| { | |
| "best_metric": 1.7579431533813477, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.018896893823077833, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 9.448446911538915e-05, | |
| "grad_norm": 1.921506404876709, | |
| "learning_rate": 1e-05, | |
| "loss": 2.9126, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 9.448446911538915e-05, | |
| "eval_loss": 3.2697126865386963, | |
| "eval_runtime": 578.8616, | |
| "eval_samples_per_second": 30.795, | |
| "eval_steps_per_second": 7.7, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0001889689382307783, | |
| "grad_norm": 2.8076276779174805, | |
| "learning_rate": 2e-05, | |
| "loss": 3.4601, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00028345340734616746, | |
| "grad_norm": 4.111767292022705, | |
| "learning_rate": 3e-05, | |
| "loss": 3.5662, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0003779378764615566, | |
| "grad_norm": 2.4972636699676514, | |
| "learning_rate": 4e-05, | |
| "loss": 3.3327, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0004724223455769458, | |
| "grad_norm": 2.4476356506347656, | |
| "learning_rate": 5e-05, | |
| "loss": 2.8492, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0005669068146923349, | |
| "grad_norm": 2.121324300765991, | |
| "learning_rate": 6e-05, | |
| "loss": 2.6541, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0006613912838077241, | |
| "grad_norm": 2.0199811458587646, | |
| "learning_rate": 7e-05, | |
| "loss": 2.6148, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0007558757529231132, | |
| "grad_norm": 1.8478460311889648, | |
| "learning_rate": 8e-05, | |
| "loss": 2.4488, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0008503602220385024, | |
| "grad_norm": 1.7431505918502808, | |
| "learning_rate": 9e-05, | |
| "loss": 2.3382, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0009448446911538916, | |
| "grad_norm": 1.6332225799560547, | |
| "learning_rate": 0.0001, | |
| "loss": 2.2087, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0010393291602692808, | |
| "grad_norm": 1.600210428237915, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 2.1525, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0011338136293846698, | |
| "grad_norm": 2.117191791534424, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.1667, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.001228298098500059, | |
| "grad_norm": 2.2468860149383545, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 2.1325, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0013227825676154481, | |
| "grad_norm": 1.6275551319122314, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 2.0035, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0014172670367308374, | |
| "grad_norm": 1.6507940292358398, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.9576, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0015117515058462264, | |
| "grad_norm": 1.3009604215621948, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 2.0143, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0016062359749616157, | |
| "grad_norm": 1.3323287963867188, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.8313, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0017007204440770047, | |
| "grad_norm": 1.3134868144989014, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.9378, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.001795204913192394, | |
| "grad_norm": 1.1137737035751343, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.8115, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0018896893823077833, | |
| "grad_norm": 1.5699925422668457, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.8515, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0019841738514231725, | |
| "grad_norm": 1.3057512044906616, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.8708, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0020786583205385616, | |
| "grad_norm": 1.3271898031234741, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.9449, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0021731427896539506, | |
| "grad_norm": 1.266333818435669, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.9315, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0022676272587693396, | |
| "grad_norm": 1.2841119766235352, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.846, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.002362111727884729, | |
| "grad_norm": 1.462406039237976, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.8218, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002456596197000118, | |
| "grad_norm": 1.4343693256378174, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.9394, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.002551080666115507, | |
| "grad_norm": 1.4013473987579346, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.9241, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0026455651352308963, | |
| "grad_norm": 1.3553178310394287, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.7259, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0027400496043462857, | |
| "grad_norm": 1.6264681816101074, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.8219, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0028345340734616748, | |
| "grad_norm": 1.4833875894546509, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.801, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.002929018542577064, | |
| "grad_norm": 1.6012096405029297, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.8659, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.003023503011692453, | |
| "grad_norm": 1.4121304750442505, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.7772, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0031179874808078423, | |
| "grad_norm": 1.473338007926941, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.706, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0032124719499232314, | |
| "grad_norm": 1.415833830833435, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.6933, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0033069564190386204, | |
| "grad_norm": 1.5390123128890991, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.9392, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0034014408881540095, | |
| "grad_norm": 3.6614441871643066, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.7784, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.003495925357269399, | |
| "grad_norm": 1.6605850458145142, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.8682, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.003590409826384788, | |
| "grad_norm": 1.6443215608596802, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.798, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.003684894295500177, | |
| "grad_norm": 2.4218122959136963, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.9146, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0037793787646155665, | |
| "grad_norm": 1.6413733959197998, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.8938, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0038738632337309556, | |
| "grad_norm": 1.6265627145767212, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.8271, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.003968347702846345, | |
| "grad_norm": 2.1023473739624023, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.1164, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.004062832171961734, | |
| "grad_norm": 1.5792863368988037, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.8294, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004157316641077123, | |
| "grad_norm": 1.7397613525390625, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.9538, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.004251801110192512, | |
| "grad_norm": 1.7604749202728271, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.0928, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.004346285579307901, | |
| "grad_norm": 1.5783957242965698, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.9636, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.004440770048423291, | |
| "grad_norm": 1.6302754878997803, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.8991, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.004535254517538679, | |
| "grad_norm": 2.5601553916931152, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.8546, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.004629738986654069, | |
| "grad_norm": 1.7190425395965576, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.873, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.004724223455769458, | |
| "grad_norm": 1.8653565645217896, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.0215, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004724223455769458, | |
| "eval_loss": 1.9359604120254517, | |
| "eval_runtime": 580.6875, | |
| "eval_samples_per_second": 30.698, | |
| "eval_steps_per_second": 7.675, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004818707924884847, | |
| "grad_norm": 3.9928219318389893, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.2698, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.004913192394000236, | |
| "grad_norm": 2.3402740955352783, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.4355, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.005007676863115625, | |
| "grad_norm": 1.7540591955184937, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.273, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005102161332231014, | |
| "grad_norm": 2.258511543273926, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 2.4268, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.005196645801346404, | |
| "grad_norm": 1.5056934356689453, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.938, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0052911302704617925, | |
| "grad_norm": 1.229217290878296, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.7706, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.005385614739577182, | |
| "grad_norm": 1.0093824863433838, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.7142, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0054800992086925715, | |
| "grad_norm": 1.0942527055740356, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.7846, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.00557458367780796, | |
| "grad_norm": 0.9831579923629761, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.7108, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0056690681469233496, | |
| "grad_norm": 1.0122381448745728, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.8403, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.005763552616038739, | |
| "grad_norm": 0.9493637084960938, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.7662, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.005858037085154128, | |
| "grad_norm": 0.9440594911575317, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.6777, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.005952521554269517, | |
| "grad_norm": 0.9995548129081726, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.7533, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.006047006023384906, | |
| "grad_norm": 0.9642996191978455, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.7805, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.006141490492500295, | |
| "grad_norm": 1.0062496662139893, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.7519, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006235974961615685, | |
| "grad_norm": 1.0581656694412231, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.7327, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.006330459430731073, | |
| "grad_norm": 1.083849549293518, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.7668, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.006424943899846463, | |
| "grad_norm": 0.9820895195007324, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.6484, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.006519428368961852, | |
| "grad_norm": 1.1138657331466675, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.8989, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.006613912838077241, | |
| "grad_norm": 1.0786190032958984, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.6859, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.00670839730719263, | |
| "grad_norm": 1.0377622842788696, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.7283, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.006802881776308019, | |
| "grad_norm": 1.203137755393982, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.8213, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.006897366245423408, | |
| "grad_norm": 1.078721523284912, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.7945, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.006991850714538798, | |
| "grad_norm": 1.117542028427124, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.7302, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0070863351836541865, | |
| "grad_norm": 1.2751412391662598, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.8452, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.007180819652769576, | |
| "grad_norm": 1.122659683227539, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.7035, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0072753041218849655, | |
| "grad_norm": 1.195285677909851, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.7232, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.007369788591000354, | |
| "grad_norm": 1.2533471584320068, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.7606, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0074642730601157435, | |
| "grad_norm": 1.1980570554733276, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.7599, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.007558757529231133, | |
| "grad_norm": 1.2624444961547852, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.8498, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.007653241998346522, | |
| "grad_norm": 1.2369712591171265, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.8208, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.007747726467461911, | |
| "grad_norm": 1.256296992301941, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.7432, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0078422109365773, | |
| "grad_norm": 1.2641733884811401, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.6733, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.00793669540569269, | |
| "grad_norm": 1.5003657341003418, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.7468, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.008031179874808078, | |
| "grad_norm": 1.4441968202590942, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.9026, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.008125664343923467, | |
| "grad_norm": 1.3647269010543823, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.7824, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.008220148813038857, | |
| "grad_norm": 1.4593603610992432, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.7798, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.008314633282154246, | |
| "grad_norm": 1.6032061576843262, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.8624, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.008409117751269636, | |
| "grad_norm": 1.4728277921676636, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.7756, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.008503602220385023, | |
| "grad_norm": 1.5030272006988525, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.9159, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.008598086689500413, | |
| "grad_norm": 1.6059142351150513, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.836, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.008692571158615802, | |
| "grad_norm": 1.453993320465088, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.7456, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.008787055627731192, | |
| "grad_norm": 1.4265638589859009, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.8045, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.008881540096846581, | |
| "grad_norm": 1.5469270944595337, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.9141, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.00897602456596197, | |
| "grad_norm": 1.7251414060592651, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.7264, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.009070509035077359, | |
| "grad_norm": 1.5912647247314453, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.8451, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.009164993504192748, | |
| "grad_norm": 1.6220170259475708, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.9897, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.009259477973308138, | |
| "grad_norm": 1.5250394344329834, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.836, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.009353962442423527, | |
| "grad_norm": 1.9486126899719238, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.8885, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.009448446911538916, | |
| "grad_norm": 1.716341495513916, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.9319, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009448446911538916, | |
| "eval_loss": 1.7917085886001587, | |
| "eval_runtime": 580.1092, | |
| "eval_samples_per_second": 30.729, | |
| "eval_steps_per_second": 7.683, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009542931380654304, | |
| "grad_norm": 1.5820826292037964, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.0011, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.009637415849769694, | |
| "grad_norm": 1.247313380241394, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.9129, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.009731900318885083, | |
| "grad_norm": 1.6755188703536987, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.0137, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.009826384788000473, | |
| "grad_norm": 1.3869831562042236, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.027, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.009920869257115862, | |
| "grad_norm": 0.9147266745567322, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6587, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.01001535372623125, | |
| "grad_norm": 1.0676195621490479, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.7996, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.01010983819534664, | |
| "grad_norm": 0.9767804145812988, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.8536, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.010204322664462029, | |
| "grad_norm": 0.9702978134155273, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.6426, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.010298807133577418, | |
| "grad_norm": 1.1226290464401245, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.7416, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.010393291602692808, | |
| "grad_norm": 1.0002644062042236, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.6968, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.010487776071808197, | |
| "grad_norm": 0.9724743366241455, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.6787, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.010582260540923585, | |
| "grad_norm": 0.9423840045928955, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.7043, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.010676745010038974, | |
| "grad_norm": 0.9745253920555115, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.7031, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.010771229479154364, | |
| "grad_norm": 0.992032527923584, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.7182, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.010865713948269753, | |
| "grad_norm": 1.0905503034591675, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.674, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.010960198417385143, | |
| "grad_norm": 0.9268785715103149, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.7155, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.01105468288650053, | |
| "grad_norm": 1.0102099180221558, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.717, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.01114916735561592, | |
| "grad_norm": 0.9986180067062378, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.6517, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.01124365182473131, | |
| "grad_norm": 1.0582340955734253, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.7724, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.011338136293846699, | |
| "grad_norm": 1.1652514934539795, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.6412, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.011432620762962089, | |
| "grad_norm": 0.9556629657745361, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.5963, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.011527105232077478, | |
| "grad_norm": 1.1353280544281006, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.8117, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.011621589701192866, | |
| "grad_norm": 1.2140469551086426, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.7815, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.011716074170308255, | |
| "grad_norm": 1.1110254526138306, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.7719, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.011810558639423645, | |
| "grad_norm": 1.1787652969360352, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.6711, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.011905043108539034, | |
| "grad_norm": 1.1851552724838257, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.6315, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.011999527577654424, | |
| "grad_norm": 1.2488592863082886, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.6918, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.012094012046769811, | |
| "grad_norm": 1.19071364402771, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.6637, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.012188496515885201, | |
| "grad_norm": 1.27628493309021, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.7564, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.01228298098500059, | |
| "grad_norm": 1.4228508472442627, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.7784, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.01237746545411598, | |
| "grad_norm": 1.529374361038208, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.8852, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.01247194992323137, | |
| "grad_norm": 1.4993687868118286, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.9408, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.012566434392346759, | |
| "grad_norm": 1.3529025316238403, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.7517, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.012660918861462147, | |
| "grad_norm": 1.6172512769699097, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.889, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.012755403330577536, | |
| "grad_norm": 1.4073165655136108, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.6619, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.012849887799692926, | |
| "grad_norm": 1.3579713106155396, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.7318, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.012944372268808315, | |
| "grad_norm": 1.3095992803573608, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.7126, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.013038856737923704, | |
| "grad_norm": 1.3842580318450928, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.6722, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.013133341207039092, | |
| "grad_norm": 1.641334891319275, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.7873, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.013227825676154482, | |
| "grad_norm": 1.4023642539978027, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.8751, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.013322310145269871, | |
| "grad_norm": 1.4509235620498657, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.8146, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.01341679461438526, | |
| "grad_norm": 1.5509487390518188, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.6868, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.01351127908350065, | |
| "grad_norm": 1.4508670568466187, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.636, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.013605763552616038, | |
| "grad_norm": 1.5073550939559937, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.8278, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.013700248021731427, | |
| "grad_norm": 1.4817103147506714, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.8222, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.013794732490846817, | |
| "grad_norm": 1.3904190063476562, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.6646, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.013889216959962206, | |
| "grad_norm": 1.4661694765090942, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.8151, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.013983701429077596, | |
| "grad_norm": 1.4789057970046997, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.7675, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.014078185898192985, | |
| "grad_norm": 1.4990551471710205, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.6692, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.014172670367308373, | |
| "grad_norm": 2.1998047828674316, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.916, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.014172670367308373, | |
| "eval_loss": 1.7659413814544678, | |
| "eval_runtime": 581.5762, | |
| "eval_samples_per_second": 30.651, | |
| "eval_steps_per_second": 7.664, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.014267154836423762, | |
| "grad_norm": 1.1709290742874146, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.9397, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.014361639305539152, | |
| "grad_norm": 1.3178352117538452, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 2.1818, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.014456123774654541, | |
| "grad_norm": 2.0690956115722656, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.2944, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.014550608243769931, | |
| "grad_norm": 1.2781531810760498, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 2.0201, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.014645092712885319, | |
| "grad_norm": 0.9701533913612366, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.7258, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.014739577182000708, | |
| "grad_norm": 0.8803207874298096, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.5666, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.014834061651116098, | |
| "grad_norm": 0.9210809469223022, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.7078, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.014928546120231487, | |
| "grad_norm": 0.993040919303894, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.7638, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.015023030589346877, | |
| "grad_norm": 0.9425827860832214, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.6668, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.015117515058462266, | |
| "grad_norm": 1.2820783853530884, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.6792, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.015211999527577654, | |
| "grad_norm": 1.0371493101119995, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.7741, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.015306483996693043, | |
| "grad_norm": 0.9637015461921692, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.6913, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.015400968465808433, | |
| "grad_norm": 0.9290022850036621, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.6807, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.015495452934923822, | |
| "grad_norm": 0.9957162737846375, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.712, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.015589937404039212, | |
| "grad_norm": 0.9981750845909119, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.7648, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0156844218731546, | |
| "grad_norm": 1.0447102785110474, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.7172, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.01577890634226999, | |
| "grad_norm": 1.0451395511627197, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.6982, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.01587339081138538, | |
| "grad_norm": 0.990496814250946, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.655, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.015967875280500768, | |
| "grad_norm": 1.0287784337997437, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.6693, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.016062359749616156, | |
| "grad_norm": 1.2052768468856812, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.672, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.016156844218731547, | |
| "grad_norm": 1.0643489360809326, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.7512, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.016251328687846935, | |
| "grad_norm": 1.0472650527954102, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.641, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.016345813156962326, | |
| "grad_norm": 1.1761748790740967, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.6721, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.016440297626077714, | |
| "grad_norm": 1.148133397102356, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.6983, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0165347820951931, | |
| "grad_norm": 1.1744489669799805, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.6267, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.016629266564308492, | |
| "grad_norm": 1.0933793783187866, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.6686, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.01672375103342388, | |
| "grad_norm": 1.1083273887634277, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.7158, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.01681823550253927, | |
| "grad_norm": 1.1390151977539062, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.8014, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.01691271997165466, | |
| "grad_norm": 1.3026089668273926, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.6812, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.017007204440770047, | |
| "grad_norm": 1.312772512435913, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.5583, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.017101688909885438, | |
| "grad_norm": 1.2901217937469482, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.7467, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.017196173379000826, | |
| "grad_norm": 1.3850208520889282, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.7443, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.017290657848116217, | |
| "grad_norm": 1.2660682201385498, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.756, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.017385142317231605, | |
| "grad_norm": 1.3568905591964722, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.8139, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.017479626786346993, | |
| "grad_norm": 1.3574552536010742, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.7441, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.017574111255462384, | |
| "grad_norm": 1.4080511331558228, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.8236, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.01766859572457777, | |
| "grad_norm": 1.3394280672073364, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.6681, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.017763080193693163, | |
| "grad_norm": 1.319677472114563, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.619, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01785756466280855, | |
| "grad_norm": 1.3819472789764404, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.6346, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.01795204913192394, | |
| "grad_norm": 1.5803827047348022, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.802, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01804653360103933, | |
| "grad_norm": 1.6066573858261108, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.9386, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.018141018070154717, | |
| "grad_norm": 1.4194458723068237, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.7029, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.01823550253927011, | |
| "grad_norm": 1.3356919288635254, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.6627, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.018329987008385496, | |
| "grad_norm": 1.7929203510284424, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.7651, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.018424471477500887, | |
| "grad_norm": 1.548923373222351, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.7475, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.018518955946616275, | |
| "grad_norm": 1.639174461364746, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.7274, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.018613440415731663, | |
| "grad_norm": 1.5783941745758057, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.7772, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.018707924884847054, | |
| "grad_norm": 1.5097144842147827, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 1.7214, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.018802409353962442, | |
| "grad_norm": 1.5874967575073242, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.7261, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.018896893823077833, | |
| "grad_norm": 1.8029627799987793, | |
| "learning_rate": 0.0, | |
| "loss": 2.0528, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.018896893823077833, | |
| "eval_loss": 1.7579431533813477, | |
| "eval_runtime": 581.9342, | |
| "eval_samples_per_second": 30.632, | |
| "eval_steps_per_second": 7.659, | |
| "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": 5.966773220278272e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |