| { | |
| "best_metric": 1.4959101676940918, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.24592683676606208, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0012296341838303104, | |
| "grad_norm": 11.878715515136719, | |
| "learning_rate": 1e-05, | |
| "loss": 8.59, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0012296341838303104, | |
| "eval_loss": 7.924351215362549, | |
| "eval_runtime": 64.9653, | |
| "eval_samples_per_second": 21.088, | |
| "eval_steps_per_second": 5.28, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.002459268367660621, | |
| "grad_norm": 11.992634773254395, | |
| "learning_rate": 2e-05, | |
| "loss": 7.4167, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0036889025514909315, | |
| "grad_norm": 10.548018455505371, | |
| "learning_rate": 3e-05, | |
| "loss": 7.0841, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.004918536735321242, | |
| "grad_norm": 10.55744457244873, | |
| "learning_rate": 4e-05, | |
| "loss": 7.3392, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006148170919151552, | |
| "grad_norm": 9.278156280517578, | |
| "learning_rate": 5e-05, | |
| "loss": 6.6823, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007377805102981863, | |
| "grad_norm": 7.4441046714782715, | |
| "learning_rate": 6e-05, | |
| "loss": 6.1153, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.008607439286812173, | |
| "grad_norm": 6.773711204528809, | |
| "learning_rate": 7e-05, | |
| "loss": 4.9143, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.009837073470642483, | |
| "grad_norm": 7.2493577003479, | |
| "learning_rate": 8e-05, | |
| "loss": 4.7519, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.011066707654472794, | |
| "grad_norm": 5.57631778717041, | |
| "learning_rate": 9e-05, | |
| "loss": 3.4949, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.012296341838303105, | |
| "grad_norm": 5.323232173919678, | |
| "learning_rate": 0.0001, | |
| "loss": 3.6786, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.013525976022133415, | |
| "grad_norm": 5.450096130371094, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 2.8657, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.014755610205963726, | |
| "grad_norm": 7.339489459991455, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.3501, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.015985244389794036, | |
| "grad_norm": 5.6941142082214355, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.7784, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.017214878573624345, | |
| "grad_norm": 3.8295962810516357, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.7433, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.018444512757454658, | |
| "grad_norm": 2.9953091144561768, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.2461, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.019674146941284967, | |
| "grad_norm": 2.84977126121521, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.5982, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02090378112511528, | |
| "grad_norm": 3.0945441722869873, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 2.0379, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.022133415308945588, | |
| "grad_norm": 2.7164065837860107, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.4218, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0233630494927759, | |
| "grad_norm": 2.107536792755127, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.4537, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.02459268367660621, | |
| "grad_norm": 2.4697482585906982, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.8444, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02582231786043652, | |
| "grad_norm": 2.4240517616271973, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.9421, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.02705195204426683, | |
| "grad_norm": 2.2246899604797363, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.7046, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.028281586228097143, | |
| "grad_norm": 2.5429322719573975, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.67, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.02951122041192745, | |
| "grad_norm": 2.7511651515960693, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.6833, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03074085459575776, | |
| "grad_norm": 2.542098045349121, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.89, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03197048877958807, | |
| "grad_norm": 2.564352512359619, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.8482, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.033200122963418385, | |
| "grad_norm": 2.605532646179199, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 2.2053, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03442975714724869, | |
| "grad_norm": 2.825662136077881, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 2.271, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.035659391331079, | |
| "grad_norm": 2.425791025161743, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.9432, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.036889025514909315, | |
| "grad_norm": 2.3046185970306396, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.5755, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03811865969873963, | |
| "grad_norm": 2.9772698879241943, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.8409, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.03934829388256993, | |
| "grad_norm": 2.5684814453125, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.081, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.040577928066400246, | |
| "grad_norm": 2.943862199783325, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 2.0085, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04180756225023056, | |
| "grad_norm": 2.787461042404175, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.5761, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.04303719643406086, | |
| "grad_norm": 2.0034701824188232, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.6916, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.044266830617891176, | |
| "grad_norm": 2.688499927520752, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.042, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.04549646480172149, | |
| "grad_norm": 2.4400579929351807, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.1838, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0467260989855518, | |
| "grad_norm": 2.571866512298584, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.386, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.047955733169382106, | |
| "grad_norm": 1.9684172868728638, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.9107, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.04918536735321242, | |
| "grad_norm": 2.2106990814208984, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 2.0839, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05041500153704273, | |
| "grad_norm": 2.17863130569458, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.2765, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.05164463572087304, | |
| "grad_norm": 2.520159959793091, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.1557, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.05287426990470335, | |
| "grad_norm": 2.488316535949707, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 2.1179, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05410390408853366, | |
| "grad_norm": 2.2110090255737305, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.2944, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.05533353827236397, | |
| "grad_norm": 2.4947621822357178, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.2529, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.056563172456194286, | |
| "grad_norm": 2.20231294631958, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.3791, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.05779280664002459, | |
| "grad_norm": 2.164745330810547, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.0467, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0590224408238549, | |
| "grad_norm": 2.1809921264648438, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.2753, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.060252075007685216, | |
| "grad_norm": 2.3916759490966797, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.475, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06148170919151552, | |
| "grad_norm": 2.225285530090332, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.3976, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06148170919151552, | |
| "eval_loss": 1.7791723012924194, | |
| "eval_runtime": 65.3428, | |
| "eval_samples_per_second": 20.966, | |
| "eval_steps_per_second": 5.249, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06271134337534584, | |
| "grad_norm": 2.3801605701446533, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.2255, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06394097755917615, | |
| "grad_norm": 2.1059653759002686, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.2806, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06517061174300645, | |
| "grad_norm": 1.9213285446166992, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.1695, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.06640024592683677, | |
| "grad_norm": 2.1097285747528076, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.3263, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.06762988011066708, | |
| "grad_norm": 2.318544387817383, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.3074, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06885951429449738, | |
| "grad_norm": 2.1936841011047363, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.746, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0700891484783277, | |
| "grad_norm": 2.146787643432617, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.4046, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.071318782662158, | |
| "grad_norm": 2.100097417831421, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.5589, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.07254841684598831, | |
| "grad_norm": 2.006561040878296, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.362, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.07377805102981863, | |
| "grad_norm": 2.735785722732544, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.4866, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07500768521364894, | |
| "grad_norm": 1.8149186372756958, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.1137, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.07623731939747926, | |
| "grad_norm": 1.6552432775497437, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.8553, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.07746695358130956, | |
| "grad_norm": 2.3033909797668457, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.4575, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.07869658776513987, | |
| "grad_norm": 1.8076468706130981, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.2089, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.07992622194897019, | |
| "grad_norm": 1.728806495666504, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.9543, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.08115585613280049, | |
| "grad_norm": 1.9694175720214844, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.224, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0823854903166308, | |
| "grad_norm": 2.006071090698242, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.2497, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.08361512450046112, | |
| "grad_norm": 2.7202627658843994, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.6642, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.08484475868429142, | |
| "grad_norm": 1.8169143199920654, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.9324, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.08607439286812173, | |
| "grad_norm": 1.8312097787857056, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.2291, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08730402705195205, | |
| "grad_norm": 2.0599782466888428, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.369, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.08853366123578235, | |
| "grad_norm": 2.4611144065856934, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.5361, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.08976329541961267, | |
| "grad_norm": 2.452723264694214, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.6963, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.09099292960344298, | |
| "grad_norm": 2.2261428833007812, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.2806, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.09222256378727328, | |
| "grad_norm": 2.154428482055664, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.4432, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0934521979711036, | |
| "grad_norm": 1.7806072235107422, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.2217, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0946818321549339, | |
| "grad_norm": 2.2291276454925537, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.7683, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.09591146633876421, | |
| "grad_norm": 2.322619676589966, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.6815, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.09714110052259453, | |
| "grad_norm": 2.1713833808898926, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.6945, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.09837073470642484, | |
| "grad_norm": 2.513861894607544, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.5102, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09960036889025516, | |
| "grad_norm": 2.748319387435913, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.5339, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.10083000307408546, | |
| "grad_norm": 2.0103747844696045, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.6207, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.10205963725791577, | |
| "grad_norm": 3.0819835662841797, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.9223, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.10328927144174609, | |
| "grad_norm": 2.741457462310791, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.2295, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.10451890562557639, | |
| "grad_norm": 2.5670251846313477, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.9684, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1057485398094067, | |
| "grad_norm": 2.1974761486053467, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.6387, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.10697817399323702, | |
| "grad_norm": 2.458648443222046, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.916, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.10820780817706732, | |
| "grad_norm": 2.3977181911468506, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 2.1755, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.10943744236089763, | |
| "grad_norm": 2.310302972793579, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.7664, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.11066707654472795, | |
| "grad_norm": 2.4010424613952637, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.8779, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11189671072855825, | |
| "grad_norm": 2.518009901046753, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.8479, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.11312634491238857, | |
| "grad_norm": 2.407428741455078, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.9454, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.11435597909621888, | |
| "grad_norm": 2.243361473083496, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.825, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.11558561328004918, | |
| "grad_norm": 2.6791443824768066, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 2.0858, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.1168152474638795, | |
| "grad_norm": 2.587371349334717, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.9513, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.1180448816477098, | |
| "grad_norm": 2.9605817794799805, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.9118, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.11927451583154011, | |
| "grad_norm": 2.8542346954345703, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.0666, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.12050415001537043, | |
| "grad_norm": 3.060283660888672, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.2217, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.12173378419920074, | |
| "grad_norm": 2.5333950519561768, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.1347, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.12296341838303104, | |
| "grad_norm": 2.796520948410034, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 2.1717, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12296341838303104, | |
| "eval_loss": 1.5793250799179077, | |
| "eval_runtime": 65.9577, | |
| "eval_samples_per_second": 20.771, | |
| "eval_steps_per_second": 5.2, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12419305256686136, | |
| "grad_norm": 2.280799150466919, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.4364, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.12542268675069168, | |
| "grad_norm": 1.8975802659988403, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.1184, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.12665232093452197, | |
| "grad_norm": 1.8655685186386108, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.1966, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.1278819551183523, | |
| "grad_norm": 2.6561853885650635, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.5542, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.1291115893021826, | |
| "grad_norm": 2.0758004188537598, | |
| "learning_rate": 5e-05, | |
| "loss": 1.337, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.1303412234860129, | |
| "grad_norm": 2.2125790119171143, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.1985, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.13157085766984322, | |
| "grad_norm": 2.1296145915985107, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.1494, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.13280049185367354, | |
| "grad_norm": 2.477501153945923, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.5914, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.13403012603750383, | |
| "grad_norm": 2.2730050086975098, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.5044, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.13525976022133415, | |
| "grad_norm": 2.106116533279419, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.4397, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13648939440516447, | |
| "grad_norm": 2.412428379058838, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.4501, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.13771902858899476, | |
| "grad_norm": 2.5878262519836426, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.8543, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.13894866277282508, | |
| "grad_norm": 1.7890974283218384, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.0142, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.1401782969566554, | |
| "grad_norm": 1.924872636795044, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.0685, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.1414079311404857, | |
| "grad_norm": 1.5426844358444214, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.0047, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.142637565324316, | |
| "grad_norm": 1.5933966636657715, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.7915, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.14386719950814633, | |
| "grad_norm": 1.889891505241394, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.8489, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.14509683369197662, | |
| "grad_norm": 1.8402427434921265, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.1365, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.14632646787580694, | |
| "grad_norm": 1.863637924194336, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.0945, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.14755610205963726, | |
| "grad_norm": 2.429023027420044, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.4027, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14878573624346758, | |
| "grad_norm": 2.0332224369049072, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.3775, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.15001537042729787, | |
| "grad_norm": 2.3549001216888428, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.7616, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.1512450046111282, | |
| "grad_norm": 1.8422634601593018, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.9636, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.1524746387949585, | |
| "grad_norm": 2.5664734840393066, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.6894, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.1537042729787888, | |
| "grad_norm": 2.370784044265747, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.4621, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.15493390716261912, | |
| "grad_norm": 2.6221935749053955, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.8478, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.15616354134644944, | |
| "grad_norm": 2.353632926940918, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.447, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.15739317553027973, | |
| "grad_norm": 2.7475156784057617, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.839, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.15862280971411005, | |
| "grad_norm": 3.5654728412628174, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.6101, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.15985244389794037, | |
| "grad_norm": 2.425671339035034, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.6681, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.16108207808177066, | |
| "grad_norm": 2.472106695175171, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.5662, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.16231171226560098, | |
| "grad_norm": 2.6592061519622803, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.6525, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.1635413464494313, | |
| "grad_norm": 3.2596681118011475, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.7005, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.1647709806332616, | |
| "grad_norm": 5.354501724243164, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.7205, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.1660006148170919, | |
| "grad_norm": 4.5541181564331055, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.4576, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.16723024900092223, | |
| "grad_norm": 2.6708881855010986, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.5948, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.16845988318475252, | |
| "grad_norm": 5.076529026031494, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.8636, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.16968951736858284, | |
| "grad_norm": 2.9999942779541016, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 2.0031, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.17091915155241316, | |
| "grad_norm": 2.340393543243408, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.7582, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.17214878573624345, | |
| "grad_norm": 2.9876410961151123, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.1751, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.17337841992007377, | |
| "grad_norm": 2.253075361251831, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.5987, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.1746080541039041, | |
| "grad_norm": 3.8176918029785156, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 2.041, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.1758376882877344, | |
| "grad_norm": 2.6594364643096924, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.8178, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.1770673224715647, | |
| "grad_norm": 3.112339735031128, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.8923, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.17829695665539502, | |
| "grad_norm": 4.239136219024658, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.0966, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.17952659083922534, | |
| "grad_norm": 2.923447370529175, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 2.1672, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.18075622502305563, | |
| "grad_norm": 3.101484775543213, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.9377, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.18198585920688595, | |
| "grad_norm": 2.7556941509246826, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.1858, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.18321549339071627, | |
| "grad_norm": 2.672124147415161, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 2.0124, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.18444512757454656, | |
| "grad_norm": 3.0054192543029785, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 2.4374, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.18444512757454656, | |
| "eval_loss": 1.507646083831787, | |
| "eval_runtime": 65.8508, | |
| "eval_samples_per_second": 20.805, | |
| "eval_steps_per_second": 5.209, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.18567476175837688, | |
| "grad_norm": 3.028813123703003, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.4659, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.1869043959422072, | |
| "grad_norm": 2.2049338817596436, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.8703, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.1881340301260375, | |
| "grad_norm": 2.5468006134033203, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.4154, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.1893636643098678, | |
| "grad_norm": 2.4568912982940674, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.3464, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.19059329849369813, | |
| "grad_norm": 2.0330543518066406, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.9793, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.19182293267752842, | |
| "grad_norm": 1.7056840658187866, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.9884, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.19305256686135874, | |
| "grad_norm": 4.244436264038086, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.3315, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.19428220104518906, | |
| "grad_norm": 2.1574676036834717, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.8982, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.19551183522901935, | |
| "grad_norm": 2.028782606124878, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.1966, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.19674146941284967, | |
| "grad_norm": 2.7471141815185547, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.4676, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.19797110359668, | |
| "grad_norm": 1.8987407684326172, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.8082, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.1992007377805103, | |
| "grad_norm": 1.409209966659546, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.7661, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.2004303719643406, | |
| "grad_norm": 1.6311520338058472, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.8921, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.20166000614817092, | |
| "grad_norm": 2.107537269592285, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.3519, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.20288964033200124, | |
| "grad_norm": 1.8441014289855957, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.9669, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.20411927451583153, | |
| "grad_norm": 1.6986143589019775, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.0068, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.20534890869966185, | |
| "grad_norm": 1.4732319116592407, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.737, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.20657854288349217, | |
| "grad_norm": 1.742377519607544, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.1452, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.20780817706732246, | |
| "grad_norm": 2.0086276531219482, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.2459, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.20903781125115278, | |
| "grad_norm": 2.198605537414551, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.1494, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.2102674454349831, | |
| "grad_norm": 2.255854845046997, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.3858, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.2114970796188134, | |
| "grad_norm": 2.4129116535186768, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.4608, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.2127267138026437, | |
| "grad_norm": 2.188180923461914, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.1779, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.21395634798647403, | |
| "grad_norm": 2.263887405395508, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.2293, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.21518598217030432, | |
| "grad_norm": 2.2432286739349365, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.241, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.21641561635413464, | |
| "grad_norm": 2.0392415523529053, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.9923, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.21764525053796496, | |
| "grad_norm": 2.3620004653930664, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.6162, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.21887488472179525, | |
| "grad_norm": 2.2898502349853516, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.5035, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.22010451890562557, | |
| "grad_norm": 2.8068838119506836, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.8557, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.2213341530894559, | |
| "grad_norm": 2.3896260261535645, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.466, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.22256378727328618, | |
| "grad_norm": 2.361085891723633, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.7573, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.2237934214571165, | |
| "grad_norm": 2.294052839279175, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.4091, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.22502305564094682, | |
| "grad_norm": 2.707016944885254, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.6148, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.22625268982477714, | |
| "grad_norm": 2.252143621444702, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.7123, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.22748232400860743, | |
| "grad_norm": 2.573137044906616, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.4542, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.22871195819243775, | |
| "grad_norm": 2.4884629249572754, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.3833, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.22994159237626807, | |
| "grad_norm": 2.8334407806396484, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.5309, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.23117122656009836, | |
| "grad_norm": 2.924685478210449, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.9376, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.23240086074392868, | |
| "grad_norm": 2.4846534729003906, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.936, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.233630494927759, | |
| "grad_norm": 2.6197221279144287, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.5968, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.2348601291115893, | |
| "grad_norm": 2.8064470291137695, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.6652, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.2360897632954196, | |
| "grad_norm": 12.523804664611816, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.5356, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.23731939747924993, | |
| "grad_norm": 2.5480988025665283, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 2.1501, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.23854903166308022, | |
| "grad_norm": 2.3857674598693848, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.4401, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.23977866584691054, | |
| "grad_norm": 2.708869457244873, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.2585, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.24100830003074086, | |
| "grad_norm": 2.6651687622070312, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.9394, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.24223793421457115, | |
| "grad_norm": 2.9050464630126953, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.1558, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.24346756839840147, | |
| "grad_norm": 6.23525333404541, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.1961, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.2446972025822318, | |
| "grad_norm": 2.542240619659424, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 2.0682, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.24592683676606208, | |
| "grad_norm": 4.225558280944824, | |
| "learning_rate": 0.0, | |
| "loss": 1.9642, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.24592683676606208, | |
| "eval_loss": 1.4959101676940918, | |
| "eval_runtime": 65.8752, | |
| "eval_samples_per_second": 20.797, | |
| "eval_steps_per_second": 5.207, | |
| "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": 1.1438577046624666e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |