| { | |
| "best_metric": 0.15422219038009644, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.8705114254624592, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004352557127312296, | |
| "grad_norm": 0.20845629274845123, | |
| "learning_rate": 1e-05, | |
| "loss": 0.1143, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004352557127312296, | |
| "eval_loss": 0.37176230549812317, | |
| "eval_runtime": 22.0337, | |
| "eval_samples_per_second": 17.564, | |
| "eval_steps_per_second": 4.402, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008705114254624592, | |
| "grad_norm": 0.22131690382957458, | |
| "learning_rate": 2e-05, | |
| "loss": 0.1578, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.013057671381936888, | |
| "grad_norm": 0.2071092277765274, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1545, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.017410228509249184, | |
| "grad_norm": 0.20054228603839874, | |
| "learning_rate": 4e-05, | |
| "loss": 0.1629, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02176278563656148, | |
| "grad_norm": 0.16518697142601013, | |
| "learning_rate": 5e-05, | |
| "loss": 0.1544, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.026115342763873776, | |
| "grad_norm": 0.15351487696170807, | |
| "learning_rate": 6e-05, | |
| "loss": 0.1549, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.030467899891186073, | |
| "grad_norm": 0.16655641794204712, | |
| "learning_rate": 7e-05, | |
| "loss": 0.1528, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03482045701849837, | |
| "grad_norm": 0.2033957540988922, | |
| "learning_rate": 8e-05, | |
| "loss": 0.1446, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03917301414581066, | |
| "grad_norm": 0.19426198303699493, | |
| "learning_rate": 9e-05, | |
| "loss": 0.1682, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04352557127312296, | |
| "grad_norm": 0.15188050270080566, | |
| "learning_rate": 0.0001, | |
| "loss": 0.1434, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04787812840043525, | |
| "grad_norm": 0.13493749499320984, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.1315, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05223068552774755, | |
| "grad_norm": 0.1329093724489212, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.1457, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.056583242655059846, | |
| "grad_norm": 0.12495986372232437, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.1257, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.060935799782372145, | |
| "grad_norm": 0.126482293009758, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.1393, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06528835690968444, | |
| "grad_norm": 0.13529592752456665, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.1624, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06964091403699674, | |
| "grad_norm": 0.11158934235572815, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.1149, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07399347116430903, | |
| "grad_norm": 0.1195780411362648, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.1313, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07834602829162132, | |
| "grad_norm": 0.12020254135131836, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.1415, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08269858541893363, | |
| "grad_norm": 0.12221608310937881, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.1426, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08705114254624592, | |
| "grad_norm": 0.13288737833499908, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.1531, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09140369967355821, | |
| "grad_norm": 0.13833260536193848, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.1565, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0957562568008705, | |
| "grad_norm": 0.12379870563745499, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.1364, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.10010881392818281, | |
| "grad_norm": 0.145259827375412, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.1681, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.1044613710554951, | |
| "grad_norm": 0.12043190002441406, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.1379, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1088139281828074, | |
| "grad_norm": 0.12238962203264236, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.1317, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11316648531011969, | |
| "grad_norm": 0.1297006756067276, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.1311, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.117519042437432, | |
| "grad_norm": 0.15071047842502594, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.1537, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.12187159956474429, | |
| "grad_norm": 0.14706623554229736, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.1368, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.12622415669205658, | |
| "grad_norm": 0.15970942378044128, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.1775, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1305767138193689, | |
| "grad_norm": 0.18217220902442932, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.1749, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13492927094668117, | |
| "grad_norm": 0.20835880935192108, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.18, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.13928182807399347, | |
| "grad_norm": 0.19409587979316711, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.1924, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.14363438520130578, | |
| "grad_norm": 0.23500393331050873, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.2071, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.14798694232861806, | |
| "grad_norm": 0.1974450945854187, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.1843, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.15233949945593037, | |
| "grad_norm": 0.24645167589187622, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.1908, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.15669205658324264, | |
| "grad_norm": 0.28018084168434143, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.2398, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.16104461371055495, | |
| "grad_norm": 0.23548993468284607, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.2099, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.16539717083786726, | |
| "grad_norm": 0.2908487915992737, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.2785, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.16974972796517954, | |
| "grad_norm": 0.2943369150161743, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.2607, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.17410228509249184, | |
| "grad_norm": 0.3620036244392395, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.245, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.17845484221980412, | |
| "grad_norm": 0.3757473826408386, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.2561, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.18280739934711643, | |
| "grad_norm": 0.34714168310165405, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.2703, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.18715995647442873, | |
| "grad_norm": 0.4004495441913605, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.1799, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.191512513601741, | |
| "grad_norm": 0.4533845782279968, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.2985, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.19586507072905332, | |
| "grad_norm": 0.45601508021354675, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.393, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.20021762785636563, | |
| "grad_norm": 0.3571247160434723, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.3085, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2045701849836779, | |
| "grad_norm": 0.5167179107666016, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.3753, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.2089227421109902, | |
| "grad_norm": 0.5352243781089783, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.3592, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.2132752992383025, | |
| "grad_norm": 0.5767673850059509, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.3925, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.2176278563656148, | |
| "grad_norm": 0.8895710706710815, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.3559, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2176278563656148, | |
| "eval_loss": 0.20801271498203278, | |
| "eval_runtime": 22.3575, | |
| "eval_samples_per_second": 17.31, | |
| "eval_steps_per_second": 4.339, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2219804134929271, | |
| "grad_norm": 0.27515241503715515, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.0879, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.22633297062023938, | |
| "grad_norm": 0.2824498116970062, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.1103, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.2306855277475517, | |
| "grad_norm": 0.19525060057640076, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.121, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.235038084874864, | |
| "grad_norm": 0.1460787057876587, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.123, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.23939064200217627, | |
| "grad_norm": 0.12723995745182037, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.133, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.24374319912948858, | |
| "grad_norm": 0.11900807917118073, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.1336, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.24809575625680086, | |
| "grad_norm": 0.10293258726596832, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.1226, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.25244831338411317, | |
| "grad_norm": 0.10009904205799103, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.1173, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.25680087051142547, | |
| "grad_norm": 0.1017303392291069, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.1312, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2611534276387378, | |
| "grad_norm": 0.10294551402330399, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.1125, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.26550598476605003, | |
| "grad_norm": 0.1070595383644104, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.1287, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.26985854189336234, | |
| "grad_norm": 0.10025951266288757, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.139, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.27421109902067464, | |
| "grad_norm": 0.09192643314599991, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.1073, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.27856365614798695, | |
| "grad_norm": 0.11544953286647797, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.1512, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.28291621327529926, | |
| "grad_norm": 0.10892773419618607, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.1414, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.28726877040261156, | |
| "grad_norm": 0.09894298017024994, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.1365, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2916213275299238, | |
| "grad_norm": 0.10541431605815887, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.1502, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2959738846572361, | |
| "grad_norm": 0.10178884863853455, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.1346, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.3003264417845484, | |
| "grad_norm": 0.09728579223155975, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.1237, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.30467899891186073, | |
| "grad_norm": 0.09369879215955734, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.1122, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.30903155603917304, | |
| "grad_norm": 0.10953373461961746, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.1456, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.3133841131664853, | |
| "grad_norm": 0.10684646666049957, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.1381, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.3177366702937976, | |
| "grad_norm": 0.11669085174798965, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.1475, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.3220892274211099, | |
| "grad_norm": 0.10585611313581467, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.1189, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3264417845484222, | |
| "grad_norm": 0.12454231828451157, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.143, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3307943416757345, | |
| "grad_norm": 0.1234443262219429, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.1539, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.33514689880304677, | |
| "grad_norm": 0.12470456212759018, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.1408, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3394994559303591, | |
| "grad_norm": 0.12070462852716446, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.133, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.3438520130576714, | |
| "grad_norm": 0.12489613145589828, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.1073, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3482045701849837, | |
| "grad_norm": 0.148418590426445, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.1529, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.352557127312296, | |
| "grad_norm": 0.1505834460258484, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.1538, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.35690968443960824, | |
| "grad_norm": 0.1543310582637787, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.1528, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.36126224156692055, | |
| "grad_norm": 0.18152011930942535, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.1747, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.36561479869423286, | |
| "grad_norm": 0.19291453063488007, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.186, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.36996735582154516, | |
| "grad_norm": 0.22318421304225922, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.2209, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.37431991294885747, | |
| "grad_norm": 0.20853181183338165, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.1594, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3786724700761698, | |
| "grad_norm": 0.24706843495368958, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.2291, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.383025027203482, | |
| "grad_norm": 0.27028441429138184, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.284, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.38737758433079433, | |
| "grad_norm": 0.2656073570251465, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.2144, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.39173014145810664, | |
| "grad_norm": 0.34104010462760925, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.2222, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.39608269858541895, | |
| "grad_norm": 0.32052871584892273, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.1821, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.40043525571273125, | |
| "grad_norm": 0.3408251702785492, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.3013, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.4047878128400435, | |
| "grad_norm": 0.3694060444831848, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.314, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.4091403699673558, | |
| "grad_norm": 0.35222458839416504, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.2424, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.4134929270946681, | |
| "grad_norm": 0.3697720766067505, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.2588, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.4178454842219804, | |
| "grad_norm": 0.40540942549705505, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.3472, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.42219804134929273, | |
| "grad_norm": 0.4851481020450592, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.3639, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.426550598476605, | |
| "grad_norm": 0.37067747116088867, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.2485, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4309031556039173, | |
| "grad_norm": 0.5414036512374878, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.3214, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4352557127312296, | |
| "grad_norm": 0.7560898065567017, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.3443, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4352557127312296, | |
| "eval_loss": 0.1710146963596344, | |
| "eval_runtime": 22.2454, | |
| "eval_samples_per_second": 17.397, | |
| "eval_steps_per_second": 4.36, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4396082698585419, | |
| "grad_norm": 0.11615023761987686, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.0623, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4439608269858542, | |
| "grad_norm": 0.15759530663490295, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.1039, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.44831338411316646, | |
| "grad_norm": 0.12545421719551086, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.0991, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.45266594124047876, | |
| "grad_norm": 0.13131660223007202, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.1296, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.45701849836779107, | |
| "grad_norm": 0.11338076740503311, | |
| "learning_rate": 5e-05, | |
| "loss": 0.1116, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.4613710554951034, | |
| "grad_norm": 0.10338016599416733, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.1046, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4657236126224157, | |
| "grad_norm": 0.11747431755065918, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.1278, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.470076169749728, | |
| "grad_norm": 0.1042664423584938, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.1117, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.47442872687704024, | |
| "grad_norm": 0.1078542098402977, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.1242, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.47878128400435255, | |
| "grad_norm": 0.0953056812286377, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.1103, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.48313384113166485, | |
| "grad_norm": 0.10049030929803848, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.1162, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.48748639825897716, | |
| "grad_norm": 0.09291423112154007, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.1064, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.49183895538628947, | |
| "grad_norm": 0.08183927834033966, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0925, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.4961915125136017, | |
| "grad_norm": 0.10217400640249252, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.1304, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.500544069640914, | |
| "grad_norm": 0.08984298259019852, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.1053, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5048966267682263, | |
| "grad_norm": 0.09127230197191238, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.1148, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5092491838955386, | |
| "grad_norm": 0.09611231833696365, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.1166, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5136017410228509, | |
| "grad_norm": 0.09848757088184357, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.1122, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.5179542981501633, | |
| "grad_norm": 0.10343743115663528, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.1282, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.5223068552774756, | |
| "grad_norm": 0.10088210552930832, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.1124, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.5266594124047879, | |
| "grad_norm": 0.09848304092884064, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.1163, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.5310119695321001, | |
| "grad_norm": 0.10799694806337357, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.1302, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5353645266594124, | |
| "grad_norm": 0.09744226187467575, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.1153, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5397170837867247, | |
| "grad_norm": 0.10120628774166107, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.1133, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.544069640914037, | |
| "grad_norm": 0.12208776921033859, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.1345, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5484221980413493, | |
| "grad_norm": 0.10760590434074402, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.1178, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5527747551686616, | |
| "grad_norm": 0.13905629515647888, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.1694, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5571273122959739, | |
| "grad_norm": 0.12248726189136505, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.1247, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5614798694232862, | |
| "grad_norm": 0.12791091203689575, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.1537, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5658324265505985, | |
| "grad_norm": 0.13339433073997498, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.1296, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5701849836779108, | |
| "grad_norm": 0.14575669169425964, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.1444, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5745375408052231, | |
| "grad_norm": 0.13964064419269562, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.1235, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5788900979325353, | |
| "grad_norm": 0.200956791639328, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.1935, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5832426550598476, | |
| "grad_norm": 0.1835886538028717, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.1682, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5875952121871599, | |
| "grad_norm": 0.19440779089927673, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.1884, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5919477693144722, | |
| "grad_norm": 0.21030087769031525, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.1648, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5963003264417845, | |
| "grad_norm": 0.27770769596099854, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.2192, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.6006528835690969, | |
| "grad_norm": 0.25254374742507935, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.1805, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.6050054406964092, | |
| "grad_norm": 0.2573109567165375, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.1884, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.6093579978237215, | |
| "grad_norm": 0.2900760769844055, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.224, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6137105549510338, | |
| "grad_norm": 0.26803362369537354, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.2029, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.6180631120783461, | |
| "grad_norm": 0.34686851501464844, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3151, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.6224156692056583, | |
| "grad_norm": 0.3472963571548462, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.2966, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.6267682263329706, | |
| "grad_norm": 0.3295505940914154, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.2848, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.6311207834602829, | |
| "grad_norm": 0.3306887745857239, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.31, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.6354733405875952, | |
| "grad_norm": 0.41762396693229675, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.3203, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.6398258977149075, | |
| "grad_norm": 0.38133504986763, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.2482, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6441784548422198, | |
| "grad_norm": 0.48297107219696045, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.3071, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6485310119695321, | |
| "grad_norm": 0.5673854947090149, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.378, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6528835690968444, | |
| "grad_norm": 0.6079791784286499, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.251, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6528835690968444, | |
| "eval_loss": 0.15901511907577515, | |
| "eval_runtime": 22.398, | |
| "eval_samples_per_second": 17.278, | |
| "eval_steps_per_second": 4.331, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6572361262241567, | |
| "grad_norm": 0.0663219541311264, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0757, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.661588683351469, | |
| "grad_norm": 0.07519705593585968, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.101, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6659412404787813, | |
| "grad_norm": 0.08506844192743301, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.1127, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6702937976060935, | |
| "grad_norm": 0.08246752619743347, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.1034, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6746463547334058, | |
| "grad_norm": 0.08656320720911026, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.1056, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6789989118607181, | |
| "grad_norm": 0.08865474909543991, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.1242, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6833514689880305, | |
| "grad_norm": 0.0883125439286232, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.1147, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.6877040261153428, | |
| "grad_norm": 0.09041907638311386, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.1076, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6920565832426551, | |
| "grad_norm": 0.08580870181322098, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.1019, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6964091403699674, | |
| "grad_norm": 0.09165311604738235, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.1191, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.7007616974972797, | |
| "grad_norm": 0.09156130254268646, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.1193, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.705114254624592, | |
| "grad_norm": 0.08939055353403091, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.1082, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.7094668117519043, | |
| "grad_norm": 0.09799733012914658, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.1187, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.7138193688792165, | |
| "grad_norm": 0.09849333018064499, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.1133, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.7181719260065288, | |
| "grad_norm": 0.09366466850042343, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.1155, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.7225244831338411, | |
| "grad_norm": 0.09362225234508514, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.1142, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.7268770402611534, | |
| "grad_norm": 0.10103306919336319, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.1246, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.7312295973884657, | |
| "grad_norm": 0.10364561527967453, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.1232, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.735582154515778, | |
| "grad_norm": 0.09509486705064774, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.1192, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.7399347116430903, | |
| "grad_norm": 0.08972634375095367, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.1125, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.7442872687704026, | |
| "grad_norm": 0.11448008567094803, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.135, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.7486398258977149, | |
| "grad_norm": 0.10920128226280212, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.1303, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7529923830250272, | |
| "grad_norm": 0.10236736387014389, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.1132, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7573449401523396, | |
| "grad_norm": 0.1080276295542717, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.1318, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7616974972796517, | |
| "grad_norm": 0.10686593502759933, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.1209, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.766050054406964, | |
| "grad_norm": 0.10954763740301132, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.1247, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7704026115342764, | |
| "grad_norm": 0.11578107625246048, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.1248, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7747551686615887, | |
| "grad_norm": 0.1304759532213211, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.1355, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.779107725788901, | |
| "grad_norm": 0.12112031877040863, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.1234, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7834602829162133, | |
| "grad_norm": 0.1396644562482834, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.1471, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7878128400435256, | |
| "grad_norm": 0.12432938814163208, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.1065, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7921653971708379, | |
| "grad_norm": 0.17179977893829346, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.1776, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7965179542981502, | |
| "grad_norm": 0.16509640216827393, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.1528, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.8008705114254625, | |
| "grad_norm": 0.2000688910484314, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.1579, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.8052230685527747, | |
| "grad_norm": 0.23999620974063873, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.1884, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.809575625680087, | |
| "grad_norm": 0.21986277401447296, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.2294, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.8139281828073993, | |
| "grad_norm": 0.29389095306396484, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.2197, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.8182807399347116, | |
| "grad_norm": 0.2735345959663391, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.1878, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.8226332970620239, | |
| "grad_norm": 0.24521368741989136, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.1607, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.8269858541893362, | |
| "grad_norm": 0.3278432786464691, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.216, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8313384113166485, | |
| "grad_norm": 0.3284049332141876, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.2048, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.8356909684439608, | |
| "grad_norm": 0.30655503273010254, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.2171, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.8400435255712732, | |
| "grad_norm": 0.3132682740688324, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.1827, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.8443960826985855, | |
| "grad_norm": 0.352733850479126, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.2007, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.8487486398258978, | |
| "grad_norm": 0.4017810821533203, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.2871, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.85310119695321, | |
| "grad_norm": 0.38403624296188354, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.3194, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8574537540805223, | |
| "grad_norm": 0.45041513442993164, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.3612, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8618063112078346, | |
| "grad_norm": 0.5342302322387695, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.3934, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8661588683351469, | |
| "grad_norm": 0.5464813113212585, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.3761, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8705114254624592, | |
| "grad_norm": 0.5738558769226074, | |
| "learning_rate": 0.0, | |
| "loss": 0.2627, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8705114254624592, | |
| "eval_loss": 0.15422219038009644, | |
| "eval_runtime": 22.3076, | |
| "eval_samples_per_second": 17.348, | |
| "eval_steps_per_second": 4.348, | |
| "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": 1.6973087530549248e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |