| { | |
| "best_metric": 0.3479, | |
| "best_model_checkpoint": "wits_checkpoints/checkpoint-63702", | |
| "epoch": 6.0, | |
| "global_step": 63702, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9800000000000004e-05, | |
| "loss": 3.3878, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9800000000000005e-05, | |
| "loss": 2.6013, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.9800000000000003e-05, | |
| "loss": 2.4783, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.98e-05, | |
| "loss": 2.4172, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.98e-05, | |
| "loss": 2.3612, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.990631210371913e-05, | |
| "loss": 2.3416, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.981167786505158e-05, | |
| "loss": 2.2736, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.971704362638403e-05, | |
| "loss": 2.2925, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.962240938771648e-05, | |
| "loss": 2.2689, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.952777514904894e-05, | |
| "loss": 2.2445, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.943314091038138e-05, | |
| "loss": 2.203, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.933850667171382e-05, | |
| "loss": 2.1892, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.924387243304628e-05, | |
| "loss": 2.1888, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.914923819437874e-05, | |
| "loss": 2.1622, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.905460395571118e-05, | |
| "loss": 2.1601, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.895996971704363e-05, | |
| "loss": 2.1562, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.886533547837609e-05, | |
| "loss": 2.1557, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.877070123970853e-05, | |
| "loss": 2.1417, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.867606700104099e-05, | |
| "loss": 2.1268, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.858143276237343e-05, | |
| "loss": 2.1194, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.848679852370588e-05, | |
| "loss": 2.1199, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.839216428503833e-05, | |
| "loss": 2.1339, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.829753004637078e-05, | |
| "loss": 2.1011, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.820289580770324e-05, | |
| "loss": 2.1096, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.810826156903568e-05, | |
| "loss": 2.0886, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.801362733036812e-05, | |
| "loss": 2.0803, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.791899309170058e-05, | |
| "loss": 2.0882, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.782435885303303e-05, | |
| "loss": 2.0889, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.772972461436548e-05, | |
| "loss": 2.0908, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.763509037569793e-05, | |
| "loss": 2.0775, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.754045613703037e-05, | |
| "loss": 2.0718, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.744582189836283e-05, | |
| "loss": 2.0858, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.735118765969529e-05, | |
| "loss": 2.0511, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.725655342102773e-05, | |
| "loss": 2.0592, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.716191918236018e-05, | |
| "loss": 2.0444, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.706728494369263e-05, | |
| "loss": 2.0513, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.697265070502508e-05, | |
| "loss": 2.0333, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.687801646635754e-05, | |
| "loss": 2.033, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.678338222768998e-05, | |
| "loss": 2.0503, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.668874798902244e-05, | |
| "loss": 2.0088, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.659411375035488e-05, | |
| "loss": 2.0314, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.649947951168733e-05, | |
| "loss": 2.02, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.640484527301979e-05, | |
| "loss": 2.0104, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.631021103435224e-05, | |
| "loss": 2.0084, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.621557679568469e-05, | |
| "loss": 1.9997, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.612094255701713e-05, | |
| "loss": 1.9974, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.602630831834958e-05, | |
| "loss": 2.0081, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.593167407968203e-05, | |
| "loss": 2.0234, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.583703984101449e-05, | |
| "loss": 2.0019, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.574240560234694e-05, | |
| "loss": 1.9859, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.564777136367938e-05, | |
| "loss": 1.9965, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.555313712501184e-05, | |
| "loss": 2.0021, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.545850288634428e-05, | |
| "loss": 1.9874, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.536386864767674e-05, | |
| "loss": 1.9913, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.526923440900918e-05, | |
| "loss": 1.9853, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.517460017034163e-05, | |
| "loss": 1.9815, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.507996593167409e-05, | |
| "loss": 1.9733, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.498533169300653e-05, | |
| "loss": 1.9825, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.489069745433899e-05, | |
| "loss": 1.9693, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.479606321567143e-05, | |
| "loss": 1.981, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.470142897700388e-05, | |
| "loss": 1.9738, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.460679473833633e-05, | |
| "loss": 1.9547, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.451216049966879e-05, | |
| "loss": 1.9699, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.441752626100124e-05, | |
| "loss": 1.9392, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.43228920223337e-05, | |
| "loss": 1.971, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.422920412605281e-05, | |
| "loss": 1.9549, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.413456988738526e-05, | |
| "loss": 1.96, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.403993564871771e-05, | |
| "loss": 1.9601, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.394530141005015e-05, | |
| "loss": 1.9535, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.385066717138261e-05, | |
| "loss": 1.9697, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.375603293271507e-05, | |
| "loss": 1.9383, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.366139869404751e-05, | |
| "loss": 1.9382, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.356676445537996e-05, | |
| "loss": 1.9561, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.34721302167124e-05, | |
| "loss": 1.9288, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.337749597804486e-05, | |
| "loss": 1.9319, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 9.328286173937732e-05, | |
| "loss": 1.944, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.318822750070976e-05, | |
| "loss": 1.9289, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.30935932620422e-05, | |
| "loss": 1.9314, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.299895902337466e-05, | |
| "loss": 1.9394, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.290432478470711e-05, | |
| "loss": 1.9239, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.280969054603957e-05, | |
| "loss": 1.9335, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.271505630737202e-05, | |
| "loss": 1.9334, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 9.262042206870445e-05, | |
| "loss": 1.9209, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.252578783003691e-05, | |
| "loss": 1.9132, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.243115359136936e-05, | |
| "loss": 1.9356, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.233651935270181e-05, | |
| "loss": 1.9078, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.224188511403427e-05, | |
| "loss": 1.9389, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.21472508753667e-05, | |
| "loss": 1.9438, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 9.205261663669916e-05, | |
| "loss": 1.939, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.195798239803162e-05, | |
| "loss": 1.9108, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.186334815936406e-05, | |
| "loss": 1.932, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.176871392069652e-05, | |
| "loss": 1.9119, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.167407968202896e-05, | |
| "loss": 1.9181, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.157944544336141e-05, | |
| "loss": 1.9145, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.148481120469387e-05, | |
| "loss": 1.9015, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.139017696602631e-05, | |
| "loss": 1.8919, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.129554272735877e-05, | |
| "loss": 1.9034, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.120090848869121e-05, | |
| "loss": 1.91, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.110627425002366e-05, | |
| "loss": 1.9128, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.101164001135611e-05, | |
| "loss": 1.902, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.091700577268857e-05, | |
| "loss": 1.8826, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.082237153402102e-05, | |
| "loss": 1.8919, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.072773729535346e-05, | |
| "loss": 1.9028, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.06331030566859e-05, | |
| "loss": 1.8857, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.053846881801836e-05, | |
| "loss": 1.9101, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.044383457935082e-05, | |
| "loss": 1.8769, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_BERT_F": 0.8193, | |
| "eval_BERT_P": 0.8208, | |
| "eval_BERT_R": 0.8185, | |
| "eval_R1": 0.5426, | |
| "eval_R2": 0.321, | |
| "eval_RL": 0.5097, | |
| "eval_RLsum": 0.5186, | |
| "eval_loss": 1.827642560005188, | |
| "eval_runtime": 113.4614, | |
| "eval_samples_per_second": 88.136, | |
| "eval_steps_per_second": 1.384, | |
| "step": 10617 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.035109302545662e-05, | |
| "loss": 1.7841, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.025645878678906e-05, | |
| "loss": 1.7802, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.01618245481215e-05, | |
| "loss": 1.7863, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.006719030945396e-05, | |
| "loss": 1.7706, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.997255607078642e-05, | |
| "loss": 1.7725, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.987792183211887e-05, | |
| "loss": 1.795, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.978328759345131e-05, | |
| "loss": 1.7763, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.968865335478377e-05, | |
| "loss": 1.7987, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.959401911611621e-05, | |
| "loss": 1.7734, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.949938487744867e-05, | |
| "loss": 1.786, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.940475063878113e-05, | |
| "loss": 1.7956, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.931011640011356e-05, | |
| "loss": 1.765, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.921548216144602e-05, | |
| "loss": 1.7879, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.912084792277846e-05, | |
| "loss": 1.7747, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.902621368411092e-05, | |
| "loss": 1.7789, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.893157944544337e-05, | |
| "loss": 1.7897, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.88369452067758e-05, | |
| "loss": 1.8051, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.874231096810826e-05, | |
| "loss": 1.7807, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.864767672944072e-05, | |
| "loss": 1.7866, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.855304249077317e-05, | |
| "loss": 1.7999, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.845840825210562e-05, | |
| "loss": 1.7748, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 8.836377401343807e-05, | |
| "loss": 1.7772, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 8.826913977477051e-05, | |
| "loss": 1.8056, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 8.817450553610297e-05, | |
| "loss": 1.7672, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 8.807987129743541e-05, | |
| "loss": 1.8047, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 8.798523705876787e-05, | |
| "loss": 1.7894, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 8.789060282010032e-05, | |
| "loss": 1.7757, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 8.779596858143276e-05, | |
| "loss": 1.7827, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 8.770133434276522e-05, | |
| "loss": 1.7911, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 8.760670010409768e-05, | |
| "loss": 1.8014, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.751206586543012e-05, | |
| "loss": 1.778, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 8.741743162676256e-05, | |
| "loss": 1.791, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 8.732279738809501e-05, | |
| "loss": 1.7901, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 8.722816314942747e-05, | |
| "loss": 1.7916, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 8.713352891075992e-05, | |
| "loss": 1.7922, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 8.703889467209237e-05, | |
| "loss": 1.7845, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 8.694426043342481e-05, | |
| "loss": 1.8001, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 8.684962619475727e-05, | |
| "loss": 1.785, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 8.675499195608971e-05, | |
| "loss": 1.802, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.666035771742217e-05, | |
| "loss": 1.7882, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.656666982114129e-05, | |
| "loss": 1.7933, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 8.647203558247374e-05, | |
| "loss": 1.7917, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 8.63774013438062e-05, | |
| "loss": 1.7825, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 8.628276710513865e-05, | |
| "loss": 1.7712, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.618813286647109e-05, | |
| "loss": 1.7599, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 8.609349862780355e-05, | |
| "loss": 1.7827, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 8.599886438913599e-05, | |
| "loss": 1.7681, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.590423015046845e-05, | |
| "loss": 1.7745, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.58095959118009e-05, | |
| "loss": 1.771, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 8.571496167313334e-05, | |
| "loss": 1.7908, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 8.56203274344658e-05, | |
| "loss": 1.7892, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 8.552569319579824e-05, | |
| "loss": 1.789, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 8.54310589571307e-05, | |
| "loss": 1.7901, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 8.533642471846314e-05, | |
| "loss": 1.7715, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 8.524179047979559e-05, | |
| "loss": 1.7816, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 8.514715624112804e-05, | |
| "loss": 1.7694, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 8.50525220024605e-05, | |
| "loss": 1.7776, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 8.495788776379295e-05, | |
| "loss": 1.7934, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 8.48632535251254e-05, | |
| "loss": 1.755, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.476861928645783e-05, | |
| "loss": 1.7771, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 8.467398504779029e-05, | |
| "loss": 1.7807, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.457935080912275e-05, | |
| "loss": 1.7867, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.448566291284186e-05, | |
| "loss": 1.7787, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 8.439102867417432e-05, | |
| "loss": 1.7719, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 8.429639443550678e-05, | |
| "loss": 1.7861, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 8.420176019683922e-05, | |
| "loss": 1.7558, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 8.410712595817167e-05, | |
| "loss": 1.7663, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 8.401249171950411e-05, | |
| "loss": 1.7819, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.391785748083657e-05, | |
| "loss": 1.7636, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.382322324216903e-05, | |
| "loss": 1.7582, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.372858900350147e-05, | |
| "loss": 1.7835, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.363395476483392e-05, | |
| "loss": 1.7795, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.353932052616637e-05, | |
| "loss": 1.7823, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.344468628749882e-05, | |
| "loss": 1.7758, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.335005204883128e-05, | |
| "loss": 1.7764, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.325541781016372e-05, | |
| "loss": 1.7771, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.316078357149616e-05, | |
| "loss": 1.7762, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 8.306614933282862e-05, | |
| "loss": 1.7685, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.297151509416107e-05, | |
| "loss": 1.7567, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.287688085549352e-05, | |
| "loss": 1.7808, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 8.278224661682598e-05, | |
| "loss": 1.7615, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 8.268761237815843e-05, | |
| "loss": 1.7585, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.259297813949087e-05, | |
| "loss": 1.7915, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.249929024321e-05, | |
| "loss": 1.7715, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.240465600454244e-05, | |
| "loss": 1.7666, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.23100217658749e-05, | |
| "loss": 1.7823, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 8.221538752720734e-05, | |
| "loss": 1.7735, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 8.21207532885398e-05, | |
| "loss": 1.7639, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 8.202611904987225e-05, | |
| "loss": 1.7581, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 8.193148481120469e-05, | |
| "loss": 1.7402, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 8.183685057253715e-05, | |
| "loss": 1.768, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 8.17422163338696e-05, | |
| "loss": 1.7646, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 8.164758209520205e-05, | |
| "loss": 1.7536, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 8.155294785653451e-05, | |
| "loss": 1.7421, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 8.145831361786694e-05, | |
| "loss": 1.7618, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.13636793791994e-05, | |
| "loss": 1.778, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 8.126904514053185e-05, | |
| "loss": 1.7714, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.11744109018643e-05, | |
| "loss": 1.7724, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 8.107977666319676e-05, | |
| "loss": 1.7576, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.09851424245292e-05, | |
| "loss": 1.7465, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 8.089050818586164e-05, | |
| "loss": 1.7562, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 8.07958739471941e-05, | |
| "loss": 1.7439, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 8.070123970852655e-05, | |
| "loss": 1.7543, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.0606605469859e-05, | |
| "loss": 1.7432, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 8.051197123119145e-05, | |
| "loss": 1.7447, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.041828333491058e-05, | |
| "loss": 1.7422, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_BERT_F": 0.8142, | |
| "eval_BERT_P": 0.8118, | |
| "eval_BERT_R": 0.8175, | |
| "eval_R1": 0.5484, | |
| "eval_R2": 0.3326, | |
| "eval_RL": 0.5142, | |
| "eval_RLsum": 0.526, | |
| "eval_loss": 1.7479418516159058, | |
| "eval_runtime": 111.6247, | |
| "eval_samples_per_second": 89.586, | |
| "eval_steps_per_second": 1.406, | |
| "step": 21234 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.032364909624302e-05, | |
| "loss": 1.6542, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.022901485757548e-05, | |
| "loss": 1.6407, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.013438061890792e-05, | |
| "loss": 1.6324, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.003974638024038e-05, | |
| "loss": 1.6509, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.994511214157282e-05, | |
| "loss": 1.6328, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.985047790290527e-05, | |
| "loss": 1.6416, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.975584366423773e-05, | |
| "loss": 1.6527, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.966120942557017e-05, | |
| "loss": 1.6546, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.956657518690263e-05, | |
| "loss": 1.6378, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.947194094823509e-05, | |
| "loss": 1.6581, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.937730670956753e-05, | |
| "loss": 1.6555, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 7.928267247089997e-05, | |
| "loss": 1.6567, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 7.918803823223243e-05, | |
| "loss": 1.654, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 7.909340399356488e-05, | |
| "loss": 1.6529, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 7.899876975489733e-05, | |
| "loss": 1.6621, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 7.890413551622978e-05, | |
| "loss": 1.668, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 7.880950127756222e-05, | |
| "loss": 1.6415, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 7.871486703889468e-05, | |
| "loss": 1.6564, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 7.862023280022712e-05, | |
| "loss": 1.653, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 7.852559856155958e-05, | |
| "loss": 1.6562, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 7.843096432289203e-05, | |
| "loss": 1.662, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 7.833633008422447e-05, | |
| "loss": 1.6417, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 7.824169584555693e-05, | |
| "loss": 1.6693, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 7.814800794927604e-05, | |
| "loss": 1.6486, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 7.80533737106085e-05, | |
| "loss": 1.6592, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 7.795873947194096e-05, | |
| "loss": 1.6648, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.78641052332734e-05, | |
| "loss": 1.6559, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.776947099460586e-05, | |
| "loss": 1.6726, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 7.76748367559383e-05, | |
| "loss": 1.6576, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 7.758020251727075e-05, | |
| "loss": 1.6655, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 7.74855682786032e-05, | |
| "loss": 1.6511, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 7.739093403993565e-05, | |
| "loss": 1.6536, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 7.729629980126811e-05, | |
| "loss": 1.6646, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.720166556260055e-05, | |
| "loss": 1.6448, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.7107031323933e-05, | |
| "loss": 1.6614, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.701334342765212e-05, | |
| "loss": 1.6532, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.691870918898458e-05, | |
| "loss": 1.671, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.682407495031703e-05, | |
| "loss": 1.6616, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.672944071164948e-05, | |
| "loss": 1.6526, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.663480647298193e-05, | |
| "loss": 1.665, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.654017223431437e-05, | |
| "loss": 1.6594, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.644553799564683e-05, | |
| "loss": 1.665, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 7.635090375697927e-05, | |
| "loss": 1.6581, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.625626951831173e-05, | |
| "loss": 1.6541, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.616163527964419e-05, | |
| "loss": 1.6697, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.606700104097663e-05, | |
| "loss": 1.6669, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 7.597236680230908e-05, | |
| "loss": 1.6664, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 7.587773256364152e-05, | |
| "loss": 1.6828, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 7.578309832497398e-05, | |
| "loss": 1.6643, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.568846408630644e-05, | |
| "loss": 1.6883, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 7.559382984763888e-05, | |
| "loss": 1.6593, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 7.549919560897133e-05, | |
| "loss": 1.6561, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.540456137030378e-05, | |
| "loss": 1.6558, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 7.530992713163623e-05, | |
| "loss": 1.6527, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 7.521529289296869e-05, | |
| "loss": 1.6563, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.512065865430113e-05, | |
| "loss": 1.6588, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.502602441563357e-05, | |
| "loss": 1.6759, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.493139017696603e-05, | |
| "loss": 1.6517, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.483675593829848e-05, | |
| "loss": 1.6702, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.474212169963093e-05, | |
| "loss": 1.6555, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.464748746096338e-05, | |
| "loss": 1.6657, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.455285322229582e-05, | |
| "loss": 1.6538, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 7.445821898362828e-05, | |
| "loss": 1.672, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 7.436358474496074e-05, | |
| "loss": 1.6703, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 7.426895050629318e-05, | |
| "loss": 1.6756, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 7.417431626762563e-05, | |
| "loss": 1.6694, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 7.407968202895807e-05, | |
| "loss": 1.6616, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 7.398504779029053e-05, | |
| "loss": 1.6559, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 7.389041355162299e-05, | |
| "loss": 1.6677, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 7.379577931295543e-05, | |
| "loss": 1.668, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 7.370114507428787e-05, | |
| "loss": 1.6657, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 7.360651083562033e-05, | |
| "loss": 1.6685, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 7.351187659695278e-05, | |
| "loss": 1.6651, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.341724235828523e-05, | |
| "loss": 1.6531, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 7.332260811961769e-05, | |
| "loss": 1.6755, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 7.322797388095014e-05, | |
| "loss": 1.6357, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 7.313333964228258e-05, | |
| "loss": 1.6642, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 7.303965174600171e-05, | |
| "loss": 1.6544, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 7.294501750733415e-05, | |
| "loss": 1.6557, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 7.285038326866661e-05, | |
| "loss": 1.6537, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 7.275574902999905e-05, | |
| "loss": 1.6657, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 7.266111479133151e-05, | |
| "loss": 1.6748, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 7.256648055266397e-05, | |
| "loss": 1.6815, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 7.24718463139964e-05, | |
| "loss": 1.6541, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 7.237721207532886e-05, | |
| "loss": 1.6779, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.22825778366613e-05, | |
| "loss": 1.6538, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 7.218794359799376e-05, | |
| "loss": 1.6755, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 7.209330935932622e-05, | |
| "loss": 1.6481, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.199867512065865e-05, | |
| "loss": 1.6627, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.19040408819911e-05, | |
| "loss": 1.6678, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.180940664332356e-05, | |
| "loss": 1.6592, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 7.171477240465601e-05, | |
| "loss": 1.6461, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 7.162013816598847e-05, | |
| "loss": 1.6494, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.15255039273209e-05, | |
| "loss": 1.6681, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.143086968865335e-05, | |
| "loss": 1.6576, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 7.133623544998581e-05, | |
| "loss": 1.659, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 7.124160121131826e-05, | |
| "loss": 1.6453, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 7.114696697265071e-05, | |
| "loss": 1.6817, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 7.105233273398316e-05, | |
| "loss": 1.6541, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 7.09576984953156e-05, | |
| "loss": 1.67, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 7.086306425664806e-05, | |
| "loss": 1.6668, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 7.076843001798052e-05, | |
| "loss": 1.6509, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 7.067474212169963e-05, | |
| "loss": 1.6596, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 7.058010788303209e-05, | |
| "loss": 1.6601, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 7.048547364436453e-05, | |
| "loss": 1.6559, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.039083940569698e-05, | |
| "loss": 1.6738, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_BERT_F": 0.7989, | |
| "eval_BERT_P": 0.786, | |
| "eval_BERT_R": 0.8157, | |
| "eval_R1": 0.5408, | |
| "eval_R2": 0.3319, | |
| "eval_RL": 0.5074, | |
| "eval_RLsum": 0.5192, | |
| "eval_loss": 1.7132327556610107, | |
| "eval_runtime": 115.0682, | |
| "eval_samples_per_second": 86.905, | |
| "eval_steps_per_second": 1.364, | |
| "step": 31851 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.029715150941612e-05, | |
| "loss": 1.586, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.020251727074856e-05, | |
| "loss": 1.5578, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.010788303208101e-05, | |
| "loss": 1.5402, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.001324879341345e-05, | |
| "loss": 1.5364, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.991861455474591e-05, | |
| "loss": 1.546, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.982398031607837e-05, | |
| "loss": 1.5478, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.972934607741081e-05, | |
| "loss": 1.5341, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.963471183874326e-05, | |
| "loss": 1.5632, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.954007760007571e-05, | |
| "loss": 1.5556, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.944544336140816e-05, | |
| "loss": 1.5686, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.935080912274062e-05, | |
| "loss": 1.5604, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.925617488407307e-05, | |
| "loss": 1.5557, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.91615406454055e-05, | |
| "loss": 1.5707, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.906690640673796e-05, | |
| "loss": 1.5569, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 6.89722721680704e-05, | |
| "loss": 1.5547, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 6.887763792940286e-05, | |
| "loss": 1.5618, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 6.878300369073532e-05, | |
| "loss": 1.5604, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 6.868836945206775e-05, | |
| "loss": 1.5614, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 6.859373521340021e-05, | |
| "loss": 1.565, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 6.849910097473267e-05, | |
| "loss": 1.5578, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 6.840446673606511e-05, | |
| "loss": 1.5507, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 6.830983249739757e-05, | |
| "loss": 1.5678, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 6.821519825873e-05, | |
| "loss": 1.5675, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 6.812056402006246e-05, | |
| "loss": 1.5484, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 6.802592978139492e-05, | |
| "loss": 1.5619, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 6.793129554272736e-05, | |
| "loss": 1.5572, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 6.783666130405982e-05, | |
| "loss": 1.5702, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 6.774202706539226e-05, | |
| "loss": 1.5541, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 6.76473928267247e-05, | |
| "loss": 1.5692, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 6.755275858805716e-05, | |
| "loss": 1.5601, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 6.745812434938962e-05, | |
| "loss": 1.5777, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 6.736349011072207e-05, | |
| "loss": 1.5749, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 6.726885587205451e-05, | |
| "loss": 1.5715, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 6.717422163338695e-05, | |
| "loss": 1.5619, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 6.707958739471941e-05, | |
| "loss": 1.5763, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 6.698495315605187e-05, | |
| "loss": 1.5734, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 6.689031891738431e-05, | |
| "loss": 1.5816, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 6.679568467871676e-05, | |
| "loss": 1.5643, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 6.670105044004922e-05, | |
| "loss": 1.5756, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 6.660641620138166e-05, | |
| "loss": 1.5827, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 6.651178196271412e-05, | |
| "loss": 1.5756, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 6.641714772404656e-05, | |
| "loss": 1.5854, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 6.6322513485379e-05, | |
| "loss": 1.5782, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 6.622787924671146e-05, | |
| "loss": 1.612, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 6.613324500804391e-05, | |
| "loss": 1.5981, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 6.603861076937637e-05, | |
| "loss": 1.5692, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 6.594397653070882e-05, | |
| "loss": 1.58, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.585028863442794e-05, | |
| "loss": 1.5951, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 6.57556543957604e-05, | |
| "loss": 1.5739, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 6.566102015709284e-05, | |
| "loss": 1.573, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 6.556638591842528e-05, | |
| "loss": 1.5624, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 6.547175167975774e-05, | |
| "loss": 1.5822, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 6.537711744109019e-05, | |
| "loss": 1.5994, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 6.528248320242264e-05, | |
| "loss": 1.5761, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 6.518784896375509e-05, | |
| "loss": 1.5654, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 6.509321472508753e-05, | |
| "loss": 1.5765, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 6.499858048641999e-05, | |
| "loss": 1.5618, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 6.490394624775245e-05, | |
| "loss": 1.5787, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 6.480931200908489e-05, | |
| "loss": 1.568, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.471467777041734e-05, | |
| "loss": 1.5724, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 6.462004353174978e-05, | |
| "loss": 1.585, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 6.452540929308224e-05, | |
| "loss": 1.595, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 6.44307750544147e-05, | |
| "loss": 1.5883, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 6.433614081574714e-05, | |
| "loss": 1.5744, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 6.42415065770796e-05, | |
| "loss": 1.5909, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 6.414687233841204e-05, | |
| "loss": 1.6012, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 6.405223809974449e-05, | |
| "loss": 1.5996, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 6.395760386107694e-05, | |
| "loss": 1.5837, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 6.386296962240939e-05, | |
| "loss": 1.5804, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 6.376833538374185e-05, | |
| "loss": 1.5834, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 6.367370114507429e-05, | |
| "loss": 1.5674, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 6.357906690640673e-05, | |
| "loss": 1.5805, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 6.348443266773919e-05, | |
| "loss": 1.5676, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 6.338979842907165e-05, | |
| "loss": 1.5772, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 6.32951641904041e-05, | |
| "loss": 1.5804, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 6.320052995173654e-05, | |
| "loss": 1.5923, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 6.3105895713069e-05, | |
| "loss": 1.5926, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 6.301126147440144e-05, | |
| "loss": 1.5947, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 6.29166272357339e-05, | |
| "loss": 1.5765, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 6.282199299706634e-05, | |
| "loss": 1.58, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 6.272735875839879e-05, | |
| "loss": 1.5761, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 6.263367086211793e-05, | |
| "loss": 1.5837, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 6.253903662345036e-05, | |
| "loss": 1.5883, | |
| "step": 40100 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 6.244440238478282e-05, | |
| "loss": 1.5805, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 6.234976814611527e-05, | |
| "loss": 1.5897, | |
| "step": 40300 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 6.225513390744772e-05, | |
| "loss": 1.5846, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 6.216049966878018e-05, | |
| "loss": 1.5672, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 6.20658654301126e-05, | |
| "loss": 1.5917, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 6.197123119144506e-05, | |
| "loss": 1.5861, | |
| "step": 40700 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 6.187659695277752e-05, | |
| "loss": 1.5601, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 6.178196271410997e-05, | |
| "loss": 1.5743, | |
| "step": 40900 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 6.168732847544242e-05, | |
| "loss": 1.564, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 6.159269423677487e-05, | |
| "loss": 1.5932, | |
| "step": 41100 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 6.149805999810731e-05, | |
| "loss": 1.5772, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 6.140342575943977e-05, | |
| "loss": 1.5885, | |
| "step": 41300 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 6.130879152077221e-05, | |
| "loss": 1.5848, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 6.121415728210467e-05, | |
| "loss": 1.5827, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 6.111952304343712e-05, | |
| "loss": 1.5894, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 6.102488880476957e-05, | |
| "loss": 1.5962, | |
| "step": 41700 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 6.093025456610202e-05, | |
| "loss": 1.5792, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 6.0835620327434476e-05, | |
| "loss": 1.5685, | |
| "step": 41900 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 6.074098608876691e-05, | |
| "loss": 1.6087, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.064635185009937e-05, | |
| "loss": 1.5927, | |
| "step": 42100 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.055171761143182e-05, | |
| "loss": 1.5878, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 6.0457083372764266e-05, | |
| "loss": 1.5851, | |
| "step": 42300 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 6.03633954764834e-05, | |
| "loss": 1.5899, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_BERT_F": 0.7955, | |
| "eval_BERT_P": 0.7803, | |
| "eval_BERT_R": 0.8157, | |
| "eval_R1": 0.5558, | |
| "eval_R2": 0.3476, | |
| "eval_RL": 0.5232, | |
| "eval_RLsum": 0.5346, | |
| "eval_loss": 1.6942824125289917, | |
| "eval_runtime": 117.0149, | |
| "eval_samples_per_second": 85.459, | |
| "eval_steps_per_second": 1.342, | |
| "step": 42468 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 6.026876123781584e-05, | |
| "loss": 1.5358, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 6.0174126999148296e-05, | |
| "loss": 1.4762, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 6.007949276048075e-05, | |
| "loss": 1.4729, | |
| "step": 42700 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.998485852181319e-05, | |
| "loss": 1.4913, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.989022428314565e-05, | |
| "loss": 1.4785, | |
| "step": 42900 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 5.97955900444781e-05, | |
| "loss": 1.4794, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 5.9700955805810544e-05, | |
| "loss": 1.5019, | |
| "step": 43100 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 5.9606321567142995e-05, | |
| "loss": 1.4861, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 5.951168732847544e-05, | |
| "loss": 1.4744, | |
| "step": 43300 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 5.94170530898079e-05, | |
| "loss": 1.4771, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 5.932241885114035e-05, | |
| "loss": 1.5019, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 5.922778461247279e-05, | |
| "loss": 1.4813, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 5.913315037380524e-05, | |
| "loss": 1.5028, | |
| "step": 43700 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 5.90385161351377e-05, | |
| "loss": 1.4876, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 5.8943881896470145e-05, | |
| "loss": 1.4936, | |
| "step": 43900 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 5.8849247657802596e-05, | |
| "loss": 1.4916, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 5.875461341913504e-05, | |
| "loss": 1.5021, | |
| "step": 44100 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 5.86599791804675e-05, | |
| "loss": 1.4823, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 5.856534494179995e-05, | |
| "loss": 1.4936, | |
| "step": 44300 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 5.847071070313239e-05, | |
| "loss": 1.4905, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 5.8376076464464844e-05, | |
| "loss": 1.4979, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 5.82814422257973e-05, | |
| "loss": 1.4852, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 5.8186807987129746e-05, | |
| "loss": 1.4889, | |
| "step": 44700 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 5.80921737484622e-05, | |
| "loss": 1.5144, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 5.7997539509794655e-05, | |
| "loss": 1.49, | |
| "step": 44900 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 5.790290527112709e-05, | |
| "loss": 1.5009, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 5.780921737484623e-05, | |
| "loss": 1.508, | |
| "step": 45100 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 5.771458313617867e-05, | |
| "loss": 1.4978, | |
| "step": 45200 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 5.761994889751112e-05, | |
| "loss": 1.4981, | |
| "step": 45300 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 5.752531465884358e-05, | |
| "loss": 1.4941, | |
| "step": 45400 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 5.743068042017602e-05, | |
| "loss": 1.4977, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 5.7336046181508475e-05, | |
| "loss": 1.4798, | |
| "step": 45600 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 5.7241411942840926e-05, | |
| "loss": 1.5163, | |
| "step": 45700 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 5.714677770417337e-05, | |
| "loss": 1.4993, | |
| "step": 45800 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 5.705214346550583e-05, | |
| "loss": 1.5128, | |
| "step": 45900 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 5.6957509226838265e-05, | |
| "loss": 1.5052, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 5.686287498817072e-05, | |
| "loss": 1.4949, | |
| "step": 46100 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 5.6768240749503174e-05, | |
| "loss": 1.5059, | |
| "step": 46200 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 5.667360651083562e-05, | |
| "loss": 1.5189, | |
| "step": 46300 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 5.6578972272168076e-05, | |
| "loss": 1.5086, | |
| "step": 46400 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 5.648433803350053e-05, | |
| "loss": 1.5097, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 5.638970379483297e-05, | |
| "loss": 1.5071, | |
| "step": 46600 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 5.629506955616542e-05, | |
| "loss": 1.4937, | |
| "step": 46700 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 5.6200435317497866e-05, | |
| "loss": 1.4931, | |
| "step": 46800 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 5.6105801078830324e-05, | |
| "loss": 1.5177, | |
| "step": 46900 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 5.6011166840162775e-05, | |
| "loss": 1.5191, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 5.5917478943881896e-05, | |
| "loss": 1.5005, | |
| "step": 47100 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 5.582284470521435e-05, | |
| "loss": 1.4894, | |
| "step": 47200 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 5.5728210466546805e-05, | |
| "loss": 1.4984, | |
| "step": 47300 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 5.563357622787925e-05, | |
| "loss": 1.5259, | |
| "step": 47400 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 5.55389419892117e-05, | |
| "loss": 1.5046, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 5.5444307750544144e-05, | |
| "loss": 1.5143, | |
| "step": 47600 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 5.53496735118766e-05, | |
| "loss": 1.4972, | |
| "step": 47700 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 5.525503927320905e-05, | |
| "loss": 1.5012, | |
| "step": 47800 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 5.51604050345415e-05, | |
| "loss": 1.5013, | |
| "step": 47900 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 5.506577079587395e-05, | |
| "loss": 1.5109, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 5.4971136557206406e-05, | |
| "loss": 1.5002, | |
| "step": 48100 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 5.487650231853885e-05, | |
| "loss": 1.5198, | |
| "step": 48200 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 5.47818680798713e-05, | |
| "loss": 1.5106, | |
| "step": 48300 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 5.468723384120376e-05, | |
| "loss": 1.5139, | |
| "step": 48400 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 5.4592599602536196e-05, | |
| "loss": 1.5, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 5.4497965363868654e-05, | |
| "loss": 1.5069, | |
| "step": 48600 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 5.44033311252011e-05, | |
| "loss": 1.5165, | |
| "step": 48700 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 5.430869688653355e-05, | |
| "loss": 1.5096, | |
| "step": 48800 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 5.421406264786601e-05, | |
| "loss": 1.5159, | |
| "step": 48900 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 5.4119428409198444e-05, | |
| "loss": 1.5074, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 5.40247941705309e-05, | |
| "loss": 1.5173, | |
| "step": 49100 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 5.393015993186335e-05, | |
| "loss": 1.5039, | |
| "step": 49200 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 5.38355256931958e-05, | |
| "loss": 1.5085, | |
| "step": 49300 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 5.3740891454528255e-05, | |
| "loss": 1.5142, | |
| "step": 49400 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 5.364625721586069e-05, | |
| "loss": 1.509, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 5.355162297719315e-05, | |
| "loss": 1.5064, | |
| "step": 49600 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 5.34569887385256e-05, | |
| "loss": 1.5169, | |
| "step": 49700 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 5.3362354499858045e-05, | |
| "loss": 1.523, | |
| "step": 49800 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 5.32677202611905e-05, | |
| "loss": 1.5003, | |
| "step": 49900 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 5.3173086022522954e-05, | |
| "loss": 1.4882, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 5.30784517838554e-05, | |
| "loss": 1.5001, | |
| "step": 50100 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 5.298381754518785e-05, | |
| "loss": 1.5085, | |
| "step": 50200 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 5.288918330652031e-05, | |
| "loss": 1.5144, | |
| "step": 50300 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 5.279454906785275e-05, | |
| "loss": 1.515, | |
| "step": 50400 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 5.26999148291852e-05, | |
| "loss": 1.5394, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 5.260528059051765e-05, | |
| "loss": 1.5241, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 5.2511592694236774e-05, | |
| "loss": 1.5094, | |
| "step": 50700 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 5.241695845556923e-05, | |
| "loss": 1.5208, | |
| "step": 50800 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 5.2322324216901676e-05, | |
| "loss": 1.5185, | |
| "step": 50900 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 5.222768997823413e-05, | |
| "loss": 1.5055, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 5.2133055739566585e-05, | |
| "loss": 1.5276, | |
| "step": 51100 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 5.203842150089903e-05, | |
| "loss": 1.5167, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 5.194378726223148e-05, | |
| "loss": 1.5258, | |
| "step": 51300 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 5.1849153023563924e-05, | |
| "loss": 1.5197, | |
| "step": 51400 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 5.1754518784896375e-05, | |
| "loss": 1.5246, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.165988454622883e-05, | |
| "loss": 1.517, | |
| "step": 51600 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 5.156525030756128e-05, | |
| "loss": 1.5076, | |
| "step": 51700 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 5.147061606889373e-05, | |
| "loss": 1.5058, | |
| "step": 51800 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 5.1375981830226186e-05, | |
| "loss": 1.5357, | |
| "step": 51900 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 5.1281347591558623e-05, | |
| "loss": 1.5151, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 5.118671335289108e-05, | |
| "loss": 1.5184, | |
| "step": 52100 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 5.1092079114223525e-05, | |
| "loss": 1.5095, | |
| "step": 52200 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 5.0997444875555976e-05, | |
| "loss": 1.5093, | |
| "step": 52300 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.0902810636888434e-05, | |
| "loss": 1.52, | |
| "step": 52400 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.080817639822087e-05, | |
| "loss": 1.5037, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 5.071354215955333e-05, | |
| "loss": 1.5203, | |
| "step": 52600 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 5.061985426327246e-05, | |
| "loss": 1.5042, | |
| "step": 52700 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 5.05252200246049e-05, | |
| "loss": 1.5167, | |
| "step": 52800 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 5.043058578593736e-05, | |
| "loss": 1.512, | |
| "step": 52900 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 5.0335951547269796e-05, | |
| "loss": 1.5099, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_BERT_F": 0.7784, | |
| "eval_BERT_P": 0.7522, | |
| "eval_BERT_R": 0.813, | |
| "eval_R1": 0.5194, | |
| "eval_R2": 0.3241, | |
| "eval_RL": 0.4874, | |
| "eval_RLsum": 0.4989, | |
| "eval_loss": 1.6852939128875732, | |
| "eval_runtime": 118.2076, | |
| "eval_samples_per_second": 84.597, | |
| "eval_steps_per_second": 1.328, | |
| "step": 53085 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 5.024226365098893e-05, | |
| "loss": 1.4963, | |
| "step": 53100 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.014762941232138e-05, | |
| "loss": 1.4269, | |
| "step": 53200 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 5.0052995173653826e-05, | |
| "loss": 1.4098, | |
| "step": 53300 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 4.9958360934986284e-05, | |
| "loss": 1.3929, | |
| "step": 53400 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 4.986372669631873e-05, | |
| "loss": 1.4166, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 4.976909245765118e-05, | |
| "loss": 1.4201, | |
| "step": 53600 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 4.967445821898363e-05, | |
| "loss": 1.443, | |
| "step": 53700 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 4.957982398031608e-05, | |
| "loss": 1.4142, | |
| "step": 53800 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 4.948518974164853e-05, | |
| "loss": 1.4207, | |
| "step": 53900 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 4.9390555502980976e-05, | |
| "loss": 1.415, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 4.9295921264313434e-05, | |
| "loss": 1.4143, | |
| "step": 54100 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 4.9201287025645885e-05, | |
| "loss": 1.4516, | |
| "step": 54200 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 4.910665278697833e-05, | |
| "loss": 1.422, | |
| "step": 54300 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 4.901201854831078e-05, | |
| "loss": 1.4181, | |
| "step": 54400 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 4.891738430964323e-05, | |
| "loss": 1.4269, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 4.882275007097568e-05, | |
| "loss": 1.4393, | |
| "step": 54600 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 4.872811583230813e-05, | |
| "loss": 1.4353, | |
| "step": 54700 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 4.8633481593640584e-05, | |
| "loss": 1.4378, | |
| "step": 54800 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 4.8538847354973035e-05, | |
| "loss": 1.4281, | |
| "step": 54900 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 4.844421311630548e-05, | |
| "loss": 1.4364, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 4.834957887763793e-05, | |
| "loss": 1.4453, | |
| "step": 55100 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 4.825494463897038e-05, | |
| "loss": 1.428, | |
| "step": 55200 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 4.816031040030283e-05, | |
| "loss": 1.4388, | |
| "step": 55300 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 4.806567616163528e-05, | |
| "loss": 1.4495, | |
| "step": 55400 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 4.797104192296773e-05, | |
| "loss": 1.4288, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 4.7876407684300185e-05, | |
| "loss": 1.4165, | |
| "step": 55600 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 4.7781773445632636e-05, | |
| "loss": 1.4271, | |
| "step": 55700 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 4.768713920696508e-05, | |
| "loss": 1.4435, | |
| "step": 55800 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 4.759250496829753e-05, | |
| "loss": 1.459, | |
| "step": 55900 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 4.749787072962998e-05, | |
| "loss": 1.4589, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 4.740323649096243e-05, | |
| "loss": 1.4366, | |
| "step": 56100 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 4.7308602252294884e-05, | |
| "loss": 1.4401, | |
| "step": 56200 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 4.721396801362733e-05, | |
| "loss": 1.4442, | |
| "step": 56300 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 4.7119333774959786e-05, | |
| "loss": 1.4414, | |
| "step": 56400 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 4.702469953629223e-05, | |
| "loss": 1.4331, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 4.693006529762468e-05, | |
| "loss": 1.4445, | |
| "step": 56600 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 4.683543105895713e-05, | |
| "loss": 1.4418, | |
| "step": 56700 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 4.674079682028958e-05, | |
| "loss": 1.4575, | |
| "step": 56800 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 4.6646162581622034e-05, | |
| "loss": 1.4358, | |
| "step": 56900 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 4.655152834295448e-05, | |
| "loss": 1.447, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 4.6456894104286936e-05, | |
| "loss": 1.4421, | |
| "step": 57100 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 4.636225986561939e-05, | |
| "loss": 1.448, | |
| "step": 57200 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 4.626762562695183e-05, | |
| "loss": 1.4538, | |
| "step": 57300 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 4.617299138828428e-05, | |
| "loss": 1.441, | |
| "step": 57400 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 4.6078357149616733e-05, | |
| "loss": 1.4521, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 4.5983722910949184e-05, | |
| "loss": 1.4388, | |
| "step": 57600 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 4.5889088672281635e-05, | |
| "loss": 1.4403, | |
| "step": 57700 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 4.579445443361408e-05, | |
| "loss": 1.434, | |
| "step": 57800 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 4.569982019494654e-05, | |
| "loss": 1.4614, | |
| "step": 57900 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 4.560518595627898e-05, | |
| "loss": 1.4574, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 4.551055171761143e-05, | |
| "loss": 1.441, | |
| "step": 58100 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 4.5415917478943884e-05, | |
| "loss": 1.4481, | |
| "step": 58200 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 4.5321283240276335e-05, | |
| "loss": 1.4533, | |
| "step": 58300 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 4.5226649001608786e-05, | |
| "loss": 1.4491, | |
| "step": 58400 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 4.5132961105327906e-05, | |
| "loss": 1.4438, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 4.503832686666036e-05, | |
| "loss": 1.4657, | |
| "step": 58600 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 4.494369262799281e-05, | |
| "loss": 1.4453, | |
| "step": 58700 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 4.484905838932526e-05, | |
| "loss": 1.4498, | |
| "step": 58800 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 4.475442415065771e-05, | |
| "loss": 1.4508, | |
| "step": 58900 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 4.4659789911990154e-05, | |
| "loss": 1.4477, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 4.456515567332261e-05, | |
| "loss": 1.4328, | |
| "step": 59100 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 4.447052143465506e-05, | |
| "loss": 1.4432, | |
| "step": 59200 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 4.437588719598751e-05, | |
| "loss": 1.4596, | |
| "step": 59300 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 4.4281252957319965e-05, | |
| "loss": 1.4558, | |
| "step": 59400 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 4.418661871865241e-05, | |
| "loss": 1.4636, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 4.409198447998486e-05, | |
| "loss": 1.4446, | |
| "step": 59600 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 4.399735024131731e-05, | |
| "loss": 1.4803, | |
| "step": 59700 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 4.390271600264976e-05, | |
| "loss": 1.4664, | |
| "step": 59800 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 4.3808081763982213e-05, | |
| "loss": 1.4365, | |
| "step": 59900 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 4.371344752531466e-05, | |
| "loss": 1.4544, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 4.361881328664711e-05, | |
| "loss": 1.4629, | |
| "step": 60100 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 4.3524179047979566e-05, | |
| "loss": 1.4479, | |
| "step": 60200 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 4.342954480931201e-05, | |
| "loss": 1.4689, | |
| "step": 60300 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 4.333491057064446e-05, | |
| "loss": 1.4503, | |
| "step": 60400 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 4.3240276331976906e-05, | |
| "loss": 1.4436, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 4.3145642093309364e-05, | |
| "loss": 1.4532, | |
| "step": 60600 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 4.3051007854641815e-05, | |
| "loss": 1.4579, | |
| "step": 60700 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 4.295637361597426e-05, | |
| "loss": 1.4457, | |
| "step": 60800 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 4.286173937730671e-05, | |
| "loss": 1.4546, | |
| "step": 60900 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 4.276710513863916e-05, | |
| "loss": 1.4429, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 4.267247089997161e-05, | |
| "loss": 1.4435, | |
| "step": 61100 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 4.257783666130406e-05, | |
| "loss": 1.4409, | |
| "step": 61200 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 4.2483202422636514e-05, | |
| "loss": 1.4432, | |
| "step": 61300 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 4.2388568183968965e-05, | |
| "loss": 1.442, | |
| "step": 61400 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 4.229393394530141e-05, | |
| "loss": 1.4587, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 4.219929970663386e-05, | |
| "loss": 1.4599, | |
| "step": 61600 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 4.210466546796632e-05, | |
| "loss": 1.4511, | |
| "step": 61700 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 4.201003122929876e-05, | |
| "loss": 1.4434, | |
| "step": 61800 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 4.191539699063121e-05, | |
| "loss": 1.4735, | |
| "step": 61900 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 4.182076275196366e-05, | |
| "loss": 1.4462, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 4.1726128513296115e-05, | |
| "loss": 1.4536, | |
| "step": 62100 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 4.1631494274628566e-05, | |
| "loss": 1.4692, | |
| "step": 62200 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 4.153686003596101e-05, | |
| "loss": 1.468, | |
| "step": 62300 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 4.144222579729346e-05, | |
| "loss": 1.4483, | |
| "step": 62400 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 4.134853790101259e-05, | |
| "loss": 1.4462, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 4.125390366234504e-05, | |
| "loss": 1.4618, | |
| "step": 62600 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 4.115926942367749e-05, | |
| "loss": 1.4566, | |
| "step": 62700 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 4.1064635185009935e-05, | |
| "loss": 1.4602, | |
| "step": 62800 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 4.097000094634239e-05, | |
| "loss": 1.4431, | |
| "step": 62900 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 4.087536670767484e-05, | |
| "loss": 1.4559, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 4.078073246900729e-05, | |
| "loss": 1.4718, | |
| "step": 63100 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 4.068609823033974e-05, | |
| "loss": 1.4524, | |
| "step": 63200 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 4.059146399167219e-05, | |
| "loss": 1.4366, | |
| "step": 63300 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 4.049682975300464e-05, | |
| "loss": 1.4507, | |
| "step": 63400 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 4.0402195514337085e-05, | |
| "loss": 1.4674, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 4.0307561275669536e-05, | |
| "loss": 1.4687, | |
| "step": 63600 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 4.0212927037001994e-05, | |
| "loss": 1.4647, | |
| "step": 63700 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_BERT_F": 0.7857, | |
| "eval_BERT_P": 0.7629, | |
| "eval_BERT_R": 0.8159, | |
| "eval_R1": 0.5521, | |
| "eval_R2": 0.3479, | |
| "eval_RL": 0.5202, | |
| "eval_RLsum": 0.5309, | |
| "eval_loss": 1.6809284687042236, | |
| "eval_runtime": 117.0136, | |
| "eval_samples_per_second": 85.46, | |
| "eval_steps_per_second": 1.342, | |
| "step": 63702 | |
| } | |
| ], | |
| "max_steps": 106170, | |
| "num_train_epochs": 10, | |
| "total_flos": 2.4856248172570214e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |