, |
|
"step": 74000 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 2.701128144362981e-05, |
|
"loss": 1.074, |
|
"step": 74500 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"learning_rate": 2.6856994741909207e-05, |
|
"loss": 1.0662, |
|
"step": 75000 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 2.6702708040188603e-05, |
|
"loss": 1.0285, |
|
"step": 75500 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 2.6548421338467995e-05, |
|
"loss": 1.0728, |
|
"step": 76000 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 2.639413463674739e-05, |
|
"loss": 1.058, |
|
"step": 76500 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 2.6239847935026783e-05, |
|
"loss": 1.0803, |
|
"step": 77000 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 2.6085561233306182e-05, |
|
"loss": 1.0426, |
|
"step": 77500 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 2.5931274531585574e-05, |
|
"loss": 1.0938, |
|
"step": 78000 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 2.5776987829864967e-05, |
|
"loss": 1.0677, |
|
"step": 78500 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 2.5622701128144366e-05, |
|
"loss": 1.0662, |
|
"step": 79000 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 2.5468414426423758e-05, |
|
"loss": 1.0569, |
|
"step": 79500 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 2.531412772470315e-05, |
|
"loss": 1.033, |
|
"step": 80000 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"learning_rate": 2.515984102298255e-05, |
|
"loss": 1.062, |
|
"step": 80500 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 2.5005554321261942e-05, |
|
"loss": 1.0434, |
|
"step": 81000 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 2.4851267619541338e-05, |
|
"loss": 1.0191, |
|
"step": 81500 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 2.4696980917820734e-05, |
|
"loss": 1.0571, |
|
"step": 82000 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 2.4542694216100126e-05, |
|
"loss": 1.0588, |
|
"step": 82500 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 2.438840751437952e-05, |
|
"loss": 1.0592, |
|
"step": 83000 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 2.4234120812658917e-05, |
|
"loss": 1.0456, |
|
"step": 83500 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 2.4079834110938313e-05, |
|
"loss": 1.0463, |
|
"step": 84000 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 2.3925547409217705e-05, |
|
"loss": 1.0612, |
|
"step": 84500 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 2.37712607074971e-05, |
|
"loss": 1.0416, |
|
"step": 85000 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 2.3616974005776497e-05, |
|
"loss": 1.0474, |
|
"step": 85500 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 2.346268730405589e-05, |
|
"loss": 1.0643, |
|
"step": 86000 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 2.3308400602335285e-05, |
|
"loss": 1.0452, |
|
"step": 86500 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 2.315411390061468e-05, |
|
"loss": 1.0124, |
|
"step": 87000 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 2.2999827198894077e-05, |
|
"loss": 1.0365, |
|
"step": 87500 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 2.284554049717347e-05, |
|
"loss": 1.046, |
|
"step": 88000 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 2.2691253795452865e-05, |
|
"loss": 1.0623, |
|
"step": 88500 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 2.2536967093732257e-05, |
|
"loss": 1.0229, |
|
"step": 89000 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 2.2382680392011653e-05, |
|
"loss": 1.0386, |
|
"step": 89500 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 2.2228393690291045e-05, |
|
"loss": 1.0264, |
|
"step": 90000 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 2.207410698857044e-05, |
|
"loss": 1.0507, |
|
"step": 90500 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 2.1919820286849836e-05, |
|
"loss": 1.0286, |
|
"step": 91000 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 2.176553358512923e-05, |
|
"loss": 1.0485, |
|
"step": 91500 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 2.1611246883408625e-05, |
|
"loss": 1.0334, |
|
"step": 92000 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 2.145696018168802e-05, |
|
"loss": 1.0558, |
|
"step": 92500 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 2.1302673479967416e-05, |
|
"loss": 1.0308, |
|
"step": 93000 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 2.114838677824681e-05, |
|
"loss": 1.0372, |
|
"step": 93500 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 2.0994100076526204e-05, |
|
"loss": 1.004, |
|
"step": 94000 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 2.08398133748056e-05, |
|
"loss": 1.0301, |
|
"step": 94500 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 2.0685526673084996e-05, |
|
"loss": 1.0352, |
|
"step": 95000 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 2.0531239971364388e-05, |
|
"loss": 1.0268, |
|
"step": 95500 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 2.0376953269643784e-05, |
|
"loss": 1.0276, |
|
"step": 96000 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 2.022266656792318e-05, |
|
"loss": 1.0155, |
|
"step": 96500 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 2.0068379866202575e-05, |
|
"loss": 1.0047, |
|
"step": 97000 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 1.9914093164481967e-05, |
|
"loss": 1.0241, |
|
"step": 97500 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 1.9759806462761363e-05, |
|
"loss": 1.0427, |
|
"step": 98000 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 1.960551976104076e-05, |
|
"loss": 1.0136, |
|
"step": 98500 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 1.945123305932015e-05, |
|
"loss": 0.9878, |
|
"step": 99000 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 1.9296946357599547e-05, |
|
"loss": 1.0195, |
|
"step": 99500 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 1.9142659655878943e-05, |
|
"loss": 1.0425, |
|
"step": 100000 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 1.898837295415834e-05, |
|
"loss": 0.9886, |
|
"step": 100500 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 1.883408625243773e-05, |
|
"loss": 1.0276, |
|
"step": 101000 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 1.8679799550717127e-05, |
|
"loss": 1.0217, |
|
"step": 101500 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 1.8525512848996522e-05, |
|
"loss": 1.0214, |
|
"step": 102000 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 1.8371226147275915e-05, |
|
"loss": 1.0278, |
|
"step": 102500 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 1.821693944555531e-05, |
|
"loss": 1.0075, |
|
"step": 103000 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 1.8062652743834703e-05, |
|
"loss": 1.011, |
|
"step": 103500 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 1.79083660421141e-05, |
|
"loss": 1.0148, |
|
"step": 104000 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 1.775407934039349e-05, |
|
"loss": 1.0112, |
|
"step": 104500 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 1.7599792638672887e-05, |
|
"loss": 1.0071, |
|
"step": 105000 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 1.7445505936952282e-05, |
|
"loss": 1.0145, |
|
"step": 105500 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 1.7291219235231678e-05, |
|
"loss": 1.0116, |
|
"step": 106000 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 1.713693253351107e-05, |
|
"loss": 1.005, |
|
"step": 106500 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 1.6982645831790466e-05, |
|
"loss": 1.0114, |
|
"step": 107000 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 1.6828359130069862e-05, |
|
"loss": 1.0087, |
|
"step": 107500 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 1.6674072428349258e-05, |
|
"loss": 1.0021, |
|
"step": 108000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 1.651978572662865e-05, |
|
"loss": 0.8941, |
|
"step": 108500 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 1.6365499024908046e-05, |
|
"loss": 0.8985, |
|
"step": 109000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 1.621121232318744e-05, |
|
"loss": 0.897, |
|
"step": 109500 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 1.6056925621466834e-05, |
|
"loss": 0.9027, |
|
"step": 110000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 1.590263891974623e-05, |
|
"loss": 0.9026, |
|
"step": 110500 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 1.5748352218025625e-05, |
|
"loss": 0.8935, |
|
"step": 111000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 1.559406551630502e-05, |
|
"loss": 0.8927, |
|
"step": 111500 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 1.5439778814584413e-05, |
|
"loss": 0.8965, |
|
"step": 112000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 1.528549211286381e-05, |
|
"loss": 0.8983, |
|
"step": 112500 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 1.5131205411143203e-05, |
|
"loss": 0.9196, |
|
"step": 113000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 1.4976918709422599e-05, |
|
"loss": 0.914, |
|
"step": 113500 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 1.4822632007701991e-05, |
|
"loss": 0.8874, |
|
"step": 114000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 1.4668345305981387e-05, |
|
"loss": 0.8932, |
|
"step": 114500 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 1.4514058604260783e-05, |
|
"loss": 0.8743, |
|
"step": 115000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 1.4359771902540175e-05, |
|
"loss": 0.8845, |
|
"step": 115500 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 1.420548520081957e-05, |
|
"loss": 0.8938, |
|
"step": 116000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 1.4051198499098966e-05, |
|
"loss": 0.9071, |
|
"step": 116500 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 1.3896911797378362e-05, |
|
"loss": 0.9015, |
|
"step": 117000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 1.3742625095657755e-05, |
|
"loss": 0.8923, |
|
"step": 117500 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 1.358833839393715e-05, |
|
"loss": 0.8793, |
|
"step": 118000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.3434051692216546e-05, |
|
"loss": 0.8973, |
|
"step": 118500 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.3279764990495942e-05, |
|
"loss": 0.8972, |
|
"step": 119000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.3125478288775334e-05, |
|
"loss": 0.8991, |
|
"step": 119500 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.297119158705473e-05, |
|
"loss": 0.8932, |
|
"step": 120000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.2816904885334124e-05, |
|
"loss": 0.8702, |
|
"step": 120500 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.266261818361352e-05, |
|
"loss": 0.8862, |
|
"step": 121000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.2508331481892912e-05, |
|
"loss": 0.889, |
|
"step": 121500 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.2354044780172308e-05, |
|
"loss": 0.8816, |
|
"step": 122000 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"learning_rate": 1.2199758078451702e-05, |
|
"loss": 0.884, |
|
"step": 122500 |
|
}, |
|
{ |
|
"epoch": 2.28, |
|
"learning_rate": 1.2045471376731097e-05, |
|
"loss": 0.8756, |
|
"step": 123000 |
|
}, |
|
{ |
|
"epoch": 2.29, |
|
"learning_rate": 1.1891184675010492e-05, |
|
"loss": 0.9071, |
|
"step": 123500 |
|
}, |
|
{ |
|
"epoch": 2.3, |
|
"learning_rate": 1.1736897973289887e-05, |
|
"loss": 0.9127, |
|
"step": 124000 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"learning_rate": 1.1582611271569281e-05, |
|
"loss": 0.8713, |
|
"step": 124500 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"learning_rate": 1.1428324569848677e-05, |
|
"loss": 0.9016, |
|
"step": 125000 |
|
}, |
|
{ |
|
"epoch": 2.32, |
|
"learning_rate": 1.1274037868128071e-05, |
|
"loss": 0.8786, |
|
"step": 125500 |
|
}, |
|
{ |
|
"epoch": 2.33, |
|
"learning_rate": 1.1119751166407467e-05, |
|
"loss": 0.8864, |
|
"step": 126000 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 1.0965464464686861e-05, |
|
"loss": 0.8945, |
|
"step": 126500 |
|
}, |
|
{ |
|
"epoch": 2.35, |
|
"learning_rate": 1.0811177762966255e-05, |
|
"loss": 0.8962, |
|
"step": 127000 |
|
}, |
|
{ |
|
"epoch": 2.36, |
|
"learning_rate": 1.0656891061245649e-05, |
|
"loss": 0.88, |
|
"step": 127500 |
|
}, |
|
{ |
|
"epoch": 2.37, |
|
"learning_rate": 1.0502604359525043e-05, |
|
"loss": 0.8745, |
|
"step": 128000 |
|
}, |
|
{ |
|
"epoch": 2.38, |
|
"learning_rate": 1.0348317657804439e-05, |
|
"loss": 0.8744, |
|
"step": 128500 |
|
}, |
|
{ |
|
"epoch": 2.39, |
|
"learning_rate": 1.0194030956083833e-05, |
|
"loss": 0.8755, |
|
"step": 129000 |
|
}, |
|
{ |
|
"epoch": 2.4, |
|
"learning_rate": 1.0039744254363228e-05, |
|
"loss": 0.8787, |
|
"step": 129500 |
|
}, |
|
{ |
|
"epoch": 2.41, |
|
"learning_rate": 9.885457552642623e-06, |
|
"loss": 0.904, |
|
"step": 130000 |
|
}, |
|
{ |
|
"epoch": 2.42, |
|
"learning_rate": 9.731170850922018e-06, |
|
"loss": 0.8856, |
|
"step": 130500 |
|
}, |
|
{ |
|
"epoch": 2.43, |
|
"learning_rate": 9.576884149201412e-06, |
|
"loss": 0.8809, |
|
"step": 131000 |
|
}, |
|
{ |
|
"epoch": 2.43, |
|
"learning_rate": 9.422597447480808e-06, |
|
"loss": 0.8919, |
|
"step": 131500 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 9.268310745760202e-06, |
|
"loss": 0.8622, |
|
"step": 132000 |
|
}, |
|
{ |
|
"epoch": 2.45, |
|
"learning_rate": 9.114024044039598e-06, |
|
"loss": 0.8899, |
|
"step": 132500 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 8.959737342318992e-06, |
|
"loss": 0.88, |
|
"step": 133000 |
|
}, |
|
{ |
|
"epoch": 2.47, |
|
"learning_rate": 8.805450640598386e-06, |
|
"loss": 0.853, |
|
"step": 133500 |
|
}, |
|
{ |
|
"epoch": 2.48, |
|
"learning_rate": 8.65116393887778e-06, |
|
"loss": 0.8702, |
|
"step": 134000 |
|
}, |
|
{ |
|
"epoch": 2.49, |
|
"learning_rate": 8.496877237157174e-06, |
|
"loss": 0.8843, |
|
"step": 134500 |
|
}, |
|
{ |
|
"epoch": 2.5, |
|
"learning_rate": 8.34259053543657e-06, |
|
"loss": 0.8832, |
|
"step": 135000 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 8.188303833715964e-06, |
|
"loss": 0.8675, |
|
"step": 135500 |
|
}, |
|
{ |
|
"epoch": 2.52, |
|
"learning_rate": 8.03401713199536e-06, |
|
"loss": 0.8834, |
|
"step": 136000 |
|
}, |
|
{ |
|
"epoch": 2.53, |
|
"learning_rate": 7.879730430274754e-06, |
|
"loss": 0.8697, |
|
"step": 136500 |
|
}, |
|
{ |
|
"epoch": 2.54, |
|
"learning_rate": 7.72544372855415e-06, |
|
"loss": 0.9034, |
|
"step": 137000 |
|
}, |
|
{ |
|
"epoch": 2.55, |
|
"learning_rate": 7.571157026833543e-06, |
|
"loss": 0.8763, |
|
"step": 137500 |
|
}, |
|
{ |
|
"epoch": 2.55, |
|
"learning_rate": 7.416870325112939e-06, |
|
"loss": 0.892, |
|
"step": 138000 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 7.262583623392333e-06, |
|
"loss": 0.8876, |
|
"step": 138500 |
|
}, |
|
{ |
|
"epoch": 2.57, |
|
"learning_rate": 7.108296921671728e-06, |
|
"loss": 0.8453, |
|
"step": 139000 |
|
}, |
|
{ |
|
"epoch": 2.58, |
|
"learning_rate": 6.954010219951122e-06, |
|
"loss": 0.8686, |
|
"step": 139500 |
|
}, |
|
{ |
|
"epoch": 2.59, |
|
"learning_rate": 6.799723518230516e-06, |
|
"loss": 0.8658, |
|
"step": 140000 |
|
}, |
|
{ |
|
"epoch": 2.6, |
|
"learning_rate": 6.645436816509912e-06, |
|
"loss": 0.8635, |
|
"step": 140500 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 6.491150114789306e-06, |
|
"loss": 0.867, |
|
"step": 141000 |
|
}, |
|
{ |
|
"epoch": 2.62, |
|
"learning_rate": 6.3368634130687016e-06, |
|
"loss": 0.8822, |
|
"step": 141500 |
|
}, |
|
{ |
|
"epoch": 2.63, |
|
"learning_rate": 6.1825767113480965e-06, |
|
"loss": 0.8764, |
|
"step": 142000 |
|
}, |
|
{ |
|
"epoch": 2.64, |
|
"learning_rate": 6.0282900096274905e-06, |
|
"loss": 0.8694, |
|
"step": 142500 |
|
}, |
|
{ |
|
"epoch": 2.65, |
|
"learning_rate": 5.8740033079068845e-06, |
|
"loss": 0.8614, |
|
"step": 143000 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 5.7197166061862794e-06, |
|
"loss": 0.8656, |
|
"step": 143500 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 5.565429904465674e-06, |
|
"loss": 0.8768, |
|
"step": 144000 |
|
}, |
|
{ |
|
"epoch": 2.68, |
|
"learning_rate": 5.411143202745069e-06, |
|
"loss": 0.8519, |
|
"step": 144500 |
|
}, |
|
{ |
|
"epoch": 2.68, |
|
"learning_rate": 5.256856501024464e-06, |
|
"loss": 0.8661, |
|
"step": 145000 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"learning_rate": 5.102569799303859e-06, |
|
"loss": 0.8587, |
|
"step": 145500 |
|
}, |
|
{ |
|
"epoch": 2.7, |
|
"learning_rate": 4.948283097583253e-06, |
|
"loss": 0.8601, |
|
"step": 146000 |
|
}, |
|
{ |
|
"epoch": 2.71, |
|
"learning_rate": 4.793996395862648e-06, |
|
"loss": 0.8748, |
|
"step": 146500 |
|
}, |
|
{ |
|
"epoch": 2.72, |
|
"learning_rate": 4.639709694142043e-06, |
|
"loss": 0.8636, |
|
"step": 147000 |
|
}, |
|
{ |
|
"epoch": 2.73, |
|
"learning_rate": 4.485422992421438e-06, |
|
"loss": 0.8647, |
|
"step": 147500 |
|
}, |
|
{ |
|
"epoch": 2.74, |
|
"learning_rate": 4.331136290700833e-06, |
|
"loss": 0.8429, |
|
"step": 148000 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 4.1768495889802275e-06, |
|
"loss": 0.8593, |
|
"step": 148500 |
|
}, |
|
{ |
|
"epoch": 2.76, |
|
"learning_rate": 4.0225628872596215e-06, |
|
"loss": 0.8497, |
|
"step": 149000 |
|
}, |
|
{ |
|
"epoch": 2.77, |
|
"learning_rate": 3.8682761855390155e-06, |
|
"loss": 0.8777, |
|
"step": 149500 |
|
}, |
|
{ |
|
"epoch": 2.78, |
|
"learning_rate": 3.713989483818411e-06, |
|
"loss": 0.8444, |
|
"step": 150000 |
|
}, |
|
{ |
|
"epoch": 2.79, |
|
"learning_rate": 3.5597027820978053e-06, |
|
"loss": 0.8481, |
|
"step": 150500 |
|
}, |
|
{ |
|
"epoch": 2.8, |
|
"learning_rate": 3.4054160803772002e-06, |
|
"loss": 0.8742, |
|
"step": 151000 |
|
}, |
|
{ |
|
"epoch": 2.8, |
|
"learning_rate": 3.251129378656595e-06, |
|
"loss": 0.8571, |
|
"step": 151500 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 3.0968426769359896e-06, |
|
"loss": 0.8782, |
|
"step": 152000 |
|
}, |
|
{ |
|
"epoch": 2.82, |
|
"learning_rate": 2.9425559752153845e-06, |
|
"loss": 0.8631, |
|
"step": 152500 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 2.7882692734947793e-06, |
|
"loss": 0.8504, |
|
"step": 153000 |
|
}, |
|
{ |
|
"epoch": 2.84, |
|
"learning_rate": 2.633982571774174e-06, |
|
"loss": 0.8702, |
|
"step": 153500 |
|
}, |
|
{ |
|
"epoch": 2.85, |
|
"learning_rate": 2.4796958700535683e-06, |
|
"loss": 0.8679, |
|
"step": 154000 |
|
}, |
|
{ |
|
"epoch": 2.86, |
|
"learning_rate": 2.325409168332963e-06, |
|
"loss": 0.8653, |
|
"step": 154500 |
|
}, |
|
{ |
|
"epoch": 2.87, |
|
"learning_rate": 2.171122466612358e-06, |
|
"loss": 0.8652, |
|
"step": 155000 |
|
}, |
|
{ |
|
"epoch": 2.88, |
|
"learning_rate": 2.0168357648917525e-06, |
|
"loss": 0.8735, |
|
"step": 155500 |
|
}, |
|
{ |
|
"epoch": 2.89, |
|
"learning_rate": 1.8625490631711474e-06, |
|
"loss": 0.8486, |
|
"step": 156000 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 1.7082623614505419e-06, |
|
"loss": 0.8519, |
|
"step": 156500 |
|
}, |
|
{ |
|
"epoch": 2.91, |
|
"learning_rate": 1.5539756597299368e-06, |
|
"loss": 0.8392, |
|
"step": 157000 |
|
}, |
|
{ |
|
"epoch": 2.92, |
|
"learning_rate": 1.3996889580093312e-06, |
|
"loss": 0.8534, |
|
"step": 157500 |
|
}, |
|
{ |
|
"epoch": 2.93, |
|
"learning_rate": 1.2454022562887261e-06, |
|
"loss": 0.8517, |
|
"step": 158000 |
|
}, |
|
{ |
|
"epoch": 2.93, |
|
"learning_rate": 1.0911155545681208e-06, |
|
"loss": 0.8441, |
|
"step": 158500 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"learning_rate": 9.368288528475154e-07, |
|
"loss": 0.8462, |
|
"step": 159000 |
|
}, |
|
{ |
|
"epoch": 2.95, |
|
"learning_rate": 7.825421511269101e-07, |
|
"loss": 0.8337, |
|
"step": 159500 |
|
}, |
|
{ |
|
"epoch": 2.96, |
|
"learning_rate": 6.282554494063048e-07, |
|
"loss": 0.8716, |
|
"step": 160000 |
|
}, |
|
{ |
|
"epoch": 2.97, |
|
"learning_rate": 4.7396874768569954e-07, |
|
"loss": 0.8624, |
|
"step": 160500 |
|
}, |
|
{ |
|
"epoch": 2.98, |
|
"learning_rate": 3.196820459650942e-07, |
|
"loss": 0.851, |
|
"step": 161000 |
|
}, |
|
{ |
|
"epoch": 2.99, |
|
"learning_rate": 1.653953442444889e-07, |
|
"loss": 0.8528, |
|
"step": 161500 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 1.1108642523883582e-08, |
|
"loss": 0.8461, |
|
"step": 162000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 162036, |
|
"total_flos": 7.352562727098778e+16, |
|
"train_loss": 1.1144006177954062, |
|
"train_runtime": 31409.6299, |
|
"train_samples_per_second": 20.635, |
|
"train_steps_per_second": 5.159 |
|
} |
|
], |
|
"max_steps": 162036, |
|
"num_train_epochs": 3, |
|
"total_flos": 7.352562727098778e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|