| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.7568590350047304, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003784295175023652, | |
| "grad_norm": 0.00010429308895254508, | |
| "learning_rate": 1e-05, | |
| "loss": 46.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.003784295175023652, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.9721, | |
| "eval_samples_per_second": 225.643, | |
| "eval_steps_per_second": 56.791, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007568590350047304, | |
| "grad_norm": 0.00010511261643841863, | |
| "learning_rate": 2e-05, | |
| "loss": 46.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.011352885525070956, | |
| "grad_norm": 8.239266026066616e-05, | |
| "learning_rate": 3e-05, | |
| "loss": 46.0, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.015137180700094607, | |
| "grad_norm": 0.0001124097325373441, | |
| "learning_rate": 4e-05, | |
| "loss": 46.0, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01892147587511826, | |
| "grad_norm": 0.00012312972103245556, | |
| "learning_rate": 5e-05, | |
| "loss": 46.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02270577105014191, | |
| "grad_norm": 0.00012848418555222452, | |
| "learning_rate": 6e-05, | |
| "loss": 46.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.026490066225165563, | |
| "grad_norm": 0.00011386930418666452, | |
| "learning_rate": 7e-05, | |
| "loss": 46.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.030274361400189215, | |
| "grad_norm": 7.296714466065168e-05, | |
| "learning_rate": 8e-05, | |
| "loss": 46.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03405865657521287, | |
| "grad_norm": 9.800998668652028e-05, | |
| "learning_rate": 9e-05, | |
| "loss": 46.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03784295175023652, | |
| "grad_norm": 0.00011683226330205798, | |
| "learning_rate": 0.0001, | |
| "loss": 46.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.041627246925260174, | |
| "grad_norm": 0.0001516055199317634, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 46.0, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04541154210028382, | |
| "grad_norm": 0.00014346589159686118, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 46.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04919583727530748, | |
| "grad_norm": 0.00014646467752754688, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 46.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.052980132450331126, | |
| "grad_norm": 0.0001113307080231607, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 46.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05676442762535478, | |
| "grad_norm": 0.00017336613382212818, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 46.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06054872280037843, | |
| "grad_norm": 0.00017019949154928327, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 46.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06433301797540208, | |
| "grad_norm": 0.00017636186385061592, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 46.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06811731315042574, | |
| "grad_norm": 0.0001949941215571016, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 46.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07190160832544938, | |
| "grad_norm": 0.0002444561687298119, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 46.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.07568590350047304, | |
| "grad_norm": 0.000181497962330468, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 46.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07947019867549669, | |
| "grad_norm": 0.00017078036034945399, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 46.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08325449385052035, | |
| "grad_norm": 0.00024027511244639754, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 46.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08703878902554399, | |
| "grad_norm": 0.00020458181097637862, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 46.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.09082308420056764, | |
| "grad_norm": 0.0002451995969749987, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 46.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0946073793755913, | |
| "grad_norm": 0.00024543306790292263, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 46.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09839167455061495, | |
| "grad_norm": 0.00018932469538412988, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 46.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.1021759697256386, | |
| "grad_norm": 0.00022809104120824486, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 46.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.10596026490066225, | |
| "grad_norm": 0.0002928116882685572, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 46.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1097445600756859, | |
| "grad_norm": 0.0002065580483758822, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 46.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.11352885525070956, | |
| "grad_norm": 0.00033147100475616753, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 46.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1173131504257332, | |
| "grad_norm": 0.00023922721447888762, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 46.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.12109744560075686, | |
| "grad_norm": 0.0003320495889056474, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 46.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.12488174077578051, | |
| "grad_norm": 0.000362045771908015, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 46.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.12866603595080417, | |
| "grad_norm": 0.0003298628726042807, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 46.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.13245033112582782, | |
| "grad_norm": 0.00023832467559259385, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 46.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.13623462630085148, | |
| "grad_norm": 0.00034446705831214786, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 46.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1400189214758751, | |
| "grad_norm": 0.0003199601487722248, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 46.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.14380321665089876, | |
| "grad_norm": 0.00027099723229184747, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 46.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.14758751182592242, | |
| "grad_norm": 0.0002793792518787086, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 46.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.15137180700094607, | |
| "grad_norm": 0.0003409314958844334, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 46.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.15515610217596973, | |
| "grad_norm": 0.000397098163375631, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 46.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.15894039735099338, | |
| "grad_norm": 0.00035342099727131426, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 46.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.16272469252601704, | |
| "grad_norm": 0.00039373149047605693, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 46.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.1665089877010407, | |
| "grad_norm": 0.00041224624146707356, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 46.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.17029328287606432, | |
| "grad_norm": 0.00032645504688844085, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 46.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.17407757805108798, | |
| "grad_norm": 0.00033724671811796725, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 46.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.17786187322611163, | |
| "grad_norm": 0.00038492504972964525, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 46.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1816461684011353, | |
| "grad_norm": 0.0004222445422783494, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 46.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.18543046357615894, | |
| "grad_norm": 0.00044694452662952244, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 46.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1892147587511826, | |
| "grad_norm": 0.0004379482415970415, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 46.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1892147587511826, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.9787, | |
| "eval_samples_per_second": 224.898, | |
| "eval_steps_per_second": 56.604, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.19299905392620625, | |
| "grad_norm": 0.00043849743087776005, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 46.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.1967833491012299, | |
| "grad_norm": 0.0005049917963333428, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 46.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.20056764427625354, | |
| "grad_norm": 0.0004644330474548042, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 46.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.2043519394512772, | |
| "grad_norm": 0.0004966144915670156, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 46.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.20813623462630085, | |
| "grad_norm": 0.0005559484125114977, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 46.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.2119205298013245, | |
| "grad_norm": 0.0006866741459816694, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 46.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.21570482497634816, | |
| "grad_norm": 0.000611191731877625, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 46.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.2194891201513718, | |
| "grad_norm": 0.000491699087433517, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 46.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.22327341532639547, | |
| "grad_norm": 0.0004499013884924352, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 46.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.22705771050141912, | |
| "grad_norm": 0.00040422569145448506, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 46.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.23084200567644275, | |
| "grad_norm": 0.0005719454493373632, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 46.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.2346263008514664, | |
| "grad_norm": 0.0006503518670797348, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 46.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.23841059602649006, | |
| "grad_norm": 0.0008147732587531209, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 46.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.24219489120151372, | |
| "grad_norm": 0.0007560556405223906, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 46.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.24597918637653737, | |
| "grad_norm": 0.000697283772751689, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 46.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.24976348155156103, | |
| "grad_norm": 0.0007711402722634375, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 46.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2535477767265847, | |
| "grad_norm": 0.0009076902642846107, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 46.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.25733207190160834, | |
| "grad_norm": 0.0008205681224353611, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 46.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.261116367076632, | |
| "grad_norm": 0.0007467827526852489, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 46.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.26490066225165565, | |
| "grad_norm": 0.0007995216874405742, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 46.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2686849574266793, | |
| "grad_norm": 0.000726785627193749, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 46.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.27246925260170296, | |
| "grad_norm": 0.0007780804298818111, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 46.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.27625354777672656, | |
| "grad_norm": 0.0007564228726550937, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 46.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2800378429517502, | |
| "grad_norm": 0.0006644019740633667, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 46.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.28382213812677387, | |
| "grad_norm": 0.0009467531344853342, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 46.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2876064333017975, | |
| "grad_norm": 0.0009459892171435058, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 46.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2913907284768212, | |
| "grad_norm": 0.0008425005944445729, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 46.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.29517502365184484, | |
| "grad_norm": 0.0007187195587903261, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 46.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2989593188268685, | |
| "grad_norm": 0.0008075591176748276, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 46.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.30274361400189215, | |
| "grad_norm": 0.0007806776557117701, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 46.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3065279091769158, | |
| "grad_norm": 0.0007592571200802922, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 46.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.31031220435193946, | |
| "grad_norm": 0.0006820546113885939, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 46.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3140964995269631, | |
| "grad_norm": 0.0007549600559286773, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 46.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.31788079470198677, | |
| "grad_norm": 0.0009110699757002294, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 46.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3216650898770104, | |
| "grad_norm": 0.0005815060576424003, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 46.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3254493850520341, | |
| "grad_norm": 0.0007678881520405412, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 46.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.32923368022705773, | |
| "grad_norm": 0.0007318141870200634, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 46.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3330179754020814, | |
| "grad_norm": 0.0007528024143539369, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 46.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.336802270577105, | |
| "grad_norm": 0.0007590457680635154, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 46.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.34058656575212864, | |
| "grad_norm": 0.0007292851223610342, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 46.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3443708609271523, | |
| "grad_norm": 0.0007912701112218201, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 46.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.34815515610217596, | |
| "grad_norm": 0.0007180406246334314, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 46.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3519394512771996, | |
| "grad_norm": 0.0008794911555014551, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 46.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.35572374645222327, | |
| "grad_norm": 0.0007236977107822895, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 46.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.3595080416272469, | |
| "grad_norm": 0.0009709982550702989, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 46.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3632923368022706, | |
| "grad_norm": 0.0008353631128557026, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 46.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.36707663197729423, | |
| "grad_norm": 0.0008676600991748273, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 46.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.3708609271523179, | |
| "grad_norm": 0.0008874779450707138, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 46.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.37464522232734154, | |
| "grad_norm": 0.000924816180486232, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 46.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.3784295175023652, | |
| "grad_norm": 0.00097953830845654, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 46.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3784295175023652, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.9694, | |
| "eval_samples_per_second": 225.955, | |
| "eval_steps_per_second": 56.87, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.38221381267738885, | |
| "grad_norm": 0.0010403641499578953, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 46.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.3859981078524125, | |
| "grad_norm": 0.0011233292752876878, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 46.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.38978240302743616, | |
| "grad_norm": 0.001115982187911868, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 46.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.3935666982024598, | |
| "grad_norm": 0.001295563648454845, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 46.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.3973509933774834, | |
| "grad_norm": 0.0010806728387251496, | |
| "learning_rate": 5e-05, | |
| "loss": 46.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.4011352885525071, | |
| "grad_norm": 0.0011615157127380371, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 46.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.40491958372753073, | |
| "grad_norm": 0.000886109541170299, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 46.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.4087038789025544, | |
| "grad_norm": 0.0007965463446453214, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 46.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.41248817407757804, | |
| "grad_norm": 0.0006533803534694016, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 46.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.4162724692526017, | |
| "grad_norm": 0.0010823339689522982, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 46.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.42005676442762535, | |
| "grad_norm": 0.0012081189779564738, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 46.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.423841059602649, | |
| "grad_norm": 0.0010268923360854387, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 46.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.42762535477767266, | |
| "grad_norm": 0.0013933356385678053, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 46.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.4314096499526963, | |
| "grad_norm": 0.001255908515304327, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 46.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.43519394512771997, | |
| "grad_norm": 0.001339130918495357, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 46.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.4389782403027436, | |
| "grad_norm": 0.0010802745819091797, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 46.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4427625354777673, | |
| "grad_norm": 0.001069769379682839, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 46.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.44654683065279094, | |
| "grad_norm": 0.001089852536097169, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 46.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.4503311258278146, | |
| "grad_norm": 0.0012072310782968998, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 46.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.45411542100283825, | |
| "grad_norm": 0.0008982583531178534, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 46.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.45789971617786185, | |
| "grad_norm": 0.0012891966616734862, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 46.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.4616840113528855, | |
| "grad_norm": 0.0008873147307895124, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 46.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.46546830652790916, | |
| "grad_norm": 0.0009927195496857166, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 46.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.4692526017029328, | |
| "grad_norm": 0.0012268994469195604, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 46.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.47303689687795647, | |
| "grad_norm": 0.001123241614550352, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 46.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.4768211920529801, | |
| "grad_norm": 0.0009914507390931249, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 46.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.4806054872280038, | |
| "grad_norm": 0.0009736844804137945, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 46.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.48438978240302744, | |
| "grad_norm": 0.0009503276087343693, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 46.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.4881740775780511, | |
| "grad_norm": 0.001040280214510858, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 46.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.49195837275307475, | |
| "grad_norm": 0.0009310917230322957, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 46.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.4957426679280984, | |
| "grad_norm": 0.0008663221378810704, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 46.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.49952696310312206, | |
| "grad_norm": 0.0011335160816088319, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 46.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5033112582781457, | |
| "grad_norm": 0.0009626140235923231, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 46.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5070955534531694, | |
| "grad_norm": 0.0009487202623859048, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 46.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.510879848628193, | |
| "grad_norm": 0.0009260333026759326, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 46.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5146641438032167, | |
| "grad_norm": 0.0010641305707395077, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 46.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5184484389782403, | |
| "grad_norm": 0.0009386951569467783, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 46.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.522232734153264, | |
| "grad_norm": 0.0009598695323802531, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 46.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5260170293282876, | |
| "grad_norm": 0.001204126514494419, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 46.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5298013245033113, | |
| "grad_norm": 0.001147297560237348, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 46.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.533585619678335, | |
| "grad_norm": 0.0009711734019219875, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 46.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5373699148533586, | |
| "grad_norm": 0.001283796038478613, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 46.0, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.5411542100283823, | |
| "grad_norm": 0.0012706165434792638, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 46.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5449385052034059, | |
| "grad_norm": 0.0008949569892138243, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 46.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5487228003784295, | |
| "grad_norm": 0.0010173728223890066, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 46.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5525070955534531, | |
| "grad_norm": 0.0012513319961726665, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 46.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5562913907284768, | |
| "grad_norm": 0.0009062588214874268, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 46.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.5600756859035004, | |
| "grad_norm": 0.0007536582998000085, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 46.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.5638599810785241, | |
| "grad_norm": 0.00113199173938483, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 46.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.5676442762535477, | |
| "grad_norm": 0.0011235671117901802, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 46.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5676442762535477, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.9729, | |
| "eval_samples_per_second": 225.551, | |
| "eval_steps_per_second": 56.768, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.0015914759133011103, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 46.0, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.575212866603595, | |
| "grad_norm": 0.0014077216619625688, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 46.0, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.5789971617786187, | |
| "grad_norm": 0.0012213258305564523, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 46.0, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.5827814569536424, | |
| "grad_norm": 0.0016379911685362458, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 46.0, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.586565752128666, | |
| "grad_norm": 0.001720768166705966, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 46.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.5903500473036897, | |
| "grad_norm": 0.001313543296419084, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 46.0, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.5941343424787133, | |
| "grad_norm": 0.0015578890452161431, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 46.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.597918637653737, | |
| "grad_norm": 0.0009314199560321867, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 46.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6017029328287606, | |
| "grad_norm": 0.0012688793940469623, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 46.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6054872280037843, | |
| "grad_norm": 0.0009555511060170829, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 46.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.609271523178808, | |
| "grad_norm": 0.0011067419545724988, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 46.0, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6130558183538316, | |
| "grad_norm": 0.001390297431498766, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 46.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.6168401135288553, | |
| "grad_norm": 0.0010950445430353284, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 46.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.6206244087038789, | |
| "grad_norm": 0.0016723081935197115, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 46.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.6244087038789026, | |
| "grad_norm": 0.0011174182873219252, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 46.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6281929990539262, | |
| "grad_norm": 0.001625697361305356, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 46.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6319772942289499, | |
| "grad_norm": 0.0013816162245348096, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 46.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.6357615894039735, | |
| "grad_norm": 0.0011888662120327353, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 46.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6395458845789972, | |
| "grad_norm": 0.0010107356356456876, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 46.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6433301797540208, | |
| "grad_norm": 0.0009662157390266657, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 46.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6471144749290445, | |
| "grad_norm": 0.000990124186500907, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 46.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6508987701040682, | |
| "grad_norm": 0.001491464558057487, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 46.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.6546830652790918, | |
| "grad_norm": 0.0011075662914663553, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 46.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.6584673604541155, | |
| "grad_norm": 0.0012964402558282018, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 46.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.6622516556291391, | |
| "grad_norm": 0.0009176530293188989, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 46.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.6660359508041628, | |
| "grad_norm": 0.0010724479798227549, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 46.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.6698202459791863, | |
| "grad_norm": 0.001172030926682055, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 46.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.67360454115421, | |
| "grad_norm": 0.0011369283311069012, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 46.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.6773888363292336, | |
| "grad_norm": 0.0013332292437553406, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 46.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.6811731315042573, | |
| "grad_norm": 0.0008293318096548319, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 46.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.684957426679281, | |
| "grad_norm": 0.0008097453392110765, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 46.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.6887417218543046, | |
| "grad_norm": 0.0008934435318224132, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 46.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.6925260170293283, | |
| "grad_norm": 0.001011797459796071, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 46.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.6963103122043519, | |
| "grad_norm": 0.0011690121609717607, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 46.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7000946073793756, | |
| "grad_norm": 0.0011907786829397082, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 46.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7038789025543992, | |
| "grad_norm": 0.0009520188323222101, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 46.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.7076631977294229, | |
| "grad_norm": 0.0008637032005935907, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 46.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.7114474929044465, | |
| "grad_norm": 0.0011205349583178759, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 46.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.7152317880794702, | |
| "grad_norm": 0.0008666692301630974, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 46.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.7190160832544938, | |
| "grad_norm": 0.000937802018597722, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 46.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.7228003784295175, | |
| "grad_norm": 0.000931553659029305, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 46.0, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.7265846736045412, | |
| "grad_norm": 0.0010818357113748789, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 46.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7303689687795648, | |
| "grad_norm": 0.0009485573973506689, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 46.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7341532639545885, | |
| "grad_norm": 0.0015266801929101348, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 46.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.7379375591296121, | |
| "grad_norm": 0.001181149622425437, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 46.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7417218543046358, | |
| "grad_norm": 0.0010459410259500146, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 46.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.7455061494796594, | |
| "grad_norm": 0.000900150160305202, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 46.0, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.7492904446546831, | |
| "grad_norm": 0.0010388368973508477, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 46.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.7530747398297067, | |
| "grad_norm": 0.0011183860478922725, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 46.0, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.7568590350047304, | |
| "grad_norm": 0.0010974981123581529, | |
| "learning_rate": 0.0, | |
| "loss": 46.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.7568590350047304, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.9678, | |
| "eval_samples_per_second": 226.142, | |
| "eval_steps_per_second": 56.917, | |
| "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": 3 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 32158462771200.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |