| { | |
| "best_metric": 0.9981662631034851, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.018976232269082974, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00018976232269082974, | |
| "grad_norm": 0.3689703047275543, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0351, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00018976232269082974, | |
| "eval_loss": 1.3198952674865723, | |
| "eval_runtime": 275.6593, | |
| "eval_samples_per_second": 32.199, | |
| "eval_steps_per_second": 8.05, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00037952464538165947, | |
| "grad_norm": 0.32817262411117554, | |
| "learning_rate": 2e-05, | |
| "loss": 1.1404, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0005692869680724893, | |
| "grad_norm": 0.4041525423526764, | |
| "learning_rate": 3e-05, | |
| "loss": 1.1584, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0007590492907633189, | |
| "grad_norm": 0.3785117268562317, | |
| "learning_rate": 4e-05, | |
| "loss": 1.0783, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0009488116134541487, | |
| "grad_norm": 0.43445149064064026, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1869, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0011385739361449785, | |
| "grad_norm": 0.3155576288700104, | |
| "learning_rate": 6e-05, | |
| "loss": 1.1816, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.001328336258835808, | |
| "grad_norm": 0.26356881856918335, | |
| "learning_rate": 7e-05, | |
| "loss": 1.1151, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0015180985815266379, | |
| "grad_norm": 0.16498738527297974, | |
| "learning_rate": 8e-05, | |
| "loss": 1.21, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0017078609042174677, | |
| "grad_norm": 0.16361583769321442, | |
| "learning_rate": 9e-05, | |
| "loss": 1.1318, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0018976232269082975, | |
| "grad_norm": 0.15728989243507385, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1683, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0020873855495991273, | |
| "grad_norm": 0.18534033000469208, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.1149, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.002277147872289957, | |
| "grad_norm": 0.18808847665786743, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.094, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0024669101949807864, | |
| "grad_norm": 0.1968209147453308, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.0957, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.002656672517671616, | |
| "grad_norm": 0.18778468668460846, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.1423, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.002846434840362446, | |
| "grad_norm": 0.18522831797599792, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.1087, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0030361971630532758, | |
| "grad_norm": 0.1853436827659607, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.1391, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0032259594857441056, | |
| "grad_norm": 0.18637384474277496, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.1179, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0034157218084349353, | |
| "grad_norm": 0.18694156408309937, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.159, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.003605484131125765, | |
| "grad_norm": 0.1744527667760849, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.1811, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.003795246453816595, | |
| "grad_norm": 0.17337507009506226, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.1214, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.003985008776507424, | |
| "grad_norm": 0.17523472011089325, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.1186, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0041747710991982545, | |
| "grad_norm": 0.1785983443260193, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.1123, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.004364533421889084, | |
| "grad_norm": 0.17681241035461426, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.0687, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.004554295744579914, | |
| "grad_norm": 0.18389420211315155, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.1033, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0047440580672707434, | |
| "grad_norm": 0.18924522399902344, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.0684, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.004933820389961573, | |
| "grad_norm": 0.17484529316425323, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.0511, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.005123582712652403, | |
| "grad_norm": 0.17750877141952515, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.1304, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.005313345035343232, | |
| "grad_norm": 0.178945392370224, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.137, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.005503107358034063, | |
| "grad_norm": 0.1712048500776291, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.0915, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.005692869680724892, | |
| "grad_norm": 0.19296303391456604, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.1328, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.005882632003415722, | |
| "grad_norm": 0.18090558052062988, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.0345, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0060723943261065515, | |
| "grad_norm": 0.1833214908838272, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.0776, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.006262156648797381, | |
| "grad_norm": 0.17969094216823578, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.1061, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.006451918971488211, | |
| "grad_norm": 0.18190151453018188, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.0681, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0066416812941790405, | |
| "grad_norm": 0.19453197717666626, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.0855, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.006831443616869871, | |
| "grad_norm": 0.18432414531707764, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.0901, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0070212059395607, | |
| "grad_norm": 0.20541809499263763, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.1055, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.00721096826225153, | |
| "grad_norm": 0.19959476590156555, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.102, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.00740073058494236, | |
| "grad_norm": 0.1939975470304489, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.1024, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.00759049290763319, | |
| "grad_norm": 0.19972407817840576, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.9908, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.007780255230324019, | |
| "grad_norm": 0.2053292840719223, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.0316, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.007970017553014849, | |
| "grad_norm": 0.21963410079479218, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.1153, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.008159779875705679, | |
| "grad_norm": 0.2149333655834198, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.1209, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.008349542198396509, | |
| "grad_norm": 0.2174244076013565, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.1614, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.008539304521087338, | |
| "grad_norm": 0.23298555612564087, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.0815, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.008729066843778168, | |
| "grad_norm": 0.26479408144950867, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.1979, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.008918829166468998, | |
| "grad_norm": 0.24461990594863892, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.0346, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.009108591489159828, | |
| "grad_norm": 0.2512701451778412, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.0319, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.009298353811850657, | |
| "grad_norm": 0.2517155110836029, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.055, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.009488116134541487, | |
| "grad_norm": 0.2827569246292114, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.1684, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.009488116134541487, | |
| "eval_loss": 1.0552674531936646, | |
| "eval_runtime": 277.3628, | |
| "eval_samples_per_second": 32.001, | |
| "eval_steps_per_second": 8.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.009677878457232317, | |
| "grad_norm": 0.21244028210639954, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.9075, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.009867640779923146, | |
| "grad_norm": 0.20778271555900574, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.045, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.010057403102613976, | |
| "grad_norm": 0.19142813980579376, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.0145, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.010247165425304806, | |
| "grad_norm": 0.1807219535112381, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.0956, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.010436927747995636, | |
| "grad_norm": 0.16223390400409698, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.0037, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.010626690070686465, | |
| "grad_norm": 0.15767407417297363, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.9815, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.010816452393377295, | |
| "grad_norm": 0.14841850101947784, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.004, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.011006214716068125, | |
| "grad_norm": 0.14706046879291534, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.0287, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.011195977038758954, | |
| "grad_norm": 0.16504324972629547, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.0654, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.011385739361449784, | |
| "grad_norm": 0.1560998260974884, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.9596, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.011575501684140614, | |
| "grad_norm": 0.16232743859291077, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.0206, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.011765264006831444, | |
| "grad_norm": 0.16730357706546783, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.9955, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.011955026329522273, | |
| "grad_norm": 0.1669314056634903, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.9638, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.012144788652213103, | |
| "grad_norm": 0.17288701236248016, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.0123, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.012334550974903933, | |
| "grad_norm": 0.1711699515581131, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.024, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.012524313297594762, | |
| "grad_norm": 0.17235831916332245, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.0528, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.012714075620285592, | |
| "grad_norm": 0.17410272359848022, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.1027, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.012903837942976422, | |
| "grad_norm": 0.16469921171665192, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.9787, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.013093600265667252, | |
| "grad_norm": 0.1830262541770935, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.9897, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.013283362588358081, | |
| "grad_norm": 0.18811868131160736, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.0582, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.013473124911048911, | |
| "grad_norm": 0.19638700783252716, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.1117, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.013662887233739741, | |
| "grad_norm": 0.18470486998558044, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.0433, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.01385264955643057, | |
| "grad_norm": 0.1824895590543747, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.047, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0140424118791214, | |
| "grad_norm": 0.18160541355609894, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.0565, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.01423217420181223, | |
| "grad_norm": 0.1841384768486023, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.9941, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01442193652450306, | |
| "grad_norm": 0.19593071937561035, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.0795, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.014611698847193889, | |
| "grad_norm": 0.19380022585391998, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.0292, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.01480146116988472, | |
| "grad_norm": 0.1812576800584793, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.0105, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.01499122349257555, | |
| "grad_norm": 0.2010316103696823, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.0638, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.01518098581526638, | |
| "grad_norm": 0.19539521634578705, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.942, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.015370748137957208, | |
| "grad_norm": 0.19889476895332336, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.0528, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.015560510460648038, | |
| "grad_norm": 0.21053452789783478, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.0796, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.01575027278333887, | |
| "grad_norm": 0.20258855819702148, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.9893, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.015940035106029697, | |
| "grad_norm": 0.20124588906764984, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.9893, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.01612979742872053, | |
| "grad_norm": 0.233975350856781, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.0696, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.016319559751411358, | |
| "grad_norm": 0.2315712422132492, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.0057, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.016509322074102186, | |
| "grad_norm": 0.23223906755447388, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.0939, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.016699084396793018, | |
| "grad_norm": 0.23546229302883148, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.0663, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.016888846719483847, | |
| "grad_norm": 0.22160938382148743, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.9972, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.017078609042174675, | |
| "grad_norm": 0.23993346095085144, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.9169, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.017268371364865507, | |
| "grad_norm": 0.23985272645950317, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.0183, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.017458133687556335, | |
| "grad_norm": 0.24388989806175232, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.0668, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.017647896010247164, | |
| "grad_norm": 0.2586608827114105, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.0306, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.017837658332937996, | |
| "grad_norm": 0.251071035861969, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.9846, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.018027420655628824, | |
| "grad_norm": 0.2634812593460083, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.0191, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.018217182978319656, | |
| "grad_norm": 0.26588213443756104, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.988, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.018406945301010485, | |
| "grad_norm": 0.2804090976715088, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.0653, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.018596707623701313, | |
| "grad_norm": 0.2770501971244812, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.0198, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.018786469946392145, | |
| "grad_norm": 0.31081750988960266, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.9818, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.018976232269082974, | |
| "grad_norm": 0.3638320863246918, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.9592, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.018976232269082974, | |
| "eval_loss": 0.9981662631034851, | |
| "eval_runtime": 277.1213, | |
| "eval_samples_per_second": 32.029, | |
| "eval_steps_per_second": 8.007, | |
| "step": 100 | |
| } | |
| ], | |
| "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": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.3421305059475456e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |