| { | |
| "best_metric": 1.2174347639083862, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.02666133439978671, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00013330667199893355, | |
| "grad_norm": 1.0745736360549927, | |
| "learning_rate": 1e-05, | |
| "loss": 1.4603, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00013330667199893355, | |
| "eval_loss": 1.7412927150726318, | |
| "eval_runtime": 893.9334, | |
| "eval_samples_per_second": 14.133, | |
| "eval_steps_per_second": 3.534, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0002666133439978671, | |
| "grad_norm": 1.1597387790679932, | |
| "learning_rate": 2e-05, | |
| "loss": 1.3804, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0003999200159968006, | |
| "grad_norm": 1.2239619493484497, | |
| "learning_rate": 3e-05, | |
| "loss": 1.4915, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0005332266879957342, | |
| "grad_norm": 1.4500244855880737, | |
| "learning_rate": 4e-05, | |
| "loss": 1.6533, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0006665333599946677, | |
| "grad_norm": 1.293084740638733, | |
| "learning_rate": 5e-05, | |
| "loss": 1.4773, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0007998400319936012, | |
| "grad_norm": 1.2047165632247925, | |
| "learning_rate": 6e-05, | |
| "loss": 1.3702, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0009331467039925348, | |
| "grad_norm": 0.8491549491882324, | |
| "learning_rate": 7e-05, | |
| "loss": 1.3265, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0010664533759914684, | |
| "grad_norm": 0.9282706379890442, | |
| "learning_rate": 8e-05, | |
| "loss": 1.2982, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0011997600479904018, | |
| "grad_norm": 0.9100210666656494, | |
| "learning_rate": 9e-05, | |
| "loss": 1.3398, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0013330667199893355, | |
| "grad_norm": 0.9498695731163025, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1851, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.001466373391988269, | |
| "grad_norm": 0.9965104460716248, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.2516, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0015996800639872025, | |
| "grad_norm": 0.8492946028709412, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.2752, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0017329867359861361, | |
| "grad_norm": 0.968284547328949, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.2821, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0018662934079850695, | |
| "grad_norm": 0.7806897163391113, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.1962, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.001999600079984003, | |
| "grad_norm": 0.8861862421035767, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.304, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.002132906751982937, | |
| "grad_norm": 0.8310438394546509, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.2358, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0022662134239818704, | |
| "grad_norm": 0.8757510781288147, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.4375, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0023995200959808036, | |
| "grad_norm": 1.1226414442062378, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.3794, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0025328267679797373, | |
| "grad_norm": 0.8201644420623779, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.2039, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.002666133439978671, | |
| "grad_norm": 0.854809045791626, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.2542, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0027994401119776045, | |
| "grad_norm": 0.9954699873924255, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.3179, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.002932746783976538, | |
| "grad_norm": 0.823428213596344, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.1712, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0030660534559754714, | |
| "grad_norm": 0.8914350271224976, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.1579, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.003199360127974405, | |
| "grad_norm": 0.8348805904388428, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.1415, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0033326667999733386, | |
| "grad_norm": 1.009377360343933, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.4218, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0034659734719722723, | |
| "grad_norm": 0.9172560572624207, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.2804, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.003599280143971206, | |
| "grad_norm": 0.826400876045227, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.1378, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.003732586815970139, | |
| "grad_norm": 0.9295492172241211, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.3835, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0038658934879690727, | |
| "grad_norm": 0.8225899934768677, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.1273, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.003999200159968006, | |
| "grad_norm": 0.8576282858848572, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.1702, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.00413250683196694, | |
| "grad_norm": 0.926065981388092, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.2486, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.004265813503965874, | |
| "grad_norm": 0.8948851823806763, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.35, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.004399120175964807, | |
| "grad_norm": 0.9296471476554871, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.2109, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.004532426847963741, | |
| "grad_norm": 0.9554625153541565, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.2104, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0046657335199626745, | |
| "grad_norm": 0.9973741769790649, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.2216, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.004799040191961607, | |
| "grad_norm": 0.9852436780929565, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.2423, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.004932346863960541, | |
| "grad_norm": 1.0341085195541382, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.1927, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0050656535359594745, | |
| "grad_norm": 1.0837221145629883, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.2843, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.005198960207958408, | |
| "grad_norm": 1.0713090896606445, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.1197, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.005332266879957342, | |
| "grad_norm": 1.299558401107788, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.4558, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0054655735519562754, | |
| "grad_norm": 1.3224207162857056, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.5213, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.005598880223955209, | |
| "grad_norm": 1.0918453931808472, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.2473, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.005732186895954143, | |
| "grad_norm": 1.2147316932678223, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.2924, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.005865493567953076, | |
| "grad_norm": 1.1717326641082764, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.1122, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.00599880023995201, | |
| "grad_norm": 1.5002002716064453, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.3528, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.006132106911950943, | |
| "grad_norm": 1.1809282302856445, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.0353, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.006265413583949876, | |
| "grad_norm": 1.3819034099578857, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.3988, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.00639872025594881, | |
| "grad_norm": 2.4021060466766357, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.4357, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.006532026927947744, | |
| "grad_norm": 3.4780421257019043, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.3491, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.006665333599946677, | |
| "grad_norm": 4.60449743270874, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.6521, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.006665333599946677, | |
| "eval_loss": 1.2686829566955566, | |
| "eval_runtime": 897.6977, | |
| "eval_samples_per_second": 14.074, | |
| "eval_steps_per_second": 3.519, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.006798640271945611, | |
| "grad_norm": 0.8523968458175659, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.207, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0069319469439445445, | |
| "grad_norm": 0.8501446843147278, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.4451, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.007065253615943478, | |
| "grad_norm": 0.6356542110443115, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.126, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.007198560287942412, | |
| "grad_norm": 0.7071014642715454, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.3005, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.007331866959941345, | |
| "grad_norm": 0.7470281720161438, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.1553, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.007465173631940278, | |
| "grad_norm": 0.7439374327659607, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.2164, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.007598480303939212, | |
| "grad_norm": 0.789055585861206, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.3809, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0077317869759381454, | |
| "grad_norm": 0.7421045303344727, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.3091, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.007865093647937079, | |
| "grad_norm": 0.7292497754096985, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.1928, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.007998400319936013, | |
| "grad_norm": 0.7535648345947266, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.3245, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.008131706991934946, | |
| "grad_norm": 0.7184985280036926, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.2755, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.00826501366393388, | |
| "grad_norm": 0.705776035785675, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.1553, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.008398320335932814, | |
| "grad_norm": 0.680077850818634, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.0244, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.008531627007931747, | |
| "grad_norm": 0.7714347839355469, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.2602, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.00866493367993068, | |
| "grad_norm": 0.8244078755378723, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.2375, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.008798240351929614, | |
| "grad_norm": 0.8791713714599609, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.2654, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.008931547023928548, | |
| "grad_norm": 0.8415705561637878, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.2845, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.009064853695927482, | |
| "grad_norm": 0.7405332326889038, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.1162, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.009198160367926415, | |
| "grad_norm": 0.7569639682769775, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.1168, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.009331467039925349, | |
| "grad_norm": 0.8448635935783386, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.3316, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.009464773711924283, | |
| "grad_norm": 0.8111772537231445, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.085, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.009598080383923215, | |
| "grad_norm": 0.8686326742172241, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.3275, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.009731387055922148, | |
| "grad_norm": 0.8204224109649658, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.203, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.009864693727921082, | |
| "grad_norm": 0.7859241962432861, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.1295, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.009998000399920015, | |
| "grad_norm": 0.8578535914421082, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.2775, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.010131307071918949, | |
| "grad_norm": 0.8531782031059265, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.1732, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.010264613743917883, | |
| "grad_norm": 0.7920311093330383, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.2004, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.010397920415916816, | |
| "grad_norm": 0.8537325263023376, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.2338, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.01053122708791575, | |
| "grad_norm": 0.8466278910636902, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.1185, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.010664533759914684, | |
| "grad_norm": 0.9058700203895569, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.229, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.010797840431913617, | |
| "grad_norm": 0.8695547580718994, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.2438, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.010931147103912551, | |
| "grad_norm": 0.9314249157905579, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.2332, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.011064453775911485, | |
| "grad_norm": 0.9494042992591858, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.2074, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.011197760447910418, | |
| "grad_norm": 0.9084473252296448, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.1922, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.011331067119909352, | |
| "grad_norm": 0.9706754088401794, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.1319, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.011464373791908285, | |
| "grad_norm": 0.9327569603919983, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.3322, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.011597680463907219, | |
| "grad_norm": 1.0464143753051758, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.287, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.011730987135906153, | |
| "grad_norm": 1.3841379880905151, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.214, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.011864293807905086, | |
| "grad_norm": 1.0763241052627563, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.1554, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.01199760047990402, | |
| "grad_norm": 0.9503827095031738, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.1839, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.012130907151902954, | |
| "grad_norm": 1.066009759902954, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.3296, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.012264213823901885, | |
| "grad_norm": 1.0369770526885986, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.153, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.012397520495900819, | |
| "grad_norm": 1.1142950057983398, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.2537, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.012530827167899753, | |
| "grad_norm": 1.1302510499954224, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.227, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.012664133839898686, | |
| "grad_norm": 1.324735403060913, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.375, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.01279744051189762, | |
| "grad_norm": 1.0952991247177124, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.0373, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.012930747183896554, | |
| "grad_norm": 1.7994344234466553, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.531, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.013064053855895487, | |
| "grad_norm": 1.9991344213485718, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.2441, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.013197360527894421, | |
| "grad_norm": 1.8803925514221191, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.238, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.013330667199893355, | |
| "grad_norm": 2.514798164367676, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.2592, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.013330667199893355, | |
| "eval_loss": 1.235968828201294, | |
| "eval_runtime": 897.8126, | |
| "eval_samples_per_second": 14.072, | |
| "eval_steps_per_second": 3.519, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.013463973871892288, | |
| "grad_norm": 0.6205466985702515, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.2382, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.013597280543891222, | |
| "grad_norm": 0.6360917091369629, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.0933, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.013730587215890155, | |
| "grad_norm": 0.6676458716392517, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.2596, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.013863893887889089, | |
| "grad_norm": 0.6764618158340454, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.2296, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.013997200559888023, | |
| "grad_norm": 0.6883326768875122, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1956, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.014130507231886956, | |
| "grad_norm": 0.6515330672264099, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.1945, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.01426381390388589, | |
| "grad_norm": 0.7443110346794128, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.2503, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.014397120575884824, | |
| "grad_norm": 0.6871118545532227, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.1663, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.014530427247883757, | |
| "grad_norm": 0.7479511499404907, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.4153, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.01466373391988269, | |
| "grad_norm": 0.6945416927337646, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.2392, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.014797040591881624, | |
| "grad_norm": 0.6980699300765991, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.3174, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.014930347263880556, | |
| "grad_norm": 0.7077943086624146, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.201, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.01506365393587949, | |
| "grad_norm": 0.7695159316062927, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.386, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.015196960607878424, | |
| "grad_norm": 0.6997578144073486, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.1997, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.015330267279877357, | |
| "grad_norm": 0.7252641916275024, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.1792, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.015463573951876291, | |
| "grad_norm": 0.7371767163276672, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.2095, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.015596880623875225, | |
| "grad_norm": 0.7858007550239563, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.3076, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.015730187295874158, | |
| "grad_norm": 0.7438164353370667, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.2489, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.015863493967873094, | |
| "grad_norm": 0.7789096236228943, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.1782, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.015996800639872025, | |
| "grad_norm": 0.7536317110061646, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.1551, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01613010731187096, | |
| "grad_norm": 0.8432468771934509, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.3212, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.016263413983869893, | |
| "grad_norm": 0.8032596707344055, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.2541, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.016396720655868825, | |
| "grad_norm": 0.8486832976341248, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.3336, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.01653002732786776, | |
| "grad_norm": 0.7489065527915955, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.057, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.016663333999866692, | |
| "grad_norm": 0.9419240355491638, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.3496, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.016796640671865627, | |
| "grad_norm": 0.8470003008842468, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.1591, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.01692994734386456, | |
| "grad_norm": 0.8417698740959167, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.1579, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.017063254015863494, | |
| "grad_norm": 0.8789008855819702, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.2494, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.017196560687862426, | |
| "grad_norm": 0.8741716742515564, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.2255, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.01732986735986136, | |
| "grad_norm": 0.8465598821640015, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.2017, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.017463174031860294, | |
| "grad_norm": 0.8675022125244141, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.1874, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.01759648070385923, | |
| "grad_norm": 0.9815752506256104, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.4558, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.01772978737585816, | |
| "grad_norm": 0.8390645384788513, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.2183, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.017863094047857096, | |
| "grad_norm": 0.9450867772102356, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.2721, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.017996400719856028, | |
| "grad_norm": 0.8508359789848328, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.1076, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.018129707391854964, | |
| "grad_norm": 0.9367918372154236, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.1246, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.018263014063853895, | |
| "grad_norm": 0.9508082270622253, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.1096, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.01839632073585283, | |
| "grad_norm": 0.9181338548660278, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.1072, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.018529627407851763, | |
| "grad_norm": 0.9841213226318359, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.1864, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.018662934079850698, | |
| "grad_norm": 0.9663724303245544, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.1096, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.01879624075184963, | |
| "grad_norm": 0.9832901358604431, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.2234, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.018929547423848565, | |
| "grad_norm": 1.0503119230270386, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.2802, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.019062854095847497, | |
| "grad_norm": 1.010730504989624, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.1259, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.01919616076784643, | |
| "grad_norm": 0.9640291929244995, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.0263, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.019329467439845364, | |
| "grad_norm": 1.2106753587722778, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.34, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.019462774111844296, | |
| "grad_norm": 1.2731579542160034, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.235, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.019596080783843232, | |
| "grad_norm": 1.3919060230255127, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.2402, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.019729387455842164, | |
| "grad_norm": 1.4603452682495117, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.115, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.0198626941278411, | |
| "grad_norm": 2.7993760108947754, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.3494, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.01999600079984003, | |
| "grad_norm": 2.829930543899536, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.0537, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01999600079984003, | |
| "eval_loss": 1.2199597358703613, | |
| "eval_runtime": 897.4455, | |
| "eval_samples_per_second": 14.078, | |
| "eval_steps_per_second": 3.52, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.020129307471838966, | |
| "grad_norm": 0.5059590339660645, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.0344, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.020262614143837898, | |
| "grad_norm": 0.6080087423324585, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.2128, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.020395920815836834, | |
| "grad_norm": 0.7243260145187378, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.3855, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.020529227487835765, | |
| "grad_norm": 0.6286957859992981, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.2555, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.0206625341598347, | |
| "grad_norm": 0.6662344336509705, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.1873, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.020795840831833633, | |
| "grad_norm": 0.6518131494522095, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.2023, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.020929147503832568, | |
| "grad_norm": 0.6527888178825378, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.1803, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.0210624541758315, | |
| "grad_norm": 0.721267580986023, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.362, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.021195760847830435, | |
| "grad_norm": 0.7137548327445984, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.4078, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.021329067519829367, | |
| "grad_norm": 0.7020995616912842, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.2114, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.021462374191828303, | |
| "grad_norm": 0.7052568793296814, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.2552, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.021595680863827234, | |
| "grad_norm": 0.702303946018219, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.0904, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.021728987535826166, | |
| "grad_norm": 0.7139044404029846, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.2769, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.021862294207825102, | |
| "grad_norm": 0.795620322227478, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.2366, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.021995600879824034, | |
| "grad_norm": 0.7649041414260864, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.135, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.02212890755182297, | |
| "grad_norm": 0.8941980004310608, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.4213, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.0222622142238219, | |
| "grad_norm": 0.7104661464691162, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.1501, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.022395520895820836, | |
| "grad_norm": 0.767971396446228, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.2285, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.022528827567819768, | |
| "grad_norm": 0.8345792889595032, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.3538, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.022662134239818704, | |
| "grad_norm": 0.7592571973800659, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.138, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.022795440911817635, | |
| "grad_norm": 0.7913587689399719, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.274, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.02292874758381657, | |
| "grad_norm": 0.8551506400108337, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.3539, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.023062054255815503, | |
| "grad_norm": 0.826088547706604, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.2791, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.023195360927814438, | |
| "grad_norm": 0.8231903314590454, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.2643, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.02332866759981337, | |
| "grad_norm": 0.8636263012886047, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.1534, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.023461974271812305, | |
| "grad_norm": 0.8509513735771179, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.1875, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.023595280943811237, | |
| "grad_norm": 0.8056614995002747, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.2276, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.023728587615810173, | |
| "grad_norm": 0.906745433807373, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.2333, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.023861894287809104, | |
| "grad_norm": 0.8647125959396362, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.1878, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.02399520095980804, | |
| "grad_norm": 1.030241847038269, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.3258, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.024128507631806972, | |
| "grad_norm": 0.8133326172828674, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.1028, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.024261814303805907, | |
| "grad_norm": 0.9469070434570312, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.2033, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.02439512097580484, | |
| "grad_norm": 1.0034525394439697, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.3276, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.02452842764780377, | |
| "grad_norm": 0.9274887442588806, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.2858, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.024661734319802706, | |
| "grad_norm": 0.9067505598068237, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.128, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.024795040991801638, | |
| "grad_norm": 0.8848114609718323, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.2095, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.024928347663800574, | |
| "grad_norm": 1.0555644035339355, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.3337, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.025061654335799505, | |
| "grad_norm": 0.9022650718688965, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.0523, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.02519496100779844, | |
| "grad_norm": 1.038295865058899, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.2955, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.025328267679797373, | |
| "grad_norm": 1.0671110153198242, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.2303, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.025461574351796308, | |
| "grad_norm": 1.0591447353363037, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.2003, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.02559488102379524, | |
| "grad_norm": 1.053215742111206, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.2151, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.025728187695794175, | |
| "grad_norm": 1.0813398361206055, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.2032, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.025861494367793107, | |
| "grad_norm": 1.0762261152267456, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.075, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.025994801039792043, | |
| "grad_norm": 1.1010886430740356, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.1568, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.026128107711790974, | |
| "grad_norm": 1.2374165058135986, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.2323, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.02626141438378991, | |
| "grad_norm": 1.644420862197876, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.2802, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.026394721055788842, | |
| "grad_norm": 1.7157775163650513, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 1.2428, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.026528027727787777, | |
| "grad_norm": 2.5402917861938477, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.2513, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.02666133439978671, | |
| "grad_norm": 3.6814701557159424, | |
| "learning_rate": 0.0, | |
| "loss": 1.0864, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02666133439978671, | |
| "eval_loss": 1.2174347639083862, | |
| "eval_runtime": 897.4909, | |
| "eval_samples_per_second": 14.077, | |
| "eval_steps_per_second": 3.52, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.84377737068544e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |