| { | |
| "best_metric": 89.26259594116004, | |
| "best_model_checkpoint": "/home/shared/dt01/temutauro/ccasimiro/berta/src/qa/roberta-base-ca-v2-qa/SQAC-ca_catalanqa/checkpoint-2458", | |
| "epoch": 10.0, | |
| "global_step": 12290, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.979658258746949e-05, | |
| "loss": 5.2373, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.959316517493898e-05, | |
| "loss": 3.9809, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.938974776240847e-05, | |
| "loss": 3.5244, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.918633034987795e-05, | |
| "loss": 3.1797, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.898291293734744e-05, | |
| "loss": 2.9253, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.877949552481693e-05, | |
| "loss": 2.585, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.857607811228642e-05, | |
| "loss": 2.3783, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.83726606997559e-05, | |
| "loss": 2.1168, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.816924328722539e-05, | |
| "loss": 1.918, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.796582587469488e-05, | |
| "loss": 1.7789, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.776240846216437e-05, | |
| "loss": 1.683, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.755899104963385e-05, | |
| "loss": 1.5405, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.7355573637103337e-05, | |
| "loss": 1.5408, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.715215622457283e-05, | |
| "loss": 1.5068, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.694873881204231e-05, | |
| "loss": 1.3649, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.6745321399511796e-05, | |
| "loss": 1.3839, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.6541903986981287e-05, | |
| "loss": 1.2455, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.633848657445077e-05, | |
| "loss": 1.3101, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.613506916192026e-05, | |
| "loss": 1.2689, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.5931651749389746e-05, | |
| "loss": 1.1822, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.5728234336859237e-05, | |
| "loss": 1.183, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.552481692432872e-05, | |
| "loss": 1.1435, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.532139951179821e-05, | |
| "loss": 1.2057, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.5117982099267696e-05, | |
| "loss": 1.1225, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_exact_match": 74.34094903339192, | |
| "eval_f1": 88.58853878556948, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4914564686737187e-05, | |
| "loss": 0.9737, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.471114727420667e-05, | |
| "loss": 0.8994, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.450772986167616e-05, | |
| "loss": 0.8886, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.4304312449145646e-05, | |
| "loss": 0.8679, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.4100895036615137e-05, | |
| "loss": 0.899, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.389747762408463e-05, | |
| "loss": 0.8763, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.369406021155411e-05, | |
| "loss": 0.8041, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.34906427990236e-05, | |
| "loss": 0.828, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.3287225386493087e-05, | |
| "loss": 0.8691, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.308380797396258e-05, | |
| "loss": 0.8637, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.288039056143206e-05, | |
| "loss": 0.8469, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.267697314890155e-05, | |
| "loss": 0.8547, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.2473555736371037e-05, | |
| "loss": 0.8116, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.227013832384053e-05, | |
| "loss": 0.8748, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.206672091131001e-05, | |
| "loss": 0.8508, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.18633034987795e-05, | |
| "loss": 0.8338, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.1659886086248987e-05, | |
| "loss": 0.8547, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.145646867371847e-05, | |
| "loss": 0.8266, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.125305126118796e-05, | |
| "loss": 0.8162, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.1049633848657446e-05, | |
| "loss": 0.8152, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.084621643612693e-05, | |
| "loss": 0.8674, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.064279902359642e-05, | |
| "loss": 0.868, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.0439381611065905e-05, | |
| "loss": 0.7845, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.0235964198535396e-05, | |
| "loss": 0.8029, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.003254678600488e-05, | |
| "loss": 0.7942, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_exact_match": 75.52724077328647, | |
| "eval_f1": 89.26259594116004, | |
| "step": 2458 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.982912937347437e-05, | |
| "loss": 0.6307, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.9625711960943855e-05, | |
| "loss": 0.5914, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.9422294548413346e-05, | |
| "loss": 0.5965, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.921887713588283e-05, | |
| "loss": 0.5843, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.901545972335232e-05, | |
| "loss": 0.6001, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 3.8812042310821805e-05, | |
| "loss": 0.5671, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 3.8608624898291296e-05, | |
| "loss": 0.5726, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 3.840520748576078e-05, | |
| "loss": 0.5599, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 3.820179007323027e-05, | |
| "loss": 0.582, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.799837266069976e-05, | |
| "loss": 0.5981, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.7794955248169246e-05, | |
| "loss": 0.572, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.759153783563874e-05, | |
| "loss": 0.5506, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.738812042310822e-05, | |
| "loss": 0.5822, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.718470301057771e-05, | |
| "loss": 0.5504, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.6981285598047196e-05, | |
| "loss": 0.6113, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.677786818551669e-05, | |
| "loss": 0.5381, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.657445077298617e-05, | |
| "loss": 0.624, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 3.637103336045566e-05, | |
| "loss": 0.5775, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.6167615947925146e-05, | |
| "loss": 0.5344, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.596419853539463e-05, | |
| "loss": 0.5982, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 3.576078112286412e-05, | |
| "loss": 0.6115, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.5557363710333605e-05, | |
| "loss": 0.5735, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 3.535394629780309e-05, | |
| "loss": 0.5682, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.515052888527258e-05, | |
| "loss": 0.5968, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_exact_match": 75.96660808435853, | |
| "eval_f1": 88.9904800819017, | |
| "step": 3687 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.4947111472742064e-05, | |
| "loss": 0.5059, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.4743694060211555e-05, | |
| "loss": 0.4033, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.454027664768104e-05, | |
| "loss": 0.4208, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.433685923515053e-05, | |
| "loss": 0.4397, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.4133441822620014e-05, | |
| "loss": 0.4268, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.3930024410089505e-05, | |
| "loss": 0.4098, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.372660699755899e-05, | |
| "loss": 0.4226, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.352318958502848e-05, | |
| "loss": 0.4164, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.3319772172497964e-05, | |
| "loss": 0.3775, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.3116354759967455e-05, | |
| "loss": 0.4004, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.291293734743694e-05, | |
| "loss": 0.4427, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.270951993490643e-05, | |
| "loss": 0.4358, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.250610252237592e-05, | |
| "loss": 0.3924, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.2302685109845405e-05, | |
| "loss": 0.4541, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.2099267697314896e-05, | |
| "loss": 0.4247, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.189585028478438e-05, | |
| "loss": 0.4114, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.169243287225387e-05, | |
| "loss": 0.4264, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.1489015459723355e-05, | |
| "loss": 0.4246, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.1285598047192846e-05, | |
| "loss": 0.4324, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.108218063466233e-05, | |
| "loss": 0.4018, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.087876322213182e-05, | |
| "loss": 0.4354, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.0675345809601305e-05, | |
| "loss": 0.4279, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.0471928397070792e-05, | |
| "loss": 0.4264, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.0268510984540277e-05, | |
| "loss": 0.4211, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.0065093572009767e-05, | |
| "loss": 0.5077, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_exact_match": 75.61511423550088, | |
| "eval_f1": 88.81911407320312, | |
| "step": 4916 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 2.986167615947925e-05, | |
| "loss": 0.302, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 2.9658258746948742e-05, | |
| "loss": 0.3048, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 2.9454841334418227e-05, | |
| "loss": 0.3303, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 2.9251423921887717e-05, | |
| "loss": 0.2985, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 2.90480065093572e-05, | |
| "loss": 0.3188, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 2.884458909682669e-05, | |
| "loss": 0.3033, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 2.8641171684296177e-05, | |
| "loss": 0.2714, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 2.8437754271765664e-05, | |
| "loss": 0.3447, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 2.8234336859235148e-05, | |
| "loss": 0.3115, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 2.803091944670464e-05, | |
| "loss": 0.3237, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 2.7827502034174123e-05, | |
| "loss": 0.3097, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 2.7624084621643614e-05, | |
| "loss": 0.3016, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 2.7420667209113098e-05, | |
| "loss": 0.3256, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 2.721724979658259e-05, | |
| "loss": 0.3051, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.7013832384052073e-05, | |
| "loss": 0.3188, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 2.6810414971521564e-05, | |
| "loss": 0.3153, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.6606997558991055e-05, | |
| "loss": 0.2839, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.640358014646054e-05, | |
| "loss": 0.3019, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.6200162733930027e-05, | |
| "loss": 0.3057, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.5996745321399514e-05, | |
| "loss": 0.3041, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 2.5793327908869e-05, | |
| "loss": 0.2825, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 2.5589910496338486e-05, | |
| "loss": 0.3363, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 2.5386493083807977e-05, | |
| "loss": 0.3117, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 2.518307567127746e-05, | |
| "loss": 0.3155, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_exact_match": 75.35149384885764, | |
| "eval_f1": 88.92853845000909, | |
| "step": 6145 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 2.4979658258746948e-05, | |
| "loss": 0.3373, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 2.4776240846216436e-05, | |
| "loss": 0.2384, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.4572823433685923e-05, | |
| "loss": 0.2404, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 2.436940602115541e-05, | |
| "loss": 0.2242, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 2.4165988608624898e-05, | |
| "loss": 0.213, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 2.3962571196094386e-05, | |
| "loss": 0.229, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 2.3759153783563877e-05, | |
| "loss": 0.2324, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 2.3555736371033364e-05, | |
| "loss": 0.2143, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 2.335231895850285e-05, | |
| "loss": 0.1819, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 2.3148901545972336e-05, | |
| "loss": 0.2446, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 2.2945484133441823e-05, | |
| "loss": 0.2254, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 2.274206672091131e-05, | |
| "loss": 0.2415, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 2.2538649308380798e-05, | |
| "loss": 0.2914, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 2.2335231895850286e-05, | |
| "loss": 0.2204, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 2.2131814483319773e-05, | |
| "loss": 0.2447, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 2.192839707078926e-05, | |
| "loss": 0.2629, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 2.1724979658258748e-05, | |
| "loss": 0.2668, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 2.1521562245728236e-05, | |
| "loss": 0.224, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 2.1318144833197723e-05, | |
| "loss": 0.2601, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 2.111472742066721e-05, | |
| "loss": 0.2408, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 2.0911310008136698e-05, | |
| "loss": 0.2212, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 2.0707892595606186e-05, | |
| "loss": 0.2445, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 2.0504475183075673e-05, | |
| "loss": 0.2413, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 2.030105777054516e-05, | |
| "loss": 0.2178, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 2.009764035801465e-05, | |
| "loss": 0.2166, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_exact_match": 75.04393673110721, | |
| "eval_f1": 88.49211373882532, | |
| "step": 7374 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.9894222945484132e-05, | |
| "loss": 0.2315, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.969080553295362e-05, | |
| "loss": 0.1622, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.9487388120423107e-05, | |
| "loss": 0.1563, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 1.9283970707892595e-05, | |
| "loss": 0.1794, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 1.9080553295362082e-05, | |
| "loss": 0.1874, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 1.887713588283157e-05, | |
| "loss": 0.1561, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 1.8673718470301057e-05, | |
| "loss": 0.1966, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 1.8470301057770545e-05, | |
| "loss": 0.1621, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 1.8266883645240032e-05, | |
| "loss": 0.1468, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 1.806346623270952e-05, | |
| "loss": 0.1831, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 1.786004882017901e-05, | |
| "loss": 0.177, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 1.76566314076485e-05, | |
| "loss": 0.1498, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 6.51, | |
| "learning_rate": 1.7453213995117982e-05, | |
| "loss": 0.1556, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 1.724979658258747e-05, | |
| "loss": 0.1928, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 1.7046379170056957e-05, | |
| "loss": 0.1776, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 1.6842961757526445e-05, | |
| "loss": 0.1697, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 1.6639544344995932e-05, | |
| "loss": 0.1782, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 1.643612693246542e-05, | |
| "loss": 0.1658, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 1.6232709519934907e-05, | |
| "loss": 0.1605, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 1.6029292107404395e-05, | |
| "loss": 0.1512, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 1.5825874694873882e-05, | |
| "loss": 0.1689, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 1.562245728234337e-05, | |
| "loss": 0.1774, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 1.5419039869812857e-05, | |
| "loss": 0.1884, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 1.5215622457282345e-05, | |
| "loss": 0.1559, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.501220504475183e-05, | |
| "loss": 0.18, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_exact_match": 73.72583479789104, | |
| "eval_f1": 87.84725743086018, | |
| "step": 8603 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.4808787632221318e-05, | |
| "loss": 0.1476, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 1.4605370219690806e-05, | |
| "loss": 0.1096, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 1.4401952807160293e-05, | |
| "loss": 0.1364, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 1.419853539462978e-05, | |
| "loss": 0.1248, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 1.3995117982099268e-05, | |
| "loss": 0.1468, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 1.3791700569568756e-05, | |
| "loss": 0.1217, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 1.3588283157038242e-05, | |
| "loss": 0.113, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 1.3384865744507729e-05, | |
| "loss": 0.1347, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 7.36, | |
| "learning_rate": 1.3181448331977217e-05, | |
| "loss": 0.1284, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 1.2978030919446704e-05, | |
| "loss": 0.1326, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 1.2774613506916192e-05, | |
| "loss": 0.12, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 1.2571196094385679e-05, | |
| "loss": 0.1201, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 1.2367778681855167e-05, | |
| "loss": 0.1273, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 1.2164361269324654e-05, | |
| "loss": 0.1191, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 1.1960943856794142e-05, | |
| "loss": 0.114, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 1.1757526444263629e-05, | |
| "loss": 0.1596, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 1.1554109031733117e-05, | |
| "loss": 0.1388, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 1.1350691619202604e-05, | |
| "loss": 0.1238, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 1.1147274206672092e-05, | |
| "loss": 0.0968, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 1.0943856794141579e-05, | |
| "loss": 0.1267, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 1.0740439381611067e-05, | |
| "loss": 0.1168, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 1.0537021969080554e-05, | |
| "loss": 0.1243, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 1.0333604556550042e-05, | |
| "loss": 0.1277, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 1.013018714401953e-05, | |
| "loss": 0.1074, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_exact_match": 74.56063268892794, | |
| "eval_f1": 88.45070096881145, | |
| "step": 9832 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 9.926769731489017e-06, | |
| "loss": 0.1002, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 9.723352318958504e-06, | |
| "loss": 0.0957, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 9.51993490642799e-06, | |
| "loss": 0.0851, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 9.316517493897477e-06, | |
| "loss": 0.0911, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 9.113100081366965e-06, | |
| "loss": 0.1098, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 8.909682668836452e-06, | |
| "loss": 0.0731, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 8.70626525630594e-06, | |
| "loss": 0.1131, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 8.502847843775427e-06, | |
| "loss": 0.1102, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 8.299430431244915e-06, | |
| "loss": 0.0748, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 8.096013018714402e-06, | |
| "loss": 0.0686, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 8.42, | |
| "learning_rate": 7.892595606183888e-06, | |
| "loss": 0.0966, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 7.689178193653377e-06, | |
| "loss": 0.1004, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 7.485760781122865e-06, | |
| "loss": 0.084, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 7.2823433685923525e-06, | |
| "loss": 0.0836, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 7.078925956061839e-06, | |
| "loss": 0.0896, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 6.875508543531327e-06, | |
| "loss": 0.0939, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 6.672091131000814e-06, | |
| "loss": 0.0965, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 6.468673718470302e-06, | |
| "loss": 0.1025, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 6.265256305939788e-06, | |
| "loss": 0.1208, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 6.061838893409276e-06, | |
| "loss": 0.1072, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 5.858421480878763e-06, | |
| "loss": 0.102, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 5.655004068348251e-06, | |
| "loss": 0.0979, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 5.451586655817738e-06, | |
| "loss": 0.1203, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 5.248169243287226e-06, | |
| "loss": 0.0906, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 5.044751830756713e-06, | |
| "loss": 0.1009, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_exact_match": 74.78031634446397, | |
| "eval_f1": 88.33201226117784, | |
| "step": 11061 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 4.8413344182262e-06, | |
| "loss": 0.094, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 4.6379170056956875e-06, | |
| "loss": 0.0629, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 4.434499593165175e-06, | |
| "loss": 0.081, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 4.2310821806346625e-06, | |
| "loss": 0.0693, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 4.02766476810415e-06, | |
| "loss": 0.0731, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 3.8242473555736375e-06, | |
| "loss": 0.0884, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 3.6208299430431246e-06, | |
| "loss": 0.0618, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 3.417412530512612e-06, | |
| "loss": 0.0915, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 3.213995117982099e-06, | |
| "loss": 0.0928, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 3.010577705451587e-06, | |
| "loss": 0.0846, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.807160292921074e-06, | |
| "loss": 0.0602, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 2.6037428803905617e-06, | |
| "loss": 0.0671, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 9.52, | |
| "learning_rate": 2.4003254678600487e-06, | |
| "loss": 0.0714, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 2.1969080553295362e-06, | |
| "loss": 0.081, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 1.9934906427990237e-06, | |
| "loss": 0.0798, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 1.790073230268511e-06, | |
| "loss": 0.0834, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 1.5866558177379985e-06, | |
| "loss": 0.0907, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 1.3832384052074858e-06, | |
| "loss": 0.0705, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 1.1798209926769731e-06, | |
| "loss": 0.0923, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 9.764035801464606e-07, | |
| "loss": 0.0656, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 7.729861676159479e-07, | |
| "loss": 0.0633, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 9.89, | |
| "learning_rate": 5.695687550854353e-07, | |
| "loss": 0.0727, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 3.661513425549227e-07, | |
| "loss": 0.0921, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 1.627339300244101e-07, | |
| "loss": 0.0661, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_exact_match": 74.47275922671353, | |
| "eval_f1": 88.1676160670247, | |
| "step": 12290 | |
| } | |
| ], | |
| "max_steps": 12290, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.4981977109796864e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |