| { | |
| "best_metric": 0.08245380222797394, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.8938547486033519, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004469273743016759, | |
| "grad_norm": 35.62481689453125, | |
| "learning_rate": 1e-05, | |
| "loss": 3.2532, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004469273743016759, | |
| "eval_loss": 1.095221757888794, | |
| "eval_runtime": 41.924, | |
| "eval_samples_per_second": 8.992, | |
| "eval_steps_per_second": 2.266, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008938547486033519, | |
| "grad_norm": 50.523406982421875, | |
| "learning_rate": 2e-05, | |
| "loss": 3.9514, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.013407821229050279, | |
| "grad_norm": 33.929325103759766, | |
| "learning_rate": 3e-05, | |
| "loss": 3.7454, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.017877094972067038, | |
| "grad_norm": 15.555086135864258, | |
| "learning_rate": 4e-05, | |
| "loss": 2.7696, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0223463687150838, | |
| "grad_norm": 13.525398254394531, | |
| "learning_rate": 5e-05, | |
| "loss": 2.1322, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.026815642458100558, | |
| "grad_norm": 17.992990493774414, | |
| "learning_rate": 6e-05, | |
| "loss": 1.7891, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03128491620111732, | |
| "grad_norm": 19.359230041503906, | |
| "learning_rate": 7e-05, | |
| "loss": 1.7914, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.035754189944134075, | |
| "grad_norm": 12.898938179016113, | |
| "learning_rate": 8e-05, | |
| "loss": 1.0366, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04022346368715084, | |
| "grad_norm": 14.949868202209473, | |
| "learning_rate": 9e-05, | |
| "loss": 1.1237, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0446927374301676, | |
| "grad_norm": 12.8253173828125, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0305, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.049162011173184354, | |
| "grad_norm": 10.631021499633789, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.9043, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.053631284916201116, | |
| "grad_norm": 9.763545036315918, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.7582, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05810055865921788, | |
| "grad_norm": 7.526438236236572, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.5253, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06256983240223464, | |
| "grad_norm": 9.544380187988281, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.7711, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0670391061452514, | |
| "grad_norm": 8.7190523147583, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.7703, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07150837988826815, | |
| "grad_norm": 7.240570545196533, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.5621, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07597765363128492, | |
| "grad_norm": 6.863866329193115, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.6333, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.08044692737430167, | |
| "grad_norm": 6.460057735443115, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.5771, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08491620111731843, | |
| "grad_norm": 6.961883544921875, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.6267, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0893854748603352, | |
| "grad_norm": 8.382084846496582, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.5625, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09385474860335195, | |
| "grad_norm": 6.16887092590332, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.4628, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.09832402234636871, | |
| "grad_norm": 9.225873947143555, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.5983, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.10279329608938548, | |
| "grad_norm": 9.558427810668945, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.5229, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.10726256983240223, | |
| "grad_norm": 9.383255958557129, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.4622, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.11173184357541899, | |
| "grad_norm": 7.553038120269775, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.4621, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11620111731843576, | |
| "grad_norm": 5.422492504119873, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.4178, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.12067039106145251, | |
| "grad_norm": 9.239324569702148, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.4406, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.12513966480446928, | |
| "grad_norm": 19.214887619018555, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.6008, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.12960893854748604, | |
| "grad_norm": 9.390142440795898, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.019, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1340782122905028, | |
| "grad_norm": 7.1718316078186035, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.6951, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13854748603351955, | |
| "grad_norm": 8.319527626037598, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.9247, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.1430167597765363, | |
| "grad_norm": 9.273724555969238, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.9412, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.14748603351955308, | |
| "grad_norm": 7.298791408538818, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.4895, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.15195530726256984, | |
| "grad_norm": 5.492515563964844, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.507, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1564245810055866, | |
| "grad_norm": 8.4094820022583, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.8276, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.16089385474860335, | |
| "grad_norm": 9.204874992370605, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.1559, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1653631284916201, | |
| "grad_norm": 5.026533126831055, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.6378, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.16983240223463686, | |
| "grad_norm": 6.076452732086182, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.6317, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.17430167597765364, | |
| "grad_norm": 6.885071277618408, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.564, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1787709497206704, | |
| "grad_norm": 6.654655456542969, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.4058, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.18324022346368715, | |
| "grad_norm": 4.889857769012451, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.394, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.1877094972067039, | |
| "grad_norm": 8.59021282196045, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.5186, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.19217877094972066, | |
| "grad_norm": 8.131345748901367, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.7604, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.19664804469273742, | |
| "grad_norm": 6.608253479003906, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.6862, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2011173184357542, | |
| "grad_norm": 5.186952590942383, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.5202, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.20558659217877095, | |
| "grad_norm": 5.894842624664307, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.5134, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2100558659217877, | |
| "grad_norm": 7.351461887359619, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.0659, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.21452513966480447, | |
| "grad_norm": 7.4132890701293945, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.7827, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.21899441340782122, | |
| "grad_norm": 5.722691535949707, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.5605, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.22346368715083798, | |
| "grad_norm": 13.774589538574219, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.0026, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22346368715083798, | |
| "eval_loss": 0.17977474629878998, | |
| "eval_runtime": 42.7426, | |
| "eval_samples_per_second": 8.82, | |
| "eval_steps_per_second": 2.223, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22793296089385476, | |
| "grad_norm": 7.665311336517334, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.3458, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.2324022346368715, | |
| "grad_norm": 6.353878498077393, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.1218, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.23687150837988827, | |
| "grad_norm": 5.1552958488464355, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.7979, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.24134078212290502, | |
| "grad_norm": 5.40798807144165, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.8484, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.24581005586592178, | |
| "grad_norm": 4.324162006378174, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.6086, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.25027932960893856, | |
| "grad_norm": 4.098971843719482, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.739, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.2547486033519553, | |
| "grad_norm": 4.3394927978515625, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.5396, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.25921787709497207, | |
| "grad_norm": 4.38415002822876, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.4487, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2636871508379888, | |
| "grad_norm": 4.5791335105896, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.5217, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2681564245810056, | |
| "grad_norm": 5.626852512359619, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.6933, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.27262569832402234, | |
| "grad_norm": 5.020511150360107, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.5472, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.2770949720670391, | |
| "grad_norm": 6.1369757652282715, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.5402, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.28156424581005585, | |
| "grad_norm": 4.541934013366699, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.3626, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.2860335195530726, | |
| "grad_norm": 3.9370110034942627, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.3981, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2905027932960894, | |
| "grad_norm": 4.38019323348999, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.4999, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.29497206703910617, | |
| "grad_norm": 4.494397163391113, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.489, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2994413407821229, | |
| "grad_norm": 3.883639335632324, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.3572, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.3039106145251397, | |
| "grad_norm": 5.072006702423096, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.4462, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.30837988826815643, | |
| "grad_norm": 5.48960542678833, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.4855, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.3128491620111732, | |
| "grad_norm": 3.713104009628296, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.3447, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.31731843575418994, | |
| "grad_norm": 5.593199253082275, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.5328, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.3217877094972067, | |
| "grad_norm": 4.817788124084473, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.3695, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.32625698324022345, | |
| "grad_norm": 3.8325271606445312, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.3874, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.3307262569832402, | |
| "grad_norm": 4.72641134262085, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.4225, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.33519553072625696, | |
| "grad_norm": 4.551783084869385, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.2202, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3396648044692737, | |
| "grad_norm": 3.0847671031951904, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.289, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.34413407821229053, | |
| "grad_norm": 6.853332042694092, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.9313, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3486033519553073, | |
| "grad_norm": 6.723824977874756, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.8523, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.35307262569832404, | |
| "grad_norm": 4.161146640777588, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.51, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3575418994413408, | |
| "grad_norm": 5.265901565551758, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.7922, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.36201117318435755, | |
| "grad_norm": 5.52878999710083, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.8681, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3664804469273743, | |
| "grad_norm": 4.830251693725586, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.5643, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.37094972067039106, | |
| "grad_norm": 5.248570919036865, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.464, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.3754189944134078, | |
| "grad_norm": 4.588041305541992, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.5442, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.37988826815642457, | |
| "grad_norm": 4.127342224121094, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.7035, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3843575418994413, | |
| "grad_norm": 3.3109381198883057, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.4833, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3888268156424581, | |
| "grad_norm": 3.9795236587524414, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.4506, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.39329608938547483, | |
| "grad_norm": 3.254094362258911, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.332, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.39776536312849164, | |
| "grad_norm": 1.8271361589431763, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.2397, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.4022346368715084, | |
| "grad_norm": 3.046233654022217, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.2517, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.40670391061452515, | |
| "grad_norm": 4.531677722930908, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.359, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.4111731843575419, | |
| "grad_norm": 5.700006484985352, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.6938, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.41564245810055866, | |
| "grad_norm": 3.9766898155212402, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.4539, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.4201117318435754, | |
| "grad_norm": 4.2325944900512695, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.3988, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.4245810055865922, | |
| "grad_norm": 3.1019227504730225, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.2806, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.42905027932960893, | |
| "grad_norm": 6.961395263671875, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.0459, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.4335195530726257, | |
| "grad_norm": 6.448850631713867, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.7186, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.43798882681564244, | |
| "grad_norm": 4.595263481140137, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.5309, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4424581005586592, | |
| "grad_norm": 3.503016471862793, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.3325, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.44692737430167595, | |
| "grad_norm": 5.690208911895752, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.6615, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.44692737430167595, | |
| "eval_loss": 0.13959139585494995, | |
| "eval_runtime": 43.0216, | |
| "eval_samples_per_second": 8.763, | |
| "eval_steps_per_second": 2.208, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.45139664804469276, | |
| "grad_norm": 6.199459075927734, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.336, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4558659217877095, | |
| "grad_norm": 4.8227410316467285, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.8468, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.46033519553072627, | |
| "grad_norm": 4.362427234649658, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.7669, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.464804469273743, | |
| "grad_norm": 4.3454766273498535, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.6913, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.4692737430167598, | |
| "grad_norm": 3.2233870029449463, | |
| "learning_rate": 5e-05, | |
| "loss": 0.482, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.47374301675977654, | |
| "grad_norm": 2.7639806270599365, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.3978, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4782122905027933, | |
| "grad_norm": 2.9535152912139893, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.4253, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.48268156424581005, | |
| "grad_norm": 3.1213266849517822, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.5537, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.4871508379888268, | |
| "grad_norm": 3.073854923248291, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.5628, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.49162011173184356, | |
| "grad_norm": 2.736854314804077, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.4288, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.4960893854748603, | |
| "grad_norm": 2.7129223346710205, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.4148, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.5005586592178771, | |
| "grad_norm": 3.133866310119629, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.4266, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.5050279329608939, | |
| "grad_norm": 2.4408538341522217, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.3005, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.5094972067039106, | |
| "grad_norm": 2.398010492324829, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.318, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.5139664804469274, | |
| "grad_norm": 3.23746395111084, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.4546, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5184357541899441, | |
| "grad_norm": 2.4386403560638428, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.2936, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5229050279329609, | |
| "grad_norm": 3.5520517826080322, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.3269, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5273743016759777, | |
| "grad_norm": 3.058534622192383, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.3643, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.5318435754189944, | |
| "grad_norm": 2.5184006690979004, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.2544, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.5363128491620112, | |
| "grad_norm": 3.625074863433838, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.3931, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.5407821229050279, | |
| "grad_norm": 4.143115520477295, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.3147, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.5452513966480447, | |
| "grad_norm": 3.0545642375946045, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.253, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5497206703910614, | |
| "grad_norm": 3.6490237712860107, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.3292, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5541899441340782, | |
| "grad_norm": 2.3374831676483154, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.1646, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5586592178770949, | |
| "grad_norm": 3.5566532611846924, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.2041, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5631284916201117, | |
| "grad_norm": 3.6130127906799316, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.2241, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5675977653631284, | |
| "grad_norm": 5.787264347076416, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.7327, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5720670391061452, | |
| "grad_norm": 5.905097484588623, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.7795, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.576536312849162, | |
| "grad_norm": 5.2323737144470215, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.5518, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5810055865921788, | |
| "grad_norm": 4.876833915710449, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.4767, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5854748603351956, | |
| "grad_norm": 7.045799732208252, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.7371, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5899441340782123, | |
| "grad_norm": 4.691097259521484, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.4452, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5944134078212291, | |
| "grad_norm": 5.455672264099121, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.3393, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5988826815642458, | |
| "grad_norm": 5.074672222137451, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.3545, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.6033519553072626, | |
| "grad_norm": 5.807733535766602, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.5254, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.6078212290502794, | |
| "grad_norm": 4.809835433959961, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.6038, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.6122905027932961, | |
| "grad_norm": 3.952597141265869, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.5192, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.6167597765363129, | |
| "grad_norm": 3.181260824203491, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.4339, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.6212290502793296, | |
| "grad_norm": 3.6808667182922363, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.4122, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.6256983240223464, | |
| "grad_norm": 2.9892947673797607, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.3642, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6301675977653631, | |
| "grad_norm": 2.677868366241455, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.3337, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.6346368715083799, | |
| "grad_norm": 3.1183557510375977, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3767, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.6391061452513966, | |
| "grad_norm": 2.5099844932556152, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.3884, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.6435754189944134, | |
| "grad_norm": 2.5012831687927246, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.3494, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.6480446927374302, | |
| "grad_norm": 3.2986056804656982, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.4629, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.6525139664804469, | |
| "grad_norm": 3.9482948780059814, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.385, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.6569832402234637, | |
| "grad_norm": 4.121823787689209, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.7082, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6614525139664804, | |
| "grad_norm": 2.9307124614715576, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.4343, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6659217877094972, | |
| "grad_norm": 2.718599319458008, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.3046, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6703910614525139, | |
| "grad_norm": 3.6085498332977295, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.5011, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6703910614525139, | |
| "eval_loss": 0.0994621068239212, | |
| "eval_runtime": 43.0147, | |
| "eval_samples_per_second": 8.764, | |
| "eval_steps_per_second": 2.209, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6748603351955307, | |
| "grad_norm": 4.863796710968018, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.9764, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6793296089385474, | |
| "grad_norm": 4.779887676239014, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.8311, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6837988826815642, | |
| "grad_norm": 4.134397983551025, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.5543, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6882681564245811, | |
| "grad_norm": 4.0554518699646, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.4894, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6927374301675978, | |
| "grad_norm": 3.0983221530914307, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.5442, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6972067039106146, | |
| "grad_norm": 2.8841450214385986, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.4463, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.7016759776536313, | |
| "grad_norm": 2.423989772796631, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.3911, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.7061452513966481, | |
| "grad_norm": 2.3909876346588135, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.3415, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.7106145251396648, | |
| "grad_norm": 2.357570171356201, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.3953, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.7150837988826816, | |
| "grad_norm": 2.9353442192077637, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.318, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.7195530726256983, | |
| "grad_norm": 2.4930405616760254, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.3126, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.7240223463687151, | |
| "grad_norm": 2.4303815364837646, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.3471, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.7284916201117319, | |
| "grad_norm": 2.444162130355835, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.2839, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.7329608938547486, | |
| "grad_norm": 2.5624821186065674, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.3557, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.7374301675977654, | |
| "grad_norm": 2.6451094150543213, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.3927, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.7418994413407821, | |
| "grad_norm": 2.3692893981933594, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.32, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.7463687150837989, | |
| "grad_norm": 2.4035532474517822, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.3318, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.7508379888268156, | |
| "grad_norm": 2.6109111309051514, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.3607, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.7553072625698324, | |
| "grad_norm": 3.1098432540893555, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.2356, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.7597765363128491, | |
| "grad_norm": 2.496288776397705, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.2809, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.7642458100558659, | |
| "grad_norm": 2.1144332885742188, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.1796, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.7687150837988826, | |
| "grad_norm": 2.4365086555480957, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.2374, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7731843575418994, | |
| "grad_norm": 3.016291618347168, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.4205, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7776536312849162, | |
| "grad_norm": 2.364156723022461, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.243, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7821229050279329, | |
| "grad_norm": 3.4438300132751465, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.2356, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7865921787709497, | |
| "grad_norm": 2.7396368980407715, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.2393, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7910614525139665, | |
| "grad_norm": 4.120048999786377, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.3642, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7955307262569833, | |
| "grad_norm": 3.089003562927246, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.5074, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 3.8327934741973877, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.4172, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.8044692737430168, | |
| "grad_norm": 3.1237096786499023, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.3909, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.8089385474860336, | |
| "grad_norm": 4.535591125488281, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.4745, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.8134078212290503, | |
| "grad_norm": 3.115571975708008, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.45, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.8178770949720671, | |
| "grad_norm": 4.075606346130371, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.4806, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.8223463687150838, | |
| "grad_norm": 3.6217126846313477, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.5044, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.8268156424581006, | |
| "grad_norm": 2.769263744354248, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.3107, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.8312849162011173, | |
| "grad_norm": 3.559227228164673, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.4518, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.8357541899441341, | |
| "grad_norm": 3.000656843185425, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.3697, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.8402234636871508, | |
| "grad_norm": 2.7184536457061768, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.2972, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.8446927374301676, | |
| "grad_norm": 2.0069222450256348, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.239, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.8491620111731844, | |
| "grad_norm": 2.8101136684417725, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.3096, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8536312849162011, | |
| "grad_norm": 2.9242937564849854, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.237, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.8581005586592179, | |
| "grad_norm": 1.7177648544311523, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.224, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.8625698324022346, | |
| "grad_norm": 2.876354455947876, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.3864, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.8670391061452514, | |
| "grad_norm": 2.442669630050659, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.4284, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.8715083798882681, | |
| "grad_norm": 2.461059331893921, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.2985, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.8759776536312849, | |
| "grad_norm": 2.5892558097839355, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.3041, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8804469273743016, | |
| "grad_norm": 3.0513784885406494, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.4899, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8849162011173184, | |
| "grad_norm": 2.80283784866333, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.4045, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8893854748603351, | |
| "grad_norm": 2.278520107269287, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.2543, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8938547486033519, | |
| "grad_norm": 2.6354920864105225, | |
| "learning_rate": 0.0, | |
| "loss": 0.3099, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8938547486033519, | |
| "eval_loss": 0.08245380222797394, | |
| "eval_runtime": 42.7558, | |
| "eval_samples_per_second": 8.818, | |
| "eval_steps_per_second": 2.222, | |
| "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": 4.420115895209165e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |