| { | |
| "best_metric": 0.5623053908348083, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.642570281124498, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00321285140562249, | |
| "grad_norm": 43.378211975097656, | |
| "learning_rate": 1e-05, | |
| "loss": 7.1917, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00321285140562249, | |
| "eval_loss": 4.8860578536987305, | |
| "eval_runtime": 45.1456, | |
| "eval_samples_per_second": 11.607, | |
| "eval_steps_per_second": 2.902, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00642570281124498, | |
| "grad_norm": 42.9906120300293, | |
| "learning_rate": 2e-05, | |
| "loss": 6.9171, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00963855421686747, | |
| "grad_norm": 21.34159278869629, | |
| "learning_rate": 3e-05, | |
| "loss": 4.0974, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01285140562248996, | |
| "grad_norm": 15.48149299621582, | |
| "learning_rate": 4e-05, | |
| "loss": 3.5558, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01606425702811245, | |
| "grad_norm": 13.900507926940918, | |
| "learning_rate": 5e-05, | |
| "loss": 3.0166, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01927710843373494, | |
| "grad_norm": 9.406023025512695, | |
| "learning_rate": 6e-05, | |
| "loss": 1.6976, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02248995983935743, | |
| "grad_norm": 8.96674633026123, | |
| "learning_rate": 7e-05, | |
| "loss": 1.675, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02570281124497992, | |
| "grad_norm": 14.493627548217773, | |
| "learning_rate": 8e-05, | |
| "loss": 1.7138, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02891566265060241, | |
| "grad_norm": 14.497711181640625, | |
| "learning_rate": 9e-05, | |
| "loss": 1.9516, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0321285140562249, | |
| "grad_norm": 5.4678850173950195, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0942, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.035341365461847386, | |
| "grad_norm": 8.695549964904785, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.1549, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03855421686746988, | |
| "grad_norm": 5.363256931304932, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.8493, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04176706827309237, | |
| "grad_norm": 6.721149921417236, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.2137, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04497991967871486, | |
| "grad_norm": 4.540046691894531, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.8473, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04819277108433735, | |
| "grad_norm": 4.353954315185547, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.7053, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05140562248995984, | |
| "grad_norm": 4.197540283203125, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.625, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.05461847389558233, | |
| "grad_norm": 4.212319850921631, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.7468, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05783132530120482, | |
| "grad_norm": 3.5598859786987305, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.949, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06104417670682731, | |
| "grad_norm": 3.6704111099243164, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.8865, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0642570281124498, | |
| "grad_norm": 3.7770400047302246, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.8484, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06746987951807229, | |
| "grad_norm": 3.669597625732422, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.8865, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.07068273092369477, | |
| "grad_norm": 4.030519485473633, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.7993, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.07389558232931727, | |
| "grad_norm": 3.771228313446045, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.7086, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07710843373493977, | |
| "grad_norm": 3.081094741821289, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.5326, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.08032128514056225, | |
| "grad_norm": 3.571728229522705, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.5236, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08353413654618475, | |
| "grad_norm": 3.535062074661255, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.4291, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.08674698795180723, | |
| "grad_norm": 3.98887038230896, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.7028, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08995983935742972, | |
| "grad_norm": 7.923908233642578, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.8625, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.09317269076305221, | |
| "grad_norm": 4.207087993621826, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.7255, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0963855421686747, | |
| "grad_norm": 4.31841516494751, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.9415, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09959839357429719, | |
| "grad_norm": 4.9078288078308105, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.0768, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.10281124497991968, | |
| "grad_norm": 2.787365674972534, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.5482, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.10602409638554217, | |
| "grad_norm": 3.606337070465088, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.041, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.10923694779116466, | |
| "grad_norm": 3.7554211616516113, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.7479, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.11244979919678715, | |
| "grad_norm": 2.6411750316619873, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.4975, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.11566265060240964, | |
| "grad_norm": 3.1252663135528564, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.6202, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.11887550200803212, | |
| "grad_norm": 4.51902437210083, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.8194, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.12208835341365462, | |
| "grad_norm": 3.887127161026001, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.78, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.12530120481927712, | |
| "grad_norm": 3.95611834526062, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.6979, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1285140562248996, | |
| "grad_norm": 5.040665626525879, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.876, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.13172690763052208, | |
| "grad_norm": 3.3330559730529785, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.7234, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.13493975903614458, | |
| "grad_norm": 4.677862644195557, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.8265, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.13815261044176708, | |
| "grad_norm": 3.277540445327759, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.5864, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.14136546184738955, | |
| "grad_norm": 3.379077196121216, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.4427, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.14457831325301204, | |
| "grad_norm": 3.9832446575164795, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.8005, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.14779116465863454, | |
| "grad_norm": 2.9706380367279053, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.6162, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.15100401606425704, | |
| "grad_norm": 4.316315650939941, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.8323, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.15421686746987953, | |
| "grad_norm": 3.8548223972320557, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.5434, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.157429718875502, | |
| "grad_norm": 3.9953150749206543, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.6779, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1606425702811245, | |
| "grad_norm": 4.220303058624268, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.9046, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1606425702811245, | |
| "eval_loss": 0.7079682350158691, | |
| "eval_runtime": 45.9464, | |
| "eval_samples_per_second": 11.405, | |
| "eval_steps_per_second": 2.851, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.163855421686747, | |
| "grad_norm": 5.201648235321045, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.1973, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.1670682730923695, | |
| "grad_norm": 5.280684471130371, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.1162, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.17028112449799196, | |
| "grad_norm": 3.7499783039093018, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.7962, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.17349397590361446, | |
| "grad_norm": 5.2729268074035645, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.0368, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.17670682730923695, | |
| "grad_norm": 3.0536630153656006, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.644, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.17991967871485945, | |
| "grad_norm": 4.1184844970703125, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.7712, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.18313253012048192, | |
| "grad_norm": 4.019740104675293, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.8859, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.18634538152610441, | |
| "grad_norm": 3.0017526149749756, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.4861, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.1895582329317269, | |
| "grad_norm": 2.8464038372039795, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.3811, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1927710843373494, | |
| "grad_norm": 3.7457327842712402, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.8073, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.19598393574297188, | |
| "grad_norm": 2.9158785343170166, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.5083, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.19919678714859437, | |
| "grad_norm": 3.6816980838775635, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.6747, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.20240963855421687, | |
| "grad_norm": 3.4483675956726074, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.6042, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.20562248995983937, | |
| "grad_norm": 3.006639242172241, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.4822, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.20883534136546184, | |
| "grad_norm": 2.9801151752471924, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.6998, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.21204819277108433, | |
| "grad_norm": 2.6671342849731445, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.4371, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.21526104417670683, | |
| "grad_norm": 3.3461883068084717, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.6186, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.21847389558232932, | |
| "grad_norm": 2.88969087600708, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.6666, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2216867469879518, | |
| "grad_norm": 2.9498941898345947, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.6094, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2248995983935743, | |
| "grad_norm": 3.6393697261810303, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.9589, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2281124497991968, | |
| "grad_norm": 3.4475810527801514, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.5965, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.23132530120481928, | |
| "grad_norm": 2.894758701324463, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.4357, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.23453815261044178, | |
| "grad_norm": 2.972968339920044, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.6122, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.23775100401606425, | |
| "grad_norm": 3.546990394592285, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.6347, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.24096385542168675, | |
| "grad_norm": 3.3609681129455566, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.7828, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.24417670682730924, | |
| "grad_norm": 3.59785795211792, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.8028, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.24738955823293174, | |
| "grad_norm": 3.608123302459717, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.739, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.25060240963855424, | |
| "grad_norm": 2.7174291610717773, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.5192, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.25381526104417673, | |
| "grad_norm": 4.14845085144043, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.5451, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.2570281124497992, | |
| "grad_norm": 3.691070556640625, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.7548, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.26024096385542167, | |
| "grad_norm": 3.6830408573150635, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.0477, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.26345381526104417, | |
| "grad_norm": 2.677612066268921, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.3927, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.26666666666666666, | |
| "grad_norm": 3.642026424407959, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.7002, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.26987951807228916, | |
| "grad_norm": 3.2168092727661133, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.5811, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.27309236947791166, | |
| "grad_norm": 3.1234962940216064, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.6275, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.27630522088353415, | |
| "grad_norm": 3.4648942947387695, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.6923, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.27951807228915665, | |
| "grad_norm": 3.4100563526153564, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.731, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.2827309236947791, | |
| "grad_norm": 2.866823673248291, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.5243, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.2859437751004016, | |
| "grad_norm": 3.5110268592834473, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.6503, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2891566265060241, | |
| "grad_norm": 3.6085054874420166, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.7454, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2923694779116466, | |
| "grad_norm": 3.804030179977417, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.8885, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.2955823293172691, | |
| "grad_norm": 3.9683756828308105, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.9552, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2987951807228916, | |
| "grad_norm": 3.0400495529174805, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.5749, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.30200803212851407, | |
| "grad_norm": 2.430614948272705, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.406, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.30522088353413657, | |
| "grad_norm": 3.800335168838501, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.7168, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.30843373493975906, | |
| "grad_norm": 2.999204158782959, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.3841, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.3116465863453815, | |
| "grad_norm": 3.8305814266204834, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.7116, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.314859437751004, | |
| "grad_norm": 2.825263261795044, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.4951, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.3180722891566265, | |
| "grad_norm": 3.391592264175415, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.5773, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.321285140562249, | |
| "grad_norm": 3.7341911792755127, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.7305, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.321285140562249, | |
| "eval_loss": 0.6347183585166931, | |
| "eval_runtime": 45.9605, | |
| "eval_samples_per_second": 11.401, | |
| "eval_steps_per_second": 2.85, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3244979919678715, | |
| "grad_norm": 3.9959065914154053, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.2413, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.327710843373494, | |
| "grad_norm": 4.994175910949707, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.1545, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.3309236947791165, | |
| "grad_norm": 2.592308759689331, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.6403, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.334136546184739, | |
| "grad_norm": 2.938979148864746, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.8006, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.3373493975903614, | |
| "grad_norm": 2.6757688522338867, | |
| "learning_rate": 5e-05, | |
| "loss": 0.5273, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3405622489959839, | |
| "grad_norm": 2.8961260318756104, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.6806, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3437751004016064, | |
| "grad_norm": 2.6824748516082764, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.642, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3469879518072289, | |
| "grad_norm": 2.843604564666748, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.5317, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.3502008032128514, | |
| "grad_norm": 3.1411731243133545, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.6839, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3534136546184739, | |
| "grad_norm": 3.153282880783081, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.7258, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3566265060240964, | |
| "grad_norm": 2.554614782333374, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.4067, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.3598393574297189, | |
| "grad_norm": 2.387897491455078, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.4426, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.36305220883534134, | |
| "grad_norm": 4.433763027191162, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.6101, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.36626506024096384, | |
| "grad_norm": 2.4086811542510986, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.4015, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.36947791164658633, | |
| "grad_norm": 3.6991384029388428, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.7094, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.37269076305220883, | |
| "grad_norm": 2.424997091293335, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.3898, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.3759036144578313, | |
| "grad_norm": 3.07243013381958, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.4756, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.3791164658634538, | |
| "grad_norm": 3.331669569015503, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.8839, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.3823293172690763, | |
| "grad_norm": 2.807100772857666, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.5148, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.3855421686746988, | |
| "grad_norm": 3.8023102283477783, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.7331, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3887550200803213, | |
| "grad_norm": 3.4350459575653076, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.6968, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.39196787148594375, | |
| "grad_norm": 2.8601789474487305, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.5103, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.39518072289156625, | |
| "grad_norm": 3.3219640254974365, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.6687, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.39839357429718875, | |
| "grad_norm": 3.554771661758423, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.0192, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.40160642570281124, | |
| "grad_norm": 3.5397536754608154, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.6995, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.40481927710843374, | |
| "grad_norm": 3.086008310317993, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.6241, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.40803212851405624, | |
| "grad_norm": 2.5428953170776367, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.4894, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.41124497991967873, | |
| "grad_norm": 4.187087535858154, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.6923, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.41445783132530123, | |
| "grad_norm": 3.550123691558838, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.7577, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.41767068273092367, | |
| "grad_norm": 3.8170347213745117, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.7507, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.42088353413654617, | |
| "grad_norm": 2.6992897987365723, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.4576, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.42409638554216866, | |
| "grad_norm": 3.4043009281158447, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.6233, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.42730923694779116, | |
| "grad_norm": 3.403949022293091, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.6223, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.43052208835341366, | |
| "grad_norm": 3.3217620849609375, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.5715, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.43373493975903615, | |
| "grad_norm": 2.954711437225342, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.6385, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.43694779116465865, | |
| "grad_norm": 4.182713031768799, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.6409, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.44016064257028115, | |
| "grad_norm": 3.129976749420166, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.6499, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.4433734939759036, | |
| "grad_norm": 2.3581554889678955, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.4461, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.4465863453815261, | |
| "grad_norm": 2.7010395526885986, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.4904, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.4497991967871486, | |
| "grad_norm": 3.3056650161743164, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.5408, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4530120481927711, | |
| "grad_norm": 4.488265514373779, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.802, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.4562248995983936, | |
| "grad_norm": 3.1086909770965576, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.5168, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.45943775100401607, | |
| "grad_norm": 3.384004831314087, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.628, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.46265060240963857, | |
| "grad_norm": 3.072906017303467, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.4627, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.46586345381526106, | |
| "grad_norm": 3.3753528594970703, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.6318, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.46907630522088356, | |
| "grad_norm": 3.0711803436279297, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.5714, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.472289156626506, | |
| "grad_norm": 3.0743980407714844, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.5818, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4755020080321285, | |
| "grad_norm": 3.5532915592193604, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.9047, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.478714859437751, | |
| "grad_norm": 3.209702968597412, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.3648, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.4819277108433735, | |
| "grad_norm": 2.8248977661132812, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.4452, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4819277108433735, | |
| "eval_loss": 0.5840469002723694, | |
| "eval_runtime": 45.9847, | |
| "eval_samples_per_second": 11.395, | |
| "eval_steps_per_second": 2.849, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.485140562248996, | |
| "grad_norm": 4.840352535247803, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.4463, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4883534136546185, | |
| "grad_norm": 3.8735439777374268, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.0224, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.491566265060241, | |
| "grad_norm": 3.095221519470215, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.6601, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.4947791164658635, | |
| "grad_norm": 3.9398765563964844, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.8681, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.4979919678714859, | |
| "grad_norm": 2.261209726333618, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.533, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.5012048192771085, | |
| "grad_norm": 3.1193084716796875, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.5615, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.5044176706827309, | |
| "grad_norm": 3.3449270725250244, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.619, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.5076305220883535, | |
| "grad_norm": 2.421292543411255, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.4291, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.5108433734939759, | |
| "grad_norm": 2.9491381645202637, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.4409, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.5140562248995983, | |
| "grad_norm": 3.581892728805542, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.5107, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5172690763052209, | |
| "grad_norm": 4.003829479217529, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.7475, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.5204819277108433, | |
| "grad_norm": 2.2960057258605957, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.4179, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.5236947791164659, | |
| "grad_norm": 3.0141780376434326, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.6065, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.5269076305220883, | |
| "grad_norm": 2.1354188919067383, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.418, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.5301204819277109, | |
| "grad_norm": 3.0823445320129395, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.6228, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.5333333333333333, | |
| "grad_norm": 2.9922661781311035, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.5633, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.5365461847389559, | |
| "grad_norm": 2.917760133743286, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.6028, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.5397590361445783, | |
| "grad_norm": 2.9899909496307373, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.5237, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.5429718875502008, | |
| "grad_norm": 3.4307661056518555, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.6825, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.5461847389558233, | |
| "grad_norm": 2.8658881187438965, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.3397, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5493975903614458, | |
| "grad_norm": 2.935495615005493, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.6252, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.5526104417670683, | |
| "grad_norm": 2.770616292953491, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.5727, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.5558232931726907, | |
| "grad_norm": 2.3446643352508545, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.4583, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.5590361445783133, | |
| "grad_norm": 2.2611119747161865, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.3853, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.5622489959839357, | |
| "grad_norm": 2.818490743637085, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.4664, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.5654618473895582, | |
| "grad_norm": 2.873927116394043, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.5385, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.5686746987951807, | |
| "grad_norm": 2.824052572250366, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.5574, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.5718875502008032, | |
| "grad_norm": 3.2166662216186523, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.445, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.5751004016064257, | |
| "grad_norm": 3.3107292652130127, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.5586, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.5783132530120482, | |
| "grad_norm": 2.9845964908599854, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.3918, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5815261044176707, | |
| "grad_norm": 2.5777885913848877, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.558, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.5847389558232932, | |
| "grad_norm": 2.727137565612793, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.447, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.5879518072289157, | |
| "grad_norm": 3.195466995239258, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.6126, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.5911646586345382, | |
| "grad_norm": 2.2888362407684326, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.3415, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5943775100401606, | |
| "grad_norm": 4.878308296203613, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.6111, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5975903614457831, | |
| "grad_norm": 2.0078675746917725, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.368, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.6008032128514056, | |
| "grad_norm": 2.9476611614227295, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.64, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.6040160642570281, | |
| "grad_norm": 3.3154850006103516, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.8211, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.6072289156626506, | |
| "grad_norm": 2.69844126701355, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.3529, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.6104417670682731, | |
| "grad_norm": 2.620026111602783, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.4808, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6136546184738956, | |
| "grad_norm": 3.183415651321411, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.5456, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.6168674698795181, | |
| "grad_norm": 3.7568700313568115, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.5221, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.6200803212851406, | |
| "grad_norm": 3.205428123474121, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.756, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.623293172690763, | |
| "grad_norm": 2.6913352012634277, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.3642, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.6265060240963856, | |
| "grad_norm": 3.261702060699463, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.5163, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.629718875502008, | |
| "grad_norm": 2.589987277984619, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.5645, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.6329317269076306, | |
| "grad_norm": 1.971290946006775, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.3183, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.636144578313253, | |
| "grad_norm": 3.6872410774230957, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.6932, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.6393574297188755, | |
| "grad_norm": 3.2588086128234863, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.5525, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.642570281124498, | |
| "grad_norm": 3.6018600463867188, | |
| "learning_rate": 0.0, | |
| "loss": 0.4481, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.642570281124498, | |
| "eval_loss": 0.5623053908348083, | |
| "eval_runtime": 45.9419, | |
| "eval_samples_per_second": 11.406, | |
| "eval_steps_per_second": 2.851, | |
| "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": 3.1421973496489574e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |