wangrongsheng's picture
commit from root
8da6a42
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.6506886454831363,
"global_step": 6000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.999996372615291e-05,
"loss": 1.4261,
"step": 10
},
{
"epoch": 0.0,
"learning_rate": 4.999985490471688e-05,
"loss": 1.3752,
"step": 20
},
{
"epoch": 0.0,
"learning_rate": 4.9999673536007715e-05,
"loss": 1.3193,
"step": 30
},
{
"epoch": 0.0,
"learning_rate": 4.999941962055173e-05,
"loss": 1.2661,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 4.9999093159085765e-05,
"loss": 1.2613,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 4.999869415255717e-05,
"loss": 1.2233,
"step": 60
},
{
"epoch": 0.01,
"learning_rate": 4.9998222602123825e-05,
"loss": 1.2184,
"step": 70
},
{
"epoch": 0.01,
"learning_rate": 4.9997678509154145e-05,
"loss": 1.2031,
"step": 80
},
{
"epoch": 0.01,
"learning_rate": 4.999706187522701e-05,
"loss": 1.1744,
"step": 90
},
{
"epoch": 0.01,
"learning_rate": 4.9996372702131866e-05,
"loss": 1.1756,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 4.9995610991868615e-05,
"loss": 1.1611,
"step": 110
},
{
"epoch": 0.01,
"learning_rate": 4.9994776746647656e-05,
"loss": 1.1384,
"step": 120
},
{
"epoch": 0.01,
"learning_rate": 4.999386996888991e-05,
"loss": 1.1275,
"step": 130
},
{
"epoch": 0.02,
"learning_rate": 4.999289066122676e-05,
"loss": 1.1227,
"step": 140
},
{
"epoch": 0.02,
"learning_rate": 4.999183882650006e-05,
"loss": 1.1135,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 4.999071446776215e-05,
"loss": 1.1253,
"step": 160
},
{
"epoch": 0.02,
"learning_rate": 4.998951758827579e-05,
"loss": 1.11,
"step": 170
},
{
"epoch": 0.02,
"learning_rate": 4.998824819151424e-05,
"loss": 1.0942,
"step": 180
},
{
"epoch": 0.02,
"learning_rate": 4.998690628116115e-05,
"loss": 1.0963,
"step": 190
},
{
"epoch": 0.02,
"learning_rate": 4.998549186111065e-05,
"loss": 1.0676,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 4.998400493546723e-05,
"loss": 1.0891,
"step": 210
},
{
"epoch": 0.02,
"learning_rate": 4.998244550854582e-05,
"loss": 1.0612,
"step": 220
},
{
"epoch": 0.02,
"learning_rate": 4.9980813584871734e-05,
"loss": 1.0726,
"step": 230
},
{
"epoch": 0.03,
"learning_rate": 4.9979109169180664e-05,
"loss": 1.0721,
"step": 240
},
{
"epoch": 0.03,
"learning_rate": 4.997733226641867e-05,
"loss": 1.0757,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 4.9975482881742155e-05,
"loss": 1.0606,
"step": 260
},
{
"epoch": 0.03,
"learning_rate": 4.9973561020517854e-05,
"loss": 1.0624,
"step": 270
},
{
"epoch": 0.03,
"learning_rate": 4.9971566688322854e-05,
"loss": 1.0503,
"step": 280
},
{
"epoch": 0.03,
"learning_rate": 4.9969499890944505e-05,
"loss": 1.0376,
"step": 290
},
{
"epoch": 0.03,
"learning_rate": 4.996736063438047e-05,
"loss": 1.0402,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 4.996514892483866e-05,
"loss": 1.0362,
"step": 310
},
{
"epoch": 0.03,
"learning_rate": 4.996286476873727e-05,
"loss": 1.0363,
"step": 320
},
{
"epoch": 0.04,
"learning_rate": 4.996050817270471e-05,
"loss": 1.0354,
"step": 330
},
{
"epoch": 0.04,
"learning_rate": 4.9958079143579594e-05,
"loss": 1.0437,
"step": 340
},
{
"epoch": 0.04,
"learning_rate": 4.9955577688410744e-05,
"loss": 1.0223,
"step": 350
},
{
"epoch": 0.04,
"learning_rate": 4.9953003814457154e-05,
"loss": 1.017,
"step": 360
},
{
"epoch": 0.04,
"learning_rate": 4.995035752918797e-05,
"loss": 1.013,
"step": 370
},
{
"epoch": 0.04,
"learning_rate": 4.994763884028246e-05,
"loss": 1.0177,
"step": 380
},
{
"epoch": 0.04,
"learning_rate": 4.994484775563002e-05,
"loss": 1.0445,
"step": 390
},
{
"epoch": 0.04,
"learning_rate": 4.9941984283330115e-05,
"loss": 1.0078,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 4.993904843169227e-05,
"loss": 1.0033,
"step": 410
},
{
"epoch": 0.05,
"learning_rate": 4.993604020923607e-05,
"loss": 1.0108,
"step": 420
},
{
"epoch": 0.05,
"learning_rate": 4.993295962469109e-05,
"loss": 1.0085,
"step": 430
},
{
"epoch": 0.05,
"learning_rate": 4.99298066869969e-05,
"loss": 1.0115,
"step": 440
},
{
"epoch": 0.05,
"learning_rate": 4.992658140530304e-05,
"loss": 0.9952,
"step": 450
},
{
"epoch": 0.05,
"learning_rate": 4.992328378896898e-05,
"loss": 1.0234,
"step": 460
},
{
"epoch": 0.05,
"learning_rate": 4.991991384756409e-05,
"loss": 0.9995,
"step": 470
},
{
"epoch": 0.05,
"learning_rate": 4.991647159086764e-05,
"loss": 0.9844,
"step": 480
},
{
"epoch": 0.05,
"learning_rate": 4.991295702886874e-05,
"loss": 0.9989,
"step": 490
},
{
"epoch": 0.05,
"learning_rate": 4.990937017176632e-05,
"loss": 1.0023,
"step": 500
},
{
"epoch": 0.06,
"learning_rate": 4.990571102996911e-05,
"loss": 0.9729,
"step": 510
},
{
"epoch": 0.06,
"learning_rate": 4.990197961409561e-05,
"loss": 0.9793,
"step": 520
},
{
"epoch": 0.06,
"learning_rate": 4.989817593497404e-05,
"loss": 0.9876,
"step": 530
},
{
"epoch": 0.06,
"learning_rate": 4.9894300003642304e-05,
"loss": 0.9883,
"step": 540
},
{
"epoch": 0.06,
"learning_rate": 4.989035183134802e-05,
"loss": 0.9997,
"step": 550
},
{
"epoch": 0.06,
"learning_rate": 4.9886331429548434e-05,
"loss": 0.9998,
"step": 560
},
{
"epoch": 0.06,
"learning_rate": 4.988223880991035e-05,
"loss": 0.988,
"step": 570
},
{
"epoch": 0.06,
"learning_rate": 4.9878073984310194e-05,
"loss": 1.0108,
"step": 580
},
{
"epoch": 0.06,
"learning_rate": 4.987383696483389e-05,
"loss": 0.9929,
"step": 590
},
{
"epoch": 0.07,
"learning_rate": 4.987126010519375e-05,
"loss": 1.0691,
"step": 600
},
{
"epoch": 0.07,
"learning_rate": 4.9866907601177285e-05,
"loss": 0.9875,
"step": 610
},
{
"epoch": 0.07,
"learning_rate": 4.986248293568848e-05,
"loss": 0.9738,
"step": 620
},
{
"epoch": 0.07,
"learning_rate": 4.9857986121567325e-05,
"loss": 0.9682,
"step": 630
},
{
"epoch": 0.07,
"learning_rate": 4.985341717186316e-05,
"loss": 0.9955,
"step": 640
},
{
"epoch": 0.07,
"learning_rate": 4.984877609983465e-05,
"loss": 0.9865,
"step": 650
},
{
"epoch": 0.07,
"learning_rate": 4.984406291894975e-05,
"loss": 0.981,
"step": 660
},
{
"epoch": 0.07,
"learning_rate": 4.9839277642885694e-05,
"loss": 0.9725,
"step": 670
},
{
"epoch": 0.07,
"learning_rate": 4.9834420285528904e-05,
"loss": 0.9681,
"step": 680
},
{
"epoch": 0.07,
"learning_rate": 4.9829490860974975e-05,
"loss": 0.9632,
"step": 690
},
{
"epoch": 0.08,
"learning_rate": 4.982448938352865e-05,
"loss": 0.9877,
"step": 700
},
{
"epoch": 0.08,
"learning_rate": 4.981941586770376e-05,
"loss": 0.9808,
"step": 710
},
{
"epoch": 0.08,
"learning_rate": 4.9814270328223166e-05,
"loss": 0.9677,
"step": 720
},
{
"epoch": 0.08,
"learning_rate": 4.980905278001876e-05,
"loss": 0.9699,
"step": 730
},
{
"epoch": 0.08,
"learning_rate": 4.980376323823139e-05,
"loss": 0.9496,
"step": 740
},
{
"epoch": 0.08,
"learning_rate": 4.97984017182108e-05,
"loss": 0.9669,
"step": 750
},
{
"epoch": 0.08,
"learning_rate": 4.979296823551565e-05,
"loss": 0.9784,
"step": 760
},
{
"epoch": 0.08,
"learning_rate": 4.978746280591338e-05,
"loss": 0.9626,
"step": 770
},
{
"epoch": 0.08,
"learning_rate": 4.978188544538026e-05,
"loss": 0.9695,
"step": 780
},
{
"epoch": 0.09,
"learning_rate": 4.977623617010127e-05,
"loss": 0.9575,
"step": 790
},
{
"epoch": 0.09,
"learning_rate": 4.977051499647008e-05,
"loss": 0.9506,
"step": 800
},
{
"epoch": 0.09,
"learning_rate": 4.9764721941089016e-05,
"loss": 0.9489,
"step": 810
},
{
"epoch": 0.09,
"learning_rate": 4.975885702076898e-05,
"loss": 0.9454,
"step": 820
},
{
"epoch": 0.09,
"learning_rate": 4.975292025252945e-05,
"loss": 0.9427,
"step": 830
},
{
"epoch": 0.09,
"learning_rate": 4.974691165359836e-05,
"loss": 0.944,
"step": 840
},
{
"epoch": 0.09,
"learning_rate": 4.974083124141211e-05,
"loss": 0.9147,
"step": 850
},
{
"epoch": 0.09,
"learning_rate": 4.9734679033615505e-05,
"loss": 0.9487,
"step": 860
},
{
"epoch": 0.09,
"learning_rate": 4.9728455048061686e-05,
"loss": 0.9595,
"step": 870
},
{
"epoch": 0.1,
"learning_rate": 4.9722159302812074e-05,
"loss": 0.9522,
"step": 880
},
{
"epoch": 0.1,
"learning_rate": 4.971579181613635e-05,
"loss": 0.9575,
"step": 890
},
{
"epoch": 0.1,
"learning_rate": 4.970935260651237e-05,
"loss": 0.9474,
"step": 900
},
{
"epoch": 0.1,
"learning_rate": 4.9702841692626125e-05,
"loss": 0.9533,
"step": 910
},
{
"epoch": 0.1,
"learning_rate": 4.969625909337169e-05,
"loss": 0.9674,
"step": 920
},
{
"epoch": 0.1,
"learning_rate": 4.968960482785116e-05,
"loss": 0.9196,
"step": 930
},
{
"epoch": 0.1,
"learning_rate": 4.96828789153746e-05,
"loss": 0.9326,
"step": 940
},
{
"epoch": 0.1,
"learning_rate": 4.967608137545998e-05,
"loss": 0.9546,
"step": 950
},
{
"epoch": 0.1,
"learning_rate": 4.966921222783315e-05,
"loss": 0.937,
"step": 960
},
{
"epoch": 0.11,
"learning_rate": 4.966227149242773e-05,
"loss": 0.9298,
"step": 970
},
{
"epoch": 0.11,
"learning_rate": 4.96552591893851e-05,
"loss": 0.944,
"step": 980
},
{
"epoch": 0.11,
"learning_rate": 4.9648175339054314e-05,
"loss": 0.9715,
"step": 990
},
{
"epoch": 0.11,
"learning_rate": 4.964101996199205e-05,
"loss": 0.9465,
"step": 1000
},
{
"epoch": 0.11,
"learning_rate": 4.963379307896255e-05,
"loss": 0.9526,
"step": 1010
},
{
"epoch": 0.11,
"learning_rate": 4.9626494710937575e-05,
"loss": 0.9149,
"step": 1020
},
{
"epoch": 0.11,
"learning_rate": 4.96191248790963e-05,
"loss": 0.9608,
"step": 1030
},
{
"epoch": 0.11,
"learning_rate": 4.961168360482531e-05,
"loss": 0.9236,
"step": 1040
},
{
"epoch": 0.11,
"learning_rate": 4.9604170909718494e-05,
"loss": 0.9455,
"step": 1050
},
{
"epoch": 0.11,
"learning_rate": 4.9596586815577e-05,
"loss": 0.9637,
"step": 1060
},
{
"epoch": 0.12,
"learning_rate": 4.958893134440915e-05,
"loss": 0.9477,
"step": 1070
},
{
"epoch": 0.12,
"learning_rate": 4.958120451843045e-05,
"loss": 0.9275,
"step": 1080
},
{
"epoch": 0.12,
"learning_rate": 4.957340636006341e-05,
"loss": 0.946,
"step": 1090
},
{
"epoch": 0.12,
"learning_rate": 4.956553689193758e-05,
"loss": 0.9226,
"step": 1100
},
{
"epoch": 0.12,
"learning_rate": 4.9557596136889425e-05,
"loss": 0.9463,
"step": 1110
},
{
"epoch": 0.12,
"learning_rate": 4.954958411796229e-05,
"loss": 0.9514,
"step": 1120
},
{
"epoch": 0.12,
"learning_rate": 4.95415008584063e-05,
"loss": 0.9169,
"step": 1130
},
{
"epoch": 0.12,
"learning_rate": 4.9533346381678346e-05,
"loss": 0.9345,
"step": 1140
},
{
"epoch": 0.12,
"learning_rate": 4.952512071144195e-05,
"loss": 0.9313,
"step": 1150
},
{
"epoch": 0.13,
"learning_rate": 4.951682387156727e-05,
"loss": 0.931,
"step": 1160
},
{
"epoch": 0.13,
"learning_rate": 4.950845588613096e-05,
"loss": 0.9292,
"step": 1170
},
{
"epoch": 0.13,
"learning_rate": 4.9500016779416124e-05,
"loss": 0.9442,
"step": 1180
},
{
"epoch": 0.13,
"learning_rate": 4.9491506575912306e-05,
"loss": 0.9307,
"step": 1190
},
{
"epoch": 0.13,
"learning_rate": 4.9482925300315296e-05,
"loss": 0.9473,
"step": 1200
},
{
"epoch": 0.13,
"learning_rate": 4.9474272977527194e-05,
"loss": 0.9279,
"step": 1210
},
{
"epoch": 0.13,
"learning_rate": 4.946554963265623e-05,
"loss": 0.9433,
"step": 1220
},
{
"epoch": 0.13,
"learning_rate": 4.9456755291016744e-05,
"loss": 0.9286,
"step": 1230
},
{
"epoch": 0.13,
"learning_rate": 4.9447889978129105e-05,
"loss": 0.9307,
"step": 1240
},
{
"epoch": 0.14,
"learning_rate": 4.943895371971963e-05,
"loss": 0.9207,
"step": 1250
},
{
"epoch": 0.14,
"learning_rate": 4.942994654172053e-05,
"loss": 0.9067,
"step": 1260
},
{
"epoch": 0.14,
"learning_rate": 4.942086847026979e-05,
"loss": 0.9253,
"step": 1270
},
{
"epoch": 0.14,
"learning_rate": 4.941171953171115e-05,
"loss": 0.9072,
"step": 1280
},
{
"epoch": 0.14,
"learning_rate": 4.9402499752593975e-05,
"loss": 0.9051,
"step": 1290
},
{
"epoch": 0.14,
"learning_rate": 4.9393209159673217e-05,
"loss": 0.9076,
"step": 1300
},
{
"epoch": 0.14,
"learning_rate": 4.938384777990932e-05,
"loss": 0.9214,
"step": 1310
},
{
"epoch": 0.14,
"learning_rate": 4.937441564046814e-05,
"loss": 0.9249,
"step": 1320
},
{
"epoch": 0.14,
"learning_rate": 4.936491276872089e-05,
"loss": 0.9379,
"step": 1330
},
{
"epoch": 0.15,
"learning_rate": 4.935533919224401e-05,
"loss": 0.933,
"step": 1340
},
{
"epoch": 0.15,
"learning_rate": 4.9345694938819146e-05,
"loss": 0.9228,
"step": 1350
},
{
"epoch": 0.15,
"learning_rate": 4.933598003643303e-05,
"loss": 0.9327,
"step": 1360
},
{
"epoch": 0.15,
"learning_rate": 4.9326194513277415e-05,
"loss": 0.9198,
"step": 1370
},
{
"epoch": 0.15,
"learning_rate": 4.931633839774898e-05,
"loss": 0.9349,
"step": 1380
},
{
"epoch": 0.15,
"learning_rate": 4.930641171844926e-05,
"loss": 0.9296,
"step": 1390
},
{
"epoch": 0.15,
"learning_rate": 4.929641450418459e-05,
"loss": 0.9283,
"step": 1400
},
{
"epoch": 0.15,
"learning_rate": 4.9286346783965925e-05,
"loss": 0.9015,
"step": 1410
},
{
"epoch": 0.15,
"learning_rate": 4.9276208587008885e-05,
"loss": 0.9227,
"step": 1420
},
{
"epoch": 0.16,
"learning_rate": 4.926599994273358e-05,
"loss": 0.9059,
"step": 1430
},
{
"epoch": 0.16,
"learning_rate": 4.925572088076454e-05,
"loss": 0.9038,
"step": 1440
},
{
"epoch": 0.16,
"learning_rate": 4.924537143093068e-05,
"loss": 0.9109,
"step": 1450
},
{
"epoch": 0.16,
"learning_rate": 4.923495162326512e-05,
"loss": 0.9094,
"step": 1460
},
{
"epoch": 0.16,
"learning_rate": 4.92244614880052e-05,
"loss": 0.9292,
"step": 1470
},
{
"epoch": 0.16,
"learning_rate": 4.921390105559233e-05,
"loss": 0.9266,
"step": 1480
},
{
"epoch": 0.16,
"learning_rate": 4.9203270356671894e-05,
"loss": 0.8977,
"step": 1490
},
{
"epoch": 0.16,
"learning_rate": 4.91925694220932e-05,
"loss": 0.9151,
"step": 1500
},
{
"epoch": 0.16,
"learning_rate": 4.918179828290939e-05,
"loss": 0.9118,
"step": 1510
},
{
"epoch": 0.16,
"learning_rate": 4.91709569703773e-05,
"loss": 0.9259,
"step": 1520
},
{
"epoch": 0.17,
"learning_rate": 4.9160045515957435e-05,
"loss": 0.9295,
"step": 1530
},
{
"epoch": 0.17,
"learning_rate": 4.914906395131381e-05,
"loss": 0.9152,
"step": 1540
},
{
"epoch": 0.17,
"learning_rate": 4.913801230831392e-05,
"loss": 0.9187,
"step": 1550
},
{
"epoch": 0.17,
"learning_rate": 4.9126890619028623e-05,
"loss": 0.9371,
"step": 1560
},
{
"epoch": 0.17,
"learning_rate": 4.9115698915732025e-05,
"loss": 0.909,
"step": 1570
},
{
"epoch": 0.17,
"learning_rate": 4.9104437230901424e-05,
"loss": 0.9044,
"step": 1580
},
{
"epoch": 0.17,
"learning_rate": 4.909310559721718e-05,
"loss": 0.9031,
"step": 1590
},
{
"epoch": 0.17,
"learning_rate": 4.908170404756267e-05,
"loss": 0.9221,
"step": 1600
},
{
"epoch": 0.17,
"learning_rate": 4.9070232615024115e-05,
"loss": 0.9227,
"step": 1610
},
{
"epoch": 0.18,
"learning_rate": 4.905869133289056e-05,
"loss": 0.928,
"step": 1620
},
{
"epoch": 0.18,
"learning_rate": 4.904708023465375e-05,
"loss": 0.8813,
"step": 1630
},
{
"epoch": 0.18,
"learning_rate": 4.903539935400803e-05,
"loss": 0.8944,
"step": 1640
},
{
"epoch": 0.18,
"learning_rate": 4.9023648724850214e-05,
"loss": 0.8901,
"step": 1650
},
{
"epoch": 0.18,
"learning_rate": 4.901182838127955e-05,
"loss": 0.923,
"step": 1660
},
{
"epoch": 0.18,
"learning_rate": 4.8999938357597594e-05,
"loss": 0.9103,
"step": 1670
},
{
"epoch": 0.18,
"learning_rate": 4.89879786883081e-05,
"loss": 0.9013,
"step": 1680
},
{
"epoch": 0.18,
"learning_rate": 4.8975949408116914e-05,
"loss": 0.8921,
"step": 1690
},
{
"epoch": 0.18,
"learning_rate": 4.8963850551931904e-05,
"loss": 0.9103,
"step": 1700
},
{
"epoch": 0.19,
"learning_rate": 4.8951682154862836e-05,
"loss": 0.9057,
"step": 1710
},
{
"epoch": 0.19,
"learning_rate": 4.893944425222128e-05,
"loss": 0.9005,
"step": 1720
},
{
"epoch": 0.19,
"learning_rate": 4.892713687952048e-05,
"loss": 0.8894,
"step": 1730
},
{
"epoch": 0.19,
"learning_rate": 4.8914760072475316e-05,
"loss": 0.9067,
"step": 1740
},
{
"epoch": 0.19,
"learning_rate": 4.890231386700214e-05,
"loss": 0.9206,
"step": 1750
},
{
"epoch": 0.19,
"learning_rate": 4.888979829921869e-05,
"loss": 0.9143,
"step": 1760
},
{
"epoch": 0.19,
"learning_rate": 4.887721340544398e-05,
"loss": 0.8992,
"step": 1770
},
{
"epoch": 0.19,
"learning_rate": 4.886455922219822e-05,
"loss": 0.9068,
"step": 1780
},
{
"epoch": 0.19,
"learning_rate": 4.885183578620268e-05,
"loss": 0.8978,
"step": 1790
},
{
"epoch": 0.2,
"learning_rate": 4.8839043134379604e-05,
"loss": 0.9198,
"step": 1800
},
{
"epoch": 0.2,
"learning_rate": 4.8826181303852075e-05,
"loss": 0.9042,
"step": 1810
},
{
"epoch": 0.2,
"learning_rate": 4.881325033194395e-05,
"loss": 0.8947,
"step": 1820
},
{
"epoch": 0.2,
"learning_rate": 4.880025025617971e-05,
"loss": 0.9095,
"step": 1830
},
{
"epoch": 0.2,
"learning_rate": 4.878718111428438e-05,
"loss": 0.9077,
"step": 1840
},
{
"epoch": 0.2,
"learning_rate": 4.8774042944183404e-05,
"loss": 0.9092,
"step": 1850
},
{
"epoch": 0.2,
"learning_rate": 4.876083578400255e-05,
"loss": 0.894,
"step": 1860
},
{
"epoch": 0.2,
"learning_rate": 4.874755967206776e-05,
"loss": 0.9111,
"step": 1870
},
{
"epoch": 0.2,
"learning_rate": 4.8734214646905086e-05,
"loss": 0.8819,
"step": 1880
},
{
"epoch": 0.2,
"learning_rate": 4.872080074724058e-05,
"loss": 0.8981,
"step": 1890
},
{
"epoch": 0.21,
"learning_rate": 4.870731801200012e-05,
"loss": 0.8966,
"step": 1900
},
{
"epoch": 0.21,
"learning_rate": 4.869376648030938e-05,
"loss": 0.8935,
"step": 1910
},
{
"epoch": 0.21,
"learning_rate": 4.868014619149364e-05,
"loss": 0.9013,
"step": 1920
},
{
"epoch": 0.21,
"learning_rate": 4.866645718507772e-05,
"loss": 0.9019,
"step": 1930
},
{
"epoch": 0.21,
"learning_rate": 4.865269950078587e-05,
"loss": 0.913,
"step": 1940
},
{
"epoch": 0.21,
"learning_rate": 4.863887317854161e-05,
"loss": 0.893,
"step": 1950
},
{
"epoch": 0.21,
"learning_rate": 4.8624978258467655e-05,
"loss": 0.8725,
"step": 1960
},
{
"epoch": 0.21,
"learning_rate": 4.861101478088578e-05,
"loss": 0.9065,
"step": 1970
},
{
"epoch": 0.21,
"learning_rate": 4.859698278631671e-05,
"loss": 0.8794,
"step": 1980
},
{
"epoch": 0.22,
"learning_rate": 4.858288231547999e-05,
"loss": 0.8948,
"step": 1990
},
{
"epoch": 0.22,
"learning_rate": 4.8568713409293906e-05,
"loss": 0.9128,
"step": 2000
},
{
"epoch": 0.22,
"learning_rate": 4.85544761088753e-05,
"loss": 0.9076,
"step": 2010
},
{
"epoch": 0.22,
"learning_rate": 4.8540170455539514e-05,
"loss": 0.8837,
"step": 2020
},
{
"epoch": 0.22,
"learning_rate": 4.8525796490800235e-05,
"loss": 0.8786,
"step": 2030
},
{
"epoch": 0.22,
"learning_rate": 4.851135425636937e-05,
"loss": 0.8859,
"step": 2040
},
{
"epoch": 0.22,
"learning_rate": 4.849684379415696e-05,
"loss": 0.8995,
"step": 2050
},
{
"epoch": 0.22,
"learning_rate": 4.8482265146271036e-05,
"loss": 0.873,
"step": 2060
},
{
"epoch": 0.22,
"learning_rate": 4.846761835501747e-05,
"loss": 0.9225,
"step": 2070
},
{
"epoch": 0.23,
"learning_rate": 4.8452903462899914e-05,
"loss": 0.9137,
"step": 2080
},
{
"epoch": 0.23,
"learning_rate": 4.8438120512619624e-05,
"loss": 0.9071,
"step": 2090
},
{
"epoch": 0.23,
"learning_rate": 4.842326954707536e-05,
"loss": 0.8829,
"step": 2100
},
{
"epoch": 0.23,
"learning_rate": 4.840835060936325e-05,
"loss": 0.8959,
"step": 2110
},
{
"epoch": 0.23,
"learning_rate": 4.8393363742776685e-05,
"loss": 0.9036,
"step": 2120
},
{
"epoch": 0.23,
"learning_rate": 4.837830899080616e-05,
"loss": 0.8982,
"step": 2130
},
{
"epoch": 0.23,
"learning_rate": 4.836318639713918e-05,
"loss": 0.8851,
"step": 2140
},
{
"epoch": 0.23,
"learning_rate": 4.834799600566011e-05,
"loss": 0.8718,
"step": 2150
},
{
"epoch": 0.23,
"learning_rate": 4.8332737860450085e-05,
"loss": 0.886,
"step": 2160
},
{
"epoch": 0.24,
"learning_rate": 4.831741200578681e-05,
"loss": 0.8869,
"step": 2170
},
{
"epoch": 0.24,
"learning_rate": 4.830201848614452e-05,
"loss": 0.8877,
"step": 2180
},
{
"epoch": 0.24,
"learning_rate": 4.828655734619377e-05,
"loss": 0.8794,
"step": 2190
},
{
"epoch": 0.24,
"learning_rate": 4.827102863080138e-05,
"loss": 0.8879,
"step": 2200
},
{
"epoch": 0.24,
"learning_rate": 4.825543238503025e-05,
"loss": 0.8873,
"step": 2210
},
{
"epoch": 0.24,
"learning_rate": 4.8239768654139236e-05,
"loss": 0.8939,
"step": 2220
},
{
"epoch": 0.24,
"learning_rate": 4.822403748358305e-05,
"loss": 0.8906,
"step": 2230
},
{
"epoch": 0.24,
"learning_rate": 4.8208238919012094e-05,
"loss": 0.8935,
"step": 2240
},
{
"epoch": 0.24,
"learning_rate": 4.819237300627234e-05,
"loss": 0.8671,
"step": 2250
},
{
"epoch": 0.25,
"learning_rate": 4.817643979140521e-05,
"loss": 0.8884,
"step": 2260
},
{
"epoch": 0.25,
"learning_rate": 4.8160439320647426e-05,
"loss": 0.8919,
"step": 2270
},
{
"epoch": 0.25,
"learning_rate": 4.814437164043087e-05,
"loss": 0.8709,
"step": 2280
},
{
"epoch": 0.25,
"learning_rate": 4.8128236797382474e-05,
"loss": 0.9036,
"step": 2290
},
{
"epoch": 0.25,
"learning_rate": 4.811203483832406e-05,
"loss": 0.884,
"step": 2300
},
{
"epoch": 0.25,
"learning_rate": 4.809576581027223e-05,
"loss": 0.8862,
"step": 2310
},
{
"epoch": 0.25,
"learning_rate": 4.8079429760438186e-05,
"loss": 0.8852,
"step": 2320
},
{
"epoch": 0.25,
"learning_rate": 4.8063026736227656e-05,
"loss": 0.9042,
"step": 2330
},
{
"epoch": 0.25,
"learning_rate": 4.804655678524068e-05,
"loss": 0.8793,
"step": 2340
},
{
"epoch": 0.25,
"learning_rate": 4.8030019955271565e-05,
"loss": 0.876,
"step": 2350
},
{
"epoch": 0.26,
"learning_rate": 4.8013416294308645e-05,
"loss": 0.8822,
"step": 2360
},
{
"epoch": 0.26,
"learning_rate": 4.799674585053422e-05,
"loss": 0.8813,
"step": 2370
},
{
"epoch": 0.26,
"learning_rate": 4.798000867232438e-05,
"loss": 0.8957,
"step": 2380
},
{
"epoch": 0.26,
"learning_rate": 4.796320480824887e-05,
"loss": 0.8813,
"step": 2390
},
{
"epoch": 0.26,
"learning_rate": 4.794633430707096e-05,
"loss": 0.8893,
"step": 2400
},
{
"epoch": 0.26,
"learning_rate": 4.792939721774728e-05,
"loss": 0.8887,
"step": 2410
},
{
"epoch": 0.26,
"learning_rate": 4.791239358942771e-05,
"loss": 0.8673,
"step": 2420
},
{
"epoch": 0.26,
"learning_rate": 4.78953234714552e-05,
"loss": 0.8823,
"step": 2430
},
{
"epoch": 0.26,
"learning_rate": 4.7878186913365676e-05,
"loss": 0.8682,
"step": 2440
},
{
"epoch": 0.27,
"learning_rate": 4.7860983964887826e-05,
"loss": 0.8696,
"step": 2450
},
{
"epoch": 0.27,
"learning_rate": 4.784371467594304e-05,
"loss": 0.892,
"step": 2460
},
{
"epoch": 0.27,
"learning_rate": 4.782637909664519e-05,
"loss": 0.8895,
"step": 2470
},
{
"epoch": 0.27,
"learning_rate": 4.7808977277300535e-05,
"loss": 0.8871,
"step": 2480
},
{
"epoch": 0.27,
"learning_rate": 4.779150926840754e-05,
"loss": 0.8726,
"step": 2490
},
{
"epoch": 0.27,
"learning_rate": 4.7773975120656765e-05,
"loss": 0.8728,
"step": 2500
},
{
"epoch": 0.27,
"learning_rate": 4.775637488493069e-05,
"loss": 0.8856,
"step": 2510
},
{
"epoch": 0.27,
"learning_rate": 4.7738708612303576e-05,
"loss": 0.8781,
"step": 2520
},
{
"epoch": 0.27,
"learning_rate": 4.77209763540413e-05,
"loss": 0.8933,
"step": 2530
},
{
"epoch": 0.28,
"learning_rate": 4.7703178161601256e-05,
"loss": 0.8646,
"step": 2540
},
{
"epoch": 0.28,
"learning_rate": 4.768531408663216e-05,
"loss": 0.8702,
"step": 2550
},
{
"epoch": 0.28,
"learning_rate": 4.7667384180973895e-05,
"loss": 0.8642,
"step": 2560
},
{
"epoch": 0.28,
"learning_rate": 4.764938849665741e-05,
"loss": 0.8651,
"step": 2570
},
{
"epoch": 0.28,
"learning_rate": 4.763132708590451e-05,
"loss": 0.8674,
"step": 2580
},
{
"epoch": 0.28,
"learning_rate": 4.761320000112775e-05,
"loss": 0.8818,
"step": 2590
},
{
"epoch": 0.28,
"learning_rate": 4.759500729493025e-05,
"loss": 0.8849,
"step": 2600
},
{
"epoch": 0.28,
"learning_rate": 4.757674902010557e-05,
"loss": 0.8696,
"step": 2610
},
{
"epoch": 0.28,
"learning_rate": 4.755842522963755e-05,
"loss": 0.8895,
"step": 2620
},
{
"epoch": 0.29,
"learning_rate": 4.7540035976700117e-05,
"loss": 0.88,
"step": 2630
},
{
"epoch": 0.29,
"learning_rate": 4.7521581314657206e-05,
"loss": 0.8869,
"step": 2640
},
{
"epoch": 0.29,
"learning_rate": 4.7503061297062535e-05,
"loss": 0.9043,
"step": 2650
},
{
"epoch": 0.29,
"learning_rate": 4.748447597765949e-05,
"loss": 0.8806,
"step": 2660
},
{
"epoch": 0.29,
"learning_rate": 4.746582541038095e-05,
"loss": 0.9031,
"step": 2670
},
{
"epoch": 0.29,
"learning_rate": 4.7447109649349145e-05,
"loss": 0.8778,
"step": 2680
},
{
"epoch": 0.29,
"learning_rate": 4.7428328748875494e-05,
"loss": 0.8805,
"step": 2690
},
{
"epoch": 0.29,
"learning_rate": 4.740948276346042e-05,
"loss": 0.876,
"step": 2700
},
{
"epoch": 0.29,
"learning_rate": 4.739057174779325e-05,
"loss": 0.8669,
"step": 2710
},
{
"epoch": 0.29,
"learning_rate": 4.737159575675201e-05,
"loss": 0.873,
"step": 2720
},
{
"epoch": 0.3,
"learning_rate": 4.7352554845403255e-05,
"loss": 0.8608,
"step": 2730
},
{
"epoch": 0.3,
"learning_rate": 4.733344906900197e-05,
"loss": 0.8711,
"step": 2740
},
{
"epoch": 0.3,
"learning_rate": 4.731427848299136e-05,
"loss": 0.8768,
"step": 2750
},
{
"epoch": 0.3,
"learning_rate": 4.729504314300268e-05,
"loss": 0.8557,
"step": 2760
},
{
"epoch": 0.3,
"learning_rate": 4.727574310485513e-05,
"loss": 0.8937,
"step": 2770
},
{
"epoch": 0.3,
"learning_rate": 4.725637842455563e-05,
"loss": 0.8685,
"step": 2780
},
{
"epoch": 0.3,
"learning_rate": 4.7236949158298703e-05,
"loss": 0.8662,
"step": 2790
},
{
"epoch": 0.3,
"learning_rate": 4.721745536246628e-05,
"loss": 0.8777,
"step": 2800
},
{
"epoch": 0.3,
"learning_rate": 4.719789709362756e-05,
"loss": 0.881,
"step": 2810
},
{
"epoch": 0.31,
"learning_rate": 4.717827440853885e-05,
"loss": 0.8807,
"step": 2820
},
{
"epoch": 0.31,
"learning_rate": 4.7158587364143344e-05,
"loss": 0.8744,
"step": 2830
},
{
"epoch": 0.31,
"learning_rate": 4.713883601757105e-05,
"loss": 0.8703,
"step": 2840
},
{
"epoch": 0.31,
"learning_rate": 4.711902042613856e-05,
"loss": 0.8672,
"step": 2850
},
{
"epoch": 0.31,
"learning_rate": 4.7099140647348874e-05,
"loss": 0.8811,
"step": 2860
},
{
"epoch": 0.31,
"learning_rate": 4.7079196738891286e-05,
"loss": 0.8776,
"step": 2870
},
{
"epoch": 0.31,
"learning_rate": 4.705918875864117e-05,
"loss": 0.8625,
"step": 2880
},
{
"epoch": 0.31,
"learning_rate": 4.703911676465986e-05,
"loss": 0.8663,
"step": 2890
},
{
"epoch": 0.31,
"learning_rate": 4.7018980815194405e-05,
"loss": 0.8666,
"step": 2900
},
{
"epoch": 0.32,
"learning_rate": 4.699878096867749e-05,
"loss": 0.8653,
"step": 2910
},
{
"epoch": 0.32,
"learning_rate": 4.697851728372721e-05,
"loss": 0.8744,
"step": 2920
},
{
"epoch": 0.32,
"learning_rate": 4.695818981914689e-05,
"loss": 0.8739,
"step": 2930
},
{
"epoch": 0.32,
"learning_rate": 4.693779863392498e-05,
"loss": 0.8724,
"step": 2940
},
{
"epoch": 0.32,
"learning_rate": 4.69173437872348e-05,
"loss": 0.865,
"step": 2950
},
{
"epoch": 0.32,
"learning_rate": 4.689682533843445e-05,
"loss": 0.8774,
"step": 2960
},
{
"epoch": 0.32,
"learning_rate": 4.6876243347066554e-05,
"loss": 0.8527,
"step": 2970
},
{
"epoch": 0.32,
"learning_rate": 4.685559787285816e-05,
"loss": 0.8809,
"step": 2980
},
{
"epoch": 0.32,
"learning_rate": 4.683488897572052e-05,
"loss": 0.8898,
"step": 2990
},
{
"epoch": 0.33,
"learning_rate": 4.681411671574898e-05,
"loss": 0.8756,
"step": 3000
},
{
"epoch": 0.33,
"learning_rate": 4.6793281153222676e-05,
"loss": 0.8809,
"step": 3010
},
{
"epoch": 0.33,
"learning_rate": 4.6772382348604524e-05,
"loss": 0.8908,
"step": 3020
},
{
"epoch": 0.33,
"learning_rate": 4.675142036254091e-05,
"loss": 0.847,
"step": 3030
},
{
"epoch": 0.33,
"learning_rate": 4.673039525586159e-05,
"loss": 0.8743,
"step": 3040
},
{
"epoch": 0.33,
"learning_rate": 4.6709307089579476e-05,
"loss": 0.8658,
"step": 3050
},
{
"epoch": 0.33,
"learning_rate": 4.668815592489049e-05,
"loss": 0.8778,
"step": 3060
},
{
"epoch": 0.33,
"learning_rate": 4.666694182317336e-05,
"loss": 0.8531,
"step": 3070
},
{
"epoch": 0.33,
"learning_rate": 4.664566484598946e-05,
"loss": 0.8645,
"step": 3080
},
{
"epoch": 0.34,
"learning_rate": 4.6624325055082595e-05,
"loss": 0.8646,
"step": 3090
},
{
"epoch": 0.34,
"learning_rate": 4.660292251237889e-05,
"loss": 0.8622,
"step": 3100
},
{
"epoch": 0.34,
"learning_rate": 4.658145727998653e-05,
"loss": 0.8676,
"step": 3110
},
{
"epoch": 0.34,
"learning_rate": 4.655992942019566e-05,
"loss": 0.8538,
"step": 3120
},
{
"epoch": 0.34,
"learning_rate": 4.653833899547814e-05,
"loss": 0.8445,
"step": 3130
},
{
"epoch": 0.34,
"learning_rate": 4.651668606848737e-05,
"loss": 0.8674,
"step": 3140
},
{
"epoch": 0.34,
"learning_rate": 4.6494970702058175e-05,
"loss": 0.8492,
"step": 3150
},
{
"epoch": 0.34,
"learning_rate": 4.6473192959206536e-05,
"loss": 0.8583,
"step": 3160
},
{
"epoch": 0.34,
"learning_rate": 4.645135290312944e-05,
"loss": 0.8727,
"step": 3170
},
{
"epoch": 0.34,
"learning_rate": 4.6429450597204736e-05,
"loss": 0.8369,
"step": 3180
},
{
"epoch": 0.35,
"learning_rate": 4.6407486104990887e-05,
"loss": 0.8857,
"step": 3190
},
{
"epoch": 0.35,
"learning_rate": 4.638545949022682e-05,
"loss": 0.8415,
"step": 3200
},
{
"epoch": 0.35,
"learning_rate": 4.6363370816831746e-05,
"loss": 0.8503,
"step": 3210
},
{
"epoch": 0.35,
"learning_rate": 4.6341220148904954e-05,
"loss": 0.8828,
"step": 3220
},
{
"epoch": 0.35,
"learning_rate": 4.631900755072564e-05,
"loss": 0.8664,
"step": 3230
},
{
"epoch": 0.35,
"learning_rate": 4.629673308675271e-05,
"loss": 0.8692,
"step": 3240
},
{
"epoch": 0.35,
"learning_rate": 4.627439682162462e-05,
"loss": 0.8622,
"step": 3250
},
{
"epoch": 0.35,
"learning_rate": 4.625199882015913e-05,
"loss": 0.8896,
"step": 3260
},
{
"epoch": 0.35,
"learning_rate": 4.6229539147353194e-05,
"loss": 0.8451,
"step": 3270
},
{
"epoch": 0.36,
"learning_rate": 4.62070178683827e-05,
"loss": 0.8462,
"step": 3280
},
{
"epoch": 0.36,
"learning_rate": 4.618443504860232e-05,
"loss": 0.8581,
"step": 3290
},
{
"epoch": 0.36,
"learning_rate": 4.616179075354532e-05,
"loss": 0.8679,
"step": 3300
},
{
"epoch": 0.36,
"learning_rate": 4.613908504892336e-05,
"loss": 0.8572,
"step": 3310
},
{
"epoch": 0.36,
"learning_rate": 4.611631800062629e-05,
"loss": 0.8519,
"step": 3320
},
{
"epoch": 0.36,
"learning_rate": 4.609348967472198e-05,
"loss": 0.8737,
"step": 3330
},
{
"epoch": 0.36,
"learning_rate": 4.6070600137456147e-05,
"loss": 0.868,
"step": 3340
},
{
"epoch": 0.36,
"learning_rate": 4.6047649455252105e-05,
"loss": 0.8675,
"step": 3350
},
{
"epoch": 0.36,
"learning_rate": 4.6024637694710614e-05,
"loss": 0.8649,
"step": 3360
},
{
"epoch": 0.37,
"learning_rate": 4.6001564922609684e-05,
"loss": 0.8622,
"step": 3370
},
{
"epoch": 0.37,
"learning_rate": 4.597843120590438e-05,
"loss": 0.8473,
"step": 3380
},
{
"epoch": 0.37,
"learning_rate": 4.59552366117266e-05,
"loss": 0.8624,
"step": 3390
},
{
"epoch": 0.37,
"learning_rate": 4.593198120738492e-05,
"loss": 0.8668,
"step": 3400
},
{
"epoch": 0.37,
"learning_rate": 4.5908665060364386e-05,
"loss": 0.8687,
"step": 3410
},
{
"epoch": 0.37,
"learning_rate": 4.58852882383263e-05,
"loss": 0.8686,
"step": 3420
},
{
"epoch": 0.37,
"learning_rate": 4.586185080910804e-05,
"loss": 0.8451,
"step": 3430
},
{
"epoch": 0.37,
"learning_rate": 4.583835284072288e-05,
"loss": 0.8617,
"step": 3440
},
{
"epoch": 0.37,
"learning_rate": 4.581479440135973e-05,
"loss": 0.8603,
"step": 3450
},
{
"epoch": 0.38,
"learning_rate": 4.579117555938304e-05,
"loss": 0.8622,
"step": 3460
},
{
"epoch": 0.38,
"learning_rate": 4.576749638333248e-05,
"loss": 0.8588,
"step": 3470
},
{
"epoch": 0.38,
"learning_rate": 4.574375694192286e-05,
"loss": 0.8318,
"step": 3480
},
{
"epoch": 0.38,
"learning_rate": 4.571995730404384e-05,
"loss": 0.8457,
"step": 3490
},
{
"epoch": 0.38,
"learning_rate": 4.569609753875976e-05,
"loss": 0.8571,
"step": 3500
},
{
"epoch": 0.38,
"learning_rate": 4.567217771530949e-05,
"loss": 0.8579,
"step": 3510
},
{
"epoch": 0.38,
"learning_rate": 4.564819790310613e-05,
"loss": 0.8474,
"step": 3520
},
{
"epoch": 0.38,
"learning_rate": 4.562415817173687e-05,
"loss": 0.8635,
"step": 3530
},
{
"epoch": 0.38,
"learning_rate": 4.560005859096282e-05,
"loss": 0.8867,
"step": 3540
},
{
"epoch": 0.38,
"learning_rate": 4.557589923071874e-05,
"loss": 0.8527,
"step": 3550
},
{
"epoch": 0.39,
"learning_rate": 4.555168016111284e-05,
"loss": 0.8579,
"step": 3560
},
{
"epoch": 0.39,
"learning_rate": 4.552740145242665e-05,
"loss": 0.8627,
"step": 3570
},
{
"epoch": 0.39,
"learning_rate": 4.550306317511474e-05,
"loss": 0.8579,
"step": 3580
},
{
"epoch": 0.39,
"learning_rate": 4.547866539980453e-05,
"loss": 0.8648,
"step": 3590
},
{
"epoch": 0.39,
"learning_rate": 4.5454208197296136e-05,
"loss": 0.8592,
"step": 3600
},
{
"epoch": 0.39,
"learning_rate": 4.542969163856208e-05,
"loss": 0.8648,
"step": 3610
},
{
"epoch": 0.39,
"learning_rate": 4.540511579474717e-05,
"loss": 0.8578,
"step": 3620
},
{
"epoch": 0.39,
"learning_rate": 4.538048073716824e-05,
"loss": 0.8463,
"step": 3630
},
{
"epoch": 0.39,
"learning_rate": 4.5355786537313946e-05,
"loss": 0.8689,
"step": 3640
},
{
"epoch": 0.4,
"learning_rate": 4.533103326684458e-05,
"loss": 0.8562,
"step": 3650
},
{
"epoch": 0.4,
"learning_rate": 4.530622099759186e-05,
"loss": 0.8449,
"step": 3660
},
{
"epoch": 0.4,
"learning_rate": 4.528134980155869e-05,
"loss": 0.8482,
"step": 3670
},
{
"epoch": 0.4,
"learning_rate": 4.525641975091899e-05,
"loss": 0.8606,
"step": 3680
},
{
"epoch": 0.4,
"learning_rate": 4.523143091801747e-05,
"loss": 0.8334,
"step": 3690
},
{
"epoch": 0.4,
"learning_rate": 4.520638337536942e-05,
"loss": 0.8408,
"step": 3700
},
{
"epoch": 0.4,
"learning_rate": 4.5181277195660495e-05,
"loss": 0.8431,
"step": 3710
},
{
"epoch": 0.4,
"learning_rate": 4.5156112451746515e-05,
"loss": 0.8542,
"step": 3720
},
{
"epoch": 0.4,
"learning_rate": 4.513088921665324e-05,
"loss": 0.8608,
"step": 3730
},
{
"epoch": 0.41,
"learning_rate": 4.510560756357619e-05,
"loss": 0.8454,
"step": 3740
},
{
"epoch": 0.41,
"learning_rate": 4.508026756588036e-05,
"loss": 0.8771,
"step": 3750
},
{
"epoch": 0.41,
"learning_rate": 4.5054869297100116e-05,
"loss": 0.8584,
"step": 3760
},
{
"epoch": 0.41,
"learning_rate": 4.502941283093888e-05,
"loss": 0.852,
"step": 3770
},
{
"epoch": 0.41,
"learning_rate": 4.500389824126896e-05,
"loss": 0.8551,
"step": 3780
},
{
"epoch": 0.41,
"learning_rate": 4.497832560213135e-05,
"loss": 0.8554,
"step": 3790
},
{
"epoch": 0.41,
"learning_rate": 4.495269498773549e-05,
"loss": 0.8437,
"step": 3800
},
{
"epoch": 0.41,
"learning_rate": 4.492700647245906e-05,
"loss": 0.8695,
"step": 3810
},
{
"epoch": 0.41,
"learning_rate": 4.4901260130847766e-05,
"loss": 0.8498,
"step": 3820
},
{
"epoch": 0.42,
"learning_rate": 4.487545603761511e-05,
"loss": 0.84,
"step": 3830
},
{
"epoch": 0.42,
"learning_rate": 4.484959426764219e-05,
"loss": 0.8524,
"step": 3840
},
{
"epoch": 0.42,
"learning_rate": 4.482367489597748e-05,
"loss": 0.868,
"step": 3850
},
{
"epoch": 0.42,
"learning_rate": 4.4797697997836604e-05,
"loss": 0.8335,
"step": 3860
},
{
"epoch": 0.42,
"learning_rate": 4.477166364860213e-05,
"loss": 0.8438,
"step": 3870
},
{
"epoch": 0.42,
"learning_rate": 4.474557192382333e-05,
"loss": 0.8461,
"step": 3880
},
{
"epoch": 0.42,
"learning_rate": 4.471942289921599e-05,
"loss": 0.8529,
"step": 3890
},
{
"epoch": 0.42,
"learning_rate": 4.469321665066216e-05,
"loss": 0.8669,
"step": 3900
},
{
"epoch": 0.42,
"learning_rate": 4.4666953254209965e-05,
"loss": 0.8252,
"step": 3910
},
{
"epoch": 0.43,
"learning_rate": 4.4640632786073357e-05,
"loss": 0.8377,
"step": 3920
},
{
"epoch": 0.43,
"learning_rate": 4.46142553226319e-05,
"loss": 0.8413,
"step": 3930
},
{
"epoch": 0.43,
"learning_rate": 4.458782094043057e-05,
"loss": 0.8481,
"step": 3940
},
{
"epoch": 0.43,
"learning_rate": 4.45613297161795e-05,
"loss": 0.8455,
"step": 3950
},
{
"epoch": 0.43,
"learning_rate": 4.4534781726753775e-05,
"loss": 0.8496,
"step": 3960
},
{
"epoch": 0.43,
"learning_rate": 4.450817704919322e-05,
"loss": 0.8552,
"step": 3970
},
{
"epoch": 0.43,
"learning_rate": 4.4481515760702155e-05,
"loss": 0.8625,
"step": 3980
},
{
"epoch": 0.43,
"learning_rate": 4.4454797938649174e-05,
"loss": 0.8525,
"step": 3990
},
{
"epoch": 0.43,
"learning_rate": 4.442802366056694e-05,
"loss": 0.8349,
"step": 4000
},
{
"epoch": 0.43,
"learning_rate": 4.440119300415193e-05,
"loss": 0.8429,
"step": 4010
},
{
"epoch": 0.44,
"learning_rate": 4.4374306047264245e-05,
"loss": 0.8593,
"step": 4020
},
{
"epoch": 0.44,
"learning_rate": 4.4347362867927344e-05,
"loss": 0.845,
"step": 4030
},
{
"epoch": 0.44,
"learning_rate": 4.432036354432786e-05,
"loss": 0.8258,
"step": 4040
},
{
"epoch": 0.44,
"learning_rate": 4.4293308154815315e-05,
"loss": 0.8676,
"step": 4050
},
{
"epoch": 0.44,
"learning_rate": 4.426619677790199e-05,
"loss": 0.8197,
"step": 4060
},
{
"epoch": 0.44,
"learning_rate": 4.4239029492262576e-05,
"loss": 0.8299,
"step": 4070
},
{
"epoch": 0.44,
"learning_rate": 4.421180637673403e-05,
"loss": 0.855,
"step": 4080
},
{
"epoch": 0.44,
"learning_rate": 4.418452751031533e-05,
"loss": 0.865,
"step": 4090
},
{
"epoch": 0.44,
"learning_rate": 4.415719297216723e-05,
"loss": 0.8631,
"step": 4100
},
{
"epoch": 0.45,
"learning_rate": 4.4129802841612036e-05,
"loss": 0.8532,
"step": 4110
},
{
"epoch": 0.45,
"learning_rate": 4.410235719813338e-05,
"loss": 0.827,
"step": 4120
},
{
"epoch": 0.45,
"learning_rate": 4.407485612137599e-05,
"loss": 0.8391,
"step": 4130
},
{
"epoch": 0.45,
"learning_rate": 4.404729969114547e-05,
"loss": 0.8395,
"step": 4140
},
{
"epoch": 0.45,
"learning_rate": 4.401968798740801e-05,
"loss": 0.8262,
"step": 4150
},
{
"epoch": 0.45,
"learning_rate": 4.3992021090290235e-05,
"loss": 0.8384,
"step": 4160
},
{
"epoch": 0.45,
"learning_rate": 4.396429908007894e-05,
"loss": 0.8218,
"step": 4170
},
{
"epoch": 0.45,
"learning_rate": 4.393652203722084e-05,
"loss": 0.8292,
"step": 4180
},
{
"epoch": 0.45,
"learning_rate": 4.390869004232234e-05,
"loss": 0.8496,
"step": 4190
},
{
"epoch": 0.46,
"learning_rate": 4.3880803176149334e-05,
"loss": 0.8454,
"step": 4200
},
{
"epoch": 0.46,
"learning_rate": 4.385286151962693e-05,
"loss": 0.8475,
"step": 4210
},
{
"epoch": 0.46,
"learning_rate": 4.3824865153839234e-05,
"loss": 0.838,
"step": 4220
},
{
"epoch": 0.46,
"learning_rate": 4.3796814160029134e-05,
"loss": 0.8539,
"step": 4230
},
{
"epoch": 0.46,
"learning_rate": 4.3768708619598e-05,
"loss": 0.8607,
"step": 4240
},
{
"epoch": 0.46,
"learning_rate": 4.3740548614105545e-05,
"loss": 0.8607,
"step": 4250
},
{
"epoch": 0.46,
"learning_rate": 4.3712334225269494e-05,
"loss": 0.8502,
"step": 4260
},
{
"epoch": 0.46,
"learning_rate": 4.3684065534965405e-05,
"loss": 0.8495,
"step": 4270
},
{
"epoch": 0.46,
"learning_rate": 4.3655742625226404e-05,
"loss": 0.8401,
"step": 4280
},
{
"epoch": 0.47,
"learning_rate": 4.362736557824297e-05,
"loss": 0.8396,
"step": 4290
},
{
"epoch": 0.47,
"learning_rate": 4.3598934476362676e-05,
"loss": 0.8483,
"step": 4300
},
{
"epoch": 0.47,
"learning_rate": 4.357044940208995e-05,
"loss": 0.8655,
"step": 4310
},
{
"epoch": 0.47,
"learning_rate": 4.354191043808585e-05,
"loss": 0.8543,
"step": 4320
},
{
"epoch": 0.47,
"learning_rate": 4.351331766716783e-05,
"loss": 0.8427,
"step": 4330
},
{
"epoch": 0.47,
"learning_rate": 4.348467117230947e-05,
"loss": 0.8545,
"step": 4340
},
{
"epoch": 0.47,
"learning_rate": 4.345597103664024e-05,
"loss": 0.8718,
"step": 4350
},
{
"epoch": 0.47,
"learning_rate": 4.34272173434453e-05,
"loss": 0.8321,
"step": 4360
},
{
"epoch": 0.47,
"learning_rate": 4.339841017616522e-05,
"loss": 0.8458,
"step": 4370
},
{
"epoch": 0.48,
"learning_rate": 4.3369549618395735e-05,
"loss": 0.8447,
"step": 4380
},
{
"epoch": 0.48,
"learning_rate": 4.3340635753887525e-05,
"loss": 0.8323,
"step": 4390
},
{
"epoch": 0.48,
"learning_rate": 4.3311668666545954e-05,
"loss": 0.8239,
"step": 4400
},
{
"epoch": 0.48,
"learning_rate": 4.328264844043084e-05,
"loss": 0.8496,
"step": 4410
},
{
"epoch": 0.48,
"learning_rate": 4.325357515975621e-05,
"loss": 0.82,
"step": 4420
},
{
"epoch": 0.48,
"learning_rate": 4.322444890889004e-05,
"loss": 0.8389,
"step": 4430
},
{
"epoch": 0.48,
"learning_rate": 4.3195269772354e-05,
"loss": 0.8532,
"step": 4440
},
{
"epoch": 0.48,
"learning_rate": 4.316603783482329e-05,
"loss": 0.8407,
"step": 4450
},
{
"epoch": 0.48,
"learning_rate": 4.313675318112627e-05,
"loss": 0.8399,
"step": 4460
},
{
"epoch": 0.48,
"learning_rate": 4.3107415896244316e-05,
"loss": 0.8303,
"step": 4470
},
{
"epoch": 0.49,
"learning_rate": 4.307802606531151e-05,
"loss": 0.846,
"step": 4480
},
{
"epoch": 0.49,
"learning_rate": 4.304858377361446e-05,
"loss": 0.8168,
"step": 4490
},
{
"epoch": 0.49,
"learning_rate": 4.301908910659195e-05,
"loss": 0.8436,
"step": 4500
},
{
"epoch": 0.49,
"learning_rate": 4.2989542149834796e-05,
"loss": 0.858,
"step": 4510
},
{
"epoch": 0.49,
"learning_rate": 4.2959942989085546e-05,
"loss": 0.8328,
"step": 4520
},
{
"epoch": 0.49,
"learning_rate": 4.2930291710238224e-05,
"loss": 0.8159,
"step": 4530
},
{
"epoch": 0.49,
"learning_rate": 4.290058839933811e-05,
"loss": 0.8253,
"step": 4540
},
{
"epoch": 0.49,
"learning_rate": 4.2870833142581476e-05,
"loss": 0.8485,
"step": 4550
},
{
"epoch": 0.49,
"learning_rate": 4.284102602631533e-05,
"loss": 0.8252,
"step": 4560
},
{
"epoch": 0.5,
"learning_rate": 4.2811167137037175e-05,
"loss": 0.8548,
"step": 4570
},
{
"epoch": 0.5,
"learning_rate": 4.278125656139476e-05,
"loss": 0.8376,
"step": 4580
},
{
"epoch": 0.5,
"learning_rate": 4.2751294386185796e-05,
"loss": 0.8225,
"step": 4590
},
{
"epoch": 0.5,
"learning_rate": 4.272128069835778e-05,
"loss": 0.8409,
"step": 4600
},
{
"epoch": 0.5,
"learning_rate": 4.2691215585007645e-05,
"loss": 0.8179,
"step": 4610
},
{
"epoch": 0.5,
"learning_rate": 4.266109913338159e-05,
"loss": 0.842,
"step": 4620
},
{
"epoch": 0.5,
"learning_rate": 4.263093143087477e-05,
"loss": 0.8751,
"step": 4630
},
{
"epoch": 0.5,
"learning_rate": 4.2600712565031074e-05,
"loss": 0.818,
"step": 4640
},
{
"epoch": 0.5,
"learning_rate": 4.257044262354288e-05,
"loss": 0.8358,
"step": 4650
},
{
"epoch": 0.51,
"learning_rate": 4.254012169425075e-05,
"loss": 0.8478,
"step": 4660
},
{
"epoch": 0.51,
"learning_rate": 4.250974986514322e-05,
"loss": 0.8399,
"step": 4670
},
{
"epoch": 0.51,
"learning_rate": 4.2479327224356556e-05,
"loss": 0.8575,
"step": 4680
},
{
"epoch": 0.51,
"learning_rate": 4.244885386017445e-05,
"loss": 0.8392,
"step": 4690
},
{
"epoch": 0.51,
"learning_rate": 4.241832986102778e-05,
"loss": 0.816,
"step": 4700
},
{
"epoch": 0.51,
"learning_rate": 4.2387755315494394e-05,
"loss": 0.8579,
"step": 4710
},
{
"epoch": 0.51,
"learning_rate": 4.2357130312298795e-05,
"loss": 0.83,
"step": 4720
},
{
"epoch": 0.51,
"learning_rate": 4.232645494031191e-05,
"loss": 0.8281,
"step": 4730
},
{
"epoch": 0.51,
"learning_rate": 4.2295729288550855e-05,
"loss": 0.8213,
"step": 4740
},
{
"epoch": 0.52,
"learning_rate": 4.226495344617863e-05,
"loss": 0.8265,
"step": 4750
},
{
"epoch": 0.52,
"learning_rate": 4.223412750250389e-05,
"loss": 0.8283,
"step": 4760
},
{
"epoch": 0.52,
"learning_rate": 4.220325154698068e-05,
"loss": 0.8392,
"step": 4770
},
{
"epoch": 0.52,
"learning_rate": 4.2172325669208176e-05,
"loss": 0.8189,
"step": 4780
},
{
"epoch": 0.52,
"learning_rate": 4.214134995893042e-05,
"loss": 0.8317,
"step": 4790
},
{
"epoch": 0.52,
"learning_rate": 4.211032450603608e-05,
"loss": 0.8369,
"step": 4800
},
{
"epoch": 0.52,
"learning_rate": 4.207924940055814e-05,
"loss": 0.8331,
"step": 4810
},
{
"epoch": 0.52,
"learning_rate": 4.204812473267371e-05,
"loss": 0.843,
"step": 4820
},
{
"epoch": 0.52,
"learning_rate": 4.201695059270367e-05,
"loss": 0.8413,
"step": 4830
},
{
"epoch": 0.52,
"learning_rate": 4.1985727071112543e-05,
"loss": 0.8208,
"step": 4840
},
{
"epoch": 0.53,
"learning_rate": 4.1954454258508086e-05,
"loss": 0.8346,
"step": 4850
},
{
"epoch": 0.53,
"learning_rate": 4.192313224564111e-05,
"loss": 0.8476,
"step": 4860
},
{
"epoch": 0.53,
"learning_rate": 4.1891761123405224e-05,
"loss": 0.8354,
"step": 4870
},
{
"epoch": 0.53,
"learning_rate": 4.1860340982836526e-05,
"loss": 0.8132,
"step": 4880
},
{
"epoch": 0.53,
"learning_rate": 4.1828871915113365e-05,
"loss": 0.8448,
"step": 4890
},
{
"epoch": 0.53,
"learning_rate": 4.1797354011556065e-05,
"loss": 0.8401,
"step": 4900
},
{
"epoch": 0.53,
"learning_rate": 4.176578736362669e-05,
"loss": 0.8241,
"step": 4910
},
{
"epoch": 0.53,
"learning_rate": 4.173417206292873e-05,
"loss": 0.8337,
"step": 4920
},
{
"epoch": 0.53,
"learning_rate": 4.170250820120687e-05,
"loss": 0.8248,
"step": 4930
},
{
"epoch": 0.54,
"learning_rate": 4.167079587034673e-05,
"loss": 0.8297,
"step": 4940
},
{
"epoch": 0.54,
"learning_rate": 4.163903516237455e-05,
"loss": 0.8414,
"step": 4950
},
{
"epoch": 0.54,
"learning_rate": 4.1607226169457e-05,
"loss": 0.8399,
"step": 4960
},
{
"epoch": 0.54,
"learning_rate": 4.157536898390082e-05,
"loss": 0.828,
"step": 4970
},
{
"epoch": 0.54,
"learning_rate": 4.154346369815263e-05,
"loss": 0.815,
"step": 4980
},
{
"epoch": 0.54,
"learning_rate": 4.1511510404798636e-05,
"loss": 0.8268,
"step": 4990
},
{
"epoch": 0.54,
"learning_rate": 4.147950919656434e-05,
"loss": 0.8168,
"step": 5000
},
{
"epoch": 0.54,
"learning_rate": 4.14474601663143e-05,
"loss": 0.8436,
"step": 5010
},
{
"epoch": 0.54,
"learning_rate": 4.141536340705184e-05,
"loss": 0.8193,
"step": 5020
},
{
"epoch": 0.55,
"learning_rate": 4.138321901191881e-05,
"loss": 0.8148,
"step": 5030
},
{
"epoch": 0.55,
"learning_rate": 4.1351027074195266e-05,
"loss": 0.8288,
"step": 5040
},
{
"epoch": 0.55,
"learning_rate": 4.131878768729924e-05,
"loss": 0.8342,
"step": 5050
},
{
"epoch": 0.55,
"learning_rate": 4.128650094478647e-05,
"loss": 0.8439,
"step": 5060
},
{
"epoch": 0.55,
"learning_rate": 4.12541669403501e-05,
"loss": 0.8242,
"step": 5070
},
{
"epoch": 0.55,
"learning_rate": 4.122178576782042e-05,
"loss": 0.8328,
"step": 5080
},
{
"epoch": 0.55,
"learning_rate": 4.118935752116463e-05,
"loss": 0.8305,
"step": 5090
},
{
"epoch": 0.55,
"learning_rate": 4.1156882294486484e-05,
"loss": 0.8179,
"step": 5100
},
{
"epoch": 0.55,
"learning_rate": 4.112436018202612e-05,
"loss": 0.8465,
"step": 5110
},
{
"epoch": 0.56,
"learning_rate": 4.109179127815968e-05,
"loss": 0.839,
"step": 5120
},
{
"epoch": 0.56,
"learning_rate": 4.1059175677399136e-05,
"loss": 0.8353,
"step": 5130
},
{
"epoch": 0.56,
"learning_rate": 4.1026513474391956e-05,
"loss": 0.8454,
"step": 5140
},
{
"epoch": 0.56,
"learning_rate": 4.099380476392083e-05,
"loss": 0.8232,
"step": 5150
},
{
"epoch": 0.56,
"learning_rate": 4.096104964090344e-05,
"loss": 0.8273,
"step": 5160
},
{
"epoch": 0.56,
"learning_rate": 4.09282482003921e-05,
"loss": 0.8308,
"step": 5170
},
{
"epoch": 0.56,
"learning_rate": 4.089540053757359e-05,
"loss": 0.8235,
"step": 5180
},
{
"epoch": 0.56,
"learning_rate": 4.0862506747768784e-05,
"loss": 0.8293,
"step": 5190
},
{
"epoch": 0.56,
"learning_rate": 4.082956692643244e-05,
"loss": 0.8282,
"step": 5200
},
{
"epoch": 0.57,
"learning_rate": 4.079658116915286e-05,
"loss": 0.8397,
"step": 5210
},
{
"epoch": 0.57,
"learning_rate": 4.076354957165169e-05,
"loss": 0.8473,
"step": 5220
},
{
"epoch": 0.57,
"learning_rate": 4.073047222978358e-05,
"loss": 0.814,
"step": 5230
},
{
"epoch": 0.57,
"learning_rate": 4.0697349239535905e-05,
"loss": 0.8313,
"step": 5240
},
{
"epoch": 0.57,
"learning_rate": 4.066418069702854e-05,
"loss": 0.8264,
"step": 5250
},
{
"epoch": 0.57,
"learning_rate": 4.0630966698513545e-05,
"loss": 0.8327,
"step": 5260
},
{
"epoch": 0.57,
"learning_rate": 4.059770734037487e-05,
"loss": 0.8266,
"step": 5270
},
{
"epoch": 0.57,
"learning_rate": 4.05644027191281e-05,
"loss": 0.8322,
"step": 5280
},
{
"epoch": 0.57,
"learning_rate": 4.053105293142019e-05,
"loss": 0.8118,
"step": 5290
},
{
"epoch": 0.57,
"learning_rate": 4.049765807402912e-05,
"loss": 0.8257,
"step": 5300
},
{
"epoch": 0.58,
"learning_rate": 4.046421824386371e-05,
"loss": 0.8144,
"step": 5310
},
{
"epoch": 0.58,
"learning_rate": 4.043073353796327e-05,
"loss": 0.8432,
"step": 5320
},
{
"epoch": 0.58,
"learning_rate": 4.0397204053497304e-05,
"loss": 0.8378,
"step": 5330
},
{
"epoch": 0.58,
"learning_rate": 4.03636298877653e-05,
"loss": 0.8297,
"step": 5340
},
{
"epoch": 0.58,
"learning_rate": 4.033001113819638e-05,
"loss": 0.8186,
"step": 5350
},
{
"epoch": 0.58,
"learning_rate": 4.029634790234906e-05,
"loss": 0.8163,
"step": 5360
},
{
"epoch": 0.58,
"learning_rate": 4.0262640277910953e-05,
"loss": 0.824,
"step": 5370
},
{
"epoch": 0.58,
"learning_rate": 4.0228888362698455e-05,
"loss": 0.8153,
"step": 5380
},
{
"epoch": 0.58,
"learning_rate": 4.019509225465654e-05,
"loss": 0.8221,
"step": 5390
},
{
"epoch": 0.59,
"learning_rate": 4.0161252051858366e-05,
"loss": 0.811,
"step": 5400
},
{
"epoch": 0.59,
"learning_rate": 4.0127367852505106e-05,
"loss": 0.8121,
"step": 5410
},
{
"epoch": 0.59,
"learning_rate": 4.0093439754925566e-05,
"loss": 0.8324,
"step": 5420
},
{
"epoch": 0.59,
"learning_rate": 4.005946785757596e-05,
"loss": 0.8439,
"step": 5430
},
{
"epoch": 0.59,
"learning_rate": 4.0025452259039596e-05,
"loss": 0.8072,
"step": 5440
},
{
"epoch": 0.59,
"learning_rate": 3.9991393058026614e-05,
"loss": 0.8491,
"step": 5450
},
{
"epoch": 0.59,
"learning_rate": 3.9957290353373666e-05,
"loss": 0.8209,
"step": 5460
},
{
"epoch": 0.59,
"learning_rate": 3.992314424404366e-05,
"loss": 0.7993,
"step": 5470
},
{
"epoch": 0.59,
"learning_rate": 3.9888954829125456e-05,
"loss": 0.8192,
"step": 5480
},
{
"epoch": 0.6,
"learning_rate": 3.985472220783358e-05,
"loss": 0.8375,
"step": 5490
},
{
"epoch": 0.6,
"learning_rate": 3.982044647950794e-05,
"loss": 0.8465,
"step": 5500
},
{
"epoch": 0.6,
"learning_rate": 3.978612774361354e-05,
"loss": 0.8105,
"step": 5510
},
{
"epoch": 0.6,
"learning_rate": 3.975176609974019e-05,
"loss": 0.8238,
"step": 5520
},
{
"epoch": 0.6,
"learning_rate": 3.9717361647602205e-05,
"loss": 0.8311,
"step": 5530
},
{
"epoch": 0.6,
"learning_rate": 3.9682914487038135e-05,
"loss": 0.8303,
"step": 5540
},
{
"epoch": 0.6,
"learning_rate": 3.9648424718010467e-05,
"loss": 0.8457,
"step": 5550
},
{
"epoch": 0.6,
"learning_rate": 3.9613892440605324e-05,
"loss": 0.8166,
"step": 5560
},
{
"epoch": 0.6,
"learning_rate": 3.957931775503219e-05,
"loss": 0.8347,
"step": 5570
},
{
"epoch": 0.61,
"learning_rate": 3.954470076162362e-05,
"loss": 0.8253,
"step": 5580
},
{
"epoch": 0.61,
"learning_rate": 3.951004156083492e-05,
"loss": 0.8439,
"step": 5590
},
{
"epoch": 0.61,
"learning_rate": 3.9475340253243906e-05,
"loss": 0.8354,
"step": 5600
},
{
"epoch": 0.61,
"learning_rate": 3.944059693955059e-05,
"loss": 0.8237,
"step": 5610
},
{
"epoch": 0.61,
"learning_rate": 3.940581172057683e-05,
"loss": 0.8298,
"step": 5620
},
{
"epoch": 0.61,
"learning_rate": 3.937098469726615e-05,
"loss": 0.8359,
"step": 5630
},
{
"epoch": 0.61,
"learning_rate": 3.933611597068333e-05,
"loss": 0.8229,
"step": 5640
},
{
"epoch": 0.61,
"learning_rate": 3.930120564201423e-05,
"loss": 0.8199,
"step": 5650
},
{
"epoch": 0.61,
"learning_rate": 3.926625381256538e-05,
"loss": 0.8275,
"step": 5660
},
{
"epoch": 0.61,
"learning_rate": 3.9231260583763786e-05,
"loss": 0.8351,
"step": 5670
},
{
"epoch": 0.62,
"learning_rate": 3.919622605715655e-05,
"loss": 0.8237,
"step": 5680
},
{
"epoch": 0.62,
"learning_rate": 3.916115033441066e-05,
"loss": 0.8156,
"step": 5690
},
{
"epoch": 0.62,
"learning_rate": 3.91260335173126e-05,
"loss": 0.8413,
"step": 5700
},
{
"epoch": 0.62,
"learning_rate": 3.909087570776816e-05,
"loss": 0.8343,
"step": 5710
},
{
"epoch": 0.62,
"learning_rate": 3.905567700780205e-05,
"loss": 0.8333,
"step": 5720
},
{
"epoch": 0.62,
"learning_rate": 3.902043751955764e-05,
"loss": 0.8067,
"step": 5730
},
{
"epoch": 0.62,
"learning_rate": 3.89851573452967e-05,
"loss": 0.8163,
"step": 5740
},
{
"epoch": 0.62,
"learning_rate": 3.894983658739902e-05,
"loss": 0.8073,
"step": 5750
},
{
"epoch": 0.62,
"learning_rate": 3.8914475348362194e-05,
"loss": 0.8144,
"step": 5760
},
{
"epoch": 0.63,
"learning_rate": 3.887907373080126e-05,
"loss": 0.8259,
"step": 5770
},
{
"epoch": 0.63,
"learning_rate": 3.884363183744847e-05,
"loss": 0.8102,
"step": 5780
},
{
"epoch": 0.63,
"learning_rate": 3.8808149771152916e-05,
"loss": 0.8212,
"step": 5790
},
{
"epoch": 0.63,
"learning_rate": 3.877262763488028e-05,
"loss": 0.8357,
"step": 5800
},
{
"epoch": 0.63,
"learning_rate": 3.8737065531712535e-05,
"loss": 0.8211,
"step": 5810
},
{
"epoch": 0.63,
"learning_rate": 3.8701463564847615e-05,
"loss": 0.8132,
"step": 5820
},
{
"epoch": 0.63,
"learning_rate": 3.866582183759915e-05,
"loss": 0.796,
"step": 5830
},
{
"epoch": 0.63,
"learning_rate": 3.8630140453396134e-05,
"loss": 0.8204,
"step": 5840
},
{
"epoch": 0.63,
"learning_rate": 3.859441951578268e-05,
"loss": 0.8276,
"step": 5850
},
{
"epoch": 0.64,
"learning_rate": 3.855865912841762e-05,
"loss": 0.8148,
"step": 5860
},
{
"epoch": 0.64,
"learning_rate": 3.852285939507432e-05,
"loss": 0.7977,
"step": 5870
},
{
"epoch": 0.64,
"learning_rate": 3.8487020419640296e-05,
"loss": 0.8046,
"step": 5880
},
{
"epoch": 0.64,
"learning_rate": 3.845114230611696e-05,
"loss": 0.8269,
"step": 5890
},
{
"epoch": 0.64,
"learning_rate": 3.841522515861927e-05,
"loss": 0.8089,
"step": 5900
},
{
"epoch": 0.64,
"learning_rate": 3.83792690813755e-05,
"loss": 0.8217,
"step": 5910
},
{
"epoch": 0.64,
"learning_rate": 3.834327417872686e-05,
"loss": 0.8226,
"step": 5920
},
{
"epoch": 0.64,
"learning_rate": 3.830724055512722e-05,
"loss": 0.7927,
"step": 5930
},
{
"epoch": 0.64,
"learning_rate": 3.827116831514286e-05,
"loss": 0.8212,
"step": 5940
},
{
"epoch": 0.65,
"learning_rate": 3.8235057563452075e-05,
"loss": 0.8079,
"step": 5950
},
{
"epoch": 0.65,
"learning_rate": 3.819890840484494e-05,
"loss": 0.8186,
"step": 5960
},
{
"epoch": 0.65,
"learning_rate": 3.816272094422298e-05,
"loss": 0.8124,
"step": 5970
},
{
"epoch": 0.65,
"learning_rate": 3.812649528659887e-05,
"loss": 0.8115,
"step": 5980
},
{
"epoch": 0.65,
"learning_rate": 3.809023153709612e-05,
"loss": 0.8117,
"step": 5990
},
{
"epoch": 0.65,
"learning_rate": 3.80539298009488e-05,
"loss": 0.7797,
"step": 6000
}
],
"max_steps": 18442,
"num_train_epochs": 2,
"total_flos": 6.206345368989336e+18,
"trial_name": null,
"trial_params": null
}