| { | |
| "best_metric": NaN, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.021961128802020422, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00010980564401010212, | |
| "grad_norm": 2.6502325534820557, | |
| "learning_rate": 1e-05, | |
| "loss": 8.6012, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00010980564401010212, | |
| "eval_loss": NaN, | |
| "eval_runtime": 740.8185, | |
| "eval_samples_per_second": 20.704, | |
| "eval_steps_per_second": 5.177, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00021961128802020424, | |
| "grad_norm": 2.9273834228515625, | |
| "learning_rate": 2e-05, | |
| "loss": 8.2653, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00032941693203030636, | |
| "grad_norm": 3.2805614471435547, | |
| "learning_rate": 3e-05, | |
| "loss": 8.8424, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0004392225760404085, | |
| "grad_norm": 3.303515672683716, | |
| "learning_rate": 4e-05, | |
| "loss": 8.7725, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0005490282200505106, | |
| "grad_norm": 3.171604633331299, | |
| "learning_rate": 5e-05, | |
| "loss": 8.7924, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0006588338640606127, | |
| "grad_norm": 2.6296401023864746, | |
| "learning_rate": 6e-05, | |
| "loss": 8.3232, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0007686395080707148, | |
| "grad_norm": 2.6075048446655273, | |
| "learning_rate": 7e-05, | |
| "loss": 8.3647, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.000878445152080817, | |
| "grad_norm": 2.7028496265411377, | |
| "learning_rate": 8e-05, | |
| "loss": 8.284, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.000988250796090919, | |
| "grad_norm": 3.0198540687561035, | |
| "learning_rate": 9e-05, | |
| "loss": 8.6266, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0010980564401010212, | |
| "grad_norm": 3.171236515045166, | |
| "learning_rate": 0.0001, | |
| "loss": 8.5796, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0012078620841111233, | |
| "grad_norm": 2.9306540489196777, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 8.211, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0013176677281212254, | |
| "grad_norm": 3.1336894035339355, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 8.2714, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0014274733721313276, | |
| "grad_norm": 2.9940574169158936, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 8.0159, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0015372790161414297, | |
| "grad_norm": 3.42741322517395, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 8.2993, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0016470846601515318, | |
| "grad_norm": 3.2383780479431152, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 7.72, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.001756890304161634, | |
| "grad_norm": 4.075361728668213, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 8.6122, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.001866695948171736, | |
| "grad_norm": 3.733229160308838, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 7.9517, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.001976501592181838, | |
| "grad_norm": 3.940094232559204, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 7.9017, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.00208630723619194, | |
| "grad_norm": 4.151432991027832, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 8.371, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0021961128802020424, | |
| "grad_norm": 5.207062244415283, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 7.3331, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0023059185242121443, | |
| "grad_norm": 4.164064407348633, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 7.332, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0024157241682222466, | |
| "grad_norm": 6.835943222045898, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 7.8871, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0025255298122323485, | |
| "grad_norm": 5.428422927856445, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 7.5314, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.002635335456242451, | |
| "grad_norm": 5.860774040222168, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 7.5008, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0027451411002525528, | |
| "grad_norm": 6.325087070465088, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 7.3492, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002854946744262655, | |
| "grad_norm": 5.724707126617432, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 7.5614, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.002964752388272757, | |
| "grad_norm": 6.806790351867676, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 6.1704, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0030745580322828594, | |
| "grad_norm": 7.080099105834961, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 6.7244, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0031843636762929613, | |
| "grad_norm": 4.642933368682861, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 6.0953, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0032941693203030636, | |
| "grad_norm": 6.909359931945801, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 7.7606, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0034039749643131655, | |
| "grad_norm": 6.136366367340088, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 4.6353, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.003513780608323268, | |
| "grad_norm": 7.052574157714844, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 5.5296, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0036235862523333697, | |
| "grad_norm": 5.715820789337158, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.8394, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.003733391896343472, | |
| "grad_norm": 5.992306709289551, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 4.3861, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.003843197540353574, | |
| "grad_norm": 6.2420783042907715, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 4.7804, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.003953003184363676, | |
| "grad_norm": 4.731659412384033, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.4961, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.004062808828373779, | |
| "grad_norm": 4.582451343536377, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 3.0759, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.00417261447238388, | |
| "grad_norm": 10.449984550476074, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 6.4343, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0042824201163939824, | |
| "grad_norm": 3.6828055381774902, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.9859, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.004392225760404085, | |
| "grad_norm": 17.16545295715332, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 7.9194, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004502031404414187, | |
| "grad_norm": 4.975934028625488, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.8401, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.004611837048424289, | |
| "grad_norm": 3.7905521392822266, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.9404, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.004721642692434391, | |
| "grad_norm": 5.671470642089844, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 4.0305, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004831448336444493, | |
| "grad_norm": 4.986380100250244, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.4824, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.004941253980454596, | |
| "grad_norm": 3.115936040878296, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.2185, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.005051059624464697, | |
| "grad_norm": 6.371460914611816, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 3.6094, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.005160865268474799, | |
| "grad_norm": 6.855349063873291, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.3127, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.005270670912484902, | |
| "grad_norm": 4.935694694519043, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.3315, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.005380476556495004, | |
| "grad_norm": 6.071768760681152, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.9731, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0054902822005051055, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0054902822005051055, | |
| "eval_loss": NaN, | |
| "eval_runtime": 745.9902, | |
| "eval_samples_per_second": 20.561, | |
| "eval_steps_per_second": 5.141, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005600087844515208, | |
| "grad_norm": 9.470258712768555, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 9.6709, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.00570989348852531, | |
| "grad_norm": 8.197243690490723, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 9.6274, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0058196991325354126, | |
| "grad_norm": 5.740113735198975, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 8.8491, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005929504776545514, | |
| "grad_norm": 3.950591802597046, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 8.6568, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.006039310420555616, | |
| "grad_norm": 2.908627986907959, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 8.2575, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.006149116064565719, | |
| "grad_norm": 2.370631217956543, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 7.9929, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.006258921708575821, | |
| "grad_norm": 2.2208831310272217, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 7.9466, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0063687273525859225, | |
| "grad_norm": 2.1500656604766846, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 7.9672, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.006478532996596025, | |
| "grad_norm": 2.2153913974761963, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 7.5905, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.006588338640606127, | |
| "grad_norm": 2.5188639163970947, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 7.6952, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0066981442846162295, | |
| "grad_norm": 2.7013847827911377, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 8.4593, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.006807949928626331, | |
| "grad_norm": 2.931748867034912, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 8.0846, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.006917755572636433, | |
| "grad_norm": 2.9140090942382812, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 7.757, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.007027561216646536, | |
| "grad_norm": 2.9066874980926514, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 7.8689, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.007137366860656638, | |
| "grad_norm": 2.986344575881958, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 7.5816, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0072471725046667395, | |
| "grad_norm": 2.930072784423828, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 8.1042, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.007356978148676842, | |
| "grad_norm": 2.986283779144287, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 7.5451, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.007466783792686944, | |
| "grad_norm": 3.8496253490448, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 7.969, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0075765894366970465, | |
| "grad_norm": 3.796609878540039, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 7.6358, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.007686395080707148, | |
| "grad_norm": 4.6060309410095215, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 7.4349, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.00779620072471725, | |
| "grad_norm": 4.086380958557129, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 6.8807, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.007906006368727353, | |
| "grad_norm": 4.672285556793213, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 8.2162, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.008015812012737454, | |
| "grad_norm": 4.407833576202393, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 7.2127, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.008125617656747557, | |
| "grad_norm": 4.553558349609375, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 7.1724, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.008235423300757659, | |
| "grad_norm": 4.706977844238281, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 5.8799, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.00834522894476776, | |
| "grad_norm": 4.596185684204102, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 5.4282, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.008455034588777863, | |
| "grad_norm": 6.931064128875732, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 7.9549, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.008564840232787965, | |
| "grad_norm": 6.019664764404297, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 3.8056, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.008674645876798068, | |
| "grad_norm": 7.432868003845215, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 7.0909, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.00878445152080817, | |
| "grad_norm": 6.675373077392578, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 3.9711, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.008894257164818271, | |
| "grad_norm": 6.31569242477417, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 6.1075, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.009004062808828374, | |
| "grad_norm": 10.232958793640137, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 4.651, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.009113868452838476, | |
| "grad_norm": 4.610867977142334, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 3.1149, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.009223674096848577, | |
| "grad_norm": 4.2287492752075195, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.0559, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.00933347974085868, | |
| "grad_norm": 3.4553091526031494, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.9526, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.009443285384868782, | |
| "grad_norm": 5.008721351623535, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 3.6585, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.009553091028878885, | |
| "grad_norm": 7.621737480163574, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 4.5919, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.009662896672888987, | |
| "grad_norm": 4.2493720054626465, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.9427, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.009772702316899088, | |
| "grad_norm": 8.625337600708008, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 3.6866, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.009882507960909191, | |
| "grad_norm": 5.66685676574707, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 2.8351, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.009992313604919293, | |
| "grad_norm": 5.009742736816406, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.8861, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.010102119248929394, | |
| "grad_norm": 3.5267746448516846, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.0906, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.010211924892939497, | |
| "grad_norm": 2.047354221343994, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.1326, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.010321730536949599, | |
| "grad_norm": 2.4679694175720215, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.2005, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.010431536180959702, | |
| "grad_norm": 0.4190016984939575, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0269, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.010541341824969803, | |
| "grad_norm": 3.6314680576324463, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.5857, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.010651147468979905, | |
| "grad_norm": 2.758847236633301, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.1722, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.010760953112990008, | |
| "grad_norm": 1.3835760354995728, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0857, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.01087075875700011, | |
| "grad_norm": 9.623651504516602, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.3477, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.010980564401010211, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.010980564401010211, | |
| "eval_loss": NaN, | |
| "eval_runtime": 745.7596, | |
| "eval_samples_per_second": 20.567, | |
| "eval_steps_per_second": 5.142, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.011090370045020314, | |
| "grad_norm": 20.27688217163086, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 10.5539, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.011200175689030416, | |
| "grad_norm": 20.22991180419922, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 10.5449, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.011309981333040519, | |
| "grad_norm": 16.139301300048828, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 10.2872, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.01141978697705062, | |
| "grad_norm": 10.393228530883789, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 9.6018, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.011529592621060722, | |
| "grad_norm": 6.897857189178467, | |
| "learning_rate": 5e-05, | |
| "loss": 9.0137, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.011639398265070825, | |
| "grad_norm": 5.242559909820557, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 8.7974, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.011749203909080927, | |
| "grad_norm": 4.246593475341797, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 8.8507, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.011859009553091028, | |
| "grad_norm": 3.383063316345215, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 8.149, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.011968815197101131, | |
| "grad_norm": 2.915780544281006, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 8.2273, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.012078620841111233, | |
| "grad_norm": 2.8346691131591797, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 8.0532, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.012188426485121336, | |
| "grad_norm": 2.5639634132385254, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 7.7036, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.012298232129131437, | |
| "grad_norm": 2.2797608375549316, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 8.2839, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.012408037773141539, | |
| "grad_norm": 2.458200693130493, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 7.4049, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.012517843417151642, | |
| "grad_norm": 2.569010019302368, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 7.8835, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.012627649061161744, | |
| "grad_norm": 2.4975454807281494, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 7.3549, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.012737454705171845, | |
| "grad_norm": 2.657583475112915, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 7.4873, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.012847260349181948, | |
| "grad_norm": 3.0164661407470703, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 7.5112, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.01295706599319205, | |
| "grad_norm": 2.9580395221710205, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 7.2151, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.013066871637202153, | |
| "grad_norm": 3.4198410511016846, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 7.6243, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.013176677281212254, | |
| "grad_norm": 3.8369998931884766, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 6.8254, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.013286482925222356, | |
| "grad_norm": 3.758122205734253, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 7.497, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.013396288569232459, | |
| "grad_norm": 3.8267016410827637, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 6.6575, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.01350609421324256, | |
| "grad_norm": 4.286101341247559, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 7.3846, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.013615899857252662, | |
| "grad_norm": 4.737072944641113, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 6.9982, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.013725705501262765, | |
| "grad_norm": 4.438685417175293, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 7.1705, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.013835511145272867, | |
| "grad_norm": 5.58498477935791, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 7.2268, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.01394531678928297, | |
| "grad_norm": 4.726098537445068, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 4.5447, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.014055122433293071, | |
| "grad_norm": 5.851497173309326, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 7.0821, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.014164928077303173, | |
| "grad_norm": 4.897258758544922, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 5.6782, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.014274733721313276, | |
| "grad_norm": 6.970987319946289, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 6.1244, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.014384539365323377, | |
| "grad_norm": 6.077361106872559, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 2.6013, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.014494345009333479, | |
| "grad_norm": 4.792599201202393, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 2.9484, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.014604150653343582, | |
| "grad_norm": 5.143320560455322, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 3.6824, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.014713956297353684, | |
| "grad_norm": 6.228075981140137, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 6.0007, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.014823761941363787, | |
| "grad_norm": 5.384539604187012, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 4.8531, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.014933567585373888, | |
| "grad_norm": 6.0412774085998535, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 4.3535, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.01504337322938399, | |
| "grad_norm": 3.7027535438537598, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.2463, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.015153178873394093, | |
| "grad_norm": 5.037492275238037, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 4.0731, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.015262984517404194, | |
| "grad_norm": 7.468751430511475, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 5.2297, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.015372790161414296, | |
| "grad_norm": 5.828507423400879, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 4.7216, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.015482595805424399, | |
| "grad_norm": 3.6966893672943115, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.6202, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.0155924014494345, | |
| "grad_norm": 7.077216148376465, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 3.4661, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.015702207093444602, | |
| "grad_norm": 7.210454940795898, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 3.1293, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.015812012737454705, | |
| "grad_norm": 6.044134140014648, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 2.1328, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.01592181838146481, | |
| "grad_norm": 6.361577033996582, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.3974, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.016031624025474908, | |
| "grad_norm": 17.18918228149414, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.4077, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.01614142966948501, | |
| "grad_norm": 4.894886016845703, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.4673, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.016251235313495115, | |
| "grad_norm": 1.4776272773742676, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.3222, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.016361040957505214, | |
| "grad_norm": 0.5147504210472107, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.2668, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.016470846601515318, | |
| "grad_norm": 2.5885744094848633, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.6849, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.016470846601515318, | |
| "eval_loss": NaN, | |
| "eval_runtime": 746.4299, | |
| "eval_samples_per_second": 20.548, | |
| "eval_steps_per_second": 5.138, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01658065224552542, | |
| "grad_norm": 2.927521228790283, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 8.3447, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.01669045788953552, | |
| "grad_norm": 3.0052969455718994, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 8.1073, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.016800263533545624, | |
| "grad_norm": 2.9719297885894775, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 8.2021, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.016910069177555727, | |
| "grad_norm": 2.8135275840759277, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 7.9668, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.01701987482156583, | |
| "grad_norm": 2.9337542057037354, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 7.8755, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.01712968046557593, | |
| "grad_norm": 2.8828859329223633, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 7.8647, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.017239486109586033, | |
| "grad_norm": 2.829685926437378, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 8.1085, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.017349291753596136, | |
| "grad_norm": 2.603776216506958, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 7.9814, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.017459097397606236, | |
| "grad_norm": 2.792379856109619, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 7.9484, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.01756890304161634, | |
| "grad_norm": 3.043826103210449, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 7.9742, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.017678708685626442, | |
| "grad_norm": 2.8475284576416016, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 8.4931, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.017788514329636542, | |
| "grad_norm": 2.9101884365081787, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 7.9352, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.017898319973646645, | |
| "grad_norm": 2.878385543823242, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 7.753, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.01800812561765675, | |
| "grad_norm": 3.1888694763183594, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 8.3528, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.018117931261666848, | |
| "grad_norm": 3.141446113586426, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 7.6143, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.01822773690567695, | |
| "grad_norm": 3.249999761581421, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 7.7441, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.018337542549687055, | |
| "grad_norm": 3.3556692600250244, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 7.6016, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.018447348193697154, | |
| "grad_norm": 3.636415719985962, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 7.6953, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.018557153837707258, | |
| "grad_norm": 3.624875545501709, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 8.2286, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.01866695948171736, | |
| "grad_norm": 3.8759829998016357, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 8.2271, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.018776765125727464, | |
| "grad_norm": 3.9147400856018066, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 7.9848, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.018886570769737564, | |
| "grad_norm": 4.543432235717773, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 7.08, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.018996376413747667, | |
| "grad_norm": 4.144062042236328, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 7.2179, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.01910618205775777, | |
| "grad_norm": 5.5954461097717285, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 7.3824, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.01921598770176787, | |
| "grad_norm": 4.188710689544678, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 7.2055, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.019325793345777973, | |
| "grad_norm": 6.450707912445068, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 7.3443, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.019435598989788076, | |
| "grad_norm": 4.754418849945068, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 4.9993, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.019545404633798176, | |
| "grad_norm": 4.2252326011657715, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 5.8137, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.01965521027780828, | |
| "grad_norm": 5.499410629272461, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 5.684, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.019765015921818382, | |
| "grad_norm": 4.098621845245361, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 3.0545, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.019874821565828482, | |
| "grad_norm": 5.947284698486328, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 6.0958, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.019984627209838585, | |
| "grad_norm": 9.094640731811523, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 5.0735, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.02009443285384869, | |
| "grad_norm": 4.276609897613525, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 3.6866, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.02020423849785879, | |
| "grad_norm": 5.701272010803223, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 2.8552, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.02031404414186889, | |
| "grad_norm": 2.7438712120056152, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 2.5842, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.020423849785878995, | |
| "grad_norm": 3.399106502532959, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 2.1209, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.020533655429889098, | |
| "grad_norm": 8.859334945678711, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 6.3239, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.020643461073899198, | |
| "grad_norm": 4.675004482269287, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 4.034, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.0207532667179093, | |
| "grad_norm": 3.336203098297119, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.999, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.020863072361919404, | |
| "grad_norm": 6.969828128814697, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 5.9507, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.020972878005929504, | |
| "grad_norm": 3.6757795810699463, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.8426, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.021082683649939607, | |
| "grad_norm": 2.9385955333709717, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.4529, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.02119248929394971, | |
| "grad_norm": 2.350724935531616, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.2031, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.02130229493795981, | |
| "grad_norm": 7.855735778808594, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 4.696, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.021412100581969913, | |
| "grad_norm": 2.9840855598449707, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.8918, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.021521906225980016, | |
| "grad_norm": 5.21516752243042, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.0816, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.021631711869990116, | |
| "grad_norm": 4.186825752258301, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.6831, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.02174151751400022, | |
| "grad_norm": 9.229358673095703, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 5.1912, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.021851323158010322, | |
| "grad_norm": 4.757203102111816, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.8366, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.021961128802020422, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.021961128802020422, | |
| "eval_loss": NaN, | |
| "eval_runtime": 746.4106, | |
| "eval_samples_per_second": 20.549, | |
| "eval_steps_per_second": 5.138, | |
| "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": 1.5164471555968205e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |