| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 27.796988998262883, | |
| "global_step": 120000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.669214732453091e-06, | |
| "loss": 49.8699, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7355802640722724e-05, | |
| "loss": 41.2735, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.6042390548992352e-05, | |
| "loss": 36.9489, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.472897845726199e-05, | |
| "loss": 33.9779, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.341556636553162e-05, | |
| "loss": 31.7843, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.210215427380125e-05, | |
| "loss": 30.3453, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.0788742182070875e-05, | |
| "loss": 29.3517, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.945795691452397e-05, | |
| "loss": 28.5657, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.81445448227936e-05, | |
| "loss": 27.9059, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.683113273106324e-05, | |
| "loss": 27.256, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.550034746351632e-05, | |
| "loss": 26.6368, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.00010418693537178596, | |
| "loss": 26.0616, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00011287352328005559, | |
| "loss": 25.5779, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.00012154273801250867, | |
| "loss": 25.154, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0001302293259207783, | |
| "loss": 24.757, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00013891591382904795, | |
| "loss": 24.3901, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0001476025017373176, | |
| "loss": 24.0289, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0001562890896455872, | |
| "loss": 23.7116, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.0001649756775538568, | |
| "loss": 23.3895, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.00017366226546212648, | |
| "loss": 23.1314, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 0.00018234885337039609, | |
| "loss": 22.8528, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 0.00019103544127866572, | |
| "loss": 22.5796, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.00019972202918693536, | |
| "loss": 22.3478, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 0.000208408617095205, | |
| "loss": 22.1018, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.0002170952050034746, | |
| "loss": 21.8626, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 0.0002257644197359277, | |
| "loss": 21.6681, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 0.00023445100764419736, | |
| "loss": 21.4255, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 0.00024313759555246697, | |
| "loss": 21.2724, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 0.0002518241834607366, | |
| "loss": 21.0849, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 0.00026051077136900625, | |
| "loss": 20.8968, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 0.00026919735927727584, | |
| "loss": 20.7233, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 0.0002777970813064628, | |
| "loss": 20.5867, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 0.00028648366921473243, | |
| "loss": 20.4368, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 0.00029517025712300207, | |
| "loss": 20.3374, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 0.000299571461663192, | |
| "loss": 20.2117, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 0.00029860628522893984, | |
| "loss": 20.0622, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 0.00029764110879468763, | |
| "loss": 19.9614, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 0.0002966759323604355, | |
| "loss": 19.8524, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 0.0002957107559261833, | |
| "loss": 19.7578, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 0.00029474557949193107, | |
| "loss": 19.6587, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 0.0002937804030576789, | |
| "loss": 19.5385, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 0.0002928152266234267, | |
| "loss": 19.4285, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 0.00029185005018917456, | |
| "loss": 19.3376, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 0.0002908848737549224, | |
| "loss": 19.2065, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 0.0002899196973206702, | |
| "loss": 19.1089, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 0.000288954520886418, | |
| "loss": 19.0411, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 0.0002879893444521658, | |
| "loss": 18.9864, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 0.00028702416801791364, | |
| "loss": 18.9157, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 0.0002860589915836615, | |
| "loss": 18.8606, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 0.0002850938151494093, | |
| "loss": 18.8049, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 0.0002841286387151571, | |
| "loss": 18.7415, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 0.0002831634622809049, | |
| "loss": 18.695, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 0.0002821982858466527, | |
| "loss": 18.5979, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 0.00028123310941240056, | |
| "loss": 18.5566, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 0.0002802698633310169, | |
| "loss": 18.5145, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 0.0002793046868967647, | |
| "loss": 18.465, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 0.0002783395104625125, | |
| "loss": 18.4232, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 0.00027737433402826034, | |
| "loss": 18.368, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 0.00027640915759400813, | |
| "loss": 18.315, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 0.000275443981159756, | |
| "loss": 18.2882, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 0.00027447880472550383, | |
| "loss": 18.2239, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 0.0002735136282912516, | |
| "loss": 18.163, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 0.0002725503822098679, | |
| "loss": 18.1283, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 0.00027158520577561575, | |
| "loss": 18.1016, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 0.0002706200293413636, | |
| "loss": 18.0663, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "learning_rate": 0.0002696567832599799, | |
| "loss": 18.0263, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 7.76, | |
| "learning_rate": 0.00026869160682572773, | |
| "loss": 17.9946, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 0.00026772643039147553, | |
| "loss": 17.9719, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 0.0002667612539572234, | |
| "loss": 17.931, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 0.00026579607752297117, | |
| "loss": 17.8646, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 0.000264830901088719, | |
| "loss": 17.8358, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 0.0002638676550073353, | |
| "loss": 17.8081, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 0.00026290247857308315, | |
| "loss": 17.7928, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "learning_rate": 0.00026193730213883095, | |
| "loss": 17.7663, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 0.0002609721257045788, | |
| "loss": 17.7224, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 0.0002600069492703266, | |
| "loss": 17.7134, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 0.0002590417728360744, | |
| "loss": 17.6934, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 0.00025807659640182223, | |
| "loss": 17.662, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 0.0002571114199675701, | |
| "loss": 17.5886, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "learning_rate": 0.00025615010423905485, | |
| "loss": 17.5854, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 0.0002551849278048027, | |
| "loss": 17.5689, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 0.00025421975137055055, | |
| "loss": 17.5545, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 9.61, | |
| "learning_rate": 0.00025325650528916683, | |
| "loss": 17.5269, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 0.00025229132885491463, | |
| "loss": 17.5051, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 0.0002513261524206625, | |
| "loss": 17.4922, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 0.0002503609759864103, | |
| "loss": 17.4682, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 10.08, | |
| "learning_rate": 0.0002493957995521581, | |
| "loss": 17.4151, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 10.19, | |
| "learning_rate": 0.0002484306231179059, | |
| "loss": 17.3825, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 10.31, | |
| "learning_rate": 0.00024746544668365376, | |
| "loss": 17.3681, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "learning_rate": 0.00024650027024940155, | |
| "loss": 17.341, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 10.54, | |
| "learning_rate": 0.0002455350938151494, | |
| "loss": 17.3344, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 10.66, | |
| "learning_rate": 0.0002445699173808972, | |
| "loss": 17.316, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 10.77, | |
| "learning_rate": 0.00024360474094664502, | |
| "loss": 17.3002, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 10.89, | |
| "learning_rate": 0.00024264149486526136, | |
| "loss": 17.279, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 0.00024167631843100918, | |
| "loss": 17.2714, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 0.00024071114199675697, | |
| "loss": 17.2032, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 11.23, | |
| "learning_rate": 0.00023974596556250482, | |
| "loss": 17.1982, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 11.35, | |
| "learning_rate": 0.00023878271948112113, | |
| "loss": 17.1764, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 11.47, | |
| "learning_rate": 0.00023781754304686895, | |
| "loss": 17.1697, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 0.00023685236661261674, | |
| "loss": 17.1619, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 0.0002358871901783646, | |
| "loss": 17.1475, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 0.0002349239440969809, | |
| "loss": 17.1354, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 0.00023395876766272873, | |
| "loss": 17.1193, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 12.05, | |
| "learning_rate": 0.00023299359122847657, | |
| "loss": 17.0976, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "learning_rate": 0.00023203034514709289, | |
| "loss": 17.0559, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 12.28, | |
| "learning_rate": 0.00023106516871284068, | |
| "loss": 17.0478, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 12.39, | |
| "learning_rate": 0.0002300999922785885, | |
| "loss": 17.0457, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 12.51, | |
| "learning_rate": 0.00022913481584433635, | |
| "loss": 17.0296, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 12.62, | |
| "learning_rate": 0.00022816963941008414, | |
| "loss": 17.0185, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 12.74, | |
| "learning_rate": 0.00022720639332870045, | |
| "loss": 17.012, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 12.86, | |
| "learning_rate": 0.0002262431472473168, | |
| "loss": 16.9982, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 0.00022527990116593313, | |
| "loss": 16.981, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 13.09, | |
| "learning_rate": 0.00022431472473168093, | |
| "loss": 16.9492, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "learning_rate": 0.00022335147865029726, | |
| "loss": 16.9352, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 13.32, | |
| "learning_rate": 0.00022238630221604506, | |
| "loss": 16.9178, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 0.0002214211257817929, | |
| "loss": 16.9156, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 13.55, | |
| "learning_rate": 0.00022045787970040922, | |
| "loss": 16.9093, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 13.67, | |
| "learning_rate": 0.00021949270326615704, | |
| "loss": 16.9013, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 13.78, | |
| "learning_rate": 0.00021852752683190483, | |
| "loss": 16.8895, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 13.9, | |
| "learning_rate": 0.00021756235039765268, | |
| "loss": 16.8733, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 0.0002165971739634005, | |
| "loss": 16.881, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 14.13, | |
| "learning_rate": 0.0002156319975291483, | |
| "loss": 16.8109, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 14.25, | |
| "learning_rate": 0.00021466682109489614, | |
| "loss": 16.8187, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 14.36, | |
| "learning_rate": 0.00021370164466064394, | |
| "loss": 16.8078, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 14.48, | |
| "learning_rate": 0.00021273646822639176, | |
| "loss": 16.8041, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 14.59, | |
| "learning_rate": 0.0002117712917921396, | |
| "loss": 16.7933, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 14.71, | |
| "learning_rate": 0.0002108061153578874, | |
| "loss": 16.7964, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 14.83, | |
| "learning_rate": 0.00020984093892363522, | |
| "loss": 16.778, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 0.00020887576248938302, | |
| "loss": 16.7764, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 15.06, | |
| "learning_rate": 0.00020791058605513086, | |
| "loss": 16.7603, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "learning_rate": 0.00020694540962087868, | |
| "loss": 16.7251, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 15.29, | |
| "learning_rate": 0.00020598023318662648, | |
| "loss": 16.712, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 15.4, | |
| "learning_rate": 0.00020501505675237433, | |
| "loss": 16.7072, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 15.52, | |
| "learning_rate": 0.00020404988031812215, | |
| "loss": 16.7106, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 0.00020308470388386994, | |
| "loss": 16.7037, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 15.75, | |
| "learning_rate": 0.0002021195274496178, | |
| "loss": 16.6918, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 15.87, | |
| "learning_rate": 0.00020115435101536558, | |
| "loss": 16.6886, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 15.98, | |
| "learning_rate": 0.0002001891745811134, | |
| "loss": 16.6947, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 16.1, | |
| "learning_rate": 0.00019922399814686125, | |
| "loss": 16.6516, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 16.21, | |
| "learning_rate": 0.00019825882171260905, | |
| "loss": 16.6403, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 16.33, | |
| "learning_rate": 0.00019729364527835687, | |
| "loss": 16.6343, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 16.45, | |
| "learning_rate": 0.00019632846884410466, | |
| "loss": 16.6257, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 16.56, | |
| "learning_rate": 0.00019536522276272103, | |
| "loss": 16.6181, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 16.68, | |
| "learning_rate": 0.00019440004632846882, | |
| "loss": 16.6204, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 16.79, | |
| "learning_rate": 0.00019343486989421664, | |
| "loss": 16.6101, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 16.91, | |
| "learning_rate": 0.00019247162381283295, | |
| "loss": 16.6116, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 0.0001915083777314493, | |
| "loss": 16.6047, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 17.14, | |
| "learning_rate": 0.0001905432012971971, | |
| "loss": 16.5427, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 17.26, | |
| "learning_rate": 0.00018957802486294493, | |
| "loss": 16.5453, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 17.37, | |
| "learning_rate": 0.00018861284842869276, | |
| "loss": 16.5489, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 17.49, | |
| "learning_rate": 0.00018764960234730907, | |
| "loss": 16.5505, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 17.6, | |
| "learning_rate": 0.0001866844259130569, | |
| "loss": 16.5423, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 17.72, | |
| "learning_rate": 0.0001857192494788047, | |
| "loss": 16.543, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 17.84, | |
| "learning_rate": 0.00018475407304455253, | |
| "loss": 16.5402, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 17.95, | |
| "learning_rate": 0.00018378889661030035, | |
| "loss": 16.5373, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 18.07, | |
| "learning_rate": 0.00018282372017604814, | |
| "loss": 16.5045, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 18.18, | |
| "learning_rate": 0.000181858543741796, | |
| "loss": 16.4782, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 18.3, | |
| "learning_rate": 0.00018089336730754381, | |
| "loss": 16.4817, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 18.42, | |
| "learning_rate": 0.0001799281908732916, | |
| "loss": 16.4737, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 18.53, | |
| "learning_rate": 0.00017896494479190797, | |
| "loss": 16.4797, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 18.65, | |
| "learning_rate": 0.00017799976835765577, | |
| "loss": 16.4755, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 18.76, | |
| "learning_rate": 0.0001770345919234036, | |
| "loss": 16.464, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 18.88, | |
| "learning_rate": 0.00017606941548915138, | |
| "loss": 16.4599, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 18.99, | |
| "learning_rate": 0.00017510423905489923, | |
| "loss": 16.4511, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 19.11, | |
| "learning_rate": 0.00017413906262064705, | |
| "loss": 16.42, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 19.23, | |
| "learning_rate": 0.00017317581653926336, | |
| "loss": 16.4039, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 19.34, | |
| "learning_rate": 0.00017221064010501116, | |
| "loss": 16.4083, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 19.46, | |
| "learning_rate": 0.000171245463670759, | |
| "loss": 16.413, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 19.57, | |
| "learning_rate": 0.00017028028723650683, | |
| "loss": 16.4069, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 19.69, | |
| "learning_rate": 0.00016931511080225462, | |
| "loss": 16.3972, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 19.81, | |
| "learning_rate": 0.00016834993436800247, | |
| "loss": 16.4007, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 19.92, | |
| "learning_rate": 0.0001673847579337503, | |
| "loss": 16.3945, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 20.04, | |
| "learning_rate": 0.00016641958149949808, | |
| "loss": 16.3928, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 20.15, | |
| "learning_rate": 0.00016545440506524593, | |
| "loss": 16.3468, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 20.27, | |
| "learning_rate": 0.00016448922863099373, | |
| "loss": 16.3459, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 20.38, | |
| "learning_rate": 0.00016352598254961006, | |
| "loss": 16.3429, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 20.5, | |
| "learning_rate": 0.00016256080611535786, | |
| "loss": 16.3394, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 20.62, | |
| "learning_rate": 0.0001615956296811057, | |
| "loss": 16.3417, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 20.73, | |
| "learning_rate": 0.0001606304532468535, | |
| "loss": 16.3369, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 20.85, | |
| "learning_rate": 0.00015966527681260132, | |
| "loss": 16.3392, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 20.96, | |
| "learning_rate": 0.00015870203073121763, | |
| "loss": 16.3402, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 21.08, | |
| "learning_rate": 0.00015773685429696548, | |
| "loss": 16.3009, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 21.2, | |
| "learning_rate": 0.00015677167786271327, | |
| "loss": 16.2924, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 21.31, | |
| "learning_rate": 0.0001558065014284611, | |
| "loss": 16.2908, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 21.43, | |
| "learning_rate": 0.00015484132499420894, | |
| "loss": 16.289, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 21.54, | |
| "learning_rate": 0.00015387614855995674, | |
| "loss": 16.2874, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 21.66, | |
| "learning_rate": 0.00015291097212570456, | |
| "loss": 16.2869, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 21.77, | |
| "learning_rate": 0.00015194772604432087, | |
| "loss": 16.2867, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 21.89, | |
| "learning_rate": 0.00015098254961006872, | |
| "loss": 16.2779, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 0.00015002123388155355, | |
| "loss": 16.2885, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 22.12, | |
| "learning_rate": 0.00014905605744730134, | |
| "loss": 16.2251, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 22.24, | |
| "learning_rate": 0.0001480908810130492, | |
| "loss": 16.247, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 22.35, | |
| "learning_rate": 0.00014712570457879698, | |
| "loss": 16.2315, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 22.47, | |
| "learning_rate": 0.0001461605281445448, | |
| "loss": 16.2464, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 22.59, | |
| "learning_rate": 0.00014519535171029263, | |
| "loss": 16.2326, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 22.7, | |
| "learning_rate": 0.00014423017527604045, | |
| "loss": 16.2327, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 22.82, | |
| "learning_rate": 0.00014326499884178827, | |
| "loss": 16.2269, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 22.93, | |
| "learning_rate": 0.0001422998224075361, | |
| "loss": 16.2186, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 23.05, | |
| "learning_rate": 0.0001413346459732839, | |
| "loss": 16.2302, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 23.16, | |
| "learning_rate": 0.0001403694695390317, | |
| "loss": 16.1879, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 23.28, | |
| "learning_rate": 0.00013940622345764804, | |
| "loss": 16.1834, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 23.4, | |
| "learning_rate": 0.00013844104702339586, | |
| "loss": 16.1779, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 23.51, | |
| "learning_rate": 0.00013747587058914368, | |
| "loss": 16.1835, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 23.63, | |
| "learning_rate": 0.00013651455486062851, | |
| "loss": 16.1829, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 23.74, | |
| "learning_rate": 0.00013554937842637633, | |
| "loss": 16.1814, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 23.86, | |
| "learning_rate": 0.00013458420199212413, | |
| "loss": 16.1864, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 23.97, | |
| "learning_rate": 0.00013361902555787198, | |
| "loss": 16.1784, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 24.09, | |
| "learning_rate": 0.0001326538491236198, | |
| "loss": 16.149, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 24.21, | |
| "learning_rate": 0.0001316886726893676, | |
| "loss": 16.1459, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 24.32, | |
| "learning_rate": 0.0001307234962551154, | |
| "loss": 16.1374, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 24.44, | |
| "learning_rate": 0.00012975831982086326, | |
| "loss": 16.143, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 24.55, | |
| "learning_rate": 0.00012879314338661105, | |
| "loss": 16.1267, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 24.67, | |
| "learning_rate": 0.00012782796695235888, | |
| "loss": 16.1363, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 24.79, | |
| "learning_rate": 0.0001268666512238437, | |
| "loss": 16.1328, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 24.9, | |
| "learning_rate": 0.00012590147478959153, | |
| "loss": 16.1314, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 0.00012493629835533935, | |
| "loss": 16.1427, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 25.13, | |
| "learning_rate": 0.00012397112192108717, | |
| "loss": 16.0975, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 25.25, | |
| "learning_rate": 0.000123005945486835, | |
| "loss": 16.091, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 25.36, | |
| "learning_rate": 0.00012204076905258281, | |
| "loss": 16.0959, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 25.48, | |
| "learning_rate": 0.00012107559261833062, | |
| "loss": 16.0919, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 25.6, | |
| "learning_rate": 0.00012011041618407844, | |
| "loss": 16.077, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 25.71, | |
| "learning_rate": 0.00011914523974982626, | |
| "loss": 16.0942, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 25.83, | |
| "learning_rate": 0.00011818006331557408, | |
| "loss": 16.0786, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 25.94, | |
| "learning_rate": 0.00011721681723419039, | |
| "loss": 16.0852, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 26.06, | |
| "learning_rate": 0.00011625164079993821, | |
| "loss": 16.0732, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 26.18, | |
| "learning_rate": 0.00011528646436568605, | |
| "loss": 16.0445, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 26.29, | |
| "learning_rate": 0.00011432128793143386, | |
| "loss": 16.043, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 26.41, | |
| "learning_rate": 0.00011335611149718168, | |
| "loss": 16.0484, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 26.52, | |
| "learning_rate": 0.00011239093506292948, | |
| "loss": 16.0494, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 26.64, | |
| "learning_rate": 0.00011142575862867732, | |
| "loss": 16.043, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 26.75, | |
| "learning_rate": 0.00011046058219442514, | |
| "loss": 16.0415, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 26.87, | |
| "learning_rate": 0.00010949540576017295, | |
| "loss": 16.032, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 26.99, | |
| "learning_rate": 0.00010853022932592077, | |
| "loss": 16.0346, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 27.1, | |
| "learning_rate": 0.00010756505289166857, | |
| "loss": 16.0145, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 27.22, | |
| "learning_rate": 0.00010659987645741641, | |
| "loss": 16.0032, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 27.33, | |
| "learning_rate": 0.00010563470002316423, | |
| "loss": 15.9992, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 27.45, | |
| "learning_rate": 0.00010467145394178054, | |
| "loss": 16.0033, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 27.57, | |
| "learning_rate": 0.00010370627750752838, | |
| "loss": 15.9964, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 27.68, | |
| "learning_rate": 0.00010274303142614469, | |
| "loss": 15.9994, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 27.8, | |
| "learning_rate": 0.00010177785499189251, | |
| "loss": 16.0, | |
| "step": 120000 | |
| } | |
| ], | |
| "max_steps": 172680, | |
| "num_train_epochs": 40, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |