| { | |
| "best_metric": 3.9518187046051025, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.48661800486618007, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0024330900243309003, | |
| "grad_norm": 0.4150792956352234, | |
| "learning_rate": 1e-05, | |
| "loss": 4.6213, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0024330900243309003, | |
| "eval_loss": 6.324268817901611, | |
| "eval_runtime": 9.1642, | |
| "eval_samples_per_second": 75.621, | |
| "eval_steps_per_second": 18.987, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004866180048661801, | |
| "grad_norm": 0.467136412858963, | |
| "learning_rate": 2e-05, | |
| "loss": 4.7926, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0072992700729927005, | |
| "grad_norm": 0.42653557658195496, | |
| "learning_rate": 3e-05, | |
| "loss": 4.7343, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.009732360097323601, | |
| "grad_norm": 0.4761233925819397, | |
| "learning_rate": 4e-05, | |
| "loss": 4.9593, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.012165450121654502, | |
| "grad_norm": 0.46191975474357605, | |
| "learning_rate": 5e-05, | |
| "loss": 4.8657, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.014598540145985401, | |
| "grad_norm": 0.4583083391189575, | |
| "learning_rate": 6e-05, | |
| "loss": 4.9244, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0170316301703163, | |
| "grad_norm": 0.47342315316200256, | |
| "learning_rate": 7e-05, | |
| "loss": 4.9183, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.019464720194647202, | |
| "grad_norm": 0.443309485912323, | |
| "learning_rate": 8e-05, | |
| "loss": 4.9222, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.021897810218978103, | |
| "grad_norm": 0.43505600094795227, | |
| "learning_rate": 9e-05, | |
| "loss": 5.1087, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.024330900243309004, | |
| "grad_norm": 0.426425039768219, | |
| "learning_rate": 0.0001, | |
| "loss": 5.1403, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0267639902676399, | |
| "grad_norm": 0.501082181930542, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 5.2618, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.029197080291970802, | |
| "grad_norm": 0.5287079215049744, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 5.2651, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.031630170316301706, | |
| "grad_norm": 0.5229122638702393, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 5.1472, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0340632603406326, | |
| "grad_norm": 0.5195987820625305, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 5.2136, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0364963503649635, | |
| "grad_norm": 0.440207302570343, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 5.169, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.038929440389294405, | |
| "grad_norm": 0.4922386407852173, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 5.0637, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0413625304136253, | |
| "grad_norm": 0.4565863311290741, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 5.2095, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.043795620437956206, | |
| "grad_norm": 0.4655221104621887, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 5.1914, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.046228710462287104, | |
| "grad_norm": 0.509451150894165, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 5.2361, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.04866180048661801, | |
| "grad_norm": 0.49580907821655273, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 5.1846, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.051094890510948905, | |
| "grad_norm": 0.5232206583023071, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 5.2727, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0535279805352798, | |
| "grad_norm": 0.5294134616851807, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 5.4143, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.05596107055961071, | |
| "grad_norm": 0.5824503898620605, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 5.3089, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.058394160583941604, | |
| "grad_norm": 0.5405100584030151, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 5.1332, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.06082725060827251, | |
| "grad_norm": 0.5429709553718567, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 5.3275, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.06326034063260341, | |
| "grad_norm": 0.5520122647285461, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 5.2802, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.06569343065693431, | |
| "grad_norm": 0.598534107208252, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 5.2446, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0681265206812652, | |
| "grad_norm": 0.5517746806144714, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 5.1139, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0705596107055961, | |
| "grad_norm": 0.5965620875358582, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 5.2888, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.072992700729927, | |
| "grad_norm": 0.6049797534942627, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 5.2226, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07542579075425791, | |
| "grad_norm": 0.5679780840873718, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 5.2251, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07785888077858881, | |
| "grad_norm": 0.6170575022697449, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 5.1787, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.08029197080291971, | |
| "grad_norm": 0.6049054265022278, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 5.0508, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0827250608272506, | |
| "grad_norm": 0.6631417274475098, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 5.0091, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0851581508515815, | |
| "grad_norm": 0.7571232318878174, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 5.0599, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08759124087591241, | |
| "grad_norm": 0.8026711940765381, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 5.0529, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.09002433090024331, | |
| "grad_norm": 0.696685791015625, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 4.9539, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.09245742092457421, | |
| "grad_norm": 0.6417451500892639, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 4.9461, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0948905109489051, | |
| "grad_norm": 0.6456137895584106, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 4.9195, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.09732360097323602, | |
| "grad_norm": 0.795696496963501, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 5.0444, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09975669099756691, | |
| "grad_norm": 0.6867913603782654, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 4.8225, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.10218978102189781, | |
| "grad_norm": 0.6948617696762085, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 5.0534, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.10462287104622871, | |
| "grad_norm": 0.7717512845993042, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 4.854, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.1070559610705596, | |
| "grad_norm": 0.8118676543235779, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 5.1043, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.10948905109489052, | |
| "grad_norm": 0.8161230087280273, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 4.8464, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.11192214111922141, | |
| "grad_norm": 0.8852051496505737, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 5.1701, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.11435523114355231, | |
| "grad_norm": 0.9199775457382202, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 4.7208, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.11678832116788321, | |
| "grad_norm": 0.8746785521507263, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 4.6893, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1192214111922141, | |
| "grad_norm": 1.082961082458496, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 5.4306, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.12165450121654502, | |
| "grad_norm": 3.5441510677337646, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 5.636, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12165450121654502, | |
| "eval_loss": 4.712304592132568, | |
| "eval_runtime": 9.175, | |
| "eval_samples_per_second": 75.532, | |
| "eval_steps_per_second": 18.965, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12408759124087591, | |
| "grad_norm": 1.7757056951522827, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 4.5692, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.12652068126520682, | |
| "grad_norm": 1.6864593029022217, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 4.5767, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.12895377128953772, | |
| "grad_norm": 1.4113835096359253, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 4.55, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.13138686131386862, | |
| "grad_norm": 1.090064525604248, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 4.5021, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.13381995133819952, | |
| "grad_norm": 0.9537211656570435, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 4.4824, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1362530413625304, | |
| "grad_norm": 0.8344225883483887, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 4.3658, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1386861313868613, | |
| "grad_norm": 0.6728500723838806, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 4.2867, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1411192214111922, | |
| "grad_norm": 0.5792769193649292, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 4.3504, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.1435523114355231, | |
| "grad_norm": 0.4859144687652588, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 4.1665, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.145985401459854, | |
| "grad_norm": 0.4857715666294098, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 4.1907, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14841849148418493, | |
| "grad_norm": 0.605350136756897, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 4.2865, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.15085158150851583, | |
| "grad_norm": 0.63717120885849, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 4.3976, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.15328467153284672, | |
| "grad_norm": 0.6708590984344482, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 4.3357, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.15571776155717762, | |
| "grad_norm": 0.5998736023902893, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 4.1715, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.15815085158150852, | |
| "grad_norm": 0.4863462746143341, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 4.2261, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.16058394160583941, | |
| "grad_norm": 0.4783068299293518, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 4.2536, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.1630170316301703, | |
| "grad_norm": 0.5137869715690613, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 4.2038, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.1654501216545012, | |
| "grad_norm": 0.518310546875, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 4.2425, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1678832116788321, | |
| "grad_norm": 0.48414185643196106, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 4.2913, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.170316301703163, | |
| "grad_norm": 0.5184882879257202, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 4.3359, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17274939172749393, | |
| "grad_norm": 0.5165735483169556, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 4.3767, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.17518248175182483, | |
| "grad_norm": 0.508715808391571, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 4.1377, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.17761557177615572, | |
| "grad_norm": 0.5977687239646912, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 4.3855, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.18004866180048662, | |
| "grad_norm": 0.557275116443634, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 4.271, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.18248175182481752, | |
| "grad_norm": 0.6329896450042725, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 4.4042, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.18491484184914841, | |
| "grad_norm": 0.5772005319595337, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 4.4691, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1873479318734793, | |
| "grad_norm": 0.6361019015312195, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 4.3219, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.1897810218978102, | |
| "grad_norm": 0.7251425385475159, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 4.2401, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.1922141119221411, | |
| "grad_norm": 0.6354728937149048, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 4.445, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.19464720194647203, | |
| "grad_norm": 0.6783463358879089, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 4.245, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19708029197080293, | |
| "grad_norm": 0.7329065203666687, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 4.1838, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.19951338199513383, | |
| "grad_norm": 0.6637375950813293, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 4.1502, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.20194647201946472, | |
| "grad_norm": 0.7039035558700562, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 4.2224, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.20437956204379562, | |
| "grad_norm": 0.7096616625785828, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 4.247, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.20681265206812652, | |
| "grad_norm": 0.7598410248756409, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 4.2627, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.20924574209245742, | |
| "grad_norm": 0.7319227457046509, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 4.1582, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.2116788321167883, | |
| "grad_norm": 0.8318302035331726, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 4.2787, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.2141119221411192, | |
| "grad_norm": 0.7554244995117188, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 4.1737, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.21654501216545013, | |
| "grad_norm": 0.8304018974304199, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 4.247, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.21897810218978103, | |
| "grad_norm": 0.7371103167533875, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 4.0402, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.22141119221411193, | |
| "grad_norm": 0.8232166767120361, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 4.3153, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.22384428223844283, | |
| "grad_norm": 0.796526312828064, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 4.2081, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.22627737226277372, | |
| "grad_norm": 0.9313976168632507, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 4.3794, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.22871046228710462, | |
| "grad_norm": 0.8513618111610413, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 4.4181, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.23114355231143552, | |
| "grad_norm": 0.8770570755004883, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 4.1982, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.23357664233576642, | |
| "grad_norm": 0.9517497420310974, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 4.3447, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2360097323600973, | |
| "grad_norm": 1.0358669757843018, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 4.0967, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.2384428223844282, | |
| "grad_norm": 1.1383416652679443, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 4.3606, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.24087591240875914, | |
| "grad_norm": 1.243801236152649, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 4.5482, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.24330900243309003, | |
| "grad_norm": 1.8168234825134277, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 5.0682, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24330900243309003, | |
| "eval_loss": 4.185164451599121, | |
| "eval_runtime": 9.1234, | |
| "eval_samples_per_second": 75.959, | |
| "eval_steps_per_second": 19.072, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24574209245742093, | |
| "grad_norm": 0.9722921252250671, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 3.8778, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.24817518248175183, | |
| "grad_norm": 1.0037745237350464, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 4.0331, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.25060827250608275, | |
| "grad_norm": 0.9996768236160278, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 3.9632, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.25304136253041365, | |
| "grad_norm": 0.982207715511322, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 4.0827, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.25547445255474455, | |
| "grad_norm": 0.8831742405891418, | |
| "learning_rate": 5e-05, | |
| "loss": 4.0007, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.25790754257907544, | |
| "grad_norm": 0.8144031763076782, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 4.0727, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.26034063260340634, | |
| "grad_norm": 0.7167903780937195, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 3.9359, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.26277372262773724, | |
| "grad_norm": 0.6442274451255798, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 3.9402, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.26520681265206814, | |
| "grad_norm": 0.6302825808525085, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 4.052, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.26763990267639903, | |
| "grad_norm": 0.576882541179657, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 3.9945, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.27007299270072993, | |
| "grad_norm": 0.6141225099563599, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 3.944, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.2725060827250608, | |
| "grad_norm": 0.5857487320899963, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 3.8652, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.2749391727493917, | |
| "grad_norm": 0.6030862927436829, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 3.9192, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.2773722627737226, | |
| "grad_norm": 0.5923975110054016, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 3.688, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.2798053527980535, | |
| "grad_norm": 0.4928789436817169, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 3.9625, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2822384428223844, | |
| "grad_norm": 0.4602852165699005, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 3.8324, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.2846715328467153, | |
| "grad_norm": 0.5370597839355469, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 4.0492, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2871046228710462, | |
| "grad_norm": 0.5495446920394897, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 3.9739, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2895377128953771, | |
| "grad_norm": 0.5667747259140015, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 3.9028, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.291970802919708, | |
| "grad_norm": 0.6401398777961731, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 3.9936, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2944038929440389, | |
| "grad_norm": 0.6706793904304504, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 4.0466, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.29683698296836986, | |
| "grad_norm": 0.6430131196975708, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 4.1019, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.29927007299270075, | |
| "grad_norm": 0.6576083302497864, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 3.9591, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.30170316301703165, | |
| "grad_norm": 0.6542022824287415, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 4.0623, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.30413625304136255, | |
| "grad_norm": 0.6278736591339111, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 3.999, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.30656934306569344, | |
| "grad_norm": 0.6900716423988342, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 4.0718, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.30900243309002434, | |
| "grad_norm": 0.630337119102478, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 3.9535, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.31143552311435524, | |
| "grad_norm": 0.7066243290901184, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 3.9906, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.31386861313868614, | |
| "grad_norm": 0.6288943290710449, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 4.0945, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.31630170316301703, | |
| "grad_norm": 0.6536298394203186, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 3.9335, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.31873479318734793, | |
| "grad_norm": 0.6657919883728027, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 3.847, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.32116788321167883, | |
| "grad_norm": 0.7174845337867737, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 4.0905, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.3236009732360097, | |
| "grad_norm": 0.8272973895072937, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 3.7112, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.3260340632603406, | |
| "grad_norm": 0.8045414686203003, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 3.882, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3284671532846715, | |
| "grad_norm": 0.83411705493927, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 3.8097, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3309002433090024, | |
| "grad_norm": 0.7868964076042175, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 4.2899, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.7803187370300293, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 3.8324, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3357664233576642, | |
| "grad_norm": 0.8379628658294678, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 4.2062, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.3381995133819951, | |
| "grad_norm": 0.8804911375045776, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 3.889, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.340632603406326, | |
| "grad_norm": 0.9127257466316223, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 4.0018, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.34306569343065696, | |
| "grad_norm": 0.9609770774841309, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 3.7671, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.34549878345498786, | |
| "grad_norm": 0.9258947372436523, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 3.9541, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.34793187347931875, | |
| "grad_norm": 0.9964628219604492, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 4.0396, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.35036496350364965, | |
| "grad_norm": 0.9827728867530823, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 4.0154, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.35279805352798055, | |
| "grad_norm": 1.044978141784668, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 4.0177, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.35523114355231145, | |
| "grad_norm": 1.166678547859192, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 4.2204, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.35766423357664234, | |
| "grad_norm": 1.2606854438781738, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 4.3236, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.36009732360097324, | |
| "grad_norm": 1.2951610088348389, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 4.4471, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.36253041362530414, | |
| "grad_norm": 1.4271166324615479, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 4.4436, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.36496350364963503, | |
| "grad_norm": 2.168100595474243, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 4.6958, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.36496350364963503, | |
| "eval_loss": 3.9819765090942383, | |
| "eval_runtime": 9.1306, | |
| "eval_samples_per_second": 75.899, | |
| "eval_steps_per_second": 19.057, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.36739659367396593, | |
| "grad_norm": 0.47898080945014954, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 3.7986, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.36982968369829683, | |
| "grad_norm": 0.4894225597381592, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 3.7199, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.3722627737226277, | |
| "grad_norm": 0.5630130767822266, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 3.9137, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.3746958637469586, | |
| "grad_norm": 0.6036057472229004, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 3.7623, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.3771289537712895, | |
| "grad_norm": 0.6120648980140686, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 3.8333, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.3795620437956204, | |
| "grad_norm": 0.6142074465751648, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 3.8089, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.3819951338199513, | |
| "grad_norm": 0.622135579586029, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 3.72, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3844282238442822, | |
| "grad_norm": 0.6999398469924927, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 3.7453, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.38686131386861317, | |
| "grad_norm": 0.6760008931159973, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 3.7665, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.38929440389294406, | |
| "grad_norm": 0.7160153985023499, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 3.7742, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.39172749391727496, | |
| "grad_norm": 0.8535175323486328, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 3.8287, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.39416058394160586, | |
| "grad_norm": 0.828071653842926, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 3.6624, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.39659367396593675, | |
| "grad_norm": 0.7289974093437195, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 3.906, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.39902676399026765, | |
| "grad_norm": 0.6275280117988586, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 3.7206, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.40145985401459855, | |
| "grad_norm": 0.6391129493713379, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 3.8872, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.40389294403892945, | |
| "grad_norm": 0.6067607402801514, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 3.849, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.40632603406326034, | |
| "grad_norm": 0.5415106415748596, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 3.747, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.40875912408759124, | |
| "grad_norm": 0.6563166379928589, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 3.7465, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.41119221411192214, | |
| "grad_norm": 0.6457042694091797, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 3.8773, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.41362530413625304, | |
| "grad_norm": 0.6696534156799316, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 3.9295, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.41605839416058393, | |
| "grad_norm": 0.6423199772834778, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 4.0534, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.41849148418491483, | |
| "grad_norm": 0.6438478231430054, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 3.8166, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.4209245742092457, | |
| "grad_norm": 0.7035348415374756, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 3.7722, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.4233576642335766, | |
| "grad_norm": 0.7330677509307861, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 3.8362, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.4257907542579075, | |
| "grad_norm": 0.6888772249221802, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 3.891, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4282238442822384, | |
| "grad_norm": 0.6441717147827148, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 3.9347, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.4306569343065693, | |
| "grad_norm": 0.6894862651824951, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 3.8289, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.43309002433090027, | |
| "grad_norm": 0.7187209129333496, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 3.8644, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.43552311435523117, | |
| "grad_norm": 0.7581349015235901, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 3.8101, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.43795620437956206, | |
| "grad_norm": 0.7134523391723633, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 3.939, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.44038929440389296, | |
| "grad_norm": 0.6464361548423767, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 3.9297, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.44282238442822386, | |
| "grad_norm": 0.7212158441543579, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 3.9269, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.44525547445255476, | |
| "grad_norm": 0.6693487763404846, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 3.9262, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.44768856447688565, | |
| "grad_norm": 0.7347055673599243, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 3.7302, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.45012165450121655, | |
| "grad_norm": 0.7876744866371155, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 3.8746, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.45255474452554745, | |
| "grad_norm": 0.746986448764801, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 4.0241, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.45498783454987834, | |
| "grad_norm": 0.8166369795799255, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 3.8364, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.45742092457420924, | |
| "grad_norm": 0.7827574014663696, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 3.8354, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.45985401459854014, | |
| "grad_norm": 0.813870906829834, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 4.0202, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.46228710462287104, | |
| "grad_norm": 0.8696313500404358, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 3.885, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.46472019464720193, | |
| "grad_norm": 0.8160954117774963, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 3.9773, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.46715328467153283, | |
| "grad_norm": 0.891071617603302, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 4.0924, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.46958637469586373, | |
| "grad_norm": 1.0203135013580322, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 4.043, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.4720194647201946, | |
| "grad_norm": 0.944577157497406, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 3.8596, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.4744525547445255, | |
| "grad_norm": 1.005293607711792, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 4.0578, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.4768856447688564, | |
| "grad_norm": 1.2007620334625244, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 4.4006, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.4793187347931874, | |
| "grad_norm": 1.1512655019760132, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 4.1126, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.48175182481751827, | |
| "grad_norm": 1.3942410945892334, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 4.4196, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.48418491484184917, | |
| "grad_norm": 1.4426358938217163, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 4.5422, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.48661800486618007, | |
| "grad_norm": 2.294565200805664, | |
| "learning_rate": 0.0, | |
| "loss": 4.7856, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.48661800486618007, | |
| "eval_loss": 3.9518187046051025, | |
| "eval_runtime": 9.1537, | |
| "eval_samples_per_second": 75.707, | |
| "eval_steps_per_second": 19.009, | |
| "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": 4969073850974208.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |