| { | |
| "best_metric": 1.1956850290298462, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.09519276534983341, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00047596382674916705, | |
| "grad_norm": 18.821937561035156, | |
| "learning_rate": 1.0170000000000001e-05, | |
| "loss": 3.7617, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00047596382674916705, | |
| "eval_loss": 2.0851309299468994, | |
| "eval_runtime": 94.1092, | |
| "eval_samples_per_second": 9.404, | |
| "eval_steps_per_second": 2.359, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0009519276534983341, | |
| "grad_norm": 20.83098602294922, | |
| "learning_rate": 2.0340000000000002e-05, | |
| "loss": 3.8855, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0014278914802475012, | |
| "grad_norm": 18.45833969116211, | |
| "learning_rate": 3.051e-05, | |
| "loss": 4.0341, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0019038553069966682, | |
| "grad_norm": 15.618721961975098, | |
| "learning_rate": 4.0680000000000004e-05, | |
| "loss": 3.4497, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.002379819133745835, | |
| "grad_norm": 14.598923683166504, | |
| "learning_rate": 5.085e-05, | |
| "loss": 3.2374, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0028557829604950024, | |
| "grad_norm": 16.081514358520508, | |
| "learning_rate": 6.102e-05, | |
| "loss": 3.4533, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0033317467872441696, | |
| "grad_norm": 14.391841888427734, | |
| "learning_rate": 7.119e-05, | |
| "loss": 3.2119, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0038077106139933364, | |
| "grad_norm": 13.710631370544434, | |
| "learning_rate": 8.136000000000001e-05, | |
| "loss": 3.3193, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.004283674440742504, | |
| "grad_norm": 12.184770584106445, | |
| "learning_rate": 9.153000000000001e-05, | |
| "loss": 3.0283, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00475963826749167, | |
| "grad_norm": 11.110732078552246, | |
| "learning_rate": 0.0001017, | |
| "loss": 3.0447, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.005235602094240838, | |
| "grad_norm": 13.03386402130127, | |
| "learning_rate": 0.00010116473684210527, | |
| "loss": 3.0503, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.005711565920990005, | |
| "grad_norm": 11.393197059631348, | |
| "learning_rate": 0.00010062947368421052, | |
| "loss": 3.1731, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.006187529747739172, | |
| "grad_norm": 11.392594337463379, | |
| "learning_rate": 0.00010009421052631579, | |
| "loss": 2.7808, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.006663493574488339, | |
| "grad_norm": 11.060273170471191, | |
| "learning_rate": 9.955894736842107e-05, | |
| "loss": 2.9735, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.007139457401237506, | |
| "grad_norm": 10.616848945617676, | |
| "learning_rate": 9.902368421052632e-05, | |
| "loss": 2.8027, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.007615421227986673, | |
| "grad_norm": 11.916224479675293, | |
| "learning_rate": 9.848842105263159e-05, | |
| "loss": 3.3891, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.00809138505473584, | |
| "grad_norm": 11.447843551635742, | |
| "learning_rate": 9.795315789473685e-05, | |
| "loss": 3.1294, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.008567348881485007, | |
| "grad_norm": 9.17795467376709, | |
| "learning_rate": 9.74178947368421e-05, | |
| "loss": 2.7995, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.009043312708234174, | |
| "grad_norm": 10.779789924621582, | |
| "learning_rate": 9.688263157894737e-05, | |
| "loss": 3.2716, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.00951927653498334, | |
| "grad_norm": 10.012481689453125, | |
| "learning_rate": 9.634736842105264e-05, | |
| "loss": 2.9491, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.009995240361732508, | |
| "grad_norm": 10.166115760803223, | |
| "learning_rate": 9.58121052631579e-05, | |
| "loss": 3.024, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.010471204188481676, | |
| "grad_norm": 9.625226020812988, | |
| "learning_rate": 9.527684210526317e-05, | |
| "loss": 2.9964, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.010947168015230843, | |
| "grad_norm": 11.936861991882324, | |
| "learning_rate": 9.474157894736843e-05, | |
| "loss": 3.0226, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.01142313184198001, | |
| "grad_norm": 9.899079322814941, | |
| "learning_rate": 9.420631578947368e-05, | |
| "loss": 3.0418, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.011899095668729176, | |
| "grad_norm": 9.566582679748535, | |
| "learning_rate": 9.367105263157895e-05, | |
| "loss": 2.5971, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.012375059495478343, | |
| "grad_norm": 9.53864860534668, | |
| "learning_rate": 9.313578947368422e-05, | |
| "loss": 3.0019, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01285102332222751, | |
| "grad_norm": 13.411975860595703, | |
| "learning_rate": 9.260052631578948e-05, | |
| "loss": 2.7033, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.013326987148976678, | |
| "grad_norm": 26.838424682617188, | |
| "learning_rate": 9.206526315789475e-05, | |
| "loss": 2.5882, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.013802950975725845, | |
| "grad_norm": 11.076753616333008, | |
| "learning_rate": 9.153000000000001e-05, | |
| "loss": 2.9107, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.014278914802475012, | |
| "grad_norm": 10.66457748413086, | |
| "learning_rate": 9.099473684210527e-05, | |
| "loss": 3.1288, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.014754878629224179, | |
| "grad_norm": 11.16288948059082, | |
| "learning_rate": 9.045947368421053e-05, | |
| "loss": 2.913, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.015230842455973346, | |
| "grad_norm": 9.841339111328125, | |
| "learning_rate": 8.99242105263158e-05, | |
| "loss": 2.8651, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.015706806282722512, | |
| "grad_norm": 9.972783088684082, | |
| "learning_rate": 8.938894736842105e-05, | |
| "loss": 2.6382, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.01618277010947168, | |
| "grad_norm": 8.628253936767578, | |
| "learning_rate": 8.885368421052633e-05, | |
| "loss": 2.8631, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.016658733936220846, | |
| "grad_norm": 9.764812469482422, | |
| "learning_rate": 8.83184210526316e-05, | |
| "loss": 3.1602, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.017134697762970014, | |
| "grad_norm": 9.357998847961426, | |
| "learning_rate": 8.778315789473685e-05, | |
| "loss": 3.232, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.017610661589719183, | |
| "grad_norm": 8.898050308227539, | |
| "learning_rate": 8.724789473684211e-05, | |
| "loss": 3.064, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.018086625416468348, | |
| "grad_norm": 8.951642990112305, | |
| "learning_rate": 8.671263157894738e-05, | |
| "loss": 2.6545, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.018562589243217516, | |
| "grad_norm": 8.89450454711914, | |
| "learning_rate": 8.617736842105263e-05, | |
| "loss": 2.8535, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.01903855306996668, | |
| "grad_norm": 8.89578628540039, | |
| "learning_rate": 8.56421052631579e-05, | |
| "loss": 2.775, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01951451689671585, | |
| "grad_norm": 11.88039493560791, | |
| "learning_rate": 8.510684210526316e-05, | |
| "loss": 2.9885, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.019990480723465015, | |
| "grad_norm": 9.176898956298828, | |
| "learning_rate": 8.457157894736843e-05, | |
| "loss": 2.7476, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.020466444550214184, | |
| "grad_norm": 10.42405891418457, | |
| "learning_rate": 8.403631578947369e-05, | |
| "loss": 3.1671, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.020942408376963352, | |
| "grad_norm": 9.580604553222656, | |
| "learning_rate": 8.350105263157896e-05, | |
| "loss": 2.7088, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.021418372203712517, | |
| "grad_norm": 11.169540405273438, | |
| "learning_rate": 8.296578947368421e-05, | |
| "loss": 3.0075, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.021894336030461686, | |
| "grad_norm": 8.268192291259766, | |
| "learning_rate": 8.243052631578948e-05, | |
| "loss": 2.5109, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02237029985721085, | |
| "grad_norm": 9.292203903198242, | |
| "learning_rate": 8.189526315789474e-05, | |
| "loss": 2.9182, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02284626368396002, | |
| "grad_norm": 10.59662914276123, | |
| "learning_rate": 8.136000000000001e-05, | |
| "loss": 3.2872, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.023322227510709188, | |
| "grad_norm": 10.883740425109863, | |
| "learning_rate": 8.082473684210527e-05, | |
| "loss": 3.0321, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.023798191337458353, | |
| "grad_norm": 8.746244430541992, | |
| "learning_rate": 8.028947368421054e-05, | |
| "loss": 2.9888, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.023798191337458353, | |
| "eval_loss": 1.4506478309631348, | |
| "eval_runtime": 94.223, | |
| "eval_samples_per_second": 9.393, | |
| "eval_steps_per_second": 2.356, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02427415516420752, | |
| "grad_norm": 8.132503509521484, | |
| "learning_rate": 7.975421052631579e-05, | |
| "loss": 2.8369, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.024750118990956686, | |
| "grad_norm": 7.467820644378662, | |
| "learning_rate": 7.921894736842106e-05, | |
| "loss": 3.1423, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.025226082817705855, | |
| "grad_norm": 6.4784369468688965, | |
| "learning_rate": 7.868368421052632e-05, | |
| "loss": 2.7428, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.02570204664445502, | |
| "grad_norm": 6.803327560424805, | |
| "learning_rate": 7.814842105263157e-05, | |
| "loss": 2.8238, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.02617801047120419, | |
| "grad_norm": 9.669547080993652, | |
| "learning_rate": 7.761315789473685e-05, | |
| "loss": 2.7933, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.026653974297953357, | |
| "grad_norm": 9.685455322265625, | |
| "learning_rate": 7.70778947368421e-05, | |
| "loss": 3.0942, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.027129938124702522, | |
| "grad_norm": 8.498844146728516, | |
| "learning_rate": 7.654263157894737e-05, | |
| "loss": 3.0961, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.02760590195145169, | |
| "grad_norm": 7.0263776779174805, | |
| "learning_rate": 7.600736842105264e-05, | |
| "loss": 2.9287, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.028081865778200855, | |
| "grad_norm": 7.7478742599487305, | |
| "learning_rate": 7.54721052631579e-05, | |
| "loss": 2.7414, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.028557829604950024, | |
| "grad_norm": 6.691257953643799, | |
| "learning_rate": 7.493684210526315e-05, | |
| "loss": 2.6123, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.029033793431699192, | |
| "grad_norm": 7.181465148925781, | |
| "learning_rate": 7.440157894736843e-05, | |
| "loss": 2.6851, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.029509757258448358, | |
| "grad_norm": 6.531435012817383, | |
| "learning_rate": 7.386631578947369e-05, | |
| "loss": 2.6416, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.029985721085197526, | |
| "grad_norm": 9.599217414855957, | |
| "learning_rate": 7.333105263157895e-05, | |
| "loss": 2.7074, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.03046168491194669, | |
| "grad_norm": 7.4359846115112305, | |
| "learning_rate": 7.279578947368422e-05, | |
| "loss": 2.8352, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03093764873869586, | |
| "grad_norm": 6.91318941116333, | |
| "learning_rate": 7.226052631578947e-05, | |
| "loss": 2.5786, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.031413612565445025, | |
| "grad_norm": 8.484053611755371, | |
| "learning_rate": 7.172526315789474e-05, | |
| "loss": 3.0093, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0318895763921942, | |
| "grad_norm": 7.760731220245361, | |
| "learning_rate": 7.119e-05, | |
| "loss": 2.8831, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03236554021894336, | |
| "grad_norm": 7.7634100914001465, | |
| "learning_rate": 7.065473684210527e-05, | |
| "loss": 2.8134, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.03284150404569253, | |
| "grad_norm": 15.038714408874512, | |
| "learning_rate": 7.011947368421053e-05, | |
| "loss": 2.446, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03331746787244169, | |
| "grad_norm": 6.84593391418457, | |
| "learning_rate": 6.95842105263158e-05, | |
| "loss": 2.8314, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.033793431699190864, | |
| "grad_norm": 6.8339667320251465, | |
| "learning_rate": 6.904894736842105e-05, | |
| "loss": 2.507, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03426939552594003, | |
| "grad_norm": 8.750052452087402, | |
| "learning_rate": 6.851368421052632e-05, | |
| "loss": 2.637, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.034745359352689194, | |
| "grad_norm": 7.736267566680908, | |
| "learning_rate": 6.797842105263158e-05, | |
| "loss": 2.748, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.035221323179438366, | |
| "grad_norm": 7.89774227142334, | |
| "learning_rate": 6.744315789473685e-05, | |
| "loss": 2.7948, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03569728700618753, | |
| "grad_norm": 7.224119663238525, | |
| "learning_rate": 6.690789473684211e-05, | |
| "loss": 2.7275, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.036173250832936696, | |
| "grad_norm": 7.180510520935059, | |
| "learning_rate": 6.637263157894738e-05, | |
| "loss": 2.5776, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.03664921465968586, | |
| "grad_norm": 6.440933704376221, | |
| "learning_rate": 6.583736842105263e-05, | |
| "loss": 2.241, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.03712517848643503, | |
| "grad_norm": 8.913047790527344, | |
| "learning_rate": 6.53021052631579e-05, | |
| "loss": 2.9477, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0376011423131842, | |
| "grad_norm": 7.736593723297119, | |
| "learning_rate": 6.476684210526316e-05, | |
| "loss": 2.8696, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03807710613993336, | |
| "grad_norm": 7.847418308258057, | |
| "learning_rate": 6.423157894736841e-05, | |
| "loss": 2.3813, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.038553069966682535, | |
| "grad_norm": 9.366430282592773, | |
| "learning_rate": 6.369631578947368e-05, | |
| "loss": 2.8983, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0390290337934317, | |
| "grad_norm": 6.681727886199951, | |
| "learning_rate": 6.316105263157896e-05, | |
| "loss": 2.4428, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.039504997620180865, | |
| "grad_norm": 8.800130844116211, | |
| "learning_rate": 6.262578947368421e-05, | |
| "loss": 2.9551, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.03998096144693003, | |
| "grad_norm": 8.04470157623291, | |
| "learning_rate": 6.209052631578948e-05, | |
| "loss": 2.7357, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0404569252736792, | |
| "grad_norm": 6.852024555206299, | |
| "learning_rate": 6.155526315789474e-05, | |
| "loss": 2.6536, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.04093288910042837, | |
| "grad_norm": 7.370736598968506, | |
| "learning_rate": 6.102e-05, | |
| "loss": 2.6235, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.04140885292717753, | |
| "grad_norm": 7.6142401695251465, | |
| "learning_rate": 6.048473684210526e-05, | |
| "loss": 2.7034, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.041884816753926704, | |
| "grad_norm": 7.17495059967041, | |
| "learning_rate": 5.9949473684210527e-05, | |
| "loss": 2.7207, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.04236078058067587, | |
| "grad_norm": 7.216758728027344, | |
| "learning_rate": 5.94142105263158e-05, | |
| "loss": 2.4448, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.042836744407425034, | |
| "grad_norm": 7.9468817710876465, | |
| "learning_rate": 5.887894736842106e-05, | |
| "loss": 2.9818, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.043312708234174206, | |
| "grad_norm": 8.539376258850098, | |
| "learning_rate": 5.834368421052632e-05, | |
| "loss": 2.4128, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.04378867206092337, | |
| "grad_norm": 8.345818519592285, | |
| "learning_rate": 5.780842105263158e-05, | |
| "loss": 2.7606, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.044264635887672536, | |
| "grad_norm": 8.508038520812988, | |
| "learning_rate": 5.727315789473684e-05, | |
| "loss": 2.6035, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0447405997144217, | |
| "grad_norm": 8.71206283569336, | |
| "learning_rate": 5.673789473684211e-05, | |
| "loss": 2.9024, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.04521656354117087, | |
| "grad_norm": 7.256693363189697, | |
| "learning_rate": 5.620263157894738e-05, | |
| "loss": 2.5654, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.04569252736792004, | |
| "grad_norm": 6.628811359405518, | |
| "learning_rate": 5.566736842105264e-05, | |
| "loss": 2.0765, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0461684911946692, | |
| "grad_norm": 9.192995071411133, | |
| "learning_rate": 5.51321052631579e-05, | |
| "loss": 3.0441, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.046644455021418375, | |
| "grad_norm": 9.181817054748535, | |
| "learning_rate": 5.459684210526316e-05, | |
| "loss": 2.8811, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.04712041884816754, | |
| "grad_norm": 9.44265079498291, | |
| "learning_rate": 5.406157894736842e-05, | |
| "loss": 2.6583, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.047596382674916705, | |
| "grad_norm": 9.560362815856934, | |
| "learning_rate": 5.352631578947368e-05, | |
| "loss": 2.5805, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.047596382674916705, | |
| "eval_loss": 1.34207284450531, | |
| "eval_runtime": 94.197, | |
| "eval_samples_per_second": 9.395, | |
| "eval_steps_per_second": 2.357, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04807234650166587, | |
| "grad_norm": 7.516977787017822, | |
| "learning_rate": 5.299105263157895e-05, | |
| "loss": 2.5181, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.04854831032841504, | |
| "grad_norm": 7.009253025054932, | |
| "learning_rate": 5.245578947368422e-05, | |
| "loss": 2.6719, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.04902427415516421, | |
| "grad_norm": 6.846374988555908, | |
| "learning_rate": 5.192052631578948e-05, | |
| "loss": 2.5637, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.04950023798191337, | |
| "grad_norm": 5.941534519195557, | |
| "learning_rate": 5.1385263157894744e-05, | |
| "loss": 2.2845, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.049976201808662545, | |
| "grad_norm": 7.417452812194824, | |
| "learning_rate": 5.085e-05, | |
| "loss": 3.1752, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.05045216563541171, | |
| "grad_norm": 6.396870136260986, | |
| "learning_rate": 5.031473684210526e-05, | |
| "loss": 2.5322, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.050928129462160875, | |
| "grad_norm": 6.419276237487793, | |
| "learning_rate": 4.9779473684210534e-05, | |
| "loss": 2.5846, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.05140409328891004, | |
| "grad_norm": 9.035964012145996, | |
| "learning_rate": 4.924421052631579e-05, | |
| "loss": 2.8623, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.05188005711565921, | |
| "grad_norm": 6.224599361419678, | |
| "learning_rate": 4.870894736842105e-05, | |
| "loss": 2.3845, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.05235602094240838, | |
| "grad_norm": 7.3065409660339355, | |
| "learning_rate": 4.817368421052632e-05, | |
| "loss": 2.7384, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.05283198476915754, | |
| "grad_norm": 6.494760036468506, | |
| "learning_rate": 4.763842105263158e-05, | |
| "loss": 2.3574, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.053307948595906714, | |
| "grad_norm": 5.581643104553223, | |
| "learning_rate": 4.710315789473684e-05, | |
| "loss": 2.2699, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.05378391242265588, | |
| "grad_norm": 6.040269374847412, | |
| "learning_rate": 4.656789473684211e-05, | |
| "loss": 2.5859, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.054259876249405044, | |
| "grad_norm": 6.788252353668213, | |
| "learning_rate": 4.6032631578947374e-05, | |
| "loss": 2.4009, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.05473584007615421, | |
| "grad_norm": 7.658879280090332, | |
| "learning_rate": 4.549736842105263e-05, | |
| "loss": 2.0763, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.05521180390290338, | |
| "grad_norm": 9.23805046081543, | |
| "learning_rate": 4.49621052631579e-05, | |
| "loss": 2.2703, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.055687767729652546, | |
| "grad_norm": 7.436956882476807, | |
| "learning_rate": 4.4426842105263164e-05, | |
| "loss": 2.7847, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.05616373155640171, | |
| "grad_norm": 8.176994323730469, | |
| "learning_rate": 4.389157894736842e-05, | |
| "loss": 2.5287, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.05663969538315088, | |
| "grad_norm": 7.547792434692383, | |
| "learning_rate": 4.335631578947369e-05, | |
| "loss": 2.81, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.05711565920990005, | |
| "grad_norm": 7.150010108947754, | |
| "learning_rate": 4.282105263157895e-05, | |
| "loss": 2.7525, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05759162303664921, | |
| "grad_norm": 7.117188930511475, | |
| "learning_rate": 4.228578947368421e-05, | |
| "loss": 2.5581, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.058067586863398385, | |
| "grad_norm": 6.114779472351074, | |
| "learning_rate": 4.175052631578948e-05, | |
| "loss": 2.5203, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.05854355069014755, | |
| "grad_norm": 6.867531776428223, | |
| "learning_rate": 4.121526315789474e-05, | |
| "loss": 2.6818, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.059019514516896715, | |
| "grad_norm": 7.646029949188232, | |
| "learning_rate": 4.0680000000000004e-05, | |
| "loss": 2.6518, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.05949547834364588, | |
| "grad_norm": 8.080942153930664, | |
| "learning_rate": 4.014473684210527e-05, | |
| "loss": 2.8059, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05997144217039505, | |
| "grad_norm": 6.926398277282715, | |
| "learning_rate": 3.960947368421053e-05, | |
| "loss": 2.5314, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.06044740599714422, | |
| "grad_norm": 7.169958114624023, | |
| "learning_rate": 3.907421052631579e-05, | |
| "loss": 2.5107, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.06092336982389338, | |
| "grad_norm": 7.661369323730469, | |
| "learning_rate": 3.853894736842105e-05, | |
| "loss": 2.3918, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.061399333650642554, | |
| "grad_norm": 7.439793586730957, | |
| "learning_rate": 3.800368421052632e-05, | |
| "loss": 2.8541, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.06187529747739172, | |
| "grad_norm": 6.659045696258545, | |
| "learning_rate": 3.746842105263158e-05, | |
| "loss": 2.5837, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.062351261304140884, | |
| "grad_norm": 8.68233585357666, | |
| "learning_rate": 3.693315789473684e-05, | |
| "loss": 2.5465, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.06282722513089005, | |
| "grad_norm": 7.132856845855713, | |
| "learning_rate": 3.639789473684211e-05, | |
| "loss": 2.4912, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.06330318895763921, | |
| "grad_norm": 7.528649806976318, | |
| "learning_rate": 3.586263157894737e-05, | |
| "loss": 2.7235, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.0637791527843884, | |
| "grad_norm": 6.1327667236328125, | |
| "learning_rate": 3.5327368421052633e-05, | |
| "loss": 2.4542, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.06425511661113756, | |
| "grad_norm": 6.09030294418335, | |
| "learning_rate": 3.47921052631579e-05, | |
| "loss": 2.3305, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.06473108043788672, | |
| "grad_norm": 6.725249290466309, | |
| "learning_rate": 3.425684210526316e-05, | |
| "loss": 2.4557, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.06520704426463589, | |
| "grad_norm": 6.322257041931152, | |
| "learning_rate": 3.3721578947368424e-05, | |
| "loss": 2.2034, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.06568300809138505, | |
| "grad_norm": 5.7170305252075195, | |
| "learning_rate": 3.318631578947369e-05, | |
| "loss": 2.1993, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.06615897191813422, | |
| "grad_norm": 7.491881370544434, | |
| "learning_rate": 3.265105263157895e-05, | |
| "loss": 2.6061, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.06663493574488338, | |
| "grad_norm": 7.218158721923828, | |
| "learning_rate": 3.211578947368421e-05, | |
| "loss": 2.622, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.06711089957163256, | |
| "grad_norm": 10.26785659790039, | |
| "learning_rate": 3.158052631578948e-05, | |
| "loss": 2.6993, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.06758686339838173, | |
| "grad_norm": 7.6761393547058105, | |
| "learning_rate": 3.104526315789474e-05, | |
| "loss": 2.6623, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.06806282722513089, | |
| "grad_norm": 7.0337395668029785, | |
| "learning_rate": 3.051e-05, | |
| "loss": 2.632, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.06853879105188006, | |
| "grad_norm": 7.8988189697265625, | |
| "learning_rate": 2.9974736842105263e-05, | |
| "loss": 3.1195, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.06901475487862922, | |
| "grad_norm": 8.312713623046875, | |
| "learning_rate": 2.943947368421053e-05, | |
| "loss": 2.5157, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.06949071870537839, | |
| "grad_norm": 7.518918514251709, | |
| "learning_rate": 2.890421052631579e-05, | |
| "loss": 2.7354, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.06996668253212755, | |
| "grad_norm": 7.5307841300964355, | |
| "learning_rate": 2.8368947368421054e-05, | |
| "loss": 2.5474, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.07044264635887673, | |
| "grad_norm": 8.257736206054688, | |
| "learning_rate": 2.783368421052632e-05, | |
| "loss": 2.8553, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.0709186101856259, | |
| "grad_norm": 6.292891979217529, | |
| "learning_rate": 2.729842105263158e-05, | |
| "loss": 2.3596, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.07139457401237506, | |
| "grad_norm": 9.312799453735352, | |
| "learning_rate": 2.676315789473684e-05, | |
| "loss": 2.5738, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07139457401237506, | |
| "eval_loss": 1.2638894319534302, | |
| "eval_runtime": 94.1331, | |
| "eval_samples_per_second": 9.402, | |
| "eval_steps_per_second": 2.358, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07187053783912423, | |
| "grad_norm": 5.808437824249268, | |
| "learning_rate": 2.622789473684211e-05, | |
| "loss": 2.4866, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.07234650166587339, | |
| "grad_norm": 6.367058753967285, | |
| "learning_rate": 2.5692631578947372e-05, | |
| "loss": 2.5468, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.07282246549262256, | |
| "grad_norm": 7.049677848815918, | |
| "learning_rate": 2.515736842105263e-05, | |
| "loss": 2.6916, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.07329842931937172, | |
| "grad_norm": 6.486673831939697, | |
| "learning_rate": 2.4622105263157897e-05, | |
| "loss": 2.5986, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.0737743931461209, | |
| "grad_norm": 6.483848571777344, | |
| "learning_rate": 2.408684210526316e-05, | |
| "loss": 2.6228, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.07425035697287007, | |
| "grad_norm": 8.12568473815918, | |
| "learning_rate": 2.355157894736842e-05, | |
| "loss": 3.0707, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.07472632079961923, | |
| "grad_norm": 6.405067443847656, | |
| "learning_rate": 2.3016315789473687e-05, | |
| "loss": 2.654, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.0752022846263684, | |
| "grad_norm": 5.833227634429932, | |
| "learning_rate": 2.248105263157895e-05, | |
| "loss": 2.4285, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.07567824845311756, | |
| "grad_norm": 5.303393840789795, | |
| "learning_rate": 2.194578947368421e-05, | |
| "loss": 2.2755, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.07615421227986673, | |
| "grad_norm": 6.173555850982666, | |
| "learning_rate": 2.1410526315789474e-05, | |
| "loss": 2.4719, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.07663017610661589, | |
| "grad_norm": 5.758234977722168, | |
| "learning_rate": 2.087526315789474e-05, | |
| "loss": 2.6648, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.07710613993336507, | |
| "grad_norm": 5.576732158660889, | |
| "learning_rate": 2.0340000000000002e-05, | |
| "loss": 2.5195, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.07758210376011423, | |
| "grad_norm": 6.40316104888916, | |
| "learning_rate": 1.9804736842105264e-05, | |
| "loss": 2.6616, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.0780580675868634, | |
| "grad_norm": 5.3443427085876465, | |
| "learning_rate": 1.9269473684210526e-05, | |
| "loss": 2.1508, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.07853403141361257, | |
| "grad_norm": 7.0513014793396, | |
| "learning_rate": 1.873421052631579e-05, | |
| "loss": 2.3545, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.07900999524036173, | |
| "grad_norm": 6.783944129943848, | |
| "learning_rate": 1.8198947368421054e-05, | |
| "loss": 2.361, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.0794859590671109, | |
| "grad_norm": 7.000679016113281, | |
| "learning_rate": 1.7663684210526317e-05, | |
| "loss": 2.6073, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.07996192289386006, | |
| "grad_norm": 5.986188888549805, | |
| "learning_rate": 1.712842105263158e-05, | |
| "loss": 2.3954, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.08043788672060924, | |
| "grad_norm": 6.529272556304932, | |
| "learning_rate": 1.6593157894736845e-05, | |
| "loss": 2.7285, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.0809138505473584, | |
| "grad_norm": 5.4884772300720215, | |
| "learning_rate": 1.6057894736842104e-05, | |
| "loss": 2.1842, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.08138981437410757, | |
| "grad_norm": 6.433114528656006, | |
| "learning_rate": 1.552263157894737e-05, | |
| "loss": 2.3042, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.08186577820085673, | |
| "grad_norm": 6.97841215133667, | |
| "learning_rate": 1.4987368421052632e-05, | |
| "loss": 2.2691, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.0823417420276059, | |
| "grad_norm": 6.401273727416992, | |
| "learning_rate": 1.4452105263157896e-05, | |
| "loss": 2.5178, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.08281770585435506, | |
| "grad_norm": 6.540492534637451, | |
| "learning_rate": 1.391684210526316e-05, | |
| "loss": 2.2453, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.08329366968110423, | |
| "grad_norm": 6.713190078735352, | |
| "learning_rate": 1.338157894736842e-05, | |
| "loss": 2.2148, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.08376963350785341, | |
| "grad_norm": 7.125333309173584, | |
| "learning_rate": 1.2846315789473686e-05, | |
| "loss": 2.635, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.08424559733460257, | |
| "grad_norm": 7.104244709014893, | |
| "learning_rate": 1.2311052631578948e-05, | |
| "loss": 2.5499, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.08472156116135174, | |
| "grad_norm": 7.630082130432129, | |
| "learning_rate": 1.177578947368421e-05, | |
| "loss": 2.3113, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.0851975249881009, | |
| "grad_norm": 7.205958366394043, | |
| "learning_rate": 1.1240526315789475e-05, | |
| "loss": 2.3001, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.08567348881485007, | |
| "grad_norm": 6.465893268585205, | |
| "learning_rate": 1.0705263157894737e-05, | |
| "loss": 2.4646, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.08614945264159923, | |
| "grad_norm": 6.484694957733154, | |
| "learning_rate": 1.0170000000000001e-05, | |
| "loss": 2.3538, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.08662541646834841, | |
| "grad_norm": 6.740143775939941, | |
| "learning_rate": 9.634736842105263e-06, | |
| "loss": 2.1774, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.08710138029509758, | |
| "grad_norm": 6.847339153289795, | |
| "learning_rate": 9.099473684210527e-06, | |
| "loss": 2.5101, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.08757734412184674, | |
| "grad_norm": 7.767678260803223, | |
| "learning_rate": 8.56421052631579e-06, | |
| "loss": 2.5202, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.08805330794859591, | |
| "grad_norm": 8.59636116027832, | |
| "learning_rate": 8.028947368421052e-06, | |
| "loss": 2.7538, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.08852927177534507, | |
| "grad_norm": 8.195631980895996, | |
| "learning_rate": 7.493684210526316e-06, | |
| "loss": 2.4046, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.08900523560209424, | |
| "grad_norm": 6.393590927124023, | |
| "learning_rate": 6.95842105263158e-06, | |
| "loss": 2.2026, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.0894811994288434, | |
| "grad_norm": 6.6723551750183105, | |
| "learning_rate": 6.423157894736843e-06, | |
| "loss": 2.1685, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.08995716325559258, | |
| "grad_norm": 7.914825439453125, | |
| "learning_rate": 5.887894736842105e-06, | |
| "loss": 2.5318, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.09043312708234175, | |
| "grad_norm": 5.79133415222168, | |
| "learning_rate": 5.3526315789473684e-06, | |
| "loss": 2.1259, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.09090909090909091, | |
| "grad_norm": 8.203691482543945, | |
| "learning_rate": 4.817368421052632e-06, | |
| "loss": 2.6837, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.09138505473584008, | |
| "grad_norm": 8.923236846923828, | |
| "learning_rate": 4.282105263157895e-06, | |
| "loss": 2.6346, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.09186101856258924, | |
| "grad_norm": 7.227407932281494, | |
| "learning_rate": 3.746842105263158e-06, | |
| "loss": 2.5861, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.0923369823893384, | |
| "grad_norm": 6.976375102996826, | |
| "learning_rate": 3.2115789473684215e-06, | |
| "loss": 2.2964, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.09281294621608757, | |
| "grad_norm": 6.509735584259033, | |
| "learning_rate": 2.6763157894736842e-06, | |
| "loss": 2.341, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.09328891004283675, | |
| "grad_norm": 8.12691879272461, | |
| "learning_rate": 2.1410526315789474e-06, | |
| "loss": 2.4133, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.09376487386958592, | |
| "grad_norm": 7.400885581970215, | |
| "learning_rate": 1.6057894736842107e-06, | |
| "loss": 2.6121, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.09424083769633508, | |
| "grad_norm": 6.809811592102051, | |
| "learning_rate": 1.0705263157894737e-06, | |
| "loss": 2.4394, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.09471680152308425, | |
| "grad_norm": 7.949410915374756, | |
| "learning_rate": 5.352631578947368e-07, | |
| "loss": 2.7608, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.09519276534983341, | |
| "grad_norm": 8.497697830200195, | |
| "learning_rate": 0.0, | |
| "loss": 2.583, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09519276534983341, | |
| "eval_loss": 1.1956850290298462, | |
| "eval_runtime": 94.1756, | |
| "eval_samples_per_second": 9.397, | |
| "eval_steps_per_second": 2.357, | |
| "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": 7.15499809800192e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |