|
{ |
|
"best_metric": 0.5150178074836731, |
|
"best_model_checkpoint": "miner_id_24/checkpoint-200", |
|
"epoch": 0.6235385814497272, |
|
"eval_steps": 50, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.003117692907248636, |
|
"grad_norm": 11.655372619628906, |
|
"learning_rate": 1e-05, |
|
"loss": 9.4725, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.003117692907248636, |
|
"eval_loss": 2.0170581340789795, |
|
"eval_runtime": 19.8075, |
|
"eval_samples_per_second": 27.262, |
|
"eval_steps_per_second": 6.816, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.006235385814497272, |
|
"grad_norm": 13.125012397766113, |
|
"learning_rate": 2e-05, |
|
"loss": 11.8029, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.009353078721745909, |
|
"grad_norm": 12.141226768493652, |
|
"learning_rate": 3e-05, |
|
"loss": 11.4664, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.012470771628994544, |
|
"grad_norm": 14.667614936828613, |
|
"learning_rate": 4e-05, |
|
"loss": 12.0023, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.01558846453624318, |
|
"grad_norm": 12.08423137664795, |
|
"learning_rate": 5e-05, |
|
"loss": 10.5206, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.018706157443491817, |
|
"grad_norm": 11.249734878540039, |
|
"learning_rate": 6e-05, |
|
"loss": 9.5772, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.021823850350740453, |
|
"grad_norm": 22.513164520263672, |
|
"learning_rate": 7e-05, |
|
"loss": 9.5923, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.02494154325798909, |
|
"grad_norm": 18.979618072509766, |
|
"learning_rate": 8e-05, |
|
"loss": 9.3301, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.028059236165237724, |
|
"grad_norm": 29.20528221130371, |
|
"learning_rate": 9e-05, |
|
"loss": 10.607, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.03117692907248636, |
|
"grad_norm": 25.135574340820312, |
|
"learning_rate": 0.0001, |
|
"loss": 11.5226, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.034294621979735, |
|
"grad_norm": 15.285947799682617, |
|
"learning_rate": 9.999316524962345e-05, |
|
"loss": 8.4817, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.037412314886983634, |
|
"grad_norm": 15.414291381835938, |
|
"learning_rate": 9.997266286704631e-05, |
|
"loss": 7.4645, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.04053000779423227, |
|
"grad_norm": 13.062957763671875, |
|
"learning_rate": 9.993849845741524e-05, |
|
"loss": 5.7704, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.043647700701480906, |
|
"grad_norm": 13.44833755493164, |
|
"learning_rate": 9.989068136093873e-05, |
|
"loss": 6.9123, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.04676539360872954, |
|
"grad_norm": 12.443500518798828, |
|
"learning_rate": 9.98292246503335e-05, |
|
"loss": 5.7389, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.04988308651597818, |
|
"grad_norm": 12.959155082702637, |
|
"learning_rate": 9.975414512725057e-05, |
|
"loss": 5.9033, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.05300077942322681, |
|
"grad_norm": 9.476544380187988, |
|
"learning_rate": 9.966546331768191e-05, |
|
"loss": 5.9961, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.05611847233047545, |
|
"grad_norm": 11.186797142028809, |
|
"learning_rate": 9.956320346634876e-05, |
|
"loss": 5.6498, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.059236165237724084, |
|
"grad_norm": 11.144340515136719, |
|
"learning_rate": 9.944739353007344e-05, |
|
"loss": 5.4165, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.06235385814497272, |
|
"grad_norm": 12.207180976867676, |
|
"learning_rate": 9.931806517013612e-05, |
|
"loss": 6.3502, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.06547155105222136, |
|
"grad_norm": 12.063061714172363, |
|
"learning_rate": 9.917525374361912e-05, |
|
"loss": 5.482, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.06858924395947, |
|
"grad_norm": 13.639694213867188, |
|
"learning_rate": 9.901899829374047e-05, |
|
"loss": 5.8621, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.07170693686671863, |
|
"grad_norm": 10.871453285217285, |
|
"learning_rate": 9.884934153917997e-05, |
|
"loss": 5.3057, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.07482462977396727, |
|
"grad_norm": 11.467005729675293, |
|
"learning_rate": 9.86663298624003e-05, |
|
"loss": 4.4601, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.0779423226812159, |
|
"grad_norm": 11.330914497375488, |
|
"learning_rate": 9.847001329696653e-05, |
|
"loss": 4.9252, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.08106001558846454, |
|
"grad_norm": 11.97061538696289, |
|
"learning_rate": 9.826044551386744e-05, |
|
"loss": 4.8823, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.08417770849571317, |
|
"grad_norm": 13.585055351257324, |
|
"learning_rate": 9.803768380684242e-05, |
|
"loss": 5.3709, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.08729540140296181, |
|
"grad_norm": 13.680033683776855, |
|
"learning_rate": 9.780178907671789e-05, |
|
"loss": 5.6623, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.09041309431021044, |
|
"grad_norm": 14.604742050170898, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 5.3659, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.09353078721745908, |
|
"grad_norm": 13.014104843139648, |
|
"learning_rate": 9.729086208503174e-05, |
|
"loss": 5.0338, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.09664848012470771, |
|
"grad_norm": 13.416587829589844, |
|
"learning_rate": 9.701596950580806e-05, |
|
"loss": 4.5345, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.09976617303195635, |
|
"grad_norm": 13.330890655517578, |
|
"learning_rate": 9.672822322997305e-05, |
|
"loss": 4.7284, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.10288386593920498, |
|
"grad_norm": 14.181489944458008, |
|
"learning_rate": 9.642770192448536e-05, |
|
"loss": 5.1683, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.10600155884645363, |
|
"grad_norm": 14.32101058959961, |
|
"learning_rate": 9.611448774886924e-05, |
|
"loss": 4.4735, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.10911925175370225, |
|
"grad_norm": 11.41262435913086, |
|
"learning_rate": 9.578866633275288e-05, |
|
"loss": 4.555, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.1122369446609509, |
|
"grad_norm": 11.651115417480469, |
|
"learning_rate": 9.545032675245813e-05, |
|
"loss": 4.3732, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.11535463756819954, |
|
"grad_norm": 10.54432487487793, |
|
"learning_rate": 9.509956150664796e-05, |
|
"loss": 4.2542, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.11847233047544817, |
|
"grad_norm": 13.021109580993652, |
|
"learning_rate": 9.473646649103818e-05, |
|
"loss": 4.5108, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.12159002338269681, |
|
"grad_norm": 13.503449440002441, |
|
"learning_rate": 9.43611409721806e-05, |
|
"loss": 4.2284, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.12470771628994544, |
|
"grad_norm": 13.191174507141113, |
|
"learning_rate": 9.397368756032445e-05, |
|
"loss": 4.2834, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.12782540919719407, |
|
"grad_norm": 12.584697723388672, |
|
"learning_rate": 9.357421218136386e-05, |
|
"loss": 3.8461, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.1309431021044427, |
|
"grad_norm": 14.472663879394531, |
|
"learning_rate": 9.316282404787871e-05, |
|
"loss": 3.8354, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.13406079501169135, |
|
"grad_norm": 12.175851821899414, |
|
"learning_rate": 9.273963562927695e-05, |
|
"loss": 3.9488, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.13717848791894, |
|
"grad_norm": 28.482507705688477, |
|
"learning_rate": 9.230476262104677e-05, |
|
"loss": 3.8271, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.1402961808261886, |
|
"grad_norm": 18.594249725341797, |
|
"learning_rate": 9.185832391312644e-05, |
|
"loss": 4.1931, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.14341387373343725, |
|
"grad_norm": 15.749979019165039, |
|
"learning_rate": 9.140044155740101e-05, |
|
"loss": 3.1982, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.1465315666406859, |
|
"grad_norm": 20.301923751831055, |
|
"learning_rate": 9.093124073433463e-05, |
|
"loss": 4.3817, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.14964925954793454, |
|
"grad_norm": 13.996861457824707, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 4.1606, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.15276695245518315, |
|
"grad_norm": 19.21599006652832, |
|
"learning_rate": 8.995939984474624e-05, |
|
"loss": 4.6406, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.1558846453624318, |
|
"grad_norm": 19.336666107177734, |
|
"learning_rate": 8.945702546981969e-05, |
|
"loss": 4.3495, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.1558846453624318, |
|
"eval_loss": 0.745703935623169, |
|
"eval_runtime": 20.391, |
|
"eval_samples_per_second": 26.482, |
|
"eval_steps_per_second": 6.621, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.15900233826968044, |
|
"grad_norm": 16.644689559936523, |
|
"learning_rate": 8.894386393810563e-05, |
|
"loss": 3.5753, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.16212003117692908, |
|
"grad_norm": 13.698346138000488, |
|
"learning_rate": 8.842005554284296e-05, |
|
"loss": 3.5221, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.1652377240841777, |
|
"grad_norm": 10.113664627075195, |
|
"learning_rate": 8.788574348801675e-05, |
|
"loss": 3.4487, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.16835541699142634, |
|
"grad_norm": 9.818008422851562, |
|
"learning_rate": 8.73410738492077e-05, |
|
"loss": 2.9978, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.17147310989867498, |
|
"grad_norm": 7.929912090301514, |
|
"learning_rate": 8.678619553365659e-05, |
|
"loss": 2.7355, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.17459080280592362, |
|
"grad_norm": 9.08836555480957, |
|
"learning_rate": 8.622126023955446e-05, |
|
"loss": 3.0724, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.17770849571317227, |
|
"grad_norm": 8.289327621459961, |
|
"learning_rate": 8.564642241456986e-05, |
|
"loss": 3.1148, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.18082618862042088, |
|
"grad_norm": 8.683759689331055, |
|
"learning_rate": 8.506183921362443e-05, |
|
"loss": 2.7112, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.18394388152766952, |
|
"grad_norm": 8.794864654541016, |
|
"learning_rate": 8.44676704559283e-05, |
|
"loss": 3.2929, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.18706157443491817, |
|
"grad_norm": 7.364131927490234, |
|
"learning_rate": 8.386407858128706e-05, |
|
"loss": 2.63, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.1901792673421668, |
|
"grad_norm": 7.907564163208008, |
|
"learning_rate": 8.32512286056924e-05, |
|
"loss": 3.0691, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.19329696024941542, |
|
"grad_norm": 9.792932510375977, |
|
"learning_rate": 8.262928807620843e-05, |
|
"loss": 3.1184, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.19641465315666407, |
|
"grad_norm": 7.965488910675049, |
|
"learning_rate": 8.199842702516583e-05, |
|
"loss": 3.3138, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.1995323460639127, |
|
"grad_norm": 9.244006156921387, |
|
"learning_rate": 8.135881792367686e-05, |
|
"loss": 2.995, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.20265003897116135, |
|
"grad_norm": 6.818788051605225, |
|
"learning_rate": 8.07106356344834e-05, |
|
"loss": 2.6801, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.20576773187840997, |
|
"grad_norm": 8.18008804321289, |
|
"learning_rate": 8.005405736415126e-05, |
|
"loss": 2.8639, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.2088854247856586, |
|
"grad_norm": 8.462689399719238, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 3.3518, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.21200311769290725, |
|
"grad_norm": 9.65335464477539, |
|
"learning_rate": 7.871643313414718e-05, |
|
"loss": 3.0363, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.2151208106001559, |
|
"grad_norm": 9.894681930541992, |
|
"learning_rate": 7.803575286758364e-05, |
|
"loss": 3.6108, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.2182385035074045, |
|
"grad_norm": 10.462113380432129, |
|
"learning_rate": 7.734740790612136e-05, |
|
"loss": 3.4783, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.22135619641465315, |
|
"grad_norm": 8.34600830078125, |
|
"learning_rate": 7.66515864363997e-05, |
|
"loss": 3.0312, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.2244738893219018, |
|
"grad_norm": 8.083913803100586, |
|
"learning_rate": 7.594847868906076e-05, |
|
"loss": 3.1196, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.22759158222915044, |
|
"grad_norm": 9.153026580810547, |
|
"learning_rate": 7.52382768867422e-05, |
|
"loss": 3.2304, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.23070927513639908, |
|
"grad_norm": 18.286788940429688, |
|
"learning_rate": 7.452117519152542e-05, |
|
"loss": 3.3481, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.2338269680436477, |
|
"grad_norm": 12.73767375946045, |
|
"learning_rate": 7.379736965185368e-05, |
|
"loss": 3.5366, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.23694466095089634, |
|
"grad_norm": 11.967613220214844, |
|
"learning_rate": 7.30670581489344e-05, |
|
"loss": 3.4847, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.24006235385814498, |
|
"grad_norm": 9.898238182067871, |
|
"learning_rate": 7.233044034264034e-05, |
|
"loss": 3.4925, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.24318004676539362, |
|
"grad_norm": 13.920319557189941, |
|
"learning_rate": 7.158771761692464e-05, |
|
"loss": 3.8911, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.24629773967264224, |
|
"grad_norm": 14.240066528320312, |
|
"learning_rate": 7.083909302476453e-05, |
|
"loss": 3.9988, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.24941543257989088, |
|
"grad_norm": 12.303492546081543, |
|
"learning_rate": 7.008477123264848e-05, |
|
"loss": 4.0639, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.2525331254871395, |
|
"grad_norm": 11.494243621826172, |
|
"learning_rate": 6.932495846462261e-05, |
|
"loss": 3.8558, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.25565081839438814, |
|
"grad_norm": 11.226067543029785, |
|
"learning_rate": 6.855986244591104e-05, |
|
"loss": 3.3342, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.2587685113016368, |
|
"grad_norm": 11.755266189575195, |
|
"learning_rate": 6.778969234612584e-05, |
|
"loss": 4.2033, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.2618862042088854, |
|
"grad_norm": 8.904754638671875, |
|
"learning_rate": 6.701465872208216e-05, |
|
"loss": 3.4204, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.26500389711613404, |
|
"grad_norm": 11.56294059753418, |
|
"learning_rate": 6.623497346023418e-05, |
|
"loss": 3.26, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.2681215900233827, |
|
"grad_norm": 10.843947410583496, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 3.6785, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.2712392829306313, |
|
"grad_norm": 9.812710762023926, |
|
"learning_rate": 6.466250186922325e-05, |
|
"loss": 3.1796, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.27435697583788, |
|
"grad_norm": 12.650871276855469, |
|
"learning_rate": 6.387014543809223e-05, |
|
"loss": 3.7719, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.2774746687451286, |
|
"grad_norm": 9.260048866271973, |
|
"learning_rate": 6.307399704769099e-05, |
|
"loss": 3.3953, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.2805923616523772, |
|
"grad_norm": 10.217822074890137, |
|
"learning_rate": 6.227427435703997e-05, |
|
"loss": 3.2167, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.2837100545596259, |
|
"grad_norm": 11.022683143615723, |
|
"learning_rate": 6.147119600233758e-05, |
|
"loss": 3.2912, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.2868277474668745, |
|
"grad_norm": 12.414704322814941, |
|
"learning_rate": 6.066498153718735e-05, |
|
"loss": 3.328, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.2899454403741232, |
|
"grad_norm": 11.644399642944336, |
|
"learning_rate": 5.985585137257401e-05, |
|
"loss": 3.1574, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.2930631332813718, |
|
"grad_norm": 19.45965003967285, |
|
"learning_rate": 5.90440267166055e-05, |
|
"loss": 2.4187, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.2961808261886204, |
|
"grad_norm": 9.440485000610352, |
|
"learning_rate": 5.8229729514036705e-05, |
|
"loss": 2.7639, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.2992985190958691, |
|
"grad_norm": 10.892629623413086, |
|
"learning_rate": 5.74131823855921e-05, |
|
"loss": 2.6128, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.3024162120031177, |
|
"grad_norm": 11.69962215423584, |
|
"learning_rate": 5.6594608567103456e-05, |
|
"loss": 3.0135, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.3055339049103663, |
|
"grad_norm": 16.776004791259766, |
|
"learning_rate": 5.577423184847932e-05, |
|
"loss": 3.2701, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.308651597817615, |
|
"grad_norm": 10.856680870056152, |
|
"learning_rate": 5.495227651252315e-05, |
|
"loss": 3.1861, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.3117692907248636, |
|
"grad_norm": 23.68087387084961, |
|
"learning_rate": 5.4128967273616625e-05, |
|
"loss": 4.1984, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.3117692907248636, |
|
"eval_loss": 0.6143662333488464, |
|
"eval_runtime": 20.4247, |
|
"eval_samples_per_second": 26.439, |
|
"eval_steps_per_second": 6.61, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.31488698363211226, |
|
"grad_norm": 6.1397271156311035, |
|
"learning_rate": 5.330452921628497e-05, |
|
"loss": 2.3444, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.3180046765393609, |
|
"grad_norm": 7.607326030731201, |
|
"learning_rate": 5.247918773366112e-05, |
|
"loss": 2.9072, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.3211223694466095, |
|
"grad_norm": 6.6434407234191895, |
|
"learning_rate": 5.165316846586541e-05, |
|
"loss": 2.3472, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.32424006235385816, |
|
"grad_norm": 7.280537128448486, |
|
"learning_rate": 5.0826697238317935e-05, |
|
"loss": 2.4164, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.3273577552611068, |
|
"grad_norm": 6.675552845001221, |
|
"learning_rate": 5e-05, |
|
"loss": 2.3209, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.3304754481683554, |
|
"grad_norm": 6.807125568389893, |
|
"learning_rate": 4.917330276168208e-05, |
|
"loss": 2.3718, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.33359314107560406, |
|
"grad_norm": 6.205484867095947, |
|
"learning_rate": 4.834683153413459e-05, |
|
"loss": 2.4963, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.3367108339828527, |
|
"grad_norm": 6.572414875030518, |
|
"learning_rate": 4.7520812266338885e-05, |
|
"loss": 2.6112, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.33982852689010135, |
|
"grad_norm": 6.466701984405518, |
|
"learning_rate": 4.669547078371504e-05, |
|
"loss": 2.1945, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.34294621979734996, |
|
"grad_norm": 5.984310150146484, |
|
"learning_rate": 4.5871032726383386e-05, |
|
"loss": 2.8097, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.3460639127045986, |
|
"grad_norm": 10.94200325012207, |
|
"learning_rate": 4.504772348747687e-05, |
|
"loss": 2.4516, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.34918160561184725, |
|
"grad_norm": 11.749654769897461, |
|
"learning_rate": 4.4225768151520694e-05, |
|
"loss": 2.4898, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.35229929851909586, |
|
"grad_norm": 7.231618881225586, |
|
"learning_rate": 4.3405391432896555e-05, |
|
"loss": 2.5991, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.35541699142634453, |
|
"grad_norm": 8.064313888549805, |
|
"learning_rate": 4.2586817614407895e-05, |
|
"loss": 2.6607, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.35853468433359315, |
|
"grad_norm": 14.305150032043457, |
|
"learning_rate": 4.17702704859633e-05, |
|
"loss": 2.7776, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.36165237724084176, |
|
"grad_norm": 7.72219705581665, |
|
"learning_rate": 4.095597328339452e-05, |
|
"loss": 2.4705, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.36477007014809043, |
|
"grad_norm": 7.625950813293457, |
|
"learning_rate": 4.0144148627425993e-05, |
|
"loss": 2.4539, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.36788776305533905, |
|
"grad_norm": 8.687618255615234, |
|
"learning_rate": 3.933501846281267e-05, |
|
"loss": 3.2093, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.37100545596258766, |
|
"grad_norm": 7.8179826736450195, |
|
"learning_rate": 3.852880399766243e-05, |
|
"loss": 3.1214, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.37412314886983633, |
|
"grad_norm": 11.393646240234375, |
|
"learning_rate": 3.772572564296005e-05, |
|
"loss": 3.2904, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.37724084177708495, |
|
"grad_norm": 9.927339553833008, |
|
"learning_rate": 3.6926002952309016e-05, |
|
"loss": 3.8075, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.3803585346843336, |
|
"grad_norm": 7.990683555603027, |
|
"learning_rate": 3.612985456190778e-05, |
|
"loss": 2.7959, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.38347622759158223, |
|
"grad_norm": 7.790266513824463, |
|
"learning_rate": 3.533749813077677e-05, |
|
"loss": 2.7875, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.38659392049883085, |
|
"grad_norm": 8.719717025756836, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 3.7968, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.3897116134060795, |
|
"grad_norm": 10.316503524780273, |
|
"learning_rate": 3.3765026539765834e-05, |
|
"loss": 3.2443, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.39282930631332813, |
|
"grad_norm": 8.768962860107422, |
|
"learning_rate": 3.298534127791785e-05, |
|
"loss": 3.6746, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.3959469992205768, |
|
"grad_norm": 9.664812088012695, |
|
"learning_rate": 3.221030765387417e-05, |
|
"loss": 3.1677, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.3990646921278254, |
|
"grad_norm": 9.670389175415039, |
|
"learning_rate": 3.144013755408895e-05, |
|
"loss": 3.4062, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.40218238503507403, |
|
"grad_norm": 9.162301063537598, |
|
"learning_rate": 3.0675041535377405e-05, |
|
"loss": 3.6435, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.4053000779423227, |
|
"grad_norm": 7.748847961425781, |
|
"learning_rate": 2.991522876735154e-05, |
|
"loss": 3.178, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.4084177708495713, |
|
"grad_norm": 9.13791561126709, |
|
"learning_rate": 2.916090697523549e-05, |
|
"loss": 3.3075, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.41153546375681993, |
|
"grad_norm": 9.217190742492676, |
|
"learning_rate": 2.8412282383075363e-05, |
|
"loss": 3.5296, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.4146531566640686, |
|
"grad_norm": 10.093830108642578, |
|
"learning_rate": 2.766955965735968e-05, |
|
"loss": 3.282, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.4177708495713172, |
|
"grad_norm": 11.196274757385254, |
|
"learning_rate": 2.693294185106562e-05, |
|
"loss": 3.1394, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.4208885424785659, |
|
"grad_norm": 9.726496696472168, |
|
"learning_rate": 2.6202630348146324e-05, |
|
"loss": 3.7191, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.4240062353858145, |
|
"grad_norm": 10.877890586853027, |
|
"learning_rate": 2.547882480847461e-05, |
|
"loss": 3.2328, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.4271239282930631, |
|
"grad_norm": 10.798422813415527, |
|
"learning_rate": 2.476172311325783e-05, |
|
"loss": 3.3357, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.4302416212003118, |
|
"grad_norm": 9.386134147644043, |
|
"learning_rate": 2.405152131093926e-05, |
|
"loss": 3.3181, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.4333593141075604, |
|
"grad_norm": 12.355273246765137, |
|
"learning_rate": 2.3348413563600325e-05, |
|
"loss": 2.9209, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.436477007014809, |
|
"grad_norm": 10.783023834228516, |
|
"learning_rate": 2.2652592093878666e-05, |
|
"loss": 3.6177, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.4395946999220577, |
|
"grad_norm": 16.665035247802734, |
|
"learning_rate": 2.196424713241637e-05, |
|
"loss": 2.9883, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.4427123928293063, |
|
"grad_norm": 10.434625625610352, |
|
"learning_rate": 2.128356686585282e-05, |
|
"loss": 2.5966, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.44583008573655497, |
|
"grad_norm": 9.988262176513672, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 2.3265, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.4489477786438036, |
|
"grad_norm": 10.1835355758667, |
|
"learning_rate": 1.9945942635848748e-05, |
|
"loss": 2.6702, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.4520654715510522, |
|
"grad_norm": 12.877456665039062, |
|
"learning_rate": 1.928936436551661e-05, |
|
"loss": 3.1513, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.45518316445830087, |
|
"grad_norm": 11.23567008972168, |
|
"learning_rate": 1.8641182076323148e-05, |
|
"loss": 2.9379, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.4583008573655495, |
|
"grad_norm": 11.280649185180664, |
|
"learning_rate": 1.800157297483417e-05, |
|
"loss": 2.5896, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.46141855027279816, |
|
"grad_norm": 12.652727127075195, |
|
"learning_rate": 1.7370711923791567e-05, |
|
"loss": 3.0991, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.46453624318004677, |
|
"grad_norm": 12.5499906539917, |
|
"learning_rate": 1.6748771394307585e-05, |
|
"loss": 2.7149, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.4676539360872954, |
|
"grad_norm": 22.550954818725586, |
|
"learning_rate": 1.6135921418712956e-05, |
|
"loss": 3.8598, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.4676539360872954, |
|
"eval_loss": 0.5526705980300903, |
|
"eval_runtime": 20.4183, |
|
"eval_samples_per_second": 26.447, |
|
"eval_steps_per_second": 6.612, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.47077162899454406, |
|
"grad_norm": 5.510451316833496, |
|
"learning_rate": 1.553232954407171e-05, |
|
"loss": 1.7517, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.47388932190179267, |
|
"grad_norm": 5.776265621185303, |
|
"learning_rate": 1.4938160786375572e-05, |
|
"loss": 2.2235, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.4770070148090413, |
|
"grad_norm": 5.69129753112793, |
|
"learning_rate": 1.435357758543015e-05, |
|
"loss": 2.0681, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.48012470771628996, |
|
"grad_norm": 7.5511956214904785, |
|
"learning_rate": 1.3778739760445552e-05, |
|
"loss": 2.3379, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.48324240062353857, |
|
"grad_norm": 6.128115177154541, |
|
"learning_rate": 1.3213804466343421e-05, |
|
"loss": 1.9243, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.48636009353078724, |
|
"grad_norm": 5.709255218505859, |
|
"learning_rate": 1.2658926150792322e-05, |
|
"loss": 2.0376, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.48947778643803586, |
|
"grad_norm": 6.6855950355529785, |
|
"learning_rate": 1.2114256511983274e-05, |
|
"loss": 2.2901, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.49259547934528447, |
|
"grad_norm": 6.084838390350342, |
|
"learning_rate": 1.157994445715706e-05, |
|
"loss": 2.2606, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.49571317225253314, |
|
"grad_norm": 6.58294677734375, |
|
"learning_rate": 1.1056136061894384e-05, |
|
"loss": 2.2749, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.49883086515978176, |
|
"grad_norm": 7.350609302520752, |
|
"learning_rate": 1.0542974530180327e-05, |
|
"loss": 2.2602, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.5019485580670304, |
|
"grad_norm": 6.449454307556152, |
|
"learning_rate": 1.0040600155253765e-05, |
|
"loss": 2.0813, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.505066250974279, |
|
"grad_norm": 10.626570701599121, |
|
"learning_rate": 9.549150281252633e-06, |
|
"loss": 2.2476, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.5081839438815277, |
|
"grad_norm": 5.894722938537598, |
|
"learning_rate": 9.068759265665384e-06, |
|
"loss": 2.2141, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.5113016367887763, |
|
"grad_norm": 8.111690521240234, |
|
"learning_rate": 8.599558442598998e-06, |
|
"loss": 2.0128, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.514419329696025, |
|
"grad_norm": 6.200974464416504, |
|
"learning_rate": 8.141676086873572e-06, |
|
"loss": 2.2972, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.5175370226032736, |
|
"grad_norm": 7.870783805847168, |
|
"learning_rate": 7.695237378953223e-06, |
|
"loss": 2.4488, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.5206547155105222, |
|
"grad_norm": 7.808360576629639, |
|
"learning_rate": 7.260364370723044e-06, |
|
"loss": 2.2896, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.5237724084177708, |
|
"grad_norm": 7.943413257598877, |
|
"learning_rate": 6.837175952121306e-06, |
|
"loss": 2.3406, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.5268901013250195, |
|
"grad_norm": 9.041855812072754, |
|
"learning_rate": 6.425787818636131e-06, |
|
"loss": 2.8324, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.5300077942322681, |
|
"grad_norm": 8.971168518066406, |
|
"learning_rate": 6.026312439675552e-06, |
|
"loss": 3.0823, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.5331254871395168, |
|
"grad_norm": 7.470766544342041, |
|
"learning_rate": 5.6388590278194096e-06, |
|
"loss": 2.7118, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.5362431800467654, |
|
"grad_norm": 9.016336441040039, |
|
"learning_rate": 5.263533508961827e-06, |
|
"loss": 2.9919, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.539360872954014, |
|
"grad_norm": 8.753948211669922, |
|
"learning_rate": 4.900438493352055e-06, |
|
"loss": 3.1282, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.5424785658612626, |
|
"grad_norm": 9.623739242553711, |
|
"learning_rate": 4.549673247541875e-06, |
|
"loss": 2.9055, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.5455962587685113, |
|
"grad_norm": 9.623494148254395, |
|
"learning_rate": 4.2113336672471245e-06, |
|
"loss": 3.3859, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.54871395167576, |
|
"grad_norm": 11.163936614990234, |
|
"learning_rate": 3.885512251130763e-06, |
|
"loss": 2.9277, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.5518316445830086, |
|
"grad_norm": 8.889883995056152, |
|
"learning_rate": 3.5722980755146517e-06, |
|
"loss": 3.0582, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.5549493374902572, |
|
"grad_norm": 14.302506446838379, |
|
"learning_rate": 3.271776770026963e-06, |
|
"loss": 3.8435, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.5580670303975058, |
|
"grad_norm": 9.2825345993042, |
|
"learning_rate": 2.9840304941919415e-06, |
|
"loss": 3.2624, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.5611847233047544, |
|
"grad_norm": 9.157449722290039, |
|
"learning_rate": 2.7091379149682685e-06, |
|
"loss": 3.5069, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.5643024162120032, |
|
"grad_norm": 7.781318187713623, |
|
"learning_rate": 2.4471741852423237e-06, |
|
"loss": 3.0756, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.5674201091192518, |
|
"grad_norm": 10.207369804382324, |
|
"learning_rate": 2.1982109232821178e-06, |
|
"loss": 3.7111, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.5705378020265004, |
|
"grad_norm": 8.891037940979004, |
|
"learning_rate": 1.962316193157593e-06, |
|
"loss": 3.7993, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.573655494933749, |
|
"grad_norm": 9.740682601928711, |
|
"learning_rate": 1.7395544861325718e-06, |
|
"loss": 3.6118, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.5767731878409976, |
|
"grad_norm": 12.479891777038574, |
|
"learning_rate": 1.5299867030334814e-06, |
|
"loss": 3.3783, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.5798908807482464, |
|
"grad_norm": 16.013286590576172, |
|
"learning_rate": 1.333670137599713e-06, |
|
"loss": 2.8446, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.583008573655495, |
|
"grad_norm": 9.353748321533203, |
|
"learning_rate": 1.1506584608200367e-06, |
|
"loss": 3.1683, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.5861262665627436, |
|
"grad_norm": 9.315102577209473, |
|
"learning_rate": 9.810017062595322e-07, |
|
"loss": 3.1097, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.5892439594699922, |
|
"grad_norm": 14.705044746398926, |
|
"learning_rate": 8.247462563808817e-07, |
|
"loss": 2.7454, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.5923616523772408, |
|
"grad_norm": 8.641880989074707, |
|
"learning_rate": 6.819348298638839e-07, |
|
"loss": 2.7142, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.5954793452844894, |
|
"grad_norm": 11.792092323303223, |
|
"learning_rate": 5.526064699265753e-07, |
|
"loss": 2.6724, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.5985970381917382, |
|
"grad_norm": 10.994759559631348, |
|
"learning_rate": 4.367965336512403e-07, |
|
"loss": 2.4745, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.6017147310989868, |
|
"grad_norm": 13.135787010192871, |
|
"learning_rate": 3.3453668231809286e-07, |
|
"loss": 2.7174, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.6048324240062354, |
|
"grad_norm": 10.710105895996094, |
|
"learning_rate": 2.458548727494292e-07, |
|
"loss": 3.0424, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.607950116913484, |
|
"grad_norm": 11.313312530517578, |
|
"learning_rate": 1.7077534966650766e-07, |
|
"loss": 2.7834, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.6110678098207326, |
|
"grad_norm": 12.285223007202148, |
|
"learning_rate": 1.0931863906127327e-07, |
|
"loss": 2.9568, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.6141855027279813, |
|
"grad_norm": 11.848814964294434, |
|
"learning_rate": 6.150154258476315e-08, |
|
"loss": 2.5463, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.61730319563523, |
|
"grad_norm": 13.349742889404297, |
|
"learning_rate": 2.7337132953697554e-08, |
|
"loss": 3.4006, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.6204208885424786, |
|
"grad_norm": 16.632017135620117, |
|
"learning_rate": 6.834750376549792e-09, |
|
"loss": 3.4796, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.6235385814497272, |
|
"grad_norm": 19.081388473510742, |
|
"learning_rate": 0.0, |
|
"loss": 3.6177, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.6235385814497272, |
|
"eval_loss": 0.5150178074836731, |
|
"eval_runtime": 20.4207, |
|
"eval_samples_per_second": 26.444, |
|
"eval_steps_per_second": 6.611, |
|
"step": 200 |
|
} |
|
], |
|
"logging_steps": 1, |
|
"max_steps": 200, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 50, |
|
"stateful_callbacks": { |
|
"EarlyStoppingCallback": { |
|
"args": { |
|
"early_stopping_patience": 5, |
|
"early_stopping_threshold": 0.0 |
|
}, |
|
"attributes": { |
|
"early_stopping_patience_counter": 0 |
|
} |
|
}, |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": true |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 1.169711928705024e+17, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|