| { | |
| "best_metric": 0.1285061538219452, | |
| "best_model_checkpoint": "YituTech/conv-bert-base/subtaskA/0/checkpoint-23952", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 71856, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9860832776664444e-05, | |
| "loss": 0.4097, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9721665553328883e-05, | |
| "loss": 0.276, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9582498329993322e-05, | |
| "loss": 0.2464, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.944333110665776e-05, | |
| "loss": 0.2459, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9304163883322203e-05, | |
| "loss": 0.2172, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9164996659986642e-05, | |
| "loss": 0.1828, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.902582943665108e-05, | |
| "loss": 0.1904, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.888666221331552e-05, | |
| "loss": 0.1698, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8747494989979962e-05, | |
| "loss": 0.1837, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.86083277666444e-05, | |
| "loss": 0.1677, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8469160543308843e-05, | |
| "loss": 0.1765, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.832999331997328e-05, | |
| "loss": 0.1783, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.819082609663772e-05, | |
| "loss": 0.1346, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.805165887330216e-05, | |
| "loss": 0.1596, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.7912491649966602e-05, | |
| "loss": 0.1575, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.777332442663104e-05, | |
| "loss": 0.1452, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.763415720329548e-05, | |
| "loss": 0.125, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.7494989979959922e-05, | |
| "loss": 0.148, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.735582275662436e-05, | |
| "loss": 0.1533, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.72166555332888e-05, | |
| "loss": 0.1372, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.707748830995324e-05, | |
| "loss": 0.1328, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.693832108661768e-05, | |
| "loss": 0.1125, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.679915386328212e-05, | |
| "loss": 0.1268, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.6659986639946563e-05, | |
| "loss": 0.1365, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.6520819416611002e-05, | |
| "loss": 0.1361, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.638165219327544e-05, | |
| "loss": 0.1489, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.624248496993988e-05, | |
| "loss": 0.1171, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6103317746604322e-05, | |
| "loss": 0.1216, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.596415052326876e-05, | |
| "loss": 0.1007, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.58249832999332e-05, | |
| "loss": 0.117, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5685816076597642e-05, | |
| "loss": 0.109, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.554664885326208e-05, | |
| "loss": 0.1248, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5407481629926523e-05, | |
| "loss": 0.1184, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.5268314406590962e-05, | |
| "loss": 0.1271, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.5129147183255401e-05, | |
| "loss": 0.1157, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.498997995991984e-05, | |
| "loss": 0.1137, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.485081273658428e-05, | |
| "loss": 0.0987, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.471164551324872e-05, | |
| "loss": 0.0976, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.4572478289913162e-05, | |
| "loss": 0.1048, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.44333110665776e-05, | |
| "loss": 0.1118, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4294143843242041e-05, | |
| "loss": 0.0962, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.415497661990648e-05, | |
| "loss": 0.0826, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.4015809396570921e-05, | |
| "loss": 0.1102, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.3876642173235361e-05, | |
| "loss": 0.0911, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.37374749498998e-05, | |
| "loss": 0.1144, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.3598307726564241e-05, | |
| "loss": 0.0771, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.345914050322868e-05, | |
| "loss": 0.0949, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_f1": 0.9746158984635939, | |
| "eval_loss": 0.1285061538219452, | |
| "eval_runtime": 361.2457, | |
| "eval_samples_per_second": 66.304, | |
| "eval_steps_per_second": 16.576, | |
| "step": 23952 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.331997327989312e-05, | |
| "loss": 0.0795, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.318080605655756e-05, | |
| "loss": 0.0369, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.3041638833222002e-05, | |
| "loss": 0.0526, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.290247160988644e-05, | |
| "loss": 0.0374, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.2763304386550881e-05, | |
| "loss": 0.0543, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.262413716321532e-05, | |
| "loss": 0.0695, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.248496993987976e-05, | |
| "loss": 0.0488, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.23458027165442e-05, | |
| "loss": 0.0454, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.220663549320864e-05, | |
| "loss": 0.0478, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.206746826987308e-05, | |
| "loss": 0.0433, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.192830104653752e-05, | |
| "loss": 0.0391, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.1789133823201962e-05, | |
| "loss": 0.0472, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.16499665998664e-05, | |
| "loss": 0.0646, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.1510799376530842e-05, | |
| "loss": 0.0503, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.137163215319528e-05, | |
| "loss": 0.0407, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.1232464929859721e-05, | |
| "loss": 0.0703, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.109329770652416e-05, | |
| "loss": 0.0545, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.09541304831886e-05, | |
| "loss": 0.0434, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.081496325985304e-05, | |
| "loss": 0.0536, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.067579603651748e-05, | |
| "loss": 0.0519, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.053662881318192e-05, | |
| "loss": 0.0388, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.039746158984636e-05, | |
| "loss": 0.0369, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.0258294366510799e-05, | |
| "loss": 0.0536, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0119127143175241e-05, | |
| "loss": 0.067, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.97995991983968e-06, | |
| "loss": 0.0461, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 9.84079269650412e-06, | |
| "loss": 0.07, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.70162547316856e-06, | |
| "loss": 0.0457, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.562458249833e-06, | |
| "loss": 0.05, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.423291026497439e-06, | |
| "loss": 0.0446, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.28412380316188e-06, | |
| "loss": 0.0508, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.14495657982632e-06, | |
| "loss": 0.0289, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 9.005789356490761e-06, | |
| "loss": 0.0424, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.8666221331552e-06, | |
| "loss": 0.0418, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.72745490981964e-06, | |
| "loss": 0.0415, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.588287686484081e-06, | |
| "loss": 0.0359, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 8.44912046314852e-06, | |
| "loss": 0.0559, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.30995323981296e-06, | |
| "loss": 0.0438, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 8.1707860164774e-06, | |
| "loss": 0.0317, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.03161879314184e-06, | |
| "loss": 0.0319, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.892451569806279e-06, | |
| "loss": 0.0395, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 7.75328434647072e-06, | |
| "loss": 0.0579, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 7.614117123135159e-06, | |
| "loss": 0.0354, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 7.474949899799599e-06, | |
| "loss": 0.0527, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 7.33578267646404e-06, | |
| "loss": 0.0519, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.1966154531284795e-06, | |
| "loss": 0.0304, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 7.057448229792919e-06, | |
| "loss": 0.0436, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.91828100645736e-06, | |
| "loss": 0.0354, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.7791137831218005e-06, | |
| "loss": 0.0367, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_f1": 0.9587925851703407, | |
| "eval_loss": 0.2794201374053955, | |
| "eval_runtime": 367.1663, | |
| "eval_samples_per_second": 65.235, | |
| "eval_steps_per_second": 16.309, | |
| "step": 47904 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.63994655978624e-06, | |
| "loss": 0.0416, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.50077933645068e-06, | |
| "loss": 0.0047, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.36161211311512e-06, | |
| "loss": 0.012, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.22244488977956e-06, | |
| "loss": 0.0146, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.083277666443999e-06, | |
| "loss": 0.013, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.944110443108439e-06, | |
| "loss": 0.0071, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.80494321977288e-06, | |
| "loss": 0.0166, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.6657759964373195e-06, | |
| "loss": 0.0137, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.526608773101759e-06, | |
| "loss": 0.0073, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.387441549766199e-06, | |
| "loss": 0.0133, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5.248274326430639e-06, | |
| "loss": 0.0147, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.1091071030950786e-06, | |
| "loss": 0.0334, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.969939879759519e-06, | |
| "loss": 0.0151, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.830772656423959e-06, | |
| "loss": 0.0073, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.6916054330883996e-06, | |
| "loss": 0.004, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.552438209752839e-06, | |
| "loss": 0.0156, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.41327098641728e-06, | |
| "loss": 0.0081, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.27410376308172e-06, | |
| "loss": 0.0122, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.1349365397461595e-06, | |
| "loss": 0.0155, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.995769316410599e-06, | |
| "loss": 0.0119, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.856602093075039e-06, | |
| "loss": 0.0225, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.717434869739479e-06, | |
| "loss": 0.0107, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.578267646403919e-06, | |
| "loss": 0.0035, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.4391004230683596e-06, | |
| "loss": 0.015, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.2999331997327993e-06, | |
| "loss": 0.0064, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 3.160765976397239e-06, | |
| "loss": 0.0036, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 3.0215987530616793e-06, | |
| "loss": 0.0137, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.882431529726119e-06, | |
| "loss": 0.0124, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 2.743264306390559e-06, | |
| "loss": 0.0128, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 2.604097083054999e-06, | |
| "loss": 0.0094, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.464929859719439e-06, | |
| "loss": 0.0032, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 2.325762636383879e-06, | |
| "loss": 0.0022, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 2.186595413048319e-06, | |
| "loss": 0.0175, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 2.047428189712759e-06, | |
| "loss": 0.0111, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.908260966377199e-06, | |
| "loss": 0.0164, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.769093743041639e-06, | |
| "loss": 0.0143, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.629926519706079e-06, | |
| "loss": 0.0143, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.4907592963705188e-06, | |
| "loss": 0.0094, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.3515920730349588e-06, | |
| "loss": 0.0194, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.2124248496993988e-06, | |
| "loss": 0.0084, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.073257626363839e-06, | |
| "loss": 0.0049, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.340904030282788e-07, | |
| "loss": 0.0048, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.949231796927188e-07, | |
| "loss": 0.0095, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 6.557559563571589e-07, | |
| "loss": 0.0069, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 5.165887330215988e-07, | |
| "loss": 0.0043, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.7742150968603874e-07, | |
| "loss": 0.002, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.3825428635047877e-07, | |
| "loss": 0.0041, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 9.908706301491873e-08, | |
| "loss": 0.0062, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_f1": 0.9778306613226453, | |
| "eval_loss": 0.17954404652118683, | |
| "eval_runtime": 362.9189, | |
| "eval_samples_per_second": 65.998, | |
| "eval_steps_per_second": 16.5, | |
| "step": 71856 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 71856, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 7.278683421643776e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |