Davlan's picture
Upload trainer_state.json
0bf7cd3
raw
history blame contribute delete
No virus
35.1 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9999893658878953,
"global_step": 141054,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.982276291349412e-05,
"loss": 2.0436,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 4.964552582698825e-05,
"loss": 2.0643,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 4.946828874048237e-05,
"loss": 2.0537,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 4.929105165397649e-05,
"loss": 2.0425,
"step": 2000
},
{
"epoch": 0.05,
"learning_rate": 4.911381456747062e-05,
"loss": 2.0306,
"step": 2500
},
{
"epoch": 0.06,
"learning_rate": 4.893657748096474e-05,
"loss": 2.0203,
"step": 3000
},
{
"epoch": 0.07,
"learning_rate": 4.875934039445886e-05,
"loss": 2.0241,
"step": 3500
},
{
"epoch": 0.09,
"learning_rate": 4.858210330795299e-05,
"loss": 2.0036,
"step": 4000
},
{
"epoch": 0.1,
"learning_rate": 4.840486622144711e-05,
"loss": 2.0077,
"step": 4500
},
{
"epoch": 0.11,
"learning_rate": 4.822762913494123e-05,
"loss": 1.9801,
"step": 5000
},
{
"epoch": 0.12,
"learning_rate": 4.805039204843536e-05,
"loss": 1.9798,
"step": 5500
},
{
"epoch": 0.13,
"learning_rate": 4.787315496192948e-05,
"loss": 1.97,
"step": 6000
},
{
"epoch": 0.14,
"learning_rate": 4.76959178754236e-05,
"loss": 1.9671,
"step": 6500
},
{
"epoch": 0.15,
"learning_rate": 4.751868078891773e-05,
"loss": 1.9642,
"step": 7000
},
{
"epoch": 0.16,
"learning_rate": 4.734144370241184e-05,
"loss": 1.9536,
"step": 7500
},
{
"epoch": 0.17,
"learning_rate": 4.716420661590596e-05,
"loss": 1.9517,
"step": 8000
},
{
"epoch": 0.18,
"learning_rate": 4.698696952940009e-05,
"loss": 1.9323,
"step": 8500
},
{
"epoch": 0.19,
"learning_rate": 4.680973244289421e-05,
"loss": 1.9254,
"step": 9000
},
{
"epoch": 0.2,
"learning_rate": 4.663249535638833e-05,
"loss": 1.9289,
"step": 9500
},
{
"epoch": 0.21,
"learning_rate": 4.645525826988246e-05,
"loss": 1.9323,
"step": 10000
},
{
"epoch": 0.22,
"learning_rate": 4.627802118337658e-05,
"loss": 1.9083,
"step": 10500
},
{
"epoch": 0.23,
"learning_rate": 4.61007840968707e-05,
"loss": 1.9122,
"step": 11000
},
{
"epoch": 0.24,
"learning_rate": 4.592354701036483e-05,
"loss": 1.906,
"step": 11500
},
{
"epoch": 0.26,
"learning_rate": 4.574630992385895e-05,
"loss": 1.8886,
"step": 12000
},
{
"epoch": 0.27,
"learning_rate": 4.556907283735307e-05,
"loss": 1.8898,
"step": 12500
},
{
"epoch": 0.28,
"learning_rate": 4.5391835750847196e-05,
"loss": 1.8977,
"step": 13000
},
{
"epoch": 0.29,
"learning_rate": 4.521459866434132e-05,
"loss": 1.8815,
"step": 13500
},
{
"epoch": 0.3,
"learning_rate": 4.503736157783544e-05,
"loss": 1.862,
"step": 14000
},
{
"epoch": 0.31,
"learning_rate": 4.4860124491329565e-05,
"loss": 1.8589,
"step": 14500
},
{
"epoch": 0.32,
"learning_rate": 4.4682887404823686e-05,
"loss": 1.8701,
"step": 15000
},
{
"epoch": 0.33,
"learning_rate": 4.450565031831781e-05,
"loss": 1.8786,
"step": 15500
},
{
"epoch": 0.34,
"learning_rate": 4.4328413231811934e-05,
"loss": 1.8626,
"step": 16000
},
{
"epoch": 0.35,
"learning_rate": 4.4151176145306055e-05,
"loss": 1.8633,
"step": 16500
},
{
"epoch": 0.36,
"learning_rate": 4.3973939058800176e-05,
"loss": 1.8495,
"step": 17000
},
{
"epoch": 0.37,
"learning_rate": 4.3796701972294304e-05,
"loss": 1.8559,
"step": 17500
},
{
"epoch": 0.38,
"learning_rate": 4.3619464885788424e-05,
"loss": 1.8567,
"step": 18000
},
{
"epoch": 0.39,
"learning_rate": 4.3442227799282545e-05,
"loss": 1.8484,
"step": 18500
},
{
"epoch": 0.4,
"learning_rate": 4.326499071277667e-05,
"loss": 1.8423,
"step": 19000
},
{
"epoch": 0.41,
"learning_rate": 4.308775362627079e-05,
"loss": 1.8311,
"step": 19500
},
{
"epoch": 0.43,
"learning_rate": 4.2910516539764914e-05,
"loss": 1.8396,
"step": 20000
},
{
"epoch": 0.44,
"learning_rate": 4.273327945325904e-05,
"loss": 1.8335,
"step": 20500
},
{
"epoch": 0.45,
"learning_rate": 4.255604236675316e-05,
"loss": 1.8276,
"step": 21000
},
{
"epoch": 0.46,
"learning_rate": 4.237880528024728e-05,
"loss": 1.8172,
"step": 21500
},
{
"epoch": 0.47,
"learning_rate": 4.220156819374141e-05,
"loss": 1.8305,
"step": 22000
},
{
"epoch": 0.48,
"learning_rate": 4.2024331107235525e-05,
"loss": 1.8229,
"step": 22500
},
{
"epoch": 0.49,
"learning_rate": 4.184709402072965e-05,
"loss": 1.8122,
"step": 23000
},
{
"epoch": 0.5,
"learning_rate": 4.166985693422377e-05,
"loss": 1.8046,
"step": 23500
},
{
"epoch": 0.51,
"learning_rate": 4.1492619847717894e-05,
"loss": 1.8239,
"step": 24000
},
{
"epoch": 0.52,
"learning_rate": 4.131538276121202e-05,
"loss": 1.8021,
"step": 24500
},
{
"epoch": 0.53,
"learning_rate": 4.113814567470614e-05,
"loss": 1.7977,
"step": 25000
},
{
"epoch": 0.54,
"learning_rate": 4.096090858820026e-05,
"loss": 1.7961,
"step": 25500
},
{
"epoch": 0.55,
"learning_rate": 4.078367150169439e-05,
"loss": 1.8011,
"step": 26000
},
{
"epoch": 0.56,
"learning_rate": 4.060643441518851e-05,
"loss": 1.804,
"step": 26500
},
{
"epoch": 0.57,
"learning_rate": 4.042919732868263e-05,
"loss": 1.7952,
"step": 27000
},
{
"epoch": 0.58,
"learning_rate": 4.025196024217676e-05,
"loss": 1.8001,
"step": 27500
},
{
"epoch": 0.6,
"learning_rate": 4.007472315567088e-05,
"loss": 1.7974,
"step": 28000
},
{
"epoch": 0.61,
"learning_rate": 3.9897486069165e-05,
"loss": 1.788,
"step": 28500
},
{
"epoch": 0.62,
"learning_rate": 3.972024898265912e-05,
"loss": 1.7909,
"step": 29000
},
{
"epoch": 0.63,
"learning_rate": 3.954301189615325e-05,
"loss": 1.7823,
"step": 29500
},
{
"epoch": 0.64,
"learning_rate": 3.936577480964737e-05,
"loss": 1.78,
"step": 30000
},
{
"epoch": 0.65,
"learning_rate": 3.918853772314149e-05,
"loss": 1.7798,
"step": 30500
},
{
"epoch": 0.66,
"learning_rate": 3.901130063663562e-05,
"loss": 1.7754,
"step": 31000
},
{
"epoch": 0.67,
"learning_rate": 3.883406355012974e-05,
"loss": 1.7721,
"step": 31500
},
{
"epoch": 0.68,
"learning_rate": 3.865682646362386e-05,
"loss": 1.7589,
"step": 32000
},
{
"epoch": 0.69,
"learning_rate": 3.847958937711799e-05,
"loss": 1.7657,
"step": 32500
},
{
"epoch": 0.7,
"learning_rate": 3.830235229061211e-05,
"loss": 1.7759,
"step": 33000
},
{
"epoch": 0.71,
"learning_rate": 3.812511520410623e-05,
"loss": 1.754,
"step": 33500
},
{
"epoch": 0.72,
"learning_rate": 3.7947878117600356e-05,
"loss": 1.7579,
"step": 34000
},
{
"epoch": 0.73,
"learning_rate": 3.777064103109448e-05,
"loss": 1.757,
"step": 34500
},
{
"epoch": 0.74,
"learning_rate": 3.75934039445886e-05,
"loss": 1.7523,
"step": 35000
},
{
"epoch": 0.76,
"learning_rate": 3.7416166858082725e-05,
"loss": 1.756,
"step": 35500
},
{
"epoch": 0.77,
"learning_rate": 3.7238929771576846e-05,
"loss": 1.7364,
"step": 36000
},
{
"epoch": 0.78,
"learning_rate": 3.706169268507097e-05,
"loss": 1.7498,
"step": 36500
},
{
"epoch": 0.79,
"learning_rate": 3.6884455598565095e-05,
"loss": 1.7457,
"step": 37000
},
{
"epoch": 0.8,
"learning_rate": 3.670721851205921e-05,
"loss": 1.739,
"step": 37500
},
{
"epoch": 0.81,
"learning_rate": 3.6529981425553336e-05,
"loss": 1.7246,
"step": 38000
},
{
"epoch": 0.82,
"learning_rate": 3.635274433904746e-05,
"loss": 1.7457,
"step": 38500
},
{
"epoch": 0.83,
"learning_rate": 3.617550725254158e-05,
"loss": 1.7435,
"step": 39000
},
{
"epoch": 0.84,
"learning_rate": 3.5998270166035705e-05,
"loss": 1.7304,
"step": 39500
},
{
"epoch": 0.85,
"learning_rate": 3.5821033079529826e-05,
"loss": 1.7383,
"step": 40000
},
{
"epoch": 0.86,
"learning_rate": 3.564379599302395e-05,
"loss": 1.7259,
"step": 40500
},
{
"epoch": 0.87,
"learning_rate": 3.5466558906518074e-05,
"loss": 1.711,
"step": 41000
},
{
"epoch": 0.88,
"learning_rate": 3.5289321820012195e-05,
"loss": 1.7196,
"step": 41500
},
{
"epoch": 0.89,
"learning_rate": 3.5112084733506316e-05,
"loss": 1.7187,
"step": 42000
},
{
"epoch": 0.9,
"learning_rate": 3.493484764700044e-05,
"loss": 1.7163,
"step": 42500
},
{
"epoch": 0.91,
"learning_rate": 3.4757610560494564e-05,
"loss": 1.7087,
"step": 43000
},
{
"epoch": 0.93,
"learning_rate": 3.4580373473988685e-05,
"loss": 1.7098,
"step": 43500
},
{
"epoch": 0.94,
"learning_rate": 3.440313638748281e-05,
"loss": 1.7014,
"step": 44000
},
{
"epoch": 0.95,
"learning_rate": 3.422589930097693e-05,
"loss": 1.7167,
"step": 44500
},
{
"epoch": 0.96,
"learning_rate": 3.4048662214471054e-05,
"loss": 1.7057,
"step": 45000
},
{
"epoch": 0.97,
"learning_rate": 3.387142512796518e-05,
"loss": 1.7069,
"step": 45500
},
{
"epoch": 0.98,
"learning_rate": 3.36941880414593e-05,
"loss": 1.6955,
"step": 46000
},
{
"epoch": 0.99,
"learning_rate": 3.351695095495342e-05,
"loss": 1.7014,
"step": 46500
},
{
"epoch": 1.0,
"learning_rate": 3.333971386844755e-05,
"loss": 1.6953,
"step": 47000
},
{
"epoch": 1.01,
"learning_rate": 3.316247678194167e-05,
"loss": 1.6936,
"step": 47500
},
{
"epoch": 1.02,
"learning_rate": 3.298523969543579e-05,
"loss": 1.6826,
"step": 48000
},
{
"epoch": 1.03,
"learning_rate": 3.280800260892992e-05,
"loss": 1.6808,
"step": 48500
},
{
"epoch": 1.04,
"learning_rate": 3.263076552242404e-05,
"loss": 1.6764,
"step": 49000
},
{
"epoch": 1.05,
"learning_rate": 3.245352843591816e-05,
"loss": 1.6748,
"step": 49500
},
{
"epoch": 1.06,
"learning_rate": 3.227629134941229e-05,
"loss": 1.6665,
"step": 50000
},
{
"epoch": 1.07,
"learning_rate": 3.209905426290641e-05,
"loss": 1.6758,
"step": 50500
},
{
"epoch": 1.08,
"learning_rate": 3.192181717640053e-05,
"loss": 1.6733,
"step": 51000
},
{
"epoch": 1.1,
"learning_rate": 3.174458008989466e-05,
"loss": 1.6694,
"step": 51500
},
{
"epoch": 1.11,
"learning_rate": 3.156734300338878e-05,
"loss": 1.6604,
"step": 52000
},
{
"epoch": 1.12,
"learning_rate": 3.139010591688289e-05,
"loss": 1.6634,
"step": 52500
},
{
"epoch": 1.13,
"learning_rate": 3.121286883037702e-05,
"loss": 1.6636,
"step": 53000
},
{
"epoch": 1.14,
"learning_rate": 3.103563174387114e-05,
"loss": 1.6613,
"step": 53500
},
{
"epoch": 1.15,
"learning_rate": 3.085839465736526e-05,
"loss": 1.6623,
"step": 54000
},
{
"epoch": 1.16,
"learning_rate": 3.068115757085939e-05,
"loss": 1.6722,
"step": 54500
},
{
"epoch": 1.17,
"learning_rate": 3.050392048435351e-05,
"loss": 1.6587,
"step": 55000
},
{
"epoch": 1.18,
"learning_rate": 3.0326683397847634e-05,
"loss": 1.6717,
"step": 55500
},
{
"epoch": 1.19,
"learning_rate": 3.0149446311341755e-05,
"loss": 1.6598,
"step": 56000
},
{
"epoch": 1.2,
"learning_rate": 2.997220922483588e-05,
"loss": 1.6506,
"step": 56500
},
{
"epoch": 1.21,
"learning_rate": 2.9794972138330003e-05,
"loss": 1.6472,
"step": 57000
},
{
"epoch": 1.22,
"learning_rate": 2.9617735051824124e-05,
"loss": 1.6674,
"step": 57500
},
{
"epoch": 1.23,
"learning_rate": 2.9440497965318248e-05,
"loss": 1.6408,
"step": 58000
},
{
"epoch": 1.24,
"learning_rate": 2.9263260878812372e-05,
"loss": 1.6329,
"step": 58500
},
{
"epoch": 1.25,
"learning_rate": 2.9086023792306493e-05,
"loss": 1.6562,
"step": 59000
},
{
"epoch": 1.27,
"learning_rate": 2.8908786705800617e-05,
"loss": 1.634,
"step": 59500
},
{
"epoch": 1.28,
"learning_rate": 2.873154961929474e-05,
"loss": 1.6334,
"step": 60000
},
{
"epoch": 1.29,
"learning_rate": 2.8554312532788862e-05,
"loss": 1.6417,
"step": 60500
},
{
"epoch": 1.3,
"learning_rate": 2.8377075446282986e-05,
"loss": 1.6395,
"step": 61000
},
{
"epoch": 1.31,
"learning_rate": 2.819983835977711e-05,
"loss": 1.624,
"step": 61500
},
{
"epoch": 1.32,
"learning_rate": 2.802260127327123e-05,
"loss": 1.6343,
"step": 62000
},
{
"epoch": 1.33,
"learning_rate": 2.7845364186765355e-05,
"loss": 1.6215,
"step": 62500
},
{
"epoch": 1.34,
"learning_rate": 2.766812710025948e-05,
"loss": 1.6286,
"step": 63000
},
{
"epoch": 1.35,
"learning_rate": 2.74908900137536e-05,
"loss": 1.6241,
"step": 63500
},
{
"epoch": 1.36,
"learning_rate": 2.7313652927247724e-05,
"loss": 1.6174,
"step": 64000
},
{
"epoch": 1.37,
"learning_rate": 2.7136415840741848e-05,
"loss": 1.6319,
"step": 64500
},
{
"epoch": 1.38,
"learning_rate": 2.695917875423597e-05,
"loss": 1.631,
"step": 65000
},
{
"epoch": 1.39,
"learning_rate": 2.6781941667730093e-05,
"loss": 1.6171,
"step": 65500
},
{
"epoch": 1.4,
"learning_rate": 2.6604704581224217e-05,
"loss": 1.6285,
"step": 66000
},
{
"epoch": 1.41,
"learning_rate": 2.6427467494718338e-05,
"loss": 1.6145,
"step": 66500
},
{
"epoch": 1.42,
"learning_rate": 2.6250230408212455e-05,
"loss": 1.6179,
"step": 67000
},
{
"epoch": 1.44,
"learning_rate": 2.607299332170658e-05,
"loss": 1.6129,
"step": 67500
},
{
"epoch": 1.45,
"learning_rate": 2.5895756235200704e-05,
"loss": 1.6064,
"step": 68000
},
{
"epoch": 1.46,
"learning_rate": 2.5718519148694824e-05,
"loss": 1.6021,
"step": 68500
},
{
"epoch": 1.47,
"learning_rate": 2.554128206218895e-05,
"loss": 1.6124,
"step": 69000
},
{
"epoch": 1.48,
"learning_rate": 2.5364044975683073e-05,
"loss": 1.5979,
"step": 69500
},
{
"epoch": 1.49,
"learning_rate": 2.5186807889177193e-05,
"loss": 1.5919,
"step": 70000
},
{
"epoch": 1.5,
"learning_rate": 2.5009570802671318e-05,
"loss": 1.5978,
"step": 70500
},
{
"epoch": 1.51,
"learning_rate": 2.4832333716165442e-05,
"loss": 1.6017,
"step": 71000
},
{
"epoch": 1.52,
"learning_rate": 2.4655096629659563e-05,
"loss": 1.5942,
"step": 71500
},
{
"epoch": 1.53,
"learning_rate": 2.4477859543153687e-05,
"loss": 1.593,
"step": 72000
},
{
"epoch": 1.54,
"learning_rate": 2.4300622456647807e-05,
"loss": 1.5917,
"step": 72500
},
{
"epoch": 1.55,
"learning_rate": 2.412338537014193e-05,
"loss": 1.601,
"step": 73000
},
{
"epoch": 1.56,
"learning_rate": 2.3946148283636056e-05,
"loss": 1.5954,
"step": 73500
},
{
"epoch": 1.57,
"learning_rate": 2.3768911197130177e-05,
"loss": 1.6106,
"step": 74000
},
{
"epoch": 1.58,
"learning_rate": 2.35916741106243e-05,
"loss": 1.5943,
"step": 74500
},
{
"epoch": 1.6,
"learning_rate": 2.3414437024118425e-05,
"loss": 1.5948,
"step": 75000
},
{
"epoch": 1.61,
"learning_rate": 2.3237199937612546e-05,
"loss": 1.5931,
"step": 75500
},
{
"epoch": 1.62,
"learning_rate": 2.305996285110667e-05,
"loss": 1.5947,
"step": 76000
},
{
"epoch": 1.63,
"learning_rate": 2.2882725764600794e-05,
"loss": 1.5955,
"step": 76500
},
{
"epoch": 1.64,
"learning_rate": 2.2705488678094915e-05,
"loss": 1.5811,
"step": 77000
},
{
"epoch": 1.65,
"learning_rate": 2.252825159158904e-05,
"loss": 1.5752,
"step": 77500
},
{
"epoch": 1.66,
"learning_rate": 2.2351014505083163e-05,
"loss": 1.5704,
"step": 78000
},
{
"epoch": 1.67,
"learning_rate": 2.2173777418577284e-05,
"loss": 1.5792,
"step": 78500
},
{
"epoch": 1.68,
"learning_rate": 2.1996540332071404e-05,
"loss": 1.575,
"step": 79000
},
{
"epoch": 1.69,
"learning_rate": 2.181930324556553e-05,
"loss": 1.5668,
"step": 79500
},
{
"epoch": 1.7,
"learning_rate": 2.1642066159059653e-05,
"loss": 1.569,
"step": 80000
},
{
"epoch": 1.71,
"learning_rate": 2.1464829072553774e-05,
"loss": 1.5741,
"step": 80500
},
{
"epoch": 1.72,
"learning_rate": 2.1287591986047898e-05,
"loss": 1.5781,
"step": 81000
},
{
"epoch": 1.73,
"learning_rate": 2.1110354899542022e-05,
"loss": 1.5666,
"step": 81500
},
{
"epoch": 1.74,
"learning_rate": 2.0933117813036143e-05,
"loss": 1.5685,
"step": 82000
},
{
"epoch": 1.75,
"learning_rate": 2.0755880726530267e-05,
"loss": 1.5557,
"step": 82500
},
{
"epoch": 1.77,
"learning_rate": 2.057864364002439e-05,
"loss": 1.5706,
"step": 83000
},
{
"epoch": 1.78,
"learning_rate": 2.040140655351851e-05,
"loss": 1.5627,
"step": 83500
},
{
"epoch": 1.79,
"learning_rate": 2.0224169467012636e-05,
"loss": 1.5706,
"step": 84000
},
{
"epoch": 1.8,
"learning_rate": 2.004693238050676e-05,
"loss": 1.557,
"step": 84500
},
{
"epoch": 1.81,
"learning_rate": 1.986969529400088e-05,
"loss": 1.5696,
"step": 85000
},
{
"epoch": 1.82,
"learning_rate": 1.9692458207495e-05,
"loss": 1.5635,
"step": 85500
},
{
"epoch": 1.83,
"learning_rate": 1.9515221120989126e-05,
"loss": 1.5541,
"step": 86000
},
{
"epoch": 1.84,
"learning_rate": 1.9337984034483246e-05,
"loss": 1.5651,
"step": 86500
},
{
"epoch": 1.85,
"learning_rate": 1.916074694797737e-05,
"loss": 1.5499,
"step": 87000
},
{
"epoch": 1.86,
"learning_rate": 1.8983509861471495e-05,
"loss": 1.5538,
"step": 87500
},
{
"epoch": 1.87,
"learning_rate": 1.8806272774965615e-05,
"loss": 1.5566,
"step": 88000
},
{
"epoch": 1.88,
"learning_rate": 1.862903568845974e-05,
"loss": 1.5582,
"step": 88500
},
{
"epoch": 1.89,
"learning_rate": 1.8451798601953864e-05,
"loss": 1.5361,
"step": 89000
},
{
"epoch": 1.9,
"learning_rate": 1.8274561515447984e-05,
"loss": 1.5418,
"step": 89500
},
{
"epoch": 1.91,
"learning_rate": 1.809732442894211e-05,
"loss": 1.548,
"step": 90000
},
{
"epoch": 1.92,
"learning_rate": 1.7920087342436233e-05,
"loss": 1.5393,
"step": 90500
},
{
"epoch": 1.94,
"learning_rate": 1.7742850255930354e-05,
"loss": 1.5325,
"step": 91000
},
{
"epoch": 1.95,
"learning_rate": 1.7565613169424478e-05,
"loss": 1.5491,
"step": 91500
},
{
"epoch": 1.96,
"learning_rate": 1.7388376082918602e-05,
"loss": 1.5407,
"step": 92000
},
{
"epoch": 1.97,
"learning_rate": 1.7211138996412723e-05,
"loss": 1.5403,
"step": 92500
},
{
"epoch": 1.98,
"learning_rate": 1.7033901909906843e-05,
"loss": 1.542,
"step": 93000
},
{
"epoch": 1.99,
"learning_rate": 1.6856664823400967e-05,
"loss": 1.5457,
"step": 93500
},
{
"epoch": 2.0,
"learning_rate": 1.6679427736895088e-05,
"loss": 1.5323,
"step": 94000
},
{
"epoch": 2.01,
"learning_rate": 1.6502190650389212e-05,
"loss": 1.5367,
"step": 94500
},
{
"epoch": 2.02,
"learning_rate": 1.6324953563883337e-05,
"loss": 1.5148,
"step": 95000
},
{
"epoch": 2.03,
"learning_rate": 1.6147716477377457e-05,
"loss": 1.5193,
"step": 95500
},
{
"epoch": 2.04,
"learning_rate": 1.597047939087158e-05,
"loss": 1.5195,
"step": 96000
},
{
"epoch": 2.05,
"learning_rate": 1.5793242304365706e-05,
"loss": 1.5254,
"step": 96500
},
{
"epoch": 2.06,
"learning_rate": 1.5616005217859826e-05,
"loss": 1.5246,
"step": 97000
},
{
"epoch": 2.07,
"learning_rate": 1.543876813135395e-05,
"loss": 1.5204,
"step": 97500
},
{
"epoch": 2.08,
"learning_rate": 1.5261531044848075e-05,
"loss": 1.5094,
"step": 98000
},
{
"epoch": 2.09,
"learning_rate": 1.5084293958342197e-05,
"loss": 1.509,
"step": 98500
},
{
"epoch": 2.11,
"learning_rate": 1.490705687183632e-05,
"loss": 1.513,
"step": 99000
},
{
"epoch": 2.12,
"learning_rate": 1.4729819785330442e-05,
"loss": 1.5191,
"step": 99500
},
{
"epoch": 2.13,
"learning_rate": 1.4552582698824566e-05,
"loss": 1.5117,
"step": 100000
},
{
"epoch": 2.14,
"learning_rate": 1.4375345612318685e-05,
"loss": 1.5069,
"step": 100500
},
{
"epoch": 2.15,
"learning_rate": 1.419810852581281e-05,
"loss": 1.5158,
"step": 101000
},
{
"epoch": 2.16,
"learning_rate": 1.4020871439306932e-05,
"loss": 1.5108,
"step": 101500
},
{
"epoch": 2.17,
"learning_rate": 1.3843634352801054e-05,
"loss": 1.5124,
"step": 102000
},
{
"epoch": 2.18,
"learning_rate": 1.3666397266295178e-05,
"loss": 1.5056,
"step": 102500
},
{
"epoch": 2.19,
"learning_rate": 1.3489160179789301e-05,
"loss": 1.4889,
"step": 103000
},
{
"epoch": 2.2,
"learning_rate": 1.3311923093283423e-05,
"loss": 1.5026,
"step": 103500
},
{
"epoch": 2.21,
"learning_rate": 1.3134686006777548e-05,
"loss": 1.49,
"step": 104000
},
{
"epoch": 2.22,
"learning_rate": 1.295744892027167e-05,
"loss": 1.489,
"step": 104500
},
{
"epoch": 2.23,
"learning_rate": 1.2780211833765792e-05,
"loss": 1.4951,
"step": 105000
},
{
"epoch": 2.24,
"learning_rate": 1.2602974747259917e-05,
"loss": 1.5005,
"step": 105500
},
{
"epoch": 2.25,
"learning_rate": 1.2425737660754037e-05,
"loss": 1.4931,
"step": 106000
},
{
"epoch": 2.27,
"learning_rate": 1.224850057424816e-05,
"loss": 1.4855,
"step": 106500
},
{
"epoch": 2.28,
"learning_rate": 1.2071263487742284e-05,
"loss": 1.4974,
"step": 107000
},
{
"epoch": 2.29,
"learning_rate": 1.1894026401236406e-05,
"loss": 1.4875,
"step": 107500
},
{
"epoch": 2.3,
"learning_rate": 1.1716789314730529e-05,
"loss": 1.4956,
"step": 108000
},
{
"epoch": 2.31,
"learning_rate": 1.1539552228224653e-05,
"loss": 1.4903,
"step": 108500
},
{
"epoch": 2.32,
"learning_rate": 1.1362315141718775e-05,
"loss": 1.4907,
"step": 109000
},
{
"epoch": 2.33,
"learning_rate": 1.1185078055212898e-05,
"loss": 1.4926,
"step": 109500
},
{
"epoch": 2.34,
"learning_rate": 1.100784096870702e-05,
"loss": 1.497,
"step": 110000
},
{
"epoch": 2.35,
"learning_rate": 1.0830603882201143e-05,
"loss": 1.4787,
"step": 110500
},
{
"epoch": 2.36,
"learning_rate": 1.0653366795695265e-05,
"loss": 1.4861,
"step": 111000
},
{
"epoch": 2.37,
"learning_rate": 1.047612970918939e-05,
"loss": 1.4778,
"step": 111500
},
{
"epoch": 2.38,
"learning_rate": 1.0298892622683512e-05,
"loss": 1.4771,
"step": 112000
},
{
"epoch": 2.39,
"learning_rate": 1.0121655536177634e-05,
"loss": 1.4742,
"step": 112500
},
{
"epoch": 2.4,
"learning_rate": 9.944418449671758e-06,
"loss": 1.474,
"step": 113000
},
{
"epoch": 2.41,
"learning_rate": 9.76718136316588e-06,
"loss": 1.4711,
"step": 113500
},
{
"epoch": 2.42,
"learning_rate": 9.589944276660003e-06,
"loss": 1.4739,
"step": 114000
},
{
"epoch": 2.44,
"learning_rate": 9.412707190154126e-06,
"loss": 1.4801,
"step": 114500
},
{
"epoch": 2.45,
"learning_rate": 9.235470103648248e-06,
"loss": 1.483,
"step": 115000
},
{
"epoch": 2.46,
"learning_rate": 9.05823301714237e-06,
"loss": 1.4634,
"step": 115500
},
{
"epoch": 2.47,
"learning_rate": 8.880995930636495e-06,
"loss": 1.4659,
"step": 116000
},
{
"epoch": 2.48,
"learning_rate": 8.703758844130617e-06,
"loss": 1.4737,
"step": 116500
},
{
"epoch": 2.49,
"learning_rate": 8.52652175762474e-06,
"loss": 1.4677,
"step": 117000
},
{
"epoch": 2.5,
"learning_rate": 8.349284671118862e-06,
"loss": 1.4774,
"step": 117500
},
{
"epoch": 2.51,
"learning_rate": 8.172047584612985e-06,
"loss": 1.4721,
"step": 118000
},
{
"epoch": 2.52,
"learning_rate": 7.994810498107109e-06,
"loss": 1.4608,
"step": 118500
},
{
"epoch": 2.53,
"learning_rate": 7.817573411601231e-06,
"loss": 1.4774,
"step": 119000
},
{
"epoch": 2.54,
"learning_rate": 7.640336325095354e-06,
"loss": 1.4637,
"step": 119500
},
{
"epoch": 2.55,
"learning_rate": 7.463099238589477e-06,
"loss": 1.471,
"step": 120000
},
{
"epoch": 2.56,
"learning_rate": 7.2858621520836e-06,
"loss": 1.4588,
"step": 120500
},
{
"epoch": 2.57,
"learning_rate": 7.108625065577722e-06,
"loss": 1.4634,
"step": 121000
},
{
"epoch": 2.58,
"learning_rate": 6.9313879790718444e-06,
"loss": 1.4518,
"step": 121500
},
{
"epoch": 2.59,
"learning_rate": 6.754150892565968e-06,
"loss": 1.4592,
"step": 122000
},
{
"epoch": 2.61,
"learning_rate": 6.576913806060091e-06,
"loss": 1.4546,
"step": 122500
},
{
"epoch": 2.62,
"learning_rate": 6.3996767195542135e-06,
"loss": 1.4585,
"step": 123000
},
{
"epoch": 2.63,
"learning_rate": 6.222439633048337e-06,
"loss": 1.4586,
"step": 123500
},
{
"epoch": 2.64,
"learning_rate": 6.045202546542459e-06,
"loss": 1.4446,
"step": 124000
},
{
"epoch": 2.65,
"learning_rate": 5.867965460036582e-06,
"loss": 1.4616,
"step": 124500
},
{
"epoch": 2.66,
"learning_rate": 5.690728373530705e-06,
"loss": 1.4522,
"step": 125000
},
{
"epoch": 2.67,
"learning_rate": 5.5134912870248275e-06,
"loss": 1.4471,
"step": 125500
},
{
"epoch": 2.68,
"learning_rate": 5.33625420051895e-06,
"loss": 1.445,
"step": 126000
},
{
"epoch": 2.69,
"learning_rate": 5.159017114013073e-06,
"loss": 1.4569,
"step": 126500
},
{
"epoch": 2.7,
"learning_rate": 4.9817800275071965e-06,
"loss": 1.4547,
"step": 127000
},
{
"epoch": 2.71,
"learning_rate": 4.804542941001319e-06,
"loss": 1.4406,
"step": 127500
},
{
"epoch": 2.72,
"learning_rate": 4.6273058544954414e-06,
"loss": 1.4471,
"step": 128000
},
{
"epoch": 2.73,
"learning_rate": 4.450068767989565e-06,
"loss": 1.4512,
"step": 128500
},
{
"epoch": 2.74,
"learning_rate": 4.272831681483687e-06,
"loss": 1.4474,
"step": 129000
},
{
"epoch": 2.75,
"learning_rate": 4.09559459497781e-06,
"loss": 1.4383,
"step": 129500
},
{
"epoch": 2.76,
"learning_rate": 3.918357508471933e-06,
"loss": 1.4401,
"step": 130000
},
{
"epoch": 2.78,
"learning_rate": 3.741120421966056e-06,
"loss": 1.4321,
"step": 130500
},
{
"epoch": 2.79,
"learning_rate": 3.5638833354601787e-06,
"loss": 1.4491,
"step": 131000
},
{
"epoch": 2.8,
"learning_rate": 3.386646248954301e-06,
"loss": 1.447,
"step": 131500
},
{
"epoch": 2.81,
"learning_rate": 3.209409162448424e-06,
"loss": 1.4563,
"step": 132000
},
{
"epoch": 2.82,
"learning_rate": 3.032172075942547e-06,
"loss": 1.4401,
"step": 132500
},
{
"epoch": 2.83,
"learning_rate": 2.8549349894366698e-06,
"loss": 1.4321,
"step": 133000
},
{
"epoch": 2.84,
"learning_rate": 2.6776979029307927e-06,
"loss": 1.4395,
"step": 133500
},
{
"epoch": 2.85,
"learning_rate": 2.5004608164249155e-06,
"loss": 1.4371,
"step": 134000
},
{
"epoch": 2.86,
"learning_rate": 2.323223729919038e-06,
"loss": 1.4242,
"step": 134500
},
{
"epoch": 2.87,
"learning_rate": 2.1459866434131613e-06,
"loss": 1.4374,
"step": 135000
},
{
"epoch": 2.88,
"learning_rate": 1.9687495569072838e-06,
"loss": 1.4228,
"step": 135500
},
{
"epoch": 2.89,
"learning_rate": 1.7915124704014068e-06,
"loss": 1.438,
"step": 136000
},
{
"epoch": 2.9,
"learning_rate": 1.6142753838955295e-06,
"loss": 1.438,
"step": 136500
},
{
"epoch": 2.91,
"learning_rate": 1.4370382973896522e-06,
"loss": 1.4338,
"step": 137000
},
{
"epoch": 2.92,
"learning_rate": 1.259801210883775e-06,
"loss": 1.4321,
"step": 137500
},
{
"epoch": 2.94,
"learning_rate": 1.082564124377898e-06,
"loss": 1.4252,
"step": 138000
},
{
"epoch": 2.95,
"learning_rate": 9.053270378720207e-07,
"loss": 1.4386,
"step": 138500
},
{
"epoch": 2.96,
"learning_rate": 7.280899513661436e-07,
"loss": 1.4353,
"step": 139000
},
{
"epoch": 2.97,
"learning_rate": 5.508528648602662e-07,
"loss": 1.4376,
"step": 139500
},
{
"epoch": 2.98,
"learning_rate": 3.736157783543891e-07,
"loss": 1.4459,
"step": 140000
},
{
"epoch": 2.99,
"learning_rate": 1.9637869184851192e-07,
"loss": 1.4323,
"step": 140500
},
{
"epoch": 3.0,
"learning_rate": 1.9141605342634737e-08,
"loss": 1.4375,
"step": 141000
},
{
"epoch": 3.0,
"step": 141054,
"total_flos": 1.3156191282659328e+18,
"train_loss": 1.6363766058878637,
"train_runtime": 156761.3641,
"train_samples_per_second": 8.998,
"train_steps_per_second": 0.9
}
],
"max_steps": 141054,
"num_train_epochs": 3,
"total_flos": 1.3156191282659328e+18,
"trial_name": null,
"trial_params": null
}