| { | |
| "best_metric": 3.17742919921875, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.11245431543435479, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.000562271577171774, | |
| "grad_norm": 3.3372039794921875, | |
| "learning_rate": 1e-05, | |
| "loss": 12.4849, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.000562271577171774, | |
| "eval_loss": 3.3527166843414307, | |
| "eval_runtime": 17.637, | |
| "eval_samples_per_second": 169.813, | |
| "eval_steps_per_second": 42.468, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.001124543154343548, | |
| "grad_norm": 3.7176997661590576, | |
| "learning_rate": 2e-05, | |
| "loss": 12.3758, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0016868147315153219, | |
| "grad_norm": 3.826338768005371, | |
| "learning_rate": 3e-05, | |
| "loss": 12.6623, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002249086308687096, | |
| "grad_norm": 3.477376699447632, | |
| "learning_rate": 4e-05, | |
| "loss": 12.0175, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0028113578858588698, | |
| "grad_norm": 3.2802834510803223, | |
| "learning_rate": 5e-05, | |
| "loss": 12.4318, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0033736294630306437, | |
| "grad_norm": 3.2721400260925293, | |
| "learning_rate": 6e-05, | |
| "loss": 12.5461, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.003935901040202418, | |
| "grad_norm": 3.382446765899658, | |
| "learning_rate": 7e-05, | |
| "loss": 12.0155, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.004498172617374192, | |
| "grad_norm": 3.6472439765930176, | |
| "learning_rate": 8e-05, | |
| "loss": 12.1008, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.005060444194545966, | |
| "grad_norm": 4.051973819732666, | |
| "learning_rate": 9e-05, | |
| "loss": 12.5926, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0056227157717177395, | |
| "grad_norm": 3.8108744621276855, | |
| "learning_rate": 0.0001, | |
| "loss": 12.5172, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.006184987348889514, | |
| "grad_norm": 3.69618821144104, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 12.283, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0067472589260612875, | |
| "grad_norm": 3.352055311203003, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 12.1051, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.007309530503233062, | |
| "grad_norm": 3.3662631511688232, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 12.6878, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.007871802080404836, | |
| "grad_norm": 3.475145101547241, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 13.3067, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.00843407365757661, | |
| "grad_norm": 3.408506393432617, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 12.3699, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.008996345234748383, | |
| "grad_norm": 3.5367794036865234, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 12.7921, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.009558616811920157, | |
| "grad_norm": 3.3267605304718018, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 12.7811, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.010120888389091932, | |
| "grad_norm": 3.1486122608184814, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 12.5525, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.010683159966263706, | |
| "grad_norm": 3.204435348510742, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 13.0802, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.011245431543435479, | |
| "grad_norm": 2.965869426727295, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 12.6219, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.011807703120607253, | |
| "grad_norm": 3.132817268371582, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 12.9339, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.012369974697779028, | |
| "grad_norm": 2.9751439094543457, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 12.5705, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.012932246274950801, | |
| "grad_norm": 3.1358697414398193, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 12.6131, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.013494517852122575, | |
| "grad_norm": 3.1683080196380615, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 12.4933, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.014056789429294348, | |
| "grad_norm": 3.2347140312194824, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 12.4986, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.014619061006466124, | |
| "grad_norm": 3.1125946044921875, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 12.4543, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.015181332583637897, | |
| "grad_norm": 3.037888288497925, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 12.1705, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.015743604160809672, | |
| "grad_norm": 3.2020065784454346, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 12.4811, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.016305875737981444, | |
| "grad_norm": 2.967869281768799, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 12.1209, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01686814731515322, | |
| "grad_norm": 3.1968579292297363, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 12.2455, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01743041889232499, | |
| "grad_norm": 3.2552599906921387, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 12.0455, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.017992690469496767, | |
| "grad_norm": 3.5462772846221924, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 12.1891, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.018554962046668542, | |
| "grad_norm": 3.96854829788208, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 12.2904, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.019117233623840314, | |
| "grad_norm": 9.371252059936523, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 15.0445, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01967950520101209, | |
| "grad_norm": 9.009614944458008, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 15.1092, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.020241776778183864, | |
| "grad_norm": 9.642281532287598, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 15.4099, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.020804048355355636, | |
| "grad_norm": 8.54543685913086, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 15.4823, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02136631993252741, | |
| "grad_norm": 7.349517345428467, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 15.2429, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.021928591509699186, | |
| "grad_norm": 7.565888404846191, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 15.5465, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.022490863086870958, | |
| "grad_norm": 6.928400039672852, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 15.0218, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.023053134664042733, | |
| "grad_norm": 7.2679762840271, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 14.6989, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.023615406241214505, | |
| "grad_norm": 7.323550224304199, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 14.6985, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.02417767781838628, | |
| "grad_norm": 7.29262113571167, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 14.6745, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.024739949395558056, | |
| "grad_norm": 7.692154407501221, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 14.4694, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.025302220972729828, | |
| "grad_norm": 7.802225112915039, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 14.8889, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.025864492549901603, | |
| "grad_norm": 7.938776016235352, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 14.5935, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.026426764127073378, | |
| "grad_norm": 7.427006721496582, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 14.0019, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02698903570424515, | |
| "grad_norm": 7.957850456237793, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 14.4919, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.027551307281416925, | |
| "grad_norm": 8.102659225463867, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 14.3474, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.028113578858588697, | |
| "grad_norm": 8.571439743041992, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 13.8711, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.028113578858588697, | |
| "eval_loss": 3.3471429347991943, | |
| "eval_runtime": 17.4913, | |
| "eval_samples_per_second": 171.228, | |
| "eval_steps_per_second": 42.821, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.028675850435760472, | |
| "grad_norm": 16.4812068939209, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 13.421, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.029238122012932247, | |
| "grad_norm": 12.559088706970215, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 13.2086, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.02980039359010402, | |
| "grad_norm": 9.944561958312988, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 12.9132, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.030362665167275794, | |
| "grad_norm": 7.988941192626953, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 12.3812, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03092493674444757, | |
| "grad_norm": 6.001148223876953, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 12.2565, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.031487208321619345, | |
| "grad_norm": 4.097027778625488, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 12.3601, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03204947989879112, | |
| "grad_norm": 3.436241388320923, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 12.3088, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03261175147596289, | |
| "grad_norm": 3.31573224067688, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 12.5382, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03317402305313467, | |
| "grad_norm": 3.3345532417297363, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 12.5246, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03373629463030644, | |
| "grad_norm": 3.238633394241333, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 12.7151, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03429856620747821, | |
| "grad_norm": 3.048017978668213, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 12.0159, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.03486083778464998, | |
| "grad_norm": 2.7373361587524414, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 12.2511, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03542310936182176, | |
| "grad_norm": 2.6594111919403076, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 12.3634, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.03598538093899353, | |
| "grad_norm": 2.713179588317871, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 12.419, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.036547652516165305, | |
| "grad_norm": 2.453629732131958, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 12.245, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.037109924093337084, | |
| "grad_norm": 2.778935432434082, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 12.5839, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.037672195670508855, | |
| "grad_norm": 2.6000263690948486, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 12.3091, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03823446724768063, | |
| "grad_norm": 2.613262176513672, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 12.3092, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.038796738824852406, | |
| "grad_norm": 2.73067569732666, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 12.6266, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03935901040202418, | |
| "grad_norm": 2.5190043449401855, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 13.0876, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03992128197919595, | |
| "grad_norm": 2.574352979660034, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 12.2462, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.04048355355636773, | |
| "grad_norm": 2.501310110092163, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 12.7794, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0410458251335395, | |
| "grad_norm": 2.5318753719329834, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 12.7336, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.04160809671071127, | |
| "grad_norm": 2.469571828842163, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 12.351, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.04217036828788305, | |
| "grad_norm": 2.513092279434204, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 12.6951, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04273263986505482, | |
| "grad_norm": 2.5848042964935303, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 12.614, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.043294911442226594, | |
| "grad_norm": 2.5222952365875244, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 12.5491, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.04385718301939837, | |
| "grad_norm": 2.7168142795562744, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 12.1427, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.044419454596570145, | |
| "grad_norm": 2.5504140853881836, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 12.1761, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.044981726173741916, | |
| "grad_norm": 2.584420680999756, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 12.3628, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04554399775091369, | |
| "grad_norm": 2.6327569484710693, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 11.9888, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.04610626932808547, | |
| "grad_norm": 2.6863629817962646, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 12.3543, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.04666854090525724, | |
| "grad_norm": 3.053786516189575, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 12.0475, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.04723081248242901, | |
| "grad_norm": 4.870243549346924, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 13.3879, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.04779308405960079, | |
| "grad_norm": 7.5411176681518555, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 14.7573, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.04835535563677256, | |
| "grad_norm": 6.51435661315918, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 14.2007, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.04891762721394433, | |
| "grad_norm": 6.183172225952148, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 14.5157, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.04947989879111611, | |
| "grad_norm": 6.011676788330078, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 14.7387, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.05004217036828788, | |
| "grad_norm": 5.791100978851318, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 13.8414, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.050604441945459655, | |
| "grad_norm": 5.774806499481201, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 14.2778, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.051166713522631434, | |
| "grad_norm": 5.833765029907227, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 14.2561, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.051728985099803206, | |
| "grad_norm": 5.757104873657227, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 14.6949, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.05229125667697498, | |
| "grad_norm": 5.482483386993408, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 13.5382, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.052853528254146756, | |
| "grad_norm": 5.702921390533447, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 13.421, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.05341579983131853, | |
| "grad_norm": 5.867873191833496, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 14.3085, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0539780714084903, | |
| "grad_norm": 6.226456642150879, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 14.3549, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.05454034298566207, | |
| "grad_norm": 6.726083755493164, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 14.3766, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05510261456283385, | |
| "grad_norm": 6.9603776931762695, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 14.5228, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.05566488614000562, | |
| "grad_norm": 7.549854755401611, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 13.9763, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.056227157717177394, | |
| "grad_norm": 8.101652145385742, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 14.7456, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.056227157717177394, | |
| "eval_loss": 3.2219655513763428, | |
| "eval_runtime": 17.4998, | |
| "eval_samples_per_second": 171.145, | |
| "eval_steps_per_second": 42.8, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05678942929434917, | |
| "grad_norm": 4.253426551818848, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 12.8723, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.057351700871520944, | |
| "grad_norm": 3.9202003479003906, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 12.5568, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.057913972448692716, | |
| "grad_norm": 3.8417775630950928, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 12.3865, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.058476244025864495, | |
| "grad_norm": 3.405397891998291, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 12.2352, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.059038515603036266, | |
| "grad_norm": 3.101681709289551, | |
| "learning_rate": 5e-05, | |
| "loss": 12.4573, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.05960078718020804, | |
| "grad_norm": 3.01593017578125, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 12.3869, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.06016305875737982, | |
| "grad_norm": 2.5413131713867188, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 12.4314, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.06072533033455159, | |
| "grad_norm": 2.533918857574463, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 12.6719, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.06128760191172336, | |
| "grad_norm": 2.4495816230773926, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 12.3535, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.06184987348889514, | |
| "grad_norm": 2.3273746967315674, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 12.2099, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06241214506606691, | |
| "grad_norm": 2.261077880859375, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 12.1137, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.06297441664323869, | |
| "grad_norm": 2.3837461471557617, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 12.153, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.06353668822041046, | |
| "grad_norm": 2.4928786754608154, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 12.4768, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.06409895979758223, | |
| "grad_norm": 2.2755861282348633, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 12.2836, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.064661231374754, | |
| "grad_norm": 2.392712354660034, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 12.2861, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06522350295192578, | |
| "grad_norm": 2.348325490951538, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 12.5431, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06578577452909755, | |
| "grad_norm": 2.284317970275879, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 11.9487, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06634804610626933, | |
| "grad_norm": 2.251986265182495, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 12.3718, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.0669103176834411, | |
| "grad_norm": 2.172983169555664, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 12.0939, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.06747258926061288, | |
| "grad_norm": 2.21427059173584, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 12.1701, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06803486083778465, | |
| "grad_norm": 2.2606608867645264, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 12.4348, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.06859713241495642, | |
| "grad_norm": 2.405468702316284, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 12.1653, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0691594039921282, | |
| "grad_norm": 2.3203816413879395, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 12.7836, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.06972167556929997, | |
| "grad_norm": 2.3274612426757812, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 12.7792, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.07028394714647175, | |
| "grad_norm": 2.3437340259552, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 12.5349, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07084621872364352, | |
| "grad_norm": 2.256972312927246, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 12.3255, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0714084903008153, | |
| "grad_norm": 2.30515718460083, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 12.2751, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.07197076187798707, | |
| "grad_norm": 2.284391403198242, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 12.0432, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.07253303345515884, | |
| "grad_norm": 2.3369052410125732, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 12.3668, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.07309530503233061, | |
| "grad_norm": 2.372157096862793, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 12.428, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0736575766095024, | |
| "grad_norm": 2.453045129776001, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 12.2144, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07421984818667417, | |
| "grad_norm": 2.5379128456115723, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 11.9208, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.07478211976384594, | |
| "grad_norm": 2.6493380069732666, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 12.1517, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07534439134101771, | |
| "grad_norm": 2.9889986515045166, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 11.8513, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.07590666291818948, | |
| "grad_norm": 3.625115156173706, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 13.3785, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07646893449536125, | |
| "grad_norm": 5.983370304107666, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 14.8017, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.07703120607253304, | |
| "grad_norm": 6.331634998321533, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 15.3347, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.07759347764970481, | |
| "grad_norm": 5.848264217376709, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 13.9854, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.07815574922687658, | |
| "grad_norm": 6.60196590423584, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 14.8867, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.07871802080404836, | |
| "grad_norm": 6.4960222244262695, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 13.9309, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07928029238122013, | |
| "grad_norm": 6.343400478363037, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 14.6485, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.0798425639583919, | |
| "grad_norm": 6.459204196929932, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 14.5744, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.08040483553556367, | |
| "grad_norm": 5.965136528015137, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 14.2422, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.08096710711273546, | |
| "grad_norm": 5.720399379730225, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 14.1928, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.08152937868990723, | |
| "grad_norm": 6.280338287353516, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 14.3255, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.082091650267079, | |
| "grad_norm": 6.046666622161865, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 14.6397, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.08265392184425077, | |
| "grad_norm": 6.060023307800293, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 13.7786, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.08321619342142254, | |
| "grad_norm": 6.446502685546875, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 14.4263, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.08377846499859432, | |
| "grad_norm": 6.9296555519104, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 13.9868, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.0843407365757661, | |
| "grad_norm": 7.5446319580078125, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 15.1491, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0843407365757661, | |
| "eval_loss": 3.1836774349212646, | |
| "eval_runtime": 17.4879, | |
| "eval_samples_per_second": 171.262, | |
| "eval_steps_per_second": 42.83, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08490300815293787, | |
| "grad_norm": 2.5498669147491455, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 12.4063, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.08546527973010964, | |
| "grad_norm": 2.699005603790283, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 12.5751, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.08602755130728142, | |
| "grad_norm": 2.4554402828216553, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 12.2151, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.08658982288445319, | |
| "grad_norm": 2.4071147441864014, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 11.7472, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.08715209446162496, | |
| "grad_norm": 2.530582904815674, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 11.7858, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.08771436603879675, | |
| "grad_norm": 2.480393409729004, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 11.7175, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.08827663761596852, | |
| "grad_norm": 2.3280837535858154, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 12.4496, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.08883890919314029, | |
| "grad_norm": 2.605595827102661, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 12.88, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.08940118077031206, | |
| "grad_norm": 2.460703134536743, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 11.7988, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.08996345234748383, | |
| "grad_norm": 2.192172050476074, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 12.0285, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0905257239246556, | |
| "grad_norm": 2.3072140216827393, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 12.657, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.09108799550182738, | |
| "grad_norm": 2.224512815475464, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 12.5217, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.09165026707899916, | |
| "grad_norm": 2.1440680027008057, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 12.5891, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.09221253865617093, | |
| "grad_norm": 2.150700569152832, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 12.1435, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.0927748102333427, | |
| "grad_norm": 2.138625144958496, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 12.156, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.09333708181051448, | |
| "grad_norm": 2.2841179370880127, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 12.912, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.09389935338768625, | |
| "grad_norm": 2.275604486465454, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 11.9028, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.09446162496485802, | |
| "grad_norm": 2.324662446975708, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 13.1071, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.0950238965420298, | |
| "grad_norm": 2.2576868534088135, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 12.7031, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.09558616811920158, | |
| "grad_norm": 2.2652509212493896, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 12.1913, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09614843969637335, | |
| "grad_norm": 2.3637468814849854, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 12.6637, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.09671071127354512, | |
| "grad_norm": 2.173945665359497, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 12.6121, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.0972729828507169, | |
| "grad_norm": 2.34234356880188, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 12.4375, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.09783525442788867, | |
| "grad_norm": 2.3364017009735107, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 12.4843, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.09839752600506045, | |
| "grad_norm": 2.230928421020508, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 12.5065, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09895979758223222, | |
| "grad_norm": 2.349921703338623, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 12.1496, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.099522069159404, | |
| "grad_norm": 2.278562545776367, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 12.3653, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.10008434073657577, | |
| "grad_norm": 2.464076280593872, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 12.3393, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.10064661231374754, | |
| "grad_norm": 2.396714687347412, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 12.2902, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.10120888389091931, | |
| "grad_norm": 2.555405378341675, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 12.8033, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.10177115546809108, | |
| "grad_norm": 2.629749298095703, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 11.9488, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.10233342704526287, | |
| "grad_norm": 2.619595766067505, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 11.9838, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.10289569862243464, | |
| "grad_norm": 2.9505879878997803, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 11.8494, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.10345797019960641, | |
| "grad_norm": 3.3083057403564453, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 11.8519, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.10402024177677818, | |
| "grad_norm": 3.8817014694213867, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 13.4171, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.10458251335394995, | |
| "grad_norm": 4.491519927978516, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 14.0314, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.10514478493112173, | |
| "grad_norm": 4.502686023712158, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 13.7681, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.10570705650829351, | |
| "grad_norm": 4.6369500160217285, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 14.1065, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.10626932808546528, | |
| "grad_norm": 4.749708652496338, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 14.0934, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.10683159966263706, | |
| "grad_norm": 4.917808532714844, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 13.7322, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.10739387123980883, | |
| "grad_norm": 5.4032769203186035, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 14.2303, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.1079561428169806, | |
| "grad_norm": 5.083553791046143, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 14.4168, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.10851841439415237, | |
| "grad_norm": 5.418875217437744, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 13.8963, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.10908068597132414, | |
| "grad_norm": 5.617504596710205, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 13.8051, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10964295754849593, | |
| "grad_norm": 5.5453104972839355, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 14.4726, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1102052291256677, | |
| "grad_norm": 5.646365165710449, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 14.2539, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.11076750070283947, | |
| "grad_norm": 6.305400371551514, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 13.8898, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.11132977228001124, | |
| "grad_norm": 6.601794242858887, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 13.9095, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.11189204385718302, | |
| "grad_norm": 6.902284145355225, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 14.2948, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.11245431543435479, | |
| "grad_norm": 7.8821258544921875, | |
| "learning_rate": 0.0, | |
| "loss": 14.6044, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11245431543435479, | |
| "eval_loss": 3.17742919921875, | |
| "eval_runtime": 17.6713, | |
| "eval_samples_per_second": 169.484, | |
| "eval_steps_per_second": 42.385, | |
| "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": 3776116363886592.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |