| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.11757789535567313, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005878894767783657, | |
| "grad_norm": 1.2760219760821201e-05, | |
| "learning_rate": 1e-05, | |
| "loss": 11.5, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0011757789535567313, | |
| "grad_norm": 1.9830104065476917e-05, | |
| "learning_rate": 2e-05, | |
| "loss": 11.5, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001763668430335097, | |
| "grad_norm": 2.6221930966130458e-05, | |
| "learning_rate": 3e-05, | |
| "loss": 11.5, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0023515579071134627, | |
| "grad_norm": 1.1205795999558177e-05, | |
| "learning_rate": 4e-05, | |
| "loss": 11.5, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0029394473838918285, | |
| "grad_norm": 1.3619834135170095e-05, | |
| "learning_rate": 5e-05, | |
| "loss": 11.5, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.003527336860670194, | |
| "grad_norm": 1.283156052522827e-05, | |
| "learning_rate": 6e-05, | |
| "loss": 11.5, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00411522633744856, | |
| "grad_norm": 1.6724487068131566e-05, | |
| "learning_rate": 7e-05, | |
| "loss": 11.5, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.004703115814226925, | |
| "grad_norm": 1.116675412049517e-05, | |
| "learning_rate": 8e-05, | |
| "loss": 11.5, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.005291005291005291, | |
| "grad_norm": 1.5011505638540257e-05, | |
| "learning_rate": 9e-05, | |
| "loss": 11.5, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.005878894767783657, | |
| "grad_norm": 1.3602088074549101e-05, | |
| "learning_rate": 0.0001, | |
| "loss": 11.5, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.006466784244562022, | |
| "grad_norm": 1.6538682757527567e-05, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 11.5, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.007054673721340388, | |
| "grad_norm": 2.307661452505272e-05, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 11.5, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.007642563198118754, | |
| "grad_norm": 2.1508752979571e-05, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 11.5, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.00823045267489712, | |
| "grad_norm": 2.3830240024835803e-05, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 11.5, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.008818342151675485, | |
| "grad_norm": 2.025593857979402e-05, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 11.5, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.00940623162845385, | |
| "grad_norm": 2.8292639399296604e-05, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 11.5, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.009994121105232217, | |
| "grad_norm": 2.6970761609845795e-05, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 11.5, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.010582010582010581, | |
| "grad_norm": 2.4985509298858233e-05, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 11.5, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.011169900058788948, | |
| "grad_norm": 3.422730878810398e-05, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 11.5, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.011757789535567314, | |
| "grad_norm": 3.088069934165105e-05, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 11.5, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.012345679012345678, | |
| "grad_norm": 2.8788177587557584e-05, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 11.5, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.012933568489124045, | |
| "grad_norm": 3.932521212846041e-05, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 11.5, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01352145796590241, | |
| "grad_norm": 2.8881346224807203e-05, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 11.5, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.014109347442680775, | |
| "grad_norm": 2.785995093290694e-05, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 11.5, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.014697236919459141, | |
| "grad_norm": 4.185497891739942e-05, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 11.5, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.015285126396237508, | |
| "grad_norm": 3.3871710911625996e-05, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 11.5, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.015873015873015872, | |
| "grad_norm": 3.2904215913731605e-05, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 11.5, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.01646090534979424, | |
| "grad_norm": 4.673474904848263e-05, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 11.5, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.017048794826572605, | |
| "grad_norm": 5.259223689790815e-05, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 11.5, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01763668430335097, | |
| "grad_norm": 6.387756002368405e-05, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 11.5, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.018224573780129337, | |
| "grad_norm": 5.021028846385889e-05, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 11.5, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0188124632569077, | |
| "grad_norm": 4.695380266639404e-05, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 11.5, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.019400352733686066, | |
| "grad_norm": 5.083639553049579e-05, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 11.5, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.019988242210464434, | |
| "grad_norm": 3.5037428460782394e-05, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 11.5, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0205761316872428, | |
| "grad_norm": 6.02395593887195e-05, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 11.5, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.021164021164021163, | |
| "grad_norm": 5.03252049384173e-05, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 11.5, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.02175191064079953, | |
| "grad_norm": 4.802745388587937e-05, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 11.5, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.022339800117577895, | |
| "grad_norm": 7.288879714906216e-05, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 11.5, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.02292768959435626, | |
| "grad_norm": 7.677981921005994e-05, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 11.5, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.023515579071134628, | |
| "grad_norm": 6.031816519680433e-05, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 11.5, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.024103468547912992, | |
| "grad_norm": 5.640002200379968e-05, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 11.5, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.024691358024691357, | |
| "grad_norm": 7.154404738685116e-05, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 11.5, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.025279247501469725, | |
| "grad_norm": 7.722323789494112e-05, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 11.5, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.02586713697824809, | |
| "grad_norm": 7.110222213668749e-05, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 11.5, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.026455026455026454, | |
| "grad_norm": 6.41928636468947e-05, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 11.5, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02704291593180482, | |
| "grad_norm": 7.909386476967484e-05, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 11.5, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.027630805408583186, | |
| "grad_norm": 8.1085272540804e-05, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 11.5, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02821869488536155, | |
| "grad_norm": 0.00010535357432672754, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 11.5, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.02880658436213992, | |
| "grad_norm": 9.938155562849715e-05, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 11.5, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.029394473838918283, | |
| "grad_norm": 0.00017812225269153714, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 11.5, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.029394473838918283, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 13.7532, | |
| "eval_samples_per_second": 208.315, | |
| "eval_steps_per_second": 52.133, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.029982363315696647, | |
| "grad_norm": 8.085326408036053e-05, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 11.5, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.030570252792475015, | |
| "grad_norm": 6.59500656183809e-05, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 11.5, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03115814226925338, | |
| "grad_norm": 9.271319868275896e-05, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 11.5, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.031746031746031744, | |
| "grad_norm": 7.636727241333574e-05, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 11.5, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03233392122281011, | |
| "grad_norm": 8.564830932300538e-05, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 11.5, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03292181069958848, | |
| "grad_norm": 6.765264697605744e-05, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 11.5, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03350970017636684, | |
| "grad_norm": 8.938118844525889e-05, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 11.5, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03409758965314521, | |
| "grad_norm": 7.598628872074187e-05, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 11.5, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03468547912992358, | |
| "grad_norm": 9.138462337432429e-05, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 11.5, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03527336860670194, | |
| "grad_norm": 9.17932266020216e-05, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 11.5, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.035861258083480306, | |
| "grad_norm": 0.00011113197251688689, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 11.5, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.036449147560258674, | |
| "grad_norm": 0.0001215208467328921, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 11.5, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.037037037037037035, | |
| "grad_norm": 0.00011370470747351646, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 11.5, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0376249265138154, | |
| "grad_norm": 0.00010293946252204478, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 11.5, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03821281599059377, | |
| "grad_norm": 0.00010239002585876733, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 11.5, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03880070546737213, | |
| "grad_norm": 0.00011284730135230348, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 11.5, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0393885949441505, | |
| "grad_norm": 0.00010800938616739586, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 11.5, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03997648442092887, | |
| "grad_norm": 0.00012077321298420429, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 11.5, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.04056437389770723, | |
| "grad_norm": 9.716762724565342e-05, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 11.5, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0411522633744856, | |
| "grad_norm": 9.754186612553895e-05, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 11.5, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.041740152851263965, | |
| "grad_norm": 0.00013152298924978822, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 11.5, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.042328042328042326, | |
| "grad_norm": 0.00013756412954535335, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 11.5, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.042915931804820694, | |
| "grad_norm": 0.00011569081107154489, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 11.5, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.04350382128159906, | |
| "grad_norm": 0.00012273195898160338, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 11.5, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.04409171075837742, | |
| "grad_norm": 0.00012811228225473315, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 11.5, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04467960023515579, | |
| "grad_norm": 0.00012965736095793545, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 11.5, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.04526748971193416, | |
| "grad_norm": 0.00012737682845909148, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 11.5, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.04585537918871252, | |
| "grad_norm": 0.00015797096421010792, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 11.5, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.04644326866549089, | |
| "grad_norm": 0.00022033862478565425, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 11.5, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.047031158142269255, | |
| "grad_norm": 0.0001866580860223621, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 11.5, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.047619047619047616, | |
| "grad_norm": 0.00018126929353456944, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 11.5, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.048206937095825984, | |
| "grad_norm": 0.00021677513723261654, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 11.5, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.04879482657260435, | |
| "grad_norm": 0.0002002787950914353, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 11.5, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.04938271604938271, | |
| "grad_norm": 0.0001375633291900158, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 11.5, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.04997060552616108, | |
| "grad_norm": 0.0001406855444656685, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 11.5, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.05055849500293945, | |
| "grad_norm": 0.00016629204037599266, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 11.5, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.05114638447971781, | |
| "grad_norm": 0.0001735340483719483, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 11.5, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.05173427395649618, | |
| "grad_norm": 0.0001282371231354773, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 11.5, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.052322163433274546, | |
| "grad_norm": 0.00019439813331700861, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 11.5, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.05291005291005291, | |
| "grad_norm": 0.0002183008036809042, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 11.5, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.053497942386831275, | |
| "grad_norm": 0.0002012772747548297, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 11.5, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.05408583186360964, | |
| "grad_norm": 0.00022741183056496084, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 11.5, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.054673721340388004, | |
| "grad_norm": 0.00018610736879054457, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 11.5, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.05526161081716637, | |
| "grad_norm": 0.00023018394131213427, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 11.5, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.05584950029394474, | |
| "grad_norm": 0.0001871351123554632, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 11.5, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0564373897707231, | |
| "grad_norm": 0.00014184312021825463, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 11.5, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.05702527924750147, | |
| "grad_norm": 0.00019797985441982746, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 11.5, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05761316872427984, | |
| "grad_norm": 0.0002441562246531248, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 11.5, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0582010582010582, | |
| "grad_norm": 0.0001735905243549496, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 11.5, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.058788947677836566, | |
| "grad_norm": 0.00039850923349149525, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 11.5, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.058788947677836566, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 14.1119, | |
| "eval_samples_per_second": 203.02, | |
| "eval_steps_per_second": 50.808, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.059376837154614934, | |
| "grad_norm": 0.00023723703634459525, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 11.5, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.059964726631393295, | |
| "grad_norm": 0.00020033006148878485, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 11.5, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.06055261610817166, | |
| "grad_norm": 0.00019261125999037176, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 11.5, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.06114050558495003, | |
| "grad_norm": 0.00016912569117266685, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 11.5, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.06172839506172839, | |
| "grad_norm": 0.00015685826656408608, | |
| "learning_rate": 5e-05, | |
| "loss": 11.5, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.06231628453850676, | |
| "grad_norm": 0.0002085861488012597, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 11.5, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.06290417401528513, | |
| "grad_norm": 0.00018315795750822872, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 11.5, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.06349206349206349, | |
| "grad_norm": 0.00023191703076008707, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 11.5, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.06407995296884186, | |
| "grad_norm": 0.00026227752096019685, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 11.5, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.06466784244562022, | |
| "grad_norm": 0.00023102114209905267, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 11.5, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06525573192239859, | |
| "grad_norm": 0.00024366994330193847, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 11.5, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.06584362139917696, | |
| "grad_norm": 0.00025131963775493205, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 11.5, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.06643151087595532, | |
| "grad_norm": 0.00024602666962891817, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 11.5, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.06701940035273368, | |
| "grad_norm": 0.00025449582608416677, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 11.5, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.06760728982951206, | |
| "grad_norm": 0.0003207982226740569, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 11.5, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06819517930629042, | |
| "grad_norm": 0.0002616269630379975, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 11.5, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06878306878306878, | |
| "grad_norm": 0.0003154428268317133, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 11.5, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06937095825984715, | |
| "grad_norm": 0.00033450601040385664, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 11.5, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.06995884773662552, | |
| "grad_norm": 0.00028725163429044187, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 11.5, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.07054673721340388, | |
| "grad_norm": 0.00020756223239004612, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 11.5, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07113462669018225, | |
| "grad_norm": 0.0002340238424949348, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 11.5, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.07172251616696061, | |
| "grad_norm": 0.000275884463917464, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 11.5, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.07231040564373897, | |
| "grad_norm": 0.0002855132916010916, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 11.5, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.07289829512051735, | |
| "grad_norm": 0.00037962698843330145, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 11.5, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.07348618459729571, | |
| "grad_norm": 0.00030194406281225383, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 11.5, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07407407407407407, | |
| "grad_norm": 0.0002575613325461745, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 11.5, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.07466196355085244, | |
| "grad_norm": 0.000314484495902434, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 11.5, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0752498530276308, | |
| "grad_norm": 0.0002988316700793803, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 11.5, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.07583774250440917, | |
| "grad_norm": 0.0003064018383156508, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 11.5, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.07642563198118754, | |
| "grad_norm": 0.0002705864899326116, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 11.5, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0770135214579659, | |
| "grad_norm": 0.00034659693483263254, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 11.5, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07760141093474426, | |
| "grad_norm": 0.00021152559202164412, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 11.5, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.07818930041152264, | |
| "grad_norm": 0.00030989633523859084, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 11.5, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.078777189888301, | |
| "grad_norm": 0.00028353952802717686, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 11.5, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.07936507936507936, | |
| "grad_norm": 0.0003137715393677354, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 11.5, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07995296884185774, | |
| "grad_norm": 0.00028044532518833876, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 11.5, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.0805408583186361, | |
| "grad_norm": 0.00019913016876671463, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 11.5, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.08112874779541446, | |
| "grad_norm": 0.0002480549446772784, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 11.5, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.08171663727219283, | |
| "grad_norm": 0.00034313430660404265, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 11.5, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.0823045267489712, | |
| "grad_norm": 0.0003088296507485211, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 11.5, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08289241622574955, | |
| "grad_norm": 0.00035947730066254735, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 11.5, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.08348030570252793, | |
| "grad_norm": 0.0003780677798204124, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 11.5, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.08406819517930629, | |
| "grad_norm": 0.00027284034877084196, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 11.5, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.08465608465608465, | |
| "grad_norm": 0.00019406316278036684, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 11.5, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.08524397413286303, | |
| "grad_norm": 0.00047140681999735534, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 11.5, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.08583186360964139, | |
| "grad_norm": 0.0002869810559786856, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 11.5, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.08641975308641975, | |
| "grad_norm": 0.00032051026937551796, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 11.5, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.08700764256319812, | |
| "grad_norm": 0.0003906071069650352, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 11.5, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.08759553203997648, | |
| "grad_norm": 0.0004939660429954529, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 11.5, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.08818342151675485, | |
| "grad_norm": 0.0008581305737607181, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 11.5, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08818342151675485, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 13.7743, | |
| "eval_samples_per_second": 207.997, | |
| "eval_steps_per_second": 52.054, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08877131099353322, | |
| "grad_norm": 0.00035099269007332623, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 11.5, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.08935920047031158, | |
| "grad_norm": 0.00027555919950827956, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 11.5, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.08994708994708994, | |
| "grad_norm": 0.00020382685761433095, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 11.5, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.09053497942386832, | |
| "grad_norm": 0.00028512062272056937, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 11.5, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.09112286890064668, | |
| "grad_norm": 0.0002565366739872843, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 11.5, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.09171075837742504, | |
| "grad_norm": 0.0002958117111120373, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 11.5, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.09229864785420341, | |
| "grad_norm": 0.00019792199600487947, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 11.5, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.09288653733098177, | |
| "grad_norm": 0.000278434221399948, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 11.5, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.09347442680776014, | |
| "grad_norm": 0.00041769936797209084, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 11.5, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.09406231628453851, | |
| "grad_norm": 0.0003284548001829535, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 11.5, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09465020576131687, | |
| "grad_norm": 0.0003623607917688787, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 11.5, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.09523809523809523, | |
| "grad_norm": 0.0003717868239618838, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 11.5, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.09582598471487361, | |
| "grad_norm": 0.00030257663456723094, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 11.5, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.09641387419165197, | |
| "grad_norm": 0.0002942180435638875, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 11.5, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.09700176366843033, | |
| "grad_norm": 0.0004222776915412396, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 11.5, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0975896531452087, | |
| "grad_norm": 0.0002578916901256889, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 11.5, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.09817754262198707, | |
| "grad_norm": 0.0003998030151706189, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 11.5, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.09876543209876543, | |
| "grad_norm": 0.0002618937869556248, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 11.5, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.0993533215755438, | |
| "grad_norm": 0.00040995507151819766, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 11.5, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.09994121105232216, | |
| "grad_norm": 0.0003908443031832576, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 11.5, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.10052910052910052, | |
| "grad_norm": 0.0002970104687847197, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 11.5, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.1011169900058789, | |
| "grad_norm": 0.00043291100882925093, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 11.5, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.10170487948265726, | |
| "grad_norm": 0.00031029590172693133, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 11.5, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.10229276895943562, | |
| "grad_norm": 0.0002891482727136463, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 11.5, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.102880658436214, | |
| "grad_norm": 0.00044630267075262964, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 11.5, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.10346854791299236, | |
| "grad_norm": 0.0003182612417731434, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 11.5, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.10405643738977072, | |
| "grad_norm": 0.00030209519900381565, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 11.5, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.10464432686654909, | |
| "grad_norm": 0.00042158199357800186, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 11.5, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.10523221634332745, | |
| "grad_norm": 0.00034510300611145794, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 11.5, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.10582010582010581, | |
| "grad_norm": 0.0003283740079496056, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 11.5, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.10640799529688419, | |
| "grad_norm": 0.00040273010381497443, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 11.5, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.10699588477366255, | |
| "grad_norm": 0.00047694615204818547, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 11.5, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.10758377425044091, | |
| "grad_norm": 0.0003430490614846349, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 11.5, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.10817166372721929, | |
| "grad_norm": 0.0003051633248105645, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 11.5, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.10875955320399765, | |
| "grad_norm": 0.00035164973814971745, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 11.5, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.10934744268077601, | |
| "grad_norm": 0.00036952606751583517, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 11.5, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.10993533215755438, | |
| "grad_norm": 0.0003788024769164622, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 11.5, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.11052322163433274, | |
| "grad_norm": 0.0003268010332249105, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 11.5, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.1111111111111111, | |
| "grad_norm": 0.0003560074546840042, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 11.5, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.11169900058788948, | |
| "grad_norm": 0.00037741457344964147, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 11.5, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.11228689006466784, | |
| "grad_norm": 0.0003082571492996067, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 11.5, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.1128747795414462, | |
| "grad_norm": 0.00028507207753136754, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 11.5, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.11346266901822458, | |
| "grad_norm": 0.0003684939001686871, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 11.5, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.11405055849500294, | |
| "grad_norm": 0.00031668058363720775, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 11.5, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.1146384479717813, | |
| "grad_norm": 0.0003494382544886321, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 11.5, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.11522633744855967, | |
| "grad_norm": 0.0003476620477158576, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 11.5, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.11581422692533803, | |
| "grad_norm": 0.0003302322584204376, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 11.5, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.1164021164021164, | |
| "grad_norm": 0.00038469716673716903, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 11.5, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.11699000587889477, | |
| "grad_norm": 0.0005184710607863963, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 11.5, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.11757789535567313, | |
| "grad_norm": 0.00041971690370701253, | |
| "learning_rate": 0.0, | |
| "loss": 11.5, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11757789535567313, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 14.4724, | |
| "eval_samples_per_second": 197.963, | |
| "eval_steps_per_second": 49.542, | |
| "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": 3 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 36982232186880.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |