| { | |
| "best_metric": 0.31977999210357666, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.007874325760856726, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 3.937162880428363e-05, | |
| "grad_norm": 0.1473366916179657, | |
| "learning_rate": 1e-05, | |
| "loss": 0.4842, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 3.937162880428363e-05, | |
| "eval_loss": 0.6386224031448364, | |
| "eval_runtime": 1995.7368, | |
| "eval_samples_per_second": 21.435, | |
| "eval_steps_per_second": 5.359, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 7.874325760856726e-05, | |
| "grad_norm": 0.1501932293176651, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4723, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0001181148864128509, | |
| "grad_norm": 0.15301698446273804, | |
| "learning_rate": 3e-05, | |
| "loss": 0.465, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00015748651521713453, | |
| "grad_norm": 0.15504051744937897, | |
| "learning_rate": 4e-05, | |
| "loss": 0.4613, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00019685814402141817, | |
| "grad_norm": 0.17802225053310394, | |
| "learning_rate": 5e-05, | |
| "loss": 0.4573, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0002362297728257018, | |
| "grad_norm": 0.16927273571491241, | |
| "learning_rate": 6e-05, | |
| "loss": 0.4521, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00027560140162998545, | |
| "grad_norm": 0.17416727542877197, | |
| "learning_rate": 7e-05, | |
| "loss": 0.4688, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00031497303043426906, | |
| "grad_norm": 0.19252535700798035, | |
| "learning_rate": 8e-05, | |
| "loss": 0.4869, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0003543446592385527, | |
| "grad_norm": 0.21720175445079803, | |
| "learning_rate": 9e-05, | |
| "loss": 0.4392, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00039371628804283634, | |
| "grad_norm": 0.20339831709861755, | |
| "learning_rate": 0.0001, | |
| "loss": 0.4673, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.00043308791684711995, | |
| "grad_norm": 0.188543438911438, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.4404, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0004724595456514036, | |
| "grad_norm": 0.1840866506099701, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.4308, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0005118311744556872, | |
| "grad_norm": 0.1921333223581314, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.4258, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0005512028032599709, | |
| "grad_norm": 0.1822589933872223, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.4197, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0005905744320642545, | |
| "grad_norm": 0.17518241703510284, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.4372, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0006299460608685381, | |
| "grad_norm": 0.16869305074214935, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.4088, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0006693176896728218, | |
| "grad_norm": 0.18317943811416626, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.4121, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0007086893184771055, | |
| "grad_norm": 0.1914009302854538, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.4111, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.000748060947281389, | |
| "grad_norm": 0.17840923368930817, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.4001, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0007874325760856727, | |
| "grad_norm": 0.19282005727291107, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.4029, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0008268042048899563, | |
| "grad_norm": 0.2061193436384201, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.3986, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0008661758336942399, | |
| "grad_norm": 0.20973272621631622, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.4132, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0009055474624985236, | |
| "grad_norm": 0.20912733674049377, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.393, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0009449190913028072, | |
| "grad_norm": 0.20087695121765137, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.3943, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0009842907201070908, | |
| "grad_norm": 0.2549019157886505, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.4077, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0010236623489113745, | |
| "grad_norm": 0.21350446343421936, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.3767, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0010630339777156581, | |
| "grad_norm": 0.2284364104270935, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.3941, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0011024056065199418, | |
| "grad_norm": 0.20515067875385284, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.3623, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0011417772353242255, | |
| "grad_norm": 0.22672885656356812, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.3734, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.001181148864128509, | |
| "grad_norm": 0.2223028987646103, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.3831, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0012205204929327926, | |
| "grad_norm": 0.22611401975154877, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.3728, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0012598921217370762, | |
| "grad_norm": 0.23607665300369263, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.387, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.00129926375054136, | |
| "grad_norm": 0.24884606897830963, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.3784, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0013386353793456436, | |
| "grad_norm": 0.2478945106267929, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.3671, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0013780070081499272, | |
| "grad_norm": 0.2744622230529785, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.3852, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.001417378636954211, | |
| "grad_norm": 0.2842346727848053, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.3984, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0014567502657584943, | |
| "grad_norm": 0.3080979585647583, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.4233, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.001496121894562778, | |
| "grad_norm": 0.32813560962677, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.3906, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0015354935233670617, | |
| "grad_norm": 0.3377130627632141, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.3866, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0015748651521713453, | |
| "grad_norm": 0.38843271136283875, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.4104, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.001614236780975629, | |
| "grad_norm": 0.3918204605579376, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.4434, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0016536084097799127, | |
| "grad_norm": 0.4351727068424225, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.4251, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0016929800385841961, | |
| "grad_norm": 0.5044992566108704, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.4586, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0017323516673884798, | |
| "grad_norm": 0.4990330636501312, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.4392, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0017717232961927635, | |
| "grad_norm": 0.5468814373016357, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.5086, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0018110949249970471, | |
| "grad_norm": 0.6528873443603516, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.4761, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0018504665538013308, | |
| "grad_norm": 0.7145413160324097, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.4981, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0018898381826056145, | |
| "grad_norm": 0.8748348951339722, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.4825, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0019292098114098981, | |
| "grad_norm": 1.0368186235427856, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.5143, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0019685814402141816, | |
| "grad_norm": 1.806100606918335, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.736, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0019685814402141816, | |
| "eval_loss": 0.4826611280441284, | |
| "eval_runtime": 2007.0233, | |
| "eval_samples_per_second": 21.314, | |
| "eval_steps_per_second": 5.329, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0020079530690184652, | |
| "grad_norm": 1.0947028398513794, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.5242, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.002047324697822749, | |
| "grad_norm": 0.5624424815177917, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.4548, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0020866963266270326, | |
| "grad_norm": 0.2854950428009033, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.4229, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0021260679554313162, | |
| "grad_norm": 0.22005236148834229, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.3959, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0021654395842356, | |
| "grad_norm": 0.19781219959259033, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.3665, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0022048112130398836, | |
| "grad_norm": 0.22003516554832458, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.3942, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0022441828418441672, | |
| "grad_norm": 0.216252401471138, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.3848, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.002283554470648451, | |
| "grad_norm": 0.21777230501174927, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.3959, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0023229260994527346, | |
| "grad_norm": 0.19707585871219635, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.3946, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.002362297728257018, | |
| "grad_norm": 0.1851402223110199, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.3931, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0024016693570613015, | |
| "grad_norm": 0.18103234469890594, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.3735, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.002441040985865585, | |
| "grad_norm": 0.1712011694908142, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.3537, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.002480412614669869, | |
| "grad_norm": 0.1616038978099823, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.3538, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0025197842434741525, | |
| "grad_norm": 0.17303889989852905, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.3652, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.002559155872278436, | |
| "grad_norm": 0.16791899502277374, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.3413, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.00259852750108272, | |
| "grad_norm": 0.1742825210094452, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.3591, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0026378991298870035, | |
| "grad_norm": 0.17425836622714996, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.3635, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.002677270758691287, | |
| "grad_norm": 0.17511220276355743, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.3388, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.002716642387495571, | |
| "grad_norm": 0.18989098072052002, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.3329, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0027560140162998545, | |
| "grad_norm": 0.18335279822349548, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.3634, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.002795385645104138, | |
| "grad_norm": 0.18777930736541748, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.3692, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.002834757273908422, | |
| "grad_norm": 0.18051090836524963, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.3142, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.002874128902712705, | |
| "grad_norm": 0.1953955888748169, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.3689, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0029135005315169887, | |
| "grad_norm": 0.1748814731836319, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.3368, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0029528721603212724, | |
| "grad_norm": 0.1853368729352951, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.3304, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.002992243789125556, | |
| "grad_norm": 0.18087562918663025, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.3334, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0030316154179298397, | |
| "grad_norm": 0.19581517577171326, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.332, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0030709870467341234, | |
| "grad_norm": 0.19563588500022888, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.3247, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.003110358675538407, | |
| "grad_norm": 0.21290044486522675, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.3609, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0031497303043426907, | |
| "grad_norm": 0.21674351394176483, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.354, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0031891019331469744, | |
| "grad_norm": 0.2117038518190384, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.3241, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.003228473561951258, | |
| "grad_norm": 0.22736124694347382, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.348, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0032678451907555417, | |
| "grad_norm": 0.23222015798091888, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.3394, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0033072168195598254, | |
| "grad_norm": 0.2689361870288849, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.3648, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.003346588448364109, | |
| "grad_norm": 0.24705283343791962, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.3355, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0033859600771683923, | |
| "grad_norm": 0.27266886830329895, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.3538, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.003425331705972676, | |
| "grad_norm": 0.29763978719711304, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.3448, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0034647033347769596, | |
| "grad_norm": 0.2855512797832489, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.3248, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0035040749635812433, | |
| "grad_norm": 0.32624325156211853, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.3722, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.003543446592385527, | |
| "grad_norm": 0.3541692793369293, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.3849, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0035828182211898106, | |
| "grad_norm": 0.3887896239757538, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.3821, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0036221898499940943, | |
| "grad_norm": 0.3992058336734772, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.3537, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.003661561478798378, | |
| "grad_norm": 0.4371652901172638, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.3808, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0037009331076026616, | |
| "grad_norm": 0.4928395748138428, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.3659, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0037403047364069453, | |
| "grad_norm": 0.6321387887001038, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.4392, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.003779676365211229, | |
| "grad_norm": 0.563918948173523, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.3722, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0038190479940155126, | |
| "grad_norm": 0.8165143728256226, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.4351, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0038584196228197963, | |
| "grad_norm": 1.0398285388946533, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.5044, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0038977912516240795, | |
| "grad_norm": 0.9911220669746399, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.5425, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.003937162880428363, | |
| "grad_norm": 1.6528089046478271, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.5023, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.003937162880428363, | |
| "eval_loss": 0.3683101534843445, | |
| "eval_runtime": 2003.406, | |
| "eval_samples_per_second": 21.353, | |
| "eval_steps_per_second": 5.338, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.003976534509232647, | |
| "grad_norm": 0.26657024025917053, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.3908, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.0040159061380369305, | |
| "grad_norm": 0.2466580718755722, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.3603, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.004055277766841214, | |
| "grad_norm": 0.23641517758369446, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.3722, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.004094649395645498, | |
| "grad_norm": 0.2994157671928406, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.3512, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0041340210244497815, | |
| "grad_norm": 0.2846444249153137, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3321, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.004173392653254065, | |
| "grad_norm": 0.2845078110694885, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.3675, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.004212764282058349, | |
| "grad_norm": 0.2055649608373642, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.3502, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.0042521359108626325, | |
| "grad_norm": 0.1906391680240631, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.3404, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.004291507539666916, | |
| "grad_norm": 0.1860475391149521, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.3344, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0043308791684712, | |
| "grad_norm": 0.18555738031864166, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.3305, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0043702507972754835, | |
| "grad_norm": 0.17265374958515167, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.3406, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.004409622426079767, | |
| "grad_norm": 0.18200285732746124, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.3268, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.004448994054884051, | |
| "grad_norm": 0.19033251702785492, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.3339, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0044883656836883345, | |
| "grad_norm": 0.17964288592338562, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.3245, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.004527737312492618, | |
| "grad_norm": 0.17820070683956146, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.3052, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.004567108941296902, | |
| "grad_norm": 0.19013427197933197, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.3423, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.0046064805701011855, | |
| "grad_norm": 0.18814770877361298, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.3079, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.004645852198905469, | |
| "grad_norm": 0.20084728300571442, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.3473, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.004685223827709752, | |
| "grad_norm": 0.18739016354084015, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.3218, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.004724595456514036, | |
| "grad_norm": 0.17784281075000763, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.3113, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.004763967085318319, | |
| "grad_norm": 0.20050129294395447, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.3167, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.004803338714122603, | |
| "grad_norm": 0.19596628844738007, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.3359, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.004842710342926887, | |
| "grad_norm": 0.19288507103919983, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.3275, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.00488208197173117, | |
| "grad_norm": 0.19886533915996552, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.3084, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.004921453600535454, | |
| "grad_norm": 0.20502859354019165, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.3176, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.004960825229339738, | |
| "grad_norm": 0.19561141729354858, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.3103, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.005000196858144021, | |
| "grad_norm": 0.20386740565299988, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.3262, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.005039568486948305, | |
| "grad_norm": 0.20647361874580383, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.2928, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.005078940115752589, | |
| "grad_norm": 0.21491894125938416, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.2924, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.005118311744556872, | |
| "grad_norm": 0.22360754013061523, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.3037, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.005157683373361156, | |
| "grad_norm": 0.23474211990833282, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.3147, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.00519705500216544, | |
| "grad_norm": 0.2296513170003891, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.2977, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.005236426630969723, | |
| "grad_norm": 0.25748276710510254, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.3343, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.005275798259774007, | |
| "grad_norm": 0.2597656548023224, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.3016, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.005315169888578291, | |
| "grad_norm": 0.24971474707126617, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.3027, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.005354541517382574, | |
| "grad_norm": 0.2793932259082794, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.3127, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.005393913146186858, | |
| "grad_norm": 0.28339770436286926, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.2968, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.005433284774991142, | |
| "grad_norm": 0.28424811363220215, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.3255, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.005472656403795425, | |
| "grad_norm": 0.32081717252731323, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.3317, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.005512028032599709, | |
| "grad_norm": 0.3422316610813141, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.3119, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.005551399661403993, | |
| "grad_norm": 0.3551773726940155, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.302, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.005590771290208276, | |
| "grad_norm": 0.4034905731678009, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3604, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.00563014291901256, | |
| "grad_norm": 0.44910168647766113, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.3203, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.005669514547816844, | |
| "grad_norm": 0.5052204728126526, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.3396, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.005708886176621126, | |
| "grad_norm": 0.5537760853767395, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.3932, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.00574825780542541, | |
| "grad_norm": 0.5518684387207031, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.342, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.005787629434229694, | |
| "grad_norm": 0.5985448956489563, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.3385, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.005827001063033977, | |
| "grad_norm": 0.7689756751060486, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.4143, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.005866372691838261, | |
| "grad_norm": 0.8520933985710144, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.4063, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.005905744320642545, | |
| "grad_norm": 1.3966264724731445, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.4739, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005905744320642545, | |
| "eval_loss": 0.32784563302993774, | |
| "eval_runtime": 2006.7661, | |
| "eval_samples_per_second": 21.317, | |
| "eval_steps_per_second": 5.329, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005945115949446828, | |
| "grad_norm": 0.1980389505624771, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.3431, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.005984487578251112, | |
| "grad_norm": 0.21009264886379242, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.3294, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.006023859207055396, | |
| "grad_norm": 0.19129659235477448, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.3527, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.006063230835859679, | |
| "grad_norm": 0.19261766970157623, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.3279, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.006102602464663963, | |
| "grad_norm": 0.21031305193901062, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.3152, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.006141974093468247, | |
| "grad_norm": 0.1911141574382782, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.305, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.00618134572227253, | |
| "grad_norm": 0.19922904670238495, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.3164, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.006220717351076814, | |
| "grad_norm": 0.17719711363315582, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.2898, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.006260088979881098, | |
| "grad_norm": 0.1834423691034317, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.3052, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.006299460608685381, | |
| "grad_norm": 0.19143041968345642, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.3004, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.006338832237489665, | |
| "grad_norm": 0.19105836749076843, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.3004, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.006378203866293949, | |
| "grad_norm": 0.18660342693328857, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.3291, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.006417575495098232, | |
| "grad_norm": 0.18928730487823486, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.2937, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.006456947123902516, | |
| "grad_norm": 0.18767684698104858, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.2991, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.0064963187527068, | |
| "grad_norm": 0.19432400166988373, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.3069, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.006535690381511083, | |
| "grad_norm": 0.1916324943304062, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.3045, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.006575062010315367, | |
| "grad_norm": 0.1847633421421051, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.2928, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.006614433639119651, | |
| "grad_norm": 0.19647832214832306, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.2997, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.006653805267923934, | |
| "grad_norm": 0.1917283535003662, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.2891, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.006693176896728218, | |
| "grad_norm": 0.19362805783748627, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.2855, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.006732548525532501, | |
| "grad_norm": 0.2030460387468338, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.3008, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.0067719201543367845, | |
| "grad_norm": 0.19901475310325623, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.2823, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.006811291783141068, | |
| "grad_norm": 0.2076079100370407, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.304, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.006850663411945352, | |
| "grad_norm": 0.2194894552230835, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.3082, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0068900350407496355, | |
| "grad_norm": 0.21589970588684082, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.2918, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.006929406669553919, | |
| "grad_norm": 0.2137822061777115, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.306, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.006968778298358203, | |
| "grad_norm": 0.21752864122390747, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.2966, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.0070081499271624865, | |
| "grad_norm": 0.23289024829864502, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.3155, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.00704752155596677, | |
| "grad_norm": 0.22937798500061035, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.3086, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.007086893184771054, | |
| "grad_norm": 0.2303188145160675, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.2987, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0071262648135753375, | |
| "grad_norm": 0.24133794009685516, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.3011, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.007165636442379621, | |
| "grad_norm": 0.25773799419403076, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.2869, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.007205008071183905, | |
| "grad_norm": 0.2506771683692932, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.3007, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.0072443796999881885, | |
| "grad_norm": 0.2606258690357208, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.3206, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.007283751328792472, | |
| "grad_norm": 0.2843955457210541, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.3079, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.007323122957596756, | |
| "grad_norm": 0.29535767436027527, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.322, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.0073624945864010395, | |
| "grad_norm": 0.2905426025390625, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.3118, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.007401866215205323, | |
| "grad_norm": 0.308418333530426, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.3303, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.007441237844009607, | |
| "grad_norm": 0.3305756449699402, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.3066, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.0074806094728138905, | |
| "grad_norm": 0.3653267025947571, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.3329, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.007519981101618174, | |
| "grad_norm": 0.37498483061790466, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.3383, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.007559352730422458, | |
| "grad_norm": 0.3785691559314728, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.2991, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.0075987243592267415, | |
| "grad_norm": 0.42409980297088623, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.3424, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.007638095988031025, | |
| "grad_norm": 0.48371052742004395, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.391, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.007677467616835309, | |
| "grad_norm": 0.5230972170829773, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.3592, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0077168392456395925, | |
| "grad_norm": 0.5310702323913574, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.3495, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.007756210874443875, | |
| "grad_norm": 0.6330666542053223, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.3257, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.007795582503248159, | |
| "grad_norm": 0.9409751892089844, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.4532, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.007834954132052444, | |
| "grad_norm": 1.0332026481628418, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.4504, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.007874325760856726, | |
| "grad_norm": 1.4735428094863892, | |
| "learning_rate": 0.0, | |
| "loss": 0.5622, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.007874325760856726, | |
| "eval_loss": 0.31977999210357666, | |
| "eval_runtime": 2004.5375, | |
| "eval_samples_per_second": 21.341, | |
| "eval_steps_per_second": 5.335, | |
| "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": 1.439681531609088e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |