| { | |
| "best_metric": 1.6850194931030273, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.038248231019315355, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00019124115509657678, | |
| "grad_norm": 13.741756439208984, | |
| "learning_rate": 1e-05, | |
| "loss": 8.6301, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00019124115509657678, | |
| "eval_loss": 2.477055072784424, | |
| "eval_runtime": 951.1783, | |
| "eval_samples_per_second": 9.259, | |
| "eval_steps_per_second": 2.315, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00038248231019315357, | |
| "grad_norm": 17.47914695739746, | |
| "learning_rate": 2e-05, | |
| "loss": 9.8839, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0005737234652897303, | |
| "grad_norm": 12.253708839416504, | |
| "learning_rate": 3e-05, | |
| "loss": 9.0649, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0007649646203863071, | |
| "grad_norm": 14.06733512878418, | |
| "learning_rate": 4e-05, | |
| "loss": 9.6933, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0009562057754828839, | |
| "grad_norm": 14.397188186645508, | |
| "learning_rate": 5e-05, | |
| "loss": 9.3439, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0011474469305794606, | |
| "grad_norm": 10.559906959533691, | |
| "learning_rate": 6e-05, | |
| "loss": 8.7796, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0013386880856760374, | |
| "grad_norm": 9.843297958374023, | |
| "learning_rate": 7e-05, | |
| "loss": 8.2437, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0015299292407726143, | |
| "grad_norm": 10.483661651611328, | |
| "learning_rate": 8e-05, | |
| "loss": 7.6632, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0017211703958691911, | |
| "grad_norm": 8.265406608581543, | |
| "learning_rate": 9e-05, | |
| "loss": 8.6506, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0019124115509657677, | |
| "grad_norm": 25.431543350219727, | |
| "learning_rate": 0.0001, | |
| "loss": 8.3187, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0021036527060623446, | |
| "grad_norm": 14.77153205871582, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 7.606, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.002294893861158921, | |
| "grad_norm": 542.9242553710938, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 8.8296, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0024861350162554982, | |
| "grad_norm": 1506.7982177734375, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 10.4389, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.002677376171352075, | |
| "grad_norm": 833.0812377929688, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 9.2056, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.002868617326448652, | |
| "grad_norm": 114.60613250732422, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 8.0894, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0030598584815452285, | |
| "grad_norm": 43.345279693603516, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 8.2857, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.003251099636641805, | |
| "grad_norm": 24.452478408813477, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 7.8226, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0034423407917383822, | |
| "grad_norm": 8.936249732971191, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 7.7073, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.003633581946834959, | |
| "grad_norm": 8.189146995544434, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 8.3366, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0038248231019315355, | |
| "grad_norm": 10.676236152648926, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 7.1745, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.004016064257028112, | |
| "grad_norm": 8.025724411010742, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 7.9372, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.004207305412124689, | |
| "grad_norm": 6.501006603240967, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 8.1329, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.004398546567221266, | |
| "grad_norm": 6.821082592010498, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 7.832, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.004589787722317842, | |
| "grad_norm": 6.214833736419678, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 7.0542, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.004781028877414419, | |
| "grad_norm": 7.392245292663574, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 6.4323, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0049722700325109965, | |
| "grad_norm": 5.831737041473389, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 8.0168, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0051635111876075735, | |
| "grad_norm": 5.859936714172363, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 8.126, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.00535475234270415, | |
| "grad_norm": 6.41107702255249, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 7.0687, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.005545993497800727, | |
| "grad_norm": 4.913946628570557, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 7.3068, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.005737234652897304, | |
| "grad_norm": 5.120382308959961, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 7.2256, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.00592847580799388, | |
| "grad_norm": 5.55898380279541, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 7.9069, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.006119716963090457, | |
| "grad_norm": 4.908207893371582, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 6.881, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.006310958118187034, | |
| "grad_norm": 5.752783298492432, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 5.963, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.00650219927328361, | |
| "grad_norm": 5.540194034576416, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 6.7955, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.006693440428380187, | |
| "grad_norm": 5.210753917694092, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 6.8377, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0068846815834767644, | |
| "grad_norm": 6.288083076477051, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 7.1198, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.007075922738573341, | |
| "grad_norm": 5.827837944030762, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 7.1235, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.007267163893669918, | |
| "grad_norm": 5.469511985778809, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 6.384, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.007458405048766495, | |
| "grad_norm": 4.952732086181641, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 6.5827, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.007649646203863071, | |
| "grad_norm": 13.096961975097656, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 6.8915, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.007840887358959648, | |
| "grad_norm": 5.062122821807861, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 7.1196, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.008032128514056224, | |
| "grad_norm": 5.489322185516357, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 6.7514, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.008223369669152802, | |
| "grad_norm": 6.432612895965576, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 7.4563, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.008414610824249378, | |
| "grad_norm": 5.161056041717529, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 7.6811, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.008605851979345954, | |
| "grad_norm": 5.6301164627075195, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 6.5877, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.008797093134442532, | |
| "grad_norm": 4.932016849517822, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 6.754, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.008988334289539109, | |
| "grad_norm": 5.741036415100098, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 6.9514, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.009179575444635685, | |
| "grad_norm": 5.429935455322266, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 6.6378, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.009370816599732263, | |
| "grad_norm": 5.711150646209717, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 6.4125, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.009562057754828839, | |
| "grad_norm": 5.637681484222412, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 7.0322, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.009562057754828839, | |
| "eval_loss": 1.7700098752975464, | |
| "eval_runtime": 955.9137, | |
| "eval_samples_per_second": 9.213, | |
| "eval_steps_per_second": 2.304, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.009753298909925417, | |
| "grad_norm": 6.44619083404541, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 7.7286, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.009944540065021993, | |
| "grad_norm": 5.049639701843262, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 7.6365, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.01013578122011857, | |
| "grad_norm": 5.096268653869629, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 7.4237, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.010327022375215147, | |
| "grad_norm": 5.7089924812316895, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 7.6252, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.010518263530311723, | |
| "grad_norm": 5.822720050811768, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 7.1462, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0107095046854083, | |
| "grad_norm": 14.382341384887695, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 8.2893, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.010900745840504877, | |
| "grad_norm": 4.712852954864502, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 6.4838, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.011091986995601454, | |
| "grad_norm": 7.484261989593506, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 7.3246, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.01128322815069803, | |
| "grad_norm": 4.904869556427002, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 7.0472, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.011474469305794608, | |
| "grad_norm": 4.846579551696777, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 7.4069, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.011665710460891184, | |
| "grad_norm": 5.9781341552734375, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 7.3087, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.01185695161598776, | |
| "grad_norm": 5.264899730682373, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 7.3775, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.012048192771084338, | |
| "grad_norm": 5.397994041442871, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 7.6606, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.012239433926180914, | |
| "grad_norm": 4.566377639770508, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 7.344, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.01243067508127749, | |
| "grad_norm": 4.5347113609313965, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 7.4656, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.012621916236374068, | |
| "grad_norm": 5.169002056121826, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 7.5279, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.012813157391470644, | |
| "grad_norm": 5.186399459838867, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 7.3893, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.01300439854656722, | |
| "grad_norm": 4.876247406005859, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 7.1706, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.013195639701663799, | |
| "grad_norm": 5.117702484130859, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 6.7098, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.013386880856760375, | |
| "grad_norm": 4.702174186706543, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 6.6773, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.013578122011856951, | |
| "grad_norm": 4.804377555847168, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 7.4112, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.013769363166953529, | |
| "grad_norm": 4.555037975311279, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 7.6652, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.013960604322050105, | |
| "grad_norm": 4.946444034576416, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 6.9994, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.014151845477146681, | |
| "grad_norm": 4.153842449188232, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 6.4639, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.01434308663224326, | |
| "grad_norm": 5.230395317077637, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 7.4758, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.014534327787339835, | |
| "grad_norm": 4.902325630187988, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 6.7119, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.014725568942436412, | |
| "grad_norm": 4.707330226898193, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 7.2524, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.01491681009753299, | |
| "grad_norm": 4.5713324546813965, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 6.9364, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.015108051252629566, | |
| "grad_norm": 4.042422771453857, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 6.8059, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.015299292407726142, | |
| "grad_norm": 4.8989481925964355, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 7.0185, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01549053356282272, | |
| "grad_norm": 4.450347900390625, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 6.9879, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.015681774717919296, | |
| "grad_norm": 5.513014793395996, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 6.9122, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.015873015873015872, | |
| "grad_norm": 4.920793533325195, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 7.3208, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.01606425702811245, | |
| "grad_norm": 4.554596424102783, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 6.5676, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.016255498183209028, | |
| "grad_norm": 8.47518539428711, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 6.4218, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.016446739338305604, | |
| "grad_norm": 5.609909534454346, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 7.3215, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.01663798049340218, | |
| "grad_norm": 5.034678936004639, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 6.3352, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.016829221648498757, | |
| "grad_norm": 4.783220291137695, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 6.7082, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.017020462803595333, | |
| "grad_norm": 5.034786701202393, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 7.4092, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.01721170395869191, | |
| "grad_norm": 6.396137714385986, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 6.6575, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.01740294511378849, | |
| "grad_norm": 5.503007888793945, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 6.5431, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.017594186268885065, | |
| "grad_norm": 6.695736885070801, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 6.8052, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.01778542742398164, | |
| "grad_norm": 13.365432739257812, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 7.067, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.017976668579078217, | |
| "grad_norm": 6.272519111633301, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 7.0269, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.018167909734174793, | |
| "grad_norm": 4.763078212738037, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 6.5489, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.01835915088927137, | |
| "grad_norm": 5.521965980529785, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 6.5887, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.01855039204436795, | |
| "grad_norm": 4.557949542999268, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 6.3241, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.018741633199464525, | |
| "grad_norm": 4.075445652008057, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 6.2914, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0189328743545611, | |
| "grad_norm": 4.799678802490234, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 6.4466, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.019124115509657678, | |
| "grad_norm": 5.99141788482666, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 6.7024, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.019124115509657678, | |
| "eval_loss": 1.7271993160247803, | |
| "eval_runtime": 955.325, | |
| "eval_samples_per_second": 9.219, | |
| "eval_steps_per_second": 2.305, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.019315356664754254, | |
| "grad_norm": 4.395586013793945, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 7.5353, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.019506597819850834, | |
| "grad_norm": 4.341134071350098, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 6.827, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.01969783897494741, | |
| "grad_norm": 4.626712799072266, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 7.497, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.019889080130043986, | |
| "grad_norm": 4.3909077644348145, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 7.3894, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.020080321285140562, | |
| "grad_norm": 4.454318523406982, | |
| "learning_rate": 5e-05, | |
| "loss": 7.1685, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.02027156244023714, | |
| "grad_norm": 4.618101119995117, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 7.1355, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.020462803595333715, | |
| "grad_norm": 4.525501251220703, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 7.7377, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.020654044750430294, | |
| "grad_norm": 4.296500205993652, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 6.0578, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.02084528590552687, | |
| "grad_norm": 4.238190650939941, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 7.0223, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.021036527060623447, | |
| "grad_norm": 4.434668064117432, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 7.5787, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.021227768215720023, | |
| "grad_norm": 4.760210037231445, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 7.2316, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.0214190093708166, | |
| "grad_norm": 4.090402603149414, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 6.7225, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.021610250525913175, | |
| "grad_norm": 5.954319477081299, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 6.7379, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.021801491681009755, | |
| "grad_norm": 4.6670098304748535, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 7.3773, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.02199273283610633, | |
| "grad_norm": 4.392358303070068, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 6.8273, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.022183973991202907, | |
| "grad_norm": 4.784904479980469, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 7.2869, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.022375215146299483, | |
| "grad_norm": 4.397310733795166, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 6.4987, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.02256645630139606, | |
| "grad_norm": 4.161839962005615, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 7.2913, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.022757697456492636, | |
| "grad_norm": 4.547800064086914, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 7.0776, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.022948938611589215, | |
| "grad_norm": 4.405834197998047, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 7.1581, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.02314017976668579, | |
| "grad_norm": 4.7855072021484375, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 7.0802, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.023331420921782368, | |
| "grad_norm": 4.423474311828613, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 6.3145, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.023522662076878944, | |
| "grad_norm": 4.4300007820129395, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 7.1603, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.02371390323197552, | |
| "grad_norm": 13.551433563232422, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 6.8777, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.023905144387072096, | |
| "grad_norm": 4.480701446533203, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 6.8104, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.024096385542168676, | |
| "grad_norm": 4.077260971069336, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 6.4012, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.024287626697265252, | |
| "grad_norm": 4.8227458000183105, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 7.3057, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.02447886785236183, | |
| "grad_norm": 4.45031213760376, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 7.3441, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.024670109007458405, | |
| "grad_norm": 5.03026008605957, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 7.2416, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.02486135016255498, | |
| "grad_norm": 5.069452285766602, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 6.7119, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.025052591317651557, | |
| "grad_norm": 7.743057727813721, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 6.2863, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.025243832472748137, | |
| "grad_norm": 4.760660648345947, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 6.7764, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.025435073627844713, | |
| "grad_norm": 4.3669962882995605, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 6.3107, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.02562631478294129, | |
| "grad_norm": 4.6456499099731445, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 7.0249, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.025817555938037865, | |
| "grad_norm": 4.724009990692139, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 6.8711, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.02600879709313444, | |
| "grad_norm": 4.439990520477295, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 6.9839, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.02620003824823102, | |
| "grad_norm": 4.0352654457092285, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 6.2291, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.026391279403327597, | |
| "grad_norm": 4.637573719024658, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 6.8318, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.026582520558424173, | |
| "grad_norm": 4.376157283782959, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 6.0131, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.02677376171352075, | |
| "grad_norm": 4.183705806732178, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 7.2248, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.026965002868617326, | |
| "grad_norm": 3.9999494552612305, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 6.015, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.027156244023713902, | |
| "grad_norm": 4.598987102508545, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 6.9662, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.02734748517881048, | |
| "grad_norm": 3.976923704147339, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 5.7888, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.027538726333907058, | |
| "grad_norm": 4.858301639556885, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 7.0739, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.027729967489003634, | |
| "grad_norm": 5.00473690032959, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 7.0943, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.02792120864410021, | |
| "grad_norm": 4.407371997833252, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 6.2305, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.028112449799196786, | |
| "grad_norm": 4.396157264709473, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 6.5071, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.028303690954293362, | |
| "grad_norm": 4.96712589263916, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 5.6791, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.028494932109389942, | |
| "grad_norm": 4.8805623054504395, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 6.7981, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.02868617326448652, | |
| "grad_norm": 11.735734939575195, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 6.3148, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02868617326448652, | |
| "eval_loss": 1.694478154182434, | |
| "eval_runtime": 955.5702, | |
| "eval_samples_per_second": 9.216, | |
| "eval_steps_per_second": 2.304, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.028877414419583095, | |
| "grad_norm": 4.778436660766602, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 7.1983, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.02906865557467967, | |
| "grad_norm": 4.229771614074707, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 6.5959, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.029259896729776247, | |
| "grad_norm": 3.8028786182403564, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 6.3808, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.029451137884872823, | |
| "grad_norm": 4.6749958992004395, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 7.0537, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.029642379039969403, | |
| "grad_norm": 4.087666988372803, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 6.9317, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.02983362019506598, | |
| "grad_norm": 4.702816486358643, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 7.2036, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.030024861350162555, | |
| "grad_norm": 4.155583381652832, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 6.8133, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.03021610250525913, | |
| "grad_norm": 4.186269760131836, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 6.5739, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.030407343660355707, | |
| "grad_norm": 3.8579230308532715, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 6.7984, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.030598584815452284, | |
| "grad_norm": 4.169841766357422, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 6.3643, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.030789825970548863, | |
| "grad_norm": 4.905203342437744, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 7.0719, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.03098106712564544, | |
| "grad_norm": 4.120387554168701, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 6.7636, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.031172308280742016, | |
| "grad_norm": 8.308049201965332, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 6.7767, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.03136354943583859, | |
| "grad_norm": 4.300602912902832, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 7.0377, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.03155479059093517, | |
| "grad_norm": 4.625697135925293, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 8.0359, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.031746031746031744, | |
| "grad_norm": 5.824817180633545, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 6.7455, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.03193727290112832, | |
| "grad_norm": 3.6922667026519775, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 6.0934, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.0321285140562249, | |
| "grad_norm": 4.206804275512695, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 6.8294, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.03231975521132148, | |
| "grad_norm": 4.29739236831665, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 6.9077, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.032510996366418056, | |
| "grad_norm": 4.441958427429199, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 7.3585, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.03270223752151463, | |
| "grad_norm": 4.338846206665039, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 6.6852, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.03289347867661121, | |
| "grad_norm": 4.238076210021973, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 6.9762, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.033084719831707785, | |
| "grad_norm": 4.270300388336182, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 6.8412, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.03327596098680436, | |
| "grad_norm": 4.070036888122559, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 6.8448, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.03346720214190094, | |
| "grad_norm": 4.254011631011963, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 5.6192, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03365844329699751, | |
| "grad_norm": 4.03452205657959, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 7.0482, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.03384968445209409, | |
| "grad_norm": 4.573181629180908, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 6.6305, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.034040925607190665, | |
| "grad_norm": 5.104208469390869, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 6.9264, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.03423216676228724, | |
| "grad_norm": 5.245705604553223, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 7.1783, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.03442340791738382, | |
| "grad_norm": 4.223027229309082, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 6.6798, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0346146490724804, | |
| "grad_norm": 4.715947151184082, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 6.3754, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.03480589022757698, | |
| "grad_norm": 4.150231838226318, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 6.6499, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.03499713138267355, | |
| "grad_norm": 4.431682586669922, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 6.6471, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.03518837253777013, | |
| "grad_norm": 4.977612018585205, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 6.6979, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.035379613692866706, | |
| "grad_norm": 4.197941303253174, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 6.9976, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.03557085484796328, | |
| "grad_norm": 4.350857734680176, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 6.4757, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.03576209600305986, | |
| "grad_norm": 4.050764083862305, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 6.7683, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.035953337158156434, | |
| "grad_norm": 4.6614789962768555, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 6.851, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.03614457831325301, | |
| "grad_norm": 4.091433525085449, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 6.5398, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.03633581946834959, | |
| "grad_norm": 4.645418643951416, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 6.7257, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.03652706062344616, | |
| "grad_norm": 4.160548210144043, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 7.0893, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.03671830177854274, | |
| "grad_norm": 4.167484283447266, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 6.363, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.03690954293363932, | |
| "grad_norm": 5.07224178314209, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 6.8514, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.0371007840887359, | |
| "grad_norm": 4.517183780670166, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 6.4508, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.037292025243832475, | |
| "grad_norm": 4.487797737121582, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 6.9985, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.03748326639892905, | |
| "grad_norm": 4.516474723815918, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 6.0556, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.03767450755402563, | |
| "grad_norm": 5.170241832733154, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 5.8307, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.0378657487091222, | |
| "grad_norm": 5.109738349914551, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 6.4203, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.03805698986421878, | |
| "grad_norm": 5.513034343719482, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 6.1713, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.038248231019315355, | |
| "grad_norm": 5.409274578094482, | |
| "learning_rate": 0.0, | |
| "loss": 6.2335, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.038248231019315355, | |
| "eval_loss": 1.6850194931030273, | |
| "eval_runtime": 956.7925, | |
| "eval_samples_per_second": 9.205, | |
| "eval_steps_per_second": 2.301, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.64182013067264e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |