|
{ |
|
"best_metric": 1.2930774688720703, |
|
"best_model_checkpoint": "miner_id_24/checkpoint-200", |
|
"epoch": 0.8385744234800838, |
|
"eval_steps": 50, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0041928721174004195, |
|
"grad_norm": 26.132644653320312, |
|
"learning_rate": 1e-05, |
|
"loss": 11.9012, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0041928721174004195, |
|
"eval_loss": 3.0544967651367188, |
|
"eval_runtime": 44.0725, |
|
"eval_samples_per_second": 9.121, |
|
"eval_steps_per_second": 2.292, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.008385744234800839, |
|
"grad_norm": 36.28742980957031, |
|
"learning_rate": 2e-05, |
|
"loss": 13.0845, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.012578616352201259, |
|
"grad_norm": 45.885284423828125, |
|
"learning_rate": 3e-05, |
|
"loss": 14.6054, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.016771488469601678, |
|
"grad_norm": 42.03292465209961, |
|
"learning_rate": 4e-05, |
|
"loss": 13.9683, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.020964360587002098, |
|
"grad_norm": 47.916175842285156, |
|
"learning_rate": 5e-05, |
|
"loss": 10.7715, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.025157232704402517, |
|
"grad_norm": 20.316499710083008, |
|
"learning_rate": 6e-05, |
|
"loss": 8.3264, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.029350104821802937, |
|
"grad_norm": 21.001514434814453, |
|
"learning_rate": 7e-05, |
|
"loss": 7.8696, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.033542976939203356, |
|
"grad_norm": 19.75449562072754, |
|
"learning_rate": 8e-05, |
|
"loss": 6.6757, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.03773584905660377, |
|
"grad_norm": 22.971229553222656, |
|
"learning_rate": 9e-05, |
|
"loss": 6.8329, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.041928721174004195, |
|
"grad_norm": 21.109153747558594, |
|
"learning_rate": 0.0001, |
|
"loss": 6.7707, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.04612159329140461, |
|
"grad_norm": 18.65835189819336, |
|
"learning_rate": 9.999316524962345e-05, |
|
"loss": 6.3268, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.050314465408805034, |
|
"grad_norm": 17.413570404052734, |
|
"learning_rate": 9.997266286704631e-05, |
|
"loss": 5.7432, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.05450733752620545, |
|
"grad_norm": 17.838520050048828, |
|
"learning_rate": 9.993849845741524e-05, |
|
"loss": 5.8578, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.05870020964360587, |
|
"grad_norm": 21.145803451538086, |
|
"learning_rate": 9.989068136093873e-05, |
|
"loss": 5.4755, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.06289308176100629, |
|
"grad_norm": 15.322203636169434, |
|
"learning_rate": 9.98292246503335e-05, |
|
"loss": 5.9927, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.06708595387840671, |
|
"grad_norm": 13.84201431274414, |
|
"learning_rate": 9.975414512725057e-05, |
|
"loss": 5.5105, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.07127882599580712, |
|
"grad_norm": 13.809650421142578, |
|
"learning_rate": 9.966546331768191e-05, |
|
"loss": 5.3261, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.07547169811320754, |
|
"grad_norm": 14.959165573120117, |
|
"learning_rate": 9.956320346634876e-05, |
|
"loss": 5.2148, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.07966457023060797, |
|
"grad_norm": 15.26698112487793, |
|
"learning_rate": 9.944739353007344e-05, |
|
"loss": 5.5136, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.08385744234800839, |
|
"grad_norm": 16.680856704711914, |
|
"learning_rate": 9.931806517013612e-05, |
|
"loss": 5.2218, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.0880503144654088, |
|
"grad_norm": 14.962255477905273, |
|
"learning_rate": 9.917525374361912e-05, |
|
"loss": 4.9713, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.09224318658280922, |
|
"grad_norm": 16.719751358032227, |
|
"learning_rate": 9.901899829374047e-05, |
|
"loss": 5.0492, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.09643605870020965, |
|
"grad_norm": 13.605834007263184, |
|
"learning_rate": 9.884934153917997e-05, |
|
"loss": 5.1194, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.10062893081761007, |
|
"grad_norm": 14.808418273925781, |
|
"learning_rate": 9.86663298624003e-05, |
|
"loss": 4.6639, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.10482180293501048, |
|
"grad_norm": 13.41181755065918, |
|
"learning_rate": 9.847001329696653e-05, |
|
"loss": 5.3685, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.1090146750524109, |
|
"grad_norm": 15.861624717712402, |
|
"learning_rate": 9.826044551386744e-05, |
|
"loss": 4.9872, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.11320754716981132, |
|
"grad_norm": 14.191696166992188, |
|
"learning_rate": 9.803768380684242e-05, |
|
"loss": 5.2217, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.11740041928721175, |
|
"grad_norm": 14.087152481079102, |
|
"learning_rate": 9.780178907671789e-05, |
|
"loss": 4.8128, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.12159329140461216, |
|
"grad_norm": 13.271366119384766, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 4.8184, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.12578616352201258, |
|
"grad_norm": 14.966277122497559, |
|
"learning_rate": 9.729086208503174e-05, |
|
"loss": 4.9615, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.129979035639413, |
|
"grad_norm": 12.936847686767578, |
|
"learning_rate": 9.701596950580806e-05, |
|
"loss": 4.8281, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.13417190775681342, |
|
"grad_norm": 12.324258804321289, |
|
"learning_rate": 9.672822322997305e-05, |
|
"loss": 4.7617, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.13836477987421383, |
|
"grad_norm": 13.070696830749512, |
|
"learning_rate": 9.642770192448536e-05, |
|
"loss": 5.0061, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.14255765199161424, |
|
"grad_norm": 14.148125648498535, |
|
"learning_rate": 9.611448774886924e-05, |
|
"loss": 4.9349, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.14675052410901468, |
|
"grad_norm": 12.829872131347656, |
|
"learning_rate": 9.578866633275288e-05, |
|
"loss": 5.1966, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.1509433962264151, |
|
"grad_norm": 15.064430236816406, |
|
"learning_rate": 9.545032675245813e-05, |
|
"loss": 5.1889, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.15513626834381553, |
|
"grad_norm": 13.483580589294434, |
|
"learning_rate": 9.509956150664796e-05, |
|
"loss": 4.9406, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.15932914046121593, |
|
"grad_norm": 14.467133522033691, |
|
"learning_rate": 9.473646649103818e-05, |
|
"loss": 4.6874, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.16352201257861634, |
|
"grad_norm": 14.77626895904541, |
|
"learning_rate": 9.43611409721806e-05, |
|
"loss": 5.4034, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.16771488469601678, |
|
"grad_norm": 13.509198188781738, |
|
"learning_rate": 9.397368756032445e-05, |
|
"loss": 5.0737, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.1719077568134172, |
|
"grad_norm": 14.06533432006836, |
|
"learning_rate": 9.357421218136386e-05, |
|
"loss": 4.9448, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.1761006289308176, |
|
"grad_norm": 14.199156761169434, |
|
"learning_rate": 9.316282404787871e-05, |
|
"loss": 5.2226, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.18029350104821804, |
|
"grad_norm": 14.681893348693848, |
|
"learning_rate": 9.273963562927695e-05, |
|
"loss": 5.2379, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.18448637316561844, |
|
"grad_norm": 15.566058158874512, |
|
"learning_rate": 9.230476262104677e-05, |
|
"loss": 5.2613, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.18867924528301888, |
|
"grad_norm": 13.801526069641113, |
|
"learning_rate": 9.185832391312644e-05, |
|
"loss": 5.461, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.1928721174004193, |
|
"grad_norm": 18.503665924072266, |
|
"learning_rate": 9.140044155740101e-05, |
|
"loss": 5.5382, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.1970649895178197, |
|
"grad_norm": 233.49794006347656, |
|
"learning_rate": 9.093124073433463e-05, |
|
"loss": 5.9307, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.20125786163522014, |
|
"grad_norm": 85.2809066772461, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 5.412, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.20545073375262055, |
|
"grad_norm": 18.1092529296875, |
|
"learning_rate": 8.995939984474624e-05, |
|
"loss": 5.5137, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.20964360587002095, |
|
"grad_norm": 17.491830825805664, |
|
"learning_rate": 8.945702546981969e-05, |
|
"loss": 5.7121, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.20964360587002095, |
|
"eval_loss": 1.4252018928527832, |
|
"eval_runtime": 45.3289, |
|
"eval_samples_per_second": 8.869, |
|
"eval_steps_per_second": 2.228, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.2138364779874214, |
|
"grad_norm": 17.839067459106445, |
|
"learning_rate": 8.894386393810563e-05, |
|
"loss": 8.458, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.2180293501048218, |
|
"grad_norm": 16.638402938842773, |
|
"learning_rate": 8.842005554284296e-05, |
|
"loss": 8.5195, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.2222222222222222, |
|
"grad_norm": 18.481019973754883, |
|
"learning_rate": 8.788574348801675e-05, |
|
"loss": 7.6538, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.22641509433962265, |
|
"grad_norm": 15.877742767333984, |
|
"learning_rate": 8.73410738492077e-05, |
|
"loss": 8.4381, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.23060796645702306, |
|
"grad_norm": 18.584312438964844, |
|
"learning_rate": 8.678619553365659e-05, |
|
"loss": 7.5717, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.2348008385744235, |
|
"grad_norm": 15.255925178527832, |
|
"learning_rate": 8.622126023955446e-05, |
|
"loss": 6.4571, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.2389937106918239, |
|
"grad_norm": 14.540227890014648, |
|
"learning_rate": 8.564642241456986e-05, |
|
"loss": 6.3747, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.2431865828092243, |
|
"grad_norm": 64.81522369384766, |
|
"learning_rate": 8.506183921362443e-05, |
|
"loss": 6.495, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.24737945492662475, |
|
"grad_norm": 17.637868881225586, |
|
"learning_rate": 8.44676704559283e-05, |
|
"loss": 6.1676, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.25157232704402516, |
|
"grad_norm": 16.184396743774414, |
|
"learning_rate": 8.386407858128706e-05, |
|
"loss": 6.3003, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.2557651991614256, |
|
"grad_norm": 14.975420951843262, |
|
"learning_rate": 8.32512286056924e-05, |
|
"loss": 5.7779, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.259958071278826, |
|
"grad_norm": 12.618855476379395, |
|
"learning_rate": 8.262928807620843e-05, |
|
"loss": 5.7407, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.2641509433962264, |
|
"grad_norm": 20.052886962890625, |
|
"learning_rate": 8.199842702516583e-05, |
|
"loss": 5.6884, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.26834381551362685, |
|
"grad_norm": 13.004949569702148, |
|
"learning_rate": 8.135881792367686e-05, |
|
"loss": 5.2275, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.27253668763102723, |
|
"grad_norm": 12.57992172241211, |
|
"learning_rate": 8.07106356344834e-05, |
|
"loss": 5.3673, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.27672955974842767, |
|
"grad_norm": 12.553001403808594, |
|
"learning_rate": 8.005405736415126e-05, |
|
"loss": 5.4621, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.2809224318658281, |
|
"grad_norm": 14.872114181518555, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 5.331, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.2851153039832285, |
|
"grad_norm": 13.390371322631836, |
|
"learning_rate": 7.871643313414718e-05, |
|
"loss": 5.3942, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.2893081761006289, |
|
"grad_norm": 12.523579597473145, |
|
"learning_rate": 7.803575286758364e-05, |
|
"loss": 5.4148, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.29350104821802936, |
|
"grad_norm": 12.635327339172363, |
|
"learning_rate": 7.734740790612136e-05, |
|
"loss": 5.3922, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.2976939203354298, |
|
"grad_norm": 11.972217559814453, |
|
"learning_rate": 7.66515864363997e-05, |
|
"loss": 4.5972, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.3018867924528302, |
|
"grad_norm": 11.309503555297852, |
|
"learning_rate": 7.594847868906076e-05, |
|
"loss": 5.3812, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.3060796645702306, |
|
"grad_norm": 13.033599853515625, |
|
"learning_rate": 7.52382768867422e-05, |
|
"loss": 4.6551, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.31027253668763105, |
|
"grad_norm": 12.664169311523438, |
|
"learning_rate": 7.452117519152542e-05, |
|
"loss": 5.5423, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.31446540880503143, |
|
"grad_norm": 13.637231826782227, |
|
"learning_rate": 7.379736965185368e-05, |
|
"loss": 5.3014, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.31865828092243187, |
|
"grad_norm": 12.818574905395508, |
|
"learning_rate": 7.30670581489344e-05, |
|
"loss": 5.273, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.3228511530398323, |
|
"grad_norm": 12.218353271484375, |
|
"learning_rate": 7.233044034264034e-05, |
|
"loss": 5.3526, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.3270440251572327, |
|
"grad_norm": 11.470541000366211, |
|
"learning_rate": 7.158771761692464e-05, |
|
"loss": 4.8623, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.3312368972746331, |
|
"grad_norm": 11.281693458557129, |
|
"learning_rate": 7.083909302476453e-05, |
|
"loss": 5.1589, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.33542976939203356, |
|
"grad_norm": 12.148499488830566, |
|
"learning_rate": 7.008477123264848e-05, |
|
"loss": 4.6785, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.33962264150943394, |
|
"grad_norm": 12.78722095489502, |
|
"learning_rate": 6.932495846462261e-05, |
|
"loss": 4.869, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.3438155136268344, |
|
"grad_norm": 11.150115013122559, |
|
"learning_rate": 6.855986244591104e-05, |
|
"loss": 4.8388, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.3480083857442348, |
|
"grad_norm": 11.860318183898926, |
|
"learning_rate": 6.778969234612584e-05, |
|
"loss": 5.3106, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.3522012578616352, |
|
"grad_norm": 12.364052772521973, |
|
"learning_rate": 6.701465872208216e-05, |
|
"loss": 4.795, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.35639412997903563, |
|
"grad_norm": 15.005729675292969, |
|
"learning_rate": 6.623497346023418e-05, |
|
"loss": 5.5391, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.36058700209643607, |
|
"grad_norm": 12.835009574890137, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 4.8894, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.36477987421383645, |
|
"grad_norm": 12.573246002197266, |
|
"learning_rate": 6.466250186922325e-05, |
|
"loss": 4.6682, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.3689727463312369, |
|
"grad_norm": 12.361988067626953, |
|
"learning_rate": 6.387014543809223e-05, |
|
"loss": 4.6504, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.3731656184486373, |
|
"grad_norm": 13.073786735534668, |
|
"learning_rate": 6.307399704769099e-05, |
|
"loss": 4.7555, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.37735849056603776, |
|
"grad_norm": 11.83043098449707, |
|
"learning_rate": 6.227427435703997e-05, |
|
"loss": 4.4787, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.38155136268343814, |
|
"grad_norm": 11.655750274658203, |
|
"learning_rate": 6.147119600233758e-05, |
|
"loss": 4.8748, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.3857442348008386, |
|
"grad_norm": 14.224935531616211, |
|
"learning_rate": 6.066498153718735e-05, |
|
"loss": 5.0494, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.389937106918239, |
|
"grad_norm": 11.915684700012207, |
|
"learning_rate": 5.985585137257401e-05, |
|
"loss": 4.7808, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.3941299790356394, |
|
"grad_norm": 13.685681343078613, |
|
"learning_rate": 5.90440267166055e-05, |
|
"loss": 4.7294, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.39832285115303984, |
|
"grad_norm": 13.151580810546875, |
|
"learning_rate": 5.8229729514036705e-05, |
|
"loss": 5.6233, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.4025157232704403, |
|
"grad_norm": 16.578044891357422, |
|
"learning_rate": 5.74131823855921e-05, |
|
"loss": 5.0588, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.40670859538784065, |
|
"grad_norm": 13.421004295349121, |
|
"learning_rate": 5.6594608567103456e-05, |
|
"loss": 5.2463, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.4109014675052411, |
|
"grad_norm": 13.570477485656738, |
|
"learning_rate": 5.577423184847932e-05, |
|
"loss": 5.4574, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.41509433962264153, |
|
"grad_norm": 13.61705493927002, |
|
"learning_rate": 5.495227651252315e-05, |
|
"loss": 5.8082, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.4192872117400419, |
|
"grad_norm": 14.726000785827637, |
|
"learning_rate": 5.4128967273616625e-05, |
|
"loss": 6.2693, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.4192872117400419, |
|
"eval_loss": 1.33364999294281, |
|
"eval_runtime": 44.4249, |
|
"eval_samples_per_second": 9.049, |
|
"eval_steps_per_second": 2.274, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.42348008385744235, |
|
"grad_norm": 12.140310287475586, |
|
"learning_rate": 5.330452921628497e-05, |
|
"loss": 7.4081, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.4276729559748428, |
|
"grad_norm": 12.952072143554688, |
|
"learning_rate": 5.247918773366112e-05, |
|
"loss": 7.5179, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.43186582809224316, |
|
"grad_norm": 14.149676322937012, |
|
"learning_rate": 5.165316846586541e-05, |
|
"loss": 7.5148, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.4360587002096436, |
|
"grad_norm": 13.430567741394043, |
|
"learning_rate": 5.0826697238317935e-05, |
|
"loss": 7.2354, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.44025157232704404, |
|
"grad_norm": 14.334609031677246, |
|
"learning_rate": 5e-05, |
|
"loss": 7.6918, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.4444444444444444, |
|
"grad_norm": 13.022397994995117, |
|
"learning_rate": 4.917330276168208e-05, |
|
"loss": 6.1959, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.44863731656184486, |
|
"grad_norm": 12.501748085021973, |
|
"learning_rate": 4.834683153413459e-05, |
|
"loss": 6.224, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.4528301886792453, |
|
"grad_norm": 11.310043334960938, |
|
"learning_rate": 4.7520812266338885e-05, |
|
"loss": 6.0405, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.4570230607966457, |
|
"grad_norm": 10.855735778808594, |
|
"learning_rate": 4.669547078371504e-05, |
|
"loss": 5.9622, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.4612159329140461, |
|
"grad_norm": 12.147499084472656, |
|
"learning_rate": 4.5871032726383386e-05, |
|
"loss": 5.5303, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.46540880503144655, |
|
"grad_norm": 12.338438987731934, |
|
"learning_rate": 4.504772348747687e-05, |
|
"loss": 5.5863, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.469601677148847, |
|
"grad_norm": 11.445416450500488, |
|
"learning_rate": 4.4225768151520694e-05, |
|
"loss": 5.0989, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.47379454926624737, |
|
"grad_norm": 11.623116493225098, |
|
"learning_rate": 4.3405391432896555e-05, |
|
"loss": 5.4732, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.4779874213836478, |
|
"grad_norm": 13.362641334533691, |
|
"learning_rate": 4.2586817614407895e-05, |
|
"loss": 5.1787, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.48218029350104824, |
|
"grad_norm": 11.479286193847656, |
|
"learning_rate": 4.17702704859633e-05, |
|
"loss": 4.9233, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.4863731656184486, |
|
"grad_norm": 12.21015739440918, |
|
"learning_rate": 4.095597328339452e-05, |
|
"loss": 4.7571, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.49056603773584906, |
|
"grad_norm": 12.11827278137207, |
|
"learning_rate": 4.0144148627425993e-05, |
|
"loss": 5.1819, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.4947589098532495, |
|
"grad_norm": 11.634247779846191, |
|
"learning_rate": 3.933501846281267e-05, |
|
"loss": 4.7911, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.4989517819706499, |
|
"grad_norm": 12.284260749816895, |
|
"learning_rate": 3.852880399766243e-05, |
|
"loss": 5.0751, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.5031446540880503, |
|
"grad_norm": 11.85743236541748, |
|
"learning_rate": 3.772572564296005e-05, |
|
"loss": 4.9025, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.5073375262054507, |
|
"grad_norm": 12.35157299041748, |
|
"learning_rate": 3.6926002952309016e-05, |
|
"loss": 4.8108, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.5115303983228512, |
|
"grad_norm": 11.492472648620605, |
|
"learning_rate": 3.612985456190778e-05, |
|
"loss": 5.0554, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.5157232704402516, |
|
"grad_norm": 12.209033966064453, |
|
"learning_rate": 3.533749813077677e-05, |
|
"loss": 5.1948, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.519916142557652, |
|
"grad_norm": 13.564550399780273, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 4.9508, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.5241090146750524, |
|
"grad_norm": 13.559568405151367, |
|
"learning_rate": 3.3765026539765834e-05, |
|
"loss": 4.9293, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.5283018867924528, |
|
"grad_norm": 10.4879789352417, |
|
"learning_rate": 3.298534127791785e-05, |
|
"loss": 4.4226, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.5324947589098532, |
|
"grad_norm": 11.57494831085205, |
|
"learning_rate": 3.221030765387417e-05, |
|
"loss": 4.815, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.5366876310272537, |
|
"grad_norm": 11.823269844055176, |
|
"learning_rate": 3.144013755408895e-05, |
|
"loss": 4.8705, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.5408805031446541, |
|
"grad_norm": 12.186006546020508, |
|
"learning_rate": 3.0675041535377405e-05, |
|
"loss": 4.7831, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.5450733752620545, |
|
"grad_norm": 12.419806480407715, |
|
"learning_rate": 2.991522876735154e-05, |
|
"loss": 5.1625, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.549266247379455, |
|
"grad_norm": 12.010101318359375, |
|
"learning_rate": 2.916090697523549e-05, |
|
"loss": 4.7556, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.5534591194968553, |
|
"grad_norm": 21.347789764404297, |
|
"learning_rate": 2.8412282383075363e-05, |
|
"loss": 4.572, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.5576519916142557, |
|
"grad_norm": 11.743422508239746, |
|
"learning_rate": 2.766955965735968e-05, |
|
"loss": 4.5518, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.5618448637316562, |
|
"grad_norm": 14.868667602539062, |
|
"learning_rate": 2.693294185106562e-05, |
|
"loss": 4.8989, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.5660377358490566, |
|
"grad_norm": 11.682171821594238, |
|
"learning_rate": 2.6202630348146324e-05, |
|
"loss": 4.9817, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.570230607966457, |
|
"grad_norm": 11.996781349182129, |
|
"learning_rate": 2.547882480847461e-05, |
|
"loss": 4.9972, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.5744234800838575, |
|
"grad_norm": 11.404191970825195, |
|
"learning_rate": 2.476172311325783e-05, |
|
"loss": 4.9415, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.5786163522012578, |
|
"grad_norm": 12.217618942260742, |
|
"learning_rate": 2.405152131093926e-05, |
|
"loss": 5.0178, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.5828092243186582, |
|
"grad_norm": 14.083375930786133, |
|
"learning_rate": 2.3348413563600325e-05, |
|
"loss": 4.9024, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.5870020964360587, |
|
"grad_norm": 12.773621559143066, |
|
"learning_rate": 2.2652592093878666e-05, |
|
"loss": 4.526, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.5911949685534591, |
|
"grad_norm": 11.781105995178223, |
|
"learning_rate": 2.196424713241637e-05, |
|
"loss": 4.9392, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.5953878406708596, |
|
"grad_norm": 12.22276496887207, |
|
"learning_rate": 2.128356686585282e-05, |
|
"loss": 4.8146, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.59958071278826, |
|
"grad_norm": 11.05539608001709, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 5.0974, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.6037735849056604, |
|
"grad_norm": 19.491600036621094, |
|
"learning_rate": 1.9945942635848748e-05, |
|
"loss": 5.1393, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.6079664570230608, |
|
"grad_norm": 11.561285972595215, |
|
"learning_rate": 1.928936436551661e-05, |
|
"loss": 4.8279, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.6121593291404612, |
|
"grad_norm": 13.252482414245605, |
|
"learning_rate": 1.8641182076323148e-05, |
|
"loss": 4.9525, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.6163522012578616, |
|
"grad_norm": 11.3013916015625, |
|
"learning_rate": 1.800157297483417e-05, |
|
"loss": 5.0606, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.6205450733752621, |
|
"grad_norm": 12.376293182373047, |
|
"learning_rate": 1.7370711923791567e-05, |
|
"loss": 5.5769, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.6247379454926625, |
|
"grad_norm": 12.117962837219238, |
|
"learning_rate": 1.6748771394307585e-05, |
|
"loss": 5.041, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.6289308176100629, |
|
"grad_norm": 14.499025344848633, |
|
"learning_rate": 1.6135921418712956e-05, |
|
"loss": 5.4035, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.6289308176100629, |
|
"eval_loss": 1.3053677082061768, |
|
"eval_runtime": 44.7304, |
|
"eval_samples_per_second": 8.987, |
|
"eval_steps_per_second": 2.258, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.6331236897274634, |
|
"grad_norm": 10.202156066894531, |
|
"learning_rate": 1.553232954407171e-05, |
|
"loss": 7.219, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.6373165618448637, |
|
"grad_norm": 12.22631549835205, |
|
"learning_rate": 1.4938160786375572e-05, |
|
"loss": 7.1389, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.6415094339622641, |
|
"grad_norm": 12.613519668579102, |
|
"learning_rate": 1.435357758543015e-05, |
|
"loss": 6.7047, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.6457023060796646, |
|
"grad_norm": 16.083316802978516, |
|
"learning_rate": 1.3778739760445552e-05, |
|
"loss": 6.975, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.649895178197065, |
|
"grad_norm": 17.691789627075195, |
|
"learning_rate": 1.3213804466343421e-05, |
|
"loss": 7.6637, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.6540880503144654, |
|
"grad_norm": 11.660400390625, |
|
"learning_rate": 1.2658926150792322e-05, |
|
"loss": 6.6873, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.6582809224318659, |
|
"grad_norm": 10.926519393920898, |
|
"learning_rate": 1.2114256511983274e-05, |
|
"loss": 5.615, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.6624737945492662, |
|
"grad_norm": 12.457261085510254, |
|
"learning_rate": 1.157994445715706e-05, |
|
"loss": 6.7217, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.6666666666666666, |
|
"grad_norm": 12.392708778381348, |
|
"learning_rate": 1.1056136061894384e-05, |
|
"loss": 4.9746, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.6708595387840671, |
|
"grad_norm": 12.87796688079834, |
|
"learning_rate": 1.0542974530180327e-05, |
|
"loss": 5.8785, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.6750524109014675, |
|
"grad_norm": 10.879608154296875, |
|
"learning_rate": 1.0040600155253765e-05, |
|
"loss": 4.8618, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.6792452830188679, |
|
"grad_norm": 11.552359580993652, |
|
"learning_rate": 9.549150281252633e-06, |
|
"loss": 5.7525, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.6834381551362684, |
|
"grad_norm": 10.915603637695312, |
|
"learning_rate": 9.068759265665384e-06, |
|
"loss": 5.3969, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.6876310272536688, |
|
"grad_norm": 11.90835952758789, |
|
"learning_rate": 8.599558442598998e-06, |
|
"loss": 4.3017, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.6918238993710691, |
|
"grad_norm": 16.585939407348633, |
|
"learning_rate": 8.141676086873572e-06, |
|
"loss": 4.7759, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.6960167714884696, |
|
"grad_norm": 11.607003211975098, |
|
"learning_rate": 7.695237378953223e-06, |
|
"loss": 4.7341, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.70020964360587, |
|
"grad_norm": 13.098814964294434, |
|
"learning_rate": 7.260364370723044e-06, |
|
"loss": 5.1048, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.7044025157232704, |
|
"grad_norm": 10.588081359863281, |
|
"learning_rate": 6.837175952121306e-06, |
|
"loss": 4.9007, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.7085953878406709, |
|
"grad_norm": 11.971794128417969, |
|
"learning_rate": 6.425787818636131e-06, |
|
"loss": 4.5698, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.7127882599580713, |
|
"grad_norm": 11.241368293762207, |
|
"learning_rate": 6.026312439675552e-06, |
|
"loss": 4.9665, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.7169811320754716, |
|
"grad_norm": 11.0311918258667, |
|
"learning_rate": 5.6388590278194096e-06, |
|
"loss": 5.2078, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.7211740041928721, |
|
"grad_norm": 10.63940715789795, |
|
"learning_rate": 5.263533508961827e-06, |
|
"loss": 5.0156, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.7253668763102725, |
|
"grad_norm": 10.94045639038086, |
|
"learning_rate": 4.900438493352055e-06, |
|
"loss": 4.7311, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.7295597484276729, |
|
"grad_norm": 10.13798999786377, |
|
"learning_rate": 4.549673247541875e-06, |
|
"loss": 4.5984, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.7337526205450734, |
|
"grad_norm": 11.286879539489746, |
|
"learning_rate": 4.2113336672471245e-06, |
|
"loss": 4.4663, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.7379454926624738, |
|
"grad_norm": 11.34493350982666, |
|
"learning_rate": 3.885512251130763e-06, |
|
"loss": 4.9219, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.7421383647798742, |
|
"grad_norm": 10.747261047363281, |
|
"learning_rate": 3.5722980755146517e-06, |
|
"loss": 4.6713, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.7463312368972747, |
|
"grad_norm": 12.748205184936523, |
|
"learning_rate": 3.271776770026963e-06, |
|
"loss": 4.5862, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.750524109014675, |
|
"grad_norm": 10.732872009277344, |
|
"learning_rate": 2.9840304941919415e-06, |
|
"loss": 4.7369, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.7547169811320755, |
|
"grad_norm": 11.290228843688965, |
|
"learning_rate": 2.7091379149682685e-06, |
|
"loss": 4.7538, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.7589098532494759, |
|
"grad_norm": 10.670109748840332, |
|
"learning_rate": 2.4471741852423237e-06, |
|
"loss": 4.8439, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.7631027253668763, |
|
"grad_norm": 13.244367599487305, |
|
"learning_rate": 2.1982109232821178e-06, |
|
"loss": 4.8907, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.7672955974842768, |
|
"grad_norm": 11.110193252563477, |
|
"learning_rate": 1.962316193157593e-06, |
|
"loss": 4.8642, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.7714884696016772, |
|
"grad_norm": 13.409936904907227, |
|
"learning_rate": 1.7395544861325718e-06, |
|
"loss": 4.6917, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.7756813417190775, |
|
"grad_norm": 13.237275123596191, |
|
"learning_rate": 1.5299867030334814e-06, |
|
"loss": 4.4056, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.779874213836478, |
|
"grad_norm": 11.482529640197754, |
|
"learning_rate": 1.333670137599713e-06, |
|
"loss": 4.7563, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.7840670859538784, |
|
"grad_norm": 13.115639686584473, |
|
"learning_rate": 1.1506584608200367e-06, |
|
"loss": 5.1429, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.7882599580712788, |
|
"grad_norm": 12.13857650756836, |
|
"learning_rate": 9.810017062595322e-07, |
|
"loss": 4.9606, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.7924528301886793, |
|
"grad_norm": 11.677776336669922, |
|
"learning_rate": 8.247462563808817e-07, |
|
"loss": 4.5443, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.7966457023060797, |
|
"grad_norm": 11.770204544067383, |
|
"learning_rate": 6.819348298638839e-07, |
|
"loss": 4.5537, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.80083857442348, |
|
"grad_norm": 13.534977912902832, |
|
"learning_rate": 5.526064699265753e-07, |
|
"loss": 4.8516, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.8050314465408805, |
|
"grad_norm": 11.312360763549805, |
|
"learning_rate": 4.367965336512403e-07, |
|
"loss": 4.4825, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.8092243186582809, |
|
"grad_norm": 10.874757766723633, |
|
"learning_rate": 3.3453668231809286e-07, |
|
"loss": 4.6596, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.8134171907756813, |
|
"grad_norm": 13.562448501586914, |
|
"learning_rate": 2.458548727494292e-07, |
|
"loss": 4.8363, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.8176100628930818, |
|
"grad_norm": 11.777942657470703, |
|
"learning_rate": 1.7077534966650766e-07, |
|
"loss": 4.7338, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.8218029350104822, |
|
"grad_norm": 13.375640869140625, |
|
"learning_rate": 1.0931863906127327e-07, |
|
"loss": 4.6219, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.8259958071278826, |
|
"grad_norm": 11.146055221557617, |
|
"learning_rate": 6.150154258476315e-08, |
|
"loss": 4.6043, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.8301886792452831, |
|
"grad_norm": 12.50096321105957, |
|
"learning_rate": 2.7337132953697554e-08, |
|
"loss": 5.2926, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.8343815513626834, |
|
"grad_norm": 11.735309600830078, |
|
"learning_rate": 6.834750376549792e-09, |
|
"loss": 4.6868, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.8385744234800838, |
|
"grad_norm": 12.703553199768066, |
|
"learning_rate": 0.0, |
|
"loss": 5.3169, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.8385744234800838, |
|
"eval_loss": 1.2930774688720703, |
|
"eval_runtime": 44.654, |
|
"eval_samples_per_second": 9.003, |
|
"eval_steps_per_second": 2.262, |
|
"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": 3.158928104840233e+17, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|