samtuckervegan's picture
Upload folder using huggingface_hub
967740c verified
{
"best_metric": 0.01995982974767685,
"best_model_checkpoint": "trustworthiness-longformer/checkpoint-1995",
"epoch": 3.0,
"eval_steps": 500,
"global_step": 1995,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03759398496240601,
"grad_norm": 9.034588813781738,
"learning_rate": 4.229323308270677e-06,
"loss": 0.5584,
"step": 25
},
{
"epoch": 0.07518796992481203,
"grad_norm": 1.967883586883545,
"learning_rate": 8.1203007518797e-06,
"loss": 0.1203,
"step": 50
},
{
"epoch": 0.11278195488721804,
"grad_norm": 1.5318199396133423,
"learning_rate": 1.2349624060150378e-05,
"loss": 0.0532,
"step": 75
},
{
"epoch": 0.15037593984962405,
"grad_norm": 0.6738603115081787,
"learning_rate": 1.6578947368421053e-05,
"loss": 0.0438,
"step": 100
},
{
"epoch": 0.18796992481203006,
"grad_norm": 2.247737169265747,
"learning_rate": 2.080827067669173e-05,
"loss": 0.0443,
"step": 125
},
{
"epoch": 0.22556390977443608,
"grad_norm": 1.5706892013549805,
"learning_rate": 2.5037593984962406e-05,
"loss": 0.0422,
"step": 150
},
{
"epoch": 0.2631578947368421,
"grad_norm": 1.9436813592910767,
"learning_rate": 2.9266917293233085e-05,
"loss": 0.0403,
"step": 175
},
{
"epoch": 0.3007518796992481,
"grad_norm": 1.7137566804885864,
"learning_rate": 3.349624060150376e-05,
"loss": 0.0389,
"step": 200
},
{
"epoch": 0.3383458646616541,
"grad_norm": 1.5335218906402588,
"learning_rate": 3.772556390977444e-05,
"loss": 0.0369,
"step": 225
},
{
"epoch": 0.37593984962406013,
"grad_norm": 2.1619019508361816,
"learning_rate": 4.195488721804511e-05,
"loss": 0.0357,
"step": 250
},
{
"epoch": 0.41353383458646614,
"grad_norm": 0.8143567442893982,
"learning_rate": 4.486842105263158e-05,
"loss": 0.034,
"step": 275
},
{
"epoch": 0.45112781954887216,
"grad_norm": 0.34969833493232727,
"learning_rate": 4.4398496240601505e-05,
"loss": 0.0309,
"step": 300
},
{
"epoch": 0.48872180451127817,
"grad_norm": 0.6551300883293152,
"learning_rate": 4.392857142857143e-05,
"loss": 0.0315,
"step": 325
},
{
"epoch": 0.5263157894736842,
"grad_norm": 1.1032406091690063,
"learning_rate": 4.3458646616541354e-05,
"loss": 0.0309,
"step": 350
},
{
"epoch": 0.5639097744360902,
"grad_norm": 0.3101660907268524,
"learning_rate": 4.2988721804511285e-05,
"loss": 0.03,
"step": 375
},
{
"epoch": 0.6015037593984962,
"grad_norm": 0.9715960621833801,
"learning_rate": 4.251879699248121e-05,
"loss": 0.029,
"step": 400
},
{
"epoch": 0.6390977443609023,
"grad_norm": 1.2566590309143066,
"learning_rate": 4.2048872180451127e-05,
"loss": 0.0311,
"step": 425
},
{
"epoch": 0.6766917293233082,
"grad_norm": 0.9328742027282715,
"learning_rate": 4.157894736842105e-05,
"loss": 0.0283,
"step": 450
},
{
"epoch": 0.7142857142857143,
"grad_norm": 1.1680524349212646,
"learning_rate": 4.110902255639098e-05,
"loss": 0.0279,
"step": 475
},
{
"epoch": 0.7518796992481203,
"grad_norm": 0.6984049081802368,
"learning_rate": 4.0639097744360906e-05,
"loss": 0.0276,
"step": 500
},
{
"epoch": 0.7894736842105263,
"grad_norm": 0.7427186965942383,
"learning_rate": 4.016917293233083e-05,
"loss": 0.0284,
"step": 525
},
{
"epoch": 0.8270676691729323,
"grad_norm": 0.9322423338890076,
"learning_rate": 3.9699248120300755e-05,
"loss": 0.0267,
"step": 550
},
{
"epoch": 0.8646616541353384,
"grad_norm": 0.5467056035995483,
"learning_rate": 3.922932330827068e-05,
"loss": 0.0274,
"step": 575
},
{
"epoch": 0.9022556390977443,
"grad_norm": 0.7492877840995789,
"learning_rate": 3.87593984962406e-05,
"loss": 0.0266,
"step": 600
},
{
"epoch": 0.9398496240601504,
"grad_norm": 1.0156948566436768,
"learning_rate": 3.828947368421053e-05,
"loss": 0.0257,
"step": 625
},
{
"epoch": 0.9774436090225563,
"grad_norm": 2.6121363639831543,
"learning_rate": 3.781954887218045e-05,
"loss": 0.0255,
"step": 650
},
{
"epoch": 1.0,
"eval_explained_variance": 0.43071454763412476,
"eval_loss": 0.021834315732121468,
"eval_mae": 0.11838787794113159,
"eval_mse": 0.021832076832652092,
"eval_r2": 0.4306418299674988,
"eval_rmse": 0.14775681653531958,
"eval_runtime": 333.3711,
"eval_samples_per_second": 63.818,
"eval_steps_per_second": 0.999,
"step": 665
},
{
"epoch": 1.0150375939849625,
"grad_norm": 1.001768708229065,
"learning_rate": 3.734962406015038e-05,
"loss": 0.0248,
"step": 675
},
{
"epoch": 1.0526315789473684,
"grad_norm": 0.44310006499290466,
"learning_rate": 3.687969924812031e-05,
"loss": 0.0248,
"step": 700
},
{
"epoch": 1.0902255639097744,
"grad_norm": 0.7214398384094238,
"learning_rate": 3.6409774436090224e-05,
"loss": 0.0241,
"step": 725
},
{
"epoch": 1.1278195488721805,
"grad_norm": 0.6446682810783386,
"learning_rate": 3.593984962406015e-05,
"loss": 0.0238,
"step": 750
},
{
"epoch": 1.1654135338345863,
"grad_norm": 0.539253830909729,
"learning_rate": 3.546992481203008e-05,
"loss": 0.0261,
"step": 775
},
{
"epoch": 1.2030075187969924,
"grad_norm": 0.4309803247451782,
"learning_rate": 3.5000000000000004e-05,
"loss": 0.0243,
"step": 800
},
{
"epoch": 1.2406015037593985,
"grad_norm": 0.2850622534751892,
"learning_rate": 3.453007518796993e-05,
"loss": 0.0239,
"step": 825
},
{
"epoch": 1.2781954887218046,
"grad_norm": 0.4944869577884674,
"learning_rate": 3.406015037593985e-05,
"loss": 0.0248,
"step": 850
},
{
"epoch": 1.3157894736842106,
"grad_norm": 0.26742395758628845,
"learning_rate": 3.3590225563909776e-05,
"loss": 0.0234,
"step": 875
},
{
"epoch": 1.3533834586466165,
"grad_norm": 1.0373315811157227,
"learning_rate": 3.31203007518797e-05,
"loss": 0.0243,
"step": 900
},
{
"epoch": 1.3909774436090225,
"grad_norm": 0.6992561221122742,
"learning_rate": 3.2650375939849625e-05,
"loss": 0.0234,
"step": 925
},
{
"epoch": 1.4285714285714286,
"grad_norm": 0.6922001242637634,
"learning_rate": 3.218045112781955e-05,
"loss": 0.024,
"step": 950
},
{
"epoch": 1.4661654135338344,
"grad_norm": 0.7946988940238953,
"learning_rate": 3.171052631578948e-05,
"loss": 0.0261,
"step": 975
},
{
"epoch": 1.5037593984962405,
"grad_norm": 0.24744807183742523,
"learning_rate": 3.1240601503759404e-05,
"loss": 0.0234,
"step": 1000
},
{
"epoch": 1.5413533834586466,
"grad_norm": 0.6126484870910645,
"learning_rate": 3.077067669172932e-05,
"loss": 0.0237,
"step": 1025
},
{
"epoch": 1.5789473684210527,
"grad_norm": 0.9163932800292969,
"learning_rate": 3.030075187969925e-05,
"loss": 0.0251,
"step": 1050
},
{
"epoch": 1.6165413533834587,
"grad_norm": 0.42940250039100647,
"learning_rate": 2.9830827067669177e-05,
"loss": 0.025,
"step": 1075
},
{
"epoch": 1.6541353383458648,
"grad_norm": 0.21495333313941956,
"learning_rate": 2.93609022556391e-05,
"loss": 0.0229,
"step": 1100
},
{
"epoch": 1.6917293233082706,
"grad_norm": 0.20642122626304626,
"learning_rate": 2.8890977443609026e-05,
"loss": 0.0223,
"step": 1125
},
{
"epoch": 1.7293233082706767,
"grad_norm": 0.2935808002948761,
"learning_rate": 2.8421052631578946e-05,
"loss": 0.0228,
"step": 1150
},
{
"epoch": 1.7669172932330826,
"grad_norm": 0.6718006730079651,
"learning_rate": 2.7951127819548878e-05,
"loss": 0.0228,
"step": 1175
},
{
"epoch": 1.8045112781954886,
"grad_norm": 0.2039840817451477,
"learning_rate": 2.74812030075188e-05,
"loss": 0.0235,
"step": 1200
},
{
"epoch": 1.8421052631578947,
"grad_norm": 0.6539952754974365,
"learning_rate": 2.7011278195488723e-05,
"loss": 0.0232,
"step": 1225
},
{
"epoch": 1.8796992481203008,
"grad_norm": 0.20303431153297424,
"learning_rate": 2.6541353383458647e-05,
"loss": 0.0243,
"step": 1250
},
{
"epoch": 1.9172932330827068,
"grad_norm": 1.0404738187789917,
"learning_rate": 2.6071428571428574e-05,
"loss": 0.0235,
"step": 1275
},
{
"epoch": 1.954887218045113,
"grad_norm": 0.4455120861530304,
"learning_rate": 2.56015037593985e-05,
"loss": 0.0234,
"step": 1300
},
{
"epoch": 1.9924812030075187,
"grad_norm": 0.29621434211730957,
"learning_rate": 2.5131578947368423e-05,
"loss": 0.0233,
"step": 1325
},
{
"epoch": 2.0,
"eval_explained_variance": 0.4605059027671814,
"eval_loss": 0.020697500556707382,
"eval_mae": 0.11327706277370453,
"eval_mse": 0.020693965256214142,
"eval_r2": 0.4603225588798523,
"eval_rmse": 0.14385397198622685,
"eval_runtime": 332.1992,
"eval_samples_per_second": 64.043,
"eval_steps_per_second": 1.002,
"step": 1330
},
{
"epoch": 2.030075187969925,
"grad_norm": 0.17958290874958038,
"learning_rate": 2.4661654135338347e-05,
"loss": 0.0235,
"step": 1350
},
{
"epoch": 2.0676691729323307,
"grad_norm": 0.27772992849349976,
"learning_rate": 2.4191729323308275e-05,
"loss": 0.0223,
"step": 1375
},
{
"epoch": 2.1052631578947367,
"grad_norm": 0.1573406457901001,
"learning_rate": 2.37218045112782e-05,
"loss": 0.0214,
"step": 1400
},
{
"epoch": 2.142857142857143,
"grad_norm": 0.704254686832428,
"learning_rate": 2.3251879699248123e-05,
"loss": 0.0224,
"step": 1425
},
{
"epoch": 2.180451127819549,
"grad_norm": 0.3219689726829529,
"learning_rate": 2.2781954887218044e-05,
"loss": 0.0219,
"step": 1450
},
{
"epoch": 2.218045112781955,
"grad_norm": 1.3933064937591553,
"learning_rate": 2.2312030075187972e-05,
"loss": 0.022,
"step": 1475
},
{
"epoch": 2.255639097744361,
"grad_norm": 0.4605341851711273,
"learning_rate": 2.1842105263157896e-05,
"loss": 0.0221,
"step": 1500
},
{
"epoch": 2.293233082706767,
"grad_norm": 0.23276148736476898,
"learning_rate": 2.137218045112782e-05,
"loss": 0.0216,
"step": 1525
},
{
"epoch": 2.3308270676691727,
"grad_norm": 0.602206289768219,
"learning_rate": 2.0902255639097745e-05,
"loss": 0.0211,
"step": 1550
},
{
"epoch": 2.3684210526315788,
"grad_norm": 0.3370234966278076,
"learning_rate": 2.0432330827067672e-05,
"loss": 0.0208,
"step": 1575
},
{
"epoch": 2.406015037593985,
"grad_norm": 0.3906821310520172,
"learning_rate": 1.9962406015037593e-05,
"loss": 0.0202,
"step": 1600
},
{
"epoch": 2.443609022556391,
"grad_norm": 0.26114577054977417,
"learning_rate": 1.949248120300752e-05,
"loss": 0.0221,
"step": 1625
},
{
"epoch": 2.481203007518797,
"grad_norm": 0.215437650680542,
"learning_rate": 1.9022556390977445e-05,
"loss": 0.023,
"step": 1650
},
{
"epoch": 2.518796992481203,
"grad_norm": 0.7732388377189636,
"learning_rate": 1.855263157894737e-05,
"loss": 0.0217,
"step": 1675
},
{
"epoch": 2.556390977443609,
"grad_norm": 0.2342628836631775,
"learning_rate": 1.8082706766917293e-05,
"loss": 0.0219,
"step": 1700
},
{
"epoch": 2.593984962406015,
"grad_norm": 0.49042457342147827,
"learning_rate": 1.761278195488722e-05,
"loss": 0.0213,
"step": 1725
},
{
"epoch": 2.6315789473684212,
"grad_norm": 0.6491354703903198,
"learning_rate": 1.7142857142857142e-05,
"loss": 0.0215,
"step": 1750
},
{
"epoch": 2.6691729323308273,
"grad_norm": 0.4228556454181671,
"learning_rate": 1.667293233082707e-05,
"loss": 0.0211,
"step": 1775
},
{
"epoch": 2.706766917293233,
"grad_norm": 0.25106143951416016,
"learning_rate": 1.6203007518796994e-05,
"loss": 0.0209,
"step": 1800
},
{
"epoch": 2.744360902255639,
"grad_norm": 0.3744986355304718,
"learning_rate": 1.5733082706766918e-05,
"loss": 0.0214,
"step": 1825
},
{
"epoch": 2.781954887218045,
"grad_norm": 1.0658032894134521,
"learning_rate": 1.5263157894736842e-05,
"loss": 0.022,
"step": 1850
},
{
"epoch": 2.819548872180451,
"grad_norm": 0.6378421187400818,
"learning_rate": 1.4793233082706768e-05,
"loss": 0.0212,
"step": 1875
},
{
"epoch": 2.857142857142857,
"grad_norm": 0.972565770149231,
"learning_rate": 1.4323308270676692e-05,
"loss": 0.0219,
"step": 1900
},
{
"epoch": 2.8947368421052633,
"grad_norm": 0.43301036953926086,
"learning_rate": 1.3853383458646618e-05,
"loss": 0.0213,
"step": 1925
},
{
"epoch": 2.932330827067669,
"grad_norm": 0.6031925082206726,
"learning_rate": 1.3383458646616541e-05,
"loss": 0.0217,
"step": 1950
},
{
"epoch": 2.969924812030075,
"grad_norm": 0.20304885506629944,
"learning_rate": 1.2913533834586467e-05,
"loss": 0.0209,
"step": 1975
},
{
"epoch": 3.0,
"eval_explained_variance": 0.4826643466949463,
"eval_loss": 0.01995982974767685,
"eval_mae": 0.11011234670877457,
"eval_mse": 0.019954577088356018,
"eval_r2": 0.4796050786972046,
"eval_rmse": 0.14126067070616655,
"eval_runtime": 332.8065,
"eval_samples_per_second": 63.926,
"eval_steps_per_second": 1.001,
"step": 1995
}
],
"logging_steps": 25,
"max_steps": 2660,
"num_input_tokens_seen": 0,
"num_train_epochs": 4,
"save_steps": 500,
"stateful_callbacks": {
"EarlyStoppingCallback": {
"args": {
"early_stopping_patience": 5,
"early_stopping_threshold": 0.01
},
"attributes": {
"early_stopping_patience_counter": 2
}
},
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 6.709296778916659e+17,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}