| { | |
| "best_metric": 10.55037784576416, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.020064205457463884, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00010032102728731942, | |
| "grad_norm": 1.5991774797439575, | |
| "learning_rate": 1.009e-05, | |
| "loss": 22.2622, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00010032102728731942, | |
| "eval_loss": 11.123811721801758, | |
| "eval_runtime": 70.5623, | |
| "eval_samples_per_second": 59.493, | |
| "eval_steps_per_second": 14.88, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00020064205457463884, | |
| "grad_norm": 1.434697151184082, | |
| "learning_rate": 2.018e-05, | |
| "loss": 22.2601, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00030096308186195825, | |
| "grad_norm": 1.767243504524231, | |
| "learning_rate": 3.027e-05, | |
| "loss": 22.2774, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0004012841091492777, | |
| "grad_norm": 1.4239779710769653, | |
| "learning_rate": 4.036e-05, | |
| "loss": 22.2314, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0005016051364365971, | |
| "grad_norm": 1.3848161697387695, | |
| "learning_rate": 5.045e-05, | |
| "loss": 22.2259, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0006019261637239165, | |
| "grad_norm": 1.6321865320205688, | |
| "learning_rate": 6.054e-05, | |
| "loss": 22.2408, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0007022471910112359, | |
| "grad_norm": 1.7302271127700806, | |
| "learning_rate": 7.062999999999999e-05, | |
| "loss": 22.2005, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0008025682182985554, | |
| "grad_norm": 1.7554914951324463, | |
| "learning_rate": 8.072e-05, | |
| "loss": 22.1973, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0009028892455858748, | |
| "grad_norm": 1.476845145225525, | |
| "learning_rate": 9.081e-05, | |
| "loss": 22.1782, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0010032102728731941, | |
| "grad_norm": 1.6162638664245605, | |
| "learning_rate": 0.0001009, | |
| "loss": 22.1558, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0011035313001605137, | |
| "grad_norm": 1.559161901473999, | |
| "learning_rate": 0.00010036894736842106, | |
| "loss": 22.1244, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.001203852327447833, | |
| "grad_norm": 1.6010931730270386, | |
| "learning_rate": 9.98378947368421e-05, | |
| "loss": 22.0848, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0013041733547351525, | |
| "grad_norm": 1.7077960968017578, | |
| "learning_rate": 9.930684210526315e-05, | |
| "loss": 22.0494, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0014044943820224719, | |
| "grad_norm": 1.5489342212677002, | |
| "learning_rate": 9.877578947368421e-05, | |
| "loss": 22.0367, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0015048154093097914, | |
| "grad_norm": 1.1338545083999634, | |
| "learning_rate": 9.824473684210527e-05, | |
| "loss": 22.0596, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0016051364365971107, | |
| "grad_norm": 1.444381833076477, | |
| "learning_rate": 9.771368421052632e-05, | |
| "loss": 21.9966, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0017054574638844303, | |
| "grad_norm": 1.4157575368881226, | |
| "learning_rate": 9.718263157894736e-05, | |
| "loss": 21.9681, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0018057784911717496, | |
| "grad_norm": 1.5229800939559937, | |
| "learning_rate": 9.665157894736842e-05, | |
| "loss": 21.9163, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.001906099518459069, | |
| "grad_norm": 1.5339312553405762, | |
| "learning_rate": 9.612052631578948e-05, | |
| "loss": 21.8712, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0020064205457463883, | |
| "grad_norm": 1.4681302309036255, | |
| "learning_rate": 9.558947368421052e-05, | |
| "loss": 21.9278, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.002106741573033708, | |
| "grad_norm": 1.7636168003082275, | |
| "learning_rate": 9.505842105263159e-05, | |
| "loss": 21.7252, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0022070626003210273, | |
| "grad_norm": 1.4351588487625122, | |
| "learning_rate": 9.452736842105263e-05, | |
| "loss": 21.7812, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.002307383627608347, | |
| "grad_norm": 1.4266163110733032, | |
| "learning_rate": 9.399631578947368e-05, | |
| "loss": 21.767, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.002407704654895666, | |
| "grad_norm": 1.3205598592758179, | |
| "learning_rate": 9.346526315789474e-05, | |
| "loss": 21.742, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0025080256821829855, | |
| "grad_norm": 1.32651948928833, | |
| "learning_rate": 9.293421052631578e-05, | |
| "loss": 21.7368, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002608346709470305, | |
| "grad_norm": 1.4364937543869019, | |
| "learning_rate": 9.240315789473684e-05, | |
| "loss": 21.647, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.002708667736757624, | |
| "grad_norm": 1.3655853271484375, | |
| "learning_rate": 9.18721052631579e-05, | |
| "loss": 21.644, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0028089887640449437, | |
| "grad_norm": 1.3614681959152222, | |
| "learning_rate": 9.134105263157895e-05, | |
| "loss": 21.5887, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0029093097913322633, | |
| "grad_norm": 1.360560655593872, | |
| "learning_rate": 9.081e-05, | |
| "loss": 21.555, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.003009630818619583, | |
| "grad_norm": 1.2450507879257202, | |
| "learning_rate": 9.027894736842105e-05, | |
| "loss": 21.5753, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.003109951845906902, | |
| "grad_norm": 1.2793240547180176, | |
| "learning_rate": 8.97478947368421e-05, | |
| "loss": 21.5722, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0032102728731942215, | |
| "grad_norm": 1.0861470699310303, | |
| "learning_rate": 8.921684210526316e-05, | |
| "loss": 21.5705, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.003310593900481541, | |
| "grad_norm": 1.2511563301086426, | |
| "learning_rate": 8.86857894736842e-05, | |
| "loss": 21.4441, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0034109149277688606, | |
| "grad_norm": 0.9954004287719727, | |
| "learning_rate": 8.815473684210527e-05, | |
| "loss": 21.6331, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0035112359550561797, | |
| "grad_norm": 1.0669211149215698, | |
| "learning_rate": 8.762368421052631e-05, | |
| "loss": 21.4872, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0036115569823434992, | |
| "grad_norm": 1.0056594610214233, | |
| "learning_rate": 8.709263157894737e-05, | |
| "loss": 21.4733, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0037118780096308188, | |
| "grad_norm": 1.016147255897522, | |
| "learning_rate": 8.656157894736843e-05, | |
| "loss": 21.4347, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.003812199036918138, | |
| "grad_norm": 1.1539530754089355, | |
| "learning_rate": 8.603052631578947e-05, | |
| "loss": 21.3633, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.003912520064205458, | |
| "grad_norm": 0.9206962585449219, | |
| "learning_rate": 8.549947368421052e-05, | |
| "loss": 21.4637, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0040128410914927765, | |
| "grad_norm": 0.833474338054657, | |
| "learning_rate": 8.496842105263158e-05, | |
| "loss": 21.4302, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004113162118780096, | |
| "grad_norm": 0.8578913807868958, | |
| "learning_rate": 8.443736842105264e-05, | |
| "loss": 21.4463, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.004213483146067416, | |
| "grad_norm": 1.1002817153930664, | |
| "learning_rate": 8.390631578947369e-05, | |
| "loss": 21.2379, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.004313804173354735, | |
| "grad_norm": 0.831058144569397, | |
| "learning_rate": 8.337526315789473e-05, | |
| "loss": 21.4246, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004414125200642055, | |
| "grad_norm": 0.7859891057014465, | |
| "learning_rate": 8.284421052631579e-05, | |
| "loss": 21.318, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.004514446227929374, | |
| "grad_norm": 0.7401735782623291, | |
| "learning_rate": 8.231315789473685e-05, | |
| "loss": 21.3669, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.004614767255216694, | |
| "grad_norm": 0.7849454879760742, | |
| "learning_rate": 8.178210526315789e-05, | |
| "loss": 21.3329, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0047150882825040125, | |
| "grad_norm": 0.6906037330627441, | |
| "learning_rate": 8.125105263157894e-05, | |
| "loss": 21.6264, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.004815409309791332, | |
| "grad_norm": 0.7339415550231934, | |
| "learning_rate": 8.072e-05, | |
| "loss": 21.5479, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0049157303370786515, | |
| "grad_norm": 0.7654588222503662, | |
| "learning_rate": 8.018894736842106e-05, | |
| "loss": 21.4293, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.005016051364365971, | |
| "grad_norm": 0.6686012148857117, | |
| "learning_rate": 7.965789473684211e-05, | |
| "loss": 21.401, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005016051364365971, | |
| "eval_loss": 10.654853820800781, | |
| "eval_runtime": 70.5622, | |
| "eval_samples_per_second": 59.494, | |
| "eval_steps_per_second": 14.88, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005116372391653291, | |
| "grad_norm": 0.6304632425308228, | |
| "learning_rate": 7.912684210526315e-05, | |
| "loss": 21.4006, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.00521669341894061, | |
| "grad_norm": 0.8711794018745422, | |
| "learning_rate": 7.859578947368421e-05, | |
| "loss": 21.125, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.00531701444622793, | |
| "grad_norm": 0.6818077564239502, | |
| "learning_rate": 7.806473684210527e-05, | |
| "loss": 21.2897, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005417335473515248, | |
| "grad_norm": 0.7950479388237, | |
| "learning_rate": 7.753368421052631e-05, | |
| "loss": 21.1913, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.005517656500802568, | |
| "grad_norm": 0.6777336597442627, | |
| "learning_rate": 7.700263157894738e-05, | |
| "loss": 21.3268, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0056179775280898875, | |
| "grad_norm": 0.9595808386802673, | |
| "learning_rate": 7.647157894736842e-05, | |
| "loss": 21.0679, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.005718298555377207, | |
| "grad_norm": 0.7060173153877258, | |
| "learning_rate": 7.594052631578948e-05, | |
| "loss": 21.2321, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.005818619582664527, | |
| "grad_norm": 0.6887240409851074, | |
| "learning_rate": 7.540947368421053e-05, | |
| "loss": 21.1762, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.005918940609951846, | |
| "grad_norm": 0.6797646880149841, | |
| "learning_rate": 7.487842105263157e-05, | |
| "loss": 21.3383, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.006019261637239166, | |
| "grad_norm": 0.7469924092292786, | |
| "learning_rate": 7.434736842105263e-05, | |
| "loss": 21.1953, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.006119582664526484, | |
| "grad_norm": 0.8638947606086731, | |
| "learning_rate": 7.381631578947368e-05, | |
| "loss": 21.1223, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.006219903691813804, | |
| "grad_norm": 0.7506782412528992, | |
| "learning_rate": 7.328526315789474e-05, | |
| "loss": 21.1332, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.006320224719101123, | |
| "grad_norm": 0.7623918652534485, | |
| "learning_rate": 7.27542105263158e-05, | |
| "loss": 21.2479, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.006420545746388443, | |
| "grad_norm": 0.6180063486099243, | |
| "learning_rate": 7.222315789473684e-05, | |
| "loss": 21.2955, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0065208667736757625, | |
| "grad_norm": 0.7025743722915649, | |
| "learning_rate": 7.16921052631579e-05, | |
| "loss": 21.0339, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006621187800963082, | |
| "grad_norm": 0.6287188529968262, | |
| "learning_rate": 7.116105263157895e-05, | |
| "loss": 21.2094, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.006721508828250402, | |
| "grad_norm": 0.6974558234214783, | |
| "learning_rate": 7.062999999999999e-05, | |
| "loss": 21.4861, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.006821829855537721, | |
| "grad_norm": 0.6953954696655273, | |
| "learning_rate": 7.009894736842106e-05, | |
| "loss": 21.3575, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.00692215088282504, | |
| "grad_norm": 0.7836357951164246, | |
| "learning_rate": 6.95678947368421e-05, | |
| "loss": 21.5824, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.007022471910112359, | |
| "grad_norm": 0.5490475296974182, | |
| "learning_rate": 6.903684210526316e-05, | |
| "loss": 21.3933, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.007122792937399679, | |
| "grad_norm": 0.5662202835083008, | |
| "learning_rate": 6.850578947368422e-05, | |
| "loss": 21.1143, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0072231139646869984, | |
| "grad_norm": 0.6373079419136047, | |
| "learning_rate": 6.797473684210526e-05, | |
| "loss": 21.1235, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.007323434991974318, | |
| "grad_norm": 0.7429666519165039, | |
| "learning_rate": 6.744368421052631e-05, | |
| "loss": 21.1772, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0074237560192616375, | |
| "grad_norm": 0.5452073812484741, | |
| "learning_rate": 6.691263157894736e-05, | |
| "loss": 21.2562, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.007524077046548957, | |
| "grad_norm": 0.5880535244941711, | |
| "learning_rate": 6.638157894736843e-05, | |
| "loss": 21.2677, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.007624398073836276, | |
| "grad_norm": 1.1386655569076538, | |
| "learning_rate": 6.585052631578948e-05, | |
| "loss": 21.0751, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.007724719101123595, | |
| "grad_norm": 0.6015380024909973, | |
| "learning_rate": 6.531947368421052e-05, | |
| "loss": 21.2662, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.007825040128410916, | |
| "grad_norm": 0.6043453216552734, | |
| "learning_rate": 6.478842105263158e-05, | |
| "loss": 21.0845, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.007925361155698234, | |
| "grad_norm": 0.715646505355835, | |
| "learning_rate": 6.425736842105264e-05, | |
| "loss": 21.035, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.008025682182985553, | |
| "grad_norm": 0.4522017538547516, | |
| "learning_rate": 6.372631578947368e-05, | |
| "loss": 21.2421, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.008126003210272873, | |
| "grad_norm": 0.591495156288147, | |
| "learning_rate": 6.319526315789473e-05, | |
| "loss": 21.2218, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.008226324237560192, | |
| "grad_norm": 0.8266862034797668, | |
| "learning_rate": 6.266421052631579e-05, | |
| "loss": 21.3115, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.008326645264847513, | |
| "grad_norm": 0.5965524911880493, | |
| "learning_rate": 6.213315789473685e-05, | |
| "loss": 21.1308, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.008426966292134831, | |
| "grad_norm": 0.5218135118484497, | |
| "learning_rate": 6.16021052631579e-05, | |
| "loss": 21.2285, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.008527287319422152, | |
| "grad_norm": 0.6091591715812683, | |
| "learning_rate": 6.107105263157894e-05, | |
| "loss": 21.2382, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.00862760834670947, | |
| "grad_norm": 0.5717049241065979, | |
| "learning_rate": 6.054e-05, | |
| "loss": 21.3019, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.008727929373996789, | |
| "grad_norm": 0.49166449904441833, | |
| "learning_rate": 6.000894736842105e-05, | |
| "loss": 21.1254, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.00882825040128411, | |
| "grad_norm": 0.6972517967224121, | |
| "learning_rate": 5.94778947368421e-05, | |
| "loss": 21.1412, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.008928571428571428, | |
| "grad_norm": 0.508397102355957, | |
| "learning_rate": 5.894684210526316e-05, | |
| "loss": 21.1448, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.009028892455858748, | |
| "grad_norm": 0.5189328193664551, | |
| "learning_rate": 5.841578947368421e-05, | |
| "loss": 21.3218, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.009129213483146067, | |
| "grad_norm": 0.5766506195068359, | |
| "learning_rate": 5.7884736842105265e-05, | |
| "loss": 21.1138, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.009229534510433388, | |
| "grad_norm": 0.4988974630832672, | |
| "learning_rate": 5.7353684210526314e-05, | |
| "loss": 21.0883, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.009329855537720706, | |
| "grad_norm": 0.6053217053413391, | |
| "learning_rate": 5.6822631578947364e-05, | |
| "loss": 21.1423, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.009430176565008025, | |
| "grad_norm": 0.6934704184532166, | |
| "learning_rate": 5.629157894736842e-05, | |
| "loss": 21.2583, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.009530497592295345, | |
| "grad_norm": 0.5622691512107849, | |
| "learning_rate": 5.576052631578948e-05, | |
| "loss": 21.2342, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.009630818619582664, | |
| "grad_norm": 0.5352026224136353, | |
| "learning_rate": 5.522947368421053e-05, | |
| "loss": 21.1231, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.009731139646869984, | |
| "grad_norm": 0.4841010868549347, | |
| "learning_rate": 5.469842105263158e-05, | |
| "loss": 21.1777, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.009831460674157303, | |
| "grad_norm": 0.5576900839805603, | |
| "learning_rate": 5.416736842105263e-05, | |
| "loss": 21.1561, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.009931781701444624, | |
| "grad_norm": 0.5335264205932617, | |
| "learning_rate": 5.3636315789473685e-05, | |
| "loss": 21.2953, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.010032102728731942, | |
| "grad_norm": 0.5405234098434448, | |
| "learning_rate": 5.3105263157894734e-05, | |
| "loss": 21.3645, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.010032102728731942, | |
| "eval_loss": 10.584715843200684, | |
| "eval_runtime": 70.5882, | |
| "eval_samples_per_second": 59.472, | |
| "eval_steps_per_second": 14.875, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01013242375601926, | |
| "grad_norm": 0.6677088141441345, | |
| "learning_rate": 5.257421052631578e-05, | |
| "loss": 21.1299, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.010232744783306581, | |
| "grad_norm": 0.5807069540023804, | |
| "learning_rate": 5.2043157894736846e-05, | |
| "loss": 21.0427, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0103330658105939, | |
| "grad_norm": 0.501835823059082, | |
| "learning_rate": 5.1512105263157895e-05, | |
| "loss": 21.1984, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.01043338683788122, | |
| "grad_norm": 0.7399368286132812, | |
| "learning_rate": 5.098105263157895e-05, | |
| "loss": 21.1552, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.010533707865168539, | |
| "grad_norm": 0.8803501725196838, | |
| "learning_rate": 5.045e-05, | |
| "loss": 20.9362, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.01063402889245586, | |
| "grad_norm": 0.6539490818977356, | |
| "learning_rate": 4.991894736842105e-05, | |
| "loss": 21.3907, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.010734349919743178, | |
| "grad_norm": 0.8279363512992859, | |
| "learning_rate": 4.9387894736842105e-05, | |
| "loss": 21.1683, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.010834670947030497, | |
| "grad_norm": 0.7174587845802307, | |
| "learning_rate": 4.885684210526316e-05, | |
| "loss": 21.0967, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.010934991974317817, | |
| "grad_norm": 0.5338941216468811, | |
| "learning_rate": 4.832578947368421e-05, | |
| "loss": 21.2763, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.011035313001605136, | |
| "grad_norm": 0.5613208413124084, | |
| "learning_rate": 4.779473684210526e-05, | |
| "loss": 21.3178, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.011135634028892456, | |
| "grad_norm": 0.5921754240989685, | |
| "learning_rate": 4.7263684210526315e-05, | |
| "loss": 21.1902, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.011235955056179775, | |
| "grad_norm": 0.6494777798652649, | |
| "learning_rate": 4.673263157894737e-05, | |
| "loss": 21.1112, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.011336276083467095, | |
| "grad_norm": 0.6433371901512146, | |
| "learning_rate": 4.620157894736842e-05, | |
| "loss": 21.1761, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.011436597110754414, | |
| "grad_norm": 0.915080726146698, | |
| "learning_rate": 4.5670526315789475e-05, | |
| "loss": 21.0194, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.011536918138041733, | |
| "grad_norm": 1.0387072563171387, | |
| "learning_rate": 4.5139473684210524e-05, | |
| "loss": 20.9303, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.011637239165329053, | |
| "grad_norm": 0.6426976919174194, | |
| "learning_rate": 4.460842105263158e-05, | |
| "loss": 21.3885, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.011737560192616372, | |
| "grad_norm": 0.923250138759613, | |
| "learning_rate": 4.4077368421052636e-05, | |
| "loss": 20.8614, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.011837881219903692, | |
| "grad_norm": 0.5725429058074951, | |
| "learning_rate": 4.3546315789473685e-05, | |
| "loss": 21.1852, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.011938202247191011, | |
| "grad_norm": 0.656847357749939, | |
| "learning_rate": 4.3015263157894734e-05, | |
| "loss": 21.3078, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.012038523274478331, | |
| "grad_norm": 0.5193572640419006, | |
| "learning_rate": 4.248421052631579e-05, | |
| "loss": 21.1464, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01213884430176565, | |
| "grad_norm": 0.7125611901283264, | |
| "learning_rate": 4.1953157894736846e-05, | |
| "loss": 21.1635, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.012239165329052969, | |
| "grad_norm": 0.523034930229187, | |
| "learning_rate": 4.1422105263157895e-05, | |
| "loss": 21.1683, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.012339486356340289, | |
| "grad_norm": 0.540019154548645, | |
| "learning_rate": 4.0891052631578944e-05, | |
| "loss": 21.0405, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.012439807383627608, | |
| "grad_norm": 0.6048979759216309, | |
| "learning_rate": 4.036e-05, | |
| "loss": 20.9969, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.012540128410914928, | |
| "grad_norm": 0.6617588996887207, | |
| "learning_rate": 3.9828947368421056e-05, | |
| "loss": 20.9615, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.012640449438202247, | |
| "grad_norm": 0.5792971253395081, | |
| "learning_rate": 3.9297894736842105e-05, | |
| "loss": 21.0814, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.012740770465489567, | |
| "grad_norm": 0.8640639185905457, | |
| "learning_rate": 3.8766842105263154e-05, | |
| "loss": 20.9608, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.012841091492776886, | |
| "grad_norm": 0.6493314504623413, | |
| "learning_rate": 3.823578947368421e-05, | |
| "loss": 21.0183, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.012941412520064205, | |
| "grad_norm": 0.5835341811180115, | |
| "learning_rate": 3.7704736842105265e-05, | |
| "loss": 21.247, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.013041733547351525, | |
| "grad_norm": 0.5780165791511536, | |
| "learning_rate": 3.7173684210526315e-05, | |
| "loss": 21.1309, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.013142054574638844, | |
| "grad_norm": 0.603012204170227, | |
| "learning_rate": 3.664263157894737e-05, | |
| "loss": 21.105, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.013242375601926164, | |
| "grad_norm": 0.5957277417182922, | |
| "learning_rate": 3.611157894736842e-05, | |
| "loss": 21.1341, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.013342696629213483, | |
| "grad_norm": 0.5657469034194946, | |
| "learning_rate": 3.5580526315789475e-05, | |
| "loss": 21.2018, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.013443017656500803, | |
| "grad_norm": 0.5668490529060364, | |
| "learning_rate": 3.504947368421053e-05, | |
| "loss": 21.2215, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.013543338683788122, | |
| "grad_norm": 0.6400595903396606, | |
| "learning_rate": 3.451842105263158e-05, | |
| "loss": 21.1004, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.013643659711075442, | |
| "grad_norm": 0.5542194247245789, | |
| "learning_rate": 3.398736842105263e-05, | |
| "loss": 21.1221, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.013743980738362761, | |
| "grad_norm": 0.5827724933624268, | |
| "learning_rate": 3.345631578947368e-05, | |
| "loss": 21.1213, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.01384430176565008, | |
| "grad_norm": 0.5147905945777893, | |
| "learning_rate": 3.292526315789474e-05, | |
| "loss": 21.1524, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0139446227929374, | |
| "grad_norm": 0.6220738291740417, | |
| "learning_rate": 3.239421052631579e-05, | |
| "loss": 21.0253, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.014044943820224719, | |
| "grad_norm": 0.6777515411376953, | |
| "learning_rate": 3.186315789473684e-05, | |
| "loss": 21.3317, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.014145264847512039, | |
| "grad_norm": 0.6565226912498474, | |
| "learning_rate": 3.1332105263157895e-05, | |
| "loss": 20.9974, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.014245585874799358, | |
| "grad_norm": 0.6572129130363464, | |
| "learning_rate": 3.080105263157895e-05, | |
| "loss": 21.1355, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.014345906902086678, | |
| "grad_norm": 0.6094454526901245, | |
| "learning_rate": 3.027e-05, | |
| "loss": 21.1024, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.014446227929373997, | |
| "grad_norm": 0.5109902024269104, | |
| "learning_rate": 2.973894736842105e-05, | |
| "loss": 21.224, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.014546548956661316, | |
| "grad_norm": 0.7738908529281616, | |
| "learning_rate": 2.9207894736842105e-05, | |
| "loss": 21.4256, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.014646869983948636, | |
| "grad_norm": 0.5986607670783997, | |
| "learning_rate": 2.8676842105263157e-05, | |
| "loss": 21.2798, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.014747191011235955, | |
| "grad_norm": 0.6626487970352173, | |
| "learning_rate": 2.814578947368421e-05, | |
| "loss": 21.1379, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.014847512038523275, | |
| "grad_norm": 0.6779626607894897, | |
| "learning_rate": 2.7614736842105266e-05, | |
| "loss": 21.0229, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.014947833065810594, | |
| "grad_norm": 0.6698882579803467, | |
| "learning_rate": 2.7083684210526315e-05, | |
| "loss": 21.0818, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.015048154093097914, | |
| "grad_norm": 0.8434391021728516, | |
| "learning_rate": 2.6552631578947367e-05, | |
| "loss": 21.5482, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.015048154093097914, | |
| "eval_loss": 10.558324813842773, | |
| "eval_runtime": 70.5565, | |
| "eval_samples_per_second": 59.498, | |
| "eval_steps_per_second": 14.882, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.015148475120385233, | |
| "grad_norm": 0.8424636721611023, | |
| "learning_rate": 2.6021578947368423e-05, | |
| "loss": 20.936, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.015248796147672551, | |
| "grad_norm": 0.5448256731033325, | |
| "learning_rate": 2.5490526315789475e-05, | |
| "loss": 21.1742, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.015349117174959872, | |
| "grad_norm": 0.6612237095832825, | |
| "learning_rate": 2.4959473684210524e-05, | |
| "loss": 21.0709, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.01544943820224719, | |
| "grad_norm": 0.550390899181366, | |
| "learning_rate": 2.442842105263158e-05, | |
| "loss": 20.8933, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.015549759229534511, | |
| "grad_norm": 0.6066597700119019, | |
| "learning_rate": 2.389736842105263e-05, | |
| "loss": 20.9679, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.01565008025682183, | |
| "grad_norm": 0.595045804977417, | |
| "learning_rate": 2.3366315789473685e-05, | |
| "loss": 21.0887, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.01575040128410915, | |
| "grad_norm": 0.6363713145256042, | |
| "learning_rate": 2.2835263157894738e-05, | |
| "loss": 21.3898, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.01585072231139647, | |
| "grad_norm": 0.5982836484909058, | |
| "learning_rate": 2.230421052631579e-05, | |
| "loss": 21.1671, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.015951043338683787, | |
| "grad_norm": 0.7963234782218933, | |
| "learning_rate": 2.1773157894736843e-05, | |
| "loss": 20.9716, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.016051364365971106, | |
| "grad_norm": 0.647216796875, | |
| "learning_rate": 2.1242105263157895e-05, | |
| "loss": 21.1608, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.016151685393258428, | |
| "grad_norm": 0.5022075772285461, | |
| "learning_rate": 2.0711052631578947e-05, | |
| "loss": 21.1819, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.016252006420545747, | |
| "grad_norm": 0.5094108581542969, | |
| "learning_rate": 2.018e-05, | |
| "loss": 21.1568, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.016352327447833066, | |
| "grad_norm": 0.5434950590133667, | |
| "learning_rate": 1.9648947368421052e-05, | |
| "loss": 20.9586, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.016452648475120384, | |
| "grad_norm": 0.6874385476112366, | |
| "learning_rate": 1.9117894736842105e-05, | |
| "loss": 21.0375, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.016552969502407703, | |
| "grad_norm": 0.49630945920944214, | |
| "learning_rate": 1.8586842105263157e-05, | |
| "loss": 21.0352, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.016653290529695025, | |
| "grad_norm": 0.6111531257629395, | |
| "learning_rate": 1.805578947368421e-05, | |
| "loss": 21.2822, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.016753611556982344, | |
| "grad_norm": 0.5392615795135498, | |
| "learning_rate": 1.7524736842105266e-05, | |
| "loss": 21.1524, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.016853932584269662, | |
| "grad_norm": 0.5594942569732666, | |
| "learning_rate": 1.6993684210526315e-05, | |
| "loss": 21.1931, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.01695425361155698, | |
| "grad_norm": 0.5756310224533081, | |
| "learning_rate": 1.646263157894737e-05, | |
| "loss": 21.2089, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.017054574638844303, | |
| "grad_norm": 0.783043622970581, | |
| "learning_rate": 1.593157894736842e-05, | |
| "loss": 20.8715, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.017154895666131622, | |
| "grad_norm": 0.4898316562175751, | |
| "learning_rate": 1.5400526315789475e-05, | |
| "loss": 21.1136, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.01725521669341894, | |
| "grad_norm": 0.6354514360427856, | |
| "learning_rate": 1.4869473684210524e-05, | |
| "loss": 21.089, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.01735553772070626, | |
| "grad_norm": 0.6032927632331848, | |
| "learning_rate": 1.4338421052631579e-05, | |
| "loss": 21.0936, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.017455858747993578, | |
| "grad_norm": 0.6571072936058044, | |
| "learning_rate": 1.3807368421052633e-05, | |
| "loss": 20.9723, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0175561797752809, | |
| "grad_norm": 0.5904538035392761, | |
| "learning_rate": 1.3276315789473684e-05, | |
| "loss": 21.287, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.01765650080256822, | |
| "grad_norm": 0.5847862362861633, | |
| "learning_rate": 1.2745263157894738e-05, | |
| "loss": 21.1966, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.017756821829855537, | |
| "grad_norm": 0.5706862211227417, | |
| "learning_rate": 1.221421052631579e-05, | |
| "loss": 20.9706, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.017857142857142856, | |
| "grad_norm": 0.5410795211791992, | |
| "learning_rate": 1.1683157894736843e-05, | |
| "loss": 21.2563, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.01795746388443018, | |
| "grad_norm": 0.5394900441169739, | |
| "learning_rate": 1.1152105263157895e-05, | |
| "loss": 20.9992, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.018057784911717497, | |
| "grad_norm": 0.45912498235702515, | |
| "learning_rate": 1.0621052631578948e-05, | |
| "loss": 21.2137, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.018158105939004816, | |
| "grad_norm": 0.5860676765441895, | |
| "learning_rate": 1.009e-05, | |
| "loss": 21.1744, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.018258426966292134, | |
| "grad_norm": 0.7394751310348511, | |
| "learning_rate": 9.558947368421052e-06, | |
| "loss": 20.9416, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.018358747993579453, | |
| "grad_norm": 0.6703020334243774, | |
| "learning_rate": 9.027894736842105e-06, | |
| "loss": 20.9801, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.018459069020866775, | |
| "grad_norm": 0.5259845852851868, | |
| "learning_rate": 8.496842105263157e-06, | |
| "loss": 21.0672, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.018559390048154094, | |
| "grad_norm": 0.47938376665115356, | |
| "learning_rate": 7.96578947368421e-06, | |
| "loss": 21.0379, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.018659711075441412, | |
| "grad_norm": 0.6665632128715515, | |
| "learning_rate": 7.434736842105262e-06, | |
| "loss": 21.3934, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.01876003210272873, | |
| "grad_norm": 0.6356412172317505, | |
| "learning_rate": 6.903684210526316e-06, | |
| "loss": 21.3995, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.01886035313001605, | |
| "grad_norm": 0.5531170964241028, | |
| "learning_rate": 6.372631578947369e-06, | |
| "loss": 21.177, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.018960674157303372, | |
| "grad_norm": 0.5264145731925964, | |
| "learning_rate": 5.841578947368421e-06, | |
| "loss": 21.2467, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.01906099518459069, | |
| "grad_norm": 0.5184823870658875, | |
| "learning_rate": 5.310526315789474e-06, | |
| "loss": 21.0768, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01916131621187801, | |
| "grad_norm": 0.601334810256958, | |
| "learning_rate": 4.779473684210526e-06, | |
| "loss": 21.0318, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.019261637239165328, | |
| "grad_norm": 0.6639525890350342, | |
| "learning_rate": 4.248421052631579e-06, | |
| "loss": 20.9996, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.01936195826645265, | |
| "grad_norm": 0.47377097606658936, | |
| "learning_rate": 3.717368421052631e-06, | |
| "loss": 21.0164, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.01946227929373997, | |
| "grad_norm": 0.6908702254295349, | |
| "learning_rate": 3.1863157894736844e-06, | |
| "loss": 20.9247, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.019562600321027288, | |
| "grad_norm": 0.5369330048561096, | |
| "learning_rate": 2.655263157894737e-06, | |
| "loss": 21.3861, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.019662921348314606, | |
| "grad_norm": 0.6818935871124268, | |
| "learning_rate": 2.1242105263157893e-06, | |
| "loss": 21.0936, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.019763242375601925, | |
| "grad_norm": 0.6107151508331299, | |
| "learning_rate": 1.5931578947368422e-06, | |
| "loss": 21.0526, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.019863563402889247, | |
| "grad_norm": 0.6597663760185242, | |
| "learning_rate": 1.0621052631578947e-06, | |
| "loss": 21.1064, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.019963884430176566, | |
| "grad_norm": 0.6686668992042542, | |
| "learning_rate": 5.310526315789473e-07, | |
| "loss": 21.0729, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.020064205457463884, | |
| "grad_norm": 0.7321626543998718, | |
| "learning_rate": 0.0, | |
| "loss": 21.4709, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.020064205457463884, | |
| "eval_loss": 10.55037784576416, | |
| "eval_runtime": 70.588, | |
| "eval_samples_per_second": 59.472, | |
| "eval_steps_per_second": 14.875, | |
| "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": 11943700070400.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |