| { | |
| "best_metric": 1.4074208736419678, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.1172504763300601, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005862523816503004, | |
| "grad_norm": 10.790068626403809, | |
| "learning_rate": 1e-05, | |
| "loss": 7.5692, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0005862523816503004, | |
| "eval_loss": 1.6656559705734253, | |
| "eval_runtime": 219.6882, | |
| "eval_samples_per_second": 13.078, | |
| "eval_steps_per_second": 3.273, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0011725047633006009, | |
| "grad_norm": 13.74497127532959, | |
| "learning_rate": 2e-05, | |
| "loss": 7.597, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0017587571449509014, | |
| "grad_norm": 7.357011318206787, | |
| "learning_rate": 3e-05, | |
| "loss": 6.9459, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0023450095266012018, | |
| "grad_norm": 4.228321075439453, | |
| "learning_rate": 4e-05, | |
| "loss": 5.9879, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0029312619082515023, | |
| "grad_norm": 3.995352029800415, | |
| "learning_rate": 5e-05, | |
| "loss": 5.753, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.003517514289901803, | |
| "grad_norm": 5.382306098937988, | |
| "learning_rate": 6e-05, | |
| "loss": 5.2265, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.004103766671552103, | |
| "grad_norm": 3.1584348678588867, | |
| "learning_rate": 7e-05, | |
| "loss": 5.2468, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0046900190532024036, | |
| "grad_norm": 3.3004236221313477, | |
| "learning_rate": 8e-05, | |
| "loss": 5.5786, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.005276271434852704, | |
| "grad_norm": 5.079125881195068, | |
| "learning_rate": 9e-05, | |
| "loss": 5.344, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.005862523816503005, | |
| "grad_norm": 3.284970760345459, | |
| "learning_rate": 0.0001, | |
| "loss": 5.2413, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.006448776198153305, | |
| "grad_norm": 3.535767078399658, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 4.6326, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.007035028579803606, | |
| "grad_norm": 2.8821194171905518, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 5.4681, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.007621280961453906, | |
| "grad_norm": 3.262932300567627, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 5.2982, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.008207533343104206, | |
| "grad_norm": 2.9881551265716553, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 5.1053, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.008793785724754507, | |
| "grad_norm": 3.3377323150634766, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 5.1543, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.009380038106404807, | |
| "grad_norm": 4.034030914306641, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 5.7338, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.009966290488055108, | |
| "grad_norm": 3.1818180084228516, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 5.6751, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.010552542869705408, | |
| "grad_norm": 3.360145330429077, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 5.2622, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.011138795251355709, | |
| "grad_norm": 2.7319986820220947, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 5.5215, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01172504763300601, | |
| "grad_norm": 3.641176700592041, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 5.4878, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01231130001465631, | |
| "grad_norm": 4.311220169067383, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 5.1884, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.01289755239630661, | |
| "grad_norm": 3.1506717205047607, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 4.8736, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.013483804777956911, | |
| "grad_norm": 4.287765026092529, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 5.5322, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.014070057159607212, | |
| "grad_norm": 4.625144958496094, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 5.439, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.014656309541257512, | |
| "grad_norm": 3.5584959983825684, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 5.7938, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.015242561922907813, | |
| "grad_norm": 3.129434823989868, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 5.6654, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01582881430455811, | |
| "grad_norm": 3.3299741744995117, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 6.0676, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.016415066686208412, | |
| "grad_norm": 3.497907876968384, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 5.6101, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.017001319067858713, | |
| "grad_norm": 3.3982856273651123, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 5.7022, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.017587571449509013, | |
| "grad_norm": 3.324096441268921, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 5.8835, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.018173823831159314, | |
| "grad_norm": 4.0818657875061035, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 5.4996, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.018760076212809614, | |
| "grad_norm": 3.6348886489868164, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 5.315, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.019346328594459915, | |
| "grad_norm": 3.49786114692688, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 6.5054, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.019932580976110215, | |
| "grad_norm": 3.781508684158325, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 5.586, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.020518833357760516, | |
| "grad_norm": 3.5593245029449463, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 6.0674, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.021105085739410816, | |
| "grad_norm": 4.080981731414795, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 5.1878, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.021691338121061117, | |
| "grad_norm": 4.535512924194336, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 6.0247, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.022277590502711418, | |
| "grad_norm": 4.254596710205078, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 5.2376, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.022863842884361718, | |
| "grad_norm": 3.7518248558044434, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 5.0529, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.02345009526601202, | |
| "grad_norm": 3.532460927963257, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 5.6947, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02403634764766232, | |
| "grad_norm": 3.9862725734710693, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 5.0275, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.02462260002931262, | |
| "grad_norm": 4.2363996505737305, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 6.0092, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.02520885241096292, | |
| "grad_norm": 4.181338787078857, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 6.353, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.02579510479261322, | |
| "grad_norm": 3.624467372894287, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 6.1128, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.02638135717426352, | |
| "grad_norm": 4.963085651397705, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 5.7576, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.026967609555913822, | |
| "grad_norm": 4.274786472320557, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 5.7228, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.027553861937564122, | |
| "grad_norm": 4.970536231994629, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 6.0499, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.028140114319214423, | |
| "grad_norm": 5.130246162414551, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 6.6606, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.028726366700864724, | |
| "grad_norm": 4.710168361663818, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 5.8428, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.029312619082515024, | |
| "grad_norm": 5.519105434417725, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 6.9914, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.029312619082515024, | |
| "eval_loss": 1.5797480344772339, | |
| "eval_runtime": 220.6377, | |
| "eval_samples_per_second": 13.021, | |
| "eval_steps_per_second": 3.259, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.029898871464165325, | |
| "grad_norm": 22.379613876342773, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 7.3968, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.030485123845815625, | |
| "grad_norm": 4.978916168212891, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 6.3433, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.031071376227465926, | |
| "grad_norm": 2.9572982788085938, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 5.0145, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.03165762860911622, | |
| "grad_norm": 2.445566177368164, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 4.924, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03224388099076653, | |
| "grad_norm": 2.6169345378875732, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 5.3782, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.032830133372416824, | |
| "grad_norm": 2.9169564247131348, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 5.3261, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03341638575406713, | |
| "grad_norm": 2.4703657627105713, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 5.0871, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.034002638135717425, | |
| "grad_norm": 2.2803452014923096, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 5.2733, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03458889051736773, | |
| "grad_norm": 2.362588882446289, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 5.878, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.035175142899018026, | |
| "grad_norm": 2.4082915782928467, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 5.5574, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03576139528066833, | |
| "grad_norm": 2.3328030109405518, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 4.9929, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.03634764766231863, | |
| "grad_norm": 2.7598860263824463, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 4.8533, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03693390004396893, | |
| "grad_norm": 2.2203588485717773, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 5.5964, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.03752015242561923, | |
| "grad_norm": 3.713977813720703, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 5.5132, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03810640480726953, | |
| "grad_norm": 2.6936147212982178, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 5.0176, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03869265718891983, | |
| "grad_norm": 2.662566900253296, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 4.8258, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.039278909570570134, | |
| "grad_norm": 2.557370662689209, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 5.1206, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03986516195222043, | |
| "grad_norm": 2.9368200302124023, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 5.6975, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.040451414333870735, | |
| "grad_norm": 2.838097333908081, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 5.9734, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.04103766671552103, | |
| "grad_norm": 2.5451722145080566, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 5.9255, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04162391909717133, | |
| "grad_norm": 2.319904088973999, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 5.4505, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.04221017147882163, | |
| "grad_norm": 2.583216905593872, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 5.8085, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.04279642386047193, | |
| "grad_norm": 2.594492197036743, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 5.4678, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.043382676242122234, | |
| "grad_norm": 2.4850809574127197, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 5.6634, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.04396892862377253, | |
| "grad_norm": 2.865980863571167, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 5.6128, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.044555181005422835, | |
| "grad_norm": 2.6326639652252197, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 5.291, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.04514143338707313, | |
| "grad_norm": 3.4590983390808105, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 5.3853, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.045727685768723436, | |
| "grad_norm": 3.297605276107788, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 5.5379, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.04631393815037373, | |
| "grad_norm": 3.1267645359039307, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 5.6926, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.04690019053202404, | |
| "grad_norm": 2.7454946041107178, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 5.9848, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.047486442913674334, | |
| "grad_norm": 2.782520055770874, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 6.2118, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.04807269529532464, | |
| "grad_norm": 3.1769840717315674, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 5.6901, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.048658947676974935, | |
| "grad_norm": 3.196343421936035, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 5.5507, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.04924520005862524, | |
| "grad_norm": 3.4306981563568115, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 5.4153, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.04983145244027554, | |
| "grad_norm": 3.2837769985198975, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 5.9451, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.05041770482192584, | |
| "grad_norm": 3.1064860820770264, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 5.8944, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.05100395720357614, | |
| "grad_norm": 2.9329092502593994, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 6.0029, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.05159020958522644, | |
| "grad_norm": 3.3825430870056152, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 5.7889, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.05217646196687674, | |
| "grad_norm": 3.9273457527160645, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 6.0082, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.05276271434852704, | |
| "grad_norm": 3.520132541656494, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 6.082, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05334896673017734, | |
| "grad_norm": 3.166968822479248, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 5.8006, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.053935219111827644, | |
| "grad_norm": 3.3038203716278076, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 5.7244, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.05452147149347794, | |
| "grad_norm": 3.7368345260620117, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 6.0354, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.055107723875128245, | |
| "grad_norm": 3.527898073196411, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 5.8237, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.05569397625677854, | |
| "grad_norm": 3.3549344539642334, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 6.2314, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.056280228638428846, | |
| "grad_norm": 4.186188220977783, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 5.8171, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.05686648102007914, | |
| "grad_norm": 4.852516174316406, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 6.041, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05745273340172945, | |
| "grad_norm": 4.0406813621521, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 6.0713, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.058038985783379744, | |
| "grad_norm": 4.6055684089660645, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 6.9265, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.05862523816503005, | |
| "grad_norm": 4.351954460144043, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 6.2784, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05862523816503005, | |
| "eval_loss": 1.4935845136642456, | |
| "eval_runtime": 220.3196, | |
| "eval_samples_per_second": 13.04, | |
| "eval_steps_per_second": 3.263, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.059211490546680345, | |
| "grad_norm": 9.79791259765625, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 7.2952, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.05979774292833065, | |
| "grad_norm": 5.464766502380371, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 6.8585, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.060383995309980946, | |
| "grad_norm": 2.3852691650390625, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 5.6773, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.06097024769163125, | |
| "grad_norm": 1.962914228439331, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 5.5284, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.06155650007328155, | |
| "grad_norm": 1.9527621269226074, | |
| "learning_rate": 5e-05, | |
| "loss": 4.8592, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.06214275245493185, | |
| "grad_norm": 2.1389286518096924, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 5.8863, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.06272900483658214, | |
| "grad_norm": 2.2018775939941406, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 4.9969, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.06331525721823245, | |
| "grad_norm": 2.167783260345459, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 5.0025, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.06390150959988275, | |
| "grad_norm": 2.3369650840759277, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 5.2025, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.06448776198153305, | |
| "grad_norm": 2.4558358192443848, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 5.5729, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06507401436318334, | |
| "grad_norm": 2.8261640071868896, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 5.6174, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.06566026674483365, | |
| "grad_norm": 2.195913553237915, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 5.6233, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.06624651912648395, | |
| "grad_norm": 3.3884429931640625, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 5.3429, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.06683277150813426, | |
| "grad_norm": 2.5708835124969482, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 5.4112, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.06741902388978455, | |
| "grad_norm": 2.4030284881591797, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 5.5757, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06800527627143485, | |
| "grad_norm": 2.3454060554504395, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 5.2617, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06859152865308515, | |
| "grad_norm": 2.5141963958740234, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 5.5981, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06917778103473546, | |
| "grad_norm": 2.672102212905884, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 5.4798, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.06976403341638575, | |
| "grad_norm": 3.4174704551696777, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 5.8241, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.07035028579803605, | |
| "grad_norm": 3.075779914855957, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 5.5416, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07093653817968636, | |
| "grad_norm": 2.761280059814453, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 6.0596, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.07152279056133666, | |
| "grad_norm": 4.131796360015869, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 5.8447, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.07210904294298695, | |
| "grad_norm": 2.837559223175049, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 5.4954, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.07269529532463725, | |
| "grad_norm": 2.8359930515289307, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 5.9241, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.07328154770628756, | |
| "grad_norm": 2.619260311126709, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 5.4931, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07386780008793786, | |
| "grad_norm": 3.2195513248443604, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 5.9272, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.07445405246958815, | |
| "grad_norm": 3.0083348751068115, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 5.4017, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.07504030485123846, | |
| "grad_norm": 2.5995583534240723, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 5.2977, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.07562655723288876, | |
| "grad_norm": 2.560567855834961, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 5.5441, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.07621280961453906, | |
| "grad_norm": 2.335968017578125, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 4.9449, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07679906199618936, | |
| "grad_norm": 2.7276153564453125, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 5.5969, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07738531437783966, | |
| "grad_norm": 2.488765239715576, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 5.5089, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.07797156675948996, | |
| "grad_norm": 2.9440059661865234, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 6.0276, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07855781914114027, | |
| "grad_norm": 2.68922758102417, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 6.5581, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.07914407152279056, | |
| "grad_norm": 6.488649368286133, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 5.0164, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07973032390444086, | |
| "grad_norm": 3.010985851287842, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 5.4577, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.08031657628609117, | |
| "grad_norm": 3.4043078422546387, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 5.5744, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.08090282866774147, | |
| "grad_norm": 2.711311101913452, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 5.5894, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.08148908104939176, | |
| "grad_norm": 3.1877779960632324, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 5.6716, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.08207533343104206, | |
| "grad_norm": 3.315678119659424, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 5.4803, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08266158581269237, | |
| "grad_norm": 3.313769578933716, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 6.0857, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.08324783819434266, | |
| "grad_norm": 4.4940948486328125, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 5.6264, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.08383409057599296, | |
| "grad_norm": 3.380831241607666, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 6.185, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.08442034295764327, | |
| "grad_norm": 4.156842231750488, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 6.1022, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.08500659533929357, | |
| "grad_norm": 6.136685371398926, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 5.7662, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.08559284772094386, | |
| "grad_norm": 3.665255308151245, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 6.3179, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.08617910010259416, | |
| "grad_norm": 3.6559813022613525, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 6.5996, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.08676535248424447, | |
| "grad_norm": 4.137120246887207, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 6.3454, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.08735160486589477, | |
| "grad_norm": 3.7026383876800537, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 6.7606, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.08793785724754506, | |
| "grad_norm": 4.525060653686523, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 7.3796, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08793785724754506, | |
| "eval_loss": 1.4337459802627563, | |
| "eval_runtime": 220.6933, | |
| "eval_samples_per_second": 13.018, | |
| "eval_steps_per_second": 3.258, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08852410962919537, | |
| "grad_norm": 3.3688249588012695, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 7.5831, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.08911036201084567, | |
| "grad_norm": 3.6086671352386475, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 6.3033, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.08969661439249597, | |
| "grad_norm": 3.4656951427459717, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 6.119, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.09028286677414626, | |
| "grad_norm": 2.5899691581726074, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 5.5605, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.09086911915579657, | |
| "grad_norm": 1.8436894416809082, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 5.4129, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.09145537153744687, | |
| "grad_norm": 2.8457846641540527, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 4.918, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.09204162391909718, | |
| "grad_norm": 1.8666106462478638, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 5.8424, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.09262787630074747, | |
| "grad_norm": 1.856207251548767, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 5.2801, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.09321412868239777, | |
| "grad_norm": 2.319056510925293, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 4.3907, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.09380038106404807, | |
| "grad_norm": 2.191591739654541, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 4.6345, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09438663344569838, | |
| "grad_norm": 2.2874443531036377, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 4.7251, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.09497288582734867, | |
| "grad_norm": 2.7336089611053467, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 5.1857, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.09555913820899897, | |
| "grad_norm": 2.612793207168579, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 5.0392, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.09614539059064928, | |
| "grad_norm": 2.5980582237243652, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 5.5705, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.09673164297229958, | |
| "grad_norm": 2.2050070762634277, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 5.6282, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.09731789535394987, | |
| "grad_norm": 2.4113943576812744, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 5.4068, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.09790414773560017, | |
| "grad_norm": 3.2364604473114014, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 4.9435, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.09849040011725048, | |
| "grad_norm": 2.5000576972961426, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 5.3538, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.09907665249890078, | |
| "grad_norm": 4.209569931030273, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 5.2727, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.09966290488055107, | |
| "grad_norm": 2.4928956031799316, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 4.9601, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.10024915726220138, | |
| "grad_norm": 2.3254218101501465, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 5.7522, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.10083540964385168, | |
| "grad_norm": 2.6643009185791016, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 5.3077, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.10142166202550199, | |
| "grad_norm": 2.2104291915893555, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 5.4648, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.10200791440715228, | |
| "grad_norm": 2.7385032176971436, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 5.3233, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.10259416678880258, | |
| "grad_norm": 2.514474630355835, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 5.4704, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.10318041917045288, | |
| "grad_norm": 2.497436285018921, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 5.6888, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.10376667155210317, | |
| "grad_norm": 2.505242109298706, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 5.9652, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.10435292393375348, | |
| "grad_norm": 2.7065258026123047, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 6.0959, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.10493917631540378, | |
| "grad_norm": 3.1412546634674072, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 6.3321, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.10552542869705409, | |
| "grad_norm": 2.5169122219085693, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 5.5505, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.10611168107870438, | |
| "grad_norm": 2.7921128273010254, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 6.1436, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.10669793346035468, | |
| "grad_norm": 2.6502459049224854, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 5.7164, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.10728418584200498, | |
| "grad_norm": 2.849736452102661, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 5.5606, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.10787043822365529, | |
| "grad_norm": 2.7651307582855225, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 5.8407, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.10845669060530558, | |
| "grad_norm": 2.776927947998047, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 5.288, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.10904294298695588, | |
| "grad_norm": 2.73980712890625, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 6.0007, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.10962919536860619, | |
| "grad_norm": 3.1120150089263916, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 5.8268, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.11021544775025649, | |
| "grad_norm": 3.4573416709899902, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 5.9987, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.11080170013190678, | |
| "grad_norm": 3.418168306350708, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 6.6219, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.11138795251355708, | |
| "grad_norm": 2.986539602279663, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 5.6802, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.11197420489520739, | |
| "grad_norm": 3.3587698936462402, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 6.2014, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.11256045727685769, | |
| "grad_norm": 3.8468992710113525, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 6.0434, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.11314670965850798, | |
| "grad_norm": 3.0598528385162354, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 6.0316, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.11373296204015829, | |
| "grad_norm": 3.564034938812256, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 6.2851, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.11431921442180859, | |
| "grad_norm": 3.7003591060638428, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 6.0102, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1149054668034589, | |
| "grad_norm": 3.6101059913635254, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 5.9999, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.11549171918510918, | |
| "grad_norm": 3.709547996520996, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 7.0065, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.11607797156675949, | |
| "grad_norm": 3.9578347206115723, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 6.1499, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.11666422394840979, | |
| "grad_norm": 4.960752964019775, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 6.0623, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.1172504763300601, | |
| "grad_norm": 4.955458641052246, | |
| "learning_rate": 0.0, | |
| "loss": 6.9019, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1172504763300601, | |
| "eval_loss": 1.4074208736419678, | |
| "eval_runtime": 220.6205, | |
| "eval_samples_per_second": 13.022, | |
| "eval_steps_per_second": 3.259, | |
| "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": 2.9455509164025446e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |