yusr9
/

RADAR / last-checkpoint /trainer_state.json
yusr9's picture
Duplicate from MahmoodAnaam/RADAR
d3ad200
Raw
History Blame Contribute Delete
235 kB
{
"best_global_step": 11000,
"best_metric": 0.978,
"best_model_checkpoint": "/content/drive/MyDrive/radar/RADAR/outputs/checkpoints/checkpoint-11000",
"epoch": 9.966777408637874,
"eval_steps": 500,
"global_step": 12000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.008305647840531562,
"grad_norm": 16.83112907409668,
"learning_rate": 4.972375690607735e-07,
"loss": 0.05300590395927429,
"step": 10
},
{
"epoch": 0.016611295681063124,
"grad_norm": 24.510848999023438,
"learning_rate": 1.0497237569060774e-06,
"loss": 0.08961113691329955,
"step": 20
},
{
"epoch": 0.024916943521594685,
"grad_norm": 19.370145797729492,
"learning_rate": 1.6022099447513815e-06,
"loss": 0.06312323808670044,
"step": 30
},
{
"epoch": 0.03322259136212625,
"grad_norm": 16.086774826049805,
"learning_rate": 2.1546961325966854e-06,
"loss": 0.16393563747406006,
"step": 40
},
{
"epoch": 0.04152823920265781,
"grad_norm": 22.249469757080078,
"learning_rate": 2.707182320441989e-06,
"loss": 0.32108685970306394,
"step": 50
},
{
"epoch": 0.04983388704318937,
"grad_norm": 1.438634991645813,
"learning_rate": 3.2596685082872933e-06,
"loss": 0.09590144157409668,
"step": 60
},
{
"epoch": 0.05813953488372093,
"grad_norm": 2.2608940601348877,
"learning_rate": 3.812154696132597e-06,
"loss": 0.0668237864971161,
"step": 70
},
{
"epoch": 0.0664451827242525,
"grad_norm": 9.070174217224121,
"learning_rate": 4.364640883977901e-06,
"loss": 0.03858251571655273,
"step": 80
},
{
"epoch": 0.07475083056478406,
"grad_norm": 33.99458694458008,
"learning_rate": 4.9171270718232054e-06,
"loss": 0.14961743354797363,
"step": 90
},
{
"epoch": 0.08305647840531562,
"grad_norm": 26.488201141357422,
"learning_rate": 5.469613259668509e-06,
"loss": 0.29830121994018555,
"step": 100
},
{
"epoch": 0.09136212624584718,
"grad_norm": 6.779278755187988,
"learning_rate": 6.0220994475138124e-06,
"loss": 0.10294698476791382,
"step": 110
},
{
"epoch": 0.09966777408637874,
"grad_norm": 22.20450782775879,
"learning_rate": 6.574585635359117e-06,
"loss": 0.1000058650970459,
"step": 120
},
{
"epoch": 0.1079734219269103,
"grad_norm": 12.729887008666992,
"learning_rate": 7.12707182320442e-06,
"loss": 0.06427679061889649,
"step": 130
},
{
"epoch": 0.11627906976744186,
"grad_norm": 9.187322616577148,
"learning_rate": 7.679558011049725e-06,
"loss": 0.14388138055801392,
"step": 140
},
{
"epoch": 0.12458471760797342,
"grad_norm": 11.621493339538574,
"learning_rate": 8.232044198895029e-06,
"loss": 0.21061787605285645,
"step": 150
},
{
"epoch": 0.132890365448505,
"grad_norm": 25.235427856445312,
"learning_rate": 8.784530386740332e-06,
"loss": 0.10970768928527833,
"step": 160
},
{
"epoch": 0.14119601328903655,
"grad_norm": 19.270275115966797,
"learning_rate": 9.337016574585636e-06,
"loss": 0.0883016049861908,
"step": 170
},
{
"epoch": 0.14950166112956811,
"grad_norm": 4.681619644165039,
"learning_rate": 9.88950276243094e-06,
"loss": 0.06994540095329285,
"step": 180
},
{
"epoch": 0.15780730897009967,
"grad_norm": 10.913886070251465,
"learning_rate": 1.0441988950276245e-05,
"loss": 0.20744791030883789,
"step": 190
},
{
"epoch": 0.16611295681063123,
"grad_norm": 39.400001525878906,
"learning_rate": 1.0994475138121547e-05,
"loss": 0.2875197410583496,
"step": 200
},
{
"epoch": 0.1744186046511628,
"grad_norm": 10.333728790283203,
"learning_rate": 1.1546961325966852e-05,
"loss": 0.13045910596847535,
"step": 210
},
{
"epoch": 0.18272425249169436,
"grad_norm": 5.740739345550537,
"learning_rate": 1.2099447513812154e-05,
"loss": 0.08207477331161499,
"step": 220
},
{
"epoch": 0.19102990033222592,
"grad_norm": 9.670475006103516,
"learning_rate": 1.265193370165746e-05,
"loss": 0.09993141889572144,
"step": 230
},
{
"epoch": 0.19933554817275748,
"grad_norm": 9.050890922546387,
"learning_rate": 1.3204419889502765e-05,
"loss": 0.17004809379577637,
"step": 240
},
{
"epoch": 0.20764119601328904,
"grad_norm": 42.47259521484375,
"learning_rate": 1.3756906077348067e-05,
"loss": 0.28405323028564455,
"step": 250
},
{
"epoch": 0.2159468438538206,
"grad_norm": 10.077814102172852,
"learning_rate": 1.4309392265193372e-05,
"loss": 0.06517388224601746,
"step": 260
},
{
"epoch": 0.22425249169435216,
"grad_norm": 47.007381439208984,
"learning_rate": 1.4861878453038676e-05,
"loss": 0.08509432077407837,
"step": 270
},
{
"epoch": 0.23255813953488372,
"grad_norm": 20.001007080078125,
"learning_rate": 1.541436464088398e-05,
"loss": 0.10799007415771485,
"step": 280
},
{
"epoch": 0.24086378737541528,
"grad_norm": 34.18750762939453,
"learning_rate": 1.5966850828729285e-05,
"loss": 0.18408216238021852,
"step": 290
},
{
"epoch": 0.24916943521594684,
"grad_norm": 12.465804100036621,
"learning_rate": 1.6519337016574586e-05,
"loss": 0.25962674617767334,
"step": 300
},
{
"epoch": 0.2574750830564784,
"grad_norm": 1.5916982889175415,
"learning_rate": 1.707182320441989e-05,
"loss": 0.07179607152938842,
"step": 310
},
{
"epoch": 0.26578073089701,
"grad_norm": 2.1843109130859375,
"learning_rate": 1.7624309392265194e-05,
"loss": 0.06425885558128357,
"step": 320
},
{
"epoch": 0.27408637873754155,
"grad_norm": 2.3891329765319824,
"learning_rate": 1.81767955801105e-05,
"loss": 0.08510435819625854,
"step": 330
},
{
"epoch": 0.2823920265780731,
"grad_norm": 8.76103401184082,
"learning_rate": 1.8729281767955803e-05,
"loss": 0.21063413619995117,
"step": 340
},
{
"epoch": 0.29069767441860467,
"grad_norm": 26.145353317260742,
"learning_rate": 1.9281767955801107e-05,
"loss": 0.32075982093811034,
"step": 350
},
{
"epoch": 0.29900332225913623,
"grad_norm": 20.650766372680664,
"learning_rate": 1.9834254143646408e-05,
"loss": 0.10961083173751832,
"step": 360
},
{
"epoch": 0.3073089700996678,
"grad_norm": 4.302230358123779,
"learning_rate": 1.9999982269182506e-05,
"loss": 0.07394000887870789,
"step": 370
},
{
"epoch": 0.31561461794019935,
"grad_norm": 1.942402958869934,
"learning_rate": 1.999989542451349e-05,
"loss": 0.06208068132400513,
"step": 380
},
{
"epoch": 0.3239202657807309,
"grad_norm": 5.765994071960449,
"learning_rate": 1.99997362099399e-05,
"loss": 0.12262409925460815,
"step": 390
},
{
"epoch": 0.33222591362126247,
"grad_norm": 34.15365219116211,
"learning_rate": 1.9999504626613983e-05,
"loss": 0.32247633934020997,
"step": 400
},
{
"epoch": 0.34053156146179403,
"grad_norm": 17.080778121948242,
"learning_rate": 1.9999200676211728e-05,
"loss": 0.08944156765937805,
"step": 410
},
{
"epoch": 0.3488372093023256,
"grad_norm": 2.397739887237549,
"learning_rate": 1.999882436093284e-05,
"loss": 0.06061902642250061,
"step": 420
},
{
"epoch": 0.35714285714285715,
"grad_norm": 0.11066510528326035,
"learning_rate": 1.9998375683500734e-05,
"loss": 0.021029716730117796,
"step": 430
},
{
"epoch": 0.3654485049833887,
"grad_norm": 12.773712158203125,
"learning_rate": 1.9997854647162524e-05,
"loss": 0.223671817779541,
"step": 440
},
{
"epoch": 0.37375415282392027,
"grad_norm": 21.977563858032227,
"learning_rate": 1.9997261255688983e-05,
"loss": 0.24787373542785646,
"step": 450
},
{
"epoch": 0.38205980066445183,
"grad_norm": 1.5935643911361694,
"learning_rate": 1.9996595513374527e-05,
"loss": 0.08940771222114563,
"step": 460
},
{
"epoch": 0.3903654485049834,
"grad_norm": 5.3650383949279785,
"learning_rate": 1.999585742503717e-05,
"loss": 0.07068525552749634,
"step": 470
},
{
"epoch": 0.39867109634551495,
"grad_norm": 1.2879642248153687,
"learning_rate": 1.999504699601851e-05,
"loss": 0.056727123260498044,
"step": 480
},
{
"epoch": 0.4069767441860465,
"grad_norm": 8.278883934020996,
"learning_rate": 1.9994164232183678e-05,
"loss": 0.22892401218414307,
"step": 490
},
{
"epoch": 0.4152823920265781,
"grad_norm": 20.857627868652344,
"learning_rate": 1.999320913992129e-05,
"loss": 0.2103438377380371,
"step": 500
},
{
"epoch": 0.4152823920265781,
"eval_brier": 0.928,
"eval_c@1": 0.904,
"eval_f05u": 0.952,
"eval_f1": 0.902,
"eval_loss": 0.1814199686050415,
"eval_mean": 0.934,
"eval_roc-auc": 0.982,
"eval_runtime": 72.3861,
"eval_samples_per_second": 254.51,
"eval_steps_per_second": 3.979,
"step": 500
},
{
"epoch": 0.42358803986710963,
"grad_norm": 4.860109329223633,
"learning_rate": 1.9992181726143415e-05,
"loss": 0.051990830898284913,
"step": 510
},
{
"epoch": 0.4318936877076412,
"grad_norm": 2.0726921558380127,
"learning_rate": 1.9991081998285514e-05,
"loss": 0.06843258142471313,
"step": 520
},
{
"epoch": 0.44019933554817275,
"grad_norm": 9.351646423339844,
"learning_rate": 1.9989909964306388e-05,
"loss": 0.10711724758148193,
"step": 530
},
{
"epoch": 0.4485049833887043,
"grad_norm": 8.241874694824219,
"learning_rate": 1.998866563268812e-05,
"loss": 0.16665782928466796,
"step": 540
},
{
"epoch": 0.4568106312292359,
"grad_norm": 15.618956565856934,
"learning_rate": 1.9987349012436038e-05,
"loss": 0.15495831966400148,
"step": 550
},
{
"epoch": 0.46511627906976744,
"grad_norm": 6.535603046417236,
"learning_rate": 1.998596011307859e-05,
"loss": 0.04994972050189972,
"step": 560
},
{
"epoch": 0.473421926910299,
"grad_norm": 0.7167952656745911,
"learning_rate": 1.998449894466734e-05,
"loss": 0.06999388933181763,
"step": 570
},
{
"epoch": 0.48172757475083056,
"grad_norm": 3.554610252380371,
"learning_rate": 1.9982965517776867e-05,
"loss": 0.032020002603530884,
"step": 580
},
{
"epoch": 0.4900332225913621,
"grad_norm": 7.1078033447265625,
"learning_rate": 1.998135984350467e-05,
"loss": 0.1670602560043335,
"step": 590
},
{
"epoch": 0.4983388704318937,
"grad_norm": 13.27259635925293,
"learning_rate": 1.9979681933471127e-05,
"loss": 0.2317356586456299,
"step": 600
},
{
"epoch": 0.5066445182724253,
"grad_norm": 1.831939458847046,
"learning_rate": 1.997793179981939e-05,
"loss": 0.08199180364608764,
"step": 610
},
{
"epoch": 0.5149501661129569,
"grad_norm": 6.776217460632324,
"learning_rate": 1.9976109455215284e-05,
"loss": 0.03575927019119263,
"step": 620
},
{
"epoch": 0.5232558139534884,
"grad_norm": 15.310006141662598,
"learning_rate": 1.9974214912847235e-05,
"loss": 0.1283599853515625,
"step": 630
},
{
"epoch": 0.53156146179402,
"grad_norm": 7.563514709472656,
"learning_rate": 1.9972248186426175e-05,
"loss": 0.12110563516616821,
"step": 640
},
{
"epoch": 0.5398671096345515,
"grad_norm": 6.085082530975342,
"learning_rate": 1.9970209290185436e-05,
"loss": 0.13210896253585816,
"step": 650
},
{
"epoch": 0.5481727574750831,
"grad_norm": 2.3826427459716797,
"learning_rate": 1.996809823888064e-05,
"loss": 0.06088674068450928,
"step": 660
},
{
"epoch": 0.5564784053156147,
"grad_norm": 0.09302367269992828,
"learning_rate": 1.996591504778961e-05,
"loss": 0.020438264310359954,
"step": 670
},
{
"epoch": 0.5647840531561462,
"grad_norm": 11.22698974609375,
"learning_rate": 1.9963659732712242e-05,
"loss": 0.06598439812660217,
"step": 680
},
{
"epoch": 0.5730897009966778,
"grad_norm": 1.0729060173034668,
"learning_rate": 1.9961332309970398e-05,
"loss": 0.13380913734436034,
"step": 690
},
{
"epoch": 0.5813953488372093,
"grad_norm": 12.22038459777832,
"learning_rate": 1.9958932796407794e-05,
"loss": 0.16085643768310548,
"step": 700
},
{
"epoch": 0.5897009966777409,
"grad_norm": 7.082716941833496,
"learning_rate": 1.9956461209389863e-05,
"loss": 0.06574858427047729,
"step": 710
},
{
"epoch": 0.5980066445182725,
"grad_norm": 1.9633880853652954,
"learning_rate": 1.9953917566803652e-05,
"loss": 0.06811402440071106,
"step": 720
},
{
"epoch": 0.606312292358804,
"grad_norm": 0.07808863371610641,
"learning_rate": 1.9951301887057662e-05,
"loss": 0.045835795998573306,
"step": 730
},
{
"epoch": 0.6146179401993356,
"grad_norm": 4.953616142272949,
"learning_rate": 1.994861418908175e-05,
"loss": 0.11979286670684815,
"step": 740
},
{
"epoch": 0.6229235880398671,
"grad_norm": 3.3133060932159424,
"learning_rate": 1.9945854492326956e-05,
"loss": 0.0893496334552765,
"step": 750
},
{
"epoch": 0.6312292358803987,
"grad_norm": 2.264364242553711,
"learning_rate": 1.994302281676539e-05,
"loss": 0.04600924849510193,
"step": 760
},
{
"epoch": 0.6395348837209303,
"grad_norm": 0.8064838647842407,
"learning_rate": 1.9940119182890078e-05,
"loss": 0.05525304079055786,
"step": 770
},
{
"epoch": 0.6478405315614618,
"grad_norm": 2.455915689468384,
"learning_rate": 1.99371436117148e-05,
"loss": 0.036086255311965944,
"step": 780
},
{
"epoch": 0.6561461794019934,
"grad_norm": 2.6588404178619385,
"learning_rate": 1.9934096124773976e-05,
"loss": 0.20075614452362062,
"step": 790
},
{
"epoch": 0.6644518272425249,
"grad_norm": 13.506540298461914,
"learning_rate": 1.9930976744122463e-05,
"loss": 0.1896299362182617,
"step": 800
},
{
"epoch": 0.6727574750830565,
"grad_norm": 1.5946769714355469,
"learning_rate": 1.992778549233543e-05,
"loss": 0.07047351002693177,
"step": 810
},
{
"epoch": 0.6810631229235881,
"grad_norm": 1.7787715196609497,
"learning_rate": 1.992452239250817e-05,
"loss": 0.07980353236198426,
"step": 820
},
{
"epoch": 0.6893687707641196,
"grad_norm": 0.3226272761821747,
"learning_rate": 1.9921187468255954e-05,
"loss": 0.049928227066993715,
"step": 830
},
{
"epoch": 0.6976744186046512,
"grad_norm": 5.392436981201172,
"learning_rate": 1.991778074371385e-05,
"loss": 0.12367438077926636,
"step": 840
},
{
"epoch": 0.7059800664451827,
"grad_norm": 7.643642425537109,
"learning_rate": 1.991430224353655e-05,
"loss": 0.13326771259307862,
"step": 850
},
{
"epoch": 0.7142857142857143,
"grad_norm": 2.352473735809326,
"learning_rate": 1.9910751992898192e-05,
"loss": 0.028580480813980104,
"step": 860
},
{
"epoch": 0.7225913621262459,
"grad_norm": 2.6504147052764893,
"learning_rate": 1.9907130017492176e-05,
"loss": 0.02747690975666046,
"step": 870
},
{
"epoch": 0.7308970099667774,
"grad_norm": 9.994357109069824,
"learning_rate": 1.990343634353097e-05,
"loss": 0.08383561372756958,
"step": 880
},
{
"epoch": 0.739202657807309,
"grad_norm": 8.853647232055664,
"learning_rate": 1.9899670997745943e-05,
"loss": 0.08635759353637695,
"step": 890
},
{
"epoch": 0.7475083056478405,
"grad_norm": 20.926956176757812,
"learning_rate": 1.9895834007387144e-05,
"loss": 0.17967108488082886,
"step": 900
},
{
"epoch": 0.7558139534883721,
"grad_norm": 1.024949312210083,
"learning_rate": 1.9891925400223132e-05,
"loss": 0.037137174606323244,
"step": 910
},
{
"epoch": 0.7641196013289037,
"grad_norm": 4.252371788024902,
"learning_rate": 1.988794520454076e-05,
"loss": 0.054977518320083615,
"step": 920
},
{
"epoch": 0.7724252491694352,
"grad_norm": 0.2748623192310333,
"learning_rate": 1.988389344914496e-05,
"loss": 0.03901148438453674,
"step": 930
},
{
"epoch": 0.7807308970099668,
"grad_norm": 1.716922402381897,
"learning_rate": 1.9879770163358554e-05,
"loss": 0.16883451938629152,
"step": 940
},
{
"epoch": 0.7890365448504983,
"grad_norm": 13.949922561645508,
"learning_rate": 1.9875575377022037e-05,
"loss": 0.13278028964996338,
"step": 950
},
{
"epoch": 0.7973421926910299,
"grad_norm": 0.58612060546875,
"learning_rate": 1.9871309120493352e-05,
"loss": 0.05191807150840759,
"step": 960
},
{
"epoch": 0.8056478405315615,
"grad_norm": 0.22168058156967163,
"learning_rate": 1.9866971424647685e-05,
"loss": 0.043825897574424746,
"step": 970
},
{
"epoch": 0.813953488372093,
"grad_norm": 0.05641980841755867,
"learning_rate": 1.9862562320877227e-05,
"loss": 0.03552835881710052,
"step": 980
},
{
"epoch": 0.8222591362126246,
"grad_norm": 14.803342819213867,
"learning_rate": 1.9858081841090944e-05,
"loss": 0.11208894252777099,
"step": 990
},
{
"epoch": 0.8305647840531561,
"grad_norm": 12.898427963256836,
"learning_rate": 1.9853530017714374e-05,
"loss": 0.21021549701690673,
"step": 1000
},
{
"epoch": 0.8305647840531561,
"eval_brier": 0.93,
"eval_c@1": 0.909,
"eval_f05u": 0.945,
"eval_f1": 0.91,
"eval_loss": 0.18657569587230682,
"eval_mean": 0.934,
"eval_roc-auc": 0.977,
"eval_runtime": 69.0871,
"eval_samples_per_second": 266.663,
"eval_steps_per_second": 4.169,
"step": 1000
},
{
"epoch": 0.8388704318936877,
"grad_norm": 10.441580772399902,
"learning_rate": 1.9848906883689364e-05,
"loss": 0.09427762031555176,
"step": 1010
},
{
"epoch": 0.8471760797342193,
"grad_norm": 10.603070259094238,
"learning_rate": 1.9844212472473834e-05,
"loss": 0.04516265392303467,
"step": 1020
},
{
"epoch": 0.8554817275747508,
"grad_norm": 1.0459258556365967,
"learning_rate": 1.983944681804156e-05,
"loss": 0.0815069019794464,
"step": 1030
},
{
"epoch": 0.8637873754152824,
"grad_norm": 2.7023723125457764,
"learning_rate": 1.9834609954881884e-05,
"loss": 0.10050047636032104,
"step": 1040
},
{
"epoch": 0.872093023255814,
"grad_norm": 3.6623501777648926,
"learning_rate": 1.9829701917999513e-05,
"loss": 0.13907005786895751,
"step": 1050
},
{
"epoch": 0.8803986710963455,
"grad_norm": 7.458986759185791,
"learning_rate": 1.982472274291423e-05,
"loss": 0.08346118927001953,
"step": 1060
},
{
"epoch": 0.8887043189368771,
"grad_norm": 0.37509089708328247,
"learning_rate": 1.981967246566066e-05,
"loss": 0.01990487277507782,
"step": 1070
},
{
"epoch": 0.8970099667774086,
"grad_norm": 3.439479351043701,
"learning_rate": 1.9814551122787994e-05,
"loss": 0.039832589030265805,
"step": 1080
},
{
"epoch": 0.9053156146179402,
"grad_norm": 4.903800964355469,
"learning_rate": 1.9809358751359724e-05,
"loss": 0.15382176637649536,
"step": 1090
},
{
"epoch": 0.9136212624584718,
"grad_norm": 6.491225719451904,
"learning_rate": 1.9804095388953384e-05,
"loss": 0.09639735817909241,
"step": 1100
},
{
"epoch": 0.9219269102990033,
"grad_norm": 0.31971296668052673,
"learning_rate": 1.9798761073660287e-05,
"loss": 0.029558992385864256,
"step": 1110
},
{
"epoch": 0.9302325581395349,
"grad_norm": 1.9831352233886719,
"learning_rate": 1.979335584408521e-05,
"loss": 0.05827125310897827,
"step": 1120
},
{
"epoch": 0.9385382059800664,
"grad_norm": 0.9004706740379333,
"learning_rate": 1.978787973934617e-05,
"loss": 0.03546359837055206,
"step": 1130
},
{
"epoch": 0.946843853820598,
"grad_norm": 5.776724815368652,
"learning_rate": 1.9782332799074094e-05,
"loss": 0.10988500118255615,
"step": 1140
},
{
"epoch": 0.9551495016611296,
"grad_norm": 8.462262153625488,
"learning_rate": 1.9776715063412553e-05,
"loss": 0.18307191133499146,
"step": 1150
},
{
"epoch": 0.9634551495016611,
"grad_norm": 0.4519129991531372,
"learning_rate": 1.9771026573017472e-05,
"loss": 0.07031915783882141,
"step": 1160
},
{
"epoch": 0.9717607973421927,
"grad_norm": 2.075129508972168,
"learning_rate": 1.976526736905684e-05,
"loss": 0.018192824721336365,
"step": 1170
},
{
"epoch": 0.9800664451827242,
"grad_norm": 3.2885875701904297,
"learning_rate": 1.9759437493210386e-05,
"loss": 0.059246021509170535,
"step": 1180
},
{
"epoch": 0.9883720930232558,
"grad_norm": 16.03324317932129,
"learning_rate": 1.975353698766932e-05,
"loss": 0.07062767148017883,
"step": 1190
},
{
"epoch": 0.9966777408637874,
"grad_norm": 5.529783248901367,
"learning_rate": 1.9747565895135973e-05,
"loss": 0.15401830673217773,
"step": 1200
},
{
"epoch": 1.004983388704319,
"grad_norm": 0.18491490185260773,
"learning_rate": 1.9741524258823554e-05,
"loss": 0.07995923161506653,
"step": 1210
},
{
"epoch": 1.0132890365448506,
"grad_norm": 2.728106737136841,
"learning_rate": 1.9735412122455777e-05,
"loss": 0.008018089830875397,
"step": 1220
},
{
"epoch": 1.021594684385382,
"grad_norm": 1.0241869688034058,
"learning_rate": 1.9729229530266576e-05,
"loss": 0.030219340324401857,
"step": 1230
},
{
"epoch": 1.0299003322259137,
"grad_norm": 7.063983917236328,
"learning_rate": 1.972297652699978e-05,
"loss": 0.07526764869689942,
"step": 1240
},
{
"epoch": 1.0382059800664452,
"grad_norm": 2.7647554874420166,
"learning_rate": 1.971665315790879e-05,
"loss": 0.08511782884597778,
"step": 1250
},
{
"epoch": 1.0465116279069768,
"grad_norm": 0.05360090732574463,
"learning_rate": 1.9710259468756247e-05,
"loss": 0.04494154155254364,
"step": 1260
},
{
"epoch": 1.0548172757475083,
"grad_norm": 0.6909206509590149,
"learning_rate": 1.97037955058137e-05,
"loss": 0.0197439044713974,
"step": 1270
},
{
"epoch": 1.06312292358804,
"grad_norm": 1.1125959157943726,
"learning_rate": 1.969726131586127e-05,
"loss": 0.03601456880569458,
"step": 1280
},
{
"epoch": 1.0714285714285714,
"grad_norm": 8.950166702270508,
"learning_rate": 1.9690656946187333e-05,
"loss": 0.06061711311340332,
"step": 1290
},
{
"epoch": 1.079734219269103,
"grad_norm": 10.442480087280273,
"learning_rate": 1.9683982444588143e-05,
"loss": 0.0630322515964508,
"step": 1300
},
{
"epoch": 1.0880398671096345,
"grad_norm": 0.17640690505504608,
"learning_rate": 1.9677237859367512e-05,
"loss": 0.08533826470375061,
"step": 1310
},
{
"epoch": 1.0963455149501662,
"grad_norm": 10.902055740356445,
"learning_rate": 1.967042323933645e-05,
"loss": 0.041460072994232176,
"step": 1320
},
{
"epoch": 1.1046511627906976,
"grad_norm": 0.6280453205108643,
"learning_rate": 1.9663538633812806e-05,
"loss": 0.026225927472114562,
"step": 1330
},
{
"epoch": 1.1129568106312293,
"grad_norm": 7.400638103485107,
"learning_rate": 1.9656584092620936e-05,
"loss": 0.10436614751815795,
"step": 1340
},
{
"epoch": 1.1212624584717608,
"grad_norm": 9.567360877990723,
"learning_rate": 1.9649559666091304e-05,
"loss": 0.09774322509765625,
"step": 1350
},
{
"epoch": 1.1295681063122924,
"grad_norm": 11.050363540649414,
"learning_rate": 1.964246540506016e-05,
"loss": 0.09643864035606384,
"step": 1360
},
{
"epoch": 1.1378737541528239,
"grad_norm": 1.9213371276855469,
"learning_rate": 1.9635301360869132e-05,
"loss": 0.04072706401348114,
"step": 1370
},
{
"epoch": 1.1461794019933556,
"grad_norm": 0.3345423936843872,
"learning_rate": 1.9628067585364882e-05,
"loss": 0.02469405233860016,
"step": 1380
},
{
"epoch": 1.154485049833887,
"grad_norm": 16.516342163085938,
"learning_rate": 1.962076413089873e-05,
"loss": 0.1438913583755493,
"step": 1390
},
{
"epoch": 1.1627906976744187,
"grad_norm": 1.104395866394043,
"learning_rate": 1.9613391050326245e-05,
"loss": 0.11620582342147827,
"step": 1400
},
{
"epoch": 1.1710963455149501,
"grad_norm": 9.292811393737793,
"learning_rate": 1.9605948397006913e-05,
"loss": 0.1055580735206604,
"step": 1410
},
{
"epoch": 1.1794019933554818,
"grad_norm": 9.632716178894043,
"learning_rate": 1.9598436224803703e-05,
"loss": 0.04468507170677185,
"step": 1420
},
{
"epoch": 1.1877076411960132,
"grad_norm": 1.043423056602478,
"learning_rate": 1.9590854588082697e-05,
"loss": 0.0349301666021347,
"step": 1430
},
{
"epoch": 1.196013289036545,
"grad_norm": 2.9193689823150635,
"learning_rate": 1.958320354171271e-05,
"loss": 0.06454914212226867,
"step": 1440
},
{
"epoch": 1.2043189368770764,
"grad_norm": 0.6662460565567017,
"learning_rate": 1.9575483141064863e-05,
"loss": 0.04076001644134521,
"step": 1450
},
{
"epoch": 1.212624584717608,
"grad_norm": 0.23810455203056335,
"learning_rate": 1.9567693442012214e-05,
"loss": 0.056306701898574826,
"step": 1460
},
{
"epoch": 1.2209302325581395,
"grad_norm": 4.045562744140625,
"learning_rate": 1.9559834500929324e-05,
"loss": 0.036572623252868655,
"step": 1470
},
{
"epoch": 1.2292358803986712,
"grad_norm": 5.869875431060791,
"learning_rate": 1.955190637469187e-05,
"loss": 0.039019054174423216,
"step": 1480
},
{
"epoch": 1.2375415282392026,
"grad_norm": 5.444359302520752,
"learning_rate": 1.9543909120676224e-05,
"loss": 0.027615264058113098,
"step": 1490
},
{
"epoch": 1.2458471760797343,
"grad_norm": 4.1580705642700195,
"learning_rate": 1.953584279675905e-05,
"loss": 0.08209576010704041,
"step": 1500
},
{
"epoch": 1.2458471760797343,
"eval_brier": 0.94,
"eval_c@1": 0.921,
"eval_f05u": 0.905,
"eval_f1": 0.93,
"eval_loss": 0.2049339860677719,
"eval_mean": 0.937,
"eval_roc-auc": 0.989,
"eval_runtime": 69.2458,
"eval_samples_per_second": 266.052,
"eval_steps_per_second": 4.159,
"step": 1500
},
{
"epoch": 1.2541528239202657,
"grad_norm": 5.411282539367676,
"learning_rate": 1.952770746131686e-05,
"loss": 0.09549391269683838,
"step": 1510
},
{
"epoch": 1.2624584717607974,
"grad_norm": 2.6220927238464355,
"learning_rate": 1.951950317322562e-05,
"loss": 0.05993903279304504,
"step": 1520
},
{
"epoch": 1.2707641196013288,
"grad_norm": 17.44796371459961,
"learning_rate": 1.9511229991860304e-05,
"loss": 0.02594410181045532,
"step": 1530
},
{
"epoch": 1.2790697674418605,
"grad_norm": 3.954331159591675,
"learning_rate": 1.950288797709447e-05,
"loss": 0.09088499546051025,
"step": 1540
},
{
"epoch": 1.287375415282392,
"grad_norm": 8.130070686340332,
"learning_rate": 1.949447718929983e-05,
"loss": 0.12227284908294678,
"step": 1550
},
{
"epoch": 1.2956810631229236,
"grad_norm": 0.1791795790195465,
"learning_rate": 1.948599768934582e-05,
"loss": 0.08279802203178406,
"step": 1560
},
{
"epoch": 1.303986710963455,
"grad_norm": 0.27110907435417175,
"learning_rate": 1.9477449538599132e-05,
"loss": 0.04039774239063263,
"step": 1570
},
{
"epoch": 1.3122923588039868,
"grad_norm": 0.05300529673695564,
"learning_rate": 1.9468832798923305e-05,
"loss": 0.00991017147898674,
"step": 1580
},
{
"epoch": 1.3205980066445182,
"grad_norm": 3.4683456420898438,
"learning_rate": 1.946014753267825e-05,
"loss": 0.05042205452919006,
"step": 1590
},
{
"epoch": 1.3289036544850499,
"grad_norm": 1.2810217142105103,
"learning_rate": 1.9451393802719815e-05,
"loss": 0.08756383657455444,
"step": 1600
},
{
"epoch": 1.3372093023255813,
"grad_norm": 0.41384729743003845,
"learning_rate": 1.944257167239932e-05,
"loss": 0.06056967377662659,
"step": 1610
},
{
"epoch": 1.345514950166113,
"grad_norm": 6.993170738220215,
"learning_rate": 1.943368120556311e-05,
"loss": 0.033351412415504454,
"step": 1620
},
{
"epoch": 1.3538205980066444,
"grad_norm": 0.06484860181808472,
"learning_rate": 1.942472246655208e-05,
"loss": 0.03970024585723877,
"step": 1630
},
{
"epoch": 1.3621262458471761,
"grad_norm": 0.5462992191314697,
"learning_rate": 1.9415695520201216e-05,
"loss": 0.0880971074104309,
"step": 1640
},
{
"epoch": 1.3704318936877076,
"grad_norm": 4.575297832489014,
"learning_rate": 1.9406600431839123e-05,
"loss": 0.10081875324249268,
"step": 1650
},
{
"epoch": 1.3787375415282392,
"grad_norm": 1.5679502487182617,
"learning_rate": 1.9397437267287565e-05,
"loss": 0.09293895959854126,
"step": 1660
},
{
"epoch": 1.3870431893687707,
"grad_norm": 0.10219092667102814,
"learning_rate": 1.9388206092860962e-05,
"loss": 0.0185856893658638,
"step": 1670
},
{
"epoch": 1.3953488372093024,
"grad_norm": 21.064393997192383,
"learning_rate": 1.9378906975365935e-05,
"loss": 0.050033068656921385,
"step": 1680
},
{
"epoch": 1.4036544850498338,
"grad_norm": 0.2731074392795563,
"learning_rate": 1.936953998210082e-05,
"loss": 0.033242452144622806,
"step": 1690
},
{
"epoch": 1.4119601328903655,
"grad_norm": 8.639477729797363,
"learning_rate": 1.9360105180855153e-05,
"loss": 0.0725834846496582,
"step": 1700
},
{
"epoch": 1.420265780730897,
"grad_norm": 0.026840293779969215,
"learning_rate": 1.9350602639909227e-05,
"loss": 0.03226782977581024,
"step": 1710
},
{
"epoch": 1.4285714285714286,
"grad_norm": 16.719680786132812,
"learning_rate": 1.9341032428033552e-05,
"loss": 0.05763497352600098,
"step": 1720
},
{
"epoch": 1.43687707641196,
"grad_norm": 0.12961284816265106,
"learning_rate": 1.933139461448839e-05,
"loss": 0.03973042368888855,
"step": 1730
},
{
"epoch": 1.4451827242524917,
"grad_norm": 2.3053228855133057,
"learning_rate": 1.9321689269023232e-05,
"loss": 0.027554985880851746,
"step": 1740
},
{
"epoch": 1.4534883720930232,
"grad_norm": 6.428056716918945,
"learning_rate": 1.9311916461876303e-05,
"loss": 0.09585081934928893,
"step": 1750
},
{
"epoch": 1.4617940199335548,
"grad_norm": 1.7588001489639282,
"learning_rate": 1.930207626377406e-05,
"loss": 0.08310455083847046,
"step": 1760
},
{
"epoch": 1.4700996677740865,
"grad_norm": 0.06497257202863693,
"learning_rate": 1.929216874593066e-05,
"loss": 0.0100330650806427,
"step": 1770
},
{
"epoch": 1.478405315614618,
"grad_norm": 0.06480560451745987,
"learning_rate": 1.9282193980047475e-05,
"loss": 0.02021673172712326,
"step": 1780
},
{
"epoch": 1.4867109634551494,
"grad_norm": 7.9334187507629395,
"learning_rate": 1.9272152038312537e-05,
"loss": 0.05561460852622986,
"step": 1790
},
{
"epoch": 1.495016611295681,
"grad_norm": 1.0112978219985962,
"learning_rate": 1.9262042993400042e-05,
"loss": 0.08957358002662659,
"step": 1800
},
{
"epoch": 1.5033222591362128,
"grad_norm": 0.8256621956825256,
"learning_rate": 1.9251866918469824e-05,
"loss": 0.08316249847412109,
"step": 1810
},
{
"epoch": 1.5116279069767442,
"grad_norm": 2.5284905433654785,
"learning_rate": 1.9241623887166803e-05,
"loss": 0.07930291891098022,
"step": 1820
},
{
"epoch": 1.5199335548172757,
"grad_norm": 4.260557651519775,
"learning_rate": 1.923131397362048e-05,
"loss": 0.02844361066818237,
"step": 1830
},
{
"epoch": 1.5282392026578073,
"grad_norm": 6.121240139007568,
"learning_rate": 1.922093725244438e-05,
"loss": 0.030609488487243652,
"step": 1840
},
{
"epoch": 1.536544850498339,
"grad_norm": 1.5962550640106201,
"learning_rate": 1.9210493798735514e-05,
"loss": 0.04409843683242798,
"step": 1850
},
{
"epoch": 1.5448504983388704,
"grad_norm": 2.6924915313720703,
"learning_rate": 1.9199983688073855e-05,
"loss": 0.11589599847793579,
"step": 1860
},
{
"epoch": 1.553156146179402,
"grad_norm": 1.626353144645691,
"learning_rate": 1.9189406996521763e-05,
"loss": 0.034252825379371646,
"step": 1870
},
{
"epoch": 1.5614617940199336,
"grad_norm": 0.059950847178697586,
"learning_rate": 1.9178763800623462e-05,
"loss": 0.021375831961631776,
"step": 1880
},
{
"epoch": 1.5697674418604652,
"grad_norm": 10.785876274108887,
"learning_rate": 1.9168054177404455e-05,
"loss": 0.08135627508163452,
"step": 1890
},
{
"epoch": 1.5780730897009967,
"grad_norm": 4.941102504730225,
"learning_rate": 1.9157278204371002e-05,
"loss": 0.08478119969367981,
"step": 1900
},
{
"epoch": 1.5863787375415281,
"grad_norm": 0.6561166048049927,
"learning_rate": 1.914643595950953e-05,
"loss": 0.07132599353790284,
"step": 1910
},
{
"epoch": 1.5946843853820598,
"grad_norm": 0.13666100800037384,
"learning_rate": 1.913552752128608e-05,
"loss": 0.04639454483985901,
"step": 1920
},
{
"epoch": 1.6029900332225915,
"grad_norm": 1.1386218070983887,
"learning_rate": 1.912455296864574e-05,
"loss": 0.02302076816558838,
"step": 1930
},
{
"epoch": 1.611295681063123,
"grad_norm": 3.7991814613342285,
"learning_rate": 1.9113512381012074e-05,
"loss": 0.07212566137313843,
"step": 1940
},
{
"epoch": 1.6196013289036544,
"grad_norm": 1.7103837728500366,
"learning_rate": 1.9102405838286552e-05,
"loss": 0.06245579719543457,
"step": 1950
},
{
"epoch": 1.627906976744186,
"grad_norm": 1.7986949682235718,
"learning_rate": 1.909123342084795e-05,
"loss": 0.06121249198913574,
"step": 1960
},
{
"epoch": 1.6362126245847177,
"grad_norm": 10.097555160522461,
"learning_rate": 1.9079995209551797e-05,
"loss": 0.06147882342338562,
"step": 1970
},
{
"epoch": 1.6445182724252492,
"grad_norm": 2.3163371086120605,
"learning_rate": 1.9068691285729764e-05,
"loss": 0.0443866103887558,
"step": 1980
},
{
"epoch": 1.6528239202657806,
"grad_norm": 7.275723934173584,
"learning_rate": 1.9057321731189105e-05,
"loss": 0.058395135402679446,
"step": 1990
},
{
"epoch": 1.6611295681063123,
"grad_norm": 0.7260637283325195,
"learning_rate": 1.9045886628212033e-05,
"loss": 0.04758671224117279,
"step": 2000
},
{
"epoch": 1.6611295681063123,
"eval_brier": 0.969,
"eval_c@1": 0.961,
"eval_f05u": 0.979,
"eval_f1": 0.962,
"eval_loss": 0.086772121489048,
"eval_mean": 0.973,
"eval_roc-auc": 0.993,
"eval_runtime": 69.3117,
"eval_samples_per_second": 265.799,
"eval_steps_per_second": 4.155,
"step": 2000
},
{
"epoch": 1.669435215946844,
"grad_norm": 1.015745997428894,
"learning_rate": 1.9034386059555144e-05,
"loss": 0.05637862086296082,
"step": 2010
},
{
"epoch": 1.6777408637873754,
"grad_norm": 0.1752169281244278,
"learning_rate": 1.902282010844881e-05,
"loss": 0.01659790724515915,
"step": 2020
},
{
"epoch": 1.6860465116279069,
"grad_norm": 6.977785587310791,
"learning_rate": 1.90111888585966e-05,
"loss": 0.04397172033786774,
"step": 2030
},
{
"epoch": 1.6943521594684385,
"grad_norm": 6.9239678382873535,
"learning_rate": 1.899949239417462e-05,
"loss": 0.035265779495239256,
"step": 2040
},
{
"epoch": 1.7026578073089702,
"grad_norm": 10.678945541381836,
"learning_rate": 1.8987730799830966e-05,
"loss": 0.04530561566352844,
"step": 2050
},
{
"epoch": 1.7109634551495017,
"grad_norm": 11.756209373474121,
"learning_rate": 1.897590416068507e-05,
"loss": 0.03778558969497681,
"step": 2060
},
{
"epoch": 1.719269102990033,
"grad_norm": 0.08435416966676712,
"learning_rate": 1.8964012562327108e-05,
"loss": 0.03003433644771576,
"step": 2070
},
{
"epoch": 1.7275747508305648,
"grad_norm": 0.061264973133802414,
"learning_rate": 1.895205609081736e-05,
"loss": 0.013656507432460784,
"step": 2080
},
{
"epoch": 1.7358803986710964,
"grad_norm": 0.13846907019615173,
"learning_rate": 1.89400348326856e-05,
"loss": 0.035211169719696046,
"step": 2090
},
{
"epoch": 1.744186046511628,
"grad_norm": 1.5302634239196777,
"learning_rate": 1.892794887493047e-05,
"loss": 0.07737213373184204,
"step": 2100
},
{
"epoch": 1.7524916943521593,
"grad_norm": 1.1371228694915771,
"learning_rate": 1.8915798305018845e-05,
"loss": 0.08640394806861877,
"step": 2110
},
{
"epoch": 1.760797342192691,
"grad_norm": 9.73719310760498,
"learning_rate": 1.89035832108852e-05,
"loss": 0.04082046747207642,
"step": 2120
},
{
"epoch": 1.7691029900332227,
"grad_norm": 1.2219951152801514,
"learning_rate": 1.8891303680930987e-05,
"loss": 0.023467710614204405,
"step": 2130
},
{
"epoch": 1.7774086378737541,
"grad_norm": 4.731136322021484,
"learning_rate": 1.8878959804023966e-05,
"loss": 0.06853907704353332,
"step": 2140
},
{
"epoch": 1.7857142857142856,
"grad_norm": 6.506364822387695,
"learning_rate": 1.88665516694976e-05,
"loss": 0.07831427454948425,
"step": 2150
},
{
"epoch": 1.7940199335548173,
"grad_norm": 1.6120778322219849,
"learning_rate": 1.8854079367150374e-05,
"loss": 0.05418685674667358,
"step": 2160
},
{
"epoch": 1.802325581395349,
"grad_norm": 2.708667039871216,
"learning_rate": 1.8841542987245164e-05,
"loss": 0.03725705742835998,
"step": 2170
},
{
"epoch": 1.8106312292358804,
"grad_norm": 0.14558850228786469,
"learning_rate": 1.882894262050859e-05,
"loss": 0.017846406996250154,
"step": 2180
},
{
"epoch": 1.8189368770764118,
"grad_norm": 1.5724000930786133,
"learning_rate": 1.8816278358130326e-05,
"loss": 0.0399004191160202,
"step": 2190
},
{
"epoch": 1.8272425249169435,
"grad_norm": 3.6788978576660156,
"learning_rate": 1.8803550291762485e-05,
"loss": 0.0954790711402893,
"step": 2200
},
{
"epoch": 1.8355481727574752,
"grad_norm": 1.1718946695327759,
"learning_rate": 1.879075851351892e-05,
"loss": 0.045383217930793765,
"step": 2210
},
{
"epoch": 1.8438538205980066,
"grad_norm": 0.3081478476524353,
"learning_rate": 1.8777903115974582e-05,
"loss": 0.03361668586730957,
"step": 2220
},
{
"epoch": 1.852159468438538,
"grad_norm": 0.06557083129882812,
"learning_rate": 1.8764984192164834e-05,
"loss": 0.004345468059182167,
"step": 2230
},
{
"epoch": 1.8604651162790697,
"grad_norm": 4.540051460266113,
"learning_rate": 1.875200183558477e-05,
"loss": 0.018223312497138978,
"step": 2240
},
{
"epoch": 1.8687707641196014,
"grad_norm": 5.702383041381836,
"learning_rate": 1.873895614018858e-05,
"loss": 0.04835923612117767,
"step": 2250
},
{
"epoch": 1.8770764119601329,
"grad_norm": 0.4880691468715668,
"learning_rate": 1.872584720038881e-05,
"loss": 0.1330762267112732,
"step": 2260
},
{
"epoch": 1.8853820598006643,
"grad_norm": 0.038495179265737534,
"learning_rate": 1.8712675111055733e-05,
"loss": 0.029408469796180725,
"step": 2270
},
{
"epoch": 1.893687707641196,
"grad_norm": 1.8155107498168945,
"learning_rate": 1.8699439967516627e-05,
"loss": 0.04721686244010925,
"step": 2280
},
{
"epoch": 1.9019933554817277,
"grad_norm": 0.045139130204916,
"learning_rate": 1.8686141865555104e-05,
"loss": 0.06960549354553222,
"step": 2290
},
{
"epoch": 1.910299003322259,
"grad_norm": 1.949537992477417,
"learning_rate": 1.8672780901410406e-05,
"loss": 0.06557999849319458,
"step": 2300
},
{
"epoch": 1.9186046511627906,
"grad_norm": 8.499605178833008,
"learning_rate": 1.865935717177671e-05,
"loss": 0.0804720401763916,
"step": 2310
},
{
"epoch": 1.9269102990033222,
"grad_norm": 1.1870088577270508,
"learning_rate": 1.8645870773802445e-05,
"loss": 0.048428884148597716,
"step": 2320
},
{
"epoch": 1.935215946843854,
"grad_norm": 2.916827440261841,
"learning_rate": 1.8632321805089555e-05,
"loss": 0.03602609932422638,
"step": 2330
},
{
"epoch": 1.9435215946843853,
"grad_norm": 4.362746715545654,
"learning_rate": 1.8618710363692826e-05,
"loss": 0.08624266386032105,
"step": 2340
},
{
"epoch": 1.9518272425249168,
"grad_norm": 3.9474129676818848,
"learning_rate": 1.8605036548119165e-05,
"loss": 0.050232100486755374,
"step": 2350
},
{
"epoch": 1.9601328903654485,
"grad_norm": 7.967952728271484,
"learning_rate": 1.859130045732687e-05,
"loss": 0.09872875809669494,
"step": 2360
},
{
"epoch": 1.9684385382059801,
"grad_norm": 3.190403938293457,
"learning_rate": 1.8577502190724947e-05,
"loss": 0.01989797055721283,
"step": 2370
},
{
"epoch": 1.9767441860465116,
"grad_norm": 0.3850468695163727,
"learning_rate": 1.856364184817236e-05,
"loss": 0.009820028394460677,
"step": 2380
},
{
"epoch": 1.985049833887043,
"grad_norm": 2.8296544551849365,
"learning_rate": 1.854971952997733e-05,
"loss": 0.0412475973367691,
"step": 2390
},
{
"epoch": 1.9933554817275747,
"grad_norm": 7.176604747772217,
"learning_rate": 1.8535735336896582e-05,
"loss": 0.08098717927932739,
"step": 2400
},
{
"epoch": 2.0016611295681064,
"grad_norm": 0.7510102987289429,
"learning_rate": 1.8521689370134657e-05,
"loss": 0.08423384428024291,
"step": 2410
},
{
"epoch": 2.009966777408638,
"grad_norm": 1.2272533178329468,
"learning_rate": 1.8507581731343144e-05,
"loss": 0.016476579010486603,
"step": 2420
},
{
"epoch": 2.0182724252491693,
"grad_norm": 0.013697882182896137,
"learning_rate": 1.8493412522619956e-05,
"loss": 0.03602006435394287,
"step": 2430
},
{
"epoch": 2.026578073089701,
"grad_norm": 0.037652477622032166,
"learning_rate": 1.8479181846508595e-05,
"loss": 0.022071529924869538,
"step": 2440
},
{
"epoch": 2.0348837209302326,
"grad_norm": 0.34081098437309265,
"learning_rate": 1.8464889805997406e-05,
"loss": 0.031250619888305665,
"step": 2450
},
{
"epoch": 2.043189368770764,
"grad_norm": 0.4983048439025879,
"learning_rate": 1.845053650451883e-05,
"loss": 0.031300300359725954,
"step": 2460
},
{
"epoch": 2.0514950166112955,
"grad_norm": 1.673167109489441,
"learning_rate": 1.843612204594866e-05,
"loss": 0.00861935168504715,
"step": 2470
},
{
"epoch": 2.0598006644518274,
"grad_norm": 4.775228977203369,
"learning_rate": 1.842164653460528e-05,
"loss": 0.02194608747959137,
"step": 2480
},
{
"epoch": 2.068106312292359,
"grad_norm": 0.9463720917701721,
"learning_rate": 1.8407110075248933e-05,
"loss": 0.027666324377059938,
"step": 2490
},
{
"epoch": 2.0764119601328903,
"grad_norm": 0.1325935274362564,
"learning_rate": 1.839251277308093e-05,
"loss": 0.05505828857421875,
"step": 2500
},
{
"epoch": 2.0764119601328903,
"eval_brier": 0.972,
"eval_c@1": 0.966,
"eval_f05u": 0.971,
"eval_f1": 0.968,
"eval_loss": 0.09322851151227951,
"eval_mean": 0.974,
"eval_roc-auc": 0.994,
"eval_runtime": 69.1528,
"eval_samples_per_second": 266.41,
"eval_steps_per_second": 4.165,
"step": 2500
},
{
"epoch": 2.0847176079734218,
"grad_norm": 0.13980907201766968,
"learning_rate": 1.8377854733742917e-05,
"loss": 0.009164278209209443,
"step": 2510
},
{
"epoch": 2.0930232558139537,
"grad_norm": 14.210821151733398,
"learning_rate": 1.8363136063316088e-05,
"loss": 0.010683315992355346,
"step": 2520
},
{
"epoch": 2.101328903654485,
"grad_norm": 4.6713032722473145,
"learning_rate": 1.834835686832044e-05,
"loss": 0.03733667433261871,
"step": 2530
},
{
"epoch": 2.1096345514950166,
"grad_norm": 0.5860873460769653,
"learning_rate": 1.833351725571398e-05,
"loss": 0.02269638329744339,
"step": 2540
},
{
"epoch": 2.117940199335548,
"grad_norm": 2.200061798095703,
"learning_rate": 1.831861733289197e-05,
"loss": 0.02127670347690582,
"step": 2550
},
{
"epoch": 2.12624584717608,
"grad_norm": 0.05895443633198738,
"learning_rate": 1.8303657207686137e-05,
"loss": 0.037612468004226685,
"step": 2560
},
{
"epoch": 2.1345514950166113,
"grad_norm": 0.008358967490494251,
"learning_rate": 1.8288636988363896e-05,
"loss": 0.014630788564682006,
"step": 2570
},
{
"epoch": 2.142857142857143,
"grad_norm": 0.016367830336093903,
"learning_rate": 1.827355678362756e-05,
"loss": 0.01680670529603958,
"step": 2580
},
{
"epoch": 2.1511627906976742,
"grad_norm": 0.05327534303069115,
"learning_rate": 1.825841670261358e-05,
"loss": 0.020807208120822908,
"step": 2590
},
{
"epoch": 2.159468438538206,
"grad_norm": 1.0469518899917603,
"learning_rate": 1.8243216854891718e-05,
"loss": 0.04530949294567108,
"step": 2600
},
{
"epoch": 2.1677740863787376,
"grad_norm": 0.5313214659690857,
"learning_rate": 1.8227957350464274e-05,
"loss": 0.011926326900720596,
"step": 2610
},
{
"epoch": 2.176079734219269,
"grad_norm": 0.31928905844688416,
"learning_rate": 1.82126382997653e-05,
"loss": 0.016019934415817262,
"step": 2620
},
{
"epoch": 2.1843853820598005,
"grad_norm": 0.03908394277095795,
"learning_rate": 1.819725981365977e-05,
"loss": 0.019885535538196563,
"step": 2630
},
{
"epoch": 2.1926910299003324,
"grad_norm": 0.03837399184703827,
"learning_rate": 1.8181822003442806e-05,
"loss": 0.011250430345535278,
"step": 2640
},
{
"epoch": 2.200996677740864,
"grad_norm": 0.10344582796096802,
"learning_rate": 1.816632498083887e-05,
"loss": 0.029060634970664977,
"step": 2650
},
{
"epoch": 2.2093023255813953,
"grad_norm": 0.06564024090766907,
"learning_rate": 1.8150768858000928e-05,
"loss": 0.03479950726032257,
"step": 2660
},
{
"epoch": 2.2176079734219267,
"grad_norm": 4.611400604248047,
"learning_rate": 1.813515374750968e-05,
"loss": 0.04098433554172516,
"step": 2670
},
{
"epoch": 2.2259136212624586,
"grad_norm": 0.09970167279243469,
"learning_rate": 1.8119479762372706e-05,
"loss": 0.007959491014480591,
"step": 2680
},
{
"epoch": 2.23421926910299,
"grad_norm": 0.029041744768619537,
"learning_rate": 1.8103747016023677e-05,
"loss": 0.028758007287979125,
"step": 2690
},
{
"epoch": 2.2425249169435215,
"grad_norm": 0.06805432587862015,
"learning_rate": 1.8087955622321517e-05,
"loss": 0.03202657103538513,
"step": 2700
},
{
"epoch": 2.250830564784053,
"grad_norm": 0.09453340619802475,
"learning_rate": 1.8072105695549588e-05,
"loss": 0.035026749968528746,
"step": 2710
},
{
"epoch": 2.259136212624585,
"grad_norm": 0.02279595658183098,
"learning_rate": 1.8056197350414857e-05,
"loss": 0.04423519670963287,
"step": 2720
},
{
"epoch": 2.2674418604651163,
"grad_norm": 1.4064537286758423,
"learning_rate": 1.8040230702047076e-05,
"loss": 0.015961700677871705,
"step": 2730
},
{
"epoch": 2.2757475083056478,
"grad_norm": 1.6543560028076172,
"learning_rate": 1.802420586599792e-05,
"loss": 0.024043741822242736,
"step": 2740
},
{
"epoch": 2.284053156146179,
"grad_norm": 0.7950565218925476,
"learning_rate": 1.8008122958240206e-05,
"loss": 0.011675900220870972,
"step": 2750
},
{
"epoch": 2.292358803986711,
"grad_norm": 13.708898544311523,
"learning_rate": 1.7991982095166988e-05,
"loss": 0.05354692339897156,
"step": 2760
},
{
"epoch": 2.3006644518272426,
"grad_norm": 0.21422316133975983,
"learning_rate": 1.7975783393590767e-05,
"loss": 0.03575671315193176,
"step": 2770
},
{
"epoch": 2.308970099667774,
"grad_norm": 0.06105324625968933,
"learning_rate": 1.7959526970742615e-05,
"loss": 0.0365283340215683,
"step": 2780
},
{
"epoch": 2.3172757475083055,
"grad_norm": 5.618215560913086,
"learning_rate": 1.7943212944271334e-05,
"loss": 0.0424439936876297,
"step": 2790
},
{
"epoch": 2.3255813953488373,
"grad_norm": 0.47717127203941345,
"learning_rate": 1.7926841432242625e-05,
"loss": 0.017289258539676666,
"step": 2800
},
{
"epoch": 2.333887043189369,
"grad_norm": 0.36353611946105957,
"learning_rate": 1.7910412553138193e-05,
"loss": 0.035215374827384946,
"step": 2810
},
{
"epoch": 2.3421926910299002,
"grad_norm": 0.7377805113792419,
"learning_rate": 1.7893926425854924e-05,
"loss": 0.018869006633758546,
"step": 2820
},
{
"epoch": 2.350498338870432,
"grad_norm": 0.009361806325614452,
"learning_rate": 1.7877383169704016e-05,
"loss": 0.003294094651937485,
"step": 2830
},
{
"epoch": 2.3588039867109636,
"grad_norm": 0.00782567448914051,
"learning_rate": 1.7860782904410103e-05,
"loss": 0.008531749993562699,
"step": 2840
},
{
"epoch": 2.367109634551495,
"grad_norm": 0.11934903264045715,
"learning_rate": 1.7844125750110405e-05,
"loss": 0.031217479705810548,
"step": 2850
},
{
"epoch": 2.3754152823920265,
"grad_norm": 0.7901199460029602,
"learning_rate": 1.7827411827353853e-05,
"loss": 0.03432442247867584,
"step": 2860
},
{
"epoch": 2.383720930232558,
"grad_norm": 0.06995333731174469,
"learning_rate": 1.7810641257100206e-05,
"loss": 0.01116345077753067,
"step": 2870
},
{
"epoch": 2.39202657807309,
"grad_norm": 0.5269324779510498,
"learning_rate": 1.7793814160719186e-05,
"loss": 0.013663582503795624,
"step": 2880
},
{
"epoch": 2.4003322259136213,
"grad_norm": 0.04618137329816818,
"learning_rate": 1.777693065998962e-05,
"loss": 0.010402342677116394,
"step": 2890
},
{
"epoch": 2.4086378737541527,
"grad_norm": 0.05698714405298233,
"learning_rate": 1.77599908770985e-05,
"loss": 0.022911915183067323,
"step": 2900
},
{
"epoch": 2.4169435215946846,
"grad_norm": 38.83896255493164,
"learning_rate": 1.7742994934640162e-05,
"loss": 0.02522394061088562,
"step": 2910
},
{
"epoch": 2.425249169435216,
"grad_norm": 0.1632380336523056,
"learning_rate": 1.772594295561537e-05,
"loss": 0.008206085860729217,
"step": 2920
},
{
"epoch": 2.4335548172757475,
"grad_norm": 1.6289150714874268,
"learning_rate": 1.770883506343042e-05,
"loss": 0.025161007046699525,
"step": 2930
},
{
"epoch": 2.441860465116279,
"grad_norm": 0.14310209453105927,
"learning_rate": 1.769167138189626e-05,
"loss": 0.0016272267326712607,
"step": 2940
},
{
"epoch": 2.4501661129568104,
"grad_norm": 0.083904929459095,
"learning_rate": 1.7674452035227586e-05,
"loss": 0.008397594094276428,
"step": 2950
},
{
"epoch": 2.4584717607973423,
"grad_norm": 0.6206247210502625,
"learning_rate": 1.7657177148041953e-05,
"loss": 0.06879042387008667,
"step": 2960
},
{
"epoch": 2.4667774086378738,
"grad_norm": 3.5987038612365723,
"learning_rate": 1.763984684535886e-05,
"loss": 0.04402000904083252,
"step": 2970
},
{
"epoch": 2.475083056478405,
"grad_norm": 0.16094276309013367,
"learning_rate": 1.7622461252598855e-05,
"loss": 0.0159262090921402,
"step": 2980
},
{
"epoch": 2.483388704318937,
"grad_norm": 2.2372312545776367,
"learning_rate": 1.760502049558262e-05,
"loss": 0.03023003935813904,
"step": 2990
},
{
"epoch": 2.4916943521594686,
"grad_norm": 0.7739448547363281,
"learning_rate": 1.758752470053007e-05,
"loss": 0.03061208426952362,
"step": 3000
},
{
"epoch": 2.4916943521594686,
"eval_brier": 0.969,
"eval_c@1": 0.963,
"eval_f05u": 0.959,
"eval_f1": 0.966,
"eval_loss": 0.11809011548757553,
"eval_mean": 0.97,
"eval_roc-auc": 0.995,
"eval_runtime": 69.3107,
"eval_samples_per_second": 265.803,
"eval_steps_per_second": 4.155,
"step": 3000
},
{
"epoch": 2.5,
"grad_norm": 2.1887381076812744,
"learning_rate": 1.7569973994059428e-05,
"loss": 0.01838263273239136,
"step": 3010
},
{
"epoch": 2.5083056478405314,
"grad_norm": 0.016387421637773514,
"learning_rate": 1.755236850318632e-05,
"loss": 0.009334413707256317,
"step": 3020
},
{
"epoch": 2.516611295681063,
"grad_norm": 2.8197641372680664,
"learning_rate": 1.753470835532285e-05,
"loss": 0.03760683834552765,
"step": 3030
},
{
"epoch": 2.524916943521595,
"grad_norm": 4.842959403991699,
"learning_rate": 1.751699367827667e-05,
"loss": 0.004269677400588989,
"step": 3040
},
{
"epoch": 2.5332225913621262,
"grad_norm": 2.8849728107452393,
"learning_rate": 1.749922460025007e-05,
"loss": 0.012648674845695495,
"step": 3050
},
{
"epoch": 2.5415282392026577,
"grad_norm": 11.304584503173828,
"learning_rate": 1.748140124983905e-05,
"loss": 0.05325874090194702,
"step": 3060
},
{
"epoch": 2.5498338870431896,
"grad_norm": 5.426615238189697,
"learning_rate": 1.746352375603237e-05,
"loss": 0.06052457690238953,
"step": 3070
},
{
"epoch": 2.558139534883721,
"grad_norm": 0.06185510754585266,
"learning_rate": 1.7445592248210634e-05,
"loss": 0.022402627766132353,
"step": 3080
},
{
"epoch": 2.5664451827242525,
"grad_norm": 0.4922488331794739,
"learning_rate": 1.7427606856145354e-05,
"loss": 0.020776252448558807,
"step": 3090
},
{
"epoch": 2.574750830564784,
"grad_norm": 0.3334231972694397,
"learning_rate": 1.7409567709998002e-05,
"loss": 0.06371630430221557,
"step": 3100
},
{
"epoch": 2.5830564784053154,
"grad_norm": 0.7910224199295044,
"learning_rate": 1.7391474940319068e-05,
"loss": 0.05364013314247131,
"step": 3110
},
{
"epoch": 2.5913621262458473,
"grad_norm": 0.006404863204807043,
"learning_rate": 1.7373328678047125e-05,
"loss": 0.018032759428024292,
"step": 3120
},
{
"epoch": 2.5996677740863787,
"grad_norm": 0.14521823823451996,
"learning_rate": 1.7355129054507873e-05,
"loss": 0.01069008857011795,
"step": 3130
},
{
"epoch": 2.60797342192691,
"grad_norm": 1.408150315284729,
"learning_rate": 1.733687620141319e-05,
"loss": 0.018587549030780793,
"step": 3140
},
{
"epoch": 2.616279069767442,
"grad_norm": 2.27775502204895,
"learning_rate": 1.7318570250860186e-05,
"loss": 0.045328712463378905,
"step": 3150
},
{
"epoch": 2.6245847176079735,
"grad_norm": 0.17508935928344727,
"learning_rate": 1.7300211335330228e-05,
"loss": 0.05949342846870422,
"step": 3160
},
{
"epoch": 2.632890365448505,
"grad_norm": 0.14915131032466888,
"learning_rate": 1.7281799587688008e-05,
"loss": 0.03284069001674652,
"step": 3170
},
{
"epoch": 2.6411960132890364,
"grad_norm": 0.03940584138035774,
"learning_rate": 1.7263335141180562e-05,
"loss": 0.011027585715055466,
"step": 3180
},
{
"epoch": 2.649501661129568,
"grad_norm": 0.054431017488241196,
"learning_rate": 1.7244818129436305e-05,
"loss": 0.012432112544775008,
"step": 3190
},
{
"epoch": 2.6578073089700998,
"grad_norm": 2.976839065551758,
"learning_rate": 1.722624868646408e-05,
"loss": 0.042336732149124146,
"step": 3200
},
{
"epoch": 2.666112956810631,
"grad_norm": 0.021991165354847908,
"learning_rate": 1.7207626946652173e-05,
"loss": 0.04990316927433014,
"step": 3210
},
{
"epoch": 2.6744186046511627,
"grad_norm": 0.032778576016426086,
"learning_rate": 1.7188953044767347e-05,
"loss": 0.022113136947155,
"step": 3220
},
{
"epoch": 2.6827242524916945,
"grad_norm": 7.216454982757568,
"learning_rate": 1.7170227115953865e-05,
"loss": 0.02170370817184448,
"step": 3230
},
{
"epoch": 2.691029900332226,
"grad_norm": 0.04913568124175072,
"learning_rate": 1.715144929573251e-05,
"loss": 0.02592366933822632,
"step": 3240
},
{
"epoch": 2.6993355481727574,
"grad_norm": 5.563662052154541,
"learning_rate": 1.7132619719999613e-05,
"loss": 0.01383553296327591,
"step": 3250
},
{
"epoch": 2.707641196013289,
"grad_norm": 0.36335670948028564,
"learning_rate": 1.7113738525026056e-05,
"loss": 0.026103276014328002,
"step": 3260
},
{
"epoch": 2.7159468438538203,
"grad_norm": 0.11070427298545837,
"learning_rate": 1.7094805847456296e-05,
"loss": 0.01853642761707306,
"step": 3270
},
{
"epoch": 2.7242524916943522,
"grad_norm": 0.05721287056803703,
"learning_rate": 1.7075821824307373e-05,
"loss": 0.029401224851608277,
"step": 3280
},
{
"epoch": 2.7325581395348837,
"grad_norm": 0.34963458776474,
"learning_rate": 1.705678659296792e-05,
"loss": 0.02173706144094467,
"step": 3290
},
{
"epoch": 2.740863787375415,
"grad_norm": 1.248716115951538,
"learning_rate": 1.7037700291197162e-05,
"loss": 0.01624070107936859,
"step": 3300
},
{
"epoch": 2.749169435215947,
"grad_norm": 0.019111966714262962,
"learning_rate": 1.7018563057123928e-05,
"loss": 0.03283540606498718,
"step": 3310
},
{
"epoch": 2.7574750830564785,
"grad_norm": 2.0715911388397217,
"learning_rate": 1.699937502924565e-05,
"loss": 0.010497630387544633,
"step": 3320
},
{
"epoch": 2.76578073089701,
"grad_norm": 0.14188338816165924,
"learning_rate": 1.698013634642735e-05,
"loss": 0.01480579972267151,
"step": 3330
},
{
"epoch": 2.7740863787375414,
"grad_norm": 0.03354760259389877,
"learning_rate": 1.696084714790065e-05,
"loss": 0.0029901012778282164,
"step": 3340
},
{
"epoch": 2.782392026578073,
"grad_norm": 0.08278927206993103,
"learning_rate": 1.694150757326276e-05,
"loss": 0.008234208077192306,
"step": 3350
},
{
"epoch": 2.7906976744186047,
"grad_norm": 0.30753204226493835,
"learning_rate": 1.6922117762475456e-05,
"loss": 0.01795431524515152,
"step": 3360
},
{
"epoch": 2.799003322259136,
"grad_norm": 0.31697142124176025,
"learning_rate": 1.690267785586408e-05,
"loss": 0.02110057771205902,
"step": 3370
},
{
"epoch": 2.8073089700996676,
"grad_norm": 0.003904345678165555,
"learning_rate": 1.6883187994116524e-05,
"loss": 0.013659822940826415,
"step": 3380
},
{
"epoch": 2.8156146179401995,
"grad_norm": 0.282507061958313,
"learning_rate": 1.6863648318282205e-05,
"loss": 0.02813073992729187,
"step": 3390
},
{
"epoch": 2.823920265780731,
"grad_norm": 1.7346644401550293,
"learning_rate": 1.6844058969771052e-05,
"loss": 0.0042308449745178224,
"step": 3400
},
{
"epoch": 2.8322259136212624,
"grad_norm": 0.06720908731222153,
"learning_rate": 1.6824420090352475e-05,
"loss": 0.0016714630648493767,
"step": 3410
},
{
"epoch": 2.840531561461794,
"grad_norm": 0.11770165711641312,
"learning_rate": 1.680473182215434e-05,
"loss": 0.020023314654827117,
"step": 3420
},
{
"epoch": 2.8488372093023253,
"grad_norm": 5.829126834869385,
"learning_rate": 1.678499430766195e-05,
"loss": 0.00045651253312826154,
"step": 3430
},
{
"epoch": 2.857142857142857,
"grad_norm": 2.2346742153167725,
"learning_rate": 1.6765207689717003e-05,
"loss": 0.024983009696006774,
"step": 3440
},
{
"epoch": 2.8654485049833887,
"grad_norm": 1.5816477537155151,
"learning_rate": 1.6745372111516557e-05,
"loss": 0.03904221951961517,
"step": 3450
},
{
"epoch": 2.87375415282392,
"grad_norm": 0.28397709131240845,
"learning_rate": 1.6725487716612007e-05,
"loss": 0.01208788827061653,
"step": 3460
},
{
"epoch": 2.882059800664452,
"grad_norm": 0.016718782484531403,
"learning_rate": 1.6705554648908037e-05,
"loss": 0.005875730887055397,
"step": 3470
},
{
"epoch": 2.8903654485049834,
"grad_norm": 0.18707916140556335,
"learning_rate": 1.668557305266157e-05,
"loss": 0.016105698049068452,
"step": 3480
},
{
"epoch": 2.898671096345515,
"grad_norm": 0.09086672961711884,
"learning_rate": 1.6665543072480744e-05,
"loss": 0.012864015996456146,
"step": 3490
},
{
"epoch": 2.9069767441860463,
"grad_norm": 10.208978652954102,
"learning_rate": 1.664546485332385e-05,
"loss": 0.029475682973861696,
"step": 3500
},
{
"epoch": 2.9069767441860463,
"eval_brier": 0.973,
"eval_c@1": 0.969,
"eval_f05u": 0.975,
"eval_f1": 0.971,
"eval_loss": 0.09430566430091858,
"eval_mean": 0.976,
"eval_roc-auc": 0.994,
"eval_runtime": 69.257,
"eval_samples_per_second": 266.009,
"eval_steps_per_second": 4.158,
"step": 3500
},
{
"epoch": 2.915282392026578,
"grad_norm": 2.053638458251953,
"learning_rate": 1.6625338540498285e-05,
"loss": 0.02082843631505966,
"step": 3510
},
{
"epoch": 2.9235880398671097,
"grad_norm": 0.05038900673389435,
"learning_rate": 1.6605164279659508e-05,
"loss": 0.01825982630252838,
"step": 3520
},
{
"epoch": 2.931893687707641,
"grad_norm": 1.7743593454360962,
"learning_rate": 1.6584942216809978e-05,
"loss": 0.008388277888298035,
"step": 3530
},
{
"epoch": 2.940199335548173,
"grad_norm": 11.47936725616455,
"learning_rate": 1.6564672498298103e-05,
"loss": 0.03784654140472412,
"step": 3540
},
{
"epoch": 2.9485049833887045,
"grad_norm": 0.07685818523168564,
"learning_rate": 1.654435527081718e-05,
"loss": 0.05816956162452698,
"step": 3550
},
{
"epoch": 2.956810631229236,
"grad_norm": 0.05088737607002258,
"learning_rate": 1.652399068140432e-05,
"loss": 0.04174873232841492,
"step": 3560
},
{
"epoch": 2.9651162790697674,
"grad_norm": 0.5170846581459045,
"learning_rate": 1.6503578877439403e-05,
"loss": 0.028970912098884583,
"step": 3570
},
{
"epoch": 2.973421926910299,
"grad_norm": 1.6754320859909058,
"learning_rate": 1.6483120006644014e-05,
"loss": 0.004614223912358284,
"step": 3580
},
{
"epoch": 2.9817275747508307,
"grad_norm": 19.248220443725586,
"learning_rate": 1.6462614217080347e-05,
"loss": 0.027228891849517822,
"step": 3590
},
{
"epoch": 2.990033222591362,
"grad_norm": 0.023427477106451988,
"learning_rate": 1.6442061657150163e-05,
"loss": 0.04394007623195648,
"step": 3600
},
{
"epoch": 2.9983388704318936,
"grad_norm": 18.01772689819336,
"learning_rate": 1.642146247559369e-05,
"loss": 0.09833118915557862,
"step": 3610
},
{
"epoch": 3.006644518272425,
"grad_norm": 1.7291778326034546,
"learning_rate": 1.640081682148857e-05,
"loss": 0.1273277997970581,
"step": 3620
},
{
"epoch": 3.014950166112957,
"grad_norm": 0.11676451563835144,
"learning_rate": 1.6380124844248772e-05,
"loss": 0.04351269900798797,
"step": 3630
},
{
"epoch": 3.0232558139534884,
"grad_norm": 3.6196670532226562,
"learning_rate": 1.6359386693623496e-05,
"loss": 0.019785214960575104,
"step": 3640
},
{
"epoch": 3.03156146179402,
"grad_norm": 3.2134315967559814,
"learning_rate": 1.6338602519696115e-05,
"loss": 0.026179811358451842,
"step": 3650
},
{
"epoch": 3.0398671096345513,
"grad_norm": 1.1688365936279297,
"learning_rate": 1.631777247288307e-05,
"loss": 0.02903587818145752,
"step": 3660
},
{
"epoch": 3.048172757475083,
"grad_norm": 0.2987384796142578,
"learning_rate": 1.6296896703932784e-05,
"loss": 0.04616578221321106,
"step": 3670
},
{
"epoch": 3.0564784053156147,
"grad_norm": 0.045798394829034805,
"learning_rate": 1.6275975363924578e-05,
"loss": 0.02623589038848877,
"step": 3680
},
{
"epoch": 3.064784053156146,
"grad_norm": 0.13879990577697754,
"learning_rate": 1.625500860426758e-05,
"loss": 0.004535328969359398,
"step": 3690
},
{
"epoch": 3.0730897009966776,
"grad_norm": 0.010658717714250088,
"learning_rate": 1.6233996576699614e-05,
"loss": 0.017601104080677034,
"step": 3700
},
{
"epoch": 3.0813953488372094,
"grad_norm": 0.24265863001346588,
"learning_rate": 1.6212939433286106e-05,
"loss": 0.03232440054416656,
"step": 3710
},
{
"epoch": 3.089700996677741,
"grad_norm": 0.25457021594047546,
"learning_rate": 1.6191837326419e-05,
"loss": 0.04634428024291992,
"step": 3720
},
{
"epoch": 3.0980066445182723,
"grad_norm": 0.10299156606197357,
"learning_rate": 1.6170690408815644e-05,
"loss": 0.012248477339744568,
"step": 3730
},
{
"epoch": 3.106312292358804,
"grad_norm": 1.0204440355300903,
"learning_rate": 1.6149498833517672e-05,
"loss": 0.024753575026988984,
"step": 3740
},
{
"epoch": 3.1146179401993357,
"grad_norm": 0.02626228705048561,
"learning_rate": 1.612826275388992e-05,
"loss": 0.0205381378531456,
"step": 3750
},
{
"epoch": 3.122923588039867,
"grad_norm": 0.037694402039051056,
"learning_rate": 1.61069823236193e-05,
"loss": 0.009481345862150192,
"step": 3760
},
{
"epoch": 3.1312292358803986,
"grad_norm": 0.921624481678009,
"learning_rate": 1.608565769671369e-05,
"loss": 0.03205291926860809,
"step": 3770
},
{
"epoch": 3.13953488372093,
"grad_norm": 1.130007266998291,
"learning_rate": 1.6064289027500833e-05,
"loss": 0.03451147675514221,
"step": 3780
},
{
"epoch": 3.147840531561462,
"grad_norm": 1.1052535772323608,
"learning_rate": 1.6042876470627195e-05,
"loss": 0.030246618390083312,
"step": 3790
},
{
"epoch": 3.1561461794019934,
"grad_norm": 0.011895990930497646,
"learning_rate": 1.602142018105687e-05,
"loss": 0.023467601835727693,
"step": 3800
},
{
"epoch": 3.164451827242525,
"grad_norm": 5.539122104644775,
"learning_rate": 1.5999920314070442e-05,
"loss": 0.004532170668244362,
"step": 3810
},
{
"epoch": 3.1727574750830563,
"grad_norm": 0.29301217198371887,
"learning_rate": 1.5978377025263865e-05,
"loss": 0.026630595326423645,
"step": 3820
},
{
"epoch": 3.181063122923588,
"grad_norm": 0.9117482304573059,
"learning_rate": 1.595679047054736e-05,
"loss": 0.025602924823760986,
"step": 3830
},
{
"epoch": 3.1893687707641196,
"grad_norm": 0.004922331776469946,
"learning_rate": 1.5935160806144238e-05,
"loss": 0.037556436657905576,
"step": 3840
},
{
"epoch": 3.197674418604651,
"grad_norm": 1.040724754333496,
"learning_rate": 1.5913488188589815e-05,
"loss": 0.010435221344232559,
"step": 3850
},
{
"epoch": 3.2059800664451825,
"grad_norm": 0.02429218962788582,
"learning_rate": 1.589177277473026e-05,
"loss": 0.005302220210433006,
"step": 3860
},
{
"epoch": 3.2142857142857144,
"grad_norm": 0.4984890818595886,
"learning_rate": 1.5870014721721457e-05,
"loss": 0.01583288609981537,
"step": 3870
},
{
"epoch": 3.222591362126246,
"grad_norm": 2.0418243408203125,
"learning_rate": 1.5848214187027873e-05,
"loss": 0.016842301189899444,
"step": 3880
},
{
"epoch": 3.2308970099667773,
"grad_norm": 1.155386209487915,
"learning_rate": 1.5826371328421425e-05,
"loss": 0.03409614264965057,
"step": 3890
},
{
"epoch": 3.2392026578073088,
"grad_norm": 1.8344030380249023,
"learning_rate": 1.580448630398032e-05,
"loss": 0.03606152236461639,
"step": 3900
},
{
"epoch": 3.2475083056478407,
"grad_norm": 0.6460553407669067,
"learning_rate": 1.578255927208793e-05,
"loss": 0.005451107397675514,
"step": 3910
},
{
"epoch": 3.255813953488372,
"grad_norm": 1.2683637142181396,
"learning_rate": 1.576059039143163e-05,
"loss": 0.03246237635612488,
"step": 3920
},
{
"epoch": 3.2641196013289036,
"grad_norm": 0.06928703933954239,
"learning_rate": 1.573857982100167e-05,
"loss": 0.03475654423236847,
"step": 3930
},
{
"epoch": 3.2724252491694354,
"grad_norm": 0.15729838609695435,
"learning_rate": 1.571652772009e-05,
"loss": 0.021227990090847016,
"step": 3940
},
{
"epoch": 3.280730897009967,
"grad_norm": 1.0783848762512207,
"learning_rate": 1.5694434248289138e-05,
"loss": 0.02351810485124588,
"step": 3950
},
{
"epoch": 3.2890365448504983,
"grad_norm": 0.05579701066017151,
"learning_rate": 1.567229956549099e-05,
"loss": 0.006204274669289589,
"step": 3960
},
{
"epoch": 3.29734219269103,
"grad_norm": 1.2872422933578491,
"learning_rate": 1.5650123831885725e-05,
"loss": 0.03524799346923828,
"step": 3970
},
{
"epoch": 3.3056478405315612,
"grad_norm": 0.0025577384512871504,
"learning_rate": 1.5627907207960594e-05,
"loss": 0.03910149931907654,
"step": 3980
},
{
"epoch": 3.313953488372093,
"grad_norm": 0.09006145596504211,
"learning_rate": 1.5605649854498776e-05,
"loss": 0.028400149941444398,
"step": 3990
},
{
"epoch": 3.3222591362126246,
"grad_norm": 0.014801884070038795,
"learning_rate": 1.558335193257821e-05,
"loss": 0.034506013989448546,
"step": 4000
},
{
"epoch": 3.3222591362126246,
"eval_brier": 0.962,
"eval_c@1": 0.955,
"eval_f05u": 0.972,
"eval_f1": 0.957,
"eval_loss": 0.13631200790405273,
"eval_mean": 0.967,
"eval_roc-auc": 0.989,
"eval_runtime": 69.2326,
"eval_samples_per_second": 266.103,
"eval_steps_per_second": 4.16,
"step": 4000
},
{
"epoch": 3.330564784053156,
"grad_norm": 0.16002759337425232,
"learning_rate": 1.5561013603570424e-05,
"loss": 0.005246283113956451,
"step": 4010
},
{
"epoch": 3.338870431893688,
"grad_norm": 0.35845229029655457,
"learning_rate": 1.553863502913939e-05,
"loss": 0.014781996607780457,
"step": 4020
},
{
"epoch": 3.3471760797342194,
"grad_norm": 0.7351519465446472,
"learning_rate": 1.551621637124033e-05,
"loss": 0.04960608184337616,
"step": 4030
},
{
"epoch": 3.355481727574751,
"grad_norm": 3.097801923751831,
"learning_rate": 1.5493757792118554e-05,
"loss": 0.009672880172729492,
"step": 4040
},
{
"epoch": 3.3637873754152823,
"grad_norm": 4.237391471862793,
"learning_rate": 1.5471259454308283e-05,
"loss": 0.01681814044713974,
"step": 4050
},
{
"epoch": 3.3720930232558137,
"grad_norm": 0.042575765401124954,
"learning_rate": 1.544872152063148e-05,
"loss": 0.01027262881398201,
"step": 4060
},
{
"epoch": 3.3803986710963456,
"grad_norm": 0.37154197692871094,
"learning_rate": 1.542614415419665e-05,
"loss": 0.02080361694097519,
"step": 4070
},
{
"epoch": 3.388704318936877,
"grad_norm": 0.002169161569327116,
"learning_rate": 1.5403527518397702e-05,
"loss": 0.005771913379430771,
"step": 4080
},
{
"epoch": 3.3970099667774085,
"grad_norm": 2.173062562942505,
"learning_rate": 1.5380871776912715e-05,
"loss": 0.03047383725643158,
"step": 4090
},
{
"epoch": 3.4053156146179404,
"grad_norm": 0.3199882209300995,
"learning_rate": 1.5358177093702798e-05,
"loss": 0.027394527196884157,
"step": 4100
},
{
"epoch": 3.413621262458472,
"grad_norm": 0.025860700756311417,
"learning_rate": 1.5335443633010866e-05,
"loss": 0.0011777013540267944,
"step": 4110
},
{
"epoch": 3.4219269102990033,
"grad_norm": 0.17993803322315216,
"learning_rate": 1.5312671559360474e-05,
"loss": 0.033722972869873045,
"step": 4120
},
{
"epoch": 3.4302325581395348,
"grad_norm": 1.2421197891235352,
"learning_rate": 1.5289861037554636e-05,
"loss": 0.028782588243484498,
"step": 4130
},
{
"epoch": 3.438538205980066,
"grad_norm": 0.22248853743076324,
"learning_rate": 1.52670122326746e-05,
"loss": 0.03898056745529175,
"step": 4140
},
{
"epoch": 3.446843853820598,
"grad_norm": 2.0566086769104004,
"learning_rate": 1.5244125310078688e-05,
"loss": 0.020737478137016298,
"step": 4150
},
{
"epoch": 3.4551495016611296,
"grad_norm": 0.014175049029290676,
"learning_rate": 1.5221200435401055e-05,
"loss": 0.019598761200904848,
"step": 4160
},
{
"epoch": 3.463455149501661,
"grad_norm": 8.482834815979004,
"learning_rate": 1.5198237774550548e-05,
"loss": 0.055632084608078,
"step": 4170
},
{
"epoch": 3.471760797342193,
"grad_norm": 0.08094330877065659,
"learning_rate": 1.5175237493709463e-05,
"loss": 0.037429016828536985,
"step": 4180
},
{
"epoch": 3.4800664451827243,
"grad_norm": 2.0313167572021484,
"learning_rate": 1.515219975933235e-05,
"loss": 0.028607624769210815,
"step": 4190
},
{
"epoch": 3.488372093023256,
"grad_norm": 0.779374897480011,
"learning_rate": 1.5129124738144812e-05,
"loss": 0.023052971065044402,
"step": 4200
},
{
"epoch": 3.4966777408637872,
"grad_norm": 0.14733143150806427,
"learning_rate": 1.5106012597142307e-05,
"loss": 0.014119112491607666,
"step": 4210
},
{
"epoch": 3.5049833887043187,
"grad_norm": 4.563329696655273,
"learning_rate": 1.5082863503588932e-05,
"loss": 0.04059003591537476,
"step": 4220
},
{
"epoch": 3.5132890365448506,
"grad_norm": 2.1403205394744873,
"learning_rate": 1.5059677625016204e-05,
"loss": 0.019231592118740082,
"step": 4230
},
{
"epoch": 3.521594684385382,
"grad_norm": 1.079723834991455,
"learning_rate": 1.5036455129221863e-05,
"loss": 0.009590108692646027,
"step": 4240
},
{
"epoch": 3.5299003322259135,
"grad_norm": 0.31128084659576416,
"learning_rate": 1.5013196184268646e-05,
"loss": 0.02513335943222046,
"step": 4250
},
{
"epoch": 3.5382059800664454,
"grad_norm": 0.06056709587574005,
"learning_rate": 1.498990095848307e-05,
"loss": 0.01343282163143158,
"step": 4260
},
{
"epoch": 3.546511627906977,
"grad_norm": 2.509009599685669,
"learning_rate": 1.4966569620454233e-05,
"loss": 0.013725298643112182,
"step": 4270
},
{
"epoch": 3.5548172757475083,
"grad_norm": 0.0038351700641214848,
"learning_rate": 1.4943202339032565e-05,
"loss": 0.03707950413227081,
"step": 4280
},
{
"epoch": 3.5631229235880397,
"grad_norm": 3.0641913414001465,
"learning_rate": 1.4919799283328633e-05,
"loss": 0.020812991261482238,
"step": 4290
},
{
"epoch": 3.571428571428571,
"grad_norm": 0.0017618021229282022,
"learning_rate": 1.4896360622711895e-05,
"loss": 0.005693419277668,
"step": 4300
},
{
"epoch": 3.579734219269103,
"grad_norm": 0.015993591398000717,
"learning_rate": 1.4872886526809486e-05,
"loss": 0.004864529147744179,
"step": 4310
},
{
"epoch": 3.5880398671096345,
"grad_norm": 2.977465867996216,
"learning_rate": 1.4849377165504999e-05,
"loss": 0.018113422393798827,
"step": 4320
},
{
"epoch": 3.596345514950166,
"grad_norm": 2.881437301635742,
"learning_rate": 1.4825832708937231e-05,
"loss": 0.0352090984582901,
"step": 4330
},
{
"epoch": 3.604651162790698,
"grad_norm": 2.6090734004974365,
"learning_rate": 1.4802253327498967e-05,
"loss": 0.021929188072681426,
"step": 4340
},
{
"epoch": 3.6129568106312293,
"grad_norm": 2.509809970855713,
"learning_rate": 1.477863919183576e-05,
"loss": 0.014704853296279907,
"step": 4350
},
{
"epoch": 3.6212624584717608,
"grad_norm": 1.6907615661621094,
"learning_rate": 1.4754990472844663e-05,
"loss": 0.016171935200691222,
"step": 4360
},
{
"epoch": 3.629568106312292,
"grad_norm": 16.249902725219727,
"learning_rate": 1.4731307341673025e-05,
"loss": 0.05183131098747253,
"step": 4370
},
{
"epoch": 3.6378737541528237,
"grad_norm": 1.8737428188323975,
"learning_rate": 1.470758996971723e-05,
"loss": 0.03585072159767151,
"step": 4380
},
{
"epoch": 3.6461794019933556,
"grad_norm": 2.861999750137329,
"learning_rate": 1.4683838528621465e-05,
"loss": 0.014898061752319336,
"step": 4390
},
{
"epoch": 3.654485049833887,
"grad_norm": 1.5936429500579834,
"learning_rate": 1.4660053190276489e-05,
"loss": 0.034113848209381105,
"step": 4400
},
{
"epoch": 3.6627906976744184,
"grad_norm": 1.4495152235031128,
"learning_rate": 1.4636234126818356e-05,
"loss": 0.008258716017007828,
"step": 4410
},
{
"epoch": 3.6710963455149503,
"grad_norm": 0.2400159239768982,
"learning_rate": 1.4612381510627214e-05,
"loss": 0.016240420937538146,
"step": 4420
},
{
"epoch": 3.679401993355482,
"grad_norm": 0.08169445395469666,
"learning_rate": 1.4588495514326023e-05,
"loss": 0.03803572058677673,
"step": 4430
},
{
"epoch": 3.6877076411960132,
"grad_norm": 0.9175194501876831,
"learning_rate": 1.4564576310779313e-05,
"loss": 0.0373820960521698,
"step": 4440
},
{
"epoch": 3.6960132890365447,
"grad_norm": 0.006045251619070768,
"learning_rate": 1.4540624073091945e-05,
"loss": 0.02483644485473633,
"step": 4450
},
{
"epoch": 3.704318936877076,
"grad_norm": 10.36731243133545,
"learning_rate": 1.4516638974607844e-05,
"loss": 0.02520762085914612,
"step": 4460
},
{
"epoch": 3.712624584717608,
"grad_norm": 15.544453620910645,
"learning_rate": 1.449262118890876e-05,
"loss": 0.03825234174728394,
"step": 4470
},
{
"epoch": 3.7209302325581395,
"grad_norm": 12.39433765411377,
"learning_rate": 1.4468570889812994e-05,
"loss": 0.039244210720062254,
"step": 4480
},
{
"epoch": 3.729235880398671,
"grad_norm": 1.8302897214889526,
"learning_rate": 1.444448825137415e-05,
"loss": 0.03714895248413086,
"step": 4490
},
{
"epoch": 3.737541528239203,
"grad_norm": 7.93354606628418,
"learning_rate": 1.4420373447879877e-05,
"loss": 0.05549731254577637,
"step": 4500
},
{
"epoch": 3.737541528239203,
"eval_brier": 0.964,
"eval_c@1": 0.958,
"eval_f05u": 0.976,
"eval_f1": 0.96,
"eval_loss": 0.13261988759040833,
"eval_mean": 0.97,
"eval_roc-auc": 0.991,
"eval_runtime": 69.2133,
"eval_samples_per_second": 266.177,
"eval_steps_per_second": 4.161,
"step": 4500
},
{
"epoch": 3.7458471760797343,
"grad_norm": 10.519672393798828,
"learning_rate": 1.43962266538506e-05,
"loss": 0.017514997720718385,
"step": 4510
},
{
"epoch": 3.7541528239202657,
"grad_norm": 0.30851686000823975,
"learning_rate": 1.4372048044038265e-05,
"loss": 0.020422892272472383,
"step": 4520
},
{
"epoch": 3.762458471760797,
"grad_norm": 0.2577485740184784,
"learning_rate": 1.4347837793425075e-05,
"loss": 0.024028824269771577,
"step": 4530
},
{
"epoch": 3.770764119601329,
"grad_norm": 1.4899733066558838,
"learning_rate": 1.432359607722221e-05,
"loss": 0.01535840928554535,
"step": 4540
},
{
"epoch": 3.7790697674418605,
"grad_norm": 0.02254222333431244,
"learning_rate": 1.4299323070868574e-05,
"loss": 0.02634434401988983,
"step": 4550
},
{
"epoch": 3.787375415282392,
"grad_norm": 1.4629563093185425,
"learning_rate": 1.4275018950029522e-05,
"loss": 0.012480440735816955,
"step": 4560
},
{
"epoch": 3.795681063122924,
"grad_norm": 0.8149392604827881,
"learning_rate": 1.4250683890595583e-05,
"loss": 0.0376665472984314,
"step": 4570
},
{
"epoch": 3.8039867109634553,
"grad_norm": 0.6940101981163025,
"learning_rate": 1.4226318068681191e-05,
"loss": 0.03424222469329834,
"step": 4580
},
{
"epoch": 3.8122923588039868,
"grad_norm": 0.2269517481327057,
"learning_rate": 1.4201921660623407e-05,
"loss": 0.0228074312210083,
"step": 4590
},
{
"epoch": 3.820598006644518,
"grad_norm": 10.251777648925781,
"learning_rate": 1.4177494842980657e-05,
"loss": 0.016562822461128234,
"step": 4600
},
{
"epoch": 3.8289036544850497,
"grad_norm": 1.0324342250823975,
"learning_rate": 1.4153037792531432e-05,
"loss": 0.016989651322364806,
"step": 4610
},
{
"epoch": 3.8372093023255816,
"grad_norm": 1.4060617685317993,
"learning_rate": 1.4128550686273021e-05,
"loss": 0.0122324638068676,
"step": 4620
},
{
"epoch": 3.845514950166113,
"grad_norm": 0.9087185263633728,
"learning_rate": 1.410403370142024e-05,
"loss": 0.019722691178321837,
"step": 4630
},
{
"epoch": 3.8538205980066444,
"grad_norm": 0.00458538020029664,
"learning_rate": 1.4079487015404122e-05,
"loss": 0.031716141104698184,
"step": 4640
},
{
"epoch": 3.8621262458471763,
"grad_norm": 2.5713677406311035,
"learning_rate": 1.4054910805870657e-05,
"loss": 0.026626569032669068,
"step": 4650
},
{
"epoch": 3.870431893687708,
"grad_norm": 0.016461262479424477,
"learning_rate": 1.4030305250679501e-05,
"loss": 0.0038371700793504715,
"step": 4660
},
{
"epoch": 3.8787375415282392,
"grad_norm": 0.10149026662111282,
"learning_rate": 1.4005670527902688e-05,
"loss": 0.032883715629577634,
"step": 4670
},
{
"epoch": 3.8870431893687707,
"grad_norm": 0.22568564116954803,
"learning_rate": 1.3981006815823332e-05,
"loss": 0.025000736117362976,
"step": 4680
},
{
"epoch": 3.895348837209302,
"grad_norm": 5.45408296585083,
"learning_rate": 1.395631429293435e-05,
"loss": 0.04558796882629394,
"step": 4690
},
{
"epoch": 3.903654485049834,
"grad_norm": 4.78015661239624,
"learning_rate": 1.3931593137937162e-05,
"loss": 0.01910436004400253,
"step": 4700
},
{
"epoch": 3.9119601328903655,
"grad_norm": 0.3174411356449127,
"learning_rate": 1.3906843529740402e-05,
"loss": 0.014453631639480592,
"step": 4710
},
{
"epoch": 3.920265780730897,
"grad_norm": 0.9285358190536499,
"learning_rate": 1.3882065647458622e-05,
"loss": 0.0379861444234848,
"step": 4720
},
{
"epoch": 3.928571428571429,
"grad_norm": 1.424649953842163,
"learning_rate": 1.3857259670410994e-05,
"loss": 0.008829368650913239,
"step": 4730
},
{
"epoch": 3.9368770764119603,
"grad_norm": 0.034531790763139725,
"learning_rate": 1.3832425778120015e-05,
"loss": 0.02229728102684021,
"step": 4740
},
{
"epoch": 3.9451827242524917,
"grad_norm": 0.01387101598083973,
"learning_rate": 1.3807564150310211e-05,
"loss": 0.018911705911159517,
"step": 4750
},
{
"epoch": 3.953488372093023,
"grad_norm": 6.543034553527832,
"learning_rate": 1.3782674966906817e-05,
"loss": 0.01977894902229309,
"step": 4760
},
{
"epoch": 3.9617940199335546,
"grad_norm": 8.277469635009766,
"learning_rate": 1.3757758408034503e-05,
"loss": 0.057103073596954344,
"step": 4770
},
{
"epoch": 3.9700996677740865,
"grad_norm": 1.6037797927856445,
"learning_rate": 1.373281465401606e-05,
"loss": 0.011168943345546722,
"step": 4780
},
{
"epoch": 3.978405315614618,
"grad_norm": 1.8816139698028564,
"learning_rate": 1.370784388537108e-05,
"loss": 0.04256616830825806,
"step": 4790
},
{
"epoch": 3.9867109634551494,
"grad_norm": 0.39578917622566223,
"learning_rate": 1.3682846282814663e-05,
"loss": 0.018482643365859985,
"step": 4800
},
{
"epoch": 3.9950166112956813,
"grad_norm": 0.12206543982028961,
"learning_rate": 1.365782202725612e-05,
"loss": 0.002756812050938606,
"step": 4810
},
{
"epoch": 4.003322259136213,
"grad_norm": 0.005353462416678667,
"learning_rate": 1.363277129979764e-05,
"loss": 0.03520027995109558,
"step": 4820
},
{
"epoch": 4.011627906976744,
"grad_norm": 1.1566811800003052,
"learning_rate": 1.3607694281732994e-05,
"loss": 0.06577891111373901,
"step": 4830
},
{
"epoch": 4.019933554817276,
"grad_norm": 0.02166386879980564,
"learning_rate": 1.3582591154546225e-05,
"loss": 0.03531816899776459,
"step": 4840
},
{
"epoch": 4.028239202657807,
"grad_norm": 0.22449389100074768,
"learning_rate": 1.3557462099910319e-05,
"loss": 0.027242958545684814,
"step": 4850
},
{
"epoch": 4.0365448504983386,
"grad_norm": 0.011027490720152855,
"learning_rate": 1.3532307299685905e-05,
"loss": 0.02250935584306717,
"step": 4860
},
{
"epoch": 4.04485049833887,
"grad_norm": 1.7733840942382812,
"learning_rate": 1.3507126935919944e-05,
"loss": 0.021818952262401582,
"step": 4870
},
{
"epoch": 4.053156146179402,
"grad_norm": 0.8993972539901733,
"learning_rate": 1.3481921190844386e-05,
"loss": 0.03043949604034424,
"step": 4880
},
{
"epoch": 4.061461794019934,
"grad_norm": 0.055961064994335175,
"learning_rate": 1.3456690246874882e-05,
"loss": 0.017452344298362732,
"step": 4890
},
{
"epoch": 4.069767441860465,
"grad_norm": 0.7132924795150757,
"learning_rate": 1.3431434286609435e-05,
"loss": 0.0379785031080246,
"step": 4900
},
{
"epoch": 4.078073089700997,
"grad_norm": 1.0427615642547607,
"learning_rate": 1.3406153492827103e-05,
"loss": 0.0008547148667275905,
"step": 4910
},
{
"epoch": 4.086378737541528,
"grad_norm": 1.5046710968017578,
"learning_rate": 1.3380848048486665e-05,
"loss": 0.03196828067302704,
"step": 4920
},
{
"epoch": 4.09468438538206,
"grad_norm": 2.3594045639038086,
"learning_rate": 1.3355518136725296e-05,
"loss": 0.05024482607841492,
"step": 4930
},
{
"epoch": 4.102990033222591,
"grad_norm": 3.284416675567627,
"learning_rate": 1.3330163940857237e-05,
"loss": 0.022133491933345795,
"step": 4940
},
{
"epoch": 4.1112956810631225,
"grad_norm": 0.07618211954832077,
"learning_rate": 1.3304785644372484e-05,
"loss": 0.006939659267663956,
"step": 4950
},
{
"epoch": 4.119601328903655,
"grad_norm": 0.0068997573107481,
"learning_rate": 1.327938343093545e-05,
"loss": 0.010384534299373627,
"step": 4960
},
{
"epoch": 4.127906976744186,
"grad_norm": 7.505576133728027,
"learning_rate": 1.3253957484383625e-05,
"loss": 0.017249995470047,
"step": 4970
},
{
"epoch": 4.136212624584718,
"grad_norm": 1.9831697940826416,
"learning_rate": 1.3228507988726275e-05,
"loss": 0.006020967662334442,
"step": 4980
},
{
"epoch": 4.144518272425249,
"grad_norm": 1.6030547618865967,
"learning_rate": 1.320303512814308e-05,
"loss": 0.060809260606765746,
"step": 4990
},
{
"epoch": 4.152823920265781,
"grad_norm": 1.171621561050415,
"learning_rate": 1.3177539086982818e-05,
"loss": 0.04928867816925049,
"step": 5000
},
{
"epoch": 4.152823920265781,
"eval_brier": 0.96,
"eval_c@1": 0.954,
"eval_f05u": 0.963,
"eval_f1": 0.957,
"eval_loss": 0.15997743606567383,
"eval_mean": 0.965,
"eval_roc-auc": 0.991,
"eval_runtime": 69.2513,
"eval_samples_per_second": 266.031,
"eval_steps_per_second": 4.159,
"step": 5000
},
{
"epoch": 4.161129568106312,
"grad_norm": 0.01830478198826313,
"learning_rate": 1.3152020049762017e-05,
"loss": 0.025147449970245362,
"step": 5010
},
{
"epoch": 4.1694352159468435,
"grad_norm": 2.3963396549224854,
"learning_rate": 1.312647820116365e-05,
"loss": 0.02699088752269745,
"step": 5020
},
{
"epoch": 4.177740863787376,
"grad_norm": 0.7355082631111145,
"learning_rate": 1.3100913726035757e-05,
"loss": 0.03964338004589081,
"step": 5030
},
{
"epoch": 4.186046511627907,
"grad_norm": 0.8636714816093445,
"learning_rate": 1.3075326809390137e-05,
"loss": 0.030416178703308105,
"step": 5040
},
{
"epoch": 4.194352159468439,
"grad_norm": 1.0738918781280518,
"learning_rate": 1.3049717636400998e-05,
"loss": 0.011282040923833846,
"step": 5050
},
{
"epoch": 4.20265780730897,
"grad_norm": 1.6882463693618774,
"learning_rate": 1.3024086392403616e-05,
"loss": 0.037940877676010135,
"step": 5060
},
{
"epoch": 4.210963455149502,
"grad_norm": 0.2809486985206604,
"learning_rate": 1.2998433262893004e-05,
"loss": 0.00866124927997589,
"step": 5070
},
{
"epoch": 4.219269102990033,
"grad_norm": 0.26813581585884094,
"learning_rate": 1.2972758433522554e-05,
"loss": 0.050227463245391846,
"step": 5080
},
{
"epoch": 4.2275747508305646,
"grad_norm": 0.22498486936092377,
"learning_rate": 1.2947062090102707e-05,
"loss": 0.013359680771827698,
"step": 5090
},
{
"epoch": 4.235880398671096,
"grad_norm": 0.6791846752166748,
"learning_rate": 1.2921344418599593e-05,
"loss": 0.013483336567878723,
"step": 5100
},
{
"epoch": 4.2441860465116275,
"grad_norm": 0.007354901172220707,
"learning_rate": 1.289560560513371e-05,
"loss": 0.024316991865634918,
"step": 5110
},
{
"epoch": 4.25249169435216,
"grad_norm": 1.3346052169799805,
"learning_rate": 1.2869845835978547e-05,
"loss": 0.017172685265541075,
"step": 5120
},
{
"epoch": 4.260797342192691,
"grad_norm": 7.70705509185791,
"learning_rate": 1.2844065297559263e-05,
"loss": 0.02509717345237732,
"step": 5130
},
{
"epoch": 4.269102990033223,
"grad_norm": 0.002678833669051528,
"learning_rate": 1.2818264176451315e-05,
"loss": 0.01910307854413986,
"step": 5140
},
{
"epoch": 4.277408637873754,
"grad_norm": 0.08240567147731781,
"learning_rate": 1.2792442659379121e-05,
"loss": 0.020165109634399415,
"step": 5150
},
{
"epoch": 4.285714285714286,
"grad_norm": 0.03600895777344704,
"learning_rate": 1.2766600933214715e-05,
"loss": 0.008698666095733642,
"step": 5160
},
{
"epoch": 4.294019933554817,
"grad_norm": 0.8765185475349426,
"learning_rate": 1.2740739184976376e-05,
"loss": 0.025190740823745728,
"step": 5170
},
{
"epoch": 4.3023255813953485,
"grad_norm": 2.9410600662231445,
"learning_rate": 1.2714857601827282e-05,
"loss": 0.031428667902946475,
"step": 5180
},
{
"epoch": 4.310631229235881,
"grad_norm": 4.09833288192749,
"learning_rate": 1.2688956371074166e-05,
"loss": 0.005366896092891693,
"step": 5190
},
{
"epoch": 4.318936877076412,
"grad_norm": 1.1638123989105225,
"learning_rate": 1.2663035680165945e-05,
"loss": 0.023147153854370116,
"step": 5200
},
{
"epoch": 4.327242524916944,
"grad_norm": 1.6218690872192383,
"learning_rate": 1.263709571669238e-05,
"loss": 0.017925117909908295,
"step": 5210
},
{
"epoch": 4.335548172757475,
"grad_norm": 2.0784993171691895,
"learning_rate": 1.2611136668382694e-05,
"loss": 0.009900534152984619,
"step": 5220
},
{
"epoch": 4.343853820598007,
"grad_norm": 0.046291958540678024,
"learning_rate": 1.2585158723104246e-05,
"loss": 0.04691925644874573,
"step": 5230
},
{
"epoch": 4.352159468438538,
"grad_norm": 4.04836893081665,
"learning_rate": 1.2559162068861132e-05,
"loss": 0.014167265594005584,
"step": 5240
},
{
"epoch": 4.3604651162790695,
"grad_norm": 0.0019111237488687038,
"learning_rate": 1.2533146893792867e-05,
"loss": 0.01547088474035263,
"step": 5250
},
{
"epoch": 4.368770764119601,
"grad_norm": 0.005684494040906429,
"learning_rate": 1.2507113386172991e-05,
"loss": 0.009821711480617524,
"step": 5260
},
{
"epoch": 4.377076411960133,
"grad_norm": 0.3991302251815796,
"learning_rate": 1.2481061734407725e-05,
"loss": 0.011389210075139999,
"step": 5270
},
{
"epoch": 4.385382059800665,
"grad_norm": 0.4083890914916992,
"learning_rate": 1.2454992127034588e-05,
"loss": 0.03720539212226868,
"step": 5280
},
{
"epoch": 4.393687707641196,
"grad_norm": 0.0017885260749608278,
"learning_rate": 1.2428904752721056e-05,
"loss": 0.04555285573005676,
"step": 5290
},
{
"epoch": 4.401993355481728,
"grad_norm": 0.6126091480255127,
"learning_rate": 1.2402799800263183e-05,
"loss": 0.04678228497505188,
"step": 5300
},
{
"epoch": 4.410299003322259,
"grad_norm": 0.48896387219429016,
"learning_rate": 1.2376677458584239e-05,
"loss": 0.009043212234973907,
"step": 5310
},
{
"epoch": 4.4186046511627906,
"grad_norm": 2.2833175659179688,
"learning_rate": 1.2350537916733335e-05,
"loss": 0.026138657331466676,
"step": 5320
},
{
"epoch": 4.426910299003322,
"grad_norm": 0.6726020574569702,
"learning_rate": 1.2324381363884061e-05,
"loss": 0.029479256272315978,
"step": 5330
},
{
"epoch": 4.4352159468438535,
"grad_norm": 0.23614397644996643,
"learning_rate": 1.2298207989333128e-05,
"loss": 0.02873944044113159,
"step": 5340
},
{
"epoch": 4.443521594684386,
"grad_norm": 0.7234976291656494,
"learning_rate": 1.227201798249897e-05,
"loss": 0.01281222254037857,
"step": 5350
},
{
"epoch": 4.451827242524917,
"grad_norm": 0.08033420890569687,
"learning_rate": 1.2245811532920408e-05,
"loss": 0.040075987577438354,
"step": 5360
},
{
"epoch": 4.460132890365449,
"grad_norm": 1.5413765907287598,
"learning_rate": 1.2219588830255247e-05,
"loss": 0.011501920223236085,
"step": 5370
},
{
"epoch": 4.46843853820598,
"grad_norm": 1.4198194742202759,
"learning_rate": 1.2193350064278917e-05,
"loss": 0.062130576372146605,
"step": 5380
},
{
"epoch": 4.476744186046512,
"grad_norm": 1.3127896785736084,
"learning_rate": 1.2167095424883106e-05,
"loss": 0.016672512888908385,
"step": 5390
},
{
"epoch": 4.485049833887043,
"grad_norm": 0.0029651224613189697,
"learning_rate": 1.2140825102074373e-05,
"loss": 0.017946107685565947,
"step": 5400
},
{
"epoch": 4.4933554817275745,
"grad_norm": 0.005830972455441952,
"learning_rate": 1.2114539285972786e-05,
"loss": 0.003538551554083824,
"step": 5410
},
{
"epoch": 4.501661129568106,
"grad_norm": 6.904330730438232,
"learning_rate": 1.2088238166810533e-05,
"loss": 0.01914323717355728,
"step": 5420
},
{
"epoch": 4.509966777408637,
"grad_norm": 0.0020963172428309917,
"learning_rate": 1.2061921934930552e-05,
"loss": 0.010284826159477234,
"step": 5430
},
{
"epoch": 4.51827242524917,
"grad_norm": 4.655736923217773,
"learning_rate": 1.2035590780785153e-05,
"loss": 0.061952787637710574,
"step": 5440
},
{
"epoch": 4.526578073089701,
"grad_norm": 0.3065240979194641,
"learning_rate": 1.2009244894934641e-05,
"loss": 0.009478168934583664,
"step": 5450
},
{
"epoch": 4.534883720930233,
"grad_norm": 6.8561859130859375,
"learning_rate": 1.1982884468045937e-05,
"loss": 0.009291090071201324,
"step": 5460
},
{
"epoch": 4.543189368770764,
"grad_norm": 0.23533843457698822,
"learning_rate": 1.195650969089119e-05,
"loss": 0.02321425825357437,
"step": 5470
},
{
"epoch": 4.5514950166112955,
"grad_norm": 0.7539715766906738,
"learning_rate": 1.1930120754346409e-05,
"loss": 0.017818546295166014,
"step": 5480
},
{
"epoch": 4.559800664451827,
"grad_norm": 2.451885461807251,
"learning_rate": 1.1903717849390071e-05,
"loss": 0.013781924545764924,
"step": 5490
},
{
"epoch": 4.568106312292358,
"grad_norm": 2.3447372913360596,
"learning_rate": 1.1877301167101747e-05,
"loss": 0.011286266148090363,
"step": 5500
},
{
"epoch": 4.568106312292358,
"eval_brier": 0.966,
"eval_c@1": 0.96,
"eval_f05u": 0.97,
"eval_f1": 0.962,
"eval_loss": 0.13212688267230988,
"eval_mean": 0.97,
"eval_roc-auc": 0.992,
"eval_runtime": 69.2775,
"eval_samples_per_second": 265.931,
"eval_steps_per_second": 4.157,
"step": 5500
},
{
"epoch": 4.576411960132891,
"grad_norm": 2.822310447692871,
"learning_rate": 1.1850870898660712e-05,
"loss": 0.02732907235622406,
"step": 5510
},
{
"epoch": 4.584717607973422,
"grad_norm": 0.24702440202236176,
"learning_rate": 1.1824427235344571e-05,
"loss": 0.009680543839931489,
"step": 5520
},
{
"epoch": 4.593023255813954,
"grad_norm": 0.11351430416107178,
"learning_rate": 1.179797036852786e-05,
"loss": 0.024537718296051024,
"step": 5530
},
{
"epoch": 4.601328903654485,
"grad_norm": 0.1742720752954483,
"learning_rate": 1.1771500489680676e-05,
"loss": 0.014485788345336915,
"step": 5540
},
{
"epoch": 4.6096345514950166,
"grad_norm": 1.921797275543213,
"learning_rate": 1.1745017790367282e-05,
"loss": 0.037458720803260806,
"step": 5550
},
{
"epoch": 4.617940199335548,
"grad_norm": 0.06870638579130173,
"learning_rate": 1.1718522462244726e-05,
"loss": 0.00331995002925396,
"step": 5560
},
{
"epoch": 4.6262458471760795,
"grad_norm": 0.5255098342895508,
"learning_rate": 1.1692014697061448e-05,
"loss": 0.013584165275096894,
"step": 5570
},
{
"epoch": 4.634551495016611,
"grad_norm": 1.848103642463684,
"learning_rate": 1.16654946866559e-05,
"loss": 0.0253132164478302,
"step": 5580
},
{
"epoch": 4.642857142857143,
"grad_norm": 1.2474435567855835,
"learning_rate": 1.1638962622955152e-05,
"loss": 0.0319608062505722,
"step": 5590
},
{
"epoch": 4.651162790697675,
"grad_norm": 0.04684052988886833,
"learning_rate": 1.1612418697973497e-05,
"loss": 0.023048269748687743,
"step": 5600
},
{
"epoch": 4.659468438538206,
"grad_norm": 0.013490485958755016,
"learning_rate": 1.1585863103811083e-05,
"loss": 0.022633382678031923,
"step": 5610
},
{
"epoch": 4.667774086378738,
"grad_norm": 0.0029091646429151297,
"learning_rate": 1.1559296032652502e-05,
"loss": 0.010863151401281357,
"step": 5620
},
{
"epoch": 4.676079734219269,
"grad_norm": 5.4041242599487305,
"learning_rate": 1.1532717676765402e-05,
"loss": 0.018604715168476105,
"step": 5630
},
{
"epoch": 4.6843853820598005,
"grad_norm": 0.03612590953707695,
"learning_rate": 1.1506128228499099e-05,
"loss": 0.0155234694480896,
"step": 5640
},
{
"epoch": 4.692691029900332,
"grad_norm": 2.8312246799468994,
"learning_rate": 1.1479527880283193e-05,
"loss": 0.025063151121139528,
"step": 5650
},
{
"epoch": 4.700996677740864,
"grad_norm": 0.003810290480032563,
"learning_rate": 1.1452916824626163e-05,
"loss": 0.007783669233322144,
"step": 5660
},
{
"epoch": 4.709302325581396,
"grad_norm": 0.29940053820610046,
"learning_rate": 1.142629525411398e-05,
"loss": 0.01714521199464798,
"step": 5670
},
{
"epoch": 4.717607973421927,
"grad_norm": 0.03696191683411598,
"learning_rate": 1.1399663361408707e-05,
"loss": 0.018399737775325775,
"step": 5680
},
{
"epoch": 4.725913621262459,
"grad_norm": 0.12434878200292587,
"learning_rate": 1.1373021339247114e-05,
"loss": 0.006360460072755814,
"step": 5690
},
{
"epoch": 4.73421926910299,
"grad_norm": 0.4497016668319702,
"learning_rate": 1.1346369380439279e-05,
"loss": 0.021831679344177245,
"step": 5700
},
{
"epoch": 4.7425249169435215,
"grad_norm": 2.9221436977386475,
"learning_rate": 1.1319707677867193e-05,
"loss": 0.015007516741752625,
"step": 5710
},
{
"epoch": 4.750830564784053,
"grad_norm": 2.3445451259613037,
"learning_rate": 1.1293036424483355e-05,
"loss": 0.012582153081893921,
"step": 5720
},
{
"epoch": 4.759136212624584,
"grad_norm": 1.3311432600021362,
"learning_rate": 1.1266355813309396e-05,
"loss": 0.024350470304489134,
"step": 5730
},
{
"epoch": 4.767441860465116,
"grad_norm": 1.7179744243621826,
"learning_rate": 1.1239666037434657e-05,
"loss": 0.022610074281692503,
"step": 5740
},
{
"epoch": 4.775747508305648,
"grad_norm": 4.0053391456604,
"learning_rate": 1.1212967290014814e-05,
"loss": 0.008827207982540131,
"step": 5750
},
{
"epoch": 4.78405315614618,
"grad_norm": 0.004406826104968786,
"learning_rate": 1.118625976427047e-05,
"loss": 0.019537603855133055,
"step": 5760
},
{
"epoch": 4.792358803986711,
"grad_norm": 1.1199983358383179,
"learning_rate": 1.1159543653485754e-05,
"loss": 0.0031148988753557207,
"step": 5770
},
{
"epoch": 4.8006644518272426,
"grad_norm": 1.8627395629882812,
"learning_rate": 1.1132819151006926e-05,
"loss": 0.03158606886863709,
"step": 5780
},
{
"epoch": 4.808970099667774,
"grad_norm": 0.04168757423758507,
"learning_rate": 1.1106086450240978e-05,
"loss": 0.019741442799568177,
"step": 5790
},
{
"epoch": 4.8172757475083055,
"grad_norm": 0.5385501980781555,
"learning_rate": 1.1079345744654235e-05,
"loss": 0.015669572353363036,
"step": 5800
},
{
"epoch": 4.825581395348837,
"grad_norm": 0.06651749461889267,
"learning_rate": 1.105259722777095e-05,
"loss": 0.02334149330854416,
"step": 5810
},
{
"epoch": 4.833887043189369,
"grad_norm": 0.24134045839309692,
"learning_rate": 1.1025841093171909e-05,
"loss": 0.006554661691188813,
"step": 5820
},
{
"epoch": 4.842192691029901,
"grad_norm": 1.2022056579589844,
"learning_rate": 1.0999077534493033e-05,
"loss": 0.02444618344306946,
"step": 5830
},
{
"epoch": 4.850498338870432,
"grad_norm": 0.11877620220184326,
"learning_rate": 1.0972306745423957e-05,
"loss": 0.028384533524513245,
"step": 5840
},
{
"epoch": 4.858803986710964,
"grad_norm": 0.27428197860717773,
"learning_rate": 1.0945528919706658e-05,
"loss": 0.02846103012561798,
"step": 5850
},
{
"epoch": 4.867109634551495,
"grad_norm": 0.15349681675434113,
"learning_rate": 1.0918744251134028e-05,
"loss": 0.011068707704544068,
"step": 5860
},
{
"epoch": 4.8754152823920265,
"grad_norm": 0.17310120165348053,
"learning_rate": 1.0891952933548493e-05,
"loss": 0.015930719673633575,
"step": 5870
},
{
"epoch": 4.883720930232558,
"grad_norm": 14.155986785888672,
"learning_rate": 1.0865155160840582e-05,
"loss": 0.05431943535804749,
"step": 5880
},
{
"epoch": 4.892026578073089,
"grad_norm": 8.113669395446777,
"learning_rate": 1.083835112694755e-05,
"loss": 0.05063359141349792,
"step": 5890
},
{
"epoch": 4.900332225913621,
"grad_norm": 0.07618654519319534,
"learning_rate": 1.0811541025851966e-05,
"loss": 0.018904516100883485,
"step": 5900
},
{
"epoch": 4.908637873754153,
"grad_norm": 0.0057159713469445705,
"learning_rate": 1.0784725051580302e-05,
"loss": 0.005965299531817436,
"step": 5910
},
{
"epoch": 4.916943521594685,
"grad_norm": 0.22818350791931152,
"learning_rate": 1.0757903398201541e-05,
"loss": 0.01754939407110214,
"step": 5920
},
{
"epoch": 4.925249169435216,
"grad_norm": 19.128080368041992,
"learning_rate": 1.073107625982576e-05,
"loss": 0.022441479563713073,
"step": 5930
},
{
"epoch": 4.9335548172757475,
"grad_norm": 0.07472891360521317,
"learning_rate": 1.070424383060273e-05,
"loss": 0.03778038620948791,
"step": 5940
},
{
"epoch": 4.941860465116279,
"grad_norm": 1.5706290006637573,
"learning_rate": 1.0677406304720523e-05,
"loss": 0.0034949634224176406,
"step": 5950
},
{
"epoch": 4.95016611295681,
"grad_norm": 0.04871722683310509,
"learning_rate": 1.0650563876404086e-05,
"loss": 0.01854807436466217,
"step": 5960
},
{
"epoch": 4.958471760797342,
"grad_norm": 0.5312822461128235,
"learning_rate": 1.0623716739913847e-05,
"loss": 0.001561350654810667,
"step": 5970
},
{
"epoch": 4.966777408637874,
"grad_norm": 0.07376973330974579,
"learning_rate": 1.0596865089544311e-05,
"loss": 0.0032320842146873474,
"step": 5980
},
{
"epoch": 4.975083056478406,
"grad_norm": 0.0012221125653013587,
"learning_rate": 1.0570009119622641e-05,
"loss": 0.014906331896781921,
"step": 5990
},
{
"epoch": 4.983388704318937,
"grad_norm": 0.09429141134023666,
"learning_rate": 1.0543149024507273e-05,
"loss": 0.007438968122005463,
"step": 6000
},
{
"epoch": 4.983388704318937,
"eval_brier": 0.962,
"eval_c@1": 0.956,
"eval_f05u": 0.971,
"eval_f1": 0.958,
"eval_loss": 0.1529388278722763,
"eval_mean": 0.968,
"eval_roc-auc": 0.99,
"eval_runtime": 69.3253,
"eval_samples_per_second": 265.747,
"eval_steps_per_second": 4.154,
"step": 6000
},
{
"epoch": 4.9916943521594686,
"grad_norm": 0.003674514591693878,
"learning_rate": 1.0516284998586497e-05,
"loss": 0.013652998208999633,
"step": 6010
},
{
"epoch": 5.0,
"grad_norm": 0.001616625115275383,
"learning_rate": 1.0489417236277035e-05,
"loss": 0.02531648278236389,
"step": 6020
},
{
"epoch": 5.0083056478405314,
"grad_norm": 0.5862334966659546,
"learning_rate": 1.0462545932022664e-05,
"loss": 0.015471625328063964,
"step": 6030
},
{
"epoch": 5.016611295681063,
"grad_norm": 0.0656253844499588,
"learning_rate": 1.0435671280292792e-05,
"loss": 0.023716352880001068,
"step": 6040
},
{
"epoch": 5.024916943521594,
"grad_norm": 0.2014501690864563,
"learning_rate": 1.040879347558105e-05,
"loss": 0.04303444623947143,
"step": 6050
},
{
"epoch": 5.033222591362127,
"grad_norm": 0.12357684224843979,
"learning_rate": 1.0381912712403888e-05,
"loss": 0.009110794961452484,
"step": 6060
},
{
"epoch": 5.041528239202658,
"grad_norm": 1.4266451597213745,
"learning_rate": 1.0355029185299166e-05,
"loss": 0.016558359563350677,
"step": 6070
},
{
"epoch": 5.04983388704319,
"grad_norm": 2.3012773990631104,
"learning_rate": 1.032814308882475e-05,
"loss": 0.04482868313789368,
"step": 6080
},
{
"epoch": 5.058139534883721,
"grad_norm": 0.33360257744789124,
"learning_rate": 1.0301254617557095e-05,
"loss": 0.01765061020851135,
"step": 6090
},
{
"epoch": 5.0664451827242525,
"grad_norm": 0.269838809967041,
"learning_rate": 1.0274363966089848e-05,
"loss": 0.03180038332939148,
"step": 6100
},
{
"epoch": 5.074750830564784,
"grad_norm": 0.016482122242450714,
"learning_rate": 1.0247471329032434e-05,
"loss": 0.0039018712937831878,
"step": 6110
},
{
"epoch": 5.083056478405315,
"grad_norm": 0.007582697086036205,
"learning_rate": 1.0220576901008642e-05,
"loss": 0.012031578272581101,
"step": 6120
},
{
"epoch": 5.091362126245847,
"grad_norm": 0.30735695362091064,
"learning_rate": 1.0193680876655235e-05,
"loss": 0.015549731254577637,
"step": 6130
},
{
"epoch": 5.099667774086379,
"grad_norm": 2.096459150314331,
"learning_rate": 1.0166783450620504e-05,
"loss": 0.011670152842998504,
"step": 6140
},
{
"epoch": 5.107973421926911,
"grad_norm": 0.1962430477142334,
"learning_rate": 1.0139884817562915e-05,
"loss": 0.034085988998413086,
"step": 6150
},
{
"epoch": 5.116279069767442,
"grad_norm": 6.137411594390869,
"learning_rate": 1.0112985172149652e-05,
"loss": 0.028455090522766114,
"step": 6160
},
{
"epoch": 5.1245847176079735,
"grad_norm": 0.003732080338522792,
"learning_rate": 1.008608470905522e-05,
"loss": 0.01228947788476944,
"step": 6170
},
{
"epoch": 5.132890365448505,
"grad_norm": 0.27746227383613586,
"learning_rate": 1.0059183622960058e-05,
"loss": 0.04811334013938904,
"step": 6180
},
{
"epoch": 5.141196013289036,
"grad_norm": 12.70612621307373,
"learning_rate": 1.0032282108549101e-05,
"loss": 0.06266883611679078,
"step": 6190
},
{
"epoch": 5.149501661129568,
"grad_norm": 10.436833381652832,
"learning_rate": 1.0005380360510393e-05,
"loss": 0.029998233914375304,
"step": 6200
},
{
"epoch": 5.157807308970099,
"grad_norm": 0.0021893042139708996,
"learning_rate": 9.978478573533658e-06,
"loss": 0.04633872807025909,
"step": 6210
},
{
"epoch": 5.166112956810632,
"grad_norm": 0.01333579607307911,
"learning_rate": 9.951576942308906e-06,
"loss": 0.009243057668209076,
"step": 6220
},
{
"epoch": 5.174418604651163,
"grad_norm": 1.6492170095443726,
"learning_rate": 9.924675661525029e-06,
"loss": 0.014937382936477662,
"step": 6230
},
{
"epoch": 5.1827242524916945,
"grad_norm": 1.5642340183258057,
"learning_rate": 9.897774925868371e-06,
"loss": 0.01730722635984421,
"step": 6240
},
{
"epoch": 5.191029900332226,
"grad_norm": 0.16060325503349304,
"learning_rate": 9.870874930021335e-06,
"loss": 0.013254204392433166,
"step": 6250
},
{
"epoch": 5.1993355481727574,
"grad_norm": 0.0033668838441371918,
"learning_rate": 9.843975868660974e-06,
"loss": 0.005051122605800628,
"step": 6260
},
{
"epoch": 5.207641196013289,
"grad_norm": 4.608306884765625,
"learning_rate": 9.817077936457566e-06,
"loss": 0.014579175412654877,
"step": 6270
},
{
"epoch": 5.21594684385382,
"grad_norm": 0.2961117923259735,
"learning_rate": 9.790181328073231e-06,
"loss": 0.018054383993148803,
"step": 6280
},
{
"epoch": 5.224252491694352,
"grad_norm": 0.1454981416463852,
"learning_rate": 9.763286238160505e-06,
"loss": 0.0120973601937294,
"step": 6290
},
{
"epoch": 5.232558139534884,
"grad_norm": 0.6783755421638489,
"learning_rate": 9.736392861360925e-06,
"loss": 0.021133965253829955,
"step": 6300
},
{
"epoch": 5.240863787375416,
"grad_norm": 0.00108309555798769,
"learning_rate": 9.709501392303645e-06,
"loss": 0.016107510030269622,
"step": 6310
},
{
"epoch": 5.249169435215947,
"grad_norm": 2.476327419281006,
"learning_rate": 9.682612025604e-06,
"loss": 0.022406360507011412,
"step": 6320
},
{
"epoch": 5.2574750830564785,
"grad_norm": 0.9213047027587891,
"learning_rate": 9.65572495586212e-06,
"loss": 0.05495978593826294,
"step": 6330
},
{
"epoch": 5.26578073089701,
"grad_norm": 1.9122434854507446,
"learning_rate": 9.628840377661506e-06,
"loss": 0.023163509368896485,
"step": 6340
},
{
"epoch": 5.274086378737541,
"grad_norm": 2.968592643737793,
"learning_rate": 9.601958485567621e-06,
"loss": 0.012122733891010285,
"step": 6350
},
{
"epoch": 5.282392026578073,
"grad_norm": 2.8721532821655273,
"learning_rate": 9.57507947412651e-06,
"loss": 0.02536485493183136,
"step": 6360
},
{
"epoch": 5.290697674418604,
"grad_norm": 0.002760088536888361,
"learning_rate": 9.548203537863342e-06,
"loss": 0.009196484833955765,
"step": 6370
},
{
"epoch": 5.299003322259137,
"grad_norm": 10.133779525756836,
"learning_rate": 9.521330871281055e-06,
"loss": 0.029024860262870787,
"step": 6380
},
{
"epoch": 5.307308970099668,
"grad_norm": 0.06383432447910309,
"learning_rate": 9.494461668858911e-06,
"loss": 0.045620501041412354,
"step": 6390
},
{
"epoch": 5.3156146179401995,
"grad_norm": 2.252612590789795,
"learning_rate": 9.467596125051104e-06,
"loss": 0.020950838923454285,
"step": 6400
},
{
"epoch": 5.323920265780731,
"grad_norm": 0.004589166957885027,
"learning_rate": 9.440734434285353e-06,
"loss": 0.009040661901235581,
"step": 6410
},
{
"epoch": 5.332225913621262,
"grad_norm": 0.0076341466046869755,
"learning_rate": 9.413876790961491e-06,
"loss": 0.013610967993736267,
"step": 6420
},
{
"epoch": 5.340531561461794,
"grad_norm": 4.5356526374816895,
"learning_rate": 9.387023389450058e-06,
"loss": 0.02646231949329376,
"step": 6430
},
{
"epoch": 5.348837209302325,
"grad_norm": 0.0021297375205904245,
"learning_rate": 9.360174424090902e-06,
"loss": 0.018574891984462737,
"step": 6440
},
{
"epoch": 5.357142857142857,
"grad_norm": 3.6231789588928223,
"learning_rate": 9.33333008919175e-06,
"loss": 0.011462628096342086,
"step": 6450
},
{
"epoch": 5.365448504983389,
"grad_norm": 3.046170711517334,
"learning_rate": 9.306490579026838e-06,
"loss": 0.013014721870422363,
"step": 6460
},
{
"epoch": 5.3737541528239205,
"grad_norm": 0.7817786335945129,
"learning_rate": 9.279656087835475e-06,
"loss": 0.004163151234388351,
"step": 6470
},
{
"epoch": 5.382059800664452,
"grad_norm": 0.27560335397720337,
"learning_rate": 9.252826809820645e-06,
"loss": 0.008296001702547073,
"step": 6480
},
{
"epoch": 5.3903654485049834,
"grad_norm": 2.140653610229492,
"learning_rate": 9.22600293914761e-06,
"loss": 0.031283590197563174,
"step": 6490
},
{
"epoch": 5.398671096345515,
"grad_norm": 0.4135418236255646,
"learning_rate": 9.199184669942498e-06,
"loss": 0.05154951810836792,
"step": 6500
},
{
"epoch": 5.398671096345515,
"eval_brier": 0.963,
"eval_c@1": 0.958,
"eval_f05u": 0.973,
"eval_f1": 0.96,
"eval_loss": 0.15943123400211334,
"eval_mean": 0.969,
"eval_roc-auc": 0.99,
"eval_runtime": 69.4261,
"eval_samples_per_second": 265.361,
"eval_steps_per_second": 4.148,
"step": 6500
},
{
"epoch": 5.406976744186046,
"grad_norm": 0.002762638730928302,
"learning_rate": 9.172372196290892e-06,
"loss": 0.001893816702067852,
"step": 6510
},
{
"epoch": 5.415282392026578,
"grad_norm": 0.0029988118913024664,
"learning_rate": 9.145565712236444e-06,
"loss": 0.000711390282958746,
"step": 6520
},
{
"epoch": 5.423588039867109,
"grad_norm": 0.1753683090209961,
"learning_rate": 9.118765411779444e-06,
"loss": 0.03069241940975189,
"step": 6530
},
{
"epoch": 5.431893687707642,
"grad_norm": 7.485196113586426,
"learning_rate": 9.091971488875448e-06,
"loss": 0.01752369850873947,
"step": 6540
},
{
"epoch": 5.440199335548173,
"grad_norm": 0.7450178861618042,
"learning_rate": 9.065184137433837e-06,
"loss": 0.0281473308801651,
"step": 6550
},
{
"epoch": 5.4485049833887045,
"grad_norm": 0.3487071394920349,
"learning_rate": 9.038403551316455e-06,
"loss": 0.018239806592464446,
"step": 6560
},
{
"epoch": 5.456810631229236,
"grad_norm": 1.172838807106018,
"learning_rate": 9.011629924336172e-06,
"loss": 0.003979488462209702,
"step": 6570
},
{
"epoch": 5.465116279069767,
"grad_norm": 0.5687670707702637,
"learning_rate": 8.984863450255494e-06,
"loss": 0.013243064284324646,
"step": 6580
},
{
"epoch": 5.473421926910299,
"grad_norm": 2.026305675506592,
"learning_rate": 8.95810432278517e-06,
"loss": 0.012412568181753158,
"step": 6590
},
{
"epoch": 5.48172757475083,
"grad_norm": 0.1178358718752861,
"learning_rate": 8.931352735582767e-06,
"loss": 0.013863597810268403,
"step": 6600
},
{
"epoch": 5.490033222591362,
"grad_norm": 0.001700263237580657,
"learning_rate": 8.9046088822513e-06,
"loss": 0.009028357267379761,
"step": 6610
},
{
"epoch": 5.498338870431894,
"grad_norm": 0.0027973002288490534,
"learning_rate": 8.877872956337803e-06,
"loss": 0.011504148691892624,
"step": 6620
},
{
"epoch": 5.5066445182724255,
"grad_norm": 1.4627960920333862,
"learning_rate": 8.851145151331932e-06,
"loss": 0.02455365061759949,
"step": 6630
},
{
"epoch": 5.514950166112957,
"grad_norm": 0.0016517736949026585,
"learning_rate": 8.824425660664589e-06,
"loss": 0.013532018661499024,
"step": 6640
},
{
"epoch": 5.523255813953488,
"grad_norm": 0.0016886944649741054,
"learning_rate": 8.79771467770649e-06,
"loss": 0.020568014681339265,
"step": 6650
},
{
"epoch": 5.53156146179402,
"grad_norm": 3.9619431495666504,
"learning_rate": 8.771012395766784e-06,
"loss": 0.02581743597984314,
"step": 6660
},
{
"epoch": 5.539867109634551,
"grad_norm": 0.0056344387121498585,
"learning_rate": 8.744319008091657e-06,
"loss": 0.014075009524822235,
"step": 6670
},
{
"epoch": 5.548172757475083,
"grad_norm": 3.7406468391418457,
"learning_rate": 8.717634707862916e-06,
"loss": 0.015134890377521516,
"step": 6680
},
{
"epoch": 5.556478405315614,
"grad_norm": 0.11011675745248795,
"learning_rate": 8.690959688196614e-06,
"loss": 0.038392487168312076,
"step": 6690
},
{
"epoch": 5.5647840531561465,
"grad_norm": 0.14210830628871918,
"learning_rate": 8.664294142141623e-06,
"loss": 0.014288631081581116,
"step": 6700
},
{
"epoch": 5.573089700996678,
"grad_norm": 0.011836647056043148,
"learning_rate": 8.637638262678265e-06,
"loss": 0.03645042777061462,
"step": 6710
},
{
"epoch": 5.5813953488372094,
"grad_norm": 0.007830922491848469,
"learning_rate": 8.610992242716914e-06,
"loss": 0.00012227758998051286,
"step": 6720
},
{
"epoch": 5.589700996677741,
"grad_norm": 0.012702585197985172,
"learning_rate": 8.584356275096564e-06,
"loss": 0.020489972829818726,
"step": 6730
},
{
"epoch": 5.598006644518272,
"grad_norm": 0.3409964144229889,
"learning_rate": 8.557730552583481e-06,
"loss": 0.021626809239387514,
"step": 6740
},
{
"epoch": 5.606312292358804,
"grad_norm": 0.6491411924362183,
"learning_rate": 8.53111526786978e-06,
"loss": 0.027367493510246275,
"step": 6750
},
{
"epoch": 5.614617940199335,
"grad_norm": 9.169612884521484,
"learning_rate": 8.504510613572036e-06,
"loss": 0.030298405885696413,
"step": 6760
},
{
"epoch": 5.622923588039868,
"grad_norm": 2.571241855621338,
"learning_rate": 8.477916782229893e-06,
"loss": 0.0046318769454956055,
"step": 6770
},
{
"epoch": 5.631229235880399,
"grad_norm": 0.16384539008140564,
"learning_rate": 8.451333966304666e-06,
"loss": 0.032301408052444455,
"step": 6780
},
{
"epoch": 5.6395348837209305,
"grad_norm": 0.13168157637119293,
"learning_rate": 8.424762358177955e-06,
"loss": 0.015387143194675445,
"step": 6790
},
{
"epoch": 5.647840531561462,
"grad_norm": 1.2372331619262695,
"learning_rate": 8.398202150150238e-06,
"loss": 0.01876421719789505,
"step": 6800
},
{
"epoch": 5.656146179401993,
"grad_norm": 0.002671166555956006,
"learning_rate": 8.371653534439505e-06,
"loss": 0.020773571729660035,
"step": 6810
},
{
"epoch": 5.664451827242525,
"grad_norm": 0.0014251646352931857,
"learning_rate": 8.34511670317984e-06,
"loss": 0.005695940181612969,
"step": 6820
},
{
"epoch": 5.672757475083056,
"grad_norm": 0.7163535356521606,
"learning_rate": 8.318591848420046e-06,
"loss": 0.0328060507774353,
"step": 6830
},
{
"epoch": 5.681063122923588,
"grad_norm": 0.22294257581233978,
"learning_rate": 8.292079162122252e-06,
"loss": 0.003689458966255188,
"step": 6840
},
{
"epoch": 5.689368770764119,
"grad_norm": 3.362990140914917,
"learning_rate": 8.265578836160523e-06,
"loss": 0.029373672604560853,
"step": 6850
},
{
"epoch": 5.6976744186046515,
"grad_norm": 1.8103702068328857,
"learning_rate": 8.239091062319469e-06,
"loss": 0.025295084714889525,
"step": 6860
},
{
"epoch": 5.705980066445183,
"grad_norm": 0.0027610426768660545,
"learning_rate": 8.212616032292866e-06,
"loss": 0.007607991248369217,
"step": 6870
},
{
"epoch": 5.714285714285714,
"grad_norm": 4.924838066101074,
"learning_rate": 8.186153937682248e-06,
"loss": 0.030681931972503663,
"step": 6880
},
{
"epoch": 5.722591362126246,
"grad_norm": 13.518671035766602,
"learning_rate": 8.159704969995553e-06,
"loss": 0.04611906409263611,
"step": 6890
},
{
"epoch": 5.730897009966777,
"grad_norm": 0.04365919157862663,
"learning_rate": 8.133269320645702e-06,
"loss": 0.017498983442783354,
"step": 6900
},
{
"epoch": 5.739202657807309,
"grad_norm": 0.2871593236923218,
"learning_rate": 8.106847180949243e-06,
"loss": 0.007226487249135971,
"step": 6910
},
{
"epoch": 5.74750830564784,
"grad_norm": 0.005341998301446438,
"learning_rate": 8.080438742124949e-06,
"loss": 0.004703724756836891,
"step": 6920
},
{
"epoch": 5.7558139534883725,
"grad_norm": 0.6335068941116333,
"learning_rate": 8.054044195292435e-06,
"loss": 0.010668383538722992,
"step": 6930
},
{
"epoch": 5.764119601328904,
"grad_norm": 0.026266956701874733,
"learning_rate": 8.027663731470787e-06,
"loss": 0.01587436646223068,
"step": 6940
},
{
"epoch": 5.7724252491694354,
"grad_norm": 0.2619726061820984,
"learning_rate": 8.001297541577155e-06,
"loss": 0.010323958098888397,
"step": 6950
},
{
"epoch": 5.780730897009967,
"grad_norm": 0.0020163620356470346,
"learning_rate": 7.974945816425412e-06,
"loss": 0.0024325501173734663,
"step": 6960
},
{
"epoch": 5.789036544850498,
"grad_norm": 0.003413138445466757,
"learning_rate": 7.948608746724735e-06,
"loss": 9.869699133560061e-05,
"step": 6970
},
{
"epoch": 5.79734219269103,
"grad_norm": 9.29476261138916,
"learning_rate": 7.92228652307823e-06,
"loss": 0.057475173473358156,
"step": 6980
},
{
"epoch": 5.805647840531561,
"grad_norm": 0.22015738487243652,
"learning_rate": 7.895979335981578e-06,
"loss": 0.0277383953332901,
"step": 6990
},
{
"epoch": 5.813953488372093,
"grad_norm": 0.6662735342979431,
"learning_rate": 7.869687375821627e-06,
"loss": 0.005917861685156822,
"step": 7000
},
{
"epoch": 5.813953488372093,
"eval_brier": 0.964,
"eval_c@1": 0.959,
"eval_f05u": 0.973,
"eval_f1": 0.961,
"eval_loss": 0.15328434109687805,
"eval_mean": 0.97,
"eval_roc-auc": 0.991,
"eval_runtime": 69.2438,
"eval_samples_per_second": 266.06,
"eval_steps_per_second": 4.159,
"step": 7000
},
{
"epoch": 5.822259136212624,
"grad_norm": 0.0058500999584794044,
"learning_rate": 7.843410832875035e-06,
"loss": 0.0010274063795804977,
"step": 7010
},
{
"epoch": 5.8305647840531565,
"grad_norm": 0.0022387949284166098,
"learning_rate": 7.817149897306881e-06,
"loss": 0.008160777390003204,
"step": 7020
},
{
"epoch": 5.838870431893688,
"grad_norm": 0.18496069312095642,
"learning_rate": 7.790904759169292e-06,
"loss": 0.02355058640241623,
"step": 7030
},
{
"epoch": 5.847176079734219,
"grad_norm": 0.18093812465667725,
"learning_rate": 7.764675608400073e-06,
"loss": 0.03586549460887909,
"step": 7040
},
{
"epoch": 5.855481727574751,
"grad_norm": 2.4179556369781494,
"learning_rate": 7.738462634821314e-06,
"loss": 0.034015771746635434,
"step": 7050
},
{
"epoch": 5.863787375415282,
"grad_norm": 0.007284322287887335,
"learning_rate": 7.71226602813804e-06,
"loss": 0.009248374402523041,
"step": 7060
},
{
"epoch": 5.872093023255814,
"grad_norm": 0.004407364875078201,
"learning_rate": 7.686085977936837e-06,
"loss": 0.00012948120711371304,
"step": 7070
},
{
"epoch": 5.880398671096345,
"grad_norm": 0.23228105902671814,
"learning_rate": 7.659922673684446e-06,
"loss": 0.01536521315574646,
"step": 7080
},
{
"epoch": 5.8887043189368775,
"grad_norm": 6.431649684906006,
"learning_rate": 7.63377630472644e-06,
"loss": 0.02363522946834564,
"step": 7090
},
{
"epoch": 5.897009966777409,
"grad_norm": 0.16404373943805695,
"learning_rate": 7.607647060285814e-06,
"loss": 0.013303206861019134,
"step": 7100
},
{
"epoch": 5.90531561461794,
"grad_norm": 3.739145517349243,
"learning_rate": 7.5815351294616414e-06,
"loss": 0.016325636208057402,
"step": 7110
},
{
"epoch": 5.913621262458472,
"grad_norm": 0.011894511990249157,
"learning_rate": 7.555440701227695e-06,
"loss": 0.00029810983687639236,
"step": 7120
},
{
"epoch": 5.921926910299003,
"grad_norm": 0.3672759234905243,
"learning_rate": 7.52936396443107e-06,
"loss": 0.02186768054962158,
"step": 7130
},
{
"epoch": 5.930232558139535,
"grad_norm": 1.5637317895889282,
"learning_rate": 7.503305107790846e-06,
"loss": 0.04363935887813568,
"step": 7140
},
{
"epoch": 5.938538205980066,
"grad_norm": 0.6385419964790344,
"learning_rate": 7.477264319896682e-06,
"loss": 0.021274478733539583,
"step": 7150
},
{
"epoch": 5.946843853820598,
"grad_norm": 0.012678664177656174,
"learning_rate": 7.451241789207488e-06,
"loss": 0.0032838169485330583,
"step": 7160
},
{
"epoch": 5.955149501661129,
"grad_norm": 0.01598937250673771,
"learning_rate": 7.425237704050039e-06,
"loss": 0.001255387905985117,
"step": 7170
},
{
"epoch": 5.9634551495016614,
"grad_norm": 0.5728321671485901,
"learning_rate": 7.399252252617617e-06,
"loss": 0.043770810961723326,
"step": 7180
},
{
"epoch": 5.971760797342193,
"grad_norm": 1.4228415489196777,
"learning_rate": 7.373285622968656e-06,
"loss": 0.03225944936275482,
"step": 7190
},
{
"epoch": 5.980066445182724,
"grad_norm": 4.1248674392700195,
"learning_rate": 7.3473380030253685e-06,
"loss": 0.01650337278842926,
"step": 7200
},
{
"epoch": 5.988372093023256,
"grad_norm": 0.04176795482635498,
"learning_rate": 7.321409580572396e-06,
"loss": 0.005239410325884819,
"step": 7210
},
{
"epoch": 5.996677740863787,
"grad_norm": 0.0033232630230486393,
"learning_rate": 7.295500543255456e-06,
"loss": 0.0006531004793941975,
"step": 7220
},
{
"epoch": 6.004983388704319,
"grad_norm": 0.001787588931620121,
"learning_rate": 7.269611078579951e-06,
"loss": 0.0020497867837548254,
"step": 7230
},
{
"epoch": 6.01328903654485,
"grad_norm": 1.474507212638855,
"learning_rate": 7.24374137390966e-06,
"loss": 0.03556830585002899,
"step": 7240
},
{
"epoch": 6.0215946843853825,
"grad_norm": 0.12092769891023636,
"learning_rate": 7.217891616465339e-06,
"loss": 0.03157237470149994,
"step": 7250
},
{
"epoch": 6.029900332225914,
"grad_norm": 0.014265786856412888,
"learning_rate": 7.192061993323397e-06,
"loss": 0.020114152133464812,
"step": 7260
},
{
"epoch": 6.038205980066445,
"grad_norm": 0.0032013391610234976,
"learning_rate": 7.166252691414526e-06,
"loss": 0.0008751790039241314,
"step": 7270
},
{
"epoch": 6.046511627906977,
"grad_norm": 0.3105766773223877,
"learning_rate": 7.140463897522347e-06,
"loss": 0.028306761384010316,
"step": 7280
},
{
"epoch": 6.054817275747508,
"grad_norm": 1.0390666723251343,
"learning_rate": 7.114695798282072e-06,
"loss": 0.02048112452030182,
"step": 7290
},
{
"epoch": 6.06312292358804,
"grad_norm": 0.736840009689331,
"learning_rate": 7.0889485801791335e-06,
"loss": 0.010562019795179367,
"step": 7300
},
{
"epoch": 6.071428571428571,
"grad_norm": 0.002257751300930977,
"learning_rate": 7.063222429547854e-06,
"loss": 0.024713443219661714,
"step": 7310
},
{
"epoch": 6.079734219269103,
"grad_norm": 2.116347551345825,
"learning_rate": 7.037517532570092e-06,
"loss": 0.0138460174202919,
"step": 7320
},
{
"epoch": 6.088039867109635,
"grad_norm": 1.5315171480178833,
"learning_rate": 7.011834075273879e-06,
"loss": 0.017610684037208557,
"step": 7330
},
{
"epoch": 6.096345514950166,
"grad_norm": 0.8159580230712891,
"learning_rate": 6.986172243532101e-06,
"loss": 0.009701524674892426,
"step": 7340
},
{
"epoch": 6.104651162790698,
"grad_norm": 0.4105885624885559,
"learning_rate": 6.960532223061126e-06,
"loss": 0.007899902015924453,
"step": 7350
},
{
"epoch": 6.112956810631229,
"grad_norm": 0.0025956020690500736,
"learning_rate": 6.93491419941948e-06,
"loss": 0.009152834862470626,
"step": 7360
},
{
"epoch": 6.121262458471761,
"grad_norm": 0.003840164514258504,
"learning_rate": 6.909318358006498e-06,
"loss": 0.0017187679186463356,
"step": 7370
},
{
"epoch": 6.129568106312292,
"grad_norm": 2.537564277648926,
"learning_rate": 6.883744884060972e-06,
"loss": 0.020690998435020445,
"step": 7380
},
{
"epoch": 6.137873754152824,
"grad_norm": 0.0015193113358691335,
"learning_rate": 6.858193962659832e-06,
"loss": 0.016765663027763368,
"step": 7390
},
{
"epoch": 6.146179401993355,
"grad_norm": 0.11541413515806198,
"learning_rate": 6.832665778716775e-06,
"loss": 0.01714625060558319,
"step": 7400
},
{
"epoch": 6.1544850498338874,
"grad_norm": 0.005307809915393591,
"learning_rate": 6.8071605169809626e-06,
"loss": 0.005107355117797851,
"step": 7410
},
{
"epoch": 6.162790697674419,
"grad_norm": 0.003973027225583792,
"learning_rate": 6.781678362035663e-06,
"loss": 9.050400112755596e-05,
"step": 7420
},
{
"epoch": 6.17109634551495,
"grad_norm": 0.007702766451984644,
"learning_rate": 6.756219498296912e-06,
"loss": 0.018383122980594635,
"step": 7430
},
{
"epoch": 6.179401993355482,
"grad_norm": 0.7614707350730896,
"learning_rate": 6.730784110012197e-06,
"loss": 0.0024529168382287025,
"step": 7440
},
{
"epoch": 6.187707641196013,
"grad_norm": 1.42940092086792,
"learning_rate": 6.705372381259095e-06,
"loss": 0.013600359857082366,
"step": 7450
},
{
"epoch": 6.196013289036545,
"grad_norm": 0.0012264583492651582,
"learning_rate": 6.6799844959439765e-06,
"loss": 0.012189332395792007,
"step": 7460
},
{
"epoch": 6.204318936877076,
"grad_norm": 1.2350224256515503,
"learning_rate": 6.6546206378006486e-06,
"loss": 0.02500038743019104,
"step": 7470
},
{
"epoch": 6.212624584717608,
"grad_norm": 9.297430992126465,
"learning_rate": 6.629280990389018e-06,
"loss": 0.04516852200031281,
"step": 7480
},
{
"epoch": 6.22093023255814,
"grad_norm": 0.15006913244724274,
"learning_rate": 6.6039657370938025e-06,
"loss": 0.028004831075668334,
"step": 7490
},
{
"epoch": 6.229235880398671,
"grad_norm": 19.135421752929688,
"learning_rate": 6.578675061123155e-06,
"loss": 0.01740252673625946,
"step": 7500
},
{
"epoch": 6.229235880398671,
"eval_brier": 0.963,
"eval_c@1": 0.958,
"eval_f05u": 0.976,
"eval_f1": 0.96,
"eval_loss": 0.14886559545993805,
"eval_mean": 0.97,
"eval_roc-auc": 0.991,
"eval_runtime": 69.3145,
"eval_samples_per_second": 265.788,
"eval_steps_per_second": 4.155,
"step": 7500
},
{
"epoch": 6.237541528239203,
"grad_norm": 3.230306386947632,
"learning_rate": 6.553409145507375e-06,
"loss": 0.017020581662654875,
"step": 7510
},
{
"epoch": 6.245847176079734,
"grad_norm": 0.002402634359896183,
"learning_rate": 6.528168173097563e-06,
"loss": 0.018050453066825865,
"step": 7520
},
{
"epoch": 6.254152823920266,
"grad_norm": 0.21895131468772888,
"learning_rate": 6.502952326564307e-06,
"loss": 0.021868236362934113,
"step": 7530
},
{
"epoch": 6.262458471760797,
"grad_norm": 0.04882687330245972,
"learning_rate": 6.477761788396357e-06,
"loss": 0.007655378431081772,
"step": 7540
},
{
"epoch": 6.270764119601329,
"grad_norm": 1.855291724205017,
"learning_rate": 6.4525967408992986e-06,
"loss": 0.04031583070755005,
"step": 7550
},
{
"epoch": 6.27906976744186,
"grad_norm": 0.0021352756302803755,
"learning_rate": 6.42745736619425e-06,
"loss": 0.011861710250377655,
"step": 7560
},
{
"epoch": 6.287375415282392,
"grad_norm": 0.32317888736724854,
"learning_rate": 6.40234384621653e-06,
"loss": 0.0062631569802761074,
"step": 7570
},
{
"epoch": 6.295681063122924,
"grad_norm": 4.248029708862305,
"learning_rate": 6.377256362714339e-06,
"loss": 0.015436901152133942,
"step": 7580
},
{
"epoch": 6.303986710963455,
"grad_norm": 1.0710715055465698,
"learning_rate": 6.3521950972474575e-06,
"loss": 0.015983884036540986,
"step": 7590
},
{
"epoch": 6.312292358803987,
"grad_norm": 0.40718525648117065,
"learning_rate": 6.327160231185918e-06,
"loss": 0.028572919964790344,
"step": 7600
},
{
"epoch": 6.320598006644518,
"grad_norm": 0.0019481098279356956,
"learning_rate": 6.302151945708703e-06,
"loss": 0.014387263357639313,
"step": 7610
},
{
"epoch": 6.32890365448505,
"grad_norm": 0.0019651392940431833,
"learning_rate": 6.277170421802432e-06,
"loss": 0.009500369429588318,
"step": 7620
},
{
"epoch": 6.337209302325581,
"grad_norm": 1.2498829364776611,
"learning_rate": 6.252215840260043e-06,
"loss": 0.03341265022754669,
"step": 7630
},
{
"epoch": 6.3455149501661126,
"grad_norm": 10.818826675415039,
"learning_rate": 6.227288381679496e-06,
"loss": 0.046141260862350465,
"step": 7640
},
{
"epoch": 6.353820598006645,
"grad_norm": 2.388810396194458,
"learning_rate": 6.20238822646245e-06,
"loss": 0.01637554168701172,
"step": 7650
},
{
"epoch": 6.362126245847176,
"grad_norm": 0.010045533068478107,
"learning_rate": 6.17751555481298e-06,
"loss": 0.020306165516376495,
"step": 7660
},
{
"epoch": 6.370431893687708,
"grad_norm": 0.0031976604368537664,
"learning_rate": 6.1526705467362614e-06,
"loss": 0.00996808186173439,
"step": 7670
},
{
"epoch": 6.378737541528239,
"grad_norm": 8.072118759155273,
"learning_rate": 6.12785338203725e-06,
"loss": 0.07115018367767334,
"step": 7680
},
{
"epoch": 6.387043189368771,
"grad_norm": 0.3611799478530884,
"learning_rate": 6.103064240319416e-06,
"loss": 0.025847119092941285,
"step": 7690
},
{
"epoch": 6.395348837209302,
"grad_norm": 1.010109543800354,
"learning_rate": 6.0783033009834125e-06,
"loss": 0.009891898930072784,
"step": 7700
},
{
"epoch": 6.403654485049834,
"grad_norm": 0.0012372470228001475,
"learning_rate": 6.053570743225798e-06,
"loss": 0.031172019243240357,
"step": 7710
},
{
"epoch": 6.411960132890365,
"grad_norm": 1.7768527269363403,
"learning_rate": 6.028866746037734e-06,
"loss": 0.023508161306381226,
"step": 7720
},
{
"epoch": 6.420265780730897,
"grad_norm": 12.104537963867188,
"learning_rate": 6.004191488203668e-06,
"loss": 0.04133204519748688,
"step": 7730
},
{
"epoch": 6.428571428571429,
"grad_norm": 0.2450614720582962,
"learning_rate": 5.9795451483000885e-06,
"loss": 0.040730857849121095,
"step": 7740
},
{
"epoch": 6.43687707641196,
"grad_norm": 0.2536294758319855,
"learning_rate": 5.9549279046941765e-06,
"loss": 0.046179857850074765,
"step": 7750
},
{
"epoch": 6.445182724252492,
"grad_norm": 1.2255915403366089,
"learning_rate": 5.9303399355425554e-06,
"loss": 0.02125682383775711,
"step": 7760
},
{
"epoch": 6.453488372093023,
"grad_norm": 0.11651555448770523,
"learning_rate": 5.9057814187899865e-06,
"loss": 0.008241599053144455,
"step": 7770
},
{
"epoch": 6.461794019933555,
"grad_norm": 0.0017291768454015255,
"learning_rate": 5.881252532168077e-06,
"loss": 0.008094576746225357,
"step": 7780
},
{
"epoch": 6.470099667774086,
"grad_norm": 0.0015593268908560276,
"learning_rate": 5.8567534531940044e-06,
"loss": 0.019118666648864746,
"step": 7790
},
{
"epoch": 6.4784053156146175,
"grad_norm": 0.8568136692047119,
"learning_rate": 5.8322843591692225e-06,
"loss": 0.032859981060028076,
"step": 7800
},
{
"epoch": 6.48671096345515,
"grad_norm": 0.32784509658813477,
"learning_rate": 5.807845427178178e-06,
"loss": 0.005178544670343399,
"step": 7810
},
{
"epoch": 6.495016611295681,
"grad_norm": 0.002425646409392357,
"learning_rate": 5.783436834087046e-06,
"loss": 0.0038521364331245422,
"step": 7820
},
{
"epoch": 6.503322259136213,
"grad_norm": 5.8360700607299805,
"learning_rate": 5.759058756542426e-06,
"loss": 0.0313642144203186,
"step": 7830
},
{
"epoch": 6.511627906976744,
"grad_norm": 0.6223105788230896,
"learning_rate": 5.734711370970079e-06,
"loss": 0.008595378696918487,
"step": 7840
},
{
"epoch": 6.519933554817276,
"grad_norm": 0.08045630156993866,
"learning_rate": 5.710394853573637e-06,
"loss": 0.021549156308174132,
"step": 7850
},
{
"epoch": 6.528239202657807,
"grad_norm": 0.0018593433778733015,
"learning_rate": 5.686109380333356e-06,
"loss": 0.015463992953300476,
"step": 7860
},
{
"epoch": 6.5365448504983386,
"grad_norm": 0.002435112139210105,
"learning_rate": 5.661855127004805e-06,
"loss": 0.007215242832899094,
"step": 7870
},
{
"epoch": 6.544850498338871,
"grad_norm": 1.4132130146026611,
"learning_rate": 5.637632269117623e-06,
"loss": 0.015145997703075408,
"step": 7880
},
{
"epoch": 6.553156146179402,
"grad_norm": 0.021744348108768463,
"learning_rate": 5.613440981974232e-06,
"loss": 0.014030510187149048,
"step": 7890
},
{
"epoch": 6.561461794019934,
"grad_norm": 0.1795722097158432,
"learning_rate": 5.589281440648573e-06,
"loss": 0.020741964876651763,
"step": 7900
},
{
"epoch": 6.569767441860465,
"grad_norm": 1.851261854171753,
"learning_rate": 5.5651538199848545e-06,
"loss": 0.03851528465747833,
"step": 7910
},
{
"epoch": 6.578073089700997,
"grad_norm": 0.0019687809981405735,
"learning_rate": 5.541058294596258e-06,
"loss": 0.004892564564943314,
"step": 7920
},
{
"epoch": 6.586378737541528,
"grad_norm": 0.22509530186653137,
"learning_rate": 5.5169950388636906e-06,
"loss": 0.014504489302635194,
"step": 7930
},
{
"epoch": 6.59468438538206,
"grad_norm": 1.3335944414138794,
"learning_rate": 5.492964226934536e-06,
"loss": 0.029465362429618835,
"step": 7940
},
{
"epoch": 6.602990033222591,
"grad_norm": 0.15977507829666138,
"learning_rate": 5.4689660327213545e-06,
"loss": 0.033452790975570676,
"step": 7950
},
{
"epoch": 6.6112956810631225,
"grad_norm": 0.8709096908569336,
"learning_rate": 5.445000629900674e-06,
"loss": 0.015643228590488435,
"step": 7960
},
{
"epoch": 6.619601328903655,
"grad_norm": 0.0027918138075619936,
"learning_rate": 5.421068191911707e-06,
"loss": 0.0045252062380313875,
"step": 7970
},
{
"epoch": 6.627906976744186,
"grad_norm": 0.36342477798461914,
"learning_rate": 5.397168891955073e-06,
"loss": 0.0021900072693824767,
"step": 7980
},
{
"epoch": 6.636212624584718,
"grad_norm": 0.0970240980386734,
"learning_rate": 5.373302902991601e-06,
"loss": 0.021340471506118775,
"step": 7990
},
{
"epoch": 6.644518272425249,
"grad_norm": 0.1519928127527237,
"learning_rate": 5.3494703977410245e-06,
"loss": 0.023007088899612428,
"step": 8000
},
{
"epoch": 6.644518272425249,
"eval_brier": 0.966,
"eval_c@1": 0.961,
"eval_f05u": 0.973,
"eval_f1": 0.963,
"eval_loss": 0.1465037614107132,
"eval_mean": 0.971,
"eval_roc-auc": 0.991,
"eval_runtime": 69.243,
"eval_samples_per_second": 266.063,
"eval_steps_per_second": 4.159,
"step": 8000
},
{
"epoch": 6.652823920265781,
"grad_norm": 0.0025060612242668867,
"learning_rate": 5.325671548680757e-06,
"loss": 0.010099204629659653,
"step": 8010
},
{
"epoch": 6.661129568106312,
"grad_norm": 0.003020784817636013,
"learning_rate": 5.301906528044658e-06,
"loss": 0.0030381690710783006,
"step": 8020
},
{
"epoch": 6.6694352159468435,
"grad_norm": 1.8054370880126953,
"learning_rate": 5.2781755078217345e-06,
"loss": 0.03165494799613953,
"step": 8030
},
{
"epoch": 6.677740863787376,
"grad_norm": 0.0012213119771331549,
"learning_rate": 5.254478659754962e-06,
"loss": 0.021227551996707915,
"step": 8040
},
{
"epoch": 6.686046511627907,
"grad_norm": 0.042880091816186905,
"learning_rate": 5.230816155339995e-06,
"loss": 0.0283914715051651,
"step": 8050
},
{
"epoch": 6.694352159468439,
"grad_norm": 0.0010312939994037151,
"learning_rate": 5.207188165823938e-06,
"loss": 0.003574676439166069,
"step": 8060
},
{
"epoch": 6.70265780730897,
"grad_norm": 0.3460790514945984,
"learning_rate": 5.18359486220412e-06,
"loss": 0.013510248064994812,
"step": 8070
},
{
"epoch": 6.710963455149502,
"grad_norm": 0.6914936304092407,
"learning_rate": 5.160036415226839e-06,
"loss": 0.015883123874664305,
"step": 8080
},
{
"epoch": 6.719269102990033,
"grad_norm": 15.786023139953613,
"learning_rate": 5.1365129953861315e-06,
"loss": 0.01682654321193695,
"step": 8090
},
{
"epoch": 6.7275747508305646,
"grad_norm": 1.921210765838623,
"learning_rate": 5.113024772922539e-06,
"loss": 0.02145901620388031,
"step": 8100
},
{
"epoch": 6.735880398671096,
"grad_norm": 0.005192174576222897,
"learning_rate": 5.08957191782189e-06,
"loss": 0.003841530904173851,
"step": 8110
},
{
"epoch": 6.7441860465116275,
"grad_norm": 0.0016935857711359859,
"learning_rate": 5.066154599814043e-06,
"loss": 0.00965556725859642,
"step": 8120
},
{
"epoch": 6.75249169435216,
"grad_norm": 0.0009309044689871371,
"learning_rate": 5.042772988371677e-06,
"loss": 0.01184718832373619,
"step": 8130
},
{
"epoch": 6.760797342192691,
"grad_norm": 3.068490743637085,
"learning_rate": 5.019427252709061e-06,
"loss": 0.011495631933212281,
"step": 8140
},
{
"epoch": 6.769102990033223,
"grad_norm": 0.12647555768489838,
"learning_rate": 4.9961175617808254e-06,
"loss": 0.015127480030059814,
"step": 8150
},
{
"epoch": 6.777408637873754,
"grad_norm": 0.010977406054735184,
"learning_rate": 4.972844084280751e-06,
"loss": 0.03067091703414917,
"step": 8160
},
{
"epoch": 6.785714285714286,
"grad_norm": 0.004081060644239187,
"learning_rate": 4.949606988640532e-06,
"loss": 0.010084083676338196,
"step": 8170
},
{
"epoch": 6.794019933554817,
"grad_norm": 18.21247100830078,
"learning_rate": 4.92640644302856e-06,
"loss": 0.02862970232963562,
"step": 8180
},
{
"epoch": 6.8023255813953485,
"grad_norm": 0.36443743109703064,
"learning_rate": 4.903242615348727e-06,
"loss": 0.023679034411907197,
"step": 8190
},
{
"epoch": 6.810631229235881,
"grad_norm": 0.15573111176490784,
"learning_rate": 4.880115673239182e-06,
"loss": 0.02482050210237503,
"step": 8200
},
{
"epoch": 6.818936877076412,
"grad_norm": 0.05145455151796341,
"learning_rate": 4.857025784071131e-06,
"loss": 0.020545361936092375,
"step": 8210
},
{
"epoch": 6.827242524916944,
"grad_norm": 0.0019702017307281494,
"learning_rate": 4.833973114947641e-06,
"loss": 0.00281891580671072,
"step": 8220
},
{
"epoch": 6.835548172757475,
"grad_norm": 0.07389739900827408,
"learning_rate": 4.8109578327023865e-06,
"loss": 0.014239233732223511,
"step": 8230
},
{
"epoch": 6.843853820598007,
"grad_norm": 0.06578060984611511,
"learning_rate": 4.7879801038985e-06,
"loss": 0.013229538500308991,
"step": 8240
},
{
"epoch": 6.852159468438538,
"grad_norm": 0.36115917563438416,
"learning_rate": 4.76504009482732e-06,
"loss": 0.028942742943763734,
"step": 8250
},
{
"epoch": 6.8604651162790695,
"grad_norm": 0.003231554990634322,
"learning_rate": 4.7421379715072055e-06,
"loss": 0.0021589316427707674,
"step": 8260
},
{
"epoch": 6.868770764119601,
"grad_norm": 1.7378817796707153,
"learning_rate": 4.719273899682347e-06,
"loss": 0.011100687831640244,
"step": 8270
},
{
"epoch": 6.877076411960132,
"grad_norm": 0.0012914666440337896,
"learning_rate": 4.696448044821539e-06,
"loss": 0.007763214409351349,
"step": 8280
},
{
"epoch": 6.885382059800665,
"grad_norm": 0.09640122205018997,
"learning_rate": 4.6736605721169985e-06,
"loss": 0.015946491062641142,
"step": 8290
},
{
"epoch": 6.893687707641196,
"grad_norm": 0.1830437332391739,
"learning_rate": 4.6509116464831865e-06,
"loss": 0.025258588790893554,
"step": 8300
},
{
"epoch": 6.901993355481728,
"grad_norm": 0.0037035609129816294,
"learning_rate": 4.628201432555564e-06,
"loss": 0.013042263686656952,
"step": 8310
},
{
"epoch": 6.910299003322259,
"grad_norm": 0.0029025475960224867,
"learning_rate": 4.605530094689464e-06,
"loss": 0.006066365912556648,
"step": 8320
},
{
"epoch": 6.9186046511627906,
"grad_norm": 3.2578377723693848,
"learning_rate": 4.582897796958856e-06,
"loss": 0.015941482782363892,
"step": 8330
},
{
"epoch": 6.926910299003322,
"grad_norm": 2.1002743244171143,
"learning_rate": 4.56030470315517e-06,
"loss": 0.020402732491493224,
"step": 8340
},
{
"epoch": 6.9352159468438535,
"grad_norm": 0.004095923621207476,
"learning_rate": 4.537750976786127e-06,
"loss": 0.012997651100158691,
"step": 8350
},
{
"epoch": 6.943521594684386,
"grad_norm": 0.0027635928709059954,
"learning_rate": 4.5152367810745346e-06,
"loss": 0.021826334297657013,
"step": 8360
},
{
"epoch": 6.951827242524917,
"grad_norm": 1.912922739982605,
"learning_rate": 4.492762278957114e-06,
"loss": 0.01737479567527771,
"step": 8370
},
{
"epoch": 6.960132890365449,
"grad_norm": 4.833646774291992,
"learning_rate": 4.4703276330833234e-06,
"loss": 0.026379984617233277,
"step": 8380
},
{
"epoch": 6.96843853820598,
"grad_norm": 0.10403254628181458,
"learning_rate": 4.447933005814182e-06,
"loss": 0.007414986193180084,
"step": 8390
},
{
"epoch": 6.976744186046512,
"grad_norm": 0.30322086811065674,
"learning_rate": 4.425578559221088e-06,
"loss": 0.027910080552101136,
"step": 8400
},
{
"epoch": 6.985049833887043,
"grad_norm": 2.7283883094787598,
"learning_rate": 4.403264455084647e-06,
"loss": 0.06774747371673584,
"step": 8410
},
{
"epoch": 6.9933554817275745,
"grad_norm": 0.009378539398312569,
"learning_rate": 4.380990854893509e-06,
"loss": 0.0005745979025959969,
"step": 8420
},
{
"epoch": 7.001661129568106,
"grad_norm": 3.630213975906372,
"learning_rate": 4.3587579198431874e-06,
"loss": 0.010593216121196746,
"step": 8430
},
{
"epoch": 7.009966777408638,
"grad_norm": 0.23145924508571625,
"learning_rate": 4.33656581083491e-06,
"loss": 0.02232535183429718,
"step": 8440
},
{
"epoch": 7.01827242524917,
"grad_norm": 0.2621423900127411,
"learning_rate": 4.3144146884744355e-06,
"loss": 0.028794461488723756,
"step": 8450
},
{
"epoch": 7.026578073089701,
"grad_norm": 0.1324620544910431,
"learning_rate": 4.292304713070895e-06,
"loss": 0.023257753252983092,
"step": 8460
},
{
"epoch": 7.034883720930233,
"grad_norm": 0.0013552814489230514,
"learning_rate": 4.270236044635657e-06,
"loss": 0.009090982377529144,
"step": 8470
},
{
"epoch": 7.043189368770764,
"grad_norm": 2.971010446548462,
"learning_rate": 4.248208842881118e-06,
"loss": 0.00317855142056942,
"step": 8480
},
{
"epoch": 7.0514950166112955,
"grad_norm": 5.764839172363281,
"learning_rate": 4.226223267219606e-06,
"loss": 0.03147430717945099,
"step": 8490
},
{
"epoch": 7.059800664451827,
"grad_norm": 3.030252456665039,
"learning_rate": 4.204279476762182e-06,
"loss": 0.012821225821971894,
"step": 8500
},
{
"epoch": 7.059800664451827,
"eval_brier": 0.967,
"eval_c@1": 0.962,
"eval_f05u": 0.974,
"eval_f1": 0.964,
"eval_loss": 0.14606554806232452,
"eval_mean": 0.972,
"eval_roc-auc": 0.991,
"eval_runtime": 69.2051,
"eval_samples_per_second": 266.209,
"eval_steps_per_second": 4.162,
"step": 8500
},
{
"epoch": 7.068106312292358,
"grad_norm": 0.15249097347259521,
"learning_rate": 4.182377630317505e-06,
"loss": 0.0239920511841774,
"step": 8510
},
{
"epoch": 7.076411960132891,
"grad_norm": 1.1104687452316284,
"learning_rate": 4.160517886390694e-06,
"loss": 0.029243892431259154,
"step": 8520
},
{
"epoch": 7.084717607973422,
"grad_norm": 1.4209219217300415,
"learning_rate": 4.138700403182157e-06,
"loss": 0.0012476762756705284,
"step": 8530
},
{
"epoch": 7.093023255813954,
"grad_norm": 0.39821135997772217,
"learning_rate": 4.116925338586461e-06,
"loss": 0.02069404423236847,
"step": 8540
},
{
"epoch": 7.101328903654485,
"grad_norm": 0.27982714772224426,
"learning_rate": 4.095192850191198e-06,
"loss": 0.01682550609111786,
"step": 8550
},
{
"epoch": 7.1096345514950166,
"grad_norm": 0.15819349884986877,
"learning_rate": 4.07350309527581e-06,
"loss": 0.006231273710727692,
"step": 8560
},
{
"epoch": 7.117940199335548,
"grad_norm": 0.0013464418007060885,
"learning_rate": 4.0518562308104995e-06,
"loss": 0.008467202633619308,
"step": 8570
},
{
"epoch": 7.1262458471760795,
"grad_norm": 0.04739031568169594,
"learning_rate": 4.03025241345505e-06,
"loss": 0.010148204118013381,
"step": 8580
},
{
"epoch": 7.134551495016611,
"grad_norm": 6.338361740112305,
"learning_rate": 4.008691799557717e-06,
"loss": 0.02180970013141632,
"step": 8590
},
{
"epoch": 7.142857142857143,
"grad_norm": 0.918834924697876,
"learning_rate": 3.987174545154095e-06,
"loss": 0.03664710223674774,
"step": 8600
},
{
"epoch": 7.151162790697675,
"grad_norm": 0.001258784788660705,
"learning_rate": 3.965700805965971e-06,
"loss": 0.03319277465343475,
"step": 8610
},
{
"epoch": 7.159468438538206,
"grad_norm": 0.001618512556888163,
"learning_rate": 3.9442707374002185e-06,
"loss": 0.008543861657381057,
"step": 8620
},
{
"epoch": 7.167774086378738,
"grad_norm": 0.7446807622909546,
"learning_rate": 3.922884494547658e-06,
"loss": 0.0017049346119165421,
"step": 8630
},
{
"epoch": 7.176079734219269,
"grad_norm": 0.23487555980682373,
"learning_rate": 3.9015422321819475e-06,
"loss": 0.011183664202690125,
"step": 8640
},
{
"epoch": 7.1843853820598005,
"grad_norm": 0.5079420208930969,
"learning_rate": 3.880244104758451e-06,
"loss": 0.013415685296058655,
"step": 8650
},
{
"epoch": 7.192691029900332,
"grad_norm": 0.6529298424720764,
"learning_rate": 3.858990266413123e-06,
"loss": 0.01553943008184433,
"step": 8660
},
{
"epoch": 7.200996677740863,
"grad_norm": 0.0024889109190553427,
"learning_rate": 3.8377808709613995e-06,
"loss": 0.01116732954978943,
"step": 8670
},
{
"epoch": 7.209302325581396,
"grad_norm": 0.058576200157403946,
"learning_rate": 3.816616071897073e-06,
"loss": 0.013654626905918121,
"step": 8680
},
{
"epoch": 7.217607973421927,
"grad_norm": 0.13021883368492126,
"learning_rate": 3.7954960223912006e-06,
"loss": 0.02838684618473053,
"step": 8690
},
{
"epoch": 7.225913621262459,
"grad_norm": 0.12662293016910553,
"learning_rate": 3.774420875290976e-06,
"loss": 0.025519010424613953,
"step": 8700
},
{
"epoch": 7.23421926910299,
"grad_norm": 0.4974972605705261,
"learning_rate": 3.7533907831186287e-06,
"loss": 0.016147831082344057,
"step": 8710
},
{
"epoch": 7.2425249169435215,
"grad_norm": 0.29633164405822754,
"learning_rate": 3.732405898070336e-06,
"loss": 0.010277282446622849,
"step": 8720
},
{
"epoch": 7.250830564784053,
"grad_norm": 0.19185616075992584,
"learning_rate": 3.7114663720150856e-06,
"loss": 0.004892281442880631,
"step": 8730
},
{
"epoch": 7.259136212624584,
"grad_norm": 0.268136590719223,
"learning_rate": 3.690572356493619e-06,
"loss": 0.030184558033943175,
"step": 8740
},
{
"epoch": 7.267441860465116,
"grad_norm": 3.44865345954895,
"learning_rate": 3.669724002717315e-06,
"loss": 0.03916491270065307,
"step": 8750
},
{
"epoch": 7.275747508305648,
"grad_norm": 1.0811718702316284,
"learning_rate": 3.6489214615670777e-06,
"loss": 0.052170008420944214,
"step": 8760
},
{
"epoch": 7.28405315614618,
"grad_norm": 0.0022155882325023413,
"learning_rate": 3.62816488359228e-06,
"loss": 0.004978426545858383,
"step": 8770
},
{
"epoch": 7.292358803986711,
"grad_norm": 0.03842281177639961,
"learning_rate": 3.6074544190096483e-06,
"loss": 0.015476545691490174,
"step": 8780
},
{
"epoch": 7.3006644518272426,
"grad_norm": 0.18393191695213318,
"learning_rate": 3.5867902177021807e-06,
"loss": 0.022972114384174347,
"step": 8790
},
{
"epoch": 7.308970099667774,
"grad_norm": 8.703371047973633,
"learning_rate": 3.5661724292180755e-06,
"loss": 0.03231580257415771,
"step": 8800
},
{
"epoch": 7.3172757475083055,
"grad_norm": 0.619775116443634,
"learning_rate": 3.5456012027696276e-06,
"loss": 0.009065944701433182,
"step": 8810
},
{
"epoch": 7.325581395348837,
"grad_norm": 0.002593767363578081,
"learning_rate": 3.525076687232164e-06,
"loss": 0.000641344441100955,
"step": 8820
},
{
"epoch": 7.333887043189369,
"grad_norm": 0.04139446094632149,
"learning_rate": 3.5045990311429612e-06,
"loss": 0.007180669158697128,
"step": 8830
},
{
"epoch": 7.342192691029901,
"grad_norm": 0.14698491990566254,
"learning_rate": 3.484168382700167e-06,
"loss": 0.016698645055294038,
"step": 8840
},
{
"epoch": 7.350498338870432,
"grad_norm": 1.8153339624404907,
"learning_rate": 3.463784889761742e-06,
"loss": 0.03294073343276978,
"step": 8850
},
{
"epoch": 7.358803986710964,
"grad_norm": 0.063026063144207,
"learning_rate": 3.4434486998443704e-06,
"loss": 0.013478732109069825,
"step": 8860
},
{
"epoch": 7.367109634551495,
"grad_norm": 0.0021197758615016937,
"learning_rate": 3.423159960122403e-06,
"loss": 0.00783894509077072,
"step": 8870
},
{
"epoch": 7.3754152823920265,
"grad_norm": 0.03898961469531059,
"learning_rate": 3.402918817426787e-06,
"loss": 0.006163169071078301,
"step": 8880
},
{
"epoch": 7.383720930232558,
"grad_norm": 0.9577112793922424,
"learning_rate": 3.3827254182440182e-06,
"loss": 0.02714129090309143,
"step": 8890
},
{
"epoch": 7.392026578073089,
"grad_norm": 0.40597739815711975,
"learning_rate": 3.3625799087150603e-06,
"loss": 0.01383037269115448,
"step": 8900
},
{
"epoch": 7.400332225913621,
"grad_norm": 0.44809389114379883,
"learning_rate": 3.3424824346342945e-06,
"loss": 0.02304709702730179,
"step": 8910
},
{
"epoch": 7.408637873754153,
"grad_norm": 0.0027849802281707525,
"learning_rate": 3.322433141448472e-06,
"loss": 0.0009386043064296245,
"step": 8920
},
{
"epoch": 7.416943521594685,
"grad_norm": 0.05637195333838463,
"learning_rate": 3.302432174255652e-06,
"loss": 0.0017526715993881226,
"step": 8930
},
{
"epoch": 7.425249169435216,
"grad_norm": 0.04584479331970215,
"learning_rate": 3.2824796778041614e-06,
"loss": 0.03774779438972473,
"step": 8940
},
{
"epoch": 7.4335548172757475,
"grad_norm": 0.8431316614151001,
"learning_rate": 3.262575796491533e-06,
"loss": 0.040053775906562804,
"step": 8950
},
{
"epoch": 7.441860465116279,
"grad_norm": 1.3370380401611328,
"learning_rate": 3.24272067436347e-06,
"loss": 0.011968208849430085,
"step": 8960
},
{
"epoch": 7.45016611295681,
"grad_norm": 2.7681705951690674,
"learning_rate": 3.2229144551128143e-06,
"loss": 0.01593385934829712,
"step": 8970
},
{
"epoch": 7.458471760797342,
"grad_norm": 0.6313722133636475,
"learning_rate": 3.2031572820784708e-06,
"loss": 0.009059444814920426,
"step": 8980
},
{
"epoch": 7.466777408637874,
"grad_norm": 0.30437061190605164,
"learning_rate": 3.1834492982444154e-06,
"loss": 0.006292606890201569,
"step": 8990
},
{
"epoch": 7.475083056478406,
"grad_norm": 0.4104173481464386,
"learning_rate": 3.1637906462386357e-06,
"loss": 0.04081603586673736,
"step": 9000
},
{
"epoch": 7.475083056478406,
"eval_brier": 0.966,
"eval_c@1": 0.961,
"eval_f05u": 0.974,
"eval_f1": 0.963,
"eval_loss": 0.1476876437664032,
"eval_mean": 0.971,
"eval_roc-auc": 0.991,
"eval_runtime": 69.3915,
"eval_samples_per_second": 265.493,
"eval_steps_per_second": 4.15,
"step": 9000
},
{
"epoch": 7.483388704318937,
"grad_norm": 7.053452968597412,
"learning_rate": 3.1441814683320835e-06,
"loss": 0.017213717103004456,
"step": 9010
},
{
"epoch": 7.4916943521594686,
"grad_norm": 0.0025665026623755693,
"learning_rate": 3.124621906437687e-06,
"loss": 0.011792691797018051,
"step": 9020
},
{
"epoch": 7.5,
"grad_norm": 0.062440987676382065,
"learning_rate": 3.1051121021092857e-06,
"loss": 0.011086343973875045,
"step": 9030
},
{
"epoch": 7.5083056478405314,
"grad_norm": 0.0011189680080860853,
"learning_rate": 3.0856521965406194e-06,
"loss": 0.04765462577342987,
"step": 9040
},
{
"epoch": 7.516611295681063,
"grad_norm": 0.08503086864948273,
"learning_rate": 3.066242330564321e-06,
"loss": 0.018283794820308685,
"step": 9050
},
{
"epoch": 7.524916943521594,
"grad_norm": 0.030199186876416206,
"learning_rate": 3.0468826446508683e-06,
"loss": 0.02109091281890869,
"step": 9060
},
{
"epoch": 7.533222591362126,
"grad_norm": 0.002193137537688017,
"learning_rate": 3.0275732789075897e-06,
"loss": 0.020866675674915312,
"step": 9070
},
{
"epoch": 7.541528239202658,
"grad_norm": 0.09009260684251785,
"learning_rate": 3.008314373077642e-06,
"loss": 0.0014830844476819039,
"step": 9080
},
{
"epoch": 7.54983388704319,
"grad_norm": 0.21655766665935516,
"learning_rate": 2.989106066538998e-06,
"loss": 0.007241271436214447,
"step": 9090
},
{
"epoch": 7.558139534883721,
"grad_norm": 0.014798282645642757,
"learning_rate": 2.9699484983034475e-06,
"loss": 0.005920373648405075,
"step": 9100
},
{
"epoch": 7.5664451827242525,
"grad_norm": 5.816130638122559,
"learning_rate": 2.9508418070155777e-06,
"loss": 0.02567654848098755,
"step": 9110
},
{
"epoch": 7.574750830564784,
"grad_norm": 0.0016904993681237102,
"learning_rate": 2.9317861309517757e-06,
"loss": 0.011639448255300522,
"step": 9120
},
{
"epoch": 7.583056478405315,
"grad_norm": 0.2148173600435257,
"learning_rate": 2.912781608019226e-06,
"loss": 0.0017349058762192726,
"step": 9130
},
{
"epoch": 7.591362126245847,
"grad_norm": 2.029493808746338,
"learning_rate": 2.893828375754927e-06,
"loss": 0.004039783775806427,
"step": 9140
},
{
"epoch": 7.599667774086379,
"grad_norm": 0.05651300400495529,
"learning_rate": 2.874926571324668e-06,
"loss": 0.03266519010066986,
"step": 9150
},
{
"epoch": 7.607973421926911,
"grad_norm": 0.20539076626300812,
"learning_rate": 2.8560763315220542e-06,
"loss": 0.016243191063404085,
"step": 9160
},
{
"epoch": 7.616279069767442,
"grad_norm": 0.001784057356417179,
"learning_rate": 2.83727779276753e-06,
"loss": 0.0005050742067396641,
"step": 9170
},
{
"epoch": 7.6245847176079735,
"grad_norm": 0.8474004864692688,
"learning_rate": 2.8185310911073504e-06,
"loss": 0.017673566937446594,
"step": 9180
},
{
"epoch": 7.632890365448505,
"grad_norm": 1.9227701425552368,
"learning_rate": 2.7998363622126432e-06,
"loss": 0.04959536790847778,
"step": 9190
},
{
"epoch": 7.641196013289036,
"grad_norm": 15.710461616516113,
"learning_rate": 2.781193741378396e-06,
"loss": 0.025568419694900514,
"step": 9200
},
{
"epoch": 7.649501661129568,
"grad_norm": 8.397846221923828,
"learning_rate": 2.7626033635224847e-06,
"loss": 0.04204786717891693,
"step": 9210
},
{
"epoch": 7.657807308970099,
"grad_norm": 0.004305708222091198,
"learning_rate": 2.7440653631847103e-06,
"loss": 0.016460147500038148,
"step": 9220
},
{
"epoch": 7.666112956810631,
"grad_norm": 1.461538314819336,
"learning_rate": 2.7255798745258044e-06,
"loss": 0.009647966921329498,
"step": 9230
},
{
"epoch": 7.674418604651163,
"grad_norm": 13.192046165466309,
"learning_rate": 2.707147031326468e-06,
"loss": 0.031538856029510495,
"step": 9240
},
{
"epoch": 7.6827242524916945,
"grad_norm": 0.0040161446668207645,
"learning_rate": 2.6887669669864146e-06,
"loss": 0.008157265931367874,
"step": 9250
},
{
"epoch": 7.691029900332226,
"grad_norm": 0.3406631648540497,
"learning_rate": 2.6704398145233746e-06,
"loss": 0.02682391107082367,
"step": 9260
},
{
"epoch": 7.6993355481727574,
"grad_norm": 0.0013381581520661712,
"learning_rate": 2.652165706572171e-06,
"loss": 0.012091658264398574,
"step": 9270
},
{
"epoch": 7.707641196013289,
"grad_norm": 0.6766977906227112,
"learning_rate": 2.633944775383729e-06,
"loss": 0.0013755757361650466,
"step": 9280
},
{
"epoch": 7.71594684385382,
"grad_norm": 1.444471836090088,
"learning_rate": 2.615777152824128e-06,
"loss": 0.04140243232250214,
"step": 9290
},
{
"epoch": 7.724252491694352,
"grad_norm": 0.14766523241996765,
"learning_rate": 2.597662970373662e-06,
"loss": 0.010613334178924561,
"step": 9300
},
{
"epoch": 7.732558139534884,
"grad_norm": 0.1722085028886795,
"learning_rate": 2.5796023591258645e-06,
"loss": 0.016667431592941283,
"step": 9310
},
{
"epoch": 7.740863787375416,
"grad_norm": 0.0021270038560032845,
"learning_rate": 2.5615954497865727e-06,
"loss": 0.011866636574268341,
"step": 9320
},
{
"epoch": 7.749169435215947,
"grad_norm": 0.4936175048351288,
"learning_rate": 2.543642372672981e-06,
"loss": 0.013976240158081054,
"step": 9330
},
{
"epoch": 7.7574750830564785,
"grad_norm": 1.179762363433838,
"learning_rate": 2.5257432577126926e-06,
"loss": 0.018904729187488555,
"step": 9340
},
{
"epoch": 7.76578073089701,
"grad_norm": 2.503565549850464,
"learning_rate": 2.5078982344427915e-06,
"loss": 0.028677886724472045,
"step": 9350
},
{
"epoch": 7.774086378737541,
"grad_norm": 0.023880483582615852,
"learning_rate": 2.4901074320088904e-06,
"loss": 0.030220001935958862,
"step": 9360
},
{
"epoch": 7.782392026578073,
"grad_norm": 0.0009466466144658625,
"learning_rate": 2.4723709791642013e-06,
"loss": 0.007851962000131607,
"step": 9370
},
{
"epoch": 7.790697674418604,
"grad_norm": 0.3103953003883362,
"learning_rate": 2.4546890042686046e-06,
"loss": 0.01569589376449585,
"step": 9380
},
{
"epoch": 7.799003322259136,
"grad_norm": 4.8353166580200195,
"learning_rate": 2.4370616352877276e-06,
"loss": 0.034313428401947024,
"step": 9390
},
{
"epoch": 7.807308970099668,
"grad_norm": 0.16469524800777435,
"learning_rate": 2.4194889997920024e-06,
"loss": 0.019627228379249573,
"step": 9400
},
{
"epoch": 7.8156146179401995,
"grad_norm": 0.0011383763048797846,
"learning_rate": 2.4019712249557516e-06,
"loss": 0.003571658208966255,
"step": 9410
},
{
"epoch": 7.823920265780731,
"grad_norm": 0.0031425883062183857,
"learning_rate": 2.384508437556279e-06,
"loss": 0.012809967994689942,
"step": 9420
},
{
"epoch": 7.832225913621262,
"grad_norm": 0.34275543689727783,
"learning_rate": 2.3671007639729195e-06,
"loss": 0.0009221424348652363,
"step": 9430
},
{
"epoch": 7.840531561461794,
"grad_norm": 0.24381868541240692,
"learning_rate": 2.349748330186169e-06,
"loss": 0.021050933003425597,
"step": 9440
},
{
"epoch": 7.848837209302325,
"grad_norm": 1.5793051719665527,
"learning_rate": 2.3324512617767337e-06,
"loss": 0.02660844922065735,
"step": 9450
},
{
"epoch": 7.857142857142857,
"grad_norm": 0.0608382374048233,
"learning_rate": 2.3152096839246398e-06,
"loss": 0.027494984865188598,
"step": 9460
},
{
"epoch": 7.865448504983389,
"grad_norm": 0.0014419383369386196,
"learning_rate": 2.298023721408331e-06,
"loss": 0.006231945753097534,
"step": 9470
},
{
"epoch": 7.8737541528239205,
"grad_norm": 10.373838424682617,
"learning_rate": 2.2808934986037545e-06,
"loss": 0.016826078295707703,
"step": 9480
},
{
"epoch": 7.882059800664452,
"grad_norm": 0.48430129885673523,
"learning_rate": 2.2638191394834596e-06,
"loss": 0.03645863831043243,
"step": 9490
},
{
"epoch": 7.8903654485049834,
"grad_norm": 0.3142700493335724,
"learning_rate": 2.2468007676157232e-06,
"loss": 0.005658136680722237,
"step": 9500
},
{
"epoch": 7.8903654485049834,
"eval_brier": 0.967,
"eval_c@1": 0.962,
"eval_f05u": 0.974,
"eval_f1": 0.964,
"eval_loss": 0.14827124774456024,
"eval_mean": 0.972,
"eval_roc-auc": 0.991,
"eval_runtime": 69.2065,
"eval_samples_per_second": 266.203,
"eval_steps_per_second": 4.161,
"step": 9500
},
{
"epoch": 7.898671096345515,
"grad_norm": 20.80424690246582,
"learning_rate": 2.2298385061636162e-06,
"loss": 0.043711668252944945,
"step": 9510
},
{
"epoch": 7.906976744186046,
"grad_norm": 0.008154784329235554,
"learning_rate": 2.212932477884154e-06,
"loss": 0.005724793672561646,
"step": 9520
},
{
"epoch": 7.915282392026578,
"grad_norm": 5.501669406890869,
"learning_rate": 2.196082805127381e-06,
"loss": 0.036894920468330386,
"step": 9530
},
{
"epoch": 7.923588039867109,
"grad_norm": 3.6727075576782227,
"learning_rate": 2.1792896098354887e-06,
"loss": 0.015390019118785857,
"step": 9540
},
{
"epoch": 7.931893687707642,
"grad_norm": 1.5521719455718994,
"learning_rate": 2.162553013541948e-06,
"loss": 0.011379404366016388,
"step": 9550
},
{
"epoch": 7.940199335548173,
"grad_norm": 0.6106346845626831,
"learning_rate": 2.145873137370611e-06,
"loss": 0.024071736633777617,
"step": 9560
},
{
"epoch": 7.9485049833887045,
"grad_norm": 0.0013641950208693743,
"learning_rate": 2.1292501020348445e-06,
"loss": 0.0028417007997632027,
"step": 9570
},
{
"epoch": 7.956810631229236,
"grad_norm": 0.0007085945107974112,
"learning_rate": 2.112684027836651e-06,
"loss": 0.008503476530313492,
"step": 9580
},
{
"epoch": 7.965116279069767,
"grad_norm": 3.62831449508667,
"learning_rate": 2.0961750346658117e-06,
"loss": 0.029273918271064757,
"step": 9590
},
{
"epoch": 7.973421926910299,
"grad_norm": 0.032878801226615906,
"learning_rate": 2.079723241998999e-06,
"loss": 0.009919158369302749,
"step": 9600
},
{
"epoch": 7.98172757475083,
"grad_norm": 0.16056130826473236,
"learning_rate": 2.0633287688989247e-06,
"loss": 0.020749954879283904,
"step": 9610
},
{
"epoch": 7.990033222591363,
"grad_norm": 3.7980868816375732,
"learning_rate": 2.046991734013475e-06,
"loss": 0.011913499981164932,
"step": 9620
},
{
"epoch": 7.998338870431894,
"grad_norm": 2.1085927486419678,
"learning_rate": 2.030712255574848e-06,
"loss": 0.0013692128472030164,
"step": 9630
},
{
"epoch": 8.006644518272426,
"grad_norm": 0.001301309559494257,
"learning_rate": 2.0144904513987142e-06,
"loss": 0.020541465282440184,
"step": 9640
},
{
"epoch": 8.014950166112957,
"grad_norm": 0.04368695244193077,
"learning_rate": 1.9983264388833368e-06,
"loss": 0.024039453268051146,
"step": 9650
},
{
"epoch": 8.023255813953488,
"grad_norm": 0.07038889080286026,
"learning_rate": 1.982220335008743e-06,
"loss": 0.03967412412166595,
"step": 9660
},
{
"epoch": 8.03156146179402,
"grad_norm": 0.004006698727607727,
"learning_rate": 1.96617225633588e-06,
"loss": 0.005919493734836578,
"step": 9670
},
{
"epoch": 8.039867109634551,
"grad_norm": 0.0018698818748816848,
"learning_rate": 1.950182319005743e-06,
"loss": 0.0006796402856707573,
"step": 9680
},
{
"epoch": 8.048172757475083,
"grad_norm": 7.567103862762451,
"learning_rate": 1.934250638738572e-06,
"loss": 0.026210534572601318,
"step": 9690
},
{
"epoch": 8.056478405315614,
"grad_norm": 4.661642074584961,
"learning_rate": 1.9183773308329958e-06,
"loss": 0.037559321522712706,
"step": 9700
},
{
"epoch": 8.064784053156146,
"grad_norm": 5.240715980529785,
"learning_rate": 1.9025625101651824e-06,
"loss": 0.030109569430351257,
"step": 9710
},
{
"epoch": 8.073089700996677,
"grad_norm": 0.0009405300370417535,
"learning_rate": 1.8868062911880446e-06,
"loss": 0.009581131488084793,
"step": 9720
},
{
"epoch": 8.081395348837209,
"grad_norm": 2.3925936222076416,
"learning_rate": 1.8711087879303757e-06,
"loss": 0.004385541379451752,
"step": 9730
},
{
"epoch": 8.08970099667774,
"grad_norm": 1.739111304283142,
"learning_rate": 1.855470113996045e-06,
"loss": 0.026923447847366333,
"step": 9740
},
{
"epoch": 8.098006644518273,
"grad_norm": 0.4548828601837158,
"learning_rate": 1.8398903825631776e-06,
"loss": 0.014367961883544922,
"step": 9750
},
{
"epoch": 8.106312292358805,
"grad_norm": 0.1681939661502838,
"learning_rate": 1.8243697063833087e-06,
"loss": 0.01353817731142044,
"step": 9760
},
{
"epoch": 8.114617940199336,
"grad_norm": 0.0006201230571605265,
"learning_rate": 1.8089081977806077e-06,
"loss": 0.004853256419301033,
"step": 9770
},
{
"epoch": 8.122923588039868,
"grad_norm": 0.0008420735248364508,
"learning_rate": 1.7935059686510303e-06,
"loss": 0.00014945025322958826,
"step": 9780
},
{
"epoch": 8.131229235880399,
"grad_norm": 0.08449747413396835,
"learning_rate": 1.7781631304615265e-06,
"loss": 0.004913771525025368,
"step": 9790
},
{
"epoch": 8.13953488372093,
"grad_norm": 0.12251075357198715,
"learning_rate": 1.762879794249236e-06,
"loss": 0.03128696382045746,
"step": 9800
},
{
"epoch": 8.147840531561462,
"grad_norm": 0.04864448681473732,
"learning_rate": 1.7476560706206703e-06,
"loss": 0.009850189089775085,
"step": 9810
},
{
"epoch": 8.156146179401993,
"grad_norm": 0.001526945037767291,
"learning_rate": 1.732492069750925e-06,
"loss": 0.01179589331150055,
"step": 9820
},
{
"epoch": 8.164451827242525,
"grad_norm": 0.0008230694220401347,
"learning_rate": 1.7173879013828743e-06,
"loss": 0.02024877518415451,
"step": 9830
},
{
"epoch": 8.172757475083056,
"grad_norm": 1.857641339302063,
"learning_rate": 1.7023436748263889e-06,
"loss": 0.01102503165602684,
"step": 9840
},
{
"epoch": 8.181063122923588,
"grad_norm": 0.010417718440294266,
"learning_rate": 1.6873594989575292e-06,
"loss": 0.04865381419658661,
"step": 9850
},
{
"epoch": 8.18936877076412,
"grad_norm": 0.0010410118848085403,
"learning_rate": 1.672435482217767e-06,
"loss": 0.015278685092926025,
"step": 9860
},
{
"epoch": 8.19767441860465,
"grad_norm": 0.003762697335332632,
"learning_rate": 1.6575717326131978e-06,
"loss": 0.0025358112528920175,
"step": 9870
},
{
"epoch": 8.205980066445182,
"grad_norm": 0.7878845930099487,
"learning_rate": 1.6427683577137598e-06,
"loss": 0.0023263685405254366,
"step": 9880
},
{
"epoch": 8.214285714285714,
"grad_norm": 1.1144360303878784,
"learning_rate": 1.6280254646524584e-06,
"loss": 0.013678796589374542,
"step": 9890
},
{
"epoch": 8.222591362126245,
"grad_norm": 9.173377990722656,
"learning_rate": 1.6133431601245864e-06,
"loss": 0.030849790573120116,
"step": 9900
},
{
"epoch": 8.230897009966778,
"grad_norm": 1.0350559949874878,
"learning_rate": 1.598721550386947e-06,
"loss": 0.045111915469169615,
"step": 9910
},
{
"epoch": 8.23920265780731,
"grad_norm": 0.006207343190908432,
"learning_rate": 1.5841607412571048e-06,
"loss": 0.038734915852546695,
"step": 9920
},
{
"epoch": 8.247508305647841,
"grad_norm": 0.0021894723176956177,
"learning_rate": 1.5696608381125944e-06,
"loss": 0.003963911905884743,
"step": 9930
},
{
"epoch": 8.255813953488373,
"grad_norm": 0.20098142325878143,
"learning_rate": 1.5552219458901717e-06,
"loss": 0.006362013518810272,
"step": 9940
},
{
"epoch": 8.264119601328904,
"grad_norm": 0.24506373703479767,
"learning_rate": 1.5408441690850628e-06,
"loss": 0.012679998576641083,
"step": 9950
},
{
"epoch": 8.272425249169435,
"grad_norm": 4.701864242553711,
"learning_rate": 1.5265276117501781e-06,
"loss": 0.015318554639816285,
"step": 9960
},
{
"epoch": 8.280730897009967,
"grad_norm": 1.0198193788528442,
"learning_rate": 1.5122723774953984e-06,
"loss": 0.03276909589767456,
"step": 9970
},
{
"epoch": 8.289036544850498,
"grad_norm": 0.0008463190170004964,
"learning_rate": 1.4980785694867938e-06,
"loss": 7.541636587120593e-05,
"step": 9980
},
{
"epoch": 8.29734219269103,
"grad_norm": 0.030637215822935104,
"learning_rate": 1.48394629044589e-06,
"loss": 0.0014228528365492822,
"step": 9990
},
{
"epoch": 8.305647840531561,
"grad_norm": 0.028038810938596725,
"learning_rate": 1.4698756426489302e-06,
"loss": 0.026153796911239625,
"step": 10000
},
{
"epoch": 8.305647840531561,
"eval_brier": 0.968,
"eval_c@1": 0.965,
"eval_f05u": 0.975,
"eval_f1": 0.966,
"eval_loss": 0.1502263844013214,
"eval_mean": 0.973,
"eval_roc-auc": 0.991,
"eval_runtime": 69.299,
"eval_samples_per_second": 265.848,
"eval_steps_per_second": 4.156,
"step": 10000
},
{
"epoch": 8.313953488372093,
"grad_norm": 4.307137489318848,
"learning_rate": 1.4558667279261185e-06,
"loss": 0.011831691116094589,
"step": 10010
},
{
"epoch": 8.322259136212624,
"grad_norm": 3.225126028060913,
"learning_rate": 1.4419196476608976e-06,
"loss": 0.010735167562961579,
"step": 10020
},
{
"epoch": 8.330564784053156,
"grad_norm": 4.529547214508057,
"learning_rate": 1.428034502789215e-06,
"loss": 0.019614629447460175,
"step": 10030
},
{
"epoch": 8.338870431893687,
"grad_norm": 11.928061485290527,
"learning_rate": 1.4142113937987744e-06,
"loss": 0.020872780680656434,
"step": 10040
},
{
"epoch": 8.347176079734218,
"grad_norm": 0.11286621540784836,
"learning_rate": 1.4004504207283354e-06,
"loss": 0.006154395639896393,
"step": 10050
},
{
"epoch": 8.355481727574752,
"grad_norm": 0.46928834915161133,
"learning_rate": 1.3867516831669692e-06,
"loss": 0.006838122010231018,
"step": 10060
},
{
"epoch": 8.363787375415283,
"grad_norm": 0.001144749578088522,
"learning_rate": 1.373115280253343e-06,
"loss": 0.0020782141014933585,
"step": 10070
},
{
"epoch": 8.372093023255815,
"grad_norm": 16.727407455444336,
"learning_rate": 1.3595413106750099e-06,
"loss": 0.01910678595304489,
"step": 10080
},
{
"epoch": 8.380398671096346,
"grad_norm": 0.4497424364089966,
"learning_rate": 1.3460298726676856e-06,
"loss": 0.015307162702083588,
"step": 10090
},
{
"epoch": 8.388704318936878,
"grad_norm": 0.3464401960372925,
"learning_rate": 1.3325810640145388e-06,
"loss": 0.02306944876909256,
"step": 10100
},
{
"epoch": 8.397009966777409,
"grad_norm": 0.01774173602461815,
"learning_rate": 1.3191949820454853e-06,
"loss": 0.0036763854324817656,
"step": 10110
},
{
"epoch": 8.40531561461794,
"grad_norm": 0.0566246397793293,
"learning_rate": 1.3058717236364903e-06,
"loss": 0.003263864666223526,
"step": 10120
},
{
"epoch": 8.413621262458472,
"grad_norm": 0.0011068865424022079,
"learning_rate": 1.292611385208854e-06,
"loss": 8.122787694446743e-05,
"step": 10130
},
{
"epoch": 8.421926910299003,
"grad_norm": 3.1502633094787598,
"learning_rate": 1.2794140627285235e-06,
"loss": 0.013773564994335175,
"step": 10140
},
{
"epoch": 8.430232558139535,
"grad_norm": 0.046208497136831284,
"learning_rate": 1.266279851705392e-06,
"loss": 0.05285004377365112,
"step": 10150
},
{
"epoch": 8.438538205980066,
"grad_norm": 0.005908760242164135,
"learning_rate": 1.2532088471926129e-06,
"loss": 0.014551596343517303,
"step": 10160
},
{
"epoch": 8.446843853820598,
"grad_norm": 0.0011245819041505456,
"learning_rate": 1.240201143785914e-06,
"loss": 0.03371871411800385,
"step": 10170
},
{
"epoch": 8.455149501661129,
"grad_norm": 0.000950102461501956,
"learning_rate": 1.2272568356229041e-06,
"loss": 0.00018040918512269856,
"step": 10180
},
{
"epoch": 8.46345514950166,
"grad_norm": 0.3537558615207672,
"learning_rate": 1.2143760163823947e-06,
"loss": 0.010649013519287109,
"step": 10190
},
{
"epoch": 8.471760797342192,
"grad_norm": 0.14442045986652374,
"learning_rate": 1.2015587792837335e-06,
"loss": 0.04385151267051697,
"step": 10200
},
{
"epoch": 8.480066445182723,
"grad_norm": 0.04695224016904831,
"learning_rate": 1.1888052170861032e-06,
"loss": 0.017310434579849245,
"step": 10210
},
{
"epoch": 8.488372093023255,
"grad_norm": 1.5139660835266113,
"learning_rate": 1.1761154220878846e-06,
"loss": 0.0013035543262958527,
"step": 10220
},
{
"epoch": 8.496677740863788,
"grad_norm": 0.0016350011574104428,
"learning_rate": 1.1634894861259605e-06,
"loss": 0.0013858130201697349,
"step": 10230
},
{
"epoch": 8.50498338870432,
"grad_norm": 0.06126336008310318,
"learning_rate": 1.150927500575063e-06,
"loss": 0.043176472187042236,
"step": 10240
},
{
"epoch": 8.513289036544851,
"grad_norm": 0.05165313184261322,
"learning_rate": 1.1384295563471158e-06,
"loss": 0.003516947850584984,
"step": 10250
},
{
"epoch": 8.521594684385382,
"grad_norm": 0.41692858934402466,
"learning_rate": 1.1259957438905666e-06,
"loss": 0.02686992883682251,
"step": 10260
},
{
"epoch": 8.529900332225914,
"grad_norm": 0.001009366475045681,
"learning_rate": 1.1136261531897364e-06,
"loss": 0.01156647726893425,
"step": 10270
},
{
"epoch": 8.538205980066445,
"grad_norm": 0.0009983720956370234,
"learning_rate": 1.1013208737641778e-06,
"loss": 6.817497196607292e-05,
"step": 10280
},
{
"epoch": 8.546511627906977,
"grad_norm": 0.051713522523641586,
"learning_rate": 1.0890799946680019e-06,
"loss": 0.027422842383384705,
"step": 10290
},
{
"epoch": 8.554817275747508,
"grad_norm": 30.517282485961914,
"learning_rate": 1.0769036044892678e-06,
"loss": 0.006294220685958862,
"step": 10300
},
{
"epoch": 8.56312292358804,
"grad_norm": 1.1368924379348755,
"learning_rate": 1.0647917913493144e-06,
"loss": 0.007869845628738404,
"step": 10310
},
{
"epoch": 8.571428571428571,
"grad_norm": 0.0017508171731606126,
"learning_rate": 1.0527446429021327e-06,
"loss": 0.0036784913390874863,
"step": 10320
},
{
"epoch": 8.579734219269103,
"grad_norm": 0.0011447499273344874,
"learning_rate": 1.0407622463337364e-06,
"loss": 0.00010180356912314892,
"step": 10330
},
{
"epoch": 8.588039867109634,
"grad_norm": 0.08397141098976135,
"learning_rate": 1.0288446883615189e-06,
"loss": 0.018365205824375154,
"step": 10340
},
{
"epoch": 8.596345514950166,
"grad_norm": 0.006722956895828247,
"learning_rate": 1.0169920552336376e-06,
"loss": 0.009779255837202072,
"step": 10350
},
{
"epoch": 8.604651162790697,
"grad_norm": 1.6663017272949219,
"learning_rate": 1.0052044327283794e-06,
"loss": 0.024825190007686616,
"step": 10360
},
{
"epoch": 8.612956810631228,
"grad_norm": 1.009850025177002,
"learning_rate": 9.934819061535517e-07,
"loss": 0.00916222259402275,
"step": 10370
},
{
"epoch": 8.621262458471762,
"grad_norm": 0.0005849698791280389,
"learning_rate": 9.81824560345853e-07,
"loss": 0.02619781196117401,
"step": 10380
},
{
"epoch": 8.629568106312293,
"grad_norm": 2.5450963973999023,
"learning_rate": 9.702324796702657e-07,
"loss": 0.010657285153865815,
"step": 10390
},
{
"epoch": 8.637873754152825,
"grad_norm": 2.804811954498291,
"learning_rate": 9.58705748019444e-07,
"loss": 0.014039421081542968,
"step": 10400
},
{
"epoch": 8.646179401993356,
"grad_norm": 0.019374903291463852,
"learning_rate": 9.472444488131061e-07,
"loss": 0.012065500766038895,
"step": 10410
},
{
"epoch": 8.654485049833887,
"grad_norm": 0.004422264639288187,
"learning_rate": 9.358486649974363e-07,
"loss": 0.02042807936668396,
"step": 10420
},
{
"epoch": 8.662790697674419,
"grad_norm": 0.004711292218416929,
"learning_rate": 9.245184790444727e-07,
"loss": 6.644466193392873e-05,
"step": 10430
},
{
"epoch": 8.67109634551495,
"grad_norm": 0.03553232550621033,
"learning_rate": 9.1325397295152e-07,
"loss": 0.015925948321819306,
"step": 10440
},
{
"epoch": 8.679401993355482,
"grad_norm": 1.0422600507736206,
"learning_rate": 9.020552282405581e-07,
"loss": 0.016211824119091035,
"step": 10450
},
{
"epoch": 8.687707641196013,
"grad_norm": 0.06379421800374985,
"learning_rate": 8.909223259576371e-07,
"loss": 0.0005402436945587397,
"step": 10460
},
{
"epoch": 8.696013289036545,
"grad_norm": 0.0011933192145079374,
"learning_rate": 8.798553466723092e-07,
"loss": 0.0007861870341002941,
"step": 10470
},
{
"epoch": 8.704318936877076,
"grad_norm": 0.0077102952636778355,
"learning_rate": 8.688543704770402e-07,
"loss": 9.638404008001089e-05,
"step": 10480
},
{
"epoch": 8.712624584717608,
"grad_norm": 1.5011637210845947,
"learning_rate": 8.579194769866161e-07,
"loss": 0.027115678787231444,
"step": 10490
},
{
"epoch": 8.720930232558139,
"grad_norm": 1.135060429573059,
"learning_rate": 8.470507453375876e-07,
"loss": 0.019812630116939546,
"step": 10500
},
{
"epoch": 8.720930232558139,
"eval_brier": 0.972,
"eval_c@1": 0.97,
"eval_f05u": 0.98,
"eval_f1": 0.971,
"eval_loss": 0.13822266459465027,
"eval_mean": 0.977,
"eval_roc-auc": 0.992,
"eval_runtime": 69.3853,
"eval_samples_per_second": 265.518,
"eval_steps_per_second": 4.151,
"step": 10500
},
{
"epoch": 8.72923588039867,
"grad_norm": 0.016085628420114517,
"learning_rate": 8.362482541876827e-07,
"loss": 0.029357168078422546,
"step": 10510
},
{
"epoch": 8.737541528239202,
"grad_norm": 0.0010006368393078446,
"learning_rate": 8.255120817152451e-07,
"loss": 0.011326415091753006,
"step": 10520
},
{
"epoch": 8.745847176079733,
"grad_norm": 0.0009730742312967777,
"learning_rate": 8.148423056186672e-07,
"loss": 0.0004337939899414778,
"step": 10530
},
{
"epoch": 8.754152823920267,
"grad_norm": 0.18613412976264954,
"learning_rate": 8.042390031158198e-07,
"loss": 0.023708954453468323,
"step": 10540
},
{
"epoch": 8.762458471760798,
"grad_norm": 3.4708197116851807,
"learning_rate": 7.937022509435089e-07,
"loss": 0.017179006338119508,
"step": 10550
},
{
"epoch": 8.77076411960133,
"grad_norm": 1.296779990196228,
"learning_rate": 7.832321253569041e-07,
"loss": 0.01114661544561386,
"step": 10560
},
{
"epoch": 8.779069767441861,
"grad_norm": 0.03814178332686424,
"learning_rate": 7.72828702128997e-07,
"loss": 0.001047772355377674,
"step": 10570
},
{
"epoch": 8.787375415282392,
"grad_norm": 0.0013747934717684984,
"learning_rate": 7.624920565500505e-07,
"loss": 0.005552845820784569,
"step": 10580
},
{
"epoch": 8.795681063122924,
"grad_norm": 22.074169158935547,
"learning_rate": 7.522222634270527e-07,
"loss": 0.028385037183761598,
"step": 10590
},
{
"epoch": 8.803986710963455,
"grad_norm": 0.024553779512643814,
"learning_rate": 7.420193970831746e-07,
"loss": 0.023264481127262114,
"step": 10600
},
{
"epoch": 8.812292358803987,
"grad_norm": 0.014030910097062588,
"learning_rate": 7.318835313572336e-07,
"loss": 0.0120232954621315,
"step": 10610
},
{
"epoch": 8.820598006644518,
"grad_norm": 0.0007124674157239497,
"learning_rate": 7.218147396031638e-07,
"loss": 0.00033702084328979254,
"step": 10620
},
{
"epoch": 8.82890365448505,
"grad_norm": 0.0013316874392330647,
"learning_rate": 7.118130946894753e-07,
"loss": 7.000082405284047e-05,
"step": 10630
},
{
"epoch": 8.837209302325581,
"grad_norm": 0.40317463874816895,
"learning_rate": 7.018786689987322e-07,
"loss": 0.0012300459668040276,
"step": 10640
},
{
"epoch": 8.845514950166113,
"grad_norm": 0.7749882340431213,
"learning_rate": 6.920115344270328e-07,
"loss": 0.01664312481880188,
"step": 10650
},
{
"epoch": 8.853820598006644,
"grad_norm": 0.26123353838920593,
"learning_rate": 6.822117623834789e-07,
"loss": 0.03449989557266235,
"step": 10660
},
{
"epoch": 8.862126245847175,
"grad_norm": 0.0007906881510280073,
"learning_rate": 6.724794237896748e-07,
"loss": 0.002543853782117367,
"step": 10670
},
{
"epoch": 8.870431893687707,
"grad_norm": 0.0005801045917905867,
"learning_rate": 6.628145890791937e-07,
"loss": 6.418936536647379e-05,
"step": 10680
},
{
"epoch": 8.878737541528238,
"grad_norm": 0.0938602015376091,
"learning_rate": 6.532173281970844e-07,
"loss": 0.0023986753076314926,
"step": 10690
},
{
"epoch": 8.887043189368772,
"grad_norm": 0.04829617962241173,
"learning_rate": 6.436877105993622e-07,
"loss": 0.0014003468677401543,
"step": 10700
},
{
"epoch": 8.895348837209303,
"grad_norm": 0.09765971451997757,
"learning_rate": 6.342258052524941e-07,
"loss": 0.0063482411205768585,
"step": 10710
},
{
"epoch": 8.903654485049834,
"grad_norm": 13.843339920043945,
"learning_rate": 6.24831680632918e-07,
"loss": 0.003428398817777634,
"step": 10720
},
{
"epoch": 8.911960132890366,
"grad_norm": 0.00944607239216566,
"learning_rate": 6.155054047265385e-07,
"loss": 0.00011292543495073914,
"step": 10730
},
{
"epoch": 8.920265780730897,
"grad_norm": 0.0007181317778304219,
"learning_rate": 6.062470450282243e-07,
"loss": 0.015462943911552429,
"step": 10740
},
{
"epoch": 8.928571428571429,
"grad_norm": 0.21927116811275482,
"learning_rate": 5.970566685413426e-07,
"loss": 0.02101568877696991,
"step": 10750
},
{
"epoch": 8.93687707641196,
"grad_norm": 0.0008117577526718378,
"learning_rate": 5.879343417772532e-07,
"loss": 0.013798975944519043,
"step": 10760
},
{
"epoch": 8.945182724252492,
"grad_norm": 0.0008244899217970669,
"learning_rate": 5.788801307548375e-07,
"loss": 0.016404911875724792,
"step": 10770
},
{
"epoch": 8.953488372093023,
"grad_norm": 0.0020234824623912573,
"learning_rate": 5.698941010000214e-07,
"loss": 0.0004574351944029331,
"step": 10780
},
{
"epoch": 8.961794019933555,
"grad_norm": 4.830206871032715,
"learning_rate": 5.609763175452965e-07,
"loss": 0.004152967035770417,
"step": 10790
},
{
"epoch": 8.970099667774086,
"grad_norm": 0.0008019914384931326,
"learning_rate": 5.521268449292505e-07,
"loss": 0.0052707776427268985,
"step": 10800
},
{
"epoch": 8.978405315614618,
"grad_norm": 2.695244789123535,
"learning_rate": 5.433457471961023e-07,
"loss": 0.02068837285041809,
"step": 10810
},
{
"epoch": 8.986710963455149,
"grad_norm": 0.0008493319619446993,
"learning_rate": 5.346330878952355e-07,
"loss": 0.003194114938378334,
"step": 10820
},
{
"epoch": 8.99501661129568,
"grad_norm": 0.0016333531821146607,
"learning_rate": 5.259889300807442e-07,
"loss": 9.348376770503818e-05,
"step": 10830
},
{
"epoch": 9.003322259136212,
"grad_norm": 0.017966577783226967,
"learning_rate": 5.174133363109702e-07,
"loss": 0.00023921006359159947,
"step": 10840
},
{
"epoch": 9.011627906976743,
"grad_norm": 1.7474874258041382,
"learning_rate": 5.089063686480511e-07,
"loss": 0.021680545806884766,
"step": 10850
},
{
"epoch": 9.019933554817277,
"grad_norm": 1.3520294427871704,
"learning_rate": 5.004680886574742e-07,
"loss": 0.009152665734291077,
"step": 10860
},
{
"epoch": 9.028239202657808,
"grad_norm": 0.006455077324062586,
"learning_rate": 4.92098557407632e-07,
"loss": 0.0037008613348007203,
"step": 10870
},
{
"epoch": 9.03654485049834,
"grad_norm": 0.0024395848158746958,
"learning_rate": 4.837978354693751e-07,
"loss": 6.576470332220197e-05,
"step": 10880
},
{
"epoch": 9.044850498338871,
"grad_norm": 0.10780571401119232,
"learning_rate": 4.7556598291557565e-07,
"loss": 0.00496029257774353,
"step": 10890
},
{
"epoch": 9.053156146179402,
"grad_norm": 0.0077397264540195465,
"learning_rate": 4.6740305932069795e-07,
"loss": 0.004276130720973015,
"step": 10900
},
{
"epoch": 9.061461794019934,
"grad_norm": 0.07076343148946762,
"learning_rate": 4.5930912376035643e-07,
"loss": 0.0017954776063561439,
"step": 10910
},
{
"epoch": 9.069767441860465,
"grad_norm": 2.265702962875366,
"learning_rate": 4.5128423481090033e-07,
"loss": 0.023255904018878937,
"step": 10920
},
{
"epoch": 9.078073089700997,
"grad_norm": 0.0012375566875562072,
"learning_rate": 4.43328450548981e-07,
"loss": 5.862920661456883e-05,
"step": 10930
},
{
"epoch": 9.086378737541528,
"grad_norm": 0.040299899876117706,
"learning_rate": 4.3544182855113416e-07,
"loss": 0.0005702963098883629,
"step": 10940
},
{
"epoch": 9.09468438538206,
"grad_norm": 0.12448187172412872,
"learning_rate": 4.276244258933693e-07,
"loss": 0.002195780538022518,
"step": 10950
},
{
"epoch": 9.102990033222591,
"grad_norm": 0.6855204701423645,
"learning_rate": 4.198762991507399e-07,
"loss": 0.02160533666610718,
"step": 10960
},
{
"epoch": 9.111295681063122,
"grad_norm": 0.30196768045425415,
"learning_rate": 4.1219750439695505e-07,
"loss": 0.006671889126300812,
"step": 10970
},
{
"epoch": 9.119601328903654,
"grad_norm": 0.0010014506988227367,
"learning_rate": 4.04588097203964e-07,
"loss": 0.0006922510918229818,
"step": 10980
},
{
"epoch": 9.127906976744185,
"grad_norm": 0.37651070952415466,
"learning_rate": 3.970481326415432e-07,
"loss": 0.0008104128763079643,
"step": 10990
},
{
"epoch": 9.136212624584717,
"grad_norm": 0.005356124602258205,
"learning_rate": 3.8957766527691786e-07,
"loss": 0.011119528859853744,
"step": 11000
},
{
"epoch": 9.136212624584717,
"eval_brier": 0.973,
"eval_c@1": 0.971,
"eval_f05u": 0.981,
"eval_f1": 0.973,
"eval_loss": 0.1350637674331665,
"eval_mean": 0.978,
"eval_roc-auc": 0.992,
"eval_runtime": 69.4523,
"eval_samples_per_second": 265.261,
"eval_steps_per_second": 4.147,
"step": 11000
},
{
"epoch": 9.144518272425248,
"grad_norm": 0.02880888432264328,
"learning_rate": 3.8217674917435555e-07,
"loss": 0.0016112534329295158,
"step": 11010
},
{
"epoch": 9.152823920265782,
"grad_norm": 0.01123111229389906,
"learning_rate": 3.74845437894773e-07,
"loss": 0.0017155883833765984,
"step": 11020
},
{
"epoch": 9.161129568106313,
"grad_norm": 0.0031563404481858015,
"learning_rate": 3.675837844953578e-07,
"loss": 5.774711025878787e-05,
"step": 11030
},
{
"epoch": 9.169435215946844,
"grad_norm": 0.06368283182382584,
"learning_rate": 3.6039184152917406e-07,
"loss": 0.0003356437664479017,
"step": 11040
},
{
"epoch": 9.177740863787376,
"grad_norm": 0.03403898701071739,
"learning_rate": 3.532696610447883e-07,
"loss": 0.018103402853012086,
"step": 11050
},
{
"epoch": 9.186046511627907,
"grad_norm": 0.5248129963874817,
"learning_rate": 3.4621729458589324e-07,
"loss": 0.00858813375234604,
"step": 11060
},
{
"epoch": 9.194352159468439,
"grad_norm": 0.07681553065776825,
"learning_rate": 3.39234793190929e-07,
"loss": 0.016493289172649382,
"step": 11070
},
{
"epoch": 9.20265780730897,
"grad_norm": 0.0010311717633157969,
"learning_rate": 3.3232220739272234e-07,
"loss": 6.358857499435543e-05,
"step": 11080
},
{
"epoch": 9.210963455149502,
"grad_norm": 0.003700330387800932,
"learning_rate": 3.254795872181138e-07,
"loss": 0.011542771011590958,
"step": 11090
},
{
"epoch": 9.219269102990033,
"grad_norm": 0.000969013839494437,
"learning_rate": 3.1870698218759655e-07,
"loss": 0.001096019335091114,
"step": 11100
},
{
"epoch": 9.227574750830565,
"grad_norm": 0.01006294321268797,
"learning_rate": 3.120044413149614e-07,
"loss": 0.011244919896125794,
"step": 11110
},
{
"epoch": 9.235880398671096,
"grad_norm": 0.02737296000123024,
"learning_rate": 3.0537201310694155e-07,
"loss": 0.0015948859974741936,
"step": 11120
},
{
"epoch": 9.244186046511627,
"grad_norm": 0.00350496219471097,
"learning_rate": 2.9880974556285804e-07,
"loss": 0.0034240052103996278,
"step": 11130
},
{
"epoch": 9.252491694352159,
"grad_norm": 0.44182488322257996,
"learning_rate": 2.9231768617427267e-07,
"loss": 0.010205399245023727,
"step": 11140
},
{
"epoch": 9.26079734219269,
"grad_norm": 0.04421304166316986,
"learning_rate": 2.858958819246538e-07,
"loss": 0.02851686179637909,
"step": 11150
},
{
"epoch": 9.269102990033222,
"grad_norm": 0.104521743953228,
"learning_rate": 2.7954437928901736e-07,
"loss": 0.011275745183229446,
"step": 11160
},
{
"epoch": 9.277408637873755,
"grad_norm": 0.0011418936774134636,
"learning_rate": 2.7326322423361195e-07,
"loss": 0.012401804327964783,
"step": 11170
},
{
"epoch": 9.285714285714286,
"grad_norm": 0.0012771759647876024,
"learning_rate": 2.6705246221557124e-07,
"loss": 0.0012776992283761502,
"step": 11180
},
{
"epoch": 9.294019933554818,
"grad_norm": 0.02292698249220848,
"learning_rate": 2.6091213818259076e-07,
"loss": 0.0002426925115287304,
"step": 11190
},
{
"epoch": 9.30232558139535,
"grad_norm": 1.425937533378601,
"learning_rate": 2.548422965726016e-07,
"loss": 0.009351170808076858,
"step": 11200
},
{
"epoch": 9.31063122923588,
"grad_norm": 0.13734589517116547,
"learning_rate": 2.488429813134485e-07,
"loss": 0.0028193186968564986,
"step": 11210
},
{
"epoch": 9.318936877076412,
"grad_norm": 0.8627015948295593,
"learning_rate": 2.4291423582257223e-07,
"loss": 0.0022154562175273895,
"step": 11220
},
{
"epoch": 9.327242524916944,
"grad_norm": 0.006032244302332401,
"learning_rate": 2.370561030066998e-07,
"loss": 0.004900708794593811,
"step": 11230
},
{
"epoch": 9.335548172757475,
"grad_norm": 0.044071223586797714,
"learning_rate": 2.3126862526152038e-07,
"loss": 0.011845049262046815,
"step": 11240
},
{
"epoch": 9.343853820598007,
"grad_norm": 1.2974224090576172,
"learning_rate": 2.2555184447139668e-07,
"loss": 0.022230133414268494,
"step": 11250
},
{
"epoch": 9.352159468438538,
"grad_norm": 0.11188158392906189,
"learning_rate": 2.1990580200904944e-07,
"loss": 0.004698529839515686,
"step": 11260
},
{
"epoch": 9.36046511627907,
"grad_norm": 0.1672002673149109,
"learning_rate": 2.14330538735259e-07,
"loss": 0.0015591046772897243,
"step": 11270
},
{
"epoch": 9.368770764119601,
"grad_norm": 1.4730030298233032,
"learning_rate": 2.0882609499857653e-07,
"loss": 0.009047257155179978,
"step": 11280
},
{
"epoch": 9.377076411960132,
"grad_norm": 0.06319593638181686,
"learning_rate": 2.0339251063502541e-07,
"loss": 0.015721666812896728,
"step": 11290
},
{
"epoch": 9.385382059800664,
"grad_norm": 1.6366090774536133,
"learning_rate": 1.9802982496781366e-07,
"loss": 0.009605227410793305,
"step": 11300
},
{
"epoch": 9.393687707641195,
"grad_norm": 0.07986780256032944,
"learning_rate": 1.9273807680705193e-07,
"loss": 0.026491966843605042,
"step": 11310
},
{
"epoch": 9.401993355481727,
"grad_norm": 0.12056667357683182,
"learning_rate": 1.8751730444947381e-07,
"loss": 0.012095175683498383,
"step": 11320
},
{
"epoch": 9.41029900332226,
"grad_norm": 0.0025509411934763193,
"learning_rate": 1.8236754567815374e-07,
"loss": 0.00012878261040896177,
"step": 11330
},
{
"epoch": 9.418604651162791,
"grad_norm": 0.1650812178850174,
"learning_rate": 1.7728883776223503e-07,
"loss": 0.011297813802957534,
"step": 11340
},
{
"epoch": 9.426910299003323,
"grad_norm": 3.9979779720306396,
"learning_rate": 1.722812174566635e-07,
"loss": 0.009340720623731614,
"step": 11350
},
{
"epoch": 9.435215946843854,
"grad_norm": 0.1160222515463829,
"learning_rate": 1.6734472100191746e-07,
"loss": 0.01668257862329483,
"step": 11360
},
{
"epoch": 9.443521594684386,
"grad_norm": 0.16997136175632477,
"learning_rate": 1.6247938412375043e-07,
"loss": 0.010060159862041474,
"step": 11370
},
{
"epoch": 9.451827242524917,
"grad_norm": 0.004492861684411764,
"learning_rate": 1.5768524203292558e-07,
"loss": 0.011055050790309906,
"step": 11380
},
{
"epoch": 9.460132890365449,
"grad_norm": 2.8961880207061768,
"learning_rate": 1.5296232942496603e-07,
"loss": 0.008695921301841736,
"step": 11390
},
{
"epoch": 9.46843853820598,
"grad_norm": 1.773732304573059,
"learning_rate": 1.4831068047990503e-07,
"loss": 0.01720489114522934,
"step": 11400
},
{
"epoch": 9.476744186046512,
"grad_norm": 1.9357893466949463,
"learning_rate": 1.4373032886203065e-07,
"loss": 0.007002674788236618,
"step": 11410
},
{
"epoch": 9.485049833887043,
"grad_norm": 0.07272736728191376,
"learning_rate": 1.3922130771965248e-07,
"loss": 0.0043906763195991514,
"step": 11420
},
{
"epoch": 9.493355481727574,
"grad_norm": 0.0014025394339114428,
"learning_rate": 1.3478364968485536e-07,
"loss": 5.8618158800527456e-05,
"step": 11430
},
{
"epoch": 9.501661129568106,
"grad_norm": 0.008486457169055939,
"learning_rate": 1.3041738687326055e-07,
"loss": 0.00012943275505676864,
"step": 11440
},
{
"epoch": 9.509966777408637,
"grad_norm": 0.045450642704963684,
"learning_rate": 1.2612255088380488e-07,
"loss": 0.0035385094583034516,
"step": 11450
},
{
"epoch": 9.518272425249169,
"grad_norm": 0.018507055938243866,
"learning_rate": 1.218991727984975e-07,
"loss": 0.0034021966159343718,
"step": 11460
},
{
"epoch": 9.5265780730897,
"grad_norm": 0.0010659305844455957,
"learning_rate": 1.1774728318220574e-07,
"loss": 0.013690489530563354,
"step": 11470
},
{
"epoch": 9.534883720930232,
"grad_norm": 0.0011029911693185568,
"learning_rate": 1.1366691208243186e-07,
"loss": 0.004185367003083229,
"step": 11480
},
{
"epoch": 9.543189368770765,
"grad_norm": 0.008363594301044941,
"learning_rate": 1.0965808902908659e-07,
"loss": 0.000700355926528573,
"step": 11490
},
{
"epoch": 9.551495016611296,
"grad_norm": 0.010572860017418861,
"learning_rate": 1.0572084303429153e-07,
"loss": 0.001556174550205469,
"step": 11500
},
{
"epoch": 9.551495016611296,
"eval_brier": 0.973,
"eval_c@1": 0.971,
"eval_f05u": 0.982,
"eval_f1": 0.973,
"eval_loss": 0.13495969772338867,
"eval_mean": 0.978,
"eval_roc-auc": 0.992,
"eval_runtime": 69.5097,
"eval_samples_per_second": 265.042,
"eval_steps_per_second": 4.143,
"step": 11500
},
{
"epoch": 9.559800664451828,
"grad_norm": 0.036029886454343796,
"learning_rate": 1.0185520259215487e-07,
"loss": 0.0023395881056785584,
"step": 11510
},
{
"epoch": 9.56810631229236,
"grad_norm": 0.10781078785657883,
"learning_rate": 9.806119567857153e-08,
"loss": 0.0003543411614373326,
"step": 11520
},
{
"epoch": 9.57641196013289,
"grad_norm": 0.001593824359588325,
"learning_rate": 9.433884975102226e-08,
"loss": 5.926559097133577e-05,
"step": 11530
},
{
"epoch": 9.584717607973422,
"grad_norm": 0.013292987830936909,
"learning_rate": 9.06881917483693e-08,
"loss": 0.00044385655783116815,
"step": 11540
},
{
"epoch": 9.593023255813954,
"grad_norm": 0.000703663972672075,
"learning_rate": 8.710924809066546e-08,
"loss": 0.0030287588015198708,
"step": 11550
},
{
"epoch": 9.601328903654485,
"grad_norm": 0.022231318056583405,
"learning_rate": 8.360204467896427e-08,
"loss": 0.017375482618808745,
"step": 11560
},
{
"epoch": 9.609634551495017,
"grad_norm": 2.488351583480835,
"learning_rate": 8.01666068951279e-08,
"loss": 0.002250361256301403,
"step": 11570
},
{
"epoch": 9.617940199335548,
"grad_norm": 0.017226671800017357,
"learning_rate": 7.68029596016484e-08,
"loss": 0.0012241391465067863,
"step": 11580
},
{
"epoch": 9.62624584717608,
"grad_norm": 0.15982404351234436,
"learning_rate": 7.351112714146125e-08,
"loss": 0.0005029088817536831,
"step": 11590
},
{
"epoch": 9.634551495016611,
"grad_norm": 0.04742132127285004,
"learning_rate": 7.029113333777649e-08,
"loss": 0.02843661904335022,
"step": 11600
},
{
"epoch": 9.642857142857142,
"grad_norm": 0.05109667778015137,
"learning_rate": 6.71430014939034e-08,
"loss": 0.013497960567474366,
"step": 11610
},
{
"epoch": 9.651162790697674,
"grad_norm": 0.005027054343372583,
"learning_rate": 6.406675439307952e-08,
"loss": 0.0016751263290643692,
"step": 11620
},
{
"epoch": 9.659468438538205,
"grad_norm": 0.0021046719048172235,
"learning_rate": 6.106241429830739e-08,
"loss": 0.007513735443353653,
"step": 11630
},
{
"epoch": 9.667774086378738,
"grad_norm": 2.6134467124938965,
"learning_rate": 5.813000295219695e-08,
"loss": 0.003876357898116112,
"step": 11640
},
{
"epoch": 9.67607973421927,
"grad_norm": 5.531118392944336,
"learning_rate": 5.526954157680342e-08,
"loss": 0.010698853433132172,
"step": 11650
},
{
"epoch": 9.684385382059801,
"grad_norm": 0.0008058390812948346,
"learning_rate": 5.2481050873471884e-08,
"loss": 0.001887832023203373,
"step": 11660
},
{
"epoch": 9.692691029900333,
"grad_norm": 0.5099664926528931,
"learning_rate": 4.976455102269628e-08,
"loss": 0.00881224349141121,
"step": 11670
},
{
"epoch": 9.700996677740864,
"grad_norm": 0.005670701619237661,
"learning_rate": 4.712006168396621e-08,
"loss": 7.717470289207995e-05,
"step": 11680
},
{
"epoch": 9.709302325581396,
"grad_norm": 8.067323684692383,
"learning_rate": 4.45476019956248e-08,
"loss": 0.012497730553150177,
"step": 11690
},
{
"epoch": 9.717607973421927,
"grad_norm": 0.1126004233956337,
"learning_rate": 4.2047190574733276e-08,
"loss": 0.0005513341166079044,
"step": 11700
},
{
"epoch": 9.725913621262459,
"grad_norm": 0.034886494278907776,
"learning_rate": 3.961884551693329e-08,
"loss": 0.014814093708992004,
"step": 11710
},
{
"epoch": 9.73421926910299,
"grad_norm": 0.051566146314144135,
"learning_rate": 3.726258439632036e-08,
"loss": 0.014201661944389344,
"step": 11720
},
{
"epoch": 9.742524916943522,
"grad_norm": 0.0011892920592799783,
"learning_rate": 3.497842426531173e-08,
"loss": 6.91100605763495e-05,
"step": 11730
},
{
"epoch": 9.750830564784053,
"grad_norm": 0.0008599160355515778,
"learning_rate": 3.276638165452539e-08,
"loss": 0.004660710692405701,
"step": 11740
},
{
"epoch": 9.759136212624584,
"grad_norm": 0.01840898208320141,
"learning_rate": 3.062647257265905e-08,
"loss": 0.010904261469841003,
"step": 11750
},
{
"epoch": 9.767441860465116,
"grad_norm": 0.003409146796911955,
"learning_rate": 2.8558712506377983e-08,
"loss": 0.0009562863036990165,
"step": 11760
},
{
"epoch": 9.775747508305647,
"grad_norm": 0.020226627588272095,
"learning_rate": 2.6563116420196268e-08,
"loss": 0.02036612331867218,
"step": 11770
},
{
"epoch": 9.784053156146179,
"grad_norm": 0.0018576461588963866,
"learning_rate": 2.463969875637573e-08,
"loss": 6.618564948439598e-05,
"step": 11780
},
{
"epoch": 9.79235880398671,
"grad_norm": 0.037959206849336624,
"learning_rate": 2.278847343481827e-08,
"loss": 0.0003085340140387416,
"step": 11790
},
{
"epoch": 9.800664451827242,
"grad_norm": 0.03762241080403328,
"learning_rate": 2.1009453852962603e-08,
"loss": 0.014483200013637542,
"step": 11800
},
{
"epoch": 9.808970099667775,
"grad_norm": 0.1814233958721161,
"learning_rate": 1.930265288569211e-08,
"loss": 0.01672428399324417,
"step": 11810
},
{
"epoch": 9.817275747508306,
"grad_norm": 0.0013242058921605349,
"learning_rate": 1.7668082885237137e-08,
"loss": 0.03881125748157501,
"step": 11820
},
{
"epoch": 9.825581395348838,
"grad_norm": 0.0018302457174286246,
"learning_rate": 1.610575568109063e-08,
"loss": 0.00010088413255289197,
"step": 11830
},
{
"epoch": 9.83388704318937,
"grad_norm": 0.011404566466808319,
"learning_rate": 1.4615682579915968e-08,
"loss": 0.0028445499017834665,
"step": 11840
},
{
"epoch": 9.8421926910299,
"grad_norm": 0.2529490888118744,
"learning_rate": 1.3197874365472596e-08,
"loss": 0.031901615858078006,
"step": 11850
},
{
"epoch": 9.850498338870432,
"grad_norm": 1.3335745334625244,
"learning_rate": 1.1852341298530522e-08,
"loss": 0.027505967020988464,
"step": 11860
},
{
"epoch": 9.858803986710964,
"grad_norm": 0.8897879719734192,
"learning_rate": 1.0579093116802607e-08,
"loss": 0.017324808239936828,
"step": 11870
},
{
"epoch": 9.867109634551495,
"grad_norm": 0.0015733162872493267,
"learning_rate": 9.37813903486906e-09,
"loss": 6.080536986701191e-05,
"step": 11880
},
{
"epoch": 9.875415282392026,
"grad_norm": 0.07180562615394592,
"learning_rate": 8.249487744115269e-09,
"loss": 0.0021126432344317435,
"step": 11890
},
{
"epoch": 9.883720930232558,
"grad_norm": 1.0551310777664185,
"learning_rate": 7.193147412665192e-09,
"loss": 0.004447254538536072,
"step": 11900
},
{
"epoch": 9.89202657807309,
"grad_norm": 0.05793992429971695,
"learning_rate": 6.209125685323614e-09,
"loss": 0.016958166658878327,
"step": 11910
},
{
"epoch": 9.90033222591362,
"grad_norm": 0.029679453000426292,
"learning_rate": 5.29742968352065e-09,
"loss": 0.0010822580195963382,
"step": 11920
},
{
"epoch": 9.908637873754152,
"grad_norm": 0.0013658910756930709,
"learning_rate": 4.458066005259554e-09,
"loss": 7.04719393979758e-05,
"step": 11930
},
{
"epoch": 9.916943521594684,
"grad_norm": 0.03027878887951374,
"learning_rate": 3.6910407250723145e-09,
"loss": 0.00042849513702094557,
"step": 11940
},
{
"epoch": 9.925249169435215,
"grad_norm": 0.004070840310305357,
"learning_rate": 2.996359393971915e-09,
"loss": 0.0011183662340044976,
"step": 11950
},
{
"epoch": 9.933554817275748,
"grad_norm": 0.0029106398578733206,
"learning_rate": 2.374027039411253e-09,
"loss": 0.009823792427778245,
"step": 11960
},
{
"epoch": 9.94186046511628,
"grad_norm": 2.4805774688720703,
"learning_rate": 1.8240481652509467e-09,
"loss": 0.010463573783636094,
"step": 11970
},
{
"epoch": 9.950166112956811,
"grad_norm": 0.008324177004396915,
"learning_rate": 1.3464267517260266e-09,
"loss": 0.0014796962030231952,
"step": 11980
},
{
"epoch": 9.958471760797343,
"grad_norm": 1.4040313959121704,
"learning_rate": 9.411662554137391e-10,
"loss": 0.0008210563100874424,
"step": 11990
},
{
"epoch": 9.966777408637874,
"grad_norm": 0.18574158847332,
"learning_rate": 6.082696092102325e-10,
"loss": 0.0029134985059499742,
"step": 12000
},
{
"epoch": 9.966777408637874,
"eval_brier": 0.973,
"eval_c@1": 0.971,
"eval_f05u": 0.982,
"eval_f1": 0.973,
"eval_loss": 0.13517948985099792,
"eval_mean": 0.978,
"eval_roc-auc": 0.992,
"eval_runtime": 69.4363,
"eval_samples_per_second": 265.322,
"eval_steps_per_second": 4.148,
"step": 12000
}
],
"logging_steps": 10,
"max_steps": 12040,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 3.671051704468599e+17,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}