| { | |
| "best_metric": 0.08518081903457642, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.4260249554367201, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0071301247771836, | |
| "grad_norm": 4.84234619140625, | |
| "learning_rate": 1e-05, | |
| "loss": 2.2642, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0071301247771836, | |
| "eval_loss": 3.368955373764038, | |
| "eval_runtime": 17.5527, | |
| "eval_samples_per_second": 13.445, | |
| "eval_steps_per_second": 3.361, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0142602495543672, | |
| "grad_norm": 4.224218368530273, | |
| "learning_rate": 2e-05, | |
| "loss": 2.4353, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0213903743315508, | |
| "grad_norm": 5.9760003089904785, | |
| "learning_rate": 3e-05, | |
| "loss": 2.4882, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0285204991087344, | |
| "grad_norm": 4.934040546417236, | |
| "learning_rate": 4e-05, | |
| "loss": 2.3519, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.035650623885918005, | |
| "grad_norm": 3.3341965675354004, | |
| "learning_rate": 5e-05, | |
| "loss": 2.1404, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0427807486631016, | |
| "grad_norm": 3.6450459957122803, | |
| "learning_rate": 6e-05, | |
| "loss": 1.7935, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.049910873440285206, | |
| "grad_norm": 6.697213649749756, | |
| "learning_rate": 7e-05, | |
| "loss": 1.7521, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0570409982174688, | |
| "grad_norm": 6.886330604553223, | |
| "learning_rate": 8e-05, | |
| "loss": 1.8535, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.06417112299465241, | |
| "grad_norm": 5.420178413391113, | |
| "learning_rate": 9e-05, | |
| "loss": 1.7291, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.07130124777183601, | |
| "grad_norm": 4.2837605476379395, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1731, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0784313725490196, | |
| "grad_norm": 8.207549095153809, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.5613, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0855614973262032, | |
| "grad_norm": 4.694660663604736, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.3187, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.09269162210338681, | |
| "grad_norm": 5.1253132820129395, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.0932, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.09982174688057041, | |
| "grad_norm": 5.843274116516113, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.3912, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.10695187165775401, | |
| "grad_norm": 13.196741104125977, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.1644, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.1140819964349376, | |
| "grad_norm": 7.18977165222168, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.3183, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.12121212121212122, | |
| "grad_norm": 5.043786525726318, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.161, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.12834224598930483, | |
| "grad_norm": 4.29835844039917, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.8264, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.1354723707664884, | |
| "grad_norm": 4.659335136413574, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.636, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.14260249554367202, | |
| "grad_norm": 11.284347534179688, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 2.0547, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1497326203208556, | |
| "grad_norm": 7.5890607833862305, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.3708, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1568627450980392, | |
| "grad_norm": 15.984006881713867, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.378, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.16399286987522282, | |
| "grad_norm": 13.209542274475098, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.3666, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.1711229946524064, | |
| "grad_norm": 8.18061637878418, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.4046, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.17825311942959002, | |
| "grad_norm": 8.380427360534668, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.6977, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.18538324420677363, | |
| "grad_norm": 6.196962356567383, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.0766, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.1925133689839572, | |
| "grad_norm": 3.1025502681732178, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.8189, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.19964349376114082, | |
| "grad_norm": 4.083560466766357, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.8789, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.20677361853832443, | |
| "grad_norm": 17.323400497436523, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.5116, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.21390374331550802, | |
| "grad_norm": 15.207404136657715, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.8476, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.22103386809269163, | |
| "grad_norm": 7.809723854064941, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.3606, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.2281639928698752, | |
| "grad_norm": 17.65806007385254, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.196, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.23529411764705882, | |
| "grad_norm": 11.293682098388672, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.3326, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.24242424242424243, | |
| "grad_norm": 7.203268051147461, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.9895, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.24955436720142601, | |
| "grad_norm": 8.514725685119629, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.8213, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.25668449197860965, | |
| "grad_norm": 17.586591720581055, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 3.0231, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.2638146167557932, | |
| "grad_norm": 12.793550491333008, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.4191, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.2709447415329768, | |
| "grad_norm": 9.30056095123291, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.6867, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.27807486631016043, | |
| "grad_norm": 4.500026702880859, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.869, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.28520499108734404, | |
| "grad_norm": 2.7901341915130615, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.638, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.29233511586452765, | |
| "grad_norm": 2.9541563987731934, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.5747, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.2994652406417112, | |
| "grad_norm": 2.571490526199341, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.5393, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.3065953654188948, | |
| "grad_norm": 4.626973628997803, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.7337, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.3137254901960784, | |
| "grad_norm": 2.8816380500793457, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.5261, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.32085561497326204, | |
| "grad_norm": 2.204977035522461, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.3566, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.32798573975044565, | |
| "grad_norm": 5.014333248138428, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.6197, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.33511586452762926, | |
| "grad_norm": 3.4855005741119385, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.5684, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.3422459893048128, | |
| "grad_norm": 2.7158308029174805, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.4849, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.3493761140819964, | |
| "grad_norm": 2.5199692249298096, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.2222, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.35650623885918004, | |
| "grad_norm": 3.8967249393463135, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.3396, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.35650623885918004, | |
| "eval_loss": 0.6471731066703796, | |
| "eval_runtime": 17.9219, | |
| "eval_samples_per_second": 13.168, | |
| "eval_steps_per_second": 3.292, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.36363636363636365, | |
| "grad_norm": 2.9278488159179688, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.2589, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.37076648841354726, | |
| "grad_norm": 5.670351505279541, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.2536, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.3778966131907308, | |
| "grad_norm": 3.0680196285247803, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.256, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.3850267379679144, | |
| "grad_norm": 2.5128862857818604, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.3629, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.39215686274509803, | |
| "grad_norm": 4.349164962768555, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.4441, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.39928698752228164, | |
| "grad_norm": 5.842977046966553, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.3586, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.40641711229946526, | |
| "grad_norm": 6.305400848388672, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.4329, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.41354723707664887, | |
| "grad_norm": 5.4536638259887695, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.6472, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.4206773618538324, | |
| "grad_norm": 4.1123433113098145, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.3898, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.42780748663101603, | |
| "grad_norm": 10.140790939331055, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.4289, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.43493761140819964, | |
| "grad_norm": 3.6327106952667236, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.256, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.44206773618538325, | |
| "grad_norm": 4.8620991706848145, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.3176, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.44919786096256686, | |
| "grad_norm": 5.55952262878418, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.5289, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.4563279857397504, | |
| "grad_norm": 3.2413785457611084, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.4002, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.46345811051693403, | |
| "grad_norm": 4.864469051361084, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.3336, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.47058823529411764, | |
| "grad_norm": 6.035539150238037, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.6054, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.47771836007130125, | |
| "grad_norm": 3.9230198860168457, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.4514, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.48484848484848486, | |
| "grad_norm": 17.25653839111328, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.3989, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.4919786096256685, | |
| "grad_norm": 3.696359157562256, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.4349, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.49910873440285203, | |
| "grad_norm": 5.974207878112793, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.4928, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.5062388591800356, | |
| "grad_norm": 13.7036714553833, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.9155, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.5133689839572193, | |
| "grad_norm": 10.811036109924316, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.4933, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.5204991087344029, | |
| "grad_norm": 8.134906768798828, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.1337, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.5276292335115864, | |
| "grad_norm": 4.494138717651367, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.7427, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.5347593582887701, | |
| "grad_norm": 4.554021835327148, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.4558, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.5418894830659536, | |
| "grad_norm": 3.2674827575683594, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.4465, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.5490196078431373, | |
| "grad_norm": 2.9735140800476074, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.3446, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.5561497326203209, | |
| "grad_norm": 4.049015045166016, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.5291, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.5632798573975044, | |
| "grad_norm": 4.700780391693115, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.3606, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.5704099821746881, | |
| "grad_norm": 2.244403123855591, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.3, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.5775401069518716, | |
| "grad_norm": 1.7338509559631348, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.2294, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.5846702317290553, | |
| "grad_norm": 3.3586502075195312, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.3078, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.5918003565062389, | |
| "grad_norm": 1.6617681980133057, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.1767, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.5989304812834224, | |
| "grad_norm": 1.8029754161834717, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.1475, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.6060606060606061, | |
| "grad_norm": 2.047156572341919, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.1415, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.6131907308377896, | |
| "grad_norm": 1.8421618938446045, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.1069, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.6203208556149733, | |
| "grad_norm": 2.745173692703247, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.3016, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.6274509803921569, | |
| "grad_norm": 3.796250104904175, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.2596, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.6345811051693404, | |
| "grad_norm": 2.5031118392944336, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.1715, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.6417112299465241, | |
| "grad_norm": 5.365614414215088, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.245, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.6488413547237076, | |
| "grad_norm": 2.6736977100372314, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.2014, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.6559714795008913, | |
| "grad_norm": 2.335885763168335, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.1751, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.6631016042780749, | |
| "grad_norm": 2.2748329639434814, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.2003, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.6702317290552585, | |
| "grad_norm": 3.269821882247925, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.2293, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.6773618538324421, | |
| "grad_norm": 2.4277210235595703, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.2263, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.6844919786096256, | |
| "grad_norm": 3.116123676300049, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.2205, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.6916221033868093, | |
| "grad_norm": 1.997261881828308, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.2223, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.6987522281639929, | |
| "grad_norm": 2.6265411376953125, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.289, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.7058823529411765, | |
| "grad_norm": 2.684760093688965, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.2045, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.7130124777183601, | |
| "grad_norm": 3.4225356578826904, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.2637, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.7130124777183601, | |
| "eval_loss": 0.29322880506515503, | |
| "eval_runtime": 17.9313, | |
| "eval_samples_per_second": 13.161, | |
| "eval_steps_per_second": 3.29, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.7201426024955436, | |
| "grad_norm": 2.119136333465576, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.2142, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 6.241914749145508, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.3946, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.7344028520499108, | |
| "grad_norm": 6.558347225189209, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.3159, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.7415329768270945, | |
| "grad_norm": 2.7997732162475586, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.282, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.7486631016042781, | |
| "grad_norm": 3.6334173679351807, | |
| "learning_rate": 5e-05, | |
| "loss": 0.4715, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.7557932263814616, | |
| "grad_norm": 9.802560806274414, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.8793, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.7629233511586453, | |
| "grad_norm": 6.028875350952148, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.6313, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.7700534759358288, | |
| "grad_norm": 3.834984302520752, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.5179, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.7771836007130125, | |
| "grad_norm": 5.074844837188721, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.4993, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.7843137254901961, | |
| "grad_norm": 4.519394874572754, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.353, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.7914438502673797, | |
| "grad_norm": 1.713512897491455, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.2382, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.7985739750445633, | |
| "grad_norm": 1.4872967004776, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.1583, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.8057040998217468, | |
| "grad_norm": 2.2665367126464844, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.1713, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.8128342245989305, | |
| "grad_norm": 3.3006982803344727, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.2197, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.8199643493761141, | |
| "grad_norm": 1.460872769355774, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.2034, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.8270944741532977, | |
| "grad_norm": 2.812030076980591, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.2966, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.8342245989304813, | |
| "grad_norm": 1.875169277191162, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.1972, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.8413547237076648, | |
| "grad_norm": 1.2524548768997192, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.1614, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.8484848484848485, | |
| "grad_norm": 1.9770088195800781, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.25, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.8556149732620321, | |
| "grad_norm": 1.616072177886963, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.203, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.8627450980392157, | |
| "grad_norm": 2.159360647201538, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.1621, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.8698752228163993, | |
| "grad_norm": 1.9339467287063599, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.1249, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.8770053475935828, | |
| "grad_norm": 1.5325121879577637, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.2229, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.8841354723707665, | |
| "grad_norm": 3.2308831214904785, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.1301, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.8912655971479501, | |
| "grad_norm": 1.1866799592971802, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0882, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.8983957219251337, | |
| "grad_norm": 1.7538654804229736, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.1651, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.9055258467023173, | |
| "grad_norm": 1.3078639507293701, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0684, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.9126559714795008, | |
| "grad_norm": 1.6905395984649658, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.1976, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.9197860962566845, | |
| "grad_norm": 3.5100855827331543, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.2035, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.9269162210338681, | |
| "grad_norm": 3.0280840396881104, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.1964, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.9340463458110517, | |
| "grad_norm": 4.221747398376465, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.2683, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.9411764705882353, | |
| "grad_norm": 1.9113394021987915, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.1909, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.948306595365419, | |
| "grad_norm": 2.266374349594116, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.2751, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.9554367201426025, | |
| "grad_norm": 7.384371280670166, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.2176, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.9625668449197861, | |
| "grad_norm": 2.6050963401794434, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.2767, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.9696969696969697, | |
| "grad_norm": 2.806575298309326, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.268, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.9768270944741533, | |
| "grad_norm": 2.965360403060913, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.1621, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.983957219251337, | |
| "grad_norm": 2.001814603805542, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.2763, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.9910873440285205, | |
| "grad_norm": 2.735741138458252, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.2692, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.9982174688057041, | |
| "grad_norm": 2.4110593795776367, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.3448, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.0053475935828877, | |
| "grad_norm": 5.351468086242676, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.5775, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.0124777183600713, | |
| "grad_norm": 2.348818063735962, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.2322, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.0196078431372548, | |
| "grad_norm": 1.7678096294403076, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.2372, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.0267379679144386, | |
| "grad_norm": 1.8597091436386108, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.2592, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.0338680926916222, | |
| "grad_norm": 1.3026940822601318, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.2226, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.0409982174688057, | |
| "grad_norm": 1.5875272750854492, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.2352, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.0481283422459893, | |
| "grad_norm": 1.7961909770965576, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.1683, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.0552584670231728, | |
| "grad_norm": 3.0041821002960205, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.1749, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.0623885918003566, | |
| "grad_norm": 1.6207810640335083, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.1075, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.0695187165775402, | |
| "grad_norm": 1.3290566205978394, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.1015, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.0695187165775402, | |
| "eval_loss": 0.12634782493114471, | |
| "eval_runtime": 17.8904, | |
| "eval_samples_per_second": 13.191, | |
| "eval_steps_per_second": 3.298, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.0766488413547237, | |
| "grad_norm": 1.3415688276290894, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.1495, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.0837789661319073, | |
| "grad_norm": 1.26509690284729, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0756, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.0909090909090908, | |
| "grad_norm": 0.8995376229286194, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0922, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.0980392156862746, | |
| "grad_norm": 0.7741421461105347, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.1309, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.1051693404634582, | |
| "grad_norm": 1.002313256263733, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.1067, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.1122994652406417, | |
| "grad_norm": 1.1984858512878418, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0788, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.1194295900178253, | |
| "grad_norm": 0.9591169357299805, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0878, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.1265597147950088, | |
| "grad_norm": 1.0988893508911133, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.1426, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.1336898395721926, | |
| "grad_norm": 0.8119126558303833, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0915, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.1408199643493762, | |
| "grad_norm": 0.8352858424186707, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.1179, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.1479500891265597, | |
| "grad_norm": 0.6990809440612793, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0853, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.1550802139037433, | |
| "grad_norm": 0.5923833847045898, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.061, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.1622103386809268, | |
| "grad_norm": 0.9815614223480225, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0908, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.1693404634581106, | |
| "grad_norm": 1.124747395515442, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.1203, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.1764705882352942, | |
| "grad_norm": 0.8811934590339661, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0818, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.1836007130124777, | |
| "grad_norm": 0.6433349847793579, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0662, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.1907308377896613, | |
| "grad_norm": 0.6433457732200623, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.0864, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.1978609625668448, | |
| "grad_norm": 1.1961352825164795, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.2094, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.2049910873440286, | |
| "grad_norm": 1.4341731071472168, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.1736, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.2121212121212122, | |
| "grad_norm": 0.6444214582443237, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.1015, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.2192513368983957, | |
| "grad_norm": 0.8121644854545593, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.118, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.2263814616755793, | |
| "grad_norm": 0.8862426280975342, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0899, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.2335115864527628, | |
| "grad_norm": 0.699418306350708, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0786, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.2406417112299466, | |
| "grad_norm": 0.7339772582054138, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0647, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.2477718360071302, | |
| "grad_norm": 1.8839962482452393, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.3744, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.2549019607843137, | |
| "grad_norm": 2.035961389541626, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.2301, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.2620320855614973, | |
| "grad_norm": 1.8421082496643066, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0912, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.2691622103386808, | |
| "grad_norm": 1.3662892580032349, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.1209, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.2762923351158646, | |
| "grad_norm": 0.7958939671516418, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0675, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.2834224598930482, | |
| "grad_norm": 0.5694426894187927, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0491, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.2905525846702317, | |
| "grad_norm": 0.7167963981628418, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.1371, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.2976827094474153, | |
| "grad_norm": 1.4464503526687622, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.1121, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.3048128342245988, | |
| "grad_norm": 0.7022302150726318, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0552, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.3119429590017826, | |
| "grad_norm": 0.5985070466995239, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0499, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.3190730837789661, | |
| "grad_norm": 0.4780137240886688, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.052, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.3262032085561497, | |
| "grad_norm": 0.5075286626815796, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0885, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.5456973910331726, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0432, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.3404634581105168, | |
| "grad_norm": 0.4035959243774414, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.0239, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.3475935828877006, | |
| "grad_norm": 0.6267684698104858, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.1171, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.3547237076648841, | |
| "grad_norm": 0.5383210182189941, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0935, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.3618538324420677, | |
| "grad_norm": 0.3904607892036438, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0171, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.3689839572192513, | |
| "grad_norm": 0.7520592212677002, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0653, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.3761140819964348, | |
| "grad_norm": 0.5954171419143677, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0703, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.3832442067736186, | |
| "grad_norm": 0.5096472501754761, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0591, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.3903743315508021, | |
| "grad_norm": 0.5481864213943481, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0826, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.3975044563279857, | |
| "grad_norm": 0.3931596875190735, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0409, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.4046345811051695, | |
| "grad_norm": 0.5922490358352661, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0608, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.4117647058823528, | |
| "grad_norm": 1.7752002477645874, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.1108, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.4188948306595366, | |
| "grad_norm": 0.7069717645645142, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.1026, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.4260249554367201, | |
| "grad_norm": 0.5625461339950562, | |
| "learning_rate": 0.0, | |
| "loss": 0.0645, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.4260249554367201, | |
| "eval_loss": 0.08518081903457642, | |
| "eval_runtime": 17.884, | |
| "eval_samples_per_second": 13.196, | |
| "eval_steps_per_second": 3.299, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "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": 3.01514151551828e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |