| { | |
| "best_metric": 2.272221326828003, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.11743981209630065, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005871990604815032, | |
| "grad_norm": 0.3585398495197296, | |
| "learning_rate": 1e-05, | |
| "loss": 1.9092, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0005871990604815032, | |
| "eval_loss": 2.6059012413024902, | |
| "eval_runtime": 211.3894, | |
| "eval_samples_per_second": 13.567, | |
| "eval_steps_per_second": 3.392, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0011743981209630064, | |
| "grad_norm": 0.44403478503227234, | |
| "learning_rate": 2e-05, | |
| "loss": 2.3702, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0017615971814445098, | |
| "grad_norm": 0.5587884783744812, | |
| "learning_rate": 3e-05, | |
| "loss": 2.3615, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002348796241926013, | |
| "grad_norm": 0.42974773049354553, | |
| "learning_rate": 4e-05, | |
| "loss": 2.2911, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.002935995302407516, | |
| "grad_norm": 0.5544111132621765, | |
| "learning_rate": 5e-05, | |
| "loss": 2.4218, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0035231943628890195, | |
| "grad_norm": 0.40412697196006775, | |
| "learning_rate": 6e-05, | |
| "loss": 2.4455, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.004110393423370522, | |
| "grad_norm": 0.43639397621154785, | |
| "learning_rate": 7e-05, | |
| "loss": 2.1368, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.004697592483852026, | |
| "grad_norm": 0.5224621891975403, | |
| "learning_rate": 8e-05, | |
| "loss": 2.4772, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.005284791544333529, | |
| "grad_norm": 0.48703229427337646, | |
| "learning_rate": 9e-05, | |
| "loss": 2.3771, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.005871990604815032, | |
| "grad_norm": 0.50340336561203, | |
| "learning_rate": 0.0001, | |
| "loss": 2.3702, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0064591896652965355, | |
| "grad_norm": 0.5083776712417603, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 2.4099, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.007046388725778039, | |
| "grad_norm": 0.516063928604126, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.3497, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.007633587786259542, | |
| "grad_norm": 0.5593276619911194, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 2.2577, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.008220786846741044, | |
| "grad_norm": 0.5569037795066833, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 2.2737, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.008807985907222548, | |
| "grad_norm": 0.5233477354049683, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 2.0727, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.009395184967704051, | |
| "grad_norm": 0.4936712682247162, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 2.2392, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.009982384028185555, | |
| "grad_norm": 0.5432935953140259, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 2.2063, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.010569583088667059, | |
| "grad_norm": 0.5101926922798157, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 2.3664, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.011156782149148562, | |
| "grad_norm": 0.4693228006362915, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 2.4251, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.011743981209630064, | |
| "grad_norm": 0.5099998116493225, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 2.4441, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.012331180270111567, | |
| "grad_norm": 0.4890613853931427, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 2.4425, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.012918379330593071, | |
| "grad_norm": 0.5973317623138428, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 2.244, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.013505578391074574, | |
| "grad_norm": 0.4584880769252777, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 2.1937, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.014092777451556078, | |
| "grad_norm": 0.49515780806541443, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 2.1623, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.014679976512037582, | |
| "grad_norm": 0.4772243797779083, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 2.3036, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.015267175572519083, | |
| "grad_norm": 0.567406415939331, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 2.3457, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.015854374633000587, | |
| "grad_norm": 0.5111720561981201, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 2.1962, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.01644157369348209, | |
| "grad_norm": 0.5667634010314941, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 2.3624, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.017028772753963594, | |
| "grad_norm": 0.5059242844581604, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 2.2448, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.017615971814445096, | |
| "grad_norm": 0.5338335633277893, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 2.1469, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0182031708749266, | |
| "grad_norm": 0.5313173532485962, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 2.22, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.018790369935408103, | |
| "grad_norm": 0.5506137609481812, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.2337, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.019377568995889608, | |
| "grad_norm": 0.6006201505661011, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 2.1631, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.01996476805637111, | |
| "grad_norm": 0.569302499294281, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.1311, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.020551967116852612, | |
| "grad_norm": 0.6763449907302856, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 2.2711, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.021139166177334117, | |
| "grad_norm": 0.6627179384231567, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.2971, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.02172636523781562, | |
| "grad_norm": 0.6061697602272034, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.1958, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.022313564298297124, | |
| "grad_norm": 0.6627572774887085, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.1163, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.022900763358778626, | |
| "grad_norm": 0.6236354112625122, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.2256, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.023487962419260128, | |
| "grad_norm": 0.8026069402694702, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 2.5571, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.024075161479741633, | |
| "grad_norm": 0.6401921510696411, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.2009, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.024662360540223135, | |
| "grad_norm": 0.8171783685684204, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.3007, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.02524955960070464, | |
| "grad_norm": 0.7011866569519043, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 2.1921, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.025836758661186142, | |
| "grad_norm": 0.7044079899787903, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.2387, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.026423957721667644, | |
| "grad_norm": 0.7611087560653687, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.2652, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02701115678214915, | |
| "grad_norm": 0.8123400807380676, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.3523, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02759835584263065, | |
| "grad_norm": 0.8470413088798523, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.3106, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.028185554903112156, | |
| "grad_norm": 0.8900328874588013, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.2621, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.028772753963593658, | |
| "grad_norm": 1.0494381189346313, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.252, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.029359953024075163, | |
| "grad_norm": 1.3739956617355347, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.4864, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.029359953024075163, | |
| "eval_loss": 2.3259637355804443, | |
| "eval_runtime": 212.5837, | |
| "eval_samples_per_second": 13.491, | |
| "eval_steps_per_second": 3.373, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.029947152084556665, | |
| "grad_norm": 0.4676729738712311, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.3446, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.030534351145038167, | |
| "grad_norm": 0.7390397787094116, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.7258, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.031121550205519672, | |
| "grad_norm": 0.5213898420333862, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.275, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.031708749266001174, | |
| "grad_norm": 0.4595617949962616, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 2.3287, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03229594832648268, | |
| "grad_norm": 0.3932957947254181, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 2.1543, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03288314738696418, | |
| "grad_norm": 0.37599682807922363, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 2.3102, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03347034644744568, | |
| "grad_norm": 0.40485823154449463, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 2.2687, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03405754550792719, | |
| "grad_norm": 0.3882083594799042, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 2.3044, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03464474456840869, | |
| "grad_norm": 0.37490957975387573, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 2.294, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03523194362889019, | |
| "grad_norm": 0.3471771776676178, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 2.1857, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0358191426893717, | |
| "grad_norm": 0.441429078578949, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 2.2638, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0364063417498532, | |
| "grad_norm": 0.44914543628692627, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 2.4695, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0369935408103347, | |
| "grad_norm": 0.44081437587738037, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 2.221, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.037580739870816206, | |
| "grad_norm": 0.48773181438446045, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 2.4275, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03816793893129771, | |
| "grad_norm": 0.45748454332351685, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 2.3778, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.038755137991779216, | |
| "grad_norm": 0.3910389542579651, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 2.236, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.039342337052260715, | |
| "grad_norm": 0.45711007714271545, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 2.2863, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03992953611274222, | |
| "grad_norm": 0.4611903131008148, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 2.2532, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.040516735173223725, | |
| "grad_norm": 0.4584813117980957, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.8862, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.041103934233705224, | |
| "grad_norm": 0.4744361340999603, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 2.0616, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04169113329418673, | |
| "grad_norm": 0.4039895832538605, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 2.1366, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.042278332354668234, | |
| "grad_norm": 0.43253761529922485, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 2.1594, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.04286553141514973, | |
| "grad_norm": 0.5091691017150879, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 2.1627, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.04345273047563124, | |
| "grad_norm": 0.4764954745769501, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 2.1508, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.04403992953611274, | |
| "grad_norm": 0.49679040908813477, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 2.284, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04462712859659425, | |
| "grad_norm": 0.5358265042304993, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 2.331, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.04521432765707575, | |
| "grad_norm": 0.5226441025733948, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 2.2219, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.04580152671755725, | |
| "grad_norm": 0.5147382616996765, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.9872, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.04638872577803876, | |
| "grad_norm": 0.5225254893302917, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 2.0704, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.046975924838520255, | |
| "grad_norm": 0.5500540137290955, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 2.2335, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04756312389900176, | |
| "grad_norm": 0.5460954308509827, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 2.3627, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.048150322959483266, | |
| "grad_norm": 0.5873216986656189, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 2.0461, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.04873752201996477, | |
| "grad_norm": 0.5911714434623718, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 2.2967, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.04932472108044627, | |
| "grad_norm": 0.5971319675445557, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.3029, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.049911920140927775, | |
| "grad_norm": 0.644014298915863, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 2.3323, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.05049911920140928, | |
| "grad_norm": 0.6283090710639954, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 2.2745, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.05108631826189078, | |
| "grad_norm": 0.6029326319694519, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 2.4444, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.051673517322372284, | |
| "grad_norm": 0.6873989701271057, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 2.3448, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.05226071638285379, | |
| "grad_norm": 0.6990385055541992, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 2.3652, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.05284791544333529, | |
| "grad_norm": 0.6101760864257812, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 2.1092, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05343511450381679, | |
| "grad_norm": 0.6388440132141113, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 2.3159, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0540223135642983, | |
| "grad_norm": 0.707058310508728, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 2.3384, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0546095126247798, | |
| "grad_norm": 0.7057024836540222, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 2.2795, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0551967116852613, | |
| "grad_norm": 0.691746175289154, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 2.3113, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.05578391074574281, | |
| "grad_norm": 0.8184278011322021, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 2.349, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.05637110980622431, | |
| "grad_norm": 0.7519662976264954, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 2.5017, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.05695830886670581, | |
| "grad_norm": 0.7975960373878479, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.1257, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.057545507927187316, | |
| "grad_norm": 0.9032453298568726, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.5218, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.05813270698766882, | |
| "grad_norm": 1.034710168838501, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.2292, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.058719906048150326, | |
| "grad_norm": 1.1983397006988525, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 2.5094, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.058719906048150326, | |
| "eval_loss": 2.288942575454712, | |
| "eval_runtime": 212.8731, | |
| "eval_samples_per_second": 13.473, | |
| "eval_steps_per_second": 3.368, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.059307105108631825, | |
| "grad_norm": 0.3455316424369812, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.3934, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.05989430416911333, | |
| "grad_norm": 0.4336451292037964, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 2.5784, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.060481503229594835, | |
| "grad_norm": 0.42090684175491333, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.2209, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.061068702290076333, | |
| "grad_norm": 0.42842987179756165, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.4051, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.06165590135055784, | |
| "grad_norm": 0.43164893984794617, | |
| "learning_rate": 5e-05, | |
| "loss": 2.2905, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.062243100411039344, | |
| "grad_norm": 0.4323105216026306, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 2.3339, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.06283029947152084, | |
| "grad_norm": 0.37328964471817017, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 2.235, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.06341749853200235, | |
| "grad_norm": 0.3794722259044647, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 2.149, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.06400469759248385, | |
| "grad_norm": 0.41767585277557373, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 2.328, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.06459189665296536, | |
| "grad_norm": 0.4514842927455902, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 2.1455, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06517909571344686, | |
| "grad_norm": 0.4334796071052551, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 2.3044, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.06576629477392835, | |
| "grad_norm": 0.41178274154663086, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 2.3794, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.06635349383440986, | |
| "grad_norm": 0.4317842721939087, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 2.3946, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.06694069289489137, | |
| "grad_norm": 0.3932304084300995, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 2.303, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.06752789195537287, | |
| "grad_norm": 0.39550313353538513, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 2.1241, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06811509101585438, | |
| "grad_norm": 0.4141201972961426, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 2.1957, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06870229007633588, | |
| "grad_norm": 0.4293804466724396, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 2.4381, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06928948913681739, | |
| "grad_norm": 0.43157073855400085, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 2.2291, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.06987668819729888, | |
| "grad_norm": 0.43993470072746277, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 2.1711, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.07046388725778038, | |
| "grad_norm": 0.45777156949043274, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 2.1552, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07105108631826189, | |
| "grad_norm": 0.46806374192237854, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 2.2253, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.0716382853787434, | |
| "grad_norm": 0.46243131160736084, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 2.1679, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0722254844392249, | |
| "grad_norm": 0.4713747203350067, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 2.3878, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0728126834997064, | |
| "grad_norm": 0.476028174161911, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 2.238, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.07339988256018791, | |
| "grad_norm": 0.44079750776290894, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 2.2885, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0739870816206694, | |
| "grad_norm": 0.4735269248485565, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 2.2382, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0745742806811509, | |
| "grad_norm": 0.4485139846801758, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 2.235, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.07516147974163241, | |
| "grad_norm": 0.5356859564781189, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 2.2264, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.07574867880211392, | |
| "grad_norm": 0.4632677733898163, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 2.1834, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.07633587786259542, | |
| "grad_norm": 0.5824257731437683, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 2.2549, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07692307692307693, | |
| "grad_norm": 0.4714558720588684, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 2.2587, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07751027598355843, | |
| "grad_norm": 0.5169317126274109, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 2.4345, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.07809747504403992, | |
| "grad_norm": 0.5915501117706299, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 2.4772, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07868467410452143, | |
| "grad_norm": 0.5694190859794617, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 2.3725, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.07927187316500293, | |
| "grad_norm": 0.5802375078201294, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 2.3258, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07985907222548444, | |
| "grad_norm": 0.6188993453979492, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 2.1346, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.08044627128596595, | |
| "grad_norm": 0.5669954419136047, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 2.1948, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.08103347034644745, | |
| "grad_norm": 0.5923700928688049, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 2.1646, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.08162066940692896, | |
| "grad_norm": 0.6563289165496826, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 2.2861, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.08220786846741045, | |
| "grad_norm": 0.6520470380783081, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.2801, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08279506752789195, | |
| "grad_norm": 0.7126626968383789, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 2.3817, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.08338226658837346, | |
| "grad_norm": 0.673728883266449, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 2.2924, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.08396946564885496, | |
| "grad_norm": 0.6358984708786011, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 2.1587, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.08455666470933647, | |
| "grad_norm": 0.6919076442718506, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 2.3918, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.08514386376981797, | |
| "grad_norm": 0.7889071106910706, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.3347, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.08573106283029946, | |
| "grad_norm": 0.7909374833106995, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 2.3217, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.08631826189078097, | |
| "grad_norm": 0.8272628784179688, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 2.4197, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.08690546095126248, | |
| "grad_norm": 0.8077692985534668, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.3445, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.08749266001174398, | |
| "grad_norm": 0.9683035016059875, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 2.3626, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.08807985907222549, | |
| "grad_norm": 1.163604736328125, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 2.5513, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08807985907222549, | |
| "eval_loss": 2.274216651916504, | |
| "eval_runtime": 213.2036, | |
| "eval_samples_per_second": 13.452, | |
| "eval_steps_per_second": 3.363, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08866705813270699, | |
| "grad_norm": 0.2833799421787262, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 2.0172, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.0892542571931885, | |
| "grad_norm": 0.37640324234962463, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 2.4128, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.08984145625366999, | |
| "grad_norm": 0.32096025347709656, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.0291, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.0904286553141515, | |
| "grad_norm": 0.3295486867427826, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 2.1376, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.091015854374633, | |
| "grad_norm": 0.3889763057231903, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 2.0726, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.0916030534351145, | |
| "grad_norm": 0.31805261969566345, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 2.3021, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.09219025249559601, | |
| "grad_norm": 0.34619516134262085, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 2.2774, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.09277745155607751, | |
| "grad_norm": 0.39927756786346436, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 2.0669, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.09336465061655902, | |
| "grad_norm": 0.4164828956127167, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 2.2204, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.09395184967704051, | |
| "grad_norm": 0.4049561619758606, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 2.2342, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09453904873752202, | |
| "grad_norm": 0.3853633999824524, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 2.2424, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.09512624779800352, | |
| "grad_norm": 0.367906391620636, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 2.1947, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.09571344685848503, | |
| "grad_norm": 0.4073794484138489, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 2.2868, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.09630064591896653, | |
| "grad_norm": 0.40216854214668274, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 2.1953, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.09688784497944804, | |
| "grad_norm": 0.40012452006340027, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 2.1738, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.09747504403992954, | |
| "grad_norm": 0.4341662526130676, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 2.2553, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.09806224310041103, | |
| "grad_norm": 0.47913673520088196, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 2.4157, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.09864944216089254, | |
| "grad_norm": 0.3952808380126953, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 2.1639, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.09923664122137404, | |
| "grad_norm": 0.36768385767936707, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 2.1705, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.09982384028185555, | |
| "grad_norm": 0.44818249344825745, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 2.2073, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.10041103934233706, | |
| "grad_norm": 0.455099880695343, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 2.4348, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.10099823840281856, | |
| "grad_norm": 0.4131917953491211, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 2.2421, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.10158543746330007, | |
| "grad_norm": 0.45736247301101685, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 2.3026, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.10217263652378156, | |
| "grad_norm": 0.4742620289325714, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 2.173, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.10275983558426306, | |
| "grad_norm": 0.5509735941886902, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 2.3251, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.10334703464474457, | |
| "grad_norm": 0.503662645816803, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 2.3581, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.10393423370522607, | |
| "grad_norm": 0.48229506611824036, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 2.3267, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.10452143276570758, | |
| "grad_norm": 0.5001346468925476, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 2.2507, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.10510863182618908, | |
| "grad_norm": 0.48915812373161316, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 2.3136, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.10569583088667057, | |
| "grad_norm": 0.5704674124717712, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 2.3694, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.10628302994715208, | |
| "grad_norm": 0.5387436747550964, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 2.2656, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.10687022900763359, | |
| "grad_norm": 0.5216829776763916, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 2.1838, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.10745742806811509, | |
| "grad_norm": 0.5243988633155823, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 2.2033, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.1080446271285966, | |
| "grad_norm": 0.5314877033233643, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 2.2078, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.1086318261890781, | |
| "grad_norm": 0.5182973742485046, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 2.2261, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1092190252495596, | |
| "grad_norm": 0.5910155177116394, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 2.2255, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.1098062243100411, | |
| "grad_norm": 0.5409195423126221, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 2.2809, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1103934233705226, | |
| "grad_norm": 0.6241095066070557, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 2.3466, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.11098062243100411, | |
| "grad_norm": 0.6316850185394287, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 2.0514, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.11156782149148561, | |
| "grad_norm": 0.6229705810546875, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.9125, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.11215502055196712, | |
| "grad_norm": 0.645051896572113, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 2.3178, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.11274221961244862, | |
| "grad_norm": 0.6645575165748596, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 2.2531, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.11332941867293013, | |
| "grad_norm": 0.780520498752594, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 2.1451, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.11391661773341162, | |
| "grad_norm": 0.7688522338867188, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 2.126, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.11450381679389313, | |
| "grad_norm": 0.7503068447113037, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.2481, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.11509101585437463, | |
| "grad_norm": 0.768911600112915, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 2.4033, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.11567821491485614, | |
| "grad_norm": 0.8218636512756348, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.4349, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.11626541397533764, | |
| "grad_norm": 0.9500688314437866, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.5446, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.11685261303581915, | |
| "grad_norm": 1.0050915479660034, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 2.4041, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.11743981209630065, | |
| "grad_norm": 1.283392310142517, | |
| "learning_rate": 0.0, | |
| "loss": 2.6971, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11743981209630065, | |
| "eval_loss": 2.272221326828003, | |
| "eval_runtime": 213.2333, | |
| "eval_samples_per_second": 13.45, | |
| "eval_steps_per_second": 3.363, | |
| "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": 2.936200135232717e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |