| { | |
| "best_metric": 1.9791635274887085, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.3982080637132902, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.001991040318566451, | |
| "grad_norm": 4.1927170753479, | |
| "learning_rate": 1e-05, | |
| "loss": 7.5775, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.001991040318566451, | |
| "eval_loss": 2.145913600921631, | |
| "eval_runtime": 96.1101, | |
| "eval_samples_per_second": 8.802, | |
| "eval_steps_per_second": 2.206, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.003982080637132902, | |
| "grad_norm": 2.4473557472229004, | |
| "learning_rate": 2e-05, | |
| "loss": 7.4194, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.005973120955699353, | |
| "grad_norm": 2.427483320236206, | |
| "learning_rate": 3e-05, | |
| "loss": 8.2488, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.007964161274265804, | |
| "grad_norm": 2.429349899291992, | |
| "learning_rate": 4e-05, | |
| "loss": 7.8777, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.009955201592832254, | |
| "grad_norm": 1.9974398612976074, | |
| "learning_rate": 5e-05, | |
| "loss": 7.6429, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.011946241911398706, | |
| "grad_norm": 1.8494004011154175, | |
| "learning_rate": 6e-05, | |
| "loss": 7.6779, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.013937282229965157, | |
| "grad_norm": 2.304295063018799, | |
| "learning_rate": 7e-05, | |
| "loss": 8.4867, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.015928322548531607, | |
| "grad_norm": 2.185741662979126, | |
| "learning_rate": 8e-05, | |
| "loss": 7.4555, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.017919362867098058, | |
| "grad_norm": 2.4035897254943848, | |
| "learning_rate": 9e-05, | |
| "loss": 7.973, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.019910403185664508, | |
| "grad_norm": 2.2464182376861572, | |
| "learning_rate": 0.0001, | |
| "loss": 7.2217, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.021901443504230962, | |
| "grad_norm": 1.9933096170425415, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 7.8099, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.023892483822797413, | |
| "grad_norm": 2.151336908340454, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 7.6336, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.025883524141363863, | |
| "grad_norm": 1.963433027267456, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 7.2683, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.027874564459930314, | |
| "grad_norm": 2.293760061264038, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 7.7981, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.029865604778496764, | |
| "grad_norm": 2.2580461502075195, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 8.0424, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.031856645097063215, | |
| "grad_norm": 1.8460721969604492, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 8.1952, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.033847685415629665, | |
| "grad_norm": 2.0165481567382812, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 8.1019, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.035838725734196115, | |
| "grad_norm": 1.986831784248352, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 7.6396, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.037829766052762566, | |
| "grad_norm": 1.853471040725708, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 7.9641, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.039820806371329016, | |
| "grad_norm": 2.076463460922241, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 8.2408, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.041811846689895474, | |
| "grad_norm": 2.391432285308838, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 7.9523, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.043802887008461924, | |
| "grad_norm": 2.024435520172119, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 8.5568, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.045793927327028375, | |
| "grad_norm": 2.76054048538208, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 7.8038, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.047784967645594825, | |
| "grad_norm": 2.3286190032958984, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 8.223, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.049776007964161276, | |
| "grad_norm": 2.2358877658843994, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 8.4138, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.051767048282727726, | |
| "grad_norm": 2.3133726119995117, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 8.2893, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05375808860129418, | |
| "grad_norm": 2.184105396270752, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 7.8866, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.05574912891986063, | |
| "grad_norm": 2.3296058177948, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 7.5064, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.05774016923842708, | |
| "grad_norm": 2.1816248893737793, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 8.2198, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.05973120955699353, | |
| "grad_norm": 2.2896475791931152, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 7.8416, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06172224987555998, | |
| "grad_norm": 2.2507057189941406, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 7.5694, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.06371329019412643, | |
| "grad_norm": 2.1473896503448486, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 8.3502, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06570433051269288, | |
| "grad_norm": 2.3653652667999268, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 7.9, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06769537083125933, | |
| "grad_norm": 2.322619915008545, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 8.4125, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.06968641114982578, | |
| "grad_norm": 2.472818374633789, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 8.0791, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.07167745146839223, | |
| "grad_norm": 2.5918126106262207, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 8.2419, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07366849178695868, | |
| "grad_norm": 2.5423145294189453, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 8.3602, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.07565953210552513, | |
| "grad_norm": 2.554898262023926, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 8.3687, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.07765057242409158, | |
| "grad_norm": 2.7162516117095947, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 8.2161, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.07964161274265803, | |
| "grad_norm": 2.6858348846435547, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 8.3233, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08163265306122448, | |
| "grad_norm": 3.176021099090576, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 8.5164, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.08362369337979095, | |
| "grad_norm": 2.9552512168884277, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 8.2171, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0856147336983574, | |
| "grad_norm": 2.777254343032837, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 8.3166, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.08760577401692385, | |
| "grad_norm": 2.599973201751709, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 8.4785, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0895968143354903, | |
| "grad_norm": 2.908027410507202, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 8.5118, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.09158785465405675, | |
| "grad_norm": 3.1602694988250732, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 8.5529, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0935788949726232, | |
| "grad_norm": 3.2083675861358643, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 8.1375, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.09556993529118965, | |
| "grad_norm": 3.468794107437134, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 8.5935, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0975609756097561, | |
| "grad_norm": 3.2282445430755615, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 8.1305, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.09955201592832255, | |
| "grad_norm": 4.116318225860596, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 8.1791, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09955201592832255, | |
| "eval_loss": 2.018760919570923, | |
| "eval_runtime": 97.5424, | |
| "eval_samples_per_second": 8.673, | |
| "eval_steps_per_second": 2.173, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.101543056246889, | |
| "grad_norm": 2.086719036102295, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 7.5437, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.10353409656545545, | |
| "grad_norm": 2.604001760482788, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 7.8617, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.1055251368840219, | |
| "grad_norm": 2.0230860710144043, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 7.4105, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.10751617720258835, | |
| "grad_norm": 1.772068977355957, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 7.8178, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.1095072175211548, | |
| "grad_norm": 1.5971081256866455, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 7.6503, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.11149825783972125, | |
| "grad_norm": 1.6913212537765503, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 7.8556, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1134892981582877, | |
| "grad_norm": 1.7140363454818726, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 7.8953, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.11548033847685416, | |
| "grad_norm": 1.679578185081482, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 7.7101, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.1174713787954206, | |
| "grad_norm": 1.6545398235321045, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 8.1807, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.11946241911398706, | |
| "grad_norm": 1.6474525928497314, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 7.8882, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1214534594325535, | |
| "grad_norm": 1.7631162405014038, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 7.7177, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.12344449975111996, | |
| "grad_norm": 1.675569772720337, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 8.0263, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.1254355400696864, | |
| "grad_norm": 1.725884199142456, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 8.3627, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.12742658038825286, | |
| "grad_norm": 1.8299144506454468, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 7.6133, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.1294176207068193, | |
| "grad_norm": 1.852302074432373, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 8.1276, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.13140866102538576, | |
| "grad_norm": 1.8702949285507202, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 7.8192, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.1333997013439522, | |
| "grad_norm": 1.9719868898391724, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 7.848, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.13539074166251866, | |
| "grad_norm": 1.9231232404708862, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 8.0177, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1373817819810851, | |
| "grad_norm": 1.9018663167953491, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 8.3228, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.13937282229965156, | |
| "grad_norm": 1.9371843338012695, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 7.5295, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.141363862618218, | |
| "grad_norm": 1.9617316722869873, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 7.6659, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.14335490293678446, | |
| "grad_norm": 1.9091986417770386, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 8.3663, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.1453459432553509, | |
| "grad_norm": 2.0275633335113525, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 7.9593, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.14733698357391736, | |
| "grad_norm": 1.8941304683685303, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 8.2681, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.1493280238924838, | |
| "grad_norm": 1.9514762163162231, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 8.4182, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.15131906421105026, | |
| "grad_norm": 2.091099739074707, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 7.8102, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.15331010452961671, | |
| "grad_norm": 2.341341495513916, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 7.5617, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.15530114484818316, | |
| "grad_norm": 2.552053451538086, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 8.1795, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15729218516674962, | |
| "grad_norm": 2.1133904457092285, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 7.7972, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.15928322548531607, | |
| "grad_norm": 2.4498162269592285, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 7.508, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.16127426580388252, | |
| "grad_norm": 2.088576078414917, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 8.273, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.16326530612244897, | |
| "grad_norm": 2.265047311782837, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 7.7166, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.16525634644101542, | |
| "grad_norm": 2.288518190383911, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 8.0692, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.1672473867595819, | |
| "grad_norm": 2.1535747051239014, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 8.0313, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.16923842707814835, | |
| "grad_norm": 2.1333887577056885, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 8.2886, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1712294673967148, | |
| "grad_norm": 2.2846364974975586, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 7.8147, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.17322050771528125, | |
| "grad_norm": 2.418757677078247, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 8.0926, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.1752115480338477, | |
| "grad_norm": 2.1804139614105225, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 7.7739, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.17720258835241415, | |
| "grad_norm": 2.639673948287964, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 8.1968, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.1791936286709806, | |
| "grad_norm": 2.676497220993042, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 8.1601, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.18118466898954705, | |
| "grad_norm": 2.6993329524993896, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 8.0585, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.1831757093081135, | |
| "grad_norm": 2.4620790481567383, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 8.3164, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.18516674962667995, | |
| "grad_norm": 2.335737943649292, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 8.0618, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.1871577899452464, | |
| "grad_norm": 2.859421730041504, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 8.7879, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.18914883026381285, | |
| "grad_norm": 2.8204166889190674, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 8.5739, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.1911398705823793, | |
| "grad_norm": 2.632925271987915, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 8.1831, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.19313091090094575, | |
| "grad_norm": 3.0627779960632324, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 7.7687, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.1951219512195122, | |
| "grad_norm": 2.720961570739746, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 8.2168, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.19711299153807865, | |
| "grad_norm": 3.122340202331543, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 8.1334, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.1991040318566451, | |
| "grad_norm": 4.294248104095459, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 8.3605, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1991040318566451, | |
| "eval_loss": 1.998111367225647, | |
| "eval_runtime": 96.7561, | |
| "eval_samples_per_second": 8.744, | |
| "eval_steps_per_second": 2.191, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.20109507217521155, | |
| "grad_norm": 2.1155364513397217, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 7.8629, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.203086112493778, | |
| "grad_norm": 2.213440418243408, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 7.8151, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.20507715281234445, | |
| "grad_norm": 2.1461501121520996, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 7.7148, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2070681931309109, | |
| "grad_norm": 1.5757580995559692, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 7.4002, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20905923344947736, | |
| "grad_norm": 1.685456395149231, | |
| "learning_rate": 5e-05, | |
| "loss": 7.4404, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2110502737680438, | |
| "grad_norm": 1.515172004699707, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 7.5646, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.21304131408661026, | |
| "grad_norm": 1.4339511394500732, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 8.2984, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.2150323544051767, | |
| "grad_norm": 2.0430703163146973, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 7.6342, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.21702339472374316, | |
| "grad_norm": 1.7122292518615723, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 7.9827, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.2190144350423096, | |
| "grad_norm": 1.5583425760269165, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 7.5668, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.22100547536087606, | |
| "grad_norm": 1.6786147356033325, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 7.8009, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.2229965156794425, | |
| "grad_norm": 1.6007399559020996, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 7.7134, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.22498755599800896, | |
| "grad_norm": 1.6516001224517822, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 7.9129, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.2269785963165754, | |
| "grad_norm": 1.6413806676864624, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 8.169, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.22896963663514186, | |
| "grad_norm": 1.6705026626586914, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 7.9665, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2309606769537083, | |
| "grad_norm": 1.6851974725723267, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 8.301, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.23295171727227476, | |
| "grad_norm": 1.7316675186157227, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 7.6395, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2349427575908412, | |
| "grad_norm": 1.7614164352416992, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 7.5071, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.23693379790940766, | |
| "grad_norm": 1.9808844327926636, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 7.9048, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.2389248382279741, | |
| "grad_norm": 1.7101449966430664, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 8.2639, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.24091587854654056, | |
| "grad_norm": 2.5709660053253174, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 7.6495, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.242906918865107, | |
| "grad_norm": 1.8031582832336426, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 8.3719, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.24489795918367346, | |
| "grad_norm": 1.863646388053894, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 8.2802, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.24688899950223991, | |
| "grad_norm": 1.7736895084381104, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 7.8002, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.24888003982080636, | |
| "grad_norm": 1.8948849439620972, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 8.2247, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2508710801393728, | |
| "grad_norm": 2.0049726963043213, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 8.0219, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.25286212045793927, | |
| "grad_norm": 2.08969783782959, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 7.4093, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.2548531607765057, | |
| "grad_norm": 2.0703985691070557, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 7.4586, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.25684420109507217, | |
| "grad_norm": 2.0214223861694336, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 8.4485, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.2588352414136386, | |
| "grad_norm": 1.9811716079711914, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 8.339, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.26082628173220507, | |
| "grad_norm": 2.365393877029419, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 7.3709, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2628173220507715, | |
| "grad_norm": 2.078462600708008, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 7.6712, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.26480836236933797, | |
| "grad_norm": 2.1712512969970703, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 7.7467, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.2667994026879044, | |
| "grad_norm": 2.5774106979370117, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 7.1528, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.26879044300647087, | |
| "grad_norm": 2.1006438732147217, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 8.5904, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2707814833250373, | |
| "grad_norm": 2.1961002349853516, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 7.8344, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.27277252364360377, | |
| "grad_norm": 2.1322438716888428, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 8.0909, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.2747635639621702, | |
| "grad_norm": 2.1785125732421875, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 8.4496, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.27675460428073667, | |
| "grad_norm": 2.217470645904541, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 8.2863, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.2787456445993031, | |
| "grad_norm": 2.28104567527771, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 8.542, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.28073668491786957, | |
| "grad_norm": 2.2927000522613525, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 8.3274, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.282727725236436, | |
| "grad_norm": 2.4151034355163574, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 8.0263, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.2847187655550025, | |
| "grad_norm": 2.414428949356079, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 7.6101, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.2867098058735689, | |
| "grad_norm": 2.436314344406128, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 8.1157, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.2887008461921354, | |
| "grad_norm": 2.676873207092285, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 7.9548, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.2906918865107018, | |
| "grad_norm": 2.914842367172241, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 8.1441, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.2926829268292683, | |
| "grad_norm": 2.72320556640625, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 8.342, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.2946739671478347, | |
| "grad_norm": 3.204787492752075, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 8.4702, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.2966650074664012, | |
| "grad_norm": 3.3982326984405518, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 8.643, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.2986560477849676, | |
| "grad_norm": 3.9501748085021973, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 8.2676, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2986560477849676, | |
| "eval_loss": 1.9845175743103027, | |
| "eval_runtime": 97.5477, | |
| "eval_samples_per_second": 8.673, | |
| "eval_steps_per_second": 2.173, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3006470881035341, | |
| "grad_norm": 1.3534071445465088, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 7.2213, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.3026381284221005, | |
| "grad_norm": 1.603489875793457, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 7.6319, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.304629168740667, | |
| "grad_norm": 1.4831784963607788, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 7.3738, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.30662020905923343, | |
| "grad_norm": 1.6179431676864624, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 7.5672, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.3086112493777999, | |
| "grad_norm": 1.6321980953216553, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 8.1407, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.31060228969636633, | |
| "grad_norm": 1.6661617755889893, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 8.0479, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.3125933300149328, | |
| "grad_norm": 1.4491256475448608, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 7.5624, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.31458437033349923, | |
| "grad_norm": 1.5583367347717285, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 7.6323, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.3165754106520657, | |
| "grad_norm": 1.589257836341858, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 7.5016, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.31856645097063213, | |
| "grad_norm": 1.644655466079712, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 7.8556, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3205574912891986, | |
| "grad_norm": 1.787083625793457, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 7.7584, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.32254853160776503, | |
| "grad_norm": 1.6175934076309204, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 7.5051, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.3245395719263315, | |
| "grad_norm": 1.5703297853469849, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 8.0459, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.32653061224489793, | |
| "grad_norm": 1.683521032333374, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 7.5684, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.3285216525634644, | |
| "grad_norm": 1.8564302921295166, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 7.3875, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.33051269288203083, | |
| "grad_norm": 1.5569313764572144, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 8.039, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.3325037332005973, | |
| "grad_norm": 1.8560819625854492, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 7.2681, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.3344947735191638, | |
| "grad_norm": 1.7723695039749146, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 7.6396, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.33648581383773024, | |
| "grad_norm": 1.6379140615463257, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 7.7666, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.3384768541562967, | |
| "grad_norm": 1.9762147665023804, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 7.9558, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.34046789447486314, | |
| "grad_norm": 1.8538239002227783, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 7.7814, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.3424589347934296, | |
| "grad_norm": 1.7401517629623413, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 7.9638, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.34444997511199604, | |
| "grad_norm": 1.9669773578643799, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 8.1031, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.3464410154305625, | |
| "grad_norm": 1.7937490940093994, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 7.506, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.34843205574912894, | |
| "grad_norm": 1.8896270990371704, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 7.4828, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3504230960676954, | |
| "grad_norm": 1.9170331954956055, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 7.7761, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.35241413638626184, | |
| "grad_norm": 1.947373390197754, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 8.3507, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.3544051767048283, | |
| "grad_norm": 1.9769765138626099, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 8.3762, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.35639621702339475, | |
| "grad_norm": 2.2252533435821533, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 7.6009, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.3583872573419612, | |
| "grad_norm": 1.9072762727737427, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 8.2758, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.36037829766052765, | |
| "grad_norm": 1.9121025800704956, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 7.8636, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.3623693379790941, | |
| "grad_norm": 2.0679497718811035, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 8.4165, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.36436037829766055, | |
| "grad_norm": 2.0888831615448, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 7.4964, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.366351418616227, | |
| "grad_norm": 2.4088799953460693, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 7.7784, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.36834245893479345, | |
| "grad_norm": 2.1652021408081055, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 8.3183, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.3703334992533599, | |
| "grad_norm": 2.2491962909698486, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 7.7902, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.37232453957192635, | |
| "grad_norm": 2.227860450744629, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 8.0956, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.3743155798904928, | |
| "grad_norm": 2.2770512104034424, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 8.1727, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.37630662020905925, | |
| "grad_norm": 2.307990789413452, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 8.3193, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.3782976605276257, | |
| "grad_norm": 2.996755838394165, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 8.1987, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.38028870084619215, | |
| "grad_norm": 2.6594462394714355, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 8.2504, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.3822797411647586, | |
| "grad_norm": 2.307396173477173, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 8.3347, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.38427078148332505, | |
| "grad_norm": 2.410108804702759, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 8.219, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.3862618218018915, | |
| "grad_norm": 2.6365981101989746, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 8.2585, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.38825286212045795, | |
| "grad_norm": 2.4511048793792725, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 8.3757, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.3902439024390244, | |
| "grad_norm": 2.7019314765930176, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 7.8581, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.39223494275759085, | |
| "grad_norm": 2.7035484313964844, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 8.0482, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.3942259830761573, | |
| "grad_norm": 3.0126864910125732, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 8.5532, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.39621702339472376, | |
| "grad_norm": 3.180725336074829, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 7.7343, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.3982080637132902, | |
| "grad_norm": 3.9281198978424072, | |
| "learning_rate": 0.0, | |
| "loss": 7.9991, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3982080637132902, | |
| "eval_loss": 1.9791635274887085, | |
| "eval_runtime": 97.1636, | |
| "eval_samples_per_second": 8.707, | |
| "eval_steps_per_second": 2.182, | |
| "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": 4.763667909102797e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |