| { | |
| "best_metric": 10.299591064453125, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.022470647716420426, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00011235323858210213, | |
| "grad_norm": 0.7895200848579407, | |
| "learning_rate": 1e-05, | |
| "loss": 41.5096, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00011235323858210213, | |
| "eval_loss": 10.377470016479492, | |
| "eval_runtime": 33.6003, | |
| "eval_samples_per_second": 446.127, | |
| "eval_steps_per_second": 111.547, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00022470647716420425, | |
| "grad_norm": 0.9913174510002136, | |
| "learning_rate": 2e-05, | |
| "loss": 41.5086, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00033705971574630637, | |
| "grad_norm": 0.7206482291221619, | |
| "learning_rate": 3e-05, | |
| "loss": 41.5074, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0004494129543284085, | |
| "grad_norm": 0.662643313407898, | |
| "learning_rate": 4e-05, | |
| "loss": 41.5175, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0005617661929105106, | |
| "grad_norm": 0.7218053936958313, | |
| "learning_rate": 5e-05, | |
| "loss": 41.5079, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0006741194314926127, | |
| "grad_norm": 0.6399303674697876, | |
| "learning_rate": 6e-05, | |
| "loss": 41.5161, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0007864726700747149, | |
| "grad_norm": 0.7161200642585754, | |
| "learning_rate": 7e-05, | |
| "loss": 41.5059, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.000898825908656817, | |
| "grad_norm": 0.6918237805366516, | |
| "learning_rate": 8e-05, | |
| "loss": 41.4916, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0010111791472389193, | |
| "grad_norm": 0.7956416606903076, | |
| "learning_rate": 9e-05, | |
| "loss": 41.4964, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0011235323858210212, | |
| "grad_norm": 0.7649855613708496, | |
| "learning_rate": 0.0001, | |
| "loss": 41.4898, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0012358856244031233, | |
| "grad_norm": 0.7298173308372498, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 41.4934, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0013482388629852255, | |
| "grad_norm": 0.7867535352706909, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 41.4863, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0014605921015673276, | |
| "grad_norm": 0.7472080588340759, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 41.4949, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0015729453401494297, | |
| "grad_norm": 0.8375926613807678, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 41.473, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0016852985787315319, | |
| "grad_norm": 0.9645178318023682, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 41.4645, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.001797651817313634, | |
| "grad_norm": 0.873009979724884, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 41.4774, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0019100050558957362, | |
| "grad_norm": 1.0542374849319458, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 41.465, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0020223582944778385, | |
| "grad_norm": 1.1131774187088013, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 41.4576, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0021347115330599404, | |
| "grad_norm": 1.0464072227478027, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 41.4605, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0022470647716420424, | |
| "grad_norm": 1.1138325929641724, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 41.4469, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0023594180102241447, | |
| "grad_norm": 1.220119595527649, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 41.4464, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0024717712488062466, | |
| "grad_norm": 1.2734307050704956, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 41.4181, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.002584124487388349, | |
| "grad_norm": 1.2603442668914795, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 41.4222, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.002696477725970451, | |
| "grad_norm": 1.3185153007507324, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 41.4272, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0028088309645525533, | |
| "grad_norm": 1.395378828048706, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 41.395, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002921184203134655, | |
| "grad_norm": 1.6009327173233032, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 41.3968, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0030335374417167576, | |
| "grad_norm": 1.40463125705719, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 41.4032, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0031458906802988595, | |
| "grad_norm": 1.5156066417694092, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 41.3763, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.003258243918880962, | |
| "grad_norm": 1.4074209928512573, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 41.3804, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0033705971574630638, | |
| "grad_norm": 1.4947224855422974, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 41.3735, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.003482950396045166, | |
| "grad_norm": 1.530563473701477, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 41.3528, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.003595303634627268, | |
| "grad_norm": 1.4782463312149048, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 41.3481, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0037076568732093704, | |
| "grad_norm": 1.3531701564788818, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 41.3561, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0038200101117914723, | |
| "grad_norm": 1.4341721534729004, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 41.3418, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.003932363350373574, | |
| "grad_norm": 1.305496096611023, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 41.3488, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.004044716588955677, | |
| "grad_norm": 1.396409273147583, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 41.3284, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.004157069827537779, | |
| "grad_norm": 1.3208515644073486, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 41.319, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.004269423066119881, | |
| "grad_norm": 1.1543635129928589, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 41.3125, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.004381776304701983, | |
| "grad_norm": 1.211503505706787, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 41.2789, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.004494129543284085, | |
| "grad_norm": 1.1140539646148682, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 41.2995, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0046064827818661875, | |
| "grad_norm": 1.090532898902893, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 41.2762, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0047188360204482895, | |
| "grad_norm": 1.0199702978134155, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 41.2719, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.004831189259030391, | |
| "grad_norm": 0.9350115060806274, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 41.261, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004943542497612493, | |
| "grad_norm": 0.8293979167938232, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 41.2873, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.005055895736194596, | |
| "grad_norm": 0.8430503010749817, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 41.2577, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.005168248974776698, | |
| "grad_norm": 0.7887256145477295, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 41.268, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0052806022133588, | |
| "grad_norm": 0.7197476029396057, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 41.2757, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.005392955451940902, | |
| "grad_norm": 0.6839483380317688, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 41.2757, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.005505308690523005, | |
| "grad_norm": 0.6647058725357056, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 41.2651, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.005617661929105107, | |
| "grad_norm": 0.6115227341651917, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 41.2641, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005617661929105107, | |
| "eval_loss": 10.310123443603516, | |
| "eval_runtime": 33.4566, | |
| "eval_samples_per_second": 448.043, | |
| "eval_steps_per_second": 112.026, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0057300151676872085, | |
| "grad_norm": 0.6150594353675842, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 41.2491, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.00584236840626931, | |
| "grad_norm": 0.8687107563018799, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 41.2237, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.005954721644851413, | |
| "grad_norm": 0.6435952186584473, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 41.2483, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.006067074883433515, | |
| "grad_norm": 0.5283822417259216, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 41.2414, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.006179428122015617, | |
| "grad_norm": 0.4348739981651306, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 41.246, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.006291781360597719, | |
| "grad_norm": 0.5016875267028809, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 41.2213, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.006404134599179822, | |
| "grad_norm": 0.4395579993724823, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 41.2227, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.006516487837761924, | |
| "grad_norm": 0.34251657128334045, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 41.2453, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.006628841076344026, | |
| "grad_norm": 0.3462662994861603, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 41.2241, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0067411943149261275, | |
| "grad_norm": 0.3944358825683594, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 41.2466, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0068535475535082295, | |
| "grad_norm": 0.32348212599754333, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 41.2382, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.006965900792090332, | |
| "grad_norm": 0.3583115041255951, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 41.2094, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.007078254030672434, | |
| "grad_norm": 0.32442328333854675, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 41.226, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.007190607269254536, | |
| "grad_norm": 0.3217829763889313, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 41.2424, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.007302960507836638, | |
| "grad_norm": 0.2875845730304718, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 41.2034, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.007415313746418741, | |
| "grad_norm": 0.33438655734062195, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 41.233, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.007527666985000843, | |
| "grad_norm": 0.2421981543302536, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 41.2046, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.007640020223582945, | |
| "grad_norm": 0.3216751217842102, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 41.2256, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.007752373462165047, | |
| "grad_norm": 0.24948103725910187, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 41.2153, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.007864726700747149, | |
| "grad_norm": 0.2150658518075943, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 41.2212, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.007977079939329251, | |
| "grad_norm": 0.23460710048675537, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 41.2334, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.008089433177911354, | |
| "grad_norm": 0.2817407548427582, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 41.2255, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.008201786416493455, | |
| "grad_norm": 0.2687859535217285, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 41.2259, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.008314139655075558, | |
| "grad_norm": 0.23189355432987213, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 41.1947, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.008426492893657659, | |
| "grad_norm": 0.27154698967933655, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 41.2248, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.008538846132239762, | |
| "grad_norm": 0.23277752101421356, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 41.2132, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.008651199370821865, | |
| "grad_norm": 0.2773878276348114, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 41.1993, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.008763552609403966, | |
| "grad_norm": 0.24345317482948303, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 41.2276, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.008875905847986068, | |
| "grad_norm": 0.2572079002857208, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 41.22, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.00898825908656817, | |
| "grad_norm": 0.3237912058830261, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 41.2147, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.009100612325150272, | |
| "grad_norm": 0.33336275815963745, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 41.1953, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.009212965563732375, | |
| "grad_norm": 0.29136842489242554, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 41.2258, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.009325318802314476, | |
| "grad_norm": 0.2960778772830963, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 41.2313, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.009437672040896579, | |
| "grad_norm": 0.22867703437805176, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 41.2079, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.009550025279478682, | |
| "grad_norm": 0.21541887521743774, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 41.2244, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.009662378518060783, | |
| "grad_norm": 0.2785476744174957, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 41.215, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.009774731756642886, | |
| "grad_norm": 0.29661858081817627, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 41.2145, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.009887084995224987, | |
| "grad_norm": 0.22034361958503723, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 41.1994, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.00999943823380709, | |
| "grad_norm": 0.2419750690460205, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 41.2124, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.010111791472389192, | |
| "grad_norm": 0.29091349244117737, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 41.2122, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.010224144710971293, | |
| "grad_norm": 0.2149897962808609, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 41.2126, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.010336497949553396, | |
| "grad_norm": 0.2682669460773468, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 41.2212, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.010448851188135499, | |
| "grad_norm": 0.26379066705703735, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 41.2084, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0105612044267176, | |
| "grad_norm": 0.29168516397476196, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 41.2061, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.010673557665299703, | |
| "grad_norm": 0.2567015588283539, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 41.2245, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.010785910903881804, | |
| "grad_norm": 0.25660285353660583, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 41.1983, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.010898264142463907, | |
| "grad_norm": 0.2569798231124878, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 41.1958, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.01101061738104601, | |
| "grad_norm": 0.3190382421016693, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 41.2096, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.01112297061962811, | |
| "grad_norm": 0.3212287127971649, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 41.233, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.011235323858210213, | |
| "grad_norm": 0.38987791538238525, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 41.2378, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.011235323858210213, | |
| "eval_loss": 10.302728652954102, | |
| "eval_runtime": 33.4303, | |
| "eval_samples_per_second": 448.396, | |
| "eval_steps_per_second": 112.114, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.011347677096792314, | |
| "grad_norm": 0.48272770643234253, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 41.1792, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.011460030335374417, | |
| "grad_norm": 0.6370357871055603, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 41.1893, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.01157238357395652, | |
| "grad_norm": 0.48485246300697327, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 41.2053, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.01168473681253862, | |
| "grad_norm": 0.3337991535663605, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 41.2021, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.011797090051120724, | |
| "grad_norm": 0.35370728373527527, | |
| "learning_rate": 5e-05, | |
| "loss": 41.2011, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.011909443289702826, | |
| "grad_norm": 0.2889159023761749, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 41.2157, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.012021796528284927, | |
| "grad_norm": 0.2480829358100891, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 41.21, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.01213414976686703, | |
| "grad_norm": 0.37383803725242615, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 41.1819, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.012246503005449131, | |
| "grad_norm": 0.24803395569324493, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 41.224, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.012358856244031234, | |
| "grad_norm": 0.24491317570209503, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 41.2214, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.012471209482613337, | |
| "grad_norm": 0.32773756980895996, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 41.1884, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.012583562721195438, | |
| "grad_norm": 0.2478872686624527, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 41.2284, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.01269591595977754, | |
| "grad_norm": 0.312561571598053, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 41.2039, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.012808269198359644, | |
| "grad_norm": 0.22887419164180756, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 41.2022, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.012920622436941745, | |
| "grad_norm": 0.2290506660938263, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 41.203, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.013032975675523847, | |
| "grad_norm": 0.33003920316696167, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 41.2017, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.013145328914105948, | |
| "grad_norm": 0.22017890214920044, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 41.2118, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.013257682152688051, | |
| "grad_norm": 0.30501383543014526, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 41.2039, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.013370035391270154, | |
| "grad_norm": 0.22522194683551788, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 41.2196, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.013482388629852255, | |
| "grad_norm": 0.28780046105384827, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 41.2061, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.013594741868434358, | |
| "grad_norm": 0.25612780451774597, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 41.2006, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.013707095107016459, | |
| "grad_norm": 0.2527892291545868, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 41.2054, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.013819448345598562, | |
| "grad_norm": 0.2517006993293762, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 41.1992, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.013931801584180665, | |
| "grad_norm": 0.23754863440990448, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 41.2099, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.014044154822762766, | |
| "grad_norm": 0.17713980376720428, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 41.1868, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.014156508061344868, | |
| "grad_norm": 0.24930499494075775, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 41.2081, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.014268861299926971, | |
| "grad_norm": 0.25248026847839355, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 41.2137, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.014381214538509072, | |
| "grad_norm": 0.29542839527130127, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 41.2147, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.014493567777091175, | |
| "grad_norm": 0.280669242143631, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 41.2003, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.014605921015673276, | |
| "grad_norm": 0.23233091831207275, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 41.2121, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.014718274254255379, | |
| "grad_norm": 0.29116538166999817, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 41.1926, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.014830627492837482, | |
| "grad_norm": 0.24843358993530273, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 41.1912, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.014942980731419583, | |
| "grad_norm": 0.2787754535675049, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 41.1993, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.015055333970001685, | |
| "grad_norm": 0.2419159710407257, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 41.2065, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.015167687208583788, | |
| "grad_norm": 0.24362392723560333, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 41.203, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.01528004044716589, | |
| "grad_norm": 0.3140493333339691, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 41.2071, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.015392393685747992, | |
| "grad_norm": 0.2296908050775528, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 41.2123, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.015504746924330093, | |
| "grad_norm": 0.26365184783935547, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 41.1996, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.015617100162912196, | |
| "grad_norm": 0.2602303922176361, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 41.2149, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.015729453401494297, | |
| "grad_norm": 0.27265465259552, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 41.2025, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0158418066400764, | |
| "grad_norm": 0.26106128096580505, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 41.2142, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.015954159878658503, | |
| "grad_norm": 0.23775847256183624, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 41.21, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.016066513117240604, | |
| "grad_norm": 0.34526339173316956, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 41.2214, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.016178866355822708, | |
| "grad_norm": 0.2909034788608551, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 41.2204, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.01629121959440481, | |
| "grad_norm": 0.2971731126308441, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 41.1924, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.01640357283298691, | |
| "grad_norm": 0.30135422945022583, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 41.207, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.01651592607156901, | |
| "grad_norm": 0.3517146110534668, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 41.2104, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.016628279310151116, | |
| "grad_norm": 0.36096063256263733, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 41.2063, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.016740632548733217, | |
| "grad_norm": 0.35298091173171997, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 41.2134, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.016852985787315318, | |
| "grad_norm": 0.4046837091445923, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 41.2229, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.016852985787315318, | |
| "eval_loss": 10.299948692321777, | |
| "eval_runtime": 33.4249, | |
| "eval_samples_per_second": 448.469, | |
| "eval_steps_per_second": 112.132, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.016965339025897423, | |
| "grad_norm": 0.555423378944397, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 41.2134, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.017077692264479524, | |
| "grad_norm": 0.9470621347427368, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 41.1938, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.017190045503061625, | |
| "grad_norm": 0.3544475734233856, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 41.1992, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.01730239874164373, | |
| "grad_norm": 0.3507763147354126, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 41.1992, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.01741475198022583, | |
| "grad_norm": 0.19748862087726593, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 41.1931, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.01752710521880793, | |
| "grad_norm": 0.24769514799118042, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 41.1975, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.017639458457390036, | |
| "grad_norm": 0.26125532388687134, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 41.2074, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.017751811695972137, | |
| "grad_norm": 0.23368193209171295, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 41.2148, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.017864164934554238, | |
| "grad_norm": 0.30703428387641907, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 41.1982, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.01797651817313634, | |
| "grad_norm": 0.3379155993461609, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 41.1927, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.018088871411718443, | |
| "grad_norm": 0.2728177309036255, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 41.1945, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.018201224650300545, | |
| "grad_norm": 0.25143495202064514, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 41.1999, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.018313577888882646, | |
| "grad_norm": 0.21429970860481262, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 41.2015, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.01842593112746475, | |
| "grad_norm": 0.18250975012779236, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 41.1981, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.01853828436604685, | |
| "grad_norm": 0.22810083627700806, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 41.1893, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.018650637604628952, | |
| "grad_norm": 0.1829170137643814, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 41.1964, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.018762990843211057, | |
| "grad_norm": 0.27752968668937683, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 41.2134, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.018875344081793158, | |
| "grad_norm": 0.29091545939445496, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 41.176, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.01898769732037526, | |
| "grad_norm": 0.18208208680152893, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 41.1901, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.019100050558957363, | |
| "grad_norm": 0.252201110124588, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 41.1909, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.019212403797539464, | |
| "grad_norm": 0.25969040393829346, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 41.2022, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.019324757036121566, | |
| "grad_norm": 0.1629827320575714, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 41.1896, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.01943711027470367, | |
| "grad_norm": 0.26366856694221497, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 41.1785, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.01954946351328577, | |
| "grad_norm": 0.2201000452041626, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 41.1834, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.019661816751867872, | |
| "grad_norm": 0.2458728551864624, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 41.1676, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.019774169990449973, | |
| "grad_norm": 0.3383263945579529, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 41.1803, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.019886523229032078, | |
| "grad_norm": 0.18288184702396393, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 41.1895, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.01999887646761418, | |
| "grad_norm": 0.22058157622814178, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 41.1888, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.02011122970619628, | |
| "grad_norm": 0.23787496984004974, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 41.1883, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.020223582944778384, | |
| "grad_norm": 0.24095582962036133, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 41.2071, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.020335936183360485, | |
| "grad_norm": 0.2212057262659073, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 41.2083, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.020448289421942586, | |
| "grad_norm": 0.2541032135486603, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 41.192, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.02056064266052469, | |
| "grad_norm": 0.2529408931732178, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 41.2063, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.020672995899106792, | |
| "grad_norm": 0.28434687852859497, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 41.1902, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.020785349137688893, | |
| "grad_norm": 0.21067148447036743, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 41.2087, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.020897702376270998, | |
| "grad_norm": 0.2179185003042221, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 41.186, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.0210100556148531, | |
| "grad_norm": 0.30646952986717224, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 41.1871, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.0211224088534352, | |
| "grad_norm": 0.22822514176368713, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 41.1927, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.0212347620920173, | |
| "grad_norm": 0.24493247270584106, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 41.2095, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.021347115330599405, | |
| "grad_norm": 0.22439813613891602, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 41.1894, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.021459468569181506, | |
| "grad_norm": 0.284005343914032, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 41.2045, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.021571821807763607, | |
| "grad_norm": 0.27038300037384033, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 41.2204, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.021684175046345712, | |
| "grad_norm": 0.24226807057857513, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 41.1851, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.021796528284927813, | |
| "grad_norm": 0.30808719992637634, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 41.1783, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.021908881523509914, | |
| "grad_norm": 0.3134829103946686, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 41.2111, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.02202123476209202, | |
| "grad_norm": 0.31079569458961487, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 41.2272, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.02213358800067412, | |
| "grad_norm": 0.3545907437801361, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 41.1982, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.02224594123925622, | |
| "grad_norm": 0.3386935591697693, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 41.2099, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.022358294477838325, | |
| "grad_norm": 0.3554247319698334, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 41.2165, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.022470647716420426, | |
| "grad_norm": 0.39326632022857666, | |
| "learning_rate": 0.0, | |
| "loss": 41.2458, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.022470647716420426, | |
| "eval_loss": 10.299591064453125, | |
| "eval_runtime": 33.5205, | |
| "eval_samples_per_second": 447.189, | |
| "eval_steps_per_second": 111.812, | |
| "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": 44841416785920.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |