|
{ |
|
"best_metric": 2.5416412353515625, |
|
"best_model_checkpoint": "miner_id_24/checkpoint-200", |
|
"epoch": 0.13717421124828533, |
|
"eval_steps": 50, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0006858710562414266, |
|
"grad_norm": 69.58672332763672, |
|
"learning_rate": 1e-05, |
|
"loss": 12.5759, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0006858710562414266, |
|
"eval_loss": 4.443713188171387, |
|
"eval_runtime": 27.7032, |
|
"eval_samples_per_second": 88.654, |
|
"eval_steps_per_second": 22.164, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0013717421124828531, |
|
"grad_norm": 71.16844940185547, |
|
"learning_rate": 2e-05, |
|
"loss": 13.0793, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.00205761316872428, |
|
"grad_norm": 63.57040023803711, |
|
"learning_rate": 3e-05, |
|
"loss": 13.3001, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.0027434842249657062, |
|
"grad_norm": 71.31071472167969, |
|
"learning_rate": 4e-05, |
|
"loss": 13.4363, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.003429355281207133, |
|
"grad_norm": 61.984378814697266, |
|
"learning_rate": 5e-05, |
|
"loss": 13.1256, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.00411522633744856, |
|
"grad_norm": 60.95974349975586, |
|
"learning_rate": 6e-05, |
|
"loss": 12.2845, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.004801097393689987, |
|
"grad_norm": 70.80992889404297, |
|
"learning_rate": 7e-05, |
|
"loss": 12.971, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.0054869684499314125, |
|
"grad_norm": 72.67168426513672, |
|
"learning_rate": 8e-05, |
|
"loss": 13.1652, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.006172839506172839, |
|
"grad_norm": 69.95319366455078, |
|
"learning_rate": 9e-05, |
|
"loss": 12.9721, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.006858710562414266, |
|
"grad_norm": 66.83712768554688, |
|
"learning_rate": 0.0001, |
|
"loss": 12.2386, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.0075445816186556925, |
|
"grad_norm": 72.75485229492188, |
|
"learning_rate": 9.999316524962345e-05, |
|
"loss": 12.2357, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.00823045267489712, |
|
"grad_norm": 92.31208038330078, |
|
"learning_rate": 9.997266286704631e-05, |
|
"loss": 12.7908, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.008916323731138546, |
|
"grad_norm": 65.39048767089844, |
|
"learning_rate": 9.993849845741524e-05, |
|
"loss": 12.2497, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.009602194787379973, |
|
"grad_norm": 68.49154663085938, |
|
"learning_rate": 9.989068136093873e-05, |
|
"loss": 12.6654, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.0102880658436214, |
|
"grad_norm": 66.04161834716797, |
|
"learning_rate": 9.98292246503335e-05, |
|
"loss": 12.4552, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.010973936899862825, |
|
"grad_norm": 56.68210983276367, |
|
"learning_rate": 9.975414512725057e-05, |
|
"loss": 12.4398, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.011659807956104253, |
|
"grad_norm": 49.466453552246094, |
|
"learning_rate": 9.966546331768191e-05, |
|
"loss": 12.1389, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.012345679012345678, |
|
"grad_norm": 46.36783981323242, |
|
"learning_rate": 9.956320346634876e-05, |
|
"loss": 12.2582, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.013031550068587106, |
|
"grad_norm": 51.09010314941406, |
|
"learning_rate": 9.944739353007344e-05, |
|
"loss": 12.5627, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.013717421124828532, |
|
"grad_norm": 52.49543380737305, |
|
"learning_rate": 9.931806517013612e-05, |
|
"loss": 12.6529, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.01440329218106996, |
|
"grad_norm": 50.82239532470703, |
|
"learning_rate": 9.917525374361912e-05, |
|
"loss": 12.0748, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.015089163237311385, |
|
"grad_norm": 72.02286529541016, |
|
"learning_rate": 9.901899829374047e-05, |
|
"loss": 12.1251, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.01577503429355281, |
|
"grad_norm": 60.11018371582031, |
|
"learning_rate": 9.884934153917997e-05, |
|
"loss": 12.3046, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.01646090534979424, |
|
"grad_norm": 64.09599304199219, |
|
"learning_rate": 9.86663298624003e-05, |
|
"loss": 12.5902, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.017146776406035666, |
|
"grad_norm": 63.138282775878906, |
|
"learning_rate": 9.847001329696653e-05, |
|
"loss": 11.8614, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.01783264746227709, |
|
"grad_norm": 63.069915771484375, |
|
"learning_rate": 9.826044551386744e-05, |
|
"loss": 11.8111, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.018518518518518517, |
|
"grad_norm": 69.37245178222656, |
|
"learning_rate": 9.803768380684242e-05, |
|
"loss": 11.8316, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.019204389574759947, |
|
"grad_norm": 78.47577667236328, |
|
"learning_rate": 9.780178907671789e-05, |
|
"loss": 11.7579, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.019890260631001373, |
|
"grad_norm": 73.54617309570312, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 12.3519, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.0205761316872428, |
|
"grad_norm": 77.54486846923828, |
|
"learning_rate": 9.729086208503174e-05, |
|
"loss": 12.1354, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.021262002743484224, |
|
"grad_norm": 93.95669555664062, |
|
"learning_rate": 9.701596950580806e-05, |
|
"loss": 11.8551, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.02194787379972565, |
|
"grad_norm": 103.21463012695312, |
|
"learning_rate": 9.672822322997305e-05, |
|
"loss": 11.2373, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.02263374485596708, |
|
"grad_norm": 94.34049224853516, |
|
"learning_rate": 9.642770192448536e-05, |
|
"loss": 12.1117, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.023319615912208505, |
|
"grad_norm": 111.072021484375, |
|
"learning_rate": 9.611448774886924e-05, |
|
"loss": 11.9094, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.02400548696844993, |
|
"grad_norm": 126.38433074951172, |
|
"learning_rate": 9.578866633275288e-05, |
|
"loss": 11.4785, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.024691358024691357, |
|
"grad_norm": 184.5917510986328, |
|
"learning_rate": 9.545032675245813e-05, |
|
"loss": 11.7936, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.025377229080932786, |
|
"grad_norm": 164.9213409423828, |
|
"learning_rate": 9.509956150664796e-05, |
|
"loss": 13.2435, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.02606310013717421, |
|
"grad_norm": 159.38638305664062, |
|
"learning_rate": 9.473646649103818e-05, |
|
"loss": 12.3651, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.026748971193415638, |
|
"grad_norm": 141.71275329589844, |
|
"learning_rate": 9.43611409721806e-05, |
|
"loss": 12.5131, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.027434842249657063, |
|
"grad_norm": 144.12167358398438, |
|
"learning_rate": 9.397368756032445e-05, |
|
"loss": 12.9262, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.028120713305898493, |
|
"grad_norm": 126.0718994140625, |
|
"learning_rate": 9.357421218136386e-05, |
|
"loss": 11.7807, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.02880658436213992, |
|
"grad_norm": 106.75911712646484, |
|
"learning_rate": 9.316282404787871e-05, |
|
"loss": 11.3693, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.029492455418381344, |
|
"grad_norm": 135.99990844726562, |
|
"learning_rate": 9.273963562927695e-05, |
|
"loss": 12.1318, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.03017832647462277, |
|
"grad_norm": 119.16165924072266, |
|
"learning_rate": 9.230476262104677e-05, |
|
"loss": 12.5354, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.030864197530864196, |
|
"grad_norm": 112.04950714111328, |
|
"learning_rate": 9.185832391312644e-05, |
|
"loss": 12.0471, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.03155006858710562, |
|
"grad_norm": 80.7934799194336, |
|
"learning_rate": 9.140044155740101e-05, |
|
"loss": 10.8909, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.03223593964334705, |
|
"grad_norm": 89.15470886230469, |
|
"learning_rate": 9.093124073433463e-05, |
|
"loss": 13.3376, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.03292181069958848, |
|
"grad_norm": 99.22161102294922, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 11.9401, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.033607681755829906, |
|
"grad_norm": 106.7686996459961, |
|
"learning_rate": 8.995939984474624e-05, |
|
"loss": 11.6899, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.03429355281207133, |
|
"grad_norm": 109.92302703857422, |
|
"learning_rate": 8.945702546981969e-05, |
|
"loss": 14.7829, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.03429355281207133, |
|
"eval_loss": 2.9941635131835938, |
|
"eval_runtime": 27.7028, |
|
"eval_samples_per_second": 88.655, |
|
"eval_steps_per_second": 22.164, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.03497942386831276, |
|
"grad_norm": 97.6561279296875, |
|
"learning_rate": 8.894386393810563e-05, |
|
"loss": 11.9808, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.03566529492455418, |
|
"grad_norm": 96.15821838378906, |
|
"learning_rate": 8.842005554284296e-05, |
|
"loss": 11.6438, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.03635116598079561, |
|
"grad_norm": 93.45790100097656, |
|
"learning_rate": 8.788574348801675e-05, |
|
"loss": 11.8972, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.037037037037037035, |
|
"grad_norm": 99.56314849853516, |
|
"learning_rate": 8.73410738492077e-05, |
|
"loss": 11.5736, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.03772290809327846, |
|
"grad_norm": 106.13288879394531, |
|
"learning_rate": 8.678619553365659e-05, |
|
"loss": 11.4556, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.038408779149519894, |
|
"grad_norm": 96.12564086914062, |
|
"learning_rate": 8.622126023955446e-05, |
|
"loss": 11.1414, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.03909465020576132, |
|
"grad_norm": 90.1689224243164, |
|
"learning_rate": 8.564642241456986e-05, |
|
"loss": 10.6768, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.039780521262002745, |
|
"grad_norm": 87.85713195800781, |
|
"learning_rate": 8.506183921362443e-05, |
|
"loss": 10.9586, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.04046639231824417, |
|
"grad_norm": 97.65484619140625, |
|
"learning_rate": 8.44676704559283e-05, |
|
"loss": 11.26, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.0411522633744856, |
|
"grad_norm": 86.68466186523438, |
|
"learning_rate": 8.386407858128706e-05, |
|
"loss": 10.6433, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.04183813443072702, |
|
"grad_norm": 90.71621704101562, |
|
"learning_rate": 8.32512286056924e-05, |
|
"loss": 10.9786, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.04252400548696845, |
|
"grad_norm": 102.1913070678711, |
|
"learning_rate": 8.262928807620843e-05, |
|
"loss": 11.2173, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.043209876543209874, |
|
"grad_norm": 107.4554672241211, |
|
"learning_rate": 8.199842702516583e-05, |
|
"loss": 11.4049, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.0438957475994513, |
|
"grad_norm": 127.98192596435547, |
|
"learning_rate": 8.135881792367686e-05, |
|
"loss": 11.8326, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.04458161865569273, |
|
"grad_norm": 142.61886596679688, |
|
"learning_rate": 8.07106356344834e-05, |
|
"loss": 10.777, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.04526748971193416, |
|
"grad_norm": 137.0384979248047, |
|
"learning_rate": 8.005405736415126e-05, |
|
"loss": 10.8837, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.045953360768175584, |
|
"grad_norm": 104.60083770751953, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 10.6117, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.04663923182441701, |
|
"grad_norm": 144.4799041748047, |
|
"learning_rate": 7.871643313414718e-05, |
|
"loss": 10.7832, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.047325102880658436, |
|
"grad_norm": 144.327392578125, |
|
"learning_rate": 7.803575286758364e-05, |
|
"loss": 10.678, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.04801097393689986, |
|
"grad_norm": 106.83956909179688, |
|
"learning_rate": 7.734740790612136e-05, |
|
"loss": 10.7744, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.04869684499314129, |
|
"grad_norm": 101.54669952392578, |
|
"learning_rate": 7.66515864363997e-05, |
|
"loss": 10.3821, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.04938271604938271, |
|
"grad_norm": 174.71713256835938, |
|
"learning_rate": 7.594847868906076e-05, |
|
"loss": 11.2289, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.050068587105624146, |
|
"grad_norm": 106.94820404052734, |
|
"learning_rate": 7.52382768867422e-05, |
|
"loss": 10.9244, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.05075445816186557, |
|
"grad_norm": 89.95751953125, |
|
"learning_rate": 7.452117519152542e-05, |
|
"loss": 10.5059, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.051440329218107, |
|
"grad_norm": 103.03185272216797, |
|
"learning_rate": 7.379736965185368e-05, |
|
"loss": 11.336, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.05212620027434842, |
|
"grad_norm": 102.09061431884766, |
|
"learning_rate": 7.30670581489344e-05, |
|
"loss": 10.6842, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.05281207133058985, |
|
"grad_norm": 92.91813659667969, |
|
"learning_rate": 7.233044034264034e-05, |
|
"loss": 10.1746, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.053497942386831275, |
|
"grad_norm": 124.47396087646484, |
|
"learning_rate": 7.158771761692464e-05, |
|
"loss": 11.1656, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.0541838134430727, |
|
"grad_norm": 151.16917419433594, |
|
"learning_rate": 7.083909302476453e-05, |
|
"loss": 11.1178, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.05486968449931413, |
|
"grad_norm": 177.20263671875, |
|
"learning_rate": 7.008477123264848e-05, |
|
"loss": 10.3579, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.05555555555555555, |
|
"grad_norm": 118.05079650878906, |
|
"learning_rate": 6.932495846462261e-05, |
|
"loss": 10.2433, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.056241426611796985, |
|
"grad_norm": 166.85604858398438, |
|
"learning_rate": 6.855986244591104e-05, |
|
"loss": 11.4896, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.05692729766803841, |
|
"grad_norm": 128.1403350830078, |
|
"learning_rate": 6.778969234612584e-05, |
|
"loss": 10.527, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.05761316872427984, |
|
"grad_norm": 146.47117614746094, |
|
"learning_rate": 6.701465872208216e-05, |
|
"loss": 11.5244, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.05829903978052126, |
|
"grad_norm": 142.8417205810547, |
|
"learning_rate": 6.623497346023418e-05, |
|
"loss": 10.6936, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.05898491083676269, |
|
"grad_norm": 108.32695007324219, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 11.2923, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.059670781893004114, |
|
"grad_norm": 158.59515380859375, |
|
"learning_rate": 6.466250186922325e-05, |
|
"loss": 10.1826, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.06035665294924554, |
|
"grad_norm": 112.02003479003906, |
|
"learning_rate": 6.387014543809223e-05, |
|
"loss": 10.3102, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.061042524005486966, |
|
"grad_norm": 155.01995849609375, |
|
"learning_rate": 6.307399704769099e-05, |
|
"loss": 11.0932, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.06172839506172839, |
|
"grad_norm": 132.44480895996094, |
|
"learning_rate": 6.227427435703997e-05, |
|
"loss": 10.1653, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.062414266117969824, |
|
"grad_norm": 109.92094421386719, |
|
"learning_rate": 6.147119600233758e-05, |
|
"loss": 11.6779, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.06310013717421124, |
|
"grad_norm": 106.69784545898438, |
|
"learning_rate": 6.066498153718735e-05, |
|
"loss": 9.6017, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.06378600823045268, |
|
"grad_norm": 164.68478393554688, |
|
"learning_rate": 5.985585137257401e-05, |
|
"loss": 10.2169, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.0644718792866941, |
|
"grad_norm": 114.99446868896484, |
|
"learning_rate": 5.90440267166055e-05, |
|
"loss": 11.2601, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.06515775034293553, |
|
"grad_norm": 98.81483459472656, |
|
"learning_rate": 5.8229729514036705e-05, |
|
"loss": 11.1502, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.06584362139917696, |
|
"grad_norm": 98.5564193725586, |
|
"learning_rate": 5.74131823855921e-05, |
|
"loss": 10.825, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.06652949245541838, |
|
"grad_norm": 109.987548828125, |
|
"learning_rate": 5.6594608567103456e-05, |
|
"loss": 11.2937, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.06721536351165981, |
|
"grad_norm": 117.61621856689453, |
|
"learning_rate": 5.577423184847932e-05, |
|
"loss": 11.0617, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.06790123456790123, |
|
"grad_norm": 83.48066711425781, |
|
"learning_rate": 5.495227651252315e-05, |
|
"loss": 11.9516, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.06858710562414266, |
|
"grad_norm": 84.93854522705078, |
|
"learning_rate": 5.4128967273616625e-05, |
|
"loss": 13.0961, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.06858710562414266, |
|
"eval_loss": 2.767896890640259, |
|
"eval_runtime": 27.6432, |
|
"eval_samples_per_second": 88.846, |
|
"eval_steps_per_second": 22.212, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.06927297668038408, |
|
"grad_norm": 116.80013275146484, |
|
"learning_rate": 5.330452921628497e-05, |
|
"loss": 10.8994, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.06995884773662552, |
|
"grad_norm": 102.24729919433594, |
|
"learning_rate": 5.247918773366112e-05, |
|
"loss": 10.5693, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.07064471879286695, |
|
"grad_norm": 118.9500503540039, |
|
"learning_rate": 5.165316846586541e-05, |
|
"loss": 11.0576, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.07133058984910837, |
|
"grad_norm": 114.09745025634766, |
|
"learning_rate": 5.0826697238317935e-05, |
|
"loss": 10.9041, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.0720164609053498, |
|
"grad_norm": 101.61298370361328, |
|
"learning_rate": 5e-05, |
|
"loss": 11.2616, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.07270233196159122, |
|
"grad_norm": 104.95634460449219, |
|
"learning_rate": 4.917330276168208e-05, |
|
"loss": 10.8302, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.07338820301783265, |
|
"grad_norm": 120.04218292236328, |
|
"learning_rate": 4.834683153413459e-05, |
|
"loss": 10.573, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.07407407407407407, |
|
"grad_norm": 133.39187622070312, |
|
"learning_rate": 4.7520812266338885e-05, |
|
"loss": 10.9659, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.0747599451303155, |
|
"grad_norm": 106.15190124511719, |
|
"learning_rate": 4.669547078371504e-05, |
|
"loss": 10.9749, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.07544581618655692, |
|
"grad_norm": 121.06906127929688, |
|
"learning_rate": 4.5871032726383386e-05, |
|
"loss": 11.165, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.07613168724279835, |
|
"grad_norm": 113.51752471923828, |
|
"learning_rate": 4.504772348747687e-05, |
|
"loss": 10.6913, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.07681755829903979, |
|
"grad_norm": 111.4140853881836, |
|
"learning_rate": 4.4225768151520694e-05, |
|
"loss": 10.9188, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.0775034293552812, |
|
"grad_norm": 108.21040344238281, |
|
"learning_rate": 4.3405391432896555e-05, |
|
"loss": 10.9079, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.07818930041152264, |
|
"grad_norm": 126.95697784423828, |
|
"learning_rate": 4.2586817614407895e-05, |
|
"loss": 11.1894, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.07887517146776406, |
|
"grad_norm": 118.61093139648438, |
|
"learning_rate": 4.17702704859633e-05, |
|
"loss": 10.6185, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.07956104252400549, |
|
"grad_norm": 83.73309326171875, |
|
"learning_rate": 4.095597328339452e-05, |
|
"loss": 10.6591, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.08024691358024691, |
|
"grad_norm": 98.3722152709961, |
|
"learning_rate": 4.0144148627425993e-05, |
|
"loss": 10.917, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.08093278463648834, |
|
"grad_norm": 83.13345336914062, |
|
"learning_rate": 3.933501846281267e-05, |
|
"loss": 10.42, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.08161865569272976, |
|
"grad_norm": 92.0757827758789, |
|
"learning_rate": 3.852880399766243e-05, |
|
"loss": 10.4654, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.0823045267489712, |
|
"grad_norm": 94.01953125, |
|
"learning_rate": 3.772572564296005e-05, |
|
"loss": 10.0288, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.08299039780521263, |
|
"grad_norm": 71.66029357910156, |
|
"learning_rate": 3.6926002952309016e-05, |
|
"loss": 10.1073, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.08367626886145405, |
|
"grad_norm": 76.70323181152344, |
|
"learning_rate": 3.612985456190778e-05, |
|
"loss": 11.1978, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.08436213991769548, |
|
"grad_norm": 89.28335571289062, |
|
"learning_rate": 3.533749813077677e-05, |
|
"loss": 10.6591, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.0850480109739369, |
|
"grad_norm": 79.22374725341797, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 10.958, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.08573388203017833, |
|
"grad_norm": 81.43111419677734, |
|
"learning_rate": 3.3765026539765834e-05, |
|
"loss": 10.8824, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.08641975308641975, |
|
"grad_norm": 88.76519775390625, |
|
"learning_rate": 3.298534127791785e-05, |
|
"loss": 10.3693, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.08710562414266118, |
|
"grad_norm": 88.95154571533203, |
|
"learning_rate": 3.221030765387417e-05, |
|
"loss": 10.4823, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.0877914951989026, |
|
"grad_norm": 83.26634216308594, |
|
"learning_rate": 3.144013755408895e-05, |
|
"loss": 10.3776, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.08847736625514403, |
|
"grad_norm": 71.29374694824219, |
|
"learning_rate": 3.0675041535377405e-05, |
|
"loss": 10.2486, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.08916323731138547, |
|
"grad_norm": 74.10224151611328, |
|
"learning_rate": 2.991522876735154e-05, |
|
"loss": 9.7498, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.08984910836762688, |
|
"grad_norm": 93.9678955078125, |
|
"learning_rate": 2.916090697523549e-05, |
|
"loss": 11.2849, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.09053497942386832, |
|
"grad_norm": 90.89468383789062, |
|
"learning_rate": 2.8412282383075363e-05, |
|
"loss": 10.5972, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.09122085048010974, |
|
"grad_norm": 89.09439849853516, |
|
"learning_rate": 2.766955965735968e-05, |
|
"loss": 11.0438, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.09190672153635117, |
|
"grad_norm": 83.01602935791016, |
|
"learning_rate": 2.693294185106562e-05, |
|
"loss": 11.3951, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.09259259259259259, |
|
"grad_norm": 97.12364196777344, |
|
"learning_rate": 2.6202630348146324e-05, |
|
"loss": 10.9804, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.09327846364883402, |
|
"grad_norm": 81.45047760009766, |
|
"learning_rate": 2.547882480847461e-05, |
|
"loss": 10.2185, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.09396433470507544, |
|
"grad_norm": 104.84080505371094, |
|
"learning_rate": 2.476172311325783e-05, |
|
"loss": 10.3834, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.09465020576131687, |
|
"grad_norm": 87.71990203857422, |
|
"learning_rate": 2.405152131093926e-05, |
|
"loss": 11.7378, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.0953360768175583, |
|
"grad_norm": 88.87088012695312, |
|
"learning_rate": 2.3348413563600325e-05, |
|
"loss": 10.5149, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.09602194787379972, |
|
"grad_norm": 87.34343719482422, |
|
"learning_rate": 2.2652592093878666e-05, |
|
"loss": 11.413, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.09670781893004116, |
|
"grad_norm": 77.03406524658203, |
|
"learning_rate": 2.196424713241637e-05, |
|
"loss": 9.7402, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.09739368998628258, |
|
"grad_norm": 89.15251922607422, |
|
"learning_rate": 2.128356686585282e-05, |
|
"loss": 9.4749, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.09807956104252401, |
|
"grad_norm": 93.92284393310547, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 10.3826, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.09876543209876543, |
|
"grad_norm": 76.36550903320312, |
|
"learning_rate": 1.9945942635848748e-05, |
|
"loss": 9.8798, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.09945130315500686, |
|
"grad_norm": 70.09993743896484, |
|
"learning_rate": 1.928936436551661e-05, |
|
"loss": 9.4255, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.10013717421124829, |
|
"grad_norm": 117.6426773071289, |
|
"learning_rate": 1.8641182076323148e-05, |
|
"loss": 11.0201, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.10082304526748971, |
|
"grad_norm": 83.5998306274414, |
|
"learning_rate": 1.800157297483417e-05, |
|
"loss": 10.8441, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.10150891632373114, |
|
"grad_norm": 81.80438232421875, |
|
"learning_rate": 1.7370711923791567e-05, |
|
"loss": 10.3606, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.10219478737997256, |
|
"grad_norm": 65.09898376464844, |
|
"learning_rate": 1.6748771394307585e-05, |
|
"loss": 9.8531, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.102880658436214, |
|
"grad_norm": 74.46653747558594, |
|
"learning_rate": 1.6135921418712956e-05, |
|
"loss": 10.4314, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.102880658436214, |
|
"eval_loss": 2.5873875617980957, |
|
"eval_runtime": 27.6802, |
|
"eval_samples_per_second": 88.728, |
|
"eval_steps_per_second": 22.182, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.10356652949245541, |
|
"grad_norm": 70.4255599975586, |
|
"learning_rate": 1.553232954407171e-05, |
|
"loss": 10.7022, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.10425240054869685, |
|
"grad_norm": 72.38617706298828, |
|
"learning_rate": 1.4938160786375572e-05, |
|
"loss": 10.7702, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.10493827160493827, |
|
"grad_norm": 69.93809509277344, |
|
"learning_rate": 1.435357758543015e-05, |
|
"loss": 10.4283, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.1056241426611797, |
|
"grad_norm": 77.07926177978516, |
|
"learning_rate": 1.3778739760445552e-05, |
|
"loss": 10.2906, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.10631001371742113, |
|
"grad_norm": 70.95220184326172, |
|
"learning_rate": 1.3213804466343421e-05, |
|
"loss": 10.3806, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.10699588477366255, |
|
"grad_norm": 65.11528015136719, |
|
"learning_rate": 1.2658926150792322e-05, |
|
"loss": 10.3674, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.10768175582990398, |
|
"grad_norm": 67.45489501953125, |
|
"learning_rate": 1.2114256511983274e-05, |
|
"loss": 10.4724, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.1083676268861454, |
|
"grad_norm": 67.21087646484375, |
|
"learning_rate": 1.157994445715706e-05, |
|
"loss": 10.1885, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.10905349794238683, |
|
"grad_norm": 71.32520294189453, |
|
"learning_rate": 1.1056136061894384e-05, |
|
"loss": 10.1943, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.10973936899862825, |
|
"grad_norm": 70.40990447998047, |
|
"learning_rate": 1.0542974530180327e-05, |
|
"loss": 10.1976, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.11042524005486969, |
|
"grad_norm": 68.82344818115234, |
|
"learning_rate": 1.0040600155253765e-05, |
|
"loss": 9.78, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.1111111111111111, |
|
"grad_norm": 66.81336975097656, |
|
"learning_rate": 9.549150281252633e-06, |
|
"loss": 9.8149, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.11179698216735254, |
|
"grad_norm": 69.47315979003906, |
|
"learning_rate": 9.068759265665384e-06, |
|
"loss": 10.1638, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.11248285322359397, |
|
"grad_norm": 68.41928100585938, |
|
"learning_rate": 8.599558442598998e-06, |
|
"loss": 10.8401, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.11316872427983539, |
|
"grad_norm": 68.94377136230469, |
|
"learning_rate": 8.141676086873572e-06, |
|
"loss": 10.735, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.11385459533607682, |
|
"grad_norm": 64.20032501220703, |
|
"learning_rate": 7.695237378953223e-06, |
|
"loss": 10.3732, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.11454046639231824, |
|
"grad_norm": 78.93729400634766, |
|
"learning_rate": 7.260364370723044e-06, |
|
"loss": 10.5986, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.11522633744855967, |
|
"grad_norm": 69.44711303710938, |
|
"learning_rate": 6.837175952121306e-06, |
|
"loss": 10.57, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.11591220850480109, |
|
"grad_norm": 84.74750518798828, |
|
"learning_rate": 6.425787818636131e-06, |
|
"loss": 10.4673, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.11659807956104253, |
|
"grad_norm": 63.21523666381836, |
|
"learning_rate": 6.026312439675552e-06, |
|
"loss": 9.6459, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.11728395061728394, |
|
"grad_norm": 63.81535720825195, |
|
"learning_rate": 5.6388590278194096e-06, |
|
"loss": 10.5836, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.11796982167352538, |
|
"grad_norm": 74.9435806274414, |
|
"learning_rate": 5.263533508961827e-06, |
|
"loss": 10.21, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.11865569272976681, |
|
"grad_norm": 80.28829193115234, |
|
"learning_rate": 4.900438493352055e-06, |
|
"loss": 9.9517, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.11934156378600823, |
|
"grad_norm": 56.12078857421875, |
|
"learning_rate": 4.549673247541875e-06, |
|
"loss": 10.2741, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.12002743484224966, |
|
"grad_norm": 60.06998062133789, |
|
"learning_rate": 4.2113336672471245e-06, |
|
"loss": 9.9261, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.12071330589849108, |
|
"grad_norm": 68.69584655761719, |
|
"learning_rate": 3.885512251130763e-06, |
|
"loss": 10.1702, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.12139917695473251, |
|
"grad_norm": 83.58258819580078, |
|
"learning_rate": 3.5722980755146517e-06, |
|
"loss": 10.7339, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.12208504801097393, |
|
"grad_norm": 67.84488677978516, |
|
"learning_rate": 3.271776770026963e-06, |
|
"loss": 10.5149, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.12277091906721536, |
|
"grad_norm": 73.4597396850586, |
|
"learning_rate": 2.9840304941919415e-06, |
|
"loss": 9.5782, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.12345679012345678, |
|
"grad_norm": 79.10492706298828, |
|
"learning_rate": 2.7091379149682685e-06, |
|
"loss": 10.292, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.12414266117969822, |
|
"grad_norm": 73.77743530273438, |
|
"learning_rate": 2.4471741852423237e-06, |
|
"loss": 11.2363, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.12482853223593965, |
|
"grad_norm": 59.835994720458984, |
|
"learning_rate": 2.1982109232821178e-06, |
|
"loss": 10.7522, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.12551440329218108, |
|
"grad_norm": 68.94219207763672, |
|
"learning_rate": 1.962316193157593e-06, |
|
"loss": 10.9599, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.1262002743484225, |
|
"grad_norm": 68.62545776367188, |
|
"learning_rate": 1.7395544861325718e-06, |
|
"loss": 9.8215, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.12688614540466392, |
|
"grad_norm": 71.54290771484375, |
|
"learning_rate": 1.5299867030334814e-06, |
|
"loss": 10.4858, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.12757201646090535, |
|
"grad_norm": 73.2717514038086, |
|
"learning_rate": 1.333670137599713e-06, |
|
"loss": 10.229, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.12825788751714678, |
|
"grad_norm": 83.46717834472656, |
|
"learning_rate": 1.1506584608200367e-06, |
|
"loss": 10.071, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.1289437585733882, |
|
"grad_norm": 61.62462615966797, |
|
"learning_rate": 9.810017062595322e-07, |
|
"loss": 9.9997, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.12962962962962962, |
|
"grad_norm": 65.07300567626953, |
|
"learning_rate": 8.247462563808817e-07, |
|
"loss": 9.7776, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.13031550068587106, |
|
"grad_norm": 70.35834503173828, |
|
"learning_rate": 6.819348298638839e-07, |
|
"loss": 9.5993, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.1310013717421125, |
|
"grad_norm": 68.59699249267578, |
|
"learning_rate": 5.526064699265753e-07, |
|
"loss": 10.057, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.13168724279835392, |
|
"grad_norm": 84.115966796875, |
|
"learning_rate": 4.367965336512403e-07, |
|
"loss": 9.7622, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.13237311385459533, |
|
"grad_norm": 72.11085510253906, |
|
"learning_rate": 3.3453668231809286e-07, |
|
"loss": 9.7314, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.13305898491083676, |
|
"grad_norm": 55.28156661987305, |
|
"learning_rate": 2.458548727494292e-07, |
|
"loss": 9.8484, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.1337448559670782, |
|
"grad_norm": 70.85758972167969, |
|
"learning_rate": 1.7077534966650766e-07, |
|
"loss": 9.7574, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.13443072702331962, |
|
"grad_norm": 66.05179595947266, |
|
"learning_rate": 1.0931863906127327e-07, |
|
"loss": 9.0654, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.13511659807956103, |
|
"grad_norm": 60.33279037475586, |
|
"learning_rate": 6.150154258476315e-08, |
|
"loss": 10.2341, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.13580246913580246, |
|
"grad_norm": 70.23603820800781, |
|
"learning_rate": 2.7337132953697554e-08, |
|
"loss": 9.7082, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.1364883401920439, |
|
"grad_norm": 60.87910842895508, |
|
"learning_rate": 6.834750376549792e-09, |
|
"loss": 10.2319, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.13717421124828533, |
|
"grad_norm": 88.05852508544922, |
|
"learning_rate": 0.0, |
|
"loss": 14.9627, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.13717421124828533, |
|
"eval_loss": 2.5416412353515625, |
|
"eval_runtime": 27.7018, |
|
"eval_samples_per_second": 88.659, |
|
"eval_steps_per_second": 22.165, |
|
"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": 1.49024359317504e+16, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|