| { | |
| "best_metric": 1.4430131912231445, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.009684526547708398, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 4.842263273854199e-05, | |
| "grad_norm": 21.036222457885742, | |
| "learning_rate": 1e-05, | |
| "loss": 14.9032, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 4.842263273854199e-05, | |
| "eval_loss": 2.380542755126953, | |
| "eval_runtime": 1543.9842, | |
| "eval_samples_per_second": 22.527, | |
| "eval_steps_per_second": 5.632, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 9.684526547708399e-05, | |
| "grad_norm": 23.958187103271484, | |
| "learning_rate": 2e-05, | |
| "loss": 21.7787, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00014526789821562599, | |
| "grad_norm": 50.999671936035156, | |
| "learning_rate": 3e-05, | |
| "loss": 22.3275, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00019369053095416797, | |
| "grad_norm": 15.309734344482422, | |
| "learning_rate": 4e-05, | |
| "loss": 20.3532, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00024211316369270999, | |
| "grad_norm": 19.989614486694336, | |
| "learning_rate": 5e-05, | |
| "loss": 23.786, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00029053579643125197, | |
| "grad_norm": 15.381088256835938, | |
| "learning_rate": 6e-05, | |
| "loss": 19.1951, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.000338958429169794, | |
| "grad_norm": 53.560001373291016, | |
| "learning_rate": 7e-05, | |
| "loss": 18.769, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00038738106190833594, | |
| "grad_norm": 21.586074829101562, | |
| "learning_rate": 8e-05, | |
| "loss": 14.7168, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00043580369464687796, | |
| "grad_norm": 12.721044540405273, | |
| "learning_rate": 9e-05, | |
| "loss": 11.3785, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00048422632738541997, | |
| "grad_norm": 17.54620361328125, | |
| "learning_rate": 0.0001, | |
| "loss": 14.1235, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0005326489601239619, | |
| "grad_norm": 110.95148468017578, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 12.0375, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0005810715928625039, | |
| "grad_norm": 20.622997283935547, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 13.8813, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.000629494225601046, | |
| "grad_norm": 46.31759262084961, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 12.2084, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.000677916858339588, | |
| "grad_norm": 17.43862533569336, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 12.192, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0007263394910781299, | |
| "grad_norm": 34.96971893310547, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 11.7192, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0007747621238166719, | |
| "grad_norm": 19.7441349029541, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 11.0306, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0008231847565552139, | |
| "grad_norm": 62.67892837524414, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 10.6573, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0008716073892937559, | |
| "grad_norm": 58.20136642456055, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 12.7056, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0009200300220322979, | |
| "grad_norm": 104.05531311035156, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 11.6294, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0009684526547708399, | |
| "grad_norm": 18.70232391357422, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 10.8203, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.001016875287509382, | |
| "grad_norm": 16.477691650390625, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 10.6831, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0010652979202479239, | |
| "grad_norm": 18.849660873413086, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 11.9745, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.001113720552986466, | |
| "grad_norm": 12.664268493652344, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 12.0293, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0011621431857250079, | |
| "grad_norm": 15.409565925598145, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 12.5849, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0012105658184635498, | |
| "grad_norm": 17.945972442626953, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 11.1719, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.001258988451202092, | |
| "grad_norm": 13.945023536682129, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 11.8341, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0013074110839406338, | |
| "grad_norm": 14.512016296386719, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 12.4799, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.001355833716679176, | |
| "grad_norm": 14.7997465133667, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 10.781, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0014042563494177178, | |
| "grad_norm": 15.816786766052246, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 10.5862, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0014526789821562597, | |
| "grad_norm": 18.61883544921875, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 12.2405, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0015011016148948019, | |
| "grad_norm": 16.041227340698242, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 10.6555, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0015495242476333438, | |
| "grad_norm": 17.19962501525879, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 11.4427, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.001597946880371886, | |
| "grad_norm": 19.055322647094727, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 10.9813, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0016463695131104278, | |
| "grad_norm": 20.501718521118164, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 8.2012, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0016947921458489697, | |
| "grad_norm": 34.35026931762695, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 8.4981, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0017432147785875118, | |
| "grad_norm": 27.589019775390625, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 8.1987, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0017916374113260537, | |
| "grad_norm": 32.47792053222656, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 8.1395, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0018400600440645959, | |
| "grad_norm": 25.601943969726562, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 9.9967, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0018884826768031378, | |
| "grad_norm": 28.293134689331055, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 8.4649, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0019369053095416799, | |
| "grad_norm": 27.98113441467285, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 8.227, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0019853279422802216, | |
| "grad_norm": 25.54633903503418, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 7.6596, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.002033750575018764, | |
| "grad_norm": 29.347736358642578, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 7.6935, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.002082173207757306, | |
| "grad_norm": 21.824691772460938, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 8.4516, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0021305958404958477, | |
| "grad_norm": 29.6697998046875, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 6.7448, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0021790184732343896, | |
| "grad_norm": 36.165809631347656, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 7.0912, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.002227441105972932, | |
| "grad_norm": 26.53007698059082, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 7.0569, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.002275863738711474, | |
| "grad_norm": 36.98576736450195, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 7.3414, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0023242863714500158, | |
| "grad_norm": 30.28927230834961, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 7.4491, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0023727090041885577, | |
| "grad_norm": 28.85490608215332, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 8.558, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0024211316369270996, | |
| "grad_norm": 35.33230972290039, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 10.2282, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0024211316369270996, | |
| "eval_loss": 1.8891305923461914, | |
| "eval_runtime": 1551.2057, | |
| "eval_samples_per_second": 22.423, | |
| "eval_steps_per_second": 5.606, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.002469554269665642, | |
| "grad_norm": 41.95703887939453, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 7.0208, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.002517976902404184, | |
| "grad_norm": 29.346487045288086, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 9.9616, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0025663995351427257, | |
| "grad_norm": 30.1339054107666, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 12.3816, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0026148221678812676, | |
| "grad_norm": 20.9738826751709, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 12.1981, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0026632448006198095, | |
| "grad_norm": 21.780534744262695, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 13.3111, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.002711667433358352, | |
| "grad_norm": 22.9322566986084, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 12.5787, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0027600900660968938, | |
| "grad_norm": 40.9758186340332, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 13.0044, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0028085126988354357, | |
| "grad_norm": 17.699081420898438, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 8.601, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0028569353315739776, | |
| "grad_norm": 15.431567192077637, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 7.9839, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0029053579643125195, | |
| "grad_norm": 20.003442764282227, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 9.0721, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.002953780597051062, | |
| "grad_norm": 23.403528213500977, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 8.1184, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0030022032297896037, | |
| "grad_norm": 24.610355377197266, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 11.5657, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0030506258625281456, | |
| "grad_norm": 12.108757972717285, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 7.8688, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0030990484952666876, | |
| "grad_norm": 15.003411293029785, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 10.0369, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0031474711280052295, | |
| "grad_norm": 15.08871841430664, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 8.7818, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.003195893760743772, | |
| "grad_norm": 14.518685340881348, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 10.1146, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0032443163934823137, | |
| "grad_norm": 18.547101974487305, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 10.0041, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0032927390262208556, | |
| "grad_norm": 14.39937686920166, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 9.7189, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0033411616589593975, | |
| "grad_norm": 18.237417221069336, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 9.6827, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0033895842916979394, | |
| "grad_norm": 16.72928237915039, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 9.8503, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0034380069244364818, | |
| "grad_norm": 13.109145164489746, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 7.8869, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0034864295571750237, | |
| "grad_norm": 14.201027870178223, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 8.3623, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0035348521899135656, | |
| "grad_norm": 13.660294532775879, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 9.8333, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0035832748226521075, | |
| "grad_norm": 15.791285514831543, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 8.6577, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.00363169745539065, | |
| "grad_norm": 16.314292907714844, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 9.029, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0036801200881291917, | |
| "grad_norm": 16.78962516784668, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 8.7836, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0037285427208677336, | |
| "grad_norm": 14.088393211364746, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 8.7852, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0037769653536062755, | |
| "grad_norm": 18.910783767700195, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 10.9484, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0038253879863448174, | |
| "grad_norm": 18.279972076416016, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 9.9709, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0038738106190833598, | |
| "grad_norm": 16.942333221435547, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 10.4576, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.003922233251821902, | |
| "grad_norm": 16.805723190307617, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 9.1141, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.003970655884560443, | |
| "grad_norm": 21.950685501098633, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 11.4159, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0040190785172989855, | |
| "grad_norm": 16.89238166809082, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 9.5154, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.004067501150037528, | |
| "grad_norm": 14.915072441101074, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 4.9179, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.004115923782776069, | |
| "grad_norm": 18.866243362426758, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 6.3148, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.004164346415514612, | |
| "grad_norm": 17.326980590820312, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 6.3618, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.004212769048253153, | |
| "grad_norm": 19.755985260009766, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 5.3807, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0042611916809916954, | |
| "grad_norm": 22.213781356811523, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 5.5742, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.004309614313730238, | |
| "grad_norm": 20.444156646728516, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 5.1767, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.004358036946468779, | |
| "grad_norm": 22.062232971191406, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 5.4209, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.004406459579207322, | |
| "grad_norm": 22.24120330810547, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 5.038, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.004454882211945864, | |
| "grad_norm": 24.152446746826172, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 4.9828, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.004503304844684405, | |
| "grad_norm": 22.805784225463867, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 5.6749, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.004551727477422948, | |
| "grad_norm": 23.759235382080078, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 6.2872, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.004600150110161489, | |
| "grad_norm": 21.070165634155273, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 6.3143, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0046485727429000315, | |
| "grad_norm": 27.341114044189453, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 5.4006, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.004696995375638574, | |
| "grad_norm": 22.522701263427734, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 5.2115, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.004745418008377115, | |
| "grad_norm": 23.514177322387695, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 5.6997, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.004793840641115658, | |
| "grad_norm": 22.86407470703125, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 8.317, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.004842263273854199, | |
| "grad_norm": 29.186965942382812, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 9.0345, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.004842263273854199, | |
| "eval_loss": 1.688473105430603, | |
| "eval_runtime": 1551.6878, | |
| "eval_samples_per_second": 22.416, | |
| "eval_steps_per_second": 5.604, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0048906859065927415, | |
| "grad_norm": 15.697155952453613, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 4.7297, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.004939108539331284, | |
| "grad_norm": 16.106801986694336, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 8.3869, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.004987531172069825, | |
| "grad_norm": 16.597978591918945, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 11.4463, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.005035953804808368, | |
| "grad_norm": 19.42620277404785, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 12.7144, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.005084376437546909, | |
| "grad_norm": 15.406960487365723, | |
| "learning_rate": 5e-05, | |
| "loss": 12.703, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.0051327990702854515, | |
| "grad_norm": 16.058740615844727, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 12.288, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.005181221703023994, | |
| "grad_norm": 20.326557159423828, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 14.2708, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.005229644335762535, | |
| "grad_norm": 13.619696617126465, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 8.6256, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.005278066968501078, | |
| "grad_norm": 21.53430938720703, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 7.909, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.005326489601239619, | |
| "grad_norm": 12.871776580810547, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 9.0985, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.005374912233978161, | |
| "grad_norm": 12.44565200805664, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 7.7149, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.005423334866716704, | |
| "grad_norm": 13.205082893371582, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 6.4102, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.005471757499455245, | |
| "grad_norm": 13.271836280822754, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 7.6428, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0055201801321937876, | |
| "grad_norm": 11.78052806854248, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 8.6454, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.005568602764932329, | |
| "grad_norm": 12.537360191345215, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 9.7592, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.005617025397670871, | |
| "grad_norm": 16.84226417541504, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 9.4038, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.005665448030409414, | |
| "grad_norm": 13.914525985717773, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 9.1214, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.005713870663147955, | |
| "grad_norm": 20.360960006713867, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 9.814, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.0057622932958864975, | |
| "grad_norm": 16.53293228149414, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 9.5071, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.005810715928625039, | |
| "grad_norm": 17.271974563598633, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 9.0974, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.005859138561363581, | |
| "grad_norm": 28.929325103759766, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 8.0589, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.005907561194102124, | |
| "grad_norm": 17.490625381469727, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 8.0824, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.005955983826840665, | |
| "grad_norm": 14.606756210327148, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 9.8295, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0060044064595792075, | |
| "grad_norm": 17.656742095947266, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 8.7699, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.006052829092317749, | |
| "grad_norm": 13.570901870727539, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 8.9588, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.006101251725056291, | |
| "grad_norm": 13.676125526428223, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 9.1766, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.006149674357794834, | |
| "grad_norm": 13.872579574584961, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 8.521, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.006198096990533375, | |
| "grad_norm": 16.32117462158203, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 9.7695, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.0062465196232719174, | |
| "grad_norm": 16.955904006958008, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 7.7272, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.006294942256010459, | |
| "grad_norm": 15.329665184020996, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 8.4103, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.006343364888749001, | |
| "grad_norm": 19.150663375854492, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 10.4638, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.006391787521487544, | |
| "grad_norm": 14.786101341247559, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 10.0501, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.006440210154226085, | |
| "grad_norm": 21.21457862854004, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 9.8941, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.006488632786964627, | |
| "grad_norm": 13.094932556152344, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 6.8504, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.006537055419703169, | |
| "grad_norm": 13.877997398376465, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 5.4123, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.006585478052441711, | |
| "grad_norm": 15.05355453491211, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 5.0412, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.0066339006851802535, | |
| "grad_norm": 25.698951721191406, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 4.9759, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.006682323317918795, | |
| "grad_norm": 16.60207748413086, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 4.929, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.006730745950657337, | |
| "grad_norm": 14.616944313049316, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 4.5399, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.006779168583395879, | |
| "grad_norm": 17.269269943237305, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 5.0089, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.006827591216134421, | |
| "grad_norm": 19.678661346435547, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 5.5994, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.0068760138488729635, | |
| "grad_norm": 21.58392333984375, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 4.7898, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.006924436481611505, | |
| "grad_norm": 21.412527084350586, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 4.7194, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.006972859114350047, | |
| "grad_norm": 17.677099227905273, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 4.7382, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.007021281747088589, | |
| "grad_norm": 21.255176544189453, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 5.6346, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.007069704379827131, | |
| "grad_norm": 18.76344871520996, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 5.2847, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.0071181270125656735, | |
| "grad_norm": 18.50588607788086, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 4.9279, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.007166549645304215, | |
| "grad_norm": 29.033336639404297, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 4.6537, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.007214972278042757, | |
| "grad_norm": 18.56321907043457, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 6.2063, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.0072633949107813, | |
| "grad_norm": 27.392192840576172, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 7.674, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0072633949107813, | |
| "eval_loss": 1.506621241569519, | |
| "eval_runtime": 1551.1378, | |
| "eval_samples_per_second": 22.424, | |
| "eval_steps_per_second": 5.606, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.007311817543519841, | |
| "grad_norm": 7.969956398010254, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 3.544, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.007360240176258383, | |
| "grad_norm": 9.882271766662598, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 4.1729, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.007408662808996925, | |
| "grad_norm": 15.573220252990723, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 10.1651, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.007457085441735467, | |
| "grad_norm": 15.387901306152344, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 12.0071, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.00750550807447401, | |
| "grad_norm": 13.318939208984375, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 10.3599, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.007553930707212551, | |
| "grad_norm": 17.2648983001709, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 10.7797, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.007602353339951093, | |
| "grad_norm": 35.52534866333008, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 12.4332, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.007650775972689635, | |
| "grad_norm": 11.675514221191406, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 6.6815, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.007699198605428177, | |
| "grad_norm": 10.801353454589844, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 6.3493, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.0077476212381667195, | |
| "grad_norm": 12.303263664245605, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 8.8866, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.007796043870905261, | |
| "grad_norm": 12.98686695098877, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 6.5817, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.007844466503643803, | |
| "grad_norm": 13.126506805419922, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 9.1023, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.007892889136382345, | |
| "grad_norm": 13.919248580932617, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 7.9798, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.007941311769120886, | |
| "grad_norm": 13.171672821044922, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 8.8413, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.00798973440185943, | |
| "grad_norm": 12.099220275878906, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 9.1222, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.008038157034597971, | |
| "grad_norm": 14.288187980651855, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 8.9465, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.008086579667336512, | |
| "grad_norm": 9.812660217285156, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 7.9193, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.008135002300075056, | |
| "grad_norm": 13.893590927124023, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 8.8289, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.008183424932813597, | |
| "grad_norm": 13.832817077636719, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 9.4699, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.008231847565552139, | |
| "grad_norm": 13.305530548095703, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 9.4601, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.008280270198290682, | |
| "grad_norm": 13.878190040588379, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 8.6175, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.008328692831029223, | |
| "grad_norm": 12.57892894744873, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 9.2335, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.008377115463767765, | |
| "grad_norm": 11.496048927307129, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 7.68, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.008425538096506306, | |
| "grad_norm": 13.143084526062012, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 9.3355, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.00847396072924485, | |
| "grad_norm": 15.608010292053223, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 10.0089, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.008522383361983391, | |
| "grad_norm": 15.516302108764648, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 9.2533, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.008570805994721932, | |
| "grad_norm": 13.103846549987793, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 9.2027, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.008619228627460476, | |
| "grad_norm": 19.22662925720215, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 8.6665, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.008667651260199017, | |
| "grad_norm": 17.343704223632812, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 9.2385, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.008716073892937558, | |
| "grad_norm": 13.398660659790039, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 8.405, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.008764496525676102, | |
| "grad_norm": 16.57634925842285, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 9.9028, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.008812919158414643, | |
| "grad_norm": 19.774003982543945, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 9.6589, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.008861341791153185, | |
| "grad_norm": 18.760805130004883, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 8.4483, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.008909764423891728, | |
| "grad_norm": 14.704105377197266, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 4.5496, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.00895818705663027, | |
| "grad_norm": 13.612820625305176, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 4.3313, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.00900660968936881, | |
| "grad_norm": 16.062952041625977, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 5.0364, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.009055032322107352, | |
| "grad_norm": 16.323183059692383, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 4.4152, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.009103454954845895, | |
| "grad_norm": 20.168174743652344, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 5.1709, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.009151877587584437, | |
| "grad_norm": 17.772062301635742, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 4.9287, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.009200300220322978, | |
| "grad_norm": 15.766966819763184, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 5.1278, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.009248722853061522, | |
| "grad_norm": 17.393341064453125, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 4.1255, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.009297145485800063, | |
| "grad_norm": 22.094804763793945, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 4.507, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.009345568118538605, | |
| "grad_norm": 21.278852462768555, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 4.8382, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.009393990751277148, | |
| "grad_norm": 23.147945404052734, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 3.6258, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.00944241338401569, | |
| "grad_norm": 17.89509391784668, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 5.1395, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.00949083601675423, | |
| "grad_norm": 23.018983840942383, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 6.1392, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.009539258649492772, | |
| "grad_norm": 26.070058822631836, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 5.5732, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.009587681282231315, | |
| "grad_norm": 19.50170135498047, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 5.8904, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.009636103914969857, | |
| "grad_norm": 21.493179321289062, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 6.1535, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.009684526547708398, | |
| "grad_norm": 23.621978759765625, | |
| "learning_rate": 0.0, | |
| "loss": 7.1968, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.009684526547708398, | |
| "eval_loss": 1.4430131912231445, | |
| "eval_runtime": 1549.8526, | |
| "eval_samples_per_second": 22.442, | |
| "eval_steps_per_second": 5.611, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.3407822982781338e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |