{ "best_metric": null, "best_model_checkpoint": null, "epoch": 6.245004513516506e-13, "eval_steps": 500, "global_step": 90000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 6.938893903907228e-17, "grad_norm": 21.603912353515625, "learning_rate": 7.499999999999999e-07, "loss": 10.2997, "step": 10 }, { "epoch": 1.3877787807814457e-16, "grad_norm": 22.350404739379883, "learning_rate": 1.4999999999999998e-06, "loss": 10.0237, "step": 20 }, { "epoch": 2.0816681711721685e-16, "grad_norm": 21.482810974121094, "learning_rate": 2.2499999999999996e-06, "loss": 9.4391, "step": 30 }, { "epoch": 2.7755575615628914e-16, "grad_norm": 17.7673397064209, "learning_rate": 2.9999999999999997e-06, "loss": 8.7461, "step": 40 }, { "epoch": 3.469446951953614e-16, "grad_norm": 14.069966316223145, "learning_rate": 3.7499999999999997e-06, "loss": 8.0342, "step": 50 }, { "epoch": 4.163336342344337e-16, "grad_norm": 13.204521179199219, "learning_rate": 4.499999999999999e-06, "loss": 7.2676, "step": 60 }, { "epoch": 4.85722573273506e-16, "grad_norm": 10.994033813476562, "learning_rate": 5.25e-06, "loss": 6.7639, "step": 70 }, { "epoch": 5.551115123125783e-16, "grad_norm": 9.357372283935547, "learning_rate": 5.999999999999999e-06, "loss": 6.4835, "step": 80 }, { "epoch": 6.245004513516506e-16, "grad_norm": 10.344311714172363, "learning_rate": 6.749999999999999e-06, "loss": 6.1994, "step": 90 }, { "epoch": 6.938893903907228e-16, "grad_norm": 10.884379386901855, "learning_rate": 7.499999999999999e-06, "loss": 6.113, "step": 100 }, { "epoch": 7.632783294297951e-16, "grad_norm": 9.956489562988281, "learning_rate": 8.249999999999999e-06, "loss": 6.082, "step": 110 }, { "epoch": 8.326672684688674e-16, "grad_norm": 10.807823181152344, "learning_rate": 8.999999999999999e-06, "loss": 6.1618, "step": 120 }, { "epoch": 9.020562075079397e-16, "grad_norm": 10.999185562133789, "learning_rate": 9.75e-06, "loss": 5.9188, "step": 130 }, { "epoch": 9.71445146547012e-16, "grad_norm": 11.480067253112793, "learning_rate": 1.05e-05, "loss": 5.8537, "step": 140 }, { "epoch": 1.0408340855860843e-15, "grad_norm": 11.149374008178711, "learning_rate": 1.1249999999999999e-05, "loss": 5.9001, "step": 150 }, { "epoch": 1.1102230246251565e-15, "grad_norm": 9.62448787689209, "learning_rate": 1.1999999999999999e-05, "loss": 5.8593, "step": 160 }, { "epoch": 1.1796119636642288e-15, "grad_norm": 8.863595962524414, "learning_rate": 1.275e-05, "loss": 5.7685, "step": 170 }, { "epoch": 1.2490009027033011e-15, "grad_norm": 11.287325859069824, "learning_rate": 1.3499999999999998e-05, "loss": 5.6345, "step": 180 }, { "epoch": 1.3183898417423734e-15, "grad_norm": 10.502348899841309, "learning_rate": 1.4249999999999999e-05, "loss": 5.5092, "step": 190 }, { "epoch": 1.3877787807814457e-15, "grad_norm": 10.425962448120117, "learning_rate": 1.4999999999999999e-05, "loss": 5.5454, "step": 200 }, { "epoch": 1.457167719820518e-15, "grad_norm": 10.038893699645996, "learning_rate": 1.5749999999999997e-05, "loss": 5.5569, "step": 210 }, { "epoch": 1.5265566588595902e-15, "grad_norm": 10.505363464355469, "learning_rate": 1.6499999999999998e-05, "loss": 5.3232, "step": 220 }, { "epoch": 1.5959455978986625e-15, "grad_norm": 10.680718421936035, "learning_rate": 1.725e-05, "loss": 5.2222, "step": 230 }, { "epoch": 1.6653345369377348e-15, "grad_norm": 9.730362892150879, "learning_rate": 1.7999999999999997e-05, "loss": 5.2259, "step": 240 }, { "epoch": 1.734723475976807e-15, "grad_norm": 10.20909595489502, "learning_rate": 1.875e-05, "loss": 5.0262, "step": 250 }, { "epoch": 1.8041124150158794e-15, "grad_norm": 8.95288372039795, "learning_rate": 1.95e-05, "loss": 4.9169, "step": 260 }, { "epoch": 1.8735013540549517e-15, "grad_norm": 10.00352954864502, "learning_rate": 2.025e-05, "loss": 4.8945, "step": 270 }, { "epoch": 1.942890293094024e-15, "grad_norm": 9.83011531829834, "learning_rate": 2.1e-05, "loss": 4.8455, "step": 280 }, { "epoch": 2.0122792321330962e-15, "grad_norm": 9.579377174377441, "learning_rate": 2.1749999999999997e-05, "loss": 4.6297, "step": 290 }, { "epoch": 2.0816681711721685e-15, "grad_norm": 7.357174873352051, "learning_rate": 2.2499999999999998e-05, "loss": 4.6716, "step": 300 }, { "epoch": 2.1510571102112408e-15, "grad_norm": 8.549370765686035, "learning_rate": 2.325e-05, "loss": 4.4728, "step": 310 }, { "epoch": 2.220446049250313e-15, "grad_norm": 7.593695640563965, "learning_rate": 2.3999999999999997e-05, "loss": 4.53, "step": 320 }, { "epoch": 2.2898349882893854e-15, "grad_norm": 7.1186676025390625, "learning_rate": 2.475e-05, "loss": 4.5274, "step": 330 }, { "epoch": 2.3592239273284576e-15, "grad_norm": 7.1292290687561035, "learning_rate": 2.55e-05, "loss": 4.1722, "step": 340 }, { "epoch": 2.42861286636753e-15, "grad_norm": 5.938808441162109, "learning_rate": 2.6249999999999998e-05, "loss": 4.2694, "step": 350 }, { "epoch": 2.4980018054066022e-15, "grad_norm": 5.501691818237305, "learning_rate": 2.6999999999999996e-05, "loss": 4.1153, "step": 360 }, { "epoch": 2.5673907444456745e-15, "grad_norm": 5.57135534286499, "learning_rate": 2.7749999999999997e-05, "loss": 4.0644, "step": 370 }, { "epoch": 2.6367796834847468e-15, "grad_norm": 5.516888618469238, "learning_rate": 2.8499999999999998e-05, "loss": 3.9032, "step": 380 }, { "epoch": 2.706168622523819e-15, "grad_norm": 4.758039474487305, "learning_rate": 2.925e-05, "loss": 3.6999, "step": 390 }, { "epoch": 2.7755575615628914e-15, "grad_norm": 4.181853294372559, "learning_rate": 2.9999999999999997e-05, "loss": 3.7477, "step": 400 }, { "epoch": 2.8449465006019636e-15, "grad_norm": 4.0829620361328125, "learning_rate": 3.0749999999999995e-05, "loss": 3.6059, "step": 410 }, { "epoch": 2.914335439641036e-15, "grad_norm": 3.986248254776001, "learning_rate": 3.149999999999999e-05, "loss": 3.5565, "step": 420 }, { "epoch": 2.9837243786801082e-15, "grad_norm": 3.1414167881011963, "learning_rate": 3.225e-05, "loss": 3.5568, "step": 430 }, { "epoch": 3.0531133177191805e-15, "grad_norm": 2.6899008750915527, "learning_rate": 3.2999999999999996e-05, "loss": 3.4629, "step": 440 }, { "epoch": 3.1225022567582528e-15, "grad_norm": 2.349440813064575, "learning_rate": 3.375e-05, "loss": 3.3943, "step": 450 }, { "epoch": 3.191891195797325e-15, "grad_norm": 2.074993848800659, "learning_rate": 3.45e-05, "loss": 3.3028, "step": 460 }, { "epoch": 3.2612801348363973e-15, "grad_norm": 1.6779011487960815, "learning_rate": 3.5249999999999996e-05, "loss": 3.3398, "step": 470 }, { "epoch": 3.3306690738754696e-15, "grad_norm": 1.4486678838729858, "learning_rate": 3.5999999999999994e-05, "loss": 3.3338, "step": 480 }, { "epoch": 3.400058012914542e-15, "grad_norm": 1.2583461999893188, "learning_rate": 3.675e-05, "loss": 3.361, "step": 490 }, { "epoch": 3.469446951953614e-15, "grad_norm": 1.058334231376648, "learning_rate": 3.75e-05, "loss": 3.1161, "step": 500 }, { "epoch": 3.5388358909926865e-15, "grad_norm": 1.054146409034729, "learning_rate": 3.8249999999999995e-05, "loss": 3.2151, "step": 510 }, { "epoch": 3.608224830031759e-15, "grad_norm": 0.894980251789093, "learning_rate": 3.9e-05, "loss": 3.2872, "step": 520 }, { "epoch": 3.677613769070831e-15, "grad_norm": 0.7013705372810364, "learning_rate": 3.975e-05, "loss": 3.114, "step": 530 }, { "epoch": 3.747002708109903e-15, "grad_norm": 0.6774550676345825, "learning_rate": 4.05e-05, "loss": 3.0174, "step": 540 }, { "epoch": 3.816391647148976e-15, "grad_norm": 0.5842148661613464, "learning_rate": 4.125e-05, "loss": 3.0517, "step": 550 }, { "epoch": 3.885780586188048e-15, "grad_norm": 0.5087082982063293, "learning_rate": 4.2e-05, "loss": 2.9807, "step": 560 }, { "epoch": 3.95516952522712e-15, "grad_norm": 0.7345365881919861, "learning_rate": 4.2749999999999996e-05, "loss": 2.9204, "step": 570 }, { "epoch": 4.0245584642661925e-15, "grad_norm": 0.5212576389312744, "learning_rate": 4.3499999999999993e-05, "loss": 2.998, "step": 580 }, { "epoch": 4.093947403305265e-15, "grad_norm": 0.5185524821281433, "learning_rate": 4.424999999999999e-05, "loss": 2.978, "step": 590 }, { "epoch": 4.163336342344337e-15, "grad_norm": 0.5526247620582581, "learning_rate": 4.4999999999999996e-05, "loss": 2.9087, "step": 600 }, { "epoch": 4.232725281383409e-15, "grad_norm": 0.6272593140602112, "learning_rate": 4.5749999999999994e-05, "loss": 2.864, "step": 610 }, { "epoch": 4.3021142204224816e-15, "grad_norm": 0.5314525961875916, "learning_rate": 4.65e-05, "loss": 2.899, "step": 620 }, { "epoch": 4.371503159461554e-15, "grad_norm": 0.8327052593231201, "learning_rate": 4.7249999999999997e-05, "loss": 2.971, "step": 630 }, { "epoch": 4.440892098500626e-15, "grad_norm": 0.503357470035553, "learning_rate": 4.7999999999999994e-05, "loss": 2.9781, "step": 640 }, { "epoch": 4.5102810375396984e-15, "grad_norm": 0.5766547322273254, "learning_rate": 4.875e-05, "loss": 2.8749, "step": 650 }, { "epoch": 4.579669976578771e-15, "grad_norm": 0.7066410779953003, "learning_rate": 4.95e-05, "loss": 2.8688, "step": 660 }, { "epoch": 4.649058915617843e-15, "grad_norm": 0.5730259418487549, "learning_rate": 5.025e-05, "loss": 2.876, "step": 670 }, { "epoch": 4.718447854656915e-15, "grad_norm": 0.5798364281654358, "learning_rate": 5.1e-05, "loss": 2.9649, "step": 680 }, { "epoch": 4.7878367936959876e-15, "grad_norm": 0.49825936555862427, "learning_rate": 5.174999999999999e-05, "loss": 2.7736, "step": 690 }, { "epoch": 4.85722573273506e-15, "grad_norm": 0.715299665927887, "learning_rate": 5.2499999999999995e-05, "loss": 2.8212, "step": 700 }, { "epoch": 4.926614671774132e-15, "grad_norm": 0.6923186779022217, "learning_rate": 5.324999999999999e-05, "loss": 2.768, "step": 710 }, { "epoch": 4.9960036108132044e-15, "grad_norm": 0.5520873069763184, "learning_rate": 5.399999999999999e-05, "loss": 2.6961, "step": 720 }, { "epoch": 5.065392549852277e-15, "grad_norm": 0.7839295268058777, "learning_rate": 5.4749999999999996e-05, "loss": 2.731, "step": 730 }, { "epoch": 5.134781488891349e-15, "grad_norm": 0.5378012657165527, "learning_rate": 5.5499999999999994e-05, "loss": 2.741, "step": 740 }, { "epoch": 5.204170427930421e-15, "grad_norm": 0.6633866429328918, "learning_rate": 5.625e-05, "loss": 2.7296, "step": 750 }, { "epoch": 5.2735593669694936e-15, "grad_norm": 0.6684882044792175, "learning_rate": 5.6999999999999996e-05, "loss": 2.6805, "step": 760 }, { "epoch": 5.342948306008566e-15, "grad_norm": 1.053054928779602, "learning_rate": 5.7749999999999994e-05, "loss": 2.7198, "step": 770 }, { "epoch": 5.412337245047638e-15, "grad_norm": 0.9652512073516846, "learning_rate": 5.85e-05, "loss": 2.6585, "step": 780 }, { "epoch": 5.4817261840867104e-15, "grad_norm": 0.5723056793212891, "learning_rate": 5.925e-05, "loss": 2.7521, "step": 790 }, { "epoch": 5.551115123125783e-15, "grad_norm": 0.7153953909873962, "learning_rate": 5.9999999999999995e-05, "loss": 2.8171, "step": 800 }, { "epoch": 5.620504062164855e-15, "grad_norm": 0.7370086908340454, "learning_rate": 6.075e-05, "loss": 2.6039, "step": 810 }, { "epoch": 5.689893001203927e-15, "grad_norm": 0.7195964455604553, "learning_rate": 6.149999999999999e-05, "loss": 2.7794, "step": 820 }, { "epoch": 5.7592819402429996e-15, "grad_norm": 1.1993961334228516, "learning_rate": 6.225e-05, "loss": 2.7232, "step": 830 }, { "epoch": 5.828670879282072e-15, "grad_norm": 0.7800962924957275, "learning_rate": 6.299999999999999e-05, "loss": 2.6843, "step": 840 }, { "epoch": 5.898059818321144e-15, "grad_norm": 0.7839193344116211, "learning_rate": 6.374999999999999e-05, "loss": 2.5716, "step": 850 }, { "epoch": 5.9674487573602164e-15, "grad_norm": 0.9984657764434814, "learning_rate": 6.45e-05, "loss": 2.5913, "step": 860 }, { "epoch": 6.036837696399289e-15, "grad_norm": 0.8234934210777283, "learning_rate": 6.525e-05, "loss": 2.5741, "step": 870 }, { "epoch": 6.106226635438361e-15, "grad_norm": 0.7158179879188538, "learning_rate": 6.599999999999999e-05, "loss": 2.5528, "step": 880 }, { "epoch": 6.175615574477433e-15, "grad_norm": 0.8789527416229248, "learning_rate": 6.675e-05, "loss": 2.5499, "step": 890 }, { "epoch": 6.2450045135165055e-15, "grad_norm": 1.3568708896636963, "learning_rate": 6.75e-05, "loss": 2.6886, "step": 900 }, { "epoch": 6.314393452555578e-15, "grad_norm": 0.6734881401062012, "learning_rate": 6.824999999999999e-05, "loss": 2.5105, "step": 910 }, { "epoch": 6.38378239159465e-15, "grad_norm": 0.4938907325267792, "learning_rate": 6.9e-05, "loss": 2.5293, "step": 920 }, { "epoch": 6.453171330633722e-15, "grad_norm": 0.7452742457389832, "learning_rate": 6.975e-05, "loss": 2.5472, "step": 930 }, { "epoch": 6.522560269672795e-15, "grad_norm": 0.7887052893638611, "learning_rate": 7.049999999999999e-05, "loss": 2.5817, "step": 940 }, { "epoch": 6.591949208711867e-15, "grad_norm": 0.9176592826843262, "learning_rate": 7.125e-05, "loss": 2.5827, "step": 950 }, { "epoch": 6.661338147750939e-15, "grad_norm": 0.8711972832679749, "learning_rate": 7.199999999999999e-05, "loss": 2.6701, "step": 960 }, { "epoch": 6.7307270867900115e-15, "grad_norm": 0.9593539237976074, "learning_rate": 7.274999999999999e-05, "loss": 2.4993, "step": 970 }, { "epoch": 6.800116025829084e-15, "grad_norm": 0.7958654761314392, "learning_rate": 7.35e-05, "loss": 2.561, "step": 980 }, { "epoch": 6.869504964868156e-15, "grad_norm": 0.7547851204872131, "learning_rate": 7.424999999999999e-05, "loss": 2.6253, "step": 990 }, { "epoch": 6.938893903907228e-15, "grad_norm": 1.3917373418807983, "learning_rate": 7.5e-05, "loss": 2.5108, "step": 1000 }, { "epoch": 7.008282842946301e-15, "grad_norm": 0.8574291467666626, "learning_rate": 7.575e-05, "loss": 2.4712, "step": 1010 }, { "epoch": 7.077671781985373e-15, "grad_norm": 1.1404145956039429, "learning_rate": 7.649999999999999e-05, "loss": 2.4707, "step": 1020 }, { "epoch": 7.147060721024445e-15, "grad_norm": 1.2319828271865845, "learning_rate": 7.725e-05, "loss": 2.4642, "step": 1030 }, { "epoch": 7.216449660063518e-15, "grad_norm": 0.9277602434158325, "learning_rate": 7.8e-05, "loss": 2.3629, "step": 1040 }, { "epoch": 7.28583859910259e-15, "grad_norm": 0.998703122138977, "learning_rate": 7.874999999999999e-05, "loss": 2.4805, "step": 1050 }, { "epoch": 7.355227538141662e-15, "grad_norm": 1.012678861618042, "learning_rate": 7.95e-05, "loss": 2.4774, "step": 1060 }, { "epoch": 7.424616477180734e-15, "grad_norm": 1.1156381368637085, "learning_rate": 8.025e-05, "loss": 2.3839, "step": 1070 }, { "epoch": 7.494005416219807e-15, "grad_norm": 0.8894811868667603, "learning_rate": 8.1e-05, "loss": 2.3982, "step": 1080 }, { "epoch": 7.563394355258879e-15, "grad_norm": 1.0015064477920532, "learning_rate": 8.175e-05, "loss": 2.4295, "step": 1090 }, { "epoch": 7.632783294297951e-15, "grad_norm": 0.8670275211334229, "learning_rate": 8.25e-05, "loss": 2.4825, "step": 1100 }, { "epoch": 7.702172233337023e-15, "grad_norm": 0.8077796697616577, "learning_rate": 8.325e-05, "loss": 2.5224, "step": 1110 }, { "epoch": 7.771561172376096e-15, "grad_norm": 0.909069836139679, "learning_rate": 8.4e-05, "loss": 2.3938, "step": 1120 }, { "epoch": 7.840950111415168e-15, "grad_norm": 1.0749292373657227, "learning_rate": 8.474999999999999e-05, "loss": 2.4737, "step": 1130 }, { "epoch": 7.91033905045424e-15, "grad_norm": 1.3984097242355347, "learning_rate": 8.549999999999999e-05, "loss": 2.4556, "step": 1140 }, { "epoch": 7.979727989493313e-15, "grad_norm": 1.0478098392486572, "learning_rate": 8.624999999999998e-05, "loss": 2.4832, "step": 1150 }, { "epoch": 8.049116928532385e-15, "grad_norm": 1.3634231090545654, "learning_rate": 8.699999999999999e-05, "loss": 2.3859, "step": 1160 }, { "epoch": 8.118505867571457e-15, "grad_norm": 1.127396583557129, "learning_rate": 8.774999999999999e-05, "loss": 2.3073, "step": 1170 }, { "epoch": 8.18789480661053e-15, "grad_norm": 0.6003841757774353, "learning_rate": 8.849999999999998e-05, "loss": 2.3594, "step": 1180 }, { "epoch": 8.257283745649602e-15, "grad_norm": 0.8309741616249084, "learning_rate": 8.924999999999999e-05, "loss": 2.3489, "step": 1190 }, { "epoch": 8.326672684688674e-15, "grad_norm": 1.0647755861282349, "learning_rate": 8.999999999999999e-05, "loss": 2.3601, "step": 1200 }, { "epoch": 8.396061623727746e-15, "grad_norm": 1.2871073484420776, "learning_rate": 9.074999999999998e-05, "loss": 2.3475, "step": 1210 }, { "epoch": 8.465450562766819e-15, "grad_norm": 1.1297383308410645, "learning_rate": 9.149999999999999e-05, "loss": 2.3378, "step": 1220 }, { "epoch": 8.534839501805891e-15, "grad_norm": 0.9135991930961609, "learning_rate": 9.224999999999999e-05, "loss": 2.3313, "step": 1230 }, { "epoch": 8.604228440844963e-15, "grad_norm": 0.9683090448379517, "learning_rate": 9.3e-05, "loss": 2.3088, "step": 1240 }, { "epoch": 8.673617379884035e-15, "grad_norm": 1.0339750051498413, "learning_rate": 9.374999999999999e-05, "loss": 2.3141, "step": 1250 }, { "epoch": 8.743006318923108e-15, "grad_norm": 0.9748283624649048, "learning_rate": 9.449999999999999e-05, "loss": 2.3705, "step": 1260 }, { "epoch": 8.81239525796218e-15, "grad_norm": 1.2086246013641357, "learning_rate": 9.525e-05, "loss": 2.4268, "step": 1270 }, { "epoch": 8.881784197001252e-15, "grad_norm": 0.9780606627464294, "learning_rate": 9.599999999999999e-05, "loss": 2.2616, "step": 1280 }, { "epoch": 8.951173136040325e-15, "grad_norm": 1.0460011959075928, "learning_rate": 9.675e-05, "loss": 2.3968, "step": 1290 }, { "epoch": 9.020562075079397e-15, "grad_norm": 0.7737791538238525, "learning_rate": 9.75e-05, "loss": 2.4006, "step": 1300 }, { "epoch": 9.089951014118469e-15, "grad_norm": 0.7884860038757324, "learning_rate": 9.824999999999999e-05, "loss": 2.2832, "step": 1310 }, { "epoch": 9.159339953157541e-15, "grad_norm": 1.0298371315002441, "learning_rate": 9.9e-05, "loss": 2.259, "step": 1320 }, { "epoch": 9.228728892196614e-15, "grad_norm": 1.0049868822097778, "learning_rate": 9.975e-05, "loss": 2.3002, "step": 1330 }, { "epoch": 9.298117831235686e-15, "grad_norm": 0.8102812767028809, "learning_rate": 0.0001005, "loss": 2.2065, "step": 1340 }, { "epoch": 9.367506770274758e-15, "grad_norm": 0.9440317153930664, "learning_rate": 0.00010125, "loss": 2.2089, "step": 1350 }, { "epoch": 9.43689570931383e-15, "grad_norm": 0.9205601811408997, "learning_rate": 0.000102, "loss": 2.2298, "step": 1360 }, { "epoch": 9.506284648352903e-15, "grad_norm": 0.7491424679756165, "learning_rate": 0.00010275, "loss": 2.3335, "step": 1370 }, { "epoch": 9.575673587391975e-15, "grad_norm": 0.967971682548523, "learning_rate": 0.00010349999999999998, "loss": 2.1904, "step": 1380 }, { "epoch": 9.645062526431047e-15, "grad_norm": 0.8644431233406067, "learning_rate": 0.00010424999999999999, "loss": 2.1913, "step": 1390 }, { "epoch": 9.71445146547012e-15, "grad_norm": 0.8563584089279175, "learning_rate": 0.00010499999999999999, "loss": 2.2521, "step": 1400 }, { "epoch": 9.783840404509192e-15, "grad_norm": 1.0663694143295288, "learning_rate": 0.00010574999999999998, "loss": 2.287, "step": 1410 }, { "epoch": 9.853229343548264e-15, "grad_norm": 1.3856794834136963, "learning_rate": 0.00010649999999999999, "loss": 2.2951, "step": 1420 }, { "epoch": 9.922618282587337e-15, "grad_norm": 0.9127955436706543, "learning_rate": 0.00010724999999999999, "loss": 2.2583, "step": 1430 }, { "epoch": 9.992007221626409e-15, "grad_norm": 0.9056940078735352, "learning_rate": 0.00010799999999999998, "loss": 2.2135, "step": 1440 }, { "epoch": 1.0061396160665481e-14, "grad_norm": 0.7638794779777527, "learning_rate": 0.00010874999999999999, "loss": 2.2702, "step": 1450 }, { "epoch": 1.0130785099704553e-14, "grad_norm": 1.3194769620895386, "learning_rate": 0.00010949999999999999, "loss": 2.3089, "step": 1460 }, { "epoch": 1.0200174038743626e-14, "grad_norm": 0.9809593558311462, "learning_rate": 0.00011024999999999998, "loss": 2.1993, "step": 1470 }, { "epoch": 1.0269562977782698e-14, "grad_norm": 0.9784894585609436, "learning_rate": 0.00011099999999999999, "loss": 2.1659, "step": 1480 }, { "epoch": 1.033895191682177e-14, "grad_norm": 0.8815931677818298, "learning_rate": 0.00011174999999999999, "loss": 2.2198, "step": 1490 }, { "epoch": 1.0408340855860843e-14, "grad_norm": 0.8990421891212463, "learning_rate": 0.0001125, "loss": 2.1296, "step": 1500 }, { "epoch": 1.0477729794899915e-14, "grad_norm": 0.8911024928092957, "learning_rate": 0.00011324999999999999, "loss": 2.1069, "step": 1510 }, { "epoch": 1.0547118733938987e-14, "grad_norm": 0.8203184008598328, "learning_rate": 0.00011399999999999999, "loss": 2.1644, "step": 1520 }, { "epoch": 1.061650767297806e-14, "grad_norm": 0.6863431334495544, "learning_rate": 0.00011475, "loss": 2.1947, "step": 1530 }, { "epoch": 1.0685896612017132e-14, "grad_norm": 0.9622281789779663, "learning_rate": 0.00011549999999999999, "loss": 2.1081, "step": 1540 }, { "epoch": 1.0755285551056204e-14, "grad_norm": 1.2254738807678223, "learning_rate": 0.00011624999999999999, "loss": 2.1878, "step": 1550 }, { "epoch": 1.0824674490095276e-14, "grad_norm": 1.0711702108383179, "learning_rate": 0.000117, "loss": 2.1141, "step": 1560 }, { "epoch": 1.0894063429134349e-14, "grad_norm": 1.1571537256240845, "learning_rate": 0.00011774999999999999, "loss": 2.2055, "step": 1570 }, { "epoch": 1.0963452368173421e-14, "grad_norm": 0.8964964151382446, "learning_rate": 0.0001185, "loss": 2.2704, "step": 1580 }, { "epoch": 1.1032841307212493e-14, "grad_norm": 1.0875707864761353, "learning_rate": 0.00011925, "loss": 2.1001, "step": 1590 }, { "epoch": 1.1102230246251565e-14, "grad_norm": 0.9118460416793823, "learning_rate": 0.00011999999999999999, "loss": 2.2217, "step": 1600 }, { "epoch": 1.1171619185290638e-14, "grad_norm": 1.3307796716690063, "learning_rate": 0.00012075, "loss": 2.1582, "step": 1610 }, { "epoch": 1.124100812432971e-14, "grad_norm": 1.0990036725997925, "learning_rate": 0.0001215, "loss": 2.1787, "step": 1620 }, { "epoch": 1.1310397063368782e-14, "grad_norm": 0.8576607704162598, "learning_rate": 0.00012225, "loss": 2.1254, "step": 1630 }, { "epoch": 1.1379786002407855e-14, "grad_norm": 0.8424851894378662, "learning_rate": 0.00012299999999999998, "loss": 2.0401, "step": 1640 }, { "epoch": 1.1449174941446927e-14, "grad_norm": 1.2170507907867432, "learning_rate": 0.00012374999999999997, "loss": 2.1238, "step": 1650 }, { "epoch": 1.1518563880485999e-14, "grad_norm": 0.8347296118736267, "learning_rate": 0.0001245, "loss": 2.0549, "step": 1660 }, { "epoch": 1.1587952819525071e-14, "grad_norm": 0.818697452545166, "learning_rate": 0.00012524999999999998, "loss": 2.0736, "step": 1670 }, { "epoch": 1.1657341758564144e-14, "grad_norm": 0.6639633178710938, "learning_rate": 0.00012599999999999997, "loss": 2.1261, "step": 1680 }, { "epoch": 1.1726730697603216e-14, "grad_norm": 1.0388716459274292, "learning_rate": 0.00012675, "loss": 2.0936, "step": 1690 }, { "epoch": 1.1796119636642288e-14, "grad_norm": 0.9513992667198181, "learning_rate": 0.00012749999999999998, "loss": 2.0938, "step": 1700 }, { "epoch": 1.186550857568136e-14, "grad_norm": 0.850586473941803, "learning_rate": 0.00012824999999999997, "loss": 2.0492, "step": 1710 }, { "epoch": 1.1934897514720433e-14, "grad_norm": 1.0611683130264282, "learning_rate": 0.000129, "loss": 2.094, "step": 1720 }, { "epoch": 1.2004286453759505e-14, "grad_norm": 0.8954281806945801, "learning_rate": 0.00012974999999999998, "loss": 2.1151, "step": 1730 }, { "epoch": 1.2073675392798577e-14, "grad_norm": 0.8540951013565063, "learning_rate": 0.0001305, "loss": 2.1586, "step": 1740 }, { "epoch": 1.214306433183765e-14, "grad_norm": 1.0360641479492188, "learning_rate": 0.00013125, "loss": 2.0286, "step": 1750 }, { "epoch": 1.2212453270876722e-14, "grad_norm": 0.7008628845214844, "learning_rate": 0.00013199999999999998, "loss": 2.1095, "step": 1760 }, { "epoch": 1.2281842209915794e-14, "grad_norm": 0.875506579875946, "learning_rate": 0.00013275, "loss": 2.1668, "step": 1770 }, { "epoch": 1.2351231148954867e-14, "grad_norm": 0.9525253772735596, "learning_rate": 0.0001335, "loss": 2.0696, "step": 1780 }, { "epoch": 1.2420620087993939e-14, "grad_norm": 0.9385840892791748, "learning_rate": 0.00013424999999999998, "loss": 2.0087, "step": 1790 }, { "epoch": 1.2490009027033011e-14, "grad_norm": 0.7559543251991272, "learning_rate": 0.000135, "loss": 2.044, "step": 1800 }, { "epoch": 1.2559397966072083e-14, "grad_norm": 0.7638208270072937, "learning_rate": 0.00013575, "loss": 2.0074, "step": 1810 }, { "epoch": 1.2628786905111156e-14, "grad_norm": 0.9758471250534058, "learning_rate": 0.00013649999999999998, "loss": 1.9645, "step": 1820 }, { "epoch": 1.2698175844150228e-14, "grad_norm": 1.090558648109436, "learning_rate": 0.00013725, "loss": 2.034, "step": 1830 }, { "epoch": 1.27675647831893e-14, "grad_norm": 0.7952758073806763, "learning_rate": 0.000138, "loss": 2.0487, "step": 1840 }, { "epoch": 1.2836953722228372e-14, "grad_norm": 1.0904946327209473, "learning_rate": 0.00013874999999999998, "loss": 1.9987, "step": 1850 }, { "epoch": 1.2906342661267445e-14, "grad_norm": 0.9957982897758484, "learning_rate": 0.0001395, "loss": 1.9881, "step": 1860 }, { "epoch": 1.2975731600306517e-14, "grad_norm": 0.8498905301094055, "learning_rate": 0.00014025, "loss": 2.0025, "step": 1870 }, { "epoch": 1.304512053934559e-14, "grad_norm": 1.1275601387023926, "learning_rate": 0.00014099999999999998, "loss": 2.0594, "step": 1880 }, { "epoch": 1.3114509478384662e-14, "grad_norm": 1.153213381767273, "learning_rate": 0.00014174999999999998, "loss": 2.0743, "step": 1890 }, { "epoch": 1.3183898417423734e-14, "grad_norm": 0.7083019614219666, "learning_rate": 0.0001425, "loss": 2.0177, "step": 1900 }, { "epoch": 1.3253287356462806e-14, "grad_norm": 0.8573106527328491, "learning_rate": 0.00014324999999999999, "loss": 2.0152, "step": 1910 }, { "epoch": 1.3322676295501878e-14, "grad_norm": 0.7648389935493469, "learning_rate": 0.00014399999999999998, "loss": 2.0271, "step": 1920 }, { "epoch": 1.339206523454095e-14, "grad_norm": 0.8933773040771484, "learning_rate": 0.00014475, "loss": 2.0949, "step": 1930 }, { "epoch": 1.3461454173580023e-14, "grad_norm": 0.8451698422431946, "learning_rate": 0.00014549999999999999, "loss": 1.9671, "step": 1940 }, { "epoch": 1.3530843112619095e-14, "grad_norm": 0.9913836717605591, "learning_rate": 0.00014624999999999998, "loss": 2.0046, "step": 1950 }, { "epoch": 1.3600232051658168e-14, "grad_norm": 0.7171565294265747, "learning_rate": 0.000147, "loss": 1.9697, "step": 1960 }, { "epoch": 1.366962099069724e-14, "grad_norm": 0.7940084338188171, "learning_rate": 0.00014774999999999999, "loss": 1.9112, "step": 1970 }, { "epoch": 1.3739009929736312e-14, "grad_norm": 0.8989030122756958, "learning_rate": 0.00014849999999999998, "loss": 1.9447, "step": 1980 }, { "epoch": 1.3808398868775384e-14, "grad_norm": 1.1493581533432007, "learning_rate": 0.00014925, "loss": 1.9683, "step": 1990 }, { "epoch": 1.3877787807814457e-14, "grad_norm": 0.7463393211364746, "learning_rate": 0.00015, "loss": 1.9684, "step": 2000 }, { "epoch": 1.3947176746853529e-14, "grad_norm": 1.0131103992462158, "learning_rate": 0.00014999999653166236, "loss": 1.9357, "step": 2010 }, { "epoch": 1.4016565685892601e-14, "grad_norm": 0.8324626684188843, "learning_rate": 0.00014999998612664985, "loss": 1.9726, "step": 2020 }, { "epoch": 1.4085954624931674e-14, "grad_norm": 1.0543251037597656, "learning_rate": 0.0001499999687849635, "loss": 1.9244, "step": 2030 }, { "epoch": 1.4155343563970746e-14, "grad_norm": 0.8291214108467102, "learning_rate": 0.0001499999445066051, "loss": 1.989, "step": 2040 }, { "epoch": 1.4224732503009818e-14, "grad_norm": 0.651785671710968, "learning_rate": 0.00014999991329157713, "loss": 2.0485, "step": 2050 }, { "epoch": 1.429412144204889e-14, "grad_norm": 0.8870018720626831, "learning_rate": 0.00014999987513988287, "loss": 1.9023, "step": 2060 }, { "epoch": 1.4363510381087963e-14, "grad_norm": 0.9270675182342529, "learning_rate": 0.00014999983005152613, "loss": 2.0293, "step": 2070 }, { "epoch": 1.4432899320127035e-14, "grad_norm": 0.9134874939918518, "learning_rate": 0.00014999977802651162, "loss": 1.9957, "step": 2080 }, { "epoch": 1.4502288259166107e-14, "grad_norm": 0.8927558660507202, "learning_rate": 0.0001499997190648447, "loss": 1.9679, "step": 2090 }, { "epoch": 1.457167719820518e-14, "grad_norm": 0.8213223814964294, "learning_rate": 0.00014999965316653136, "loss": 1.8924, "step": 2100 }, { "epoch": 1.4641066137244252e-14, "grad_norm": 1.0323517322540283, "learning_rate": 0.00014999958033157842, "loss": 1.9001, "step": 2110 }, { "epoch": 1.4710455076283324e-14, "grad_norm": 0.9897832870483398, "learning_rate": 0.00014999950055999336, "loss": 1.8939, "step": 2120 }, { "epoch": 1.4779844015322396e-14, "grad_norm": 0.8330813050270081, "learning_rate": 0.00014999941385178433, "loss": 1.8691, "step": 2130 }, { "epoch": 1.484923295436147e-14, "grad_norm": 0.8520774245262146, "learning_rate": 0.00014999932020696034, "loss": 1.8677, "step": 2140 }, { "epoch": 1.491862189340054e-14, "grad_norm": 1.1963998079299927, "learning_rate": 0.00014999921962553092, "loss": 2.0011, "step": 2150 }, { "epoch": 1.4988010832439613e-14, "grad_norm": 0.9163972735404968, "learning_rate": 0.00014999911210750646, "loss": 1.8536, "step": 2160 }, { "epoch": 1.5057399771478686e-14, "grad_norm": 0.6594530940055847, "learning_rate": 0.000149998997652898, "loss": 1.8847, "step": 2170 }, { "epoch": 1.5126788710517758e-14, "grad_norm": 1.0829864740371704, "learning_rate": 0.00014999887626171725, "loss": 1.8984, "step": 2180 }, { "epoch": 1.519617764955683e-14, "grad_norm": 0.8072304725646973, "learning_rate": 0.00014999874793397678, "loss": 1.9186, "step": 2190 }, { "epoch": 1.5265566588595902e-14, "grad_norm": 0.972638726234436, "learning_rate": 0.00014999861266968968, "loss": 1.916, "step": 2200 }, { "epoch": 1.5334955527634975e-14, "grad_norm": 0.817029595375061, "learning_rate": 0.00014999847046886993, "loss": 1.9955, "step": 2210 }, { "epoch": 1.5404344466674047e-14, "grad_norm": 1.0268336534500122, "learning_rate": 0.0001499983213315321, "loss": 1.8693, "step": 2220 }, { "epoch": 1.547373340571312e-14, "grad_norm": 1.036584734916687, "learning_rate": 0.00014999816525769152, "loss": 1.9212, "step": 2230 }, { "epoch": 1.554312234475219e-14, "grad_norm": 0.7752702832221985, "learning_rate": 0.00014999800224736425, "loss": 1.973, "step": 2240 }, { "epoch": 1.5612511283791264e-14, "grad_norm": 0.7765113711357117, "learning_rate": 0.000149997832300567, "loss": 1.8878, "step": 2250 }, { "epoch": 1.5681900222830336e-14, "grad_norm": 0.9559946060180664, "learning_rate": 0.00014999765541731728, "loss": 1.8529, "step": 2260 }, { "epoch": 1.5751289161869408e-14, "grad_norm": 0.7806987166404724, "learning_rate": 0.00014999747159763325, "loss": 1.8572, "step": 2270 }, { "epoch": 1.582067810090848e-14, "grad_norm": 1.012261986732483, "learning_rate": 0.00014999728084153378, "loss": 1.8514, "step": 2280 }, { "epoch": 1.5890067039947553e-14, "grad_norm": 0.8569822311401367, "learning_rate": 0.0001499970831490385, "loss": 1.7717, "step": 2290 }, { "epoch": 1.5959455978986625e-14, "grad_norm": 0.8552131652832031, "learning_rate": 0.00014999687852016775, "loss": 1.8606, "step": 2300 }, { "epoch": 1.6028844918025698e-14, "grad_norm": 0.9630708694458008, "learning_rate": 0.00014999666695494248, "loss": 1.8807, "step": 2310 }, { "epoch": 1.609823385706477e-14, "grad_norm": 0.9003567695617676, "learning_rate": 0.0001499964484533845, "loss": 1.805, "step": 2320 }, { "epoch": 1.6167622796103842e-14, "grad_norm": 0.7934194803237915, "learning_rate": 0.00014999622301551624, "loss": 1.8252, "step": 2330 }, { "epoch": 1.6237011735142914e-14, "grad_norm": 0.9224129319190979, "learning_rate": 0.00014999599064136087, "loss": 1.8462, "step": 2340 }, { "epoch": 1.6306400674181987e-14, "grad_norm": 0.8912016153335571, "learning_rate": 0.00014999575133094227, "loss": 1.8838, "step": 2350 }, { "epoch": 1.637578961322106e-14, "grad_norm": 0.9116982817649841, "learning_rate": 0.00014999550508428504, "loss": 1.9144, "step": 2360 }, { "epoch": 1.644517855226013e-14, "grad_norm": 0.9381203055381775, "learning_rate": 0.00014999525190141446, "loss": 1.8226, "step": 2370 }, { "epoch": 1.6514567491299204e-14, "grad_norm": 0.9559411406517029, "learning_rate": 0.0001499949917823566, "loss": 1.8918, "step": 2380 }, { "epoch": 1.6583956430338276e-14, "grad_norm": 0.8803228139877319, "learning_rate": 0.00014999472472713814, "loss": 1.8744, "step": 2390 }, { "epoch": 1.6653345369377348e-14, "grad_norm": 0.86564701795578, "learning_rate": 0.00014999445073578653, "loss": 1.8993, "step": 2400 }, { "epoch": 1.672273430841642e-14, "grad_norm": 0.9622906446456909, "learning_rate": 0.00014999416980832997, "loss": 1.8278, "step": 2410 }, { "epoch": 1.6792123247455493e-14, "grad_norm": 0.9098809361457825, "learning_rate": 0.00014999388194479726, "loss": 1.7616, "step": 2420 }, { "epoch": 1.6861512186494565e-14, "grad_norm": 0.8255011439323425, "learning_rate": 0.00014999358714521806, "loss": 1.8137, "step": 2430 }, { "epoch": 1.6930901125533637e-14, "grad_norm": 0.921724796295166, "learning_rate": 0.0001499932854096226, "loss": 1.7955, "step": 2440 }, { "epoch": 1.700029006457271e-14, "grad_norm": 0.9063284397125244, "learning_rate": 0.00014999297673804192, "loss": 1.8032, "step": 2450 }, { "epoch": 1.7069679003611782e-14, "grad_norm": 0.9060503244400024, "learning_rate": 0.00014999266113050776, "loss": 1.8046, "step": 2460 }, { "epoch": 1.7139067942650854e-14, "grad_norm": 1.0166730880737305, "learning_rate": 0.0001499923385870525, "loss": 1.7962, "step": 2470 }, { "epoch": 1.7208456881689926e-14, "grad_norm": 0.8114067316055298, "learning_rate": 0.00014999200910770932, "loss": 1.8016, "step": 2480 }, { "epoch": 1.7277845820729e-14, "grad_norm": 0.8602918386459351, "learning_rate": 0.00014999167269251208, "loss": 1.7869, "step": 2490 }, { "epoch": 1.734723475976807e-14, "grad_norm": 0.827763557434082, "learning_rate": 0.00014999132934149533, "loss": 1.7878, "step": 2500 }, { "epoch": 1.7416623698807143e-14, "grad_norm": 1.0598772764205933, "learning_rate": 0.0001499909790546944, "loss": 1.8261, "step": 2510 }, { "epoch": 1.7486012637846216e-14, "grad_norm": 0.8363236784934998, "learning_rate": 0.00014999062183214526, "loss": 1.8762, "step": 2520 }, { "epoch": 1.7555401576885288e-14, "grad_norm": 0.8335496783256531, "learning_rate": 0.0001499902576738846, "loss": 1.7549, "step": 2530 }, { "epoch": 1.762479051592436e-14, "grad_norm": 0.9701045155525208, "learning_rate": 0.00014998988657994987, "loss": 1.8627, "step": 2540 }, { "epoch": 1.7694179454963432e-14, "grad_norm": 0.7390426397323608, "learning_rate": 0.0001499895085503792, "loss": 1.868, "step": 2550 }, { "epoch": 1.7763568394002505e-14, "grad_norm": 0.6825494766235352, "learning_rate": 0.0001499891235852114, "loss": 1.7779, "step": 2560 }, { "epoch": 1.7832957333041577e-14, "grad_norm": 0.8172840476036072, "learning_rate": 0.0001499887316844861, "loss": 1.7561, "step": 2570 }, { "epoch": 1.790234627208065e-14, "grad_norm": 1.3414615392684937, "learning_rate": 0.00014998833284824353, "loss": 1.7882, "step": 2580 }, { "epoch": 1.797173521111972e-14, "grad_norm": 0.9150567650794983, "learning_rate": 0.0001499879270765247, "loss": 1.7298, "step": 2590 }, { "epoch": 1.8041124150158794e-14, "grad_norm": 0.7639792561531067, "learning_rate": 0.00014998751436937131, "loss": 1.7155, "step": 2600 }, { "epoch": 1.8110513089197866e-14, "grad_norm": 1.0061734914779663, "learning_rate": 0.00014998709472682575, "loss": 1.7318, "step": 2610 }, { "epoch": 1.8179902028236938e-14, "grad_norm": 0.8394237160682678, "learning_rate": 0.00014998666814893113, "loss": 1.8273, "step": 2620 }, { "epoch": 1.824929096727601e-14, "grad_norm": 0.9275315999984741, "learning_rate": 0.00014998623463573132, "loss": 1.7107, "step": 2630 }, { "epoch": 1.8318679906315083e-14, "grad_norm": 0.7962659597396851, "learning_rate": 0.00014998579418727087, "loss": 1.7273, "step": 2640 }, { "epoch": 1.8388068845354155e-14, "grad_norm": 0.7458086013793945, "learning_rate": 0.00014998534680359506, "loss": 1.7686, "step": 2650 }, { "epoch": 1.8457457784393227e-14, "grad_norm": 0.9098437428474426, "learning_rate": 0.0001499848924847498, "loss": 1.7906, "step": 2660 }, { "epoch": 1.85268467234323e-14, "grad_norm": 0.9882850050926208, "learning_rate": 0.00014998443123078186, "loss": 1.7988, "step": 2670 }, { "epoch": 1.8596235662471372e-14, "grad_norm": 0.9308807849884033, "learning_rate": 0.00014998396304173855, "loss": 1.7743, "step": 2680 }, { "epoch": 1.8665624601510444e-14, "grad_norm": 0.893605649471283, "learning_rate": 0.00014998348791766806, "loss": 1.7454, "step": 2690 }, { "epoch": 1.8735013540549517e-14, "grad_norm": 0.8023024797439575, "learning_rate": 0.0001499830058586192, "loss": 1.7939, "step": 2700 }, { "epoch": 1.880440247958859e-14, "grad_norm": 0.8889029622077942, "learning_rate": 0.00014998251686464147, "loss": 1.8271, "step": 2710 }, { "epoch": 1.887379141862766e-14, "grad_norm": 0.7801493406295776, "learning_rate": 0.00014998202093578517, "loss": 1.738, "step": 2720 }, { "epoch": 1.8943180357666733e-14, "grad_norm": 0.7598353028297424, "learning_rate": 0.00014998151807210124, "loss": 1.7038, "step": 2730 }, { "epoch": 1.9012569296705806e-14, "grad_norm": 1.2258628606796265, "learning_rate": 0.00014998100827364136, "loss": 1.7682, "step": 2740 }, { "epoch": 1.9081958235744878e-14, "grad_norm": 0.8115176558494568, "learning_rate": 0.0001499804915404579, "loss": 1.6798, "step": 2750 }, { "epoch": 1.915134717478395e-14, "grad_norm": 1.036686658859253, "learning_rate": 0.000149979967872604, "loss": 1.6606, "step": 2760 }, { "epoch": 1.9220736113823023e-14, "grad_norm": 0.8134544491767883, "learning_rate": 0.0001499794372701335, "loss": 1.7098, "step": 2770 }, { "epoch": 1.9290125052862095e-14, "grad_norm": 0.8433420062065125, "learning_rate": 0.00014997889973310084, "loss": 1.7411, "step": 2780 }, { "epoch": 1.9359513991901167e-14, "grad_norm": 0.944068193435669, "learning_rate": 0.00014997835526156132, "loss": 1.6728, "step": 2790 }, { "epoch": 1.942890293094024e-14, "grad_norm": 1.1273646354675293, "learning_rate": 0.00014997780385557087, "loss": 1.7441, "step": 2800 }, { "epoch": 1.9498291869979312e-14, "grad_norm": 1.0087165832519531, "learning_rate": 0.00014997724551518617, "loss": 1.6834, "step": 2810 }, { "epoch": 1.9567680809018384e-14, "grad_norm": 0.7923544645309448, "learning_rate": 0.0001499766802404646, "loss": 1.7528, "step": 2820 }, { "epoch": 1.9637069748057456e-14, "grad_norm": 0.8438127040863037, "learning_rate": 0.00014997610803146424, "loss": 1.7978, "step": 2830 }, { "epoch": 1.970645868709653e-14, "grad_norm": 0.9322770237922668, "learning_rate": 0.0001499755288882439, "loss": 1.6743, "step": 2840 }, { "epoch": 1.97758476261356e-14, "grad_norm": 1.0682945251464844, "learning_rate": 0.0001499749428108631, "loss": 1.7818, "step": 2850 }, { "epoch": 1.9845236565174673e-14, "grad_norm": 1.0479704141616821, "learning_rate": 0.00014997434979938207, "loss": 1.7317, "step": 2860 }, { "epoch": 1.9914625504213745e-14, "grad_norm": 0.9753880500793457, "learning_rate": 0.00014997374985386172, "loss": 1.7446, "step": 2870 }, { "epoch": 1.9984014443252818e-14, "grad_norm": 0.8568105697631836, "learning_rate": 0.00014997314297436373, "loss": 1.6967, "step": 2880 }, { "epoch": 2.005340338229189e-14, "grad_norm": 0.8884969353675842, "learning_rate": 0.00014997252916095047, "loss": 1.6255, "step": 2890 }, { "epoch": 2.0122792321330962e-14, "grad_norm": 1.1442170143127441, "learning_rate": 0.00014997190841368503, "loss": 1.719, "step": 2900 }, { "epoch": 2.0192181260370035e-14, "grad_norm": 0.8666963577270508, "learning_rate": 0.00014997128073263114, "loss": 1.6394, "step": 2910 }, { "epoch": 2.0261570199409107e-14, "grad_norm": 0.7022987008094788, "learning_rate": 0.0001499706461178534, "loss": 1.6554, "step": 2920 }, { "epoch": 2.033095913844818e-14, "grad_norm": 0.9575149416923523, "learning_rate": 0.00014997000456941695, "loss": 1.7066, "step": 2930 }, { "epoch": 2.040034807748725e-14, "grad_norm": 0.9893184304237366, "learning_rate": 0.00014996935608738773, "loss": 1.6786, "step": 2940 }, { "epoch": 2.0469737016526324e-14, "grad_norm": 0.9398677349090576, "learning_rate": 0.0001499687006718324, "loss": 1.6864, "step": 2950 }, { "epoch": 2.0539125955565396e-14, "grad_norm": 0.8368314504623413, "learning_rate": 0.00014996803832281833, "loss": 1.6536, "step": 2960 }, { "epoch": 2.0608514894604468e-14, "grad_norm": 0.9867850542068481, "learning_rate": 0.00014996736904041355, "loss": 1.6837, "step": 2970 }, { "epoch": 2.067790383364354e-14, "grad_norm": 0.9460932612419128, "learning_rate": 0.00014996669282468686, "loss": 1.6964, "step": 2980 }, { "epoch": 2.0747292772682613e-14, "grad_norm": 0.9411274194717407, "learning_rate": 0.00014996600967570777, "loss": 1.7318, "step": 2990 }, { "epoch": 2.0816681711721685e-14, "grad_norm": 1.0623198747634888, "learning_rate": 0.00014996531959354643, "loss": 1.6389, "step": 3000 }, { "epoch": 2.0886070650760757e-14, "grad_norm": 0.9084365963935852, "learning_rate": 0.00014996462257827378, "loss": 1.7147, "step": 3010 }, { "epoch": 2.095545958979983e-14, "grad_norm": 1.0316091775894165, "learning_rate": 0.00014996391862996146, "loss": 1.7584, "step": 3020 }, { "epoch": 2.1024848528838902e-14, "grad_norm": 1.0804544687271118, "learning_rate": 0.00014996320774868184, "loss": 1.6755, "step": 3030 }, { "epoch": 2.1094237467877974e-14, "grad_norm": 1.0081864595413208, "learning_rate": 0.0001499624899345079, "loss": 1.6193, "step": 3040 }, { "epoch": 2.1163626406917047e-14, "grad_norm": 1.08277428150177, "learning_rate": 0.0001499617651875135, "loss": 1.6514, "step": 3050 }, { "epoch": 2.123301534595612e-14, "grad_norm": 0.9237635731697083, "learning_rate": 0.00014996103350777302, "loss": 1.6361, "step": 3060 }, { "epoch": 2.130240428499519e-14, "grad_norm": 1.0666544437408447, "learning_rate": 0.00014996029489536172, "loss": 1.5834, "step": 3070 }, { "epoch": 2.1371793224034263e-14, "grad_norm": 1.0933096408843994, "learning_rate": 0.00014995954935035548, "loss": 1.6437, "step": 3080 }, { "epoch": 2.1441182163073336e-14, "grad_norm": 0.9595020413398743, "learning_rate": 0.00014995879687283094, "loss": 1.6638, "step": 3090 }, { "epoch": 2.1510571102112408e-14, "grad_norm": 1.0795753002166748, "learning_rate": 0.00014995803746286536, "loss": 1.6177, "step": 3100 }, { "epoch": 2.157996004115148e-14, "grad_norm": 1.0109508037567139, "learning_rate": 0.00014995727112053687, "loss": 1.6189, "step": 3110 }, { "epoch": 2.1649348980190553e-14, "grad_norm": 1.050767421722412, "learning_rate": 0.00014995649784592418, "loss": 1.6263, "step": 3120 }, { "epoch": 2.1718737919229625e-14, "grad_norm": 1.0827438831329346, "learning_rate": 0.00014995571763910676, "loss": 1.6731, "step": 3130 }, { "epoch": 2.1788126858268697e-14, "grad_norm": 0.9246954917907715, "learning_rate": 0.00014995493050016476, "loss": 1.68, "step": 3140 }, { "epoch": 2.185751579730777e-14, "grad_norm": 0.8015041351318359, "learning_rate": 0.00014995413642917915, "loss": 1.6386, "step": 3150 }, { "epoch": 2.1926904736346842e-14, "grad_norm": 1.0325864553451538, "learning_rate": 0.00014995333542623145, "loss": 1.6512, "step": 3160 }, { "epoch": 2.1996293675385914e-14, "grad_norm": 0.9788027405738831, "learning_rate": 0.000149952527491404, "loss": 1.6611, "step": 3170 }, { "epoch": 2.2065682614424986e-14, "grad_norm": 0.8213567137718201, "learning_rate": 0.00014995171262477983, "loss": 1.7154, "step": 3180 }, { "epoch": 2.213507155346406e-14, "grad_norm": 0.8066673874855042, "learning_rate": 0.0001499508908264427, "loss": 1.5996, "step": 3190 }, { "epoch": 2.220446049250313e-14, "grad_norm": 1.1526113748550415, "learning_rate": 0.00014995006209647706, "loss": 1.6239, "step": 3200 }, { "epoch": 2.2273849431542203e-14, "grad_norm": 0.9288071990013123, "learning_rate": 0.00014994922643496805, "loss": 1.6154, "step": 3210 }, { "epoch": 2.2343238370581275e-14, "grad_norm": 0.9600983262062073, "learning_rate": 0.00014994838384200154, "loss": 1.5574, "step": 3220 }, { "epoch": 2.2412627309620348e-14, "grad_norm": 1.1947942972183228, "learning_rate": 0.00014994753431766416, "loss": 1.5837, "step": 3230 }, { "epoch": 2.248201624865942e-14, "grad_norm": 1.1357449293136597, "learning_rate": 0.0001499466778620432, "loss": 1.6083, "step": 3240 }, { "epoch": 2.2551405187698492e-14, "grad_norm": 0.9186743497848511, "learning_rate": 0.00014994581447522667, "loss": 1.608, "step": 3250 }, { "epoch": 2.2620794126737565e-14, "grad_norm": 1.2345136404037476, "learning_rate": 0.00014994494415730328, "loss": 1.583, "step": 3260 }, { "epoch": 2.2690183065776637e-14, "grad_norm": 1.010189414024353, "learning_rate": 0.00014994406690836248, "loss": 1.6225, "step": 3270 }, { "epoch": 2.275957200481571e-14, "grad_norm": 1.076006293296814, "learning_rate": 0.00014994318272849443, "loss": 1.5718, "step": 3280 }, { "epoch": 2.282896094385478e-14, "grad_norm": 0.9936869740486145, "learning_rate": 0.00014994229161778994, "loss": 1.6281, "step": 3290 }, { "epoch": 2.2898349882893854e-14, "grad_norm": 1.0155872106552124, "learning_rate": 0.0001499413935763407, "loss": 1.6749, "step": 3300 }, { "epoch": 2.2967738821932926e-14, "grad_norm": 1.0020672082901, "learning_rate": 0.00014994048860423887, "loss": 1.5589, "step": 3310 }, { "epoch": 2.3037127760971998e-14, "grad_norm": 0.9915865659713745, "learning_rate": 0.00014993957670157755, "loss": 1.678, "step": 3320 }, { "epoch": 2.310651670001107e-14, "grad_norm": 1.123186469078064, "learning_rate": 0.00014993865786845038, "loss": 1.6475, "step": 3330 }, { "epoch": 2.3175905639050143e-14, "grad_norm": 1.0350154638290405, "learning_rate": 0.00014993773210495185, "loss": 1.6204, "step": 3340 }, { "epoch": 2.3245294578089215e-14, "grad_norm": 0.975878894329071, "learning_rate": 0.00014993679941117702, "loss": 1.5491, "step": 3350 }, { "epoch": 2.3314683517128287e-14, "grad_norm": 1.3609933853149414, "learning_rate": 0.0001499358597872218, "loss": 1.5516, "step": 3360 }, { "epoch": 2.338407245616736e-14, "grad_norm": 1.030429482460022, "learning_rate": 0.00014993491323318274, "loss": 1.5621, "step": 3370 }, { "epoch": 2.3453461395206432e-14, "grad_norm": 1.16990327835083, "learning_rate": 0.0001499339597491571, "loss": 1.5365, "step": 3380 }, { "epoch": 2.3522850334245504e-14, "grad_norm": 1.1071399450302124, "learning_rate": 0.00014993299933524284, "loss": 1.5405, "step": 3390 }, { "epoch": 2.3592239273284576e-14, "grad_norm": 1.1476824283599854, "learning_rate": 0.00014993203199153873, "loss": 1.6512, "step": 3400 }, { "epoch": 2.366162821232365e-14, "grad_norm": 1.1337790489196777, "learning_rate": 0.0001499310577181441, "loss": 1.526, "step": 3410 }, { "epoch": 2.373101715136272e-14, "grad_norm": 0.9245176911354065, "learning_rate": 0.00014993007651515914, "loss": 1.5558, "step": 3420 }, { "epoch": 2.3800406090401793e-14, "grad_norm": 0.9351624846458435, "learning_rate": 0.00014992908838268464, "loss": 1.5721, "step": 3430 }, { "epoch": 2.3869795029440866e-14, "grad_norm": 1.0754315853118896, "learning_rate": 0.00014992809332082217, "loss": 1.5758, "step": 3440 }, { "epoch": 2.3939183968479938e-14, "grad_norm": 0.9710455536842346, "learning_rate": 0.00014992709132967394, "loss": 1.5811, "step": 3450 }, { "epoch": 2.400857290751901e-14, "grad_norm": 1.1485459804534912, "learning_rate": 0.00014992608240934298, "loss": 1.6396, "step": 3460 }, { "epoch": 2.4077961846558082e-14, "grad_norm": 1.0477937459945679, "learning_rate": 0.00014992506655993297, "loss": 1.5501, "step": 3470 }, { "epoch": 2.4147350785597155e-14, "grad_norm": 1.091619849205017, "learning_rate": 0.00014992404378154825, "loss": 1.5967, "step": 3480 }, { "epoch": 2.4216739724636227e-14, "grad_norm": 0.9429634809494019, "learning_rate": 0.00014992301407429397, "loss": 1.6431, "step": 3490 }, { "epoch": 2.42861286636753e-14, "grad_norm": 1.0868749618530273, "learning_rate": 0.00014992197743827592, "loss": 1.5626, "step": 3500 }, { "epoch": 2.4355517602714372e-14, "grad_norm": 1.0939804315567017, "learning_rate": 0.00014992093387360068, "loss": 1.5255, "step": 3510 }, { "epoch": 2.4424906541753444e-14, "grad_norm": 1.0352535247802734, "learning_rate": 0.00014991988338037545, "loss": 1.5343, "step": 3520 }, { "epoch": 2.4494295480792516e-14, "grad_norm": 1.0779110193252563, "learning_rate": 0.0001499188259587082, "loss": 1.5297, "step": 3530 }, { "epoch": 2.456368441983159e-14, "grad_norm": 1.2223507165908813, "learning_rate": 0.0001499177616087076, "loss": 1.4651, "step": 3540 }, { "epoch": 2.463307335887066e-14, "grad_norm": 1.0099384784698486, "learning_rate": 0.000149916690330483, "loss": 1.5429, "step": 3550 }, { "epoch": 2.4702462297909733e-14, "grad_norm": 1.2090606689453125, "learning_rate": 0.0001499156121241445, "loss": 1.5627, "step": 3560 }, { "epoch": 2.4771851236948805e-14, "grad_norm": 1.158146619796753, "learning_rate": 0.00014991452698980294, "loss": 1.4904, "step": 3570 }, { "epoch": 2.4841240175987878e-14, "grad_norm": 0.9170963168144226, "learning_rate": 0.00014991343492756982, "loss": 1.5129, "step": 3580 }, { "epoch": 2.491062911502695e-14, "grad_norm": 0.9876478314399719, "learning_rate": 0.00014991233593755732, "loss": 1.5264, "step": 3590 }, { "epoch": 2.4980018054066022e-14, "grad_norm": 1.0892444849014282, "learning_rate": 0.00014991123001987844, "loss": 1.5573, "step": 3600 }, { "epoch": 2.5049406993105094e-14, "grad_norm": 1.1070212125778198, "learning_rate": 0.00014991011717464676, "loss": 1.5853, "step": 3610 }, { "epoch": 2.5118795932144167e-14, "grad_norm": 0.9931412935256958, "learning_rate": 0.00014990899740197674, "loss": 1.5024, "step": 3620 }, { "epoch": 2.518818487118324e-14, "grad_norm": 1.2729668617248535, "learning_rate": 0.00014990787070198337, "loss": 1.5807, "step": 3630 }, { "epoch": 2.525757381022231e-14, "grad_norm": 1.370579481124878, "learning_rate": 0.00014990673707478247, "loss": 1.575, "step": 3640 }, { "epoch": 2.5326962749261384e-14, "grad_norm": 1.1172288656234741, "learning_rate": 0.00014990559652049054, "loss": 1.593, "step": 3650 }, { "epoch": 2.5396351688300456e-14, "grad_norm": 1.480516791343689, "learning_rate": 0.00014990444903922477, "loss": 1.5179, "step": 3660 }, { "epoch": 2.5465740627339528e-14, "grad_norm": 1.069406509399414, "learning_rate": 0.00014990329463110308, "loss": 1.4535, "step": 3670 }, { "epoch": 2.55351295663786e-14, "grad_norm": 0.8658589720726013, "learning_rate": 0.00014990213329624415, "loss": 1.5129, "step": 3680 }, { "epoch": 2.5604518505417673e-14, "grad_norm": 1.0906540155410767, "learning_rate": 0.00014990096503476728, "loss": 1.4812, "step": 3690 }, { "epoch": 2.5673907444456745e-14, "grad_norm": 1.0991297960281372, "learning_rate": 0.00014989978984679254, "loss": 1.5069, "step": 3700 }, { "epoch": 2.5743296383495817e-14, "grad_norm": 1.128038763999939, "learning_rate": 0.0001498986077324407, "loss": 1.513, "step": 3710 }, { "epoch": 2.581268532253489e-14, "grad_norm": 1.3145102262496948, "learning_rate": 0.00014989741869183323, "loss": 1.497, "step": 3720 }, { "epoch": 2.5882074261573962e-14, "grad_norm": 1.1474285125732422, "learning_rate": 0.00014989622272509233, "loss": 1.5001, "step": 3730 }, { "epoch": 2.5951463200613034e-14, "grad_norm": 0.9948394894599915, "learning_rate": 0.00014989501983234094, "loss": 1.4887, "step": 3740 }, { "epoch": 2.6020852139652106e-14, "grad_norm": 1.1609777212142944, "learning_rate": 0.0001498938100137026, "loss": 1.4825, "step": 3750 }, { "epoch": 2.609024107869118e-14, "grad_norm": 1.100608229637146, "learning_rate": 0.0001498925932693017, "loss": 1.5232, "step": 3760 }, { "epoch": 2.615963001773025e-14, "grad_norm": 1.145509958267212, "learning_rate": 0.00014989136959926326, "loss": 1.5588, "step": 3770 }, { "epoch": 2.6229018956769323e-14, "grad_norm": 1.0900394916534424, "learning_rate": 0.00014989013900371302, "loss": 1.4584, "step": 3780 }, { "epoch": 2.6298407895808396e-14, "grad_norm": 1.04021418094635, "learning_rate": 0.00014988890148277746, "loss": 1.5754, "step": 3790 }, { "epoch": 2.6367796834847468e-14, "grad_norm": 1.0435367822647095, "learning_rate": 0.00014988765703658375, "loss": 1.582, "step": 3800 }, { "epoch": 2.643718577388654e-14, "grad_norm": 1.0292794704437256, "learning_rate": 0.00014988640566525977, "loss": 1.4863, "step": 3810 }, { "epoch": 2.6506574712925612e-14, "grad_norm": 1.4019991159439087, "learning_rate": 0.00014988514736893413, "loss": 1.4623, "step": 3820 }, { "epoch": 2.6575963651964685e-14, "grad_norm": 1.7095189094543457, "learning_rate": 0.0001498838821477361, "loss": 1.498, "step": 3830 }, { "epoch": 2.6645352591003757e-14, "grad_norm": 1.307321310043335, "learning_rate": 0.00014988261000179577, "loss": 1.4416, "step": 3840 }, { "epoch": 2.671474153004283e-14, "grad_norm": 1.074109435081482, "learning_rate": 0.00014988133093124384, "loss": 1.4244, "step": 3850 }, { "epoch": 2.67841304690819e-14, "grad_norm": 1.2925368547439575, "learning_rate": 0.0001498800449362117, "loss": 1.4582, "step": 3860 }, { "epoch": 2.6853519408120974e-14, "grad_norm": 1.1075389385223389, "learning_rate": 0.00014987875201683158, "loss": 1.5429, "step": 3870 }, { "epoch": 2.6922908347160046e-14, "grad_norm": 1.1025605201721191, "learning_rate": 0.00014987745217323633, "loss": 1.4254, "step": 3880 }, { "epoch": 2.6992297286199118e-14, "grad_norm": 1.0848637819290161, "learning_rate": 0.00014987614540555954, "loss": 1.4515, "step": 3890 }, { "epoch": 2.706168622523819e-14, "grad_norm": 1.0332951545715332, "learning_rate": 0.00014987483171393544, "loss": 1.4737, "step": 3900 }, { "epoch": 2.7131075164277263e-14, "grad_norm": 1.2650678157806396, "learning_rate": 0.0001498735110984991, "loss": 1.5034, "step": 3910 }, { "epoch": 2.7200464103316335e-14, "grad_norm": 1.1344175338745117, "learning_rate": 0.00014987218355938623, "loss": 1.5035, "step": 3920 }, { "epoch": 2.7269853042355408e-14, "grad_norm": 1.012808084487915, "learning_rate": 0.00014987084909673322, "loss": 1.4762, "step": 3930 }, { "epoch": 2.733924198139448e-14, "grad_norm": 1.1148784160614014, "learning_rate": 0.0001498695077106772, "loss": 1.4602, "step": 3940 }, { "epoch": 2.7408630920433552e-14, "grad_norm": 1.222556710243225, "learning_rate": 0.00014986815940135604, "loss": 1.5195, "step": 3950 }, { "epoch": 2.7478019859472624e-14, "grad_norm": 1.0864182710647583, "learning_rate": 0.00014986680416890832, "loss": 1.5485, "step": 3960 }, { "epoch": 2.7547408798511697e-14, "grad_norm": 1.0785478353500366, "learning_rate": 0.00014986544201347327, "loss": 1.4517, "step": 3970 }, { "epoch": 2.761679773755077e-14, "grad_norm": 1.0276103019714355, "learning_rate": 0.00014986407293519091, "loss": 1.4244, "step": 3980 }, { "epoch": 2.768618667658984e-14, "grad_norm": 1.2541874647140503, "learning_rate": 0.00014986269693420193, "loss": 1.488, "step": 3990 }, { "epoch": 2.7755575615628914e-14, "grad_norm": 1.2097420692443848, "learning_rate": 0.0001498613140106477, "loss": 1.4043, "step": 4000 }, { "epoch": 2.7824964554667986e-14, "grad_norm": 1.3962198495864868, "learning_rate": 0.00014985992416467034, "loss": 1.3912, "step": 4010 }, { "epoch": 2.7894353493707058e-14, "grad_norm": 1.0829370021820068, "learning_rate": 0.00014985852739641273, "loss": 1.4416, "step": 4020 }, { "epoch": 2.796374243274613e-14, "grad_norm": 1.1242831945419312, "learning_rate": 0.00014985712370601837, "loss": 1.4709, "step": 4030 }, { "epoch": 2.8033131371785203e-14, "grad_norm": 1.1386492252349854, "learning_rate": 0.00014985571309363154, "loss": 1.409, "step": 4040 }, { "epoch": 2.8102520310824275e-14, "grad_norm": 1.4364837408065796, "learning_rate": 0.00014985429555939713, "loss": 1.4791, "step": 4050 }, { "epoch": 2.8171909249863347e-14, "grad_norm": 1.1776186227798462, "learning_rate": 0.0001498528711034609, "loss": 1.4059, "step": 4060 }, { "epoch": 2.824129818890242e-14, "grad_norm": 1.3247079849243164, "learning_rate": 0.0001498514397259692, "loss": 1.4792, "step": 4070 }, { "epoch": 2.8310687127941492e-14, "grad_norm": 1.3631489276885986, "learning_rate": 0.0001498500014270691, "loss": 1.5081, "step": 4080 }, { "epoch": 2.8380076066980564e-14, "grad_norm": 1.235464334487915, "learning_rate": 0.00014984855620690847, "loss": 1.3883, "step": 4090 }, { "epoch": 2.8449465006019636e-14, "grad_norm": 1.2256449460983276, "learning_rate": 0.00014984710406563576, "loss": 1.5015, "step": 4100 }, { "epoch": 2.851885394505871e-14, "grad_norm": 1.4692617654800415, "learning_rate": 0.00014984564500340029, "loss": 1.4671, "step": 4110 }, { "epoch": 2.858824288409778e-14, "grad_norm": 1.2475314140319824, "learning_rate": 0.00014984417902035187, "loss": 1.4661, "step": 4120 }, { "epoch": 2.8657631823136853e-14, "grad_norm": 1.0781768560409546, "learning_rate": 0.00014984270611664125, "loss": 1.4219, "step": 4130 }, { "epoch": 2.8727020762175925e-14, "grad_norm": 1.120155692100525, "learning_rate": 0.00014984122629241978, "loss": 1.355, "step": 4140 }, { "epoch": 2.8796409701215e-14, "grad_norm": 1.3112698793411255, "learning_rate": 0.00014983973954783953, "loss": 1.4491, "step": 4150 }, { "epoch": 2.886579864025407e-14, "grad_norm": 1.3356902599334717, "learning_rate": 0.00014983824588305327, "loss": 1.3761, "step": 4160 }, { "epoch": 2.893518757929314e-14, "grad_norm": 1.046492099761963, "learning_rate": 0.00014983674529821454, "loss": 1.4041, "step": 4170 }, { "epoch": 2.9004576518332215e-14, "grad_norm": 1.3322831392288208, "learning_rate": 0.00014983523779347747, "loss": 1.4473, "step": 4180 }, { "epoch": 2.907396545737129e-14, "grad_norm": 1.3671907186508179, "learning_rate": 0.00014983372336899704, "loss": 1.4213, "step": 4190 }, { "epoch": 2.914335439641036e-14, "grad_norm": 1.5634770393371582, "learning_rate": 0.00014983220202492887, "loss": 1.4426, "step": 4200 }, { "epoch": 2.921274333544943e-14, "grad_norm": 1.157705307006836, "learning_rate": 0.0001498306737614293, "loss": 1.3943, "step": 4210 }, { "epoch": 2.9282132274488504e-14, "grad_norm": 1.3161952495574951, "learning_rate": 0.0001498291385786554, "loss": 1.4186, "step": 4220 }, { "epoch": 2.9351521213527576e-14, "grad_norm": 1.3655389547348022, "learning_rate": 0.0001498275964767649, "loss": 1.432, "step": 4230 }, { "epoch": 2.942091015256665e-14, "grad_norm": 1.314526081085205, "learning_rate": 0.0001498260474559163, "loss": 1.4586, "step": 4240 }, { "epoch": 2.949029909160572e-14, "grad_norm": 1.6086381673812866, "learning_rate": 0.00014982449151626875, "loss": 1.3672, "step": 4250 }, { "epoch": 2.955968803064479e-14, "grad_norm": 1.107791543006897, "learning_rate": 0.0001498229286579822, "loss": 1.448, "step": 4260 }, { "epoch": 2.9629076969683865e-14, "grad_norm": 1.0939781665802002, "learning_rate": 0.0001498213588812172, "loss": 1.4988, "step": 4270 }, { "epoch": 2.969846590872294e-14, "grad_norm": 1.4428602457046509, "learning_rate": 0.00014981978218613513, "loss": 1.4093, "step": 4280 }, { "epoch": 2.976785484776201e-14, "grad_norm": 1.5083874464035034, "learning_rate": 0.00014981819857289797, "loss": 1.3623, "step": 4290 }, { "epoch": 2.983724378680108e-14, "grad_norm": 2.09505033493042, "learning_rate": 0.00014981660804166852, "loss": 1.3952, "step": 4300 }, { "epoch": 2.9906632725840154e-14, "grad_norm": 1.2561094760894775, "learning_rate": 0.00014981501059261018, "loss": 1.3869, "step": 4310 }, { "epoch": 2.9976021664879227e-14, "grad_norm": 1.2011432647705078, "learning_rate": 0.0001498134062258871, "loss": 1.3383, "step": 4320 }, { "epoch": 3.00454106039183e-14, "grad_norm": 1.4155274629592896, "learning_rate": 0.0001498117949416642, "loss": 1.3998, "step": 4330 }, { "epoch": 3.011479954295737e-14, "grad_norm": 1.278083324432373, "learning_rate": 0.00014981017674010704, "loss": 1.4154, "step": 4340 }, { "epoch": 3.0184188481996443e-14, "grad_norm": 1.4111058712005615, "learning_rate": 0.00014980855162138193, "loss": 1.372, "step": 4350 }, { "epoch": 3.0253577421035516e-14, "grad_norm": 1.2687475681304932, "learning_rate": 0.00014980691958565585, "loss": 1.3914, "step": 4360 }, { "epoch": 3.032296636007459e-14, "grad_norm": 1.3966423273086548, "learning_rate": 0.00014980528063309655, "loss": 1.3767, "step": 4370 }, { "epoch": 3.039235529911366e-14, "grad_norm": 1.2208062410354614, "learning_rate": 0.0001498036347638724, "loss": 1.4206, "step": 4380 }, { "epoch": 3.046174423815273e-14, "grad_norm": 1.2229195833206177, "learning_rate": 0.00014980198197815263, "loss": 1.4256, "step": 4390 }, { "epoch": 3.0531133177191805e-14, "grad_norm": 1.2402780055999756, "learning_rate": 0.00014980032227610701, "loss": 1.3712, "step": 4400 }, { "epoch": 3.060052211623088e-14, "grad_norm": 1.269439935684204, "learning_rate": 0.00014979865565790615, "loss": 1.3914, "step": 4410 }, { "epoch": 3.066991105526995e-14, "grad_norm": 1.0517586469650269, "learning_rate": 0.00014979698212372127, "loss": 1.4053, "step": 4420 }, { "epoch": 3.073929999430902e-14, "grad_norm": 1.0129338502883911, "learning_rate": 0.0001497953016737244, "loss": 1.4515, "step": 4430 }, { "epoch": 3.0808688933348094e-14, "grad_norm": 1.0878732204437256, "learning_rate": 0.0001497936143080882, "loss": 1.3483, "step": 4440 }, { "epoch": 3.0878077872387166e-14, "grad_norm": 1.296628475189209, "learning_rate": 0.00014979192002698613, "loss": 1.3788, "step": 4450 }, { "epoch": 3.094746681142624e-14, "grad_norm": 1.1006441116333008, "learning_rate": 0.00014979021883059222, "loss": 1.3804, "step": 4460 }, { "epoch": 3.101685575046531e-14, "grad_norm": 1.0906528234481812, "learning_rate": 0.00014978851071908135, "loss": 1.3136, "step": 4470 }, { "epoch": 3.108624468950438e-14, "grad_norm": 1.1550437211990356, "learning_rate": 0.00014978679569262906, "loss": 1.346, "step": 4480 }, { "epoch": 3.1155633628543455e-14, "grad_norm": 1.3836573362350464, "learning_rate": 0.00014978507375141154, "loss": 1.3657, "step": 4490 }, { "epoch": 3.122502256758253e-14, "grad_norm": 1.118142008781433, "learning_rate": 0.0001497833448956058, "loss": 1.3737, "step": 4500 }, { "epoch": 3.12944115066216e-14, "grad_norm": 1.4167520999908447, "learning_rate": 0.00014978160912538949, "loss": 1.3568, "step": 4510 }, { "epoch": 3.136380044566067e-14, "grad_norm": 1.3260984420776367, "learning_rate": 0.00014977986644094097, "loss": 1.3775, "step": 4520 }, { "epoch": 3.1433189384699745e-14, "grad_norm": 1.4747165441513062, "learning_rate": 0.00014977811684243936, "loss": 1.3375, "step": 4530 }, { "epoch": 3.1502578323738817e-14, "grad_norm": 1.3495501279830933, "learning_rate": 0.00014977636033006444, "loss": 1.3827, "step": 4540 }, { "epoch": 3.157196726277789e-14, "grad_norm": 1.3271843194961548, "learning_rate": 0.00014977459690399672, "loss": 1.4258, "step": 4550 }, { "epoch": 3.164135620181696e-14, "grad_norm": 1.3226573467254639, "learning_rate": 0.00014977282656441742, "loss": 1.3079, "step": 4560 }, { "epoch": 3.1710745140856034e-14, "grad_norm": 1.3186389207839966, "learning_rate": 0.00014977104931150848, "loss": 1.4311, "step": 4570 }, { "epoch": 3.1780134079895106e-14, "grad_norm": 1.4615586996078491, "learning_rate": 0.00014976926514545253, "loss": 1.4033, "step": 4580 }, { "epoch": 3.184952301893418e-14, "grad_norm": 1.2029820680618286, "learning_rate": 0.00014976747406643292, "loss": 1.3639, "step": 4590 }, { "epoch": 3.191891195797325e-14, "grad_norm": 1.270350694656372, "learning_rate": 0.00014976567607463372, "loss": 1.3118, "step": 4600 }, { "epoch": 3.198830089701232e-14, "grad_norm": 1.6770591735839844, "learning_rate": 0.00014976387117023968, "loss": 1.3356, "step": 4610 }, { "epoch": 3.2057689836051395e-14, "grad_norm": 1.3606736660003662, "learning_rate": 0.00014976205935343632, "loss": 1.3342, "step": 4620 }, { "epoch": 3.212707877509047e-14, "grad_norm": 1.3609522581100464, "learning_rate": 0.00014976024062440978, "loss": 1.3099, "step": 4630 }, { "epoch": 3.219646771412954e-14, "grad_norm": 1.1263738870620728, "learning_rate": 0.00014975841498334703, "loss": 1.3105, "step": 4640 }, { "epoch": 3.226585665316861e-14, "grad_norm": 1.4245606660842896, "learning_rate": 0.0001497565824304356, "loss": 1.4146, "step": 4650 }, { "epoch": 3.2335245592207684e-14, "grad_norm": 1.199484944343567, "learning_rate": 0.0001497547429658639, "loss": 1.2908, "step": 4660 }, { "epoch": 3.2404634531246757e-14, "grad_norm": 1.1532188653945923, "learning_rate": 0.0001497528965898209, "loss": 1.3228, "step": 4670 }, { "epoch": 3.247402347028583e-14, "grad_norm": 1.349195957183838, "learning_rate": 0.00014975104330249638, "loss": 1.3365, "step": 4680 }, { "epoch": 3.25434124093249e-14, "grad_norm": 1.24483060836792, "learning_rate": 0.00014974918310408075, "loss": 1.3471, "step": 4690 }, { "epoch": 3.2612801348363973e-14, "grad_norm": 1.2634373903274536, "learning_rate": 0.00014974731599476525, "loss": 1.34, "step": 4700 }, { "epoch": 3.2682190287403046e-14, "grad_norm": 1.2425416707992554, "learning_rate": 0.00014974544197474167, "loss": 1.3959, "step": 4710 }, { "epoch": 3.275157922644212e-14, "grad_norm": 1.3373216390609741, "learning_rate": 0.00014974356104420267, "loss": 1.3008, "step": 4720 }, { "epoch": 3.282096816548119e-14, "grad_norm": 1.3088589906692505, "learning_rate": 0.0001497416732033415, "loss": 1.3612, "step": 4730 }, { "epoch": 3.289035710452026e-14, "grad_norm": 1.2757568359375, "learning_rate": 0.00014973977845235216, "loss": 1.4041, "step": 4740 }, { "epoch": 3.2959746043559335e-14, "grad_norm": 1.3219661712646484, "learning_rate": 0.00014973787679142938, "loss": 1.3239, "step": 4750 }, { "epoch": 3.302913498259841e-14, "grad_norm": 1.465116262435913, "learning_rate": 0.0001497359682207686, "loss": 1.3036, "step": 4760 }, { "epoch": 3.309852392163748e-14, "grad_norm": 1.3339122533798218, "learning_rate": 0.00014973405274056593, "loss": 1.3282, "step": 4770 }, { "epoch": 3.316791286067655e-14, "grad_norm": 1.3699498176574707, "learning_rate": 0.0001497321303510182, "loss": 1.2999, "step": 4780 }, { "epoch": 3.3237301799715624e-14, "grad_norm": 1.6772847175598145, "learning_rate": 0.00014973020105232303, "loss": 1.2498, "step": 4790 }, { "epoch": 3.3306690738754696e-14, "grad_norm": 1.3006882667541504, "learning_rate": 0.00014972826484467862, "loss": 1.3186, "step": 4800 }, { "epoch": 3.337607967779377e-14, "grad_norm": 1.4262330532073975, "learning_rate": 0.000149726321728284, "loss": 1.3277, "step": 4810 }, { "epoch": 3.344546861683284e-14, "grad_norm": 1.238342046737671, "learning_rate": 0.0001497243717033388, "loss": 1.2678, "step": 4820 }, { "epoch": 3.351485755587191e-14, "grad_norm": 1.2364537715911865, "learning_rate": 0.00014972241477004345, "loss": 1.2957, "step": 4830 }, { "epoch": 3.3584246494910985e-14, "grad_norm": 1.367565631866455, "learning_rate": 0.00014972045092859904, "loss": 1.3024, "step": 4840 }, { "epoch": 3.365363543395006e-14, "grad_norm": 1.3155871629714966, "learning_rate": 0.0001497184801792074, "loss": 1.332, "step": 4850 }, { "epoch": 3.372302437298913e-14, "grad_norm": 1.387068271636963, "learning_rate": 0.00014971650252207106, "loss": 1.3555, "step": 4860 }, { "epoch": 3.37924133120282e-14, "grad_norm": 1.3169015645980835, "learning_rate": 0.0001497145179573932, "loss": 1.2655, "step": 4870 }, { "epoch": 3.3861802251067274e-14, "grad_norm": 1.4567800760269165, "learning_rate": 0.00014971252648537787, "loss": 1.3429, "step": 4880 }, { "epoch": 3.3931191190106347e-14, "grad_norm": 1.6149765253067017, "learning_rate": 0.00014971052810622962, "loss": 1.34, "step": 4890 }, { "epoch": 3.400058012914542e-14, "grad_norm": 1.420358419418335, "learning_rate": 0.00014970852282015387, "loss": 1.3577, "step": 4900 }, { "epoch": 3.406996906818449e-14, "grad_norm": 1.7365856170654297, "learning_rate": 0.0001497065106273567, "loss": 1.2925, "step": 4910 }, { "epoch": 3.4139358007223564e-14, "grad_norm": 1.26491117477417, "learning_rate": 0.00014970449152804484, "loss": 1.248, "step": 4920 }, { "epoch": 3.4208746946262636e-14, "grad_norm": 1.130911946296692, "learning_rate": 0.00014970246552242585, "loss": 1.3019, "step": 4930 }, { "epoch": 3.427813588530171e-14, "grad_norm": 1.3927314281463623, "learning_rate": 0.0001497004326107079, "loss": 1.2698, "step": 4940 }, { "epoch": 3.434752482434078e-14, "grad_norm": 1.3523198366165161, "learning_rate": 0.0001496983927930999, "loss": 1.2948, "step": 4950 }, { "epoch": 3.441691376337985e-14, "grad_norm": 1.4599957466125488, "learning_rate": 0.00014969634606981148, "loss": 1.2837, "step": 4960 }, { "epoch": 3.4486302702418925e-14, "grad_norm": 1.436250925064087, "learning_rate": 0.00014969429244105295, "loss": 1.2676, "step": 4970 }, { "epoch": 3.4555691641458e-14, "grad_norm": 1.5768656730651855, "learning_rate": 0.0001496922319070354, "loss": 1.2857, "step": 4980 }, { "epoch": 3.462508058049707e-14, "grad_norm": 1.2750502824783325, "learning_rate": 0.00014969016446797056, "loss": 1.2789, "step": 4990 }, { "epoch": 3.469446951953614e-14, "grad_norm": 1.4451810121536255, "learning_rate": 0.00014968809012407087, "loss": 1.2725, "step": 5000 }, { "epoch": 3.4763858458575214e-14, "grad_norm": 1.3236846923828125, "learning_rate": 0.00014968600887554955, "loss": 1.2984, "step": 5010 }, { "epoch": 3.4833247397614286e-14, "grad_norm": 1.3108012676239014, "learning_rate": 0.00014968392072262042, "loss": 1.3375, "step": 5020 }, { "epoch": 3.490263633665336e-14, "grad_norm": 1.396248459815979, "learning_rate": 0.00014968182566549808, "loss": 1.2265, "step": 5030 }, { "epoch": 3.497202527569243e-14, "grad_norm": 1.2273263931274414, "learning_rate": 0.0001496797237043979, "loss": 1.3425, "step": 5040 }, { "epoch": 3.5041414214731503e-14, "grad_norm": 1.4335212707519531, "learning_rate": 0.00014967761483953578, "loss": 1.3505, "step": 5050 }, { "epoch": 3.5110803153770576e-14, "grad_norm": 1.276984453201294, "learning_rate": 0.00014967549907112855, "loss": 1.2621, "step": 5060 }, { "epoch": 3.518019209280965e-14, "grad_norm": 1.4695947170257568, "learning_rate": 0.00014967337639939358, "loss": 1.2541, "step": 5070 }, { "epoch": 3.524958103184872e-14, "grad_norm": 1.6610710620880127, "learning_rate": 0.00014967124682454898, "loss": 1.3036, "step": 5080 }, { "epoch": 3.531896997088779e-14, "grad_norm": 1.3429895639419556, "learning_rate": 0.00014966911034681367, "loss": 1.2411, "step": 5090 }, { "epoch": 3.5388358909926865e-14, "grad_norm": 1.2597368955612183, "learning_rate": 0.00014966696696640716, "loss": 1.2265, "step": 5100 }, { "epoch": 3.545774784896594e-14, "grad_norm": 1.3476396799087524, "learning_rate": 0.0001496648166835497, "loss": 1.244, "step": 5110 }, { "epoch": 3.552713678800501e-14, "grad_norm": 1.1684163808822632, "learning_rate": 0.0001496626594984623, "loss": 1.3025, "step": 5120 }, { "epoch": 3.559652572704408e-14, "grad_norm": 1.2176332473754883, "learning_rate": 0.00014966049541136663, "loss": 1.2108, "step": 5130 }, { "epoch": 3.5665914666083154e-14, "grad_norm": 1.3234387636184692, "learning_rate": 0.00014965832442248507, "loss": 1.2481, "step": 5140 }, { "epoch": 3.5735303605122226e-14, "grad_norm": 1.3182040452957153, "learning_rate": 0.00014965614653204078, "loss": 1.2737, "step": 5150 }, { "epoch": 3.58046925441613e-14, "grad_norm": 1.554351568222046, "learning_rate": 0.0001496539617402575, "loss": 1.2937, "step": 5160 }, { "epoch": 3.587408148320037e-14, "grad_norm": 1.638044834136963, "learning_rate": 0.00014965177004735977, "loss": 1.2844, "step": 5170 }, { "epoch": 3.594347042223944e-14, "grad_norm": 1.3348991870880127, "learning_rate": 0.00014964957145357287, "loss": 1.2647, "step": 5180 }, { "epoch": 3.6012859361278515e-14, "grad_norm": 1.450531005859375, "learning_rate": 0.0001496473659591227, "loss": 1.2445, "step": 5190 }, { "epoch": 3.608224830031759e-14, "grad_norm": 1.2430405616760254, "learning_rate": 0.00014964515356423585, "loss": 1.292, "step": 5200 }, { "epoch": 3.615163723935666e-14, "grad_norm": 1.4713777303695679, "learning_rate": 0.00014964293426913977, "loss": 1.3298, "step": 5210 }, { "epoch": 3.622102617839573e-14, "grad_norm": 1.3466085195541382, "learning_rate": 0.00014964070807406253, "loss": 1.248, "step": 5220 }, { "epoch": 3.6290415117434804e-14, "grad_norm": 1.325778603553772, "learning_rate": 0.00014963847497923284, "loss": 1.2334, "step": 5230 }, { "epoch": 3.6359804056473877e-14, "grad_norm": 1.4891599416732788, "learning_rate": 0.00014963623498488024, "loss": 1.3002, "step": 5240 }, { "epoch": 3.642919299551295e-14, "grad_norm": 1.238718032836914, "learning_rate": 0.00014963398809123487, "loss": 1.1991, "step": 5250 }, { "epoch": 3.649858193455202e-14, "grad_norm": 1.4766241312026978, "learning_rate": 0.0001496317342985277, "loss": 1.1897, "step": 5260 }, { "epoch": 3.6567970873591094e-14, "grad_norm": 1.146134376525879, "learning_rate": 0.00014962947360699027, "loss": 1.213, "step": 5270 }, { "epoch": 3.6637359812630166e-14, "grad_norm": 1.3172760009765625, "learning_rate": 0.00014962720601685494, "loss": 1.2459, "step": 5280 }, { "epoch": 3.670674875166924e-14, "grad_norm": 1.361673355102539, "learning_rate": 0.00014962493152835478, "loss": 1.2031, "step": 5290 }, { "epoch": 3.677613769070831e-14, "grad_norm": 1.560665249824524, "learning_rate": 0.00014962265014172344, "loss": 1.2811, "step": 5300 }, { "epoch": 3.684552662974738e-14, "grad_norm": 1.3961011171340942, "learning_rate": 0.00014962036185719546, "loss": 1.217, "step": 5310 }, { "epoch": 3.6914915568786455e-14, "grad_norm": 1.5635650157928467, "learning_rate": 0.00014961806667500592, "loss": 1.2675, "step": 5320 }, { "epoch": 3.698430450782553e-14, "grad_norm": 1.4974462985992432, "learning_rate": 0.00014961576459539074, "loss": 1.3032, "step": 5330 }, { "epoch": 3.70536934468646e-14, "grad_norm": 1.4079276323318481, "learning_rate": 0.00014961345561858644, "loss": 1.1835, "step": 5340 }, { "epoch": 3.712308238590367e-14, "grad_norm": 1.6846318244934082, "learning_rate": 0.00014961113974483036, "loss": 1.2979, "step": 5350 }, { "epoch": 3.7192471324942744e-14, "grad_norm": 1.5251315832138062, "learning_rate": 0.00014960881697436047, "loss": 1.2499, "step": 5360 }, { "epoch": 3.7261860263981816e-14, "grad_norm": 1.594505786895752, "learning_rate": 0.00014960648730741547, "loss": 1.2645, "step": 5370 }, { "epoch": 3.733124920302089e-14, "grad_norm": 1.5279814004898071, "learning_rate": 0.00014960415074423476, "loss": 1.2338, "step": 5380 }, { "epoch": 3.740063814205996e-14, "grad_norm": 1.442065715789795, "learning_rate": 0.00014960180728505846, "loss": 1.1812, "step": 5390 }, { "epoch": 3.7470027081099033e-14, "grad_norm": 1.515682578086853, "learning_rate": 0.0001495994569301274, "loss": 1.2566, "step": 5400 }, { "epoch": 3.7539416020138106e-14, "grad_norm": 1.4166944026947021, "learning_rate": 0.00014959709967968314, "loss": 1.1753, "step": 5410 }, { "epoch": 3.760880495917718e-14, "grad_norm": 1.131672978401184, "learning_rate": 0.0001495947355339679, "loss": 1.2021, "step": 5420 }, { "epoch": 3.767819389821625e-14, "grad_norm": 1.3473701477050781, "learning_rate": 0.00014959236449322462, "loss": 1.2271, "step": 5430 }, { "epoch": 3.774758283725532e-14, "grad_norm": 1.6018822193145752, "learning_rate": 0.00014958998655769697, "loss": 1.2082, "step": 5440 }, { "epoch": 3.7816971776294395e-14, "grad_norm": 1.6769176721572876, "learning_rate": 0.00014958760172762936, "loss": 1.2416, "step": 5450 }, { "epoch": 3.788636071533347e-14, "grad_norm": 1.5518276691436768, "learning_rate": 0.0001495852100032668, "loss": 1.2079, "step": 5460 }, { "epoch": 3.795574965437254e-14, "grad_norm": 1.4253578186035156, "learning_rate": 0.00014958281138485516, "loss": 1.2326, "step": 5470 }, { "epoch": 3.802513859341161e-14, "grad_norm": 1.665061116218567, "learning_rate": 0.00014958040587264082, "loss": 1.2228, "step": 5480 }, { "epoch": 3.8094527532450684e-14, "grad_norm": 1.3333241939544678, "learning_rate": 0.00014957799346687113, "loss": 1.2561, "step": 5490 }, { "epoch": 3.8163916471489756e-14, "grad_norm": 1.5276864767074585, "learning_rate": 0.00014957557416779387, "loss": 1.1687, "step": 5500 }, { "epoch": 3.823330541052883e-14, "grad_norm": 1.0931506156921387, "learning_rate": 0.00014957314797565773, "loss": 1.2413, "step": 5510 }, { "epoch": 3.83026943495679e-14, "grad_norm": 1.576334834098816, "learning_rate": 0.00014957071489071202, "loss": 1.2854, "step": 5520 }, { "epoch": 3.837208328860697e-14, "grad_norm": 1.684796690940857, "learning_rate": 0.00014956827491320677, "loss": 1.2255, "step": 5530 }, { "epoch": 3.8441472227646045e-14, "grad_norm": 1.2849870920181274, "learning_rate": 0.00014956582804339275, "loss": 1.1811, "step": 5540 }, { "epoch": 3.851086116668512e-14, "grad_norm": 1.5479934215545654, "learning_rate": 0.0001495633742815214, "loss": 1.2121, "step": 5550 }, { "epoch": 3.858025010572419e-14, "grad_norm": 1.394901156425476, "learning_rate": 0.00014956091362784487, "loss": 1.1843, "step": 5560 }, { "epoch": 3.864963904476326e-14, "grad_norm": 1.4731277227401733, "learning_rate": 0.00014955844608261604, "loss": 1.1414, "step": 5570 }, { "epoch": 3.8719027983802334e-14, "grad_norm": 1.5998958349227905, "learning_rate": 0.0001495559716460885, "loss": 1.2025, "step": 5580 }, { "epoch": 3.8788416922841407e-14, "grad_norm": 1.3202542066574097, "learning_rate": 0.00014955349031851655, "loss": 1.1954, "step": 5590 }, { "epoch": 3.885780586188048e-14, "grad_norm": 1.3600413799285889, "learning_rate": 0.00014955100210015517, "loss": 1.1712, "step": 5600 }, { "epoch": 3.892719480091955e-14, "grad_norm": 1.4922014474868774, "learning_rate": 0.00014954850699126, "loss": 1.1929, "step": 5610 }, { "epoch": 3.8996583739958623e-14, "grad_norm": 1.6093670129776, "learning_rate": 0.00014954600499208758, "loss": 1.1917, "step": 5620 }, { "epoch": 3.9065972678997696e-14, "grad_norm": 1.353619933128357, "learning_rate": 0.00014954349610289488, "loss": 1.2193, "step": 5630 }, { "epoch": 3.913536161803677e-14, "grad_norm": 1.3347852230072021, "learning_rate": 0.00014954098032393987, "loss": 1.2151, "step": 5640 }, { "epoch": 3.920475055707584e-14, "grad_norm": 1.2829476594924927, "learning_rate": 0.00014953845765548098, "loss": 1.167, "step": 5650 }, { "epoch": 3.927413949611491e-14, "grad_norm": 1.4588804244995117, "learning_rate": 0.00014953592809777752, "loss": 1.1882, "step": 5660 }, { "epoch": 3.9343528435153985e-14, "grad_norm": 1.4413843154907227, "learning_rate": 0.00014953339165108936, "loss": 1.2067, "step": 5670 }, { "epoch": 3.941291737419306e-14, "grad_norm": 1.1804817914962769, "learning_rate": 0.0001495308483156773, "loss": 1.2769, "step": 5680 }, { "epoch": 3.948230631323213e-14, "grad_norm": 1.272597074508667, "learning_rate": 0.00014952829809180255, "loss": 1.1812, "step": 5690 }, { "epoch": 3.95516952522712e-14, "grad_norm": 1.41383957862854, "learning_rate": 0.00014952574097972727, "loss": 1.1961, "step": 5700 }, { "epoch": 3.9621084191310274e-14, "grad_norm": 1.3033385276794434, "learning_rate": 0.00014952317697971426, "loss": 1.1964, "step": 5710 }, { "epoch": 3.9690473130349346e-14, "grad_norm": 1.3902952671051025, "learning_rate": 0.00014952060609202698, "loss": 1.1249, "step": 5720 }, { "epoch": 3.975986206938842e-14, "grad_norm": 1.7199945449829102, "learning_rate": 0.0001495180283169296, "loss": 1.1665, "step": 5730 }, { "epoch": 3.982925100842749e-14, "grad_norm": 1.3998773097991943, "learning_rate": 0.00014951544365468707, "loss": 1.1614, "step": 5740 }, { "epoch": 3.9898639947466563e-14, "grad_norm": 1.4846340417861938, "learning_rate": 0.00014951285210556498, "loss": 1.1653, "step": 5750 }, { "epoch": 3.9968028886505635e-14, "grad_norm": 1.7286142110824585, "learning_rate": 0.00014951025366982969, "loss": 1.1577, "step": 5760 }, { "epoch": 4.003741782554471e-14, "grad_norm": 1.4514340162277222, "learning_rate": 0.00014950764834774816, "loss": 1.2, "step": 5770 }, { "epoch": 4.010680676458378e-14, "grad_norm": 1.738775610923767, "learning_rate": 0.0001495050361395882, "loss": 1.1554, "step": 5780 }, { "epoch": 4.017619570362285e-14, "grad_norm": 1.3452740907669067, "learning_rate": 0.00014950241704561823, "loss": 1.1773, "step": 5790 }, { "epoch": 4.0245584642661925e-14, "grad_norm": 1.34990656375885, "learning_rate": 0.00014949979106610738, "loss": 1.2284, "step": 5800 }, { "epoch": 4.0314973581701e-14, "grad_norm": 1.5917000770568848, "learning_rate": 0.0001494971582013255, "loss": 1.1199, "step": 5810 }, { "epoch": 4.038436252074007e-14, "grad_norm": 1.465013027191162, "learning_rate": 0.00014949451845154324, "loss": 1.2354, "step": 5820 }, { "epoch": 4.045375145977914e-14, "grad_norm": 1.6721481084823608, "learning_rate": 0.0001494918718170318, "loss": 1.2192, "step": 5830 }, { "epoch": 4.0523140398818214e-14, "grad_norm": 1.544049859046936, "learning_rate": 0.00014948921829806318, "loss": 1.2002, "step": 5840 }, { "epoch": 4.0592529337857286e-14, "grad_norm": 1.5203306674957275, "learning_rate": 0.00014948655789491006, "loss": 1.1428, "step": 5850 }, { "epoch": 4.066191827689636e-14, "grad_norm": 1.6023669242858887, "learning_rate": 0.00014948389060784586, "loss": 1.1437, "step": 5860 }, { "epoch": 4.073130721593543e-14, "grad_norm": 1.5380996465682983, "learning_rate": 0.00014948121643714468, "loss": 1.1493, "step": 5870 }, { "epoch": 4.08006961549745e-14, "grad_norm": 1.4577490091323853, "learning_rate": 0.00014947853538308133, "loss": 1.1159, "step": 5880 }, { "epoch": 4.0870085094013575e-14, "grad_norm": 1.3631421327590942, "learning_rate": 0.00014947584744593134, "loss": 1.1279, "step": 5890 }, { "epoch": 4.093947403305265e-14, "grad_norm": 1.7626762390136719, "learning_rate": 0.0001494731526259709, "loss": 1.2096, "step": 5900 }, { "epoch": 4.100886297209172e-14, "grad_norm": 1.5428125858306885, "learning_rate": 0.000149470450923477, "loss": 1.1088, "step": 5910 }, { "epoch": 4.107825191113079e-14, "grad_norm": 1.195734977722168, "learning_rate": 0.00014946774233872724, "loss": 1.1463, "step": 5920 }, { "epoch": 4.1147640850169864e-14, "grad_norm": 1.2814549207687378, "learning_rate": 0.00014946502687199998, "loss": 1.1549, "step": 5930 }, { "epoch": 4.1217029789208937e-14, "grad_norm": 1.4383759498596191, "learning_rate": 0.00014946230452357427, "loss": 1.1614, "step": 5940 }, { "epoch": 4.128641872824801e-14, "grad_norm": 1.4905034303665161, "learning_rate": 0.0001494595752937299, "loss": 1.1422, "step": 5950 }, { "epoch": 4.135580766728708e-14, "grad_norm": 1.468567967414856, "learning_rate": 0.00014945683918274734, "loss": 1.2101, "step": 5960 }, { "epoch": 4.1425196606326153e-14, "grad_norm": 1.6669834852218628, "learning_rate": 0.00014945409619090773, "loss": 1.1341, "step": 5970 }, { "epoch": 4.1494585545365226e-14, "grad_norm": 1.5112745761871338, "learning_rate": 0.00014945134631849297, "loss": 1.1742, "step": 5980 }, { "epoch": 4.15639744844043e-14, "grad_norm": 1.551767349243164, "learning_rate": 0.00014944858956578565, "loss": 1.2251, "step": 5990 }, { "epoch": 4.163336342344337e-14, "grad_norm": 1.7461601495742798, "learning_rate": 0.0001494458259330691, "loss": 1.1685, "step": 6000 }, { "epoch": 4.170275236248244e-14, "grad_norm": 1.5083036422729492, "learning_rate": 0.00014944305542062731, "loss": 1.1251, "step": 6010 }, { "epoch": 4.1772141301521515e-14, "grad_norm": 1.5079352855682373, "learning_rate": 0.00014944027802874497, "loss": 1.1331, "step": 6020 }, { "epoch": 4.184153024056059e-14, "grad_norm": 1.6347131729125977, "learning_rate": 0.0001494374937577075, "loss": 1.1294, "step": 6030 }, { "epoch": 4.191091917959966e-14, "grad_norm": 1.7858411073684692, "learning_rate": 0.0001494347026078011, "loss": 1.0685, "step": 6040 }, { "epoch": 4.198030811863873e-14, "grad_norm": 1.703926920890808, "learning_rate": 0.0001494319045793125, "loss": 1.1426, "step": 6050 }, { "epoch": 4.2049697057677804e-14, "grad_norm": 1.4666630029678345, "learning_rate": 0.00014942909967252929, "loss": 1.1393, "step": 6060 }, { "epoch": 4.2119085996716876e-14, "grad_norm": 1.5640897750854492, "learning_rate": 0.00014942628788773975, "loss": 1.098, "step": 6070 }, { "epoch": 4.218847493575595e-14, "grad_norm": 1.3287544250488281, "learning_rate": 0.0001494234692252328, "loss": 1.1109, "step": 6080 }, { "epoch": 4.225786387479502e-14, "grad_norm": 1.400882601737976, "learning_rate": 0.00014942064368529806, "loss": 1.1181, "step": 6090 }, { "epoch": 4.232725281383409e-14, "grad_norm": 1.3692002296447754, "learning_rate": 0.000149417811268226, "loss": 1.1475, "step": 6100 }, { "epoch": 4.2396641752873165e-14, "grad_norm": 1.5027590990066528, "learning_rate": 0.0001494149719743076, "loss": 1.1749, "step": 6110 }, { "epoch": 4.246603069191224e-14, "grad_norm": 1.5449861288070679, "learning_rate": 0.0001494121258038347, "loss": 1.1066, "step": 6120 }, { "epoch": 4.253541963095131e-14, "grad_norm": 1.6701282262802124, "learning_rate": 0.00014940927275709974, "loss": 1.1838, "step": 6130 }, { "epoch": 4.260480856999038e-14, "grad_norm": 1.5783942937850952, "learning_rate": 0.00014940641283439594, "loss": 1.1504, "step": 6140 }, { "epoch": 4.2674197509029455e-14, "grad_norm": 1.7973560094833374, "learning_rate": 0.0001494035460360172, "loss": 1.1897, "step": 6150 }, { "epoch": 4.2743586448068527e-14, "grad_norm": 1.6395381689071655, "learning_rate": 0.00014940067236225818, "loss": 1.1364, "step": 6160 }, { "epoch": 4.28129753871076e-14, "grad_norm": 1.312443494796753, "learning_rate": 0.00014939779181341413, "loss": 1.058, "step": 6170 }, { "epoch": 4.288236432614667e-14, "grad_norm": 1.3466291427612305, "learning_rate": 0.00014939490438978104, "loss": 1.1184, "step": 6180 }, { "epoch": 4.2951753265185744e-14, "grad_norm": 1.4466900825500488, "learning_rate": 0.00014939201009165574, "loss": 1.1062, "step": 6190 }, { "epoch": 4.3021142204224816e-14, "grad_norm": 1.5658448934555054, "learning_rate": 0.0001493891089193356, "loss": 1.1287, "step": 6200 }, { "epoch": 4.309053114326389e-14, "grad_norm": 1.6477054357528687, "learning_rate": 0.00014938620087311875, "loss": 1.1095, "step": 6210 }, { "epoch": 4.315992008230296e-14, "grad_norm": 1.888909101486206, "learning_rate": 0.00014938328595330403, "loss": 1.0972, "step": 6220 }, { "epoch": 4.322930902134203e-14, "grad_norm": 1.6461572647094727, "learning_rate": 0.00014938036416019108, "loss": 1.1093, "step": 6230 }, { "epoch": 4.3298697960381105e-14, "grad_norm": 1.415482997894287, "learning_rate": 0.00014937743549408008, "loss": 1.093, "step": 6240 }, { "epoch": 4.336808689942018e-14, "grad_norm": 1.6068419218063354, "learning_rate": 0.000149374499955272, "loss": 1.095, "step": 6250 }, { "epoch": 4.343747583845925e-14, "grad_norm": 1.611606240272522, "learning_rate": 0.00014937155754406856, "loss": 1.1288, "step": 6260 }, { "epoch": 4.350686477749832e-14, "grad_norm": 1.671305775642395, "learning_rate": 0.00014936860826077208, "loss": 1.1721, "step": 6270 }, { "epoch": 4.3576253716537394e-14, "grad_norm": 1.606431245803833, "learning_rate": 0.00014936565210568566, "loss": 1.0859, "step": 6280 }, { "epoch": 4.3645642655576466e-14, "grad_norm": 1.2221152782440186, "learning_rate": 0.00014936268907911313, "loss": 1.1711, "step": 6290 }, { "epoch": 4.371503159461554e-14, "grad_norm": 1.4913781881332397, "learning_rate": 0.00014935971918135895, "loss": 1.1632, "step": 6300 }, { "epoch": 4.378442053365461e-14, "grad_norm": 1.3133372068405151, "learning_rate": 0.00014935674241272834, "loss": 1.1108, "step": 6310 }, { "epoch": 4.3853809472693683e-14, "grad_norm": 1.434231162071228, "learning_rate": 0.00014935375877352718, "loss": 1.0833, "step": 6320 }, { "epoch": 4.3923198411732756e-14, "grad_norm": 1.602030873298645, "learning_rate": 0.00014935076826406212, "loss": 1.1049, "step": 6330 }, { "epoch": 4.399258735077183e-14, "grad_norm": 1.4527682065963745, "learning_rate": 0.0001493477708846405, "loss": 1.066, "step": 6340 }, { "epoch": 4.40619762898109e-14, "grad_norm": 1.60149347782135, "learning_rate": 0.00014934476663557028, "loss": 1.062, "step": 6350 }, { "epoch": 4.413136522884997e-14, "grad_norm": 1.7654001712799072, "learning_rate": 0.00014934175551716022, "loss": 1.089, "step": 6360 }, { "epoch": 4.4200754167889045e-14, "grad_norm": 1.3022562265396118, "learning_rate": 0.0001493387375297198, "loss": 1.1328, "step": 6370 }, { "epoch": 4.427014310692812e-14, "grad_norm": 1.3601164817810059, "learning_rate": 0.0001493357126735591, "loss": 1.0445, "step": 6380 }, { "epoch": 4.433953204596719e-14, "grad_norm": 1.3066684007644653, "learning_rate": 0.00014933268094898905, "loss": 1.063, "step": 6390 }, { "epoch": 4.440892098500626e-14, "grad_norm": 1.4158310890197754, "learning_rate": 0.00014932964235632112, "loss": 1.086, "step": 6400 }, { "epoch": 4.4478309924045334e-14, "grad_norm": 1.8126593828201294, "learning_rate": 0.0001493265968958676, "loss": 1.1262, "step": 6410 }, { "epoch": 4.4547698863084406e-14, "grad_norm": 1.762926697731018, "learning_rate": 0.00014932354456794153, "loss": 1.1335, "step": 6420 }, { "epoch": 4.461708780212348e-14, "grad_norm": 1.4028321504592896, "learning_rate": 0.0001493204853728565, "loss": 1.1157, "step": 6430 }, { "epoch": 4.468647674116255e-14, "grad_norm": 1.502228021621704, "learning_rate": 0.0001493174193109269, "loss": 1.0988, "step": 6440 }, { "epoch": 4.475586568020162e-14, "grad_norm": 1.3645344972610474, "learning_rate": 0.00014931434638246787, "loss": 1.11, "step": 6450 }, { "epoch": 4.4825254619240695e-14, "grad_norm": 1.6515939235687256, "learning_rate": 0.00014931126658779514, "loss": 1.1555, "step": 6460 }, { "epoch": 4.489464355827977e-14, "grad_norm": 1.505246877670288, "learning_rate": 0.0001493081799272252, "loss": 1.095, "step": 6470 }, { "epoch": 4.496403249731884e-14, "grad_norm": 1.5108147859573364, "learning_rate": 0.00014930508640107533, "loss": 1.047, "step": 6480 }, { "epoch": 4.503342143635791e-14, "grad_norm": 1.7379156351089478, "learning_rate": 0.0001493019860096634, "loss": 1.1164, "step": 6490 }, { "epoch": 4.5102810375396984e-14, "grad_norm": 1.4867202043533325, "learning_rate": 0.000149298878753308, "loss": 1.0369, "step": 6500 }, { "epoch": 4.5172199314436057e-14, "grad_norm": 1.646683931350708, "learning_rate": 0.00014929576463232845, "loss": 1.0421, "step": 6510 }, { "epoch": 4.524158825347513e-14, "grad_norm": 1.6492705345153809, "learning_rate": 0.00014929264364704477, "loss": 1.0828, "step": 6520 }, { "epoch": 4.53109771925142e-14, "grad_norm": 1.436767339706421, "learning_rate": 0.00014928951579777774, "loss": 1.091, "step": 6530 }, { "epoch": 4.5380366131553274e-14, "grad_norm": 1.4456214904785156, "learning_rate": 0.00014928638108484876, "loss": 1.0462, "step": 6540 }, { "epoch": 4.5449755070592346e-14, "grad_norm": 1.6934441328048706, "learning_rate": 0.00014928323950857996, "loss": 1.0995, "step": 6550 }, { "epoch": 4.551914400963142e-14, "grad_norm": 1.5086290836334229, "learning_rate": 0.0001492800910692942, "loss": 1.0378, "step": 6560 }, { "epoch": 4.558853294867049e-14, "grad_norm": 1.6073063611984253, "learning_rate": 0.00014927693576731503, "loss": 1.1151, "step": 6570 }, { "epoch": 4.565792188770956e-14, "grad_norm": 1.944796085357666, "learning_rate": 0.0001492737736029667, "loss": 1.1609, "step": 6580 }, { "epoch": 4.5727310826748635e-14, "grad_norm": 1.6255817413330078, "learning_rate": 0.00014927060457657423, "loss": 1.0464, "step": 6590 }, { "epoch": 4.579669976578771e-14, "grad_norm": 2.011263847351074, "learning_rate": 0.00014926742868846317, "loss": 1.149, "step": 6600 }, { "epoch": 4.586608870482678e-14, "grad_norm": 1.6598840951919556, "learning_rate": 0.00014926424593895998, "loss": 1.0838, "step": 6610 }, { "epoch": 4.593547764386585e-14, "grad_norm": 1.5272120237350464, "learning_rate": 0.00014926105632839172, "loss": 1.1084, "step": 6620 }, { "epoch": 4.6004866582904924e-14, "grad_norm": 1.3763394355773926, "learning_rate": 0.00014925785985708617, "loss": 1.071, "step": 6630 }, { "epoch": 4.6074255521943996e-14, "grad_norm": 1.6222338676452637, "learning_rate": 0.0001492546565253718, "loss": 0.9973, "step": 6640 }, { "epoch": 4.614364446098307e-14, "grad_norm": 1.5138095617294312, "learning_rate": 0.00014925144633357783, "loss": 1.0925, "step": 6650 }, { "epoch": 4.621303340002214e-14, "grad_norm": 1.5648725032806396, "learning_rate": 0.00014924822928203413, "loss": 1.0265, "step": 6660 }, { "epoch": 4.6282422339061213e-14, "grad_norm": 1.3777111768722534, "learning_rate": 0.00014924500537107132, "loss": 1.0641, "step": 6670 }, { "epoch": 4.6351811278100286e-14, "grad_norm": 1.597206950187683, "learning_rate": 0.0001492417746010207, "loss": 1.0992, "step": 6680 }, { "epoch": 4.642120021713936e-14, "grad_norm": 1.4832708835601807, "learning_rate": 0.00014923853697221428, "loss": 1.0556, "step": 6690 }, { "epoch": 4.649058915617843e-14, "grad_norm": 1.6673011779785156, "learning_rate": 0.00014923529248498478, "loss": 1.0874, "step": 6700 }, { "epoch": 4.65599780952175e-14, "grad_norm": 1.5188250541687012, "learning_rate": 0.00014923204113966563, "loss": 1.0397, "step": 6710 }, { "epoch": 4.6629367034256575e-14, "grad_norm": 1.7064030170440674, "learning_rate": 0.00014922878293659094, "loss": 1.0687, "step": 6720 }, { "epoch": 4.669875597329565e-14, "grad_norm": 1.919374704360962, "learning_rate": 0.00014922551787609556, "loss": 1.083, "step": 6730 }, { "epoch": 4.676814491233472e-14, "grad_norm": 1.6127924919128418, "learning_rate": 0.00014922224595851502, "loss": 1.1197, "step": 6740 }, { "epoch": 4.683753385137379e-14, "grad_norm": 1.69143545627594, "learning_rate": 0.0001492189671841855, "loss": 1.0307, "step": 6750 }, { "epoch": 4.6906922790412864e-14, "grad_norm": 1.421725869178772, "learning_rate": 0.00014921568155344407, "loss": 1.0989, "step": 6760 }, { "epoch": 4.6976311729451936e-14, "grad_norm": 1.6075016260147095, "learning_rate": 0.00014921238906662827, "loss": 1.125, "step": 6770 }, { "epoch": 4.704570066849101e-14, "grad_norm": 1.638609766960144, "learning_rate": 0.0001492090897240765, "loss": 1.0596, "step": 6780 }, { "epoch": 4.711508960753008e-14, "grad_norm": 1.5289114713668823, "learning_rate": 0.00014920578352612784, "loss": 1.0122, "step": 6790 }, { "epoch": 4.718447854656915e-14, "grad_norm": 1.662838101387024, "learning_rate": 0.000149202470473122, "loss": 1.0429, "step": 6800 }, { "epoch": 4.7253867485608225e-14, "grad_norm": 1.5796163082122803, "learning_rate": 0.00014919915056539948, "loss": 1.029, "step": 6810 }, { "epoch": 4.73232564246473e-14, "grad_norm": 1.709768533706665, "learning_rate": 0.00014919582380330145, "loss": 1.0089, "step": 6820 }, { "epoch": 4.739264536368637e-14, "grad_norm": 1.9371436834335327, "learning_rate": 0.00014919249018716978, "loss": 1.0737, "step": 6830 }, { "epoch": 4.746203430272544e-14, "grad_norm": 1.30146062374115, "learning_rate": 0.00014918914971734707, "loss": 1.0628, "step": 6840 }, { "epoch": 4.7531423241764514e-14, "grad_norm": 1.8570631742477417, "learning_rate": 0.00014918580239417655, "loss": 1.0265, "step": 6850 }, { "epoch": 4.7600812180803587e-14, "grad_norm": 1.6485148668289185, "learning_rate": 0.00014918244821800227, "loss": 1.0406, "step": 6860 }, { "epoch": 4.767020111984266e-14, "grad_norm": 1.5119093656539917, "learning_rate": 0.0001491790871891689, "loss": 1.0277, "step": 6870 }, { "epoch": 4.773959005888173e-14, "grad_norm": 1.6071621179580688, "learning_rate": 0.00014917571930802184, "loss": 1.0683, "step": 6880 }, { "epoch": 4.7808978997920804e-14, "grad_norm": 1.5294808149337769, "learning_rate": 0.00014917234457490722, "loss": 1.0832, "step": 6890 }, { "epoch": 4.7878367936959876e-14, "grad_norm": 1.4667280912399292, "learning_rate": 0.0001491689629901718, "loss": 1.0472, "step": 6900 }, { "epoch": 4.794775687599895e-14, "grad_norm": 1.722569227218628, "learning_rate": 0.0001491655745541631, "loss": 1.062, "step": 6910 }, { "epoch": 4.801714581503802e-14, "grad_norm": 1.304660677909851, "learning_rate": 0.00014916217926722934, "loss": 1.0617, "step": 6920 }, { "epoch": 4.808653475407709e-14, "grad_norm": 1.25371515750885, "learning_rate": 0.00014915877712971945, "loss": 1.1047, "step": 6930 }, { "epoch": 4.8155923693116165e-14, "grad_norm": 1.1841133832931519, "learning_rate": 0.00014915536814198303, "loss": 1.0115, "step": 6940 }, { "epoch": 4.822531263215524e-14, "grad_norm": 1.6290136575698853, "learning_rate": 0.00014915195230437046, "loss": 1.0237, "step": 6950 }, { "epoch": 4.829470157119431e-14, "grad_norm": 1.442118763923645, "learning_rate": 0.0001491485296172327, "loss": 1.0405, "step": 6960 }, { "epoch": 4.836409051023338e-14, "grad_norm": 1.4184110164642334, "learning_rate": 0.00014914510008092152, "loss": 0.9834, "step": 6970 }, { "epoch": 4.8433479449272454e-14, "grad_norm": 1.4639792442321777, "learning_rate": 0.0001491416636957894, "loss": 1.0283, "step": 6980 }, { "epoch": 4.8502868388311526e-14, "grad_norm": 1.747564435005188, "learning_rate": 0.00014913822046218938, "loss": 1.035, "step": 6990 }, { "epoch": 4.85722573273506e-14, "grad_norm": 1.3700151443481445, "learning_rate": 0.00014913477038047538, "loss": 1.0141, "step": 7000 }, { "epoch": 4.864164626638967e-14, "grad_norm": 1.8251084089279175, "learning_rate": 0.0001491313134510019, "loss": 1.0068, "step": 7010 }, { "epoch": 4.8711035205428743e-14, "grad_norm": 1.973168134689331, "learning_rate": 0.00014912784967412428, "loss": 1.0566, "step": 7020 }, { "epoch": 4.8780424144467815e-14, "grad_norm": 1.8557467460632324, "learning_rate": 0.0001491243790501984, "loss": 0.9915, "step": 7030 }, { "epoch": 4.884981308350689e-14, "grad_norm": 1.5770270824432373, "learning_rate": 0.00014912090157958094, "loss": 1.0417, "step": 7040 }, { "epoch": 4.891920202254596e-14, "grad_norm": 1.5540941953659058, "learning_rate": 0.00014911741726262925, "loss": 1.0963, "step": 7050 }, { "epoch": 4.898859096158503e-14, "grad_norm": 1.4301692247390747, "learning_rate": 0.00014911392609970146, "loss": 0.9823, "step": 7060 }, { "epoch": 4.9057979900624105e-14, "grad_norm": 1.7769068479537964, "learning_rate": 0.0001491104280911563, "loss": 1.0981, "step": 7070 }, { "epoch": 4.912736883966318e-14, "grad_norm": 1.6712015867233276, "learning_rate": 0.0001491069232373532, "loss": 1.0524, "step": 7080 }, { "epoch": 4.919675777870225e-14, "grad_norm": 1.6782456636428833, "learning_rate": 0.00014910341153865237, "loss": 1.0269, "step": 7090 }, { "epoch": 4.926614671774132e-14, "grad_norm": 1.6421294212341309, "learning_rate": 0.00014909989299541473, "loss": 0.9766, "step": 7100 }, { "epoch": 4.9335535656780394e-14, "grad_norm": 1.9402964115142822, "learning_rate": 0.0001490963676080018, "loss": 0.9835, "step": 7110 }, { "epoch": 4.9404924595819466e-14, "grad_norm": 1.6834173202514648, "learning_rate": 0.00014909283537677595, "loss": 1.0012, "step": 7120 }, { "epoch": 4.947431353485854e-14, "grad_norm": 1.5281224250793457, "learning_rate": 0.00014908929630210013, "loss": 0.9675, "step": 7130 }, { "epoch": 4.954370247389761e-14, "grad_norm": 1.4570081233978271, "learning_rate": 0.000149085750384338, "loss": 0.9749, "step": 7140 }, { "epoch": 4.961309141293668e-14, "grad_norm": 1.8173484802246094, "learning_rate": 0.000149082197623854, "loss": 1.0669, "step": 7150 }, { "epoch": 4.9682480351975755e-14, "grad_norm": 1.5981179475784302, "learning_rate": 0.00014907863802101322, "loss": 0.9368, "step": 7160 }, { "epoch": 4.975186929101483e-14, "grad_norm": 1.5754978656768799, "learning_rate": 0.00014907507157618148, "loss": 1.0049, "step": 7170 }, { "epoch": 4.98212582300539e-14, "grad_norm": 1.5819036960601807, "learning_rate": 0.00014907149828972523, "loss": 1.0218, "step": 7180 }, { "epoch": 4.989064716909297e-14, "grad_norm": 1.3645238876342773, "learning_rate": 0.00014906791816201177, "loss": 1.0206, "step": 7190 }, { "epoch": 4.9960036108132044e-14, "grad_norm": 1.4233559370040894, "learning_rate": 0.00014906433119340893, "loss": 1.0149, "step": 7200 }, { "epoch": 5.0029425047171117e-14, "grad_norm": 1.6572462320327759, "learning_rate": 0.0001490607373842854, "loss": 1.0754, "step": 7210 }, { "epoch": 5.009881398621019e-14, "grad_norm": 1.5963687896728516, "learning_rate": 0.00014905713673501046, "loss": 0.9806, "step": 7220 }, { "epoch": 5.016820292524926e-14, "grad_norm": 1.6888257265090942, "learning_rate": 0.00014905352924595412, "loss": 1.0373, "step": 7230 }, { "epoch": 5.0237591864288333e-14, "grad_norm": 1.483241319656372, "learning_rate": 0.00014904991491748712, "loss": 1.0573, "step": 7240 }, { "epoch": 5.0306980803327406e-14, "grad_norm": 1.5689283609390259, "learning_rate": 0.0001490462937499809, "loss": 1.0, "step": 7250 }, { "epoch": 5.037636974236648e-14, "grad_norm": 1.5441135168075562, "learning_rate": 0.00014904266574380758, "loss": 0.9748, "step": 7260 }, { "epoch": 5.044575868140555e-14, "grad_norm": 1.375638484954834, "learning_rate": 0.00014903903089934, "loss": 0.9893, "step": 7270 }, { "epoch": 5.051514762044462e-14, "grad_norm": 1.5695685148239136, "learning_rate": 0.00014903538921695172, "loss": 0.9719, "step": 7280 }, { "epoch": 5.0584536559483695e-14, "grad_norm": 1.7745568752288818, "learning_rate": 0.00014903174069701692, "loss": 0.9123, "step": 7290 }, { "epoch": 5.065392549852277e-14, "grad_norm": 1.387685775756836, "learning_rate": 0.00014902808533991056, "loss": 0.9821, "step": 7300 }, { "epoch": 5.072331443756184e-14, "grad_norm": 1.5993086099624634, "learning_rate": 0.00014902442314600835, "loss": 0.9897, "step": 7310 }, { "epoch": 5.079270337660091e-14, "grad_norm": 1.6205099821090698, "learning_rate": 0.00014902075411568654, "loss": 0.9377, "step": 7320 }, { "epoch": 5.0862092315639984e-14, "grad_norm": 1.65731680393219, "learning_rate": 0.00014901707824932224, "loss": 0.9907, "step": 7330 }, { "epoch": 5.0931481254679056e-14, "grad_norm": 1.6752699613571167, "learning_rate": 0.0001490133955472932, "loss": 1.0004, "step": 7340 }, { "epoch": 5.100087019371813e-14, "grad_norm": 1.6765345335006714, "learning_rate": 0.00014900970600997785, "loss": 1.0206, "step": 7350 }, { "epoch": 5.10702591327572e-14, "grad_norm": 1.6786681413650513, "learning_rate": 0.00014900600963775539, "loss": 1.0334, "step": 7360 }, { "epoch": 5.113964807179627e-14, "grad_norm": 1.4011380672454834, "learning_rate": 0.00014900230643100562, "loss": 0.9647, "step": 7370 }, { "epoch": 5.1209037010835345e-14, "grad_norm": 1.7928099632263184, "learning_rate": 0.00014899859639010916, "loss": 1.0344, "step": 7380 }, { "epoch": 5.127842594987442e-14, "grad_norm": 1.685343861579895, "learning_rate": 0.00014899487951544724, "loss": 1.0111, "step": 7390 }, { "epoch": 5.134781488891349e-14, "grad_norm": 1.853797197341919, "learning_rate": 0.00014899115580740182, "loss": 1.0362, "step": 7400 }, { "epoch": 5.141720382795256e-14, "grad_norm": 1.82111656665802, "learning_rate": 0.00014898742526635558, "loss": 0.9828, "step": 7410 }, { "epoch": 5.1486592766991635e-14, "grad_norm": 1.4297012090682983, "learning_rate": 0.00014898368789269194, "loss": 0.9159, "step": 7420 }, { "epoch": 5.155598170603071e-14, "grad_norm": 1.5177209377288818, "learning_rate": 0.00014897994368679487, "loss": 0.9824, "step": 7430 }, { "epoch": 5.162537064506978e-14, "grad_norm": 1.4061439037322998, "learning_rate": 0.00014897619264904924, "loss": 0.9448, "step": 7440 }, { "epoch": 5.169475958410885e-14, "grad_norm": 1.7434873580932617, "learning_rate": 0.00014897243477984048, "loss": 0.9647, "step": 7450 }, { "epoch": 5.1764148523147924e-14, "grad_norm": 1.7755237817764282, "learning_rate": 0.00014896867007955478, "loss": 0.9659, "step": 7460 }, { "epoch": 5.1833537462186996e-14, "grad_norm": 1.5148093700408936, "learning_rate": 0.000148964898548579, "loss": 0.9315, "step": 7470 }, { "epoch": 5.190292640122607e-14, "grad_norm": 1.716980218887329, "learning_rate": 0.0001489611201873008, "loss": 0.9722, "step": 7480 }, { "epoch": 5.197231534026514e-14, "grad_norm": 1.4663524627685547, "learning_rate": 0.0001489573349961084, "loss": 0.9774, "step": 7490 }, { "epoch": 5.204170427930421e-14, "grad_norm": 1.8997145891189575, "learning_rate": 0.00014895354297539079, "loss": 0.9738, "step": 7500 }, { "epoch": 5.2111093218343285e-14, "grad_norm": 1.7451438903808594, "learning_rate": 0.00014894974412553766, "loss": 0.9855, "step": 7510 }, { "epoch": 5.218048215738236e-14, "grad_norm": 1.6771354675292969, "learning_rate": 0.0001489459384469394, "loss": 1.0328, "step": 7520 }, { "epoch": 5.224987109642143e-14, "grad_norm": 1.5617295503616333, "learning_rate": 0.00014894212593998712, "loss": 0.9352, "step": 7530 }, { "epoch": 5.23192600354605e-14, "grad_norm": 1.6316944360733032, "learning_rate": 0.0001489383066050726, "loss": 1.0292, "step": 7540 }, { "epoch": 5.2388648974499574e-14, "grad_norm": 1.5140089988708496, "learning_rate": 0.00014893448044258834, "loss": 1.0238, "step": 7550 }, { "epoch": 5.2458037913538647e-14, "grad_norm": 1.5109648704528809, "learning_rate": 0.00014893064745292756, "loss": 0.965, "step": 7560 }, { "epoch": 5.252742685257772e-14, "grad_norm": 1.6603038311004639, "learning_rate": 0.00014892680763648413, "loss": 0.9411, "step": 7570 }, { "epoch": 5.259681579161679e-14, "grad_norm": 1.741356611251831, "learning_rate": 0.00014892296099365266, "loss": 0.9711, "step": 7580 }, { "epoch": 5.2666204730655863e-14, "grad_norm": 1.7549865245819092, "learning_rate": 0.00014891910752482844, "loss": 0.9311, "step": 7590 }, { "epoch": 5.2735593669694936e-14, "grad_norm": 1.5028401613235474, "learning_rate": 0.00014891524723040749, "loss": 0.905, "step": 7600 }, { "epoch": 5.280498260873401e-14, "grad_norm": 1.668319582939148, "learning_rate": 0.0001489113801107865, "loss": 0.937, "step": 7610 }, { "epoch": 5.287437154777308e-14, "grad_norm": 1.3178852796554565, "learning_rate": 0.0001489075061663629, "loss": 0.9928, "step": 7620 }, { "epoch": 5.294376048681215e-14, "grad_norm": 1.4248154163360596, "learning_rate": 0.0001489036253975348, "loss": 0.8842, "step": 7630 }, { "epoch": 5.3013149425851225e-14, "grad_norm": 1.6834237575531006, "learning_rate": 0.000148899737804701, "loss": 0.9488, "step": 7640 }, { "epoch": 5.30825383648903e-14, "grad_norm": 1.4553626775741577, "learning_rate": 0.000148895843388261, "loss": 0.9804, "step": 7650 }, { "epoch": 5.315192730392937e-14, "grad_norm": 1.7196199893951416, "learning_rate": 0.000148891942148615, "loss": 0.9886, "step": 7660 }, { "epoch": 5.322131624296844e-14, "grad_norm": 1.860291600227356, "learning_rate": 0.00014888803408616392, "loss": 0.9837, "step": 7670 }, { "epoch": 5.3290705182007514e-14, "grad_norm": 1.565588355064392, "learning_rate": 0.00014888411920130942, "loss": 0.9772, "step": 7680 }, { "epoch": 5.3360094121046586e-14, "grad_norm": 1.4972505569458008, "learning_rate": 0.00014888019749445376, "loss": 0.938, "step": 7690 }, { "epoch": 5.342948306008566e-14, "grad_norm": 1.5817397832870483, "learning_rate": 0.00014887626896599996, "loss": 0.9836, "step": 7700 }, { "epoch": 5.349887199912473e-14, "grad_norm": 1.6830259561538696, "learning_rate": 0.0001488723336163518, "loss": 1.0149, "step": 7710 }, { "epoch": 5.35682609381638e-14, "grad_norm": 1.603892207145691, "learning_rate": 0.00014886839144591363, "loss": 0.9554, "step": 7720 }, { "epoch": 5.3637649877202875e-14, "grad_norm": 1.5990626811981201, "learning_rate": 0.0001488644424550906, "loss": 0.9289, "step": 7730 }, { "epoch": 5.370703881624195e-14, "grad_norm": 1.4782373905181885, "learning_rate": 0.00014886048664428851, "loss": 0.9758, "step": 7740 }, { "epoch": 5.377642775528102e-14, "grad_norm": 1.5879873037338257, "learning_rate": 0.0001488565240139139, "loss": 0.9088, "step": 7750 }, { "epoch": 5.384581669432009e-14, "grad_norm": 1.5485289096832275, "learning_rate": 0.00014885255456437399, "loss": 0.8933, "step": 7760 }, { "epoch": 5.3915205633359164e-14, "grad_norm": 1.5896878242492676, "learning_rate": 0.0001488485782960767, "loss": 0.9292, "step": 7770 }, { "epoch": 5.3984594572398237e-14, "grad_norm": 1.4275259971618652, "learning_rate": 0.00014884459520943063, "loss": 0.9311, "step": 7780 }, { "epoch": 5.405398351143731e-14, "grad_norm": 1.5605356693267822, "learning_rate": 0.00014884060530484514, "loss": 0.8926, "step": 7790 }, { "epoch": 5.412337245047638e-14, "grad_norm": 1.9389886856079102, "learning_rate": 0.00014883660858273025, "loss": 0.9926, "step": 7800 }, { "epoch": 5.4192761389515454e-14, "grad_norm": 1.6532649993896484, "learning_rate": 0.00014883260504349667, "loss": 0.9287, "step": 7810 }, { "epoch": 5.4262150328554526e-14, "grad_norm": 1.6964521408081055, "learning_rate": 0.0001488285946875558, "loss": 0.9753, "step": 7820 }, { "epoch": 5.43315392675936e-14, "grad_norm": 2.033000946044922, "learning_rate": 0.00014882457751531986, "loss": 1.0158, "step": 7830 }, { "epoch": 5.440092820663267e-14, "grad_norm": 1.7651774883270264, "learning_rate": 0.0001488205535272016, "loss": 0.9111, "step": 7840 }, { "epoch": 5.447031714567174e-14, "grad_norm": 1.7897096872329712, "learning_rate": 0.00014881652272361452, "loss": 0.9872, "step": 7850 }, { "epoch": 5.4539706084710815e-14, "grad_norm": 1.884938359260559, "learning_rate": 0.0001488124851049729, "loss": 0.9443, "step": 7860 }, { "epoch": 5.460909502374989e-14, "grad_norm": 1.7755769491195679, "learning_rate": 0.00014880844067169169, "loss": 0.9636, "step": 7870 }, { "epoch": 5.467848396278896e-14, "grad_norm": 1.4995214939117432, "learning_rate": 0.00014880438942418647, "loss": 0.9344, "step": 7880 }, { "epoch": 5.474787290182803e-14, "grad_norm": 1.5900394916534424, "learning_rate": 0.0001488003313628736, "loss": 0.8811, "step": 7890 }, { "epoch": 5.4817261840867104e-14, "grad_norm": 1.767905831336975, "learning_rate": 0.00014879626648817008, "loss": 0.9461, "step": 7900 }, { "epoch": 5.4886650779906176e-14, "grad_norm": 1.6974843740463257, "learning_rate": 0.00014879219480049366, "loss": 0.9046, "step": 7910 }, { "epoch": 5.495603971894525e-14, "grad_norm": 1.4110424518585205, "learning_rate": 0.00014878811630026275, "loss": 0.911, "step": 7920 }, { "epoch": 5.502542865798432e-14, "grad_norm": 1.617721438407898, "learning_rate": 0.0001487840309878965, "loss": 0.9483, "step": 7930 }, { "epoch": 5.5094817597023393e-14, "grad_norm": 1.6433985233306885, "learning_rate": 0.00014877993886381472, "loss": 0.8907, "step": 7940 }, { "epoch": 5.5164206536062466e-14, "grad_norm": 1.7467045783996582, "learning_rate": 0.000148775839928438, "loss": 0.9438, "step": 7950 }, { "epoch": 5.523359547510154e-14, "grad_norm": 1.7288267612457275, "learning_rate": 0.00014877173418218745, "loss": 0.9297, "step": 7960 }, { "epoch": 5.530298441414061e-14, "grad_norm": 1.9646748304367065, "learning_rate": 0.00014876762162548512, "loss": 0.9465, "step": 7970 }, { "epoch": 5.537237335317968e-14, "grad_norm": 1.6966099739074707, "learning_rate": 0.00014876350225875357, "loss": 0.939, "step": 7980 }, { "epoch": 5.5441762292218755e-14, "grad_norm": 1.7682116031646729, "learning_rate": 0.00014875937608241618, "loss": 0.9805, "step": 7990 }, { "epoch": 5.551115123125783e-14, "grad_norm": 1.7481449842453003, "learning_rate": 0.0001487552430968969, "loss": 0.8948, "step": 8000 }, { "epoch": 5.55805401702969e-14, "grad_norm": 1.4799022674560547, "learning_rate": 0.00014875110330262053, "loss": 0.9468, "step": 8010 }, { "epoch": 5.564992910933597e-14, "grad_norm": 1.4530788660049438, "learning_rate": 0.00014874695670001245, "loss": 0.9789, "step": 8020 }, { "epoch": 5.5719318048375044e-14, "grad_norm": 1.8860975503921509, "learning_rate": 0.00014874280328949885, "loss": 0.9305, "step": 8030 }, { "epoch": 5.5788706987414116e-14, "grad_norm": 1.7742464542388916, "learning_rate": 0.00014873864307150648, "loss": 0.8942, "step": 8040 }, { "epoch": 5.585809592645319e-14, "grad_norm": 1.6210575103759766, "learning_rate": 0.00014873447604646294, "loss": 0.9187, "step": 8050 }, { "epoch": 5.592748486549226e-14, "grad_norm": 1.6796706914901733, "learning_rate": 0.00014873030221479642, "loss": 0.8926, "step": 8060 }, { "epoch": 5.599687380453133e-14, "grad_norm": 1.5521727800369263, "learning_rate": 0.00014872612157693584, "loss": 0.8647, "step": 8070 }, { "epoch": 5.6066262743570405e-14, "grad_norm": 1.7140781879425049, "learning_rate": 0.00014872193413331084, "loss": 0.9155, "step": 8080 }, { "epoch": 5.613565168260948e-14, "grad_norm": 1.4365912675857544, "learning_rate": 0.00014871773988435174, "loss": 0.915, "step": 8090 }, { "epoch": 5.620504062164855e-14, "grad_norm": 1.8225301504135132, "learning_rate": 0.00014871353883048956, "loss": 0.8795, "step": 8100 }, { "epoch": 5.627442956068762e-14, "grad_norm": 1.7657774686813354, "learning_rate": 0.000148709330972156, "loss": 0.9115, "step": 8110 }, { "epoch": 5.6343818499726694e-14, "grad_norm": 1.6334878206253052, "learning_rate": 0.00014870511630978356, "loss": 0.9176, "step": 8120 }, { "epoch": 5.6413207438765767e-14, "grad_norm": 1.8052997589111328, "learning_rate": 0.0001487008948438053, "loss": 0.9568, "step": 8130 }, { "epoch": 5.648259637780484e-14, "grad_norm": 1.7089539766311646, "learning_rate": 0.00014869666657465504, "loss": 0.946, "step": 8140 }, { "epoch": 5.655198531684391e-14, "grad_norm": 1.3745145797729492, "learning_rate": 0.0001486924315027673, "loss": 0.9044, "step": 8150 }, { "epoch": 5.6621374255882984e-14, "grad_norm": 1.5789188146591187, "learning_rate": 0.00014868818962857734, "loss": 0.9003, "step": 8160 }, { "epoch": 5.6690763194922056e-14, "grad_norm": 1.5538816452026367, "learning_rate": 0.00014868394095252105, "loss": 0.9173, "step": 8170 }, { "epoch": 5.676015213396113e-14, "grad_norm": 1.2747383117675781, "learning_rate": 0.00014867968547503505, "loss": 0.9509, "step": 8180 }, { "epoch": 5.68295410730002e-14, "grad_norm": 1.308235764503479, "learning_rate": 0.00014867542319655665, "loss": 0.8835, "step": 8190 }, { "epoch": 5.689893001203927e-14, "grad_norm": 1.8319785594940186, "learning_rate": 0.00014867115411752388, "loss": 0.9085, "step": 8200 }, { "epoch": 5.696831895107834e-14, "grad_norm": 1.4376585483551025, "learning_rate": 0.00014866687823837543, "loss": 0.9001, "step": 8210 }, { "epoch": 5.703770789011742e-14, "grad_norm": 1.4694677591323853, "learning_rate": 0.00014866259555955077, "loss": 0.8557, "step": 8220 }, { "epoch": 5.710709682915649e-14, "grad_norm": 1.739822506904602, "learning_rate": 0.00014865830608148993, "loss": 0.8852, "step": 8230 }, { "epoch": 5.717648576819556e-14, "grad_norm": 1.7187854051589966, "learning_rate": 0.00014865400980463377, "loss": 0.8903, "step": 8240 }, { "epoch": 5.724587470723463e-14, "grad_norm": 1.5888073444366455, "learning_rate": 0.0001486497067294238, "loss": 0.8812, "step": 8250 }, { "epoch": 5.731526364627371e-14, "grad_norm": 1.9992470741271973, "learning_rate": 0.0001486453968563022, "loss": 0.8777, "step": 8260 }, { "epoch": 5.738465258531278e-14, "grad_norm": 1.756414771080017, "learning_rate": 0.00014864108018571194, "loss": 0.9252, "step": 8270 }, { "epoch": 5.745404152435185e-14, "grad_norm": 1.8830442428588867, "learning_rate": 0.00014863675671809655, "loss": 0.8865, "step": 8280 }, { "epoch": 5.752343046339092e-14, "grad_norm": 1.7648216485977173, "learning_rate": 0.00014863242645390035, "loss": 0.9345, "step": 8290 }, { "epoch": 5.759281940243e-14, "grad_norm": 1.617107629776001, "learning_rate": 0.00014862808939356838, "loss": 0.9654, "step": 8300 }, { "epoch": 5.766220834146907e-14, "grad_norm": 1.465878963470459, "learning_rate": 0.00014862374553754632, "loss": 0.8525, "step": 8310 }, { "epoch": 5.773159728050814e-14, "grad_norm": 1.7221111059188843, "learning_rate": 0.00014861939488628058, "loss": 0.9439, "step": 8320 }, { "epoch": 5.780098621954721e-14, "grad_norm": 1.6850534677505493, "learning_rate": 0.0001486150374402182, "loss": 0.9207, "step": 8330 }, { "epoch": 5.787037515858628e-14, "grad_norm": 1.5449188947677612, "learning_rate": 0.00014861067319980705, "loss": 0.8965, "step": 8340 }, { "epoch": 5.793976409762536e-14, "grad_norm": 1.6744966506958008, "learning_rate": 0.00014860630216549555, "loss": 0.8531, "step": 8350 }, { "epoch": 5.800915303666443e-14, "grad_norm": 1.9027297496795654, "learning_rate": 0.00014860192433773298, "loss": 0.8674, "step": 8360 }, { "epoch": 5.80785419757035e-14, "grad_norm": 1.5491387844085693, "learning_rate": 0.00014859753971696918, "loss": 0.8669, "step": 8370 }, { "epoch": 5.814793091474257e-14, "grad_norm": 1.7119169235229492, "learning_rate": 0.0001485931483036547, "loss": 0.8576, "step": 8380 }, { "epoch": 5.821731985378165e-14, "grad_norm": 1.4522480964660645, "learning_rate": 0.00014858875009824093, "loss": 0.8527, "step": 8390 }, { "epoch": 5.828670879282072e-14, "grad_norm": 2.0944488048553467, "learning_rate": 0.00014858434510117976, "loss": 0.9414, "step": 8400 }, { "epoch": 5.835609773185979e-14, "grad_norm": 1.6326664686203003, "learning_rate": 0.0001485799333129239, "loss": 0.8176, "step": 8410 }, { "epoch": 5.842548667089886e-14, "grad_norm": 1.3971549272537231, "learning_rate": 0.00014857551473392675, "loss": 0.8755, "step": 8420 }, { "epoch": 5.849487560993794e-14, "grad_norm": 1.698218584060669, "learning_rate": 0.00014857108936464235, "loss": 0.8982, "step": 8430 }, { "epoch": 5.856426454897701e-14, "grad_norm": 1.6616417169570923, "learning_rate": 0.00014856665720552553, "loss": 0.912, "step": 8440 }, { "epoch": 5.863365348801608e-14, "grad_norm": 1.5370365381240845, "learning_rate": 0.00014856221825703167, "loss": 0.9003, "step": 8450 }, { "epoch": 5.870304242705515e-14, "grad_norm": 1.541561245918274, "learning_rate": 0.00014855777251961702, "loss": 0.9489, "step": 8460 }, { "epoch": 5.877243136609422e-14, "grad_norm": 1.640614628791809, "learning_rate": 0.00014855331999373848, "loss": 0.8423, "step": 8470 }, { "epoch": 5.88418203051333e-14, "grad_norm": 1.8100106716156006, "learning_rate": 0.00014854886067985353, "loss": 0.8967, "step": 8480 }, { "epoch": 5.891120924417237e-14, "grad_norm": 1.7398099899291992, "learning_rate": 0.00014854439457842047, "loss": 0.927, "step": 8490 }, { "epoch": 5.898059818321144e-14, "grad_norm": 1.584211826324463, "learning_rate": 0.00014853992168989828, "loss": 0.8669, "step": 8500 }, { "epoch": 5.904998712225051e-14, "grad_norm": 1.652841329574585, "learning_rate": 0.00014853544201474657, "loss": 0.8434, "step": 8510 }, { "epoch": 5.911937606128959e-14, "grad_norm": 1.7805161476135254, "learning_rate": 0.00014853095555342576, "loss": 0.8704, "step": 8520 }, { "epoch": 5.918876500032866e-14, "grad_norm": 1.7415740489959717, "learning_rate": 0.00014852646230639684, "loss": 0.8606, "step": 8530 }, { "epoch": 5.925815393936773e-14, "grad_norm": 1.775122046470642, "learning_rate": 0.00014852196227412161, "loss": 0.8076, "step": 8540 }, { "epoch": 5.93275428784068e-14, "grad_norm": 1.7406561374664307, "learning_rate": 0.0001485174554570625, "loss": 0.8711, "step": 8550 }, { "epoch": 5.939693181744587e-14, "grad_norm": 1.766641616821289, "learning_rate": 0.00014851294185568264, "loss": 0.87, "step": 8560 }, { "epoch": 5.946632075648495e-14, "grad_norm": 1.634676456451416, "learning_rate": 0.0001485084214704459, "loss": 0.8215, "step": 8570 }, { "epoch": 5.953570969552402e-14, "grad_norm": 1.6285473108291626, "learning_rate": 0.00014850389430181682, "loss": 0.86, "step": 8580 }, { "epoch": 5.960509863456309e-14, "grad_norm": 1.6516666412353516, "learning_rate": 0.0001484993603502606, "loss": 0.8784, "step": 8590 }, { "epoch": 5.967448757360216e-14, "grad_norm": 1.6260311603546143, "learning_rate": 0.00014849481961624322, "loss": 0.9051, "step": 8600 }, { "epoch": 5.974387651264124e-14, "grad_norm": 1.615708351135254, "learning_rate": 0.00014849027210023127, "loss": 0.9151, "step": 8610 }, { "epoch": 5.981326545168031e-14, "grad_norm": 1.6598455905914307, "learning_rate": 0.0001484857178026921, "loss": 0.841, "step": 8620 }, { "epoch": 5.988265439071938e-14, "grad_norm": 1.7843338251113892, "learning_rate": 0.00014848115672409375, "loss": 0.8902, "step": 8630 }, { "epoch": 5.995204332975845e-14, "grad_norm": 1.776330590248108, "learning_rate": 0.00014847658886490492, "loss": 0.8856, "step": 8640 }, { "epoch": 6.002143226879753e-14, "grad_norm": 1.684604525566101, "learning_rate": 0.00014847201422559504, "loss": 0.9049, "step": 8650 }, { "epoch": 6.00908212078366e-14, "grad_norm": 1.8272384405136108, "learning_rate": 0.0001484674328066342, "loss": 0.837, "step": 8660 }, { "epoch": 6.016021014687567e-14, "grad_norm": 1.4384194612503052, "learning_rate": 0.00014846284460849325, "loss": 0.7971, "step": 8670 }, { "epoch": 6.022959908591474e-14, "grad_norm": 1.5380032062530518, "learning_rate": 0.00014845824963164366, "loss": 0.8679, "step": 8680 }, { "epoch": 6.029898802495381e-14, "grad_norm": 1.7192710638046265, "learning_rate": 0.00014845364787655766, "loss": 0.836, "step": 8690 }, { "epoch": 6.036837696399289e-14, "grad_norm": 1.5962748527526855, "learning_rate": 0.00014844903934370816, "loss": 0.8412, "step": 8700 }, { "epoch": 6.043776590303196e-14, "grad_norm": 1.8154973983764648, "learning_rate": 0.00014844442403356872, "loss": 0.8393, "step": 8710 }, { "epoch": 6.050715484207103e-14, "grad_norm": 1.5707809925079346, "learning_rate": 0.0001484398019466137, "loss": 0.8206, "step": 8720 }, { "epoch": 6.05765437811101e-14, "grad_norm": 1.840761423110962, "learning_rate": 0.00014843517308331803, "loss": 0.8527, "step": 8730 }, { "epoch": 6.064593272014918e-14, "grad_norm": 1.757648229598999, "learning_rate": 0.00014843053744415744, "loss": 0.8583, "step": 8740 }, { "epoch": 6.071532165918825e-14, "grad_norm": 1.7229485511779785, "learning_rate": 0.00014842589502960828, "loss": 0.8529, "step": 8750 }, { "epoch": 6.078471059822732e-14, "grad_norm": 1.7419410943984985, "learning_rate": 0.00014842124584014762, "loss": 0.8725, "step": 8760 }, { "epoch": 6.085409953726639e-14, "grad_norm": 1.6117035150527954, "learning_rate": 0.00014841658987625327, "loss": 0.9038, "step": 8770 }, { "epoch": 6.092348847630547e-14, "grad_norm": 1.5970323085784912, "learning_rate": 0.00014841192713840371, "loss": 0.8009, "step": 8780 }, { "epoch": 6.099287741534454e-14, "grad_norm": 1.5609781742095947, "learning_rate": 0.00014840725762707807, "loss": 0.9026, "step": 8790 }, { "epoch": 6.106226635438361e-14, "grad_norm": 1.7113856077194214, "learning_rate": 0.00014840258134275625, "loss": 0.9068, "step": 8800 }, { "epoch": 6.113165529342268e-14, "grad_norm": 1.4494893550872803, "learning_rate": 0.0001483978982859188, "loss": 0.8317, "step": 8810 }, { "epoch": 6.120104423246175e-14, "grad_norm": 1.7341556549072266, "learning_rate": 0.00014839320845704698, "loss": 0.8059, "step": 8820 }, { "epoch": 6.127043317150083e-14, "grad_norm": 1.8322105407714844, "learning_rate": 0.00014838851185662276, "loss": 0.8504, "step": 8830 }, { "epoch": 6.13398221105399e-14, "grad_norm": 1.6304552555084229, "learning_rate": 0.00014838380848512873, "loss": 0.8132, "step": 8840 }, { "epoch": 6.140921104957897e-14, "grad_norm": 1.6699870824813843, "learning_rate": 0.00014837909834304827, "loss": 0.8014, "step": 8850 }, { "epoch": 6.147859998861804e-14, "grad_norm": 1.6839110851287842, "learning_rate": 0.00014837438143086543, "loss": 0.8081, "step": 8860 }, { "epoch": 6.154798892765712e-14, "grad_norm": 1.6309720277786255, "learning_rate": 0.00014836965774906493, "loss": 0.8758, "step": 8870 }, { "epoch": 6.161737786669619e-14, "grad_norm": 1.4689602851867676, "learning_rate": 0.00014836492729813225, "loss": 0.7796, "step": 8880 }, { "epoch": 6.168676680573526e-14, "grad_norm": 1.6518499851226807, "learning_rate": 0.00014836019007855345, "loss": 0.8255, "step": 8890 }, { "epoch": 6.175615574477433e-14, "grad_norm": 1.4563132524490356, "learning_rate": 0.00014835544609081535, "loss": 0.8668, "step": 8900 }, { "epoch": 6.18255446838134e-14, "grad_norm": 1.7883716821670532, "learning_rate": 0.00014835069533540552, "loss": 0.8803, "step": 8910 }, { "epoch": 6.189493362285248e-14, "grad_norm": 2.1836071014404297, "learning_rate": 0.00014834593781281217, "loss": 0.8722, "step": 8920 }, { "epoch": 6.196432256189155e-14, "grad_norm": 1.4780985116958618, "learning_rate": 0.00014834117352352418, "loss": 0.8433, "step": 8930 }, { "epoch": 6.203371150093062e-14, "grad_norm": 1.6509366035461426, "learning_rate": 0.00014833640246803118, "loss": 0.8195, "step": 8940 }, { "epoch": 6.21031004399697e-14, "grad_norm": 1.574957013130188, "learning_rate": 0.00014833162464682344, "loss": 0.8713, "step": 8950 }, { "epoch": 6.217248937900877e-14, "grad_norm": 1.6771413087844849, "learning_rate": 0.00014832684006039197, "loss": 0.8955, "step": 8960 }, { "epoch": 6.224187831804784e-14, "grad_norm": 1.5668163299560547, "learning_rate": 0.00014832204870922846, "loss": 0.8211, "step": 8970 }, { "epoch": 6.231126725708691e-14, "grad_norm": 1.6396591663360596, "learning_rate": 0.0001483172505938253, "loss": 0.8073, "step": 8980 }, { "epoch": 6.238065619612598e-14, "grad_norm": 1.9295185804367065, "learning_rate": 0.00014831244571467562, "loss": 0.8715, "step": 8990 }, { "epoch": 6.245004513516506e-14, "grad_norm": 1.4524710178375244, "learning_rate": 0.00014830763407227307, "loss": 0.7928, "step": 9000 }, { "epoch": 6.251943407420413e-14, "grad_norm": 1.7525362968444824, "learning_rate": 0.00014830281566711225, "loss": 0.7891, "step": 9010 }, { "epoch": 6.25888230132432e-14, "grad_norm": 1.560866355895996, "learning_rate": 0.00014829799049968824, "loss": 0.807, "step": 9020 }, { "epoch": 6.265821195228227e-14, "grad_norm": 1.782263159751892, "learning_rate": 0.00014829315857049695, "loss": 0.8413, "step": 9030 }, { "epoch": 6.272760089132134e-14, "grad_norm": 1.4566947221755981, "learning_rate": 0.00014828831988003495, "loss": 0.7792, "step": 9040 }, { "epoch": 6.279698983036042e-14, "grad_norm": 2.3462181091308594, "learning_rate": 0.00014828347442879942, "loss": 0.8696, "step": 9050 }, { "epoch": 6.286637876939949e-14, "grad_norm": 1.7256051301956177, "learning_rate": 0.00014827862221728836, "loss": 0.8268, "step": 9060 }, { "epoch": 6.293576770843856e-14, "grad_norm": 1.7342160940170288, "learning_rate": 0.00014827376324600037, "loss": 0.8716, "step": 9070 }, { "epoch": 6.300515664747763e-14, "grad_norm": 1.769508719444275, "learning_rate": 0.00014826889751543484, "loss": 0.8956, "step": 9080 }, { "epoch": 6.30745455865167e-14, "grad_norm": 2.0159718990325928, "learning_rate": 0.0001482640250260918, "loss": 0.7927, "step": 9090 }, { "epoch": 6.314393452555578e-14, "grad_norm": 2.0468387603759766, "learning_rate": 0.0001482591457784719, "loss": 0.8891, "step": 9100 }, { "epoch": 6.321332346459485e-14, "grad_norm": 2.018437623977661, "learning_rate": 0.00014825425977307662, "loss": 0.8314, "step": 9110 }, { "epoch": 6.328271240363392e-14, "grad_norm": 1.8290477991104126, "learning_rate": 0.00014824936701040805, "loss": 0.84, "step": 9120 }, { "epoch": 6.3352101342673e-14, "grad_norm": 1.5000249147415161, "learning_rate": 0.00014824446749096902, "loss": 0.8104, "step": 9130 }, { "epoch": 6.342149028171207e-14, "grad_norm": 1.7037287950515747, "learning_rate": 0.000148239561215263, "loss": 0.7642, "step": 9140 }, { "epoch": 6.349087922075114e-14, "grad_norm": 1.9698604345321655, "learning_rate": 0.00014823464818379423, "loss": 0.8456, "step": 9150 }, { "epoch": 6.356026815979021e-14, "grad_norm": 1.5439224243164062, "learning_rate": 0.00014822972839706754, "loss": 0.7799, "step": 9160 }, { "epoch": 6.362965709882928e-14, "grad_norm": 1.2217121124267578, "learning_rate": 0.00014822480185558854, "loss": 0.7906, "step": 9170 }, { "epoch": 6.369904603786836e-14, "grad_norm": 1.6923890113830566, "learning_rate": 0.00014821986855986354, "loss": 0.8205, "step": 9180 }, { "epoch": 6.376843497690743e-14, "grad_norm": 1.7000075578689575, "learning_rate": 0.00014821492851039948, "loss": 0.799, "step": 9190 }, { "epoch": 6.38378239159465e-14, "grad_norm": 1.8833069801330566, "learning_rate": 0.00014820998170770406, "loss": 0.8214, "step": 9200 }, { "epoch": 6.390721285498557e-14, "grad_norm": 1.8590736389160156, "learning_rate": 0.0001482050281522856, "loss": 0.8282, "step": 9210 }, { "epoch": 6.397660179402465e-14, "grad_norm": 1.8877638578414917, "learning_rate": 0.00014820006784465315, "loss": 0.8293, "step": 9220 }, { "epoch": 6.404599073306372e-14, "grad_norm": 1.8827332258224487, "learning_rate": 0.00014819510078531653, "loss": 0.8371, "step": 9230 }, { "epoch": 6.411537967210279e-14, "grad_norm": 1.8325026035308838, "learning_rate": 0.0001481901269747861, "loss": 0.86, "step": 9240 }, { "epoch": 6.418476861114186e-14, "grad_norm": 1.7891998291015625, "learning_rate": 0.00014818514641357303, "loss": 0.7821, "step": 9250 }, { "epoch": 6.425415755018093e-14, "grad_norm": 1.329018473625183, "learning_rate": 0.00014818015910218914, "loss": 0.8239, "step": 9260 }, { "epoch": 6.432354648922001e-14, "grad_norm": 1.48125422000885, "learning_rate": 0.000148175165041147, "loss": 0.8467, "step": 9270 }, { "epoch": 6.439293542825908e-14, "grad_norm": 2.1634836196899414, "learning_rate": 0.00014817016423095976, "loss": 0.7989, "step": 9280 }, { "epoch": 6.446232436729815e-14, "grad_norm": 1.9797488451004028, "learning_rate": 0.0001481651566721414, "loss": 0.7754, "step": 9290 }, { "epoch": 6.453171330633722e-14, "grad_norm": 1.9431921243667603, "learning_rate": 0.00014816014236520645, "loss": 0.8173, "step": 9300 }, { "epoch": 6.46011022453763e-14, "grad_norm": 1.7105804681777954, "learning_rate": 0.00014815512131067024, "loss": 0.7957, "step": 9310 }, { "epoch": 6.467049118441537e-14, "grad_norm": 1.6939976215362549, "learning_rate": 0.0001481500935090488, "loss": 0.7564, "step": 9320 }, { "epoch": 6.473988012345444e-14, "grad_norm": 1.6710238456726074, "learning_rate": 0.00014814505896085876, "loss": 0.7857, "step": 9330 }, { "epoch": 6.480926906249351e-14, "grad_norm": 1.5692813396453857, "learning_rate": 0.00014814001766661752, "loss": 0.8038, "step": 9340 }, { "epoch": 6.487865800153259e-14, "grad_norm": 1.6710907220840454, "learning_rate": 0.00014813496962684316, "loss": 0.7728, "step": 9350 }, { "epoch": 6.494804694057166e-14, "grad_norm": 1.8215038776397705, "learning_rate": 0.00014812991484205442, "loss": 0.7926, "step": 9360 }, { "epoch": 6.501743587961073e-14, "grad_norm": 1.6474415063858032, "learning_rate": 0.0001481248533127708, "loss": 0.8069, "step": 9370 }, { "epoch": 6.50868248186498e-14, "grad_norm": 1.793882966041565, "learning_rate": 0.0001481197850395124, "loss": 0.8361, "step": 9380 }, { "epoch": 6.515621375768887e-14, "grad_norm": 1.786407709121704, "learning_rate": 0.0001481147100228001, "loss": 0.8397, "step": 9390 }, { "epoch": 6.522560269672795e-14, "grad_norm": 1.5592249631881714, "learning_rate": 0.00014810962826315542, "loss": 0.7909, "step": 9400 }, { "epoch": 6.529499163576702e-14, "grad_norm": 1.7047032117843628, "learning_rate": 0.0001481045397611006, "loss": 0.7966, "step": 9410 }, { "epoch": 6.536438057480609e-14, "grad_norm": 1.4683407545089722, "learning_rate": 0.00014809944451715855, "loss": 0.7953, "step": 9420 }, { "epoch": 6.543376951384516e-14, "grad_norm": 1.5101654529571533, "learning_rate": 0.0001480943425318529, "loss": 0.8351, "step": 9430 }, { "epoch": 6.550315845288424e-14, "grad_norm": 1.4948140382766724, "learning_rate": 0.00014808923380570796, "loss": 0.7705, "step": 9440 }, { "epoch": 6.557254739192331e-14, "grad_norm": 1.973104476928711, "learning_rate": 0.0001480841183392487, "loss": 0.802, "step": 9450 }, { "epoch": 6.564193633096238e-14, "grad_norm": 1.5749156475067139, "learning_rate": 0.00014807899613300085, "loss": 0.8215, "step": 9460 }, { "epoch": 6.571132527000145e-14, "grad_norm": 1.4972069263458252, "learning_rate": 0.0001480738671874908, "loss": 0.7598, "step": 9470 }, { "epoch": 6.578071420904053e-14, "grad_norm": 1.7243767976760864, "learning_rate": 0.00014806873150324557, "loss": 0.7766, "step": 9480 }, { "epoch": 6.58501031480796e-14, "grad_norm": 1.7579342126846313, "learning_rate": 0.00014806358908079302, "loss": 0.7614, "step": 9490 }, { "epoch": 6.591949208711867e-14, "grad_norm": 1.6407419443130493, "learning_rate": 0.00014805843992066154, "loss": 0.7801, "step": 9500 }, { "epoch": 6.598888102615774e-14, "grad_norm": 1.802369475364685, "learning_rate": 0.00014805328402338032, "loss": 0.7559, "step": 9510 }, { "epoch": 6.605826996519681e-14, "grad_norm": 1.638789176940918, "learning_rate": 0.0001480481213894792, "loss": 0.8109, "step": 9520 }, { "epoch": 6.612765890423589e-14, "grad_norm": 1.8164503574371338, "learning_rate": 0.00014804295201948874, "loss": 0.7613, "step": 9530 }, { "epoch": 6.619704784327496e-14, "grad_norm": 1.733708381652832, "learning_rate": 0.00014803777591394015, "loss": 0.8117, "step": 9540 }, { "epoch": 6.626643678231403e-14, "grad_norm": 1.626444697380066, "learning_rate": 0.00014803259307336533, "loss": 0.8507, "step": 9550 }, { "epoch": 6.63358257213531e-14, "grad_norm": 1.6189097166061401, "learning_rate": 0.00014802740349829696, "loss": 0.7387, "step": 9560 }, { "epoch": 6.640521466039218e-14, "grad_norm": 1.8640772104263306, "learning_rate": 0.00014802220718926833, "loss": 0.8215, "step": 9570 }, { "epoch": 6.647460359943125e-14, "grad_norm": 1.8455644845962524, "learning_rate": 0.0001480170041468134, "loss": 0.808, "step": 9580 }, { "epoch": 6.654399253847032e-14, "grad_norm": 1.647828459739685, "learning_rate": 0.0001480117943714669, "loss": 0.7776, "step": 9590 }, { "epoch": 6.661338147750939e-14, "grad_norm": 1.6663093566894531, "learning_rate": 0.00014800657786376422, "loss": 0.7426, "step": 9600 }, { "epoch": 6.668277041654846e-14, "grad_norm": 1.93611741065979, "learning_rate": 0.00014800135462424144, "loss": 0.7631, "step": 9610 }, { "epoch": 6.675215935558754e-14, "grad_norm": 1.9894664287567139, "learning_rate": 0.00014799612465343532, "loss": 0.7877, "step": 9620 }, { "epoch": 6.682154829462661e-14, "grad_norm": 1.801867127418518, "learning_rate": 0.00014799088795188327, "loss": 0.7466, "step": 9630 }, { "epoch": 6.689093723366568e-14, "grad_norm": 1.4175209999084473, "learning_rate": 0.00014798564452012352, "loss": 0.7279, "step": 9640 }, { "epoch": 6.696032617270475e-14, "grad_norm": 1.9562599658966064, "learning_rate": 0.00014798039435869487, "loss": 0.8162, "step": 9650 }, { "epoch": 6.702971511174383e-14, "grad_norm": 1.8310068845748901, "learning_rate": 0.0001479751374681369, "loss": 0.7217, "step": 9660 }, { "epoch": 6.70991040507829e-14, "grad_norm": 1.4878566265106201, "learning_rate": 0.00014796987384898977, "loss": 0.7491, "step": 9670 }, { "epoch": 6.716849298982197e-14, "grad_norm": 1.736346960067749, "learning_rate": 0.00014796460350179446, "loss": 0.7864, "step": 9680 }, { "epoch": 6.723788192886104e-14, "grad_norm": 1.5252172946929932, "learning_rate": 0.00014795932642709255, "loss": 0.773, "step": 9690 }, { "epoch": 6.730727086790012e-14, "grad_norm": 1.5290720462799072, "learning_rate": 0.00014795404262542635, "loss": 0.7769, "step": 9700 }, { "epoch": 6.737665980693919e-14, "grad_norm": 1.5474748611450195, "learning_rate": 0.00014794875209733885, "loss": 0.8266, "step": 9710 }, { "epoch": 6.744604874597826e-14, "grad_norm": 1.8798251152038574, "learning_rate": 0.00014794345484337373, "loss": 0.7359, "step": 9720 }, { "epoch": 6.751543768501733e-14, "grad_norm": 2.217877149581909, "learning_rate": 0.00014793815086407537, "loss": 0.7766, "step": 9730 }, { "epoch": 6.75848266240564e-14, "grad_norm": 1.5790609121322632, "learning_rate": 0.00014793284015998884, "loss": 0.816, "step": 9740 }, { "epoch": 6.765421556309548e-14, "grad_norm": 1.736182689666748, "learning_rate": 0.0001479275227316599, "loss": 0.7556, "step": 9750 }, { "epoch": 6.772360450213455e-14, "grad_norm": 1.6081117391586304, "learning_rate": 0.00014792219857963495, "loss": 0.7365, "step": 9760 }, { "epoch": 6.779299344117362e-14, "grad_norm": 1.9723862409591675, "learning_rate": 0.00014791686770446123, "loss": 0.7769, "step": 9770 }, { "epoch": 6.786238238021269e-14, "grad_norm": 1.8546388149261475, "learning_rate": 0.0001479115301066865, "loss": 0.7717, "step": 9780 }, { "epoch": 6.793177131925177e-14, "grad_norm": 1.9275010824203491, "learning_rate": 0.00014790618578685926, "loss": 0.6977, "step": 9790 }, { "epoch": 6.800116025829084e-14, "grad_norm": 1.7664377689361572, "learning_rate": 0.00014790083474552877, "loss": 0.7455, "step": 9800 }, { "epoch": 6.807054919732991e-14, "grad_norm": 1.7937942743301392, "learning_rate": 0.0001478954769832449, "loss": 0.7707, "step": 9810 }, { "epoch": 6.813993813636898e-14, "grad_norm": 1.7742493152618408, "learning_rate": 0.00014789011250055832, "loss": 0.7185, "step": 9820 }, { "epoch": 6.820932707540805e-14, "grad_norm": 1.739514946937561, "learning_rate": 0.0001478847412980202, "loss": 0.7555, "step": 9830 }, { "epoch": 6.827871601444713e-14, "grad_norm": 1.9444005489349365, "learning_rate": 0.0001478793633761826, "loss": 0.7742, "step": 9840 }, { "epoch": 6.83481049534862e-14, "grad_norm": 1.4878854751586914, "learning_rate": 0.00014787397873559813, "loss": 0.7743, "step": 9850 }, { "epoch": 6.841749389252527e-14, "grad_norm": 1.7679541110992432, "learning_rate": 0.0001478685873768202, "loss": 0.7968, "step": 9860 }, { "epoch": 6.848688283156434e-14, "grad_norm": 1.5802594423294067, "learning_rate": 0.0001478631893004028, "loss": 0.7265, "step": 9870 }, { "epoch": 6.855627177060342e-14, "grad_norm": 1.7941960096359253, "learning_rate": 0.0001478577845069007, "loss": 0.7736, "step": 9880 }, { "epoch": 6.862566070964249e-14, "grad_norm": 1.8532254695892334, "learning_rate": 0.00014785237299686933, "loss": 0.7642, "step": 9890 }, { "epoch": 6.869504964868156e-14, "grad_norm": 1.8797340393066406, "learning_rate": 0.00014784695477086478, "loss": 0.7867, "step": 9900 }, { "epoch": 6.876443858772063e-14, "grad_norm": 2.057265520095825, "learning_rate": 0.00014784152982944386, "loss": 0.738, "step": 9910 }, { "epoch": 6.88338275267597e-14, "grad_norm": 1.540207862854004, "learning_rate": 0.0001478360981731641, "loss": 0.6966, "step": 9920 }, { "epoch": 6.890321646579878e-14, "grad_norm": 1.514382243156433, "learning_rate": 0.0001478306598025837, "loss": 0.7781, "step": 9930 }, { "epoch": 6.897260540483785e-14, "grad_norm": 1.622616171836853, "learning_rate": 0.00014782521471826148, "loss": 0.742, "step": 9940 }, { "epoch": 6.904199434387692e-14, "grad_norm": 1.7631787061691284, "learning_rate": 0.000147819762920757, "loss": 0.7446, "step": 9950 }, { "epoch": 6.9111383282916e-14, "grad_norm": 1.9986532926559448, "learning_rate": 0.0001478143044106306, "loss": 0.741, "step": 9960 }, { "epoch": 6.918077222195507e-14, "grad_norm": 1.9675596952438354, "learning_rate": 0.00014780883918844316, "loss": 0.7377, "step": 9970 }, { "epoch": 6.925016116099414e-14, "grad_norm": 1.850937843322754, "learning_rate": 0.00014780336725475632, "loss": 0.7487, "step": 9980 }, { "epoch": 6.931955010003321e-14, "grad_norm": 1.787003517150879, "learning_rate": 0.0001477978886101324, "loss": 0.7659, "step": 9990 }, { "epoch": 6.938893903907228e-14, "grad_norm": 1.872416615486145, "learning_rate": 0.00014779240325513447, "loss": 0.7499, "step": 10000 }, { "epoch": 6.945832797811136e-14, "grad_norm": 1.6789631843566895, "learning_rate": 0.0001477869111903262, "loss": 0.7594, "step": 10010 }, { "epoch": 6.952771691715043e-14, "grad_norm": 2.0601003170013428, "learning_rate": 0.00014778141241627196, "loss": 0.8133, "step": 10020 }, { "epoch": 6.95971058561895e-14, "grad_norm": 1.6647976636886597, "learning_rate": 0.0001477759069335369, "loss": 0.7071, "step": 10030 }, { "epoch": 6.966649479522857e-14, "grad_norm": 1.5113708972930908, "learning_rate": 0.00014777039474268673, "loss": 0.7791, "step": 10040 }, { "epoch": 6.973588373426765e-14, "grad_norm": 1.467720627784729, "learning_rate": 0.00014776487584428793, "loss": 0.7802, "step": 10050 }, { "epoch": 6.980527267330672e-14, "grad_norm": 1.5991625785827637, "learning_rate": 0.00014775935023890767, "loss": 0.7127, "step": 10060 }, { "epoch": 6.987466161234579e-14, "grad_norm": 1.7860878705978394, "learning_rate": 0.00014775381792711377, "loss": 0.7111, "step": 10070 }, { "epoch": 6.994405055138486e-14, "grad_norm": 1.9571532011032104, "learning_rate": 0.0001477482789094748, "loss": 0.7474, "step": 10080 }, { "epoch": 7.001343949042393e-14, "grad_norm": 1.6839089393615723, "learning_rate": 0.0001477427331865599, "loss": 0.7294, "step": 10090 }, { "epoch": 7.008282842946301e-14, "grad_norm": 1.7400212287902832, "learning_rate": 0.0001477371807589391, "loss": 0.7027, "step": 10100 }, { "epoch": 7.015221736850208e-14, "grad_norm": 1.7809795141220093, "learning_rate": 0.0001477316216271829, "loss": 0.7198, "step": 10110 }, { "epoch": 7.022160630754115e-14, "grad_norm": 1.627596139907837, "learning_rate": 0.0001477260557918626, "loss": 0.7787, "step": 10120 }, { "epoch": 7.029099524658022e-14, "grad_norm": 1.5115605592727661, "learning_rate": 0.00014772048325355022, "loss": 0.6899, "step": 10130 }, { "epoch": 7.03603841856193e-14, "grad_norm": 1.6498656272888184, "learning_rate": 0.0001477149040128184, "loss": 0.711, "step": 10140 }, { "epoch": 7.042977312465837e-14, "grad_norm": 1.5264075994491577, "learning_rate": 0.0001477093180702405, "loss": 0.7574, "step": 10150 }, { "epoch": 7.049916206369744e-14, "grad_norm": 1.7721933126449585, "learning_rate": 0.0001477037254263905, "loss": 0.7585, "step": 10160 }, { "epoch": 7.056855100273651e-14, "grad_norm": 2.1747548580169678, "learning_rate": 0.00014769812608184327, "loss": 0.7648, "step": 10170 }, { "epoch": 7.063793994177558e-14, "grad_norm": 1.644742727279663, "learning_rate": 0.0001476925200371741, "loss": 0.7595, "step": 10180 }, { "epoch": 7.070732888081466e-14, "grad_norm": 1.6708656549453735, "learning_rate": 0.00014768690729295914, "loss": 0.7117, "step": 10190 }, { "epoch": 7.077671781985373e-14, "grad_norm": 1.6205227375030518, "learning_rate": 0.00014768128784977522, "loss": 0.7375, "step": 10200 }, { "epoch": 7.08461067588928e-14, "grad_norm": 1.6445764303207397, "learning_rate": 0.0001476756617081998, "loss": 0.7575, "step": 10210 }, { "epoch": 7.091549569793187e-14, "grad_norm": 1.635176420211792, "learning_rate": 0.00014767002886881102, "loss": 0.7036, "step": 10220 }, { "epoch": 7.098488463697095e-14, "grad_norm": 1.6106767654418945, "learning_rate": 0.00014766438933218776, "loss": 0.7023, "step": 10230 }, { "epoch": 7.105427357601002e-14, "grad_norm": 1.7170623540878296, "learning_rate": 0.0001476587430989096, "loss": 0.7491, "step": 10240 }, { "epoch": 7.112366251504909e-14, "grad_norm": 1.6238926649093628, "learning_rate": 0.00014765309016955678, "loss": 0.6832, "step": 10250 }, { "epoch": 7.119305145408816e-14, "grad_norm": 1.8757878541946411, "learning_rate": 0.0001476474305447102, "loss": 0.6803, "step": 10260 }, { "epoch": 7.126244039312724e-14, "grad_norm": 1.4947154521942139, "learning_rate": 0.00014764176422495147, "loss": 0.6941, "step": 10270 }, { "epoch": 7.133182933216631e-14, "grad_norm": 1.622617244720459, "learning_rate": 0.00014763609121086293, "loss": 0.7277, "step": 10280 }, { "epoch": 7.140121827120538e-14, "grad_norm": 1.510399341583252, "learning_rate": 0.00014763041150302753, "loss": 0.6754, "step": 10290 }, { "epoch": 7.147060721024445e-14, "grad_norm": 2.0051794052124023, "learning_rate": 0.00014762472510202894, "loss": 0.7441, "step": 10300 }, { "epoch": 7.153999614928352e-14, "grad_norm": 1.7692116498947144, "learning_rate": 0.0001476190320084516, "loss": 0.7064, "step": 10310 }, { "epoch": 7.16093850883226e-14, "grad_norm": 1.8945611715316772, "learning_rate": 0.00014761333222288047, "loss": 0.7378, "step": 10320 }, { "epoch": 7.167877402736167e-14, "grad_norm": 2.0911824703216553, "learning_rate": 0.00014760762574590137, "loss": 0.7898, "step": 10330 }, { "epoch": 7.174816296640074e-14, "grad_norm": 1.8985759019851685, "learning_rate": 0.00014760191257810065, "loss": 0.6961, "step": 10340 }, { "epoch": 7.181755190543981e-14, "grad_norm": 2.2164368629455566, "learning_rate": 0.0001475961927200655, "loss": 0.7633, "step": 10350 }, { "epoch": 7.188694084447889e-14, "grad_norm": 1.9855022430419922, "learning_rate": 0.00014759046617238372, "loss": 0.7081, "step": 10360 }, { "epoch": 7.195632978351796e-14, "grad_norm": 1.7786694765090942, "learning_rate": 0.00014758473293564374, "loss": 0.7092, "step": 10370 }, { "epoch": 7.202571872255703e-14, "grad_norm": 1.885247826576233, "learning_rate": 0.00014757899301043477, "loss": 0.7077, "step": 10380 }, { "epoch": 7.20951076615961e-14, "grad_norm": 1.6464756727218628, "learning_rate": 0.00014757324639734672, "loss": 0.6544, "step": 10390 }, { "epoch": 7.216449660063518e-14, "grad_norm": 1.935860276222229, "learning_rate": 0.00014756749309697008, "loss": 0.73, "step": 10400 }, { "epoch": 7.223388553967425e-14, "grad_norm": 1.6753548383712769, "learning_rate": 0.00014756173310989613, "loss": 0.6712, "step": 10410 }, { "epoch": 7.230327447871332e-14, "grad_norm": 1.3094464540481567, "learning_rate": 0.00014755596643671677, "loss": 0.6892, "step": 10420 }, { "epoch": 7.237266341775239e-14, "grad_norm": 1.612871766090393, "learning_rate": 0.0001475501930780247, "loss": 0.7104, "step": 10430 }, { "epoch": 7.244205235679146e-14, "grad_norm": 1.7919875383377075, "learning_rate": 0.00014754441303441306, "loss": 0.6879, "step": 10440 }, { "epoch": 7.251144129583054e-14, "grad_norm": 1.625044584274292, "learning_rate": 0.00014753862630647598, "loss": 0.7, "step": 10450 }, { "epoch": 7.258083023486961e-14, "grad_norm": 1.8476080894470215, "learning_rate": 0.0001475328328948081, "loss": 0.7029, "step": 10460 }, { "epoch": 7.265021917390868e-14, "grad_norm": 1.787153959274292, "learning_rate": 0.00014752703280000476, "loss": 0.7146, "step": 10470 }, { "epoch": 7.271960811294775e-14, "grad_norm": 1.8826532363891602, "learning_rate": 0.000147521226022662, "loss": 0.7071, "step": 10480 }, { "epoch": 7.278899705198683e-14, "grad_norm": 1.9152928590774536, "learning_rate": 0.0001475154125633766, "loss": 0.7585, "step": 10490 }, { "epoch": 7.28583859910259e-14, "grad_norm": 1.8734873533248901, "learning_rate": 0.000147509592422746, "loss": 0.6797, "step": 10500 }, { "epoch": 7.292777493006497e-14, "grad_norm": 1.5387792587280273, "learning_rate": 0.00014750376560136825, "loss": 0.7202, "step": 10510 }, { "epoch": 7.299716386910404e-14, "grad_norm": 1.607237696647644, "learning_rate": 0.00014749793209984215, "loss": 0.7386, "step": 10520 }, { "epoch": 7.306655280814311e-14, "grad_norm": 1.9958539009094238, "learning_rate": 0.0001474920919187672, "loss": 0.6777, "step": 10530 }, { "epoch": 7.313594174718219e-14, "grad_norm": 2.0275650024414062, "learning_rate": 0.0001474862450587436, "loss": 0.6777, "step": 10540 }, { "epoch": 7.320533068622126e-14, "grad_norm": 1.6118521690368652, "learning_rate": 0.00014748039152037218, "loss": 0.6891, "step": 10550 }, { "epoch": 7.327471962526033e-14, "grad_norm": 1.908440351486206, "learning_rate": 0.00014747453130425448, "loss": 0.6893, "step": 10560 }, { "epoch": 7.33441085642994e-14, "grad_norm": 1.8559142351150513, "learning_rate": 0.00014746866441099269, "loss": 0.6464, "step": 10570 }, { "epoch": 7.341349750333848e-14, "grad_norm": 2.0115582942962646, "learning_rate": 0.0001474627908411898, "loss": 0.6926, "step": 10580 }, { "epoch": 7.348288644237755e-14, "grad_norm": 1.6273860931396484, "learning_rate": 0.00014745691059544938, "loss": 0.6971, "step": 10590 }, { "epoch": 7.355227538141662e-14, "grad_norm": 1.807313084602356, "learning_rate": 0.0001474510236743757, "loss": 0.6693, "step": 10600 }, { "epoch": 7.362166432045569e-14, "grad_norm": 1.9227155447006226, "learning_rate": 0.0001474451300785737, "loss": 0.6823, "step": 10610 }, { "epoch": 7.369105325949477e-14, "grad_norm": 1.9558216333389282, "learning_rate": 0.00014743922980864916, "loss": 0.7082, "step": 10620 }, { "epoch": 7.376044219853384e-14, "grad_norm": 1.6356412172317505, "learning_rate": 0.0001474333228652083, "loss": 0.7277, "step": 10630 }, { "epoch": 7.382983113757291e-14, "grad_norm": 1.7641595602035522, "learning_rate": 0.00014742740924885821, "loss": 0.721, "step": 10640 }, { "epoch": 7.389922007661198e-14, "grad_norm": 1.8435031175613403, "learning_rate": 0.0001474214889602066, "loss": 0.7095, "step": 10650 }, { "epoch": 7.396860901565105e-14, "grad_norm": 1.7197883129119873, "learning_rate": 0.00014741556199986185, "loss": 0.7005, "step": 10660 }, { "epoch": 7.403799795469013e-14, "grad_norm": 1.5980031490325928, "learning_rate": 0.0001474096283684331, "loss": 0.6903, "step": 10670 }, { "epoch": 7.41073868937292e-14, "grad_norm": 1.3973413705825806, "learning_rate": 0.00014740368806653005, "loss": 0.7149, "step": 10680 }, { "epoch": 7.417677583276827e-14, "grad_norm": 1.5934303998947144, "learning_rate": 0.00014739774109476318, "loss": 0.6548, "step": 10690 }, { "epoch": 7.424616477180734e-14, "grad_norm": 1.7760168313980103, "learning_rate": 0.00014739178745374368, "loss": 0.6933, "step": 10700 }, { "epoch": 7.431555371084642e-14, "grad_norm": 1.46489679813385, "learning_rate": 0.00014738582714408333, "loss": 0.6898, "step": 10710 }, { "epoch": 7.438494264988549e-14, "grad_norm": 1.5496094226837158, "learning_rate": 0.00014737986016639465, "loss": 0.6673, "step": 10720 }, { "epoch": 7.445433158892456e-14, "grad_norm": 1.8729746341705322, "learning_rate": 0.0001473738865212909, "loss": 0.6853, "step": 10730 }, { "epoch": 7.452372052796363e-14, "grad_norm": 1.7420719861984253, "learning_rate": 0.00014736790620938588, "loss": 0.672, "step": 10740 }, { "epoch": 7.45931094670027e-14, "grad_norm": 1.6780097484588623, "learning_rate": 0.0001473619192312942, "loss": 0.6688, "step": 10750 }, { "epoch": 7.466249840604178e-14, "grad_norm": 1.8313485383987427, "learning_rate": 0.00014735592558763112, "loss": 0.6629, "step": 10760 }, { "epoch": 7.473188734508085e-14, "grad_norm": 1.9416035413742065, "learning_rate": 0.00014734992527901258, "loss": 0.7099, "step": 10770 }, { "epoch": 7.480127628411992e-14, "grad_norm": 2.0532593727111816, "learning_rate": 0.00014734391830605517, "loss": 0.6853, "step": 10780 }, { "epoch": 7.4870665223159e-14, "grad_norm": 1.7901287078857422, "learning_rate": 0.00014733790466937627, "loss": 0.7008, "step": 10790 }, { "epoch": 7.494005416219807e-14, "grad_norm": 1.6691807508468628, "learning_rate": 0.00014733188436959378, "loss": 0.7473, "step": 10800 }, { "epoch": 7.500944310123714e-14, "grad_norm": 1.955786108970642, "learning_rate": 0.00014732585740732644, "loss": 0.6732, "step": 10810 }, { "epoch": 7.507883204027621e-14, "grad_norm": 1.7100886106491089, "learning_rate": 0.00014731982378319363, "loss": 0.728, "step": 10820 }, { "epoch": 7.514822097931528e-14, "grad_norm": 2.005129098892212, "learning_rate": 0.0001473137834978154, "loss": 0.6931, "step": 10830 }, { "epoch": 7.521760991835436e-14, "grad_norm": 1.6403708457946777, "learning_rate": 0.0001473077365518124, "loss": 0.6491, "step": 10840 }, { "epoch": 7.528699885739343e-14, "grad_norm": 1.673646330833435, "learning_rate": 0.00014730168294580613, "loss": 0.6403, "step": 10850 }, { "epoch": 7.53563877964325e-14, "grad_norm": 2.026262044906616, "learning_rate": 0.00014729562268041868, "loss": 0.6572, "step": 10860 }, { "epoch": 7.542577673547157e-14, "grad_norm": 1.9117258787155151, "learning_rate": 0.0001472895557562728, "loss": 0.6681, "step": 10870 }, { "epoch": 7.549516567451064e-14, "grad_norm": 1.997230887413025, "learning_rate": 0.00014728348217399198, "loss": 0.6648, "step": 10880 }, { "epoch": 7.556455461354972e-14, "grad_norm": 1.539120078086853, "learning_rate": 0.00014727740193420043, "loss": 0.6561, "step": 10890 }, { "epoch": 7.563394355258879e-14, "grad_norm": 1.915386438369751, "learning_rate": 0.0001472713150375229, "loss": 0.7189, "step": 10900 }, { "epoch": 7.570333249162786e-14, "grad_norm": 1.8230383396148682, "learning_rate": 0.00014726522148458495, "loss": 0.6131, "step": 10910 }, { "epoch": 7.577272143066693e-14, "grad_norm": 1.5482615232467651, "learning_rate": 0.00014725912127601279, "loss": 0.6596, "step": 10920 }, { "epoch": 7.5842110369706e-14, "grad_norm": 1.6101040840148926, "learning_rate": 0.00014725301441243332, "loss": 0.6866, "step": 10930 }, { "epoch": 7.591149930874508e-14, "grad_norm": 1.7994129657745361, "learning_rate": 0.0001472469008944741, "loss": 0.7104, "step": 10940 }, { "epoch": 7.598088824778415e-14, "grad_norm": 1.6004949808120728, "learning_rate": 0.0001472407807227634, "loss": 0.6741, "step": 10950 }, { "epoch": 7.605027718682322e-14, "grad_norm": 1.8204092979431152, "learning_rate": 0.00014723465389793018, "loss": 0.7445, "step": 10960 }, { "epoch": 7.61196661258623e-14, "grad_norm": 1.917456865310669, "learning_rate": 0.00014722852042060403, "loss": 0.6759, "step": 10970 }, { "epoch": 7.618905506490137e-14, "grad_norm": 1.891422986984253, "learning_rate": 0.00014722238029141527, "loss": 0.6786, "step": 10980 }, { "epoch": 7.625844400394044e-14, "grad_norm": 1.6179174184799194, "learning_rate": 0.0001472162335109949, "loss": 0.6975, "step": 10990 }, { "epoch": 7.632783294297951e-14, "grad_norm": 1.9470535516738892, "learning_rate": 0.00014721008007997457, "loss": 0.6433, "step": 11000 }, { "epoch": 7.639722188201858e-14, "grad_norm": 1.9209139347076416, "learning_rate": 0.0001472039199989867, "loss": 0.6445, "step": 11010 }, { "epoch": 7.646661082105766e-14, "grad_norm": 1.8959945440292358, "learning_rate": 0.0001471977532686643, "loss": 0.655, "step": 11020 }, { "epoch": 7.653599976009673e-14, "grad_norm": 2.085935115814209, "learning_rate": 0.0001471915798896411, "loss": 0.6665, "step": 11030 }, { "epoch": 7.66053886991358e-14, "grad_norm": 1.9947798252105713, "learning_rate": 0.0001471853998625515, "loss": 0.6157, "step": 11040 }, { "epoch": 7.667477763817487e-14, "grad_norm": 1.7950530052185059, "learning_rate": 0.0001471792131880306, "loss": 0.6697, "step": 11050 }, { "epoch": 7.674416657721395e-14, "grad_norm": 1.9456112384796143, "learning_rate": 0.00014717301986671419, "loss": 0.6532, "step": 11060 }, { "epoch": 7.681355551625302e-14, "grad_norm": 1.823869228363037, "learning_rate": 0.0001471668198992387, "loss": 0.6238, "step": 11070 }, { "epoch": 7.688294445529209e-14, "grad_norm": 1.6623897552490234, "learning_rate": 0.00014716061328624134, "loss": 0.6458, "step": 11080 }, { "epoch": 7.695233339433116e-14, "grad_norm": 2.069861888885498, "learning_rate": 0.00014715440002835984, "loss": 0.6865, "step": 11090 }, { "epoch": 7.702172233337023e-14, "grad_norm": 1.5409854650497437, "learning_rate": 0.00014714818012623277, "loss": 0.6925, "step": 11100 }, { "epoch": 7.709111127240931e-14, "grad_norm": 1.8930684328079224, "learning_rate": 0.00014714195358049936, "loss": 0.701, "step": 11110 }, { "epoch": 7.716050021144838e-14, "grad_norm": 1.7155686616897583, "learning_rate": 0.0001471357203917994, "loss": 0.6526, "step": 11120 }, { "epoch": 7.722988915048745e-14, "grad_norm": 1.950844645500183, "learning_rate": 0.00014712948056077346, "loss": 0.7039, "step": 11130 }, { "epoch": 7.729927808952652e-14, "grad_norm": 1.9521020650863647, "learning_rate": 0.00014712323408806282, "loss": 0.6636, "step": 11140 }, { "epoch": 7.73686670285656e-14, "grad_norm": 1.7835694551467896, "learning_rate": 0.00014711698097430938, "loss": 0.6761, "step": 11150 }, { "epoch": 7.743805596760467e-14, "grad_norm": 1.9458069801330566, "learning_rate": 0.00014711072122015577, "loss": 0.6367, "step": 11160 }, { "epoch": 7.750744490664374e-14, "grad_norm": 1.5616997480392456, "learning_rate": 0.00014710445482624522, "loss": 0.5997, "step": 11170 }, { "epoch": 7.757683384568281e-14, "grad_norm": 1.327799677848816, "learning_rate": 0.00014709818179322177, "loss": 0.6468, "step": 11180 }, { "epoch": 7.764622278472189e-14, "grad_norm": 1.7982299327850342, "learning_rate": 0.00014709190212173004, "loss": 0.6468, "step": 11190 }, { "epoch": 7.771561172376096e-14, "grad_norm": 1.9547935724258423, "learning_rate": 0.0001470856158124153, "loss": 0.6575, "step": 11200 }, { "epoch": 7.778500066280003e-14, "grad_norm": 1.7772589921951294, "learning_rate": 0.00014707932286592369, "loss": 0.6412, "step": 11210 }, { "epoch": 7.78543896018391e-14, "grad_norm": 1.9637423753738403, "learning_rate": 0.00014707302328290184, "loss": 0.6288, "step": 11220 }, { "epoch": 7.792377854087817e-14, "grad_norm": 1.9232382774353027, "learning_rate": 0.00014706671706399713, "loss": 0.6577, "step": 11230 }, { "epoch": 7.799316747991725e-14, "grad_norm": 1.7557636499404907, "learning_rate": 0.0001470604042098576, "loss": 0.6587, "step": 11240 }, { "epoch": 7.806255641895632e-14, "grad_norm": 1.9909387826919556, "learning_rate": 0.00014705408472113205, "loss": 0.6651, "step": 11250 }, { "epoch": 7.813194535799539e-14, "grad_norm": 1.7129284143447876, "learning_rate": 0.00014704775859846988, "loss": 0.6546, "step": 11260 }, { "epoch": 7.820133429703446e-14, "grad_norm": 1.8904848098754883, "learning_rate": 0.00014704142584252118, "loss": 0.7052, "step": 11270 }, { "epoch": 7.827072323607354e-14, "grad_norm": 1.7734811305999756, "learning_rate": 0.00014703508645393676, "loss": 0.6347, "step": 11280 }, { "epoch": 7.834011217511261e-14, "grad_norm": 1.8089745044708252, "learning_rate": 0.0001470287404333681, "loss": 0.7031, "step": 11290 }, { "epoch": 7.840950111415168e-14, "grad_norm": 1.7830053567886353, "learning_rate": 0.00014702238778146733, "loss": 0.6875, "step": 11300 }, { "epoch": 7.847889005319075e-14, "grad_norm": 1.549267292022705, "learning_rate": 0.00014701602849888725, "loss": 0.6157, "step": 11310 }, { "epoch": 7.854827899222983e-14, "grad_norm": 1.7390127182006836, "learning_rate": 0.00014700966258628147, "loss": 0.6191, "step": 11320 }, { "epoch": 7.86176679312689e-14, "grad_norm": 2.1265435218811035, "learning_rate": 0.0001470032900443041, "loss": 0.6505, "step": 11330 }, { "epoch": 7.868705687030797e-14, "grad_norm": 1.6828279495239258, "learning_rate": 0.00014699691087361009, "loss": 0.6051, "step": 11340 }, { "epoch": 7.875644580934704e-14, "grad_norm": 1.5206729173660278, "learning_rate": 0.0001469905250748549, "loss": 0.6111, "step": 11350 }, { "epoch": 7.882583474838611e-14, "grad_norm": 1.9147469997406006, "learning_rate": 0.00014698413264869489, "loss": 0.6318, "step": 11360 }, { "epoch": 7.889522368742519e-14, "grad_norm": 1.624571442604065, "learning_rate": 0.0001469777335957869, "loss": 0.6777, "step": 11370 }, { "epoch": 7.896461262646426e-14, "grad_norm": 1.9383827447891235, "learning_rate": 0.00014697132791678853, "loss": 0.6104, "step": 11380 }, { "epoch": 7.903400156550333e-14, "grad_norm": 1.52640962600708, "learning_rate": 0.00014696491561235812, "loss": 0.633, "step": 11390 }, { "epoch": 7.91033905045424e-14, "grad_norm": 1.6418941020965576, "learning_rate": 0.0001469584966831546, "loss": 0.6722, "step": 11400 }, { "epoch": 7.917277944358148e-14, "grad_norm": 1.9073678255081177, "learning_rate": 0.0001469520711298376, "loss": 0.6883, "step": 11410 }, { "epoch": 7.924216838262055e-14, "grad_norm": 2.0333821773529053, "learning_rate": 0.00014694563895306745, "loss": 0.6445, "step": 11420 }, { "epoch": 7.931155732165962e-14, "grad_norm": 1.5806773900985718, "learning_rate": 0.0001469392001535052, "loss": 0.6563, "step": 11430 }, { "epoch": 7.938094626069869e-14, "grad_norm": 1.8054637908935547, "learning_rate": 0.00014693275473181248, "loss": 0.6524, "step": 11440 }, { "epoch": 7.945033519973776e-14, "grad_norm": 1.7205920219421387, "learning_rate": 0.0001469263026886517, "loss": 0.66, "step": 11450 }, { "epoch": 7.951972413877684e-14, "grad_norm": 2.0001189708709717, "learning_rate": 0.00014691984402468588, "loss": 0.6876, "step": 11460 }, { "epoch": 7.958911307781591e-14, "grad_norm": 1.5387972593307495, "learning_rate": 0.00014691337874057876, "loss": 0.6169, "step": 11470 }, { "epoch": 7.965850201685498e-14, "grad_norm": 1.7264119386672974, "learning_rate": 0.00014690690683699474, "loss": 0.6135, "step": 11480 }, { "epoch": 7.972789095589405e-14, "grad_norm": 1.814969778060913, "learning_rate": 0.0001469004283145989, "loss": 0.6494, "step": 11490 }, { "epoch": 7.979727989493313e-14, "grad_norm": 1.5978071689605713, "learning_rate": 0.00014689394317405705, "loss": 0.5904, "step": 11500 }, { "epoch": 7.98666688339722e-14, "grad_norm": 1.9473140239715576, "learning_rate": 0.0001468874514160356, "loss": 0.6001, "step": 11510 }, { "epoch": 7.993605777301127e-14, "grad_norm": 1.727342963218689, "learning_rate": 0.0001468809530412017, "loss": 0.6257, "step": 11520 }, { "epoch": 8.000544671205034e-14, "grad_norm": 1.585967779159546, "learning_rate": 0.00014687444805022316, "loss": 0.6401, "step": 11530 }, { "epoch": 8.007483565108942e-14, "grad_norm": 1.7930288314819336, "learning_rate": 0.00014686793644376844, "loss": 0.6096, "step": 11540 }, { "epoch": 8.014422459012849e-14, "grad_norm": 2.2578845024108887, "learning_rate": 0.00014686141822250676, "loss": 0.6616, "step": 11550 }, { "epoch": 8.021361352916756e-14, "grad_norm": 1.8147739171981812, "learning_rate": 0.0001468548933871079, "loss": 0.6301, "step": 11560 }, { "epoch": 8.028300246820663e-14, "grad_norm": 1.8474198579788208, "learning_rate": 0.00014684836193824244, "loss": 0.6781, "step": 11570 }, { "epoch": 8.03523914072457e-14, "grad_norm": 1.8720064163208008, "learning_rate": 0.00014684182387658155, "loss": 0.6732, "step": 11580 }, { "epoch": 8.042178034628478e-14, "grad_norm": 1.832761526107788, "learning_rate": 0.0001468352792027972, "loss": 0.609, "step": 11590 }, { "epoch": 8.049116928532385e-14, "grad_norm": 2.0308077335357666, "learning_rate": 0.00014682872791756186, "loss": 0.6823, "step": 11600 }, { "epoch": 8.056055822436292e-14, "grad_norm": 2.0721590518951416, "learning_rate": 0.00014682217002154885, "loss": 0.6367, "step": 11610 }, { "epoch": 8.0629947163402e-14, "grad_norm": 1.991287112236023, "learning_rate": 0.00014681560551543202, "loss": 0.6421, "step": 11620 }, { "epoch": 8.069933610244107e-14, "grad_norm": 1.7575279474258423, "learning_rate": 0.00014680903439988604, "loss": 0.6133, "step": 11630 }, { "epoch": 8.076872504148014e-14, "grad_norm": 1.748283863067627, "learning_rate": 0.00014680245667558615, "loss": 0.5759, "step": 11640 }, { "epoch": 8.083811398051921e-14, "grad_norm": 1.9124258756637573, "learning_rate": 0.00014679587234320835, "loss": 0.6326, "step": 11650 }, { "epoch": 8.090750291955828e-14, "grad_norm": 1.6935774087905884, "learning_rate": 0.00014678928140342927, "loss": 0.5901, "step": 11660 }, { "epoch": 8.097689185859736e-14, "grad_norm": 1.4243029356002808, "learning_rate": 0.00014678268385692622, "loss": 0.6068, "step": 11670 }, { "epoch": 8.104628079763643e-14, "grad_norm": 1.9324419498443604, "learning_rate": 0.0001467760797043772, "loss": 0.6389, "step": 11680 }, { "epoch": 8.11156697366755e-14, "grad_norm": 1.7607519626617432, "learning_rate": 0.00014676946894646094, "loss": 0.6124, "step": 11690 }, { "epoch": 8.118505867571457e-14, "grad_norm": 1.9704097509384155, "learning_rate": 0.00014676285158385672, "loss": 0.6377, "step": 11700 }, { "epoch": 8.125444761475364e-14, "grad_norm": 1.7701810598373413, "learning_rate": 0.00014675622761724463, "loss": 0.6126, "step": 11710 }, { "epoch": 8.132383655379272e-14, "grad_norm": 2.048444986343384, "learning_rate": 0.00014674959704730537, "loss": 0.6481, "step": 11720 }, { "epoch": 8.139322549283179e-14, "grad_norm": 1.875495433807373, "learning_rate": 0.0001467429598747203, "loss": 0.6435, "step": 11730 }, { "epoch": 8.146261443187086e-14, "grad_norm": 1.9382944107055664, "learning_rate": 0.00014673631610017158, "loss": 0.6617, "step": 11740 }, { "epoch": 8.153200337090993e-14, "grad_norm": 2.1489415168762207, "learning_rate": 0.00014672966572434185, "loss": 0.6027, "step": 11750 }, { "epoch": 8.1601392309949e-14, "grad_norm": 1.538514494895935, "learning_rate": 0.00014672300874791464, "loss": 0.6443, "step": 11760 }, { "epoch": 8.167078124898808e-14, "grad_norm": 1.7952110767364502, "learning_rate": 0.000146716345171574, "loss": 0.6639, "step": 11770 }, { "epoch": 8.174017018802715e-14, "grad_norm": 1.826385498046875, "learning_rate": 0.00014670967499600472, "loss": 0.6023, "step": 11780 }, { "epoch": 8.180955912706622e-14, "grad_norm": 2.0488674640655518, "learning_rate": 0.0001467029982218923, "loss": 0.5913, "step": 11790 }, { "epoch": 8.18789480661053e-14, "grad_norm": 1.6371275186538696, "learning_rate": 0.00014669631484992286, "loss": 0.6189, "step": 11800 }, { "epoch": 8.194833700514437e-14, "grad_norm": 1.7752704620361328, "learning_rate": 0.00014668962488078322, "loss": 0.5913, "step": 11810 }, { "epoch": 8.201772594418344e-14, "grad_norm": 1.7896560430526733, "learning_rate": 0.00014668292831516083, "loss": 0.5581, "step": 11820 }, { "epoch": 8.208711488322251e-14, "grad_norm": 1.8518203496932983, "learning_rate": 0.00014667622515374395, "loss": 0.6112, "step": 11830 }, { "epoch": 8.215650382226158e-14, "grad_norm": 1.625461459159851, "learning_rate": 0.00014666951539722142, "loss": 0.6191, "step": 11840 }, { "epoch": 8.222589276130066e-14, "grad_norm": 1.9410938024520874, "learning_rate": 0.0001466627990462827, "loss": 0.6015, "step": 11850 }, { "epoch": 8.229528170033973e-14, "grad_norm": 1.7591253519058228, "learning_rate": 0.00014665607610161806, "loss": 0.5999, "step": 11860 }, { "epoch": 8.23646706393788e-14, "grad_norm": 1.7223809957504272, "learning_rate": 0.0001466493465639184, "loss": 0.6025, "step": 11870 }, { "epoch": 8.243405957841787e-14, "grad_norm": 1.984139084815979, "learning_rate": 0.00014664261043387526, "loss": 0.6603, "step": 11880 }, { "epoch": 8.250344851745695e-14, "grad_norm": 1.8351627588272095, "learning_rate": 0.00014663586771218086, "loss": 0.6409, "step": 11890 }, { "epoch": 8.257283745649602e-14, "grad_norm": 1.6128699779510498, "learning_rate": 0.00014662911839952815, "loss": 0.613, "step": 11900 }, { "epoch": 8.264222639553509e-14, "grad_norm": 1.9926910400390625, "learning_rate": 0.00014662236249661072, "loss": 0.6278, "step": 11910 }, { "epoch": 8.271161533457416e-14, "grad_norm": 1.7165452241897583, "learning_rate": 0.00014661560000412283, "loss": 0.6194, "step": 11920 }, { "epoch": 8.278100427361323e-14, "grad_norm": 1.4922155141830444, "learning_rate": 0.0001466088309227595, "loss": 0.6436, "step": 11930 }, { "epoch": 8.285039321265231e-14, "grad_norm": 1.3964905738830566, "learning_rate": 0.00014660205525321627, "loss": 0.5683, "step": 11940 }, { "epoch": 8.291978215169138e-14, "grad_norm": 1.8347221612930298, "learning_rate": 0.00014659527299618944, "loss": 0.6108, "step": 11950 }, { "epoch": 8.298917109073045e-14, "grad_norm": 1.4889708757400513, "learning_rate": 0.00014658848415237608, "loss": 0.6147, "step": 11960 }, { "epoch": 8.305856002976952e-14, "grad_norm": 1.4178546667099, "learning_rate": 0.00014658168872247378, "loss": 0.5565, "step": 11970 }, { "epoch": 8.31279489688086e-14, "grad_norm": 1.849022626876831, "learning_rate": 0.00014657488670718092, "loss": 0.5833, "step": 11980 }, { "epoch": 8.319733790784767e-14, "grad_norm": 1.8982609510421753, "learning_rate": 0.00014656807810719643, "loss": 0.5866, "step": 11990 }, { "epoch": 8.326672684688674e-14, "grad_norm": 1.8296819925308228, "learning_rate": 0.0001465612629232201, "loss": 0.5974, "step": 12000 }, { "epoch": 8.333611578592581e-14, "grad_norm": 2.0990192890167236, "learning_rate": 0.00014655444115595228, "loss": 0.5879, "step": 12010 }, { "epoch": 8.340550472496489e-14, "grad_norm": 1.6368311643600464, "learning_rate": 0.00014654761280609395, "loss": 0.603, "step": 12020 }, { "epoch": 8.347489366400396e-14, "grad_norm": 1.9391672611236572, "learning_rate": 0.0001465407778743469, "loss": 0.5795, "step": 12030 }, { "epoch": 8.354428260304303e-14, "grad_norm": 2.0748159885406494, "learning_rate": 0.00014653393636141347, "loss": 0.6466, "step": 12040 }, { "epoch": 8.36136715420821e-14, "grad_norm": 1.6568641662597656, "learning_rate": 0.00014652708826799675, "loss": 0.6602, "step": 12050 }, { "epoch": 8.368306048112117e-14, "grad_norm": 1.8242896795272827, "learning_rate": 0.0001465202335948005, "loss": 0.5831, "step": 12060 }, { "epoch": 8.375244942016025e-14, "grad_norm": 1.804728388786316, "learning_rate": 0.00014651337234252915, "loss": 0.6652, "step": 12070 }, { "epoch": 8.382183835919932e-14, "grad_norm": 1.935641884803772, "learning_rate": 0.00014650650451188778, "loss": 0.623, "step": 12080 }, { "epoch": 8.389122729823839e-14, "grad_norm": 1.801060676574707, "learning_rate": 0.00014649963010358215, "loss": 0.586, "step": 12090 }, { "epoch": 8.396061623727746e-14, "grad_norm": 1.6459389925003052, "learning_rate": 0.00014649274911831875, "loss": 0.5486, "step": 12100 }, { "epoch": 8.403000517631654e-14, "grad_norm": 2.2724111080169678, "learning_rate": 0.00014648586155680473, "loss": 0.5834, "step": 12110 }, { "epoch": 8.409939411535561e-14, "grad_norm": 1.864217758178711, "learning_rate": 0.00014647896741974782, "loss": 0.5784, "step": 12120 }, { "epoch": 8.416878305439468e-14, "grad_norm": 1.7782739400863647, "learning_rate": 0.00014647206670785655, "loss": 0.5557, "step": 12130 }, { "epoch": 8.423817199343375e-14, "grad_norm": 1.5777384042739868, "learning_rate": 0.00014646515942184004, "loss": 0.5532, "step": 12140 }, { "epoch": 8.430756093247282e-14, "grad_norm": 2.237774610519409, "learning_rate": 0.0001464582455624082, "loss": 0.6349, "step": 12150 }, { "epoch": 8.43769498715119e-14, "grad_norm": 1.8874484300613403, "learning_rate": 0.00014645132513027142, "loss": 0.5489, "step": 12160 }, { "epoch": 8.444633881055097e-14, "grad_norm": 1.515108346939087, "learning_rate": 0.000146444398126141, "loss": 0.5923, "step": 12170 }, { "epoch": 8.451572774959004e-14, "grad_norm": 1.6060832738876343, "learning_rate": 0.0001464374645507287, "loss": 0.5778, "step": 12180 }, { "epoch": 8.458511668862911e-14, "grad_norm": 1.8125596046447754, "learning_rate": 0.00014643052440474713, "loss": 0.6066, "step": 12190 }, { "epoch": 8.465450562766819e-14, "grad_norm": 1.6969876289367676, "learning_rate": 0.00014642357768890942, "loss": 0.6124, "step": 12200 }, { "epoch": 8.472389456670726e-14, "grad_norm": 1.6685600280761719, "learning_rate": 0.0001464166244039295, "loss": 0.6409, "step": 12210 }, { "epoch": 8.479328350574633e-14, "grad_norm": 2.0508852005004883, "learning_rate": 0.00014640966455052197, "loss": 0.5889, "step": 12220 }, { "epoch": 8.48626724447854e-14, "grad_norm": 1.9384311437606812, "learning_rate": 0.00014640269812940198, "loss": 0.6074, "step": 12230 }, { "epoch": 8.493206138382448e-14, "grad_norm": 1.6965117454528809, "learning_rate": 0.0001463957251412855, "loss": 0.6244, "step": 12240 }, { "epoch": 8.500145032286355e-14, "grad_norm": 1.7365933656692505, "learning_rate": 0.00014638874558688906, "loss": 0.5631, "step": 12250 }, { "epoch": 8.507083926190262e-14, "grad_norm": 1.726814866065979, "learning_rate": 0.00014638175946692994, "loss": 0.5505, "step": 12260 }, { "epoch": 8.514022820094169e-14, "grad_norm": 2.0869436264038086, "learning_rate": 0.00014637476678212608, "loss": 0.5865, "step": 12270 }, { "epoch": 8.520961713998076e-14, "grad_norm": 1.8720430135726929, "learning_rate": 0.00014636776753319613, "loss": 0.5706, "step": 12280 }, { "epoch": 8.527900607901984e-14, "grad_norm": 1.9910624027252197, "learning_rate": 0.0001463607617208593, "loss": 0.5215, "step": 12290 }, { "epoch": 8.534839501805891e-14, "grad_norm": 1.876868724822998, "learning_rate": 0.00014635374934583557, "loss": 0.5673, "step": 12300 }, { "epoch": 8.541778395709798e-14, "grad_norm": 1.9793531894683838, "learning_rate": 0.00014634673040884557, "loss": 0.5826, "step": 12310 }, { "epoch": 8.548717289613705e-14, "grad_norm": 1.8160724639892578, "learning_rate": 0.00014633970491061062, "loss": 0.5562, "step": 12320 }, { "epoch": 8.555656183517613e-14, "grad_norm": 1.7849628925323486, "learning_rate": 0.00014633267285185268, "loss": 0.5776, "step": 12330 }, { "epoch": 8.56259507742152e-14, "grad_norm": 1.789126992225647, "learning_rate": 0.00014632563423329443, "loss": 0.5726, "step": 12340 }, { "epoch": 8.569533971325427e-14, "grad_norm": 1.8279370069503784, "learning_rate": 0.00014631858905565914, "loss": 0.6106, "step": 12350 }, { "epoch": 8.576472865229334e-14, "grad_norm": 1.758277416229248, "learning_rate": 0.00014631153731967087, "loss": 0.6168, "step": 12360 }, { "epoch": 8.583411759133242e-14, "grad_norm": 1.868533730506897, "learning_rate": 0.0001463044790260543, "loss": 0.5564, "step": 12370 }, { "epoch": 8.590350653037149e-14, "grad_norm": 2.15696382522583, "learning_rate": 0.00014629741417553476, "loss": 0.6365, "step": 12380 }, { "epoch": 8.597289546941056e-14, "grad_norm": 2.0377767086029053, "learning_rate": 0.00014629034276883823, "loss": 0.6002, "step": 12390 }, { "epoch": 8.604228440844963e-14, "grad_norm": 1.9457752704620361, "learning_rate": 0.00014628326480669145, "loss": 0.6042, "step": 12400 }, { "epoch": 8.61116733474887e-14, "grad_norm": 2.1396522521972656, "learning_rate": 0.00014627618028982182, "loss": 0.5521, "step": 12410 }, { "epoch": 8.618106228652778e-14, "grad_norm": 1.5338581800460815, "learning_rate": 0.00014626908921895734, "loss": 0.5246, "step": 12420 }, { "epoch": 8.625045122556685e-14, "grad_norm": 1.3009228706359863, "learning_rate": 0.00014626199159482672, "loss": 0.583, "step": 12430 }, { "epoch": 8.631984016460592e-14, "grad_norm": 1.8689589500427246, "learning_rate": 0.0001462548874181594, "loss": 0.5479, "step": 12440 }, { "epoch": 8.638922910364499e-14, "grad_norm": 1.7979040145874023, "learning_rate": 0.00014624777668968537, "loss": 0.5569, "step": 12450 }, { "epoch": 8.645861804268407e-14, "grad_norm": 1.9970425367355347, "learning_rate": 0.00014624065941013547, "loss": 0.5419, "step": 12460 }, { "epoch": 8.652800698172314e-14, "grad_norm": 1.900072693824768, "learning_rate": 0.00014623353558024103, "loss": 0.5562, "step": 12470 }, { "epoch": 8.659739592076221e-14, "grad_norm": 2.177082061767578, "learning_rate": 0.00014622640520073414, "loss": 0.586, "step": 12480 }, { "epoch": 8.666678485980128e-14, "grad_norm": 1.669804573059082, "learning_rate": 0.0001462192682723476, "loss": 0.5593, "step": 12490 }, { "epoch": 8.673617379884035e-14, "grad_norm": 2.0029194355010986, "learning_rate": 0.0001462121247958148, "loss": 0.5608, "step": 12500 }, { "epoch": 8.680556273787943e-14, "grad_norm": 1.8924312591552734, "learning_rate": 0.00014620497477186984, "loss": 0.5878, "step": 12510 }, { "epoch": 8.68749516769185e-14, "grad_norm": 1.7594841718673706, "learning_rate": 0.00014619781820124757, "loss": 0.6116, "step": 12520 }, { "epoch": 8.694434061595757e-14, "grad_norm": 1.8959667682647705, "learning_rate": 0.00014619065508468337, "loss": 0.5469, "step": 12530 }, { "epoch": 8.701372955499664e-14, "grad_norm": 1.7747634649276733, "learning_rate": 0.00014618348542291335, "loss": 0.6521, "step": 12540 }, { "epoch": 8.708311849403572e-14, "grad_norm": 1.903565526008606, "learning_rate": 0.00014617630921667435, "loss": 0.6172, "step": 12550 }, { "epoch": 8.715250743307479e-14, "grad_norm": 1.4263851642608643, "learning_rate": 0.00014616912646670384, "loss": 0.5442, "step": 12560 }, { "epoch": 8.722189637211386e-14, "grad_norm": 1.931229829788208, "learning_rate": 0.0001461619371737399, "loss": 0.5359, "step": 12570 }, { "epoch": 8.729128531115293e-14, "grad_norm": 2.060948371887207, "learning_rate": 0.0001461547413385214, "loss": 0.5756, "step": 12580 }, { "epoch": 8.7360674250192e-14, "grad_norm": 1.651002287864685, "learning_rate": 0.0001461475389617878, "loss": 0.5287, "step": 12590 }, { "epoch": 8.743006318923108e-14, "grad_norm": 1.7310506105422974, "learning_rate": 0.00014614033004427924, "loss": 0.511, "step": 12600 }, { "epoch": 8.749945212827015e-14, "grad_norm": 1.6929713487625122, "learning_rate": 0.0001461331145867366, "loss": 0.5226, "step": 12610 }, { "epoch": 8.756884106730922e-14, "grad_norm": 1.7985080480575562, "learning_rate": 0.00014612589258990134, "loss": 0.5775, "step": 12620 }, { "epoch": 8.76382300063483e-14, "grad_norm": 1.7553006410598755, "learning_rate": 0.00014611866405451563, "loss": 0.5288, "step": 12630 }, { "epoch": 8.770761894538737e-14, "grad_norm": 1.6226269006729126, "learning_rate": 0.00014611142898132235, "loss": 0.5508, "step": 12640 }, { "epoch": 8.777700788442644e-14, "grad_norm": 1.5343693494796753, "learning_rate": 0.00014610418737106496, "loss": 0.558, "step": 12650 }, { "epoch": 8.784639682346551e-14, "grad_norm": 1.9040379524230957, "learning_rate": 0.0001460969392244877, "loss": 0.5909, "step": 12660 }, { "epoch": 8.791578576250458e-14, "grad_norm": 2.3416941165924072, "learning_rate": 0.0001460896845423354, "loss": 0.5856, "step": 12670 }, { "epoch": 8.798517470154366e-14, "grad_norm": 1.518486738204956, "learning_rate": 0.0001460824233253536, "loss": 0.5627, "step": 12680 }, { "epoch": 8.805456364058273e-14, "grad_norm": 1.9263806343078613, "learning_rate": 0.00014607515557428852, "loss": 0.5719, "step": 12690 }, { "epoch": 8.81239525796218e-14, "grad_norm": 1.885764718055725, "learning_rate": 0.00014606788128988698, "loss": 0.6107, "step": 12700 }, { "epoch": 8.819334151866087e-14, "grad_norm": 1.9315112829208374, "learning_rate": 0.0001460606004728966, "loss": 0.6156, "step": 12710 }, { "epoch": 8.826273045769994e-14, "grad_norm": 1.6834007501602173, "learning_rate": 0.00014605331312406557, "loss": 0.5384, "step": 12720 }, { "epoch": 8.833211939673902e-14, "grad_norm": 1.6985210180282593, "learning_rate": 0.00014604601924414276, "loss": 0.5319, "step": 12730 }, { "epoch": 8.840150833577809e-14, "grad_norm": 1.8392934799194336, "learning_rate": 0.0001460387188338777, "loss": 0.5782, "step": 12740 }, { "epoch": 8.847089727481716e-14, "grad_norm": 1.5398123264312744, "learning_rate": 0.00014603141189402073, "loss": 0.4987, "step": 12750 }, { "epoch": 8.854028621385623e-14, "grad_norm": 1.8241780996322632, "learning_rate": 0.00014602409842532266, "loss": 0.4965, "step": 12760 }, { "epoch": 8.860967515289531e-14, "grad_norm": 1.597991943359375, "learning_rate": 0.00014601677842853507, "loss": 0.5177, "step": 12770 }, { "epoch": 8.867906409193438e-14, "grad_norm": 1.7104915380477905, "learning_rate": 0.00014600945190441022, "loss": 0.5459, "step": 12780 }, { "epoch": 8.874845303097345e-14, "grad_norm": 1.5096904039382935, "learning_rate": 0.00014600211885370103, "loss": 0.51, "step": 12790 }, { "epoch": 8.881784197001252e-14, "grad_norm": 2.1093735694885254, "learning_rate": 0.00014599477927716107, "loss": 0.568, "step": 12800 }, { "epoch": 8.88872309090516e-14, "grad_norm": 1.7477951049804688, "learning_rate": 0.0001459874331755446, "loss": 0.524, "step": 12810 }, { "epoch": 8.895661984809067e-14, "grad_norm": 1.8474655151367188, "learning_rate": 0.00014598008054960654, "loss": 0.5772, "step": 12820 }, { "epoch": 8.902600878712974e-14, "grad_norm": 1.9230769872665405, "learning_rate": 0.00014597272140010254, "loss": 0.6032, "step": 12830 }, { "epoch": 8.909539772616881e-14, "grad_norm": 1.8260226249694824, "learning_rate": 0.0001459653557277888, "loss": 0.5135, "step": 12840 }, { "epoch": 8.916478666520788e-14, "grad_norm": 2.526167392730713, "learning_rate": 0.00014595798353342227, "loss": 0.621, "step": 12850 }, { "epoch": 8.923417560424696e-14, "grad_norm": 2.3232128620147705, "learning_rate": 0.00014595060481776057, "loss": 0.5698, "step": 12860 }, { "epoch": 8.930356454328603e-14, "grad_norm": 1.9089231491088867, "learning_rate": 0.000145943219581562, "loss": 0.5628, "step": 12870 }, { "epoch": 8.93729534823251e-14, "grad_norm": 1.8234467506408691, "learning_rate": 0.00014593582782558546, "loss": 0.5342, "step": 12880 }, { "epoch": 8.944234242136417e-14, "grad_norm": 1.8338319063186646, "learning_rate": 0.00014592842955059063, "loss": 0.4996, "step": 12890 }, { "epoch": 8.951173136040325e-14, "grad_norm": 2.079850435256958, "learning_rate": 0.00014592102475733773, "loss": 0.5532, "step": 12900 }, { "epoch": 8.958112029944232e-14, "grad_norm": 1.577723741531372, "learning_rate": 0.00014591361344658775, "loss": 0.4884, "step": 12910 }, { "epoch": 8.965050923848139e-14, "grad_norm": 1.3188661336898804, "learning_rate": 0.00014590619561910233, "loss": 0.5081, "step": 12920 }, { "epoch": 8.971989817752046e-14, "grad_norm": 1.7690623998641968, "learning_rate": 0.00014589877127564374, "loss": 0.5283, "step": 12930 }, { "epoch": 8.978928711655954e-14, "grad_norm": 1.6262685060501099, "learning_rate": 0.00014589134041697498, "loss": 0.5178, "step": 12940 }, { "epoch": 8.985867605559861e-14, "grad_norm": 1.9526047706604004, "learning_rate": 0.00014588390304385963, "loss": 0.5457, "step": 12950 }, { "epoch": 8.992806499463768e-14, "grad_norm": 1.7686899900436401, "learning_rate": 0.00014587645915706206, "loss": 0.5153, "step": 12960 }, { "epoch": 8.999745393367675e-14, "grad_norm": 1.9296541213989258, "learning_rate": 0.00014586900875734722, "loss": 0.5444, "step": 12970 }, { "epoch": 9.006684287271582e-14, "grad_norm": 2.204080820083618, "learning_rate": 0.00014586155184548072, "loss": 0.5592, "step": 12980 }, { "epoch": 9.01362318117549e-14, "grad_norm": 1.742918848991394, "learning_rate": 0.00014585408842222893, "loss": 0.574, "step": 12990 }, { "epoch": 9.020562075079397e-14, "grad_norm": 1.9207394123077393, "learning_rate": 0.00014584661848835882, "loss": 0.508, "step": 13000 }, { "epoch": 9.027500968983304e-14, "grad_norm": 1.7133293151855469, "learning_rate": 0.00014583914204463803, "loss": 0.5755, "step": 13010 }, { "epoch": 9.034439862887211e-14, "grad_norm": 1.6706287860870361, "learning_rate": 0.00014583165909183484, "loss": 0.5861, "step": 13020 }, { "epoch": 9.041378756791119e-14, "grad_norm": 2.1353256702423096, "learning_rate": 0.00014582416963071832, "loss": 0.5239, "step": 13030 }, { "epoch": 9.048317650695026e-14, "grad_norm": 1.8304259777069092, "learning_rate": 0.00014581667366205808, "loss": 0.506, "step": 13040 }, { "epoch": 9.055256544598933e-14, "grad_norm": 1.6840101480484009, "learning_rate": 0.00014580917118662445, "loss": 0.5317, "step": 13050 }, { "epoch": 9.06219543850284e-14, "grad_norm": 1.9074162244796753, "learning_rate": 0.00014580166220518845, "loss": 0.5101, "step": 13060 }, { "epoch": 9.069134332406747e-14, "grad_norm": 1.7532092332839966, "learning_rate": 0.0001457941467185217, "loss": 0.4656, "step": 13070 }, { "epoch": 9.076073226310655e-14, "grad_norm": 1.944380521774292, "learning_rate": 0.0001457866247273966, "loss": 0.5112, "step": 13080 }, { "epoch": 9.083012120214562e-14, "grad_norm": 1.7294270992279053, "learning_rate": 0.00014577909623258607, "loss": 0.5185, "step": 13090 }, { "epoch": 9.089951014118469e-14, "grad_norm": 1.8654406070709229, "learning_rate": 0.00014577156123486386, "loss": 0.5102, "step": 13100 }, { "epoch": 9.096889908022376e-14, "grad_norm": 2.01185941696167, "learning_rate": 0.00014576401973500424, "loss": 0.512, "step": 13110 }, { "epoch": 9.103828801926284e-14, "grad_norm": 1.746657371520996, "learning_rate": 0.00014575647173378228, "loss": 0.5025, "step": 13120 }, { "epoch": 9.110767695830191e-14, "grad_norm": 1.9053512811660767, "learning_rate": 0.0001457489172319736, "loss": 0.5487, "step": 13130 }, { "epoch": 9.117706589734098e-14, "grad_norm": 1.883307695388794, "learning_rate": 0.00014574135623035455, "loss": 0.5577, "step": 13140 }, { "epoch": 9.124645483638005e-14, "grad_norm": 1.4428898096084595, "learning_rate": 0.0001457337887297022, "loss": 0.4998, "step": 13150 }, { "epoch": 9.131584377541913e-14, "grad_norm": 1.9143213033676147, "learning_rate": 0.00014572621473079415, "loss": 0.522, "step": 13160 }, { "epoch": 9.13852327144582e-14, "grad_norm": 1.6298106908798218, "learning_rate": 0.00014571863423440878, "loss": 0.5371, "step": 13170 }, { "epoch": 9.145462165349727e-14, "grad_norm": 1.4857451915740967, "learning_rate": 0.0001457110472413251, "loss": 0.5672, "step": 13180 }, { "epoch": 9.152401059253634e-14, "grad_norm": 1.443803310394287, "learning_rate": 0.0001457034537523228, "loss": 0.491, "step": 13190 }, { "epoch": 9.159339953157541e-14, "grad_norm": 1.6803009510040283, "learning_rate": 0.00014569585376818226, "loss": 0.5123, "step": 13200 }, { "epoch": 9.166278847061449e-14, "grad_norm": 1.5666348934173584, "learning_rate": 0.0001456882472896844, "loss": 0.5267, "step": 13210 }, { "epoch": 9.173217740965356e-14, "grad_norm": 1.6019494533538818, "learning_rate": 0.00014568063431761102, "loss": 0.4819, "step": 13220 }, { "epoch": 9.180156634869263e-14, "grad_norm": 1.7549004554748535, "learning_rate": 0.00014567301485274439, "loss": 0.4955, "step": 13230 }, { "epoch": 9.18709552877317e-14, "grad_norm": 1.8631086349487305, "learning_rate": 0.00014566538889586756, "loss": 0.4937, "step": 13240 }, { "epoch": 9.194034422677078e-14, "grad_norm": 1.810246229171753, "learning_rate": 0.00014565775644776423, "loss": 0.5147, "step": 13250 }, { "epoch": 9.200973316580985e-14, "grad_norm": 2.199460744857788, "learning_rate": 0.00014565011750921868, "loss": 0.5059, "step": 13260 }, { "epoch": 9.207912210484892e-14, "grad_norm": 1.8092135190963745, "learning_rate": 0.00014564247208101604, "loss": 0.5252, "step": 13270 }, { "epoch": 9.214851104388799e-14, "grad_norm": 1.9690837860107422, "learning_rate": 0.0001456348201639419, "loss": 0.484, "step": 13280 }, { "epoch": 9.221789998292707e-14, "grad_norm": 1.7890839576721191, "learning_rate": 0.00014562716175878267, "loss": 0.5245, "step": 13290 }, { "epoch": 9.228728892196614e-14, "grad_norm": 1.7083340883255005, "learning_rate": 0.00014561949686632536, "loss": 0.5629, "step": 13300 }, { "epoch": 9.235667786100521e-14, "grad_norm": 1.5658704042434692, "learning_rate": 0.00014561182548735764, "loss": 0.4642, "step": 13310 }, { "epoch": 9.242606680004428e-14, "grad_norm": 1.9546990394592285, "learning_rate": 0.00014560414762266788, "loss": 0.5661, "step": 13320 }, { "epoch": 9.249545573908335e-14, "grad_norm": 1.9428247213363647, "learning_rate": 0.0001455964632730451, "loss": 0.5423, "step": 13330 }, { "epoch": 9.256484467812243e-14, "grad_norm": 1.91390061378479, "learning_rate": 0.00014558877243927898, "loss": 0.5076, "step": 13340 }, { "epoch": 9.26342336171615e-14, "grad_norm": 1.6540240049362183, "learning_rate": 0.00014558107512215986, "loss": 0.474, "step": 13350 }, { "epoch": 9.270362255620057e-14, "grad_norm": 2.3850674629211426, "learning_rate": 0.00014557337132247878, "loss": 0.4917, "step": 13360 }, { "epoch": 9.277301149523964e-14, "grad_norm": 2.0765769481658936, "learning_rate": 0.00014556566104102742, "loss": 0.5001, "step": 13370 }, { "epoch": 9.284240043427872e-14, "grad_norm": 1.7414309978485107, "learning_rate": 0.00014555794427859815, "loss": 0.4746, "step": 13380 }, { "epoch": 9.291178937331779e-14, "grad_norm": 1.708103895187378, "learning_rate": 0.00014555022103598394, "loss": 0.4773, "step": 13390 }, { "epoch": 9.298117831235686e-14, "grad_norm": 2.1132917404174805, "learning_rate": 0.0001455424913139785, "loss": 0.5459, "step": 13400 }, { "epoch": 9.305056725139593e-14, "grad_norm": 1.9388973712921143, "learning_rate": 0.0001455347551133762, "loss": 0.4743, "step": 13410 }, { "epoch": 9.3119956190435e-14, "grad_norm": 1.5122756958007812, "learning_rate": 0.000145527012434972, "loss": 0.5025, "step": 13420 }, { "epoch": 9.318934512947408e-14, "grad_norm": 1.708476185798645, "learning_rate": 0.00014551926327956163, "loss": 0.5056, "step": 13430 }, { "epoch": 9.325873406851315e-14, "grad_norm": 1.6370904445648193, "learning_rate": 0.00014551150764794145, "loss": 0.5065, "step": 13440 }, { "epoch": 9.332812300755222e-14, "grad_norm": 1.5455464124679565, "learning_rate": 0.00014550374554090842, "loss": 0.5026, "step": 13450 }, { "epoch": 9.33975119465913e-14, "grad_norm": 1.578701376914978, "learning_rate": 0.00014549597695926025, "loss": 0.5494, "step": 13460 }, { "epoch": 9.346690088563037e-14, "grad_norm": 1.745391607284546, "learning_rate": 0.00014548820190379526, "loss": 0.4685, "step": 13470 }, { "epoch": 9.353628982466944e-14, "grad_norm": 2.0206429958343506, "learning_rate": 0.0001454804203753125, "loss": 0.5309, "step": 13480 }, { "epoch": 9.360567876370851e-14, "grad_norm": 1.7269880771636963, "learning_rate": 0.00014547263237461156, "loss": 0.5417, "step": 13490 }, { "epoch": 9.367506770274758e-14, "grad_norm": 1.8156310319900513, "learning_rate": 0.00014546483790249287, "loss": 0.4834, "step": 13500 }, { "epoch": 9.374445664178666e-14, "grad_norm": 1.9445420503616333, "learning_rate": 0.00014545703695975737, "loss": 0.4701, "step": 13510 }, { "epoch": 9.381384558082573e-14, "grad_norm": 1.979134440422058, "learning_rate": 0.0001454492295472068, "loss": 0.5059, "step": 13520 }, { "epoch": 9.38832345198648e-14, "grad_norm": 1.9943641424179077, "learning_rate": 0.00014544141566564338, "loss": 0.4915, "step": 13530 }, { "epoch": 9.395262345890387e-14, "grad_norm": 2.0147037506103516, "learning_rate": 0.00014543359531587024, "loss": 0.4382, "step": 13540 }, { "epoch": 9.402201239794294e-14, "grad_norm": 1.9555226564407349, "learning_rate": 0.00014542576849869095, "loss": 0.5076, "step": 13550 }, { "epoch": 9.409140133698202e-14, "grad_norm": 2.1542837619781494, "learning_rate": 0.00014541793521490986, "loss": 0.5074, "step": 13560 }, { "epoch": 9.416079027602109e-14, "grad_norm": 1.573736548423767, "learning_rate": 0.000145410095465332, "loss": 0.4784, "step": 13570 }, { "epoch": 9.423017921506016e-14, "grad_norm": 1.7546331882476807, "learning_rate": 0.00014540224925076295, "loss": 0.4957, "step": 13580 }, { "epoch": 9.429956815409923e-14, "grad_norm": 1.7481378316879272, "learning_rate": 0.00014539439657200912, "loss": 0.4954, "step": 13590 }, { "epoch": 9.43689570931383e-14, "grad_norm": 1.5542930364608765, "learning_rate": 0.00014538653742987738, "loss": 0.501, "step": 13600 }, { "epoch": 9.443834603217738e-14, "grad_norm": 1.817026972770691, "learning_rate": 0.0001453786718251755, "loss": 0.5227, "step": 13610 }, { "epoch": 9.450773497121645e-14, "grad_norm": 1.716010332107544, "learning_rate": 0.00014537079975871173, "loss": 0.4623, "step": 13620 }, { "epoch": 9.457712391025552e-14, "grad_norm": 1.9755781888961792, "learning_rate": 0.00014536292123129503, "loss": 0.5102, "step": 13630 }, { "epoch": 9.46465128492946e-14, "grad_norm": 2.0019102096557617, "learning_rate": 0.00014535503624373512, "loss": 0.5091, "step": 13640 }, { "epoch": 9.471590178833367e-14, "grad_norm": 1.8190650939941406, "learning_rate": 0.0001453471447968422, "loss": 0.5173, "step": 13650 }, { "epoch": 9.478529072737274e-14, "grad_norm": 2.149968385696411, "learning_rate": 0.00014533924689142736, "loss": 0.4706, "step": 13660 }, { "epoch": 9.485467966641181e-14, "grad_norm": 1.7202515602111816, "learning_rate": 0.00014533134252830212, "loss": 0.4439, "step": 13670 }, { "epoch": 9.492406860545088e-14, "grad_norm": 1.3774373531341553, "learning_rate": 0.00014532343170827884, "loss": 0.5127, "step": 13680 }, { "epoch": 9.499345754448996e-14, "grad_norm": 1.6376941204071045, "learning_rate": 0.00014531551443217045, "loss": 0.4771, "step": 13690 }, { "epoch": 9.506284648352903e-14, "grad_norm": 2.1280462741851807, "learning_rate": 0.0001453075907007906, "loss": 0.4733, "step": 13700 }, { "epoch": 9.51322354225681e-14, "grad_norm": 2.1267406940460205, "learning_rate": 0.00014529966051495354, "loss": 0.4912, "step": 13710 }, { "epoch": 9.520162436160717e-14, "grad_norm": 2.0959317684173584, "learning_rate": 0.00014529172387547427, "loss": 0.4828, "step": 13720 }, { "epoch": 9.527101330064625e-14, "grad_norm": 1.9174953699111938, "learning_rate": 0.00014528378078316836, "loss": 0.4948, "step": 13730 }, { "epoch": 9.534040223968532e-14, "grad_norm": 1.8229044675827026, "learning_rate": 0.0001452758312388521, "loss": 0.4922, "step": 13740 }, { "epoch": 9.540979117872439e-14, "grad_norm": 1.7982683181762695, "learning_rate": 0.00014526787524334243, "loss": 0.4706, "step": 13750 }, { "epoch": 9.547918011776346e-14, "grad_norm": 1.806432843208313, "learning_rate": 0.00014525991279745696, "loss": 0.4778, "step": 13760 }, { "epoch": 9.554856905680253e-14, "grad_norm": 1.7961784601211548, "learning_rate": 0.00014525194390201395, "loss": 0.5101, "step": 13770 }, { "epoch": 9.561795799584161e-14, "grad_norm": 1.6512629985809326, "learning_rate": 0.00014524396855783233, "loss": 0.4397, "step": 13780 }, { "epoch": 9.568734693488068e-14, "grad_norm": 1.8228294849395752, "learning_rate": 0.0001452359867657317, "loss": 0.5212, "step": 13790 }, { "epoch": 9.575673587391975e-14, "grad_norm": 1.710288166999817, "learning_rate": 0.00014522799852653223, "loss": 0.5146, "step": 13800 }, { "epoch": 9.582612481295882e-14, "grad_norm": 1.4125797748565674, "learning_rate": 0.000145220003841055, "loss": 0.4481, "step": 13810 }, { "epoch": 9.58955137519979e-14, "grad_norm": 1.7084463834762573, "learning_rate": 0.00014521200271012144, "loss": 0.4484, "step": 13820 }, { "epoch": 9.596490269103697e-14, "grad_norm": 2.562086582183838, "learning_rate": 0.00014520399513455386, "loss": 0.4981, "step": 13830 }, { "epoch": 9.603429163007604e-14, "grad_norm": 1.9363263845443726, "learning_rate": 0.00014519598111517514, "loss": 0.4819, "step": 13840 }, { "epoch": 9.610368056911511e-14, "grad_norm": 1.5592763423919678, "learning_rate": 0.00014518796065280887, "loss": 0.4454, "step": 13850 }, { "epoch": 9.617306950815419e-14, "grad_norm": 1.9152923822402954, "learning_rate": 0.00014517993374827925, "loss": 0.4562, "step": 13860 }, { "epoch": 9.624245844719326e-14, "grad_norm": 1.7481296062469482, "learning_rate": 0.0001451719004024112, "loss": 0.5229, "step": 13870 }, { "epoch": 9.631184738623233e-14, "grad_norm": 1.860676884651184, "learning_rate": 0.0001451638606160302, "loss": 0.4562, "step": 13880 }, { "epoch": 9.63812363252714e-14, "grad_norm": 1.6480166912078857, "learning_rate": 0.00014515581438996256, "loss": 0.4634, "step": 13890 }, { "epoch": 9.645062526431047e-14, "grad_norm": 1.5378928184509277, "learning_rate": 0.0001451477617250351, "loss": 0.4894, "step": 13900 }, { "epoch": 9.652001420334955e-14, "grad_norm": 1.7672655582427979, "learning_rate": 0.00014513970262207537, "loss": 0.4856, "step": 13910 }, { "epoch": 9.658940314238862e-14, "grad_norm": 2.019899606704712, "learning_rate": 0.00014513163708191156, "loss": 0.4756, "step": 13920 }, { "epoch": 9.665879208142769e-14, "grad_norm": 1.333742618560791, "learning_rate": 0.00014512356510537256, "loss": 0.4625, "step": 13930 }, { "epoch": 9.672818102046676e-14, "grad_norm": 1.7457395792007446, "learning_rate": 0.00014511548669328785, "loss": 0.4547, "step": 13940 }, { "epoch": 9.679756995950584e-14, "grad_norm": 1.7249329090118408, "learning_rate": 0.0001451074018464876, "loss": 0.4998, "step": 13950 }, { "epoch": 9.686695889854491e-14, "grad_norm": 1.744535207748413, "learning_rate": 0.00014509931056580272, "loss": 0.5057, "step": 13960 }, { "epoch": 9.693634783758398e-14, "grad_norm": 1.676255464553833, "learning_rate": 0.00014509121285206464, "loss": 0.4415, "step": 13970 }, { "epoch": 9.700573677662305e-14, "grad_norm": 1.632596731185913, "learning_rate": 0.0001450831087061056, "loss": 0.4403, "step": 13980 }, { "epoch": 9.707512571566213e-14, "grad_norm": 2.1650424003601074, "learning_rate": 0.00014507499812875836, "loss": 0.5135, "step": 13990 }, { "epoch": 9.71445146547012e-14, "grad_norm": 1.5834444761276245, "learning_rate": 0.00014506688112085644, "loss": 0.4516, "step": 14000 }, { "epoch": 9.721390359374027e-14, "grad_norm": 1.8523372411727905, "learning_rate": 0.00014505875768323402, "loss": 0.4315, "step": 14010 }, { "epoch": 9.728329253277934e-14, "grad_norm": 1.639618992805481, "learning_rate": 0.00014505062781672585, "loss": 0.4661, "step": 14020 }, { "epoch": 9.735268147181841e-14, "grad_norm": 1.5737920999526978, "learning_rate": 0.00014504249152216743, "loss": 0.4837, "step": 14030 }, { "epoch": 9.742207041085749e-14, "grad_norm": 1.6471182107925415, "learning_rate": 0.0001450343488003949, "loss": 0.4429, "step": 14040 }, { "epoch": 9.749145934989656e-14, "grad_norm": 2.1322524547576904, "learning_rate": 0.00014502619965224502, "loss": 0.4871, "step": 14050 }, { "epoch": 9.756084828893563e-14, "grad_norm": 1.6367660760879517, "learning_rate": 0.00014501804407855527, "loss": 0.4461, "step": 14060 }, { "epoch": 9.76302372279747e-14, "grad_norm": 1.7700135707855225, "learning_rate": 0.00014500988208016375, "loss": 0.4708, "step": 14070 }, { "epoch": 9.769962616701378e-14, "grad_norm": 1.7651363611221313, "learning_rate": 0.00014500171365790923, "loss": 0.4901, "step": 14080 }, { "epoch": 9.776901510605285e-14, "grad_norm": 1.8301613330841064, "learning_rate": 0.00014499353881263115, "loss": 0.4201, "step": 14090 }, { "epoch": 9.783840404509192e-14, "grad_norm": 2.251863956451416, "learning_rate": 0.00014498535754516961, "loss": 0.5071, "step": 14100 }, { "epoch": 9.790779298413099e-14, "grad_norm": 2.2569563388824463, "learning_rate": 0.00014497716985636537, "loss": 0.4787, "step": 14110 }, { "epoch": 9.797718192317006e-14, "grad_norm": 1.8017748594284058, "learning_rate": 0.00014496897574705982, "loss": 0.4586, "step": 14120 }, { "epoch": 9.804657086220914e-14, "grad_norm": 1.8812689781188965, "learning_rate": 0.000144960775218095, "loss": 0.4356, "step": 14130 }, { "epoch": 9.811595980124821e-14, "grad_norm": 1.7185002565383911, "learning_rate": 0.0001449525682703137, "loss": 0.411, "step": 14140 }, { "epoch": 9.818534874028728e-14, "grad_norm": 1.9639073610305786, "learning_rate": 0.0001449443549045593, "loss": 0.4951, "step": 14150 }, { "epoch": 9.825473767932635e-14, "grad_norm": 1.9095617532730103, "learning_rate": 0.00014493613512167583, "loss": 0.4402, "step": 14160 }, { "epoch": 9.832412661836543e-14, "grad_norm": 1.514885663986206, "learning_rate": 0.00014492790892250802, "loss": 0.4482, "step": 14170 }, { "epoch": 9.83935155574045e-14, "grad_norm": 1.7884515523910522, "learning_rate": 0.00014491967630790123, "loss": 0.484, "step": 14180 }, { "epoch": 9.846290449644357e-14, "grad_norm": 1.7022219896316528, "learning_rate": 0.00014491143727870148, "loss": 0.451, "step": 14190 }, { "epoch": 9.853229343548264e-14, "grad_norm": 1.727394700050354, "learning_rate": 0.00014490319183575547, "loss": 0.463, "step": 14200 }, { "epoch": 9.860168237452172e-14, "grad_norm": 1.7394859790802002, "learning_rate": 0.0001448949399799106, "loss": 0.4458, "step": 14210 }, { "epoch": 9.867107131356079e-14, "grad_norm": 1.8306599855422974, "learning_rate": 0.00014488668171201477, "loss": 0.4605, "step": 14220 }, { "epoch": 9.874046025259986e-14, "grad_norm": 1.8661525249481201, "learning_rate": 0.0001448784170329167, "loss": 0.4468, "step": 14230 }, { "epoch": 9.880984919163893e-14, "grad_norm": 1.7209590673446655, "learning_rate": 0.00014487014594346574, "loss": 0.4728, "step": 14240 }, { "epoch": 9.8879238130678e-14, "grad_norm": 1.931809425354004, "learning_rate": 0.00014486186844451184, "loss": 0.4211, "step": 14250 }, { "epoch": 9.894862706971708e-14, "grad_norm": 1.5931119918823242, "learning_rate": 0.00014485358453690563, "loss": 0.4811, "step": 14260 }, { "epoch": 9.901801600875615e-14, "grad_norm": 1.7905843257904053, "learning_rate": 0.00014484529422149848, "loss": 0.5061, "step": 14270 }, { "epoch": 9.908740494779522e-14, "grad_norm": 2.017944812774658, "learning_rate": 0.00014483699749914225, "loss": 0.4263, "step": 14280 }, { "epoch": 9.915679388683429e-14, "grad_norm": 1.8854979276657104, "learning_rate": 0.0001448286943706896, "loss": 0.4156, "step": 14290 }, { "epoch": 9.922618282587337e-14, "grad_norm": 1.786467432975769, "learning_rate": 0.00014482038483699385, "loss": 0.4553, "step": 14300 }, { "epoch": 9.929557176491244e-14, "grad_norm": 1.83211350440979, "learning_rate": 0.00014481206889890887, "loss": 0.452, "step": 14310 }, { "epoch": 9.936496070395151e-14, "grad_norm": 1.968900203704834, "learning_rate": 0.0001448037465572893, "loss": 0.4115, "step": 14320 }, { "epoch": 9.943434964299058e-14, "grad_norm": 1.9846101999282837, "learning_rate": 0.00014479541781299034, "loss": 0.4544, "step": 14330 }, { "epoch": 9.950373858202965e-14, "grad_norm": 1.622365117073059, "learning_rate": 0.00014478708266686796, "loss": 0.4633, "step": 14340 }, { "epoch": 9.957312752106873e-14, "grad_norm": 1.8244438171386719, "learning_rate": 0.00014477874111977867, "loss": 0.4356, "step": 14350 }, { "epoch": 9.96425164601078e-14, "grad_norm": 1.735727310180664, "learning_rate": 0.0001447703931725797, "loss": 0.4345, "step": 14360 }, { "epoch": 9.971190539914687e-14, "grad_norm": 1.932239055633545, "learning_rate": 0.000144762038826129, "loss": 0.4378, "step": 14370 }, { "epoch": 9.978129433818594e-14, "grad_norm": 1.8763742446899414, "learning_rate": 0.00014475367808128498, "loss": 0.4647, "step": 14380 }, { "epoch": 9.985068327722502e-14, "grad_norm": 1.8195712566375732, "learning_rate": 0.00014474531093890697, "loss": 0.463, "step": 14390 }, { "epoch": 9.992007221626409e-14, "grad_norm": 1.5656613111495972, "learning_rate": 0.00014473693739985475, "loss": 0.4304, "step": 14400 }, { "epoch": 9.998946115530316e-14, "grad_norm": 1.7358012199401855, "learning_rate": 0.00014472855746498886, "loss": 0.449, "step": 14410 }, { "epoch": 1.0005885009434223e-13, "grad_norm": 1.837323546409607, "learning_rate": 0.00014472017113517042, "loss": 0.4671, "step": 14420 }, { "epoch": 1.001282390333813e-13, "grad_norm": 1.4301447868347168, "learning_rate": 0.00014471177841126132, "loss": 0.4901, "step": 14430 }, { "epoch": 1.0019762797242038e-13, "grad_norm": 1.3110641241073608, "learning_rate": 0.000144703379294124, "loss": 0.4016, "step": 14440 }, { "epoch": 1.0026701691145945e-13, "grad_norm": 1.7230404615402222, "learning_rate": 0.00014469497378462163, "loss": 0.4238, "step": 14450 }, { "epoch": 1.0033640585049852e-13, "grad_norm": 1.5877337455749512, "learning_rate": 0.000144686561883618, "loss": 0.4595, "step": 14460 }, { "epoch": 1.004057947895376e-13, "grad_norm": 1.40691339969635, "learning_rate": 0.00014467814359197753, "loss": 0.4176, "step": 14470 }, { "epoch": 1.0047518372857667e-13, "grad_norm": 1.7740713357925415, "learning_rate": 0.0001446697189105654, "loss": 0.4255, "step": 14480 }, { "epoch": 1.0054457266761574e-13, "grad_norm": 1.8904603719711304, "learning_rate": 0.0001446612878402473, "loss": 0.4375, "step": 14490 }, { "epoch": 1.0061396160665481e-13, "grad_norm": 1.653802514076233, "learning_rate": 0.00014465285038188969, "loss": 0.4302, "step": 14500 }, { "epoch": 1.0068335054569388e-13, "grad_norm": 1.9095039367675781, "learning_rate": 0.00014464440653635968, "loss": 0.4169, "step": 14510 }, { "epoch": 1.0075273948473296e-13, "grad_norm": 1.9127061367034912, "learning_rate": 0.00014463595630452493, "loss": 0.4414, "step": 14520 }, { "epoch": 1.0082212842377203e-13, "grad_norm": 2.0750606060028076, "learning_rate": 0.00014462749968725388, "loss": 0.4168, "step": 14530 }, { "epoch": 1.008915173628111e-13, "grad_norm": 1.7789204120635986, "learning_rate": 0.00014461903668541558, "loss": 0.4376, "step": 14540 }, { "epoch": 1.0096090630185017e-13, "grad_norm": 1.7484546899795532, "learning_rate": 0.00014461056729987975, "loss": 0.4779, "step": 14550 }, { "epoch": 1.0103029524088925e-13, "grad_norm": 1.6340715885162354, "learning_rate": 0.0001446020915315167, "loss": 0.4011, "step": 14560 }, { "epoch": 1.0109968417992832e-13, "grad_norm": 1.7629475593566895, "learning_rate": 0.0001445936093811975, "loss": 0.4731, "step": 14570 }, { "epoch": 1.0116907311896739e-13, "grad_norm": 2.139462947845459, "learning_rate": 0.00014458512084979382, "loss": 0.4766, "step": 14580 }, { "epoch": 1.0123846205800646e-13, "grad_norm": 1.8401265144348145, "learning_rate": 0.00014457662593817796, "loss": 0.4314, "step": 14590 }, { "epoch": 1.0130785099704553e-13, "grad_norm": 1.6141939163208008, "learning_rate": 0.0001445681246472229, "loss": 0.3917, "step": 14600 }, { "epoch": 1.0137723993608461e-13, "grad_norm": 2.2529399394989014, "learning_rate": 0.0001445596169778023, "loss": 0.4078, "step": 14610 }, { "epoch": 1.0144662887512368e-13, "grad_norm": 2.071216106414795, "learning_rate": 0.00014455110293079047, "loss": 0.4341, "step": 14620 }, { "epoch": 1.0151601781416275e-13, "grad_norm": 1.748245120048523, "learning_rate": 0.0001445425825070623, "loss": 0.4069, "step": 14630 }, { "epoch": 1.0158540675320182e-13, "grad_norm": 1.633109450340271, "learning_rate": 0.00014453405570749348, "loss": 0.3991, "step": 14640 }, { "epoch": 1.016547956922409e-13, "grad_norm": 2.1189966201782227, "learning_rate": 0.0001445255225329602, "loss": 0.4755, "step": 14650 }, { "epoch": 1.0172418463127997e-13, "grad_norm": 1.6352012157440186, "learning_rate": 0.00014451698298433943, "loss": 0.3827, "step": 14660 }, { "epoch": 1.0179357357031904e-13, "grad_norm": 1.4167475700378418, "learning_rate": 0.0001445084370625087, "loss": 0.4094, "step": 14670 }, { "epoch": 1.0186296250935811e-13, "grad_norm": 1.7471041679382324, "learning_rate": 0.00014449988476834626, "loss": 0.4199, "step": 14680 }, { "epoch": 1.0193235144839718e-13, "grad_norm": 1.6175038814544678, "learning_rate": 0.00014449132610273098, "loss": 0.4245, "step": 14690 }, { "epoch": 1.0200174038743626e-13, "grad_norm": 1.6283847093582153, "learning_rate": 0.0001444827610665424, "loss": 0.4156, "step": 14700 }, { "epoch": 1.0207112932647533e-13, "grad_norm": 1.6541292667388916, "learning_rate": 0.00014447418966066072, "loss": 0.4642, "step": 14710 }, { "epoch": 1.021405182655144e-13, "grad_norm": 1.9426277875900269, "learning_rate": 0.00014446561188596676, "loss": 0.4039, "step": 14720 }, { "epoch": 1.0220990720455347e-13, "grad_norm": 1.9671211242675781, "learning_rate": 0.00014445702774334206, "loss": 0.4349, "step": 14730 }, { "epoch": 1.0227929614359255e-13, "grad_norm": 1.702793836593628, "learning_rate": 0.00014444843723366871, "loss": 0.4661, "step": 14740 }, { "epoch": 1.0234868508263162e-13, "grad_norm": 1.9742450714111328, "learning_rate": 0.00014443984035782958, "loss": 0.4138, "step": 14750 }, { "epoch": 1.0241807402167069e-13, "grad_norm": 1.8860503435134888, "learning_rate": 0.00014443123711670812, "loss": 0.3985, "step": 14760 }, { "epoch": 1.0248746296070976e-13, "grad_norm": 1.888893723487854, "learning_rate": 0.00014442262751118845, "loss": 0.4231, "step": 14770 }, { "epoch": 1.0255685189974884e-13, "grad_norm": 2.0754287242889404, "learning_rate": 0.0001444140115421553, "loss": 0.4253, "step": 14780 }, { "epoch": 1.0262624083878791e-13, "grad_norm": 1.8352080583572388, "learning_rate": 0.00014440538921049414, "loss": 0.3666, "step": 14790 }, { "epoch": 1.0269562977782698e-13, "grad_norm": 1.9858156442642212, "learning_rate": 0.00014439676051709104, "loss": 0.4281, "step": 14800 }, { "epoch": 1.0276501871686605e-13, "grad_norm": 2.0522804260253906, "learning_rate": 0.0001443881254628327, "loss": 0.4276, "step": 14810 }, { "epoch": 1.0283440765590512e-13, "grad_norm": 1.626975655555725, "learning_rate": 0.00014437948404860657, "loss": 0.391, "step": 14820 }, { "epoch": 1.029037965949442e-13, "grad_norm": 1.627425193786621, "learning_rate": 0.0001443708362753006, "loss": 0.3999, "step": 14830 }, { "epoch": 1.0297318553398327e-13, "grad_norm": 1.8121074438095093, "learning_rate": 0.00014436218214380358, "loss": 0.4117, "step": 14840 }, { "epoch": 1.0304257447302234e-13, "grad_norm": 1.5409117937088013, "learning_rate": 0.00014435352165500482, "loss": 0.4151, "step": 14850 }, { "epoch": 1.0311196341206141e-13, "grad_norm": 1.8921139240264893, "learning_rate": 0.00014434485480979426, "loss": 0.4395, "step": 14860 }, { "epoch": 1.0318135235110049e-13, "grad_norm": 1.8083405494689941, "learning_rate": 0.00014433618160906265, "loss": 0.3904, "step": 14870 }, { "epoch": 1.0325074129013956e-13, "grad_norm": 2.004055976867676, "learning_rate": 0.0001443275020537012, "loss": 0.4428, "step": 14880 }, { "epoch": 1.0332013022917863e-13, "grad_norm": 1.9738460779190063, "learning_rate": 0.000144318816144602, "loss": 0.438, "step": 14890 }, { "epoch": 1.033895191682177e-13, "grad_norm": 2.0826165676116943, "learning_rate": 0.00014431012388265753, "loss": 0.4462, "step": 14900 }, { "epoch": 1.0345890810725678e-13, "grad_norm": 2.447376012802124, "learning_rate": 0.00014430142526876112, "loss": 0.4106, "step": 14910 }, { "epoch": 1.0352829704629585e-13, "grad_norm": 1.4299986362457275, "learning_rate": 0.00014429272030380667, "loss": 0.3754, "step": 14920 }, { "epoch": 1.0359768598533492e-13, "grad_norm": 1.3793294429779053, "learning_rate": 0.00014428400898868876, "loss": 0.4261, "step": 14930 }, { "epoch": 1.0366707492437399e-13, "grad_norm": 1.6968472003936768, "learning_rate": 0.0001442752913243026, "loss": 0.3995, "step": 14940 }, { "epoch": 1.0373646386341306e-13, "grad_norm": 1.8891446590423584, "learning_rate": 0.00014426656731154406, "loss": 0.3984, "step": 14950 }, { "epoch": 1.0380585280245214e-13, "grad_norm": 1.9520779848098755, "learning_rate": 0.00014425783695130973, "loss": 0.4177, "step": 14960 }, { "epoch": 1.0387524174149121e-13, "grad_norm": 1.9332494735717773, "learning_rate": 0.00014424910024449673, "loss": 0.4003, "step": 14970 }, { "epoch": 1.0394463068053028e-13, "grad_norm": 1.9502410888671875, "learning_rate": 0.00014424035719200285, "loss": 0.4147, "step": 14980 }, { "epoch": 1.0401401961956935e-13, "grad_norm": 1.536048173904419, "learning_rate": 0.00014423160779472672, "loss": 0.389, "step": 14990 }, { "epoch": 1.0408340855860843e-13, "grad_norm": 2.078852415084839, "learning_rate": 0.00014422285205356733, "loss": 0.3972, "step": 15000 }, { "epoch": 1.041527974976475e-13, "grad_norm": 1.808544635772705, "learning_rate": 0.00014421408996942453, "loss": 0.3945, "step": 15010 }, { "epoch": 1.0422218643668657e-13, "grad_norm": 1.7157059907913208, "learning_rate": 0.00014420532154319878, "loss": 0.438, "step": 15020 }, { "epoch": 1.0429157537572564e-13, "grad_norm": 1.915388584136963, "learning_rate": 0.00014419654677579115, "loss": 0.3771, "step": 15030 }, { "epoch": 1.0436096431476471e-13, "grad_norm": 1.5922266244888306, "learning_rate": 0.00014418776566810335, "loss": 0.445, "step": 15040 }, { "epoch": 1.0443035325380379e-13, "grad_norm": 1.662207007408142, "learning_rate": 0.00014417897822103785, "loss": 0.4536, "step": 15050 }, { "epoch": 1.0449974219284286e-13, "grad_norm": 1.5343735218048096, "learning_rate": 0.00014417018443549763, "loss": 0.3891, "step": 15060 }, { "epoch": 1.0456913113188193e-13, "grad_norm": 1.8765058517456055, "learning_rate": 0.0001441613843123864, "loss": 0.3892, "step": 15070 }, { "epoch": 1.04638520070921e-13, "grad_norm": 2.1808178424835205, "learning_rate": 0.00014415257785260852, "loss": 0.4194, "step": 15080 }, { "epoch": 1.0470790900996008e-13, "grad_norm": 1.7446709871292114, "learning_rate": 0.00014414376505706905, "loss": 0.3935, "step": 15090 }, { "epoch": 1.0477729794899915e-13, "grad_norm": 1.4954153299331665, "learning_rate": 0.00014413494592667352, "loss": 0.3641, "step": 15100 }, { "epoch": 1.0484668688803822e-13, "grad_norm": 1.8224188089370728, "learning_rate": 0.00014412612046232833, "loss": 0.3806, "step": 15110 }, { "epoch": 1.0491607582707729e-13, "grad_norm": 1.5136107206344604, "learning_rate": 0.0001441172886649404, "loss": 0.4409, "step": 15120 }, { "epoch": 1.0498546476611637e-13, "grad_norm": 1.6719919443130493, "learning_rate": 0.00014410845053541735, "loss": 0.3763, "step": 15130 }, { "epoch": 1.0505485370515544e-13, "grad_norm": 1.449815034866333, "learning_rate": 0.0001440996060746674, "loss": 0.3907, "step": 15140 }, { "epoch": 1.0512424264419451e-13, "grad_norm": 1.5000842809677124, "learning_rate": 0.00014409075528359948, "loss": 0.3952, "step": 15150 }, { "epoch": 1.0519363158323358e-13, "grad_norm": 2.015023708343506, "learning_rate": 0.00014408189816312314, "loss": 0.4105, "step": 15160 }, { "epoch": 1.0526302052227265e-13, "grad_norm": 2.2000865936279297, "learning_rate": 0.00014407303471414862, "loss": 0.3978, "step": 15170 }, { "epoch": 1.0533240946131173e-13, "grad_norm": 1.3515688180923462, "learning_rate": 0.00014406416493758674, "loss": 0.3925, "step": 15180 }, { "epoch": 1.054017984003508e-13, "grad_norm": 1.7239675521850586, "learning_rate": 0.000144055288834349, "loss": 0.394, "step": 15190 }, { "epoch": 1.0547118733938987e-13, "grad_norm": 1.726251482963562, "learning_rate": 0.00014404640640534758, "loss": 0.4192, "step": 15200 }, { "epoch": 1.0554057627842894e-13, "grad_norm": 1.683341145515442, "learning_rate": 0.00014403751765149528, "loss": 0.4396, "step": 15210 }, { "epoch": 1.0560996521746802e-13, "grad_norm": 1.481836199760437, "learning_rate": 0.00014402862257370557, "loss": 0.3774, "step": 15220 }, { "epoch": 1.0567935415650709e-13, "grad_norm": 1.601854681968689, "learning_rate": 0.00014401972117289254, "loss": 0.3845, "step": 15230 }, { "epoch": 1.0574874309554616e-13, "grad_norm": 1.885923147201538, "learning_rate": 0.00014401081344997095, "loss": 0.4241, "step": 15240 }, { "epoch": 1.0581813203458523e-13, "grad_norm": 1.4333876371383667, "learning_rate": 0.0001440018994058562, "loss": 0.3706, "step": 15250 }, { "epoch": 1.058875209736243e-13, "grad_norm": 1.8450332880020142, "learning_rate": 0.0001439929790414644, "loss": 0.3534, "step": 15260 }, { "epoch": 1.0595690991266338e-13, "grad_norm": 1.6301060914993286, "learning_rate": 0.00014398405235771218, "loss": 0.3859, "step": 15270 }, { "epoch": 1.0602629885170245e-13, "grad_norm": 1.589263916015625, "learning_rate": 0.00014397511935551694, "loss": 0.4038, "step": 15280 }, { "epoch": 1.0609568779074152e-13, "grad_norm": 1.599394679069519, "learning_rate": 0.0001439661800357967, "loss": 0.3707, "step": 15290 }, { "epoch": 1.061650767297806e-13, "grad_norm": 2.1095094680786133, "learning_rate": 0.00014395723439947008, "loss": 0.4158, "step": 15300 }, { "epoch": 1.0623446566881967e-13, "grad_norm": 1.8801119327545166, "learning_rate": 0.00014394828244745637, "loss": 0.3776, "step": 15310 }, { "epoch": 1.0630385460785874e-13, "grad_norm": 1.8907523155212402, "learning_rate": 0.00014393932418067554, "loss": 0.4059, "step": 15320 }, { "epoch": 1.0637324354689781e-13, "grad_norm": 2.047545909881592, "learning_rate": 0.0001439303596000482, "loss": 0.4213, "step": 15330 }, { "epoch": 1.0644263248593688e-13, "grad_norm": 1.8326255083084106, "learning_rate": 0.00014392138870649563, "loss": 0.3525, "step": 15340 }, { "epoch": 1.0651202142497596e-13, "grad_norm": 2.19541597366333, "learning_rate": 0.0001439124115009397, "loss": 0.4365, "step": 15350 }, { "epoch": 1.0658141036401503e-13, "grad_norm": 2.3077967166900635, "learning_rate": 0.0001439034279843029, "loss": 0.4126, "step": 15360 }, { "epoch": 1.066507993030541e-13, "grad_norm": 1.7208411693572998, "learning_rate": 0.0001438944381575085, "loss": 0.3912, "step": 15370 }, { "epoch": 1.0672018824209317e-13, "grad_norm": 1.7019726037979126, "learning_rate": 0.00014388544202148035, "loss": 0.3756, "step": 15380 }, { "epoch": 1.0678957718113224e-13, "grad_norm": 1.9980783462524414, "learning_rate": 0.0001438764395771429, "loss": 0.3644, "step": 15390 }, { "epoch": 1.0685896612017132e-13, "grad_norm": 2.026555061340332, "learning_rate": 0.0001438674308254213, "loss": 0.4113, "step": 15400 }, { "epoch": 1.0692835505921039e-13, "grad_norm": 1.6188535690307617, "learning_rate": 0.00014385841576724135, "loss": 0.3642, "step": 15410 }, { "epoch": 1.0699774399824946e-13, "grad_norm": 1.312746524810791, "learning_rate": 0.0001438493944035295, "loss": 0.3633, "step": 15420 }, { "epoch": 1.0706713293728853e-13, "grad_norm": 1.6944844722747803, "learning_rate": 0.0001438403667352128, "loss": 0.3951, "step": 15430 }, { "epoch": 1.071365218763276e-13, "grad_norm": 1.7134639024734497, "learning_rate": 0.00014383133276321902, "loss": 0.3637, "step": 15440 }, { "epoch": 1.0720591081536668e-13, "grad_norm": 1.81497061252594, "learning_rate": 0.00014382229248847652, "loss": 0.3964, "step": 15450 }, { "epoch": 1.0727529975440575e-13, "grad_norm": 1.8634607791900635, "learning_rate": 0.00014381324591191433, "loss": 0.3782, "step": 15460 }, { "epoch": 1.0734468869344482e-13, "grad_norm": 2.119349718093872, "learning_rate": 0.00014380419303446212, "loss": 0.402, "step": 15470 }, { "epoch": 1.074140776324839e-13, "grad_norm": 1.758280873298645, "learning_rate": 0.00014379513385705023, "loss": 0.3799, "step": 15480 }, { "epoch": 1.0748346657152297e-13, "grad_norm": 2.039473533630371, "learning_rate": 0.00014378606838060963, "loss": 0.4103, "step": 15490 }, { "epoch": 1.0755285551056204e-13, "grad_norm": 1.983949065208435, "learning_rate": 0.00014377699660607195, "loss": 0.355, "step": 15500 }, { "epoch": 1.0762224444960111e-13, "grad_norm": 1.592117428779602, "learning_rate": 0.00014376791853436946, "loss": 0.4189, "step": 15510 }, { "epoch": 1.0769163338864018e-13, "grad_norm": 1.5414996147155762, "learning_rate": 0.000143758834166435, "loss": 0.442, "step": 15520 }, { "epoch": 1.0776102232767926e-13, "grad_norm": 1.974075198173523, "learning_rate": 0.0001437497435032022, "loss": 0.366, "step": 15530 }, { "epoch": 1.0783041126671833e-13, "grad_norm": 2.187056541442871, "learning_rate": 0.00014374064654560525, "loss": 0.3638, "step": 15540 }, { "epoch": 1.078998002057574e-13, "grad_norm": 1.6903893947601318, "learning_rate": 0.00014373154329457903, "loss": 0.3975, "step": 15550 }, { "epoch": 1.0796918914479647e-13, "grad_norm": 1.8506098985671997, "learning_rate": 0.00014372243375105899, "loss": 0.3763, "step": 15560 }, { "epoch": 1.0803857808383555e-13, "grad_norm": 1.844221591949463, "learning_rate": 0.0001437133179159813, "loss": 0.3424, "step": 15570 }, { "epoch": 1.0810796702287462e-13, "grad_norm": 1.931058406829834, "learning_rate": 0.00014370419579028278, "loss": 0.3691, "step": 15580 }, { "epoch": 1.0817735596191369e-13, "grad_norm": 1.4651002883911133, "learning_rate": 0.00014369506737490082, "loss": 0.379, "step": 15590 }, { "epoch": 1.0824674490095276e-13, "grad_norm": 1.8757349252700806, "learning_rate": 0.00014368593267077357, "loss": 0.3581, "step": 15600 }, { "epoch": 1.0831613383999183e-13, "grad_norm": 1.9355021715164185, "learning_rate": 0.0001436767916788397, "loss": 0.3708, "step": 15610 }, { "epoch": 1.0838552277903091e-13, "grad_norm": 1.8483262062072754, "learning_rate": 0.0001436676444000386, "loss": 0.38, "step": 15620 }, { "epoch": 1.0845491171806998e-13, "grad_norm": 1.758249044418335, "learning_rate": 0.00014365849083531034, "loss": 0.4144, "step": 15630 }, { "epoch": 1.0852430065710905e-13, "grad_norm": 1.6624515056610107, "learning_rate": 0.00014364933098559557, "loss": 0.3858, "step": 15640 }, { "epoch": 1.0859368959614812e-13, "grad_norm": 1.5106526613235474, "learning_rate": 0.00014364016485183558, "loss": 0.3679, "step": 15650 }, { "epoch": 1.086630785351872e-13, "grad_norm": 1.8545869588851929, "learning_rate": 0.00014363099243497237, "loss": 0.3809, "step": 15660 }, { "epoch": 1.0873246747422627e-13, "grad_norm": 1.881779670715332, "learning_rate": 0.00014362181373594852, "loss": 0.4057, "step": 15670 }, { "epoch": 1.0880185641326534e-13, "grad_norm": 1.2863056659698486, "learning_rate": 0.0001436126287557073, "loss": 0.4185, "step": 15680 }, { "epoch": 1.0887124535230441e-13, "grad_norm": 1.309495210647583, "learning_rate": 0.0001436034374951926, "loss": 0.344, "step": 15690 }, { "epoch": 1.0894063429134349e-13, "grad_norm": 2.0055041313171387, "learning_rate": 0.00014359423995534895, "loss": 0.3771, "step": 15700 }, { "epoch": 1.0901002323038256e-13, "grad_norm": 1.3381903171539307, "learning_rate": 0.00014358503613712156, "loss": 0.3952, "step": 15710 }, { "epoch": 1.0907941216942163e-13, "grad_norm": 1.6012518405914307, "learning_rate": 0.00014357582604145633, "loss": 0.3568, "step": 15720 }, { "epoch": 1.091488011084607e-13, "grad_norm": 1.850258469581604, "learning_rate": 0.0001435666096692996, "loss": 0.3598, "step": 15730 }, { "epoch": 1.0921819004749977e-13, "grad_norm": 1.8361762762069702, "learning_rate": 0.0001435573870215986, "loss": 0.3617, "step": 15740 }, { "epoch": 1.0928757898653885e-13, "grad_norm": 1.7903783321380615, "learning_rate": 0.00014354815809930108, "loss": 0.3649, "step": 15750 }, { "epoch": 1.0935696792557792e-13, "grad_norm": 1.994302749633789, "learning_rate": 0.00014353892290335543, "loss": 0.3554, "step": 15760 }, { "epoch": 1.0942635686461699e-13, "grad_norm": 1.993088960647583, "learning_rate": 0.00014352968143471076, "loss": 0.3806, "step": 15770 }, { "epoch": 1.0949574580365606e-13, "grad_norm": 2.0833475589752197, "learning_rate": 0.00014352043369431671, "loss": 0.3629, "step": 15780 }, { "epoch": 1.0956513474269514e-13, "grad_norm": 1.6784272193908691, "learning_rate": 0.0001435111796831237, "loss": 0.3873, "step": 15790 }, { "epoch": 1.0963452368173421e-13, "grad_norm": 1.7375853061676025, "learning_rate": 0.00014350191940208263, "loss": 0.402, "step": 15800 }, { "epoch": 1.0970391262077328e-13, "grad_norm": 1.5425752401351929, "learning_rate": 0.00014349265285214526, "loss": 0.3358, "step": 15810 }, { "epoch": 1.0977330155981235e-13, "grad_norm": 1.9172592163085938, "learning_rate": 0.00014348338003426374, "loss": 0.4105, "step": 15820 }, { "epoch": 1.0984269049885143e-13, "grad_norm": 2.2164041996002197, "learning_rate": 0.0001434741009493911, "loss": 0.4137, "step": 15830 }, { "epoch": 1.099120794378905e-13, "grad_norm": 1.84764564037323, "learning_rate": 0.00014346481559848086, "loss": 0.3556, "step": 15840 }, { "epoch": 1.0998146837692957e-13, "grad_norm": 1.4959100484848022, "learning_rate": 0.00014345552398248724, "loss": 0.3282, "step": 15850 }, { "epoch": 1.1005085731596864e-13, "grad_norm": 2.1046578884124756, "learning_rate": 0.0001434462261023651, "loss": 0.3655, "step": 15860 }, { "epoch": 1.1012024625500771e-13, "grad_norm": 2.004833221435547, "learning_rate": 0.00014343692195907, "loss": 0.3685, "step": 15870 }, { "epoch": 1.1018963519404679e-13, "grad_norm": 1.9627074003219604, "learning_rate": 0.00014342761155355796, "loss": 0.3575, "step": 15880 }, { "epoch": 1.1025902413308586e-13, "grad_norm": 1.5124528408050537, "learning_rate": 0.0001434182948867859, "loss": 0.336, "step": 15890 }, { "epoch": 1.1032841307212493e-13, "grad_norm": 2.1452324390411377, "learning_rate": 0.00014340897195971115, "loss": 0.4026, "step": 15900 }, { "epoch": 1.10397802011164e-13, "grad_norm": 2.041642189025879, "learning_rate": 0.00014339964277329184, "loss": 0.3322, "step": 15910 }, { "epoch": 1.1046719095020308e-13, "grad_norm": 1.270796775817871, "learning_rate": 0.00014339030732848668, "loss": 0.3548, "step": 15920 }, { "epoch": 1.1053657988924215e-13, "grad_norm": 1.8269917964935303, "learning_rate": 0.00014338096562625506, "loss": 0.3663, "step": 15930 }, { "epoch": 1.1060596882828122e-13, "grad_norm": 1.7731302976608276, "learning_rate": 0.0001433716176675569, "loss": 0.3837, "step": 15940 }, { "epoch": 1.1067535776732029e-13, "grad_norm": 1.5817476511001587, "learning_rate": 0.00014336226345335298, "loss": 0.3586, "step": 15950 }, { "epoch": 1.1074474670635936e-13, "grad_norm": 1.578827142715454, "learning_rate": 0.00014335290298460445, "loss": 0.3935, "step": 15960 }, { "epoch": 1.1081413564539844e-13, "grad_norm": 1.7338427305221558, "learning_rate": 0.00014334353626227332, "loss": 0.3362, "step": 15970 }, { "epoch": 1.1088352458443751e-13, "grad_norm": 2.019024610519409, "learning_rate": 0.00014333416328732217, "loss": 0.3758, "step": 15980 }, { "epoch": 1.1095291352347658e-13, "grad_norm": 1.6736891269683838, "learning_rate": 0.0001433247840607142, "loss": 0.4142, "step": 15990 }, { "epoch": 1.1102230246251565e-13, "grad_norm": 1.8206791877746582, "learning_rate": 0.00014331539858341327, "loss": 0.3367, "step": 16000 }, { "epoch": 1.1109169140155473e-13, "grad_norm": 1.7804083824157715, "learning_rate": 0.0001433060068563839, "loss": 0.3358, "step": 16010 }, { "epoch": 1.111610803405938e-13, "grad_norm": 2.0010719299316406, "learning_rate": 0.00014329660888059121, "loss": 0.3715, "step": 16020 }, { "epoch": 1.1123046927963287e-13, "grad_norm": 1.973218560218811, "learning_rate": 0.00014328720465700103, "loss": 0.3584, "step": 16030 }, { "epoch": 1.1129985821867194e-13, "grad_norm": 1.9603890180587769, "learning_rate": 0.00014327779418657973, "loss": 0.3251, "step": 16040 }, { "epoch": 1.1136924715771102e-13, "grad_norm": 1.7948505878448486, "learning_rate": 0.00014326837747029447, "loss": 0.3528, "step": 16050 }, { "epoch": 1.1143863609675009e-13, "grad_norm": 2.0103893280029297, "learning_rate": 0.00014325895450911284, "loss": 0.3591, "step": 16060 }, { "epoch": 1.1150802503578916e-13, "grad_norm": 1.7974759340286255, "learning_rate": 0.0001432495253040033, "loss": 0.334, "step": 16070 }, { "epoch": 1.1157741397482823e-13, "grad_norm": 1.5160692930221558, "learning_rate": 0.00014324008985593483, "loss": 0.3466, "step": 16080 }, { "epoch": 1.116468029138673e-13, "grad_norm": 1.9029220342636108, "learning_rate": 0.00014323064816587703, "loss": 0.3552, "step": 16090 }, { "epoch": 1.1171619185290638e-13, "grad_norm": 1.6434223651885986, "learning_rate": 0.00014322120023480022, "loss": 0.3756, "step": 16100 }, { "epoch": 1.1178558079194545e-13, "grad_norm": 1.814003586769104, "learning_rate": 0.0001432117460636753, "loss": 0.3681, "step": 16110 }, { "epoch": 1.1185496973098452e-13, "grad_norm": 1.5957978963851929, "learning_rate": 0.00014320228565347385, "loss": 0.3329, "step": 16120 }, { "epoch": 1.119243586700236e-13, "grad_norm": 1.8402522802352905, "learning_rate": 0.00014319281900516804, "loss": 0.3671, "step": 16130 }, { "epoch": 1.1199374760906267e-13, "grad_norm": 1.9919283390045166, "learning_rate": 0.00014318334611973076, "loss": 0.374, "step": 16140 }, { "epoch": 1.1206313654810174e-13, "grad_norm": 1.607854962348938, "learning_rate": 0.00014317386699813544, "loss": 0.3884, "step": 16150 }, { "epoch": 1.1213252548714081e-13, "grad_norm": 2.257175922393799, "learning_rate": 0.00014316438164135628, "loss": 0.3375, "step": 16160 }, { "epoch": 1.1220191442617988e-13, "grad_norm": 1.5334290266036987, "learning_rate": 0.00014315489005036798, "loss": 0.3217, "step": 16170 }, { "epoch": 1.1227130336521896e-13, "grad_norm": 1.2673588991165161, "learning_rate": 0.00014314539222614602, "loss": 0.361, "step": 16180 }, { "epoch": 1.1234069230425803e-13, "grad_norm": 1.6288866996765137, "learning_rate": 0.00014313588816966638, "loss": 0.3448, "step": 16190 }, { "epoch": 1.124100812432971e-13, "grad_norm": 1.928484559059143, "learning_rate": 0.00014312637788190576, "loss": 0.3561, "step": 16200 }, { "epoch": 1.1247947018233617e-13, "grad_norm": 1.9828760623931885, "learning_rate": 0.00014311686136384154, "loss": 0.345, "step": 16210 }, { "epoch": 1.1254885912137524e-13, "grad_norm": 1.95497465133667, "learning_rate": 0.00014310733861645165, "loss": 0.3391, "step": 16220 }, { "epoch": 1.1261824806041432e-13, "grad_norm": 2.1265969276428223, "learning_rate": 0.0001430978096407147, "loss": 0.3625, "step": 16230 }, { "epoch": 1.1268763699945339e-13, "grad_norm": 1.6356000900268555, "learning_rate": 0.00014308827443760996, "loss": 0.3417, "step": 16240 }, { "epoch": 1.1275702593849246e-13, "grad_norm": 1.96791672706604, "learning_rate": 0.0001430787330081173, "loss": 0.3454, "step": 16250 }, { "epoch": 1.1282641487753153e-13, "grad_norm": 1.73987877368927, "learning_rate": 0.00014306918535321727, "loss": 0.3538, "step": 16260 }, { "epoch": 1.128958038165706e-13, "grad_norm": 1.652331829071045, "learning_rate": 0.000143059631473891, "loss": 0.3703, "step": 16270 }, { "epoch": 1.1296519275560968e-13, "grad_norm": 1.6340163946151733, "learning_rate": 0.00014305007137112037, "loss": 0.3217, "step": 16280 }, { "epoch": 1.1303458169464875e-13, "grad_norm": 1.5712095499038696, "learning_rate": 0.00014304050504588777, "loss": 0.3794, "step": 16290 }, { "epoch": 1.1310397063368782e-13, "grad_norm": 1.8094518184661865, "learning_rate": 0.00014303093249917634, "loss": 0.4008, "step": 16300 }, { "epoch": 1.131733595727269e-13, "grad_norm": 1.3709641695022583, "learning_rate": 0.00014302135373196973, "loss": 0.3289, "step": 16310 }, { "epoch": 1.1324274851176597e-13, "grad_norm": 1.702418565750122, "learning_rate": 0.00014301176874525234, "loss": 0.3318, "step": 16320 }, { "epoch": 1.1331213745080504e-13, "grad_norm": 2.1709985733032227, "learning_rate": 0.00014300217754000923, "loss": 0.3684, "step": 16330 }, { "epoch": 1.1338152638984411e-13, "grad_norm": 1.7453473806381226, "learning_rate": 0.000142992580117226, "loss": 0.3272, "step": 16340 }, { "epoch": 1.1345091532888318e-13, "grad_norm": 1.7947686910629272, "learning_rate": 0.00014298297647788894, "loss": 0.3345, "step": 16350 }, { "epoch": 1.1352030426792226e-13, "grad_norm": 1.8869065046310425, "learning_rate": 0.00014297336662298494, "loss": 0.3368, "step": 16360 }, { "epoch": 1.1358969320696133e-13, "grad_norm": 1.482727289199829, "learning_rate": 0.00014296375055350164, "loss": 0.3713, "step": 16370 }, { "epoch": 1.136590821460004e-13, "grad_norm": 1.8203368186950684, "learning_rate": 0.0001429541282704272, "loss": 0.3272, "step": 16380 }, { "epoch": 1.1372847108503947e-13, "grad_norm": 1.48373281955719, "learning_rate": 0.00014294449977475044, "loss": 0.3463, "step": 16390 }, { "epoch": 1.1379786002407855e-13, "grad_norm": 1.5540413856506348, "learning_rate": 0.00014293486506746083, "loss": 0.3442, "step": 16400 }, { "epoch": 1.1386724896311762e-13, "grad_norm": 1.9068033695220947, "learning_rate": 0.00014292522414954855, "loss": 0.3637, "step": 16410 }, { "epoch": 1.139366379021567e-13, "grad_norm": 2.208906650543213, "learning_rate": 0.00014291557702200426, "loss": 0.3526, "step": 16420 }, { "epoch": 1.1400602684119576e-13, "grad_norm": 1.4125510454177856, "learning_rate": 0.00014290592368581943, "loss": 0.3432, "step": 16430 }, { "epoch": 1.1407541578023483e-13, "grad_norm": 1.655966877937317, "learning_rate": 0.00014289626414198608, "loss": 0.331, "step": 16440 }, { "epoch": 1.141448047192739e-13, "grad_norm": 1.675534725189209, "learning_rate": 0.00014288659839149686, "loss": 0.3595, "step": 16450 }, { "epoch": 1.1421419365831298e-13, "grad_norm": 1.941994547843933, "learning_rate": 0.00014287692643534508, "loss": 0.3968, "step": 16460 }, { "epoch": 1.1428358259735205e-13, "grad_norm": 1.5383470058441162, "learning_rate": 0.00014286724827452466, "loss": 0.3224, "step": 16470 }, { "epoch": 1.1435297153639112e-13, "grad_norm": 1.7456247806549072, "learning_rate": 0.00014285756391003025, "loss": 0.3372, "step": 16480 }, { "epoch": 1.144223604754302e-13, "grad_norm": 1.7163136005401611, "learning_rate": 0.000142847873342857, "loss": 0.3817, "step": 16490 }, { "epoch": 1.1449174941446927e-13, "grad_norm": 1.5134613513946533, "learning_rate": 0.0001428381765740008, "loss": 0.3184, "step": 16500 }, { "epoch": 1.1456113835350834e-13, "grad_norm": 2.075183629989624, "learning_rate": 0.00014282847360445812, "loss": 0.3243, "step": 16510 }, { "epoch": 1.1463052729254741e-13, "grad_norm": 1.4197566509246826, "learning_rate": 0.00014281876443522613, "loss": 0.3302, "step": 16520 }, { "epoch": 1.1469991623158649e-13, "grad_norm": 1.697121262550354, "learning_rate": 0.00014280904906730258, "loss": 0.3502, "step": 16530 }, { "epoch": 1.1476930517062556e-13, "grad_norm": 1.7865769863128662, "learning_rate": 0.00014279932750168589, "loss": 0.3254, "step": 16540 }, { "epoch": 1.1483869410966463e-13, "grad_norm": 2.1286418437957764, "learning_rate": 0.00014278959973937504, "loss": 0.3659, "step": 16550 }, { "epoch": 1.149080830487037e-13, "grad_norm": 1.791611909866333, "learning_rate": 0.00014277986578136977, "loss": 0.3208, "step": 16560 }, { "epoch": 1.1497747198774277e-13, "grad_norm": 1.8985167741775513, "learning_rate": 0.00014277012562867038, "loss": 0.3607, "step": 16570 }, { "epoch": 1.1504686092678185e-13, "grad_norm": 2.03389310836792, "learning_rate": 0.00014276037928227778, "loss": 0.3709, "step": 16580 }, { "epoch": 1.1511624986582092e-13, "grad_norm": 1.8117878437042236, "learning_rate": 0.00014275062674319364, "loss": 0.3114, "step": 16590 }, { "epoch": 1.1518563880486e-13, "grad_norm": 2.062324285507202, "learning_rate": 0.00014274086801242014, "loss": 0.3658, "step": 16600 }, { "epoch": 1.1525502774389906e-13, "grad_norm": 2.2434024810791016, "learning_rate": 0.0001427311030909601, "loss": 0.3541, "step": 16610 }, { "epoch": 1.1532441668293814e-13, "grad_norm": 1.842240571975708, "learning_rate": 0.00014272133197981712, "loss": 0.3532, "step": 16620 }, { "epoch": 1.153938056219772e-13, "grad_norm": 1.7195637226104736, "learning_rate": 0.00014271155467999528, "loss": 0.3306, "step": 16630 }, { "epoch": 1.1546319456101628e-13, "grad_norm": 2.1292998790740967, "learning_rate": 0.0001427017711924993, "loss": 0.3241, "step": 16640 }, { "epoch": 1.1553258350005535e-13, "grad_norm": 1.7023004293441772, "learning_rate": 0.00014269198151833465, "loss": 0.3691, "step": 16650 }, { "epoch": 1.1560197243909442e-13, "grad_norm": 1.6633024215698242, "learning_rate": 0.00014268218565850738, "loss": 0.3203, "step": 16660 }, { "epoch": 1.156713613781335e-13, "grad_norm": 1.2562707662582397, "learning_rate": 0.0001426723836140241, "loss": 0.3324, "step": 16670 }, { "epoch": 1.1574075031717257e-13, "grad_norm": 1.6491605043411255, "learning_rate": 0.0001426625753858922, "loss": 0.3402, "step": 16680 }, { "epoch": 1.1581013925621164e-13, "grad_norm": 1.8877145051956177, "learning_rate": 0.0001426527609751196, "loss": 0.3282, "step": 16690 }, { "epoch": 1.1587952819525071e-13, "grad_norm": 1.9029465913772583, "learning_rate": 0.00014264294038271482, "loss": 0.3567, "step": 16700 }, { "epoch": 1.1594891713428979e-13, "grad_norm": 1.6828585863113403, "learning_rate": 0.00014263311360968717, "loss": 0.3217, "step": 16710 }, { "epoch": 1.1601830607332886e-13, "grad_norm": 1.8109854459762573, "learning_rate": 0.00014262328065704644, "loss": 0.3349, "step": 16720 }, { "epoch": 1.1608769501236793e-13, "grad_norm": 1.8659015893936157, "learning_rate": 0.00014261344152580318, "loss": 0.3324, "step": 16730 }, { "epoch": 1.16157083951407e-13, "grad_norm": 1.8579113483428955, "learning_rate": 0.00014260359621696845, "loss": 0.3483, "step": 16740 }, { "epoch": 1.1622647289044608e-13, "grad_norm": 1.8433513641357422, "learning_rate": 0.00014259374473155406, "loss": 0.3073, "step": 16750 }, { "epoch": 1.1629586182948515e-13, "grad_norm": 1.4951900243759155, "learning_rate": 0.00014258388707057238, "loss": 0.3513, "step": 16760 }, { "epoch": 1.1636525076852422e-13, "grad_norm": 1.6364794969558716, "learning_rate": 0.00014257402323503642, "loss": 0.386, "step": 16770 }, { "epoch": 1.164346397075633e-13, "grad_norm": 1.9342482089996338, "learning_rate": 0.00014256415322595983, "loss": 0.3196, "step": 16780 }, { "epoch": 1.1650402864660236e-13, "grad_norm": 2.024843215942383, "learning_rate": 0.000142554277044357, "loss": 0.3189, "step": 16790 }, { "epoch": 1.1657341758564144e-13, "grad_norm": 2.0302200317382812, "learning_rate": 0.00014254439469124278, "loss": 0.3654, "step": 16800 }, { "epoch": 1.166428065246805e-13, "grad_norm": 1.6572859287261963, "learning_rate": 0.00014253450616763275, "loss": 0.3332, "step": 16810 }, { "epoch": 1.1671219546371958e-13, "grad_norm": 1.851134181022644, "learning_rate": 0.00014252461147454314, "loss": 0.3063, "step": 16820 }, { "epoch": 1.1678158440275865e-13, "grad_norm": 1.9901585578918457, "learning_rate": 0.00014251471061299073, "loss": 0.3384, "step": 16830 }, { "epoch": 1.1685097334179773e-13, "grad_norm": 1.501957893371582, "learning_rate": 0.00014250480358399302, "loss": 0.3211, "step": 16840 }, { "epoch": 1.169203622808368e-13, "grad_norm": 1.8401042222976685, "learning_rate": 0.00014249489038856814, "loss": 0.3226, "step": 16850 }, { "epoch": 1.1698975121987587e-13, "grad_norm": 2.002894639968872, "learning_rate": 0.00014248497102773474, "loss": 0.3273, "step": 16860 }, { "epoch": 1.1705914015891494e-13, "grad_norm": 1.6913931369781494, "learning_rate": 0.00014247504550251227, "loss": 0.3185, "step": 16870 }, { "epoch": 1.1712852909795402e-13, "grad_norm": 1.7402065992355347, "learning_rate": 0.00014246511381392073, "loss": 0.3431, "step": 16880 }, { "epoch": 1.1719791803699309e-13, "grad_norm": 1.7324925661087036, "learning_rate": 0.0001424551759629807, "loss": 0.3331, "step": 16890 }, { "epoch": 1.1726730697603216e-13, "grad_norm": 1.392574667930603, "learning_rate": 0.00014244523195071346, "loss": 0.3124, "step": 16900 }, { "epoch": 1.1733669591507123e-13, "grad_norm": 1.8279811143875122, "learning_rate": 0.00014243528177814097, "loss": 0.3264, "step": 16910 }, { "epoch": 1.174060848541103e-13, "grad_norm": 1.7530206441879272, "learning_rate": 0.0001424253254462857, "loss": 0.3432, "step": 16920 }, { "epoch": 1.1747547379314938e-13, "grad_norm": 1.4518858194351196, "learning_rate": 0.00014241536295617085, "loss": 0.3677, "step": 16930 }, { "epoch": 1.1754486273218845e-13, "grad_norm": 1.3310619592666626, "learning_rate": 0.00014240539430882023, "loss": 0.2914, "step": 16940 }, { "epoch": 1.1761425167122752e-13, "grad_norm": 1.8807568550109863, "learning_rate": 0.00014239541950525824, "loss": 0.3298, "step": 16950 }, { "epoch": 1.176836406102666e-13, "grad_norm": 1.4816628694534302, "learning_rate": 0.00014238543854650998, "loss": 0.3641, "step": 16960 }, { "epoch": 1.1775302954930567e-13, "grad_norm": 1.4535504579544067, "learning_rate": 0.00014237545143360112, "loss": 0.3074, "step": 16970 }, { "epoch": 1.1782241848834474e-13, "grad_norm": 1.8568367958068848, "learning_rate": 0.000142365458167558, "loss": 0.3325, "step": 16980 }, { "epoch": 1.178918074273838e-13, "grad_norm": 1.7328332662582397, "learning_rate": 0.0001423554587494076, "loss": 0.3252, "step": 16990 }, { "epoch": 1.1796119636642288e-13, "grad_norm": 1.6979777812957764, "learning_rate": 0.0001423454531801775, "loss": 0.3187, "step": 17000 }, { "epoch": 1.1803058530546195e-13, "grad_norm": 2.169914484024048, "learning_rate": 0.00014233544146089591, "loss": 0.327, "step": 17010 }, { "epoch": 1.1809997424450103e-13, "grad_norm": 1.8471131324768066, "learning_rate": 0.00014232542359259172, "loss": 0.3367, "step": 17020 }, { "epoch": 1.181693631835401e-13, "grad_norm": 2.052914619445801, "learning_rate": 0.00014231539957629445, "loss": 0.3038, "step": 17030 }, { "epoch": 1.1823875212257917e-13, "grad_norm": 1.7594066858291626, "learning_rate": 0.00014230536941303417, "loss": 0.3297, "step": 17040 }, { "epoch": 1.1830814106161824e-13, "grad_norm": 1.7039954662322998, "learning_rate": 0.00014229533310384162, "loss": 0.3411, "step": 17050 }, { "epoch": 1.1837753000065732e-13, "grad_norm": 1.6464667320251465, "learning_rate": 0.00014228529064974824, "loss": 0.292, "step": 17060 }, { "epoch": 1.184469189396964e-13, "grad_norm": 1.8389662504196167, "learning_rate": 0.00014227524205178603, "loss": 0.3516, "step": 17070 }, { "epoch": 1.1851630787873546e-13, "grad_norm": 2.0815773010253906, "learning_rate": 0.00014226518731098766, "loss": 0.3529, "step": 17080 }, { "epoch": 1.1858569681777453e-13, "grad_norm": 1.7919608354568481, "learning_rate": 0.00014225512642838635, "loss": 0.3136, "step": 17090 }, { "epoch": 1.186550857568136e-13, "grad_norm": 1.6683062314987183, "learning_rate": 0.00014224505940501605, "loss": 0.2951, "step": 17100 }, { "epoch": 1.1872447469585268e-13, "grad_norm": 2.1705808639526367, "learning_rate": 0.00014223498624191133, "loss": 0.3193, "step": 17110 }, { "epoch": 1.1879386363489175e-13, "grad_norm": 1.9096113443374634, "learning_rate": 0.00014222490694010734, "loss": 0.334, "step": 17120 }, { "epoch": 1.1886325257393082e-13, "grad_norm": 1.9997411966323853, "learning_rate": 0.00014221482150063986, "loss": 0.305, "step": 17130 }, { "epoch": 1.189326415129699e-13, "grad_norm": 1.4364265203475952, "learning_rate": 0.00014220472992454536, "loss": 0.311, "step": 17140 }, { "epoch": 1.1900203045200897e-13, "grad_norm": 2.078157663345337, "learning_rate": 0.0001421946322128609, "loss": 0.3559, "step": 17150 }, { "epoch": 1.1907141939104804e-13, "grad_norm": 2.0192630290985107, "learning_rate": 0.00014218452836662417, "loss": 0.299, "step": 17160 }, { "epoch": 1.191408083300871e-13, "grad_norm": 1.3377103805541992, "learning_rate": 0.0001421744183868735, "loss": 0.3224, "step": 17170 }, { "epoch": 1.1921019726912618e-13, "grad_norm": 1.7080835103988647, "learning_rate": 0.00014216430227464785, "loss": 0.3181, "step": 17180 }, { "epoch": 1.1927958620816526e-13, "grad_norm": 1.7487682104110718, "learning_rate": 0.0001421541800309868, "loss": 0.3275, "step": 17190 }, { "epoch": 1.1934897514720433e-13, "grad_norm": 1.5349706411361694, "learning_rate": 0.00014214405165693055, "loss": 0.3117, "step": 17200 }, { "epoch": 1.194183640862434e-13, "grad_norm": 1.6735001802444458, "learning_rate": 0.00014213391715352003, "loss": 0.3346, "step": 17210 }, { "epoch": 1.1948775302528247e-13, "grad_norm": 1.8839633464813232, "learning_rate": 0.00014212377652179662, "loss": 0.3005, "step": 17220 }, { "epoch": 1.1955714196432154e-13, "grad_norm": 1.8114471435546875, "learning_rate": 0.00014211362976280248, "loss": 0.3207, "step": 17230 }, { "epoch": 1.1962653090336062e-13, "grad_norm": 1.6681708097457886, "learning_rate": 0.0001421034768775803, "loss": 0.3631, "step": 17240 }, { "epoch": 1.196959198423997e-13, "grad_norm": 1.8091624975204468, "learning_rate": 0.00014209331786717354, "loss": 0.298, "step": 17250 }, { "epoch": 1.1976530878143876e-13, "grad_norm": 1.8713428974151611, "learning_rate": 0.0001420831527326261, "loss": 0.3038, "step": 17260 }, { "epoch": 1.1983469772047783e-13, "grad_norm": 2.0564699172973633, "learning_rate": 0.00014207298147498264, "loss": 0.3378, "step": 17270 }, { "epoch": 1.199040866595169e-13, "grad_norm": 1.8830658197402954, "learning_rate": 0.00014206280409528844, "loss": 0.3144, "step": 17280 }, { "epoch": 1.1997347559855598e-13, "grad_norm": 1.8437894582748413, "learning_rate": 0.00014205262059458935, "loss": 0.286, "step": 17290 }, { "epoch": 1.2004286453759505e-13, "grad_norm": 1.8809828758239746, "learning_rate": 0.00014204243097393188, "loss": 0.3256, "step": 17300 }, { "epoch": 1.2011225347663412e-13, "grad_norm": 2.1843676567077637, "learning_rate": 0.00014203223523436324, "loss": 0.3231, "step": 17310 }, { "epoch": 1.201816424156732e-13, "grad_norm": 1.7373796701431274, "learning_rate": 0.0001420220333769311, "loss": 0.2981, "step": 17320 }, { "epoch": 1.2025103135471227e-13, "grad_norm": 1.61070716381073, "learning_rate": 0.00014201182540268394, "loss": 0.3102, "step": 17330 }, { "epoch": 1.2032042029375134e-13, "grad_norm": 1.799973487854004, "learning_rate": 0.00014200161131267076, "loss": 0.3103, "step": 17340 }, { "epoch": 1.203898092327904e-13, "grad_norm": 1.6861954927444458, "learning_rate": 0.00014199139110794115, "loss": 0.3278, "step": 17350 }, { "epoch": 1.2045919817182948e-13, "grad_norm": 1.669665813446045, "learning_rate": 0.00014198116478954552, "loss": 0.3186, "step": 17360 }, { "epoch": 1.2052858711086856e-13, "grad_norm": 1.8179724216461182, "learning_rate": 0.0001419709323585347, "loss": 0.2845, "step": 17370 }, { "epoch": 1.2059797604990763e-13, "grad_norm": 1.9571044445037842, "learning_rate": 0.00014196069381596028, "loss": 0.3312, "step": 17380 }, { "epoch": 1.206673649889467e-13, "grad_norm": 1.9201067686080933, "learning_rate": 0.00014195044916287436, "loss": 0.327, "step": 17390 }, { "epoch": 1.2073675392798577e-13, "grad_norm": 1.7665215730667114, "learning_rate": 0.00014194019840032982, "loss": 0.3474, "step": 17400 }, { "epoch": 1.2080614286702485e-13, "grad_norm": 2.2397847175598145, "learning_rate": 0.00014192994152938002, "loss": 0.2952, "step": 17410 }, { "epoch": 1.2087553180606392e-13, "grad_norm": 1.460151195526123, "learning_rate": 0.00014191967855107902, "loss": 0.2854, "step": 17420 }, { "epoch": 1.20944920745103e-13, "grad_norm": 1.3488205671310425, "learning_rate": 0.00014190940946648154, "loss": 0.3337, "step": 17430 }, { "epoch": 1.2101430968414206e-13, "grad_norm": 1.548288345336914, "learning_rate": 0.00014189913427664288, "loss": 0.2953, "step": 17440 }, { "epoch": 1.2108369862318114e-13, "grad_norm": 2.0084245204925537, "learning_rate": 0.00014188885298261895, "loss": 0.3176, "step": 17450 }, { "epoch": 1.211530875622202e-13, "grad_norm": 2.07537579536438, "learning_rate": 0.00014187856558546636, "loss": 0.3164, "step": 17460 }, { "epoch": 1.2122247650125928e-13, "grad_norm": 1.9616018533706665, "learning_rate": 0.00014186827208624223, "loss": 0.3015, "step": 17470 }, { "epoch": 1.2129186544029835e-13, "grad_norm": 1.8670326471328735, "learning_rate": 0.0001418579724860044, "loss": 0.3083, "step": 17480 }, { "epoch": 1.2136125437933742e-13, "grad_norm": 1.5639066696166992, "learning_rate": 0.00014184766678581137, "loss": 0.2953, "step": 17490 }, { "epoch": 1.214306433183765e-13, "grad_norm": 1.9097561836242676, "learning_rate": 0.00014183735498672216, "loss": 0.2963, "step": 17500 }, { "epoch": 1.2150003225741557e-13, "grad_norm": 1.8916879892349243, "learning_rate": 0.00014182703708979645, "loss": 0.3101, "step": 17510 }, { "epoch": 1.2156942119645464e-13, "grad_norm": 1.4945497512817383, "learning_rate": 0.0001418167130960946, "loss": 0.3067, "step": 17520 }, { "epoch": 1.2163881013549371e-13, "grad_norm": 1.7590841054916382, "learning_rate": 0.00014180638300667755, "loss": 0.2777, "step": 17530 }, { "epoch": 1.2170819907453279e-13, "grad_norm": 1.5739964246749878, "learning_rate": 0.0001417960468226069, "loss": 0.3409, "step": 17540 }, { "epoch": 1.2177758801357186e-13, "grad_norm": 1.7751493453979492, "learning_rate": 0.00014178570454494483, "loss": 0.354, "step": 17550 }, { "epoch": 1.2184697695261093e-13, "grad_norm": 1.5252972841262817, "learning_rate": 0.00014177535617475415, "loss": 0.2945, "step": 17560 }, { "epoch": 1.2191636589165e-13, "grad_norm": 1.6444323062896729, "learning_rate": 0.00014176500171309838, "loss": 0.283, "step": 17570 }, { "epoch": 1.2198575483068907e-13, "grad_norm": 2.2148704528808594, "learning_rate": 0.0001417546411610415, "loss": 0.3169, "step": 17580 }, { "epoch": 1.2205514376972815e-13, "grad_norm": 1.7272590398788452, "learning_rate": 0.00014174427451964832, "loss": 0.2966, "step": 17590 }, { "epoch": 1.2212453270876722e-13, "grad_norm": 1.5358726978302002, "learning_rate": 0.00014173390178998414, "loss": 0.2743, "step": 17600 }, { "epoch": 1.221939216478063e-13, "grad_norm": 2.0239858627319336, "learning_rate": 0.00014172352297311489, "loss": 0.2991, "step": 17610 }, { "epoch": 1.2226331058684536e-13, "grad_norm": 1.4490796327590942, "learning_rate": 0.00014171313807010718, "loss": 0.3302, "step": 17620 }, { "epoch": 1.2233269952588444e-13, "grad_norm": 1.5162767171859741, "learning_rate": 0.00014170274708202822, "loss": 0.2742, "step": 17630 }, { "epoch": 1.224020884649235e-13, "grad_norm": 1.379867434501648, "learning_rate": 0.00014169235000994587, "loss": 0.2773, "step": 17640 }, { "epoch": 1.2247147740396258e-13, "grad_norm": 1.338138461112976, "learning_rate": 0.00014168194685492853, "loss": 0.2885, "step": 17650 }, { "epoch": 1.2254086634300165e-13, "grad_norm": 1.869503378868103, "learning_rate": 0.00014167153761804533, "loss": 0.3144, "step": 17660 }, { "epoch": 1.2261025528204073e-13, "grad_norm": 1.933161973953247, "learning_rate": 0.00014166112230036597, "loss": 0.3097, "step": 17670 }, { "epoch": 1.226796442210798e-13, "grad_norm": 1.3558541536331177, "learning_rate": 0.00014165070090296078, "loss": 0.2843, "step": 17680 }, { "epoch": 1.2274903316011887e-13, "grad_norm": 1.5507371425628662, "learning_rate": 0.00014164027342690073, "loss": 0.2902, "step": 17690 }, { "epoch": 1.2281842209915794e-13, "grad_norm": 1.6877191066741943, "learning_rate": 0.00014162983987325742, "loss": 0.3191, "step": 17700 }, { "epoch": 1.2288781103819701e-13, "grad_norm": 2.028691291809082, "learning_rate": 0.000141619400243103, "loss": 0.3555, "step": 17710 }, { "epoch": 1.2295719997723609e-13, "grad_norm": 1.514865517616272, "learning_rate": 0.0001416089545375104, "loss": 0.287, "step": 17720 }, { "epoch": 1.2302658891627516e-13, "grad_norm": 1.5669037103652954, "learning_rate": 0.000141598502757553, "loss": 0.2843, "step": 17730 }, { "epoch": 1.2309597785531423e-13, "grad_norm": 1.8851616382598877, "learning_rate": 0.00014158804490430494, "loss": 0.3295, "step": 17740 }, { "epoch": 1.231653667943533e-13, "grad_norm": 1.4397262334823608, "learning_rate": 0.00014157758097884087, "loss": 0.2781, "step": 17750 }, { "epoch": 1.2323475573339238e-13, "grad_norm": 1.8969043493270874, "learning_rate": 0.00014156711098223615, "loss": 0.2723, "step": 17760 }, { "epoch": 1.2330414467243145e-13, "grad_norm": 1.6388556957244873, "learning_rate": 0.00014155663491556672, "loss": 0.2914, "step": 17770 }, { "epoch": 1.2337353361147052e-13, "grad_norm": 1.4391374588012695, "learning_rate": 0.0001415461527799092, "loss": 0.3021, "step": 17780 }, { "epoch": 1.234429225505096e-13, "grad_norm": 1.4459048509597778, "learning_rate": 0.00014153566457634074, "loss": 0.268, "step": 17790 }, { "epoch": 1.2351231148954867e-13, "grad_norm": 1.9468704462051392, "learning_rate": 0.0001415251703059392, "loss": 0.2912, "step": 17800 }, { "epoch": 1.2358170042858774e-13, "grad_norm": 1.55011785030365, "learning_rate": 0.00014151466996978302, "loss": 0.2595, "step": 17810 }, { "epoch": 1.236510893676268e-13, "grad_norm": 1.869748592376709, "learning_rate": 0.00014150416356895128, "loss": 0.3053, "step": 17820 }, { "epoch": 1.2372047830666588e-13, "grad_norm": 1.8185421228408813, "learning_rate": 0.00014149365110452367, "loss": 0.3131, "step": 17830 }, { "epoch": 1.2378986724570495e-13, "grad_norm": 1.830710768699646, "learning_rate": 0.0001414831325775805, "loss": 0.2594, "step": 17840 }, { "epoch": 1.2385925618474403e-13, "grad_norm": 2.071624517440796, "learning_rate": 0.0001414726079892027, "loss": 0.3164, "step": 17850 }, { "epoch": 1.239286451237831e-13, "grad_norm": 2.3410913944244385, "learning_rate": 0.00014146207734047189, "loss": 0.3066, "step": 17860 }, { "epoch": 1.2399803406282217e-13, "grad_norm": 1.9778952598571777, "learning_rate": 0.0001414515406324702, "loss": 0.3118, "step": 17870 }, { "epoch": 1.2406742300186124e-13, "grad_norm": 1.6809756755828857, "learning_rate": 0.00014144099786628046, "loss": 0.2769, "step": 17880 }, { "epoch": 1.2413681194090032e-13, "grad_norm": 1.749700903892517, "learning_rate": 0.00014143044904298612, "loss": 0.2641, "step": 17890 }, { "epoch": 1.242062008799394e-13, "grad_norm": 1.9683247804641724, "learning_rate": 0.00014141989416367124, "loss": 0.3239, "step": 17900 }, { "epoch": 1.2427558981897846e-13, "grad_norm": 1.6938629150390625, "learning_rate": 0.00014140933322942046, "loss": 0.2754, "step": 17910 }, { "epoch": 1.2434497875801753e-13, "grad_norm": 1.2430545091629028, "learning_rate": 0.0001413987662413191, "loss": 0.284, "step": 17920 }, { "epoch": 1.244143676970566e-13, "grad_norm": 1.679658055305481, "learning_rate": 0.0001413881932004531, "loss": 0.3009, "step": 17930 }, { "epoch": 1.2448375663609568e-13, "grad_norm": 1.6466819047927856, "learning_rate": 0.000141377614107909, "loss": 0.2725, "step": 17940 }, { "epoch": 1.2455314557513475e-13, "grad_norm": 1.7341560125350952, "learning_rate": 0.0001413670289647739, "loss": 0.2835, "step": 17950 }, { "epoch": 1.2462253451417382e-13, "grad_norm": 1.3702216148376465, "learning_rate": 0.0001413564377721357, "loss": 0.2522, "step": 17960 }, { "epoch": 1.246919234532129e-13, "grad_norm": 1.752394199371338, "learning_rate": 0.00014134584053108275, "loss": 0.2718, "step": 17970 }, { "epoch": 1.2476131239225197e-13, "grad_norm": 1.933655023574829, "learning_rate": 0.00014133523724270405, "loss": 0.2885, "step": 17980 }, { "epoch": 1.2483070133129104e-13, "grad_norm": 1.649552583694458, "learning_rate": 0.00014132462790808928, "loss": 0.2909, "step": 17990 }, { "epoch": 1.249000902703301e-13, "grad_norm": 1.8530762195587158, "learning_rate": 0.00014131401252832877, "loss": 0.2605, "step": 18000 }, { "epoch": 1.2496947920936918e-13, "grad_norm": 1.4199707508087158, "learning_rate": 0.00014130339110451337, "loss": 0.2904, "step": 18010 }, { "epoch": 1.2503886814840826e-13, "grad_norm": 1.6727416515350342, "learning_rate": 0.00014129276363773454, "loss": 0.3272, "step": 18020 }, { "epoch": 1.2510825708744733e-13, "grad_norm": 1.7788805961608887, "learning_rate": 0.0001412821301290845, "loss": 0.2755, "step": 18030 }, { "epoch": 1.251776460264864e-13, "grad_norm": 1.8474547863006592, "learning_rate": 0.00014127149057965597, "loss": 0.2713, "step": 18040 }, { "epoch": 1.2524703496552547e-13, "grad_norm": 1.8340920209884644, "learning_rate": 0.00014126084499054237, "loss": 0.2995, "step": 18050 }, { "epoch": 1.2531642390456454e-13, "grad_norm": 1.8334825038909912, "learning_rate": 0.00014125019336283761, "loss": 0.2904, "step": 18060 }, { "epoch": 1.2538581284360362e-13, "grad_norm": 1.6643165349960327, "learning_rate": 0.0001412395356976364, "loss": 0.2537, "step": 18070 }, { "epoch": 1.254552017826427e-13, "grad_norm": 1.9119514226913452, "learning_rate": 0.00014122887199603397, "loss": 0.29, "step": 18080 }, { "epoch": 1.2552459072168176e-13, "grad_norm": 1.3604857921600342, "learning_rate": 0.00014121820225912614, "loss": 0.2839, "step": 18090 }, { "epoch": 1.2559397966072083e-13, "grad_norm": 1.6737403869628906, "learning_rate": 0.00014120752648800943, "loss": 0.2631, "step": 18100 }, { "epoch": 1.256633685997599e-13, "grad_norm": 1.7539950609207153, "learning_rate": 0.0001411968446837809, "loss": 0.2629, "step": 18110 }, { "epoch": 1.2573275753879898e-13, "grad_norm": 1.4338805675506592, "learning_rate": 0.00014118615684753833, "loss": 0.2477, "step": 18120 }, { "epoch": 1.2580214647783805e-13, "grad_norm": 1.6255359649658203, "learning_rate": 0.00014117546298038, "loss": 0.2776, "step": 18130 }, { "epoch": 1.2587153541687712e-13, "grad_norm": 1.7147454023361206, "learning_rate": 0.00014116476308340491, "loss": 0.294, "step": 18140 }, { "epoch": 1.259409243559162e-13, "grad_norm": 1.4060701131820679, "learning_rate": 0.00014115405715771262, "loss": 0.2588, "step": 18150 }, { "epoch": 1.2601031329495527e-13, "grad_norm": 1.7932746410369873, "learning_rate": 0.00014114334520440335, "loss": 0.2785, "step": 18160 }, { "epoch": 1.2607970223399434e-13, "grad_norm": 1.5973848104476929, "learning_rate": 0.00014113262722457792, "loss": 0.2818, "step": 18170 }, { "epoch": 1.261490911730334e-13, "grad_norm": 1.442251205444336, "learning_rate": 0.00014112190321933772, "loss": 0.3217, "step": 18180 }, { "epoch": 1.2621848011207248e-13, "grad_norm": 1.1956080198287964, "learning_rate": 0.00014111117318978486, "loss": 0.2525, "step": 18190 }, { "epoch": 1.2628786905111156e-13, "grad_norm": 1.750165581703186, "learning_rate": 0.00014110043713702204, "loss": 0.2816, "step": 18200 }, { "epoch": 1.2635725799015063e-13, "grad_norm": 1.3822035789489746, "learning_rate": 0.00014108969506215248, "loss": 0.2934, "step": 18210 }, { "epoch": 1.264266469291897e-13, "grad_norm": 1.3064594268798828, "learning_rate": 0.00014107894696628018, "loss": 0.2617, "step": 18220 }, { "epoch": 1.2649603586822877e-13, "grad_norm": 1.6207410097122192, "learning_rate": 0.0001410681928505096, "loss": 0.2683, "step": 18230 }, { "epoch": 1.2656542480726785e-13, "grad_norm": 1.7517060041427612, "learning_rate": 0.00014105743271594594, "loss": 0.2714, "step": 18240 }, { "epoch": 1.2663481374630692e-13, "grad_norm": 1.402374505996704, "learning_rate": 0.00014104666656369492, "loss": 0.2664, "step": 18250 }, { "epoch": 1.26704202685346e-13, "grad_norm": 1.9894126653671265, "learning_rate": 0.000141035894394863, "loss": 0.2598, "step": 18260 }, { "epoch": 1.2677359162438506e-13, "grad_norm": 1.5224121809005737, "learning_rate": 0.00014102511621055713, "loss": 0.2701, "step": 18270 }, { "epoch": 1.2684298056342413e-13, "grad_norm": 1.6330522298812866, "learning_rate": 0.000141014332011885, "loss": 0.2314, "step": 18280 }, { "epoch": 1.269123695024632e-13, "grad_norm": 1.7087167501449585, "learning_rate": 0.00014100354179995476, "loss": 0.2691, "step": 18290 }, { "epoch": 1.2698175844150228e-13, "grad_norm": 1.5683058500289917, "learning_rate": 0.00014099274557587534, "loss": 0.2965, "step": 18300 }, { "epoch": 1.2705114738054135e-13, "grad_norm": 1.6796787977218628, "learning_rate": 0.00014098194334075624, "loss": 0.2425, "step": 18310 }, { "epoch": 1.2712053631958042e-13, "grad_norm": 1.626630425453186, "learning_rate": 0.00014097113509570748, "loss": 0.3009, "step": 18320 }, { "epoch": 1.271899252586195e-13, "grad_norm": 2.1198348999023438, "learning_rate": 0.00014096032084183984, "loss": 0.3005, "step": 18330 }, { "epoch": 1.2725931419765857e-13, "grad_norm": 1.7857742309570312, "learning_rate": 0.00014094950058026463, "loss": 0.2785, "step": 18340 }, { "epoch": 1.2732870313669764e-13, "grad_norm": 1.6163996458053589, "learning_rate": 0.00014093867431209378, "loss": 0.2476, "step": 18350 }, { "epoch": 1.2739809207573671e-13, "grad_norm": 2.268178939819336, "learning_rate": 0.0001409278420384399, "loss": 0.2634, "step": 18360 }, { "epoch": 1.2746748101477579e-13, "grad_norm": 1.8560377359390259, "learning_rate": 0.00014091700376041614, "loss": 0.2735, "step": 18370 }, { "epoch": 1.2753686995381486e-13, "grad_norm": 1.7499204874038696, "learning_rate": 0.00014090615947913635, "loss": 0.2538, "step": 18380 }, { "epoch": 1.2760625889285393e-13, "grad_norm": 1.5468909740447998, "learning_rate": 0.00014089530919571486, "loss": 0.2525, "step": 18390 }, { "epoch": 1.27675647831893e-13, "grad_norm": 2.185856342315674, "learning_rate": 0.0001408844529112668, "loss": 0.2957, "step": 18400 }, { "epoch": 1.2774503677093207e-13, "grad_norm": 1.7351452112197876, "learning_rate": 0.00014087359062690775, "loss": 0.2398, "step": 18410 }, { "epoch": 1.2781442570997115e-13, "grad_norm": 1.330352544784546, "learning_rate": 0.00014086272234375405, "loss": 0.257, "step": 18420 }, { "epoch": 1.2788381464901022e-13, "grad_norm": 1.3901991844177246, "learning_rate": 0.0001408518480629225, "loss": 0.2471, "step": 18430 }, { "epoch": 1.279532035880493e-13, "grad_norm": 1.37436044216156, "learning_rate": 0.00014084096778553068, "loss": 0.2492, "step": 18440 }, { "epoch": 1.2802259252708836e-13, "grad_norm": 1.5544151067733765, "learning_rate": 0.0001408300815126967, "loss": 0.2637, "step": 18450 }, { "epoch": 1.2809198146612744e-13, "grad_norm": 1.523976445198059, "learning_rate": 0.00014081918924553922, "loss": 0.2829, "step": 18460 }, { "epoch": 1.281613704051665e-13, "grad_norm": 1.915492296218872, "learning_rate": 0.00014080829098517763, "loss": 0.2488, "step": 18470 }, { "epoch": 1.2823075934420558e-13, "grad_norm": 1.763445496559143, "learning_rate": 0.00014079738673273195, "loss": 0.2645, "step": 18480 }, { "epoch": 1.2830014828324465e-13, "grad_norm": 1.64935302734375, "learning_rate": 0.00014078647648932267, "loss": 0.3098, "step": 18490 }, { "epoch": 1.2836953722228372e-13, "grad_norm": 1.831176519393921, "learning_rate": 0.00014077556025607108, "loss": 0.2644, "step": 18500 }, { "epoch": 1.284389261613228e-13, "grad_norm": 1.7333911657333374, "learning_rate": 0.0001407646380340989, "loss": 0.2507, "step": 18510 }, { "epoch": 1.2850831510036187e-13, "grad_norm": 1.778226613998413, "learning_rate": 0.00014075370982452864, "loss": 0.2686, "step": 18520 }, { "epoch": 1.2857770403940094e-13, "grad_norm": 1.805923342704773, "learning_rate": 0.00014074277562848328, "loss": 0.2632, "step": 18530 }, { "epoch": 1.2864709297844001e-13, "grad_norm": 1.8940526247024536, "learning_rate": 0.00014073183544708652, "loss": 0.2278, "step": 18540 }, { "epoch": 1.2871648191747909e-13, "grad_norm": 1.7407399415969849, "learning_rate": 0.00014072088928146262, "loss": 0.2652, "step": 18550 }, { "epoch": 1.2878587085651816e-13, "grad_norm": 2.0254721641540527, "learning_rate": 0.00014070993713273647, "loss": 0.2631, "step": 18560 }, { "epoch": 1.2885525979555723e-13, "grad_norm": 1.6353949308395386, "learning_rate": 0.00014069897900203358, "loss": 0.2419, "step": 18570 }, { "epoch": 1.289246487345963e-13, "grad_norm": 1.5648308992385864, "learning_rate": 0.00014068801489048005, "loss": 0.2522, "step": 18580 }, { "epoch": 1.2899403767363538e-13, "grad_norm": 1.5751519203186035, "learning_rate": 0.00014067704479920262, "loss": 0.2392, "step": 18590 }, { "epoch": 1.2906342661267445e-13, "grad_norm": 1.5240508317947388, "learning_rate": 0.00014066606872932866, "loss": 0.254, "step": 18600 }, { "epoch": 1.2913281555171352e-13, "grad_norm": 1.7418696880340576, "learning_rate": 0.00014065508668198608, "loss": 0.2832, "step": 18610 }, { "epoch": 1.292022044907526e-13, "grad_norm": 1.6128560304641724, "learning_rate": 0.00014064409865830355, "loss": 0.2399, "step": 18620 }, { "epoch": 1.2927159342979166e-13, "grad_norm": 1.8235944509506226, "learning_rate": 0.00014063310465941013, "loss": 0.2687, "step": 18630 }, { "epoch": 1.2934098236883074e-13, "grad_norm": 1.773114800453186, "learning_rate": 0.00014062210468643573, "loss": 0.2587, "step": 18640 }, { "epoch": 1.294103713078698e-13, "grad_norm": 1.6759711503982544, "learning_rate": 0.00014061109874051074, "loss": 0.2946, "step": 18650 }, { "epoch": 1.2947976024690888e-13, "grad_norm": 2.252171039581299, "learning_rate": 0.00014060008682276618, "loss": 0.2594, "step": 18660 }, { "epoch": 1.2954914918594795e-13, "grad_norm": 1.2509504556655884, "learning_rate": 0.00014058906893433368, "loss": 0.2359, "step": 18670 }, { "epoch": 1.2961853812498703e-13, "grad_norm": 1.2707663774490356, "learning_rate": 0.00014057804507634555, "loss": 0.2714, "step": 18680 }, { "epoch": 1.296879270640261e-13, "grad_norm": 1.5369131565093994, "learning_rate": 0.00014056701524993464, "loss": 0.2518, "step": 18690 }, { "epoch": 1.2975731600306517e-13, "grad_norm": 1.7494031190872192, "learning_rate": 0.0001405559794562344, "loss": 0.2492, "step": 18700 }, { "epoch": 1.2982670494210424e-13, "grad_norm": 1.649871587753296, "learning_rate": 0.000140544937696379, "loss": 0.2481, "step": 18710 }, { "epoch": 1.2989609388114332e-13, "grad_norm": 1.766791820526123, "learning_rate": 0.00014053388997150307, "loss": 0.2442, "step": 18720 }, { "epoch": 1.299654828201824e-13, "grad_norm": 1.8887056112289429, "learning_rate": 0.000140522836282742, "loss": 0.2542, "step": 18730 }, { "epoch": 1.3003487175922146e-13, "grad_norm": 1.4040321111679077, "learning_rate": 0.0001405117766312317, "loss": 0.2424, "step": 18740 }, { "epoch": 1.3010426069826053e-13, "grad_norm": 1.7502309083938599, "learning_rate": 0.00014050071101810873, "loss": 0.2302, "step": 18750 }, { "epoch": 1.301736496372996e-13, "grad_norm": 1.7455283403396606, "learning_rate": 0.00014048963944451024, "loss": 0.247, "step": 18760 }, { "epoch": 1.3024303857633868e-13, "grad_norm": 1.5527241230010986, "learning_rate": 0.00014047856191157403, "loss": 0.2747, "step": 18770 }, { "epoch": 1.3031242751537775e-13, "grad_norm": 1.614526391029358, "learning_rate": 0.0001404674784204385, "loss": 0.2255, "step": 18780 }, { "epoch": 1.3038181645441682e-13, "grad_norm": 1.3848803043365479, "learning_rate": 0.00014045638897224258, "loss": 0.2698, "step": 18790 }, { "epoch": 1.304512053934559e-13, "grad_norm": 1.601289987564087, "learning_rate": 0.00014044529356812595, "loss": 0.2741, "step": 18800 }, { "epoch": 1.3052059433249497e-13, "grad_norm": 1.5209646224975586, "learning_rate": 0.00014043419220922883, "loss": 0.252, "step": 18810 }, { "epoch": 1.3058998327153404e-13, "grad_norm": 1.596069574356079, "learning_rate": 0.00014042308489669203, "loss": 0.241, "step": 18820 }, { "epoch": 1.306593722105731e-13, "grad_norm": 1.9741519689559937, "learning_rate": 0.000140411971631657, "loss": 0.2678, "step": 18830 }, { "epoch": 1.3072876114961218e-13, "grad_norm": 1.7162832021713257, "learning_rate": 0.00014040085241526584, "loss": 0.2481, "step": 18840 }, { "epoch": 1.3079815008865125e-13, "grad_norm": 1.4070416688919067, "learning_rate": 0.0001403897272486612, "loss": 0.2232, "step": 18850 }, { "epoch": 1.3086753902769033e-13, "grad_norm": 1.6381824016571045, "learning_rate": 0.00014037859613298635, "loss": 0.2317, "step": 18860 }, { "epoch": 1.309369279667294e-13, "grad_norm": 1.2134543657302856, "learning_rate": 0.00014036745906938517, "loss": 0.258, "step": 18870 }, { "epoch": 1.3100631690576847e-13, "grad_norm": 1.5115605592727661, "learning_rate": 0.00014035631605900225, "loss": 0.2284, "step": 18880 }, { "epoch": 1.3107570584480754e-13, "grad_norm": 1.4227933883666992, "learning_rate": 0.0001403451671029826, "loss": 0.2403, "step": 18890 }, { "epoch": 1.3114509478384662e-13, "grad_norm": 1.2367302179336548, "learning_rate": 0.00014033401220247204, "loss": 0.2439, "step": 18900 }, { "epoch": 1.312144837228857e-13, "grad_norm": 1.6065062284469604, "learning_rate": 0.0001403228513586168, "loss": 0.2477, "step": 18910 }, { "epoch": 1.3128387266192476e-13, "grad_norm": 1.9474626779556274, "learning_rate": 0.00014031168457256398, "loss": 0.2551, "step": 18920 }, { "epoch": 1.3135326160096383e-13, "grad_norm": 1.2930855751037598, "learning_rate": 0.00014030051184546103, "loss": 0.2461, "step": 18930 }, { "epoch": 1.314226505400029e-13, "grad_norm": 1.5134634971618652, "learning_rate": 0.00014028933317845614, "loss": 0.2404, "step": 18940 }, { "epoch": 1.3149203947904198e-13, "grad_norm": 1.4245493412017822, "learning_rate": 0.00014027814857269812, "loss": 0.2545, "step": 18950 }, { "epoch": 1.3156142841808105e-13, "grad_norm": 1.8311511278152466, "learning_rate": 0.00014026695802933634, "loss": 0.2857, "step": 18960 }, { "epoch": 1.3163081735712012e-13, "grad_norm": 1.472915530204773, "learning_rate": 0.0001402557615495208, "loss": 0.2432, "step": 18970 }, { "epoch": 1.317002062961592e-13, "grad_norm": 1.4344178438186646, "learning_rate": 0.00014024455913440212, "loss": 0.2364, "step": 18980 }, { "epoch": 1.3176959523519827e-13, "grad_norm": 1.7758495807647705, "learning_rate": 0.00014023335078513152, "loss": 0.2666, "step": 18990 }, { "epoch": 1.3183898417423734e-13, "grad_norm": 1.328005075454712, "learning_rate": 0.00014022213650286087, "loss": 0.2287, "step": 19000 }, { "epoch": 1.319083731132764e-13, "grad_norm": 1.7745246887207031, "learning_rate": 0.00014021091628874254, "loss": 0.2144, "step": 19010 }, { "epoch": 1.3197776205231548e-13, "grad_norm": 1.3891799449920654, "learning_rate": 0.00014019969014392964, "loss": 0.2296, "step": 19020 }, { "epoch": 1.3204715099135456e-13, "grad_norm": 1.3930858373641968, "learning_rate": 0.00014018845806957578, "loss": 0.2367, "step": 19030 }, { "epoch": 1.3211653993039363e-13, "grad_norm": 1.4854233264923096, "learning_rate": 0.0001401772200668353, "loss": 0.2201, "step": 19040 }, { "epoch": 1.321859288694327e-13, "grad_norm": 2.2674407958984375, "learning_rate": 0.000140165976136863, "loss": 0.2606, "step": 19050 }, { "epoch": 1.3225531780847177e-13, "grad_norm": 1.4341148138046265, "learning_rate": 0.00014015472628081443, "loss": 0.2129, "step": 19060 }, { "epoch": 1.3232470674751085e-13, "grad_norm": 1.5682178735733032, "learning_rate": 0.00014014347049984567, "loss": 0.2433, "step": 19070 }, { "epoch": 1.3239409568654992e-13, "grad_norm": 1.8431189060211182, "learning_rate": 0.00014013220879511342, "loss": 0.2701, "step": 19080 }, { "epoch": 1.32463484625589e-13, "grad_norm": 1.7328846454620361, "learning_rate": 0.00014012094116777502, "loss": 0.2182, "step": 19090 }, { "epoch": 1.3253287356462806e-13, "grad_norm": 2.172269105911255, "learning_rate": 0.00014010966761898833, "loss": 0.2701, "step": 19100 }, { "epoch": 1.3260226250366713e-13, "grad_norm": 2.2701377868652344, "learning_rate": 0.00014009838814991196, "loss": 0.2468, "step": 19110 }, { "epoch": 1.326716514427062e-13, "grad_norm": 1.7641688585281372, "learning_rate": 0.00014008710276170502, "loss": 0.2561, "step": 19120 }, { "epoch": 1.3274104038174528e-13, "grad_norm": 1.429370403289795, "learning_rate": 0.0001400758114555272, "loss": 0.2336, "step": 19130 }, { "epoch": 1.3281042932078435e-13, "grad_norm": 1.66506028175354, "learning_rate": 0.00014006451423253895, "loss": 0.2173, "step": 19140 }, { "epoch": 1.3287981825982342e-13, "grad_norm": 1.8693960905075073, "learning_rate": 0.0001400532110939012, "loss": 0.2585, "step": 19150 }, { "epoch": 1.329492071988625e-13, "grad_norm": 1.46687650680542, "learning_rate": 0.00014004190204077553, "loss": 0.2251, "step": 19160 }, { "epoch": 1.3301859613790157e-13, "grad_norm": 1.1382837295532227, "learning_rate": 0.0001400305870743241, "loss": 0.2229, "step": 19170 }, { "epoch": 1.3308798507694064e-13, "grad_norm": 1.5619769096374512, "learning_rate": 0.0001400192661957097, "loss": 0.2392, "step": 19180 }, { "epoch": 1.331573740159797e-13, "grad_norm": 1.60714852809906, "learning_rate": 0.0001400079394060958, "loss": 0.221, "step": 19190 }, { "epoch": 1.3322676295501878e-13, "grad_norm": 1.7675917148590088, "learning_rate": 0.00013999660670664627, "loss": 0.2421, "step": 19200 }, { "epoch": 1.3329615189405786e-13, "grad_norm": 1.5503729581832886, "learning_rate": 0.00013998526809852582, "loss": 0.2335, "step": 19210 }, { "epoch": 1.3336554083309693e-13, "grad_norm": 1.6907007694244385, "learning_rate": 0.00013997392358289964, "loss": 0.2249, "step": 19220 }, { "epoch": 1.33434929772136e-13, "grad_norm": 1.7539230585098267, "learning_rate": 0.00013996257316093355, "loss": 0.2263, "step": 19230 }, { "epoch": 1.3350431871117507e-13, "grad_norm": 1.8506344556808472, "learning_rate": 0.000139951216833794, "loss": 0.2529, "step": 19240 }, { "epoch": 1.3357370765021415e-13, "grad_norm": 1.8371599912643433, "learning_rate": 0.000139939854602648, "loss": 0.2155, "step": 19250 }, { "epoch": 1.3364309658925322e-13, "grad_norm": 1.5641376972198486, "learning_rate": 0.00013992848646866325, "loss": 0.2568, "step": 19260 }, { "epoch": 1.337124855282923e-13, "grad_norm": 1.617890477180481, "learning_rate": 0.00013991711243300795, "loss": 0.2688, "step": 19270 }, { "epoch": 1.3378187446733136e-13, "grad_norm": 1.7757060527801514, "learning_rate": 0.00013990573249685095, "loss": 0.2316, "step": 19280 }, { "epoch": 1.3385126340637044e-13, "grad_norm": 1.748121976852417, "learning_rate": 0.00013989434666136178, "loss": 0.2204, "step": 19290 }, { "epoch": 1.339206523454095e-13, "grad_norm": 1.767255187034607, "learning_rate": 0.00013988295492771047, "loss": 0.244, "step": 19300 }, { "epoch": 1.3399004128444858e-13, "grad_norm": 1.7915669679641724, "learning_rate": 0.0001398715572970677, "loss": 0.2381, "step": 19310 }, { "epoch": 1.3405943022348765e-13, "grad_norm": 1.480228066444397, "learning_rate": 0.00013986015377060474, "loss": 0.2054, "step": 19320 }, { "epoch": 1.3412881916252672e-13, "grad_norm": 1.7377631664276123, "learning_rate": 0.00013984874434949348, "loss": 0.225, "step": 19330 }, { "epoch": 1.341982081015658e-13, "grad_norm": 1.313896656036377, "learning_rate": 0.00013983732903490647, "loss": 0.2266, "step": 19340 }, { "epoch": 1.3426759704060487e-13, "grad_norm": 1.6765241622924805, "learning_rate": 0.00013982590782801675, "loss": 0.2157, "step": 19350 }, { "epoch": 1.3433698597964394e-13, "grad_norm": 1.9411529302597046, "learning_rate": 0.00013981448072999806, "loss": 0.2306, "step": 19360 }, { "epoch": 1.3440637491868301e-13, "grad_norm": 1.433658480644226, "learning_rate": 0.0001398030477420247, "loss": 0.2259, "step": 19370 }, { "epoch": 1.3447576385772209e-13, "grad_norm": 1.6355050802230835, "learning_rate": 0.0001397916088652716, "loss": 0.2317, "step": 19380 }, { "epoch": 1.3454515279676116e-13, "grad_norm": 1.7436949014663696, "learning_rate": 0.00013978016410091425, "loss": 0.239, "step": 19390 }, { "epoch": 1.3461454173580023e-13, "grad_norm": 1.418057918548584, "learning_rate": 0.00013976871345012878, "loss": 0.2203, "step": 19400 }, { "epoch": 1.346839306748393e-13, "grad_norm": 1.8976993560791016, "learning_rate": 0.00013975725691409195, "loss": 0.2379, "step": 19410 }, { "epoch": 1.3475331961387838e-13, "grad_norm": 1.7285581827163696, "learning_rate": 0.00013974579449398111, "loss": 0.2507, "step": 19420 }, { "epoch": 1.3482270855291745e-13, "grad_norm": 1.3844739198684692, "learning_rate": 0.0001397343261909742, "loss": 0.2682, "step": 19430 }, { "epoch": 1.3489209749195652e-13, "grad_norm": 1.204055666923523, "learning_rate": 0.0001397228520062497, "loss": 0.2157, "step": 19440 }, { "epoch": 1.349614864309956e-13, "grad_norm": 1.5922874212265015, "learning_rate": 0.00013971137194098682, "loss": 0.2248, "step": 19450 }, { "epoch": 1.3503087537003466e-13, "grad_norm": 1.2430839538574219, "learning_rate": 0.0001396998859963653, "loss": 0.2383, "step": 19460 }, { "epoch": 1.3510026430907374e-13, "grad_norm": 1.3282335996627808, "learning_rate": 0.00013968839417356554, "loss": 0.2188, "step": 19470 }, { "epoch": 1.351696532481128e-13, "grad_norm": 1.4735538959503174, "learning_rate": 0.00013967689647376844, "loss": 0.2111, "step": 19480 }, { "epoch": 1.3523904218715188e-13, "grad_norm": 1.6087640523910522, "learning_rate": 0.00013966539289815558, "loss": 0.2108, "step": 19490 }, { "epoch": 1.3530843112619095e-13, "grad_norm": 1.442710518836975, "learning_rate": 0.00013965388344790915, "loss": 0.2141, "step": 19500 }, { "epoch": 1.3537782006523003e-13, "grad_norm": 1.933143973350525, "learning_rate": 0.00013964236812421196, "loss": 0.2141, "step": 19510 }, { "epoch": 1.354472090042691e-13, "grad_norm": 1.8268013000488281, "learning_rate": 0.0001396308469282473, "loss": 0.2381, "step": 19520 }, { "epoch": 1.3551659794330817e-13, "grad_norm": 1.7104865312576294, "learning_rate": 0.00013961931986119922, "loss": 0.2042, "step": 19530 }, { "epoch": 1.3558598688234724e-13, "grad_norm": 1.5465118885040283, "learning_rate": 0.0001396077869242523, "loss": 0.2229, "step": 19540 }, { "epoch": 1.3565537582138631e-13, "grad_norm": 1.590201497077942, "learning_rate": 0.0001395962481185917, "loss": 0.2531, "step": 19550 }, { "epoch": 1.3572476476042539e-13, "grad_norm": 1.5865511894226074, "learning_rate": 0.00013958470344540324, "loss": 0.2088, "step": 19560 }, { "epoch": 1.3579415369946446e-13, "grad_norm": 1.9219495058059692, "learning_rate": 0.00013957315290587328, "loss": 0.2678, "step": 19570 }, { "epoch": 1.3586354263850353e-13, "grad_norm": 1.8602383136749268, "learning_rate": 0.00013956159650118884, "loss": 0.258, "step": 19580 }, { "epoch": 1.359329315775426e-13, "grad_norm": 1.747678279876709, "learning_rate": 0.00013955003423253754, "loss": 0.236, "step": 19590 }, { "epoch": 1.3600232051658168e-13, "grad_norm": 1.4046849012374878, "learning_rate": 0.00013953846610110756, "loss": 0.2062, "step": 19600 }, { "epoch": 1.3607170945562075e-13, "grad_norm": 1.8028347492218018, "learning_rate": 0.0001395268921080877, "loss": 0.2096, "step": 19610 }, { "epoch": 1.3614109839465982e-13, "grad_norm": 1.6693408489227295, "learning_rate": 0.00013951531225466738, "loss": 0.2207, "step": 19620 }, { "epoch": 1.362104873336989e-13, "grad_norm": 1.6453253030776978, "learning_rate": 0.00013950372654203663, "loss": 0.2088, "step": 19630 }, { "epoch": 1.3627987627273797e-13, "grad_norm": 1.2954292297363281, "learning_rate": 0.00013949213497138602, "loss": 0.1968, "step": 19640 }, { "epoch": 1.3634926521177704e-13, "grad_norm": 1.821724772453308, "learning_rate": 0.00013948053754390678, "loss": 0.2304, "step": 19650 }, { "epoch": 1.364186541508161e-13, "grad_norm": 1.7084071636199951, "learning_rate": 0.00013946893426079073, "loss": 0.1936, "step": 19660 }, { "epoch": 1.3648804308985518e-13, "grad_norm": 1.2234588861465454, "learning_rate": 0.00013945732512323026, "loss": 0.2152, "step": 19670 }, { "epoch": 1.3655743202889425e-13, "grad_norm": 1.5975722074508667, "learning_rate": 0.00013944571013241848, "loss": 0.2174, "step": 19680 }, { "epoch": 1.3662682096793333e-13, "grad_norm": 1.4479063749313354, "learning_rate": 0.0001394340892895489, "loss": 0.2089, "step": 19690 }, { "epoch": 1.366962099069724e-13, "grad_norm": 1.3977911472320557, "learning_rate": 0.0001394224625958158, "loss": 0.2054, "step": 19700 }, { "epoch": 1.3676559884601147e-13, "grad_norm": 1.385676622390747, "learning_rate": 0.00013941083005241398, "loss": 0.2378, "step": 19710 }, { "epoch": 1.3683498778505054e-13, "grad_norm": 1.778881549835205, "learning_rate": 0.0001393991916605389, "loss": 0.2128, "step": 19720 }, { "epoch": 1.3690437672408962e-13, "grad_norm": 1.7528373003005981, "learning_rate": 0.00013938754742138655, "loss": 0.2378, "step": 19730 }, { "epoch": 1.369737656631287e-13, "grad_norm": 1.5372416973114014, "learning_rate": 0.00013937589733615358, "loss": 0.257, "step": 19740 }, { "epoch": 1.3704315460216776e-13, "grad_norm": 1.828014850616455, "learning_rate": 0.00013936424140603723, "loss": 0.2221, "step": 19750 }, { "epoch": 1.3711254354120683e-13, "grad_norm": 1.6069269180297852, "learning_rate": 0.00013935257963223528, "loss": 0.2016, "step": 19760 }, { "epoch": 1.371819324802459e-13, "grad_norm": 1.667110800743103, "learning_rate": 0.00013934091201594617, "loss": 0.2132, "step": 19770 }, { "epoch": 1.3725132141928498e-13, "grad_norm": 1.6764041185379028, "learning_rate": 0.00013932923855836896, "loss": 0.2115, "step": 19780 }, { "epoch": 1.3732071035832405e-13, "grad_norm": 1.6458953619003296, "learning_rate": 0.00013931755926070325, "loss": 0.183, "step": 19790 }, { "epoch": 1.3739009929736312e-13, "grad_norm": 1.6089006662368774, "learning_rate": 0.00013930587412414933, "loss": 0.2082, "step": 19800 }, { "epoch": 1.374594882364022e-13, "grad_norm": 1.635711908340454, "learning_rate": 0.00013929418314990796, "loss": 0.2064, "step": 19810 }, { "epoch": 1.3752887717544127e-13, "grad_norm": 1.4413493871688843, "learning_rate": 0.00013928248633918058, "loss": 0.1943, "step": 19820 }, { "epoch": 1.3759826611448034e-13, "grad_norm": 1.4470081329345703, "learning_rate": 0.00013927078369316924, "loss": 0.2069, "step": 19830 }, { "epoch": 1.376676550535194e-13, "grad_norm": 1.5590804815292358, "learning_rate": 0.00013925907521307657, "loss": 0.2026, "step": 19840 }, { "epoch": 1.3773704399255848e-13, "grad_norm": 1.3941737413406372, "learning_rate": 0.00013924736090010576, "loss": 0.2019, "step": 19850 }, { "epoch": 1.3780643293159756e-13, "grad_norm": 1.6604894399642944, "learning_rate": 0.00013923564075546068, "loss": 0.2229, "step": 19860 }, { "epoch": 1.3787582187063663e-13, "grad_norm": 1.4791662693023682, "learning_rate": 0.0001392239147803458, "loss": 0.1933, "step": 19870 }, { "epoch": 1.379452108096757e-13, "grad_norm": 1.8254674673080444, "learning_rate": 0.00013921218297596602, "loss": 0.2315, "step": 19880 }, { "epoch": 1.3801459974871477e-13, "grad_norm": 1.8056483268737793, "learning_rate": 0.00013920044534352705, "loss": 0.2304, "step": 19890 }, { "epoch": 1.3808398868775384e-13, "grad_norm": 1.8971314430236816, "learning_rate": 0.00013918870188423514, "loss": 0.239, "step": 19900 }, { "epoch": 1.3815337762679292e-13, "grad_norm": 2.1466100215911865, "learning_rate": 0.00013917695259929702, "loss": 0.2141, "step": 19910 }, { "epoch": 1.38222766565832e-13, "grad_norm": 1.2131363153457642, "learning_rate": 0.0001391651974899202, "loss": 0.1827, "step": 19920 }, { "epoch": 1.3829215550487106e-13, "grad_norm": 1.1727452278137207, "learning_rate": 0.00013915343655731267, "loss": 0.212, "step": 19930 }, { "epoch": 1.3836154444391013e-13, "grad_norm": 1.4068236351013184, "learning_rate": 0.00013914166980268303, "loss": 0.2016, "step": 19940 }, { "epoch": 1.384309333829492e-13, "grad_norm": 1.6544520854949951, "learning_rate": 0.0001391298972272405, "loss": 0.2026, "step": 19950 }, { "epoch": 1.3850032232198828e-13, "grad_norm": 1.7311147451400757, "learning_rate": 0.00013911811883219493, "loss": 0.2012, "step": 19960 }, { "epoch": 1.3856971126102735e-13, "grad_norm": 1.5734703540802002, "learning_rate": 0.0001391063346187567, "loss": 0.1912, "step": 19970 }, { "epoch": 1.3863910020006642e-13, "grad_norm": 1.6789894104003906, "learning_rate": 0.00013909454458813685, "loss": 0.2064, "step": 19980 }, { "epoch": 1.387084891391055e-13, "grad_norm": 1.3988325595855713, "learning_rate": 0.00013908274874154696, "loss": 0.1996, "step": 19990 }, { "epoch": 1.3877787807814457e-13, "grad_norm": 1.500465750694275, "learning_rate": 0.0001390709470801992, "loss": 0.188, "step": 20000 }, { "epoch": 1.3884726701718364e-13, "grad_norm": 1.6067993640899658, "learning_rate": 0.0001390591396053065, "loss": 0.1954, "step": 20010 }, { "epoch": 1.389166559562227e-13, "grad_norm": 1.4723458290100098, "learning_rate": 0.00013904732631808215, "loss": 0.2271, "step": 20020 }, { "epoch": 1.3898604489526178e-13, "grad_norm": 1.4983866214752197, "learning_rate": 0.0001390355072197402, "loss": 0.1882, "step": 20030 }, { "epoch": 1.3905543383430086e-13, "grad_norm": 1.4769229888916016, "learning_rate": 0.0001390236823114952, "loss": 0.2397, "step": 20040 }, { "epoch": 1.3912482277333993e-13, "grad_norm": 1.5674223899841309, "learning_rate": 0.0001390118515945624, "loss": 0.2375, "step": 20050 }, { "epoch": 1.39194211712379e-13, "grad_norm": 1.3987514972686768, "learning_rate": 0.00013900001507015751, "loss": 0.1982, "step": 20060 }, { "epoch": 1.3926360065141807e-13, "grad_norm": 1.4657530784606934, "learning_rate": 0.00013898817273949704, "loss": 0.1959, "step": 20070 }, { "epoch": 1.3933298959045715e-13, "grad_norm": 1.818324327468872, "learning_rate": 0.00013897632460379788, "loss": 0.203, "step": 20080 }, { "epoch": 1.3940237852949622e-13, "grad_norm": 1.5295735597610474, "learning_rate": 0.00013896447066427765, "loss": 0.195, "step": 20090 }, { "epoch": 1.394717674685353e-13, "grad_norm": 1.3156217336654663, "learning_rate": 0.00013895261092215447, "loss": 0.1875, "step": 20100 }, { "epoch": 1.3954115640757436e-13, "grad_norm": 1.6953608989715576, "learning_rate": 0.00013894074537864717, "loss": 0.1921, "step": 20110 }, { "epoch": 1.3961054534661343e-13, "grad_norm": 1.3794904947280884, "learning_rate": 0.00013892887403497511, "loss": 0.2158, "step": 20120 }, { "epoch": 1.396799342856525e-13, "grad_norm": 1.4157909154891968, "learning_rate": 0.00013891699689235824, "loss": 0.1815, "step": 20130 }, { "epoch": 1.3974932322469158e-13, "grad_norm": 1.217581868171692, "learning_rate": 0.00013890511395201713, "loss": 0.191, "step": 20140 }, { "epoch": 1.3981871216373065e-13, "grad_norm": 1.226813554763794, "learning_rate": 0.0001388932252151729, "loss": 0.197, "step": 20150 }, { "epoch": 1.3988810110276972e-13, "grad_norm": 1.4698500633239746, "learning_rate": 0.00013888133068304742, "loss": 0.1971, "step": 20160 }, { "epoch": 1.399574900418088e-13, "grad_norm": 1.9427233934402466, "learning_rate": 0.00013886943035686293, "loss": 0.2048, "step": 20170 }, { "epoch": 1.4002687898084787e-13, "grad_norm": 1.200546383857727, "learning_rate": 0.00013885752423784236, "loss": 0.2085, "step": 20180 }, { "epoch": 1.4009626791988694e-13, "grad_norm": 1.5043128728866577, "learning_rate": 0.00013884561232720933, "loss": 0.202, "step": 20190 }, { "epoch": 1.4016565685892601e-13, "grad_norm": 1.4472593069076538, "learning_rate": 0.0001388336946261879, "loss": 0.2276, "step": 20200 }, { "epoch": 1.4023504579796509e-13, "grad_norm": 1.5552386045455933, "learning_rate": 0.0001388217711360029, "loss": 0.2332, "step": 20210 }, { "epoch": 1.4030443473700416e-13, "grad_norm": 1.3368918895721436, "learning_rate": 0.00013880984185787952, "loss": 0.1991, "step": 20220 }, { "epoch": 1.4037382367604323e-13, "grad_norm": 1.3134844303131104, "learning_rate": 0.00013879790679304378, "loss": 0.1932, "step": 20230 }, { "epoch": 1.404432126150823e-13, "grad_norm": 1.5469311475753784, "learning_rate": 0.00013878596594272215, "loss": 0.2116, "step": 20240 }, { "epoch": 1.4051260155412137e-13, "grad_norm": 1.3769437074661255, "learning_rate": 0.00013877401930814174, "loss": 0.1896, "step": 20250 }, { "epoch": 1.4058199049316045e-13, "grad_norm": 1.6334459781646729, "learning_rate": 0.00013876206689053026, "loss": 0.1839, "step": 20260 }, { "epoch": 1.4065137943219952e-13, "grad_norm": 1.3582990169525146, "learning_rate": 0.00013875010869111601, "loss": 0.1942, "step": 20270 }, { "epoch": 1.407207683712386e-13, "grad_norm": 1.2537903785705566, "learning_rate": 0.00013873814471112789, "loss": 0.1977, "step": 20280 }, { "epoch": 1.4079015731027766e-13, "grad_norm": 1.2574717998504639, "learning_rate": 0.00013872617495179536, "loss": 0.1835, "step": 20290 }, { "epoch": 1.4085954624931674e-13, "grad_norm": 1.8310041427612305, "learning_rate": 0.00013871419941434855, "loss": 0.2082, "step": 20300 }, { "epoch": 1.409289351883558e-13, "grad_norm": 1.3459484577178955, "learning_rate": 0.00013870221810001807, "loss": 0.1764, "step": 20310 }, { "epoch": 1.4099832412739488e-13, "grad_norm": 1.5367748737335205, "learning_rate": 0.0001386902310100352, "loss": 0.1901, "step": 20320 }, { "epoch": 1.4106771306643395e-13, "grad_norm": 1.8830277919769287, "learning_rate": 0.0001386782381456318, "loss": 0.2163, "step": 20330 }, { "epoch": 1.4113710200547303e-13, "grad_norm": 1.6396079063415527, "learning_rate": 0.00013866623950804037, "loss": 0.1801, "step": 20340 }, { "epoch": 1.412064909445121e-13, "grad_norm": 2.208148956298828, "learning_rate": 0.0001386542350984939, "loss": 0.2262, "step": 20350 }, { "epoch": 1.4127587988355117e-13, "grad_norm": 1.946086049079895, "learning_rate": 0.00013864222491822607, "loss": 0.216, "step": 20360 }, { "epoch": 1.4134526882259024e-13, "grad_norm": 1.4824920892715454, "learning_rate": 0.0001386302089684711, "loss": 0.2077, "step": 20370 }, { "epoch": 1.4141465776162931e-13, "grad_norm": 1.5476740598678589, "learning_rate": 0.0001386181872504638, "loss": 0.2001, "step": 20380 }, { "epoch": 1.4148404670066839e-13, "grad_norm": 1.5397588014602661, "learning_rate": 0.0001386061597654396, "loss": 0.1755, "step": 20390 }, { "epoch": 1.4155343563970746e-13, "grad_norm": 1.662587285041809, "learning_rate": 0.0001385941265146345, "loss": 0.2081, "step": 20400 }, { "epoch": 1.4162282457874653e-13, "grad_norm": 1.5815485715866089, "learning_rate": 0.00013858208749928513, "loss": 0.1912, "step": 20410 }, { "epoch": 1.416922135177856e-13, "grad_norm": 1.12846040725708, "learning_rate": 0.00013857004272062865, "loss": 0.1892, "step": 20420 }, { "epoch": 1.4176160245682468e-13, "grad_norm": 1.5368242263793945, "learning_rate": 0.0001385579921799029, "loss": 0.2015, "step": 20430 }, { "epoch": 1.4183099139586375e-13, "grad_norm": 1.4661269187927246, "learning_rate": 0.0001385459358783462, "loss": 0.1913, "step": 20440 }, { "epoch": 1.4190038033490282e-13, "grad_norm": 1.5241929292678833, "learning_rate": 0.0001385338738171976, "loss": 0.2049, "step": 20450 }, { "epoch": 1.419697692739419e-13, "grad_norm": 1.34172523021698, "learning_rate": 0.00013852180599769657, "loss": 0.1857, "step": 20460 }, { "epoch": 1.4203915821298096e-13, "grad_norm": 1.4936801195144653, "learning_rate": 0.00013850973242108332, "loss": 0.1821, "step": 20470 }, { "epoch": 1.4210854715202004e-13, "grad_norm": 1.7043406963348389, "learning_rate": 0.0001384976530885986, "loss": 0.1808, "step": 20480 }, { "epoch": 1.421779360910591e-13, "grad_norm": 1.5759611129760742, "learning_rate": 0.00013848556800148375, "loss": 0.2061, "step": 20490 }, { "epoch": 1.4224732503009818e-13, "grad_norm": 1.7007288932800293, "learning_rate": 0.00013847347716098066, "loss": 0.1747, "step": 20500 }, { "epoch": 1.4231671396913725e-13, "grad_norm": 1.4732060432434082, "learning_rate": 0.0001384613805683319, "loss": 0.2116, "step": 20510 }, { "epoch": 1.4238610290817633e-13, "grad_norm": 1.4071300029754639, "learning_rate": 0.00013844927822478058, "loss": 0.2268, "step": 20520 }, { "epoch": 1.424554918472154e-13, "grad_norm": 1.7897052764892578, "learning_rate": 0.00013843717013157036, "loss": 0.1906, "step": 20530 }, { "epoch": 1.4252488078625447e-13, "grad_norm": 1.821205496788025, "learning_rate": 0.00013842505628994557, "loss": 0.1893, "step": 20540 }, { "epoch": 1.4259426972529354e-13, "grad_norm": 1.480617880821228, "learning_rate": 0.00013841293670115108, "loss": 0.1968, "step": 20550 }, { "epoch": 1.4266365866433262e-13, "grad_norm": 1.5834845304489136, "learning_rate": 0.00013840081136643243, "loss": 0.1913, "step": 20560 }, { "epoch": 1.427330476033717e-13, "grad_norm": 1.528291940689087, "learning_rate": 0.00013838868028703558, "loss": 0.177, "step": 20570 }, { "epoch": 1.4280243654241076e-13, "grad_norm": 1.7074694633483887, "learning_rate": 0.0001383765434642073, "loss": 0.1942, "step": 20580 }, { "epoch": 1.4287182548144983e-13, "grad_norm": 1.450757622718811, "learning_rate": 0.0001383644008991947, "loss": 0.199, "step": 20590 }, { "epoch": 1.429412144204889e-13, "grad_norm": 1.692602276802063, "learning_rate": 0.00013835225259324575, "loss": 0.1888, "step": 20600 }, { "epoch": 1.4301060335952798e-13, "grad_norm": 1.8443726301193237, "learning_rate": 0.0001383400985476088, "loss": 0.193, "step": 20610 }, { "epoch": 1.4307999229856705e-13, "grad_norm": 1.4420338869094849, "learning_rate": 0.00013832793876353288, "loss": 0.186, "step": 20620 }, { "epoch": 1.4314938123760612e-13, "grad_norm": 1.4226974248886108, "learning_rate": 0.00013831577324226765, "loss": 0.1871, "step": 20630 }, { "epoch": 1.432187701766452e-13, "grad_norm": 1.5973148345947266, "learning_rate": 0.00013830360198506325, "loss": 0.1936, "step": 20640 }, { "epoch": 1.4328815911568427e-13, "grad_norm": 1.1947746276855469, "learning_rate": 0.00013829142499317045, "loss": 0.1835, "step": 20650 }, { "epoch": 1.4335754805472334e-13, "grad_norm": 1.5609142780303955, "learning_rate": 0.00013827924226784069, "loss": 0.1835, "step": 20660 }, { "epoch": 1.434269369937624e-13, "grad_norm": 1.5583231449127197, "learning_rate": 0.0001382670538103259, "loss": 0.2069, "step": 20670 }, { "epoch": 1.4349632593280148e-13, "grad_norm": 1.2412151098251343, "learning_rate": 0.00013825485962187862, "loss": 0.218, "step": 20680 }, { "epoch": 1.4356571487184056e-13, "grad_norm": 1.2420296669006348, "learning_rate": 0.000138242659703752, "loss": 0.1816, "step": 20690 }, { "epoch": 1.4363510381087963e-13, "grad_norm": 1.5779426097869873, "learning_rate": 0.00013823045405719978, "loss": 0.1966, "step": 20700 }, { "epoch": 1.437044927499187e-13, "grad_norm": 1.155928373336792, "learning_rate": 0.0001382182426834763, "loss": 0.2005, "step": 20710 }, { "epoch": 1.4377388168895777e-13, "grad_norm": 1.241661548614502, "learning_rate": 0.00013820602558383644, "loss": 0.1763, "step": 20720 }, { "epoch": 1.4384327062799684e-13, "grad_norm": 1.5252103805541992, "learning_rate": 0.00013819380275953568, "loss": 0.1833, "step": 20730 }, { "epoch": 1.4391265956703592e-13, "grad_norm": 1.5249539613723755, "learning_rate": 0.00013818157421183015, "loss": 0.1812, "step": 20740 }, { "epoch": 1.43982048506075e-13, "grad_norm": 1.5087400674819946, "learning_rate": 0.0001381693399419765, "loss": 0.1932, "step": 20750 }, { "epoch": 1.4405143744511406e-13, "grad_norm": 1.9458907842636108, "learning_rate": 0.000138157099951232, "loss": 0.1868, "step": 20760 }, { "epoch": 1.4412082638415313e-13, "grad_norm": 1.814105749130249, "learning_rate": 0.00013814485424085447, "loss": 0.2026, "step": 20770 }, { "epoch": 1.441902153231922e-13, "grad_norm": 1.485484004020691, "learning_rate": 0.00013813260281210238, "loss": 0.1677, "step": 20780 }, { "epoch": 1.4425960426223128e-13, "grad_norm": 1.3023546934127808, "learning_rate": 0.00013812034566623474, "loss": 0.1734, "step": 20790 }, { "epoch": 1.4432899320127035e-13, "grad_norm": 1.455148696899414, "learning_rate": 0.00013810808280451117, "loss": 0.2059, "step": 20800 }, { "epoch": 1.4439838214030942e-13, "grad_norm": 1.3973414897918701, "learning_rate": 0.00013809581422819192, "loss": 0.1665, "step": 20810 }, { "epoch": 1.444677710793485e-13, "grad_norm": 1.4032175540924072, "learning_rate": 0.00013808353993853767, "loss": 0.1989, "step": 20820 }, { "epoch": 1.4453716001838757e-13, "grad_norm": 1.784487247467041, "learning_rate": 0.00013807125993680987, "loss": 0.2125, "step": 20830 }, { "epoch": 1.4460654895742664e-13, "grad_norm": 1.6433801651000977, "learning_rate": 0.00013805897422427044, "loss": 0.1886, "step": 20840 }, { "epoch": 1.446759378964657e-13, "grad_norm": 1.4983279705047607, "learning_rate": 0.00013804668280218197, "loss": 0.1797, "step": 20850 }, { "epoch": 1.4474532683550478e-13, "grad_norm": 2.1090025901794434, "learning_rate": 0.00013803438567180755, "loss": 0.1894, "step": 20860 }, { "epoch": 1.4481471577454386e-13, "grad_norm": 1.6589642763137817, "learning_rate": 0.00013802208283441095, "loss": 0.1906, "step": 20870 }, { "epoch": 1.4488410471358293e-13, "grad_norm": 1.6117457151412964, "learning_rate": 0.00013800977429125644, "loss": 0.1769, "step": 20880 }, { "epoch": 1.44953493652622e-13, "grad_norm": 1.2392094135284424, "learning_rate": 0.00013799746004360892, "loss": 0.168, "step": 20890 }, { "epoch": 1.4502288259166107e-13, "grad_norm": 1.9797112941741943, "learning_rate": 0.0001379851400927339, "loss": 0.2026, "step": 20900 }, { "epoch": 1.4509227153070015e-13, "grad_norm": 1.525684118270874, "learning_rate": 0.00013797281443989743, "loss": 0.1719, "step": 20910 }, { "epoch": 1.4516166046973922e-13, "grad_norm": 1.1668154001235962, "learning_rate": 0.00013796048308636612, "loss": 0.1852, "step": 20920 }, { "epoch": 1.452310494087783e-13, "grad_norm": 1.5076528787612915, "learning_rate": 0.00013794814603340728, "loss": 0.1865, "step": 20930 }, { "epoch": 1.4530043834781736e-13, "grad_norm": 1.2321621179580688, "learning_rate": 0.0001379358032822887, "loss": 0.1733, "step": 20940 }, { "epoch": 1.4536982728685643e-13, "grad_norm": 1.234960913658142, "learning_rate": 0.00013792345483427877, "loss": 0.1664, "step": 20950 }, { "epoch": 1.454392162258955e-13, "grad_norm": 1.4358619451522827, "learning_rate": 0.00013791110069064652, "loss": 0.2024, "step": 20960 }, { "epoch": 1.4550860516493458e-13, "grad_norm": 1.57178795337677, "learning_rate": 0.00013789874085266149, "loss": 0.1664, "step": 20970 }, { "epoch": 1.4557799410397365e-13, "grad_norm": 1.6166778802871704, "learning_rate": 0.0001378863753215939, "loss": 0.1839, "step": 20980 }, { "epoch": 1.4564738304301272e-13, "grad_norm": 1.4963871240615845, "learning_rate": 0.00013787400409871444, "loss": 0.2093, "step": 20990 }, { "epoch": 1.457167719820518e-13, "grad_norm": 1.6471776962280273, "learning_rate": 0.0001378616271852945, "loss": 0.1801, "step": 21000 }, { "epoch": 1.4578616092109087e-13, "grad_norm": 1.8013348579406738, "learning_rate": 0.00013784924458260601, "loss": 0.1841, "step": 21010 }, { "epoch": 1.4585554986012994e-13, "grad_norm": 1.7296448945999146, "learning_rate": 0.0001378368562919214, "loss": 0.1946, "step": 21020 }, { "epoch": 1.4592493879916901e-13, "grad_norm": 1.6815780401229858, "learning_rate": 0.0001378244623145138, "loss": 0.1835, "step": 21030 }, { "epoch": 1.4599432773820809e-13, "grad_norm": 1.761411190032959, "learning_rate": 0.00013781206265165688, "loss": 0.1596, "step": 21040 }, { "epoch": 1.4606371667724716e-13, "grad_norm": 1.5276859998703003, "learning_rate": 0.00013779965730462493, "loss": 0.1754, "step": 21050 }, { "epoch": 1.4613310561628623e-13, "grad_norm": 1.758154034614563, "learning_rate": 0.00013778724627469275, "loss": 0.1769, "step": 21060 }, { "epoch": 1.462024945553253e-13, "grad_norm": 1.2912992238998413, "learning_rate": 0.0001377748295631358, "loss": 0.164, "step": 21070 }, { "epoch": 1.4627188349436437e-13, "grad_norm": 1.418284296989441, "learning_rate": 0.00013776240717123007, "loss": 0.1794, "step": 21080 }, { "epoch": 1.4634127243340345e-13, "grad_norm": 1.506644606590271, "learning_rate": 0.00013774997910025215, "loss": 0.1782, "step": 21090 }, { "epoch": 1.4641066137244252e-13, "grad_norm": 1.2078807353973389, "learning_rate": 0.00013773754535147923, "loss": 0.1689, "step": 21100 }, { "epoch": 1.464800503114816e-13, "grad_norm": 1.5635913610458374, "learning_rate": 0.00013772510592618907, "loss": 0.1824, "step": 21110 }, { "epoch": 1.4654943925052066e-13, "grad_norm": 1.5408225059509277, "learning_rate": 0.00013771266082566002, "loss": 0.1681, "step": 21120 }, { "epoch": 1.4661882818955974e-13, "grad_norm": 1.6544259786605835, "learning_rate": 0.00013770021005117097, "loss": 0.1823, "step": 21130 }, { "epoch": 1.466882171285988e-13, "grad_norm": 1.859809160232544, "learning_rate": 0.0001376877536040015, "loss": 0.1874, "step": 21140 }, { "epoch": 1.4675760606763788e-13, "grad_norm": 1.5739080905914307, "learning_rate": 0.0001376752914854316, "loss": 0.1912, "step": 21150 }, { "epoch": 1.4682699500667695e-13, "grad_norm": 2.0056333541870117, "learning_rate": 0.00013766282369674206, "loss": 0.1717, "step": 21160 }, { "epoch": 1.4689638394571602e-13, "grad_norm": 1.20939040184021, "learning_rate": 0.0001376503502392141, "loss": 0.1646, "step": 21170 }, { "epoch": 1.469657728847551e-13, "grad_norm": 1.0969387292861938, "learning_rate": 0.00013763787111412953, "loss": 0.1883, "step": 21180 }, { "epoch": 1.4703516182379417e-13, "grad_norm": 1.3780031204223633, "learning_rate": 0.00013762538632277078, "loss": 0.172, "step": 21190 }, { "epoch": 1.4710455076283324e-13, "grad_norm": 1.4822860956192017, "learning_rate": 0.0001376128958664209, "loss": 0.175, "step": 21200 }, { "epoch": 1.4717393970187231e-13, "grad_norm": 1.5434200763702393, "learning_rate": 0.00013760039974636344, "loss": 0.1674, "step": 21210 }, { "epoch": 1.4724332864091139e-13, "grad_norm": 1.6243687868118286, "learning_rate": 0.00013758789796388258, "loss": 0.1678, "step": 21220 }, { "epoch": 1.4731271757995046e-13, "grad_norm": 1.7608634233474731, "learning_rate": 0.00013757539052026307, "loss": 0.1776, "step": 21230 }, { "epoch": 1.4738210651898953e-13, "grad_norm": 1.4528018236160278, "learning_rate": 0.00013756287741679027, "loss": 0.1762, "step": 21240 }, { "epoch": 1.474514954580286e-13, "grad_norm": 1.6013484001159668, "learning_rate": 0.00013755035865475003, "loss": 0.1671, "step": 21250 }, { "epoch": 1.4752088439706768e-13, "grad_norm": 1.401793360710144, "learning_rate": 0.00013753783423542894, "loss": 0.1626, "step": 21260 }, { "epoch": 1.4759027333610675e-13, "grad_norm": 1.4200857877731323, "learning_rate": 0.00013752530416011403, "loss": 0.1881, "step": 21270 }, { "epoch": 1.4765966227514582e-13, "grad_norm": 1.3998546600341797, "learning_rate": 0.00013751276843009295, "loss": 0.1596, "step": 21280 }, { "epoch": 1.477290512141849e-13, "grad_norm": 1.398518681526184, "learning_rate": 0.00013750022704665392, "loss": 0.1919, "step": 21290 }, { "epoch": 1.4779844015322396e-13, "grad_norm": 1.489148736000061, "learning_rate": 0.00013748768001108582, "loss": 0.2037, "step": 21300 }, { "epoch": 1.4786782909226304e-13, "grad_norm": 1.2090872526168823, "learning_rate": 0.00013747512732467803, "loss": 0.1622, "step": 21310 }, { "epoch": 1.479372180313021e-13, "grad_norm": 1.4013752937316895, "learning_rate": 0.00013746256898872054, "loss": 0.1618, "step": 21320 }, { "epoch": 1.4800660697034118e-13, "grad_norm": 1.9303805828094482, "learning_rate": 0.0001374500050045039, "loss": 0.1834, "step": 21330 }, { "epoch": 1.4807599590938025e-13, "grad_norm": 1.4557979106903076, "learning_rate": 0.00013743743537331926, "loss": 0.1719, "step": 21340 }, { "epoch": 1.4814538484841933e-13, "grad_norm": 1.4188294410705566, "learning_rate": 0.00013742486009645833, "loss": 0.1615, "step": 21350 }, { "epoch": 1.482147737874584e-13, "grad_norm": 1.6720573902130127, "learning_rate": 0.00013741227917521347, "loss": 0.1685, "step": 21360 }, { "epoch": 1.4828416272649747e-13, "grad_norm": 1.1677929162979126, "learning_rate": 0.0001373996926108775, "loss": 0.1829, "step": 21370 }, { "epoch": 1.4835355166553654e-13, "grad_norm": 1.2826924324035645, "learning_rate": 0.00013738710040474392, "loss": 0.1615, "step": 21380 }, { "epoch": 1.4842294060457561e-13, "grad_norm": 1.2959949970245361, "learning_rate": 0.00013737450255810676, "loss": 0.1693, "step": 21390 }, { "epoch": 1.4849232954361469e-13, "grad_norm": 1.2786033153533936, "learning_rate": 0.00013736189907226067, "loss": 0.1796, "step": 21400 }, { "epoch": 1.4856171848265376e-13, "grad_norm": 1.4809913635253906, "learning_rate": 0.00013734928994850083, "loss": 0.1723, "step": 21410 }, { "epoch": 1.4863110742169283e-13, "grad_norm": 1.7271045446395874, "learning_rate": 0.00013733667518812302, "loss": 0.1662, "step": 21420 }, { "epoch": 1.487004963607319e-13, "grad_norm": 1.1053634881973267, "learning_rate": 0.00013732405479242364, "loss": 0.1683, "step": 21430 }, { "epoch": 1.4876988529977098e-13, "grad_norm": 1.2967175245285034, "learning_rate": 0.00013731142876269957, "loss": 0.1635, "step": 21440 }, { "epoch": 1.4883927423881005e-13, "grad_norm": 1.3672832250595093, "learning_rate": 0.00013729879710024838, "loss": 0.1802, "step": 21450 }, { "epoch": 1.4890866317784912e-13, "grad_norm": 1.365996241569519, "learning_rate": 0.00013728615980636817, "loss": 0.1976, "step": 21460 }, { "epoch": 1.489780521168882e-13, "grad_norm": 1.1355242729187012, "learning_rate": 0.00013727351688235757, "loss": 0.1588, "step": 21470 }, { "epoch": 1.4904744105592727e-13, "grad_norm": 1.2798818349838257, "learning_rate": 0.0001372608683295159, "loss": 0.1622, "step": 21480 }, { "epoch": 1.4911682999496634e-13, "grad_norm": 1.7661277055740356, "learning_rate": 0.00013724821414914297, "loss": 0.1888, "step": 21490 }, { "epoch": 1.491862189340054e-13, "grad_norm": 1.2053446769714355, "learning_rate": 0.0001372355543425392, "loss": 0.1584, "step": 21500 }, { "epoch": 1.4925560787304448e-13, "grad_norm": 1.6729828119277954, "learning_rate": 0.00013722288891100554, "loss": 0.156, "step": 21510 }, { "epoch": 1.4932499681208355e-13, "grad_norm": 1.325523018836975, "learning_rate": 0.0001372102178558436, "loss": 0.1699, "step": 21520 }, { "epoch": 1.4939438575112263e-13, "grad_norm": 1.2301042079925537, "learning_rate": 0.00013719754117835555, "loss": 0.1656, "step": 21530 }, { "epoch": 1.494637746901617e-13, "grad_norm": 1.2667548656463623, "learning_rate": 0.00013718485887984406, "loss": 0.1551, "step": 21540 }, { "epoch": 1.4953316362920077e-13, "grad_norm": 1.9628373384475708, "learning_rate": 0.00013717217096161245, "loss": 0.1855, "step": 21550 }, { "epoch": 1.4960255256823984e-13, "grad_norm": 1.3795734643936157, "learning_rate": 0.00013715947742496467, "loss": 0.1606, "step": 21560 }, { "epoch": 1.4967194150727892e-13, "grad_norm": 1.460747480392456, "learning_rate": 0.00013714677827120507, "loss": 0.1684, "step": 21570 }, { "epoch": 1.49741330446318e-13, "grad_norm": 1.6113289594650269, "learning_rate": 0.00013713407350163874, "loss": 0.1796, "step": 21580 }, { "epoch": 1.4981071938535706e-13, "grad_norm": 1.4224910736083984, "learning_rate": 0.00013712136311757133, "loss": 0.1514, "step": 21590 }, { "epoch": 1.4988010832439613e-13, "grad_norm": 1.7575112581253052, "learning_rate": 0.00013710864712030893, "loss": 0.1761, "step": 21600 }, { "epoch": 1.499494972634352e-13, "grad_norm": 1.8757349252700806, "learning_rate": 0.0001370959255111584, "loss": 0.1752, "step": 21610 }, { "epoch": 1.5001888620247428e-13, "grad_norm": 1.4139723777770996, "learning_rate": 0.00013708319829142704, "loss": 0.1684, "step": 21620 }, { "epoch": 1.5008827514151335e-13, "grad_norm": 1.2983416318893433, "learning_rate": 0.0001370704654624228, "loss": 0.1578, "step": 21630 }, { "epoch": 1.5015766408055242e-13, "grad_norm": 1.4391160011291504, "learning_rate": 0.00013705772702545413, "loss": 0.1444, "step": 21640 }, { "epoch": 1.502270530195915e-13, "grad_norm": 1.6872400045394897, "learning_rate": 0.0001370449829818301, "loss": 0.1781, "step": 21650 }, { "epoch": 1.5029644195863057e-13, "grad_norm": 1.2601380348205566, "learning_rate": 0.00013703223333286045, "loss": 0.1563, "step": 21660 }, { "epoch": 1.5036583089766964e-13, "grad_norm": 1.024302363395691, "learning_rate": 0.00013701947807985532, "loss": 0.1625, "step": 21670 }, { "epoch": 1.504352198367087e-13, "grad_norm": 1.352124571800232, "learning_rate": 0.0001370067172241255, "loss": 0.1788, "step": 21680 }, { "epoch": 1.5050460877574778e-13, "grad_norm": 1.3742800951004028, "learning_rate": 0.00013699395076698242, "loss": 0.157, "step": 21690 }, { "epoch": 1.5057399771478686e-13, "grad_norm": 1.4490667581558228, "learning_rate": 0.000136981178709738, "loss": 0.1674, "step": 21700 }, { "epoch": 1.5064338665382593e-13, "grad_norm": 1.3534023761749268, "learning_rate": 0.00013696840105370482, "loss": 0.1603, "step": 21710 }, { "epoch": 1.50712775592865e-13, "grad_norm": 1.5683536529541016, "learning_rate": 0.0001369556178001959, "loss": 0.1658, "step": 21720 }, { "epoch": 1.5078216453190407e-13, "grad_norm": 1.458376407623291, "learning_rate": 0.00013694282895052495, "loss": 0.1592, "step": 21730 }, { "epoch": 1.5085155347094314e-13, "grad_norm": 1.530041217803955, "learning_rate": 0.00013693003450600624, "loss": 0.1785, "step": 21740 }, { "epoch": 1.5092094240998222e-13, "grad_norm": 1.4673631191253662, "learning_rate": 0.0001369172344679546, "loss": 0.1491, "step": 21750 }, { "epoch": 1.509903313490213e-13, "grad_norm": 1.2020940780639648, "learning_rate": 0.00013690442883768544, "loss": 0.1652, "step": 21760 }, { "epoch": 1.5105972028806036e-13, "grad_norm": 1.217270016670227, "learning_rate": 0.0001368916176165147, "loss": 0.1859, "step": 21770 }, { "epoch": 1.5112910922709943e-13, "grad_norm": 1.5137418508529663, "learning_rate": 0.00013687880080575897, "loss": 0.1539, "step": 21780 }, { "epoch": 1.511984981661385e-13, "grad_norm": 1.5772240161895752, "learning_rate": 0.00013686597840673533, "loss": 0.1518, "step": 21790 }, { "epoch": 1.5126788710517758e-13, "grad_norm": 1.465167760848999, "learning_rate": 0.00013685315042076153, "loss": 0.1624, "step": 21800 }, { "epoch": 1.5133727604421665e-13, "grad_norm": 1.4759265184402466, "learning_rate": 0.00013684031684915582, "loss": 0.1559, "step": 21810 }, { "epoch": 1.5140666498325572e-13, "grad_norm": 1.359091877937317, "learning_rate": 0.00013682747769323708, "loss": 0.1409, "step": 21820 }, { "epoch": 1.514760539222948e-13, "grad_norm": 1.642987847328186, "learning_rate": 0.0001368146329543247, "loss": 0.1655, "step": 21830 }, { "epoch": 1.5154544286133387e-13, "grad_norm": 1.1841685771942139, "learning_rate": 0.00013680178263373867, "loss": 0.1677, "step": 21840 }, { "epoch": 1.5161483180037294e-13, "grad_norm": 1.4139493703842163, "learning_rate": 0.00013678892673279957, "loss": 0.1534, "step": 21850 }, { "epoch": 1.51684220739412e-13, "grad_norm": 1.5098745822906494, "learning_rate": 0.00013677606525282855, "loss": 0.1528, "step": 21860 }, { "epoch": 1.5175360967845108e-13, "grad_norm": 1.3732857704162598, "learning_rate": 0.00013676319819514736, "loss": 0.1562, "step": 21870 }, { "epoch": 1.5182299861749016e-13, "grad_norm": 1.5104966163635254, "learning_rate": 0.00013675032556107822, "loss": 0.1668, "step": 21880 }, { "epoch": 1.5189238755652923e-13, "grad_norm": 1.5053534507751465, "learning_rate": 0.00013673744735194406, "loss": 0.1645, "step": 21890 }, { "epoch": 1.519617764955683e-13, "grad_norm": 1.2520101070404053, "learning_rate": 0.00013672456356906827, "loss": 0.1604, "step": 21900 }, { "epoch": 1.5203116543460737e-13, "grad_norm": 1.4670342206954956, "learning_rate": 0.0001367116742137749, "loss": 0.1611, "step": 21910 }, { "epoch": 1.5210055437364645e-13, "grad_norm": 1.327317476272583, "learning_rate": 0.00013669877928738848, "loss": 0.1642, "step": 21920 }, { "epoch": 1.5216994331268552e-13, "grad_norm": 1.0992791652679443, "learning_rate": 0.0001366858787912342, "loss": 0.1759, "step": 21930 }, { "epoch": 1.522393322517246e-13, "grad_norm": 1.1043214797973633, "learning_rate": 0.00013667297272663774, "loss": 0.1434, "step": 21940 }, { "epoch": 1.5230872119076366e-13, "grad_norm": 1.3253439664840698, "learning_rate": 0.00013666006109492551, "loss": 0.157, "step": 21950 }, { "epoch": 1.5237811012980274e-13, "grad_norm": 1.0388140678405762, "learning_rate": 0.00013664714389742427, "loss": 0.1609, "step": 21960 }, { "epoch": 1.524474990688418e-13, "grad_norm": 1.0539315938949585, "learning_rate": 0.0001366342211354615, "loss": 0.1446, "step": 21970 }, { "epoch": 1.5251688800788088e-13, "grad_norm": 1.3917452096939087, "learning_rate": 0.00013662129281036525, "loss": 0.1471, "step": 21980 }, { "epoch": 1.5258627694691995e-13, "grad_norm": 1.6238524913787842, "learning_rate": 0.00013660835892346406, "loss": 0.1553, "step": 21990 }, { "epoch": 1.5265566588595902e-13, "grad_norm": 1.4214861392974854, "learning_rate": 0.00013659541947608707, "loss": 0.1589, "step": 22000 }, { "epoch": 1.527250548249981e-13, "grad_norm": 1.6760830879211426, "learning_rate": 0.00013658247446956406, "loss": 0.1518, "step": 22010 }, { "epoch": 1.5279444376403717e-13, "grad_norm": 1.4445463418960571, "learning_rate": 0.00013656952390522531, "loss": 0.1562, "step": 22020 }, { "epoch": 1.5286383270307624e-13, "grad_norm": 1.5955499410629272, "learning_rate": 0.00013655656778440171, "loss": 0.1432, "step": 22030 }, { "epoch": 1.5293322164211531e-13, "grad_norm": 1.4832066297531128, "learning_rate": 0.00013654360610842466, "loss": 0.1564, "step": 22040 }, { "epoch": 1.5300261058115439e-13, "grad_norm": 1.5040690898895264, "learning_rate": 0.0001365306388786262, "loss": 0.1742, "step": 22050 }, { "epoch": 1.5307199952019346e-13, "grad_norm": 1.5077176094055176, "learning_rate": 0.00013651766609633896, "loss": 0.1508, "step": 22060 }, { "epoch": 1.5314138845923253e-13, "grad_norm": 1.491066813468933, "learning_rate": 0.000136504687762896, "loss": 0.1767, "step": 22070 }, { "epoch": 1.532107773982716e-13, "grad_norm": 1.6839885711669922, "learning_rate": 0.00013649170387963113, "loss": 0.1724, "step": 22080 }, { "epoch": 1.5328016633731067e-13, "grad_norm": 1.4737459421157837, "learning_rate": 0.00013647871444787861, "loss": 0.1453, "step": 22090 }, { "epoch": 1.5334955527634975e-13, "grad_norm": 1.284422755241394, "learning_rate": 0.0001364657194689733, "loss": 0.1402, "step": 22100 }, { "epoch": 1.5341894421538882e-13, "grad_norm": 1.678178071975708, "learning_rate": 0.00013645271894425064, "loss": 0.1503, "step": 22110 }, { "epoch": 1.534883331544279e-13, "grad_norm": 1.511561393737793, "learning_rate": 0.00013643971287504664, "loss": 0.1508, "step": 22120 }, { "epoch": 1.5355772209346696e-13, "grad_norm": 1.3623719215393066, "learning_rate": 0.00013642670126269788, "loss": 0.1416, "step": 22130 }, { "epoch": 1.5362711103250604e-13, "grad_norm": 1.1977053880691528, "learning_rate": 0.0001364136841085415, "loss": 0.1386, "step": 22140 }, { "epoch": 1.536964999715451e-13, "grad_norm": 1.9169758558273315, "learning_rate": 0.00013640066141391524, "loss": 0.1736, "step": 22150 }, { "epoch": 1.5376588891058418e-13, "grad_norm": 1.6782846450805664, "learning_rate": 0.00013638763318015733, "loss": 0.1419, "step": 22160 }, { "epoch": 1.5383527784962325e-13, "grad_norm": 1.043471336364746, "learning_rate": 0.00013637459940860669, "loss": 0.1503, "step": 22170 }, { "epoch": 1.5390466678866233e-13, "grad_norm": 1.235046148300171, "learning_rate": 0.0001363615601006027, "loss": 0.1466, "step": 22180 }, { "epoch": 1.539740557277014e-13, "grad_norm": 1.2453967332839966, "learning_rate": 0.00013634851525748534, "loss": 0.1515, "step": 22190 }, { "epoch": 1.5404344466674047e-13, "grad_norm": 1.2036802768707275, "learning_rate": 0.0001363354648805952, "loss": 0.1485, "step": 22200 }, { "epoch": 1.5411283360577954e-13, "grad_norm": 1.444446325302124, "learning_rate": 0.0001363224089712734, "loss": 0.1729, "step": 22210 }, { "epoch": 1.5418222254481861e-13, "grad_norm": 1.6178821325302124, "learning_rate": 0.00013630934753086163, "loss": 0.1555, "step": 22220 }, { "epoch": 1.5425161148385769e-13, "grad_norm": 1.603791356086731, "learning_rate": 0.0001362962805607022, "loss": 0.1634, "step": 22230 }, { "epoch": 1.5432100042289676e-13, "grad_norm": 1.227280616760254, "learning_rate": 0.00013628320806213787, "loss": 0.1735, "step": 22240 }, { "epoch": 1.5439038936193583e-13, "grad_norm": 1.4453811645507812, "learning_rate": 0.0001362701300365121, "loss": 0.1407, "step": 22250 }, { "epoch": 1.544597783009749e-13, "grad_norm": 1.4461793899536133, "learning_rate": 0.00013625704648516883, "loss": 0.1447, "step": 22260 }, { "epoch": 1.5452916724001398e-13, "grad_norm": 1.50182044506073, "learning_rate": 0.00013624395740945264, "loss": 0.1528, "step": 22270 }, { "epoch": 1.5459855617905305e-13, "grad_norm": 1.5967386960983276, "learning_rate": 0.00013623086281070856, "loss": 0.1475, "step": 22280 }, { "epoch": 1.5466794511809212e-13, "grad_norm": 1.458259105682373, "learning_rate": 0.00013621776269028238, "loss": 0.1284, "step": 22290 }, { "epoch": 1.547373340571312e-13, "grad_norm": 1.4482818841934204, "learning_rate": 0.00013620465704952022, "loss": 0.1499, "step": 22300 }, { "epoch": 1.5480672299617027e-13, "grad_norm": 1.6240257024765015, "learning_rate": 0.00013619154588976893, "loss": 0.1539, "step": 22310 }, { "epoch": 1.5487611193520934e-13, "grad_norm": 1.3385870456695557, "learning_rate": 0.0001361784292123759, "loss": 0.1436, "step": 22320 }, { "epoch": 1.549455008742484e-13, "grad_norm": 1.2960309982299805, "learning_rate": 0.0001361653070186891, "loss": 0.1504, "step": 22330 }, { "epoch": 1.5501488981328748e-13, "grad_norm": 1.3092037439346313, "learning_rate": 0.00013615217931005702, "loss": 0.1435, "step": 22340 }, { "epoch": 1.5508427875232655e-13, "grad_norm": 1.261133074760437, "learning_rate": 0.00013613904608782867, "loss": 0.1498, "step": 22350 }, { "epoch": 1.5515366769136563e-13, "grad_norm": 1.4168422222137451, "learning_rate": 0.0001361259073533538, "loss": 0.1601, "step": 22360 }, { "epoch": 1.552230566304047e-13, "grad_norm": 1.416181206703186, "learning_rate": 0.00013611276310798257, "loss": 0.1496, "step": 22370 }, { "epoch": 1.5529244556944377e-13, "grad_norm": 1.6786985397338867, "learning_rate": 0.0001360996133530657, "loss": 0.1696, "step": 22380 }, { "epoch": 1.5536183450848284e-13, "grad_norm": 1.6781089305877686, "learning_rate": 0.00013608645808995458, "loss": 0.1661, "step": 22390 }, { "epoch": 1.5543122344752192e-13, "grad_norm": 1.2893413305282593, "learning_rate": 0.00013607329732000118, "loss": 0.1616, "step": 22400 }, { "epoch": 1.55500612386561e-13, "grad_norm": 1.6412502527236938, "learning_rate": 0.0001360601310445579, "loss": 0.1409, "step": 22410 }, { "epoch": 1.5557000132560006e-13, "grad_norm": 1.0575237274169922, "learning_rate": 0.00013604695926497777, "loss": 0.1343, "step": 22420 }, { "epoch": 1.5563939026463913e-13, "grad_norm": 1.003075122833252, "learning_rate": 0.00013603378198261443, "loss": 0.1528, "step": 22430 }, { "epoch": 1.557087792036782e-13, "grad_norm": 1.2561532258987427, "learning_rate": 0.00013602059919882206, "loss": 0.1397, "step": 22440 }, { "epoch": 1.5577816814271728e-13, "grad_norm": 1.3332477807998657, "learning_rate": 0.00013600741091495538, "loss": 0.1408, "step": 22450 }, { "epoch": 1.5584755708175635e-13, "grad_norm": 1.5615330934524536, "learning_rate": 0.0001359942171323697, "loss": 0.1496, "step": 22460 }, { "epoch": 1.5591694602079542e-13, "grad_norm": 1.433660864830017, "learning_rate": 0.00013598101785242082, "loss": 0.1421, "step": 22470 }, { "epoch": 1.559863349598345e-13, "grad_norm": 1.7320507764816284, "learning_rate": 0.00013596781307646528, "loss": 0.1589, "step": 22480 }, { "epoch": 1.5605572389887357e-13, "grad_norm": 1.2612559795379639, "learning_rate": 0.00013595460280586, "loss": 0.1488, "step": 22490 }, { "epoch": 1.5612511283791264e-13, "grad_norm": 1.4658232927322388, "learning_rate": 0.00013594138704196257, "loss": 0.1395, "step": 22500 }, { "epoch": 1.561945017769517e-13, "grad_norm": 1.4662104845046997, "learning_rate": 0.0001359281657861311, "loss": 0.1447, "step": 22510 }, { "epoch": 1.5626389071599078e-13, "grad_norm": 1.4528124332427979, "learning_rate": 0.00013591493903972432, "loss": 0.1625, "step": 22520 }, { "epoch": 1.5633327965502986e-13, "grad_norm": 1.3954815864562988, "learning_rate": 0.00013590170680410142, "loss": 0.1417, "step": 22530 }, { "epoch": 1.5640266859406893e-13, "grad_norm": 1.3933961391448975, "learning_rate": 0.00013588846908062227, "loss": 0.1691, "step": 22540 }, { "epoch": 1.56472057533108e-13, "grad_norm": 1.3972030878067017, "learning_rate": 0.00013587522587064725, "loss": 0.1743, "step": 22550 }, { "epoch": 1.5654144647214707e-13, "grad_norm": 1.0432604551315308, "learning_rate": 0.00013586197717553727, "loss": 0.1345, "step": 22560 }, { "epoch": 1.5661083541118614e-13, "grad_norm": 1.2681171894073486, "learning_rate": 0.00013584872299665386, "loss": 0.136, "step": 22570 }, { "epoch": 1.5668022435022522e-13, "grad_norm": 1.6951791048049927, "learning_rate": 0.0001358354633353591, "loss": 0.1532, "step": 22580 }, { "epoch": 1.567496132892643e-13, "grad_norm": 1.2706894874572754, "learning_rate": 0.0001358221981930156, "loss": 0.136, "step": 22590 }, { "epoch": 1.5681900222830336e-13, "grad_norm": 1.1124855279922485, "learning_rate": 0.00013580892757098658, "loss": 0.1284, "step": 22600 }, { "epoch": 1.5688839116734243e-13, "grad_norm": 1.4331785440444946, "learning_rate": 0.00013579565147063582, "loss": 0.1335, "step": 22610 }, { "epoch": 1.569577801063815e-13, "grad_norm": 1.1028902530670166, "learning_rate": 0.00013578236989332762, "loss": 0.1555, "step": 22620 }, { "epoch": 1.5702716904542058e-13, "grad_norm": 1.2478399276733398, "learning_rate": 0.0001357690828404269, "loss": 0.1307, "step": 22630 }, { "epoch": 1.5709655798445965e-13, "grad_norm": 1.2164201736450195, "learning_rate": 0.00013575579031329903, "loss": 0.1473, "step": 22640 }, { "epoch": 1.5716594692349872e-13, "grad_norm": 1.1170300245285034, "learning_rate": 0.00013574249231331013, "loss": 0.1486, "step": 22650 }, { "epoch": 1.572353358625378e-13, "grad_norm": 1.4066152572631836, "learning_rate": 0.0001357291888418267, "loss": 0.1466, "step": 22660 }, { "epoch": 1.5730472480157687e-13, "grad_norm": 1.7252269983291626, "learning_rate": 0.00013571587990021593, "loss": 0.1471, "step": 22670 }, { "epoch": 1.5737411374061594e-13, "grad_norm": 1.180123209953308, "learning_rate": 0.00013570256548984548, "loss": 0.1541, "step": 22680 }, { "epoch": 1.57443502679655e-13, "grad_norm": 1.2731614112854004, "learning_rate": 0.00013568924561208366, "loss": 0.1461, "step": 22690 }, { "epoch": 1.5751289161869408e-13, "grad_norm": 1.4373449087142944, "learning_rate": 0.00013567592026829926, "loss": 0.1553, "step": 22700 }, { "epoch": 1.5758228055773316e-13, "grad_norm": 1.353692889213562, "learning_rate": 0.00013566258945986164, "loss": 0.1674, "step": 22710 }, { "epoch": 1.5765166949677223e-13, "grad_norm": 1.0956662893295288, "learning_rate": 0.0001356492531881408, "loss": 0.1316, "step": 22720 }, { "epoch": 1.577210584358113e-13, "grad_norm": 1.1886465549468994, "learning_rate": 0.00013563591145450722, "loss": 0.136, "step": 22730 }, { "epoch": 1.5779044737485037e-13, "grad_norm": 1.404825210571289, "learning_rate": 0.000135622564260332, "loss": 0.1539, "step": 22740 }, { "epoch": 1.5785983631388945e-13, "grad_norm": 1.0567299127578735, "learning_rate": 0.0001356092116069867, "loss": 0.1292, "step": 22750 }, { "epoch": 1.5792922525292852e-13, "grad_norm": 1.4402486085891724, "learning_rate": 0.00013559585349584363, "loss": 0.1268, "step": 22760 }, { "epoch": 1.579986141919676e-13, "grad_norm": 1.251348614692688, "learning_rate": 0.00013558248992827544, "loss": 0.1364, "step": 22770 }, { "epoch": 1.5806800313100666e-13, "grad_norm": 1.1213701963424683, "learning_rate": 0.00013556912090565549, "loss": 0.138, "step": 22780 }, { "epoch": 1.5813739207004573e-13, "grad_norm": 1.2460981607437134, "learning_rate": 0.00013555574642935763, "loss": 0.1299, "step": 22790 }, { "epoch": 1.582067810090848e-13, "grad_norm": 1.7169896364212036, "learning_rate": 0.00013554236650075636, "loss": 0.1579, "step": 22800 }, { "epoch": 1.5827616994812388e-13, "grad_norm": 1.2938991785049438, "learning_rate": 0.00013552898112122657, "loss": 0.1339, "step": 22810 }, { "epoch": 1.5834555888716295e-13, "grad_norm": 1.3660218715667725, "learning_rate": 0.00013551559029214392, "loss": 0.1427, "step": 22820 }, { "epoch": 1.5841494782620202e-13, "grad_norm": 1.4801160097122192, "learning_rate": 0.00013550219401488447, "loss": 0.1572, "step": 22830 }, { "epoch": 1.584843367652411e-13, "grad_norm": 1.5133956670761108, "learning_rate": 0.0001354887922908249, "loss": 0.1385, "step": 22840 }, { "epoch": 1.5855372570428017e-13, "grad_norm": 1.6864056587219238, "learning_rate": 0.00013547538512134248, "loss": 0.1594, "step": 22850 }, { "epoch": 1.5862311464331924e-13, "grad_norm": 1.8590505123138428, "learning_rate": 0.00013546197250781494, "loss": 0.1548, "step": 22860 }, { "epoch": 1.5869250358235831e-13, "grad_norm": 1.3486499786376953, "learning_rate": 0.00013544855445162072, "loss": 0.147, "step": 22870 }, { "epoch": 1.5876189252139739e-13, "grad_norm": 1.2505762577056885, "learning_rate": 0.00013543513095413864, "loss": 0.1317, "step": 22880 }, { "epoch": 1.5883128146043646e-13, "grad_norm": 1.3995726108551025, "learning_rate": 0.00013542170201674824, "loss": 0.1242, "step": 22890 }, { "epoch": 1.5890067039947553e-13, "grad_norm": 1.463741421699524, "learning_rate": 0.0001354082676408295, "loss": 0.1468, "step": 22900 }, { "epoch": 1.589700593385146e-13, "grad_norm": 1.2391046285629272, "learning_rate": 0.00013539482782776307, "loss": 0.129, "step": 22910 }, { "epoch": 1.5903944827755367e-13, "grad_norm": 1.0168012380599976, "learning_rate": 0.00013538138257893008, "loss": 0.131, "step": 22920 }, { "epoch": 1.5910883721659275e-13, "grad_norm": 1.3050029277801514, "learning_rate": 0.00013536793189571224, "loss": 0.1425, "step": 22930 }, { "epoch": 1.5917822615563182e-13, "grad_norm": 1.2376725673675537, "learning_rate": 0.00013535447577949178, "loss": 0.1297, "step": 22940 }, { "epoch": 1.592476150946709e-13, "grad_norm": 1.3564016819000244, "learning_rate": 0.00013534101423165161, "loss": 0.1465, "step": 22950 }, { "epoch": 1.5931700403370996e-13, "grad_norm": 1.3011674880981445, "learning_rate": 0.00013532754725357498, "loss": 0.1392, "step": 22960 }, { "epoch": 1.5938639297274904e-13, "grad_norm": 1.5294636487960815, "learning_rate": 0.00013531407484664596, "loss": 0.141, "step": 22970 }, { "epoch": 1.594557819117881e-13, "grad_norm": 1.381751537322998, "learning_rate": 0.00013530059701224897, "loss": 0.1363, "step": 22980 }, { "epoch": 1.5952517085082718e-13, "grad_norm": 1.5365196466445923, "learning_rate": 0.00013528711375176912, "loss": 0.1522, "step": 22990 }, { "epoch": 1.5959455978986625e-13, "grad_norm": 1.591871738433838, "learning_rate": 0.000135273625066592, "loss": 0.1364, "step": 23000 }, { "epoch": 1.5966394872890532e-13, "grad_norm": 1.188309669494629, "learning_rate": 0.00013526013095810377, "loss": 0.1488, "step": 23010 }, { "epoch": 1.597333376679444e-13, "grad_norm": 1.3183008432388306, "learning_rate": 0.0001352466314276912, "loss": 0.1692, "step": 23020 }, { "epoch": 1.5980272660698347e-13, "grad_norm": 1.5121012926101685, "learning_rate": 0.00013523312647674153, "loss": 0.1351, "step": 23030 }, { "epoch": 1.5987211554602254e-13, "grad_norm": 1.52003014087677, "learning_rate": 0.0001352196161066426, "loss": 0.1269, "step": 23040 }, { "epoch": 1.5994150448506161e-13, "grad_norm": 1.3346444368362427, "learning_rate": 0.00013520610031878287, "loss": 0.1397, "step": 23050 }, { "epoch": 1.6001089342410069e-13, "grad_norm": 1.411750078201294, "learning_rate": 0.00013519257911455122, "loss": 0.1298, "step": 23060 }, { "epoch": 1.6008028236313976e-13, "grad_norm": 1.211505651473999, "learning_rate": 0.00013517905249533723, "loss": 0.1167, "step": 23070 }, { "epoch": 1.6014967130217883e-13, "grad_norm": 1.4320199489593506, "learning_rate": 0.00013516552046253095, "loss": 0.1316, "step": 23080 }, { "epoch": 1.602190602412179e-13, "grad_norm": 1.1026551723480225, "learning_rate": 0.000135151983017523, "loss": 0.1367, "step": 23090 }, { "epoch": 1.6028844918025698e-13, "grad_norm": 1.29591703414917, "learning_rate": 0.00013513844016170456, "loss": 0.1265, "step": 23100 }, { "epoch": 1.6035783811929605e-13, "grad_norm": 1.5411635637283325, "learning_rate": 0.0001351248918964674, "loss": 0.1394, "step": 23110 }, { "epoch": 1.6042722705833512e-13, "grad_norm": 1.3687690496444702, "learning_rate": 0.00013511133822320377, "loss": 0.1365, "step": 23120 }, { "epoch": 1.604966159973742e-13, "grad_norm": 1.3305721282958984, "learning_rate": 0.00013509777914330655, "loss": 0.1428, "step": 23130 }, { "epoch": 1.6056600493641326e-13, "grad_norm": 1.2815340757369995, "learning_rate": 0.00013508421465816914, "loss": 0.1414, "step": 23140 }, { "epoch": 1.6063539387545234e-13, "grad_norm": 1.2714186906814575, "learning_rate": 0.00013507064476918546, "loss": 0.1362, "step": 23150 }, { "epoch": 1.607047828144914e-13, "grad_norm": 1.4725369215011597, "learning_rate": 0.0001350570694777501, "loss": 0.1452, "step": 23160 }, { "epoch": 1.6077417175353048e-13, "grad_norm": 1.4258899688720703, "learning_rate": 0.00013504348878525812, "loss": 0.1441, "step": 23170 }, { "epoch": 1.6084356069256955e-13, "grad_norm": 1.1335818767547607, "learning_rate": 0.0001350299026931051, "loss": 0.1632, "step": 23180 }, { "epoch": 1.6091294963160863e-13, "grad_norm": 1.0027052164077759, "learning_rate": 0.00013501631120268725, "loss": 0.1265, "step": 23190 }, { "epoch": 1.609823385706477e-13, "grad_norm": 1.2636704444885254, "learning_rate": 0.0001350027143154013, "loss": 0.1361, "step": 23200 }, { "epoch": 1.6105172750968677e-13, "grad_norm": 0.9794815182685852, "learning_rate": 0.00013498911203264457, "loss": 0.1405, "step": 23210 }, { "epoch": 1.6112111644872584e-13, "grad_norm": 1.0320398807525635, "learning_rate": 0.00013497550435581488, "loss": 0.1212, "step": 23220 }, { "epoch": 1.6119050538776492e-13, "grad_norm": 1.1512742042541504, "learning_rate": 0.00013496189128631064, "loss": 0.123, "step": 23230 }, { "epoch": 1.61259894326804e-13, "grad_norm": 1.362483024597168, "learning_rate": 0.00013494827282553075, "loss": 0.1281, "step": 23240 }, { "epoch": 1.6132928326584306e-13, "grad_norm": 1.1407825946807861, "learning_rate": 0.00013493464897487484, "loss": 0.1307, "step": 23250 }, { "epoch": 1.6139867220488213e-13, "grad_norm": 1.5579259395599365, "learning_rate": 0.00013492101973574283, "loss": 0.1288, "step": 23260 }, { "epoch": 1.614680611439212e-13, "grad_norm": 1.3906949758529663, "learning_rate": 0.00013490738510953544, "loss": 0.1447, "step": 23270 }, { "epoch": 1.6153745008296028e-13, "grad_norm": 1.4121527671813965, "learning_rate": 0.00013489374509765382, "loss": 0.1286, "step": 23280 }, { "epoch": 1.6160683902199935e-13, "grad_norm": 1.2066465616226196, "learning_rate": 0.00013488009970149967, "loss": 0.1369, "step": 23290 }, { "epoch": 1.6167622796103842e-13, "grad_norm": 1.2686911821365356, "learning_rate": 0.00013486644892247526, "loss": 0.1475, "step": 23300 }, { "epoch": 1.617456169000775e-13, "grad_norm": 1.3946868181228638, "learning_rate": 0.00013485279276198348, "loss": 0.1308, "step": 23310 }, { "epoch": 1.6181500583911657e-13, "grad_norm": 1.4009974002838135, "learning_rate": 0.00013483913122142762, "loss": 0.1553, "step": 23320 }, { "epoch": 1.6188439477815564e-13, "grad_norm": 1.795723557472229, "learning_rate": 0.00013482546430221168, "loss": 0.157, "step": 23330 }, { "epoch": 1.619537837171947e-13, "grad_norm": 1.3853418827056885, "learning_rate": 0.00013481179200574016, "loss": 0.1353, "step": 23340 }, { "epoch": 1.6202317265623378e-13, "grad_norm": 1.1131670475006104, "learning_rate": 0.00013479811433341805, "loss": 0.1231, "step": 23350 }, { "epoch": 1.6209256159527285e-13, "grad_norm": 1.6093813180923462, "learning_rate": 0.000134784431286651, "loss": 0.1345, "step": 23360 }, { "epoch": 1.6216195053431193e-13, "grad_norm": 1.2847806215286255, "learning_rate": 0.0001347707428668451, "loss": 0.1321, "step": 23370 }, { "epoch": 1.62231339473351e-13, "grad_norm": 1.3247967958450317, "learning_rate": 0.00013475704907540707, "loss": 0.1216, "step": 23380 }, { "epoch": 1.6230072841239007e-13, "grad_norm": 1.106207251548767, "learning_rate": 0.00013474334991374416, "loss": 0.1189, "step": 23390 }, { "epoch": 1.6237011735142914e-13, "grad_norm": 1.7725318670272827, "learning_rate": 0.00013472964538326418, "loss": 0.1487, "step": 23400 }, { "epoch": 1.6243950629046822e-13, "grad_norm": 1.3403557538986206, "learning_rate": 0.00013471593548537548, "loss": 0.1165, "step": 23410 }, { "epoch": 1.625088952295073e-13, "grad_norm": 0.9661650061607361, "learning_rate": 0.00013470222022148701, "loss": 0.1285, "step": 23420 }, { "epoch": 1.6257828416854636e-13, "grad_norm": 1.182816505432129, "learning_rate": 0.0001346884995930081, "loss": 0.133, "step": 23430 }, { "epoch": 1.6264767310758543e-13, "grad_norm": 1.2001898288726807, "learning_rate": 0.0001346747736013489, "loss": 0.1338, "step": 23440 }, { "epoch": 1.627170620466245e-13, "grad_norm": 1.2473782300949097, "learning_rate": 0.0001346610422479199, "loss": 0.1324, "step": 23450 }, { "epoch": 1.6278645098566358e-13, "grad_norm": 1.3044713735580444, "learning_rate": 0.0001346473055341322, "loss": 0.1457, "step": 23460 }, { "epoch": 1.6285583992470265e-13, "grad_norm": 1.5600261688232422, "learning_rate": 0.0001346335634613975, "loss": 0.1346, "step": 23470 }, { "epoch": 1.6292522886374172e-13, "grad_norm": 1.467065691947937, "learning_rate": 0.00013461981603112797, "loss": 0.1434, "step": 23480 }, { "epoch": 1.629946178027808e-13, "grad_norm": 1.2212945222854614, "learning_rate": 0.0001346060632447364, "loss": 0.1597, "step": 23490 }, { "epoch": 1.6306400674181987e-13, "grad_norm": 1.3408942222595215, "learning_rate": 0.0001345923051036361, "loss": 0.1285, "step": 23500 }, { "epoch": 1.6313339568085894e-13, "grad_norm": 1.3963079452514648, "learning_rate": 0.00013457854160924096, "loss": 0.1278, "step": 23510 }, { "epoch": 1.63202784619898e-13, "grad_norm": 1.5130858421325684, "learning_rate": 0.00013456477276296532, "loss": 0.1402, "step": 23520 }, { "epoch": 1.6327217355893708e-13, "grad_norm": 1.3781764507293701, "learning_rate": 0.0001345509985662242, "loss": 0.1293, "step": 23530 }, { "epoch": 1.6334156249797616e-13, "grad_norm": 1.3927228450775146, "learning_rate": 0.00013453721902043309, "loss": 0.1136, "step": 23540 }, { "epoch": 1.6341095143701523e-13, "grad_norm": 1.3583952188491821, "learning_rate": 0.00013452343412700808, "loss": 0.1311, "step": 23550 }, { "epoch": 1.634803403760543e-13, "grad_norm": 1.4774045944213867, "learning_rate": 0.00013450964388736576, "loss": 0.135, "step": 23560 }, { "epoch": 1.6354972931509337e-13, "grad_norm": 1.1321336030960083, "learning_rate": 0.00013449584830292331, "loss": 0.1149, "step": 23570 }, { "epoch": 1.6361911825413245e-13, "grad_norm": 1.2175511121749878, "learning_rate": 0.0001344820473750984, "loss": 0.1277, "step": 23580 }, { "epoch": 1.6368850719317152e-13, "grad_norm": 1.3673275709152222, "learning_rate": 0.00013446824110530933, "loss": 0.1275, "step": 23590 }, { "epoch": 1.637578961322106e-13, "grad_norm": 1.31201171875, "learning_rate": 0.0001344544294949749, "loss": 0.1325, "step": 23600 }, { "epoch": 1.6382728507124966e-13, "grad_norm": 1.4685789346694946, "learning_rate": 0.00013444061254551444, "loss": 0.1424, "step": 23610 }, { "epoch": 1.6389667401028873e-13, "grad_norm": 1.4418318271636963, "learning_rate": 0.0001344267902583479, "loss": 0.1274, "step": 23620 }, { "epoch": 1.639660629493278e-13, "grad_norm": 1.5597106218338013, "learning_rate": 0.00013441296263489572, "loss": 0.1461, "step": 23630 }, { "epoch": 1.6403545188836688e-13, "grad_norm": 1.6919686794281006, "learning_rate": 0.00013439912967657888, "loss": 0.1466, "step": 23640 }, { "epoch": 1.6410484082740595e-13, "grad_norm": 1.2710684537887573, "learning_rate": 0.000134385291384819, "loss": 0.1476, "step": 23650 }, { "epoch": 1.6417422976644502e-13, "grad_norm": 1.6835209131240845, "learning_rate": 0.0001343714477610381, "loss": 0.128, "step": 23660 }, { "epoch": 1.642436187054841e-13, "grad_norm": 1.1567965745925903, "learning_rate": 0.00013435759880665884, "loss": 0.1183, "step": 23670 }, { "epoch": 1.6431300764452317e-13, "grad_norm": 1.031693458557129, "learning_rate": 0.00013434374452310445, "loss": 0.1422, "step": 23680 }, { "epoch": 1.6438239658356224e-13, "grad_norm": 1.2803598642349243, "learning_rate": 0.00013432988491179866, "loss": 0.1262, "step": 23690 }, { "epoch": 1.644517855226013e-13, "grad_norm": 1.2936547994613647, "learning_rate": 0.00013431601997416573, "loss": 0.1276, "step": 23700 }, { "epoch": 1.6452117446164038e-13, "grad_norm": 1.4935615062713623, "learning_rate": 0.00013430214971163053, "loss": 0.1309, "step": 23710 }, { "epoch": 1.6459056340067946e-13, "grad_norm": 1.383334755897522, "learning_rate": 0.00013428827412561848, "loss": 0.1301, "step": 23720 }, { "epoch": 1.6465995233971853e-13, "grad_norm": 1.5159685611724854, "learning_rate": 0.00013427439321755548, "loss": 0.1291, "step": 23730 }, { "epoch": 1.647293412787576e-13, "grad_norm": 1.2353132963180542, "learning_rate": 0.00013426050698886797, "loss": 0.127, "step": 23740 }, { "epoch": 1.6479873021779667e-13, "grad_norm": 1.439557433128357, "learning_rate": 0.000134246615440983, "loss": 0.1218, "step": 23750 }, { "epoch": 1.6486811915683575e-13, "grad_norm": 1.4241914749145508, "learning_rate": 0.00013423271857532817, "loss": 0.1287, "step": 23760 }, { "epoch": 1.6493750809587482e-13, "grad_norm": 1.230776071548462, "learning_rate": 0.0001342188163933316, "loss": 0.1398, "step": 23770 }, { "epoch": 1.650068970349139e-13, "grad_norm": 1.2513962984085083, "learning_rate": 0.00013420490889642187, "loss": 0.1243, "step": 23780 }, { "epoch": 1.6507628597395296e-13, "grad_norm": 1.2668960094451904, "learning_rate": 0.00013419099608602832, "loss": 0.15, "step": 23790 }, { "epoch": 1.6514567491299204e-13, "grad_norm": 1.6003705263137817, "learning_rate": 0.00013417707796358063, "loss": 0.1614, "step": 23800 }, { "epoch": 1.652150638520311e-13, "grad_norm": 1.1873584985733032, "learning_rate": 0.0001341631545305091, "loss": 0.1278, "step": 23810 }, { "epoch": 1.6528445279107018e-13, "grad_norm": 1.2999825477600098, "learning_rate": 0.00013414922578824462, "loss": 0.1244, "step": 23820 }, { "epoch": 1.6535384173010925e-13, "grad_norm": 1.7319589853286743, "learning_rate": 0.00013413529173821856, "loss": 0.1347, "step": 23830 }, { "epoch": 1.6542323066914832e-13, "grad_norm": 1.2848762273788452, "learning_rate": 0.00013412135238186288, "loss": 0.127, "step": 23840 }, { "epoch": 1.654926196081874e-13, "grad_norm": 1.2438476085662842, "learning_rate": 0.00013410740772061002, "loss": 0.12, "step": 23850 }, { "epoch": 1.6556200854722647e-13, "grad_norm": 1.4114346504211426, "learning_rate": 0.00013409345775589306, "loss": 0.1218, "step": 23860 }, { "epoch": 1.6563139748626554e-13, "grad_norm": 1.2032718658447266, "learning_rate": 0.00013407950248914556, "loss": 0.141, "step": 23870 }, { "epoch": 1.6570078642530461e-13, "grad_norm": 1.3264939785003662, "learning_rate": 0.00013406554192180162, "loss": 0.1222, "step": 23880 }, { "epoch": 1.6577017536434369e-13, "grad_norm": 1.1345691680908203, "learning_rate": 0.00013405157605529593, "loss": 0.1221, "step": 23890 }, { "epoch": 1.6583956430338276e-13, "grad_norm": 1.131453037261963, "learning_rate": 0.00013403760489106372, "loss": 0.1281, "step": 23900 }, { "epoch": 1.6590895324242183e-13, "grad_norm": 1.273559808731079, "learning_rate": 0.00013402362843054067, "loss": 0.1288, "step": 23910 }, { "epoch": 1.659783421814609e-13, "grad_norm": 1.486638069152832, "learning_rate": 0.00013400964667516314, "loss": 0.1295, "step": 23920 }, { "epoch": 1.6604773112049998e-13, "grad_norm": 1.0012152194976807, "learning_rate": 0.000133995659626368, "loss": 0.1278, "step": 23930 }, { "epoch": 1.6611712005953905e-13, "grad_norm": 1.16046941280365, "learning_rate": 0.00013398166728559256, "loss": 0.1284, "step": 23940 }, { "epoch": 1.6618650899857812e-13, "grad_norm": 1.4296294450759888, "learning_rate": 0.00013396766965427477, "loss": 0.1386, "step": 23950 }, { "epoch": 1.662558979376172e-13, "grad_norm": 1.7087210416793823, "learning_rate": 0.00013395366673385313, "loss": 0.1703, "step": 23960 }, { "epoch": 1.6632528687665626e-13, "grad_norm": 1.3159468173980713, "learning_rate": 0.00013393965852576668, "loss": 0.1299, "step": 23970 }, { "epoch": 1.6639467581569534e-13, "grad_norm": 1.2051020860671997, "learning_rate": 0.00013392564503145494, "loss": 0.1261, "step": 23980 }, { "epoch": 1.664640647547344e-13, "grad_norm": 1.3180423974990845, "learning_rate": 0.00013391162625235804, "loss": 0.1358, "step": 23990 }, { "epoch": 1.6653345369377348e-13, "grad_norm": 1.0055710077285767, "learning_rate": 0.00013389760218991657, "loss": 0.1164, "step": 24000 }, { "epoch": 1.6660284263281255e-13, "grad_norm": 1.4872016906738281, "learning_rate": 0.0001338835728455718, "loss": 0.1152, "step": 24010 }, { "epoch": 1.6667223157185163e-13, "grad_norm": 1.1723037958145142, "learning_rate": 0.00013386953822076538, "loss": 0.1216, "step": 24020 }, { "epoch": 1.667416205108907e-13, "grad_norm": 1.2664484977722168, "learning_rate": 0.00013385549831693969, "loss": 0.1305, "step": 24030 }, { "epoch": 1.6681100944992977e-13, "grad_norm": 1.288387656211853, "learning_rate": 0.00013384145313553744, "loss": 0.1233, "step": 24040 }, { "epoch": 1.6688039838896884e-13, "grad_norm": 1.8285934925079346, "learning_rate": 0.00013382740267800207, "loss": 0.1385, "step": 24050 }, { "epoch": 1.6694978732800791e-13, "grad_norm": 1.1857913732528687, "learning_rate": 0.00013381334694577746, "loss": 0.1192, "step": 24060 }, { "epoch": 1.6701917626704699e-13, "grad_norm": 1.3530710935592651, "learning_rate": 0.00013379928594030804, "loss": 0.1273, "step": 24070 }, { "epoch": 1.6708856520608606e-13, "grad_norm": 1.4986189603805542, "learning_rate": 0.00013378521966303882, "loss": 0.1367, "step": 24080 }, { "epoch": 1.6715795414512513e-13, "grad_norm": 1.5118799209594727, "learning_rate": 0.0001337711481154153, "loss": 0.1186, "step": 24090 }, { "epoch": 1.672273430841642e-13, "grad_norm": 1.7458544969558716, "learning_rate": 0.0001337570712988836, "loss": 0.143, "step": 24100 }, { "epoch": 1.6729673202320328e-13, "grad_norm": 2.1575734615325928, "learning_rate": 0.00013374298921489027, "loss": 0.1398, "step": 24110 }, { "epoch": 1.6736612096224235e-13, "grad_norm": 1.5112217664718628, "learning_rate": 0.0001337289018648825, "loss": 0.1486, "step": 24120 }, { "epoch": 1.6743550990128142e-13, "grad_norm": 1.3768409490585327, "learning_rate": 0.000133714809250308, "loss": 0.1314, "step": 24130 }, { "epoch": 1.675048988403205e-13, "grad_norm": 1.3622865676879883, "learning_rate": 0.00013370071137261492, "loss": 0.1167, "step": 24140 }, { "epoch": 1.6757428777935957e-13, "grad_norm": 1.4002701044082642, "learning_rate": 0.00013368660823325211, "loss": 0.13, "step": 24150 }, { "epoch": 1.6764367671839864e-13, "grad_norm": 1.199225902557373, "learning_rate": 0.00013367249983366894, "loss": 0.1145, "step": 24160 }, { "epoch": 1.677130656574377e-13, "grad_norm": 0.9948814511299133, "learning_rate": 0.00013365838617531516, "loss": 0.1213, "step": 24170 }, { "epoch": 1.6778245459647678e-13, "grad_norm": 1.2781856060028076, "learning_rate": 0.0001336442672596412, "loss": 0.1271, "step": 24180 }, { "epoch": 1.6785184353551585e-13, "grad_norm": 1.3634743690490723, "learning_rate": 0.00013363014308809802, "loss": 0.124, "step": 24190 }, { "epoch": 1.6792123247455493e-13, "grad_norm": 1.323927640914917, "learning_rate": 0.0001336160136621371, "loss": 0.1349, "step": 24200 }, { "epoch": 1.67990621413594e-13, "grad_norm": 1.2286056280136108, "learning_rate": 0.00013360187898321045, "loss": 0.1282, "step": 24210 }, { "epoch": 1.6806001035263307e-13, "grad_norm": 1.3318709135055542, "learning_rate": 0.00013358773905277062, "loss": 0.1251, "step": 24220 }, { "epoch": 1.6812939929167214e-13, "grad_norm": 1.4376260042190552, "learning_rate": 0.0001335735938722707, "loss": 0.1217, "step": 24230 }, { "epoch": 1.6819878823071122e-13, "grad_norm": 1.4379669427871704, "learning_rate": 0.00013355944344316435, "loss": 0.1284, "step": 24240 }, { "epoch": 1.682681771697503e-13, "grad_norm": 1.5061495304107666, "learning_rate": 0.00013354528776690573, "loss": 0.1178, "step": 24250 }, { "epoch": 1.6833756610878936e-13, "grad_norm": 1.0785908699035645, "learning_rate": 0.00013353112684494955, "loss": 0.1316, "step": 24260 }, { "epoch": 1.6840695504782843e-13, "grad_norm": 1.4634933471679688, "learning_rate": 0.00013351696067875107, "loss": 0.1542, "step": 24270 }, { "epoch": 1.684763439868675e-13, "grad_norm": 1.403398871421814, "learning_rate": 0.00013350278926976613, "loss": 0.13, "step": 24280 }, { "epoch": 1.6854573292590658e-13, "grad_norm": 1.4213316440582275, "learning_rate": 0.00013348861261945098, "loss": 0.1275, "step": 24290 }, { "epoch": 1.6861512186494565e-13, "grad_norm": 1.3683372735977173, "learning_rate": 0.00013347443072926258, "loss": 0.1328, "step": 24300 }, { "epoch": 1.6868451080398472e-13, "grad_norm": 1.4396002292633057, "learning_rate": 0.00013346024360065827, "loss": 0.1217, "step": 24310 }, { "epoch": 1.687538997430238e-13, "grad_norm": 1.2567838430404663, "learning_rate": 0.00013344605123509606, "loss": 0.1063, "step": 24320 }, { "epoch": 1.6882328868206287e-13, "grad_norm": 1.5291980504989624, "learning_rate": 0.00013343185363403435, "loss": 0.1214, "step": 24330 }, { "epoch": 1.6889267762110194e-13, "grad_norm": 1.1176443099975586, "learning_rate": 0.00013341765079893223, "loss": 0.124, "step": 24340 }, { "epoch": 1.68962066560141e-13, "grad_norm": 1.342373251914978, "learning_rate": 0.00013340344273124927, "loss": 0.1224, "step": 24350 }, { "epoch": 1.6903145549918008e-13, "grad_norm": 1.3069753646850586, "learning_rate": 0.00013338922943244551, "loss": 0.1266, "step": 24360 }, { "epoch": 1.6910084443821916e-13, "grad_norm": 1.2493044137954712, "learning_rate": 0.00013337501090398166, "loss": 0.1237, "step": 24370 }, { "epoch": 1.6917023337725823e-13, "grad_norm": 1.2322450876235962, "learning_rate": 0.00013336078714731883, "loss": 0.124, "step": 24380 }, { "epoch": 1.692396223162973e-13, "grad_norm": 1.3741661310195923, "learning_rate": 0.0001333465581639188, "loss": 0.1244, "step": 24390 }, { "epoch": 1.6930901125533637e-13, "grad_norm": 1.0810022354125977, "learning_rate": 0.00013333232395524374, "loss": 0.1141, "step": 24400 }, { "epoch": 1.6937840019437544e-13, "grad_norm": 1.3576653003692627, "learning_rate": 0.00013331808452275647, "loss": 0.1247, "step": 24410 }, { "epoch": 1.6944778913341452e-13, "grad_norm": 1.2162140607833862, "learning_rate": 0.00013330383986792034, "loss": 0.1278, "step": 24420 }, { "epoch": 1.695171780724536e-13, "grad_norm": 1.1823474168777466, "learning_rate": 0.0001332895899921992, "loss": 0.1474, "step": 24430 }, { "epoch": 1.6958656701149266e-13, "grad_norm": 0.9772384166717529, "learning_rate": 0.00013327533489705743, "loss": 0.1168, "step": 24440 }, { "epoch": 1.6965595595053173e-13, "grad_norm": 1.3551186323165894, "learning_rate": 0.00013326107458395996, "loss": 0.1307, "step": 24450 }, { "epoch": 1.697253448895708e-13, "grad_norm": 1.0125473737716675, "learning_rate": 0.00013324680905437228, "loss": 0.1376, "step": 24460 }, { "epoch": 1.6979473382860988e-13, "grad_norm": 1.0197365283966064, "learning_rate": 0.0001332325383097604, "loss": 0.1153, "step": 24470 }, { "epoch": 1.6986412276764895e-13, "grad_norm": 1.2458797693252563, "learning_rate": 0.00013321826235159083, "loss": 0.1139, "step": 24480 }, { "epoch": 1.6993351170668802e-13, "grad_norm": 1.3667491674423218, "learning_rate": 0.00013320398118133066, "loss": 0.117, "step": 24490 }, { "epoch": 1.700029006457271e-13, "grad_norm": 1.1689181327819824, "learning_rate": 0.00013318969480044748, "loss": 0.1171, "step": 24500 }, { "epoch": 1.7007228958476617e-13, "grad_norm": 1.631478190422058, "learning_rate": 0.0001331754032104095, "loss": 0.122, "step": 24510 }, { "epoch": 1.7014167852380524e-13, "grad_norm": 1.4131907224655151, "learning_rate": 0.00013316110641268535, "loss": 0.1301, "step": 24520 }, { "epoch": 1.702110674628443e-13, "grad_norm": 1.3642234802246094, "learning_rate": 0.00013314680440874426, "loss": 0.1136, "step": 24530 }, { "epoch": 1.7028045640188338e-13, "grad_norm": 1.2028415203094482, "learning_rate": 0.000133132497200056, "loss": 0.1195, "step": 24540 }, { "epoch": 1.7034984534092246e-13, "grad_norm": 1.1401283740997314, "learning_rate": 0.00013311818478809086, "loss": 0.1297, "step": 24550 }, { "epoch": 1.7041923427996153e-13, "grad_norm": 1.1822882890701294, "learning_rate": 0.00013310386717431963, "loss": 0.1068, "step": 24560 }, { "epoch": 1.704886232190006e-13, "grad_norm": 1.265884280204773, "learning_rate": 0.0001330895443602137, "loss": 0.1344, "step": 24570 }, { "epoch": 1.7055801215803967e-13, "grad_norm": 1.4806301593780518, "learning_rate": 0.0001330752163472449, "loss": 0.132, "step": 24580 }, { "epoch": 1.7062740109707875e-13, "grad_norm": 1.3610044717788696, "learning_rate": 0.00013306088313688575, "loss": 0.1225, "step": 24590 }, { "epoch": 1.7069679003611782e-13, "grad_norm": 1.2777080535888672, "learning_rate": 0.00013304654473060914, "loss": 0.1148, "step": 24600 }, { "epoch": 1.707661789751569e-13, "grad_norm": 1.9627500772476196, "learning_rate": 0.0001330322011298886, "loss": 0.1248, "step": 24610 }, { "epoch": 1.7083556791419596e-13, "grad_norm": 1.335463523864746, "learning_rate": 0.00013301785233619815, "loss": 0.1272, "step": 24620 }, { "epoch": 1.7090495685323503e-13, "grad_norm": 1.268084168434143, "learning_rate": 0.00013300349835101237, "loss": 0.1144, "step": 24630 }, { "epoch": 1.709743457922741e-13, "grad_norm": 1.1150197982788086, "learning_rate": 0.0001329891391758063, "loss": 0.113, "step": 24640 }, { "epoch": 1.7104373473131318e-13, "grad_norm": 1.5627896785736084, "learning_rate": 0.0001329747748120556, "loss": 0.1308, "step": 24650 }, { "epoch": 1.7111312367035225e-13, "grad_norm": 1.3865529298782349, "learning_rate": 0.00013296040526123646, "loss": 0.1084, "step": 24660 }, { "epoch": 1.7118251260939132e-13, "grad_norm": 0.9820379018783569, "learning_rate": 0.00013294603052482553, "loss": 0.1205, "step": 24670 }, { "epoch": 1.712519015484304e-13, "grad_norm": 1.2264890670776367, "learning_rate": 0.0001329316506043001, "loss": 0.121, "step": 24680 }, { "epoch": 1.7132129048746947e-13, "grad_norm": 1.1407119035720825, "learning_rate": 0.00013291726550113787, "loss": 0.1182, "step": 24690 }, { "epoch": 1.7139067942650854e-13, "grad_norm": 1.1433329582214355, "learning_rate": 0.00013290287521681713, "loss": 0.1171, "step": 24700 }, { "epoch": 1.7146006836554761e-13, "grad_norm": 1.0542411804199219, "learning_rate": 0.00013288847975281674, "loss": 0.1243, "step": 24710 }, { "epoch": 1.7152945730458669e-13, "grad_norm": 1.3614171743392944, "learning_rate": 0.00013287407911061604, "loss": 0.1121, "step": 24720 }, { "epoch": 1.7159884624362576e-13, "grad_norm": 1.3591220378875732, "learning_rate": 0.00013285967329169492, "loss": 0.1215, "step": 24730 }, { "epoch": 1.7166823518266483e-13, "grad_norm": 1.2478532791137695, "learning_rate": 0.0001328452622975338, "loss": 0.1352, "step": 24740 }, { "epoch": 1.717376241217039e-13, "grad_norm": 1.3370963335037231, "learning_rate": 0.00013283084612961365, "loss": 0.1131, "step": 24750 }, { "epoch": 1.7180701306074297e-13, "grad_norm": 1.4419578313827515, "learning_rate": 0.00013281642478941597, "loss": 0.1162, "step": 24760 }, { "epoch": 1.7187640199978205e-13, "grad_norm": 1.6744327545166016, "learning_rate": 0.00013280199827842273, "loss": 0.1296, "step": 24770 }, { "epoch": 1.7194579093882112e-13, "grad_norm": 1.4347509145736694, "learning_rate": 0.0001327875665981165, "loss": 0.1183, "step": 24780 }, { "epoch": 1.720151798778602e-13, "grad_norm": 1.3847639560699463, "learning_rate": 0.00013277312974998038, "loss": 0.104, "step": 24790 }, { "epoch": 1.7208456881689926e-13, "grad_norm": 1.3941097259521484, "learning_rate": 0.00013275868773549796, "loss": 0.1211, "step": 24800 }, { "epoch": 1.7215395775593834e-13, "grad_norm": 1.3369346857070923, "learning_rate": 0.00013274424055615338, "loss": 0.1163, "step": 24810 }, { "epoch": 1.722233466949774e-13, "grad_norm": 1.1950114965438843, "learning_rate": 0.00013272978821343131, "loss": 0.1065, "step": 24820 }, { "epoch": 1.7229273563401648e-13, "grad_norm": 1.213114857673645, "learning_rate": 0.00013271533070881696, "loss": 0.1167, "step": 24830 }, { "epoch": 1.7236212457305555e-13, "grad_norm": 1.3211146593093872, "learning_rate": 0.00013270086804379608, "loss": 0.1203, "step": 24840 }, { "epoch": 1.7243151351209463e-13, "grad_norm": 1.180115818977356, "learning_rate": 0.00013268640021985492, "loss": 0.1194, "step": 24850 }, { "epoch": 1.725009024511337e-13, "grad_norm": 1.2611342668533325, "learning_rate": 0.00013267192723848023, "loss": 0.1277, "step": 24860 }, { "epoch": 1.7257029139017277e-13, "grad_norm": 1.1423420906066895, "learning_rate": 0.00013265744910115942, "loss": 0.1085, "step": 24870 }, { "epoch": 1.7263968032921184e-13, "grad_norm": 1.4576102495193481, "learning_rate": 0.00013264296580938026, "loss": 0.1212, "step": 24880 }, { "epoch": 1.7270906926825091e-13, "grad_norm": 1.5472830533981323, "learning_rate": 0.00013262847736463123, "loss": 0.122, "step": 24890 }, { "epoch": 1.7277845820728999e-13, "grad_norm": 1.3005807399749756, "learning_rate": 0.00013261398376840115, "loss": 0.1272, "step": 24900 }, { "epoch": 1.7284784714632906e-13, "grad_norm": 1.6802105903625488, "learning_rate": 0.0001325994850221795, "loss": 0.1082, "step": 24910 }, { "epoch": 1.7291723608536813e-13, "grad_norm": 1.08872652053833, "learning_rate": 0.00013258498112745623, "loss": 0.1067, "step": 24920 }, { "epoch": 1.729866250244072e-13, "grad_norm": 0.9423346519470215, "learning_rate": 0.0001325704720857219, "loss": 0.1277, "step": 24930 }, { "epoch": 1.7305601396344628e-13, "grad_norm": 1.259488821029663, "learning_rate": 0.00013255595789846746, "loss": 0.1097, "step": 24940 }, { "epoch": 1.7312540290248535e-13, "grad_norm": 1.3147269487380981, "learning_rate": 0.00013254143856718457, "loss": 0.1132, "step": 24950 }, { "epoch": 1.7319479184152442e-13, "grad_norm": 1.4079474210739136, "learning_rate": 0.00013252691409336518, "loss": 0.1162, "step": 24960 }, { "epoch": 1.732641807805635e-13, "grad_norm": 1.2041765451431274, "learning_rate": 0.00013251238447850204, "loss": 0.1077, "step": 24970 }, { "epoch": 1.7333356971960256e-13, "grad_norm": 1.564808964729309, "learning_rate": 0.0001324978497240882, "loss": 0.1159, "step": 24980 }, { "epoch": 1.7340295865864164e-13, "grad_norm": 1.1912305355072021, "learning_rate": 0.0001324833098316174, "loss": 0.1113, "step": 24990 }, { "epoch": 1.734723475976807e-13, "grad_norm": 1.3051865100860596, "learning_rate": 0.0001324687648025838, "loss": 0.1138, "step": 25000 }, { "epoch": 1.7354173653671978e-13, "grad_norm": 1.3526899814605713, "learning_rate": 0.00013245421463848212, "loss": 0.1156, "step": 25010 }, { "epoch": 1.7361112547575885e-13, "grad_norm": 1.126560091972351, "learning_rate": 0.00013243965934080763, "loss": 0.1251, "step": 25020 }, { "epoch": 1.7368051441479793e-13, "grad_norm": 1.1977200508117676, "learning_rate": 0.00013242509891105616, "loss": 0.1035, "step": 25030 }, { "epoch": 1.73749903353837e-13, "grad_norm": 1.1391147375106812, "learning_rate": 0.00013241053335072392, "loss": 0.1235, "step": 25040 }, { "epoch": 1.7381929229287607e-13, "grad_norm": 1.3000950813293457, "learning_rate": 0.00013239596266130784, "loss": 0.1299, "step": 25050 }, { "epoch": 1.7388868123191514e-13, "grad_norm": 1.0000401735305786, "learning_rate": 0.00013238138684430525, "loss": 0.1064, "step": 25060 }, { "epoch": 1.7395807017095422e-13, "grad_norm": 1.1787505149841309, "learning_rate": 0.00013236680590121403, "loss": 0.1037, "step": 25070 }, { "epoch": 1.740274591099933e-13, "grad_norm": 1.5936565399169922, "learning_rate": 0.0001323522198335326, "loss": 0.1174, "step": 25080 }, { "epoch": 1.7409684804903236e-13, "grad_norm": 1.1249845027923584, "learning_rate": 0.00013233762864275995, "loss": 0.1059, "step": 25090 }, { "epoch": 1.7416623698807143e-13, "grad_norm": 1.0444865226745605, "learning_rate": 0.0001323230323303955, "loss": 0.101, "step": 25100 }, { "epoch": 1.742356259271105e-13, "grad_norm": 1.2985899448394775, "learning_rate": 0.00013230843089793925, "loss": 0.1073, "step": 25110 }, { "epoch": 1.7430501486614958e-13, "grad_norm": 1.0158942937850952, "learning_rate": 0.00013229382434689173, "loss": 0.1202, "step": 25120 }, { "epoch": 1.7437440380518865e-13, "grad_norm": 1.1333653926849365, "learning_rate": 0.00013227921267875404, "loss": 0.101, "step": 25130 }, { "epoch": 1.7444379274422772e-13, "grad_norm": 1.0311483144760132, "learning_rate": 0.0001322645958950277, "loss": 0.1084, "step": 25140 }, { "epoch": 1.745131816832668e-13, "grad_norm": 1.1024121046066284, "learning_rate": 0.00013224997399721483, "loss": 0.1106, "step": 25150 }, { "epoch": 1.7458257062230587e-13, "grad_norm": 1.2821128368377686, "learning_rate": 0.00013223534698681805, "loss": 0.1157, "step": 25160 }, { "epoch": 1.7465195956134494e-13, "grad_norm": 1.519037127494812, "learning_rate": 0.0001322207148653405, "loss": 0.1155, "step": 25170 }, { "epoch": 1.74721348500384e-13, "grad_norm": 0.9736198782920837, "learning_rate": 0.0001322060776342859, "loss": 0.1132, "step": 25180 }, { "epoch": 1.7479073743942308e-13, "grad_norm": 1.139958381652832, "learning_rate": 0.00013219143529515846, "loss": 0.1088, "step": 25190 }, { "epoch": 1.7486012637846216e-13, "grad_norm": 1.143436074256897, "learning_rate": 0.00013217678784946285, "loss": 0.1181, "step": 25200 }, { "epoch": 1.7492951531750123e-13, "grad_norm": 1.2455534934997559, "learning_rate": 0.00013216213529870435, "loss": 0.1291, "step": 25210 }, { "epoch": 1.749989042565403e-13, "grad_norm": 0.9314767122268677, "learning_rate": 0.00013214747764438874, "loss": 0.1024, "step": 25220 }, { "epoch": 1.7506829319557937e-13, "grad_norm": 1.059744119644165, "learning_rate": 0.00013213281488802232, "loss": 0.105, "step": 25230 }, { "epoch": 1.7513768213461844e-13, "grad_norm": 1.3146718740463257, "learning_rate": 0.00013211814703111192, "loss": 0.1194, "step": 25240 }, { "epoch": 1.7520707107365752e-13, "grad_norm": 0.9262596368789673, "learning_rate": 0.0001321034740751649, "loss": 0.0945, "step": 25250 }, { "epoch": 1.752764600126966e-13, "grad_norm": 1.2389172315597534, "learning_rate": 0.0001320887960216891, "loss": 0.0959, "step": 25260 }, { "epoch": 1.7534584895173566e-13, "grad_norm": 1.0326571464538574, "learning_rate": 0.00013207411287219294, "loss": 0.1064, "step": 25270 }, { "epoch": 1.7541523789077473e-13, "grad_norm": 1.0176522731781006, "learning_rate": 0.00013205942462818536, "loss": 0.1083, "step": 25280 }, { "epoch": 1.754846268298138e-13, "grad_norm": 1.0921335220336914, "learning_rate": 0.0001320447312911758, "loss": 0.1021, "step": 25290 }, { "epoch": 1.7555401576885288e-13, "grad_norm": 1.503315806388855, "learning_rate": 0.0001320300328626742, "loss": 0.1172, "step": 25300 }, { "epoch": 1.7562340470789195e-13, "grad_norm": 1.2310285568237305, "learning_rate": 0.00013201532934419108, "loss": 0.1058, "step": 25310 }, { "epoch": 1.7569279364693102e-13, "grad_norm": 1.275752305984497, "learning_rate": 0.00013200062073723745, "loss": 0.1087, "step": 25320 }, { "epoch": 1.757621825859701e-13, "grad_norm": 1.4014532566070557, "learning_rate": 0.0001319859070433248, "loss": 0.1199, "step": 25330 }, { "epoch": 1.7583157152500917e-13, "grad_norm": 1.2859525680541992, "learning_rate": 0.00013197118826396528, "loss": 0.0988, "step": 25340 }, { "epoch": 1.7590096046404824e-13, "grad_norm": 1.4606021642684937, "learning_rate": 0.0001319564644006714, "loss": 0.1196, "step": 25350 }, { "epoch": 1.759703494030873e-13, "grad_norm": 1.5036470890045166, "learning_rate": 0.00013194173545495632, "loss": 0.1135, "step": 25360 }, { "epoch": 1.7603973834212638e-13, "grad_norm": 1.2189135551452637, "learning_rate": 0.00013192700142833363, "loss": 0.1136, "step": 25370 }, { "epoch": 1.7610912728116546e-13, "grad_norm": 1.1385712623596191, "learning_rate": 0.0001319122623223175, "loss": 0.103, "step": 25380 }, { "epoch": 1.7617851622020453e-13, "grad_norm": 1.2425233125686646, "learning_rate": 0.0001318975181384226, "loss": 0.0963, "step": 25390 }, { "epoch": 1.762479051592436e-13, "grad_norm": 1.2515766620635986, "learning_rate": 0.0001318827688781641, "loss": 0.1159, "step": 25400 }, { "epoch": 1.7631729409828267e-13, "grad_norm": 1.0549464225769043, "learning_rate": 0.00013186801454305772, "loss": 0.093, "step": 25410 }, { "epoch": 1.7638668303732175e-13, "grad_norm": 0.9147568941116333, "learning_rate": 0.00013185325513461976, "loss": 0.0979, "step": 25420 }, { "epoch": 1.7645607197636082e-13, "grad_norm": 1.202668309211731, "learning_rate": 0.0001318384906543669, "loss": 0.1085, "step": 25430 }, { "epoch": 1.765254609153999e-13, "grad_norm": 1.1173627376556396, "learning_rate": 0.00013182372110381646, "loss": 0.0996, "step": 25440 }, { "epoch": 1.7659484985443896e-13, "grad_norm": 1.1558938026428223, "learning_rate": 0.00013180894648448621, "loss": 0.1102, "step": 25450 }, { "epoch": 1.7666423879347803e-13, "grad_norm": 1.1026318073272705, "learning_rate": 0.00013179416679789455, "loss": 0.1048, "step": 25460 }, { "epoch": 1.767336277325171e-13, "grad_norm": 1.2153874635696411, "learning_rate": 0.00013177938204556023, "loss": 0.1075, "step": 25470 }, { "epoch": 1.7680301667155618e-13, "grad_norm": 1.2656323909759521, "learning_rate": 0.00013176459222900266, "loss": 0.102, "step": 25480 }, { "epoch": 1.7687240561059525e-13, "grad_norm": 1.2859373092651367, "learning_rate": 0.00013174979734974172, "loss": 0.113, "step": 25490 }, { "epoch": 1.7694179454963432e-13, "grad_norm": 1.3869919776916504, "learning_rate": 0.0001317349974092978, "loss": 0.0995, "step": 25500 }, { "epoch": 1.770111834886734e-13, "grad_norm": 1.0087759494781494, "learning_rate": 0.00013172019240919184, "loss": 0.111, "step": 25510 }, { "epoch": 1.7708057242771247e-13, "grad_norm": 1.1151095628738403, "learning_rate": 0.00013170538235094526, "loss": 0.1211, "step": 25520 }, { "epoch": 1.7714996136675154e-13, "grad_norm": 1.296323537826538, "learning_rate": 0.00013169056723608005, "loss": 0.1025, "step": 25530 }, { "epoch": 1.7721935030579061e-13, "grad_norm": 1.343057632446289, "learning_rate": 0.0001316757470661187, "loss": 0.098, "step": 25540 }, { "epoch": 1.7728873924482969e-13, "grad_norm": 1.3101695775985718, "learning_rate": 0.0001316609218425842, "loss": 0.1086, "step": 25550 }, { "epoch": 1.7735812818386876e-13, "grad_norm": 1.178957462310791, "learning_rate": 0.00013164609156700007, "loss": 0.103, "step": 25560 }, { "epoch": 1.7742751712290783e-13, "grad_norm": 1.0325347185134888, "learning_rate": 0.00013163125624089032, "loss": 0.089, "step": 25570 }, { "epoch": 1.774969060619469e-13, "grad_norm": 1.4195088148117065, "learning_rate": 0.0001316164158657796, "loss": 0.1043, "step": 25580 }, { "epoch": 1.7756629500098597e-13, "grad_norm": 1.0249550342559814, "learning_rate": 0.00013160157044319286, "loss": 0.1056, "step": 25590 }, { "epoch": 1.7763568394002505e-13, "grad_norm": 1.2319504022598267, "learning_rate": 0.0001315867199746558, "loss": 0.096, "step": 25600 }, { "epoch": 1.7770507287906412e-13, "grad_norm": 1.2958236932754517, "learning_rate": 0.00013157186446169453, "loss": 0.1068, "step": 25610 }, { "epoch": 1.777744618181032e-13, "grad_norm": 1.276450276374817, "learning_rate": 0.00013155700390583564, "loss": 0.103, "step": 25620 }, { "epoch": 1.7784385075714226e-13, "grad_norm": 1.096838355064392, "learning_rate": 0.0001315421383086063, "loss": 0.1073, "step": 25630 }, { "epoch": 1.7791323969618134e-13, "grad_norm": 1.1335175037384033, "learning_rate": 0.0001315272676715342, "loss": 0.1037, "step": 25640 }, { "epoch": 1.779826286352204e-13, "grad_norm": 0.971335232257843, "learning_rate": 0.0001315123919961475, "loss": 0.0988, "step": 25650 }, { "epoch": 1.7805201757425948e-13, "grad_norm": 1.2263644933700562, "learning_rate": 0.00013149751128397496, "loss": 0.1051, "step": 25660 }, { "epoch": 1.7812140651329855e-13, "grad_norm": 1.1989668607711792, "learning_rate": 0.00013148262553654575, "loss": 0.1093, "step": 25670 }, { "epoch": 1.7819079545233762e-13, "grad_norm": 0.9835513234138489, "learning_rate": 0.0001314677347553896, "loss": 0.1174, "step": 25680 }, { "epoch": 1.782601843913767e-13, "grad_norm": 0.8957812786102295, "learning_rate": 0.00013145283894203683, "loss": 0.0944, "step": 25690 }, { "epoch": 1.7832957333041577e-13, "grad_norm": 1.0768389701843262, "learning_rate": 0.0001314379380980182, "loss": 0.1015, "step": 25700 }, { "epoch": 1.7839896226945484e-13, "grad_norm": 0.9529315233230591, "learning_rate": 0.00013142303222486495, "loss": 0.1131, "step": 25710 }, { "epoch": 1.7846835120849391e-13, "grad_norm": 0.9136055707931519, "learning_rate": 0.00013140812132410898, "loss": 0.0935, "step": 25720 }, { "epoch": 1.7853774014753299e-13, "grad_norm": 1.0602242946624756, "learning_rate": 0.00013139320539728254, "loss": 0.0939, "step": 25730 }, { "epoch": 1.7860712908657206e-13, "grad_norm": 1.2591757774353027, "learning_rate": 0.00013137828444591848, "loss": 0.1002, "step": 25740 }, { "epoch": 1.7867651802561113e-13, "grad_norm": 1.036872386932373, "learning_rate": 0.0001313633584715502, "loss": 0.1011, "step": 25750 }, { "epoch": 1.787459069646502e-13, "grad_norm": 1.5465513467788696, "learning_rate": 0.00013134842747571153, "loss": 0.0988, "step": 25760 }, { "epoch": 1.7881529590368928e-13, "grad_norm": 1.3119146823883057, "learning_rate": 0.00013133349145993693, "loss": 0.1113, "step": 25770 }, { "epoch": 1.7888468484272835e-13, "grad_norm": 1.3402211666107178, "learning_rate": 0.00013131855042576126, "loss": 0.1004, "step": 25780 }, { "epoch": 1.7895407378176742e-13, "grad_norm": 1.087001085281372, "learning_rate": 0.00013130360437471993, "loss": 0.1001, "step": 25790 }, { "epoch": 1.790234627208065e-13, "grad_norm": 1.0791794061660767, "learning_rate": 0.00013128865330834893, "loss": 0.1093, "step": 25800 }, { "epoch": 1.7909285165984556e-13, "grad_norm": 0.9941415786743164, "learning_rate": 0.00013127369722818467, "loss": 0.0903, "step": 25810 }, { "epoch": 1.7916224059888464e-13, "grad_norm": 1.1986932754516602, "learning_rate": 0.0001312587361357641, "loss": 0.113, "step": 25820 }, { "epoch": 1.792316295379237e-13, "grad_norm": 1.4734926223754883, "learning_rate": 0.00013124377003262479, "loss": 0.1118, "step": 25830 }, { "epoch": 1.7930101847696278e-13, "grad_norm": 1.178462266921997, "learning_rate": 0.00013122879892030467, "loss": 0.101, "step": 25840 }, { "epoch": 1.7937040741600185e-13, "grad_norm": 0.9786094427108765, "learning_rate": 0.00013121382280034224, "loss": 0.0911, "step": 25850 }, { "epoch": 1.7943979635504093e-13, "grad_norm": 1.3913334608078003, "learning_rate": 0.00013119884167427662, "loss": 0.0988, "step": 25860 }, { "epoch": 1.7950918529408e-13, "grad_norm": 1.298830509185791, "learning_rate": 0.0001311838555436473, "loss": 0.1066, "step": 25870 }, { "epoch": 1.7957857423311907e-13, "grad_norm": 1.165212631225586, "learning_rate": 0.0001311688644099943, "loss": 0.0937, "step": 25880 }, { "epoch": 1.7964796317215814e-13, "grad_norm": 1.0013344287872314, "learning_rate": 0.00013115386827485822, "loss": 0.0914, "step": 25890 }, { "epoch": 1.7971735211119721e-13, "grad_norm": 1.4201936721801758, "learning_rate": 0.0001311388671397802, "loss": 0.1119, "step": 25900 }, { "epoch": 1.7978674105023629e-13, "grad_norm": 1.1440010070800781, "learning_rate": 0.00013112386100630174, "loss": 0.0929, "step": 25910 }, { "epoch": 1.7985612998927536e-13, "grad_norm": 0.91596919298172, "learning_rate": 0.00013110884987596505, "loss": 0.1025, "step": 25920 }, { "epoch": 1.7992551892831443e-13, "grad_norm": 1.152217984199524, "learning_rate": 0.00013109383375031267, "loss": 0.106, "step": 25930 }, { "epoch": 1.799949078673535e-13, "grad_norm": 1.0324705839157104, "learning_rate": 0.00013107881263088784, "loss": 0.1063, "step": 25940 }, { "epoch": 1.8006429680639258e-13, "grad_norm": 1.077831745147705, "learning_rate": 0.0001310637865192341, "loss": 0.0968, "step": 25950 }, { "epoch": 1.8013368574543165e-13, "grad_norm": 1.0398633480072021, "learning_rate": 0.00013104875541689574, "loss": 0.109, "step": 25960 }, { "epoch": 1.8020307468447072e-13, "grad_norm": 1.2145577669143677, "learning_rate": 0.00013103371932541734, "loss": 0.0948, "step": 25970 }, { "epoch": 1.802724636235098e-13, "grad_norm": 1.2983448505401611, "learning_rate": 0.00013101867824634415, "loss": 0.1053, "step": 25980 }, { "epoch": 1.8034185256254887e-13, "grad_norm": 1.0956933498382568, "learning_rate": 0.00013100363218122182, "loss": 0.1135, "step": 25990 }, { "epoch": 1.8041124150158794e-13, "grad_norm": 1.1460646390914917, "learning_rate": 0.0001309885811315966, "loss": 0.0969, "step": 26000 }, { "epoch": 1.80480630440627e-13, "grad_norm": 1.2497299909591675, "learning_rate": 0.00013097352509901523, "loss": 0.095, "step": 26010 }, { "epoch": 1.8055001937966608e-13, "grad_norm": 1.412459135055542, "learning_rate": 0.00013095846408502494, "loss": 0.1034, "step": 26020 }, { "epoch": 1.8061940831870515e-13, "grad_norm": 1.199131727218628, "learning_rate": 0.0001309433980911735, "loss": 0.0993, "step": 26030 }, { "epoch": 1.8068879725774423e-13, "grad_norm": 1.2302289009094238, "learning_rate": 0.00013092832711900917, "loss": 0.0871, "step": 26040 }, { "epoch": 1.807581861967833e-13, "grad_norm": 1.2128527164459229, "learning_rate": 0.0001309132511700807, "loss": 0.0998, "step": 26050 }, { "epoch": 1.8082757513582237e-13, "grad_norm": 1.3277194499969482, "learning_rate": 0.0001308981702459374, "loss": 0.1003, "step": 26060 }, { "epoch": 1.8089696407486144e-13, "grad_norm": 0.9983822703361511, "learning_rate": 0.00013088308434812903, "loss": 0.0908, "step": 26070 }, { "epoch": 1.8096635301390052e-13, "grad_norm": 1.0978436470031738, "learning_rate": 0.00013086799347820596, "loss": 0.1009, "step": 26080 }, { "epoch": 1.810357419529396e-13, "grad_norm": 1.217671275138855, "learning_rate": 0.000130852897637719, "loss": 0.1033, "step": 26090 }, { "epoch": 1.8110513089197866e-13, "grad_norm": 1.011109709739685, "learning_rate": 0.00013083779682821945, "loss": 0.1035, "step": 26100 }, { "epoch": 1.8117451983101773e-13, "grad_norm": 1.1630574464797974, "learning_rate": 0.0001308226910512592, "loss": 0.1023, "step": 26110 }, { "epoch": 1.812439087700568e-13, "grad_norm": 1.084748387336731, "learning_rate": 0.00013080758030839057, "loss": 0.0933, "step": 26120 }, { "epoch": 1.8131329770909588e-13, "grad_norm": 1.3783518075942993, "learning_rate": 0.0001307924646011664, "loss": 0.1029, "step": 26130 }, { "epoch": 1.8138268664813495e-13, "grad_norm": 1.3780888319015503, "learning_rate": 0.0001307773439311401, "loss": 0.1046, "step": 26140 }, { "epoch": 1.8145207558717402e-13, "grad_norm": 1.2319754362106323, "learning_rate": 0.00013076221829986558, "loss": 0.1068, "step": 26150 }, { "epoch": 1.815214645262131e-13, "grad_norm": 1.4005056619644165, "learning_rate": 0.00013074708770889718, "loss": 0.0917, "step": 26160 }, { "epoch": 1.8159085346525217e-13, "grad_norm": 0.9122030735015869, "learning_rate": 0.00013073195215978982, "loss": 0.0879, "step": 26170 }, { "epoch": 1.8166024240429124e-13, "grad_norm": 0.9154354929924011, "learning_rate": 0.00013071681165409895, "loss": 0.1021, "step": 26180 }, { "epoch": 1.817296313433303e-13, "grad_norm": 1.0301629304885864, "learning_rate": 0.00013070166619338046, "loss": 0.0928, "step": 26190 }, { "epoch": 1.8179902028236938e-13, "grad_norm": 1.1675856113433838, "learning_rate": 0.00013068651577919074, "loss": 0.0945, "step": 26200 }, { "epoch": 1.8186840922140846e-13, "grad_norm": 1.3172316551208496, "learning_rate": 0.0001306713604130868, "loss": 0.0996, "step": 26210 }, { "epoch": 1.8193779816044753e-13, "grad_norm": 1.2216566801071167, "learning_rate": 0.0001306562000966261, "loss": 0.0932, "step": 26220 }, { "epoch": 1.820071870994866e-13, "grad_norm": 1.4246238470077515, "learning_rate": 0.00013064103483136652, "loss": 0.0979, "step": 26230 }, { "epoch": 1.8207657603852567e-13, "grad_norm": 1.0663495063781738, "learning_rate": 0.00013062586461886656, "loss": 0.0961, "step": 26240 }, { "epoch": 1.8214596497756474e-13, "grad_norm": 1.3690556287765503, "learning_rate": 0.0001306106894606852, "loss": 0.098, "step": 26250 }, { "epoch": 1.8221535391660382e-13, "grad_norm": 1.1551963090896606, "learning_rate": 0.00013059550935838198, "loss": 0.0955, "step": 26260 }, { "epoch": 1.822847428556429e-13, "grad_norm": 1.104629397392273, "learning_rate": 0.0001305803243135168, "loss": 0.1031, "step": 26270 }, { "epoch": 1.8235413179468196e-13, "grad_norm": 1.0899055004119873, "learning_rate": 0.0001305651343276502, "loss": 0.0896, "step": 26280 }, { "epoch": 1.8242352073372103e-13, "grad_norm": 1.0802972316741943, "learning_rate": 0.0001305499394023432, "loss": 0.1055, "step": 26290 }, { "epoch": 1.824929096727601e-13, "grad_norm": 1.131181001663208, "learning_rate": 0.00013053473953915734, "loss": 0.1084, "step": 26300 }, { "epoch": 1.8256229861179918e-13, "grad_norm": 0.9201792478561401, "learning_rate": 0.00013051953473965455, "loss": 0.0915, "step": 26310 }, { "epoch": 1.8263168755083825e-13, "grad_norm": 1.025833249092102, "learning_rate": 0.00013050432500539741, "loss": 0.0865, "step": 26320 }, { "epoch": 1.8270107648987732e-13, "grad_norm": 1.3694249391555786, "learning_rate": 0.000130489110337949, "loss": 0.096, "step": 26330 }, { "epoch": 1.827704654289164e-13, "grad_norm": 1.1363694667816162, "learning_rate": 0.0001304738907388728, "loss": 0.0937, "step": 26340 }, { "epoch": 1.8283985436795547e-13, "grad_norm": 1.0210740566253662, "learning_rate": 0.0001304586662097329, "loss": 0.0868, "step": 26350 }, { "epoch": 1.8290924330699454e-13, "grad_norm": 1.2736327648162842, "learning_rate": 0.00013044343675209383, "loss": 0.0908, "step": 26360 }, { "epoch": 1.829786322460336e-13, "grad_norm": 1.0238455533981323, "learning_rate": 0.00013042820236752067, "loss": 0.106, "step": 26370 }, { "epoch": 1.8304802118507268e-13, "grad_norm": 1.0022027492523193, "learning_rate": 0.000130412963057579, "loss": 0.0874, "step": 26380 }, { "epoch": 1.8311741012411176e-13, "grad_norm": 0.9363425970077515, "learning_rate": 0.00013039771882383488, "loss": 0.0939, "step": 26390 }, { "epoch": 1.8318679906315083e-13, "grad_norm": 0.978100061416626, "learning_rate": 0.0001303824696678549, "loss": 0.0956, "step": 26400 }, { "epoch": 1.832561880021899e-13, "grad_norm": 1.2434794902801514, "learning_rate": 0.00013036721559120612, "loss": 0.1015, "step": 26410 }, { "epoch": 1.8332557694122897e-13, "grad_norm": 1.3580325841903687, "learning_rate": 0.00013035195659545618, "loss": 0.0976, "step": 26420 }, { "epoch": 1.8339496588026805e-13, "grad_norm": 0.9788486957550049, "learning_rate": 0.00013033669268217314, "loss": 0.0976, "step": 26430 }, { "epoch": 1.8346435481930712e-13, "grad_norm": 1.043886661529541, "learning_rate": 0.00013032142385292563, "loss": 0.0928, "step": 26440 }, { "epoch": 1.835337437583462e-13, "grad_norm": 1.1012427806854248, "learning_rate": 0.00013030615010928273, "loss": 0.1018, "step": 26450 }, { "epoch": 1.8360313269738526e-13, "grad_norm": 1.1833497285842896, "learning_rate": 0.0001302908714528141, "loss": 0.1093, "step": 26460 }, { "epoch": 1.8367252163642434e-13, "grad_norm": 0.9013774991035461, "learning_rate": 0.00013027558788508982, "loss": 0.0893, "step": 26470 }, { "epoch": 1.837419105754634e-13, "grad_norm": 0.94959557056427, "learning_rate": 0.00013026029940768055, "loss": 0.0872, "step": 26480 }, { "epoch": 1.8381129951450248e-13, "grad_norm": 1.1728285551071167, "learning_rate": 0.00013024500602215738, "loss": 0.0983, "step": 26490 }, { "epoch": 1.8388068845354155e-13, "grad_norm": 0.9208895564079285, "learning_rate": 0.00013022970773009194, "loss": 0.0881, "step": 26500 }, { "epoch": 1.8395007739258062e-13, "grad_norm": 1.1533912420272827, "learning_rate": 0.0001302144045330564, "loss": 0.0853, "step": 26510 }, { "epoch": 1.840194663316197e-13, "grad_norm": 0.9957232475280762, "learning_rate": 0.00013019909643262342, "loss": 0.0925, "step": 26520 }, { "epoch": 1.8408885527065877e-13, "grad_norm": 0.9806642532348633, "learning_rate": 0.0001301837834303661, "loss": 0.0978, "step": 26530 }, { "epoch": 1.8415824420969784e-13, "grad_norm": 0.9487854838371277, "learning_rate": 0.00013016846552785813, "loss": 0.0863, "step": 26540 }, { "epoch": 1.8422763314873691e-13, "grad_norm": 1.2651135921478271, "learning_rate": 0.0001301531427266736, "loss": 0.102, "step": 26550 }, { "epoch": 1.8429702208777599e-13, "grad_norm": 1.103935956954956, "learning_rate": 0.00013013781502838723, "loss": 0.0897, "step": 26560 }, { "epoch": 1.8436641102681506e-13, "grad_norm": 1.0859659910202026, "learning_rate": 0.00013012248243457413, "loss": 0.0966, "step": 26570 }, { "epoch": 1.8443579996585413e-13, "grad_norm": 1.2771179676055908, "learning_rate": 0.00013010714494681003, "loss": 0.1023, "step": 26580 }, { "epoch": 1.845051889048932e-13, "grad_norm": 1.1901531219482422, "learning_rate": 0.000130091802566671, "loss": 0.0905, "step": 26590 }, { "epoch": 1.8457457784393227e-13, "grad_norm": 1.3487049341201782, "learning_rate": 0.00013007645529573378, "loss": 0.1034, "step": 26600 }, { "epoch": 1.8464396678297135e-13, "grad_norm": 1.4534698724746704, "learning_rate": 0.00013006110313557552, "loss": 0.1016, "step": 26610 }, { "epoch": 1.8471335572201042e-13, "grad_norm": 1.221246600151062, "learning_rate": 0.0001300457460877739, "loss": 0.1001, "step": 26620 }, { "epoch": 1.847827446610495e-13, "grad_norm": 0.9679844379425049, "learning_rate": 0.0001300303841539071, "loss": 0.0874, "step": 26630 }, { "epoch": 1.8485213360008856e-13, "grad_norm": 1.0840195417404175, "learning_rate": 0.00013001501733555376, "loss": 0.0813, "step": 26640 }, { "epoch": 1.8492152253912764e-13, "grad_norm": 1.2184923887252808, "learning_rate": 0.00012999964563429313, "loss": 0.0962, "step": 26650 }, { "epoch": 1.849909114781667e-13, "grad_norm": 1.0314174890518188, "learning_rate": 0.00012998426905170483, "loss": 0.0862, "step": 26660 }, { "epoch": 1.8506030041720578e-13, "grad_norm": 0.7958972454071045, "learning_rate": 0.00012996888758936903, "loss": 0.0856, "step": 26670 }, { "epoch": 1.8512968935624485e-13, "grad_norm": 1.084032416343689, "learning_rate": 0.00012995350124886648, "loss": 0.0962, "step": 26680 }, { "epoch": 1.8519907829528393e-13, "grad_norm": 1.0565812587738037, "learning_rate": 0.00012993811003177833, "loss": 0.0905, "step": 26690 }, { "epoch": 1.85268467234323e-13, "grad_norm": 1.1335512399673462, "learning_rate": 0.00012992271393968627, "loss": 0.0955, "step": 26700 }, { "epoch": 1.8533785617336207e-13, "grad_norm": 1.0127862691879272, "learning_rate": 0.00012990731297417247, "loss": 0.0909, "step": 26710 }, { "epoch": 1.8540724511240114e-13, "grad_norm": 1.1877596378326416, "learning_rate": 0.00012989190713681966, "loss": 0.0957, "step": 26720 }, { "epoch": 1.8547663405144021e-13, "grad_norm": 1.1384557485580444, "learning_rate": 0.000129876496429211, "loss": 0.0902, "step": 26730 }, { "epoch": 1.8554602299047929e-13, "grad_norm": 1.20585036277771, "learning_rate": 0.0001298610808529302, "loss": 0.1001, "step": 26740 }, { "epoch": 1.8561541192951836e-13, "grad_norm": 1.1993380784988403, "learning_rate": 0.0001298456604095614, "loss": 0.09, "step": 26750 }, { "epoch": 1.8568480086855743e-13, "grad_norm": 0.9391906261444092, "learning_rate": 0.00012983023510068935, "loss": 0.0979, "step": 26760 }, { "epoch": 1.857541898075965e-13, "grad_norm": 1.1208977699279785, "learning_rate": 0.0001298148049278992, "loss": 0.111, "step": 26770 }, { "epoch": 1.8582357874663558e-13, "grad_norm": 1.2300900220870972, "learning_rate": 0.00012979936989277667, "loss": 0.0928, "step": 26780 }, { "epoch": 1.8589296768567465e-13, "grad_norm": 1.1174955368041992, "learning_rate": 0.00012978392999690795, "loss": 0.0831, "step": 26790 }, { "epoch": 1.8596235662471372e-13, "grad_norm": 1.0957835912704468, "learning_rate": 0.0001297684852418797, "loss": 0.0915, "step": 26800 }, { "epoch": 1.860317455637528e-13, "grad_norm": 1.1811069250106812, "learning_rate": 0.00012975303562927912, "loss": 0.0909, "step": 26810 }, { "epoch": 1.8610113450279187e-13, "grad_norm": 1.0488722324371338, "learning_rate": 0.00012973758116069392, "loss": 0.0794, "step": 26820 }, { "epoch": 1.8617052344183094e-13, "grad_norm": 1.245032548904419, "learning_rate": 0.00012972212183771228, "loss": 0.0901, "step": 26830 }, { "epoch": 1.8623991238087e-13, "grad_norm": 0.9910404086112976, "learning_rate": 0.00012970665766192288, "loss": 0.0923, "step": 26840 }, { "epoch": 1.8630930131990908e-13, "grad_norm": 1.1379848718643188, "learning_rate": 0.00012969118863491485, "loss": 0.087, "step": 26850 }, { "epoch": 1.8637869025894815e-13, "grad_norm": 1.2524570226669312, "learning_rate": 0.00012967571475827798, "loss": 0.0918, "step": 26860 }, { "epoch": 1.8644807919798723e-13, "grad_norm": 1.1465871334075928, "learning_rate": 0.0001296602360336024, "loss": 0.0873, "step": 26870 }, { "epoch": 1.865174681370263e-13, "grad_norm": 1.1040983200073242, "learning_rate": 0.00012964475246247874, "loss": 0.0981, "step": 26880 }, { "epoch": 1.8658685707606537e-13, "grad_norm": 1.2506444454193115, "learning_rate": 0.00012962926404649826, "loss": 0.0947, "step": 26890 }, { "epoch": 1.8665624601510444e-13, "grad_norm": 0.9852734208106995, "learning_rate": 0.0001296137707872526, "loss": 0.0917, "step": 26900 }, { "epoch": 1.8672563495414352e-13, "grad_norm": 1.1045382022857666, "learning_rate": 0.0001295982726863339, "loss": 0.0936, "step": 26910 }, { "epoch": 1.867950238931826e-13, "grad_norm": 1.1589806079864502, "learning_rate": 0.0001295827697453349, "loss": 0.0966, "step": 26920 }, { "epoch": 1.8686441283222166e-13, "grad_norm": 1.0507047176361084, "learning_rate": 0.00012956726196584873, "loss": 0.1078, "step": 26930 }, { "epoch": 1.8693380177126073e-13, "grad_norm": 0.8507964611053467, "learning_rate": 0.00012955174934946903, "loss": 0.0882, "step": 26940 }, { "epoch": 1.870031907102998e-13, "grad_norm": 1.042676568031311, "learning_rate": 0.00012953623189779003, "loss": 0.0868, "step": 26950 }, { "epoch": 1.8707257964933888e-13, "grad_norm": 0.9114885926246643, "learning_rate": 0.00012952070961240632, "loss": 0.0953, "step": 26960 }, { "epoch": 1.8714196858837795e-13, "grad_norm": 0.9250043630599976, "learning_rate": 0.00012950518249491312, "loss": 0.0863, "step": 26970 }, { "epoch": 1.8721135752741702e-13, "grad_norm": 1.0164796113967896, "learning_rate": 0.00012948965054690602, "loss": 0.0842, "step": 26980 }, { "epoch": 1.872807464664561e-13, "grad_norm": 1.2059979438781738, "learning_rate": 0.0001294741137699812, "loss": 0.0877, "step": 26990 }, { "epoch": 1.8735013540549517e-13, "grad_norm": 1.053693175315857, "learning_rate": 0.00012945857216573533, "loss": 0.0904, "step": 27000 }, { "epoch": 1.8741952434453424e-13, "grad_norm": 1.3986055850982666, "learning_rate": 0.0001294430257357655, "loss": 0.0878, "step": 27010 }, { "epoch": 1.874889132835733e-13, "grad_norm": 1.2118083238601685, "learning_rate": 0.0001294274744816694, "loss": 0.0989, "step": 27020 }, { "epoch": 1.8755830222261238e-13, "grad_norm": 1.1977405548095703, "learning_rate": 0.00012941191840504511, "loss": 0.0845, "step": 27030 }, { "epoch": 1.8762769116165146e-13, "grad_norm": 1.012107491493225, "learning_rate": 0.00012939635750749132, "loss": 0.0901, "step": 27040 }, { "epoch": 1.8769708010069053e-13, "grad_norm": 1.1291608810424805, "learning_rate": 0.00012938079179060708, "loss": 0.0991, "step": 27050 }, { "epoch": 1.877664690397296e-13, "grad_norm": 1.0638360977172852, "learning_rate": 0.00012936522125599206, "loss": 0.0855, "step": 27060 }, { "epoch": 1.8783585797876867e-13, "grad_norm": 1.1803584098815918, "learning_rate": 0.0001293496459052464, "loss": 0.1016, "step": 27070 }, { "epoch": 1.8790524691780774e-13, "grad_norm": 1.4547141790390015, "learning_rate": 0.00012933406573997064, "loss": 0.1016, "step": 27080 }, { "epoch": 1.8797463585684682e-13, "grad_norm": 1.261317491531372, "learning_rate": 0.00012931848076176594, "loss": 0.0955, "step": 27090 }, { "epoch": 1.880440247958859e-13, "grad_norm": 0.9876871705055237, "learning_rate": 0.00012930289097223385, "loss": 0.085, "step": 27100 }, { "epoch": 1.8811341373492496e-13, "grad_norm": 1.3894985914230347, "learning_rate": 0.00012928729637297652, "loss": 0.0865, "step": 27110 }, { "epoch": 1.8818280267396403e-13, "grad_norm": 1.2268089056015015, "learning_rate": 0.0001292716969655965, "loss": 0.0924, "step": 27120 }, { "epoch": 1.882521916130031e-13, "grad_norm": 1.1115657091140747, "learning_rate": 0.00012925609275169688, "loss": 0.0851, "step": 27130 }, { "epoch": 1.8832158055204218e-13, "grad_norm": 0.9179831743240356, "learning_rate": 0.00012924048373288126, "loss": 0.0805, "step": 27140 }, { "epoch": 1.8839096949108125e-13, "grad_norm": 1.494208812713623, "learning_rate": 0.00012922486991075364, "loss": 0.0971, "step": 27150 }, { "epoch": 1.8846035843012032e-13, "grad_norm": 1.114138126373291, "learning_rate": 0.00012920925128691866, "loss": 0.0836, "step": 27160 }, { "epoch": 1.885297473691594e-13, "grad_norm": 0.9002746343612671, "learning_rate": 0.00012919362786298135, "loss": 0.0908, "step": 27170 }, { "epoch": 1.8859913630819847e-13, "grad_norm": 1.0276561975479126, "learning_rate": 0.00012917799964054725, "loss": 0.0938, "step": 27180 }, { "epoch": 1.8866852524723754e-13, "grad_norm": 1.0349056720733643, "learning_rate": 0.00012916236662122242, "loss": 0.0916, "step": 27190 }, { "epoch": 1.887379141862766e-13, "grad_norm": 0.9344289898872375, "learning_rate": 0.00012914672880661335, "loss": 0.0839, "step": 27200 }, { "epoch": 1.8880730312531568e-13, "grad_norm": 1.0597141981124878, "learning_rate": 0.00012913108619832714, "loss": 0.0978, "step": 27210 }, { "epoch": 1.8887669206435476e-13, "grad_norm": 1.2235198020935059, "learning_rate": 0.00012911543879797128, "loss": 0.088, "step": 27220 }, { "epoch": 1.8894608100339383e-13, "grad_norm": 1.2121621370315552, "learning_rate": 0.00012909978660715374, "loss": 0.0931, "step": 27230 }, { "epoch": 1.890154699424329e-13, "grad_norm": 1.0311949253082275, "learning_rate": 0.00012908412962748308, "loss": 0.103, "step": 27240 }, { "epoch": 1.8908485888147197e-13, "grad_norm": 1.2777961492538452, "learning_rate": 0.0001290684678605683, "loss": 0.0905, "step": 27250 }, { "epoch": 1.8915424782051105e-13, "grad_norm": 1.1589505672454834, "learning_rate": 0.00012905280130801888, "loss": 0.085, "step": 27260 }, { "epoch": 1.8922363675955012e-13, "grad_norm": 1.2537922859191895, "learning_rate": 0.00012903712997144474, "loss": 0.0904, "step": 27270 }, { "epoch": 1.892930256985892e-13, "grad_norm": 1.1979681253433228, "learning_rate": 0.00012902145385245645, "loss": 0.0896, "step": 27280 }, { "epoch": 1.8936241463762826e-13, "grad_norm": 1.1579278707504272, "learning_rate": 0.00012900577295266494, "loss": 0.0765, "step": 27290 }, { "epoch": 1.8943180357666733e-13, "grad_norm": 1.0174283981323242, "learning_rate": 0.00012899008727368166, "loss": 0.0845, "step": 27300 }, { "epoch": 1.895011925157064e-13, "grad_norm": 1.2458466291427612, "learning_rate": 0.00012897439681711854, "loss": 0.0891, "step": 27310 }, { "epoch": 1.8957058145474548e-13, "grad_norm": 0.9823107719421387, "learning_rate": 0.00012895870158458807, "loss": 0.0808, "step": 27320 }, { "epoch": 1.8963997039378455e-13, "grad_norm": 1.1137994527816772, "learning_rate": 0.00012894300157770313, "loss": 0.0912, "step": 27330 }, { "epoch": 1.8970935933282362e-13, "grad_norm": 1.1259360313415527, "learning_rate": 0.00012892729679807714, "loss": 0.0901, "step": 27340 }, { "epoch": 1.897787482718627e-13, "grad_norm": 0.9746826887130737, "learning_rate": 0.00012891158724732406, "loss": 0.0906, "step": 27350 }, { "epoch": 1.8984813721090177e-13, "grad_norm": 1.0495051145553589, "learning_rate": 0.00012889587292705824, "loss": 0.0886, "step": 27360 }, { "epoch": 1.8991752614994084e-13, "grad_norm": 1.0999404191970825, "learning_rate": 0.00012888015383889458, "loss": 0.0849, "step": 27370 }, { "epoch": 1.8998691508897991e-13, "grad_norm": 1.2523927688598633, "learning_rate": 0.00012886442998444848, "loss": 0.0952, "step": 27380 }, { "epoch": 1.9005630402801899e-13, "grad_norm": 1.2442151308059692, "learning_rate": 0.00012884870136533577, "loss": 0.0946, "step": 27390 }, { "epoch": 1.9012569296705806e-13, "grad_norm": 1.1852830648422241, "learning_rate": 0.0001288329679831729, "loss": 0.0966, "step": 27400 }, { "epoch": 1.9019508190609713e-13, "grad_norm": 1.5286377668380737, "learning_rate": 0.00012881722983957663, "loss": 0.0898, "step": 27410 }, { "epoch": 1.902644708451362e-13, "grad_norm": 0.9092168807983398, "learning_rate": 0.00012880148693616433, "loss": 0.0792, "step": 27420 }, { "epoch": 1.9033385978417527e-13, "grad_norm": 0.9163597822189331, "learning_rate": 0.0001287857392745538, "loss": 0.0903, "step": 27430 }, { "epoch": 1.9040324872321435e-13, "grad_norm": 1.0367367267608643, "learning_rate": 0.0001287699868563634, "loss": 0.0867, "step": 27440 }, { "epoch": 1.9047263766225342e-13, "grad_norm": 1.0822557210922241, "learning_rate": 0.00012875422968321198, "loss": 0.0839, "step": 27450 }, { "epoch": 1.905420266012925e-13, "grad_norm": 1.197063684463501, "learning_rate": 0.0001287384677567187, "loss": 0.0847, "step": 27460 }, { "epoch": 1.9061141554033156e-13, "grad_norm": 1.1503565311431885, "learning_rate": 0.00012872270107850347, "loss": 0.0851, "step": 27470 }, { "epoch": 1.9068080447937064e-13, "grad_norm": 1.3033220767974854, "learning_rate": 0.00012870692965018647, "loss": 0.0879, "step": 27480 }, { "epoch": 1.907501934184097e-13, "grad_norm": 1.057037591934204, "learning_rate": 0.00012869115347338852, "loss": 0.0906, "step": 27490 }, { "epoch": 1.9081958235744878e-13, "grad_norm": 1.277176856994629, "learning_rate": 0.00012867537254973086, "loss": 0.0872, "step": 27500 }, { "epoch": 1.9088897129648785e-13, "grad_norm": 1.0718640089035034, "learning_rate": 0.0001286595868808352, "loss": 0.0865, "step": 27510 }, { "epoch": 1.9095836023552692e-13, "grad_norm": 1.0141774415969849, "learning_rate": 0.00012864379646832376, "loss": 0.089, "step": 27520 }, { "epoch": 1.91027749174566e-13, "grad_norm": 1.0622121095657349, "learning_rate": 0.0001286280013138193, "loss": 0.0808, "step": 27530 }, { "epoch": 1.9109713811360507e-13, "grad_norm": 1.0381072759628296, "learning_rate": 0.00012861220141894495, "loss": 0.096, "step": 27540 }, { "epoch": 1.9116652705264414e-13, "grad_norm": 1.062558650970459, "learning_rate": 0.00012859639678532446, "loss": 0.0995, "step": 27550 }, { "epoch": 1.9123591599168321e-13, "grad_norm": 1.0114381313323975, "learning_rate": 0.00012858058741458195, "loss": 0.0857, "step": 27560 }, { "epoch": 1.9130530493072229e-13, "grad_norm": 1.0276494026184082, "learning_rate": 0.00012856477330834212, "loss": 0.085, "step": 27570 }, { "epoch": 1.9137469386976136e-13, "grad_norm": 1.3536255359649658, "learning_rate": 0.0001285489544682301, "loss": 0.088, "step": 27580 }, { "epoch": 1.9144408280880043e-13, "grad_norm": 1.2166783809661865, "learning_rate": 0.0001285331308958715, "loss": 0.0862, "step": 27590 }, { "epoch": 1.915134717478395e-13, "grad_norm": 1.0353736877441406, "learning_rate": 0.00012851730259289245, "loss": 0.0814, "step": 27600 }, { "epoch": 1.9158286068687858e-13, "grad_norm": 1.1290748119354248, "learning_rate": 0.00012850146956091957, "loss": 0.0831, "step": 27610 }, { "epoch": 1.9165224962591765e-13, "grad_norm": 0.9410660266876221, "learning_rate": 0.00012848563180157995, "loss": 0.0902, "step": 27620 }, { "epoch": 1.9172163856495672e-13, "grad_norm": 0.9510087370872498, "learning_rate": 0.00012846978931650116, "loss": 0.0803, "step": 27630 }, { "epoch": 1.917910275039958e-13, "grad_norm": 0.9417691230773926, "learning_rate": 0.00012845394210731124, "loss": 0.0848, "step": 27640 }, { "epoch": 1.9186041644303486e-13, "grad_norm": 0.9286977648735046, "learning_rate": 0.00012843809017563877, "loss": 0.0905, "step": 27650 }, { "epoch": 1.9192980538207394e-13, "grad_norm": 1.1443543434143066, "learning_rate": 0.00012842223352311276, "loss": 0.0926, "step": 27660 }, { "epoch": 1.91999194321113e-13, "grad_norm": 1.2067177295684814, "learning_rate": 0.00012840637215136275, "loss": 0.0853, "step": 27670 }, { "epoch": 1.9206858326015208e-13, "grad_norm": 0.8705170750617981, "learning_rate": 0.00012839050606201874, "loss": 0.0844, "step": 27680 }, { "epoch": 1.9213797219919115e-13, "grad_norm": 0.9702244400978088, "learning_rate": 0.0001283746352567112, "loss": 0.0836, "step": 27690 }, { "epoch": 1.9220736113823023e-13, "grad_norm": 1.068798542022705, "learning_rate": 0.0001283587597370711, "loss": 0.0903, "step": 27700 }, { "epoch": 1.922767500772693e-13, "grad_norm": 1.0789673328399658, "learning_rate": 0.0001283428795047299, "loss": 0.0984, "step": 27710 }, { "epoch": 1.9234613901630837e-13, "grad_norm": 0.9784782528877258, "learning_rate": 0.00012832699456131956, "loss": 0.0844, "step": 27720 }, { "epoch": 1.9241552795534744e-13, "grad_norm": 0.8798347115516663, "learning_rate": 0.0001283111049084725, "loss": 0.0814, "step": 27730 }, { "epoch": 1.9248491689438652e-13, "grad_norm": 1.1948001384735107, "learning_rate": 0.00012829521054782159, "loss": 0.0924, "step": 27740 }, { "epoch": 1.925543058334256e-13, "grad_norm": 1.024231195449829, "learning_rate": 0.00012827931148100028, "loss": 0.0845, "step": 27750 }, { "epoch": 1.9262369477246466e-13, "grad_norm": 1.2339823246002197, "learning_rate": 0.0001282634077096424, "loss": 0.0785, "step": 27760 }, { "epoch": 1.9269308371150373e-13, "grad_norm": 0.9425860047340393, "learning_rate": 0.00012824749923538234, "loss": 0.0833, "step": 27770 }, { "epoch": 1.927624726505428e-13, "grad_norm": 0.9889612793922424, "learning_rate": 0.00012823158605985494, "loss": 0.0858, "step": 27780 }, { "epoch": 1.9283186158958188e-13, "grad_norm": 0.9167054295539856, "learning_rate": 0.00012821566818469553, "loss": 0.0792, "step": 27790 }, { "epoch": 1.9290125052862095e-13, "grad_norm": 1.3831851482391357, "learning_rate": 0.0001281997456115399, "loss": 0.0933, "step": 27800 }, { "epoch": 1.9297063946766002e-13, "grad_norm": 1.0817558765411377, "learning_rate": 0.00012818381834202433, "loss": 0.082, "step": 27810 }, { "epoch": 1.930400284066991e-13, "grad_norm": 1.1168040037155151, "learning_rate": 0.00012816788637778562, "loss": 0.0891, "step": 27820 }, { "epoch": 1.9310941734573817e-13, "grad_norm": 1.1708756685256958, "learning_rate": 0.00012815194972046106, "loss": 0.088, "step": 27830 }, { "epoch": 1.9317880628477724e-13, "grad_norm": 1.083957552909851, "learning_rate": 0.00012813600837168832, "loss": 0.0798, "step": 27840 }, { "epoch": 1.932481952238163e-13, "grad_norm": 1.2719109058380127, "learning_rate": 0.00012812006233310565, "loss": 0.0922, "step": 27850 }, { "epoch": 1.9331758416285538e-13, "grad_norm": 1.4719600677490234, "learning_rate": 0.0001281041116063518, "loss": 0.091, "step": 27860 }, { "epoch": 1.9338697310189445e-13, "grad_norm": 1.0827378034591675, "learning_rate": 0.00012808815619306588, "loss": 0.0894, "step": 27870 }, { "epoch": 1.9345636204093353e-13, "grad_norm": 1.0660572052001953, "learning_rate": 0.0001280721960948876, "loss": 0.0841, "step": 27880 }, { "epoch": 1.935257509799726e-13, "grad_norm": 1.0029748678207397, "learning_rate": 0.0001280562313134571, "loss": 0.0753, "step": 27890 }, { "epoch": 1.9359513991901167e-13, "grad_norm": 1.1654772758483887, "learning_rate": 0.00012804026185041502, "loss": 0.0892, "step": 27900 }, { "epoch": 1.9366452885805074e-13, "grad_norm": 1.0033490657806396, "learning_rate": 0.0001280242877074024, "loss": 0.0811, "step": 27910 }, { "epoch": 1.9373391779708982e-13, "grad_norm": 0.8405134081840515, "learning_rate": 0.00012800830888606097, "loss": 0.0809, "step": 27920 }, { "epoch": 1.938033067361289e-13, "grad_norm": 1.0209846496582031, "learning_rate": 0.00012799232538803268, "loss": 0.0855, "step": 27930 }, { "epoch": 1.9387269567516796e-13, "grad_norm": 1.0621728897094727, "learning_rate": 0.00012797633721496016, "loss": 0.0827, "step": 27940 }, { "epoch": 1.9394208461420703e-13, "grad_norm": 1.0951058864593506, "learning_rate": 0.00012796034436848638, "loss": 0.087, "step": 27950 }, { "epoch": 1.940114735532461e-13, "grad_norm": 1.0211071968078613, "learning_rate": 0.0001279443468502549, "loss": 0.0859, "step": 27960 }, { "epoch": 1.9408086249228518e-13, "grad_norm": 1.3005675077438354, "learning_rate": 0.0001279283446619097, "loss": 0.0881, "step": 27970 }, { "epoch": 1.9415025143132425e-13, "grad_norm": 1.1025261878967285, "learning_rate": 0.00012791233780509522, "loss": 0.0832, "step": 27980 }, { "epoch": 1.9421964037036332e-13, "grad_norm": 1.1464141607284546, "learning_rate": 0.00012789632628145648, "loss": 0.0861, "step": 27990 }, { "epoch": 1.942890293094024e-13, "grad_norm": 1.162245512008667, "learning_rate": 0.00012788031009263888, "loss": 0.0798, "step": 28000 }, { "epoch": 1.9435841824844147e-13, "grad_norm": 0.9345223307609558, "learning_rate": 0.00012786428924028832, "loss": 0.0888, "step": 28010 }, { "epoch": 1.9442780718748054e-13, "grad_norm": 1.0791071653366089, "learning_rate": 0.0001278482637260512, "loss": 0.0984, "step": 28020 }, { "epoch": 1.944971961265196e-13, "grad_norm": 1.2076786756515503, "learning_rate": 0.00012783223355157438, "loss": 0.0839, "step": 28030 }, { "epoch": 1.9456658506555868e-13, "grad_norm": 1.2230074405670166, "learning_rate": 0.00012781619871850527, "loss": 0.0794, "step": 28040 }, { "epoch": 1.9463597400459776e-13, "grad_norm": 1.040511131286621, "learning_rate": 0.0001278001592284916, "loss": 0.0828, "step": 28050 }, { "epoch": 1.9470536294363683e-13, "grad_norm": 1.0994017124176025, "learning_rate": 0.00012778411508318176, "loss": 0.0813, "step": 28060 }, { "epoch": 1.947747518826759e-13, "grad_norm": 0.9790829420089722, "learning_rate": 0.0001277680662842245, "loss": 0.0734, "step": 28070 }, { "epoch": 1.9484414082171497e-13, "grad_norm": 1.156275987625122, "learning_rate": 0.00012775201283326908, "loss": 0.0838, "step": 28080 }, { "epoch": 1.9491352976075405e-13, "grad_norm": 0.9380320906639099, "learning_rate": 0.00012773595473196528, "loss": 0.0834, "step": 28090 }, { "epoch": 1.9498291869979312e-13, "grad_norm": 1.0822221040725708, "learning_rate": 0.00012771989198196327, "loss": 0.0817, "step": 28100 }, { "epoch": 1.950523076388322e-13, "grad_norm": 1.2635339498519897, "learning_rate": 0.00012770382458491378, "loss": 0.0836, "step": 28110 }, { "epoch": 1.9512169657787126e-13, "grad_norm": 1.0825541019439697, "learning_rate": 0.00012768775254246796, "loss": 0.0848, "step": 28120 }, { "epoch": 1.9519108551691033e-13, "grad_norm": 1.1027127504348755, "learning_rate": 0.00012767167585627746, "loss": 0.0895, "step": 28130 }, { "epoch": 1.952604744559494e-13, "grad_norm": 1.0030196905136108, "learning_rate": 0.00012765559452799445, "loss": 0.0851, "step": 28140 }, { "epoch": 1.9532986339498848e-13, "grad_norm": 0.9692121744155884, "learning_rate": 0.00012763950855927152, "loss": 0.0796, "step": 28150 }, { "epoch": 1.9539925233402755e-13, "grad_norm": 1.1324546337127686, "learning_rate": 0.0001276234179517617, "loss": 0.0839, "step": 28160 }, { "epoch": 1.9546864127306662e-13, "grad_norm": 1.1253814697265625, "learning_rate": 0.00012760732270711866, "loss": 0.0892, "step": 28170 }, { "epoch": 1.955380302121057e-13, "grad_norm": 0.9183614253997803, "learning_rate": 0.0001275912228269963, "loss": 0.095, "step": 28180 }, { "epoch": 1.9560741915114477e-13, "grad_norm": 0.8505924344062805, "learning_rate": 0.00012757511831304927, "loss": 0.0799, "step": 28190 }, { "epoch": 1.9567680809018384e-13, "grad_norm": 1.0300136804580688, "learning_rate": 0.00012755900916693247, "loss": 0.082, "step": 28200 }, { "epoch": 1.957461970292229e-13, "grad_norm": 0.9029402732849121, "learning_rate": 0.00012754289539030142, "loss": 0.0833, "step": 28210 }, { "epoch": 1.9581558596826198e-13, "grad_norm": 0.8680784702301025, "learning_rate": 0.00012752677698481198, "loss": 0.0767, "step": 28220 }, { "epoch": 1.9588497490730106e-13, "grad_norm": 1.1048896312713623, "learning_rate": 0.00012751065395212065, "loss": 0.079, "step": 28230 }, { "epoch": 1.9595436384634013e-13, "grad_norm": 1.139451026916504, "learning_rate": 0.00012749452629388428, "loss": 0.0806, "step": 28240 }, { "epoch": 1.960237527853792e-13, "grad_norm": 1.0326635837554932, "learning_rate": 0.00012747839401176027, "loss": 0.0818, "step": 28250 }, { "epoch": 1.9609314172441827e-13, "grad_norm": 1.2154232263565063, "learning_rate": 0.0001274622571074064, "loss": 0.0811, "step": 28260 }, { "epoch": 1.9616253066345735e-13, "grad_norm": 1.2017481327056885, "learning_rate": 0.00012744611558248106, "loss": 0.088, "step": 28270 }, { "epoch": 1.9623191960249642e-13, "grad_norm": 1.181439757347107, "learning_rate": 0.00012742996943864303, "loss": 0.0794, "step": 28280 }, { "epoch": 1.963013085415355e-13, "grad_norm": 1.0167112350463867, "learning_rate": 0.00012741381867755153, "loss": 0.0878, "step": 28290 }, { "epoch": 1.9637069748057456e-13, "grad_norm": 1.027436375617981, "learning_rate": 0.00012739766330086637, "loss": 0.0875, "step": 28300 }, { "epoch": 1.9644008641961364e-13, "grad_norm": 1.0233538150787354, "learning_rate": 0.00012738150331024768, "loss": 0.0768, "step": 28310 }, { "epoch": 1.965094753586527e-13, "grad_norm": 1.1102185249328613, "learning_rate": 0.00012736533870735624, "loss": 0.0909, "step": 28320 }, { "epoch": 1.9657886429769178e-13, "grad_norm": 1.3650933504104614, "learning_rate": 0.00012734916949385313, "loss": 0.0928, "step": 28330 }, { "epoch": 1.9664825323673085e-13, "grad_norm": 1.0863101482391357, "learning_rate": 0.00012733299567140006, "loss": 0.082, "step": 28340 }, { "epoch": 1.9671764217576992e-13, "grad_norm": 0.9678525328636169, "learning_rate": 0.0001273168172416591, "loss": 0.0779, "step": 28350 }, { "epoch": 1.96787031114809e-13, "grad_norm": 1.2385977506637573, "learning_rate": 0.00012730063420629287, "loss": 0.0798, "step": 28360 }, { "epoch": 1.9685642005384807e-13, "grad_norm": 1.1505639553070068, "learning_rate": 0.00012728444656696442, "loss": 0.0816, "step": 28370 }, { "epoch": 1.9692580899288714e-13, "grad_norm": 1.0173720121383667, "learning_rate": 0.00012726825432533725, "loss": 0.0768, "step": 28380 }, { "epoch": 1.9699519793192621e-13, "grad_norm": 0.9200519323348999, "learning_rate": 0.00012725205748307538, "loss": 0.0728, "step": 28390 }, { "epoch": 1.9706458687096529e-13, "grad_norm": 1.3532938957214355, "learning_rate": 0.0001272358560418433, "loss": 0.0864, "step": 28400 }, { "epoch": 1.9713397581000436e-13, "grad_norm": 1.1094499826431274, "learning_rate": 0.00012721965000330593, "loss": 0.075, "step": 28410 }, { "epoch": 1.9720336474904343e-13, "grad_norm": 0.8314544558525085, "learning_rate": 0.00012720343936912875, "loss": 0.0795, "step": 28420 }, { "epoch": 1.972727536880825e-13, "grad_norm": 1.0004255771636963, "learning_rate": 0.0001271872241409776, "loss": 0.083, "step": 28430 }, { "epoch": 1.9734214262712158e-13, "grad_norm": 1.0371743440628052, "learning_rate": 0.00012717100432051888, "loss": 0.0835, "step": 28440 }, { "epoch": 1.9741153156616065e-13, "grad_norm": 0.9654641151428223, "learning_rate": 0.00012715477990941942, "loss": 0.083, "step": 28450 }, { "epoch": 1.9748092050519972e-13, "grad_norm": 1.055358648300171, "learning_rate": 0.0001271385509093465, "loss": 0.0876, "step": 28460 }, { "epoch": 1.975503094442388e-13, "grad_norm": 1.119163155555725, "learning_rate": 0.00012712231732196793, "loss": 0.0773, "step": 28470 }, { "epoch": 1.9761969838327786e-13, "grad_norm": 1.1282120943069458, "learning_rate": 0.00012710607914895197, "loss": 0.0824, "step": 28480 }, { "epoch": 1.9768908732231694e-13, "grad_norm": 1.0638850927352905, "learning_rate": 0.00012708983639196737, "loss": 0.0941, "step": 28490 }, { "epoch": 1.97758476261356e-13, "grad_norm": 1.1148706674575806, "learning_rate": 0.00012707358905268324, "loss": 0.0782, "step": 28500 }, { "epoch": 1.9782786520039508e-13, "grad_norm": 1.215890884399414, "learning_rate": 0.00012705733713276932, "loss": 0.0782, "step": 28510 }, { "epoch": 1.9789725413943415e-13, "grad_norm": 1.1932663917541504, "learning_rate": 0.00012704108063389574, "loss": 0.0829, "step": 28520 }, { "epoch": 1.9796664307847323e-13, "grad_norm": 1.0931295156478882, "learning_rate": 0.00012702481955773312, "loss": 0.0811, "step": 28530 }, { "epoch": 1.980360320175123e-13, "grad_norm": 1.0214674472808838, "learning_rate": 0.00012700855390595248, "loss": 0.0691, "step": 28540 }, { "epoch": 1.9810542095655137e-13, "grad_norm": 0.9590651988983154, "learning_rate": 0.0001269922836802254, "loss": 0.0762, "step": 28550 }, { "epoch": 1.9817480989559044e-13, "grad_norm": 1.0946062803268433, "learning_rate": 0.0001269760088822239, "loss": 0.0772, "step": 28560 }, { "epoch": 1.9824419883462951e-13, "grad_norm": 0.9644160866737366, "learning_rate": 0.0001269597295136205, "loss": 0.0721, "step": 28570 }, { "epoch": 1.9831358777366859e-13, "grad_norm": 0.9613929986953735, "learning_rate": 0.0001269434455760881, "loss": 0.0796, "step": 28580 }, { "epoch": 1.9838297671270766e-13, "grad_norm": 1.0483512878417969, "learning_rate": 0.00012692715707130015, "loss": 0.0784, "step": 28590 }, { "epoch": 1.9845236565174673e-13, "grad_norm": 1.0255701541900635, "learning_rate": 0.0001269108640009306, "loss": 0.0815, "step": 28600 }, { "epoch": 1.985217545907858e-13, "grad_norm": 1.0838736295700073, "learning_rate": 0.00012689456636665375, "loss": 0.0842, "step": 28610 }, { "epoch": 1.9859114352982488e-13, "grad_norm": 0.9813454151153564, "learning_rate": 0.00012687826417014445, "loss": 0.0777, "step": 28620 }, { "epoch": 1.9866053246886395e-13, "grad_norm": 1.1223986148834229, "learning_rate": 0.00012686195741307802, "loss": 0.0804, "step": 28630 }, { "epoch": 1.9872992140790302e-13, "grad_norm": 1.2203494310379028, "learning_rate": 0.0001268456460971302, "loss": 0.0815, "step": 28640 }, { "epoch": 1.987993103469421e-13, "grad_norm": 1.0840797424316406, "learning_rate": 0.00012682933022397727, "loss": 0.086, "step": 28650 }, { "epoch": 1.9886869928598117e-13, "grad_norm": 1.3583277463912964, "learning_rate": 0.00012681300979529593, "loss": 0.0776, "step": 28660 }, { "epoch": 1.9893808822502024e-13, "grad_norm": 1.0082718133926392, "learning_rate": 0.00012679668481276336, "loss": 0.0729, "step": 28670 }, { "epoch": 1.990074771640593e-13, "grad_norm": 0.9420090913772583, "learning_rate": 0.0001267803552780572, "loss": 0.0832, "step": 28680 }, { "epoch": 1.9907686610309838e-13, "grad_norm": 1.0277491807937622, "learning_rate": 0.00012676402119285554, "loss": 0.0822, "step": 28690 }, { "epoch": 1.9914625504213745e-13, "grad_norm": 1.000876545906067, "learning_rate": 0.000126747682558837, "loss": 0.0748, "step": 28700 }, { "epoch": 1.9921564398117653e-13, "grad_norm": 1.0594571828842163, "learning_rate": 0.00012673133937768062, "loss": 0.0738, "step": 28710 }, { "epoch": 1.992850329202156e-13, "grad_norm": 1.048385739326477, "learning_rate": 0.00012671499165106592, "loss": 0.0731, "step": 28720 }, { "epoch": 1.9935442185925467e-13, "grad_norm": 1.1991115808486938, "learning_rate": 0.00012669863938067282, "loss": 0.0773, "step": 28730 }, { "epoch": 1.9942381079829374e-13, "grad_norm": 0.9606889486312866, "learning_rate": 0.00012668228256818188, "loss": 0.077, "step": 28740 }, { "epoch": 1.9949319973733282e-13, "grad_norm": 1.111046314239502, "learning_rate": 0.00012666592121527394, "loss": 0.0737, "step": 28750 }, { "epoch": 1.995625886763719e-13, "grad_norm": 1.1310917139053345, "learning_rate": 0.0001266495553236304, "loss": 0.0785, "step": 28760 }, { "epoch": 1.9963197761541096e-13, "grad_norm": 0.9672444462776184, "learning_rate": 0.0001266331848949331, "loss": 0.0811, "step": 28770 }, { "epoch": 1.9970136655445003e-13, "grad_norm": 0.9788880348205566, "learning_rate": 0.0001266168099308644, "loss": 0.074, "step": 28780 }, { "epoch": 1.997707554934891e-13, "grad_norm": 0.9588594436645508, "learning_rate": 0.00012660043043310703, "loss": 0.0825, "step": 28790 }, { "epoch": 1.9984014443252818e-13, "grad_norm": 1.0468392372131348, "learning_rate": 0.00012658404640334425, "loss": 0.087, "step": 28800 }, { "epoch": 1.9990953337156725e-13, "grad_norm": 0.9169132113456726, "learning_rate": 0.00012656765784325983, "loss": 0.0756, "step": 28810 }, { "epoch": 1.9997892231060632e-13, "grad_norm": 1.0316805839538574, "learning_rate": 0.00012655126475453788, "loss": 0.0745, "step": 28820 }, { "epoch": 2.000483112496454e-13, "grad_norm": 1.2737044095993042, "learning_rate": 0.00012653486713886308, "loss": 0.0804, "step": 28830 }, { "epoch": 2.0011770018868447e-13, "grad_norm": 1.2355999946594238, "learning_rate": 0.0001265184649979205, "loss": 0.0796, "step": 28840 }, { "epoch": 2.0018708912772354e-13, "grad_norm": 1.042900562286377, "learning_rate": 0.00012650205833339575, "loss": 0.0785, "step": 28850 }, { "epoch": 2.002564780667626e-13, "grad_norm": 1.0713704824447632, "learning_rate": 0.00012648564714697486, "loss": 0.0743, "step": 28860 }, { "epoch": 2.0032586700580168e-13, "grad_norm": 0.8673224449157715, "learning_rate": 0.00012646923144034436, "loss": 0.0788, "step": 28870 }, { "epoch": 2.0039525594484076e-13, "grad_norm": 0.9151037931442261, "learning_rate": 0.0001264528112151912, "loss": 0.0684, "step": 28880 }, { "epoch": 2.0046464488387983e-13, "grad_norm": 0.9060225486755371, "learning_rate": 0.0001264363864732028, "loss": 0.0744, "step": 28890 }, { "epoch": 2.005340338229189e-13, "grad_norm": 0.8712586760520935, "learning_rate": 0.00012641995721606708, "loss": 0.0754, "step": 28900 }, { "epoch": 2.0060342276195797e-13, "grad_norm": 1.0913761854171753, "learning_rate": 0.00012640352344547238, "loss": 0.0777, "step": 28910 }, { "epoch": 2.0067281170099704e-13, "grad_norm": 1.1815871000289917, "learning_rate": 0.00012638708516310754, "loss": 0.0805, "step": 28920 }, { "epoch": 2.0074220064003612e-13, "grad_norm": 0.8783786296844482, "learning_rate": 0.00012637064237066187, "loss": 0.0762, "step": 28930 }, { "epoch": 2.008115895790752e-13, "grad_norm": 0.9869687557220459, "learning_rate": 0.00012635419506982506, "loss": 0.0763, "step": 28940 }, { "epoch": 2.0088097851811426e-13, "grad_norm": 1.0213245153427124, "learning_rate": 0.00012633774326228738, "loss": 0.0805, "step": 28950 }, { "epoch": 2.0095036745715333e-13, "grad_norm": 1.0636097192764282, "learning_rate": 0.00012632128694973948, "loss": 0.0877, "step": 28960 }, { "epoch": 2.010197563961924e-13, "grad_norm": 0.9139089584350586, "learning_rate": 0.00012630482613387256, "loss": 0.0731, "step": 28970 }, { "epoch": 2.0108914533523148e-13, "grad_norm": 0.8461382985115051, "learning_rate": 0.00012628836081637817, "loss": 0.0714, "step": 28980 }, { "epoch": 2.0115853427427055e-13, "grad_norm": 0.9724683165550232, "learning_rate": 0.00012627189099894835, "loss": 0.0812, "step": 28990 }, { "epoch": 2.0122792321330962e-13, "grad_norm": 0.9541454911231995, "learning_rate": 0.0001262554166832757, "loss": 0.0796, "step": 29000 }, { "epoch": 2.012973121523487e-13, "grad_norm": 1.202180027961731, "learning_rate": 0.00012623893787105318, "loss": 0.0755, "step": 29010 }, { "epoch": 2.0136670109138777e-13, "grad_norm": 0.9061958193778992, "learning_rate": 0.00012622245456397424, "loss": 0.0777, "step": 29020 }, { "epoch": 2.0143609003042684e-13, "grad_norm": 0.887723982334137, "learning_rate": 0.00012620596676373282, "loss": 0.0753, "step": 29030 }, { "epoch": 2.015054789694659e-13, "grad_norm": 0.8598712682723999, "learning_rate": 0.00012618947447202327, "loss": 0.0691, "step": 29040 }, { "epoch": 2.0157486790850498e-13, "grad_norm": 1.2076345682144165, "learning_rate": 0.00012617297769054042, "loss": 0.0798, "step": 29050 }, { "epoch": 2.0164425684754406e-13, "grad_norm": 0.9749704599380493, "learning_rate": 0.00012615647642097962, "loss": 0.0718, "step": 29060 }, { "epoch": 2.0171364578658313e-13, "grad_norm": 1.085344672203064, "learning_rate": 0.00012613997066503657, "loss": 0.0771, "step": 29070 }, { "epoch": 2.017830347256222e-13, "grad_norm": 1.0818777084350586, "learning_rate": 0.00012612346042440754, "loss": 0.0828, "step": 29080 }, { "epoch": 2.0185242366466127e-13, "grad_norm": 1.1138375997543335, "learning_rate": 0.0001261069457007892, "loss": 0.0721, "step": 29090 }, { "epoch": 2.0192181260370035e-13, "grad_norm": 1.286953330039978, "learning_rate": 0.00012609042649587871, "loss": 0.0832, "step": 29100 }, { "epoch": 2.0199120154273942e-13, "grad_norm": 1.2817775011062622, "learning_rate": 0.00012607390281137366, "loss": 0.0804, "step": 29110 }, { "epoch": 2.020605904817785e-13, "grad_norm": 1.0272959470748901, "learning_rate": 0.00012605737464897209, "loss": 0.0803, "step": 29120 }, { "epoch": 2.0212997942081756e-13, "grad_norm": 0.9406805634498596, "learning_rate": 0.0001260408420103726, "loss": 0.0746, "step": 29130 }, { "epoch": 2.0219936835985663e-13, "grad_norm": 0.9877521991729736, "learning_rate": 0.0001260243048972741, "loss": 0.0671, "step": 29140 }, { "epoch": 2.022687572988957e-13, "grad_norm": 1.1082727909088135, "learning_rate": 0.00012600776331137605, "loss": 0.0764, "step": 29150 }, { "epoch": 2.0233814623793478e-13, "grad_norm": 1.0188043117523193, "learning_rate": 0.00012599121725437835, "loss": 0.0763, "step": 29160 }, { "epoch": 2.0240753517697385e-13, "grad_norm": 0.860170841217041, "learning_rate": 0.00012597466672798142, "loss": 0.0763, "step": 29170 }, { "epoch": 2.0247692411601292e-13, "grad_norm": 0.931441605091095, "learning_rate": 0.00012595811173388605, "loss": 0.078, "step": 29180 }, { "epoch": 2.02546313055052e-13, "grad_norm": 0.9459874629974365, "learning_rate": 0.0001259415522737935, "loss": 0.073, "step": 29190 }, { "epoch": 2.0261570199409107e-13, "grad_norm": 0.9546277523040771, "learning_rate": 0.00012592498834940554, "loss": 0.0737, "step": 29200 }, { "epoch": 2.0268509093313014e-13, "grad_norm": 0.939481258392334, "learning_rate": 0.00012590841996242436, "loss": 0.0743, "step": 29210 }, { "epoch": 2.0275447987216921e-13, "grad_norm": 1.0750075578689575, "learning_rate": 0.00012589184711455265, "loss": 0.0745, "step": 29220 }, { "epoch": 2.0282386881120829e-13, "grad_norm": 1.1555768251419067, "learning_rate": 0.00012587526980749347, "loss": 0.0752, "step": 29230 }, { "epoch": 2.0289325775024736e-13, "grad_norm": 1.0612107515335083, "learning_rate": 0.00012585868804295043, "loss": 0.0784, "step": 29240 }, { "epoch": 2.0296264668928643e-13, "grad_norm": 1.1292935609817505, "learning_rate": 0.00012584210182262757, "loss": 0.0725, "step": 29250 }, { "epoch": 2.030320356283255e-13, "grad_norm": 0.8791369199752808, "learning_rate": 0.00012582551114822938, "loss": 0.0793, "step": 29260 }, { "epoch": 2.0310142456736457e-13, "grad_norm": 0.9163093566894531, "learning_rate": 0.00012580891602146082, "loss": 0.0856, "step": 29270 }, { "epoch": 2.0317081350640365e-13, "grad_norm": 1.1035462617874146, "learning_rate": 0.00012579231644402726, "loss": 0.0752, "step": 29280 }, { "epoch": 2.0324020244544272e-13, "grad_norm": 1.07320237159729, "learning_rate": 0.00012577571241763458, "loss": 0.0718, "step": 29290 }, { "epoch": 2.033095913844818e-13, "grad_norm": 0.9692752361297607, "learning_rate": 0.0001257591039439891, "loss": 0.074, "step": 29300 }, { "epoch": 2.0337898032352086e-13, "grad_norm": 1.2161539793014526, "learning_rate": 0.00012574249102479765, "loss": 0.0735, "step": 29310 }, { "epoch": 2.0344836926255994e-13, "grad_norm": 1.0517007112503052, "learning_rate": 0.00012572587366176737, "loss": 0.0687, "step": 29320 }, { "epoch": 2.03517758201599e-13, "grad_norm": 1.1587275266647339, "learning_rate": 0.00012570925185660607, "loss": 0.0773, "step": 29330 }, { "epoch": 2.0358714714063808e-13, "grad_norm": 0.9087308049201965, "learning_rate": 0.00012569262561102178, "loss": 0.0762, "step": 29340 }, { "epoch": 2.0365653607967715e-13, "grad_norm": 1.0039066076278687, "learning_rate": 0.0001256759949267232, "loss": 0.0708, "step": 29350 }, { "epoch": 2.0372592501871623e-13, "grad_norm": 1.070981740951538, "learning_rate": 0.00012565935980541932, "loss": 0.0721, "step": 29360 }, { "epoch": 2.037953139577553e-13, "grad_norm": 0.9836401343345642, "learning_rate": 0.0001256427202488197, "loss": 0.073, "step": 29370 }, { "epoch": 2.0386470289679437e-13, "grad_norm": 1.022847056388855, "learning_rate": 0.0001256260762586343, "loss": 0.0742, "step": 29380 }, { "epoch": 2.0393409183583344e-13, "grad_norm": 1.017547845840454, "learning_rate": 0.00012560942783657357, "loss": 0.0768, "step": 29390 }, { "epoch": 2.0400348077487251e-13, "grad_norm": 0.9236281514167786, "learning_rate": 0.00012559277498434837, "loss": 0.072, "step": 29400 }, { "epoch": 2.0407286971391159e-13, "grad_norm": 1.0410696268081665, "learning_rate": 0.00012557611770367004, "loss": 0.0748, "step": 29410 }, { "epoch": 2.0414225865295066e-13, "grad_norm": 0.9477593302726746, "learning_rate": 0.00012555945599625038, "loss": 0.0753, "step": 29420 }, { "epoch": 2.0421164759198973e-13, "grad_norm": 0.8950768709182739, "learning_rate": 0.00012554278986380164, "loss": 0.0801, "step": 29430 }, { "epoch": 2.042810365310288e-13, "grad_norm": 0.877460241317749, "learning_rate": 0.00012552611930803653, "loss": 0.0703, "step": 29440 }, { "epoch": 2.0435042547006788e-13, "grad_norm": 0.9797287583351135, "learning_rate": 0.0001255094443306682, "loss": 0.0733, "step": 29450 }, { "epoch": 2.0441981440910695e-13, "grad_norm": 0.8238462805747986, "learning_rate": 0.00012549276493341028, "loss": 0.0731, "step": 29460 }, { "epoch": 2.0448920334814602e-13, "grad_norm": 0.8730815649032593, "learning_rate": 0.00012547608111797683, "loss": 0.0702, "step": 29470 }, { "epoch": 2.045585922871851e-13, "grad_norm": 1.029665231704712, "learning_rate": 0.00012545939288608238, "loss": 0.0702, "step": 29480 }, { "epoch": 2.0462798122622416e-13, "grad_norm": 1.0695842504501343, "learning_rate": 0.00012544270023944184, "loss": 0.0737, "step": 29490 }, { "epoch": 2.0469737016526324e-13, "grad_norm": 0.9748758673667908, "learning_rate": 0.0001254260031797707, "loss": 0.0738, "step": 29500 }, { "epoch": 2.047667591043023e-13, "grad_norm": 1.1775007247924805, "learning_rate": 0.00012540930170878487, "loss": 0.0712, "step": 29510 }, { "epoch": 2.0483614804334138e-13, "grad_norm": 1.0251330137252808, "learning_rate": 0.00012539259582820066, "loss": 0.0755, "step": 29520 }, { "epoch": 2.0490553698238045e-13, "grad_norm": 1.040675401687622, "learning_rate": 0.0001253758855397348, "loss": 0.0689, "step": 29530 }, { "epoch": 2.0497492592141953e-13, "grad_norm": 0.9822264313697815, "learning_rate": 0.00012535917084510457, "loss": 0.0728, "step": 29540 }, { "epoch": 2.050443148604586e-13, "grad_norm": 0.9233497381210327, "learning_rate": 0.00012534245174602773, "loss": 0.0781, "step": 29550 }, { "epoch": 2.0511370379949767e-13, "grad_norm": 0.9229308366775513, "learning_rate": 0.00012532572824422236, "loss": 0.0686, "step": 29560 }, { "epoch": 2.0518309273853674e-13, "grad_norm": 1.010880947113037, "learning_rate": 0.00012530900034140707, "loss": 0.0797, "step": 29570 }, { "epoch": 2.0525248167757582e-13, "grad_norm": 1.1884275674819946, "learning_rate": 0.0001252922680393009, "loss": 0.0801, "step": 29580 }, { "epoch": 2.053218706166149e-13, "grad_norm": 1.091994047164917, "learning_rate": 0.00012527553133962336, "loss": 0.0722, "step": 29590 }, { "epoch": 2.0539125955565396e-13, "grad_norm": 0.947294294834137, "learning_rate": 0.00012525879024409447, "loss": 0.0689, "step": 29600 }, { "epoch": 2.0546064849469303e-13, "grad_norm": 1.1433732509613037, "learning_rate": 0.00012524204475443455, "loss": 0.0699, "step": 29610 }, { "epoch": 2.055300374337321e-13, "grad_norm": 0.9899142980575562, "learning_rate": 0.0001252252948723645, "loss": 0.0699, "step": 29620 }, { "epoch": 2.0559942637277118e-13, "grad_norm": 0.9550251960754395, "learning_rate": 0.00012520854059960563, "loss": 0.0688, "step": 29630 }, { "epoch": 2.0566881531181025e-13, "grad_norm": 0.9061142206192017, "learning_rate": 0.00012519178193787968, "loss": 0.0679, "step": 29640 }, { "epoch": 2.0573820425084932e-13, "grad_norm": 1.1847138404846191, "learning_rate": 0.0001251750188889089, "loss": 0.0793, "step": 29650 }, { "epoch": 2.058075931898884e-13, "grad_norm": 1.051498532295227, "learning_rate": 0.0001251582514544159, "loss": 0.0691, "step": 29660 }, { "epoch": 2.0587698212892747e-13, "grad_norm": 0.8276376128196716, "learning_rate": 0.0001251414796361239, "loss": 0.0716, "step": 29670 }, { "epoch": 2.0594637106796654e-13, "grad_norm": 0.8592110872268677, "learning_rate": 0.00012512470343575634, "loss": 0.0713, "step": 29680 }, { "epoch": 2.060157600070056e-13, "grad_norm": 0.8815467953681946, "learning_rate": 0.0001251079228550373, "loss": 0.0722, "step": 29690 }, { "epoch": 2.0608514894604468e-13, "grad_norm": 0.9255519509315491, "learning_rate": 0.00012509113789569123, "loss": 0.0715, "step": 29700 }, { "epoch": 2.0615453788508376e-13, "grad_norm": 0.9043813943862915, "learning_rate": 0.00012507434855944308, "loss": 0.0757, "step": 29710 }, { "epoch": 2.0622392682412283e-13, "grad_norm": 1.1526973247528076, "learning_rate": 0.00012505755484801818, "loss": 0.0708, "step": 29720 }, { "epoch": 2.062933157631619e-13, "grad_norm": 1.1016287803649902, "learning_rate": 0.00012504075676314234, "loss": 0.0753, "step": 29730 }, { "epoch": 2.0636270470220097e-13, "grad_norm": 0.9749714732170105, "learning_rate": 0.00012502395430654184, "loss": 0.0834, "step": 29740 }, { "epoch": 2.0643209364124004e-13, "grad_norm": 1.01514732837677, "learning_rate": 0.0001250071474799434, "loss": 0.0713, "step": 29750 }, { "epoch": 2.0650148258027912e-13, "grad_norm": 0.9645426273345947, "learning_rate": 0.00012499033628507418, "loss": 0.0688, "step": 29760 }, { "epoch": 2.065708715193182e-13, "grad_norm": 1.045177936553955, "learning_rate": 0.00012497352072366174, "loss": 0.0715, "step": 29770 }, { "epoch": 2.0664026045835726e-13, "grad_norm": 1.131277322769165, "learning_rate": 0.00012495670079743423, "loss": 0.0692, "step": 29780 }, { "epoch": 2.0670964939739633e-13, "grad_norm": 1.0013502836227417, "learning_rate": 0.00012493987650812012, "loss": 0.0637, "step": 29790 }, { "epoch": 2.067790383364354e-13, "grad_norm": 1.0149388313293457, "learning_rate": 0.00012492304785744832, "loss": 0.0726, "step": 29800 }, { "epoch": 2.0684842727547448e-13, "grad_norm": 1.0971713066101074, "learning_rate": 0.0001249062148471483, "loss": 0.0719, "step": 29810 }, { "epoch": 2.0691781621451355e-13, "grad_norm": 0.9373289942741394, "learning_rate": 0.00012488937747894994, "loss": 0.0666, "step": 29820 }, { "epoch": 2.0698720515355262e-13, "grad_norm": 0.9545922875404358, "learning_rate": 0.00012487253575458346, "loss": 0.0724, "step": 29830 }, { "epoch": 2.070565940925917e-13, "grad_norm": 0.9599703550338745, "learning_rate": 0.00012485568967577962, "loss": 0.0711, "step": 29840 }, { "epoch": 2.0712598303163077e-13, "grad_norm": 0.9514959454536438, "learning_rate": 0.00012483883924426967, "loss": 0.0716, "step": 29850 }, { "epoch": 2.0719537197066984e-13, "grad_norm": 0.98956298828125, "learning_rate": 0.0001248219844617852, "loss": 0.0736, "step": 29860 }, { "epoch": 2.072647609097089e-13, "grad_norm": 0.9691352844238281, "learning_rate": 0.00012480512533005834, "loss": 0.0673, "step": 29870 }, { "epoch": 2.0733414984874798e-13, "grad_norm": 1.195594072341919, "learning_rate": 0.00012478826185082162, "loss": 0.0757, "step": 29880 }, { "epoch": 2.0740353878778706e-13, "grad_norm": 1.274621844291687, "learning_rate": 0.00012477139402580802, "loss": 0.0768, "step": 29890 }, { "epoch": 2.0747292772682613e-13, "grad_norm": 1.0163902044296265, "learning_rate": 0.00012475452185675093, "loss": 0.08, "step": 29900 }, { "epoch": 2.075423166658652e-13, "grad_norm": 1.1183068752288818, "learning_rate": 0.00012473764534538434, "loss": 0.0718, "step": 29910 }, { "epoch": 2.0761170560490427e-13, "grad_norm": 0.8087285757064819, "learning_rate": 0.00012472076449344244, "loss": 0.0657, "step": 29920 }, { "epoch": 2.0768109454394335e-13, "grad_norm": 0.8108895421028137, "learning_rate": 0.0001247038793026601, "loss": 0.0733, "step": 29930 }, { "epoch": 2.0775048348298242e-13, "grad_norm": 1.076885461807251, "learning_rate": 0.00012468698977477247, "loss": 0.0683, "step": 29940 }, { "epoch": 2.078198724220215e-13, "grad_norm": 1.1110645532608032, "learning_rate": 0.00012467009591151523, "loss": 0.068, "step": 29950 }, { "epoch": 2.0788926136106056e-13, "grad_norm": 1.1162809133529663, "learning_rate": 0.00012465319771462455, "loss": 0.0706, "step": 29960 }, { "epoch": 2.0795865030009963e-13, "grad_norm": 1.0225876569747925, "learning_rate": 0.0001246362951858369, "loss": 0.0678, "step": 29970 }, { "epoch": 2.080280392391387e-13, "grad_norm": 1.155882477760315, "learning_rate": 0.0001246193883268893, "loss": 0.07, "step": 29980 }, { "epoch": 2.0809742817817778e-13, "grad_norm": 0.9509937763214111, "learning_rate": 0.00012460247713951922, "loss": 0.0715, "step": 29990 }, { "epoch": 2.0816681711721685e-13, "grad_norm": 1.088134527206421, "learning_rate": 0.0001245855616254645, "loss": 0.0688, "step": 30000 }, { "epoch": 2.0823620605625592e-13, "grad_norm": 1.0455971956253052, "learning_rate": 0.00012456864178646352, "loss": 0.0706, "step": 30010 }, { "epoch": 2.08305594995295e-13, "grad_norm": 0.8992842435836792, "learning_rate": 0.00012455171762425503, "loss": 0.0737, "step": 30020 }, { "epoch": 2.0837498393433407e-13, "grad_norm": 0.9411769509315491, "learning_rate": 0.00012453478914057828, "loss": 0.0658, "step": 30030 }, { "epoch": 2.0844437287337314e-13, "grad_norm": 0.9053667187690735, "learning_rate": 0.00012451785633717287, "loss": 0.0759, "step": 30040 }, { "epoch": 2.0851376181241221e-13, "grad_norm": 0.9789357781410217, "learning_rate": 0.00012450091921577898, "loss": 0.0801, "step": 30050 }, { "epoch": 2.0858315075145129e-13, "grad_norm": 0.8515798449516296, "learning_rate": 0.00012448397777813712, "loss": 0.0698, "step": 30060 }, { "epoch": 2.0865253969049036e-13, "grad_norm": 0.8986340761184692, "learning_rate": 0.00012446703202598832, "loss": 0.0682, "step": 30070 }, { "epoch": 2.0872192862952943e-13, "grad_norm": 1.1263936758041382, "learning_rate": 0.000124450081961074, "loss": 0.0706, "step": 30080 }, { "epoch": 2.087913175685685e-13, "grad_norm": 0.9514462947845459, "learning_rate": 0.00012443312758513605, "loss": 0.0677, "step": 30090 }, { "epoch": 2.0886070650760757e-13, "grad_norm": 0.9075345396995544, "learning_rate": 0.00012441616889991678, "loss": 0.0664, "step": 30100 }, { "epoch": 2.0893009544664665e-13, "grad_norm": 1.1420097351074219, "learning_rate": 0.00012439920590715897, "loss": 0.0665, "step": 30110 }, { "epoch": 2.0899948438568572e-13, "grad_norm": 0.8757203221321106, "learning_rate": 0.00012438223860860583, "loss": 0.0738, "step": 30120 }, { "epoch": 2.090688733247248e-13, "grad_norm": 0.9105010628700256, "learning_rate": 0.000124365267006001, "loss": 0.0647, "step": 30130 }, { "epoch": 2.0913826226376386e-13, "grad_norm": 0.910582959651947, "learning_rate": 0.0001243482911010886, "loss": 0.0678, "step": 30140 }, { "epoch": 2.0920765120280294e-13, "grad_norm": 0.9212582111358643, "learning_rate": 0.00012433131089561317, "loss": 0.0729, "step": 30150 }, { "epoch": 2.09277040141842e-13, "grad_norm": 1.0755435228347778, "learning_rate": 0.00012431432639131967, "loss": 0.0723, "step": 30160 }, { "epoch": 2.0934642908088108e-13, "grad_norm": 1.1286208629608154, "learning_rate": 0.00012429733758995354, "loss": 0.0722, "step": 30170 }, { "epoch": 2.0941581801992015e-13, "grad_norm": 0.820288896560669, "learning_rate": 0.00012428034449326063, "loss": 0.0698, "step": 30180 }, { "epoch": 2.0948520695895922e-13, "grad_norm": 0.892474353313446, "learning_rate": 0.00012426334710298722, "loss": 0.0678, "step": 30190 }, { "epoch": 2.095545958979983e-13, "grad_norm": 0.9043319821357727, "learning_rate": 0.0001242463454208801, "loss": 0.0712, "step": 30200 }, { "epoch": 2.0962398483703737e-13, "grad_norm": 0.9874981641769409, "learning_rate": 0.00012422933944868648, "loss": 0.0794, "step": 30210 }, { "epoch": 2.0969337377607644e-13, "grad_norm": 0.8800444006919861, "learning_rate": 0.0001242123291881539, "loss": 0.0671, "step": 30220 }, { "epoch": 2.0976276271511551e-13, "grad_norm": 0.873563826084137, "learning_rate": 0.00012419531464103053, "loss": 0.0681, "step": 30230 }, { "epoch": 2.0983215165415459e-13, "grad_norm": 1.0104987621307373, "learning_rate": 0.0001241782958090648, "loss": 0.0706, "step": 30240 }, { "epoch": 2.0990154059319366e-13, "grad_norm": 0.9022513031959534, "learning_rate": 0.00012416127269400572, "loss": 0.0662, "step": 30250 }, { "epoch": 2.0997092953223273e-13, "grad_norm": 0.9450763463973999, "learning_rate": 0.00012414424529760262, "loss": 0.0651, "step": 30260 }, { "epoch": 2.100403184712718e-13, "grad_norm": 0.8746439218521118, "learning_rate": 0.00012412721362160535, "loss": 0.0673, "step": 30270 }, { "epoch": 2.1010970741031088e-13, "grad_norm": 0.8486469388008118, "learning_rate": 0.0001241101776677642, "loss": 0.0682, "step": 30280 }, { "epoch": 2.1017909634934995e-13, "grad_norm": 0.8677980303764343, "learning_rate": 0.0001240931374378299, "loss": 0.0649, "step": 30290 }, { "epoch": 2.1024848528838902e-13, "grad_norm": 1.2445037364959717, "learning_rate": 0.00012407609293355352, "loss": 0.0734, "step": 30300 }, { "epoch": 2.103178742274281e-13, "grad_norm": 0.9894232153892517, "learning_rate": 0.00012405904415668673, "loss": 0.0685, "step": 30310 }, { "epoch": 2.1038726316646716e-13, "grad_norm": 1.0724287033081055, "learning_rate": 0.0001240419911089815, "loss": 0.0711, "step": 30320 }, { "epoch": 2.1045665210550624e-13, "grad_norm": 1.021417498588562, "learning_rate": 0.00012402493379219033, "loss": 0.0754, "step": 30330 }, { "epoch": 2.105260410445453e-13, "grad_norm": 1.018202781677246, "learning_rate": 0.00012400787220806611, "loss": 0.0679, "step": 30340 }, { "epoch": 2.1059542998358438e-13, "grad_norm": 1.2096805572509766, "learning_rate": 0.0001239908063583622, "loss": 0.0745, "step": 30350 }, { "epoch": 2.1066481892262345e-13, "grad_norm": 1.3015135526657104, "learning_rate": 0.00012397373624483237, "loss": 0.072, "step": 30360 }, { "epoch": 2.1073420786166253e-13, "grad_norm": 1.049099326133728, "learning_rate": 0.00012395666186923083, "loss": 0.0714, "step": 30370 }, { "epoch": 2.108035968007016e-13, "grad_norm": 0.9952385425567627, "learning_rate": 0.00012393958323331223, "loss": 0.0685, "step": 30380 }, { "epoch": 2.1087298573974067e-13, "grad_norm": 0.989203691482544, "learning_rate": 0.0001239225003388317, "loss": 0.0638, "step": 30390 }, { "epoch": 2.1094237467877974e-13, "grad_norm": 1.1290949583053589, "learning_rate": 0.00012390541318754475, "loss": 0.07, "step": 30400 }, { "epoch": 2.1101176361781881e-13, "grad_norm": 1.0424751043319702, "learning_rate": 0.00012388832178120734, "loss": 0.0658, "step": 30410 }, { "epoch": 2.1108115255685789e-13, "grad_norm": 0.7812362909317017, "learning_rate": 0.0001238712261215759, "loss": 0.0653, "step": 30420 }, { "epoch": 2.1115054149589696e-13, "grad_norm": 0.8987851738929749, "learning_rate": 0.00012385412621040726, "loss": 0.0702, "step": 30430 }, { "epoch": 2.1121993043493603e-13, "grad_norm": 0.9117500185966492, "learning_rate": 0.00012383702204945874, "loss": 0.0658, "step": 30440 }, { "epoch": 2.112893193739751e-13, "grad_norm": 0.9283738136291504, "learning_rate": 0.00012381991364048803, "loss": 0.0676, "step": 30450 }, { "epoch": 2.1135870831301418e-13, "grad_norm": 0.9043581485748291, "learning_rate": 0.00012380280098525325, "loss": 0.0688, "step": 30460 }, { "epoch": 2.1142809725205325e-13, "grad_norm": 0.9922744631767273, "learning_rate": 0.00012378568408551304, "loss": 0.0696, "step": 30470 }, { "epoch": 2.1149748619109232e-13, "grad_norm": 1.0182747840881348, "learning_rate": 0.00012376856294302643, "loss": 0.0697, "step": 30480 }, { "epoch": 2.115668751301314e-13, "grad_norm": 1.1053394079208374, "learning_rate": 0.00012375143755955284, "loss": 0.0734, "step": 30490 }, { "epoch": 2.1163626406917047e-13, "grad_norm": 1.1685916185379028, "learning_rate": 0.00012373430793685222, "loss": 0.0696, "step": 30500 }, { "epoch": 2.1170565300820954e-13, "grad_norm": 0.9207883477210999, "learning_rate": 0.00012371717407668486, "loss": 0.0741, "step": 30510 }, { "epoch": 2.117750419472486e-13, "grad_norm": 0.9578773379325867, "learning_rate": 0.00012370003598081156, "loss": 0.0771, "step": 30520 }, { "epoch": 2.1184443088628768e-13, "grad_norm": 1.143919587135315, "learning_rate": 0.0001236828936509935, "loss": 0.0667, "step": 30530 }, { "epoch": 2.1191381982532675e-13, "grad_norm": 1.103873610496521, "learning_rate": 0.00012366574708899238, "loss": 0.0659, "step": 30540 }, { "epoch": 2.1198320876436583e-13, "grad_norm": 0.971693754196167, "learning_rate": 0.00012364859629657022, "loss": 0.0686, "step": 30550 }, { "epoch": 2.120525977034049e-13, "grad_norm": 1.0193068981170654, "learning_rate": 0.00012363144127548956, "loss": 0.0673, "step": 30560 }, { "epoch": 2.1212198664244397e-13, "grad_norm": 1.0101701021194458, "learning_rate": 0.0001236142820275133, "loss": 0.0618, "step": 30570 }, { "epoch": 2.1219137558148304e-13, "grad_norm": 1.1403809785842896, "learning_rate": 0.00012359711855440488, "loss": 0.0663, "step": 30580 }, { "epoch": 2.1226076452052212e-13, "grad_norm": 0.8532542586326599, "learning_rate": 0.00012357995085792806, "loss": 0.0645, "step": 30590 }, { "epoch": 2.123301534595612e-13, "grad_norm": 0.9029741287231445, "learning_rate": 0.00012356277893984713, "loss": 0.0619, "step": 30600 }, { "epoch": 2.1239954239860026e-13, "grad_norm": 0.9738597273826599, "learning_rate": 0.00012354560280192675, "loss": 0.065, "step": 30610 }, { "epoch": 2.1246893133763933e-13, "grad_norm": 0.9247140288352966, "learning_rate": 0.00012352842244593202, "loss": 0.0669, "step": 30620 }, { "epoch": 2.125383202766784e-13, "grad_norm": 0.9877589344978333, "learning_rate": 0.00012351123787362854, "loss": 0.0693, "step": 30630 }, { "epoch": 2.1260770921571748e-13, "grad_norm": 0.9547553658485413, "learning_rate": 0.00012349404908678225, "loss": 0.0702, "step": 30640 }, { "epoch": 2.1267709815475655e-13, "grad_norm": 0.9458938837051392, "learning_rate": 0.00012347685608715957, "loss": 0.0673, "step": 30650 }, { "epoch": 2.1274648709379562e-13, "grad_norm": 1.0359187126159668, "learning_rate": 0.0001234596588765274, "loss": 0.0718, "step": 30660 }, { "epoch": 2.128158760328347e-13, "grad_norm": 1.0015939474105835, "learning_rate": 0.0001234424574566529, "loss": 0.0732, "step": 30670 }, { "epoch": 2.1288526497187377e-13, "grad_norm": 0.8490175604820251, "learning_rate": 0.00012342525182930392, "loss": 0.0761, "step": 30680 }, { "epoch": 2.1295465391091284e-13, "grad_norm": 0.8180879950523376, "learning_rate": 0.00012340804199624853, "loss": 0.0653, "step": 30690 }, { "epoch": 2.130240428499519e-13, "grad_norm": 0.9810343980789185, "learning_rate": 0.00012339082795925532, "loss": 0.0685, "step": 30700 }, { "epoch": 2.1309343178899098e-13, "grad_norm": 0.842278778553009, "learning_rate": 0.00012337360972009332, "loss": 0.0686, "step": 30710 }, { "epoch": 2.1316282072803006e-13, "grad_norm": 0.8127549290657043, "learning_rate": 0.00012335638728053195, "loss": 0.0647, "step": 30720 }, { "epoch": 2.1323220966706913e-13, "grad_norm": 0.9138376712799072, "learning_rate": 0.00012333916064234107, "loss": 0.0641, "step": 30730 }, { "epoch": 2.133015986061082e-13, "grad_norm": 1.0593737363815308, "learning_rate": 0.000123321929807291, "loss": 0.0659, "step": 30740 }, { "epoch": 2.1337098754514727e-13, "grad_norm": 0.887953519821167, "learning_rate": 0.00012330469477715253, "loss": 0.0633, "step": 30750 }, { "epoch": 2.1344037648418634e-13, "grad_norm": 1.077784538269043, "learning_rate": 0.00012328745555369674, "loss": 0.0614, "step": 30760 }, { "epoch": 2.1350976542322542e-13, "grad_norm": 0.9626303911209106, "learning_rate": 0.00012327021213869526, "loss": 0.0659, "step": 30770 }, { "epoch": 2.135791543622645e-13, "grad_norm": 1.030250906944275, "learning_rate": 0.00012325296453392012, "loss": 0.0617, "step": 30780 }, { "epoch": 2.1364854330130356e-13, "grad_norm": 0.9383456110954285, "learning_rate": 0.00012323571274114382, "loss": 0.0669, "step": 30790 }, { "epoch": 2.1371793224034263e-13, "grad_norm": 0.9171403646469116, "learning_rate": 0.00012321845676213918, "loss": 0.0714, "step": 30800 }, { "epoch": 2.137873211793817e-13, "grad_norm": 0.9042221903800964, "learning_rate": 0.0001232011965986796, "loss": 0.0645, "step": 30810 }, { "epoch": 2.1385671011842078e-13, "grad_norm": 0.952716052532196, "learning_rate": 0.00012318393225253877, "loss": 0.0735, "step": 30820 }, { "epoch": 2.1392609905745985e-13, "grad_norm": 1.1249825954437256, "learning_rate": 0.00012316666372549087, "loss": 0.0759, "step": 30830 }, { "epoch": 2.1399548799649892e-13, "grad_norm": 0.9223648309707642, "learning_rate": 0.00012314939101931054, "loss": 0.0698, "step": 30840 }, { "epoch": 2.14064876935538e-13, "grad_norm": 0.8844447135925293, "learning_rate": 0.0001231321141357728, "loss": 0.0636, "step": 30850 }, { "epoch": 2.1413426587457707e-13, "grad_norm": 1.092519760131836, "learning_rate": 0.0001231148330766531, "loss": 0.0659, "step": 30860 }, { "epoch": 2.1420365481361614e-13, "grad_norm": 1.0214380025863647, "learning_rate": 0.0001230975478437274, "loss": 0.0658, "step": 30870 }, { "epoch": 2.142730437526552e-13, "grad_norm": 0.9682969450950623, "learning_rate": 0.00012308025843877198, "loss": 0.0646, "step": 30880 }, { "epoch": 2.1434243269169428e-13, "grad_norm": 0.8160616755485535, "learning_rate": 0.00012306296486356365, "loss": 0.0628, "step": 30890 }, { "epoch": 2.1441182163073336e-13, "grad_norm": 1.0742954015731812, "learning_rate": 0.00012304566711987948, "loss": 0.0697, "step": 30900 }, { "epoch": 2.1448121056977243e-13, "grad_norm": 0.9230682253837585, "learning_rate": 0.00012302836520949715, "loss": 0.06, "step": 30910 }, { "epoch": 2.145505995088115e-13, "grad_norm": 0.7904887199401855, "learning_rate": 0.00012301105913419474, "loss": 0.0619, "step": 30920 }, { "epoch": 2.1461998844785057e-13, "grad_norm": 0.8881224393844604, "learning_rate": 0.00012299374889575068, "loss": 0.0634, "step": 30930 }, { "epoch": 2.1468937738688965e-13, "grad_norm": 0.9317581653594971, "learning_rate": 0.00012297643449594385, "loss": 0.0642, "step": 30940 }, { "epoch": 2.1475876632592872e-13, "grad_norm": 0.8784576058387756, "learning_rate": 0.00012295911593655358, "loss": 0.0648, "step": 30950 }, { "epoch": 2.148281552649678e-13, "grad_norm": 0.9054245948791504, "learning_rate": 0.00012294179321935964, "loss": 0.0689, "step": 30960 }, { "epoch": 2.1489754420400686e-13, "grad_norm": 0.9950438141822815, "learning_rate": 0.00012292446634614218, "loss": 0.0648, "step": 30970 }, { "epoch": 2.1496693314304594e-13, "grad_norm": 1.0597310066223145, "learning_rate": 0.00012290713531868186, "loss": 0.0702, "step": 30980 }, { "epoch": 2.15036322082085e-13, "grad_norm": 0.9893236756324768, "learning_rate": 0.00012288980013875966, "loss": 0.0787, "step": 30990 }, { "epoch": 2.1510571102112408e-13, "grad_norm": 0.9546467065811157, "learning_rate": 0.00012287246080815707, "loss": 0.0677, "step": 31000 }, { "epoch": 2.1517509996016315e-13, "grad_norm": 0.9495257139205933, "learning_rate": 0.00012285511732865596, "loss": 0.0641, "step": 31010 }, { "epoch": 2.1524448889920222e-13, "grad_norm": 0.9813787341117859, "learning_rate": 0.00012283776970203864, "loss": 0.0656, "step": 31020 }, { "epoch": 2.153138778382413e-13, "grad_norm": 1.004181981086731, "learning_rate": 0.00012282041793008786, "loss": 0.0643, "step": 31030 }, { "epoch": 2.1538326677728037e-13, "grad_norm": 0.9725863933563232, "learning_rate": 0.00012280306201458677, "loss": 0.0613, "step": 31040 }, { "epoch": 2.1545265571631944e-13, "grad_norm": 0.9274056553840637, "learning_rate": 0.000122785701957319, "loss": 0.0652, "step": 31050 }, { "epoch": 2.1552204465535851e-13, "grad_norm": 0.9676752090454102, "learning_rate": 0.00012276833776006852, "loss": 0.0645, "step": 31060 }, { "epoch": 2.1559143359439759e-13, "grad_norm": 0.8532295227050781, "learning_rate": 0.00012275096942461976, "loss": 0.0601, "step": 31070 }, { "epoch": 2.1566082253343666e-13, "grad_norm": 0.8819594979286194, "learning_rate": 0.00012273359695275766, "loss": 0.0622, "step": 31080 }, { "epoch": 2.1573021147247573e-13, "grad_norm": 0.9283459186553955, "learning_rate": 0.00012271622034626742, "loss": 0.063, "step": 31090 }, { "epoch": 2.157996004115148e-13, "grad_norm": 0.8965410590171814, "learning_rate": 0.00012269883960693484, "loss": 0.0636, "step": 31100 }, { "epoch": 2.1586898935055387e-13, "grad_norm": 0.9056936502456665, "learning_rate": 0.000122681454736546, "loss": 0.066, "step": 31110 }, { "epoch": 2.1593837828959295e-13, "grad_norm": 0.89704430103302, "learning_rate": 0.0001226640657368875, "loss": 0.0617, "step": 31120 }, { "epoch": 2.1600776722863202e-13, "grad_norm": 1.0746431350708008, "learning_rate": 0.00012264667260974632, "loss": 0.0672, "step": 31130 }, { "epoch": 2.160771561676711e-13, "grad_norm": 1.1745976209640503, "learning_rate": 0.00012262927535690988, "loss": 0.0701, "step": 31140 }, { "epoch": 2.1614654510671016e-13, "grad_norm": 1.0917390584945679, "learning_rate": 0.000122611873980166, "loss": 0.074, "step": 31150 }, { "epoch": 2.1621593404574924e-13, "grad_norm": 1.1608855724334717, "learning_rate": 0.00012259446848130294, "loss": 0.0655, "step": 31160 }, { "epoch": 2.162853229847883e-13, "grad_norm": 0.7998027205467224, "learning_rate": 0.00012257705886210946, "loss": 0.0591, "step": 31170 }, { "epoch": 2.1635471192382738e-13, "grad_norm": 0.7826114892959595, "learning_rate": 0.00012255964512437455, "loss": 0.0639, "step": 31180 }, { "epoch": 2.1642410086286645e-13, "grad_norm": 0.9944005012512207, "learning_rate": 0.00012254222726988784, "loss": 0.0623, "step": 31190 }, { "epoch": 2.1649348980190553e-13, "grad_norm": 0.9674311876296997, "learning_rate": 0.00012252480530043923, "loss": 0.0633, "step": 31200 }, { "epoch": 2.165628787409446e-13, "grad_norm": 1.0772919654846191, "learning_rate": 0.00012250737921781912, "loss": 0.0651, "step": 31210 }, { "epoch": 2.1663226767998367e-13, "grad_norm": 0.9341673851013184, "learning_rate": 0.0001224899490238183, "loss": 0.0616, "step": 31220 }, { "epoch": 2.1670165661902274e-13, "grad_norm": 1.0320885181427002, "learning_rate": 0.00012247251472022802, "loss": 0.0626, "step": 31230 }, { "epoch": 2.1677104555806181e-13, "grad_norm": 0.8834547400474548, "learning_rate": 0.0001224550763088399, "loss": 0.0614, "step": 31240 }, { "epoch": 2.1684043449710089e-13, "grad_norm": 0.9897778034210205, "learning_rate": 0.000122437633791446, "loss": 0.0603, "step": 31250 }, { "epoch": 2.1690982343613996e-13, "grad_norm": 0.9501094818115234, "learning_rate": 0.00012242018716983883, "loss": 0.0631, "step": 31260 }, { "epoch": 2.1697921237517903e-13, "grad_norm": 0.871131956577301, "learning_rate": 0.00012240273644581134, "loss": 0.065, "step": 31270 }, { "epoch": 2.170486013142181e-13, "grad_norm": 0.9002489447593689, "learning_rate": 0.0001223852816211568, "loss": 0.0611, "step": 31280 }, { "epoch": 2.1711799025325718e-13, "grad_norm": 0.8984865546226501, "learning_rate": 0.00012236782269766896, "loss": 0.0665, "step": 31290 }, { "epoch": 2.1718737919229625e-13, "grad_norm": 0.9777487516403198, "learning_rate": 0.00012235035967714204, "loss": 0.0725, "step": 31300 }, { "epoch": 2.1725676813133532e-13, "grad_norm": 0.8326346278190613, "learning_rate": 0.0001223328925613706, "loss": 0.0637, "step": 31310 }, { "epoch": 2.173261570703744e-13, "grad_norm": 0.8991476893424988, "learning_rate": 0.00012231542135214971, "loss": 0.0623, "step": 31320 }, { "epoch": 2.1739554600941347e-13, "grad_norm": 1.0750749111175537, "learning_rate": 0.00012229794605127475, "loss": 0.0631, "step": 31330 }, { "epoch": 2.1746493494845254e-13, "grad_norm": 0.9145813584327698, "learning_rate": 0.0001222804666605416, "loss": 0.0596, "step": 31340 }, { "epoch": 2.175343238874916e-13, "grad_norm": 0.8849271535873413, "learning_rate": 0.00012226298318174656, "loss": 0.0596, "step": 31350 }, { "epoch": 2.1760371282653068e-13, "grad_norm": 0.9899057149887085, "learning_rate": 0.00012224549561668633, "loss": 0.0601, "step": 31360 }, { "epoch": 2.1767310176556975e-13, "grad_norm": 0.8865023851394653, "learning_rate": 0.00012222800396715798, "loss": 0.0674, "step": 31370 }, { "epoch": 2.1774249070460883e-13, "grad_norm": 0.8466919660568237, "learning_rate": 0.0001222105082349591, "loss": 0.0596, "step": 31380 }, { "epoch": 2.178118796436479e-13, "grad_norm": 0.8054666519165039, "learning_rate": 0.0001221930084218876, "loss": 0.06, "step": 31390 }, { "epoch": 2.1788126858268697e-13, "grad_norm": 0.7952196598052979, "learning_rate": 0.00012217550452974193, "loss": 0.0615, "step": 31400 }, { "epoch": 2.1795065752172604e-13, "grad_norm": 0.9754857420921326, "learning_rate": 0.00012215799656032083, "loss": 0.0623, "step": 31410 }, { "epoch": 2.1802004646076512e-13, "grad_norm": 1.080552101135254, "learning_rate": 0.00012214048451542354, "loss": 0.0644, "step": 31420 }, { "epoch": 2.180894353998042e-13, "grad_norm": 0.811520516872406, "learning_rate": 0.00012212296839684967, "loss": 0.0618, "step": 31430 }, { "epoch": 2.1815882433884326e-13, "grad_norm": 0.8741259574890137, "learning_rate": 0.00012210544820639928, "loss": 0.0615, "step": 31440 }, { "epoch": 2.1822821327788233e-13, "grad_norm": 0.9098253846168518, "learning_rate": 0.00012208792394587288, "loss": 0.0635, "step": 31450 }, { "epoch": 2.182976022169214e-13, "grad_norm": 0.9780353903770447, "learning_rate": 0.00012207039561707133, "loss": 0.0702, "step": 31460 }, { "epoch": 2.1836699115596048e-13, "grad_norm": 0.8312398195266724, "learning_rate": 0.00012205286322179593, "loss": 0.0611, "step": 31470 }, { "epoch": 2.1843638009499955e-13, "grad_norm": 0.8262131214141846, "learning_rate": 0.00012203532676184843, "loss": 0.0615, "step": 31480 }, { "epoch": 2.1850576903403862e-13, "grad_norm": 0.9218004941940308, "learning_rate": 0.00012201778623903097, "loss": 0.0635, "step": 31490 }, { "epoch": 2.185751579730777e-13, "grad_norm": 0.8211460709571838, "learning_rate": 0.00012200024165514609, "loss": 0.0579, "step": 31500 }, { "epoch": 2.1864454691211677e-13, "grad_norm": 0.9557305574417114, "learning_rate": 0.00012198269301199677, "loss": 0.0575, "step": 31510 }, { "epoch": 2.1871393585115584e-13, "grad_norm": 0.8286266326904297, "learning_rate": 0.00012196514031138645, "loss": 0.0606, "step": 31520 }, { "epoch": 2.187833247901949e-13, "grad_norm": 0.8384743928909302, "learning_rate": 0.0001219475835551189, "loss": 0.0618, "step": 31530 }, { "epoch": 2.1885271372923398e-13, "grad_norm": 0.7692549824714661, "learning_rate": 0.00012193002274499836, "loss": 0.0579, "step": 31540 }, { "epoch": 2.1892210266827306e-13, "grad_norm": 0.9902336597442627, "learning_rate": 0.00012191245788282948, "loss": 0.0622, "step": 31550 }, { "epoch": 2.1899149160731213e-13, "grad_norm": 0.8377128839492798, "learning_rate": 0.00012189488897041732, "loss": 0.0591, "step": 31560 }, { "epoch": 2.190608805463512e-13, "grad_norm": 0.9642511606216431, "learning_rate": 0.00012187731600956739, "loss": 0.0614, "step": 31570 }, { "epoch": 2.1913026948539027e-13, "grad_norm": 0.9816409349441528, "learning_rate": 0.00012185973900208549, "loss": 0.0655, "step": 31580 }, { "epoch": 2.1919965842442934e-13, "grad_norm": 0.9828884601593018, "learning_rate": 0.00012184215794977804, "loss": 0.0597, "step": 31590 }, { "epoch": 2.1926904736346842e-13, "grad_norm": 1.0736877918243408, "learning_rate": 0.00012182457285445173, "loss": 0.065, "step": 31600 }, { "epoch": 2.193384363025075e-13, "grad_norm": 1.1571986675262451, "learning_rate": 0.00012180698371791368, "loss": 0.0639, "step": 31610 }, { "epoch": 2.1940782524154656e-13, "grad_norm": 0.9847063422203064, "learning_rate": 0.00012178939054197147, "loss": 0.064, "step": 31620 }, { "epoch": 2.1947721418058563e-13, "grad_norm": 0.9167648553848267, "learning_rate": 0.00012177179332843307, "loss": 0.062, "step": 31630 }, { "epoch": 2.195466031196247e-13, "grad_norm": 0.9551918506622314, "learning_rate": 0.00012175419207910685, "loss": 0.0578, "step": 31640 }, { "epoch": 2.1961599205866378e-13, "grad_norm": 0.9280365109443665, "learning_rate": 0.00012173658679580162, "loss": 0.062, "step": 31650 }, { "epoch": 2.1968538099770285e-13, "grad_norm": 0.8590663075447083, "learning_rate": 0.00012171897748032662, "loss": 0.0578, "step": 31660 }, { "epoch": 2.1975476993674192e-13, "grad_norm": 0.7373614311218262, "learning_rate": 0.00012170136413449147, "loss": 0.0578, "step": 31670 }, { "epoch": 2.19824158875781e-13, "grad_norm": 0.8822330236434937, "learning_rate": 0.0001216837467601062, "loss": 0.0605, "step": 31680 }, { "epoch": 2.1989354781482007e-13, "grad_norm": 0.9522877931594849, "learning_rate": 0.00012166612535898128, "loss": 0.0599, "step": 31690 }, { "epoch": 2.1996293675385914e-13, "grad_norm": 0.862821102142334, "learning_rate": 0.00012164849993292756, "loss": 0.0598, "step": 31700 }, { "epoch": 2.200323256928982e-13, "grad_norm": 0.8487995266914368, "learning_rate": 0.00012163087048375637, "loss": 0.0585, "step": 31710 }, { "epoch": 2.2010171463193728e-13, "grad_norm": 0.8942941427230835, "learning_rate": 0.00012161323701327941, "loss": 0.0603, "step": 31720 }, { "epoch": 2.2017110357097636e-13, "grad_norm": 0.9895216226577759, "learning_rate": 0.00012159559952330871, "loss": 0.0606, "step": 31730 }, { "epoch": 2.2024049251001543e-13, "grad_norm": 0.9120442271232605, "learning_rate": 0.00012157795801565691, "loss": 0.0621, "step": 31740 }, { "epoch": 2.203098814490545e-13, "grad_norm": 0.9555367231369019, "learning_rate": 0.00012156031249213688, "loss": 0.0594, "step": 31750 }, { "epoch": 2.2037927038809357e-13, "grad_norm": 0.8100427985191345, "learning_rate": 0.00012154266295456198, "loss": 0.0629, "step": 31760 }, { "epoch": 2.2044865932713265e-13, "grad_norm": 0.8282209038734436, "learning_rate": 0.00012152500940474601, "loss": 0.0657, "step": 31770 }, { "epoch": 2.2051804826617172e-13, "grad_norm": 1.0049244165420532, "learning_rate": 0.0001215073518445031, "loss": 0.0612, "step": 31780 }, { "epoch": 2.205874372052108e-13, "grad_norm": 1.0380659103393555, "learning_rate": 0.00012148969027564787, "loss": 0.0593, "step": 31790 }, { "epoch": 2.2065682614424986e-13, "grad_norm": 0.9139519929885864, "learning_rate": 0.00012147202469999531, "loss": 0.0611, "step": 31800 }, { "epoch": 2.2072621508328893e-13, "grad_norm": 0.8646926879882812, "learning_rate": 0.00012145435511936083, "loss": 0.0584, "step": 31810 }, { "epoch": 2.20795604022328e-13, "grad_norm": 0.842173159122467, "learning_rate": 0.00012143668153556027, "loss": 0.0567, "step": 31820 }, { "epoch": 2.2086499296136708e-13, "grad_norm": 0.9776089191436768, "learning_rate": 0.00012141900395040983, "loss": 0.059, "step": 31830 }, { "epoch": 2.2093438190040615e-13, "grad_norm": 0.7847867012023926, "learning_rate": 0.00012140132236572621, "loss": 0.059, "step": 31840 }, { "epoch": 2.2100377083944522e-13, "grad_norm": 0.8957677483558655, "learning_rate": 0.00012138363678332641, "loss": 0.0588, "step": 31850 }, { "epoch": 2.210731597784843e-13, "grad_norm": 0.9745914936065674, "learning_rate": 0.00012136594720502794, "loss": 0.0575, "step": 31860 }, { "epoch": 2.2114254871752337e-13, "grad_norm": 0.9454705715179443, "learning_rate": 0.00012134825363264869, "loss": 0.0588, "step": 31870 }, { "epoch": 2.2121193765656244e-13, "grad_norm": 0.9020155072212219, "learning_rate": 0.00012133055606800689, "loss": 0.0606, "step": 31880 }, { "epoch": 2.2128132659560151e-13, "grad_norm": 0.9081616997718811, "learning_rate": 0.0001213128545129213, "loss": 0.0627, "step": 31890 }, { "epoch": 2.2135071553464059e-13, "grad_norm": 0.8368297219276428, "learning_rate": 0.000121295148969211, "loss": 0.0599, "step": 31900 }, { "epoch": 2.2142010447367966e-13, "grad_norm": 0.9004431366920471, "learning_rate": 0.00012127743943869551, "loss": 0.061, "step": 31910 }, { "epoch": 2.2148949341271873e-13, "grad_norm": 0.8656052350997925, "learning_rate": 0.00012125972592319479, "loss": 0.0623, "step": 31920 }, { "epoch": 2.215588823517578e-13, "grad_norm": 0.7804728150367737, "learning_rate": 0.00012124200842452913, "loss": 0.065, "step": 31930 }, { "epoch": 2.2162827129079687e-13, "grad_norm": 0.7853375673294067, "learning_rate": 0.00012122428694451934, "loss": 0.0586, "step": 31940 }, { "epoch": 2.2169766022983595e-13, "grad_norm": 0.9405984878540039, "learning_rate": 0.0001212065614849865, "loss": 0.0605, "step": 31950 }, { "epoch": 2.2176704916887502e-13, "grad_norm": 0.7739964723587036, "learning_rate": 0.00012118883204775225, "loss": 0.0604, "step": 31960 }, { "epoch": 2.218364381079141e-13, "grad_norm": 0.8170434236526489, "learning_rate": 0.00012117109863463851, "loss": 0.0559, "step": 31970 }, { "epoch": 2.2190582704695316e-13, "grad_norm": 0.8676373958587646, "learning_rate": 0.00012115336124746771, "loss": 0.0572, "step": 31980 }, { "epoch": 2.2197521598599224e-13, "grad_norm": 0.9147117733955383, "learning_rate": 0.00012113561988806263, "loss": 0.0581, "step": 31990 }, { "epoch": 2.220446049250313e-13, "grad_norm": 0.8506485223770142, "learning_rate": 0.00012111787455824645, "loss": 0.0579, "step": 32000 }, { "epoch": 2.2211399386407038e-13, "grad_norm": 0.9346557855606079, "learning_rate": 0.00012110012525984278, "loss": 0.0577, "step": 32010 }, { "epoch": 2.2218338280310945e-13, "grad_norm": 0.9041094779968262, "learning_rate": 0.00012108237199467567, "loss": 0.0592, "step": 32020 }, { "epoch": 2.2225277174214852e-13, "grad_norm": 0.9373618364334106, "learning_rate": 0.00012106461476456951, "loss": 0.0571, "step": 32030 }, { "epoch": 2.223221606811876e-13, "grad_norm": 0.8896515369415283, "learning_rate": 0.00012104685357134913, "loss": 0.0588, "step": 32040 }, { "epoch": 2.2239154962022667e-13, "grad_norm": 0.8727226853370667, "learning_rate": 0.0001210290884168398, "loss": 0.0636, "step": 32050 }, { "epoch": 2.2246093855926574e-13, "grad_norm": 0.8473594188690186, "learning_rate": 0.00012101131930286716, "loss": 0.057, "step": 32060 }, { "epoch": 2.2253032749830481e-13, "grad_norm": 0.8808810114860535, "learning_rate": 0.00012099354623125723, "loss": 0.0632, "step": 32070 }, { "epoch": 2.2259971643734389e-13, "grad_norm": 0.9887738227844238, "learning_rate": 0.00012097576920383652, "loss": 0.0637, "step": 32080 }, { "epoch": 2.2266910537638296e-13, "grad_norm": 0.8962635397911072, "learning_rate": 0.00012095798822243184, "loss": 0.0588, "step": 32090 }, { "epoch": 2.2273849431542203e-13, "grad_norm": 0.8695021271705627, "learning_rate": 0.0001209402032888705, "loss": 0.057, "step": 32100 }, { "epoch": 2.228078832544611e-13, "grad_norm": 0.9641795754432678, "learning_rate": 0.00012092241440498017, "loss": 0.0588, "step": 32110 }, { "epoch": 2.2287727219350018e-13, "grad_norm": 0.9722954034805298, "learning_rate": 0.00012090462157258892, "loss": 0.0594, "step": 32120 }, { "epoch": 2.2294666113253925e-13, "grad_norm": 0.9812174439430237, "learning_rate": 0.00012088682479352529, "loss": 0.0558, "step": 32130 }, { "epoch": 2.2301605007157832e-13, "grad_norm": 0.8779757022857666, "learning_rate": 0.0001208690240696181, "loss": 0.0554, "step": 32140 }, { "epoch": 2.230854390106174e-13, "grad_norm": 1.0865910053253174, "learning_rate": 0.0001208512194026967, "loss": 0.0621, "step": 32150 }, { "epoch": 2.2315482794965646e-13, "grad_norm": 0.9502326250076294, "learning_rate": 0.00012083341079459081, "loss": 0.0549, "step": 32160 }, { "epoch": 2.2322421688869554e-13, "grad_norm": 0.7340664267539978, "learning_rate": 0.0001208155982471305, "loss": 0.0561, "step": 32170 }, { "epoch": 2.232936058277346e-13, "grad_norm": 0.8202730417251587, "learning_rate": 0.00012079778176214633, "loss": 0.0583, "step": 32180 }, { "epoch": 2.2336299476677368e-13, "grad_norm": 0.7857391834259033, "learning_rate": 0.00012077996134146915, "loss": 0.0588, "step": 32190 }, { "epoch": 2.2343238370581275e-13, "grad_norm": 0.8479932546615601, "learning_rate": 0.00012076213698693036, "loss": 0.0582, "step": 32200 }, { "epoch": 2.2350177264485183e-13, "grad_norm": 0.8472136855125427, "learning_rate": 0.00012074430870036167, "loss": 0.0612, "step": 32210 }, { "epoch": 2.235711615838909e-13, "grad_norm": 0.9639440774917603, "learning_rate": 0.00012072647648359518, "loss": 0.0575, "step": 32220 }, { "epoch": 2.2364055052292997e-13, "grad_norm": 0.9755998253822327, "learning_rate": 0.00012070864033846347, "loss": 0.0593, "step": 32230 }, { "epoch": 2.2370993946196904e-13, "grad_norm": 0.8721300363540649, "learning_rate": 0.00012069080026679947, "loss": 0.0643, "step": 32240 }, { "epoch": 2.2377932840100812e-13, "grad_norm": 0.9097620844841003, "learning_rate": 0.00012067295627043652, "loss": 0.0575, "step": 32250 }, { "epoch": 2.238487173400472e-13, "grad_norm": 0.9070300459861755, "learning_rate": 0.00012065510835120837, "loss": 0.0569, "step": 32260 }, { "epoch": 2.2391810627908626e-13, "grad_norm": 0.940555989742279, "learning_rate": 0.00012063725651094914, "loss": 0.0593, "step": 32270 }, { "epoch": 2.2398749521812533e-13, "grad_norm": 0.964974582195282, "learning_rate": 0.00012061940075149346, "loss": 0.0597, "step": 32280 }, { "epoch": 2.240568841571644e-13, "grad_norm": 0.9413407444953918, "learning_rate": 0.0001206015410746762, "loss": 0.0547, "step": 32290 }, { "epoch": 2.2412627309620348e-13, "grad_norm": 0.9356335997581482, "learning_rate": 0.00012058367748233279, "loss": 0.0581, "step": 32300 }, { "epoch": 2.2419566203524255e-13, "grad_norm": 0.9269062280654907, "learning_rate": 0.00012056580997629893, "loss": 0.0576, "step": 32310 }, { "epoch": 2.2426505097428162e-13, "grad_norm": 0.8451566100120544, "learning_rate": 0.00012054793855841084, "loss": 0.0545, "step": 32320 }, { "epoch": 2.243344399133207e-13, "grad_norm": 0.8841183185577393, "learning_rate": 0.00012053006323050507, "loss": 0.0565, "step": 32330 }, { "epoch": 2.2440382885235977e-13, "grad_norm": 0.8927401900291443, "learning_rate": 0.00012051218399441859, "loss": 0.0578, "step": 32340 }, { "epoch": 2.2447321779139884e-13, "grad_norm": 0.8277440071105957, "learning_rate": 0.00012049430085198872, "loss": 0.0584, "step": 32350 }, { "epoch": 2.245426067304379e-13, "grad_norm": 0.9166984558105469, "learning_rate": 0.00012047641380505331, "loss": 0.0608, "step": 32360 }, { "epoch": 2.2461199566947698e-13, "grad_norm": 0.8640173673629761, "learning_rate": 0.00012045852285545047, "loss": 0.0562, "step": 32370 }, { "epoch": 2.2468138460851605e-13, "grad_norm": 0.9776782989501953, "learning_rate": 0.0001204406280050188, "loss": 0.0603, "step": 32380 }, { "epoch": 2.2475077354755513e-13, "grad_norm": 1.0738216638565063, "learning_rate": 0.0001204227292555973, "loss": 0.0611, "step": 32390 }, { "epoch": 2.248201624865942e-13, "grad_norm": 0.9062270522117615, "learning_rate": 0.0001204048266090253, "loss": 0.0613, "step": 32400 }, { "epoch": 2.2488955142563327e-13, "grad_norm": 1.0578653812408447, "learning_rate": 0.00012038692006714257, "loss": 0.0564, "step": 32410 }, { "epoch": 2.2495894036467234e-13, "grad_norm": 0.8032293915748596, "learning_rate": 0.00012036900963178934, "loss": 0.0534, "step": 32420 }, { "epoch": 2.2502832930371142e-13, "grad_norm": 0.7608926892280579, "learning_rate": 0.00012035109530480615, "loss": 0.0578, "step": 32430 }, { "epoch": 2.250977182427505e-13, "grad_norm": 0.8688946962356567, "learning_rate": 0.00012033317708803395, "loss": 0.0577, "step": 32440 }, { "epoch": 2.2516710718178956e-13, "grad_norm": 0.8665933012962341, "learning_rate": 0.00012031525498331417, "loss": 0.0571, "step": 32450 }, { "epoch": 2.2523649612082863e-13, "grad_norm": 0.9333743453025818, "learning_rate": 0.00012029732899248856, "loss": 0.0577, "step": 32460 }, { "epoch": 2.253058850598677e-13, "grad_norm": 0.8665225505828857, "learning_rate": 0.0001202793991173993, "loss": 0.0557, "step": 32470 }, { "epoch": 2.2537527399890678e-13, "grad_norm": 0.9619240164756775, "learning_rate": 0.00012026146535988896, "loss": 0.0565, "step": 32480 }, { "epoch": 2.2544466293794585e-13, "grad_norm": 0.8643956780433655, "learning_rate": 0.00012024352772180051, "loss": 0.0568, "step": 32490 }, { "epoch": 2.2551405187698492e-13, "grad_norm": 0.9519940614700317, "learning_rate": 0.00012022558620497731, "loss": 0.0563, "step": 32500 }, { "epoch": 2.25583440816024e-13, "grad_norm": 0.9452416300773621, "learning_rate": 0.00012020764081126317, "loss": 0.0561, "step": 32510 }, { "epoch": 2.2565282975506307e-13, "grad_norm": 0.8268569111824036, "learning_rate": 0.0001201896915425022, "loss": 0.0597, "step": 32520 }, { "epoch": 2.2572221869410214e-13, "grad_norm": 0.8820160031318665, "learning_rate": 0.00012017173840053901, "loss": 0.055, "step": 32530 }, { "epoch": 2.257916076331412e-13, "grad_norm": 0.8260784149169922, "learning_rate": 0.00012015378138721856, "loss": 0.0604, "step": 32540 }, { "epoch": 2.2586099657218028e-13, "grad_norm": 0.8830448389053345, "learning_rate": 0.00012013582050438621, "loss": 0.063, "step": 32550 }, { "epoch": 2.2593038551121936e-13, "grad_norm": 0.7895841598510742, "learning_rate": 0.0001201178557538877, "loss": 0.0562, "step": 32560 }, { "epoch": 2.2599977445025843e-13, "grad_norm": 0.8509309887886047, "learning_rate": 0.00012009988713756922, "loss": 0.0545, "step": 32570 }, { "epoch": 2.260691633892975e-13, "grad_norm": 1.0049644708633423, "learning_rate": 0.00012008191465727729, "loss": 0.0561, "step": 32580 }, { "epoch": 2.2613855232833657e-13, "grad_norm": 0.8630942702293396, "learning_rate": 0.00012006393831485891, "loss": 0.054, "step": 32590 }, { "epoch": 2.2620794126737565e-13, "grad_norm": 0.7944074273109436, "learning_rate": 0.00012004595811216137, "loss": 0.054, "step": 32600 }, { "epoch": 2.2627733020641472e-13, "grad_norm": 0.8861646056175232, "learning_rate": 0.00012002797405103244, "loss": 0.0557, "step": 32610 }, { "epoch": 2.263467191454538e-13, "grad_norm": 0.772408664226532, "learning_rate": 0.00012000998613332027, "loss": 0.0595, "step": 32620 }, { "epoch": 2.2641610808449286e-13, "grad_norm": 0.8006537556648254, "learning_rate": 0.00011999199436087342, "loss": 0.0538, "step": 32630 }, { "epoch": 2.2648549702353193e-13, "grad_norm": 0.7718137502670288, "learning_rate": 0.0001199739987355408, "loss": 0.0547, "step": 32640 }, { "epoch": 2.26554885962571e-13, "grad_norm": 0.793136477470398, "learning_rate": 0.00011995599925917173, "loss": 0.0574, "step": 32650 }, { "epoch": 2.2662427490161008e-13, "grad_norm": 0.9185860753059387, "learning_rate": 0.00011993799593361593, "loss": 0.0589, "step": 32660 }, { "epoch": 2.2669366384064915e-13, "grad_norm": 1.0525367259979248, "learning_rate": 0.00011991998876072354, "loss": 0.0578, "step": 32670 }, { "epoch": 2.2676305277968822e-13, "grad_norm": 0.744719922542572, "learning_rate": 0.0001199019777423451, "loss": 0.0568, "step": 32680 }, { "epoch": 2.268324417187273e-13, "grad_norm": 0.8280267715454102, "learning_rate": 0.00011988396288033146, "loss": 0.0557, "step": 32690 }, { "epoch": 2.2690183065776637e-13, "grad_norm": 0.8526077270507812, "learning_rate": 0.00011986594417653399, "loss": 0.0581, "step": 32700 }, { "epoch": 2.2697121959680544e-13, "grad_norm": 0.8833545446395874, "learning_rate": 0.00011984792163280437, "loss": 0.0622, "step": 32710 }, { "epoch": 2.270406085358445e-13, "grad_norm": 0.8029620051383972, "learning_rate": 0.0001198298952509947, "loss": 0.0557, "step": 32720 }, { "epoch": 2.2710999747488358e-13, "grad_norm": 0.783352255821228, "learning_rate": 0.00011981186503295745, "loss": 0.0537, "step": 32730 }, { "epoch": 2.2717938641392266e-13, "grad_norm": 0.872044563293457, "learning_rate": 0.00011979383098054553, "loss": 0.0565, "step": 32740 }, { "epoch": 2.2724877535296173e-13, "grad_norm": 0.8017037510871887, "learning_rate": 0.00011977579309561218, "loss": 0.0525, "step": 32750 }, { "epoch": 2.273181642920008e-13, "grad_norm": 0.9216960072517395, "learning_rate": 0.00011975775138001113, "loss": 0.052, "step": 32760 }, { "epoch": 2.2738755323103987e-13, "grad_norm": 0.8130645155906677, "learning_rate": 0.00011973970583559642, "loss": 0.0545, "step": 32770 }, { "epoch": 2.2745694217007895e-13, "grad_norm": 0.8011841773986816, "learning_rate": 0.00011972165646422251, "loss": 0.0565, "step": 32780 }, { "epoch": 2.27526331109118e-13, "grad_norm": 0.8037225604057312, "learning_rate": 0.00011970360326774426, "loss": 0.054, "step": 32790 }, { "epoch": 2.275957200481571e-13, "grad_norm": 1.0155341625213623, "learning_rate": 0.00011968554624801692, "loss": 0.0567, "step": 32800 }, { "epoch": 2.2766510898719616e-13, "grad_norm": 0.8316033482551575, "learning_rate": 0.00011966748540689609, "loss": 0.0549, "step": 32810 }, { "epoch": 2.2773449792623524e-13, "grad_norm": 0.9134392142295837, "learning_rate": 0.00011964942074623786, "loss": 0.0581, "step": 32820 }, { "epoch": 2.278038868652743e-13, "grad_norm": 0.9193828105926514, "learning_rate": 0.00011963135226789861, "loss": 0.0584, "step": 32830 }, { "epoch": 2.278732758043134e-13, "grad_norm": 0.8517521619796753, "learning_rate": 0.0001196132799737352, "loss": 0.0531, "step": 32840 }, { "epoch": 2.2794266474335245e-13, "grad_norm": 0.9584771990776062, "learning_rate": 0.0001195952038656048, "loss": 0.0584, "step": 32850 }, { "epoch": 2.280120536823915e-13, "grad_norm": 0.9899454116821289, "learning_rate": 0.00011957712394536502, "loss": 0.0575, "step": 32860 }, { "epoch": 2.280814426214306e-13, "grad_norm": 0.8847069144248962, "learning_rate": 0.00011955904021487388, "loss": 0.058, "step": 32870 }, { "epoch": 2.2815083156046967e-13, "grad_norm": 0.7853360772132874, "learning_rate": 0.00011954095267598974, "loss": 0.0554, "step": 32880 }, { "epoch": 2.2822022049950874e-13, "grad_norm": 0.8393681049346924, "learning_rate": 0.00011952286133057137, "loss": 0.0507, "step": 32890 }, { "epoch": 2.282896094385478e-13, "grad_norm": 0.8686476945877075, "learning_rate": 0.00011950476618047796, "loss": 0.0552, "step": 32900 }, { "epoch": 2.283589983775869e-13, "grad_norm": 0.8370986580848694, "learning_rate": 0.00011948666722756905, "loss": 0.0523, "step": 32910 }, { "epoch": 2.2842838731662596e-13, "grad_norm": 0.7509298324584961, "learning_rate": 0.00011946856447370458, "loss": 0.0525, "step": 32920 }, { "epoch": 2.2849777625566503e-13, "grad_norm": 0.9517870545387268, "learning_rate": 0.00011945045792074493, "loss": 0.0555, "step": 32930 }, { "epoch": 2.285671651947041e-13, "grad_norm": 0.8765607476234436, "learning_rate": 0.0001194323475705508, "loss": 0.0541, "step": 32940 }, { "epoch": 2.286365541337432e-13, "grad_norm": 0.8124650716781616, "learning_rate": 0.0001194142334249833, "loss": 0.0555, "step": 32950 }, { "epoch": 2.2870594307278225e-13, "grad_norm": 0.8448967933654785, "learning_rate": 0.00011939611548590395, "loss": 0.0535, "step": 32960 }, { "epoch": 2.287753320118213e-13, "grad_norm": 0.8986851572990417, "learning_rate": 0.00011937799375517467, "loss": 0.0557, "step": 32970 }, { "epoch": 2.288447209508604e-13, "grad_norm": 0.9789203405380249, "learning_rate": 0.00011935986823465775, "loss": 0.0565, "step": 32980 }, { "epoch": 2.2891410988989946e-13, "grad_norm": 0.8694853186607361, "learning_rate": 0.00011934173892621586, "loss": 0.0579, "step": 32990 }, { "epoch": 2.2898349882893854e-13, "grad_norm": 0.896988034248352, "learning_rate": 0.00011932360583171202, "loss": 0.0537, "step": 33000 }, { "epoch": 2.290528877679776e-13, "grad_norm": 0.7970523834228516, "learning_rate": 0.00011930546895300976, "loss": 0.0564, "step": 33010 }, { "epoch": 2.291222767070167e-13, "grad_norm": 0.8457378149032593, "learning_rate": 0.0001192873282919729, "loss": 0.0606, "step": 33020 }, { "epoch": 2.2919166564605575e-13, "grad_norm": 0.9351410865783691, "learning_rate": 0.00011926918385046564, "loss": 0.0566, "step": 33030 }, { "epoch": 2.2926105458509483e-13, "grad_norm": 0.8886727094650269, "learning_rate": 0.00011925103563035267, "loss": 0.0535, "step": 33040 }, { "epoch": 2.293304435241339e-13, "grad_norm": 0.8005928993225098, "learning_rate": 0.00011923288363349897, "loss": 0.0538, "step": 33050 }, { "epoch": 2.2939983246317297e-13, "grad_norm": 0.8220956325531006, "learning_rate": 0.00011921472786176992, "loss": 0.0527, "step": 33060 }, { "epoch": 2.2946922140221204e-13, "grad_norm": 0.8577771782875061, "learning_rate": 0.00011919656831703132, "loss": 0.0513, "step": 33070 }, { "epoch": 2.295386103412511e-13, "grad_norm": 0.9209487438201904, "learning_rate": 0.00011917840500114939, "loss": 0.0533, "step": 33080 }, { "epoch": 2.296079992802902e-13, "grad_norm": 0.8435440063476562, "learning_rate": 0.00011916023791599064, "loss": 0.0539, "step": 33090 }, { "epoch": 2.2967738821932926e-13, "grad_norm": 0.8620036244392395, "learning_rate": 0.000119142067063422, "loss": 0.0522, "step": 33100 }, { "epoch": 2.2974677715836833e-13, "grad_norm": 0.8478751182556152, "learning_rate": 0.00011912389244531088, "loss": 0.0524, "step": 33110 }, { "epoch": 2.298161660974074e-13, "grad_norm": 0.861842155456543, "learning_rate": 0.00011910571406352495, "loss": 0.054, "step": 33120 }, { "epoch": 2.298855550364465e-13, "grad_norm": 0.8733360171318054, "learning_rate": 0.00011908753191993234, "loss": 0.0559, "step": 33130 }, { "epoch": 2.2995494397548555e-13, "grad_norm": 0.9012876749038696, "learning_rate": 0.00011906934601640156, "loss": 0.0574, "step": 33140 }, { "epoch": 2.300243329145246e-13, "grad_norm": 0.7914884686470032, "learning_rate": 0.00011905115635480148, "loss": 0.0543, "step": 33150 }, { "epoch": 2.300937218535637e-13, "grad_norm": 0.8762744665145874, "learning_rate": 0.00011903296293700137, "loss": 0.0548, "step": 33160 }, { "epoch": 2.3016311079260277e-13, "grad_norm": 0.8979458808898926, "learning_rate": 0.00011901476576487092, "loss": 0.0552, "step": 33170 }, { "epoch": 2.3023249973164184e-13, "grad_norm": 0.8087114691734314, "learning_rate": 0.0001189965648402801, "loss": 0.0581, "step": 33180 }, { "epoch": 2.303018886706809e-13, "grad_norm": 0.7592564225196838, "learning_rate": 0.00011897836016509939, "loss": 0.054, "step": 33190 }, { "epoch": 2.3037127760972e-13, "grad_norm": 0.8810161352157593, "learning_rate": 0.0001189601517411996, "loss": 0.0545, "step": 33200 }, { "epoch": 2.3044066654875905e-13, "grad_norm": 0.7163630127906799, "learning_rate": 0.00011894193957045193, "loss": 0.0531, "step": 33210 }, { "epoch": 2.3051005548779813e-13, "grad_norm": 0.7200165390968323, "learning_rate": 0.00011892372365472794, "loss": 0.0501, "step": 33220 }, { "epoch": 2.305794444268372e-13, "grad_norm": 0.8288876414299011, "learning_rate": 0.00011890550399589965, "loss": 0.0522, "step": 33230 }, { "epoch": 2.3064883336587627e-13, "grad_norm": 0.8452411890029907, "learning_rate": 0.00011888728059583934, "loss": 0.0527, "step": 33240 }, { "epoch": 2.3071822230491534e-13, "grad_norm": 0.7955312728881836, "learning_rate": 0.00011886905345641981, "loss": 0.0527, "step": 33250 }, { "epoch": 2.307876112439544e-13, "grad_norm": 0.9419277310371399, "learning_rate": 0.00011885082257951416, "loss": 0.0521, "step": 33260 }, { "epoch": 2.308570001829935e-13, "grad_norm": 0.914756715297699, "learning_rate": 0.0001188325879669959, "loss": 0.0539, "step": 33270 }, { "epoch": 2.3092638912203256e-13, "grad_norm": 0.9263339638710022, "learning_rate": 0.00011881434962073891, "loss": 0.052, "step": 33280 }, { "epoch": 2.3099577806107163e-13, "grad_norm": 0.8086922764778137, "learning_rate": 0.00011879610754261748, "loss": 0.0551, "step": 33290 }, { "epoch": 2.310651670001107e-13, "grad_norm": 0.8788591623306274, "learning_rate": 0.00011877786173450621, "loss": 0.0586, "step": 33300 }, { "epoch": 2.311345559391498e-13, "grad_norm": 0.8358376622200012, "learning_rate": 0.00011875961219828025, "loss": 0.0522, "step": 33310 }, { "epoch": 2.3120394487818885e-13, "grad_norm": 0.8298091292381287, "learning_rate": 0.00011874135893581493, "loss": 0.0568, "step": 33320 }, { "epoch": 2.312733338172279e-13, "grad_norm": 0.9163547158241272, "learning_rate": 0.0001187231019489861, "loss": 0.0562, "step": 33330 }, { "epoch": 2.31342722756267e-13, "grad_norm": 0.9379832744598389, "learning_rate": 0.00011870484123966992, "loss": 0.0548, "step": 33340 }, { "epoch": 2.3141211169530607e-13, "grad_norm": 0.812232494354248, "learning_rate": 0.00011868657680974299, "loss": 0.0523, "step": 33350 }, { "epoch": 2.3148150063434514e-13, "grad_norm": 0.9532745480537415, "learning_rate": 0.00011866830866108226, "loss": 0.0523, "step": 33360 }, { "epoch": 2.315508895733842e-13, "grad_norm": 0.8576196432113647, "learning_rate": 0.00011865003679556506, "loss": 0.0519, "step": 33370 }, { "epoch": 2.316202785124233e-13, "grad_norm": 0.8450637459754944, "learning_rate": 0.0001186317612150691, "loss": 0.0497, "step": 33380 }, { "epoch": 2.3168966745146236e-13, "grad_norm": 0.7764143943786621, "learning_rate": 0.00011861348192147252, "loss": 0.0503, "step": 33390 }, { "epoch": 2.3175905639050143e-13, "grad_norm": 0.9576692581176758, "learning_rate": 0.00011859519891665374, "loss": 0.0547, "step": 33400 }, { "epoch": 2.318284453295405e-13, "grad_norm": 0.8644888997077942, "learning_rate": 0.00011857691220249167, "loss": 0.0502, "step": 33410 }, { "epoch": 2.3189783426857957e-13, "grad_norm": 0.7400041222572327, "learning_rate": 0.00011855862178086552, "loss": 0.0518, "step": 33420 }, { "epoch": 2.3196722320761864e-13, "grad_norm": 0.8570407629013062, "learning_rate": 0.00011854032765365495, "loss": 0.0519, "step": 33430 }, { "epoch": 2.320366121466577e-13, "grad_norm": 0.8377070426940918, "learning_rate": 0.00011852202982273995, "loss": 0.0532, "step": 33440 }, { "epoch": 2.321060010856968e-13, "grad_norm": 0.7989755272865295, "learning_rate": 0.0001185037282900009, "loss": 0.0533, "step": 33450 }, { "epoch": 2.3217539002473586e-13, "grad_norm": 0.8143694996833801, "learning_rate": 0.00011848542305731858, "loss": 0.0572, "step": 33460 }, { "epoch": 2.3224477896377493e-13, "grad_norm": 0.9081676006317139, "learning_rate": 0.00011846711412657413, "loss": 0.053, "step": 33470 }, { "epoch": 2.32314167902814e-13, "grad_norm": 0.9340803623199463, "learning_rate": 0.00011844880149964908, "loss": 0.0541, "step": 33480 }, { "epoch": 2.323835568418531e-13, "grad_norm": 0.8124773502349854, "learning_rate": 0.00011843048517842535, "loss": 0.0565, "step": 33490 }, { "epoch": 2.3245294578089215e-13, "grad_norm": 0.8441075086593628, "learning_rate": 0.0001184121651647852, "loss": 0.0528, "step": 33500 }, { "epoch": 2.325223347199312e-13, "grad_norm": 0.8382510542869568, "learning_rate": 0.00011839384146061132, "loss": 0.0521, "step": 33510 }, { "epoch": 2.325917236589703e-13, "grad_norm": 0.8767039775848389, "learning_rate": 0.00011837551406778674, "loss": 0.0527, "step": 33520 }, { "epoch": 2.3266111259800937e-13, "grad_norm": 0.8590094447135925, "learning_rate": 0.00011835718298819489, "loss": 0.0511, "step": 33530 }, { "epoch": 2.3273050153704844e-13, "grad_norm": 0.9013243913650513, "learning_rate": 0.00011833884822371959, "loss": 0.0476, "step": 33540 }, { "epoch": 2.327998904760875e-13, "grad_norm": 0.8676539659500122, "learning_rate": 0.000118320509776245, "loss": 0.0518, "step": 33550 }, { "epoch": 2.328692794151266e-13, "grad_norm": 0.9195994734764099, "learning_rate": 0.00011830216764765567, "loss": 0.0508, "step": 33560 }, { "epoch": 2.3293866835416566e-13, "grad_norm": 0.7891174554824829, "learning_rate": 0.00011828382183983659, "loss": 0.0487, "step": 33570 }, { "epoch": 2.3300805729320473e-13, "grad_norm": 0.7508566379547119, "learning_rate": 0.00011826547235467301, "loss": 0.0506, "step": 33580 }, { "epoch": 2.330774462322438e-13, "grad_norm": 0.8279736638069153, "learning_rate": 0.00011824711919405066, "loss": 0.0512, "step": 33590 }, { "epoch": 2.3314683517128287e-13, "grad_norm": 0.8160027265548706, "learning_rate": 0.00011822876235985564, "loss": 0.0529, "step": 33600 }, { "epoch": 2.3321622411032195e-13, "grad_norm": 0.8962525129318237, "learning_rate": 0.00011821040185397435, "loss": 0.055, "step": 33610 }, { "epoch": 2.33285613049361e-13, "grad_norm": 0.7804820537567139, "learning_rate": 0.00011819203767829365, "loss": 0.0514, "step": 33620 }, { "epoch": 2.333550019884001e-13, "grad_norm": 0.9295476078987122, "learning_rate": 0.00011817366983470074, "loss": 0.0551, "step": 33630 }, { "epoch": 2.3342439092743916e-13, "grad_norm": 0.9490881562232971, "learning_rate": 0.0001181552983250832, "loss": 0.0542, "step": 33640 }, { "epoch": 2.3349377986647823e-13, "grad_norm": 0.9487265944480896, "learning_rate": 0.00011813692315132896, "loss": 0.0556, "step": 33650 }, { "epoch": 2.335631688055173e-13, "grad_norm": 0.9793968796730042, "learning_rate": 0.00011811854431532638, "loss": 0.0523, "step": 33660 }, { "epoch": 2.336325577445564e-13, "grad_norm": 0.7548530697822571, "learning_rate": 0.0001181001618189642, "loss": 0.0503, "step": 33670 }, { "epoch": 2.3370194668359545e-13, "grad_norm": 0.7070840001106262, "learning_rate": 0.00011808177566413147, "loss": 0.0527, "step": 33680 }, { "epoch": 2.337713356226345e-13, "grad_norm": 0.7934685945510864, "learning_rate": 0.00011806338585271765, "loss": 0.0503, "step": 33690 }, { "epoch": 2.338407245616736e-13, "grad_norm": 0.8250464200973511, "learning_rate": 0.0001180449923866126, "loss": 0.0496, "step": 33700 }, { "epoch": 2.3391011350071267e-13, "grad_norm": 0.8790379166603088, "learning_rate": 0.00011802659526770653, "loss": 0.0511, "step": 33710 }, { "epoch": 2.3397950243975174e-13, "grad_norm": 0.8702540397644043, "learning_rate": 0.00011800819449789003, "loss": 0.049, "step": 33720 }, { "epoch": 2.340488913787908e-13, "grad_norm": 0.9005327820777893, "learning_rate": 0.00011798979007905404, "loss": 0.0498, "step": 33730 }, { "epoch": 2.341182803178299e-13, "grad_norm": 0.7854928374290466, "learning_rate": 0.00011797138201308994, "loss": 0.0494, "step": 33740 }, { "epoch": 2.3418766925686896e-13, "grad_norm": 0.8377496600151062, "learning_rate": 0.00011795297030188943, "loss": 0.0505, "step": 33750 }, { "epoch": 2.3425705819590803e-13, "grad_norm": 0.8491815328598022, "learning_rate": 0.0001179345549473446, "loss": 0.0533, "step": 33760 }, { "epoch": 2.343264471349471e-13, "grad_norm": 0.8590802550315857, "learning_rate": 0.00011791613595134789, "loss": 0.0555, "step": 33770 }, { "epoch": 2.3439583607398617e-13, "grad_norm": 0.8168409466743469, "learning_rate": 0.00011789771331579218, "loss": 0.0505, "step": 33780 }, { "epoch": 2.3446522501302525e-13, "grad_norm": 0.7761601209640503, "learning_rate": 0.00011787928704257065, "loss": 0.0558, "step": 33790 }, { "epoch": 2.345346139520643e-13, "grad_norm": 0.8275895714759827, "learning_rate": 0.0001178608571335769, "loss": 0.0563, "step": 33800 }, { "epoch": 2.346040028911034e-13, "grad_norm": 0.77999347448349, "learning_rate": 0.00011784242359070491, "loss": 0.0513, "step": 33810 }, { "epoch": 2.3467339183014246e-13, "grad_norm": 0.8368788957595825, "learning_rate": 0.00011782398641584899, "loss": 0.0511, "step": 33820 }, { "epoch": 2.3474278076918154e-13, "grad_norm": 0.9922943711280823, "learning_rate": 0.00011780554561090383, "loss": 0.0531, "step": 33830 }, { "epoch": 2.348121697082206e-13, "grad_norm": 0.8517614603042603, "learning_rate": 0.00011778710117776453, "loss": 0.05, "step": 33840 }, { "epoch": 2.348815586472597e-13, "grad_norm": 0.7397392988204956, "learning_rate": 0.00011776865311832654, "loss": 0.0478, "step": 33850 }, { "epoch": 2.3495094758629875e-13, "grad_norm": 0.8187459111213684, "learning_rate": 0.00011775020143448568, "loss": 0.0481, "step": 33860 }, { "epoch": 2.350203365253378e-13, "grad_norm": 0.7182119488716125, "learning_rate": 0.00011773174612813817, "loss": 0.0522, "step": 33870 }, { "epoch": 2.350897254643769e-13, "grad_norm": 0.7264203429222107, "learning_rate": 0.00011771328720118055, "loss": 0.0474, "step": 33880 }, { "epoch": 2.3515911440341597e-13, "grad_norm": 0.7277066111564636, "learning_rate": 0.00011769482465550979, "loss": 0.0486, "step": 33890 }, { "epoch": 2.3522850334245504e-13, "grad_norm": 0.7523068189620972, "learning_rate": 0.00011767635849302318, "loss": 0.0495, "step": 33900 }, { "epoch": 2.352978922814941e-13, "grad_norm": 0.8781472444534302, "learning_rate": 0.00011765788871561842, "loss": 0.0514, "step": 33910 }, { "epoch": 2.353672812205332e-13, "grad_norm": 0.9381985068321228, "learning_rate": 0.00011763941532519357, "loss": 0.0526, "step": 33920 }, { "epoch": 2.3543667015957226e-13, "grad_norm": 0.7394860982894897, "learning_rate": 0.00011762093832364704, "loss": 0.0532, "step": 33930 }, { "epoch": 2.3550605909861133e-13, "grad_norm": 0.818760871887207, "learning_rate": 0.00011760245771287764, "loss": 0.052, "step": 33940 }, { "epoch": 2.355754480376504e-13, "grad_norm": 0.82021164894104, "learning_rate": 0.00011758397349478453, "loss": 0.0534, "step": 33950 }, { "epoch": 2.356448369766895e-13, "grad_norm": 0.813840925693512, "learning_rate": 0.00011756548567126729, "loss": 0.0557, "step": 33960 }, { "epoch": 2.3571422591572855e-13, "grad_norm": 0.771039605140686, "learning_rate": 0.00011754699424422576, "loss": 0.0512, "step": 33970 }, { "epoch": 2.357836148547676e-13, "grad_norm": 0.7766757011413574, "learning_rate": 0.00011752849921556029, "loss": 0.0508, "step": 33980 }, { "epoch": 2.358530037938067e-13, "grad_norm": 0.8463510870933533, "learning_rate": 0.00011751000058717151, "loss": 0.0532, "step": 33990 }, { "epoch": 2.3592239273284576e-13, "grad_norm": 0.7736789584159851, "learning_rate": 0.00011749149836096044, "loss": 0.0487, "step": 34000 }, { "epoch": 2.3599178167188484e-13, "grad_norm": 0.8330776691436768, "learning_rate": 0.00011747299253882846, "loss": 0.047, "step": 34010 }, { "epoch": 2.360611706109239e-13, "grad_norm": 0.7413046360015869, "learning_rate": 0.00011745448312267734, "loss": 0.0487, "step": 34020 }, { "epoch": 2.36130559549963e-13, "grad_norm": 0.7140994071960449, "learning_rate": 0.00011743597011440922, "loss": 0.0496, "step": 34030 }, { "epoch": 2.3619994848900205e-13, "grad_norm": 0.7323560118675232, "learning_rate": 0.00011741745351592657, "loss": 0.048, "step": 34040 }, { "epoch": 2.3626933742804113e-13, "grad_norm": 0.8984230160713196, "learning_rate": 0.0001173989333291323, "loss": 0.0508, "step": 34050 }, { "epoch": 2.363387263670802e-13, "grad_norm": 0.7759442925453186, "learning_rate": 0.00011738040955592962, "loss": 0.048, "step": 34060 }, { "epoch": 2.3640811530611927e-13, "grad_norm": 0.8919438123703003, "learning_rate": 0.00011736188219822214, "loss": 0.0511, "step": 34070 }, { "epoch": 2.3647750424515834e-13, "grad_norm": 0.9356898665428162, "learning_rate": 0.00011734335125791385, "loss": 0.0534, "step": 34080 }, { "epoch": 2.365468931841974e-13, "grad_norm": 0.8423986434936523, "learning_rate": 0.00011732481673690905, "loss": 0.0504, "step": 34090 }, { "epoch": 2.366162821232365e-13, "grad_norm": 0.9416558742523193, "learning_rate": 0.00011730627863711252, "loss": 0.055, "step": 34100 }, { "epoch": 2.3668567106227556e-13, "grad_norm": 0.9834628105163574, "learning_rate": 0.00011728773696042928, "loss": 0.0522, "step": 34110 }, { "epoch": 2.3675506000131463e-13, "grad_norm": 0.8512260913848877, "learning_rate": 0.00011726919170876477, "loss": 0.0521, "step": 34120 }, { "epoch": 2.368244489403537e-13, "grad_norm": 0.7863671183586121, "learning_rate": 0.00011725064288402484, "loss": 0.0508, "step": 34130 }, { "epoch": 2.368938378793928e-13, "grad_norm": 0.813810408115387, "learning_rate": 0.00011723209048811568, "loss": 0.048, "step": 34140 }, { "epoch": 2.3696322681843185e-13, "grad_norm": 0.8628058433532715, "learning_rate": 0.00011721353452294381, "loss": 0.0522, "step": 34150 }, { "epoch": 2.370326157574709e-13, "grad_norm": 0.8029714822769165, "learning_rate": 0.00011719497499041614, "loss": 0.0485, "step": 34160 }, { "epoch": 2.3710200469651e-13, "grad_norm": 0.6749922037124634, "learning_rate": 0.00011717641189243996, "loss": 0.0474, "step": 34170 }, { "epoch": 2.3717139363554907e-13, "grad_norm": 0.8103323578834534, "learning_rate": 0.00011715784523092291, "loss": 0.0493, "step": 34180 }, { "epoch": 2.3724078257458814e-13, "grad_norm": 0.8453459143638611, "learning_rate": 0.00011713927500777303, "loss": 0.0478, "step": 34190 }, { "epoch": 2.373101715136272e-13, "grad_norm": 0.7940205931663513, "learning_rate": 0.00011712070122489867, "loss": 0.0488, "step": 34200 }, { "epoch": 2.373795604526663e-13, "grad_norm": 0.7329771518707275, "learning_rate": 0.00011710212388420857, "loss": 0.0482, "step": 34210 }, { "epoch": 2.3744894939170536e-13, "grad_norm": 0.8279137015342712, "learning_rate": 0.00011708354298761187, "loss": 0.0503, "step": 34220 }, { "epoch": 2.3751833833074443e-13, "grad_norm": 0.8574737906455994, "learning_rate": 0.00011706495853701802, "loss": 0.0512, "step": 34230 }, { "epoch": 2.375877272697835e-13, "grad_norm": 0.8840752243995667, "learning_rate": 0.00011704637053433687, "loss": 0.0528, "step": 34240 }, { "epoch": 2.3765711620882257e-13, "grad_norm": 0.8476657867431641, "learning_rate": 0.00011702777898147866, "loss": 0.0498, "step": 34250 }, { "epoch": 2.3772650514786164e-13, "grad_norm": 0.7499244213104248, "learning_rate": 0.00011700918388035391, "loss": 0.0527, "step": 34260 }, { "epoch": 2.377958940869007e-13, "grad_norm": 0.7848800420761108, "learning_rate": 0.00011699058523287359, "loss": 0.0548, "step": 34270 }, { "epoch": 2.378652830259398e-13, "grad_norm": 0.8767039775848389, "learning_rate": 0.00011697198304094897, "loss": 0.0504, "step": 34280 }, { "epoch": 2.3793467196497886e-13, "grad_norm": 0.8701224327087402, "learning_rate": 0.00011695337730649176, "loss": 0.0479, "step": 34290 }, { "epoch": 2.3800406090401793e-13, "grad_norm": 0.8622602224349976, "learning_rate": 0.00011693476803141391, "loss": 0.0499, "step": 34300 }, { "epoch": 2.38073449843057e-13, "grad_norm": 0.7799948453903198, "learning_rate": 0.0001169161552176279, "loss": 0.0485, "step": 34310 }, { "epoch": 2.381428387820961e-13, "grad_norm": 0.7791817784309387, "learning_rate": 0.00011689753886704646, "loss": 0.0466, "step": 34320 }, { "epoch": 2.3821222772113515e-13, "grad_norm": 0.8497368097305298, "learning_rate": 0.00011687891898158266, "loss": 0.0479, "step": 34330 }, { "epoch": 2.382816166601742e-13, "grad_norm": 0.7473609447479248, "learning_rate": 0.00011686029556315005, "loss": 0.0482, "step": 34340 }, { "epoch": 2.383510055992133e-13, "grad_norm": 0.7969523072242737, "learning_rate": 0.00011684166861366244, "loss": 0.0466, "step": 34350 }, { "epoch": 2.3842039453825237e-13, "grad_norm": 0.8665263652801514, "learning_rate": 0.00011682303813503406, "loss": 0.0476, "step": 34360 }, { "epoch": 2.3848978347729144e-13, "grad_norm": 0.8084185123443604, "learning_rate": 0.00011680440412917946, "loss": 0.0481, "step": 34370 }, { "epoch": 2.385591724163305e-13, "grad_norm": 0.8109997510910034, "learning_rate": 0.00011678576659801358, "loss": 0.0502, "step": 34380 }, { "epoch": 2.386285613553696e-13, "grad_norm": 0.792270839214325, "learning_rate": 0.00011676712554345172, "loss": 0.0514, "step": 34390 }, { "epoch": 2.3869795029440866e-13, "grad_norm": 0.7761482000350952, "learning_rate": 0.00011674848096740954, "loss": 0.0507, "step": 34400 }, { "epoch": 2.3876733923344773e-13, "grad_norm": 0.8484029173851013, "learning_rate": 0.00011672983287180305, "loss": 0.0514, "step": 34410 }, { "epoch": 2.388367281724868e-13, "grad_norm": 0.816978931427002, "learning_rate": 0.00011671118125854863, "loss": 0.0513, "step": 34420 }, { "epoch": 2.3890611711152587e-13, "grad_norm": 0.715154767036438, "learning_rate": 0.00011669252612956305, "loss": 0.0538, "step": 34430 }, { "epoch": 2.3897550605056495e-13, "grad_norm": 0.6841785907745361, "learning_rate": 0.00011667386748676343, "loss": 0.0487, "step": 34440 }, { "epoch": 2.39044894989604e-13, "grad_norm": 0.8316085338592529, "learning_rate": 0.00011665520533206716, "loss": 0.0498, "step": 34450 }, { "epoch": 2.391142839286431e-13, "grad_norm": 0.7458159327507019, "learning_rate": 0.00011663653966739211, "loss": 0.0496, "step": 34460 }, { "epoch": 2.3918367286768216e-13, "grad_norm": 0.7086544632911682, "learning_rate": 0.00011661787049465649, "loss": 0.0473, "step": 34470 }, { "epoch": 2.3925306180672123e-13, "grad_norm": 0.7855117321014404, "learning_rate": 0.00011659919781577883, "loss": 0.0474, "step": 34480 }, { "epoch": 2.393224507457603e-13, "grad_norm": 0.8141254782676697, "learning_rate": 0.00011658052163267801, "loss": 0.0476, "step": 34490 }, { "epoch": 2.393918396847994e-13, "grad_norm": 0.7361511588096619, "learning_rate": 0.00011656184194727333, "loss": 0.0468, "step": 34500 }, { "epoch": 2.3946122862383845e-13, "grad_norm": 0.85152268409729, "learning_rate": 0.00011654315876148441, "loss": 0.0466, "step": 34510 }, { "epoch": 2.395306175628775e-13, "grad_norm": 0.8354913592338562, "learning_rate": 0.00011652447207723124, "loss": 0.0485, "step": 34520 }, { "epoch": 2.396000065019166e-13, "grad_norm": 0.8651109337806702, "learning_rate": 0.00011650578189643417, "loss": 0.0471, "step": 34530 }, { "epoch": 2.3966939544095567e-13, "grad_norm": 0.8311788439750671, "learning_rate": 0.0001164870882210139, "loss": 0.0497, "step": 34540 }, { "epoch": 2.3973878437999474e-13, "grad_norm": 0.8500776290893555, "learning_rate": 0.00011646839105289151, "loss": 0.0527, "step": 34550 }, { "epoch": 2.398081733190338e-13, "grad_norm": 0.7795536518096924, "learning_rate": 0.0001164496903939884, "loss": 0.0499, "step": 34560 }, { "epoch": 2.398775622580729e-13, "grad_norm": 0.8378130197525024, "learning_rate": 0.00011643098624622636, "loss": 0.0537, "step": 34570 }, { "epoch": 2.3994695119711196e-13, "grad_norm": 0.9401465654373169, "learning_rate": 0.00011641227861152755, "loss": 0.0531, "step": 34580 }, { "epoch": 2.4001634013615103e-13, "grad_norm": 0.8769137263298035, "learning_rate": 0.00011639356749181448, "loss": 0.0507, "step": 34590 }, { "epoch": 2.400857290751901e-13, "grad_norm": 0.7344598770141602, "learning_rate": 0.00011637485288900997, "loss": 0.0476, "step": 34600 }, { "epoch": 2.4015511801422917e-13, "grad_norm": 0.8792693614959717, "learning_rate": 0.00011635613480503727, "loss": 0.0482, "step": 34610 }, { "epoch": 2.4022450695326825e-13, "grad_norm": 0.8395033478736877, "learning_rate": 0.00011633741324181992, "loss": 0.0485, "step": 34620 }, { "epoch": 2.402938958923073e-13, "grad_norm": 0.8213008642196655, "learning_rate": 0.0001163186882012819, "loss": 0.047, "step": 34630 }, { "epoch": 2.403632848313464e-13, "grad_norm": 0.7389689683914185, "learning_rate": 0.00011629995968534743, "loss": 0.0459, "step": 34640 }, { "epoch": 2.4043267377038546e-13, "grad_norm": 0.8863536715507507, "learning_rate": 0.00011628122769594125, "loss": 0.05, "step": 34650 }, { "epoch": 2.4050206270942454e-13, "grad_norm": 0.7763416171073914, "learning_rate": 0.00011626249223498828, "loss": 0.0444, "step": 34660 }, { "epoch": 2.405714516484636e-13, "grad_norm": 0.6902033686637878, "learning_rate": 0.00011624375330441392, "loss": 0.0463, "step": 34670 }, { "epoch": 2.406408405875027e-13, "grad_norm": 0.7790897488594055, "learning_rate": 0.00011622501090614387, "loss": 0.048, "step": 34680 }, { "epoch": 2.4071022952654175e-13, "grad_norm": 0.7930765151977539, "learning_rate": 0.00011620626504210422, "loss": 0.049, "step": 34690 }, { "epoch": 2.407796184655808e-13, "grad_norm": 0.8119553923606873, "learning_rate": 0.00011618751571422139, "loss": 0.0486, "step": 34700 }, { "epoch": 2.408490074046199e-13, "grad_norm": 0.9000443816184998, "learning_rate": 0.00011616876292442215, "loss": 0.0521, "step": 34710 }, { "epoch": 2.4091839634365897e-13, "grad_norm": 0.8864666223526001, "learning_rate": 0.00011615000667463367, "loss": 0.0492, "step": 34720 }, { "epoch": 2.4098778528269804e-13, "grad_norm": 0.9026704430580139, "learning_rate": 0.00011613124696678344, "loss": 0.0511, "step": 34730 }, { "epoch": 2.410571742217371e-13, "grad_norm": 0.8109933137893677, "learning_rate": 0.0001161124838027993, "loss": 0.0542, "step": 34740 }, { "epoch": 2.411265631607762e-13, "grad_norm": 0.8617712259292603, "learning_rate": 0.00011609371718460944, "loss": 0.05, "step": 34750 }, { "epoch": 2.4119595209981526e-13, "grad_norm": 0.8210774660110474, "learning_rate": 0.00011607494711414247, "loss": 0.0476, "step": 34760 }, { "epoch": 2.4126534103885433e-13, "grad_norm": 0.8614673018455505, "learning_rate": 0.00011605617359332727, "loss": 0.049, "step": 34770 }, { "epoch": 2.413347299778934e-13, "grad_norm": 0.8252878189086914, "learning_rate": 0.00011603739662409313, "loss": 0.0486, "step": 34780 }, { "epoch": 2.414041189169325e-13, "grad_norm": 0.8458961844444275, "learning_rate": 0.00011601861620836966, "loss": 0.0458, "step": 34790 }, { "epoch": 2.4147350785597155e-13, "grad_norm": 0.7895017266273499, "learning_rate": 0.00011599983234808688, "loss": 0.0476, "step": 34800 }, { "epoch": 2.415428967950106e-13, "grad_norm": 0.8436257243156433, "learning_rate": 0.00011598104504517505, "loss": 0.0466, "step": 34810 }, { "epoch": 2.416122857340497e-13, "grad_norm": 0.760536253452301, "learning_rate": 0.00011596225430156491, "loss": 0.0443, "step": 34820 }, { "epoch": 2.4168167467308876e-13, "grad_norm": 0.7614865899085999, "learning_rate": 0.0001159434601191875, "loss": 0.0454, "step": 34830 }, { "epoch": 2.4175106361212784e-13, "grad_norm": 0.7884981632232666, "learning_rate": 0.00011592466249997419, "loss": 0.047, "step": 34840 }, { "epoch": 2.418204525511669e-13, "grad_norm": 0.7609449028968811, "learning_rate": 0.00011590586144585677, "loss": 0.0484, "step": 34850 }, { "epoch": 2.41889841490206e-13, "grad_norm": 0.8242413997650146, "learning_rate": 0.00011588705695876728, "loss": 0.0503, "step": 34860 }, { "epoch": 2.4195923042924505e-13, "grad_norm": 0.8660187721252441, "learning_rate": 0.00011586824904063823, "loss": 0.0485, "step": 34870 }, { "epoch": 2.4202861936828413e-13, "grad_norm": 0.9127811789512634, "learning_rate": 0.0001158494376934024, "loss": 0.0512, "step": 34880 }, { "epoch": 2.420980083073232e-13, "grad_norm": 0.9334484934806824, "learning_rate": 0.00011583062291899294, "loss": 0.0511, "step": 34890 }, { "epoch": 2.4216739724636227e-13, "grad_norm": 0.8662179708480835, "learning_rate": 0.00011581180471934341, "loss": 0.0536, "step": 34900 }, { "epoch": 2.4223678618540134e-13, "grad_norm": 0.9357815980911255, "learning_rate": 0.0001157929830963876, "loss": 0.0494, "step": 34910 }, { "epoch": 2.423061751244404e-13, "grad_norm": 0.7313324809074402, "learning_rate": 0.00011577415805205978, "loss": 0.0456, "step": 34920 }, { "epoch": 2.423755640634795e-13, "grad_norm": 0.6965136528015137, "learning_rate": 0.00011575532958829449, "loss": 0.0485, "step": 34930 }, { "epoch": 2.4244495300251856e-13, "grad_norm": 0.8588796854019165, "learning_rate": 0.00011573649770702665, "loss": 0.048, "step": 34940 }, { "epoch": 2.4251434194155763e-13, "grad_norm": 0.8385587930679321, "learning_rate": 0.00011571766241019155, "loss": 0.0479, "step": 34950 }, { "epoch": 2.425837308805967e-13, "grad_norm": 0.8626478314399719, "learning_rate": 0.00011569882369972479, "loss": 0.0475, "step": 34960 }, { "epoch": 2.426531198196358e-13, "grad_norm": 0.8327782154083252, "learning_rate": 0.00011567998157756233, "loss": 0.0461, "step": 34970 }, { "epoch": 2.4272250875867485e-13, "grad_norm": 0.8628079891204834, "learning_rate": 0.0001156611360456405, "loss": 0.046, "step": 34980 }, { "epoch": 2.427918976977139e-13, "grad_norm": 0.7720588445663452, "learning_rate": 0.000115642287105896, "loss": 0.046, "step": 34990 }, { "epoch": 2.42861286636753e-13, "grad_norm": 0.8453037738800049, "learning_rate": 0.00011562343476026581, "loss": 0.0467, "step": 35000 }, { "epoch": 2.4293067557579207e-13, "grad_norm": 0.8147440552711487, "learning_rate": 0.00011560457901068731, "loss": 0.0479, "step": 35010 }, { "epoch": 2.4300006451483114e-13, "grad_norm": 0.7776644825935364, "learning_rate": 0.00011558571985909826, "loss": 0.0502, "step": 35020 }, { "epoch": 2.430694534538702e-13, "grad_norm": 0.7681533694267273, "learning_rate": 0.00011556685730743668, "loss": 0.0481, "step": 35030 }, { "epoch": 2.431388423929093e-13, "grad_norm": 0.7960782051086426, "learning_rate": 0.00011554799135764104, "loss": 0.0522, "step": 35040 }, { "epoch": 2.4320823133194835e-13, "grad_norm": 0.8625665307044983, "learning_rate": 0.00011552912201165005, "loss": 0.0525, "step": 35050 }, { "epoch": 2.4327762027098743e-13, "grad_norm": 0.7683826684951782, "learning_rate": 0.0001155102492714029, "loss": 0.0492, "step": 35060 }, { "epoch": 2.433470092100265e-13, "grad_norm": 0.8227818012237549, "learning_rate": 0.000115491373138839, "loss": 0.0475, "step": 35070 }, { "epoch": 2.4341639814906557e-13, "grad_norm": 0.9266380071640015, "learning_rate": 0.00011547249361589818, "loss": 0.0485, "step": 35080 }, { "epoch": 2.4348578708810464e-13, "grad_norm": 0.8361711502075195, "learning_rate": 0.00011545361070452063, "loss": 0.0477, "step": 35090 }, { "epoch": 2.435551760271437e-13, "grad_norm": 0.7973777651786804, "learning_rate": 0.00011543472440664682, "loss": 0.0469, "step": 35100 }, { "epoch": 2.436245649661828e-13, "grad_norm": 0.8559040427207947, "learning_rate": 0.00011541583472421766, "loss": 0.0467, "step": 35110 }, { "epoch": 2.4369395390522186e-13, "grad_norm": 0.7259889841079712, "learning_rate": 0.00011539694165917433, "loss": 0.0481, "step": 35120 }, { "epoch": 2.4376334284426093e-13, "grad_norm": 0.7225173115730286, "learning_rate": 0.00011537804521345839, "loss": 0.0445, "step": 35130 }, { "epoch": 2.438327317833e-13, "grad_norm": 0.6954705715179443, "learning_rate": 0.00011535914538901173, "loss": 0.0445, "step": 35140 }, { "epoch": 2.439021207223391e-13, "grad_norm": 0.716577410697937, "learning_rate": 0.00011534024218777663, "loss": 0.0462, "step": 35150 }, { "epoch": 2.4397150966137815e-13, "grad_norm": 0.8507887721061707, "learning_rate": 0.00011532133561169566, "loss": 0.0478, "step": 35160 }, { "epoch": 2.440408986004172e-13, "grad_norm": 0.9517760872840881, "learning_rate": 0.00011530242566271179, "loss": 0.0487, "step": 35170 }, { "epoch": 2.441102875394563e-13, "grad_norm": 0.7485321760177612, "learning_rate": 0.0001152835123427683, "loss": 0.0487, "step": 35180 }, { "epoch": 2.4417967647849537e-13, "grad_norm": 0.7300138473510742, "learning_rate": 0.00011526459565380883, "loss": 0.0486, "step": 35190 }, { "epoch": 2.4424906541753444e-13, "grad_norm": 0.8109167218208313, "learning_rate": 0.00011524567559777735, "loss": 0.0514, "step": 35200 }, { "epoch": 2.443184543565735e-13, "grad_norm": 0.8425114750862122, "learning_rate": 0.0001152267521766182, "loss": 0.0526, "step": 35210 }, { "epoch": 2.443878432956126e-13, "grad_norm": 0.7816370725631714, "learning_rate": 0.00011520782539227608, "loss": 0.0487, "step": 35220 }, { "epoch": 2.4445723223465166e-13, "grad_norm": 0.7550259232521057, "learning_rate": 0.00011518889524669598, "loss": 0.0471, "step": 35230 }, { "epoch": 2.4452662117369073e-13, "grad_norm": 0.8230970501899719, "learning_rate": 0.00011516996174182327, "loss": 0.0486, "step": 35240 }, { "epoch": 2.445960101127298e-13, "grad_norm": 0.7285270094871521, "learning_rate": 0.00011515102487960368, "loss": 0.0468, "step": 35250 }, { "epoch": 2.4466539905176887e-13, "grad_norm": 0.8274895548820496, "learning_rate": 0.00011513208466198324, "loss": 0.0461, "step": 35260 }, { "epoch": 2.4473478799080794e-13, "grad_norm": 0.7240591049194336, "learning_rate": 0.00011511314109090837, "loss": 0.0473, "step": 35270 }, { "epoch": 2.44804176929847e-13, "grad_norm": 0.7413195967674255, "learning_rate": 0.00011509419416832583, "loss": 0.0471, "step": 35280 }, { "epoch": 2.448735658688861e-13, "grad_norm": 0.6979293823242188, "learning_rate": 0.00011507524389618269, "loss": 0.0438, "step": 35290 }, { "epoch": 2.4494295480792516e-13, "grad_norm": 0.8619993329048157, "learning_rate": 0.00011505629027642638, "loss": 0.046, "step": 35300 }, { "epoch": 2.4501234374696423e-13, "grad_norm": 0.741826593875885, "learning_rate": 0.00011503733331100468, "loss": 0.0448, "step": 35310 }, { "epoch": 2.450817326860033e-13, "grad_norm": 0.8326917886734009, "learning_rate": 0.00011501837300186577, "loss": 0.0471, "step": 35320 }, { "epoch": 2.451511216250424e-13, "grad_norm": 0.887649416923523, "learning_rate": 0.00011499940935095805, "loss": 0.0507, "step": 35330 }, { "epoch": 2.4522051056408145e-13, "grad_norm": 0.858023464679718, "learning_rate": 0.00011498044236023034, "loss": 0.0462, "step": 35340 }, { "epoch": 2.452898995031205e-13, "grad_norm": 0.9294561743736267, "learning_rate": 0.00011496147203163181, "loss": 0.0501, "step": 35350 }, { "epoch": 2.453592884421596e-13, "grad_norm": 0.9098287224769592, "learning_rate": 0.00011494249836711197, "loss": 0.049, "step": 35360 }, { "epoch": 2.4542867738119867e-13, "grad_norm": 0.8440823554992676, "learning_rate": 0.0001149235213686206, "loss": 0.0513, "step": 35370 }, { "epoch": 2.4549806632023774e-13, "grad_norm": 0.7574344873428345, "learning_rate": 0.00011490454103810793, "loss": 0.0482, "step": 35380 }, { "epoch": 2.455674552592768e-13, "grad_norm": 0.7852427363395691, "learning_rate": 0.0001148855573775245, "loss": 0.0457, "step": 35390 }, { "epoch": 2.456368441983159e-13, "grad_norm": 0.8851295113563538, "learning_rate": 0.00011486657038882113, "loss": 0.0478, "step": 35400 }, { "epoch": 2.4570623313735496e-13, "grad_norm": 0.7771066427230835, "learning_rate": 0.00011484758007394905, "loss": 0.0469, "step": 35410 }, { "epoch": 2.4577562207639403e-13, "grad_norm": 0.6446880102157593, "learning_rate": 0.00011482858643485983, "loss": 0.0467, "step": 35420 }, { "epoch": 2.458450110154331e-13, "grad_norm": 0.736162543296814, "learning_rate": 0.00011480958947350533, "loss": 0.0472, "step": 35430 }, { "epoch": 2.4591439995447217e-13, "grad_norm": 0.7811934351921082, "learning_rate": 0.00011479058919183779, "loss": 0.0453, "step": 35440 }, { "epoch": 2.4598378889351125e-13, "grad_norm": 0.7865127325057983, "learning_rate": 0.00011477158559180983, "loss": 0.0455, "step": 35450 }, { "epoch": 2.460531778325503e-13, "grad_norm": 0.7622998356819153, "learning_rate": 0.00011475257867537429, "loss": 0.0441, "step": 35460 }, { "epoch": 2.461225667715894e-13, "grad_norm": 0.7977428436279297, "learning_rate": 0.00011473356844448448, "loss": 0.046, "step": 35470 }, { "epoch": 2.4619195571062846e-13, "grad_norm": 0.8413758277893066, "learning_rate": 0.00011471455490109396, "loss": 0.0461, "step": 35480 }, { "epoch": 2.4626134464966754e-13, "grad_norm": 0.8629497289657593, "learning_rate": 0.0001146955380471567, "loss": 0.0497, "step": 35490 }, { "epoch": 2.463307335887066e-13, "grad_norm": 0.8945522308349609, "learning_rate": 0.00011467651788462696, "loss": 0.0469, "step": 35500 }, { "epoch": 2.464001225277457e-13, "grad_norm": 0.7739160060882568, "learning_rate": 0.00011465749441545936, "loss": 0.0489, "step": 35510 }, { "epoch": 2.4646951146678475e-13, "grad_norm": 0.8275593519210815, "learning_rate": 0.00011463846764160886, "loss": 0.0515, "step": 35520 }, { "epoch": 2.465389004058238e-13, "grad_norm": 0.9040143489837646, "learning_rate": 0.00011461943756503078, "loss": 0.0494, "step": 35530 }, { "epoch": 2.466082893448629e-13, "grad_norm": 0.7959047555923462, "learning_rate": 0.0001146004041876807, "loss": 0.0472, "step": 35540 }, { "epoch": 2.4667767828390197e-13, "grad_norm": 0.745222270488739, "learning_rate": 0.00011458136751151465, "loss": 0.0481, "step": 35550 }, { "epoch": 2.4674706722294104e-13, "grad_norm": 0.7988526821136475, "learning_rate": 0.00011456232753848894, "loss": 0.0474, "step": 35560 }, { "epoch": 2.468164561619801e-13, "grad_norm": 0.8109333515167236, "learning_rate": 0.00011454328427056018, "loss": 0.0466, "step": 35570 }, { "epoch": 2.468858451010192e-13, "grad_norm": 0.873643159866333, "learning_rate": 0.0001145242377096854, "loss": 0.0483, "step": 35580 }, { "epoch": 2.4695523404005826e-13, "grad_norm": 0.7545214891433716, "learning_rate": 0.00011450518785782191, "loss": 0.0466, "step": 35590 }, { "epoch": 2.4702462297909733e-13, "grad_norm": 0.7692100405693054, "learning_rate": 0.0001144861347169274, "loss": 0.0451, "step": 35600 }, { "epoch": 2.470940119181364e-13, "grad_norm": 0.8209536075592041, "learning_rate": 0.00011446707828895987, "loss": 0.0443, "step": 35610 }, { "epoch": 2.471634008571755e-13, "grad_norm": 0.7796837687492371, "learning_rate": 0.00011444801857587765, "loss": 0.0443, "step": 35620 }, { "epoch": 2.4723278979621455e-13, "grad_norm": 0.7921335697174072, "learning_rate": 0.00011442895557963942, "loss": 0.0458, "step": 35630 }, { "epoch": 2.473021787352536e-13, "grad_norm": 0.7864035367965698, "learning_rate": 0.00011440988930220423, "loss": 0.0469, "step": 35640 }, { "epoch": 2.473715676742927e-13, "grad_norm": 0.7793260812759399, "learning_rate": 0.0001143908197455314, "loss": 0.0462, "step": 35650 }, { "epoch": 2.4744095661333176e-13, "grad_norm": 0.7801175117492676, "learning_rate": 0.00011437174691158069, "loss": 0.0479, "step": 35660 }, { "epoch": 2.4751034555237084e-13, "grad_norm": 0.8823111653327942, "learning_rate": 0.00011435267080231205, "loss": 0.0498, "step": 35670 }, { "epoch": 2.475797344914099e-13, "grad_norm": 0.7820412516593933, "learning_rate": 0.00011433359141968587, "loss": 0.0504, "step": 35680 }, { "epoch": 2.47649123430449e-13, "grad_norm": 0.757707953453064, "learning_rate": 0.00011431450876566286, "loss": 0.047, "step": 35690 }, { "epoch": 2.4771851236948805e-13, "grad_norm": 0.8450416922569275, "learning_rate": 0.00011429542284220407, "loss": 0.0476, "step": 35700 }, { "epoch": 2.4778790130852713e-13, "grad_norm": 0.7236728668212891, "learning_rate": 0.00011427633365127088, "loss": 0.0476, "step": 35710 }, { "epoch": 2.478572902475662e-13, "grad_norm": 0.7290011644363403, "learning_rate": 0.00011425724119482498, "loss": 0.0469, "step": 35720 }, { "epoch": 2.4792667918660527e-13, "grad_norm": 0.7771588563919067, "learning_rate": 0.00011423814547482843, "loss": 0.0488, "step": 35730 }, { "epoch": 2.4799606812564434e-13, "grad_norm": 0.8076416254043579, "learning_rate": 0.0001142190464932436, "loss": 0.0474, "step": 35740 }, { "epoch": 2.480654570646834e-13, "grad_norm": 0.7726717591285706, "learning_rate": 0.00011419994425203322, "loss": 0.046, "step": 35750 }, { "epoch": 2.481348460037225e-13, "grad_norm": 0.8253093361854553, "learning_rate": 0.00011418083875316034, "loss": 0.0451, "step": 35760 }, { "epoch": 2.4820423494276156e-13, "grad_norm": 0.7994204163551331, "learning_rate": 0.00011416172999858836, "loss": 0.0458, "step": 35770 }, { "epoch": 2.4827362388180063e-13, "grad_norm": 0.8198405504226685, "learning_rate": 0.00011414261799028097, "loss": 0.0442, "step": 35780 }, { "epoch": 2.483430128208397e-13, "grad_norm": 0.7281197905540466, "learning_rate": 0.00011412350273020225, "loss": 0.0451, "step": 35790 }, { "epoch": 2.484124017598788e-13, "grad_norm": 0.7567961812019348, "learning_rate": 0.00011410438422031659, "loss": 0.0476, "step": 35800 }, { "epoch": 2.4848179069891785e-13, "grad_norm": 0.7498745322227478, "learning_rate": 0.0001140852624625887, "loss": 0.0453, "step": 35810 }, { "epoch": 2.485511796379569e-13, "grad_norm": 0.7552310228347778, "learning_rate": 0.00011406613745898365, "loss": 0.0486, "step": 35820 }, { "epoch": 2.48620568576996e-13, "grad_norm": 0.8451691269874573, "learning_rate": 0.00011404700921146684, "loss": 0.0509, "step": 35830 }, { "epoch": 2.4868995751603507e-13, "grad_norm": 0.8670942187309265, "learning_rate": 0.00011402787772200396, "loss": 0.0486, "step": 35840 }, { "epoch": 2.4875934645507414e-13, "grad_norm": 0.7839831113815308, "learning_rate": 0.00011400874299256111, "loss": 0.0458, "step": 35850 }, { "epoch": 2.488287353941132e-13, "grad_norm": 0.9557467699050903, "learning_rate": 0.00011398960502510465, "loss": 0.046, "step": 35860 }, { "epoch": 2.488981243331523e-13, "grad_norm": 0.9052034616470337, "learning_rate": 0.0001139704638216013, "loss": 0.0464, "step": 35870 }, { "epoch": 2.4896751327219135e-13, "grad_norm": 0.8548781275749207, "learning_rate": 0.00011395131938401815, "loss": 0.0469, "step": 35880 }, { "epoch": 2.4903690221123043e-13, "grad_norm": 0.7416832447052002, "learning_rate": 0.0001139321717143226, "loss": 0.0461, "step": 35890 }, { "epoch": 2.491062911502695e-13, "grad_norm": 0.8733077645301819, "learning_rate": 0.00011391302081448233, "loss": 0.0493, "step": 35900 }, { "epoch": 2.4917568008930857e-13, "grad_norm": 0.8009308576583862, "learning_rate": 0.0001138938666864654, "loss": 0.044, "step": 35910 }, { "epoch": 2.4924506902834764e-13, "grad_norm": 0.6735405325889587, "learning_rate": 0.00011387470933224019, "loss": 0.0445, "step": 35920 }, { "epoch": 2.493144579673867e-13, "grad_norm": 0.7248240113258362, "learning_rate": 0.00011385554875377544, "loss": 0.0445, "step": 35930 }, { "epoch": 2.493838469064258e-13, "grad_norm": 0.7058416604995728, "learning_rate": 0.00011383638495304019, "loss": 0.0451, "step": 35940 }, { "epoch": 2.4945323584546486e-13, "grad_norm": 0.7093080282211304, "learning_rate": 0.00011381721793200378, "loss": 0.0445, "step": 35950 }, { "epoch": 2.4952262478450393e-13, "grad_norm": 0.7275057435035706, "learning_rate": 0.000113798047692636, "loss": 0.0465, "step": 35960 }, { "epoch": 2.49592013723543e-13, "grad_norm": 0.8429529070854187, "learning_rate": 0.0001137788742369068, "loss": 0.0453, "step": 35970 }, { "epoch": 2.496614026625821e-13, "grad_norm": 0.8872484564781189, "learning_rate": 0.00011375969756678663, "loss": 0.0469, "step": 35980 }, { "epoch": 2.4973079160162115e-13, "grad_norm": 0.7857743501663208, "learning_rate": 0.00011374051768424614, "loss": 0.0503, "step": 35990 }, { "epoch": 2.498001805406602e-13, "grad_norm": 0.8099059462547302, "learning_rate": 0.00011372133459125638, "loss": 0.0472, "step": 36000 }, { "epoch": 2.498695694796993e-13, "grad_norm": 0.7683554887771606, "learning_rate": 0.00011370214828978871, "loss": 0.0454, "step": 36010 }, { "epoch": 2.4993895841873837e-13, "grad_norm": 0.8073738813400269, "learning_rate": 0.00011368295878181479, "loss": 0.046, "step": 36020 }, { "epoch": 2.5000834735777744e-13, "grad_norm": 0.921096920967102, "learning_rate": 0.0001136637660693067, "loss": 0.0464, "step": 36030 }, { "epoch": 2.500777362968165e-13, "grad_norm": 0.8846017122268677, "learning_rate": 0.00011364457015423677, "loss": 0.0445, "step": 36040 }, { "epoch": 2.501471252358556e-13, "grad_norm": 0.8138588070869446, "learning_rate": 0.00011362537103857764, "loss": 0.0474, "step": 36050 }, { "epoch": 2.5021651417489466e-13, "grad_norm": 0.8658238649368286, "learning_rate": 0.00011360616872430235, "loss": 0.0464, "step": 36060 }, { "epoch": 2.5028590311393373e-13, "grad_norm": 0.7130147814750671, "learning_rate": 0.00011358696321338425, "loss": 0.0432, "step": 36070 }, { "epoch": 2.503552920529728e-13, "grad_norm": 0.7182934880256653, "learning_rate": 0.00011356775450779697, "loss": 0.044, "step": 36080 }, { "epoch": 2.5042468099201187e-13, "grad_norm": 0.806667685508728, "learning_rate": 0.00011354854260951453, "loss": 0.044, "step": 36090 }, { "epoch": 2.5049406993105094e-13, "grad_norm": 0.7408490777015686, "learning_rate": 0.00011352932752051127, "loss": 0.045, "step": 36100 }, { "epoch": 2.5056345887009e-13, "grad_norm": 0.7712174654006958, "learning_rate": 0.0001135101092427618, "loss": 0.0454, "step": 36110 }, { "epoch": 2.506328478091291e-13, "grad_norm": 0.7543407082557678, "learning_rate": 0.00011349088777824108, "loss": 0.0429, "step": 36120 }, { "epoch": 2.5070223674816816e-13, "grad_norm": 0.849170982837677, "learning_rate": 0.00011347166312892445, "loss": 0.0468, "step": 36130 }, { "epoch": 2.5077162568720723e-13, "grad_norm": 0.9344675540924072, "learning_rate": 0.00011345243529678754, "loss": 0.0475, "step": 36140 }, { "epoch": 2.508410146262463e-13, "grad_norm": 0.8887582421302795, "learning_rate": 0.00011343320428380633, "loss": 0.0487, "step": 36150 }, { "epoch": 2.509104035652854e-13, "grad_norm": 0.9139231443405151, "learning_rate": 0.00011341397009195706, "loss": 0.0461, "step": 36160 }, { "epoch": 2.5097979250432445e-13, "grad_norm": 0.6703584790229797, "learning_rate": 0.00011339473272321635, "loss": 0.0434, "step": 36170 }, { "epoch": 2.510491814433635e-13, "grad_norm": 0.6387067437171936, "learning_rate": 0.00011337549217956116, "loss": 0.0456, "step": 36180 }, { "epoch": 2.511185703824026e-13, "grad_norm": 0.7813538312911987, "learning_rate": 0.00011335624846296875, "loss": 0.0454, "step": 36190 }, { "epoch": 2.5118795932144167e-13, "grad_norm": 0.8318049907684326, "learning_rate": 0.00011333700157541672, "loss": 0.0458, "step": 36200 }, { "epoch": 2.5125734826048074e-13, "grad_norm": 0.8805781602859497, "learning_rate": 0.00011331775151888297, "loss": 0.0465, "step": 36210 }, { "epoch": 2.513267371995198e-13, "grad_norm": 0.8301519155502319, "learning_rate": 0.00011329849829534574, "loss": 0.0445, "step": 36220 }, { "epoch": 2.513961261385589e-13, "grad_norm": 0.8323409557342529, "learning_rate": 0.0001132792419067836, "loss": 0.0448, "step": 36230 }, { "epoch": 2.5146551507759796e-13, "grad_norm": 0.7416876554489136, "learning_rate": 0.00011325998235517547, "loss": 0.0429, "step": 36240 }, { "epoch": 2.5153490401663703e-13, "grad_norm": 0.8367570638656616, "learning_rate": 0.00011324071964250055, "loss": 0.0441, "step": 36250 }, { "epoch": 2.516042929556761e-13, "grad_norm": 0.7926416993141174, "learning_rate": 0.00011322145377073837, "loss": 0.0443, "step": 36260 }, { "epoch": 2.5167368189471517e-13, "grad_norm": 0.7438994646072388, "learning_rate": 0.00011320218474186881, "loss": 0.0455, "step": 36270 }, { "epoch": 2.5174307083375425e-13, "grad_norm": 0.7523229718208313, "learning_rate": 0.00011318291255787206, "loss": 0.0422, "step": 36280 }, { "epoch": 2.518124597727933e-13, "grad_norm": 0.7646626830101013, "learning_rate": 0.00011316363722072866, "loss": 0.0469, "step": 36290 }, { "epoch": 2.518818487118324e-13, "grad_norm": 0.813685417175293, "learning_rate": 0.0001131443587324194, "loss": 0.0483, "step": 36300 }, { "epoch": 2.5195123765087146e-13, "grad_norm": 0.7680483460426331, "learning_rate": 0.0001131250770949255, "loss": 0.0451, "step": 36310 }, { "epoch": 2.5202062658991053e-13, "grad_norm": 0.7851223945617676, "learning_rate": 0.00011310579231022842, "loss": 0.0443, "step": 36320 }, { "epoch": 2.520900155289496e-13, "grad_norm": 0.872170627117157, "learning_rate": 0.00011308650438030997, "loss": 0.0446, "step": 36330 }, { "epoch": 2.521594044679887e-13, "grad_norm": 0.7305863499641418, "learning_rate": 0.00011306721330715226, "loss": 0.0437, "step": 36340 }, { "epoch": 2.5222879340702775e-13, "grad_norm": 0.7049211263656616, "learning_rate": 0.0001130479190927378, "loss": 0.0434, "step": 36350 }, { "epoch": 2.522981823460668e-13, "grad_norm": 0.8046408891677856, "learning_rate": 0.00011302862173904938, "loss": 0.0447, "step": 36360 }, { "epoch": 2.523675712851059e-13, "grad_norm": 0.7882811427116394, "learning_rate": 0.00011300932124807004, "loss": 0.0483, "step": 36370 }, { "epoch": 2.5243696022414497e-13, "grad_norm": 0.7468268275260925, "learning_rate": 0.00011299001762178324, "loss": 0.0438, "step": 36380 }, { "epoch": 2.5250634916318404e-13, "grad_norm": 0.692488431930542, "learning_rate": 0.00011297071086217273, "loss": 0.0436, "step": 36390 }, { "epoch": 2.525757381022231e-13, "grad_norm": 0.7074691653251648, "learning_rate": 0.00011295140097122257, "loss": 0.0437, "step": 36400 }, { "epoch": 2.526451270412622e-13, "grad_norm": 0.7806397676467896, "learning_rate": 0.00011293208795091717, "loss": 0.0449, "step": 36410 }, { "epoch": 2.5271451598030126e-13, "grad_norm": 0.8508927822113037, "learning_rate": 0.00011291277180324121, "loss": 0.0447, "step": 36420 }, { "epoch": 2.5278390491934033e-13, "grad_norm": 0.6568777561187744, "learning_rate": 0.00011289345253017976, "loss": 0.0434, "step": 36430 }, { "epoch": 2.528532938583794e-13, "grad_norm": 0.6976693272590637, "learning_rate": 0.00011287413013371816, "loss": 0.0434, "step": 36440 }, { "epoch": 2.5292268279741847e-13, "grad_norm": 0.7545168399810791, "learning_rate": 0.0001128548046158421, "loss": 0.0451, "step": 36450 }, { "epoch": 2.5299207173645755e-13, "grad_norm": 0.7744401693344116, "learning_rate": 0.00011283547597853755, "loss": 0.0479, "step": 36460 }, { "epoch": 2.530614606754966e-13, "grad_norm": 0.7337263226509094, "learning_rate": 0.00011281614422379085, "loss": 0.0449, "step": 36470 }, { "epoch": 2.531308496145357e-13, "grad_norm": 0.7143024802207947, "learning_rate": 0.00011279680935358863, "loss": 0.0434, "step": 36480 }, { "epoch": 2.5320023855357476e-13, "grad_norm": 0.814141571521759, "learning_rate": 0.00011277747136991785, "loss": 0.0447, "step": 36490 }, { "epoch": 2.5326962749261384e-13, "grad_norm": 0.7208200693130493, "learning_rate": 0.0001127581302747658, "loss": 0.0419, "step": 36500 }, { "epoch": 2.533390164316529e-13, "grad_norm": 0.7994378209114075, "learning_rate": 0.00011273878607012005, "loss": 0.0419, "step": 36510 }, { "epoch": 2.53408405370692e-13, "grad_norm": 0.7048152089118958, "learning_rate": 0.00011271943875796853, "loss": 0.0441, "step": 36520 }, { "epoch": 2.5347779430973105e-13, "grad_norm": 0.7034547924995422, "learning_rate": 0.00011270008834029953, "loss": 0.0455, "step": 36530 }, { "epoch": 2.535471832487701e-13, "grad_norm": 0.7125654220581055, "learning_rate": 0.00011268073481910152, "loss": 0.043, "step": 36540 }, { "epoch": 2.536165721878092e-13, "grad_norm": 0.8350516557693481, "learning_rate": 0.00011266137819636344, "loss": 0.0447, "step": 36550 }, { "epoch": 2.5368596112684827e-13, "grad_norm": 0.7522575855255127, "learning_rate": 0.00011264201847407444, "loss": 0.0423, "step": 36560 }, { "epoch": 2.5375535006588734e-13, "grad_norm": 0.7657331228256226, "learning_rate": 0.00011262265565422405, "loss": 0.0446, "step": 36570 }, { "epoch": 2.538247390049264e-13, "grad_norm": 0.8130709528923035, "learning_rate": 0.00011260328973880212, "loss": 0.0458, "step": 36580 }, { "epoch": 2.538941279439655e-13, "grad_norm": 0.7408849596977234, "learning_rate": 0.00011258392072979875, "loss": 0.0416, "step": 36590 }, { "epoch": 2.5396351688300456e-13, "grad_norm": 0.8142774105072021, "learning_rate": 0.00011256454862920447, "loss": 0.0454, "step": 36600 }, { "epoch": 2.5403290582204363e-13, "grad_norm": 0.9213258028030396, "learning_rate": 0.00011254517343901004, "loss": 0.0456, "step": 36610 }, { "epoch": 2.541022947610827e-13, "grad_norm": 0.8262038826942444, "learning_rate": 0.00011252579516120654, "loss": 0.0456, "step": 36620 }, { "epoch": 2.541716837001218e-13, "grad_norm": 0.7368107438087463, "learning_rate": 0.00011250641379778542, "loss": 0.0452, "step": 36630 }, { "epoch": 2.5424107263916085e-13, "grad_norm": 0.7294831871986389, "learning_rate": 0.00011248702935073842, "loss": 0.0415, "step": 36640 }, { "epoch": 2.543104615781999e-13, "grad_norm": 0.785311758518219, "learning_rate": 0.00011246764182205754, "loss": 0.0436, "step": 36650 }, { "epoch": 2.54379850517239e-13, "grad_norm": 0.7899620532989502, "learning_rate": 0.0001124482512137352, "loss": 0.0417, "step": 36660 }, { "epoch": 2.5444923945627806e-13, "grad_norm": 0.6585299968719482, "learning_rate": 0.00011242885752776406, "loss": 0.042, "step": 36670 }, { "epoch": 2.5451862839531714e-13, "grad_norm": 0.8031708002090454, "learning_rate": 0.00011240946076613716, "loss": 0.0445, "step": 36680 }, { "epoch": 2.545880173343562e-13, "grad_norm": 0.8570979833602905, "learning_rate": 0.00011239006093084777, "loss": 0.0442, "step": 36690 }, { "epoch": 2.546574062733953e-13, "grad_norm": 0.7573898434638977, "learning_rate": 0.00011237065802388956, "loss": 0.0452, "step": 36700 }, { "epoch": 2.5472679521243435e-13, "grad_norm": 0.7472285628318787, "learning_rate": 0.0001123512520472565, "loss": 0.0428, "step": 36710 }, { "epoch": 2.5479618415147343e-13, "grad_norm": 0.7808094024658203, "learning_rate": 0.0001123318430029428, "loss": 0.0434, "step": 36720 }, { "epoch": 2.548655730905125e-13, "grad_norm": 0.8095381855964661, "learning_rate": 0.00011231243089294306, "loss": 0.0441, "step": 36730 }, { "epoch": 2.5493496202955157e-13, "grad_norm": 0.783582866191864, "learning_rate": 0.00011229301571925221, "loss": 0.045, "step": 36740 }, { "epoch": 2.5500435096859064e-13, "grad_norm": 0.7786815166473389, "learning_rate": 0.00011227359748386543, "loss": 0.0414, "step": 36750 }, { "epoch": 2.550737399076297e-13, "grad_norm": 0.6940621733665466, "learning_rate": 0.00011225417618877825, "loss": 0.0439, "step": 36760 }, { "epoch": 2.551431288466688e-13, "grad_norm": 0.774778425693512, "learning_rate": 0.0001122347518359865, "loss": 0.0467, "step": 36770 }, { "epoch": 2.5521251778570786e-13, "grad_norm": 0.9243810772895813, "learning_rate": 0.00011221532442748635, "loss": 0.0452, "step": 36780 }, { "epoch": 2.5528190672474693e-13, "grad_norm": 0.8167636394500732, "learning_rate": 0.00011219589396527426, "loss": 0.0433, "step": 36790 }, { "epoch": 2.55351295663786e-13, "grad_norm": 0.7319698929786682, "learning_rate": 0.00011217646045134703, "loss": 0.0437, "step": 36800 }, { "epoch": 2.554206846028251e-13, "grad_norm": 0.7393870949745178, "learning_rate": 0.00011215702388770174, "loss": 0.042, "step": 36810 }, { "epoch": 2.5549007354186415e-13, "grad_norm": 0.7555144429206848, "learning_rate": 0.00011213758427633581, "loss": 0.041, "step": 36820 }, { "epoch": 2.555594624809032e-13, "grad_norm": 0.8329355120658875, "learning_rate": 0.00011211814161924693, "loss": 0.0422, "step": 36830 }, { "epoch": 2.556288514199423e-13, "grad_norm": 0.7565129399299622, "learning_rate": 0.00011209869591843319, "loss": 0.0437, "step": 36840 }, { "epoch": 2.5569824035898137e-13, "grad_norm": 0.8233445286750793, "learning_rate": 0.00011207924717589288, "loss": 0.0428, "step": 36850 }, { "epoch": 2.5576762929802044e-13, "grad_norm": 0.7969838380813599, "learning_rate": 0.00011205979539362471, "loss": 0.0432, "step": 36860 }, { "epoch": 2.558370182370595e-13, "grad_norm": 0.7922124862670898, "learning_rate": 0.00011204034057362764, "loss": 0.0429, "step": 36870 }, { "epoch": 2.559064071760986e-13, "grad_norm": 0.7932674884796143, "learning_rate": 0.00011202088271790093, "loss": 0.0442, "step": 36880 }, { "epoch": 2.5597579611513765e-13, "grad_norm": 0.7844447493553162, "learning_rate": 0.0001120014218284442, "loss": 0.0449, "step": 36890 }, { "epoch": 2.5604518505417673e-13, "grad_norm": 0.720670759677887, "learning_rate": 0.00011198195790725736, "loss": 0.0439, "step": 36900 }, { "epoch": 2.561145739932158e-13, "grad_norm": 0.7790853977203369, "learning_rate": 0.00011196249095634064, "loss": 0.0442, "step": 36910 }, { "epoch": 2.5618396293225487e-13, "grad_norm": 0.7758981585502625, "learning_rate": 0.00011194302097769454, "loss": 0.0438, "step": 36920 }, { "epoch": 2.5625335187129394e-13, "grad_norm": 0.7043635249137878, "learning_rate": 0.00011192354797331992, "loss": 0.0472, "step": 36930 }, { "epoch": 2.56322740810333e-13, "grad_norm": 0.7519229054450989, "learning_rate": 0.00011190407194521794, "loss": 0.0444, "step": 36940 }, { "epoch": 2.563921297493721e-13, "grad_norm": 0.8100257515907288, "learning_rate": 0.00011188459289539005, "loss": 0.0443, "step": 36950 }, { "epoch": 2.5646151868841116e-13, "grad_norm": 0.6960688233375549, "learning_rate": 0.00011186511082583805, "loss": 0.0437, "step": 36960 }, { "epoch": 2.5653090762745023e-13, "grad_norm": 0.6597402691841125, "learning_rate": 0.00011184562573856402, "loss": 0.0409, "step": 36970 }, { "epoch": 2.566002965664893e-13, "grad_norm": 0.733822226524353, "learning_rate": 0.00011182613763557032, "loss": 0.0419, "step": 36980 }, { "epoch": 2.566696855055284e-13, "grad_norm": 0.7811447978019714, "learning_rate": 0.00011180664651885971, "loss": 0.0419, "step": 36990 }, { "epoch": 2.5673907444456745e-13, "grad_norm": 0.726276159286499, "learning_rate": 0.00011178715239043516, "loss": 0.0425, "step": 37000 }, { "epoch": 2.568084633836065e-13, "grad_norm": 0.8593469262123108, "learning_rate": 0.00011176765525230003, "loss": 0.0425, "step": 37010 }, { "epoch": 2.568778523226456e-13, "grad_norm": 0.7737381458282471, "learning_rate": 0.00011174815510645793, "loss": 0.0434, "step": 37020 }, { "epoch": 2.5694724126168467e-13, "grad_norm": 0.8236024379730225, "learning_rate": 0.0001117286519549128, "loss": 0.0422, "step": 37030 }, { "epoch": 2.5701663020072374e-13, "grad_norm": 0.7641502618789673, "learning_rate": 0.00011170914579966893, "loss": 0.0435, "step": 37040 }, { "epoch": 2.570860191397628e-13, "grad_norm": 0.7733798027038574, "learning_rate": 0.00011168963664273082, "loss": 0.0448, "step": 37050 }, { "epoch": 2.571554080788019e-13, "grad_norm": 0.7094228267669678, "learning_rate": 0.00011167012448610338, "loss": 0.0418, "step": 37060 }, { "epoch": 2.5722479701784096e-13, "grad_norm": 0.753429651260376, "learning_rate": 0.00011165060933179183, "loss": 0.045, "step": 37070 }, { "epoch": 2.5729418595688003e-13, "grad_norm": 0.8617067337036133, "learning_rate": 0.00011163109118180157, "loss": 0.0449, "step": 37080 }, { "epoch": 2.573635748959191e-13, "grad_norm": 0.801604151725769, "learning_rate": 0.00011161157003813844, "loss": 0.0434, "step": 37090 }, { "epoch": 2.5743296383495817e-13, "grad_norm": 0.7768362760543823, "learning_rate": 0.00011159204590280853, "loss": 0.0427, "step": 37100 }, { "epoch": 2.5750235277399725e-13, "grad_norm": 0.8763707876205444, "learning_rate": 0.00011157251877781822, "loss": 0.0429, "step": 37110 }, { "epoch": 2.575717417130363e-13, "grad_norm": 0.786838710308075, "learning_rate": 0.0001115529886651743, "loss": 0.0428, "step": 37120 }, { "epoch": 2.576411306520754e-13, "grad_norm": 0.7742676734924316, "learning_rate": 0.00011153345556688373, "loss": 0.0407, "step": 37130 }, { "epoch": 2.5771051959111446e-13, "grad_norm": 0.6850247979164124, "learning_rate": 0.00011151391948495386, "loss": 0.0408, "step": 37140 }, { "epoch": 2.5777990853015353e-13, "grad_norm": 0.8583899140357971, "learning_rate": 0.00011149438042139233, "loss": 0.0444, "step": 37150 }, { "epoch": 2.578492974691926e-13, "grad_norm": 0.7740747928619385, "learning_rate": 0.00011147483837820709, "loss": 0.0402, "step": 37160 }, { "epoch": 2.579186864082317e-13, "grad_norm": 0.6468548774719238, "learning_rate": 0.00011145529335740637, "loss": 0.0424, "step": 37170 }, { "epoch": 2.5798807534727075e-13, "grad_norm": 0.7552149891853333, "learning_rate": 0.00011143574536099873, "loss": 0.0424, "step": 37180 }, { "epoch": 2.580574642863098e-13, "grad_norm": 0.7311981320381165, "learning_rate": 0.00011141619439099305, "loss": 0.0434, "step": 37190 }, { "epoch": 2.581268532253489e-13, "grad_norm": 0.7115101218223572, "learning_rate": 0.00011139664044939847, "loss": 0.0424, "step": 37200 }, { "epoch": 2.5819624216438797e-13, "grad_norm": 0.7634279131889343, "learning_rate": 0.00011137708353822446, "loss": 0.0447, "step": 37210 }, { "epoch": 2.5826563110342704e-13, "grad_norm": 0.8367338180541992, "learning_rate": 0.00011135752365948082, "loss": 0.0421, "step": 37220 }, { "epoch": 2.583350200424661e-13, "grad_norm": 0.7839717268943787, "learning_rate": 0.0001113379608151776, "loss": 0.0442, "step": 37230 }, { "epoch": 2.584044089815052e-13, "grad_norm": 0.7752593755722046, "learning_rate": 0.00011131839500732524, "loss": 0.0452, "step": 37240 }, { "epoch": 2.5847379792054426e-13, "grad_norm": 0.8267584443092346, "learning_rate": 0.00011129882623793438, "loss": 0.0424, "step": 37250 }, { "epoch": 2.5854318685958333e-13, "grad_norm": 0.7908802032470703, "learning_rate": 0.00011127925450901602, "loss": 0.0424, "step": 37260 }, { "epoch": 2.586125757986224e-13, "grad_norm": 0.7588205337524414, "learning_rate": 0.00011125967982258152, "loss": 0.0425, "step": 37270 }, { "epoch": 2.5868196473766147e-13, "grad_norm": 0.7712082266807556, "learning_rate": 0.00011124010218064239, "loss": 0.041, "step": 37280 }, { "epoch": 2.5875135367670055e-13, "grad_norm": 0.8096905946731567, "learning_rate": 0.00011122052158521062, "loss": 0.0393, "step": 37290 }, { "epoch": 2.588207426157396e-13, "grad_norm": 0.7773475646972656, "learning_rate": 0.00011120093803829836, "loss": 0.0416, "step": 37300 }, { "epoch": 2.588901315547787e-13, "grad_norm": 0.7855134010314941, "learning_rate": 0.00011118135154191817, "loss": 0.042, "step": 37310 }, { "epoch": 2.5895952049381776e-13, "grad_norm": 0.7260875701904297, "learning_rate": 0.00011116176209808282, "loss": 0.0393, "step": 37320 }, { "epoch": 2.5902890943285684e-13, "grad_norm": 0.7116843461990356, "learning_rate": 0.00011114216970880548, "loss": 0.0415, "step": 37330 }, { "epoch": 2.590982983718959e-13, "grad_norm": 0.7602555751800537, "learning_rate": 0.00011112257437609954, "loss": 0.0416, "step": 37340 }, { "epoch": 2.59167687310935e-13, "grad_norm": 0.7090222835540771, "learning_rate": 0.00011110297610197871, "loss": 0.043, "step": 37350 }, { "epoch": 2.5923707624997405e-13, "grad_norm": 0.7550398111343384, "learning_rate": 0.00011108337488845709, "loss": 0.0431, "step": 37360 }, { "epoch": 2.593064651890131e-13, "grad_norm": 0.7602975964546204, "learning_rate": 0.00011106377073754894, "loss": 0.041, "step": 37370 }, { "epoch": 2.593758541280522e-13, "grad_norm": 0.8070722818374634, "learning_rate": 0.00011104416365126892, "loss": 0.0431, "step": 37380 }, { "epoch": 2.5944524306709127e-13, "grad_norm": 0.847748875617981, "learning_rate": 0.00011102455363163195, "loss": 0.0435, "step": 37390 }, { "epoch": 2.5951463200613034e-13, "grad_norm": 0.7916213870048523, "learning_rate": 0.00011100494068065328, "loss": 0.0432, "step": 37400 }, { "epoch": 2.595840209451694e-13, "grad_norm": 0.9397351145744324, "learning_rate": 0.00011098532480034844, "loss": 0.0409, "step": 37410 }, { "epoch": 2.596534098842085e-13, "grad_norm": 0.7019903063774109, "learning_rate": 0.00011096570599273325, "loss": 0.0395, "step": 37420 }, { "epoch": 2.5972279882324756e-13, "grad_norm": 0.6537598371505737, "learning_rate": 0.00011094608425982387, "loss": 0.0424, "step": 37430 }, { "epoch": 2.5979218776228663e-13, "grad_norm": 0.7244237661361694, "learning_rate": 0.00011092645960363673, "loss": 0.0408, "step": 37440 }, { "epoch": 2.598615767013257e-13, "grad_norm": 0.7477510571479797, "learning_rate": 0.00011090683202618857, "loss": 0.0422, "step": 37450 }, { "epoch": 2.599309656403648e-13, "grad_norm": 0.7877071499824524, "learning_rate": 0.00011088720152949642, "loss": 0.0413, "step": 37460 }, { "epoch": 2.6000035457940385e-13, "grad_norm": 0.7850227952003479, "learning_rate": 0.00011086756811557765, "loss": 0.0403, "step": 37470 }, { "epoch": 2.600697435184429e-13, "grad_norm": 0.791819155216217, "learning_rate": 0.00011084793178644986, "loss": 0.0408, "step": 37480 }, { "epoch": 2.60139132457482e-13, "grad_norm": 0.7353498339653015, "learning_rate": 0.000110828292544131, "loss": 0.0405, "step": 37490 }, { "epoch": 2.6020852139652106e-13, "grad_norm": 0.8242743611335754, "learning_rate": 0.00011080865039063932, "loss": 0.0412, "step": 37500 }, { "epoch": 2.6027791033556014e-13, "grad_norm": 0.8036930561065674, "learning_rate": 0.00011078900532799335, "loss": 0.0421, "step": 37510 }, { "epoch": 2.603472992745992e-13, "grad_norm": 0.7093186378479004, "learning_rate": 0.00011076935735821193, "loss": 0.0428, "step": 37520 }, { "epoch": 2.604166882136383e-13, "grad_norm": 0.7353605031967163, "learning_rate": 0.00011074970648331417, "loss": 0.0404, "step": 37530 }, { "epoch": 2.6048607715267735e-13, "grad_norm": 0.7044395208358765, "learning_rate": 0.00011073005270531954, "loss": 0.0434, "step": 37540 }, { "epoch": 2.6055546609171643e-13, "grad_norm": 0.7400542497634888, "learning_rate": 0.00011071039602624775, "loss": 0.0433, "step": 37550 }, { "epoch": 2.606248550307555e-13, "grad_norm": 0.6493906378746033, "learning_rate": 0.00011069073644811882, "loss": 0.0406, "step": 37560 }, { "epoch": 2.6069424396979457e-13, "grad_norm": 0.7702745795249939, "learning_rate": 0.00011067107397295311, "loss": 0.0399, "step": 37570 }, { "epoch": 2.6076363290883364e-13, "grad_norm": 0.8314536809921265, "learning_rate": 0.00011065140860277124, "loss": 0.0412, "step": 37580 }, { "epoch": 2.608330218478727e-13, "grad_norm": 0.7488458156585693, "learning_rate": 0.0001106317403395941, "loss": 0.0402, "step": 37590 }, { "epoch": 2.609024107869118e-13, "grad_norm": 0.7102478742599487, "learning_rate": 0.00011061206918544291, "loss": 0.0396, "step": 37600 }, { "epoch": 2.6097179972595086e-13, "grad_norm": 0.7601866722106934, "learning_rate": 0.00011059239514233925, "loss": 0.041, "step": 37610 }, { "epoch": 2.6104118866498993e-13, "grad_norm": 0.6878970861434937, "learning_rate": 0.00011057271821230487, "loss": 0.0424, "step": 37620 }, { "epoch": 2.61110577604029e-13, "grad_norm": 0.7222480773925781, "learning_rate": 0.00011055303839736192, "loss": 0.0391, "step": 37630 }, { "epoch": 2.611799665430681e-13, "grad_norm": 0.6583291888237, "learning_rate": 0.00011053335569953277, "loss": 0.0392, "step": 37640 }, { "epoch": 2.6124935548210715e-13, "grad_norm": 0.6682080030441284, "learning_rate": 0.00011051367012084013, "loss": 0.0405, "step": 37650 }, { "epoch": 2.613187444211462e-13, "grad_norm": 0.8018766641616821, "learning_rate": 0.00011049398166330703, "loss": 0.0425, "step": 37660 }, { "epoch": 2.613881333601853e-13, "grad_norm": 0.856321394443512, "learning_rate": 0.00011047429032895672, "loss": 0.0428, "step": 37670 }, { "epoch": 2.6145752229922437e-13, "grad_norm": 0.6616285443305969, "learning_rate": 0.00011045459611981284, "loss": 0.0411, "step": 37680 }, { "epoch": 2.6152691123826344e-13, "grad_norm": 0.7155362367630005, "learning_rate": 0.00011043489903789926, "loss": 0.0409, "step": 37690 }, { "epoch": 2.615963001773025e-13, "grad_norm": 0.7241118550300598, "learning_rate": 0.00011041519908524013, "loss": 0.0422, "step": 37700 }, { "epoch": 2.616656891163416e-13, "grad_norm": 0.732657253742218, "learning_rate": 0.00011039549626385998, "loss": 0.0429, "step": 37710 }, { "epoch": 2.6173507805538065e-13, "grad_norm": 0.6830222606658936, "learning_rate": 0.00011037579057578355, "loss": 0.0404, "step": 37720 }, { "epoch": 2.6180446699441973e-13, "grad_norm": 0.6491436958312988, "learning_rate": 0.00011035608202303592, "loss": 0.0396, "step": 37730 }, { "epoch": 2.618738559334588e-13, "grad_norm": 0.7140064835548401, "learning_rate": 0.00011033637060764242, "loss": 0.0411, "step": 37740 }, { "epoch": 2.6194324487249787e-13, "grad_norm": 0.7105810642242432, "learning_rate": 0.00011031665633162871, "loss": 0.0398, "step": 37750 }, { "epoch": 2.6201263381153694e-13, "grad_norm": 0.7955533266067505, "learning_rate": 0.00011029693919702078, "loss": 0.0393, "step": 37760 }, { "epoch": 2.62082022750576e-13, "grad_norm": 0.6719236969947815, "learning_rate": 0.00011027721920584484, "loss": 0.0404, "step": 37770 }, { "epoch": 2.621514116896151e-13, "grad_norm": 0.6630440354347229, "learning_rate": 0.00011025749636012744, "loss": 0.0415, "step": 37780 }, { "epoch": 2.6222080062865416e-13, "grad_norm": 0.6708546280860901, "learning_rate": 0.00011023777066189537, "loss": 0.0396, "step": 37790 }, { "epoch": 2.6229018956769323e-13, "grad_norm": 0.8149197101593018, "learning_rate": 0.00011021804211317581, "loss": 0.0406, "step": 37800 }, { "epoch": 2.623595785067323e-13, "grad_norm": 0.713606059551239, "learning_rate": 0.00011019831071599615, "loss": 0.0399, "step": 37810 }, { "epoch": 2.624289674457714e-13, "grad_norm": 0.7786809206008911, "learning_rate": 0.0001101785764723841, "loss": 0.0422, "step": 37820 }, { "epoch": 2.6249835638481045e-13, "grad_norm": 0.8202670812606812, "learning_rate": 0.00011015883938436766, "loss": 0.0441, "step": 37830 }, { "epoch": 2.625677453238495e-13, "grad_norm": 0.7502217888832092, "learning_rate": 0.00011013909945397512, "loss": 0.0393, "step": 37840 }, { "epoch": 2.626371342628886e-13, "grad_norm": 0.8009524941444397, "learning_rate": 0.00011011935668323504, "loss": 0.0426, "step": 37850 }, { "epoch": 2.6270652320192767e-13, "grad_norm": 0.8705453276634216, "learning_rate": 0.00011009961107417636, "loss": 0.0422, "step": 37860 }, { "epoch": 2.6277591214096674e-13, "grad_norm": 0.7841110825538635, "learning_rate": 0.00011007986262882819, "loss": 0.0414, "step": 37870 }, { "epoch": 2.628453010800058e-13, "grad_norm": 0.7332037091255188, "learning_rate": 0.00011006011134922, "loss": 0.04, "step": 37880 }, { "epoch": 2.629146900190449e-13, "grad_norm": 0.7183026671409607, "learning_rate": 0.00011004035723738156, "loss": 0.0377, "step": 37890 }, { "epoch": 2.6298407895808396e-13, "grad_norm": 0.7447527050971985, "learning_rate": 0.0001100206002953429, "loss": 0.0401, "step": 37900 }, { "epoch": 2.6305346789712303e-13, "grad_norm": 0.7273229360580444, "learning_rate": 0.00011000084052513437, "loss": 0.0395, "step": 37910 }, { "epoch": 2.631228568361621e-13, "grad_norm": 0.6257280111312866, "learning_rate": 0.00010998107792878658, "loss": 0.0399, "step": 37920 }, { "epoch": 2.6319224577520117e-13, "grad_norm": 0.7365231513977051, "learning_rate": 0.00010996131250833045, "loss": 0.0409, "step": 37930 }, { "epoch": 2.6326163471424024e-13, "grad_norm": 0.7329486012458801, "learning_rate": 0.0001099415442657972, "loss": 0.0395, "step": 37940 }, { "epoch": 2.633310236532793e-13, "grad_norm": 0.7237637042999268, "learning_rate": 0.00010992177320321827, "loss": 0.0397, "step": 37950 }, { "epoch": 2.634004125923184e-13, "grad_norm": 0.6971651911735535, "learning_rate": 0.00010990199932262548, "loss": 0.0388, "step": 37960 }, { "epoch": 2.6346980153135746e-13, "grad_norm": 0.7840548753738403, "learning_rate": 0.00010988222262605092, "loss": 0.0413, "step": 37970 }, { "epoch": 2.6353919047039653e-13, "grad_norm": 0.7893508076667786, "learning_rate": 0.00010986244311552691, "loss": 0.0413, "step": 37980 }, { "epoch": 2.636085794094356e-13, "grad_norm": 0.7460846304893494, "learning_rate": 0.00010984266079308617, "loss": 0.0424, "step": 37990 }, { "epoch": 2.636779683484747e-13, "grad_norm": 0.7615764737129211, "learning_rate": 0.00010982287566076157, "loss": 0.0391, "step": 38000 }, { "epoch": 2.6374735728751375e-13, "grad_norm": 0.6650166511535645, "learning_rate": 0.00010980308772058639, "loss": 0.0415, "step": 38010 }, { "epoch": 2.638167462265528e-13, "grad_norm": 0.737170398235321, "learning_rate": 0.00010978329697459412, "loss": 0.0431, "step": 38020 }, { "epoch": 2.638861351655919e-13, "grad_norm": 0.8594196438789368, "learning_rate": 0.00010976350342481858, "loss": 0.0405, "step": 38030 }, { "epoch": 2.6395552410463097e-13, "grad_norm": 0.7690625786781311, "learning_rate": 0.00010974370707329385, "loss": 0.0389, "step": 38040 }, { "epoch": 2.6402491304367004e-13, "grad_norm": 0.7212774753570557, "learning_rate": 0.00010972390792205434, "loss": 0.0386, "step": 38050 }, { "epoch": 2.640943019827091e-13, "grad_norm": 0.7261438965797424, "learning_rate": 0.00010970410597313469, "loss": 0.0383, "step": 38060 }, { "epoch": 2.641636909217482e-13, "grad_norm": 0.7307690382003784, "learning_rate": 0.0001096843012285699, "loss": 0.0373, "step": 38070 }, { "epoch": 2.6423307986078726e-13, "grad_norm": 0.7935174107551575, "learning_rate": 0.00010966449369039518, "loss": 0.0403, "step": 38080 }, { "epoch": 2.6430246879982633e-13, "grad_norm": 0.6678755283355713, "learning_rate": 0.00010964468336064605, "loss": 0.0395, "step": 38090 }, { "epoch": 2.643718577388654e-13, "grad_norm": 0.7394644021987915, "learning_rate": 0.00010962487024135836, "loss": 0.0385, "step": 38100 }, { "epoch": 2.6444124667790447e-13, "grad_norm": 0.7204469442367554, "learning_rate": 0.00010960505433456822, "loss": 0.0384, "step": 38110 }, { "epoch": 2.6451063561694355e-13, "grad_norm": 0.7502350211143494, "learning_rate": 0.00010958523564231201, "loss": 0.0387, "step": 38120 }, { "epoch": 2.645800245559826e-13, "grad_norm": 0.7523340582847595, "learning_rate": 0.00010956541416662641, "loss": 0.0409, "step": 38130 }, { "epoch": 2.646494134950217e-13, "grad_norm": 0.7483212351799011, "learning_rate": 0.00010954558990954837, "loss": 0.0411, "step": 38140 }, { "epoch": 2.6471880243406076e-13, "grad_norm": 0.6879724264144897, "learning_rate": 0.00010952576287311516, "loss": 0.0396, "step": 38150 }, { "epoch": 2.6478819137309983e-13, "grad_norm": 0.7311705350875854, "learning_rate": 0.00010950593305936433, "loss": 0.0402, "step": 38160 }, { "epoch": 2.648575803121389e-13, "grad_norm": 0.6889681816101074, "learning_rate": 0.00010948610047033369, "loss": 0.0404, "step": 38170 }, { "epoch": 2.64926969251178e-13, "grad_norm": 0.6539055705070496, "learning_rate": 0.0001094662651080613, "loss": 0.0421, "step": 38180 }, { "epoch": 2.6499635819021705e-13, "grad_norm": 0.6608389019966125, "learning_rate": 0.00010944642697458563, "loss": 0.039, "step": 38190 }, { "epoch": 2.650657471292561e-13, "grad_norm": 0.7250911593437195, "learning_rate": 0.0001094265860719453, "loss": 0.0387, "step": 38200 }, { "epoch": 2.651351360682952e-13, "grad_norm": 0.6206827759742737, "learning_rate": 0.0001094067424021793, "loss": 0.0384, "step": 38210 }, { "epoch": 2.6520452500733427e-13, "grad_norm": 0.6640780568122864, "learning_rate": 0.00010938689596732687, "loss": 0.0367, "step": 38220 }, { "epoch": 2.6527391394637334e-13, "grad_norm": 0.7070741653442383, "learning_rate": 0.00010936704676942754, "loss": 0.0384, "step": 38230 }, { "epoch": 2.653433028854124e-13, "grad_norm": 0.7594460248947144, "learning_rate": 0.00010934719481052113, "loss": 0.0397, "step": 38240 }, { "epoch": 2.654126918244515e-13, "grad_norm": 0.6739788055419922, "learning_rate": 0.0001093273400926477, "loss": 0.0388, "step": 38250 }, { "epoch": 2.6548208076349056e-13, "grad_norm": 0.8082127571105957, "learning_rate": 0.00010930748261784768, "loss": 0.0382, "step": 38260 }, { "epoch": 2.6555146970252963e-13, "grad_norm": 0.7141673564910889, "learning_rate": 0.00010928762238816172, "loss": 0.0389, "step": 38270 }, { "epoch": 2.656208586415687e-13, "grad_norm": 0.7581074833869934, "learning_rate": 0.00010926775940563076, "loss": 0.038, "step": 38280 }, { "epoch": 2.6569024758060777e-13, "grad_norm": 0.7162062525749207, "learning_rate": 0.00010924789367229603, "loss": 0.0396, "step": 38290 }, { "epoch": 2.6575963651964685e-13, "grad_norm": 0.7469784617424011, "learning_rate": 0.00010922802519019905, "loss": 0.0417, "step": 38300 }, { "epoch": 2.658290254586859e-13, "grad_norm": 0.7288835048675537, "learning_rate": 0.0001092081539613816, "loss": 0.0384, "step": 38310 }, { "epoch": 2.65898414397725e-13, "grad_norm": 0.7261838912963867, "learning_rate": 0.00010918827998788577, "loss": 0.042, "step": 38320 }, { "epoch": 2.6596780333676406e-13, "grad_norm": 0.7841137647628784, "learning_rate": 0.00010916840327175393, "loss": 0.0406, "step": 38330 }, { "epoch": 2.6603719227580314e-13, "grad_norm": 0.7453241944313049, "learning_rate": 0.0001091485238150287, "loss": 0.0398, "step": 38340 }, { "epoch": 2.661065812148422e-13, "grad_norm": 0.6938133239746094, "learning_rate": 0.000109128641619753, "loss": 0.038, "step": 38350 }, { "epoch": 2.661759701538813e-13, "grad_norm": 0.7309705018997192, "learning_rate": 0.0001091087566879701, "loss": 0.0374, "step": 38360 }, { "epoch": 2.6624535909292035e-13, "grad_norm": 0.6943206191062927, "learning_rate": 0.00010908886902172342, "loss": 0.0372, "step": 38370 }, { "epoch": 2.663147480319594e-13, "grad_norm": 0.757266104221344, "learning_rate": 0.00010906897862305673, "loss": 0.0371, "step": 38380 }, { "epoch": 2.663841369709985e-13, "grad_norm": 0.6695705652236938, "learning_rate": 0.00010904908549401408, "loss": 0.0378, "step": 38390 }, { "epoch": 2.6645352591003757e-13, "grad_norm": 0.8496259450912476, "learning_rate": 0.00010902918963663984, "loss": 0.0411, "step": 38400 }, { "epoch": 2.6652291484907664e-13, "grad_norm": 0.7625584006309509, "learning_rate": 0.00010900929105297857, "loss": 0.0374, "step": 38410 }, { "epoch": 2.665923037881157e-13, "grad_norm": 0.6226692795753479, "learning_rate": 0.00010898938974507519, "loss": 0.0383, "step": 38420 }, { "epoch": 2.666616927271548e-13, "grad_norm": 0.6994080543518066, "learning_rate": 0.00010896948571497483, "loss": 0.0377, "step": 38430 }, { "epoch": 2.6673108166619386e-13, "grad_norm": 0.6904315948486328, "learning_rate": 0.00010894957896472298, "loss": 0.0387, "step": 38440 }, { "epoch": 2.6680047060523293e-13, "grad_norm": 0.7010007500648499, "learning_rate": 0.00010892966949636537, "loss": 0.0389, "step": 38450 }, { "epoch": 2.66869859544272e-13, "grad_norm": 0.7240594625473022, "learning_rate": 0.00010890975731194799, "loss": 0.041, "step": 38460 }, { "epoch": 2.669392484833111e-13, "grad_norm": 0.8077466487884521, "learning_rate": 0.00010888984241351711, "loss": 0.0391, "step": 38470 }, { "epoch": 2.6700863742235015e-13, "grad_norm": 0.746380627155304, "learning_rate": 0.00010886992480311932, "loss": 0.0399, "step": 38480 }, { "epoch": 2.670780263613892e-13, "grad_norm": 0.7053259015083313, "learning_rate": 0.00010885000448280146, "loss": 0.0413, "step": 38490 }, { "epoch": 2.671474153004283e-13, "grad_norm": 0.7323825359344482, "learning_rate": 0.00010883008145461065, "loss": 0.0385, "step": 38500 }, { "epoch": 2.6721680423946736e-13, "grad_norm": 0.7076690196990967, "learning_rate": 0.0001088101557205943, "loss": 0.0373, "step": 38510 }, { "epoch": 2.6728619317850644e-13, "grad_norm": 0.7192633152008057, "learning_rate": 0.00010879022728280008, "loss": 0.0374, "step": 38520 }, { "epoch": 2.673555821175455e-13, "grad_norm": 0.7428314685821533, "learning_rate": 0.00010877029614327595, "loss": 0.0372, "step": 38530 }, { "epoch": 2.674249710565846e-13, "grad_norm": 0.7637856006622314, "learning_rate": 0.00010875036230407015, "loss": 0.0359, "step": 38540 }, { "epoch": 2.6749435999562365e-13, "grad_norm": 0.748988151550293, "learning_rate": 0.00010873042576723116, "loss": 0.0389, "step": 38550 }, { "epoch": 2.6756374893466273e-13, "grad_norm": 0.7828423976898193, "learning_rate": 0.00010871048653480781, "loss": 0.0389, "step": 38560 }, { "epoch": 2.676331378737018e-13, "grad_norm": 0.7143635749816895, "learning_rate": 0.00010869054460884914, "loss": 0.0372, "step": 38570 }, { "epoch": 2.6770252681274087e-13, "grad_norm": 0.6875312328338623, "learning_rate": 0.00010867059999140454, "loss": 0.0384, "step": 38580 }, { "epoch": 2.6777191575177994e-13, "grad_norm": 0.7112404108047485, "learning_rate": 0.00010865065268452356, "loss": 0.0373, "step": 38590 }, { "epoch": 2.67841304690819e-13, "grad_norm": 0.6843749284744263, "learning_rate": 0.00010863070269025614, "loss": 0.0389, "step": 38600 }, { "epoch": 2.679106936298581e-13, "grad_norm": 0.7227162718772888, "learning_rate": 0.00010861075001065245, "loss": 0.0401, "step": 38610 }, { "epoch": 2.6798008256889716e-13, "grad_norm": 0.7273902297019958, "learning_rate": 0.00010859079464776292, "loss": 0.0383, "step": 38620 }, { "epoch": 2.6804947150793623e-13, "grad_norm": 0.7657022476196289, "learning_rate": 0.0001085708366036383, "loss": 0.0407, "step": 38630 }, { "epoch": 2.681188604469753e-13, "grad_norm": 0.7778539657592773, "learning_rate": 0.00010855087588032955, "loss": 0.0401, "step": 38640 }, { "epoch": 2.681882493860144e-13, "grad_norm": 0.7829164862632751, "learning_rate": 0.00010853091247988797, "loss": 0.0409, "step": 38650 }, { "epoch": 2.6825763832505345e-13, "grad_norm": 0.8642181158065796, "learning_rate": 0.00010851094640436514, "loss": 0.0386, "step": 38660 }, { "epoch": 2.683270272640925e-13, "grad_norm": 0.6468228101730347, "learning_rate": 0.00010849097765581283, "loss": 0.0364, "step": 38670 }, { "epoch": 2.683964162031316e-13, "grad_norm": 0.6140094995498657, "learning_rate": 0.00010847100623628318, "loss": 0.0375, "step": 38680 }, { "epoch": 2.6846580514217067e-13, "grad_norm": 0.7076430916786194, "learning_rate": 0.00010845103214782854, "loss": 0.0366, "step": 38690 }, { "epoch": 2.6853519408120974e-13, "grad_norm": 0.7657477259635925, "learning_rate": 0.00010843105539250156, "loss": 0.0375, "step": 38700 }, { "epoch": 2.686045830202488e-13, "grad_norm": 0.752364456653595, "learning_rate": 0.00010841107597235517, "loss": 0.0385, "step": 38710 }, { "epoch": 2.686739719592879e-13, "grad_norm": 0.7267125248908997, "learning_rate": 0.00010839109388944257, "loss": 0.0378, "step": 38720 }, { "epoch": 2.6874336089832696e-13, "grad_norm": 0.7988079786300659, "learning_rate": 0.00010837110914581723, "loss": 0.0385, "step": 38730 }, { "epoch": 2.6881274983736603e-13, "grad_norm": 0.6736594438552856, "learning_rate": 0.00010835112174353288, "loss": 0.0382, "step": 38740 }, { "epoch": 2.688821387764051e-13, "grad_norm": 0.7552305459976196, "learning_rate": 0.00010833113168464354, "loss": 0.0378, "step": 38750 }, { "epoch": 2.6895152771544417e-13, "grad_norm": 0.740077793598175, "learning_rate": 0.00010831113897120352, "loss": 0.0385, "step": 38760 }, { "epoch": 2.6902091665448324e-13, "grad_norm": 0.709750235080719, "learning_rate": 0.00010829114360526735, "loss": 0.0397, "step": 38770 }, { "epoch": 2.690903055935223e-13, "grad_norm": 0.700766384601593, "learning_rate": 0.00010827114558888991, "loss": 0.0375, "step": 38780 }, { "epoch": 2.691596945325614e-13, "grad_norm": 0.6673919558525085, "learning_rate": 0.00010825114492412627, "loss": 0.0408, "step": 38790 }, { "epoch": 2.6922908347160046e-13, "grad_norm": 0.733097493648529, "learning_rate": 0.00010823114161303181, "loss": 0.0416, "step": 38800 }, { "epoch": 2.6929847241063953e-13, "grad_norm": 0.6632235646247864, "learning_rate": 0.00010821113565766218, "loss": 0.0382, "step": 38810 }, { "epoch": 2.693678613496786e-13, "grad_norm": 0.7574519515037537, "learning_rate": 0.00010819112706007331, "loss": 0.0382, "step": 38820 }, { "epoch": 2.694372502887177e-13, "grad_norm": 0.8036030530929565, "learning_rate": 0.00010817111582232144, "loss": 0.0383, "step": 38830 }, { "epoch": 2.6950663922775675e-13, "grad_norm": 0.696149468421936, "learning_rate": 0.00010815110194646296, "loss": 0.0362, "step": 38840 }, { "epoch": 2.695760281667958e-13, "grad_norm": 0.6613438129425049, "learning_rate": 0.00010813108543455464, "loss": 0.0354, "step": 38850 }, { "epoch": 2.696454171058349e-13, "grad_norm": 0.7201342582702637, "learning_rate": 0.0001081110662886535, "loss": 0.0364, "step": 38860 }, { "epoch": 2.6971480604487397e-13, "grad_norm": 0.6572126746177673, "learning_rate": 0.00010809104451081682, "loss": 0.0396, "step": 38870 }, { "epoch": 2.6978419498391304e-13, "grad_norm": 0.6810831427574158, "learning_rate": 0.00010807102010310211, "loss": 0.0369, "step": 38880 }, { "epoch": 2.698535839229521e-13, "grad_norm": 0.6590979695320129, "learning_rate": 0.00010805099306756722, "loss": 0.0377, "step": 38890 }, { "epoch": 2.699229728619912e-13, "grad_norm": 0.6218493580818176, "learning_rate": 0.00010803096340627025, "loss": 0.0381, "step": 38900 }, { "epoch": 2.6999236180103026e-13, "grad_norm": 0.7553439736366272, "learning_rate": 0.00010801093112126957, "loss": 0.0392, "step": 38910 }, { "epoch": 2.7006175074006933e-13, "grad_norm": 0.8107624053955078, "learning_rate": 0.00010799089621462375, "loss": 0.0388, "step": 38920 }, { "epoch": 2.701311396791084e-13, "grad_norm": 0.6371167302131653, "learning_rate": 0.00010797085868839172, "loss": 0.0382, "step": 38930 }, { "epoch": 2.7020052861814747e-13, "grad_norm": 0.7258062362670898, "learning_rate": 0.00010795081854463268, "loss": 0.0383, "step": 38940 }, { "epoch": 2.7026991755718655e-13, "grad_norm": 0.7333642244338989, "learning_rate": 0.00010793077578540601, "loss": 0.0404, "step": 38950 }, { "epoch": 2.703393064962256e-13, "grad_norm": 0.7254760265350342, "learning_rate": 0.00010791073041277145, "loss": 0.0411, "step": 38960 }, { "epoch": 2.704086954352647e-13, "grad_norm": 0.7006233930587769, "learning_rate": 0.00010789068242878896, "loss": 0.0384, "step": 38970 }, { "epoch": 2.7047808437430376e-13, "grad_norm": 0.6811196208000183, "learning_rate": 0.00010787063183551882, "loss": 0.0374, "step": 38980 }, { "epoch": 2.7054747331334283e-13, "grad_norm": 0.6941555738449097, "learning_rate": 0.00010785057863502148, "loss": 0.0385, "step": 38990 }, { "epoch": 2.706168622523819e-13, "grad_norm": 0.6089948415756226, "learning_rate": 0.00010783052282935775, "loss": 0.0354, "step": 39000 }, { "epoch": 2.70686251191421e-13, "grad_norm": 0.731636643409729, "learning_rate": 0.00010781046442058868, "loss": 0.0353, "step": 39010 }, { "epoch": 2.7075564013046005e-13, "grad_norm": 0.6462688446044922, "learning_rate": 0.00010779040341077558, "loss": 0.0367, "step": 39020 }, { "epoch": 2.708250290694991e-13, "grad_norm": 0.6721126437187195, "learning_rate": 0.00010777033980198002, "loss": 0.0381, "step": 39030 }, { "epoch": 2.708944180085382e-13, "grad_norm": 0.7009062767028809, "learning_rate": 0.00010775027359626385, "loss": 0.0371, "step": 39040 }, { "epoch": 2.7096380694757727e-13, "grad_norm": 0.8181266784667969, "learning_rate": 0.00010773020479568921, "loss": 0.0386, "step": 39050 }, { "epoch": 2.7103319588661634e-13, "grad_norm": 0.6715419888496399, "learning_rate": 0.00010771013340231845, "loss": 0.036, "step": 39060 }, { "epoch": 2.711025848256554e-13, "grad_norm": 0.7427701354026794, "learning_rate": 0.00010769005941821424, "loss": 0.0387, "step": 39070 }, { "epoch": 2.711719737646945e-13, "grad_norm": 0.7354913949966431, "learning_rate": 0.00010766998284543946, "loss": 0.0398, "step": 39080 }, { "epoch": 2.7124136270373356e-13, "grad_norm": 0.720928430557251, "learning_rate": 0.00010764990368605734, "loss": 0.0368, "step": 39090 }, { "epoch": 2.7131075164277263e-13, "grad_norm": 0.7903634905815125, "learning_rate": 0.0001076298219421313, "loss": 0.0401, "step": 39100 }, { "epoch": 2.713801405818117e-13, "grad_norm": 0.8399421572685242, "learning_rate": 0.00010760973761572504, "loss": 0.0394, "step": 39110 }, { "epoch": 2.7144952952085077e-13, "grad_norm": 0.7468748092651367, "learning_rate": 0.00010758965070890255, "loss": 0.0389, "step": 39120 }, { "epoch": 2.7151891845988985e-13, "grad_norm": 0.7008014917373657, "learning_rate": 0.00010756956122372807, "loss": 0.038, "step": 39130 }, { "epoch": 2.715883073989289e-13, "grad_norm": 0.6916550397872925, "learning_rate": 0.00010754946916226613, "loss": 0.0362, "step": 39140 }, { "epoch": 2.71657696337968e-13, "grad_norm": 0.7078437209129333, "learning_rate": 0.00010752937452658148, "loss": 0.0377, "step": 39150 }, { "epoch": 2.7172708527700706e-13, "grad_norm": 0.6804105043411255, "learning_rate": 0.00010750927731873914, "loss": 0.0353, "step": 39160 }, { "epoch": 2.7179647421604614e-13, "grad_norm": 0.6085290908813477, "learning_rate": 0.00010748917754080445, "loss": 0.036, "step": 39170 }, { "epoch": 2.718658631550852e-13, "grad_norm": 0.67713463306427, "learning_rate": 0.0001074690751948429, "loss": 0.0377, "step": 39180 }, { "epoch": 2.719352520941243e-13, "grad_norm": 0.7305282354354858, "learning_rate": 0.00010744897028292042, "loss": 0.0374, "step": 39190 }, { "epoch": 2.7200464103316335e-13, "grad_norm": 0.7609127163887024, "learning_rate": 0.00010742886280710306, "loss": 0.0386, "step": 39200 }, { "epoch": 2.720740299722024e-13, "grad_norm": 0.7052116990089417, "learning_rate": 0.00010740875276945715, "loss": 0.037, "step": 39210 }, { "epoch": 2.721434189112415e-13, "grad_norm": 0.7559269666671753, "learning_rate": 0.00010738864017204933, "loss": 0.0373, "step": 39220 }, { "epoch": 2.7221280785028057e-13, "grad_norm": 0.7774202227592468, "learning_rate": 0.00010736852501694648, "loss": 0.0388, "step": 39230 }, { "epoch": 2.7228219678931964e-13, "grad_norm": 0.7306115031242371, "learning_rate": 0.00010734840730621576, "loss": 0.0388, "step": 39240 }, { "epoch": 2.723515857283587e-13, "grad_norm": 0.7469100952148438, "learning_rate": 0.00010732828704192456, "loss": 0.0366, "step": 39250 }, { "epoch": 2.724209746673978e-13, "grad_norm": 0.6364192366600037, "learning_rate": 0.00010730816422614055, "loss": 0.0392, "step": 39260 }, { "epoch": 2.7249036360643686e-13, "grad_norm": 0.6924206018447876, "learning_rate": 0.00010728803886093166, "loss": 0.041, "step": 39270 }, { "epoch": 2.7255975254547593e-13, "grad_norm": 0.7715107798576355, "learning_rate": 0.0001072679109483661, "loss": 0.0382, "step": 39280 }, { "epoch": 2.72629141484515e-13, "grad_norm": 0.7552563548088074, "learning_rate": 0.00010724778049051233, "loss": 0.0372, "step": 39290 }, { "epoch": 2.726985304235541e-13, "grad_norm": 0.6953726410865784, "learning_rate": 0.00010722764748943905, "loss": 0.0369, "step": 39300 }, { "epoch": 2.7276791936259315e-13, "grad_norm": 0.6958321332931519, "learning_rate": 0.00010720751194721524, "loss": 0.0359, "step": 39310 }, { "epoch": 2.728373083016322e-13, "grad_norm": 0.7109001278877258, "learning_rate": 0.00010718737386591014, "loss": 0.0346, "step": 39320 }, { "epoch": 2.729066972406713e-13, "grad_norm": 0.8152148127555847, "learning_rate": 0.00010716723324759326, "loss": 0.0365, "step": 39330 }, { "epoch": 2.7297608617971036e-13, "grad_norm": 0.6694443821907043, "learning_rate": 0.00010714709009433439, "loss": 0.0364, "step": 39340 }, { "epoch": 2.7304547511874944e-13, "grad_norm": 0.7080352306365967, "learning_rate": 0.0001071269444082035, "loss": 0.0369, "step": 39350 }, { "epoch": 2.731148640577885e-13, "grad_norm": 0.7440170049667358, "learning_rate": 0.00010710679619127087, "loss": 0.0373, "step": 39360 }, { "epoch": 2.731842529968276e-13, "grad_norm": 0.7223971486091614, "learning_rate": 0.00010708664544560709, "loss": 0.037, "step": 39370 }, { "epoch": 2.7325364193586665e-13, "grad_norm": 0.7323412895202637, "learning_rate": 0.00010706649217328294, "loss": 0.0379, "step": 39380 }, { "epoch": 2.7332303087490573e-13, "grad_norm": 0.7112335562705994, "learning_rate": 0.00010704633637636947, "loss": 0.038, "step": 39390 }, { "epoch": 2.733924198139448e-13, "grad_norm": 0.6443758010864258, "learning_rate": 0.00010702617805693801, "loss": 0.0369, "step": 39400 }, { "epoch": 2.7346180875298387e-13, "grad_norm": 0.7182528972625732, "learning_rate": 0.00010700601721706016, "loss": 0.037, "step": 39410 }, { "epoch": 2.7353119769202294e-13, "grad_norm": 0.6825970411300659, "learning_rate": 0.00010698585385880774, "loss": 0.0389, "step": 39420 }, { "epoch": 2.73600586631062e-13, "grad_norm": 0.6549422740936279, "learning_rate": 0.00010696568798425285, "loss": 0.0405, "step": 39430 }, { "epoch": 2.736699755701011e-13, "grad_norm": 0.6300342679023743, "learning_rate": 0.00010694551959546785, "loss": 0.0372, "step": 39440 }, { "epoch": 2.7373936450914016e-13, "grad_norm": 0.7117260098457336, "learning_rate": 0.00010692534869452536, "loss": 0.0379, "step": 39450 }, { "epoch": 2.7380875344817923e-13, "grad_norm": 0.6153036952018738, "learning_rate": 0.00010690517528349825, "loss": 0.0366, "step": 39460 }, { "epoch": 2.738781423872183e-13, "grad_norm": 0.6226153373718262, "learning_rate": 0.00010688499936445965, "loss": 0.035, "step": 39470 }, { "epoch": 2.739475313262574e-13, "grad_norm": 0.6820581555366516, "learning_rate": 0.00010686482093948295, "loss": 0.0358, "step": 39480 }, { "epoch": 2.7401692026529645e-13, "grad_norm": 0.7449571490287781, "learning_rate": 0.00010684464001064181, "loss": 0.0368, "step": 39490 }, { "epoch": 2.740863092043355e-13, "grad_norm": 0.6957082748413086, "learning_rate": 0.00010682445658001009, "loss": 0.036, "step": 39500 }, { "epoch": 2.741556981433746e-13, "grad_norm": 0.7729166150093079, "learning_rate": 0.00010680427064966201, "loss": 0.0363, "step": 39510 }, { "epoch": 2.7422508708241367e-13, "grad_norm": 0.7272922396659851, "learning_rate": 0.00010678408222167196, "loss": 0.0385, "step": 39520 }, { "epoch": 2.7429447602145274e-13, "grad_norm": 0.7286201119422913, "learning_rate": 0.00010676389129811461, "loss": 0.0364, "step": 39530 }, { "epoch": 2.743638649604918e-13, "grad_norm": 0.6803078055381775, "learning_rate": 0.0001067436978810649, "loss": 0.0375, "step": 39540 }, { "epoch": 2.744332538995309e-13, "grad_norm": 0.7068505883216858, "learning_rate": 0.00010672350197259804, "loss": 0.0387, "step": 39550 }, { "epoch": 2.7450264283856995e-13, "grad_norm": 0.646852433681488, "learning_rate": 0.00010670330357478942, "loss": 0.0355, "step": 39560 }, { "epoch": 2.7457203177760903e-13, "grad_norm": 0.7212904095649719, "learning_rate": 0.00010668310268971481, "loss": 0.0389, "step": 39570 }, { "epoch": 2.746414207166481e-13, "grad_norm": 0.8204472661018372, "learning_rate": 0.0001066628993194501, "loss": 0.0389, "step": 39580 }, { "epoch": 2.7471080965568717e-13, "grad_norm": 0.7798288464546204, "learning_rate": 0.00010664269346607151, "loss": 0.0381, "step": 39590 }, { "epoch": 2.7478019859472624e-13, "grad_norm": 0.7231776118278503, "learning_rate": 0.00010662248513165556, "loss": 0.0367, "step": 39600 }, { "epoch": 2.748495875337653e-13, "grad_norm": 0.7844999432563782, "learning_rate": 0.00010660227431827891, "loss": 0.0364, "step": 39610 }, { "epoch": 2.749189764728044e-13, "grad_norm": 0.7130874395370483, "learning_rate": 0.00010658206102801857, "loss": 0.0359, "step": 39620 }, { "epoch": 2.7498836541184346e-13, "grad_norm": 0.7059792280197144, "learning_rate": 0.00010656184526295175, "loss": 0.0345, "step": 39630 }, { "epoch": 2.7505775435088253e-13, "grad_norm": 0.6279985308647156, "learning_rate": 0.00010654162702515594, "loss": 0.0348, "step": 39640 }, { "epoch": 2.751271432899216e-13, "grad_norm": 0.7713756561279297, "learning_rate": 0.00010652140631670888, "loss": 0.0386, "step": 39650 }, { "epoch": 2.751965322289607e-13, "grad_norm": 0.7092790007591248, "learning_rate": 0.00010650118313968857, "loss": 0.0343, "step": 39660 }, { "epoch": 2.7526592116799975e-13, "grad_norm": 0.6430889964103699, "learning_rate": 0.00010648095749617325, "loss": 0.0364, "step": 39670 }, { "epoch": 2.753353101070388e-13, "grad_norm": 0.7042497396469116, "learning_rate": 0.00010646072938824145, "loss": 0.0371, "step": 39680 }, { "epoch": 2.754046990460779e-13, "grad_norm": 0.697342574596405, "learning_rate": 0.00010644049881797185, "loss": 0.0372, "step": 39690 }, { "epoch": 2.7547408798511697e-13, "grad_norm": 0.6727628111839294, "learning_rate": 0.00010642026578744353, "loss": 0.037, "step": 39700 }, { "epoch": 2.7554347692415604e-13, "grad_norm": 0.7056255340576172, "learning_rate": 0.00010640003029873569, "loss": 0.0387, "step": 39710 }, { "epoch": 2.756128658631951e-13, "grad_norm": 0.7583853006362915, "learning_rate": 0.00010637979235392788, "loss": 0.0362, "step": 39720 }, { "epoch": 2.756822548022342e-13, "grad_norm": 0.7548683285713196, "learning_rate": 0.00010635955195509985, "loss": 0.0372, "step": 39730 }, { "epoch": 2.7575164374127326e-13, "grad_norm": 0.7483044862747192, "learning_rate": 0.00010633930910433162, "loss": 0.039, "step": 39740 }, { "epoch": 2.7582103268031233e-13, "grad_norm": 0.7934048771858215, "learning_rate": 0.00010631906380370345, "loss": 0.0375, "step": 39750 }, { "epoch": 2.758904216193514e-13, "grad_norm": 0.7694196105003357, "learning_rate": 0.00010629881605529587, "loss": 0.0364, "step": 39760 }, { "epoch": 2.7595981055839047e-13, "grad_norm": 0.7407631874084473, "learning_rate": 0.00010627856586118965, "loss": 0.0368, "step": 39770 }, { "epoch": 2.7602919949742954e-13, "grad_norm": 0.7029005289077759, "learning_rate": 0.00010625831322346583, "loss": 0.0355, "step": 39780 }, { "epoch": 2.760985884364686e-13, "grad_norm": 0.7409451007843018, "learning_rate": 0.00010623805814420563, "loss": 0.0332, "step": 39790 }, { "epoch": 2.761679773755077e-13, "grad_norm": 0.7297446727752686, "learning_rate": 0.00010621780062549063, "loss": 0.0359, "step": 39800 }, { "epoch": 2.7623736631454676e-13, "grad_norm": 0.6932554841041565, "learning_rate": 0.00010619754066940259, "loss": 0.0353, "step": 39810 }, { "epoch": 2.7630675525358583e-13, "grad_norm": 0.6750460267066956, "learning_rate": 0.00010617727827802348, "loss": 0.0346, "step": 39820 }, { "epoch": 2.763761441926249e-13, "grad_norm": 0.6792547106742859, "learning_rate": 0.00010615701345343567, "loss": 0.0359, "step": 39830 }, { "epoch": 2.76445533131664e-13, "grad_norm": 0.726091742515564, "learning_rate": 0.00010613674619772163, "loss": 0.0367, "step": 39840 }, { "epoch": 2.7651492207070305e-13, "grad_norm": 0.751781165599823, "learning_rate": 0.00010611647651296414, "loss": 0.0371, "step": 39850 }, { "epoch": 2.765843110097421e-13, "grad_norm": 0.7874179482460022, "learning_rate": 0.00010609620440124624, "loss": 0.0392, "step": 39860 }, { "epoch": 2.766536999487812e-13, "grad_norm": 0.7091031670570374, "learning_rate": 0.0001060759298646512, "loss": 0.0362, "step": 39870 }, { "epoch": 2.7672308888782027e-13, "grad_norm": 0.7626211047172546, "learning_rate": 0.00010605565290526253, "loss": 0.0374, "step": 39880 }, { "epoch": 2.7679247782685934e-13, "grad_norm": 0.7593145370483398, "learning_rate": 0.00010603537352516402, "loss": 0.0372, "step": 39890 }, { "epoch": 2.768618667658984e-13, "grad_norm": 0.718679666519165, "learning_rate": 0.00010601509172643968, "loss": 0.0379, "step": 39900 }, { "epoch": 2.769312557049375e-13, "grad_norm": 0.8393063545227051, "learning_rate": 0.00010599480751117382, "loss": 0.0369, "step": 39910 }, { "epoch": 2.7700064464397656e-13, "grad_norm": 0.64520263671875, "learning_rate": 0.00010597452088145089, "loss": 0.0358, "step": 39920 }, { "epoch": 2.7707003358301563e-13, "grad_norm": 0.6192818880081177, "learning_rate": 0.00010595423183935569, "loss": 0.0365, "step": 39930 }, { "epoch": 2.771394225220547e-13, "grad_norm": 0.7133952975273132, "learning_rate": 0.00010593394038697324, "loss": 0.0352, "step": 39940 }, { "epoch": 2.7720881146109377e-13, "grad_norm": 0.7106195688247681, "learning_rate": 0.00010591364652638881, "loss": 0.0353, "step": 39950 }, { "epoch": 2.7727820040013285e-13, "grad_norm": 0.7348600625991821, "learning_rate": 0.0001058933502596879, "loss": 0.0357, "step": 39960 }, { "epoch": 2.773475893391719e-13, "grad_norm": 0.7013934850692749, "learning_rate": 0.00010587305158895625, "loss": 0.0344, "step": 39970 }, { "epoch": 2.77416978278211e-13, "grad_norm": 0.7679845690727234, "learning_rate": 0.00010585275051627988, "loss": 0.0355, "step": 39980 }, { "epoch": 2.7748636721725006e-13, "grad_norm": 0.6866532564163208, "learning_rate": 0.00010583244704374504, "loss": 0.0359, "step": 39990 }, { "epoch": 2.7755575615628914e-13, "grad_norm": 0.7643980979919434, "learning_rate": 0.00010581214117343825, "loss": 0.0362, "step": 40000 }, { "epoch": 2.776251450953282e-13, "grad_norm": 0.7469555139541626, "learning_rate": 0.00010579183290744625, "loss": 0.0374, "step": 40010 }, { "epoch": 2.776945340343673e-13, "grad_norm": 0.7159682512283325, "learning_rate": 0.00010577152224785598, "loss": 0.0385, "step": 40020 }, { "epoch": 2.7776392297340635e-13, "grad_norm": 0.6790052652359009, "learning_rate": 0.00010575120919675472, "loss": 0.0356, "step": 40030 }, { "epoch": 2.778333119124454e-13, "grad_norm": 0.6693634986877441, "learning_rate": 0.00010573089375622993, "loss": 0.0381, "step": 40040 }, { "epoch": 2.779027008514845e-13, "grad_norm": 0.6875048279762268, "learning_rate": 0.00010571057592836937, "loss": 0.0374, "step": 40050 }, { "epoch": 2.7797208979052357e-13, "grad_norm": 0.6271710395812988, "learning_rate": 0.00010569025571526098, "loss": 0.0356, "step": 40060 }, { "epoch": 2.7804147872956264e-13, "grad_norm": 0.6844568848609924, "learning_rate": 0.000105669933118993, "loss": 0.0363, "step": 40070 }, { "epoch": 2.781108676686017e-13, "grad_norm": 0.7256521582603455, "learning_rate": 0.00010564960814165386, "loss": 0.0367, "step": 40080 }, { "epoch": 2.781802566076408e-13, "grad_norm": 0.6763705015182495, "learning_rate": 0.00010562928078533231, "loss": 0.0348, "step": 40090 }, { "epoch": 2.7824964554667986e-13, "grad_norm": 0.665901243686676, "learning_rate": 0.00010560895105211729, "loss": 0.0344, "step": 40100 }, { "epoch": 2.7831903448571893e-13, "grad_norm": 0.7007461786270142, "learning_rate": 0.00010558861894409799, "loss": 0.0344, "step": 40110 }, { "epoch": 2.78388423424758e-13, "grad_norm": 0.6174138188362122, "learning_rate": 0.00010556828446336384, "loss": 0.0366, "step": 40120 }, { "epoch": 2.784578123637971e-13, "grad_norm": 0.6579844355583191, "learning_rate": 0.00010554794761200456, "loss": 0.0341, "step": 40130 }, { "epoch": 2.7852720130283615e-13, "grad_norm": 0.6248412728309631, "learning_rate": 0.00010552760839211001, "loss": 0.0341, "step": 40140 }, { "epoch": 2.785965902418752e-13, "grad_norm": 0.6320821642875671, "learning_rate": 0.00010550726680577042, "loss": 0.0357, "step": 40150 }, { "epoch": 2.786659791809143e-13, "grad_norm": 0.6984745264053345, "learning_rate": 0.0001054869228550762, "loss": 0.0368, "step": 40160 }, { "epoch": 2.7873536811995336e-13, "grad_norm": 0.7675280570983887, "learning_rate": 0.00010546657654211797, "loss": 0.0371, "step": 40170 }, { "epoch": 2.7880475705899244e-13, "grad_norm": 0.634484589099884, "learning_rate": 0.0001054462278689867, "loss": 0.0369, "step": 40180 }, { "epoch": 2.788741459980315e-13, "grad_norm": 0.6267728209495544, "learning_rate": 0.00010542587683777344, "loss": 0.0358, "step": 40190 }, { "epoch": 2.789435349370706e-13, "grad_norm": 0.6387699246406555, "learning_rate": 0.00010540552345056966, "loss": 0.0372, "step": 40200 }, { "epoch": 2.7901292387610965e-13, "grad_norm": 0.6984362006187439, "learning_rate": 0.00010538516770946694, "loss": 0.0378, "step": 40210 }, { "epoch": 2.7908231281514873e-13, "grad_norm": 0.6808851957321167, "learning_rate": 0.00010536480961655716, "loss": 0.0355, "step": 40220 }, { "epoch": 2.791517017541878e-13, "grad_norm": 0.6536777019500732, "learning_rate": 0.00010534444917393244, "loss": 0.0359, "step": 40230 }, { "epoch": 2.7922109069322687e-13, "grad_norm": 0.6571732759475708, "learning_rate": 0.00010532408638368511, "loss": 0.0367, "step": 40240 }, { "epoch": 2.7929047963226594e-13, "grad_norm": 0.6074491739273071, "learning_rate": 0.00010530372124790778, "loss": 0.0349, "step": 40250 }, { "epoch": 2.79359868571305e-13, "grad_norm": 0.683221161365509, "learning_rate": 0.00010528335376869328, "loss": 0.0347, "step": 40260 }, { "epoch": 2.794292575103441e-13, "grad_norm": 0.622276246547699, "learning_rate": 0.00010526298394813469, "loss": 0.0348, "step": 40270 }, { "epoch": 2.7949864644938316e-13, "grad_norm": 0.613656759262085, "learning_rate": 0.00010524261178832531, "loss": 0.0352, "step": 40280 }, { "epoch": 2.7956803538842223e-13, "grad_norm": 0.6469065546989441, "learning_rate": 0.00010522223729135873, "loss": 0.0345, "step": 40290 }, { "epoch": 2.796374243274613e-13, "grad_norm": 0.7689803242683411, "learning_rate": 0.00010520186045932872, "loss": 0.0359, "step": 40300 }, { "epoch": 2.797068132665004e-13, "grad_norm": 0.6478021740913391, "learning_rate": 0.00010518148129432931, "loss": 0.0343, "step": 40310 }, { "epoch": 2.7977620220553945e-13, "grad_norm": 0.7315558195114136, "learning_rate": 0.00010516109979845482, "loss": 0.0364, "step": 40320 }, { "epoch": 2.798455911445785e-13, "grad_norm": 0.744411289691925, "learning_rate": 0.0001051407159737997, "loss": 0.038, "step": 40330 }, { "epoch": 2.799149800836176e-13, "grad_norm": 0.7246782779693604, "learning_rate": 0.00010512032982245877, "loss": 0.036, "step": 40340 }, { "epoch": 2.7998436902265667e-13, "grad_norm": 0.7636134028434753, "learning_rate": 0.00010509994134652698, "loss": 0.038, "step": 40350 }, { "epoch": 2.8005375796169574e-13, "grad_norm": 0.7927470803260803, "learning_rate": 0.00010507955054809956, "loss": 0.037, "step": 40360 }, { "epoch": 2.801231469007348e-13, "grad_norm": 0.7167902588844299, "learning_rate": 0.00010505915742927202, "loss": 0.0366, "step": 40370 }, { "epoch": 2.801925358397739e-13, "grad_norm": 0.7068259119987488, "learning_rate": 0.00010503876199214003, "loss": 0.035, "step": 40380 }, { "epoch": 2.8026192477881295e-13, "grad_norm": 0.7164059281349182, "learning_rate": 0.00010501836423879958, "loss": 0.0341, "step": 40390 }, { "epoch": 2.8033131371785203e-13, "grad_norm": 0.7511824369430542, "learning_rate": 0.0001049979641713468, "loss": 0.0364, "step": 40400 }, { "epoch": 2.804007026568911e-13, "grad_norm": 0.6810638308525085, "learning_rate": 0.00010497756179187819, "loss": 0.0349, "step": 40410 }, { "epoch": 2.8047009159593017e-13, "grad_norm": 0.5698809027671814, "learning_rate": 0.00010495715710249035, "loss": 0.0348, "step": 40420 }, { "epoch": 2.8053948053496924e-13, "grad_norm": 0.633352518081665, "learning_rate": 0.0001049367501052802, "loss": 0.0353, "step": 40430 }, { "epoch": 2.806088694740083e-13, "grad_norm": 0.6652390360832214, "learning_rate": 0.00010491634080234488, "loss": 0.0341, "step": 40440 }, { "epoch": 2.806782584130474e-13, "grad_norm": 0.7045243978500366, "learning_rate": 0.00010489592919578176, "loss": 0.0358, "step": 40450 }, { "epoch": 2.8074764735208646e-13, "grad_norm": 0.6472264528274536, "learning_rate": 0.00010487551528768847, "loss": 0.0344, "step": 40460 }, { "epoch": 2.8081703629112553e-13, "grad_norm": 0.7128479480743408, "learning_rate": 0.00010485509908016279, "loss": 0.0351, "step": 40470 }, { "epoch": 2.808864252301646e-13, "grad_norm": 0.7406001687049866, "learning_rate": 0.00010483468057530288, "loss": 0.0351, "step": 40480 }, { "epoch": 2.809558141692037e-13, "grad_norm": 0.7963809967041016, "learning_rate": 0.00010481425977520703, "loss": 0.0376, "step": 40490 }, { "epoch": 2.8102520310824275e-13, "grad_norm": 0.7524694204330444, "learning_rate": 0.00010479383668197378, "loss": 0.0359, "step": 40500 }, { "epoch": 2.810945920472818e-13, "grad_norm": 0.6600159406661987, "learning_rate": 0.00010477341129770195, "loss": 0.0373, "step": 40510 }, { "epoch": 2.811639809863209e-13, "grad_norm": 0.6807093620300293, "learning_rate": 0.00010475298362449052, "loss": 0.0387, "step": 40520 }, { "epoch": 2.8123336992535997e-13, "grad_norm": 0.7361227869987488, "learning_rate": 0.00010473255366443879, "loss": 0.0357, "step": 40530 }, { "epoch": 2.8130275886439904e-13, "grad_norm": 0.7672550082206726, "learning_rate": 0.00010471212141964628, "loss": 0.0344, "step": 40540 }, { "epoch": 2.813721478034381e-13, "grad_norm": 0.715376079082489, "learning_rate": 0.00010469168689221267, "loss": 0.035, "step": 40550 }, { "epoch": 2.814415367424772e-13, "grad_norm": 0.7294541597366333, "learning_rate": 0.00010467125008423793, "loss": 0.0344, "step": 40560 }, { "epoch": 2.8151092568151626e-13, "grad_norm": 0.7417559623718262, "learning_rate": 0.00010465081099782229, "loss": 0.0346, "step": 40570 }, { "epoch": 2.8158031462055533e-13, "grad_norm": 0.7600880861282349, "learning_rate": 0.00010463036963506616, "loss": 0.0355, "step": 40580 }, { "epoch": 2.816497035595944e-13, "grad_norm": 0.6419711112976074, "learning_rate": 0.00010460992599807023, "loss": 0.0355, "step": 40590 }, { "epoch": 2.8171909249863347e-13, "grad_norm": 0.7419015169143677, "learning_rate": 0.00010458948008893537, "loss": 0.0346, "step": 40600 }, { "epoch": 2.8178848143767254e-13, "grad_norm": 0.7169345021247864, "learning_rate": 0.00010456903190976276, "loss": 0.0351, "step": 40610 }, { "epoch": 2.818578703767116e-13, "grad_norm": 0.6747776865959167, "learning_rate": 0.00010454858146265373, "loss": 0.0352, "step": 40620 }, { "epoch": 2.819272593157507e-13, "grad_norm": 0.6684876680374146, "learning_rate": 0.00010452812874970987, "loss": 0.0352, "step": 40630 }, { "epoch": 2.8199664825478976e-13, "grad_norm": 0.7223830819129944, "learning_rate": 0.00010450767377303307, "loss": 0.0361, "step": 40640 }, { "epoch": 2.8206603719382883e-13, "grad_norm": 0.7251724600791931, "learning_rate": 0.00010448721653472535, "loss": 0.0358, "step": 40650 }, { "epoch": 2.821354261328679e-13, "grad_norm": 0.7586217522621155, "learning_rate": 0.00010446675703688903, "loss": 0.0371, "step": 40660 }, { "epoch": 2.82204815071907e-13, "grad_norm": 0.692321240901947, "learning_rate": 0.00010444629528162663, "loss": 0.0372, "step": 40670 }, { "epoch": 2.8227420401094605e-13, "grad_norm": 0.6365525722503662, "learning_rate": 0.00010442583127104095, "loss": 0.0383, "step": 40680 }, { "epoch": 2.823435929499851e-13, "grad_norm": 0.5939146876335144, "learning_rate": 0.0001044053650072349, "loss": 0.0342, "step": 40690 }, { "epoch": 2.824129818890242e-13, "grad_norm": 0.7293074131011963, "learning_rate": 0.00010438489649231182, "loss": 0.0348, "step": 40700 }, { "epoch": 2.8248237082806327e-13, "grad_norm": 0.5917070508003235, "learning_rate": 0.00010436442572837508, "loss": 0.0345, "step": 40710 }, { "epoch": 2.8255175976710234e-13, "grad_norm": 0.5952087044715881, "learning_rate": 0.00010434395271752839, "loss": 0.0335, "step": 40720 }, { "epoch": 2.826211487061414e-13, "grad_norm": 0.7101953625679016, "learning_rate": 0.0001043234774618757, "loss": 0.0341, "step": 40730 }, { "epoch": 2.826905376451805e-13, "grad_norm": 0.7623522281646729, "learning_rate": 0.00010430299996352114, "loss": 0.0346, "step": 40740 }, { "epoch": 2.8275992658421956e-13, "grad_norm": 0.7189036011695862, "learning_rate": 0.00010428252022456911, "loss": 0.0342, "step": 40750 }, { "epoch": 2.8282931552325863e-13, "grad_norm": 0.8067599534988403, "learning_rate": 0.0001042620382471242, "loss": 0.035, "step": 40760 }, { "epoch": 2.828987044622977e-13, "grad_norm": 0.7333223223686218, "learning_rate": 0.00010424155403329123, "loss": 0.0361, "step": 40770 }, { "epoch": 2.8296809340133677e-13, "grad_norm": 0.7406851053237915, "learning_rate": 0.00010422106758517533, "loss": 0.034, "step": 40780 }, { "epoch": 2.8303748234037585e-13, "grad_norm": 0.6545450091362, "learning_rate": 0.00010420057890488175, "loss": 0.0348, "step": 40790 }, { "epoch": 2.831068712794149e-13, "grad_norm": 0.6630377769470215, "learning_rate": 0.00010418008799451606, "loss": 0.0364, "step": 40800 }, { "epoch": 2.83176260218454e-13, "grad_norm": 0.6918194890022278, "learning_rate": 0.000104159594856184, "loss": 0.0343, "step": 40810 }, { "epoch": 2.8324564915749306e-13, "grad_norm": 0.7116217017173767, "learning_rate": 0.00010413909949199156, "loss": 0.0372, "step": 40820 }, { "epoch": 2.8331503809653213e-13, "grad_norm": 0.7703463435173035, "learning_rate": 0.00010411860190404494, "loss": 0.0378, "step": 40830 }, { "epoch": 2.833844270355712e-13, "grad_norm": 0.7101000547409058, "learning_rate": 0.00010409810209445062, "loss": 0.0364, "step": 40840 }, { "epoch": 2.834538159746103e-13, "grad_norm": 0.6596708297729492, "learning_rate": 0.00010407760006531525, "loss": 0.0337, "step": 40850 }, { "epoch": 2.8352320491364935e-13, "grad_norm": 0.7424857020378113, "learning_rate": 0.00010405709581874572, "loss": 0.0339, "step": 40860 }, { "epoch": 2.835925938526884e-13, "grad_norm": 0.6915359497070312, "learning_rate": 0.00010403658935684922, "loss": 0.0341, "step": 40870 }, { "epoch": 2.836619827917275e-13, "grad_norm": 0.6652345657348633, "learning_rate": 0.00010401608068173307, "loss": 0.033, "step": 40880 }, { "epoch": 2.8373137173076657e-13, "grad_norm": 0.6455739736557007, "learning_rate": 0.00010399556979550483, "loss": 0.0337, "step": 40890 }, { "epoch": 2.8380076066980564e-13, "grad_norm": 0.8008694648742676, "learning_rate": 0.00010397505670027232, "loss": 0.0361, "step": 40900 }, { "epoch": 2.838701496088447e-13, "grad_norm": 0.7614725232124329, "learning_rate": 0.00010395454139814364, "loss": 0.0332, "step": 40910 }, { "epoch": 2.839395385478838e-13, "grad_norm": 0.6105318069458008, "learning_rate": 0.000103934023891227, "loss": 0.0344, "step": 40920 }, { "epoch": 2.8400892748692286e-13, "grad_norm": 0.6731852889060974, "learning_rate": 0.00010391350418163089, "loss": 0.0351, "step": 40930 }, { "epoch": 2.8407831642596193e-13, "grad_norm": 0.6799097061157227, "learning_rate": 0.00010389298227146405, "loss": 0.035, "step": 40940 }, { "epoch": 2.84147705365001e-13, "grad_norm": 0.6393249034881592, "learning_rate": 0.00010387245816283544, "loss": 0.0347, "step": 40950 }, { "epoch": 2.8421709430404007e-13, "grad_norm": 0.6808954477310181, "learning_rate": 0.00010385193185785418, "loss": 0.0357, "step": 40960 }, { "epoch": 2.8428648324307915e-13, "grad_norm": 0.7204254865646362, "learning_rate": 0.00010383140335862973, "loss": 0.0339, "step": 40970 }, { "epoch": 2.843558721821182e-13, "grad_norm": 0.7801540493965149, "learning_rate": 0.00010381087266727168, "loss": 0.0361, "step": 40980 }, { "epoch": 2.844252611211573e-13, "grad_norm": 0.7241537570953369, "learning_rate": 0.00010379033978588988, "loss": 0.0379, "step": 40990 }, { "epoch": 2.8449465006019636e-13, "grad_norm": 0.7522146701812744, "learning_rate": 0.0001037698047165944, "loss": 0.0363, "step": 41000 }, { "epoch": 2.8456403899923544e-13, "grad_norm": 0.7291750907897949, "learning_rate": 0.00010374926746149554, "loss": 0.0344, "step": 41010 }, { "epoch": 2.846334279382745e-13, "grad_norm": 0.724157452583313, "learning_rate": 0.00010372872802270382, "loss": 0.0346, "step": 41020 }, { "epoch": 2.847028168773136e-13, "grad_norm": 0.7319716215133667, "learning_rate": 0.00010370818640233, "loss": 0.0335, "step": 41030 }, { "epoch": 2.8477220581635265e-13, "grad_norm": 0.7180967330932617, "learning_rate": 0.00010368764260248503, "loss": 0.0323, "step": 41040 }, { "epoch": 2.848415947553917e-13, "grad_norm": 0.6749131679534912, "learning_rate": 0.00010366709662528012, "loss": 0.0341, "step": 41050 }, { "epoch": 2.849109836944308e-13, "grad_norm": 0.7131304740905762, "learning_rate": 0.0001036465484728267, "loss": 0.0342, "step": 41060 }, { "epoch": 2.8498037263346987e-13, "grad_norm": 0.7001612186431885, "learning_rate": 0.00010362599814723638, "loss": 0.0331, "step": 41070 }, { "epoch": 2.8504976157250894e-13, "grad_norm": 0.680414617061615, "learning_rate": 0.00010360544565062106, "loss": 0.035, "step": 41080 }, { "epoch": 2.85119150511548e-13, "grad_norm": 0.7093191742897034, "learning_rate": 0.00010358489098509282, "loss": 0.0353, "step": 41090 }, { "epoch": 2.851885394505871e-13, "grad_norm": 0.6600826978683472, "learning_rate": 0.00010356433415276396, "loss": 0.0356, "step": 41100 }, { "epoch": 2.8525792838962616e-13, "grad_norm": 0.6768158078193665, "learning_rate": 0.00010354377515574698, "loss": 0.0364, "step": 41110 }, { "epoch": 2.8532731732866523e-13, "grad_norm": 0.6677889823913574, "learning_rate": 0.00010352321399615472, "loss": 0.0331, "step": 41120 }, { "epoch": 2.853967062677043e-13, "grad_norm": 0.7104982733726501, "learning_rate": 0.00010350265067610012, "loss": 0.0347, "step": 41130 }, { "epoch": 2.854660952067434e-13, "grad_norm": 0.773462176322937, "learning_rate": 0.00010348208519769635, "loss": 0.0357, "step": 41140 }, { "epoch": 2.8553548414578245e-13, "grad_norm": 0.79094398021698, "learning_rate": 0.00010346151756305686, "loss": 0.0364, "step": 41150 }, { "epoch": 2.856048730848215e-13, "grad_norm": 0.879010796546936, "learning_rate": 0.0001034409477742953, "loss": 0.0351, "step": 41160 }, { "epoch": 2.856742620238606e-13, "grad_norm": 0.6439926028251648, "learning_rate": 0.00010342037583352552, "loss": 0.0333, "step": 41170 }, { "epoch": 2.8574365096289966e-13, "grad_norm": 0.5977691411972046, "learning_rate": 0.00010339980174286164, "loss": 0.034, "step": 41180 }, { "epoch": 2.8581303990193874e-13, "grad_norm": 0.7250075936317444, "learning_rate": 0.00010337922550441794, "loss": 0.0335, "step": 41190 }, { "epoch": 2.858824288409778e-13, "grad_norm": 0.7686924338340759, "learning_rate": 0.00010335864712030894, "loss": 0.0359, "step": 41200 }, { "epoch": 2.859518177800169e-13, "grad_norm": 0.7519763708114624, "learning_rate": 0.00010333806659264941, "loss": 0.0355, "step": 41210 }, { "epoch": 2.8602120671905595e-13, "grad_norm": 0.7127652168273926, "learning_rate": 0.0001033174839235543, "loss": 0.0334, "step": 41220 }, { "epoch": 2.8609059565809503e-13, "grad_norm": 0.7298749685287476, "learning_rate": 0.00010329689911513884, "loss": 0.0342, "step": 41230 }, { "epoch": 2.861599845971341e-13, "grad_norm": 0.6721780896186829, "learning_rate": 0.00010327631216951836, "loss": 0.0347, "step": 41240 }, { "epoch": 2.8622937353617317e-13, "grad_norm": 0.7490611672401428, "learning_rate": 0.00010325572308880858, "loss": 0.0349, "step": 41250 }, { "epoch": 2.8629876247521224e-13, "grad_norm": 0.7196139097213745, "learning_rate": 0.00010323513187512532, "loss": 0.0359, "step": 41260 }, { "epoch": 2.863681514142513e-13, "grad_norm": 0.6576598286628723, "learning_rate": 0.0001032145385305846, "loss": 0.0361, "step": 41270 }, { "epoch": 2.864375403532904e-13, "grad_norm": 0.6630513072013855, "learning_rate": 0.00010319394305730277, "loss": 0.0329, "step": 41280 }, { "epoch": 2.8650692929232946e-13, "grad_norm": 0.6373247504234314, "learning_rate": 0.00010317334545739627, "loss": 0.0355, "step": 41290 }, { "epoch": 2.8657631823136853e-13, "grad_norm": 0.6743766665458679, "learning_rate": 0.0001031527457329819, "loss": 0.0358, "step": 41300 }, { "epoch": 2.866457071704076e-13, "grad_norm": 0.6365957260131836, "learning_rate": 0.00010313214388617656, "loss": 0.0342, "step": 41310 }, { "epoch": 2.867150961094467e-13, "grad_norm": 0.6940137147903442, "learning_rate": 0.0001031115399190974, "loss": 0.0336, "step": 41320 }, { "epoch": 2.8678448504848575e-13, "grad_norm": 0.8147335052490234, "learning_rate": 0.0001030909338338618, "loss": 0.0346, "step": 41330 }, { "epoch": 2.868538739875248e-13, "grad_norm": 0.6763673424720764, "learning_rate": 0.00010307032563258737, "loss": 0.0329, "step": 41340 }, { "epoch": 2.869232629265639e-13, "grad_norm": 0.6681472659111023, "learning_rate": 0.00010304971531739194, "loss": 0.033, "step": 41350 }, { "epoch": 2.8699265186560297e-13, "grad_norm": 0.6778015494346619, "learning_rate": 0.0001030291028903935, "loss": 0.0341, "step": 41360 }, { "epoch": 2.8706204080464204e-13, "grad_norm": 0.6416174173355103, "learning_rate": 0.00010300848835371032, "loss": 0.0339, "step": 41370 }, { "epoch": 2.871314297436811e-13, "grad_norm": 0.6404531002044678, "learning_rate": 0.00010298787170946089, "loss": 0.032, "step": 41380 }, { "epoch": 2.872008186827202e-13, "grad_norm": 0.6286908388137817, "learning_rate": 0.00010296725295976385, "loss": 0.033, "step": 41390 }, { "epoch": 2.8727020762175925e-13, "grad_norm": 0.628956139087677, "learning_rate": 0.0001029466321067381, "loss": 0.034, "step": 41400 }, { "epoch": 2.8733959656079833e-13, "grad_norm": 0.7001970410346985, "learning_rate": 0.00010292600915250281, "loss": 0.035, "step": 41410 }, { "epoch": 2.874089854998374e-13, "grad_norm": 0.7957877516746521, "learning_rate": 0.00010290538409917723, "loss": 0.0359, "step": 41420 }, { "epoch": 2.8747837443887647e-13, "grad_norm": 0.6192899346351624, "learning_rate": 0.00010288475694888094, "loss": 0.0334, "step": 41430 }, { "epoch": 2.8754776337791554e-13, "grad_norm": 0.6373462677001953, "learning_rate": 0.0001028641277037337, "loss": 0.0333, "step": 41440 }, { "epoch": 2.876171523169546e-13, "grad_norm": 0.6475844979286194, "learning_rate": 0.00010284349636585547, "loss": 0.0345, "step": 41450 }, { "epoch": 2.876865412559937e-13, "grad_norm": 0.6622808575630188, "learning_rate": 0.00010282286293736649, "loss": 0.0356, "step": 41460 }, { "epoch": 2.8775593019503276e-13, "grad_norm": 0.6322243809700012, "learning_rate": 0.00010280222742038712, "loss": 0.0342, "step": 41470 }, { "epoch": 2.8782531913407183e-13, "grad_norm": 0.6558703184127808, "learning_rate": 0.000102781589817038, "loss": 0.0336, "step": 41480 }, { "epoch": 2.878947080731109e-13, "grad_norm": 0.6525861620903015, "learning_rate": 0.00010276095012943994, "loss": 0.0347, "step": 41490 }, { "epoch": 2.8796409701215e-13, "grad_norm": 0.615079939365387, "learning_rate": 0.000102740308359714, "loss": 0.0324, "step": 41500 }, { "epoch": 2.8803348595118905e-13, "grad_norm": 0.7026222944259644, "learning_rate": 0.00010271966450998151, "loss": 0.033, "step": 41510 }, { "epoch": 2.881028748902281e-13, "grad_norm": 0.5866784453392029, "learning_rate": 0.00010269901858236385, "loss": 0.0335, "step": 41520 }, { "epoch": 2.881722638292672e-13, "grad_norm": 0.5975731611251831, "learning_rate": 0.00010267837057898274, "loss": 0.0332, "step": 41530 }, { "epoch": 2.8824165276830627e-13, "grad_norm": 0.6165285110473633, "learning_rate": 0.0001026577205019601, "loss": 0.0318, "step": 41540 }, { "epoch": 2.8831104170734534e-13, "grad_norm": 0.7799416184425354, "learning_rate": 0.00010263706835341802, "loss": 0.034, "step": 41550 }, { "epoch": 2.883804306463844e-13, "grad_norm": 0.6497504711151123, "learning_rate": 0.00010261641413547886, "loss": 0.0325, "step": 41560 }, { "epoch": 2.884498195854235e-13, "grad_norm": 0.6919741630554199, "learning_rate": 0.00010259575785026517, "loss": 0.0346, "step": 41570 }, { "epoch": 2.8851920852446256e-13, "grad_norm": 0.7164174914360046, "learning_rate": 0.00010257509949989965, "loss": 0.0358, "step": 41580 }, { "epoch": 2.8858859746350163e-13, "grad_norm": 0.707055389881134, "learning_rate": 0.00010255443908650534, "loss": 0.0329, "step": 41590 }, { "epoch": 2.886579864025407e-13, "grad_norm": 0.7372544407844543, "learning_rate": 0.00010253377661220536, "loss": 0.0348, "step": 41600 }, { "epoch": 2.8872737534157977e-13, "grad_norm": 0.776802122592926, "learning_rate": 0.00010251311207912311, "loss": 0.0349, "step": 41610 }, { "epoch": 2.8879676428061885e-13, "grad_norm": 0.6940571665763855, "learning_rate": 0.00010249244548938223, "loss": 0.0334, "step": 41620 }, { "epoch": 2.888661532196579e-13, "grad_norm": 0.642642080783844, "learning_rate": 0.0001024717768451065, "loss": 0.0336, "step": 41630 }, { "epoch": 2.88935542158697e-13, "grad_norm": 0.6660865545272827, "learning_rate": 0.00010245110614841993, "loss": 0.0328, "step": 41640 }, { "epoch": 2.8900493109773606e-13, "grad_norm": 0.6745574474334717, "learning_rate": 0.0001024304334014468, "loss": 0.0336, "step": 41650 }, { "epoch": 2.8907432003677513e-13, "grad_norm": 0.6263169050216675, "learning_rate": 0.00010240975860631153, "loss": 0.0324, "step": 41660 }, { "epoch": 2.891437089758142e-13, "grad_norm": 0.5730962157249451, "learning_rate": 0.00010238908176513875, "loss": 0.0334, "step": 41670 }, { "epoch": 2.892130979148533e-13, "grad_norm": 0.631469190120697, "learning_rate": 0.0001023684028800534, "loss": 0.0335, "step": 41680 }, { "epoch": 2.8928248685389235e-13, "grad_norm": 0.6359084844589233, "learning_rate": 0.0001023477219531805, "loss": 0.0319, "step": 41690 }, { "epoch": 2.893518757929314e-13, "grad_norm": 0.6417659521102905, "learning_rate": 0.00010232703898664537, "loss": 0.0325, "step": 41700 }, { "epoch": 2.894212647319705e-13, "grad_norm": 0.6239885091781616, "learning_rate": 0.00010230635398257346, "loss": 0.033, "step": 41710 }, { "epoch": 2.8949065367100957e-13, "grad_norm": 0.7160294651985168, "learning_rate": 0.00010228566694309052, "loss": 0.0332, "step": 41720 }, { "epoch": 2.8956004261004864e-13, "grad_norm": 0.6914883852005005, "learning_rate": 0.00010226497787032246, "loss": 0.0338, "step": 41730 }, { "epoch": 2.896294315490877e-13, "grad_norm": 0.692836582660675, "learning_rate": 0.00010224428676639537, "loss": 0.0355, "step": 41740 }, { "epoch": 2.896988204881268e-13, "grad_norm": 0.7077901363372803, "learning_rate": 0.00010222359363343563, "loss": 0.032, "step": 41750 }, { "epoch": 2.8976820942716586e-13, "grad_norm": 0.6012435555458069, "learning_rate": 0.00010220289847356975, "loss": 0.0337, "step": 41760 }, { "epoch": 2.8983759836620493e-13, "grad_norm": 0.6507777571678162, "learning_rate": 0.00010218220128892448, "loss": 0.0358, "step": 41770 }, { "epoch": 2.89906987305244e-13, "grad_norm": 0.7304379940032959, "learning_rate": 0.00010216150208162679, "loss": 0.0327, "step": 41780 }, { "epoch": 2.8997637624428307e-13, "grad_norm": 0.7257174253463745, "learning_rate": 0.00010214080085380385, "loss": 0.0314, "step": 41790 }, { "epoch": 2.9004576518332215e-13, "grad_norm": 0.656329333782196, "learning_rate": 0.00010212009760758302, "loss": 0.0334, "step": 41800 }, { "epoch": 2.901151541223612e-13, "grad_norm": 0.6361106634140015, "learning_rate": 0.00010209939234509188, "loss": 0.0318, "step": 41810 }, { "epoch": 2.901845430614003e-13, "grad_norm": 0.6764311194419861, "learning_rate": 0.00010207868506845823, "loss": 0.0319, "step": 41820 }, { "epoch": 2.9025393200043936e-13, "grad_norm": 0.7691108584403992, "learning_rate": 0.00010205797577981004, "loss": 0.0333, "step": 41830 }, { "epoch": 2.9032332093947844e-13, "grad_norm": 0.645847737789154, "learning_rate": 0.00010203726448127556, "loss": 0.0326, "step": 41840 }, { "epoch": 2.903927098785175e-13, "grad_norm": 0.6843918561935425, "learning_rate": 0.00010201655117498316, "loss": 0.0317, "step": 41850 }, { "epoch": 2.904620988175566e-13, "grad_norm": 0.688572347164154, "learning_rate": 0.00010199583586306145, "loss": 0.0319, "step": 41860 }, { "epoch": 2.9053148775659565e-13, "grad_norm": 0.6445980668067932, "learning_rate": 0.00010197511854763927, "loss": 0.0329, "step": 41870 }, { "epoch": 2.906008766956347e-13, "grad_norm": 0.6763575077056885, "learning_rate": 0.00010195439923084563, "loss": 0.034, "step": 41880 }, { "epoch": 2.906702656346738e-13, "grad_norm": 0.6982194781303406, "learning_rate": 0.00010193367791480978, "loss": 0.0341, "step": 41890 }, { "epoch": 2.9073965457371287e-13, "grad_norm": 0.6515775918960571, "learning_rate": 0.00010191295460166116, "loss": 0.0335, "step": 41900 }, { "epoch": 2.9080904351275194e-13, "grad_norm": 0.6720160841941833, "learning_rate": 0.0001018922292935294, "loss": 0.0331, "step": 41910 }, { "epoch": 2.90878432451791e-13, "grad_norm": 0.6629319190979004, "learning_rate": 0.00010187150199254434, "loss": 0.034, "step": 41920 }, { "epoch": 2.909478213908301e-13, "grad_norm": 0.5893653631210327, "learning_rate": 0.00010185077270083605, "loss": 0.0343, "step": 41930 }, { "epoch": 2.9101721032986916e-13, "grad_norm": 0.5831941962242126, "learning_rate": 0.00010183004142053475, "loss": 0.0316, "step": 41940 }, { "epoch": 2.9108659926890823e-13, "grad_norm": 0.6937285661697388, "learning_rate": 0.000101809308153771, "loss": 0.0324, "step": 41950 }, { "epoch": 2.911559882079473e-13, "grad_norm": 0.6109587550163269, "learning_rate": 0.00010178857290267534, "loss": 0.0325, "step": 41960 }, { "epoch": 2.912253771469864e-13, "grad_norm": 0.5872533321380615, "learning_rate": 0.00010176783566937871, "loss": 0.0309, "step": 41970 }, { "epoch": 2.9129476608602545e-13, "grad_norm": 0.6698512434959412, "learning_rate": 0.00010174709645601218, "loss": 0.032, "step": 41980 }, { "epoch": 2.913641550250645e-13, "grad_norm": 0.6950770616531372, "learning_rate": 0.00010172635526470701, "loss": 0.0326, "step": 41990 }, { "epoch": 2.914335439641036e-13, "grad_norm": 0.663817822933197, "learning_rate": 0.00010170561209759467, "loss": 0.0321, "step": 42000 }, { "epoch": 2.9150293290314266e-13, "grad_norm": 0.741736888885498, "learning_rate": 0.00010168486695680689, "loss": 0.0314, "step": 42010 }, { "epoch": 2.9157232184218174e-13, "grad_norm": 0.7179228663444519, "learning_rate": 0.00010166411984447552, "loss": 0.0328, "step": 42020 }, { "epoch": 2.916417107812208e-13, "grad_norm": 0.7061957716941833, "learning_rate": 0.00010164337076273265, "loss": 0.032, "step": 42030 }, { "epoch": 2.917110997202599e-13, "grad_norm": 0.6627841591835022, "learning_rate": 0.00010162261971371056, "loss": 0.0338, "step": 42040 }, { "epoch": 2.9178048865929895e-13, "grad_norm": 0.6662027835845947, "learning_rate": 0.00010160186669954178, "loss": 0.0348, "step": 42050 }, { "epoch": 2.9184987759833803e-13, "grad_norm": 0.6780067086219788, "learning_rate": 0.000101581111722359, "loss": 0.0324, "step": 42060 }, { "epoch": 2.919192665373771e-13, "grad_norm": 0.6636269688606262, "learning_rate": 0.00010156035478429509, "loss": 0.0341, "step": 42070 }, { "epoch": 2.9198865547641617e-13, "grad_norm": 0.6985071301460266, "learning_rate": 0.00010153959588748315, "loss": 0.0341, "step": 42080 }, { "epoch": 2.9205804441545524e-13, "grad_norm": 0.664534330368042, "learning_rate": 0.00010151883503405651, "loss": 0.0324, "step": 42090 }, { "epoch": 2.921274333544943e-13, "grad_norm": 0.6088982224464417, "learning_rate": 0.00010149807222614861, "loss": 0.0305, "step": 42100 }, { "epoch": 2.921968222935334e-13, "grad_norm": 0.7572206854820251, "learning_rate": 0.00010147730746589322, "loss": 0.0315, "step": 42110 }, { "epoch": 2.9226621123257246e-13, "grad_norm": 0.7130482792854309, "learning_rate": 0.0001014565407554242, "loss": 0.0312, "step": 42120 }, { "epoch": 2.9233560017161153e-13, "grad_norm": 0.6676921248435974, "learning_rate": 0.0001014357720968757, "loss": 0.0308, "step": 42130 }, { "epoch": 2.924049891106506e-13, "grad_norm": 0.6645044684410095, "learning_rate": 0.00010141500149238195, "loss": 0.0315, "step": 42140 }, { "epoch": 2.924743780496897e-13, "grad_norm": 0.7510046362876892, "learning_rate": 0.00010139422894407752, "loss": 0.0338, "step": 42150 }, { "epoch": 2.9254376698872875e-13, "grad_norm": 0.6913697719573975, "learning_rate": 0.00010137345445409707, "loss": 0.0312, "step": 42160 }, { "epoch": 2.926131559277678e-13, "grad_norm": 0.58387291431427, "learning_rate": 0.00010135267802457552, "loss": 0.0319, "step": 42170 }, { "epoch": 2.926825448668069e-13, "grad_norm": 0.680457592010498, "learning_rate": 0.00010133189965764798, "loss": 0.0324, "step": 42180 }, { "epoch": 2.9275193380584597e-13, "grad_norm": 0.6964467763900757, "learning_rate": 0.00010131111935544972, "loss": 0.0328, "step": 42190 }, { "epoch": 2.9282132274488504e-13, "grad_norm": 0.6496818661689758, "learning_rate": 0.00010129033712011626, "loss": 0.0339, "step": 42200 }, { "epoch": 2.928907116839241e-13, "grad_norm": 0.6742785573005676, "learning_rate": 0.0001012695529537833, "loss": 0.0356, "step": 42210 }, { "epoch": 2.929601006229632e-13, "grad_norm": 0.7365615367889404, "learning_rate": 0.00010124876685858675, "loss": 0.0329, "step": 42220 }, { "epoch": 2.9302948956200225e-13, "grad_norm": 0.7121560573577881, "learning_rate": 0.00010122797883666266, "loss": 0.0328, "step": 42230 }, { "epoch": 2.9309887850104133e-13, "grad_norm": 0.6653189659118652, "learning_rate": 0.00010120718889014736, "loss": 0.0347, "step": 42240 }, { "epoch": 2.931682674400804e-13, "grad_norm": 0.6910677552223206, "learning_rate": 0.00010118639702117732, "loss": 0.0318, "step": 42250 }, { "epoch": 2.9323765637911947e-13, "grad_norm": 0.6424713134765625, "learning_rate": 0.00010116560323188924, "loss": 0.0305, "step": 42260 }, { "epoch": 2.9330704531815854e-13, "grad_norm": 0.6985456943511963, "learning_rate": 0.00010114480752442003, "loss": 0.0313, "step": 42270 }, { "epoch": 2.933764342571976e-13, "grad_norm": 0.6598065495491028, "learning_rate": 0.00010112400990090673, "loss": 0.0309, "step": 42280 }, { "epoch": 2.934458231962367e-13, "grad_norm": 0.6916919350624084, "learning_rate": 0.00010110321036348665, "loss": 0.0297, "step": 42290 }, { "epoch": 2.9351521213527576e-13, "grad_norm": 0.6716853976249695, "learning_rate": 0.00010108240891429722, "loss": 0.0317, "step": 42300 }, { "epoch": 2.9358460107431483e-13, "grad_norm": 0.6890207529067993, "learning_rate": 0.00010106160555547617, "loss": 0.0322, "step": 42310 }, { "epoch": 2.936539900133539e-13, "grad_norm": 0.6182098388671875, "learning_rate": 0.00010104080028916132, "loss": 0.0305, "step": 42320 }, { "epoch": 2.93723378952393e-13, "grad_norm": 0.5835895538330078, "learning_rate": 0.0001010199931174908, "loss": 0.0309, "step": 42330 }, { "epoch": 2.9379276789143205e-13, "grad_norm": 0.6401651501655579, "learning_rate": 0.0001009991840426028, "loss": 0.0321, "step": 42340 }, { "epoch": 2.938621568304711e-13, "grad_norm": 0.6477852463722229, "learning_rate": 0.00010097837306663579, "loss": 0.0333, "step": 42350 }, { "epoch": 2.939315457695102e-13, "grad_norm": 0.7303540110588074, "learning_rate": 0.00010095756019172847, "loss": 0.0343, "step": 42360 }, { "epoch": 2.9400093470854927e-13, "grad_norm": 0.6744301319122314, "learning_rate": 0.00010093674542001964, "loss": 0.0327, "step": 42370 }, { "epoch": 2.9407032364758834e-13, "grad_norm": 0.7125586271286011, "learning_rate": 0.00010091592875364833, "loss": 0.0339, "step": 42380 }, { "epoch": 2.941397125866274e-13, "grad_norm": 0.7435663342475891, "learning_rate": 0.00010089511019475384, "loss": 0.0335, "step": 42390 }, { "epoch": 2.942091015256665e-13, "grad_norm": 0.7075325846672058, "learning_rate": 0.00010087428974547555, "loss": 0.0337, "step": 42400 }, { "epoch": 2.9427849046470556e-13, "grad_norm": 0.7364488244056702, "learning_rate": 0.00010085346740795309, "loss": 0.0316, "step": 42410 }, { "epoch": 2.9434787940374463e-13, "grad_norm": 0.6025335192680359, "learning_rate": 0.00010083264318432629, "loss": 0.0302, "step": 42420 }, { "epoch": 2.944172683427837e-13, "grad_norm": 0.5457717776298523, "learning_rate": 0.00010081181707673516, "loss": 0.0315, "step": 42430 }, { "epoch": 2.9448665728182277e-13, "grad_norm": 0.5848764181137085, "learning_rate": 0.00010079098908731992, "loss": 0.0298, "step": 42440 }, { "epoch": 2.9455604622086184e-13, "grad_norm": 0.6720796823501587, "learning_rate": 0.00010077015921822092, "loss": 0.0306, "step": 42450 }, { "epoch": 2.946254351599009e-13, "grad_norm": 0.6923845410346985, "learning_rate": 0.0001007493274715788, "loss": 0.0313, "step": 42460 }, { "epoch": 2.9469482409894e-13, "grad_norm": 0.6544526815414429, "learning_rate": 0.00010072849384953436, "loss": 0.0307, "step": 42470 }, { "epoch": 2.9476421303797906e-13, "grad_norm": 0.6665274500846863, "learning_rate": 0.00010070765835422854, "loss": 0.0308, "step": 42480 }, { "epoch": 2.9483360197701813e-13, "grad_norm": 0.5974776744842529, "learning_rate": 0.00010068682098780253, "loss": 0.0305, "step": 42490 }, { "epoch": 2.949029909160572e-13, "grad_norm": 0.631452202796936, "learning_rate": 0.00010066598175239768, "loss": 0.0307, "step": 42500 }, { "epoch": 2.949723798550963e-13, "grad_norm": 0.6706948280334473, "learning_rate": 0.00010064514065015555, "loss": 0.0333, "step": 42510 }, { "epoch": 2.9504176879413535e-13, "grad_norm": 0.6743252277374268, "learning_rate": 0.0001006242976832179, "loss": 0.0341, "step": 42520 }, { "epoch": 2.951111577331744e-13, "grad_norm": 0.6788199543952942, "learning_rate": 0.00010060345285372667, "loss": 0.0323, "step": 42530 }, { "epoch": 2.951805466722135e-13, "grad_norm": 0.6290085911750793, "learning_rate": 0.00010058260616382398, "loss": 0.0346, "step": 42540 }, { "epoch": 2.9524993561125257e-13, "grad_norm": 0.6632705926895142, "learning_rate": 0.00010056175761565214, "loss": 0.0343, "step": 42550 }, { "epoch": 2.9531932455029164e-13, "grad_norm": 0.6353968977928162, "learning_rate": 0.00010054090721135368, "loss": 0.0321, "step": 42560 }, { "epoch": 2.953887134893307e-13, "grad_norm": 0.6441598534584045, "learning_rate": 0.0001005200549530713, "loss": 0.0307, "step": 42570 }, { "epoch": 2.954581024283698e-13, "grad_norm": 0.7282390594482422, "learning_rate": 0.00010049920084294792, "loss": 0.0311, "step": 42580 }, { "epoch": 2.9552749136740886e-13, "grad_norm": 0.6076737642288208, "learning_rate": 0.00010047834488312658, "loss": 0.0298, "step": 42590 }, { "epoch": 2.9559688030644793e-13, "grad_norm": 0.5893675088882446, "learning_rate": 0.00010045748707575056, "loss": 0.029, "step": 42600 }, { "epoch": 2.95666269245487e-13, "grad_norm": 0.659096360206604, "learning_rate": 0.00010043662742296337, "loss": 0.0303, "step": 42610 }, { "epoch": 2.9573565818452607e-13, "grad_norm": 0.5769816040992737, "learning_rate": 0.0001004157659269086, "loss": 0.0315, "step": 42620 }, { "epoch": 2.9580504712356515e-13, "grad_norm": 0.6022499203681946, "learning_rate": 0.00010039490258973013, "loss": 0.0294, "step": 42630 }, { "epoch": 2.958744360626042e-13, "grad_norm": 0.5768997073173523, "learning_rate": 0.00010037403741357198, "loss": 0.0298, "step": 42640 }, { "epoch": 2.959438250016433e-13, "grad_norm": 0.5892210006713867, "learning_rate": 0.00010035317040057838, "loss": 0.0305, "step": 42650 }, { "epoch": 2.9601321394068236e-13, "grad_norm": 0.6787297129631042, "learning_rate": 0.00010033230155289374, "loss": 0.032, "step": 42660 }, { "epoch": 2.9608260287972143e-13, "grad_norm": 0.7406107783317566, "learning_rate": 0.00010031143087266267, "loss": 0.0329, "step": 42670 }, { "epoch": 2.961519918187605e-13, "grad_norm": 0.603127121925354, "learning_rate": 0.00010029055836202993, "loss": 0.0334, "step": 42680 }, { "epoch": 2.962213807577996e-13, "grad_norm": 0.6762076020240784, "learning_rate": 0.00010026968402314051, "loss": 0.0331, "step": 42690 }, { "epoch": 2.9629076969683865e-13, "grad_norm": 0.6441278457641602, "learning_rate": 0.00010024880785813958, "loss": 0.0332, "step": 42700 }, { "epoch": 2.963601586358777e-13, "grad_norm": 0.6548441648483276, "learning_rate": 0.00010022792986917247, "loss": 0.0347, "step": 42710 }, { "epoch": 2.964295475749168e-13, "grad_norm": 0.642418384552002, "learning_rate": 0.00010020705005838476, "loss": 0.0322, "step": 42720 }, { "epoch": 2.9649893651395587e-13, "grad_norm": 0.577645480632782, "learning_rate": 0.00010018616842792213, "loss": 0.0304, "step": 42730 }, { "epoch": 2.9656832545299494e-13, "grad_norm": 0.6317281126976013, "learning_rate": 0.00010016528497993051, "loss": 0.0316, "step": 42740 }, { "epoch": 2.96637714392034e-13, "grad_norm": 0.5871382355690002, "learning_rate": 0.000100144399716556, "loss": 0.029, "step": 42750 }, { "epoch": 2.967071033310731e-13, "grad_norm": 0.6772845983505249, "learning_rate": 0.00010012351263994487, "loss": 0.0288, "step": 42760 }, { "epoch": 2.9677649227011216e-13, "grad_norm": 0.5956639647483826, "learning_rate": 0.0001001026237522436, "loss": 0.0308, "step": 42770 }, { "epoch": 2.9684588120915123e-13, "grad_norm": 0.6021177768707275, "learning_rate": 0.0001000817330555989, "loss": 0.0302, "step": 42780 }, { "epoch": 2.969152701481903e-13, "grad_norm": 0.5915908217430115, "learning_rate": 0.00010006084055215753, "loss": 0.0292, "step": 42790 }, { "epoch": 2.9698465908722937e-13, "grad_norm": 0.6959138512611389, "learning_rate": 0.00010003994624406656, "loss": 0.0311, "step": 42800 }, { "epoch": 2.9705404802626845e-13, "grad_norm": 0.6476137638092041, "learning_rate": 0.00010001905013347321, "loss": 0.0297, "step": 42810 }, { "epoch": 2.971234369653075e-13, "grad_norm": 0.7185014486312866, "learning_rate": 9.99981522225249e-05, "loss": 0.0314, "step": 42820 }, { "epoch": 2.971928259043466e-13, "grad_norm": 0.7616618275642395, "learning_rate": 9.997725251336916e-05, "loss": 0.0335, "step": 42830 }, { "epoch": 2.9726221484338566e-13, "grad_norm": 0.6922137141227722, "learning_rate": 9.995635100815379e-05, "loss": 0.0316, "step": 42840 }, { "epoch": 2.9733160378242474e-13, "grad_norm": 0.7251935601234436, "learning_rate": 9.993544770902676e-05, "loss": 0.0341, "step": 42850 }, { "epoch": 2.974009927214638e-13, "grad_norm": 0.7430274486541748, "learning_rate": 9.99145426181362e-05, "loss": 0.0331, "step": 42860 }, { "epoch": 2.974703816605029e-13, "grad_norm": 0.67939293384552, "learning_rate": 9.989363573763039e-05, "loss": 0.0332, "step": 42870 }, { "epoch": 2.9753977059954195e-13, "grad_norm": 0.6451032161712646, "learning_rate": 9.98727270696579e-05, "loss": 0.0322, "step": 42880 }, { "epoch": 2.97609159538581e-13, "grad_norm": 0.6530218124389648, "learning_rate": 9.985181661636739e-05, "loss": 0.0303, "step": 42890 }, { "epoch": 2.976785484776201e-13, "grad_norm": 0.630229115486145, "learning_rate": 9.983090437990775e-05, "loss": 0.0304, "step": 42900 }, { "epoch": 2.9774793741665917e-13, "grad_norm": 0.6091335415840149, "learning_rate": 9.980999036242802e-05, "loss": 0.0288, "step": 42910 }, { "epoch": 2.9781732635569824e-13, "grad_norm": 0.5532868504524231, "learning_rate": 9.978907456607746e-05, "loss": 0.0293, "step": 42920 }, { "epoch": 2.978867152947373e-13, "grad_norm": 0.609160840511322, "learning_rate": 9.976815699300548e-05, "loss": 0.0308, "step": 42930 }, { "epoch": 2.979561042337764e-13, "grad_norm": 0.6294161081314087, "learning_rate": 9.974723764536169e-05, "loss": 0.0297, "step": 42940 }, { "epoch": 2.9802549317281546e-13, "grad_norm": 0.6781303882598877, "learning_rate": 9.972631652529587e-05, "loss": 0.0306, "step": 42950 }, { "epoch": 2.9809488211185453e-13, "grad_norm": 0.6073896288871765, "learning_rate": 9.970539363495798e-05, "loss": 0.0295, "step": 42960 }, { "epoch": 2.981642710508936e-13, "grad_norm": 0.6580345034599304, "learning_rate": 9.96844689764982e-05, "loss": 0.0303, "step": 42970 }, { "epoch": 2.982336599899327e-13, "grad_norm": 0.6975750923156738, "learning_rate": 9.966354255206687e-05, "loss": 0.0307, "step": 42980 }, { "epoch": 2.9830304892897175e-13, "grad_norm": 0.7263681888580322, "learning_rate": 9.964261436381448e-05, "loss": 0.0325, "step": 42990 }, { "epoch": 2.983724378680108e-13, "grad_norm": 0.7108066082000732, "learning_rate": 9.962168441389173e-05, "loss": 0.0305, "step": 43000 }, { "epoch": 2.984418268070499e-13, "grad_norm": 0.6086004376411438, "learning_rate": 9.960075270444952e-05, "loss": 0.0325, "step": 43010 }, { "epoch": 2.9851121574608896e-13, "grad_norm": 0.6793447136878967, "learning_rate": 9.957981923763888e-05, "loss": 0.0344, "step": 43020 }, { "epoch": 2.9858060468512804e-13, "grad_norm": 0.7304160594940186, "learning_rate": 9.955888401561109e-05, "loss": 0.0323, "step": 43030 }, { "epoch": 2.986499936241671e-13, "grad_norm": 0.6786839365959167, "learning_rate": 9.95379470405175e-05, "loss": 0.0312, "step": 43040 }, { "epoch": 2.987193825632062e-13, "grad_norm": 0.6175243258476257, "learning_rate": 9.95170083145098e-05, "loss": 0.0316, "step": 43050 }, { "epoch": 2.9878877150224525e-13, "grad_norm": 0.6104013919830322, "learning_rate": 9.94960678397397e-05, "loss": 0.0295, "step": 43060 }, { "epoch": 2.9885816044128433e-13, "grad_norm": 0.611754834651947, "learning_rate": 9.947512561835917e-05, "loss": 0.0284, "step": 43070 }, { "epoch": 2.989275493803234e-13, "grad_norm": 0.6514986753463745, "learning_rate": 9.945418165252036e-05, "loss": 0.0298, "step": 43080 }, { "epoch": 2.9899693831936247e-13, "grad_norm": 0.5714368224143982, "learning_rate": 9.943323594437562e-05, "loss": 0.0297, "step": 43090 }, { "epoch": 2.9906632725840154e-13, "grad_norm": 0.6512865424156189, "learning_rate": 9.941228849607739e-05, "loss": 0.0293, "step": 43100 }, { "epoch": 2.991357161974406e-13, "grad_norm": 0.663731038570404, "learning_rate": 9.939133930977838e-05, "loss": 0.0298, "step": 43110 }, { "epoch": 2.992051051364797e-13, "grad_norm": 0.617608368396759, "learning_rate": 9.937038838763143e-05, "loss": 0.03, "step": 43120 }, { "epoch": 2.9927449407551876e-13, "grad_norm": 0.6404427886009216, "learning_rate": 9.934943573178959e-05, "loss": 0.0309, "step": 43130 }, { "epoch": 2.9934388301455783e-13, "grad_norm": 0.6390275359153748, "learning_rate": 9.932848134440606e-05, "loss": 0.0306, "step": 43140 }, { "epoch": 2.994132719535969e-13, "grad_norm": 0.6432192921638489, "learning_rate": 9.930752522763424e-05, "loss": 0.0312, "step": 43150 }, { "epoch": 2.99482660892636e-13, "grad_norm": 0.6934531331062317, "learning_rate": 9.92865673836277e-05, "loss": 0.0311, "step": 43160 }, { "epoch": 2.9955204983167505e-13, "grad_norm": 0.662616491317749, "learning_rate": 9.926560781454016e-05, "loss": 0.032, "step": 43170 }, { "epoch": 2.996214387707141e-13, "grad_norm": 0.630315899848938, "learning_rate": 9.924464652252557e-05, "loss": 0.0339, "step": 43180 }, { "epoch": 2.996908277097532e-13, "grad_norm": 0.621777355670929, "learning_rate": 9.922368350973798e-05, "loss": 0.0311, "step": 43190 }, { "epoch": 2.9976021664879227e-13, "grad_norm": 0.6702921390533447, "learning_rate": 9.920271877833171e-05, "loss": 0.0318, "step": 43200 }, { "epoch": 2.9982960558783134e-13, "grad_norm": 0.5715734958648682, "learning_rate": 9.918175233046123e-05, "loss": 0.0313, "step": 43210 }, { "epoch": 2.998989945268704e-13, "grad_norm": 0.5512141585350037, "learning_rate": 9.916078416828112e-05, "loss": 0.0284, "step": 43220 }, { "epoch": 2.999683834659095e-13, "grad_norm": 0.6317028403282166, "learning_rate": 9.91398142939462e-05, "loss": 0.0292, "step": 43230 }, { "epoch": 3.0003777240494856e-13, "grad_norm": 0.6963872909545898, "learning_rate": 9.911884270961147e-05, "loss": 0.0297, "step": 43240 }, { "epoch": 3.0010716134398763e-13, "grad_norm": 0.6076423525810242, "learning_rate": 9.909786941743207e-05, "loss": 0.029, "step": 43250 }, { "epoch": 3.001765502830267e-13, "grad_norm": 0.7131039500236511, "learning_rate": 9.907689441956334e-05, "loss": 0.0293, "step": 43260 }, { "epoch": 3.0024593922206577e-13, "grad_norm": 0.6608413457870483, "learning_rate": 9.905591771816079e-05, "loss": 0.0306, "step": 43270 }, { "epoch": 3.0031532816110484e-13, "grad_norm": 0.6463358402252197, "learning_rate": 9.903493931538008e-05, "loss": 0.0301, "step": 43280 }, { "epoch": 3.003847171001439e-13, "grad_norm": 0.6230844259262085, "learning_rate": 9.90139592133771e-05, "loss": 0.0304, "step": 43290 }, { "epoch": 3.00454106039183e-13, "grad_norm": 0.621405303478241, "learning_rate": 9.899297741430783e-05, "loss": 0.0316, "step": 43300 }, { "epoch": 3.0052349497822206e-13, "grad_norm": 0.619528591632843, "learning_rate": 9.897199392032855e-05, "loss": 0.0298, "step": 43310 }, { "epoch": 3.0059288391726113e-13, "grad_norm": 0.6666383147239685, "learning_rate": 9.895100873359559e-05, "loss": 0.0324, "step": 43320 }, { "epoch": 3.006622728563002e-13, "grad_norm": 0.7400027513504028, "learning_rate": 9.89300218562655e-05, "loss": 0.0324, "step": 43330 }, { "epoch": 3.007316617953393e-13, "grad_norm": 0.7641668915748596, "learning_rate": 9.890903329049505e-05, "loss": 0.0323, "step": 43340 }, { "epoch": 3.0080105073437835e-13, "grad_norm": 0.7026227712631226, "learning_rate": 9.88880430384411e-05, "loss": 0.0311, "step": 43350 }, { "epoch": 3.008704396734174e-13, "grad_norm": 0.7484764456748962, "learning_rate": 9.886705110226074e-05, "loss": 0.0314, "step": 43360 }, { "epoch": 3.009398286124565e-13, "grad_norm": 0.6828743815422058, "learning_rate": 9.884605748411124e-05, "loss": 0.0313, "step": 43370 }, { "epoch": 3.0100921755149557e-13, "grad_norm": 0.6509206891059875, "learning_rate": 9.882506218614998e-05, "loss": 0.029, "step": 43380 }, { "epoch": 3.0107860649053464e-13, "grad_norm": 0.6001594066619873, "learning_rate": 9.880406521053456e-05, "loss": 0.0287, "step": 43390 }, { "epoch": 3.011479954295737e-13, "grad_norm": 0.7444248199462891, "learning_rate": 9.878306655942277e-05, "loss": 0.0313, "step": 43400 }, { "epoch": 3.012173843686128e-13, "grad_norm": 0.6509586572647095, "learning_rate": 9.876206623497256e-05, "loss": 0.0279, "step": 43410 }, { "epoch": 3.0128677330765186e-13, "grad_norm": 0.587126612663269, "learning_rate": 9.874106423934199e-05, "loss": 0.0298, "step": 43420 }, { "epoch": 3.0135616224669093e-13, "grad_norm": 0.624123752117157, "learning_rate": 9.872006057468939e-05, "loss": 0.03, "step": 43430 }, { "epoch": 3.0142555118573e-13, "grad_norm": 0.6040538549423218, "learning_rate": 9.86990552431732e-05, "loss": 0.0305, "step": 43440 }, { "epoch": 3.0149494012476907e-13, "grad_norm": 0.5962653756141663, "learning_rate": 9.8678048246952e-05, "loss": 0.03, "step": 43450 }, { "epoch": 3.0156432906380815e-13, "grad_norm": 0.6188221573829651, "learning_rate": 9.865703958818466e-05, "loss": 0.031, "step": 43460 }, { "epoch": 3.016337180028472e-13, "grad_norm": 0.7173623442649841, "learning_rate": 9.86360292690301e-05, "loss": 0.0295, "step": 43470 }, { "epoch": 3.017031069418863e-13, "grad_norm": 0.6654420495033264, "learning_rate": 9.861501729164747e-05, "loss": 0.0309, "step": 43480 }, { "epoch": 3.0177249588092536e-13, "grad_norm": 0.6772895455360413, "learning_rate": 9.859400365819606e-05, "loss": 0.033, "step": 43490 }, { "epoch": 3.0184188481996443e-13, "grad_norm": 0.7226951718330383, "learning_rate": 9.857298837083536e-05, "loss": 0.0321, "step": 43500 }, { "epoch": 3.019112737590035e-13, "grad_norm": 0.7105949521064758, "learning_rate": 9.855197143172503e-05, "loss": 0.0315, "step": 43510 }, { "epoch": 3.019806626980426e-13, "grad_norm": 0.6890221834182739, "learning_rate": 9.85309528430249e-05, "loss": 0.0314, "step": 43520 }, { "epoch": 3.0205005163708165e-13, "grad_norm": 0.669043779373169, "learning_rate": 9.850993260689491e-05, "loss": 0.0311, "step": 43530 }, { "epoch": 3.021194405761207e-13, "grad_norm": 0.697148859500885, "learning_rate": 9.848891072549524e-05, "loss": 0.0288, "step": 43540 }, { "epoch": 3.021888295151598e-13, "grad_norm": 0.671267569065094, "learning_rate": 9.846788720098621e-05, "loss": 0.0298, "step": 43550 }, { "epoch": 3.0225821845419887e-13, "grad_norm": 0.6704240441322327, "learning_rate": 9.844686203552833e-05, "loss": 0.03, "step": 43560 }, { "epoch": 3.0232760739323794e-13, "grad_norm": 0.6060469150543213, "learning_rate": 9.842583523128227e-05, "loss": 0.0281, "step": 43570 }, { "epoch": 3.02396996332277e-13, "grad_norm": 0.6417121291160583, "learning_rate": 9.840480679040881e-05, "loss": 0.0296, "step": 43580 }, { "epoch": 3.024663852713161e-13, "grad_norm": 0.6505590677261353, "learning_rate": 9.838377671506904e-05, "loss": 0.0307, "step": 43590 }, { "epoch": 3.0253577421035516e-13, "grad_norm": 0.6532476544380188, "learning_rate": 9.836274500742402e-05, "loss": 0.0305, "step": 43600 }, { "epoch": 3.0260516314939423e-13, "grad_norm": 0.6250622868537903, "learning_rate": 9.834171166963514e-05, "loss": 0.0312, "step": 43610 }, { "epoch": 3.026745520884333e-13, "grad_norm": 0.6196281909942627, "learning_rate": 9.832067670386392e-05, "loss": 0.0295, "step": 43620 }, { "epoch": 3.0274394102747237e-13, "grad_norm": 0.6891713738441467, "learning_rate": 9.8299640112272e-05, "loss": 0.0304, "step": 43630 }, { "epoch": 3.0281332996651145e-13, "grad_norm": 0.6664081811904907, "learning_rate": 9.827860189702123e-05, "loss": 0.0308, "step": 43640 }, { "epoch": 3.028827189055505e-13, "grad_norm": 0.67582768201828, "learning_rate": 9.825756206027361e-05, "loss": 0.0326, "step": 43650 }, { "epoch": 3.029521078445896e-13, "grad_norm": 0.7834383845329285, "learning_rate": 9.823652060419131e-05, "loss": 0.0313, "step": 43660 }, { "epoch": 3.0302149678362866e-13, "grad_norm": 0.5952619314193726, "learning_rate": 9.821547753093668e-05, "loss": 0.0298, "step": 43670 }, { "epoch": 3.0309088572266774e-13, "grad_norm": 0.5641648173332214, "learning_rate": 9.81944328426722e-05, "loss": 0.0306, "step": 43680 }, { "epoch": 3.031602746617068e-13, "grad_norm": 0.5916054248809814, "learning_rate": 9.817338654156056e-05, "loss": 0.0299, "step": 43690 }, { "epoch": 3.032296636007459e-13, "grad_norm": 0.638672947883606, "learning_rate": 9.815233862976456e-05, "loss": 0.0302, "step": 43700 }, { "epoch": 3.0329905253978495e-13, "grad_norm": 0.6621240973472595, "learning_rate": 9.813128910944725e-05, "loss": 0.0298, "step": 43710 }, { "epoch": 3.03368441478824e-13, "grad_norm": 0.631158709526062, "learning_rate": 9.811023798277176e-05, "loss": 0.0286, "step": 43720 }, { "epoch": 3.034378304178631e-13, "grad_norm": 0.6463781595230103, "learning_rate": 9.808918525190144e-05, "loss": 0.0288, "step": 43730 }, { "epoch": 3.0350721935690217e-13, "grad_norm": 0.6139430403709412, "learning_rate": 9.806813091899977e-05, "loss": 0.0293, "step": 43740 }, { "epoch": 3.0357660829594124e-13, "grad_norm": 0.6857646107673645, "learning_rate": 9.804707498623045e-05, "loss": 0.0299, "step": 43750 }, { "epoch": 3.036459972349803e-13, "grad_norm": 0.6766214370727539, "learning_rate": 9.802601745575723e-05, "loss": 0.0307, "step": 43760 }, { "epoch": 3.037153861740194e-13, "grad_norm": 0.6250574588775635, "learning_rate": 9.800495832974417e-05, "loss": 0.0312, "step": 43770 }, { "epoch": 3.0378477511305846e-13, "grad_norm": 0.6474721431732178, "learning_rate": 9.798389761035539e-05, "loss": 0.029, "step": 43780 }, { "epoch": 3.0385416405209753e-13, "grad_norm": 0.6125244498252869, "learning_rate": 9.796283529975521e-05, "loss": 0.031, "step": 43790 }, { "epoch": 3.039235529911366e-13, "grad_norm": 0.6279101967811584, "learning_rate": 9.794177140010811e-05, "loss": 0.0313, "step": 43800 }, { "epoch": 3.039929419301757e-13, "grad_norm": 0.6213160157203674, "learning_rate": 9.792070591357873e-05, "loss": 0.0304, "step": 43810 }, { "epoch": 3.0406233086921475e-13, "grad_norm": 0.6835373640060425, "learning_rate": 9.789963884233188e-05, "loss": 0.0303, "step": 43820 }, { "epoch": 3.041317198082538e-13, "grad_norm": 0.7448740601539612, "learning_rate": 9.787857018853253e-05, "loss": 0.031, "step": 43830 }, { "epoch": 3.042011087472929e-13, "grad_norm": 0.6429094672203064, "learning_rate": 9.785749995434582e-05, "loss": 0.0296, "step": 43840 }, { "epoch": 3.0427049768633196e-13, "grad_norm": 0.5934818387031555, "learning_rate": 9.783642814193705e-05, "loss": 0.029, "step": 43850 }, { "epoch": 3.0433988662537104e-13, "grad_norm": 0.6539469957351685, "learning_rate": 9.781535475347166e-05, "loss": 0.03, "step": 43860 }, { "epoch": 3.044092755644101e-13, "grad_norm": 0.5647606253623962, "learning_rate": 9.779427979111525e-05, "loss": 0.0311, "step": 43870 }, { "epoch": 3.044786645034492e-13, "grad_norm": 0.5728639960289001, "learning_rate": 9.777320325703364e-05, "loss": 0.0277, "step": 43880 }, { "epoch": 3.0454805344248825e-13, "grad_norm": 0.5707852244377136, "learning_rate": 9.775212515339276e-05, "loss": 0.0283, "step": 43890 }, { "epoch": 3.0461744238152733e-13, "grad_norm": 0.6360562443733215, "learning_rate": 9.77310454823587e-05, "loss": 0.03, "step": 43900 }, { "epoch": 3.046868313205664e-13, "grad_norm": 0.6682756543159485, "learning_rate": 9.770996424609773e-05, "loss": 0.0304, "step": 43910 }, { "epoch": 3.0475622025960547e-13, "grad_norm": 0.73850017786026, "learning_rate": 9.768888144677629e-05, "loss": 0.0316, "step": 43920 }, { "epoch": 3.0482560919864454e-13, "grad_norm": 0.5928657054901123, "learning_rate": 9.766779708656094e-05, "loss": 0.0301, "step": 43930 }, { "epoch": 3.048949981376836e-13, "grad_norm": 0.6453010439872742, "learning_rate": 9.764671116761844e-05, "loss": 0.0296, "step": 43940 }, { "epoch": 3.049643870767227e-13, "grad_norm": 0.6323205232620239, "learning_rate": 9.76256236921157e-05, "loss": 0.0309, "step": 43950 }, { "epoch": 3.0503377601576176e-13, "grad_norm": 0.6708399057388306, "learning_rate": 9.760453466221978e-05, "loss": 0.032, "step": 43960 }, { "epoch": 3.0510316495480083e-13, "grad_norm": 0.6178486347198486, "learning_rate": 9.758344408009792e-05, "loss": 0.0304, "step": 43970 }, { "epoch": 3.051725538938399e-13, "grad_norm": 0.6250672340393066, "learning_rate": 9.756235194791748e-05, "loss": 0.03, "step": 43980 }, { "epoch": 3.05241942832879e-13, "grad_norm": 0.6354818344116211, "learning_rate": 9.7541258267846e-05, "loss": 0.0308, "step": 43990 }, { "epoch": 3.0531133177191805e-13, "grad_norm": 0.5801980495452881, "learning_rate": 9.75201630420512e-05, "loss": 0.029, "step": 44000 }, { "epoch": 3.053807207109571e-13, "grad_norm": 0.6879044771194458, "learning_rate": 9.749906627270099e-05, "loss": 0.0288, "step": 44010 }, { "epoch": 3.054501096499962e-13, "grad_norm": 0.5896983742713928, "learning_rate": 9.747796796196331e-05, "loss": 0.0292, "step": 44020 }, { "epoch": 3.0551949858903527e-13, "grad_norm": 0.58832848072052, "learning_rate": 9.745686811200639e-05, "loss": 0.0303, "step": 44030 }, { "epoch": 3.0558888752807434e-13, "grad_norm": 0.5444359183311462, "learning_rate": 9.743576672499853e-05, "loss": 0.0276, "step": 44040 }, { "epoch": 3.056582764671134e-13, "grad_norm": 0.6877012848854065, "learning_rate": 9.741466380310826e-05, "loss": 0.029, "step": 44050 }, { "epoch": 3.057276654061525e-13, "grad_norm": 0.6234868168830872, "learning_rate": 9.739355934850424e-05, "loss": 0.0283, "step": 44060 }, { "epoch": 3.0579705434519155e-13, "grad_norm": 0.6526547074317932, "learning_rate": 9.737245336335524e-05, "loss": 0.0293, "step": 44070 }, { "epoch": 3.0586644328423063e-13, "grad_norm": 0.6845564246177673, "learning_rate": 9.735134584983026e-05, "loss": 0.0315, "step": 44080 }, { "epoch": 3.059358322232697e-13, "grad_norm": 0.6523955464363098, "learning_rate": 9.73302368100984e-05, "loss": 0.0291, "step": 44090 }, { "epoch": 3.0600522116230877e-13, "grad_norm": 0.7568230032920837, "learning_rate": 9.730912624632898e-05, "loss": 0.031, "step": 44100 }, { "epoch": 3.0607461010134784e-13, "grad_norm": 0.7216835618019104, "learning_rate": 9.728801416069143e-05, "loss": 0.0304, "step": 44110 }, { "epoch": 3.061439990403869e-13, "grad_norm": 0.6743960380554199, "learning_rate": 9.726690055535532e-05, "loss": 0.0308, "step": 44120 }, { "epoch": 3.06213387979426e-13, "grad_norm": 0.6407556533813477, "learning_rate": 9.724578543249042e-05, "loss": 0.0301, "step": 44130 }, { "epoch": 3.0628277691846506e-13, "grad_norm": 0.6450367569923401, "learning_rate": 9.722466879426662e-05, "loss": 0.0288, "step": 44140 }, { "epoch": 3.0635216585750413e-13, "grad_norm": 0.6679278612136841, "learning_rate": 9.7203550642854e-05, "loss": 0.03, "step": 44150 }, { "epoch": 3.064215547965432e-13, "grad_norm": 0.608702540397644, "learning_rate": 9.718243098042278e-05, "loss": 0.028, "step": 44160 }, { "epoch": 3.064909437355823e-13, "grad_norm": 0.5459784269332886, "learning_rate": 9.71613098091433e-05, "loss": 0.0287, "step": 44170 }, { "epoch": 3.0656033267462135e-13, "grad_norm": 0.6198534369468689, "learning_rate": 9.714018713118617e-05, "loss": 0.0294, "step": 44180 }, { "epoch": 3.066297216136604e-13, "grad_norm": 0.6391916871070862, "learning_rate": 9.711906294872198e-05, "loss": 0.0285, "step": 44190 }, { "epoch": 3.066991105526995e-13, "grad_norm": 0.6183887124061584, "learning_rate": 9.709793726392162e-05, "loss": 0.0285, "step": 44200 }, { "epoch": 3.0676849949173857e-13, "grad_norm": 0.5892254710197449, "learning_rate": 9.707681007895608e-05, "loss": 0.0282, "step": 44210 }, { "epoch": 3.0683788843077764e-13, "grad_norm": 0.6502280235290527, "learning_rate": 9.705568139599648e-05, "loss": 0.029, "step": 44220 }, { "epoch": 3.069072773698167e-13, "grad_norm": 0.7025021910667419, "learning_rate": 9.703455121721417e-05, "loss": 0.0295, "step": 44230 }, { "epoch": 3.069766663088558e-13, "grad_norm": 0.6490840911865234, "learning_rate": 9.701341954478056e-05, "loss": 0.0303, "step": 44240 }, { "epoch": 3.0704605524789486e-13, "grad_norm": 0.6608484983444214, "learning_rate": 9.699228638086729e-05, "loss": 0.0289, "step": 44250 }, { "epoch": 3.0711544418693393e-13, "grad_norm": 0.5842410922050476, "learning_rate": 9.697115172764606e-05, "loss": 0.0305, "step": 44260 }, { "epoch": 3.07184833125973e-13, "grad_norm": 0.6294547319412231, "learning_rate": 9.695001558728885e-05, "loss": 0.0317, "step": 44270 }, { "epoch": 3.0725422206501207e-13, "grad_norm": 0.7233222126960754, "learning_rate": 9.692887796196773e-05, "loss": 0.0304, "step": 44280 }, { "epoch": 3.0732361100405114e-13, "grad_norm": 0.6759064197540283, "learning_rate": 9.690773885385486e-05, "loss": 0.0291, "step": 44290 }, { "epoch": 3.073929999430902e-13, "grad_norm": 0.6335878372192383, "learning_rate": 9.688659826512266e-05, "loss": 0.0294, "step": 44300 }, { "epoch": 3.074623888821293e-13, "grad_norm": 0.6355639100074768, "learning_rate": 9.686545619794365e-05, "loss": 0.0287, "step": 44310 }, { "epoch": 3.0753177782116836e-13, "grad_norm": 0.6187055110931396, "learning_rate": 9.68443126544905e-05, "loss": 0.0276, "step": 44320 }, { "epoch": 3.0760116676020743e-13, "grad_norm": 0.6429044008255005, "learning_rate": 9.682316763693603e-05, "loss": 0.0286, "step": 44330 }, { "epoch": 3.076705556992465e-13, "grad_norm": 0.606601357460022, "learning_rate": 9.680202114745324e-05, "loss": 0.0289, "step": 44340 }, { "epoch": 3.077399446382856e-13, "grad_norm": 0.635152280330658, "learning_rate": 9.678087318821525e-05, "loss": 0.0279, "step": 44350 }, { "epoch": 3.0780933357732465e-13, "grad_norm": 0.6379396915435791, "learning_rate": 9.675972376139533e-05, "loss": 0.0272, "step": 44360 }, { "epoch": 3.078787225163637e-13, "grad_norm": 0.6124791502952576, "learning_rate": 9.67385728691669e-05, "loss": 0.0279, "step": 44370 }, { "epoch": 3.079481114554028e-13, "grad_norm": 0.6357905268669128, "learning_rate": 9.671742051370359e-05, "loss": 0.0291, "step": 44380 }, { "epoch": 3.0801750039444187e-13, "grad_norm": 0.61397784948349, "learning_rate": 9.669626669717908e-05, "loss": 0.0291, "step": 44390 }, { "epoch": 3.0808688933348094e-13, "grad_norm": 0.6251562833786011, "learning_rate": 9.667511142176731e-05, "loss": 0.0289, "step": 44400 }, { "epoch": 3.0815627827252e-13, "grad_norm": 0.6534612774848938, "learning_rate": 9.66539546896423e-05, "loss": 0.0293, "step": 44410 }, { "epoch": 3.082256672115591e-13, "grad_norm": 0.5750834941864014, "learning_rate": 9.663279650297818e-05, "loss": 0.0296, "step": 44420 }, { "epoch": 3.0829505615059816e-13, "grad_norm": 0.5699148774147034, "learning_rate": 9.661163686394938e-05, "loss": 0.0313, "step": 44430 }, { "epoch": 3.0836444508963723e-13, "grad_norm": 0.6025230884552002, "learning_rate": 9.659047577473028e-05, "loss": 0.0297, "step": 44440 }, { "epoch": 3.084338340286763e-13, "grad_norm": 0.6698747873306274, "learning_rate": 9.656931323749557e-05, "loss": 0.0295, "step": 44450 }, { "epoch": 3.0850322296771537e-13, "grad_norm": 0.5777143239974976, "learning_rate": 9.654814925442003e-05, "loss": 0.0291, "step": 44460 }, { "epoch": 3.0857261190675445e-13, "grad_norm": 0.5575671195983887, "learning_rate": 9.652698382767854e-05, "loss": 0.028, "step": 44470 }, { "epoch": 3.086420008457935e-13, "grad_norm": 0.640675961971283, "learning_rate": 9.650581695944622e-05, "loss": 0.028, "step": 44480 }, { "epoch": 3.087113897848326e-13, "grad_norm": 0.6542297005653381, "learning_rate": 9.64846486518983e-05, "loss": 0.0286, "step": 44490 }, { "epoch": 3.0878077872387166e-13, "grad_norm": 0.5666239857673645, "learning_rate": 9.646347890721011e-05, "loss": 0.0284, "step": 44500 }, { "epoch": 3.0885016766291074e-13, "grad_norm": 0.6380034685134888, "learning_rate": 9.644230772755724e-05, "loss": 0.0274, "step": 44510 }, { "epoch": 3.089195566019498e-13, "grad_norm": 0.6174631714820862, "learning_rate": 9.64211351151153e-05, "loss": 0.0275, "step": 44520 }, { "epoch": 3.089889455409889e-13, "grad_norm": 0.6196539998054504, "learning_rate": 9.639996107206011e-05, "loss": 0.0276, "step": 44530 }, { "epoch": 3.0905833448002795e-13, "grad_norm": 0.5821378827095032, "learning_rate": 9.637878560056768e-05, "loss": 0.0287, "step": 44540 }, { "epoch": 3.09127723419067e-13, "grad_norm": 0.5859251618385315, "learning_rate": 9.635760870281408e-05, "loss": 0.0301, "step": 44550 }, { "epoch": 3.091971123581061e-13, "grad_norm": 0.6059665679931641, "learning_rate": 9.633643038097555e-05, "loss": 0.0281, "step": 44560 }, { "epoch": 3.0926650129714517e-13, "grad_norm": 0.6295042634010315, "learning_rate": 9.631525063722852e-05, "loss": 0.0298, "step": 44570 }, { "epoch": 3.0933589023618424e-13, "grad_norm": 0.6647229790687561, "learning_rate": 9.629406947374953e-05, "loss": 0.0305, "step": 44580 }, { "epoch": 3.094052791752233e-13, "grad_norm": 0.6455488204956055, "learning_rate": 9.627288689271528e-05, "loss": 0.0294, "step": 44590 }, { "epoch": 3.094746681142624e-13, "grad_norm": 0.6379033327102661, "learning_rate": 9.62517028963026e-05, "loss": 0.0288, "step": 44600 }, { "epoch": 3.0954405705330146e-13, "grad_norm": 0.7086061239242554, "learning_rate": 9.623051748668849e-05, "loss": 0.0286, "step": 44610 }, { "epoch": 3.0961344599234053e-13, "grad_norm": 0.6460553407669067, "learning_rate": 9.620933066605004e-05, "loss": 0.0286, "step": 44620 }, { "epoch": 3.096828349313796e-13, "grad_norm": 0.659558892250061, "learning_rate": 9.618814243656459e-05, "loss": 0.0279, "step": 44630 }, { "epoch": 3.097522238704187e-13, "grad_norm": 0.6091549396514893, "learning_rate": 9.616695280040952e-05, "loss": 0.0282, "step": 44640 }, { "epoch": 3.0982161280945775e-13, "grad_norm": 0.7002169489860535, "learning_rate": 9.61457617597624e-05, "loss": 0.0294, "step": 44650 }, { "epoch": 3.098910017484968e-13, "grad_norm": 0.6228944063186646, "learning_rate": 9.612456931680093e-05, "loss": 0.0269, "step": 44660 }, { "epoch": 3.099603906875359e-13, "grad_norm": 0.5166511535644531, "learning_rate": 9.610337547370298e-05, "loss": 0.0273, "step": 44670 }, { "epoch": 3.1002977962657496e-13, "grad_norm": 0.5958866477012634, "learning_rate": 9.608218023264654e-05, "loss": 0.0274, "step": 44680 }, { "epoch": 3.1009916856561404e-13, "grad_norm": 0.5570656061172485, "learning_rate": 9.606098359580975e-05, "loss": 0.0274, "step": 44690 }, { "epoch": 3.101685575046531e-13, "grad_norm": 0.5686476826667786, "learning_rate": 9.60397855653709e-05, "loss": 0.0277, "step": 44700 }, { "epoch": 3.102379464436922e-13, "grad_norm": 0.608399510383606, "learning_rate": 9.601858614350842e-05, "loss": 0.0296, "step": 44710 }, { "epoch": 3.1030733538273125e-13, "grad_norm": 0.6355164051055908, "learning_rate": 9.599738533240088e-05, "loss": 0.0277, "step": 44720 }, { "epoch": 3.1037672432177033e-13, "grad_norm": 0.621671736240387, "learning_rate": 9.597618313422696e-05, "loss": 0.0284, "step": 44730 }, { "epoch": 3.104461132608094e-13, "grad_norm": 0.6007792949676514, "learning_rate": 9.595497955116556e-05, "loss": 0.0297, "step": 44740 }, { "epoch": 3.1051550219984847e-13, "grad_norm": 0.6344917416572571, "learning_rate": 9.593377458539564e-05, "loss": 0.0287, "step": 44750 }, { "epoch": 3.1058489113888754e-13, "grad_norm": 0.6291694045066833, "learning_rate": 9.591256823909637e-05, "loss": 0.0286, "step": 44760 }, { "epoch": 3.106542800779266e-13, "grad_norm": 0.6340035200119019, "learning_rate": 9.589136051444705e-05, "loss": 0.0286, "step": 44770 }, { "epoch": 3.107236690169657e-13, "grad_norm": 0.638484537601471, "learning_rate": 9.587015141362708e-05, "loss": 0.0287, "step": 44780 }, { "epoch": 3.1079305795600476e-13, "grad_norm": 0.6583892703056335, "learning_rate": 9.584894093881598e-05, "loss": 0.0271, "step": 44790 }, { "epoch": 3.1086244689504383e-13, "grad_norm": 0.6273078918457031, "learning_rate": 9.582772909219351e-05, "loss": 0.0288, "step": 44800 }, { "epoch": 3.109318358340829e-13, "grad_norm": 0.6542932391166687, "learning_rate": 9.580651587593952e-05, "loss": 0.0285, "step": 44810 }, { "epoch": 3.11001224773122e-13, "grad_norm": 0.577697217464447, "learning_rate": 9.578530129223398e-05, "loss": 0.0261, "step": 44820 }, { "epoch": 3.1107061371216105e-13, "grad_norm": 0.5822138786315918, "learning_rate": 9.576408534325702e-05, "loss": 0.0276, "step": 44830 }, { "epoch": 3.111400026512001e-13, "grad_norm": 0.6004728078842163, "learning_rate": 9.574286803118891e-05, "loss": 0.0272, "step": 44840 }, { "epoch": 3.112093915902392e-13, "grad_norm": 0.5650214552879333, "learning_rate": 9.572164935821005e-05, "loss": 0.0276, "step": 44850 }, { "epoch": 3.1127878052927827e-13, "grad_norm": 0.6230296492576599, "learning_rate": 9.570042932650102e-05, "loss": 0.0284, "step": 44860 }, { "epoch": 3.1134816946831734e-13, "grad_norm": 0.6030629277229309, "learning_rate": 9.567920793824247e-05, "loss": 0.0272, "step": 44870 }, { "epoch": 3.114175584073564e-13, "grad_norm": 0.6619849801063538, "learning_rate": 9.565798519561527e-05, "loss": 0.0289, "step": 44880 }, { "epoch": 3.114869473463955e-13, "grad_norm": 0.6936413645744324, "learning_rate": 9.563676110080036e-05, "loss": 0.0291, "step": 44890 }, { "epoch": 3.1155633628543455e-13, "grad_norm": 0.6470478177070618, "learning_rate": 9.561553565597883e-05, "loss": 0.029, "step": 44900 }, { "epoch": 3.1162572522447363e-13, "grad_norm": 0.6937232613563538, "learning_rate": 9.559430886333194e-05, "loss": 0.0282, "step": 44910 }, { "epoch": 3.116951141635127e-13, "grad_norm": 0.5586386919021606, "learning_rate": 9.557308072504106e-05, "loss": 0.0275, "step": 44920 }, { "epoch": 3.1176450310255177e-13, "grad_norm": 0.5355756878852844, "learning_rate": 9.555185124328775e-05, "loss": 0.0282, "step": 44930 }, { "epoch": 3.1183389204159084e-13, "grad_norm": 0.5909143090248108, "learning_rate": 9.553062042025365e-05, "loss": 0.0275, "step": 44940 }, { "epoch": 3.119032809806299e-13, "grad_norm": 0.6404951810836792, "learning_rate": 9.550938825812054e-05, "loss": 0.0281, "step": 44950 }, { "epoch": 3.11972669919669e-13, "grad_norm": 0.6860111951828003, "learning_rate": 9.548815475907038e-05, "loss": 0.0287, "step": 44960 }, { "epoch": 3.1204205885870806e-13, "grad_norm": 0.6115120053291321, "learning_rate": 9.546691992528522e-05, "loss": 0.0274, "step": 44970 }, { "epoch": 3.1211144779774713e-13, "grad_norm": 0.6618744134902954, "learning_rate": 9.544568375894728e-05, "loss": 0.0273, "step": 44980 }, { "epoch": 3.121808367367862e-13, "grad_norm": 0.5947539210319519, "learning_rate": 9.542444626223894e-05, "loss": 0.0271, "step": 44990 }, { "epoch": 3.122502256758253e-13, "grad_norm": 0.6270471215248108, "learning_rate": 9.540320743734263e-05, "loss": 0.027, "step": 45000 }, { "epoch": 3.1231961461486435e-13, "grad_norm": 0.6286842823028564, "learning_rate": 9.5381967286441e-05, "loss": 0.0275, "step": 45010 }, { "epoch": 3.123890035539034e-13, "grad_norm": 0.6078721284866333, "learning_rate": 9.536072581171678e-05, "loss": 0.0283, "step": 45020 }, { "epoch": 3.124583924929425e-13, "grad_norm": 0.6294148564338684, "learning_rate": 9.533948301535287e-05, "loss": 0.0267, "step": 45030 }, { "epoch": 3.1252778143198157e-13, "grad_norm": 0.6120913028717041, "learning_rate": 9.531823889953233e-05, "loss": 0.0292, "step": 45040 }, { "epoch": 3.1259717037102064e-13, "grad_norm": 0.63112872838974, "learning_rate": 9.52969934664383e-05, "loss": 0.0297, "step": 45050 }, { "epoch": 3.126665593100597e-13, "grad_norm": 0.6068421602249146, "learning_rate": 9.527574671825405e-05, "loss": 0.0276, "step": 45060 }, { "epoch": 3.127359482490988e-13, "grad_norm": 0.660443127155304, "learning_rate": 9.525449865716307e-05, "loss": 0.0276, "step": 45070 }, { "epoch": 3.1280533718813786e-13, "grad_norm": 0.6864109635353088, "learning_rate": 9.523324928534889e-05, "loss": 0.0281, "step": 45080 }, { "epoch": 3.1287472612717693e-13, "grad_norm": 0.6041078567504883, "learning_rate": 9.521199860499524e-05, "loss": 0.0268, "step": 45090 }, { "epoch": 3.12944115066216e-13, "grad_norm": 0.5653950572013855, "learning_rate": 9.519074661828594e-05, "loss": 0.0264, "step": 45100 }, { "epoch": 3.1301350400525507e-13, "grad_norm": 0.6482602953910828, "learning_rate": 9.516949332740498e-05, "loss": 0.0272, "step": 45110 }, { "epoch": 3.1308289294429414e-13, "grad_norm": 0.8062494993209839, "learning_rate": 9.514823873453642e-05, "loss": 0.0296, "step": 45120 }, { "epoch": 3.131522818833332e-13, "grad_norm": 0.5550071597099304, "learning_rate": 9.512698284186455e-05, "loss": 0.0266, "step": 45130 }, { "epoch": 3.132216708223723e-13, "grad_norm": 0.5755850672721863, "learning_rate": 9.510572565157369e-05, "loss": 0.027, "step": 45140 }, { "epoch": 3.1329105976141136e-13, "grad_norm": 0.5630673766136169, "learning_rate": 9.508446716584843e-05, "loss": 0.0272, "step": 45150 }, { "epoch": 3.1336044870045043e-13, "grad_norm": 0.6291399002075195, "learning_rate": 9.506320738687332e-05, "loss": 0.0272, "step": 45160 }, { "epoch": 3.134298376394895e-13, "grad_norm": 0.6847826838493347, "learning_rate": 9.504194631683322e-05, "loss": 0.0279, "step": 45170 }, { "epoch": 3.134992265785286e-13, "grad_norm": 0.5400248169898987, "learning_rate": 9.502068395791298e-05, "loss": 0.0275, "step": 45180 }, { "epoch": 3.1356861551756765e-13, "grad_norm": 0.6055764555931091, "learning_rate": 9.499942031229762e-05, "loss": 0.0269, "step": 45190 }, { "epoch": 3.136380044566067e-13, "grad_norm": 0.6230140924453735, "learning_rate": 9.497815538217239e-05, "loss": 0.0284, "step": 45200 }, { "epoch": 3.137073933956458e-13, "grad_norm": 0.6098566055297852, "learning_rate": 9.495688916972251e-05, "loss": 0.0294, "step": 45210 }, { "epoch": 3.1377678233468487e-13, "grad_norm": 0.619138240814209, "learning_rate": 9.493562167713345e-05, "loss": 0.0276, "step": 45220 }, { "epoch": 3.1384617127372394e-13, "grad_norm": 0.6068234443664551, "learning_rate": 9.491435290659077e-05, "loss": 0.0269, "step": 45230 }, { "epoch": 3.13915560212763e-13, "grad_norm": 0.6199199557304382, "learning_rate": 9.489308286028019e-05, "loss": 0.0282, "step": 45240 }, { "epoch": 3.139849491518021e-13, "grad_norm": 0.5672628283500671, "learning_rate": 9.48718115403875e-05, "loss": 0.0267, "step": 45250 }, { "epoch": 3.1405433809084116e-13, "grad_norm": 0.6714605689048767, "learning_rate": 9.485053894909868e-05, "loss": 0.0263, "step": 45260 }, { "epoch": 3.1412372702988023e-13, "grad_norm": 0.5798627138137817, "learning_rate": 9.482926508859982e-05, "loss": 0.0278, "step": 45270 }, { "epoch": 3.141931159689193e-13, "grad_norm": 0.6016163229942322, "learning_rate": 9.480798996107712e-05, "loss": 0.0282, "step": 45280 }, { "epoch": 3.1426250490795837e-13, "grad_norm": 0.5542532801628113, "learning_rate": 9.478671356871698e-05, "loss": 0.0263, "step": 45290 }, { "epoch": 3.1433189384699745e-13, "grad_norm": 0.6543612480163574, "learning_rate": 9.476543591370584e-05, "loss": 0.0274, "step": 45300 }, { "epoch": 3.144012827860365e-13, "grad_norm": 0.5754370093345642, "learning_rate": 9.47441569982303e-05, "loss": 0.0262, "step": 45310 }, { "epoch": 3.144706717250756e-13, "grad_norm": 0.6020365953445435, "learning_rate": 9.472287682447713e-05, "loss": 0.0272, "step": 45320 }, { "epoch": 3.1454006066411466e-13, "grad_norm": 0.6366329193115234, "learning_rate": 9.47015953946332e-05, "loss": 0.0287, "step": 45330 }, { "epoch": 3.1460944960315373e-13, "grad_norm": 0.603252649307251, "learning_rate": 9.468031271088549e-05, "loss": 0.0262, "step": 45340 }, { "epoch": 3.146788385421928e-13, "grad_norm": 0.6798623204231262, "learning_rate": 9.465902877542111e-05, "loss": 0.0283, "step": 45350 }, { "epoch": 3.147482274812319e-13, "grad_norm": 0.6721739768981934, "learning_rate": 9.463774359042735e-05, "loss": 0.0283, "step": 45360 }, { "epoch": 3.1481761642027095e-13, "grad_norm": 0.6240401268005371, "learning_rate": 9.461645715809156e-05, "loss": 0.0283, "step": 45370 }, { "epoch": 3.1488700535931e-13, "grad_norm": 0.6228607892990112, "learning_rate": 9.459516948060132e-05, "loss": 0.0279, "step": 45380 }, { "epoch": 3.149563942983491e-13, "grad_norm": 0.6507647037506104, "learning_rate": 9.457388056014417e-05, "loss": 0.0263, "step": 45390 }, { "epoch": 3.1502578323738817e-13, "grad_norm": 0.6370537877082825, "learning_rate": 9.455259039890796e-05, "loss": 0.0277, "step": 45400 }, { "epoch": 3.1509517217642724e-13, "grad_norm": 0.6132138967514038, "learning_rate": 9.453129899908056e-05, "loss": 0.027, "step": 45410 }, { "epoch": 3.151645611154663e-13, "grad_norm": 0.5196923613548279, "learning_rate": 9.451000636284999e-05, "loss": 0.0263, "step": 45420 }, { "epoch": 3.152339500545054e-13, "grad_norm": 0.6498682498931885, "learning_rate": 9.448871249240441e-05, "loss": 0.0283, "step": 45430 }, { "epoch": 3.1530333899354446e-13, "grad_norm": 0.620496928691864, "learning_rate": 9.446741738993204e-05, "loss": 0.0274, "step": 45440 }, { "epoch": 3.1537272793258353e-13, "grad_norm": 0.6268341541290283, "learning_rate": 9.444612105762136e-05, "loss": 0.0268, "step": 45450 }, { "epoch": 3.154421168716226e-13, "grad_norm": 0.5992475748062134, "learning_rate": 9.442482349766088e-05, "loss": 0.0264, "step": 45460 }, { "epoch": 3.1551150581066167e-13, "grad_norm": 0.591097354888916, "learning_rate": 9.44035247122392e-05, "loss": 0.0268, "step": 45470 }, { "epoch": 3.1558089474970075e-13, "grad_norm": 0.6596060991287231, "learning_rate": 9.438222470354518e-05, "loss": 0.0272, "step": 45480 }, { "epoch": 3.156502836887398e-13, "grad_norm": 0.6180923581123352, "learning_rate": 9.43609234737677e-05, "loss": 0.0272, "step": 45490 }, { "epoch": 3.157196726277789e-13, "grad_norm": 0.6236153244972229, "learning_rate": 9.433962102509575e-05, "loss": 0.0261, "step": 45500 }, { "epoch": 3.1578906156681796e-13, "grad_norm": 0.5474125146865845, "learning_rate": 9.431831735971854e-05, "loss": 0.0277, "step": 45510 }, { "epoch": 3.1585845050585704e-13, "grad_norm": 0.5668367147445679, "learning_rate": 9.429701247982533e-05, "loss": 0.029, "step": 45520 }, { "epoch": 3.159278394448961e-13, "grad_norm": 0.6532398462295532, "learning_rate": 9.427570638760558e-05, "loss": 0.0272, "step": 45530 }, { "epoch": 3.159972283839352e-13, "grad_norm": 0.6337958574295044, "learning_rate": 9.425439908524873e-05, "loss": 0.0271, "step": 45540 }, { "epoch": 3.1606661732297425e-13, "grad_norm": 0.5949790477752686, "learning_rate": 9.423309057494448e-05, "loss": 0.0277, "step": 45550 }, { "epoch": 3.161360062620133e-13, "grad_norm": 0.6056762933731079, "learning_rate": 9.421178085888264e-05, "loss": 0.0271, "step": 45560 }, { "epoch": 3.162053952010524e-13, "grad_norm": 0.5988747477531433, "learning_rate": 9.419046993925307e-05, "loss": 0.0259, "step": 45570 }, { "epoch": 3.1627478414009147e-13, "grad_norm": 0.6453702449798584, "learning_rate": 9.416915781824583e-05, "loss": 0.0267, "step": 45580 }, { "epoch": 3.1634417307913054e-13, "grad_norm": 0.5872904658317566, "learning_rate": 9.414784449805106e-05, "loss": 0.0283, "step": 45590 }, { "epoch": 3.164135620181696e-13, "grad_norm": 0.6125164031982422, "learning_rate": 9.412652998085902e-05, "loss": 0.0269, "step": 45600 }, { "epoch": 3.164829509572087e-13, "grad_norm": 0.6247146725654602, "learning_rate": 9.410521426886012e-05, "loss": 0.0268, "step": 45610 }, { "epoch": 3.1655233989624776e-13, "grad_norm": 0.6059311032295227, "learning_rate": 9.408389736424491e-05, "loss": 0.0265, "step": 45620 }, { "epoch": 3.1662172883528683e-13, "grad_norm": 0.5983449816703796, "learning_rate": 9.406257926920401e-05, "loss": 0.0273, "step": 45630 }, { "epoch": 3.166911177743259e-13, "grad_norm": 0.5962136387825012, "learning_rate": 9.404125998592817e-05, "loss": 0.0275, "step": 45640 }, { "epoch": 3.16760506713365e-13, "grad_norm": 0.5659757852554321, "learning_rate": 9.40199395166083e-05, "loss": 0.0263, "step": 45650 }, { "epoch": 3.1682989565240405e-13, "grad_norm": 0.6013432145118713, "learning_rate": 9.399861786343539e-05, "loss": 0.0266, "step": 45660 }, { "epoch": 3.168992845914431e-13, "grad_norm": 0.5965667963027954, "learning_rate": 9.39772950286006e-05, "loss": 0.0268, "step": 45670 }, { "epoch": 3.169686735304822e-13, "grad_norm": 0.5539023280143738, "learning_rate": 9.395597101429514e-05, "loss": 0.0281, "step": 45680 }, { "epoch": 3.1703806246952126e-13, "grad_norm": 0.5482134819030762, "learning_rate": 9.393464582271044e-05, "loss": 0.0261, "step": 45690 }, { "epoch": 3.1710745140856034e-13, "grad_norm": 0.5881778597831726, "learning_rate": 9.391331945603795e-05, "loss": 0.0274, "step": 45700 }, { "epoch": 3.171768403475994e-13, "grad_norm": 0.5259479880332947, "learning_rate": 9.389199191646932e-05, "loss": 0.0276, "step": 45710 }, { "epoch": 3.172462292866385e-13, "grad_norm": 0.5536108016967773, "learning_rate": 9.387066320619624e-05, "loss": 0.0261, "step": 45720 }, { "epoch": 3.1731561822567755e-13, "grad_norm": 0.588038444519043, "learning_rate": 9.384933332741061e-05, "loss": 0.0264, "step": 45730 }, { "epoch": 3.1738500716471663e-13, "grad_norm": 0.6502087116241455, "learning_rate": 9.382800228230441e-05, "loss": 0.0268, "step": 45740 }, { "epoch": 3.174543961037557e-13, "grad_norm": 0.6318624019622803, "learning_rate": 9.38066700730697e-05, "loss": 0.027, "step": 45750 }, { "epoch": 3.1752378504279477e-13, "grad_norm": 0.6447232365608215, "learning_rate": 9.378533670189871e-05, "loss": 0.0267, "step": 45760 }, { "epoch": 3.1759317398183384e-13, "grad_norm": 0.594931960105896, "learning_rate": 9.376400217098379e-05, "loss": 0.0272, "step": 45770 }, { "epoch": 3.176625629208729e-13, "grad_norm": 0.6281158328056335, "learning_rate": 9.37426664825174e-05, "loss": 0.0264, "step": 45780 }, { "epoch": 3.17731951859912e-13, "grad_norm": 0.5917810201644897, "learning_rate": 9.37213296386921e-05, "loss": 0.0266, "step": 45790 }, { "epoch": 3.1780134079895106e-13, "grad_norm": 0.5729508399963379, "learning_rate": 9.369999164170058e-05, "loss": 0.028, "step": 45800 }, { "epoch": 3.1787072973799013e-13, "grad_norm": 0.5851069092750549, "learning_rate": 9.367865249373564e-05, "loss": 0.0257, "step": 45810 }, { "epoch": 3.179401186770292e-13, "grad_norm": 0.598071277141571, "learning_rate": 9.365731219699024e-05, "loss": 0.0276, "step": 45820 }, { "epoch": 3.180095076160683e-13, "grad_norm": 0.6250261068344116, "learning_rate": 9.36359707536574e-05, "loss": 0.0271, "step": 45830 }, { "epoch": 3.1807889655510735e-13, "grad_norm": 0.6358203887939453, "learning_rate": 9.361462816593028e-05, "loss": 0.0264, "step": 45840 }, { "epoch": 3.181482854941464e-13, "grad_norm": 0.6164734959602356, "learning_rate": 9.359328443600223e-05, "loss": 0.0252, "step": 45850 }, { "epoch": 3.182176744331855e-13, "grad_norm": 0.66863614320755, "learning_rate": 9.357193956606657e-05, "loss": 0.0261, "step": 45860 }, { "epoch": 3.1828706337222457e-13, "grad_norm": 0.6065816879272461, "learning_rate": 9.355059355831685e-05, "loss": 0.0267, "step": 45870 }, { "epoch": 3.1835645231126364e-13, "grad_norm": 0.599337637424469, "learning_rate": 9.35292464149467e-05, "loss": 0.0256, "step": 45880 }, { "epoch": 3.184258412503027e-13, "grad_norm": 0.5513173937797546, "learning_rate": 9.350789813814987e-05, "loss": 0.0256, "step": 45890 }, { "epoch": 3.184952301893418e-13, "grad_norm": 0.679758608341217, "learning_rate": 9.348654873012021e-05, "loss": 0.028, "step": 45900 }, { "epoch": 3.1856461912838085e-13, "grad_norm": 0.6407814621925354, "learning_rate": 9.346519819305177e-05, "loss": 0.0259, "step": 45910 }, { "epoch": 3.1863400806741993e-13, "grad_norm": 0.5380088686943054, "learning_rate": 9.344384652913858e-05, "loss": 0.0269, "step": 45920 }, { "epoch": 3.18703397006459e-13, "grad_norm": 0.5754967331886292, "learning_rate": 9.34224937405749e-05, "loss": 0.0268, "step": 45930 }, { "epoch": 3.1877278594549807e-13, "grad_norm": 0.537575364112854, "learning_rate": 9.340113982955501e-05, "loss": 0.0268, "step": 45940 }, { "epoch": 3.1884217488453714e-13, "grad_norm": 0.5687212347984314, "learning_rate": 9.337978479827341e-05, "loss": 0.0265, "step": 45950 }, { "epoch": 3.189115638235762e-13, "grad_norm": 0.5848460793495178, "learning_rate": 9.335842864892465e-05, "loss": 0.0278, "step": 45960 }, { "epoch": 3.189809527626153e-13, "grad_norm": 0.6443522572517395, "learning_rate": 9.333707138370338e-05, "loss": 0.0257, "step": 45970 }, { "epoch": 3.1905034170165436e-13, "grad_norm": 0.6169628500938416, "learning_rate": 9.33157130048044e-05, "loss": 0.026, "step": 45980 }, { "epoch": 3.1911973064069343e-13, "grad_norm": 0.5594403743743896, "learning_rate": 9.329435351442266e-05, "loss": 0.0267, "step": 45990 }, { "epoch": 3.191891195797325e-13, "grad_norm": 0.6220539808273315, "learning_rate": 9.327299291475309e-05, "loss": 0.0259, "step": 46000 }, { "epoch": 3.192585085187716e-13, "grad_norm": 0.6468659043312073, "learning_rate": 9.325163120799091e-05, "loss": 0.0257, "step": 46010 }, { "epoch": 3.1932789745781065e-13, "grad_norm": 0.6565113067626953, "learning_rate": 9.323026839633132e-05, "loss": 0.026, "step": 46020 }, { "epoch": 3.193972863968497e-13, "grad_norm": 0.6103517413139343, "learning_rate": 9.320890448196971e-05, "loss": 0.026, "step": 46030 }, { "epoch": 3.194666753358888e-13, "grad_norm": 0.62332683801651, "learning_rate": 9.318753946710154e-05, "loss": 0.0249, "step": 46040 }, { "epoch": 3.1953606427492787e-13, "grad_norm": 0.6065613031387329, "learning_rate": 9.31661733539224e-05, "loss": 0.0259, "step": 46050 }, { "epoch": 3.1960545321396694e-13, "grad_norm": 0.6580379009246826, "learning_rate": 9.314480614462798e-05, "loss": 0.0269, "step": 46060 }, { "epoch": 3.19674842153006e-13, "grad_norm": 0.5755429267883301, "learning_rate": 9.312343784141412e-05, "loss": 0.0257, "step": 46070 }, { "epoch": 3.197442310920451e-13, "grad_norm": 0.5792060494422913, "learning_rate": 9.310206844647671e-05, "loss": 0.0264, "step": 46080 }, { "epoch": 3.1981362003108416e-13, "grad_norm": 0.597564160823822, "learning_rate": 9.308069796201182e-05, "loss": 0.026, "step": 46090 }, { "epoch": 3.1988300897012323e-13, "grad_norm": 0.5512620210647583, "learning_rate": 9.305932639021559e-05, "loss": 0.0264, "step": 46100 }, { "epoch": 3.199523979091623e-13, "grad_norm": 0.6276112198829651, "learning_rate": 9.303795373328425e-05, "loss": 0.0276, "step": 46110 }, { "epoch": 3.2002178684820137e-13, "grad_norm": 0.5818544626235962, "learning_rate": 9.301657999341422e-05, "loss": 0.0259, "step": 46120 }, { "epoch": 3.2009117578724045e-13, "grad_norm": 0.6218962669372559, "learning_rate": 9.299520517280198e-05, "loss": 0.0265, "step": 46130 }, { "epoch": 3.201605647262795e-13, "grad_norm": 0.6070834994316101, "learning_rate": 9.297382927364409e-05, "loss": 0.0261, "step": 46140 }, { "epoch": 3.202299536653186e-13, "grad_norm": 0.5991001129150391, "learning_rate": 9.295245229813726e-05, "loss": 0.0272, "step": 46150 }, { "epoch": 3.2029934260435766e-13, "grad_norm": 0.6731329560279846, "learning_rate": 9.293107424847833e-05, "loss": 0.0252, "step": 46160 }, { "epoch": 3.2036873154339673e-13, "grad_norm": 0.5608423352241516, "learning_rate": 9.290969512686426e-05, "loss": 0.0247, "step": 46170 }, { "epoch": 3.204381204824358e-13, "grad_norm": 0.52590411901474, "learning_rate": 9.288831493549203e-05, "loss": 0.0264, "step": 46180 }, { "epoch": 3.205075094214749e-13, "grad_norm": 0.5892799496650696, "learning_rate": 9.28669336765588e-05, "loss": 0.0251, "step": 46190 }, { "epoch": 3.2057689836051395e-13, "grad_norm": 0.6029329299926758, "learning_rate": 9.28455513522618e-05, "loss": 0.0265, "step": 46200 }, { "epoch": 3.20646287299553e-13, "grad_norm": 0.6232374310493469, "learning_rate": 9.282416796479847e-05, "loss": 0.0261, "step": 46210 }, { "epoch": 3.207156762385921e-13, "grad_norm": 0.6211069226264954, "learning_rate": 9.280278351636623e-05, "loss": 0.0252, "step": 46220 }, { "epoch": 3.2078506517763117e-13, "grad_norm": 0.6431276202201843, "learning_rate": 9.278139800916269e-05, "loss": 0.0262, "step": 46230 }, { "epoch": 3.2085445411667024e-13, "grad_norm": 0.5893673896789551, "learning_rate": 9.276001144538555e-05, "loss": 0.0265, "step": 46240 }, { "epoch": 3.209238430557093e-13, "grad_norm": 0.6838138103485107, "learning_rate": 9.273862382723257e-05, "loss": 0.0259, "step": 46250 }, { "epoch": 3.209932319947484e-13, "grad_norm": 0.629554033279419, "learning_rate": 9.271723515690167e-05, "loss": 0.0263, "step": 46260 }, { "epoch": 3.2106262093378746e-13, "grad_norm": 0.5935543775558472, "learning_rate": 9.26958454365909e-05, "loss": 0.0271, "step": 46270 }, { "epoch": 3.2113200987282653e-13, "grad_norm": 0.6011525988578796, "learning_rate": 9.267445466849835e-05, "loss": 0.0255, "step": 46280 }, { "epoch": 3.212013988118656e-13, "grad_norm": 0.5529034733772278, "learning_rate": 9.265306285482231e-05, "loss": 0.027, "step": 46290 }, { "epoch": 3.2127078775090467e-13, "grad_norm": 0.5571406483650208, "learning_rate": 9.263166999776109e-05, "loss": 0.0269, "step": 46300 }, { "epoch": 3.2134017668994375e-13, "grad_norm": 0.5285430550575256, "learning_rate": 9.261027609951311e-05, "loss": 0.0255, "step": 46310 }, { "epoch": 3.214095656289828e-13, "grad_norm": 0.562929630279541, "learning_rate": 9.258888116227696e-05, "loss": 0.0246, "step": 46320 }, { "epoch": 3.214789545680219e-13, "grad_norm": 0.655827522277832, "learning_rate": 9.256748518825131e-05, "loss": 0.0255, "step": 46330 }, { "epoch": 3.2154834350706096e-13, "grad_norm": 0.5734859704971313, "learning_rate": 9.254608817963489e-05, "loss": 0.0258, "step": 46340 }, { "epoch": 3.2161773244610004e-13, "grad_norm": 0.5948672890663147, "learning_rate": 9.252469013862662e-05, "loss": 0.0253, "step": 46350 }, { "epoch": 3.216871213851391e-13, "grad_norm": 0.607109785079956, "learning_rate": 9.250329106742544e-05, "loss": 0.0256, "step": 46360 }, { "epoch": 3.217565103241782e-13, "grad_norm": 0.528352677822113, "learning_rate": 9.248189096823047e-05, "loss": 0.0267, "step": 46370 }, { "epoch": 3.2182589926321725e-13, "grad_norm": 0.5821390151977539, "learning_rate": 9.246048984324089e-05, "loss": 0.0247, "step": 46380 }, { "epoch": 3.218952882022563e-13, "grad_norm": 0.5660746097564697, "learning_rate": 9.2439087694656e-05, "loss": 0.025, "step": 46390 }, { "epoch": 3.219646771412954e-13, "grad_norm": 0.5301609635353088, "learning_rate": 9.241768452467522e-05, "loss": 0.0263, "step": 46400 }, { "epoch": 3.2203406608033447e-13, "grad_norm": 0.5998611450195312, "learning_rate": 9.239628033549801e-05, "loss": 0.0267, "step": 46410 }, { "epoch": 3.2210345501937354e-13, "grad_norm": 0.6892158389091492, "learning_rate": 9.237487512932405e-05, "loss": 0.0263, "step": 46420 }, { "epoch": 3.221728439584126e-13, "grad_norm": 0.5558634400367737, "learning_rate": 9.2353468908353e-05, "loss": 0.0263, "step": 46430 }, { "epoch": 3.222422328974517e-13, "grad_norm": 0.6009860634803772, "learning_rate": 9.23320616747847e-05, "loss": 0.0258, "step": 46440 }, { "epoch": 3.2231162183649076e-13, "grad_norm": 0.5880527496337891, "learning_rate": 9.23106534308191e-05, "loss": 0.026, "step": 46450 }, { "epoch": 3.2238101077552983e-13, "grad_norm": 0.5958804488182068, "learning_rate": 9.228924417865621e-05, "loss": 0.0273, "step": 46460 }, { "epoch": 3.224503997145689e-13, "grad_norm": 0.5675816535949707, "learning_rate": 9.226783392049616e-05, "loss": 0.0249, "step": 46470 }, { "epoch": 3.22519788653608e-13, "grad_norm": 0.5440008640289307, "learning_rate": 9.22464226585392e-05, "loss": 0.0246, "step": 46480 }, { "epoch": 3.2258917759264705e-13, "grad_norm": 0.5887829661369324, "learning_rate": 9.222501039498566e-05, "loss": 0.0255, "step": 46490 }, { "epoch": 3.226585665316861e-13, "grad_norm": 0.5350705981254578, "learning_rate": 9.220359713203602e-05, "loss": 0.0245, "step": 46500 }, { "epoch": 3.227279554707252e-13, "grad_norm": 0.5907357931137085, "learning_rate": 9.218218287189076e-05, "loss": 0.0245, "step": 46510 }, { "epoch": 3.2279734440976426e-13, "grad_norm": 0.5376310348510742, "learning_rate": 9.216076761675056e-05, "loss": 0.0255, "step": 46520 }, { "epoch": 3.2286673334880334e-13, "grad_norm": 0.549482524394989, "learning_rate": 9.213935136881617e-05, "loss": 0.0254, "step": 46530 }, { "epoch": 3.229361222878424e-13, "grad_norm": 0.5464864373207092, "learning_rate": 9.211793413028847e-05, "loss": 0.0246, "step": 46540 }, { "epoch": 3.230055112268815e-13, "grad_norm": 0.6508569121360779, "learning_rate": 9.209651590336837e-05, "loss": 0.0259, "step": 46550 }, { "epoch": 3.2307490016592055e-13, "grad_norm": 0.5721670985221863, "learning_rate": 9.207509669025694e-05, "loss": 0.0253, "step": 46560 }, { "epoch": 3.2314428910495963e-13, "grad_norm": 0.5651334524154663, "learning_rate": 9.205367649315537e-05, "loss": 0.0263, "step": 46570 }, { "epoch": 3.232136780439987e-13, "grad_norm": 0.6280550360679626, "learning_rate": 9.203225531426486e-05, "loss": 0.0267, "step": 46580 }, { "epoch": 3.2328306698303777e-13, "grad_norm": 0.6060566306114197, "learning_rate": 9.201083315578681e-05, "loss": 0.0246, "step": 46590 }, { "epoch": 3.2335245592207684e-13, "grad_norm": 0.6674287915229797, "learning_rate": 9.198941001992268e-05, "loss": 0.0272, "step": 46600 }, { "epoch": 3.234218448611159e-13, "grad_norm": 0.687288224697113, "learning_rate": 9.196798590887404e-05, "loss": 0.0257, "step": 46610 }, { "epoch": 3.23491233800155e-13, "grad_norm": 0.6069150567054749, "learning_rate": 9.194656082484252e-05, "loss": 0.0257, "step": 46620 }, { "epoch": 3.2356062273919406e-13, "grad_norm": 0.5568283200263977, "learning_rate": 9.192513477002988e-05, "loss": 0.025, "step": 46630 }, { "epoch": 3.2363001167823313e-13, "grad_norm": 0.5776405334472656, "learning_rate": 9.190370774663799e-05, "loss": 0.0237, "step": 46640 }, { "epoch": 3.236994006172722e-13, "grad_norm": 0.5948516726493835, "learning_rate": 9.188227975686882e-05, "loss": 0.0248, "step": 46650 }, { "epoch": 3.237687895563113e-13, "grad_norm": 0.5488926768302917, "learning_rate": 9.186085080292443e-05, "loss": 0.0236, "step": 46660 }, { "epoch": 3.2383817849535035e-13, "grad_norm": 0.49698519706726074, "learning_rate": 9.183942088700695e-05, "loss": 0.0246, "step": 46670 }, { "epoch": 3.239075674343894e-13, "grad_norm": 0.5703169703483582, "learning_rate": 9.181799001131867e-05, "loss": 0.0253, "step": 46680 }, { "epoch": 3.239769563734285e-13, "grad_norm": 0.6021973490715027, "learning_rate": 9.179655817806197e-05, "loss": 0.0243, "step": 46690 }, { "epoch": 3.2404634531246757e-13, "grad_norm": 0.603164792060852, "learning_rate": 9.177512538943924e-05, "loss": 0.0254, "step": 46700 }, { "epoch": 3.2411573425150664e-13, "grad_norm": 0.5743541717529297, "learning_rate": 9.175369164765306e-05, "loss": 0.0251, "step": 46710 }, { "epoch": 3.241851231905457e-13, "grad_norm": 0.6174156069755554, "learning_rate": 9.173225695490608e-05, "loss": 0.0262, "step": 46720 }, { "epoch": 3.242545121295848e-13, "grad_norm": 0.6214199662208557, "learning_rate": 9.171082131340108e-05, "loss": 0.0258, "step": 46730 }, { "epoch": 3.2432390106862385e-13, "grad_norm": 0.6113168597221375, "learning_rate": 9.168938472534083e-05, "loss": 0.0267, "step": 46740 }, { "epoch": 3.2439329000766293e-13, "grad_norm": 0.6276829242706299, "learning_rate": 9.166794719292833e-05, "loss": 0.0251, "step": 46750 }, { "epoch": 3.24462678946702e-13, "grad_norm": 0.5300297737121582, "learning_rate": 9.164650871836662e-05, "loss": 0.0264, "step": 46760 }, { "epoch": 3.2453206788574107e-13, "grad_norm": 0.5583218336105347, "learning_rate": 9.162506930385883e-05, "loss": 0.0267, "step": 46770 }, { "epoch": 3.2460145682478014e-13, "grad_norm": 0.6232838034629822, "learning_rate": 9.160362895160817e-05, "loss": 0.0247, "step": 46780 }, { "epoch": 3.246708457638192e-13, "grad_norm": 0.6037397384643555, "learning_rate": 9.1582187663818e-05, "loss": 0.0241, "step": 46790 }, { "epoch": 3.247402347028583e-13, "grad_norm": 0.5917115211486816, "learning_rate": 9.156074544269174e-05, "loss": 0.0246, "step": 46800 }, { "epoch": 3.2480962364189736e-13, "grad_norm": 0.5492146611213684, "learning_rate": 9.15393022904329e-05, "loss": 0.0242, "step": 46810 }, { "epoch": 3.2487901258093643e-13, "grad_norm": 0.5751981735229492, "learning_rate": 9.151785820924511e-05, "loss": 0.0231, "step": 46820 }, { "epoch": 3.249484015199755e-13, "grad_norm": 0.6149616241455078, "learning_rate": 9.149641320133207e-05, "loss": 0.0248, "step": 46830 }, { "epoch": 3.250177904590146e-13, "grad_norm": 0.5406672358512878, "learning_rate": 9.147496726889761e-05, "loss": 0.0248, "step": 46840 }, { "epoch": 3.2508717939805365e-13, "grad_norm": 0.5834351778030396, "learning_rate": 9.145352041414562e-05, "loss": 0.0242, "step": 46850 }, { "epoch": 3.251565683370927e-13, "grad_norm": 0.6227256655693054, "learning_rate": 9.143207263928006e-05, "loss": 0.0245, "step": 46860 }, { "epoch": 3.252259572761318e-13, "grad_norm": 0.5756033062934875, "learning_rate": 9.141062394650509e-05, "loss": 0.0251, "step": 46870 }, { "epoch": 3.2529534621517087e-13, "grad_norm": 0.5991106629371643, "learning_rate": 9.138917433802487e-05, "loss": 0.0261, "step": 46880 }, { "epoch": 3.2536473515420994e-13, "grad_norm": 0.594515323638916, "learning_rate": 9.136772381604366e-05, "loss": 0.0264, "step": 46890 }, { "epoch": 3.25434124093249e-13, "grad_norm": 0.5445026159286499, "learning_rate": 9.134627238276587e-05, "loss": 0.0252, "step": 46900 }, { "epoch": 3.255035130322881e-13, "grad_norm": 0.5848159790039062, "learning_rate": 9.132482004039593e-05, "loss": 0.0253, "step": 46910 }, { "epoch": 3.2557290197132716e-13, "grad_norm": 0.5351141691207886, "learning_rate": 9.130336679113844e-05, "loss": 0.0258, "step": 46920 }, { "epoch": 3.2564229091036623e-13, "grad_norm": 0.5069361329078674, "learning_rate": 9.128191263719803e-05, "loss": 0.0262, "step": 46930 }, { "epoch": 3.257116798494053e-13, "grad_norm": 0.5148259997367859, "learning_rate": 9.126045758077946e-05, "loss": 0.0241, "step": 46940 }, { "epoch": 3.2578106878844437e-13, "grad_norm": 0.6021772027015686, "learning_rate": 9.123900162408758e-05, "loss": 0.0244, "step": 46950 }, { "epoch": 3.2585045772748344e-13, "grad_norm": 0.5206978917121887, "learning_rate": 9.12175447693273e-05, "loss": 0.0246, "step": 46960 }, { "epoch": 3.259198466665225e-13, "grad_norm": 0.501653790473938, "learning_rate": 9.119608701870364e-05, "loss": 0.0235, "step": 46970 }, { "epoch": 3.259892356055616e-13, "grad_norm": 0.5787191987037659, "learning_rate": 9.117462837442174e-05, "loss": 0.0242, "step": 46980 }, { "epoch": 3.2605862454460066e-13, "grad_norm": 0.6334424614906311, "learning_rate": 9.115316883868682e-05, "loss": 0.0246, "step": 46990 }, { "epoch": 3.2612801348363973e-13, "grad_norm": 0.5476463437080383, "learning_rate": 9.113170841370415e-05, "loss": 0.0242, "step": 47000 }, { "epoch": 3.261974024226788e-13, "grad_norm": 0.626733660697937, "learning_rate": 9.111024710167914e-05, "loss": 0.0244, "step": 47010 }, { "epoch": 3.262667913617179e-13, "grad_norm": 0.6105243563652039, "learning_rate": 9.108878490481727e-05, "loss": 0.0252, "step": 47020 }, { "epoch": 3.2633618030075695e-13, "grad_norm": 0.6480897665023804, "learning_rate": 9.106732182532411e-05, "loss": 0.0246, "step": 47030 }, { "epoch": 3.26405569239796e-13, "grad_norm": 0.580447256565094, "learning_rate": 9.104585786540535e-05, "loss": 0.0259, "step": 47040 }, { "epoch": 3.264749581788351e-13, "grad_norm": 0.5817169547080994, "learning_rate": 9.102439302726671e-05, "loss": 0.0263, "step": 47050 }, { "epoch": 3.2654434711787417e-13, "grad_norm": 0.5902524590492249, "learning_rate": 9.100292731311405e-05, "loss": 0.0242, "step": 47060 }, { "epoch": 3.2661373605691324e-13, "grad_norm": 0.6300840973854065, "learning_rate": 9.098146072515333e-05, "loss": 0.026, "step": 47070 }, { "epoch": 3.266831249959523e-13, "grad_norm": 0.6152220368385315, "learning_rate": 9.095999326559054e-05, "loss": 0.0257, "step": 47080 }, { "epoch": 3.267525139349914e-13, "grad_norm": 0.5964124202728271, "learning_rate": 9.09385249366318e-05, "loss": 0.0252, "step": 47090 }, { "epoch": 3.2682190287403046e-13, "grad_norm": 0.5657888054847717, "learning_rate": 9.091705574048333e-05, "loss": 0.0241, "step": 47100 }, { "epoch": 3.2689129181306953e-13, "grad_norm": 0.6145143508911133, "learning_rate": 9.089558567935143e-05, "loss": 0.0242, "step": 47110 }, { "epoch": 3.269606807521086e-13, "grad_norm": 0.5546163320541382, "learning_rate": 9.087411475544246e-05, "loss": 0.0243, "step": 47120 }, { "epoch": 3.2703006969114767e-13, "grad_norm": 0.5671824216842651, "learning_rate": 9.08526429709629e-05, "loss": 0.0235, "step": 47130 }, { "epoch": 3.2709945863018675e-13, "grad_norm": 0.5321338772773743, "learning_rate": 9.083117032811932e-05, "loss": 0.0234, "step": 47140 }, { "epoch": 3.271688475692258e-13, "grad_norm": 0.6408839225769043, "learning_rate": 9.080969682911838e-05, "loss": 0.025, "step": 47150 }, { "epoch": 3.272382365082649e-13, "grad_norm": 0.5603000521659851, "learning_rate": 9.07882224761668e-05, "loss": 0.0229, "step": 47160 }, { "epoch": 3.2730762544730396e-13, "grad_norm": 0.47553789615631104, "learning_rate": 9.076674727147138e-05, "loss": 0.0239, "step": 47170 }, { "epoch": 3.2737701438634303e-13, "grad_norm": 0.5668596625328064, "learning_rate": 9.074527121723907e-05, "loss": 0.0249, "step": 47180 }, { "epoch": 3.274464033253821e-13, "grad_norm": 0.5789200663566589, "learning_rate": 9.072379431567682e-05, "loss": 0.0252, "step": 47190 }, { "epoch": 3.275157922644212e-13, "grad_norm": 0.5578937530517578, "learning_rate": 9.070231656899178e-05, "loss": 0.0251, "step": 47200 }, { "epoch": 3.2758518120346025e-13, "grad_norm": 0.5398679375648499, "learning_rate": 9.068083797939106e-05, "loss": 0.026, "step": 47210 }, { "epoch": 3.276545701424993e-13, "grad_norm": 0.6260742545127869, "learning_rate": 9.065935854908197e-05, "loss": 0.0247, "step": 47220 }, { "epoch": 3.277239590815384e-13, "grad_norm": 0.6390319466590881, "learning_rate": 9.063787828027183e-05, "loss": 0.0249, "step": 47230 }, { "epoch": 3.2779334802057747e-13, "grad_norm": 0.5869815349578857, "learning_rate": 9.061639717516807e-05, "loss": 0.0259, "step": 47240 }, { "epoch": 3.2786273695961654e-13, "grad_norm": 0.6066563725471497, "learning_rate": 9.059491523597823e-05, "loss": 0.0247, "step": 47250 }, { "epoch": 3.279321258986556e-13, "grad_norm": 0.6152095198631287, "learning_rate": 9.057343246490987e-05, "loss": 0.0242, "step": 47260 }, { "epoch": 3.280015148376947e-13, "grad_norm": 0.6060041785240173, "learning_rate": 9.055194886417076e-05, "loss": 0.0243, "step": 47270 }, { "epoch": 3.2807090377673376e-13, "grad_norm": 0.591062605381012, "learning_rate": 9.053046443596858e-05, "loss": 0.0242, "step": 47280 }, { "epoch": 3.2814029271577283e-13, "grad_norm": 0.564928412437439, "learning_rate": 9.050897918251123e-05, "loss": 0.0225, "step": 47290 }, { "epoch": 3.282096816548119e-13, "grad_norm": 0.5686840415000916, "learning_rate": 9.048749310600666e-05, "loss": 0.0241, "step": 47300 }, { "epoch": 3.2827907059385097e-13, "grad_norm": 0.5862485766410828, "learning_rate": 9.04660062086629e-05, "loss": 0.0242, "step": 47310 }, { "epoch": 3.2834845953289005e-13, "grad_norm": 0.558156430721283, "learning_rate": 9.044451849268805e-05, "loss": 0.0229, "step": 47320 }, { "epoch": 3.284178484719291e-13, "grad_norm": 0.5405188798904419, "learning_rate": 9.042302996029031e-05, "loss": 0.024, "step": 47330 }, { "epoch": 3.284872374109682e-13, "grad_norm": 0.5664898753166199, "learning_rate": 9.040154061367799e-05, "loss": 0.0241, "step": 47340 }, { "epoch": 3.2855662635000726e-13, "grad_norm": 0.5685330629348755, "learning_rate": 9.03800504550594e-05, "loss": 0.0251, "step": 47350 }, { "epoch": 3.2862601528904634e-13, "grad_norm": 0.5801804661750793, "learning_rate": 9.035855948664303e-05, "loss": 0.0255, "step": 47360 }, { "epoch": 3.286954042280854e-13, "grad_norm": 0.5550025701522827, "learning_rate": 9.033706771063743e-05, "loss": 0.0243, "step": 47370 }, { "epoch": 3.287647931671245e-13, "grad_norm": 0.6078742742538452, "learning_rate": 9.031557512925113e-05, "loss": 0.0256, "step": 47380 }, { "epoch": 3.2883418210616355e-13, "grad_norm": 0.5842233896255493, "learning_rate": 9.02940817446929e-05, "loss": 0.0248, "step": 47390 }, { "epoch": 3.289035710452026e-13, "grad_norm": 0.6193001866340637, "learning_rate": 9.02725875591715e-05, "loss": 0.0254, "step": 47400 }, { "epoch": 3.289729599842417e-13, "grad_norm": 0.6793241500854492, "learning_rate": 9.02510925748958e-05, "loss": 0.0241, "step": 47410 }, { "epoch": 3.2904234892328077e-13, "grad_norm": 0.540825366973877, "learning_rate": 9.022959679407471e-05, "loss": 0.0236, "step": 47420 }, { "epoch": 3.2911173786231984e-13, "grad_norm": 0.5357517600059509, "learning_rate": 9.020810021891731e-05, "loss": 0.0241, "step": 47430 }, { "epoch": 3.291811268013589e-13, "grad_norm": 0.5593471527099609, "learning_rate": 9.018660285163268e-05, "loss": 0.0234, "step": 47440 }, { "epoch": 3.29250515740398e-13, "grad_norm": 0.5842500925064087, "learning_rate": 9.016510469443e-05, "loss": 0.0236, "step": 47450 }, { "epoch": 3.2931990467943706e-13, "grad_norm": 0.6196332573890686, "learning_rate": 9.014360574951853e-05, "loss": 0.0238, "step": 47460 }, { "epoch": 3.2938929361847613e-13, "grad_norm": 0.5584472417831421, "learning_rate": 9.012210601910768e-05, "loss": 0.0233, "step": 47470 }, { "epoch": 3.294586825575152e-13, "grad_norm": 0.5902983546257019, "learning_rate": 9.010060550540684e-05, "loss": 0.0234, "step": 47480 }, { "epoch": 3.295280714965543e-13, "grad_norm": 0.5574569702148438, "learning_rate": 9.00791042106255e-05, "loss": 0.0238, "step": 47490 }, { "epoch": 3.2959746043559335e-13, "grad_norm": 0.6034954190254211, "learning_rate": 9.005760213697328e-05, "loss": 0.0237, "step": 47500 }, { "epoch": 3.296668493746324e-13, "grad_norm": 0.5672005414962769, "learning_rate": 9.003609928665983e-05, "loss": 0.025, "step": 47510 }, { "epoch": 3.297362383136715e-13, "grad_norm": 0.5594006180763245, "learning_rate": 9.001459566189497e-05, "loss": 0.0256, "step": 47520 }, { "epoch": 3.2980562725271056e-13, "grad_norm": 0.5814996361732483, "learning_rate": 8.999309126488846e-05, "loss": 0.0237, "step": 47530 }, { "epoch": 3.2987501619174964e-13, "grad_norm": 0.5535615682601929, "learning_rate": 8.997158609785022e-05, "loss": 0.0259, "step": 47540 }, { "epoch": 3.299444051307887e-13, "grad_norm": 0.5240840315818787, "learning_rate": 8.995008016299028e-05, "loss": 0.0254, "step": 47550 }, { "epoch": 3.300137940698278e-13, "grad_norm": 0.497249037027359, "learning_rate": 8.992857346251865e-05, "loss": 0.0244, "step": 47560 }, { "epoch": 3.3008318300886685e-13, "grad_norm": 0.5931419134140015, "learning_rate": 8.990706599864555e-05, "loss": 0.024, "step": 47570 }, { "epoch": 3.3015257194790593e-13, "grad_norm": 0.6752386689186096, "learning_rate": 8.988555777358116e-05, "loss": 0.0244, "step": 47580 }, { "epoch": 3.30221960886945e-13, "grad_norm": 0.5852775573730469, "learning_rate": 8.986404878953579e-05, "loss": 0.0238, "step": 47590 }, { "epoch": 3.3029134982598407e-13, "grad_norm": 0.5269185304641724, "learning_rate": 8.984253904871983e-05, "loss": 0.0227, "step": 47600 }, { "epoch": 3.3036073876502314e-13, "grad_norm": 0.60691899061203, "learning_rate": 8.982102855334375e-05, "loss": 0.024, "step": 47610 }, { "epoch": 3.304301277040622e-13, "grad_norm": 0.5067854523658752, "learning_rate": 8.979951730561806e-05, "loss": 0.0243, "step": 47620 }, { "epoch": 3.304995166431013e-13, "grad_norm": 0.5216547250747681, "learning_rate": 8.977800530775339e-05, "loss": 0.023, "step": 47630 }, { "epoch": 3.3056890558214036e-13, "grad_norm": 0.5072141885757446, "learning_rate": 8.975649256196044e-05, "loss": 0.0229, "step": 47640 }, { "epoch": 3.3063829452117943e-13, "grad_norm": 0.5061946511268616, "learning_rate": 8.973497907044996e-05, "loss": 0.0235, "step": 47650 }, { "epoch": 3.307076834602185e-13, "grad_norm": 0.5837818384170532, "learning_rate": 8.971346483543281e-05, "loss": 0.0238, "step": 47660 }, { "epoch": 3.307770723992576e-13, "grad_norm": 0.6351330280303955, "learning_rate": 8.969194985911994e-05, "loss": 0.0246, "step": 47670 }, { "epoch": 3.3084646133829665e-13, "grad_norm": 0.5138067007064819, "learning_rate": 8.96704341437223e-05, "loss": 0.0244, "step": 47680 }, { "epoch": 3.309158502773357e-13, "grad_norm": 0.5572799444198608, "learning_rate": 8.964891769145098e-05, "loss": 0.0242, "step": 47690 }, { "epoch": 3.309852392163748e-13, "grad_norm": 0.5474033951759338, "learning_rate": 8.962740050451717e-05, "loss": 0.0246, "step": 47700 }, { "epoch": 3.3105462815541387e-13, "grad_norm": 0.5573053956031799, "learning_rate": 8.9605882585132e-05, "loss": 0.0248, "step": 47710 }, { "epoch": 3.3112401709445294e-13, "grad_norm": 0.5032818913459778, "learning_rate": 8.958436393550688e-05, "loss": 0.0236, "step": 47720 }, { "epoch": 3.31193406033492e-13, "grad_norm": 0.5427598357200623, "learning_rate": 8.956284455785312e-05, "loss": 0.0241, "step": 47730 }, { "epoch": 3.312627949725311e-13, "grad_norm": 0.5670055747032166, "learning_rate": 8.954132445438219e-05, "loss": 0.0248, "step": 47740 }, { "epoch": 3.3133218391157016e-13, "grad_norm": 0.5169335603713989, "learning_rate": 8.95198036273056e-05, "loss": 0.0235, "step": 47750 }, { "epoch": 3.3140157285060923e-13, "grad_norm": 0.5997944474220276, "learning_rate": 8.949828207883497e-05, "loss": 0.0235, "step": 47760 }, { "epoch": 3.314709617896483e-13, "grad_norm": 0.5458863377571106, "learning_rate": 8.9476759811182e-05, "loss": 0.0239, "step": 47770 }, { "epoch": 3.3154035072868737e-13, "grad_norm": 0.5847116708755493, "learning_rate": 8.945523682655837e-05, "loss": 0.024, "step": 47780 }, { "epoch": 3.3160973966772644e-13, "grad_norm": 0.5256556272506714, "learning_rate": 8.943371312717595e-05, "loss": 0.0229, "step": 47790 }, { "epoch": 3.316791286067655e-13, "grad_norm": 0.6346076130867004, "learning_rate": 8.941218871524661e-05, "loss": 0.024, "step": 47800 }, { "epoch": 3.317485175458046e-13, "grad_norm": 0.5202842950820923, "learning_rate": 8.939066359298236e-05, "loss": 0.0226, "step": 47810 }, { "epoch": 3.3181790648484366e-13, "grad_norm": 0.5997431874275208, "learning_rate": 8.93691377625952e-05, "loss": 0.0246, "step": 47820 }, { "epoch": 3.3188729542388273e-13, "grad_norm": 0.6159689426422119, "learning_rate": 8.934761122629723e-05, "loss": 0.0251, "step": 47830 }, { "epoch": 3.319566843629218e-13, "grad_norm": 0.5960099101066589, "learning_rate": 8.932608398630068e-05, "loss": 0.0235, "step": 47840 }, { "epoch": 3.320260733019609e-13, "grad_norm": 0.6459105610847473, "learning_rate": 8.93045560448178e-05, "loss": 0.0255, "step": 47850 }, { "epoch": 3.3209546224099995e-13, "grad_norm": 0.632943332195282, "learning_rate": 8.92830274040609e-05, "loss": 0.0242, "step": 47860 }, { "epoch": 3.32164851180039e-13, "grad_norm": 0.5638312101364136, "learning_rate": 8.926149806624241e-05, "loss": 0.0241, "step": 47870 }, { "epoch": 3.322342401190781e-13, "grad_norm": 0.5675579309463501, "learning_rate": 8.923996803357479e-05, "loss": 0.0235, "step": 47880 }, { "epoch": 3.3230362905811717e-13, "grad_norm": 0.5786958336830139, "learning_rate": 8.921843730827055e-05, "loss": 0.0233, "step": 47890 }, { "epoch": 3.3237301799715624e-13, "grad_norm": 0.5796501040458679, "learning_rate": 8.919690589254238e-05, "loss": 0.0239, "step": 47900 }, { "epoch": 3.324424069361953e-13, "grad_norm": 0.5545475482940674, "learning_rate": 8.917537378860292e-05, "loss": 0.0231, "step": 47910 }, { "epoch": 3.325117958752344e-13, "grad_norm": 0.4738120436668396, "learning_rate": 8.915384099866493e-05, "loss": 0.0234, "step": 47920 }, { "epoch": 3.3258118481427346e-13, "grad_norm": 0.5672345161437988, "learning_rate": 8.913230752494126e-05, "loss": 0.024, "step": 47930 }, { "epoch": 3.3265057375331253e-13, "grad_norm": 0.5672672986984253, "learning_rate": 8.911077336964479e-05, "loss": 0.0235, "step": 47940 }, { "epoch": 3.327199626923516e-13, "grad_norm": 0.555691123008728, "learning_rate": 8.908923853498849e-05, "loss": 0.0234, "step": 47950 }, { "epoch": 3.3278935163139067e-13, "grad_norm": 0.5457942485809326, "learning_rate": 8.906770302318544e-05, "loss": 0.0232, "step": 47960 }, { "epoch": 3.3285874057042975e-13, "grad_norm": 0.587817370891571, "learning_rate": 8.904616683644867e-05, "loss": 0.0235, "step": 47970 }, { "epoch": 3.329281295094688e-13, "grad_norm": 0.6005582809448242, "learning_rate": 8.902462997699142e-05, "loss": 0.0243, "step": 47980 }, { "epoch": 3.329975184485079e-13, "grad_norm": 0.6146189570426941, "learning_rate": 8.900309244702694e-05, "loss": 0.0246, "step": 47990 }, { "epoch": 3.3306690738754696e-13, "grad_norm": 0.6045655608177185, "learning_rate": 8.898155424876851e-05, "loss": 0.0238, "step": 48000 }, { "epoch": 3.3313629632658603e-13, "grad_norm": 0.535637617111206, "learning_rate": 8.896001538442955e-05, "loss": 0.0254, "step": 48010 }, { "epoch": 3.332056852656251e-13, "grad_norm": 0.5318540334701538, "learning_rate": 8.893847585622348e-05, "loss": 0.0251, "step": 48020 }, { "epoch": 3.332750742046642e-13, "grad_norm": 0.6313303112983704, "learning_rate": 8.891693566636383e-05, "loss": 0.0237, "step": 48030 }, { "epoch": 3.3334446314370325e-13, "grad_norm": 0.610626220703125, "learning_rate": 8.889539481706422e-05, "loss": 0.0227, "step": 48040 }, { "epoch": 3.334138520827423e-13, "grad_norm": 0.5624651908874512, "learning_rate": 8.887385331053826e-05, "loss": 0.0234, "step": 48050 }, { "epoch": 3.334832410217814e-13, "grad_norm": 0.5670626759529114, "learning_rate": 8.885231114899972e-05, "loss": 0.0227, "step": 48060 }, { "epoch": 3.3355262996082047e-13, "grad_norm": 0.5719210505485535, "learning_rate": 8.883076833466236e-05, "loss": 0.0226, "step": 48070 }, { "epoch": 3.3362201889985954e-13, "grad_norm": 0.5853833556175232, "learning_rate": 8.880922486974005e-05, "loss": 0.0237, "step": 48080 }, { "epoch": 3.336914078388986e-13, "grad_norm": 0.5212323069572449, "learning_rate": 8.878768075644673e-05, "loss": 0.0237, "step": 48090 }, { "epoch": 3.337607967779377e-13, "grad_norm": 0.5439566969871521, "learning_rate": 8.876613599699636e-05, "loss": 0.0229, "step": 48100 }, { "epoch": 3.3383018571697676e-13, "grad_norm": 0.5710490345954895, "learning_rate": 8.874459059360302e-05, "loss": 0.0227, "step": 48110 }, { "epoch": 3.3389957465601583e-13, "grad_norm": 0.5326021909713745, "learning_rate": 8.872304454848085e-05, "loss": 0.023, "step": 48120 }, { "epoch": 3.339689635950549e-13, "grad_norm": 0.5538378357887268, "learning_rate": 8.870149786384403e-05, "loss": 0.0232, "step": 48130 }, { "epoch": 3.3403835253409397e-13, "grad_norm": 0.5479506850242615, "learning_rate": 8.867995054190682e-05, "loss": 0.0243, "step": 48140 }, { "epoch": 3.3410774147313305e-13, "grad_norm": 0.5380809307098389, "learning_rate": 8.865840258488352e-05, "loss": 0.0238, "step": 48150 }, { "epoch": 3.341771304121721e-13, "grad_norm": 0.6195347905158997, "learning_rate": 8.863685399498853e-05, "loss": 0.0246, "step": 48160 }, { "epoch": 3.342465193512112e-13, "grad_norm": 0.5627621412277222, "learning_rate": 8.861530477443631e-05, "loss": 0.0242, "step": 48170 }, { "epoch": 3.3431590829025026e-13, "grad_norm": 0.48808157444000244, "learning_rate": 8.859375492544138e-05, "loss": 0.025, "step": 48180 }, { "epoch": 3.3438529722928934e-13, "grad_norm": 0.4971482753753662, "learning_rate": 8.857220445021832e-05, "loss": 0.023, "step": 48190 }, { "epoch": 3.344546861683284e-13, "grad_norm": 0.5961081385612488, "learning_rate": 8.855065335098177e-05, "loss": 0.0231, "step": 48200 }, { "epoch": 3.345240751073675e-13, "grad_norm": 0.48598387837409973, "learning_rate": 8.852910162994643e-05, "loss": 0.0228, "step": 48210 }, { "epoch": 3.3459346404640655e-13, "grad_norm": 0.513951301574707, "learning_rate": 8.850754928932709e-05, "loss": 0.0223, "step": 48220 }, { "epoch": 3.346628529854456e-13, "grad_norm": 0.5627040863037109, "learning_rate": 8.84859963313386e-05, "loss": 0.0224, "step": 48230 }, { "epoch": 3.347322419244847e-13, "grad_norm": 0.5894806385040283, "learning_rate": 8.846444275819586e-05, "loss": 0.0232, "step": 48240 }, { "epoch": 3.3480163086352377e-13, "grad_norm": 0.5502176880836487, "learning_rate": 8.844288857211382e-05, "loss": 0.0233, "step": 48250 }, { "epoch": 3.3487101980256284e-13, "grad_norm": 0.5847805738449097, "learning_rate": 8.84213337753075e-05, "loss": 0.0224, "step": 48260 }, { "epoch": 3.349404087416019e-13, "grad_norm": 0.5742717385292053, "learning_rate": 8.8399778369992e-05, "loss": 0.0232, "step": 48270 }, { "epoch": 3.35009797680641e-13, "grad_norm": 0.5772313475608826, "learning_rate": 8.837822235838248e-05, "loss": 0.0227, "step": 48280 }, { "epoch": 3.3507918661968006e-13, "grad_norm": 0.5277940034866333, "learning_rate": 8.835666574269415e-05, "loss": 0.0235, "step": 48290 }, { "epoch": 3.3514857555871913e-13, "grad_norm": 0.5506172776222229, "learning_rate": 8.833510852514228e-05, "loss": 0.0244, "step": 48300 }, { "epoch": 3.352179644977582e-13, "grad_norm": 0.5506837368011475, "learning_rate": 8.831355070794223e-05, "loss": 0.0229, "step": 48310 }, { "epoch": 3.352873534367973e-13, "grad_norm": 0.5730078816413879, "learning_rate": 8.829199229330938e-05, "loss": 0.0251, "step": 48320 }, { "epoch": 3.3535674237583635e-13, "grad_norm": 0.6235454678535461, "learning_rate": 8.827043328345922e-05, "loss": 0.025, "step": 48330 }, { "epoch": 3.354261313148754e-13, "grad_norm": 0.5804219841957092, "learning_rate": 8.824887368060726e-05, "loss": 0.0238, "step": 48340 }, { "epoch": 3.354955202539145e-13, "grad_norm": 0.5371835827827454, "learning_rate": 8.822731348696907e-05, "loss": 0.0226, "step": 48350 }, { "epoch": 3.3556490919295356e-13, "grad_norm": 0.6014106869697571, "learning_rate": 8.820575270476031e-05, "loss": 0.0227, "step": 48360 }, { "epoch": 3.3563429813199264e-13, "grad_norm": 0.5652573704719543, "learning_rate": 8.818419133619669e-05, "loss": 0.0225, "step": 48370 }, { "epoch": 3.357036870710317e-13, "grad_norm": 0.5389593243598938, "learning_rate": 8.816262938349394e-05, "loss": 0.0221, "step": 48380 }, { "epoch": 3.357730760100708e-13, "grad_norm": 0.5459481477737427, "learning_rate": 8.814106684886792e-05, "loss": 0.0223, "step": 48390 }, { "epoch": 3.3584246494910985e-13, "grad_norm": 0.6725991368293762, "learning_rate": 8.811950373453453e-05, "loss": 0.0243, "step": 48400 }, { "epoch": 3.3591185388814893e-13, "grad_norm": 0.580207347869873, "learning_rate": 8.809794004270967e-05, "loss": 0.0222, "step": 48410 }, { "epoch": 3.35981242827188e-13, "grad_norm": 0.4747016429901123, "learning_rate": 8.80763757756094e-05, "loss": 0.0223, "step": 48420 }, { "epoch": 3.3605063176622707e-13, "grad_norm": 0.5684108138084412, "learning_rate": 8.805481093544974e-05, "loss": 0.0235, "step": 48430 }, { "epoch": 3.3612002070526614e-13, "grad_norm": 0.5055317878723145, "learning_rate": 8.803324552444682e-05, "loss": 0.0227, "step": 48440 }, { "epoch": 3.361894096443052e-13, "grad_norm": 0.5334353446960449, "learning_rate": 8.801167954481687e-05, "loss": 0.024, "step": 48450 }, { "epoch": 3.362587985833443e-13, "grad_norm": 0.5582849383354187, "learning_rate": 8.799011299877603e-05, "loss": 0.0239, "step": 48460 }, { "epoch": 3.3632818752238336e-13, "grad_norm": 0.6164410710334778, "learning_rate": 8.79685458885407e-05, "loss": 0.0233, "step": 48470 }, { "epoch": 3.3639757646142243e-13, "grad_norm": 0.6195483207702637, "learning_rate": 8.794697821632717e-05, "loss": 0.0247, "step": 48480 }, { "epoch": 3.364669654004615e-13, "grad_norm": 0.5563749670982361, "learning_rate": 8.792540998435187e-05, "loss": 0.0257, "step": 48490 }, { "epoch": 3.365363543395006e-13, "grad_norm": 0.5804353356361389, "learning_rate": 8.79038411948313e-05, "loss": 0.0238, "step": 48500 }, { "epoch": 3.3660574327853965e-13, "grad_norm": 0.6064050197601318, "learning_rate": 8.788227184998196e-05, "loss": 0.023, "step": 48510 }, { "epoch": 3.366751322175787e-13, "grad_norm": 0.5686772465705872, "learning_rate": 8.786070195202045e-05, "loss": 0.0228, "step": 48520 }, { "epoch": 3.367445211566178e-13, "grad_norm": 0.5476861000061035, "learning_rate": 8.783913150316339e-05, "loss": 0.0223, "step": 48530 }, { "epoch": 3.3681391009565687e-13, "grad_norm": 0.5927801132202148, "learning_rate": 8.781756050562751e-05, "loss": 0.0218, "step": 48540 }, { "epoch": 3.3688329903469594e-13, "grad_norm": 0.5642041563987732, "learning_rate": 8.779598896162954e-05, "loss": 0.0225, "step": 48550 }, { "epoch": 3.36952687973735e-13, "grad_norm": 0.6215547323226929, "learning_rate": 8.777441687338629e-05, "loss": 0.0232, "step": 48560 }, { "epoch": 3.370220769127741e-13, "grad_norm": 0.5629910826683044, "learning_rate": 8.775284424311466e-05, "loss": 0.0226, "step": 48570 }, { "epoch": 3.3709146585181315e-13, "grad_norm": 0.5472050309181213, "learning_rate": 8.773127107303155e-05, "loss": 0.0225, "step": 48580 }, { "epoch": 3.3716085479085223e-13, "grad_norm": 0.5743930339813232, "learning_rate": 8.770969736535395e-05, "loss": 0.0231, "step": 48590 }, { "epoch": 3.372302437298913e-13, "grad_norm": 0.5249804854393005, "learning_rate": 8.768812312229887e-05, "loss": 0.0235, "step": 48600 }, { "epoch": 3.3729963266893037e-13, "grad_norm": 0.5491690635681152, "learning_rate": 8.766654834608342e-05, "loss": 0.0237, "step": 48610 }, { "epoch": 3.3736902160796944e-13, "grad_norm": 0.5413094162940979, "learning_rate": 8.764497303892475e-05, "loss": 0.0228, "step": 48620 }, { "epoch": 3.374384105470085e-13, "grad_norm": 0.6076018214225769, "learning_rate": 8.762339720304003e-05, "loss": 0.0241, "step": 48630 }, { "epoch": 3.375077994860476e-13, "grad_norm": 0.6422993540763855, "learning_rate": 8.760182084064654e-05, "loss": 0.0241, "step": 48640 }, { "epoch": 3.3757718842508666e-13, "grad_norm": 0.5652147531509399, "learning_rate": 8.758024395396159e-05, "loss": 0.0247, "step": 48650 }, { "epoch": 3.3764657736412573e-13, "grad_norm": 0.6309148073196411, "learning_rate": 8.755866654520251e-05, "loss": 0.0235, "step": 48660 }, { "epoch": 3.377159663031648e-13, "grad_norm": 0.51901775598526, "learning_rate": 8.753708861658672e-05, "loss": 0.0225, "step": 48670 }, { "epoch": 3.377853552422039e-13, "grad_norm": 0.4743160903453827, "learning_rate": 8.751551017033174e-05, "loss": 0.0224, "step": 48680 }, { "epoch": 3.3785474418124295e-13, "grad_norm": 0.5203929543495178, "learning_rate": 8.749393120865504e-05, "loss": 0.0221, "step": 48690 }, { "epoch": 3.37924133120282e-13, "grad_norm": 0.5343957543373108, "learning_rate": 8.74723517337742e-05, "loss": 0.0218, "step": 48700 }, { "epoch": 3.379935220593211e-13, "grad_norm": 0.5618872046470642, "learning_rate": 8.745077174790686e-05, "loss": 0.0226, "step": 48710 }, { "epoch": 3.3806291099836017e-13, "grad_norm": 0.565697431564331, "learning_rate": 8.742919125327067e-05, "loss": 0.0224, "step": 48720 }, { "epoch": 3.3813229993739924e-13, "grad_norm": 0.584374725818634, "learning_rate": 8.74076102520834e-05, "loss": 0.023, "step": 48730 }, { "epoch": 3.382016888764383e-13, "grad_norm": 0.5626105070114136, "learning_rate": 8.738602874656281e-05, "loss": 0.0232, "step": 48740 }, { "epoch": 3.382710778154774e-13, "grad_norm": 0.5682989358901978, "learning_rate": 8.736444673892676e-05, "loss": 0.0225, "step": 48750 }, { "epoch": 3.3834046675451646e-13, "grad_norm": 0.5551961064338684, "learning_rate": 8.734286423139309e-05, "loss": 0.023, "step": 48760 }, { "epoch": 3.3840985569355553e-13, "grad_norm": 0.5235211849212646, "learning_rate": 8.732128122617978e-05, "loss": 0.0237, "step": 48770 }, { "epoch": 3.384792446325946e-13, "grad_norm": 0.5461137294769287, "learning_rate": 8.729969772550483e-05, "loss": 0.0223, "step": 48780 }, { "epoch": 3.3854863357163367e-13, "grad_norm": 0.542151927947998, "learning_rate": 8.727811373158626e-05, "loss": 0.0238, "step": 48790 }, { "epoch": 3.3861802251067274e-13, "grad_norm": 0.5758242607116699, "learning_rate": 8.725652924664212e-05, "loss": 0.0244, "step": 48800 }, { "epoch": 3.386874114497118e-13, "grad_norm": 0.5083637833595276, "learning_rate": 8.723494427289063e-05, "loss": 0.023, "step": 48810 }, { "epoch": 3.387568003887509e-13, "grad_norm": 0.5781888961791992, "learning_rate": 8.721335881254994e-05, "loss": 0.0233, "step": 48820 }, { "epoch": 3.3882618932778996e-13, "grad_norm": 0.6342253088951111, "learning_rate": 8.719177286783826e-05, "loss": 0.023, "step": 48830 }, { "epoch": 3.3889557826682903e-13, "grad_norm": 0.5373969078063965, "learning_rate": 8.717018644097395e-05, "loss": 0.0219, "step": 48840 }, { "epoch": 3.389649672058681e-13, "grad_norm": 0.5044919848442078, "learning_rate": 8.714859953417532e-05, "loss": 0.0213, "step": 48850 }, { "epoch": 3.390343561449072e-13, "grad_norm": 0.5662953853607178, "learning_rate": 8.712701214966073e-05, "loss": 0.0214, "step": 48860 }, { "epoch": 3.3910374508394625e-13, "grad_norm": 0.49889227747917175, "learning_rate": 8.710542428964868e-05, "loss": 0.0229, "step": 48870 }, { "epoch": 3.391731340229853e-13, "grad_norm": 0.49986758828163147, "learning_rate": 8.708383595635762e-05, "loss": 0.0218, "step": 48880 }, { "epoch": 3.392425229620244e-13, "grad_norm": 0.4925785958766937, "learning_rate": 8.706224715200608e-05, "loss": 0.0221, "step": 48890 }, { "epoch": 3.3931191190106347e-13, "grad_norm": 0.5063982605934143, "learning_rate": 8.704065787881267e-05, "loss": 0.0229, "step": 48900 }, { "epoch": 3.3938130084010254e-13, "grad_norm": 0.5651050209999084, "learning_rate": 8.701906813899601e-05, "loss": 0.023, "step": 48910 }, { "epoch": 3.394506897791416e-13, "grad_norm": 0.6093182563781738, "learning_rate": 8.699747793477479e-05, "loss": 0.0236, "step": 48920 }, { "epoch": 3.395200787181807e-13, "grad_norm": 0.5102638006210327, "learning_rate": 8.697588726836773e-05, "loss": 0.0227, "step": 48930 }, { "epoch": 3.3958946765721976e-13, "grad_norm": 0.5593448281288147, "learning_rate": 8.695429614199361e-05, "loss": 0.0226, "step": 48940 }, { "epoch": 3.3965885659625883e-13, "grad_norm": 0.5223065614700317, "learning_rate": 8.693270455787125e-05, "loss": 0.0229, "step": 48950 }, { "epoch": 3.397282455352979e-13, "grad_norm": 0.5538185238838196, "learning_rate": 8.691111251821956e-05, "loss": 0.0243, "step": 48960 }, { "epoch": 3.3979763447433697e-13, "grad_norm": 0.5383997559547424, "learning_rate": 8.68895200252574e-05, "loss": 0.0226, "step": 48970 }, { "epoch": 3.3986702341337605e-13, "grad_norm": 0.5348732471466064, "learning_rate": 8.686792708120375e-05, "loss": 0.0226, "step": 48980 }, { "epoch": 3.399364123524151e-13, "grad_norm": 0.5363768339157104, "learning_rate": 8.684633368827764e-05, "loss": 0.023, "step": 48990 }, { "epoch": 3.400058012914542e-13, "grad_norm": 0.4846382141113281, "learning_rate": 8.682473984869813e-05, "loss": 0.0219, "step": 49000 }, { "epoch": 3.4007519023049326e-13, "grad_norm": 0.5649611949920654, "learning_rate": 8.68031455646843e-05, "loss": 0.0214, "step": 49010 }, { "epoch": 3.4014457916953234e-13, "grad_norm": 0.5282170176506042, "learning_rate": 8.67815508384553e-05, "loss": 0.0217, "step": 49020 }, { "epoch": 3.402139681085714e-13, "grad_norm": 0.5273364186286926, "learning_rate": 8.675995567223036e-05, "loss": 0.022, "step": 49030 }, { "epoch": 3.402833570476105e-13, "grad_norm": 0.47757819294929504, "learning_rate": 8.673836006822865e-05, "loss": 0.0213, "step": 49040 }, { "epoch": 3.4035274598664955e-13, "grad_norm": 0.5810680985450745, "learning_rate": 8.671676402866954e-05, "loss": 0.0231, "step": 49050 }, { "epoch": 3.404221349256886e-13, "grad_norm": 0.5362492203712463, "learning_rate": 8.669516755577227e-05, "loss": 0.0223, "step": 49060 }, { "epoch": 3.404915238647277e-13, "grad_norm": 0.5599052309989929, "learning_rate": 8.667357065175629e-05, "loss": 0.0229, "step": 49070 }, { "epoch": 3.4056091280376677e-13, "grad_norm": 0.5900490283966064, "learning_rate": 8.665197331884099e-05, "loss": 0.0243, "step": 49080 }, { "epoch": 3.4063030174280584e-13, "grad_norm": 0.5621170997619629, "learning_rate": 8.663037555924581e-05, "loss": 0.0221, "step": 49090 }, { "epoch": 3.406996906818449e-13, "grad_norm": 0.630993127822876, "learning_rate": 8.66087773751903e-05, "loss": 0.0239, "step": 49100 }, { "epoch": 3.40769079620884e-13, "grad_norm": 0.6104215979576111, "learning_rate": 8.658717876889398e-05, "loss": 0.0226, "step": 49110 }, { "epoch": 3.4083846855992306e-13, "grad_norm": 0.5678355693817139, "learning_rate": 8.656557974257645e-05, "loss": 0.0229, "step": 49120 }, { "epoch": 3.4090785749896213e-13, "grad_norm": 0.5488075017929077, "learning_rate": 8.654398029845733e-05, "loss": 0.0226, "step": 49130 }, { "epoch": 3.409772464380012e-13, "grad_norm": 0.5841380953788757, "learning_rate": 8.652238043875631e-05, "loss": 0.0217, "step": 49140 }, { "epoch": 3.410466353770403e-13, "grad_norm": 0.5444135069847107, "learning_rate": 8.650078016569315e-05, "loss": 0.0226, "step": 49150 }, { "epoch": 3.4111602431607935e-13, "grad_norm": 0.4990423023700714, "learning_rate": 8.647917948148753e-05, "loss": 0.0223, "step": 49160 }, { "epoch": 3.411854132551184e-13, "grad_norm": 0.4702761471271515, "learning_rate": 8.645757838835933e-05, "loss": 0.0217, "step": 49170 }, { "epoch": 3.412548021941575e-13, "grad_norm": 0.5328130722045898, "learning_rate": 8.643597688852837e-05, "loss": 0.0222, "step": 49180 }, { "epoch": 3.4132419113319656e-13, "grad_norm": 0.5599326491355896, "learning_rate": 8.641437498421453e-05, "loss": 0.0213, "step": 49190 }, { "epoch": 3.4139358007223564e-13, "grad_norm": 0.5621533989906311, "learning_rate": 8.639277267763777e-05, "loss": 0.0223, "step": 49200 }, { "epoch": 3.414629690112747e-13, "grad_norm": 0.5221828818321228, "learning_rate": 8.637116997101804e-05, "loss": 0.0219, "step": 49210 }, { "epoch": 3.415323579503138e-13, "grad_norm": 0.5403250455856323, "learning_rate": 8.63495668665754e-05, "loss": 0.0224, "step": 49220 }, { "epoch": 3.4160174688935285e-13, "grad_norm": 0.6035546064376831, "learning_rate": 8.632796336652982e-05, "loss": 0.023, "step": 49230 }, { "epoch": 3.4167113582839193e-13, "grad_norm": 0.5797938108444214, "learning_rate": 8.630635947310147e-05, "loss": 0.0233, "step": 49240 }, { "epoch": 3.41740524767431e-13, "grad_norm": 0.5772759318351746, "learning_rate": 8.628475518851043e-05, "loss": 0.0218, "step": 49250 }, { "epoch": 3.4180991370647007e-13, "grad_norm": 0.5182526111602783, "learning_rate": 8.626315051497693e-05, "loss": 0.0231, "step": 49260 }, { "epoch": 3.4187930264550914e-13, "grad_norm": 0.5326244831085205, "learning_rate": 8.624154545472116e-05, "loss": 0.0237, "step": 49270 }, { "epoch": 3.419486915845482e-13, "grad_norm": 0.5834130048751831, "learning_rate": 8.621994000996338e-05, "loss": 0.0223, "step": 49280 }, { "epoch": 3.420180805235873e-13, "grad_norm": 0.6038919687271118, "learning_rate": 8.619833418292389e-05, "loss": 0.0227, "step": 49290 }, { "epoch": 3.4208746946262636e-13, "grad_norm": 0.5880574584007263, "learning_rate": 8.617672797582301e-05, "loss": 0.0225, "step": 49300 }, { "epoch": 3.4215685840166543e-13, "grad_norm": 0.5340772271156311, "learning_rate": 8.615512139088113e-05, "loss": 0.0217, "step": 49310 }, { "epoch": 3.422262473407045e-13, "grad_norm": 0.5244141817092896, "learning_rate": 8.613351443031864e-05, "loss": 0.0216, "step": 49320 }, { "epoch": 3.422956362797436e-13, "grad_norm": 0.5820050239562988, "learning_rate": 8.611190709635604e-05, "loss": 0.022, "step": 49330 }, { "epoch": 3.4236502521878265e-13, "grad_norm": 0.4929552376270294, "learning_rate": 8.609029939121376e-05, "loss": 0.0219, "step": 49340 }, { "epoch": 3.424344141578217e-13, "grad_norm": 0.5477803349494934, "learning_rate": 8.606869131711236e-05, "loss": 0.0211, "step": 49350 }, { "epoch": 3.425038030968608e-13, "grad_norm": 0.5707849860191345, "learning_rate": 8.60470828762724e-05, "loss": 0.0215, "step": 49360 }, { "epoch": 3.4257319203589987e-13, "grad_norm": 0.5603575110435486, "learning_rate": 8.602547407091447e-05, "loss": 0.0217, "step": 49370 }, { "epoch": 3.4264258097493894e-13, "grad_norm": 0.594012975692749, "learning_rate": 8.600386490325927e-05, "loss": 0.0225, "step": 49380 }, { "epoch": 3.42711969913978e-13, "grad_norm": 0.5598976016044617, "learning_rate": 8.598225537552737e-05, "loss": 0.0235, "step": 49390 }, { "epoch": 3.427813588530171e-13, "grad_norm": 0.5337756872177124, "learning_rate": 8.596064548993959e-05, "loss": 0.0219, "step": 49400 }, { "epoch": 3.4285074779205615e-13, "grad_norm": 0.5743016004562378, "learning_rate": 8.593903524871663e-05, "loss": 0.0225, "step": 49410 }, { "epoch": 3.4292013673109523e-13, "grad_norm": 0.6383458971977234, "learning_rate": 8.59174246540793e-05, "loss": 0.0229, "step": 49420 }, { "epoch": 3.429895256701343e-13, "grad_norm": 0.49604079127311707, "learning_rate": 8.589581370824841e-05, "loss": 0.0231, "step": 49430 }, { "epoch": 3.4305891460917337e-13, "grad_norm": 0.4905713200569153, "learning_rate": 8.587420241344484e-05, "loss": 0.0216, "step": 49440 }, { "epoch": 3.4312830354821244e-13, "grad_norm": 0.5852630734443665, "learning_rate": 8.585259077188945e-05, "loss": 0.0225, "step": 49450 }, { "epoch": 3.431976924872515e-13, "grad_norm": 0.4970642328262329, "learning_rate": 8.58309787858032e-05, "loss": 0.0226, "step": 49460 }, { "epoch": 3.432670814262906e-13, "grad_norm": 0.49491336941719055, "learning_rate": 8.580936645740705e-05, "loss": 0.0213, "step": 49470 }, { "epoch": 3.4333647036532966e-13, "grad_norm": 0.5423033237457275, "learning_rate": 8.5787753788922e-05, "loss": 0.022, "step": 49480 }, { "epoch": 3.4340585930436873e-13, "grad_norm": 0.5859965682029724, "learning_rate": 8.576614078256908e-05, "loss": 0.0221, "step": 49490 }, { "epoch": 3.434752482434078e-13, "grad_norm": 0.5346181988716125, "learning_rate": 8.574452744056941e-05, "loss": 0.0212, "step": 49500 }, { "epoch": 3.435446371824469e-13, "grad_norm": 0.5632449388504028, "learning_rate": 8.572291376514404e-05, "loss": 0.0212, "step": 49510 }, { "epoch": 3.4361402612148595e-13, "grad_norm": 0.5601570010185242, "learning_rate": 8.570129975851415e-05, "loss": 0.0224, "step": 49520 }, { "epoch": 3.43683415060525e-13, "grad_norm": 0.5761575698852539, "learning_rate": 8.567968542290087e-05, "loss": 0.0219, "step": 49530 }, { "epoch": 3.437528039995641e-13, "grad_norm": 0.5380463600158691, "learning_rate": 8.565807076052546e-05, "loss": 0.0228, "step": 49540 }, { "epoch": 3.4382219293860317e-13, "grad_norm": 0.5395721793174744, "learning_rate": 8.563645577360917e-05, "loss": 0.0235, "step": 49550 }, { "epoch": 3.4389158187764224e-13, "grad_norm": 0.5200127363204956, "learning_rate": 8.561484046437321e-05, "loss": 0.0214, "step": 49560 }, { "epoch": 3.439609708166813e-13, "grad_norm": 0.5568419098854065, "learning_rate": 8.559322483503892e-05, "loss": 0.0229, "step": 49570 }, { "epoch": 3.440303597557204e-13, "grad_norm": 0.5782740712165833, "learning_rate": 8.557160888782767e-05, "loss": 0.023, "step": 49580 }, { "epoch": 3.4409974869475946e-13, "grad_norm": 0.5516549944877625, "learning_rate": 8.55499926249608e-05, "loss": 0.0218, "step": 49590 }, { "epoch": 3.4416913763379853e-13, "grad_norm": 0.538648784160614, "learning_rate": 8.552837604865977e-05, "loss": 0.021, "step": 49600 }, { "epoch": 3.442385265728376e-13, "grad_norm": 0.613886296749115, "learning_rate": 8.550675916114594e-05, "loss": 0.0219, "step": 49610 }, { "epoch": 3.4430791551187667e-13, "grad_norm": 0.5397483110427856, "learning_rate": 8.548514196464083e-05, "loss": 0.0216, "step": 49620 }, { "epoch": 3.4437730445091574e-13, "grad_norm": 0.5362643599510193, "learning_rate": 8.546352446136594e-05, "loss": 0.0217, "step": 49630 }, { "epoch": 3.444466933899548e-13, "grad_norm": 0.5065056085586548, "learning_rate": 8.544190665354281e-05, "loss": 0.0212, "step": 49640 }, { "epoch": 3.445160823289939e-13, "grad_norm": 0.5998517274856567, "learning_rate": 8.5420288543393e-05, "loss": 0.0232, "step": 49650 }, { "epoch": 3.4458547126803296e-13, "grad_norm": 0.5608064532279968, "learning_rate": 8.539867013313809e-05, "loss": 0.0209, "step": 49660 }, { "epoch": 3.4465486020707203e-13, "grad_norm": 0.48158106207847595, "learning_rate": 8.537705142499975e-05, "loss": 0.0216, "step": 49670 }, { "epoch": 3.447242491461111e-13, "grad_norm": 0.5246418118476868, "learning_rate": 8.535543242119958e-05, "loss": 0.0217, "step": 49680 }, { "epoch": 3.447936380851502e-13, "grad_norm": 0.5259914398193359, "learning_rate": 8.533381312395931e-05, "loss": 0.0221, "step": 49690 }, { "epoch": 3.4486302702418925e-13, "grad_norm": 0.5067688226699829, "learning_rate": 8.531219353550064e-05, "loss": 0.0221, "step": 49700 }, { "epoch": 3.449324159632283e-13, "grad_norm": 0.5353577136993408, "learning_rate": 8.529057365804535e-05, "loss": 0.023, "step": 49710 }, { "epoch": 3.450018049022674e-13, "grad_norm": 0.5721082091331482, "learning_rate": 8.526895349381518e-05, "loss": 0.0213, "step": 49720 }, { "epoch": 3.4507119384130647e-13, "grad_norm": 0.564868152141571, "learning_rate": 8.524733304503195e-05, "loss": 0.0219, "step": 49730 }, { "epoch": 3.4514058278034554e-13, "grad_norm": 0.5432714223861694, "learning_rate": 8.522571231391751e-05, "loss": 0.0227, "step": 49740 }, { "epoch": 3.452099717193846e-13, "grad_norm": 0.544259786605835, "learning_rate": 8.520409130269375e-05, "loss": 0.0211, "step": 49750 }, { "epoch": 3.452793606584237e-13, "grad_norm": 0.5574213862419128, "learning_rate": 8.518247001358251e-05, "loss": 0.0215, "step": 49760 }, { "epoch": 3.4534874959746276e-13, "grad_norm": 0.5926068425178528, "learning_rate": 8.516084844880574e-05, "loss": 0.0216, "step": 49770 }, { "epoch": 3.4541813853650183e-13, "grad_norm": 0.5532689094543457, "learning_rate": 8.513922661058538e-05, "loss": 0.0211, "step": 49780 }, { "epoch": 3.454875274755409e-13, "grad_norm": 0.5577003359794617, "learning_rate": 8.511760450114343e-05, "loss": 0.0206, "step": 49790 }, { "epoch": 3.4555691641457997e-13, "grad_norm": 0.5513697862625122, "learning_rate": 8.509598212270187e-05, "loss": 0.0221, "step": 49800 }, { "epoch": 3.4562630535361905e-13, "grad_norm": 0.5600746870040894, "learning_rate": 8.50743594774828e-05, "loss": 0.0217, "step": 49810 }, { "epoch": 3.456956942926581e-13, "grad_norm": 0.5720741152763367, "learning_rate": 8.50527365677082e-05, "loss": 0.021, "step": 49820 }, { "epoch": 3.457650832316972e-13, "grad_norm": 0.497000515460968, "learning_rate": 8.503111339560021e-05, "loss": 0.0208, "step": 49830 }, { "epoch": 3.4583447217073626e-13, "grad_norm": 0.5123056173324585, "learning_rate": 8.500948996338093e-05, "loss": 0.0214, "step": 49840 }, { "epoch": 3.4590386110977533e-13, "grad_norm": 0.5106721520423889, "learning_rate": 8.498786627327252e-05, "loss": 0.0221, "step": 49850 }, { "epoch": 3.459732500488144e-13, "grad_norm": 0.5427759885787964, "learning_rate": 8.496624232749715e-05, "loss": 0.0226, "step": 49860 }, { "epoch": 3.460426389878535e-13, "grad_norm": 0.5215252637863159, "learning_rate": 8.4944618128277e-05, "loss": 0.0218, "step": 49870 }, { "epoch": 3.4611202792689255e-13, "grad_norm": 0.5695480704307556, "learning_rate": 8.492299367783431e-05, "loss": 0.0225, "step": 49880 }, { "epoch": 3.461814168659316e-13, "grad_norm": 0.5531876683235168, "learning_rate": 8.490136897839132e-05, "loss": 0.0221, "step": 49890 }, { "epoch": 3.462508058049707e-13, "grad_norm": 0.5477227568626404, "learning_rate": 8.487974403217028e-05, "loss": 0.0222, "step": 49900 }, { "epoch": 3.4632019474400977e-13, "grad_norm": 0.6073189377784729, "learning_rate": 8.485811884139352e-05, "loss": 0.021, "step": 49910 }, { "epoch": 3.4638958368304884e-13, "grad_norm": 0.48903417587280273, "learning_rate": 8.483649340828336e-05, "loss": 0.0207, "step": 49920 }, { "epoch": 3.464589726220879e-13, "grad_norm": 0.47742021083831787, "learning_rate": 8.481486773506214e-05, "loss": 0.0216, "step": 49930 }, { "epoch": 3.46528361561127e-13, "grad_norm": 0.61330646276474, "learning_rate": 8.479324182395226e-05, "loss": 0.0216, "step": 49940 }, { "epoch": 3.4659775050016606e-13, "grad_norm": 0.5275806784629822, "learning_rate": 8.47716156771761e-05, "loss": 0.0212, "step": 49950 }, { "epoch": 3.4666713943920513e-13, "grad_norm": 0.5530470609664917, "learning_rate": 8.474998929695607e-05, "loss": 0.0218, "step": 49960 }, { "epoch": 3.467365283782442e-13, "grad_norm": 0.5516512393951416, "learning_rate": 8.472836268551464e-05, "loss": 0.0213, "step": 49970 }, { "epoch": 3.4680591731728327e-13, "grad_norm": 0.5607737898826599, "learning_rate": 8.470673584507428e-05, "loss": 0.021, "step": 49980 }, { "epoch": 3.4687530625632235e-13, "grad_norm": 0.5168374180793762, "learning_rate": 8.468510877785747e-05, "loss": 0.0209, "step": 49990 }, { "epoch": 3.469446951953614e-13, "grad_norm": 0.5589137077331543, "learning_rate": 8.466348148608673e-05, "loss": 0.0213, "step": 50000 }, { "epoch": 3.470140841344005e-13, "grad_norm": 0.5404887199401855, "learning_rate": 8.46418539719846e-05, "loss": 0.0224, "step": 50010 }, { "epoch": 3.4708347307343956e-13, "grad_norm": 0.5358009338378906, "learning_rate": 8.462022623777365e-05, "loss": 0.0226, "step": 50020 }, { "epoch": 3.4715286201247864e-13, "grad_norm": 0.5699650049209595, "learning_rate": 8.459859828567647e-05, "loss": 0.021, "step": 50030 }, { "epoch": 3.472222509515177e-13, "grad_norm": 0.5207070112228394, "learning_rate": 8.457697011791567e-05, "loss": 0.0224, "step": 50040 }, { "epoch": 3.472916398905568e-13, "grad_norm": 0.503705620765686, "learning_rate": 8.455534173671387e-05, "loss": 0.0228, "step": 50050 }, { "epoch": 3.4736102882959585e-13, "grad_norm": 0.49905672669410706, "learning_rate": 8.453371314429373e-05, "loss": 0.0212, "step": 50060 }, { "epoch": 3.474304177686349e-13, "grad_norm": 0.5533127188682556, "learning_rate": 8.451208434287792e-05, "loss": 0.0206, "step": 50070 }, { "epoch": 3.47499806707674e-13, "grad_norm": 0.6028702855110168, "learning_rate": 8.449045533468913e-05, "loss": 0.0214, "step": 50080 }, { "epoch": 3.4756919564671307e-13, "grad_norm": 0.505291759967804, "learning_rate": 8.44688261219501e-05, "loss": 0.0212, "step": 50090 }, { "epoch": 3.4763858458575214e-13, "grad_norm": 0.506644606590271, "learning_rate": 8.444719670688353e-05, "loss": 0.022, "step": 50100 }, { "epoch": 3.477079735247912e-13, "grad_norm": 0.5590987801551819, "learning_rate": 8.442556709171223e-05, "loss": 0.0212, "step": 50110 }, { "epoch": 3.477773624638303e-13, "grad_norm": 0.5024212002754211, "learning_rate": 8.440393727865893e-05, "loss": 0.0239, "step": 50120 }, { "epoch": 3.4784675140286936e-13, "grad_norm": 0.5187482833862305, "learning_rate": 8.438230726994645e-05, "loss": 0.0206, "step": 50130 }, { "epoch": 3.4791614034190843e-13, "grad_norm": 0.4822043776512146, "learning_rate": 8.43606770677976e-05, "loss": 0.0207, "step": 50140 }, { "epoch": 3.479855292809475e-13, "grad_norm": 0.5129780173301697, "learning_rate": 8.433904667443525e-05, "loss": 0.0215, "step": 50150 }, { "epoch": 3.480549182199866e-13, "grad_norm": 0.5413611531257629, "learning_rate": 8.431741609208223e-05, "loss": 0.0214, "step": 50160 }, { "epoch": 3.4812430715902565e-13, "grad_norm": 0.596983790397644, "learning_rate": 8.429578532296142e-05, "loss": 0.0222, "step": 50170 }, { "epoch": 3.481936960980647e-13, "grad_norm": 0.472932904958725, "learning_rate": 8.427415436929575e-05, "loss": 0.0218, "step": 50180 }, { "epoch": 3.482630850371038e-13, "grad_norm": 0.5257861018180847, "learning_rate": 8.425252323330811e-05, "loss": 0.0218, "step": 50190 }, { "epoch": 3.4833247397614286e-13, "grad_norm": 0.5564894080162048, "learning_rate": 8.423089191722143e-05, "loss": 0.0219, "step": 50200 }, { "epoch": 3.4840186291518194e-13, "grad_norm": 0.5233202576637268, "learning_rate": 8.42092604232587e-05, "loss": 0.0231, "step": 50210 }, { "epoch": 3.48471251854221e-13, "grad_norm": 0.4724739193916321, "learning_rate": 8.418762875364285e-05, "loss": 0.0208, "step": 50220 }, { "epoch": 3.485406407932601e-13, "grad_norm": 0.49588996171951294, "learning_rate": 8.416599691059687e-05, "loss": 0.0205, "step": 50230 }, { "epoch": 3.4861002973229915e-13, "grad_norm": 0.5472085475921631, "learning_rate": 8.414436489634379e-05, "loss": 0.0213, "step": 50240 }, { "epoch": 3.4867941867133823e-13, "grad_norm": 0.4791385531425476, "learning_rate": 8.412273271310664e-05, "loss": 0.0206, "step": 50250 }, { "epoch": 3.487488076103773e-13, "grad_norm": 0.5575357675552368, "learning_rate": 8.410110036310845e-05, "loss": 0.0211, "step": 50260 }, { "epoch": 3.4881819654941637e-13, "grad_norm": 0.48662951588630676, "learning_rate": 8.407946784857231e-05, "loss": 0.0208, "step": 50270 }, { "epoch": 3.4888758548845544e-13, "grad_norm": 0.46033602952957153, "learning_rate": 8.405783517172123e-05, "loss": 0.0212, "step": 50280 }, { "epoch": 3.489569744274945e-13, "grad_norm": 0.4989683926105499, "learning_rate": 8.403620233477839e-05, "loss": 0.0202, "step": 50290 }, { "epoch": 3.490263633665336e-13, "grad_norm": 0.6299213767051697, "learning_rate": 8.401456933996683e-05, "loss": 0.0213, "step": 50300 }, { "epoch": 3.4909575230557266e-13, "grad_norm": 0.503720223903656, "learning_rate": 8.399293618950971e-05, "loss": 0.0206, "step": 50310 }, { "epoch": 3.4916514124461173e-13, "grad_norm": 0.5602874159812927, "learning_rate": 8.397130288563019e-05, "loss": 0.0211, "step": 50320 }, { "epoch": 3.492345301836508e-13, "grad_norm": 0.5662518739700317, "learning_rate": 8.394966943055139e-05, "loss": 0.0221, "step": 50330 }, { "epoch": 3.493039191226899e-13, "grad_norm": 0.534523069858551, "learning_rate": 8.392803582649649e-05, "loss": 0.0211, "step": 50340 }, { "epoch": 3.4937330806172895e-13, "grad_norm": 0.60425865650177, "learning_rate": 8.39064020756887e-05, "loss": 0.0226, "step": 50350 }, { "epoch": 3.49442697000768e-13, "grad_norm": 0.6164950132369995, "learning_rate": 8.38847681803512e-05, "loss": 0.0223, "step": 50360 }, { "epoch": 3.495120859398071e-13, "grad_norm": 0.5601711273193359, "learning_rate": 8.386313414270724e-05, "loss": 0.0215, "step": 50370 }, { "epoch": 3.4958147487884617e-13, "grad_norm": 0.5277507305145264, "learning_rate": 8.384149996498002e-05, "loss": 0.0208, "step": 50380 }, { "epoch": 3.4965086381788524e-13, "grad_norm": 0.5349122881889343, "learning_rate": 8.381986564939282e-05, "loss": 0.0201, "step": 50390 }, { "epoch": 3.497202527569243e-13, "grad_norm": 0.5525115132331848, "learning_rate": 8.379823119816887e-05, "loss": 0.0208, "step": 50400 }, { "epoch": 3.497896416959634e-13, "grad_norm": 0.4915927052497864, "learning_rate": 8.377659661353149e-05, "loss": 0.0206, "step": 50410 }, { "epoch": 3.4985903063500245e-13, "grad_norm": 0.4410668909549713, "learning_rate": 8.375496189770396e-05, "loss": 0.0203, "step": 50420 }, { "epoch": 3.4992841957404153e-13, "grad_norm": 0.5178558826446533, "learning_rate": 8.373332705290954e-05, "loss": 0.0209, "step": 50430 }, { "epoch": 3.499978085130806e-13, "grad_norm": 0.5216947793960571, "learning_rate": 8.371169208137157e-05, "loss": 0.0201, "step": 50440 }, { "epoch": 3.5006719745211967e-13, "grad_norm": 0.535340428352356, "learning_rate": 8.369005698531338e-05, "loss": 0.0204, "step": 50450 }, { "epoch": 3.5013658639115874e-13, "grad_norm": 0.538616955280304, "learning_rate": 8.366842176695835e-05, "loss": 0.0203, "step": 50460 }, { "epoch": 3.502059753301978e-13, "grad_norm": 0.5492199659347534, "learning_rate": 8.364678642852978e-05, "loss": 0.0211, "step": 50470 }, { "epoch": 3.502753642692369e-13, "grad_norm": 0.5580446720123291, "learning_rate": 8.362515097225106e-05, "loss": 0.0209, "step": 50480 }, { "epoch": 3.5034475320827596e-13, "grad_norm": 0.5499285459518433, "learning_rate": 8.360351540034558e-05, "loss": 0.0217, "step": 50490 }, { "epoch": 3.5041414214731503e-13, "grad_norm": 0.5528894662857056, "learning_rate": 8.358187971503674e-05, "loss": 0.0205, "step": 50500 }, { "epoch": 3.504835310863541e-13, "grad_norm": 0.48441439867019653, "learning_rate": 8.35602439185479e-05, "loss": 0.0218, "step": 50510 }, { "epoch": 3.505529200253932e-13, "grad_norm": 0.49637746810913086, "learning_rate": 8.353860801310252e-05, "loss": 0.023, "step": 50520 }, { "epoch": 3.5062230896443225e-13, "grad_norm": 0.5746678113937378, "learning_rate": 8.3516972000924e-05, "loss": 0.0207, "step": 50530 }, { "epoch": 3.506916979034713e-13, "grad_norm": 0.5738468170166016, "learning_rate": 8.349533588423576e-05, "loss": 0.0204, "step": 50540 }, { "epoch": 3.507610868425104e-13, "grad_norm": 0.4985070526599884, "learning_rate": 8.34736996652613e-05, "loss": 0.0202, "step": 50550 }, { "epoch": 3.5083047578154947e-13, "grad_norm": 0.5150710344314575, "learning_rate": 8.345206334622405e-05, "loss": 0.0204, "step": 50560 }, { "epoch": 3.5089986472058854e-13, "grad_norm": 0.5063187479972839, "learning_rate": 8.343042692934744e-05, "loss": 0.0194, "step": 50570 }, { "epoch": 3.509692536596276e-13, "grad_norm": 0.5621926188468933, "learning_rate": 8.3408790416855e-05, "loss": 0.0206, "step": 50580 }, { "epoch": 3.510386425986667e-13, "grad_norm": 0.5012655854225159, "learning_rate": 8.338715381097022e-05, "loss": 0.0205, "step": 50590 }, { "epoch": 3.5110803153770576e-13, "grad_norm": 0.49904635548591614, "learning_rate": 8.336551711391659e-05, "loss": 0.0198, "step": 50600 }, { "epoch": 3.5117742047674483e-13, "grad_norm": 0.5164356231689453, "learning_rate": 8.334388032791758e-05, "loss": 0.0201, "step": 50610 }, { "epoch": 3.512468094157839e-13, "grad_norm": 0.5486159324645996, "learning_rate": 8.332224345519677e-05, "loss": 0.0204, "step": 50620 }, { "epoch": 3.5131619835482297e-13, "grad_norm": 0.5459868907928467, "learning_rate": 8.330060649797765e-05, "loss": 0.0215, "step": 50630 }, { "epoch": 3.5138558729386205e-13, "grad_norm": 0.5239611268043518, "learning_rate": 8.327896945848374e-05, "loss": 0.0213, "step": 50640 }, { "epoch": 3.514549762329011e-13, "grad_norm": 0.5044001936912537, "learning_rate": 8.32573323389386e-05, "loss": 0.0208, "step": 50650 }, { "epoch": 3.515243651719402e-13, "grad_norm": 0.5527635812759399, "learning_rate": 8.323569514156578e-05, "loss": 0.0211, "step": 50660 }, { "epoch": 3.5159375411097926e-13, "grad_norm": 0.5414304733276367, "learning_rate": 8.321405786858883e-05, "loss": 0.0217, "step": 50670 }, { "epoch": 3.5166314305001833e-13, "grad_norm": 0.4521651566028595, "learning_rate": 8.319242052223133e-05, "loss": 0.0221, "step": 50680 }, { "epoch": 3.517325319890574e-13, "grad_norm": 0.46692314743995667, "learning_rate": 8.317078310471685e-05, "loss": 0.0204, "step": 50690 }, { "epoch": 3.518019209280965e-13, "grad_norm": 0.517399251461029, "learning_rate": 8.314914561826896e-05, "loss": 0.0203, "step": 50700 }, { "epoch": 3.5187130986713555e-13, "grad_norm": 0.4788537323474884, "learning_rate": 8.312750806511126e-05, "loss": 0.0203, "step": 50710 }, { "epoch": 3.519406988061746e-13, "grad_norm": 0.4685424864292145, "learning_rate": 8.310587044746736e-05, "loss": 0.0192, "step": 50720 }, { "epoch": 3.520100877452137e-13, "grad_norm": 0.5146034359931946, "learning_rate": 8.308423276756083e-05, "loss": 0.0194, "step": 50730 }, { "epoch": 3.5207947668425277e-13, "grad_norm": 0.5617352724075317, "learning_rate": 8.30625950276153e-05, "loss": 0.0203, "step": 50740 }, { "epoch": 3.5214886562329184e-13, "grad_norm": 0.48804187774658203, "learning_rate": 8.304095722985436e-05, "loss": 0.0199, "step": 50750 }, { "epoch": 3.522182545623309e-13, "grad_norm": 0.5285152792930603, "learning_rate": 8.301931937650166e-05, "loss": 0.0196, "step": 50760 }, { "epoch": 3.5228764350137e-13, "grad_norm": 0.5183587670326233, "learning_rate": 8.299768146978081e-05, "loss": 0.0201, "step": 50770 }, { "epoch": 3.5235703244040906e-13, "grad_norm": 0.5300604104995728, "learning_rate": 8.297604351191548e-05, "loss": 0.0206, "step": 50780 }, { "epoch": 3.5242642137944813e-13, "grad_norm": 0.5038935542106628, "learning_rate": 8.295440550512922e-05, "loss": 0.0211, "step": 50790 }, { "epoch": 3.524958103184872e-13, "grad_norm": 0.5192740559577942, "learning_rate": 8.293276745164574e-05, "loss": 0.0224, "step": 50800 }, { "epoch": 3.5256519925752627e-13, "grad_norm": 0.5053485035896301, "learning_rate": 8.291112935368868e-05, "loss": 0.0202, "step": 50810 }, { "epoch": 3.5263458819656535e-13, "grad_norm": 0.5621658563613892, "learning_rate": 8.288949121348167e-05, "loss": 0.0219, "step": 50820 }, { "epoch": 3.527039771356044e-13, "grad_norm": 0.581274151802063, "learning_rate": 8.28678530332484e-05, "loss": 0.0218, "step": 50830 }, { "epoch": 3.527733660746435e-13, "grad_norm": 0.5253773927688599, "learning_rate": 8.284621481521249e-05, "loss": 0.021, "step": 50840 }, { "epoch": 3.5284275501368256e-13, "grad_norm": 0.5336890816688538, "learning_rate": 8.282457656159764e-05, "loss": 0.0203, "step": 50850 }, { "epoch": 3.5291214395272164e-13, "grad_norm": 0.5808971524238586, "learning_rate": 8.280293827462748e-05, "loss": 0.0202, "step": 50860 }, { "epoch": 3.529815328917607e-13, "grad_norm": 0.5374729633331299, "learning_rate": 8.278129995652569e-05, "loss": 0.0197, "step": 50870 }, { "epoch": 3.530509218307998e-13, "grad_norm": 0.5055311918258667, "learning_rate": 8.275966160951596e-05, "loss": 0.0194, "step": 50880 }, { "epoch": 3.5312031076983885e-13, "grad_norm": 0.47235509753227234, "learning_rate": 8.273802323582198e-05, "loss": 0.0192, "step": 50890 }, { "epoch": 3.531896997088779e-13, "grad_norm": 0.594506025314331, "learning_rate": 8.271638483766739e-05, "loss": 0.0204, "step": 50900 }, { "epoch": 3.53259088647917e-13, "grad_norm": 0.5313072800636292, "learning_rate": 8.269474641727589e-05, "loss": 0.0191, "step": 50910 }, { "epoch": 3.5332847758695607e-13, "grad_norm": 0.4251670837402344, "learning_rate": 8.267310797687116e-05, "loss": 0.0194, "step": 50920 }, { "epoch": 3.5339786652599514e-13, "grad_norm": 0.49633222818374634, "learning_rate": 8.26514695186769e-05, "loss": 0.0198, "step": 50930 }, { "epoch": 3.534672554650342e-13, "grad_norm": 0.49278756976127625, "learning_rate": 8.262983104491677e-05, "loss": 0.0205, "step": 50940 }, { "epoch": 3.535366444040733e-13, "grad_norm": 0.4881819486618042, "learning_rate": 8.260819255781452e-05, "loss": 0.0209, "step": 50950 }, { "epoch": 3.5360603334311236e-13, "grad_norm": 0.495283305644989, "learning_rate": 8.258655405959379e-05, "loss": 0.022, "step": 50960 }, { "epoch": 3.5367542228215143e-13, "grad_norm": 0.5754495859146118, "learning_rate": 8.256491555247826e-05, "loss": 0.0207, "step": 50970 }, { "epoch": 3.537448112211905e-13, "grad_norm": 0.5671079754829407, "learning_rate": 8.254327703869167e-05, "loss": 0.021, "step": 50980 }, { "epoch": 3.538142001602296e-13, "grad_norm": 0.5485640168190002, "learning_rate": 8.252163852045768e-05, "loss": 0.0214, "step": 50990 }, { "epoch": 3.5388358909926865e-13, "grad_norm": 0.5542250871658325, "learning_rate": 8.25e-05, "loss": 0.0205, "step": 51000 }, { "epoch": 3.539529780383077e-13, "grad_norm": 0.5072221755981445, "learning_rate": 8.247836147954234e-05, "loss": 0.0202, "step": 51010 }, { "epoch": 3.540223669773468e-13, "grad_norm": 0.536436915397644, "learning_rate": 8.245672296130833e-05, "loss": 0.0204, "step": 51020 }, { "epoch": 3.5409175591638586e-13, "grad_norm": 0.5309360027313232, "learning_rate": 8.243508444752175e-05, "loss": 0.0198, "step": 51030 }, { "epoch": 3.5416114485542494e-13, "grad_norm": 0.5062083005905151, "learning_rate": 8.241344594040621e-05, "loss": 0.0187, "step": 51040 }, { "epoch": 3.54230533794464e-13, "grad_norm": 0.5330078601837158, "learning_rate": 8.239180744218549e-05, "loss": 0.0195, "step": 51050 }, { "epoch": 3.542999227335031e-13, "grad_norm": 0.5595748424530029, "learning_rate": 8.237016895508323e-05, "loss": 0.0197, "step": 51060 }, { "epoch": 3.5436931167254215e-13, "grad_norm": 0.47119131684303284, "learning_rate": 8.23485304813231e-05, "loss": 0.019, "step": 51070 }, { "epoch": 3.5443870061158123e-13, "grad_norm": 0.49038609862327576, "learning_rate": 8.232689202312884e-05, "loss": 0.0195, "step": 51080 }, { "epoch": 3.545080895506203e-13, "grad_norm": 0.5461416244506836, "learning_rate": 8.230525358272413e-05, "loss": 0.0201, "step": 51090 }, { "epoch": 3.5457747848965937e-13, "grad_norm": 0.5004540681838989, "learning_rate": 8.228361516233264e-05, "loss": 0.0203, "step": 51100 }, { "epoch": 3.5464686742869844e-13, "grad_norm": 0.5162896513938904, "learning_rate": 8.226197676417806e-05, "loss": 0.0218, "step": 51110 }, { "epoch": 3.547162563677375e-13, "grad_norm": 0.4973120093345642, "learning_rate": 8.224033839048404e-05, "loss": 0.0201, "step": 51120 }, { "epoch": 3.547856453067766e-13, "grad_norm": 0.5412853956222534, "learning_rate": 8.221870004347432e-05, "loss": 0.0211, "step": 51130 }, { "epoch": 3.5485503424581566e-13, "grad_norm": 0.5662870407104492, "learning_rate": 8.219706172537252e-05, "loss": 0.0211, "step": 51140 }, { "epoch": 3.5492442318485473e-13, "grad_norm": 0.5488186478614807, "learning_rate": 8.217542343840238e-05, "loss": 0.021, "step": 51150 }, { "epoch": 3.549938121238938e-13, "grad_norm": 0.6297606229782104, "learning_rate": 8.21537851847875e-05, "loss": 0.0201, "step": 51160 }, { "epoch": 3.550632010629329e-13, "grad_norm": 0.4891548454761505, "learning_rate": 8.21321469667516e-05, "loss": 0.0196, "step": 51170 }, { "epoch": 3.5513259000197195e-13, "grad_norm": 0.4483289420604706, "learning_rate": 8.211050878651831e-05, "loss": 0.0207, "step": 51180 }, { "epoch": 3.55201978941011e-13, "grad_norm": 0.4963017404079437, "learning_rate": 8.208887064631133e-05, "loss": 0.0196, "step": 51190 }, { "epoch": 3.552713678800501e-13, "grad_norm": 0.5048074722290039, "learning_rate": 8.206723254835426e-05, "loss": 0.019, "step": 51200 }, { "epoch": 3.5534075681908917e-13, "grad_norm": 0.5181759595870972, "learning_rate": 8.204559449487078e-05, "loss": 0.0197, "step": 51210 }, { "epoch": 3.5541014575812824e-13, "grad_norm": 0.5336496233940125, "learning_rate": 8.202395648808452e-05, "loss": 0.0191, "step": 51220 }, { "epoch": 3.554795346971673e-13, "grad_norm": 0.5446047186851501, "learning_rate": 8.200231853021919e-05, "loss": 0.0196, "step": 51230 }, { "epoch": 3.555489236362064e-13, "grad_norm": 0.4930143654346466, "learning_rate": 8.198068062349831e-05, "loss": 0.0193, "step": 51240 }, { "epoch": 3.5561831257524545e-13, "grad_norm": 0.5383409261703491, "learning_rate": 8.195904277014563e-05, "loss": 0.0196, "step": 51250 }, { "epoch": 3.5568770151428453e-13, "grad_norm": 0.5325444936752319, "learning_rate": 8.19374049723847e-05, "loss": 0.0203, "step": 51260 }, { "epoch": 3.557570904533236e-13, "grad_norm": 0.5413357615470886, "learning_rate": 8.191576723243917e-05, "loss": 0.0213, "step": 51270 }, { "epoch": 3.5582647939236267e-13, "grad_norm": 0.523770272731781, "learning_rate": 8.189412955253263e-05, "loss": 0.0196, "step": 51280 }, { "epoch": 3.5589586833140174e-13, "grad_norm": 0.49768856167793274, "learning_rate": 8.187249193488874e-05, "loss": 0.0214, "step": 51290 }, { "epoch": 3.559652572704408e-13, "grad_norm": 0.4990467429161072, "learning_rate": 8.185085438173102e-05, "loss": 0.021, "step": 51300 }, { "epoch": 3.560346462094799e-13, "grad_norm": 0.4768422842025757, "learning_rate": 8.182921689528315e-05, "loss": 0.02, "step": 51310 }, { "epoch": 3.5610403514851896e-13, "grad_norm": 0.5412383079528809, "learning_rate": 8.180757947776866e-05, "loss": 0.0198, "step": 51320 }, { "epoch": 3.5617342408755803e-13, "grad_norm": 0.6304457187652588, "learning_rate": 8.178594213141117e-05, "loss": 0.0207, "step": 51330 }, { "epoch": 3.562428130265971e-13, "grad_norm": 0.4911624789237976, "learning_rate": 8.17643048584342e-05, "loss": 0.0196, "step": 51340 }, { "epoch": 3.563122019656362e-13, "grad_norm": 0.47107744216918945, "learning_rate": 8.17426676610614e-05, "loss": 0.0192, "step": 51350 }, { "epoch": 3.5638159090467525e-13, "grad_norm": 0.48941850662231445, "learning_rate": 8.172103054151624e-05, "loss": 0.0189, "step": 51360 }, { "epoch": 3.564509798437143e-13, "grad_norm": 0.462238073348999, "learning_rate": 8.169939350202235e-05, "loss": 0.0202, "step": 51370 }, { "epoch": 3.565203687827534e-13, "grad_norm": 0.47678422927856445, "learning_rate": 8.16777565448032e-05, "loss": 0.0189, "step": 51380 }, { "epoch": 3.5658975772179247e-13, "grad_norm": 0.4561671316623688, "learning_rate": 8.16561196720824e-05, "loss": 0.0191, "step": 51390 }, { "epoch": 3.5665914666083154e-13, "grad_norm": 0.4658696949481964, "learning_rate": 8.163448288608339e-05, "loss": 0.0195, "step": 51400 }, { "epoch": 3.567285355998706e-13, "grad_norm": 0.5183248519897461, "learning_rate": 8.161284618902978e-05, "loss": 0.0198, "step": 51410 }, { "epoch": 3.567979245389097e-13, "grad_norm": 0.5826743245124817, "learning_rate": 8.159120958314496e-05, "loss": 0.0202, "step": 51420 }, { "epoch": 3.5686731347794876e-13, "grad_norm": 0.4717167615890503, "learning_rate": 8.156957307065255e-05, "loss": 0.02, "step": 51430 }, { "epoch": 3.5693670241698783e-13, "grad_norm": 0.5113921165466309, "learning_rate": 8.154793665377595e-05, "loss": 0.02, "step": 51440 }, { "epoch": 3.570060913560269e-13, "grad_norm": 0.5014706254005432, "learning_rate": 8.15263003347387e-05, "loss": 0.0204, "step": 51450 }, { "epoch": 3.5707548029506597e-13, "grad_norm": 0.49090471863746643, "learning_rate": 8.150466411576425e-05, "loss": 0.0209, "step": 51460 }, { "epoch": 3.5714486923410504e-13, "grad_norm": 0.476012647151947, "learning_rate": 8.148302799907602e-05, "loss": 0.0196, "step": 51470 }, { "epoch": 3.572142581731441e-13, "grad_norm": 0.4929237365722656, "learning_rate": 8.146139198689752e-05, "loss": 0.0197, "step": 51480 }, { "epoch": 3.572836471121832e-13, "grad_norm": 0.5960633158683777, "learning_rate": 8.14397560814521e-05, "loss": 0.0208, "step": 51490 }, { "epoch": 3.5735303605122226e-13, "grad_norm": 0.4975767433643341, "learning_rate": 8.141812028496328e-05, "loss": 0.0195, "step": 51500 }, { "epoch": 3.5742242499026133e-13, "grad_norm": 0.5248620510101318, "learning_rate": 8.139648459965442e-05, "loss": 0.0193, "step": 51510 }, { "epoch": 3.574918139293004e-13, "grad_norm": 0.4614652991294861, "learning_rate": 8.137484902774895e-05, "loss": 0.0192, "step": 51520 }, { "epoch": 3.575612028683395e-13, "grad_norm": 0.46380940079689026, "learning_rate": 8.135321357147022e-05, "loss": 0.0196, "step": 51530 }, { "epoch": 3.5763059180737855e-13, "grad_norm": 0.4607473909854889, "learning_rate": 8.133157823304167e-05, "loss": 0.019, "step": 51540 }, { "epoch": 3.576999807464176e-13, "grad_norm": 0.5590870976448059, "learning_rate": 8.130994301468662e-05, "loss": 0.0196, "step": 51550 }, { "epoch": 3.577693696854567e-13, "grad_norm": 0.5036726593971252, "learning_rate": 8.128830791862845e-05, "loss": 0.0187, "step": 51560 }, { "epoch": 3.5783875862449577e-13, "grad_norm": 0.5169402360916138, "learning_rate": 8.126667294709046e-05, "loss": 0.0197, "step": 51570 }, { "epoch": 3.5790814756353484e-13, "grad_norm": 0.5474885106086731, "learning_rate": 8.124503810229607e-05, "loss": 0.0209, "step": 51580 }, { "epoch": 3.579775365025739e-13, "grad_norm": 0.5531172752380371, "learning_rate": 8.12234033864685e-05, "loss": 0.0193, "step": 51590 }, { "epoch": 3.58046925441613e-13, "grad_norm": 0.5910763144493103, "learning_rate": 8.120176880183111e-05, "loss": 0.0205, "step": 51600 }, { "epoch": 3.5811631438065206e-13, "grad_norm": 0.5710684657096863, "learning_rate": 8.118013435060718e-05, "loss": 0.02, "step": 51610 }, { "epoch": 3.5818570331969113e-13, "grad_norm": 0.5400375723838806, "learning_rate": 8.115850003502e-05, "loss": 0.0199, "step": 51620 }, { "epoch": 3.582550922587302e-13, "grad_norm": 0.5003958940505981, "learning_rate": 8.113686585729277e-05, "loss": 0.0193, "step": 51630 }, { "epoch": 3.5832448119776927e-13, "grad_norm": 0.5153312087059021, "learning_rate": 8.111523181964882e-05, "loss": 0.0193, "step": 51640 }, { "epoch": 3.5839387013680835e-13, "grad_norm": 0.5624439716339111, "learning_rate": 8.10935979243113e-05, "loss": 0.0203, "step": 51650 }, { "epoch": 3.584632590758474e-13, "grad_norm": 0.5289732217788696, "learning_rate": 8.107196417350351e-05, "loss": 0.0192, "step": 51660 }, { "epoch": 3.585326480148865e-13, "grad_norm": 0.43417972326278687, "learning_rate": 8.105033056944862e-05, "loss": 0.0193, "step": 51670 }, { "epoch": 3.5860203695392556e-13, "grad_norm": 0.5052714943885803, "learning_rate": 8.102869711436984e-05, "loss": 0.02, "step": 51680 }, { "epoch": 3.5867142589296463e-13, "grad_norm": 0.5107351541519165, "learning_rate": 8.100706381049026e-05, "loss": 0.0194, "step": 51690 }, { "epoch": 3.587408148320037e-13, "grad_norm": 0.4910464882850647, "learning_rate": 8.098543066003317e-05, "loss": 0.0192, "step": 51700 }, { "epoch": 3.588102037710428e-13, "grad_norm": 0.4969979226589203, "learning_rate": 8.096379766522161e-05, "loss": 0.0187, "step": 51710 }, { "epoch": 3.5887959271008185e-13, "grad_norm": 0.510028064250946, "learning_rate": 8.094216482827876e-05, "loss": 0.0192, "step": 51720 }, { "epoch": 3.589489816491209e-13, "grad_norm": 0.5371729135513306, "learning_rate": 8.092053215142769e-05, "loss": 0.0196, "step": 51730 }, { "epoch": 3.5901837058816e-13, "grad_norm": 0.5135911107063293, "learning_rate": 8.089889963689155e-05, "loss": 0.0203, "step": 51740 }, { "epoch": 3.5908775952719907e-13, "grad_norm": 0.5426809787750244, "learning_rate": 8.087726728689335e-05, "loss": 0.0194, "step": 51750 }, { "epoch": 3.5915714846623814e-13, "grad_norm": 0.48949432373046875, "learning_rate": 8.085563510365621e-05, "loss": 0.0199, "step": 51760 }, { "epoch": 3.592265374052772e-13, "grad_norm": 0.49874234199523926, "learning_rate": 8.083400308940312e-05, "loss": 0.0207, "step": 51770 }, { "epoch": 3.592959263443163e-13, "grad_norm": 0.5553842186927795, "learning_rate": 8.081237124635715e-05, "loss": 0.0201, "step": 51780 }, { "epoch": 3.5936531528335536e-13, "grad_norm": 0.5667583346366882, "learning_rate": 8.079073957674128e-05, "loss": 0.0192, "step": 51790 }, { "epoch": 3.5943470422239443e-13, "grad_norm": 0.5186908841133118, "learning_rate": 8.076910808277857e-05, "loss": 0.02, "step": 51800 }, { "epoch": 3.595040931614335e-13, "grad_norm": 0.5100565552711487, "learning_rate": 8.074747676669189e-05, "loss": 0.0199, "step": 51810 }, { "epoch": 3.5957348210047257e-13, "grad_norm": 0.5348549485206604, "learning_rate": 8.072584563070424e-05, "loss": 0.0188, "step": 51820 }, { "epoch": 3.5964287103951165e-13, "grad_norm": 0.5544939637184143, "learning_rate": 8.070421467703855e-05, "loss": 0.0198, "step": 51830 }, { "epoch": 3.597122599785507e-13, "grad_norm": 0.4801314175128937, "learning_rate": 8.068258390791777e-05, "loss": 0.0202, "step": 51840 }, { "epoch": 3.597816489175898e-13, "grad_norm": 0.5336490273475647, "learning_rate": 8.066095332556475e-05, "loss": 0.0191, "step": 51850 }, { "epoch": 3.5985103785662886e-13, "grad_norm": 0.5180385112762451, "learning_rate": 8.06393229322024e-05, "loss": 0.019, "step": 51860 }, { "epoch": 3.5992042679566794e-13, "grad_norm": 0.49208158254623413, "learning_rate": 8.061769273005354e-05, "loss": 0.019, "step": 51870 }, { "epoch": 3.59989815734707e-13, "grad_norm": 0.5132974982261658, "learning_rate": 8.059606272134107e-05, "loss": 0.0195, "step": 51880 }, { "epoch": 3.600592046737461e-13, "grad_norm": 0.4960877299308777, "learning_rate": 8.057443290828777e-05, "loss": 0.0196, "step": 51890 }, { "epoch": 3.6012859361278515e-13, "grad_norm": 0.462104469537735, "learning_rate": 8.055280329311646e-05, "loss": 0.0197, "step": 51900 }, { "epoch": 3.601979825518242e-13, "grad_norm": 0.5213698744773865, "learning_rate": 8.053117387804992e-05, "loss": 0.0193, "step": 51910 }, { "epoch": 3.602673714908633e-13, "grad_norm": 0.4807513952255249, "learning_rate": 8.050954466531084e-05, "loss": 0.0198, "step": 51920 }, { "epoch": 3.6033676042990237e-13, "grad_norm": 0.45045679807662964, "learning_rate": 8.04879156571221e-05, "loss": 0.0206, "step": 51930 }, { "epoch": 3.6040614936894144e-13, "grad_norm": 0.44461601972579956, "learning_rate": 8.046628685570627e-05, "loss": 0.0186, "step": 51940 }, { "epoch": 3.604755383079805e-13, "grad_norm": 0.5403242707252502, "learning_rate": 8.044465826328616e-05, "loss": 0.0194, "step": 51950 }, { "epoch": 3.605449272470196e-13, "grad_norm": 0.49109378457069397, "learning_rate": 8.042302988208433e-05, "loss": 0.0197, "step": 51960 }, { "epoch": 3.6061431618605866e-13, "grad_norm": 0.4720197021961212, "learning_rate": 8.040140171432353e-05, "loss": 0.0191, "step": 51970 }, { "epoch": 3.6068370512509773e-13, "grad_norm": 0.48723945021629333, "learning_rate": 8.037977376222634e-05, "loss": 0.0192, "step": 51980 }, { "epoch": 3.607530940641368e-13, "grad_norm": 0.5523606538772583, "learning_rate": 8.03581460280154e-05, "loss": 0.0197, "step": 51990 }, { "epoch": 3.608224830031759e-13, "grad_norm": 0.5176199078559875, "learning_rate": 8.033651851391327e-05, "loss": 0.0191, "step": 52000 }, { "epoch": 3.6089187194221495e-13, "grad_norm": 0.5670444965362549, "learning_rate": 8.031489122214256e-05, "loss": 0.019, "step": 52010 }, { "epoch": 3.60961260881254e-13, "grad_norm": 0.5113219618797302, "learning_rate": 8.029326415492571e-05, "loss": 0.0191, "step": 52020 }, { "epoch": 3.610306498202931e-13, "grad_norm": 0.5102171301841736, "learning_rate": 8.027163731448537e-05, "loss": 0.0186, "step": 52030 }, { "epoch": 3.6110003875933216e-13, "grad_norm": 0.4795995354652405, "learning_rate": 8.025001070304393e-05, "loss": 0.0191, "step": 52040 }, { "epoch": 3.6116942769837124e-13, "grad_norm": 0.4871818423271179, "learning_rate": 8.022838432282392e-05, "loss": 0.0196, "step": 52050 }, { "epoch": 3.612388166374103e-13, "grad_norm": 0.5055462121963501, "learning_rate": 8.020675817604774e-05, "loss": 0.0186, "step": 52060 }, { "epoch": 3.613082055764494e-13, "grad_norm": 0.5219749212265015, "learning_rate": 8.018513226493787e-05, "loss": 0.0201, "step": 52070 }, { "epoch": 3.6137759451548845e-13, "grad_norm": 0.5431345105171204, "learning_rate": 8.016350659171664e-05, "loss": 0.0199, "step": 52080 }, { "epoch": 3.6144698345452753e-13, "grad_norm": 0.5095189213752747, "learning_rate": 8.014188115860649e-05, "loss": 0.0197, "step": 52090 }, { "epoch": 3.615163723935666e-13, "grad_norm": 0.4636484384536743, "learning_rate": 8.012025596782972e-05, "loss": 0.018, "step": 52100 }, { "epoch": 3.6158576133260567e-13, "grad_norm": 0.5657482743263245, "learning_rate": 8.009863102160872e-05, "loss": 0.019, "step": 52110 }, { "epoch": 3.6165515027164474e-13, "grad_norm": 0.5844066739082336, "learning_rate": 8.007700632216569e-05, "loss": 0.019, "step": 52120 }, { "epoch": 3.617245392106838e-13, "grad_norm": 0.5260088443756104, "learning_rate": 8.0055381871723e-05, "loss": 0.0188, "step": 52130 }, { "epoch": 3.617939281497229e-13, "grad_norm": 0.48184892535209656, "learning_rate": 8.003375767250285e-05, "loss": 0.0192, "step": 52140 }, { "epoch": 3.6186331708876196e-13, "grad_norm": 0.5808162689208984, "learning_rate": 8.001213372672748e-05, "loss": 0.0204, "step": 52150 }, { "epoch": 3.6193270602780103e-13, "grad_norm": 0.5282675623893738, "learning_rate": 7.999051003661904e-05, "loss": 0.0189, "step": 52160 }, { "epoch": 3.620020949668401e-13, "grad_norm": 0.4820133149623871, "learning_rate": 7.996888660439979e-05, "loss": 0.0191, "step": 52170 }, { "epoch": 3.620714839058792e-13, "grad_norm": 0.49576205015182495, "learning_rate": 7.99472634322918e-05, "loss": 0.0189, "step": 52180 }, { "epoch": 3.6214087284491825e-13, "grad_norm": 0.48679858446121216, "learning_rate": 7.99256405225172e-05, "loss": 0.019, "step": 52190 }, { "epoch": 3.622102617839573e-13, "grad_norm": 0.4590948820114136, "learning_rate": 7.99040178772981e-05, "loss": 0.019, "step": 52200 }, { "epoch": 3.622796507229964e-13, "grad_norm": 0.5012454986572266, "learning_rate": 7.988239549885658e-05, "loss": 0.0203, "step": 52210 }, { "epoch": 3.6234903966203547e-13, "grad_norm": 0.5440102815628052, "learning_rate": 7.98607733894146e-05, "loss": 0.0187, "step": 52220 }, { "epoch": 3.6241842860107454e-13, "grad_norm": 0.5307241678237915, "learning_rate": 7.983915155119426e-05, "loss": 0.0195, "step": 52230 }, { "epoch": 3.624878175401136e-13, "grad_norm": 0.4920610785484314, "learning_rate": 7.981752998641749e-05, "loss": 0.02, "step": 52240 }, { "epoch": 3.625572064791527e-13, "grad_norm": 0.5133256316184998, "learning_rate": 7.979590869730625e-05, "loss": 0.0192, "step": 52250 }, { "epoch": 3.6262659541819176e-13, "grad_norm": 0.49406787753105164, "learning_rate": 7.977428768608246e-05, "loss": 0.0183, "step": 52260 }, { "epoch": 3.6269598435723083e-13, "grad_norm": 0.5345662236213684, "learning_rate": 7.975266695496804e-05, "loss": 0.0191, "step": 52270 }, { "epoch": 3.627653732962699e-13, "grad_norm": 0.540333092212677, "learning_rate": 7.97310465061848e-05, "loss": 0.0189, "step": 52280 }, { "epoch": 3.6283476223530897e-13, "grad_norm": 0.5203160643577576, "learning_rate": 7.970942634195465e-05, "loss": 0.0181, "step": 52290 }, { "epoch": 3.6290415117434804e-13, "grad_norm": 0.5040625333786011, "learning_rate": 7.968780646449933e-05, "loss": 0.0191, "step": 52300 }, { "epoch": 3.629735401133871e-13, "grad_norm": 0.5044330954551697, "learning_rate": 7.966618687604069e-05, "loss": 0.0191, "step": 52310 }, { "epoch": 3.630429290524262e-13, "grad_norm": 0.48973557353019714, "learning_rate": 7.96445675788004e-05, "loss": 0.0184, "step": 52320 }, { "epoch": 3.6311231799146526e-13, "grad_norm": 0.4843593239784241, "learning_rate": 7.962294857500026e-05, "loss": 0.0182, "step": 52330 }, { "epoch": 3.6318170693050433e-13, "grad_norm": 0.4897971749305725, "learning_rate": 7.96013298668619e-05, "loss": 0.0188, "step": 52340 }, { "epoch": 3.632510958695434e-13, "grad_norm": 0.47415661811828613, "learning_rate": 7.9579711456607e-05, "loss": 0.0192, "step": 52350 }, { "epoch": 3.633204848085825e-13, "grad_norm": 0.4718707799911499, "learning_rate": 7.955809334645718e-05, "loss": 0.0196, "step": 52360 }, { "epoch": 3.6338987374762155e-13, "grad_norm": 0.4939899742603302, "learning_rate": 7.953647553863406e-05, "loss": 0.0189, "step": 52370 }, { "epoch": 3.634592626866606e-13, "grad_norm": 0.5531481504440308, "learning_rate": 7.951485803535917e-05, "loss": 0.0194, "step": 52380 }, { "epoch": 3.635286516256997e-13, "grad_norm": 0.5501035451889038, "learning_rate": 7.949324083885406e-05, "loss": 0.0195, "step": 52390 }, { "epoch": 3.6359804056473877e-13, "grad_norm": 0.5186317563056946, "learning_rate": 7.947162395134027e-05, "loss": 0.0196, "step": 52400 }, { "epoch": 3.6366742950377784e-13, "grad_norm": 0.5752972960472107, "learning_rate": 7.945000737503918e-05, "loss": 0.0185, "step": 52410 }, { "epoch": 3.637368184428169e-13, "grad_norm": 0.45923498272895813, "learning_rate": 7.942839111217234e-05, "loss": 0.0177, "step": 52420 }, { "epoch": 3.63806207381856e-13, "grad_norm": 0.41857555508613586, "learning_rate": 7.940677516496106e-05, "loss": 0.0186, "step": 52430 }, { "epoch": 3.6387559632089506e-13, "grad_norm": 0.4733777344226837, "learning_rate": 7.938515953562681e-05, "loss": 0.0185, "step": 52440 }, { "epoch": 3.6394498525993413e-13, "grad_norm": 0.498624712228775, "learning_rate": 7.936354422639083e-05, "loss": 0.0187, "step": 52450 }, { "epoch": 3.640143741989732e-13, "grad_norm": 0.5429264307022095, "learning_rate": 7.934192923947454e-05, "loss": 0.0188, "step": 52460 }, { "epoch": 3.6408376313801227e-13, "grad_norm": 0.5128287672996521, "learning_rate": 7.932031457709912e-05, "loss": 0.0184, "step": 52470 }, { "epoch": 3.6415315207705135e-13, "grad_norm": 0.5124810338020325, "learning_rate": 7.929870024148587e-05, "loss": 0.0184, "step": 52480 }, { "epoch": 3.642225410160904e-13, "grad_norm": 0.4863266944885254, "learning_rate": 7.927708623485596e-05, "loss": 0.0185, "step": 52490 }, { "epoch": 3.642919299551295e-13, "grad_norm": 0.496317058801651, "learning_rate": 7.92554725594306e-05, "loss": 0.0184, "step": 52500 }, { "epoch": 3.6436131889416856e-13, "grad_norm": 0.5170130729675293, "learning_rate": 7.92338592174309e-05, "loss": 0.0192, "step": 52510 }, { "epoch": 3.6443070783320763e-13, "grad_norm": 0.4963432252407074, "learning_rate": 7.921224621107801e-05, "loss": 0.0196, "step": 52520 }, { "epoch": 3.645000967722467e-13, "grad_norm": 0.4874110221862793, "learning_rate": 7.919063354259295e-05, "loss": 0.0186, "step": 52530 }, { "epoch": 3.645694857112858e-13, "grad_norm": 0.5008116364479065, "learning_rate": 7.916902121419682e-05, "loss": 0.0196, "step": 52540 }, { "epoch": 3.6463887465032485e-13, "grad_norm": 0.4828372895717621, "learning_rate": 7.914740922811055e-05, "loss": 0.0198, "step": 52550 }, { "epoch": 3.647082635893639e-13, "grad_norm": 0.46896684169769287, "learning_rate": 7.912579758655518e-05, "loss": 0.0193, "step": 52560 }, { "epoch": 3.64777652528403e-13, "grad_norm": 0.5084020495414734, "learning_rate": 7.910418629175159e-05, "loss": 0.0183, "step": 52570 }, { "epoch": 3.6484704146744207e-13, "grad_norm": 0.5649613738059998, "learning_rate": 7.90825753459207e-05, "loss": 0.0189, "step": 52580 }, { "epoch": 3.6491643040648114e-13, "grad_norm": 0.5024621486663818, "learning_rate": 7.906096475128334e-05, "loss": 0.018, "step": 52590 }, { "epoch": 3.649858193455202e-13, "grad_norm": 0.4404276907444, "learning_rate": 7.903935451006041e-05, "loss": 0.0181, "step": 52600 }, { "epoch": 3.650552082845593e-13, "grad_norm": 0.4852730333805084, "learning_rate": 7.90177446244726e-05, "loss": 0.018, "step": 52610 }, { "epoch": 3.6512459722359836e-13, "grad_norm": 0.505782961845398, "learning_rate": 7.899613509674075e-05, "loss": 0.0196, "step": 52620 }, { "epoch": 3.6519398616263743e-13, "grad_norm": 0.45590639114379883, "learning_rate": 7.89745259290855e-05, "loss": 0.0181, "step": 52630 }, { "epoch": 3.652633751016765e-13, "grad_norm": 0.43210721015930176, "learning_rate": 7.89529171237276e-05, "loss": 0.0179, "step": 52640 }, { "epoch": 3.6533276404071557e-13, "grad_norm": 0.47125452756881714, "learning_rate": 7.893130868288763e-05, "loss": 0.0185, "step": 52650 }, { "epoch": 3.6540215297975465e-13, "grad_norm": 0.5178036689758301, "learning_rate": 7.890970060878624e-05, "loss": 0.0186, "step": 52660 }, { "epoch": 3.654715419187937e-13, "grad_norm": 0.5484645962715149, "learning_rate": 7.888809290364396e-05, "loss": 0.0192, "step": 52670 }, { "epoch": 3.655409308578328e-13, "grad_norm": 0.4479076862335205, "learning_rate": 7.886648556968133e-05, "loss": 0.0191, "step": 52680 }, { "epoch": 3.6561031979687186e-13, "grad_norm": 0.47525545954704285, "learning_rate": 7.884487860911886e-05, "loss": 0.0187, "step": 52690 }, { "epoch": 3.6567970873591094e-13, "grad_norm": 0.485935777425766, "learning_rate": 7.882327202417699e-05, "loss": 0.019, "step": 52700 }, { "epoch": 3.6574909767495e-13, "grad_norm": 0.6892603039741516, "learning_rate": 7.880166581707608e-05, "loss": 0.0215, "step": 52710 }, { "epoch": 3.658184866139891e-13, "grad_norm": 0.458909809589386, "learning_rate": 7.878005999003662e-05, "loss": 0.0187, "step": 52720 }, { "epoch": 3.6588787555302815e-13, "grad_norm": 0.4745040833950043, "learning_rate": 7.875845454527881e-05, "loss": 0.0184, "step": 52730 }, { "epoch": 3.659572644920672e-13, "grad_norm": 0.5070935487747192, "learning_rate": 7.873684948502307e-05, "loss": 0.0192, "step": 52740 }, { "epoch": 3.660266534311063e-13, "grad_norm": 0.4459710717201233, "learning_rate": 7.871524481148954e-05, "loss": 0.018, "step": 52750 }, { "epoch": 3.6609604237014537e-13, "grad_norm": 0.5008001923561096, "learning_rate": 7.869364052689853e-05, "loss": 0.0175, "step": 52760 }, { "epoch": 3.6616543130918444e-13, "grad_norm": 0.4580323100090027, "learning_rate": 7.867203663347018e-05, "loss": 0.018, "step": 52770 }, { "epoch": 3.662348202482235e-13, "grad_norm": 0.45940425992012024, "learning_rate": 7.865043313342462e-05, "loss": 0.0186, "step": 52780 }, { "epoch": 3.663042091872626e-13, "grad_norm": 0.4915725588798523, "learning_rate": 7.862883002898194e-05, "loss": 0.0179, "step": 52790 }, { "epoch": 3.6637359812630166e-13, "grad_norm": 0.5396838784217834, "learning_rate": 7.860722732236223e-05, "loss": 0.0188, "step": 52800 }, { "epoch": 3.6644298706534073e-13, "grad_norm": 0.460480272769928, "learning_rate": 7.858562501578546e-05, "loss": 0.0182, "step": 52810 }, { "epoch": 3.665123760043798e-13, "grad_norm": 0.5141297578811646, "learning_rate": 7.856402311147164e-05, "loss": 0.0186, "step": 52820 }, { "epoch": 3.665817649434189e-13, "grad_norm": 0.5512241721153259, "learning_rate": 7.85424216116407e-05, "loss": 0.0195, "step": 52830 }, { "epoch": 3.6665115388245795e-13, "grad_norm": 0.5236852765083313, "learning_rate": 7.852082051851247e-05, "loss": 0.0185, "step": 52840 }, { "epoch": 3.66720542821497e-13, "grad_norm": 0.556244969367981, "learning_rate": 7.84992198343069e-05, "loss": 0.0194, "step": 52850 }, { "epoch": 3.667899317605361e-13, "grad_norm": 0.5411030650138855, "learning_rate": 7.847761956124369e-05, "loss": 0.0187, "step": 52860 }, { "epoch": 3.6685932069957516e-13, "grad_norm": 0.4945814609527588, "learning_rate": 7.845601970154268e-05, "loss": 0.0196, "step": 52870 }, { "epoch": 3.6692870963861424e-13, "grad_norm": 0.4700498580932617, "learning_rate": 7.843442025742355e-05, "loss": 0.0184, "step": 52880 }, { "epoch": 3.669980985776533e-13, "grad_norm": 0.49120664596557617, "learning_rate": 7.841282123110603e-05, "loss": 0.0174, "step": 52890 }, { "epoch": 3.670674875166924e-13, "grad_norm": 0.5270082950592041, "learning_rate": 7.83912226248097e-05, "loss": 0.0187, "step": 52900 }, { "epoch": 3.6713687645573145e-13, "grad_norm": 0.45403531193733215, "learning_rate": 7.836962444075417e-05, "loss": 0.0174, "step": 52910 }, { "epoch": 3.6720626539477053e-13, "grad_norm": 0.4093081057071686, "learning_rate": 7.834802668115901e-05, "loss": 0.0175, "step": 52920 }, { "epoch": 3.672756543338096e-13, "grad_norm": 0.5004528760910034, "learning_rate": 7.832642934824372e-05, "loss": 0.0189, "step": 52930 }, { "epoch": 3.6734504327284867e-13, "grad_norm": 0.5032355785369873, "learning_rate": 7.830483244422773e-05, "loss": 0.0178, "step": 52940 }, { "epoch": 3.6741443221188774e-13, "grad_norm": 0.4967758059501648, "learning_rate": 7.82832359713305e-05, "loss": 0.0183, "step": 52950 }, { "epoch": 3.674838211509268e-13, "grad_norm": 0.47304829955101013, "learning_rate": 7.826163993177135e-05, "loss": 0.018, "step": 52960 }, { "epoch": 3.675532100899659e-13, "grad_norm": 0.47070497274398804, "learning_rate": 7.824004432776967e-05, "loss": 0.0182, "step": 52970 }, { "epoch": 3.6762259902900496e-13, "grad_norm": 0.5364718437194824, "learning_rate": 7.821844916154468e-05, "loss": 0.0186, "step": 52980 }, { "epoch": 3.6769198796804403e-13, "grad_norm": 0.5010282397270203, "learning_rate": 7.819685443531571e-05, "loss": 0.019, "step": 52990 }, { "epoch": 3.677613769070831e-13, "grad_norm": 0.499502569437027, "learning_rate": 7.817526015130187e-05, "loss": 0.0182, "step": 53000 }, { "epoch": 3.678307658461222e-13, "grad_norm": 0.46644094586372375, "learning_rate": 7.815366631172235e-05, "loss": 0.0193, "step": 53010 }, { "epoch": 3.6790015478516125e-13, "grad_norm": 0.4761579632759094, "learning_rate": 7.813207291879625e-05, "loss": 0.0194, "step": 53020 }, { "epoch": 3.679695437242003e-13, "grad_norm": 0.5245131254196167, "learning_rate": 7.811047997474262e-05, "loss": 0.0186, "step": 53030 }, { "epoch": 3.680389326632394e-13, "grad_norm": 0.5154314637184143, "learning_rate": 7.808888748178044e-05, "loss": 0.0176, "step": 53040 }, { "epoch": 3.6810832160227847e-13, "grad_norm": 0.49082887172698975, "learning_rate": 7.806729544212875e-05, "loss": 0.0179, "step": 53050 }, { "epoch": 3.6817771054131754e-13, "grad_norm": 0.4765118360519409, "learning_rate": 7.804570385800637e-05, "loss": 0.0181, "step": 53060 }, { "epoch": 3.682470994803566e-13, "grad_norm": 0.4630770683288574, "learning_rate": 7.802411273163227e-05, "loss": 0.017, "step": 53070 }, { "epoch": 3.683164884193957e-13, "grad_norm": 0.5255570411682129, "learning_rate": 7.800252206522519e-05, "loss": 0.0182, "step": 53080 }, { "epoch": 3.6838587735843475e-13, "grad_norm": 0.45643189549446106, "learning_rate": 7.798093186100398e-05, "loss": 0.0182, "step": 53090 }, { "epoch": 3.6845526629747383e-13, "grad_norm": 0.4844566583633423, "learning_rate": 7.795934212118733e-05, "loss": 0.0176, "step": 53100 }, { "epoch": 3.685246552365129e-13, "grad_norm": 0.4763311743736267, "learning_rate": 7.793775284799392e-05, "loss": 0.0175, "step": 53110 }, { "epoch": 3.6859404417555197e-13, "grad_norm": 0.492738276720047, "learning_rate": 7.791616404364238e-05, "loss": 0.0178, "step": 53120 }, { "epoch": 3.6866343311459104e-13, "grad_norm": 0.48052728176116943, "learning_rate": 7.789457571035132e-05, "loss": 0.0183, "step": 53130 }, { "epoch": 3.687328220536301e-13, "grad_norm": 0.4795387387275696, "learning_rate": 7.787298785033925e-05, "loss": 0.0183, "step": 53140 }, { "epoch": 3.688022109926692e-13, "grad_norm": 0.4514838755130768, "learning_rate": 7.785140046582468e-05, "loss": 0.0185, "step": 53150 }, { "epoch": 3.6887159993170826e-13, "grad_norm": 0.5030040144920349, "learning_rate": 7.782981355902602e-05, "loss": 0.0185, "step": 53160 }, { "epoch": 3.6894098887074733e-13, "grad_norm": 0.4688705801963806, "learning_rate": 7.780822713216174e-05, "loss": 0.0188, "step": 53170 }, { "epoch": 3.690103778097864e-13, "grad_norm": 0.45395249128341675, "learning_rate": 7.778664118745006e-05, "loss": 0.0196, "step": 53180 }, { "epoch": 3.690797667488255e-13, "grad_norm": 0.4276584982872009, "learning_rate": 7.776505572710937e-05, "loss": 0.0178, "step": 53190 }, { "epoch": 3.6914915568786455e-13, "grad_norm": 0.4957522749900818, "learning_rate": 7.774347075335786e-05, "loss": 0.0177, "step": 53200 }, { "epoch": 3.692185446269036e-13, "grad_norm": 0.4415682554244995, "learning_rate": 7.772188626841374e-05, "loss": 0.0179, "step": 53210 }, { "epoch": 3.692879335659427e-13, "grad_norm": 0.40513524413108826, "learning_rate": 7.770030227449517e-05, "loss": 0.017, "step": 53220 }, { "epoch": 3.6935732250498177e-13, "grad_norm": 0.4684157073497772, "learning_rate": 7.767871877382022e-05, "loss": 0.0171, "step": 53230 }, { "epoch": 3.6942671144402084e-13, "grad_norm": 0.5126807689666748, "learning_rate": 7.765713576860689e-05, "loss": 0.0181, "step": 53240 }, { "epoch": 3.694961003830599e-13, "grad_norm": 0.4815499782562256, "learning_rate": 7.763555326107325e-05, "loss": 0.018, "step": 53250 }, { "epoch": 3.69565489322099e-13, "grad_norm": 0.5498165488243103, "learning_rate": 7.761397125343716e-05, "loss": 0.0176, "step": 53260 }, { "epoch": 3.6963487826113806e-13, "grad_norm": 0.49146392941474915, "learning_rate": 7.759238974791661e-05, "loss": 0.0179, "step": 53270 }, { "epoch": 3.6970426720017713e-13, "grad_norm": 0.4904935657978058, "learning_rate": 7.757080874672934e-05, "loss": 0.0173, "step": 53280 }, { "epoch": 3.697736561392162e-13, "grad_norm": 0.457709938287735, "learning_rate": 7.754922825209314e-05, "loss": 0.0179, "step": 53290 }, { "epoch": 3.6984304507825527e-13, "grad_norm": 0.4719298481941223, "learning_rate": 7.752764826622583e-05, "loss": 0.0187, "step": 53300 }, { "epoch": 3.6991243401729434e-13, "grad_norm": 0.4852328300476074, "learning_rate": 7.750606879134496e-05, "loss": 0.0178, "step": 53310 }, { "epoch": 3.699818229563334e-13, "grad_norm": 0.48794856667518616, "learning_rate": 7.748448982966827e-05, "loss": 0.0191, "step": 53320 }, { "epoch": 3.700512118953725e-13, "grad_norm": 0.5482064485549927, "learning_rate": 7.746291138341328e-05, "loss": 0.019, "step": 53330 }, { "epoch": 3.7012060083441156e-13, "grad_norm": 0.5002661347389221, "learning_rate": 7.744133345479749e-05, "loss": 0.0181, "step": 53340 }, { "epoch": 3.7018998977345063e-13, "grad_norm": 0.48537832498550415, "learning_rate": 7.741975604603842e-05, "loss": 0.0176, "step": 53350 }, { "epoch": 3.702593787124897e-13, "grad_norm": 0.5278927683830261, "learning_rate": 7.739817915935347e-05, "loss": 0.0175, "step": 53360 }, { "epoch": 3.703287676515288e-13, "grad_norm": 0.5158731937408447, "learning_rate": 7.737660279695997e-05, "loss": 0.0177, "step": 53370 }, { "epoch": 3.7039815659056785e-13, "grad_norm": 0.4633285701274872, "learning_rate": 7.735502696107527e-05, "loss": 0.0169, "step": 53380 }, { "epoch": 3.704675455296069e-13, "grad_norm": 0.4290820360183716, "learning_rate": 7.733345165391658e-05, "loss": 0.017, "step": 53390 }, { "epoch": 3.70536934468646e-13, "grad_norm": 0.5362357497215271, "learning_rate": 7.731187687770113e-05, "loss": 0.019, "step": 53400 }, { "epoch": 3.7060632340768507e-13, "grad_norm": 0.4841296076774597, "learning_rate": 7.729030263464605e-05, "loss": 0.0172, "step": 53410 }, { "epoch": 3.7067571234672414e-13, "grad_norm": 0.4184737503528595, "learning_rate": 7.726872892696845e-05, "loss": 0.0178, "step": 53420 }, { "epoch": 3.707451012857632e-13, "grad_norm": 0.4867669641971588, "learning_rate": 7.724715575688534e-05, "loss": 0.0175, "step": 53430 }, { "epoch": 3.708144902248023e-13, "grad_norm": 0.4489465653896332, "learning_rate": 7.72255831266137e-05, "loss": 0.0175, "step": 53440 }, { "epoch": 3.7088387916384136e-13, "grad_norm": 0.4241325259208679, "learning_rate": 7.720401103837048e-05, "loss": 0.0177, "step": 53450 }, { "epoch": 3.7095326810288043e-13, "grad_norm": 0.4646032154560089, "learning_rate": 7.718243949437252e-05, "loss": 0.0185, "step": 53460 }, { "epoch": 3.710226570419195e-13, "grad_norm": 0.5288805365562439, "learning_rate": 7.716086849683661e-05, "loss": 0.018, "step": 53470 }, { "epoch": 3.7109204598095857e-13, "grad_norm": 0.5332158803939819, "learning_rate": 7.713929804797956e-05, "loss": 0.0188, "step": 53480 }, { "epoch": 3.7116143491999765e-13, "grad_norm": 0.47258156538009644, "learning_rate": 7.711772815001804e-05, "loss": 0.0186, "step": 53490 }, { "epoch": 3.712308238590367e-13, "grad_norm": 0.4790956974029541, "learning_rate": 7.70961588051687e-05, "loss": 0.0182, "step": 53500 }, { "epoch": 3.713002127980758e-13, "grad_norm": 0.4995194375514984, "learning_rate": 7.707459001564812e-05, "loss": 0.0171, "step": 53510 }, { "epoch": 3.7136960173711486e-13, "grad_norm": 0.4827887713909149, "learning_rate": 7.705302178367283e-05, "loss": 0.0173, "step": 53520 }, { "epoch": 3.7143899067615394e-13, "grad_norm": 0.49167537689208984, "learning_rate": 7.70314541114593e-05, "loss": 0.0173, "step": 53530 }, { "epoch": 3.71508379615193e-13, "grad_norm": 0.47436460852622986, "learning_rate": 7.700988700122397e-05, "loss": 0.0162, "step": 53540 }, { "epoch": 3.715777685542321e-13, "grad_norm": 0.4768994152545929, "learning_rate": 7.698832045518314e-05, "loss": 0.0172, "step": 53550 }, { "epoch": 3.7164715749327115e-13, "grad_norm": 0.5126230716705322, "learning_rate": 7.696675447555318e-05, "loss": 0.0178, "step": 53560 }, { "epoch": 3.717165464323102e-13, "grad_norm": 0.458350270986557, "learning_rate": 7.694518906455025e-05, "loss": 0.0171, "step": 53570 }, { "epoch": 3.717859353713493e-13, "grad_norm": 0.4549701511859894, "learning_rate": 7.69236242243906e-05, "loss": 0.0177, "step": 53580 }, { "epoch": 3.7185532431038837e-13, "grad_norm": 0.4772760570049286, "learning_rate": 7.69020599572903e-05, "loss": 0.0174, "step": 53590 }, { "epoch": 3.7192471324942744e-13, "grad_norm": 0.44690707325935364, "learning_rate": 7.688049626546547e-05, "loss": 0.0176, "step": 53600 }, { "epoch": 3.719941021884665e-13, "grad_norm": 0.48503828048706055, "learning_rate": 7.685893315113205e-05, "loss": 0.0182, "step": 53610 }, { "epoch": 3.720634911275056e-13, "grad_norm": 0.46825411915779114, "learning_rate": 7.683737061650605e-05, "loss": 0.0176, "step": 53620 }, { "epoch": 3.7213288006654466e-13, "grad_norm": 0.5257481336593628, "learning_rate": 7.68158086638033e-05, "loss": 0.0185, "step": 53630 }, { "epoch": 3.7220226900558373e-13, "grad_norm": 0.5549166202545166, "learning_rate": 7.679424729523969e-05, "loss": 0.0187, "step": 53640 }, { "epoch": 3.722716579446228e-13, "grad_norm": 0.47904399037361145, "learning_rate": 7.677268651303091e-05, "loss": 0.0187, "step": 53650 }, { "epoch": 3.7234104688366187e-13, "grad_norm": 0.5293837189674377, "learning_rate": 7.675112631939276e-05, "loss": 0.0173, "step": 53660 }, { "epoch": 3.7241043582270095e-13, "grad_norm": 0.42937302589416504, "learning_rate": 7.672956671654076e-05, "loss": 0.0167, "step": 53670 }, { "epoch": 3.7247982476174e-13, "grad_norm": 0.40463361144065857, "learning_rate": 7.670800770669062e-05, "loss": 0.017, "step": 53680 }, { "epoch": 3.725492137007791e-13, "grad_norm": 0.46209949254989624, "learning_rate": 7.668644929205775e-05, "loss": 0.0172, "step": 53690 }, { "epoch": 3.7261860263981816e-13, "grad_norm": 0.492366760969162, "learning_rate": 7.666489147485772e-05, "loss": 0.0168, "step": 53700 }, { "epoch": 3.7268799157885724e-13, "grad_norm": 0.5173071026802063, "learning_rate": 7.664333425730584e-05, "loss": 0.0172, "step": 53710 }, { "epoch": 3.727573805178963e-13, "grad_norm": 0.5045473575592041, "learning_rate": 7.662177764161752e-05, "loss": 0.0173, "step": 53720 }, { "epoch": 3.728267694569354e-13, "grad_norm": 0.509494423866272, "learning_rate": 7.660022163000801e-05, "loss": 0.0176, "step": 53730 }, { "epoch": 3.7289615839597445e-13, "grad_norm": 0.4523133933544159, "learning_rate": 7.65786662246925e-05, "loss": 0.0178, "step": 53740 }, { "epoch": 3.7296554733501353e-13, "grad_norm": 0.5121173858642578, "learning_rate": 7.65571114278862e-05, "loss": 0.0172, "step": 53750 }, { "epoch": 3.730349362740526e-13, "grad_norm": 0.4921945035457611, "learning_rate": 7.653555724180413e-05, "loss": 0.0177, "step": 53760 }, { "epoch": 3.7310432521309167e-13, "grad_norm": 0.4929356873035431, "learning_rate": 7.65140036686614e-05, "loss": 0.0185, "step": 53770 }, { "epoch": 3.7317371415213074e-13, "grad_norm": 0.4852578043937683, "learning_rate": 7.64924507106729e-05, "loss": 0.0174, "step": 53780 }, { "epoch": 3.732431030911698e-13, "grad_norm": 0.47542211413383484, "learning_rate": 7.647089837005358e-05, "loss": 0.0189, "step": 53790 }, { "epoch": 3.733124920302089e-13, "grad_norm": 0.48522165417671204, "learning_rate": 7.644934664901823e-05, "loss": 0.019, "step": 53800 }, { "epoch": 3.7338188096924796e-13, "grad_norm": 0.4272958040237427, "learning_rate": 7.64277955497817e-05, "loss": 0.0174, "step": 53810 }, { "epoch": 3.7345126990828703e-13, "grad_norm": 0.47629231214523315, "learning_rate": 7.640624507455863e-05, "loss": 0.0172, "step": 53820 }, { "epoch": 3.735206588473261e-13, "grad_norm": 0.5200065970420837, "learning_rate": 7.63846952255637e-05, "loss": 0.0174, "step": 53830 }, { "epoch": 3.735900477863652e-13, "grad_norm": 0.4622580111026764, "learning_rate": 7.636314600501147e-05, "loss": 0.0166, "step": 53840 }, { "epoch": 3.7365943672540425e-13, "grad_norm": 0.40832066535949707, "learning_rate": 7.63415974151165e-05, "loss": 0.0164, "step": 53850 }, { "epoch": 3.737288256644433e-13, "grad_norm": 0.4880216419696808, "learning_rate": 7.632004945809318e-05, "loss": 0.0164, "step": 53860 }, { "epoch": 3.737982146034824e-13, "grad_norm": 0.46240171790122986, "learning_rate": 7.629850213615596e-05, "loss": 0.0178, "step": 53870 }, { "epoch": 3.7386760354252146e-13, "grad_norm": 0.4414355456829071, "learning_rate": 7.627695545151913e-05, "loss": 0.0168, "step": 53880 }, { "epoch": 3.7393699248156054e-13, "grad_norm": 0.44464293122291565, "learning_rate": 7.625540940639698e-05, "loss": 0.0177, "step": 53890 }, { "epoch": 3.740063814205996e-13, "grad_norm": 0.47826093435287476, "learning_rate": 7.623386400300364e-05, "loss": 0.0182, "step": 53900 }, { "epoch": 3.740757703596387e-13, "grad_norm": 0.4790838062763214, "learning_rate": 7.621231924355329e-05, "loss": 0.0178, "step": 53910 }, { "epoch": 3.7414515929867775e-13, "grad_norm": 0.5179612636566162, "learning_rate": 7.619077513025995e-05, "loss": 0.0179, "step": 53920 }, { "epoch": 3.7421454823771683e-13, "grad_norm": 0.4425625205039978, "learning_rate": 7.616923166533765e-05, "loss": 0.0179, "step": 53930 }, { "epoch": 3.742839371767559e-13, "grad_norm": 0.49023687839508057, "learning_rate": 7.614768885100028e-05, "loss": 0.0177, "step": 53940 }, { "epoch": 3.7435332611579497e-13, "grad_norm": 0.4915149211883545, "learning_rate": 7.612614668946173e-05, "loss": 0.0184, "step": 53950 }, { "epoch": 3.7442271505483404e-13, "grad_norm": 0.4762732684612274, "learning_rate": 7.610460518293578e-05, "loss": 0.0193, "step": 53960 }, { "epoch": 3.744921039938731e-13, "grad_norm": 0.4605860710144043, "learning_rate": 7.608306433363617e-05, "loss": 0.0175, "step": 53970 }, { "epoch": 3.745614929329122e-13, "grad_norm": 0.44704288244247437, "learning_rate": 7.606152414377653e-05, "loss": 0.017, "step": 53980 }, { "epoch": 3.7463088187195126e-13, "grad_norm": 0.48771169781684875, "learning_rate": 7.603998461557045e-05, "loss": 0.0173, "step": 53990 }, { "epoch": 3.7470027081099033e-13, "grad_norm": 0.42398080229759216, "learning_rate": 7.601844575123147e-05, "loss": 0.0167, "step": 54000 }, { "epoch": 3.747696597500294e-13, "grad_norm": 0.4551076292991638, "learning_rate": 7.599690755297306e-05, "loss": 0.0163, "step": 54010 }, { "epoch": 3.748390486890685e-13, "grad_norm": 0.4417365491390228, "learning_rate": 7.597537002300855e-05, "loss": 0.0169, "step": 54020 }, { "epoch": 3.7490843762810755e-13, "grad_norm": 0.43411099910736084, "learning_rate": 7.595383316355133e-05, "loss": 0.0171, "step": 54030 }, { "epoch": 3.749778265671466e-13, "grad_norm": 0.4634115397930145, "learning_rate": 7.593229697681456e-05, "loss": 0.0174, "step": 54040 }, { "epoch": 3.750472155061857e-13, "grad_norm": 0.5458539128303528, "learning_rate": 7.591076146501148e-05, "loss": 0.018, "step": 54050 }, { "epoch": 3.7511660444522477e-13, "grad_norm": 0.4371199607849121, "learning_rate": 7.58892266303552e-05, "loss": 0.0172, "step": 54060 }, { "epoch": 3.7518599338426384e-13, "grad_norm": 0.4684154987335205, "learning_rate": 7.586769247505874e-05, "loss": 0.0175, "step": 54070 }, { "epoch": 3.752553823233029e-13, "grad_norm": 0.49249622225761414, "learning_rate": 7.584615900133505e-05, "loss": 0.0183, "step": 54080 }, { "epoch": 3.75324771262342e-13, "grad_norm": 0.4803415834903717, "learning_rate": 7.582462621139707e-05, "loss": 0.0168, "step": 54090 }, { "epoch": 3.7539416020138106e-13, "grad_norm": 0.5237522721290588, "learning_rate": 7.580309410745762e-05, "loss": 0.018, "step": 54100 }, { "epoch": 3.7546354914042013e-13, "grad_norm": 0.5519092082977295, "learning_rate": 7.578156269172945e-05, "loss": 0.0183, "step": 54110 }, { "epoch": 3.755329380794592e-13, "grad_norm": 0.4618622958660126, "learning_rate": 7.576003196642521e-05, "loss": 0.0177, "step": 54120 }, { "epoch": 3.7560232701849827e-13, "grad_norm": 0.45254117250442505, "learning_rate": 7.573850193375759e-05, "loss": 0.0172, "step": 54130 }, { "epoch": 3.7567171595753734e-13, "grad_norm": 0.4838229715824127, "learning_rate": 7.571697259593908e-05, "loss": 0.0164, "step": 54140 }, { "epoch": 3.757411048965764e-13, "grad_norm": 0.5141808986663818, "learning_rate": 7.56954439551822e-05, "loss": 0.0171, "step": 54150 }, { "epoch": 3.758104938356155e-13, "grad_norm": 0.4328046441078186, "learning_rate": 7.567391601369929e-05, "loss": 0.0164, "step": 54160 }, { "epoch": 3.7587988277465456e-13, "grad_norm": 0.3865774869918823, "learning_rate": 7.565238877370277e-05, "loss": 0.0163, "step": 54170 }, { "epoch": 3.7594927171369363e-13, "grad_norm": 0.44808584451675415, "learning_rate": 7.563086223740481e-05, "loss": 0.0173, "step": 54180 }, { "epoch": 3.760186606527327e-13, "grad_norm": 0.4595172703266144, "learning_rate": 7.560933640701764e-05, "loss": 0.0168, "step": 54190 }, { "epoch": 3.760880495917718e-13, "grad_norm": 0.45157772302627563, "learning_rate": 7.558781128475339e-05, "loss": 0.0175, "step": 54200 }, { "epoch": 3.7615743853081085e-13, "grad_norm": 0.4559052586555481, "learning_rate": 7.556628687282405e-05, "loss": 0.0172, "step": 54210 }, { "epoch": 3.762268274698499e-13, "grad_norm": 0.5074588060379028, "learning_rate": 7.554476317344166e-05, "loss": 0.0175, "step": 54220 }, { "epoch": 3.76296216408889e-13, "grad_norm": 0.47070902585983276, "learning_rate": 7.552324018881803e-05, "loss": 0.0173, "step": 54230 }, { "epoch": 3.7636560534792807e-13, "grad_norm": 0.4599663019180298, "learning_rate": 7.550171792116503e-05, "loss": 0.0177, "step": 54240 }, { "epoch": 3.7643499428696714e-13, "grad_norm": 0.48055222630500793, "learning_rate": 7.54801963726944e-05, "loss": 0.0163, "step": 54250 }, { "epoch": 3.765043832260062e-13, "grad_norm": 0.42274725437164307, "learning_rate": 7.545867554561783e-05, "loss": 0.0179, "step": 54260 }, { "epoch": 3.765737721650453e-13, "grad_norm": 0.46116042137145996, "learning_rate": 7.543715544214688e-05, "loss": 0.0189, "step": 54270 }, { "epoch": 3.7664316110408436e-13, "grad_norm": 0.4969404935836792, "learning_rate": 7.541563606449314e-05, "loss": 0.0172, "step": 54280 }, { "epoch": 3.7671255004312343e-13, "grad_norm": 0.465044766664505, "learning_rate": 7.539411741486799e-05, "loss": 0.0166, "step": 54290 }, { "epoch": 3.767819389821625e-13, "grad_norm": 0.4744855463504791, "learning_rate": 7.537259949548286e-05, "loss": 0.0168, "step": 54300 }, { "epoch": 3.7685132792120157e-13, "grad_norm": 0.47110140323638916, "learning_rate": 7.535108230854902e-05, "loss": 0.0165, "step": 54310 }, { "epoch": 3.7692071686024065e-13, "grad_norm": 0.46484652161598206, "learning_rate": 7.53295658562777e-05, "loss": 0.0162, "step": 54320 }, { "epoch": 3.769901057992797e-13, "grad_norm": 0.5110781788825989, "learning_rate": 7.530805014088006e-05, "loss": 0.0167, "step": 54330 }, { "epoch": 3.770594947383188e-13, "grad_norm": 0.4323353171348572, "learning_rate": 7.52865351645672e-05, "loss": 0.0167, "step": 54340 }, { "epoch": 3.7712888367735786e-13, "grad_norm": 0.4610573649406433, "learning_rate": 7.526502092955004e-05, "loss": 0.0164, "step": 54350 }, { "epoch": 3.7719827261639693e-13, "grad_norm": 0.4518487751483917, "learning_rate": 7.524350743803957e-05, "loss": 0.0165, "step": 54360 }, { "epoch": 3.77267661555436e-13, "grad_norm": 0.49619174003601074, "learning_rate": 7.522199469224661e-05, "loss": 0.0175, "step": 54370 }, { "epoch": 3.773370504944751e-13, "grad_norm": 0.4854128658771515, "learning_rate": 7.520048269438195e-05, "loss": 0.017, "step": 54380 }, { "epoch": 3.7740643943351415e-13, "grad_norm": 0.4473860263824463, "learning_rate": 7.517897144665625e-05, "loss": 0.0176, "step": 54390 }, { "epoch": 3.774758283725532e-13, "grad_norm": 0.44583553075790405, "learning_rate": 7.515746095128017e-05, "loss": 0.0174, "step": 54400 }, { "epoch": 3.775452173115923e-13, "grad_norm": 0.48375439643859863, "learning_rate": 7.51359512104642e-05, "loss": 0.0174, "step": 54410 }, { "epoch": 3.7761460625063137e-13, "grad_norm": 0.44383472204208374, "learning_rate": 7.511444222641884e-05, "loss": 0.0177, "step": 54420 }, { "epoch": 3.7768399518967044e-13, "grad_norm": 0.4263160228729248, "learning_rate": 7.509293400135444e-05, "loss": 0.0185, "step": 54430 }, { "epoch": 3.777533841287095e-13, "grad_norm": 0.39835724234580994, "learning_rate": 7.507142653748135e-05, "loss": 0.0165, "step": 54440 }, { "epoch": 3.778227730677486e-13, "grad_norm": 0.4924413561820984, "learning_rate": 7.504991983700971e-05, "loss": 0.0167, "step": 54450 }, { "epoch": 3.7789216200678766e-13, "grad_norm": 0.41791361570358276, "learning_rate": 7.502841390214978e-05, "loss": 0.0169, "step": 54460 }, { "epoch": 3.7796155094582673e-13, "grad_norm": 0.4202394485473633, "learning_rate": 7.500690873511154e-05, "loss": 0.0164, "step": 54470 }, { "epoch": 3.780309398848658e-13, "grad_norm": 0.4576888680458069, "learning_rate": 7.498540433810505e-05, "loss": 0.0162, "step": 54480 }, { "epoch": 3.7810032882390487e-13, "grad_norm": 0.4607239365577698, "learning_rate": 7.496390071334013e-05, "loss": 0.0165, "step": 54490 }, { "epoch": 3.7816971776294395e-13, "grad_norm": 0.4562227427959442, "learning_rate": 7.494239786302672e-05, "loss": 0.0165, "step": 54500 }, { "epoch": 3.78239106701983e-13, "grad_norm": 0.5203714370727539, "learning_rate": 7.49208957893745e-05, "loss": 0.0162, "step": 54510 }, { "epoch": 3.783084956410221e-13, "grad_norm": 0.46112945675849915, "learning_rate": 7.489939449459318e-05, "loss": 0.0169, "step": 54520 }, { "epoch": 3.7837788458006116e-13, "grad_norm": 0.4737503230571747, "learning_rate": 7.48778939808923e-05, "loss": 0.0168, "step": 54530 }, { "epoch": 3.7844727351910024e-13, "grad_norm": 0.47005268931388855, "learning_rate": 7.485639425048144e-05, "loss": 0.017, "step": 54540 }, { "epoch": 3.785166624581393e-13, "grad_norm": 0.4464413821697235, "learning_rate": 7.483489530556999e-05, "loss": 0.0175, "step": 54550 }, { "epoch": 3.785860513971784e-13, "grad_norm": 0.4453490674495697, "learning_rate": 7.481339714836731e-05, "loss": 0.0165, "step": 54560 }, { "epoch": 3.7865544033621745e-13, "grad_norm": 0.4774115979671478, "learning_rate": 7.479189978108266e-05, "loss": 0.0176, "step": 54570 }, { "epoch": 3.787248292752565e-13, "grad_norm": 0.5226303339004517, "learning_rate": 7.477040320592526e-05, "loss": 0.0177, "step": 54580 }, { "epoch": 3.787942182142956e-13, "grad_norm": 0.4782789945602417, "learning_rate": 7.474890742510418e-05, "loss": 0.0171, "step": 54590 }, { "epoch": 3.7886360715333467e-13, "grad_norm": 0.4462185502052307, "learning_rate": 7.47274124408285e-05, "loss": 0.0161, "step": 54600 }, { "epoch": 3.7893299609237374e-13, "grad_norm": 0.516609787940979, "learning_rate": 7.470591825530709e-05, "loss": 0.0169, "step": 54610 }, { "epoch": 3.790023850314128e-13, "grad_norm": 0.48930057883262634, "learning_rate": 7.468442487074886e-05, "loss": 0.0169, "step": 54620 }, { "epoch": 3.790717739704519e-13, "grad_norm": 0.44771629571914673, "learning_rate": 7.466293228936261e-05, "loss": 0.0163, "step": 54630 }, { "epoch": 3.7914116290949096e-13, "grad_norm": 0.4129778742790222, "learning_rate": 7.464144051335696e-05, "loss": 0.0159, "step": 54640 }, { "epoch": 3.7921055184853003e-13, "grad_norm": 0.5090532302856445, "learning_rate": 7.461994954494059e-05, "loss": 0.0167, "step": 54650 }, { "epoch": 3.792799407875691e-13, "grad_norm": 0.47681403160095215, "learning_rate": 7.459845938632201e-05, "loss": 0.016, "step": 54660 }, { "epoch": 3.793493297266082e-13, "grad_norm": 0.39814481139183044, "learning_rate": 7.457697003970969e-05, "loss": 0.0161, "step": 54670 }, { "epoch": 3.7941871866564725e-13, "grad_norm": 0.4842764437198639, "learning_rate": 7.455548150731193e-05, "loss": 0.0168, "step": 54680 }, { "epoch": 3.794881076046863e-13, "grad_norm": 0.42961880564689636, "learning_rate": 7.45339937913371e-05, "loss": 0.0169, "step": 54690 }, { "epoch": 3.795574965437254e-13, "grad_norm": 0.4480929374694824, "learning_rate": 7.451250689399333e-05, "loss": 0.0175, "step": 54700 }, { "epoch": 3.7962688548276446e-13, "grad_norm": 0.46269935369491577, "learning_rate": 7.449102081748877e-05, "loss": 0.0178, "step": 54710 }, { "epoch": 3.7969627442180354e-13, "grad_norm": 0.5089104175567627, "learning_rate": 7.446953556403142e-05, "loss": 0.0168, "step": 54720 }, { "epoch": 3.797656633608426e-13, "grad_norm": 0.49539658427238464, "learning_rate": 7.444805113582926e-05, "loss": 0.0172, "step": 54730 }, { "epoch": 3.798350522998817e-13, "grad_norm": 0.47205987572669983, "learning_rate": 7.442656753509013e-05, "loss": 0.0175, "step": 54740 }, { "epoch": 3.7990444123892075e-13, "grad_norm": 0.462840735912323, "learning_rate": 7.440508476402176e-05, "loss": 0.0167, "step": 54750 }, { "epoch": 3.7997383017795983e-13, "grad_norm": 0.4828065037727356, "learning_rate": 7.438360282483192e-05, "loss": 0.0164, "step": 54760 }, { "epoch": 3.800432191169989e-13, "grad_norm": 0.48826566338539124, "learning_rate": 7.436212171972819e-05, "loss": 0.0166, "step": 54770 }, { "epoch": 3.8011260805603797e-13, "grad_norm": 0.4972902536392212, "learning_rate": 7.434064145091803e-05, "loss": 0.0167, "step": 54780 }, { "epoch": 3.8018199699507704e-13, "grad_norm": 0.4871586263179779, "learning_rate": 7.431916202060894e-05, "loss": 0.0157, "step": 54790 }, { "epoch": 3.802513859341161e-13, "grad_norm": 0.450539767742157, "learning_rate": 7.429768343100822e-05, "loss": 0.0162, "step": 54800 }, { "epoch": 3.803207748731552e-13, "grad_norm": 0.48765721917152405, "learning_rate": 7.427620568432319e-05, "loss": 0.0164, "step": 54810 }, { "epoch": 3.8039016381219426e-13, "grad_norm": 0.45028194785118103, "learning_rate": 7.425472878276094e-05, "loss": 0.0158, "step": 54820 }, { "epoch": 3.8045955275123333e-13, "grad_norm": 0.44141682982444763, "learning_rate": 7.423325272852862e-05, "loss": 0.0161, "step": 54830 }, { "epoch": 3.805289416902724e-13, "grad_norm": 0.49336788058280945, "learning_rate": 7.42117775238332e-05, "loss": 0.0167, "step": 54840 }, { "epoch": 3.805983306293115e-13, "grad_norm": 0.47102081775665283, "learning_rate": 7.419030317088162e-05, "loss": 0.017, "step": 54850 }, { "epoch": 3.8066771956835055e-13, "grad_norm": 0.47963792085647583, "learning_rate": 7.416882967188065e-05, "loss": 0.0173, "step": 54860 }, { "epoch": 3.807371085073896e-13, "grad_norm": 0.44044041633605957, "learning_rate": 7.414735702903708e-05, "loss": 0.0168, "step": 54870 }, { "epoch": 3.808064974464287e-13, "grad_norm": 0.4962373375892639, "learning_rate": 7.412588524455753e-05, "loss": 0.0172, "step": 54880 }, { "epoch": 3.8087588638546777e-13, "grad_norm": 0.5070865154266357, "learning_rate": 7.410441432064857e-05, "loss": 0.0171, "step": 54890 }, { "epoch": 3.8094527532450684e-13, "grad_norm": 0.4612414836883545, "learning_rate": 7.408294425951665e-05, "loss": 0.017, "step": 54900 }, { "epoch": 3.810146642635459e-13, "grad_norm": 0.5206678509712219, "learning_rate": 7.40614750633682e-05, "loss": 0.0162, "step": 54910 }, { "epoch": 3.81084053202585e-13, "grad_norm": 0.4268079698085785, "learning_rate": 7.404000673440945e-05, "loss": 0.0159, "step": 54920 }, { "epoch": 3.8115344214162405e-13, "grad_norm": 0.3969945013523102, "learning_rate": 7.401853927484668e-05, "loss": 0.0164, "step": 54930 }, { "epoch": 3.8122283108066313e-13, "grad_norm": 0.4357655644416809, "learning_rate": 7.399707268688593e-05, "loss": 0.0163, "step": 54940 }, { "epoch": 3.812922200197022e-13, "grad_norm": 0.4530888497829437, "learning_rate": 7.397560697273329e-05, "loss": 0.016, "step": 54950 }, { "epoch": 3.8136160895874127e-13, "grad_norm": 0.4664982557296753, "learning_rate": 7.395414213459465e-05, "loss": 0.0159, "step": 54960 }, { "epoch": 3.8143099789778034e-13, "grad_norm": 0.47281551361083984, "learning_rate": 7.393267817467589e-05, "loss": 0.0156, "step": 54970 }, { "epoch": 3.815003868368194e-13, "grad_norm": 0.4566817581653595, "learning_rate": 7.391121509518272e-05, "loss": 0.016, "step": 54980 }, { "epoch": 3.815697757758585e-13, "grad_norm": 0.4373587965965271, "learning_rate": 7.388975289832086e-05, "loss": 0.0161, "step": 54990 }, { "epoch": 3.8163916471489756e-13, "grad_norm": 0.470460444688797, "learning_rate": 7.386829158629583e-05, "loss": 0.0159, "step": 55000 }, { "epoch": 3.8170855365393663e-13, "grad_norm": 0.46771344542503357, "learning_rate": 7.384683116131318e-05, "loss": 0.0165, "step": 55010 }, { "epoch": 3.817779425929757e-13, "grad_norm": 0.48092642426490784, "learning_rate": 7.382537162557823e-05, "loss": 0.017, "step": 55020 }, { "epoch": 3.818473315320148e-13, "grad_norm": 0.46194300055503845, "learning_rate": 7.380391298129634e-05, "loss": 0.0158, "step": 55030 }, { "epoch": 3.8191672047105385e-13, "grad_norm": 0.4518159031867981, "learning_rate": 7.378245523067268e-05, "loss": 0.017, "step": 55040 }, { "epoch": 3.819861094100929e-13, "grad_norm": 0.44070160388946533, "learning_rate": 7.376099837591242e-05, "loss": 0.017, "step": 55050 }, { "epoch": 3.82055498349132e-13, "grad_norm": 0.4044795632362366, "learning_rate": 7.373954241922052e-05, "loss": 0.016, "step": 55060 }, { "epoch": 3.8212488728817107e-13, "grad_norm": 0.44738152623176575, "learning_rate": 7.371808736280196e-05, "loss": 0.0156, "step": 55070 }, { "epoch": 3.8219427622721014e-13, "grad_norm": 0.4970778822898865, "learning_rate": 7.369663320886153e-05, "loss": 0.0161, "step": 55080 }, { "epoch": 3.822636651662492e-13, "grad_norm": 0.45646458864212036, "learning_rate": 7.367517995960406e-05, "loss": 0.0158, "step": 55090 }, { "epoch": 3.823330541052883e-13, "grad_norm": 0.41835954785346985, "learning_rate": 7.365372761723415e-05, "loss": 0.0159, "step": 55100 }, { "epoch": 3.8240244304432736e-13, "grad_norm": 0.4750367999076843, "learning_rate": 7.363227618395632e-05, "loss": 0.0159, "step": 55110 }, { "epoch": 3.8247183198336643e-13, "grad_norm": 0.4189508259296417, "learning_rate": 7.361082566197513e-05, "loss": 0.0166, "step": 55120 }, { "epoch": 3.825412209224055e-13, "grad_norm": 0.4362517297267914, "learning_rate": 7.358937605349489e-05, "loss": 0.0151, "step": 55130 }, { "epoch": 3.8261060986144457e-13, "grad_norm": 0.4048622250556946, "learning_rate": 7.356792736071994e-05, "loss": 0.0155, "step": 55140 }, { "epoch": 3.8267999880048365e-13, "grad_norm": 0.4044731557369232, "learning_rate": 7.354647958585439e-05, "loss": 0.0158, "step": 55150 }, { "epoch": 3.827493877395227e-13, "grad_norm": 0.4523946940898895, "learning_rate": 7.35250327311024e-05, "loss": 0.0165, "step": 55160 }, { "epoch": 3.828187766785618e-13, "grad_norm": 0.5152435302734375, "learning_rate": 7.350358679866792e-05, "loss": 0.0167, "step": 55170 }, { "epoch": 3.8288816561760086e-13, "grad_norm": 0.430132657289505, "learning_rate": 7.348214179075489e-05, "loss": 0.0164, "step": 55180 }, { "epoch": 3.8295755455663993e-13, "grad_norm": 0.4488770365715027, "learning_rate": 7.34606977095671e-05, "loss": 0.0165, "step": 55190 }, { "epoch": 3.83026943495679e-13, "grad_norm": 0.4458865523338318, "learning_rate": 7.343925455730827e-05, "loss": 0.017, "step": 55200 }, { "epoch": 3.830963324347181e-13, "grad_norm": 0.4571939706802368, "learning_rate": 7.341781233618198e-05, "loss": 0.0172, "step": 55210 }, { "epoch": 3.8316572137375715e-13, "grad_norm": 0.41893407702445984, "learning_rate": 7.339637104839184e-05, "loss": 0.0163, "step": 55220 }, { "epoch": 3.832351103127962e-13, "grad_norm": 0.4542102515697479, "learning_rate": 7.337493069614117e-05, "loss": 0.0159, "step": 55230 }, { "epoch": 3.833044992518353e-13, "grad_norm": 0.45297324657440186, "learning_rate": 7.335349128163339e-05, "loss": 0.0163, "step": 55240 }, { "epoch": 3.8337388819087437e-13, "grad_norm": 0.4180162847042084, "learning_rate": 7.333205280707166e-05, "loss": 0.0156, "step": 55250 }, { "epoch": 3.8344327712991344e-13, "grad_norm": 0.44924241304397583, "learning_rate": 7.331061527465918e-05, "loss": 0.0158, "step": 55260 }, { "epoch": 3.835126660689525e-13, "grad_norm": 0.42139488458633423, "learning_rate": 7.328917868659895e-05, "loss": 0.016, "step": 55270 }, { "epoch": 3.835820550079916e-13, "grad_norm": 0.4090135991573334, "learning_rate": 7.326774304509392e-05, "loss": 0.0157, "step": 55280 }, { "epoch": 3.8365144394703066e-13, "grad_norm": 0.4120146930217743, "learning_rate": 7.324630835234695e-05, "loss": 0.0151, "step": 55290 }, { "epoch": 3.8372083288606973e-13, "grad_norm": 0.4951382875442505, "learning_rate": 7.322487461056079e-05, "loss": 0.0158, "step": 55300 }, { "epoch": 3.837902218251088e-13, "grad_norm": 0.4428543448448181, "learning_rate": 7.320344182193803e-05, "loss": 0.0155, "step": 55310 }, { "epoch": 3.8385961076414787e-13, "grad_norm": 0.47422170639038086, "learning_rate": 7.318200998868133e-05, "loss": 0.0165, "step": 55320 }, { "epoch": 3.8392899970318695e-13, "grad_norm": 0.46884724497795105, "learning_rate": 7.316057911299302e-05, "loss": 0.0171, "step": 55330 }, { "epoch": 3.83998388642226e-13, "grad_norm": 0.4679160714149475, "learning_rate": 7.313914919707558e-05, "loss": 0.0159, "step": 55340 }, { "epoch": 3.840677775812651e-13, "grad_norm": 0.5017527341842651, "learning_rate": 7.311772024313117e-05, "loss": 0.0173, "step": 55350 }, { "epoch": 3.8413716652030416e-13, "grad_norm": 0.501653254032135, "learning_rate": 7.309629225336202e-05, "loss": 0.0167, "step": 55360 }, { "epoch": 3.8420655545934324e-13, "grad_norm": 0.45739880204200745, "learning_rate": 7.307486522997012e-05, "loss": 0.0164, "step": 55370 }, { "epoch": 3.842759443983823e-13, "grad_norm": 0.4282846450805664, "learning_rate": 7.305343917515748e-05, "loss": 0.0156, "step": 55380 }, { "epoch": 3.843453333374214e-13, "grad_norm": 0.4579041004180908, "learning_rate": 7.303201409112596e-05, "loss": 0.0153, "step": 55390 }, { "epoch": 3.8441472227646045e-13, "grad_norm": 0.49165114760398865, "learning_rate": 7.301058998007732e-05, "loss": 0.0161, "step": 55400 }, { "epoch": 3.844841112154995e-13, "grad_norm": 0.44104763865470886, "learning_rate": 7.298916684421317e-05, "loss": 0.0156, "step": 55410 }, { "epoch": 3.845535001545386e-13, "grad_norm": 0.36843207478523254, "learning_rate": 7.296774468573512e-05, "loss": 0.0158, "step": 55420 }, { "epoch": 3.8462288909357767e-13, "grad_norm": 0.4301425814628601, "learning_rate": 7.294632350684462e-05, "loss": 0.016, "step": 55430 }, { "epoch": 3.8469227803261674e-13, "grad_norm": 0.45360827445983887, "learning_rate": 7.292490330974305e-05, "loss": 0.0155, "step": 55440 }, { "epoch": 3.847616669716558e-13, "grad_norm": 0.4418373107910156, "learning_rate": 7.290348409663161e-05, "loss": 0.0155, "step": 55450 }, { "epoch": 3.848310559106949e-13, "grad_norm": 0.43493449687957764, "learning_rate": 7.288206586971152e-05, "loss": 0.0153, "step": 55460 }, { "epoch": 3.8490044484973396e-13, "grad_norm": 0.45937347412109375, "learning_rate": 7.286064863118379e-05, "loss": 0.016, "step": 55470 }, { "epoch": 3.8496983378877303e-13, "grad_norm": 0.5047304630279541, "learning_rate": 7.283923238324944e-05, "loss": 0.0162, "step": 55480 }, { "epoch": 3.850392227278121e-13, "grad_norm": 0.4471980929374695, "learning_rate": 7.281781712810924e-05, "loss": 0.0165, "step": 55490 }, { "epoch": 3.851086116668512e-13, "grad_norm": 0.4794996976852417, "learning_rate": 7.279640286796398e-05, "loss": 0.016, "step": 55500 }, { "epoch": 3.8517800060589025e-13, "grad_norm": 0.46587052941322327, "learning_rate": 7.277498960501431e-05, "loss": 0.0169, "step": 55510 }, { "epoch": 3.852473895449293e-13, "grad_norm": 0.4203658401966095, "learning_rate": 7.275357734146078e-05, "loss": 0.0169, "step": 55520 }, { "epoch": 3.853167784839684e-13, "grad_norm": 0.4553566575050354, "learning_rate": 7.273216607950381e-05, "loss": 0.0157, "step": 55530 }, { "epoch": 3.8538616742300746e-13, "grad_norm": 0.4808772802352905, "learning_rate": 7.271075582134377e-05, "loss": 0.0153, "step": 55540 }, { "epoch": 3.8545555636204654e-13, "grad_norm": 0.44745397567749023, "learning_rate": 7.268934656918091e-05, "loss": 0.0158, "step": 55550 }, { "epoch": 3.855249453010856e-13, "grad_norm": 0.4412488639354706, "learning_rate": 7.26679383252153e-05, "loss": 0.0155, "step": 55560 }, { "epoch": 3.855943342401247e-13, "grad_norm": 0.44928503036499023, "learning_rate": 7.264653109164701e-05, "loss": 0.0154, "step": 55570 }, { "epoch": 3.8566372317916375e-13, "grad_norm": 0.4568135440349579, "learning_rate": 7.262512487067595e-05, "loss": 0.0155, "step": 55580 }, { "epoch": 3.8573311211820283e-13, "grad_norm": 0.4164244830608368, "learning_rate": 7.2603719664502e-05, "loss": 0.0157, "step": 55590 }, { "epoch": 3.858025010572419e-13, "grad_norm": 0.42011743783950806, "learning_rate": 7.258231547532478e-05, "loss": 0.0148, "step": 55600 }, { "epoch": 3.8587188999628097e-13, "grad_norm": 0.4663582146167755, "learning_rate": 7.2560912305344e-05, "loss": 0.0152, "step": 55610 }, { "epoch": 3.8594127893532004e-13, "grad_norm": 0.4266122579574585, "learning_rate": 7.253951015675911e-05, "loss": 0.0153, "step": 55620 }, { "epoch": 3.860106678743591e-13, "grad_norm": 0.4522101879119873, "learning_rate": 7.251810903176955e-05, "loss": 0.0158, "step": 55630 }, { "epoch": 3.860800568133982e-13, "grad_norm": 0.4641442894935608, "learning_rate": 7.249670893257456e-05, "loss": 0.0158, "step": 55640 }, { "epoch": 3.8614944575243726e-13, "grad_norm": 0.4337977170944214, "learning_rate": 7.24753098613734e-05, "loss": 0.0158, "step": 55650 }, { "epoch": 3.8621883469147633e-13, "grad_norm": 0.4783169627189636, "learning_rate": 7.245391182036511e-05, "loss": 0.0163, "step": 55660 }, { "epoch": 3.862882236305154e-13, "grad_norm": 0.42656323313713074, "learning_rate": 7.24325148117487e-05, "loss": 0.0161, "step": 55670 }, { "epoch": 3.863576125695545e-13, "grad_norm": 0.40153542160987854, "learning_rate": 7.241111883772304e-05, "loss": 0.0166, "step": 55680 }, { "epoch": 3.8642700150859355e-13, "grad_norm": 0.38213324546813965, "learning_rate": 7.238972390048689e-05, "loss": 0.0155, "step": 55690 }, { "epoch": 3.864963904476326e-13, "grad_norm": 0.4665980637073517, "learning_rate": 7.236833000223891e-05, "loss": 0.0155, "step": 55700 }, { "epoch": 3.865657793866717e-13, "grad_norm": 0.387023389339447, "learning_rate": 7.234693714517769e-05, "loss": 0.0155, "step": 55710 }, { "epoch": 3.8663516832571077e-13, "grad_norm": 0.4155510365962982, "learning_rate": 7.232554533150163e-05, "loss": 0.0153, "step": 55720 }, { "epoch": 3.8670455726474984e-13, "grad_norm": 0.434192031621933, "learning_rate": 7.230415456340913e-05, "loss": 0.0154, "step": 55730 }, { "epoch": 3.867739462037889e-13, "grad_norm": 0.46447399258613586, "learning_rate": 7.228276484309833e-05, "loss": 0.0156, "step": 55740 }, { "epoch": 3.86843335142828e-13, "grad_norm": 0.43449845910072327, "learning_rate": 7.226137617276746e-05, "loss": 0.0157, "step": 55750 }, { "epoch": 3.8691272408186705e-13, "grad_norm": 0.4747503697872162, "learning_rate": 7.223998855461445e-05, "loss": 0.0154, "step": 55760 }, { "epoch": 3.8698211302090613e-13, "grad_norm": 0.4493538737297058, "learning_rate": 7.221860199083731e-05, "loss": 0.0156, "step": 55770 }, { "epoch": 3.870515019599452e-13, "grad_norm": 0.45785173773765564, "learning_rate": 7.219721648363375e-05, "loss": 0.0153, "step": 55780 }, { "epoch": 3.8712089089898427e-13, "grad_norm": 0.4279797673225403, "learning_rate": 7.217583203520153e-05, "loss": 0.0162, "step": 55790 }, { "epoch": 3.8719027983802334e-13, "grad_norm": 0.440555214881897, "learning_rate": 7.215444864773816e-05, "loss": 0.0163, "step": 55800 }, { "epoch": 3.872596687770624e-13, "grad_norm": 0.4251851439476013, "learning_rate": 7.21330663234412e-05, "loss": 0.0157, "step": 55810 }, { "epoch": 3.873290577161015e-13, "grad_norm": 0.45445647835731506, "learning_rate": 7.211168506450797e-05, "loss": 0.0171, "step": 55820 }, { "epoch": 3.8739844665514056e-13, "grad_norm": 0.4893704950809479, "learning_rate": 7.209030487313574e-05, "loss": 0.0165, "step": 55830 }, { "epoch": 3.8746783559417963e-13, "grad_norm": 0.44830262660980225, "learning_rate": 7.206892575152163e-05, "loss": 0.0159, "step": 55840 }, { "epoch": 3.875372245332187e-13, "grad_norm": 0.43196582794189453, "learning_rate": 7.204754770186273e-05, "loss": 0.0152, "step": 55850 }, { "epoch": 3.876066134722578e-13, "grad_norm": 0.48839059472084045, "learning_rate": 7.202617072635591e-05, "loss": 0.0152, "step": 55860 }, { "epoch": 3.8767600241129685e-13, "grad_norm": 0.44684478640556335, "learning_rate": 7.200479482719802e-05, "loss": 0.0152, "step": 55870 }, { "epoch": 3.877453913503359e-13, "grad_norm": 0.45685213804244995, "learning_rate": 7.198342000658576e-05, "loss": 0.015, "step": 55880 }, { "epoch": 3.87814780289375e-13, "grad_norm": 0.3987896144390106, "learning_rate": 7.196204626671573e-05, "loss": 0.0152, "step": 55890 }, { "epoch": 3.8788416922841407e-13, "grad_norm": 0.48840799927711487, "learning_rate": 7.19406736097844e-05, "loss": 0.0162, "step": 55900 }, { "epoch": 3.8795355816745314e-13, "grad_norm": 0.44337666034698486, "learning_rate": 7.191930203798818e-05, "loss": 0.0151, "step": 55910 }, { "epoch": 3.880229471064922e-13, "grad_norm": 0.3859133720397949, "learning_rate": 7.189793155352326e-05, "loss": 0.0149, "step": 55920 }, { "epoch": 3.880923360455313e-13, "grad_norm": 0.43542107939720154, "learning_rate": 7.187656215858588e-05, "loss": 0.0157, "step": 55930 }, { "epoch": 3.8816172498457036e-13, "grad_norm": 0.4336422085762024, "learning_rate": 7.185519385537199e-05, "loss": 0.0155, "step": 55940 }, { "epoch": 3.8823111392360943e-13, "grad_norm": 0.40504786372184753, "learning_rate": 7.18338266460776e-05, "loss": 0.0156, "step": 55950 }, { "epoch": 3.883005028626485e-13, "grad_norm": 0.4182085692882538, "learning_rate": 7.181246053289845e-05, "loss": 0.0165, "step": 55960 }, { "epoch": 3.8836989180168757e-13, "grad_norm": 0.47633999586105347, "learning_rate": 7.179109551803028e-05, "loss": 0.0158, "step": 55970 }, { "epoch": 3.8843928074072664e-13, "grad_norm": 0.4960346519947052, "learning_rate": 7.176973160366866e-05, "loss": 0.0162, "step": 55980 }, { "epoch": 3.885086696797657e-13, "grad_norm": 0.4496540129184723, "learning_rate": 7.174836879200909e-05, "loss": 0.0165, "step": 55990 }, { "epoch": 3.885780586188048e-13, "grad_norm": 0.4459231495857239, "learning_rate": 7.172700708524691e-05, "loss": 0.0156, "step": 56000 }, { "epoch": 3.8864744755784386e-13, "grad_norm": 0.43527787923812866, "learning_rate": 7.170564648557736e-05, "loss": 0.0152, "step": 56010 }, { "epoch": 3.8871683649688293e-13, "grad_norm": 0.4528762102127075, "learning_rate": 7.16842869951956e-05, "loss": 0.0153, "step": 56020 }, { "epoch": 3.88786225435922e-13, "grad_norm": 0.4548131227493286, "learning_rate": 7.166292861629662e-05, "loss": 0.0151, "step": 56030 }, { "epoch": 3.888556143749611e-13, "grad_norm": 0.43560269474983215, "learning_rate": 7.164157135107537e-05, "loss": 0.0146, "step": 56040 }, { "epoch": 3.8892500331400015e-13, "grad_norm": 0.4347342848777771, "learning_rate": 7.162021520172659e-05, "loss": 0.0152, "step": 56050 }, { "epoch": 3.889943922530392e-13, "grad_norm": 0.45726901292800903, "learning_rate": 7.159886017044498e-05, "loss": 0.0152, "step": 56060 }, { "epoch": 3.890637811920783e-13, "grad_norm": 0.40003442764282227, "learning_rate": 7.15775062594251e-05, "loss": 0.0149, "step": 56070 }, { "epoch": 3.8913317013111737e-13, "grad_norm": 0.41750699281692505, "learning_rate": 7.155615347086143e-05, "loss": 0.0151, "step": 56080 }, { "epoch": 3.8920255907015644e-13, "grad_norm": 0.42072537541389465, "learning_rate": 7.153480180694823e-05, "loss": 0.0152, "step": 56090 }, { "epoch": 3.892719480091955e-13, "grad_norm": 0.4072836637496948, "learning_rate": 7.151345126987977e-05, "loss": 0.0158, "step": 56100 }, { "epoch": 3.893413369482346e-13, "grad_norm": 0.45844894647598267, "learning_rate": 7.149210186185013e-05, "loss": 0.0157, "step": 56110 }, { "epoch": 3.8941072588727366e-13, "grad_norm": 0.4288014769554138, "learning_rate": 7.14707535850533e-05, "loss": 0.0155, "step": 56120 }, { "epoch": 3.8948011482631273e-13, "grad_norm": 0.4775060713291168, "learning_rate": 7.144940644168315e-05, "loss": 0.0164, "step": 56130 }, { "epoch": 3.895495037653518e-13, "grad_norm": 0.49656859040260315, "learning_rate": 7.142806043393345e-05, "loss": 0.0159, "step": 56140 }, { "epoch": 3.8961889270439087e-13, "grad_norm": 0.45077407360076904, "learning_rate": 7.140671556399779e-05, "loss": 0.0162, "step": 56150 }, { "epoch": 3.8968828164342995e-13, "grad_norm": 0.48960015177726746, "learning_rate": 7.138537183406969e-05, "loss": 0.0152, "step": 56160 }, { "epoch": 3.89757670582469e-13, "grad_norm": 0.4169079661369324, "learning_rate": 7.13640292463426e-05, "loss": 0.0147, "step": 56170 }, { "epoch": 3.898270595215081e-13, "grad_norm": 0.4019482433795929, "learning_rate": 7.134268780300977e-05, "loss": 0.0155, "step": 56180 }, { "epoch": 3.8989644846054716e-13, "grad_norm": 0.4450223743915558, "learning_rate": 7.132134750626436e-05, "loss": 0.0151, "step": 56190 }, { "epoch": 3.8996583739958623e-13, "grad_norm": 0.46733954548835754, "learning_rate": 7.130000835829944e-05, "loss": 0.0153, "step": 56200 }, { "epoch": 3.900352263386253e-13, "grad_norm": 0.4565531015396118, "learning_rate": 7.127867036130789e-05, "loss": 0.0153, "step": 56210 }, { "epoch": 3.901046152776644e-13, "grad_norm": 0.4428942799568176, "learning_rate": 7.12573335174826e-05, "loss": 0.015, "step": 56220 }, { "epoch": 3.9017400421670345e-13, "grad_norm": 0.4647255539894104, "learning_rate": 7.123599782901617e-05, "loss": 0.015, "step": 56230 }, { "epoch": 3.902433931557425e-13, "grad_norm": 0.41901347041130066, "learning_rate": 7.121466329810127e-05, "loss": 0.015, "step": 56240 }, { "epoch": 3.903127820947816e-13, "grad_norm": 0.4670256972312927, "learning_rate": 7.119332992693029e-05, "loss": 0.0155, "step": 56250 }, { "epoch": 3.9038217103382067e-13, "grad_norm": 0.44787096977233887, "learning_rate": 7.117199771769558e-05, "loss": 0.0155, "step": 56260 }, { "epoch": 3.9045155997285974e-13, "grad_norm": 0.4428877532482147, "learning_rate": 7.115066667258937e-05, "loss": 0.0161, "step": 56270 }, { "epoch": 3.905209489118988e-13, "grad_norm": 0.4336393475532532, "learning_rate": 7.112933679380376e-05, "loss": 0.0154, "step": 56280 }, { "epoch": 3.905903378509379e-13, "grad_norm": 0.4153941571712494, "learning_rate": 7.110800808353068e-05, "loss": 0.0162, "step": 56290 }, { "epoch": 3.9065972678997696e-13, "grad_norm": 0.4313300848007202, "learning_rate": 7.108668054396205e-05, "loss": 0.0164, "step": 56300 }, { "epoch": 3.9072911572901603e-13, "grad_norm": 0.3811050057411194, "learning_rate": 7.106535417728956e-05, "loss": 0.015, "step": 56310 }, { "epoch": 3.907985046680551e-13, "grad_norm": 0.43672922253608704, "learning_rate": 7.104402898570485e-05, "loss": 0.015, "step": 56320 }, { "epoch": 3.9086789360709417e-13, "grad_norm": 0.49879372119903564, "learning_rate": 7.102270497139938e-05, "loss": 0.0152, "step": 56330 }, { "epoch": 3.9093728254613325e-13, "grad_norm": 0.4296831488609314, "learning_rate": 7.10013821365646e-05, "loss": 0.0148, "step": 56340 }, { "epoch": 3.910066714851723e-13, "grad_norm": 0.40705427527427673, "learning_rate": 7.09800604833917e-05, "loss": 0.0147, "step": 56350 }, { "epoch": 3.910760604242114e-13, "grad_norm": 0.4687160551548004, "learning_rate": 7.095874001407182e-05, "loss": 0.015, "step": 56360 }, { "epoch": 3.9114544936325046e-13, "grad_norm": 0.4048539996147156, "learning_rate": 7.093742073079599e-05, "loss": 0.0153, "step": 56370 }, { "epoch": 3.9121483830228954e-13, "grad_norm": 0.41617441177368164, "learning_rate": 7.091610263575508e-05, "loss": 0.0143, "step": 56380 }, { "epoch": 3.912842272413286e-13, "grad_norm": 0.4182735085487366, "learning_rate": 7.089478573113984e-05, "loss": 0.0147, "step": 56390 }, { "epoch": 3.913536161803677e-13, "grad_norm": 0.4363378584384918, "learning_rate": 7.087347001914098e-05, "loss": 0.0154, "step": 56400 }, { "epoch": 3.9142300511940675e-13, "grad_norm": 0.4318457543849945, "learning_rate": 7.085215550194893e-05, "loss": 0.0155, "step": 56410 }, { "epoch": 3.914923940584458e-13, "grad_norm": 0.46839621663093567, "learning_rate": 7.083084218175417e-05, "loss": 0.0157, "step": 56420 }, { "epoch": 3.915617829974849e-13, "grad_norm": 0.38013797998428345, "learning_rate": 7.080953006074691e-05, "loss": 0.0157, "step": 56430 }, { "epoch": 3.9163117193652397e-13, "grad_norm": 0.44086623191833496, "learning_rate": 7.078821914111736e-05, "loss": 0.0151, "step": 56440 }, { "epoch": 3.9170056087556304e-13, "grad_norm": 0.4386201798915863, "learning_rate": 7.076690942505552e-05, "loss": 0.0157, "step": 56450 }, { "epoch": 3.917699498146021e-13, "grad_norm": 0.44787922501564026, "learning_rate": 7.074560091475127e-05, "loss": 0.016, "step": 56460 }, { "epoch": 3.918393387536412e-13, "grad_norm": 0.40596306324005127, "learning_rate": 7.072429361239444e-05, "loss": 0.0153, "step": 56470 }, { "epoch": 3.9190872769268026e-13, "grad_norm": 0.4174996614456177, "learning_rate": 7.070298752017464e-05, "loss": 0.0151, "step": 56480 }, { "epoch": 3.9197811663171933e-13, "grad_norm": 0.4357069134712219, "learning_rate": 7.068168264028144e-05, "loss": 0.0151, "step": 56490 }, { "epoch": 3.920475055707584e-13, "grad_norm": 0.41008269786834717, "learning_rate": 7.066037897490425e-05, "loss": 0.0153, "step": 56500 }, { "epoch": 3.921168945097975e-13, "grad_norm": 0.44887158274650574, "learning_rate": 7.063907652623232e-05, "loss": 0.0147, "step": 56510 }, { "epoch": 3.9218628344883655e-13, "grad_norm": 0.40428757667541504, "learning_rate": 7.061777529645481e-05, "loss": 0.0149, "step": 56520 }, { "epoch": 3.922556723878756e-13, "grad_norm": 0.41951295733451843, "learning_rate": 7.05964752877608e-05, "loss": 0.015, "step": 56530 }, { "epoch": 3.923250613269147e-13, "grad_norm": 0.38077035546302795, "learning_rate": 7.057517650233912e-05, "loss": 0.0145, "step": 56540 }, { "epoch": 3.9239445026595376e-13, "grad_norm": 0.4603797495365143, "learning_rate": 7.055387894237864e-05, "loss": 0.015, "step": 56550 }, { "epoch": 3.9246383920499284e-13, "grad_norm": 0.40778848528862, "learning_rate": 7.053258261006795e-05, "loss": 0.015, "step": 56560 }, { "epoch": 3.925332281440319e-13, "grad_norm": 0.44215431809425354, "learning_rate": 7.051128750759561e-05, "loss": 0.0154, "step": 56570 }, { "epoch": 3.92602617083071e-13, "grad_norm": 0.455243855714798, "learning_rate": 7.048999363715001e-05, "loss": 0.0157, "step": 56580 }, { "epoch": 3.9267200602211005e-13, "grad_norm": 0.4369215965270996, "learning_rate": 7.046870100091942e-05, "loss": 0.0151, "step": 56590 }, { "epoch": 3.9274139496114913e-13, "grad_norm": 0.5058269500732422, "learning_rate": 7.044740960109202e-05, "loss": 0.016, "step": 56600 }, { "epoch": 3.928107839001882e-13, "grad_norm": 0.4909345805644989, "learning_rate": 7.042611943985583e-05, "loss": 0.0154, "step": 56610 }, { "epoch": 3.9288017283922727e-13, "grad_norm": 0.4342595636844635, "learning_rate": 7.040483051939868e-05, "loss": 0.0153, "step": 56620 }, { "epoch": 3.9294956177826634e-13, "grad_norm": 0.40795594453811646, "learning_rate": 7.038354284190843e-05, "loss": 0.0152, "step": 56630 }, { "epoch": 3.930189507173054e-13, "grad_norm": 0.4372982978820801, "learning_rate": 7.036225640957264e-05, "loss": 0.0145, "step": 56640 }, { "epoch": 3.930883396563445e-13, "grad_norm": 0.4582301378250122, "learning_rate": 7.03409712245789e-05, "loss": 0.0156, "step": 56650 }, { "epoch": 3.9315772859538356e-13, "grad_norm": 0.433635413646698, "learning_rate": 7.031968728911453e-05, "loss": 0.0145, "step": 56660 }, { "epoch": 3.9322711753442263e-13, "grad_norm": 0.37060025334358215, "learning_rate": 7.029840460536681e-05, "loss": 0.0146, "step": 56670 }, { "epoch": 3.932965064734617e-13, "grad_norm": 0.4029880166053772, "learning_rate": 7.027712317552286e-05, "loss": 0.0151, "step": 56680 }, { "epoch": 3.933658954125008e-13, "grad_norm": 0.43690505623817444, "learning_rate": 7.025584300176969e-05, "loss": 0.0146, "step": 56690 }, { "epoch": 3.9343528435153985e-13, "grad_norm": 0.44804713129997253, "learning_rate": 7.023456408629416e-05, "loss": 0.0149, "step": 56700 }, { "epoch": 3.935046732905789e-13, "grad_norm": 0.4127626419067383, "learning_rate": 7.021328643128302e-05, "loss": 0.0143, "step": 56710 }, { "epoch": 3.93574062229618e-13, "grad_norm": 0.4582853615283966, "learning_rate": 7.019201003892285e-05, "loss": 0.0156, "step": 56720 }, { "epoch": 3.9364345116865707e-13, "grad_norm": 0.46384140849113464, "learning_rate": 7.017073491140018e-05, "loss": 0.0153, "step": 56730 }, { "epoch": 3.9371284010769614e-13, "grad_norm": 0.4601602852344513, "learning_rate": 7.01494610509013e-05, "loss": 0.0156, "step": 56740 }, { "epoch": 3.937822290467352e-13, "grad_norm": 0.46088993549346924, "learning_rate": 7.01281884596125e-05, "loss": 0.0148, "step": 56750 }, { "epoch": 3.938516179857743e-13, "grad_norm": 0.4131361246109009, "learning_rate": 7.01069171397198e-05, "loss": 0.0158, "step": 56760 }, { "epoch": 3.9392100692481335e-13, "grad_norm": 0.42925024032592773, "learning_rate": 7.008564709340922e-05, "loss": 0.0161, "step": 56770 }, { "epoch": 3.9399039586385243e-13, "grad_norm": 0.4675041139125824, "learning_rate": 7.006437832286654e-05, "loss": 0.0151, "step": 56780 }, { "epoch": 3.940597848028915e-13, "grad_norm": 0.43873900175094604, "learning_rate": 7.004311083027748e-05, "loss": 0.0149, "step": 56790 }, { "epoch": 3.9412917374193057e-13, "grad_norm": 0.42449554800987244, "learning_rate": 7.002184461782761e-05, "loss": 0.0151, "step": 56800 }, { "epoch": 3.9419856268096964e-13, "grad_norm": 0.4432221055030823, "learning_rate": 7.000057968770237e-05, "loss": 0.0152, "step": 56810 }, { "epoch": 3.942679516200087e-13, "grad_norm": 0.43157264590263367, "learning_rate": 6.997931604208701e-05, "loss": 0.0142, "step": 56820 }, { "epoch": 3.943373405590478e-13, "grad_norm": 0.4365767538547516, "learning_rate": 6.995805368316678e-05, "loss": 0.0146, "step": 56830 }, { "epoch": 3.9440672949808686e-13, "grad_norm": 0.3928064703941345, "learning_rate": 6.993679261312664e-05, "loss": 0.0149, "step": 56840 }, { "epoch": 3.9447611843712593e-13, "grad_norm": 0.4556015133857727, "learning_rate": 6.991553283415157e-05, "loss": 0.0146, "step": 56850 }, { "epoch": 3.94545507376165e-13, "grad_norm": 0.45335668325424194, "learning_rate": 6.989427434842628e-05, "loss": 0.0144, "step": 56860 }, { "epoch": 3.946148963152041e-13, "grad_norm": 0.42711931467056274, "learning_rate": 6.987301715813545e-05, "loss": 0.0153, "step": 56870 }, { "epoch": 3.9468428525424315e-13, "grad_norm": 0.4403322637081146, "learning_rate": 6.985176126546357e-05, "loss": 0.0152, "step": 56880 }, { "epoch": 3.947536741932822e-13, "grad_norm": 0.4640282690525055, "learning_rate": 6.983050667259505e-05, "loss": 0.0161, "step": 56890 }, { "epoch": 3.948230631323213e-13, "grad_norm": 0.41736626625061035, "learning_rate": 6.980925338171407e-05, "loss": 0.0151, "step": 56900 }, { "epoch": 3.9489245207136037e-13, "grad_norm": 0.4536428451538086, "learning_rate": 6.978800139500476e-05, "loss": 0.015, "step": 56910 }, { "epoch": 3.9496184101039944e-13, "grad_norm": 0.41467857360839844, "learning_rate": 6.976675071465111e-05, "loss": 0.0154, "step": 56920 }, { "epoch": 3.950312299494385e-13, "grad_norm": 0.39238741993904114, "learning_rate": 6.974550134283692e-05, "loss": 0.0159, "step": 56930 }, { "epoch": 3.951006188884776e-13, "grad_norm": 0.4258521795272827, "learning_rate": 6.972425328174594e-05, "loss": 0.0146, "step": 56940 }, { "epoch": 3.9517000782751666e-13, "grad_norm": 0.42735788226127625, "learning_rate": 6.97030065335617e-05, "loss": 0.0149, "step": 56950 }, { "epoch": 3.9523939676655573e-13, "grad_norm": 0.4038934111595154, "learning_rate": 6.968176110046768e-05, "loss": 0.0147, "step": 56960 }, { "epoch": 3.953087857055948e-13, "grad_norm": 0.38536423444747925, "learning_rate": 6.96605169846471e-05, "loss": 0.015, "step": 56970 }, { "epoch": 3.9537817464463387e-13, "grad_norm": 0.3932158648967743, "learning_rate": 6.963927418828325e-05, "loss": 0.0144, "step": 56980 }, { "epoch": 3.9544756358367295e-13, "grad_norm": 0.4392944276332855, "learning_rate": 6.9618032713559e-05, "loss": 0.0147, "step": 56990 }, { "epoch": 3.95516952522712e-13, "grad_norm": 0.4040606915950775, "learning_rate": 6.959679256265738e-05, "loss": 0.0143, "step": 57000 }, { "epoch": 3.955863414617511e-13, "grad_norm": 0.4628385007381439, "learning_rate": 6.957555373776106e-05, "loss": 0.0146, "step": 57010 }, { "epoch": 3.9565573040079016e-13, "grad_norm": 0.4242909550666809, "learning_rate": 6.95543162410527e-05, "loss": 0.0145, "step": 57020 }, { "epoch": 3.9572511933982923e-13, "grad_norm": 0.4189840853214264, "learning_rate": 6.953308007471478e-05, "loss": 0.0142, "step": 57030 }, { "epoch": 3.957945082788683e-13, "grad_norm": 0.3918236792087555, "learning_rate": 6.951184524092964e-05, "loss": 0.0149, "step": 57040 }, { "epoch": 3.958638972179074e-13, "grad_norm": 0.40740594267845154, "learning_rate": 6.949061174187946e-05, "loss": 0.0153, "step": 57050 }, { "epoch": 3.9593328615694645e-13, "grad_norm": 0.4120291769504547, "learning_rate": 6.946937957974636e-05, "loss": 0.0143, "step": 57060 }, { "epoch": 3.960026750959855e-13, "grad_norm": 0.4184180796146393, "learning_rate": 6.944814875671225e-05, "loss": 0.0156, "step": 57070 }, { "epoch": 3.960720640350246e-13, "grad_norm": 0.4614017605781555, "learning_rate": 6.942691927495894e-05, "loss": 0.0153, "step": 57080 }, { "epoch": 3.9614145297406367e-13, "grad_norm": 0.4297547936439514, "learning_rate": 6.940569113666807e-05, "loss": 0.0151, "step": 57090 }, { "epoch": 3.9621084191310274e-13, "grad_norm": 0.4139978885650635, "learning_rate": 6.938446434402119e-05, "loss": 0.0145, "step": 57100 }, { "epoch": 3.962802308521418e-13, "grad_norm": 0.4635542333126068, "learning_rate": 6.936323889919964e-05, "loss": 0.0146, "step": 57110 }, { "epoch": 3.963496197911809e-13, "grad_norm": 0.4333750307559967, "learning_rate": 6.934201480438473e-05, "loss": 0.015, "step": 57120 }, { "epoch": 3.9641900873021996e-13, "grad_norm": 0.4310522973537445, "learning_rate": 6.932079206175752e-05, "loss": 0.0146, "step": 57130 }, { "epoch": 3.9648839766925903e-13, "grad_norm": 0.3905532658100128, "learning_rate": 6.929957067349899e-05, "loss": 0.0145, "step": 57140 }, { "epoch": 3.965577866082981e-13, "grad_norm": 0.47215235233306885, "learning_rate": 6.927835064178992e-05, "loss": 0.0149, "step": 57150 }, { "epoch": 3.9662717554733717e-13, "grad_norm": 0.4373227059841156, "learning_rate": 6.925713196881109e-05, "loss": 0.014, "step": 57160 }, { "epoch": 3.9669656448637625e-13, "grad_norm": 0.3623383641242981, "learning_rate": 6.923591465674296e-05, "loss": 0.0144, "step": 57170 }, { "epoch": 3.967659534254153e-13, "grad_norm": 0.4165416955947876, "learning_rate": 6.921469870776602e-05, "loss": 0.0143, "step": 57180 }, { "epoch": 3.968353423644544e-13, "grad_norm": 0.3943047523498535, "learning_rate": 6.919348412406045e-05, "loss": 0.0143, "step": 57190 }, { "epoch": 3.9690473130349346e-13, "grad_norm": 0.3963940739631653, "learning_rate": 6.917227090780648e-05, "loss": 0.0148, "step": 57200 }, { "epoch": 3.9697412024253254e-13, "grad_norm": 0.3964156210422516, "learning_rate": 6.915105906118399e-05, "loss": 0.015, "step": 57210 }, { "epoch": 3.970435091815716e-13, "grad_norm": 0.4563462734222412, "learning_rate": 6.912984858637293e-05, "loss": 0.0146, "step": 57220 }, { "epoch": 3.971128981206107e-13, "grad_norm": 0.44456925988197327, "learning_rate": 6.910863948555295e-05, "loss": 0.0149, "step": 57230 }, { "epoch": 3.9718228705964975e-13, "grad_norm": 0.44994962215423584, "learning_rate": 6.908743176090361e-05, "loss": 0.0154, "step": 57240 }, { "epoch": 3.972516759986888e-13, "grad_norm": 0.4185008108615875, "learning_rate": 6.906622541460433e-05, "loss": 0.0146, "step": 57250 }, { "epoch": 3.973210649377279e-13, "grad_norm": 0.43882977962493896, "learning_rate": 6.904502044883444e-05, "loss": 0.0142, "step": 57260 }, { "epoch": 3.9739045387676697e-13, "grad_norm": 0.4675801694393158, "learning_rate": 6.902381686577303e-05, "loss": 0.0143, "step": 57270 }, { "epoch": 3.9745984281580604e-13, "grad_norm": 0.44657430052757263, "learning_rate": 6.900261466759914e-05, "loss": 0.0144, "step": 57280 }, { "epoch": 3.975292317548451e-13, "grad_norm": 0.4550873637199402, "learning_rate": 6.898141385649158e-05, "loss": 0.0137, "step": 57290 }, { "epoch": 3.975986206938842e-13, "grad_norm": 0.41476088762283325, "learning_rate": 6.896021443462907e-05, "loss": 0.0144, "step": 57300 }, { "epoch": 3.9766800963292326e-13, "grad_norm": 0.4287893772125244, "learning_rate": 6.893901640419023e-05, "loss": 0.0143, "step": 57310 }, { "epoch": 3.9773739857196233e-13, "grad_norm": 0.41624078154563904, "learning_rate": 6.891781976735345e-05, "loss": 0.0138, "step": 57320 }, { "epoch": 3.978067875110014e-13, "grad_norm": 0.3956306576728821, "learning_rate": 6.889662452629702e-05, "loss": 0.0142, "step": 57330 }, { "epoch": 3.978761764500405e-13, "grad_norm": 0.40677058696746826, "learning_rate": 6.887543068319906e-05, "loss": 0.0141, "step": 57340 }, { "epoch": 3.9794556538907955e-13, "grad_norm": 0.41273507475852966, "learning_rate": 6.88542382402376e-05, "loss": 0.0147, "step": 57350 }, { "epoch": 3.980149543281186e-13, "grad_norm": 0.42305639386177063, "learning_rate": 6.883304719959049e-05, "loss": 0.0149, "step": 57360 }, { "epoch": 3.980843432671577e-13, "grad_norm": 0.4161817133426666, "learning_rate": 6.881185756343543e-05, "loss": 0.0142, "step": 57370 }, { "epoch": 3.9815373220619676e-13, "grad_norm": 0.44471654295921326, "learning_rate": 6.879066933394994e-05, "loss": 0.0148, "step": 57380 }, { "epoch": 3.9822312114523584e-13, "grad_norm": 0.45363178849220276, "learning_rate": 6.876948251331152e-05, "loss": 0.015, "step": 57390 }, { "epoch": 3.982925100842749e-13, "grad_norm": 0.43336793780326843, "learning_rate": 6.87482971036974e-05, "loss": 0.0151, "step": 57400 }, { "epoch": 3.98361899023314e-13, "grad_norm": 0.468639999628067, "learning_rate": 6.872711310728474e-05, "loss": 0.0146, "step": 57410 }, { "epoch": 3.9843128796235305e-13, "grad_norm": 0.399952232837677, "learning_rate": 6.870593052625045e-05, "loss": 0.014, "step": 57420 }, { "epoch": 3.9850067690139213e-13, "grad_norm": 0.3680015504360199, "learning_rate": 6.868474936277148e-05, "loss": 0.0141, "step": 57430 }, { "epoch": 3.985700658404312e-13, "grad_norm": 0.4158458113670349, "learning_rate": 6.866356961902445e-05, "loss": 0.0142, "step": 57440 }, { "epoch": 3.9863945477947027e-13, "grad_norm": 0.45401182770729065, "learning_rate": 6.864239129718592e-05, "loss": 0.014, "step": 57450 }, { "epoch": 3.9870884371850934e-13, "grad_norm": 0.4258561432361603, "learning_rate": 6.862121439943231e-05, "loss": 0.0141, "step": 57460 }, { "epoch": 3.987782326575484e-13, "grad_norm": 0.4141712784767151, "learning_rate": 6.860003892793989e-05, "loss": 0.0139, "step": 57470 }, { "epoch": 3.988476215965875e-13, "grad_norm": 0.4617779552936554, "learning_rate": 6.85788648848847e-05, "loss": 0.0141, "step": 57480 }, { "epoch": 3.9891701053562656e-13, "grad_norm": 0.41368526220321655, "learning_rate": 6.855769227244277e-05, "loss": 0.0139, "step": 57490 }, { "epoch": 3.9898639947466563e-13, "grad_norm": 0.4462350308895111, "learning_rate": 6.853652109278986e-05, "loss": 0.0145, "step": 57500 }, { "epoch": 3.990557884137047e-13, "grad_norm": 0.4380151331424713, "learning_rate": 6.851535134810173e-05, "loss": 0.0148, "step": 57510 }, { "epoch": 3.991251773527438e-13, "grad_norm": 0.41457033157348633, "learning_rate": 6.849418304055377e-05, "loss": 0.0154, "step": 57520 }, { "epoch": 3.9919456629178285e-13, "grad_norm": 0.4198453724384308, "learning_rate": 6.847301617232148e-05, "loss": 0.0142, "step": 57530 }, { "epoch": 3.992639552308219e-13, "grad_norm": 0.4181012809276581, "learning_rate": 6.845185074558e-05, "loss": 0.0151, "step": 57540 }, { "epoch": 3.99333344169861e-13, "grad_norm": 0.41439008712768555, "learning_rate": 6.843068676250443e-05, "loss": 0.0152, "step": 57550 }, { "epoch": 3.9940273310890007e-13, "grad_norm": 0.4078505337238312, "learning_rate": 6.840952422526972e-05, "loss": 0.0145, "step": 57560 }, { "epoch": 3.9947212204793914e-13, "grad_norm": 0.4217565059661865, "learning_rate": 6.838836313605064e-05, "loss": 0.0141, "step": 57570 }, { "epoch": 3.995415109869782e-13, "grad_norm": 0.47821855545043945, "learning_rate": 6.836720349702178e-05, "loss": 0.0143, "step": 57580 }, { "epoch": 3.996108999260173e-13, "grad_norm": 0.4049832224845886, "learning_rate": 6.83460453103577e-05, "loss": 0.0136, "step": 57590 }, { "epoch": 3.9968028886505635e-13, "grad_norm": 0.3929227888584137, "learning_rate": 6.832488857823267e-05, "loss": 0.0137, "step": 57600 }, { "epoch": 3.9974967780409543e-13, "grad_norm": 0.411957323551178, "learning_rate": 6.83037333028209e-05, "loss": 0.0142, "step": 57610 }, { "epoch": 3.998190667431345e-13, "grad_norm": 0.36621329188346863, "learning_rate": 6.82825794862964e-05, "loss": 0.0144, "step": 57620 }, { "epoch": 3.9988845568217357e-13, "grad_norm": 0.3827258348464966, "learning_rate": 6.826142713083309e-05, "loss": 0.0135, "step": 57630 }, { "epoch": 3.9995784462121264e-13, "grad_norm": 0.3739221394062042, "learning_rate": 6.824027623860467e-05, "loss": 0.0137, "step": 57640 }, { "epoch": 4.000272335602517e-13, "grad_norm": 0.3888283371925354, "learning_rate": 6.821912681178475e-05, "loss": 0.0142, "step": 57650 }, { "epoch": 4.000966224992908e-13, "grad_norm": 0.45020735263824463, "learning_rate": 6.819797885254676e-05, "loss": 0.0145, "step": 57660 }, { "epoch": 4.0016601143832986e-13, "grad_norm": 0.49309056997299194, "learning_rate": 6.817683236306394e-05, "loss": 0.0154, "step": 57670 }, { "epoch": 4.0023540037736893e-13, "grad_norm": 0.39010870456695557, "learning_rate": 6.815568734550949e-05, "loss": 0.0142, "step": 57680 }, { "epoch": 4.00304789316408e-13, "grad_norm": 0.3949480354785919, "learning_rate": 6.813454380205635e-05, "loss": 0.0143, "step": 57690 }, { "epoch": 4.003741782554471e-13, "grad_norm": 0.4118918180465698, "learning_rate": 6.811340173487731e-05, "loss": 0.0149, "step": 57700 }, { "epoch": 4.0044356719448615e-13, "grad_norm": 0.4277634620666504, "learning_rate": 6.809226114614513e-05, "loss": 0.0155, "step": 57710 }, { "epoch": 4.005129561335252e-13, "grad_norm": 0.3900010883808136, "learning_rate": 6.807112203803226e-05, "loss": 0.0146, "step": 57720 }, { "epoch": 4.005823450725643e-13, "grad_norm": 0.3765146732330322, "learning_rate": 6.804998441271114e-05, "loss": 0.0138, "step": 57730 }, { "epoch": 4.0065173401160337e-13, "grad_norm": 0.426493763923645, "learning_rate": 6.802884827235391e-05, "loss": 0.0146, "step": 57740 }, { "epoch": 4.0072112295064244e-13, "grad_norm": 0.40401798486709595, "learning_rate": 6.800771361913273e-05, "loss": 0.0137, "step": 57750 }, { "epoch": 4.007905118896815e-13, "grad_norm": 0.4212826192378998, "learning_rate": 6.798658045521944e-05, "loss": 0.0137, "step": 57760 }, { "epoch": 4.008599008287206e-13, "grad_norm": 0.37939518690109253, "learning_rate": 6.796544878278583e-05, "loss": 0.014, "step": 57770 }, { "epoch": 4.0092928976775966e-13, "grad_norm": 0.371828556060791, "learning_rate": 6.794431860400349e-05, "loss": 0.0139, "step": 57780 }, { "epoch": 4.0099867870679873e-13, "grad_norm": 0.3874775767326355, "learning_rate": 6.792318992104392e-05, "loss": 0.0133, "step": 57790 }, { "epoch": 4.010680676458378e-13, "grad_norm": 0.45589303970336914, "learning_rate": 6.790206273607835e-05, "loss": 0.0144, "step": 57800 }, { "epoch": 4.0113745658487687e-13, "grad_norm": 0.41384899616241455, "learning_rate": 6.788093705127802e-05, "loss": 0.0142, "step": 57810 }, { "epoch": 4.0120684552391594e-13, "grad_norm": 0.4228958785533905, "learning_rate": 6.785981286881385e-05, "loss": 0.0144, "step": 57820 }, { "epoch": 4.01276234462955e-13, "grad_norm": 0.42065027356147766, "learning_rate": 6.783869019085667e-05, "loss": 0.015, "step": 57830 }, { "epoch": 4.013456234019941e-13, "grad_norm": 0.4128718972206116, "learning_rate": 6.781756901957724e-05, "loss": 0.0137, "step": 57840 }, { "epoch": 4.0141501234103316e-13, "grad_norm": 0.4563153386116028, "learning_rate": 6.779644935714599e-05, "loss": 0.0149, "step": 57850 }, { "epoch": 4.0148440128007223e-13, "grad_norm": 0.48293963074684143, "learning_rate": 6.777533120573339e-05, "loss": 0.0144, "step": 57860 }, { "epoch": 4.015537902191113e-13, "grad_norm": 0.418451189994812, "learning_rate": 6.775421456750958e-05, "loss": 0.0149, "step": 57870 }, { "epoch": 4.016231791581504e-13, "grad_norm": 0.39542967081069946, "learning_rate": 6.773309944464468e-05, "loss": 0.0146, "step": 57880 }, { "epoch": 4.0169256809718945e-13, "grad_norm": 0.45584413409233093, "learning_rate": 6.771198583930857e-05, "loss": 0.0142, "step": 57890 }, { "epoch": 4.017619570362285e-13, "grad_norm": 0.4419142007827759, "learning_rate": 6.7690873753671e-05, "loss": 0.0144, "step": 57900 }, { "epoch": 4.018313459752676e-13, "grad_norm": 0.42624136805534363, "learning_rate": 6.766976318990157e-05, "loss": 0.0137, "step": 57910 }, { "epoch": 4.0190073491430667e-13, "grad_norm": 0.3503129482269287, "learning_rate": 6.764865415016976e-05, "loss": 0.0141, "step": 57920 }, { "epoch": 4.0197012385334574e-13, "grad_norm": 0.3685849606990814, "learning_rate": 6.762754663664476e-05, "loss": 0.0141, "step": 57930 }, { "epoch": 4.020395127923848e-13, "grad_norm": 0.4072343409061432, "learning_rate": 6.760644065149578e-05, "loss": 0.0139, "step": 57940 }, { "epoch": 4.021089017314239e-13, "grad_norm": 0.408720999956131, "learning_rate": 6.758533619689172e-05, "loss": 0.0138, "step": 57950 }, { "epoch": 4.0217829067046296e-13, "grad_norm": 0.40746134519577026, "learning_rate": 6.756423327500147e-05, "loss": 0.014, "step": 57960 }, { "epoch": 4.0224767960950203e-13, "grad_norm": 0.4426896870136261, "learning_rate": 6.754313188799361e-05, "loss": 0.0143, "step": 57970 }, { "epoch": 4.023170685485411e-13, "grad_norm": 0.4399579167366028, "learning_rate": 6.752203203803669e-05, "loss": 0.0145, "step": 57980 }, { "epoch": 4.0238645748758017e-13, "grad_norm": 0.4241209924221039, "learning_rate": 6.750093372729901e-05, "loss": 0.0149, "step": 57990 }, { "epoch": 4.0245584642661925e-13, "grad_norm": 0.45219913125038147, "learning_rate": 6.747983695794878e-05, "loss": 0.0139, "step": 58000 }, { "epoch": 4.025252353656583e-13, "grad_norm": 0.37937626242637634, "learning_rate": 6.745874173215399e-05, "loss": 0.0146, "step": 58010 }, { "epoch": 4.025946243046974e-13, "grad_norm": 0.4072069525718689, "learning_rate": 6.743764805208254e-05, "loss": 0.0147, "step": 58020 }, { "epoch": 4.0266401324373646e-13, "grad_norm": 0.44306349754333496, "learning_rate": 6.741655591990208e-05, "loss": 0.0143, "step": 58030 }, { "epoch": 4.0273340218277554e-13, "grad_norm": 0.4412842094898224, "learning_rate": 6.739546533778022e-05, "loss": 0.0138, "step": 58040 }, { "epoch": 4.028027911218146e-13, "grad_norm": 0.40085819363594055, "learning_rate": 6.737437630788428e-05, "loss": 0.014, "step": 58050 }, { "epoch": 4.028721800608537e-13, "grad_norm": 0.42393332719802856, "learning_rate": 6.735328883238155e-05, "loss": 0.0138, "step": 58060 }, { "epoch": 4.0294156899989275e-13, "grad_norm": 0.4182624816894531, "learning_rate": 6.733220291343903e-05, "loss": 0.0137, "step": 58070 }, { "epoch": 4.030109579389318e-13, "grad_norm": 0.464653342962265, "learning_rate": 6.731111855322371e-05, "loss": 0.0141, "step": 58080 }, { "epoch": 4.030803468779709e-13, "grad_norm": 0.3769427537918091, "learning_rate": 6.729003575390225e-05, "loss": 0.0138, "step": 58090 }, { "epoch": 4.0314973581700997e-13, "grad_norm": 0.4304334819316864, "learning_rate": 6.72689545176413e-05, "loss": 0.0137, "step": 58100 }, { "epoch": 4.0321912475604904e-13, "grad_norm": 0.45158982276916504, "learning_rate": 6.724787484660724e-05, "loss": 0.0142, "step": 58110 }, { "epoch": 4.032885136950881e-13, "grad_norm": 0.4089983403682709, "learning_rate": 6.722679674296634e-05, "loss": 0.0138, "step": 58120 }, { "epoch": 4.033579026341272e-13, "grad_norm": 0.398776113986969, "learning_rate": 6.720572020888472e-05, "loss": 0.0141, "step": 58130 }, { "epoch": 4.0342729157316626e-13, "grad_norm": 0.4063652753829956, "learning_rate": 6.718464524652834e-05, "loss": 0.0147, "step": 58140 }, { "epoch": 4.0349668051220533e-13, "grad_norm": 0.39153823256492615, "learning_rate": 6.716357185806293e-05, "loss": 0.014, "step": 58150 }, { "epoch": 4.035660694512444e-13, "grad_norm": 0.43882134556770325, "learning_rate": 6.714250004565416e-05, "loss": 0.0142, "step": 58160 }, { "epoch": 4.0363545839028347e-13, "grad_norm": 0.3834071755409241, "learning_rate": 6.712142981146743e-05, "loss": 0.0143, "step": 58170 }, { "epoch": 4.0370484732932255e-13, "grad_norm": 0.37758058309555054, "learning_rate": 6.710036115766811e-05, "loss": 0.0149, "step": 58180 }, { "epoch": 4.037742362683616e-13, "grad_norm": 0.3534920811653137, "learning_rate": 6.707929408642126e-05, "loss": 0.0137, "step": 58190 }, { "epoch": 4.038436252074007e-13, "grad_norm": 0.4264662563800812, "learning_rate": 6.705822859989189e-05, "loss": 0.0137, "step": 58200 }, { "epoch": 4.0391301414643976e-13, "grad_norm": 0.36163854598999023, "learning_rate": 6.703716470024479e-05, "loss": 0.0138, "step": 58210 }, { "epoch": 4.0398240308547884e-13, "grad_norm": 0.38455355167388916, "learning_rate": 6.701610238964462e-05, "loss": 0.0137, "step": 58220 }, { "epoch": 4.040517920245179e-13, "grad_norm": 0.40065595507621765, "learning_rate": 6.699504167025582e-05, "loss": 0.0141, "step": 58230 }, { "epoch": 4.04121180963557e-13, "grad_norm": 0.4075939655303955, "learning_rate": 6.697398254424277e-05, "loss": 0.0137, "step": 58240 }, { "epoch": 4.0419056990259605e-13, "grad_norm": 0.4015392065048218, "learning_rate": 6.695292501376954e-05, "loss": 0.0136, "step": 58250 }, { "epoch": 4.0425995884163513e-13, "grad_norm": 0.4214920401573181, "learning_rate": 6.693186908100023e-05, "loss": 0.0134, "step": 58260 }, { "epoch": 4.043293477806742e-13, "grad_norm": 0.41022083163261414, "learning_rate": 6.691081474809856e-05, "loss": 0.014, "step": 58270 }, { "epoch": 4.0439873671971327e-13, "grad_norm": 0.4181958734989166, "learning_rate": 6.688976201722824e-05, "loss": 0.0133, "step": 58280 }, { "epoch": 4.0446812565875234e-13, "grad_norm": 0.4100355803966522, "learning_rate": 6.686871089055275e-05, "loss": 0.0138, "step": 58290 }, { "epoch": 4.045375145977914e-13, "grad_norm": 0.4056140184402466, "learning_rate": 6.684766137023541e-05, "loss": 0.0146, "step": 58300 }, { "epoch": 4.046069035368305e-13, "grad_norm": 0.3617168366909027, "learning_rate": 6.682661345843944e-05, "loss": 0.0138, "step": 58310 }, { "epoch": 4.0467629247586956e-13, "grad_norm": 0.4023165702819824, "learning_rate": 6.680556715732781e-05, "loss": 0.0148, "step": 58320 }, { "epoch": 4.0474568141490863e-13, "grad_norm": 0.45406314730644226, "learning_rate": 6.678452246906332e-05, "loss": 0.0144, "step": 58330 }, { "epoch": 4.048150703539477e-13, "grad_norm": 0.4425419867038727, "learning_rate": 6.676347939580869e-05, "loss": 0.0144, "step": 58340 }, { "epoch": 4.048844592929868e-13, "grad_norm": 0.4028211832046509, "learning_rate": 6.67424379397264e-05, "loss": 0.0135, "step": 58350 }, { "epoch": 4.0495384823202585e-13, "grad_norm": 0.4316137433052063, "learning_rate": 6.672139810297876e-05, "loss": 0.0136, "step": 58360 }, { "epoch": 4.050232371710649e-13, "grad_norm": 0.40604284405708313, "learning_rate": 6.670035988772801e-05, "loss": 0.0135, "step": 58370 }, { "epoch": 4.05092626110104e-13, "grad_norm": 0.41607794165611267, "learning_rate": 6.667932329613607e-05, "loss": 0.0139, "step": 58380 }, { "epoch": 4.0516201504914306e-13, "grad_norm": 0.3791644871234894, "learning_rate": 6.665828833036487e-05, "loss": 0.0138, "step": 58390 }, { "epoch": 4.0523140398818214e-13, "grad_norm": 0.43319472670555115, "learning_rate": 6.663725499257597e-05, "loss": 0.0141, "step": 58400 }, { "epoch": 4.053007929272212e-13, "grad_norm": 0.4303855299949646, "learning_rate": 6.661622328493099e-05, "loss": 0.0132, "step": 58410 }, { "epoch": 4.053701818662603e-13, "grad_norm": 0.3602091073989868, "learning_rate": 6.659519320959118e-05, "loss": 0.0133, "step": 58420 }, { "epoch": 4.0543957080529935e-13, "grad_norm": 0.3889369070529938, "learning_rate": 6.657416476871773e-05, "loss": 0.0136, "step": 58430 }, { "epoch": 4.0550895974433843e-13, "grad_norm": 0.39211130142211914, "learning_rate": 6.655313796447166e-05, "loss": 0.0134, "step": 58440 }, { "epoch": 4.055783486833775e-13, "grad_norm": 0.3854549527168274, "learning_rate": 6.653211279901379e-05, "loss": 0.0137, "step": 58450 }, { "epoch": 4.0564773762241657e-13, "grad_norm": 0.39625030755996704, "learning_rate": 6.651108927450475e-05, "loss": 0.0144, "step": 58460 }, { "epoch": 4.0571712656145564e-13, "grad_norm": 0.4508553147315979, "learning_rate": 6.649006739310509e-05, "loss": 0.0134, "step": 58470 }, { "epoch": 4.057865155004947e-13, "grad_norm": 0.43042969703674316, "learning_rate": 6.64690471569751e-05, "loss": 0.014, "step": 58480 }, { "epoch": 4.058559044395338e-13, "grad_norm": 0.38030144572257996, "learning_rate": 6.644802856827495e-05, "loss": 0.0142, "step": 58490 }, { "epoch": 4.0592529337857286e-13, "grad_norm": 0.4090728163719177, "learning_rate": 6.642701162916461e-05, "loss": 0.0142, "step": 58500 }, { "epoch": 4.0599468231761193e-13, "grad_norm": 0.4244824945926666, "learning_rate": 6.640599634180394e-05, "loss": 0.0134, "step": 58510 }, { "epoch": 4.06064071256651e-13, "grad_norm": 0.4310479462146759, "learning_rate": 6.638498270835251e-05, "loss": 0.0135, "step": 58520 }, { "epoch": 4.061334601956901e-13, "grad_norm": 0.424546480178833, "learning_rate": 6.636397073096989e-05, "loss": 0.0133, "step": 58530 }, { "epoch": 4.0620284913472915e-13, "grad_norm": 0.4647945463657379, "learning_rate": 6.634296041181534e-05, "loss": 0.0133, "step": 58540 }, { "epoch": 4.062722380737682e-13, "grad_norm": 0.39978456497192383, "learning_rate": 6.6321951753048e-05, "loss": 0.0135, "step": 58550 }, { "epoch": 4.063416270128073e-13, "grad_norm": 0.40911298990249634, "learning_rate": 6.630094475682679e-05, "loss": 0.0134, "step": 58560 }, { "epoch": 4.0641101595184637e-13, "grad_norm": 0.3885899484157562, "learning_rate": 6.627993942531061e-05, "loss": 0.0132, "step": 58570 }, { "epoch": 4.0648040489088544e-13, "grad_norm": 0.37165728211402893, "learning_rate": 6.625893576065797e-05, "loss": 0.0131, "step": 58580 }, { "epoch": 4.065497938299245e-13, "grad_norm": 0.402981698513031, "learning_rate": 6.623793376502744e-05, "loss": 0.0132, "step": 58590 }, { "epoch": 4.066191827689636e-13, "grad_norm": 0.380336195230484, "learning_rate": 6.62169334405772e-05, "loss": 0.0133, "step": 58600 }, { "epoch": 4.0668857170800266e-13, "grad_norm": 0.4026171565055847, "learning_rate": 6.619593478946544e-05, "loss": 0.014, "step": 58610 }, { "epoch": 4.0675796064704173e-13, "grad_norm": 0.3991316854953766, "learning_rate": 6.617493781385001e-05, "loss": 0.0137, "step": 58620 }, { "epoch": 4.068273495860808e-13, "grad_norm": 0.42211297154426575, "learning_rate": 6.615394251588876e-05, "loss": 0.0138, "step": 58630 }, { "epoch": 4.0689673852511987e-13, "grad_norm": 0.4297184348106384, "learning_rate": 6.613294889773926e-05, "loss": 0.0141, "step": 58640 }, { "epoch": 4.0696612746415894e-13, "grad_norm": 0.42591068148612976, "learning_rate": 6.611195696155891e-05, "loss": 0.0142, "step": 58650 }, { "epoch": 4.07035516403198e-13, "grad_norm": 0.4571468234062195, "learning_rate": 6.609096670950494e-05, "loss": 0.0137, "step": 58660 }, { "epoch": 4.071049053422371e-13, "grad_norm": 0.36638063192367554, "learning_rate": 6.60699781437345e-05, "loss": 0.0128, "step": 58670 }, { "epoch": 4.0717429428127616e-13, "grad_norm": 0.34877273440361023, "learning_rate": 6.604899126640441e-05, "loss": 0.0134, "step": 58680 }, { "epoch": 4.0724368322031523e-13, "grad_norm": 0.40231195092201233, "learning_rate": 6.602800607967145e-05, "loss": 0.0134, "step": 58690 }, { "epoch": 4.073130721593543e-13, "grad_norm": 0.41688111424446106, "learning_rate": 6.600702258569213e-05, "loss": 0.0134, "step": 58700 }, { "epoch": 4.073824610983934e-13, "grad_norm": 0.426654189825058, "learning_rate": 6.59860407866229e-05, "loss": 0.0136, "step": 58710 }, { "epoch": 4.0745185003743245e-13, "grad_norm": 0.39790940284729004, "learning_rate": 6.596506068461992e-05, "loss": 0.0134, "step": 58720 }, { "epoch": 4.075212389764715e-13, "grad_norm": 0.41834887862205505, "learning_rate": 6.594408228183921e-05, "loss": 0.0133, "step": 58730 }, { "epoch": 4.075906279155106e-13, "grad_norm": 0.3945387899875641, "learning_rate": 6.592310558043666e-05, "loss": 0.0133, "step": 58740 }, { "epoch": 4.0766001685454967e-13, "grad_norm": 0.4242449402809143, "learning_rate": 6.590213058256793e-05, "loss": 0.0128, "step": 58750 }, { "epoch": 4.0772940579358874e-13, "grad_norm": 0.38701456785202026, "learning_rate": 6.588115729038852e-05, "loss": 0.0136, "step": 58760 }, { "epoch": 4.077987947326278e-13, "grad_norm": 0.3883788585662842, "learning_rate": 6.586018570605379e-05, "loss": 0.0141, "step": 58770 }, { "epoch": 4.078681836716669e-13, "grad_norm": 0.40584617853164673, "learning_rate": 6.583921583171889e-05, "loss": 0.0137, "step": 58780 }, { "epoch": 4.0793757261070596e-13, "grad_norm": 0.40673205256462097, "learning_rate": 6.58182476695388e-05, "loss": 0.0143, "step": 58790 }, { "epoch": 4.0800696154974503e-13, "grad_norm": 0.40646934509277344, "learning_rate": 6.579728122166829e-05, "loss": 0.0146, "step": 58800 }, { "epoch": 4.080763504887841e-13, "grad_norm": 0.3595012426376343, "learning_rate": 6.577631649026202e-05, "loss": 0.0136, "step": 58810 }, { "epoch": 4.0814573942782317e-13, "grad_norm": 0.40747177600860596, "learning_rate": 6.575535347747446e-05, "loss": 0.0134, "step": 58820 }, { "epoch": 4.0821512836686225e-13, "grad_norm": 0.4273371994495392, "learning_rate": 6.573439218545984e-05, "loss": 0.0134, "step": 58830 }, { "epoch": 4.082845173059013e-13, "grad_norm": 0.396469384431839, "learning_rate": 6.571343261637232e-05, "loss": 0.013, "step": 58840 }, { "epoch": 4.083539062449404e-13, "grad_norm": 0.3808845579624176, "learning_rate": 6.569247477236574e-05, "loss": 0.0131, "step": 58850 }, { "epoch": 4.0842329518397946e-13, "grad_norm": 0.39700183272361755, "learning_rate": 6.567151865559394e-05, "loss": 0.0133, "step": 58860 }, { "epoch": 4.0849268412301853e-13, "grad_norm": 0.35213950276374817, "learning_rate": 6.56505642682104e-05, "loss": 0.0137, "step": 58870 }, { "epoch": 4.085620730620576e-13, "grad_norm": 0.35905835032463074, "learning_rate": 6.562961161236857e-05, "loss": 0.0129, "step": 58880 }, { "epoch": 4.086314620010967e-13, "grad_norm": 0.3485119044780731, "learning_rate": 6.560866069022162e-05, "loss": 0.013, "step": 58890 }, { "epoch": 4.0870085094013575e-13, "grad_norm": 0.4130946695804596, "learning_rate": 6.558771150392263e-05, "loss": 0.0132, "step": 58900 }, { "epoch": 4.087702398791748e-13, "grad_norm": 0.3862975835800171, "learning_rate": 6.556676405562438e-05, "loss": 0.0142, "step": 58910 }, { "epoch": 4.088396288182139e-13, "grad_norm": 0.4322080910205841, "learning_rate": 6.554581834747963e-05, "loss": 0.0137, "step": 58920 }, { "epoch": 4.0890901775725297e-13, "grad_norm": 0.3542990982532501, "learning_rate": 6.552487438164083e-05, "loss": 0.0136, "step": 58930 }, { "epoch": 4.0897840669629204e-13, "grad_norm": 0.378222793340683, "learning_rate": 6.550393216026033e-05, "loss": 0.0135, "step": 58940 }, { "epoch": 4.090477956353311e-13, "grad_norm": 0.39774850010871887, "learning_rate": 6.54829916854902e-05, "loss": 0.014, "step": 58950 }, { "epoch": 4.091171845743702e-13, "grad_norm": 0.40024909377098083, "learning_rate": 6.54620529594825e-05, "loss": 0.014, "step": 58960 }, { "epoch": 4.0918657351340926e-13, "grad_norm": 0.3695693910121918, "learning_rate": 6.544111598438891e-05, "loss": 0.0137, "step": 58970 }, { "epoch": 4.0925596245244833e-13, "grad_norm": 0.387448787689209, "learning_rate": 6.542018076236111e-05, "loss": 0.0129, "step": 58980 }, { "epoch": 4.093253513914874e-13, "grad_norm": 0.4106977880001068, "learning_rate": 6.539924729555047e-05, "loss": 0.0133, "step": 58990 }, { "epoch": 4.0939474033052647e-13, "grad_norm": 0.3670305013656616, "learning_rate": 6.537831558610825e-05, "loss": 0.0129, "step": 59000 }, { "epoch": 4.0946412926956555e-13, "grad_norm": 0.40044716000556946, "learning_rate": 6.53573856361855e-05, "loss": 0.0128, "step": 59010 }, { "epoch": 4.095335182086046e-13, "grad_norm": 0.33848366141319275, "learning_rate": 6.533645744793313e-05, "loss": 0.0128, "step": 59020 }, { "epoch": 4.096029071476437e-13, "grad_norm": 0.35968485474586487, "learning_rate": 6.531553102350176e-05, "loss": 0.0135, "step": 59030 }, { "epoch": 4.0967229608668276e-13, "grad_norm": 0.36819592118263245, "learning_rate": 6.529460636504201e-05, "loss": 0.0127, "step": 59040 }, { "epoch": 4.0974168502572184e-13, "grad_norm": 0.44015631079673767, "learning_rate": 6.527368347470412e-05, "loss": 0.0132, "step": 59050 }, { "epoch": 4.098110739647609e-13, "grad_norm": 0.3869715631008148, "learning_rate": 6.525276235463831e-05, "loss": 0.0134, "step": 59060 }, { "epoch": 4.098804629038e-13, "grad_norm": 0.40561458468437195, "learning_rate": 6.523184300699452e-05, "loss": 0.0137, "step": 59070 }, { "epoch": 4.0994985184283905e-13, "grad_norm": 0.44949543476104736, "learning_rate": 6.521092543392254e-05, "loss": 0.0136, "step": 59080 }, { "epoch": 4.100192407818781e-13, "grad_norm": 0.3991803824901581, "learning_rate": 6.519000963757195e-05, "loss": 0.0134, "step": 59090 }, { "epoch": 4.100886297209172e-13, "grad_norm": 0.44397252798080444, "learning_rate": 6.516909562009225e-05, "loss": 0.0135, "step": 59100 }, { "epoch": 4.1015801865995627e-13, "grad_norm": 0.44632619619369507, "learning_rate": 6.514818338363257e-05, "loss": 0.0138, "step": 59110 }, { "epoch": 4.1022740759899534e-13, "grad_norm": 0.3903624713420868, "learning_rate": 6.51272729303421e-05, "loss": 0.0137, "step": 59120 }, { "epoch": 4.102967965380344e-13, "grad_norm": 0.39937642216682434, "learning_rate": 6.510636426236959e-05, "loss": 0.0132, "step": 59130 }, { "epoch": 4.103661854770735e-13, "grad_norm": 0.39612728357315063, "learning_rate": 6.508545738186382e-05, "loss": 0.0127, "step": 59140 }, { "epoch": 4.1043557441611256e-13, "grad_norm": 0.4078879952430725, "learning_rate": 6.506455229097322e-05, "loss": 0.0133, "step": 59150 }, { "epoch": 4.1050496335515163e-13, "grad_norm": 0.37224480509757996, "learning_rate": 6.50436489918462e-05, "loss": 0.0129, "step": 59160 }, { "epoch": 4.105743522941907e-13, "grad_norm": 0.3414113223552704, "learning_rate": 6.502274748663085e-05, "loss": 0.0126, "step": 59170 }, { "epoch": 4.106437412332298e-13, "grad_norm": 0.3998844623565674, "learning_rate": 6.50018477774751e-05, "loss": 0.0135, "step": 59180 }, { "epoch": 4.1071313017226885e-13, "grad_norm": 0.374756783246994, "learning_rate": 6.498094986652679e-05, "loss": 0.0127, "step": 59190 }, { "epoch": 4.107825191113079e-13, "grad_norm": 0.38612130284309387, "learning_rate": 6.496005375593344e-05, "loss": 0.0131, "step": 59200 }, { "epoch": 4.10851908050347e-13, "grad_norm": 0.3915155529975891, "learning_rate": 6.49391594478425e-05, "loss": 0.0126, "step": 59210 }, { "epoch": 4.1092129698938606e-13, "grad_norm": 0.40161240100860596, "learning_rate": 6.49182669444011e-05, "loss": 0.0132, "step": 59220 }, { "epoch": 4.1099068592842514e-13, "grad_norm": 0.42547696828842163, "learning_rate": 6.489737624775639e-05, "loss": 0.0132, "step": 59230 }, { "epoch": 4.110600748674642e-13, "grad_norm": 0.40007367730140686, "learning_rate": 6.487648736005512e-05, "loss": 0.0137, "step": 59240 }, { "epoch": 4.111294638065033e-13, "grad_norm": 0.40161648392677307, "learning_rate": 6.485560028344402e-05, "loss": 0.0129, "step": 59250 }, { "epoch": 4.1119885274554235e-13, "grad_norm": 0.3589463233947754, "learning_rate": 6.483471502006949e-05, "loss": 0.0135, "step": 59260 }, { "epoch": 4.1126824168458143e-13, "grad_norm": 0.37951698899269104, "learning_rate": 6.481383157207788e-05, "loss": 0.0139, "step": 59270 }, { "epoch": 4.113376306236205e-13, "grad_norm": 0.4037768840789795, "learning_rate": 6.479294994161524e-05, "loss": 0.0133, "step": 59280 }, { "epoch": 4.1140701956265957e-13, "grad_norm": 0.3825221657752991, "learning_rate": 6.477207013082752e-05, "loss": 0.013, "step": 59290 }, { "epoch": 4.1147640850169864e-13, "grad_norm": 0.3733770251274109, "learning_rate": 6.475119214186042e-05, "loss": 0.0129, "step": 59300 }, { "epoch": 4.115457974407377e-13, "grad_norm": 0.3949357867240906, "learning_rate": 6.47303159768595e-05, "loss": 0.0129, "step": 59310 }, { "epoch": 4.116151863797768e-13, "grad_norm": 0.3879771828651428, "learning_rate": 6.470944163797007e-05, "loss": 0.0125, "step": 59320 }, { "epoch": 4.1168457531881586e-13, "grad_norm": 0.42295753955841064, "learning_rate": 6.468856912733734e-05, "loss": 0.0127, "step": 59330 }, { "epoch": 4.1175396425785493e-13, "grad_norm": 0.35396626591682434, "learning_rate": 6.466769844710625e-05, "loss": 0.0131, "step": 59340 }, { "epoch": 4.11823353196894e-13, "grad_norm": 0.3936145007610321, "learning_rate": 6.464682959942162e-05, "loss": 0.0127, "step": 59350 }, { "epoch": 4.118927421359331e-13, "grad_norm": 0.4186612367630005, "learning_rate": 6.462596258642802e-05, "loss": 0.0126, "step": 59360 }, { "epoch": 4.1196213107497215e-13, "grad_norm": 0.433896005153656, "learning_rate": 6.46050974102699e-05, "loss": 0.0134, "step": 59370 }, { "epoch": 4.120315200140112e-13, "grad_norm": 0.4166755676269531, "learning_rate": 6.45842340730914e-05, "loss": 0.0133, "step": 59380 }, { "epoch": 4.121009089530503e-13, "grad_norm": 0.39657217264175415, "learning_rate": 6.456337257703666e-05, "loss": 0.0133, "step": 59390 }, { "epoch": 4.1217029789208937e-13, "grad_norm": 0.38028520345687866, "learning_rate": 6.454251292424944e-05, "loss": 0.0134, "step": 59400 }, { "epoch": 4.1223968683112844e-13, "grad_norm": 0.3849521279335022, "learning_rate": 6.452165511687342e-05, "loss": 0.0129, "step": 59410 }, { "epoch": 4.123090757701675e-13, "grad_norm": 0.39472293853759766, "learning_rate": 6.450079915705208e-05, "loss": 0.0135, "step": 59420 }, { "epoch": 4.123784647092066e-13, "grad_norm": 0.3438981771469116, "learning_rate": 6.44799450469287e-05, "loss": 0.0136, "step": 59430 }, { "epoch": 4.1244785364824565e-13, "grad_norm": 0.3548378646373749, "learning_rate": 6.44590927886463e-05, "loss": 0.0131, "step": 59440 }, { "epoch": 4.1251724258728473e-13, "grad_norm": 0.4216485023498535, "learning_rate": 6.443824238434786e-05, "loss": 0.0132, "step": 59450 }, { "epoch": 4.125866315263238e-13, "grad_norm": 0.353209525346756, "learning_rate": 6.4417393836176e-05, "loss": 0.0129, "step": 59460 }, { "epoch": 4.1265602046536287e-13, "grad_norm": 0.3676789104938507, "learning_rate": 6.439654714627333e-05, "loss": 0.013, "step": 59470 }, { "epoch": 4.1272540940440194e-13, "grad_norm": 0.3958030045032501, "learning_rate": 6.437570231678206e-05, "loss": 0.0129, "step": 59480 }, { "epoch": 4.12794798343441e-13, "grad_norm": 0.4284670352935791, "learning_rate": 6.435485934984443e-05, "loss": 0.0131, "step": 59490 }, { "epoch": 4.128641872824801e-13, "grad_norm": 0.3765457272529602, "learning_rate": 6.43340182476023e-05, "loss": 0.0127, "step": 59500 }, { "epoch": 4.1293357622151916e-13, "grad_norm": 0.44276162981987, "learning_rate": 6.431317901219746e-05, "loss": 0.0127, "step": 59510 }, { "epoch": 4.1300296516055823e-13, "grad_norm": 0.405403196811676, "learning_rate": 6.429234164577145e-05, "loss": 0.0133, "step": 59520 }, { "epoch": 4.130723540995973e-13, "grad_norm": 0.3730766177177429, "learning_rate": 6.427150615046564e-05, "loss": 0.013, "step": 59530 }, { "epoch": 4.131417430386364e-13, "grad_norm": 0.39303141832351685, "learning_rate": 6.425067252842117e-05, "loss": 0.0134, "step": 59540 }, { "epoch": 4.1321113197767545e-13, "grad_norm": 0.3489309847354889, "learning_rate": 6.422984078177908e-05, "loss": 0.0133, "step": 59550 }, { "epoch": 4.132805209167145e-13, "grad_norm": 0.37903526425361633, "learning_rate": 6.420901091268008e-05, "loss": 0.0126, "step": 59560 }, { "epoch": 4.133499098557536e-13, "grad_norm": 0.3770083785057068, "learning_rate": 6.418818292326484e-05, "loss": 0.0135, "step": 59570 }, { "epoch": 4.1341929879479267e-13, "grad_norm": 0.42512884736061096, "learning_rate": 6.41673568156737e-05, "loss": 0.0132, "step": 59580 }, { "epoch": 4.1348868773383174e-13, "grad_norm": 0.4105325937271118, "learning_rate": 6.414653259204691e-05, "loss": 0.0134, "step": 59590 }, { "epoch": 4.135580766728708e-13, "grad_norm": 0.3682454228401184, "learning_rate": 6.412571025452445e-05, "loss": 0.0126, "step": 59600 }, { "epoch": 4.136274656119099e-13, "grad_norm": 0.4189709722995758, "learning_rate": 6.410488980524616e-05, "loss": 0.0126, "step": 59610 }, { "epoch": 4.1369685455094896e-13, "grad_norm": 0.40490779280662537, "learning_rate": 6.408407124635164e-05, "loss": 0.0126, "step": 59620 }, { "epoch": 4.1376624348998803e-13, "grad_norm": 0.41686350107192993, "learning_rate": 6.406325457998036e-05, "loss": 0.0127, "step": 59630 }, { "epoch": 4.138356324290271e-13, "grad_norm": 0.35959571599960327, "learning_rate": 6.404243980827153e-05, "loss": 0.0126, "step": 59640 }, { "epoch": 4.1390502136806617e-13, "grad_norm": 0.4305672347545624, "learning_rate": 6.402162693336421e-05, "loss": 0.0136, "step": 59650 }, { "epoch": 4.1397441030710525e-13, "grad_norm": 0.3907923996448517, "learning_rate": 6.400081595739723e-05, "loss": 0.0126, "step": 59660 }, { "epoch": 4.140437992461443e-13, "grad_norm": 0.36560893058776855, "learning_rate": 6.39800068825092e-05, "loss": 0.0128, "step": 59670 }, { "epoch": 4.141131881851834e-13, "grad_norm": 0.3830989599227905, "learning_rate": 6.395919971083868e-05, "loss": 0.0134, "step": 59680 }, { "epoch": 4.1418257712422246e-13, "grad_norm": 0.3578548729419708, "learning_rate": 6.393839444452383e-05, "loss": 0.013, "step": 59690 }, { "epoch": 4.1425196606326153e-13, "grad_norm": 0.35648632049560547, "learning_rate": 6.391759108570278e-05, "loss": 0.013, "step": 59700 }, { "epoch": 4.143213550023006e-13, "grad_norm": 0.37653711438179016, "learning_rate": 6.389678963651335e-05, "loss": 0.0134, "step": 59710 }, { "epoch": 4.143907439413397e-13, "grad_norm": 0.42285382747650146, "learning_rate": 6.387599009909327e-05, "loss": 0.0126, "step": 59720 }, { "epoch": 4.1446013288037875e-13, "grad_norm": 0.42346200346946716, "learning_rate": 6.385519247557997e-05, "loss": 0.0128, "step": 59730 }, { "epoch": 4.145295218194178e-13, "grad_norm": 0.3919401168823242, "learning_rate": 6.383439676811073e-05, "loss": 0.0135, "step": 59740 }, { "epoch": 4.145989107584569e-13, "grad_norm": 0.37839797139167786, "learning_rate": 6.381360297882267e-05, "loss": 0.0128, "step": 59750 }, { "epoch": 4.1466829969749597e-13, "grad_norm": 0.3675868809223175, "learning_rate": 6.379281110985267e-05, "loss": 0.0126, "step": 59760 }, { "epoch": 4.1473768863653504e-13, "grad_norm": 0.40141671895980835, "learning_rate": 6.377202116333734e-05, "loss": 0.0129, "step": 59770 }, { "epoch": 4.148070775755741e-13, "grad_norm": 0.41552606225013733, "learning_rate": 6.375123314141329e-05, "loss": 0.0126, "step": 59780 }, { "epoch": 4.148764665146132e-13, "grad_norm": 0.39737042784690857, "learning_rate": 6.373044704621668e-05, "loss": 0.0124, "step": 59790 }, { "epoch": 4.1494585545365226e-13, "grad_norm": 0.39039474725723267, "learning_rate": 6.370966287988374e-05, "loss": 0.0128, "step": 59800 }, { "epoch": 4.1501524439269133e-13, "grad_norm": 0.3937103748321533, "learning_rate": 6.368888064455026e-05, "loss": 0.0129, "step": 59810 }, { "epoch": 4.150846333317304e-13, "grad_norm": 0.38824400305747986, "learning_rate": 6.366810034235202e-05, "loss": 0.0126, "step": 59820 }, { "epoch": 4.1515402227076947e-13, "grad_norm": 0.3740406334400177, "learning_rate": 6.364732197542448e-05, "loss": 0.0131, "step": 59830 }, { "epoch": 4.1522341120980855e-13, "grad_norm": 0.40364789962768555, "learning_rate": 6.362654554590293e-05, "loss": 0.0128, "step": 59840 }, { "epoch": 4.152928001488476e-13, "grad_norm": 0.3836596608161926, "learning_rate": 6.360577105592248e-05, "loss": 0.0128, "step": 59850 }, { "epoch": 4.153621890878867e-13, "grad_norm": 0.3887212872505188, "learning_rate": 6.358499850761805e-05, "loss": 0.013, "step": 59860 }, { "epoch": 4.1543157802692576e-13, "grad_norm": 0.38160479068756104, "learning_rate": 6.35642279031243e-05, "loss": 0.0126, "step": 59870 }, { "epoch": 4.1550096696596484e-13, "grad_norm": 0.41429826617240906, "learning_rate": 6.354345924457578e-05, "loss": 0.0134, "step": 59880 }, { "epoch": 4.155703559050039e-13, "grad_norm": 0.4106445908546448, "learning_rate": 6.352269253410675e-05, "loss": 0.0131, "step": 59890 }, { "epoch": 4.15639744844043e-13, "grad_norm": 0.3748418688774109, "learning_rate": 6.350192777385138e-05, "loss": 0.0133, "step": 59900 }, { "epoch": 4.1570913378308205e-13, "grad_norm": 0.4186420738697052, "learning_rate": 6.348116496594349e-05, "loss": 0.0129, "step": 59910 }, { "epoch": 4.157785227221211e-13, "grad_norm": 0.3493199050426483, "learning_rate": 6.346040411251685e-05, "loss": 0.0124, "step": 59920 }, { "epoch": 4.158479116611602e-13, "grad_norm": 0.3513109087944031, "learning_rate": 6.343964521570491e-05, "loss": 0.0127, "step": 59930 }, { "epoch": 4.1591730060019927e-13, "grad_norm": 0.37526512145996094, "learning_rate": 6.341888827764099e-05, "loss": 0.0124, "step": 59940 }, { "epoch": 4.1598668953923834e-13, "grad_norm": 0.4001331627368927, "learning_rate": 6.339813330045819e-05, "loss": 0.0125, "step": 59950 }, { "epoch": 4.160560784782774e-13, "grad_norm": 0.38554689288139343, "learning_rate": 6.337738028628942e-05, "loss": 0.0132, "step": 59960 }, { "epoch": 4.161254674173165e-13, "grad_norm": 0.3872356712818146, "learning_rate": 6.335662923726735e-05, "loss": 0.0122, "step": 59970 }, { "epoch": 4.1619485635635556e-13, "grad_norm": 0.47053053975105286, "learning_rate": 6.333588015552447e-05, "loss": 0.0125, "step": 59980 }, { "epoch": 4.1626424529539463e-13, "grad_norm": 0.3781798481941223, "learning_rate": 6.331513304319309e-05, "loss": 0.013, "step": 59990 }, { "epoch": 4.163336342344337e-13, "grad_norm": 0.4012109935283661, "learning_rate": 6.329438790240531e-05, "loss": 0.0125, "step": 60000 }, { "epoch": 4.164030231734728e-13, "grad_norm": 0.4141157269477844, "learning_rate": 6.327364473529297e-05, "loss": 0.013, "step": 60010 }, { "epoch": 4.1647241211251185e-13, "grad_norm": 0.37243983149528503, "learning_rate": 6.325290354398781e-05, "loss": 0.0131, "step": 60020 }, { "epoch": 4.165418010515509e-13, "grad_norm": 0.3741716742515564, "learning_rate": 6.323216433062126e-05, "loss": 0.0124, "step": 60030 }, { "epoch": 4.1661118999059e-13, "grad_norm": 0.39346545934677124, "learning_rate": 6.321142709732466e-05, "loss": 0.0133, "step": 60040 }, { "epoch": 4.1668057892962906e-13, "grad_norm": 0.39337044954299927, "learning_rate": 6.319069184622902e-05, "loss": 0.0138, "step": 60050 }, { "epoch": 4.1674996786866814e-13, "grad_norm": 0.35488060116767883, "learning_rate": 6.316995857946523e-05, "loss": 0.013, "step": 60060 }, { "epoch": 4.168193568077072e-13, "grad_norm": 0.3747446537017822, "learning_rate": 6.314922729916395e-05, "loss": 0.0122, "step": 60070 }, { "epoch": 4.168887457467463e-13, "grad_norm": 0.41318026185035706, "learning_rate": 6.312849800745566e-05, "loss": 0.0124, "step": 60080 }, { "epoch": 4.1695813468578535e-13, "grad_norm": 0.37613919377326965, "learning_rate": 6.310777070647063e-05, "loss": 0.0121, "step": 60090 }, { "epoch": 4.1702752362482443e-13, "grad_norm": 0.3730739653110504, "learning_rate": 6.308704539833884e-05, "loss": 0.0122, "step": 60100 }, { "epoch": 4.170969125638635e-13, "grad_norm": 0.39427703619003296, "learning_rate": 6.306632208519022e-05, "loss": 0.0122, "step": 60110 }, { "epoch": 4.1716630150290257e-13, "grad_norm": 0.34340184926986694, "learning_rate": 6.304560076915436e-05, "loss": 0.0134, "step": 60120 }, { "epoch": 4.1723569044194164e-13, "grad_norm": 0.3620416820049286, "learning_rate": 6.302488145236074e-05, "loss": 0.0121, "step": 60130 }, { "epoch": 4.173050793809807e-13, "grad_norm": 0.3489490747451782, "learning_rate": 6.300416413693853e-05, "loss": 0.0125, "step": 60140 }, { "epoch": 4.173744683200198e-13, "grad_norm": 0.3735385835170746, "learning_rate": 6.298344882501684e-05, "loss": 0.0131, "step": 60150 }, { "epoch": 4.1744385725905886e-13, "grad_norm": 0.3703104257583618, "learning_rate": 6.296273551872444e-05, "loss": 0.0128, "step": 60160 }, { "epoch": 4.1751324619809793e-13, "grad_norm": 0.42918866872787476, "learning_rate": 6.294202422018994e-05, "loss": 0.0129, "step": 60170 }, { "epoch": 4.17582635137137e-13, "grad_norm": 0.3388691544532776, "learning_rate": 6.292131493154177e-05, "loss": 0.0129, "step": 60180 }, { "epoch": 4.176520240761761e-13, "grad_norm": 0.38362374901771545, "learning_rate": 6.290060765490812e-05, "loss": 0.0125, "step": 60190 }, { "epoch": 4.1772141301521515e-13, "grad_norm": 0.3639390468597412, "learning_rate": 6.287990239241698e-05, "loss": 0.013, "step": 60200 }, { "epoch": 4.177908019542542e-13, "grad_norm": 0.40243589878082275, "learning_rate": 6.285919914619616e-05, "loss": 0.0134, "step": 60210 }, { "epoch": 4.178601908932933e-13, "grad_norm": 0.35123035311698914, "learning_rate": 6.283849791837321e-05, "loss": 0.0124, "step": 60220 }, { "epoch": 4.1792957983233237e-13, "grad_norm": 0.33194535970687866, "learning_rate": 6.281779871107551e-05, "loss": 0.012, "step": 60230 }, { "epoch": 4.1799896877137144e-13, "grad_norm": 0.3816789388656616, "learning_rate": 6.279710152643025e-05, "loss": 0.0124, "step": 60240 }, { "epoch": 4.180683577104105e-13, "grad_norm": 0.3555672764778137, "learning_rate": 6.277640636656436e-05, "loss": 0.0119, "step": 60250 }, { "epoch": 4.181377466494496e-13, "grad_norm": 0.4037970006465912, "learning_rate": 6.275571323360461e-05, "loss": 0.0121, "step": 60260 }, { "epoch": 4.1820713558848865e-13, "grad_norm": 0.33996254205703735, "learning_rate": 6.273502212967754e-05, "loss": 0.0129, "step": 60270 }, { "epoch": 4.1827652452752773e-13, "grad_norm": 0.3480628728866577, "learning_rate": 6.271433305690948e-05, "loss": 0.013, "step": 60280 }, { "epoch": 4.183459134665668e-13, "grad_norm": 0.37435901165008545, "learning_rate": 6.269364601742654e-05, "loss": 0.0121, "step": 60290 }, { "epoch": 4.1841530240560587e-13, "grad_norm": 0.4637153446674347, "learning_rate": 6.267296101335463e-05, "loss": 0.0132, "step": 60300 }, { "epoch": 4.1848469134464494e-13, "grad_norm": 0.4018581807613373, "learning_rate": 6.265227804681949e-05, "loss": 0.0127, "step": 60310 }, { "epoch": 4.18554080283684e-13, "grad_norm": 0.3949790298938751, "learning_rate": 6.263159711994656e-05, "loss": 0.0125, "step": 60320 }, { "epoch": 4.186234692227231e-13, "grad_norm": 0.36742475628852844, "learning_rate": 6.261091823486122e-05, "loss": 0.0133, "step": 60330 }, { "epoch": 4.1869285816176216e-13, "grad_norm": 0.3915010988712311, "learning_rate": 6.259024139368845e-05, "loss": 0.0125, "step": 60340 }, { "epoch": 4.1876224710080123e-13, "grad_norm": 0.404476135969162, "learning_rate": 6.25695665985532e-05, "loss": 0.013, "step": 60350 }, { "epoch": 4.188316360398403e-13, "grad_norm": 0.4095889627933502, "learning_rate": 6.254889385158004e-05, "loss": 0.0127, "step": 60360 }, { "epoch": 4.189010249788794e-13, "grad_norm": 0.37967392802238464, "learning_rate": 6.25282231548935e-05, "loss": 0.0128, "step": 60370 }, { "epoch": 4.1897041391791845e-13, "grad_norm": 0.3586665093898773, "learning_rate": 6.250755451061775e-05, "loss": 0.0122, "step": 60380 }, { "epoch": 4.190398028569575e-13, "grad_norm": 0.3652138411998749, "learning_rate": 6.248688792087689e-05, "loss": 0.0117, "step": 60390 }, { "epoch": 4.191091917959966e-13, "grad_norm": 0.40149521827697754, "learning_rate": 6.246622338779463e-05, "loss": 0.0124, "step": 60400 }, { "epoch": 4.1917858073503567e-13, "grad_norm": 0.3844377398490906, "learning_rate": 6.244556091349466e-05, "loss": 0.0122, "step": 60410 }, { "epoch": 4.1924796967407474e-13, "grad_norm": 0.3228718936443329, "learning_rate": 6.24249005001003e-05, "loss": 0.0121, "step": 60420 }, { "epoch": 4.193173586131138e-13, "grad_norm": 0.3477684557437897, "learning_rate": 6.240424214973483e-05, "loss": 0.0131, "step": 60430 }, { "epoch": 4.193867475521529e-13, "grad_norm": 0.3556922376155853, "learning_rate": 6.23835858645211e-05, "loss": 0.0124, "step": 60440 }, { "epoch": 4.1945613649119196e-13, "grad_norm": 0.3913474977016449, "learning_rate": 6.236293164658197e-05, "loss": 0.0124, "step": 60450 }, { "epoch": 4.1952552543023103e-13, "grad_norm": 0.3479597270488739, "learning_rate": 6.234227949803989e-05, "loss": 0.0121, "step": 60460 }, { "epoch": 4.195949143692701e-13, "grad_norm": 0.39468899369239807, "learning_rate": 6.232162942101726e-05, "loss": 0.0125, "step": 60470 }, { "epoch": 4.1966430330830917e-13, "grad_norm": 0.4254477620124817, "learning_rate": 6.230098141763615e-05, "loss": 0.0126, "step": 60480 }, { "epoch": 4.1973369224734824e-13, "grad_norm": 0.381817489862442, "learning_rate": 6.228033549001852e-05, "loss": 0.0132, "step": 60490 }, { "epoch": 4.198030811863873e-13, "grad_norm": 0.3770514130592346, "learning_rate": 6.225969164028595e-05, "loss": 0.0123, "step": 60500 }, { "epoch": 4.198724701254264e-13, "grad_norm": 0.34101632237434387, "learning_rate": 6.223904987056005e-05, "loss": 0.0126, "step": 60510 }, { "epoch": 4.1994185906446546e-13, "grad_norm": 0.3630281686782837, "learning_rate": 6.221841018296198e-05, "loss": 0.0131, "step": 60520 }, { "epoch": 4.2001124800350453e-13, "grad_norm": 0.37383899092674255, "learning_rate": 6.219777257961289e-05, "loss": 0.0124, "step": 60530 }, { "epoch": 4.200806369425436e-13, "grad_norm": 0.3648926913738251, "learning_rate": 6.217713706263351e-05, "loss": 0.0118, "step": 60540 }, { "epoch": 4.201500258815827e-13, "grad_norm": 0.3767421245574951, "learning_rate": 6.21565036341445e-05, "loss": 0.0123, "step": 60550 }, { "epoch": 4.2021941482062175e-13, "grad_norm": 0.3699710965156555, "learning_rate": 6.213587229626632e-05, "loss": 0.0122, "step": 60560 }, { "epoch": 4.202888037596608e-13, "grad_norm": 0.3850472867488861, "learning_rate": 6.211524305111906e-05, "loss": 0.0124, "step": 60570 }, { "epoch": 4.203581926986999e-13, "grad_norm": 0.43244120478630066, "learning_rate": 6.209461590082278e-05, "loss": 0.0127, "step": 60580 }, { "epoch": 4.2042758163773897e-13, "grad_norm": 0.36051222681999207, "learning_rate": 6.207399084749719e-05, "loss": 0.0123, "step": 60590 }, { "epoch": 4.2049697057677804e-13, "grad_norm": 0.38250890374183655, "learning_rate": 6.205336789326189e-05, "loss": 0.0125, "step": 60600 }, { "epoch": 4.205663595158171e-13, "grad_norm": 0.3808608651161194, "learning_rate": 6.203274704023614e-05, "loss": 0.0122, "step": 60610 }, { "epoch": 4.206357484548562e-13, "grad_norm": 0.3701058328151703, "learning_rate": 6.201212829053911e-05, "loss": 0.0121, "step": 60620 }, { "epoch": 4.2070513739389526e-13, "grad_norm": 0.43147021532058716, "learning_rate": 6.199151164628965e-05, "loss": 0.0129, "step": 60630 }, { "epoch": 4.2077452633293433e-13, "grad_norm": 0.3918412923812866, "learning_rate": 6.19708971096065e-05, "loss": 0.0127, "step": 60640 }, { "epoch": 4.208439152719734e-13, "grad_norm": 0.3705867826938629, "learning_rate": 6.195028468260806e-05, "loss": 0.0124, "step": 60650 }, { "epoch": 4.2091330421101247e-13, "grad_norm": 0.3572317957878113, "learning_rate": 6.192967436741264e-05, "loss": 0.0127, "step": 60660 }, { "epoch": 4.2098269315005155e-13, "grad_norm": 0.42390602827072144, "learning_rate": 6.19090661661382e-05, "loss": 0.0123, "step": 60670 }, { "epoch": 4.210520820890906e-13, "grad_norm": 0.34442147612571716, "learning_rate": 6.188846008090262e-05, "loss": 0.0129, "step": 60680 }, { "epoch": 4.211214710281297e-13, "grad_norm": 0.33868053555488586, "learning_rate": 6.186785611382344e-05, "loss": 0.012, "step": 60690 }, { "epoch": 4.2119085996716876e-13, "grad_norm": 0.3729186952114105, "learning_rate": 6.184725426701812e-05, "loss": 0.0121, "step": 60700 }, { "epoch": 4.2126024890620783e-13, "grad_norm": 0.33124950528144836, "learning_rate": 6.18266545426037e-05, "loss": 0.0124, "step": 60710 }, { "epoch": 4.213296378452469e-13, "grad_norm": 0.34845200181007385, "learning_rate": 6.180605694269726e-05, "loss": 0.0116, "step": 60720 }, { "epoch": 4.21399026784286e-13, "grad_norm": 0.3573269844055176, "learning_rate": 6.17854614694154e-05, "loss": 0.0124, "step": 60730 }, { "epoch": 4.2146841572332505e-13, "grad_norm": 0.372366726398468, "learning_rate": 6.17648681248747e-05, "loss": 0.0125, "step": 60740 }, { "epoch": 4.215378046623641e-13, "grad_norm": 0.37740007042884827, "learning_rate": 6.174427691119139e-05, "loss": 0.0123, "step": 60750 }, { "epoch": 4.216071936014032e-13, "grad_norm": 0.4107911288738251, "learning_rate": 6.172368783048162e-05, "loss": 0.0121, "step": 60760 }, { "epoch": 4.2167658254044227e-13, "grad_norm": 0.4010809361934662, "learning_rate": 6.170310088486116e-05, "loss": 0.0124, "step": 60770 }, { "epoch": 4.2174597147948134e-13, "grad_norm": 0.3813934624195099, "learning_rate": 6.168251607644568e-05, "loss": 0.0116, "step": 60780 }, { "epoch": 4.218153604185204e-13, "grad_norm": 0.38649794459342957, "learning_rate": 6.166193340735056e-05, "loss": 0.0123, "step": 60790 }, { "epoch": 4.218847493575595e-13, "grad_norm": 0.36449936032295227, "learning_rate": 6.164135287969106e-05, "loss": 0.0129, "step": 60800 }, { "epoch": 4.2195413829659856e-13, "grad_norm": 0.35765039920806885, "learning_rate": 6.162077449558206e-05, "loss": 0.0125, "step": 60810 }, { "epoch": 4.2202352723563763e-13, "grad_norm": 0.39329713582992554, "learning_rate": 6.160019825713836e-05, "loss": 0.0126, "step": 60820 }, { "epoch": 4.220929161746767e-13, "grad_norm": 0.40433722734451294, "learning_rate": 6.157962416647446e-05, "loss": 0.0127, "step": 60830 }, { "epoch": 4.2216230511371577e-13, "grad_norm": 0.3869937062263489, "learning_rate": 6.15590522257047e-05, "loss": 0.0126, "step": 60840 }, { "epoch": 4.2223169405275485e-13, "grad_norm": 0.3510833978652954, "learning_rate": 6.153848243694313e-05, "loss": 0.0118, "step": 60850 }, { "epoch": 4.223010829917939e-13, "grad_norm": 0.39117106795310974, "learning_rate": 6.151791480230365e-05, "loss": 0.0122, "step": 60860 }, { "epoch": 4.22370471930833e-13, "grad_norm": 0.39710912108421326, "learning_rate": 6.149734932389988e-05, "loss": 0.012, "step": 60870 }, { "epoch": 4.2243986086987206e-13, "grad_norm": 0.3726847767829895, "learning_rate": 6.147678600384527e-05, "loss": 0.012, "step": 60880 }, { "epoch": 4.2250924980891114e-13, "grad_norm": 0.3661019504070282, "learning_rate": 6.145622484425297e-05, "loss": 0.0123, "step": 60890 }, { "epoch": 4.225786387479502e-13, "grad_norm": 0.4046143889427185, "learning_rate": 6.143566584723606e-05, "loss": 0.0131, "step": 60900 }, { "epoch": 4.226480276869893e-13, "grad_norm": 0.37117910385131836, "learning_rate": 6.141510901490718e-05, "loss": 0.0118, "step": 60910 }, { "epoch": 4.2271741662602835e-13, "grad_norm": 0.3013368248939514, "learning_rate": 6.139455434937892e-05, "loss": 0.0119, "step": 60920 }, { "epoch": 4.227868055650674e-13, "grad_norm": 0.3629961907863617, "learning_rate": 6.13740018527636e-05, "loss": 0.0119, "step": 60930 }, { "epoch": 4.228561945041065e-13, "grad_norm": 0.3537103831768036, "learning_rate": 6.13534515271733e-05, "loss": 0.0122, "step": 60940 }, { "epoch": 4.2292558344314557e-13, "grad_norm": 0.35580500960350037, "learning_rate": 6.133290337471987e-05, "loss": 0.0122, "step": 60950 }, { "epoch": 4.2299497238218464e-13, "grad_norm": 0.3617808520793915, "learning_rate": 6.131235739751497e-05, "loss": 0.0125, "step": 60960 }, { "epoch": 4.230643613212237e-13, "grad_norm": 0.40696290135383606, "learning_rate": 6.129181359767e-05, "loss": 0.012, "step": 60970 }, { "epoch": 4.231337502602628e-13, "grad_norm": 0.39778417348861694, "learning_rate": 6.127127197729618e-05, "loss": 0.0123, "step": 60980 }, { "epoch": 4.2320313919930186e-13, "grad_norm": 0.3964042067527771, "learning_rate": 6.125073253850446e-05, "loss": 0.0126, "step": 60990 }, { "epoch": 4.2327252813834093e-13, "grad_norm": 0.3762957453727722, "learning_rate": 6.123019528340561e-05, "loss": 0.0123, "step": 61000 }, { "epoch": 4.2334191707738e-13, "grad_norm": 0.4044255316257477, "learning_rate": 6.120966021411014e-05, "loss": 0.012, "step": 61010 }, { "epoch": 4.234113060164191e-13, "grad_norm": 0.39368048310279846, "learning_rate": 6.11891273327283e-05, "loss": 0.0122, "step": 61020 }, { "epoch": 4.2348069495545815e-13, "grad_norm": 0.37560775876045227, "learning_rate": 6.116859664137027e-05, "loss": 0.0119, "step": 61030 }, { "epoch": 4.235500838944972e-13, "grad_norm": 0.3981151282787323, "learning_rate": 6.114806814214581e-05, "loss": 0.0121, "step": 61040 }, { "epoch": 4.236194728335363e-13, "grad_norm": 0.38676008582115173, "learning_rate": 6.112754183716459e-05, "loss": 0.0124, "step": 61050 }, { "epoch": 4.2368886177257536e-13, "grad_norm": 0.37760159373283386, "learning_rate": 6.110701772853595e-05, "loss": 0.0124, "step": 61060 }, { "epoch": 4.2375825071161444e-13, "grad_norm": 0.37996000051498413, "learning_rate": 6.108649581836913e-05, "loss": 0.0118, "step": 61070 }, { "epoch": 4.238276396506535e-13, "grad_norm": 0.38411659002304077, "learning_rate": 6.1065976108773e-05, "loss": 0.0119, "step": 61080 }, { "epoch": 4.238970285896926e-13, "grad_norm": 0.3510439693927765, "learning_rate": 6.104545860185637e-05, "loss": 0.0119, "step": 61090 }, { "epoch": 4.2396641752873165e-13, "grad_norm": 0.35197746753692627, "learning_rate": 6.102494329972766e-05, "loss": 0.012, "step": 61100 }, { "epoch": 4.2403580646777073e-13, "grad_norm": 0.3808964192867279, "learning_rate": 6.100443020449519e-05, "loss": 0.0125, "step": 61110 }, { "epoch": 4.241051954068098e-13, "grad_norm": 0.3529256582260132, "learning_rate": 6.098391931826694e-05, "loss": 0.0122, "step": 61120 }, { "epoch": 4.2417458434584887e-13, "grad_norm": 0.39880242943763733, "learning_rate": 6.0963410643150776e-05, "loss": 0.0126, "step": 61130 }, { "epoch": 4.2424397328488794e-13, "grad_norm": 0.4025323688983917, "learning_rate": 6.094290418125426e-05, "loss": 0.0128, "step": 61140 }, { "epoch": 4.24313362223927e-13, "grad_norm": 0.3745158612728119, "learning_rate": 6.0922399934684766e-05, "loss": 0.0125, "step": 61150 }, { "epoch": 4.243827511629661e-13, "grad_norm": 0.42155885696411133, "learning_rate": 6.0901897905549384e-05, "loss": 0.012, "step": 61160 }, { "epoch": 4.2445214010200516e-13, "grad_norm": 0.3404020369052887, "learning_rate": 6.088139809595506e-05, "loss": 0.0117, "step": 61170 }, { "epoch": 4.2452152904104423e-13, "grad_norm": 0.2961239218711853, "learning_rate": 6.086090050800845e-05, "loss": 0.012, "step": 61180 }, { "epoch": 4.245909179800833e-13, "grad_norm": 0.36970967054367065, "learning_rate": 6.0840405143816004e-05, "loss": 0.0118, "step": 61190 }, { "epoch": 4.246603069191224e-13, "grad_norm": 0.36786550283432007, "learning_rate": 6.081991200548392e-05, "loss": 0.0128, "step": 61200 }, { "epoch": 4.2472969585816145e-13, "grad_norm": 0.37537872791290283, "learning_rate": 6.0799421095118246e-05, "loss": 0.0121, "step": 61210 }, { "epoch": 4.247990847972005e-13, "grad_norm": 0.3818269670009613, "learning_rate": 6.077893241482465e-05, "loss": 0.012, "step": 61220 }, { "epoch": 4.248684737362396e-13, "grad_norm": 0.36727938055992126, "learning_rate": 6.075844596670875e-05, "loss": 0.0118, "step": 61230 }, { "epoch": 4.2493786267527867e-13, "grad_norm": 0.36338895559310913, "learning_rate": 6.07379617528758e-05, "loss": 0.0122, "step": 61240 }, { "epoch": 4.2500725161431774e-13, "grad_norm": 0.38993608951568604, "learning_rate": 6.0717479775430886e-05, "loss": 0.0115, "step": 61250 }, { "epoch": 4.250766405533568e-13, "grad_norm": 0.3652445077896118, "learning_rate": 6.069700003647884e-05, "loss": 0.0121, "step": 61260 }, { "epoch": 4.251460294923959e-13, "grad_norm": 0.36016806960105896, "learning_rate": 6.067652253812428e-05, "loss": 0.0123, "step": 61270 }, { "epoch": 4.2521541843143495e-13, "grad_norm": 0.36087629199028015, "learning_rate": 6.065604728247159e-05, "loss": 0.0124, "step": 61280 }, { "epoch": 4.2528480737047403e-13, "grad_norm": 0.3700447082519531, "learning_rate": 6.0635574271624935e-05, "loss": 0.0126, "step": 61290 }, { "epoch": 4.253541963095131e-13, "grad_norm": 0.3627805709838867, "learning_rate": 6.061510350768817e-05, "loss": 0.0126, "step": 61300 }, { "epoch": 4.2542358524855217e-13, "grad_norm": 0.3492460548877716, "learning_rate": 6.059463499276508e-05, "loss": 0.0121, "step": 61310 }, { "epoch": 4.2549297418759124e-13, "grad_norm": 0.3798457682132721, "learning_rate": 6.057416872895905e-05, "loss": 0.0117, "step": 61320 }, { "epoch": 4.255623631266303e-13, "grad_norm": 0.3870060443878174, "learning_rate": 6.0553704718373346e-05, "loss": 0.012, "step": 61330 }, { "epoch": 4.256317520656694e-13, "grad_norm": 0.3572353422641754, "learning_rate": 6.053324296311096e-05, "loss": 0.0116, "step": 61340 }, { "epoch": 4.2570114100470846e-13, "grad_norm": 0.35640788078308105, "learning_rate": 6.0512783465274644e-05, "loss": 0.0115, "step": 61350 }, { "epoch": 4.2577052994374753e-13, "grad_norm": 0.36455923318862915, "learning_rate": 6.049232622696693e-05, "loss": 0.0128, "step": 61360 }, { "epoch": 4.258399188827866e-13, "grad_norm": 0.3367912173271179, "learning_rate": 6.047187125029012e-05, "loss": 0.0125, "step": 61370 }, { "epoch": 4.259093078218257e-13, "grad_norm": 0.3221932649612427, "learning_rate": 6.045141853734627e-05, "loss": 0.0115, "step": 61380 }, { "epoch": 4.2597869676086475e-13, "grad_norm": 0.31275174021720886, "learning_rate": 6.0430968090237256e-05, "loss": 0.0118, "step": 61390 }, { "epoch": 4.260480856999038e-13, "grad_norm": 0.3362356424331665, "learning_rate": 6.0410519911064605e-05, "loss": 0.0123, "step": 61400 }, { "epoch": 4.261174746389429e-13, "grad_norm": 0.37017419934272766, "learning_rate": 6.039007400192977e-05, "loss": 0.0118, "step": 61410 }, { "epoch": 4.2618686357798197e-13, "grad_norm": 0.4053521752357483, "learning_rate": 6.036963036493382e-05, "loss": 0.0127, "step": 61420 }, { "epoch": 4.2625625251702104e-13, "grad_norm": 0.35259220004081726, "learning_rate": 6.034918900217769e-05, "loss": 0.0119, "step": 61430 }, { "epoch": 4.263256414560601e-13, "grad_norm": 0.3444765508174896, "learning_rate": 6.032874991576207e-05, "loss": 0.0122, "step": 61440 }, { "epoch": 4.263950303950992e-13, "grad_norm": 0.37417685985565186, "learning_rate": 6.030831310778733e-05, "loss": 0.0124, "step": 61450 }, { "epoch": 4.2646441933413826e-13, "grad_norm": 0.3695544898509979, "learning_rate": 6.0287878580353734e-05, "loss": 0.0126, "step": 61460 }, { "epoch": 4.2653380827317733e-13, "grad_norm": 0.3219206929206848, "learning_rate": 6.026744633556119e-05, "loss": 0.0118, "step": 61470 }, { "epoch": 4.266031972122164e-13, "grad_norm": 0.3412145972251892, "learning_rate": 6.0247016375509484e-05, "loss": 0.0117, "step": 61480 }, { "epoch": 4.2667258615125547e-13, "grad_norm": 0.354159951210022, "learning_rate": 6.022658870229805e-05, "loss": 0.012, "step": 61490 }, { "epoch": 4.2674197509029455e-13, "grad_norm": 0.33709341287612915, "learning_rate": 6.020616331802624e-05, "loss": 0.0118, "step": 61500 }, { "epoch": 4.268113640293336e-13, "grad_norm": 0.39689382910728455, "learning_rate": 6.0185740224792974e-05, "loss": 0.0118, "step": 61510 }, { "epoch": 4.268807529683727e-13, "grad_norm": 0.351919561624527, "learning_rate": 6.016531942469714e-05, "loss": 0.0127, "step": 61520 }, { "epoch": 4.2695014190741176e-13, "grad_norm": 0.3329906761646271, "learning_rate": 6.0144900919837205e-05, "loss": 0.0121, "step": 61530 }, { "epoch": 4.2701953084645083e-13, "grad_norm": 0.3274400234222412, "learning_rate": 6.0124484712311564e-05, "loss": 0.0117, "step": 61540 }, { "epoch": 4.270889197854899e-13, "grad_norm": 0.3947077691555023, "learning_rate": 6.0104070804218234e-05, "loss": 0.012, "step": 61550 }, { "epoch": 4.27158308724529e-13, "grad_norm": 0.3456667363643646, "learning_rate": 6.008365919765512e-05, "loss": 0.0121, "step": 61560 }, { "epoch": 4.2722769766356805e-13, "grad_norm": 0.3829488158226013, "learning_rate": 6.0063249894719796e-05, "loss": 0.0118, "step": 61570 }, { "epoch": 4.272970866026071e-13, "grad_norm": 0.3855654001235962, "learning_rate": 6.004284289750965e-05, "loss": 0.0129, "step": 61580 }, { "epoch": 4.273664755416462e-13, "grad_norm": 0.37017330527305603, "learning_rate": 6.0022438208121796e-05, "loss": 0.0117, "step": 61590 }, { "epoch": 4.2743586448068527e-13, "grad_norm": 0.3897968828678131, "learning_rate": 6.000203582865319e-05, "loss": 0.0124, "step": 61600 }, { "epoch": 4.2750525341972434e-13, "grad_norm": 0.3977571129798889, "learning_rate": 5.9981635761200426e-05, "loss": 0.0122, "step": 61610 }, { "epoch": 4.275746423587634e-13, "grad_norm": 0.3893416225910187, "learning_rate": 5.996123800785998e-05, "loss": 0.0122, "step": 61620 }, { "epoch": 4.276440312978025e-13, "grad_norm": 0.3702080249786377, "learning_rate": 5.9940842570727976e-05, "loss": 0.012, "step": 61630 }, { "epoch": 4.2771342023684156e-13, "grad_norm": 0.3741549849510193, "learning_rate": 5.9920449451900436e-05, "loss": 0.0117, "step": 61640 }, { "epoch": 4.2778280917588063e-13, "grad_norm": 0.3803284764289856, "learning_rate": 5.990005865347302e-05, "loss": 0.0118, "step": 61650 }, { "epoch": 4.278521981149197e-13, "grad_norm": 0.36180561780929565, "learning_rate": 5.987967017754125e-05, "loss": 0.0114, "step": 61660 }, { "epoch": 4.2792158705395877e-13, "grad_norm": 0.3064892888069153, "learning_rate": 5.98592840262003e-05, "loss": 0.0119, "step": 61670 }, { "epoch": 4.2799097599299785e-13, "grad_norm": 0.3438263237476349, "learning_rate": 5.9838900201545195e-05, "loss": 0.0125, "step": 61680 }, { "epoch": 4.280603649320369e-13, "grad_norm": 0.3568100035190582, "learning_rate": 5.981851870567067e-05, "loss": 0.0113, "step": 61690 }, { "epoch": 4.28129753871076e-13, "grad_norm": 0.3720785975456238, "learning_rate": 5.9798139540671294e-05, "loss": 0.0118, "step": 61700 }, { "epoch": 4.2819914281011506e-13, "grad_norm": 0.35333219170570374, "learning_rate": 5.977776270864127e-05, "loss": 0.0119, "step": 61710 }, { "epoch": 4.2826853174915414e-13, "grad_norm": 0.37341076135635376, "learning_rate": 5.975738821167468e-05, "loss": 0.012, "step": 61720 }, { "epoch": 4.283379206881932e-13, "grad_norm": 0.4000820517539978, "learning_rate": 5.9737016051865296e-05, "loss": 0.0117, "step": 61730 }, { "epoch": 4.284073096272323e-13, "grad_norm": 0.40166008472442627, "learning_rate": 5.9716646231306716e-05, "loss": 0.0124, "step": 61740 }, { "epoch": 4.2847669856627135e-13, "grad_norm": 0.3923890292644501, "learning_rate": 5.9696278752092215e-05, "loss": 0.0117, "step": 61750 }, { "epoch": 4.285460875053104e-13, "grad_norm": 0.32922694087028503, "learning_rate": 5.9675913616314885e-05, "loss": 0.0124, "step": 61760 }, { "epoch": 4.286154764443495e-13, "grad_norm": 0.3658309280872345, "learning_rate": 5.965555082606756e-05, "loss": 0.0123, "step": 61770 }, { "epoch": 4.2868486538338857e-13, "grad_norm": 0.3860469460487366, "learning_rate": 5.963519038344283e-05, "loss": 0.0121, "step": 61780 }, { "epoch": 4.2875425432242764e-13, "grad_norm": 0.39559513330459595, "learning_rate": 5.9614832290533044e-05, "loss": 0.0116, "step": 61790 }, { "epoch": 4.288236432614667e-13, "grad_norm": 0.35658127069473267, "learning_rate": 5.9594476549430335e-05, "loss": 0.0122, "step": 61800 }, { "epoch": 4.288930322005058e-13, "grad_norm": 0.34007832407951355, "learning_rate": 5.957412316222654e-05, "loss": 0.0117, "step": 61810 }, { "epoch": 4.2896242113954486e-13, "grad_norm": 0.37189212441444397, "learning_rate": 5.955377213101331e-05, "loss": 0.0114, "step": 61820 }, { "epoch": 4.2903181007858393e-13, "grad_norm": 0.38932004570961, "learning_rate": 5.9533423457881994e-05, "loss": 0.0121, "step": 61830 }, { "epoch": 4.29101199017623e-13, "grad_norm": 0.3252717852592468, "learning_rate": 5.95130771449238e-05, "loss": 0.012, "step": 61840 }, { "epoch": 4.291705879566621e-13, "grad_norm": 0.35395193099975586, "learning_rate": 5.9492733194229565e-05, "loss": 0.0114, "step": 61850 }, { "epoch": 4.2923997689570115e-13, "grad_norm": 0.35428911447525024, "learning_rate": 5.9472391607889985e-05, "loss": 0.0113, "step": 61860 }, { "epoch": 4.293093658347402e-13, "grad_norm": 0.3514023423194885, "learning_rate": 5.9452052387995444e-05, "loss": 0.0115, "step": 61870 }, { "epoch": 4.293787547737793e-13, "grad_norm": 0.3580729365348816, "learning_rate": 5.943171553663614e-05, "loss": 0.012, "step": 61880 }, { "epoch": 4.2944814371281836e-13, "grad_norm": 0.3508799076080322, "learning_rate": 5.941138105590202e-05, "loss": 0.0119, "step": 61890 }, { "epoch": 4.2951753265185744e-13, "grad_norm": 0.338222473859787, "learning_rate": 5.93910489478827e-05, "loss": 0.0118, "step": 61900 }, { "epoch": 4.295869215908965e-13, "grad_norm": 0.38696932792663574, "learning_rate": 5.9370719214667676e-05, "loss": 0.012, "step": 61910 }, { "epoch": 4.296563105299356e-13, "grad_norm": 0.33621910214424133, "learning_rate": 5.9350391858346125e-05, "loss": 0.0119, "step": 61920 }, { "epoch": 4.2972569946897465e-13, "grad_norm": 0.3149675130844116, "learning_rate": 5.933006688100702e-05, "loss": 0.0122, "step": 61930 }, { "epoch": 4.2979508840801373e-13, "grad_norm": 0.3392586410045624, "learning_rate": 5.930974428473901e-05, "loss": 0.0118, "step": 61940 }, { "epoch": 4.298644773470528e-13, "grad_norm": 0.3687262535095215, "learning_rate": 5.928942407163065e-05, "loss": 0.0116, "step": 61950 }, { "epoch": 4.2993386628609187e-13, "grad_norm": 0.3318782150745392, "learning_rate": 5.926910624377005e-05, "loss": 0.0119, "step": 61960 }, { "epoch": 4.3000325522513094e-13, "grad_norm": 0.32855454087257385, "learning_rate": 5.924879080324529e-05, "loss": 0.0115, "step": 61970 }, { "epoch": 4.3007264416417e-13, "grad_norm": 0.3555127680301666, "learning_rate": 5.9228477752144015e-05, "loss": 0.0115, "step": 61980 }, { "epoch": 4.301420331032091e-13, "grad_norm": 0.43046724796295166, "learning_rate": 5.920816709255377e-05, "loss": 0.0122, "step": 61990 }, { "epoch": 4.3021142204224816e-13, "grad_norm": 0.3462871313095093, "learning_rate": 5.9187858826561734e-05, "loss": 0.0116, "step": 62000 }, { "epoch": 4.3028081098128723e-13, "grad_norm": 0.3828408122062683, "learning_rate": 5.916755295625494e-05, "loss": 0.0115, "step": 62010 }, { "epoch": 4.303501999203263e-13, "grad_norm": 0.3657103478908539, "learning_rate": 5.91472494837201e-05, "loss": 0.0119, "step": 62020 }, { "epoch": 4.304195888593654e-13, "grad_norm": 0.3841564953327179, "learning_rate": 5.912694841104377e-05, "loss": 0.0114, "step": 62030 }, { "epoch": 4.3048897779840445e-13, "grad_norm": 0.3405803442001343, "learning_rate": 5.91066497403121e-05, "loss": 0.0118, "step": 62040 }, { "epoch": 4.305583667374435e-13, "grad_norm": 0.3478187918663025, "learning_rate": 5.9086353473611204e-05, "loss": 0.0122, "step": 62050 }, { "epoch": 4.306277556764826e-13, "grad_norm": 0.3372678756713867, "learning_rate": 5.9066059613026754e-05, "loss": 0.0114, "step": 62060 }, { "epoch": 4.3069714461552167e-13, "grad_norm": 0.3788280487060547, "learning_rate": 5.904576816064433e-05, "loss": 0.0122, "step": 62070 }, { "epoch": 4.3076653355456074e-13, "grad_norm": 0.37060728669166565, "learning_rate": 5.902547911854911e-05, "loss": 0.012, "step": 62080 }, { "epoch": 4.308359224935998e-13, "grad_norm": 0.3482286334037781, "learning_rate": 5.9005192488826204e-05, "loss": 0.0117, "step": 62090 }, { "epoch": 4.309053114326389e-13, "grad_norm": 0.32581326365470886, "learning_rate": 5.89849082735603e-05, "loss": 0.0115, "step": 62100 }, { "epoch": 4.3097470037167795e-13, "grad_norm": 0.3813166618347168, "learning_rate": 5.896462647483598e-05, "loss": 0.0114, "step": 62110 }, { "epoch": 4.3104408931071703e-13, "grad_norm": 0.35309547185897827, "learning_rate": 5.8944347094737456e-05, "loss": 0.0114, "step": 62120 }, { "epoch": 4.311134782497561e-13, "grad_norm": 0.3609031140804291, "learning_rate": 5.892407013534882e-05, "loss": 0.0111, "step": 62130 }, { "epoch": 4.3118286718879517e-13, "grad_norm": 0.32274940609931946, "learning_rate": 5.890379559875375e-05, "loss": 0.0111, "step": 62140 }, { "epoch": 4.3125225612783424e-13, "grad_norm": 0.399610698223114, "learning_rate": 5.888352348703586e-05, "loss": 0.0132, "step": 62150 }, { "epoch": 4.313216450668733e-13, "grad_norm": 0.37607184052467346, "learning_rate": 5.8863253802278356e-05, "loss": 0.0112, "step": 62160 }, { "epoch": 4.313910340059124e-13, "grad_norm": 0.3452204763889313, "learning_rate": 5.884298654656433e-05, "loss": 0.0117, "step": 62170 }, { "epoch": 4.3146042294495146e-13, "grad_norm": 0.3704957962036133, "learning_rate": 5.882272172197649e-05, "loss": 0.0116, "step": 62180 }, { "epoch": 4.3152981188399053e-13, "grad_norm": 0.3474009335041046, "learning_rate": 5.880245933059742e-05, "loss": 0.0115, "step": 62190 }, { "epoch": 4.315992008230296e-13, "grad_norm": 0.34023502469062805, "learning_rate": 5.878219937450936e-05, "loss": 0.0118, "step": 62200 }, { "epoch": 4.316685897620687e-13, "grad_norm": 0.3381686210632324, "learning_rate": 5.876194185579437e-05, "loss": 0.0122, "step": 62210 }, { "epoch": 4.3173797870110775e-13, "grad_norm": 0.3943679630756378, "learning_rate": 5.874168677653417e-05, "loss": 0.0117, "step": 62220 }, { "epoch": 4.318073676401468e-13, "grad_norm": 0.3719496726989746, "learning_rate": 5.872143413881033e-05, "loss": 0.0118, "step": 62230 }, { "epoch": 4.318767565791859e-13, "grad_norm": 0.38190510869026184, "learning_rate": 5.87011839447041e-05, "loss": 0.0122, "step": 62240 }, { "epoch": 4.3194614551822497e-13, "grad_norm": 0.356886088848114, "learning_rate": 5.8680936196296546e-05, "loss": 0.0115, "step": 62250 }, { "epoch": 4.3201553445726404e-13, "grad_norm": 0.3588060438632965, "learning_rate": 5.866069089566837e-05, "loss": 0.0116, "step": 62260 }, { "epoch": 4.320849233963031e-13, "grad_norm": 0.36875900626182556, "learning_rate": 5.8640448044900146e-05, "loss": 0.0117, "step": 62270 }, { "epoch": 4.321543123353422e-13, "grad_norm": 0.3518502116203308, "learning_rate": 5.86202076460721e-05, "loss": 0.0115, "step": 62280 }, { "epoch": 4.3222370127438126e-13, "grad_norm": 0.36238980293273926, "learning_rate": 5.859996970126431e-05, "loss": 0.0111, "step": 62290 }, { "epoch": 4.3229309021342033e-13, "grad_norm": 0.3593444526195526, "learning_rate": 5.8579734212556464e-05, "loss": 0.0119, "step": 62300 }, { "epoch": 4.323624791524594e-13, "grad_norm": 0.3708542585372925, "learning_rate": 5.855950118202814e-05, "loss": 0.0118, "step": 62310 }, { "epoch": 4.3243186809149847e-13, "grad_norm": 0.3392954170703888, "learning_rate": 5.853927061175856e-05, "loss": 0.011, "step": 62320 }, { "epoch": 4.3250125703053754e-13, "grad_norm": 0.3268490135669708, "learning_rate": 5.8519042503826745e-05, "loss": 0.0114, "step": 62330 }, { "epoch": 4.325706459695766e-13, "grad_norm": 0.3575190007686615, "learning_rate": 5.849881686031141e-05, "loss": 0.0115, "step": 62340 }, { "epoch": 4.326400349086157e-13, "grad_norm": 0.3314041197299957, "learning_rate": 5.847859368329111e-05, "loss": 0.0116, "step": 62350 }, { "epoch": 4.3270942384765476e-13, "grad_norm": 0.34486910700798035, "learning_rate": 5.845837297484408e-05, "loss": 0.012, "step": 62360 }, { "epoch": 4.3277881278669383e-13, "grad_norm": 0.3539644479751587, "learning_rate": 5.8438154737048246e-05, "loss": 0.0113, "step": 62370 }, { "epoch": 4.328482017257329e-13, "grad_norm": 0.36319977045059204, "learning_rate": 5.841793897198145e-05, "loss": 0.0118, "step": 62380 }, { "epoch": 4.32917590664772e-13, "grad_norm": 0.38633230328559875, "learning_rate": 5.8397725681721084e-05, "loss": 0.0121, "step": 62390 }, { "epoch": 4.3298697960381105e-13, "grad_norm": 0.38228821754455566, "learning_rate": 5.837751486834445e-05, "loss": 0.0123, "step": 62400 }, { "epoch": 4.330563685428501e-13, "grad_norm": 0.4014406204223633, "learning_rate": 5.8357306533928466e-05, "loss": 0.0115, "step": 62410 }, { "epoch": 4.331257574818892e-13, "grad_norm": 0.34704577922821045, "learning_rate": 5.833710068054991e-05, "loss": 0.0115, "step": 62420 }, { "epoch": 4.3319514642092827e-13, "grad_norm": 0.31622400879859924, "learning_rate": 5.831689731028518e-05, "loss": 0.0114, "step": 62430 }, { "epoch": 4.3326453535996734e-13, "grad_norm": 0.35335594415664673, "learning_rate": 5.829669642521056e-05, "loss": 0.0114, "step": 62440 }, { "epoch": 4.333339242990064e-13, "grad_norm": 0.35526734590530396, "learning_rate": 5.827649802740196e-05, "loss": 0.0114, "step": 62450 }, { "epoch": 4.334033132380455e-13, "grad_norm": 0.374310702085495, "learning_rate": 5.82563021189351e-05, "loss": 0.0118, "step": 62460 }, { "epoch": 4.3347270217708456e-13, "grad_norm": 0.36202579736709595, "learning_rate": 5.823610870188537e-05, "loss": 0.0116, "step": 62470 }, { "epoch": 4.3354209111612363e-13, "grad_norm": 0.35447490215301514, "learning_rate": 5.8215917778328044e-05, "loss": 0.0113, "step": 62480 }, { "epoch": 4.336114800551627e-13, "grad_norm": 0.33920660614967346, "learning_rate": 5.819572935033797e-05, "loss": 0.0114, "step": 62490 }, { "epoch": 4.3368086899420177e-13, "grad_norm": 0.36459583044052124, "learning_rate": 5.8175543419989906e-05, "loss": 0.0113, "step": 62500 }, { "epoch": 4.3375025793324085e-13, "grad_norm": 0.3307925760746002, "learning_rate": 5.815535998935819e-05, "loss": 0.0114, "step": 62510 }, { "epoch": 4.338196468722799e-13, "grad_norm": 0.3361212909221649, "learning_rate": 5.813517906051705e-05, "loss": 0.0115, "step": 62520 }, { "epoch": 4.33889035811319e-13, "grad_norm": 0.36660492420196533, "learning_rate": 5.811500063554034e-05, "loss": 0.0113, "step": 62530 }, { "epoch": 4.3395842475035806e-13, "grad_norm": 0.32042592763900757, "learning_rate": 5.809482471650176e-05, "loss": 0.0119, "step": 62540 }, { "epoch": 4.3402781368939714e-13, "grad_norm": 0.3573472499847412, "learning_rate": 5.807465130547464e-05, "loss": 0.0121, "step": 62550 }, { "epoch": 4.340972026284362e-13, "grad_norm": 0.3098123073577881, "learning_rate": 5.8054480404532144e-05, "loss": 0.0112, "step": 62560 }, { "epoch": 4.341665915674753e-13, "grad_norm": 0.36360839009284973, "learning_rate": 5.803431201574715e-05, "loss": 0.0111, "step": 62570 }, { "epoch": 4.3423598050651435e-13, "grad_norm": 0.39085742831230164, "learning_rate": 5.801414614119226e-05, "loss": 0.0114, "step": 62580 }, { "epoch": 4.343053694455534e-13, "grad_norm": 0.36471128463745117, "learning_rate": 5.799398278293982e-05, "loss": 0.011, "step": 62590 }, { "epoch": 4.343747583845925e-13, "grad_norm": 0.3372747302055359, "learning_rate": 5.797382194306197e-05, "loss": 0.011, "step": 62600 }, { "epoch": 4.3444414732363157e-13, "grad_norm": 0.3684907853603363, "learning_rate": 5.795366362363051e-05, "loss": 0.0112, "step": 62610 }, { "epoch": 4.3451353626267064e-13, "grad_norm": 0.32030847668647766, "learning_rate": 5.7933507826717054e-05, "loss": 0.0125, "step": 62620 }, { "epoch": 4.345829252017097e-13, "grad_norm": 0.34561535716056824, "learning_rate": 5.791335455439288e-05, "loss": 0.0112, "step": 62630 }, { "epoch": 4.346523141407488e-13, "grad_norm": 0.3356643319129944, "learning_rate": 5.789320380872911e-05, "loss": 0.0112, "step": 62640 }, { "epoch": 4.3472170307978786e-13, "grad_norm": 0.3263397216796875, "learning_rate": 5.78730555917965e-05, "loss": 0.0115, "step": 62650 }, { "epoch": 4.3479109201882693e-13, "grad_norm": 0.34678182005882263, "learning_rate": 5.785290990566561e-05, "loss": 0.0115, "step": 62660 }, { "epoch": 4.34860480957866e-13, "grad_norm": 0.3791603446006775, "learning_rate": 5.78327667524067e-05, "loss": 0.0114, "step": 62670 }, { "epoch": 4.3492986989690507e-13, "grad_norm": 0.34360867738723755, "learning_rate": 5.781262613408984e-05, "loss": 0.0113, "step": 62680 }, { "epoch": 4.3499925883594415e-13, "grad_norm": 0.33011704683303833, "learning_rate": 5.779248805278474e-05, "loss": 0.0113, "step": 62690 }, { "epoch": 4.350686477749832e-13, "grad_norm": 0.34545549750328064, "learning_rate": 5.777235251056095e-05, "loss": 0.0115, "step": 62700 }, { "epoch": 4.351380367140223e-13, "grad_norm": 0.36016637086868286, "learning_rate": 5.7752219509487646e-05, "loss": 0.012, "step": 62710 }, { "epoch": 4.3520742565306136e-13, "grad_norm": 0.3330512046813965, "learning_rate": 5.773208905163388e-05, "loss": 0.0113, "step": 62720 }, { "epoch": 4.3527681459210044e-13, "grad_norm": 0.32725077867507935, "learning_rate": 5.771196113906831e-05, "loss": 0.0113, "step": 62730 }, { "epoch": 4.353462035311395e-13, "grad_norm": 0.32723960280418396, "learning_rate": 5.769183577385945e-05, "loss": 0.0114, "step": 62740 }, { "epoch": 4.354155924701786e-13, "grad_norm": 0.3307805359363556, "learning_rate": 5.767171295807544e-05, "loss": 0.011, "step": 62750 }, { "epoch": 4.3548498140921765e-13, "grad_norm": 0.3458457291126251, "learning_rate": 5.765159269378423e-05, "loss": 0.0108, "step": 62760 }, { "epoch": 4.3555437034825673e-13, "grad_norm": 0.3162112236022949, "learning_rate": 5.7631474983053504e-05, "loss": 0.0112, "step": 62770 }, { "epoch": 4.356237592872958e-13, "grad_norm": 0.3339323103427887, "learning_rate": 5.761135982795069e-05, "loss": 0.0116, "step": 62780 }, { "epoch": 4.3569314822633487e-13, "grad_norm": 0.30756235122680664, "learning_rate": 5.759124723054284e-05, "loss": 0.0109, "step": 62790 }, { "epoch": 4.3576253716537394e-13, "grad_norm": 0.39561834931373596, "learning_rate": 5.757113719289695e-05, "loss": 0.0113, "step": 62800 }, { "epoch": 4.35831926104413e-13, "grad_norm": 0.32803893089294434, "learning_rate": 5.755102971707959e-05, "loss": 0.0111, "step": 62810 }, { "epoch": 4.359013150434521e-13, "grad_norm": 0.3590516149997711, "learning_rate": 5.7530924805157076e-05, "loss": 0.0113, "step": 62820 }, { "epoch": 4.3597070398249116e-13, "grad_norm": 0.3725723922252655, "learning_rate": 5.751082245919559e-05, "loss": 0.0116, "step": 62830 }, { "epoch": 4.3604009292153023e-13, "grad_norm": 0.34834936261177063, "learning_rate": 5.7490722681260856e-05, "loss": 0.0108, "step": 62840 }, { "epoch": 4.361094818605693e-13, "grad_norm": 0.38251835107803345, "learning_rate": 5.747062547341854e-05, "loss": 0.0118, "step": 62850 }, { "epoch": 4.361788707996084e-13, "grad_norm": 0.38464468717575073, "learning_rate": 5.7450530837733864e-05, "loss": 0.0118, "step": 62860 }, { "epoch": 4.3624825973864745e-13, "grad_norm": 0.36731767654418945, "learning_rate": 5.743043877627191e-05, "loss": 0.0116, "step": 62870 }, { "epoch": 4.363176486776865e-13, "grad_norm": 0.3538719713687897, "learning_rate": 5.7410349291097444e-05, "loss": 0.0113, "step": 62880 }, { "epoch": 4.363870376167256e-13, "grad_norm": 0.3409591019153595, "learning_rate": 5.739026238427496e-05, "loss": 0.0109, "step": 62890 }, { "epoch": 4.3645642655576466e-13, "grad_norm": 0.36762160062789917, "learning_rate": 5.73701780578687e-05, "loss": 0.0113, "step": 62900 }, { "epoch": 4.3652581549480374e-13, "grad_norm": 0.3298957943916321, "learning_rate": 5.735009631394266e-05, "loss": 0.011, "step": 62910 }, { "epoch": 4.365952044338428e-13, "grad_norm": 0.2862321734428406, "learning_rate": 5.733001715456053e-05, "loss": 0.0111, "step": 62920 }, { "epoch": 4.366645933728819e-13, "grad_norm": 0.3505696654319763, "learning_rate": 5.730994058178578e-05, "loss": 0.0113, "step": 62930 }, { "epoch": 4.3673398231192095e-13, "grad_norm": 0.33141881227493286, "learning_rate": 5.728986659768154e-05, "loss": 0.011, "step": 62940 }, { "epoch": 4.3680337125096003e-13, "grad_norm": 0.3444478213787079, "learning_rate": 5.726979520431079e-05, "loss": 0.011, "step": 62950 }, { "epoch": 4.368727601899991e-13, "grad_norm": 0.34716299176216125, "learning_rate": 5.724972640373614e-05, "loss": 0.0111, "step": 62960 }, { "epoch": 4.3694214912903817e-13, "grad_norm": 0.3356727361679077, "learning_rate": 5.722966019801998e-05, "loss": 0.0112, "step": 62970 }, { "epoch": 4.3701153806807724e-13, "grad_norm": 0.35089707374572754, "learning_rate": 5.720959658922442e-05, "loss": 0.011, "step": 62980 }, { "epoch": 4.370809270071163e-13, "grad_norm": 0.3530139923095703, "learning_rate": 5.718953557941132e-05, "loss": 0.0112, "step": 62990 }, { "epoch": 4.371503159461554e-13, "grad_norm": 0.3646751940250397, "learning_rate": 5.7169477170642254e-05, "loss": 0.011, "step": 63000 }, { "epoch": 4.3721970488519446e-13, "grad_norm": 0.306592732667923, "learning_rate": 5.714942136497853e-05, "loss": 0.0115, "step": 63010 }, { "epoch": 4.3728909382423353e-13, "grad_norm": 0.3442261517047882, "learning_rate": 5.7129368164481196e-05, "loss": 0.012, "step": 63020 }, { "epoch": 4.373584827632726e-13, "grad_norm": 0.37473922967910767, "learning_rate": 5.710931757121104e-05, "loss": 0.0113, "step": 63030 }, { "epoch": 4.374278717023117e-13, "grad_norm": 0.36793842911720276, "learning_rate": 5.708926958722854e-05, "loss": 0.0111, "step": 63040 }, { "epoch": 4.3749726064135075e-13, "grad_norm": 0.32315659523010254, "learning_rate": 5.706922421459399e-05, "loss": 0.011, "step": 63050 }, { "epoch": 4.375666495803898e-13, "grad_norm": 0.34530600905418396, "learning_rate": 5.7049181455367315e-05, "loss": 0.0111, "step": 63060 }, { "epoch": 4.376360385194289e-13, "grad_norm": 0.33669084310531616, "learning_rate": 5.702914131160827e-05, "loss": 0.0107, "step": 63070 }, { "epoch": 4.3770542745846797e-13, "grad_norm": 0.3542836010456085, "learning_rate": 5.700910378537625e-05, "loss": 0.0113, "step": 63080 }, { "epoch": 4.3777481639750704e-13, "grad_norm": 0.31463953852653503, "learning_rate": 5.698906887873042e-05, "loss": 0.0109, "step": 63090 }, { "epoch": 4.378442053365461e-13, "grad_norm": 0.3322543799877167, "learning_rate": 5.6969036593729724e-05, "loss": 0.0108, "step": 63100 }, { "epoch": 4.379135942755852e-13, "grad_norm": 0.3475959599018097, "learning_rate": 5.694900693243276e-05, "loss": 0.0106, "step": 63110 }, { "epoch": 4.3798298321462426e-13, "grad_norm": 0.3628882169723511, "learning_rate": 5.6928979896897864e-05, "loss": 0.0111, "step": 63120 }, { "epoch": 4.3805237215366333e-13, "grad_norm": 0.32247230410575867, "learning_rate": 5.690895548918318e-05, "loss": 0.0113, "step": 63130 }, { "epoch": 4.381217610927024e-13, "grad_norm": 0.33929330110549927, "learning_rate": 5.688893371134647e-05, "loss": 0.0111, "step": 63140 }, { "epoch": 4.3819115003174147e-13, "grad_norm": 0.31210416555404663, "learning_rate": 5.686891456544534e-05, "loss": 0.011, "step": 63150 }, { "epoch": 4.3826053897078054e-13, "grad_norm": 0.33031028509140015, "learning_rate": 5.684889805353702e-05, "loss": 0.0114, "step": 63160 }, { "epoch": 4.383299279098196e-13, "grad_norm": 0.331421822309494, "learning_rate": 5.6828884177678554e-05, "loss": 0.0114, "step": 63170 }, { "epoch": 4.383993168488587e-13, "grad_norm": 0.3068122863769531, "learning_rate": 5.6808872939926665e-05, "loss": 0.0119, "step": 63180 }, { "epoch": 4.3846870578789776e-13, "grad_norm": 0.2962513566017151, "learning_rate": 5.678886434233781e-05, "loss": 0.011, "step": 63190 }, { "epoch": 4.3853809472693683e-13, "grad_norm": 0.3509257435798645, "learning_rate": 5.676885838696818e-05, "loss": 0.0111, "step": 63200 }, { "epoch": 4.386074836659759e-13, "grad_norm": 0.2870328426361084, "learning_rate": 5.674885507587375e-05, "loss": 0.0109, "step": 63210 }, { "epoch": 4.38676872605015e-13, "grad_norm": 0.29917195439338684, "learning_rate": 5.6728854411110076e-05, "loss": 0.0108, "step": 63220 }, { "epoch": 4.3874626154405405e-13, "grad_norm": 0.3436031639575958, "learning_rate": 5.670885639473263e-05, "loss": 0.0108, "step": 63230 }, { "epoch": 4.388156504830931e-13, "grad_norm": 0.3350675106048584, "learning_rate": 5.668886102879646e-05, "loss": 0.0111, "step": 63240 }, { "epoch": 4.388850394221322e-13, "grad_norm": 0.3166504502296448, "learning_rate": 5.666886831535644e-05, "loss": 0.0109, "step": 63250 }, { "epoch": 4.3895442836117127e-13, "grad_norm": 0.3434446454048157, "learning_rate": 5.664887825646712e-05, "loss": 0.0107, "step": 63260 }, { "epoch": 4.3902381730021034e-13, "grad_norm": 0.3611193597316742, "learning_rate": 5.662889085418276e-05, "loss": 0.011, "step": 63270 }, { "epoch": 4.390932062392494e-13, "grad_norm": 0.3359088599681854, "learning_rate": 5.6608906110557426e-05, "loss": 0.011, "step": 63280 }, { "epoch": 4.391625951782885e-13, "grad_norm": 0.338031530380249, "learning_rate": 5.658892402764481e-05, "loss": 0.0109, "step": 63290 }, { "epoch": 4.3923198411732756e-13, "grad_norm": 0.35009047389030457, "learning_rate": 5.6568944607498454e-05, "loss": 0.0115, "step": 63300 }, { "epoch": 4.3930137305636663e-13, "grad_norm": 0.33693593740463257, "learning_rate": 5.6548967852171476e-05, "loss": 0.0109, "step": 63310 }, { "epoch": 4.393707619954057e-13, "grad_norm": 0.3630075454711914, "learning_rate": 5.6528993763716816e-05, "loss": 0.0114, "step": 63320 }, { "epoch": 4.3944015093444477e-13, "grad_norm": 0.35778293013572693, "learning_rate": 5.650902234418716e-05, "loss": 0.0114, "step": 63330 }, { "epoch": 4.3950953987348385e-13, "grad_norm": 0.35436299443244934, "learning_rate": 5.6489053595634875e-05, "loss": 0.0111, "step": 63340 }, { "epoch": 4.395789288125229e-13, "grad_norm": 0.3213707208633423, "learning_rate": 5.6469087520112005e-05, "loss": 0.0106, "step": 63350 }, { "epoch": 4.39648317751562e-13, "grad_norm": 0.3398258686065674, "learning_rate": 5.6449124119670455e-05, "loss": 0.0107, "step": 63360 }, { "epoch": 4.3971770669060106e-13, "grad_norm": 0.32614830136299133, "learning_rate": 5.642916339636171e-05, "loss": 0.0111, "step": 63370 }, { "epoch": 4.3978709562964013e-13, "grad_norm": 0.3286227881908417, "learning_rate": 5.640920535223709e-05, "loss": 0.0104, "step": 63380 }, { "epoch": 4.398564845686792e-13, "grad_norm": 0.30728182196617126, "learning_rate": 5.6389249989347547e-05, "loss": 0.0106, "step": 63390 }, { "epoch": 4.399258735077183e-13, "grad_norm": 0.3673912584781647, "learning_rate": 5.636929730974386e-05, "loss": 0.0113, "step": 63400 }, { "epoch": 4.3999526244675735e-13, "grad_norm": 0.334514319896698, "learning_rate": 5.634934731547644e-05, "loss": 0.0105, "step": 63410 }, { "epoch": 4.400646513857964e-13, "grad_norm": 0.2869037687778473, "learning_rate": 5.632940000859548e-05, "loss": 0.0106, "step": 63420 }, { "epoch": 4.401340403248355e-13, "grad_norm": 0.3542172610759735, "learning_rate": 5.630945539115084e-05, "loss": 0.0109, "step": 63430 }, { "epoch": 4.4020342926387457e-13, "grad_norm": 0.3454854190349579, "learning_rate": 5.62895134651922e-05, "loss": 0.011, "step": 63440 }, { "epoch": 4.4027281820291364e-13, "grad_norm": 0.30618003010749817, "learning_rate": 5.626957423276884e-05, "loss": 0.0107, "step": 63450 }, { "epoch": 4.403422071419527e-13, "grad_norm": 0.320640504360199, "learning_rate": 5.6249637695929864e-05, "loss": 0.0112, "step": 63460 }, { "epoch": 4.404115960809918e-13, "grad_norm": 0.3539236783981323, "learning_rate": 5.6229703856724044e-05, "loss": 0.0107, "step": 63470 }, { "epoch": 4.4048098502003086e-13, "grad_norm": 0.36652740836143494, "learning_rate": 5.620977271719992e-05, "loss": 0.0113, "step": 63480 }, { "epoch": 4.4055037395906993e-13, "grad_norm": 0.33539149165153503, "learning_rate": 5.6189844279405685e-05, "loss": 0.0116, "step": 63490 }, { "epoch": 4.40619762898109e-13, "grad_norm": 0.3351951539516449, "learning_rate": 5.616991854538933e-05, "loss": 0.0109, "step": 63500 }, { "epoch": 4.4068915183714807e-13, "grad_norm": 0.31638771295547485, "learning_rate": 5.614999551719852e-05, "loss": 0.0107, "step": 63510 }, { "epoch": 4.4075854077618715e-13, "grad_norm": 0.3533623516559601, "learning_rate": 5.6130075196880664e-05, "loss": 0.0108, "step": 63520 }, { "epoch": 4.408279297152262e-13, "grad_norm": 0.3293606638908386, "learning_rate": 5.6110157586482874e-05, "loss": 0.0106, "step": 63530 }, { "epoch": 4.408973186542653e-13, "grad_norm": 0.3548642098903656, "learning_rate": 5.6090242688052014e-05, "loss": 0.0101, "step": 63540 }, { "epoch": 4.4096670759330436e-13, "grad_norm": 0.3136599659919739, "learning_rate": 5.607033050363462e-05, "loss": 0.0106, "step": 63550 }, { "epoch": 4.4103609653234344e-13, "grad_norm": 0.35528260469436646, "learning_rate": 5.605042103527701e-05, "loss": 0.0107, "step": 63560 }, { "epoch": 4.411054854713825e-13, "grad_norm": 0.3309308588504791, "learning_rate": 5.6030514285025154e-05, "loss": 0.0102, "step": 63570 }, { "epoch": 4.411748744104216e-13, "grad_norm": 0.31780773401260376, "learning_rate": 5.6010610254924814e-05, "loss": 0.0106, "step": 63580 }, { "epoch": 4.4124426334946065e-13, "grad_norm": 0.3260852098464966, "learning_rate": 5.599070894702141e-05, "loss": 0.011, "step": 63590 }, { "epoch": 4.413136522884997e-13, "grad_norm": 0.3194820284843445, "learning_rate": 5.597081036336016e-05, "loss": 0.0109, "step": 63600 }, { "epoch": 4.413830412275388e-13, "grad_norm": 0.33370599150657654, "learning_rate": 5.595091450598588e-05, "loss": 0.0111, "step": 63610 }, { "epoch": 4.4145243016657787e-13, "grad_norm": 0.31117933988571167, "learning_rate": 5.593102137694327e-05, "loss": 0.0105, "step": 63620 }, { "epoch": 4.4152181910561694e-13, "grad_norm": 0.3525335192680359, "learning_rate": 5.591113097827658e-05, "loss": 0.011, "step": 63630 }, { "epoch": 4.41591208044656e-13, "grad_norm": 0.36916282773017883, "learning_rate": 5.58912433120299e-05, "loss": 0.0113, "step": 63640 }, { "epoch": 4.416605969836951e-13, "grad_norm": 0.3382030129432678, "learning_rate": 5.5871358380246954e-05, "loss": 0.0114, "step": 63650 }, { "epoch": 4.4172998592273416e-13, "grad_norm": 0.3745100498199463, "learning_rate": 5.585147618497128e-05, "loss": 0.0109, "step": 63660 }, { "epoch": 4.4179937486177323e-13, "grad_norm": 0.31278231739997864, "learning_rate": 5.583159672824606e-05, "loss": 0.0103, "step": 63670 }, { "epoch": 4.418687638008123e-13, "grad_norm": 0.28774455189704895, "learning_rate": 5.581172001211423e-05, "loss": 0.0106, "step": 63680 }, { "epoch": 4.419381527398514e-13, "grad_norm": 0.32712695002555847, "learning_rate": 5.579184603861838e-05, "loss": 0.0106, "step": 63690 }, { "epoch": 4.4200754167889045e-13, "grad_norm": 0.34020179510116577, "learning_rate": 5.5771974809800966e-05, "loss": 0.0105, "step": 63700 }, { "epoch": 4.420769306179295e-13, "grad_norm": 0.32447969913482666, "learning_rate": 5.575210632770399e-05, "loss": 0.0107, "step": 63710 }, { "epoch": 4.421463195569686e-13, "grad_norm": 0.3084345757961273, "learning_rate": 5.5732240594369244e-05, "loss": 0.0106, "step": 63720 }, { "epoch": 4.4221570849600766e-13, "grad_norm": 0.3533952832221985, "learning_rate": 5.571237761183828e-05, "loss": 0.0104, "step": 63730 }, { "epoch": 4.4228509743504674e-13, "grad_norm": 0.31084364652633667, "learning_rate": 5.5692517382152306e-05, "loss": 0.0107, "step": 63740 }, { "epoch": 4.423544863740858e-13, "grad_norm": 0.34632349014282227, "learning_rate": 5.5672659907352285e-05, "loss": 0.0107, "step": 63750 }, { "epoch": 4.424238753131249e-13, "grad_norm": 0.3287542164325714, "learning_rate": 5.565280518947888e-05, "loss": 0.0107, "step": 63760 }, { "epoch": 4.4249326425216395e-13, "grad_norm": 0.3099934458732605, "learning_rate": 5.563295323057247e-05, "loss": 0.0111, "step": 63770 }, { "epoch": 4.4256265319120303e-13, "grad_norm": 0.3167427182197571, "learning_rate": 5.5613104032673126e-05, "loss": 0.0103, "step": 63780 }, { "epoch": 4.426320421302421e-13, "grad_norm": 0.29567429423332214, "learning_rate": 5.559325759782069e-05, "loss": 0.011, "step": 63790 }, { "epoch": 4.4270143106928117e-13, "grad_norm": 0.3466021716594696, "learning_rate": 5.5573413928054686e-05, "loss": 0.0112, "step": 63800 }, { "epoch": 4.4277082000832024e-13, "grad_norm": 0.2827315330505371, "learning_rate": 5.555357302541438e-05, "loss": 0.0106, "step": 63810 }, { "epoch": 4.428402089473593e-13, "grad_norm": 0.3271568715572357, "learning_rate": 5.553373489193868e-05, "loss": 0.0106, "step": 63820 }, { "epoch": 4.429095978863984e-13, "grad_norm": 0.35760048031806946, "learning_rate": 5.551389952966633e-05, "loss": 0.0104, "step": 63830 }, { "epoch": 4.4297898682543746e-13, "grad_norm": 0.30204305052757263, "learning_rate": 5.549406694063567e-05, "loss": 0.0106, "step": 63840 }, { "epoch": 4.4304837576447653e-13, "grad_norm": 0.3015279173851013, "learning_rate": 5.547423712688484e-05, "loss": 0.0102, "step": 63850 }, { "epoch": 4.431177647035156e-13, "grad_norm": 0.32360854744911194, "learning_rate": 5.545441009045162e-05, "loss": 0.0104, "step": 63860 }, { "epoch": 4.431871536425547e-13, "grad_norm": 0.2918195128440857, "learning_rate": 5.5434585833373604e-05, "loss": 0.0107, "step": 63870 }, { "epoch": 4.4325654258159375e-13, "grad_norm": 0.326748788356781, "learning_rate": 5.541476435768799e-05, "loss": 0.0104, "step": 63880 }, { "epoch": 4.433259315206328e-13, "grad_norm": 0.30323848128318787, "learning_rate": 5.53949456654318e-05, "loss": 0.0103, "step": 63890 }, { "epoch": 4.433953204596719e-13, "grad_norm": 0.30037274956703186, "learning_rate": 5.537512975864164e-05, "loss": 0.0109, "step": 63900 }, { "epoch": 4.4346470939871097e-13, "grad_norm": 0.330942302942276, "learning_rate": 5.5355316639353945e-05, "loss": 0.0109, "step": 63910 }, { "epoch": 4.4353409833775004e-13, "grad_norm": 0.34833934903144836, "learning_rate": 5.533550630960482e-05, "loss": 0.0106, "step": 63920 }, { "epoch": 4.436034872767891e-13, "grad_norm": 0.3073365390300751, "learning_rate": 5.531569877143012e-05, "loss": 0.0107, "step": 63930 }, { "epoch": 4.436728762158282e-13, "grad_norm": 0.3087747097015381, "learning_rate": 5.529589402686529e-05, "loss": 0.0106, "step": 63940 }, { "epoch": 4.4374226515486725e-13, "grad_norm": 0.3290553092956543, "learning_rate": 5.527609207794566e-05, "loss": 0.0111, "step": 63950 }, { "epoch": 4.4381165409390633e-13, "grad_norm": 0.306947261095047, "learning_rate": 5.525629292670613e-05, "loss": 0.0112, "step": 63960 }, { "epoch": 4.438810430329454e-13, "grad_norm": 0.3184467852115631, "learning_rate": 5.523649657518144e-05, "loss": 0.0108, "step": 63970 }, { "epoch": 4.4395043197198447e-13, "grad_norm": 0.2941265106201172, "learning_rate": 5.521670302540587e-05, "loss": 0.0109, "step": 63980 }, { "epoch": 4.4401982091102354e-13, "grad_norm": 0.3367195427417755, "learning_rate": 5.519691227941361e-05, "loss": 0.0105, "step": 63990 }, { "epoch": 4.440892098500626e-13, "grad_norm": 0.30720391869544983, "learning_rate": 5.517712433923841e-05, "loss": 0.0104, "step": 64000 }, { "epoch": 4.441585987891017e-13, "grad_norm": 0.32857823371887207, "learning_rate": 5.515733920691383e-05, "loss": 0.0101, "step": 64010 }, { "epoch": 4.4422798772814076e-13, "grad_norm": 0.2856352925300598, "learning_rate": 5.513755688447307e-05, "loss": 0.0105, "step": 64020 }, { "epoch": 4.4429737666717983e-13, "grad_norm": 0.29716405272483826, "learning_rate": 5.511777737394908e-05, "loss": 0.0105, "step": 64030 }, { "epoch": 4.443667656062189e-13, "grad_norm": 0.31049755215644836, "learning_rate": 5.50980006773745e-05, "loss": 0.0103, "step": 64040 }, { "epoch": 4.44436154545258e-13, "grad_norm": 0.34219762682914734, "learning_rate": 5.507822679678173e-05, "loss": 0.0107, "step": 64050 }, { "epoch": 4.4450554348429705e-13, "grad_norm": 0.3240765333175659, "learning_rate": 5.5058455734202824e-05, "loss": 0.0104, "step": 64060 }, { "epoch": 4.445749324233361e-13, "grad_norm": 0.3013385236263275, "learning_rate": 5.503868749166956e-05, "loss": 0.0109, "step": 64070 }, { "epoch": 4.446443213623752e-13, "grad_norm": 0.3252449631690979, "learning_rate": 5.5018922071213404e-05, "loss": 0.0111, "step": 64080 }, { "epoch": 4.4471371030141427e-13, "grad_norm": 0.33742600679397583, "learning_rate": 5.4999159474865625e-05, "loss": 0.0104, "step": 64090 }, { "epoch": 4.4478309924045334e-13, "grad_norm": 0.3597673773765564, "learning_rate": 5.497939970465707e-05, "loss": 0.0107, "step": 64100 }, { "epoch": 4.448524881794924e-13, "grad_norm": 0.3990754783153534, "learning_rate": 5.4959642762618446e-05, "loss": 0.0113, "step": 64110 }, { "epoch": 4.449218771185315e-13, "grad_norm": 0.34482836723327637, "learning_rate": 5.4939888650779994e-05, "loss": 0.0108, "step": 64120 }, { "epoch": 4.4499126605757056e-13, "grad_norm": 0.2994837164878845, "learning_rate": 5.492013737117183e-05, "loss": 0.0104, "step": 64130 }, { "epoch": 4.4506065499660963e-13, "grad_norm": 0.3205737769603729, "learning_rate": 5.490038892582364e-05, "loss": 0.0105, "step": 64140 }, { "epoch": 4.451300439356487e-13, "grad_norm": 0.3434416651725769, "learning_rate": 5.4880643316764954e-05, "loss": 0.0105, "step": 64150 }, { "epoch": 4.4519943287468777e-13, "grad_norm": 0.31197088956832886, "learning_rate": 5.48609005460249e-05, "loss": 0.0101, "step": 64160 }, { "epoch": 4.4526882181372684e-13, "grad_norm": 0.29012182354927063, "learning_rate": 5.484116061563233e-05, "loss": 0.0103, "step": 64170 }, { "epoch": 4.453382107527659e-13, "grad_norm": 0.2821494936943054, "learning_rate": 5.48214235276159e-05, "loss": 0.0103, "step": 64180 }, { "epoch": 4.45407599691805e-13, "grad_norm": 0.3044007122516632, "learning_rate": 5.4801689284003836e-05, "loss": 0.0104, "step": 64190 }, { "epoch": 4.4547698863084406e-13, "grad_norm": 0.330048143863678, "learning_rate": 5.478195788682419e-05, "loss": 0.0106, "step": 64200 }, { "epoch": 4.4554637756988313e-13, "grad_norm": 0.32819220423698425, "learning_rate": 5.4762229338104614e-05, "loss": 0.0105, "step": 64210 }, { "epoch": 4.456157665089222e-13, "grad_norm": 0.33636024594306946, "learning_rate": 5.474250363987258e-05, "loss": 0.0104, "step": 64220 }, { "epoch": 4.456851554479613e-13, "grad_norm": 0.3554540276527405, "learning_rate": 5.4722780794155156e-05, "loss": 0.011, "step": 64230 }, { "epoch": 4.4575454438700035e-13, "grad_norm": 0.3256145715713501, "learning_rate": 5.470306080297924e-05, "loss": 0.0106, "step": 64240 }, { "epoch": 4.458239333260394e-13, "grad_norm": 0.34765729308128357, "learning_rate": 5.4683343668371266e-05, "loss": 0.0104, "step": 64250 }, { "epoch": 4.458933222650785e-13, "grad_norm": 0.31306976079940796, "learning_rate": 5.46636293923576e-05, "loss": 0.0106, "step": 64260 }, { "epoch": 4.4596271120411757e-13, "grad_norm": 0.3053697943687439, "learning_rate": 5.464391797696409e-05, "loss": 0.0108, "step": 64270 }, { "epoch": 4.4603210014315664e-13, "grad_norm": 0.36450469493865967, "learning_rate": 5.462420942421645e-05, "loss": 0.0105, "step": 64280 }, { "epoch": 4.461014890821957e-13, "grad_norm": 0.3279739320278168, "learning_rate": 5.460450373614002e-05, "loss": 0.0104, "step": 64290 }, { "epoch": 4.461708780212348e-13, "grad_norm": 0.316686749458313, "learning_rate": 5.458480091475987e-05, "loss": 0.0107, "step": 64300 }, { "epoch": 4.4624026696027386e-13, "grad_norm": 0.3156580328941345, "learning_rate": 5.4565100962100744e-05, "loss": 0.0101, "step": 64310 }, { "epoch": 4.4630965589931293e-13, "grad_norm": 0.31009623408317566, "learning_rate": 5.454540388018716e-05, "loss": 0.0099, "step": 64320 }, { "epoch": 4.46379044838352e-13, "grad_norm": 0.3593516945838928, "learning_rate": 5.452570967104327e-05, "loss": 0.0105, "step": 64330 }, { "epoch": 4.4644843377739107e-13, "grad_norm": 0.2939136326313019, "learning_rate": 5.450601833669299e-05, "loss": 0.0106, "step": 64340 }, { "epoch": 4.4651782271643015e-13, "grad_norm": 0.3237610459327698, "learning_rate": 5.448632987915987e-05, "loss": 0.0104, "step": 64350 }, { "epoch": 4.465872116554692e-13, "grad_norm": 0.30458444356918335, "learning_rate": 5.446664430046726e-05, "loss": 0.0101, "step": 64360 }, { "epoch": 4.466566005945083e-13, "grad_norm": 0.30634549260139465, "learning_rate": 5.4446961602638086e-05, "loss": 0.0106, "step": 64370 }, { "epoch": 4.4672598953354736e-13, "grad_norm": 0.3676679730415344, "learning_rate": 5.442728178769512e-05, "loss": 0.0106, "step": 64380 }, { "epoch": 4.4679537847258644e-13, "grad_norm": 0.31291139125823975, "learning_rate": 5.440760485766072e-05, "loss": 0.0105, "step": 64390 }, { "epoch": 4.468647674116255e-13, "grad_norm": 0.29216447472572327, "learning_rate": 5.438793081455708e-05, "loss": 0.0106, "step": 64400 }, { "epoch": 4.469341563506646e-13, "grad_norm": 0.3106144368648529, "learning_rate": 5.436825966040587e-05, "loss": 0.0102, "step": 64410 }, { "epoch": 4.4700354528970365e-13, "grad_norm": 0.31568920612335205, "learning_rate": 5.4348591397228764e-05, "loss": 0.0105, "step": 64420 }, { "epoch": 4.470729342287427e-13, "grad_norm": 0.3057469129562378, "learning_rate": 5.432892602704687e-05, "loss": 0.0109, "step": 64430 }, { "epoch": 4.471423231677818e-13, "grad_norm": 0.2548050880432129, "learning_rate": 5.4309263551881156e-05, "loss": 0.0103, "step": 64440 }, { "epoch": 4.4721171210682087e-13, "grad_norm": 0.36109399795532227, "learning_rate": 5.428960397375223e-05, "loss": 0.0106, "step": 64450 }, { "epoch": 4.4728110104585994e-13, "grad_norm": 0.28653720021247864, "learning_rate": 5.4269947294680454e-05, "loss": 0.0101, "step": 64460 }, { "epoch": 4.47350489984899e-13, "grad_norm": 0.27962109446525574, "learning_rate": 5.425029351668581e-05, "loss": 0.01, "step": 64470 }, { "epoch": 4.474198789239381e-13, "grad_norm": 0.3303075134754181, "learning_rate": 5.423064264178807e-05, "loss": 0.0101, "step": 64480 }, { "epoch": 4.4748926786297716e-13, "grad_norm": 0.33052584528923035, "learning_rate": 5.421099467200663e-05, "loss": 0.0107, "step": 64490 }, { "epoch": 4.4755865680201623e-13, "grad_norm": 0.3051033616065979, "learning_rate": 5.419134960936066e-05, "loss": 0.0103, "step": 64500 }, { "epoch": 4.476280457410553e-13, "grad_norm": 0.34933701157569885, "learning_rate": 5.4171707455868986e-05, "loss": 0.0101, "step": 64510 }, { "epoch": 4.476974346800944e-13, "grad_norm": 0.31120234727859497, "learning_rate": 5.415206821355013e-05, "loss": 0.0103, "step": 64520 }, { "epoch": 4.4776682361913345e-13, "grad_norm": 0.33162084221839905, "learning_rate": 5.413243188442233e-05, "loss": 0.0101, "step": 64530 }, { "epoch": 4.478362125581725e-13, "grad_norm": 0.31282705068588257, "learning_rate": 5.411279847050354e-05, "loss": 0.0104, "step": 64540 }, { "epoch": 4.479056014972116e-13, "grad_norm": 0.3072776198387146, "learning_rate": 5.409316797381141e-05, "loss": 0.0108, "step": 64550 }, { "epoch": 4.4797499043625066e-13, "grad_norm": 0.2983379065990448, "learning_rate": 5.407354039636328e-05, "loss": 0.0103, "step": 64560 }, { "epoch": 4.4804437937528974e-13, "grad_norm": 0.3338513970375061, "learning_rate": 5.405391574017611e-05, "loss": 0.0106, "step": 64570 }, { "epoch": 4.481137683143288e-13, "grad_norm": 0.34582576155662537, "learning_rate": 5.4034294007266745e-05, "loss": 0.0102, "step": 64580 }, { "epoch": 4.481831572533679e-13, "grad_norm": 0.34537234902381897, "learning_rate": 5.401467519965154e-05, "loss": 0.011, "step": 64590 }, { "epoch": 4.4825254619240695e-13, "grad_norm": 0.29927653074264526, "learning_rate": 5.399505931934673e-05, "loss": 0.0103, "step": 64600 }, { "epoch": 4.4832193513144603e-13, "grad_norm": 0.358028382062912, "learning_rate": 5.397544636836801e-05, "loss": 0.0105, "step": 64610 }, { "epoch": 4.483913240704851e-13, "grad_norm": 0.3030480146408081, "learning_rate": 5.395583634873108e-05, "loss": 0.0102, "step": 64620 }, { "epoch": 4.4846071300952417e-13, "grad_norm": 0.32389846444129944, "learning_rate": 5.3936229262451044e-05, "loss": 0.0098, "step": 64630 }, { "epoch": 4.4853010194856324e-13, "grad_norm": 0.2872280478477478, "learning_rate": 5.39166251115429e-05, "loss": 0.01, "step": 64640 }, { "epoch": 4.485994908876023e-13, "grad_norm": 0.3478406071662903, "learning_rate": 5.3897023898021267e-05, "loss": 0.0106, "step": 64650 }, { "epoch": 4.486688798266414e-13, "grad_norm": 0.319305419921875, "learning_rate": 5.387742562390046e-05, "loss": 0.0097, "step": 64660 }, { "epoch": 4.4873826876568046e-13, "grad_norm": 0.2792222201824188, "learning_rate": 5.385783029119452e-05, "loss": 0.0102, "step": 64670 }, { "epoch": 4.4880765770471953e-13, "grad_norm": 0.28643089532852173, "learning_rate": 5.383823790191716e-05, "loss": 0.01, "step": 64680 }, { "epoch": 4.488770466437586e-13, "grad_norm": 0.28850221633911133, "learning_rate": 5.381864845808187e-05, "loss": 0.0099, "step": 64690 }, { "epoch": 4.489464355827977e-13, "grad_norm": 0.2942553162574768, "learning_rate": 5.379906196170163e-05, "loss": 0.01, "step": 64700 }, { "epoch": 4.4901582452183675e-13, "grad_norm": 0.328262060880661, "learning_rate": 5.3779478414789414e-05, "loss": 0.011, "step": 64710 }, { "epoch": 4.490852134608758e-13, "grad_norm": 0.35414060950279236, "learning_rate": 5.375989781935759e-05, "loss": 0.0101, "step": 64720 }, { "epoch": 4.491546023999149e-13, "grad_norm": 0.3354805111885071, "learning_rate": 5.37403201774185e-05, "loss": 0.0105, "step": 64730 }, { "epoch": 4.4922399133895397e-13, "grad_norm": 0.3188210129737854, "learning_rate": 5.3720745490983934e-05, "loss": 0.0104, "step": 64740 }, { "epoch": 4.4929338027799304e-13, "grad_norm": 0.3344137668609619, "learning_rate": 5.370117376206562e-05, "loss": 0.011, "step": 64750 }, { "epoch": 4.493627692170321e-13, "grad_norm": 0.3154471218585968, "learning_rate": 5.368160499267474e-05, "loss": 0.0101, "step": 64760 }, { "epoch": 4.494321581560712e-13, "grad_norm": 0.31455862522125244, "learning_rate": 5.3662039184822395e-05, "loss": 0.0102, "step": 64770 }, { "epoch": 4.4950154709511025e-13, "grad_norm": 0.3338375389575958, "learning_rate": 5.364247634051919e-05, "loss": 0.0099, "step": 64780 }, { "epoch": 4.4957093603414933e-13, "grad_norm": 0.3235466182231903, "learning_rate": 5.362291646177553e-05, "loss": 0.0097, "step": 64790 }, { "epoch": 4.496403249731884e-13, "grad_norm": 0.30903977155685425, "learning_rate": 5.360335955060155e-05, "loss": 0.0102, "step": 64800 }, { "epoch": 4.4970971391222747e-13, "grad_norm": 0.325932115316391, "learning_rate": 5.3583805609006945e-05, "loss": 0.0102, "step": 64810 }, { "epoch": 4.4977910285126654e-13, "grad_norm": 0.32151222229003906, "learning_rate": 5.3564254639001255e-05, "loss": 0.0097, "step": 64820 }, { "epoch": 4.498484917903056e-13, "grad_norm": 0.27749279141426086, "learning_rate": 5.354470664259363e-05, "loss": 0.01, "step": 64830 }, { "epoch": 4.499178807293447e-13, "grad_norm": 0.32359302043914795, "learning_rate": 5.3525161621792894e-05, "loss": 0.0099, "step": 64840 }, { "epoch": 4.4998726966838376e-13, "grad_norm": 0.2956250011920929, "learning_rate": 5.350561957860765e-05, "loss": 0.0099, "step": 64850 }, { "epoch": 4.5005665860742283e-13, "grad_norm": 0.3187384009361267, "learning_rate": 5.34860805150461e-05, "loss": 0.0101, "step": 64860 }, { "epoch": 4.501260475464619e-13, "grad_norm": 0.3615292012691498, "learning_rate": 5.346654443311625e-05, "loss": 0.0106, "step": 64870 }, { "epoch": 4.50195436485501e-13, "grad_norm": 0.35114678740501404, "learning_rate": 5.344701133482568e-05, "loss": 0.0103, "step": 64880 }, { "epoch": 4.5026482542454005e-13, "grad_norm": 0.35997921228408813, "learning_rate": 5.3427481222181774e-05, "loss": 0.0103, "step": 64890 }, { "epoch": 4.503342143635791e-13, "grad_norm": 0.34932568669319153, "learning_rate": 5.3407954097191456e-05, "loss": 0.0119, "step": 64900 }, { "epoch": 4.504036033026182e-13, "grad_norm": 0.3317350149154663, "learning_rate": 5.3388429961861577e-05, "loss": 0.0101, "step": 64910 }, { "epoch": 4.5047299224165727e-13, "grad_norm": 0.28347867727279663, "learning_rate": 5.336890881819841e-05, "loss": 0.0099, "step": 64920 }, { "epoch": 4.5054238118069634e-13, "grad_norm": 0.2705812454223633, "learning_rate": 5.334939066820817e-05, "loss": 0.0104, "step": 64930 }, { "epoch": 4.506117701197354e-13, "grad_norm": 0.30118271708488464, "learning_rate": 5.3329875513896574e-05, "loss": 0.0098, "step": 64940 }, { "epoch": 4.506811590587745e-13, "grad_norm": 0.3083434998989105, "learning_rate": 5.3310363357269176e-05, "loss": 0.01, "step": 64950 }, { "epoch": 4.5075054799781356e-13, "grad_norm": 0.31891682744026184, "learning_rate": 5.329085420033106e-05, "loss": 0.0102, "step": 64960 }, { "epoch": 4.5081993693685263e-13, "grad_norm": 0.2993044853210449, "learning_rate": 5.3271348045087195e-05, "loss": 0.01, "step": 64970 }, { "epoch": 4.508893258758917e-13, "grad_norm": 0.3118281364440918, "learning_rate": 5.325184489354206e-05, "loss": 0.01, "step": 64980 }, { "epoch": 4.5095871481493077e-13, "grad_norm": 0.32676443457603455, "learning_rate": 5.323234474769998e-05, "loss": 0.01, "step": 64990 }, { "epoch": 4.5102810375396984e-13, "grad_norm": 0.36902332305908203, "learning_rate": 5.321284760956483e-05, "loss": 0.0098, "step": 65000 }, { "epoch": 4.510974926930089e-13, "grad_norm": 0.34204137325286865, "learning_rate": 5.319335348114028e-05, "loss": 0.0099, "step": 65010 }, { "epoch": 4.51166881632048e-13, "grad_norm": 0.30899402499198914, "learning_rate": 5.3173862364429656e-05, "loss": 0.0101, "step": 65020 }, { "epoch": 4.5123627057108706e-13, "grad_norm": 0.3038466274738312, "learning_rate": 5.315437426143597e-05, "loss": 0.0106, "step": 65030 }, { "epoch": 4.5130565951012613e-13, "grad_norm": 0.30346769094467163, "learning_rate": 5.313488917416193e-05, "loss": 0.0105, "step": 65040 }, { "epoch": 4.513750484491652e-13, "grad_norm": 0.31900107860565186, "learning_rate": 5.311540710460993e-05, "loss": 0.0102, "step": 65050 }, { "epoch": 4.514444373882043e-13, "grad_norm": 0.26892396807670593, "learning_rate": 5.309592805478205e-05, "loss": 0.0107, "step": 65060 }, { "epoch": 4.5151382632724335e-13, "grad_norm": 0.3298605680465698, "learning_rate": 5.307645202668007e-05, "loss": 0.0099, "step": 65070 }, { "epoch": 4.515832152662824e-13, "grad_norm": 0.3572522699832916, "learning_rate": 5.305697902230548e-05, "loss": 0.0101, "step": 65080 }, { "epoch": 4.516526042053215e-13, "grad_norm": 0.30124810338020325, "learning_rate": 5.303750904365936e-05, "loss": 0.0099, "step": 65090 }, { "epoch": 4.5172199314436057e-13, "grad_norm": 0.34662356972694397, "learning_rate": 5.301804209274263e-05, "loss": 0.0097, "step": 65100 }, { "epoch": 4.5179138208339964e-13, "grad_norm": 0.3161762058734894, "learning_rate": 5.299857817155581e-05, "loss": 0.0095, "step": 65110 }, { "epoch": 4.518607710224387e-13, "grad_norm": 0.2848176062107086, "learning_rate": 5.297911728209907e-05, "loss": 0.0104, "step": 65120 }, { "epoch": 4.519301599614778e-13, "grad_norm": 0.2771387994289398, "learning_rate": 5.295965942637236e-05, "loss": 0.0098, "step": 65130 }, { "epoch": 4.5199954890051686e-13, "grad_norm": 0.2818545997142792, "learning_rate": 5.2940204606375284e-05, "loss": 0.0099, "step": 65140 }, { "epoch": 4.5206893783955593e-13, "grad_norm": 0.2770088315010071, "learning_rate": 5.292075282410712e-05, "loss": 0.01, "step": 65150 }, { "epoch": 4.52138326778595e-13, "grad_norm": 0.3122367858886719, "learning_rate": 5.290130408156681e-05, "loss": 0.01, "step": 65160 }, { "epoch": 4.5220771571763407e-13, "grad_norm": 0.36003637313842773, "learning_rate": 5.2881858380753064e-05, "loss": 0.0105, "step": 65170 }, { "epoch": 4.5227710465667315e-13, "grad_norm": 0.2862752377986908, "learning_rate": 5.286241572366419e-05, "loss": 0.0107, "step": 65180 }, { "epoch": 4.523464935957122e-13, "grad_norm": 0.31494763493537903, "learning_rate": 5.2842976112298256e-05, "loss": 0.01, "step": 65190 }, { "epoch": 4.524158825347513e-13, "grad_norm": 0.3036027252674103, "learning_rate": 5.282353954865297e-05, "loss": 0.01, "step": 65200 }, { "epoch": 4.5248527147379036e-13, "grad_norm": 0.34337809681892395, "learning_rate": 5.2804106034725716e-05, "loss": 0.0105, "step": 65210 }, { "epoch": 4.5255466041282943e-13, "grad_norm": 0.28888508677482605, "learning_rate": 5.278467557251367e-05, "loss": 0.0101, "step": 65220 }, { "epoch": 4.526240493518685e-13, "grad_norm": 0.2883574664592743, "learning_rate": 5.276524816401348e-05, "loss": 0.0099, "step": 65230 }, { "epoch": 4.526934382909076e-13, "grad_norm": 0.31699907779693604, "learning_rate": 5.274582381122177e-05, "loss": 0.0101, "step": 65240 }, { "epoch": 4.5276282722994665e-13, "grad_norm": 0.29111406207084656, "learning_rate": 5.272640251613456e-05, "loss": 0.0098, "step": 65250 }, { "epoch": 4.528322161689857e-13, "grad_norm": 0.3022870719432831, "learning_rate": 5.27069842807478e-05, "loss": 0.0095, "step": 65260 }, { "epoch": 4.529016051080248e-13, "grad_norm": 0.27688735723495483, "learning_rate": 5.2687569107056916e-05, "loss": 0.0098, "step": 65270 }, { "epoch": 4.5297099404706387e-13, "grad_norm": 0.28300610184669495, "learning_rate": 5.266815699705721e-05, "loss": 0.01, "step": 65280 }, { "epoch": 4.5304038298610294e-13, "grad_norm": 0.2606264054775238, "learning_rate": 5.2648747952743484e-05, "loss": 0.0097, "step": 65290 }, { "epoch": 4.53109771925142e-13, "grad_norm": 0.3233979046344757, "learning_rate": 5.2629341976110414e-05, "loss": 0.0103, "step": 65300 }, { "epoch": 4.531791608641811e-13, "grad_norm": 0.2935120463371277, "learning_rate": 5.260993906915221e-05, "loss": 0.0096, "step": 65310 }, { "epoch": 4.5324854980322016e-13, "grad_norm": 0.32151007652282715, "learning_rate": 5.2590539233862846e-05, "loss": 0.0099, "step": 65320 }, { "epoch": 4.5331793874225923e-13, "grad_norm": 0.3384244740009308, "learning_rate": 5.2571142472235936e-05, "loss": 0.0108, "step": 65330 }, { "epoch": 4.533873276812983e-13, "grad_norm": 0.3434867858886719, "learning_rate": 5.25517487862648e-05, "loss": 0.01, "step": 65340 }, { "epoch": 4.5345671662033737e-13, "grad_norm": 0.34799882769584656, "learning_rate": 5.2532358177942456e-05, "loss": 0.0106, "step": 65350 }, { "epoch": 4.5352610555937645e-13, "grad_norm": 0.32869914174079895, "learning_rate": 5.251297064926159e-05, "loss": 0.0099, "step": 65360 }, { "epoch": 4.535954944984155e-13, "grad_norm": 0.3042822480201721, "learning_rate": 5.2493586202214564e-05, "loss": 0.0098, "step": 65370 }, { "epoch": 4.536648834374546e-13, "grad_norm": 0.29265227913856506, "learning_rate": 5.247420483879344e-05, "loss": 0.0098, "step": 65380 }, { "epoch": 4.5373427237649366e-13, "grad_norm": 0.2746927738189697, "learning_rate": 5.2454826560989944e-05, "loss": 0.0098, "step": 65390 }, { "epoch": 4.5380366131553274e-13, "grad_norm": 0.31556060910224915, "learning_rate": 5.2435451370795505e-05, "loss": 0.01, "step": 65400 }, { "epoch": 4.538730502545718e-13, "grad_norm": 0.2874470055103302, "learning_rate": 5.241607927020121e-05, "loss": 0.0095, "step": 65410 }, { "epoch": 4.539424391936109e-13, "grad_norm": 0.26702168583869934, "learning_rate": 5.239671026119787e-05, "loss": 0.0095, "step": 65420 }, { "epoch": 4.5401182813264995e-13, "grad_norm": 0.30332833528518677, "learning_rate": 5.237734434577593e-05, "loss": 0.0099, "step": 65430 }, { "epoch": 4.54081217071689e-13, "grad_norm": 0.3089481294155121, "learning_rate": 5.235798152592557e-05, "loss": 0.0096, "step": 65440 }, { "epoch": 4.541506060107281e-13, "grad_norm": 0.31057292222976685, "learning_rate": 5.2338621803636554e-05, "loss": 0.0097, "step": 65450 }, { "epoch": 4.5421999494976717e-13, "grad_norm": 0.2923828363418579, "learning_rate": 5.231926518089848e-05, "loss": 0.0098, "step": 65460 }, { "epoch": 4.5428938388880624e-13, "grad_norm": 0.2907666265964508, "learning_rate": 5.2299911659700465e-05, "loss": 0.0097, "step": 65470 }, { "epoch": 4.543587728278453e-13, "grad_norm": 0.32486405968666077, "learning_rate": 5.228056124203145e-05, "loss": 0.0103, "step": 65480 }, { "epoch": 4.544281617668844e-13, "grad_norm": 0.3119240403175354, "learning_rate": 5.226121392987992e-05, "loss": 0.0101, "step": 65490 }, { "epoch": 4.5449755070592346e-13, "grad_norm": 0.3754616975784302, "learning_rate": 5.22418697252342e-05, "loss": 0.0099, "step": 65500 }, { "epoch": 4.5456693964496253e-13, "grad_norm": 0.28071776032447815, "learning_rate": 5.222252863008214e-05, "loss": 0.01, "step": 65510 }, { "epoch": 4.546363285840016e-13, "grad_norm": 0.28604644536972046, "learning_rate": 5.220319064641137e-05, "loss": 0.0104, "step": 65520 }, { "epoch": 4.547057175230407e-13, "grad_norm": 0.3712790906429291, "learning_rate": 5.2183855776209145e-05, "loss": 0.0098, "step": 65530 }, { "epoch": 4.547751064620797e-13, "grad_norm": 0.3805556893348694, "learning_rate": 5.216452402146244e-05, "loss": 0.0101, "step": 65540 }, { "epoch": 4.548444954011188e-13, "grad_norm": 0.29448577761650085, "learning_rate": 5.214519538415792e-05, "loss": 0.0097, "step": 65550 }, { "epoch": 4.549138843401579e-13, "grad_norm": 0.3233986794948578, "learning_rate": 5.212586986628182e-05, "loss": 0.0097, "step": 65560 }, { "epoch": 4.54983273279197e-13, "grad_norm": 0.30704864859580994, "learning_rate": 5.210654746982026e-05, "loss": 0.0094, "step": 65570 }, { "epoch": 4.55052662218236e-13, "grad_norm": 0.342268705368042, "learning_rate": 5.2087228196758755e-05, "loss": 0.0096, "step": 65580 }, { "epoch": 4.551220511572751e-13, "grad_norm": 0.26770707964897156, "learning_rate": 5.2067912049082845e-05, "loss": 0.0098, "step": 65590 }, { "epoch": 4.551914400963142e-13, "grad_norm": 0.30570900440216064, "learning_rate": 5.20485990287774e-05, "loss": 0.0096, "step": 65600 }, { "epoch": 4.552608290353533e-13, "grad_norm": 0.31497371196746826, "learning_rate": 5.202928913782729e-05, "loss": 0.0098, "step": 65610 }, { "epoch": 4.553302179743923e-13, "grad_norm": 0.2986777126789093, "learning_rate": 5.200998237821673e-05, "loss": 0.0098, "step": 65620 }, { "epoch": 4.553996069134314e-13, "grad_norm": 0.2846105992794037, "learning_rate": 5.199067875192997e-05, "loss": 0.0097, "step": 65630 }, { "epoch": 4.554689958524705e-13, "grad_norm": 0.299781858921051, "learning_rate": 5.1971378260950626e-05, "loss": 0.0101, "step": 65640 }, { "epoch": 4.555383847915095e-13, "grad_norm": 0.25842607021331787, "learning_rate": 5.1952080907262186e-05, "loss": 0.0097, "step": 65650 }, { "epoch": 4.556077737305486e-13, "grad_norm": 0.31699925661087036, "learning_rate": 5.193278669284773e-05, "loss": 0.0098, "step": 65660 }, { "epoch": 4.556771626695877e-13, "grad_norm": 0.2969532608985901, "learning_rate": 5.1913495619690046e-05, "loss": 0.0097, "step": 65670 }, { "epoch": 4.557465516086268e-13, "grad_norm": 0.2722286581993103, "learning_rate": 5.189420768977159e-05, "loss": 0.01, "step": 65680 }, { "epoch": 4.558159405476658e-13, "grad_norm": 0.27898111939430237, "learning_rate": 5.187492290507451e-05, "loss": 0.0095, "step": 65690 }, { "epoch": 4.558853294867049e-13, "grad_norm": 0.31490784883499146, "learning_rate": 5.185564126758059e-05, "loss": 0.01, "step": 65700 }, { "epoch": 4.55954718425744e-13, "grad_norm": 0.2700248956680298, "learning_rate": 5.183636277927135e-05, "loss": 0.0097, "step": 65710 }, { "epoch": 4.56024107364783e-13, "grad_norm": 0.26993727684020996, "learning_rate": 5.1817087442127936e-05, "loss": 0.0095, "step": 65720 }, { "epoch": 4.560934963038221e-13, "grad_norm": 0.292436420917511, "learning_rate": 5.179781525813118e-05, "loss": 0.0094, "step": 65730 }, { "epoch": 4.561628852428612e-13, "grad_norm": 0.30739542841911316, "learning_rate": 5.177854622926163e-05, "loss": 0.0097, "step": 65740 }, { "epoch": 4.562322741819003e-13, "grad_norm": 0.2568281292915344, "learning_rate": 5.175928035749944e-05, "loss": 0.0095, "step": 65750 }, { "epoch": 4.563016631209393e-13, "grad_norm": 0.30263733863830566, "learning_rate": 5.174001764482451e-05, "loss": 0.0094, "step": 65760 }, { "epoch": 4.563710520599784e-13, "grad_norm": 0.2984101474285126, "learning_rate": 5.1720758093216396e-05, "loss": 0.0097, "step": 65770 }, { "epoch": 4.564404409990175e-13, "grad_norm": 0.30389297008514404, "learning_rate": 5.170150170465424e-05, "loss": 0.0096, "step": 65780 }, { "epoch": 4.565098299380566e-13, "grad_norm": 0.28483596444129944, "learning_rate": 5.168224848111705e-05, "loss": 0.0096, "step": 65790 }, { "epoch": 4.565792188770956e-13, "grad_norm": 0.284747838973999, "learning_rate": 5.166299842458326e-05, "loss": 0.0101, "step": 65800 }, { "epoch": 4.566486078161347e-13, "grad_norm": 0.28607410192489624, "learning_rate": 5.164375153703125e-05, "loss": 0.0092, "step": 65810 }, { "epoch": 4.567179967551738e-13, "grad_norm": 0.29948848485946655, "learning_rate": 5.1624507820438816e-05, "loss": 0.0102, "step": 65820 }, { "epoch": 4.567873856942128e-13, "grad_norm": 0.3440502882003784, "learning_rate": 5.1605267276783643e-05, "loss": 0.0098, "step": 65830 }, { "epoch": 4.568567746332519e-13, "grad_norm": 0.3283948004245758, "learning_rate": 5.158602990804295e-05, "loss": 0.0096, "step": 65840 }, { "epoch": 4.56926163572291e-13, "grad_norm": 0.27816301584243774, "learning_rate": 5.156679571619368e-05, "loss": 0.0096, "step": 65850 }, { "epoch": 4.569955525113301e-13, "grad_norm": 0.30626076459884644, "learning_rate": 5.154756470321244e-05, "loss": 0.0095, "step": 65860 }, { "epoch": 4.570649414503691e-13, "grad_norm": 0.3080982267856598, "learning_rate": 5.1528336871075535e-05, "loss": 0.0096, "step": 65870 }, { "epoch": 4.571343303894082e-13, "grad_norm": 0.2953481078147888, "learning_rate": 5.150911222175892e-05, "loss": 0.0095, "step": 65880 }, { "epoch": 4.572037193284473e-13, "grad_norm": 0.29024335741996765, "learning_rate": 5.148989075723821e-05, "loss": 0.0094, "step": 65890 }, { "epoch": 4.572731082674864e-13, "grad_norm": 0.334625244140625, "learning_rate": 5.1470672479488734e-05, "loss": 0.0098, "step": 65900 }, { "epoch": 4.573424972065254e-13, "grad_norm": 0.29631003737449646, "learning_rate": 5.145145739048544e-05, "loss": 0.0092, "step": 65910 }, { "epoch": 4.574118861455645e-13, "grad_norm": 0.24200914800167084, "learning_rate": 5.143224549220301e-05, "loss": 0.0095, "step": 65920 }, { "epoch": 4.574812750846036e-13, "grad_norm": 0.3076699674129486, "learning_rate": 5.141303678661573e-05, "loss": 0.01, "step": 65930 }, { "epoch": 4.575506640236426e-13, "grad_norm": 0.295560359954834, "learning_rate": 5.139383127569762e-05, "loss": 0.0097, "step": 65940 }, { "epoch": 4.576200529626817e-13, "grad_norm": 0.2835594713687897, "learning_rate": 5.137462896142234e-05, "loss": 0.01, "step": 65950 }, { "epoch": 4.576894419017208e-13, "grad_norm": 0.29180189967155457, "learning_rate": 5.135542984576322e-05, "loss": 0.0099, "step": 65960 }, { "epoch": 4.577588308407599e-13, "grad_norm": 0.341827392578125, "learning_rate": 5.1336233930693296e-05, "loss": 0.0095, "step": 65970 }, { "epoch": 4.578282197797989e-13, "grad_norm": 0.3160187304019928, "learning_rate": 5.13170412181852e-05, "loss": 0.0098, "step": 65980 }, { "epoch": 4.57897608718838e-13, "grad_norm": 0.2897961735725403, "learning_rate": 5.12978517102113e-05, "loss": 0.0098, "step": 65990 }, { "epoch": 4.579669976578771e-13, "grad_norm": 0.3117341697216034, "learning_rate": 5.1278665408743635e-05, "loss": 0.0095, "step": 66000 }, { "epoch": 4.580363865969161e-13, "grad_norm": 0.2945639491081238, "learning_rate": 5.125948231575387e-05, "loss": 0.0095, "step": 66010 }, { "epoch": 4.581057755359552e-13, "grad_norm": 0.29841625690460205, "learning_rate": 5.124030243321337e-05, "loss": 0.0095, "step": 66020 }, { "epoch": 4.581751644749943e-13, "grad_norm": 0.2983202338218689, "learning_rate": 5.122112576309318e-05, "loss": 0.0093, "step": 66030 }, { "epoch": 4.582445534140334e-13, "grad_norm": 0.30881717801094055, "learning_rate": 5.120195230736401e-05, "loss": 0.0093, "step": 66040 }, { "epoch": 4.583139423530724e-13, "grad_norm": 0.2839643061161041, "learning_rate": 5.11827820679962e-05, "loss": 0.0094, "step": 66050 }, { "epoch": 4.583833312921115e-13, "grad_norm": 0.30790141224861145, "learning_rate": 5.116361504695981e-05, "loss": 0.0094, "step": 66060 }, { "epoch": 4.584527202311506e-13, "grad_norm": 0.27678921818733215, "learning_rate": 5.1144451246224554e-05, "loss": 0.0092, "step": 66070 }, { "epoch": 4.585221091701897e-13, "grad_norm": 0.2769167423248291, "learning_rate": 5.1125290667759795e-05, "loss": 0.0098, "step": 66080 }, { "epoch": 4.585914981092287e-13, "grad_norm": 0.3463283181190491, "learning_rate": 5.11061333135346e-05, "loss": 0.0095, "step": 66090 }, { "epoch": 4.586608870482678e-13, "grad_norm": 0.2479957640171051, "learning_rate": 5.108697918551769e-05, "loss": 0.0097, "step": 66100 }, { "epoch": 4.587302759873069e-13, "grad_norm": 0.3055058717727661, "learning_rate": 5.106782828567738e-05, "loss": 0.0103, "step": 66110 }, { "epoch": 4.587996649263459e-13, "grad_norm": 0.3048938512802124, "learning_rate": 5.1048680615981844e-05, "loss": 0.0097, "step": 66120 }, { "epoch": 4.58869053865385e-13, "grad_norm": 0.3382197320461273, "learning_rate": 5.102953617839866e-05, "loss": 0.0099, "step": 66130 }, { "epoch": 4.589384428044241e-13, "grad_norm": 0.30055177211761475, "learning_rate": 5.1010394974895375e-05, "loss": 0.0099, "step": 66140 }, { "epoch": 4.590078317434632e-13, "grad_norm": 0.3169719874858856, "learning_rate": 5.099125700743888e-05, "loss": 0.0098, "step": 66150 }, { "epoch": 4.590772206825022e-13, "grad_norm": 0.3380206525325775, "learning_rate": 5.097212227799605e-05, "loss": 0.0095, "step": 66160 }, { "epoch": 4.591466096215413e-13, "grad_norm": 0.2802402973175049, "learning_rate": 5.095299078853318e-05, "loss": 0.0094, "step": 66170 }, { "epoch": 4.592159985605804e-13, "grad_norm": 0.26928114891052246, "learning_rate": 5.093386254101636e-05, "loss": 0.0095, "step": 66180 }, { "epoch": 4.592853874996194e-13, "grad_norm": 0.2996124029159546, "learning_rate": 5.0914737537411306e-05, "loss": 0.0094, "step": 66190 }, { "epoch": 4.593547764386585e-13, "grad_norm": 0.30544522404670715, "learning_rate": 5.089561577968342e-05, "loss": 0.0094, "step": 66200 }, { "epoch": 4.594241653776976e-13, "grad_norm": 0.3240213394165039, "learning_rate": 5.087649726979775e-05, "loss": 0.0095, "step": 66210 }, { "epoch": 4.594935543167367e-13, "grad_norm": 0.34066399931907654, "learning_rate": 5.085738200971903e-05, "loss": 0.0093, "step": 66220 }, { "epoch": 4.595629432557757e-13, "grad_norm": 0.2964239716529846, "learning_rate": 5.083827000141163e-05, "loss": 0.0097, "step": 66230 }, { "epoch": 4.596323321948148e-13, "grad_norm": 0.28944113850593567, "learning_rate": 5.0819161246839634e-05, "loss": 0.0099, "step": 66240 }, { "epoch": 4.597017211338539e-13, "grad_norm": 0.32240790128707886, "learning_rate": 5.0800055747966764e-05, "loss": 0.0096, "step": 66250 }, { "epoch": 4.59771110072893e-13, "grad_norm": 0.3177180588245392, "learning_rate": 5.0780953506756385e-05, "loss": 0.0099, "step": 66260 }, { "epoch": 4.59840499011932e-13, "grad_norm": 0.2806799113750458, "learning_rate": 5.0761854525171554e-05, "loss": 0.0096, "step": 66270 }, { "epoch": 4.599098879509711e-13, "grad_norm": 0.31224265694618225, "learning_rate": 5.0742758805174995e-05, "loss": 0.0095, "step": 66280 }, { "epoch": 4.599792768900102e-13, "grad_norm": 0.30667707324028015, "learning_rate": 5.0723666348729106e-05, "loss": 0.01, "step": 66290 }, { "epoch": 4.600486658290492e-13, "grad_norm": 0.2746841609477997, "learning_rate": 5.070457715779592e-05, "loss": 0.0098, "step": 66300 }, { "epoch": 4.601180547680883e-13, "grad_norm": 0.2713482081890106, "learning_rate": 5.0685491234337116e-05, "loss": 0.0094, "step": 66310 }, { "epoch": 4.601874437071274e-13, "grad_norm": 0.2902302145957947, "learning_rate": 5.066640858031413e-05, "loss": 0.0095, "step": 66320 }, { "epoch": 4.602568326461665e-13, "grad_norm": 0.32663291692733765, "learning_rate": 5.0647329197687933e-05, "loss": 0.0093, "step": 66330 }, { "epoch": 4.603262215852055e-13, "grad_norm": 0.29176589846611023, "learning_rate": 5.062825308841931e-05, "loss": 0.0094, "step": 66340 }, { "epoch": 4.603956105242446e-13, "grad_norm": 0.306302011013031, "learning_rate": 5.060918025446855e-05, "loss": 0.0091, "step": 66350 }, { "epoch": 4.604649994632837e-13, "grad_norm": 0.2824706435203552, "learning_rate": 5.059011069779577e-05, "loss": 0.0093, "step": 66360 }, { "epoch": 4.605343884023227e-13, "grad_norm": 0.26791542768478394, "learning_rate": 5.057104442036053e-05, "loss": 0.0096, "step": 66370 }, { "epoch": 4.606037773413618e-13, "grad_norm": 0.29248037934303284, "learning_rate": 5.0551981424122354e-05, "loss": 0.0094, "step": 66380 }, { "epoch": 4.606731662804009e-13, "grad_norm": 0.2747436463832855, "learning_rate": 5.053292171104013e-05, "loss": 0.0096, "step": 66390 }, { "epoch": 4.6074255521944e-13, "grad_norm": 0.25657063722610474, "learning_rate": 5.051386528307259e-05, "loss": 0.0099, "step": 66400 }, { "epoch": 4.60811944158479e-13, "grad_norm": 0.2915452718734741, "learning_rate": 5.049481214217808e-05, "loss": 0.0098, "step": 66410 }, { "epoch": 4.608813330975181e-13, "grad_norm": 0.3305865526199341, "learning_rate": 5.0475762290314605e-05, "loss": 0.0103, "step": 66420 }, { "epoch": 4.609507220365572e-13, "grad_norm": 0.28432512283325195, "learning_rate": 5.045671572943983e-05, "loss": 0.0097, "step": 66430 }, { "epoch": 4.610201109755963e-13, "grad_norm": 0.3296300768852234, "learning_rate": 5.043767246151106e-05, "loss": 0.0096, "step": 66440 }, { "epoch": 4.610894999146353e-13, "grad_norm": 0.2747706174850464, "learning_rate": 5.0418632488485364e-05, "loss": 0.0098, "step": 66450 }, { "epoch": 4.611588888536744e-13, "grad_norm": 0.3132672607898712, "learning_rate": 5.0399595812319264e-05, "loss": 0.0099, "step": 66460 }, { "epoch": 4.612282777927135e-13, "grad_norm": 0.3057853877544403, "learning_rate": 5.038056243496923e-05, "loss": 0.0094, "step": 66470 }, { "epoch": 4.612976667317525e-13, "grad_norm": 0.2556556761264801, "learning_rate": 5.03615323583911e-05, "loss": 0.0094, "step": 66480 }, { "epoch": 4.613670556707916e-13, "grad_norm": 0.28686875104904175, "learning_rate": 5.0342505584540644e-05, "loss": 0.0094, "step": 66490 }, { "epoch": 4.614364446098307e-13, "grad_norm": 0.24491091072559357, "learning_rate": 5.0323482115373054e-05, "loss": 0.0092, "step": 66500 }, { "epoch": 4.615058335488698e-13, "grad_norm": 0.3277246356010437, "learning_rate": 5.0304461952843306e-05, "loss": 0.0091, "step": 66510 }, { "epoch": 4.615752224879088e-13, "grad_norm": 0.2701321244239807, "learning_rate": 5.028544509890605e-05, "loss": 0.0092, "step": 66520 }, { "epoch": 4.616446114269479e-13, "grad_norm": 0.26988762617111206, "learning_rate": 5.0266431555515535e-05, "loss": 0.0093, "step": 66530 }, { "epoch": 4.61714000365987e-13, "grad_norm": 0.2687923312187195, "learning_rate": 5.024742132462572e-05, "loss": 0.0093, "step": 66540 }, { "epoch": 4.61783389305026e-13, "grad_norm": 0.3228974938392639, "learning_rate": 5.0228414408190185e-05, "loss": 0.0099, "step": 66550 }, { "epoch": 4.618527782440651e-13, "grad_norm": 0.2718633711338043, "learning_rate": 5.0209410808162193e-05, "loss": 0.0095, "step": 66560 }, { "epoch": 4.619221671831042e-13, "grad_norm": 0.29051730036735535, "learning_rate": 5.019041052649466e-05, "loss": 0.0102, "step": 66570 }, { "epoch": 4.619915561221433e-13, "grad_norm": 0.3007703721523285, "learning_rate": 5.017141356514017e-05, "loss": 0.0098, "step": 66580 }, { "epoch": 4.620609450611823e-13, "grad_norm": 0.3049209415912628, "learning_rate": 5.015241992605093e-05, "loss": 0.0094, "step": 66590 }, { "epoch": 4.621303340002214e-13, "grad_norm": 0.3163311779499054, "learning_rate": 5.013342961117887e-05, "loss": 0.0097, "step": 66600 }, { "epoch": 4.621997229392605e-13, "grad_norm": 0.308834969997406, "learning_rate": 5.0114442622475505e-05, "loss": 0.0095, "step": 66610 }, { "epoch": 4.622691118782996e-13, "grad_norm": 0.3000316917896271, "learning_rate": 5.009545896189205e-05, "loss": 0.0095, "step": 66620 }, { "epoch": 4.623385008173386e-13, "grad_norm": 0.2873081564903259, "learning_rate": 5.007647863137941e-05, "loss": 0.0095, "step": 66630 }, { "epoch": 4.624078897563777e-13, "grad_norm": 0.29452669620513916, "learning_rate": 5.005750163288804e-05, "loss": 0.009, "step": 66640 }, { "epoch": 4.624772786954168e-13, "grad_norm": 0.30748170614242554, "learning_rate": 5.00385279683682e-05, "loss": 0.0092, "step": 66650 }, { "epoch": 4.625466676344558e-13, "grad_norm": 0.29520848393440247, "learning_rate": 5.001955763976964e-05, "loss": 0.0091, "step": 66660 }, { "epoch": 4.626160565734949e-13, "grad_norm": 0.27072322368621826, "learning_rate": 5.000059064904196e-05, "loss": 0.0091, "step": 66670 }, { "epoch": 4.62685445512534e-13, "grad_norm": 0.2756080627441406, "learning_rate": 4.9981626998134206e-05, "loss": 0.0093, "step": 66680 }, { "epoch": 4.627548344515731e-13, "grad_norm": 0.30937501788139343, "learning_rate": 4.99626666889953e-05, "loss": 0.0093, "step": 66690 }, { "epoch": 4.628242233906121e-13, "grad_norm": 0.2922912538051605, "learning_rate": 4.994370972357358e-05, "loss": 0.0093, "step": 66700 }, { "epoch": 4.628936123296512e-13, "grad_norm": 0.29258233308792114, "learning_rate": 4.992475610381731e-05, "loss": 0.0096, "step": 66710 }, { "epoch": 4.629630012686903e-13, "grad_norm": 0.30808642506599426, "learning_rate": 4.9905805831674166e-05, "loss": 0.0096, "step": 66720 }, { "epoch": 4.630323902077293e-13, "grad_norm": 0.31175798177719116, "learning_rate": 4.988685890909161e-05, "loss": 0.0097, "step": 66730 }, { "epoch": 4.631017791467684e-13, "grad_norm": 0.28750208020210266, "learning_rate": 4.986791533801674e-05, "loss": 0.0095, "step": 66740 }, { "epoch": 4.631711680858075e-13, "grad_norm": 0.30179572105407715, "learning_rate": 4.984897512039631e-05, "loss": 0.0092, "step": 66750 }, { "epoch": 4.632405570248466e-13, "grad_norm": 0.2924845218658447, "learning_rate": 4.983003825817671e-05, "loss": 0.0096, "step": 66760 }, { "epoch": 4.633099459638856e-13, "grad_norm": 0.2736211121082306, "learning_rate": 4.981110475330401e-05, "loss": 0.0101, "step": 66770 }, { "epoch": 4.633793349029247e-13, "grad_norm": 0.36085888743400574, "learning_rate": 4.979217460772391e-05, "loss": 0.0095, "step": 66780 }, { "epoch": 4.634487238419638e-13, "grad_norm": 0.2790614366531372, "learning_rate": 4.9773247823381777e-05, "loss": 0.009, "step": 66790 }, { "epoch": 4.635181127810029e-13, "grad_norm": 0.27531668543815613, "learning_rate": 4.9754324402222626e-05, "loss": 0.0093, "step": 66800 }, { "epoch": 4.635875017200419e-13, "grad_norm": 0.28229328989982605, "learning_rate": 4.973540434619115e-05, "loss": 0.0089, "step": 66810 }, { "epoch": 4.63656890659081e-13, "grad_norm": 0.2883852422237396, "learning_rate": 4.971648765723168e-05, "loss": 0.009, "step": 66820 }, { "epoch": 4.637262795981201e-13, "grad_norm": 0.31070879101753235, "learning_rate": 4.969757433728821e-05, "loss": 0.0092, "step": 66830 }, { "epoch": 4.637956685371591e-13, "grad_norm": 0.28066450357437134, "learning_rate": 4.9678664388304315e-05, "loss": 0.0093, "step": 66840 }, { "epoch": 4.638650574761982e-13, "grad_norm": 0.2965380847454071, "learning_rate": 4.9659757812223374e-05, "loss": 0.0093, "step": 66850 }, { "epoch": 4.639344464152373e-13, "grad_norm": 0.3031172752380371, "learning_rate": 4.9640854610988245e-05, "loss": 0.0093, "step": 66860 }, { "epoch": 4.640038353542764e-13, "grad_norm": 0.27414602041244507, "learning_rate": 4.962195478654162e-05, "loss": 0.0095, "step": 66870 }, { "epoch": 4.640732242933154e-13, "grad_norm": 0.27438607811927795, "learning_rate": 4.9603058340825666e-05, "loss": 0.0095, "step": 66880 }, { "epoch": 4.641426132323545e-13, "grad_norm": 0.26100531220436096, "learning_rate": 4.958416527578234e-05, "loss": 0.0096, "step": 66890 }, { "epoch": 4.642120021713936e-13, "grad_norm": 0.2850794196128845, "learning_rate": 4.956527559335317e-05, "loss": 0.0093, "step": 66900 }, { "epoch": 4.642813911104327e-13, "grad_norm": 0.29629215598106384, "learning_rate": 4.954638929547938e-05, "loss": 0.0094, "step": 66910 }, { "epoch": 4.643507800494717e-13, "grad_norm": 0.2904321551322937, "learning_rate": 4.9527506384101814e-05, "loss": 0.0094, "step": 66920 }, { "epoch": 4.644201689885108e-13, "grad_norm": 0.260958194732666, "learning_rate": 4.9508626861160996e-05, "loss": 0.0098, "step": 66930 }, { "epoch": 4.644895579275499e-13, "grad_norm": 0.28992515802383423, "learning_rate": 4.94897507285971e-05, "loss": 0.0094, "step": 66940 }, { "epoch": 4.645589468665889e-13, "grad_norm": 0.29869264364242554, "learning_rate": 4.947087798834993e-05, "loss": 0.0092, "step": 66950 }, { "epoch": 4.64628335805628e-13, "grad_norm": 0.24849051237106323, "learning_rate": 4.9452008642358984e-05, "loss": 0.0091, "step": 66960 }, { "epoch": 4.646977247446671e-13, "grad_norm": 0.2671033442020416, "learning_rate": 4.94331426925633e-05, "loss": 0.0089, "step": 66970 }, { "epoch": 4.647671136837062e-13, "grad_norm": 0.27565819025039673, "learning_rate": 4.941428014090176e-05, "loss": 0.009, "step": 66980 }, { "epoch": 4.648365026227452e-13, "grad_norm": 0.3095563054084778, "learning_rate": 4.939542098931267e-05, "loss": 0.0093, "step": 66990 }, { "epoch": 4.649058915617843e-13, "grad_norm": 0.2605685293674469, "learning_rate": 4.937656523973421e-05, "loss": 0.0091, "step": 67000 }, { "epoch": 4.649752805008234e-13, "grad_norm": 0.3101874589920044, "learning_rate": 4.9357712894104e-05, "loss": 0.009, "step": 67010 }, { "epoch": 4.650446694398624e-13, "grad_norm": 0.2722776234149933, "learning_rate": 4.9338863954359504e-05, "loss": 0.0094, "step": 67020 }, { "epoch": 4.651140583789015e-13, "grad_norm": 0.307711660861969, "learning_rate": 4.9320018422437657e-05, "loss": 0.0092, "step": 67030 }, { "epoch": 4.651834473179406e-13, "grad_norm": 0.28965163230895996, "learning_rate": 4.930117630027523e-05, "loss": 0.0095, "step": 67040 }, { "epoch": 4.652528362569797e-13, "grad_norm": 0.25412487983703613, "learning_rate": 4.928233758980846e-05, "loss": 0.0096, "step": 67050 }, { "epoch": 4.653222251960187e-13, "grad_norm": 0.26955369114875793, "learning_rate": 4.926350229297335e-05, "loss": 0.0088, "step": 67060 }, { "epoch": 4.653916141350578e-13, "grad_norm": 0.30734023451805115, "learning_rate": 4.924467041170551e-05, "loss": 0.0097, "step": 67070 }, { "epoch": 4.654610030740969e-13, "grad_norm": 0.3031429350376129, "learning_rate": 4.922584194794022e-05, "loss": 0.0095, "step": 67080 }, { "epoch": 4.65530392013136e-13, "grad_norm": 0.27925166487693787, "learning_rate": 4.920701690361239e-05, "loss": 0.0091, "step": 67090 }, { "epoch": 4.65599780952175e-13, "grad_norm": 0.2704828083515167, "learning_rate": 4.918819528065659e-05, "loss": 0.0091, "step": 67100 }, { "epoch": 4.656691698912141e-13, "grad_norm": 0.2971114218235016, "learning_rate": 4.916937708100704e-05, "loss": 0.0089, "step": 67110 }, { "epoch": 4.657385588302532e-13, "grad_norm": 0.2878607511520386, "learning_rate": 4.915056230659758e-05, "loss": 0.0089, "step": 67120 }, { "epoch": 4.658079477692922e-13, "grad_norm": 0.2796492278575897, "learning_rate": 4.9131750959361747e-05, "loss": 0.009, "step": 67130 }, { "epoch": 4.658773367083313e-13, "grad_norm": 0.2588178217411041, "learning_rate": 4.91129430412327e-05, "loss": 0.0089, "step": 67140 }, { "epoch": 4.659467256473704e-13, "grad_norm": 0.29487353563308716, "learning_rate": 4.909413855414323e-05, "loss": 0.0095, "step": 67150 }, { "epoch": 4.660161145864095e-13, "grad_norm": 0.2923057973384857, "learning_rate": 4.907533750002578e-05, "loss": 0.0088, "step": 67160 }, { "epoch": 4.660855035254485e-13, "grad_norm": 0.2338510900735855, "learning_rate": 4.9056539880812486e-05, "loss": 0.0089, "step": 67170 }, { "epoch": 4.661548924644876e-13, "grad_norm": 0.273405522108078, "learning_rate": 4.903774569843509e-05, "loss": 0.0091, "step": 67180 }, { "epoch": 4.662242814035267e-13, "grad_norm": 0.31486427783966064, "learning_rate": 4.901895495482493e-05, "loss": 0.0095, "step": 67190 }, { "epoch": 4.662936703425657e-13, "grad_norm": 0.2754594385623932, "learning_rate": 4.9000167651913136e-05, "loss": 0.0092, "step": 67200 }, { "epoch": 4.663630592816048e-13, "grad_norm": 0.27608436346054077, "learning_rate": 4.8981383791630296e-05, "loss": 0.0095, "step": 67210 }, { "epoch": 4.664324482206439e-13, "grad_norm": 0.2942158281803131, "learning_rate": 4.8962603375906865e-05, "loss": 0.0088, "step": 67220 }, { "epoch": 4.66501837159683e-13, "grad_norm": 0.31109440326690674, "learning_rate": 4.89438264066727e-05, "loss": 0.0094, "step": 67230 }, { "epoch": 4.66571226098722e-13, "grad_norm": 0.3156225383281708, "learning_rate": 4.892505288585752e-05, "loss": 0.0096, "step": 67240 }, { "epoch": 4.666406150377611e-13, "grad_norm": 0.28829094767570496, "learning_rate": 4.8906282815390555e-05, "loss": 0.0091, "step": 67250 }, { "epoch": 4.667100039768002e-13, "grad_norm": 0.27905964851379395, "learning_rate": 4.88875161972007e-05, "loss": 0.0091, "step": 67260 }, { "epoch": 4.667793929158393e-13, "grad_norm": 0.2973497211933136, "learning_rate": 4.886875303321655e-05, "loss": 0.009, "step": 67270 }, { "epoch": 4.668487818548783e-13, "grad_norm": 0.29267236590385437, "learning_rate": 4.8849993325366316e-05, "loss": 0.009, "step": 67280 }, { "epoch": 4.669181707939174e-13, "grad_norm": 0.30408549308776855, "learning_rate": 4.883123707557783e-05, "loss": 0.0089, "step": 67290 }, { "epoch": 4.669875597329565e-13, "grad_norm": 0.28185251355171204, "learning_rate": 4.8812484285778607e-05, "loss": 0.0089, "step": 67300 }, { "epoch": 4.670569486719955e-13, "grad_norm": 0.2922477126121521, "learning_rate": 4.879373495789577e-05, "loss": 0.0094, "step": 67310 }, { "epoch": 4.671263376110346e-13, "grad_norm": 0.26556551456451416, "learning_rate": 4.8774989093856126e-05, "loss": 0.0088, "step": 67320 }, { "epoch": 4.671957265500737e-13, "grad_norm": 0.2824840247631073, "learning_rate": 4.875624669558607e-05, "loss": 0.0091, "step": 67330 }, { "epoch": 4.672651154891128e-13, "grad_norm": 0.26494356989860535, "learning_rate": 4.873750776501171e-05, "loss": 0.009, "step": 67340 }, { "epoch": 4.673345044281518e-13, "grad_norm": 0.2774065136909485, "learning_rate": 4.871877230405877e-05, "loss": 0.0095, "step": 67350 }, { "epoch": 4.674038933671909e-13, "grad_norm": 0.2800314426422119, "learning_rate": 4.870004031465253e-05, "loss": 0.0093, "step": 67360 }, { "epoch": 4.6747328230623e-13, "grad_norm": 0.25381922721862793, "learning_rate": 4.8681311798718114e-05, "loss": 0.0092, "step": 67370 }, { "epoch": 4.67542671245269e-13, "grad_norm": 0.3179624378681183, "learning_rate": 4.866258675818007e-05, "loss": 0.009, "step": 67380 }, { "epoch": 4.676120601843081e-13, "grad_norm": 0.29380670189857483, "learning_rate": 4.8643865194962733e-05, "loss": 0.0093, "step": 67390 }, { "epoch": 4.676814491233472e-13, "grad_norm": 0.29001572728157043, "learning_rate": 4.8625147110990025e-05, "loss": 0.0092, "step": 67400 }, { "epoch": 4.677508380623863e-13, "grad_norm": 0.3062751591205597, "learning_rate": 4.8606432508185525e-05, "loss": 0.009, "step": 67410 }, { "epoch": 4.678202270014253e-13, "grad_norm": 0.25052928924560547, "learning_rate": 4.858772138847244e-05, "loss": 0.0088, "step": 67420 }, { "epoch": 4.678896159404644e-13, "grad_norm": 0.23660103976726532, "learning_rate": 4.856901375377363e-05, "loss": 0.0088, "step": 67430 }, { "epoch": 4.679590048795035e-13, "grad_norm": 0.2647271156311035, "learning_rate": 4.8550309606011606e-05, "loss": 0.009, "step": 67440 }, { "epoch": 4.680283938185426e-13, "grad_norm": 0.27918899059295654, "learning_rate": 4.853160894710849e-05, "loss": 0.0088, "step": 67450 }, { "epoch": 4.680977827575816e-13, "grad_norm": 0.28505805134773254, "learning_rate": 4.851291177898609e-05, "loss": 0.0091, "step": 67460 }, { "epoch": 4.681671716966207e-13, "grad_norm": 0.25989067554473877, "learning_rate": 4.849421810356583e-05, "loss": 0.009, "step": 67470 }, { "epoch": 4.682365606356598e-13, "grad_norm": 0.30030056834220886, "learning_rate": 4.847552792276875e-05, "loss": 0.0089, "step": 67480 }, { "epoch": 4.683059495746988e-13, "grad_norm": 0.2576723098754883, "learning_rate": 4.8456841238515586e-05, "loss": 0.009, "step": 67490 }, { "epoch": 4.683753385137379e-13, "grad_norm": 0.29640084505081177, "learning_rate": 4.843815805272667e-05, "loss": 0.0092, "step": 67500 }, { "epoch": 4.68444727452777e-13, "grad_norm": 0.27476564049720764, "learning_rate": 4.8419478367322e-05, "loss": 0.0091, "step": 67510 }, { "epoch": 4.685141163918161e-13, "grad_norm": 0.2622460722923279, "learning_rate": 4.840080218422118e-05, "loss": 0.0091, "step": 67520 }, { "epoch": 4.685835053308551e-13, "grad_norm": 0.26573097705841064, "learning_rate": 4.838212950534353e-05, "loss": 0.0092, "step": 67530 }, { "epoch": 4.686528942698942e-13, "grad_norm": 0.25946325063705444, "learning_rate": 4.8363460332607866e-05, "loss": 0.009, "step": 67540 }, { "epoch": 4.687222832089333e-13, "grad_norm": 0.2873457372188568, "learning_rate": 4.8344794667932854e-05, "loss": 0.0094, "step": 67550 }, { "epoch": 4.687916721479723e-13, "grad_norm": 0.23813489079475403, "learning_rate": 4.832613251323656e-05, "loss": 0.0093, "step": 67560 }, { "epoch": 4.688610610870114e-13, "grad_norm": 0.2673458158969879, "learning_rate": 4.830747387043693e-05, "loss": 0.0088, "step": 67570 }, { "epoch": 4.689304500260505e-13, "grad_norm": 0.3251059055328369, "learning_rate": 4.828881874145135e-05, "loss": 0.0089, "step": 67580 }, { "epoch": 4.689998389650896e-13, "grad_norm": 0.2742953896522522, "learning_rate": 4.827016712819694e-05, "loss": 0.0087, "step": 67590 }, { "epoch": 4.690692279041286e-13, "grad_norm": 0.2685050368309021, "learning_rate": 4.8251519032590456e-05, "loss": 0.0087, "step": 67600 }, { "epoch": 4.691386168431677e-13, "grad_norm": 0.28410598635673523, "learning_rate": 4.823287445654827e-05, "loss": 0.0088, "step": 67610 }, { "epoch": 4.692080057822068e-13, "grad_norm": 0.25369781255722046, "learning_rate": 4.8214233401986414e-05, "loss": 0.0096, "step": 67620 }, { "epoch": 4.692773947212459e-13, "grad_norm": 0.2638530135154724, "learning_rate": 4.8195595870820536e-05, "loss": 0.0089, "step": 67630 }, { "epoch": 4.693467836602849e-13, "grad_norm": 0.26374179124832153, "learning_rate": 4.817696186496593e-05, "loss": 0.0088, "step": 67640 }, { "epoch": 4.69416172599324e-13, "grad_norm": 0.2646406590938568, "learning_rate": 4.815833138633754e-05, "loss": 0.009, "step": 67650 }, { "epoch": 4.694855615383631e-13, "grad_norm": 0.2947632968425751, "learning_rate": 4.813970443684994e-05, "loss": 0.0091, "step": 67660 }, { "epoch": 4.695549504774021e-13, "grad_norm": 0.28645285964012146, "learning_rate": 4.8121081018417326e-05, "loss": 0.0094, "step": 67670 }, { "epoch": 4.696243394164412e-13, "grad_norm": 0.23962268233299255, "learning_rate": 4.8102461132953546e-05, "loss": 0.0092, "step": 67680 }, { "epoch": 4.696937283554803e-13, "grad_norm": 0.2616957128047943, "learning_rate": 4.808384478237209e-05, "loss": 0.0091, "step": 67690 }, { "epoch": 4.697631172945194e-13, "grad_norm": 0.2779390513896942, "learning_rate": 4.8065231968586065e-05, "loss": 0.009, "step": 67700 }, { "epoch": 4.698325062335584e-13, "grad_norm": 0.2857542634010315, "learning_rate": 4.804662269350826e-05, "loss": 0.0095, "step": 67710 }, { "epoch": 4.699018951725975e-13, "grad_norm": 0.24074165523052216, "learning_rate": 4.8028016959051014e-05, "loss": 0.0088, "step": 67720 }, { "epoch": 4.699712841116366e-13, "grad_norm": 0.23932169377803802, "learning_rate": 4.800941476712643e-05, "loss": 0.0088, "step": 67730 }, { "epoch": 4.700406730506757e-13, "grad_norm": 0.2819550335407257, "learning_rate": 4.799081611964607e-05, "loss": 0.009, "step": 67740 }, { "epoch": 4.701100619897147e-13, "grad_norm": 0.24025024473667145, "learning_rate": 4.7972221018521336e-05, "loss": 0.0085, "step": 67750 }, { "epoch": 4.701794509287538e-13, "grad_norm": 0.27014243602752686, "learning_rate": 4.795362946566309e-05, "loss": 0.0088, "step": 67760 }, { "epoch": 4.702488398677929e-13, "grad_norm": 0.2647782564163208, "learning_rate": 4.7935041462981975e-05, "loss": 0.0089, "step": 67770 }, { "epoch": 4.703182288068319e-13, "grad_norm": 0.2634406089782715, "learning_rate": 4.7916457012388104e-05, "loss": 0.0092, "step": 67780 }, { "epoch": 4.70387617745871e-13, "grad_norm": 0.24471630156040192, "learning_rate": 4.7897876115791425e-05, "loss": 0.0086, "step": 67790 }, { "epoch": 4.704570066849101e-13, "grad_norm": 0.3042019307613373, "learning_rate": 4.787929877510134e-05, "loss": 0.0091, "step": 67800 }, { "epoch": 4.705263956239492e-13, "grad_norm": 0.25518709421157837, "learning_rate": 4.786072499222697e-05, "loss": 0.009, "step": 67810 }, { "epoch": 4.705957845629882e-13, "grad_norm": 0.28060752153396606, "learning_rate": 4.7842154769077085e-05, "loss": 0.0092, "step": 67820 }, { "epoch": 4.706651735020273e-13, "grad_norm": 0.28563064336776733, "learning_rate": 4.7823588107560026e-05, "loss": 0.0098, "step": 67830 }, { "epoch": 4.707345624410664e-13, "grad_norm": 0.2745561897754669, "learning_rate": 4.780502500958388e-05, "loss": 0.0091, "step": 67840 }, { "epoch": 4.708039513801054e-13, "grad_norm": 0.31400465965270996, "learning_rate": 4.7786465477056184e-05, "loss": 0.0094, "step": 67850 }, { "epoch": 4.708733403191445e-13, "grad_norm": 0.3019389510154724, "learning_rate": 4.7767909511884336e-05, "loss": 0.0091, "step": 67860 }, { "epoch": 4.709427292581836e-13, "grad_norm": 0.310271680355072, "learning_rate": 4.774935711597514e-05, "loss": 0.0094, "step": 67870 }, { "epoch": 4.710121181972227e-13, "grad_norm": 0.2740729749202728, "learning_rate": 4.7730808291235225e-05, "loss": 0.0088, "step": 67880 }, { "epoch": 4.710815071362617e-13, "grad_norm": 0.28356629610061646, "learning_rate": 4.7712263039570716e-05, "loss": 0.0087, "step": 67890 }, { "epoch": 4.711508960753008e-13, "grad_norm": 0.27897417545318604, "learning_rate": 4.76937213628875e-05, "loss": 0.0088, "step": 67900 }, { "epoch": 4.712202850143399e-13, "grad_norm": 0.23818424344062805, "learning_rate": 4.767518326309094e-05, "loss": 0.0086, "step": 67910 }, { "epoch": 4.71289673953379e-13, "grad_norm": 0.23314176499843597, "learning_rate": 4.7656648742086167e-05, "loss": 0.0088, "step": 67920 }, { "epoch": 4.71359062892418e-13, "grad_norm": 0.27129891514778137, "learning_rate": 4.763811780177786e-05, "loss": 0.0093, "step": 67930 }, { "epoch": 4.714284518314571e-13, "grad_norm": 0.2668531537055969, "learning_rate": 4.7619590444070385e-05, "loss": 0.0088, "step": 67940 }, { "epoch": 4.714978407704962e-13, "grad_norm": 0.26437386870384216, "learning_rate": 4.76010666708677e-05, "loss": 0.0087, "step": 67950 }, { "epoch": 4.715672297095352e-13, "grad_norm": 0.32269948720932007, "learning_rate": 4.758254648407343e-05, "loss": 0.0091, "step": 67960 }, { "epoch": 4.716366186485743e-13, "grad_norm": 0.26862528920173645, "learning_rate": 4.756402988559079e-05, "loss": 0.0094, "step": 67970 }, { "epoch": 4.717060075876134e-13, "grad_norm": 0.27999600768089294, "learning_rate": 4.754551687732267e-05, "loss": 0.009, "step": 67980 }, { "epoch": 4.717753965266525e-13, "grad_norm": 0.29012051224708557, "learning_rate": 4.752700746117153e-05, "loss": 0.0094, "step": 67990 }, { "epoch": 4.718447854656915e-13, "grad_norm": 0.26984086632728577, "learning_rate": 4.750850163903955e-05, "loss": 0.0087, "step": 68000 }, { "epoch": 4.719141744047306e-13, "grad_norm": 0.260538786649704, "learning_rate": 4.7489999412828475e-05, "loss": 0.0092, "step": 68010 }, { "epoch": 4.719835633437697e-13, "grad_norm": 0.25651830434799194, "learning_rate": 4.7471500784439694e-05, "loss": 0.0089, "step": 68020 }, { "epoch": 4.720529522828087e-13, "grad_norm": 0.30189430713653564, "learning_rate": 4.7453005755774215e-05, "loss": 0.0089, "step": 68030 }, { "epoch": 4.721223412218478e-13, "grad_norm": 0.27757057547569275, "learning_rate": 4.7434514328732734e-05, "loss": 0.0087, "step": 68040 }, { "epoch": 4.721917301608869e-13, "grad_norm": 0.27055346965789795, "learning_rate": 4.741602650521545e-05, "loss": 0.0088, "step": 68050 }, { "epoch": 4.72261119099926e-13, "grad_norm": 0.2656692564487457, "learning_rate": 4.739754228712237e-05, "loss": 0.0088, "step": 68060 }, { "epoch": 4.72330508038965e-13, "grad_norm": 0.3355685770511627, "learning_rate": 4.7379061676352944e-05, "loss": 0.0086, "step": 68070 }, { "epoch": 4.723998969780041e-13, "grad_norm": 0.3053951561450958, "learning_rate": 4.7360584674806434e-05, "loss": 0.0088, "step": 68080 }, { "epoch": 4.724692859170432e-13, "grad_norm": 0.2639189660549164, "learning_rate": 4.734211128438155e-05, "loss": 0.009, "step": 68090 }, { "epoch": 4.725386748560823e-13, "grad_norm": 0.2597377896308899, "learning_rate": 4.7323641506976806e-05, "loss": 0.0088, "step": 68100 }, { "epoch": 4.726080637951213e-13, "grad_norm": 0.2820436954498291, "learning_rate": 4.730517534449018e-05, "loss": 0.0086, "step": 68110 }, { "epoch": 4.726774527341604e-13, "grad_norm": 0.2890341877937317, "learning_rate": 4.728671279881943e-05, "loss": 0.0092, "step": 68120 }, { "epoch": 4.727468416731995e-13, "grad_norm": 0.28037524223327637, "learning_rate": 4.726825387186182e-05, "loss": 0.0093, "step": 68130 }, { "epoch": 4.728162306122385e-13, "grad_norm": 0.28797903656959534, "learning_rate": 4.7249798565514294e-05, "loss": 0.0091, "step": 68140 }, { "epoch": 4.728856195512776e-13, "grad_norm": 0.2547895610332489, "learning_rate": 4.7231346881673446e-05, "loss": 0.0091, "step": 68150 }, { "epoch": 4.729550084903167e-13, "grad_norm": 0.2819385528564453, "learning_rate": 4.721289882223547e-05, "loss": 0.0089, "step": 68160 }, { "epoch": 4.730243974293558e-13, "grad_norm": 0.27228134870529175, "learning_rate": 4.719445438909617e-05, "loss": 0.009, "step": 68170 }, { "epoch": 4.730937863683948e-13, "grad_norm": 0.2690829932689667, "learning_rate": 4.717601358415101e-05, "loss": 0.009, "step": 68180 }, { "epoch": 4.731631753074339e-13, "grad_norm": 0.2542329728603363, "learning_rate": 4.715757640929507e-05, "loss": 0.0089, "step": 68190 }, { "epoch": 4.73232564246473e-13, "grad_norm": 0.2953173816204071, "learning_rate": 4.713914286642307e-05, "loss": 0.0088, "step": 68200 }, { "epoch": 4.73301953185512e-13, "grad_norm": 0.23877979815006256, "learning_rate": 4.712071295742933e-05, "loss": 0.0087, "step": 68210 }, { "epoch": 4.733713421245511e-13, "grad_norm": 0.238652303814888, "learning_rate": 4.710228668420781e-05, "loss": 0.0084, "step": 68220 }, { "epoch": 4.734407310635902e-13, "grad_norm": 0.25197091698646545, "learning_rate": 4.708386404865209e-05, "loss": 0.0087, "step": 68230 }, { "epoch": 4.735101200026293e-13, "grad_norm": 0.25745412707328796, "learning_rate": 4.706544505265541e-05, "loss": 0.0089, "step": 68240 }, { "epoch": 4.735795089416683e-13, "grad_norm": 0.26135849952697754, "learning_rate": 4.704702969811058e-05, "loss": 0.0087, "step": 68250 }, { "epoch": 4.736488978807074e-13, "grad_norm": 0.26607003808021545, "learning_rate": 4.702861798691006e-05, "loss": 0.0087, "step": 68260 }, { "epoch": 4.737182868197465e-13, "grad_norm": 0.28726550936698914, "learning_rate": 4.7010209920945954e-05, "loss": 0.0088, "step": 68270 }, { "epoch": 4.737876757587856e-13, "grad_norm": 0.2668599784374237, "learning_rate": 4.699180550210998e-05, "loss": 0.009, "step": 68280 }, { "epoch": 4.738570646978246e-13, "grad_norm": 0.2686091661453247, "learning_rate": 4.697340473229347e-05, "loss": 0.0092, "step": 68290 }, { "epoch": 4.739264536368637e-13, "grad_norm": 0.26124057173728943, "learning_rate": 4.69550076133874e-05, "loss": 0.0095, "step": 68300 }, { "epoch": 4.739958425759028e-13, "grad_norm": 0.24760939180850983, "learning_rate": 4.6936614147282345e-05, "loss": 0.0086, "step": 68310 }, { "epoch": 4.740652315149418e-13, "grad_norm": 0.27389782667160034, "learning_rate": 4.691822433586853e-05, "loss": 0.0092, "step": 68320 }, { "epoch": 4.741346204539809e-13, "grad_norm": 0.2433856874704361, "learning_rate": 4.68998381810358e-05, "loss": 0.0089, "step": 68330 }, { "epoch": 4.7420400939302e-13, "grad_norm": 0.2832699120044708, "learning_rate": 4.6881455684673596e-05, "loss": 0.0088, "step": 68340 }, { "epoch": 4.742733983320591e-13, "grad_norm": 0.260914146900177, "learning_rate": 4.686307684867103e-05, "loss": 0.0086, "step": 68350 }, { "epoch": 4.743427872710981e-13, "grad_norm": 0.2663213014602661, "learning_rate": 4.6844701674916806e-05, "loss": 0.0086, "step": 68360 }, { "epoch": 4.744121762101372e-13, "grad_norm": 0.24222560226917267, "learning_rate": 4.682633016529928e-05, "loss": 0.0084, "step": 68370 }, { "epoch": 4.744815651491763e-13, "grad_norm": 0.26069727540016174, "learning_rate": 4.680796232170634e-05, "loss": 0.0083, "step": 68380 }, { "epoch": 4.745509540882153e-13, "grad_norm": 0.24303239583969116, "learning_rate": 4.678959814602566e-05, "loss": 0.0086, "step": 68390 }, { "epoch": 4.746203430272544e-13, "grad_norm": 0.29559004306793213, "learning_rate": 4.677123764014434e-05, "loss": 0.0092, "step": 68400 }, { "epoch": 4.746897319662935e-13, "grad_norm": 0.27077415585517883, "learning_rate": 4.675288080594934e-05, "loss": 0.0087, "step": 68410 }, { "epoch": 4.747591209053326e-13, "grad_norm": 0.21597139537334442, "learning_rate": 4.673452764532697e-05, "loss": 0.0089, "step": 68420 }, { "epoch": 4.748285098443716e-13, "grad_norm": 0.2720465660095215, "learning_rate": 4.671617816016343e-05, "loss": 0.0087, "step": 68430 }, { "epoch": 4.748978987834107e-13, "grad_norm": 0.2583448588848114, "learning_rate": 4.66978323523443e-05, "loss": 0.009, "step": 68440 }, { "epoch": 4.749672877224498e-13, "grad_norm": 0.2522432804107666, "learning_rate": 4.667949022375502e-05, "loss": 0.0089, "step": 68450 }, { "epoch": 4.750366766614889e-13, "grad_norm": 0.2558937668800354, "learning_rate": 4.666115177628041e-05, "loss": 0.0091, "step": 68460 }, { "epoch": 4.751060656005279e-13, "grad_norm": 0.2632719874382019, "learning_rate": 4.66428170118051e-05, "loss": 0.0085, "step": 68470 }, { "epoch": 4.75175454539567e-13, "grad_norm": 0.30198436975479126, "learning_rate": 4.662448593221325e-05, "loss": 0.0091, "step": 68480 }, { "epoch": 4.752448434786061e-13, "grad_norm": 0.2751086950302124, "learning_rate": 4.660615853938867e-05, "loss": 0.0087, "step": 68490 }, { "epoch": 4.753142324176451e-13, "grad_norm": 0.27120620012283325, "learning_rate": 4.6587834835214786e-05, "loss": 0.0087, "step": 68500 }, { "epoch": 4.753836213566842e-13, "grad_norm": 0.2582520842552185, "learning_rate": 4.6569514821574634e-05, "loss": 0.0087, "step": 68510 }, { "epoch": 4.754530102957233e-13, "grad_norm": 0.26272109150886536, "learning_rate": 4.65511985003509e-05, "loss": 0.0085, "step": 68520 }, { "epoch": 4.755223992347624e-13, "grad_norm": 0.24710695445537567, "learning_rate": 4.6532885873425845e-05, "loss": 0.0086, "step": 68530 }, { "epoch": 4.755917881738014e-13, "grad_norm": 0.27172210812568665, "learning_rate": 4.65145769426814e-05, "loss": 0.0082, "step": 68540 }, { "epoch": 4.756611771128405e-13, "grad_norm": 0.2932058572769165, "learning_rate": 4.649627170999908e-05, "loss": 0.0088, "step": 68550 }, { "epoch": 4.757305660518796e-13, "grad_norm": 0.2778204381465912, "learning_rate": 4.647797017726003e-05, "loss": 0.0089, "step": 68560 }, { "epoch": 4.757999549909186e-13, "grad_norm": 0.24840667843818665, "learning_rate": 4.645967234634505e-05, "loss": 0.0088, "step": 68570 }, { "epoch": 4.758693439299577e-13, "grad_norm": 0.261974036693573, "learning_rate": 4.644137821913446e-05, "loss": 0.0089, "step": 68580 }, { "epoch": 4.759387328689968e-13, "grad_norm": 0.25366753339767456, "learning_rate": 4.6423087797508344e-05, "loss": 0.0085, "step": 68590 }, { "epoch": 4.760081218080359e-13, "grad_norm": 0.24724318087100983, "learning_rate": 4.640480108334624e-05, "loss": 0.0091, "step": 68600 }, { "epoch": 4.760775107470749e-13, "grad_norm": 0.264845609664917, "learning_rate": 4.63865180785275e-05, "loss": 0.009, "step": 68610 }, { "epoch": 4.76146899686114e-13, "grad_norm": 0.23541474342346191, "learning_rate": 4.636823878493087e-05, "loss": 0.0087, "step": 68620 }, { "epoch": 4.762162886251531e-13, "grad_norm": 0.28573358058929443, "learning_rate": 4.634996320443494e-05, "loss": 0.009, "step": 68630 }, { "epoch": 4.762856775641922e-13, "grad_norm": 0.2636360824108124, "learning_rate": 4.633169133891771e-05, "loss": 0.0086, "step": 68640 }, { "epoch": 4.763550665032312e-13, "grad_norm": 0.2580375671386719, "learning_rate": 4.6313423190256996e-05, "loss": 0.0088, "step": 68650 }, { "epoch": 4.764244554422703e-13, "grad_norm": 0.28308677673339844, "learning_rate": 4.6295158760330074e-05, "loss": 0.0084, "step": 68660 }, { "epoch": 4.764938443813094e-13, "grad_norm": 0.24969817698001862, "learning_rate": 4.6276898051013906e-05, "loss": 0.0084, "step": 68670 }, { "epoch": 4.765632333203484e-13, "grad_norm": 0.22549563646316528, "learning_rate": 4.625864106418507e-05, "loss": 0.0085, "step": 68680 }, { "epoch": 4.766326222593875e-13, "grad_norm": 0.25162971019744873, "learning_rate": 4.624038780171975e-05, "loss": 0.0083, "step": 68690 }, { "epoch": 4.767020111984266e-13, "grad_norm": 0.2574358284473419, "learning_rate": 4.6222138265493787e-05, "loss": 0.0086, "step": 68700 }, { "epoch": 4.767714001374657e-13, "grad_norm": 0.2824746370315552, "learning_rate": 4.620389245738253e-05, "loss": 0.0087, "step": 68710 }, { "epoch": 4.768407890765047e-13, "grad_norm": 0.2504040598869324, "learning_rate": 4.618565037926112e-05, "loss": 0.0084, "step": 68720 }, { "epoch": 4.769101780155438e-13, "grad_norm": 0.2932237982749939, "learning_rate": 4.6167412033004104e-05, "loss": 0.0089, "step": 68730 }, { "epoch": 4.769795669545829e-13, "grad_norm": 0.24511966109275818, "learning_rate": 4.6149177420485856e-05, "loss": 0.0085, "step": 68740 }, { "epoch": 4.77048955893622e-13, "grad_norm": 0.3002612888813019, "learning_rate": 4.613094654358017e-05, "loss": 0.0086, "step": 68750 }, { "epoch": 4.77118344832661e-13, "grad_norm": 0.25061798095703125, "learning_rate": 4.611271940416066e-05, "loss": 0.0088, "step": 68760 }, { "epoch": 4.771877337717001e-13, "grad_norm": 0.25415316224098206, "learning_rate": 4.609449600410035e-05, "loss": 0.009, "step": 68770 }, { "epoch": 4.772571227107392e-13, "grad_norm": 0.2622629404067993, "learning_rate": 4.6076276345272055e-05, "loss": 0.0085, "step": 68780 }, { "epoch": 4.773265116497782e-13, "grad_norm": 0.25040245056152344, "learning_rate": 4.605806042954808e-05, "loss": 0.0087, "step": 68790 }, { "epoch": 4.773959005888173e-13, "grad_norm": 0.27253857254981995, "learning_rate": 4.60398482588004e-05, "loss": 0.0088, "step": 68800 }, { "epoch": 4.774652895278564e-13, "grad_norm": 0.2443271279335022, "learning_rate": 4.602163983490061e-05, "loss": 0.0087, "step": 68810 }, { "epoch": 4.775346784668955e-13, "grad_norm": 0.27153974771499634, "learning_rate": 4.600343515971991e-05, "loss": 0.0084, "step": 68820 }, { "epoch": 4.776040674059345e-13, "grad_norm": 0.26017317175865173, "learning_rate": 4.5985234235129103e-05, "loss": 0.0088, "step": 68830 }, { "epoch": 4.776734563449736e-13, "grad_norm": 0.2562647759914398, "learning_rate": 4.596703706299862e-05, "loss": 0.0083, "step": 68840 }, { "epoch": 4.777428452840127e-13, "grad_norm": 0.23639404773712158, "learning_rate": 4.594884364519852e-05, "loss": 0.0083, "step": 68850 }, { "epoch": 4.778122342230517e-13, "grad_norm": 0.26219144463539124, "learning_rate": 4.593065398359843e-05, "loss": 0.0085, "step": 68860 }, { "epoch": 4.778816231620908e-13, "grad_norm": 0.2593964636325836, "learning_rate": 4.591246808006764e-05, "loss": 0.0087, "step": 68870 }, { "epoch": 4.779510121011299e-13, "grad_norm": 0.2284373790025711, "learning_rate": 4.589428593647504e-05, "loss": 0.0082, "step": 68880 }, { "epoch": 4.78020401040169e-13, "grad_norm": 0.23363997042179108, "learning_rate": 4.587610755468912e-05, "loss": 0.0084, "step": 68890 }, { "epoch": 4.78089789979208e-13, "grad_norm": 0.22328175604343414, "learning_rate": 4.585793293657801e-05, "loss": 0.0085, "step": 68900 }, { "epoch": 4.781591789182471e-13, "grad_norm": 0.26659315824508667, "learning_rate": 4.5839762084009375e-05, "loss": 0.0084, "step": 68910 }, { "epoch": 4.782285678572862e-13, "grad_norm": 0.3116467297077179, "learning_rate": 4.582159499885063e-05, "loss": 0.009, "step": 68920 }, { "epoch": 4.782979567963253e-13, "grad_norm": 0.2209859937429428, "learning_rate": 4.5803431682968656e-05, "loss": 0.0089, "step": 68930 }, { "epoch": 4.783673457353643e-13, "grad_norm": 0.23865284025669098, "learning_rate": 4.578527213823009e-05, "loss": 0.0086, "step": 68940 }, { "epoch": 4.784367346744034e-13, "grad_norm": 0.24033771455287933, "learning_rate": 4.5767116366501014e-05, "loss": 0.0086, "step": 68950 }, { "epoch": 4.785061236134425e-13, "grad_norm": 0.2644426226615906, "learning_rate": 4.5748964369647326e-05, "loss": 0.0087, "step": 68960 }, { "epoch": 4.785755125524815e-13, "grad_norm": 0.2539658844470978, "learning_rate": 4.573081614953432e-05, "loss": 0.0084, "step": 68970 }, { "epoch": 4.786449014915206e-13, "grad_norm": 0.2567020058631897, "learning_rate": 4.57126717080271e-05, "loss": 0.0084, "step": 68980 }, { "epoch": 4.787142904305597e-13, "grad_norm": 0.2386203557252884, "learning_rate": 4.5694531046990236e-05, "loss": 0.0082, "step": 68990 }, { "epoch": 4.787836793695988e-13, "grad_norm": 0.26355502009391785, "learning_rate": 4.567639416828797e-05, "loss": 0.0083, "step": 69000 }, { "epoch": 4.788530683086378e-13, "grad_norm": 0.2513032555580139, "learning_rate": 4.565826107378415e-05, "loss": 0.0082, "step": 69010 }, { "epoch": 4.789224572476769e-13, "grad_norm": 0.23341378569602966, "learning_rate": 4.564013176534224e-05, "loss": 0.0083, "step": 69020 }, { "epoch": 4.78991846186716e-13, "grad_norm": 0.22873520851135254, "learning_rate": 4.5622006244825306e-05, "loss": 0.0082, "step": 69030 }, { "epoch": 4.79061235125755e-13, "grad_norm": 0.22370609641075134, "learning_rate": 4.5603884514096024e-05, "loss": 0.008, "step": 69040 }, { "epoch": 4.791306240647941e-13, "grad_norm": 0.27049732208251953, "learning_rate": 4.5585766575016685e-05, "loss": 0.0086, "step": 69050 }, { "epoch": 4.792000130038332e-13, "grad_norm": 0.23705819249153137, "learning_rate": 4.55676524294492e-05, "loss": 0.0081, "step": 69060 }, { "epoch": 4.792694019428723e-13, "grad_norm": 0.26275065541267395, "learning_rate": 4.554954207925505e-05, "loss": 0.0086, "step": 69070 }, { "epoch": 4.793387908819113e-13, "grad_norm": 0.2665783166885376, "learning_rate": 4.5531435526295395e-05, "loss": 0.0087, "step": 69080 }, { "epoch": 4.794081798209504e-13, "grad_norm": 0.2437864989042282, "learning_rate": 4.551333277243094e-05, "loss": 0.0084, "step": 69090 }, { "epoch": 4.794775687599895e-13, "grad_norm": 0.29262351989746094, "learning_rate": 4.549523381952202e-05, "loss": 0.0089, "step": 69100 }, { "epoch": 4.795469576990286e-13, "grad_norm": 0.3060847520828247, "learning_rate": 4.5477138669428606e-05, "loss": 0.0084, "step": 69110 }, { "epoch": 4.796163466380676e-13, "grad_norm": 0.2729071378707886, "learning_rate": 4.545904732401027e-05, "loss": 0.0089, "step": 69120 }, { "epoch": 4.796857355771067e-13, "grad_norm": 0.25399458408355713, "learning_rate": 4.5440959785126096e-05, "loss": 0.0084, "step": 69130 }, { "epoch": 4.797551245161458e-13, "grad_norm": 0.262144535779953, "learning_rate": 4.542287605463497e-05, "loss": 0.0082, "step": 69140 }, { "epoch": 4.798245134551848e-13, "grad_norm": 0.2596310079097748, "learning_rate": 4.540479613439519e-05, "loss": 0.0082, "step": 69150 }, { "epoch": 4.798939023942239e-13, "grad_norm": 0.24251557886600494, "learning_rate": 4.5386720026264796e-05, "loss": 0.0081, "step": 69160 }, { "epoch": 4.79963291333263e-13, "grad_norm": 0.21863685548305511, "learning_rate": 4.536864773210137e-05, "loss": 0.0084, "step": 69170 }, { "epoch": 4.800326802723021e-13, "grad_norm": 0.24055179953575134, "learning_rate": 4.5350579253762135e-05, "loss": 0.0086, "step": 69180 }, { "epoch": 4.801020692113411e-13, "grad_norm": 0.24971961975097656, "learning_rate": 4.533251459310389e-05, "loss": 0.0082, "step": 69190 }, { "epoch": 4.801714581503802e-13, "grad_norm": 0.2659221887588501, "learning_rate": 4.5314453751983083e-05, "loss": 0.0082, "step": 69200 }, { "epoch": 4.802408470894193e-13, "grad_norm": 0.25182273983955383, "learning_rate": 4.5296396732255725e-05, "loss": 0.0085, "step": 69210 }, { "epoch": 4.803102360284583e-13, "grad_norm": 0.25799882411956787, "learning_rate": 4.5278343535777464e-05, "loss": 0.0084, "step": 69220 }, { "epoch": 4.803796249674974e-13, "grad_norm": 0.2767460346221924, "learning_rate": 4.526029416440356e-05, "loss": 0.0087, "step": 69230 }, { "epoch": 4.804490139065365e-13, "grad_norm": 0.27448615431785583, "learning_rate": 4.524224861998885e-05, "loss": 0.0089, "step": 69240 }, { "epoch": 4.805184028455756e-13, "grad_norm": 0.26419520378112793, "learning_rate": 4.522420690438782e-05, "loss": 0.0085, "step": 69250 }, { "epoch": 4.805877917846146e-13, "grad_norm": 0.22561492025852203, "learning_rate": 4.5206169019454466e-05, "loss": 0.0087, "step": 69260 }, { "epoch": 4.806571807236537e-13, "grad_norm": 0.27087441086769104, "learning_rate": 4.5188134967042576e-05, "loss": 0.0091, "step": 69270 }, { "epoch": 4.807265696626928e-13, "grad_norm": 0.2887558937072754, "learning_rate": 4.5170104749005296e-05, "loss": 0.0085, "step": 69280 }, { "epoch": 4.807959586017319e-13, "grad_norm": 0.29892995953559875, "learning_rate": 4.515207836719565e-05, "loss": 0.0082, "step": 69290 }, { "epoch": 4.808653475407709e-13, "grad_norm": 0.2678721249103546, "learning_rate": 4.5134055823466e-05, "loss": 0.0083, "step": 69300 }, { "epoch": 4.8093473647981e-13, "grad_norm": 0.26273682713508606, "learning_rate": 4.511603711966854e-05, "loss": 0.0081, "step": 69310 }, { "epoch": 4.810041254188491e-13, "grad_norm": 0.2670562267303467, "learning_rate": 4.509802225765492e-05, "loss": 0.0084, "step": 69320 }, { "epoch": 4.810735143578881e-13, "grad_norm": 0.3207748532295227, "learning_rate": 4.508001123927647e-05, "loss": 0.0087, "step": 69330 }, { "epoch": 4.811429032969272e-13, "grad_norm": 0.2523707151412964, "learning_rate": 4.506200406638408e-05, "loss": 0.0084, "step": 69340 }, { "epoch": 4.812122922359663e-13, "grad_norm": 0.24142126739025116, "learning_rate": 4.504400074082829e-05, "loss": 0.0083, "step": 69350 }, { "epoch": 4.812816811750054e-13, "grad_norm": 0.2546161413192749, "learning_rate": 4.502600126445921e-05, "loss": 0.0083, "step": 69360 }, { "epoch": 4.813510701140444e-13, "grad_norm": 0.25863415002822876, "learning_rate": 4.500800563912657e-05, "loss": 0.0086, "step": 69370 }, { "epoch": 4.814204590530835e-13, "grad_norm": 0.28291618824005127, "learning_rate": 4.499001386667971e-05, "loss": 0.0085, "step": 69380 }, { "epoch": 4.814898479921226e-13, "grad_norm": 0.2955322563648224, "learning_rate": 4.4972025948967546e-05, "loss": 0.0088, "step": 69390 }, { "epoch": 4.815592369311616e-13, "grad_norm": 0.2462141066789627, "learning_rate": 4.495404188783863e-05, "loss": 0.0087, "step": 69400 }, { "epoch": 4.816286258702007e-13, "grad_norm": 0.26421433687210083, "learning_rate": 4.49360616851411e-05, "loss": 0.0088, "step": 69410 }, { "epoch": 4.816980148092398e-13, "grad_norm": 0.24168068170547485, "learning_rate": 4.491808534272269e-05, "loss": 0.0087, "step": 69420 }, { "epoch": 4.817674037482789e-13, "grad_norm": 0.22570960223674774, "learning_rate": 4.4900112862430766e-05, "loss": 0.0091, "step": 69430 }, { "epoch": 4.818367926873179e-13, "grad_norm": 0.22781679034233093, "learning_rate": 4.4882144246112276e-05, "loss": 0.0087, "step": 69440 }, { "epoch": 4.81906181626357e-13, "grad_norm": 0.25646018981933594, "learning_rate": 4.48641794956138e-05, "loss": 0.0085, "step": 69450 }, { "epoch": 4.819755705653961e-13, "grad_norm": 0.22850123047828674, "learning_rate": 4.484621861278141e-05, "loss": 0.0085, "step": 69460 }, { "epoch": 4.820449595044352e-13, "grad_norm": 0.22684279084205627, "learning_rate": 4.4828261599460984e-05, "loss": 0.0081, "step": 69470 }, { "epoch": 4.821143484434742e-13, "grad_norm": 0.2376236766576767, "learning_rate": 4.481030845749777e-05, "loss": 0.0083, "step": 69480 }, { "epoch": 4.821837373825133e-13, "grad_norm": 0.267703652381897, "learning_rate": 4.479235918873683e-05, "loss": 0.0087, "step": 69490 }, { "epoch": 4.822531263215524e-13, "grad_norm": 0.24122656881809235, "learning_rate": 4.477441379502266e-05, "loss": 0.0084, "step": 69500 }, { "epoch": 4.823225152605914e-13, "grad_norm": 0.265511155128479, "learning_rate": 4.475647227819948e-05, "loss": 0.0082, "step": 69510 }, { "epoch": 4.823919041996305e-13, "grad_norm": 0.2595348358154297, "learning_rate": 4.4738534640111036e-05, "loss": 0.0082, "step": 69520 }, { "epoch": 4.824612931386696e-13, "grad_norm": 0.26313573122024536, "learning_rate": 4.4720600882600694e-05, "loss": 0.0087, "step": 69530 }, { "epoch": 4.825306820777087e-13, "grad_norm": 0.24702607095241547, "learning_rate": 4.470267100751142e-05, "loss": 0.0082, "step": 69540 }, { "epoch": 4.826000710167477e-13, "grad_norm": 0.26340094208717346, "learning_rate": 4.468474501668581e-05, "loss": 0.0087, "step": 69550 }, { "epoch": 4.826694599557868e-13, "grad_norm": 0.25945332646369934, "learning_rate": 4.466682291196603e-05, "loss": 0.0084, "step": 69560 }, { "epoch": 4.827388488948259e-13, "grad_norm": 0.23761974275112152, "learning_rate": 4.464890469519385e-05, "loss": 0.0085, "step": 69570 }, { "epoch": 4.82808237833865e-13, "grad_norm": 0.24557678401470184, "learning_rate": 4.463099036821065e-05, "loss": 0.0086, "step": 69580 }, { "epoch": 4.82877626772904e-13, "grad_norm": 0.3134388327598572, "learning_rate": 4.461307993285741e-05, "loss": 0.0087, "step": 69590 }, { "epoch": 4.829470157119431e-13, "grad_norm": 0.27603888511657715, "learning_rate": 4.4595173390974703e-05, "loss": 0.0085, "step": 69600 }, { "epoch": 4.830164046509822e-13, "grad_norm": 0.2714596390724182, "learning_rate": 4.457727074440269e-05, "loss": 0.0088, "step": 69610 }, { "epoch": 4.830857935900212e-13, "grad_norm": 0.25731900334358215, "learning_rate": 4.45593719949812e-05, "loss": 0.0081, "step": 69620 }, { "epoch": 4.831551825290603e-13, "grad_norm": 0.23936626315116882, "learning_rate": 4.454147714454952e-05, "loss": 0.0078, "step": 69630 }, { "epoch": 4.832245714680994e-13, "grad_norm": 0.23130136728286743, "learning_rate": 4.452358619494671e-05, "loss": 0.0083, "step": 69640 }, { "epoch": 4.832939604071385e-13, "grad_norm": 0.2625180780887604, "learning_rate": 4.4505699148011276e-05, "loss": 0.0088, "step": 69650 }, { "epoch": 4.833633493461775e-13, "grad_norm": 0.22242343425750732, "learning_rate": 4.448781600558143e-05, "loss": 0.0083, "step": 69660 }, { "epoch": 4.834327382852166e-13, "grad_norm": 0.22950845956802368, "learning_rate": 4.446993676949493e-05, "loss": 0.0081, "step": 69670 }, { "epoch": 4.835021272242557e-13, "grad_norm": 0.28298360109329224, "learning_rate": 4.445206144158916e-05, "loss": 0.0082, "step": 69680 }, { "epoch": 4.835715161632947e-13, "grad_norm": 0.2339596152305603, "learning_rate": 4.443419002370107e-05, "loss": 0.0084, "step": 69690 }, { "epoch": 4.836409051023338e-13, "grad_norm": 0.20773787796497345, "learning_rate": 4.441632251766722e-05, "loss": 0.0081, "step": 69700 }, { "epoch": 4.837102940413729e-13, "grad_norm": 0.23787568509578705, "learning_rate": 4.43984589253238e-05, "loss": 0.0088, "step": 69710 }, { "epoch": 4.83779682980412e-13, "grad_norm": 0.2624454200267792, "learning_rate": 4.438059924850655e-05, "loss": 0.0082, "step": 69720 }, { "epoch": 4.83849071919451e-13, "grad_norm": 0.27552109956741333, "learning_rate": 4.436274348905084e-05, "loss": 0.0084, "step": 69730 }, { "epoch": 4.839184608584901e-13, "grad_norm": 0.24775007367134094, "learning_rate": 4.434489164879163e-05, "loss": 0.0088, "step": 69740 }, { "epoch": 4.839878497975292e-13, "grad_norm": 0.26603639125823975, "learning_rate": 4.432704372956348e-05, "loss": 0.0082, "step": 69750 }, { "epoch": 4.840572387365683e-13, "grad_norm": 0.2905757427215576, "learning_rate": 4.430919973320052e-05, "loss": 0.0084, "step": 69760 }, { "epoch": 4.841266276756073e-13, "grad_norm": 0.2515723407268524, "learning_rate": 4.4291359661536515e-05, "loss": 0.0085, "step": 69770 }, { "epoch": 4.841960166146464e-13, "grad_norm": 0.25669392943382263, "learning_rate": 4.427352351640483e-05, "loss": 0.0082, "step": 69780 }, { "epoch": 4.842654055536855e-13, "grad_norm": 0.25629687309265137, "learning_rate": 4.425569129963832e-05, "loss": 0.008, "step": 69790 }, { "epoch": 4.843347944927245e-13, "grad_norm": 0.2476060688495636, "learning_rate": 4.423786301306965e-05, "loss": 0.0096, "step": 69800 }, { "epoch": 4.844041834317636e-13, "grad_norm": 0.24697183072566986, "learning_rate": 4.422003865853083e-05, "loss": 0.0083, "step": 69810 }, { "epoch": 4.844735723708027e-13, "grad_norm": 0.2399291843175888, "learning_rate": 4.420221823785369e-05, "loss": 0.0082, "step": 69820 }, { "epoch": 4.845429613098418e-13, "grad_norm": 0.22545184195041656, "learning_rate": 4.4184401752869487e-05, "loss": 0.008, "step": 69830 }, { "epoch": 4.846123502488808e-13, "grad_norm": 0.25184279680252075, "learning_rate": 4.41665892054092e-05, "loss": 0.0083, "step": 69840 }, { "epoch": 4.846817391879199e-13, "grad_norm": 0.23015651106834412, "learning_rate": 4.4148780597303265e-05, "loss": 0.0084, "step": 69850 }, { "epoch": 4.84751128126959e-13, "grad_norm": 0.2296769767999649, "learning_rate": 4.413097593038189e-05, "loss": 0.0083, "step": 69860 }, { "epoch": 4.84820517065998e-13, "grad_norm": 0.2365695685148239, "learning_rate": 4.411317520647472e-05, "loss": 0.0083, "step": 69870 }, { "epoch": 4.848899060050371e-13, "grad_norm": 0.2503095865249634, "learning_rate": 4.409537842741106e-05, "loss": 0.0086, "step": 69880 }, { "epoch": 4.849592949440762e-13, "grad_norm": 0.25083038210868835, "learning_rate": 4.407758559501983e-05, "loss": 0.0084, "step": 69890 }, { "epoch": 4.850286838831153e-13, "grad_norm": 0.257229208946228, "learning_rate": 4.405979671112948e-05, "loss": 0.0089, "step": 69900 }, { "epoch": 4.850980728221543e-13, "grad_norm": 0.2686103880405426, "learning_rate": 4.404201177756815e-05, "loss": 0.0081, "step": 69910 }, { "epoch": 4.851674617611934e-13, "grad_norm": 0.30065983533859253, "learning_rate": 4.4024230796163475e-05, "loss": 0.0091, "step": 69920 }, { "epoch": 4.852368507002325e-13, "grad_norm": 0.22570958733558655, "learning_rate": 4.400645376874275e-05, "loss": 0.0087, "step": 69930 }, { "epoch": 4.853062396392716e-13, "grad_norm": 0.27094706892967224, "learning_rate": 4.398868069713283e-05, "loss": 0.0082, "step": 69940 }, { "epoch": 4.853756285783106e-13, "grad_norm": 0.24811981618404388, "learning_rate": 4.3970911583160174e-05, "loss": 0.0088, "step": 69950 }, { "epoch": 4.854450175173497e-13, "grad_norm": 0.2844648063182831, "learning_rate": 4.3953146428650845e-05, "loss": 0.0096, "step": 69960 }, { "epoch": 4.855144064563888e-13, "grad_norm": 0.2598603069782257, "learning_rate": 4.3935385235430475e-05, "loss": 0.0083, "step": 69970 }, { "epoch": 4.855837953954278e-13, "grad_norm": 0.2716936767101288, "learning_rate": 4.3917628005324345e-05, "loss": 0.0082, "step": 69980 }, { "epoch": 4.856531843344669e-13, "grad_norm": 0.24372248351573944, "learning_rate": 4.3899874740157195e-05, "loss": 0.0081, "step": 69990 }, { "epoch": 4.85722573273506e-13, "grad_norm": 0.25736647844314575, "learning_rate": 4.388212544175356e-05, "loss": 0.0082, "step": 70000 }, { "epoch": 4.857919622125451e-13, "grad_norm": 0.2491678148508072, "learning_rate": 4.386438011193736e-05, "loss": 0.0083, "step": 70010 }, { "epoch": 4.858613511515841e-13, "grad_norm": 0.24495451152324677, "learning_rate": 4.3846638752532285e-05, "loss": 0.0086, "step": 70020 }, { "epoch": 4.859307400906232e-13, "grad_norm": 0.2605110704898834, "learning_rate": 4.3828901365361454e-05, "loss": 0.0081, "step": 70030 }, { "epoch": 4.860001290296623e-13, "grad_norm": 0.2519713342189789, "learning_rate": 4.381116795224775e-05, "loss": 0.0088, "step": 70040 }, { "epoch": 4.860695179687013e-13, "grad_norm": 0.2277379333972931, "learning_rate": 4.379343851501347e-05, "loss": 0.0085, "step": 70050 }, { "epoch": 4.861389069077404e-13, "grad_norm": 0.2128090262413025, "learning_rate": 4.3775713055480676e-05, "loss": 0.0082, "step": 70060 }, { "epoch": 4.862082958467795e-13, "grad_norm": 0.2244141399860382, "learning_rate": 4.375799157547086e-05, "loss": 0.0081, "step": 70070 }, { "epoch": 4.862776847858186e-13, "grad_norm": 0.2632037401199341, "learning_rate": 4.3740274076805215e-05, "loss": 0.0093, "step": 70080 }, { "epoch": 4.863470737248576e-13, "grad_norm": 0.2677481472492218, "learning_rate": 4.372256056130448e-05, "loss": 0.0085, "step": 70090 }, { "epoch": 4.864164626638967e-13, "grad_norm": 0.23299247026443481, "learning_rate": 4.3704851030789e-05, "loss": 0.0082, "step": 70100 }, { "epoch": 4.864858516029358e-13, "grad_norm": 0.2533228099346161, "learning_rate": 4.3687145487078714e-05, "loss": 0.0086, "step": 70110 }, { "epoch": 4.865552405419749e-13, "grad_norm": 0.2335035353899002, "learning_rate": 4.36694439319931e-05, "loss": 0.0096, "step": 70120 }, { "epoch": 4.866246294810139e-13, "grad_norm": 0.2355967015028, "learning_rate": 4.365174636735134e-05, "loss": 0.008, "step": 70130 }, { "epoch": 4.86694018420053e-13, "grad_norm": 0.2067989706993103, "learning_rate": 4.3634052794972044e-05, "loss": 0.008, "step": 70140 }, { "epoch": 4.867634073590921e-13, "grad_norm": 0.2465122938156128, "learning_rate": 4.36163632166736e-05, "loss": 0.008, "step": 70150 }, { "epoch": 4.868327962981311e-13, "grad_norm": 0.23916824162006378, "learning_rate": 4.3598677634273785e-05, "loss": 0.0083, "step": 70160 }, { "epoch": 4.869021852371702e-13, "grad_norm": 0.2352510392665863, "learning_rate": 4.3580996049590165e-05, "loss": 0.0081, "step": 70170 }, { "epoch": 4.869715741762093e-13, "grad_norm": 0.2403838187456131, "learning_rate": 4.356331846443972e-05, "loss": 0.0082, "step": 70180 }, { "epoch": 4.870409631152484e-13, "grad_norm": 0.22717738151550293, "learning_rate": 4.3545644880639175e-05, "loss": 0.0083, "step": 70190 }, { "epoch": 4.871103520542874e-13, "grad_norm": 0.24590347707271576, "learning_rate": 4.35279753000047e-05, "loss": 0.0082, "step": 70200 }, { "epoch": 4.871797409933265e-13, "grad_norm": 0.2482406497001648, "learning_rate": 4.351030972435213e-05, "loss": 0.0085, "step": 70210 }, { "epoch": 4.872491299323656e-13, "grad_norm": 0.2726372182369232, "learning_rate": 4.349264815549691e-05, "loss": 0.008, "step": 70220 }, { "epoch": 4.873185188714046e-13, "grad_norm": 0.22625721991062164, "learning_rate": 4.347499059525399e-05, "loss": 0.0084, "step": 70230 }, { "epoch": 4.873879078104437e-13, "grad_norm": 0.25436079502105713, "learning_rate": 4.3457337045438015e-05, "loss": 0.0093, "step": 70240 }, { "epoch": 4.874572967494828e-13, "grad_norm": 0.22670233249664307, "learning_rate": 4.343968750786311e-05, "loss": 0.0083, "step": 70250 }, { "epoch": 4.875266856885219e-13, "grad_norm": 0.2656444311141968, "learning_rate": 4.342204198434309e-05, "loss": 0.0083, "step": 70260 }, { "epoch": 4.875960746275609e-13, "grad_norm": 0.2480727732181549, "learning_rate": 4.3404400476691265e-05, "loss": 0.0088, "step": 70270 }, { "epoch": 4.876654635666e-13, "grad_norm": 0.2250194102525711, "learning_rate": 4.33867629867206e-05, "loss": 0.0082, "step": 70280 }, { "epoch": 4.877348525056391e-13, "grad_norm": 0.2429918348789215, "learning_rate": 4.3369129516243614e-05, "loss": 0.0079, "step": 70290 }, { "epoch": 4.878042414446782e-13, "grad_norm": 0.28200584650039673, "learning_rate": 4.335150006707242e-05, "loss": 0.0079, "step": 70300 }, { "epoch": 4.878736303837172e-13, "grad_norm": 0.2644076943397522, "learning_rate": 4.3333874641018736e-05, "loss": 0.0081, "step": 70310 }, { "epoch": 4.879430193227563e-13, "grad_norm": 0.24726314842700958, "learning_rate": 4.331625323989379e-05, "loss": 0.008, "step": 70320 }, { "epoch": 4.880124082617954e-13, "grad_norm": 0.26796138286590576, "learning_rate": 4.329863586550854e-05, "loss": 0.0084, "step": 70330 }, { "epoch": 4.880817972008344e-13, "grad_norm": 0.2330552190542221, "learning_rate": 4.328102251967334e-05, "loss": 0.0079, "step": 70340 }, { "epoch": 4.881511861398735e-13, "grad_norm": 0.25059598684310913, "learning_rate": 4.326341320419837e-05, "loss": 0.0086, "step": 70350 }, { "epoch": 4.882205750789126e-13, "grad_norm": 0.27360108494758606, "learning_rate": 4.324580792089313e-05, "loss": 0.0083, "step": 70360 }, { "epoch": 4.882899640179517e-13, "grad_norm": 0.2431475669145584, "learning_rate": 4.322820667156693e-05, "loss": 0.0084, "step": 70370 }, { "epoch": 4.883593529569907e-13, "grad_norm": 0.21898002922534943, "learning_rate": 4.32106094580285e-05, "loss": 0.0084, "step": 70380 }, { "epoch": 4.884287418960298e-13, "grad_norm": 0.25118428468704224, "learning_rate": 4.319301628208632e-05, "loss": 0.0085, "step": 70390 }, { "epoch": 4.884981308350689e-13, "grad_norm": 0.2567228674888611, "learning_rate": 4.317542714554826e-05, "loss": 0.0084, "step": 70400 }, { "epoch": 4.88567519774108e-13, "grad_norm": 0.24342377483844757, "learning_rate": 4.315784205022194e-05, "loss": 0.008, "step": 70410 }, { "epoch": 4.88636908713147e-13, "grad_norm": 0.20788881182670593, "learning_rate": 4.314026099791449e-05, "loss": 0.0081, "step": 70420 }, { "epoch": 4.887062976521861e-13, "grad_norm": 0.2521955668926239, "learning_rate": 4.312268399043262e-05, "loss": 0.0089, "step": 70430 }, { "epoch": 4.887756865912252e-13, "grad_norm": 0.2523502707481384, "learning_rate": 4.3105111029582665e-05, "loss": 0.0082, "step": 70440 }, { "epoch": 4.888450755302642e-13, "grad_norm": 0.22030295431613922, "learning_rate": 4.308754211717051e-05, "loss": 0.0079, "step": 70450 }, { "epoch": 4.889144644693033e-13, "grad_norm": 0.23468776047229767, "learning_rate": 4.306997725500163e-05, "loss": 0.0077, "step": 70460 }, { "epoch": 4.889838534083424e-13, "grad_norm": 0.22765499353408813, "learning_rate": 4.3052416444881076e-05, "loss": 0.0081, "step": 70470 }, { "epoch": 4.890532423473815e-13, "grad_norm": 0.2410126030445099, "learning_rate": 4.3034859688613525e-05, "loss": 0.0081, "step": 70480 }, { "epoch": 4.891226312864205e-13, "grad_norm": 0.23816490173339844, "learning_rate": 4.301730698800319e-05, "loss": 0.008, "step": 70490 }, { "epoch": 4.891920202254596e-13, "grad_norm": 0.24748365581035614, "learning_rate": 4.2999758344853885e-05, "loss": 0.0079, "step": 70500 }, { "epoch": 4.892614091644987e-13, "grad_norm": 0.2163534015417099, "learning_rate": 4.298221376096901e-05, "loss": 0.0084, "step": 70510 }, { "epoch": 4.893307981035377e-13, "grad_norm": 0.24019113183021545, "learning_rate": 4.2964673238151565e-05, "loss": 0.0084, "step": 70520 }, { "epoch": 4.894001870425768e-13, "grad_norm": 0.2616007924079895, "learning_rate": 4.294713677820405e-05, "loss": 0.0084, "step": 70530 }, { "epoch": 4.894695759816159e-13, "grad_norm": 0.24271827936172485, "learning_rate": 4.292960438292867e-05, "loss": 0.008, "step": 70540 }, { "epoch": 4.89538964920655e-13, "grad_norm": 0.2264832854270935, "learning_rate": 4.2912076054127106e-05, "loss": 0.0083, "step": 70550 }, { "epoch": 4.89608353859694e-13, "grad_norm": 0.25215837359428406, "learning_rate": 4.2894551793600706e-05, "loss": 0.008, "step": 70560 }, { "epoch": 4.896777427987331e-13, "grad_norm": 0.2532516121864319, "learning_rate": 4.287703160315033e-05, "loss": 0.008, "step": 70570 }, { "epoch": 4.897471317377722e-13, "grad_norm": 0.3155503273010254, "learning_rate": 4.2859515484576476e-05, "loss": 0.0084, "step": 70580 }, { "epoch": 4.898165206768113e-13, "grad_norm": 0.2111993134021759, "learning_rate": 4.284200343967916e-05, "loss": 0.0084, "step": 70590 }, { "epoch": 4.898859096158503e-13, "grad_norm": 0.2390824556350708, "learning_rate": 4.282449547025807e-05, "loss": 0.0079, "step": 70600 }, { "epoch": 4.899552985548894e-13, "grad_norm": 0.2125551700592041, "learning_rate": 4.280699157811237e-05, "loss": 0.0078, "step": 70610 }, { "epoch": 4.900246874939285e-13, "grad_norm": 0.23293320834636688, "learning_rate": 4.2789491765040906e-05, "loss": 0.0076, "step": 70620 }, { "epoch": 4.900940764329675e-13, "grad_norm": 0.21729165315628052, "learning_rate": 4.277199603284201e-05, "loss": 0.0081, "step": 70630 }, { "epoch": 4.901634653720066e-13, "grad_norm": 0.2237137109041214, "learning_rate": 4.275450438331369e-05, "loss": 0.008, "step": 70640 }, { "epoch": 4.902328543110457e-13, "grad_norm": 0.231721892952919, "learning_rate": 4.273701681825342e-05, "loss": 0.008, "step": 70650 }, { "epoch": 4.903022432500848e-13, "grad_norm": 0.3112575113773346, "learning_rate": 4.2719533339458395e-05, "loss": 0.0083, "step": 70660 }, { "epoch": 4.903716321891238e-13, "grad_norm": 0.23451998829841614, "learning_rate": 4.270205394872524e-05, "loss": 0.0082, "step": 70670 }, { "epoch": 4.904410211281629e-13, "grad_norm": 0.2060495913028717, "learning_rate": 4.2684578647850305e-05, "loss": 0.0086, "step": 70680 }, { "epoch": 4.90510410067202e-13, "grad_norm": 0.2079218178987503, "learning_rate": 4.266710743862937e-05, "loss": 0.0085, "step": 70690 }, { "epoch": 4.90579799006241e-13, "grad_norm": 0.23385024070739746, "learning_rate": 4.2649640322857973e-05, "loss": 0.008, "step": 70700 }, { "epoch": 4.906491879452801e-13, "grad_norm": 0.18985074758529663, "learning_rate": 4.263217730233103e-05, "loss": 0.0079, "step": 70710 }, { "epoch": 4.907185768843192e-13, "grad_norm": 0.21749158203601837, "learning_rate": 4.2614718378843224e-05, "loss": 0.0076, "step": 70720 }, { "epoch": 4.907879658233583e-13, "grad_norm": 0.20493321120738983, "learning_rate": 4.259726355418867e-05, "loss": 0.0076, "step": 70730 }, { "epoch": 4.908573547623973e-13, "grad_norm": 0.30771398544311523, "learning_rate": 4.257981283016114e-05, "loss": 0.0084, "step": 70740 }, { "epoch": 4.909267437014364e-13, "grad_norm": 0.2052977830171585, "learning_rate": 4.2562366208553984e-05, "loss": 0.0078, "step": 70750 }, { "epoch": 4.909961326404755e-13, "grad_norm": 0.21750205755233765, "learning_rate": 4.254492369116009e-05, "loss": 0.0076, "step": 70760 }, { "epoch": 4.910655215795146e-13, "grad_norm": 0.2203167974948883, "learning_rate": 4.2527485279771975e-05, "loss": 0.0075, "step": 70770 }, { "epoch": 4.911349105185536e-13, "grad_norm": 0.21375979483127594, "learning_rate": 4.251005097618169e-05, "loss": 0.0079, "step": 70780 }, { "epoch": 4.912042994575927e-13, "grad_norm": 0.22749704122543335, "learning_rate": 4.249262078218087e-05, "loss": 0.0081, "step": 70790 }, { "epoch": 4.912736883966318e-13, "grad_norm": 0.21277563273906708, "learning_rate": 4.247519469956076e-05, "loss": 0.0081, "step": 70800 }, { "epoch": 4.913430773356708e-13, "grad_norm": 0.2216232717037201, "learning_rate": 4.245777273011216e-05, "loss": 0.0079, "step": 70810 }, { "epoch": 4.914124662747099e-13, "grad_norm": 0.2392214685678482, "learning_rate": 4.244035487562544e-05, "loss": 0.009, "step": 70820 }, { "epoch": 4.91481855213749e-13, "grad_norm": 0.29728400707244873, "learning_rate": 4.2422941137890535e-05, "loss": 0.0085, "step": 70830 }, { "epoch": 4.915512441527881e-13, "grad_norm": 0.24193204939365387, "learning_rate": 4.240553151869703e-05, "loss": 0.0081, "step": 70840 }, { "epoch": 4.916206330918271e-13, "grad_norm": 0.23074690997600555, "learning_rate": 4.238812601983399e-05, "loss": 0.0081, "step": 70850 }, { "epoch": 4.916900220308662e-13, "grad_norm": 0.24430373311042786, "learning_rate": 4.2370724643090144e-05, "loss": 0.0078, "step": 70860 }, { "epoch": 4.917594109699053e-13, "grad_norm": 0.23891285061836243, "learning_rate": 4.235332739025367e-05, "loss": 0.0076, "step": 70870 }, { "epoch": 4.918287999089443e-13, "grad_norm": 0.23765131831169128, "learning_rate": 4.2335934263112505e-05, "loss": 0.0077, "step": 70880 }, { "epoch": 4.918981888479834e-13, "grad_norm": 0.22562864422798157, "learning_rate": 4.231854526345398e-05, "loss": 0.0075, "step": 70890 }, { "epoch": 4.919675777870225e-13, "grad_norm": 0.2421031892299652, "learning_rate": 4.230116039306517e-05, "loss": 0.0086, "step": 70900 }, { "epoch": 4.920369667260616e-13, "grad_norm": 0.2246357798576355, "learning_rate": 4.228377965373255e-05, "loss": 0.0076, "step": 70910 }, { "epoch": 4.921063556651006e-13, "grad_norm": 0.2062016725540161, "learning_rate": 4.226640304724235e-05, "loss": 0.0077, "step": 70920 }, { "epoch": 4.921757446041397e-13, "grad_norm": 0.23189491033554077, "learning_rate": 4.224903057538023e-05, "loss": 0.0075, "step": 70930 }, { "epoch": 4.922451335431788e-13, "grad_norm": 0.22467412054538727, "learning_rate": 4.223166223993149e-05, "loss": 0.0086, "step": 70940 }, { "epoch": 4.923145224822179e-13, "grad_norm": 0.22244375944137573, "learning_rate": 4.2214298042681e-05, "loss": 0.0085, "step": 70950 }, { "epoch": 4.923839114212569e-13, "grad_norm": 0.3736104965209961, "learning_rate": 4.219693798541321e-05, "loss": 0.0084, "step": 70960 }, { "epoch": 4.92453300360296e-13, "grad_norm": 0.31775039434432983, "learning_rate": 4.217958206991216e-05, "loss": 0.0082, "step": 70970 }, { "epoch": 4.925226892993351e-13, "grad_norm": 0.2708466947078705, "learning_rate": 4.2162230297961345e-05, "loss": 0.0082, "step": 70980 }, { "epoch": 4.925920782383741e-13, "grad_norm": 0.2545161545276642, "learning_rate": 4.214488267134406e-05, "loss": 0.0085, "step": 70990 }, { "epoch": 4.926614671774132e-13, "grad_norm": 0.24470287561416626, "learning_rate": 4.2127539191842926e-05, "loss": 0.0081, "step": 71000 }, { "epoch": 4.927308561164523e-13, "grad_norm": 0.2282605916261673, "learning_rate": 4.2110199861240345e-05, "loss": 0.0079, "step": 71010 }, { "epoch": 4.928002450554914e-13, "grad_norm": 0.21101051568984985, "learning_rate": 4.209286468131812e-05, "loss": 0.0076, "step": 71020 }, { "epoch": 4.928696339945304e-13, "grad_norm": 0.25169637799263, "learning_rate": 4.207553365385781e-05, "loss": 0.0076, "step": 71030 }, { "epoch": 4.929390229335695e-13, "grad_norm": 0.22246432304382324, "learning_rate": 4.2058206780640356e-05, "loss": 0.0073, "step": 71040 }, { "epoch": 4.930084118726086e-13, "grad_norm": 0.21538594365119934, "learning_rate": 4.204088406344643e-05, "loss": 0.0077, "step": 71050 }, { "epoch": 4.930778008116476e-13, "grad_norm": 0.20688055455684662, "learning_rate": 4.202356550405617e-05, "loss": 0.008, "step": 71060 }, { "epoch": 4.931471897506867e-13, "grad_norm": 0.24117010831832886, "learning_rate": 4.2006251104249336e-05, "loss": 0.0076, "step": 71070 }, { "epoch": 4.932165786897258e-13, "grad_norm": 0.26559939980506897, "learning_rate": 4.198894086580526e-05, "loss": 0.0075, "step": 71080 }, { "epoch": 4.932859676287649e-13, "grad_norm": 0.22787141799926758, "learning_rate": 4.197163479050284e-05, "loss": 0.0077, "step": 71090 }, { "epoch": 4.933553565678039e-13, "grad_norm": 0.22502397000789642, "learning_rate": 4.195433288012053e-05, "loss": 0.0079, "step": 71100 }, { "epoch": 4.93424745506843e-13, "grad_norm": 0.2116807997226715, "learning_rate": 4.193703513643637e-05, "loss": 0.0084, "step": 71110 }, { "epoch": 4.934941344458821e-13, "grad_norm": 0.23429566621780396, "learning_rate": 4.1919741561227995e-05, "loss": 0.0083, "step": 71120 }, { "epoch": 4.935635233849212e-13, "grad_norm": 0.2811720073223114, "learning_rate": 4.190245215627258e-05, "loss": 0.0083, "step": 71130 }, { "epoch": 4.936329123239602e-13, "grad_norm": 0.2888329029083252, "learning_rate": 4.188516692334686e-05, "loss": 0.0082, "step": 71140 }, { "epoch": 4.937023012629993e-13, "grad_norm": 0.2590448260307312, "learning_rate": 4.1867885864227186e-05, "loss": 0.0084, "step": 71150 }, { "epoch": 4.937716902020384e-13, "grad_norm": 0.24853849411010742, "learning_rate": 4.185060898068945e-05, "loss": 0.0079, "step": 71160 }, { "epoch": 4.938410791410774e-13, "grad_norm": 0.2015428990125656, "learning_rate": 4.183333627450912e-05, "loss": 0.0076, "step": 71170 }, { "epoch": 4.939104680801165e-13, "grad_norm": 0.18929561972618103, "learning_rate": 4.1816067747461227e-05, "loss": 0.0076, "step": 71180 }, { "epoch": 4.939798570191556e-13, "grad_norm": 0.3449893593788147, "learning_rate": 4.1798803401320414e-05, "loss": 0.0077, "step": 71190 }, { "epoch": 4.940492459581947e-13, "grad_norm": 0.2332705706357956, "learning_rate": 4.178154323786079e-05, "loss": 0.0073, "step": 71200 }, { "epoch": 4.941186348972337e-13, "grad_norm": 0.2766117453575134, "learning_rate": 4.176428725885619e-05, "loss": 0.0079, "step": 71210 }, { "epoch": 4.941880238362728e-13, "grad_norm": 0.2496979981660843, "learning_rate": 4.1747035466079846e-05, "loss": 0.0077, "step": 71220 }, { "epoch": 4.942574127753119e-13, "grad_norm": 0.24256868660449982, "learning_rate": 4.1729787861304754e-05, "loss": 0.0076, "step": 71230 }, { "epoch": 4.94326801714351e-13, "grad_norm": 0.22520747780799866, "learning_rate": 4.171254444630325e-05, "loss": 0.0077, "step": 71240 }, { "epoch": 4.9439619065339e-13, "grad_norm": 0.24872912466526031, "learning_rate": 4.16953052228475e-05, "loss": 0.0081, "step": 71250 }, { "epoch": 4.944655795924291e-13, "grad_norm": 0.2782672941684723, "learning_rate": 4.167807019270898e-05, "loss": 0.0082, "step": 71260 }, { "epoch": 4.945349685314682e-13, "grad_norm": 0.2502304017543793, "learning_rate": 4.166083935765893e-05, "loss": 0.008, "step": 71270 }, { "epoch": 4.946043574705072e-13, "grad_norm": 0.21192315220832825, "learning_rate": 4.164361271946805e-05, "loss": 0.0077, "step": 71280 }, { "epoch": 4.946737464095463e-13, "grad_norm": 0.22027504444122314, "learning_rate": 4.162639027990668e-05, "loss": 0.0081, "step": 71290 }, { "epoch": 4.947431353485854e-13, "grad_norm": 0.2203008532524109, "learning_rate": 4.160917204074466e-05, "loss": 0.0084, "step": 71300 }, { "epoch": 4.948125242876245e-13, "grad_norm": 0.2152726948261261, "learning_rate": 4.1591958003751444e-05, "loss": 0.0078, "step": 71310 }, { "epoch": 4.948819132266635e-13, "grad_norm": 0.2517750561237335, "learning_rate": 4.1574748170696056e-05, "loss": 0.0077, "step": 71320 }, { "epoch": 4.949513021657026e-13, "grad_norm": 0.24026042222976685, "learning_rate": 4.155754254334706e-05, "loss": 0.0075, "step": 71330 }, { "epoch": 4.950206911047417e-13, "grad_norm": 0.21505919098854065, "learning_rate": 4.15403411234726e-05, "loss": 0.0073, "step": 71340 }, { "epoch": 4.950900800437807e-13, "grad_norm": 0.20882020890712738, "learning_rate": 4.1523143912840394e-05, "loss": 0.0076, "step": 71350 }, { "epoch": 4.951594689828198e-13, "grad_norm": 0.2864571809768677, "learning_rate": 4.1505950913217725e-05, "loss": 0.0074, "step": 71360 }, { "epoch": 4.952288579218589e-13, "grad_norm": 0.23261211812496185, "learning_rate": 4.148876212637143e-05, "loss": 0.0082, "step": 71370 }, { "epoch": 4.95298246860898e-13, "grad_norm": 0.21237047016620636, "learning_rate": 4.1471577554067935e-05, "loss": 0.0075, "step": 71380 }, { "epoch": 4.95367635799937e-13, "grad_norm": 0.1878514140844345, "learning_rate": 4.145439719807326e-05, "loss": 0.0075, "step": 71390 }, { "epoch": 4.954370247389761e-13, "grad_norm": 0.26561427116394043, "learning_rate": 4.143722106015286e-05, "loss": 0.0082, "step": 71400 }, { "epoch": 4.955064136780152e-13, "grad_norm": 0.23229382932186127, "learning_rate": 4.142004914207194e-05, "loss": 0.0082, "step": 71410 }, { "epoch": 4.955758026170543e-13, "grad_norm": 0.24562817811965942, "learning_rate": 4.1402881445595144e-05, "loss": 0.0083, "step": 71420 }, { "epoch": 4.956451915560933e-13, "grad_norm": 0.20108111202716827, "learning_rate": 4.13857179724867e-05, "loss": 0.0076, "step": 71430 }, { "epoch": 4.957145804951324e-13, "grad_norm": 0.20020397007465363, "learning_rate": 4.136855872451046e-05, "loss": 0.0075, "step": 71440 }, { "epoch": 4.957839694341715e-13, "grad_norm": 0.19649259746074677, "learning_rate": 4.135140370342978e-05, "loss": 0.008, "step": 71450 }, { "epoch": 4.958533583732105e-13, "grad_norm": 0.2220538854598999, "learning_rate": 4.133425291100762e-05, "loss": 0.0087, "step": 71460 }, { "epoch": 4.959227473122496e-13, "grad_norm": 0.2073451280593872, "learning_rate": 4.1317106349006476e-05, "loss": 0.0078, "step": 71470 }, { "epoch": 4.959921362512887e-13, "grad_norm": 0.21814747154712677, "learning_rate": 4.129996401918843e-05, "loss": 0.0077, "step": 71480 }, { "epoch": 4.960615251903278e-13, "grad_norm": 0.23360660672187805, "learning_rate": 4.128282592331514e-05, "loss": 0.0074, "step": 71490 }, { "epoch": 4.961309141293668e-13, "grad_norm": 0.2051049917936325, "learning_rate": 4.126569206314779e-05, "loss": 0.0075, "step": 71500 }, { "epoch": 4.962003030684059e-13, "grad_norm": 0.22998195886611938, "learning_rate": 4.124856244044715e-05, "loss": 0.0079, "step": 71510 }, { "epoch": 4.96269692007445e-13, "grad_norm": 0.20054033398628235, "learning_rate": 4.1231437056973605e-05, "loss": 0.0074, "step": 71520 }, { "epoch": 4.96339080946484e-13, "grad_norm": 0.2073090672492981, "learning_rate": 4.121431591448694e-05, "loss": 0.0078, "step": 71530 }, { "epoch": 4.964084698855231e-13, "grad_norm": 0.21945928037166595, "learning_rate": 4.119719901474675e-05, "loss": 0.0074, "step": 71540 }, { "epoch": 4.964778588245622e-13, "grad_norm": 0.24383482336997986, "learning_rate": 4.118008635951197e-05, "loss": 0.0076, "step": 71550 }, { "epoch": 4.965472477636013e-13, "grad_norm": 0.2641074061393738, "learning_rate": 4.116297795054127e-05, "loss": 0.0078, "step": 71560 }, { "epoch": 4.966166367026403e-13, "grad_norm": 0.1933012157678604, "learning_rate": 4.1145873789592704e-05, "loss": 0.0086, "step": 71570 }, { "epoch": 4.966860256416794e-13, "grad_norm": 0.2690528631210327, "learning_rate": 4.11287738784241e-05, "loss": 0.0082, "step": 71580 }, { "epoch": 4.967554145807185e-13, "grad_norm": 0.21927976608276367, "learning_rate": 4.1111678218792664e-05, "loss": 0.0076, "step": 71590 }, { "epoch": 4.968248035197576e-13, "grad_norm": 0.2738482654094696, "learning_rate": 4.109458681245527e-05, "loss": 0.0084, "step": 71600 }, { "epoch": 4.968941924587966e-13, "grad_norm": 0.2661706209182739, "learning_rate": 4.107749966116831e-05, "loss": 0.0098, "step": 71610 }, { "epoch": 4.969635813978357e-13, "grad_norm": 0.2405339777469635, "learning_rate": 4.106041676668777e-05, "loss": 0.0083, "step": 71620 }, { "epoch": 4.970329703368748e-13, "grad_norm": 0.23000003397464752, "learning_rate": 4.104333813076918e-05, "loss": 0.0079, "step": 71630 }, { "epoch": 4.971023592759138e-13, "grad_norm": 0.22884850203990936, "learning_rate": 4.102626375516764e-05, "loss": 0.0077, "step": 71640 }, { "epoch": 4.971717482149529e-13, "grad_norm": 0.2667182385921478, "learning_rate": 4.100919364163779e-05, "loss": 0.0075, "step": 71650 }, { "epoch": 4.97241137153992e-13, "grad_norm": 0.2571485936641693, "learning_rate": 4.099212779193388e-05, "loss": 0.0087, "step": 71660 }, { "epoch": 4.973105260930311e-13, "grad_norm": 0.18684649467468262, "learning_rate": 4.097506620780965e-05, "loss": 0.0074, "step": 71670 }, { "epoch": 4.973799150320701e-13, "grad_norm": 0.2333659678697586, "learning_rate": 4.095800889101848e-05, "loss": 0.008, "step": 71680 }, { "epoch": 4.974493039711092e-13, "grad_norm": 0.26112815737724304, "learning_rate": 4.094095584331327e-05, "loss": 0.0075, "step": 71690 }, { "epoch": 4.975186929101483e-13, "grad_norm": 0.24247026443481445, "learning_rate": 4.092390706644646e-05, "loss": 0.0076, "step": 71700 }, { "epoch": 4.975880818491873e-13, "grad_norm": 0.19976182281970978, "learning_rate": 4.09068625621701e-05, "loss": 0.0074, "step": 71710 }, { "epoch": 4.976574707882264e-13, "grad_norm": 0.22929047048091888, "learning_rate": 4.088982233223579e-05, "loss": 0.0076, "step": 71720 }, { "epoch": 4.977268597272655e-13, "grad_norm": 0.27247023582458496, "learning_rate": 4.087278637839462e-05, "loss": 0.0084, "step": 71730 }, { "epoch": 4.977962486663046e-13, "grad_norm": 0.22224968671798706, "learning_rate": 4.085575470239739e-05, "loss": 0.0079, "step": 71740 }, { "epoch": 4.978656376053436e-13, "grad_norm": 0.23562300205230713, "learning_rate": 4.083872730599428e-05, "loss": 0.0074, "step": 71750 }, { "epoch": 4.979350265443827e-13, "grad_norm": 0.21099023520946503, "learning_rate": 4.082170419093519e-05, "loss": 0.0083, "step": 71760 }, { "epoch": 4.980044154834218e-13, "grad_norm": 0.21015463769435883, "learning_rate": 4.080468535896945e-05, "loss": 0.0108, "step": 71770 }, { "epoch": 4.980738044224609e-13, "grad_norm": 0.24713435769081116, "learning_rate": 4.078767081184609e-05, "loss": 0.008, "step": 71780 }, { "epoch": 4.981431933614999e-13, "grad_norm": 0.20791742205619812, "learning_rate": 4.07706605513135e-05, "loss": 0.008, "step": 71790 }, { "epoch": 4.98212582300539e-13, "grad_norm": 0.21306073665618896, "learning_rate": 4.075365457911987e-05, "loss": 0.0075, "step": 71800 }, { "epoch": 4.982819712395781e-13, "grad_norm": 0.2258022576570511, "learning_rate": 4.0736652897012764e-05, "loss": 0.0073, "step": 71810 }, { "epoch": 4.983513601786171e-13, "grad_norm": 0.21678262948989868, "learning_rate": 4.071965550673937e-05, "loss": 0.0105, "step": 71820 }, { "epoch": 4.984207491176562e-13, "grad_norm": 0.22402147948741913, "learning_rate": 4.070266241004646e-05, "loss": 0.0078, "step": 71830 }, { "epoch": 4.984901380566953e-13, "grad_norm": 0.2242993414402008, "learning_rate": 4.068567360868033e-05, "loss": 0.0078, "step": 71840 }, { "epoch": 4.985595269957344e-13, "grad_norm": 0.24306723475456238, "learning_rate": 4.066868910438682e-05, "loss": 0.0074, "step": 71850 }, { "epoch": 4.986289159347734e-13, "grad_norm": 0.20798802375793457, "learning_rate": 4.0651708898911375e-05, "loss": 0.0074, "step": 71860 }, { "epoch": 4.986983048738125e-13, "grad_norm": 0.2522946000099182, "learning_rate": 4.063473299399898e-05, "loss": 0.0074, "step": 71870 }, { "epoch": 4.987676938128516e-13, "grad_norm": 0.23396234214305878, "learning_rate": 4.061776139139417e-05, "loss": 0.0077, "step": 71880 }, { "epoch": 4.988370827518906e-13, "grad_norm": 0.20198240876197815, "learning_rate": 4.060079409284104e-05, "loss": 0.0078, "step": 71890 }, { "epoch": 4.989064716909297e-13, "grad_norm": 0.2335481196641922, "learning_rate": 4.058383110008321e-05, "loss": 0.0075, "step": 71900 }, { "epoch": 4.989758606299688e-13, "grad_norm": 0.21442915499210358, "learning_rate": 4.056687241486396e-05, "loss": 0.0074, "step": 71910 }, { "epoch": 4.990452495690079e-13, "grad_norm": 0.7497925758361816, "learning_rate": 4.054991803892597e-05, "loss": 0.0102, "step": 71920 }, { "epoch": 4.991146385080469e-13, "grad_norm": 0.18596023321151733, "learning_rate": 4.053296797401167e-05, "loss": 0.0081, "step": 71930 }, { "epoch": 4.99184027447086e-13, "grad_norm": 0.20082831382751465, "learning_rate": 4.0516022221862855e-05, "loss": 0.0074, "step": 71940 }, { "epoch": 4.992534163861251e-13, "grad_norm": 0.2541281580924988, "learning_rate": 4.049908078422101e-05, "loss": 0.008, "step": 71950 }, { "epoch": 4.993228053251642e-13, "grad_norm": 0.2565300762653351, "learning_rate": 4.0482143662827114e-05, "loss": 0.0074, "step": 71960 }, { "epoch": 4.993921942642032e-13, "grad_norm": 0.17828215658664703, "learning_rate": 4.046521085942173e-05, "loss": 0.007, "step": 71970 }, { "epoch": 4.994615832032423e-13, "grad_norm": 0.2024649679660797, "learning_rate": 4.044828237574496e-05, "loss": 0.0093, "step": 71980 }, { "epoch": 4.995309721422814e-13, "grad_norm": 0.196929469704628, "learning_rate": 4.043135821353648e-05, "loss": 0.0075, "step": 71990 }, { "epoch": 4.996003610813204e-13, "grad_norm": 0.20472244918346405, "learning_rate": 4.0414438374535494e-05, "loss": 0.0073, "step": 72000 }, { "epoch": 4.996697500203595e-13, "grad_norm": 0.23511460423469543, "learning_rate": 4.039752286048079e-05, "loss": 0.0071, "step": 72010 }, { "epoch": 4.997391389593986e-13, "grad_norm": 0.2098323553800583, "learning_rate": 4.038061167311069e-05, "loss": 0.0073, "step": 72020 }, { "epoch": 4.998085278984377e-13, "grad_norm": 0.2106613963842392, "learning_rate": 4.03637048141631e-05, "loss": 0.0075, "step": 72030 }, { "epoch": 4.998779168374767e-13, "grad_norm": 0.2022550255060196, "learning_rate": 4.034680228537544e-05, "loss": 0.0079, "step": 72040 }, { "epoch": 4.999473057765158e-13, "grad_norm": 0.21072600781917572, "learning_rate": 4.0329904088484754e-05, "loss": 0.0075, "step": 72050 }, { "epoch": 5.000166947155549e-13, "grad_norm": 0.23629161715507507, "learning_rate": 4.031301022522751e-05, "loss": 0.0074, "step": 72060 }, { "epoch": 5.000860836545939e-13, "grad_norm": 0.23773589730262756, "learning_rate": 4.0296120697339915e-05, "loss": 0.0076, "step": 72070 }, { "epoch": 5.00155472593633e-13, "grad_norm": 0.3098701536655426, "learning_rate": 4.027923550655754e-05, "loss": 0.0093, "step": 72080 }, { "epoch": 5.002248615326721e-13, "grad_norm": 0.22183218598365784, "learning_rate": 4.026235465461568e-05, "loss": 0.0076, "step": 72090 }, { "epoch": 5.002942504717112e-13, "grad_norm": 0.19583551585674286, "learning_rate": 4.024547814324902e-05, "loss": 0.0076, "step": 72100 }, { "epoch": 5.003636394107502e-13, "grad_norm": 0.22308124601840973, "learning_rate": 4.022860597419199e-05, "loss": 0.0085, "step": 72110 }, { "epoch": 5.004330283497893e-13, "grad_norm": 0.19396895170211792, "learning_rate": 4.021173814917835e-05, "loss": 0.0072, "step": 72120 }, { "epoch": 5.005024172888284e-13, "grad_norm": 0.246995210647583, "learning_rate": 4.0194874669941654e-05, "loss": 0.008, "step": 72130 }, { "epoch": 5.005718062278675e-13, "grad_norm": 0.1976284384727478, "learning_rate": 4.0178015538214785e-05, "loss": 0.0072, "step": 72140 }, { "epoch": 5.006411951669065e-13, "grad_norm": 0.2204473912715912, "learning_rate": 4.016116075573034e-05, "loss": 0.0075, "step": 72150 }, { "epoch": 5.007105841059456e-13, "grad_norm": 0.1959783136844635, "learning_rate": 4.014431032422037e-05, "loss": 0.0074, "step": 72160 }, { "epoch": 5.007799730449847e-13, "grad_norm": 0.2030830830335617, "learning_rate": 4.012746424541655e-05, "loss": 0.0073, "step": 72170 }, { "epoch": 5.008493619840237e-13, "grad_norm": 0.2068309336900711, "learning_rate": 4.011062252105007e-05, "loss": 0.0072, "step": 72180 }, { "epoch": 5.009187509230628e-13, "grad_norm": 0.22144058346748352, "learning_rate": 4.009378515285167e-05, "loss": 0.0076, "step": 72190 }, { "epoch": 5.009881398621019e-13, "grad_norm": 0.1661754846572876, "learning_rate": 4.007695214255165e-05, "loss": 0.0076, "step": 72200 }, { "epoch": 5.01057528801141e-13, "grad_norm": 0.20186232030391693, "learning_rate": 4.006012349187989e-05, "loss": 0.0076, "step": 72210 }, { "epoch": 5.0112691774018e-13, "grad_norm": 0.20587413012981415, "learning_rate": 4.004329920256576e-05, "loss": 0.0071, "step": 72220 }, { "epoch": 5.011963066792191e-13, "grad_norm": 0.22385020554065704, "learning_rate": 4.002647927633824e-05, "loss": 0.0074, "step": 72230 }, { "epoch": 5.012656956182582e-13, "grad_norm": 0.20294788479804993, "learning_rate": 4.000966371492583e-05, "loss": 0.0091, "step": 72240 }, { "epoch": 5.013350845572973e-13, "grad_norm": 0.22286586463451385, "learning_rate": 3.99928525200566e-05, "loss": 0.0078, "step": 72250 }, { "epoch": 5.014044734963363e-13, "grad_norm": 0.2263411283493042, "learning_rate": 3.997604569345815e-05, "loss": 0.0076, "step": 72260 }, { "epoch": 5.014738624353754e-13, "grad_norm": 0.2244614213705063, "learning_rate": 3.995924323685767e-05, "loss": 0.0075, "step": 72270 }, { "epoch": 5.015432513744145e-13, "grad_norm": 0.21316419541835785, "learning_rate": 3.994244515198181e-05, "loss": 0.007, "step": 72280 }, { "epoch": 5.016126403134535e-13, "grad_norm": 0.2733892500400543, "learning_rate": 3.992565144055693e-05, "loss": 0.0079, "step": 72290 }, { "epoch": 5.016820292524926e-13, "grad_norm": 0.2507301867008209, "learning_rate": 3.9908862104308736e-05, "loss": 0.0077, "step": 72300 }, { "epoch": 5.017514181915317e-13, "grad_norm": 0.23983700573444366, "learning_rate": 3.98920771449627e-05, "loss": 0.0074, "step": 72310 }, { "epoch": 5.018208071305708e-13, "grad_norm": 0.2204606682062149, "learning_rate": 3.987529656424365e-05, "loss": 0.0071, "step": 72320 }, { "epoch": 5.018901960696098e-13, "grad_norm": 0.2031804770231247, "learning_rate": 3.985852036387611e-05, "loss": 0.0072, "step": 72330 }, { "epoch": 5.019595850086489e-13, "grad_norm": 0.22919593751430511, "learning_rate": 3.984174854558407e-05, "loss": 0.0073, "step": 72340 }, { "epoch": 5.02028973947688e-13, "grad_norm": 0.21253053843975067, "learning_rate": 3.98249811110911e-05, "loss": 0.0077, "step": 72350 }, { "epoch": 5.02098362886727e-13, "grad_norm": 0.20082232356071472, "learning_rate": 3.9808218062120304e-05, "loss": 0.0078, "step": 72360 }, { "epoch": 5.021677518257661e-13, "grad_norm": 0.26188981533050537, "learning_rate": 3.979145940039438e-05, "loss": 0.0075, "step": 72370 }, { "epoch": 5.022371407648052e-13, "grad_norm": 0.23832884430885315, "learning_rate": 3.977470512763552e-05, "loss": 0.0074, "step": 72380 }, { "epoch": 5.023065297038443e-13, "grad_norm": 0.3460409343242645, "learning_rate": 3.975795524556544e-05, "loss": 0.0085, "step": 72390 }, { "epoch": 5.023759186428833e-13, "grad_norm": 0.22608670592308044, "learning_rate": 3.974120975590555e-05, "loss": 0.0078, "step": 72400 }, { "epoch": 5.024453075819224e-13, "grad_norm": 0.24965547025203705, "learning_rate": 3.972446866037661e-05, "loss": 0.0081, "step": 72410 }, { "epoch": 5.025146965209615e-13, "grad_norm": 0.22004689276218414, "learning_rate": 3.97077319606991e-05, "loss": 0.0076, "step": 72420 }, { "epoch": 5.025840854600006e-13, "grad_norm": 0.2199837565422058, "learning_rate": 3.9690999658592925e-05, "loss": 0.0073, "step": 72430 }, { "epoch": 5.026534743990396e-13, "grad_norm": 0.39324092864990234, "learning_rate": 3.967427175577765e-05, "loss": 0.0076, "step": 72440 }, { "epoch": 5.027228633380787e-13, "grad_norm": 0.1815018653869629, "learning_rate": 3.965754825397226e-05, "loss": 0.0081, "step": 72450 }, { "epoch": 5.027922522771178e-13, "grad_norm": 0.21698138117790222, "learning_rate": 3.964082915489542e-05, "loss": 0.0074, "step": 72460 }, { "epoch": 5.028616412161568e-13, "grad_norm": 0.20958979427814484, "learning_rate": 3.962411446026522e-05, "loss": 0.0076, "step": 72470 }, { "epoch": 5.029310301551959e-13, "grad_norm": 0.21201686561107635, "learning_rate": 3.960740417179937e-05, "loss": 0.0071, "step": 72480 }, { "epoch": 5.03000419094235e-13, "grad_norm": 0.1682957410812378, "learning_rate": 3.959069829121513e-05, "loss": 0.0074, "step": 72490 }, { "epoch": 5.030698080332741e-13, "grad_norm": 0.22629782557487488, "learning_rate": 3.957399682022928e-05, "loss": 0.0075, "step": 72500 }, { "epoch": 5.031391969723131e-13, "grad_norm": 0.22418344020843506, "learning_rate": 3.955729976055816e-05, "loss": 0.0078, "step": 72510 }, { "epoch": 5.032085859113522e-13, "grad_norm": 0.19148865342140198, "learning_rate": 3.954060711391764e-05, "loss": 0.0076, "step": 72520 }, { "epoch": 5.032779748503913e-13, "grad_norm": 0.21954698860645294, "learning_rate": 3.9523918882023166e-05, "loss": 0.0075, "step": 72530 }, { "epoch": 5.033473637894303e-13, "grad_norm": 0.18736632168293, "learning_rate": 3.95072350665897e-05, "loss": 0.0072, "step": 72540 }, { "epoch": 5.034167527284694e-13, "grad_norm": 0.19568894803524017, "learning_rate": 3.949055566933177e-05, "loss": 0.0082, "step": 72550 }, { "epoch": 5.034861416675085e-13, "grad_norm": 0.18714098632335663, "learning_rate": 3.947388069196344e-05, "loss": 0.0074, "step": 72560 }, { "epoch": 5.035555306065476e-13, "grad_norm": 0.21821610629558563, "learning_rate": 3.945721013619833e-05, "loss": 0.0075, "step": 72570 }, { "epoch": 5.036249195455866e-13, "grad_norm": 0.1862587034702301, "learning_rate": 3.9440544003749595e-05, "loss": 0.0072, "step": 72580 }, { "epoch": 5.036943084846257e-13, "grad_norm": 0.19354769587516785, "learning_rate": 3.9423882296329934e-05, "loss": 0.007, "step": 72590 }, { "epoch": 5.037636974236648e-13, "grad_norm": 0.20490242540836334, "learning_rate": 3.940722501565163e-05, "loss": 0.0067, "step": 72600 }, { "epoch": 5.038330863627039e-13, "grad_norm": 0.22441893815994263, "learning_rate": 3.939057216342641e-05, "loss": 0.0075, "step": 72610 }, { "epoch": 5.039024753017429e-13, "grad_norm": 0.21171513199806213, "learning_rate": 3.937392374136569e-05, "loss": 0.0074, "step": 72620 }, { "epoch": 5.03971864240782e-13, "grad_norm": 0.20861685276031494, "learning_rate": 3.9357279751180266e-05, "loss": 0.0072, "step": 72630 }, { "epoch": 5.040412531798211e-13, "grad_norm": 0.22092919051647186, "learning_rate": 3.9340640194580676e-05, "loss": 0.007, "step": 72640 }, { "epoch": 5.041106421188601e-13, "grad_norm": 0.17988885939121246, "learning_rate": 3.932400507327678e-05, "loss": 0.0073, "step": 72650 }, { "epoch": 5.041800310578992e-13, "grad_norm": 0.2468172013759613, "learning_rate": 3.9307374388978215e-05, "loss": 0.0079, "step": 72660 }, { "epoch": 5.042494199969383e-13, "grad_norm": 0.23504583537578583, "learning_rate": 3.9290748143393934e-05, "loss": 0.0075, "step": 72670 }, { "epoch": 5.043188089359774e-13, "grad_norm": 0.23860283195972443, "learning_rate": 3.9274126338232604e-05, "loss": 0.0077, "step": 72680 }, { "epoch": 5.043881978750164e-13, "grad_norm": 0.18773749470710754, "learning_rate": 3.925750897520235e-05, "loss": 0.0072, "step": 72690 }, { "epoch": 5.044575868140555e-13, "grad_norm": 0.2086046040058136, "learning_rate": 3.924089605601087e-05, "loss": 0.0071, "step": 72700 }, { "epoch": 5.045269757530946e-13, "grad_norm": 0.20826013386249542, "learning_rate": 3.92242875823654e-05, "loss": 0.0081, "step": 72710 }, { "epoch": 5.045963646921336e-13, "grad_norm": 0.2081318497657776, "learning_rate": 3.9207683555972736e-05, "loss": 0.0074, "step": 72720 }, { "epoch": 5.046657536311727e-13, "grad_norm": 0.20474757254123688, "learning_rate": 3.9191083978539175e-05, "loss": 0.0073, "step": 72730 }, { "epoch": 5.047351425702118e-13, "grad_norm": 0.20529378950595856, "learning_rate": 3.91744888517706e-05, "loss": 0.0072, "step": 72740 }, { "epoch": 5.048045315092509e-13, "grad_norm": 0.20228774845600128, "learning_rate": 3.915789817737241e-05, "loss": 0.0069, "step": 72750 }, { "epoch": 5.048739204482899e-13, "grad_norm": 0.22721944749355316, "learning_rate": 3.914131195704955e-05, "loss": 0.0069, "step": 72760 }, { "epoch": 5.04943309387329e-13, "grad_norm": 0.1972881257534027, "learning_rate": 3.912473019250651e-05, "loss": 0.0078, "step": 72770 }, { "epoch": 5.050126983263681e-13, "grad_norm": 0.2054509073495865, "learning_rate": 3.910815288544734e-05, "loss": 0.0078, "step": 72780 }, { "epoch": 5.050820872654072e-13, "grad_norm": 0.1837097704410553, "learning_rate": 3.909158003757563e-05, "loss": 0.0071, "step": 72790 }, { "epoch": 5.051514762044462e-13, "grad_norm": 0.21665264666080475, "learning_rate": 3.9075011650594465e-05, "loss": 0.0073, "step": 72800 }, { "epoch": 5.052208651434853e-13, "grad_norm": 0.18979574739933014, "learning_rate": 3.90584477262065e-05, "loss": 0.0072, "step": 72810 }, { "epoch": 5.052902540825244e-13, "grad_norm": 0.22880718111991882, "learning_rate": 3.9041888266113964e-05, "loss": 0.008, "step": 72820 }, { "epoch": 5.053596430215634e-13, "grad_norm": 0.22786712646484375, "learning_rate": 3.9025333272018584e-05, "loss": 0.0079, "step": 72830 }, { "epoch": 5.054290319606025e-13, "grad_norm": 0.19903074204921722, "learning_rate": 3.900878274562164e-05, "loss": 0.0074, "step": 72840 }, { "epoch": 5.054984208996416e-13, "grad_norm": 0.25019845366477966, "learning_rate": 3.899223668862397e-05, "loss": 0.0078, "step": 72850 }, { "epoch": 5.055678098386807e-13, "grad_norm": 0.22188647091388702, "learning_rate": 3.897569510272592e-05, "loss": 0.0074, "step": 72860 }, { "epoch": 5.056371987777197e-13, "grad_norm": 0.20813937485218048, "learning_rate": 3.895915798962742e-05, "loss": 0.0078, "step": 72870 }, { "epoch": 5.057065877167588e-13, "grad_norm": 0.2219725102186203, "learning_rate": 3.894262535102789e-05, "loss": 0.0073, "step": 72880 }, { "epoch": 5.057759766557979e-13, "grad_norm": 0.18599487841129303, "learning_rate": 3.892609718862634e-05, "loss": 0.007, "step": 72890 }, { "epoch": 5.058453655948369e-13, "grad_norm": 0.2642306685447693, "learning_rate": 3.890957350412128e-05, "loss": 0.0071, "step": 72900 }, { "epoch": 5.05914754533876e-13, "grad_norm": 0.18017007410526276, "learning_rate": 3.889305429921077e-05, "loss": 0.0068, "step": 72910 }, { "epoch": 5.059841434729151e-13, "grad_norm": 0.16393138468265533, "learning_rate": 3.887653957559244e-05, "loss": 0.0073, "step": 72920 }, { "epoch": 5.060535324119542e-13, "grad_norm": 0.20666110515594482, "learning_rate": 3.886002933496344e-05, "loss": 0.0073, "step": 72930 }, { "epoch": 5.061229213509932e-13, "grad_norm": 0.20898239314556122, "learning_rate": 3.8843523579020384e-05, "loss": 0.0074, "step": 72940 }, { "epoch": 5.061923102900323e-13, "grad_norm": 0.2051830142736435, "learning_rate": 3.882702230945959e-05, "loss": 0.0071, "step": 72950 }, { "epoch": 5.062616992290714e-13, "grad_norm": 0.19710169732570648, "learning_rate": 3.881052552797673e-05, "loss": 0.0072, "step": 72960 }, { "epoch": 5.063310881681105e-13, "grad_norm": 0.20460155606269836, "learning_rate": 3.8794033236267195e-05, "loss": 0.0071, "step": 72970 }, { "epoch": 5.064004771071495e-13, "grad_norm": 0.24088336527347565, "learning_rate": 3.877754543602573e-05, "loss": 0.0082, "step": 72980 }, { "epoch": 5.064698660461886e-13, "grad_norm": 0.19162394106388092, "learning_rate": 3.876106212894682e-05, "loss": 0.0078, "step": 72990 }, { "epoch": 5.065392549852277e-13, "grad_norm": 0.2586126923561096, "learning_rate": 3.874458331672429e-05, "loss": 0.0076, "step": 73000 }, { "epoch": 5.066086439242667e-13, "grad_norm": 0.189431831240654, "learning_rate": 3.872810900105163e-05, "loss": 0.0077, "step": 73010 }, { "epoch": 5.066780328633058e-13, "grad_norm": 0.20959340035915375, "learning_rate": 3.871163918362183e-05, "loss": 0.0078, "step": 73020 }, { "epoch": 5.067474218023449e-13, "grad_norm": 0.2783803641796112, "learning_rate": 3.869517386612744e-05, "loss": 0.0074, "step": 73030 }, { "epoch": 5.06816810741384e-13, "grad_norm": 0.20801781117916107, "learning_rate": 3.867871305026049e-05, "loss": 0.0072, "step": 73040 }, { "epoch": 5.06886199680423e-13, "grad_norm": 0.20137028396129608, "learning_rate": 3.86622567377126e-05, "loss": 0.0072, "step": 73050 }, { "epoch": 5.069555886194621e-13, "grad_norm": 0.18527880311012268, "learning_rate": 3.864580493017492e-05, "loss": 0.007, "step": 73060 }, { "epoch": 5.070249775585012e-13, "grad_norm": 0.21667008101940155, "learning_rate": 3.8629357629338137e-05, "loss": 0.0068, "step": 73070 }, { "epoch": 5.070943664975402e-13, "grad_norm": 0.23108696937561035, "learning_rate": 3.861291483689245e-05, "loss": 0.0073, "step": 73080 }, { "epoch": 5.071637554365793e-13, "grad_norm": 0.24004340171813965, "learning_rate": 3.8596476554527605e-05, "loss": 0.0071, "step": 73090 }, { "epoch": 5.072331443756184e-13, "grad_norm": 0.19417116045951843, "learning_rate": 3.8580042783932904e-05, "loss": 0.0071, "step": 73100 }, { "epoch": 5.073025333146575e-13, "grad_norm": 0.20840926468372345, "learning_rate": 3.856361352679718e-05, "loss": 0.0071, "step": 73110 }, { "epoch": 5.073719222536965e-13, "grad_norm": 0.2013937383890152, "learning_rate": 3.854718878480878e-05, "loss": 0.0071, "step": 73120 }, { "epoch": 5.074413111927356e-13, "grad_norm": 0.24372199177742004, "learning_rate": 3.8530768559655634e-05, "loss": 0.0074, "step": 73130 }, { "epoch": 5.075107001317747e-13, "grad_norm": 0.21327507495880127, "learning_rate": 3.851435285302511e-05, "loss": 0.0081, "step": 73140 }, { "epoch": 5.075800890708138e-13, "grad_norm": 0.19376221299171448, "learning_rate": 3.849794166660425e-05, "loss": 0.0075, "step": 73150 }, { "epoch": 5.076494780098528e-13, "grad_norm": 0.1855311393737793, "learning_rate": 3.848153500207948e-05, "loss": 0.0074, "step": 73160 }, { "epoch": 5.077188669488919e-13, "grad_norm": 0.20521797239780426, "learning_rate": 3.846513286113694e-05, "loss": 0.0076, "step": 73170 }, { "epoch": 5.07788255887931e-13, "grad_norm": 0.1861760914325714, "learning_rate": 3.844873524546211e-05, "loss": 0.0075, "step": 73180 }, { "epoch": 5.0785764482697e-13, "grad_norm": 0.17788830399513245, "learning_rate": 3.843234215674017e-05, "loss": 0.0072, "step": 73190 }, { "epoch": 5.079270337660091e-13, "grad_norm": 0.21448875963687897, "learning_rate": 3.8415953596655704e-05, "loss": 0.0071, "step": 73200 }, { "epoch": 5.079964227050482e-13, "grad_norm": 0.17874670028686523, "learning_rate": 3.839956956689297e-05, "loss": 0.007, "step": 73210 }, { "epoch": 5.080658116440873e-13, "grad_norm": 0.17660081386566162, "learning_rate": 3.83831900691356e-05, "loss": 0.0068, "step": 73220 }, { "epoch": 5.081352005831263e-13, "grad_norm": 0.20773403346538544, "learning_rate": 3.836681510506688e-05, "loss": 0.0069, "step": 73230 }, { "epoch": 5.082045895221654e-13, "grad_norm": 0.228367879986763, "learning_rate": 3.83504446763696e-05, "loss": 0.0073, "step": 73240 }, { "epoch": 5.082739784612045e-13, "grad_norm": 0.21370752155780792, "learning_rate": 3.833407878472607e-05, "loss": 0.007, "step": 73250 }, { "epoch": 5.083433674002436e-13, "grad_norm": 0.2159319669008255, "learning_rate": 3.831771743181814e-05, "loss": 0.0068, "step": 73260 }, { "epoch": 5.084127563392826e-13, "grad_norm": 0.19860893487930298, "learning_rate": 3.8301360619327156e-05, "loss": 0.0071, "step": 73270 }, { "epoch": 5.084821452783217e-13, "grad_norm": 0.19113780558109283, "learning_rate": 3.82850083489341e-05, "loss": 0.0071, "step": 73280 }, { "epoch": 5.085515342173608e-13, "grad_norm": 0.21781708300113678, "learning_rate": 3.826866062231938e-05, "loss": 0.0074, "step": 73290 }, { "epoch": 5.086209231563998e-13, "grad_norm": 0.21344026923179626, "learning_rate": 3.8252317441163e-05, "loss": 0.0078, "step": 73300 }, { "epoch": 5.086903120954389e-13, "grad_norm": 0.18718063831329346, "learning_rate": 3.823597880714444e-05, "loss": 0.0072, "step": 73310 }, { "epoch": 5.08759701034478e-13, "grad_norm": 0.1821633279323578, "learning_rate": 3.821964472194282e-05, "loss": 0.0074, "step": 73320 }, { "epoch": 5.088290899735171e-13, "grad_norm": 0.22287370264530182, "learning_rate": 3.820331518723665e-05, "loss": 0.0073, "step": 73330 }, { "epoch": 5.088984789125561e-13, "grad_norm": 0.19782060384750366, "learning_rate": 3.818699020470406e-05, "loss": 0.007, "step": 73340 }, { "epoch": 5.089678678515952e-13, "grad_norm": 0.19609631597995758, "learning_rate": 3.817066977602273e-05, "loss": 0.0069, "step": 73350 }, { "epoch": 5.090372567906343e-13, "grad_norm": 0.21344462037086487, "learning_rate": 3.81543539028698e-05, "loss": 0.0071, "step": 73360 }, { "epoch": 5.091066457296733e-13, "grad_norm": 0.18618470430374146, "learning_rate": 3.813804258692199e-05, "loss": 0.0067, "step": 73370 }, { "epoch": 5.091760346687124e-13, "grad_norm": 0.2045094072818756, "learning_rate": 3.8121735829855546e-05, "loss": 0.0066, "step": 73380 }, { "epoch": 5.092454236077515e-13, "grad_norm": 0.1898142397403717, "learning_rate": 3.810543363334626e-05, "loss": 0.0069, "step": 73390 }, { "epoch": 5.093148125467906e-13, "grad_norm": 0.24216201901435852, "learning_rate": 3.8089135999069395e-05, "loss": 0.0071, "step": 73400 }, { "epoch": 5.093842014858296e-13, "grad_norm": 0.18897303938865662, "learning_rate": 3.807284292869982e-05, "loss": 0.0068, "step": 73410 }, { "epoch": 5.094535904248687e-13, "grad_norm": 0.17371760308742523, "learning_rate": 3.805655442391189e-05, "loss": 0.0069, "step": 73420 }, { "epoch": 5.095229793639078e-13, "grad_norm": 0.20562611520290375, "learning_rate": 3.80402704863795e-05, "loss": 0.0072, "step": 73430 }, { "epoch": 5.095923683029469e-13, "grad_norm": 0.18853405117988586, "learning_rate": 3.802399111777609e-05, "loss": 0.0071, "step": 73440 }, { "epoch": 5.096617572419859e-13, "grad_norm": 0.19643312692642212, "learning_rate": 3.80077163197746e-05, "loss": 0.0073, "step": 73450 }, { "epoch": 5.09731146181025e-13, "grad_norm": 0.20492711663246155, "learning_rate": 3.799144609404755e-05, "loss": 0.0075, "step": 73460 }, { "epoch": 5.098005351200641e-13, "grad_norm": 0.20878776907920837, "learning_rate": 3.797518044226689e-05, "loss": 0.0073, "step": 73470 }, { "epoch": 5.098699240591031e-13, "grad_norm": 0.18306879699230194, "learning_rate": 3.795891936610426e-05, "loss": 0.0074, "step": 73480 }, { "epoch": 5.099393129981422e-13, "grad_norm": 0.20920579135417938, "learning_rate": 3.794266286723065e-05, "loss": 0.0073, "step": 73490 }, { "epoch": 5.100087019371813e-13, "grad_norm": 0.2245684564113617, "learning_rate": 3.7926410947316746e-05, "loss": 0.007, "step": 73500 }, { "epoch": 5.100780908762204e-13, "grad_norm": 0.2247379571199417, "learning_rate": 3.791016360803262e-05, "loss": 0.0071, "step": 73510 }, { "epoch": 5.101474798152594e-13, "grad_norm": 0.21295157074928284, "learning_rate": 3.789392085104802e-05, "loss": 0.0069, "step": 73520 }, { "epoch": 5.102168687542985e-13, "grad_norm": 0.20694586634635925, "learning_rate": 3.787768267803204e-05, "loss": 0.0067, "step": 73530 }, { "epoch": 5.102862576933376e-13, "grad_norm": 0.23159363865852356, "learning_rate": 3.786144909065351e-05, "loss": 0.0066, "step": 73540 }, { "epoch": 5.103556466323766e-13, "grad_norm": 0.21833626925945282, "learning_rate": 3.7845220090580595e-05, "loss": 0.007, "step": 73550 }, { "epoch": 5.104250355714157e-13, "grad_norm": 0.23242850601673126, "learning_rate": 3.782899567948113e-05, "loss": 0.007, "step": 73560 }, { "epoch": 5.104944245104548e-13, "grad_norm": 0.1963491439819336, "learning_rate": 3.781277585902239e-05, "loss": 0.0069, "step": 73570 }, { "epoch": 5.105638134494939e-13, "grad_norm": 0.18799564242362976, "learning_rate": 3.779656063087125e-05, "loss": 0.0068, "step": 73580 }, { "epoch": 5.106332023885329e-13, "grad_norm": 0.1852056086063385, "learning_rate": 3.778034999669405e-05, "loss": 0.0072, "step": 73590 }, { "epoch": 5.10702591327572e-13, "grad_norm": 0.1886712610721588, "learning_rate": 3.776414395815671e-05, "loss": 0.0074, "step": 73600 }, { "epoch": 5.107719802666111e-13, "grad_norm": 0.21694137156009674, "learning_rate": 3.7747942516924604e-05, "loss": 0.0074, "step": 73610 }, { "epoch": 5.108413692056502e-13, "grad_norm": 0.22317545115947723, "learning_rate": 3.7731745674662745e-05, "loss": 0.0072, "step": 73620 }, { "epoch": 5.109107581446892e-13, "grad_norm": 0.19980330765247345, "learning_rate": 3.771555343303557e-05, "loss": 0.0076, "step": 73630 }, { "epoch": 5.109801470837283e-13, "grad_norm": 0.20917987823486328, "learning_rate": 3.76993657937071e-05, "loss": 0.0072, "step": 73640 }, { "epoch": 5.110495360227674e-13, "grad_norm": 0.19344587624073029, "learning_rate": 3.768318275834087e-05, "loss": 0.0073, "step": 73650 }, { "epoch": 5.111189249618064e-13, "grad_norm": 0.24616476893424988, "learning_rate": 3.7667004328599935e-05, "loss": 0.0069, "step": 73660 }, { "epoch": 5.111883139008455e-13, "grad_norm": 0.18349526822566986, "learning_rate": 3.765083050614684e-05, "loss": 0.0071, "step": 73670 }, { "epoch": 5.112577028398846e-13, "grad_norm": 0.17309872806072235, "learning_rate": 3.7634661292643785e-05, "loss": 0.0069, "step": 73680 }, { "epoch": 5.113270917789237e-13, "grad_norm": 0.1947578638792038, "learning_rate": 3.761849668975232e-05, "loss": 0.0067, "step": 73690 }, { "epoch": 5.113964807179627e-13, "grad_norm": 0.18312904238700867, "learning_rate": 3.760233669913366e-05, "loss": 0.007, "step": 73700 }, { "epoch": 5.114658696570018e-13, "grad_norm": 0.2052687108516693, "learning_rate": 3.7586181322448466e-05, "loss": 0.007, "step": 73710 }, { "epoch": 5.115352585960409e-13, "grad_norm": 0.21913598477840424, "learning_rate": 3.7570030561356974e-05, "loss": 0.007, "step": 73720 }, { "epoch": 5.116046475350799e-13, "grad_norm": 0.20779265463352203, "learning_rate": 3.755388441751892e-05, "loss": 0.0071, "step": 73730 }, { "epoch": 5.11674036474119e-13, "grad_norm": 0.20501314103603363, "learning_rate": 3.753774289259358e-05, "loss": 0.0069, "step": 73740 }, { "epoch": 5.117434254131581e-13, "grad_norm": 0.20660161972045898, "learning_rate": 3.752160598823974e-05, "loss": 0.0072, "step": 73750 }, { "epoch": 5.118128143521972e-13, "grad_norm": 0.22077049314975739, "learning_rate": 3.7505473706115716e-05, "loss": 0.0074, "step": 73760 }, { "epoch": 5.118822032912362e-13, "grad_norm": 0.18432672321796417, "learning_rate": 3.748934604787935e-05, "loss": 0.0071, "step": 73770 }, { "epoch": 5.119515922302753e-13, "grad_norm": 0.19946354627609253, "learning_rate": 3.747322301518801e-05, "loss": 0.0071, "step": 73780 }, { "epoch": 5.120209811693144e-13, "grad_norm": 0.17975766956806183, "learning_rate": 3.745710460969861e-05, "loss": 0.0072, "step": 73790 }, { "epoch": 5.120903701083535e-13, "grad_norm": 0.2062721848487854, "learning_rate": 3.744099083306751e-05, "loss": 0.0073, "step": 73800 }, { "epoch": 5.121597590473925e-13, "grad_norm": 0.2003089040517807, "learning_rate": 3.7424881686950736e-05, "loss": 0.0072, "step": 73810 }, { "epoch": 5.122291479864316e-13, "grad_norm": 0.21260473132133484, "learning_rate": 3.740877717300366e-05, "loss": 0.0069, "step": 73820 }, { "epoch": 5.122985369254707e-13, "grad_norm": 0.21102769672870636, "learning_rate": 3.739267729288136e-05, "loss": 0.0071, "step": 73830 }, { "epoch": 5.123679258645097e-13, "grad_norm": 0.18109411001205444, "learning_rate": 3.7376582048238263e-05, "loss": 0.0069, "step": 73840 }, { "epoch": 5.124373148035488e-13, "grad_norm": 0.1712174415588379, "learning_rate": 3.73604914407285e-05, "loss": 0.0066, "step": 73850 }, { "epoch": 5.125067037425879e-13, "grad_norm": 0.21086683869361877, "learning_rate": 3.734440547200553e-05, "loss": 0.007, "step": 73860 }, { "epoch": 5.12576092681627e-13, "grad_norm": 0.18110451102256775, "learning_rate": 3.732832414372253e-05, "loss": 0.0072, "step": 73870 }, { "epoch": 5.12645481620666e-13, "grad_norm": 0.20008587837219238, "learning_rate": 3.731224745753205e-05, "loss": 0.007, "step": 73880 }, { "epoch": 5.127148705597051e-13, "grad_norm": 0.184193953871727, "learning_rate": 3.729617541508623e-05, "loss": 0.0072, "step": 73890 }, { "epoch": 5.127842594987442e-13, "grad_norm": 0.18994036316871643, "learning_rate": 3.728010801803673e-05, "loss": 0.0072, "step": 73900 }, { "epoch": 5.128536484377832e-13, "grad_norm": 0.25485554337501526, "learning_rate": 3.726404526803472e-05, "loss": 0.0074, "step": 73910 }, { "epoch": 5.129230373768223e-13, "grad_norm": 0.2044658660888672, "learning_rate": 3.72479871667309e-05, "loss": 0.0073, "step": 73920 }, { "epoch": 5.129924263158614e-13, "grad_norm": 0.16207271814346313, "learning_rate": 3.723193371577549e-05, "loss": 0.0073, "step": 73930 }, { "epoch": 5.130618152549005e-13, "grad_norm": 0.20278382301330566, "learning_rate": 3.721588491681822e-05, "loss": 0.0074, "step": 73940 }, { "epoch": 5.131312041939395e-13, "grad_norm": 0.19731701910495758, "learning_rate": 3.719984077150838e-05, "loss": 0.0073, "step": 73950 }, { "epoch": 5.132005931329786e-13, "grad_norm": 0.22107556462287903, "learning_rate": 3.7183801281494735e-05, "loss": 0.0073, "step": 73960 }, { "epoch": 5.132699820720177e-13, "grad_norm": 0.18027295172214508, "learning_rate": 3.716776644842559e-05, "loss": 0.007, "step": 73970 }, { "epoch": 5.133393710110568e-13, "grad_norm": 0.21631348133087158, "learning_rate": 3.715173627394879e-05, "loss": 0.0068, "step": 73980 }, { "epoch": 5.134087599500958e-13, "grad_norm": 0.18582968413829803, "learning_rate": 3.7135710759711696e-05, "loss": 0.0069, "step": 73990 }, { "epoch": 5.134781488891349e-13, "grad_norm": 0.1925685554742813, "learning_rate": 3.711968990736111e-05, "loss": 0.0068, "step": 74000 }, { "epoch": 5.13547537828174e-13, "grad_norm": 0.18950441479682922, "learning_rate": 3.710367371854352e-05, "loss": 0.0065, "step": 74010 }, { "epoch": 5.13616926767213e-13, "grad_norm": 0.19195134937763214, "learning_rate": 3.7087662194904744e-05, "loss": 0.0072, "step": 74020 }, { "epoch": 5.136863157062521e-13, "grad_norm": 0.21031539142131805, "learning_rate": 3.707165533809032e-05, "loss": 0.0075, "step": 74030 }, { "epoch": 5.137557046452912e-13, "grad_norm": 0.1753561645746231, "learning_rate": 3.705565314974508e-05, "loss": 0.0069, "step": 74040 }, { "epoch": 5.138250935843303e-13, "grad_norm": 0.2392679750919342, "learning_rate": 3.7039655631513614e-05, "loss": 0.0073, "step": 74050 }, { "epoch": 5.138944825233693e-13, "grad_norm": 0.18196609616279602, "learning_rate": 3.702366278503982e-05, "loss": 0.0071, "step": 74060 }, { "epoch": 5.139638714624084e-13, "grad_norm": 0.1972610503435135, "learning_rate": 3.70076746119673e-05, "loss": 0.0073, "step": 74070 }, { "epoch": 5.140332604014475e-13, "grad_norm": 0.25528815388679504, "learning_rate": 3.699169111393902e-05, "loss": 0.0077, "step": 74080 }, { "epoch": 5.141026493404866e-13, "grad_norm": 0.24963466823101044, "learning_rate": 3.697571229259756e-05, "loss": 0.0071, "step": 74090 }, { "epoch": 5.141720382795256e-13, "grad_norm": 0.20963037014007568, "learning_rate": 3.6959738149584984e-05, "loss": 0.0073, "step": 74100 }, { "epoch": 5.142414272185647e-13, "grad_norm": 0.21366380155086517, "learning_rate": 3.694376868654289e-05, "loss": 0.0071, "step": 74110 }, { "epoch": 5.143108161576038e-13, "grad_norm": 0.1979323923587799, "learning_rate": 3.692780390511239e-05, "loss": 0.0071, "step": 74120 }, { "epoch": 5.143802050966428e-13, "grad_norm": 0.18444399535655975, "learning_rate": 3.69118438069341e-05, "loss": 0.007, "step": 74130 }, { "epoch": 5.144495940356819e-13, "grad_norm": 0.19986654818058014, "learning_rate": 3.6895888393648216e-05, "loss": 0.0066, "step": 74140 }, { "epoch": 5.14518982974721e-13, "grad_norm": 0.17262916266918182, "learning_rate": 3.6879937666894315e-05, "loss": 0.0069, "step": 74150 }, { "epoch": 5.145883719137601e-13, "grad_norm": 0.18626274168491364, "learning_rate": 3.686399162831168e-05, "loss": 0.0067, "step": 74160 }, { "epoch": 5.146577608527991e-13, "grad_norm": 0.17052069306373596, "learning_rate": 3.684805027953893e-05, "loss": 0.0067, "step": 74170 }, { "epoch": 5.147271497918382e-13, "grad_norm": 0.22803717851638794, "learning_rate": 3.683211362221437e-05, "loss": 0.0075, "step": 74180 }, { "epoch": 5.147965387308773e-13, "grad_norm": 0.18612004816532135, "learning_rate": 3.681618165797565e-05, "loss": 0.0069, "step": 74190 }, { "epoch": 5.148659276699163e-13, "grad_norm": 0.17613568902015686, "learning_rate": 3.6800254388460126e-05, "loss": 0.0069, "step": 74200 }, { "epoch": 5.149353166089554e-13, "grad_norm": 0.2046521008014679, "learning_rate": 3.6784331815304487e-05, "loss": 0.0071, "step": 74210 }, { "epoch": 5.150047055479945e-13, "grad_norm": 0.2261815369129181, "learning_rate": 3.6768413940145064e-05, "loss": 0.007, "step": 74220 }, { "epoch": 5.150740944870336e-13, "grad_norm": 0.21885471045970917, "learning_rate": 3.6752500764617655e-05, "loss": 0.0072, "step": 74230 }, { "epoch": 5.151434834260726e-13, "grad_norm": 0.19741623103618622, "learning_rate": 3.6736592290357603e-05, "loss": 0.0073, "step": 74240 }, { "epoch": 5.152128723651117e-13, "grad_norm": 0.20852044224739075, "learning_rate": 3.672068851899973e-05, "loss": 0.0071, "step": 74250 }, { "epoch": 5.152822613041508e-13, "grad_norm": 0.1818351149559021, "learning_rate": 3.670478945217841e-05, "loss": 0.007, "step": 74260 }, { "epoch": 5.153516502431899e-13, "grad_norm": 0.17300499975681305, "learning_rate": 3.6688895091527514e-05, "loss": 0.0071, "step": 74270 }, { "epoch": 5.154210391822289e-13, "grad_norm": 0.18868187069892883, "learning_rate": 3.6673005438680443e-05, "loss": 0.0068, "step": 74280 }, { "epoch": 5.15490428121268e-13, "grad_norm": 0.1750366985797882, "learning_rate": 3.665712049527009e-05, "loss": 0.0068, "step": 74290 }, { "epoch": 5.155598170603071e-13, "grad_norm": 0.1960323005914688, "learning_rate": 3.664124026292891e-05, "loss": 0.0066, "step": 74300 }, { "epoch": 5.156292059993461e-13, "grad_norm": 0.19155797362327576, "learning_rate": 3.66253647432888e-05, "loss": 0.0069, "step": 74310 }, { "epoch": 5.156985949383852e-13, "grad_norm": 0.2176945060491562, "learning_rate": 3.6609493937981257e-05, "loss": 0.0066, "step": 74320 }, { "epoch": 5.157679838774243e-13, "grad_norm": 0.20123378932476044, "learning_rate": 3.659362784863722e-05, "loss": 0.0067, "step": 74330 }, { "epoch": 5.158373728164634e-13, "grad_norm": 0.18609696626663208, "learning_rate": 3.657776647688723e-05, "loss": 0.0072, "step": 74340 }, { "epoch": 5.159067617555024e-13, "grad_norm": 0.21672601997852325, "learning_rate": 3.656190982436121e-05, "loss": 0.007, "step": 74350 }, { "epoch": 5.159761506945415e-13, "grad_norm": 0.18780139088630676, "learning_rate": 3.6546057892688756e-05, "loss": 0.0067, "step": 74360 }, { "epoch": 5.160455396335806e-13, "grad_norm": 0.2087203562259674, "learning_rate": 3.6530210683498816e-05, "loss": 0.007, "step": 74370 }, { "epoch": 5.161149285726196e-13, "grad_norm": 0.19263465702533722, "learning_rate": 3.6514368198420046e-05, "loss": 0.0069, "step": 74380 }, { "epoch": 5.161843175116587e-13, "grad_norm": 0.20731231570243835, "learning_rate": 3.64985304390804e-05, "loss": 0.0073, "step": 74390 }, { "epoch": 5.162537064506978e-13, "grad_norm": 0.16653332114219666, "learning_rate": 3.648269740710754e-05, "loss": 0.0071, "step": 74400 }, { "epoch": 5.163230953897369e-13, "grad_norm": 0.19945375621318817, "learning_rate": 3.64668691041285e-05, "loss": 0.007, "step": 74410 }, { "epoch": 5.163924843287759e-13, "grad_norm": 0.19464868307113647, "learning_rate": 3.6451045531769904e-05, "loss": 0.0069, "step": 74420 }, { "epoch": 5.16461873267815e-13, "grad_norm": 0.17891253530979156, "learning_rate": 3.6435226691657876e-05, "loss": 0.0071, "step": 74430 }, { "epoch": 5.165312622068541e-13, "grad_norm": 0.16337011754512787, "learning_rate": 3.641941258541804e-05, "loss": 0.0068, "step": 74440 }, { "epoch": 5.166006511458932e-13, "grad_norm": 0.17340154945850372, "learning_rate": 3.640360321467553e-05, "loss": 0.0068, "step": 74450 }, { "epoch": 5.166700400849322e-13, "grad_norm": 0.15398035943508148, "learning_rate": 3.638779858105503e-05, "loss": 0.0069, "step": 74460 }, { "epoch": 5.167394290239713e-13, "grad_norm": 0.13692963123321533, "learning_rate": 3.63719986861807e-05, "loss": 0.0066, "step": 74470 }, { "epoch": 5.168088179630104e-13, "grad_norm": 0.18725420534610748, "learning_rate": 3.635620353167622e-05, "loss": 0.0066, "step": 74480 }, { "epoch": 5.168782069020494e-13, "grad_norm": 0.19829161465168, "learning_rate": 3.6340413119164794e-05, "loss": 0.0067, "step": 74490 }, { "epoch": 5.169475958410885e-13, "grad_norm": 0.19881287217140198, "learning_rate": 3.632462745026914e-05, "loss": 0.007, "step": 74500 }, { "epoch": 5.170169847801276e-13, "grad_norm": 0.20684915781021118, "learning_rate": 3.630884652661146e-05, "loss": 0.0071, "step": 74510 }, { "epoch": 5.170863737191667e-13, "grad_norm": 0.1913360208272934, "learning_rate": 3.6293070349813505e-05, "loss": 0.0069, "step": 74520 }, { "epoch": 5.171557626582057e-13, "grad_norm": 0.19892609119415283, "learning_rate": 3.627729892149653e-05, "loss": 0.0069, "step": 74530 }, { "epoch": 5.172251515972448e-13, "grad_norm": 0.2187400907278061, "learning_rate": 3.62615322432813e-05, "loss": 0.007, "step": 74540 }, { "epoch": 5.172945405362839e-13, "grad_norm": 0.20103849470615387, "learning_rate": 3.624577031678803e-05, "loss": 0.0075, "step": 74550 }, { "epoch": 5.173639294753229e-13, "grad_norm": 0.1951591819524765, "learning_rate": 3.623001314363658e-05, "loss": 0.007, "step": 74560 }, { "epoch": 5.17433318414362e-13, "grad_norm": 0.1999409794807434, "learning_rate": 3.621426072544616e-05, "loss": 0.0071, "step": 74570 }, { "epoch": 5.175027073534011e-13, "grad_norm": 0.18774482607841492, "learning_rate": 3.619851306383569e-05, "loss": 0.007, "step": 74580 }, { "epoch": 5.175720962924402e-13, "grad_norm": 0.21221835911273956, "learning_rate": 3.6182770160423364e-05, "loss": 0.0072, "step": 74590 }, { "epoch": 5.176414852314792e-13, "grad_norm": 0.17620238661766052, "learning_rate": 3.616703201682711e-05, "loss": 0.0066, "step": 74600 }, { "epoch": 5.177108741705183e-13, "grad_norm": 0.1977871209383011, "learning_rate": 3.61512986346642e-05, "loss": 0.0068, "step": 74610 }, { "epoch": 5.177802631095574e-13, "grad_norm": 0.25347670912742615, "learning_rate": 3.613557001555152e-05, "loss": 0.007, "step": 74620 }, { "epoch": 5.178496520485965e-13, "grad_norm": 0.2047416716814041, "learning_rate": 3.611984616110542e-05, "loss": 0.0067, "step": 74630 }, { "epoch": 5.179190409876355e-13, "grad_norm": 0.1719169020652771, "learning_rate": 3.610412707294176e-05, "loss": 0.0067, "step": 74640 }, { "epoch": 5.179884299266746e-13, "grad_norm": 0.2031312733888626, "learning_rate": 3.608841275267594e-05, "loss": 0.007, "step": 74650 }, { "epoch": 5.180578188657137e-13, "grad_norm": 0.20401531457901, "learning_rate": 3.6072703201922835e-05, "loss": 0.0071, "step": 74660 }, { "epoch": 5.181272078047527e-13, "grad_norm": 0.18003034591674805, "learning_rate": 3.605699842229689e-05, "loss": 0.007, "step": 74670 }, { "epoch": 5.181965967437918e-13, "grad_norm": 0.17515701055526733, "learning_rate": 3.604129841541192e-05, "loss": 0.007, "step": 74680 }, { "epoch": 5.182659856828309e-13, "grad_norm": 0.17847959697246552, "learning_rate": 3.602560318288146e-05, "loss": 0.007, "step": 74690 }, { "epoch": 5.1833537462187e-13, "grad_norm": 0.1987888216972351, "learning_rate": 3.600991272631832e-05, "loss": 0.0072, "step": 74700 }, { "epoch": 5.18404763560909e-13, "grad_norm": 0.19216163456439972, "learning_rate": 3.599422704733506e-05, "loss": 0.0075, "step": 74710 }, { "epoch": 5.184741524999481e-13, "grad_norm": 0.19933000206947327, "learning_rate": 3.597854614754352e-05, "loss": 0.007, "step": 74720 }, { "epoch": 5.185435414389872e-13, "grad_norm": 0.20332731306552887, "learning_rate": 3.596287002855525e-05, "loss": 0.0071, "step": 74730 }, { "epoch": 5.186129303780262e-13, "grad_norm": 0.17850393056869507, "learning_rate": 3.594719869198114e-05, "loss": 0.0071, "step": 74740 }, { "epoch": 5.186823193170653e-13, "grad_norm": 0.1996726542711258, "learning_rate": 3.5931532139431714e-05, "loss": 0.0068, "step": 74750 }, { "epoch": 5.187517082561044e-13, "grad_norm": 0.19896714389324188, "learning_rate": 3.5915870372516915e-05, "loss": 0.0066, "step": 74760 }, { "epoch": 5.188210971951435e-13, "grad_norm": 0.18640482425689697, "learning_rate": 3.5900213392846266e-05, "loss": 0.007, "step": 74770 }, { "epoch": 5.188904861341825e-13, "grad_norm": 0.21023161709308624, "learning_rate": 3.588456120202875e-05, "loss": 0.0067, "step": 74780 }, { "epoch": 5.189598750732216e-13, "grad_norm": 0.18113680183887482, "learning_rate": 3.586891380167286e-05, "loss": 0.0065, "step": 74790 }, { "epoch": 5.190292640122607e-13, "grad_norm": 0.18182125687599182, "learning_rate": 3.585327119338663e-05, "loss": 0.0066, "step": 74800 }, { "epoch": 5.190986529512998e-13, "grad_norm": 0.1771525740623474, "learning_rate": 3.583763337877759e-05, "loss": 0.0068, "step": 74810 }, { "epoch": 5.191680418903388e-13, "grad_norm": 0.16178016364574432, "learning_rate": 3.5822000359452746e-05, "loss": 0.0072, "step": 74820 }, { "epoch": 5.192374308293779e-13, "grad_norm": 0.2208496779203415, "learning_rate": 3.580637213701864e-05, "loss": 0.0071, "step": 74830 }, { "epoch": 5.19306819768417e-13, "grad_norm": 0.1908951848745346, "learning_rate": 3.579074871308132e-05, "loss": 0.0072, "step": 74840 }, { "epoch": 5.19376208707456e-13, "grad_norm": 0.20335163176059723, "learning_rate": 3.577513008924633e-05, "loss": 0.0071, "step": 74850 }, { "epoch": 5.194455976464951e-13, "grad_norm": 0.19055867195129395, "learning_rate": 3.575951626711874e-05, "loss": 0.007, "step": 74860 }, { "epoch": 5.195149865855342e-13, "grad_norm": 0.18488438427448273, "learning_rate": 3.574390724830312e-05, "loss": 0.0072, "step": 74870 }, { "epoch": 5.195843755245733e-13, "grad_norm": 0.1892748773097992, "learning_rate": 3.572830303440347e-05, "loss": 0.007, "step": 74880 }, { "epoch": 5.196537644636123e-13, "grad_norm": 0.22693641483783722, "learning_rate": 3.571270362702348e-05, "loss": 0.0073, "step": 74890 }, { "epoch": 5.197231534026514e-13, "grad_norm": 0.20719462633132935, "learning_rate": 3.569710902776612e-05, "loss": 0.0071, "step": 74900 }, { "epoch": 5.197925423416905e-13, "grad_norm": 0.21221601963043213, "learning_rate": 3.568151923823407e-05, "loss": 0.0069, "step": 74910 }, { "epoch": 5.198619312807296e-13, "grad_norm": 0.19086404144763947, "learning_rate": 3.566593426002933e-05, "loss": 0.0067, "step": 74920 }, { "epoch": 5.199313202197686e-13, "grad_norm": 0.17978344857692719, "learning_rate": 3.5650354094753606e-05, "loss": 0.0068, "step": 74930 }, { "epoch": 5.200007091588077e-13, "grad_norm": 0.14326868951320648, "learning_rate": 3.56347787440079e-05, "loss": 0.0067, "step": 74940 }, { "epoch": 5.200700980978468e-13, "grad_norm": 0.19859656691551208, "learning_rate": 3.561920820939291e-05, "loss": 0.0067, "step": 74950 }, { "epoch": 5.201394870368858e-13, "grad_norm": 0.20172829926013947, "learning_rate": 3.560364249250869e-05, "loss": 0.0066, "step": 74960 }, { "epoch": 5.202088759759249e-13, "grad_norm": 0.19477315247058868, "learning_rate": 3.5588081594954885e-05, "loss": 0.0068, "step": 74970 }, { "epoch": 5.20278264914964e-13, "grad_norm": 0.20056748390197754, "learning_rate": 3.5572525518330605e-05, "loss": 0.007, "step": 74980 }, { "epoch": 5.203476538540031e-13, "grad_norm": 0.1915747970342636, "learning_rate": 3.555697426423449e-05, "loss": 0.0072, "step": 74990 }, { "epoch": 5.204170427930421e-13, "grad_norm": 0.19083908200263977, "learning_rate": 3.554142783426467e-05, "loss": 0.007, "step": 75000 }, { "epoch": 5.204864317320812e-13, "grad_norm": 0.18540795147418976, "learning_rate": 3.552588623001878e-05, "loss": 0.0069, "step": 75010 }, { "epoch": 5.205558206711203e-13, "grad_norm": 0.17485012114048004, "learning_rate": 3.5510349453093975e-05, "loss": 0.0071, "step": 75020 }, { "epoch": 5.206252096101593e-13, "grad_norm": 0.21701110899448395, "learning_rate": 3.549481750508688e-05, "loss": 0.0068, "step": 75030 }, { "epoch": 5.206945985491984e-13, "grad_norm": 0.1950228214263916, "learning_rate": 3.5479290387593654e-05, "loss": 0.0072, "step": 75040 }, { "epoch": 5.207639874882375e-13, "grad_norm": 0.17314890027046204, "learning_rate": 3.546376810220996e-05, "loss": 0.0068, "step": 75050 }, { "epoch": 5.208333764272766e-13, "grad_norm": 0.1535077691078186, "learning_rate": 3.544825065053096e-05, "loss": 0.0069, "step": 75060 }, { "epoch": 5.209027653663156e-13, "grad_norm": 0.1592196375131607, "learning_rate": 3.543273803415128e-05, "loss": 0.0066, "step": 75070 }, { "epoch": 5.209721543053547e-13, "grad_norm": 0.17350710928440094, "learning_rate": 3.54172302546651e-05, "loss": 0.0066, "step": 75080 }, { "epoch": 5.210415432443938e-13, "grad_norm": 0.18523728847503662, "learning_rate": 3.540172731366609e-05, "loss": 0.0066, "step": 75090 }, { "epoch": 5.211109321834329e-13, "grad_norm": 0.16712692379951477, "learning_rate": 3.538622921274741e-05, "loss": 0.0065, "step": 75100 }, { "epoch": 5.211803211224719e-13, "grad_norm": 0.18357530236244202, "learning_rate": 3.537073595350174e-05, "loss": 0.0066, "step": 75110 }, { "epoch": 5.21249710061511e-13, "grad_norm": 0.17180149257183075, "learning_rate": 3.5355247537521245e-05, "loss": 0.0067, "step": 75120 }, { "epoch": 5.213190990005501e-13, "grad_norm": 0.1380210816860199, "learning_rate": 3.533976396639762e-05, "loss": 0.0064, "step": 75130 }, { "epoch": 5.213884879395891e-13, "grad_norm": 0.17242363095283508, "learning_rate": 3.532428524172201e-05, "loss": 0.0068, "step": 75140 }, { "epoch": 5.214578768786282e-13, "grad_norm": 0.18934492766857147, "learning_rate": 3.5308811365085125e-05, "loss": 0.007, "step": 75150 }, { "epoch": 5.215272658176673e-13, "grad_norm": 0.1926550716161728, "learning_rate": 3.529334233807713e-05, "loss": 0.0069, "step": 75160 }, { "epoch": 5.215966547567064e-13, "grad_norm": 0.18560031056404114, "learning_rate": 3.5277878162287714e-05, "loss": 0.0067, "step": 75170 }, { "epoch": 5.216660436957454e-13, "grad_norm": 0.19868001341819763, "learning_rate": 3.5262418839306054e-05, "loss": 0.007, "step": 75180 }, { "epoch": 5.217354326347845e-13, "grad_norm": 0.1491418033838272, "learning_rate": 3.5246964370720854e-05, "loss": 0.0067, "step": 75190 }, { "epoch": 5.218048215738236e-13, "grad_norm": 0.18742415308952332, "learning_rate": 3.523151475812031e-05, "loss": 0.0072, "step": 75200 }, { "epoch": 5.218742105128626e-13, "grad_norm": 0.21175961196422577, "learning_rate": 3.521607000309204e-05, "loss": 0.0069, "step": 75210 }, { "epoch": 5.219435994519017e-13, "grad_norm": 0.1800520420074463, "learning_rate": 3.520063010722333e-05, "loss": 0.0066, "step": 75220 }, { "epoch": 5.220129883909408e-13, "grad_norm": 0.16827885806560516, "learning_rate": 3.5185195072100775e-05, "loss": 0.0066, "step": 75230 }, { "epoch": 5.220823773299799e-13, "grad_norm": 0.16498883068561554, "learning_rate": 3.516976489931066e-05, "loss": 0.0068, "step": 75240 }, { "epoch": 5.221517662690189e-13, "grad_norm": 0.17438602447509766, "learning_rate": 3.515433959043858e-05, "loss": 0.0065, "step": 75250 }, { "epoch": 5.22221155208058e-13, "grad_norm": 0.1901312917470932, "learning_rate": 3.5138919147069824e-05, "loss": 0.0066, "step": 75260 }, { "epoch": 5.222905441470971e-13, "grad_norm": 0.16317667067050934, "learning_rate": 3.5123503570788986e-05, "loss": 0.0066, "step": 75270 }, { "epoch": 5.223599330861362e-13, "grad_norm": 0.15013918280601501, "learning_rate": 3.510809286318034e-05, "loss": 0.0065, "step": 75280 }, { "epoch": 5.224293220251752e-13, "grad_norm": 0.2010171264410019, "learning_rate": 3.509268702582752e-05, "loss": 0.0064, "step": 75290 }, { "epoch": 5.224987109642143e-13, "grad_norm": 0.1715899407863617, "learning_rate": 3.507728606031373e-05, "loss": 0.0068, "step": 75300 }, { "epoch": 5.225680999032534e-13, "grad_norm": 0.20052629709243774, "learning_rate": 3.5061889968221665e-05, "loss": 0.0067, "step": 75310 }, { "epoch": 5.226374888422924e-13, "grad_norm": 0.18614766001701355, "learning_rate": 3.504649875113351e-05, "loss": 0.007, "step": 75320 }, { "epoch": 5.227068777813315e-13, "grad_norm": 0.20240731537342072, "learning_rate": 3.503111241063095e-05, "loss": 0.007, "step": 75330 }, { "epoch": 5.227762667203706e-13, "grad_norm": 0.159679114818573, "learning_rate": 3.501573094829518e-05, "loss": 0.0066, "step": 75340 }, { "epoch": 5.228456556594097e-13, "grad_norm": 0.19748616218566895, "learning_rate": 3.5000354365706873e-05, "loss": 0.0068, "step": 75350 }, { "epoch": 5.229150445984487e-13, "grad_norm": 0.20836812257766724, "learning_rate": 3.4984982664446216e-05, "loss": 0.0074, "step": 75360 }, { "epoch": 5.229844335374878e-13, "grad_norm": 0.18790820240974426, "learning_rate": 3.496961584609289e-05, "loss": 0.0067, "step": 75370 }, { "epoch": 5.230538224765269e-13, "grad_norm": 0.18847963213920593, "learning_rate": 3.495425391222608e-05, "loss": 0.0065, "step": 75380 }, { "epoch": 5.231232114155659e-13, "grad_norm": 0.1806817352771759, "learning_rate": 3.493889686442446e-05, "loss": 0.0064, "step": 75390 }, { "epoch": 5.23192600354605e-13, "grad_norm": 0.16785188019275665, "learning_rate": 3.492354470426622e-05, "loss": 0.0064, "step": 75400 }, { "epoch": 5.232619892936441e-13, "grad_norm": 0.18762005865573883, "learning_rate": 3.490819743332899e-05, "loss": 0.0064, "step": 75410 }, { "epoch": 5.233313782326832e-13, "grad_norm": 0.1605347990989685, "learning_rate": 3.489285505318999e-05, "loss": 0.0068, "step": 75420 }, { "epoch": 5.234007671717222e-13, "grad_norm": 0.19140230119228363, "learning_rate": 3.487751756542584e-05, "loss": 0.0064, "step": 75430 }, { "epoch": 5.234701561107613e-13, "grad_norm": 0.16466137766838074, "learning_rate": 3.486218497161278e-05, "loss": 0.0064, "step": 75440 }, { "epoch": 5.235395450498004e-13, "grad_norm": 0.1754085123538971, "learning_rate": 3.484685727332638e-05, "loss": 0.0065, "step": 75450 }, { "epoch": 5.236089339888395e-13, "grad_norm": 0.20250152051448822, "learning_rate": 3.4831534472141884e-05, "loss": 0.0068, "step": 75460 }, { "epoch": 5.236783229278785e-13, "grad_norm": 0.24707435071468353, "learning_rate": 3.481621656963386e-05, "loss": 0.0068, "step": 75470 }, { "epoch": 5.237477118669176e-13, "grad_norm": 0.18363431096076965, "learning_rate": 3.4800903567376565e-05, "loss": 0.0069, "step": 75480 }, { "epoch": 5.238171008059567e-13, "grad_norm": 0.19027625024318695, "learning_rate": 3.478559546694357e-05, "loss": 0.0068, "step": 75490 }, { "epoch": 5.238864897449957e-13, "grad_norm": 0.18414106965065002, "learning_rate": 3.4770292269908044e-05, "loss": 0.0067, "step": 75500 }, { "epoch": 5.239558786840348e-13, "grad_norm": 0.17451874911785126, "learning_rate": 3.4754993977842626e-05, "loss": 0.0067, "step": 75510 }, { "epoch": 5.240252676230739e-13, "grad_norm": 0.15782538056373596, "learning_rate": 3.473970059231946e-05, "loss": 0.0068, "step": 75520 }, { "epoch": 5.24094656562113e-13, "grad_norm": 0.17434963583946228, "learning_rate": 3.472441211491019e-05, "loss": 0.0065, "step": 75530 }, { "epoch": 5.24164045501152e-13, "grad_norm": 0.1778373420238495, "learning_rate": 3.47091285471859e-05, "loss": 0.0063, "step": 75540 }, { "epoch": 5.242334344401911e-13, "grad_norm": 0.17372511327266693, "learning_rate": 3.469384989071728e-05, "loss": 0.0065, "step": 75550 }, { "epoch": 5.243028233792302e-13, "grad_norm": 0.17601752281188965, "learning_rate": 3.4678576147074374e-05, "loss": 0.0066, "step": 75560 }, { "epoch": 5.243722123182692e-13, "grad_norm": 0.1759599894285202, "learning_rate": 3.4663307317826884e-05, "loss": 0.0064, "step": 75570 }, { "epoch": 5.244416012573083e-13, "grad_norm": 0.19550445675849915, "learning_rate": 3.464804340454382e-05, "loss": 0.007, "step": 75580 }, { "epoch": 5.245109901963474e-13, "grad_norm": 0.16042251884937286, "learning_rate": 3.463278440879389e-05, "loss": 0.0064, "step": 75590 }, { "epoch": 5.245803791353865e-13, "grad_norm": 0.1794966161251068, "learning_rate": 3.461753033214509e-05, "loss": 0.0062, "step": 75600 }, { "epoch": 5.246497680744255e-13, "grad_norm": 0.19530507922172546, "learning_rate": 3.460228117616512e-05, "loss": 0.0064, "step": 75610 }, { "epoch": 5.247191570134646e-13, "grad_norm": 0.18589842319488525, "learning_rate": 3.4587036942421e-05, "loss": 0.0066, "step": 75620 }, { "epoch": 5.247885459525037e-13, "grad_norm": 0.15874838829040527, "learning_rate": 3.457179763247931e-05, "loss": 0.0066, "step": 75630 }, { "epoch": 5.248579348915428e-13, "grad_norm": 0.1736285537481308, "learning_rate": 3.455656324790615e-05, "loss": 0.0066, "step": 75640 }, { "epoch": 5.249273238305818e-13, "grad_norm": 0.17147506773471832, "learning_rate": 3.4541333790267094e-05, "loss": 0.0066, "step": 75650 }, { "epoch": 5.249967127696209e-13, "grad_norm": 0.15973897278308868, "learning_rate": 3.4526109261127185e-05, "loss": 0.0066, "step": 75660 }, { "epoch": 5.2506610170866e-13, "grad_norm": 0.14329583942890167, "learning_rate": 3.4510889662050986e-05, "loss": 0.0067, "step": 75670 }, { "epoch": 5.25135490647699e-13, "grad_norm": 0.16283445060253143, "learning_rate": 3.449567499460256e-05, "loss": 0.0067, "step": 75680 }, { "epoch": 5.252048795867381e-13, "grad_norm": 0.16452349722385406, "learning_rate": 3.4480465260345436e-05, "loss": 0.0065, "step": 75690 }, { "epoch": 5.252742685257772e-13, "grad_norm": 0.19350019097328186, "learning_rate": 3.446526046084266e-05, "loss": 0.0064, "step": 75700 }, { "epoch": 5.253436574648163e-13, "grad_norm": 0.16067607700824738, "learning_rate": 3.4450060597656766e-05, "loss": 0.0063, "step": 75710 }, { "epoch": 5.254130464038553e-13, "grad_norm": 0.17967097461223602, "learning_rate": 3.4434865672349765e-05, "loss": 0.0065, "step": 75720 }, { "epoch": 5.254824353428944e-13, "grad_norm": 0.17389637231826782, "learning_rate": 3.44196756864832e-05, "loss": 0.0066, "step": 75730 }, { "epoch": 5.255518242819335e-13, "grad_norm": 0.16728389263153076, "learning_rate": 3.440449064161801e-05, "loss": 0.0065, "step": 75740 }, { "epoch": 5.256212132209725e-13, "grad_norm": 0.18127617239952087, "learning_rate": 3.438931053931478e-05, "loss": 0.0063, "step": 75750 }, { "epoch": 5.256906021600116e-13, "grad_norm": 0.17192843556404114, "learning_rate": 3.437413538113342e-05, "loss": 0.0062, "step": 75760 }, { "epoch": 5.257599910990507e-13, "grad_norm": 0.16841988265514374, "learning_rate": 3.43589651686335e-05, "loss": 0.0065, "step": 75770 }, { "epoch": 5.258293800380898e-13, "grad_norm": 0.17637452483177185, "learning_rate": 3.4343799903373905e-05, "loss": 0.0065, "step": 75780 }, { "epoch": 5.258987689771288e-13, "grad_norm": 0.16758863627910614, "learning_rate": 3.432863958691319e-05, "loss": 0.0065, "step": 75790 }, { "epoch": 5.259681579161679e-13, "grad_norm": 0.17364080250263214, "learning_rate": 3.431348422080922e-05, "loss": 0.0067, "step": 75800 }, { "epoch": 5.26037546855207e-13, "grad_norm": 0.1740828901529312, "learning_rate": 3.4298333806619555e-05, "loss": 0.0063, "step": 75810 }, { "epoch": 5.261069357942461e-13, "grad_norm": 0.16107752919197083, "learning_rate": 3.4283188345901054e-05, "loss": 0.0065, "step": 75820 }, { "epoch": 5.261763247332851e-13, "grad_norm": 0.18516209721565247, "learning_rate": 3.4268047840210164e-05, "loss": 0.0066, "step": 75830 }, { "epoch": 5.262457136723242e-13, "grad_norm": 0.16196729242801666, "learning_rate": 3.425291229110281e-05, "loss": 0.0065, "step": 75840 }, { "epoch": 5.263151026113633e-13, "grad_norm": 0.15088732540607452, "learning_rate": 3.4237781700134416e-05, "loss": 0.0067, "step": 75850 }, { "epoch": 5.263844915504023e-13, "grad_norm": 0.19316397607326508, "learning_rate": 3.4222656068859875e-05, "loss": 0.0062, "step": 75860 }, { "epoch": 5.264538804894414e-13, "grad_norm": 0.1957567185163498, "learning_rate": 3.42075353988336e-05, "loss": 0.0063, "step": 75870 }, { "epoch": 5.265232694284805e-13, "grad_norm": 0.16444151103496552, "learning_rate": 3.419241969160944e-05, "loss": 0.0061, "step": 75880 }, { "epoch": 5.265926583675196e-13, "grad_norm": 0.1495315581560135, "learning_rate": 3.41773089487408e-05, "loss": 0.0063, "step": 75890 }, { "epoch": 5.266620473065586e-13, "grad_norm": 0.18387579917907715, "learning_rate": 3.416220317178053e-05, "loss": 0.0064, "step": 75900 }, { "epoch": 5.267314362455977e-13, "grad_norm": 0.1725621521472931, "learning_rate": 3.414710236228099e-05, "loss": 0.0061, "step": 75910 }, { "epoch": 5.268008251846368e-13, "grad_norm": 0.16245174407958984, "learning_rate": 3.413200652179402e-05, "loss": 0.0063, "step": 75920 }, { "epoch": 5.268702141236759e-13, "grad_norm": 0.21448175609111786, "learning_rate": 3.4116915651870945e-05, "loss": 0.0065, "step": 75930 }, { "epoch": 5.269396030627149e-13, "grad_norm": 0.17363010346889496, "learning_rate": 3.410182975406261e-05, "loss": 0.0065, "step": 75940 }, { "epoch": 5.27008992001754e-13, "grad_norm": 0.18083254992961884, "learning_rate": 3.408674882991932e-05, "loss": 0.0064, "step": 75950 }, { "epoch": 5.270783809407931e-13, "grad_norm": 0.1793965995311737, "learning_rate": 3.407167288099085e-05, "loss": 0.0067, "step": 75960 }, { "epoch": 5.271477698798321e-13, "grad_norm": 0.19917713105678558, "learning_rate": 3.40566019088265e-05, "loss": 0.0066, "step": 75970 }, { "epoch": 5.272171588188712e-13, "grad_norm": 0.17608699202537537, "learning_rate": 3.404153591497505e-05, "loss": 0.0065, "step": 75980 }, { "epoch": 5.272865477579103e-13, "grad_norm": 0.17789523303508759, "learning_rate": 3.4026474900984764e-05, "loss": 0.0069, "step": 75990 }, { "epoch": 5.273559366969494e-13, "grad_norm": 0.1539900153875351, "learning_rate": 3.401141886840339e-05, "loss": 0.0067, "step": 76000 }, { "epoch": 5.274253256359884e-13, "grad_norm": 0.19766780734062195, "learning_rate": 3.3996367818778186e-05, "loss": 0.0062, "step": 76010 }, { "epoch": 5.274947145750275e-13, "grad_norm": 0.20373517274856567, "learning_rate": 3.398132175365587e-05, "loss": 0.0062, "step": 76020 }, { "epoch": 5.275641035140666e-13, "grad_norm": 0.20679517090320587, "learning_rate": 3.396628067458266e-05, "loss": 0.0063, "step": 76030 }, { "epoch": 5.276334924531056e-13, "grad_norm": 0.23342905938625336, "learning_rate": 3.3951244583104265e-05, "loss": 0.0062, "step": 76040 }, { "epoch": 5.277028813921447e-13, "grad_norm": 0.16737349331378937, "learning_rate": 3.393621348076587e-05, "loss": 0.0063, "step": 76050 }, { "epoch": 5.277722703311838e-13, "grad_norm": 0.1720198392868042, "learning_rate": 3.3921187369112185e-05, "loss": 0.0062, "step": 76060 }, { "epoch": 5.278416592702229e-13, "grad_norm": 0.17577245831489563, "learning_rate": 3.3906166249687306e-05, "loss": 0.0062, "step": 76070 }, { "epoch": 5.279110482092619e-13, "grad_norm": 0.16725018620491028, "learning_rate": 3.389115012403497e-05, "loss": 0.0064, "step": 76080 }, { "epoch": 5.27980437148301e-13, "grad_norm": 0.1700775921344757, "learning_rate": 3.3876138993698243e-05, "loss": 0.0064, "step": 76090 }, { "epoch": 5.280498260873401e-13, "grad_norm": 0.1697947084903717, "learning_rate": 3.386113286021983e-05, "loss": 0.007, "step": 76100 }, { "epoch": 5.281192150263792e-13, "grad_norm": 0.230595663189888, "learning_rate": 3.384613172514176e-05, "loss": 0.0067, "step": 76110 }, { "epoch": 5.281886039654182e-13, "grad_norm": 0.17733590304851532, "learning_rate": 3.383113559000572e-05, "loss": 0.0065, "step": 76120 }, { "epoch": 5.282579929044573e-13, "grad_norm": 0.19560734927654266, "learning_rate": 3.3816144456352705e-05, "loss": 0.0068, "step": 76130 }, { "epoch": 5.283273818434964e-13, "grad_norm": 0.18832233548164368, "learning_rate": 3.380115832572338e-05, "loss": 0.0064, "step": 76140 }, { "epoch": 5.283967707825354e-13, "grad_norm": 0.17657609283924103, "learning_rate": 3.378617719965773e-05, "loss": 0.0067, "step": 76150 }, { "epoch": 5.284661597215745e-13, "grad_norm": 0.2098413109779358, "learning_rate": 3.377120107969533e-05, "loss": 0.0065, "step": 76160 }, { "epoch": 5.285355486606136e-13, "grad_norm": 0.16905559599399567, "learning_rate": 3.3756229967375214e-05, "loss": 0.0062, "step": 76170 }, { "epoch": 5.286049375996527e-13, "grad_norm": 0.13544192910194397, "learning_rate": 3.374126386423588e-05, "loss": 0.0061, "step": 76180 }, { "epoch": 5.286743265386917e-13, "grad_norm": 0.14285337924957275, "learning_rate": 3.372630277181534e-05, "loss": 0.0061, "step": 76190 }, { "epoch": 5.287437154777308e-13, "grad_norm": 0.15661348402500153, "learning_rate": 3.371134669165108e-05, "loss": 0.0064, "step": 76200 }, { "epoch": 5.288131044167699e-13, "grad_norm": 0.17445974051952362, "learning_rate": 3.369639562528006e-05, "loss": 0.0062, "step": 76210 }, { "epoch": 5.288824933558089e-13, "grad_norm": 0.16794759035110474, "learning_rate": 3.368144957423873e-05, "loss": 0.0061, "step": 76220 }, { "epoch": 5.28951882294848e-13, "grad_norm": 0.18667352199554443, "learning_rate": 3.366650854006305e-05, "loss": 0.0062, "step": 76230 }, { "epoch": 5.290212712338871e-13, "grad_norm": 0.16370762884616852, "learning_rate": 3.365157252428843e-05, "loss": 0.0065, "step": 76240 }, { "epoch": 5.290906601729262e-13, "grad_norm": 0.2175005078315735, "learning_rate": 3.363664152844979e-05, "loss": 0.0066, "step": 76250 }, { "epoch": 5.291600491119652e-13, "grad_norm": 0.195443257689476, "learning_rate": 3.362171555408151e-05, "loss": 0.0068, "step": 76260 }, { "epoch": 5.292294380510043e-13, "grad_norm": 0.22014324367046356, "learning_rate": 3.360679460271746e-05, "loss": 0.0068, "step": 76270 }, { "epoch": 5.292988269900434e-13, "grad_norm": 0.18412433564662933, "learning_rate": 3.359187867589104e-05, "loss": 0.0066, "step": 76280 }, { "epoch": 5.293682159290825e-13, "grad_norm": 0.14072763919830322, "learning_rate": 3.357696777513502e-05, "loss": 0.0067, "step": 76290 }, { "epoch": 5.294376048681215e-13, "grad_norm": 0.15405842661857605, "learning_rate": 3.35620619019818e-05, "loss": 0.0065, "step": 76300 }, { "epoch": 5.295069938071606e-13, "grad_norm": 0.15109097957611084, "learning_rate": 3.354716105796314e-05, "loss": 0.0071, "step": 76310 }, { "epoch": 5.295763827461997e-13, "grad_norm": 0.1522584855556488, "learning_rate": 3.353226524461038e-05, "loss": 0.0067, "step": 76320 }, { "epoch": 5.296457716852387e-13, "grad_norm": 0.17508043348789215, "learning_rate": 3.3517374463454234e-05, "loss": 0.0064, "step": 76330 }, { "epoch": 5.297151606242778e-13, "grad_norm": 0.17437057197093964, "learning_rate": 3.350248871602504e-05, "loss": 0.0062, "step": 76340 }, { "epoch": 5.297845495633169e-13, "grad_norm": 0.18883833289146423, "learning_rate": 3.348760800385247e-05, "loss": 0.0062, "step": 76350 }, { "epoch": 5.29853938502356e-13, "grad_norm": 0.15941400825977325, "learning_rate": 3.347273232846579e-05, "loss": 0.0062, "step": 76360 }, { "epoch": 5.29923327441395e-13, "grad_norm": 0.17699921131134033, "learning_rate": 3.3457861691393676e-05, "loss": 0.0063, "step": 76370 }, { "epoch": 5.299927163804341e-13, "grad_norm": 0.12789027392864227, "learning_rate": 3.344299609416435e-05, "loss": 0.006, "step": 76380 }, { "epoch": 5.300621053194732e-13, "grad_norm": 0.14548760652542114, "learning_rate": 3.342813553830546e-05, "loss": 0.0063, "step": 76390 }, { "epoch": 5.301314942585122e-13, "grad_norm": 0.19076035916805267, "learning_rate": 3.341328002534418e-05, "loss": 0.0063, "step": 76400 }, { "epoch": 5.302008831975513e-13, "grad_norm": 0.17172430455684662, "learning_rate": 3.3398429556807145e-05, "loss": 0.0068, "step": 76410 }, { "epoch": 5.302702721365904e-13, "grad_norm": 0.17817915976047516, "learning_rate": 3.3383584134220413e-05, "loss": 0.0065, "step": 76420 }, { "epoch": 5.303396610756295e-13, "grad_norm": 0.18615256249904633, "learning_rate": 3.336874375910968e-05, "loss": 0.0068, "step": 76430 }, { "epoch": 5.304090500146685e-13, "grad_norm": 0.15390515327453613, "learning_rate": 3.335390843299993e-05, "loss": 0.0066, "step": 76440 }, { "epoch": 5.304784389537076e-13, "grad_norm": 0.16073596477508545, "learning_rate": 3.333907815741581e-05, "loss": 0.0065, "step": 76450 }, { "epoch": 5.305478278927467e-13, "grad_norm": 0.19758571684360504, "learning_rate": 3.332425293388128e-05, "loss": 0.0068, "step": 76460 }, { "epoch": 5.306172168317858e-13, "grad_norm": 0.14558617770671844, "learning_rate": 3.330943276391994e-05, "loss": 0.0069, "step": 76470 }, { "epoch": 5.306866057708248e-13, "grad_norm": 0.1735854148864746, "learning_rate": 3.329461764905473e-05, "loss": 0.0067, "step": 76480 }, { "epoch": 5.307559947098639e-13, "grad_norm": 0.21518807113170624, "learning_rate": 3.327980759080817e-05, "loss": 0.0065, "step": 76490 }, { "epoch": 5.30825383648903e-13, "grad_norm": 0.16428501904010773, "learning_rate": 3.32650025907022e-05, "loss": 0.0063, "step": 76500 }, { "epoch": 5.30894772587942e-13, "grad_norm": 0.18427440524101257, "learning_rate": 3.3250202650258284e-05, "loss": 0.0062, "step": 76510 }, { "epoch": 5.309641615269811e-13, "grad_norm": 0.1494743525981903, "learning_rate": 3.323540777099734e-05, "loss": 0.0065, "step": 76520 }, { "epoch": 5.310335504660202e-13, "grad_norm": 0.17919163405895233, "learning_rate": 3.3220617954439766e-05, "loss": 0.0062, "step": 76530 }, { "epoch": 5.311029394050593e-13, "grad_norm": 0.18188366293907166, "learning_rate": 3.320583320210545e-05, "loss": 0.0063, "step": 76540 }, { "epoch": 5.311723283440983e-13, "grad_norm": 0.19927601516246796, "learning_rate": 3.3191053515513765e-05, "loss": 0.0065, "step": 76550 }, { "epoch": 5.312417172831374e-13, "grad_norm": 0.161680668592453, "learning_rate": 3.3176278896183534e-05, "loss": 0.0067, "step": 76560 }, { "epoch": 5.313111062221765e-13, "grad_norm": 0.1683005690574646, "learning_rate": 3.31615093456331e-05, "loss": 0.0066, "step": 76570 }, { "epoch": 5.313804951612155e-13, "grad_norm": 0.1739349365234375, "learning_rate": 3.314674486538024e-05, "loss": 0.0068, "step": 76580 }, { "epoch": 5.314498841002546e-13, "grad_norm": 0.1946493685245514, "learning_rate": 3.313198545694226e-05, "loss": 0.0068, "step": 76590 }, { "epoch": 5.315192730392937e-13, "grad_norm": 0.17588892579078674, "learning_rate": 3.3117231121835905e-05, "loss": 0.0069, "step": 76600 }, { "epoch": 5.315886619783328e-13, "grad_norm": 0.17892630398273468, "learning_rate": 3.3102481861577436e-05, "loss": 0.0067, "step": 76610 }, { "epoch": 5.316580509173718e-13, "grad_norm": 0.19039775431156158, "learning_rate": 3.30877376776825e-05, "loss": 0.0071, "step": 76620 }, { "epoch": 5.317274398564109e-13, "grad_norm": 0.16355609893798828, "learning_rate": 3.3072998571666386e-05, "loss": 0.0068, "step": 76630 }, { "epoch": 5.3179682879545e-13, "grad_norm": 0.19808095693588257, "learning_rate": 3.3058264545043675e-05, "loss": 0.0065, "step": 76640 }, { "epoch": 5.318662177344891e-13, "grad_norm": 0.178902730345726, "learning_rate": 3.3043535599328596e-05, "loss": 0.0067, "step": 76650 }, { "epoch": 5.319356066735281e-13, "grad_norm": 0.16137023270130157, "learning_rate": 3.302881173603471e-05, "loss": 0.0062, "step": 76660 }, { "epoch": 5.320049956125672e-13, "grad_norm": 0.14341214299201965, "learning_rate": 3.3014092956675194e-05, "loss": 0.0065, "step": 76670 }, { "epoch": 5.320743845516063e-13, "grad_norm": 0.19370388984680176, "learning_rate": 3.2999379262762573e-05, "loss": 0.0063, "step": 76680 }, { "epoch": 5.321437734906453e-13, "grad_norm": 0.1857534497976303, "learning_rate": 3.298467065580893e-05, "loss": 0.0061, "step": 76690 }, { "epoch": 5.322131624296844e-13, "grad_norm": 0.14290709793567657, "learning_rate": 3.29699671373258e-05, "loss": 0.0064, "step": 76700 }, { "epoch": 5.322825513687235e-13, "grad_norm": 0.16607899963855743, "learning_rate": 3.2955268708824206e-05, "loss": 0.0064, "step": 76710 }, { "epoch": 5.323519403077626e-13, "grad_norm": 0.17440703511238098, "learning_rate": 3.2940575371814625e-05, "loss": 0.0071, "step": 76720 }, { "epoch": 5.324213292468016e-13, "grad_norm": 0.17966048419475555, "learning_rate": 3.292588712780704e-05, "loss": 0.0068, "step": 76730 }, { "epoch": 5.324907181858407e-13, "grad_norm": 0.14881783723831177, "learning_rate": 3.291120397831088e-05, "loss": 0.0069, "step": 76740 }, { "epoch": 5.325601071248798e-13, "grad_norm": 0.18126246333122253, "learning_rate": 3.28965259248351e-05, "loss": 0.007, "step": 76750 }, { "epoch": 5.326294960639189e-13, "grad_norm": 0.15592584013938904, "learning_rate": 3.2881852968888066e-05, "loss": 0.0067, "step": 76760 }, { "epoch": 5.326988850029579e-13, "grad_norm": 0.15491165220737457, "learning_rate": 3.286718511197767e-05, "loss": 0.0069, "step": 76770 }, { "epoch": 5.32768273941997e-13, "grad_norm": 0.21135807037353516, "learning_rate": 3.285252235561125e-05, "loss": 0.0074, "step": 76780 }, { "epoch": 5.328376628810361e-13, "grad_norm": 0.17348268628120422, "learning_rate": 3.283786470129564e-05, "loss": 0.0066, "step": 76790 }, { "epoch": 5.329070518200751e-13, "grad_norm": 0.15421819686889648, "learning_rate": 3.2823212150537135e-05, "loss": 0.0064, "step": 76800 }, { "epoch": 5.329764407591142e-13, "grad_norm": 0.18576441705226898, "learning_rate": 3.2808564704841544e-05, "loss": 0.0065, "step": 76810 }, { "epoch": 5.330458296981533e-13, "grad_norm": 0.14488346874713898, "learning_rate": 3.279392236571406e-05, "loss": 0.0064, "step": 76820 }, { "epoch": 5.331152186371924e-13, "grad_norm": 0.19726620614528656, "learning_rate": 3.2779285134659474e-05, "loss": 0.0065, "step": 76830 }, { "epoch": 5.331846075762314e-13, "grad_norm": 0.1509862095117569, "learning_rate": 3.276465301318193e-05, "loss": 0.0066, "step": 76840 }, { "epoch": 5.332539965152705e-13, "grad_norm": 0.17503081262111664, "learning_rate": 3.2750026002785174e-05, "loss": 0.0062, "step": 76850 }, { "epoch": 5.333233854543096e-13, "grad_norm": 0.1602964848279953, "learning_rate": 3.273540410497228e-05, "loss": 0.0062, "step": 76860 }, { "epoch": 5.333927743933486e-13, "grad_norm": 0.15931060910224915, "learning_rate": 3.272078732124596e-05, "loss": 0.0064, "step": 76870 }, { "epoch": 5.334621633323877e-13, "grad_norm": 0.18053410947322845, "learning_rate": 3.270617565310825e-05, "loss": 0.0068, "step": 76880 }, { "epoch": 5.335315522714268e-13, "grad_norm": 0.1692662239074707, "learning_rate": 3.2691569102060754e-05, "loss": 0.0065, "step": 76890 }, { "epoch": 5.336009412104659e-13, "grad_norm": 0.16268318891525269, "learning_rate": 3.267696766960451e-05, "loss": 0.0066, "step": 76900 }, { "epoch": 5.336703301495049e-13, "grad_norm": 0.1452028751373291, "learning_rate": 3.266237135724005e-05, "loss": 0.0071, "step": 76910 }, { "epoch": 5.33739719088544e-13, "grad_norm": 0.22069256007671356, "learning_rate": 3.264778016646738e-05, "loss": 0.0068, "step": 76920 }, { "epoch": 5.338091080275831e-13, "grad_norm": 0.1545569896697998, "learning_rate": 3.2633194098785964e-05, "loss": 0.007, "step": 76930 }, { "epoch": 5.338784969666222e-13, "grad_norm": 0.15866132080554962, "learning_rate": 3.261861315569476e-05, "loss": 0.0069, "step": 76940 }, { "epoch": 5.339478859056612e-13, "grad_norm": 0.17892518639564514, "learning_rate": 3.2604037338692144e-05, "loss": 0.0064, "step": 76950 }, { "epoch": 5.340172748447003e-13, "grad_norm": 0.15414804220199585, "learning_rate": 3.2589466649276074e-05, "loss": 0.0063, "step": 76960 }, { "epoch": 5.340866637837394e-13, "grad_norm": 0.13815440237522125, "learning_rate": 3.2574901088943835e-05, "loss": 0.0059, "step": 76970 }, { "epoch": 5.341560527227784e-13, "grad_norm": 0.16457681357860565, "learning_rate": 3.256034065919236e-05, "loss": 0.0062, "step": 76980 }, { "epoch": 5.342254416618175e-13, "grad_norm": 0.1626458466053009, "learning_rate": 3.2545785361517865e-05, "loss": 0.0065, "step": 76990 }, { "epoch": 5.342948306008566e-13, "grad_norm": 0.13703173398971558, "learning_rate": 3.2531235197416214e-05, "loss": 0.0062, "step": 77000 }, { "epoch": 5.343642195398957e-13, "grad_norm": 0.1751158982515335, "learning_rate": 3.251669016838259e-05, "loss": 0.0062, "step": 77010 }, { "epoch": 5.344336084789347e-13, "grad_norm": 0.1497434675693512, "learning_rate": 3.2502150275911786e-05, "loss": 0.0062, "step": 77020 }, { "epoch": 5.345029974179738e-13, "grad_norm": 0.18154002726078033, "learning_rate": 3.248761552149797e-05, "loss": 0.0065, "step": 77030 }, { "epoch": 5.345723863570129e-13, "grad_norm": 0.16300669312477112, "learning_rate": 3.247308590663481e-05, "loss": 0.0066, "step": 77040 }, { "epoch": 5.346417752960519e-13, "grad_norm": 0.1877572238445282, "learning_rate": 3.245856143281545e-05, "loss": 0.0066, "step": 77050 }, { "epoch": 5.34711164235091e-13, "grad_norm": 0.1532183587551117, "learning_rate": 3.244404210153252e-05, "loss": 0.007, "step": 77060 }, { "epoch": 5.347805531741301e-13, "grad_norm": 0.15027067065238953, "learning_rate": 3.2429527914278106e-05, "loss": 0.0065, "step": 77070 }, { "epoch": 5.348499421131692e-13, "grad_norm": 0.1724022477865219, "learning_rate": 3.241501887254376e-05, "loss": 0.0066, "step": 77080 }, { "epoch": 5.349193310522082e-13, "grad_norm": 0.15191486477851868, "learning_rate": 3.240051497782051e-05, "loss": 0.0079, "step": 77090 }, { "epoch": 5.349887199912473e-13, "grad_norm": 0.1453164517879486, "learning_rate": 3.238601623159885e-05, "loss": 0.0066, "step": 77100 }, { "epoch": 5.350581089302864e-13, "grad_norm": 0.18567898869514465, "learning_rate": 3.237152263536878e-05, "loss": 0.0064, "step": 77110 }, { "epoch": 5.351274978693255e-13, "grad_norm": 0.19104741513729095, "learning_rate": 3.2357034190619705e-05, "loss": 0.0062, "step": 77120 }, { "epoch": 5.351968868083645e-13, "grad_norm": 0.19021901488304138, "learning_rate": 3.2342550898840566e-05, "loss": 0.0059, "step": 77130 }, { "epoch": 5.352662757474036e-13, "grad_norm": 0.17173586785793304, "learning_rate": 3.2328072761519763e-05, "loss": 0.0069, "step": 77140 }, { "epoch": 5.353356646864427e-13, "grad_norm": 0.19916631281375885, "learning_rate": 3.231359978014507e-05, "loss": 0.0064, "step": 77150 }, { "epoch": 5.354050536254817e-13, "grad_norm": 0.17383436858654022, "learning_rate": 3.229913195620392e-05, "loss": 0.0062, "step": 77160 }, { "epoch": 5.354744425645208e-13, "grad_norm": 0.18069271743297577, "learning_rate": 3.2284669291183006e-05, "loss": 0.0061, "step": 77170 }, { "epoch": 5.355438315035599e-13, "grad_norm": 0.1462547332048416, "learning_rate": 3.2270211786568684e-05, "loss": 0.0063, "step": 77180 }, { "epoch": 5.35613220442599e-13, "grad_norm": 0.18571825325489044, "learning_rate": 3.22557594438466e-05, "loss": 0.0065, "step": 77190 }, { "epoch": 5.35682609381638e-13, "grad_norm": 0.15560677647590637, "learning_rate": 3.224131226450203e-05, "loss": 0.0064, "step": 77200 }, { "epoch": 5.357519983206771e-13, "grad_norm": 0.19905225932598114, "learning_rate": 3.222687025001959e-05, "loss": 0.0064, "step": 77210 }, { "epoch": 5.358213872597162e-13, "grad_norm": 0.15951934456825256, "learning_rate": 3.2212433401883476e-05, "loss": 0.0065, "step": 77220 }, { "epoch": 5.358907761987552e-13, "grad_norm": 0.16682839393615723, "learning_rate": 3.219800172157726e-05, "loss": 0.0063, "step": 77230 }, { "epoch": 5.359601651377943e-13, "grad_norm": 0.17214523255825043, "learning_rate": 3.218357521058402e-05, "loss": 0.0063, "step": 77240 }, { "epoch": 5.360295540768334e-13, "grad_norm": 0.19061759114265442, "learning_rate": 3.216915387038632e-05, "loss": 0.0083, "step": 77250 }, { "epoch": 5.360989430158725e-13, "grad_norm": 0.25089728832244873, "learning_rate": 3.2154737702466175e-05, "loss": 0.0065, "step": 77260 }, { "epoch": 5.361683319549115e-13, "grad_norm": 0.17555007338523865, "learning_rate": 3.214032670830507e-05, "loss": 0.0063, "step": 77270 }, { "epoch": 5.362377208939506e-13, "grad_norm": 0.15403485298156738, "learning_rate": 3.212592088938396e-05, "loss": 0.0062, "step": 77280 }, { "epoch": 5.363071098329897e-13, "grad_norm": 0.1688356250524521, "learning_rate": 3.2111520247183256e-05, "loss": 0.0059, "step": 77290 }, { "epoch": 5.363764987720288e-13, "grad_norm": 0.1444079726934433, "learning_rate": 3.2097124783182865e-05, "loss": 0.0063, "step": 77300 }, { "epoch": 5.364458877110678e-13, "grad_norm": 0.17519159615039825, "learning_rate": 3.208273449886214e-05, "loss": 0.0063, "step": 77310 }, { "epoch": 5.365152766501069e-13, "grad_norm": 0.1769205778837204, "learning_rate": 3.20683493956999e-05, "loss": 0.006, "step": 77320 }, { "epoch": 5.36584665589146e-13, "grad_norm": 0.18472306430339813, "learning_rate": 3.2053969475174455e-05, "loss": 0.006, "step": 77330 }, { "epoch": 5.36654054528185e-13, "grad_norm": 0.15326270461082458, "learning_rate": 3.2039594738763516e-05, "loss": 0.0063, "step": 77340 }, { "epoch": 5.367234434672241e-13, "grad_norm": 0.1401645392179489, "learning_rate": 3.202522518794438e-05, "loss": 0.0066, "step": 77350 }, { "epoch": 5.367928324062632e-13, "grad_norm": 0.1507928967475891, "learning_rate": 3.20108608241937e-05, "loss": 0.0065, "step": 77360 }, { "epoch": 5.368622213453023e-13, "grad_norm": 0.1565588265657425, "learning_rate": 3.1996501648987646e-05, "loss": 0.0062, "step": 77370 }, { "epoch": 5.369316102843413e-13, "grad_norm": 0.17120252549648285, "learning_rate": 3.198214766380184e-05, "loss": 0.0063, "step": 77380 }, { "epoch": 5.370009992233804e-13, "grad_norm": 0.19438248872756958, "learning_rate": 3.196779887011139e-05, "loss": 0.0064, "step": 77390 }, { "epoch": 5.370703881624195e-13, "grad_norm": 0.15528719127178192, "learning_rate": 3.195345526939085e-05, "loss": 0.0078, "step": 77400 }, { "epoch": 5.371397771014585e-13, "grad_norm": 0.1951855570077896, "learning_rate": 3.193911686311426e-05, "loss": 0.0063, "step": 77410 }, { "epoch": 5.372091660404976e-13, "grad_norm": 0.1432962566614151, "learning_rate": 3.192478365275508e-05, "loss": 0.0064, "step": 77420 }, { "epoch": 5.372785549795367e-13, "grad_norm": 0.13812687993049622, "learning_rate": 3.191045563978632e-05, "loss": 0.0065, "step": 77430 }, { "epoch": 5.373479439185758e-13, "grad_norm": 0.17295435070991516, "learning_rate": 3.189613282568037e-05, "loss": 0.0059, "step": 77440 }, { "epoch": 5.374173328576148e-13, "grad_norm": 0.17762474715709686, "learning_rate": 3.188181521190914e-05, "loss": 0.0061, "step": 77450 }, { "epoch": 5.374867217966539e-13, "grad_norm": 0.1673123687505722, "learning_rate": 3.186750279994398e-05, "loss": 0.0061, "step": 77460 }, { "epoch": 5.37556110735693e-13, "grad_norm": 0.17001880705356598, "learning_rate": 3.185319559125573e-05, "loss": 0.0062, "step": 77470 }, { "epoch": 5.376254996747321e-13, "grad_norm": 0.17287209630012512, "learning_rate": 3.1838893587314626e-05, "loss": 0.006, "step": 77480 }, { "epoch": 5.376948886137711e-13, "grad_norm": 0.17887412011623383, "learning_rate": 3.18245967895905e-05, "loss": 0.0063, "step": 77490 }, { "epoch": 5.377642775528102e-13, "grad_norm": 0.14810636639595032, "learning_rate": 3.181030519955249e-05, "loss": 0.0063, "step": 77500 }, { "epoch": 5.378336664918493e-13, "grad_norm": 0.13460537791252136, "learning_rate": 3.1796018818669345e-05, "loss": 0.0066, "step": 77510 }, { "epoch": 5.379030554308883e-13, "grad_norm": 0.17987525463104248, "learning_rate": 3.178173764840916e-05, "loss": 0.0062, "step": 77520 }, { "epoch": 5.379724443699274e-13, "grad_norm": 0.13447873294353485, "learning_rate": 3.1767461690239604e-05, "loss": 0.0063, "step": 77530 }, { "epoch": 5.380418333089665e-13, "grad_norm": 0.18176575005054474, "learning_rate": 3.175319094562768e-05, "loss": 0.0066, "step": 77540 }, { "epoch": 5.381112222480056e-13, "grad_norm": 0.18918980658054352, "learning_rate": 3.173892541604002e-05, "loss": 0.0067, "step": 77550 }, { "epoch": 5.381806111870446e-13, "grad_norm": 0.17445091903209686, "learning_rate": 3.172466510294256e-05, "loss": 0.0071, "step": 77560 }, { "epoch": 5.382500001260837e-13, "grad_norm": 0.1442602276802063, "learning_rate": 3.171041000780079e-05, "loss": 0.0063, "step": 77570 }, { "epoch": 5.383193890651228e-13, "grad_norm": 0.17874810099601746, "learning_rate": 3.169616013207964e-05, "loss": 0.0063, "step": 77580 }, { "epoch": 5.383887780041618e-13, "grad_norm": 0.18675872683525085, "learning_rate": 3.1681915477243516e-05, "loss": 0.0061, "step": 77590 }, { "epoch": 5.384581669432009e-13, "grad_norm": 0.21062807738780975, "learning_rate": 3.166767604475626e-05, "loss": 0.0061, "step": 77600 }, { "epoch": 5.3852755588224e-13, "grad_norm": 0.197170227766037, "learning_rate": 3.1653441836081214e-05, "loss": 0.0063, "step": 77610 }, { "epoch": 5.385969448212791e-13, "grad_norm": 0.1713196188211441, "learning_rate": 3.163921285268116e-05, "loss": 0.0063, "step": 77620 }, { "epoch": 5.386663337603181e-13, "grad_norm": 0.20285247266292572, "learning_rate": 3.162498909601833e-05, "loss": 0.0061, "step": 77630 }, { "epoch": 5.387357226993572e-13, "grad_norm": 0.16317938268184662, "learning_rate": 3.161077056755447e-05, "loss": 0.006, "step": 77640 }, { "epoch": 5.388051116383963e-13, "grad_norm": 0.1644166111946106, "learning_rate": 3.1596557268750724e-05, "loss": 0.0063, "step": 77650 }, { "epoch": 5.388745005774354e-13, "grad_norm": 0.16850364208221436, "learning_rate": 3.158234920106775e-05, "loss": 0.0066, "step": 77660 }, { "epoch": 5.389438895164744e-13, "grad_norm": 0.15933063626289368, "learning_rate": 3.1568146365965626e-05, "loss": 0.0064, "step": 77670 }, { "epoch": 5.390132784555135e-13, "grad_norm": 0.14359235763549805, "learning_rate": 3.1553948764903936e-05, "loss": 0.0061, "step": 77680 }, { "epoch": 5.390826673945526e-13, "grad_norm": 0.16924025118350983, "learning_rate": 3.153975639934172e-05, "loss": 0.0065, "step": 77690 }, { "epoch": 5.391520563335916e-13, "grad_norm": 0.14603079855442047, "learning_rate": 3.1525569270737395e-05, "loss": 0.0066, "step": 77700 }, { "epoch": 5.392214452726307e-13, "grad_norm": 0.42506253719329834, "learning_rate": 3.1511387380548995e-05, "loss": 0.0075, "step": 77710 }, { "epoch": 5.392908342116698e-13, "grad_norm": 0.15285974740982056, "learning_rate": 3.1497210730233854e-05, "loss": 0.0064, "step": 77720 }, { "epoch": 5.393602231507089e-13, "grad_norm": 0.14580371975898743, "learning_rate": 3.148303932124891e-05, "loss": 0.0067, "step": 77730 }, { "epoch": 5.394296120897479e-13, "grad_norm": 0.1659649759531021, "learning_rate": 3.146887315505043e-05, "loss": 0.006, "step": 77740 }, { "epoch": 5.39499001028787e-13, "grad_norm": 0.1520204246044159, "learning_rate": 3.1454712233094273e-05, "loss": 0.0058, "step": 77750 }, { "epoch": 5.395683899678261e-13, "grad_norm": 0.20241142809391022, "learning_rate": 3.1440556556835656e-05, "loss": 0.0067, "step": 77760 }, { "epoch": 5.396377789068652e-13, "grad_norm": 0.1519535332918167, "learning_rate": 3.142640612772931e-05, "loss": 0.0066, "step": 77770 }, { "epoch": 5.397071678459042e-13, "grad_norm": 0.16058658063411713, "learning_rate": 3.1412260947229384e-05, "loss": 0.0062, "step": 77780 }, { "epoch": 5.397765567849433e-13, "grad_norm": 0.1433069407939911, "learning_rate": 3.1398121016789546e-05, "loss": 0.0062, "step": 77790 }, { "epoch": 5.398459457239824e-13, "grad_norm": 0.1688191294670105, "learning_rate": 3.13839863378629e-05, "loss": 0.0062, "step": 77800 }, { "epoch": 5.399153346630214e-13, "grad_norm": 0.20413340628147125, "learning_rate": 3.1369856911901955e-05, "loss": 0.0063, "step": 77810 }, { "epoch": 5.399847236020605e-13, "grad_norm": 0.1596611589193344, "learning_rate": 3.13557327403588e-05, "loss": 0.0066, "step": 77820 }, { "epoch": 5.400541125410996e-13, "grad_norm": 0.1917760968208313, "learning_rate": 3.134161382468483e-05, "loss": 0.0065, "step": 77830 }, { "epoch": 5.401235014801387e-13, "grad_norm": 0.1723620891571045, "learning_rate": 3.1327500166331075e-05, "loss": 0.0061, "step": 77840 }, { "epoch": 5.401928904191777e-13, "grad_norm": 0.1984543800354004, "learning_rate": 3.131339176674785e-05, "loss": 0.0068, "step": 77850 }, { "epoch": 5.402622793582168e-13, "grad_norm": 0.1918564885854721, "learning_rate": 3.129928862738508e-05, "loss": 0.0065, "step": 77860 }, { "epoch": 5.403316682972559e-13, "grad_norm": 0.1614772528409958, "learning_rate": 3.128519074969202e-05, "loss": 0.0073, "step": 77870 }, { "epoch": 5.404010572362949e-13, "grad_norm": 0.16706882417201996, "learning_rate": 3.1271098135117504e-05, "loss": 0.0063, "step": 77880 }, { "epoch": 5.40470446175334e-13, "grad_norm": 0.14971524477005005, "learning_rate": 3.125701078510974e-05, "loss": 0.0061, "step": 77890 }, { "epoch": 5.405398351143731e-13, "grad_norm": 0.13065780699253082, "learning_rate": 3.124292870111641e-05, "loss": 0.0058, "step": 77900 }, { "epoch": 5.406092240534122e-13, "grad_norm": 0.2960812747478485, "learning_rate": 3.122885188458469e-05, "loss": 0.0065, "step": 77910 }, { "epoch": 5.406786129924512e-13, "grad_norm": 0.16824311017990112, "learning_rate": 3.1214780336961175e-05, "loss": 0.006, "step": 77920 }, { "epoch": 5.407480019314903e-13, "grad_norm": 0.17311646044254303, "learning_rate": 3.120071405969195e-05, "loss": 0.0063, "step": 77930 }, { "epoch": 5.408173908705294e-13, "grad_norm": 0.16576138138771057, "learning_rate": 3.118665305422253e-05, "loss": 0.0063, "step": 77940 }, { "epoch": 5.408867798095685e-13, "grad_norm": 0.1713404655456543, "learning_rate": 3.1172597321997913e-05, "loss": 0.0062, "step": 77950 }, { "epoch": 5.409561687486075e-13, "grad_norm": 0.15568459033966064, "learning_rate": 3.115854686446255e-05, "loss": 0.006, "step": 77960 }, { "epoch": 5.410255576876466e-13, "grad_norm": 0.1772022396326065, "learning_rate": 3.114450168306032e-05, "loss": 0.0064, "step": 77970 }, { "epoch": 5.410949466266857e-13, "grad_norm": 0.16217875480651855, "learning_rate": 3.11304617792346e-05, "loss": 0.0065, "step": 77980 }, { "epoch": 5.411643355657247e-13, "grad_norm": 0.1626279354095459, "learning_rate": 3.1116427154428214e-05, "loss": 0.0063, "step": 77990 }, { "epoch": 5.412337245047638e-13, "grad_norm": 0.17078134417533875, "learning_rate": 3.1102397810083426e-05, "loss": 0.0061, "step": 78000 }, { "epoch": 5.413031134438029e-13, "grad_norm": 0.1505424678325653, "learning_rate": 3.108837374764197e-05, "loss": 0.0069, "step": 78010 }, { "epoch": 5.41372502382842e-13, "grad_norm": 0.15265311300754547, "learning_rate": 3.1074354968545064e-05, "loss": 0.0071, "step": 78020 }, { "epoch": 5.41441891321881e-13, "grad_norm": 0.1845298707485199, "learning_rate": 3.10603414742333e-05, "loss": 0.0068, "step": 78030 }, { "epoch": 5.415112802609201e-13, "grad_norm": 0.2039126306772232, "learning_rate": 3.104633326614685e-05, "loss": 0.0065, "step": 78040 }, { "epoch": 5.415806691999592e-13, "grad_norm": 0.12997910380363464, "learning_rate": 3.1032330345725205e-05, "loss": 0.0062, "step": 78050 }, { "epoch": 5.416500581389982e-13, "grad_norm": 0.16651716828346252, "learning_rate": 3.101833271440745e-05, "loss": 0.006, "step": 78060 }, { "epoch": 5.417194470780373e-13, "grad_norm": 0.13874560594558716, "learning_rate": 3.100434037363199e-05, "loss": 0.006, "step": 78070 }, { "epoch": 5.417888360170764e-13, "grad_norm": 0.23870649933815002, "learning_rate": 3.099035332483684e-05, "loss": 0.0062, "step": 78080 }, { "epoch": 5.418582249561155e-13, "grad_norm": 0.15551887452602386, "learning_rate": 3.097637156945931e-05, "loss": 0.0064, "step": 78090 }, { "epoch": 5.419276138951545e-13, "grad_norm": 0.15431281924247742, "learning_rate": 3.096239510893629e-05, "loss": 0.0063, "step": 78100 }, { "epoch": 5.419970028341936e-13, "grad_norm": 0.17375178635120392, "learning_rate": 3.094842394470406e-05, "loss": 0.0062, "step": 78110 }, { "epoch": 5.420663917732327e-13, "grad_norm": 0.14151446521282196, "learning_rate": 3.0934458078198365e-05, "loss": 0.0061, "step": 78120 }, { "epoch": 5.421357807122718e-13, "grad_norm": 0.16765739023685455, "learning_rate": 3.0920497510854435e-05, "loss": 0.0062, "step": 78130 }, { "epoch": 5.422051696513108e-13, "grad_norm": 0.17057353258132935, "learning_rate": 3.090654224410692e-05, "loss": 0.0064, "step": 78140 }, { "epoch": 5.422745585903499e-13, "grad_norm": 0.16055524349212646, "learning_rate": 3.0892592279389956e-05, "loss": 0.006, "step": 78150 }, { "epoch": 5.42343947529389e-13, "grad_norm": 0.1462721973657608, "learning_rate": 3.087864761813711e-05, "loss": 0.0062, "step": 78160 }, { "epoch": 5.42413336468428e-13, "grad_norm": 0.1671866476535797, "learning_rate": 3.086470826178141e-05, "loss": 0.0064, "step": 78170 }, { "epoch": 5.424827254074671e-13, "grad_norm": 0.1433231234550476, "learning_rate": 3.0850774211755345e-05, "loss": 0.0075, "step": 78180 }, { "epoch": 5.425521143465062e-13, "grad_norm": 0.14766955375671387, "learning_rate": 3.0836845469490866e-05, "loss": 0.0062, "step": 78190 }, { "epoch": 5.426215032855453e-13, "grad_norm": 0.1624150425195694, "learning_rate": 3.082292203641935e-05, "loss": 0.0065, "step": 78200 }, { "epoch": 5.426908922245843e-13, "grad_norm": 0.11006657034158707, "learning_rate": 3.080900391397166e-05, "loss": 0.0058, "step": 78210 }, { "epoch": 5.427602811636234e-13, "grad_norm": 0.15111497044563293, "learning_rate": 3.079509110357811e-05, "loss": 0.0059, "step": 78220 }, { "epoch": 5.428296701026625e-13, "grad_norm": 0.1261935830116272, "learning_rate": 3.078118360666842e-05, "loss": 0.0059, "step": 78230 }, { "epoch": 5.428990590417015e-13, "grad_norm": 0.16480869054794312, "learning_rate": 3.076728142467183e-05, "loss": 0.0061, "step": 78240 }, { "epoch": 5.429684479807406e-13, "grad_norm": 0.13423281908035278, "learning_rate": 3.0753384559017e-05, "loss": 0.0064, "step": 78250 }, { "epoch": 5.430378369197797e-13, "grad_norm": 0.1606614887714386, "learning_rate": 3.073949301113204e-05, "loss": 0.006, "step": 78260 }, { "epoch": 5.431072258588188e-13, "grad_norm": 0.15709444880485535, "learning_rate": 3.072560678244452e-05, "loss": 0.0062, "step": 78270 }, { "epoch": 5.431766147978578e-13, "grad_norm": 0.15330635011196136, "learning_rate": 3.07117258743815e-05, "loss": 0.0062, "step": 78280 }, { "epoch": 5.432460037368969e-13, "grad_norm": 0.1446211189031601, "learning_rate": 3.069785028836943e-05, "loss": 0.0062, "step": 78290 }, { "epoch": 5.43315392675936e-13, "grad_norm": 0.13774096965789795, "learning_rate": 3.0683980025834246e-05, "loss": 0.0063, "step": 78300 }, { "epoch": 5.433847816149751e-13, "grad_norm": 0.1723395586013794, "learning_rate": 3.067011508820134e-05, "loss": 0.0059, "step": 78310 }, { "epoch": 5.434541705540141e-13, "grad_norm": 0.15064597129821777, "learning_rate": 3.0656255476895545e-05, "loss": 0.0063, "step": 78320 }, { "epoch": 5.435235594930532e-13, "grad_norm": 0.20087218284606934, "learning_rate": 3.064240119334115e-05, "loss": 0.0063, "step": 78330 }, { "epoch": 5.435929484320923e-13, "grad_norm": 0.15639106929302216, "learning_rate": 3.062855223896191e-05, "loss": 0.0064, "step": 78340 }, { "epoch": 5.436623373711313e-13, "grad_norm": 0.1421820968389511, "learning_rate": 3.0614708615181014e-05, "loss": 0.0059, "step": 78350 }, { "epoch": 5.437317263101704e-13, "grad_norm": 0.14588595926761627, "learning_rate": 3.0600870323421084e-05, "loss": 0.0062, "step": 78360 }, { "epoch": 5.438011152492095e-13, "grad_norm": 0.16749972105026245, "learning_rate": 3.058703736510428e-05, "loss": 0.0059, "step": 78370 }, { "epoch": 5.438705041882486e-13, "grad_norm": 0.1614316999912262, "learning_rate": 3.0573209741652066e-05, "loss": 0.006, "step": 78380 }, { "epoch": 5.439398931272876e-13, "grad_norm": 0.15368634462356567, "learning_rate": 3.055938745448555e-05, "loss": 0.0059, "step": 78390 }, { "epoch": 5.440092820663267e-13, "grad_norm": 0.19387421011924744, "learning_rate": 3.054557050502509e-05, "loss": 0.0063, "step": 78400 }, { "epoch": 5.440786710053658e-13, "grad_norm": 0.16649356484413147, "learning_rate": 3.0531758894690676e-05, "loss": 0.006, "step": 78410 }, { "epoch": 5.441480599444048e-13, "grad_norm": 0.12896209955215454, "learning_rate": 3.05179526249016e-05, "loss": 0.0059, "step": 78420 }, { "epoch": 5.442174488834439e-13, "grad_norm": 0.19692960381507874, "learning_rate": 3.0504151697076702e-05, "loss": 0.0063, "step": 78430 }, { "epoch": 5.44286837822483e-13, "grad_norm": 0.14583390951156616, "learning_rate": 3.0490356112634235e-05, "loss": 0.0062, "step": 78440 }, { "epoch": 5.443562267615221e-13, "grad_norm": 0.14117619395256042, "learning_rate": 3.0476565872991908e-05, "loss": 0.0068, "step": 78450 }, { "epoch": 5.444256157005611e-13, "grad_norm": 0.15764272212982178, "learning_rate": 3.0462780979566896e-05, "loss": 0.0063, "step": 78460 }, { "epoch": 5.444950046396002e-13, "grad_norm": 0.1654536873102188, "learning_rate": 3.0449001433775796e-05, "loss": 0.006, "step": 78470 }, { "epoch": 5.445643935786393e-13, "grad_norm": 0.13994058966636658, "learning_rate": 3.0435227237034672e-05, "loss": 0.0063, "step": 78480 }, { "epoch": 5.446337825176784e-13, "grad_norm": 0.16330017149448395, "learning_rate": 3.0421458390759052e-05, "loss": 0.0065, "step": 78490 }, { "epoch": 5.447031714567174e-13, "grad_norm": 0.13151970505714417, "learning_rate": 3.0407694896363876e-05, "loss": 0.0061, "step": 78500 }, { "epoch": 5.447725603957565e-13, "grad_norm": 0.1749110221862793, "learning_rate": 3.0393936755263577e-05, "loss": 0.0063, "step": 78510 }, { "epoch": 5.448419493347956e-13, "grad_norm": 0.13523001968860626, "learning_rate": 3.0380183968872017e-05, "loss": 0.0058, "step": 78520 }, { "epoch": 5.449113382738346e-13, "grad_norm": 0.1464131623506546, "learning_rate": 3.0366436538602498e-05, "loss": 0.0058, "step": 78530 }, { "epoch": 5.449807272128737e-13, "grad_norm": 0.17036549746990204, "learning_rate": 3.0352694465867784e-05, "loss": 0.0058, "step": 78540 }, { "epoch": 5.450501161519128e-13, "grad_norm": 0.18454451858997345, "learning_rate": 3.0338957752080108e-05, "loss": 0.0059, "step": 78550 }, { "epoch": 5.451195050909519e-13, "grad_norm": 0.12814563512802124, "learning_rate": 3.0325226398651083e-05, "loss": 0.0061, "step": 78560 }, { "epoch": 5.451888940299909e-13, "grad_norm": 0.18150798976421356, "learning_rate": 3.031150040699188e-05, "loss": 0.0061, "step": 78570 }, { "epoch": 5.4525828296903e-13, "grad_norm": 0.1785961091518402, "learning_rate": 3.029777977851299e-05, "loss": 0.006, "step": 78580 }, { "epoch": 5.453276719080691e-13, "grad_norm": 0.1304413080215454, "learning_rate": 3.0284064514624498e-05, "loss": 0.0062, "step": 78590 }, { "epoch": 5.453970608471082e-13, "grad_norm": 0.14211155474185944, "learning_rate": 3.0270354616735787e-05, "loss": 0.0063, "step": 78600 }, { "epoch": 5.454664497861472e-13, "grad_norm": 0.14098946750164032, "learning_rate": 3.025665008625583e-05, "loss": 0.0069, "step": 78610 }, { "epoch": 5.455358387251863e-13, "grad_norm": 0.16898120939731598, "learning_rate": 3.0242950924592913e-05, "loss": 0.0059, "step": 78620 }, { "epoch": 5.456052276642254e-13, "grad_norm": 0.16173303127288818, "learning_rate": 3.0229257133154912e-05, "loss": 0.0067, "step": 78630 }, { "epoch": 5.456746166032644e-13, "grad_norm": 0.1613481044769287, "learning_rate": 3.0215568713349005e-05, "loss": 0.0063, "step": 78640 }, { "epoch": 5.457440055423035e-13, "grad_norm": 0.15771782398223877, "learning_rate": 3.020188566658194e-05, "loss": 0.0065, "step": 78650 }, { "epoch": 5.458133944813426e-13, "grad_norm": 0.16836807131767273, "learning_rate": 3.018820799425984e-05, "loss": 0.0061, "step": 78660 }, { "epoch": 5.458827834203817e-13, "grad_norm": 0.12101577967405319, "learning_rate": 3.0174535697788296e-05, "loss": 0.0063, "step": 78670 }, { "epoch": 5.459521723594207e-13, "grad_norm": 0.14164894819259644, "learning_rate": 3.016086877857238e-05, "loss": 0.0058, "step": 78680 }, { "epoch": 5.460215612984598e-13, "grad_norm": 0.1703297197818756, "learning_rate": 3.0147207238016527e-05, "loss": 0.006, "step": 78690 }, { "epoch": 5.460909502374989e-13, "grad_norm": 0.17124350368976593, "learning_rate": 3.0133551077524733e-05, "loss": 0.0059, "step": 78700 }, { "epoch": 5.461603391765379e-13, "grad_norm": 0.21480077505111694, "learning_rate": 3.0119900298500322e-05, "loss": 0.0063, "step": 78710 }, { "epoch": 5.46229728115577e-13, "grad_norm": 0.16680693626403809, "learning_rate": 3.010625490234618e-05, "loss": 0.0062, "step": 78720 }, { "epoch": 5.462991170546161e-13, "grad_norm": 0.1468997448682785, "learning_rate": 3.009261489046453e-05, "loss": 0.0061, "step": 78730 }, { "epoch": 5.463685059936552e-13, "grad_norm": 0.17295695841312408, "learning_rate": 3.007898026425716e-05, "loss": 0.0063, "step": 78740 }, { "epoch": 5.464378949326942e-13, "grad_norm": 0.1532568484544754, "learning_rate": 3.0065351025125184e-05, "loss": 0.0065, "step": 78750 }, { "epoch": 5.465072838717333e-13, "grad_norm": 0.16114568710327148, "learning_rate": 3.005172717446924e-05, "loss": 0.0065, "step": 78760 }, { "epoch": 5.465766728107724e-13, "grad_norm": 0.15208837389945984, "learning_rate": 3.0038108713689373e-05, "loss": 0.0069, "step": 78770 }, { "epoch": 5.466460617498115e-13, "grad_norm": 0.17806482315063477, "learning_rate": 3.0024495644185124e-05, "loss": 0.0061, "step": 78780 }, { "epoch": 5.467154506888505e-13, "grad_norm": 0.17332129180431366, "learning_rate": 3.001088796735542e-05, "loss": 0.0072, "step": 78790 }, { "epoch": 5.467848396278896e-13, "grad_norm": 0.13928189873695374, "learning_rate": 2.9997285684598673e-05, "loss": 0.0067, "step": 78800 }, { "epoch": 5.468542285669287e-13, "grad_norm": 0.18052273988723755, "learning_rate": 2.998368879731274e-05, "loss": 0.0065, "step": 78810 }, { "epoch": 5.469236175059677e-13, "grad_norm": 0.22073093056678772, "learning_rate": 2.9970097306894885e-05, "loss": 0.0063, "step": 78820 }, { "epoch": 5.469930064450068e-13, "grad_norm": 0.16078516840934753, "learning_rate": 2.9956511214741873e-05, "loss": 0.0061, "step": 78830 }, { "epoch": 5.470623953840459e-13, "grad_norm": 0.15358388423919678, "learning_rate": 2.9942930522249862e-05, "loss": 0.0059, "step": 78840 }, { "epoch": 5.47131784323085e-13, "grad_norm": 0.17253702878952026, "learning_rate": 2.9929355230814505e-05, "loss": 0.0058, "step": 78850 }, { "epoch": 5.47201173262124e-13, "grad_norm": 0.14707662165164948, "learning_rate": 2.9915785341830854e-05, "loss": 0.0059, "step": 78860 }, { "epoch": 5.472705622011631e-13, "grad_norm": 0.12370654940605164, "learning_rate": 2.990222085669343e-05, "loss": 0.0064, "step": 78870 }, { "epoch": 5.473399511402022e-13, "grad_norm": 0.15604227781295776, "learning_rate": 2.9888661776796227e-05, "loss": 0.0062, "step": 78880 }, { "epoch": 5.474093400792412e-13, "grad_norm": 0.13109168410301208, "learning_rate": 2.9875108103532576e-05, "loss": 0.0063, "step": 78890 }, { "epoch": 5.474787290182803e-13, "grad_norm": 0.1863136887550354, "learning_rate": 2.9861559838295426e-05, "loss": 0.0068, "step": 78900 }, { "epoch": 5.475481179573194e-13, "grad_norm": 0.17549656331539154, "learning_rate": 2.9848016982476968e-05, "loss": 0.0065, "step": 78910 }, { "epoch": 5.476175068963585e-13, "grad_norm": 0.20398926734924316, "learning_rate": 2.9834479537469035e-05, "loss": 0.0066, "step": 78920 }, { "epoch": 5.476868958353975e-13, "grad_norm": 0.1414707601070404, "learning_rate": 2.982094750466274e-05, "loss": 0.0065, "step": 78930 }, { "epoch": 5.477562847744366e-13, "grad_norm": 0.15439490973949432, "learning_rate": 2.980742088544876e-05, "loss": 0.0062, "step": 78940 }, { "epoch": 5.478256737134757e-13, "grad_norm": 0.17855602502822876, "learning_rate": 2.9793899681217117e-05, "loss": 0.0073, "step": 78950 }, { "epoch": 5.478950626525148e-13, "grad_norm": 0.15891864895820618, "learning_rate": 2.9780383893357393e-05, "loss": 0.0066, "step": 78960 }, { "epoch": 5.479644515915538e-13, "grad_norm": 0.15374287962913513, "learning_rate": 2.9766873523258483e-05, "loss": 0.0066, "step": 78970 }, { "epoch": 5.480338405305929e-13, "grad_norm": 0.12508299946784973, "learning_rate": 2.9753368572308807e-05, "loss": 0.0064, "step": 78980 }, { "epoch": 5.48103229469632e-13, "grad_norm": 0.1614958792924881, "learning_rate": 2.9739869041896216e-05, "loss": 0.0059, "step": 78990 }, { "epoch": 5.48172618408671e-13, "grad_norm": 0.1319250911474228, "learning_rate": 2.972637493340799e-05, "loss": 0.0059, "step": 79000 }, { "epoch": 5.482420073477101e-13, "grad_norm": 0.17923615872859955, "learning_rate": 2.971288624823086e-05, "loss": 0.0059, "step": 79010 }, { "epoch": 5.483113962867492e-13, "grad_norm": 0.15194649994373322, "learning_rate": 2.9699402987751003e-05, "loss": 0.0062, "step": 79020 }, { "epoch": 5.483807852257883e-13, "grad_norm": 0.1298903524875641, "learning_rate": 2.968592515335403e-05, "loss": 0.0062, "step": 79030 }, { "epoch": 5.484501741648273e-13, "grad_norm": 0.1526252180337906, "learning_rate": 2.9672452746425e-05, "loss": 0.0063, "step": 79040 }, { "epoch": 5.485195631038664e-13, "grad_norm": 0.18662108480930328, "learning_rate": 2.96589857683484e-05, "loss": 0.0062, "step": 79050 }, { "epoch": 5.485889520429055e-13, "grad_norm": 0.18492421507835388, "learning_rate": 2.964552422050819e-05, "loss": 0.0065, "step": 79060 }, { "epoch": 5.486583409819445e-13, "grad_norm": 0.15572118759155273, "learning_rate": 2.9632068104287746e-05, "loss": 0.0067, "step": 79070 }, { "epoch": 5.487277299209836e-13, "grad_norm": 0.16319355368614197, "learning_rate": 2.9618617421069908e-05, "loss": 0.0067, "step": 79080 }, { "epoch": 5.487971188600227e-13, "grad_norm": 0.15131168067455292, "learning_rate": 2.9605172172236895e-05, "loss": 0.0061, "step": 79090 }, { "epoch": 5.488665077990618e-13, "grad_norm": 0.18526232242584229, "learning_rate": 2.9591732359170492e-05, "loss": 0.0067, "step": 79100 }, { "epoch": 5.489358967381008e-13, "grad_norm": 0.17672833800315857, "learning_rate": 2.9578297983251752e-05, "loss": 0.007, "step": 79110 }, { "epoch": 5.490052856771399e-13, "grad_norm": 0.13465720415115356, "learning_rate": 2.9564869045861363e-05, "loss": 0.0066, "step": 79120 }, { "epoch": 5.49074674616179e-13, "grad_norm": 0.14885835349559784, "learning_rate": 2.9551445548379304e-05, "loss": 0.0066, "step": 79130 }, { "epoch": 5.491440635552181e-13, "grad_norm": 0.15336258709430695, "learning_rate": 2.9538027492185053e-05, "loss": 0.0063, "step": 79140 }, { "epoch": 5.492134524942571e-13, "grad_norm": 0.15552069246768951, "learning_rate": 2.9524614878657526e-05, "loss": 0.006, "step": 79150 }, { "epoch": 5.492828414332962e-13, "grad_norm": 0.12477695196866989, "learning_rate": 2.9511207709175078e-05, "loss": 0.0058, "step": 79160 }, { "epoch": 5.493522303723353e-13, "grad_norm": 0.13627034425735474, "learning_rate": 2.9497805985115517e-05, "loss": 0.006, "step": 79170 }, { "epoch": 5.494216193113743e-13, "grad_norm": 0.1521320641040802, "learning_rate": 2.9484409707856066e-05, "loss": 0.0063, "step": 79180 }, { "epoch": 5.494910082504134e-13, "grad_norm": 0.14839424192905426, "learning_rate": 2.9471018878773402e-05, "loss": 0.0061, "step": 79190 }, { "epoch": 5.495603971894525e-13, "grad_norm": 0.12802134454250336, "learning_rate": 2.9457633499243643e-05, "loss": 0.0061, "step": 79200 }, { "epoch": 5.496297861284916e-13, "grad_norm": 0.13085278868675232, "learning_rate": 2.944425357064236e-05, "loss": 0.0061, "step": 79210 }, { "epoch": 5.496991750675306e-13, "grad_norm": 0.17019616067409515, "learning_rate": 2.9430879094344497e-05, "loss": 0.0064, "step": 79220 }, { "epoch": 5.497685640065697e-13, "grad_norm": 0.18688057363033295, "learning_rate": 2.9417510071724566e-05, "loss": 0.007, "step": 79230 }, { "epoch": 5.498379529456088e-13, "grad_norm": 0.1586703658103943, "learning_rate": 2.9404146504156366e-05, "loss": 0.0065, "step": 79240 }, { "epoch": 5.499073418846478e-13, "grad_norm": 0.13725504279136658, "learning_rate": 2.939078839301329e-05, "loss": 0.0061, "step": 79250 }, { "epoch": 5.499767308236869e-13, "grad_norm": 0.22882461547851562, "learning_rate": 2.9377435739668004e-05, "loss": 0.0072, "step": 79260 }, { "epoch": 5.50046119762726e-13, "grad_norm": 0.14505280554294586, "learning_rate": 2.9364088545492775e-05, "loss": 0.0075, "step": 79270 }, { "epoch": 5.501155087017651e-13, "grad_norm": 0.15871627628803253, "learning_rate": 2.935074681185919e-05, "loss": 0.0065, "step": 79280 }, { "epoch": 5.501848976408041e-13, "grad_norm": 0.13616304099559784, "learning_rate": 2.933741054013836e-05, "loss": 0.0062, "step": 79290 }, { "epoch": 5.502542865798432e-13, "grad_norm": 0.14203399419784546, "learning_rate": 2.932407973170076e-05, "loss": 0.0063, "step": 79300 }, { "epoch": 5.503236755188823e-13, "grad_norm": 0.13040386140346527, "learning_rate": 2.9310754387916344e-05, "loss": 0.0059, "step": 79310 }, { "epoch": 5.503930644579214e-13, "grad_norm": 0.14268486201763153, "learning_rate": 2.92974345101545e-05, "loss": 0.0056, "step": 79320 }, { "epoch": 5.504624533969604e-13, "grad_norm": 0.1836567223072052, "learning_rate": 2.928412009978405e-05, "loss": 0.006, "step": 79330 }, { "epoch": 5.505318423359995e-13, "grad_norm": 0.14514906704425812, "learning_rate": 2.9270811158173274e-05, "loss": 0.0061, "step": 79340 }, { "epoch": 5.506012312750386e-13, "grad_norm": 0.13947634398937225, "learning_rate": 2.9257507686689856e-05, "loss": 0.006, "step": 79350 }, { "epoch": 5.506706202140776e-13, "grad_norm": 0.1453571617603302, "learning_rate": 2.9244209686700945e-05, "loss": 0.006, "step": 79360 }, { "epoch": 5.507400091531167e-13, "grad_norm": 0.13923192024230957, "learning_rate": 2.923091715957311e-05, "loss": 0.0063, "step": 79370 }, { "epoch": 5.508093980921558e-13, "grad_norm": 0.1441570520401001, "learning_rate": 2.9217630106672363e-05, "loss": 0.0062, "step": 79380 }, { "epoch": 5.508787870311949e-13, "grad_norm": 0.15947192907333374, "learning_rate": 2.9204348529364166e-05, "loss": 0.0065, "step": 79390 }, { "epoch": 5.509481759702339e-13, "grad_norm": 0.1452464461326599, "learning_rate": 2.919107242901339e-05, "loss": 0.0064, "step": 79400 }, { "epoch": 5.51017564909273e-13, "grad_norm": 0.1417127102613449, "learning_rate": 2.9177801806984382e-05, "loss": 0.0062, "step": 79410 }, { "epoch": 5.510869538483121e-13, "grad_norm": 0.16045315563678741, "learning_rate": 2.9164536664640896e-05, "loss": 0.0069, "step": 79420 }, { "epoch": 5.511563427873512e-13, "grad_norm": 0.13124768435955048, "learning_rate": 2.9151277003346146e-05, "loss": 0.0074, "step": 79430 }, { "epoch": 5.512257317263902e-13, "grad_norm": 0.1330002397298813, "learning_rate": 2.9138022824462716e-05, "loss": 0.0062, "step": 79440 }, { "epoch": 5.512951206654293e-13, "grad_norm": 0.14081628620624542, "learning_rate": 2.912477412935276e-05, "loss": 0.0061, "step": 79450 }, { "epoch": 5.513645096044684e-13, "grad_norm": 0.1427016258239746, "learning_rate": 2.91115309193777e-05, "loss": 0.006, "step": 79460 }, { "epoch": 5.514338985435074e-13, "grad_norm": 0.11845952272415161, "learning_rate": 2.909829319589856e-05, "loss": 0.0059, "step": 79470 }, { "epoch": 5.515032874825465e-13, "grad_norm": 0.1580101102590561, "learning_rate": 2.908506096027566e-05, "loss": 0.0058, "step": 79480 }, { "epoch": 5.515726764215856e-13, "grad_norm": 0.2154001146554947, "learning_rate": 2.9071834213868875e-05, "loss": 0.0062, "step": 79490 }, { "epoch": 5.516420653606247e-13, "grad_norm": 0.15073111653327942, "learning_rate": 2.9058612958037424e-05, "loss": 0.0062, "step": 79500 }, { "epoch": 5.517114542996637e-13, "grad_norm": 0.19816260039806366, "learning_rate": 2.9045397194139992e-05, "loss": 0.0063, "step": 79510 }, { "epoch": 5.517808432387028e-13, "grad_norm": 0.14022545516490936, "learning_rate": 2.903218692353471e-05, "loss": 0.0062, "step": 79520 }, { "epoch": 5.518502321777419e-13, "grad_norm": 0.20509572327136993, "learning_rate": 2.901898214757916e-05, "loss": 0.0062, "step": 79530 }, { "epoch": 5.519196211167809e-13, "grad_norm": 0.1409214437007904, "learning_rate": 2.9005782867630315e-05, "loss": 0.0063, "step": 79540 }, { "epoch": 5.5198901005582e-13, "grad_norm": 0.1545439511537552, "learning_rate": 2.899258908504462e-05, "loss": 0.0065, "step": 79550 }, { "epoch": 5.520583989948591e-13, "grad_norm": 0.16378867626190186, "learning_rate": 2.897940080117792e-05, "loss": 0.0062, "step": 79560 }, { "epoch": 5.521277879338982e-13, "grad_norm": 0.12953031063079834, "learning_rate": 2.8966218017385544e-05, "loss": 0.0063, "step": 79570 }, { "epoch": 5.521971768729372e-13, "grad_norm": 0.19890804588794708, "learning_rate": 2.8953040735022206e-05, "loss": 0.0068, "step": 79580 }, { "epoch": 5.522665658119763e-13, "grad_norm": 0.1209481954574585, "learning_rate": 2.8939868955442095e-05, "loss": 0.0067, "step": 79590 }, { "epoch": 5.523359547510154e-13, "grad_norm": 0.16368286311626434, "learning_rate": 2.8926702679998823e-05, "loss": 0.0063, "step": 79600 }, { "epoch": 5.524053436900545e-13, "grad_norm": 0.14394111931324005, "learning_rate": 2.891354191004538e-05, "loss": 0.0064, "step": 79610 }, { "epoch": 5.524747326290935e-13, "grad_norm": 0.1693994402885437, "learning_rate": 2.8900386646934306e-05, "loss": 0.0059, "step": 79620 }, { "epoch": 5.525441215681326e-13, "grad_norm": 0.13471990823745728, "learning_rate": 2.8887236892017463e-05, "loss": 0.0059, "step": 79630 }, { "epoch": 5.526135105071717e-13, "grad_norm": 0.11307079344987869, "learning_rate": 2.8874092646646206e-05, "loss": 0.006, "step": 79640 }, { "epoch": 5.526828994462107e-13, "grad_norm": 0.14203877747058868, "learning_rate": 2.886095391217131e-05, "loss": 0.0063, "step": 79650 }, { "epoch": 5.527522883852498e-13, "grad_norm": 0.17664454877376556, "learning_rate": 2.8847820689942993e-05, "loss": 0.0063, "step": 79660 }, { "epoch": 5.528216773242889e-13, "grad_norm": 0.13167470693588257, "learning_rate": 2.8834692981310883e-05, "loss": 0.0063, "step": 79670 }, { "epoch": 5.52891066263328e-13, "grad_norm": 0.14628303050994873, "learning_rate": 2.882157078762407e-05, "loss": 0.0063, "step": 79680 }, { "epoch": 5.52960455202367e-13, "grad_norm": 0.1481662541627884, "learning_rate": 2.8808454110231053e-05, "loss": 0.0061, "step": 79690 }, { "epoch": 5.530298441414061e-13, "grad_norm": 0.1326954960823059, "learning_rate": 2.8795342950479787e-05, "loss": 0.0067, "step": 79700 }, { "epoch": 5.530992330804452e-13, "grad_norm": 0.1198628693819046, "learning_rate": 2.8782237309717633e-05, "loss": 0.0063, "step": 79710 }, { "epoch": 5.531686220194842e-13, "grad_norm": 0.13337907195091248, "learning_rate": 2.8769137189291413e-05, "loss": 0.0061, "step": 79720 }, { "epoch": 5.532380109585233e-13, "grad_norm": 0.1650388240814209, "learning_rate": 2.8756042590547358e-05, "loss": 0.0063, "step": 79730 }, { "epoch": 5.533073998975624e-13, "grad_norm": 0.1287394016981125, "learning_rate": 2.8742953514831152e-05, "loss": 0.0065, "step": 79740 }, { "epoch": 5.533767888366015e-13, "grad_norm": 0.13287626206874847, "learning_rate": 2.8729869963487888e-05, "loss": 0.0063, "step": 79750 }, { "epoch": 5.534461777756405e-13, "grad_norm": 0.15486517548561096, "learning_rate": 2.8716791937862135e-05, "loss": 0.0062, "step": 79760 }, { "epoch": 5.535155667146796e-13, "grad_norm": 0.12909603118896484, "learning_rate": 2.8703719439297798e-05, "loss": 0.006, "step": 79770 }, { "epoch": 5.535849556537187e-13, "grad_norm": 0.13316583633422852, "learning_rate": 2.8690652469138358e-05, "loss": 0.0061, "step": 79780 }, { "epoch": 5.536543445927578e-13, "grad_norm": 0.14457768201828003, "learning_rate": 2.8677591028726584e-05, "loss": 0.0062, "step": 79790 }, { "epoch": 5.537237335317968e-13, "grad_norm": 0.14046482741832733, "learning_rate": 2.8664535119404804e-05, "loss": 0.0067, "step": 79800 }, { "epoch": 5.537931224708359e-13, "grad_norm": 0.1781223714351654, "learning_rate": 2.865148474251465e-05, "loss": 0.0064, "step": 79810 }, { "epoch": 5.53862511409875e-13, "grad_norm": 0.17446760833263397, "learning_rate": 2.863843989939732e-05, "loss": 0.0065, "step": 79820 }, { "epoch": 5.53931900348914e-13, "grad_norm": 0.15826241672039032, "learning_rate": 2.862540059139332e-05, "loss": 0.0062, "step": 79830 }, { "epoch": 5.540012892879531e-13, "grad_norm": 0.16514933109283447, "learning_rate": 2.861236681984266e-05, "loss": 0.0064, "step": 79840 }, { "epoch": 5.540706782269922e-13, "grad_norm": 0.1301426887512207, "learning_rate": 2.859933858608476e-05, "loss": 0.006, "step": 79850 }, { "epoch": 5.541400671660313e-13, "grad_norm": 0.19336658716201782, "learning_rate": 2.8586315891458486e-05, "loss": 0.0067, "step": 79860 }, { "epoch": 5.542094561050703e-13, "grad_norm": 0.1441114842891693, "learning_rate": 2.857329873730211e-05, "loss": 0.0063, "step": 79870 }, { "epoch": 5.542788450441094e-13, "grad_norm": 0.16135352849960327, "learning_rate": 2.856028712495335e-05, "loss": 0.0061, "step": 79880 }, { "epoch": 5.543482339831485e-13, "grad_norm": 0.12974555790424347, "learning_rate": 2.8547281055749355e-05, "loss": 0.0064, "step": 79890 }, { "epoch": 5.544176229221875e-13, "grad_norm": 0.17141243815422058, "learning_rate": 2.8534280531026697e-05, "loss": 0.0065, "step": 79900 }, { "epoch": 5.544870118612266e-13, "grad_norm": 0.17993561923503876, "learning_rate": 2.8521285552121384e-05, "loss": 0.0062, "step": 79910 }, { "epoch": 5.545564008002657e-13, "grad_norm": 0.10866876691579819, "learning_rate": 2.850829612036886e-05, "loss": 0.006, "step": 79920 }, { "epoch": 5.546257897393048e-13, "grad_norm": 0.11504978686571121, "learning_rate": 2.8495312237103974e-05, "loss": 0.006, "step": 79930 }, { "epoch": 5.546951786783438e-13, "grad_norm": 0.1502632349729538, "learning_rate": 2.8482333903661034e-05, "loss": 0.0061, "step": 79940 }, { "epoch": 5.547645676173829e-13, "grad_norm": 0.15352439880371094, "learning_rate": 2.8469361121373766e-05, "loss": 0.0064, "step": 79950 }, { "epoch": 5.54833956556422e-13, "grad_norm": 0.13537003099918365, "learning_rate": 2.8456393891575335e-05, "loss": 0.0068, "step": 79960 }, { "epoch": 5.549033454954611e-13, "grad_norm": 0.16484986245632172, "learning_rate": 2.844343221559828e-05, "loss": 0.0061, "step": 79970 }, { "epoch": 5.549727344345001e-13, "grad_norm": 0.13111460208892822, "learning_rate": 2.8430476094774692e-05, "loss": 0.0063, "step": 79980 }, { "epoch": 5.550421233735392e-13, "grad_norm": 0.12462030351161957, "learning_rate": 2.841752553043593e-05, "loss": 0.0063, "step": 79990 }, { "epoch": 5.551115123125783e-13, "grad_norm": 0.14344331622123718, "learning_rate": 2.8404580523912935e-05, "loss": 0.0061, "step": 80000 }, { "epoch": 5.551809012516173e-13, "grad_norm": 0.21036407351493835, "learning_rate": 2.839164107653594e-05, "loss": 0.0062, "step": 80010 }, { "epoch": 5.552502901906564e-13, "grad_norm": 0.12574440240859985, "learning_rate": 2.8378707189634762e-05, "loss": 0.0064, "step": 80020 }, { "epoch": 5.553196791296955e-13, "grad_norm": 0.13946574926376343, "learning_rate": 2.8365778864538464e-05, "loss": 0.0061, "step": 80030 }, { "epoch": 5.553890680687346e-13, "grad_norm": 0.14115217328071594, "learning_rate": 2.835285610257572e-05, "loss": 0.0062, "step": 80040 }, { "epoch": 5.554584570077736e-13, "grad_norm": 0.11657387763261795, "learning_rate": 2.8339938905074488e-05, "loss": 0.0065, "step": 80050 }, { "epoch": 5.555278459468127e-13, "grad_norm": 0.13586238026618958, "learning_rate": 2.8327027273362224e-05, "loss": 0.0061, "step": 80060 }, { "epoch": 5.555972348858518e-13, "grad_norm": 0.14868804812431335, "learning_rate": 2.8314121208765803e-05, "loss": 0.006, "step": 80070 }, { "epoch": 5.556666238248908e-13, "grad_norm": 0.1452762931585312, "learning_rate": 2.8301220712611523e-05, "loss": 0.006, "step": 80080 }, { "epoch": 5.557360127639299e-13, "grad_norm": 0.1284928023815155, "learning_rate": 2.828832578622513e-05, "loss": 0.0061, "step": 80090 }, { "epoch": 5.55805401702969e-13, "grad_norm": 0.15173271298408508, "learning_rate": 2.8275436430931717e-05, "loss": 0.0059, "step": 80100 }, { "epoch": 5.558747906420081e-13, "grad_norm": 0.14884550869464874, "learning_rate": 2.826255264805595e-05, "loss": 0.0062, "step": 80110 }, { "epoch": 5.559441795810471e-13, "grad_norm": 0.13780759274959564, "learning_rate": 2.8249674438921757e-05, "loss": 0.0068, "step": 80120 }, { "epoch": 5.560135685200862e-13, "grad_norm": 0.13694587349891663, "learning_rate": 2.8236801804852648e-05, "loss": 0.0059, "step": 80130 }, { "epoch": 5.560829574591253e-13, "grad_norm": 0.12587769329547882, "learning_rate": 2.8223934747171414e-05, "loss": 0.0061, "step": 80140 }, { "epoch": 5.561523463981644e-13, "grad_norm": 0.1548135131597519, "learning_rate": 2.8211073267200427e-05, "loss": 0.0062, "step": 80150 }, { "epoch": 5.562217353372034e-13, "grad_norm": 0.1556343287229538, "learning_rate": 2.8198217366261343e-05, "loss": 0.0061, "step": 80160 }, { "epoch": 5.562911242762425e-13, "grad_norm": 0.1423150897026062, "learning_rate": 2.8185367045675312e-05, "loss": 0.0063, "step": 80170 }, { "epoch": 5.563605132152816e-13, "grad_norm": 0.15407031774520874, "learning_rate": 2.817252230676292e-05, "loss": 0.0064, "step": 80180 }, { "epoch": 5.564299021543206e-13, "grad_norm": 0.1552441418170929, "learning_rate": 2.8159683150844165e-05, "loss": 0.0063, "step": 80190 }, { "epoch": 5.564992910933597e-13, "grad_norm": 0.1273980438709259, "learning_rate": 2.8146849579238463e-05, "loss": 0.0062, "step": 80200 }, { "epoch": 5.565686800323988e-13, "grad_norm": 0.13972420990467072, "learning_rate": 2.8134021593264657e-05, "loss": 0.0061, "step": 80210 }, { "epoch": 5.566380689714379e-13, "grad_norm": 0.13638882339000702, "learning_rate": 2.812119919424103e-05, "loss": 0.006, "step": 80220 }, { "epoch": 5.567074579104769e-13, "grad_norm": 0.1522969901561737, "learning_rate": 2.8108382383485294e-05, "loss": 0.006, "step": 80230 }, { "epoch": 5.56776846849516e-13, "grad_norm": 0.11555840075016022, "learning_rate": 2.809557116231456e-05, "loss": 0.0059, "step": 80240 }, { "epoch": 5.568462357885551e-13, "grad_norm": 0.09418584406375885, "learning_rate": 2.8082765532045378e-05, "loss": 0.0061, "step": 80250 }, { "epoch": 5.569156247275941e-13, "grad_norm": 0.13780616223812103, "learning_rate": 2.8069965493993733e-05, "loss": 0.006, "step": 80260 }, { "epoch": 5.569850136666332e-13, "grad_norm": 0.13988330960273743, "learning_rate": 2.8057171049475036e-05, "loss": 0.0066, "step": 80270 }, { "epoch": 5.570544026056723e-13, "grad_norm": 0.15009461343288422, "learning_rate": 2.8044382199804098e-05, "loss": 0.0063, "step": 80280 }, { "epoch": 5.571237915447114e-13, "grad_norm": 0.1435912549495697, "learning_rate": 2.803159894629521e-05, "loss": 0.0059, "step": 80290 }, { "epoch": 5.571931804837504e-13, "grad_norm": 0.14348487555980682, "learning_rate": 2.8018821290261973e-05, "loss": 0.006, "step": 80300 }, { "epoch": 5.572625694227895e-13, "grad_norm": 0.13724945485591888, "learning_rate": 2.800604923301757e-05, "loss": 0.0061, "step": 80310 }, { "epoch": 5.573319583618286e-13, "grad_norm": 0.13410189747810364, "learning_rate": 2.7993282775874478e-05, "loss": 0.0061, "step": 80320 }, { "epoch": 5.574013473008677e-13, "grad_norm": 0.14047424495220184, "learning_rate": 2.79805219201447e-05, "loss": 0.0067, "step": 80330 }, { "epoch": 5.574707362399067e-13, "grad_norm": 0.15651270747184753, "learning_rate": 2.7967766667139538e-05, "loss": 0.006, "step": 80340 }, { "epoch": 5.575401251789458e-13, "grad_norm": 0.14581120014190674, "learning_rate": 2.7955017018169872e-05, "loss": 0.0064, "step": 80350 }, { "epoch": 5.576095141179849e-13, "grad_norm": 0.12895439565181732, "learning_rate": 2.7942272974545858e-05, "loss": 0.0064, "step": 80360 }, { "epoch": 5.576789030570239e-13, "grad_norm": 0.1384688764810562, "learning_rate": 2.7929534537577213e-05, "loss": 0.0059, "step": 80370 }, { "epoch": 5.57748291996063e-13, "grad_norm": 0.14531610906124115, "learning_rate": 2.7916801708572947e-05, "loss": 0.0059, "step": 80380 }, { "epoch": 5.578176809351021e-13, "grad_norm": 0.14335401356220245, "learning_rate": 2.7904074488841594e-05, "loss": 0.006, "step": 80390 }, { "epoch": 5.578870698741412e-13, "grad_norm": 0.17086414992809296, "learning_rate": 2.7891352879691055e-05, "loss": 0.006, "step": 80400 }, { "epoch": 5.579564588131802e-13, "grad_norm": 0.1477409303188324, "learning_rate": 2.7878636882428675e-05, "loss": 0.006, "step": 80410 }, { "epoch": 5.580258477522193e-13, "grad_norm": 0.14566338062286377, "learning_rate": 2.786592649836124e-05, "loss": 0.0058, "step": 80420 }, { "epoch": 5.580952366912584e-13, "grad_norm": 0.13674215972423553, "learning_rate": 2.7853221728794917e-05, "loss": 0.0063, "step": 80430 }, { "epoch": 5.581646256302975e-13, "grad_norm": 0.15109188854694366, "learning_rate": 2.784052257503533e-05, "loss": 0.0062, "step": 80440 }, { "epoch": 5.582340145693365e-13, "grad_norm": 0.14343257248401642, "learning_rate": 2.782782903838751e-05, "loss": 0.0058, "step": 80450 }, { "epoch": 5.583034035083756e-13, "grad_norm": 0.11797402799129486, "learning_rate": 2.7815141120155926e-05, "loss": 0.006, "step": 80460 }, { "epoch": 5.583727924474147e-13, "grad_norm": 0.14414647221565247, "learning_rate": 2.7802458821644448e-05, "loss": 0.006, "step": 80470 }, { "epoch": 5.584421813864537e-13, "grad_norm": 0.1459723860025406, "learning_rate": 2.7789782144156375e-05, "loss": 0.0065, "step": 80480 }, { "epoch": 5.585115703254928e-13, "grad_norm": 0.15111292898654938, "learning_rate": 2.777711108899446e-05, "loss": 0.0062, "step": 80490 }, { "epoch": 5.585809592645319e-13, "grad_norm": 0.12416911125183105, "learning_rate": 2.7764445657460815e-05, "loss": 0.0059, "step": 80500 }, { "epoch": 5.58650348203571e-13, "grad_norm": 0.12504860758781433, "learning_rate": 2.7751785850857033e-05, "loss": 0.0061, "step": 80510 }, { "epoch": 5.5871973714261e-13, "grad_norm": 0.1278969794511795, "learning_rate": 2.773913167048409e-05, "loss": 0.0064, "step": 80520 }, { "epoch": 5.587891260816491e-13, "grad_norm": 0.1478598266839981, "learning_rate": 2.7726483117642414e-05, "loss": 0.0058, "step": 80530 }, { "epoch": 5.588585150206882e-13, "grad_norm": 0.14328734576702118, "learning_rate": 2.771384019363184e-05, "loss": 0.0059, "step": 80540 }, { "epoch": 5.589279039597272e-13, "grad_norm": 0.16071346402168274, "learning_rate": 2.770120289975161e-05, "loss": 0.0061, "step": 80550 }, { "epoch": 5.589972928987663e-13, "grad_norm": 0.14259295165538788, "learning_rate": 2.7688571237300422e-05, "loss": 0.0059, "step": 80560 }, { "epoch": 5.590666818378054e-13, "grad_norm": 0.14315254986286163, "learning_rate": 2.767594520757637e-05, "loss": 0.0057, "step": 80570 }, { "epoch": 5.591360707768445e-13, "grad_norm": 0.20429584383964539, "learning_rate": 2.7663324811876972e-05, "loss": 0.0061, "step": 80580 }, { "epoch": 5.592054597158835e-13, "grad_norm": 0.13827694952487946, "learning_rate": 2.765071005149917e-05, "loss": 0.0058, "step": 80590 }, { "epoch": 5.592748486549226e-13, "grad_norm": 0.12841850519180298, "learning_rate": 2.763810092773932e-05, "loss": 0.0058, "step": 80600 }, { "epoch": 5.593442375939617e-13, "grad_norm": 0.144059956073761, "learning_rate": 2.7625497441893213e-05, "loss": 0.0059, "step": 80610 }, { "epoch": 5.594136265330008e-13, "grad_norm": 0.15601783990859985, "learning_rate": 2.761289959525608e-05, "loss": 0.0059, "step": 80620 }, { "epoch": 5.594830154720398e-13, "grad_norm": 0.12996520102024078, "learning_rate": 2.7600307389122477e-05, "loss": 0.0064, "step": 80630 }, { "epoch": 5.595524044110789e-13, "grad_norm": 0.13288277387619019, "learning_rate": 2.758772082478653e-05, "loss": 0.0066, "step": 80640 }, { "epoch": 5.59621793350118e-13, "grad_norm": 0.14348091185092926, "learning_rate": 2.7575139903541635e-05, "loss": 0.0061, "step": 80650 }, { "epoch": 5.59691182289157e-13, "grad_norm": 0.1365506500005722, "learning_rate": 2.7562564626680732e-05, "loss": 0.006, "step": 80660 }, { "epoch": 5.597605712281961e-13, "grad_norm": 0.11659976840019226, "learning_rate": 2.754999499549608e-05, "loss": 0.006, "step": 80670 }, { "epoch": 5.598299601672352e-13, "grad_norm": 0.15796777606010437, "learning_rate": 2.753743101127945e-05, "loss": 0.0062, "step": 80680 }, { "epoch": 5.598993491062743e-13, "grad_norm": 0.10959349572658539, "learning_rate": 2.752487267532194e-05, "loss": 0.0057, "step": 80690 }, { "epoch": 5.599687380453133e-13, "grad_norm": 0.13083580136299133, "learning_rate": 2.751231998891417e-05, "loss": 0.006, "step": 80700 }, { "epoch": 5.600381269843524e-13, "grad_norm": 0.10655312240123749, "learning_rate": 2.7499772953346072e-05, "loss": 0.0058, "step": 80710 }, { "epoch": 5.601075159233915e-13, "grad_norm": 0.09757020324468613, "learning_rate": 2.7487231569907065e-05, "loss": 0.0057, "step": 80720 }, { "epoch": 5.601769048624305e-13, "grad_norm": 0.11151410639286041, "learning_rate": 2.747469583988598e-05, "loss": 0.0057, "step": 80730 }, { "epoch": 5.602462938014696e-13, "grad_norm": 0.18239504098892212, "learning_rate": 2.7462165764571055e-05, "loss": 0.006, "step": 80740 }, { "epoch": 5.603156827405087e-13, "grad_norm": 0.13176649808883667, "learning_rate": 2.744964134524994e-05, "loss": 0.0058, "step": 80750 }, { "epoch": 5.603850716795478e-13, "grad_norm": 0.15437687933444977, "learning_rate": 2.7437122583209732e-05, "loss": 0.0058, "step": 80760 }, { "epoch": 5.604544606185868e-13, "grad_norm": 0.13415446877479553, "learning_rate": 2.7424609479736913e-05, "loss": 0.0058, "step": 80770 }, { "epoch": 5.605238495576259e-13, "grad_norm": 0.11590395867824554, "learning_rate": 2.7412102036117407e-05, "loss": 0.0059, "step": 80780 }, { "epoch": 5.60593238496665e-13, "grad_norm": 0.1474672257900238, "learning_rate": 2.739960025363655e-05, "loss": 0.006, "step": 80790 }, { "epoch": 5.606626274357041e-13, "grad_norm": 0.15880055725574493, "learning_rate": 2.7387104133579086e-05, "loss": 0.0063, "step": 80800 }, { "epoch": 5.607320163747431e-13, "grad_norm": 0.12040431797504425, "learning_rate": 2.7374613677229194e-05, "loss": 0.0059, "step": 80810 }, { "epoch": 5.608014053137822e-13, "grad_norm": 0.13387736678123474, "learning_rate": 2.736212888587048e-05, "loss": 0.0061, "step": 80820 }, { "epoch": 5.608707942528213e-13, "grad_norm": 0.14328418672084808, "learning_rate": 2.7349649760785888e-05, "loss": 0.006, "step": 80830 }, { "epoch": 5.609401831918603e-13, "grad_norm": 0.11740589141845703, "learning_rate": 2.7337176303257925e-05, "loss": 0.0061, "step": 80840 }, { "epoch": 5.610095721308994e-13, "grad_norm": 0.12409884482622147, "learning_rate": 2.732470851456836e-05, "loss": 0.0057, "step": 80850 }, { "epoch": 5.610789610699385e-13, "grad_norm": 0.13643281161785126, "learning_rate": 2.7312246395998523e-05, "loss": 0.006, "step": 80860 }, { "epoch": 5.611483500089776e-13, "grad_norm": 0.1338299959897995, "learning_rate": 2.7299789948829007e-05, "loss": 0.0058, "step": 80870 }, { "epoch": 5.612177389480166e-13, "grad_norm": 0.10430609434843063, "learning_rate": 2.7287339174339983e-05, "loss": 0.0059, "step": 80880 }, { "epoch": 5.612871278870557e-13, "grad_norm": 0.14106114208698273, "learning_rate": 2.7274894073810904e-05, "loss": 0.0058, "step": 80890 }, { "epoch": 5.613565168260948e-13, "grad_norm": 0.16323544085025787, "learning_rate": 2.7262454648520757e-05, "loss": 0.0061, "step": 80900 }, { "epoch": 5.614259057651338e-13, "grad_norm": 0.10709664970636368, "learning_rate": 2.7250020899747836e-05, "loss": 0.0057, "step": 80910 }, { "epoch": 5.614952947041729e-13, "grad_norm": 0.12067084014415741, "learning_rate": 2.7237592828769924e-05, "loss": 0.0059, "step": 80920 }, { "epoch": 5.61564683643212e-13, "grad_norm": 0.13756373524665833, "learning_rate": 2.7225170436864194e-05, "loss": 0.0059, "step": 80930 }, { "epoch": 5.616340725822511e-13, "grad_norm": 0.14184486865997314, "learning_rate": 2.7212753725307228e-05, "loss": 0.006, "step": 80940 }, { "epoch": 5.617034615212901e-13, "grad_norm": 0.11597050726413727, "learning_rate": 2.7200342695375072e-05, "loss": 0.0062, "step": 80950 }, { "epoch": 5.617728504603292e-13, "grad_norm": 0.132301926612854, "learning_rate": 2.7187937348343095e-05, "loss": 0.0061, "step": 80960 }, { "epoch": 5.618422393993683e-13, "grad_norm": 0.23432160913944244, "learning_rate": 2.717553768548621e-05, "loss": 0.0061, "step": 80970 }, { "epoch": 5.619116283384074e-13, "grad_norm": 0.2702256739139557, "learning_rate": 2.7163143708078604e-05, "loss": 0.006, "step": 80980 }, { "epoch": 5.619810172774464e-13, "grad_norm": 0.13602256774902344, "learning_rate": 2.715075541739402e-05, "loss": 0.0059, "step": 80990 }, { "epoch": 5.620504062164855e-13, "grad_norm": 0.1521623432636261, "learning_rate": 2.713837281470547e-05, "loss": 0.0057, "step": 81000 }, { "epoch": 5.621197951555246e-13, "grad_norm": 0.12068050354719162, "learning_rate": 2.7125995901285547e-05, "loss": 0.0059, "step": 81010 }, { "epoch": 5.621891840945636e-13, "grad_norm": 0.1386909931898117, "learning_rate": 2.711362467840609e-05, "loss": 0.0058, "step": 81020 }, { "epoch": 5.622585730336027e-13, "grad_norm": 0.18765297532081604, "learning_rate": 2.71012591473385e-05, "loss": 0.0061, "step": 81030 }, { "epoch": 5.623279619726418e-13, "grad_norm": 0.1333843171596527, "learning_rate": 2.7088899309353483e-05, "loss": 0.0057, "step": 81040 }, { "epoch": 5.623973509116809e-13, "grad_norm": 0.137503981590271, "learning_rate": 2.7076545165721227e-05, "loss": 0.006, "step": 81050 }, { "epoch": 5.624667398507199e-13, "grad_norm": 0.1527857780456543, "learning_rate": 2.706419671771131e-05, "loss": 0.0055, "step": 81060 }, { "epoch": 5.62536128789759e-13, "grad_norm": 0.10300601273775101, "learning_rate": 2.7051853966592714e-05, "loss": 0.0056, "step": 81070 }, { "epoch": 5.626055177287981e-13, "grad_norm": 0.11369865387678146, "learning_rate": 2.7039516913633862e-05, "loss": 0.0058, "step": 81080 }, { "epoch": 5.626749066678371e-13, "grad_norm": 0.12338189035654068, "learning_rate": 2.7027185560102573e-05, "loss": 0.0057, "step": 81090 }, { "epoch": 5.627442956068762e-13, "grad_norm": 0.13665255904197693, "learning_rate": 2.7014859907266086e-05, "loss": 0.0059, "step": 81100 }, { "epoch": 5.628136845459153e-13, "grad_norm": 0.13376769423484802, "learning_rate": 2.7002539956391058e-05, "loss": 0.0064, "step": 81110 }, { "epoch": 5.628830734849544e-13, "grad_norm": 0.1878439038991928, "learning_rate": 2.6990225708743542e-05, "loss": 0.0059, "step": 81120 }, { "epoch": 5.629524624239934e-13, "grad_norm": 0.15521292388439178, "learning_rate": 2.697791716558904e-05, "loss": 0.0061, "step": 81130 }, { "epoch": 5.630218513630325e-13, "grad_norm": 0.17220443487167358, "learning_rate": 2.6965614328192426e-05, "loss": 0.0063, "step": 81140 }, { "epoch": 5.630912403020716e-13, "grad_norm": 0.1164737194776535, "learning_rate": 2.695331719781804e-05, "loss": 0.0059, "step": 81150 }, { "epoch": 5.631606292411107e-13, "grad_norm": 0.11410751193761826, "learning_rate": 2.6941025775729544e-05, "loss": 0.0057, "step": 81160 }, { "epoch": 5.632300181801497e-13, "grad_norm": 0.09950246661901474, "learning_rate": 2.6928740063190136e-05, "loss": 0.0057, "step": 81170 }, { "epoch": 5.632994071191888e-13, "grad_norm": 0.10306872427463531, "learning_rate": 2.6916460061462313e-05, "loss": 0.0057, "step": 81180 }, { "epoch": 5.633687960582279e-13, "grad_norm": 0.11047536134719849, "learning_rate": 2.6904185771808095e-05, "loss": 0.0058, "step": 81190 }, { "epoch": 5.634381849972669e-13, "grad_norm": 0.1403772234916687, "learning_rate": 2.6891917195488783e-05, "loss": 0.0056, "step": 81200 }, { "epoch": 5.63507573936306e-13, "grad_norm": 0.13199971616268158, "learning_rate": 2.6879654333765242e-05, "loss": 0.006, "step": 81210 }, { "epoch": 5.635769628753451e-13, "grad_norm": 0.13522553443908691, "learning_rate": 2.6867397187897598e-05, "loss": 0.0055, "step": 81220 }, { "epoch": 5.636463518143842e-13, "grad_norm": 0.11928708851337433, "learning_rate": 2.6855145759145523e-05, "loss": 0.0056, "step": 81230 }, { "epoch": 5.637157407534232e-13, "grad_norm": 0.10016711801290512, "learning_rate": 2.684290004876801e-05, "loss": 0.0056, "step": 81240 }, { "epoch": 5.637851296924623e-13, "grad_norm": 0.1138167455792427, "learning_rate": 2.6830660058023496e-05, "loss": 0.0058, "step": 81250 }, { "epoch": 5.638545186315014e-13, "grad_norm": 0.4461274743080139, "learning_rate": 2.681842578816984e-05, "loss": 0.0063, "step": 81260 }, { "epoch": 5.639239075705405e-13, "grad_norm": 0.11171583086252213, "learning_rate": 2.680619724046431e-05, "loss": 0.006, "step": 81270 }, { "epoch": 5.639932965095795e-13, "grad_norm": 0.15235106647014618, "learning_rate": 2.6793974416163564e-05, "loss": 0.0058, "step": 81280 }, { "epoch": 5.640626854486186e-13, "grad_norm": 0.12347110360860825, "learning_rate": 2.678175731652369e-05, "loss": 0.006, "step": 81290 }, { "epoch": 5.641320743876577e-13, "grad_norm": 0.12440482527017593, "learning_rate": 2.676954594280019e-05, "loss": 0.0061, "step": 81300 }, { "epoch": 5.642014633266967e-13, "grad_norm": 0.1161777451634407, "learning_rate": 2.6757340296247976e-05, "loss": 0.0058, "step": 81310 }, { "epoch": 5.642708522657358e-13, "grad_norm": 0.2360321581363678, "learning_rate": 2.6745140378121368e-05, "loss": 0.0062, "step": 81320 }, { "epoch": 5.643402412047749e-13, "grad_norm": 0.13114257156848907, "learning_rate": 2.6732946189674084e-05, "loss": 0.0057, "step": 81330 }, { "epoch": 5.64409630143814e-13, "grad_norm": 0.12523192167282104, "learning_rate": 2.672075773215928e-05, "loss": 0.0056, "step": 81340 }, { "epoch": 5.64479019082853e-13, "grad_norm": 0.11555995792150497, "learning_rate": 2.6708575006829507e-05, "loss": 0.0057, "step": 81350 }, { "epoch": 5.645484080218921e-13, "grad_norm": 0.1450289785861969, "learning_rate": 2.6696398014936737e-05, "loss": 0.0056, "step": 81360 }, { "epoch": 5.646177969609312e-13, "grad_norm": 0.1264878362417221, "learning_rate": 2.668422675773234e-05, "loss": 0.006, "step": 81370 }, { "epoch": 5.646871858999702e-13, "grad_norm": 0.10396093875169754, "learning_rate": 2.6672061236467067e-05, "loss": 0.0055, "step": 81380 }, { "epoch": 5.647565748390093e-13, "grad_norm": 0.11894220858812332, "learning_rate": 2.6659901452391184e-05, "loss": 0.0057, "step": 81390 }, { "epoch": 5.648259637780484e-13, "grad_norm": 0.11545303463935852, "learning_rate": 2.6647747406754245e-05, "loss": 0.0057, "step": 81400 }, { "epoch": 5.648953527170875e-13, "grad_norm": 0.16226238012313843, "learning_rate": 2.6635599100805283e-05, "loss": 0.006, "step": 81410 }, { "epoch": 5.649647416561265e-13, "grad_norm": 0.17892464995384216, "learning_rate": 2.662345653579272e-05, "loss": 0.0063, "step": 81420 }, { "epoch": 5.650341305951656e-13, "grad_norm": 0.10978937149047852, "learning_rate": 2.6611319712964402e-05, "loss": 0.0058, "step": 81430 }, { "epoch": 5.651035195342047e-13, "grad_norm": 0.12240718305110931, "learning_rate": 2.6599188633567568e-05, "loss": 0.0062, "step": 81440 }, { "epoch": 5.651729084732438e-13, "grad_norm": 0.1451931595802307, "learning_rate": 2.6587063298848894e-05, "loss": 0.0064, "step": 81450 }, { "epoch": 5.652422974122828e-13, "grad_norm": 0.1540079116821289, "learning_rate": 2.6574943710054414e-05, "loss": 0.0061, "step": 81460 }, { "epoch": 5.653116863513219e-13, "grad_norm": 0.15449294447898865, "learning_rate": 2.656282986842964e-05, "loss": 0.0059, "step": 81470 }, { "epoch": 5.65381075290361e-13, "grad_norm": 0.11829805374145508, "learning_rate": 2.655072177521943e-05, "loss": 0.0059, "step": 81480 }, { "epoch": 5.654504642294e-13, "grad_norm": 0.118361696600914, "learning_rate": 2.6538619431668094e-05, "loss": 0.0058, "step": 81490 }, { "epoch": 5.655198531684391e-13, "grad_norm": 0.11075611412525177, "learning_rate": 2.6526522839019353e-05, "loss": 0.0057, "step": 81500 }, { "epoch": 5.655892421074782e-13, "grad_norm": 0.1254684031009674, "learning_rate": 2.651443199851625e-05, "loss": 0.0055, "step": 81510 }, { "epoch": 5.656586310465173e-13, "grad_norm": 0.12918786704540253, "learning_rate": 2.650234691140141e-05, "loss": 0.0056, "step": 81520 }, { "epoch": 5.657280199855563e-13, "grad_norm": 0.1346176117658615, "learning_rate": 2.6490267578916664e-05, "loss": 0.0058, "step": 81530 }, { "epoch": 5.657974089245954e-13, "grad_norm": 0.11258888244628906, "learning_rate": 2.647819400230344e-05, "loss": 0.0056, "step": 81540 }, { "epoch": 5.658667978636345e-13, "grad_norm": 0.13366954028606415, "learning_rate": 2.6466126182802404e-05, "loss": 0.006, "step": 81550 }, { "epoch": 5.659361868026735e-13, "grad_norm": 0.1221756562590599, "learning_rate": 2.6454064121653782e-05, "loss": 0.0059, "step": 81560 }, { "epoch": 5.660055757417126e-13, "grad_norm": 0.1604648381471634, "learning_rate": 2.64420078200971e-05, "loss": 0.0059, "step": 81570 }, { "epoch": 5.660749646807517e-13, "grad_norm": 0.12702196836471558, "learning_rate": 2.642995727937133e-05, "loss": 0.0065, "step": 81580 }, { "epoch": 5.661443536197908e-13, "grad_norm": 0.10594471544027328, "learning_rate": 2.6417912500714863e-05, "loss": 0.0058, "step": 81590 }, { "epoch": 5.662137425588298e-13, "grad_norm": 0.1372162401676178, "learning_rate": 2.6405873485365485e-05, "loss": 0.0062, "step": 81600 }, { "epoch": 5.662831314978689e-13, "grad_norm": 0.15675032138824463, "learning_rate": 2.6393840234560388e-05, "loss": 0.0064, "step": 81610 }, { "epoch": 5.66352520436908e-13, "grad_norm": 0.13967403769493103, "learning_rate": 2.6381812749536194e-05, "loss": 0.0059, "step": 81620 }, { "epoch": 5.664219093759471e-13, "grad_norm": 0.13282841444015503, "learning_rate": 2.636979103152889e-05, "loss": 0.0058, "step": 81630 }, { "epoch": 5.664912983149861e-13, "grad_norm": 0.15940356254577637, "learning_rate": 2.63577750817739e-05, "loss": 0.0056, "step": 81640 }, { "epoch": 5.665606872540252e-13, "grad_norm": 0.12812256813049316, "learning_rate": 2.634576490150607e-05, "loss": 0.0056, "step": 81650 }, { "epoch": 5.666300761930643e-13, "grad_norm": 0.11868511885404587, "learning_rate": 2.633376049195961e-05, "loss": 0.0055, "step": 81660 }, { "epoch": 5.666994651321033e-13, "grad_norm": 0.10312440991401672, "learning_rate": 2.6321761854368167e-05, "loss": 0.0055, "step": 81670 }, { "epoch": 5.667688540711424e-13, "grad_norm": 0.13076095283031464, "learning_rate": 2.630976898996479e-05, "loss": 0.0055, "step": 81680 }, { "epoch": 5.668382430101815e-13, "grad_norm": 0.12751919031143188, "learning_rate": 2.6297781899981928e-05, "loss": 0.0055, "step": 81690 }, { "epoch": 5.669076319492206e-13, "grad_norm": 0.10026243329048157, "learning_rate": 2.6285800585651463e-05, "loss": 0.0056, "step": 81700 }, { "epoch": 5.669770208882596e-13, "grad_norm": 0.16613979637622833, "learning_rate": 2.6273825048204612e-05, "loss": 0.0058, "step": 81710 }, { "epoch": 5.670464098272987e-13, "grad_norm": 0.16944630444049835, "learning_rate": 2.6261855288872103e-05, "loss": 0.0058, "step": 81720 }, { "epoch": 5.671157987663378e-13, "grad_norm": 0.13923586905002594, "learning_rate": 2.624989130888396e-05, "loss": 0.006, "step": 81730 }, { "epoch": 5.671851877053768e-13, "grad_norm": 0.15757158398628235, "learning_rate": 2.6237933109469737e-05, "loss": 0.0059, "step": 81740 }, { "epoch": 5.672545766444159e-13, "grad_norm": 0.14428801834583282, "learning_rate": 2.6225980691858244e-05, "loss": 0.0057, "step": 81750 }, { "epoch": 5.67323965583455e-13, "grad_norm": 0.115754634141922, "learning_rate": 2.621403405727786e-05, "loss": 0.0063, "step": 81760 }, { "epoch": 5.673933545224941e-13, "grad_norm": 0.11897885054349899, "learning_rate": 2.6202093206956226e-05, "loss": 0.0059, "step": 81770 }, { "epoch": 5.674627434615331e-13, "grad_norm": 0.15085773169994354, "learning_rate": 2.6190158142120477e-05, "loss": 0.0056, "step": 81780 }, { "epoch": 5.675321324005722e-13, "grad_norm": 0.1487092822790146, "learning_rate": 2.617822886399712e-05, "loss": 0.0056, "step": 81790 }, { "epoch": 5.676015213396113e-13, "grad_norm": 0.1075444370508194, "learning_rate": 2.616630537381207e-05, "loss": 0.0058, "step": 81800 }, { "epoch": 5.676709102786504e-13, "grad_norm": 0.10115885734558105, "learning_rate": 2.615438767279066e-05, "loss": 0.0054, "step": 81810 }, { "epoch": 5.677402992176894e-13, "grad_norm": 0.13528655469417572, "learning_rate": 2.614247576215762e-05, "loss": 0.0055, "step": 81820 }, { "epoch": 5.678096881567285e-13, "grad_norm": 0.13911586999893188, "learning_rate": 2.613056964313708e-05, "loss": 0.0054, "step": 81830 }, { "epoch": 5.678790770957676e-13, "grad_norm": 0.10846641659736633, "learning_rate": 2.6118669316952576e-05, "loss": 0.0052, "step": 81840 }, { "epoch": 5.679484660348066e-13, "grad_norm": 0.11482922732830048, "learning_rate": 2.6106774784827055e-05, "loss": 0.0053, "step": 81850 }, { "epoch": 5.680178549738457e-13, "grad_norm": 0.13881340622901917, "learning_rate": 2.6094886047982856e-05, "loss": 0.0055, "step": 81860 }, { "epoch": 5.680872439128848e-13, "grad_norm": 0.18684600293636322, "learning_rate": 2.608300310764177e-05, "loss": 0.0057, "step": 81870 }, { "epoch": 5.681566328519239e-13, "grad_norm": 0.1240893304347992, "learning_rate": 2.6071125965024884e-05, "loss": 0.0058, "step": 81880 }, { "epoch": 5.682260217909629e-13, "grad_norm": 0.14087598025798798, "learning_rate": 2.6059254621352837e-05, "loss": 0.0059, "step": 81890 }, { "epoch": 5.68295410730002e-13, "grad_norm": 0.08823767304420471, "learning_rate": 2.604738907784554e-05, "loss": 0.0057, "step": 81900 }, { "epoch": 5.683647996690411e-13, "grad_norm": 0.1318943202495575, "learning_rate": 2.6035529335722374e-05, "loss": 0.0061, "step": 81910 }, { "epoch": 5.684341886080801e-13, "grad_norm": 0.14446063339710236, "learning_rate": 2.6023675396202122e-05, "loss": 0.0057, "step": 81920 }, { "epoch": 5.685035775471192e-13, "grad_norm": 0.14688636362552643, "learning_rate": 2.6011827260502955e-05, "loss": 0.0058, "step": 81930 }, { "epoch": 5.685729664861583e-13, "grad_norm": 0.11903318017721176, "learning_rate": 2.599998492984246e-05, "loss": 0.0056, "step": 81940 }, { "epoch": 5.686423554251974e-13, "grad_norm": 0.1487322896718979, "learning_rate": 2.5988148405437603e-05, "loss": 0.0056, "step": 81950 }, { "epoch": 5.687117443642364e-13, "grad_norm": 0.12663298845291138, "learning_rate": 2.5976317688504796e-05, "loss": 0.0056, "step": 81960 }, { "epoch": 5.687811333032755e-13, "grad_norm": 0.09817813336849213, "learning_rate": 2.5964492780259802e-05, "loss": 0.0056, "step": 81970 }, { "epoch": 5.688505222423146e-13, "grad_norm": 0.1441580057144165, "learning_rate": 2.5952673681917836e-05, "loss": 0.0055, "step": 81980 }, { "epoch": 5.689199111813537e-13, "grad_norm": 0.1557295173406601, "learning_rate": 2.594086039469349e-05, "loss": 0.0054, "step": 81990 }, { "epoch": 5.689893001203927e-13, "grad_norm": 0.08948507905006409, "learning_rate": 2.592905291980076e-05, "loss": 0.0053, "step": 82000 }, { "epoch": 5.690586890594318e-13, "grad_norm": 0.13117560744285583, "learning_rate": 2.5917251258453038e-05, "loss": 0.0053, "step": 82010 }, { "epoch": 5.691280779984709e-13, "grad_norm": 0.1262354999780655, "learning_rate": 2.5905455411863145e-05, "loss": 0.0058, "step": 82020 }, { "epoch": 5.691974669375099e-13, "grad_norm": 0.27364814281463623, "learning_rate": 2.58936653812433e-05, "loss": 0.0061, "step": 82030 }, { "epoch": 5.69266855876549e-13, "grad_norm": 0.1449144333600998, "learning_rate": 2.588188116780506e-05, "loss": 0.0057, "step": 82040 }, { "epoch": 5.693362448155881e-13, "grad_norm": 0.12951169908046722, "learning_rate": 2.5870102772759495e-05, "loss": 0.0057, "step": 82050 }, { "epoch": 5.694056337546272e-13, "grad_norm": 0.11168958246707916, "learning_rate": 2.5858330197316964e-05, "loss": 0.0057, "step": 82060 }, { "epoch": 5.694750226936662e-13, "grad_norm": 0.15003499388694763, "learning_rate": 2.5846563442687336e-05, "loss": 0.0059, "step": 82070 }, { "epoch": 5.695444116327053e-13, "grad_norm": 0.17909179627895355, "learning_rate": 2.5834802510079782e-05, "loss": 0.0057, "step": 82080 }, { "epoch": 5.696138005717444e-13, "grad_norm": 0.1270732283592224, "learning_rate": 2.5823047400702968e-05, "loss": 0.0057, "step": 82090 }, { "epoch": 5.696831895107834e-13, "grad_norm": 0.11839916557073593, "learning_rate": 2.5811298115764853e-05, "loss": 0.0055, "step": 82100 }, { "epoch": 5.697525784498225e-13, "grad_norm": 0.13151948153972626, "learning_rate": 2.579955465647293e-05, "loss": 0.0057, "step": 82110 }, { "epoch": 5.698219673888616e-13, "grad_norm": 0.11534707248210907, "learning_rate": 2.578781702403397e-05, "loss": 0.0054, "step": 82120 }, { "epoch": 5.698913563279007e-13, "grad_norm": 0.12176872044801712, "learning_rate": 2.577608521965422e-05, "loss": 0.006, "step": 82130 }, { "epoch": 5.699607452669397e-13, "grad_norm": 0.12070873379707336, "learning_rate": 2.5764359244539294e-05, "loss": 0.0052, "step": 82140 }, { "epoch": 5.700301342059788e-13, "grad_norm": 0.15020188689231873, "learning_rate": 2.5752639099894222e-05, "loss": 0.0054, "step": 82150 }, { "epoch": 5.700995231450179e-13, "grad_norm": 0.1340407133102417, "learning_rate": 2.5740924786923425e-05, "loss": 0.0056, "step": 82160 }, { "epoch": 5.70168912084057e-13, "grad_norm": 0.11092182993888855, "learning_rate": 2.572921630683075e-05, "loss": 0.0055, "step": 82170 }, { "epoch": 5.70238301023096e-13, "grad_norm": 0.10950350761413574, "learning_rate": 2.5717513660819406e-05, "loss": 0.0057, "step": 82180 }, { "epoch": 5.703076899621351e-13, "grad_norm": 0.12667876482009888, "learning_rate": 2.570581685009203e-05, "loss": 0.0059, "step": 82190 }, { "epoch": 5.703770789011742e-13, "grad_norm": 0.11900647729635239, "learning_rate": 2.5694125875850662e-05, "loss": 0.0055, "step": 82200 }, { "epoch": 5.704464678402132e-13, "grad_norm": 0.1409771740436554, "learning_rate": 2.5682440739296715e-05, "loss": 0.0058, "step": 82210 }, { "epoch": 5.705158567792523e-13, "grad_norm": 0.18886902928352356, "learning_rate": 2.5670761441631024e-05, "loss": 0.0058, "step": 82220 }, { "epoch": 5.705852457182914e-13, "grad_norm": 0.122271828353405, "learning_rate": 2.565908798405383e-05, "loss": 0.0056, "step": 82230 }, { "epoch": 5.706546346573305e-13, "grad_norm": 0.2687402665615082, "learning_rate": 2.5647420367764724e-05, "loss": 0.0059, "step": 82240 }, { "epoch": 5.707240235963695e-13, "grad_norm": 0.17731662094593048, "learning_rate": 2.5635758593962794e-05, "loss": 0.0057, "step": 82250 }, { "epoch": 5.707934125354086e-13, "grad_norm": 0.15623392164707184, "learning_rate": 2.56241026638464e-05, "loss": 0.0056, "step": 82260 }, { "epoch": 5.708628014744477e-13, "grad_norm": 0.12163075059652328, "learning_rate": 2.561245257861345e-05, "loss": 0.0057, "step": 82270 }, { "epoch": 5.709321904134868e-13, "grad_norm": 0.12667131423950195, "learning_rate": 2.5600808339461084e-05, "loss": 0.0055, "step": 82280 }, { "epoch": 5.710015793525258e-13, "grad_norm": 0.1397876888513565, "learning_rate": 2.5589169947586015e-05, "loss": 0.0054, "step": 82290 }, { "epoch": 5.710709682915649e-13, "grad_norm": 0.11956975609064102, "learning_rate": 2.5577537404184195e-05, "loss": 0.0053, "step": 82300 }, { "epoch": 5.71140357230604e-13, "grad_norm": 0.1894119679927826, "learning_rate": 2.5565910710451106e-05, "loss": 0.0053, "step": 82310 }, { "epoch": 5.71209746169643e-13, "grad_norm": 0.12175155431032181, "learning_rate": 2.5554289867581536e-05, "loss": 0.0055, "step": 82320 }, { "epoch": 5.712791351086821e-13, "grad_norm": 0.09441204369068146, "learning_rate": 2.5542674876769724e-05, "loss": 0.0056, "step": 82330 }, { "epoch": 5.713485240477212e-13, "grad_norm": 0.12958063185214996, "learning_rate": 2.5531065739209276e-05, "loss": 0.0058, "step": 82340 }, { "epoch": 5.714179129867603e-13, "grad_norm": 0.11670687049627304, "learning_rate": 2.5519462456093223e-05, "loss": 0.0057, "step": 82350 }, { "epoch": 5.714873019257993e-13, "grad_norm": 0.1410544216632843, "learning_rate": 2.5507865028613997e-05, "loss": 0.0056, "step": 82360 }, { "epoch": 5.715566908648384e-13, "grad_norm": 0.1306973248720169, "learning_rate": 2.5496273457963367e-05, "loss": 0.0057, "step": 82370 }, { "epoch": 5.716260798038775e-13, "grad_norm": 0.11605740338563919, "learning_rate": 2.5484687745332606e-05, "loss": 0.0057, "step": 82380 }, { "epoch": 5.716954687429165e-13, "grad_norm": 0.13498954474925995, "learning_rate": 2.5473107891912274e-05, "loss": 0.0057, "step": 82390 }, { "epoch": 5.717648576819556e-13, "grad_norm": 0.13145482540130615, "learning_rate": 2.5461533898892435e-05, "loss": 0.0061, "step": 82400 }, { "epoch": 5.718342466209947e-13, "grad_norm": 0.17135752737522125, "learning_rate": 2.5449965767462435e-05, "loss": 0.0058, "step": 82410 }, { "epoch": 5.719036355600338e-13, "grad_norm": 0.11076022684574127, "learning_rate": 2.5438403498811145e-05, "loss": 0.0056, "step": 82420 }, { "epoch": 5.719730244990728e-13, "grad_norm": 0.10168895125389099, "learning_rate": 2.5426847094126696e-05, "loss": 0.0058, "step": 82430 }, { "epoch": 5.720424134381119e-13, "grad_norm": 0.1476949006319046, "learning_rate": 2.541529655459676e-05, "loss": 0.0056, "step": 82440 }, { "epoch": 5.72111802377151e-13, "grad_norm": 0.09558334201574326, "learning_rate": 2.5403751881408294e-05, "loss": 0.0053, "step": 82450 }, { "epoch": 5.721811913161901e-13, "grad_norm": 0.12882906198501587, "learning_rate": 2.53922130757477e-05, "loss": 0.0054, "step": 82460 }, { "epoch": 5.722505802552291e-13, "grad_norm": 0.23248997330665588, "learning_rate": 2.538068013880076e-05, "loss": 0.0057, "step": 82470 }, { "epoch": 5.723199691942682e-13, "grad_norm": 0.12213397026062012, "learning_rate": 2.5369153071752685e-05, "loss": 0.0055, "step": 82480 }, { "epoch": 5.723893581333073e-13, "grad_norm": 0.09593990445137024, "learning_rate": 2.5357631875788043e-05, "loss": 0.0058, "step": 82490 }, { "epoch": 5.724587470723463e-13, "grad_norm": 0.14831571280956268, "learning_rate": 2.534611655209082e-05, "loss": 0.0056, "step": 82500 }, { "epoch": 5.725281360113854e-13, "grad_norm": 0.14816239476203918, "learning_rate": 2.5334607101844398e-05, "loss": 0.0055, "step": 82510 }, { "epoch": 5.725975249504245e-13, "grad_norm": 0.10532985627651215, "learning_rate": 2.532310352623155e-05, "loss": 0.0058, "step": 82520 }, { "epoch": 5.726669138894636e-13, "grad_norm": 0.10291223973035812, "learning_rate": 2.5311605826434455e-05, "loss": 0.0055, "step": 82530 }, { "epoch": 5.727363028285026e-13, "grad_norm": 0.13229230046272278, "learning_rate": 2.5300114003634666e-05, "loss": 0.0059, "step": 82540 }, { "epoch": 5.728056917675417e-13, "grad_norm": 0.11346873641014099, "learning_rate": 2.5288628059013162e-05, "loss": 0.0055, "step": 82550 }, { "epoch": 5.728750807065808e-13, "grad_norm": 0.11785896122455597, "learning_rate": 2.5277147993750296e-05, "loss": 0.0057, "step": 82560 }, { "epoch": 5.729444696456198e-13, "grad_norm": 0.11003702878952026, "learning_rate": 2.52656738090258e-05, "loss": 0.0059, "step": 82570 }, { "epoch": 5.730138585846589e-13, "grad_norm": 0.12334344536066055, "learning_rate": 2.5254205506018885e-05, "loss": 0.0055, "step": 82580 }, { "epoch": 5.73083247523698e-13, "grad_norm": 0.12919360399246216, "learning_rate": 2.524274308590802e-05, "loss": 0.0056, "step": 82590 }, { "epoch": 5.731526364627371e-13, "grad_norm": 0.09679032862186432, "learning_rate": 2.5231286549871223e-05, "loss": 0.0053, "step": 82600 }, { "epoch": 5.732220254017761e-13, "grad_norm": 0.148740753531456, "learning_rate": 2.5219835899085756e-05, "loss": 0.0053, "step": 82610 }, { "epoch": 5.732914143408152e-13, "grad_norm": 0.11697467416524887, "learning_rate": 2.520839113472843e-05, "loss": 0.0054, "step": 82620 }, { "epoch": 5.733608032798543e-13, "grad_norm": 0.19641315937042236, "learning_rate": 2.5196952257975293e-05, "loss": 0.0057, "step": 82630 }, { "epoch": 5.734301922188934e-13, "grad_norm": 0.1280863881111145, "learning_rate": 2.5185519270001937e-05, "loss": 0.0056, "step": 82640 }, { "epoch": 5.734995811579324e-13, "grad_norm": 0.09259147197008133, "learning_rate": 2.5174092171983238e-05, "loss": 0.0058, "step": 82650 }, { "epoch": 5.735689700969715e-13, "grad_norm": 0.11149263381958008, "learning_rate": 2.516267096509353e-05, "loss": 0.0056, "step": 82660 }, { "epoch": 5.736383590360106e-13, "grad_norm": 0.1637006253004074, "learning_rate": 2.5151255650506502e-05, "loss": 0.0055, "step": 82670 }, { "epoch": 5.737077479750496e-13, "grad_norm": 0.12141619622707367, "learning_rate": 2.513984622939526e-05, "loss": 0.0059, "step": 82680 }, { "epoch": 5.737771369140887e-13, "grad_norm": 0.11608537286520004, "learning_rate": 2.512844270293231e-05, "loss": 0.0055, "step": 82690 }, { "epoch": 5.738465258531278e-13, "grad_norm": 0.1483110934495926, "learning_rate": 2.5117045072289525e-05, "loss": 0.0058, "step": 82700 }, { "epoch": 5.739159147921669e-13, "grad_norm": 0.12329035997390747, "learning_rate": 2.5105653338638203e-05, "loss": 0.0056, "step": 82710 }, { "epoch": 5.739853037312059e-13, "grad_norm": 0.13333217799663544, "learning_rate": 2.5094267503149024e-05, "loss": 0.0059, "step": 82720 }, { "epoch": 5.74054692670245e-13, "grad_norm": 0.12189057469367981, "learning_rate": 2.508288756699204e-05, "loss": 0.0057, "step": 82730 }, { "epoch": 5.741240816092841e-13, "grad_norm": 0.11234313994646072, "learning_rate": 2.5071513531336742e-05, "loss": 0.0054, "step": 82740 }, { "epoch": 5.741934705483231e-13, "grad_norm": 0.10292470455169678, "learning_rate": 2.506014539735197e-05, "loss": 0.0056, "step": 82750 }, { "epoch": 5.742628594873622e-13, "grad_norm": 0.10312889516353607, "learning_rate": 2.5048783166205984e-05, "loss": 0.0054, "step": 82760 }, { "epoch": 5.743322484264013e-13, "grad_norm": 0.10033424943685532, "learning_rate": 2.5037426839066443e-05, "loss": 0.0055, "step": 82770 }, { "epoch": 5.744016373654404e-13, "grad_norm": 0.1216273382306099, "learning_rate": 2.502607641710036e-05, "loss": 0.0055, "step": 82780 }, { "epoch": 5.744710263044794e-13, "grad_norm": 0.20144037902355194, "learning_rate": 2.501473190147418e-05, "loss": 0.0058, "step": 82790 }, { "epoch": 5.745404152435185e-13, "grad_norm": 0.15246671438217163, "learning_rate": 2.500339329335372e-05, "loss": 0.0058, "step": 82800 }, { "epoch": 5.746098041825576e-13, "grad_norm": 0.09216224402189255, "learning_rate": 2.4992060593904217e-05, "loss": 0.0057, "step": 82810 }, { "epoch": 5.746791931215967e-13, "grad_norm": 0.13680365681648254, "learning_rate": 2.4980733804290275e-05, "loss": 0.0056, "step": 82820 }, { "epoch": 5.747485820606357e-13, "grad_norm": 0.16719937324523926, "learning_rate": 2.4969412925675882e-05, "loss": 0.006, "step": 82830 }, { "epoch": 5.748179709996748e-13, "grad_norm": 0.15831628441810608, "learning_rate": 2.4958097959224456e-05, "loss": 0.006, "step": 82840 }, { "epoch": 5.748873599387139e-13, "grad_norm": 0.13587641716003418, "learning_rate": 2.4946788906098778e-05, "loss": 0.006, "step": 82850 }, { "epoch": 5.749567488777529e-13, "grad_norm": 0.14593078196048737, "learning_rate": 2.4935485767461027e-05, "loss": 0.0058, "step": 82860 }, { "epoch": 5.75026137816792e-13, "grad_norm": 0.14503946900367737, "learning_rate": 2.4924188544472773e-05, "loss": 0.0055, "step": 82870 }, { "epoch": 5.750955267558311e-13, "grad_norm": 0.15353858470916748, "learning_rate": 2.4912897238294995e-05, "loss": 0.0059, "step": 82880 }, { "epoch": 5.751649156948702e-13, "grad_norm": 0.10045052319765091, "learning_rate": 2.4901611850088055e-05, "loss": 0.0055, "step": 82890 }, { "epoch": 5.752343046339092e-13, "grad_norm": 0.1355588287115097, "learning_rate": 2.489033238101165e-05, "loss": 0.0054, "step": 82900 }, { "epoch": 5.753036935729483e-13, "grad_norm": 0.18168626725673676, "learning_rate": 2.4879058832224997e-05, "loss": 0.0057, "step": 82910 }, { "epoch": 5.753730825119874e-13, "grad_norm": 0.10890203714370728, "learning_rate": 2.4867791204886558e-05, "loss": 0.0055, "step": 82920 }, { "epoch": 5.754424714510264e-13, "grad_norm": 0.10620288550853729, "learning_rate": 2.4856529500154325e-05, "loss": 0.0054, "step": 82930 }, { "epoch": 5.755118603900655e-13, "grad_norm": 0.18084396421909332, "learning_rate": 2.484527371918555e-05, "loss": 0.0059, "step": 82940 }, { "epoch": 5.755812493291046e-13, "grad_norm": 0.16506589949131012, "learning_rate": 2.483402386313699e-05, "loss": 0.0059, "step": 82950 }, { "epoch": 5.756506382681437e-13, "grad_norm": 0.1479841023683548, "learning_rate": 2.4822779933164693e-05, "loss": 0.0058, "step": 82960 }, { "epoch": 5.757200272071827e-13, "grad_norm": 0.12195029109716415, "learning_rate": 2.4811541930424205e-05, "loss": 0.0058, "step": 82970 }, { "epoch": 5.757894161462218e-13, "grad_norm": 0.12041187286376953, "learning_rate": 2.4800309856070367e-05, "loss": 0.0058, "step": 82980 }, { "epoch": 5.758588050852609e-13, "grad_norm": 0.1437549591064453, "learning_rate": 2.478908371125746e-05, "loss": 0.0058, "step": 82990 }, { "epoch": 5.759281940243e-13, "grad_norm": 0.1352679282426834, "learning_rate": 2.4777863497139138e-05, "loss": 0.0059, "step": 83000 }, { "epoch": 5.75997582963339e-13, "grad_norm": 0.10496477037668228, "learning_rate": 2.4766649214868455e-05, "loss": 0.0059, "step": 83010 }, { "epoch": 5.760669719023781e-13, "grad_norm": 0.15053308010101318, "learning_rate": 2.4755440865597873e-05, "loss": 0.0057, "step": 83020 }, { "epoch": 5.761363608414172e-13, "grad_norm": 0.13104689121246338, "learning_rate": 2.4744238450479192e-05, "loss": 0.0055, "step": 83030 }, { "epoch": 5.762057497804562e-13, "grad_norm": 0.1150982454419136, "learning_rate": 2.473304197066366e-05, "loss": 0.0055, "step": 83040 }, { "epoch": 5.762751387194953e-13, "grad_norm": 0.11092741042375565, "learning_rate": 2.4721851427301875e-05, "loss": 0.0055, "step": 83050 }, { "epoch": 5.763445276585344e-13, "grad_norm": 0.13655385375022888, "learning_rate": 2.4710666821543843e-05, "loss": 0.0054, "step": 83060 }, { "epoch": 5.764139165975735e-13, "grad_norm": 0.11347392201423645, "learning_rate": 2.4699488154538956e-05, "loss": 0.0052, "step": 83070 }, { "epoch": 5.764833055366125e-13, "grad_norm": 0.1553630381822586, "learning_rate": 2.468831542743601e-05, "loss": 0.0055, "step": 83080 }, { "epoch": 5.765526944756516e-13, "grad_norm": 0.12430556863546371, "learning_rate": 2.4677148641383152e-05, "loss": 0.0054, "step": 83090 }, { "epoch": 5.766220834146907e-13, "grad_norm": 0.17808715999126434, "learning_rate": 2.4665987797527953e-05, "loss": 0.0058, "step": 83100 }, { "epoch": 5.766914723537298e-13, "grad_norm": 0.1165616363286972, "learning_rate": 2.465483289701739e-05, "loss": 0.0061, "step": 83110 }, { "epoch": 5.767608612927688e-13, "grad_norm": 0.11884475499391556, "learning_rate": 2.4643683940997737e-05, "loss": 0.0059, "step": 83120 }, { "epoch": 5.768302502318079e-13, "grad_norm": 0.12908373773097992, "learning_rate": 2.4632540930614805e-05, "loss": 0.0058, "step": 83130 }, { "epoch": 5.76899639170847e-13, "grad_norm": 0.12657895684242249, "learning_rate": 2.4621403867013634e-05, "loss": 0.0056, "step": 83140 }, { "epoch": 5.76969028109886e-13, "grad_norm": 0.16590961813926697, "learning_rate": 2.46102727513388e-05, "loss": 0.0058, "step": 83150 }, { "epoch": 5.770384170489251e-13, "grad_norm": 0.1508064866065979, "learning_rate": 2.459914758473413e-05, "loss": 0.0058, "step": 83160 }, { "epoch": 5.771078059879642e-13, "grad_norm": 0.12749837338924408, "learning_rate": 2.4588028368342972e-05, "loss": 0.0058, "step": 83170 }, { "epoch": 5.771771949270033e-13, "grad_norm": 0.11491411924362183, "learning_rate": 2.4576915103307963e-05, "loss": 0.0056, "step": 83180 }, { "epoch": 5.772465838660423e-13, "grad_norm": 0.12217976897954941, "learning_rate": 2.4565807790771173e-05, "loss": 0.0055, "step": 83190 }, { "epoch": 5.773159728050814e-13, "grad_norm": 0.11131974309682846, "learning_rate": 2.4554706431874036e-05, "loss": 0.0056, "step": 83200 }, { "epoch": 5.773853617441205e-13, "grad_norm": 0.13831539452075958, "learning_rate": 2.4543611027757415e-05, "loss": 0.0056, "step": 83210 }, { "epoch": 5.774547506831595e-13, "grad_norm": 0.12075912207365036, "learning_rate": 2.4532521579561532e-05, "loss": 0.0053, "step": 83220 }, { "epoch": 5.775241396221986e-13, "grad_norm": 0.13579872250556946, "learning_rate": 2.452143808842596e-05, "loss": 0.0055, "step": 83230 }, { "epoch": 5.775935285612377e-13, "grad_norm": 0.11442164331674576, "learning_rate": 2.4510360555489762e-05, "loss": 0.0054, "step": 83240 }, { "epoch": 5.776629175002768e-13, "grad_norm": 0.1531831920146942, "learning_rate": 2.4499288981891267e-05, "loss": 0.0053, "step": 83250 }, { "epoch": 5.777323064393158e-13, "grad_norm": 0.16944897174835205, "learning_rate": 2.44882233687683e-05, "loss": 0.0062, "step": 83260 }, { "epoch": 5.778016953783549e-13, "grad_norm": 0.21415963768959045, "learning_rate": 2.4477163717257988e-05, "loss": 0.0064, "step": 83270 }, { "epoch": 5.77871084317394e-13, "grad_norm": 0.12346015125513077, "learning_rate": 2.4466110028496926e-05, "loss": 0.0056, "step": 83280 }, { "epoch": 5.779404732564331e-13, "grad_norm": 0.13482655584812164, "learning_rate": 2.4455062303620998e-05, "loss": 0.0059, "step": 83290 }, { "epoch": 5.780098621954721e-13, "grad_norm": 0.115457683801651, "learning_rate": 2.4444020543765587e-05, "loss": 0.0056, "step": 83300 }, { "epoch": 5.780792511345112e-13, "grad_norm": 0.11905848234891891, "learning_rate": 2.443298475006536e-05, "loss": 0.0056, "step": 83310 }, { "epoch": 5.781486400735503e-13, "grad_norm": 0.10660360753536224, "learning_rate": 2.4421954923654438e-05, "loss": 0.0057, "step": 83320 }, { "epoch": 5.782180290125893e-13, "grad_norm": 0.10765537619590759, "learning_rate": 2.441093106566629e-05, "loss": 0.0056, "step": 83330 }, { "epoch": 5.782874179516284e-13, "grad_norm": 0.10218805819749832, "learning_rate": 2.439991317723381e-05, "loss": 0.0055, "step": 83340 }, { "epoch": 5.783568068906675e-13, "grad_norm": 0.10455664992332458, "learning_rate": 2.438890125948925e-05, "loss": 0.0055, "step": 83350 }, { "epoch": 5.784261958297066e-13, "grad_norm": 0.10468633472919464, "learning_rate": 2.4377895313564247e-05, "loss": 0.0055, "step": 83360 }, { "epoch": 5.784955847687456e-13, "grad_norm": 0.11417359113693237, "learning_rate": 2.4366895340589846e-05, "loss": 0.0056, "step": 83370 }, { "epoch": 5.785649737077847e-13, "grad_norm": 0.10912920534610748, "learning_rate": 2.435590134169646e-05, "loss": 0.0052, "step": 83380 }, { "epoch": 5.786343626468238e-13, "grad_norm": 0.1351131647825241, "learning_rate": 2.4344913318013895e-05, "loss": 0.0054, "step": 83390 }, { "epoch": 5.787037515858628e-13, "grad_norm": 0.11834019422531128, "learning_rate": 2.4333931270671336e-05, "loss": 0.0055, "step": 83400 }, { "epoch": 5.787731405249019e-13, "grad_norm": 0.10847096890211105, "learning_rate": 2.4322955200797364e-05, "loss": 0.0054, "step": 83410 }, { "epoch": 5.78842529463941e-13, "grad_norm": 0.1349087953567505, "learning_rate": 2.431198510951994e-05, "loss": 0.0063, "step": 83420 }, { "epoch": 5.789119184029801e-13, "grad_norm": 0.12164418399333954, "learning_rate": 2.4301020997966407e-05, "loss": 0.0058, "step": 83430 }, { "epoch": 5.789813073420191e-13, "grad_norm": 0.09310482442378998, "learning_rate": 2.429006286726353e-05, "loss": 0.0059, "step": 83440 }, { "epoch": 5.790506962810582e-13, "grad_norm": 0.12355178594589233, "learning_rate": 2.4279110718537358e-05, "loss": 0.0056, "step": 83450 }, { "epoch": 5.791200852200973e-13, "grad_norm": 0.11077497899532318, "learning_rate": 2.4268164552913473e-05, "loss": 0.0058, "step": 83460 }, { "epoch": 5.791894741591364e-13, "grad_norm": 0.1208110973238945, "learning_rate": 2.42572243715167e-05, "loss": 0.0057, "step": 83470 }, { "epoch": 5.792588630981754e-13, "grad_norm": 0.11721932142972946, "learning_rate": 2.4246290175471364e-05, "loss": 0.0055, "step": 83480 }, { "epoch": 5.793282520372145e-13, "grad_norm": 0.13298161327838898, "learning_rate": 2.4235361965901074e-05, "loss": 0.0055, "step": 83490 }, { "epoch": 5.793976409762536e-13, "grad_norm": 0.13487011194229126, "learning_rate": 2.4224439743928934e-05, "loss": 0.0057, "step": 83500 }, { "epoch": 5.794670299152926e-13, "grad_norm": 0.13626858592033386, "learning_rate": 2.4213523510677317e-05, "loss": 0.0055, "step": 83510 }, { "epoch": 5.795364188543317e-13, "grad_norm": 0.1504824161529541, "learning_rate": 2.4202613267268053e-05, "loss": 0.0057, "step": 83520 }, { "epoch": 5.796058077933708e-13, "grad_norm": 0.1364983171224594, "learning_rate": 2.4191709014822354e-05, "loss": 0.0055, "step": 83530 }, { "epoch": 5.796751967324099e-13, "grad_norm": 0.140411838889122, "learning_rate": 2.418081075446078e-05, "loss": 0.0054, "step": 83540 }, { "epoch": 5.797445856714489e-13, "grad_norm": 0.1119951531291008, "learning_rate": 2.4169918487303316e-05, "loss": 0.0054, "step": 83550 }, { "epoch": 5.79813974610488e-13, "grad_norm": 0.133401021361351, "learning_rate": 2.41590322144693e-05, "loss": 0.0054, "step": 83560 }, { "epoch": 5.798833635495271e-13, "grad_norm": 0.14199066162109375, "learning_rate": 2.414815193707747e-05, "loss": 0.0055, "step": 83570 }, { "epoch": 5.799527524885661e-13, "grad_norm": 0.1595320999622345, "learning_rate": 2.413727765624594e-05, "loss": 0.006, "step": 83580 }, { "epoch": 5.800221414276052e-13, "grad_norm": 0.15795451402664185, "learning_rate": 2.412640937309222e-05, "loss": 0.006, "step": 83590 }, { "epoch": 5.800915303666443e-13, "grad_norm": 0.11795055121183395, "learning_rate": 2.4115547088733188e-05, "loss": 0.0056, "step": 83600 }, { "epoch": 5.801609193056834e-13, "grad_norm": 0.14862145483493805, "learning_rate": 2.4104690804285104e-05, "loss": 0.0058, "step": 83610 }, { "epoch": 5.802303082447224e-13, "grad_norm": 0.11756731569766998, "learning_rate": 2.409384052086365e-05, "loss": 0.0055, "step": 83620 }, { "epoch": 5.802996971837615e-13, "grad_norm": 0.1389184594154358, "learning_rate": 2.4082996239583834e-05, "loss": 0.0055, "step": 83630 }, { "epoch": 5.803690861228006e-13, "grad_norm": 0.12522554397583008, "learning_rate": 2.407215796156009e-05, "loss": 0.0057, "step": 83640 }, { "epoch": 5.804384750618397e-13, "grad_norm": 0.18189173936843872, "learning_rate": 2.4061325687906198e-05, "loss": 0.0057, "step": 83650 }, { "epoch": 5.805078640008787e-13, "grad_norm": 0.11092527955770493, "learning_rate": 2.4050499419735373e-05, "loss": 0.0054, "step": 83660 }, { "epoch": 5.805772529399178e-13, "grad_norm": 0.10050440579652786, "learning_rate": 2.4039679158160156e-05, "loss": 0.0056, "step": 83670 }, { "epoch": 5.806466418789569e-13, "grad_norm": 0.13347087800502777, "learning_rate": 2.4028864904292508e-05, "loss": 0.0055, "step": 83680 }, { "epoch": 5.807160308179959e-13, "grad_norm": 0.10837093740701675, "learning_rate": 2.401805665924377e-05, "loss": 0.0054, "step": 83690 }, { "epoch": 5.80785419757035e-13, "grad_norm": 0.11939991265535355, "learning_rate": 2.4007254424124637e-05, "loss": 0.0054, "step": 83700 }, { "epoch": 5.808548086960741e-13, "grad_norm": 0.11787251383066177, "learning_rate": 2.399645820004523e-05, "loss": 0.0055, "step": 83710 }, { "epoch": 5.809241976351132e-13, "grad_norm": 0.11240992695093155, "learning_rate": 2.398566798811502e-05, "loss": 0.0054, "step": 83720 }, { "epoch": 5.809935865741522e-13, "grad_norm": 0.1327337771654129, "learning_rate": 2.397488378944285e-05, "loss": 0.0054, "step": 83730 }, { "epoch": 5.810629755131913e-13, "grad_norm": 0.12886744737625122, "learning_rate": 2.3964105605136994e-05, "loss": 0.0058, "step": 83740 }, { "epoch": 5.811323644522304e-13, "grad_norm": 0.14235782623291016, "learning_rate": 2.3953333436305065e-05, "loss": 0.0061, "step": 83750 }, { "epoch": 5.812017533912694e-13, "grad_norm": 0.1039617732167244, "learning_rate": 2.394256728405406e-05, "loss": 0.0056, "step": 83760 }, { "epoch": 5.812711423303085e-13, "grad_norm": 0.10022851079702377, "learning_rate": 2.3931807149490408e-05, "loss": 0.0056, "step": 83770 }, { "epoch": 5.813405312693476e-13, "grad_norm": 0.09031115472316742, "learning_rate": 2.392105303371982e-05, "loss": 0.0054, "step": 83780 }, { "epoch": 5.814099202083867e-13, "grad_norm": 0.14309801161289215, "learning_rate": 2.391030493784751e-05, "loss": 0.0057, "step": 83790 }, { "epoch": 5.814793091474257e-13, "grad_norm": 0.10110513865947723, "learning_rate": 2.389956286297795e-05, "loss": 0.0055, "step": 83800 }, { "epoch": 5.815486980864648e-13, "grad_norm": 0.11554224789142609, "learning_rate": 2.3888826810215127e-05, "loss": 0.0057, "step": 83810 }, { "epoch": 5.816180870255039e-13, "grad_norm": 0.11934128403663635, "learning_rate": 2.387809678066226e-05, "loss": 0.0054, "step": 83820 }, { "epoch": 5.81687475964543e-13, "grad_norm": 0.1301143318414688, "learning_rate": 2.3867372775422103e-05, "loss": 0.0056, "step": 83830 }, { "epoch": 5.81756864903582e-13, "grad_norm": 0.12066357582807541, "learning_rate": 2.3856654795596654e-05, "loss": 0.0055, "step": 83840 }, { "epoch": 5.818262538426211e-13, "grad_norm": 0.10354752093553543, "learning_rate": 2.384594284228738e-05, "loss": 0.0053, "step": 83850 }, { "epoch": 5.818956427816602e-13, "grad_norm": 0.12822121381759644, "learning_rate": 2.383523691659509e-05, "loss": 0.0054, "step": 83860 }, { "epoch": 5.819650317206992e-13, "grad_norm": 0.1326318234205246, "learning_rate": 2.3824537019619993e-05, "loss": 0.0054, "step": 83870 }, { "epoch": 5.820344206597383e-13, "grad_norm": 0.12342336028814316, "learning_rate": 2.381384315246167e-05, "loss": 0.0054, "step": 83880 }, { "epoch": 5.821038095987774e-13, "grad_norm": 0.1392608880996704, "learning_rate": 2.380315531621908e-05, "loss": 0.0057, "step": 83890 }, { "epoch": 5.821731985378165e-13, "grad_norm": 0.13779962062835693, "learning_rate": 2.3792473511990565e-05, "loss": 0.0057, "step": 83900 }, { "epoch": 5.822425874768555e-13, "grad_norm": 0.13159480690956116, "learning_rate": 2.378179774087384e-05, "loss": 0.0063, "step": 83910 }, { "epoch": 5.823119764158946e-13, "grad_norm": 0.14912280440330505, "learning_rate": 2.377112800396601e-05, "loss": 0.0057, "step": 83920 }, { "epoch": 5.823813653549337e-13, "grad_norm": 0.13414739072322845, "learning_rate": 2.3760464302363575e-05, "loss": 0.0056, "step": 83930 }, { "epoch": 5.824507542939728e-13, "grad_norm": 0.10736484825611115, "learning_rate": 2.3749806637162358e-05, "loss": 0.0054, "step": 83940 }, { "epoch": 5.825201432330118e-13, "grad_norm": 0.1185276210308075, "learning_rate": 2.3739155009457636e-05, "loss": 0.0056, "step": 83950 }, { "epoch": 5.825895321720509e-13, "grad_norm": 0.11778222024440765, "learning_rate": 2.3728509420344016e-05, "loss": 0.0055, "step": 83960 }, { "epoch": 5.8265892111109e-13, "grad_norm": 0.10191567242145538, "learning_rate": 2.37178698709155e-05, "loss": 0.0055, "step": 83970 }, { "epoch": 5.82728310050129e-13, "grad_norm": 0.11124271154403687, "learning_rate": 2.3707236362265445e-05, "loss": 0.0055, "step": 83980 }, { "epoch": 5.827976989891681e-13, "grad_norm": 0.12280629575252533, "learning_rate": 2.369660889548666e-05, "loss": 0.0055, "step": 83990 }, { "epoch": 5.828670879282072e-13, "grad_norm": 0.10051633417606354, "learning_rate": 2.3685987471671216e-05, "loss": 0.0054, "step": 84000 }, { "epoch": 5.829364768672463e-13, "grad_norm": 0.13805125653743744, "learning_rate": 2.3675372091910692e-05, "loss": 0.0053, "step": 84010 }, { "epoch": 5.830058658062853e-13, "grad_norm": 0.11398834735155106, "learning_rate": 2.3664762757295932e-05, "loss": 0.0053, "step": 84020 }, { "epoch": 5.830752547453244e-13, "grad_norm": 0.10759180039167404, "learning_rate": 2.3654159468917263e-05, "loss": 0.0054, "step": 84030 }, { "epoch": 5.831446436843635e-13, "grad_norm": 0.10121466964483261, "learning_rate": 2.3643562227864278e-05, "loss": 0.0053, "step": 84040 }, { "epoch": 5.832140326234025e-13, "grad_norm": 0.1399226039648056, "learning_rate": 2.363297103522607e-05, "loss": 0.0054, "step": 84050 }, { "epoch": 5.832834215624416e-13, "grad_norm": 0.1533665508031845, "learning_rate": 2.3622385892091005e-05, "loss": 0.0058, "step": 84060 }, { "epoch": 5.833528105014807e-13, "grad_norm": 0.12543021142482758, "learning_rate": 2.3611806799546894e-05, "loss": 0.0056, "step": 84070 }, { "epoch": 5.834221994405198e-13, "grad_norm": 0.12812037765979767, "learning_rate": 2.3601233758680882e-05, "loss": 0.0057, "step": 84080 }, { "epoch": 5.834915883795588e-13, "grad_norm": 0.12656022608280182, "learning_rate": 2.3590666770579528e-05, "loss": 0.0054, "step": 84090 }, { "epoch": 5.835609773185979e-13, "grad_norm": 0.1067778691649437, "learning_rate": 2.3580105836328753e-05, "loss": 0.0055, "step": 84100 }, { "epoch": 5.83630366257637e-13, "grad_norm": 0.14991208910942078, "learning_rate": 2.3569550957013855e-05, "loss": 0.0055, "step": 84110 }, { "epoch": 5.836997551966761e-13, "grad_norm": 0.1056179478764534, "learning_rate": 2.355900213371952e-05, "loss": 0.0055, "step": 84120 }, { "epoch": 5.837691441357151e-13, "grad_norm": 0.11839089542627335, "learning_rate": 2.3548459367529792e-05, "loss": 0.0055, "step": 84130 }, { "epoch": 5.838385330747542e-13, "grad_norm": 0.10305705666542053, "learning_rate": 2.353792265952812e-05, "loss": 0.0054, "step": 84140 }, { "epoch": 5.839079220137933e-13, "grad_norm": 0.09597080945968628, "learning_rate": 2.3527392010797284e-05, "loss": 0.0055, "step": 84150 }, { "epoch": 5.839773109528323e-13, "grad_norm": 0.12188680469989777, "learning_rate": 2.3516867422419517e-05, "loss": 0.0053, "step": 84160 }, { "epoch": 5.840466998918714e-13, "grad_norm": 0.1386547088623047, "learning_rate": 2.350634889547633e-05, "loss": 0.0052, "step": 84170 }, { "epoch": 5.841160888309105e-13, "grad_norm": 0.11035417765378952, "learning_rate": 2.3495836431048726e-05, "loss": 0.0054, "step": 84180 }, { "epoch": 5.841854777699496e-13, "grad_norm": 0.10806547105312347, "learning_rate": 2.3485330030216978e-05, "loss": 0.0053, "step": 84190 }, { "epoch": 5.842548667089886e-13, "grad_norm": 0.12082256376743317, "learning_rate": 2.3474829694060793e-05, "loss": 0.0053, "step": 84200 }, { "epoch": 5.843242556480277e-13, "grad_norm": 0.10704367607831955, "learning_rate": 2.3464335423659254e-05, "loss": 0.0054, "step": 84210 }, { "epoch": 5.843936445870668e-13, "grad_norm": 0.12266946583986282, "learning_rate": 2.34538472200908e-05, "loss": 0.0057, "step": 84220 }, { "epoch": 5.844630335261058e-13, "grad_norm": 0.12226992845535278, "learning_rate": 2.344336508443326e-05, "loss": 0.0057, "step": 84230 }, { "epoch": 5.845324224651449e-13, "grad_norm": 0.1141209825873375, "learning_rate": 2.3432889017763847e-05, "loss": 0.0055, "step": 84240 }, { "epoch": 5.84601811404184e-13, "grad_norm": 0.1219196543097496, "learning_rate": 2.3422419021159123e-05, "loss": 0.0054, "step": 84250 }, { "epoch": 5.846712003432231e-13, "grad_norm": 0.12255875766277313, "learning_rate": 2.341195509569506e-05, "loss": 0.0055, "step": 84260 }, { "epoch": 5.847405892822621e-13, "grad_norm": 0.11456136405467987, "learning_rate": 2.340149724244697e-05, "loss": 0.0055, "step": 84270 }, { "epoch": 5.848099782213012e-13, "grad_norm": 0.11383797973394394, "learning_rate": 2.3391045462489578e-05, "loss": 0.0055, "step": 84280 }, { "epoch": 5.848793671603403e-13, "grad_norm": 0.1647845357656479, "learning_rate": 2.3380599756896958e-05, "loss": 0.0055, "step": 84290 }, { "epoch": 5.849487560993794e-13, "grad_norm": 0.09959574043750763, "learning_rate": 2.3370160126742586e-05, "loss": 0.0056, "step": 84300 }, { "epoch": 5.850181450384184e-13, "grad_norm": 0.0926700085401535, "learning_rate": 2.3359726573099254e-05, "loss": 0.0054, "step": 84310 }, { "epoch": 5.850875339774575e-13, "grad_norm": 0.12402776628732681, "learning_rate": 2.334929909703922e-05, "loss": 0.0054, "step": 84320 }, { "epoch": 5.851569229164966e-13, "grad_norm": 0.10852045565843582, "learning_rate": 2.3338877699634017e-05, "loss": 0.0053, "step": 84330 }, { "epoch": 5.852263118555356e-13, "grad_norm": 0.10532776266336441, "learning_rate": 2.332846238195468e-05, "loss": 0.0054, "step": 84340 }, { "epoch": 5.852957007945747e-13, "grad_norm": 0.09825335443019867, "learning_rate": 2.3318053145071466e-05, "loss": 0.0053, "step": 84350 }, { "epoch": 5.853650897336138e-13, "grad_norm": 0.09906308352947235, "learning_rate": 2.3307649990054146e-05, "loss": 0.0054, "step": 84360 }, { "epoch": 5.854344786726529e-13, "grad_norm": 0.15808521211147308, "learning_rate": 2.329725291797175e-05, "loss": 0.0054, "step": 84370 }, { "epoch": 5.855038676116919e-13, "grad_norm": 0.13441278040409088, "learning_rate": 2.3286861929892813e-05, "loss": 0.0058, "step": 84380 }, { "epoch": 5.85573256550731e-13, "grad_norm": 0.11796606332063675, "learning_rate": 2.3276477026885107e-05, "loss": 0.0057, "step": 84390 }, { "epoch": 5.856426454897701e-13, "grad_norm": 0.12644651532173157, "learning_rate": 2.326609821001586e-05, "loss": 0.0055, "step": 84400 }, { "epoch": 5.857120344288091e-13, "grad_norm": 0.12239290773868561, "learning_rate": 2.3255725480351666e-05, "loss": 0.0054, "step": 84410 }, { "epoch": 5.857814233678482e-13, "grad_norm": 0.08824998885393143, "learning_rate": 2.3245358838958478e-05, "loss": 0.0055, "step": 84420 }, { "epoch": 5.858508123068873e-13, "grad_norm": 0.11172791570425034, "learning_rate": 2.323499828690163e-05, "loss": 0.0056, "step": 84430 }, { "epoch": 5.859202012459264e-13, "grad_norm": 0.0776275172829628, "learning_rate": 2.3224643825245832e-05, "loss": 0.0055, "step": 84440 }, { "epoch": 5.859895901849654e-13, "grad_norm": 0.10552860796451569, "learning_rate": 2.3214295455055168e-05, "loss": 0.0056, "step": 84450 }, { "epoch": 5.860589791240045e-13, "grad_norm": 0.08615682274103165, "learning_rate": 2.3203953177393085e-05, "loss": 0.0055, "step": 84460 }, { "epoch": 5.861283680630436e-13, "grad_norm": 0.11868342757225037, "learning_rate": 2.3193616993322423e-05, "loss": 0.0054, "step": 84470 }, { "epoch": 5.861977570020827e-13, "grad_norm": 0.1163485199213028, "learning_rate": 2.3183286903905376e-05, "loss": 0.0055, "step": 84480 }, { "epoch": 5.862671459411217e-13, "grad_norm": 0.10413923859596252, "learning_rate": 2.317296291020353e-05, "loss": 0.0053, "step": 84490 }, { "epoch": 5.863365348801608e-13, "grad_norm": 0.11341197043657303, "learning_rate": 2.316264501327784e-05, "loss": 0.0054, "step": 84500 }, { "epoch": 5.864059238191999e-13, "grad_norm": 0.10950444638729095, "learning_rate": 2.315233321418861e-05, "loss": 0.0053, "step": 84510 }, { "epoch": 5.864753127582389e-13, "grad_norm": 0.11111404001712799, "learning_rate": 2.3142027513995574e-05, "loss": 0.0053, "step": 84520 }, { "epoch": 5.86544701697278e-13, "grad_norm": 0.12889692187309265, "learning_rate": 2.3131727913757756e-05, "loss": 0.0054, "step": 84530 }, { "epoch": 5.866140906363171e-13, "grad_norm": 0.12518489360809326, "learning_rate": 2.3121434414533643e-05, "loss": 0.0057, "step": 84540 }, { "epoch": 5.866834795753562e-13, "grad_norm": 0.14742080867290497, "learning_rate": 2.3111147017381013e-05, "loss": 0.0057, "step": 84550 }, { "epoch": 5.867528685143952e-13, "grad_norm": 0.10426969826221466, "learning_rate": 2.31008657233571e-05, "loss": 0.0054, "step": 84560 }, { "epoch": 5.868222574534343e-13, "grad_norm": 0.10681517422199249, "learning_rate": 2.309059053351842e-05, "loss": 0.0058, "step": 84570 }, { "epoch": 5.868916463924734e-13, "grad_norm": 0.13114967942237854, "learning_rate": 2.3080321448920958e-05, "loss": 0.0055, "step": 84580 }, { "epoch": 5.869610353315124e-13, "grad_norm": 0.12079273909330368, "learning_rate": 2.307005847061998e-05, "loss": 0.0056, "step": 84590 }, { "epoch": 5.870304242705515e-13, "grad_norm": 0.11737986654043198, "learning_rate": 2.3059801599670184e-05, "loss": 0.0055, "step": 84600 }, { "epoch": 5.870998132095906e-13, "grad_norm": 0.10394811630249023, "learning_rate": 2.304955083712563e-05, "loss": 0.0057, "step": 84610 }, { "epoch": 5.871692021486297e-13, "grad_norm": 0.13103890419006348, "learning_rate": 2.3039306184039725e-05, "loss": 0.0055, "step": 84620 }, { "epoch": 5.872385910876687e-13, "grad_norm": 0.09964247047901154, "learning_rate": 2.3029067641465297e-05, "loss": 0.0054, "step": 84630 }, { "epoch": 5.873079800267078e-13, "grad_norm": 0.10111097991466522, "learning_rate": 2.301883521045447e-05, "loss": 0.0054, "step": 84640 }, { "epoch": 5.873773689657469e-13, "grad_norm": 0.13427965342998505, "learning_rate": 2.3008608892058837e-05, "loss": 0.0053, "step": 84650 }, { "epoch": 5.87446757904786e-13, "grad_norm": 0.11931797862052917, "learning_rate": 2.2998388687329256e-05, "loss": 0.0052, "step": 84660 }, { "epoch": 5.87516146843825e-13, "grad_norm": 0.10249102860689163, "learning_rate": 2.298817459731607e-05, "loss": 0.0053, "step": 84670 }, { "epoch": 5.875855357828641e-13, "grad_norm": 0.10728256404399872, "learning_rate": 2.2977966623068884e-05, "loss": 0.0053, "step": 84680 }, { "epoch": 5.876549247219032e-13, "grad_norm": 0.1238049641251564, "learning_rate": 2.296776476563677e-05, "loss": 0.0054, "step": 84690 }, { "epoch": 5.877243136609422e-13, "grad_norm": 0.14959760010242462, "learning_rate": 2.2957569026068087e-05, "loss": 0.0057, "step": 84700 }, { "epoch": 5.877937025999813e-13, "grad_norm": 0.09552153944969177, "learning_rate": 2.2947379405410642e-05, "loss": 0.0058, "step": 84710 }, { "epoch": 5.878630915390204e-13, "grad_norm": 0.13100609183311462, "learning_rate": 2.293719590471156e-05, "loss": 0.0054, "step": 84720 }, { "epoch": 5.879324804780595e-13, "grad_norm": 0.11958502978086472, "learning_rate": 2.292701852501735e-05, "loss": 0.0058, "step": 84730 }, { "epoch": 5.880018694170985e-13, "grad_norm": 0.1269761025905609, "learning_rate": 2.2916847267373897e-05, "loss": 0.0056, "step": 84740 }, { "epoch": 5.880712583561376e-13, "grad_norm": 0.12928810715675354, "learning_rate": 2.2906682132826465e-05, "loss": 0.0056, "step": 84750 }, { "epoch": 5.881406472951767e-13, "grad_norm": 0.10551415383815765, "learning_rate": 2.2896523122419673e-05, "loss": 0.0055, "step": 84760 }, { "epoch": 5.882100362342157e-13, "grad_norm": 0.11920235306024551, "learning_rate": 2.2886370237197522e-05, "loss": 0.0058, "step": 84770 }, { "epoch": 5.882794251732548e-13, "grad_norm": 0.13916628062725067, "learning_rate": 2.2876223478203376e-05, "loss": 0.0054, "step": 84780 }, { "epoch": 5.883488141122939e-13, "grad_norm": 0.12254754453897476, "learning_rate": 2.2866082846479976e-05, "loss": 0.0053, "step": 84790 }, { "epoch": 5.88418203051333e-13, "grad_norm": 0.09169048070907593, "learning_rate": 2.2855948343069418e-05, "loss": 0.0053, "step": 84800 }, { "epoch": 5.88487591990372e-13, "grad_norm": 0.0933244600892067, "learning_rate": 2.284581996901319e-05, "loss": 0.0053, "step": 84810 }, { "epoch": 5.885569809294111e-13, "grad_norm": 0.11452054232358932, "learning_rate": 2.2835697725352142e-05, "loss": 0.0055, "step": 84820 }, { "epoch": 5.886263698684502e-13, "grad_norm": 0.15123066306114197, "learning_rate": 2.282558161312649e-05, "loss": 0.0054, "step": 84830 }, { "epoch": 5.886957588074893e-13, "grad_norm": 0.1254274696111679, "learning_rate": 2.2815471633375818e-05, "loss": 0.0054, "step": 84840 }, { "epoch": 5.887651477465283e-13, "grad_norm": 0.1355838030576706, "learning_rate": 2.2805367787139102e-05, "loss": 0.0058, "step": 84850 }, { "epoch": 5.888345366855674e-13, "grad_norm": 0.14616753160953522, "learning_rate": 2.279527007545462e-05, "loss": 0.0058, "step": 84860 }, { "epoch": 5.889039256246065e-13, "grad_norm": 0.15050801634788513, "learning_rate": 2.2785178499360132e-05, "loss": 0.0059, "step": 84870 }, { "epoch": 5.889733145636455e-13, "grad_norm": 0.12992289662361145, "learning_rate": 2.2775093059892652e-05, "loss": 0.0057, "step": 84880 }, { "epoch": 5.890427035026846e-13, "grad_norm": 0.12657661736011505, "learning_rate": 2.2765013758088662e-05, "loss": 0.0056, "step": 84890 }, { "epoch": 5.891120924417237e-13, "grad_norm": 0.14053697884082794, "learning_rate": 2.2754940594983912e-05, "loss": 0.0058, "step": 84900 }, { "epoch": 5.891814813807628e-13, "grad_norm": 0.14241623878479004, "learning_rate": 2.2744873571613645e-05, "loss": 0.0055, "step": 84910 }, { "epoch": 5.892508703198018e-13, "grad_norm": 0.12218177318572998, "learning_rate": 2.2734812689012348e-05, "loss": 0.0056, "step": 84920 }, { "epoch": 5.893202592588409e-13, "grad_norm": 0.09607502818107605, "learning_rate": 2.272475794821396e-05, "loss": 0.0054, "step": 84930 }, { "epoch": 5.8938964819788e-13, "grad_norm": 0.0919882133603096, "learning_rate": 2.271470935025174e-05, "loss": 0.0054, "step": 84940 }, { "epoch": 5.894590371369191e-13, "grad_norm": 0.11911485344171524, "learning_rate": 2.270466689615836e-05, "loss": 0.0054, "step": 84950 }, { "epoch": 5.895284260759581e-13, "grad_norm": 0.12917107343673706, "learning_rate": 2.2694630586965833e-05, "loss": 0.0054, "step": 84960 }, { "epoch": 5.895978150149972e-13, "grad_norm": 0.11199160665273666, "learning_rate": 2.268460042370555e-05, "loss": 0.0053, "step": 84970 }, { "epoch": 5.896672039540363e-13, "grad_norm": 0.14924518764019012, "learning_rate": 2.2674576407408247e-05, "loss": 0.0054, "step": 84980 }, { "epoch": 5.897365928930753e-13, "grad_norm": 0.13879795372486115, "learning_rate": 2.2664558539104065e-05, "loss": 0.0054, "step": 84990 }, { "epoch": 5.898059818321144e-13, "grad_norm": 0.15385378897190094, "learning_rate": 2.2654546819822492e-05, "loss": 0.0057, "step": 85000 }, { "epoch": 5.898753707711535e-13, "grad_norm": 0.13285760581493378, "learning_rate": 2.2644541250592395e-05, "loss": 0.0057, "step": 85010 }, { "epoch": 5.899447597101926e-13, "grad_norm": 0.13087540864944458, "learning_rate": 2.2634541832441982e-05, "loss": 0.0058, "step": 85020 }, { "epoch": 5.900141486492316e-13, "grad_norm": 0.15388645231723785, "learning_rate": 2.262454856639887e-05, "loss": 0.0056, "step": 85030 }, { "epoch": 5.900835375882707e-13, "grad_norm": 0.11079951375722885, "learning_rate": 2.2614561453490024e-05, "loss": 0.0056, "step": 85040 }, { "epoch": 5.901529265273098e-13, "grad_norm": 0.12687823176383972, "learning_rate": 2.260458049474175e-05, "loss": 0.0056, "step": 85050 }, { "epoch": 5.902223154663488e-13, "grad_norm": 0.09231813251972198, "learning_rate": 2.2594605691179765e-05, "loss": 0.0056, "step": 85060 }, { "epoch": 5.902917044053879e-13, "grad_norm": 0.14201785624027252, "learning_rate": 2.2584637043829132e-05, "loss": 0.0056, "step": 85070 }, { "epoch": 5.90361093344427e-13, "grad_norm": 0.134184792637825, "learning_rate": 2.2574674553714282e-05, "loss": 0.0057, "step": 85080 }, { "epoch": 5.904304822834661e-13, "grad_norm": 0.10174340754747391, "learning_rate": 2.2564718221859017e-05, "loss": 0.0055, "step": 85090 }, { "epoch": 5.904998712225051e-13, "grad_norm": 0.10431765019893646, "learning_rate": 2.2554768049286515e-05, "loss": 0.0054, "step": 85100 }, { "epoch": 5.905692601615442e-13, "grad_norm": 0.10839825123548508, "learning_rate": 2.2544824037019297e-05, "loss": 0.0053, "step": 85110 }, { "epoch": 5.906386491005833e-13, "grad_norm": 0.10737103968858719, "learning_rate": 2.253488618607927e-05, "loss": 0.0054, "step": 85120 }, { "epoch": 5.907080380396224e-13, "grad_norm": 0.10442556440830231, "learning_rate": 2.2524954497487698e-05, "loss": 0.0053, "step": 85130 }, { "epoch": 5.907774269786614e-13, "grad_norm": 0.10496652871370316, "learning_rate": 2.2515028972265227e-05, "loss": 0.0054, "step": 85140 }, { "epoch": 5.908468159177005e-13, "grad_norm": 0.11340752989053726, "learning_rate": 2.2505109611431858e-05, "loss": 0.0054, "step": 85150 }, { "epoch": 5.909162048567396e-13, "grad_norm": 0.14486335217952728, "learning_rate": 2.249519641600696e-05, "loss": 0.0056, "step": 85160 }, { "epoch": 5.909855937957786e-13, "grad_norm": 0.11836440116167068, "learning_rate": 2.248528938700925e-05, "loss": 0.0057, "step": 85170 }, { "epoch": 5.910549827348177e-13, "grad_norm": 0.1050361692905426, "learning_rate": 2.2475388525456864e-05, "loss": 0.006, "step": 85180 }, { "epoch": 5.911243716738568e-13, "grad_norm": 0.14046865701675415, "learning_rate": 2.246549383236722e-05, "loss": 0.0056, "step": 85190 }, { "epoch": 5.911937606128959e-13, "grad_norm": 0.12040051817893982, "learning_rate": 2.245560530875721e-05, "loss": 0.0056, "step": 85200 }, { "epoch": 5.912631495519349e-13, "grad_norm": 0.10577807575464249, "learning_rate": 2.244572295564298e-05, "loss": 0.0057, "step": 85210 }, { "epoch": 5.91332538490974e-13, "grad_norm": 0.11113426089286804, "learning_rate": 2.2435846774040146e-05, "loss": 0.0055, "step": 85220 }, { "epoch": 5.914019274300131e-13, "grad_norm": 0.10412438213825226, "learning_rate": 2.242597676496358e-05, "loss": 0.0055, "step": 85230 }, { "epoch": 5.914713163690521e-13, "grad_norm": 0.11457469314336777, "learning_rate": 2.2416112929427643e-05, "loss": 0.0057, "step": 85240 }, { "epoch": 5.915407053080912e-13, "grad_norm": 0.11736525595188141, "learning_rate": 2.240625526844595e-05, "loss": 0.0055, "step": 85250 }, { "epoch": 5.916100942471303e-13, "grad_norm": 0.09327264130115509, "learning_rate": 2.2396403783031547e-05, "loss": 0.0052, "step": 85260 }, { "epoch": 5.916794831861694e-13, "grad_norm": 0.08730565756559372, "learning_rate": 2.238655847419683e-05, "loss": 0.0052, "step": 85270 }, { "epoch": 5.917488721252084e-13, "grad_norm": 0.1021362915635109, "learning_rate": 2.2376719342953544e-05, "loss": 0.0053, "step": 85280 }, { "epoch": 5.918182610642475e-13, "grad_norm": 0.10279243439435959, "learning_rate": 2.2366886390312836e-05, "loss": 0.0052, "step": 85290 }, { "epoch": 5.918876500032866e-13, "grad_norm": 0.11460310965776443, "learning_rate": 2.235705961728517e-05, "loss": 0.0054, "step": 85300 }, { "epoch": 5.919570389423257e-13, "grad_norm": 0.08365902304649353, "learning_rate": 2.2347239024880416e-05, "loss": 0.0055, "step": 85310 }, { "epoch": 5.920264278813647e-13, "grad_norm": 0.10920702666044235, "learning_rate": 2.2337424614107787e-05, "loss": 0.0056, "step": 85320 }, { "epoch": 5.920958168204038e-13, "grad_norm": 0.13147196173667908, "learning_rate": 2.232761638597587e-05, "loss": 0.0058, "step": 85330 }, { "epoch": 5.921652057594429e-13, "grad_norm": 0.12655018270015717, "learning_rate": 2.231781434149261e-05, "loss": 0.0059, "step": 85340 }, { "epoch": 5.922345946984819e-13, "grad_norm": 0.11570872366428375, "learning_rate": 2.2308018481665316e-05, "loss": 0.0058, "step": 85350 }, { "epoch": 5.92303983637521e-13, "grad_norm": 0.11095482856035233, "learning_rate": 2.2298228807500672e-05, "loss": 0.0055, "step": 85360 }, { "epoch": 5.923733725765601e-13, "grad_norm": 0.1279756873846054, "learning_rate": 2.2288445320004722e-05, "loss": 0.0056, "step": 85370 }, { "epoch": 5.924427615155992e-13, "grad_norm": 0.09676486253738403, "learning_rate": 2.2278668020182867e-05, "loss": 0.0058, "step": 85380 }, { "epoch": 5.925121504546382e-13, "grad_norm": 0.10271425545215607, "learning_rate": 2.2268896909039857e-05, "loss": 0.0055, "step": 85390 }, { "epoch": 5.925815393936773e-13, "grad_norm": 0.136086568236351, "learning_rate": 2.225913198757986e-05, "loss": 0.0056, "step": 85400 }, { "epoch": 5.926509283327164e-13, "grad_norm": 0.09580027312040329, "learning_rate": 2.224937325680635e-05, "loss": 0.0052, "step": 85410 }, { "epoch": 5.927203172717554e-13, "grad_norm": 0.08710163831710815, "learning_rate": 2.223962071772221e-05, "loss": 0.0053, "step": 85420 }, { "epoch": 5.927897062107945e-13, "grad_norm": 0.08941365033388138, "learning_rate": 2.2229874371329626e-05, "loss": 0.0051, "step": 85430 }, { "epoch": 5.928590951498336e-13, "grad_norm": 0.13311134278774261, "learning_rate": 2.2220134218630243e-05, "loss": 0.0053, "step": 85440 }, { "epoch": 5.929284840888727e-13, "grad_norm": 0.12166989594697952, "learning_rate": 2.221040026062495e-05, "loss": 0.0051, "step": 85450 }, { "epoch": 5.929978730279117e-13, "grad_norm": 0.11069293320178986, "learning_rate": 2.2200672498314127e-05, "loss": 0.0052, "step": 85460 }, { "epoch": 5.930672619669508e-13, "grad_norm": 0.10031212121248245, "learning_rate": 2.2190950932697406e-05, "loss": 0.0055, "step": 85470 }, { "epoch": 5.931366509059899e-13, "grad_norm": 0.12920765578746796, "learning_rate": 2.218123556477384e-05, "loss": 0.0055, "step": 85480 }, { "epoch": 5.93206039845029e-13, "grad_norm": 0.13460838794708252, "learning_rate": 2.2171526395541842e-05, "loss": 0.0057, "step": 85490 }, { "epoch": 5.93275428784068e-13, "grad_norm": 0.10448992997407913, "learning_rate": 2.2161823425999176e-05, "loss": 0.0056, "step": 85500 }, { "epoch": 5.933448177231071e-13, "grad_norm": 0.09818495810031891, "learning_rate": 2.215212665714299e-05, "loss": 0.0055, "step": 85510 }, { "epoch": 5.934142066621462e-13, "grad_norm": 0.08816060423851013, "learning_rate": 2.214243608996973e-05, "loss": 0.0056, "step": 85520 }, { "epoch": 5.934835956011852e-13, "grad_norm": 0.11807670444250107, "learning_rate": 2.213275172547532e-05, "loss": 0.0055, "step": 85530 }, { "epoch": 5.935529845402243e-13, "grad_norm": 0.10955849289894104, "learning_rate": 2.212307356465491e-05, "loss": 0.0055, "step": 85540 }, { "epoch": 5.936223734792634e-13, "grad_norm": 0.10726197808980942, "learning_rate": 2.2113401608503143e-05, "loss": 0.0056, "step": 85550 }, { "epoch": 5.936917624183025e-13, "grad_norm": 0.1306956708431244, "learning_rate": 2.21037358580139e-05, "loss": 0.0053, "step": 85560 }, { "epoch": 5.937611513573415e-13, "grad_norm": 0.09523846954107285, "learning_rate": 2.209407631418056e-05, "loss": 0.0052, "step": 85570 }, { "epoch": 5.938305402963806e-13, "grad_norm": 0.10897060483694077, "learning_rate": 2.2084422977995735e-05, "loss": 0.0052, "step": 85580 }, { "epoch": 5.938999292354197e-13, "grad_norm": 0.09532852470874786, "learning_rate": 2.2074775850451476e-05, "loss": 0.0051, "step": 85590 }, { "epoch": 5.939693181744587e-13, "grad_norm": 0.10859381407499313, "learning_rate": 2.206513493253917e-05, "loss": 0.0052, "step": 85600 }, { "epoch": 5.940387071134978e-13, "grad_norm": 0.10622972249984741, "learning_rate": 2.2055500225249573e-05, "loss": 0.0052, "step": 85610 }, { "epoch": 5.941080960525369e-13, "grad_norm": 0.13008780777454376, "learning_rate": 2.2045871729572807e-05, "loss": 0.0052, "step": 85620 }, { "epoch": 5.94177484991576e-13, "grad_norm": 0.1018339991569519, "learning_rate": 2.203624944649835e-05, "loss": 0.0053, "step": 85630 }, { "epoch": 5.94246873930615e-13, "grad_norm": 0.11274457722902298, "learning_rate": 2.2026633377015026e-05, "loss": 0.0054, "step": 85640 }, { "epoch": 5.943162628696541e-13, "grad_norm": 0.09181591868400574, "learning_rate": 2.2017023522111053e-05, "loss": 0.0056, "step": 85650 }, { "epoch": 5.943856518086932e-13, "grad_norm": 0.12375239282846451, "learning_rate": 2.200741988277399e-05, "loss": 0.0056, "step": 85660 }, { "epoch": 5.944550407477323e-13, "grad_norm": 0.13442936539649963, "learning_rate": 2.1997822459990754e-05, "loss": 0.0055, "step": 85670 }, { "epoch": 5.945244296867713e-13, "grad_norm": 0.11137532442808151, "learning_rate": 2.1988231254747634e-05, "loss": 0.0055, "step": 85680 }, { "epoch": 5.945938186258104e-13, "grad_norm": 0.11477656662464142, "learning_rate": 2.197864626803027e-05, "loss": 0.0055, "step": 85690 }, { "epoch": 5.946632075648495e-13, "grad_norm": 0.11792299151420593, "learning_rate": 2.1969067500823666e-05, "loss": 0.0055, "step": 85700 }, { "epoch": 5.947325965038885e-13, "grad_norm": 0.09788572043180466, "learning_rate": 2.1959494954112226e-05, "loss": 0.0056, "step": 85710 }, { "epoch": 5.948019854429276e-13, "grad_norm": 0.09278085827827454, "learning_rate": 2.1949928628879616e-05, "loss": 0.0053, "step": 85720 }, { "epoch": 5.948713743819667e-13, "grad_norm": 0.13876119256019592, "learning_rate": 2.1940368526108975e-05, "loss": 0.0053, "step": 85730 }, { "epoch": 5.949407633210058e-13, "grad_norm": 0.1363796442747116, "learning_rate": 2.1930814646782712e-05, "loss": 0.0052, "step": 85740 }, { "epoch": 5.950101522600448e-13, "grad_norm": 0.11791849136352539, "learning_rate": 2.1921266991882696e-05, "loss": 0.0052, "step": 85750 }, { "epoch": 5.950795411990839e-13, "grad_norm": 0.09839321672916412, "learning_rate": 2.1911725562390023e-05, "loss": 0.005, "step": 85760 }, { "epoch": 5.95148930138123e-13, "grad_norm": 0.12498551607131958, "learning_rate": 2.190219035928529e-05, "loss": 0.0051, "step": 85770 }, { "epoch": 5.95218319077162e-13, "grad_norm": 0.12315601110458374, "learning_rate": 2.189266138354833e-05, "loss": 0.0052, "step": 85780 }, { "epoch": 5.952877080162011e-13, "grad_norm": 0.1807239055633545, "learning_rate": 2.188313863615845e-05, "loss": 0.0053, "step": 85790 }, { "epoch": 5.953570969552402e-13, "grad_norm": 0.12252122908830643, "learning_rate": 2.1873622118094215e-05, "loss": 0.0054, "step": 85800 }, { "epoch": 5.954264858942793e-13, "grad_norm": 0.12403353303670883, "learning_rate": 2.1864111830333618e-05, "loss": 0.0055, "step": 85810 }, { "epoch": 5.954958748333183e-13, "grad_norm": 0.10676810890436172, "learning_rate": 2.1854607773853978e-05, "loss": 0.0055, "step": 85820 }, { "epoch": 5.955652637723574e-13, "grad_norm": 0.1192975789308548, "learning_rate": 2.1845109949631998e-05, "loss": 0.0054, "step": 85830 }, { "epoch": 5.956346527113965e-13, "grad_norm": 0.1112888753414154, "learning_rate": 2.1835618358643705e-05, "loss": 0.0054, "step": 85840 }, { "epoch": 5.957040416504356e-13, "grad_norm": 0.10089079290628433, "learning_rate": 2.1826133001864538e-05, "loss": 0.0054, "step": 85850 }, { "epoch": 5.957734305894746e-13, "grad_norm": 0.12071019411087036, "learning_rate": 2.1816653880269235e-05, "loss": 0.0054, "step": 85860 }, { "epoch": 5.958428195285137e-13, "grad_norm": 0.09216578304767609, "learning_rate": 2.1807180994831945e-05, "loss": 0.0054, "step": 85870 }, { "epoch": 5.959122084675528e-13, "grad_norm": 0.0870710164308548, "learning_rate": 2.1797714346526144e-05, "loss": 0.0052, "step": 85880 }, { "epoch": 5.959815974065918e-13, "grad_norm": 0.09090206027030945, "learning_rate": 2.1788253936324685e-05, "loss": 0.0052, "step": 85890 }, { "epoch": 5.960509863456309e-13, "grad_norm": 0.13705383241176605, "learning_rate": 2.1778799765199762e-05, "loss": 0.0052, "step": 85900 }, { "epoch": 5.9612037528467e-13, "grad_norm": 0.10655245184898376, "learning_rate": 2.1769351834122953e-05, "loss": 0.0052, "step": 85910 }, { "epoch": 5.961897642237091e-13, "grad_norm": 0.09454690665006638, "learning_rate": 2.1759910144065148e-05, "loss": 0.005, "step": 85920 }, { "epoch": 5.962591531627481e-13, "grad_norm": 0.1343589872121811, "learning_rate": 2.175047469599668e-05, "loss": 0.0051, "step": 85930 }, { "epoch": 5.963285421017872e-13, "grad_norm": 0.1436736136674881, "learning_rate": 2.174104549088714e-05, "loss": 0.0053, "step": 85940 }, { "epoch": 5.963979310408263e-13, "grad_norm": 0.0853751078248024, "learning_rate": 2.173162252970554e-05, "loss": 0.0053, "step": 85950 }, { "epoch": 5.964673199798654e-13, "grad_norm": 0.12385519593954086, "learning_rate": 2.1722205813420244e-05, "loss": 0.0056, "step": 85960 }, { "epoch": 5.965367089189044e-13, "grad_norm": 0.10669916123151779, "learning_rate": 2.171279534299896e-05, "loss": 0.0055, "step": 85970 }, { "epoch": 5.966060978579435e-13, "grad_norm": 0.10062506794929504, "learning_rate": 2.170339111940877e-05, "loss": 0.0056, "step": 85980 }, { "epoch": 5.966754867969826e-13, "grad_norm": 0.0957406535744667, "learning_rate": 2.169399314361609e-05, "loss": 0.0054, "step": 85990 }, { "epoch": 5.967448757360216e-13, "grad_norm": 0.12109778076410294, "learning_rate": 2.1684601416586716e-05, "loss": 0.0053, "step": 86000 }, { "epoch": 5.968142646750607e-13, "grad_norm": 0.12282992899417877, "learning_rate": 2.1675215939285785e-05, "loss": 0.0054, "step": 86010 }, { "epoch": 5.968836536140998e-13, "grad_norm": 0.09007307887077332, "learning_rate": 2.166583671267781e-05, "loss": 0.0054, "step": 86020 }, { "epoch": 5.969530425531389e-13, "grad_norm": 0.12357962876558304, "learning_rate": 2.1656463737726663e-05, "loss": 0.0053, "step": 86030 }, { "epoch": 5.970224314921779e-13, "grad_norm": 0.10580126196146011, "learning_rate": 2.1647097015395545e-05, "loss": 0.0052, "step": 86040 }, { "epoch": 5.97091820431217e-13, "grad_norm": 0.10971516370773315, "learning_rate": 2.1637736546647024e-05, "loss": 0.0053, "step": 86050 }, { "epoch": 5.971612093702561e-13, "grad_norm": 0.09931385517120361, "learning_rate": 2.1628382332443073e-05, "loss": 0.0051, "step": 86060 }, { "epoch": 5.972305983092951e-13, "grad_norm": 0.11652436852455139, "learning_rate": 2.1619034373744933e-05, "loss": 0.0051, "step": 86070 }, { "epoch": 5.972999872483342e-13, "grad_norm": 0.1170896366238594, "learning_rate": 2.1609692671513293e-05, "loss": 0.005, "step": 86080 }, { "epoch": 5.973693761873733e-13, "grad_norm": 0.0922447219491005, "learning_rate": 2.1600357226708128e-05, "loss": 0.0051, "step": 86090 }, { "epoch": 5.974387651264124e-13, "grad_norm": 0.10109364986419678, "learning_rate": 2.1591028040288838e-05, "loss": 0.0054, "step": 86100 }, { "epoch": 5.975081540654514e-13, "grad_norm": 0.11625254899263382, "learning_rate": 2.158170511321409e-05, "loss": 0.0055, "step": 86110 }, { "epoch": 5.975775430044905e-13, "grad_norm": 0.12415158748626709, "learning_rate": 2.1572388446442018e-05, "loss": 0.0056, "step": 86120 }, { "epoch": 5.976469319435296e-13, "grad_norm": 0.12714603543281555, "learning_rate": 2.1563078040930014e-05, "loss": 0.0056, "step": 86130 }, { "epoch": 5.977163208825687e-13, "grad_norm": 0.1206597238779068, "learning_rate": 2.1553773897634868e-05, "loss": 0.0055, "step": 86140 }, { "epoch": 5.977857098216077e-13, "grad_norm": 0.10509136319160461, "learning_rate": 2.154447601751274e-05, "loss": 0.0053, "step": 86150 }, { "epoch": 5.978550987606468e-13, "grad_norm": 0.12005975842475891, "learning_rate": 2.153518440151913e-05, "loss": 0.0052, "step": 86160 }, { "epoch": 5.979244876996859e-13, "grad_norm": 0.08559776842594147, "learning_rate": 2.152589905060889e-05, "loss": 0.0053, "step": 86170 }, { "epoch": 5.979938766387249e-13, "grad_norm": 0.08199596405029297, "learning_rate": 2.151661996573624e-05, "loss": 0.0053, "step": 86180 }, { "epoch": 5.98063265577764e-13, "grad_norm": 0.11194854974746704, "learning_rate": 2.1507347147854753e-05, "loss": 0.0054, "step": 86190 }, { "epoch": 5.981326545168031e-13, "grad_norm": 0.09472525119781494, "learning_rate": 2.1498080597917344e-05, "loss": 0.0051, "step": 86200 }, { "epoch": 5.982020434558422e-13, "grad_norm": 0.09031786024570465, "learning_rate": 2.14888203168763e-05, "loss": 0.0052, "step": 86210 }, { "epoch": 5.982714323948812e-13, "grad_norm": 0.11692198365926743, "learning_rate": 2.1479566305683268e-05, "loss": 0.0052, "step": 86220 }, { "epoch": 5.983408213339203e-13, "grad_norm": 0.0962272584438324, "learning_rate": 2.1470318565289233e-05, "loss": 0.0051, "step": 86230 }, { "epoch": 5.984102102729594e-13, "grad_norm": 0.12132100760936737, "learning_rate": 2.1461077096644552e-05, "loss": 0.0051, "step": 86240 }, { "epoch": 5.984795992119984e-13, "grad_norm": 0.10776499658823013, "learning_rate": 2.1451841900698904e-05, "loss": 0.0053, "step": 86250 }, { "epoch": 5.985489881510375e-13, "grad_norm": 0.10715362429618835, "learning_rate": 2.144261297840139e-05, "loss": 0.0054, "step": 86260 }, { "epoch": 5.986183770900766e-13, "grad_norm": 0.117158442735672, "learning_rate": 2.1433390330700375e-05, "loss": 0.0057, "step": 86270 }, { "epoch": 5.986877660291157e-13, "grad_norm": 0.1368023008108139, "learning_rate": 2.1424173958543693e-05, "loss": 0.0055, "step": 86280 }, { "epoch": 5.987571549681547e-13, "grad_norm": 0.10760488361120224, "learning_rate": 2.1414963862878402e-05, "loss": 0.0056, "step": 86290 }, { "epoch": 5.988265439071938e-13, "grad_norm": 0.1272798329591751, "learning_rate": 2.140576004465104e-05, "loss": 0.0054, "step": 86300 }, { "epoch": 5.988959328462329e-13, "grad_norm": 0.10478188842535019, "learning_rate": 2.1396562504807395e-05, "loss": 0.0052, "step": 86310 }, { "epoch": 5.98965321785272e-13, "grad_norm": 0.12021257728338242, "learning_rate": 2.1387371244292703e-05, "loss": 0.0052, "step": 86320 }, { "epoch": 5.99034710724311e-13, "grad_norm": 0.11931538581848145, "learning_rate": 2.137818626405148e-05, "loss": 0.0054, "step": 86330 }, { "epoch": 5.991040996633501e-13, "grad_norm": 0.09852541983127594, "learning_rate": 2.1369007565027623e-05, "loss": 0.0053, "step": 86340 }, { "epoch": 5.991734886023892e-13, "grad_norm": 0.12126970291137695, "learning_rate": 2.1359835148164403e-05, "loss": 0.0053, "step": 86350 }, { "epoch": 5.992428775414282e-13, "grad_norm": 0.11012348532676697, "learning_rate": 2.135066901440442e-05, "loss": 0.0052, "step": 86360 }, { "epoch": 5.993122664804673e-13, "grad_norm": 0.09973309189081192, "learning_rate": 2.1341509164689633e-05, "loss": 0.0053, "step": 86370 }, { "epoch": 5.993816554195064e-13, "grad_norm": 0.08973921835422516, "learning_rate": 2.133235559996136e-05, "loss": 0.0051, "step": 86380 }, { "epoch": 5.994510443585455e-13, "grad_norm": 0.12104592472314835, "learning_rate": 2.13232083211603e-05, "loss": 0.0051, "step": 86390 }, { "epoch": 5.995204332975845e-13, "grad_norm": 0.12104229629039764, "learning_rate": 2.1314067329226422e-05, "loss": 0.0052, "step": 86400 }, { "epoch": 5.995898222366236e-13, "grad_norm": 0.1334744691848755, "learning_rate": 2.1304932625099166e-05, "loss": 0.0052, "step": 86410 }, { "epoch": 5.996592111756627e-13, "grad_norm": 0.08055876195430756, "learning_rate": 2.1295804209717207e-05, "loss": 0.0054, "step": 86420 }, { "epoch": 5.997286001147017e-13, "grad_norm": 0.12882652878761292, "learning_rate": 2.1286682084018683e-05, "loss": 0.0058, "step": 86430 }, { "epoch": 5.997979890537408e-13, "grad_norm": 0.1157495453953743, "learning_rate": 2.1277566248940996e-05, "loss": 0.0054, "step": 86440 }, { "epoch": 5.998673779927799e-13, "grad_norm": 0.1290120631456375, "learning_rate": 2.1268456705420973e-05, "loss": 0.0055, "step": 86450 }, { "epoch": 5.99936766931819e-13, "grad_norm": 0.11668439209461212, "learning_rate": 2.125935345439473e-05, "loss": 0.0052, "step": 86460 }, { "epoch": 6.00006155870858e-13, "grad_norm": 0.09195995330810547, "learning_rate": 2.1250256496797795e-05, "loss": 0.0052, "step": 86470 }, { "epoch": 6.000755448098971e-13, "grad_norm": 0.10645625740289688, "learning_rate": 2.1241165833564995e-05, "loss": 0.0053, "step": 86480 }, { "epoch": 6.001449337489362e-13, "grad_norm": 0.09778375178575516, "learning_rate": 2.1232081465630553e-05, "loss": 0.0052, "step": 86490 }, { "epoch": 6.002143226879753e-13, "grad_norm": 0.13115501403808594, "learning_rate": 2.122300339392804e-05, "loss": 0.0054, "step": 86500 }, { "epoch": 6.002837116270143e-13, "grad_norm": 0.10175903141498566, "learning_rate": 2.121393161939034e-05, "loss": 0.0052, "step": 86510 }, { "epoch": 6.003531005660534e-13, "grad_norm": 0.1009463518857956, "learning_rate": 2.120486614294975e-05, "loss": 0.0052, "step": 86520 }, { "epoch": 6.004224895050925e-13, "grad_norm": 0.08577249199151993, "learning_rate": 2.1195806965537862e-05, "loss": 0.0052, "step": 86530 }, { "epoch": 6.004918784441315e-13, "grad_norm": 0.08277016133069992, "learning_rate": 2.118675408808567e-05, "loss": 0.0049, "step": 86540 }, { "epoch": 6.005612673831706e-13, "grad_norm": 0.1350523978471756, "learning_rate": 2.1177707511523473e-05, "loss": 0.0051, "step": 86550 }, { "epoch": 6.006306563222097e-13, "grad_norm": 0.1264684945344925, "learning_rate": 2.116866723678097e-05, "loss": 0.0052, "step": 86560 }, { "epoch": 6.007000452612488e-13, "grad_norm": 0.12979833781719208, "learning_rate": 2.1159633264787185e-05, "loss": 0.0054, "step": 86570 }, { "epoch": 6.007694342002878e-13, "grad_norm": 0.10561966896057129, "learning_rate": 2.11506055964705e-05, "loss": 0.0054, "step": 86580 }, { "epoch": 6.008388231393269e-13, "grad_norm": 0.09782511740922928, "learning_rate": 2.1141584232758647e-05, "loss": 0.0056, "step": 86590 }, { "epoch": 6.00908212078366e-13, "grad_norm": 0.119982048869133, "learning_rate": 2.113256917457869e-05, "loss": 0.0055, "step": 86600 }, { "epoch": 6.00977601017405e-13, "grad_norm": 0.15058687329292297, "learning_rate": 2.1123560422857105e-05, "loss": 0.0054, "step": 86610 }, { "epoch": 6.010469899564441e-13, "grad_norm": 0.11914623528718948, "learning_rate": 2.1114557978519645e-05, "loss": 0.0053, "step": 86620 }, { "epoch": 6.011163788954832e-13, "grad_norm": 0.12453510612249374, "learning_rate": 2.1105561842491485e-05, "loss": 0.0053, "step": 86630 }, { "epoch": 6.011857678345223e-13, "grad_norm": 0.1589144915342331, "learning_rate": 2.109657201569708e-05, "loss": 0.0054, "step": 86640 }, { "epoch": 6.012551567735613e-13, "grad_norm": 0.12474829703569412, "learning_rate": 2.1087588499060315e-05, "loss": 0.0053, "step": 86650 }, { "epoch": 6.013245457126004e-13, "grad_norm": 0.12231559306383133, "learning_rate": 2.1078611293504365e-05, "loss": 0.0054, "step": 86660 }, { "epoch": 6.013939346516395e-13, "grad_norm": 0.09905297309160233, "learning_rate": 2.106964039995177e-05, "loss": 0.0053, "step": 86670 }, { "epoch": 6.014633235906786e-13, "grad_norm": 0.10768955200910568, "learning_rate": 2.1060675819324448e-05, "loss": 0.0052, "step": 86680 }, { "epoch": 6.015327125297176e-13, "grad_norm": 0.08949847519397736, "learning_rate": 2.105171755254363e-05, "loss": 0.0052, "step": 86690 }, { "epoch": 6.016021014687567e-13, "grad_norm": 0.09011844545602798, "learning_rate": 2.1042765600529926e-05, "loss": 0.005, "step": 86700 }, { "epoch": 6.016714904077958e-13, "grad_norm": 0.09785208851099014, "learning_rate": 2.1033819964203296e-05, "loss": 0.0051, "step": 86710 }, { "epoch": 6.017408793468348e-13, "grad_norm": 0.12090494483709335, "learning_rate": 2.1024880644483037e-05, "loss": 0.0052, "step": 86720 }, { "epoch": 6.018102682858739e-13, "grad_norm": 0.20521001517772675, "learning_rate": 2.1015947642287793e-05, "loss": 0.0054, "step": 86730 }, { "epoch": 6.01879657224913e-13, "grad_norm": 0.14138972759246826, "learning_rate": 2.1007020958535587e-05, "loss": 0.0054, "step": 86740 }, { "epoch": 6.019490461639521e-13, "grad_norm": 0.12109556049108505, "learning_rate": 2.0998100594143758e-05, "loss": 0.0057, "step": 86750 }, { "epoch": 6.020184351029911e-13, "grad_norm": 0.12056434899568558, "learning_rate": 2.0989186550029027e-05, "loss": 0.0057, "step": 86760 }, { "epoch": 6.020878240420302e-13, "grad_norm": 0.15416301786899567, "learning_rate": 2.0980278827107442e-05, "loss": 0.0053, "step": 86770 }, { "epoch": 6.021572129810693e-13, "grad_norm": 0.13469110429286957, "learning_rate": 2.0971377426294414e-05, "loss": 0.0053, "step": 86780 }, { "epoch": 6.022266019201084e-13, "grad_norm": 0.09112714231014252, "learning_rate": 2.096248234850471e-05, "loss": 0.0052, "step": 86790 }, { "epoch": 6.022959908591474e-13, "grad_norm": 0.12347374856472015, "learning_rate": 2.0953593594652402e-05, "loss": 0.0054, "step": 86800 }, { "epoch": 6.023653797981865e-13, "grad_norm": 0.14897747337818146, "learning_rate": 2.0944711165650997e-05, "loss": 0.0052, "step": 86810 }, { "epoch": 6.024347687372256e-13, "grad_norm": 0.10399709641933441, "learning_rate": 2.093583506241325e-05, "loss": 0.0052, "step": 86820 }, { "epoch": 6.025041576762646e-13, "grad_norm": 0.10290136188268661, "learning_rate": 2.092696528585137e-05, "loss": 0.0052, "step": 86830 }, { "epoch": 6.025735466153037e-13, "grad_norm": 0.09756439924240112, "learning_rate": 2.091810183687682e-05, "loss": 0.0053, "step": 86840 }, { "epoch": 6.026429355543428e-13, "grad_norm": 0.13992857933044434, "learning_rate": 2.0909244716400504e-05, "loss": 0.005, "step": 86850 }, { "epoch": 6.027123244933819e-13, "grad_norm": 0.12048851698637009, "learning_rate": 2.090039392533259e-05, "loss": 0.005, "step": 86860 }, { "epoch": 6.027817134324209e-13, "grad_norm": 0.17806527018547058, "learning_rate": 2.089154946458265e-05, "loss": 0.0052, "step": 86870 }, { "epoch": 6.0285110237146e-13, "grad_norm": 0.09631280601024628, "learning_rate": 2.0882711335059582e-05, "loss": 0.0051, "step": 86880 }, { "epoch": 6.029204913104991e-13, "grad_norm": 0.09053270518779755, "learning_rate": 2.087387953767165e-05, "loss": 0.0055, "step": 86890 }, { "epoch": 6.029898802495381e-13, "grad_norm": 0.1180492490530014, "learning_rate": 2.0865054073326462e-05, "loss": 0.0056, "step": 86900 }, { "epoch": 6.030592691885772e-13, "grad_norm": 0.11188258975744247, "learning_rate": 2.0856234942930956e-05, "loss": 0.0057, "step": 86910 }, { "epoch": 6.031286581276163e-13, "grad_norm": 0.11733626574277878, "learning_rate": 2.0847422147391458e-05, "loss": 0.0056, "step": 86920 }, { "epoch": 6.031980470666554e-13, "grad_norm": 0.12266135960817337, "learning_rate": 2.083861568761359e-05, "loss": 0.0054, "step": 86930 }, { "epoch": 6.032674360056944e-13, "grad_norm": 0.10007048398256302, "learning_rate": 2.0829815564502386e-05, "loss": 0.0055, "step": 86940 }, { "epoch": 6.033368249447335e-13, "grad_norm": 0.1478486806154251, "learning_rate": 2.0821021778962158e-05, "loss": 0.0054, "step": 86950 }, { "epoch": 6.034062138837726e-13, "grad_norm": 0.11818113178014755, "learning_rate": 2.0812234331896643e-05, "loss": 0.0054, "step": 86960 }, { "epoch": 6.034756028228117e-13, "grad_norm": 0.10387904942035675, "learning_rate": 2.0803453224208852e-05, "loss": 0.0055, "step": 86970 }, { "epoch": 6.035449917618507e-13, "grad_norm": 0.13429786264896393, "learning_rate": 2.079467845680122e-05, "loss": 0.0052, "step": 86980 }, { "epoch": 6.036143807008898e-13, "grad_norm": 0.15900279581546783, "learning_rate": 2.0785910030575454e-05, "loss": 0.0053, "step": 86990 }, { "epoch": 6.036837696399289e-13, "grad_norm": 0.09825659543275833, "learning_rate": 2.0777147946432663e-05, "loss": 0.0053, "step": 87000 }, { "epoch": 6.037531585789679e-13, "grad_norm": 0.12425234168767929, "learning_rate": 2.076839220527329e-05, "loss": 0.0049, "step": 87010 }, { "epoch": 6.03822547518007e-13, "grad_norm": 0.1478385329246521, "learning_rate": 2.075964280799711e-05, "loss": 0.0052, "step": 87020 }, { "epoch": 6.038919364570461e-13, "grad_norm": 0.09671075642108917, "learning_rate": 2.075089975550328e-05, "loss": 0.0052, "step": 87030 }, { "epoch": 6.039613253960852e-13, "grad_norm": 0.0927228257060051, "learning_rate": 2.074216304869027e-05, "loss": 0.0053, "step": 87040 }, { "epoch": 6.040307143351242e-13, "grad_norm": 0.1120164543390274, "learning_rate": 2.0733432688455916e-05, "loss": 0.0055, "step": 87050 }, { "epoch": 6.041001032741633e-13, "grad_norm": 0.09663476049900055, "learning_rate": 2.0724708675697396e-05, "loss": 0.0054, "step": 87060 }, { "epoch": 6.041694922132024e-13, "grad_norm": 0.0835016593337059, "learning_rate": 2.0715991011311236e-05, "loss": 0.0056, "step": 87070 }, { "epoch": 6.042388811522414e-13, "grad_norm": 0.1263507753610611, "learning_rate": 2.070727969619333e-05, "loss": 0.0055, "step": 87080 }, { "epoch": 6.043082700912805e-13, "grad_norm": 0.12996871769428253, "learning_rate": 2.0698574731238875e-05, "loss": 0.0055, "step": 87090 }, { "epoch": 6.043776590303196e-13, "grad_norm": 0.1457197219133377, "learning_rate": 2.0689876117342463e-05, "loss": 0.0054, "step": 87100 }, { "epoch": 6.044470479693587e-13, "grad_norm": 0.11626511812210083, "learning_rate": 2.0681183855398e-05, "loss": 0.0054, "step": 87110 }, { "epoch": 6.045164369083977e-13, "grad_norm": 0.10015798360109329, "learning_rate": 2.0672497946298765e-05, "loss": 0.0053, "step": 87120 }, { "epoch": 6.045858258474368e-13, "grad_norm": 0.12378542125225067, "learning_rate": 2.066381839093734e-05, "loss": 0.0052, "step": 87130 }, { "epoch": 6.046552147864759e-13, "grad_norm": 0.07353420555591583, "learning_rate": 2.0655145190205722e-05, "loss": 0.0051, "step": 87140 }, { "epoch": 6.04724603725515e-13, "grad_norm": 0.11829481273889542, "learning_rate": 2.064647834499518e-05, "loss": 0.0053, "step": 87150 }, { "epoch": 6.04793992664554e-13, "grad_norm": 0.11708678305149078, "learning_rate": 2.0637817856196408e-05, "loss": 0.005, "step": 87160 }, { "epoch": 6.048633816035931e-13, "grad_norm": 0.08916398882865906, "learning_rate": 2.0629163724699365e-05, "loss": 0.0053, "step": 87170 }, { "epoch": 6.049327705426322e-13, "grad_norm": 0.15853922069072723, "learning_rate": 2.062051595139343e-05, "loss": 0.0053, "step": 87180 }, { "epoch": 6.050021594816712e-13, "grad_norm": 0.12489848583936691, "learning_rate": 2.0611874537167272e-05, "loss": 0.0053, "step": 87190 }, { "epoch": 6.050715484207103e-13, "grad_norm": 0.10968372225761414, "learning_rate": 2.0603239482908956e-05, "loss": 0.0053, "step": 87200 }, { "epoch": 6.051409373597494e-13, "grad_norm": 0.08750620484352112, "learning_rate": 2.0594610789505853e-05, "loss": 0.0055, "step": 87210 }, { "epoch": 6.052103262987885e-13, "grad_norm": 0.11254166811704636, "learning_rate": 2.0585988457844682e-05, "loss": 0.0055, "step": 87220 }, { "epoch": 6.052797152378275e-13, "grad_norm": 0.10077877342700958, "learning_rate": 2.0577372488811547e-05, "loss": 0.0054, "step": 87230 }, { "epoch": 6.053491041768666e-13, "grad_norm": 0.13364963233470917, "learning_rate": 2.0568762883291864e-05, "loss": 0.0055, "step": 87240 }, { "epoch": 6.054184931159057e-13, "grad_norm": 0.13528217375278473, "learning_rate": 2.0560159642170397e-05, "loss": 0.0055, "step": 87250 }, { "epoch": 6.054878820549447e-13, "grad_norm": 0.12181858718395233, "learning_rate": 2.055156276633127e-05, "loss": 0.0054, "step": 87260 }, { "epoch": 6.055572709939838e-13, "grad_norm": 0.07936770468950272, "learning_rate": 2.0542972256657945e-05, "loss": 0.0055, "step": 87270 }, { "epoch": 6.056266599330229e-13, "grad_norm": 0.09134580194950104, "learning_rate": 2.053438811403323e-05, "loss": 0.0054, "step": 87280 }, { "epoch": 6.05696048872062e-13, "grad_norm": 0.09488523006439209, "learning_rate": 2.052581033933928e-05, "loss": 0.0052, "step": 87290 }, { "epoch": 6.05765437811101e-13, "grad_norm": 0.11199706792831421, "learning_rate": 2.051723893345759e-05, "loss": 0.0051, "step": 87300 }, { "epoch": 6.058348267501401e-13, "grad_norm": 0.09166327118873596, "learning_rate": 2.050867389726902e-05, "loss": 0.0051, "step": 87310 }, { "epoch": 6.059042156891792e-13, "grad_norm": 0.13738422095775604, "learning_rate": 2.050011523165374e-05, "loss": 0.0049, "step": 87320 }, { "epoch": 6.059736046282183e-13, "grad_norm": 0.13132551312446594, "learning_rate": 2.0491562937491294e-05, "loss": 0.0055, "step": 87330 }, { "epoch": 6.060429935672573e-13, "grad_norm": 0.10519963502883911, "learning_rate": 2.0483017015660565e-05, "loss": 0.0053, "step": 87340 }, { "epoch": 6.061123825062964e-13, "grad_norm": 0.11397509276866913, "learning_rate": 2.0474477467039787e-05, "loss": 0.0052, "step": 87350 }, { "epoch": 6.061817714453355e-13, "grad_norm": 0.1169770285487175, "learning_rate": 2.0465944292506515e-05, "loss": 0.0055, "step": 87360 }, { "epoch": 6.062511603843745e-13, "grad_norm": 0.11256023496389389, "learning_rate": 2.045741749293767e-05, "loss": 0.0053, "step": 87370 }, { "epoch": 6.063205493234136e-13, "grad_norm": 0.12508714199066162, "learning_rate": 2.044889706920953e-05, "loss": 0.0055, "step": 87380 }, { "epoch": 6.063899382624527e-13, "grad_norm": 0.12106213718652725, "learning_rate": 2.044038302219768e-05, "loss": 0.0053, "step": 87390 }, { "epoch": 6.064593272014918e-13, "grad_norm": 0.12587840855121613, "learning_rate": 2.0431875352777088e-05, "loss": 0.0055, "step": 87400 }, { "epoch": 6.065287161405308e-13, "grad_norm": 0.13960041105747223, "learning_rate": 2.0423374061822032e-05, "loss": 0.0055, "step": 87410 }, { "epoch": 6.065981050795699e-13, "grad_norm": 0.10673782974481583, "learning_rate": 2.0414879150206164e-05, "loss": 0.0054, "step": 87420 }, { "epoch": 6.06667494018609e-13, "grad_norm": 0.10733841359615326, "learning_rate": 2.0406390618802465e-05, "loss": 0.0054, "step": 87430 }, { "epoch": 6.06736882957648e-13, "grad_norm": 0.09933536499738693, "learning_rate": 2.0397908468483266e-05, "loss": 0.0053, "step": 87440 }, { "epoch": 6.068062718966871e-13, "grad_norm": 0.10606475174427032, "learning_rate": 2.0389432700120252e-05, "loss": 0.005, "step": 87450 }, { "epoch": 6.068756608357262e-13, "grad_norm": 0.10161250084638596, "learning_rate": 2.0380963314584393e-05, "loss": 0.0051, "step": 87460 }, { "epoch": 6.069450497747653e-13, "grad_norm": 0.12783488631248474, "learning_rate": 2.037250031274611e-05, "loss": 0.0052, "step": 87470 }, { "epoch": 6.070144387138043e-13, "grad_norm": 0.1415909230709076, "learning_rate": 2.0364043695475064e-05, "loss": 0.005, "step": 87480 }, { "epoch": 6.070838276528434e-13, "grad_norm": 0.11075910180807114, "learning_rate": 2.0355593463640336e-05, "loss": 0.0052, "step": 87490 }, { "epoch": 6.071532165918825e-13, "grad_norm": 0.16521279513835907, "learning_rate": 2.034714961811029e-05, "loss": 0.0052, "step": 87500 }, { "epoch": 6.072226055309216e-13, "grad_norm": 0.12706829607486725, "learning_rate": 2.0338712159752692e-05, "loss": 0.0053, "step": 87510 }, { "epoch": 6.072919944699606e-13, "grad_norm": 0.13036367297172546, "learning_rate": 2.033028108943459e-05, "loss": 0.0054, "step": 87520 }, { "epoch": 6.073613834089997e-13, "grad_norm": 0.08626888692378998, "learning_rate": 2.0321856408022448e-05, "loss": 0.0053, "step": 87530 }, { "epoch": 6.074307723480388e-13, "grad_norm": 0.10886262357234955, "learning_rate": 2.031343811638199e-05, "loss": 0.0054, "step": 87540 }, { "epoch": 6.075001612870778e-13, "grad_norm": 0.0967084988951683, "learning_rate": 2.0305026215378348e-05, "loss": 0.0052, "step": 87550 }, { "epoch": 6.075695502261169e-13, "grad_norm": 0.12164229154586792, "learning_rate": 2.0296620705875973e-05, "loss": 0.0054, "step": 87560 }, { "epoch": 6.07638939165156e-13, "grad_norm": 0.1499662697315216, "learning_rate": 2.0288221588738664e-05, "loss": 0.0056, "step": 87570 }, { "epoch": 6.077083281041951e-13, "grad_norm": 0.10140480846166611, "learning_rate": 2.027982886482956e-05, "loss": 0.0052, "step": 87580 }, { "epoch": 6.077777170432341e-13, "grad_norm": 0.11239281296730042, "learning_rate": 2.027144253501114e-05, "loss": 0.0053, "step": 87590 }, { "epoch": 6.078471059822732e-13, "grad_norm": 0.09510983526706696, "learning_rate": 2.026306260014524e-05, "loss": 0.0052, "step": 87600 }, { "epoch": 6.079164949213123e-13, "grad_norm": 0.10430819541215897, "learning_rate": 2.025468906109302e-05, "loss": 0.0049, "step": 87610 }, { "epoch": 6.079858838603514e-13, "grad_norm": 0.10147561132907867, "learning_rate": 2.024632191871499e-05, "loss": 0.0053, "step": 87620 }, { "epoch": 6.080552727993904e-13, "grad_norm": 0.1284509301185608, "learning_rate": 2.023796117387101e-05, "loss": 0.0051, "step": 87630 }, { "epoch": 6.081246617384295e-13, "grad_norm": 0.10975814610719681, "learning_rate": 2.0229606827420272e-05, "loss": 0.005, "step": 87640 }, { "epoch": 6.081940506774686e-13, "grad_norm": 0.17146651446819305, "learning_rate": 2.0221258880221332e-05, "loss": 0.0052, "step": 87650 }, { "epoch": 6.082634396165076e-13, "grad_norm": 0.1549869030714035, "learning_rate": 2.021291733313202e-05, "loss": 0.0054, "step": 87660 }, { "epoch": 6.083328285555467e-13, "grad_norm": 0.11137007176876068, "learning_rate": 2.0204582187009635e-05, "loss": 0.0055, "step": 87670 }, { "epoch": 6.084022174945858e-13, "grad_norm": 0.10218925774097443, "learning_rate": 2.019625344271068e-05, "loss": 0.0053, "step": 87680 }, { "epoch": 6.084716064336249e-13, "grad_norm": 0.11483129113912582, "learning_rate": 2.0187931101091107e-05, "loss": 0.0053, "step": 87690 }, { "epoch": 6.085409953726639e-13, "grad_norm": 0.09832162410020828, "learning_rate": 2.0179615163006127e-05, "loss": 0.0054, "step": 87700 }, { "epoch": 6.08610384311703e-13, "grad_norm": 0.1392710655927658, "learning_rate": 2.017130562931037e-05, "loss": 0.0053, "step": 87710 }, { "epoch": 6.086797732507421e-13, "grad_norm": 0.1364269107580185, "learning_rate": 2.0163002500857726e-05, "loss": 0.0056, "step": 87720 }, { "epoch": 6.087491621897811e-13, "grad_norm": 0.10168374329805374, "learning_rate": 2.015470577850152e-05, "loss": 0.0054, "step": 87730 }, { "epoch": 6.088185511288202e-13, "grad_norm": 0.1153247281908989, "learning_rate": 2.0146415463094337e-05, "loss": 0.0054, "step": 87740 }, { "epoch": 6.088879400678593e-13, "grad_norm": 0.07884837687015533, "learning_rate": 2.0138131555488145e-05, "loss": 0.0052, "step": 87750 }, { "epoch": 6.089573290068984e-13, "grad_norm": 0.13628840446472168, "learning_rate": 2.012985405653424e-05, "loss": 0.005, "step": 87760 }, { "epoch": 6.090267179459374e-13, "grad_norm": 0.08632305264472961, "learning_rate": 2.0121582967083274e-05, "loss": 0.0051, "step": 87770 }, { "epoch": 6.090961068849765e-13, "grad_norm": 0.10152330249547958, "learning_rate": 2.0113318287985227e-05, "loss": 0.0053, "step": 87780 }, { "epoch": 6.091654958240156e-13, "grad_norm": 0.08544814586639404, "learning_rate": 2.0105060020089398e-05, "loss": 0.0049, "step": 87790 }, { "epoch": 6.092348847630547e-13, "grad_norm": 0.08736035227775574, "learning_rate": 2.0096808164244502e-05, "loss": 0.0052, "step": 87800 }, { "epoch": 6.093042737020937e-13, "grad_norm": 0.12770304083824158, "learning_rate": 2.008856272129849e-05, "loss": 0.0053, "step": 87810 }, { "epoch": 6.093736626411328e-13, "grad_norm": 0.14974220097064972, "learning_rate": 2.0080323692098767e-05, "loss": 0.0053, "step": 87820 }, { "epoch": 6.094430515801719e-13, "grad_norm": 0.13756000995635986, "learning_rate": 2.0072091077491967e-05, "loss": 0.0054, "step": 87830 }, { "epoch": 6.095124405192109e-13, "grad_norm": 0.08665229380130768, "learning_rate": 2.0063864878324168e-05, "loss": 0.0052, "step": 87840 }, { "epoch": 6.0958182945825e-13, "grad_norm": 0.11235019564628601, "learning_rate": 2.0055645095440686e-05, "loss": 0.0055, "step": 87850 }, { "epoch": 6.096512183972891e-13, "grad_norm": 0.20279955863952637, "learning_rate": 2.004743172968629e-05, "loss": 0.0053, "step": 87860 }, { "epoch": 6.097206073363282e-13, "grad_norm": 0.11228043586015701, "learning_rate": 2.003922478190499e-05, "loss": 0.0054, "step": 87870 }, { "epoch": 6.097899962753672e-13, "grad_norm": 0.09718836098909378, "learning_rate": 2.0031024252940193e-05, "loss": 0.0052, "step": 87880 }, { "epoch": 6.098593852144063e-13, "grad_norm": 0.08045466989278793, "learning_rate": 2.0022830143634626e-05, "loss": 0.0053, "step": 87890 }, { "epoch": 6.099287741534454e-13, "grad_norm": 0.09639301896095276, "learning_rate": 2.001464245483037e-05, "loss": 0.0053, "step": 87900 }, { "epoch": 6.099981630924844e-13, "grad_norm": 0.07045109570026398, "learning_rate": 2.0006461187368827e-05, "loss": 0.005, "step": 87910 }, { "epoch": 6.100675520315235e-13, "grad_norm": 0.09982877969741821, "learning_rate": 1.999828634209075e-05, "loss": 0.005, "step": 87920 }, { "epoch": 6.101369409705626e-13, "grad_norm": 0.0798180103302002, "learning_rate": 1.9990117919836232e-05, "loss": 0.0049, "step": 87930 }, { "epoch": 6.102063299096017e-13, "grad_norm": 0.09131360799074173, "learning_rate": 1.998195592144472e-05, "loss": 0.005, "step": 87940 }, { "epoch": 6.102757188486407e-13, "grad_norm": 0.11366615444421768, "learning_rate": 1.9973800347754965e-05, "loss": 0.005, "step": 87950 }, { "epoch": 6.103451077876798e-13, "grad_norm": 0.13078142702579498, "learning_rate": 1.9965651199605096e-05, "loss": 0.005, "step": 87960 }, { "epoch": 6.104144967267189e-13, "grad_norm": 0.08555015176534653, "learning_rate": 1.9957508477832555e-05, "loss": 0.005, "step": 87970 }, { "epoch": 6.10483885665758e-13, "grad_norm": 0.13789711892604828, "learning_rate": 1.994937218327415e-05, "loss": 0.0052, "step": 87980 }, { "epoch": 6.10553274604797e-13, "grad_norm": 0.10466960817575455, "learning_rate": 1.9941242316765968e-05, "loss": 0.0055, "step": 87990 }, { "epoch": 6.106226635438361e-13, "grad_norm": 0.1431806981563568, "learning_rate": 1.993311887914354e-05, "loss": 0.0054, "step": 88000 }, { "epoch": 6.106920524828752e-13, "grad_norm": 0.07097034901380539, "learning_rate": 1.9925001871241622e-05, "loss": 0.0052, "step": 88010 }, { "epoch": 6.107614414219142e-13, "grad_norm": 0.10513704270124435, "learning_rate": 1.9916891293894403e-05, "loss": 0.0051, "step": 88020 }, { "epoch": 6.108308303609533e-13, "grad_norm": 0.08811065554618835, "learning_rate": 1.9908787147935337e-05, "loss": 0.0052, "step": 88030 }, { "epoch": 6.109002192999924e-13, "grad_norm": 0.09716255962848663, "learning_rate": 1.9900689434197282e-05, "loss": 0.0054, "step": 88040 }, { "epoch": 6.109696082390315e-13, "grad_norm": 0.08671221137046814, "learning_rate": 1.9892598153512376e-05, "loss": 0.0053, "step": 88050 }, { "epoch": 6.110389971780705e-13, "grad_norm": 0.11751877516508102, "learning_rate": 1.9884513306712153e-05, "loss": 0.0051, "step": 88060 }, { "epoch": 6.111083861171096e-13, "grad_norm": 0.07944921404123306, "learning_rate": 1.987643489462744e-05, "loss": 0.005, "step": 88070 }, { "epoch": 6.111777750561487e-13, "grad_norm": 0.12346870452165604, "learning_rate": 1.9868362918088417e-05, "loss": 0.0049, "step": 88080 }, { "epoch": 6.112471639951877e-13, "grad_norm": 0.09343855828046799, "learning_rate": 1.9860297377924612e-05, "loss": 0.0051, "step": 88090 }, { "epoch": 6.113165529342268e-13, "grad_norm": 0.1611308753490448, "learning_rate": 1.985223827496488e-05, "loss": 0.005, "step": 88100 }, { "epoch": 6.113859418732659e-13, "grad_norm": 0.12693443894386292, "learning_rate": 1.9844185610037416e-05, "loss": 0.0049, "step": 88110 }, { "epoch": 6.11455330812305e-13, "grad_norm": 0.1074792817234993, "learning_rate": 1.9836139383969764e-05, "loss": 0.005, "step": 88120 }, { "epoch": 6.11524719751344e-13, "grad_norm": 0.14505988359451294, "learning_rate": 1.9828099597588796e-05, "loss": 0.0052, "step": 88130 }, { "epoch": 6.115941086903831e-13, "grad_norm": 0.1043282151222229, "learning_rate": 1.9820066251720737e-05, "loss": 0.0054, "step": 88140 }, { "epoch": 6.116634976294222e-13, "grad_norm": 0.11571280658245087, "learning_rate": 1.981203934719112e-05, "loss": 0.0053, "step": 88150 }, { "epoch": 6.117328865684613e-13, "grad_norm": 0.09226811677217484, "learning_rate": 1.980401888482484e-05, "loss": 0.0052, "step": 88160 }, { "epoch": 6.118022755075003e-13, "grad_norm": 0.11081677675247192, "learning_rate": 1.9796004865446128e-05, "loss": 0.0051, "step": 88170 }, { "epoch": 6.118716644465394e-13, "grad_norm": 0.07673675566911697, "learning_rate": 1.978799728987855e-05, "loss": 0.0053, "step": 88180 }, { "epoch": 6.119410533855785e-13, "grad_norm": 0.10243485122919083, "learning_rate": 1.9779996158945004e-05, "loss": 0.0051, "step": 88190 }, { "epoch": 6.120104423246175e-13, "grad_norm": 0.13120950758457184, "learning_rate": 1.9772001473467747e-05, "loss": 0.0052, "step": 88200 }, { "epoch": 6.120798312636566e-13, "grad_norm": 0.12225906550884247, "learning_rate": 1.976401323426832e-05, "loss": 0.0053, "step": 88210 }, { "epoch": 6.121492202026957e-13, "grad_norm": 0.06587961316108704, "learning_rate": 1.9756031442167675e-05, "loss": 0.0051, "step": 88220 }, { "epoch": 6.122186091417348e-13, "grad_norm": 0.10945107042789459, "learning_rate": 1.9748056097986046e-05, "loss": 0.005, "step": 88230 }, { "epoch": 6.122879980807738e-13, "grad_norm": 0.11383730173110962, "learning_rate": 1.9740087202543035e-05, "loss": 0.0049, "step": 88240 }, { "epoch": 6.123573870198129e-13, "grad_norm": 0.07218223065137863, "learning_rate": 1.9732124756657563e-05, "loss": 0.0051, "step": 88250 }, { "epoch": 6.12426775958852e-13, "grad_norm": 0.09107700735330582, "learning_rate": 1.9724168761147895e-05, "loss": 0.0049, "step": 88260 }, { "epoch": 6.12496164897891e-13, "grad_norm": 0.10143899917602539, "learning_rate": 1.971621921683164e-05, "loss": 0.0049, "step": 88270 }, { "epoch": 6.125655538369301e-13, "grad_norm": 0.1066390872001648, "learning_rate": 1.970827612452573e-05, "loss": 0.0051, "step": 88280 }, { "epoch": 6.126349427759692e-13, "grad_norm": 0.11141803115606308, "learning_rate": 1.9700339485046446e-05, "loss": 0.005, "step": 88290 }, { "epoch": 6.127043317150083e-13, "grad_norm": 0.073282390832901, "learning_rate": 1.9692409299209394e-05, "loss": 0.0053, "step": 88300 }, { "epoch": 6.127737206540473e-13, "grad_norm": 0.10448415577411652, "learning_rate": 1.9684485567829546e-05, "loss": 0.0052, "step": 88310 }, { "epoch": 6.128431095930864e-13, "grad_norm": 0.10872334986925125, "learning_rate": 1.9676568291721155e-05, "loss": 0.0052, "step": 88320 }, { "epoch": 6.129124985321255e-13, "grad_norm": 0.09720342606306076, "learning_rate": 1.9668657471697875e-05, "loss": 0.0053, "step": 88330 }, { "epoch": 6.129818874711646e-13, "grad_norm": 0.10862530767917633, "learning_rate": 1.9660753108572637e-05, "loss": 0.0051, "step": 88340 }, { "epoch": 6.130512764102036e-13, "grad_norm": 0.10921318829059601, "learning_rate": 1.965285520315777e-05, "loss": 0.0051, "step": 88350 }, { "epoch": 6.131206653492427e-13, "grad_norm": 0.10930231213569641, "learning_rate": 1.964496375626487e-05, "loss": 0.0052, "step": 88360 }, { "epoch": 6.131900542882818e-13, "grad_norm": 0.11931112408638, "learning_rate": 1.9637078768704946e-05, "loss": 0.0051, "step": 88370 }, { "epoch": 6.132594432273208e-13, "grad_norm": 0.11383254081010818, "learning_rate": 1.9629200241288257e-05, "loss": 0.0049, "step": 88380 }, { "epoch": 6.133288321663599e-13, "grad_norm": 0.10758722573518753, "learning_rate": 1.9621328174824495e-05, "loss": 0.0049, "step": 88390 }, { "epoch": 6.13398221105399e-13, "grad_norm": 0.10936049371957779, "learning_rate": 1.96134625701226e-05, "loss": 0.005, "step": 88400 }, { "epoch": 6.134676100444381e-13, "grad_norm": 0.11797133833169937, "learning_rate": 1.96056034279909e-05, "loss": 0.0051, "step": 88410 }, { "epoch": 6.135369989834771e-13, "grad_norm": 0.09707728028297424, "learning_rate": 1.9597750749237038e-05, "loss": 0.005, "step": 88420 }, { "epoch": 6.136063879225162e-13, "grad_norm": 0.11148081719875336, "learning_rate": 1.958990453466801e-05, "loss": 0.0049, "step": 88430 }, { "epoch": 6.136757768615553e-13, "grad_norm": 0.07123936712741852, "learning_rate": 1.958206478509012e-05, "loss": 0.005, "step": 88440 }, { "epoch": 6.137451658005944e-13, "grad_norm": 0.10108265280723572, "learning_rate": 1.957423150130904e-05, "loss": 0.0051, "step": 88450 }, { "epoch": 6.138145547396334e-13, "grad_norm": 0.09956569969654083, "learning_rate": 1.9566404684129755e-05, "loss": 0.0054, "step": 88460 }, { "epoch": 6.138839436786725e-13, "grad_norm": 0.08124953508377075, "learning_rate": 1.955858433435659e-05, "loss": 0.0052, "step": 88470 }, { "epoch": 6.139533326177116e-13, "grad_norm": 0.12873975932598114, "learning_rate": 1.9550770452793202e-05, "loss": 0.0053, "step": 88480 }, { "epoch": 6.140227215567506e-13, "grad_norm": 0.10265558212995529, "learning_rate": 1.9542963040242604e-05, "loss": 0.0052, "step": 88490 }, { "epoch": 6.140921104957897e-13, "grad_norm": 0.11340861767530441, "learning_rate": 1.9535162097507128e-05, "loss": 0.005, "step": 88500 }, { "epoch": 6.141614994348288e-13, "grad_norm": 0.11586227267980576, "learning_rate": 1.9527367625388425e-05, "loss": 0.005, "step": 88510 }, { "epoch": 6.142308883738679e-13, "grad_norm": 0.09909862279891968, "learning_rate": 1.9519579624687515e-05, "loss": 0.0052, "step": 88520 }, { "epoch": 6.143002773129069e-13, "grad_norm": 0.09353501349687576, "learning_rate": 1.9511798096204745e-05, "loss": 0.0051, "step": 88530 }, { "epoch": 6.14369666251946e-13, "grad_norm": 0.09264354407787323, "learning_rate": 1.9504023040739745e-05, "loss": 0.005, "step": 88540 }, { "epoch": 6.144390551909851e-13, "grad_norm": 0.09224719554185867, "learning_rate": 1.9496254459091575e-05, "loss": 0.0049, "step": 88550 }, { "epoch": 6.145084441300241e-13, "grad_norm": 0.09670115262269974, "learning_rate": 1.9488492352058538e-05, "loss": 0.005, "step": 88560 }, { "epoch": 6.145778330690632e-13, "grad_norm": 0.06024320796132088, "learning_rate": 1.9480736720438347e-05, "loss": 0.005, "step": 88570 }, { "epoch": 6.146472220081023e-13, "grad_norm": 0.12326578050851822, "learning_rate": 1.9472987565027974e-05, "loss": 0.0048, "step": 88580 }, { "epoch": 6.147166109471414e-13, "grad_norm": 0.12167659401893616, "learning_rate": 1.9465244886623797e-05, "loss": 0.0049, "step": 88590 }, { "epoch": 6.147859998861804e-13, "grad_norm": 0.1094147339463234, "learning_rate": 1.9457508686021488e-05, "loss": 0.0049, "step": 88600 }, { "epoch": 6.148553888252195e-13, "grad_norm": 0.10210287570953369, "learning_rate": 1.9449778964016056e-05, "loss": 0.0053, "step": 88610 }, { "epoch": 6.149247777642586e-13, "grad_norm": 0.13134463131427765, "learning_rate": 1.9442055721401852e-05, "loss": 0.0054, "step": 88620 }, { "epoch": 6.149941667032977e-13, "grad_norm": 0.179733008146286, "learning_rate": 1.943433895897256e-05, "loss": 0.0052, "step": 88630 }, { "epoch": 6.150635556423367e-13, "grad_norm": 0.11131781339645386, "learning_rate": 1.94266286775212e-05, "loss": 0.0052, "step": 88640 }, { "epoch": 6.151329445813758e-13, "grad_norm": 0.11647067964076996, "learning_rate": 1.9418924877840122e-05, "loss": 0.0053, "step": 88650 }, { "epoch": 6.152023335204149e-13, "grad_norm": 0.12068600207567215, "learning_rate": 1.9411227560721017e-05, "loss": 0.0051, "step": 88660 }, { "epoch": 6.152717224594539e-13, "grad_norm": 0.14281277358531952, "learning_rate": 1.9403536726954887e-05, "loss": 0.0052, "step": 88670 }, { "epoch": 6.15341111398493e-13, "grad_norm": 0.08272840082645416, "learning_rate": 1.93958523773321e-05, "loss": 0.0051, "step": 88680 }, { "epoch": 6.154105003375321e-13, "grad_norm": 0.14895020425319672, "learning_rate": 1.9388174512642338e-05, "loss": 0.0051, "step": 88690 }, { "epoch": 6.154798892765712e-13, "grad_norm": 0.11433063447475433, "learning_rate": 1.938050313367463e-05, "loss": 0.0049, "step": 88700 }, { "epoch": 6.155492782156102e-13, "grad_norm": 0.11272274702787399, "learning_rate": 1.9372838241217303e-05, "loss": 0.0049, "step": 88710 }, { "epoch": 6.156186671546493e-13, "grad_norm": 0.07937031239271164, "learning_rate": 1.936517983605809e-05, "loss": 0.0049, "step": 88720 }, { "epoch": 6.156880560936884e-13, "grad_norm": 0.08945588767528534, "learning_rate": 1.935752791898396e-05, "loss": 0.0049, "step": 88730 }, { "epoch": 6.157574450327274e-13, "grad_norm": 0.1461302936077118, "learning_rate": 1.9349882490781298e-05, "loss": 0.0049, "step": 88740 }, { "epoch": 6.158268339717665e-13, "grad_norm": 0.11693978309631348, "learning_rate": 1.9342243552235778e-05, "loss": 0.005, "step": 88750 }, { "epoch": 6.158962229108056e-13, "grad_norm": 0.14488163590431213, "learning_rate": 1.9334611104132434e-05, "loss": 0.0049, "step": 88760 }, { "epoch": 6.159656118498447e-13, "grad_norm": 0.12613309919834137, "learning_rate": 1.9326985147255592e-05, "loss": 0.0052, "step": 88770 }, { "epoch": 6.160350007888837e-13, "grad_norm": 0.11330727487802505, "learning_rate": 1.931936568238897e-05, "loss": 0.0052, "step": 88780 }, { "epoch": 6.161043897279228e-13, "grad_norm": 0.09319525957107544, "learning_rate": 1.9311752710315567e-05, "loss": 0.0053, "step": 88790 }, { "epoch": 6.161737786669619e-13, "grad_norm": 0.09761946648359299, "learning_rate": 1.930414623181774e-05, "loss": 0.0056, "step": 88800 }, { "epoch": 6.16243167606001e-13, "grad_norm": 0.10295873880386353, "learning_rate": 1.929654624767717e-05, "loss": 0.0053, "step": 88810 }, { "epoch": 6.1631255654504e-13, "grad_norm": 0.17657499015331268, "learning_rate": 1.9288952758674875e-05, "loss": 0.0052, "step": 88820 }, { "epoch": 6.163819454840791e-13, "grad_norm": 0.08503354340791702, "learning_rate": 1.9281365765591206e-05, "loss": 0.0051, "step": 88830 }, { "epoch": 6.164513344231182e-13, "grad_norm": 0.08686326444149017, "learning_rate": 1.927378526920584e-05, "loss": 0.005, "step": 88840 }, { "epoch": 6.165207233621572e-13, "grad_norm": 0.07995645701885223, "learning_rate": 1.9266211270297804e-05, "loss": 0.0051, "step": 88850 }, { "epoch": 6.165901123011963e-13, "grad_norm": 0.12166281789541245, "learning_rate": 1.925864376964544e-05, "loss": 0.0049, "step": 88860 }, { "epoch": 6.166595012402354e-13, "grad_norm": 0.06475821137428284, "learning_rate": 1.9251082768026396e-05, "loss": 0.005, "step": 88870 }, { "epoch": 6.167288901792745e-13, "grad_norm": 0.10685697942972183, "learning_rate": 1.924352826621773e-05, "loss": 0.0048, "step": 88880 }, { "epoch": 6.167982791183135e-13, "grad_norm": 0.10796620696783066, "learning_rate": 1.9235980264995737e-05, "loss": 0.0048, "step": 88890 }, { "epoch": 6.168676680573526e-13, "grad_norm": 0.06614427268505096, "learning_rate": 1.9228438765136138e-05, "loss": 0.0049, "step": 88900 }, { "epoch": 6.169370569963917e-13, "grad_norm": 0.08344034105539322, "learning_rate": 1.9220903767413903e-05, "loss": 0.0049, "step": 88910 }, { "epoch": 6.170064459354307e-13, "grad_norm": 0.09908666461706161, "learning_rate": 1.9213375272603405e-05, "loss": 0.0051, "step": 88920 }, { "epoch": 6.170758348744698e-13, "grad_norm": 0.24217018485069275, "learning_rate": 1.920585328147828e-05, "loss": 0.0055, "step": 88930 }, { "epoch": 6.171452238135089e-13, "grad_norm": 0.08094517886638641, "learning_rate": 1.919833779481155e-05, "loss": 0.0051, "step": 88940 }, { "epoch": 6.17214612752548e-13, "grad_norm": 0.08157701045274734, "learning_rate": 1.9190828813375533e-05, "loss": 0.0053, "step": 88950 }, { "epoch": 6.17284001691587e-13, "grad_norm": 0.11577226966619492, "learning_rate": 1.918332633794191e-05, "loss": 0.0054, "step": 88960 }, { "epoch": 6.173533906306261e-13, "grad_norm": 0.11940257251262665, "learning_rate": 1.9175830369281675e-05, "loss": 0.0052, "step": 88970 }, { "epoch": 6.174227795696652e-13, "grad_norm": 0.11558021605014801, "learning_rate": 1.916834090816514e-05, "loss": 0.0052, "step": 88980 }, { "epoch": 6.174921685087043e-13, "grad_norm": 0.10169629007577896, "learning_rate": 1.9160857955361973e-05, "loss": 0.0053, "step": 88990 }, { "epoch": 6.175615574477433e-13, "grad_norm": 0.07156495749950409, "learning_rate": 1.9153381511641168e-05, "loss": 0.005, "step": 89000 }, { "epoch": 6.176309463867824e-13, "grad_norm": 0.11512205749750137, "learning_rate": 1.9145911577771052e-05, "loss": 0.0051, "step": 89010 }, { "epoch": 6.177003353258215e-13, "grad_norm": 0.09507051110267639, "learning_rate": 1.9138448154519254e-05, "loss": 0.0049, "step": 89020 }, { "epoch": 6.177697242648605e-13, "grad_norm": 0.07116558402776718, "learning_rate": 1.9130991242652777e-05, "loss": 0.005, "step": 89030 }, { "epoch": 6.178391132038996e-13, "grad_norm": 0.0972541943192482, "learning_rate": 1.9123540842937926e-05, "loss": 0.005, "step": 89040 }, { "epoch": 6.179085021429387e-13, "grad_norm": 0.1286972314119339, "learning_rate": 1.9116096956140353e-05, "loss": 0.0049, "step": 89050 }, { "epoch": 6.179778910819778e-13, "grad_norm": 0.12320684641599655, "learning_rate": 1.9108659583025036e-05, "loss": 0.0051, "step": 89060 }, { "epoch": 6.180472800210168e-13, "grad_norm": 0.0812196284532547, "learning_rate": 1.9101228724356247e-05, "loss": 0.0049, "step": 89070 }, { "epoch": 6.181166689600559e-13, "grad_norm": 0.08125194162130356, "learning_rate": 1.909380438089767e-05, "loss": 0.0051, "step": 89080 }, { "epoch": 6.18186057899095e-13, "grad_norm": 0.10525073111057281, "learning_rate": 1.9086386553412228e-05, "loss": 0.0052, "step": 89090 }, { "epoch": 6.18255446838134e-13, "grad_norm": 0.11317172646522522, "learning_rate": 1.9078975242662264e-05, "loss": 0.0051, "step": 89100 }, { "epoch": 6.183248357771731e-13, "grad_norm": 0.10012117773294449, "learning_rate": 1.907157044940936e-05, "loss": 0.0054, "step": 89110 }, { "epoch": 6.183942247162122e-13, "grad_norm": 0.12302170693874359, "learning_rate": 1.9064172174414512e-05, "loss": 0.0052, "step": 89120 }, { "epoch": 6.184636136552513e-13, "grad_norm": 0.13124366104602814, "learning_rate": 1.9056780418437985e-05, "loss": 0.0054, "step": 89130 }, { "epoch": 6.185330025942903e-13, "grad_norm": 0.10218732059001923, "learning_rate": 1.90493951822394e-05, "loss": 0.0056, "step": 89140 }, { "epoch": 6.186023915333294e-13, "grad_norm": 0.09521826356649399, "learning_rate": 1.9042016466577713e-05, "loss": 0.0051, "step": 89150 }, { "epoch": 6.186717804723685e-13, "grad_norm": 0.0901232585310936, "learning_rate": 1.9034644272211194e-05, "loss": 0.0051, "step": 89160 }, { "epoch": 6.187411694114076e-13, "grad_norm": 0.0765138491988182, "learning_rate": 1.9027278599897454e-05, "loss": 0.0052, "step": 89170 }, { "epoch": 6.188105583504466e-13, "grad_norm": 0.11728421598672867, "learning_rate": 1.9019919450393435e-05, "loss": 0.0051, "step": 89180 }, { "epoch": 6.188799472894857e-13, "grad_norm": 0.08364700525999069, "learning_rate": 1.9012566824455398e-05, "loss": 0.0052, "step": 89190 }, { "epoch": 6.189493362285248e-13, "grad_norm": 0.08920610696077347, "learning_rate": 1.900522072283893e-05, "loss": 0.0052, "step": 89200 }, { "epoch": 6.190187251675638e-13, "grad_norm": 0.14256705343723297, "learning_rate": 1.8997881146298986e-05, "loss": 0.0054, "step": 89210 }, { "epoch": 6.190881141066029e-13, "grad_norm": 0.0883970856666565, "learning_rate": 1.899054809558978e-05, "loss": 0.0053, "step": 89220 }, { "epoch": 6.19157503045642e-13, "grad_norm": 0.08858611434698105, "learning_rate": 1.8983221571464943e-05, "loss": 0.0051, "step": 89230 }, { "epoch": 6.192268919846811e-13, "grad_norm": 0.08859515190124512, "learning_rate": 1.8975901574677345e-05, "loss": 0.0052, "step": 89240 }, { "epoch": 6.192962809237201e-13, "grad_norm": 0.11143546551465988, "learning_rate": 1.8968588105979272e-05, "loss": 0.0054, "step": 89250 }, { "epoch": 6.193656698627592e-13, "grad_norm": 0.08980704843997955, "learning_rate": 1.896128116612226e-05, "loss": 0.0053, "step": 89260 }, { "epoch": 6.194350588017983e-13, "grad_norm": 0.10723426192998886, "learning_rate": 1.8953980755857247e-05, "loss": 0.0052, "step": 89270 }, { "epoch": 6.195044477408373e-13, "grad_norm": 0.09360603988170624, "learning_rate": 1.894668687593443e-05, "loss": 0.0052, "step": 89280 }, { "epoch": 6.195738366798764e-13, "grad_norm": 0.11095739156007767, "learning_rate": 1.8939399527103383e-05, "loss": 0.0052, "step": 89290 }, { "epoch": 6.196432256189155e-13, "grad_norm": 0.114882692694664, "learning_rate": 1.8932118710112992e-05, "loss": 0.0052, "step": 89300 }, { "epoch": 6.197126145579546e-13, "grad_norm": 0.1007324680685997, "learning_rate": 1.8924844425711484e-05, "loss": 0.0051, "step": 89310 }, { "epoch": 6.197820034969936e-13, "grad_norm": 0.09732320159673691, "learning_rate": 1.891757667464639e-05, "loss": 0.0052, "step": 89320 }, { "epoch": 6.198513924360327e-13, "grad_norm": 0.10452030599117279, "learning_rate": 1.8910315457664592e-05, "loss": 0.0051, "step": 89330 }, { "epoch": 6.199207813750718e-13, "grad_norm": 0.08704341948032379, "learning_rate": 1.8903060775512297e-05, "loss": 0.0051, "step": 89340 }, { "epoch": 6.199901703141109e-13, "grad_norm": 0.12410597503185272, "learning_rate": 1.8895812628935027e-05, "loss": 0.0051, "step": 89350 }, { "epoch": 6.200595592531499e-13, "grad_norm": 0.11075472831726074, "learning_rate": 1.8888571018677654e-05, "loss": 0.0052, "step": 89360 }, { "epoch": 6.20128948192189e-13, "grad_norm": 0.08742687851190567, "learning_rate": 1.888133594548435e-05, "loss": 0.0054, "step": 89370 }, { "epoch": 6.201983371312281e-13, "grad_norm": 0.08696740865707397, "learning_rate": 1.8874107410098647e-05, "loss": 0.0052, "step": 89380 }, { "epoch": 6.202677260702671e-13, "grad_norm": 0.10185439139604568, "learning_rate": 1.8866885413263397e-05, "loss": 0.0053, "step": 89390 }, { "epoch": 6.203371150093062e-13, "grad_norm": 0.10394277423620224, "learning_rate": 1.8859669955720722e-05, "loss": 0.0056, "step": 89400 }, { "epoch": 6.204065039483453e-13, "grad_norm": 0.08755791187286377, "learning_rate": 1.8852461038212197e-05, "loss": 0.0051, "step": 89410 }, { "epoch": 6.204758928873844e-13, "grad_norm": 0.11839032173156738, "learning_rate": 1.8845258661478582e-05, "loss": 0.0052, "step": 89420 }, { "epoch": 6.205452818264234e-13, "grad_norm": 0.08283891528844833, "learning_rate": 1.8838062826260085e-05, "loss": 0.0053, "step": 89430 }, { "epoch": 6.206146707654625e-13, "grad_norm": 0.08277387171983719, "learning_rate": 1.8830873533296154e-05, "loss": 0.0051, "step": 89440 }, { "epoch": 6.206840597045016e-13, "grad_norm": 0.13502340018749237, "learning_rate": 1.8823690783325627e-05, "loss": 0.0051, "step": 89450 }, { "epoch": 6.207534486435407e-13, "grad_norm": 0.07282912731170654, "learning_rate": 1.8816514577086617e-05, "loss": 0.0053, "step": 89460 }, { "epoch": 6.208228375825797e-13, "grad_norm": 0.08263463526964188, "learning_rate": 1.8809344915316627e-05, "loss": 0.005, "step": 89470 }, { "epoch": 6.208922265216188e-13, "grad_norm": 0.0906217098236084, "learning_rate": 1.880218179875242e-05, "loss": 0.0052, "step": 89480 }, { "epoch": 6.209616154606579e-13, "grad_norm": 0.1006147488951683, "learning_rate": 1.8795025228130124e-05, "loss": 0.0051, "step": 89490 }, { "epoch": 6.210310043996969e-13, "grad_norm": 0.09785891324281693, "learning_rate": 1.8787875204185185e-05, "loss": 0.005, "step": 89500 }, { "epoch": 6.21100393338736e-13, "grad_norm": 0.08329293876886368, "learning_rate": 1.8780731727652398e-05, "loss": 0.005, "step": 89510 }, { "epoch": 6.211697822777751e-13, "grad_norm": 0.0794922336935997, "learning_rate": 1.8773594799265852e-05, "loss": 0.0051, "step": 89520 }, { "epoch": 6.212391712168142e-13, "grad_norm": 0.1052350178360939, "learning_rate": 1.8766464419758973e-05, "loss": 0.0053, "step": 89530 }, { "epoch": 6.213085601558532e-13, "grad_norm": 0.10605458915233612, "learning_rate": 1.8759340589864537e-05, "loss": 0.0053, "step": 89540 }, { "epoch": 6.213779490948923e-13, "grad_norm": 0.11925844848155975, "learning_rate": 1.8752223310314602e-05, "loss": 0.0055, "step": 89550 }, { "epoch": 6.214473380339314e-13, "grad_norm": 0.09425685554742813, "learning_rate": 1.87451125818406e-05, "loss": 0.0057, "step": 89560 }, { "epoch": 6.215167269729704e-13, "grad_norm": 0.10061188042163849, "learning_rate": 1.8738008405173265e-05, "loss": 0.0052, "step": 89570 }, { "epoch": 6.215861159120095e-13, "grad_norm": 0.13005924224853516, "learning_rate": 1.8730910781042665e-05, "loss": 0.0054, "step": 89580 }, { "epoch": 6.216555048510486e-13, "grad_norm": 0.10385201871395111, "learning_rate": 1.872381971017817e-05, "loss": 0.0053, "step": 89590 }, { "epoch": 6.217248937900877e-13, "grad_norm": 0.10429397225379944, "learning_rate": 1.8716735193308533e-05, "loss": 0.0051, "step": 89600 }, { "epoch": 6.217942827291267e-13, "grad_norm": 0.10308411717414856, "learning_rate": 1.8709657231161763e-05, "loss": 0.0052, "step": 89610 }, { "epoch": 6.218636716681658e-13, "grad_norm": 0.10099636763334274, "learning_rate": 1.870258582446525e-05, "loss": 0.0051, "step": 89620 }, { "epoch": 6.219330606072049e-13, "grad_norm": 0.06872563064098358, "learning_rate": 1.869552097394569e-05, "loss": 0.005, "step": 89630 }, { "epoch": 6.22002449546244e-13, "grad_norm": 0.10331161320209503, "learning_rate": 1.8688462680329103e-05, "loss": 0.0051, "step": 89640 }, { "epoch": 6.22071838485283e-13, "grad_norm": 0.07261158525943756, "learning_rate": 1.8681410944340835e-05, "loss": 0.005, "step": 89650 }, { "epoch": 6.221412274243221e-13, "grad_norm": 0.07991491258144379, "learning_rate": 1.867436576670557e-05, "loss": 0.0049, "step": 89660 }, { "epoch": 6.222106163633612e-13, "grad_norm": 0.1849789023399353, "learning_rate": 1.8667327148147306e-05, "loss": 0.0051, "step": 89670 }, { "epoch": 6.222800053024002e-13, "grad_norm": 0.09683290868997574, "learning_rate": 1.866029508938937e-05, "loss": 0.0051, "step": 89680 }, { "epoch": 6.223493942414393e-13, "grad_norm": 0.11180226504802704, "learning_rate": 1.8653269591154413e-05, "loss": 0.0052, "step": 89690 }, { "epoch": 6.224187831804784e-13, "grad_norm": 0.10121311247348785, "learning_rate": 1.8646250654164417e-05, "loss": 0.0056, "step": 89700 }, { "epoch": 6.224881721195175e-13, "grad_norm": 0.130515456199646, "learning_rate": 1.863923827914069e-05, "loss": 0.0061, "step": 89710 }, { "epoch": 6.225575610585565e-13, "grad_norm": 0.07594870030879974, "learning_rate": 1.863223246680387e-05, "loss": 0.0052, "step": 89720 }, { "epoch": 6.226269499975956e-13, "grad_norm": 0.11475086957216263, "learning_rate": 1.8625233217873882e-05, "loss": 0.0052, "step": 89730 }, { "epoch": 6.226963389366347e-13, "grad_norm": 0.08680588006973267, "learning_rate": 1.861824053307004e-05, "loss": 0.0055, "step": 89740 }, { "epoch": 6.227657278756737e-13, "grad_norm": 0.12102308124303818, "learning_rate": 1.861125441311093e-05, "loss": 0.0053, "step": 89750 }, { "epoch": 6.228351168147128e-13, "grad_norm": 0.09055336564779282, "learning_rate": 1.860427485871451e-05, "loss": 0.005, "step": 89760 }, { "epoch": 6.229045057537519e-13, "grad_norm": 0.11754113435745239, "learning_rate": 1.8597301870598002e-05, "loss": 0.0051, "step": 89770 }, { "epoch": 6.22973894692791e-13, "grad_norm": 0.10403163731098175, "learning_rate": 1.8590335449478033e-05, "loss": 0.005, "step": 89780 }, { "epoch": 6.2304328363183e-13, "grad_norm": 0.1012902706861496, "learning_rate": 1.858337559607046e-05, "loss": 0.005, "step": 89790 }, { "epoch": 6.231126725708691e-13, "grad_norm": 0.16704757511615753, "learning_rate": 1.857642231109057e-05, "loss": 0.0052, "step": 89800 }, { "epoch": 6.231820615099082e-13, "grad_norm": 0.06774799525737762, "learning_rate": 1.856947559525288e-05, "loss": 0.0049, "step": 89810 }, { "epoch": 6.232514504489473e-13, "grad_norm": 0.11606753617525101, "learning_rate": 1.8562535449271286e-05, "loss": 0.0051, "step": 89820 }, { "epoch": 6.233208393879863e-13, "grad_norm": 0.11075150221586227, "learning_rate": 1.8555601873859008e-05, "loss": 0.0055, "step": 89830 }, { "epoch": 6.233902283270254e-13, "grad_norm": 0.08517253398895264, "learning_rate": 1.8548674869728558e-05, "loss": 0.0051, "step": 89840 }, { "epoch": 6.234596172660645e-13, "grad_norm": 0.07298902422189713, "learning_rate": 1.854175443759181e-05, "loss": 0.0052, "step": 89850 }, { "epoch": 6.235290062051035e-13, "grad_norm": 0.11563093960285187, "learning_rate": 1.8534840578159936e-05, "loss": 0.0054, "step": 89860 }, { "epoch": 6.235983951441426e-13, "grad_norm": 0.10742247104644775, "learning_rate": 1.8527933292143445e-05, "loss": 0.0061, "step": 89870 }, { "epoch": 6.236677840831817e-13, "grad_norm": 0.10200807452201843, "learning_rate": 1.8521032580252177e-05, "loss": 0.0053, "step": 89880 }, { "epoch": 6.237371730222208e-13, "grad_norm": 0.10461524873971939, "learning_rate": 1.8514138443195275e-05, "loss": 0.0052, "step": 89890 }, { "epoch": 6.238065619612598e-13, "grad_norm": 0.08083292841911316, "learning_rate": 1.8507250881681224e-05, "loss": 0.0055, "step": 89900 }, { "epoch": 6.238759509002989e-13, "grad_norm": 0.12841445207595825, "learning_rate": 1.8500369896417833e-05, "loss": 0.0051, "step": 89910 }, { "epoch": 6.23945339839338e-13, "grad_norm": 0.0858033299446106, "learning_rate": 1.8493495488112217e-05, "loss": 0.0051, "step": 89920 }, { "epoch": 6.24014728778377e-13, "grad_norm": 0.10322804003953934, "learning_rate": 1.8486627657470844e-05, "loss": 0.0051, "step": 89930 }, { "epoch": 6.240841177174161e-13, "grad_norm": 0.14251786470413208, "learning_rate": 1.847976640519949e-05, "loss": 0.0051, "step": 89940 }, { "epoch": 6.241535066564552e-13, "grad_norm": 0.08825992792844772, "learning_rate": 1.8472911732003234e-05, "loss": 0.005, "step": 89950 }, { "epoch": 6.242228955954943e-13, "grad_norm": 0.09310390055179596, "learning_rate": 1.846606363858653e-05, "loss": 0.0051, "step": 89960 }, { "epoch": 6.242922845345333e-13, "grad_norm": 0.07522889971733093, "learning_rate": 1.8459222125653087e-05, "loss": 0.0051, "step": 89970 }, { "epoch": 6.243616734735724e-13, "grad_norm": 0.08356285095214844, "learning_rate": 1.845238719390603e-05, "loss": 0.005, "step": 89980 }, { "epoch": 6.244310624126115e-13, "grad_norm": 0.06873329728841782, "learning_rate": 1.8445558844047702e-05, "loss": 0.0057, "step": 89990 }, { "epoch": 6.245004513516506e-13, "grad_norm": 0.10935172438621521, "learning_rate": 1.8438737076779865e-05, "loss": 0.0051, "step": 90000 } ], "logging_steps": 10, "max_steps": 100000, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 5000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 3.34142449385472e+18, "train_batch_size": 8, "trial_name": null, "trial_params": null }