{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0816681711721685e-13, "eval_steps": 500, "global_step": 60000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 3.469446951953614e-17, "grad_norm": 2.9650447368621826, "learning_rate": 1.25e-06, "loss": 10.4961, "step": 10 }, { "epoch": 6.938893903907228e-17, "grad_norm": 2.7266697883605957, "learning_rate": 2.5e-06, "loss": 10.447, "step": 20 }, { "epoch": 1.0408340855860843e-16, "grad_norm": 3.4442460536956787, "learning_rate": 3.75e-06, "loss": 10.3415, "step": 30 }, { "epoch": 1.3877787807814457e-16, "grad_norm": 3.1194708347320557, "learning_rate": 5e-06, "loss": 10.1604, "step": 40 }, { "epoch": 1.734723475976807e-16, "grad_norm": 2.316713333129883, "learning_rate": 6.25e-06, "loss": 9.9157, "step": 50 }, { "epoch": 2.0816681711721685e-16, "grad_norm": 1.8460209369659424, "learning_rate": 7.5e-06, "loss": 9.726, "step": 60 }, { "epoch": 2.42861286636753e-16, "grad_norm": 1.7614326477050781, "learning_rate": 8.750000000000001e-06, "loss": 9.5575, "step": 70 }, { "epoch": 2.7755575615628914e-16, "grad_norm": 1.6680262088775635, "learning_rate": 1e-05, "loss": 9.4509, "step": 80 }, { "epoch": 3.122502256758253e-16, "grad_norm": 1.6773438453674316, "learning_rate": 1.1249999999999999e-05, "loss": 9.3629, "step": 90 }, { "epoch": 3.469446951953614e-16, "grad_norm": 1.6793391704559326, "learning_rate": 1.25e-05, "loss": 9.2789, "step": 100 }, { "epoch": 3.8163916471489756e-16, "grad_norm": 1.6503127813339233, "learning_rate": 1.375e-05, "loss": 9.1765, "step": 110 }, { "epoch": 4.163336342344337e-16, "grad_norm": 1.413922667503357, "learning_rate": 1.5e-05, "loss": 9.0796, "step": 120 }, { "epoch": 4.510281037539698e-16, "grad_norm": 1.6549742221832275, "learning_rate": 1.6250000000000002e-05, "loss": 8.9366, "step": 130 }, { "epoch": 4.85722573273506e-16, "grad_norm": 1.6644552946090698, "learning_rate": 1.7500000000000002e-05, "loss": 8.8063, "step": 140 }, { "epoch": 5.204170427930421e-16, "grad_norm": 1.6726809740066528, "learning_rate": 1.875e-05, "loss": 8.6547, "step": 150 }, { "epoch": 5.551115123125783e-16, "grad_norm": 1.3429627418518066, "learning_rate": 2e-05, "loss": 8.4825, "step": 160 }, { "epoch": 5.898059818321144e-16, "grad_norm": 1.3818439245224, "learning_rate": 2.125e-05, "loss": 8.3539, "step": 170 }, { "epoch": 6.245004513516506e-16, "grad_norm": 1.2484933137893677, "learning_rate": 2.2499999999999998e-05, "loss": 8.2253, "step": 180 }, { "epoch": 6.591949208711867e-16, "grad_norm": 1.3431599140167236, "learning_rate": 2.375e-05, "loss": 8.0984, "step": 190 }, { "epoch": 6.938893903907228e-16, "grad_norm": 1.2077322006225586, "learning_rate": 2.5e-05, "loss": 7.9794, "step": 200 }, { "epoch": 7.28583859910259e-16, "grad_norm": 1.2059931755065918, "learning_rate": 2.625e-05, "loss": 7.8712, "step": 210 }, { "epoch": 7.632783294297951e-16, "grad_norm": 1.0400508642196655, "learning_rate": 2.75e-05, "loss": 7.7504, "step": 220 }, { "epoch": 7.979727989493313e-16, "grad_norm": 1.722244143486023, "learning_rate": 2.875e-05, "loss": 7.6519, "step": 230 }, { "epoch": 8.326672684688674e-16, "grad_norm": 0.9291983842849731, "learning_rate": 3e-05, "loss": 7.5326, "step": 240 }, { "epoch": 8.673617379884035e-16, "grad_norm": 1.5803189277648926, "learning_rate": 3.125e-05, "loss": 7.4771, "step": 250 }, { "epoch": 9.020562075079397e-16, "grad_norm": 1.2373018264770508, "learning_rate": 3.2500000000000004e-05, "loss": 7.3619, "step": 260 }, { "epoch": 9.367506770274758e-16, "grad_norm": 1.5016615390777588, "learning_rate": 3.375e-05, "loss": 7.3008, "step": 270 }, { "epoch": 9.71445146547012e-16, "grad_norm": 2.444209337234497, "learning_rate": 3.5000000000000004e-05, "loss": 7.2271, "step": 280 }, { "epoch": 1.0061396160665481e-15, "grad_norm": 1.6730331182479858, "learning_rate": 3.625e-05, "loss": 7.1369, "step": 290 }, { "epoch": 1.0408340855860843e-15, "grad_norm": 1.073745608329773, "learning_rate": 3.75e-05, "loss": 7.06, "step": 300 }, { "epoch": 1.0755285551056204e-15, "grad_norm": 2.2394232749938965, "learning_rate": 3.875e-05, "loss": 7.0158, "step": 310 }, { "epoch": 1.1102230246251565e-15, "grad_norm": 3.680305004119873, "learning_rate": 4e-05, "loss": 6.9461, "step": 320 }, { "epoch": 1.1449174941446927e-15, "grad_norm": 2.0185418128967285, "learning_rate": 4.125e-05, "loss": 6.9186, "step": 330 }, { "epoch": 1.1796119636642288e-15, "grad_norm": 3.4356203079223633, "learning_rate": 4.25e-05, "loss": 6.8343, "step": 340 }, { "epoch": 1.214306433183765e-15, "grad_norm": 2.132859230041504, "learning_rate": 4.375e-05, "loss": 6.7868, "step": 350 }, { "epoch": 1.2490009027033011e-15, "grad_norm": 2.238842010498047, "learning_rate": 4.4999999999999996e-05, "loss": 6.7227, "step": 360 }, { "epoch": 1.2836953722228372e-15, "grad_norm": 3.1938469409942627, "learning_rate": 4.625e-05, "loss": 6.6739, "step": 370 }, { "epoch": 1.3183898417423734e-15, "grad_norm": 2.551469564437866, "learning_rate": 4.75e-05, "loss": 6.6253, "step": 380 }, { "epoch": 1.3530843112619095e-15, "grad_norm": 1.2076300382614136, "learning_rate": 4.8750000000000006e-05, "loss": 6.5931, "step": 390 }, { "epoch": 1.3877787807814457e-15, "grad_norm": 1.6426105499267578, "learning_rate": 5e-05, "loss": 6.5104, "step": 400 }, { "epoch": 1.4224732503009818e-15, "grad_norm": 1.842815637588501, "learning_rate": 5.125e-05, "loss": 6.4636, "step": 410 }, { "epoch": 1.457167719820518e-15, "grad_norm": 2.054093599319458, "learning_rate": 5.25e-05, "loss": 6.3933, "step": 420 }, { "epoch": 1.4918621893400541e-15, "grad_norm": 2.4590888023376465, "learning_rate": 5.375e-05, "loss": 6.3629, "step": 430 }, { "epoch": 1.5265566588595902e-15, "grad_norm": 3.189297914505005, "learning_rate": 5.5e-05, "loss": 6.3123, "step": 440 }, { "epoch": 1.5612511283791264e-15, "grad_norm": 2.3856773376464844, "learning_rate": 5.6250000000000005e-05, "loss": 6.2728, "step": 450 }, { "epoch": 1.5959455978986625e-15, "grad_norm": 3.2710888385772705, "learning_rate": 5.75e-05, "loss": 6.2031, "step": 460 }, { "epoch": 1.6306400674181987e-15, "grad_norm": 1.8460352420806885, "learning_rate": 5.875e-05, "loss": 6.203, "step": 470 }, { "epoch": 1.6653345369377348e-15, "grad_norm": 2.2415506839752197, "learning_rate": 6e-05, "loss": 6.1427, "step": 480 }, { "epoch": 1.700029006457271e-15, "grad_norm": 1.6626116037368774, "learning_rate": 6.125e-05, "loss": 6.1043, "step": 490 }, { "epoch": 1.734723475976807e-15, "grad_norm": 1.8126109838485718, "learning_rate": 6.25e-05, "loss": 6.0609, "step": 500 }, { "epoch": 1.7694179454963432e-15, "grad_norm": 2.3724911212921143, "learning_rate": 6.375e-05, "loss": 5.9916, "step": 510 }, { "epoch": 1.8041124150158794e-15, "grad_norm": 1.9395197629928589, "learning_rate": 6.500000000000001e-05, "loss": 5.9928, "step": 520 }, { "epoch": 1.8388068845354155e-15, "grad_norm": 2.246708869934082, "learning_rate": 6.625000000000001e-05, "loss": 5.9645, "step": 530 }, { "epoch": 1.8735013540549517e-15, "grad_norm": 1.8528831005096436, "learning_rate": 6.75e-05, "loss": 5.8988, "step": 540 }, { "epoch": 1.908195823574488e-15, "grad_norm": 1.8359224796295166, "learning_rate": 6.875e-05, "loss": 5.8728, "step": 550 }, { "epoch": 1.942890293094024e-15, "grad_norm": 2.5528690814971924, "learning_rate": 7.000000000000001e-05, "loss": 5.8439, "step": 560 }, { "epoch": 1.97758476261356e-15, "grad_norm": 2.0963072776794434, "learning_rate": 7.125e-05, "loss": 5.8277, "step": 570 }, { "epoch": 2.0122792321330962e-15, "grad_norm": 2.849673271179199, "learning_rate": 7.25e-05, "loss": 5.7307, "step": 580 }, { "epoch": 2.0469737016526324e-15, "grad_norm": 2.4977662563323975, "learning_rate": 7.375e-05, "loss": 5.7902, "step": 590 }, { "epoch": 2.0816681711721685e-15, "grad_norm": 2.3966333866119385, "learning_rate": 7.5e-05, "loss": 5.7587, "step": 600 }, { "epoch": 2.1163626406917047e-15, "grad_norm": 2.4535083770751953, "learning_rate": 7.625e-05, "loss": 5.7153, "step": 610 }, { "epoch": 2.1510571102112408e-15, "grad_norm": 2.380038022994995, "learning_rate": 7.75e-05, "loss": 5.7053, "step": 620 }, { "epoch": 2.185751579730777e-15, "grad_norm": 1.4641388654708862, "learning_rate": 7.875e-05, "loss": 5.641, "step": 630 }, { "epoch": 2.220446049250313e-15, "grad_norm": 1.4955519437789917, "learning_rate": 8e-05, "loss": 5.5903, "step": 640 }, { "epoch": 2.2551405187698492e-15, "grad_norm": 3.2592275142669678, "learning_rate": 8.125000000000001e-05, "loss": 5.6062, "step": 650 }, { "epoch": 2.2898349882893854e-15, "grad_norm": 1.638158917427063, "learning_rate": 8.25e-05, "loss": 5.5805, "step": 660 }, { "epoch": 2.3245294578089215e-15, "grad_norm": 2.124303102493286, "learning_rate": 8.375e-05, "loss": 5.5433, "step": 670 }, { "epoch": 2.3592239273284576e-15, "grad_norm": 2.1727302074432373, "learning_rate": 8.5e-05, "loss": 5.4754, "step": 680 }, { "epoch": 2.3939183968479938e-15, "grad_norm": 2.236020565032959, "learning_rate": 8.625e-05, "loss": 5.5287, "step": 690 }, { "epoch": 2.42861286636753e-15, "grad_norm": 1.7151645421981812, "learning_rate": 8.75e-05, "loss": 5.4679, "step": 700 }, { "epoch": 2.463307335887066e-15, "grad_norm": 2.30198335647583, "learning_rate": 8.875e-05, "loss": 5.4288, "step": 710 }, { "epoch": 2.4980018054066022e-15, "grad_norm": 2.2258923053741455, "learning_rate": 8.999999999999999e-05, "loss": 5.4713, "step": 720 }, { "epoch": 2.5326962749261384e-15, "grad_norm": 1.9565815925598145, "learning_rate": 9.125e-05, "loss": 5.4248, "step": 730 }, { "epoch": 2.5673907444456745e-15, "grad_norm": 2.261991500854492, "learning_rate": 9.25e-05, "loss": 5.3978, "step": 740 }, { "epoch": 2.6020852139652106e-15, "grad_norm": 2.2596771717071533, "learning_rate": 9.375e-05, "loss": 5.3716, "step": 750 }, { "epoch": 2.6367796834847468e-15, "grad_norm": 2.3100645542144775, "learning_rate": 9.5e-05, "loss": 5.3151, "step": 760 }, { "epoch": 2.671474153004283e-15, "grad_norm": 1.5338190793991089, "learning_rate": 9.625000000000001e-05, "loss": 5.3351, "step": 770 }, { "epoch": 2.706168622523819e-15, "grad_norm": 1.9649896621704102, "learning_rate": 9.750000000000001e-05, "loss": 5.3045, "step": 780 }, { "epoch": 2.7408630920433552e-15, "grad_norm": 1.8696657419204712, "learning_rate": 9.875e-05, "loss": 5.2856, "step": 790 }, { "epoch": 2.7755575615628914e-15, "grad_norm": 1.5760527849197388, "learning_rate": 0.0001, "loss": 5.258, "step": 800 }, { "epoch": 2.8102520310824275e-15, "grad_norm": 2.010075092315674, "learning_rate": 0.00010125000000000001, "loss": 5.2447, "step": 810 }, { "epoch": 2.8449465006019636e-15, "grad_norm": 2.556671619415283, "learning_rate": 0.0001025, "loss": 5.2054, "step": 820 }, { "epoch": 2.8796409701214998e-15, "grad_norm": 1.9800317287445068, "learning_rate": 0.00010375, "loss": 5.1718, "step": 830 }, { "epoch": 2.914335439641036e-15, "grad_norm": 2.0150325298309326, "learning_rate": 0.000105, "loss": 5.1584, "step": 840 }, { "epoch": 2.949029909160572e-15, "grad_norm": 1.6884160041809082, "learning_rate": 0.00010625, "loss": 5.1606, "step": 850 }, { "epoch": 2.9837243786801082e-15, "grad_norm": 2.1996312141418457, "learning_rate": 0.0001075, "loss": 5.1728, "step": 860 }, { "epoch": 3.0184188481996443e-15, "grad_norm": 1.9362671375274658, "learning_rate": 0.00010875, "loss": 5.0865, "step": 870 }, { "epoch": 3.0531133177191805e-15, "grad_norm": 1.6913535594940186, "learning_rate": 0.00011, "loss": 5.0838, "step": 880 }, { "epoch": 3.0878077872387166e-15, "grad_norm": 2.0805556774139404, "learning_rate": 0.00011125000000000001, "loss": 5.0846, "step": 890 }, { "epoch": 3.1225022567582528e-15, "grad_norm": 1.6517592668533325, "learning_rate": 0.00011250000000000001, "loss": 5.017, "step": 900 }, { "epoch": 3.157196726277789e-15, "grad_norm": 1.830851435661316, "learning_rate": 0.00011375, "loss": 5.0122, "step": 910 }, { "epoch": 3.191891195797325e-15, "grad_norm": 1.8484851121902466, "learning_rate": 0.000115, "loss": 5.049, "step": 920 }, { "epoch": 3.226585665316861e-15, "grad_norm": 2.1999258995056152, "learning_rate": 0.00011625, "loss": 5.001, "step": 930 }, { "epoch": 3.2612801348363973e-15, "grad_norm": 1.4078352451324463, "learning_rate": 0.0001175, "loss": 4.9722, "step": 940 }, { "epoch": 3.2959746043559335e-15, "grad_norm": 2.032158613204956, "learning_rate": 0.00011875, "loss": 4.9543, "step": 950 }, { "epoch": 3.3306690738754696e-15, "grad_norm": 1.7583720684051514, "learning_rate": 0.00012, "loss": 4.9433, "step": 960 }, { "epoch": 3.3653635433950058e-15, "grad_norm": 1.423781394958496, "learning_rate": 0.00012124999999999999, "loss": 4.9037, "step": 970 }, { "epoch": 3.400058012914542e-15, "grad_norm": 1.9152929782867432, "learning_rate": 0.0001225, "loss": 4.8994, "step": 980 }, { "epoch": 3.434752482434078e-15, "grad_norm": 1.3977447748184204, "learning_rate": 0.00012375, "loss": 4.9067, "step": 990 }, { "epoch": 3.469446951953614e-15, "grad_norm": 1.7368465662002563, "learning_rate": 0.000125, "loss": 4.8658, "step": 1000 }, { "epoch": 3.5041414214731503e-15, "grad_norm": 1.4774492979049683, "learning_rate": 0.00012499999555227336, "loss": 4.8578, "step": 1010 }, { "epoch": 3.5388358909926865e-15, "grad_norm": 1.5713034868240356, "learning_rate": 0.0001249999822090941, "loss": 4.7922, "step": 1020 }, { "epoch": 3.573530360512223e-15, "grad_norm": 1.3736143112182617, "learning_rate": 0.00012499995997046438, "loss": 4.7758, "step": 1030 }, { "epoch": 3.608224830031759e-15, "grad_norm": 1.2564338445663452, "learning_rate": 0.0001249999288363877, "loss": 4.7618, "step": 1040 }, { "epoch": 3.642919299551295e-15, "grad_norm": 1.7350471019744873, "learning_rate": 0.00012499988880686897, "loss": 4.7535, "step": 1050 }, { "epoch": 3.677613769070831e-15, "grad_norm": 1.4064443111419678, "learning_rate": 0.00012499983988191452, "loss": 4.72, "step": 1060 }, { "epoch": 3.712308238590367e-15, "grad_norm": 1.518574595451355, "learning_rate": 0.00012499978206153208, "loss": 4.7201, "step": 1070 }, { "epoch": 3.747002708109903e-15, "grad_norm": 1.4150875806808472, "learning_rate": 0.00012499971534573085, "loss": 4.6716, "step": 1080 }, { "epoch": 3.7816971776294395e-15, "grad_norm": 1.9536768198013306, "learning_rate": 0.0001249996397345213, "loss": 4.7172, "step": 1090 }, { "epoch": 3.816391647148976e-15, "grad_norm": 1.393886923789978, "learning_rate": 0.00012499955522791544, "loss": 4.6288, "step": 1100 }, { "epoch": 3.851086116668512e-15, "grad_norm": 1.855498194694519, "learning_rate": 0.00012499946182592664, "loss": 4.655, "step": 1110 }, { "epoch": 3.885780586188048e-15, "grad_norm": 1.303644061088562, "learning_rate": 0.00012499935952856964, "loss": 4.6732, "step": 1120 }, { "epoch": 3.920475055707584e-15, "grad_norm": 1.4335750341415405, "learning_rate": 0.0001249992483358606, "loss": 4.6339, "step": 1130 }, { "epoch": 3.95516952522712e-15, "grad_norm": 1.4705201387405396, "learning_rate": 0.00012499912824781717, "loss": 4.6139, "step": 1140 }, { "epoch": 3.989863994746656e-15, "grad_norm": 1.8620712757110596, "learning_rate": 0.00012499899926445826, "loss": 4.6399, "step": 1150 }, { "epoch": 4.0245584642661925e-15, "grad_norm": 1.787693977355957, "learning_rate": 0.00012499886138580434, "loss": 4.5913, "step": 1160 }, { "epoch": 4.0592529337857286e-15, "grad_norm": 1.5172003507614136, "learning_rate": 0.0001249987146118772, "loss": 4.5397, "step": 1170 }, { "epoch": 4.093947403305265e-15, "grad_norm": 1.3738051652908325, "learning_rate": 0.0001249985589427, "loss": 4.5184, "step": 1180 }, { "epoch": 4.128641872824801e-15, "grad_norm": 1.4118515253067017, "learning_rate": 0.00012499839437829743, "loss": 4.5054, "step": 1190 }, { "epoch": 4.163336342344337e-15, "grad_norm": 1.4125251770019531, "learning_rate": 0.00012499822091869547, "loss": 4.5171, "step": 1200 }, { "epoch": 4.198030811863873e-15, "grad_norm": 1.3282525539398193, "learning_rate": 0.00012499803856392153, "loss": 4.503, "step": 1210 }, { "epoch": 4.232725281383409e-15, "grad_norm": 1.2331408262252808, "learning_rate": 0.0001249978473140045, "loss": 4.495, "step": 1220 }, { "epoch": 4.2674197509029455e-15, "grad_norm": 1.5739362239837646, "learning_rate": 0.00012499764716897462, "loss": 4.4237, "step": 1230 }, { "epoch": 4.3021142204224816e-15, "grad_norm": 1.2533175945281982, "learning_rate": 0.00012499743812886352, "loss": 4.4466, "step": 1240 }, { "epoch": 4.336808689942018e-15, "grad_norm": 1.3963618278503418, "learning_rate": 0.00012499722019370424, "loss": 4.4651, "step": 1250 }, { "epoch": 4.371503159461554e-15, "grad_norm": 1.3796100616455078, "learning_rate": 0.00012499699336353126, "loss": 4.4456, "step": 1260 }, { "epoch": 4.40619762898109e-15, "grad_norm": 1.2395864725112915, "learning_rate": 0.0001249967576383805, "loss": 4.396, "step": 1270 }, { "epoch": 4.440892098500626e-15, "grad_norm": 1.3255988359451294, "learning_rate": 0.00012499651301828914, "loss": 4.3895, "step": 1280 }, { "epoch": 4.475586568020162e-15, "grad_norm": 1.2580350637435913, "learning_rate": 0.00012499625950329593, "loss": 4.3952, "step": 1290 }, { "epoch": 4.5102810375396984e-15, "grad_norm": 1.7051551342010498, "learning_rate": 0.00012499599709344097, "loss": 4.4146, "step": 1300 }, { "epoch": 4.5449755070592346e-15, "grad_norm": 1.6887810230255127, "learning_rate": 0.00012499572578876574, "loss": 4.3906, "step": 1310 }, { "epoch": 4.579669976578771e-15, "grad_norm": 1.3832134008407593, "learning_rate": 0.00012499544558931315, "loss": 4.3235, "step": 1320 }, { "epoch": 4.614364446098307e-15, "grad_norm": 1.2897875308990479, "learning_rate": 0.0001249951564951275, "loss": 4.3326, "step": 1330 }, { "epoch": 4.649058915617843e-15, "grad_norm": 1.369277834892273, "learning_rate": 0.00012499485850625447, "loss": 4.3606, "step": 1340 }, { "epoch": 4.683753385137379e-15, "grad_norm": 1.3328986167907715, "learning_rate": 0.00012499455162274126, "loss": 4.3017, "step": 1350 }, { "epoch": 4.718447854656915e-15, "grad_norm": 1.3932201862335205, "learning_rate": 0.0001249942358446364, "loss": 4.2776, "step": 1360 }, { "epoch": 4.7531423241764514e-15, "grad_norm": 1.3236936330795288, "learning_rate": 0.00012499391117198975, "loss": 4.2895, "step": 1370 }, { "epoch": 4.7878367936959876e-15, "grad_norm": 1.2384673357009888, "learning_rate": 0.0001249935776048527, "loss": 4.2817, "step": 1380 }, { "epoch": 4.822531263215524e-15, "grad_norm": 1.3051435947418213, "learning_rate": 0.000124993235143278, "loss": 4.285, "step": 1390 }, { "epoch": 4.85722573273506e-15, "grad_norm": 1.1847801208496094, "learning_rate": 0.00012499288378731982, "loss": 4.2601, "step": 1400 }, { "epoch": 4.891920202254596e-15, "grad_norm": 1.327549695968628, "learning_rate": 0.0001249925235370337, "loss": 4.2651, "step": 1410 }, { "epoch": 4.926614671774132e-15, "grad_norm": 1.2897961139678955, "learning_rate": 0.00012499215439247665, "loss": 4.2658, "step": 1420 }, { "epoch": 4.961309141293668e-15, "grad_norm": 1.382060170173645, "learning_rate": 0.00012499177635370699, "loss": 4.2425, "step": 1430 }, { "epoch": 4.9960036108132044e-15, "grad_norm": 1.1921894550323486, "learning_rate": 0.00012499138942078454, "loss": 4.2182, "step": 1440 }, { "epoch": 5.0306980803327406e-15, "grad_norm": 1.1907954216003418, "learning_rate": 0.0001249909935937705, "loss": 4.1612, "step": 1450 }, { "epoch": 5.065392549852277e-15, "grad_norm": 1.176125407218933, "learning_rate": 0.00012499058887272742, "loss": 4.1921, "step": 1460 }, { "epoch": 5.100087019371813e-15, "grad_norm": 1.2113888263702393, "learning_rate": 0.00012499017525771934, "loss": 4.1782, "step": 1470 }, { "epoch": 5.134781488891349e-15, "grad_norm": 1.2233693599700928, "learning_rate": 0.00012498975274881165, "loss": 4.182, "step": 1480 }, { "epoch": 5.169475958410885e-15, "grad_norm": 1.2346550226211548, "learning_rate": 0.0001249893213460712, "loss": 4.1604, "step": 1490 }, { "epoch": 5.204170427930421e-15, "grad_norm": 1.2451103925704956, "learning_rate": 0.00012498888104956618, "loss": 4.1485, "step": 1500 }, { "epoch": 5.2388648974499574e-15, "grad_norm": 1.2079665660858154, "learning_rate": 0.00012498843185936625, "loss": 4.174, "step": 1510 }, { "epoch": 5.2735593669694936e-15, "grad_norm": 1.2002967596054077, "learning_rate": 0.0001249879737755424, "loss": 4.1516, "step": 1520 }, { "epoch": 5.30825383648903e-15, "grad_norm": 1.4664313793182373, "learning_rate": 0.00012498750679816711, "loss": 4.1264, "step": 1530 }, { "epoch": 5.342948306008566e-15, "grad_norm": 1.0170352458953857, "learning_rate": 0.0001249870309273142, "loss": 4.1262, "step": 1540 }, { "epoch": 5.377642775528102e-15, "grad_norm": 1.2074439525604248, "learning_rate": 0.00012498654616305895, "loss": 4.091, "step": 1550 }, { "epoch": 5.412337245047638e-15, "grad_norm": 1.1782931089401245, "learning_rate": 0.00012498605250547802, "loss": 4.1052, "step": 1560 }, { "epoch": 5.447031714567174e-15, "grad_norm": 1.0114202499389648, "learning_rate": 0.00012498554995464948, "loss": 4.0755, "step": 1570 }, { "epoch": 5.4817261840867104e-15, "grad_norm": 1.287811279296875, "learning_rate": 0.00012498503851065277, "loss": 4.0476, "step": 1580 }, { "epoch": 5.5164206536062466e-15, "grad_norm": 1.2855843305587769, "learning_rate": 0.0001249845181735688, "loss": 4.0412, "step": 1590 }, { "epoch": 5.551115123125783e-15, "grad_norm": 1.090482234954834, "learning_rate": 0.00012498398894347985, "loss": 4.0784, "step": 1600 }, { "epoch": 5.585809592645319e-15, "grad_norm": 1.0627268552780151, "learning_rate": 0.0001249834508204696, "loss": 4.0521, "step": 1610 }, { "epoch": 5.620504062164855e-15, "grad_norm": 1.113208293914795, "learning_rate": 0.0001249829038046232, "loss": 4.0331, "step": 1620 }, { "epoch": 5.655198531684391e-15, "grad_norm": 1.1350445747375488, "learning_rate": 0.0001249823478960271, "loss": 3.9818, "step": 1630 }, { "epoch": 5.689893001203927e-15, "grad_norm": 1.36904776096344, "learning_rate": 0.00012498178309476923, "loss": 3.9643, "step": 1640 }, { "epoch": 5.7245874707234634e-15, "grad_norm": 1.180347204208374, "learning_rate": 0.00012498120940093892, "loss": 4.0249, "step": 1650 }, { "epoch": 5.7592819402429996e-15, "grad_norm": 0.9944842457771301, "learning_rate": 0.00012498062681462685, "loss": 4.0005, "step": 1660 }, { "epoch": 5.793976409762536e-15, "grad_norm": 1.0845718383789062, "learning_rate": 0.00012498003533592523, "loss": 3.9519, "step": 1670 }, { "epoch": 5.828670879282072e-15, "grad_norm": 0.9461311101913452, "learning_rate": 0.00012497943496492754, "loss": 4.0007, "step": 1680 }, { "epoch": 5.863365348801608e-15, "grad_norm": 1.4032315015792847, "learning_rate": 0.00012497882570172873, "loss": 3.968, "step": 1690 }, { "epoch": 5.898059818321144e-15, "grad_norm": 1.22874915599823, "learning_rate": 0.00012497820754642515, "loss": 3.9251, "step": 1700 }, { "epoch": 5.93275428784068e-15, "grad_norm": 1.0901143550872803, "learning_rate": 0.00012497758049911456, "loss": 3.9381, "step": 1710 }, { "epoch": 5.9674487573602164e-15, "grad_norm": 0.9730834364891052, "learning_rate": 0.00012497694455989615, "loss": 3.9167, "step": 1720 }, { "epoch": 6.0021432268797525e-15, "grad_norm": 0.9602221846580505, "learning_rate": 0.00012497629972887044, "loss": 3.9218, "step": 1730 }, { "epoch": 6.036837696399289e-15, "grad_norm": 1.2867379188537598, "learning_rate": 0.00012497564600613944, "loss": 3.9117, "step": 1740 }, { "epoch": 6.071532165918825e-15, "grad_norm": 1.098429560661316, "learning_rate": 0.00012497498339180652, "loss": 3.8809, "step": 1750 }, { "epoch": 6.106226635438361e-15, "grad_norm": 1.2760214805603027, "learning_rate": 0.00012497431188597646, "loss": 3.8981, "step": 1760 }, { "epoch": 6.140921104957897e-15, "grad_norm": 1.1986569166183472, "learning_rate": 0.00012497363148875545, "loss": 3.8554, "step": 1770 }, { "epoch": 6.175615574477433e-15, "grad_norm": 1.2420878410339355, "learning_rate": 0.00012497294220025108, "loss": 3.8573, "step": 1780 }, { "epoch": 6.2103100439969694e-15, "grad_norm": 1.204250454902649, "learning_rate": 0.0001249722440205724, "loss": 3.8267, "step": 1790 }, { "epoch": 6.2450045135165055e-15, "grad_norm": 1.0219510793685913, "learning_rate": 0.0001249715369498298, "loss": 3.7965, "step": 1800 }, { "epoch": 6.279698983036042e-15, "grad_norm": 1.0054283142089844, "learning_rate": 0.00012497082098813505, "loss": 3.8242, "step": 1810 }, { "epoch": 6.314393452555578e-15, "grad_norm": 1.1228842735290527, "learning_rate": 0.00012497009613560145, "loss": 3.8334, "step": 1820 }, { "epoch": 6.349087922075114e-15, "grad_norm": 0.9138529300689697, "learning_rate": 0.00012496936239234357, "loss": 3.7868, "step": 1830 }, { "epoch": 6.38378239159465e-15, "grad_norm": 0.9028067588806152, "learning_rate": 0.0001249686197584775, "loss": 3.7481, "step": 1840 }, { "epoch": 6.418476861114186e-15, "grad_norm": 1.0702089071273804, "learning_rate": 0.0001249678682341206, "loss": 3.7874, "step": 1850 }, { "epoch": 6.453171330633722e-15, "grad_norm": 0.9645017385482788, "learning_rate": 0.0001249671078193918, "loss": 3.774, "step": 1860 }, { "epoch": 6.4878658001532585e-15, "grad_norm": 1.1562224626541138, "learning_rate": 0.0001249663385144113, "loss": 3.7205, "step": 1870 }, { "epoch": 6.522560269672795e-15, "grad_norm": 0.9911544322967529, "learning_rate": 0.0001249655603193008, "loss": 3.759, "step": 1880 }, { "epoch": 6.557254739192331e-15, "grad_norm": 0.8554500341415405, "learning_rate": 0.00012496477323418334, "loss": 3.7598, "step": 1890 }, { "epoch": 6.591949208711867e-15, "grad_norm": 1.3611812591552734, "learning_rate": 0.00012496397725918337, "loss": 3.7585, "step": 1900 }, { "epoch": 6.626643678231403e-15, "grad_norm": 1.0532523393630981, "learning_rate": 0.0001249631723944268, "loss": 3.7138, "step": 1910 }, { "epoch": 6.661338147750939e-15, "grad_norm": 1.048390507698059, "learning_rate": 0.0001249623586400409, "loss": 3.7379, "step": 1920 }, { "epoch": 6.696032617270475e-15, "grad_norm": 0.9820449948310852, "learning_rate": 0.00012496153599615435, "loss": 3.7021, "step": 1930 }, { "epoch": 6.7307270867900115e-15, "grad_norm": 0.9594332575798035, "learning_rate": 0.00012496070446289727, "loss": 3.697, "step": 1940 }, { "epoch": 6.765421556309548e-15, "grad_norm": 1.1087936162948608, "learning_rate": 0.00012495986404040117, "loss": 3.6939, "step": 1950 }, { "epoch": 6.800116025829084e-15, "grad_norm": 0.9202387928962708, "learning_rate": 0.0001249590147287989, "loss": 3.6615, "step": 1960 }, { "epoch": 6.83481049534862e-15, "grad_norm": 1.1009349822998047, "learning_rate": 0.0001249581565282248, "loss": 3.6478, "step": 1970 }, { "epoch": 6.869504964868156e-15, "grad_norm": 0.8833752870559692, "learning_rate": 0.00012495728943881458, "loss": 3.6564, "step": 1980 }, { "epoch": 6.904199434387692e-15, "grad_norm": 1.0989856719970703, "learning_rate": 0.00012495641346070538, "loss": 3.6285, "step": 1990 }, { "epoch": 6.938893903907228e-15, "grad_norm": 1.0148354768753052, "learning_rate": 0.00012495552859403573, "loss": 3.612, "step": 2000 }, { "epoch": 6.9735883734267645e-15, "grad_norm": 0.8948827385902405, "learning_rate": 0.00012495463483894556, "loss": 3.6257, "step": 2010 }, { "epoch": 7.008282842946301e-15, "grad_norm": 1.0413926839828491, "learning_rate": 0.0001249537321955762, "loss": 3.6526, "step": 2020 }, { "epoch": 7.042977312465837e-15, "grad_norm": 0.9643189311027527, "learning_rate": 0.00012495282066407037, "loss": 3.6171, "step": 2030 }, { "epoch": 7.077671781985373e-15, "grad_norm": 0.9083977341651917, "learning_rate": 0.00012495190024457226, "loss": 3.5783, "step": 2040 }, { "epoch": 7.112366251504909e-15, "grad_norm": 1.0824311971664429, "learning_rate": 0.00012495097093722743, "loss": 3.6228, "step": 2050 }, { "epoch": 7.147060721024445e-15, "grad_norm": 0.972273588180542, "learning_rate": 0.00012495003274218283, "loss": 3.6096, "step": 2060 }, { "epoch": 7.181755190543981e-15, "grad_norm": 0.958733320236206, "learning_rate": 0.0001249490856595868, "loss": 3.5518, "step": 2070 }, { "epoch": 7.216449660063518e-15, "grad_norm": 0.9249112010002136, "learning_rate": 0.00012494812968958917, "loss": 3.5677, "step": 2080 }, { "epoch": 7.251144129583054e-15, "grad_norm": 1.052067756652832, "learning_rate": 0.00012494716483234106, "loss": 3.5801, "step": 2090 }, { "epoch": 7.28583859910259e-15, "grad_norm": 0.9500466585159302, "learning_rate": 0.00012494619108799508, "loss": 3.5742, "step": 2100 }, { "epoch": 7.320533068622126e-15, "grad_norm": 0.888433039188385, "learning_rate": 0.00012494520845670522, "loss": 3.5483, "step": 2110 }, { "epoch": 7.355227538141662e-15, "grad_norm": 0.9463629722595215, "learning_rate": 0.0001249442169386269, "loss": 3.5998, "step": 2120 }, { "epoch": 7.389922007661198e-15, "grad_norm": 0.9361791014671326, "learning_rate": 0.00012494321653391688, "loss": 3.5482, "step": 2130 }, { "epoch": 7.424616477180734e-15, "grad_norm": 0.9349877834320068, "learning_rate": 0.00012494220724273337, "loss": 3.5663, "step": 2140 }, { "epoch": 7.45931094670027e-15, "grad_norm": 0.9149386882781982, "learning_rate": 0.00012494118906523598, "loss": 3.5403, "step": 2150 }, { "epoch": 7.494005416219807e-15, "grad_norm": 1.0850200653076172, "learning_rate": 0.00012494016200158576, "loss": 3.5317, "step": 2160 }, { "epoch": 7.528699885739343e-15, "grad_norm": 0.991774320602417, "learning_rate": 0.0001249391260519451, "loss": 3.5255, "step": 2170 }, { "epoch": 7.563394355258879e-15, "grad_norm": 1.0342744588851929, "learning_rate": 0.00012493808121647785, "loss": 3.5157, "step": 2180 }, { "epoch": 7.598088824778415e-15, "grad_norm": 0.9991723895072937, "learning_rate": 0.00012493702749534922, "loss": 3.5123, "step": 2190 }, { "epoch": 7.632783294297951e-15, "grad_norm": 0.9972164034843445, "learning_rate": 0.0001249359648887258, "loss": 3.4793, "step": 2200 }, { "epoch": 7.667477763817487e-15, "grad_norm": 0.8239566087722778, "learning_rate": 0.00012493489339677577, "loss": 3.5202, "step": 2210 }, { "epoch": 7.702172233337023e-15, "grad_norm": 0.8215254545211792, "learning_rate": 0.00012493381301966844, "loss": 3.4924, "step": 2220 }, { "epoch": 7.73686670285656e-15, "grad_norm": 1.2948344945907593, "learning_rate": 0.00012493272375757472, "loss": 3.5306, "step": 2230 }, { "epoch": 7.771561172376096e-15, "grad_norm": 0.9287869930267334, "learning_rate": 0.00012493162561066687, "loss": 3.5125, "step": 2240 }, { "epoch": 7.806255641895632e-15, "grad_norm": 0.9031260013580322, "learning_rate": 0.00012493051857911854, "loss": 3.4875, "step": 2250 }, { "epoch": 7.840950111415168e-15, "grad_norm": 0.9185664653778076, "learning_rate": 0.00012492940266310481, "loss": 3.4955, "step": 2260 }, { "epoch": 7.875644580934704e-15, "grad_norm": 0.8454828858375549, "learning_rate": 0.00012492827786280214, "loss": 3.4473, "step": 2270 }, { "epoch": 7.91033905045424e-15, "grad_norm": 0.9080525636672974, "learning_rate": 0.0001249271441783884, "loss": 3.4786, "step": 2280 }, { "epoch": 7.945033519973776e-15, "grad_norm": 0.9131040573120117, "learning_rate": 0.0001249260016100429, "loss": 3.4458, "step": 2290 }, { "epoch": 7.979727989493313e-15, "grad_norm": 0.9473633170127869, "learning_rate": 0.00012492485015794632, "loss": 3.4098, "step": 2300 }, { "epoch": 8.014422459012849e-15, "grad_norm": 0.9682335257530212, "learning_rate": 0.00012492368982228071, "loss": 3.4409, "step": 2310 }, { "epoch": 8.049116928532385e-15, "grad_norm": 0.8564648628234863, "learning_rate": 0.00012492252060322966, "loss": 3.4205, "step": 2320 }, { "epoch": 8.083811398051921e-15, "grad_norm": 0.9404218196868896, "learning_rate": 0.00012492134250097794, "loss": 3.468, "step": 2330 }, { "epoch": 8.118505867571457e-15, "grad_norm": 0.8040472865104675, "learning_rate": 0.00012492015551571198, "loss": 3.4182, "step": 2340 }, { "epoch": 8.153200337090993e-15, "grad_norm": 0.9431834816932678, "learning_rate": 0.0001249189596476194, "loss": 3.4193, "step": 2350 }, { "epoch": 8.18789480661053e-15, "grad_norm": 0.8354449272155762, "learning_rate": 0.0001249177548968894, "loss": 3.452, "step": 2360 }, { "epoch": 8.222589276130066e-15, "grad_norm": 0.9927154183387756, "learning_rate": 0.00012491654126371246, "loss": 3.4074, "step": 2370 }, { "epoch": 8.257283745649602e-15, "grad_norm": 0.866419792175293, "learning_rate": 0.00012491531874828048, "loss": 3.3784, "step": 2380 }, { "epoch": 8.291978215169138e-15, "grad_norm": 0.864596962928772, "learning_rate": 0.00012491408735078682, "loss": 3.4237, "step": 2390 }, { "epoch": 8.326672684688674e-15, "grad_norm": 0.8692665696144104, "learning_rate": 0.0001249128470714262, "loss": 3.3939, "step": 2400 }, { "epoch": 8.36136715420821e-15, "grad_norm": 0.7357878088951111, "learning_rate": 0.0001249115979103948, "loss": 3.3764, "step": 2410 }, { "epoch": 8.396061623727746e-15, "grad_norm": 0.7842319011688232, "learning_rate": 0.0001249103398678901, "loss": 3.3997, "step": 2420 }, { "epoch": 8.430756093247282e-15, "grad_norm": 0.8886202573776245, "learning_rate": 0.00012490907294411107, "loss": 3.365, "step": 2430 }, { "epoch": 8.465450562766819e-15, "grad_norm": 0.9134165048599243, "learning_rate": 0.00012490779713925811, "loss": 3.3988, "step": 2440 }, { "epoch": 8.500145032286355e-15, "grad_norm": 0.8116360306739807, "learning_rate": 0.00012490651245353294, "loss": 3.3715, "step": 2450 }, { "epoch": 8.534839501805891e-15, "grad_norm": 0.9030981659889221, "learning_rate": 0.00012490521888713872, "loss": 3.3979, "step": 2460 }, { "epoch": 8.569533971325427e-15, "grad_norm": 0.8228681087493896, "learning_rate": 0.00012490391644028, "loss": 3.3764, "step": 2470 }, { "epoch": 8.604228440844963e-15, "grad_norm": 0.8020610809326172, "learning_rate": 0.0001249026051131628, "loss": 3.3886, "step": 2480 }, { "epoch": 8.6389229103645e-15, "grad_norm": 1.0466738939285278, "learning_rate": 0.00012490128490599443, "loss": 3.3997, "step": 2490 }, { "epoch": 8.673617379884035e-15, "grad_norm": 0.9045177102088928, "learning_rate": 0.00012489995581898373, "loss": 3.333, "step": 2500 }, { "epoch": 8.708311849403572e-15, "grad_norm": 0.8391941785812378, "learning_rate": 0.00012489861785234085, "loss": 3.3264, "step": 2510 }, { "epoch": 8.743006318923108e-15, "grad_norm": 0.8131318688392639, "learning_rate": 0.0001248972710062774, "loss": 3.3349, "step": 2520 }, { "epoch": 8.777700788442644e-15, "grad_norm": 0.8956968784332275, "learning_rate": 0.00012489591528100634, "loss": 3.3765, "step": 2530 }, { "epoch": 8.81239525796218e-15, "grad_norm": 0.8356823325157166, "learning_rate": 0.00012489455067674212, "loss": 3.3303, "step": 2540 }, { "epoch": 8.847089727481716e-15, "grad_norm": 0.7715813517570496, "learning_rate": 0.0001248931771937005, "loss": 3.3419, "step": 2550 }, { "epoch": 8.881784197001252e-15, "grad_norm": 0.8786708116531372, "learning_rate": 0.00012489179483209865, "loss": 3.3086, "step": 2560 }, { "epoch": 8.916478666520788e-15, "grad_norm": 0.9560003280639648, "learning_rate": 0.00012489040359215525, "loss": 3.3382, "step": 2570 }, { "epoch": 8.951173136040325e-15, "grad_norm": 0.8346591591835022, "learning_rate": 0.00012488900347409028, "loss": 3.3606, "step": 2580 }, { "epoch": 8.98586760555986e-15, "grad_norm": 0.9229335784912109, "learning_rate": 0.00012488759447812513, "loss": 3.3199, "step": 2590 }, { "epoch": 9.020562075079397e-15, "grad_norm": 0.819379985332489, "learning_rate": 0.0001248861766044827, "loss": 3.2917, "step": 2600 }, { "epoch": 9.055256544598933e-15, "grad_norm": 0.9512423276901245, "learning_rate": 0.00012488474985338714, "loss": 3.3208, "step": 2610 }, { "epoch": 9.089951014118469e-15, "grad_norm": 0.7904150485992432, "learning_rate": 0.0001248833142250641, "loss": 3.2871, "step": 2620 }, { "epoch": 9.124645483638005e-15, "grad_norm": 0.8217881321907043, "learning_rate": 0.0001248818697197406, "loss": 3.3407, "step": 2630 }, { "epoch": 9.159339953157541e-15, "grad_norm": 0.8349485397338867, "learning_rate": 0.00012488041633764512, "loss": 3.3241, "step": 2640 }, { "epoch": 9.194034422677078e-15, "grad_norm": 0.8650612831115723, "learning_rate": 0.00012487895407900745, "loss": 3.2836, "step": 2650 }, { "epoch": 9.228728892196614e-15, "grad_norm": 0.8204360604286194, "learning_rate": 0.00012487748294405884, "loss": 3.2879, "step": 2660 }, { "epoch": 9.26342336171615e-15, "grad_norm": 0.7803131341934204, "learning_rate": 0.000124876002933032, "loss": 3.2708, "step": 2670 }, { "epoch": 9.298117831235686e-15, "grad_norm": 0.7937780022621155, "learning_rate": 0.0001248745140461609, "loss": 3.2979, "step": 2680 }, { "epoch": 9.332812300755222e-15, "grad_norm": 0.786794126033783, "learning_rate": 0.000124873016283681, "loss": 3.3095, "step": 2690 }, { "epoch": 9.367506770274758e-15, "grad_norm": 0.9007452726364136, "learning_rate": 0.0001248715096458292, "loss": 3.2877, "step": 2700 }, { "epoch": 9.402201239794294e-15, "grad_norm": 0.8332339525222778, "learning_rate": 0.00012486999413284378, "loss": 3.262, "step": 2710 }, { "epoch": 9.43689570931383e-15, "grad_norm": 0.8420783877372742, "learning_rate": 0.00012486846974496435, "loss": 3.2808, "step": 2720 }, { "epoch": 9.471590178833367e-15, "grad_norm": 0.7920785546302795, "learning_rate": 0.000124866936482432, "loss": 3.2958, "step": 2730 }, { "epoch": 9.506284648352903e-15, "grad_norm": 0.7988773584365845, "learning_rate": 0.0001248653943454892, "loss": 3.2738, "step": 2740 }, { "epoch": 9.540979117872439e-15, "grad_norm": 0.8747711777687073, "learning_rate": 0.00012486384333437986, "loss": 3.268, "step": 2750 }, { "epoch": 9.575673587391975e-15, "grad_norm": 0.6877643465995789, "learning_rate": 0.0001248622834493492, "loss": 3.2286, "step": 2760 }, { "epoch": 9.610368056911511e-15, "grad_norm": 0.726291835308075, "learning_rate": 0.00012486071469064393, "loss": 3.2957, "step": 2770 }, { "epoch": 9.645062526431047e-15, "grad_norm": 0.7729490399360657, "learning_rate": 0.00012485913705851211, "loss": 3.2905, "step": 2780 }, { "epoch": 9.679756995950584e-15, "grad_norm": 0.7480143904685974, "learning_rate": 0.00012485755055320332, "loss": 3.2628, "step": 2790 }, { "epoch": 9.71445146547012e-15, "grad_norm": 0.8002223372459412, "learning_rate": 0.00012485595517496833, "loss": 3.215, "step": 2800 }, { "epoch": 9.749145934989656e-15, "grad_norm": 0.8052833080291748, "learning_rate": 0.0001248543509240595, "loss": 3.212, "step": 2810 }, { "epoch": 9.783840404509192e-15, "grad_norm": 0.7762216329574585, "learning_rate": 0.00012485273780073056, "loss": 3.2522, "step": 2820 }, { "epoch": 9.818534874028728e-15, "grad_norm": 0.8273215293884277, "learning_rate": 0.00012485111580523654, "loss": 3.2747, "step": 2830 }, { "epoch": 9.853229343548264e-15, "grad_norm": 0.7497468590736389, "learning_rate": 0.00012484948493783399, "loss": 3.2257, "step": 2840 }, { "epoch": 9.8879238130678e-15, "grad_norm": 0.8332861065864563, "learning_rate": 0.0001248478451987808, "loss": 3.2171, "step": 2850 }, { "epoch": 9.922618282587337e-15, "grad_norm": 0.7868185639381409, "learning_rate": 0.00012484619658833628, "loss": 3.2377, "step": 2860 }, { "epoch": 9.957312752106873e-15, "grad_norm": 0.8009334802627563, "learning_rate": 0.00012484453910676114, "loss": 3.1884, "step": 2870 }, { "epoch": 9.992007221626409e-15, "grad_norm": 0.7414532899856567, "learning_rate": 0.00012484287275431753, "loss": 3.2756, "step": 2880 }, { "epoch": 1.0026701691145945e-14, "grad_norm": 0.7201011180877686, "learning_rate": 0.00012484119753126892, "loss": 3.219, "step": 2890 }, { "epoch": 1.0061396160665481e-14, "grad_norm": 0.8258967995643616, "learning_rate": 0.00012483951343788028, "loss": 3.2697, "step": 2900 }, { "epoch": 1.0096090630185017e-14, "grad_norm": 0.8723838925361633, "learning_rate": 0.0001248378204744179, "loss": 3.2168, "step": 2910 }, { "epoch": 1.0130785099704553e-14, "grad_norm": 0.8283339738845825, "learning_rate": 0.00012483611864114953, "loss": 3.258, "step": 2920 }, { "epoch": 1.016547956922409e-14, "grad_norm": 0.8674870133399963, "learning_rate": 0.00012483440793834428, "loss": 3.1909, "step": 2930 }, { "epoch": 1.0200174038743626e-14, "grad_norm": 0.8387137055397034, "learning_rate": 0.0001248326883662727, "loss": 3.1885, "step": 2940 }, { "epoch": 1.0234868508263162e-14, "grad_norm": 0.7276076078414917, "learning_rate": 0.0001248309599252067, "loss": 3.1412, "step": 2950 }, { "epoch": 1.0269562977782698e-14, "grad_norm": 0.8058744668960571, "learning_rate": 0.00012482922261541966, "loss": 3.2163, "step": 2960 }, { "epoch": 1.0304257447302234e-14, "grad_norm": 0.8478225469589233, "learning_rate": 0.00012482747643718627, "loss": 3.1966, "step": 2970 }, { "epoch": 1.033895191682177e-14, "grad_norm": 0.7253317832946777, "learning_rate": 0.0001248257213907827, "loss": 3.1938, "step": 2980 }, { "epoch": 1.0373646386341306e-14, "grad_norm": 0.7397022843360901, "learning_rate": 0.00012482395747648653, "loss": 3.2005, "step": 2990 }, { "epoch": 1.0408340855860843e-14, "grad_norm": 0.8318946361541748, "learning_rate": 0.00012482218469457663, "loss": 3.184, "step": 3000 }, { "epoch": 1.0443035325380379e-14, "grad_norm": 0.7288562059402466, "learning_rate": 0.00012482040304533342, "loss": 3.2116, "step": 3010 }, { "epoch": 1.0477729794899915e-14, "grad_norm": 0.7220286726951599, "learning_rate": 0.00012481861252903862, "loss": 3.2132, "step": 3020 }, { "epoch": 1.0512424264419451e-14, "grad_norm": 0.8244631886482239, "learning_rate": 0.0001248168131459754, "loss": 3.179, "step": 3030 }, { "epoch": 1.0547118733938987e-14, "grad_norm": 0.7161905765533447, "learning_rate": 0.0001248150048964283, "loss": 3.1869, "step": 3040 }, { "epoch": 1.0581813203458523e-14, "grad_norm": 0.880398690700531, "learning_rate": 0.00012481318778068327, "loss": 3.1897, "step": 3050 }, { "epoch": 1.061650767297806e-14, "grad_norm": 0.8268193006515503, "learning_rate": 0.00012481136179902773, "loss": 3.1524, "step": 3060 }, { "epoch": 1.0651202142497596e-14, "grad_norm": 0.7551537156105042, "learning_rate": 0.00012480952695175036, "loss": 3.1608, "step": 3070 }, { "epoch": 1.0685896612017132e-14, "grad_norm": 0.6582223176956177, "learning_rate": 0.00012480768323914137, "loss": 3.1905, "step": 3080 }, { "epoch": 1.0720591081536668e-14, "grad_norm": 0.7432381510734558, "learning_rate": 0.00012480583066149233, "loss": 3.1832, "step": 3090 }, { "epoch": 1.0755285551056204e-14, "grad_norm": 0.7773494124412537, "learning_rate": 0.00012480396921909622, "loss": 3.2064, "step": 3100 }, { "epoch": 1.078998002057574e-14, "grad_norm": 0.7896692752838135, "learning_rate": 0.00012480209891224737, "loss": 3.125, "step": 3110 }, { "epoch": 1.0824674490095276e-14, "grad_norm": 0.7044252753257751, "learning_rate": 0.00012480021974124159, "loss": 3.183, "step": 3120 }, { "epoch": 1.0859368959614812e-14, "grad_norm": 0.7290381789207458, "learning_rate": 0.00012479833170637604, "loss": 3.1689, "step": 3130 }, { "epoch": 1.0894063429134349e-14, "grad_norm": 0.7230690121650696, "learning_rate": 0.0001247964348079493, "loss": 3.166, "step": 3140 }, { "epoch": 1.0928757898653885e-14, "grad_norm": 0.7418133616447449, "learning_rate": 0.00012479452904626135, "loss": 3.1478, "step": 3150 }, { "epoch": 1.0963452368173421e-14, "grad_norm": 0.7461750507354736, "learning_rate": 0.00012479261442161356, "loss": 3.1556, "step": 3160 }, { "epoch": 1.0998146837692957e-14, "grad_norm": 0.7212677597999573, "learning_rate": 0.0001247906909343087, "loss": 3.1521, "step": 3170 }, { "epoch": 1.1032841307212493e-14, "grad_norm": 0.7130588293075562, "learning_rate": 0.00012478875858465099, "loss": 3.1717, "step": 3180 }, { "epoch": 1.106753577673203e-14, "grad_norm": 0.9168184399604797, "learning_rate": 0.00012478681737294601, "loss": 3.1594, "step": 3190 }, { "epoch": 1.1102230246251565e-14, "grad_norm": 0.7010019421577454, "learning_rate": 0.0001247848672995007, "loss": 3.1446, "step": 3200 }, { "epoch": 1.1136924715771102e-14, "grad_norm": 0.7739977240562439, "learning_rate": 0.0001247829083646235, "loss": 3.1707, "step": 3210 }, { "epoch": 1.1171619185290638e-14, "grad_norm": 0.6979143023490906, "learning_rate": 0.00012478094056862413, "loss": 3.1356, "step": 3220 }, { "epoch": 1.1206313654810174e-14, "grad_norm": 0.7540343999862671, "learning_rate": 0.0001247789639118139, "loss": 3.0953, "step": 3230 }, { "epoch": 1.124100812432971e-14, "grad_norm": 0.7760114073753357, "learning_rate": 0.00012477697839450528, "loss": 3.1579, "step": 3240 }, { "epoch": 1.1275702593849246e-14, "grad_norm": 0.7897104620933533, "learning_rate": 0.00012477498401701232, "loss": 3.1324, "step": 3250 }, { "epoch": 1.1310397063368782e-14, "grad_norm": 0.7589520812034607, "learning_rate": 0.0001247729807796504, "loss": 3.1521, "step": 3260 }, { "epoch": 1.1345091532888318e-14, "grad_norm": 0.8075765371322632, "learning_rate": 0.00012477096868273633, "loss": 3.1161, "step": 3270 }, { "epoch": 1.1379786002407855e-14, "grad_norm": 0.761272668838501, "learning_rate": 0.00012476894772658827, "loss": 3.1173, "step": 3280 }, { "epoch": 1.141448047192739e-14, "grad_norm": 0.7620368003845215, "learning_rate": 0.00012476691791152587, "loss": 3.1482, "step": 3290 }, { "epoch": 1.1449174941446927e-14, "grad_norm": 0.6672408580780029, "learning_rate": 0.0001247648792378701, "loss": 3.1034, "step": 3300 }, { "epoch": 1.1483869410966463e-14, "grad_norm": 0.7323631048202515, "learning_rate": 0.00012476283170594334, "loss": 3.1547, "step": 3310 }, { "epoch": 1.1518563880485999e-14, "grad_norm": 0.7070632576942444, "learning_rate": 0.00012476077531606942, "loss": 3.122, "step": 3320 }, { "epoch": 1.1553258350005535e-14, "grad_norm": 0.7476520538330078, "learning_rate": 0.00012475871006857353, "loss": 3.0888, "step": 3330 }, { "epoch": 1.1587952819525071e-14, "grad_norm": 0.718359649181366, "learning_rate": 0.00012475663596378227, "loss": 3.0729, "step": 3340 }, { "epoch": 1.1622647289044608e-14, "grad_norm": 0.6925711631774902, "learning_rate": 0.00012475455300202363, "loss": 3.0973, "step": 3350 }, { "epoch": 1.1657341758564144e-14, "grad_norm": 0.6826291680335999, "learning_rate": 0.00012475246118362701, "loss": 3.1146, "step": 3360 }, { "epoch": 1.169203622808368e-14, "grad_norm": 0.7196283936500549, "learning_rate": 0.00012475036050892325, "loss": 3.13, "step": 3370 }, { "epoch": 1.1726730697603216e-14, "grad_norm": 0.7545862197875977, "learning_rate": 0.00012474825097824453, "loss": 3.1113, "step": 3380 }, { "epoch": 1.1761425167122752e-14, "grad_norm": 0.6151329278945923, "learning_rate": 0.00012474613259192443, "loss": 3.0882, "step": 3390 }, { "epoch": 1.1796119636642288e-14, "grad_norm": 0.7248510718345642, "learning_rate": 0.00012474400535029801, "loss": 3.1441, "step": 3400 }, { "epoch": 1.1830814106161824e-14, "grad_norm": 0.7816692590713501, "learning_rate": 0.00012474186925370162, "loss": 3.1188, "step": 3410 }, { "epoch": 1.186550857568136e-14, "grad_norm": 0.7221320867538452, "learning_rate": 0.0001247397243024731, "loss": 3.0861, "step": 3420 }, { "epoch": 1.1900203045200897e-14, "grad_norm": 0.6229228377342224, "learning_rate": 0.00012473757049695164, "loss": 3.1001, "step": 3430 }, { "epoch": 1.1934897514720433e-14, "grad_norm": 0.7787955403327942, "learning_rate": 0.00012473540783747785, "loss": 3.1153, "step": 3440 }, { "epoch": 1.1969591984239969e-14, "grad_norm": 0.7670498490333557, "learning_rate": 0.00012473323632439376, "loss": 3.0933, "step": 3450 }, { "epoch": 1.2004286453759505e-14, "grad_norm": 0.6864545941352844, "learning_rate": 0.00012473105595804272, "loss": 3.0867, "step": 3460 }, { "epoch": 1.2038980923279041e-14, "grad_norm": 0.7928248643875122, "learning_rate": 0.00012472886673876957, "loss": 3.0766, "step": 3470 }, { "epoch": 1.2073675392798577e-14, "grad_norm": 0.7140690684318542, "learning_rate": 0.00012472666866692052, "loss": 3.1037, "step": 3480 }, { "epoch": 1.2108369862318114e-14, "grad_norm": 0.7960715889930725, "learning_rate": 0.00012472446174284317, "loss": 3.1122, "step": 3490 }, { "epoch": 1.214306433183765e-14, "grad_norm": 0.6798316836357117, "learning_rate": 0.00012472224596688655, "loss": 3.0693, "step": 3500 }, { "epoch": 1.2177758801357186e-14, "grad_norm": 0.7184551358222961, "learning_rate": 0.00012472002133940102, "loss": 3.0804, "step": 3510 }, { "epoch": 1.2212453270876722e-14, "grad_norm": 0.6529607772827148, "learning_rate": 0.00012471778786073842, "loss": 3.0766, "step": 3520 }, { "epoch": 1.2247147740396258e-14, "grad_norm": 0.7676675319671631, "learning_rate": 0.00012471554553125194, "loss": 3.0744, "step": 3530 }, { "epoch": 1.2281842209915794e-14, "grad_norm": 0.7851840853691101, "learning_rate": 0.00012471329435129617, "loss": 3.0638, "step": 3540 }, { "epoch": 1.231653667943533e-14, "grad_norm": 0.7693812847137451, "learning_rate": 0.00012471103432122715, "loss": 3.1136, "step": 3550 }, { "epoch": 1.2351231148954867e-14, "grad_norm": 0.6595890522003174, "learning_rate": 0.00012470876544140228, "loss": 3.0377, "step": 3560 }, { "epoch": 1.2385925618474403e-14, "grad_norm": 0.6804799437522888, "learning_rate": 0.00012470648771218033, "loss": 3.0777, "step": 3570 }, { "epoch": 1.2420620087993939e-14, "grad_norm": 0.708682656288147, "learning_rate": 0.00012470420113392153, "loss": 3.0709, "step": 3580 }, { "epoch": 1.2455314557513475e-14, "grad_norm": 0.7101171612739563, "learning_rate": 0.00012470190570698747, "loss": 3.0794, "step": 3590 }, { "epoch": 1.2490009027033011e-14, "grad_norm": 0.7054678797721863, "learning_rate": 0.0001246996014317412, "loss": 3.0477, "step": 3600 }, { "epoch": 1.2524703496552547e-14, "grad_norm": 0.7510730028152466, "learning_rate": 0.00012469728830854704, "loss": 3.057, "step": 3610 }, { "epoch": 1.2559397966072083e-14, "grad_norm": 0.6728399991989136, "learning_rate": 0.00012469496633777085, "loss": 3.0661, "step": 3620 }, { "epoch": 1.259409243559162e-14, "grad_norm": 0.6882238984107971, "learning_rate": 0.00012469263551977982, "loss": 3.0691, "step": 3630 }, { "epoch": 1.2628786905111156e-14, "grad_norm": 0.6853118538856506, "learning_rate": 0.00012469029585494251, "loss": 3.1002, "step": 3640 }, { "epoch": 1.2663481374630692e-14, "grad_norm": 0.6914693117141724, "learning_rate": 0.00012468794734362898, "loss": 3.0565, "step": 3650 }, { "epoch": 1.2698175844150228e-14, "grad_norm": 0.673748791217804, "learning_rate": 0.00012468558998621055, "loss": 3.0594, "step": 3660 }, { "epoch": 1.2732870313669764e-14, "grad_norm": 0.6834123730659485, "learning_rate": 0.0001246832237830601, "loss": 3.0657, "step": 3670 }, { "epoch": 1.27675647831893e-14, "grad_norm": 0.7288254499435425, "learning_rate": 0.00012468084873455177, "loss": 3.0325, "step": 3680 }, { "epoch": 1.2802259252708836e-14, "grad_norm": 0.6663938760757446, "learning_rate": 0.00012467846484106118, "loss": 3.0182, "step": 3690 }, { "epoch": 1.2836953722228372e-14, "grad_norm": 0.6900312900543213, "learning_rate": 0.0001246760721029653, "loss": 3.0622, "step": 3700 }, { "epoch": 1.2871648191747909e-14, "grad_norm": 0.6715964674949646, "learning_rate": 0.00012467367052064254, "loss": 3.0465, "step": 3710 }, { "epoch": 1.2906342661267445e-14, "grad_norm": 0.6792755126953125, "learning_rate": 0.00012467126009447268, "loss": 3.0486, "step": 3720 }, { "epoch": 1.2941037130786981e-14, "grad_norm": 0.6889999508857727, "learning_rate": 0.0001246688408248369, "loss": 3.0633, "step": 3730 }, { "epoch": 1.2975731600306517e-14, "grad_norm": 0.7507778406143188, "learning_rate": 0.0001246664127121178, "loss": 3.0589, "step": 3740 }, { "epoch": 1.3010426069826053e-14, "grad_norm": 0.785260796546936, "learning_rate": 0.00012466397575669935, "loss": 3.0201, "step": 3750 }, { "epoch": 1.304512053934559e-14, "grad_norm": 0.7369335889816284, "learning_rate": 0.00012466152995896696, "loss": 3.0194, "step": 3760 }, { "epoch": 1.3079815008865125e-14, "grad_norm": 0.7063583135604858, "learning_rate": 0.0001246590753193074, "loss": 3.0212, "step": 3770 }, { "epoch": 1.3114509478384662e-14, "grad_norm": 0.7327459454536438, "learning_rate": 0.00012465661183810886, "loss": 3.0432, "step": 3780 }, { "epoch": 1.3149203947904198e-14, "grad_norm": 0.671994686126709, "learning_rate": 0.00012465413951576087, "loss": 3.0037, "step": 3790 }, { "epoch": 1.3183898417423734e-14, "grad_norm": 0.7931689023971558, "learning_rate": 0.00012465165835265447, "loss": 3.0759, "step": 3800 }, { "epoch": 1.321859288694327e-14, "grad_norm": 0.6893624663352966, "learning_rate": 0.00012464916834918199, "loss": 3.0365, "step": 3810 }, { "epoch": 1.3253287356462806e-14, "grad_norm": 0.7476165294647217, "learning_rate": 0.00012464666950573724, "loss": 3.0465, "step": 3820 }, { "epoch": 1.3287981825982342e-14, "grad_norm": 0.6562722325325012, "learning_rate": 0.00012464416182271534, "loss": 3.0297, "step": 3830 }, { "epoch": 1.3322676295501878e-14, "grad_norm": 0.7290176153182983, "learning_rate": 0.0001246416453005129, "loss": 3.0461, "step": 3840 }, { "epoch": 1.3357370765021415e-14, "grad_norm": 0.7958412766456604, "learning_rate": 0.00012463911993952788, "loss": 3.0429, "step": 3850 }, { "epoch": 1.339206523454095e-14, "grad_norm": 0.7099930047988892, "learning_rate": 0.00012463658574015963, "loss": 2.9997, "step": 3860 }, { "epoch": 1.3426759704060487e-14, "grad_norm": 0.6824677586555481, "learning_rate": 0.00012463404270280894, "loss": 3.0016, "step": 3870 }, { "epoch": 1.3461454173580023e-14, "grad_norm": 0.7381972074508667, "learning_rate": 0.00012463149082787793, "loss": 3.0158, "step": 3880 }, { "epoch": 1.3496148643099559e-14, "grad_norm": 0.7286034822463989, "learning_rate": 0.00012462893011577018, "loss": 3.0281, "step": 3890 }, { "epoch": 1.3530843112619095e-14, "grad_norm": 0.6532379388809204, "learning_rate": 0.00012462636056689063, "loss": 3.01, "step": 3900 }, { "epoch": 1.3565537582138631e-14, "grad_norm": 0.7265604734420776, "learning_rate": 0.00012462378218164568, "loss": 2.9586, "step": 3910 }, { "epoch": 1.3600232051658168e-14, "grad_norm": 0.7170725464820862, "learning_rate": 0.000124621194960443, "loss": 3.0039, "step": 3920 }, { "epoch": 1.3634926521177704e-14, "grad_norm": 0.7389524579048157, "learning_rate": 0.0001246185989036918, "loss": 3.0106, "step": 3930 }, { "epoch": 1.366962099069724e-14, "grad_norm": 0.6490112543106079, "learning_rate": 0.00012461599401180262, "loss": 3.0187, "step": 3940 }, { "epoch": 1.3704315460216776e-14, "grad_norm": 0.6700443625450134, "learning_rate": 0.00012461338028518735, "loss": 3.0239, "step": 3950 }, { "epoch": 1.3739009929736312e-14, "grad_norm": 0.6895691752433777, "learning_rate": 0.00012461075772425938, "loss": 3.0483, "step": 3960 }, { "epoch": 1.3773704399255848e-14, "grad_norm": 0.6426942944526672, "learning_rate": 0.00012460812632943344, "loss": 3.0123, "step": 3970 }, { "epoch": 1.3808398868775384e-14, "grad_norm": 0.7045239806175232, "learning_rate": 0.00012460548610112565, "loss": 3.0633, "step": 3980 }, { "epoch": 1.384309333829492e-14, "grad_norm": 0.7273630499839783, "learning_rate": 0.00012460283703975353, "loss": 3.0271, "step": 3990 }, { "epoch": 1.3877787807814457e-14, "grad_norm": 0.6235355138778687, "learning_rate": 0.000124600179145736, "loss": 3.0032, "step": 4000 }, { "epoch": 1.3912482277333993e-14, "grad_norm": 0.8024575710296631, "learning_rate": 0.00012459751241949342, "loss": 3.0002, "step": 4010 }, { "epoch": 1.3947176746853529e-14, "grad_norm": 0.6960814595222473, "learning_rate": 0.0001245948368614475, "loss": 3.0072, "step": 4020 }, { "epoch": 1.3981871216373065e-14, "grad_norm": 0.7123613953590393, "learning_rate": 0.0001245921524720213, "loss": 3.0125, "step": 4030 }, { "epoch": 1.4016565685892601e-14, "grad_norm": 0.6465147733688354, "learning_rate": 0.00012458945925163944, "loss": 2.9801, "step": 4040 }, { "epoch": 1.4051260155412137e-14, "grad_norm": 0.766997754573822, "learning_rate": 0.00012458675720072773, "loss": 3.0143, "step": 4050 }, { "epoch": 1.4085954624931674e-14, "grad_norm": 0.6654321551322937, "learning_rate": 0.00012458404631971354, "loss": 2.9965, "step": 4060 }, { "epoch": 1.412064909445121e-14, "grad_norm": 0.6323001384735107, "learning_rate": 0.00012458132660902554, "loss": 3.0071, "step": 4070 }, { "epoch": 1.4155343563970746e-14, "grad_norm": 0.790391743183136, "learning_rate": 0.00012457859806909381, "loss": 2.9942, "step": 4080 }, { "epoch": 1.4190038033490282e-14, "grad_norm": 0.6655433773994446, "learning_rate": 0.0001245758607003499, "loss": 2.9962, "step": 4090 }, { "epoch": 1.4224732503009818e-14, "grad_norm": 0.7439237833023071, "learning_rate": 0.00012457311450322664, "loss": 3.049, "step": 4100 }, { "epoch": 1.4259426972529354e-14, "grad_norm": 0.6412649154663086, "learning_rate": 0.00012457035947815838, "loss": 2.9676, "step": 4110 }, { "epoch": 1.429412144204889e-14, "grad_norm": 0.6341674327850342, "learning_rate": 0.00012456759562558077, "loss": 3.022, "step": 4120 }, { "epoch": 1.4328815911568427e-14, "grad_norm": 0.6322102546691895, "learning_rate": 0.0001245648229459309, "loss": 2.9499, "step": 4130 }, { "epoch": 1.4363510381087963e-14, "grad_norm": 0.6494558453559875, "learning_rate": 0.00012456204143964722, "loss": 2.9976, "step": 4140 }, { "epoch": 1.43982048506075e-14, "grad_norm": 0.7101941108703613, "learning_rate": 0.00012455925110716961, "loss": 3.0186, "step": 4150 }, { "epoch": 1.4432899320127035e-14, "grad_norm": 0.7319994568824768, "learning_rate": 0.00012455645194893938, "loss": 2.9911, "step": 4160 }, { "epoch": 1.446759378964657e-14, "grad_norm": 0.6181273460388184, "learning_rate": 0.00012455364396539916, "loss": 2.9671, "step": 4170 }, { "epoch": 1.4502288259166107e-14, "grad_norm": 0.6182767748832703, "learning_rate": 0.00012455082715699298, "loss": 2.9797, "step": 4180 }, { "epoch": 1.4536982728685643e-14, "grad_norm": 0.6621291637420654, "learning_rate": 0.00012454800152416633, "loss": 2.9832, "step": 4190 }, { "epoch": 1.457167719820518e-14, "grad_norm": 0.7200559973716736, "learning_rate": 0.00012454516706736607, "loss": 2.9667, "step": 4200 }, { "epoch": 1.4606371667724716e-14, "grad_norm": 0.7144989967346191, "learning_rate": 0.0001245423237870404, "loss": 2.9483, "step": 4210 }, { "epoch": 1.4641066137244252e-14, "grad_norm": 0.6361348628997803, "learning_rate": 0.000124539471683639, "loss": 2.9564, "step": 4220 }, { "epoch": 1.4675760606763788e-14, "grad_norm": 0.7094681859016418, "learning_rate": 0.00012453661075761288, "loss": 2.9637, "step": 4230 }, { "epoch": 1.4710455076283324e-14, "grad_norm": 0.6563013195991516, "learning_rate": 0.0001245337410094145, "loss": 2.9862, "step": 4240 }, { "epoch": 1.474514954580286e-14, "grad_norm": 0.7973436713218689, "learning_rate": 0.00012453086243949767, "loss": 2.9711, "step": 4250 }, { "epoch": 1.4779844015322396e-14, "grad_norm": 0.6667431592941284, "learning_rate": 0.0001245279750483176, "loss": 2.9671, "step": 4260 }, { "epoch": 1.4814538484841933e-14, "grad_norm": 0.6637325882911682, "learning_rate": 0.00012452507883633091, "loss": 3.0024, "step": 4270 }, { "epoch": 1.484923295436147e-14, "grad_norm": 0.6902765035629272, "learning_rate": 0.00012452217380399562, "loss": 2.9848, "step": 4280 }, { "epoch": 1.4883927423881005e-14, "grad_norm": 0.6676066517829895, "learning_rate": 0.00012451925995177115, "loss": 2.968, "step": 4290 }, { "epoch": 1.491862189340054e-14, "grad_norm": 0.68576979637146, "learning_rate": 0.00012451633728011828, "loss": 2.9368, "step": 4300 }, { "epoch": 1.4953316362920077e-14, "grad_norm": 0.6960861086845398, "learning_rate": 0.00012451340578949918, "loss": 2.9944, "step": 4310 }, { "epoch": 1.4988010832439613e-14, "grad_norm": 0.6907550096511841, "learning_rate": 0.0001245104654803775, "loss": 2.9824, "step": 4320 }, { "epoch": 1.502270530195915e-14, "grad_norm": 0.660068690776825, "learning_rate": 0.00012450751635321818, "loss": 2.9757, "step": 4330 }, { "epoch": 1.5057399771478686e-14, "grad_norm": 0.6573997139930725, "learning_rate": 0.00012450455840848763, "loss": 2.9512, "step": 4340 }, { "epoch": 1.5092094240998222e-14, "grad_norm": 0.6628772020339966, "learning_rate": 0.00012450159164665358, "loss": 2.9597, "step": 4350 }, { "epoch": 1.5126788710517758e-14, "grad_norm": 0.6509264707565308, "learning_rate": 0.00012449861606818525, "loss": 2.9452, "step": 4360 }, { "epoch": 1.5161483180037294e-14, "grad_norm": 0.7052345275878906, "learning_rate": 0.00012449563167355317, "loss": 2.9774, "step": 4370 }, { "epoch": 1.519617764955683e-14, "grad_norm": 0.6775937676429749, "learning_rate": 0.00012449263846322928, "loss": 2.9467, "step": 4380 }, { "epoch": 1.5230872119076366e-14, "grad_norm": 0.6212556958198547, "learning_rate": 0.00012448963643768698, "loss": 2.976, "step": 4390 }, { "epoch": 1.5265566588595902e-14, "grad_norm": 0.6026918292045593, "learning_rate": 0.00012448662559740096, "loss": 2.8954, "step": 4400 }, { "epoch": 1.530026105811544e-14, "grad_norm": 0.6559715867042542, "learning_rate": 0.0001244836059428474, "loss": 2.9687, "step": 4410 }, { "epoch": 1.5334955527634975e-14, "grad_norm": 0.7170567512512207, "learning_rate": 0.00012448057747450384, "loss": 2.9653, "step": 4420 }, { "epoch": 1.536964999715451e-14, "grad_norm": 0.6450586318969727, "learning_rate": 0.00012447754019284915, "loss": 2.9486, "step": 4430 }, { "epoch": 1.5404344466674047e-14, "grad_norm": 0.6832867860794067, "learning_rate": 0.0001244744940983637, "loss": 2.9616, "step": 4440 }, { "epoch": 1.5439038936193583e-14, "grad_norm": 0.7528901100158691, "learning_rate": 0.0001244714391915292, "loss": 2.9915, "step": 4450 }, { "epoch": 1.547373340571312e-14, "grad_norm": 0.6651819348335266, "learning_rate": 0.00012446837547282872, "loss": 2.9502, "step": 4460 }, { "epoch": 1.5508427875232655e-14, "grad_norm": 0.697746992111206, "learning_rate": 0.00012446530294274682, "loss": 2.9282, "step": 4470 }, { "epoch": 1.554312234475219e-14, "grad_norm": 0.6883642077445984, "learning_rate": 0.00012446222160176934, "loss": 2.9767, "step": 4480 }, { "epoch": 1.5577816814271728e-14, "grad_norm": 0.67514568567276, "learning_rate": 0.0001244591314503836, "loss": 2.9328, "step": 4490 }, { "epoch": 1.5612511283791264e-14, "grad_norm": 0.6155393719673157, "learning_rate": 0.00012445603248907824, "loss": 2.9279, "step": 4500 }, { "epoch": 1.56472057533108e-14, "grad_norm": 0.6523802280426025, "learning_rate": 0.00012445292471834337, "loss": 2.945, "step": 4510 }, { "epoch": 1.5681900222830336e-14, "grad_norm": 0.6453002691268921, "learning_rate": 0.00012444980813867046, "loss": 2.9758, "step": 4520 }, { "epoch": 1.5716594692349872e-14, "grad_norm": 0.6484992504119873, "learning_rate": 0.00012444668275055236, "loss": 2.9306, "step": 4530 }, { "epoch": 1.5751289161869408e-14, "grad_norm": 0.6696417927742004, "learning_rate": 0.00012444354855448332, "loss": 2.9522, "step": 4540 }, { "epoch": 1.5785983631388945e-14, "grad_norm": 0.5923197269439697, "learning_rate": 0.000124440405550959, "loss": 2.9383, "step": 4550 }, { "epoch": 1.582067810090848e-14, "grad_norm": 0.6410911083221436, "learning_rate": 0.0001244372537404764, "loss": 2.9401, "step": 4560 }, { "epoch": 1.5855372570428017e-14, "grad_norm": 0.6767243146896362, "learning_rate": 0.000124434093123534, "loss": 2.9276, "step": 4570 }, { "epoch": 1.5890067039947553e-14, "grad_norm": 0.6143051981925964, "learning_rate": 0.00012443092370063163, "loss": 2.912, "step": 4580 }, { "epoch": 1.592476150946709e-14, "grad_norm": 0.6822436451911926, "learning_rate": 0.00012442774547227043, "loss": 2.9309, "step": 4590 }, { "epoch": 1.5959455978986625e-14, "grad_norm": 0.6930224895477295, "learning_rate": 0.0001244245584389531, "loss": 2.9278, "step": 4600 }, { "epoch": 1.599415044850616e-14, "grad_norm": 0.625003457069397, "learning_rate": 0.00012442136260118358, "loss": 2.9273, "step": 4610 }, { "epoch": 1.6028844918025698e-14, "grad_norm": 0.7277611494064331, "learning_rate": 0.0001244181579594673, "loss": 2.9698, "step": 4620 }, { "epoch": 1.6063539387545234e-14, "grad_norm": 0.6799565553665161, "learning_rate": 0.00012441494451431105, "loss": 2.9425, "step": 4630 }, { "epoch": 1.609823385706477e-14, "grad_norm": 0.7252404093742371, "learning_rate": 0.00012441172226622296, "loss": 2.9508, "step": 4640 }, { "epoch": 1.6132928326584306e-14, "grad_norm": 0.682142972946167, "learning_rate": 0.00012440849121571264, "loss": 2.906, "step": 4650 }, { "epoch": 1.6167622796103842e-14, "grad_norm": 0.7829553484916687, "learning_rate": 0.00012440525136329106, "loss": 2.9439, "step": 4660 }, { "epoch": 1.6202317265623378e-14, "grad_norm": 0.6608889698982239, "learning_rate": 0.00012440200270947055, "loss": 2.936, "step": 4670 }, { "epoch": 1.6237011735142914e-14, "grad_norm": 0.6656187772750854, "learning_rate": 0.00012439874525476487, "loss": 2.8845, "step": 4680 }, { "epoch": 1.627170620466245e-14, "grad_norm": 0.6729143857955933, "learning_rate": 0.00012439547899968913, "loss": 2.9277, "step": 4690 }, { "epoch": 1.6306400674181987e-14, "grad_norm": 0.7461599111557007, "learning_rate": 0.00012439220394475992, "loss": 2.896, "step": 4700 }, { "epoch": 1.6341095143701523e-14, "grad_norm": 0.653232216835022, "learning_rate": 0.00012438892009049513, "loss": 2.906, "step": 4710 }, { "epoch": 1.637578961322106e-14, "grad_norm": 0.6327627897262573, "learning_rate": 0.00012438562743741406, "loss": 2.8935, "step": 4720 }, { "epoch": 1.6410484082740595e-14, "grad_norm": 0.6152717471122742, "learning_rate": 0.00012438232598603743, "loss": 2.8998, "step": 4730 }, { "epoch": 1.644517855226013e-14, "grad_norm": 0.627433717250824, "learning_rate": 0.00012437901573688733, "loss": 2.9308, "step": 4740 }, { "epoch": 1.6479873021779667e-14, "grad_norm": 0.66085284948349, "learning_rate": 0.00012437569669048723, "loss": 2.945, "step": 4750 }, { "epoch": 1.6514567491299204e-14, "grad_norm": 0.6678071022033691, "learning_rate": 0.00012437236884736206, "loss": 2.9264, "step": 4760 }, { "epoch": 1.654926196081874e-14, "grad_norm": 0.705095648765564, "learning_rate": 0.00012436903220803804, "loss": 2.905, "step": 4770 }, { "epoch": 1.6583956430338276e-14, "grad_norm": 0.6709147691726685, "learning_rate": 0.00012436568677304282, "loss": 2.9208, "step": 4780 }, { "epoch": 1.6618650899857812e-14, "grad_norm": 0.6072796583175659, "learning_rate": 0.00012436233254290552, "loss": 2.9294, "step": 4790 }, { "epoch": 1.6653345369377348e-14, "grad_norm": 0.6151487231254578, "learning_rate": 0.00012435896951815654, "loss": 2.9011, "step": 4800 }, { "epoch": 1.6688039838896884e-14, "grad_norm": 0.6451088190078735, "learning_rate": 0.0001243555976993277, "loss": 2.9041, "step": 4810 }, { "epoch": 1.672273430841642e-14, "grad_norm": 0.6668460369110107, "learning_rate": 0.00012435221708695225, "loss": 2.8972, "step": 4820 }, { "epoch": 1.6757428777935957e-14, "grad_norm": 0.6194532513618469, "learning_rate": 0.0001243488276815648, "loss": 2.8888, "step": 4830 }, { "epoch": 1.6792123247455493e-14, "grad_norm": 0.6643616557121277, "learning_rate": 0.00012434542948370133, "loss": 2.9321, "step": 4840 }, { "epoch": 1.682681771697503e-14, "grad_norm": 0.930942714214325, "learning_rate": 0.0001243420224938993, "loss": 2.9282, "step": 4850 }, { "epoch": 1.6861512186494565e-14, "grad_norm": 0.6564489603042603, "learning_rate": 0.0001243386067126974, "loss": 2.9303, "step": 4860 }, { "epoch": 1.68962066560141e-14, "grad_norm": 0.6609086394309998, "learning_rate": 0.00012433518214063587, "loss": 2.8856, "step": 4870 }, { "epoch": 1.6930901125533637e-14, "grad_norm": 0.681067168712616, "learning_rate": 0.00012433174877825626, "loss": 2.9399, "step": 4880 }, { "epoch": 1.6965595595053173e-14, "grad_norm": 0.620232880115509, "learning_rate": 0.00012432830662610156, "loss": 2.9347, "step": 4890 }, { "epoch": 1.700029006457271e-14, "grad_norm": 0.674216091632843, "learning_rate": 0.00012432485568471607, "loss": 2.9054, "step": 4900 }, { "epoch": 1.7034984534092246e-14, "grad_norm": 0.6145908236503601, "learning_rate": 0.00012432139595464555, "loss": 2.9109, "step": 4910 }, { "epoch": 1.7069679003611782e-14, "grad_norm": 0.5697910785675049, "learning_rate": 0.00012431792743643712, "loss": 2.8799, "step": 4920 }, { "epoch": 1.7104373473131318e-14, "grad_norm": 0.6260907053947449, "learning_rate": 0.0001243144501306393, "loss": 2.9033, "step": 4930 }, { "epoch": 1.7139067942650854e-14, "grad_norm": 0.7109225392341614, "learning_rate": 0.000124310964037802, "loss": 2.8997, "step": 4940 }, { "epoch": 1.717376241217039e-14, "grad_norm": 0.5778696537017822, "learning_rate": 0.00012430746915847651, "loss": 2.8694, "step": 4950 }, { "epoch": 1.7208456881689926e-14, "grad_norm": 0.6547163128852844, "learning_rate": 0.0001243039654932155, "loss": 2.87, "step": 4960 }, { "epoch": 1.7243151351209463e-14, "grad_norm": 0.5900265574455261, "learning_rate": 0.00012430045304257306, "loss": 2.8675, "step": 4970 }, { "epoch": 1.7277845820729e-14, "grad_norm": 0.6545571088790894, "learning_rate": 0.00012429693180710465, "loss": 2.8489, "step": 4980 }, { "epoch": 1.7312540290248535e-14, "grad_norm": 0.6126354932785034, "learning_rate": 0.00012429340178736714, "loss": 2.8966, "step": 4990 }, { "epoch": 1.734723475976807e-14, "grad_norm": 0.6317709684371948, "learning_rate": 0.00012428986298391876, "loss": 2.902, "step": 5000 }, { "epoch": 1.7381929229287607e-14, "grad_norm": 0.6308935284614563, "learning_rate": 0.00012428631539731914, "loss": 2.8895, "step": 5010 }, { "epoch": 1.7416623698807143e-14, "grad_norm": 0.6461998820304871, "learning_rate": 0.00012428275902812928, "loss": 2.8602, "step": 5020 }, { "epoch": 1.745131816832668e-14, "grad_norm": 0.5970547199249268, "learning_rate": 0.00012427919387691163, "loss": 2.8886, "step": 5030 }, { "epoch": 1.7486012637846216e-14, "grad_norm": 0.6639490127563477, "learning_rate": 0.00012427561994422996, "loss": 2.8976, "step": 5040 }, { "epoch": 1.7520707107365752e-14, "grad_norm": 0.6429731845855713, "learning_rate": 0.00012427203723064945, "loss": 2.9007, "step": 5050 }, { "epoch": 1.7555401576885288e-14, "grad_norm": 0.5911052823066711, "learning_rate": 0.0001242684457367367, "loss": 2.906, "step": 5060 }, { "epoch": 1.7590096046404824e-14, "grad_norm": 0.6038382053375244, "learning_rate": 0.00012426484546305966, "loss": 2.8914, "step": 5070 }, { "epoch": 1.762479051592436e-14, "grad_norm": 0.6138734221458435, "learning_rate": 0.0001242612364101877, "loss": 2.8841, "step": 5080 }, { "epoch": 1.7659484985443896e-14, "grad_norm": 0.5827398896217346, "learning_rate": 0.00012425761857869154, "loss": 2.8641, "step": 5090 }, { "epoch": 1.7694179454963432e-14, "grad_norm": 0.6348038911819458, "learning_rate": 0.00012425399196914328, "loss": 2.8433, "step": 5100 }, { "epoch": 1.772887392448297e-14, "grad_norm": 0.6564421653747559, "learning_rate": 0.00012425035658211652, "loss": 2.843, "step": 5110 }, { "epoch": 1.7763568394002505e-14, "grad_norm": 0.6284332275390625, "learning_rate": 0.0001242467124181861, "loss": 2.8578, "step": 5120 }, { "epoch": 1.779826286352204e-14, "grad_norm": 0.6046045422554016, "learning_rate": 0.00012424305947792832, "loss": 2.8547, "step": 5130 }, { "epoch": 1.7832957333041577e-14, "grad_norm": 0.6275157928466797, "learning_rate": 0.0001242393977619209, "loss": 2.9054, "step": 5140 }, { "epoch": 1.7867651802561113e-14, "grad_norm": 0.6477947235107422, "learning_rate": 0.00012423572727074283, "loss": 2.8567, "step": 5150 }, { "epoch": 1.790234627208065e-14, "grad_norm": 0.7141586542129517, "learning_rate": 0.00012423204800497467, "loss": 2.8511, "step": 5160 }, { "epoch": 1.7937040741600185e-14, "grad_norm": 0.6430010795593262, "learning_rate": 0.00012422835996519818, "loss": 2.8503, "step": 5170 }, { "epoch": 1.797173521111972e-14, "grad_norm": 0.6338449120521545, "learning_rate": 0.00012422466315199662, "loss": 2.8828, "step": 5180 }, { "epoch": 1.8006429680639258e-14, "grad_norm": 0.634334921836853, "learning_rate": 0.00012422095756595462, "loss": 2.8691, "step": 5190 }, { "epoch": 1.8041124150158794e-14, "grad_norm": 0.6007212400436401, "learning_rate": 0.00012421724320765818, "loss": 2.874, "step": 5200 }, { "epoch": 1.807581861967833e-14, "grad_norm": 0.6862580180168152, "learning_rate": 0.0001242135200776947, "loss": 2.8969, "step": 5210 }, { "epoch": 1.8110513089197866e-14, "grad_norm": 0.5748586654663086, "learning_rate": 0.00012420978817665292, "loss": 2.8735, "step": 5220 }, { "epoch": 1.8145207558717402e-14, "grad_norm": 0.6087973713874817, "learning_rate": 0.0001242060475051231, "loss": 2.8977, "step": 5230 }, { "epoch": 1.8179902028236938e-14, "grad_norm": 0.5898349285125732, "learning_rate": 0.00012420229806369668, "loss": 2.879, "step": 5240 }, { "epoch": 1.8214596497756474e-14, "grad_norm": 0.6444045305252075, "learning_rate": 0.00012419853985296667, "loss": 2.832, "step": 5250 }, { "epoch": 1.824929096727601e-14, "grad_norm": 0.633687436580658, "learning_rate": 0.0001241947728735274, "loss": 2.8747, "step": 5260 }, { "epoch": 1.8283985436795547e-14, "grad_norm": 0.6253474950790405, "learning_rate": 0.00012419099712597455, "loss": 2.853, "step": 5270 }, { "epoch": 1.8318679906315083e-14, "grad_norm": 0.6750002503395081, "learning_rate": 0.00012418721261090528, "loss": 2.8355, "step": 5280 }, { "epoch": 1.835337437583462e-14, "grad_norm": 1.5764212608337402, "learning_rate": 0.000124183419328918, "loss": 2.8633, "step": 5290 }, { "epoch": 1.8388068845354155e-14, "grad_norm": 0.5921177864074707, "learning_rate": 0.00012417961728061267, "loss": 2.8772, "step": 5300 }, { "epoch": 1.842276331487369e-14, "grad_norm": 0.6291271448135376, "learning_rate": 0.0001241758064665905, "loss": 2.8566, "step": 5310 }, { "epoch": 1.8457457784393227e-14, "grad_norm": 0.6287446618080139, "learning_rate": 0.00012417198688745412, "loss": 2.8894, "step": 5320 }, { "epoch": 1.8492152253912764e-14, "grad_norm": 0.6294393539428711, "learning_rate": 0.0001241681585438076, "loss": 2.8845, "step": 5330 }, { "epoch": 1.85268467234323e-14, "grad_norm": 0.6331487894058228, "learning_rate": 0.00012416432143625637, "loss": 2.8598, "step": 5340 }, { "epoch": 1.8561541192951836e-14, "grad_norm": 0.674874484539032, "learning_rate": 0.00012416047556540718, "loss": 2.8613, "step": 5350 }, { "epoch": 1.8596235662471372e-14, "grad_norm": 0.5660271048545837, "learning_rate": 0.00012415662093186828, "loss": 2.8657, "step": 5360 }, { "epoch": 1.8630930131990908e-14, "grad_norm": 0.6504572629928589, "learning_rate": 0.0001241527575362492, "loss": 2.8539, "step": 5370 }, { "epoch": 1.8665624601510444e-14, "grad_norm": 0.6455576419830322, "learning_rate": 0.00012414888537916097, "loss": 2.8297, "step": 5380 }, { "epoch": 1.870031907102998e-14, "grad_norm": 0.641903281211853, "learning_rate": 0.00012414500446121586, "loss": 2.8527, "step": 5390 }, { "epoch": 1.8735013540549517e-14, "grad_norm": 0.618110179901123, "learning_rate": 0.00012414111478302765, "loss": 2.8685, "step": 5400 }, { "epoch": 1.8769708010069053e-14, "grad_norm": 0.5739728212356567, "learning_rate": 0.00012413721634521143, "loss": 2.8576, "step": 5410 }, { "epoch": 1.880440247958859e-14, "grad_norm": 0.6063082814216614, "learning_rate": 0.00012413330914838374, "loss": 2.877, "step": 5420 }, { "epoch": 1.8839096949108125e-14, "grad_norm": 0.6450121402740479, "learning_rate": 0.00012412939319316244, "loss": 2.8514, "step": 5430 }, { "epoch": 1.887379141862766e-14, "grad_norm": 0.6283537149429321, "learning_rate": 0.00012412546848016682, "loss": 2.8327, "step": 5440 }, { "epoch": 1.8908485888147197e-14, "grad_norm": 0.6947651505470276, "learning_rate": 0.00012412153501001753, "loss": 2.8722, "step": 5450 }, { "epoch": 1.8943180357666733e-14, "grad_norm": 0.6342113614082336, "learning_rate": 0.00012411759278333664, "loss": 2.8614, "step": 5460 }, { "epoch": 1.897787482718627e-14, "grad_norm": 0.6199445128440857, "learning_rate": 0.00012411364180074753, "loss": 2.8295, "step": 5470 }, { "epoch": 1.9012569296705806e-14, "grad_norm": 0.6593498587608337, "learning_rate": 0.00012410968206287508, "loss": 2.9103, "step": 5480 }, { "epoch": 1.9047263766225342e-14, "grad_norm": 0.5745114088058472, "learning_rate": 0.00012410571357034542, "loss": 2.8773, "step": 5490 }, { "epoch": 1.9081958235744878e-14, "grad_norm": 0.5985862016677856, "learning_rate": 0.00012410173632378615, "loss": 2.848, "step": 5500 }, { "epoch": 1.9116652705264414e-14, "grad_norm": 0.6325246095657349, "learning_rate": 0.0001240977503238263, "loss": 2.8385, "step": 5510 }, { "epoch": 1.915134717478395e-14, "grad_norm": 0.6150628924369812, "learning_rate": 0.00012409375557109612, "loss": 2.8565, "step": 5520 }, { "epoch": 1.9186041644303486e-14, "grad_norm": 0.6192411780357361, "learning_rate": 0.00012408975206622743, "loss": 2.8271, "step": 5530 }, { "epoch": 1.9220736113823023e-14, "grad_norm": 0.5619994401931763, "learning_rate": 0.00012408573980985331, "loss": 2.8591, "step": 5540 }, { "epoch": 1.925543058334256e-14, "grad_norm": 0.6149645447731018, "learning_rate": 0.0001240817188026083, "loss": 2.8295, "step": 5550 }, { "epoch": 1.9290125052862095e-14, "grad_norm": 0.6415473818778992, "learning_rate": 0.00012407768904512822, "loss": 2.832, "step": 5560 }, { "epoch": 1.932481952238163e-14, "grad_norm": 0.6282888054847717, "learning_rate": 0.0001240736505380504, "loss": 2.8583, "step": 5570 }, { "epoch": 1.9359513991901167e-14, "grad_norm": 0.6100879311561584, "learning_rate": 0.00012406960328201346, "loss": 2.8677, "step": 5580 }, { "epoch": 1.9394208461420703e-14, "grad_norm": 0.6105282306671143, "learning_rate": 0.0001240655472776575, "loss": 2.8586, "step": 5590 }, { "epoch": 1.942890293094024e-14, "grad_norm": 0.6249734163284302, "learning_rate": 0.00012406148252562386, "loss": 2.8233, "step": 5600 }, { "epoch": 1.9463597400459776e-14, "grad_norm": 0.6065224409103394, "learning_rate": 0.0001240574090265554, "loss": 2.8247, "step": 5610 }, { "epoch": 1.9498291869979312e-14, "grad_norm": 0.5984311699867249, "learning_rate": 0.00012405332678109627, "loss": 2.8603, "step": 5620 }, { "epoch": 1.9532986339498848e-14, "grad_norm": 0.6282042264938354, "learning_rate": 0.00012404923578989208, "loss": 2.803, "step": 5630 }, { "epoch": 1.9567680809018384e-14, "grad_norm": 0.5800025463104248, "learning_rate": 0.00012404513605358976, "loss": 2.8092, "step": 5640 }, { "epoch": 1.960237527853792e-14, "grad_norm": 0.627775251865387, "learning_rate": 0.00012404102757283765, "loss": 2.8279, "step": 5650 }, { "epoch": 1.9637069748057456e-14, "grad_norm": 0.6285877823829651, "learning_rate": 0.0001240369103482855, "loss": 2.8429, "step": 5660 }, { "epoch": 1.9671764217576992e-14, "grad_norm": 0.5945749282836914, "learning_rate": 0.00012403278438058438, "loss": 2.827, "step": 5670 }, { "epoch": 1.970645868709653e-14, "grad_norm": 0.6339585185050964, "learning_rate": 0.00012402864967038678, "loss": 2.8683, "step": 5680 }, { "epoch": 1.9741153156616065e-14, "grad_norm": 0.6999961733818054, "learning_rate": 0.0001240245062183466, "loss": 2.8214, "step": 5690 }, { "epoch": 1.97758476261356e-14, "grad_norm": 0.5739607810974121, "learning_rate": 0.00012402035402511906, "loss": 2.8184, "step": 5700 }, { "epoch": 1.9810542095655137e-14, "grad_norm": 0.6444486975669861, "learning_rate": 0.00012401619309136079, "loss": 2.8174, "step": 5710 }, { "epoch": 1.9845236565174673e-14, "grad_norm": 0.6257888674736023, "learning_rate": 0.0001240120234177298, "loss": 2.8035, "step": 5720 }, { "epoch": 1.987993103469421e-14, "grad_norm": 0.600119948387146, "learning_rate": 0.00012400784500488554, "loss": 2.8408, "step": 5730 }, { "epoch": 1.9914625504213745e-14, "grad_norm": 0.7033205628395081, "learning_rate": 0.00012400365785348875, "loss": 2.8637, "step": 5740 }, { "epoch": 1.9949319973733282e-14, "grad_norm": 0.6281517744064331, "learning_rate": 0.00012399946196420158, "loss": 2.8114, "step": 5750 }, { "epoch": 1.9984014443252818e-14, "grad_norm": 0.6357794404029846, "learning_rate": 0.0001239952573376876, "loss": 2.8168, "step": 5760 }, { "epoch": 2.0018708912772354e-14, "grad_norm": 0.6197851896286011, "learning_rate": 0.0001239910439746117, "loss": 2.8363, "step": 5770 }, { "epoch": 2.005340338229189e-14, "grad_norm": 0.6617996692657471, "learning_rate": 0.00012398682187564027, "loss": 2.8533, "step": 5780 }, { "epoch": 2.0088097851811426e-14, "grad_norm": 0.603771448135376, "learning_rate": 0.0001239825910414409, "loss": 2.8585, "step": 5790 }, { "epoch": 2.0122792321330962e-14, "grad_norm": 0.5812570452690125, "learning_rate": 0.00012397835147268273, "loss": 2.8288, "step": 5800 }, { "epoch": 2.01574867908505e-14, "grad_norm": 0.5876151919364929, "learning_rate": 0.00012397410317003618, "loss": 2.8128, "step": 5810 }, { "epoch": 2.0192181260370035e-14, "grad_norm": 0.6807602643966675, "learning_rate": 0.00012396984613417307, "loss": 2.8356, "step": 5820 }, { "epoch": 2.022687572988957e-14, "grad_norm": 0.639345645904541, "learning_rate": 0.00012396558036576662, "loss": 2.8469, "step": 5830 }, { "epoch": 2.0261570199409107e-14, "grad_norm": 0.5763149857521057, "learning_rate": 0.00012396130586549145, "loss": 2.8604, "step": 5840 }, { "epoch": 2.0296264668928643e-14, "grad_norm": 0.61338210105896, "learning_rate": 0.0001239570226340235, "loss": 2.7819, "step": 5850 }, { "epoch": 2.033095913844818e-14, "grad_norm": 0.6552673578262329, "learning_rate": 0.00012395273067204016, "loss": 2.8324, "step": 5860 }, { "epoch": 2.0365653607967715e-14, "grad_norm": 0.5748652815818787, "learning_rate": 0.00012394842998022018, "loss": 2.8241, "step": 5870 }, { "epoch": 2.040034807748725e-14, "grad_norm": 0.6022285223007202, "learning_rate": 0.00012394412055924362, "loss": 2.8576, "step": 5880 }, { "epoch": 2.0435042547006788e-14, "grad_norm": 0.6179032921791077, "learning_rate": 0.00012393980240979203, "loss": 2.8123, "step": 5890 }, { "epoch": 2.0469737016526324e-14, "grad_norm": 0.6196229457855225, "learning_rate": 0.00012393547553254825, "loss": 2.8278, "step": 5900 }, { "epoch": 2.050443148604586e-14, "grad_norm": 0.6168218851089478, "learning_rate": 0.00012393113992819656, "loss": 2.8137, "step": 5910 }, { "epoch": 2.0539125955565396e-14, "grad_norm": 0.610420823097229, "learning_rate": 0.00012392679559742258, "loss": 2.8339, "step": 5920 }, { "epoch": 2.0573820425084932e-14, "grad_norm": 0.5912334322929382, "learning_rate": 0.00012392244254091334, "loss": 2.781, "step": 5930 }, { "epoch": 2.0608514894604468e-14, "grad_norm": 0.5794424414634705, "learning_rate": 0.00012391808075935725, "loss": 2.8105, "step": 5940 }, { "epoch": 2.0643209364124004e-14, "grad_norm": 0.5825474858283997, "learning_rate": 0.00012391371025344408, "loss": 2.8061, "step": 5950 }, { "epoch": 2.067790383364354e-14, "grad_norm": 0.6089785099029541, "learning_rate": 0.00012390933102386496, "loss": 2.8504, "step": 5960 }, { "epoch": 2.0712598303163077e-14, "grad_norm": 0.6193978786468506, "learning_rate": 0.0001239049430713125, "loss": 2.8388, "step": 5970 }, { "epoch": 2.0747292772682613e-14, "grad_norm": 0.5813872814178467, "learning_rate": 0.00012390054639648053, "loss": 2.8201, "step": 5980 }, { "epoch": 2.078198724220215e-14, "grad_norm": 0.5832180380821228, "learning_rate": 0.0001238961410000644, "loss": 2.821, "step": 5990 }, { "epoch": 2.0816681711721685e-14, "grad_norm": 0.5708989500999451, "learning_rate": 0.0001238917268827608, "loss": 2.7599, "step": 6000 }, { "epoch": 2.085137618124122e-14, "grad_norm": 0.6428094506263733, "learning_rate": 0.0001238873040452677, "loss": 2.7868, "step": 6010 }, { "epoch": 2.0886070650760757e-14, "grad_norm": 0.6414648294448853, "learning_rate": 0.00012388287248828465, "loss": 2.7709, "step": 6020 }, { "epoch": 2.0920765120280294e-14, "grad_norm": 0.6225854754447937, "learning_rate": 0.00012387843221251237, "loss": 2.8044, "step": 6030 }, { "epoch": 2.095545958979983e-14, "grad_norm": 0.6116979718208313, "learning_rate": 0.00012387398321865307, "loss": 2.7937, "step": 6040 }, { "epoch": 2.0990154059319366e-14, "grad_norm": 0.6234277486801147, "learning_rate": 0.00012386952550741038, "loss": 2.7908, "step": 6050 }, { "epoch": 2.1024848528838902e-14, "grad_norm": 0.6102809906005859, "learning_rate": 0.00012386505907948918, "loss": 2.8163, "step": 6060 }, { "epoch": 2.1059542998358438e-14, "grad_norm": 0.608622670173645, "learning_rate": 0.0001238605839355958, "loss": 2.7869, "step": 6070 }, { "epoch": 2.1094237467877974e-14, "grad_norm": 0.6176615357398987, "learning_rate": 0.00012385610007643798, "loss": 2.8188, "step": 6080 }, { "epoch": 2.112893193739751e-14, "grad_norm": 0.6223378777503967, "learning_rate": 0.0001238516075027248, "loss": 2.8031, "step": 6090 }, { "epoch": 2.1163626406917047e-14, "grad_norm": 0.6040530204772949, "learning_rate": 0.00012384710621516672, "loss": 2.8022, "step": 6100 }, { "epoch": 2.1198320876436583e-14, "grad_norm": 0.6011039614677429, "learning_rate": 0.00012384259621447553, "loss": 2.7691, "step": 6110 }, { "epoch": 2.123301534595612e-14, "grad_norm": 0.5895322561264038, "learning_rate": 0.00012383807750136451, "loss": 2.7932, "step": 6120 }, { "epoch": 2.1267709815475655e-14, "grad_norm": 0.6420459747314453, "learning_rate": 0.00012383355007654825, "loss": 2.8098, "step": 6130 }, { "epoch": 2.130240428499519e-14, "grad_norm": 0.6566873788833618, "learning_rate": 0.00012382901394074268, "loss": 2.7858, "step": 6140 }, { "epoch": 2.1337098754514727e-14, "grad_norm": 0.5898364782333374, "learning_rate": 0.0001238244690946652, "loss": 2.7977, "step": 6150 }, { "epoch": 2.1371793224034263e-14, "grad_norm": 0.5980865359306335, "learning_rate": 0.0001238199155390345, "loss": 2.7549, "step": 6160 }, { "epoch": 2.14064876935538e-14, "grad_norm": 0.6302582025527954, "learning_rate": 0.00012381535327457072, "loss": 2.7859, "step": 6170 }, { "epoch": 2.1441182163073336e-14, "grad_norm": 0.5613805055618286, "learning_rate": 0.00012381078230199534, "loss": 2.774, "step": 6180 }, { "epoch": 2.1475876632592872e-14, "grad_norm": 0.5955912470817566, "learning_rate": 0.00012380620262203117, "loss": 2.7872, "step": 6190 }, { "epoch": 2.1510571102112408e-14, "grad_norm": 0.5910189151763916, "learning_rate": 0.00012380161423540252, "loss": 2.7723, "step": 6200 }, { "epoch": 2.1545265571631944e-14, "grad_norm": 0.584894597530365, "learning_rate": 0.00012379701714283493, "loss": 2.8089, "step": 6210 }, { "epoch": 2.157996004115148e-14, "grad_norm": 0.5542137622833252, "learning_rate": 0.00012379241134505544, "loss": 2.8126, "step": 6220 }, { "epoch": 2.1614654510671016e-14, "grad_norm": 0.5854094624519348, "learning_rate": 0.0001237877968427924, "loss": 2.7798, "step": 6230 }, { "epoch": 2.1649348980190553e-14, "grad_norm": 0.648425281047821, "learning_rate": 0.00012378317363677556, "loss": 2.7636, "step": 6240 }, { "epoch": 2.168404344971009e-14, "grad_norm": 0.5945795178413391, "learning_rate": 0.00012377854172773605, "loss": 2.7874, "step": 6250 }, { "epoch": 2.1718737919229625e-14, "grad_norm": 0.6185757517814636, "learning_rate": 0.00012377390111640636, "loss": 2.7791, "step": 6260 }, { "epoch": 2.175343238874916e-14, "grad_norm": 0.5691210627555847, "learning_rate": 0.00012376925180352034, "loss": 2.7999, "step": 6270 }, { "epoch": 2.1788126858268697e-14, "grad_norm": 0.6154516935348511, "learning_rate": 0.00012376459378981325, "loss": 2.7648, "step": 6280 }, { "epoch": 2.1822821327788233e-14, "grad_norm": 0.602311372756958, "learning_rate": 0.00012375992707602172, "loss": 2.8407, "step": 6290 }, { "epoch": 2.185751579730777e-14, "grad_norm": 0.6028624773025513, "learning_rate": 0.00012375525166288378, "loss": 2.813, "step": 6300 }, { "epoch": 2.1892210266827306e-14, "grad_norm": 0.5635526776313782, "learning_rate": 0.00012375056755113876, "loss": 2.7697, "step": 6310 }, { "epoch": 2.1926904736346842e-14, "grad_norm": 0.5648844242095947, "learning_rate": 0.00012374587474152742, "loss": 2.777, "step": 6320 }, { "epoch": 2.1961599205866378e-14, "grad_norm": 0.5999817252159119, "learning_rate": 0.0001237411732347919, "loss": 2.7845, "step": 6330 }, { "epoch": 2.1996293675385914e-14, "grad_norm": 0.5786688327789307, "learning_rate": 0.00012373646303167571, "loss": 2.7517, "step": 6340 }, { "epoch": 2.203098814490545e-14, "grad_norm": 0.5822632312774658, "learning_rate": 0.0001237317441329237, "loss": 2.8104, "step": 6350 }, { "epoch": 2.2065682614424986e-14, "grad_norm": 0.5697758793830872, "learning_rate": 0.00012372701653928217, "loss": 2.773, "step": 6360 }, { "epoch": 2.2100377083944522e-14, "grad_norm": 0.5976788997650146, "learning_rate": 0.00012372228025149873, "loss": 2.7856, "step": 6370 }, { "epoch": 2.213507155346406e-14, "grad_norm": 0.6103385090827942, "learning_rate": 0.00012371753527032234, "loss": 2.7375, "step": 6380 }, { "epoch": 2.2169766022983595e-14, "grad_norm": 0.6408196091651917, "learning_rate": 0.00012371278159650342, "loss": 2.7789, "step": 6390 }, { "epoch": 2.220446049250313e-14, "grad_norm": 0.5938745141029358, "learning_rate": 0.0001237080192307937, "loss": 2.7624, "step": 6400 }, { "epoch": 2.2239154962022667e-14, "grad_norm": 0.5632643103599548, "learning_rate": 0.00012370324817394634, "loss": 2.8005, "step": 6410 }, { "epoch": 2.2273849431542203e-14, "grad_norm": 0.6022831201553345, "learning_rate": 0.00012369846842671584, "loss": 2.7967, "step": 6420 }, { "epoch": 2.230854390106174e-14, "grad_norm": 0.5827109217643738, "learning_rate": 0.00012369367998985804, "loss": 2.7983, "step": 6430 }, { "epoch": 2.2343238370581275e-14, "grad_norm": 0.5579185485839844, "learning_rate": 0.00012368888286413016, "loss": 2.8037, "step": 6440 }, { "epoch": 2.237793284010081e-14, "grad_norm": 0.6156270503997803, "learning_rate": 0.00012368407705029094, "loss": 2.7951, "step": 6450 }, { "epoch": 2.2412627309620348e-14, "grad_norm": 0.6143196225166321, "learning_rate": 0.0001236792625491003, "loss": 2.7499, "step": 6460 }, { "epoch": 2.2447321779139884e-14, "grad_norm": 0.5745688676834106, "learning_rate": 0.0001236744393613196, "loss": 2.746, "step": 6470 }, { "epoch": 2.248201624865942e-14, "grad_norm": 0.6295422911643982, "learning_rate": 0.0001236696074877116, "loss": 2.7846, "step": 6480 }, { "epoch": 2.2516710718178956e-14, "grad_norm": 0.6230496764183044, "learning_rate": 0.00012366476692904043, "loss": 2.7298, "step": 6490 }, { "epoch": 2.2551405187698492e-14, "grad_norm": 0.6398041248321533, "learning_rate": 0.0001236599176860716, "loss": 2.7723, "step": 6500 }, { "epoch": 2.2586099657218028e-14, "grad_norm": 0.6491246819496155, "learning_rate": 0.00012365505975957196, "loss": 2.7881, "step": 6510 }, { "epoch": 2.2620794126737565e-14, "grad_norm": 0.6155674457550049, "learning_rate": 0.00012365019315030974, "loss": 2.8093, "step": 6520 }, { "epoch": 2.26554885962571e-14, "grad_norm": 0.630222737789154, "learning_rate": 0.00012364531785905453, "loss": 2.7238, "step": 6530 }, { "epoch": 2.2690183065776637e-14, "grad_norm": 0.5848940014839172, "learning_rate": 0.00012364043388657738, "loss": 2.7661, "step": 6540 }, { "epoch": 2.2724877535296173e-14, "grad_norm": 0.6199184656143188, "learning_rate": 0.0001236355412336506, "loss": 2.7645, "step": 6550 }, { "epoch": 2.275957200481571e-14, "grad_norm": 0.6436838507652283, "learning_rate": 0.00012363063990104793, "loss": 2.7458, "step": 6560 }, { "epoch": 2.2794266474335245e-14, "grad_norm": 0.6143272519111633, "learning_rate": 0.00012362572988954448, "loss": 2.8118, "step": 6570 }, { "epoch": 2.282896094385478e-14, "grad_norm": 0.5796267986297607, "learning_rate": 0.00012362081119991669, "loss": 2.7506, "step": 6580 }, { "epoch": 2.2863655413374318e-14, "grad_norm": 0.6154047846794128, "learning_rate": 0.00012361588383294245, "loss": 2.7583, "step": 6590 }, { "epoch": 2.2898349882893854e-14, "grad_norm": 0.5823302865028381, "learning_rate": 0.00012361094778940102, "loss": 2.7758, "step": 6600 }, { "epoch": 2.293304435241339e-14, "grad_norm": 0.597959578037262, "learning_rate": 0.00012360600307007288, "loss": 2.7891, "step": 6610 }, { "epoch": 2.2967738821932926e-14, "grad_norm": 0.5578334331512451, "learning_rate": 0.0001236010496757401, "loss": 2.7634, "step": 6620 }, { "epoch": 2.3002433291452462e-14, "grad_norm": 0.5919632315635681, "learning_rate": 0.00012359608760718595, "loss": 2.763, "step": 6630 }, { "epoch": 2.3037127760971998e-14, "grad_norm": 0.6613178253173828, "learning_rate": 0.00012359111686519516, "loss": 2.7496, "step": 6640 }, { "epoch": 2.3071822230491534e-14, "grad_norm": 0.5555776357650757, "learning_rate": 0.00012358613745055383, "loss": 2.8292, "step": 6650 }, { "epoch": 2.310651670001107e-14, "grad_norm": 0.6103131771087646, "learning_rate": 0.00012358114936404942, "loss": 2.7302, "step": 6660 }, { "epoch": 2.3141211169530607e-14, "grad_norm": 0.599263072013855, "learning_rate": 0.00012357615260647067, "loss": 2.7521, "step": 6670 }, { "epoch": 2.3175905639050143e-14, "grad_norm": 0.5573865175247192, "learning_rate": 0.00012357114717860787, "loss": 2.7692, "step": 6680 }, { "epoch": 2.321060010856968e-14, "grad_norm": 0.591641366481781, "learning_rate": 0.00012356613308125254, "loss": 2.7558, "step": 6690 }, { "epoch": 2.3245294578089215e-14, "grad_norm": 0.7042614221572876, "learning_rate": 0.00012356111031519763, "loss": 2.7208, "step": 6700 }, { "epoch": 2.327998904760875e-14, "grad_norm": 0.5957866907119751, "learning_rate": 0.00012355607888123743, "loss": 2.6941, "step": 6710 }, { "epoch": 2.3314683517128287e-14, "grad_norm": 0.5923128128051758, "learning_rate": 0.00012355103878016762, "loss": 2.7404, "step": 6720 }, { "epoch": 2.3349377986647823e-14, "grad_norm": 0.5869559049606323, "learning_rate": 0.00012354599001278526, "loss": 2.7705, "step": 6730 }, { "epoch": 2.338407245616736e-14, "grad_norm": 0.6349188089370728, "learning_rate": 0.0001235409325798888, "loss": 2.7812, "step": 6740 }, { "epoch": 2.3418766925686896e-14, "grad_norm": 0.5657222270965576, "learning_rate": 0.00012353586648227797, "loss": 2.7488, "step": 6750 }, { "epoch": 2.3453461395206432e-14, "grad_norm": 0.5683046579360962, "learning_rate": 0.00012353079172075396, "loss": 2.7788, "step": 6760 }, { "epoch": 2.3488155864725968e-14, "grad_norm": 0.5673462152481079, "learning_rate": 0.00012352570829611927, "loss": 2.7746, "step": 6770 }, { "epoch": 2.3522850334245504e-14, "grad_norm": 0.6285842061042786, "learning_rate": 0.00012352061620917788, "loss": 2.7744, "step": 6780 }, { "epoch": 2.355754480376504e-14, "grad_norm": 0.5994075536727905, "learning_rate": 0.00012351551546073495, "loss": 2.7494, "step": 6790 }, { "epoch": 2.3592239273284576e-14, "grad_norm": 0.579120397567749, "learning_rate": 0.0001235104060515972, "loss": 2.7581, "step": 6800 }, { "epoch": 2.3626933742804113e-14, "grad_norm": 0.5559671521186829, "learning_rate": 0.00012350528798257262, "loss": 2.7592, "step": 6810 }, { "epoch": 2.366162821232365e-14, "grad_norm": 0.5771757364273071, "learning_rate": 0.00012350016125447057, "loss": 2.7738, "step": 6820 }, { "epoch": 2.3696322681843185e-14, "grad_norm": 0.5947537422180176, "learning_rate": 0.00012349502586810182, "loss": 2.7781, "step": 6830 }, { "epoch": 2.373101715136272e-14, "grad_norm": 0.5973154306411743, "learning_rate": 0.00012348988182427848, "loss": 2.7644, "step": 6840 }, { "epoch": 2.3765711620882257e-14, "grad_norm": 0.5740877985954285, "learning_rate": 0.00012348472912381402, "loss": 2.7506, "step": 6850 }, { "epoch": 2.3800406090401793e-14, "grad_norm": 0.5710994005203247, "learning_rate": 0.00012347956776752333, "loss": 2.7282, "step": 6860 }, { "epoch": 2.383510055992133e-14, "grad_norm": 0.6055555939674377, "learning_rate": 0.0001234743977562226, "loss": 2.748, "step": 6870 }, { "epoch": 2.3869795029440866e-14, "grad_norm": 0.6032747030258179, "learning_rate": 0.0001234692190907294, "loss": 2.7718, "step": 6880 }, { "epoch": 2.3904489498960402e-14, "grad_norm": 0.5790963172912598, "learning_rate": 0.00012346403177186278, "loss": 2.7434, "step": 6890 }, { "epoch": 2.3939183968479938e-14, "grad_norm": 1.0198516845703125, "learning_rate": 0.000123458835800443, "loss": 2.7439, "step": 6900 }, { "epoch": 2.3973878437999474e-14, "grad_norm": 0.6181248426437378, "learning_rate": 0.0001234536311772918, "loss": 2.7246, "step": 6910 }, { "epoch": 2.400857290751901e-14, "grad_norm": 0.5797358155250549, "learning_rate": 0.0001234484179032322, "loss": 2.7425, "step": 6920 }, { "epoch": 2.4043267377038546e-14, "grad_norm": 0.5769255757331848, "learning_rate": 0.00012344319597908863, "loss": 2.7803, "step": 6930 }, { "epoch": 2.4077961846558082e-14, "grad_norm": 0.5734415650367737, "learning_rate": 0.00012343796540568693, "loss": 2.7299, "step": 6940 }, { "epoch": 2.411265631607762e-14, "grad_norm": 0.5857387185096741, "learning_rate": 0.00012343272618385427, "loss": 2.7619, "step": 6950 }, { "epoch": 2.4147350785597155e-14, "grad_norm": 0.646302342414856, "learning_rate": 0.00012342747831441917, "loss": 2.7442, "step": 6960 }, { "epoch": 2.418204525511669e-14, "grad_norm": 0.5781094431877136, "learning_rate": 0.00012342222179821156, "loss": 2.706, "step": 6970 }, { "epoch": 2.4216739724636227e-14, "grad_norm": 0.5467093586921692, "learning_rate": 0.00012341695663606268, "loss": 2.7291, "step": 6980 }, { "epoch": 2.4251434194155763e-14, "grad_norm": 0.572467029094696, "learning_rate": 0.00012341168282880516, "loss": 2.7462, "step": 6990 }, { "epoch": 2.42861286636753e-14, "grad_norm": 0.570917546749115, "learning_rate": 0.00012340640037727306, "loss": 2.736, "step": 7000 }, { "epoch": 2.4320823133194835e-14, "grad_norm": 0.5814106464385986, "learning_rate": 0.00012340110928230172, "loss": 2.735, "step": 7010 }, { "epoch": 2.4355517602714372e-14, "grad_norm": 0.5690702199935913, "learning_rate": 0.0001233958095447279, "loss": 2.7362, "step": 7020 }, { "epoch": 2.4390212072233908e-14, "grad_norm": 0.621868371963501, "learning_rate": 0.0001233905011653897, "loss": 2.7474, "step": 7030 }, { "epoch": 2.4424906541753444e-14, "grad_norm": 0.6047300696372986, "learning_rate": 0.00012338518414512655, "loss": 2.706, "step": 7040 }, { "epoch": 2.445960101127298e-14, "grad_norm": 0.6763775944709778, "learning_rate": 0.00012337985848477934, "loss": 2.7683, "step": 7050 }, { "epoch": 2.4494295480792516e-14, "grad_norm": 0.569762110710144, "learning_rate": 0.0001233745241851903, "loss": 2.7552, "step": 7060 }, { "epoch": 2.4528989950312052e-14, "grad_norm": 0.5659019351005554, "learning_rate": 0.00012336918124720295, "loss": 2.7482, "step": 7070 }, { "epoch": 2.456368441983159e-14, "grad_norm": 0.5698880553245544, "learning_rate": 0.00012336382967166224, "loss": 2.7362, "step": 7080 }, { "epoch": 2.4598378889351125e-14, "grad_norm": 0.5826675295829773, "learning_rate": 0.00012335846945941452, "loss": 2.7545, "step": 7090 }, { "epoch": 2.463307335887066e-14, "grad_norm": 0.5701731443405151, "learning_rate": 0.00012335310061130738, "loss": 2.7302, "step": 7100 }, { "epoch": 2.4667767828390197e-14, "grad_norm": 0.5824063420295715, "learning_rate": 0.00012334772312818995, "loss": 2.7653, "step": 7110 }, { "epoch": 2.4702462297909733e-14, "grad_norm": 0.5795401334762573, "learning_rate": 0.00012334233701091256, "loss": 2.7416, "step": 7120 }, { "epoch": 2.473715676742927e-14, "grad_norm": 0.5750593543052673, "learning_rate": 0.00012333694226032701, "loss": 2.7447, "step": 7130 }, { "epoch": 2.4771851236948805e-14, "grad_norm": 0.6570457220077515, "learning_rate": 0.00012333153887728645, "loss": 2.693, "step": 7140 }, { "epoch": 2.480654570646834e-14, "grad_norm": 0.6170352697372437, "learning_rate": 0.00012332612686264535, "loss": 2.7063, "step": 7150 }, { "epoch": 2.4841240175987878e-14, "grad_norm": 0.5621455311775208, "learning_rate": 0.00012332070621725956, "loss": 2.7277, "step": 7160 }, { "epoch": 2.4875934645507414e-14, "grad_norm": 0.6283536553382874, "learning_rate": 0.00012331527694198636, "loss": 2.6995, "step": 7170 }, { "epoch": 2.491062911502695e-14, "grad_norm": 0.5938434600830078, "learning_rate": 0.00012330983903768427, "loss": 2.7957, "step": 7180 }, { "epoch": 2.4945323584546486e-14, "grad_norm": 0.5632232427597046, "learning_rate": 0.00012330439250521333, "loss": 2.7403, "step": 7190 }, { "epoch": 2.4980018054066022e-14, "grad_norm": 0.6330715417861938, "learning_rate": 0.00012329893734543482, "loss": 2.7192, "step": 7200 }, { "epoch": 2.5014712523585558e-14, "grad_norm": 0.5441268086433411, "learning_rate": 0.0001232934735592114, "loss": 2.6966, "step": 7210 }, { "epoch": 2.5049406993105094e-14, "grad_norm": 0.5736047625541687, "learning_rate": 0.00012328800114740716, "loss": 2.7234, "step": 7220 }, { "epoch": 2.508410146262463e-14, "grad_norm": 0.5625598430633545, "learning_rate": 0.00012328252011088752, "loss": 2.7559, "step": 7230 }, { "epoch": 2.5118795932144167e-14, "grad_norm": 0.5650346875190735, "learning_rate": 0.00012327703045051923, "loss": 2.747, "step": 7240 }, { "epoch": 2.5153490401663703e-14, "grad_norm": 0.5809916257858276, "learning_rate": 0.00012327153216717048, "loss": 2.7553, "step": 7250 }, { "epoch": 2.518818487118324e-14, "grad_norm": 0.5478555560112, "learning_rate": 0.00012326602526171071, "loss": 2.7235, "step": 7260 }, { "epoch": 2.5222879340702775e-14, "grad_norm": 0.5453422665596008, "learning_rate": 0.00012326050973501083, "loss": 2.7345, "step": 7270 }, { "epoch": 2.525757381022231e-14, "grad_norm": 0.6005186438560486, "learning_rate": 0.00012325498558794307, "loss": 2.7067, "step": 7280 }, { "epoch": 2.5292268279741847e-14, "grad_norm": 0.5725247859954834, "learning_rate": 0.00012324945282138098, "loss": 2.738, "step": 7290 }, { "epoch": 2.5326962749261384e-14, "grad_norm": 0.5643709301948547, "learning_rate": 0.0001232439114361996, "loss": 2.7066, "step": 7300 }, { "epoch": 2.536165721878092e-14, "grad_norm": 0.5257111191749573, "learning_rate": 0.00012323836143327522, "loss": 2.7213, "step": 7310 }, { "epoch": 2.5396351688300456e-14, "grad_norm": 0.6025110483169556, "learning_rate": 0.0001232328028134855, "loss": 2.7572, "step": 7320 }, { "epoch": 2.5431046157819992e-14, "grad_norm": 0.5421791672706604, "learning_rate": 0.00012322723557770953, "loss": 2.7074, "step": 7330 }, { "epoch": 2.5465740627339528e-14, "grad_norm": 0.5835829377174377, "learning_rate": 0.00012322165972682766, "loss": 2.7329, "step": 7340 }, { "epoch": 2.5500435096859064e-14, "grad_norm": 0.5800181031227112, "learning_rate": 0.00012321607526172173, "loss": 2.6736, "step": 7350 }, { "epoch": 2.55351295663786e-14, "grad_norm": 0.5680585503578186, "learning_rate": 0.00012321048218327482, "loss": 2.7246, "step": 7360 }, { "epoch": 2.5569824035898137e-14, "grad_norm": 0.5471453666687012, "learning_rate": 0.00012320488049237147, "loss": 2.7152, "step": 7370 }, { "epoch": 2.5604518505417673e-14, "grad_norm": 0.5314862132072449, "learning_rate": 0.00012319927018989753, "loss": 2.6953, "step": 7380 }, { "epoch": 2.563921297493721e-14, "grad_norm": 0.6064063906669617, "learning_rate": 0.00012319365127674018, "loss": 2.7237, "step": 7390 }, { "epoch": 2.5673907444456745e-14, "grad_norm": 0.559303879737854, "learning_rate": 0.00012318802375378807, "loss": 2.7405, "step": 7400 }, { "epoch": 2.570860191397628e-14, "grad_norm": 0.5577559471130371, "learning_rate": 0.00012318238762193108, "loss": 2.7203, "step": 7410 }, { "epoch": 2.5743296383495817e-14, "grad_norm": 0.5860854983329773, "learning_rate": 0.00012317674288206057, "loss": 2.7036, "step": 7420 }, { "epoch": 2.5777990853015353e-14, "grad_norm": 0.6219355463981628, "learning_rate": 0.00012317108953506915, "loss": 2.7522, "step": 7430 }, { "epoch": 2.581268532253489e-14, "grad_norm": 0.5643932223320007, "learning_rate": 0.0001231654275818509, "loss": 2.7567, "step": 7440 }, { "epoch": 2.5847379792054426e-14, "grad_norm": 0.5741448402404785, "learning_rate": 0.0001231597570233012, "loss": 2.724, "step": 7450 }, { "epoch": 2.5882074261573962e-14, "grad_norm": 0.5942105650901794, "learning_rate": 0.00012315407786031677, "loss": 2.6867, "step": 7460 }, { "epoch": 2.5916768731093498e-14, "grad_norm": 0.5577355623245239, "learning_rate": 0.00012314839009379576, "loss": 2.6847, "step": 7470 }, { "epoch": 2.5951463200613034e-14, "grad_norm": 0.6128631830215454, "learning_rate": 0.0001231426937246376, "loss": 2.7059, "step": 7480 }, { "epoch": 2.598615767013257e-14, "grad_norm": 0.5741069316864014, "learning_rate": 0.00012313698875374314, "loss": 2.7302, "step": 7490 }, { "epoch": 2.6020852139652106e-14, "grad_norm": 0.6051983833312988, "learning_rate": 0.0001231312751820146, "loss": 2.715, "step": 7500 }, { "epoch": 2.6055546609171643e-14, "grad_norm": 0.5421214699745178, "learning_rate": 0.0001231255530103555, "loss": 2.7362, "step": 7510 }, { "epoch": 2.609024107869118e-14, "grad_norm": 0.584946870803833, "learning_rate": 0.00012311982223967076, "loss": 2.7274, "step": 7520 }, { "epoch": 2.6124935548210715e-14, "grad_norm": 0.6235318779945374, "learning_rate": 0.00012311408287086662, "loss": 2.7205, "step": 7530 }, { "epoch": 2.615963001773025e-14, "grad_norm": 0.6149477362632751, "learning_rate": 0.0001231083349048508, "loss": 2.73, "step": 7540 }, { "epoch": 2.6194324487249787e-14, "grad_norm": 0.5890522003173828, "learning_rate": 0.00012310257834253216, "loss": 2.725, "step": 7550 }, { "epoch": 2.6229018956769323e-14, "grad_norm": 0.5587625503540039, "learning_rate": 0.00012309681318482119, "loss": 2.7054, "step": 7560 }, { "epoch": 2.626371342628886e-14, "grad_norm": 0.6507006883621216, "learning_rate": 0.0001230910394326295, "loss": 2.6922, "step": 7570 }, { "epoch": 2.6298407895808396e-14, "grad_norm": 0.5716079473495483, "learning_rate": 0.0001230852570868702, "loss": 2.7204, "step": 7580 }, { "epoch": 2.6333102365327932e-14, "grad_norm": 0.5947009325027466, "learning_rate": 0.00012307946614845773, "loss": 2.7619, "step": 7590 }, { "epoch": 2.6367796834847468e-14, "grad_norm": 0.5918928980827332, "learning_rate": 0.00012307366661830786, "loss": 2.7096, "step": 7600 }, { "epoch": 2.6402491304367004e-14, "grad_norm": 0.6017585396766663, "learning_rate": 0.0001230678584973377, "loss": 2.7274, "step": 7610 }, { "epoch": 2.643718577388654e-14, "grad_norm": 0.5627968907356262, "learning_rate": 0.00012306204178646584, "loss": 2.7079, "step": 7620 }, { "epoch": 2.6471880243406076e-14, "grad_norm": 0.5690865516662598, "learning_rate": 0.00012305621648661205, "loss": 2.7091, "step": 7630 }, { "epoch": 2.6506574712925612e-14, "grad_norm": 0.5826201438903809, "learning_rate": 0.00012305038259869763, "loss": 2.7154, "step": 7640 }, { "epoch": 2.654126918244515e-14, "grad_norm": 0.5735682845115662, "learning_rate": 0.00012304454012364508, "loss": 2.7265, "step": 7650 }, { "epoch": 2.6575963651964685e-14, "grad_norm": 0.5647796988487244, "learning_rate": 0.00012303868906237843, "loss": 2.6855, "step": 7660 }, { "epoch": 2.661065812148422e-14, "grad_norm": 0.5775138139724731, "learning_rate": 0.0001230328294158229, "loss": 2.7063, "step": 7670 }, { "epoch": 2.6645352591003757e-14, "grad_norm": 0.5702304840087891, "learning_rate": 0.00012302696118490514, "loss": 2.6862, "step": 7680 }, { "epoch": 2.6680047060523293e-14, "grad_norm": 0.5837470293045044, "learning_rate": 0.00012302108437055323, "loss": 2.6967, "step": 7690 }, { "epoch": 2.671474153004283e-14, "grad_norm": 0.6450745463371277, "learning_rate": 0.0001230151989736965, "loss": 2.7094, "step": 7700 }, { "epoch": 2.6749435999562365e-14, "grad_norm": 0.5832573175430298, "learning_rate": 0.00012300930499526563, "loss": 2.7016, "step": 7710 }, { "epoch": 2.67841304690819e-14, "grad_norm": 0.5531536340713501, "learning_rate": 0.00012300340243619277, "loss": 2.7232, "step": 7720 }, { "epoch": 2.6818824938601438e-14, "grad_norm": 0.6044182777404785, "learning_rate": 0.00012299749129741132, "loss": 2.6943, "step": 7730 }, { "epoch": 2.6853519408120974e-14, "grad_norm": 0.5934939980506897, "learning_rate": 0.0001229915715798561, "loss": 2.7138, "step": 7740 }, { "epoch": 2.688821387764051e-14, "grad_norm": 0.5704272389411926, "learning_rate": 0.00012298564328446323, "loss": 2.6902, "step": 7750 }, { "epoch": 2.6922908347160046e-14, "grad_norm": 0.5770971179008484, "learning_rate": 0.00012297970641217027, "loss": 2.7127, "step": 7760 }, { "epoch": 2.6957602816679582e-14, "grad_norm": 0.5804921984672546, "learning_rate": 0.00012297376096391603, "loss": 2.7327, "step": 7770 }, { "epoch": 2.6992297286199118e-14, "grad_norm": 0.6150832176208496, "learning_rate": 0.00012296780694064075, "loss": 2.7028, "step": 7780 }, { "epoch": 2.7026991755718655e-14, "grad_norm": 0.6145310401916504, "learning_rate": 0.00012296184434328603, "loss": 2.7334, "step": 7790 }, { "epoch": 2.706168622523819e-14, "grad_norm": 0.5813900828361511, "learning_rate": 0.00012295587317279476, "loss": 2.7214, "step": 7800 }, { "epoch": 2.7096380694757727e-14, "grad_norm": 0.5688285231590271, "learning_rate": 0.00012294989343011127, "loss": 2.7155, "step": 7810 }, { "epoch": 2.7131075164277263e-14, "grad_norm": 0.568423330783844, "learning_rate": 0.0001229439051161812, "loss": 2.7214, "step": 7820 }, { "epoch": 2.71657696337968e-14, "grad_norm": 0.6008029580116272, "learning_rate": 0.00012293790823195153, "loss": 2.7187, "step": 7830 }, { "epoch": 2.7200464103316335e-14, "grad_norm": 0.5742303133010864, "learning_rate": 0.00012293190277837062, "loss": 2.6822, "step": 7840 }, { "epoch": 2.723515857283587e-14, "grad_norm": 0.5873542428016663, "learning_rate": 0.0001229258887563882, "loss": 2.7041, "step": 7850 }, { "epoch": 2.7269853042355408e-14, "grad_norm": 0.6187730431556702, "learning_rate": 0.0001229198661669553, "loss": 2.7311, "step": 7860 }, { "epoch": 2.7304547511874944e-14, "grad_norm": 0.5639210939407349, "learning_rate": 0.0001229138350110244, "loss": 2.6857, "step": 7870 }, { "epoch": 2.733924198139448e-14, "grad_norm": 0.5946155786514282, "learning_rate": 0.00012290779528954923, "loss": 2.6811, "step": 7880 }, { "epoch": 2.7373936450914016e-14, "grad_norm": 0.6033107042312622, "learning_rate": 0.0001229017470034849, "loss": 2.7342, "step": 7890 }, { "epoch": 2.7408630920433552e-14, "grad_norm": 0.551990270614624, "learning_rate": 0.00012289569015378794, "loss": 2.7082, "step": 7900 }, { "epoch": 2.7443325389953088e-14, "grad_norm": 0.5572894215583801, "learning_rate": 0.00012288962474141615, "loss": 2.7065, "step": 7910 }, { "epoch": 2.7478019859472624e-14, "grad_norm": 0.5644770860671997, "learning_rate": 0.0001228835507673288, "loss": 2.6439, "step": 7920 }, { "epoch": 2.751271432899216e-14, "grad_norm": 0.5705562233924866, "learning_rate": 0.00012287746823248634, "loss": 2.7343, "step": 7930 }, { "epoch": 2.7547408798511697e-14, "grad_norm": 0.5350631475448608, "learning_rate": 0.00012287137713785072, "loss": 2.7196, "step": 7940 }, { "epoch": 2.7582103268031233e-14, "grad_norm": 0.5523934364318848, "learning_rate": 0.00012286527748438516, "loss": 2.7396, "step": 7950 }, { "epoch": 2.761679773755077e-14, "grad_norm": 0.5986164212226868, "learning_rate": 0.00012285916927305432, "loss": 2.6991, "step": 7960 }, { "epoch": 2.7651492207070305e-14, "grad_norm": 0.5188223719596863, "learning_rate": 0.00012285305250482414, "loss": 2.7232, "step": 7970 }, { "epoch": 2.768618667658984e-14, "grad_norm": 0.5866278409957886, "learning_rate": 0.00012284692718066191, "loss": 2.6461, "step": 7980 }, { "epoch": 2.7720881146109377e-14, "grad_norm": 0.6075391173362732, "learning_rate": 0.00012284079330153634, "loss": 2.7117, "step": 7990 }, { "epoch": 2.7755575615628914e-14, "grad_norm": 0.5969737768173218, "learning_rate": 0.0001228346508684174, "loss": 2.6775, "step": 8000 }, { "epoch": 2.779027008514845e-14, "grad_norm": 0.5980135798454285, "learning_rate": 0.0001228284998822765, "loss": 2.6846, "step": 8010 }, { "epoch": 2.7824964554667986e-14, "grad_norm": 0.5240525603294373, "learning_rate": 0.00012282234034408635, "loss": 2.7148, "step": 8020 }, { "epoch": 2.7859659024187522e-14, "grad_norm": 0.5938628911972046, "learning_rate": 0.00012281617225482104, "loss": 2.7032, "step": 8030 }, { "epoch": 2.7894353493707058e-14, "grad_norm": 0.5991101861000061, "learning_rate": 0.00012280999561545597, "loss": 2.6775, "step": 8040 }, { "epoch": 2.7929047963226594e-14, "grad_norm": 0.6069301962852478, "learning_rate": 0.000122803810426968, "loss": 2.6744, "step": 8050 }, { "epoch": 2.796374243274613e-14, "grad_norm": 0.588005781173706, "learning_rate": 0.00012279761669033517, "loss": 2.701, "step": 8060 }, { "epoch": 2.7998436902265667e-14, "grad_norm": 0.5970097184181213, "learning_rate": 0.000122791414406537, "loss": 2.6829, "step": 8070 }, { "epoch": 2.8033131371785203e-14, "grad_norm": 0.5444302558898926, "learning_rate": 0.00012278520357655433, "loss": 2.6926, "step": 8080 }, { "epoch": 2.806782584130474e-14, "grad_norm": 0.5803437829017639, "learning_rate": 0.0001227789842013694, "loss": 2.7117, "step": 8090 }, { "epoch": 2.8102520310824275e-14, "grad_norm": 0.5844326019287109, "learning_rate": 0.00012277275628196565, "loss": 2.6914, "step": 8100 }, { "epoch": 2.813721478034381e-14, "grad_norm": 0.5499669313430786, "learning_rate": 0.00012276651981932805, "loss": 2.6675, "step": 8110 }, { "epoch": 2.8171909249863347e-14, "grad_norm": 0.5634953379631042, "learning_rate": 0.00012276027481444282, "loss": 2.6827, "step": 8120 }, { "epoch": 2.8206603719382883e-14, "grad_norm": 0.5684443712234497, "learning_rate": 0.00012275402126829752, "loss": 2.6976, "step": 8130 }, { "epoch": 2.824129818890242e-14, "grad_norm": 0.592460572719574, "learning_rate": 0.00012274775918188116, "loss": 2.6713, "step": 8140 }, { "epoch": 2.8275992658421956e-14, "grad_norm": 0.5342685580253601, "learning_rate": 0.000122741488556184, "loss": 2.6415, "step": 8150 }, { "epoch": 2.8310687127941492e-14, "grad_norm": 0.5845648646354675, "learning_rate": 0.00012273520939219768, "loss": 2.74, "step": 8160 }, { "epoch": 2.8345381597461028e-14, "grad_norm": 0.5626019835472107, "learning_rate": 0.0001227289216909152, "loss": 2.688, "step": 8170 }, { "epoch": 2.8380076066980564e-14, "grad_norm": 0.6154710650444031, "learning_rate": 0.0001227226254533309, "loss": 2.6612, "step": 8180 }, { "epoch": 2.84147705365001e-14, "grad_norm": 0.5516773462295532, "learning_rate": 0.0001227163206804405, "loss": 2.7218, "step": 8190 }, { "epoch": 2.8449465006019636e-14, "grad_norm": 0.5680599808692932, "learning_rate": 0.00012271000737324104, "loss": 2.6318, "step": 8200 }, { "epoch": 2.848415947553917e-14, "grad_norm": 0.6498519778251648, "learning_rate": 0.00012270368553273086, "loss": 2.7052, "step": 8210 }, { "epoch": 2.851885394505871e-14, "grad_norm": 0.571983814239502, "learning_rate": 0.0001226973551599098, "loss": 2.6634, "step": 8220 }, { "epoch": 2.8553548414578245e-14, "grad_norm": 0.5949233770370483, "learning_rate": 0.00012269101625577886, "loss": 2.7187, "step": 8230 }, { "epoch": 2.858824288409778e-14, "grad_norm": 0.5886818766593933, "learning_rate": 0.00012268466882134056, "loss": 2.6904, "step": 8240 }, { "epoch": 2.862293735361732e-14, "grad_norm": 0.8348990678787231, "learning_rate": 0.00012267831285759864, "loss": 2.6906, "step": 8250 }, { "epoch": 2.8657631823136853e-14, "grad_norm": 0.5469231605529785, "learning_rate": 0.00012267194836555826, "loss": 2.6787, "step": 8260 }, { "epoch": 2.869232629265639e-14, "grad_norm": 0.5840815305709839, "learning_rate": 0.00012266557534622592, "loss": 2.6876, "step": 8270 }, { "epoch": 2.8727020762175925e-14, "grad_norm": 0.5527530312538147, "learning_rate": 0.00012265919380060946, "loss": 2.725, "step": 8280 }, { "epoch": 2.876171523169546e-14, "grad_norm": 0.5877522230148315, "learning_rate": 0.00012265280372971804, "loss": 2.674, "step": 8290 }, { "epoch": 2.8796409701215e-14, "grad_norm": 0.6059237718582153, "learning_rate": 0.0001226464051345622, "loss": 2.6989, "step": 8300 }, { "epoch": 2.8831104170734534e-14, "grad_norm": 0.5534751415252686, "learning_rate": 0.00012263999801615384, "loss": 2.6798, "step": 8310 }, { "epoch": 2.886579864025407e-14, "grad_norm": 0.5741186141967773, "learning_rate": 0.00012263358237550617, "loss": 2.6826, "step": 8320 }, { "epoch": 2.8900493109773606e-14, "grad_norm": 0.5818377733230591, "learning_rate": 0.0001226271582136338, "loss": 2.7183, "step": 8330 }, { "epoch": 2.893518757929314e-14, "grad_norm": 0.5563935041427612, "learning_rate": 0.0001226207255315526, "loss": 2.7142, "step": 8340 }, { "epoch": 2.896988204881268e-14, "grad_norm": 0.5422356128692627, "learning_rate": 0.00012261428433027992, "loss": 2.6905, "step": 8350 }, { "epoch": 2.9004576518332215e-14, "grad_norm": 0.5527451038360596, "learning_rate": 0.00012260783461083433, "loss": 2.7234, "step": 8360 }, { "epoch": 2.903927098785175e-14, "grad_norm": 0.6017501354217529, "learning_rate": 0.0001226013763742358, "loss": 2.7047, "step": 8370 }, { "epoch": 2.907396545737129e-14, "grad_norm": 0.5606989860534668, "learning_rate": 0.00012259490962150565, "loss": 2.6873, "step": 8380 }, { "epoch": 2.910865992689082e-14, "grad_norm": 0.5723716616630554, "learning_rate": 0.00012258843435366656, "loss": 2.6908, "step": 8390 }, { "epoch": 2.914335439641036e-14, "grad_norm": 0.6364874243736267, "learning_rate": 0.0001225819505717425, "loss": 2.661, "step": 8400 }, { "epoch": 2.9178048865929895e-14, "grad_norm": 0.5418698191642761, "learning_rate": 0.00012257545827675886, "loss": 2.6914, "step": 8410 }, { "epoch": 2.921274333544943e-14, "grad_norm": 0.5674570798873901, "learning_rate": 0.00012256895746974234, "loss": 2.6632, "step": 8420 }, { "epoch": 2.924743780496897e-14, "grad_norm": 0.5606125593185425, "learning_rate": 0.00012256244815172092, "loss": 2.6258, "step": 8430 }, { "epoch": 2.9282132274488504e-14, "grad_norm": 0.5548431277275085, "learning_rate": 0.00012255593032372409, "loss": 2.6939, "step": 8440 }, { "epoch": 2.931682674400804e-14, "grad_norm": 0.5820953249931335, "learning_rate": 0.00012254940398678254, "loss": 2.6809, "step": 8450 }, { "epoch": 2.9351521213527576e-14, "grad_norm": 0.5409359931945801, "learning_rate": 0.00012254286914192836, "loss": 2.6734, "step": 8460 }, { "epoch": 2.938621568304711e-14, "grad_norm": 0.5668644309043884, "learning_rate": 0.00012253632579019497, "loss": 2.6974, "step": 8470 }, { "epoch": 2.942091015256665e-14, "grad_norm": 0.610866904258728, "learning_rate": 0.00012252977393261715, "loss": 2.6881, "step": 8480 }, { "epoch": 2.9455604622086184e-14, "grad_norm": 0.5846846699714661, "learning_rate": 0.000122523213570231, "loss": 2.6332, "step": 8490 }, { "epoch": 2.949029909160572e-14, "grad_norm": 0.5613436698913574, "learning_rate": 0.00012251664470407404, "loss": 2.7131, "step": 8500 }, { "epoch": 2.9524993561125257e-14, "grad_norm": 0.5695031881332397, "learning_rate": 0.00012251006733518503, "loss": 2.67, "step": 8510 }, { "epoch": 2.955968803064479e-14, "grad_norm": 0.5879989862442017, "learning_rate": 0.00012250348146460415, "loss": 2.6649, "step": 8520 }, { "epoch": 2.959438250016433e-14, "grad_norm": 0.5663895010948181, "learning_rate": 0.0001224968870933729, "loss": 2.6996, "step": 8530 }, { "epoch": 2.9629076969683865e-14, "grad_norm": 0.5668023228645325, "learning_rate": 0.0001224902842225341, "loss": 2.6843, "step": 8540 }, { "epoch": 2.96637714392034e-14, "grad_norm": 0.5497010946273804, "learning_rate": 0.00012248367285313193, "loss": 2.6652, "step": 8550 }, { "epoch": 2.969846590872294e-14, "grad_norm": 0.5846459865570068, "learning_rate": 0.00012247705298621198, "loss": 2.6673, "step": 8560 }, { "epoch": 2.9733160378242474e-14, "grad_norm": 0.5710161924362183, "learning_rate": 0.00012247042462282107, "loss": 2.6924, "step": 8570 }, { "epoch": 2.976785484776201e-14, "grad_norm": 0.5730729103088379, "learning_rate": 0.00012246378776400744, "loss": 2.646, "step": 8580 }, { "epoch": 2.9802549317281546e-14, "grad_norm": 0.5892428755760193, "learning_rate": 0.00012245714241082063, "loss": 2.7087, "step": 8590 }, { "epoch": 2.983724378680108e-14, "grad_norm": 0.5558162927627563, "learning_rate": 0.00012245048856431156, "loss": 2.6776, "step": 8600 }, { "epoch": 2.987193825632062e-14, "grad_norm": 0.5673137903213501, "learning_rate": 0.0001224438262255325, "loss": 2.6224, "step": 8610 }, { "epoch": 2.9906632725840154e-14, "grad_norm": 0.5773476958274841, "learning_rate": 0.00012243715539553702, "loss": 2.649, "step": 8620 }, { "epoch": 2.994132719535969e-14, "grad_norm": 0.5818929076194763, "learning_rate": 0.00012243047607538005, "loss": 2.6584, "step": 8630 }, { "epoch": 2.9976021664879227e-14, "grad_norm": 0.5667370557785034, "learning_rate": 0.00012242378826611786, "loss": 2.6612, "step": 8640 }, { "epoch": 3.001071613439876e-14, "grad_norm": 0.5747900605201721, "learning_rate": 0.0001224170919688081, "loss": 2.6718, "step": 8650 }, { "epoch": 3.00454106039183e-14, "grad_norm": 0.5726996660232544, "learning_rate": 0.0001224103871845097, "loss": 2.707, "step": 8660 }, { "epoch": 3.0080105073437835e-14, "grad_norm": 0.5508043766021729, "learning_rate": 0.000122403673914283, "loss": 2.6401, "step": 8670 }, { "epoch": 3.011479954295737e-14, "grad_norm": 0.583243191242218, "learning_rate": 0.0001223969521591896, "loss": 2.6221, "step": 8680 }, { "epoch": 3.014949401247691e-14, "grad_norm": 0.5679004192352295, "learning_rate": 0.00012239022192029253, "loss": 2.7069, "step": 8690 }, { "epoch": 3.0184188481996443e-14, "grad_norm": 0.6159049868583679, "learning_rate": 0.0001223834831986561, "loss": 2.6701, "step": 8700 }, { "epoch": 3.021888295151598e-14, "grad_norm": 0.5989751219749451, "learning_rate": 0.00012237673599534596, "loss": 2.7013, "step": 8710 }, { "epoch": 3.0253577421035516e-14, "grad_norm": 0.5724371075630188, "learning_rate": 0.00012236998031142917, "loss": 2.6607, "step": 8720 }, { "epoch": 3.028827189055505e-14, "grad_norm": 0.5319761633872986, "learning_rate": 0.00012236321614797407, "loss": 2.5941, "step": 8730 }, { "epoch": 3.032296636007459e-14, "grad_norm": 0.542374312877655, "learning_rate": 0.00012235644350605033, "loss": 2.6759, "step": 8740 }, { "epoch": 3.0357660829594124e-14, "grad_norm": 0.54795902967453, "learning_rate": 0.00012234966238672897, "loss": 2.7001, "step": 8750 }, { "epoch": 3.039235529911366e-14, "grad_norm": 0.5563028454780579, "learning_rate": 0.0001223428727910824, "loss": 2.6897, "step": 8760 }, { "epoch": 3.0427049768633196e-14, "grad_norm": 0.5498706102371216, "learning_rate": 0.00012233607472018437, "loss": 2.6805, "step": 8770 }, { "epoch": 3.046174423815273e-14, "grad_norm": 0.5624290704727173, "learning_rate": 0.00012232926817510987, "loss": 2.6746, "step": 8780 }, { "epoch": 3.049643870767227e-14, "grad_norm": 0.5394318699836731, "learning_rate": 0.0001223224531569353, "loss": 2.598, "step": 8790 }, { "epoch": 3.0531133177191805e-14, "grad_norm": 0.5328729748725891, "learning_rate": 0.00012231562966673847, "loss": 2.6754, "step": 8800 }, { "epoch": 3.056582764671134e-14, "grad_norm": 1.2187495231628418, "learning_rate": 0.00012230879770559834, "loss": 2.6881, "step": 8810 }, { "epoch": 3.060052211623088e-14, "grad_norm": 0.6069213151931763, "learning_rate": 0.00012230195727459543, "loss": 2.6864, "step": 8820 }, { "epoch": 3.0635216585750413e-14, "grad_norm": 0.55827397108078, "learning_rate": 0.00012229510837481145, "loss": 2.673, "step": 8830 }, { "epoch": 3.066991105526995e-14, "grad_norm": 0.5615711808204651, "learning_rate": 0.0001222882510073295, "loss": 2.6563, "step": 8840 }, { "epoch": 3.0704605524789486e-14, "grad_norm": 0.5916841626167297, "learning_rate": 0.00012228138517323398, "loss": 2.7015, "step": 8850 }, { "epoch": 3.073929999430902e-14, "grad_norm": 0.5886059403419495, "learning_rate": 0.00012227451087361074, "loss": 2.6472, "step": 8860 }, { "epoch": 3.077399446382856e-14, "grad_norm": 0.5575699806213379, "learning_rate": 0.00012226762810954684, "loss": 2.6831, "step": 8870 }, { "epoch": 3.0808688933348094e-14, "grad_norm": 0.5235202312469482, "learning_rate": 0.00012226073688213074, "loss": 2.6963, "step": 8880 }, { "epoch": 3.084338340286763e-14, "grad_norm": 0.551873505115509, "learning_rate": 0.0001222538371924522, "loss": 2.6238, "step": 8890 }, { "epoch": 3.0878077872387166e-14, "grad_norm": 0.551823616027832, "learning_rate": 0.0001222469290416024, "loss": 2.6597, "step": 8900 }, { "epoch": 3.09127723419067e-14, "grad_norm": 0.5748608112335205, "learning_rate": 0.00012224001243067375, "loss": 2.6428, "step": 8910 }, { "epoch": 3.094746681142624e-14, "grad_norm": 0.5467037558555603, "learning_rate": 0.00012223308736076009, "loss": 2.6774, "step": 8920 }, { "epoch": 3.0982161280945775e-14, "grad_norm": 0.5781031847000122, "learning_rate": 0.00012222615383295656, "loss": 2.6604, "step": 8930 }, { "epoch": 3.101685575046531e-14, "grad_norm": 0.5848052501678467, "learning_rate": 0.00012221921184835962, "loss": 2.6239, "step": 8940 }, { "epoch": 3.105155021998485e-14, "grad_norm": 0.5537866353988647, "learning_rate": 0.0001222122614080671, "loss": 2.6, "step": 8950 }, { "epoch": 3.108624468950438e-14, "grad_norm": 0.5576179027557373, "learning_rate": 0.00012220530251317812, "loss": 2.6444, "step": 8960 }, { "epoch": 3.112093915902392e-14, "grad_norm": 0.5529995560646057, "learning_rate": 0.00012219833516479323, "loss": 2.6658, "step": 8970 }, { "epoch": 3.1155633628543455e-14, "grad_norm": 0.5421051979064941, "learning_rate": 0.0001221913593640142, "loss": 2.6503, "step": 8980 }, { "epoch": 3.119032809806299e-14, "grad_norm": 0.5637949705123901, "learning_rate": 0.00012218437511194423, "loss": 2.6205, "step": 8990 }, { "epoch": 3.122502256758253e-14, "grad_norm": 0.5615357756614685, "learning_rate": 0.0001221773824096878, "loss": 2.6471, "step": 9000 }, { "epoch": 3.1259717037102064e-14, "grad_norm": 0.5570152997970581, "learning_rate": 0.00012217038125835071, "loss": 2.6795, "step": 9010 }, { "epoch": 3.12944115066216e-14, "grad_norm": 0.6177634000778198, "learning_rate": 0.0001221633716590402, "loss": 2.7038, "step": 9020 }, { "epoch": 3.1329105976141136e-14, "grad_norm": 0.5722036957740784, "learning_rate": 0.00012215635361286474, "loss": 2.6983, "step": 9030 }, { "epoch": 3.136380044566067e-14, "grad_norm": 0.5542957782745361, "learning_rate": 0.00012214932712093418, "loss": 2.6767, "step": 9040 }, { "epoch": 3.139849491518021e-14, "grad_norm": 0.5635998249053955, "learning_rate": 0.0001221422921843597, "loss": 2.6891, "step": 9050 }, { "epoch": 3.1433189384699745e-14, "grad_norm": 0.5611237287521362, "learning_rate": 0.00012213524880425383, "loss": 2.702, "step": 9060 }, { "epoch": 3.146788385421928e-14, "grad_norm": 0.5180429220199585, "learning_rate": 0.0001221281969817304, "loss": 2.6863, "step": 9070 }, { "epoch": 3.1502578323738817e-14, "grad_norm": 0.5857835412025452, "learning_rate": 0.0001221211367179046, "loss": 2.6354, "step": 9080 }, { "epoch": 3.153727279325835e-14, "grad_norm": 0.5531266927719116, "learning_rate": 0.00012211406801389293, "loss": 2.6281, "step": 9090 }, { "epoch": 3.157196726277789e-14, "grad_norm": 0.5656772255897522, "learning_rate": 0.0001221069908708133, "loss": 2.6601, "step": 9100 }, { "epoch": 3.1606661732297425e-14, "grad_norm": 0.5882046818733215, "learning_rate": 0.0001220999052897848, "loss": 2.6608, "step": 9110 }, { "epoch": 3.164135620181696e-14, "grad_norm": 0.5936713814735413, "learning_rate": 0.00012209281127192808, "loss": 2.6473, "step": 9120 }, { "epoch": 3.16760506713365e-14, "grad_norm": 0.5584852695465088, "learning_rate": 0.00012208570881836492, "loss": 2.6559, "step": 9130 }, { "epoch": 3.1710745140856034e-14, "grad_norm": 0.5620651245117188, "learning_rate": 0.00012207859793021851, "loss": 2.6619, "step": 9140 }, { "epoch": 3.174543961037557e-14, "grad_norm": 0.549185037612915, "learning_rate": 0.00012207147860861343, "loss": 2.6756, "step": 9150 }, { "epoch": 3.1780134079895106e-14, "grad_norm": 0.580562949180603, "learning_rate": 0.00012206435085467546, "loss": 2.6502, "step": 9160 }, { "epoch": 3.181482854941464e-14, "grad_norm": 0.5719506144523621, "learning_rate": 0.00012205721466953188, "loss": 2.6266, "step": 9170 }, { "epoch": 3.184952301893418e-14, "grad_norm": 0.5752028822898865, "learning_rate": 0.00012205007005431113, "loss": 2.6668, "step": 9180 }, { "epoch": 3.1884217488453714e-14, "grad_norm": 0.5765613913536072, "learning_rate": 0.00012204291701014312, "loss": 2.6726, "step": 9190 }, { "epoch": 3.191891195797325e-14, "grad_norm": 0.6549952030181885, "learning_rate": 0.00012203575553815905, "loss": 2.6123, "step": 9200 }, { "epoch": 3.1953606427492787e-14, "grad_norm": 0.5555936694145203, "learning_rate": 0.00012202858563949141, "loss": 2.6564, "step": 9210 }, { "epoch": 3.198830089701232e-14, "grad_norm": 0.540915846824646, "learning_rate": 0.0001220214073152741, "loss": 2.6215, "step": 9220 }, { "epoch": 3.202299536653186e-14, "grad_norm": 0.5766614079475403, "learning_rate": 0.00012201422056664228, "loss": 2.635, "step": 9230 }, { "epoch": 3.2057689836051395e-14, "grad_norm": 0.5567623376846313, "learning_rate": 0.00012200702539473246, "loss": 2.6599, "step": 9240 }, { "epoch": 3.209238430557093e-14, "grad_norm": 0.5546400547027588, "learning_rate": 0.00012199982180068253, "loss": 2.6233, "step": 9250 }, { "epoch": 3.212707877509047e-14, "grad_norm": 0.5220249891281128, "learning_rate": 0.00012199260978563165, "loss": 2.6335, "step": 9260 }, { "epoch": 3.2161773244610004e-14, "grad_norm": 0.5695894956588745, "learning_rate": 0.00012198538935072034, "loss": 2.6441, "step": 9270 }, { "epoch": 3.219646771412954e-14, "grad_norm": 0.5464606881141663, "learning_rate": 0.00012197816049709049, "loss": 2.6283, "step": 9280 }, { "epoch": 3.2231162183649076e-14, "grad_norm": 0.5551155805587769, "learning_rate": 0.00012197092322588523, "loss": 2.6417, "step": 9290 }, { "epoch": 3.226585665316861e-14, "grad_norm": 0.5350834727287292, "learning_rate": 0.00012196367753824908, "loss": 2.6507, "step": 9300 }, { "epoch": 3.230055112268815e-14, "grad_norm": 0.5904157757759094, "learning_rate": 0.0001219564234353279, "loss": 2.6422, "step": 9310 }, { "epoch": 3.2335245592207684e-14, "grad_norm": 0.5731456875801086, "learning_rate": 0.00012194916091826886, "loss": 2.6211, "step": 9320 }, { "epoch": 3.236994006172722e-14, "grad_norm": 0.5498260259628296, "learning_rate": 0.00012194188998822047, "loss": 2.6377, "step": 9330 }, { "epoch": 3.2404634531246757e-14, "grad_norm": 0.5684872269630432, "learning_rate": 0.00012193461064633255, "loss": 2.6302, "step": 9340 }, { "epoch": 3.243932900076629e-14, "grad_norm": 0.562574565410614, "learning_rate": 0.00012192732289375626, "loss": 2.6455, "step": 9350 }, { "epoch": 3.247402347028583e-14, "grad_norm": 0.5885260105133057, "learning_rate": 0.00012192002673164413, "loss": 2.6344, "step": 9360 }, { "epoch": 3.2508717939805365e-14, "grad_norm": 0.5488312840461731, "learning_rate": 0.00012191272216114995, "loss": 2.6706, "step": 9370 }, { "epoch": 3.25434124093249e-14, "grad_norm": 0.5654180645942688, "learning_rate": 0.0001219054091834289, "loss": 2.6656, "step": 9380 }, { "epoch": 3.257810687884444e-14, "grad_norm": 0.5743809342384338, "learning_rate": 0.00012189808779963744, "loss": 2.6602, "step": 9390 }, { "epoch": 3.2612801348363973e-14, "grad_norm": 0.5739467144012451, "learning_rate": 0.00012189075801093341, "loss": 2.6667, "step": 9400 }, { "epoch": 3.264749581788351e-14, "grad_norm": 0.5663201212882996, "learning_rate": 0.00012188341981847592, "loss": 2.6585, "step": 9410 }, { "epoch": 3.2682190287403046e-14, "grad_norm": 0.5551659464836121, "learning_rate": 0.00012187607322342546, "loss": 2.6168, "step": 9420 }, { "epoch": 3.271688475692258e-14, "grad_norm": 0.5350239276885986, "learning_rate": 0.00012186871822694385, "loss": 2.6613, "step": 9430 }, { "epoch": 3.275157922644212e-14, "grad_norm": 0.5728340148925781, "learning_rate": 0.0001218613548301942, "loss": 2.6287, "step": 9440 }, { "epoch": 3.2786273695961654e-14, "grad_norm": 0.5597618222236633, "learning_rate": 0.00012185398303434098, "loss": 2.6484, "step": 9450 }, { "epoch": 3.282096816548119e-14, "grad_norm": 0.5319777131080627, "learning_rate": 0.00012184660284054994, "loss": 2.621, "step": 9460 }, { "epoch": 3.2855662635000726e-14, "grad_norm": 0.5528765916824341, "learning_rate": 0.00012183921424998823, "loss": 2.5949, "step": 9470 }, { "epoch": 3.289035710452026e-14, "grad_norm": 0.5720252394676208, "learning_rate": 0.00012183181726382428, "loss": 2.6348, "step": 9480 }, { "epoch": 3.29250515740398e-14, "grad_norm": 0.597533106803894, "learning_rate": 0.00012182441188322785, "loss": 2.6342, "step": 9490 }, { "epoch": 3.2959746043559335e-14, "grad_norm": 0.5717959403991699, "learning_rate": 0.00012181699810937007, "loss": 2.6187, "step": 9500 }, { "epoch": 3.299444051307887e-14, "grad_norm": 0.6093868613243103, "learning_rate": 0.00012180957594342334, "loss": 2.631, "step": 9510 }, { "epoch": 3.302913498259841e-14, "grad_norm": 0.5646700859069824, "learning_rate": 0.0001218021453865614, "loss": 2.6287, "step": 9520 }, { "epoch": 3.3063829452117943e-14, "grad_norm": 0.548583984375, "learning_rate": 0.00012179470643995935, "loss": 2.6607, "step": 9530 }, { "epoch": 3.309852392163748e-14, "grad_norm": 0.5663267374038696, "learning_rate": 0.00012178725910479359, "loss": 2.6321, "step": 9540 }, { "epoch": 3.3133218391157016e-14, "grad_norm": 0.5751446485519409, "learning_rate": 0.00012177980338224186, "loss": 2.6502, "step": 9550 }, { "epoch": 3.316791286067655e-14, "grad_norm": 0.5875082612037659, "learning_rate": 0.0001217723392734832, "loss": 2.5709, "step": 9560 }, { "epoch": 3.320260733019609e-14, "grad_norm": 0.5595256090164185, "learning_rate": 0.00012176486677969799, "loss": 2.5983, "step": 9570 }, { "epoch": 3.3237301799715624e-14, "grad_norm": 0.5367342829704285, "learning_rate": 0.00012175738590206796, "loss": 2.6153, "step": 9580 }, { "epoch": 3.327199626923516e-14, "grad_norm": 0.5533463358879089, "learning_rate": 0.00012174989664177614, "loss": 2.6648, "step": 9590 }, { "epoch": 3.3306690738754696e-14, "grad_norm": 0.5985952019691467, "learning_rate": 0.00012174239900000692, "loss": 2.6487, "step": 9600 }, { "epoch": 3.334138520827423e-14, "grad_norm": 0.5100234150886536, "learning_rate": 0.00012173489297794594, "loss": 2.6231, "step": 9610 }, { "epoch": 3.337607967779377e-14, "grad_norm": 0.5480148196220398, "learning_rate": 0.00012172737857678022, "loss": 2.5996, "step": 9620 }, { "epoch": 3.3410774147313305e-14, "grad_norm": 0.5434893369674683, "learning_rate": 0.00012171985579769812, "loss": 2.6622, "step": 9630 }, { "epoch": 3.344546861683284e-14, "grad_norm": 0.54547119140625, "learning_rate": 0.00012171232464188929, "loss": 2.6571, "step": 9640 }, { "epoch": 3.348016308635238e-14, "grad_norm": 0.5238019227981567, "learning_rate": 0.00012170478511054472, "loss": 2.6418, "step": 9650 }, { "epoch": 3.351485755587191e-14, "grad_norm": 0.5617318749427795, "learning_rate": 0.00012169723720485675, "loss": 2.6355, "step": 9660 }, { "epoch": 3.354955202539145e-14, "grad_norm": 0.5515934228897095, "learning_rate": 0.00012168968092601897, "loss": 2.662, "step": 9670 }, { "epoch": 3.3584246494910985e-14, "grad_norm": 0.5660345554351807, "learning_rate": 0.00012168211627522635, "loss": 2.6264, "step": 9680 }, { "epoch": 3.361894096443052e-14, "grad_norm": 0.5841648578643799, "learning_rate": 0.00012167454325367519, "loss": 2.6663, "step": 9690 }, { "epoch": 3.365363543395006e-14, "grad_norm": 0.5720034837722778, "learning_rate": 0.00012166696186256309, "loss": 2.6172, "step": 9700 }, { "epoch": 3.3688329903469594e-14, "grad_norm": 0.5382291078567505, "learning_rate": 0.00012165937210308899, "loss": 2.6066, "step": 9710 }, { "epoch": 3.372302437298913e-14, "grad_norm": 0.5801699161529541, "learning_rate": 0.00012165177397645312, "loss": 2.6327, "step": 9720 }, { "epoch": 3.3757718842508666e-14, "grad_norm": 0.5597705841064453, "learning_rate": 0.00012164416748385711, "loss": 2.6067, "step": 9730 }, { "epoch": 3.37924133120282e-14, "grad_norm": 0.5449910759925842, "learning_rate": 0.0001216365526265038, "loss": 2.6436, "step": 9740 }, { "epoch": 3.382710778154774e-14, "grad_norm": 0.5628539323806763, "learning_rate": 0.00012162892940559745, "loss": 2.665, "step": 9750 }, { "epoch": 3.3861802251067274e-14, "grad_norm": 0.542974054813385, "learning_rate": 0.00012162129782234362, "loss": 2.6462, "step": 9760 }, { "epoch": 3.389649672058681e-14, "grad_norm": 0.5859957337379456, "learning_rate": 0.00012161365787794913, "loss": 2.6237, "step": 9770 }, { "epoch": 3.3931191190106347e-14, "grad_norm": 0.5761734247207642, "learning_rate": 0.00012160600957362222, "loss": 2.6634, "step": 9780 }, { "epoch": 3.396588565962588e-14, "grad_norm": 0.5522316694259644, "learning_rate": 0.00012159835291057236, "loss": 2.6219, "step": 9790 }, { "epoch": 3.400058012914542e-14, "grad_norm": 0.5559985041618347, "learning_rate": 0.00012159068789001044, "loss": 2.6618, "step": 9800 }, { "epoch": 3.4035274598664955e-14, "grad_norm": 0.7429655194282532, "learning_rate": 0.00012158301451314855, "loss": 2.6024, "step": 9810 }, { "epoch": 3.406996906818449e-14, "grad_norm": 0.563667893409729, "learning_rate": 0.00012157533278120023, "loss": 2.6171, "step": 9820 }, { "epoch": 3.410466353770403e-14, "grad_norm": 0.5564098954200745, "learning_rate": 0.00012156764269538026, "loss": 2.6318, "step": 9830 }, { "epoch": 3.4139358007223564e-14, "grad_norm": 0.5968917012214661, "learning_rate": 0.00012155994425690474, "loss": 2.6304, "step": 9840 }, { "epoch": 3.41740524767431e-14, "grad_norm": 0.5802537202835083, "learning_rate": 0.00012155223746699113, "loss": 2.6405, "step": 9850 }, { "epoch": 3.4208746946262636e-14, "grad_norm": 0.6104631423950195, "learning_rate": 0.00012154452232685818, "loss": 2.6582, "step": 9860 }, { "epoch": 3.424344141578217e-14, "grad_norm": 0.5636932253837585, "learning_rate": 0.000121536798837726, "loss": 2.6222, "step": 9870 }, { "epoch": 3.427813588530171e-14, "grad_norm": 0.5644193887710571, "learning_rate": 0.00012152906700081597, "loss": 2.6005, "step": 9880 }, { "epoch": 3.4312830354821244e-14, "grad_norm": 0.5523471236228943, "learning_rate": 0.00012152132681735083, "loss": 2.6533, "step": 9890 }, { "epoch": 3.434752482434078e-14, "grad_norm": 0.5553683042526245, "learning_rate": 0.00012151357828855459, "loss": 2.6527, "step": 9900 }, { "epoch": 3.4382219293860317e-14, "grad_norm": 0.5407199859619141, "learning_rate": 0.00012150582141565264, "loss": 2.5983, "step": 9910 }, { "epoch": 3.441691376337985e-14, "grad_norm": 0.5622711181640625, "learning_rate": 0.00012149805619987167, "loss": 2.6487, "step": 9920 }, { "epoch": 3.445160823289939e-14, "grad_norm": 0.5716853141784668, "learning_rate": 0.00012149028264243966, "loss": 2.658, "step": 9930 }, { "epoch": 3.4486302702418925e-14, "grad_norm": 0.5754017233848572, "learning_rate": 0.00012148250074458596, "loss": 2.5955, "step": 9940 }, { "epoch": 3.452099717193846e-14, "grad_norm": 0.5677022337913513, "learning_rate": 0.00012147471050754117, "loss": 2.6195, "step": 9950 }, { "epoch": 3.4555691641458e-14, "grad_norm": 0.5266816020011902, "learning_rate": 0.00012146691193253729, "loss": 2.6204, "step": 9960 }, { "epoch": 3.4590386110977533e-14, "grad_norm": 0.5604323744773865, "learning_rate": 0.00012145910502080757, "loss": 2.5897, "step": 9970 }, { "epoch": 3.462508058049707e-14, "grad_norm": 0.5449842214584351, "learning_rate": 0.0001214512897735866, "loss": 2.6233, "step": 9980 }, { "epoch": 3.4659775050016606e-14, "grad_norm": 0.5615988969802856, "learning_rate": 0.00012144346619211032, "loss": 2.6395, "step": 9990 }, { "epoch": 3.469446951953614e-14, "grad_norm": 0.5592764616012573, "learning_rate": 0.00012143563427761595, "loss": 2.6549, "step": 10000 }, { "epoch": 3.472916398905568e-14, "grad_norm": 0.5349418520927429, "learning_rate": 0.00012142779403134204, "loss": 2.635, "step": 10010 }, { "epoch": 3.4763858458575214e-14, "grad_norm": 0.5817193388938904, "learning_rate": 0.00012141994545452845, "loss": 2.6335, "step": 10020 }, { "epoch": 3.479855292809475e-14, "grad_norm": 0.559002161026001, "learning_rate": 0.00012141208854841637, "loss": 2.6205, "step": 10030 }, { "epoch": 3.4833247397614286e-14, "grad_norm": 0.5750009417533875, "learning_rate": 0.00012140422331424831, "loss": 2.6058, "step": 10040 }, { "epoch": 3.486794186713382e-14, "grad_norm": 0.5333647727966309, "learning_rate": 0.0001213963497532681, "loss": 2.6119, "step": 10050 }, { "epoch": 3.490263633665336e-14, "grad_norm": 0.5548805594444275, "learning_rate": 0.00012138846786672085, "loss": 2.617, "step": 10060 }, { "epoch": 3.4937330806172895e-14, "grad_norm": 0.549152672290802, "learning_rate": 0.00012138057765585302, "loss": 2.6547, "step": 10070 }, { "epoch": 3.497202527569243e-14, "grad_norm": 0.5450000762939453, "learning_rate": 0.00012137267912191237, "loss": 2.5918, "step": 10080 }, { "epoch": 3.500671974521197e-14, "grad_norm": 0.5352359414100647, "learning_rate": 0.00012136477226614802, "loss": 2.6314, "step": 10090 }, { "epoch": 3.5041414214731503e-14, "grad_norm": 0.6112861037254333, "learning_rate": 0.00012135685708981033, "loss": 2.6332, "step": 10100 }, { "epoch": 3.507610868425104e-14, "grad_norm": 0.590126633644104, "learning_rate": 0.00012134893359415103, "loss": 2.6294, "step": 10110 }, { "epoch": 3.5110803153770576e-14, "grad_norm": 0.5565000176429749, "learning_rate": 0.00012134100178042318, "loss": 2.6165, "step": 10120 }, { "epoch": 3.514549762329011e-14, "grad_norm": 0.5644953846931458, "learning_rate": 0.00012133306164988107, "loss": 2.6488, "step": 10130 }, { "epoch": 3.518019209280965e-14, "grad_norm": 0.7008234858512878, "learning_rate": 0.00012132511320378042, "loss": 2.5983, "step": 10140 }, { "epoch": 3.5214886562329184e-14, "grad_norm": 0.5759028792381287, "learning_rate": 0.00012131715644337818, "loss": 2.6325, "step": 10150 }, { "epoch": 3.524958103184872e-14, "grad_norm": 0.5520719289779663, "learning_rate": 0.00012130919136993265, "loss": 2.6199, "step": 10160 }, { "epoch": 3.5284275501368256e-14, "grad_norm": 0.5636146664619446, "learning_rate": 0.00012130121798470343, "loss": 2.6518, "step": 10170 }, { "epoch": 3.531896997088779e-14, "grad_norm": 0.5818845629692078, "learning_rate": 0.00012129323628895148, "loss": 2.6562, "step": 10180 }, { "epoch": 3.535366444040733e-14, "grad_norm": 0.5608141422271729, "learning_rate": 0.00012128524628393899, "loss": 2.6449, "step": 10190 }, { "epoch": 3.5388358909926865e-14, "grad_norm": 0.5501791834831238, "learning_rate": 0.00012127724797092951, "loss": 2.641, "step": 10200 }, { "epoch": 3.54230533794464e-14, "grad_norm": 0.5205058455467224, "learning_rate": 0.00012126924135118794, "loss": 2.6228, "step": 10210 }, { "epoch": 3.545774784896594e-14, "grad_norm": 0.5496019721031189, "learning_rate": 0.00012126122642598044, "loss": 2.6296, "step": 10220 }, { "epoch": 3.549244231848547e-14, "grad_norm": 0.5422104597091675, "learning_rate": 0.00012125320319657447, "loss": 2.6144, "step": 10230 }, { "epoch": 3.552713678800501e-14, "grad_norm": 0.5842846035957336, "learning_rate": 0.0001212451716642389, "loss": 2.6127, "step": 10240 }, { "epoch": 3.5561831257524545e-14, "grad_norm": 0.5422183871269226, "learning_rate": 0.0001212371318302438, "loss": 2.5888, "step": 10250 }, { "epoch": 3.559652572704408e-14, "grad_norm": 0.5695863366127014, "learning_rate": 0.00012122908369586063, "loss": 2.6197, "step": 10260 }, { "epoch": 3.563122019656362e-14, "grad_norm": 0.6287125945091248, "learning_rate": 0.00012122102726236208, "loss": 2.6356, "step": 10270 }, { "epoch": 3.5665914666083154e-14, "grad_norm": 0.5702443718910217, "learning_rate": 0.00012121296253102228, "loss": 2.6147, "step": 10280 }, { "epoch": 3.570060913560269e-14, "grad_norm": 0.5915464162826538, "learning_rate": 0.00012120488950311654, "loss": 2.6292, "step": 10290 }, { "epoch": 3.5735303605122226e-14, "grad_norm": 0.5436525940895081, "learning_rate": 0.00012119680817992155, "loss": 2.6108, "step": 10300 }, { "epoch": 3.576999807464176e-14, "grad_norm": 0.5847308039665222, "learning_rate": 0.00012118871856271533, "loss": 2.5815, "step": 10310 }, { "epoch": 3.58046925441613e-14, "grad_norm": 0.5649065375328064, "learning_rate": 0.00012118062065277717, "loss": 2.5854, "step": 10320 }, { "epoch": 3.5839387013680835e-14, "grad_norm": 0.5282293558120728, "learning_rate": 0.00012117251445138767, "loss": 2.5968, "step": 10330 }, { "epoch": 3.587408148320037e-14, "grad_norm": 0.5436049103736877, "learning_rate": 0.00012116439995982876, "loss": 2.6342, "step": 10340 }, { "epoch": 3.590877595271991e-14, "grad_norm": 0.5505560040473938, "learning_rate": 0.0001211562771793837, "loss": 2.5946, "step": 10350 }, { "epoch": 3.594347042223944e-14, "grad_norm": 0.7935149669647217, "learning_rate": 0.00012114814611133699, "loss": 2.6009, "step": 10360 }, { "epoch": 3.597816489175898e-14, "grad_norm": 0.5632250905036926, "learning_rate": 0.00012114000675697455, "loss": 2.5909, "step": 10370 }, { "epoch": 3.6012859361278515e-14, "grad_norm": 0.5430962443351746, "learning_rate": 0.0001211318591175835, "loss": 2.6145, "step": 10380 }, { "epoch": 3.604755383079805e-14, "grad_norm": 0.5550293326377869, "learning_rate": 0.00012112370319445233, "loss": 2.6101, "step": 10390 }, { "epoch": 3.608224830031759e-14, "grad_norm": 0.5625216960906982, "learning_rate": 0.00012111553898887088, "loss": 2.6119, "step": 10400 }, { "epoch": 3.6116942769837124e-14, "grad_norm": 0.5645459890365601, "learning_rate": 0.00012110736650213015, "loss": 2.6231, "step": 10410 }, { "epoch": 3.615163723935666e-14, "grad_norm": 0.5542747974395752, "learning_rate": 0.00012109918573552262, "loss": 2.593, "step": 10420 }, { "epoch": 3.6186331708876196e-14, "grad_norm": 0.5197889804840088, "learning_rate": 0.000121090996690342, "loss": 2.6088, "step": 10430 }, { "epoch": 3.622102617839573e-14, "grad_norm": 0.5293579697608948, "learning_rate": 0.0001210827993678833, "loss": 2.5843, "step": 10440 }, { "epoch": 3.625572064791527e-14, "grad_norm": 0.560478687286377, "learning_rate": 0.00012107459376944287, "loss": 2.5825, "step": 10450 }, { "epoch": 3.6290415117434804e-14, "grad_norm": 0.5376710295677185, "learning_rate": 0.00012106637989631834, "loss": 2.6191, "step": 10460 }, { "epoch": 3.632510958695434e-14, "grad_norm": 0.5210549831390381, "learning_rate": 0.00012105815774980869, "loss": 2.6055, "step": 10470 }, { "epoch": 3.6359804056473877e-14, "grad_norm": 0.5574628710746765, "learning_rate": 0.00012104992733121415, "loss": 2.6159, "step": 10480 }, { "epoch": 3.639449852599341e-14, "grad_norm": 0.606914222240448, "learning_rate": 0.0001210416886418363, "loss": 2.6319, "step": 10490 }, { "epoch": 3.642919299551295e-14, "grad_norm": 0.5881217122077942, "learning_rate": 0.00012103344168297801, "loss": 2.6528, "step": 10500 }, { "epoch": 3.6463887465032485e-14, "grad_norm": 0.5776757597923279, "learning_rate": 0.0001210251864559435, "loss": 2.6052, "step": 10510 }, { "epoch": 3.649858193455202e-14, "grad_norm": 0.5661537051200867, "learning_rate": 0.00012101692296203824, "loss": 2.6123, "step": 10520 }, { "epoch": 3.653327640407156e-14, "grad_norm": 0.5847659707069397, "learning_rate": 0.00012100865120256903, "loss": 2.6053, "step": 10530 }, { "epoch": 3.6567970873591094e-14, "grad_norm": 0.5542734861373901, "learning_rate": 0.00012100037117884397, "loss": 2.6168, "step": 10540 }, { "epoch": 3.660266534311063e-14, "grad_norm": 0.5225227475166321, "learning_rate": 0.00012099208289217248, "loss": 2.5921, "step": 10550 }, { "epoch": 3.6637359812630166e-14, "grad_norm": 0.5730896592140198, "learning_rate": 0.00012098378634386528, "loss": 2.5783, "step": 10560 }, { "epoch": 3.66720542821497e-14, "grad_norm": 0.5391912460327148, "learning_rate": 0.00012097548153523442, "loss": 2.5599, "step": 10570 }, { "epoch": 3.670674875166924e-14, "grad_norm": 0.5337449908256531, "learning_rate": 0.00012096716846759318, "loss": 2.5759, "step": 10580 }, { "epoch": 3.6741443221188774e-14, "grad_norm": 0.5401464700698853, "learning_rate": 0.00012095884714225628, "loss": 2.5905, "step": 10590 }, { "epoch": 3.677613769070831e-14, "grad_norm": 0.5429476499557495, "learning_rate": 0.00012095051756053959, "loss": 2.6245, "step": 10600 }, { "epoch": 3.6810832160227847e-14, "grad_norm": 0.5727413892745972, "learning_rate": 0.00012094217972376042, "loss": 2.6357, "step": 10610 }, { "epoch": 3.684552662974738e-14, "grad_norm": 0.5658982396125793, "learning_rate": 0.00012093383363323727, "loss": 2.5721, "step": 10620 }, { "epoch": 3.688022109926692e-14, "grad_norm": 0.5996747016906738, "learning_rate": 0.00012092547929029004, "loss": 2.6237, "step": 10630 }, { "epoch": 3.6914915568786455e-14, "grad_norm": 0.5585694909095764, "learning_rate": 0.0001209171166962399, "loss": 2.6402, "step": 10640 }, { "epoch": 3.694961003830599e-14, "grad_norm": 0.5825451612472534, "learning_rate": 0.00012090874585240931, "loss": 2.6224, "step": 10650 }, { "epoch": 3.698430450782553e-14, "grad_norm": 0.5629118084907532, "learning_rate": 0.00012090036676012204, "loss": 2.6116, "step": 10660 }, { "epoch": 3.7018998977345063e-14, "grad_norm": 0.5307306051254272, "learning_rate": 0.0001208919794207032, "loss": 2.6074, "step": 10670 }, { "epoch": 3.70536934468646e-14, "grad_norm": 0.5950667858123779, "learning_rate": 0.00012088358383547914, "loss": 2.6363, "step": 10680 }, { "epoch": 3.7088387916384136e-14, "grad_norm": 0.55521160364151, "learning_rate": 0.00012087518000577756, "loss": 2.638, "step": 10690 }, { "epoch": 3.712308238590367e-14, "grad_norm": 0.5378021597862244, "learning_rate": 0.00012086676793292747, "loss": 2.6071, "step": 10700 }, { "epoch": 3.715777685542321e-14, "grad_norm": 0.548244833946228, "learning_rate": 0.00012085834761825914, "loss": 2.6425, "step": 10710 }, { "epoch": 3.7192471324942744e-14, "grad_norm": 0.5572342872619629, "learning_rate": 0.00012084991906310421, "loss": 2.6271, "step": 10720 }, { "epoch": 3.722716579446228e-14, "grad_norm": 0.5526764392852783, "learning_rate": 0.00012084148226879554, "loss": 2.6345, "step": 10730 }, { "epoch": 3.7261860263981816e-14, "grad_norm": 0.5432220697402954, "learning_rate": 0.00012083303723666738, "loss": 2.6295, "step": 10740 }, { "epoch": 3.729655473350135e-14, "grad_norm": 0.5489490032196045, "learning_rate": 0.00012082458396805518, "loss": 2.6482, "step": 10750 }, { "epoch": 3.733124920302089e-14, "grad_norm": 0.5437672138214111, "learning_rate": 0.00012081612246429581, "loss": 2.593, "step": 10760 }, { "epoch": 3.7365943672540425e-14, "grad_norm": 0.5519899725914001, "learning_rate": 0.00012080765272672735, "loss": 2.6093, "step": 10770 }, { "epoch": 3.740063814205996e-14, "grad_norm": 0.5485339760780334, "learning_rate": 0.00012079917475668923, "loss": 2.6408, "step": 10780 }, { "epoch": 3.74353326115795e-14, "grad_norm": 0.5540599226951599, "learning_rate": 0.00012079068855552218, "loss": 2.6254, "step": 10790 }, { "epoch": 3.7470027081099033e-14, "grad_norm": 0.550097644329071, "learning_rate": 0.00012078219412456818, "loss": 2.5729, "step": 10800 }, { "epoch": 3.750472155061857e-14, "grad_norm": 0.6055213809013367, "learning_rate": 0.0001207736914651706, "loss": 2.5688, "step": 10810 }, { "epoch": 3.7539416020138106e-14, "grad_norm": 0.5484846234321594, "learning_rate": 0.00012076518057867402, "loss": 2.5635, "step": 10820 }, { "epoch": 3.757411048965764e-14, "grad_norm": 0.588966965675354, "learning_rate": 0.0001207566614664244, "loss": 2.5915, "step": 10830 }, { "epoch": 3.760880495917718e-14, "grad_norm": 0.5600298047065735, "learning_rate": 0.00012074813412976891, "loss": 2.6043, "step": 10840 }, { "epoch": 3.7643499428696714e-14, "grad_norm": 0.5488225817680359, "learning_rate": 0.00012073959857005613, "loss": 2.6074, "step": 10850 }, { "epoch": 3.767819389821625e-14, "grad_norm": 0.5587607026100159, "learning_rate": 0.00012073105478863587, "loss": 2.6249, "step": 10860 }, { "epoch": 3.7712888367735786e-14, "grad_norm": 0.5290815234184265, "learning_rate": 0.00012072250278685925, "loss": 2.6331, "step": 10870 }, { "epoch": 3.774758283725532e-14, "grad_norm": 0.5708407759666443, "learning_rate": 0.00012071394256607868, "loss": 2.6363, "step": 10880 }, { "epoch": 3.778227730677486e-14, "grad_norm": 0.5720111131668091, "learning_rate": 0.00012070537412764793, "loss": 2.6146, "step": 10890 }, { "epoch": 3.7816971776294395e-14, "grad_norm": 0.5615537166595459, "learning_rate": 0.00012069679747292198, "loss": 2.5766, "step": 10900 }, { "epoch": 3.785166624581393e-14, "grad_norm": 0.5666632056236267, "learning_rate": 0.00012068821260325719, "loss": 2.6067, "step": 10910 }, { "epoch": 3.788636071533347e-14, "grad_norm": 0.5345425009727478, "learning_rate": 0.00012067961952001114, "loss": 2.6205, "step": 10920 }, { "epoch": 3.7921055184853e-14, "grad_norm": 0.5444106459617615, "learning_rate": 0.00012067101822454276, "loss": 2.5849, "step": 10930 }, { "epoch": 3.795574965437254e-14, "grad_norm": 0.5313660502433777, "learning_rate": 0.00012066240871821234, "loss": 2.6059, "step": 10940 }, { "epoch": 3.7990444123892075e-14, "grad_norm": 0.5680751204490662, "learning_rate": 0.0001206537910023813, "loss": 2.6202, "step": 10950 }, { "epoch": 3.802513859341161e-14, "grad_norm": 0.5371715426445007, "learning_rate": 0.00012064516507841252, "loss": 2.5892, "step": 10960 }, { "epoch": 3.805983306293115e-14, "grad_norm": 0.5551703572273254, "learning_rate": 0.00012063653094767011, "loss": 2.5933, "step": 10970 }, { "epoch": 3.8094527532450684e-14, "grad_norm": 0.5563033223152161, "learning_rate": 0.00012062788861151946, "loss": 2.6096, "step": 10980 }, { "epoch": 3.812922200197022e-14, "grad_norm": 0.5769312977790833, "learning_rate": 0.0001206192380713273, "loss": 2.6267, "step": 10990 }, { "epoch": 3.8163916471489756e-14, "grad_norm": 0.5728552341461182, "learning_rate": 0.00012061057932846161, "loss": 2.6246, "step": 11000 }, { "epoch": 3.819861094100929e-14, "grad_norm": 0.5569671392440796, "learning_rate": 0.00012060191238429175, "loss": 2.5917, "step": 11010 }, { "epoch": 3.823330541052883e-14, "grad_norm": 0.5582177042961121, "learning_rate": 0.00012059323724018828, "loss": 2.5815, "step": 11020 }, { "epoch": 3.8267999880048365e-14, "grad_norm": 0.5287896394729614, "learning_rate": 0.00012058455389752313, "loss": 2.5804, "step": 11030 }, { "epoch": 3.83026943495679e-14, "grad_norm": 0.5462626814842224, "learning_rate": 0.00012057586235766945, "loss": 2.5918, "step": 11040 }, { "epoch": 3.8337388819087437e-14, "grad_norm": 0.5342673659324646, "learning_rate": 0.00012056716262200175, "loss": 2.5695, "step": 11050 }, { "epoch": 3.837208328860697e-14, "grad_norm": 0.566508412361145, "learning_rate": 0.00012055845469189585, "loss": 2.5835, "step": 11060 }, { "epoch": 3.840677775812651e-14, "grad_norm": 0.5749298334121704, "learning_rate": 0.00012054973856872879, "loss": 2.6113, "step": 11070 }, { "epoch": 3.8441472227646045e-14, "grad_norm": 0.5610042810440063, "learning_rate": 0.00012054101425387897, "loss": 2.6247, "step": 11080 }, { "epoch": 3.847616669716558e-14, "grad_norm": 0.6043697595596313, "learning_rate": 0.00012053228174872609, "loss": 2.6042, "step": 11090 }, { "epoch": 3.851086116668512e-14, "grad_norm": 0.5326177477836609, "learning_rate": 0.00012052354105465107, "loss": 2.5801, "step": 11100 }, { "epoch": 3.8545555636204654e-14, "grad_norm": 0.5784082412719727, "learning_rate": 0.00012051479217303624, "loss": 2.5532, "step": 11110 }, { "epoch": 3.858025010572419e-14, "grad_norm": 0.5561894774436951, "learning_rate": 0.00012050603510526509, "loss": 2.5772, "step": 11120 }, { "epoch": 3.8614944575243726e-14, "grad_norm": 0.5524073839187622, "learning_rate": 0.00012049726985272252, "loss": 2.585, "step": 11130 }, { "epoch": 3.864963904476326e-14, "grad_norm": 0.5423831939697266, "learning_rate": 0.00012048849641679466, "loss": 2.595, "step": 11140 }, { "epoch": 3.86843335142828e-14, "grad_norm": 0.5639083385467529, "learning_rate": 0.00012047971479886896, "loss": 2.5817, "step": 11150 }, { "epoch": 3.8719027983802334e-14, "grad_norm": 0.5555867552757263, "learning_rate": 0.00012047092500033416, "loss": 2.6169, "step": 11160 }, { "epoch": 3.875372245332187e-14, "grad_norm": 0.5506484508514404, "learning_rate": 0.00012046212702258028, "loss": 2.5713, "step": 11170 }, { "epoch": 3.8788416922841407e-14, "grad_norm": 0.5442196726799011, "learning_rate": 0.00012045332086699865, "loss": 2.5846, "step": 11180 }, { "epoch": 3.882311139236094e-14, "grad_norm": 0.6276108622550964, "learning_rate": 0.00012044450653498191, "loss": 2.6022, "step": 11190 }, { "epoch": 3.885780586188048e-14, "grad_norm": 0.5636003613471985, "learning_rate": 0.00012043568402792392, "loss": 2.5841, "step": 11200 }, { "epoch": 3.8892500331400015e-14, "grad_norm": 0.5845822095870972, "learning_rate": 0.00012042685334721996, "loss": 2.6043, "step": 11210 }, { "epoch": 3.892719480091955e-14, "grad_norm": 0.52003413438797, "learning_rate": 0.00012041801449426645, "loss": 2.5847, "step": 11220 }, { "epoch": 3.896188927043909e-14, "grad_norm": 0.5558779835700989, "learning_rate": 0.0001204091674704612, "loss": 2.6012, "step": 11230 }, { "epoch": 3.8996583739958623e-14, "grad_norm": 0.5502667427062988, "learning_rate": 0.00012040031227720331, "loss": 2.6013, "step": 11240 }, { "epoch": 3.903127820947816e-14, "grad_norm": 0.5477631092071533, "learning_rate": 0.00012039144891589315, "loss": 2.5749, "step": 11250 }, { "epoch": 3.9065972678997696e-14, "grad_norm": 0.543838620185852, "learning_rate": 0.00012038257738793237, "loss": 2.5787, "step": 11260 }, { "epoch": 3.910066714851723e-14, "grad_norm": 0.5582623481750488, "learning_rate": 0.00012037369769472393, "loss": 2.6006, "step": 11270 }, { "epoch": 3.913536161803677e-14, "grad_norm": 0.5380273461341858, "learning_rate": 0.0001203648098376721, "loss": 2.6126, "step": 11280 }, { "epoch": 3.9170056087556304e-14, "grad_norm": 0.5521540641784668, "learning_rate": 0.00012035591381818236, "loss": 2.5669, "step": 11290 }, { "epoch": 3.920475055707584e-14, "grad_norm": 0.5741785764694214, "learning_rate": 0.0001203470096376616, "loss": 2.5823, "step": 11300 }, { "epoch": 3.9239445026595376e-14, "grad_norm": 0.546096920967102, "learning_rate": 0.00012033809729751793, "loss": 2.5862, "step": 11310 }, { "epoch": 3.927413949611491e-14, "grad_norm": 0.5233967304229736, "learning_rate": 0.00012032917679916073, "loss": 2.5602, "step": 11320 }, { "epoch": 3.930883396563445e-14, "grad_norm": 0.5461140871047974, "learning_rate": 0.00012032024814400072, "loss": 2.5626, "step": 11330 }, { "epoch": 3.9343528435153985e-14, "grad_norm": 0.5452801585197449, "learning_rate": 0.00012031131133344989, "loss": 2.6252, "step": 11340 }, { "epoch": 3.937822290467352e-14, "grad_norm": 0.5621305704116821, "learning_rate": 0.00012030236636892149, "loss": 2.5854, "step": 11350 }, { "epoch": 3.941291737419306e-14, "grad_norm": 0.5446386337280273, "learning_rate": 0.00012029341325183012, "loss": 2.5904, "step": 11360 }, { "epoch": 3.9447611843712593e-14, "grad_norm": 0.5620809197425842, "learning_rate": 0.00012028445198359166, "loss": 2.6118, "step": 11370 }, { "epoch": 3.948230631323213e-14, "grad_norm": 0.5497198104858398, "learning_rate": 0.00012027548256562321, "loss": 2.591, "step": 11380 }, { "epoch": 3.9517000782751666e-14, "grad_norm": 0.5528308153152466, "learning_rate": 0.00012026650499934324, "loss": 2.5622, "step": 11390 }, { "epoch": 3.95516952522712e-14, "grad_norm": 0.5670973658561707, "learning_rate": 0.00012025751928617145, "loss": 2.6337, "step": 11400 }, { "epoch": 3.958638972179074e-14, "grad_norm": 0.5377684235572815, "learning_rate": 0.00012024852542752886, "loss": 2.5809, "step": 11410 }, { "epoch": 3.9621084191310274e-14, "grad_norm": 0.5608730316162109, "learning_rate": 0.00012023952342483777, "loss": 2.5881, "step": 11420 }, { "epoch": 3.965577866082981e-14, "grad_norm": 0.5398859977722168, "learning_rate": 0.0001202305132795218, "loss": 2.5799, "step": 11430 }, { "epoch": 3.9690473130349346e-14, "grad_norm": 0.5563932657241821, "learning_rate": 0.00012022149499300579, "loss": 2.5715, "step": 11440 }, { "epoch": 3.972516759986888e-14, "grad_norm": 0.5408849716186523, "learning_rate": 0.00012021246856671591, "loss": 2.5634, "step": 11450 }, { "epoch": 3.975986206938842e-14, "grad_norm": 0.5421640276908875, "learning_rate": 0.00012020343400207964, "loss": 2.5913, "step": 11460 }, { "epoch": 3.9794556538907955e-14, "grad_norm": 0.5459408760070801, "learning_rate": 0.00012019439130052567, "loss": 2.5453, "step": 11470 }, { "epoch": 3.982925100842749e-14, "grad_norm": 0.5121273994445801, "learning_rate": 0.00012018534046348405, "loss": 2.6049, "step": 11480 }, { "epoch": 3.986394547794703e-14, "grad_norm": 0.5439488887786865, "learning_rate": 0.00012017628149238613, "loss": 2.5835, "step": 11490 }, { "epoch": 3.9898639947466563e-14, "grad_norm": 0.5245824456214905, "learning_rate": 0.00012016721438866444, "loss": 2.5976, "step": 11500 }, { "epoch": 3.99333344169861e-14, "grad_norm": 0.5365587472915649, "learning_rate": 0.00012015813915375293, "loss": 2.6047, "step": 11510 }, { "epoch": 3.9968028886505635e-14, "grad_norm": 0.5417706370353699, "learning_rate": 0.00012014905578908669, "loss": 2.5853, "step": 11520 }, { "epoch": 4.000272335602517e-14, "grad_norm": 0.5575244426727295, "learning_rate": 0.00012013996429610226, "loss": 2.5919, "step": 11530 }, { "epoch": 4.003741782554471e-14, "grad_norm": 0.5546166896820068, "learning_rate": 0.00012013086467623734, "loss": 2.5868, "step": 11540 }, { "epoch": 4.0072112295064244e-14, "grad_norm": 0.5590048432350159, "learning_rate": 0.00012012175693093094, "loss": 2.5782, "step": 11550 }, { "epoch": 4.010680676458378e-14, "grad_norm": 0.5884609818458557, "learning_rate": 0.00012011264106162341, "loss": 2.6167, "step": 11560 }, { "epoch": 4.0141501234103316e-14, "grad_norm": 0.5702016353607178, "learning_rate": 0.00012010351706975634, "loss": 2.5838, "step": 11570 }, { "epoch": 4.017619570362285e-14, "grad_norm": 0.525166928768158, "learning_rate": 0.00012009438495677256, "loss": 2.5596, "step": 11580 }, { "epoch": 4.021089017314239e-14, "grad_norm": 0.5347808599472046, "learning_rate": 0.00012008524472411631, "loss": 2.6122, "step": 11590 }, { "epoch": 4.0245584642661925e-14, "grad_norm": 0.531851053237915, "learning_rate": 0.00012007609637323298, "loss": 2.5879, "step": 11600 }, { "epoch": 4.028027911218146e-14, "grad_norm": 0.5621383190155029, "learning_rate": 0.00012006693990556936, "loss": 2.5698, "step": 11610 }, { "epoch": 4.0314973581701e-14, "grad_norm": 0.537147045135498, "learning_rate": 0.0001200577753225734, "loss": 2.558, "step": 11620 }, { "epoch": 4.034966805122053e-14, "grad_norm": 0.5917002558708191, "learning_rate": 0.00012004860262569446, "loss": 2.5539, "step": 11630 }, { "epoch": 4.038436252074007e-14, "grad_norm": 0.6091615557670593, "learning_rate": 0.00012003942181638307, "loss": 2.5699, "step": 11640 }, { "epoch": 4.0419056990259605e-14, "grad_norm": 0.5468105673789978, "learning_rate": 0.00012003023289609114, "loss": 2.5638, "step": 11650 }, { "epoch": 4.045375145977914e-14, "grad_norm": 0.539151668548584, "learning_rate": 0.0001200210358662718, "loss": 2.588, "step": 11660 }, { "epoch": 4.048844592929868e-14, "grad_norm": 0.5409261584281921, "learning_rate": 0.00012001183072837946, "loss": 2.6127, "step": 11670 }, { "epoch": 4.0523140398818214e-14, "grad_norm": 0.5254801511764526, "learning_rate": 0.00012000261748386987, "loss": 2.5669, "step": 11680 }, { "epoch": 4.055783486833775e-14, "grad_norm": 0.5562078356742859, "learning_rate": 0.00011999339613420002, "loss": 2.5372, "step": 11690 }, { "epoch": 4.0592529337857286e-14, "grad_norm": 0.576702356338501, "learning_rate": 0.00011998416668082817, "loss": 2.5698, "step": 11700 }, { "epoch": 4.062722380737682e-14, "grad_norm": 0.5529664754867554, "learning_rate": 0.00011997492912521389, "loss": 2.6157, "step": 11710 }, { "epoch": 4.066191827689636e-14, "grad_norm": 0.553093433380127, "learning_rate": 0.00011996568346881801, "loss": 2.5643, "step": 11720 }, { "epoch": 4.0696612746415894e-14, "grad_norm": 0.5511648058891296, "learning_rate": 0.00011995642971310265, "loss": 2.5719, "step": 11730 }, { "epoch": 4.073130721593543e-14, "grad_norm": 0.5893524885177612, "learning_rate": 0.00011994716785953123, "loss": 2.6082, "step": 11740 }, { "epoch": 4.0766001685454967e-14, "grad_norm": 0.6157523989677429, "learning_rate": 0.00011993789790956842, "loss": 2.5877, "step": 11750 }, { "epoch": 4.08006961549745e-14, "grad_norm": 0.5966951847076416, "learning_rate": 0.00011992861986468018, "loss": 2.5674, "step": 11760 }, { "epoch": 4.083539062449404e-14, "grad_norm": 0.5405528545379639, "learning_rate": 0.00011991933372633376, "loss": 2.5975, "step": 11770 }, { "epoch": 4.0870085094013575e-14, "grad_norm": 0.5619329810142517, "learning_rate": 0.00011991003949599767, "loss": 2.5572, "step": 11780 }, { "epoch": 4.090477956353311e-14, "grad_norm": 0.5892471671104431, "learning_rate": 0.00011990073717514174, "loss": 2.5924, "step": 11790 }, { "epoch": 4.093947403305265e-14, "grad_norm": 0.5538279414176941, "learning_rate": 0.00011989142676523702, "loss": 2.5657, "step": 11800 }, { "epoch": 4.0974168502572184e-14, "grad_norm": 0.5286985039710999, "learning_rate": 0.0001198821082677559, "loss": 2.5583, "step": 11810 }, { "epoch": 4.100886297209172e-14, "grad_norm": 0.546241283416748, "learning_rate": 0.000119872781684172, "loss": 2.615, "step": 11820 }, { "epoch": 4.1043557441611256e-14, "grad_norm": 0.5642609596252441, "learning_rate": 0.00011986344701596025, "loss": 2.5899, "step": 11830 }, { "epoch": 4.107825191113079e-14, "grad_norm": 0.6441773176193237, "learning_rate": 0.00011985410426459684, "loss": 2.5678, "step": 11840 }, { "epoch": 4.111294638065033e-14, "grad_norm": 0.554463267326355, "learning_rate": 0.00011984475343155923, "loss": 2.5819, "step": 11850 }, { "epoch": 4.1147640850169864e-14, "grad_norm": 0.5494024157524109, "learning_rate": 0.0001198353945183262, "loss": 2.574, "step": 11860 }, { "epoch": 4.11823353196894e-14, "grad_norm": 0.5484917163848877, "learning_rate": 0.00011982602752637778, "loss": 2.612, "step": 11870 }, { "epoch": 4.1217029789208937e-14, "grad_norm": 0.5077998042106628, "learning_rate": 0.00011981665245719527, "loss": 2.5664, "step": 11880 }, { "epoch": 4.125172425872847e-14, "grad_norm": 0.5697404146194458, "learning_rate": 0.00011980726931226125, "loss": 2.5647, "step": 11890 }, { "epoch": 4.128641872824801e-14, "grad_norm": 0.539943277835846, "learning_rate": 0.0001197978780930596, "loss": 2.5783, "step": 11900 }, { "epoch": 4.1321113197767545e-14, "grad_norm": 0.5540561676025391, "learning_rate": 0.00011978847880107546, "loss": 2.5728, "step": 11910 }, { "epoch": 4.135580766728708e-14, "grad_norm": 0.5263074636459351, "learning_rate": 0.00011977907143779522, "loss": 2.5198, "step": 11920 }, { "epoch": 4.139050213680662e-14, "grad_norm": 0.572586178779602, "learning_rate": 0.00011976965600470661, "loss": 2.5555, "step": 11930 }, { "epoch": 4.1425196606326153e-14, "grad_norm": 0.5449743270874023, "learning_rate": 0.00011976023250329857, "loss": 2.5371, "step": 11940 }, { "epoch": 4.145989107584569e-14, "grad_norm": 0.6047948002815247, "learning_rate": 0.00011975080093506138, "loss": 2.6102, "step": 11950 }, { "epoch": 4.1494585545365226e-14, "grad_norm": 0.5717654228210449, "learning_rate": 0.00011974136130148652, "loss": 2.5773, "step": 11960 }, { "epoch": 4.152928001488476e-14, "grad_norm": 0.5705273151397705, "learning_rate": 0.00011973191360406684, "loss": 2.5718, "step": 11970 }, { "epoch": 4.15639744844043e-14, "grad_norm": 0.5969630479812622, "learning_rate": 0.00011972245784429633, "loss": 2.5981, "step": 11980 }, { "epoch": 4.1598668953923834e-14, "grad_norm": 0.5442354083061218, "learning_rate": 0.00011971299402367042, "loss": 2.6011, "step": 11990 }, { "epoch": 4.163336342344337e-14, "grad_norm": 0.545595645904541, "learning_rate": 0.00011970352214368569, "loss": 2.593, "step": 12000 }, { "epoch": 4.1668057892962906e-14, "grad_norm": 0.5431531071662903, "learning_rate": 0.00011969404220584004, "loss": 2.58, "step": 12010 }, { "epoch": 4.170275236248244e-14, "grad_norm": 0.5382686257362366, "learning_rate": 0.00011968455421163263, "loss": 2.5772, "step": 12020 }, { "epoch": 4.173744683200198e-14, "grad_norm": 0.5286597609519958, "learning_rate": 0.00011967505816256394, "loss": 2.5536, "step": 12030 }, { "epoch": 4.1772141301521515e-14, "grad_norm": 0.555290937423706, "learning_rate": 0.00011966555406013568, "loss": 2.6215, "step": 12040 }, { "epoch": 4.180683577104105e-14, "grad_norm": 0.5584155321121216, "learning_rate": 0.0001196560419058508, "loss": 2.5744, "step": 12050 }, { "epoch": 4.184153024056059e-14, "grad_norm": 0.5442929863929749, "learning_rate": 0.00011964652170121361, "loss": 2.5697, "step": 12060 }, { "epoch": 4.1876224710080123e-14, "grad_norm": 0.5342869758605957, "learning_rate": 0.00011963699344772962, "loss": 2.5606, "step": 12070 }, { "epoch": 4.191091917959966e-14, "grad_norm": 0.5867231488227844, "learning_rate": 0.00011962745714690567, "loss": 2.6183, "step": 12080 }, { "epoch": 4.1945613649119196e-14, "grad_norm": 0.5683425664901733, "learning_rate": 0.00011961791280024982, "loss": 2.5806, "step": 12090 }, { "epoch": 4.198030811863873e-14, "grad_norm": 0.5399709343910217, "learning_rate": 0.00011960836040927142, "loss": 2.5705, "step": 12100 }, { "epoch": 4.201500258815827e-14, "grad_norm": 0.5142285227775574, "learning_rate": 0.00011959879997548115, "loss": 2.5704, "step": 12110 }, { "epoch": 4.2049697057677804e-14, "grad_norm": 0.5363353490829468, "learning_rate": 0.00011958923150039085, "loss": 2.6147, "step": 12120 }, { "epoch": 4.208439152719734e-14, "grad_norm": 0.5711092948913574, "learning_rate": 0.0001195796549855137, "loss": 2.5441, "step": 12130 }, { "epoch": 4.2119085996716876e-14, "grad_norm": 0.5700768828392029, "learning_rate": 0.00011957007043236419, "loss": 2.5431, "step": 12140 }, { "epoch": 4.215378046623641e-14, "grad_norm": 0.553406834602356, "learning_rate": 0.000119560477842458, "loss": 2.5644, "step": 12150 }, { "epoch": 4.218847493575595e-14, "grad_norm": 0.5756966471672058, "learning_rate": 0.00011955087721731211, "loss": 2.5645, "step": 12160 }, { "epoch": 4.2223169405275485e-14, "grad_norm": 0.5456786751747131, "learning_rate": 0.00011954126855844477, "loss": 2.5102, "step": 12170 }, { "epoch": 4.225786387479502e-14, "grad_norm": 0.5663999319076538, "learning_rate": 0.00011953165186737555, "loss": 2.5618, "step": 12180 }, { "epoch": 4.229255834431456e-14, "grad_norm": 0.5544654130935669, "learning_rate": 0.0001195220271456252, "loss": 2.5367, "step": 12190 }, { "epoch": 4.232725281383409e-14, "grad_norm": 0.5503784418106079, "learning_rate": 0.00011951239439471581, "loss": 2.5791, "step": 12200 }, { "epoch": 4.236194728335363e-14, "grad_norm": 0.5194251537322998, "learning_rate": 0.0001195027536161707, "loss": 2.56, "step": 12210 }, { "epoch": 4.2396641752873165e-14, "grad_norm": 0.5451456308364868, "learning_rate": 0.00011949310481151452, "loss": 2.6052, "step": 12220 }, { "epoch": 4.24313362223927e-14, "grad_norm": 0.5330513119697571, "learning_rate": 0.00011948344798227307, "loss": 2.5511, "step": 12230 }, { "epoch": 4.246603069191224e-14, "grad_norm": 0.5135144591331482, "learning_rate": 0.00011947378312997359, "loss": 2.5582, "step": 12240 }, { "epoch": 4.2500725161431774e-14, "grad_norm": 0.5606111288070679, "learning_rate": 0.0001194641102561444, "loss": 2.5685, "step": 12250 }, { "epoch": 4.253541963095131e-14, "grad_norm": 0.5538490414619446, "learning_rate": 0.00011945442936231523, "loss": 2.5814, "step": 12260 }, { "epoch": 4.2570114100470846e-14, "grad_norm": 0.5380173921585083, "learning_rate": 0.00011944474045001703, "loss": 2.5713, "step": 12270 }, { "epoch": 4.260480856999038e-14, "grad_norm": 0.5578806400299072, "learning_rate": 0.00011943504352078203, "loss": 2.5466, "step": 12280 }, { "epoch": 4.263950303950992e-14, "grad_norm": 0.5822352170944214, "learning_rate": 0.00011942533857614367, "loss": 2.574, "step": 12290 }, { "epoch": 4.2674197509029455e-14, "grad_norm": 0.5446638464927673, "learning_rate": 0.00011941562561763674, "loss": 2.5496, "step": 12300 }, { "epoch": 4.270889197854899e-14, "grad_norm": 0.5514087677001953, "learning_rate": 0.00011940590464679724, "loss": 2.5772, "step": 12310 }, { "epoch": 4.2743586448068527e-14, "grad_norm": 0.5503462553024292, "learning_rate": 0.00011939617566516248, "loss": 2.5492, "step": 12320 }, { "epoch": 4.277828091758806e-14, "grad_norm": 0.5609793663024902, "learning_rate": 0.000119386438674271, "loss": 2.5697, "step": 12330 }, { "epoch": 4.28129753871076e-14, "grad_norm": 0.5404332280158997, "learning_rate": 0.00011937669367566265, "loss": 2.5513, "step": 12340 }, { "epoch": 4.2847669856627135e-14, "grad_norm": 0.5572315454483032, "learning_rate": 0.00011936694067087845, "loss": 2.5669, "step": 12350 }, { "epoch": 4.288236432614667e-14, "grad_norm": 0.5386480093002319, "learning_rate": 0.00011935717966146081, "loss": 2.521, "step": 12360 }, { "epoch": 4.291705879566621e-14, "grad_norm": 0.5758058428764343, "learning_rate": 0.00011934741064895335, "loss": 2.5402, "step": 12370 }, { "epoch": 4.2951753265185744e-14, "grad_norm": 0.5335978865623474, "learning_rate": 0.00011933763363490095, "loss": 2.5763, "step": 12380 }, { "epoch": 4.298644773470528e-14, "grad_norm": 0.5485419631004333, "learning_rate": 0.00011932784862084972, "loss": 2.5944, "step": 12390 }, { "epoch": 4.3021142204224816e-14, "grad_norm": 0.5326485633850098, "learning_rate": 0.00011931805560834714, "loss": 2.5689, "step": 12400 }, { "epoch": 4.305583667374435e-14, "grad_norm": 0.5450775623321533, "learning_rate": 0.00011930825459894183, "loss": 2.5603, "step": 12410 }, { "epoch": 4.309053114326389e-14, "grad_norm": 0.5459516048431396, "learning_rate": 0.00011929844559418378, "loss": 2.5111, "step": 12420 }, { "epoch": 4.3125225612783424e-14, "grad_norm": 0.5413381457328796, "learning_rate": 0.00011928862859562419, "loss": 2.5713, "step": 12430 }, { "epoch": 4.315992008230296e-14, "grad_norm": 0.5578582286834717, "learning_rate": 0.00011927880360481551, "loss": 2.5772, "step": 12440 }, { "epoch": 4.3194614551822497e-14, "grad_norm": 0.545028805732727, "learning_rate": 0.00011926897062331149, "loss": 2.5885, "step": 12450 }, { "epoch": 4.322930902134203e-14, "grad_norm": 0.5457649827003479, "learning_rate": 0.00011925912965266716, "loss": 2.5687, "step": 12460 }, { "epoch": 4.326400349086157e-14, "grad_norm": 0.543074369430542, "learning_rate": 0.00011924928069443874, "loss": 2.5598, "step": 12470 }, { "epoch": 4.3298697960381105e-14, "grad_norm": 0.5741780400276184, "learning_rate": 0.0001192394237501838, "loss": 2.5875, "step": 12480 }, { "epoch": 4.333339242990064e-14, "grad_norm": 0.5951067805290222, "learning_rate": 0.00011922955882146109, "loss": 2.5825, "step": 12490 }, { "epoch": 4.336808689942018e-14, "grad_norm": 0.550504207611084, "learning_rate": 0.0001192196859098307, "loss": 2.5716, "step": 12500 }, { "epoch": 4.3402781368939714e-14, "grad_norm": 0.5217750668525696, "learning_rate": 0.00011920980501685391, "loss": 2.5397, "step": 12510 }, { "epoch": 4.343747583845925e-14, "grad_norm": 0.5482033491134644, "learning_rate": 0.00011919991614409332, "loss": 2.5652, "step": 12520 }, { "epoch": 4.3472170307978786e-14, "grad_norm": 0.544506311416626, "learning_rate": 0.00011919001929311279, "loss": 2.5249, "step": 12530 }, { "epoch": 4.350686477749832e-14, "grad_norm": 0.5416708588600159, "learning_rate": 0.00011918011446547737, "loss": 2.5702, "step": 12540 }, { "epoch": 4.354155924701786e-14, "grad_norm": 0.5372009873390198, "learning_rate": 0.00011917020166275347, "loss": 2.5318, "step": 12550 }, { "epoch": 4.3576253716537394e-14, "grad_norm": 0.5477988719940186, "learning_rate": 0.0001191602808865087, "loss": 2.6156, "step": 12560 }, { "epoch": 4.361094818605693e-14, "grad_norm": 0.6170890927314758, "learning_rate": 0.00011915035213831195, "loss": 2.5704, "step": 12570 }, { "epoch": 4.3645642655576466e-14, "grad_norm": 0.5358986258506775, "learning_rate": 0.00011914041541973335, "loss": 2.5697, "step": 12580 }, { "epoch": 4.3680337125096e-14, "grad_norm": 0.576802134513855, "learning_rate": 0.0001191304707323443, "loss": 2.5289, "step": 12590 }, { "epoch": 4.371503159461554e-14, "grad_norm": 0.5836868286132812, "learning_rate": 0.0001191205180777175, "loss": 2.6033, "step": 12600 }, { "epoch": 4.3749726064135075e-14, "grad_norm": 0.5993554592132568, "learning_rate": 0.00011911055745742683, "loss": 2.5472, "step": 12610 }, { "epoch": 4.378442053365461e-14, "grad_norm": 0.5555655360221863, "learning_rate": 0.00011910058887304754, "loss": 2.5773, "step": 12620 }, { "epoch": 4.381911500317415e-14, "grad_norm": 0.5551588535308838, "learning_rate": 0.00011909061232615603, "loss": 2.591, "step": 12630 }, { "epoch": 4.3853809472693683e-14, "grad_norm": 0.5497987270355225, "learning_rate": 0.00011908062781833002, "loss": 2.5379, "step": 12640 }, { "epoch": 4.388850394221322e-14, "grad_norm": 0.5513547658920288, "learning_rate": 0.00011907063535114845, "loss": 2.5692, "step": 12650 }, { "epoch": 4.3923198411732756e-14, "grad_norm": 0.5665968060493469, "learning_rate": 0.00011906063492619158, "loss": 2.533, "step": 12660 }, { "epoch": 4.395789288125229e-14, "grad_norm": 0.5445855259895325, "learning_rate": 0.00011905062654504087, "loss": 2.5477, "step": 12670 }, { "epoch": 4.399258735077183e-14, "grad_norm": 0.5625795125961304, "learning_rate": 0.00011904061020927906, "loss": 2.5808, "step": 12680 }, { "epoch": 4.4027281820291364e-14, "grad_norm": 0.5551760792732239, "learning_rate": 0.00011903058592049014, "loss": 2.5157, "step": 12690 }, { "epoch": 4.40619762898109e-14, "grad_norm": 0.5484020113945007, "learning_rate": 0.00011902055368025938, "loss": 2.5472, "step": 12700 }, { "epoch": 4.4096670759330436e-14, "grad_norm": 0.5605646967887878, "learning_rate": 0.00011901051349017329, "loss": 2.5651, "step": 12710 }, { "epoch": 4.413136522884997e-14, "grad_norm": 0.5442164540290833, "learning_rate": 0.00011900046535181967, "loss": 2.5699, "step": 12720 }, { "epoch": 4.416605969836951e-14, "grad_norm": 0.5594670176506042, "learning_rate": 0.00011899040926678746, "loss": 2.5565, "step": 12730 }, { "epoch": 4.4200754167889045e-14, "grad_norm": 0.5440113544464111, "learning_rate": 0.00011898034523666703, "loss": 2.5732, "step": 12740 }, { "epoch": 4.423544863740858e-14, "grad_norm": 0.5331368446350098, "learning_rate": 0.00011897027326304988, "loss": 2.5062, "step": 12750 }, { "epoch": 4.427014310692812e-14, "grad_norm": 0.5319521427154541, "learning_rate": 0.00011896019334752881, "loss": 2.5629, "step": 12760 }, { "epoch": 4.4304837576447653e-14, "grad_norm": 0.5547661781311035, "learning_rate": 0.00011895010549169786, "loss": 2.5633, "step": 12770 }, { "epoch": 4.433953204596719e-14, "grad_norm": 0.5399665832519531, "learning_rate": 0.00011894000969715238, "loss": 2.5315, "step": 12780 }, { "epoch": 4.4374226515486725e-14, "grad_norm": 0.5404934883117676, "learning_rate": 0.00011892990596548888, "loss": 2.5485, "step": 12790 }, { "epoch": 4.440892098500626e-14, "grad_norm": 0.5540802478790283, "learning_rate": 0.00011891979429830521, "loss": 2.536, "step": 12800 }, { "epoch": 4.44436154545258e-14, "grad_norm": 0.5392959713935852, "learning_rate": 0.00011890967469720045, "loss": 2.5545, "step": 12810 }, { "epoch": 4.4478309924045334e-14, "grad_norm": 0.5366972088813782, "learning_rate": 0.00011889954716377489, "loss": 2.6046, "step": 12820 }, { "epoch": 4.451300439356487e-14, "grad_norm": 0.5446826219558716, "learning_rate": 0.00011888941169963015, "loss": 2.5403, "step": 12830 }, { "epoch": 4.4547698863084406e-14, "grad_norm": 0.5543631315231323, "learning_rate": 0.00011887926830636906, "loss": 2.5539, "step": 12840 }, { "epoch": 4.458239333260394e-14, "grad_norm": 0.5544484853744507, "learning_rate": 0.00011886911698559569, "loss": 2.5636, "step": 12850 }, { "epoch": 4.461708780212348e-14, "grad_norm": 0.5458822250366211, "learning_rate": 0.0001188589577389154, "loss": 2.5697, "step": 12860 }, { "epoch": 4.4651782271643015e-14, "grad_norm": 0.5490422248840332, "learning_rate": 0.00011884879056793479, "loss": 2.5463, "step": 12870 }, { "epoch": 4.468647674116255e-14, "grad_norm": 0.5526748299598694, "learning_rate": 0.00011883861547426171, "loss": 2.5622, "step": 12880 }, { "epoch": 4.472117121068209e-14, "grad_norm": 0.5283459424972534, "learning_rate": 0.00011882843245950524, "loss": 2.535, "step": 12890 }, { "epoch": 4.475586568020162e-14, "grad_norm": 0.5361932516098022, "learning_rate": 0.00011881824152527577, "loss": 2.5145, "step": 12900 }, { "epoch": 4.479056014972116e-14, "grad_norm": 0.5794731974601746, "learning_rate": 0.00011880804267318488, "loss": 2.5802, "step": 12910 }, { "epoch": 4.4825254619240695e-14, "grad_norm": 7.104547500610352, "learning_rate": 0.00011879783590484547, "loss": 2.5705, "step": 12920 }, { "epoch": 4.485994908876023e-14, "grad_norm": 0.5765463709831238, "learning_rate": 0.00011878762122187162, "loss": 2.5578, "step": 12930 }, { "epoch": 4.489464355827977e-14, "grad_norm": 0.5447003841400146, "learning_rate": 0.00011877739862587868, "loss": 2.5446, "step": 12940 }, { "epoch": 4.4929338027799304e-14, "grad_norm": 0.525709867477417, "learning_rate": 0.0001187671681184833, "loss": 2.5813, "step": 12950 }, { "epoch": 4.496403249731884e-14, "grad_norm": 0.5599184632301331, "learning_rate": 0.00011875692970130331, "loss": 2.5187, "step": 12960 }, { "epoch": 4.4998726966838376e-14, "grad_norm": 0.5611594915390015, "learning_rate": 0.00011874668337595786, "loss": 2.5478, "step": 12970 }, { "epoch": 4.503342143635791e-14, "grad_norm": 0.5437379479408264, "learning_rate": 0.00011873642914406734, "loss": 2.5292, "step": 12980 }, { "epoch": 4.506811590587745e-14, "grad_norm": 0.5407551527023315, "learning_rate": 0.00011872616700725332, "loss": 2.5449, "step": 12990 }, { "epoch": 4.5102810375396984e-14, "grad_norm": 0.5224987268447876, "learning_rate": 0.00011871589696713867, "loss": 2.5819, "step": 13000 }, { "epoch": 4.513750484491652e-14, "grad_norm": 0.5913327932357788, "learning_rate": 0.00011870561902534752, "loss": 2.542, "step": 13010 }, { "epoch": 4.5172199314436057e-14, "grad_norm": 0.5573772192001343, "learning_rate": 0.00011869533318350527, "loss": 2.5744, "step": 13020 }, { "epoch": 4.520689378395559e-14, "grad_norm": 0.6021262407302856, "learning_rate": 0.00011868503944323849, "loss": 2.5295, "step": 13030 }, { "epoch": 4.524158825347513e-14, "grad_norm": 0.5644800066947937, "learning_rate": 0.00011867473780617508, "loss": 2.5944, "step": 13040 }, { "epoch": 4.5276282722994665e-14, "grad_norm": 0.567125141620636, "learning_rate": 0.00011866442827394414, "loss": 2.5581, "step": 13050 }, { "epoch": 4.53109771925142e-14, "grad_norm": 0.5159335136413574, "learning_rate": 0.00011865411084817602, "loss": 2.5284, "step": 13060 }, { "epoch": 4.534567166203374e-14, "grad_norm": 0.5951399803161621, "learning_rate": 0.00011864378553050235, "loss": 2.5814, "step": 13070 }, { "epoch": 4.5380366131553274e-14, "grad_norm": 0.5903134942054749, "learning_rate": 0.000118633452322556, "loss": 2.5368, "step": 13080 }, { "epoch": 4.541506060107281e-14, "grad_norm": 0.5388596653938293, "learning_rate": 0.00011862311122597105, "loss": 2.5379, "step": 13090 }, { "epoch": 4.5449755070592346e-14, "grad_norm": 0.5322829484939575, "learning_rate": 0.00011861276224238288, "loss": 2.5366, "step": 13100 }, { "epoch": 4.548444954011188e-14, "grad_norm": 0.5227189660072327, "learning_rate": 0.00011860240537342808, "loss": 2.5262, "step": 13110 }, { "epoch": 4.551914400963142e-14, "grad_norm": 0.5645536184310913, "learning_rate": 0.00011859204062074451, "loss": 2.5399, "step": 13120 }, { "epoch": 4.5553838479150954e-14, "grad_norm": 0.5606644153594971, "learning_rate": 0.00011858166798597123, "loss": 2.5406, "step": 13130 }, { "epoch": 4.558853294867049e-14, "grad_norm": 0.5144214034080505, "learning_rate": 0.00011857128747074865, "loss": 2.5503, "step": 13140 }, { "epoch": 4.5623227418190027e-14, "grad_norm": 0.5377312898635864, "learning_rate": 0.0001185608990767183, "loss": 2.5131, "step": 13150 }, { "epoch": 4.565792188770956e-14, "grad_norm": 0.5467860102653503, "learning_rate": 0.00011855050280552303, "loss": 2.5501, "step": 13160 }, { "epoch": 4.56926163572291e-14, "grad_norm": 0.5503011345863342, "learning_rate": 0.00011854009865880695, "loss": 2.5418, "step": 13170 }, { "epoch": 4.5727310826748635e-14, "grad_norm": 0.5237338542938232, "learning_rate": 0.00011852968663821534, "loss": 2.543, "step": 13180 }, { "epoch": 4.576200529626817e-14, "grad_norm": 0.5474035143852234, "learning_rate": 0.0001185192667453948, "loss": 2.5356, "step": 13190 }, { "epoch": 4.579669976578771e-14, "grad_norm": 0.5297275185585022, "learning_rate": 0.00011850883898199316, "loss": 2.566, "step": 13200 }, { "epoch": 4.5831394235307243e-14, "grad_norm": 0.5565235614776611, "learning_rate": 0.00011849840334965945, "loss": 2.5405, "step": 13210 }, { "epoch": 4.586608870482678e-14, "grad_norm": 0.5631586909294128, "learning_rate": 0.00011848795985004399, "loss": 2.5561, "step": 13220 }, { "epoch": 4.5900783174346316e-14, "grad_norm": 0.555429220199585, "learning_rate": 0.0001184775084847983, "loss": 2.5432, "step": 13230 }, { "epoch": 4.593547764386585e-14, "grad_norm": 0.5860438346862793, "learning_rate": 0.00011846704925557523, "loss": 2.5861, "step": 13240 }, { "epoch": 4.597017211338539e-14, "grad_norm": 0.5342402458190918, "learning_rate": 0.00011845658216402878, "loss": 2.5281, "step": 13250 }, { "epoch": 4.6004866582904924e-14, "grad_norm": 0.5432714223861694, "learning_rate": 0.00011844610721181424, "loss": 2.5266, "step": 13260 }, { "epoch": 4.603956105242446e-14, "grad_norm": 0.5556492209434509, "learning_rate": 0.00011843562440058813, "loss": 2.5047, "step": 13270 }, { "epoch": 4.6074255521943996e-14, "grad_norm": 0.5264197587966919, "learning_rate": 0.00011842513373200822, "loss": 2.5583, "step": 13280 }, { "epoch": 4.610894999146353e-14, "grad_norm": 0.5377755761146545, "learning_rate": 0.00011841463520773352, "loss": 2.5307, "step": 13290 }, { "epoch": 4.614364446098307e-14, "grad_norm": 0.5470077991485596, "learning_rate": 0.00011840412882942426, "loss": 2.5354, "step": 13300 }, { "epoch": 4.6178338930502605e-14, "grad_norm": 0.5155746936798096, "learning_rate": 0.00011839361459874198, "loss": 2.5659, "step": 13310 }, { "epoch": 4.621303340002214e-14, "grad_norm": 0.5799461603164673, "learning_rate": 0.00011838309251734937, "loss": 2.5316, "step": 13320 }, { "epoch": 4.624772786954168e-14, "grad_norm": 0.5353295803070068, "learning_rate": 0.00011837256258691043, "loss": 2.5629, "step": 13330 }, { "epoch": 4.6282422339061213e-14, "grad_norm": 0.5570316910743713, "learning_rate": 0.00011836202480909036, "loss": 2.5389, "step": 13340 }, { "epoch": 4.631711680858075e-14, "grad_norm": 0.5827422738075256, "learning_rate": 0.00011835147918555562, "loss": 2.5162, "step": 13350 }, { "epoch": 4.6351811278100286e-14, "grad_norm": 0.5675159096717834, "learning_rate": 0.00011834092571797394, "loss": 2.5267, "step": 13360 }, { "epoch": 4.638650574761982e-14, "grad_norm": 0.5278079509735107, "learning_rate": 0.00011833036440801424, "loss": 2.5438, "step": 13370 }, { "epoch": 4.642120021713936e-14, "grad_norm": 0.5489444732666016, "learning_rate": 0.00011831979525734667, "loss": 2.5509, "step": 13380 }, { "epoch": 4.6455894686658894e-14, "grad_norm": 0.5457969903945923, "learning_rate": 0.00011830921826764272, "loss": 2.5517, "step": 13390 }, { "epoch": 4.649058915617843e-14, "grad_norm": 0.574812114238739, "learning_rate": 0.000118298633440575, "loss": 2.5399, "step": 13400 }, { "epoch": 4.6525283625697966e-14, "grad_norm": 0.600325345993042, "learning_rate": 0.0001182880407778174, "loss": 2.5549, "step": 13410 }, { "epoch": 4.65599780952175e-14, "grad_norm": 0.5630314350128174, "learning_rate": 0.0001182774402810451, "loss": 2.5399, "step": 13420 }, { "epoch": 4.659467256473704e-14, "grad_norm": 0.5261701345443726, "learning_rate": 0.00011826683195193442, "loss": 2.5478, "step": 13430 }, { "epoch": 4.6629367034256575e-14, "grad_norm": 0.5507047772407532, "learning_rate": 0.00011825621579216304, "loss": 2.5468, "step": 13440 }, { "epoch": 4.666406150377611e-14, "grad_norm": 0.560580849647522, "learning_rate": 0.0001182455918034098, "loss": 2.5742, "step": 13450 }, { "epoch": 4.669875597329565e-14, "grad_norm": 0.5592156052589417, "learning_rate": 0.00011823495998735474, "loss": 2.5307, "step": 13460 }, { "epoch": 4.673345044281518e-14, "grad_norm": 0.542646050453186, "learning_rate": 0.00011822432034567925, "loss": 2.5397, "step": 13470 }, { "epoch": 4.676814491233472e-14, "grad_norm": 0.6085817217826843, "learning_rate": 0.00011821367288006588, "loss": 2.5542, "step": 13480 }, { "epoch": 4.6802839381854255e-14, "grad_norm": 0.5192282199859619, "learning_rate": 0.0001182030175921984, "loss": 2.5331, "step": 13490 }, { "epoch": 4.683753385137379e-14, "grad_norm": 0.557971715927124, "learning_rate": 0.00011819235448376189, "loss": 2.5191, "step": 13500 }, { "epoch": 4.687222832089333e-14, "grad_norm": 0.7041448950767517, "learning_rate": 0.00011818168355644264, "loss": 2.546, "step": 13510 }, { "epoch": 4.6906922790412864e-14, "grad_norm": 0.5469762682914734, "learning_rate": 0.00011817100481192814, "loss": 2.5468, "step": 13520 }, { "epoch": 4.69416172599324e-14, "grad_norm": 0.5436945557594299, "learning_rate": 0.00011816031825190715, "loss": 2.5615, "step": 13530 }, { "epoch": 4.6976311729451936e-14, "grad_norm": 0.5268374085426331, "learning_rate": 0.00011814962387806966, "loss": 2.5444, "step": 13540 }, { "epoch": 4.701100619897147e-14, "grad_norm": 0.5531018376350403, "learning_rate": 0.00011813892169210687, "loss": 2.5217, "step": 13550 }, { "epoch": 4.704570066849101e-14, "grad_norm": 0.5561888217926025, "learning_rate": 0.00011812821169571127, "loss": 2.529, "step": 13560 }, { "epoch": 4.7080395138010545e-14, "grad_norm": 0.5564419031143188, "learning_rate": 0.00011811749389057654, "loss": 2.5508, "step": 13570 }, { "epoch": 4.711508960753008e-14, "grad_norm": 0.5525124669075012, "learning_rate": 0.0001181067682783976, "loss": 2.5095, "step": 13580 }, { "epoch": 4.7149784077049617e-14, "grad_norm": 0.5535767674446106, "learning_rate": 0.00011809603486087064, "loss": 2.5357, "step": 13590 }, { "epoch": 4.718447854656915e-14, "grad_norm": 0.5175579190254211, "learning_rate": 0.00011808529363969304, "loss": 2.5362, "step": 13600 }, { "epoch": 4.721917301608869e-14, "grad_norm": 0.544497013092041, "learning_rate": 0.00011807454461656342, "loss": 2.5258, "step": 13610 }, { "epoch": 4.7253867485608225e-14, "grad_norm": 0.5184549689292908, "learning_rate": 0.00011806378779318165, "loss": 2.5387, "step": 13620 }, { "epoch": 4.728856195512776e-14, "grad_norm": 0.5528564453125, "learning_rate": 0.00011805302317124884, "loss": 2.5671, "step": 13630 }, { "epoch": 4.73232564246473e-14, "grad_norm": 0.5679931044578552, "learning_rate": 0.00011804225075246733, "loss": 2.4949, "step": 13640 }, { "epoch": 4.7357950894166834e-14, "grad_norm": 0.5582529902458191, "learning_rate": 0.00011803147053854068, "loss": 2.5477, "step": 13650 }, { "epoch": 4.739264536368637e-14, "grad_norm": 0.570791482925415, "learning_rate": 0.00011802068253117367, "loss": 2.5386, "step": 13660 }, { "epoch": 4.7427339833205906e-14, "grad_norm": 0.5492092967033386, "learning_rate": 0.00011800988673207233, "loss": 2.5189, "step": 13670 }, { "epoch": 4.746203430272544e-14, "grad_norm": 0.5511994957923889, "learning_rate": 0.00011799908314294395, "loss": 2.5212, "step": 13680 }, { "epoch": 4.749672877224498e-14, "grad_norm": 0.5775168538093567, "learning_rate": 0.00011798827176549701, "loss": 2.5033, "step": 13690 }, { "epoch": 4.7531423241764514e-14, "grad_norm": 0.5638816356658936, "learning_rate": 0.00011797745260144124, "loss": 2.527, "step": 13700 }, { "epoch": 4.756611771128405e-14, "grad_norm": 0.5594960451126099, "learning_rate": 0.00011796662565248758, "loss": 2.5616, "step": 13710 }, { "epoch": 4.7600812180803587e-14, "grad_norm": 0.568459153175354, "learning_rate": 0.00011795579092034825, "loss": 2.5446, "step": 13720 }, { "epoch": 4.763550665032312e-14, "grad_norm": 0.538184642791748, "learning_rate": 0.00011794494840673666, "loss": 2.5655, "step": 13730 }, { "epoch": 4.767020111984266e-14, "grad_norm": 0.5524560809135437, "learning_rate": 0.00011793409811336743, "loss": 2.5376, "step": 13740 }, { "epoch": 4.7704895589362195e-14, "grad_norm": 0.5642053484916687, "learning_rate": 0.00011792324004195646, "loss": 2.498, "step": 13750 }, { "epoch": 4.773959005888173e-14, "grad_norm": 0.5865797996520996, "learning_rate": 0.00011791237419422089, "loss": 2.5506, "step": 13760 }, { "epoch": 4.777428452840127e-14, "grad_norm": 0.5437580347061157, "learning_rate": 0.00011790150057187901, "loss": 2.5064, "step": 13770 }, { "epoch": 4.7808978997920804e-14, "grad_norm": 0.5406655669212341, "learning_rate": 0.00011789061917665042, "loss": 2.5491, "step": 13780 }, { "epoch": 4.784367346744034e-14, "grad_norm": 0.5137045979499817, "learning_rate": 0.00011787973001025591, "loss": 2.4832, "step": 13790 }, { "epoch": 4.7878367936959876e-14, "grad_norm": 0.5401195883750916, "learning_rate": 0.00011786883307441752, "loss": 2.5326, "step": 13800 }, { "epoch": 4.791306240647941e-14, "grad_norm": 0.5628296136856079, "learning_rate": 0.0001178579283708585, "loss": 2.5517, "step": 13810 }, { "epoch": 4.794775687599895e-14, "grad_norm": 0.5861107707023621, "learning_rate": 0.00011784701590130332, "loss": 2.5319, "step": 13820 }, { "epoch": 4.7982451345518484e-14, "grad_norm": 0.6918483972549438, "learning_rate": 0.0001178360956674777, "loss": 2.5749, "step": 13830 }, { "epoch": 4.801714581503802e-14, "grad_norm": 0.5728329420089722, "learning_rate": 0.00011782516767110861, "loss": 2.4835, "step": 13840 }, { "epoch": 4.8051840284557557e-14, "grad_norm": 0.5569953918457031, "learning_rate": 0.00011781423191392418, "loss": 2.5176, "step": 13850 }, { "epoch": 4.808653475407709e-14, "grad_norm": 0.5549789071083069, "learning_rate": 0.0001178032883976538, "loss": 2.5098, "step": 13860 }, { "epoch": 4.812122922359663e-14, "grad_norm": 0.5618237257003784, "learning_rate": 0.00011779233712402813, "loss": 2.5478, "step": 13870 }, { "epoch": 4.8155923693116165e-14, "grad_norm": 0.5335788130760193, "learning_rate": 0.00011778137809477901, "loss": 2.5139, "step": 13880 }, { "epoch": 4.81906181626357e-14, "grad_norm": 0.5943315625190735, "learning_rate": 0.00011777041131163953, "loss": 2.522, "step": 13890 }, { "epoch": 4.822531263215524e-14, "grad_norm": 0.5334214568138123, "learning_rate": 0.00011775943677634394, "loss": 2.5352, "step": 13900 }, { "epoch": 4.8260007101674773e-14, "grad_norm": 0.5536807179450989, "learning_rate": 0.0001177484544906278, "loss": 2.5749, "step": 13910 }, { "epoch": 4.829470157119431e-14, "grad_norm": 0.5392267107963562, "learning_rate": 0.00011773746445622789, "loss": 2.5006, "step": 13920 }, { "epoch": 4.8329396040713846e-14, "grad_norm": 0.5562549233436584, "learning_rate": 0.00011772646667488213, "loss": 2.5412, "step": 13930 }, { "epoch": 4.836409051023338e-14, "grad_norm": 0.5462127327919006, "learning_rate": 0.00011771546114832979, "loss": 2.5356, "step": 13940 }, { "epoch": 4.839878497975292e-14, "grad_norm": 0.5644789934158325, "learning_rate": 0.00011770444787831123, "loss": 2.4976, "step": 13950 }, { "epoch": 4.8433479449272454e-14, "grad_norm": 0.5676937103271484, "learning_rate": 0.00011769342686656816, "loss": 2.5187, "step": 13960 }, { "epoch": 4.846817391879199e-14, "grad_norm": 0.5907922387123108, "learning_rate": 0.00011768239811484344, "loss": 2.5435, "step": 13970 }, { "epoch": 4.8502868388311526e-14, "grad_norm": 0.5474948287010193, "learning_rate": 0.00011767136162488118, "loss": 2.5563, "step": 13980 }, { "epoch": 4.853756285783106e-14, "grad_norm": 0.5348525643348694, "learning_rate": 0.00011766031739842668, "loss": 2.5499, "step": 13990 }, { "epoch": 4.85722573273506e-14, "grad_norm": 0.5447676777839661, "learning_rate": 0.00011764926543722652, "loss": 2.5085, "step": 14000 }, { "epoch": 4.8606951796870135e-14, "grad_norm": 0.5584738254547119, "learning_rate": 0.00011763820574302844, "loss": 2.5459, "step": 14010 }, { "epoch": 4.864164626638967e-14, "grad_norm": 0.5280540585517883, "learning_rate": 0.00011762713831758147, "loss": 2.4983, "step": 14020 }, { "epoch": 4.867634073590921e-14, "grad_norm": 0.5689372420310974, "learning_rate": 0.00011761606316263581, "loss": 2.5364, "step": 14030 }, { "epoch": 4.8711035205428743e-14, "grad_norm": 0.5632480382919312, "learning_rate": 0.0001176049802799429, "loss": 2.5038, "step": 14040 }, { "epoch": 4.874572967494828e-14, "grad_norm": 0.5558111667633057, "learning_rate": 0.00011759388967125543, "loss": 2.5271, "step": 14050 }, { "epoch": 4.8780424144467815e-14, "grad_norm": 0.5551756024360657, "learning_rate": 0.00011758279133832724, "loss": 2.508, "step": 14060 }, { "epoch": 4.881511861398735e-14, "grad_norm": 0.5468283891677856, "learning_rate": 0.00011757168528291344, "loss": 2.5244, "step": 14070 }, { "epoch": 4.884981308350689e-14, "grad_norm": 0.5360465049743652, "learning_rate": 0.00011756057150677038, "loss": 2.519, "step": 14080 }, { "epoch": 4.8884507553026424e-14, "grad_norm": 0.5384688377380371, "learning_rate": 0.00011754945001165561, "loss": 2.4744, "step": 14090 }, { "epoch": 4.891920202254596e-14, "grad_norm": 0.5573025941848755, "learning_rate": 0.00011753832079932788, "loss": 2.5108, "step": 14100 }, { "epoch": 4.8953896492065496e-14, "grad_norm": 0.5718728303909302, "learning_rate": 0.0001175271838715472, "loss": 2.5174, "step": 14110 }, { "epoch": 4.898859096158503e-14, "grad_norm": 0.557177722454071, "learning_rate": 0.00011751603923007476, "loss": 2.5246, "step": 14120 }, { "epoch": 4.902328543110457e-14, "grad_norm": 0.5193539261817932, "learning_rate": 0.000117504886876673, "loss": 2.5064, "step": 14130 }, { "epoch": 4.9057979900624105e-14, "grad_norm": 0.5165082812309265, "learning_rate": 0.00011749372681310556, "loss": 2.52, "step": 14140 }, { "epoch": 4.909267437014364e-14, "grad_norm": 0.5403398275375366, "learning_rate": 0.00011748255904113731, "loss": 2.5157, "step": 14150 }, { "epoch": 4.912736883966318e-14, "grad_norm": 0.5203744173049927, "learning_rate": 0.00011747138356253436, "loss": 2.5027, "step": 14160 }, { "epoch": 4.916206330918271e-14, "grad_norm": 0.5413910746574402, "learning_rate": 0.00011746020037906397, "loss": 2.484, "step": 14170 }, { "epoch": 4.919675777870225e-14, "grad_norm": 0.5811920166015625, "learning_rate": 0.0001174490094924947, "loss": 2.5499, "step": 14180 }, { "epoch": 4.9231452248221785e-14, "grad_norm": 0.550523042678833, "learning_rate": 0.00011743781090459629, "loss": 2.5435, "step": 14190 }, { "epoch": 4.926614671774132e-14, "grad_norm": 0.5374614596366882, "learning_rate": 0.00011742660461713969, "loss": 2.4986, "step": 14200 }, { "epoch": 4.930084118726086e-14, "grad_norm": 0.560251772403717, "learning_rate": 0.0001174153906318971, "loss": 2.5181, "step": 14210 }, { "epoch": 4.9335535656780394e-14, "grad_norm": 0.5231373310089111, "learning_rate": 0.00011740416895064188, "loss": 2.5252, "step": 14220 }, { "epoch": 4.937023012629993e-14, "grad_norm": 0.5296940207481384, "learning_rate": 0.00011739293957514864, "loss": 2.5349, "step": 14230 }, { "epoch": 4.9404924595819466e-14, "grad_norm": 0.5440646409988403, "learning_rate": 0.00011738170250719324, "loss": 2.5141, "step": 14240 }, { "epoch": 4.9439619065339e-14, "grad_norm": 0.5452266931533813, "learning_rate": 0.00011737045774855272, "loss": 2.5511, "step": 14250 }, { "epoch": 4.947431353485854e-14, "grad_norm": 0.5793325901031494, "learning_rate": 0.00011735920530100533, "loss": 2.4995, "step": 14260 }, { "epoch": 4.9509008004378074e-14, "grad_norm": 0.5755615234375, "learning_rate": 0.00011734794516633056, "loss": 2.5098, "step": 14270 }, { "epoch": 4.954370247389761e-14, "grad_norm": 0.5606346130371094, "learning_rate": 0.00011733667734630908, "loss": 2.4835, "step": 14280 }, { "epoch": 4.9578396943417147e-14, "grad_norm": 0.5302919149398804, "learning_rate": 0.00011732540184272284, "loss": 2.5316, "step": 14290 }, { "epoch": 4.961309141293668e-14, "grad_norm": 0.5558198690414429, "learning_rate": 0.0001173141186573549, "loss": 2.5182, "step": 14300 }, { "epoch": 4.964778588245622e-14, "grad_norm": 0.5240646600723267, "learning_rate": 0.00011730282779198965, "loss": 2.5565, "step": 14310 }, { "epoch": 4.9682480351975755e-14, "grad_norm": 0.591363787651062, "learning_rate": 0.00011729152924841266, "loss": 2.4956, "step": 14320 }, { "epoch": 4.971717482149529e-14, "grad_norm": 0.5587567090988159, "learning_rate": 0.00011728022302841064, "loss": 2.4973, "step": 14330 }, { "epoch": 4.975186929101483e-14, "grad_norm": 0.5328636169433594, "learning_rate": 0.00011726890913377162, "loss": 2.5186, "step": 14340 }, { "epoch": 4.9786563760534364e-14, "grad_norm": 0.5454198718070984, "learning_rate": 0.00011725758756628474, "loss": 2.4948, "step": 14350 }, { "epoch": 4.98212582300539e-14, "grad_norm": 0.5464425086975098, "learning_rate": 0.00011724625832774047, "loss": 2.5424, "step": 14360 }, { "epoch": 4.9855952699573436e-14, "grad_norm": 0.5812118053436279, "learning_rate": 0.0001172349214199304, "loss": 2.4893, "step": 14370 }, { "epoch": 4.989064716909297e-14, "grad_norm": 0.5706376433372498, "learning_rate": 0.00011722357684464736, "loss": 2.5581, "step": 14380 }, { "epoch": 4.992534163861251e-14, "grad_norm": 0.5470327138900757, "learning_rate": 0.00011721222460368541, "loss": 2.5068, "step": 14390 }, { "epoch": 4.9960036108132044e-14, "grad_norm": 0.5680293440818787, "learning_rate": 0.00011720086469883982, "loss": 2.4929, "step": 14400 }, { "epoch": 4.999473057765158e-14, "grad_norm": 0.5574185252189636, "learning_rate": 0.00011718949713190702, "loss": 2.5106, "step": 14410 }, { "epoch": 5.0029425047171117e-14, "grad_norm": 0.5403979420661926, "learning_rate": 0.00011717812190468475, "loss": 2.501, "step": 14420 }, { "epoch": 5.006411951669065e-14, "grad_norm": 0.5305143594741821, "learning_rate": 0.00011716673901897183, "loss": 2.513, "step": 14430 }, { "epoch": 5.009881398621019e-14, "grad_norm": 0.5774564146995544, "learning_rate": 0.00011715534847656843, "loss": 2.5108, "step": 14440 }, { "epoch": 5.0133508455729725e-14, "grad_norm": 0.5324370265007019, "learning_rate": 0.00011714395027927585, "loss": 2.5155, "step": 14450 }, { "epoch": 5.016820292524926e-14, "grad_norm": 0.5454780459403992, "learning_rate": 0.00011713254442889661, "loss": 2.5103, "step": 14460 }, { "epoch": 5.02028973947688e-14, "grad_norm": 0.5531365871429443, "learning_rate": 0.00011712113092723445, "loss": 2.478, "step": 14470 }, { "epoch": 5.0237591864288333e-14, "grad_norm": 0.5655040144920349, "learning_rate": 0.0001171097097760943, "loss": 2.5431, "step": 14480 }, { "epoch": 5.027228633380787e-14, "grad_norm": 0.5472716689109802, "learning_rate": 0.00011709828097728234, "loss": 2.5329, "step": 14490 }, { "epoch": 5.0306980803327406e-14, "grad_norm": 0.5462208986282349, "learning_rate": 0.00011708684453260594, "loss": 2.4932, "step": 14500 }, { "epoch": 5.034167527284694e-14, "grad_norm": 0.5515497326850891, "learning_rate": 0.00011707540044387366, "loss": 2.5319, "step": 14510 }, { "epoch": 5.037636974236648e-14, "grad_norm": 0.5347950458526611, "learning_rate": 0.00011706394871289529, "loss": 2.5216, "step": 14520 }, { "epoch": 5.0411064211886014e-14, "grad_norm": 0.534930408000946, "learning_rate": 0.0001170524893414818, "loss": 2.5266, "step": 14530 }, { "epoch": 5.044575868140555e-14, "grad_norm": 0.558034360408783, "learning_rate": 0.00011704102233144543, "loss": 2.5468, "step": 14540 }, { "epoch": 5.0480453150925086e-14, "grad_norm": 0.5500307083129883, "learning_rate": 0.00011702954768459956, "loss": 2.4736, "step": 14550 }, { "epoch": 5.051514762044462e-14, "grad_norm": 0.5484467148780823, "learning_rate": 0.00011701806540275883, "loss": 2.5376, "step": 14560 }, { "epoch": 5.054984208996416e-14, "grad_norm": 0.5651101469993591, "learning_rate": 0.00011700657548773903, "loss": 2.5272, "step": 14570 }, { "epoch": 5.0584536559483695e-14, "grad_norm": 0.5475466251373291, "learning_rate": 0.00011699507794135723, "loss": 2.533, "step": 14580 }, { "epoch": 5.061923102900323e-14, "grad_norm": 0.5710880756378174, "learning_rate": 0.00011698357276543163, "loss": 2.53, "step": 14590 }, { "epoch": 5.065392549852277e-14, "grad_norm": 0.5552012920379639, "learning_rate": 0.00011697205996178172, "loss": 2.5275, "step": 14600 }, { "epoch": 5.0688619968042303e-14, "grad_norm": 0.5503589510917664, "learning_rate": 0.00011696053953222812, "loss": 2.4958, "step": 14610 }, { "epoch": 5.072331443756184e-14, "grad_norm": 0.5384141802787781, "learning_rate": 0.00011694901147859267, "loss": 2.5271, "step": 14620 }, { "epoch": 5.0758008907081376e-14, "grad_norm": 0.542513906955719, "learning_rate": 0.00011693747580269848, "loss": 2.5129, "step": 14630 }, { "epoch": 5.079270337660091e-14, "grad_norm": 0.5588251948356628, "learning_rate": 0.00011692593250636978, "loss": 2.4979, "step": 14640 }, { "epoch": 5.082739784612045e-14, "grad_norm": 0.5618288516998291, "learning_rate": 0.00011691438159143208, "loss": 2.5429, "step": 14650 }, { "epoch": 5.0862092315639984e-14, "grad_norm": 0.5308624505996704, "learning_rate": 0.00011690282305971199, "loss": 2.503, "step": 14660 }, { "epoch": 5.089678678515952e-14, "grad_norm": 0.5707845687866211, "learning_rate": 0.00011689125691303745, "loss": 2.5069, "step": 14670 }, { "epoch": 5.0931481254679056e-14, "grad_norm": 0.5209592580795288, "learning_rate": 0.00011687968315323756, "loss": 2.5696, "step": 14680 }, { "epoch": 5.096617572419859e-14, "grad_norm": 0.5431550741195679, "learning_rate": 0.00011686810178214255, "loss": 2.5192, "step": 14690 }, { "epoch": 5.100087019371813e-14, "grad_norm": 0.5760151743888855, "learning_rate": 0.00011685651280158396, "loss": 2.5269, "step": 14700 }, { "epoch": 5.1035564663237665e-14, "grad_norm": 0.5483358502388, "learning_rate": 0.00011684491621339448, "loss": 2.4956, "step": 14710 }, { "epoch": 5.10702591327572e-14, "grad_norm": 0.5828441977500916, "learning_rate": 0.00011683331201940799, "loss": 2.5099, "step": 14720 }, { "epoch": 5.110495360227674e-14, "grad_norm": 0.5561049580574036, "learning_rate": 0.00011682170022145962, "loss": 2.5841, "step": 14730 }, { "epoch": 5.113964807179627e-14, "grad_norm": 0.5596514344215393, "learning_rate": 0.00011681008082138566, "loss": 2.5343, "step": 14740 }, { "epoch": 5.117434254131581e-14, "grad_norm": 0.5458812713623047, "learning_rate": 0.00011679845382102361, "loss": 2.501, "step": 14750 }, { "epoch": 5.1209037010835345e-14, "grad_norm": 0.5616704821586609, "learning_rate": 0.0001167868192222122, "loss": 2.5287, "step": 14760 }, { "epoch": 5.124373148035488e-14, "grad_norm": 0.5318477153778076, "learning_rate": 0.00011677517702679136, "loss": 2.4885, "step": 14770 }, { "epoch": 5.127842594987442e-14, "grad_norm": 0.5794629454612732, "learning_rate": 0.00011676352723660215, "loss": 2.492, "step": 14780 }, { "epoch": 5.1313120419393954e-14, "grad_norm": 0.5603069067001343, "learning_rate": 0.00011675186985348691, "loss": 2.4931, "step": 14790 }, { "epoch": 5.134781488891349e-14, "grad_norm": 0.5422003269195557, "learning_rate": 0.00011674020487928918, "loss": 2.526, "step": 14800 }, { "epoch": 5.1382509358433026e-14, "grad_norm": 0.5387911796569824, "learning_rate": 0.00011672853231585363, "loss": 2.4961, "step": 14810 }, { "epoch": 5.141720382795256e-14, "grad_norm": 0.60600745677948, "learning_rate": 0.00011671685216502619, "loss": 2.4973, "step": 14820 }, { "epoch": 5.14518982974721e-14, "grad_norm": 0.5796007513999939, "learning_rate": 0.00011670516442865399, "loss": 2.5163, "step": 14830 }, { "epoch": 5.1486592766991635e-14, "grad_norm": 0.5268006324768066, "learning_rate": 0.00011669346910858534, "loss": 2.5351, "step": 14840 }, { "epoch": 5.152128723651117e-14, "grad_norm": 0.5556192398071289, "learning_rate": 0.00011668176620666974, "loss": 2.5241, "step": 14850 }, { "epoch": 5.155598170603071e-14, "grad_norm": 0.5406909584999084, "learning_rate": 0.00011667005572475792, "loss": 2.5152, "step": 14860 }, { "epoch": 5.159067617555024e-14, "grad_norm": 0.5537069439888, "learning_rate": 0.00011665833766470179, "loss": 2.5039, "step": 14870 }, { "epoch": 5.162537064506978e-14, "grad_norm": 0.5618975758552551, "learning_rate": 0.00011664661202835446, "loss": 2.5282, "step": 14880 }, { "epoch": 5.1660065114589315e-14, "grad_norm": 0.5810341835021973, "learning_rate": 0.00011663487881757022, "loss": 2.5161, "step": 14890 }, { "epoch": 5.169475958410885e-14, "grad_norm": 0.5427200794219971, "learning_rate": 0.0001166231380342046, "loss": 2.5231, "step": 14900 }, { "epoch": 5.172945405362839e-14, "grad_norm": 0.5494481921195984, "learning_rate": 0.00011661138968011428, "loss": 2.5221, "step": 14910 }, { "epoch": 5.1764148523147924e-14, "grad_norm": 0.5780762434005737, "learning_rate": 0.0001165996337571572, "loss": 2.4757, "step": 14920 }, { "epoch": 5.179884299266746e-14, "grad_norm": 0.5406098961830139, "learning_rate": 0.00011658787026719243, "loss": 2.5029, "step": 14930 }, { "epoch": 5.1833537462186996e-14, "grad_norm": 0.5801511406898499, "learning_rate": 0.00011657609921208025, "loss": 2.5449, "step": 14940 }, { "epoch": 5.186823193170653e-14, "grad_norm": 0.5530770421028137, "learning_rate": 0.00011656432059368218, "loss": 2.5018, "step": 14950 }, { "epoch": 5.190292640122607e-14, "grad_norm": 0.5439819693565369, "learning_rate": 0.0001165525344138609, "loss": 2.4968, "step": 14960 }, { "epoch": 5.1937620870745604e-14, "grad_norm": 0.5440488457679749, "learning_rate": 0.00011654074067448028, "loss": 2.5235, "step": 14970 }, { "epoch": 5.197231534026514e-14, "grad_norm": 0.5448414087295532, "learning_rate": 0.00011652893937740542, "loss": 2.5188, "step": 14980 }, { "epoch": 5.2007009809784677e-14, "grad_norm": 0.5244982838630676, "learning_rate": 0.00011651713052450258, "loss": 2.4978, "step": 14990 }, { "epoch": 5.204170427930421e-14, "grad_norm": 0.5322825312614441, "learning_rate": 0.00011650531411763923, "loss": 2.4898, "step": 15000 }, { "epoch": 5.207639874882375e-14, "grad_norm": 0.5383532047271729, "learning_rate": 0.00011649349015868402, "loss": 2.5188, "step": 15010 }, { "epoch": 5.2111093218343285e-14, "grad_norm": 0.5573674440383911, "learning_rate": 0.00011648165864950683, "loss": 2.5157, "step": 15020 }, { "epoch": 5.214578768786282e-14, "grad_norm": 0.5469521880149841, "learning_rate": 0.0001164698195919787, "loss": 2.4961, "step": 15030 }, { "epoch": 5.218048215738236e-14, "grad_norm": 0.5638744235038757, "learning_rate": 0.00011645797298797188, "loss": 2.5228, "step": 15040 }, { "epoch": 5.2215176626901894e-14, "grad_norm": 0.5407611131668091, "learning_rate": 0.0001164461188393598, "loss": 2.5288, "step": 15050 }, { "epoch": 5.224987109642143e-14, "grad_norm": 0.5250468254089355, "learning_rate": 0.0001164342571480171, "loss": 2.5174, "step": 15060 }, { "epoch": 5.2284565565940966e-14, "grad_norm": 0.5413164496421814, "learning_rate": 0.0001164223879158196, "loss": 2.4643, "step": 15070 }, { "epoch": 5.23192600354605e-14, "grad_norm": 0.5435682535171509, "learning_rate": 0.00011641051114464432, "loss": 2.4947, "step": 15080 }, { "epoch": 5.235395450498004e-14, "grad_norm": 0.5611423850059509, "learning_rate": 0.00011639862683636946, "loss": 2.5438, "step": 15090 }, { "epoch": 5.2388648974499574e-14, "grad_norm": 2.601862668991089, "learning_rate": 0.00011638673499287444, "loss": 2.5167, "step": 15100 }, { "epoch": 5.242334344401911e-14, "grad_norm": 0.5561999678611755, "learning_rate": 0.00011637483561603982, "loss": 2.5221, "step": 15110 }, { "epoch": 5.2458037913538647e-14, "grad_norm": 0.5256353616714478, "learning_rate": 0.00011636292870774742, "loss": 2.4869, "step": 15120 }, { "epoch": 5.249273238305818e-14, "grad_norm": 0.5370330214500427, "learning_rate": 0.0001163510142698802, "loss": 2.5327, "step": 15130 }, { "epoch": 5.252742685257772e-14, "grad_norm": 0.5968655347824097, "learning_rate": 0.00011633909230432232, "loss": 2.5305, "step": 15140 }, { "epoch": 5.2562121322097255e-14, "grad_norm": 0.5393545031547546, "learning_rate": 0.00011632716281295916, "loss": 2.527, "step": 15150 }, { "epoch": 5.259681579161679e-14, "grad_norm": 0.5704300403594971, "learning_rate": 0.00011631522579767721, "loss": 2.5393, "step": 15160 }, { "epoch": 5.263151026113633e-14, "grad_norm": 0.5609644055366516, "learning_rate": 0.00011630328126036428, "loss": 2.5321, "step": 15170 }, { "epoch": 5.2666204730655863e-14, "grad_norm": 0.5650519132614136, "learning_rate": 0.00011629132920290925, "loss": 2.524, "step": 15180 }, { "epoch": 5.27008992001754e-14, "grad_norm": 0.5300900340080261, "learning_rate": 0.00011627936962720225, "loss": 2.5298, "step": 15190 }, { "epoch": 5.2735593669694936e-14, "grad_norm": 0.5552693009376526, "learning_rate": 0.00011626740253513457, "loss": 2.5401, "step": 15200 }, { "epoch": 5.277028813921447e-14, "grad_norm": 0.548653781414032, "learning_rate": 0.00011625542792859869, "loss": 2.5117, "step": 15210 }, { "epoch": 5.280498260873401e-14, "grad_norm": 0.542378306388855, "learning_rate": 0.00011624344580948836, "loss": 2.4627, "step": 15220 }, { "epoch": 5.2839677078253544e-14, "grad_norm": 0.5396348834037781, "learning_rate": 0.00011623145617969838, "loss": 2.5065, "step": 15230 }, { "epoch": 5.287437154777308e-14, "grad_norm": 0.5589211583137512, "learning_rate": 0.00011621945904112481, "loss": 2.4866, "step": 15240 }, { "epoch": 5.2909066017292616e-14, "grad_norm": 0.5727335214614868, "learning_rate": 0.00011620745439566494, "loss": 2.5516, "step": 15250 }, { "epoch": 5.294376048681215e-14, "grad_norm": 0.5933425426483154, "learning_rate": 0.00011619544224521715, "loss": 2.4976, "step": 15260 }, { "epoch": 5.297845495633169e-14, "grad_norm": 0.5591787695884705, "learning_rate": 0.00011618342259168109, "loss": 2.4819, "step": 15270 }, { "epoch": 5.3013149425851225e-14, "grad_norm": 0.5721724033355713, "learning_rate": 0.00011617139543695756, "loss": 2.5467, "step": 15280 }, { "epoch": 5.304784389537076e-14, "grad_norm": 0.728571891784668, "learning_rate": 0.00011615936078294856, "loss": 2.5112, "step": 15290 }, { "epoch": 5.30825383648903e-14, "grad_norm": 0.5556759238243103, "learning_rate": 0.00011614731863155725, "loss": 2.4805, "step": 15300 }, { "epoch": 5.3117232834409833e-14, "grad_norm": 0.5708962678909302, "learning_rate": 0.00011613526898468797, "loss": 2.4657, "step": 15310 }, { "epoch": 5.315192730392937e-14, "grad_norm": 0.5371319055557251, "learning_rate": 0.00011612321184424633, "loss": 2.495, "step": 15320 }, { "epoch": 5.3186621773448906e-14, "grad_norm": 0.553253710269928, "learning_rate": 0.00011611114721213903, "loss": 2.5051, "step": 15330 }, { "epoch": 5.322131624296844e-14, "grad_norm": 0.5782899260520935, "learning_rate": 0.00011609907509027397, "loss": 2.4745, "step": 15340 }, { "epoch": 5.325601071248798e-14, "grad_norm": 0.5512089729309082, "learning_rate": 0.00011608699548056027, "loss": 2.5184, "step": 15350 }, { "epoch": 5.3290705182007514e-14, "grad_norm": 0.526251494884491, "learning_rate": 0.0001160749083849082, "loss": 2.4858, "step": 15360 }, { "epoch": 5.332539965152705e-14, "grad_norm": 0.5322789549827576, "learning_rate": 0.00011606281380522926, "loss": 2.4609, "step": 15370 }, { "epoch": 5.3360094121046586e-14, "grad_norm": 0.562379002571106, "learning_rate": 0.00011605071174343608, "loss": 2.5023, "step": 15380 }, { "epoch": 5.339478859056612e-14, "grad_norm": 0.5400394797325134, "learning_rate": 0.00011603860220144252, "loss": 2.5239, "step": 15390 }, { "epoch": 5.342948306008566e-14, "grad_norm": 0.5241631269454956, "learning_rate": 0.00011602648518116357, "loss": 2.4845, "step": 15400 }, { "epoch": 5.3464177529605195e-14, "grad_norm": 0.5675278902053833, "learning_rate": 0.00011601436068451548, "loss": 2.5258, "step": 15410 }, { "epoch": 5.349887199912473e-14, "grad_norm": 0.5353706479072571, "learning_rate": 0.00011600222871341555, "loss": 2.5437, "step": 15420 }, { "epoch": 5.353356646864427e-14, "grad_norm": 0.5702682733535767, "learning_rate": 0.00011599008926978242, "loss": 2.5165, "step": 15430 }, { "epoch": 5.35682609381638e-14, "grad_norm": 0.5188799500465393, "learning_rate": 0.00011597794235553583, "loss": 2.4939, "step": 15440 }, { "epoch": 5.360295540768334e-14, "grad_norm": 0.538637638092041, "learning_rate": 0.00011596578797259668, "loss": 2.4927, "step": 15450 }, { "epoch": 5.3637649877202875e-14, "grad_norm": 0.5373274087905884, "learning_rate": 0.00011595362612288712, "loss": 2.4805, "step": 15460 }, { "epoch": 5.367234434672241e-14, "grad_norm": 0.5507751107215881, "learning_rate": 0.0001159414568083304, "loss": 2.4996, "step": 15470 }, { "epoch": 5.370703881624195e-14, "grad_norm": 0.5184004902839661, "learning_rate": 0.00011592928003085104, "loss": 2.515, "step": 15480 }, { "epoch": 5.3741733285761484e-14, "grad_norm": 0.5465449690818787, "learning_rate": 0.00011591709579237464, "loss": 2.4802, "step": 15490 }, { "epoch": 5.377642775528102e-14, "grad_norm": 0.5581454634666443, "learning_rate": 0.00011590490409482805, "loss": 2.5497, "step": 15500 }, { "epoch": 5.3811122224800556e-14, "grad_norm": 0.5553013682365417, "learning_rate": 0.00011589270494013931, "loss": 2.473, "step": 15510 }, { "epoch": 5.384581669432009e-14, "grad_norm": 0.5349869728088379, "learning_rate": 0.00011588049833023758, "loss": 2.497, "step": 15520 }, { "epoch": 5.388051116383963e-14, "grad_norm": 0.547757625579834, "learning_rate": 0.00011586828426705324, "loss": 2.5063, "step": 15530 }, { "epoch": 5.3915205633359164e-14, "grad_norm": 0.5429736375808716, "learning_rate": 0.00011585606275251785, "loss": 2.503, "step": 15540 }, { "epoch": 5.39499001028787e-14, "grad_norm": 0.5406195521354675, "learning_rate": 0.00011584383378856412, "loss": 2.489, "step": 15550 }, { "epoch": 5.3984594572398237e-14, "grad_norm": 0.5373625159263611, "learning_rate": 0.00011583159737712598, "loss": 2.4817, "step": 15560 }, { "epoch": 5.401928904191777e-14, "grad_norm": 0.5241339206695557, "learning_rate": 0.00011581935352013848, "loss": 2.4961, "step": 15570 }, { "epoch": 5.405398351143731e-14, "grad_norm": 0.5474107265472412, "learning_rate": 0.00011580710221953789, "loss": 2.4409, "step": 15580 }, { "epoch": 5.4088677980956845e-14, "grad_norm": 0.5671996474266052, "learning_rate": 0.00011579484347726167, "loss": 2.4788, "step": 15590 }, { "epoch": 5.412337245047638e-14, "grad_norm": 0.5471214056015015, "learning_rate": 0.00011578257729524841, "loss": 2.5286, "step": 15600 }, { "epoch": 5.415806691999592e-14, "grad_norm": 0.5571038722991943, "learning_rate": 0.0001157703036754379, "loss": 2.4943, "step": 15610 }, { "epoch": 5.4192761389515454e-14, "grad_norm": 0.5351354479789734, "learning_rate": 0.00011575802261977114, "loss": 2.5135, "step": 15620 }, { "epoch": 5.422745585903499e-14, "grad_norm": 0.5626742839813232, "learning_rate": 0.00011574573413019024, "loss": 2.5136, "step": 15630 }, { "epoch": 5.4262150328554526e-14, "grad_norm": 0.5506417751312256, "learning_rate": 0.00011573343820863854, "loss": 2.4725, "step": 15640 }, { "epoch": 5.429684479807406e-14, "grad_norm": 0.5576821565628052, "learning_rate": 0.00011572113485706051, "loss": 2.5231, "step": 15650 }, { "epoch": 5.43315392675936e-14, "grad_norm": 0.584098219871521, "learning_rate": 0.00011570882407740183, "loss": 2.5042, "step": 15660 }, { "epoch": 5.4366233737113134e-14, "grad_norm": 0.5544558763504028, "learning_rate": 0.00011569650587160934, "loss": 2.5033, "step": 15670 }, { "epoch": 5.440092820663267e-14, "grad_norm": 0.5675811171531677, "learning_rate": 0.0001156841802416311, "loss": 2.5085, "step": 15680 }, { "epoch": 5.4435622676152207e-14, "grad_norm": 0.5911427140235901, "learning_rate": 0.00011567184718941623, "loss": 2.4889, "step": 15690 }, { "epoch": 5.447031714567174e-14, "grad_norm": 0.5484492778778076, "learning_rate": 0.00011565950671691513, "loss": 2.4931, "step": 15700 }, { "epoch": 5.450501161519128e-14, "grad_norm": 0.5447160005569458, "learning_rate": 0.00011564715882607938, "loss": 2.5137, "step": 15710 }, { "epoch": 5.4539706084710815e-14, "grad_norm": 0.544312596321106, "learning_rate": 0.00011563480351886163, "loss": 2.4838, "step": 15720 }, { "epoch": 5.457440055423035e-14, "grad_norm": 0.5380884408950806, "learning_rate": 0.00011562244079721578, "loss": 2.5112, "step": 15730 }, { "epoch": 5.460909502374989e-14, "grad_norm": 0.5663769841194153, "learning_rate": 0.00011561007066309693, "loss": 2.5242, "step": 15740 }, { "epoch": 5.4643789493269423e-14, "grad_norm": 0.5510218739509583, "learning_rate": 0.00011559769311846126, "loss": 2.496, "step": 15750 }, { "epoch": 5.467848396278896e-14, "grad_norm": 0.5449652671813965, "learning_rate": 0.00011558530816526619, "loss": 2.4735, "step": 15760 }, { "epoch": 5.4713178432308496e-14, "grad_norm": 0.5448790192604065, "learning_rate": 0.0001155729158054703, "loss": 2.4734, "step": 15770 }, { "epoch": 5.474787290182803e-14, "grad_norm": 0.5590184926986694, "learning_rate": 0.00011556051604103332, "loss": 2.4914, "step": 15780 }, { "epoch": 5.478256737134757e-14, "grad_norm": 0.5799431800842285, "learning_rate": 0.00011554810887391619, "loss": 2.5285, "step": 15790 }, { "epoch": 5.4817261840867104e-14, "grad_norm": 0.5402566194534302, "learning_rate": 0.00011553569430608098, "loss": 2.4853, "step": 15800 }, { "epoch": 5.485195631038664e-14, "grad_norm": 0.5569892525672913, "learning_rate": 0.00011552327233949094, "loss": 2.4882, "step": 15810 }, { "epoch": 5.4886650779906176e-14, "grad_norm": 0.535874605178833, "learning_rate": 0.0001155108429761105, "loss": 2.5163, "step": 15820 }, { "epoch": 5.492134524942571e-14, "grad_norm": 0.5610423684120178, "learning_rate": 0.00011549840621790529, "loss": 2.5428, "step": 15830 }, { "epoch": 5.495603971894525e-14, "grad_norm": 0.551434338092804, "learning_rate": 0.00011548596206684203, "loss": 2.5159, "step": 15840 }, { "epoch": 5.4990734188464785e-14, "grad_norm": 0.558402419090271, "learning_rate": 0.00011547351052488868, "loss": 2.4809, "step": 15850 }, { "epoch": 5.502542865798432e-14, "grad_norm": 0.5814874768257141, "learning_rate": 0.00011546105159401435, "loss": 2.5136, "step": 15860 }, { "epoch": 5.506012312750386e-14, "grad_norm": 0.5655152201652527, "learning_rate": 0.00011544858527618929, "loss": 2.5351, "step": 15870 }, { "epoch": 5.5094817597023393e-14, "grad_norm": 0.5704218149185181, "learning_rate": 0.00011543611157338494, "loss": 2.5045, "step": 15880 }, { "epoch": 5.512951206654293e-14, "grad_norm": 0.5586896538734436, "learning_rate": 0.00011542363048757394, "loss": 2.5074, "step": 15890 }, { "epoch": 5.5164206536062466e-14, "grad_norm": 0.5544239282608032, "learning_rate": 0.00011541114202073004, "loss": 2.5103, "step": 15900 }, { "epoch": 5.5198901005582e-14, "grad_norm": 0.5495901703834534, "learning_rate": 0.0001153986461748282, "loss": 2.4933, "step": 15910 }, { "epoch": 5.523359547510154e-14, "grad_norm": 0.5628910660743713, "learning_rate": 0.00011538614295184452, "loss": 2.5053, "step": 15920 }, { "epoch": 5.5268289944621074e-14, "grad_norm": 0.5430458784103394, "learning_rate": 0.00011537363235375628, "loss": 2.5256, "step": 15930 }, { "epoch": 5.530298441414061e-14, "grad_norm": 0.5835579633712769, "learning_rate": 0.00011536111438254191, "loss": 2.4856, "step": 15940 }, { "epoch": 5.5337678883660146e-14, "grad_norm": 0.5471209287643433, "learning_rate": 0.00011534858904018106, "loss": 2.476, "step": 15950 }, { "epoch": 5.537237335317968e-14, "grad_norm": 0.5979980826377869, "learning_rate": 0.00011533605632865444, "loss": 2.5243, "step": 15960 }, { "epoch": 5.540706782269922e-14, "grad_norm": 0.5353455543518066, "learning_rate": 0.00011532351624994406, "loss": 2.4942, "step": 15970 }, { "epoch": 5.5441762292218755e-14, "grad_norm": 0.5332491993904114, "learning_rate": 0.00011531096880603295, "loss": 2.5217, "step": 15980 }, { "epoch": 5.547645676173829e-14, "grad_norm": 0.5608599185943604, "learning_rate": 0.00011529841399890545, "loss": 2.4913, "step": 15990 }, { "epoch": 5.551115123125783e-14, "grad_norm": 0.54685378074646, "learning_rate": 0.00011528585183054698, "loss": 2.4796, "step": 16000 }, { "epoch": 5.5545845700777363e-14, "grad_norm": 0.5187146067619324, "learning_rate": 0.00011527328230294409, "loss": 2.4957, "step": 16010 }, { "epoch": 5.55805401702969e-14, "grad_norm": 0.5689557790756226, "learning_rate": 0.00011526070541808461, "loss": 2.4564, "step": 16020 }, { "epoch": 5.5615234639816435e-14, "grad_norm": 0.5214100480079651, "learning_rate": 0.00011524812117795744, "loss": 2.4781, "step": 16030 }, { "epoch": 5.564992910933597e-14, "grad_norm": 0.5324169397354126, "learning_rate": 0.00011523552958455263, "loss": 2.4763, "step": 16040 }, { "epoch": 5.568462357885551e-14, "grad_norm": 0.5649765133857727, "learning_rate": 0.0001152229306398615, "loss": 2.4977, "step": 16050 }, { "epoch": 5.5719318048375044e-14, "grad_norm": 0.5398749113082886, "learning_rate": 0.00011521032434587641, "loss": 2.5228, "step": 16060 }, { "epoch": 5.575401251789458e-14, "grad_norm": 0.5495522618293762, "learning_rate": 0.00011519771070459098, "loss": 2.511, "step": 16070 }, { "epoch": 5.5788706987414116e-14, "grad_norm": 0.5528058409690857, "learning_rate": 0.0001151850897179999, "loss": 2.5048, "step": 16080 }, { "epoch": 5.582340145693365e-14, "grad_norm": 0.5318600535392761, "learning_rate": 0.00011517246138809911, "loss": 2.4832, "step": 16090 }, { "epoch": 5.585809592645319e-14, "grad_norm": 0.5462633967399597, "learning_rate": 0.00011515982571688567, "loss": 2.478, "step": 16100 }, { "epoch": 5.5892790395972725e-14, "grad_norm": 0.515731930732727, "learning_rate": 0.00011514718270635776, "loss": 2.4351, "step": 16110 }, { "epoch": 5.592748486549226e-14, "grad_norm": 0.5500914454460144, "learning_rate": 0.00011513453235851484, "loss": 2.5008, "step": 16120 }, { "epoch": 5.59621793350118e-14, "grad_norm": 0.5694894790649414, "learning_rate": 0.00011512187467535737, "loss": 2.5273, "step": 16130 }, { "epoch": 5.599687380453133e-14, "grad_norm": 0.5228210687637329, "learning_rate": 0.0001151092096588871, "loss": 2.4961, "step": 16140 }, { "epoch": 5.603156827405087e-14, "grad_norm": 0.5344275832176208, "learning_rate": 0.0001150965373111069, "loss": 2.5009, "step": 16150 }, { "epoch": 5.6066262743570405e-14, "grad_norm": 0.5413697957992554, "learning_rate": 0.00011508385763402076, "loss": 2.5125, "step": 16160 }, { "epoch": 5.610095721308994e-14, "grad_norm": 0.5384553670883179, "learning_rate": 0.0001150711706296339, "loss": 2.5283, "step": 16170 }, { "epoch": 5.613565168260948e-14, "grad_norm": 0.5648044347763062, "learning_rate": 0.00011505847629995263, "loss": 2.4842, "step": 16180 }, { "epoch": 5.6170346152129014e-14, "grad_norm": 0.5511257648468018, "learning_rate": 0.00011504577464698446, "loss": 2.4769, "step": 16190 }, { "epoch": 5.620504062164855e-14, "grad_norm": 0.5068755149841309, "learning_rate": 0.00011503306567273803, "loss": 2.4661, "step": 16200 }, { "epoch": 5.6239735091168086e-14, "grad_norm": 0.54970383644104, "learning_rate": 0.00011502034937922319, "loss": 2.5129, "step": 16210 }, { "epoch": 5.627442956068762e-14, "grad_norm": 0.5751581788063049, "learning_rate": 0.0001150076257684509, "loss": 2.4791, "step": 16220 }, { "epoch": 5.630912403020716e-14, "grad_norm": 0.5881938934326172, "learning_rate": 0.00011499489484243328, "loss": 2.5279, "step": 16230 }, { "epoch": 5.6343818499726694e-14, "grad_norm": 0.5610857009887695, "learning_rate": 0.00011498215660318359, "loss": 2.5079, "step": 16240 }, { "epoch": 5.637851296924623e-14, "grad_norm": 0.5309084057807922, "learning_rate": 0.00011496941105271632, "loss": 2.5293, "step": 16250 }, { "epoch": 5.6413207438765767e-14, "grad_norm": 0.5334853529930115, "learning_rate": 0.00011495665819304706, "loss": 2.4742, "step": 16260 }, { "epoch": 5.64479019082853e-14, "grad_norm": 0.5598465800285339, "learning_rate": 0.00011494389802619255, "loss": 2.4868, "step": 16270 }, { "epoch": 5.648259637780484e-14, "grad_norm": 0.5407698154449463, "learning_rate": 0.00011493113055417072, "loss": 2.4896, "step": 16280 }, { "epoch": 5.6517290847324375e-14, "grad_norm": 0.5182095170021057, "learning_rate": 0.00011491835577900062, "loss": 2.5157, "step": 16290 }, { "epoch": 5.655198531684391e-14, "grad_norm": 0.5508682727813721, "learning_rate": 0.00011490557370270246, "loss": 2.49, "step": 16300 }, { "epoch": 5.658667978636345e-14, "grad_norm": 0.5251641869544983, "learning_rate": 0.00011489278432729762, "loss": 2.4896, "step": 16310 }, { "epoch": 5.6621374255882984e-14, "grad_norm": 0.5304814577102661, "learning_rate": 0.00011487998765480865, "loss": 2.4587, "step": 16320 }, { "epoch": 5.665606872540252e-14, "grad_norm": 0.5284743309020996, "learning_rate": 0.00011486718368725923, "loss": 2.5149, "step": 16330 }, { "epoch": 5.6690763194922056e-14, "grad_norm": 0.5339013338088989, "learning_rate": 0.00011485437242667418, "loss": 2.5335, "step": 16340 }, { "epoch": 5.672545766444159e-14, "grad_norm": 0.5618588924407959, "learning_rate": 0.00011484155387507952, "loss": 2.5283, "step": 16350 }, { "epoch": 5.676015213396113e-14, "grad_norm": 0.5562686324119568, "learning_rate": 0.00011482872803450233, "loss": 2.5035, "step": 16360 }, { "epoch": 5.6794846603480664e-14, "grad_norm": 0.5329679846763611, "learning_rate": 0.000114815894906971, "loss": 2.5299, "step": 16370 }, { "epoch": 5.68295410730002e-14, "grad_norm": 0.543014407157898, "learning_rate": 0.00011480305449451489, "loss": 2.5135, "step": 16380 }, { "epoch": 5.686423554251974e-14, "grad_norm": 0.6081127524375916, "learning_rate": 0.00011479020679916466, "loss": 2.5165, "step": 16390 }, { "epoch": 5.689893001203927e-14, "grad_norm": 0.6089504957199097, "learning_rate": 0.00011477735182295204, "loss": 2.5206, "step": 16400 }, { "epoch": 5.693362448155881e-14, "grad_norm": 0.5742416381835938, "learning_rate": 0.00011476448956790992, "loss": 2.4753, "step": 16410 }, { "epoch": 5.696831895107834e-14, "grad_norm": 0.5457869172096252, "learning_rate": 0.0001147516200360724, "loss": 2.4619, "step": 16420 }, { "epoch": 5.700301342059788e-14, "grad_norm": 0.5561283826828003, "learning_rate": 0.00011473874322947464, "loss": 2.4897, "step": 16430 }, { "epoch": 5.703770789011742e-14, "grad_norm": 0.565039336681366, "learning_rate": 0.00011472585915015301, "loss": 2.4965, "step": 16440 }, { "epoch": 5.707240235963695e-14, "grad_norm": 0.5497084259986877, "learning_rate": 0.000114712967800145, "loss": 2.5074, "step": 16450 }, { "epoch": 5.710709682915649e-14, "grad_norm": 0.5580193400382996, "learning_rate": 0.00011470006918148931, "loss": 2.505, "step": 16460 }, { "epoch": 5.714179129867603e-14, "grad_norm": 0.5521441102027893, "learning_rate": 0.00011468716329622571, "loss": 2.5007, "step": 16470 }, { "epoch": 5.717648576819556e-14, "grad_norm": 0.5553426742553711, "learning_rate": 0.00011467425014639517, "loss": 2.5102, "step": 16480 }, { "epoch": 5.72111802377151e-14, "grad_norm": 0.5171317458152771, "learning_rate": 0.00011466132973403979, "loss": 2.4891, "step": 16490 }, { "epoch": 5.724587470723463e-14, "grad_norm": 0.5315152406692505, "learning_rate": 0.0001146484020612028, "loss": 2.5327, "step": 16500 }, { "epoch": 5.728056917675417e-14, "grad_norm": 0.5565922856330872, "learning_rate": 0.00011463546712992864, "loss": 2.4808, "step": 16510 }, { "epoch": 5.731526364627371e-14, "grad_norm": 0.5500588417053223, "learning_rate": 0.00011462252494226284, "loss": 2.4693, "step": 16520 }, { "epoch": 5.734995811579324e-14, "grad_norm": 0.5519455671310425, "learning_rate": 0.00011460957550025208, "loss": 2.4883, "step": 16530 }, { "epoch": 5.738465258531278e-14, "grad_norm": 0.6704633831977844, "learning_rate": 0.00011459661880594423, "loss": 2.4782, "step": 16540 }, { "epoch": 5.741934705483231e-14, "grad_norm": 0.556472897529602, "learning_rate": 0.00011458365486138826, "loss": 2.501, "step": 16550 }, { "epoch": 5.745404152435185e-14, "grad_norm": 0.5518902540206909, "learning_rate": 0.00011457068366863431, "loss": 2.5226, "step": 16560 }, { "epoch": 5.748873599387139e-14, "grad_norm": 0.5732284188270569, "learning_rate": 0.00011455770522973367, "loss": 2.4854, "step": 16570 }, { "epoch": 5.752343046339092e-14, "grad_norm": 0.5415456295013428, "learning_rate": 0.00011454471954673877, "loss": 2.4813, "step": 16580 }, { "epoch": 5.755812493291046e-14, "grad_norm": 0.5698407888412476, "learning_rate": 0.00011453172662170315, "loss": 2.4694, "step": 16590 }, { "epoch": 5.759281940243e-14, "grad_norm": 0.5658152103424072, "learning_rate": 0.00011451872645668158, "loss": 2.5067, "step": 16600 }, { "epoch": 5.762751387194953e-14, "grad_norm": 0.5285978317260742, "learning_rate": 0.0001145057190537299, "loss": 2.4579, "step": 16610 }, { "epoch": 5.766220834146907e-14, "grad_norm": 0.5427430272102356, "learning_rate": 0.00011449270441490511, "loss": 2.475, "step": 16620 }, { "epoch": 5.76969028109886e-14, "grad_norm": 0.5593248605728149, "learning_rate": 0.00011447968254226537, "loss": 2.4964, "step": 16630 }, { "epoch": 5.773159728050814e-14, "grad_norm": 0.5866517424583435, "learning_rate": 0.00011446665343787, "loss": 2.501, "step": 16640 }, { "epoch": 5.776629175002768e-14, "grad_norm": 0.5646384954452515, "learning_rate": 0.0001144536171037794, "loss": 2.4813, "step": 16650 }, { "epoch": 5.780098621954721e-14, "grad_norm": 0.5438405275344849, "learning_rate": 0.00011444057354205518, "loss": 2.5147, "step": 16660 }, { "epoch": 5.783568068906675e-14, "grad_norm": 0.5522012710571289, "learning_rate": 0.00011442752275476006, "loss": 2.4762, "step": 16670 }, { "epoch": 5.787037515858628e-14, "grad_norm": 0.5386776924133301, "learning_rate": 0.00011441446474395792, "loss": 2.4804, "step": 16680 }, { "epoch": 5.790506962810582e-14, "grad_norm": 0.5814762711524963, "learning_rate": 0.00011440139951171375, "loss": 2.4538, "step": 16690 }, { "epoch": 5.793976409762536e-14, "grad_norm": 0.5474389791488647, "learning_rate": 0.00011438832706009373, "loss": 2.5006, "step": 16700 }, { "epoch": 5.797445856714489e-14, "grad_norm": 0.5166049003601074, "learning_rate": 0.00011437524739116515, "loss": 2.4924, "step": 16710 }, { "epoch": 5.800915303666443e-14, "grad_norm": 0.5811395049095154, "learning_rate": 0.00011436216050699642, "loss": 2.522, "step": 16720 }, { "epoch": 5.804384750618397e-14, "grad_norm": 0.521093487739563, "learning_rate": 0.00011434906640965714, "loss": 2.4596, "step": 16730 }, { "epoch": 5.80785419757035e-14, "grad_norm": 0.5518320798873901, "learning_rate": 0.00011433596510121805, "loss": 2.4669, "step": 16740 }, { "epoch": 5.811323644522304e-14, "grad_norm": 0.5457063913345337, "learning_rate": 0.00011432285658375096, "loss": 2.4951, "step": 16750 }, { "epoch": 5.814793091474257e-14, "grad_norm": 0.5701975226402283, "learning_rate": 0.00011430974085932891, "loss": 2.4609, "step": 16760 }, { "epoch": 5.818262538426211e-14, "grad_norm": 0.5812790989875793, "learning_rate": 0.000114296617930026, "loss": 2.4859, "step": 16770 }, { "epoch": 5.821731985378165e-14, "grad_norm": 0.5384799838066101, "learning_rate": 0.00011428348779791755, "loss": 2.507, "step": 16780 }, { "epoch": 5.825201432330118e-14, "grad_norm": 0.5531455874443054, "learning_rate": 0.00011427035046507997, "loss": 2.5005, "step": 16790 }, { "epoch": 5.828670879282072e-14, "grad_norm": 0.5611482858657837, "learning_rate": 0.00011425720593359079, "loss": 2.5337, "step": 16800 }, { "epoch": 5.832140326234025e-14, "grad_norm": 0.5793189406394958, "learning_rate": 0.00011424405420552872, "loss": 2.4405, "step": 16810 }, { "epoch": 5.835609773185979e-14, "grad_norm": 0.5642656087875366, "learning_rate": 0.00011423089528297362, "loss": 2.4737, "step": 16820 }, { "epoch": 5.839079220137933e-14, "grad_norm": 0.5434162020683289, "learning_rate": 0.0001142177291680064, "loss": 2.4704, "step": 16830 }, { "epoch": 5.842548667089886e-14, "grad_norm": 0.5394679307937622, "learning_rate": 0.00011420455586270921, "loss": 2.4663, "step": 16840 }, { "epoch": 5.84601811404184e-14, "grad_norm": 0.5374225974082947, "learning_rate": 0.0001141913753691653, "loss": 2.4856, "step": 16850 }, { "epoch": 5.849487560993794e-14, "grad_norm": 3.4168365001678467, "learning_rate": 0.00011417818768945902, "loss": 2.4995, "step": 16860 }, { "epoch": 5.852957007945747e-14, "grad_norm": 0.5839749574661255, "learning_rate": 0.0001141649928256759, "loss": 2.4797, "step": 16870 }, { "epoch": 5.856426454897701e-14, "grad_norm": 0.5590096712112427, "learning_rate": 0.00011415179077990263, "loss": 2.4878, "step": 16880 }, { "epoch": 5.859895901849654e-14, "grad_norm": 0.5381380915641785, "learning_rate": 0.00011413858155422693, "loss": 2.4904, "step": 16890 }, { "epoch": 5.863365348801608e-14, "grad_norm": 0.5705340504646301, "learning_rate": 0.0001141253651507378, "loss": 2.4891, "step": 16900 }, { "epoch": 5.866834795753562e-14, "grad_norm": 0.525648832321167, "learning_rate": 0.00011411214157152525, "loss": 2.465, "step": 16910 }, { "epoch": 5.870304242705515e-14, "grad_norm": 0.6125407218933105, "learning_rate": 0.0001140989108186805, "loss": 2.4577, "step": 16920 }, { "epoch": 5.873773689657469e-14, "grad_norm": 0.5468223094940186, "learning_rate": 0.00011408567289429584, "loss": 2.4511, "step": 16930 }, { "epoch": 5.877243136609422e-14, "grad_norm": 0.5530538558959961, "learning_rate": 0.00011407242780046481, "loss": 2.4894, "step": 16940 }, { "epoch": 5.880712583561376e-14, "grad_norm": 0.5418038368225098, "learning_rate": 0.00011405917553928192, "loss": 2.5023, "step": 16950 }, { "epoch": 5.88418203051333e-14, "grad_norm": 0.5497557520866394, "learning_rate": 0.00011404591611284296, "loss": 2.5287, "step": 16960 }, { "epoch": 5.887651477465283e-14, "grad_norm": 0.5541995167732239, "learning_rate": 0.00011403264952324478, "loss": 2.4687, "step": 16970 }, { "epoch": 5.891120924417237e-14, "grad_norm": 0.5397704243659973, "learning_rate": 0.00011401937577258538, "loss": 2.4639, "step": 16980 }, { "epoch": 5.89459037136919e-14, "grad_norm": 0.5901837944984436, "learning_rate": 0.00011400609486296387, "loss": 2.4953, "step": 16990 }, { "epoch": 5.898059818321144e-14, "grad_norm": 0.5335520505905151, "learning_rate": 0.00011399280679648052, "loss": 2.4294, "step": 17000 }, { "epoch": 5.901529265273098e-14, "grad_norm": 0.6006693243980408, "learning_rate": 0.00011397951157523674, "loss": 2.5049, "step": 17010 }, { "epoch": 5.904998712225051e-14, "grad_norm": 0.5375439524650574, "learning_rate": 0.00011396620920133503, "loss": 2.4723, "step": 17020 }, { "epoch": 5.908468159177005e-14, "grad_norm": 0.5482585430145264, "learning_rate": 0.00011395289967687908, "loss": 2.468, "step": 17030 }, { "epoch": 5.911937606128959e-14, "grad_norm": 0.5648439526557922, "learning_rate": 0.00011393958300397362, "loss": 2.4675, "step": 17040 }, { "epoch": 5.915407053080912e-14, "grad_norm": 0.5421444177627563, "learning_rate": 0.00011392625918472465, "loss": 2.4532, "step": 17050 }, { "epoch": 5.918876500032866e-14, "grad_norm": 0.5478857755661011, "learning_rate": 0.00011391292822123912, "loss": 2.4781, "step": 17060 }, { "epoch": 5.92234594698482e-14, "grad_norm": 0.5480595827102661, "learning_rate": 0.0001138995901156253, "loss": 2.464, "step": 17070 }, { "epoch": 5.925815393936773e-14, "grad_norm": 0.5684863328933716, "learning_rate": 0.00011388624486999242, "loss": 2.4622, "step": 17080 }, { "epoch": 5.929284840888727e-14, "grad_norm": 0.5539911389350891, "learning_rate": 0.00011387289248645096, "loss": 2.4791, "step": 17090 }, { "epoch": 5.93275428784068e-14, "grad_norm": 0.5682486295700073, "learning_rate": 0.00011385953296711251, "loss": 2.4952, "step": 17100 }, { "epoch": 5.936223734792634e-14, "grad_norm": 0.5798810720443726, "learning_rate": 0.00011384616631408968, "loss": 2.4957, "step": 17110 }, { "epoch": 5.939693181744587e-14, "grad_norm": 0.5579007267951965, "learning_rate": 0.00011383279252949634, "loss": 2.4589, "step": 17120 }, { "epoch": 5.943162628696541e-14, "grad_norm": 0.5450924634933472, "learning_rate": 0.00011381941161544746, "loss": 2.4434, "step": 17130 }, { "epoch": 5.946632075648495e-14, "grad_norm": 0.5505945682525635, "learning_rate": 0.00011380602357405909, "loss": 2.4574, "step": 17140 }, { "epoch": 5.950101522600448e-14, "grad_norm": 0.5366917252540588, "learning_rate": 0.00011379262840744842, "loss": 2.4381, "step": 17150 }, { "epoch": 5.953570969552402e-14, "grad_norm": 0.5892321467399597, "learning_rate": 0.00011377922611773381, "loss": 2.4669, "step": 17160 }, { "epoch": 5.957040416504356e-14, "grad_norm": 0.5248239636421204, "learning_rate": 0.00011376581670703469, "loss": 2.4834, "step": 17170 }, { "epoch": 5.960509863456309e-14, "grad_norm": 0.553449273109436, "learning_rate": 0.00011375240017747166, "loss": 2.4664, "step": 17180 }, { "epoch": 5.963979310408263e-14, "grad_norm": 0.5663995146751404, "learning_rate": 0.00011373897653116644, "loss": 2.4682, "step": 17190 }, { "epoch": 5.967448757360216e-14, "grad_norm": 0.5519457459449768, "learning_rate": 0.00011372554577024185, "loss": 2.4647, "step": 17200 }, { "epoch": 5.97091820431217e-14, "grad_norm": 0.5353186130523682, "learning_rate": 0.00011371210789682183, "loss": 2.4405, "step": 17210 }, { "epoch": 5.974387651264124e-14, "grad_norm": 0.564741849899292, "learning_rate": 0.0001136986629130315, "loss": 2.4818, "step": 17220 }, { "epoch": 5.977857098216077e-14, "grad_norm": 0.5345650911331177, "learning_rate": 0.00011368521082099705, "loss": 2.4677, "step": 17230 }, { "epoch": 5.981326545168031e-14, "grad_norm": 0.5836304426193237, "learning_rate": 0.0001136717516228458, "loss": 2.4565, "step": 17240 }, { "epoch": 5.984795992119984e-14, "grad_norm": 0.577659547328949, "learning_rate": 0.00011365828532070623, "loss": 2.4807, "step": 17250 }, { "epoch": 5.988265439071938e-14, "grad_norm": 0.5304780006408691, "learning_rate": 0.00011364481191670791, "loss": 2.4729, "step": 17260 }, { "epoch": 5.991734886023892e-14, "grad_norm": 0.5473693609237671, "learning_rate": 0.00011363133141298153, "loss": 2.4654, "step": 17270 }, { "epoch": 5.995204332975845e-14, "grad_norm": 0.5311798453330994, "learning_rate": 0.00011361784381165895, "loss": 2.4794, "step": 17280 }, { "epoch": 5.998673779927799e-14, "grad_norm": 0.5482014417648315, "learning_rate": 0.00011360434911487307, "loss": 2.4681, "step": 17290 }, { "epoch": 6.002143226879753e-14, "grad_norm": 0.5429633855819702, "learning_rate": 0.000113590847324758, "loss": 2.4756, "step": 17300 }, { "epoch": 6.005612673831706e-14, "grad_norm": 0.5682976841926575, "learning_rate": 0.00011357733844344891, "loss": 2.4888, "step": 17310 }, { "epoch": 6.00908212078366e-14, "grad_norm": 0.5963857769966125, "learning_rate": 0.00011356382247308212, "loss": 2.5019, "step": 17320 }, { "epoch": 6.012551567735613e-14, "grad_norm": 0.5744146704673767, "learning_rate": 0.00011355029941579509, "loss": 2.4515, "step": 17330 }, { "epoch": 6.016021014687567e-14, "grad_norm": 0.5864333510398865, "learning_rate": 0.00011353676927372635, "loss": 2.4612, "step": 17340 }, { "epoch": 6.01949046163952e-14, "grad_norm": 0.5463144779205322, "learning_rate": 0.00011352323204901556, "loss": 2.4824, "step": 17350 }, { "epoch": 6.022959908591474e-14, "grad_norm": 0.5235787630081177, "learning_rate": 0.00011350968774380353, "loss": 2.4939, "step": 17360 }, { "epoch": 6.026429355543428e-14, "grad_norm": 0.5445688366889954, "learning_rate": 0.00011349613636023217, "loss": 2.5044, "step": 17370 }, { "epoch": 6.029898802495381e-14, "grad_norm": 0.5213082432746887, "learning_rate": 0.00011348257790044456, "loss": 2.4463, "step": 17380 }, { "epoch": 6.033368249447335e-14, "grad_norm": 0.5507453680038452, "learning_rate": 0.00011346901236658479, "loss": 2.4896, "step": 17390 }, { "epoch": 6.036837696399289e-14, "grad_norm": 0.5196648836135864, "learning_rate": 0.00011345543976079813, "loss": 2.5017, "step": 17400 }, { "epoch": 6.040307143351242e-14, "grad_norm": 0.5638561248779297, "learning_rate": 0.00011344186008523105, "loss": 2.4789, "step": 17410 }, { "epoch": 6.043776590303196e-14, "grad_norm": 0.5577253103256226, "learning_rate": 0.00011342827334203099, "loss": 2.4944, "step": 17420 }, { "epoch": 6.04724603725515e-14, "grad_norm": 0.5490243434906006, "learning_rate": 0.00011341467953334658, "loss": 2.5079, "step": 17430 }, { "epoch": 6.050715484207103e-14, "grad_norm": 0.5342481136322021, "learning_rate": 0.0001134010786613276, "loss": 2.474, "step": 17440 }, { "epoch": 6.054184931159057e-14, "grad_norm": 0.5534713268280029, "learning_rate": 0.00011338747072812487, "loss": 2.5249, "step": 17450 }, { "epoch": 6.05765437811101e-14, "grad_norm": 0.5779215693473816, "learning_rate": 0.00011337385573589039, "loss": 2.4997, "step": 17460 }, { "epoch": 6.061123825062964e-14, "grad_norm": 0.5355042219161987, "learning_rate": 0.00011336023368677726, "loss": 2.4689, "step": 17470 }, { "epoch": 6.064593272014918e-14, "grad_norm": 0.5315300822257996, "learning_rate": 0.00011334660458293966, "loss": 2.4416, "step": 17480 }, { "epoch": 6.068062718966871e-14, "grad_norm": 0.5515336990356445, "learning_rate": 0.00011333296842653294, "loss": 2.4917, "step": 17490 }, { "epoch": 6.071532165918825e-14, "grad_norm": 0.5328466892242432, "learning_rate": 0.00011331932521971355, "loss": 2.5049, "step": 17500 }, { "epoch": 6.075001612870778e-14, "grad_norm": 0.5679963231086731, "learning_rate": 0.00011330567496463903, "loss": 2.4673, "step": 17510 }, { "epoch": 6.078471059822732e-14, "grad_norm": 0.5823729038238525, "learning_rate": 0.00011329201766346806, "loss": 2.4839, "step": 17520 }, { "epoch": 6.081940506774686e-14, "grad_norm": 0.571517288684845, "learning_rate": 0.00011327835331836039, "loss": 2.4526, "step": 17530 }, { "epoch": 6.085409953726639e-14, "grad_norm": 0.541929304599762, "learning_rate": 0.00011326468193147698, "loss": 2.4764, "step": 17540 }, { "epoch": 6.088879400678593e-14, "grad_norm": 0.5698961019515991, "learning_rate": 0.00011325100350497977, "loss": 2.5109, "step": 17550 }, { "epoch": 6.092348847630547e-14, "grad_norm": 0.6059116125106812, "learning_rate": 0.00011323731804103193, "loss": 2.462, "step": 17560 }, { "epoch": 6.0958182945825e-14, "grad_norm": 0.6018331050872803, "learning_rate": 0.00011322362554179773, "loss": 2.4949, "step": 17570 }, { "epoch": 6.099287741534454e-14, "grad_norm": 0.5577785968780518, "learning_rate": 0.00011320992600944244, "loss": 2.4898, "step": 17580 }, { "epoch": 6.102757188486407e-14, "grad_norm": 0.5515245199203491, "learning_rate": 0.00011319621944613259, "loss": 2.4769, "step": 17590 }, { "epoch": 6.106226635438361e-14, "grad_norm": 0.5454210638999939, "learning_rate": 0.0001131825058540357, "loss": 2.485, "step": 17600 }, { "epoch": 6.109696082390315e-14, "grad_norm": 0.573880672454834, "learning_rate": 0.00011316878523532052, "loss": 2.4911, "step": 17610 }, { "epoch": 6.113165529342268e-14, "grad_norm": 0.5328211784362793, "learning_rate": 0.00011315505759215682, "loss": 2.4499, "step": 17620 }, { "epoch": 6.116634976294222e-14, "grad_norm": 0.543708324432373, "learning_rate": 0.00011314132292671547, "loss": 2.4903, "step": 17630 }, { "epoch": 6.120104423246175e-14, "grad_norm": 0.597334086894989, "learning_rate": 0.00011312758124116852, "loss": 2.4971, "step": 17640 }, { "epoch": 6.123573870198129e-14, "grad_norm": 0.5273880958557129, "learning_rate": 0.00011311383253768912, "loss": 2.4487, "step": 17650 }, { "epoch": 6.127043317150083e-14, "grad_norm": 0.5402578115463257, "learning_rate": 0.00011310007681845151, "loss": 2.4596, "step": 17660 }, { "epoch": 6.130512764102036e-14, "grad_norm": 0.5660821199417114, "learning_rate": 0.00011308631408563099, "loss": 2.475, "step": 17670 }, { "epoch": 6.13398221105399e-14, "grad_norm": 0.5563436150550842, "learning_rate": 0.00011307254434140404, "loss": 2.4664, "step": 17680 }, { "epoch": 6.137451658005944e-14, "grad_norm": 0.5369067788124084, "learning_rate": 0.00011305876758794827, "loss": 2.4708, "step": 17690 }, { "epoch": 6.140921104957897e-14, "grad_norm": 0.5460045337677002, "learning_rate": 0.0001130449838274423, "loss": 2.4687, "step": 17700 }, { "epoch": 6.144390551909851e-14, "grad_norm": 0.5354008078575134, "learning_rate": 0.00011303119306206592, "loss": 2.4525, "step": 17710 }, { "epoch": 6.147859998861804e-14, "grad_norm": 0.5652183890342712, "learning_rate": 0.00011301739529400004, "loss": 2.4845, "step": 17720 }, { "epoch": 6.151329445813758e-14, "grad_norm": 0.5476153492927551, "learning_rate": 0.00011300359052542666, "loss": 2.477, "step": 17730 }, { "epoch": 6.154798892765712e-14, "grad_norm": 0.5392038226127625, "learning_rate": 0.00011298977875852887, "loss": 2.4871, "step": 17740 }, { "epoch": 6.158268339717665e-14, "grad_norm": 0.5333921313285828, "learning_rate": 0.00011297595999549089, "loss": 2.4618, "step": 17750 }, { "epoch": 6.161737786669619e-14, "grad_norm": 0.5443691611289978, "learning_rate": 0.00011296213423849802, "loss": 2.4937, "step": 17760 }, { "epoch": 6.165207233621572e-14, "grad_norm": 0.5508635640144348, "learning_rate": 0.00011294830148973672, "loss": 2.4428, "step": 17770 }, { "epoch": 6.168676680573526e-14, "grad_norm": 0.540499746799469, "learning_rate": 0.00011293446175139449, "loss": 2.5109, "step": 17780 }, { "epoch": 6.17214612752548e-14, "grad_norm": 0.5328560471534729, "learning_rate": 0.00011292061502565998, "loss": 2.4767, "step": 17790 }, { "epoch": 6.175615574477433e-14, "grad_norm": 0.5726773142814636, "learning_rate": 0.00011290676131472293, "loss": 2.4732, "step": 17800 }, { "epoch": 6.179085021429387e-14, "grad_norm": 0.5373123288154602, "learning_rate": 0.00011289290062077417, "loss": 2.4563, "step": 17810 }, { "epoch": 6.18255446838134e-14, "grad_norm": 0.5860594511032104, "learning_rate": 0.00011287903294600568, "loss": 2.4732, "step": 17820 }, { "epoch": 6.186023915333294e-14, "grad_norm": 0.5218111276626587, "learning_rate": 0.00011286515829261049, "loss": 2.4755, "step": 17830 }, { "epoch": 6.189493362285248e-14, "grad_norm": 0.5783238410949707, "learning_rate": 0.00011285127666278277, "loss": 2.4488, "step": 17840 }, { "epoch": 6.192962809237201e-14, "grad_norm": 0.5698190927505493, "learning_rate": 0.00011283738805871773, "loss": 2.4749, "step": 17850 }, { "epoch": 6.196432256189155e-14, "grad_norm": 0.596146285533905, "learning_rate": 0.00011282349248261182, "loss": 2.4556, "step": 17860 }, { "epoch": 6.199901703141109e-14, "grad_norm": 0.532749354839325, "learning_rate": 0.00011280958993666245, "loss": 2.4572, "step": 17870 }, { "epoch": 6.203371150093062e-14, "grad_norm": 0.5491549968719482, "learning_rate": 0.00011279568042306818, "loss": 2.4978, "step": 17880 }, { "epoch": 6.206840597045016e-14, "grad_norm": 0.5425570011138916, "learning_rate": 0.00011278176394402873, "loss": 2.4752, "step": 17890 }, { "epoch": 6.21031004399697e-14, "grad_norm": 0.5683391094207764, "learning_rate": 0.00011276784050174482, "loss": 2.4528, "step": 17900 }, { "epoch": 6.213779490948923e-14, "grad_norm": 0.5635042190551758, "learning_rate": 0.00011275391009841836, "loss": 2.4007, "step": 17910 }, { "epoch": 6.217248937900877e-14, "grad_norm": 0.6092345714569092, "learning_rate": 0.0001127399727362523, "loss": 2.468, "step": 17920 }, { "epoch": 6.22071838485283e-14, "grad_norm": 0.5409399271011353, "learning_rate": 0.00011272602841745069, "loss": 2.4508, "step": 17930 }, { "epoch": 6.224187831804784e-14, "grad_norm": 0.5653491020202637, "learning_rate": 0.00011271207714421878, "loss": 2.4593, "step": 17940 }, { "epoch": 6.227657278756737e-14, "grad_norm": 0.5450578331947327, "learning_rate": 0.00011269811891876276, "loss": 2.4718, "step": 17950 }, { "epoch": 6.231126725708691e-14, "grad_norm": 0.5584098696708679, "learning_rate": 0.00011268415374329007, "loss": 2.4637, "step": 17960 }, { "epoch": 6.234596172660645e-14, "grad_norm": 0.5727172493934631, "learning_rate": 0.00011267018162000913, "loss": 2.4495, "step": 17970 }, { "epoch": 6.238065619612598e-14, "grad_norm": 0.541060209274292, "learning_rate": 0.00011265620255112954, "loss": 2.4373, "step": 17980 }, { "epoch": 6.241535066564552e-14, "grad_norm": 0.5480059385299683, "learning_rate": 0.00011264221653886198, "loss": 2.4374, "step": 17990 }, { "epoch": 6.245004513516506e-14, "grad_norm": 0.5503222346305847, "learning_rate": 0.00011262822358541818, "loss": 2.4553, "step": 18000 }, { "epoch": 6.248473960468459e-14, "grad_norm": 0.5992000699043274, "learning_rate": 0.00011261422369301104, "loss": 2.4676, "step": 18010 }, { "epoch": 6.251943407420413e-14, "grad_norm": 0.5458118915557861, "learning_rate": 0.00011260021686385451, "loss": 2.4932, "step": 18020 }, { "epoch": 6.255412854372366e-14, "grad_norm": 0.5618975162506104, "learning_rate": 0.00011258620310016365, "loss": 2.4863, "step": 18030 }, { "epoch": 6.25888230132432e-14, "grad_norm": 0.5647193789482117, "learning_rate": 0.0001125721824041546, "loss": 2.4848, "step": 18040 }, { "epoch": 6.262351748276274e-14, "grad_norm": 0.5733253359794617, "learning_rate": 0.00011255815477804464, "loss": 2.461, "step": 18050 }, { "epoch": 6.265821195228227e-14, "grad_norm": 0.5526642203330994, "learning_rate": 0.0001125441202240521, "loss": 2.4747, "step": 18060 }, { "epoch": 6.269290642180181e-14, "grad_norm": 0.5308496356010437, "learning_rate": 0.00011253007874439643, "loss": 2.4547, "step": 18070 }, { "epoch": 6.272760089132134e-14, "grad_norm": 0.5855588912963867, "learning_rate": 0.00011251603034129815, "loss": 2.4427, "step": 18080 }, { "epoch": 6.276229536084088e-14, "grad_norm": 0.530491292476654, "learning_rate": 0.00011250197501697894, "loss": 2.4391, "step": 18090 }, { "epoch": 6.279698983036042e-14, "grad_norm": 0.5185192823410034, "learning_rate": 0.0001124879127736615, "loss": 2.4589, "step": 18100 }, { "epoch": 6.283168429987995e-14, "grad_norm": 0.5752148032188416, "learning_rate": 0.00011247384361356962, "loss": 2.4689, "step": 18110 }, { "epoch": 6.286637876939949e-14, "grad_norm": 0.5574005842208862, "learning_rate": 0.00011245976753892826, "loss": 2.475, "step": 18120 }, { "epoch": 6.290107323891903e-14, "grad_norm": 0.5788752436637878, "learning_rate": 0.00011244568455196345, "loss": 2.4298, "step": 18130 }, { "epoch": 6.293576770843856e-14, "grad_norm": 0.5493202805519104, "learning_rate": 0.00011243159465490225, "loss": 2.4541, "step": 18140 }, { "epoch": 6.29704621779581e-14, "grad_norm": 0.5668098330497742, "learning_rate": 0.00011241749784997285, "loss": 2.4291, "step": 18150 }, { "epoch": 6.300515664747763e-14, "grad_norm": 0.5645730495452881, "learning_rate": 0.00011240339413940457, "loss": 2.459, "step": 18160 }, { "epoch": 6.303985111699717e-14, "grad_norm": 0.5409650802612305, "learning_rate": 0.00011238928352542778, "loss": 2.4662, "step": 18170 }, { "epoch": 6.30745455865167e-14, "grad_norm": 0.5812675952911377, "learning_rate": 0.00011237516601027395, "loss": 2.4237, "step": 18180 }, { "epoch": 6.310924005603624e-14, "grad_norm": 0.5264400839805603, "learning_rate": 0.00011236104159617565, "loss": 2.4595, "step": 18190 }, { "epoch": 6.314393452555578e-14, "grad_norm": 0.5421088337898254, "learning_rate": 0.00011234691028536652, "loss": 2.4311, "step": 18200 }, { "epoch": 6.317862899507531e-14, "grad_norm": 0.5764902234077454, "learning_rate": 0.00011233277208008131, "loss": 2.4578, "step": 18210 }, { "epoch": 6.321332346459485e-14, "grad_norm": 0.5411604046821594, "learning_rate": 0.00011231862698255587, "loss": 2.4754, "step": 18220 }, { "epoch": 6.324801793411439e-14, "grad_norm": 0.5425264835357666, "learning_rate": 0.00011230447499502709, "loss": 2.4475, "step": 18230 }, { "epoch": 6.328271240363392e-14, "grad_norm": 0.5495361685752869, "learning_rate": 0.00011229031611973304, "loss": 2.4649, "step": 18240 }, { "epoch": 6.331740687315346e-14, "grad_norm": 0.5685843825340271, "learning_rate": 0.00011227615035891277, "loss": 2.467, "step": 18250 }, { "epoch": 6.3352101342673e-14, "grad_norm": 0.5484800934791565, "learning_rate": 0.00011226197771480652, "loss": 2.4931, "step": 18260 }, { "epoch": 6.338679581219253e-14, "grad_norm": 0.5607442259788513, "learning_rate": 0.00011224779818965554, "loss": 2.4517, "step": 18270 }, { "epoch": 6.342149028171207e-14, "grad_norm": 0.5729661583900452, "learning_rate": 0.0001122336117857022, "loss": 2.5001, "step": 18280 }, { "epoch": 6.34561847512316e-14, "grad_norm": 0.5813125967979431, "learning_rate": 0.00011221941850518997, "loss": 2.4706, "step": 18290 }, { "epoch": 6.349087922075114e-14, "grad_norm": 0.5906779766082764, "learning_rate": 0.00011220521835036337, "loss": 2.468, "step": 18300 }, { "epoch": 6.352557369027068e-14, "grad_norm": 0.548352837562561, "learning_rate": 0.00011219101132346808, "loss": 2.4487, "step": 18310 }, { "epoch": 6.356026815979021e-14, "grad_norm": 0.5352654457092285, "learning_rate": 0.00011217679742675079, "loss": 2.4443, "step": 18320 }, { "epoch": 6.359496262930975e-14, "grad_norm": 0.5581517815589905, "learning_rate": 0.00011216257666245928, "loss": 2.4889, "step": 18330 }, { "epoch": 6.362965709882928e-14, "grad_norm": 0.5488559603691101, "learning_rate": 0.0001121483490328425, "loss": 2.4261, "step": 18340 }, { "epoch": 6.366435156834882e-14, "grad_norm": 0.5435758233070374, "learning_rate": 0.00011213411454015039, "loss": 2.4492, "step": 18350 }, { "epoch": 6.369904603786836e-14, "grad_norm": 0.5585180521011353, "learning_rate": 0.00011211987318663402, "loss": 2.4623, "step": 18360 }, { "epoch": 6.373374050738789e-14, "grad_norm": 0.545933723449707, "learning_rate": 0.00011210562497454552, "loss": 2.4567, "step": 18370 }, { "epoch": 6.376843497690743e-14, "grad_norm": 0.5426208972930908, "learning_rate": 0.00011209136990613814, "loss": 2.4223, "step": 18380 }, { "epoch": 6.380312944642696e-14, "grad_norm": 0.5708092451095581, "learning_rate": 0.0001120771079836662, "loss": 2.4671, "step": 18390 }, { "epoch": 6.38378239159465e-14, "grad_norm": 0.6105402112007141, "learning_rate": 0.0001120628392093851, "loss": 2.445, "step": 18400 }, { "epoch": 6.387251838546604e-14, "grad_norm": 0.5775161385536194, "learning_rate": 0.00011204856358555132, "loss": 2.482, "step": 18410 }, { "epoch": 6.390721285498557e-14, "grad_norm": 0.5142259001731873, "learning_rate": 0.00011203428111442244, "loss": 2.446, "step": 18420 }, { "epoch": 6.394190732450511e-14, "grad_norm": 0.5621064305305481, "learning_rate": 0.00011201999179825706, "loss": 2.4683, "step": 18430 }, { "epoch": 6.397660179402465e-14, "grad_norm": 0.5369477868080139, "learning_rate": 0.00011200569563931497, "loss": 2.4345, "step": 18440 }, { "epoch": 6.401129626354418e-14, "grad_norm": 0.605520486831665, "learning_rate": 0.00011199139263985694, "loss": 2.4565, "step": 18450 }, { "epoch": 6.404599073306372e-14, "grad_norm": 0.5395402312278748, "learning_rate": 0.00011197708280214492, "loss": 2.4634, "step": 18460 }, { "epoch": 6.408068520258325e-14, "grad_norm": 0.5307486653327942, "learning_rate": 0.00011196276612844184, "loss": 2.4639, "step": 18470 }, { "epoch": 6.411537967210279e-14, "grad_norm": 0.5269826650619507, "learning_rate": 0.00011194844262101176, "loss": 2.4535, "step": 18480 }, { "epoch": 6.415007414162233e-14, "grad_norm": 0.5264390110969543, "learning_rate": 0.00011193411228211986, "loss": 2.4645, "step": 18490 }, { "epoch": 6.418476861114186e-14, "grad_norm": 0.6164603233337402, "learning_rate": 0.00011191977511403232, "loss": 2.4521, "step": 18500 }, { "epoch": 6.42194630806614e-14, "grad_norm": 0.571496307849884, "learning_rate": 0.00011190543111901646, "loss": 2.4437, "step": 18510 }, { "epoch": 6.425415755018093e-14, "grad_norm": 0.5794401168823242, "learning_rate": 0.00011189108029934065, "loss": 2.4798, "step": 18520 }, { "epoch": 6.428885201970047e-14, "grad_norm": 0.5441077947616577, "learning_rate": 0.00011187672265727435, "loss": 2.4238, "step": 18530 }, { "epoch": 6.432354648922001e-14, "grad_norm": 0.5612691044807434, "learning_rate": 0.00011186235819508811, "loss": 2.4651, "step": 18540 }, { "epoch": 6.435824095873954e-14, "grad_norm": 0.5415865778923035, "learning_rate": 0.00011184798691505352, "loss": 2.467, "step": 18550 }, { "epoch": 6.439293542825908e-14, "grad_norm": 0.5529482960700989, "learning_rate": 0.0001118336088194433, "loss": 2.464, "step": 18560 }, { "epoch": 6.442762989777862e-14, "grad_norm": 0.5468656420707703, "learning_rate": 0.0001118192239105312, "loss": 2.4649, "step": 18570 }, { "epoch": 6.446232436729815e-14, "grad_norm": 0.5621579885482788, "learning_rate": 0.00011180483219059209, "loss": 2.4592, "step": 18580 }, { "epoch": 6.449701883681769e-14, "grad_norm": 0.5354814529418945, "learning_rate": 0.00011179043366190189, "loss": 2.4939, "step": 18590 }, { "epoch": 6.453171330633722e-14, "grad_norm": 0.5615726709365845, "learning_rate": 0.00011177602832673761, "loss": 2.4631, "step": 18600 }, { "epoch": 6.456640777585676e-14, "grad_norm": 0.574696958065033, "learning_rate": 0.0001117616161873773, "loss": 2.4459, "step": 18610 }, { "epoch": 6.46011022453763e-14, "grad_norm": 0.5767737030982971, "learning_rate": 0.00011174719724610014, "loss": 2.4668, "step": 18620 }, { "epoch": 6.463579671489583e-14, "grad_norm": 0.5558457374572754, "learning_rate": 0.00011173277150518638, "loss": 2.45, "step": 18630 }, { "epoch": 6.467049118441537e-14, "grad_norm": 0.5616034865379333, "learning_rate": 0.00011171833896691728, "loss": 2.4347, "step": 18640 }, { "epoch": 6.47051856539349e-14, "grad_norm": 0.5832933187484741, "learning_rate": 0.00011170389963357525, "loss": 2.4626, "step": 18650 }, { "epoch": 6.473988012345444e-14, "grad_norm": 0.5956979990005493, "learning_rate": 0.00011168945350744375, "loss": 2.4662, "step": 18660 }, { "epoch": 6.477457459297398e-14, "grad_norm": 1.79346764087677, "learning_rate": 0.00011167500059080729, "loss": 2.4978, "step": 18670 }, { "epoch": 6.480926906249351e-14, "grad_norm": 0.5509292483329773, "learning_rate": 0.00011166054088595149, "loss": 2.4736, "step": 18680 }, { "epoch": 6.484396353201305e-14, "grad_norm": 0.5462419390678406, "learning_rate": 0.00011164607439516304, "loss": 2.4372, "step": 18690 }, { "epoch": 6.487865800153259e-14, "grad_norm": 0.5649743676185608, "learning_rate": 0.00011163160112072967, "loss": 2.4494, "step": 18700 }, { "epoch": 6.491335247105212e-14, "grad_norm": 0.6353193521499634, "learning_rate": 0.0001116171210649402, "loss": 2.4359, "step": 18710 }, { "epoch": 6.494804694057166e-14, "grad_norm": 0.5467793345451355, "learning_rate": 0.00011160263423008454, "loss": 2.4599, "step": 18720 }, { "epoch": 6.49827414100912e-14, "grad_norm": 0.5529032945632935, "learning_rate": 0.00011158814061845367, "loss": 2.449, "step": 18730 }, { "epoch": 6.501743587961073e-14, "grad_norm": 0.5653977394104004, "learning_rate": 0.00011157364023233961, "loss": 2.4568, "step": 18740 }, { "epoch": 6.505213034913027e-14, "grad_norm": 0.56126868724823, "learning_rate": 0.00011155913307403547, "loss": 2.4259, "step": 18750 }, { "epoch": 6.50868248186498e-14, "grad_norm": 0.5474733710289001, "learning_rate": 0.00011154461914583543, "loss": 2.4582, "step": 18760 }, { "epoch": 6.512151928816934e-14, "grad_norm": 0.5351760387420654, "learning_rate": 0.00011153009845003476, "loss": 2.4218, "step": 18770 }, { "epoch": 6.515621375768887e-14, "grad_norm": 0.5825449824333191, "learning_rate": 0.00011151557098892978, "loss": 2.4126, "step": 18780 }, { "epoch": 6.519090822720841e-14, "grad_norm": 0.5795783400535583, "learning_rate": 0.00011150103676481786, "loss": 2.4552, "step": 18790 }, { "epoch": 6.522560269672795e-14, "grad_norm": 0.5454334616661072, "learning_rate": 0.00011148649577999748, "loss": 2.4613, "step": 18800 }, { "epoch": 6.526029716624748e-14, "grad_norm": 0.5439749956130981, "learning_rate": 0.00011147194803676819, "loss": 2.4625, "step": 18810 }, { "epoch": 6.529499163576702e-14, "grad_norm": 0.5642018914222717, "learning_rate": 0.00011145739353743055, "loss": 2.4774, "step": 18820 }, { "epoch": 6.532968610528656e-14, "grad_norm": 0.553961455821991, "learning_rate": 0.00011144283228428626, "loss": 2.4668, "step": 18830 }, { "epoch": 6.536438057480609e-14, "grad_norm": 0.5630622506141663, "learning_rate": 0.00011142826427963805, "loss": 2.4281, "step": 18840 }, { "epoch": 6.539907504432563e-14, "grad_norm": 0.5307106375694275, "learning_rate": 0.00011141368952578972, "loss": 2.4412, "step": 18850 }, { "epoch": 6.543376951384516e-14, "grad_norm": 0.545993983745575, "learning_rate": 0.00011139910802504612, "loss": 2.4617, "step": 18860 }, { "epoch": 6.54684639833647e-14, "grad_norm": 0.5470076203346252, "learning_rate": 0.00011138451977971325, "loss": 2.4576, "step": 18870 }, { "epoch": 6.550315845288424e-14, "grad_norm": 0.5418696403503418, "learning_rate": 0.00011136992479209805, "loss": 2.4498, "step": 18880 }, { "epoch": 6.553785292240377e-14, "grad_norm": 0.5659899711608887, "learning_rate": 0.00011135532306450862, "loss": 2.4313, "step": 18890 }, { "epoch": 6.557254739192331e-14, "grad_norm": 0.540299654006958, "learning_rate": 0.00011134071459925413, "loss": 2.4156, "step": 18900 }, { "epoch": 6.560724186144284e-14, "grad_norm": 0.5442661643028259, "learning_rate": 0.00011132609939864471, "loss": 2.3884, "step": 18910 }, { "epoch": 6.564193633096238e-14, "grad_norm": 0.5622106790542603, "learning_rate": 0.00011131147746499169, "loss": 2.4835, "step": 18920 }, { "epoch": 6.567663080048192e-14, "grad_norm": 0.5685534477233887, "learning_rate": 0.00011129684880060738, "loss": 2.4684, "step": 18930 }, { "epoch": 6.571132527000145e-14, "grad_norm": 0.5346983671188354, "learning_rate": 0.00011128221340780516, "loss": 2.499, "step": 18940 }, { "epoch": 6.574601973952099e-14, "grad_norm": 0.5241139531135559, "learning_rate": 0.00011126757128889953, "loss": 2.4799, "step": 18950 }, { "epoch": 6.578071420904053e-14, "grad_norm": 0.5421934723854065, "learning_rate": 0.00011125292244620598, "loss": 2.4349, "step": 18960 }, { "epoch": 6.581540867856006e-14, "grad_norm": 0.556093692779541, "learning_rate": 0.0001112382668820411, "loss": 2.4635, "step": 18970 }, { "epoch": 6.58501031480796e-14, "grad_norm": 0.5471346378326416, "learning_rate": 0.0001112236045987226, "loss": 2.4375, "step": 18980 }, { "epoch": 6.588479761759913e-14, "grad_norm": 0.5473746061325073, "learning_rate": 0.0001112089355985691, "loss": 2.4349, "step": 18990 }, { "epoch": 6.591949208711867e-14, "grad_norm": 0.5909225940704346, "learning_rate": 0.00011119425988390044, "loss": 2.4247, "step": 19000 }, { "epoch": 6.59541865566382e-14, "grad_norm": 0.5499498844146729, "learning_rate": 0.00011117957745703746, "loss": 2.4469, "step": 19010 }, { "epoch": 6.598888102615774e-14, "grad_norm": 0.545418918132782, "learning_rate": 0.00011116488832030202, "loss": 2.4371, "step": 19020 }, { "epoch": 6.602357549567728e-14, "grad_norm": 0.5435922145843506, "learning_rate": 0.00011115019247601712, "loss": 2.4317, "step": 19030 }, { "epoch": 6.605826996519681e-14, "grad_norm": 0.5518187880516052, "learning_rate": 0.00011113548992650674, "loss": 2.441, "step": 19040 }, { "epoch": 6.609296443471635e-14, "grad_norm": 0.5362842679023743, "learning_rate": 0.00011112078067409599, "loss": 2.4709, "step": 19050 }, { "epoch": 6.612765890423589e-14, "grad_norm": 0.5993102192878723, "learning_rate": 0.00011110606472111103, "loss": 2.4436, "step": 19060 }, { "epoch": 6.616235337375542e-14, "grad_norm": 0.5321890115737915, "learning_rate": 0.000111091342069879, "loss": 2.4751, "step": 19070 }, { "epoch": 6.619704784327496e-14, "grad_norm": 0.5349705219268799, "learning_rate": 0.00011107661272272825, "loss": 2.4512, "step": 19080 }, { "epoch": 6.62317423127945e-14, "grad_norm": 0.5906382203102112, "learning_rate": 0.000111061876681988, "loss": 2.4324, "step": 19090 }, { "epoch": 6.626643678231403e-14, "grad_norm": 0.5592759847640991, "learning_rate": 0.0001110471339499887, "loss": 2.4086, "step": 19100 }, { "epoch": 6.630113125183357e-14, "grad_norm": 0.5751851797103882, "learning_rate": 0.00011103238452906175, "loss": 2.4285, "step": 19110 }, { "epoch": 6.63358257213531e-14, "grad_norm": 0.562889039516449, "learning_rate": 0.00011101762842153966, "loss": 2.4335, "step": 19120 }, { "epoch": 6.637052019087264e-14, "grad_norm": 0.5629086494445801, "learning_rate": 0.00011100286562975599, "loss": 2.4284, "step": 19130 }, { "epoch": 6.640521466039218e-14, "grad_norm": 0.552021861076355, "learning_rate": 0.00011098809615604533, "loss": 2.4303, "step": 19140 }, { "epoch": 6.643990912991171e-14, "grad_norm": 0.5585321187973022, "learning_rate": 0.00011097332000274333, "loss": 2.4558, "step": 19150 }, { "epoch": 6.647460359943125e-14, "grad_norm": 0.5363466143608093, "learning_rate": 0.00011095853717218675, "loss": 2.4555, "step": 19160 }, { "epoch": 6.650929806895078e-14, "grad_norm": 0.5767779350280762, "learning_rate": 0.00011094374766671334, "loss": 2.446, "step": 19170 }, { "epoch": 6.654399253847032e-14, "grad_norm": 0.5697128772735596, "learning_rate": 0.00011092895148866195, "loss": 2.4565, "step": 19180 }, { "epoch": 6.657868700798986e-14, "grad_norm": 0.5487766861915588, "learning_rate": 0.00011091414864037249, "loss": 2.4233, "step": 19190 }, { "epoch": 6.661338147750939e-14, "grad_norm": 0.5335415601730347, "learning_rate": 0.00011089933912418583, "loss": 2.4572, "step": 19200 }, { "epoch": 6.664807594702893e-14, "grad_norm": 0.5832814574241638, "learning_rate": 0.00011088452294244403, "loss": 2.4034, "step": 19210 }, { "epoch": 6.668277041654846e-14, "grad_norm": 0.5642853379249573, "learning_rate": 0.00011086970009749011, "loss": 2.4445, "step": 19220 }, { "epoch": 6.6717464886068e-14, "grad_norm": 0.5754279494285583, "learning_rate": 0.00011085487059166821, "loss": 2.4589, "step": 19230 }, { "epoch": 6.675215935558754e-14, "grad_norm": 0.5799453258514404, "learning_rate": 0.00011084003442732346, "loss": 2.4348, "step": 19240 }, { "epoch": 6.678685382510707e-14, "grad_norm": 0.5478160381317139, "learning_rate": 0.00011082519160680206, "loss": 2.436, "step": 19250 }, { "epoch": 6.682154829462661e-14, "grad_norm": 0.5433496236801147, "learning_rate": 0.00011081034213245132, "loss": 2.4306, "step": 19260 }, { "epoch": 6.685624276414615e-14, "grad_norm": 0.5446781516075134, "learning_rate": 0.00011079548600661951, "loss": 2.4325, "step": 19270 }, { "epoch": 6.689093723366568e-14, "grad_norm": 0.5293609499931335, "learning_rate": 0.00011078062323165603, "loss": 2.4079, "step": 19280 }, { "epoch": 6.692563170318522e-14, "grad_norm": 0.5382822155952454, "learning_rate": 0.00011076575380991129, "loss": 2.41, "step": 19290 }, { "epoch": 6.696032617270475e-14, "grad_norm": 0.5610021948814392, "learning_rate": 0.00011075087774373674, "loss": 2.453, "step": 19300 }, { "epoch": 6.699502064222429e-14, "grad_norm": 0.5634414553642273, "learning_rate": 0.00011073599503548495, "loss": 2.3878, "step": 19310 }, { "epoch": 6.702971511174383e-14, "grad_norm": 0.5277231335639954, "learning_rate": 0.00011072110568750946, "loss": 2.4315, "step": 19320 }, { "epoch": 6.706440958126336e-14, "grad_norm": 0.5444855093955994, "learning_rate": 0.00011070620970216487, "loss": 2.4398, "step": 19330 }, { "epoch": 6.70991040507829e-14, "grad_norm": 0.5612186789512634, "learning_rate": 0.0001106913070818069, "loss": 2.4575, "step": 19340 }, { "epoch": 6.713379852030243e-14, "grad_norm": 0.5690255761146545, "learning_rate": 0.00011067639782879223, "loss": 2.4258, "step": 19350 }, { "epoch": 6.716849298982197e-14, "grad_norm": 0.5430018305778503, "learning_rate": 0.00011066148194547863, "loss": 2.4896, "step": 19360 }, { "epoch": 6.720318745934151e-14, "grad_norm": 0.5949823260307312, "learning_rate": 0.00011064655943422499, "loss": 2.4542, "step": 19370 }, { "epoch": 6.723788192886104e-14, "grad_norm": 0.5454648733139038, "learning_rate": 0.00011063163029739108, "loss": 2.4696, "step": 19380 }, { "epoch": 6.727257639838058e-14, "grad_norm": 0.560668408870697, "learning_rate": 0.00011061669453733789, "loss": 2.4499, "step": 19390 }, { "epoch": 6.730727086790012e-14, "grad_norm": 0.5490331053733826, "learning_rate": 0.00011060175215642733, "loss": 2.485, "step": 19400 }, { "epoch": 6.734196533741965e-14, "grad_norm": 0.5486035346984863, "learning_rate": 0.00011058680315702243, "loss": 2.4384, "step": 19410 }, { "epoch": 6.737665980693919e-14, "grad_norm": 0.5826327204704285, "learning_rate": 0.00011057184754148725, "loss": 2.452, "step": 19420 }, { "epoch": 6.741135427645872e-14, "grad_norm": 0.5675860047340393, "learning_rate": 0.00011055688531218689, "loss": 2.4473, "step": 19430 }, { "epoch": 6.744604874597826e-14, "grad_norm": 0.5704185366630554, "learning_rate": 0.00011054191647148748, "loss": 2.4351, "step": 19440 }, { "epoch": 6.74807432154978e-14, "grad_norm": 0.5495672225952148, "learning_rate": 0.00011052694102175622, "loss": 2.4495, "step": 19450 }, { "epoch": 6.751543768501733e-14, "grad_norm": 0.5418678522109985, "learning_rate": 0.00011051195896536137, "loss": 2.4305, "step": 19460 }, { "epoch": 6.755013215453687e-14, "grad_norm": 0.5466930270195007, "learning_rate": 0.00011049697030467222, "loss": 2.423, "step": 19470 }, { "epoch": 6.75848266240564e-14, "grad_norm": 0.5865089893341064, "learning_rate": 0.00011048197504205904, "loss": 2.4642, "step": 19480 }, { "epoch": 6.761952109357594e-14, "grad_norm": 0.5280202031135559, "learning_rate": 0.00011046697317989325, "loss": 2.456, "step": 19490 }, { "epoch": 6.765421556309548e-14, "grad_norm": 0.5504178404808044, "learning_rate": 0.00011045196472054729, "loss": 2.4597, "step": 19500 }, { "epoch": 6.768891003261501e-14, "grad_norm": 0.5624188780784607, "learning_rate": 0.00011043694966639453, "loss": 2.4431, "step": 19510 }, { "epoch": 6.772360450213455e-14, "grad_norm": 0.5418493747711182, "learning_rate": 0.00011042192801980955, "loss": 2.4435, "step": 19520 }, { "epoch": 6.775829897165409e-14, "grad_norm": 0.5526648759841919, "learning_rate": 0.00011040689978316787, "loss": 2.4106, "step": 19530 }, { "epoch": 6.779299344117362e-14, "grad_norm": 0.5792639255523682, "learning_rate": 0.00011039186495884606, "loss": 2.4494, "step": 19540 }, { "epoch": 6.782768791069316e-14, "grad_norm": 0.5706257820129395, "learning_rate": 0.00011037682354922177, "loss": 2.5151, "step": 19550 }, { "epoch": 6.786238238021269e-14, "grad_norm": 0.5643436312675476, "learning_rate": 0.00011036177555667366, "loss": 2.4523, "step": 19560 }, { "epoch": 6.789707684973223e-14, "grad_norm": 0.5787794589996338, "learning_rate": 0.00011034672098358145, "loss": 2.4341, "step": 19570 }, { "epoch": 6.793177131925177e-14, "grad_norm": 0.5340955853462219, "learning_rate": 0.00011033165983232587, "loss": 2.426, "step": 19580 }, { "epoch": 6.79664657887713e-14, "grad_norm": 0.5407457947731018, "learning_rate": 0.00011031659210528874, "loss": 2.4146, "step": 19590 }, { "epoch": 6.800116025829084e-14, "grad_norm": 0.5550571084022522, "learning_rate": 0.00011030151780485287, "loss": 2.4232, "step": 19600 }, { "epoch": 6.803585472781037e-14, "grad_norm": 0.5606695413589478, "learning_rate": 0.00011028643693340213, "loss": 2.4346, "step": 19610 }, { "epoch": 6.807054919732991e-14, "grad_norm": 0.5338703989982605, "learning_rate": 0.00011027134949332145, "loss": 2.4198, "step": 19620 }, { "epoch": 6.810524366684945e-14, "grad_norm": 0.578831672668457, "learning_rate": 0.00011025625548699675, "loss": 2.4276, "step": 19630 }, { "epoch": 6.813993813636898e-14, "grad_norm": 0.5443309545516968, "learning_rate": 0.00011024115491681506, "loss": 2.4527, "step": 19640 }, { "epoch": 6.817463260588852e-14, "grad_norm": 0.5369391441345215, "learning_rate": 0.00011022604778516435, "loss": 2.4739, "step": 19650 }, { "epoch": 6.820932707540805e-14, "grad_norm": 0.5584915280342102, "learning_rate": 0.00011021093409443372, "loss": 2.4644, "step": 19660 }, { "epoch": 6.824402154492759e-14, "grad_norm": 0.5516752600669861, "learning_rate": 0.00011019581384701327, "loss": 2.4659, "step": 19670 }, { "epoch": 6.827871601444713e-14, "grad_norm": 0.5436469912528992, "learning_rate": 0.00011018068704529412, "loss": 2.4572, "step": 19680 }, { "epoch": 6.831341048396666e-14, "grad_norm": 0.6561904549598694, "learning_rate": 0.00011016555369166844, "loss": 2.4268, "step": 19690 }, { "epoch": 6.83481049534862e-14, "grad_norm": 0.5561536550521851, "learning_rate": 0.00011015041378852948, "loss": 2.3956, "step": 19700 }, { "epoch": 6.838279942300574e-14, "grad_norm": 0.5438426733016968, "learning_rate": 0.00011013526733827144, "loss": 2.445, "step": 19710 }, { "epoch": 6.841749389252527e-14, "grad_norm": 0.579338788986206, "learning_rate": 0.0001101201143432896, "loss": 2.4409, "step": 19720 }, { "epoch": 6.845218836204481e-14, "grad_norm": 0.6125617623329163, "learning_rate": 0.00011010495480598032, "loss": 2.4453, "step": 19730 }, { "epoch": 6.848688283156434e-14, "grad_norm": 0.5424051284790039, "learning_rate": 0.00011008978872874091, "loss": 2.4431, "step": 19740 }, { "epoch": 6.852157730108388e-14, "grad_norm": 0.5434989333152771, "learning_rate": 0.00011007461611396978, "loss": 2.435, "step": 19750 }, { "epoch": 6.855627177060342e-14, "grad_norm": 0.5475049018859863, "learning_rate": 0.00011005943696406633, "loss": 2.448, "step": 19760 }, { "epoch": 6.859096624012295e-14, "grad_norm": 0.5691448450088501, "learning_rate": 0.000110044251281431, "loss": 2.4189, "step": 19770 }, { "epoch": 6.862566070964249e-14, "grad_norm": 0.5542029738426208, "learning_rate": 0.00011002905906846532, "loss": 2.4322, "step": 19780 }, { "epoch": 6.866035517916202e-14, "grad_norm": 0.5835248231887817, "learning_rate": 0.00011001386032757176, "loss": 2.4157, "step": 19790 }, { "epoch": 6.869504964868156e-14, "grad_norm": 0.5740675330162048, "learning_rate": 0.00010999865506115392, "loss": 2.4252, "step": 19800 }, { "epoch": 6.87297441182011e-14, "grad_norm": 0.5550439953804016, "learning_rate": 0.00010998344327161632, "loss": 2.4349, "step": 19810 }, { "epoch": 6.876443858772063e-14, "grad_norm": 0.5705216526985168, "learning_rate": 0.00010996822496136462, "loss": 2.4869, "step": 19820 }, { "epoch": 6.879913305724017e-14, "grad_norm": 0.5545175075531006, "learning_rate": 0.00010995300013280546, "loss": 2.4357, "step": 19830 }, { "epoch": 6.88338275267597e-14, "grad_norm": 0.545316755771637, "learning_rate": 0.0001099377687883465, "loss": 2.4148, "step": 19840 }, { "epoch": 6.886852199627924e-14, "grad_norm": 0.5755236744880676, "learning_rate": 0.00010992253093039644, "loss": 2.3955, "step": 19850 }, { "epoch": 6.890321646579878e-14, "grad_norm": 0.5612711310386658, "learning_rate": 0.00010990728656136504, "loss": 2.4383, "step": 19860 }, { "epoch": 6.893791093531831e-14, "grad_norm": 0.5408034920692444, "learning_rate": 0.00010989203568366305, "loss": 2.4351, "step": 19870 }, { "epoch": 6.897260540483785e-14, "grad_norm": 0.5979433655738831, "learning_rate": 0.00010987677829970228, "loss": 2.4303, "step": 19880 }, { "epoch": 6.900729987435739e-14, "grad_norm": 0.5638786554336548, "learning_rate": 0.00010986151441189555, "loss": 2.4515, "step": 19890 }, { "epoch": 6.904199434387692e-14, "grad_norm": 0.5688802003860474, "learning_rate": 0.0001098462440226567, "loss": 2.4452, "step": 19900 }, { "epoch": 6.907668881339646e-14, "grad_norm": 0.5593293905258179, "learning_rate": 0.00010983096713440061, "loss": 2.4383, "step": 19910 }, { "epoch": 6.9111383282916e-14, "grad_norm": 0.5829849243164062, "learning_rate": 0.00010981568374954318, "loss": 2.4492, "step": 19920 }, { "epoch": 6.914607775243553e-14, "grad_norm": 0.634812593460083, "learning_rate": 0.00010980039387050137, "loss": 2.4233, "step": 19930 }, { "epoch": 6.918077222195507e-14, "grad_norm": 0.5573734641075134, "learning_rate": 0.00010978509749969317, "loss": 2.4713, "step": 19940 }, { "epoch": 6.92154666914746e-14, "grad_norm": 0.5528643131256104, "learning_rate": 0.00010976979463953748, "loss": 2.4266, "step": 19950 }, { "epoch": 6.925016116099414e-14, "grad_norm": 0.5551659464836121, "learning_rate": 0.0001097544852924544, "loss": 2.4428, "step": 19960 }, { "epoch": 6.928485563051368e-14, "grad_norm": 0.5367408990859985, "learning_rate": 0.00010973916946086492, "loss": 2.4435, "step": 19970 }, { "epoch": 6.931955010003321e-14, "grad_norm": 0.5567007064819336, "learning_rate": 0.00010972384714719116, "loss": 2.4572, "step": 19980 }, { "epoch": 6.935424456955275e-14, "grad_norm": 0.5454758405685425, "learning_rate": 0.00010970851835385616, "loss": 2.4722, "step": 19990 }, { "epoch": 6.938893903907228e-14, "grad_norm": 0.5598333477973938, "learning_rate": 0.00010969318308328404, "loss": 2.4251, "step": 20000 }, { "epoch": 6.942363350859182e-14, "grad_norm": 0.5563428401947021, "learning_rate": 0.00010967784133789998, "loss": 2.4671, "step": 20010 }, { "epoch": 6.945832797811136e-14, "grad_norm": 0.5398301482200623, "learning_rate": 0.0001096624931201301, "loss": 2.4401, "step": 20020 }, { "epoch": 6.949302244763089e-14, "grad_norm": 0.5770026445388794, "learning_rate": 0.00010964713843240164, "loss": 2.4511, "step": 20030 }, { "epoch": 6.952771691715043e-14, "grad_norm": 0.6415372490882874, "learning_rate": 0.00010963177727714275, "loss": 2.4448, "step": 20040 }, { "epoch": 6.956241138666996e-14, "grad_norm": 0.5793150067329407, "learning_rate": 0.00010961640965678273, "loss": 2.4487, "step": 20050 }, { "epoch": 6.95971058561895e-14, "grad_norm": 0.5750124454498291, "learning_rate": 0.00010960103557375177, "loss": 2.4296, "step": 20060 }, { "epoch": 6.963180032570904e-14, "grad_norm": 0.9373526573181152, "learning_rate": 0.0001095856550304812, "loss": 2.4603, "step": 20070 }, { "epoch": 6.966649479522857e-14, "grad_norm": 0.5860679745674133, "learning_rate": 0.0001095702680294033, "loss": 2.4342, "step": 20080 }, { "epoch": 6.970118926474811e-14, "grad_norm": 0.5422905683517456, "learning_rate": 0.00010955487457295139, "loss": 2.4354, "step": 20090 }, { "epoch": 6.973588373426765e-14, "grad_norm": 0.5795828700065613, "learning_rate": 0.00010953947466355982, "loss": 2.4644, "step": 20100 }, { "epoch": 6.977057820378718e-14, "grad_norm": 0.5652391910552979, "learning_rate": 0.00010952406830366396, "loss": 2.3953, "step": 20110 }, { "epoch": 6.980527267330672e-14, "grad_norm": 0.5525239109992981, "learning_rate": 0.00010950865549570017, "loss": 2.4557, "step": 20120 }, { "epoch": 6.983996714282625e-14, "grad_norm": 0.5732763409614563, "learning_rate": 0.00010949323624210586, "loss": 2.4295, "step": 20130 }, { "epoch": 6.987466161234579e-14, "grad_norm": 0.5221094489097595, "learning_rate": 0.00010947781054531948, "loss": 2.4491, "step": 20140 }, { "epoch": 6.990935608186533e-14, "grad_norm": 0.5613672137260437, "learning_rate": 0.00010946237840778044, "loss": 2.4527, "step": 20150 }, { "epoch": 6.994405055138486e-14, "grad_norm": 0.5592731833457947, "learning_rate": 0.0001094469398319292, "loss": 2.4861, "step": 20160 }, { "epoch": 6.99787450209044e-14, "grad_norm": 0.5756745934486389, "learning_rate": 0.00010943149482020726, "loss": 2.4369, "step": 20170 }, { "epoch": 7.001343949042393e-14, "grad_norm": 0.5417027473449707, "learning_rate": 0.0001094160433750571, "loss": 2.4663, "step": 20180 }, { "epoch": 7.004813395994347e-14, "grad_norm": 0.5763527750968933, "learning_rate": 0.00010940058549892223, "loss": 2.409, "step": 20190 }, { "epoch": 7.008282842946301e-14, "grad_norm": 0.5514048337936401, "learning_rate": 0.0001093851211942472, "loss": 2.4404, "step": 20200 }, { "epoch": 7.011752289898254e-14, "grad_norm": 0.5562760829925537, "learning_rate": 0.00010936965046347752, "loss": 2.4295, "step": 20210 }, { "epoch": 7.015221736850208e-14, "grad_norm": 0.5574753880500793, "learning_rate": 0.0001093541733090598, "loss": 2.4203, "step": 20220 }, { "epoch": 7.018691183802162e-14, "grad_norm": 0.5391328930854797, "learning_rate": 0.00010933868973344157, "loss": 2.4225, "step": 20230 }, { "epoch": 7.022160630754115e-14, "grad_norm": 0.5352974534034729, "learning_rate": 0.00010932319973907147, "loss": 2.4156, "step": 20240 }, { "epoch": 7.025630077706069e-14, "grad_norm": 0.6245367527008057, "learning_rate": 0.00010930770332839908, "loss": 2.4194, "step": 20250 }, { "epoch": 7.029099524658022e-14, "grad_norm": 0.554970383644104, "learning_rate": 0.00010929220050387501, "loss": 2.4576, "step": 20260 }, { "epoch": 7.032568971609976e-14, "grad_norm": 0.5391042828559875, "learning_rate": 0.00010927669126795096, "loss": 2.4262, "step": 20270 }, { "epoch": 7.03603841856193e-14, "grad_norm": 0.5368862748146057, "learning_rate": 0.00010926117562307951, "loss": 2.4335, "step": 20280 }, { "epoch": 7.039507865513883e-14, "grad_norm": 0.5406222939491272, "learning_rate": 0.00010924565357171438, "loss": 2.4538, "step": 20290 }, { "epoch": 7.042977312465837e-14, "grad_norm": 0.6371257305145264, "learning_rate": 0.00010923012511631023, "loss": 2.4176, "step": 20300 }, { "epoch": 7.04644675941779e-14, "grad_norm": 0.5191178917884827, "learning_rate": 0.00010921459025932272, "loss": 2.4415, "step": 20310 }, { "epoch": 7.049916206369744e-14, "grad_norm": 0.5780346989631653, "learning_rate": 0.00010919904900320859, "loss": 2.3983, "step": 20320 }, { "epoch": 7.053385653321698e-14, "grad_norm": 0.5772722363471985, "learning_rate": 0.00010918350135042557, "loss": 2.4053, "step": 20330 }, { "epoch": 7.056855100273651e-14, "grad_norm": 0.5628174543380737, "learning_rate": 0.00010916794730343233, "loss": 2.4504, "step": 20340 }, { "epoch": 7.060324547225605e-14, "grad_norm": 0.5600860714912415, "learning_rate": 0.00010915238686468867, "loss": 2.3989, "step": 20350 }, { "epoch": 7.063793994177558e-14, "grad_norm": 0.5484347939491272, "learning_rate": 0.00010913682003665531, "loss": 2.4362, "step": 20360 }, { "epoch": 7.067263441129512e-14, "grad_norm": 0.5184027552604675, "learning_rate": 0.00010912124682179402, "loss": 2.4393, "step": 20370 }, { "epoch": 7.070732888081466e-14, "grad_norm": 0.5499021410942078, "learning_rate": 0.00010910566722256755, "loss": 2.4542, "step": 20380 }, { "epoch": 7.074202335033419e-14, "grad_norm": 0.5516461730003357, "learning_rate": 0.0001090900812414397, "loss": 2.441, "step": 20390 }, { "epoch": 7.077671781985373e-14, "grad_norm": 0.5502073168754578, "learning_rate": 0.00010907448888087524, "loss": 2.4162, "step": 20400 }, { "epoch": 7.081141228937327e-14, "grad_norm": 0.5385432839393616, "learning_rate": 0.00010905889014334, "loss": 2.4319, "step": 20410 }, { "epoch": 7.08461067588928e-14, "grad_norm": 0.568942129611969, "learning_rate": 0.00010904328503130075, "loss": 2.4319, "step": 20420 }, { "epoch": 7.088080122841234e-14, "grad_norm": 0.5344502329826355, "learning_rate": 0.00010902767354722532, "loss": 2.4444, "step": 20430 }, { "epoch": 7.091549569793187e-14, "grad_norm": 0.5594977736473083, "learning_rate": 0.00010901205569358255, "loss": 2.4407, "step": 20440 }, { "epoch": 7.095019016745141e-14, "grad_norm": 0.5883808135986328, "learning_rate": 0.00010899643147284223, "loss": 2.4596, "step": 20450 }, { "epoch": 7.098488463697095e-14, "grad_norm": 0.5792665481567383, "learning_rate": 0.00010898080088747521, "loss": 2.4431, "step": 20460 }, { "epoch": 7.101957910649048e-14, "grad_norm": 0.573493480682373, "learning_rate": 0.00010896516393995335, "loss": 2.4238, "step": 20470 }, { "epoch": 7.105427357601002e-14, "grad_norm": 0.5614354610443115, "learning_rate": 0.0001089495206327495, "loss": 2.4182, "step": 20480 }, { "epoch": 7.108896804552955e-14, "grad_norm": 0.5580729246139526, "learning_rate": 0.00010893387096833749, "loss": 2.4666, "step": 20490 }, { "epoch": 7.112366251504909e-14, "grad_norm": 0.5755344033241272, "learning_rate": 0.00010891821494919218, "loss": 2.3891, "step": 20500 }, { "epoch": 7.115835698456863e-14, "grad_norm": 0.5764667987823486, "learning_rate": 0.00010890255257778946, "loss": 2.4407, "step": 20510 }, { "epoch": 7.119305145408816e-14, "grad_norm": 0.5575677156448364, "learning_rate": 0.00010888688385660618, "loss": 2.4351, "step": 20520 }, { "epoch": 7.12277459236077e-14, "grad_norm": 0.5678972601890564, "learning_rate": 0.00010887120878812022, "loss": 2.4696, "step": 20530 }, { "epoch": 7.126244039312724e-14, "grad_norm": 0.532089352607727, "learning_rate": 0.00010885552737481045, "loss": 2.4262, "step": 20540 }, { "epoch": 7.129713486264677e-14, "grad_norm": 0.5331153869628906, "learning_rate": 0.00010883983961915676, "loss": 2.432, "step": 20550 }, { "epoch": 7.133182933216631e-14, "grad_norm": 0.5499382615089417, "learning_rate": 0.00010882414552364003, "loss": 2.4267, "step": 20560 }, { "epoch": 7.136652380168584e-14, "grad_norm": 0.5561090707778931, "learning_rate": 0.00010880844509074215, "loss": 2.4304, "step": 20570 }, { "epoch": 7.140121827120538e-14, "grad_norm": 0.5624654293060303, "learning_rate": 0.00010879273832294599, "loss": 2.4358, "step": 20580 }, { "epoch": 7.143591274072492e-14, "grad_norm": 0.567119300365448, "learning_rate": 0.00010877702522273549, "loss": 2.413, "step": 20590 }, { "epoch": 7.147060721024445e-14, "grad_norm": 0.5649868249893188, "learning_rate": 0.00010876130579259548, "loss": 2.4871, "step": 20600 }, { "epoch": 7.150530167976399e-14, "grad_norm": 0.5730873346328735, "learning_rate": 0.00010874558003501187, "loss": 2.47, "step": 20610 }, { "epoch": 7.153999614928352e-14, "grad_norm": 0.5635114312171936, "learning_rate": 0.00010872984795247158, "loss": 2.419, "step": 20620 }, { "epoch": 7.157469061880306e-14, "grad_norm": 0.5462237596511841, "learning_rate": 0.00010871410954746247, "loss": 2.4163, "step": 20630 }, { "epoch": 7.16093850883226e-14, "grad_norm": 0.5256242752075195, "learning_rate": 0.00010869836482247345, "loss": 2.4566, "step": 20640 }, { "epoch": 7.164407955784213e-14, "grad_norm": 0.5466334223747253, "learning_rate": 0.00010868261377999443, "loss": 2.4205, "step": 20650 }, { "epoch": 7.167877402736167e-14, "grad_norm": 0.5399599671363831, "learning_rate": 0.00010866685642251627, "loss": 2.4532, "step": 20660 }, { "epoch": 7.17134684968812e-14, "grad_norm": 0.5552499294281006, "learning_rate": 0.00010865109275253088, "loss": 2.4496, "step": 20670 }, { "epoch": 7.174816296640074e-14, "grad_norm": 0.574782133102417, "learning_rate": 0.00010863532277253113, "loss": 2.4215, "step": 20680 }, { "epoch": 7.178285743592028e-14, "grad_norm": 0.5563462376594543, "learning_rate": 0.00010861954648501093, "loss": 2.4498, "step": 20690 }, { "epoch": 7.181755190543981e-14, "grad_norm": 0.5810015201568604, "learning_rate": 0.00010860376389246516, "loss": 2.4289, "step": 20700 }, { "epoch": 7.185224637495935e-14, "grad_norm": 0.5394448041915894, "learning_rate": 0.00010858797499738968, "loss": 2.466, "step": 20710 }, { "epoch": 7.188694084447889e-14, "grad_norm": 0.5740752816200256, "learning_rate": 0.00010857217980228138, "loss": 2.4632, "step": 20720 }, { "epoch": 7.192163531399842e-14, "grad_norm": 0.5515329241752625, "learning_rate": 0.00010855637830963813, "loss": 2.4376, "step": 20730 }, { "epoch": 7.195632978351796e-14, "grad_norm": 0.5889381170272827, "learning_rate": 0.00010854057052195882, "loss": 2.4576, "step": 20740 }, { "epoch": 7.19910242530375e-14, "grad_norm": 0.5686904191970825, "learning_rate": 0.0001085247564417433, "loss": 2.4205, "step": 20750 }, { "epoch": 7.202571872255703e-14, "grad_norm": 0.5450237393379211, "learning_rate": 0.00010850893607149244, "loss": 2.4378, "step": 20760 }, { "epoch": 7.206041319207657e-14, "grad_norm": 0.5487020015716553, "learning_rate": 0.00010849310941370808, "loss": 2.4415, "step": 20770 }, { "epoch": 7.20951076615961e-14, "grad_norm": 0.6151769757270813, "learning_rate": 0.00010847727647089307, "loss": 2.4515, "step": 20780 }, { "epoch": 7.212980213111564e-14, "grad_norm": 0.5449238419532776, "learning_rate": 0.00010846143724555125, "loss": 2.4177, "step": 20790 }, { "epoch": 7.216449660063518e-14, "grad_norm": 0.5700823068618774, "learning_rate": 0.00010844559174018749, "loss": 2.457, "step": 20800 }, { "epoch": 7.219919107015471e-14, "grad_norm": 0.573756754398346, "learning_rate": 0.0001084297399573076, "loss": 2.4424, "step": 20810 }, { "epoch": 7.223388553967425e-14, "grad_norm": 0.5646795034408569, "learning_rate": 0.0001084138818994184, "loss": 2.426, "step": 20820 }, { "epoch": 7.226858000919378e-14, "grad_norm": 0.5403907299041748, "learning_rate": 0.00010839801756902768, "loss": 2.4501, "step": 20830 }, { "epoch": 7.230327447871332e-14, "grad_norm": 0.5632953643798828, "learning_rate": 0.0001083821469686443, "loss": 2.448, "step": 20840 }, { "epoch": 7.233796894823286e-14, "grad_norm": 0.5549387335777283, "learning_rate": 0.00010836627010077805, "loss": 2.4473, "step": 20850 }, { "epoch": 7.237266341775239e-14, "grad_norm": 0.5712152123451233, "learning_rate": 0.00010835038696793966, "loss": 2.437, "step": 20860 }, { "epoch": 7.240735788727193e-14, "grad_norm": 0.6256892681121826, "learning_rate": 0.00010833449757264096, "loss": 2.4263, "step": 20870 }, { "epoch": 7.244205235679146e-14, "grad_norm": 0.5532128810882568, "learning_rate": 0.00010831860191739474, "loss": 2.4242, "step": 20880 }, { "epoch": 7.2476746826311e-14, "grad_norm": 0.5772107839584351, "learning_rate": 0.00010830270000471472, "loss": 2.4302, "step": 20890 }, { "epoch": 7.251144129583054e-14, "grad_norm": 0.5303003191947937, "learning_rate": 0.00010828679183711567, "loss": 2.4257, "step": 20900 }, { "epoch": 7.254613576535007e-14, "grad_norm": 0.5627037882804871, "learning_rate": 0.00010827087741711334, "loss": 2.4033, "step": 20910 }, { "epoch": 7.258083023486961e-14, "grad_norm": 0.5503320097923279, "learning_rate": 0.00010825495674722442, "loss": 2.4436, "step": 20920 }, { "epoch": 7.261552470438914e-14, "grad_norm": 0.5558784604072571, "learning_rate": 0.00010823902982996665, "loss": 2.4497, "step": 20930 }, { "epoch": 7.265021917390868e-14, "grad_norm": 0.5719415545463562, "learning_rate": 0.00010822309666785875, "loss": 2.4281, "step": 20940 }, { "epoch": 7.268491364342822e-14, "grad_norm": 0.5624500513076782, "learning_rate": 0.0001082071572634204, "loss": 2.4251, "step": 20950 }, { "epoch": 7.271960811294775e-14, "grad_norm": 0.5435826182365417, "learning_rate": 0.00010819121161917228, "loss": 2.4367, "step": 20960 }, { "epoch": 7.275430258246729e-14, "grad_norm": 0.5639973282814026, "learning_rate": 0.00010817525973763604, "loss": 2.4786, "step": 20970 }, { "epoch": 7.278899705198683e-14, "grad_norm": 0.5541608929634094, "learning_rate": 0.00010815930162133437, "loss": 2.4134, "step": 20980 }, { "epoch": 7.282369152150636e-14, "grad_norm": 0.5812622308731079, "learning_rate": 0.00010814333727279087, "loss": 2.407, "step": 20990 }, { "epoch": 7.28583859910259e-14, "grad_norm": 0.540570855140686, "learning_rate": 0.00010812736669453017, "loss": 2.4609, "step": 21000 }, { "epoch": 7.289308046054543e-14, "grad_norm": 0.5507962703704834, "learning_rate": 0.00010811138988907791, "loss": 2.4014, "step": 21010 }, { "epoch": 7.292777493006497e-14, "grad_norm": 0.5828272700309753, "learning_rate": 0.00010809540685896065, "loss": 2.3891, "step": 21020 }, { "epoch": 7.296246939958451e-14, "grad_norm": 0.5829244256019592, "learning_rate": 0.000108079417606706, "loss": 2.4194, "step": 21030 }, { "epoch": 7.299716386910404e-14, "grad_norm": 0.5354094505310059, "learning_rate": 0.00010806342213484249, "loss": 2.4192, "step": 21040 }, { "epoch": 7.303185833862358e-14, "grad_norm": 0.5853525400161743, "learning_rate": 0.0001080474204458997, "loss": 2.4284, "step": 21050 }, { "epoch": 7.306655280814311e-14, "grad_norm": 0.5578125715255737, "learning_rate": 0.00010803141254240812, "loss": 2.4325, "step": 21060 }, { "epoch": 7.310124727766265e-14, "grad_norm": 0.5802693367004395, "learning_rate": 0.00010801539842689929, "loss": 2.442, "step": 21070 }, { "epoch": 7.313594174718219e-14, "grad_norm": 0.545564591884613, "learning_rate": 0.0001079993781019057, "loss": 2.436, "step": 21080 }, { "epoch": 7.317063621670172e-14, "grad_norm": 0.5660610198974609, "learning_rate": 0.00010798335156996082, "loss": 2.4158, "step": 21090 }, { "epoch": 7.320533068622126e-14, "grad_norm": 0.5945662260055542, "learning_rate": 0.00010796731883359913, "loss": 2.4249, "step": 21100 }, { "epoch": 7.32400251557408e-14, "grad_norm": 0.5659404397010803, "learning_rate": 0.00010795127989535603, "loss": 2.4237, "step": 21110 }, { "epoch": 7.327471962526033e-14, "grad_norm": 0.5473706126213074, "learning_rate": 0.00010793523475776797, "loss": 2.4657, "step": 21120 }, { "epoch": 7.330941409477987e-14, "grad_norm": 0.5529412031173706, "learning_rate": 0.00010791918342337234, "loss": 2.4614, "step": 21130 }, { "epoch": 7.33441085642994e-14, "grad_norm": 0.5552354454994202, "learning_rate": 0.00010790312589470753, "loss": 2.4067, "step": 21140 }, { "epoch": 7.337880303381894e-14, "grad_norm": 0.6031678318977356, "learning_rate": 0.00010788706217431288, "loss": 2.4782, "step": 21150 }, { "epoch": 7.341349750333848e-14, "grad_norm": 0.5480990409851074, "learning_rate": 0.00010787099226472875, "loss": 2.4449, "step": 21160 }, { "epoch": 7.344819197285801e-14, "grad_norm": 0.5656017065048218, "learning_rate": 0.00010785491616849644, "loss": 2.4247, "step": 21170 }, { "epoch": 7.348288644237755e-14, "grad_norm": 0.6009307503700256, "learning_rate": 0.00010783883388815827, "loss": 2.4537, "step": 21180 }, { "epoch": 7.351758091189708e-14, "grad_norm": 0.5398609638214111, "learning_rate": 0.0001078227454262575, "loss": 2.4531, "step": 21190 }, { "epoch": 7.355227538141662e-14, "grad_norm": 0.5468766689300537, "learning_rate": 0.00010780665078533837, "loss": 2.4121, "step": 21200 }, { "epoch": 7.358696985093616e-14, "grad_norm": 0.5621956586837769, "learning_rate": 0.0001077905499679461, "loss": 2.4246, "step": 21210 }, { "epoch": 7.362166432045569e-14, "grad_norm": 0.582600474357605, "learning_rate": 0.00010777444297662695, "loss": 2.4383, "step": 21220 }, { "epoch": 7.365635878997523e-14, "grad_norm": 0.5576606392860413, "learning_rate": 0.00010775832981392805, "loss": 2.4481, "step": 21230 }, { "epoch": 7.369105325949477e-14, "grad_norm": 0.5254770517349243, "learning_rate": 0.00010774221048239758, "loss": 2.3976, "step": 21240 }, { "epoch": 7.37257477290143e-14, "grad_norm": 0.5561068654060364, "learning_rate": 0.00010772608498458468, "loss": 2.4125, "step": 21250 }, { "epoch": 7.376044219853384e-14, "grad_norm": 0.6335052847862244, "learning_rate": 0.00010770995332303943, "loss": 2.4555, "step": 21260 }, { "epoch": 7.379513666805337e-14, "grad_norm": 0.5665112733840942, "learning_rate": 0.00010769381550031295, "loss": 2.4161, "step": 21270 }, { "epoch": 7.382983113757291e-14, "grad_norm": 0.5567308068275452, "learning_rate": 0.00010767767151895727, "loss": 2.4352, "step": 21280 }, { "epoch": 7.386452560709245e-14, "grad_norm": 0.5421197414398193, "learning_rate": 0.00010766152138152542, "loss": 2.4543, "step": 21290 }, { "epoch": 7.389922007661198e-14, "grad_norm": 0.550743043422699, "learning_rate": 0.00010764536509057144, "loss": 2.427, "step": 21300 }, { "epoch": 7.393391454613152e-14, "grad_norm": 0.5535509586334229, "learning_rate": 0.00010762920264865027, "loss": 2.4267, "step": 21310 }, { "epoch": 7.396860901565105e-14, "grad_norm": 0.5652389526367188, "learning_rate": 0.00010761303405831788, "loss": 2.426, "step": 21320 }, { "epoch": 7.400330348517059e-14, "grad_norm": 0.5553126931190491, "learning_rate": 0.00010759685932213118, "loss": 2.4144, "step": 21330 }, { "epoch": 7.403799795469013e-14, "grad_norm": 3.5392518043518066, "learning_rate": 0.00010758067844264809, "loss": 2.4376, "step": 21340 }, { "epoch": 7.407269242420966e-14, "grad_norm": 0.5936217904090881, "learning_rate": 0.00010756449142242744, "loss": 2.4278, "step": 21350 }, { "epoch": 7.41073868937292e-14, "grad_norm": 0.5622357726097107, "learning_rate": 0.00010754829826402911, "loss": 2.462, "step": 21360 }, { "epoch": 7.414208136324874e-14, "grad_norm": 0.5386602878570557, "learning_rate": 0.00010753209897001388, "loss": 2.4623, "step": 21370 }, { "epoch": 7.417677583276827e-14, "grad_norm": 0.5503466129302979, "learning_rate": 0.00010751589354294352, "loss": 2.4312, "step": 21380 }, { "epoch": 7.421147030228781e-14, "grad_norm": 0.5840374231338501, "learning_rate": 0.00010749968198538082, "loss": 2.4175, "step": 21390 }, { "epoch": 7.424616477180734e-14, "grad_norm": 0.5653924942016602, "learning_rate": 0.00010748346429988946, "loss": 2.4262, "step": 21400 }, { "epoch": 7.428085924132688e-14, "grad_norm": 0.5475180149078369, "learning_rate": 0.00010746724048903416, "loss": 2.4146, "step": 21410 }, { "epoch": 7.431555371084642e-14, "grad_norm": 0.5964974164962769, "learning_rate": 0.00010745101055538053, "loss": 2.387, "step": 21420 }, { "epoch": 7.435024818036595e-14, "grad_norm": 0.5630491971969604, "learning_rate": 0.00010743477450149525, "loss": 2.4104, "step": 21430 }, { "epoch": 7.438494264988549e-14, "grad_norm": 0.5541214346885681, "learning_rate": 0.00010741853232994585, "loss": 2.4172, "step": 21440 }, { "epoch": 7.441963711940502e-14, "grad_norm": 0.5510262250900269, "learning_rate": 0.00010740228404330095, "loss": 2.4353, "step": 21450 }, { "epoch": 7.445433158892456e-14, "grad_norm": 0.568884015083313, "learning_rate": 0.00010738602964413004, "loss": 2.4229, "step": 21460 }, { "epoch": 7.44890260584441e-14, "grad_norm": 0.5538864731788635, "learning_rate": 0.0001073697691350036, "loss": 2.3997, "step": 21470 }, { "epoch": 7.452372052796363e-14, "grad_norm": 0.5608435869216919, "learning_rate": 0.00010735350251849314, "loss": 2.4313, "step": 21480 }, { "epoch": 7.455841499748317e-14, "grad_norm": 0.5811085104942322, "learning_rate": 0.00010733722979717106, "loss": 2.4181, "step": 21490 }, { "epoch": 7.45931094670027e-14, "grad_norm": 0.5421110987663269, "learning_rate": 0.00010732095097361075, "loss": 2.434, "step": 21500 }, { "epoch": 7.462780393652224e-14, "grad_norm": 0.5452114343643188, "learning_rate": 0.00010730466605038656, "loss": 2.4041, "step": 21510 }, { "epoch": 7.466249840604178e-14, "grad_norm": 0.5836210250854492, "learning_rate": 0.0001072883750300738, "loss": 2.4076, "step": 21520 }, { "epoch": 7.469719287556131e-14, "grad_norm": 0.5845527648925781, "learning_rate": 0.00010727207791524877, "loss": 2.401, "step": 21530 }, { "epoch": 7.473188734508085e-14, "grad_norm": 0.5357938408851624, "learning_rate": 0.00010725577470848874, "loss": 2.4517, "step": 21540 }, { "epoch": 7.476658181460039e-14, "grad_norm": 0.581990122795105, "learning_rate": 0.0001072394654123719, "loss": 2.4158, "step": 21550 }, { "epoch": 7.480127628411992e-14, "grad_norm": 0.5174565315246582, "learning_rate": 0.0001072231500294774, "loss": 2.4234, "step": 21560 }, { "epoch": 7.483597075363946e-14, "grad_norm": 0.5839492082595825, "learning_rate": 0.00010720682856238541, "loss": 2.4364, "step": 21570 }, { "epoch": 7.4870665223159e-14, "grad_norm": 0.5509362816810608, "learning_rate": 0.00010719050101367703, "loss": 2.4324, "step": 21580 }, { "epoch": 7.490535969267853e-14, "grad_norm": 0.5403429269790649, "learning_rate": 0.00010717416738593431, "loss": 2.3703, "step": 21590 }, { "epoch": 7.494005416219807e-14, "grad_norm": 0.549527645111084, "learning_rate": 0.00010715782768174029, "loss": 2.4049, "step": 21600 }, { "epoch": 7.49747486317176e-14, "grad_norm": 0.540643036365509, "learning_rate": 0.00010714148190367892, "loss": 2.4315, "step": 21610 }, { "epoch": 7.500944310123714e-14, "grad_norm": 0.5272877216339111, "learning_rate": 0.00010712513005433517, "loss": 2.4312, "step": 21620 }, { "epoch": 7.504413757075667e-14, "grad_norm": 0.5796213150024414, "learning_rate": 0.00010710877213629493, "loss": 2.42, "step": 21630 }, { "epoch": 7.507883204027621e-14, "grad_norm": 0.560926616191864, "learning_rate": 0.00010709240815214508, "loss": 2.4501, "step": 21640 }, { "epoch": 7.511352650979575e-14, "grad_norm": 0.5490780472755432, "learning_rate": 0.00010707603810447342, "loss": 2.3764, "step": 21650 }, { "epoch": 7.514822097931528e-14, "grad_norm": 0.5347467064857483, "learning_rate": 0.00010705966199586875, "loss": 2.4086, "step": 21660 }, { "epoch": 7.518291544883482e-14, "grad_norm": 0.5576143860816956, "learning_rate": 0.00010704327982892083, "loss": 2.4118, "step": 21670 }, { "epoch": 7.521760991835436e-14, "grad_norm": 0.5485594272613525, "learning_rate": 0.00010702689160622031, "loss": 2.4302, "step": 21680 }, { "epoch": 7.525230438787389e-14, "grad_norm": 0.589754581451416, "learning_rate": 0.00010701049733035888, "loss": 2.3935, "step": 21690 }, { "epoch": 7.528699885739343e-14, "grad_norm": 0.559894323348999, "learning_rate": 0.00010699409700392915, "loss": 2.4238, "step": 21700 }, { "epoch": 7.532169332691296e-14, "grad_norm": 0.5246466994285583, "learning_rate": 0.00010697769062952468, "loss": 2.3951, "step": 21710 }, { "epoch": 7.53563877964325e-14, "grad_norm": 0.5784648060798645, "learning_rate": 0.00010696127820974002, "loss": 2.4041, "step": 21720 }, { "epoch": 7.539108226595204e-14, "grad_norm": 0.5885919332504272, "learning_rate": 0.00010694485974717062, "loss": 2.4666, "step": 21730 }, { "epoch": 7.542577673547157e-14, "grad_norm": 0.5705073475837708, "learning_rate": 0.00010692843524441293, "loss": 2.4375, "step": 21740 }, { "epoch": 7.546047120499111e-14, "grad_norm": 0.6160881519317627, "learning_rate": 0.00010691200470406435, "loss": 2.4627, "step": 21750 }, { "epoch": 7.549516567451064e-14, "grad_norm": 0.5927947163581848, "learning_rate": 0.00010689556812872324, "loss": 2.4151, "step": 21760 }, { "epoch": 7.552986014403018e-14, "grad_norm": 0.5427994132041931, "learning_rate": 0.00010687912552098888, "loss": 2.4235, "step": 21770 }, { "epoch": 7.556455461354972e-14, "grad_norm": 0.547138512134552, "learning_rate": 0.00010686267688346155, "loss": 2.4236, "step": 21780 }, { "epoch": 7.559924908306925e-14, "grad_norm": 0.589713454246521, "learning_rate": 0.00010684622221874243, "loss": 2.4412, "step": 21790 }, { "epoch": 7.563394355258879e-14, "grad_norm": 0.583284854888916, "learning_rate": 0.0001068297615294337, "loss": 2.3823, "step": 21800 }, { "epoch": 7.566863802210833e-14, "grad_norm": 0.5507752299308777, "learning_rate": 0.0001068132948181385, "loss": 2.4283, "step": 21810 }, { "epoch": 7.570333249162786e-14, "grad_norm": 0.563817024230957, "learning_rate": 0.00010679682208746086, "loss": 2.4357, "step": 21820 }, { "epoch": 7.57380269611474e-14, "grad_norm": 0.5399883389472961, "learning_rate": 0.00010678034334000582, "loss": 2.3751, "step": 21830 }, { "epoch": 7.577272143066693e-14, "grad_norm": 0.5662234425544739, "learning_rate": 0.00010676385857837933, "loss": 2.4228, "step": 21840 }, { "epoch": 7.580741590018647e-14, "grad_norm": 0.5556972622871399, "learning_rate": 0.00010674736780518832, "loss": 2.3904, "step": 21850 }, { "epoch": 7.5842110369706e-14, "grad_norm": 0.5507402420043945, "learning_rate": 0.00010673087102304072, "loss": 2.4151, "step": 21860 }, { "epoch": 7.587680483922554e-14, "grad_norm": 0.5560979247093201, "learning_rate": 0.0001067143682345453, "loss": 2.432, "step": 21870 }, { "epoch": 7.591149930874508e-14, "grad_norm": 0.5664848685264587, "learning_rate": 0.00010669785944231183, "loss": 2.4317, "step": 21880 }, { "epoch": 7.594619377826461e-14, "grad_norm": 0.5852212309837341, "learning_rate": 0.00010668134464895105, "loss": 2.4379, "step": 21890 }, { "epoch": 7.598088824778415e-14, "grad_norm": 0.5939696431159973, "learning_rate": 0.00010666482385707461, "loss": 2.3975, "step": 21900 }, { "epoch": 7.601558271730369e-14, "grad_norm": 0.5809732675552368, "learning_rate": 0.00010664829706929519, "loss": 2.4241, "step": 21910 }, { "epoch": 7.605027718682322e-14, "grad_norm": 0.561225414276123, "learning_rate": 0.0001066317642882263, "loss": 2.3896, "step": 21920 }, { "epoch": 7.608497165634276e-14, "grad_norm": 0.5618250966072083, "learning_rate": 0.00010661522551648248, "loss": 2.4054, "step": 21930 }, { "epoch": 7.61196661258623e-14, "grad_norm": 0.552345335483551, "learning_rate": 0.00010659868075667918, "loss": 2.4418, "step": 21940 }, { "epoch": 7.615436059538183e-14, "grad_norm": 0.5853887796401978, "learning_rate": 0.00010658213001143282, "loss": 2.4535, "step": 21950 }, { "epoch": 7.618905506490137e-14, "grad_norm": 0.5877192616462708, "learning_rate": 0.00010656557328336076, "loss": 2.4428, "step": 21960 }, { "epoch": 7.62237495344209e-14, "grad_norm": 0.5775336027145386, "learning_rate": 0.00010654901057508132, "loss": 2.3864, "step": 21970 }, { "epoch": 7.625844400394044e-14, "grad_norm": 0.564199686050415, "learning_rate": 0.00010653244188921373, "loss": 2.4127, "step": 21980 }, { "epoch": 7.629313847345998e-14, "grad_norm": 0.5575755834579468, "learning_rate": 0.0001065158672283782, "loss": 2.4067, "step": 21990 }, { "epoch": 7.632783294297951e-14, "grad_norm": 0.5890095233917236, "learning_rate": 0.00010649928659519583, "loss": 2.4173, "step": 22000 }, { "epoch": 7.636252741249905e-14, "grad_norm": 0.5618470311164856, "learning_rate": 0.00010648269999228874, "loss": 2.4283, "step": 22010 }, { "epoch": 7.639722188201858e-14, "grad_norm": 0.5851930379867554, "learning_rate": 0.00010646610742227995, "loss": 2.4665, "step": 22020 }, { "epoch": 7.643191635153812e-14, "grad_norm": 0.5690003037452698, "learning_rate": 0.00010644950888779345, "loss": 2.4181, "step": 22030 }, { "epoch": 7.646661082105766e-14, "grad_norm": 0.5726107954978943, "learning_rate": 0.00010643290439145413, "loss": 2.3878, "step": 22040 }, { "epoch": 7.650130529057719e-14, "grad_norm": 0.5613391399383545, "learning_rate": 0.00010641629393588786, "loss": 2.4134, "step": 22050 }, { "epoch": 7.653599976009673e-14, "grad_norm": 0.5503520965576172, "learning_rate": 0.00010639967752372142, "loss": 2.3929, "step": 22060 }, { "epoch": 7.657069422961627e-14, "grad_norm": 0.5408121347427368, "learning_rate": 0.00010638305515758259, "loss": 2.4233, "step": 22070 }, { "epoch": 7.66053886991358e-14, "grad_norm": 0.5676050186157227, "learning_rate": 0.00010636642684010001, "loss": 2.407, "step": 22080 }, { "epoch": 7.664008316865534e-14, "grad_norm": 0.5416102409362793, "learning_rate": 0.00010634979257390335, "loss": 2.4211, "step": 22090 }, { "epoch": 7.667477763817487e-14, "grad_norm": 0.5392609238624573, "learning_rate": 0.00010633315236162318, "loss": 2.4313, "step": 22100 }, { "epoch": 7.670947210769441e-14, "grad_norm": 0.5926995277404785, "learning_rate": 0.00010631650620589094, "loss": 2.3928, "step": 22110 }, { "epoch": 7.674416657721395e-14, "grad_norm": 0.5564664602279663, "learning_rate": 0.00010629985410933916, "loss": 2.3593, "step": 22120 }, { "epoch": 7.677886104673348e-14, "grad_norm": 0.5699015855789185, "learning_rate": 0.00010628319607460115, "loss": 2.4441, "step": 22130 }, { "epoch": 7.681355551625302e-14, "grad_norm": 0.5868661999702454, "learning_rate": 0.00010626653210431127, "loss": 2.4195, "step": 22140 }, { "epoch": 7.684824998577255e-14, "grad_norm": 0.5359159708023071, "learning_rate": 0.00010624986220110479, "loss": 2.4239, "step": 22150 }, { "epoch": 7.688294445529209e-14, "grad_norm": 0.5825529098510742, "learning_rate": 0.0001062331863676179, "loss": 2.4662, "step": 22160 }, { "epoch": 7.691763892481163e-14, "grad_norm": 0.5990020632743835, "learning_rate": 0.00010621650460648774, "loss": 2.4426, "step": 22170 }, { "epoch": 7.695233339433116e-14, "grad_norm": 0.5650686621665955, "learning_rate": 0.0001061998169203524, "loss": 2.3835, "step": 22180 }, { "epoch": 7.69870278638507e-14, "grad_norm": 0.5319480895996094, "learning_rate": 0.00010618312331185088, "loss": 2.3911, "step": 22190 }, { "epoch": 7.702172233337023e-14, "grad_norm": 0.5857933163642883, "learning_rate": 0.00010616642378362314, "loss": 2.4221, "step": 22200 }, { "epoch": 7.705641680288977e-14, "grad_norm": 0.5672991871833801, "learning_rate": 0.00010614971833831006, "loss": 2.4207, "step": 22210 }, { "epoch": 7.709111127240931e-14, "grad_norm": 0.6259801983833313, "learning_rate": 0.00010613300697855346, "loss": 2.4234, "step": 22220 }, { "epoch": 7.712580574192884e-14, "grad_norm": 0.5575423240661621, "learning_rate": 0.0001061162897069961, "loss": 2.4253, "step": 22230 }, { "epoch": 7.716050021144838e-14, "grad_norm": 0.5459088087081909, "learning_rate": 0.00010609956652628169, "loss": 2.4087, "step": 22240 }, { "epoch": 7.719519468096792e-14, "grad_norm": 0.5627111196517944, "learning_rate": 0.00010608283743905482, "loss": 2.3725, "step": 22250 }, { "epoch": 7.722988915048745e-14, "grad_norm": 0.5505663752555847, "learning_rate": 0.00010606610244796108, "loss": 2.3977, "step": 22260 }, { "epoch": 7.726458362000699e-14, "grad_norm": 0.5851065516471863, "learning_rate": 0.00010604936155564694, "loss": 2.4267, "step": 22270 }, { "epoch": 7.729927808952652e-14, "grad_norm": 0.5614489316940308, "learning_rate": 0.0001060326147647599, "loss": 2.4174, "step": 22280 }, { "epoch": 7.733397255904606e-14, "grad_norm": 0.5653488636016846, "learning_rate": 0.0001060158620779482, "loss": 2.3915, "step": 22290 }, { "epoch": 7.73686670285656e-14, "grad_norm": 0.5696859955787659, "learning_rate": 0.00010599910349786123, "loss": 2.3887, "step": 22300 }, { "epoch": 7.740336149808513e-14, "grad_norm": 0.5636772513389587, "learning_rate": 0.0001059823390271492, "loss": 2.429, "step": 22310 }, { "epoch": 7.743805596760467e-14, "grad_norm": 0.5941615104675293, "learning_rate": 0.00010596556866846321, "loss": 2.4454, "step": 22320 }, { "epoch": 7.74727504371242e-14, "grad_norm": 0.5533768534660339, "learning_rate": 0.00010594879242445542, "loss": 2.4007, "step": 22330 }, { "epoch": 7.750744490664374e-14, "grad_norm": 0.5750634670257568, "learning_rate": 0.00010593201029777881, "loss": 2.4022, "step": 22340 }, { "epoch": 7.754213937616328e-14, "grad_norm": 0.5896473526954651, "learning_rate": 0.00010591522229108734, "loss": 2.4175, "step": 22350 }, { "epoch": 7.757683384568281e-14, "grad_norm": 0.618707001209259, "learning_rate": 0.00010589842840703589, "loss": 2.4209, "step": 22360 }, { "epoch": 7.761152831520235e-14, "grad_norm": 0.5642974376678467, "learning_rate": 0.00010588162864828024, "loss": 2.3813, "step": 22370 }, { "epoch": 7.764622278472189e-14, "grad_norm": 0.5348756909370422, "learning_rate": 0.00010586482301747716, "loss": 2.4174, "step": 22380 }, { "epoch": 7.768091725424142e-14, "grad_norm": 0.525809109210968, "learning_rate": 0.00010584801151728431, "loss": 2.3819, "step": 22390 }, { "epoch": 7.771561172376096e-14, "grad_norm": 0.5709630846977234, "learning_rate": 0.00010583119415036028, "loss": 2.4027, "step": 22400 }, { "epoch": 7.77503061932805e-14, "grad_norm": 0.5519557595252991, "learning_rate": 0.0001058143709193646, "loss": 2.4348, "step": 22410 }, { "epoch": 7.778500066280003e-14, "grad_norm": 0.5858150124549866, "learning_rate": 0.00010579754182695771, "loss": 2.4255, "step": 22420 }, { "epoch": 7.781969513231957e-14, "grad_norm": 0.5620788931846619, "learning_rate": 0.00010578070687580098, "loss": 2.4332, "step": 22430 }, { "epoch": 7.78543896018391e-14, "grad_norm": 0.5795358419418335, "learning_rate": 0.00010576386606855671, "loss": 2.4362, "step": 22440 }, { "epoch": 7.788908407135864e-14, "grad_norm": 0.5455554723739624, "learning_rate": 0.00010574701940788817, "loss": 2.4106, "step": 22450 }, { "epoch": 7.792377854087817e-14, "grad_norm": 0.5405575037002563, "learning_rate": 0.00010573016689645947, "loss": 2.3621, "step": 22460 }, { "epoch": 7.795847301039771e-14, "grad_norm": 0.583018958568573, "learning_rate": 0.00010571330853693569, "loss": 2.3874, "step": 22470 }, { "epoch": 7.799316747991725e-14, "grad_norm": 0.5560513138771057, "learning_rate": 0.00010569644433198285, "loss": 2.4064, "step": 22480 }, { "epoch": 7.802786194943678e-14, "grad_norm": 0.5643375515937805, "learning_rate": 0.00010567957428426787, "loss": 2.4312, "step": 22490 }, { "epoch": 7.806255641895632e-14, "grad_norm": 0.5736744403839111, "learning_rate": 0.00010566269839645863, "loss": 2.399, "step": 22500 }, { "epoch": 7.809725088847586e-14, "grad_norm": 0.5631886124610901, "learning_rate": 0.00010564581667122387, "loss": 2.4162, "step": 22510 }, { "epoch": 7.813194535799539e-14, "grad_norm": 0.5617097616195679, "learning_rate": 0.00010562892911123332, "loss": 2.4548, "step": 22520 }, { "epoch": 7.816663982751493e-14, "grad_norm": 0.5816309452056885, "learning_rate": 0.00010561203571915758, "loss": 2.4393, "step": 22530 }, { "epoch": 7.820133429703446e-14, "grad_norm": 0.5810979604721069, "learning_rate": 0.0001055951364976682, "loss": 2.4045, "step": 22540 }, { "epoch": 7.8236028766554e-14, "grad_norm": 0.5839025378227234, "learning_rate": 0.00010557823144943764, "loss": 2.4035, "step": 22550 }, { "epoch": 7.827072323607354e-14, "grad_norm": 0.5589423775672913, "learning_rate": 0.0001055613205771393, "loss": 2.3954, "step": 22560 }, { "epoch": 7.830541770559307e-14, "grad_norm": 0.5849512815475464, "learning_rate": 0.0001055444038834475, "loss": 2.4206, "step": 22570 }, { "epoch": 7.834011217511261e-14, "grad_norm": 0.5565935373306274, "learning_rate": 0.00010552748137103743, "loss": 2.4193, "step": 22580 }, { "epoch": 7.837480664463214e-14, "grad_norm": 0.5685550570487976, "learning_rate": 0.00010551055304258526, "loss": 2.4016, "step": 22590 }, { "epoch": 7.840950111415168e-14, "grad_norm": 0.5611292123794556, "learning_rate": 0.00010549361890076808, "loss": 2.3848, "step": 22600 }, { "epoch": 7.844419558367122e-14, "grad_norm": 0.5980010628700256, "learning_rate": 0.00010547667894826388, "loss": 2.4357, "step": 22610 }, { "epoch": 7.847889005319075e-14, "grad_norm": 0.5726418495178223, "learning_rate": 0.00010545973318775153, "loss": 2.4304, "step": 22620 }, { "epoch": 7.851358452271029e-14, "grad_norm": 0.5392782092094421, "learning_rate": 0.00010544278162191087, "loss": 2.4253, "step": 22630 }, { "epoch": 7.854827899222983e-14, "grad_norm": 0.5640112161636353, "learning_rate": 0.00010542582425342265, "loss": 2.3866, "step": 22640 }, { "epoch": 7.858297346174936e-14, "grad_norm": 0.5423575043678284, "learning_rate": 0.00010540886108496855, "loss": 2.4032, "step": 22650 }, { "epoch": 7.86176679312689e-14, "grad_norm": 0.5870696902275085, "learning_rate": 0.00010539189211923112, "loss": 2.4172, "step": 22660 }, { "epoch": 7.865236240078843e-14, "grad_norm": 0.5571466684341431, "learning_rate": 0.00010537491735889387, "loss": 2.4577, "step": 22670 }, { "epoch": 7.868705687030797e-14, "grad_norm": 0.5407306551933289, "learning_rate": 0.0001053579368066412, "loss": 2.4154, "step": 22680 }, { "epoch": 7.87217513398275e-14, "grad_norm": 0.5995076894760132, "learning_rate": 0.00010534095046515846, "loss": 2.4367, "step": 22690 }, { "epoch": 7.875644580934704e-14, "grad_norm": 0.5267719030380249, "learning_rate": 0.00010532395833713187, "loss": 2.4412, "step": 22700 }, { "epoch": 7.879114027886658e-14, "grad_norm": 0.5496965050697327, "learning_rate": 0.00010530696042524859, "loss": 2.4267, "step": 22710 }, { "epoch": 7.882583474838611e-14, "grad_norm": 0.5798371434211731, "learning_rate": 0.00010528995673219672, "loss": 2.4207, "step": 22720 }, { "epoch": 7.886052921790565e-14, "grad_norm": 0.5824645757675171, "learning_rate": 0.00010527294726066523, "loss": 2.4068, "step": 22730 }, { "epoch": 7.889522368742519e-14, "grad_norm": 0.5484040379524231, "learning_rate": 0.00010525593201334402, "loss": 2.428, "step": 22740 }, { "epoch": 7.892991815694472e-14, "grad_norm": 0.5848569869995117, "learning_rate": 0.0001052389109929239, "loss": 2.4526, "step": 22750 }, { "epoch": 7.896461262646426e-14, "grad_norm": 0.5540977716445923, "learning_rate": 0.0001052218842020966, "loss": 2.4106, "step": 22760 }, { "epoch": 7.89993070959838e-14, "grad_norm": 0.5985759496688843, "learning_rate": 0.0001052048516435548, "loss": 2.4148, "step": 22770 }, { "epoch": 7.903400156550333e-14, "grad_norm": 0.5590077638626099, "learning_rate": 0.000105187813319992, "loss": 2.4475, "step": 22780 }, { "epoch": 7.906869603502287e-14, "grad_norm": 0.551064670085907, "learning_rate": 0.00010517076923410268, "loss": 2.4356, "step": 22790 }, { "epoch": 7.91033905045424e-14, "grad_norm": 0.5601693987846375, "learning_rate": 0.00010515371938858221, "loss": 2.3884, "step": 22800 }, { "epoch": 7.913808497406194e-14, "grad_norm": 0.5586779117584229, "learning_rate": 0.0001051366637861269, "loss": 2.3922, "step": 22810 }, { "epoch": 7.917277944358148e-14, "grad_norm": 0.5598430633544922, "learning_rate": 0.00010511960242943391, "loss": 2.4269, "step": 22820 }, { "epoch": 7.920747391310101e-14, "grad_norm": 0.5710048079490662, "learning_rate": 0.00010510253532120139, "loss": 2.4228, "step": 22830 }, { "epoch": 7.924216838262055e-14, "grad_norm": 0.5463669300079346, "learning_rate": 0.00010508546246412833, "loss": 2.4209, "step": 22840 }, { "epoch": 7.927686285214008e-14, "grad_norm": 0.5783661603927612, "learning_rate": 0.00010506838386091464, "loss": 2.403, "step": 22850 }, { "epoch": 7.931155732165962e-14, "grad_norm": 0.5629075765609741, "learning_rate": 0.00010505129951426118, "loss": 2.4017, "step": 22860 }, { "epoch": 7.934625179117916e-14, "grad_norm": 0.612630307674408, "learning_rate": 0.0001050342094268697, "loss": 2.4163, "step": 22870 }, { "epoch": 7.938094626069869e-14, "grad_norm": 0.5521957874298096, "learning_rate": 0.00010501711360144282, "loss": 2.4011, "step": 22880 }, { "epoch": 7.941564073021823e-14, "grad_norm": 0.583936333656311, "learning_rate": 0.00010500001204068413, "loss": 2.4008, "step": 22890 }, { "epoch": 7.945033519973776e-14, "grad_norm": 0.6069756150245667, "learning_rate": 0.00010498290474729806, "loss": 2.442, "step": 22900 }, { "epoch": 7.94850296692573e-14, "grad_norm": 0.5887277722358704, "learning_rate": 0.00010496579172399, "loss": 2.42, "step": 22910 }, { "epoch": 7.951972413877684e-14, "grad_norm": 0.5480983257293701, "learning_rate": 0.00010494867297346623, "loss": 2.4177, "step": 22920 }, { "epoch": 7.955441860829637e-14, "grad_norm": 0.5942115783691406, "learning_rate": 0.00010493154849843394, "loss": 2.422, "step": 22930 }, { "epoch": 7.958911307781591e-14, "grad_norm": 0.5642284750938416, "learning_rate": 0.0001049144183016012, "loss": 2.3865, "step": 22940 }, { "epoch": 7.962380754733545e-14, "grad_norm": 0.5751197934150696, "learning_rate": 0.00010489728238567701, "loss": 2.4089, "step": 22950 }, { "epoch": 7.965850201685498e-14, "grad_norm": 0.5938237309455872, "learning_rate": 0.00010488014075337126, "loss": 2.4161, "step": 22960 }, { "epoch": 7.969319648637452e-14, "grad_norm": 0.5826483964920044, "learning_rate": 0.00010486299340739478, "loss": 2.4196, "step": 22970 }, { "epoch": 7.972789095589405e-14, "grad_norm": 0.5470030307769775, "learning_rate": 0.00010484584035045925, "loss": 2.3949, "step": 22980 }, { "epoch": 7.976258542541359e-14, "grad_norm": 0.5774021148681641, "learning_rate": 0.00010482868158527729, "loss": 2.4363, "step": 22990 }, { "epoch": 7.979727989493313e-14, "grad_norm": 0.6131924390792847, "learning_rate": 0.00010481151711456238, "loss": 2.4081, "step": 23000 }, { "epoch": 7.983197436445266e-14, "grad_norm": 0.5839693546295166, "learning_rate": 0.00010479434694102898, "loss": 2.3895, "step": 23010 }, { "epoch": 7.98666688339722e-14, "grad_norm": 0.574853777885437, "learning_rate": 0.00010477717106739236, "loss": 2.3964, "step": 23020 }, { "epoch": 7.990136330349173e-14, "grad_norm": 1.1332004070281982, "learning_rate": 0.00010475998949636878, "loss": 2.4368, "step": 23030 }, { "epoch": 7.993605777301127e-14, "grad_norm": 0.5717352628707886, "learning_rate": 0.00010474280223067536, "loss": 2.4461, "step": 23040 }, { "epoch": 7.997075224253081e-14, "grad_norm": 0.5333948731422424, "learning_rate": 0.00010472560927303004, "loss": 2.373, "step": 23050 }, { "epoch": 8.000544671205034e-14, "grad_norm": 0.5803001523017883, "learning_rate": 0.00010470841062615184, "loss": 2.4464, "step": 23060 }, { "epoch": 8.004014118156988e-14, "grad_norm": 0.5816470384597778, "learning_rate": 0.0001046912062927605, "loss": 2.4158, "step": 23070 }, { "epoch": 8.007483565108942e-14, "grad_norm": 0.5522683262825012, "learning_rate": 0.00010467399627557679, "loss": 2.3894, "step": 23080 }, { "epoch": 8.010953012060895e-14, "grad_norm": 0.5676687955856323, "learning_rate": 0.0001046567805773223, "loss": 2.3919, "step": 23090 }, { "epoch": 8.014422459012849e-14, "grad_norm": 0.5815120935440063, "learning_rate": 0.00010463955920071953, "loss": 2.4115, "step": 23100 }, { "epoch": 8.017891905964802e-14, "grad_norm": 0.5737789869308472, "learning_rate": 0.00010462233214849194, "loss": 2.4223, "step": 23110 }, { "epoch": 8.021361352916756e-14, "grad_norm": 0.5788572430610657, "learning_rate": 0.0001046050994233638, "loss": 2.4125, "step": 23120 }, { "epoch": 8.02483079986871e-14, "grad_norm": 0.5915061831474304, "learning_rate": 0.00010458786102806033, "loss": 2.4395, "step": 23130 }, { "epoch": 8.028300246820663e-14, "grad_norm": 0.5722375512123108, "learning_rate": 0.00010457061696530763, "loss": 2.4192, "step": 23140 }, { "epoch": 8.031769693772617e-14, "grad_norm": 0.5616556406021118, "learning_rate": 0.00010455336723783271, "loss": 2.4054, "step": 23150 }, { "epoch": 8.03523914072457e-14, "grad_norm": 0.5696420073509216, "learning_rate": 0.00010453611184836345, "loss": 2.4522, "step": 23160 }, { "epoch": 8.038708587676524e-14, "grad_norm": 0.5347952842712402, "learning_rate": 0.00010451885079962867, "loss": 2.37, "step": 23170 }, { "epoch": 8.042178034628478e-14, "grad_norm": 0.5616056323051453, "learning_rate": 0.000104501584094358, "loss": 2.428, "step": 23180 }, { "epoch": 8.045647481580431e-14, "grad_norm": 0.5363370180130005, "learning_rate": 0.00010448431173528209, "loss": 2.388, "step": 23190 }, { "epoch": 8.049116928532385e-14, "grad_norm": 0.5648899674415588, "learning_rate": 0.00010446703372513236, "loss": 2.4142, "step": 23200 }, { "epoch": 8.052586375484339e-14, "grad_norm": 0.5652140378952026, "learning_rate": 0.0001044497500666412, "loss": 2.3855, "step": 23210 }, { "epoch": 8.056055822436292e-14, "grad_norm": 0.5987458825111389, "learning_rate": 0.0001044324607625419, "loss": 2.4344, "step": 23220 }, { "epoch": 8.059525269388246e-14, "grad_norm": 0.5907345414161682, "learning_rate": 0.00010441516581556856, "loss": 2.4188, "step": 23230 }, { "epoch": 8.0629947163402e-14, "grad_norm": 0.5354942679405212, "learning_rate": 0.00010439786522845626, "loss": 2.4289, "step": 23240 }, { "epoch": 8.066464163292153e-14, "grad_norm": 0.5604778528213501, "learning_rate": 0.00010438055900394094, "loss": 2.4156, "step": 23250 }, { "epoch": 8.069933610244107e-14, "grad_norm": 1.1294664144515991, "learning_rate": 0.00010436324714475944, "loss": 2.4233, "step": 23260 }, { "epoch": 8.07340305719606e-14, "grad_norm": 0.6244608163833618, "learning_rate": 0.00010434592965364945, "loss": 2.4509, "step": 23270 }, { "epoch": 8.076872504148014e-14, "grad_norm": 0.5516549348831177, "learning_rate": 0.00010432860653334957, "loss": 2.3743, "step": 23280 }, { "epoch": 8.080341951099967e-14, "grad_norm": 0.571266770362854, "learning_rate": 0.00010431127778659936, "loss": 2.4182, "step": 23290 }, { "epoch": 8.083811398051921e-14, "grad_norm": 0.5740324258804321, "learning_rate": 0.00010429394341613917, "loss": 2.4098, "step": 23300 }, { "epoch": 8.087280845003875e-14, "grad_norm": 0.5406126976013184, "learning_rate": 0.00010427660342471031, "loss": 2.4045, "step": 23310 }, { "epoch": 8.090750291955828e-14, "grad_norm": 0.5914402008056641, "learning_rate": 0.00010425925781505493, "loss": 2.3888, "step": 23320 }, { "epoch": 8.094219738907782e-14, "grad_norm": 0.5540059804916382, "learning_rate": 0.00010424190658991606, "loss": 2.4002, "step": 23330 }, { "epoch": 8.097689185859736e-14, "grad_norm": 0.5488845705986023, "learning_rate": 0.0001042245497520377, "loss": 2.4416, "step": 23340 }, { "epoch": 8.101158632811689e-14, "grad_norm": 0.5789303779602051, "learning_rate": 0.00010420718730416468, "loss": 2.3929, "step": 23350 }, { "epoch": 8.104628079763643e-14, "grad_norm": 0.573334276676178, "learning_rate": 0.00010418981924904268, "loss": 2.4121, "step": 23360 }, { "epoch": 8.108097526715596e-14, "grad_norm": 0.5734925270080566, "learning_rate": 0.00010417244558941835, "loss": 2.4356, "step": 23370 }, { "epoch": 8.11156697366755e-14, "grad_norm": 0.584602415561676, "learning_rate": 0.00010415506632803917, "loss": 2.4049, "step": 23380 }, { "epoch": 8.115036420619504e-14, "grad_norm": 0.5551133751869202, "learning_rate": 0.0001041376814676535, "loss": 2.3936, "step": 23390 }, { "epoch": 8.118505867571457e-14, "grad_norm": 0.5685889720916748, "learning_rate": 0.00010412029101101064, "loss": 2.3763, "step": 23400 }, { "epoch": 8.121975314523411e-14, "grad_norm": 0.563317060470581, "learning_rate": 0.00010410289496086072, "loss": 2.4268, "step": 23410 }, { "epoch": 8.125444761475364e-14, "grad_norm": 0.5623689889907837, "learning_rate": 0.0001040854933199548, "loss": 2.392, "step": 23420 }, { "epoch": 8.128914208427318e-14, "grad_norm": 0.5624549984931946, "learning_rate": 0.00010406808609104475, "loss": 2.4031, "step": 23430 }, { "epoch": 8.132383655379272e-14, "grad_norm": 0.5498312711715698, "learning_rate": 0.00010405067327688343, "loss": 2.3841, "step": 23440 }, { "epoch": 8.135853102331225e-14, "grad_norm": 0.551084041595459, "learning_rate": 0.0001040332548802245, "loss": 2.3995, "step": 23450 }, { "epoch": 8.139322549283179e-14, "grad_norm": 0.536048173904419, "learning_rate": 0.00010401583090382253, "loss": 2.4041, "step": 23460 }, { "epoch": 8.142791996235133e-14, "grad_norm": 0.5692119598388672, "learning_rate": 0.00010399840135043297, "loss": 2.4168, "step": 23470 }, { "epoch": 8.146261443187086e-14, "grad_norm": 0.5698443651199341, "learning_rate": 0.00010398096622281215, "loss": 2.4174, "step": 23480 }, { "epoch": 8.14973089013904e-14, "grad_norm": 0.582249641418457, "learning_rate": 0.0001039635255237173, "loss": 2.4407, "step": 23490 }, { "epoch": 8.153200337090993e-14, "grad_norm": 0.5521010160446167, "learning_rate": 0.00010394607925590651, "loss": 2.4033, "step": 23500 }, { "epoch": 8.156669784042947e-14, "grad_norm": 0.607707679271698, "learning_rate": 0.00010392862742213874, "loss": 2.3925, "step": 23510 }, { "epoch": 8.1601392309949e-14, "grad_norm": 0.5624300837516785, "learning_rate": 0.00010391117002517387, "loss": 2.392, "step": 23520 }, { "epoch": 8.163608677946854e-14, "grad_norm": 0.587111234664917, "learning_rate": 0.00010389370706777265, "loss": 2.3827, "step": 23530 }, { "epoch": 8.167078124898808e-14, "grad_norm": 0.5644682049751282, "learning_rate": 0.00010387623855269665, "loss": 2.4332, "step": 23540 }, { "epoch": 8.170547571850761e-14, "grad_norm": 0.5632292628288269, "learning_rate": 0.0001038587644827084, "loss": 2.4029, "step": 23550 }, { "epoch": 8.174017018802715e-14, "grad_norm": 0.5684829354286194, "learning_rate": 0.00010384128486057126, "loss": 2.4192, "step": 23560 }, { "epoch": 8.177486465754669e-14, "grad_norm": 0.5730515122413635, "learning_rate": 0.00010382379968904949, "loss": 2.3641, "step": 23570 }, { "epoch": 8.180955912706622e-14, "grad_norm": 0.5708633065223694, "learning_rate": 0.00010380630897090823, "loss": 2.4063, "step": 23580 }, { "epoch": 8.184425359658576e-14, "grad_norm": 0.5960639119148254, "learning_rate": 0.00010378881270891346, "loss": 2.4438, "step": 23590 }, { "epoch": 8.18789480661053e-14, "grad_norm": 0.5870510339736938, "learning_rate": 0.00010377131090583208, "loss": 2.4155, "step": 23600 }, { "epoch": 8.191364253562483e-14, "grad_norm": 0.543451726436615, "learning_rate": 0.00010375380356443184, "loss": 2.3741, "step": 23610 }, { "epoch": 8.194833700514437e-14, "grad_norm": 0.5462096929550171, "learning_rate": 0.00010373629068748141, "loss": 2.4128, "step": 23620 }, { "epoch": 8.19830314746639e-14, "grad_norm": 0.5592319369316101, "learning_rate": 0.00010371877227775024, "loss": 2.3862, "step": 23630 }, { "epoch": 8.201772594418344e-14, "grad_norm": 0.5507820248603821, "learning_rate": 0.00010370124833800876, "loss": 2.4009, "step": 23640 }, { "epoch": 8.205242041370298e-14, "grad_norm": 0.5404307842254639, "learning_rate": 0.00010368371887102822, "loss": 2.404, "step": 23650 }, { "epoch": 8.208711488322251e-14, "grad_norm": 0.5695003271102905, "learning_rate": 0.00010366618387958074, "loss": 2.3959, "step": 23660 }, { "epoch": 8.212180935274205e-14, "grad_norm": 0.5530596971511841, "learning_rate": 0.00010364864336643934, "loss": 2.4024, "step": 23670 }, { "epoch": 8.215650382226158e-14, "grad_norm": 0.5607258677482605, "learning_rate": 0.00010363109733437791, "loss": 2.4277, "step": 23680 }, { "epoch": 8.219119829178112e-14, "grad_norm": 0.643597424030304, "learning_rate": 0.0001036135457861712, "loss": 2.4012, "step": 23690 }, { "epoch": 8.222589276130066e-14, "grad_norm": 0.5466392636299133, "learning_rate": 0.0001035959887245948, "loss": 2.4099, "step": 23700 }, { "epoch": 8.226058723082019e-14, "grad_norm": 0.6259854435920715, "learning_rate": 0.00010357842615242527, "loss": 2.4081, "step": 23710 }, { "epoch": 8.229528170033973e-14, "grad_norm": 0.5660258531570435, "learning_rate": 0.00010356085807243996, "loss": 2.3921, "step": 23720 }, { "epoch": 8.232997616985926e-14, "grad_norm": 0.57262122631073, "learning_rate": 0.00010354328448741707, "loss": 2.4019, "step": 23730 }, { "epoch": 8.23646706393788e-14, "grad_norm": 0.5618054270744324, "learning_rate": 0.00010352570540013576, "loss": 2.4297, "step": 23740 }, { "epoch": 8.239936510889834e-14, "grad_norm": 0.5563138127326965, "learning_rate": 0.00010350812081337597, "loss": 2.3897, "step": 23750 }, { "epoch": 8.243405957841787e-14, "grad_norm": 0.5620490312576294, "learning_rate": 0.00010349053072991858, "loss": 2.4118, "step": 23760 }, { "epoch": 8.246875404793741e-14, "grad_norm": 0.5861133337020874, "learning_rate": 0.00010347293515254533, "loss": 2.3803, "step": 23770 }, { "epoch": 8.250344851745695e-14, "grad_norm": 0.5699045062065125, "learning_rate": 0.00010345533408403877, "loss": 2.4262, "step": 23780 }, { "epoch": 8.253814298697648e-14, "grad_norm": 0.5582321882247925, "learning_rate": 0.00010343772752718236, "loss": 2.4276, "step": 23790 }, { "epoch": 8.257283745649602e-14, "grad_norm": 0.561233639717102, "learning_rate": 0.00010342011548476045, "loss": 2.4046, "step": 23800 }, { "epoch": 8.260753192601555e-14, "grad_norm": 0.5731855034828186, "learning_rate": 0.00010340249795955821, "loss": 2.4012, "step": 23810 }, { "epoch": 8.264222639553509e-14, "grad_norm": 0.5810763835906982, "learning_rate": 0.0001033848749543617, "loss": 2.3552, "step": 23820 }, { "epoch": 8.267692086505463e-14, "grad_norm": 0.5919564962387085, "learning_rate": 0.00010336724647195786, "loss": 2.4055, "step": 23830 }, { "epoch": 8.271161533457416e-14, "grad_norm": 0.5938172340393066, "learning_rate": 0.00010334961251513448, "loss": 2.4238, "step": 23840 }, { "epoch": 8.27463098040937e-14, "grad_norm": 0.5628291368484497, "learning_rate": 0.00010333197308668023, "loss": 2.3954, "step": 23850 }, { "epoch": 8.278100427361323e-14, "grad_norm": 0.5736475586891174, "learning_rate": 0.0001033143281893846, "loss": 2.361, "step": 23860 }, { "epoch": 8.281569874313277e-14, "grad_norm": 0.5687358379364014, "learning_rate": 0.000103296677826038, "loss": 2.3928, "step": 23870 }, { "epoch": 8.285039321265231e-14, "grad_norm": 0.5356073379516602, "learning_rate": 0.0001032790219994317, "loss": 2.3976, "step": 23880 }, { "epoch": 8.288508768217184e-14, "grad_norm": 0.5726895332336426, "learning_rate": 0.00010326136071235778, "loss": 2.4337, "step": 23890 }, { "epoch": 8.291978215169138e-14, "grad_norm": 0.5835265517234802, "learning_rate": 0.00010324369396760925, "loss": 2.3961, "step": 23900 }, { "epoch": 8.295447662121092e-14, "grad_norm": 0.5793695449829102, "learning_rate": 0.00010322602176797993, "loss": 2.4144, "step": 23910 }, { "epoch": 8.298917109073045e-14, "grad_norm": 0.562233567237854, "learning_rate": 0.00010320834411626455, "loss": 2.3914, "step": 23920 }, { "epoch": 8.302386556024999e-14, "grad_norm": 0.5831064581871033, "learning_rate": 0.00010319066101525866, "loss": 2.4285, "step": 23930 }, { "epoch": 8.305856002976952e-14, "grad_norm": 0.5716841816902161, "learning_rate": 0.00010317297246775871, "loss": 2.3916, "step": 23940 }, { "epoch": 8.309325449928906e-14, "grad_norm": 0.5547429323196411, "learning_rate": 0.00010315527847656196, "loss": 2.3716, "step": 23950 }, { "epoch": 8.31279489688086e-14, "grad_norm": 0.5551949143409729, "learning_rate": 0.0001031375790444666, "loss": 2.3636, "step": 23960 }, { "epoch": 8.316264343832813e-14, "grad_norm": 0.5652115345001221, "learning_rate": 0.00010311987417427163, "loss": 2.3741, "step": 23970 }, { "epoch": 8.319733790784767e-14, "grad_norm": 0.554884672164917, "learning_rate": 0.00010310216386877688, "loss": 2.4012, "step": 23980 }, { "epoch": 8.32320323773672e-14, "grad_norm": 0.6124816536903381, "learning_rate": 0.00010308444813078316, "loss": 2.3842, "step": 23990 }, { "epoch": 8.326672684688674e-14, "grad_norm": 0.5621520280838013, "learning_rate": 0.000103066726963092, "loss": 2.3999, "step": 24000 }, { "epoch": 8.330142131640628e-14, "grad_norm": 0.5833355784416199, "learning_rate": 0.00010304900036850588, "loss": 2.3778, "step": 24010 }, { "epoch": 8.333611578592581e-14, "grad_norm": 0.5719831585884094, "learning_rate": 0.0001030312683498281, "loss": 2.4119, "step": 24020 }, { "epoch": 8.337081025544535e-14, "grad_norm": 0.5962221026420593, "learning_rate": 0.00010301353090986282, "loss": 2.4369, "step": 24030 }, { "epoch": 8.340550472496489e-14, "grad_norm": 0.5634356737136841, "learning_rate": 0.00010299578805141506, "loss": 2.3743, "step": 24040 }, { "epoch": 8.344019919448442e-14, "grad_norm": 0.5537898540496826, "learning_rate": 0.00010297803977729073, "loss": 2.4134, "step": 24050 }, { "epoch": 8.347489366400396e-14, "grad_norm": 0.5737374424934387, "learning_rate": 0.00010296028609029652, "loss": 2.4149, "step": 24060 }, { "epoch": 8.350958813352349e-14, "grad_norm": 0.5745027661323547, "learning_rate": 0.00010294252699324008, "loss": 2.3693, "step": 24070 }, { "epoch": 8.354428260304303e-14, "grad_norm": 0.5496186017990112, "learning_rate": 0.0001029247624889298, "loss": 2.3747, "step": 24080 }, { "epoch": 8.357897707256257e-14, "grad_norm": 0.5425242185592651, "learning_rate": 0.00010290699258017504, "loss": 2.4361, "step": 24090 }, { "epoch": 8.36136715420821e-14, "grad_norm": 0.5500790476799011, "learning_rate": 0.00010288921726978591, "loss": 2.397, "step": 24100 }, { "epoch": 8.364836601160164e-14, "grad_norm": 0.5583398342132568, "learning_rate": 0.00010287143656057344, "loss": 2.434, "step": 24110 }, { "epoch": 8.368306048112117e-14, "grad_norm": 0.5412720441818237, "learning_rate": 0.00010285365045534951, "loss": 2.3817, "step": 24120 }, { "epoch": 8.371775495064071e-14, "grad_norm": 0.5730198621749878, "learning_rate": 0.00010283585895692683, "loss": 2.3866, "step": 24130 }, { "epoch": 8.375244942016025e-14, "grad_norm": 0.5927741527557373, "learning_rate": 0.00010281806206811893, "loss": 2.4037, "step": 24140 }, { "epoch": 8.378714388967978e-14, "grad_norm": 0.5369418263435364, "learning_rate": 0.0001028002597917403, "loss": 2.4034, "step": 24150 }, { "epoch": 8.382183835919932e-14, "grad_norm": 0.5640385150909424, "learning_rate": 0.0001027824521306062, "loss": 2.3773, "step": 24160 }, { "epoch": 8.385653282871885e-14, "grad_norm": 0.5930258631706238, "learning_rate": 0.00010276463908753274, "loss": 2.3997, "step": 24170 }, { "epoch": 8.389122729823839e-14, "grad_norm": 0.5872297883033752, "learning_rate": 0.0001027468206653369, "loss": 2.4018, "step": 24180 }, { "epoch": 8.392592176775793e-14, "grad_norm": 0.5470549464225769, "learning_rate": 0.00010272899686683652, "loss": 2.3782, "step": 24190 }, { "epoch": 8.396061623727746e-14, "grad_norm": 0.5381404757499695, "learning_rate": 0.0001027111676948503, "loss": 2.4079, "step": 24200 }, { "epoch": 8.3995310706797e-14, "grad_norm": 0.6271170973777771, "learning_rate": 0.00010269333315219772, "loss": 2.4121, "step": 24210 }, { "epoch": 8.403000517631654e-14, "grad_norm": 0.5508876442909241, "learning_rate": 0.00010267549324169921, "loss": 2.4035, "step": 24220 }, { "epoch": 8.406469964583607e-14, "grad_norm": 0.6120626330375671, "learning_rate": 0.00010265764796617598, "loss": 2.4056, "step": 24230 }, { "epoch": 8.409939411535561e-14, "grad_norm": 0.5882008671760559, "learning_rate": 0.00010263979732845008, "loss": 2.3619, "step": 24240 }, { "epoch": 8.413408858487514e-14, "grad_norm": 0.5898140668869019, "learning_rate": 0.00010262194133134448, "loss": 2.3899, "step": 24250 }, { "epoch": 8.416878305439468e-14, "grad_norm": 0.5778892040252686, "learning_rate": 0.0001026040799776829, "loss": 2.4072, "step": 24260 }, { "epoch": 8.420347752391422e-14, "grad_norm": 0.580388069152832, "learning_rate": 0.00010258621327029003, "loss": 2.3959, "step": 24270 }, { "epoch": 8.423817199343375e-14, "grad_norm": 0.5329089760780334, "learning_rate": 0.00010256834121199128, "loss": 2.3962, "step": 24280 }, { "epoch": 8.427286646295329e-14, "grad_norm": 0.5560459494590759, "learning_rate": 0.00010255046380561297, "loss": 2.3856, "step": 24290 }, { "epoch": 8.430756093247282e-14, "grad_norm": 0.5478885173797607, "learning_rate": 0.00010253258105398226, "loss": 2.3994, "step": 24300 }, { "epoch": 8.434225540199236e-14, "grad_norm": 0.5516511797904968, "learning_rate": 0.00010251469295992715, "loss": 2.4048, "step": 24310 }, { "epoch": 8.43769498715119e-14, "grad_norm": 0.5447450280189514, "learning_rate": 0.00010249679952627654, "loss": 2.3992, "step": 24320 }, { "epoch": 8.441164434103143e-14, "grad_norm": 0.5715891122817993, "learning_rate": 0.00010247890075586005, "loss": 2.4033, "step": 24330 }, { "epoch": 8.444633881055097e-14, "grad_norm": 0.5725030899047852, "learning_rate": 0.00010246099665150824, "loss": 2.3699, "step": 24340 }, { "epoch": 8.44810332800705e-14, "grad_norm": 0.5882763266563416, "learning_rate": 0.00010244308721605252, "loss": 2.3648, "step": 24350 }, { "epoch": 8.451572774959004e-14, "grad_norm": 0.564552903175354, "learning_rate": 0.00010242517245232507, "loss": 2.3824, "step": 24360 }, { "epoch": 8.455042221910958e-14, "grad_norm": 0.5525565147399902, "learning_rate": 0.00010240725236315897, "loss": 2.357, "step": 24370 }, { "epoch": 8.458511668862911e-14, "grad_norm": 0.5407058596611023, "learning_rate": 0.00010238932695138813, "loss": 2.4329, "step": 24380 }, { "epoch": 8.461981115814865e-14, "grad_norm": 0.5667784214019775, "learning_rate": 0.00010237139621984732, "loss": 2.3996, "step": 24390 }, { "epoch": 8.465450562766819e-14, "grad_norm": 0.5763686299324036, "learning_rate": 0.0001023534601713721, "loss": 2.4018, "step": 24400 }, { "epoch": 8.468920009718772e-14, "grad_norm": 0.610221803188324, "learning_rate": 0.00010233551880879892, "loss": 2.405, "step": 24410 }, { "epoch": 8.472389456670726e-14, "grad_norm": 0.5778310894966125, "learning_rate": 0.00010231757213496506, "loss": 2.4018, "step": 24420 }, { "epoch": 8.47585890362268e-14, "grad_norm": 0.5739724040031433, "learning_rate": 0.00010229962015270861, "loss": 2.4215, "step": 24430 }, { "epoch": 8.479328350574633e-14, "grad_norm": 0.5686819553375244, "learning_rate": 0.00010228166286486852, "loss": 2.3691, "step": 24440 }, { "epoch": 8.482797797526587e-14, "grad_norm": 0.5771732330322266, "learning_rate": 0.00010226370027428458, "loss": 2.4146, "step": 24450 }, { "epoch": 8.48626724447854e-14, "grad_norm": 0.5816347002983093, "learning_rate": 0.00010224573238379746, "loss": 2.4139, "step": 24460 }, { "epoch": 8.489736691430494e-14, "grad_norm": 0.5224953293800354, "learning_rate": 0.00010222775919624859, "loss": 2.3818, "step": 24470 }, { "epoch": 8.493206138382448e-14, "grad_norm": 0.5733636021614075, "learning_rate": 0.00010220978071448027, "loss": 2.3749, "step": 24480 }, { "epoch": 8.496675585334401e-14, "grad_norm": 0.6106364130973816, "learning_rate": 0.00010219179694133565, "loss": 2.3968, "step": 24490 }, { "epoch": 8.500145032286355e-14, "grad_norm": 0.5879061818122864, "learning_rate": 0.00010217380787965872, "loss": 2.428, "step": 24500 }, { "epoch": 8.503614479238308e-14, "grad_norm": 0.555372953414917, "learning_rate": 0.00010215581353229428, "loss": 2.3903, "step": 24510 }, { "epoch": 8.507083926190262e-14, "grad_norm": 0.5746854543685913, "learning_rate": 0.000102137813902088, "loss": 2.385, "step": 24520 }, { "epoch": 8.510553373142216e-14, "grad_norm": 0.5938072204589844, "learning_rate": 0.00010211980899188636, "loss": 2.3942, "step": 24530 }, { "epoch": 8.514022820094169e-14, "grad_norm": 0.6026631593704224, "learning_rate": 0.00010210179880453665, "loss": 2.3775, "step": 24540 }, { "epoch": 8.517492267046123e-14, "grad_norm": 0.5644952058792114, "learning_rate": 0.00010208378334288709, "loss": 2.3822, "step": 24550 }, { "epoch": 8.520961713998076e-14, "grad_norm": 0.5786916017532349, "learning_rate": 0.00010206576260978662, "loss": 2.4133, "step": 24560 }, { "epoch": 8.52443116095003e-14, "grad_norm": 0.5570988059043884, "learning_rate": 0.00010204773660808507, "loss": 2.3827, "step": 24570 }, { "epoch": 8.527900607901984e-14, "grad_norm": 0.5604352951049805, "learning_rate": 0.0001020297053406331, "loss": 2.3865, "step": 24580 }, { "epoch": 8.531370054853937e-14, "grad_norm": 0.5527738928794861, "learning_rate": 0.00010201166881028222, "loss": 2.3959, "step": 24590 }, { "epoch": 8.534839501805891e-14, "grad_norm": 0.5626962184906006, "learning_rate": 0.00010199362701988474, "loss": 2.3768, "step": 24600 }, { "epoch": 8.538308948757845e-14, "grad_norm": 0.5977851748466492, "learning_rate": 0.00010197557997229378, "loss": 2.4124, "step": 24610 }, { "epoch": 8.541778395709798e-14, "grad_norm": 0.5619433522224426, "learning_rate": 0.00010195752767036338, "loss": 2.3905, "step": 24620 }, { "epoch": 8.545247842661752e-14, "grad_norm": 0.5996397137641907, "learning_rate": 0.00010193947011694831, "loss": 2.3826, "step": 24630 }, { "epoch": 8.548717289613705e-14, "grad_norm": 0.5996801853179932, "learning_rate": 0.00010192140731490426, "loss": 2.3989, "step": 24640 }, { "epoch": 8.552186736565659e-14, "grad_norm": 0.5616782903671265, "learning_rate": 0.0001019033392670877, "loss": 2.4308, "step": 24650 }, { "epoch": 8.555656183517613e-14, "grad_norm": 0.5750167965888977, "learning_rate": 0.0001018852659763559, "loss": 2.3732, "step": 24660 }, { "epoch": 8.559125630469566e-14, "grad_norm": 0.5604051947593689, "learning_rate": 0.000101867187445567, "loss": 2.3981, "step": 24670 }, { "epoch": 8.56259507742152e-14, "grad_norm": 0.5602191686630249, "learning_rate": 0.00010184910367758, "loss": 2.3933, "step": 24680 }, { "epoch": 8.566064524373473e-14, "grad_norm": 0.5581074953079224, "learning_rate": 0.00010183101467525466, "loss": 2.4053, "step": 24690 }, { "epoch": 8.569533971325427e-14, "grad_norm": 0.570049524307251, "learning_rate": 0.00010181292044145162, "loss": 2.3335, "step": 24700 }, { "epoch": 8.573003418277381e-14, "grad_norm": 0.5644198656082153, "learning_rate": 0.00010179482097903232, "loss": 2.3997, "step": 24710 }, { "epoch": 8.576472865229334e-14, "grad_norm": 0.6072389483451843, "learning_rate": 0.00010177671629085902, "loss": 2.377, "step": 24720 }, { "epoch": 8.579942312181288e-14, "grad_norm": 0.5512850284576416, "learning_rate": 0.00010175860637979486, "loss": 2.3945, "step": 24730 }, { "epoch": 8.583411759133242e-14, "grad_norm": 0.5728508830070496, "learning_rate": 0.00010174049124870371, "loss": 2.4099, "step": 24740 }, { "epoch": 8.586881206085195e-14, "grad_norm": 0.5428751111030579, "learning_rate": 0.0001017223709004504, "loss": 2.4315, "step": 24750 }, { "epoch": 8.590350653037149e-14, "grad_norm": 0.5815463662147522, "learning_rate": 0.00010170424533790041, "loss": 2.3853, "step": 24760 }, { "epoch": 8.593820099989102e-14, "grad_norm": 0.5421085953712463, "learning_rate": 0.00010168611456392023, "loss": 2.3557, "step": 24770 }, { "epoch": 8.597289546941056e-14, "grad_norm": 0.5787822008132935, "learning_rate": 0.00010166797858137702, "loss": 2.3702, "step": 24780 }, { "epoch": 8.60075899389301e-14, "grad_norm": 0.5658475756645203, "learning_rate": 0.00010164983739313887, "loss": 2.3898, "step": 24790 }, { "epoch": 8.604228440844963e-14, "grad_norm": 0.554672122001648, "learning_rate": 0.00010163169100207464, "loss": 2.4007, "step": 24800 }, { "epoch": 8.607697887796917e-14, "grad_norm": 0.6038801670074463, "learning_rate": 0.00010161353941105405, "loss": 2.405, "step": 24810 }, { "epoch": 8.61116733474887e-14, "grad_norm": 0.5495851039886475, "learning_rate": 0.00010159538262294756, "loss": 2.3931, "step": 24820 }, { "epoch": 8.614636781700824e-14, "grad_norm": 0.5886641144752502, "learning_rate": 0.00010157722064062652, "loss": 2.3943, "step": 24830 }, { "epoch": 8.618106228652778e-14, "grad_norm": 0.5838031768798828, "learning_rate": 0.00010155905346696316, "loss": 2.3971, "step": 24840 }, { "epoch": 8.621575675604731e-14, "grad_norm": 0.6451982855796814, "learning_rate": 0.00010154088110483041, "loss": 2.3992, "step": 24850 }, { "epoch": 8.625045122556685e-14, "grad_norm": 0.5684218406677246, "learning_rate": 0.00010152270355710207, "loss": 2.4112, "step": 24860 }, { "epoch": 8.628514569508638e-14, "grad_norm": 0.5534093379974365, "learning_rate": 0.00010150452082665276, "loss": 2.3806, "step": 24870 }, { "epoch": 8.631984016460592e-14, "grad_norm": 0.5858080387115479, "learning_rate": 0.00010148633291635793, "loss": 2.3524, "step": 24880 }, { "epoch": 8.635453463412546e-14, "grad_norm": 0.5635209679603577, "learning_rate": 0.00010146813982909387, "loss": 2.4027, "step": 24890 }, { "epoch": 8.638922910364499e-14, "grad_norm": 0.5859189033508301, "learning_rate": 0.00010144994156773763, "loss": 2.402, "step": 24900 }, { "epoch": 8.642392357316453e-14, "grad_norm": 0.5513394474983215, "learning_rate": 0.00010143173813516712, "loss": 2.4044, "step": 24910 }, { "epoch": 8.645861804268407e-14, "grad_norm": 0.599636435508728, "learning_rate": 0.00010141352953426104, "loss": 2.3225, "step": 24920 }, { "epoch": 8.64933125122036e-14, "grad_norm": 0.5991833806037903, "learning_rate": 0.00010139531576789894, "loss": 2.3708, "step": 24930 }, { "epoch": 8.652800698172314e-14, "grad_norm": 0.5812159776687622, "learning_rate": 0.00010137709683896117, "loss": 2.3785, "step": 24940 }, { "epoch": 8.656270145124267e-14, "grad_norm": 0.6139986515045166, "learning_rate": 0.00010135887275032888, "loss": 2.3862, "step": 24950 }, { "epoch": 8.659739592076221e-14, "grad_norm": 0.5794146656990051, "learning_rate": 0.00010134064350488407, "loss": 2.3911, "step": 24960 }, { "epoch": 8.663209039028175e-14, "grad_norm": 0.5727996826171875, "learning_rate": 0.00010132240910550951, "loss": 2.4248, "step": 24970 }, { "epoch": 8.666678485980128e-14, "grad_norm": 0.5811457633972168, "learning_rate": 0.00010130416955508886, "loss": 2.4078, "step": 24980 }, { "epoch": 8.670147932932082e-14, "grad_norm": 0.5865065455436707, "learning_rate": 0.00010128592485650651, "loss": 2.3955, "step": 24990 }, { "epoch": 8.673617379884035e-14, "grad_norm": 0.5513076782226562, "learning_rate": 0.00010126767501264772, "loss": 2.4437, "step": 25000 }, { "epoch": 8.677086826835989e-14, "grad_norm": 0.5856971740722656, "learning_rate": 0.00010124942002639854, "loss": 2.3815, "step": 25010 }, { "epoch": 8.680556273787943e-14, "grad_norm": 0.5708604454994202, "learning_rate": 0.00010123115990064586, "loss": 2.3858, "step": 25020 }, { "epoch": 8.684025720739896e-14, "grad_norm": 0.5713416337966919, "learning_rate": 0.00010121289463827733, "loss": 2.4082, "step": 25030 }, { "epoch": 8.68749516769185e-14, "grad_norm": 0.5514233708381653, "learning_rate": 0.00010119462424218142, "loss": 2.3724, "step": 25040 }, { "epoch": 8.690964614643804e-14, "grad_norm": 0.5739091038703918, "learning_rate": 0.0001011763487152475, "loss": 2.3872, "step": 25050 }, { "epoch": 8.694434061595757e-14, "grad_norm": 0.5708903670310974, "learning_rate": 0.00010115806806036566, "loss": 2.4125, "step": 25060 }, { "epoch": 8.697903508547711e-14, "grad_norm": 0.5702176094055176, "learning_rate": 0.00010113978228042683, "loss": 2.3719, "step": 25070 }, { "epoch": 8.701372955499664e-14, "grad_norm": 0.5559750199317932, "learning_rate": 0.00010112149137832274, "loss": 2.4053, "step": 25080 }, { "epoch": 8.704842402451618e-14, "grad_norm": 0.5534878373146057, "learning_rate": 0.00010110319535694594, "loss": 2.388, "step": 25090 }, { "epoch": 8.708311849403572e-14, "grad_norm": 0.5453763008117676, "learning_rate": 0.0001010848942191898, "loss": 2.3928, "step": 25100 }, { "epoch": 8.711781296355525e-14, "grad_norm": 0.5495046377182007, "learning_rate": 0.00010106658796794848, "loss": 2.4265, "step": 25110 }, { "epoch": 8.715250743307479e-14, "grad_norm": 0.5632814168930054, "learning_rate": 0.00010104827660611694, "loss": 2.4011, "step": 25120 }, { "epoch": 8.718720190259432e-14, "grad_norm": 0.63042813539505, "learning_rate": 0.000101029960136591, "loss": 2.3744, "step": 25130 }, { "epoch": 8.722189637211386e-14, "grad_norm": 0.5926727652549744, "learning_rate": 0.00010101163856226723, "loss": 2.4175, "step": 25140 }, { "epoch": 8.72565908416334e-14, "grad_norm": 0.5500116348266602, "learning_rate": 0.00010099331188604302, "loss": 2.3656, "step": 25150 }, { "epoch": 8.729128531115293e-14, "grad_norm": 0.5631020665168762, "learning_rate": 0.0001009749801108166, "loss": 2.4189, "step": 25160 }, { "epoch": 8.732597978067247e-14, "grad_norm": 0.5522215366363525, "learning_rate": 0.00010095664323948696, "loss": 2.389, "step": 25170 }, { "epoch": 8.7360674250192e-14, "grad_norm": 0.5846664905548096, "learning_rate": 0.00010093830127495393, "loss": 2.3944, "step": 25180 }, { "epoch": 8.739536871971154e-14, "grad_norm": 0.5379654169082642, "learning_rate": 0.00010091995422011814, "loss": 2.3725, "step": 25190 }, { "epoch": 8.743006318923108e-14, "grad_norm": 0.5426070094108582, "learning_rate": 0.000100901602077881, "loss": 2.3908, "step": 25200 }, { "epoch": 8.746475765875061e-14, "grad_norm": 0.5691808462142944, "learning_rate": 0.00010088324485114476, "loss": 2.4089, "step": 25210 }, { "epoch": 8.749945212827015e-14, "grad_norm": 0.545245885848999, "learning_rate": 0.00010086488254281247, "loss": 2.339, "step": 25220 }, { "epoch": 8.753414659778969e-14, "grad_norm": 0.5780717134475708, "learning_rate": 0.0001008465151557879, "loss": 2.3963, "step": 25230 }, { "epoch": 8.756884106730922e-14, "grad_norm": 0.5712110996246338, "learning_rate": 0.0001008281426929758, "loss": 2.3628, "step": 25240 }, { "epoch": 8.760353553682876e-14, "grad_norm": 0.5786333084106445, "learning_rate": 0.00010080976515728155, "loss": 2.3967, "step": 25250 }, { "epoch": 8.76382300063483e-14, "grad_norm": 0.6028172373771667, "learning_rate": 0.00010079138255161139, "loss": 2.3853, "step": 25260 }, { "epoch": 8.767292447586783e-14, "grad_norm": 0.5892795920372009, "learning_rate": 0.00010077299487887243, "loss": 2.3712, "step": 25270 }, { "epoch": 8.770761894538737e-14, "grad_norm": 0.5813065767288208, "learning_rate": 0.00010075460214197246, "loss": 2.4208, "step": 25280 }, { "epoch": 8.77423134149069e-14, "grad_norm": 0.5719956159591675, "learning_rate": 0.00010073620434382016, "loss": 2.4037, "step": 25290 }, { "epoch": 8.777700788442644e-14, "grad_norm": 0.5505186319351196, "learning_rate": 0.000100717801487325, "loss": 2.3879, "step": 25300 }, { "epoch": 8.781170235394598e-14, "grad_norm": 0.5519489049911499, "learning_rate": 0.0001006993935753972, "loss": 2.3719, "step": 25310 }, { "epoch": 8.784639682346551e-14, "grad_norm": 0.5606244802474976, "learning_rate": 0.00010068098061094782, "loss": 2.3917, "step": 25320 }, { "epoch": 8.788109129298505e-14, "grad_norm": 0.5586726069450378, "learning_rate": 0.00010066256259688873, "loss": 2.4248, "step": 25330 }, { "epoch": 8.791578576250458e-14, "grad_norm": 0.6390678882598877, "learning_rate": 0.00010064413953613259, "loss": 2.3975, "step": 25340 }, { "epoch": 8.795048023202412e-14, "grad_norm": 0.5520294904708862, "learning_rate": 0.0001006257114315928, "loss": 2.3759, "step": 25350 }, { "epoch": 8.798517470154366e-14, "grad_norm": 0.5688536167144775, "learning_rate": 0.00010060727828618364, "loss": 2.4016, "step": 25360 }, { "epoch": 8.801986917106319e-14, "grad_norm": 0.5656577944755554, "learning_rate": 0.00010058884010282018, "loss": 2.3975, "step": 25370 }, { "epoch": 8.805456364058273e-14, "grad_norm": 0.5530942678451538, "learning_rate": 0.00010057039688441819, "loss": 2.3864, "step": 25380 }, { "epoch": 8.808925811010226e-14, "grad_norm": 0.5796802043914795, "learning_rate": 0.00010055194863389437, "loss": 2.3745, "step": 25390 }, { "epoch": 8.81239525796218e-14, "grad_norm": 0.5756041407585144, "learning_rate": 0.0001005334953541661, "loss": 2.3888, "step": 25400 }, { "epoch": 8.815864704914134e-14, "grad_norm": 0.577934980392456, "learning_rate": 0.00010051503704815164, "loss": 2.3684, "step": 25410 }, { "epoch": 8.819334151866087e-14, "grad_norm": 0.5569955110549927, "learning_rate": 0.00010049657371877002, "loss": 2.3664, "step": 25420 }, { "epoch": 8.822803598818041e-14, "grad_norm": 0.5970001816749573, "learning_rate": 0.00010047810536894104, "loss": 2.4066, "step": 25430 }, { "epoch": 8.826273045769994e-14, "grad_norm": 0.5611277222633362, "learning_rate": 0.00010045963200158531, "loss": 2.3838, "step": 25440 }, { "epoch": 8.829742492721948e-14, "grad_norm": 0.55644690990448, "learning_rate": 0.00010044115361962423, "loss": 2.4027, "step": 25450 }, { "epoch": 8.833211939673902e-14, "grad_norm": 0.5618312954902649, "learning_rate": 0.00010042267022598002, "loss": 2.4234, "step": 25460 }, { "epoch": 8.836681386625855e-14, "grad_norm": 0.5939270853996277, "learning_rate": 0.00010040418182357564, "loss": 2.3966, "step": 25470 }, { "epoch": 8.840150833577809e-14, "grad_norm": 0.6014487147331238, "learning_rate": 0.0001003856884153349, "loss": 2.3751, "step": 25480 }, { "epoch": 8.843620280529763e-14, "grad_norm": 0.6413789987564087, "learning_rate": 0.00010036719000418235, "loss": 2.3868, "step": 25490 }, { "epoch": 8.847089727481716e-14, "grad_norm": 0.5685141682624817, "learning_rate": 0.00010034868659304337, "loss": 2.3992, "step": 25500 }, { "epoch": 8.85055917443367e-14, "grad_norm": 0.584909200668335, "learning_rate": 0.0001003301781848441, "loss": 2.3766, "step": 25510 }, { "epoch": 8.854028621385623e-14, "grad_norm": 0.5715394020080566, "learning_rate": 0.00010031166478251149, "loss": 2.3819, "step": 25520 }, { "epoch": 8.857498068337577e-14, "grad_norm": 0.5716086626052856, "learning_rate": 0.00010029314638897329, "loss": 2.408, "step": 25530 }, { "epoch": 8.860967515289531e-14, "grad_norm": 0.5671064853668213, "learning_rate": 0.00010027462300715801, "loss": 2.3716, "step": 25540 }, { "epoch": 8.864436962241484e-14, "grad_norm": 0.6352173089981079, "learning_rate": 0.00010025609463999496, "loss": 2.4194, "step": 25550 }, { "epoch": 8.867906409193438e-14, "grad_norm": 0.5575957298278809, "learning_rate": 0.00010023756129041426, "loss": 2.3848, "step": 25560 }, { "epoch": 8.871375856145391e-14, "grad_norm": 0.5679643154144287, "learning_rate": 0.00010021902296134678, "loss": 2.3785, "step": 25570 }, { "epoch": 8.874845303097345e-14, "grad_norm": 0.5638715624809265, "learning_rate": 0.00010020047965572422, "loss": 2.3851, "step": 25580 }, { "epoch": 8.878314750049299e-14, "grad_norm": 0.5886834263801575, "learning_rate": 0.00010018193137647901, "loss": 2.3869, "step": 25590 }, { "epoch": 8.881784197001252e-14, "grad_norm": 0.5994704365730286, "learning_rate": 0.00010016337812654444, "loss": 2.3661, "step": 25600 }, { "epoch": 8.885253643953206e-14, "grad_norm": 0.5637876987457275, "learning_rate": 0.00010014481990885452, "loss": 2.3993, "step": 25610 }, { "epoch": 8.88872309090516e-14, "grad_norm": 0.5666521787643433, "learning_rate": 0.00010012625672634408, "loss": 2.3774, "step": 25620 }, { "epoch": 8.892192537857113e-14, "grad_norm": 0.5720869302749634, "learning_rate": 0.00010010768858194873, "loss": 2.4085, "step": 25630 }, { "epoch": 8.895661984809067e-14, "grad_norm": 0.544829249382019, "learning_rate": 0.00010008911547860487, "loss": 2.3954, "step": 25640 }, { "epoch": 8.89913143176102e-14, "grad_norm": 0.5669230818748474, "learning_rate": 0.00010007053741924964, "loss": 2.4052, "step": 25650 }, { "epoch": 8.902600878712974e-14, "grad_norm": 0.5818740129470825, "learning_rate": 0.00010005195440682104, "loss": 2.3689, "step": 25660 }, { "epoch": 8.906070325664928e-14, "grad_norm": 0.5515725612640381, "learning_rate": 0.0001000333664442578, "loss": 2.3958, "step": 25670 }, { "epoch": 8.909539772616881e-14, "grad_norm": 0.5459871888160706, "learning_rate": 0.00010001477353449943, "loss": 2.4058, "step": 25680 }, { "epoch": 8.913009219568835e-14, "grad_norm": 0.580108642578125, "learning_rate": 9.999617568048627e-05, "loss": 2.347, "step": 25690 }, { "epoch": 8.916478666520788e-14, "grad_norm": 0.5690877437591553, "learning_rate": 9.99775728851594e-05, "loss": 2.379, "step": 25700 }, { "epoch": 8.919948113472742e-14, "grad_norm": 0.5942381024360657, "learning_rate": 9.995896515146069e-05, "loss": 2.3744, "step": 25710 }, { "epoch": 8.923417560424696e-14, "grad_norm": 0.5407899618148804, "learning_rate": 9.994035248233277e-05, "loss": 2.3683, "step": 25720 }, { "epoch": 8.926887007376649e-14, "grad_norm": 0.5577166676521301, "learning_rate": 9.99217348807191e-05, "loss": 2.3648, "step": 25730 }, { "epoch": 8.930356454328603e-14, "grad_norm": 0.5734179615974426, "learning_rate": 9.990311234956389e-05, "loss": 2.3668, "step": 25740 }, { "epoch": 8.933825901280557e-14, "grad_norm": 0.5852835774421692, "learning_rate": 9.988448489181214e-05, "loss": 2.3822, "step": 25750 }, { "epoch": 8.93729534823251e-14, "grad_norm": 0.5744771361351013, "learning_rate": 9.986585251040959e-05, "loss": 2.4034, "step": 25760 }, { "epoch": 8.940764795184464e-14, "grad_norm": 0.5708554983139038, "learning_rate": 9.984721520830283e-05, "loss": 2.3961, "step": 25770 }, { "epoch": 8.944234242136417e-14, "grad_norm": 0.5805282592773438, "learning_rate": 9.982857298843919e-05, "loss": 2.3857, "step": 25780 }, { "epoch": 8.947703689088371e-14, "grad_norm": 0.57811039686203, "learning_rate": 9.980992585376672e-05, "loss": 2.3943, "step": 25790 }, { "epoch": 8.951173136040325e-14, "grad_norm": 0.5582066178321838, "learning_rate": 9.979127380723435e-05, "loss": 2.382, "step": 25800 }, { "epoch": 8.954642582992278e-14, "grad_norm": 0.5212278366088867, "learning_rate": 9.977261685179174e-05, "loss": 2.3988, "step": 25810 }, { "epoch": 8.958112029944232e-14, "grad_norm": 0.5996244549751282, "learning_rate": 9.975395499038932e-05, "loss": 2.3511, "step": 25820 }, { "epoch": 8.961581476896185e-14, "grad_norm": 0.5702414512634277, "learning_rate": 9.973528822597831e-05, "loss": 2.4036, "step": 25830 }, { "epoch": 8.965050923848139e-14, "grad_norm": 0.565920889377594, "learning_rate": 9.971661656151069e-05, "loss": 2.3341, "step": 25840 }, { "epoch": 8.968520370800093e-14, "grad_norm": 0.5846312642097473, "learning_rate": 9.969793999993919e-05, "loss": 2.409, "step": 25850 }, { "epoch": 8.971989817752046e-14, "grad_norm": 0.5534244179725647, "learning_rate": 9.96792585442174e-05, "loss": 2.4065, "step": 25860 }, { "epoch": 8.975459264704e-14, "grad_norm": 0.5511627793312073, "learning_rate": 9.966057219729962e-05, "loss": 2.411, "step": 25870 }, { "epoch": 8.978928711655954e-14, "grad_norm": 0.5677987337112427, "learning_rate": 9.964188096214093e-05, "loss": 2.389, "step": 25880 }, { "epoch": 8.982398158607907e-14, "grad_norm": 0.5706063508987427, "learning_rate": 9.962318484169716e-05, "loss": 2.4025, "step": 25890 }, { "epoch": 8.985867605559861e-14, "grad_norm": 0.5633090138435364, "learning_rate": 9.9604483838925e-05, "loss": 2.4128, "step": 25900 }, { "epoch": 8.989337052511814e-14, "grad_norm": 0.6293523907661438, "learning_rate": 9.95857779567818e-05, "loss": 2.3965, "step": 25910 }, { "epoch": 8.992806499463768e-14, "grad_norm": 0.5634257793426514, "learning_rate": 9.956706719822576e-05, "loss": 2.391, "step": 25920 }, { "epoch": 8.996275946415722e-14, "grad_norm": 0.5691273212432861, "learning_rate": 9.954835156621583e-05, "loss": 2.3626, "step": 25930 }, { "epoch": 8.999745393367675e-14, "grad_norm": 0.5525830984115601, "learning_rate": 9.952963106371168e-05, "loss": 2.3963, "step": 25940 }, { "epoch": 9.003214840319629e-14, "grad_norm": 0.5777362585067749, "learning_rate": 9.951090569367384e-05, "loss": 2.3749, "step": 25950 }, { "epoch": 9.006684287271582e-14, "grad_norm": 0.5822543501853943, "learning_rate": 9.949217545906357e-05, "loss": 2.3671, "step": 25960 }, { "epoch": 9.010153734223536e-14, "grad_norm": 0.5539024472236633, "learning_rate": 9.947344036284289e-05, "loss": 2.3715, "step": 25970 }, { "epoch": 9.01362318117549e-14, "grad_norm": 0.5777491331100464, "learning_rate": 9.945470040797455e-05, "loss": 2.3635, "step": 25980 }, { "epoch": 9.017092628127443e-14, "grad_norm": 0.5875590443611145, "learning_rate": 9.943595559742218e-05, "loss": 2.3355, "step": 25990 }, { "epoch": 9.020562075079397e-14, "grad_norm": 0.5927296280860901, "learning_rate": 9.941720593415005e-05, "loss": 2.3674, "step": 26000 }, { "epoch": 9.02403152203135e-14, "grad_norm": 0.5617555379867554, "learning_rate": 9.93984514211233e-05, "loss": 2.3905, "step": 26010 }, { "epoch": 9.027500968983304e-14, "grad_norm": 0.5657986998558044, "learning_rate": 9.937969206130778e-05, "loss": 2.3972, "step": 26020 }, { "epoch": 9.030970415935258e-14, "grad_norm": 0.5614043474197388, "learning_rate": 9.936092785767014e-05, "loss": 2.3518, "step": 26030 }, { "epoch": 9.034439862887211e-14, "grad_norm": 0.5545433759689331, "learning_rate": 9.934215881317776e-05, "loss": 2.3774, "step": 26040 }, { "epoch": 9.037909309839165e-14, "grad_norm": 0.5844190716743469, "learning_rate": 9.93233849307988e-05, "loss": 2.3815, "step": 26050 }, { "epoch": 9.041378756791119e-14, "grad_norm": 0.571398913860321, "learning_rate": 9.930460621350216e-05, "loss": 2.3787, "step": 26060 }, { "epoch": 9.044848203743072e-14, "grad_norm": 0.5893279314041138, "learning_rate": 9.92858226642576e-05, "loss": 2.4011, "step": 26070 }, { "epoch": 9.048317650695026e-14, "grad_norm": 0.5689113140106201, "learning_rate": 9.926703428603551e-05, "loss": 2.4001, "step": 26080 }, { "epoch": 9.05178709764698e-14, "grad_norm": 0.5700855255126953, "learning_rate": 9.924824108180714e-05, "loss": 2.387, "step": 26090 }, { "epoch": 9.055256544598933e-14, "grad_norm": 0.557524561882019, "learning_rate": 9.92294430545445e-05, "loss": 2.3636, "step": 26100 }, { "epoch": 9.058725991550887e-14, "grad_norm": 0.566113293170929, "learning_rate": 9.92106402072203e-05, "loss": 2.3439, "step": 26110 }, { "epoch": 9.06219543850284e-14, "grad_norm": 0.5610853433609009, "learning_rate": 9.919183254280806e-05, "loss": 2.366, "step": 26120 }, { "epoch": 9.065664885454794e-14, "grad_norm": 0.5761110782623291, "learning_rate": 9.917302006428204e-05, "loss": 2.3641, "step": 26130 }, { "epoch": 9.069134332406747e-14, "grad_norm": 0.5876413583755493, "learning_rate": 9.915420277461728e-05, "loss": 2.343, "step": 26140 }, { "epoch": 9.072603779358701e-14, "grad_norm": 0.5548674464225769, "learning_rate": 9.913538067678958e-05, "loss": 2.3909, "step": 26150 }, { "epoch": 9.076073226310655e-14, "grad_norm": 0.6105297803878784, "learning_rate": 9.911655377377548e-05, "loss": 2.3624, "step": 26160 }, { "epoch": 9.079542673262608e-14, "grad_norm": 0.6085036993026733, "learning_rate": 9.909772206855232e-05, "loss": 2.3798, "step": 26170 }, { "epoch": 9.083012120214562e-14, "grad_norm": 0.58851158618927, "learning_rate": 9.907888556409813e-05, "loss": 2.3507, "step": 26180 }, { "epoch": 9.086481567166516e-14, "grad_norm": 0.5430507063865662, "learning_rate": 9.906004426339176e-05, "loss": 2.3783, "step": 26190 }, { "epoch": 9.089951014118469e-14, "grad_norm": 0.5690869688987732, "learning_rate": 9.90411981694128e-05, "loss": 2.3774, "step": 26200 }, { "epoch": 9.093420461070423e-14, "grad_norm": 0.581724226474762, "learning_rate": 9.902234728514159e-05, "loss": 2.3919, "step": 26210 }, { "epoch": 9.096889908022376e-14, "grad_norm": 0.6033334136009216, "learning_rate": 9.900349161355924e-05, "loss": 2.3808, "step": 26220 }, { "epoch": 9.10035935497433e-14, "grad_norm": 0.6234849691390991, "learning_rate": 9.898463115764763e-05, "loss": 2.3816, "step": 26230 }, { "epoch": 9.103828801926284e-14, "grad_norm": 0.5939595699310303, "learning_rate": 9.896576592038937e-05, "loss": 2.3961, "step": 26240 }, { "epoch": 9.107298248878237e-14, "grad_norm": 0.5694277882575989, "learning_rate": 9.894689590476782e-05, "loss": 2.4018, "step": 26250 }, { "epoch": 9.110767695830191e-14, "grad_norm": 0.5888341069221497, "learning_rate": 9.892802111376713e-05, "loss": 2.371, "step": 26260 }, { "epoch": 9.114237142782144e-14, "grad_norm": 0.5316814184188843, "learning_rate": 9.890914155037214e-05, "loss": 2.368, "step": 26270 }, { "epoch": 9.117706589734098e-14, "grad_norm": 0.5661939382553101, "learning_rate": 9.889025721756856e-05, "loss": 2.3683, "step": 26280 }, { "epoch": 9.121176036686052e-14, "grad_norm": 0.5753399133682251, "learning_rate": 9.887136811834274e-05, "loss": 2.3455, "step": 26290 }, { "epoch": 9.124645483638005e-14, "grad_norm": 0.5685588717460632, "learning_rate": 9.885247425568184e-05, "loss": 2.364, "step": 26300 }, { "epoch": 9.128114930589959e-14, "grad_norm": 0.5941200256347656, "learning_rate": 9.883357563257376e-05, "loss": 2.3947, "step": 26310 }, { "epoch": 9.131584377541913e-14, "grad_norm": 0.5716224908828735, "learning_rate": 9.881467225200715e-05, "loss": 2.3627, "step": 26320 }, { "epoch": 9.135053824493866e-14, "grad_norm": 0.5516136884689331, "learning_rate": 9.879576411697144e-05, "loss": 2.3921, "step": 26330 }, { "epoch": 9.13852327144582e-14, "grad_norm": 0.5762332081794739, "learning_rate": 9.877685123045675e-05, "loss": 2.4066, "step": 26340 }, { "epoch": 9.141992718397773e-14, "grad_norm": 0.5684521794319153, "learning_rate": 9.875793359545401e-05, "loss": 2.3827, "step": 26350 }, { "epoch": 9.145462165349727e-14, "grad_norm": 0.5455964803695679, "learning_rate": 9.873901121495489e-05, "loss": 2.3675, "step": 26360 }, { "epoch": 9.14893161230168e-14, "grad_norm": 0.5769913196563721, "learning_rate": 9.872008409195178e-05, "loss": 2.4092, "step": 26370 }, { "epoch": 9.152401059253634e-14, "grad_norm": 0.6008603572845459, "learning_rate": 9.870115222943788e-05, "loss": 2.3762, "step": 26380 }, { "epoch": 9.155870506205588e-14, "grad_norm": 0.6010853052139282, "learning_rate": 9.868221563040706e-05, "loss": 2.36, "step": 26390 }, { "epoch": 9.159339953157541e-14, "grad_norm": 0.5582424402236938, "learning_rate": 9.8663274297854e-05, "loss": 2.3449, "step": 26400 }, { "epoch": 9.162809400109495e-14, "grad_norm": 0.6030482053756714, "learning_rate": 9.86443282347741e-05, "loss": 2.42, "step": 26410 }, { "epoch": 9.166278847061449e-14, "grad_norm": 0.5933235287666321, "learning_rate": 9.862537744416354e-05, "loss": 2.3578, "step": 26420 }, { "epoch": 9.169748294013402e-14, "grad_norm": 0.5908945202827454, "learning_rate": 9.860642192901919e-05, "loss": 2.4361, "step": 26430 }, { "epoch": 9.173217740965356e-14, "grad_norm": 0.5704118609428406, "learning_rate": 9.858746169233872e-05, "loss": 2.3654, "step": 26440 }, { "epoch": 9.17668718791731e-14, "grad_norm": 0.5572944283485413, "learning_rate": 9.856849673712053e-05, "loss": 2.3524, "step": 26450 }, { "epoch": 9.180156634869263e-14, "grad_norm": 0.5578535795211792, "learning_rate": 9.854952706636376e-05, "loss": 2.394, "step": 26460 }, { "epoch": 9.183626081821217e-14, "grad_norm": 0.5803686380386353, "learning_rate": 9.85305526830683e-05, "loss": 2.379, "step": 26470 }, { "epoch": 9.18709552877317e-14, "grad_norm": 0.6014369130134583, "learning_rate": 9.851157359023479e-05, "loss": 2.3787, "step": 26480 }, { "epoch": 9.190564975725124e-14, "grad_norm": 0.5663735270500183, "learning_rate": 9.84925897908646e-05, "loss": 2.4103, "step": 26490 }, { "epoch": 9.194034422677078e-14, "grad_norm": 0.6000224947929382, "learning_rate": 9.847360128795984e-05, "loss": 2.408, "step": 26500 }, { "epoch": 9.197503869629031e-14, "grad_norm": 0.5870617032051086, "learning_rate": 9.845460808452343e-05, "loss": 2.4056, "step": 26510 }, { "epoch": 9.200973316580985e-14, "grad_norm": 0.5662401914596558, "learning_rate": 9.843561018355894e-05, "loss": 2.359, "step": 26520 }, { "epoch": 9.204442763532938e-14, "grad_norm": 0.5683334469795227, "learning_rate": 9.841660758807073e-05, "loss": 2.3943, "step": 26530 }, { "epoch": 9.207912210484892e-14, "grad_norm": 0.5936341881752014, "learning_rate": 9.839760030106392e-05, "loss": 2.3728, "step": 26540 }, { "epoch": 9.211381657436846e-14, "grad_norm": 0.5594950914382935, "learning_rate": 9.837858832554431e-05, "loss": 2.3677, "step": 26550 }, { "epoch": 9.214851104388799e-14, "grad_norm": 0.5718585252761841, "learning_rate": 9.835957166451851e-05, "loss": 2.3724, "step": 26560 }, { "epoch": 9.218320551340753e-14, "grad_norm": 0.5967565774917603, "learning_rate": 9.834055032099382e-05, "loss": 2.3733, "step": 26570 }, { "epoch": 9.221789998292707e-14, "grad_norm": 0.5899174809455872, "learning_rate": 9.832152429797835e-05, "loss": 2.3774, "step": 26580 }, { "epoch": 9.22525944524466e-14, "grad_norm": 0.5733642578125, "learning_rate": 9.830249359848082e-05, "loss": 2.3771, "step": 26590 }, { "epoch": 9.228728892196614e-14, "grad_norm": 0.5655060410499573, "learning_rate": 9.828345822551083e-05, "loss": 2.4005, "step": 26600 }, { "epoch": 9.232198339148567e-14, "grad_norm": 0.563005805015564, "learning_rate": 9.826441818207867e-05, "loss": 2.3696, "step": 26610 }, { "epoch": 9.235667786100521e-14, "grad_norm": 0.5723924040794373, "learning_rate": 9.824537347119531e-05, "loss": 2.3874, "step": 26620 }, { "epoch": 9.239137233052475e-14, "grad_norm": 0.5860955715179443, "learning_rate": 9.822632409587255e-05, "loss": 2.3757, "step": 26630 }, { "epoch": 9.242606680004428e-14, "grad_norm": 0.5581047534942627, "learning_rate": 9.820727005912285e-05, "loss": 2.3572, "step": 26640 }, { "epoch": 9.246076126956382e-14, "grad_norm": 0.5756823420524597, "learning_rate": 9.81882113639595e-05, "loss": 2.3387, "step": 26650 }, { "epoch": 9.249545573908335e-14, "grad_norm": 0.5699936151504517, "learning_rate": 9.816914801339638e-05, "loss": 2.363, "step": 26660 }, { "epoch": 9.253015020860289e-14, "grad_norm": 0.5774894952774048, "learning_rate": 9.815008001044827e-05, "loss": 2.3842, "step": 26670 }, { "epoch": 9.256484467812243e-14, "grad_norm": 0.5802525877952576, "learning_rate": 9.813100735813057e-05, "loss": 2.3613, "step": 26680 }, { "epoch": 9.259953914764196e-14, "grad_norm": 0.5755798816680908, "learning_rate": 9.811193005945948e-05, "loss": 2.4235, "step": 26690 }, { "epoch": 9.26342336171615e-14, "grad_norm": 0.5471661686897278, "learning_rate": 9.809284811745189e-05, "loss": 2.3737, "step": 26700 }, { "epoch": 9.266892808668103e-14, "grad_norm": 0.5589843988418579, "learning_rate": 9.807376153512548e-05, "loss": 2.3557, "step": 26710 }, { "epoch": 9.270362255620057e-14, "grad_norm": 0.5519663691520691, "learning_rate": 9.805467031549859e-05, "loss": 2.3207, "step": 26720 }, { "epoch": 9.273831702572011e-14, "grad_norm": 0.6167529225349426, "learning_rate": 9.803557446159035e-05, "loss": 2.3864, "step": 26730 }, { "epoch": 9.277301149523964e-14, "grad_norm": 0.5761629343032837, "learning_rate": 9.801647397642059e-05, "loss": 2.3918, "step": 26740 }, { "epoch": 9.280770596475918e-14, "grad_norm": 0.617478609085083, "learning_rate": 9.799736886300988e-05, "loss": 2.4029, "step": 26750 }, { "epoch": 9.284240043427872e-14, "grad_norm": 0.5786637663841248, "learning_rate": 9.797825912437959e-05, "loss": 2.3752, "step": 26760 }, { "epoch": 9.287709490379825e-14, "grad_norm": 0.5757376551628113, "learning_rate": 9.79591447635517e-05, "loss": 2.3475, "step": 26770 }, { "epoch": 9.291178937331779e-14, "grad_norm": 0.6311630606651306, "learning_rate": 9.7940025783549e-05, "loss": 2.3733, "step": 26780 }, { "epoch": 9.294648384283732e-14, "grad_norm": 0.5906806588172913, "learning_rate": 9.792090218739499e-05, "loss": 2.4126, "step": 26790 }, { "epoch": 9.298117831235686e-14, "grad_norm": 0.5862739086151123, "learning_rate": 9.790177397811391e-05, "loss": 2.3959, "step": 26800 }, { "epoch": 9.30158727818764e-14, "grad_norm": 0.5703856348991394, "learning_rate": 9.78826411587307e-05, "loss": 2.3974, "step": 26810 }, { "epoch": 9.305056725139593e-14, "grad_norm": 0.5540105700492859, "learning_rate": 9.786350373227107e-05, "loss": 2.3762, "step": 26820 }, { "epoch": 9.308526172091547e-14, "grad_norm": 0.5527639389038086, "learning_rate": 9.784436170176144e-05, "loss": 2.3628, "step": 26830 }, { "epoch": 9.3119956190435e-14, "grad_norm": 0.5689169764518738, "learning_rate": 9.782521507022895e-05, "loss": 2.415, "step": 26840 }, { "epoch": 9.315465065995454e-14, "grad_norm": 0.5591270923614502, "learning_rate": 9.780606384070147e-05, "loss": 2.4027, "step": 26850 }, { "epoch": 9.318934512947408e-14, "grad_norm": 0.567287027835846, "learning_rate": 9.778690801620765e-05, "loss": 2.3512, "step": 26860 }, { "epoch": 9.322403959899361e-14, "grad_norm": 0.5710867643356323, "learning_rate": 9.776774759977675e-05, "loss": 2.3734, "step": 26870 }, { "epoch": 9.325873406851315e-14, "grad_norm": 0.5622454881668091, "learning_rate": 9.774858259443885e-05, "loss": 2.4088, "step": 26880 }, { "epoch": 9.329342853803269e-14, "grad_norm": 0.5648418068885803, "learning_rate": 9.772941300322474e-05, "loss": 2.3831, "step": 26890 }, { "epoch": 9.332812300755222e-14, "grad_norm": 0.5914862155914307, "learning_rate": 9.771023882916593e-05, "loss": 2.3948, "step": 26900 }, { "epoch": 9.336281747707176e-14, "grad_norm": 0.5581477880477905, "learning_rate": 9.76910600752946e-05, "loss": 2.3499, "step": 26910 }, { "epoch": 9.33975119465913e-14, "grad_norm": 0.5662215948104858, "learning_rate": 9.767187674464378e-05, "loss": 2.3778, "step": 26920 }, { "epoch": 9.343220641611083e-14, "grad_norm": 0.5891581773757935, "learning_rate": 9.765268884024711e-05, "loss": 2.4088, "step": 26930 }, { "epoch": 9.346690088563037e-14, "grad_norm": 0.5874652862548828, "learning_rate": 9.7633496365139e-05, "loss": 2.3461, "step": 26940 }, { "epoch": 9.35015953551499e-14, "grad_norm": 0.5963525176048279, "learning_rate": 9.761429932235457e-05, "loss": 2.3672, "step": 26950 }, { "epoch": 9.353628982466944e-14, "grad_norm": 0.5786092281341553, "learning_rate": 9.759509771492966e-05, "loss": 2.3579, "step": 26960 }, { "epoch": 9.357098429418897e-14, "grad_norm": 0.6231219172477722, "learning_rate": 9.757589154590085e-05, "loss": 2.4037, "step": 26970 }, { "epoch": 9.360567876370851e-14, "grad_norm": 0.5576334595680237, "learning_rate": 9.755668081830542e-05, "loss": 2.4069, "step": 26980 }, { "epoch": 9.364037323322805e-14, "grad_norm": 0.5817937850952148, "learning_rate": 9.753746553518139e-05, "loss": 2.3946, "step": 26990 }, { "epoch": 9.367506770274758e-14, "grad_norm": 0.5914424061775208, "learning_rate": 9.75182456995675e-05, "loss": 2.3528, "step": 27000 }, { "epoch": 9.370976217226712e-14, "grad_norm": 0.5661783814430237, "learning_rate": 9.749902131450318e-05, "loss": 2.3604, "step": 27010 }, { "epoch": 9.374445664178666e-14, "grad_norm": 0.5829393267631531, "learning_rate": 9.747979238302864e-05, "loss": 2.3586, "step": 27020 }, { "epoch": 9.377915111130619e-14, "grad_norm": 0.5448939204216003, "learning_rate": 9.746055890818471e-05, "loss": 2.3237, "step": 27030 }, { "epoch": 9.381384558082573e-14, "grad_norm": 0.5770891904830933, "learning_rate": 9.744132089301305e-05, "loss": 2.3575, "step": 27040 }, { "epoch": 9.384854005034526e-14, "grad_norm": 0.6064008474349976, "learning_rate": 9.742207834055594e-05, "loss": 2.3616, "step": 27050 }, { "epoch": 9.38832345198648e-14, "grad_norm": 0.5848496556282043, "learning_rate": 9.740283125385648e-05, "loss": 2.385, "step": 27060 }, { "epoch": 9.391792898938434e-14, "grad_norm": 0.5516229271888733, "learning_rate": 9.738357963595838e-05, "loss": 2.3702, "step": 27070 }, { "epoch": 9.395262345890387e-14, "grad_norm": 0.5650082230567932, "learning_rate": 9.736432348990617e-05, "loss": 2.3482, "step": 27080 }, { "epoch": 9.398731792842341e-14, "grad_norm": 0.5574127435684204, "learning_rate": 9.734506281874498e-05, "loss": 2.3899, "step": 27090 }, { "epoch": 9.402201239794294e-14, "grad_norm": 0.5920735001564026, "learning_rate": 9.732579762552079e-05, "loss": 2.3926, "step": 27100 }, { "epoch": 9.405670686746248e-14, "grad_norm": 0.6036304235458374, "learning_rate": 9.730652791328017e-05, "loss": 2.3493, "step": 27110 }, { "epoch": 9.409140133698202e-14, "grad_norm": 0.585379958152771, "learning_rate": 9.728725368507045e-05, "loss": 2.3597, "step": 27120 }, { "epoch": 9.412609580650155e-14, "grad_norm": 0.5722930431365967, "learning_rate": 9.726797494393975e-05, "loss": 2.3605, "step": 27130 }, { "epoch": 9.416079027602109e-14, "grad_norm": 0.5764927864074707, "learning_rate": 9.724869169293677e-05, "loss": 2.3685, "step": 27140 }, { "epoch": 9.419548474554063e-14, "grad_norm": 0.5665320754051208, "learning_rate": 9.722940393511104e-05, "loss": 2.3608, "step": 27150 }, { "epoch": 9.423017921506016e-14, "grad_norm": 0.5916764736175537, "learning_rate": 9.721011167351271e-05, "loss": 2.3829, "step": 27160 }, { "epoch": 9.42648736845797e-14, "grad_norm": 0.5768653750419617, "learning_rate": 9.719081491119272e-05, "loss": 2.3407, "step": 27170 }, { "epoch": 9.429956815409923e-14, "grad_norm": 0.5662758946418762, "learning_rate": 9.717151365120266e-05, "loss": 2.3598, "step": 27180 }, { "epoch": 9.433426262361877e-14, "grad_norm": 0.5649036169052124, "learning_rate": 9.715220789659489e-05, "loss": 2.3553, "step": 27190 }, { "epoch": 9.43689570931383e-14, "grad_norm": 0.5650597810745239, "learning_rate": 9.713289765042241e-05, "loss": 2.3608, "step": 27200 }, { "epoch": 9.440365156265784e-14, "grad_norm": 0.5699713826179504, "learning_rate": 9.7113582915739e-05, "loss": 2.3385, "step": 27210 }, { "epoch": 9.443834603217738e-14, "grad_norm": 0.5610552430152893, "learning_rate": 9.709426369559909e-05, "loss": 2.3698, "step": 27220 }, { "epoch": 9.447304050169691e-14, "grad_norm": 0.5557568073272705, "learning_rate": 9.707493999305788e-05, "loss": 2.4103, "step": 27230 }, { "epoch": 9.450773497121645e-14, "grad_norm": 0.5765385627746582, "learning_rate": 9.705561181117122e-05, "loss": 2.3604, "step": 27240 }, { "epoch": 9.454242944073599e-14, "grad_norm": 0.5731558799743652, "learning_rate": 9.703627915299573e-05, "loss": 2.3311, "step": 27250 }, { "epoch": 9.457712391025552e-14, "grad_norm": 0.5674732327461243, "learning_rate": 9.701694202158866e-05, "loss": 2.3519, "step": 27260 }, { "epoch": 9.461181837977506e-14, "grad_norm": 0.57768714427948, "learning_rate": 9.699760042000804e-05, "loss": 2.3799, "step": 27270 }, { "epoch": 9.46465128492946e-14, "grad_norm": 0.6114889979362488, "learning_rate": 9.697825435131257e-05, "loss": 2.38, "step": 27280 }, { "epoch": 9.468120731881413e-14, "grad_norm": 0.60740727186203, "learning_rate": 9.695890381856168e-05, "loss": 2.34, "step": 27290 }, { "epoch": 9.471590178833367e-14, "grad_norm": 0.5442131757736206, "learning_rate": 9.693954882481546e-05, "loss": 2.391, "step": 27300 }, { "epoch": 9.47505962578532e-14, "grad_norm": 0.574073314666748, "learning_rate": 9.692018937313475e-05, "loss": 2.4077, "step": 27310 }, { "epoch": 9.478529072737274e-14, "grad_norm": 0.55948805809021, "learning_rate": 9.690082546658107e-05, "loss": 2.3612, "step": 27320 }, { "epoch": 9.481998519689228e-14, "grad_norm": 0.6142717003822327, "learning_rate": 9.688145710821669e-05, "loss": 2.362, "step": 27330 }, { "epoch": 9.485467966641181e-14, "grad_norm": 0.5908925533294678, "learning_rate": 9.686208430110452e-05, "loss": 2.3529, "step": 27340 }, { "epoch": 9.488937413593135e-14, "grad_norm": 0.5796312093734741, "learning_rate": 9.684270704830821e-05, "loss": 2.3692, "step": 27350 }, { "epoch": 9.492406860545088e-14, "grad_norm": 0.601659893989563, "learning_rate": 9.682332535289209e-05, "loss": 2.386, "step": 27360 }, { "epoch": 9.495876307497042e-14, "grad_norm": 0.5859995484352112, "learning_rate": 9.680393921792124e-05, "loss": 2.3723, "step": 27370 }, { "epoch": 9.499345754448996e-14, "grad_norm": 0.569171130657196, "learning_rate": 9.678454864646138e-05, "loss": 2.3447, "step": 27380 }, { "epoch": 9.502815201400949e-14, "grad_norm": 0.5575588345527649, "learning_rate": 9.676515364157898e-05, "loss": 2.3638, "step": 27390 }, { "epoch": 9.506284648352903e-14, "grad_norm": 0.5764550566673279, "learning_rate": 9.674575420634119e-05, "loss": 2.3432, "step": 27400 }, { "epoch": 9.509754095304856e-14, "grad_norm": 0.5980959534645081, "learning_rate": 9.672635034381586e-05, "loss": 2.3523, "step": 27410 }, { "epoch": 9.51322354225681e-14, "grad_norm": 0.5719417929649353, "learning_rate": 9.670694205707154e-05, "loss": 2.3645, "step": 27420 }, { "epoch": 9.516692989208764e-14, "grad_norm": 0.6183213591575623, "learning_rate": 9.66875293491775e-05, "loss": 2.3974, "step": 27430 }, { "epoch": 9.520162436160717e-14, "grad_norm": 0.598259687423706, "learning_rate": 9.666811222320366e-05, "loss": 2.3824, "step": 27440 }, { "epoch": 9.523631883112671e-14, "grad_norm": 0.5956482887268066, "learning_rate": 9.664869068222071e-05, "loss": 2.3985, "step": 27450 }, { "epoch": 9.527101330064625e-14, "grad_norm": 0.5603923201560974, "learning_rate": 9.662926472929998e-05, "loss": 2.3843, "step": 27460 }, { "epoch": 9.530570777016578e-14, "grad_norm": 0.5516121983528137, "learning_rate": 9.660983436751351e-05, "loss": 2.3434, "step": 27470 }, { "epoch": 9.534040223968532e-14, "grad_norm": 0.5622056722640991, "learning_rate": 9.659039959993406e-05, "loss": 2.3461, "step": 27480 }, { "epoch": 9.537509670920485e-14, "grad_norm": 0.5720241665840149, "learning_rate": 9.657096042963507e-05, "loss": 2.3738, "step": 27490 }, { "epoch": 9.540979117872439e-14, "grad_norm": 0.585159957408905, "learning_rate": 9.655151685969066e-05, "loss": 2.3356, "step": 27500 }, { "epoch": 9.544448564824393e-14, "grad_norm": 0.588363766670227, "learning_rate": 9.653206889317569e-05, "loss": 2.3702, "step": 27510 }, { "epoch": 9.547918011776346e-14, "grad_norm": 0.5737068057060242, "learning_rate": 9.651261653316567e-05, "loss": 2.3558, "step": 27520 }, { "epoch": 9.5513874587283e-14, "grad_norm": 0.5626336932182312, "learning_rate": 9.649315978273683e-05, "loss": 2.3934, "step": 27530 }, { "epoch": 9.554856905680253e-14, "grad_norm": 0.5713600516319275, "learning_rate": 9.647369864496607e-05, "loss": 2.3405, "step": 27540 }, { "epoch": 9.558326352632207e-14, "grad_norm": 0.5776804685592651, "learning_rate": 9.645423312293103e-05, "loss": 2.36, "step": 27550 }, { "epoch": 9.561795799584161e-14, "grad_norm": 0.5758205056190491, "learning_rate": 9.643476321971001e-05, "loss": 2.3384, "step": 27560 }, { "epoch": 9.565265246536114e-14, "grad_norm": 0.5797953605651855, "learning_rate": 9.641528893838199e-05, "loss": 2.3424, "step": 27570 }, { "epoch": 9.568734693488068e-14, "grad_norm": 0.6062913537025452, "learning_rate": 9.639581028202666e-05, "loss": 2.4121, "step": 27580 }, { "epoch": 9.572204140440022e-14, "grad_norm": 0.565854012966156, "learning_rate": 9.637632725372443e-05, "loss": 2.3498, "step": 27590 }, { "epoch": 9.575673587391975e-14, "grad_norm": 0.5743780136108398, "learning_rate": 9.635683985655633e-05, "loss": 2.3787, "step": 27600 }, { "epoch": 9.579143034343929e-14, "grad_norm": 0.574257493019104, "learning_rate": 9.633734809360418e-05, "loss": 2.3664, "step": 27610 }, { "epoch": 9.582612481295882e-14, "grad_norm": 0.5868635773658752, "learning_rate": 9.631785196795035e-05, "loss": 2.3721, "step": 27620 }, { "epoch": 9.586081928247836e-14, "grad_norm": 0.587618350982666, "learning_rate": 9.629835148267808e-05, "loss": 2.3619, "step": 27630 }, { "epoch": 9.58955137519979e-14, "grad_norm": 0.6046120524406433, "learning_rate": 9.627884664087116e-05, "loss": 2.3412, "step": 27640 }, { "epoch": 9.593020822151743e-14, "grad_norm": 0.5643002986907959, "learning_rate": 9.62593374456141e-05, "loss": 2.3614, "step": 27650 }, { "epoch": 9.596490269103697e-14, "grad_norm": 0.5837509036064148, "learning_rate": 9.62398238999921e-05, "loss": 2.4157, "step": 27660 }, { "epoch": 9.59995971605565e-14, "grad_norm": 0.5649028420448303, "learning_rate": 9.622030600709114e-05, "loss": 2.3568, "step": 27670 }, { "epoch": 9.603429163007604e-14, "grad_norm": 0.5506762862205505, "learning_rate": 9.620078376999772e-05, "loss": 2.3385, "step": 27680 }, { "epoch": 9.606898609959558e-14, "grad_norm": 0.5807803869247437, "learning_rate": 9.618125719179914e-05, "loss": 2.3903, "step": 27690 }, { "epoch": 9.610368056911511e-14, "grad_norm": 0.5881657600402832, "learning_rate": 9.616172627558336e-05, "loss": 2.3512, "step": 27700 }, { "epoch": 9.613837503863465e-14, "grad_norm": 0.5831103920936584, "learning_rate": 9.614219102443905e-05, "loss": 2.3441, "step": 27710 }, { "epoch": 9.617306950815419e-14, "grad_norm": 0.5939533710479736, "learning_rate": 9.612265144145552e-05, "loss": 2.3033, "step": 27720 }, { "epoch": 9.620776397767372e-14, "grad_norm": 0.5579217076301575, "learning_rate": 9.61031075297228e-05, "loss": 2.3282, "step": 27730 }, { "epoch": 9.624245844719326e-14, "grad_norm": 0.5684774518013, "learning_rate": 9.608355929233156e-05, "loss": 2.3705, "step": 27740 }, { "epoch": 9.62771529167128e-14, "grad_norm": 0.5690931081771851, "learning_rate": 9.606400673237322e-05, "loss": 2.3708, "step": 27750 }, { "epoch": 9.631184738623233e-14, "grad_norm": 0.5769988894462585, "learning_rate": 9.604444985293983e-05, "loss": 2.3513, "step": 27760 }, { "epoch": 9.634654185575187e-14, "grad_norm": 0.6014742255210876, "learning_rate": 9.602488865712417e-05, "loss": 2.4005, "step": 27770 }, { "epoch": 9.63812363252714e-14, "grad_norm": 0.5829352140426636, "learning_rate": 9.600532314801961e-05, "loss": 2.3641, "step": 27780 }, { "epoch": 9.641593079479094e-14, "grad_norm": 0.5782846212387085, "learning_rate": 9.598575332872034e-05, "loss": 2.3851, "step": 27790 }, { "epoch": 9.645062526431047e-14, "grad_norm": 0.5796065330505371, "learning_rate": 9.596617920232112e-05, "loss": 2.3591, "step": 27800 }, { "epoch": 9.648531973383001e-14, "grad_norm": 0.5778412818908691, "learning_rate": 9.594660077191742e-05, "loss": 2.3719, "step": 27810 }, { "epoch": 9.652001420334955e-14, "grad_norm": 0.564953625202179, "learning_rate": 9.592701804060545e-05, "loss": 2.3527, "step": 27820 }, { "epoch": 9.655470867286908e-14, "grad_norm": 0.5803773999214172, "learning_rate": 9.590743101148201e-05, "loss": 2.3788, "step": 27830 }, { "epoch": 9.658940314238862e-14, "grad_norm": 0.5610167384147644, "learning_rate": 9.588783968764461e-05, "loss": 2.3992, "step": 27840 }, { "epoch": 9.662409761190816e-14, "grad_norm": 0.5793769955635071, "learning_rate": 9.586824407219148e-05, "loss": 2.3541, "step": 27850 }, { "epoch": 9.665879208142769e-14, "grad_norm": 0.5699241757392883, "learning_rate": 9.584864416822147e-05, "loss": 2.354, "step": 27860 }, { "epoch": 9.669348655094723e-14, "grad_norm": 0.5740676522254944, "learning_rate": 9.582903997883419e-05, "loss": 2.3617, "step": 27870 }, { "epoch": 9.672818102046676e-14, "grad_norm": 0.5838427543640137, "learning_rate": 9.58094315071298e-05, "loss": 2.3556, "step": 27880 }, { "epoch": 9.67628754899863e-14, "grad_norm": 0.5831942558288574, "learning_rate": 9.578981875620926e-05, "loss": 2.3707, "step": 27890 }, { "epoch": 9.679756995950584e-14, "grad_norm": 0.5893644690513611, "learning_rate": 9.577020172917415e-05, "loss": 2.3674, "step": 27900 }, { "epoch": 9.683226442902537e-14, "grad_norm": 0.5981907844543457, "learning_rate": 9.575058042912673e-05, "loss": 2.3638, "step": 27910 }, { "epoch": 9.686695889854491e-14, "grad_norm": 0.5833572745323181, "learning_rate": 9.573095485916992e-05, "loss": 2.3475, "step": 27920 }, { "epoch": 9.690165336806444e-14, "grad_norm": 0.5528524518013, "learning_rate": 9.571132502240735e-05, "loss": 2.3604, "step": 27930 }, { "epoch": 9.693634783758398e-14, "grad_norm": 0.5459917187690735, "learning_rate": 9.569169092194333e-05, "loss": 2.3834, "step": 27940 }, { "epoch": 9.697104230710352e-14, "grad_norm": 0.5671799778938293, "learning_rate": 9.56720525608828e-05, "loss": 2.3545, "step": 27950 }, { "epoch": 9.700573677662305e-14, "grad_norm": 0.5588716864585876, "learning_rate": 9.565240994233141e-05, "loss": 2.3691, "step": 27960 }, { "epoch": 9.704043124614259e-14, "grad_norm": 0.5973741412162781, "learning_rate": 9.563276306939544e-05, "loss": 2.3796, "step": 27970 }, { "epoch": 9.707512571566213e-14, "grad_norm": 0.5891607999801636, "learning_rate": 9.561311194518192e-05, "loss": 2.3397, "step": 27980 }, { "epoch": 9.710982018518166e-14, "grad_norm": 0.5727447867393494, "learning_rate": 9.559345657279848e-05, "loss": 2.3214, "step": 27990 }, { "epoch": 9.71445146547012e-14, "grad_norm": 0.5990419387817383, "learning_rate": 9.557379695535348e-05, "loss": 2.3641, "step": 28000 }, { "epoch": 9.717920912422073e-14, "grad_norm": 0.5753552913665771, "learning_rate": 9.555413309595585e-05, "loss": 2.3309, "step": 28010 }, { "epoch": 9.721390359374027e-14, "grad_norm": 0.5739324688911438, "learning_rate": 9.553446499771531e-05, "loss": 2.3852, "step": 28020 }, { "epoch": 9.72485980632598e-14, "grad_norm": 0.5447441339492798, "learning_rate": 9.551479266374221e-05, "loss": 2.3487, "step": 28030 }, { "epoch": 9.728329253277934e-14, "grad_norm": 0.5804396271705627, "learning_rate": 9.549511609714754e-05, "loss": 2.3571, "step": 28040 }, { "epoch": 9.731798700229888e-14, "grad_norm": 0.5819075107574463, "learning_rate": 9.547543530104298e-05, "loss": 2.3383, "step": 28050 }, { "epoch": 9.735268147181841e-14, "grad_norm": 0.5684731006622314, "learning_rate": 9.545575027854089e-05, "loss": 2.3773, "step": 28060 }, { "epoch": 9.738737594133795e-14, "grad_norm": 0.5807982087135315, "learning_rate": 9.543606103275426e-05, "loss": 2.3692, "step": 28070 }, { "epoch": 9.742207041085749e-14, "grad_norm": 0.5599341988563538, "learning_rate": 9.54163675667968e-05, "loss": 2.3774, "step": 28080 }, { "epoch": 9.745676488037702e-14, "grad_norm": 0.5603864192962646, "learning_rate": 9.539666988378286e-05, "loss": 2.3623, "step": 28090 }, { "epoch": 9.749145934989656e-14, "grad_norm": 0.579388439655304, "learning_rate": 9.537696798682744e-05, "loss": 2.3531, "step": 28100 }, { "epoch": 9.75261538194161e-14, "grad_norm": 0.5869835615158081, "learning_rate": 9.535726187904627e-05, "loss": 2.3768, "step": 28110 }, { "epoch": 9.756084828893563e-14, "grad_norm": 0.5635539889335632, "learning_rate": 9.533755156355564e-05, "loss": 2.3677, "step": 28120 }, { "epoch": 9.759554275845517e-14, "grad_norm": 0.5730419754981995, "learning_rate": 9.53178370434726e-05, "loss": 2.3677, "step": 28130 }, { "epoch": 9.76302372279747e-14, "grad_norm": 0.5699201822280884, "learning_rate": 9.529811832191486e-05, "loss": 2.3523, "step": 28140 }, { "epoch": 9.766493169749424e-14, "grad_norm": 0.5800976753234863, "learning_rate": 9.527839540200071e-05, "loss": 2.3531, "step": 28150 }, { "epoch": 9.769962616701378e-14, "grad_norm": 0.5700372457504272, "learning_rate": 9.525866828684919e-05, "loss": 2.3131, "step": 28160 }, { "epoch": 9.773432063653331e-14, "grad_norm": 0.5870076417922974, "learning_rate": 9.523893697957996e-05, "loss": 2.342, "step": 28170 }, { "epoch": 9.776901510605285e-14, "grad_norm": 0.5606963038444519, "learning_rate": 9.521920148331336e-05, "loss": 2.367, "step": 28180 }, { "epoch": 9.780370957557238e-14, "grad_norm": 0.5948479771614075, "learning_rate": 9.519946180117042e-05, "loss": 2.3705, "step": 28190 }, { "epoch": 9.783840404509192e-14, "grad_norm": 0.5794645547866821, "learning_rate": 9.517971793627275e-05, "loss": 2.3886, "step": 28200 }, { "epoch": 9.787309851461146e-14, "grad_norm": 0.6151121258735657, "learning_rate": 9.515996989174273e-05, "loss": 2.3408, "step": 28210 }, { "epoch": 9.790779298413099e-14, "grad_norm": 0.6213765740394592, "learning_rate": 9.514021767070327e-05, "loss": 2.3467, "step": 28220 }, { "epoch": 9.794248745365053e-14, "grad_norm": 0.5979655981063843, "learning_rate": 9.512046127627808e-05, "loss": 2.3199, "step": 28230 }, { "epoch": 9.797718192317006e-14, "grad_norm": 0.5912374258041382, "learning_rate": 9.51007007115914e-05, "loss": 2.3829, "step": 28240 }, { "epoch": 9.80118763926896e-14, "grad_norm": 0.5808934569358826, "learning_rate": 9.508093597976828e-05, "loss": 2.376, "step": 28250 }, { "epoch": 9.804657086220914e-14, "grad_norm": 0.5754021406173706, "learning_rate": 9.506116708393427e-05, "loss": 2.3867, "step": 28260 }, { "epoch": 9.808126533172867e-14, "grad_norm": 0.5991708040237427, "learning_rate": 9.504139402721568e-05, "loss": 2.3567, "step": 28270 }, { "epoch": 9.811595980124821e-14, "grad_norm": 0.567314863204956, "learning_rate": 9.502161681273943e-05, "loss": 2.362, "step": 28280 }, { "epoch": 9.815065427076775e-14, "grad_norm": 0.5805442333221436, "learning_rate": 9.500183544363315e-05, "loss": 2.3435, "step": 28290 }, { "epoch": 9.818534874028728e-14, "grad_norm": 0.5939908623695374, "learning_rate": 9.498204992302506e-05, "loss": 2.3543, "step": 28300 }, { "epoch": 9.822004320980682e-14, "grad_norm": 0.579798698425293, "learning_rate": 9.496226025404407e-05, "loss": 2.3545, "step": 28310 }, { "epoch": 9.825473767932635e-14, "grad_norm": 0.5657044649124146, "learning_rate": 9.494246643981978e-05, "loss": 2.3501, "step": 28320 }, { "epoch": 9.828943214884589e-14, "grad_norm": 0.5765859484672546, "learning_rate": 9.492266848348237e-05, "loss": 2.3736, "step": 28330 }, { "epoch": 9.832412661836543e-14, "grad_norm": 0.5684407949447632, "learning_rate": 9.490286638816273e-05, "loss": 2.3645, "step": 28340 }, { "epoch": 9.835882108788496e-14, "grad_norm": 0.591932475566864, "learning_rate": 9.488306015699242e-05, "loss": 2.3675, "step": 28350 }, { "epoch": 9.83935155574045e-14, "grad_norm": 0.5881394147872925, "learning_rate": 9.486324979310358e-05, "loss": 2.3028, "step": 28360 }, { "epoch": 9.842821002692403e-14, "grad_norm": 0.5909896492958069, "learning_rate": 9.484343529962909e-05, "loss": 2.341, "step": 28370 }, { "epoch": 9.846290449644357e-14, "grad_norm": 0.5661938190460205, "learning_rate": 9.482361667970242e-05, "loss": 2.3599, "step": 28380 }, { "epoch": 9.849759896596311e-14, "grad_norm": 0.609021782875061, "learning_rate": 9.48037939364577e-05, "loss": 2.3486, "step": 28390 }, { "epoch": 9.853229343548264e-14, "grad_norm": 0.5711994171142578, "learning_rate": 9.478396707302975e-05, "loss": 2.3374, "step": 28400 }, { "epoch": 9.856698790500218e-14, "grad_norm": 0.5709974765777588, "learning_rate": 9.476413609255399e-05, "loss": 2.3705, "step": 28410 }, { "epoch": 9.860168237452172e-14, "grad_norm": 0.5695263147354126, "learning_rate": 9.474430099816656e-05, "loss": 2.3569, "step": 28420 }, { "epoch": 9.863637684404125e-14, "grad_norm": 0.5821710228919983, "learning_rate": 9.47244617930042e-05, "loss": 2.3422, "step": 28430 }, { "epoch": 9.867107131356079e-14, "grad_norm": 0.6006764769554138, "learning_rate": 9.470461848020429e-05, "loss": 2.3874, "step": 28440 }, { "epoch": 9.870576578308032e-14, "grad_norm": 0.5640271306037903, "learning_rate": 9.468477106290487e-05, "loss": 2.3768, "step": 28450 }, { "epoch": 9.874046025259986e-14, "grad_norm": 0.5820425152778625, "learning_rate": 9.466491954424465e-05, "loss": 2.3739, "step": 28460 }, { "epoch": 9.87751547221194e-14, "grad_norm": 0.5680769085884094, "learning_rate": 9.4645063927363e-05, "loss": 2.345, "step": 28470 }, { "epoch": 9.880984919163893e-14, "grad_norm": 0.5899831056594849, "learning_rate": 9.462520421539988e-05, "loss": 2.317, "step": 28480 }, { "epoch": 9.884454366115847e-14, "grad_norm": 0.5975916981697083, "learning_rate": 9.460534041149595e-05, "loss": 2.3583, "step": 28490 }, { "epoch": 9.8879238130678e-14, "grad_norm": 0.5845494270324707, "learning_rate": 9.45854725187925e-05, "loss": 2.3551, "step": 28500 }, { "epoch": 9.891393260019754e-14, "grad_norm": 0.5637778043746948, "learning_rate": 9.456560054043148e-05, "loss": 2.3436, "step": 28510 }, { "epoch": 9.894862706971708e-14, "grad_norm": 0.6088965535163879, "learning_rate": 9.454572447955545e-05, "loss": 2.375, "step": 28520 }, { "epoch": 9.898332153923661e-14, "grad_norm": 0.5805349349975586, "learning_rate": 9.452584433930764e-05, "loss": 2.3593, "step": 28530 }, { "epoch": 9.901801600875615e-14, "grad_norm": 0.5648502707481384, "learning_rate": 9.450596012283192e-05, "loss": 2.3559, "step": 28540 }, { "epoch": 9.905271047827569e-14, "grad_norm": 0.5721808671951294, "learning_rate": 9.448607183327283e-05, "loss": 2.3879, "step": 28550 }, { "epoch": 9.908740494779522e-14, "grad_norm": 0.6036391854286194, "learning_rate": 9.446617947377551e-05, "loss": 2.3814, "step": 28560 }, { "epoch": 9.912209941731476e-14, "grad_norm": 0.5935460925102234, "learning_rate": 9.444628304748577e-05, "loss": 2.3458, "step": 28570 }, { "epoch": 9.915679388683429e-14, "grad_norm": 0.5807209610939026, "learning_rate": 9.442638255755007e-05, "loss": 2.3144, "step": 28580 }, { "epoch": 9.919148835635383e-14, "grad_norm": 0.5798979997634888, "learning_rate": 9.440647800711549e-05, "loss": 2.3859, "step": 28590 }, { "epoch": 9.922618282587337e-14, "grad_norm": 0.6180003881454468, "learning_rate": 9.438656939932977e-05, "loss": 2.382, "step": 28600 }, { "epoch": 9.92608772953929e-14, "grad_norm": 0.5989835262298584, "learning_rate": 9.436665673734126e-05, "loss": 2.3372, "step": 28610 }, { "epoch": 9.929557176491244e-14, "grad_norm": 0.6025972962379456, "learning_rate": 9.434674002429901e-05, "loss": 2.3713, "step": 28620 }, { "epoch": 9.933026623443197e-14, "grad_norm": 0.5693135261535645, "learning_rate": 9.432681926335267e-05, "loss": 2.3634, "step": 28630 }, { "epoch": 9.936496070395151e-14, "grad_norm": 0.5647591352462769, "learning_rate": 9.430689445765252e-05, "loss": 2.3677, "step": 28640 }, { "epoch": 9.939965517347105e-14, "grad_norm": 0.5733875036239624, "learning_rate": 9.42869656103495e-05, "loss": 2.3357, "step": 28650 }, { "epoch": 9.943434964299058e-14, "grad_norm": 0.5685036778450012, "learning_rate": 9.426703272459521e-05, "loss": 2.3802, "step": 28660 }, { "epoch": 9.946904411251012e-14, "grad_norm": 0.572609543800354, "learning_rate": 9.424709580354185e-05, "loss": 2.3526, "step": 28670 }, { "epoch": 9.950373858202965e-14, "grad_norm": 0.5870956778526306, "learning_rate": 9.422715485034226e-05, "loss": 2.3406, "step": 28680 }, { "epoch": 9.953843305154919e-14, "grad_norm": 0.5740899443626404, "learning_rate": 9.420720986814993e-05, "loss": 2.3435, "step": 28690 }, { "epoch": 9.957312752106873e-14, "grad_norm": 0.5445432662963867, "learning_rate": 9.418726086011901e-05, "loss": 2.3417, "step": 28700 }, { "epoch": 9.960782199058826e-14, "grad_norm": 0.5698163509368896, "learning_rate": 9.416730782940423e-05, "loss": 2.3575, "step": 28710 }, { "epoch": 9.96425164601078e-14, "grad_norm": 0.5602871179580688, "learning_rate": 9.414735077916104e-05, "loss": 2.3321, "step": 28720 }, { "epoch": 9.967721092962734e-14, "grad_norm": 0.5973138213157654, "learning_rate": 9.412738971254543e-05, "loss": 2.3799, "step": 28730 }, { "epoch": 9.971190539914687e-14, "grad_norm": 0.582879900932312, "learning_rate": 9.410742463271408e-05, "loss": 2.3362, "step": 28740 }, { "epoch": 9.974659986866641e-14, "grad_norm": 0.5859866738319397, "learning_rate": 9.408745554282431e-05, "loss": 2.3789, "step": 28750 }, { "epoch": 9.978129433818594e-14, "grad_norm": 0.5716578960418701, "learning_rate": 9.406748244603407e-05, "loss": 2.3244, "step": 28760 }, { "epoch": 9.981598880770548e-14, "grad_norm": 0.6018020510673523, "learning_rate": 9.40475053455019e-05, "loss": 2.3555, "step": 28770 }, { "epoch": 9.985068327722502e-14, "grad_norm": 0.5790575742721558, "learning_rate": 9.402752424438701e-05, "loss": 2.3766, "step": 28780 }, { "epoch": 9.988537774674455e-14, "grad_norm": 0.5796124935150146, "learning_rate": 9.400753914584927e-05, "loss": 2.3691, "step": 28790 }, { "epoch": 9.992007221626409e-14, "grad_norm": 0.5955498218536377, "learning_rate": 9.398755005304911e-05, "loss": 2.3599, "step": 28800 }, { "epoch": 9.995476668578362e-14, "grad_norm": 0.5634669661521912, "learning_rate": 9.396755696914769e-05, "loss": 2.3923, "step": 28810 }, { "epoch": 9.998946115530316e-14, "grad_norm": 0.5919613838195801, "learning_rate": 9.394755989730667e-05, "loss": 2.3673, "step": 28820 }, { "epoch": 1.000241556248227e-13, "grad_norm": 0.595602810382843, "learning_rate": 9.392755884068848e-05, "loss": 2.3688, "step": 28830 }, { "epoch": 1.0005885009434223e-13, "grad_norm": 0.5964133739471436, "learning_rate": 9.390755380245608e-05, "loss": 2.3683, "step": 28840 }, { "epoch": 1.0009354456386177e-13, "grad_norm": 0.586860716342926, "learning_rate": 9.388754478577312e-05, "loss": 2.3461, "step": 28850 }, { "epoch": 1.001282390333813e-13, "grad_norm": 0.5916602611541748, "learning_rate": 9.386753179380382e-05, "loss": 2.3598, "step": 28860 }, { "epoch": 1.0016293350290084e-13, "grad_norm": 0.5989728569984436, "learning_rate": 9.384751482971308e-05, "loss": 2.3719, "step": 28870 }, { "epoch": 1.0019762797242038e-13, "grad_norm": 0.6144014596939087, "learning_rate": 9.382749389666643e-05, "loss": 2.3715, "step": 28880 }, { "epoch": 1.0023232244193991e-13, "grad_norm": 0.5536962151527405, "learning_rate": 9.380746899782997e-05, "loss": 2.3067, "step": 28890 }, { "epoch": 1.0026701691145945e-13, "grad_norm": 0.5847498178482056, "learning_rate": 9.378744013637048e-05, "loss": 2.3713, "step": 28900 }, { "epoch": 1.0030171138097899e-13, "grad_norm": 0.5745286345481873, "learning_rate": 9.376740731545537e-05, "loss": 2.3625, "step": 28910 }, { "epoch": 1.0033640585049852e-13, "grad_norm": 0.5817280411720276, "learning_rate": 9.374737053825265e-05, "loss": 2.3706, "step": 28920 }, { "epoch": 1.0037110032001806e-13, "grad_norm": 0.591966986656189, "learning_rate": 9.372732980793093e-05, "loss": 2.3518, "step": 28930 }, { "epoch": 1.004057947895376e-13, "grad_norm": 0.576919436454773, "learning_rate": 9.370728512765954e-05, "loss": 2.3414, "step": 28940 }, { "epoch": 1.0044048925905713e-13, "grad_norm": 0.6146889328956604, "learning_rate": 9.368723650060832e-05, "loss": 2.3784, "step": 28950 }, { "epoch": 1.0047518372857667e-13, "grad_norm": 0.5870780348777771, "learning_rate": 9.36671839299478e-05, "loss": 2.3854, "step": 28960 }, { "epoch": 1.005098781980962e-13, "grad_norm": 0.5877739191055298, "learning_rate": 9.364712741884916e-05, "loss": 2.377, "step": 28970 }, { "epoch": 1.0054457266761574e-13, "grad_norm": 0.6395822763442993, "learning_rate": 9.362706697048411e-05, "loss": 2.3744, "step": 28980 }, { "epoch": 1.0057926713713528e-13, "grad_norm": 0.5696932077407837, "learning_rate": 9.360700258802506e-05, "loss": 2.3106, "step": 28990 }, { "epoch": 1.0061396160665481e-13, "grad_norm": 0.5591946244239807, "learning_rate": 9.358693427464504e-05, "loss": 2.3541, "step": 29000 }, { "epoch": 1.0064865607617435e-13, "grad_norm": 0.5812931656837463, "learning_rate": 9.356686203351767e-05, "loss": 2.3452, "step": 29010 }, { "epoch": 1.0068335054569388e-13, "grad_norm": 0.5560579299926758, "learning_rate": 9.354678586781718e-05, "loss": 2.3562, "step": 29020 }, { "epoch": 1.0071804501521342e-13, "grad_norm": 0.5915319323539734, "learning_rate": 9.352670578071844e-05, "loss": 2.3423, "step": 29030 }, { "epoch": 1.0075273948473296e-13, "grad_norm": 0.5898007154464722, "learning_rate": 9.350662177539698e-05, "loss": 2.3718, "step": 29040 }, { "epoch": 1.0078743395425249e-13, "grad_norm": 0.5776564478874207, "learning_rate": 9.348653385502889e-05, "loss": 2.3325, "step": 29050 }, { "epoch": 1.0082212842377203e-13, "grad_norm": 0.5753403902053833, "learning_rate": 9.346644202279089e-05, "loss": 2.3464, "step": 29060 }, { "epoch": 1.0085682289329156e-13, "grad_norm": 0.6309239268302917, "learning_rate": 9.344634628186036e-05, "loss": 2.3699, "step": 29070 }, { "epoch": 1.008915173628111e-13, "grad_norm": 0.63422691822052, "learning_rate": 9.342624663541527e-05, "loss": 2.3459, "step": 29080 }, { "epoch": 1.0092621183233064e-13, "grad_norm": 0.590779185295105, "learning_rate": 9.340614308663415e-05, "loss": 2.3423, "step": 29090 }, { "epoch": 1.0096090630185017e-13, "grad_norm": 0.5801606774330139, "learning_rate": 9.338603563869628e-05, "loss": 2.3806, "step": 29100 }, { "epoch": 1.0099560077136971e-13, "grad_norm": 0.597601592540741, "learning_rate": 9.336592429478143e-05, "loss": 2.352, "step": 29110 }, { "epoch": 1.0103029524088925e-13, "grad_norm": 0.5855535268783569, "learning_rate": 9.334580905807002e-05, "loss": 2.3236, "step": 29120 }, { "epoch": 1.0106498971040878e-13, "grad_norm": 0.5751462578773499, "learning_rate": 9.332568993174316e-05, "loss": 2.3672, "step": 29130 }, { "epoch": 1.0109968417992832e-13, "grad_norm": 0.5945692658424377, "learning_rate": 9.330556691898248e-05, "loss": 2.3448, "step": 29140 }, { "epoch": 1.0113437864944785e-13, "grad_norm": 0.5829809904098511, "learning_rate": 9.328544002297026e-05, "loss": 2.3461, "step": 29150 }, { "epoch": 1.0116907311896739e-13, "grad_norm": 0.5835426449775696, "learning_rate": 9.326530924688942e-05, "loss": 2.3275, "step": 29160 }, { "epoch": 1.0120376758848693e-13, "grad_norm": 0.5905778408050537, "learning_rate": 9.324517459392344e-05, "loss": 2.377, "step": 29170 }, { "epoch": 1.0123846205800646e-13, "grad_norm": 0.5702304840087891, "learning_rate": 9.322503606725644e-05, "loss": 2.3673, "step": 29180 }, { "epoch": 1.01273156527526e-13, "grad_norm": 0.5844725966453552, "learning_rate": 9.32048936700732e-05, "loss": 2.3275, "step": 29190 }, { "epoch": 1.0130785099704553e-13, "grad_norm": 0.5855835676193237, "learning_rate": 9.318474740555901e-05, "loss": 2.3596, "step": 29200 }, { "epoch": 1.0134254546656507e-13, "grad_norm": 0.5885027647018433, "learning_rate": 9.316459727689986e-05, "loss": 2.3513, "step": 29210 }, { "epoch": 1.0137723993608461e-13, "grad_norm": 0.5901160836219788, "learning_rate": 9.314444328728232e-05, "loss": 2.3374, "step": 29220 }, { "epoch": 1.0141193440560414e-13, "grad_norm": 0.5759366154670715, "learning_rate": 9.312428543989355e-05, "loss": 2.3516, "step": 29230 }, { "epoch": 1.0144662887512368e-13, "grad_norm": 0.5766516327857971, "learning_rate": 9.310412373792138e-05, "loss": 2.3573, "step": 29240 }, { "epoch": 1.0148132334464322e-13, "grad_norm": 0.5886015295982361, "learning_rate": 9.308395818455415e-05, "loss": 2.3742, "step": 29250 }, { "epoch": 1.0151601781416275e-13, "grad_norm": 0.5962815880775452, "learning_rate": 9.306378878298091e-05, "loss": 2.3645, "step": 29260 }, { "epoch": 1.0155071228368229e-13, "grad_norm": 0.579572319984436, "learning_rate": 9.304361553639129e-05, "loss": 2.3299, "step": 29270 }, { "epoch": 1.0158540675320182e-13, "grad_norm": 0.597817599773407, "learning_rate": 9.302343844797548e-05, "loss": 2.361, "step": 29280 }, { "epoch": 1.0162010122272136e-13, "grad_norm": 0.5877052545547485, "learning_rate": 9.300325752092431e-05, "loss": 2.3753, "step": 29290 }, { "epoch": 1.016547956922409e-13, "grad_norm": 0.5759891867637634, "learning_rate": 9.298307275842923e-05, "loss": 2.3633, "step": 29300 }, { "epoch": 1.0168949016176043e-13, "grad_norm": 0.5805674195289612, "learning_rate": 9.296288416368233e-05, "loss": 2.3566, "step": 29310 }, { "epoch": 1.0172418463127997e-13, "grad_norm": 0.6074081659317017, "learning_rate": 9.294269173987623e-05, "loss": 2.3598, "step": 29320 }, { "epoch": 1.017588791007995e-13, "grad_norm": 0.5659328699111938, "learning_rate": 9.292249549020415e-05, "loss": 2.3517, "step": 29330 }, { "epoch": 1.0179357357031904e-13, "grad_norm": 0.6011695265769958, "learning_rate": 9.290229541786001e-05, "loss": 2.3883, "step": 29340 }, { "epoch": 1.0182826803983858e-13, "grad_norm": 0.5982118844985962, "learning_rate": 9.288209152603821e-05, "loss": 2.386, "step": 29350 }, { "epoch": 1.0186296250935811e-13, "grad_norm": 0.5899274945259094, "learning_rate": 9.286188381793392e-05, "loss": 2.3629, "step": 29360 }, { "epoch": 1.0189765697887765e-13, "grad_norm": 0.5650566220283508, "learning_rate": 9.284167229674273e-05, "loss": 2.3821, "step": 29370 }, { "epoch": 1.0193235144839718e-13, "grad_norm": 0.5920373201370239, "learning_rate": 9.282145696566095e-05, "loss": 2.3785, "step": 29380 }, { "epoch": 1.0196704591791672e-13, "grad_norm": 0.5602938532829285, "learning_rate": 9.280123782788547e-05, "loss": 2.3512, "step": 29390 }, { "epoch": 1.0200174038743626e-13, "grad_norm": 0.5854268670082092, "learning_rate": 9.278101488661378e-05, "loss": 2.3645, "step": 29400 }, { "epoch": 1.0203643485695579e-13, "grad_norm": 0.5830360054969788, "learning_rate": 9.276078814504391e-05, "loss": 2.3439, "step": 29410 }, { "epoch": 1.0207112932647533e-13, "grad_norm": 0.6071446537971497, "learning_rate": 9.27405576063746e-05, "loss": 2.3549, "step": 29420 }, { "epoch": 1.0210582379599487e-13, "grad_norm": 0.5610299110412598, "learning_rate": 9.272032327380512e-05, "loss": 2.3308, "step": 29430 }, { "epoch": 1.021405182655144e-13, "grad_norm": 0.5651683807373047, "learning_rate": 9.270008515053534e-05, "loss": 2.3285, "step": 29440 }, { "epoch": 1.0217521273503394e-13, "grad_norm": 0.5878199934959412, "learning_rate": 9.267984323976576e-05, "loss": 2.3279, "step": 29450 }, { "epoch": 1.0220990720455347e-13, "grad_norm": 0.6335925459861755, "learning_rate": 9.265959754469745e-05, "loss": 2.3623, "step": 29460 }, { "epoch": 1.0224460167407301e-13, "grad_norm": 0.5903370976448059, "learning_rate": 9.263934806853212e-05, "loss": 2.3819, "step": 29470 }, { "epoch": 1.0227929614359255e-13, "grad_norm": 0.5807398557662964, "learning_rate": 9.261909481447204e-05, "loss": 2.3439, "step": 29480 }, { "epoch": 1.0231399061311208e-13, "grad_norm": 0.6066139340400696, "learning_rate": 9.259883778572008e-05, "loss": 2.3736, "step": 29490 }, { "epoch": 1.0234868508263162e-13, "grad_norm": 0.6177418231964111, "learning_rate": 9.257857698547972e-05, "loss": 2.311, "step": 29500 }, { "epoch": 1.0238337955215115e-13, "grad_norm": 0.5718498229980469, "learning_rate": 9.255831241695501e-05, "loss": 2.3903, "step": 29510 }, { "epoch": 1.0241807402167069e-13, "grad_norm": 0.5614590644836426, "learning_rate": 9.253804408335062e-05, "loss": 2.3521, "step": 29520 }, { "epoch": 1.0245276849119023e-13, "grad_norm": 0.5935940742492676, "learning_rate": 9.251777198787186e-05, "loss": 2.33, "step": 29530 }, { "epoch": 1.0248746296070976e-13, "grad_norm": 0.6043438911437988, "learning_rate": 9.249749613372455e-05, "loss": 2.3723, "step": 29540 }, { "epoch": 1.025221574302293e-13, "grad_norm": 0.5965611934661865, "learning_rate": 9.247721652411514e-05, "loss": 2.336, "step": 29550 }, { "epoch": 1.0255685189974884e-13, "grad_norm": 0.610763430595398, "learning_rate": 9.245693316225069e-05, "loss": 2.3708, "step": 29560 }, { "epoch": 1.0259154636926837e-13, "grad_norm": 0.5803080797195435, "learning_rate": 9.24366460513388e-05, "loss": 2.3235, "step": 29570 }, { "epoch": 1.0262624083878791e-13, "grad_norm": 0.5738401412963867, "learning_rate": 9.241635519458778e-05, "loss": 2.3866, "step": 29580 }, { "epoch": 1.0266093530830744e-13, "grad_norm": 0.5859739184379578, "learning_rate": 9.239606059520637e-05, "loss": 2.3185, "step": 29590 }, { "epoch": 1.0269562977782698e-13, "grad_norm": 0.6109182834625244, "learning_rate": 9.237576225640403e-05, "loss": 2.3671, "step": 29600 }, { "epoch": 1.0273032424734652e-13, "grad_norm": 0.5693517923355103, "learning_rate": 9.235546018139076e-05, "loss": 2.3745, "step": 29610 }, { "epoch": 1.0276501871686605e-13, "grad_norm": 0.5761893391609192, "learning_rate": 9.233515437337716e-05, "loss": 2.3296, "step": 29620 }, { "epoch": 1.0279971318638559e-13, "grad_norm": 0.6115656495094299, "learning_rate": 9.231484483557445e-05, "loss": 2.3235, "step": 29630 }, { "epoch": 1.0283440765590512e-13, "grad_norm": 0.6088634729385376, "learning_rate": 9.229453157119435e-05, "loss": 2.3323, "step": 29640 }, { "epoch": 1.0286910212542466e-13, "grad_norm": 0.606875479221344, "learning_rate": 9.227421458344925e-05, "loss": 2.322, "step": 29650 }, { "epoch": 1.029037965949442e-13, "grad_norm": 0.6023401021957397, "learning_rate": 9.225389387555212e-05, "loss": 2.3556, "step": 29660 }, { "epoch": 1.0293849106446373e-13, "grad_norm": 0.588534951210022, "learning_rate": 9.223356945071649e-05, "loss": 2.3817, "step": 29670 }, { "epoch": 1.0297318553398327e-13, "grad_norm": 0.5848342180252075, "learning_rate": 9.22132413121565e-05, "loss": 2.3483, "step": 29680 }, { "epoch": 1.030078800035028e-13, "grad_norm": 0.5601363182067871, "learning_rate": 9.21929094630869e-05, "loss": 2.3029, "step": 29690 }, { "epoch": 1.0304257447302234e-13, "grad_norm": 0.5894670486450195, "learning_rate": 9.217257390672292e-05, "loss": 2.3706, "step": 29700 }, { "epoch": 1.0307726894254188e-13, "grad_norm": 0.6000224351882935, "learning_rate": 9.215223464628054e-05, "loss": 2.3646, "step": 29710 }, { "epoch": 1.0311196341206141e-13, "grad_norm": 0.5677781105041504, "learning_rate": 9.213189168497617e-05, "loss": 2.3644, "step": 29720 }, { "epoch": 1.0314665788158095e-13, "grad_norm": 0.6016218066215515, "learning_rate": 9.211154502602691e-05, "loss": 2.3553, "step": 29730 }, { "epoch": 1.0318135235110049e-13, "grad_norm": 0.6066521406173706, "learning_rate": 9.209119467265039e-05, "loss": 2.3164, "step": 29740 }, { "epoch": 1.0321604682062002e-13, "grad_norm": 0.551516056060791, "learning_rate": 9.207084062806488e-05, "loss": 2.3123, "step": 29750 }, { "epoch": 1.0325074129013956e-13, "grad_norm": 0.586746096611023, "learning_rate": 9.205048289548914e-05, "loss": 2.3977, "step": 29760 }, { "epoch": 1.032854357596591e-13, "grad_norm": 0.5753178000450134, "learning_rate": 9.203012147814263e-05, "loss": 2.3611, "step": 29770 }, { "epoch": 1.0332013022917863e-13, "grad_norm": 0.5851069092750549, "learning_rate": 9.200975637924531e-05, "loss": 2.377, "step": 29780 }, { "epoch": 1.0335482469869817e-13, "grad_norm": 0.5809356570243835, "learning_rate": 9.19893876020177e-05, "loss": 2.3665, "step": 29790 }, { "epoch": 1.033895191682177e-13, "grad_norm": 0.5639350414276123, "learning_rate": 9.196901514968103e-05, "loss": 2.301, "step": 29800 }, { "epoch": 1.0342421363773724e-13, "grad_norm": 0.5706053972244263, "learning_rate": 9.194863902545697e-05, "loss": 2.335, "step": 29810 }, { "epoch": 1.0345890810725678e-13, "grad_norm": 0.585981011390686, "learning_rate": 9.192825923256786e-05, "loss": 2.3597, "step": 29820 }, { "epoch": 1.0349360257677631e-13, "grad_norm": 0.628234326839447, "learning_rate": 9.190787577423654e-05, "loss": 2.3823, "step": 29830 }, { "epoch": 1.0352829704629585e-13, "grad_norm": 0.5800325870513916, "learning_rate": 9.188748865368651e-05, "loss": 2.3591, "step": 29840 }, { "epoch": 1.0356299151581538e-13, "grad_norm": 0.5806323885917664, "learning_rate": 9.186709787414184e-05, "loss": 2.3358, "step": 29850 }, { "epoch": 1.0359768598533492e-13, "grad_norm": 0.5706075429916382, "learning_rate": 9.184670343882711e-05, "loss": 2.3924, "step": 29860 }, { "epoch": 1.0363238045485446e-13, "grad_norm": 0.56962651014328, "learning_rate": 9.182630535096759e-05, "loss": 2.3458, "step": 29870 }, { "epoch": 1.0366707492437399e-13, "grad_norm": 0.5865389704704285, "learning_rate": 9.180590361378897e-05, "loss": 2.3531, "step": 29880 }, { "epoch": 1.0370176939389353e-13, "grad_norm": 0.5645785927772522, "learning_rate": 9.178549823051769e-05, "loss": 2.3954, "step": 29890 }, { "epoch": 1.0373646386341306e-13, "grad_norm": 0.5819410681724548, "learning_rate": 9.176508920438066e-05, "loss": 2.3211, "step": 29900 }, { "epoch": 1.037711583329326e-13, "grad_norm": 0.5969740152359009, "learning_rate": 9.174467653860536e-05, "loss": 2.3543, "step": 29910 }, { "epoch": 1.0380585280245214e-13, "grad_norm": 0.6031556129455566, "learning_rate": 9.172426023641994e-05, "loss": 2.3264, "step": 29920 }, { "epoch": 1.0384054727197167e-13, "grad_norm": 0.6587806344032288, "learning_rate": 9.170384030105302e-05, "loss": 2.3648, "step": 29930 }, { "epoch": 1.0387524174149121e-13, "grad_norm": 0.5714554786682129, "learning_rate": 9.168341673573382e-05, "loss": 2.3363, "step": 29940 }, { "epoch": 1.0390993621101074e-13, "grad_norm": 0.7571824193000793, "learning_rate": 9.16629895436922e-05, "loss": 2.3587, "step": 29950 }, { "epoch": 1.0394463068053028e-13, "grad_norm": 0.5812775492668152, "learning_rate": 9.164255872815854e-05, "loss": 2.2696, "step": 29960 }, { "epoch": 1.0397932515004982e-13, "grad_norm": 0.6101423501968384, "learning_rate": 9.162212429236373e-05, "loss": 2.3466, "step": 29970 }, { "epoch": 1.0401401961956935e-13, "grad_norm": 0.6341891884803772, "learning_rate": 9.160168623953938e-05, "loss": 2.3335, "step": 29980 }, { "epoch": 1.0404871408908889e-13, "grad_norm": 0.5540817975997925, "learning_rate": 9.158124457291754e-05, "loss": 2.3385, "step": 29990 }, { "epoch": 1.0408340855860843e-13, "grad_norm": 0.5984048843383789, "learning_rate": 9.156079929573092e-05, "loss": 2.328, "step": 30000 }, { "epoch": 1.0411810302812796e-13, "grad_norm": 0.5993669629096985, "learning_rate": 9.154035041121272e-05, "loss": 2.3372, "step": 30010 }, { "epoch": 1.041527974976475e-13, "grad_norm": 0.5818987488746643, "learning_rate": 9.151989792259681e-05, "loss": 2.3462, "step": 30020 }, { "epoch": 1.0418749196716703e-13, "grad_norm": 0.5737876296043396, "learning_rate": 9.149944183311755e-05, "loss": 2.3837, "step": 30030 }, { "epoch": 1.0422218643668657e-13, "grad_norm": 0.602026104927063, "learning_rate": 9.147898214600987e-05, "loss": 2.3836, "step": 30040 }, { "epoch": 1.0425688090620611e-13, "grad_norm": 0.5764596462249756, "learning_rate": 9.145851886450933e-05, "loss": 2.3786, "step": 30050 }, { "epoch": 1.0429157537572564e-13, "grad_norm": 0.5881123542785645, "learning_rate": 9.143805199185201e-05, "loss": 2.3883, "step": 30060 }, { "epoch": 1.0432626984524518e-13, "grad_norm": 0.5358340740203857, "learning_rate": 9.141758153127456e-05, "loss": 2.3604, "step": 30070 }, { "epoch": 1.0436096431476471e-13, "grad_norm": 0.5957754850387573, "learning_rate": 9.139710748601422e-05, "loss": 2.3704, "step": 30080 }, { "epoch": 1.0439565878428425e-13, "grad_norm": 0.6084723472595215, "learning_rate": 9.137662985930878e-05, "loss": 2.3771, "step": 30090 }, { "epoch": 1.0443035325380379e-13, "grad_norm": 0.5587078928947449, "learning_rate": 9.13561486543966e-05, "loss": 2.3031, "step": 30100 }, { "epoch": 1.0446504772332332e-13, "grad_norm": 0.585382878780365, "learning_rate": 9.13356638745166e-05, "loss": 2.3432, "step": 30110 }, { "epoch": 1.0449974219284286e-13, "grad_norm": 0.5795041918754578, "learning_rate": 9.131517552290829e-05, "loss": 2.3679, "step": 30120 }, { "epoch": 1.045344366623624e-13, "grad_norm": 0.5795407891273499, "learning_rate": 9.129468360281171e-05, "loss": 2.3479, "step": 30130 }, { "epoch": 1.0456913113188193e-13, "grad_norm": 0.5848731398582458, "learning_rate": 9.127418811746748e-05, "loss": 2.3386, "step": 30140 }, { "epoch": 1.0460382560140147e-13, "grad_norm": 0.5772559642791748, "learning_rate": 9.125368907011678e-05, "loss": 2.3641, "step": 30150 }, { "epoch": 1.04638520070921e-13, "grad_norm": 0.5995484590530396, "learning_rate": 9.123318646400134e-05, "loss": 2.3723, "step": 30160 }, { "epoch": 1.0467321454044054e-13, "grad_norm": 0.5947049856185913, "learning_rate": 9.121268030236355e-05, "loss": 2.3611, "step": 30170 }, { "epoch": 1.0470790900996008e-13, "grad_norm": 0.6182209849357605, "learning_rate": 9.119217058844622e-05, "loss": 2.3205, "step": 30180 }, { "epoch": 1.0474260347947961e-13, "grad_norm": 0.5693410634994507, "learning_rate": 9.117165732549278e-05, "loss": 2.3728, "step": 30190 }, { "epoch": 1.0477729794899915e-13, "grad_norm": 0.5922750234603882, "learning_rate": 9.115114051674724e-05, "loss": 2.3477, "step": 30200 }, { "epoch": 1.0481199241851868e-13, "grad_norm": 0.6340181827545166, "learning_rate": 9.113062016545416e-05, "loss": 2.3168, "step": 30210 }, { "epoch": 1.0484668688803822e-13, "grad_norm": 0.6070259213447571, "learning_rate": 9.111009627485866e-05, "loss": 2.3535, "step": 30220 }, { "epoch": 1.0488138135755776e-13, "grad_norm": 0.5639597177505493, "learning_rate": 9.108956884820641e-05, "loss": 2.3252, "step": 30230 }, { "epoch": 1.0491607582707729e-13, "grad_norm": 0.5895862579345703, "learning_rate": 9.106903788874366e-05, "loss": 2.3412, "step": 30240 }, { "epoch": 1.0495077029659683e-13, "grad_norm": 0.6062889695167542, "learning_rate": 9.104850339971716e-05, "loss": 2.3394, "step": 30250 }, { "epoch": 1.0498546476611637e-13, "grad_norm": 0.5941148996353149, "learning_rate": 9.102796538437431e-05, "loss": 2.3371, "step": 30260 }, { "epoch": 1.050201592356359e-13, "grad_norm": 0.5911526679992676, "learning_rate": 9.100742384596301e-05, "loss": 2.3175, "step": 30270 }, { "epoch": 1.0505485370515544e-13, "grad_norm": 0.5716204047203064, "learning_rate": 9.098687878773172e-05, "loss": 2.3509, "step": 30280 }, { "epoch": 1.0508954817467497e-13, "grad_norm": 0.5773632526397705, "learning_rate": 9.096633021292947e-05, "loss": 2.3363, "step": 30290 }, { "epoch": 1.0512424264419451e-13, "grad_norm": 0.5931911468505859, "learning_rate": 9.094577812480585e-05, "loss": 2.3393, "step": 30300 }, { "epoch": 1.0515893711371405e-13, "grad_norm": 0.5579511523246765, "learning_rate": 9.092522252661094e-05, "loss": 2.3227, "step": 30310 }, { "epoch": 1.0519363158323358e-13, "grad_norm": 0.5940195322036743, "learning_rate": 9.090466342159549e-05, "loss": 2.3587, "step": 30320 }, { "epoch": 1.0522832605275312e-13, "grad_norm": 0.5771260261535645, "learning_rate": 9.088410081301074e-05, "loss": 2.3215, "step": 30330 }, { "epoch": 1.0526302052227265e-13, "grad_norm": 0.5863942503929138, "learning_rate": 9.086353470410848e-05, "loss": 2.3231, "step": 30340 }, { "epoch": 1.0529771499179219e-13, "grad_norm": 0.5685732960700989, "learning_rate": 9.084296509814105e-05, "loss": 2.3558, "step": 30350 }, { "epoch": 1.0533240946131173e-13, "grad_norm": 0.6237274408340454, "learning_rate": 9.082239199836136e-05, "loss": 2.3345, "step": 30360 }, { "epoch": 1.0536710393083126e-13, "grad_norm": 0.5696809887886047, "learning_rate": 9.080181540802291e-05, "loss": 2.3522, "step": 30370 }, { "epoch": 1.054017984003508e-13, "grad_norm": 0.5893709659576416, "learning_rate": 9.078123533037964e-05, "loss": 2.3409, "step": 30380 }, { "epoch": 1.0543649286987034e-13, "grad_norm": 0.6016566157341003, "learning_rate": 9.076065176868616e-05, "loss": 2.3718, "step": 30390 }, { "epoch": 1.0547118733938987e-13, "grad_norm": 0.589857280254364, "learning_rate": 9.074006472619759e-05, "loss": 2.3659, "step": 30400 }, { "epoch": 1.0550588180890941e-13, "grad_norm": 0.5865256786346436, "learning_rate": 9.071947420616955e-05, "loss": 2.3681, "step": 30410 }, { "epoch": 1.0554057627842894e-13, "grad_norm": 0.6048421263694763, "learning_rate": 9.069888021185828e-05, "loss": 2.3677, "step": 30420 }, { "epoch": 1.0557527074794848e-13, "grad_norm": 0.6006256937980652, "learning_rate": 9.067828274652055e-05, "loss": 2.3101, "step": 30430 }, { "epoch": 1.0560996521746802e-13, "grad_norm": 0.5789134502410889, "learning_rate": 9.065768181341368e-05, "loss": 2.3502, "step": 30440 }, { "epoch": 1.0564465968698755e-13, "grad_norm": 0.5704207420349121, "learning_rate": 9.063707741579549e-05, "loss": 2.3586, "step": 30450 }, { "epoch": 1.0567935415650709e-13, "grad_norm": 0.5834181904792786, "learning_rate": 9.06164695569244e-05, "loss": 2.349, "step": 30460 }, { "epoch": 1.0571404862602662e-13, "grad_norm": 0.5908387899398804, "learning_rate": 9.05958582400594e-05, "loss": 2.3204, "step": 30470 }, { "epoch": 1.0574874309554616e-13, "grad_norm": 0.6040387153625488, "learning_rate": 9.057524346845998e-05, "loss": 2.3043, "step": 30480 }, { "epoch": 1.057834375650657e-13, "grad_norm": 0.5714357495307922, "learning_rate": 9.055462524538616e-05, "loss": 2.3373, "step": 30490 }, { "epoch": 1.0581813203458523e-13, "grad_norm": 0.5556373596191406, "learning_rate": 9.053400357409856e-05, "loss": 2.3105, "step": 30500 }, { "epoch": 1.0585282650410477e-13, "grad_norm": 0.5870271921157837, "learning_rate": 9.051337845785832e-05, "loss": 2.3341, "step": 30510 }, { "epoch": 1.058875209736243e-13, "grad_norm": 0.5887905955314636, "learning_rate": 9.04927498999271e-05, "loss": 2.3486, "step": 30520 }, { "epoch": 1.0592221544314384e-13, "grad_norm": 0.5796964764595032, "learning_rate": 9.047211790356716e-05, "loss": 2.3197, "step": 30530 }, { "epoch": 1.0595690991266338e-13, "grad_norm": 0.5851061940193176, "learning_rate": 9.045148247204126e-05, "loss": 2.3618, "step": 30540 }, { "epoch": 1.0599160438218291e-13, "grad_norm": 0.5913552641868591, "learning_rate": 9.043084360861268e-05, "loss": 2.3329, "step": 30550 }, { "epoch": 1.0602629885170245e-13, "grad_norm": 0.5577903389930725, "learning_rate": 9.041020131654535e-05, "loss": 2.3733, "step": 30560 }, { "epoch": 1.0606099332122199e-13, "grad_norm": 0.557208240032196, "learning_rate": 9.038955559910364e-05, "loss": 2.3057, "step": 30570 }, { "epoch": 1.0609568779074152e-13, "grad_norm": 0.5712530612945557, "learning_rate": 9.036890645955245e-05, "loss": 2.2826, "step": 30580 }, { "epoch": 1.0613038226026106e-13, "grad_norm": 0.5983570218086243, "learning_rate": 9.034825390115734e-05, "loss": 2.3171, "step": 30590 }, { "epoch": 1.061650767297806e-13, "grad_norm": 0.5886116623878479, "learning_rate": 9.032759792718428e-05, "loss": 2.362, "step": 30600 }, { "epoch": 1.0619977119930013e-13, "grad_norm": 0.5948327779769897, "learning_rate": 9.030693854089985e-05, "loss": 2.3712, "step": 30610 }, { "epoch": 1.0623446566881967e-13, "grad_norm": 0.6335188150405884, "learning_rate": 9.028627574557113e-05, "loss": 2.3233, "step": 30620 }, { "epoch": 1.062691601383392e-13, "grad_norm": 0.583411455154419, "learning_rate": 9.026560954446582e-05, "loss": 2.3289, "step": 30630 }, { "epoch": 1.0630385460785874e-13, "grad_norm": 0.5811537504196167, "learning_rate": 9.024493994085204e-05, "loss": 2.3451, "step": 30640 }, { "epoch": 1.0633854907737827e-13, "grad_norm": 0.593433678150177, "learning_rate": 9.022426693799856e-05, "loss": 2.3542, "step": 30650 }, { "epoch": 1.0637324354689781e-13, "grad_norm": 0.5622756481170654, "learning_rate": 9.020359053917461e-05, "loss": 2.3647, "step": 30660 }, { "epoch": 1.0640793801641735e-13, "grad_norm": 0.5671822428703308, "learning_rate": 9.018291074765e-05, "loss": 2.3435, "step": 30670 }, { "epoch": 1.0644263248593688e-13, "grad_norm": 0.6165722012519836, "learning_rate": 9.016222756669502e-05, "loss": 2.3647, "step": 30680 }, { "epoch": 1.0647732695545642e-13, "grad_norm": 0.6312336325645447, "learning_rate": 9.014154099958057e-05, "loss": 2.3662, "step": 30690 }, { "epoch": 1.0651202142497596e-13, "grad_norm": 0.5911554098129272, "learning_rate": 9.012085104957808e-05, "loss": 2.3543, "step": 30700 }, { "epoch": 1.0654671589449549e-13, "grad_norm": 0.6015543937683105, "learning_rate": 9.010015771995946e-05, "loss": 2.3072, "step": 30710 }, { "epoch": 1.0658141036401503e-13, "grad_norm": 0.6194292306900024, "learning_rate": 9.007946101399715e-05, "loss": 2.3529, "step": 30720 }, { "epoch": 1.0661610483353456e-13, "grad_norm": 0.608399510383606, "learning_rate": 9.00587609349642e-05, "loss": 2.3294, "step": 30730 }, { "epoch": 1.066507993030541e-13, "grad_norm": 0.598363995552063, "learning_rate": 9.003805748613412e-05, "loss": 2.3309, "step": 30740 }, { "epoch": 1.0668549377257364e-13, "grad_norm": 0.593966543674469, "learning_rate": 9.001735067078103e-05, "loss": 2.3385, "step": 30750 }, { "epoch": 1.0672018824209317e-13, "grad_norm": 0.5881968140602112, "learning_rate": 8.99966404921795e-05, "loss": 2.3576, "step": 30760 }, { "epoch": 1.0675488271161271e-13, "grad_norm": 0.5956518650054932, "learning_rate": 8.997592695360466e-05, "loss": 2.3671, "step": 30770 }, { "epoch": 1.0678957718113224e-13, "grad_norm": 0.586590588092804, "learning_rate": 8.995521005833215e-05, "loss": 2.3417, "step": 30780 }, { "epoch": 1.0682427165065178e-13, "grad_norm": 0.5952486991882324, "learning_rate": 8.993448980963824e-05, "loss": 2.3571, "step": 30790 }, { "epoch": 1.0685896612017132e-13, "grad_norm": 0.5821198225021362, "learning_rate": 8.99137662107996e-05, "loss": 2.3586, "step": 30800 }, { "epoch": 1.0689366058969085e-13, "grad_norm": 0.5708397030830383, "learning_rate": 8.989303926509355e-05, "loss": 2.3595, "step": 30810 }, { "epoch": 1.0692835505921039e-13, "grad_norm": 0.5954495668411255, "learning_rate": 8.987230897579784e-05, "loss": 2.3498, "step": 30820 }, { "epoch": 1.0696304952872993e-13, "grad_norm": 0.5586220622062683, "learning_rate": 8.985157534619077e-05, "loss": 2.2834, "step": 30830 }, { "epoch": 1.0699774399824946e-13, "grad_norm": 0.5884944796562195, "learning_rate": 8.983083837955119e-05, "loss": 2.3617, "step": 30840 }, { "epoch": 1.07032438467769e-13, "grad_norm": 0.6065415143966675, "learning_rate": 8.98100980791585e-05, "loss": 2.3312, "step": 30850 }, { "epoch": 1.0706713293728853e-13, "grad_norm": 0.6056885123252869, "learning_rate": 8.978935444829259e-05, "loss": 2.346, "step": 30860 }, { "epoch": 1.0710182740680807e-13, "grad_norm": 0.6000056266784668, "learning_rate": 8.976860749023386e-05, "loss": 2.3532, "step": 30870 }, { "epoch": 1.071365218763276e-13, "grad_norm": 0.5999758243560791, "learning_rate": 8.97478572082633e-05, "loss": 2.3478, "step": 30880 }, { "epoch": 1.0717121634584714e-13, "grad_norm": 0.6029338240623474, "learning_rate": 8.972710360566238e-05, "loss": 2.3432, "step": 30890 }, { "epoch": 1.0720591081536668e-13, "grad_norm": 0.5980414152145386, "learning_rate": 8.97063466857131e-05, "loss": 2.3401, "step": 30900 }, { "epoch": 1.0724060528488621e-13, "grad_norm": 0.5728577971458435, "learning_rate": 8.968558645169797e-05, "loss": 2.3749, "step": 30910 }, { "epoch": 1.0727529975440575e-13, "grad_norm": 0.6101979613304138, "learning_rate": 8.966482290690007e-05, "loss": 2.3428, "step": 30920 }, { "epoch": 1.0730999422392529e-13, "grad_norm": 0.6095596551895142, "learning_rate": 8.964405605460295e-05, "loss": 2.3549, "step": 30930 }, { "epoch": 1.0734468869344482e-13, "grad_norm": 0.5913894176483154, "learning_rate": 8.962328589809072e-05, "loss": 2.3697, "step": 30940 }, { "epoch": 1.0737938316296436e-13, "grad_norm": 0.5929524302482605, "learning_rate": 8.960251244064801e-05, "loss": 2.3142, "step": 30950 }, { "epoch": 1.074140776324839e-13, "grad_norm": 0.5692691802978516, "learning_rate": 8.958173568555997e-05, "loss": 2.3234, "step": 30960 }, { "epoch": 1.0744877210200343e-13, "grad_norm": 0.5882210731506348, "learning_rate": 8.956095563611224e-05, "loss": 2.3551, "step": 30970 }, { "epoch": 1.0748346657152297e-13, "grad_norm": 0.6401045918464661, "learning_rate": 8.954017229559102e-05, "loss": 2.3043, "step": 30980 }, { "epoch": 1.075181610410425e-13, "grad_norm": 0.5800861120223999, "learning_rate": 8.951938566728302e-05, "loss": 2.3402, "step": 30990 }, { "epoch": 1.0755285551056204e-13, "grad_norm": 0.6000980138778687, "learning_rate": 8.949859575447544e-05, "loss": 2.3329, "step": 31000 }, { "epoch": 1.0758754998008158e-13, "grad_norm": 0.5892881155014038, "learning_rate": 8.947780256045606e-05, "loss": 2.3697, "step": 31010 }, { "epoch": 1.0762224444960111e-13, "grad_norm": 0.5997738838195801, "learning_rate": 8.945700608851315e-05, "loss": 2.3627, "step": 31020 }, { "epoch": 1.0765693891912065e-13, "grad_norm": 0.5815600156784058, "learning_rate": 8.943620634193544e-05, "loss": 2.3601, "step": 31030 }, { "epoch": 1.0769163338864018e-13, "grad_norm": 0.5916813015937805, "learning_rate": 8.94154033240123e-05, "loss": 2.3293, "step": 31040 }, { "epoch": 1.0772632785815972e-13, "grad_norm": 0.610858142375946, "learning_rate": 8.939459703803349e-05, "loss": 2.3345, "step": 31050 }, { "epoch": 1.0776102232767926e-13, "grad_norm": 0.5760012865066528, "learning_rate": 8.937378748728938e-05, "loss": 2.3164, "step": 31060 }, { "epoch": 1.0779571679719879e-13, "grad_norm": 0.5939036011695862, "learning_rate": 8.935297467507079e-05, "loss": 2.3434, "step": 31070 }, { "epoch": 1.0783041126671833e-13, "grad_norm": 0.5845237374305725, "learning_rate": 8.933215860466912e-05, "loss": 2.3526, "step": 31080 }, { "epoch": 1.0786510573623787e-13, "grad_norm": 0.6111212968826294, "learning_rate": 8.931133927937625e-05, "loss": 2.3434, "step": 31090 }, { "epoch": 1.078998002057574e-13, "grad_norm": 0.5700090527534485, "learning_rate": 8.929051670248454e-05, "loss": 2.3403, "step": 31100 }, { "epoch": 1.0793449467527694e-13, "grad_norm": 0.598056435585022, "learning_rate": 8.926969087728694e-05, "loss": 2.3488, "step": 31110 }, { "epoch": 1.0796918914479647e-13, "grad_norm": 0.60148686170578, "learning_rate": 8.924886180707688e-05, "loss": 2.3594, "step": 31120 }, { "epoch": 1.0800388361431601e-13, "grad_norm": 0.5787108540534973, "learning_rate": 8.922802949514826e-05, "loss": 2.3347, "step": 31130 }, { "epoch": 1.0803857808383555e-13, "grad_norm": 0.57475346326828, "learning_rate": 8.920719394479557e-05, "loss": 2.3771, "step": 31140 }, { "epoch": 1.0807327255335508e-13, "grad_norm": 0.6234696507453918, "learning_rate": 8.918635515931374e-05, "loss": 2.3, "step": 31150 }, { "epoch": 1.0810796702287462e-13, "grad_norm": 0.591560423374176, "learning_rate": 8.916551314199824e-05, "loss": 2.348, "step": 31160 }, { "epoch": 1.0814266149239415e-13, "grad_norm": 0.6416382193565369, "learning_rate": 8.91446678961451e-05, "loss": 2.3622, "step": 31170 }, { "epoch": 1.0817735596191369e-13, "grad_norm": 0.6039542555809021, "learning_rate": 8.91238194250508e-05, "loss": 2.3665, "step": 31180 }, { "epoch": 1.0821205043143323e-13, "grad_norm": 0.5984228849411011, "learning_rate": 8.910296773201232e-05, "loss": 2.3715, "step": 31190 }, { "epoch": 1.0824674490095276e-13, "grad_norm": 0.5825446844100952, "learning_rate": 8.908211282032719e-05, "loss": 2.3566, "step": 31200 }, { "epoch": 1.082814393704723e-13, "grad_norm": 0.5782743096351624, "learning_rate": 8.906125469329344e-05, "loss": 2.3555, "step": 31210 }, { "epoch": 1.0831613383999183e-13, "grad_norm": 0.5951467156410217, "learning_rate": 8.904039335420961e-05, "loss": 2.3576, "step": 31220 }, { "epoch": 1.0835082830951137e-13, "grad_norm": 0.6685452461242676, "learning_rate": 8.901952880637472e-05, "loss": 2.3465, "step": 31230 }, { "epoch": 1.0838552277903091e-13, "grad_norm": 0.5631248950958252, "learning_rate": 8.899866105308834e-05, "loss": 2.3396, "step": 31240 }, { "epoch": 1.0842021724855044e-13, "grad_norm": 0.5736950635910034, "learning_rate": 8.897779009765053e-05, "loss": 2.3334, "step": 31250 }, { "epoch": 1.0845491171806998e-13, "grad_norm": 0.5873597860336304, "learning_rate": 8.89569159433618e-05, "loss": 2.3165, "step": 31260 }, { "epoch": 1.0848960618758952e-13, "grad_norm": 0.5864097476005554, "learning_rate": 8.893603859352328e-05, "loss": 2.3386, "step": 31270 }, { "epoch": 1.0852430065710905e-13, "grad_norm": 0.5721928477287292, "learning_rate": 8.891515805143652e-05, "loss": 2.3553, "step": 31280 }, { "epoch": 1.0855899512662859e-13, "grad_norm": 0.5972686409950256, "learning_rate": 8.889427432040358e-05, "loss": 2.3407, "step": 31290 }, { "epoch": 1.0859368959614812e-13, "grad_norm": 0.5930706262588501, "learning_rate": 8.887338740372707e-05, "loss": 2.3569, "step": 31300 }, { "epoch": 1.0862838406566766e-13, "grad_norm": 0.5675032734870911, "learning_rate": 8.885249730471008e-05, "loss": 2.3497, "step": 31310 }, { "epoch": 1.086630785351872e-13, "grad_norm": 0.6050297617912292, "learning_rate": 8.883160402665616e-05, "loss": 2.3059, "step": 31320 }, { "epoch": 1.0869777300470673e-13, "grad_norm": 0.5850178599357605, "learning_rate": 8.881070757286945e-05, "loss": 2.3147, "step": 31330 }, { "epoch": 1.0873246747422627e-13, "grad_norm": 0.5872731804847717, "learning_rate": 8.878980794665452e-05, "loss": 2.3306, "step": 31340 }, { "epoch": 1.087671619437458e-13, "grad_norm": 0.5891874432563782, "learning_rate": 8.876890515131644e-05, "loss": 2.3299, "step": 31350 }, { "epoch": 1.0880185641326534e-13, "grad_norm": 0.6075451374053955, "learning_rate": 8.874799919016085e-05, "loss": 2.3422, "step": 31360 }, { "epoch": 1.0883655088278488e-13, "grad_norm": 0.5910305976867676, "learning_rate": 8.872709006649385e-05, "loss": 2.3224, "step": 31370 }, { "epoch": 1.0887124535230441e-13, "grad_norm": 0.5873690247535706, "learning_rate": 8.870617778362199e-05, "loss": 2.3134, "step": 31380 }, { "epoch": 1.0890593982182395e-13, "grad_norm": 0.573495626449585, "learning_rate": 8.868526234485242e-05, "loss": 2.3373, "step": 31390 }, { "epoch": 1.0894063429134349e-13, "grad_norm": 0.5943721532821655, "learning_rate": 8.866434375349269e-05, "loss": 2.3351, "step": 31400 }, { "epoch": 1.0897532876086302e-13, "grad_norm": 0.6160727739334106, "learning_rate": 8.864342201285096e-05, "loss": 2.36, "step": 31410 }, { "epoch": 1.0901002323038256e-13, "grad_norm": 0.7114790081977844, "learning_rate": 8.862249712623575e-05, "loss": 2.3292, "step": 31420 }, { "epoch": 1.090447176999021e-13, "grad_norm": 0.6139883399009705, "learning_rate": 8.860156909695617e-05, "loss": 2.3335, "step": 31430 }, { "epoch": 1.0907941216942163e-13, "grad_norm": 0.5663664937019348, "learning_rate": 8.858063792832184e-05, "loss": 2.3327, "step": 31440 }, { "epoch": 1.0911410663894117e-13, "grad_norm": 0.5788684487342834, "learning_rate": 8.855970362364282e-05, "loss": 2.3188, "step": 31450 }, { "epoch": 1.091488011084607e-13, "grad_norm": 0.5764639973640442, "learning_rate": 8.853876618622973e-05, "loss": 2.3183, "step": 31460 }, { "epoch": 1.0918349557798024e-13, "grad_norm": 0.583072304725647, "learning_rate": 8.851782561939358e-05, "loss": 2.3465, "step": 31470 }, { "epoch": 1.0921819004749977e-13, "grad_norm": 0.5679745674133301, "learning_rate": 8.849688192644596e-05, "loss": 2.3259, "step": 31480 }, { "epoch": 1.0925288451701931e-13, "grad_norm": 0.6544737219810486, "learning_rate": 8.847593511069895e-05, "loss": 2.3464, "step": 31490 }, { "epoch": 1.0928757898653885e-13, "grad_norm": 0.5888381600379944, "learning_rate": 8.845498517546512e-05, "loss": 2.352, "step": 31500 }, { "epoch": 1.0932227345605838e-13, "grad_norm": 0.5988656282424927, "learning_rate": 8.84340321240575e-05, "loss": 2.3896, "step": 31510 }, { "epoch": 1.0935696792557792e-13, "grad_norm": 0.617055356502533, "learning_rate": 8.841307595978964e-05, "loss": 2.3315, "step": 31520 }, { "epoch": 1.0939166239509746e-13, "grad_norm": 0.6120641231536865, "learning_rate": 8.839211668597557e-05, "loss": 2.3453, "step": 31530 }, { "epoch": 1.0942635686461699e-13, "grad_norm": 0.5944246649742126, "learning_rate": 8.837115430592983e-05, "loss": 2.3315, "step": 31540 }, { "epoch": 1.0946105133413653e-13, "grad_norm": 0.5942283272743225, "learning_rate": 8.835018882296744e-05, "loss": 2.3434, "step": 31550 }, { "epoch": 1.0949574580365606e-13, "grad_norm": 0.5871239900588989, "learning_rate": 8.832922024040392e-05, "loss": 2.3365, "step": 31560 }, { "epoch": 1.095304402731756e-13, "grad_norm": 0.6008467674255371, "learning_rate": 8.830824856155524e-05, "loss": 2.2939, "step": 31570 }, { "epoch": 1.0956513474269514e-13, "grad_norm": 0.605073094367981, "learning_rate": 8.828727378973791e-05, "loss": 2.3028, "step": 31580 }, { "epoch": 1.0959982921221467e-13, "grad_norm": 0.6166130304336548, "learning_rate": 8.826629592826891e-05, "loss": 2.3137, "step": 31590 }, { "epoch": 1.0963452368173421e-13, "grad_norm": 0.6153894662857056, "learning_rate": 8.824531498046571e-05, "loss": 2.3215, "step": 31600 }, { "epoch": 1.0966921815125374e-13, "grad_norm": 0.6248331069946289, "learning_rate": 8.822433094964627e-05, "loss": 2.326, "step": 31610 }, { "epoch": 1.0970391262077328e-13, "grad_norm": 0.5755918622016907, "learning_rate": 8.820334383912901e-05, "loss": 2.3752, "step": 31620 }, { "epoch": 1.0973860709029282e-13, "grad_norm": 0.5839005708694458, "learning_rate": 8.818235365223288e-05, "loss": 2.3167, "step": 31630 }, { "epoch": 1.0977330155981235e-13, "grad_norm": 0.5920534729957581, "learning_rate": 8.81613603922773e-05, "loss": 2.3521, "step": 31640 }, { "epoch": 1.0980799602933189e-13, "grad_norm": 0.6210189461708069, "learning_rate": 8.814036406258217e-05, "loss": 2.3712, "step": 31650 }, { "epoch": 1.0984269049885143e-13, "grad_norm": 0.6018071174621582, "learning_rate": 8.811936466646787e-05, "loss": 2.3424, "step": 31660 }, { "epoch": 1.0987738496837096e-13, "grad_norm": 0.5916240811347961, "learning_rate": 8.80983622072553e-05, "loss": 2.3548, "step": 31670 }, { "epoch": 1.099120794378905e-13, "grad_norm": 0.5814312696456909, "learning_rate": 8.807735668826577e-05, "loss": 2.3334, "step": 31680 }, { "epoch": 1.0994677390741003e-13, "grad_norm": 0.6008515357971191, "learning_rate": 8.805634811282117e-05, "loss": 2.3548, "step": 31690 }, { "epoch": 1.0998146837692957e-13, "grad_norm": 0.6022180914878845, "learning_rate": 8.80353364842438e-05, "loss": 2.3679, "step": 31700 }, { "epoch": 1.100161628464491e-13, "grad_norm": 0.5874930620193481, "learning_rate": 8.801432180585647e-05, "loss": 2.3576, "step": 31710 }, { "epoch": 1.1005085731596864e-13, "grad_norm": 0.6100168228149414, "learning_rate": 8.799330408098249e-05, "loss": 2.3703, "step": 31720 }, { "epoch": 1.1008555178548818e-13, "grad_norm": 0.5967262983322144, "learning_rate": 8.797228331294562e-05, "loss": 2.3581, "step": 31730 }, { "epoch": 1.1012024625500771e-13, "grad_norm": 0.6135166883468628, "learning_rate": 8.795125950507009e-05, "loss": 2.3604, "step": 31740 }, { "epoch": 1.1015494072452725e-13, "grad_norm": 0.6370359063148499, "learning_rate": 8.793023266068065e-05, "loss": 2.3513, "step": 31750 }, { "epoch": 1.1018963519404679e-13, "grad_norm": 0.5988125801086426, "learning_rate": 8.790920278310254e-05, "loss": 2.3988, "step": 31760 }, { "epoch": 1.1022432966356632e-13, "grad_norm": 0.6250596642494202, "learning_rate": 8.788816987566143e-05, "loss": 2.3448, "step": 31770 }, { "epoch": 1.1025902413308586e-13, "grad_norm": 0.5770289897918701, "learning_rate": 8.786713394168349e-05, "loss": 2.3358, "step": 31780 }, { "epoch": 1.102937186026054e-13, "grad_norm": 0.6002679467201233, "learning_rate": 8.784609498449539e-05, "loss": 2.3708, "step": 31790 }, { "epoch": 1.1032841307212493e-13, "grad_norm": 0.5923312902450562, "learning_rate": 8.782505300742424e-05, "loss": 2.367, "step": 31800 }, { "epoch": 1.1036310754164447e-13, "grad_norm": 0.5817593336105347, "learning_rate": 8.780400801379765e-05, "loss": 2.3302, "step": 31810 }, { "epoch": 1.10397802011164e-13, "grad_norm": 0.59783536195755, "learning_rate": 8.77829600069437e-05, "loss": 2.3035, "step": 31820 }, { "epoch": 1.1043249648068354e-13, "grad_norm": 0.582152247428894, "learning_rate": 8.776190899019098e-05, "loss": 2.3056, "step": 31830 }, { "epoch": 1.1046719095020308e-13, "grad_norm": 0.6369392275810242, "learning_rate": 8.774085496686851e-05, "loss": 2.3056, "step": 31840 }, { "epoch": 1.1050188541972261e-13, "grad_norm": 0.5846228003501892, "learning_rate": 8.77197979403058e-05, "loss": 2.3487, "step": 31850 }, { "epoch": 1.1053657988924215e-13, "grad_norm": 0.5828734040260315, "learning_rate": 8.769873791383283e-05, "loss": 2.3331, "step": 31860 }, { "epoch": 1.1057127435876168e-13, "grad_norm": 0.6194456219673157, "learning_rate": 8.767767489078008e-05, "loss": 2.3635, "step": 31870 }, { "epoch": 1.1060596882828122e-13, "grad_norm": 0.5689435005187988, "learning_rate": 8.765660887447848e-05, "loss": 2.3458, "step": 31880 }, { "epoch": 1.1064066329780076e-13, "grad_norm": 0.5862482786178589, "learning_rate": 8.763553986825943e-05, "loss": 2.389, "step": 31890 }, { "epoch": 1.1067535776732029e-13, "grad_norm": 0.6186788082122803, "learning_rate": 8.761446787545482e-05, "loss": 2.3401, "step": 31900 }, { "epoch": 1.1071005223683983e-13, "grad_norm": 0.5990809202194214, "learning_rate": 8.759339289939698e-05, "loss": 2.345, "step": 31910 }, { "epoch": 1.1074474670635936e-13, "grad_norm": 0.6078212261199951, "learning_rate": 8.757231494341879e-05, "loss": 2.3422, "step": 31920 }, { "epoch": 1.107794411758789e-13, "grad_norm": 0.5754959583282471, "learning_rate": 8.755123401085352e-05, "loss": 2.3314, "step": 31930 }, { "epoch": 1.1081413564539844e-13, "grad_norm": 0.5747816562652588, "learning_rate": 8.753015010503491e-05, "loss": 2.3549, "step": 31940 }, { "epoch": 1.1084883011491797e-13, "grad_norm": 0.5732548832893372, "learning_rate": 8.750906322929724e-05, "loss": 2.3387, "step": 31950 }, { "epoch": 1.1088352458443751e-13, "grad_norm": 0.5888025760650635, "learning_rate": 8.748797338697517e-05, "loss": 2.3571, "step": 31960 }, { "epoch": 1.1091821905395705e-13, "grad_norm": 0.5856858491897583, "learning_rate": 8.746688058140394e-05, "loss": 2.3406, "step": 31970 }, { "epoch": 1.1095291352347658e-13, "grad_norm": 0.5792394280433655, "learning_rate": 8.744578481591912e-05, "loss": 2.3259, "step": 31980 }, { "epoch": 1.1098760799299612e-13, "grad_norm": 0.5979260206222534, "learning_rate": 8.74246860938569e-05, "loss": 2.344, "step": 31990 }, { "epoch": 1.1102230246251565e-13, "grad_norm": 0.609341561794281, "learning_rate": 8.740358441855382e-05, "loss": 2.369, "step": 32000 }, { "epoch": 1.1105699693203519e-13, "grad_norm": 0.6035370230674744, "learning_rate": 8.738247979334694e-05, "loss": 2.3447, "step": 32010 }, { "epoch": 1.1109169140155473e-13, "grad_norm": 0.6082351207733154, "learning_rate": 8.736137222157376e-05, "loss": 2.3465, "step": 32020 }, { "epoch": 1.1112638587107426e-13, "grad_norm": 0.6050905585289001, "learning_rate": 8.734026170657228e-05, "loss": 2.3392, "step": 32030 }, { "epoch": 1.111610803405938e-13, "grad_norm": 0.6075258851051331, "learning_rate": 8.731914825168093e-05, "loss": 2.3115, "step": 32040 }, { "epoch": 1.1119577481011333e-13, "grad_norm": 0.5751078128814697, "learning_rate": 8.729803186023861e-05, "loss": 2.33, "step": 32050 }, { "epoch": 1.1123046927963287e-13, "grad_norm": 0.5614445209503174, "learning_rate": 8.727691253558474e-05, "loss": 2.3375, "step": 32060 }, { "epoch": 1.1126516374915241e-13, "grad_norm": 0.5823469161987305, "learning_rate": 8.725579028105912e-05, "loss": 2.3083, "step": 32070 }, { "epoch": 1.1129985821867194e-13, "grad_norm": 0.576866626739502, "learning_rate": 8.723466510000209e-05, "loss": 2.3403, "step": 32080 }, { "epoch": 1.1133455268819148e-13, "grad_norm": 0.5870134830474854, "learning_rate": 8.721353699575436e-05, "loss": 2.3581, "step": 32090 }, { "epoch": 1.1136924715771102e-13, "grad_norm": 0.6039745211601257, "learning_rate": 8.719240597165721e-05, "loss": 2.3678, "step": 32100 }, { "epoch": 1.1140394162723055e-13, "grad_norm": 0.5819171071052551, "learning_rate": 8.717127203105232e-05, "loss": 2.3179, "step": 32110 }, { "epoch": 1.1143863609675009e-13, "grad_norm": 0.5725260376930237, "learning_rate": 8.715013517728181e-05, "loss": 2.3099, "step": 32120 }, { "epoch": 1.1147333056626962e-13, "grad_norm": 0.59647536277771, "learning_rate": 8.712899541368831e-05, "loss": 2.3215, "step": 32130 }, { "epoch": 1.1150802503578916e-13, "grad_norm": 0.6367320418357849, "learning_rate": 8.710785274361489e-05, "loss": 2.3133, "step": 32140 }, { "epoch": 1.115427195053087e-13, "grad_norm": 0.6083545088768005, "learning_rate": 8.708670717040509e-05, "loss": 2.3488, "step": 32150 }, { "epoch": 1.1157741397482823e-13, "grad_norm": 0.6025922894477844, "learning_rate": 8.70655586974029e-05, "loss": 2.3289, "step": 32160 }, { "epoch": 1.1161210844434777e-13, "grad_norm": 0.6146324276924133, "learning_rate": 8.704440732795275e-05, "loss": 2.3591, "step": 32170 }, { "epoch": 1.116468029138673e-13, "grad_norm": 0.5768133997917175, "learning_rate": 8.702325306539958e-05, "loss": 2.3181, "step": 32180 }, { "epoch": 1.1168149738338684e-13, "grad_norm": 0.6475382447242737, "learning_rate": 8.700209591308873e-05, "loss": 2.3133, "step": 32190 }, { "epoch": 1.1171619185290638e-13, "grad_norm": 0.6000733971595764, "learning_rate": 8.698093587436602e-05, "loss": 2.3259, "step": 32200 }, { "epoch": 1.1175088632242591e-13, "grad_norm": 0.641428530216217, "learning_rate": 8.695977295257774e-05, "loss": 2.3682, "step": 32210 }, { "epoch": 1.1178558079194545e-13, "grad_norm": 0.595223069190979, "learning_rate": 8.69386071510706e-05, "loss": 2.3315, "step": 32220 }, { "epoch": 1.1182027526146499e-13, "grad_norm": 0.6009179353713989, "learning_rate": 8.691743847319182e-05, "loss": 2.3232, "step": 32230 }, { "epoch": 1.1185496973098452e-13, "grad_norm": 0.582416296005249, "learning_rate": 8.689626692228904e-05, "loss": 2.3474, "step": 32240 }, { "epoch": 1.1188966420050406e-13, "grad_norm": 0.552489161491394, "learning_rate": 8.687509250171034e-05, "loss": 2.3352, "step": 32250 }, { "epoch": 1.119243586700236e-13, "grad_norm": 0.6223734617233276, "learning_rate": 8.68539152148043e-05, "loss": 2.3323, "step": 32260 }, { "epoch": 1.1195905313954313e-13, "grad_norm": 0.6170471906661987, "learning_rate": 8.683273506491989e-05, "loss": 2.3377, "step": 32270 }, { "epoch": 1.1199374760906267e-13, "grad_norm": 0.5819860100746155, "learning_rate": 8.681155205540659e-05, "loss": 2.3039, "step": 32280 }, { "epoch": 1.120284420785822e-13, "grad_norm": 0.5791371464729309, "learning_rate": 8.67903661896143e-05, "loss": 2.3827, "step": 32290 }, { "epoch": 1.1206313654810174e-13, "grad_norm": 0.6172679662704468, "learning_rate": 8.67691774708934e-05, "loss": 2.3129, "step": 32300 }, { "epoch": 1.1209783101762127e-13, "grad_norm": 0.5947421193122864, "learning_rate": 8.674798590259469e-05, "loss": 2.3263, "step": 32310 }, { "epoch": 1.1213252548714081e-13, "grad_norm": 0.5950635075569153, "learning_rate": 8.672679148806944e-05, "loss": 2.3465, "step": 32320 }, { "epoch": 1.1216721995666035e-13, "grad_norm": 0.5664499402046204, "learning_rate": 8.670559423066935e-05, "loss": 2.2905, "step": 32330 }, { "epoch": 1.1220191442617988e-13, "grad_norm": 0.5842919945716858, "learning_rate": 8.668439413374661e-05, "loss": 2.3218, "step": 32340 }, { "epoch": 1.1223660889569942e-13, "grad_norm": 0.6150602698326111, "learning_rate": 8.666319120065381e-05, "loss": 2.3607, "step": 32350 }, { "epoch": 1.1227130336521896e-13, "grad_norm": 0.5823174715042114, "learning_rate": 8.664198543474404e-05, "loss": 2.3377, "step": 32360 }, { "epoch": 1.1230599783473849e-13, "grad_norm": 0.5761433839797974, "learning_rate": 8.662077683937075e-05, "loss": 2.3322, "step": 32370 }, { "epoch": 1.1234069230425803e-13, "grad_norm": 0.593967854976654, "learning_rate": 8.659956541788798e-05, "loss": 2.3763, "step": 32380 }, { "epoch": 1.1237538677377756e-13, "grad_norm": 0.5918429493904114, "learning_rate": 8.657835117365005e-05, "loss": 2.322, "step": 32390 }, { "epoch": 1.124100812432971e-13, "grad_norm": 0.5808221101760864, "learning_rate": 8.655713411001186e-05, "loss": 2.336, "step": 32400 }, { "epoch": 1.1244477571281664e-13, "grad_norm": 0.5856638550758362, "learning_rate": 8.653591423032871e-05, "loss": 2.3322, "step": 32410 }, { "epoch": 1.1247947018233617e-13, "grad_norm": 0.5769846439361572, "learning_rate": 8.651469153795633e-05, "loss": 2.3639, "step": 32420 }, { "epoch": 1.1251416465185571e-13, "grad_norm": 0.6318490505218506, "learning_rate": 8.649346603625089e-05, "loss": 2.3227, "step": 32430 }, { "epoch": 1.1254885912137524e-13, "grad_norm": 0.5895079374313354, "learning_rate": 8.647223772856902e-05, "loss": 2.3329, "step": 32440 }, { "epoch": 1.1258355359089478e-13, "grad_norm": 0.5816254615783691, "learning_rate": 8.645100661826783e-05, "loss": 2.3074, "step": 32450 }, { "epoch": 1.1261824806041432e-13, "grad_norm": 0.601548969745636, "learning_rate": 8.64297727087048e-05, "loss": 2.313, "step": 32460 }, { "epoch": 1.1265294252993385e-13, "grad_norm": 0.6211124658584595, "learning_rate": 8.640853600323792e-05, "loss": 2.33, "step": 32470 }, { "epoch": 1.1268763699945339e-13, "grad_norm": 0.5681113004684448, "learning_rate": 8.638729650522556e-05, "loss": 2.3345, "step": 32480 }, { "epoch": 1.1272233146897292e-13, "grad_norm": 0.5861620306968689, "learning_rate": 8.636605421802659e-05, "loss": 2.3553, "step": 32490 }, { "epoch": 1.1275702593849246e-13, "grad_norm": 0.6326673626899719, "learning_rate": 8.634480914500028e-05, "loss": 2.3115, "step": 32500 }, { "epoch": 1.12791720408012e-13, "grad_norm": 0.57658451795578, "learning_rate": 8.632356128950638e-05, "loss": 2.3416, "step": 32510 }, { "epoch": 1.1282641487753153e-13, "grad_norm": 0.5991392731666565, "learning_rate": 8.6302310654905e-05, "loss": 2.3472, "step": 32520 }, { "epoch": 1.1286110934705107e-13, "grad_norm": 0.5847806930541992, "learning_rate": 8.628105724455681e-05, "loss": 2.3467, "step": 32530 }, { "epoch": 1.128958038165706e-13, "grad_norm": 0.6060876250267029, "learning_rate": 8.625980106182282e-05, "loss": 2.3497, "step": 32540 }, { "epoch": 1.1293049828609014e-13, "grad_norm": 0.6131573915481567, "learning_rate": 8.62385421100645e-05, "loss": 2.3392, "step": 32550 }, { "epoch": 1.1296519275560968e-13, "grad_norm": 0.5964147448539734, "learning_rate": 8.62172803926438e-05, "loss": 2.3167, "step": 32560 }, { "epoch": 1.1299988722512921e-13, "grad_norm": 0.6015008687973022, "learning_rate": 8.619601591292308e-05, "loss": 2.3483, "step": 32570 }, { "epoch": 1.1303458169464875e-13, "grad_norm": 0.6042729020118713, "learning_rate": 8.617474867426511e-05, "loss": 2.369, "step": 32580 }, { "epoch": 1.1306927616416829e-13, "grad_norm": 0.577364981174469, "learning_rate": 8.615347868003312e-05, "loss": 2.3198, "step": 32590 }, { "epoch": 1.1310397063368782e-13, "grad_norm": 0.586683452129364, "learning_rate": 8.61322059335908e-05, "loss": 2.3382, "step": 32600 }, { "epoch": 1.1313866510320736e-13, "grad_norm": 0.6006174683570862, "learning_rate": 8.611093043830224e-05, "loss": 2.3386, "step": 32610 }, { "epoch": 1.131733595727269e-13, "grad_norm": 0.5834852457046509, "learning_rate": 8.608965219753195e-05, "loss": 2.3375, "step": 32620 }, { "epoch": 1.1320805404224643e-13, "grad_norm": 0.581831693649292, "learning_rate": 8.606837121464495e-05, "loss": 2.3037, "step": 32630 }, { "epoch": 1.1324274851176597e-13, "grad_norm": 0.5977972745895386, "learning_rate": 8.604708749300661e-05, "loss": 2.3047, "step": 32640 }, { "epoch": 1.132774429812855e-13, "grad_norm": 0.6056487560272217, "learning_rate": 8.602580103598279e-05, "loss": 2.3415, "step": 32650 }, { "epoch": 1.1331213745080504e-13, "grad_norm": 0.5995120406150818, "learning_rate": 8.600451184693974e-05, "loss": 2.316, "step": 32660 }, { "epoch": 1.1334683192032458e-13, "grad_norm": 0.6234210133552551, "learning_rate": 8.598321992924418e-05, "loss": 2.3259, "step": 32670 }, { "epoch": 1.1338152638984411e-13, "grad_norm": 0.5776862502098083, "learning_rate": 8.59619252862632e-05, "loss": 2.3474, "step": 32680 }, { "epoch": 1.1341622085936365e-13, "grad_norm": 0.6421903371810913, "learning_rate": 8.594062792136442e-05, "loss": 2.344, "step": 32690 }, { "epoch": 1.1345091532888318e-13, "grad_norm": 0.5684354901313782, "learning_rate": 8.59193278379158e-05, "loss": 2.2932, "step": 32700 }, { "epoch": 1.1348560979840272e-13, "grad_norm": 0.6104498505592346, "learning_rate": 8.589802503928579e-05, "loss": 2.3094, "step": 32710 }, { "epoch": 1.1352030426792226e-13, "grad_norm": 0.6248301267623901, "learning_rate": 8.587671952884322e-05, "loss": 2.3598, "step": 32720 }, { "epoch": 1.1355499873744179e-13, "grad_norm": 0.5973992347717285, "learning_rate": 8.58554113099574e-05, "loss": 2.3449, "step": 32730 }, { "epoch": 1.1358969320696133e-13, "grad_norm": 0.5842437744140625, "learning_rate": 8.583410038599802e-05, "loss": 2.3027, "step": 32740 }, { "epoch": 1.1362438767648086e-13, "grad_norm": 0.6046479940414429, "learning_rate": 8.581278676033522e-05, "loss": 2.3532, "step": 32750 }, { "epoch": 1.136590821460004e-13, "grad_norm": 0.5962740182876587, "learning_rate": 8.579147043633959e-05, "loss": 2.3434, "step": 32760 }, { "epoch": 1.1369377661551994e-13, "grad_norm": 0.5760523676872253, "learning_rate": 8.577015141738209e-05, "loss": 2.333, "step": 32770 }, { "epoch": 1.1372847108503947e-13, "grad_norm": 0.563156247138977, "learning_rate": 8.574882970683413e-05, "loss": 2.3728, "step": 32780 }, { "epoch": 1.13763165554559e-13, "grad_norm": 0.5739744305610657, "learning_rate": 8.572750530806762e-05, "loss": 2.2883, "step": 32790 }, { "epoch": 1.1379786002407855e-13, "grad_norm": 0.6004202365875244, "learning_rate": 8.570617822445478e-05, "loss": 2.3658, "step": 32800 }, { "epoch": 1.1383255449359808e-13, "grad_norm": 0.6041077971458435, "learning_rate": 8.568484845936829e-05, "loss": 2.3147, "step": 32810 }, { "epoch": 1.1386724896311762e-13, "grad_norm": 0.6607598066329956, "learning_rate": 8.56635160161813e-05, "loss": 2.3193, "step": 32820 }, { "epoch": 1.1390194343263715e-13, "grad_norm": 0.6087676882743835, "learning_rate": 8.564218089826736e-05, "loss": 2.3371, "step": 32830 }, { "epoch": 1.139366379021567e-13, "grad_norm": 0.5764495134353638, "learning_rate": 8.562084310900042e-05, "loss": 2.3557, "step": 32840 }, { "epoch": 1.1397133237167623e-13, "grad_norm": 0.5904257893562317, "learning_rate": 8.559950265175485e-05, "loss": 2.3478, "step": 32850 }, { "epoch": 1.1400602684119576e-13, "grad_norm": 0.5922452807426453, "learning_rate": 8.55781595299055e-05, "loss": 2.3345, "step": 32860 }, { "epoch": 1.140407213107153e-13, "grad_norm": 0.5808495283126831, "learning_rate": 8.555681374682754e-05, "loss": 2.3273, "step": 32870 }, { "epoch": 1.1407541578023483e-13, "grad_norm": 0.6119394898414612, "learning_rate": 8.553546530589669e-05, "loss": 2.3427, "step": 32880 }, { "epoch": 1.1411011024975437e-13, "grad_norm": 0.61001056432724, "learning_rate": 8.551411421048899e-05, "loss": 2.363, "step": 32890 }, { "epoch": 1.141448047192739e-13, "grad_norm": 0.6261263489723206, "learning_rate": 8.549276046398093e-05, "loss": 2.303, "step": 32900 }, { "epoch": 1.1417949918879344e-13, "grad_norm": 0.5977233648300171, "learning_rate": 8.547140406974943e-05, "loss": 2.3362, "step": 32910 }, { "epoch": 1.1421419365831298e-13, "grad_norm": 0.5719186067581177, "learning_rate": 8.54500450311718e-05, "loss": 2.3313, "step": 32920 }, { "epoch": 1.1424888812783252e-13, "grad_norm": 0.5783168077468872, "learning_rate": 8.542868335162581e-05, "loss": 2.2997, "step": 32930 }, { "epoch": 1.1428358259735205e-13, "grad_norm": 0.5915437340736389, "learning_rate": 8.540731903448963e-05, "loss": 2.3002, "step": 32940 }, { "epoch": 1.143182770668716e-13, "grad_norm": 0.5763083100318909, "learning_rate": 8.538595208314181e-05, "loss": 2.2825, "step": 32950 }, { "epoch": 1.1435297153639112e-13, "grad_norm": 0.6393998265266418, "learning_rate": 8.53645825009614e-05, "loss": 2.3632, "step": 32960 }, { "epoch": 1.1438766600591066e-13, "grad_norm": 0.5975817441940308, "learning_rate": 8.534321029132777e-05, "loss": 2.36, "step": 32970 }, { "epoch": 1.144223604754302e-13, "grad_norm": 0.6034948229789734, "learning_rate": 8.532183545762077e-05, "loss": 2.3315, "step": 32980 }, { "epoch": 1.1445705494494973e-13, "grad_norm": 0.5972079634666443, "learning_rate": 8.530045800322065e-05, "loss": 2.3445, "step": 32990 }, { "epoch": 1.1449174941446927e-13, "grad_norm": 0.5887584090232849, "learning_rate": 8.527907793150806e-05, "loss": 2.3449, "step": 33000 }, { "epoch": 1.145264438839888e-13, "grad_norm": 0.5701847076416016, "learning_rate": 8.525769524586408e-05, "loss": 2.3696, "step": 33010 }, { "epoch": 1.1456113835350834e-13, "grad_norm": 0.6037826538085938, "learning_rate": 8.523630994967019e-05, "loss": 2.3483, "step": 33020 }, { "epoch": 1.1459583282302788e-13, "grad_norm": 0.5811011791229248, "learning_rate": 8.52149220463083e-05, "loss": 2.3054, "step": 33030 }, { "epoch": 1.1463052729254741e-13, "grad_norm": 0.5950972437858582, "learning_rate": 8.51935315391607e-05, "loss": 2.3446, "step": 33040 }, { "epoch": 1.1466522176206695e-13, "grad_norm": 0.61382657289505, "learning_rate": 8.517213843161017e-05, "loss": 2.354, "step": 33050 }, { "epoch": 1.1469991623158649e-13, "grad_norm": 0.6061109304428101, "learning_rate": 8.515074272703979e-05, "loss": 2.3401, "step": 33060 }, { "epoch": 1.1473461070110602e-13, "grad_norm": 0.5721532106399536, "learning_rate": 8.512934442883313e-05, "loss": 2.3247, "step": 33070 }, { "epoch": 1.1476930517062556e-13, "grad_norm": 0.5870599746704102, "learning_rate": 8.510794354037414e-05, "loss": 2.3375, "step": 33080 }, { "epoch": 1.148039996401451e-13, "grad_norm": 0.5998032689094543, "learning_rate": 8.508654006504719e-05, "loss": 2.3494, "step": 33090 }, { "epoch": 1.1483869410966463e-13, "grad_norm": 0.5960907936096191, "learning_rate": 8.506513400623706e-05, "loss": 2.3482, "step": 33100 }, { "epoch": 1.1487338857918417e-13, "grad_norm": 0.5672832131385803, "learning_rate": 8.504372536732892e-05, "loss": 2.34, "step": 33110 }, { "epoch": 1.149080830487037e-13, "grad_norm": 0.5981654524803162, "learning_rate": 8.502231415170839e-05, "loss": 2.3404, "step": 33120 }, { "epoch": 1.1494277751822324e-13, "grad_norm": 0.6753188371658325, "learning_rate": 8.50009003627614e-05, "loss": 2.3414, "step": 33130 }, { "epoch": 1.1497747198774277e-13, "grad_norm": 0.5921972393989563, "learning_rate": 8.497948400387447e-05, "loss": 2.3426, "step": 33140 }, { "epoch": 1.150121664572623e-13, "grad_norm": 0.5945360660552979, "learning_rate": 8.495806507843431e-05, "loss": 2.3038, "step": 33150 }, { "epoch": 1.1504686092678185e-13, "grad_norm": 0.6582672595977783, "learning_rate": 8.493664358982819e-05, "loss": 2.3091, "step": 33160 }, { "epoch": 1.1508155539630138e-13, "grad_norm": 0.6038356423377991, "learning_rate": 8.491521954144371e-05, "loss": 2.3269, "step": 33170 }, { "epoch": 1.1511624986582092e-13, "grad_norm": 0.6041941046714783, "learning_rate": 8.489379293666894e-05, "loss": 2.2833, "step": 33180 }, { "epoch": 1.1515094433534045e-13, "grad_norm": 0.5963372588157654, "learning_rate": 8.487236377889223e-05, "loss": 2.3327, "step": 33190 }, { "epoch": 1.1518563880486e-13, "grad_norm": 0.5875493288040161, "learning_rate": 8.485093207150253e-05, "loss": 2.315, "step": 33200 }, { "epoch": 1.1522033327437953e-13, "grad_norm": 0.6146724820137024, "learning_rate": 8.4829497817889e-05, "loss": 2.3004, "step": 33210 }, { "epoch": 1.1525502774389906e-13, "grad_norm": 0.5886492133140564, "learning_rate": 8.48080610214413e-05, "loss": 2.3371, "step": 33220 }, { "epoch": 1.152897222134186e-13, "grad_norm": 0.6277220845222473, "learning_rate": 8.478662168554949e-05, "loss": 2.3332, "step": 33230 }, { "epoch": 1.1532441668293814e-13, "grad_norm": 0.592812716960907, "learning_rate": 8.476517981360397e-05, "loss": 2.344, "step": 33240 }, { "epoch": 1.1535911115245767e-13, "grad_norm": 0.5849651098251343, "learning_rate": 8.474373540899563e-05, "loss": 2.3053, "step": 33250 }, { "epoch": 1.153938056219772e-13, "grad_norm": 0.5951617360115051, "learning_rate": 8.472228847511575e-05, "loss": 2.3525, "step": 33260 }, { "epoch": 1.1542850009149674e-13, "grad_norm": 0.580687403678894, "learning_rate": 8.47008390153559e-05, "loss": 2.3278, "step": 33270 }, { "epoch": 1.1546319456101628e-13, "grad_norm": 0.5772156119346619, "learning_rate": 8.467938703310818e-05, "loss": 2.3578, "step": 33280 }, { "epoch": 1.1549788903053582e-13, "grad_norm": 0.5971308946609497, "learning_rate": 8.465793253176501e-05, "loss": 2.2968, "step": 33290 }, { "epoch": 1.1553258350005535e-13, "grad_norm": 0.5901183485984802, "learning_rate": 8.463647551471926e-05, "loss": 2.3333, "step": 33300 }, { "epoch": 1.155672779695749e-13, "grad_norm": 0.5834811329841614, "learning_rate": 8.461501598536417e-05, "loss": 2.3167, "step": 33310 }, { "epoch": 1.1560197243909442e-13, "grad_norm": 0.5917845368385315, "learning_rate": 8.459355394709334e-05, "loss": 2.3629, "step": 33320 }, { "epoch": 1.1563666690861396e-13, "grad_norm": 0.5862854719161987, "learning_rate": 8.457208940330086e-05, "loss": 2.3029, "step": 33330 }, { "epoch": 1.156713613781335e-13, "grad_norm": 0.60304856300354, "learning_rate": 8.455062235738111e-05, "loss": 2.3268, "step": 33340 }, { "epoch": 1.1570605584765303e-13, "grad_norm": 0.5645644664764404, "learning_rate": 8.452915281272896e-05, "loss": 2.3215, "step": 33350 }, { "epoch": 1.1574075031717257e-13, "grad_norm": 0.5992483496665955, "learning_rate": 8.450768077273964e-05, "loss": 2.3271, "step": 33360 }, { "epoch": 1.157754447866921e-13, "grad_norm": 0.5806235074996948, "learning_rate": 8.448620624080874e-05, "loss": 2.2956, "step": 33370 }, { "epoch": 1.1581013925621164e-13, "grad_norm": 0.6046596169471741, "learning_rate": 8.446472922033227e-05, "loss": 2.2988, "step": 33380 }, { "epoch": 1.1584483372573118e-13, "grad_norm": 0.6022421717643738, "learning_rate": 8.444324971470665e-05, "loss": 2.348, "step": 33390 }, { "epoch": 1.1587952819525071e-13, "grad_norm": 0.6186346411705017, "learning_rate": 8.442176772732867e-05, "loss": 2.3519, "step": 33400 }, { "epoch": 1.1591422266477025e-13, "grad_norm": 0.5943505764007568, "learning_rate": 8.440028326159554e-05, "loss": 2.3205, "step": 33410 }, { "epoch": 1.1594891713428979e-13, "grad_norm": 0.6223346590995789, "learning_rate": 8.437879632090484e-05, "loss": 2.3447, "step": 33420 }, { "epoch": 1.1598361160380932e-13, "grad_norm": 0.5750603079795837, "learning_rate": 8.435730690865452e-05, "loss": 2.2972, "step": 33430 }, { "epoch": 1.1601830607332886e-13, "grad_norm": 0.6166639924049377, "learning_rate": 8.433581502824296e-05, "loss": 2.3149, "step": 33440 }, { "epoch": 1.160530005428484e-13, "grad_norm": 0.588918149471283, "learning_rate": 8.43143206830689e-05, "loss": 2.3456, "step": 33450 }, { "epoch": 1.1608769501236793e-13, "grad_norm": 0.6154412031173706, "learning_rate": 8.429282387653153e-05, "loss": 2.3383, "step": 33460 }, { "epoch": 1.1612238948188747e-13, "grad_norm": 0.6153872013092041, "learning_rate": 8.427132461203035e-05, "loss": 2.3474, "step": 33470 }, { "epoch": 1.16157083951407e-13, "grad_norm": 0.5913094878196716, "learning_rate": 8.424982289296529e-05, "loss": 2.3073, "step": 33480 }, { "epoch": 1.1619177842092654e-13, "grad_norm": 0.603609025478363, "learning_rate": 8.422831872273667e-05, "loss": 2.3675, "step": 33490 }, { "epoch": 1.1622647289044608e-13, "grad_norm": 0.6127690672874451, "learning_rate": 8.420681210474516e-05, "loss": 2.34, "step": 33500 }, { "epoch": 1.162611673599656e-13, "grad_norm": 0.6197881698608398, "learning_rate": 8.41853030423919e-05, "loss": 2.3372, "step": 33510 }, { "epoch": 1.1629586182948515e-13, "grad_norm": 0.6211634874343872, "learning_rate": 8.416379153907831e-05, "loss": 2.3109, "step": 33520 }, { "epoch": 1.1633055629900468e-13, "grad_norm": 0.5848438143730164, "learning_rate": 8.41422775982063e-05, "loss": 2.3188, "step": 33530 }, { "epoch": 1.1636525076852422e-13, "grad_norm": 0.5859118103981018, "learning_rate": 8.412076122317805e-05, "loss": 2.3181, "step": 33540 }, { "epoch": 1.1639994523804376e-13, "grad_norm": 0.6013963222503662, "learning_rate": 8.409924241739625e-05, "loss": 2.3227, "step": 33550 }, { "epoch": 1.164346397075633e-13, "grad_norm": 0.6054528951644897, "learning_rate": 8.407772118426388e-05, "loss": 2.3152, "step": 33560 }, { "epoch": 1.1646933417708283e-13, "grad_norm": 0.584378719329834, "learning_rate": 8.405619752718433e-05, "loss": 2.3552, "step": 33570 }, { "epoch": 1.1650402864660236e-13, "grad_norm": 0.6241729259490967, "learning_rate": 8.403467144956141e-05, "loss": 2.3596, "step": 33580 }, { "epoch": 1.165387231161219e-13, "grad_norm": 0.6088836193084717, "learning_rate": 8.401314295479927e-05, "loss": 2.3591, "step": 33590 }, { "epoch": 1.1657341758564144e-13, "grad_norm": 0.5820246934890747, "learning_rate": 8.399161204630248e-05, "loss": 2.2886, "step": 33600 }, { "epoch": 1.1660811205516097e-13, "grad_norm": 0.6030177474021912, "learning_rate": 8.39700787274759e-05, "loss": 2.3371, "step": 33610 }, { "epoch": 1.166428065246805e-13, "grad_norm": 0.600084662437439, "learning_rate": 8.394854300172493e-05, "loss": 2.3359, "step": 33620 }, { "epoch": 1.1667750099420005e-13, "grad_norm": 0.6024853587150574, "learning_rate": 8.39270048724552e-05, "loss": 2.3469, "step": 33630 }, { "epoch": 1.1671219546371958e-13, "grad_norm": 0.6131813526153564, "learning_rate": 8.390546434307277e-05, "loss": 2.3662, "step": 33640 }, { "epoch": 1.1674688993323912e-13, "grad_norm": 0.5608351826667786, "learning_rate": 8.388392141698413e-05, "loss": 2.2987, "step": 33650 }, { "epoch": 1.1678158440275865e-13, "grad_norm": 0.6119819283485413, "learning_rate": 8.38623760975961e-05, "loss": 2.3425, "step": 33660 }, { "epoch": 1.168162788722782e-13, "grad_norm": 0.5903018116950989, "learning_rate": 8.384082838831588e-05, "loss": 2.2873, "step": 33670 }, { "epoch": 1.1685097334179773e-13, "grad_norm": 0.6084502339363098, "learning_rate": 8.381927829255104e-05, "loss": 2.3395, "step": 33680 }, { "epoch": 1.1688566781131726e-13, "grad_norm": 0.6038215756416321, "learning_rate": 8.379772581370956e-05, "loss": 2.3473, "step": 33690 }, { "epoch": 1.169203622808368e-13, "grad_norm": 0.6264404058456421, "learning_rate": 8.377617095519979e-05, "loss": 2.3005, "step": 33700 }, { "epoch": 1.1695505675035633e-13, "grad_norm": 0.5808203816413879, "learning_rate": 8.375461372043042e-05, "loss": 2.3015, "step": 33710 }, { "epoch": 1.1698975121987587e-13, "grad_norm": 0.5837031602859497, "learning_rate": 8.373305411281053e-05, "loss": 2.3445, "step": 33720 }, { "epoch": 1.170244456893954e-13, "grad_norm": 0.6118404269218445, "learning_rate": 8.371149213574962e-05, "loss": 2.3265, "step": 33730 }, { "epoch": 1.1705914015891494e-13, "grad_norm": 0.5775696635246277, "learning_rate": 8.368992779265753e-05, "loss": 2.3318, "step": 33740 }, { "epoch": 1.1709383462843448e-13, "grad_norm": 0.5787637829780579, "learning_rate": 8.366836108694444e-05, "loss": 2.3444, "step": 33750 }, { "epoch": 1.1712852909795402e-13, "grad_norm": 0.5998949408531189, "learning_rate": 8.364679202202098e-05, "loss": 2.2859, "step": 33760 }, { "epoch": 1.1716322356747355e-13, "grad_norm": 0.5909065008163452, "learning_rate": 8.362522060129808e-05, "loss": 2.3243, "step": 33770 }, { "epoch": 1.1719791803699309e-13, "grad_norm": 0.5917692184448242, "learning_rate": 8.360364682818711e-05, "loss": 2.3266, "step": 33780 }, { "epoch": 1.1723261250651262e-13, "grad_norm": 0.5726926326751709, "learning_rate": 8.358207070609975e-05, "loss": 2.304, "step": 33790 }, { "epoch": 1.1726730697603216e-13, "grad_norm": 0.5929071307182312, "learning_rate": 8.356049223844807e-05, "loss": 2.3236, "step": 33800 }, { "epoch": 1.173020014455517e-13, "grad_norm": 0.5749326348304749, "learning_rate": 8.353891142864455e-05, "loss": 2.3222, "step": 33810 }, { "epoch": 1.1733669591507123e-13, "grad_norm": 0.5981829762458801, "learning_rate": 8.351732828010196e-05, "loss": 2.3224, "step": 33820 }, { "epoch": 1.1737139038459077e-13, "grad_norm": 0.5882853269577026, "learning_rate": 8.349574279623355e-05, "loss": 2.3451, "step": 33830 }, { "epoch": 1.174060848541103e-13, "grad_norm": 0.5952692627906799, "learning_rate": 8.347415498045284e-05, "loss": 2.3446, "step": 33840 }, { "epoch": 1.1744077932362984e-13, "grad_norm": 0.6108142137527466, "learning_rate": 8.345256483617378e-05, "loss": 2.334, "step": 33850 }, { "epoch": 1.1747547379314938e-13, "grad_norm": 0.6039605736732483, "learning_rate": 8.343097236681064e-05, "loss": 2.3281, "step": 33860 }, { "epoch": 1.175101682626689e-13, "grad_norm": 0.5702954530715942, "learning_rate": 8.340937757577808e-05, "loss": 2.3408, "step": 33870 }, { "epoch": 1.1754486273218845e-13, "grad_norm": 0.5794089436531067, "learning_rate": 8.338778046649116e-05, "loss": 2.337, "step": 33880 }, { "epoch": 1.1757955720170798e-13, "grad_norm": 0.5912054777145386, "learning_rate": 8.336618104236525e-05, "loss": 2.3275, "step": 33890 }, { "epoch": 1.1761425167122752e-13, "grad_norm": 0.6030844449996948, "learning_rate": 8.334457930681616e-05, "loss": 2.3173, "step": 33900 }, { "epoch": 1.1764894614074706e-13, "grad_norm": 0.5957759618759155, "learning_rate": 8.332297526325993e-05, "loss": 2.3014, "step": 33910 }, { "epoch": 1.176836406102666e-13, "grad_norm": 0.6011395454406738, "learning_rate": 8.330136891511317e-05, "loss": 2.3099, "step": 33920 }, { "epoch": 1.1771833507978613e-13, "grad_norm": 0.6034684181213379, "learning_rate": 8.327976026579262e-05, "loss": 2.364, "step": 33930 }, { "epoch": 1.1775302954930567e-13, "grad_norm": 0.6023189425468445, "learning_rate": 8.325814931871559e-05, "loss": 2.342, "step": 33940 }, { "epoch": 1.177877240188252e-13, "grad_norm": 0.6079905033111572, "learning_rate": 8.323653607729962e-05, "loss": 2.3136, "step": 33950 }, { "epoch": 1.1782241848834474e-13, "grad_norm": 0.6294716596603394, "learning_rate": 8.321492054496268e-05, "loss": 2.336, "step": 33960 }, { "epoch": 1.1785711295786427e-13, "grad_norm": 0.6083453297615051, "learning_rate": 8.319330272512306e-05, "loss": 2.3111, "step": 33970 }, { "epoch": 1.178918074273838e-13, "grad_norm": 0.5708293318748474, "learning_rate": 8.317168262119943e-05, "loss": 2.3273, "step": 33980 }, { "epoch": 1.1792650189690335e-13, "grad_norm": 0.5853406190872192, "learning_rate": 8.315006023661085e-05, "loss": 2.2977, "step": 33990 }, { "epoch": 1.1796119636642288e-13, "grad_norm": 0.5928559303283691, "learning_rate": 8.31284355747767e-05, "loss": 2.3254, "step": 34000 }, { "epoch": 1.1799589083594242e-13, "grad_norm": 0.5653216242790222, "learning_rate": 8.310680863911674e-05, "loss": 2.3483, "step": 34010 }, { "epoch": 1.1803058530546195e-13, "grad_norm": 0.6033141613006592, "learning_rate": 8.308517943305108e-05, "loss": 2.3322, "step": 34020 }, { "epoch": 1.180652797749815e-13, "grad_norm": 0.5965766310691833, "learning_rate": 8.306354796000018e-05, "loss": 2.3217, "step": 34030 }, { "epoch": 1.1809997424450103e-13, "grad_norm": 0.6003820300102234, "learning_rate": 8.304191422338488e-05, "loss": 2.3325, "step": 34040 }, { "epoch": 1.1813466871402056e-13, "grad_norm": 0.6356145739555359, "learning_rate": 8.302027822662638e-05, "loss": 2.3269, "step": 34050 }, { "epoch": 1.181693631835401e-13, "grad_norm": 0.5943249464035034, "learning_rate": 8.299863997314622e-05, "loss": 2.3276, "step": 34060 }, { "epoch": 1.1820405765305964e-13, "grad_norm": 0.6640463471412659, "learning_rate": 8.297699946636629e-05, "loss": 2.3324, "step": 34070 }, { "epoch": 1.1823875212257917e-13, "grad_norm": 0.6082382202148438, "learning_rate": 8.295535670970888e-05, "loss": 2.3332, "step": 34080 }, { "epoch": 1.182734465920987e-13, "grad_norm": 0.597669780254364, "learning_rate": 8.293371170659656e-05, "loss": 2.3238, "step": 34090 }, { "epoch": 1.1830814106161824e-13, "grad_norm": 0.5712488293647766, "learning_rate": 8.291206446045235e-05, "loss": 2.3353, "step": 34100 }, { "epoch": 1.1834283553113778e-13, "grad_norm": 0.607682466506958, "learning_rate": 8.289041497469956e-05, "loss": 2.3061, "step": 34110 }, { "epoch": 1.1837753000065732e-13, "grad_norm": 0.6312044262886047, "learning_rate": 8.286876325276184e-05, "loss": 2.3044, "step": 34120 }, { "epoch": 1.1841222447017685e-13, "grad_norm": 0.5910617113113403, "learning_rate": 8.284710929806326e-05, "loss": 2.2992, "step": 34130 }, { "epoch": 1.184469189396964e-13, "grad_norm": 0.5869236588478088, "learning_rate": 8.282545311402817e-05, "loss": 2.3366, "step": 34140 }, { "epoch": 1.1848161340921592e-13, "grad_norm": 0.5886296033859253, "learning_rate": 8.280379470408135e-05, "loss": 2.3176, "step": 34150 }, { "epoch": 1.1851630787873546e-13, "grad_norm": 0.5687978863716125, "learning_rate": 8.278213407164789e-05, "loss": 2.3167, "step": 34160 }, { "epoch": 1.18551002348255e-13, "grad_norm": 0.5726280212402344, "learning_rate": 8.27604712201532e-05, "loss": 2.2904, "step": 34170 }, { "epoch": 1.1858569681777453e-13, "grad_norm": 0.6041830778121948, "learning_rate": 8.27388061530231e-05, "loss": 2.3433, "step": 34180 }, { "epoch": 1.1862039128729407e-13, "grad_norm": 0.5893668532371521, "learning_rate": 8.271713887368371e-05, "loss": 2.3215, "step": 34190 }, { "epoch": 1.186550857568136e-13, "grad_norm": 0.5812632441520691, "learning_rate": 8.269546938556154e-05, "loss": 2.3087, "step": 34200 }, { "epoch": 1.1868978022633314e-13, "grad_norm": 0.6026266813278198, "learning_rate": 8.267379769208341e-05, "loss": 2.3513, "step": 34210 }, { "epoch": 1.1872447469585268e-13, "grad_norm": 0.594331681728363, "learning_rate": 8.265212379667656e-05, "loss": 2.3006, "step": 34220 }, { "epoch": 1.1875916916537221e-13, "grad_norm": 0.580424964427948, "learning_rate": 8.26304477027685e-05, "loss": 2.3096, "step": 34230 }, { "epoch": 1.1879386363489175e-13, "grad_norm": 0.5810263752937317, "learning_rate": 8.260876941378711e-05, "loss": 2.3677, "step": 34240 }, { "epoch": 1.1882855810441129e-13, "grad_norm": 0.6025339365005493, "learning_rate": 8.258708893316064e-05, "loss": 2.3334, "step": 34250 }, { "epoch": 1.1886325257393082e-13, "grad_norm": 0.6306710839271545, "learning_rate": 8.256540626431766e-05, "loss": 2.3498, "step": 34260 }, { "epoch": 1.1889794704345036e-13, "grad_norm": 0.5874931812286377, "learning_rate": 8.254372141068713e-05, "loss": 2.2759, "step": 34270 }, { "epoch": 1.189326415129699e-13, "grad_norm": 0.6241834759712219, "learning_rate": 8.252203437569826e-05, "loss": 2.3071, "step": 34280 }, { "epoch": 1.1896733598248943e-13, "grad_norm": 0.592319905757904, "learning_rate": 8.250034516278071e-05, "loss": 2.3039, "step": 34290 }, { "epoch": 1.1900203045200897e-13, "grad_norm": 0.5904476046562195, "learning_rate": 8.247865377536444e-05, "loss": 2.3375, "step": 34300 }, { "epoch": 1.190367249215285e-13, "grad_norm": 0.5878010392189026, "learning_rate": 8.245696021687976e-05, "loss": 2.3039, "step": 34310 }, { "epoch": 1.1907141939104804e-13, "grad_norm": 0.6068446636199951, "learning_rate": 8.24352644907573e-05, "loss": 2.3273, "step": 34320 }, { "epoch": 1.1910611386056758e-13, "grad_norm": 0.6101313829421997, "learning_rate": 8.241356660042809e-05, "loss": 2.3247, "step": 34330 }, { "epoch": 1.191408083300871e-13, "grad_norm": 0.5843600034713745, "learning_rate": 8.239186654932342e-05, "loss": 2.3284, "step": 34340 }, { "epoch": 1.1917550279960665e-13, "grad_norm": 0.5869078636169434, "learning_rate": 8.237016434087497e-05, "loss": 2.3319, "step": 34350 }, { "epoch": 1.1921019726912618e-13, "grad_norm": 0.6119171977043152, "learning_rate": 8.23484599785148e-05, "loss": 2.3198, "step": 34360 }, { "epoch": 1.1924489173864572e-13, "grad_norm": 0.5951277017593384, "learning_rate": 8.232675346567523e-05, "loss": 2.3203, "step": 34370 }, { "epoch": 1.1927958620816526e-13, "grad_norm": 0.5828631520271301, "learning_rate": 8.230504480578897e-05, "loss": 2.3116, "step": 34380 }, { "epoch": 1.193142806776848e-13, "grad_norm": 0.6096369624137878, "learning_rate": 8.228333400228905e-05, "loss": 2.3186, "step": 34390 }, { "epoch": 1.1934897514720433e-13, "grad_norm": 0.6473119258880615, "learning_rate": 8.226162105860888e-05, "loss": 2.3296, "step": 34400 }, { "epoch": 1.1938366961672386e-13, "grad_norm": 0.6203206181526184, "learning_rate": 8.223990597818212e-05, "loss": 2.3452, "step": 34410 }, { "epoch": 1.194183640862434e-13, "grad_norm": 0.5951830744743347, "learning_rate": 8.221818876444286e-05, "loss": 2.3076, "step": 34420 }, { "epoch": 1.1945305855576294e-13, "grad_norm": 0.5758978724479675, "learning_rate": 8.219646942082548e-05, "loss": 2.3487, "step": 34430 }, { "epoch": 1.1948775302528247e-13, "grad_norm": 0.593814492225647, "learning_rate": 8.21747479507647e-05, "loss": 2.3112, "step": 34440 }, { "epoch": 1.19522447494802e-13, "grad_norm": 0.6096377372741699, "learning_rate": 8.215302435769561e-05, "loss": 2.348, "step": 34450 }, { "epoch": 1.1955714196432154e-13, "grad_norm": 0.62547767162323, "learning_rate": 8.213129864505357e-05, "loss": 2.3367, "step": 34460 }, { "epoch": 1.1959183643384108e-13, "grad_norm": 0.5887045860290527, "learning_rate": 8.210957081627436e-05, "loss": 2.3348, "step": 34470 }, { "epoch": 1.1962653090336062e-13, "grad_norm": 0.63025963306427, "learning_rate": 8.208784087479402e-05, "loss": 2.2948, "step": 34480 }, { "epoch": 1.1966122537288015e-13, "grad_norm": 0.6012529730796814, "learning_rate": 8.206610882404896e-05, "loss": 2.3375, "step": 34490 }, { "epoch": 1.196959198423997e-13, "grad_norm": 0.5694349408149719, "learning_rate": 8.204437466747591e-05, "loss": 2.3268, "step": 34500 }, { "epoch": 1.1973061431191923e-13, "grad_norm": 0.5737162828445435, "learning_rate": 8.202263840851194e-05, "loss": 2.305, "step": 34510 }, { "epoch": 1.1976530878143876e-13, "grad_norm": 0.5883069634437561, "learning_rate": 8.200090005059445e-05, "loss": 2.3169, "step": 34520 }, { "epoch": 1.198000032509583e-13, "grad_norm": 0.5992470383644104, "learning_rate": 8.197915959716119e-05, "loss": 2.3265, "step": 34530 }, { "epoch": 1.1983469772047783e-13, "grad_norm": 0.6069474220275879, "learning_rate": 8.195741705165023e-05, "loss": 2.3371, "step": 34540 }, { "epoch": 1.1986939218999737e-13, "grad_norm": 0.6193127036094666, "learning_rate": 8.193567241749996e-05, "loss": 2.316, "step": 34550 }, { "epoch": 1.199040866595169e-13, "grad_norm": 0.5879985690116882, "learning_rate": 8.191392569814909e-05, "loss": 2.3404, "step": 34560 }, { "epoch": 1.1993878112903644e-13, "grad_norm": 0.58694988489151, "learning_rate": 8.189217689703667e-05, "loss": 2.2921, "step": 34570 }, { "epoch": 1.1997347559855598e-13, "grad_norm": 0.5961881279945374, "learning_rate": 8.187042601760211e-05, "loss": 2.3179, "step": 34580 }, { "epoch": 1.2000817006807551e-13, "grad_norm": 0.5978996753692627, "learning_rate": 8.184867306328514e-05, "loss": 2.3232, "step": 34590 }, { "epoch": 1.2004286453759505e-13, "grad_norm": 0.6301842331886292, "learning_rate": 8.182691803752573e-05, "loss": 2.3013, "step": 34600 }, { "epoch": 1.2007755900711459e-13, "grad_norm": 0.6604142189025879, "learning_rate": 8.180516094376433e-05, "loss": 2.3451, "step": 34610 }, { "epoch": 1.2011225347663412e-13, "grad_norm": 0.6442047357559204, "learning_rate": 8.178340178544161e-05, "loss": 2.2978, "step": 34620 }, { "epoch": 1.2014694794615366e-13, "grad_norm": 0.5749270915985107, "learning_rate": 8.176164056599859e-05, "loss": 2.3175, "step": 34630 }, { "epoch": 1.201816424156732e-13, "grad_norm": 0.6130125522613525, "learning_rate": 8.173987728887662e-05, "loss": 2.3433, "step": 34640 }, { "epoch": 1.2021633688519273e-13, "grad_norm": 0.6132310628890991, "learning_rate": 8.171811195751737e-05, "loss": 2.2993, "step": 34650 }, { "epoch": 1.2025103135471227e-13, "grad_norm": 0.6018722057342529, "learning_rate": 8.169634457536284e-05, "loss": 2.3239, "step": 34660 }, { "epoch": 1.202857258242318e-13, "grad_norm": 0.6065945625305176, "learning_rate": 8.167457514585536e-05, "loss": 2.3137, "step": 34670 }, { "epoch": 1.2032042029375134e-13, "grad_norm": 0.6220952868461609, "learning_rate": 8.165280367243756e-05, "loss": 2.3131, "step": 34680 }, { "epoch": 1.2035511476327088e-13, "grad_norm": 0.5888240933418274, "learning_rate": 8.163103015855246e-05, "loss": 2.3138, "step": 34690 }, { "epoch": 1.203898092327904e-13, "grad_norm": 0.61471027135849, "learning_rate": 8.16092546076433e-05, "loss": 2.3074, "step": 34700 }, { "epoch": 1.2042450370230995e-13, "grad_norm": 0.6118069887161255, "learning_rate": 8.158747702315374e-05, "loss": 2.2893, "step": 34710 }, { "epoch": 1.2045919817182948e-13, "grad_norm": 0.5979725122451782, "learning_rate": 8.156569740852768e-05, "loss": 2.3359, "step": 34720 }, { "epoch": 1.2049389264134902e-13, "grad_norm": 0.5815098285675049, "learning_rate": 8.154391576720938e-05, "loss": 2.3302, "step": 34730 }, { "epoch": 1.2052858711086856e-13, "grad_norm": 0.6027605533599854, "learning_rate": 8.152213210264345e-05, "loss": 2.3462, "step": 34740 }, { "epoch": 1.205632815803881e-13, "grad_norm": 0.5979508757591248, "learning_rate": 8.150034641827475e-05, "loss": 2.3159, "step": 34750 }, { "epoch": 1.2059797604990763e-13, "grad_norm": 0.5807327032089233, "learning_rate": 8.147855871754854e-05, "loss": 2.3112, "step": 34760 }, { "epoch": 1.2063267051942717e-13, "grad_norm": 0.6179565787315369, "learning_rate": 8.145676900391033e-05, "loss": 2.3125, "step": 34770 }, { "epoch": 1.206673649889467e-13, "grad_norm": 0.6009407043457031, "learning_rate": 8.143497728080598e-05, "loss": 2.2931, "step": 34780 }, { "epoch": 1.2070205945846624e-13, "grad_norm": 0.600666880607605, "learning_rate": 8.141318355168167e-05, "loss": 2.3388, "step": 34790 }, { "epoch": 1.2073675392798577e-13, "grad_norm": 0.58729487657547, "learning_rate": 8.13913878199839e-05, "loss": 2.2884, "step": 34800 }, { "epoch": 1.207714483975053e-13, "grad_norm": 0.5964505672454834, "learning_rate": 8.136959008915945e-05, "loss": 2.2989, "step": 34810 }, { "epoch": 1.2080614286702485e-13, "grad_norm": 0.6046568155288696, "learning_rate": 8.134779036265547e-05, "loss": 2.3525, "step": 34820 }, { "epoch": 1.2084083733654438e-13, "grad_norm": 0.5878017544746399, "learning_rate": 8.132598864391935e-05, "loss": 2.3189, "step": 34830 }, { "epoch": 1.2087553180606392e-13, "grad_norm": 0.6019262671470642, "learning_rate": 8.130418493639891e-05, "loss": 2.3504, "step": 34840 }, { "epoch": 1.2091022627558345e-13, "grad_norm": 0.611096978187561, "learning_rate": 8.12823792435422e-05, "loss": 2.2846, "step": 34850 }, { "epoch": 1.20944920745103e-13, "grad_norm": 0.5847552418708801, "learning_rate": 8.126057156879757e-05, "loss": 2.3043, "step": 34860 }, { "epoch": 1.2097961521462253e-13, "grad_norm": 0.6166864037513733, "learning_rate": 8.123876191561374e-05, "loss": 2.3093, "step": 34870 }, { "epoch": 1.2101430968414206e-13, "grad_norm": 0.7008529901504517, "learning_rate": 8.121695028743974e-05, "loss": 2.3105, "step": 34880 }, { "epoch": 1.210490041536616e-13, "grad_norm": 0.5716992020606995, "learning_rate": 8.119513668772484e-05, "loss": 2.3213, "step": 34890 }, { "epoch": 1.2108369862318114e-13, "grad_norm": 0.5982014536857605, "learning_rate": 8.11733211199187e-05, "loss": 2.2933, "step": 34900 }, { "epoch": 1.2111839309270067e-13, "grad_norm": 0.5805603861808777, "learning_rate": 8.115150358747128e-05, "loss": 2.3381, "step": 34910 }, { "epoch": 1.211530875622202e-13, "grad_norm": 0.6089751124382019, "learning_rate": 8.112968409383283e-05, "loss": 2.3294, "step": 34920 }, { "epoch": 1.2118778203173974e-13, "grad_norm": 0.6191920042037964, "learning_rate": 8.110786264245388e-05, "loss": 2.311, "step": 34930 }, { "epoch": 1.2122247650125928e-13, "grad_norm": 0.5825940370559692, "learning_rate": 8.108603923678535e-05, "loss": 2.3417, "step": 34940 }, { "epoch": 1.2125717097077882e-13, "grad_norm": 0.582037091255188, "learning_rate": 8.10642138802784e-05, "loss": 2.2937, "step": 34950 }, { "epoch": 1.2129186544029835e-13, "grad_norm": 0.6209961175918579, "learning_rate": 8.104238657638452e-05, "loss": 2.308, "step": 34960 }, { "epoch": 1.213265599098179e-13, "grad_norm": 0.6033042073249817, "learning_rate": 8.102055732855552e-05, "loss": 2.3074, "step": 34970 }, { "epoch": 1.2136125437933742e-13, "grad_norm": 0.5965479612350464, "learning_rate": 8.099872614024351e-05, "loss": 2.3186, "step": 34980 }, { "epoch": 1.2139594884885696e-13, "grad_norm": 0.5925340056419373, "learning_rate": 8.097689301490088e-05, "loss": 2.3396, "step": 34990 }, { "epoch": 1.214306433183765e-13, "grad_norm": 0.5857922434806824, "learning_rate": 8.095505795598039e-05, "loss": 2.3433, "step": 35000 }, { "epoch": 1.2146533778789603e-13, "grad_norm": 0.6125169396400452, "learning_rate": 8.093322096693504e-05, "loss": 2.2846, "step": 35010 }, { "epoch": 1.2150003225741557e-13, "grad_norm": 0.6135491728782654, "learning_rate": 8.091138205121817e-05, "loss": 2.3505, "step": 35020 }, { "epoch": 1.215347267269351e-13, "grad_norm": 0.586594820022583, "learning_rate": 8.088954121228341e-05, "loss": 2.2775, "step": 35030 }, { "epoch": 1.2156942119645464e-13, "grad_norm": 0.5991926193237305, "learning_rate": 8.086769845358472e-05, "loss": 2.3364, "step": 35040 }, { "epoch": 1.2160411566597418e-13, "grad_norm": 0.5845337510108948, "learning_rate": 8.08458537785763e-05, "loss": 2.2855, "step": 35050 }, { "epoch": 1.2163881013549371e-13, "grad_norm": 0.5960389971733093, "learning_rate": 8.082400719071276e-05, "loss": 2.3318, "step": 35060 }, { "epoch": 1.2167350460501325e-13, "grad_norm": 0.6122472286224365, "learning_rate": 8.080215869344889e-05, "loss": 2.3141, "step": 35070 }, { "epoch": 1.2170819907453279e-13, "grad_norm": 0.6197759509086609, "learning_rate": 8.078030829023988e-05, "loss": 2.2876, "step": 35080 }, { "epoch": 1.2174289354405232e-13, "grad_norm": 0.5945320129394531, "learning_rate": 8.075845598454117e-05, "loss": 2.3296, "step": 35090 }, { "epoch": 1.2177758801357186e-13, "grad_norm": 0.6236541867256165, "learning_rate": 8.07366017798085e-05, "loss": 2.2652, "step": 35100 }, { "epoch": 1.218122824830914e-13, "grad_norm": 0.6073376536369324, "learning_rate": 8.071474567949797e-05, "loss": 2.2892, "step": 35110 }, { "epoch": 1.2184697695261093e-13, "grad_norm": 0.593651533126831, "learning_rate": 8.069288768706588e-05, "loss": 2.3156, "step": 35120 }, { "epoch": 1.2188167142213047e-13, "grad_norm": 0.5887266397476196, "learning_rate": 8.067102780596893e-05, "loss": 2.2831, "step": 35130 }, { "epoch": 1.2191636589165e-13, "grad_norm": 0.5945258140563965, "learning_rate": 8.064916603966403e-05, "loss": 2.3293, "step": 35140 }, { "epoch": 1.2195106036116954e-13, "grad_norm": 0.6166735291481018, "learning_rate": 8.062730239160846e-05, "loss": 2.309, "step": 35150 }, { "epoch": 1.2198575483068907e-13, "grad_norm": 0.6546661257743835, "learning_rate": 8.060543686525974e-05, "loss": 2.308, "step": 35160 }, { "epoch": 1.220204493002086e-13, "grad_norm": 1.004837155342102, "learning_rate": 8.058356946407576e-05, "loss": 2.3152, "step": 35170 }, { "epoch": 1.2205514376972815e-13, "grad_norm": 0.6036854982376099, "learning_rate": 8.056170019151461e-05, "loss": 2.2534, "step": 35180 }, { "epoch": 1.2208983823924768e-13, "grad_norm": 0.6282046437263489, "learning_rate": 8.053982905103477e-05, "loss": 2.3136, "step": 35190 }, { "epoch": 1.2212453270876722e-13, "grad_norm": 0.5974205732345581, "learning_rate": 8.051795604609494e-05, "loss": 2.3547, "step": 35200 }, { "epoch": 1.2215922717828676e-13, "grad_norm": 0.5900152921676636, "learning_rate": 8.049608118015415e-05, "loss": 2.3041, "step": 35210 }, { "epoch": 1.221939216478063e-13, "grad_norm": 0.5786914229393005, "learning_rate": 8.047420445667176e-05, "loss": 2.3, "step": 35220 }, { "epoch": 1.2222861611732583e-13, "grad_norm": 0.6081931591033936, "learning_rate": 8.045232587910732e-05, "loss": 2.3276, "step": 35230 }, { "epoch": 1.2226331058684536e-13, "grad_norm": 0.5748016834259033, "learning_rate": 8.04304454509208e-05, "loss": 2.3123, "step": 35240 }, { "epoch": 1.222980050563649e-13, "grad_norm": 0.5962873101234436, "learning_rate": 8.04085631755724e-05, "loss": 2.3053, "step": 35250 }, { "epoch": 1.2233269952588444e-13, "grad_norm": 0.6048467755317688, "learning_rate": 8.038667905652256e-05, "loss": 2.3535, "step": 35260 }, { "epoch": 1.2236739399540397e-13, "grad_norm": 0.6050984859466553, "learning_rate": 8.03647930972321e-05, "loss": 2.2839, "step": 35270 }, { "epoch": 1.224020884649235e-13, "grad_norm": 0.6370023488998413, "learning_rate": 8.03429053011621e-05, "loss": 2.3208, "step": 35280 }, { "epoch": 1.2243678293444304e-13, "grad_norm": 0.6000770926475525, "learning_rate": 8.032101567177393e-05, "loss": 2.2771, "step": 35290 }, { "epoch": 1.2247147740396258e-13, "grad_norm": 0.5913699269294739, "learning_rate": 8.029912421252919e-05, "loss": 2.3108, "step": 35300 }, { "epoch": 1.2250617187348212e-13, "grad_norm": 0.6054161787033081, "learning_rate": 8.027723092688992e-05, "loss": 2.3115, "step": 35310 }, { "epoch": 1.2254086634300165e-13, "grad_norm": 0.611118495464325, "learning_rate": 8.025533581831827e-05, "loss": 2.3329, "step": 35320 }, { "epoch": 1.225755608125212e-13, "grad_norm": 0.6302390098571777, "learning_rate": 8.023343889027683e-05, "loss": 2.3186, "step": 35330 }, { "epoch": 1.2261025528204073e-13, "grad_norm": 0.5775547027587891, "learning_rate": 8.021154014622837e-05, "loss": 2.289, "step": 35340 }, { "epoch": 1.2264494975156026e-13, "grad_norm": 0.5980232357978821, "learning_rate": 8.0189639589636e-05, "loss": 2.2989, "step": 35350 }, { "epoch": 1.226796442210798e-13, "grad_norm": 0.5787455439567566, "learning_rate": 8.01677372239631e-05, "loss": 2.308, "step": 35360 }, { "epoch": 1.2271433869059933e-13, "grad_norm": 0.6018754243850708, "learning_rate": 8.014583305267332e-05, "loss": 2.3023, "step": 35370 }, { "epoch": 1.2274903316011887e-13, "grad_norm": 0.5741833448410034, "learning_rate": 8.012392707923068e-05, "loss": 2.3194, "step": 35380 }, { "epoch": 1.227837276296384e-13, "grad_norm": 0.5917328596115112, "learning_rate": 8.010201930709935e-05, "loss": 2.316, "step": 35390 }, { "epoch": 1.2281842209915794e-13, "grad_norm": 0.5954431295394897, "learning_rate": 8.00801097397439e-05, "loss": 2.335, "step": 35400 }, { "epoch": 1.2285311656867748e-13, "grad_norm": 0.6210041046142578, "learning_rate": 8.005819838062912e-05, "loss": 2.288, "step": 35410 }, { "epoch": 1.2288781103819701e-13, "grad_norm": 0.6328104734420776, "learning_rate": 8.003628523322009e-05, "loss": 2.2906, "step": 35420 }, { "epoch": 1.2292250550771655e-13, "grad_norm": 0.6024813652038574, "learning_rate": 8.001437030098223e-05, "loss": 2.3055, "step": 35430 }, { "epoch": 1.2295719997723609e-13, "grad_norm": 0.5727076530456543, "learning_rate": 7.999245358738116e-05, "loss": 2.2989, "step": 35440 }, { "epoch": 1.2299189444675562e-13, "grad_norm": 0.5979491472244263, "learning_rate": 7.997053509588284e-05, "loss": 2.3294, "step": 35450 }, { "epoch": 1.2302658891627516e-13, "grad_norm": 0.6071661710739136, "learning_rate": 7.994861482995347e-05, "loss": 2.2904, "step": 35460 }, { "epoch": 1.230612833857947e-13, "grad_norm": 0.6009554266929626, "learning_rate": 7.992669279305957e-05, "loss": 2.3137, "step": 35470 }, { "epoch": 1.2309597785531423e-13, "grad_norm": 0.6217548847198486, "learning_rate": 7.99047689886679e-05, "loss": 2.2606, "step": 35480 }, { "epoch": 1.2313067232483377e-13, "grad_norm": 0.613007664680481, "learning_rate": 7.988284342024554e-05, "loss": 2.3193, "step": 35490 }, { "epoch": 1.231653667943533e-13, "grad_norm": 0.6087950468063354, "learning_rate": 7.986091609125984e-05, "loss": 2.3603, "step": 35500 }, { "epoch": 1.2320006126387284e-13, "grad_norm": 0.6099095940589905, "learning_rate": 7.98389870051784e-05, "loss": 2.2947, "step": 35510 }, { "epoch": 1.2323475573339238e-13, "grad_norm": 0.603653609752655, "learning_rate": 7.981705616546912e-05, "loss": 2.3105, "step": 35520 }, { "epoch": 1.232694502029119e-13, "grad_norm": 0.5904843211174011, "learning_rate": 7.979512357560016e-05, "loss": 2.3362, "step": 35530 }, { "epoch": 1.2330414467243145e-13, "grad_norm": 0.6422297358512878, "learning_rate": 7.977318923903999e-05, "loss": 2.2966, "step": 35540 }, { "epoch": 1.2333883914195098e-13, "grad_norm": 0.5753820538520813, "learning_rate": 7.975125315925734e-05, "loss": 2.3389, "step": 35550 }, { "epoch": 1.2337353361147052e-13, "grad_norm": 0.5970420837402344, "learning_rate": 7.972931533972118e-05, "loss": 2.2956, "step": 35560 }, { "epoch": 1.2340822808099006e-13, "grad_norm": 0.6015148162841797, "learning_rate": 7.970737578390084e-05, "loss": 2.3071, "step": 35570 }, { "epoch": 1.234429225505096e-13, "grad_norm": 0.5993735790252686, "learning_rate": 7.968543449526581e-05, "loss": 2.3286, "step": 35580 }, { "epoch": 1.2347761702002913e-13, "grad_norm": 0.6092479825019836, "learning_rate": 7.966349147728597e-05, "loss": 2.312, "step": 35590 }, { "epoch": 1.2351231148954867e-13, "grad_norm": 0.6032568216323853, "learning_rate": 7.964154673343137e-05, "loss": 2.3458, "step": 35600 }, { "epoch": 1.235470059590682e-13, "grad_norm": 0.597997784614563, "learning_rate": 7.961960026717245e-05, "loss": 2.3215, "step": 35610 }, { "epoch": 1.2358170042858774e-13, "grad_norm": 0.6146364808082581, "learning_rate": 7.959765208197979e-05, "loss": 2.307, "step": 35620 }, { "epoch": 1.2361639489810727e-13, "grad_norm": 0.5752798914909363, "learning_rate": 7.957570218132433e-05, "loss": 2.3156, "step": 35630 }, { "epoch": 1.236510893676268e-13, "grad_norm": 0.5937431454658508, "learning_rate": 7.955375056867727e-05, "loss": 2.2953, "step": 35640 }, { "epoch": 1.2368578383714635e-13, "grad_norm": 0.6029908061027527, "learning_rate": 7.953179724751004e-05, "loss": 2.3343, "step": 35650 }, { "epoch": 1.2372047830666588e-13, "grad_norm": 0.6168636679649353, "learning_rate": 7.95098422212944e-05, "loss": 2.3089, "step": 35660 }, { "epoch": 1.2375517277618542e-13, "grad_norm": 0.5951492190361023, "learning_rate": 7.948788549350233e-05, "loss": 2.3169, "step": 35670 }, { "epoch": 1.2378986724570495e-13, "grad_norm": 0.5915054678916931, "learning_rate": 7.946592706760611e-05, "loss": 2.2995, "step": 35680 }, { "epoch": 1.238245617152245e-13, "grad_norm": 0.5892814993858337, "learning_rate": 7.944396694707824e-05, "loss": 2.3685, "step": 35690 }, { "epoch": 1.2385925618474403e-13, "grad_norm": 0.6216738224029541, "learning_rate": 7.942200513539158e-05, "loss": 2.3104, "step": 35700 }, { "epoch": 1.2389395065426356e-13, "grad_norm": 0.5814478993415833, "learning_rate": 7.940004163601914e-05, "loss": 2.3323, "step": 35710 }, { "epoch": 1.239286451237831e-13, "grad_norm": 0.6138721704483032, "learning_rate": 7.937807645243433e-05, "loss": 2.2925, "step": 35720 }, { "epoch": 1.2396333959330263e-13, "grad_norm": 0.6168862581253052, "learning_rate": 7.93561095881107e-05, "loss": 2.3325, "step": 35730 }, { "epoch": 1.2399803406282217e-13, "grad_norm": 0.615189254283905, "learning_rate": 7.933414104652213e-05, "loss": 2.3418, "step": 35740 }, { "epoch": 1.240327285323417e-13, "grad_norm": 0.6090540885925293, "learning_rate": 7.931217083114277e-05, "loss": 2.3277, "step": 35750 }, { "epoch": 1.2406742300186124e-13, "grad_norm": 0.569099485874176, "learning_rate": 7.929019894544703e-05, "loss": 2.3168, "step": 35760 }, { "epoch": 1.2410211747138078e-13, "grad_norm": 0.6032038331031799, "learning_rate": 7.926822539290952e-05, "loss": 2.3147, "step": 35770 }, { "epoch": 1.2413681194090032e-13, "grad_norm": 0.5917912125587463, "learning_rate": 7.924625017700523e-05, "loss": 2.311, "step": 35780 }, { "epoch": 1.2417150641041985e-13, "grad_norm": 0.6024013757705688, "learning_rate": 7.922427330120931e-05, "loss": 2.2985, "step": 35790 }, { "epoch": 1.242062008799394e-13, "grad_norm": 0.6017213463783264, "learning_rate": 7.920229476899723e-05, "loss": 2.2687, "step": 35800 }, { "epoch": 1.2424089534945892e-13, "grad_norm": 0.6229382157325745, "learning_rate": 7.918031458384474e-05, "loss": 2.3241, "step": 35810 }, { "epoch": 1.2427558981897846e-13, "grad_norm": 0.6019366383552551, "learning_rate": 7.915833274922777e-05, "loss": 2.3163, "step": 35820 }, { "epoch": 1.24310284288498e-13, "grad_norm": 0.5955382585525513, "learning_rate": 7.913634926862255e-05, "loss": 2.2949, "step": 35830 }, { "epoch": 1.2434497875801753e-13, "grad_norm": 0.6018106937408447, "learning_rate": 7.911436414550563e-05, "loss": 2.3032, "step": 35840 }, { "epoch": 1.2437967322753707e-13, "grad_norm": 0.605509340763092, "learning_rate": 7.909237738335371e-05, "loss": 2.3445, "step": 35850 }, { "epoch": 1.244143676970566e-13, "grad_norm": 0.6093212366104126, "learning_rate": 7.907038898564385e-05, "loss": 2.3059, "step": 35860 }, { "epoch": 1.2444906216657614e-13, "grad_norm": 0.5713951587677002, "learning_rate": 7.904839895585329e-05, "loss": 2.3311, "step": 35870 }, { "epoch": 1.2448375663609568e-13, "grad_norm": 0.5945079922676086, "learning_rate": 7.90264072974596e-05, "loss": 2.2998, "step": 35880 }, { "epoch": 1.2451845110561521e-13, "grad_norm": 0.5727508664131165, "learning_rate": 7.900441401394057e-05, "loss": 2.3145, "step": 35890 }, { "epoch": 1.2455314557513475e-13, "grad_norm": 0.5990481376647949, "learning_rate": 7.898241910877421e-05, "loss": 2.306, "step": 35900 }, { "epoch": 1.2458784004465429e-13, "grad_norm": 0.5969821214675903, "learning_rate": 7.896042258543885e-05, "loss": 2.3262, "step": 35910 }, { "epoch": 1.2462253451417382e-13, "grad_norm": 0.5936622023582458, "learning_rate": 7.893842444741306e-05, "loss": 2.3228, "step": 35920 }, { "epoch": 1.2465722898369336e-13, "grad_norm": 0.6057032346725464, "learning_rate": 7.891642469817566e-05, "loss": 2.3002, "step": 35930 }, { "epoch": 1.246919234532129e-13, "grad_norm": 0.6086731553077698, "learning_rate": 7.889442334120569e-05, "loss": 2.3135, "step": 35940 }, { "epoch": 1.2472661792273243e-13, "grad_norm": 0.586713969707489, "learning_rate": 7.887242037998248e-05, "loss": 2.3088, "step": 35950 }, { "epoch": 1.2476131239225197e-13, "grad_norm": 0.6089972853660583, "learning_rate": 7.885041581798564e-05, "loss": 2.3161, "step": 35960 }, { "epoch": 1.247960068617715e-13, "grad_norm": 0.6057736277580261, "learning_rate": 7.882840965869497e-05, "loss": 2.3048, "step": 35970 }, { "epoch": 1.2483070133129104e-13, "grad_norm": 0.5827393531799316, "learning_rate": 7.880640190559059e-05, "loss": 2.3317, "step": 35980 }, { "epoch": 1.2486539580081057e-13, "grad_norm": 0.6163725256919861, "learning_rate": 7.878439256215281e-05, "loss": 2.3088, "step": 35990 }, { "epoch": 1.249000902703301e-13, "grad_norm": 0.625928521156311, "learning_rate": 7.876238163186222e-05, "loss": 2.2789, "step": 36000 }, { "epoch": 1.2493478473984965e-13, "grad_norm": 0.6282992959022522, "learning_rate": 7.874036911819964e-05, "loss": 2.3322, "step": 36010 }, { "epoch": 1.2496947920936918e-13, "grad_norm": 0.6050147414207458, "learning_rate": 7.871835502464621e-05, "loss": 2.2834, "step": 36020 }, { "epoch": 1.2500417367888872e-13, "grad_norm": 0.5953363180160522, "learning_rate": 7.869633935468322e-05, "loss": 2.3383, "step": 36030 }, { "epoch": 1.2503886814840826e-13, "grad_norm": 0.6266483068466187, "learning_rate": 7.867432211179229e-05, "loss": 2.3197, "step": 36040 }, { "epoch": 1.250735626179278e-13, "grad_norm": 0.605323851108551, "learning_rate": 7.865230329945524e-05, "loss": 2.2938, "step": 36050 }, { "epoch": 1.2510825708744733e-13, "grad_norm": 0.612338662147522, "learning_rate": 7.863028292115417e-05, "loss": 2.3321, "step": 36060 }, { "epoch": 1.2514295155696686e-13, "grad_norm": 0.6106471419334412, "learning_rate": 7.86082609803714e-05, "loss": 2.3193, "step": 36070 }, { "epoch": 1.251776460264864e-13, "grad_norm": 0.6159601211547852, "learning_rate": 7.858623748058954e-05, "loss": 2.2936, "step": 36080 }, { "epoch": 1.2521234049600594e-13, "grad_norm": 0.6316839456558228, "learning_rate": 7.856421242529138e-05, "loss": 2.2899, "step": 36090 }, { "epoch": 1.2524703496552547e-13, "grad_norm": 0.598496675491333, "learning_rate": 7.854218581795999e-05, "loss": 2.3206, "step": 36100 }, { "epoch": 1.25281729435045e-13, "grad_norm": 0.5904721021652222, "learning_rate": 7.85201576620787e-05, "loss": 2.3496, "step": 36110 }, { "epoch": 1.2531642390456454e-13, "grad_norm": 0.6277822852134705, "learning_rate": 7.84981279611311e-05, "loss": 2.3112, "step": 36120 }, { "epoch": 1.2535111837408408e-13, "grad_norm": 0.6200352907180786, "learning_rate": 7.847609671860097e-05, "loss": 2.2842, "step": 36130 }, { "epoch": 1.2538581284360362e-13, "grad_norm": 0.5984392762184143, "learning_rate": 7.845406393797237e-05, "loss": 2.3186, "step": 36140 }, { "epoch": 1.2542050731312315e-13, "grad_norm": 0.625886082649231, "learning_rate": 7.84320296227296e-05, "loss": 2.3281, "step": 36150 }, { "epoch": 1.254552017826427e-13, "grad_norm": 0.5917491912841797, "learning_rate": 7.840999377635717e-05, "loss": 2.34, "step": 36160 }, { "epoch": 1.2548989625216223e-13, "grad_norm": 0.616455078125, "learning_rate": 7.838795640233989e-05, "loss": 2.3558, "step": 36170 }, { "epoch": 1.2552459072168176e-13, "grad_norm": 0.6031914353370667, "learning_rate": 7.836591750416275e-05, "loss": 2.3106, "step": 36180 }, { "epoch": 1.255592851912013e-13, "grad_norm": 0.6058579087257385, "learning_rate": 7.834387708531106e-05, "loss": 2.3441, "step": 36190 }, { "epoch": 1.2559397966072083e-13, "grad_norm": 0.5961235761642456, "learning_rate": 7.832183514927027e-05, "loss": 2.3452, "step": 36200 }, { "epoch": 1.2562867413024037e-13, "grad_norm": 0.6017611026763916, "learning_rate": 7.829979169952616e-05, "loss": 2.3162, "step": 36210 }, { "epoch": 1.256633685997599e-13, "grad_norm": 0.5636517405509949, "learning_rate": 7.827774673956467e-05, "loss": 2.3004, "step": 36220 }, { "epoch": 1.2569806306927944e-13, "grad_norm": 0.5987327694892883, "learning_rate": 7.825570027287208e-05, "loss": 2.2937, "step": 36230 }, { "epoch": 1.2573275753879898e-13, "grad_norm": 0.6080207228660583, "learning_rate": 7.823365230293482e-05, "loss": 2.3079, "step": 36240 }, { "epoch": 1.2576745200831851e-13, "grad_norm": 0.5991127490997314, "learning_rate": 7.821160283323957e-05, "loss": 2.2956, "step": 36250 }, { "epoch": 1.2580214647783805e-13, "grad_norm": 0.6068345904350281, "learning_rate": 7.818955186727327e-05, "loss": 2.3416, "step": 36260 }, { "epoch": 1.2583684094735759e-13, "grad_norm": 0.6201028823852539, "learning_rate": 7.81674994085231e-05, "loss": 2.3131, "step": 36270 }, { "epoch": 1.2587153541687712e-13, "grad_norm": 0.6234039664268494, "learning_rate": 7.814544546047648e-05, "loss": 2.2953, "step": 36280 }, { "epoch": 1.2590622988639666e-13, "grad_norm": 0.6292524337768555, "learning_rate": 7.812339002662102e-05, "loss": 2.3102, "step": 36290 }, { "epoch": 1.259409243559162e-13, "grad_norm": 0.599340558052063, "learning_rate": 7.810133311044464e-05, "loss": 2.3572, "step": 36300 }, { "epoch": 1.2597561882543573e-13, "grad_norm": 0.6114088892936707, "learning_rate": 7.807927471543541e-05, "loss": 2.2818, "step": 36310 }, { "epoch": 1.2601031329495527e-13, "grad_norm": 0.5809634923934937, "learning_rate": 7.80572148450817e-05, "loss": 2.2746, "step": 36320 }, { "epoch": 1.260450077644748e-13, "grad_norm": 0.5918647050857544, "learning_rate": 7.803515350287206e-05, "loss": 2.2993, "step": 36330 }, { "epoch": 1.2607970223399434e-13, "grad_norm": 0.6175738573074341, "learning_rate": 7.801309069229531e-05, "loss": 2.3166, "step": 36340 }, { "epoch": 1.2611439670351388e-13, "grad_norm": 0.6896723508834839, "learning_rate": 7.799102641684054e-05, "loss": 2.2787, "step": 36350 }, { "epoch": 1.261490911730334e-13, "grad_norm": 0.6165565252304077, "learning_rate": 7.796896067999697e-05, "loss": 2.3032, "step": 36360 }, { "epoch": 1.2618378564255295e-13, "grad_norm": 0.5855021476745605, "learning_rate": 7.794689348525413e-05, "loss": 2.2921, "step": 36370 }, { "epoch": 1.2621848011207248e-13, "grad_norm": 0.6182008981704712, "learning_rate": 7.792482483610176e-05, "loss": 2.295, "step": 36380 }, { "epoch": 1.2625317458159202e-13, "grad_norm": 0.6349795460700989, "learning_rate": 7.790275473602982e-05, "loss": 2.298, "step": 36390 }, { "epoch": 1.2628786905111156e-13, "grad_norm": 0.5821319222450256, "learning_rate": 7.788068318852849e-05, "loss": 2.3188, "step": 36400 }, { "epoch": 1.263225635206311e-13, "grad_norm": 0.6105663180351257, "learning_rate": 7.785861019708824e-05, "loss": 2.3256, "step": 36410 }, { "epoch": 1.2635725799015063e-13, "grad_norm": 0.6063022017478943, "learning_rate": 7.783653576519965e-05, "loss": 2.2954, "step": 36420 }, { "epoch": 1.2639195245967016e-13, "grad_norm": 0.6263741850852966, "learning_rate": 7.781445989635366e-05, "loss": 2.351, "step": 36430 }, { "epoch": 1.264266469291897e-13, "grad_norm": 0.5968576669692993, "learning_rate": 7.779238259404139e-05, "loss": 2.3304, "step": 36440 }, { "epoch": 1.2646134139870924e-13, "grad_norm": 0.6310900449752808, "learning_rate": 7.777030386175414e-05, "loss": 2.3084, "step": 36450 }, { "epoch": 1.2649603586822877e-13, "grad_norm": 0.6014912724494934, "learning_rate": 7.774822370298347e-05, "loss": 2.2946, "step": 36460 }, { "epoch": 1.265307303377483e-13, "grad_norm": 0.6027376651763916, "learning_rate": 7.77261421212212e-05, "loss": 2.2847, "step": 36470 }, { "epoch": 1.2656542480726785e-13, "grad_norm": 0.6043545007705688, "learning_rate": 7.77040591199593e-05, "loss": 2.3047, "step": 36480 }, { "epoch": 1.2660011927678738e-13, "grad_norm": 0.614563524723053, "learning_rate": 7.768197470269004e-05, "loss": 2.3236, "step": 36490 }, { "epoch": 1.2663481374630692e-13, "grad_norm": 0.5711690783500671, "learning_rate": 7.765988887290585e-05, "loss": 2.3016, "step": 36500 }, { "epoch": 1.2666950821582645e-13, "grad_norm": 0.6207473278045654, "learning_rate": 7.763780163409944e-05, "loss": 2.3401, "step": 36510 }, { "epoch": 1.26704202685346e-13, "grad_norm": 0.6009278893470764, "learning_rate": 7.761571298976371e-05, "loss": 2.2956, "step": 36520 }, { "epoch": 1.2673889715486553e-13, "grad_norm": 0.627299964427948, "learning_rate": 7.759362294339179e-05, "loss": 2.3232, "step": 36530 }, { "epoch": 1.2677359162438506e-13, "grad_norm": 0.6158077716827393, "learning_rate": 7.7571531498477e-05, "loss": 2.2857, "step": 36540 }, { "epoch": 1.268082860939046e-13, "grad_norm": 0.6299452185630798, "learning_rate": 7.7549438658513e-05, "loss": 2.3056, "step": 36550 }, { "epoch": 1.2684298056342413e-13, "grad_norm": 0.6028612852096558, "learning_rate": 7.752734442699348e-05, "loss": 2.3117, "step": 36560 }, { "epoch": 1.2687767503294367e-13, "grad_norm": 0.6219512224197388, "learning_rate": 7.750524880741251e-05, "loss": 2.2781, "step": 36570 }, { "epoch": 1.269123695024632e-13, "grad_norm": 0.6690372228622437, "learning_rate": 7.748315180326428e-05, "loss": 2.3185, "step": 36580 }, { "epoch": 1.2694706397198274e-13, "grad_norm": 0.5839430689811707, "learning_rate": 7.746105341804329e-05, "loss": 2.2634, "step": 36590 }, { "epoch": 1.2698175844150228e-13, "grad_norm": 0.6059889197349548, "learning_rate": 7.743895365524418e-05, "loss": 2.2661, "step": 36600 }, { "epoch": 1.2701645291102182e-13, "grad_norm": 0.5999433994293213, "learning_rate": 7.741685251836186e-05, "loss": 2.3147, "step": 36610 }, { "epoch": 1.2705114738054135e-13, "grad_norm": 0.6178340911865234, "learning_rate": 7.739475001089142e-05, "loss": 2.3098, "step": 36620 }, { "epoch": 1.270858418500609e-13, "grad_norm": 0.6218565702438354, "learning_rate": 7.737264613632818e-05, "loss": 2.323, "step": 36630 }, { "epoch": 1.2712053631958042e-13, "grad_norm": 0.6134302020072937, "learning_rate": 7.735054089816764e-05, "loss": 2.3147, "step": 36640 }, { "epoch": 1.2715523078909996e-13, "grad_norm": 0.5972164869308472, "learning_rate": 7.732843429990562e-05, "loss": 2.2992, "step": 36650 }, { "epoch": 1.271899252586195e-13, "grad_norm": 0.6023012399673462, "learning_rate": 7.730632634503805e-05, "loss": 2.307, "step": 36660 }, { "epoch": 1.2722461972813903e-13, "grad_norm": 0.6164330840110779, "learning_rate": 7.728421703706113e-05, "loss": 2.315, "step": 36670 }, { "epoch": 1.2725931419765857e-13, "grad_norm": 0.6095483899116516, "learning_rate": 7.726210637947127e-05, "loss": 2.3158, "step": 36680 }, { "epoch": 1.272940086671781e-13, "grad_norm": 0.6063002943992615, "learning_rate": 7.723999437576502e-05, "loss": 2.3281, "step": 36690 }, { "epoch": 1.2732870313669764e-13, "grad_norm": 0.5768128633499146, "learning_rate": 7.721788102943927e-05, "loss": 2.2954, "step": 36700 }, { "epoch": 1.2736339760621718e-13, "grad_norm": 0.6308834552764893, "learning_rate": 7.7195766343991e-05, "loss": 2.3218, "step": 36710 }, { "epoch": 1.2739809207573671e-13, "grad_norm": 0.618783175945282, "learning_rate": 7.717365032291751e-05, "loss": 2.3066, "step": 36720 }, { "epoch": 1.2743278654525625e-13, "grad_norm": 0.6069857478141785, "learning_rate": 7.71515329697162e-05, "loss": 2.288, "step": 36730 }, { "epoch": 1.2746748101477579e-13, "grad_norm": 0.5987468957901001, "learning_rate": 7.712941428788481e-05, "loss": 2.3032, "step": 36740 }, { "epoch": 1.2750217548429532e-13, "grad_norm": 0.6213812232017517, "learning_rate": 7.710729428092116e-05, "loss": 2.3537, "step": 36750 }, { "epoch": 1.2753686995381486e-13, "grad_norm": 0.6154086589813232, "learning_rate": 7.708517295232337e-05, "loss": 2.3125, "step": 36760 }, { "epoch": 1.275715644233344e-13, "grad_norm": 0.6330193877220154, "learning_rate": 7.706305030558972e-05, "loss": 2.2928, "step": 36770 }, { "epoch": 1.2760625889285393e-13, "grad_norm": 0.6069002151489258, "learning_rate": 7.704092634421872e-05, "loss": 2.2661, "step": 36780 }, { "epoch": 1.2764095336237347e-13, "grad_norm": 0.6050984859466553, "learning_rate": 7.70188010717091e-05, "loss": 2.3144, "step": 36790 }, { "epoch": 1.27675647831893e-13, "grad_norm": 0.5956056118011475, "learning_rate": 7.699667449155977e-05, "loss": 2.3162, "step": 36800 }, { "epoch": 1.2771034230141254e-13, "grad_norm": 0.586917519569397, "learning_rate": 7.697454660726982e-05, "loss": 2.3251, "step": 36810 }, { "epoch": 1.2774503677093207e-13, "grad_norm": 0.6264936923980713, "learning_rate": 7.695241742233866e-05, "loss": 2.2808, "step": 36820 }, { "epoch": 1.277797312404516e-13, "grad_norm": 0.6059373021125793, "learning_rate": 7.693028694026579e-05, "loss": 2.3282, "step": 36830 }, { "epoch": 1.2781442570997115e-13, "grad_norm": 0.6147292852401733, "learning_rate": 7.690815516455096e-05, "loss": 2.3317, "step": 36840 }, { "epoch": 1.2784912017949068e-13, "grad_norm": 0.6351568102836609, "learning_rate": 7.68860220986941e-05, "loss": 2.3362, "step": 36850 }, { "epoch": 1.2788381464901022e-13, "grad_norm": 0.6137797236442566, "learning_rate": 7.686388774619537e-05, "loss": 2.3051, "step": 36860 }, { "epoch": 1.2791850911852976e-13, "grad_norm": 1.2548397779464722, "learning_rate": 7.684175211055517e-05, "loss": 2.3249, "step": 36870 }, { "epoch": 1.279532035880493e-13, "grad_norm": 0.6052848100662231, "learning_rate": 7.6819615195274e-05, "loss": 2.3297, "step": 36880 }, { "epoch": 1.2798789805756883e-13, "grad_norm": 0.6110605001449585, "learning_rate": 7.679747700385265e-05, "loss": 2.3268, "step": 36890 }, { "epoch": 1.2802259252708836e-13, "grad_norm": 0.6186817288398743, "learning_rate": 7.67753375397921e-05, "loss": 2.3094, "step": 36900 }, { "epoch": 1.280572869966079e-13, "grad_norm": 0.6252396702766418, "learning_rate": 7.67531968065935e-05, "loss": 2.3195, "step": 36910 }, { "epoch": 1.2809198146612744e-13, "grad_norm": 0.6048965454101562, "learning_rate": 7.673105480775822e-05, "loss": 2.2974, "step": 36920 }, { "epoch": 1.2812667593564697e-13, "grad_norm": 0.6076198816299438, "learning_rate": 7.67089115467878e-05, "loss": 2.3106, "step": 36930 }, { "epoch": 1.281613704051665e-13, "grad_norm": 0.5931867957115173, "learning_rate": 7.668676702718407e-05, "loss": 2.3283, "step": 36940 }, { "epoch": 1.2819606487468604e-13, "grad_norm": 0.6101574897766113, "learning_rate": 7.666462125244892e-05, "loss": 2.3255, "step": 36950 }, { "epoch": 1.2823075934420558e-13, "grad_norm": 0.6132187843322754, "learning_rate": 7.664247422608455e-05, "loss": 2.2783, "step": 36960 }, { "epoch": 1.2826545381372512e-13, "grad_norm": 0.6032510995864868, "learning_rate": 7.662032595159333e-05, "loss": 2.3193, "step": 36970 }, { "epoch": 1.2830014828324465e-13, "grad_norm": 0.6114382147789001, "learning_rate": 7.659817643247781e-05, "loss": 2.308, "step": 36980 }, { "epoch": 1.283348427527642e-13, "grad_norm": 0.6223697662353516, "learning_rate": 7.657602567224075e-05, "loss": 2.3673, "step": 36990 }, { "epoch": 1.2836953722228372e-13, "grad_norm": 0.5863919258117676, "learning_rate": 7.65538736743851e-05, "loss": 2.3027, "step": 37000 }, { "epoch": 1.2840423169180326e-13, "grad_norm": 0.6228392124176025, "learning_rate": 7.6531720442414e-05, "loss": 2.2832, "step": 37010 }, { "epoch": 1.284389261613228e-13, "grad_norm": 0.6233537197113037, "learning_rate": 7.650956597983077e-05, "loss": 2.3039, "step": 37020 }, { "epoch": 1.2847362063084233e-13, "grad_norm": 0.6201449632644653, "learning_rate": 7.648741029013902e-05, "loss": 2.2881, "step": 37030 }, { "epoch": 1.2850831510036187e-13, "grad_norm": 0.6037846803665161, "learning_rate": 7.646525337684241e-05, "loss": 2.3079, "step": 37040 }, { "epoch": 1.285430095698814e-13, "grad_norm": 0.6248581409454346, "learning_rate": 7.644309524344489e-05, "loss": 2.2775, "step": 37050 }, { "epoch": 1.2857770403940094e-13, "grad_norm": 0.6139913201332092, "learning_rate": 7.642093589345059e-05, "loss": 2.2715, "step": 37060 }, { "epoch": 1.2861239850892048e-13, "grad_norm": 0.6071272492408752, "learning_rate": 7.639877533036379e-05, "loss": 2.374, "step": 37070 }, { "epoch": 1.2864709297844001e-13, "grad_norm": 0.5987386107444763, "learning_rate": 7.637661355768905e-05, "loss": 2.3008, "step": 37080 }, { "epoch": 1.2868178744795955e-13, "grad_norm": 0.6196722388267517, "learning_rate": 7.6354450578931e-05, "loss": 2.3153, "step": 37090 }, { "epoch": 1.2871648191747909e-13, "grad_norm": 0.61617112159729, "learning_rate": 7.633228639759455e-05, "loss": 2.312, "step": 37100 }, { "epoch": 1.2875117638699862e-13, "grad_norm": 0.6440759897232056, "learning_rate": 7.631012101718479e-05, "loss": 2.2903, "step": 37110 }, { "epoch": 1.2878587085651816e-13, "grad_norm": 0.626526951789856, "learning_rate": 7.628795444120697e-05, "loss": 2.2573, "step": 37120 }, { "epoch": 1.288205653260377e-13, "grad_norm": 0.6160299181938171, "learning_rate": 7.626578667316651e-05, "loss": 2.3079, "step": 37130 }, { "epoch": 1.2885525979555723e-13, "grad_norm": 0.5936102867126465, "learning_rate": 7.624361771656911e-05, "loss": 2.325, "step": 37140 }, { "epoch": 1.2888995426507677e-13, "grad_norm": 0.6160911321640015, "learning_rate": 7.622144757492057e-05, "loss": 2.3209, "step": 37150 }, { "epoch": 1.289246487345963e-13, "grad_norm": 0.6111903190612793, "learning_rate": 7.619927625172692e-05, "loss": 2.2974, "step": 37160 }, { "epoch": 1.2895934320411584e-13, "grad_norm": 0.6349260210990906, "learning_rate": 7.617710375049435e-05, "loss": 2.3305, "step": 37170 }, { "epoch": 1.2899403767363538e-13, "grad_norm": 0.6367468237876892, "learning_rate": 7.615493007472923e-05, "loss": 2.3391, "step": 37180 }, { "epoch": 1.290287321431549e-13, "grad_norm": 0.5937991738319397, "learning_rate": 7.61327552279382e-05, "loss": 2.3173, "step": 37190 }, { "epoch": 1.2906342661267445e-13, "grad_norm": 0.6296828985214233, "learning_rate": 7.611057921362796e-05, "loss": 2.3057, "step": 37200 }, { "epoch": 1.2909812108219398e-13, "grad_norm": 0.6053290367126465, "learning_rate": 7.608840203530547e-05, "loss": 2.2657, "step": 37210 }, { "epoch": 1.2913281555171352e-13, "grad_norm": 0.5945888161659241, "learning_rate": 7.606622369647789e-05, "loss": 2.3147, "step": 37220 }, { "epoch": 1.2916751002123306e-13, "grad_norm": 0.6504223942756653, "learning_rate": 7.604404420065249e-05, "loss": 2.2915, "step": 37230 }, { "epoch": 1.292022044907526e-13, "grad_norm": 0.648664653301239, "learning_rate": 7.60218635513368e-05, "loss": 2.3245, "step": 37240 }, { "epoch": 1.2923689896027213e-13, "grad_norm": 0.6109389066696167, "learning_rate": 7.599968175203848e-05, "loss": 2.3391, "step": 37250 }, { "epoch": 1.2927159342979166e-13, "grad_norm": 0.6254541873931885, "learning_rate": 7.597749880626541e-05, "loss": 2.3427, "step": 37260 }, { "epoch": 1.293062878993112e-13, "grad_norm": 0.6082358956336975, "learning_rate": 7.59553147175256e-05, "loss": 2.3101, "step": 37270 }, { "epoch": 1.2934098236883074e-13, "grad_norm": 0.6259357333183289, "learning_rate": 7.59331294893273e-05, "loss": 2.2963, "step": 37280 }, { "epoch": 1.2937567683835027e-13, "grad_norm": 0.6010357141494751, "learning_rate": 7.59109431251789e-05, "loss": 2.2972, "step": 37290 }, { "epoch": 1.294103713078698e-13, "grad_norm": 0.6209099292755127, "learning_rate": 7.588875562858899e-05, "loss": 2.344, "step": 37300 }, { "epoch": 1.2944506577738935e-13, "grad_norm": 0.5946083664894104, "learning_rate": 7.586656700306633e-05, "loss": 2.2918, "step": 37310 }, { "epoch": 1.2947976024690888e-13, "grad_norm": 0.6152932047843933, "learning_rate": 7.584437725211986e-05, "loss": 2.3049, "step": 37320 }, { "epoch": 1.2951445471642842e-13, "grad_norm": 0.6205236911773682, "learning_rate": 7.58221863792587e-05, "loss": 2.2757, "step": 37330 }, { "epoch": 1.2954914918594795e-13, "grad_norm": 0.6335858702659607, "learning_rate": 7.579999438799212e-05, "loss": 2.3488, "step": 37340 }, { "epoch": 1.295838436554675e-13, "grad_norm": 0.6297491788864136, "learning_rate": 7.577780128182963e-05, "loss": 2.3191, "step": 37350 }, { "epoch": 1.2961853812498703e-13, "grad_norm": 0.5890154838562012, "learning_rate": 7.575560706428087e-05, "loss": 2.3483, "step": 37360 }, { "epoch": 1.2965323259450656e-13, "grad_norm": 0.6185715198516846, "learning_rate": 7.573341173885563e-05, "loss": 2.3172, "step": 37370 }, { "epoch": 1.296879270640261e-13, "grad_norm": 0.5989314913749695, "learning_rate": 7.571121530906397e-05, "loss": 2.3567, "step": 37380 }, { "epoch": 1.2972262153354563e-13, "grad_norm": 0.6163512468338013, "learning_rate": 7.5689017778416e-05, "loss": 2.3065, "step": 37390 }, { "epoch": 1.2975731600306517e-13, "grad_norm": 0.6195485591888428, "learning_rate": 7.566681915042213e-05, "loss": 2.3395, "step": 37400 }, { "epoch": 1.297920104725847e-13, "grad_norm": 0.588651716709137, "learning_rate": 7.564461942859282e-05, "loss": 2.3347, "step": 37410 }, { "epoch": 1.2982670494210424e-13, "grad_norm": 0.6207393407821655, "learning_rate": 7.562241861643882e-05, "loss": 2.2851, "step": 37420 }, { "epoch": 1.2986139941162378e-13, "grad_norm": 0.5779405236244202, "learning_rate": 7.560021671747099e-05, "loss": 2.2976, "step": 37430 }, { "epoch": 1.2989609388114332e-13, "grad_norm": 0.6067055463790894, "learning_rate": 7.557801373520031e-05, "loss": 2.3247, "step": 37440 }, { "epoch": 1.2993078835066285e-13, "grad_norm": 0.6015934348106384, "learning_rate": 7.555580967313807e-05, "loss": 2.2898, "step": 37450 }, { "epoch": 1.299654828201824e-13, "grad_norm": 0.6269647479057312, "learning_rate": 7.55336045347956e-05, "loss": 2.3207, "step": 37460 }, { "epoch": 1.3000017728970192e-13, "grad_norm": 0.5939361453056335, "learning_rate": 7.551139832368446e-05, "loss": 2.3098, "step": 37470 }, { "epoch": 1.3003487175922146e-13, "grad_norm": 0.6105335354804993, "learning_rate": 7.54891910433164e-05, "loss": 2.3411, "step": 37480 }, { "epoch": 1.30069566228741e-13, "grad_norm": 0.5953258275985718, "learning_rate": 7.546698269720327e-05, "loss": 2.3175, "step": 37490 }, { "epoch": 1.3010426069826053e-13, "grad_norm": 0.6148907542228699, "learning_rate": 7.544477328885716e-05, "loss": 2.3032, "step": 37500 }, { "epoch": 1.3013895516778007e-13, "grad_norm": 0.6202893853187561, "learning_rate": 7.542256282179027e-05, "loss": 2.3541, "step": 37510 }, { "epoch": 1.301736496372996e-13, "grad_norm": 0.5872066617012024, "learning_rate": 7.540035129951501e-05, "loss": 2.3102, "step": 37520 }, { "epoch": 1.3020834410681914e-13, "grad_norm": 0.608401894569397, "learning_rate": 7.537813872554393e-05, "loss": 2.2943, "step": 37530 }, { "epoch": 1.3024303857633868e-13, "grad_norm": 0.5766288638114929, "learning_rate": 7.535592510338978e-05, "loss": 2.2936, "step": 37540 }, { "epoch": 1.3027773304585821e-13, "grad_norm": 0.6026673913002014, "learning_rate": 7.533371043656542e-05, "loss": 2.2864, "step": 37550 }, { "epoch": 1.3031242751537775e-13, "grad_norm": 0.6063592433929443, "learning_rate": 7.531149472858396e-05, "loss": 2.2988, "step": 37560 }, { "epoch": 1.3034712198489729e-13, "grad_norm": 0.6328795552253723, "learning_rate": 7.528927798295857e-05, "loss": 2.3578, "step": 37570 }, { "epoch": 1.3038181645441682e-13, "grad_norm": 0.6035875678062439, "learning_rate": 7.526706020320266e-05, "loss": 2.3149, "step": 37580 }, { "epoch": 1.3041651092393636e-13, "grad_norm": 0.6141307950019836, "learning_rate": 7.524484139282979e-05, "loss": 2.3067, "step": 37590 }, { "epoch": 1.304512053934559e-13, "grad_norm": 0.5928457975387573, "learning_rate": 7.522262155535363e-05, "loss": 2.2864, "step": 37600 }, { "epoch": 1.3048589986297543e-13, "grad_norm": 0.606218159198761, "learning_rate": 7.52004006942881e-05, "loss": 2.3106, "step": 37610 }, { "epoch": 1.3052059433249497e-13, "grad_norm": 0.6114345192909241, "learning_rate": 7.517817881314724e-05, "loss": 2.3083, "step": 37620 }, { "epoch": 1.305552888020145e-13, "grad_norm": 0.6183407306671143, "learning_rate": 7.515595591544524e-05, "loss": 2.2843, "step": 37630 }, { "epoch": 1.3058998327153404e-13, "grad_norm": 0.612787663936615, "learning_rate": 7.513373200469643e-05, "loss": 2.3093, "step": 37640 }, { "epoch": 1.3062467774105357e-13, "grad_norm": 0.674845278263092, "learning_rate": 7.511150708441537e-05, "loss": 2.3154, "step": 37650 }, { "epoch": 1.306593722105731e-13, "grad_norm": 0.6176592111587524, "learning_rate": 7.508928115811673e-05, "loss": 2.2949, "step": 37660 }, { "epoch": 1.3069406668009265e-13, "grad_norm": 0.6094842553138733, "learning_rate": 7.506705422931532e-05, "loss": 2.3383, "step": 37670 }, { "epoch": 1.3072876114961218e-13, "grad_norm": 0.6022486090660095, "learning_rate": 7.50448263015262e-05, "loss": 2.2874, "step": 37680 }, { "epoch": 1.3076345561913172e-13, "grad_norm": 0.6122115850448608, "learning_rate": 7.502259737826446e-05, "loss": 2.2918, "step": 37690 }, { "epoch": 1.3079815008865125e-13, "grad_norm": 0.6088767647743225, "learning_rate": 7.500036746304545e-05, "loss": 2.295, "step": 37700 }, { "epoch": 1.308328445581708e-13, "grad_norm": 0.6451987028121948, "learning_rate": 7.497813655938461e-05, "loss": 2.3125, "step": 37710 }, { "epoch": 1.3086753902769033e-13, "grad_norm": 0.6173652410507202, "learning_rate": 7.49559046707976e-05, "loss": 2.3016, "step": 37720 }, { "epoch": 1.3090223349720986e-13, "grad_norm": 0.6707383990287781, "learning_rate": 7.493367180080018e-05, "loss": 2.3344, "step": 37730 }, { "epoch": 1.309369279667294e-13, "grad_norm": 0.5967534780502319, "learning_rate": 7.49114379529083e-05, "loss": 2.3028, "step": 37740 }, { "epoch": 1.3097162243624894e-13, "grad_norm": 0.6143549680709839, "learning_rate": 7.488920313063806e-05, "loss": 2.2761, "step": 37750 }, { "epoch": 1.3100631690576847e-13, "grad_norm": 0.6018725633621216, "learning_rate": 7.486696733750567e-05, "loss": 2.3156, "step": 37760 }, { "epoch": 1.31041011375288e-13, "grad_norm": 0.6166786551475525, "learning_rate": 7.484473057702755e-05, "loss": 2.2713, "step": 37770 }, { "epoch": 1.3107570584480754e-13, "grad_norm": 0.6066465377807617, "learning_rate": 7.482249285272026e-05, "loss": 2.3228, "step": 37780 }, { "epoch": 1.3111040031432708e-13, "grad_norm": 0.6073412895202637, "learning_rate": 7.480025416810051e-05, "loss": 2.328, "step": 37790 }, { "epoch": 1.3114509478384662e-13, "grad_norm": 0.5881743431091309, "learning_rate": 7.477801452668513e-05, "loss": 2.3021, "step": 37800 }, { "epoch": 1.3117978925336615e-13, "grad_norm": 0.6155763864517212, "learning_rate": 7.475577393199115e-05, "loss": 2.3554, "step": 37810 }, { "epoch": 1.312144837228857e-13, "grad_norm": 0.5811368823051453, "learning_rate": 7.473353238753571e-05, "loss": 2.2852, "step": 37820 }, { "epoch": 1.3124917819240522e-13, "grad_norm": 0.700000524520874, "learning_rate": 7.471128989683613e-05, "loss": 2.3001, "step": 37830 }, { "epoch": 1.3128387266192476e-13, "grad_norm": 0.6637058258056641, "learning_rate": 7.468904646340989e-05, "loss": 2.268, "step": 37840 }, { "epoch": 1.313185671314443e-13, "grad_norm": 0.6185637712478638, "learning_rate": 7.466680209077454e-05, "loss": 2.3088, "step": 37850 }, { "epoch": 1.3135326160096383e-13, "grad_norm": 0.6016209721565247, "learning_rate": 7.46445567824479e-05, "loss": 2.2864, "step": 37860 }, { "epoch": 1.3138795607048337e-13, "grad_norm": 0.617913007736206, "learning_rate": 7.462231054194782e-05, "loss": 2.3176, "step": 37870 }, { "epoch": 1.314226505400029e-13, "grad_norm": 0.5979971885681152, "learning_rate": 7.46000633727924e-05, "loss": 2.314, "step": 37880 }, { "epoch": 1.3145734500952244e-13, "grad_norm": 0.6054664254188538, "learning_rate": 7.45778152784998e-05, "loss": 2.3143, "step": 37890 }, { "epoch": 1.3149203947904198e-13, "grad_norm": 0.6029255986213684, "learning_rate": 7.455556626258841e-05, "loss": 2.3024, "step": 37900 }, { "epoch": 1.3152673394856151e-13, "grad_norm": 0.5980185866355896, "learning_rate": 7.453331632857667e-05, "loss": 2.303, "step": 37910 }, { "epoch": 1.3156142841808105e-13, "grad_norm": 0.6161489486694336, "learning_rate": 7.451106547998322e-05, "loss": 2.2955, "step": 37920 }, { "epoch": 1.3159612288760059e-13, "grad_norm": 0.6199895739555359, "learning_rate": 7.448881372032686e-05, "loss": 2.2663, "step": 37930 }, { "epoch": 1.3163081735712012e-13, "grad_norm": 0.6492649912834167, "learning_rate": 7.446656105312651e-05, "loss": 2.2874, "step": 37940 }, { "epoch": 1.3166551182663966e-13, "grad_norm": 0.6084995269775391, "learning_rate": 7.444430748190126e-05, "loss": 2.3014, "step": 37950 }, { "epoch": 1.317002062961592e-13, "grad_norm": 0.6417068839073181, "learning_rate": 7.44220530101703e-05, "loss": 2.3218, "step": 37960 }, { "epoch": 1.3173490076567873e-13, "grad_norm": 0.6148139834403992, "learning_rate": 7.439979764145296e-05, "loss": 2.3476, "step": 37970 }, { "epoch": 1.3176959523519827e-13, "grad_norm": 0.645601749420166, "learning_rate": 7.437754137926876e-05, "loss": 2.3082, "step": 37980 }, { "epoch": 1.318042897047178e-13, "grad_norm": 0.5900384187698364, "learning_rate": 7.435528422713735e-05, "loss": 2.2795, "step": 37990 }, { "epoch": 1.3183898417423734e-13, "grad_norm": 0.6088452339172363, "learning_rate": 7.433302618857847e-05, "loss": 2.3141, "step": 38000 }, { "epoch": 1.3187367864375688e-13, "grad_norm": 0.6385581493377686, "learning_rate": 7.431076726711207e-05, "loss": 2.2756, "step": 38010 }, { "epoch": 1.319083731132764e-13, "grad_norm": 0.6093170046806335, "learning_rate": 7.428850746625819e-05, "loss": 2.2853, "step": 38020 }, { "epoch": 1.3194306758279595e-13, "grad_norm": 0.639186441898346, "learning_rate": 7.426624678953703e-05, "loss": 2.3207, "step": 38030 }, { "epoch": 1.3197776205231548e-13, "grad_norm": 0.6346942782402039, "learning_rate": 7.424398524046891e-05, "loss": 2.3016, "step": 38040 }, { "epoch": 1.3201245652183502e-13, "grad_norm": 0.6279592514038086, "learning_rate": 7.422172282257434e-05, "loss": 2.278, "step": 38050 }, { "epoch": 1.3204715099135456e-13, "grad_norm": 0.5936598777770996, "learning_rate": 7.41994595393739e-05, "loss": 2.2784, "step": 38060 }, { "epoch": 1.320818454608741e-13, "grad_norm": 0.6195902824401855, "learning_rate": 7.417719539438835e-05, "loss": 2.2998, "step": 38070 }, { "epoch": 1.3211653993039363e-13, "grad_norm": 0.6291100382804871, "learning_rate": 7.415493039113854e-05, "loss": 2.2909, "step": 38080 }, { "epoch": 1.3215123439991316e-13, "grad_norm": 0.5949569344520569, "learning_rate": 7.413266453314553e-05, "loss": 2.2889, "step": 38090 }, { "epoch": 1.321859288694327e-13, "grad_norm": 0.5924923419952393, "learning_rate": 7.411039782393046e-05, "loss": 2.2817, "step": 38100 }, { "epoch": 1.3222062333895224e-13, "grad_norm": 0.5906124114990234, "learning_rate": 7.408813026701462e-05, "loss": 2.319, "step": 38110 }, { "epoch": 1.3225531780847177e-13, "grad_norm": 0.6039515733718872, "learning_rate": 7.406586186591943e-05, "loss": 2.3133, "step": 38120 }, { "epoch": 1.322900122779913e-13, "grad_norm": 0.5903822779655457, "learning_rate": 7.404359262416645e-05, "loss": 2.2925, "step": 38130 }, { "epoch": 1.3232470674751085e-13, "grad_norm": 0.6362204551696777, "learning_rate": 7.402132254527734e-05, "loss": 2.286, "step": 38140 }, { "epoch": 1.3235940121703038e-13, "grad_norm": 0.596290111541748, "learning_rate": 7.399905163277395e-05, "loss": 2.2716, "step": 38150 }, { "epoch": 1.3239409568654992e-13, "grad_norm": 0.6069778203964233, "learning_rate": 7.397677989017825e-05, "loss": 2.2831, "step": 38160 }, { "epoch": 1.3242879015606945e-13, "grad_norm": 0.6069668531417847, "learning_rate": 7.395450732101228e-05, "loss": 2.2606, "step": 38170 }, { "epoch": 1.32463484625589e-13, "grad_norm": 0.6265324354171753, "learning_rate": 7.39322339287983e-05, "loss": 2.2997, "step": 38180 }, { "epoch": 1.3249817909510853e-13, "grad_norm": 0.601424515247345, "learning_rate": 7.390995971705862e-05, "loss": 2.3035, "step": 38190 }, { "epoch": 1.3253287356462806e-13, "grad_norm": 0.6037221550941467, "learning_rate": 7.388768468931572e-05, "loss": 2.2774, "step": 38200 }, { "epoch": 1.325675680341476e-13, "grad_norm": 0.6231350898742676, "learning_rate": 7.386540884909223e-05, "loss": 2.3175, "step": 38210 }, { "epoch": 1.3260226250366713e-13, "grad_norm": 0.6584796905517578, "learning_rate": 7.384313219991085e-05, "loss": 2.2506, "step": 38220 }, { "epoch": 1.3263695697318667e-13, "grad_norm": 0.6258801221847534, "learning_rate": 7.382085474529448e-05, "loss": 2.3074, "step": 38230 }, { "epoch": 1.326716514427062e-13, "grad_norm": 0.6262336373329163, "learning_rate": 7.379857648876605e-05, "loss": 2.2961, "step": 38240 }, { "epoch": 1.3270634591222574e-13, "grad_norm": 0.6118462681770325, "learning_rate": 7.377629743384872e-05, "loss": 2.2819, "step": 38250 }, { "epoch": 1.3274104038174528e-13, "grad_norm": 0.625617265701294, "learning_rate": 7.375401758406571e-05, "loss": 2.2922, "step": 38260 }, { "epoch": 1.3277573485126482e-13, "grad_norm": 0.654167115688324, "learning_rate": 7.373173694294042e-05, "loss": 2.3066, "step": 38270 }, { "epoch": 1.3281042932078435e-13, "grad_norm": 0.6246086359024048, "learning_rate": 7.370945551399628e-05, "loss": 2.306, "step": 38280 }, { "epoch": 1.3284512379030389e-13, "grad_norm": 0.6791098117828369, "learning_rate": 7.368717330075698e-05, "loss": 2.291, "step": 38290 }, { "epoch": 1.3287981825982342e-13, "grad_norm": 1.0688037872314453, "learning_rate": 7.366489030674618e-05, "loss": 2.3117, "step": 38300 }, { "epoch": 1.3291451272934296e-13, "grad_norm": 0.6132121682167053, "learning_rate": 7.364260653548781e-05, "loss": 2.279, "step": 38310 }, { "epoch": 1.329492071988625e-13, "grad_norm": 0.6275956034660339, "learning_rate": 7.362032199050584e-05, "loss": 2.312, "step": 38320 }, { "epoch": 1.3298390166838203e-13, "grad_norm": 0.6643993258476257, "learning_rate": 7.359803667532435e-05, "loss": 2.3279, "step": 38330 }, { "epoch": 1.3301859613790157e-13, "grad_norm": 0.6378934979438782, "learning_rate": 7.35757505934676e-05, "loss": 2.3038, "step": 38340 }, { "epoch": 1.330532906074211e-13, "grad_norm": 0.5938676595687866, "learning_rate": 7.355346374845994e-05, "loss": 2.2653, "step": 38350 }, { "epoch": 1.3308798507694064e-13, "grad_norm": 0.6195676326751709, "learning_rate": 7.353117614382583e-05, "loss": 2.3102, "step": 38360 }, { "epoch": 1.3312267954646018e-13, "grad_norm": 0.6030203700065613, "learning_rate": 7.350888778308988e-05, "loss": 2.3007, "step": 38370 }, { "epoch": 1.331573740159797e-13, "grad_norm": 0.5964051485061646, "learning_rate": 7.348659866977678e-05, "loss": 2.3149, "step": 38380 }, { "epoch": 1.3319206848549925e-13, "grad_norm": 0.602323055267334, "learning_rate": 7.346430880741139e-05, "loss": 2.3081, "step": 38390 }, { "epoch": 1.3322676295501878e-13, "grad_norm": 0.632226288318634, "learning_rate": 7.344201819951863e-05, "loss": 2.3065, "step": 38400 }, { "epoch": 1.3326145742453832e-13, "grad_norm": 0.6128791570663452, "learning_rate": 7.341972684962356e-05, "loss": 2.3258, "step": 38410 }, { "epoch": 1.3329615189405786e-13, "grad_norm": 0.6391457319259644, "learning_rate": 7.33974347612514e-05, "loss": 2.2891, "step": 38420 }, { "epoch": 1.333308463635774e-13, "grad_norm": 0.6299259662628174, "learning_rate": 7.337514193792747e-05, "loss": 2.3211, "step": 38430 }, { "epoch": 1.3336554083309693e-13, "grad_norm": 0.6272913813591003, "learning_rate": 7.335284838317713e-05, "loss": 2.3541, "step": 38440 }, { "epoch": 1.3340023530261647e-13, "grad_norm": 0.6029478311538696, "learning_rate": 7.333055410052594e-05, "loss": 2.2926, "step": 38450 }, { "epoch": 1.33434929772136e-13, "grad_norm": 0.6271802186965942, "learning_rate": 7.330825909349954e-05, "loss": 2.3394, "step": 38460 }, { "epoch": 1.3346962424165554e-13, "grad_norm": 0.6031838059425354, "learning_rate": 7.328596336562371e-05, "loss": 2.3039, "step": 38470 }, { "epoch": 1.3350431871117507e-13, "grad_norm": 0.5991477966308594, "learning_rate": 7.326366692042432e-05, "loss": 2.3303, "step": 38480 }, { "epoch": 1.335390131806946e-13, "grad_norm": 0.6044168472290039, "learning_rate": 7.324136976142737e-05, "loss": 2.2796, "step": 38490 }, { "epoch": 1.3357370765021415e-13, "grad_norm": 0.5898423194885254, "learning_rate": 7.321907189215894e-05, "loss": 2.2922, "step": 38500 }, { "epoch": 1.3360840211973368e-13, "grad_norm": 0.6347872018814087, "learning_rate": 7.319677331614526e-05, "loss": 2.3292, "step": 38510 }, { "epoch": 1.3364309658925322e-13, "grad_norm": 0.6247830986976624, "learning_rate": 7.317447403691267e-05, "loss": 2.2863, "step": 38520 }, { "epoch": 1.3367779105877275e-13, "grad_norm": 0.6485874056816101, "learning_rate": 7.31521740579876e-05, "loss": 2.3127, "step": 38530 }, { "epoch": 1.337124855282923e-13, "grad_norm": 0.6301644444465637, "learning_rate": 7.312987338289657e-05, "loss": 2.2992, "step": 38540 }, { "epoch": 1.3374717999781183e-13, "grad_norm": 0.633389949798584, "learning_rate": 7.310757201516632e-05, "loss": 2.2866, "step": 38550 }, { "epoch": 1.3378187446733136e-13, "grad_norm": 0.6127995848655701, "learning_rate": 7.308526995832351e-05, "loss": 2.3247, "step": 38560 }, { "epoch": 1.338165689368509e-13, "grad_norm": 0.643616259098053, "learning_rate": 7.306296721589509e-05, "loss": 2.317, "step": 38570 }, { "epoch": 1.3385126340637044e-13, "grad_norm": 0.6048904061317444, "learning_rate": 7.304066379140804e-05, "loss": 2.2859, "step": 38580 }, { "epoch": 1.3388595787588997e-13, "grad_norm": 0.5940736532211304, "learning_rate": 7.301835968838945e-05, "loss": 2.3237, "step": 38590 }, { "epoch": 1.339206523454095e-13, "grad_norm": 0.6173995137214661, "learning_rate": 7.299605491036651e-05, "loss": 2.3052, "step": 38600 }, { "epoch": 1.3395534681492904e-13, "grad_norm": 0.6732543110847473, "learning_rate": 7.297374946086653e-05, "loss": 2.328, "step": 38610 }, { "epoch": 1.3399004128444858e-13, "grad_norm": 0.6988582611083984, "learning_rate": 7.295144334341694e-05, "loss": 2.3236, "step": 38620 }, { "epoch": 1.3402473575396812e-13, "grad_norm": 0.6142970323562622, "learning_rate": 7.292913656154525e-05, "loss": 2.2935, "step": 38630 }, { "epoch": 1.3405943022348765e-13, "grad_norm": 0.6367724537849426, "learning_rate": 7.290682911877908e-05, "loss": 2.2954, "step": 38640 }, { "epoch": 1.340941246930072e-13, "grad_norm": 0.6547672748565674, "learning_rate": 7.288452101864617e-05, "loss": 2.3113, "step": 38650 }, { "epoch": 1.3412881916252672e-13, "grad_norm": 0.6236430406570435, "learning_rate": 7.286221226467436e-05, "loss": 2.3096, "step": 38660 }, { "epoch": 1.3416351363204626e-13, "grad_norm": 0.5777832269668579, "learning_rate": 7.283990286039156e-05, "loss": 2.264, "step": 38670 }, { "epoch": 1.341982081015658e-13, "grad_norm": 0.5944259166717529, "learning_rate": 7.281759280932584e-05, "loss": 2.2898, "step": 38680 }, { "epoch": 1.3423290257108533e-13, "grad_norm": 0.6304221749305725, "learning_rate": 7.279528211500534e-05, "loss": 2.3033, "step": 38690 }, { "epoch": 1.3426759704060487e-13, "grad_norm": 0.6166908740997314, "learning_rate": 7.277297078095829e-05, "loss": 2.2932, "step": 38700 }, { "epoch": 1.343022915101244e-13, "grad_norm": 0.6207402348518372, "learning_rate": 7.275065881071303e-05, "loss": 2.3075, "step": 38710 }, { "epoch": 1.3433698597964394e-13, "grad_norm": 0.6186254024505615, "learning_rate": 7.272834620779804e-05, "loss": 2.3068, "step": 38720 }, { "epoch": 1.3437168044916348e-13, "grad_norm": 0.5970202088356018, "learning_rate": 7.270603297574181e-05, "loss": 2.2737, "step": 38730 }, { "epoch": 1.3440637491868301e-13, "grad_norm": 0.6233776211738586, "learning_rate": 7.268371911807304e-05, "loss": 2.3123, "step": 38740 }, { "epoch": 1.3444106938820255e-13, "grad_norm": 0.6118336915969849, "learning_rate": 7.266140463832045e-05, "loss": 2.3074, "step": 38750 }, { "epoch": 1.3447576385772209e-13, "grad_norm": 0.5968732237815857, "learning_rate": 7.26390895400129e-05, "loss": 2.3248, "step": 38760 }, { "epoch": 1.3451045832724162e-13, "grad_norm": 0.6368276476860046, "learning_rate": 7.261677382667929e-05, "loss": 2.2957, "step": 38770 }, { "epoch": 1.3454515279676116e-13, "grad_norm": 0.6128309965133667, "learning_rate": 7.259445750184872e-05, "loss": 2.281, "step": 38780 }, { "epoch": 1.345798472662807e-13, "grad_norm": 0.644919753074646, "learning_rate": 7.257214056905026e-05, "loss": 2.3189, "step": 38790 }, { "epoch": 1.3461454173580023e-13, "grad_norm": 0.6394919753074646, "learning_rate": 7.254982303181318e-05, "loss": 2.3317, "step": 38800 }, { "epoch": 1.3464923620531977e-13, "grad_norm": 0.6119078993797302, "learning_rate": 7.252750489366678e-05, "loss": 2.2875, "step": 38810 }, { "epoch": 1.346839306748393e-13, "grad_norm": 0.6165124177932739, "learning_rate": 7.250518615814052e-05, "loss": 2.3041, "step": 38820 }, { "epoch": 1.3471862514435884e-13, "grad_norm": 0.5892268419265747, "learning_rate": 7.248286682876387e-05, "loss": 2.2945, "step": 38830 }, { "epoch": 1.3475331961387838e-13, "grad_norm": 0.6062673330307007, "learning_rate": 7.24605469090665e-05, "loss": 2.3084, "step": 38840 }, { "epoch": 1.347880140833979e-13, "grad_norm": 0.6434024572372437, "learning_rate": 7.243822640257805e-05, "loss": 2.2809, "step": 38850 }, { "epoch": 1.3482270855291745e-13, "grad_norm": 0.6032641530036926, "learning_rate": 7.241590531282837e-05, "loss": 2.2822, "step": 38860 }, { "epoch": 1.3485740302243698e-13, "grad_norm": 0.6190606951713562, "learning_rate": 7.239358364334728e-05, "loss": 2.3128, "step": 38870 }, { "epoch": 1.3489209749195652e-13, "grad_norm": 0.6546017527580261, "learning_rate": 7.237126139766484e-05, "loss": 2.2678, "step": 38880 }, { "epoch": 1.3492679196147606e-13, "grad_norm": 0.6412737965583801, "learning_rate": 7.234893857931105e-05, "loss": 2.3119, "step": 38890 }, { "epoch": 1.349614864309956e-13, "grad_norm": 0.6128380298614502, "learning_rate": 7.232661519181612e-05, "loss": 2.3177, "step": 38900 }, { "epoch": 1.3499618090051513e-13, "grad_norm": 0.6073296070098877, "learning_rate": 7.230429123871029e-05, "loss": 2.3006, "step": 38910 }, { "epoch": 1.3503087537003466e-13, "grad_norm": 0.5773903727531433, "learning_rate": 7.228196672352389e-05, "loss": 2.2453, "step": 38920 }, { "epoch": 1.350655698395542e-13, "grad_norm": 0.6141188740730286, "learning_rate": 7.225964164978735e-05, "loss": 2.2714, "step": 38930 }, { "epoch": 1.3510026430907374e-13, "grad_norm": 0.6325246691703796, "learning_rate": 7.22373160210312e-05, "loss": 2.3006, "step": 38940 }, { "epoch": 1.3513495877859327e-13, "grad_norm": 0.6051325798034668, "learning_rate": 7.221498984078603e-05, "loss": 2.2808, "step": 38950 }, { "epoch": 1.351696532481128e-13, "grad_norm": 0.628747820854187, "learning_rate": 7.219266311258255e-05, "loss": 2.3152, "step": 38960 }, { "epoch": 1.3520434771763234e-13, "grad_norm": 0.6240924000740051, "learning_rate": 7.217033583995156e-05, "loss": 2.2639, "step": 38970 }, { "epoch": 1.3523904218715188e-13, "grad_norm": 0.6156002879142761, "learning_rate": 7.214800802642386e-05, "loss": 2.2856, "step": 38980 }, { "epoch": 1.3527373665667142e-13, "grad_norm": 0.5972217917442322, "learning_rate": 7.212567967553045e-05, "loss": 2.2859, "step": 38990 }, { "epoch": 1.3530843112619095e-13, "grad_norm": 0.6216511726379395, "learning_rate": 7.210335079080237e-05, "loss": 2.2827, "step": 39000 }, { "epoch": 1.353431255957105e-13, "grad_norm": 0.614439845085144, "learning_rate": 7.208102137577073e-05, "loss": 2.2457, "step": 39010 }, { "epoch": 1.3537782006523003e-13, "grad_norm": 0.5978015661239624, "learning_rate": 7.205869143396672e-05, "loss": 2.3242, "step": 39020 }, { "epoch": 1.3541251453474956e-13, "grad_norm": 0.6114648580551147, "learning_rate": 7.203636096892165e-05, "loss": 2.3116, "step": 39030 }, { "epoch": 1.354472090042691e-13, "grad_norm": 0.6229153871536255, "learning_rate": 7.201402998416686e-05, "loss": 2.327, "step": 39040 }, { "epoch": 1.3548190347378863e-13, "grad_norm": 0.6266801953315735, "learning_rate": 7.199169848323385e-05, "loss": 2.314, "step": 39050 }, { "epoch": 1.3551659794330817e-13, "grad_norm": 0.6166887283325195, "learning_rate": 7.196936646965413e-05, "loss": 2.2965, "step": 39060 }, { "epoch": 1.355512924128277e-13, "grad_norm": 0.6035652160644531, "learning_rate": 7.194703394695932e-05, "loss": 2.2962, "step": 39070 }, { "epoch": 1.3558598688234724e-13, "grad_norm": 0.6092222332954407, "learning_rate": 7.19247009186811e-05, "loss": 2.2947, "step": 39080 }, { "epoch": 1.3562068135186678e-13, "grad_norm": 0.6388653516769409, "learning_rate": 7.190236738835124e-05, "loss": 2.2854, "step": 39090 }, { "epoch": 1.3565537582138631e-13, "grad_norm": 0.5954583883285522, "learning_rate": 7.188003335950165e-05, "loss": 2.2711, "step": 39100 }, { "epoch": 1.3569007029090585e-13, "grad_norm": 0.6145684123039246, "learning_rate": 7.185769883566422e-05, "loss": 2.2797, "step": 39110 }, { "epoch": 1.3572476476042539e-13, "grad_norm": 0.6277069449424744, "learning_rate": 7.183536382037094e-05, "loss": 2.3062, "step": 39120 }, { "epoch": 1.3575945922994492e-13, "grad_norm": 0.6074790358543396, "learning_rate": 7.181302831715396e-05, "loss": 2.3087, "step": 39130 }, { "epoch": 1.3579415369946446e-13, "grad_norm": 0.6543247699737549, "learning_rate": 7.179069232954541e-05, "loss": 2.2373, "step": 39140 }, { "epoch": 1.35828848168984e-13, "grad_norm": 0.6366881132125854, "learning_rate": 7.176835586107753e-05, "loss": 2.3183, "step": 39150 }, { "epoch": 1.3586354263850353e-13, "grad_norm": 0.6212483048439026, "learning_rate": 7.174601891528267e-05, "loss": 2.2611, "step": 39160 }, { "epoch": 1.3589823710802307e-13, "grad_norm": 0.6379342675209045, "learning_rate": 7.172368149569319e-05, "loss": 2.2779, "step": 39170 }, { "epoch": 1.359329315775426e-13, "grad_norm": 0.6143132448196411, "learning_rate": 7.17013436058416e-05, "loss": 2.2897, "step": 39180 }, { "epoch": 1.3596762604706214e-13, "grad_norm": 0.6110664010047913, "learning_rate": 7.16790052492604e-05, "loss": 2.251, "step": 39190 }, { "epoch": 1.3600232051658168e-13, "grad_norm": 0.6162256598472595, "learning_rate": 7.165666642948226e-05, "loss": 2.2942, "step": 39200 }, { "epoch": 1.360370149861012e-13, "grad_norm": 0.5968922972679138, "learning_rate": 7.163432715003983e-05, "loss": 2.2897, "step": 39210 }, { "epoch": 1.3607170945562075e-13, "grad_norm": 0.6209198236465454, "learning_rate": 7.16119874144659e-05, "loss": 2.2843, "step": 39220 }, { "epoch": 1.3610640392514028e-13, "grad_norm": 0.5977610349655151, "learning_rate": 7.158964722629328e-05, "loss": 2.2987, "step": 39230 }, { "epoch": 1.3614109839465982e-13, "grad_norm": 0.6055030226707458, "learning_rate": 7.156730658905491e-05, "loss": 2.2705, "step": 39240 }, { "epoch": 1.3617579286417936e-13, "grad_norm": 0.6213457584381104, "learning_rate": 7.154496550628375e-05, "loss": 2.3011, "step": 39250 }, { "epoch": 1.362104873336989e-13, "grad_norm": 0.6668248176574707, "learning_rate": 7.152262398151285e-05, "loss": 2.2982, "step": 39260 }, { "epoch": 1.3624518180321843e-13, "grad_norm": 0.6091310381889343, "learning_rate": 7.150028201827534e-05, "loss": 2.3322, "step": 39270 }, { "epoch": 1.3627987627273797e-13, "grad_norm": 0.6031999588012695, "learning_rate": 7.147793962010439e-05, "loss": 2.3013, "step": 39280 }, { "epoch": 1.363145707422575e-13, "grad_norm": 0.6128405332565308, "learning_rate": 7.145559679053329e-05, "loss": 2.3113, "step": 39290 }, { "epoch": 1.3634926521177704e-13, "grad_norm": 0.6227239370346069, "learning_rate": 7.143325353309535e-05, "loss": 2.3126, "step": 39300 }, { "epoch": 1.3638395968129657e-13, "grad_norm": 0.6082547307014465, "learning_rate": 7.141090985132394e-05, "loss": 2.3048, "step": 39310 }, { "epoch": 1.364186541508161e-13, "grad_norm": 0.6061961650848389, "learning_rate": 7.138856574875256e-05, "loss": 2.2975, "step": 39320 }, { "epoch": 1.3645334862033565e-13, "grad_norm": 0.6106687784194946, "learning_rate": 7.136622122891471e-05, "loss": 2.2808, "step": 39330 }, { "epoch": 1.3648804308985518e-13, "grad_norm": 0.6284006237983704, "learning_rate": 7.1343876295344e-05, "loss": 2.2851, "step": 39340 }, { "epoch": 1.3652273755937472e-13, "grad_norm": 0.6165644526481628, "learning_rate": 7.132153095157408e-05, "loss": 2.2881, "step": 39350 }, { "epoch": 1.3655743202889425e-13, "grad_norm": 0.6166632175445557, "learning_rate": 7.129918520113867e-05, "loss": 2.2692, "step": 39360 }, { "epoch": 1.365921264984138e-13, "grad_norm": 0.6037120819091797, "learning_rate": 7.127683904757157e-05, "loss": 2.2967, "step": 39370 }, { "epoch": 1.3662682096793333e-13, "grad_norm": 0.6435449123382568, "learning_rate": 7.125449249440663e-05, "loss": 2.2938, "step": 39380 }, { "epoch": 1.3666151543745286e-13, "grad_norm": 0.6062971353530884, "learning_rate": 7.123214554517774e-05, "loss": 2.2549, "step": 39390 }, { "epoch": 1.366962099069724e-13, "grad_norm": 0.6130158305168152, "learning_rate": 7.120979820341893e-05, "loss": 2.3072, "step": 39400 }, { "epoch": 1.3673090437649194e-13, "grad_norm": 0.5942577719688416, "learning_rate": 7.118745047266417e-05, "loss": 2.2853, "step": 39410 }, { "epoch": 1.3676559884601147e-13, "grad_norm": 1.0460453033447266, "learning_rate": 7.116510235644765e-05, "loss": 2.3042, "step": 39420 }, { "epoch": 1.36800293315531e-13, "grad_norm": 0.6197386980056763, "learning_rate": 7.114275385830345e-05, "loss": 2.2904, "step": 39430 }, { "epoch": 1.3683498778505054e-13, "grad_norm": 0.5987205505371094, "learning_rate": 7.112040498176582e-05, "loss": 2.2969, "step": 39440 }, { "epoch": 1.3686968225457008e-13, "grad_norm": 0.6367272734642029, "learning_rate": 7.109805573036905e-05, "loss": 2.2721, "step": 39450 }, { "epoch": 1.3690437672408962e-13, "grad_norm": 0.6078767776489258, "learning_rate": 7.10757061076475e-05, "loss": 2.2954, "step": 39460 }, { "epoch": 1.3693907119360915e-13, "grad_norm": 0.6100460290908813, "learning_rate": 7.105335611713551e-05, "loss": 2.2905, "step": 39470 }, { "epoch": 1.369737656631287e-13, "grad_norm": 0.6408981084823608, "learning_rate": 7.103100576236761e-05, "loss": 2.3185, "step": 39480 }, { "epoch": 1.3700846013264822e-13, "grad_norm": 0.6388390064239502, "learning_rate": 7.100865504687827e-05, "loss": 2.2817, "step": 39490 }, { "epoch": 1.3704315460216776e-13, "grad_norm": 0.6139464378356934, "learning_rate": 7.098630397420208e-05, "loss": 2.2937, "step": 39500 }, { "epoch": 1.370778490716873e-13, "grad_norm": 0.6254363059997559, "learning_rate": 7.096395254787365e-05, "loss": 2.2868, "step": 39510 }, { "epoch": 1.3711254354120683e-13, "grad_norm": 0.6274831891059875, "learning_rate": 7.094160077142769e-05, "loss": 2.279, "step": 39520 }, { "epoch": 1.3714723801072637e-13, "grad_norm": 0.6189901232719421, "learning_rate": 7.091924864839895e-05, "loss": 2.3152, "step": 39530 }, { "epoch": 1.371819324802459e-13, "grad_norm": 0.5995326042175293, "learning_rate": 7.089689618232222e-05, "loss": 2.258, "step": 39540 }, { "epoch": 1.3721662694976544e-13, "grad_norm": 0.6258880496025085, "learning_rate": 7.087454337673231e-05, "loss": 2.3026, "step": 39550 }, { "epoch": 1.3725132141928498e-13, "grad_norm": 0.6412018537521362, "learning_rate": 7.085219023516416e-05, "loss": 2.2925, "step": 39560 }, { "epoch": 1.3728601588880451e-13, "grad_norm": 0.6275628209114075, "learning_rate": 7.082983676115273e-05, "loss": 2.3063, "step": 39570 }, { "epoch": 1.3732071035832405e-13, "grad_norm": 0.6122655272483826, "learning_rate": 7.080748295823303e-05, "loss": 2.2935, "step": 39580 }, { "epoch": 1.3735540482784359e-13, "grad_norm": 0.6326655745506287, "learning_rate": 7.078512882994008e-05, "loss": 2.2909, "step": 39590 }, { "epoch": 1.3739009929736312e-13, "grad_norm": 0.6060925722122192, "learning_rate": 7.076277437980904e-05, "loss": 2.3364, "step": 39600 }, { "epoch": 1.3742479376688266e-13, "grad_norm": 0.5933321714401245, "learning_rate": 7.074041961137504e-05, "loss": 2.3039, "step": 39610 }, { "epoch": 1.374594882364022e-13, "grad_norm": 0.6098291277885437, "learning_rate": 7.071806452817336e-05, "loss": 2.2692, "step": 39620 }, { "epoch": 1.3749418270592173e-13, "grad_norm": 0.6057484149932861, "learning_rate": 7.069570913373916e-05, "loss": 2.2984, "step": 39630 }, { "epoch": 1.3752887717544127e-13, "grad_norm": 0.6007506847381592, "learning_rate": 7.067335343160784e-05, "loss": 2.29, "step": 39640 }, { "epoch": 1.375635716449608e-13, "grad_norm": 0.6402588486671448, "learning_rate": 7.065099742531473e-05, "loss": 2.2794, "step": 39650 }, { "epoch": 1.3759826611448034e-13, "grad_norm": 0.6066094636917114, "learning_rate": 7.062864111839525e-05, "loss": 2.3048, "step": 39660 }, { "epoch": 1.3763296058399987e-13, "grad_norm": 0.6339110732078552, "learning_rate": 7.060628451438481e-05, "loss": 2.2908, "step": 39670 }, { "epoch": 1.376676550535194e-13, "grad_norm": 0.6385108232498169, "learning_rate": 7.058392761681898e-05, "loss": 2.3244, "step": 39680 }, { "epoch": 1.3770234952303895e-13, "grad_norm": 0.6399056911468506, "learning_rate": 7.056157042923329e-05, "loss": 2.3073, "step": 39690 }, { "epoch": 1.3773704399255848e-13, "grad_norm": 0.6034976243972778, "learning_rate": 7.053921295516332e-05, "loss": 2.2689, "step": 39700 }, { "epoch": 1.3777173846207802e-13, "grad_norm": 0.5959464907646179, "learning_rate": 7.051685519814471e-05, "loss": 2.2932, "step": 39710 }, { "epoch": 1.3780643293159756e-13, "grad_norm": 0.6106297969818115, "learning_rate": 7.04944971617132e-05, "loss": 2.2755, "step": 39720 }, { "epoch": 1.378411274011171e-13, "grad_norm": 0.6141250729560852, "learning_rate": 7.047213884940444e-05, "loss": 2.2977, "step": 39730 }, { "epoch": 1.3787582187063663e-13, "grad_norm": 0.6119880676269531, "learning_rate": 7.044978026475426e-05, "loss": 2.2704, "step": 39740 }, { "epoch": 1.3791051634015616e-13, "grad_norm": 0.6144484877586365, "learning_rate": 7.042742141129846e-05, "loss": 2.3099, "step": 39750 }, { "epoch": 1.379452108096757e-13, "grad_norm": 0.6381356716156006, "learning_rate": 7.040506229257292e-05, "loss": 2.272, "step": 39760 }, { "epoch": 1.3797990527919524e-13, "grad_norm": 0.6096616387367249, "learning_rate": 7.03827029121135e-05, "loss": 2.2695, "step": 39770 }, { "epoch": 1.3801459974871477e-13, "grad_norm": 0.6142352223396301, "learning_rate": 7.03603432734562e-05, "loss": 2.2578, "step": 39780 }, { "epoch": 1.380492942182343e-13, "grad_norm": 0.5891954898834229, "learning_rate": 7.033798338013694e-05, "loss": 2.2728, "step": 39790 }, { "epoch": 1.3808398868775384e-13, "grad_norm": 0.5990816950798035, "learning_rate": 7.03156232356918e-05, "loss": 2.2744, "step": 39800 }, { "epoch": 1.3811868315727338e-13, "grad_norm": 0.6115233302116394, "learning_rate": 7.02932628436568e-05, "loss": 2.3085, "step": 39810 }, { "epoch": 1.3815337762679292e-13, "grad_norm": 0.6205354332923889, "learning_rate": 7.02709022075681e-05, "loss": 2.2937, "step": 39820 }, { "epoch": 1.3818807209631245e-13, "grad_norm": 0.6210078597068787, "learning_rate": 7.024854133096178e-05, "loss": 2.3082, "step": 39830 }, { "epoch": 1.38222766565832e-13, "grad_norm": 0.5921348929405212, "learning_rate": 7.022618021737404e-05, "loss": 2.3126, "step": 39840 }, { "epoch": 1.3825746103535153e-13, "grad_norm": 0.5947177410125732, "learning_rate": 7.020381887034111e-05, "loss": 2.2847, "step": 39850 }, { "epoch": 1.3829215550487106e-13, "grad_norm": 0.661346435546875, "learning_rate": 7.018145729339927e-05, "loss": 2.2562, "step": 39860 }, { "epoch": 1.383268499743906e-13, "grad_norm": 0.650934100151062, "learning_rate": 7.015909549008477e-05, "loss": 2.3057, "step": 39870 }, { "epoch": 1.3836154444391013e-13, "grad_norm": 0.6010775566101074, "learning_rate": 7.013673346393394e-05, "loss": 2.2841, "step": 39880 }, { "epoch": 1.3839623891342967e-13, "grad_norm": 0.5994139313697815, "learning_rate": 7.011437121848314e-05, "loss": 2.2933, "step": 39890 }, { "epoch": 1.384309333829492e-13, "grad_norm": 0.6205776929855347, "learning_rate": 7.00920087572688e-05, "loss": 2.2802, "step": 39900 }, { "epoch": 1.3846562785246874e-13, "grad_norm": 0.6409497261047363, "learning_rate": 7.00696460838273e-05, "loss": 2.2592, "step": 39910 }, { "epoch": 1.3850032232198828e-13, "grad_norm": 0.6084522604942322, "learning_rate": 7.004728320169516e-05, "loss": 2.2426, "step": 39920 }, { "epoch": 1.3853501679150781e-13, "grad_norm": 0.6070494651794434, "learning_rate": 7.002492011440884e-05, "loss": 2.2859, "step": 39930 }, { "epoch": 1.3856971126102735e-13, "grad_norm": 0.6244845390319824, "learning_rate": 7.000255682550487e-05, "loss": 2.298, "step": 39940 }, { "epoch": 1.3860440573054689e-13, "grad_norm": 0.62662273645401, "learning_rate": 6.998019333851981e-05, "loss": 2.3361, "step": 39950 }, { "epoch": 1.3863910020006642e-13, "grad_norm": 0.603812038898468, "learning_rate": 6.995782965699028e-05, "loss": 2.3365, "step": 39960 }, { "epoch": 1.3867379466958596e-13, "grad_norm": 0.643650233745575, "learning_rate": 6.993546578445289e-05, "loss": 2.2861, "step": 39970 }, { "epoch": 1.387084891391055e-13, "grad_norm": 0.6367295384407043, "learning_rate": 6.991310172444429e-05, "loss": 2.277, "step": 39980 }, { "epoch": 1.3874318360862503e-13, "grad_norm": 0.6126594543457031, "learning_rate": 6.989073748050114e-05, "loss": 2.2803, "step": 39990 }, { "epoch": 1.3877787807814457e-13, "grad_norm": 0.655472993850708, "learning_rate": 6.98683730561602e-05, "loss": 2.282, "step": 40000 }, { "epoch": 1.388125725476641e-13, "grad_norm": 0.6234102249145508, "learning_rate": 6.984600845495819e-05, "loss": 2.2763, "step": 40010 }, { "epoch": 1.3884726701718364e-13, "grad_norm": 0.6353353261947632, "learning_rate": 6.982364368043187e-05, "loss": 2.2735, "step": 40020 }, { "epoch": 1.3888196148670318e-13, "grad_norm": 0.5878025889396667, "learning_rate": 6.980127873611804e-05, "loss": 2.2664, "step": 40030 }, { "epoch": 1.389166559562227e-13, "grad_norm": 0.6493603587150574, "learning_rate": 6.977891362555354e-05, "loss": 2.2818, "step": 40040 }, { "epoch": 1.3895135042574225e-13, "grad_norm": 0.6366462111473083, "learning_rate": 6.975654835227518e-05, "loss": 2.276, "step": 40050 }, { "epoch": 1.3898604489526178e-13, "grad_norm": 0.6282851696014404, "learning_rate": 6.97341829198199e-05, "loss": 2.2903, "step": 40060 }, { "epoch": 1.3902073936478132e-13, "grad_norm": 0.5984408855438232, "learning_rate": 6.971181733172453e-05, "loss": 2.3071, "step": 40070 }, { "epoch": 1.3905543383430086e-13, "grad_norm": 0.6272803544998169, "learning_rate": 6.968945159152604e-05, "loss": 2.2982, "step": 40080 }, { "epoch": 1.390901283038204e-13, "grad_norm": 0.6379275321960449, "learning_rate": 6.966708570276138e-05, "loss": 2.3108, "step": 40090 }, { "epoch": 1.3912482277333993e-13, "grad_norm": 0.6234877705574036, "learning_rate": 6.964471966896751e-05, "loss": 2.2698, "step": 40100 }, { "epoch": 1.3915951724285947e-13, "grad_norm": 0.6401616334915161, "learning_rate": 6.962235349368142e-05, "loss": 2.2491, "step": 40110 }, { "epoch": 1.39194211712379e-13, "grad_norm": 0.59464430809021, "learning_rate": 6.959998718044013e-05, "loss": 2.2591, "step": 40120 }, { "epoch": 1.3922890618189854e-13, "grad_norm": 0.623931348323822, "learning_rate": 6.957762073278068e-05, "loss": 2.2746, "step": 40130 }, { "epoch": 1.3926360065141807e-13, "grad_norm": 0.5993025302886963, "learning_rate": 6.955525415424013e-05, "loss": 2.2847, "step": 40140 }, { "epoch": 1.392982951209376e-13, "grad_norm": 0.6007147431373596, "learning_rate": 6.953288744835558e-05, "loss": 2.2569, "step": 40150 }, { "epoch": 1.3933298959045715e-13, "grad_norm": 0.6259012818336487, "learning_rate": 6.951052061866411e-05, "loss": 2.2745, "step": 40160 }, { "epoch": 1.3936768405997668e-13, "grad_norm": 0.6140263080596924, "learning_rate": 6.948815366870287e-05, "loss": 2.3239, "step": 40170 }, { "epoch": 1.3940237852949622e-13, "grad_norm": 0.6145274043083191, "learning_rate": 6.946578660200898e-05, "loss": 2.3082, "step": 40180 }, { "epoch": 1.3943707299901575e-13, "grad_norm": 0.6079286932945251, "learning_rate": 6.944341942211959e-05, "loss": 2.2698, "step": 40190 }, { "epoch": 1.394717674685353e-13, "grad_norm": 0.61369788646698, "learning_rate": 6.942105213257189e-05, "loss": 2.2667, "step": 40200 }, { "epoch": 1.3950646193805483e-13, "grad_norm": 0.5971001982688904, "learning_rate": 6.939868473690306e-05, "loss": 2.2852, "step": 40210 }, { "epoch": 1.3954115640757436e-13, "grad_norm": 0.6412456631660461, "learning_rate": 6.937631723865033e-05, "loss": 2.2859, "step": 40220 }, { "epoch": 1.395758508770939e-13, "grad_norm": 0.6350859999656677, "learning_rate": 6.935394964135092e-05, "loss": 2.2471, "step": 40230 }, { "epoch": 1.3961054534661343e-13, "grad_norm": 0.6188556551933289, "learning_rate": 6.933158194854207e-05, "loss": 2.3016, "step": 40240 }, { "epoch": 1.3964523981613297e-13, "grad_norm": 0.6344431638717651, "learning_rate": 6.930921416376105e-05, "loss": 2.2689, "step": 40250 }, { "epoch": 1.396799342856525e-13, "grad_norm": 0.6043437719345093, "learning_rate": 6.928684629054512e-05, "loss": 2.2869, "step": 40260 }, { "epoch": 1.3971462875517204e-13, "grad_norm": 0.6430048942565918, "learning_rate": 6.926447833243156e-05, "loss": 2.279, "step": 40270 }, { "epoch": 1.3974932322469158e-13, "grad_norm": 0.6062501668930054, "learning_rate": 6.92421102929577e-05, "loss": 2.2788, "step": 40280 }, { "epoch": 1.3978401769421112e-13, "grad_norm": 0.5806398987770081, "learning_rate": 6.921974217566083e-05, "loss": 2.2431, "step": 40290 }, { "epoch": 1.3981871216373065e-13, "grad_norm": 0.6272449493408203, "learning_rate": 6.919737398407827e-05, "loss": 2.2602, "step": 40300 }, { "epoch": 1.398534066332502e-13, "grad_norm": 0.6370559930801392, "learning_rate": 6.917500572174736e-05, "loss": 2.283, "step": 40310 }, { "epoch": 1.3988810110276972e-13, "grad_norm": 0.6480666995048523, "learning_rate": 6.915263739220545e-05, "loss": 2.2523, "step": 40320 }, { "epoch": 1.3992279557228926e-13, "grad_norm": 0.6035826206207275, "learning_rate": 6.913026899898992e-05, "loss": 2.258, "step": 40330 }, { "epoch": 1.399574900418088e-13, "grad_norm": 0.6129710078239441, "learning_rate": 6.910790054563812e-05, "loss": 2.2758, "step": 40340 }, { "epoch": 1.3999218451132833e-13, "grad_norm": 0.6302993893623352, "learning_rate": 6.908553203568743e-05, "loss": 2.2998, "step": 40350 }, { "epoch": 1.4002687898084787e-13, "grad_norm": 0.6142181754112244, "learning_rate": 6.906316347267524e-05, "loss": 2.2644, "step": 40360 }, { "epoch": 1.400615734503674e-13, "grad_norm": 0.6429085731506348, "learning_rate": 6.904079486013892e-05, "loss": 2.3021, "step": 40370 }, { "epoch": 1.4009626791988694e-13, "grad_norm": 0.612223744392395, "learning_rate": 6.901842620161592e-05, "loss": 2.3223, "step": 40380 }, { "epoch": 1.4013096238940648e-13, "grad_norm": 0.6290953159332275, "learning_rate": 6.89960575006436e-05, "loss": 2.2974, "step": 40390 }, { "epoch": 1.4016565685892601e-13, "grad_norm": 0.6707074046134949, "learning_rate": 6.897368876075944e-05, "loss": 2.2988, "step": 40400 }, { "epoch": 1.4020035132844555e-13, "grad_norm": 0.6398902535438538, "learning_rate": 6.895131998550081e-05, "loss": 2.3203, "step": 40410 }, { "epoch": 1.4023504579796509e-13, "grad_norm": 0.6171675324440002, "learning_rate": 6.892895117840514e-05, "loss": 2.2809, "step": 40420 }, { "epoch": 1.4026974026748462e-13, "grad_norm": 0.6207858324050903, "learning_rate": 6.890658234300991e-05, "loss": 2.2912, "step": 40430 }, { "epoch": 1.4030443473700416e-13, "grad_norm": 0.6148186922073364, "learning_rate": 6.88842134828525e-05, "loss": 2.3107, "step": 40440 }, { "epoch": 1.403391292065237e-13, "grad_norm": 0.6478198766708374, "learning_rate": 6.886184460147042e-05, "loss": 2.246, "step": 40450 }, { "epoch": 1.4037382367604323e-13, "grad_norm": 0.608615517616272, "learning_rate": 6.883947570240104e-05, "loss": 2.319, "step": 40460 }, { "epoch": 1.4040851814556277e-13, "grad_norm": 0.6212443113327026, "learning_rate": 6.881710678918188e-05, "loss": 2.2651, "step": 40470 }, { "epoch": 1.404432126150823e-13, "grad_norm": 0.6220203638076782, "learning_rate": 6.879473786535034e-05, "loss": 2.2696, "step": 40480 }, { "epoch": 1.4047790708460184e-13, "grad_norm": 0.6330322623252869, "learning_rate": 6.877236893444389e-05, "loss": 2.3146, "step": 40490 }, { "epoch": 1.4051260155412137e-13, "grad_norm": 0.6430771350860596, "learning_rate": 6.875e-05, "loss": 2.2845, "step": 40500 }, { "epoch": 1.405472960236409e-13, "grad_norm": 0.6109170317649841, "learning_rate": 6.87276310655561e-05, "loss": 2.3014, "step": 40510 }, { "epoch": 1.4058199049316045e-13, "grad_norm": 0.627488911151886, "learning_rate": 6.870526213464967e-05, "loss": 2.2816, "step": 40520 }, { "epoch": 1.4061668496267998e-13, "grad_norm": 0.6411418914794922, "learning_rate": 6.868289321081814e-05, "loss": 2.2863, "step": 40530 }, { "epoch": 1.4065137943219952e-13, "grad_norm": 0.6108585596084595, "learning_rate": 6.866052429759896e-05, "loss": 2.2645, "step": 40540 }, { "epoch": 1.4068607390171906e-13, "grad_norm": 0.662269115447998, "learning_rate": 6.86381553985296e-05, "loss": 2.3021, "step": 40550 }, { "epoch": 1.407207683712386e-13, "grad_norm": 0.6790271401405334, "learning_rate": 6.861578651714749e-05, "loss": 2.3236, "step": 40560 }, { "epoch": 1.4075546284075813e-13, "grad_norm": 0.6279926896095276, "learning_rate": 6.859341765699011e-05, "loss": 2.3166, "step": 40570 }, { "epoch": 1.4079015731027766e-13, "grad_norm": 0.6341143846511841, "learning_rate": 6.857104882159486e-05, "loss": 2.2796, "step": 40580 }, { "epoch": 1.408248517797972e-13, "grad_norm": 0.6014166474342346, "learning_rate": 6.854868001449923e-05, "loss": 2.2618, "step": 40590 }, { "epoch": 1.4085954624931674e-13, "grad_norm": 0.6102643013000488, "learning_rate": 6.85263112392406e-05, "loss": 2.2718, "step": 40600 }, { "epoch": 1.4089424071883627e-13, "grad_norm": 0.6077476739883423, "learning_rate": 6.85039424993564e-05, "loss": 2.2533, "step": 40610 }, { "epoch": 1.409289351883558e-13, "grad_norm": 0.6178117990493774, "learning_rate": 6.84815737983841e-05, "loss": 2.2671, "step": 40620 }, { "epoch": 1.4096362965787534e-13, "grad_norm": 0.6193773746490479, "learning_rate": 6.845920513986107e-05, "loss": 2.2751, "step": 40630 }, { "epoch": 1.4099832412739488e-13, "grad_norm": 0.6763565540313721, "learning_rate": 6.84368365273248e-05, "loss": 2.2828, "step": 40640 }, { "epoch": 1.4103301859691442e-13, "grad_norm": 0.6179067492485046, "learning_rate": 6.841446796431258e-05, "loss": 2.2842, "step": 40650 }, { "epoch": 1.4106771306643395e-13, "grad_norm": 0.6729224324226379, "learning_rate": 6.839209945436188e-05, "loss": 2.3211, "step": 40660 }, { "epoch": 1.411024075359535e-13, "grad_norm": 0.6242539286613464, "learning_rate": 6.836973100101009e-05, "loss": 2.2885, "step": 40670 }, { "epoch": 1.4113710200547303e-13, "grad_norm": 0.6082979440689087, "learning_rate": 6.834736260779454e-05, "loss": 2.3193, "step": 40680 }, { "epoch": 1.4117179647499256e-13, "grad_norm": 0.6288167834281921, "learning_rate": 6.832499427825266e-05, "loss": 2.2954, "step": 40690 }, { "epoch": 1.412064909445121e-13, "grad_norm": 0.6243518590927124, "learning_rate": 6.830262601592176e-05, "loss": 2.2824, "step": 40700 }, { "epoch": 1.4124118541403163e-13, "grad_norm": 0.6118938326835632, "learning_rate": 6.82802578243392e-05, "loss": 2.2459, "step": 40710 }, { "epoch": 1.4127587988355117e-13, "grad_norm": 0.6230709552764893, "learning_rate": 6.825788970704231e-05, "loss": 2.2812, "step": 40720 }, { "epoch": 1.413105743530707e-13, "grad_norm": 0.6133674383163452, "learning_rate": 6.823552166756842e-05, "loss": 2.2875, "step": 40730 }, { "epoch": 1.4134526882259024e-13, "grad_norm": 0.6288545727729797, "learning_rate": 6.821315370945489e-05, "loss": 2.2904, "step": 40740 }, { "epoch": 1.4137996329210978e-13, "grad_norm": 0.6084797978401184, "learning_rate": 6.819078583623896e-05, "loss": 2.2745, "step": 40750 }, { "epoch": 1.4141465776162931e-13, "grad_norm": 0.6034690737724304, "learning_rate": 6.816841805145795e-05, "loss": 2.2706, "step": 40760 }, { "epoch": 1.4144935223114885e-13, "grad_norm": 0.628845751285553, "learning_rate": 6.81460503586491e-05, "loss": 2.2952, "step": 40770 }, { "epoch": 1.4148404670066839e-13, "grad_norm": 0.6777082085609436, "learning_rate": 6.812368276134968e-05, "loss": 2.2988, "step": 40780 }, { "epoch": 1.4151874117018792e-13, "grad_norm": 0.5957391858100891, "learning_rate": 6.810131526309696e-05, "loss": 2.2732, "step": 40790 }, { "epoch": 1.4155343563970746e-13, "grad_norm": 0.5920467972755432, "learning_rate": 6.807894786742814e-05, "loss": 2.2553, "step": 40800 }, { "epoch": 1.41588130109227e-13, "grad_norm": 0.622519850730896, "learning_rate": 6.805658057788045e-05, "loss": 2.2588, "step": 40810 }, { "epoch": 1.4162282457874653e-13, "grad_norm": 0.6054553389549255, "learning_rate": 6.803421339799104e-05, "loss": 2.2731, "step": 40820 }, { "epoch": 1.4165751904826607e-13, "grad_norm": 0.6159979104995728, "learning_rate": 6.801184633129713e-05, "loss": 2.3129, "step": 40830 }, { "epoch": 1.416922135177856e-13, "grad_norm": 0.6365079879760742, "learning_rate": 6.79894793813359e-05, "loss": 2.2952, "step": 40840 }, { "epoch": 1.4172690798730514e-13, "grad_norm": 0.620419979095459, "learning_rate": 6.796711255164443e-05, "loss": 2.2692, "step": 40850 }, { "epoch": 1.4176160245682468e-13, "grad_norm": 0.6165825724601746, "learning_rate": 6.794474584575988e-05, "loss": 2.2459, "step": 40860 }, { "epoch": 1.417962969263442e-13, "grad_norm": 0.6290144324302673, "learning_rate": 6.792237926721933e-05, "loss": 2.282, "step": 40870 }, { "epoch": 1.4183099139586375e-13, "grad_norm": 0.6274662613868713, "learning_rate": 6.790001281955988e-05, "loss": 2.2555, "step": 40880 }, { "epoch": 1.4186568586538328e-13, "grad_norm": 0.6290456652641296, "learning_rate": 6.787764650631861e-05, "loss": 2.2951, "step": 40890 }, { "epoch": 1.4190038033490282e-13, "grad_norm": 0.6233773827552795, "learning_rate": 6.78552803310325e-05, "loss": 2.2882, "step": 40900 }, { "epoch": 1.4193507480442236e-13, "grad_norm": 0.615822434425354, "learning_rate": 6.783291429723865e-05, "loss": 2.2864, "step": 40910 }, { "epoch": 1.419697692739419e-13, "grad_norm": 0.6160604953765869, "learning_rate": 6.781054840847396e-05, "loss": 2.2773, "step": 40920 }, { "epoch": 1.4200446374346143e-13, "grad_norm": 0.6511644124984741, "learning_rate": 6.778818266827547e-05, "loss": 2.2908, "step": 40930 }, { "epoch": 1.4203915821298096e-13, "grad_norm": 0.6600939631462097, "learning_rate": 6.776581708018013e-05, "loss": 2.293, "step": 40940 }, { "epoch": 1.420738526825005e-13, "grad_norm": 0.615871250629425, "learning_rate": 6.77434516477248e-05, "loss": 2.2914, "step": 40950 }, { "epoch": 1.4210854715202004e-13, "grad_norm": 0.629979133605957, "learning_rate": 6.772108637444649e-05, "loss": 2.2885, "step": 40960 }, { "epoch": 1.4214324162153957e-13, "grad_norm": 0.6123164296150208, "learning_rate": 6.769872126388197e-05, "loss": 2.2645, "step": 40970 }, { "epoch": 1.421779360910591e-13, "grad_norm": 0.6639456748962402, "learning_rate": 6.767635631956812e-05, "loss": 2.3021, "step": 40980 }, { "epoch": 1.4221263056057865e-13, "grad_norm": 0.6486911177635193, "learning_rate": 6.765399154504183e-05, "loss": 2.289, "step": 40990 }, { "epoch": 1.4224732503009818e-13, "grad_norm": 0.6346791982650757, "learning_rate": 6.76316269438398e-05, "loss": 2.3229, "step": 41000 }, { "epoch": 1.4228201949961772e-13, "grad_norm": 0.6152845621109009, "learning_rate": 6.760926251949887e-05, "loss": 2.2909, "step": 41010 }, { "epoch": 1.4231671396913725e-13, "grad_norm": 0.6351901292800903, "learning_rate": 6.758689827555573e-05, "loss": 2.2888, "step": 41020 }, { "epoch": 1.423514084386568e-13, "grad_norm": 0.6271422505378723, "learning_rate": 6.756453421554713e-05, "loss": 2.3002, "step": 41030 }, { "epoch": 1.4238610290817633e-13, "grad_norm": 0.6204388737678528, "learning_rate": 6.754217034300973e-05, "loss": 2.2889, "step": 41040 }, { "epoch": 1.4242079737769586e-13, "grad_norm": 0.626214861869812, "learning_rate": 6.75198066614802e-05, "loss": 2.2879, "step": 41050 }, { "epoch": 1.424554918472154e-13, "grad_norm": 0.6089880466461182, "learning_rate": 6.749744317449516e-05, "loss": 2.291, "step": 41060 }, { "epoch": 1.4249018631673493e-13, "grad_norm": 0.6109974384307861, "learning_rate": 6.747507988559118e-05, "loss": 2.2821, "step": 41070 }, { "epoch": 1.4252488078625447e-13, "grad_norm": 0.6213722825050354, "learning_rate": 6.745271679830485e-05, "loss": 2.3018, "step": 41080 }, { "epoch": 1.42559575255774e-13, "grad_norm": 0.6266905665397644, "learning_rate": 6.74303539161727e-05, "loss": 2.2892, "step": 41090 }, { "epoch": 1.4259426972529354e-13, "grad_norm": 0.6276528239250183, "learning_rate": 6.740799124273121e-05, "loss": 2.2783, "step": 41100 }, { "epoch": 1.4262896419481308e-13, "grad_norm": 0.6549265384674072, "learning_rate": 6.738562878151686e-05, "loss": 2.2978, "step": 41110 }, { "epoch": 1.4266365866433262e-13, "grad_norm": 0.6365146636962891, "learning_rate": 6.736326653606606e-05, "loss": 2.2649, "step": 41120 }, { "epoch": 1.4269835313385215e-13, "grad_norm": 0.5989809632301331, "learning_rate": 6.734090450991525e-05, "loss": 2.2811, "step": 41130 }, { "epoch": 1.427330476033717e-13, "grad_norm": 0.6508694291114807, "learning_rate": 6.731854270660074e-05, "loss": 2.3001, "step": 41140 }, { "epoch": 1.4276774207289122e-13, "grad_norm": 0.6260665655136108, "learning_rate": 6.729618112965888e-05, "loss": 2.2782, "step": 41150 }, { "epoch": 1.4280243654241076e-13, "grad_norm": 0.6204610466957092, "learning_rate": 6.727381978262597e-05, "loss": 2.3049, "step": 41160 }, { "epoch": 1.428371310119303e-13, "grad_norm": 0.6224971413612366, "learning_rate": 6.725145866903824e-05, "loss": 2.3045, "step": 41170 }, { "epoch": 1.4287182548144983e-13, "grad_norm": 0.650421142578125, "learning_rate": 6.722909779243192e-05, "loss": 2.3166, "step": 41180 }, { "epoch": 1.4290651995096937e-13, "grad_norm": 0.6154957413673401, "learning_rate": 6.72067371563432e-05, "loss": 2.3034, "step": 41190 }, { "epoch": 1.429412144204889e-13, "grad_norm": 0.6213278770446777, "learning_rate": 6.71843767643082e-05, "loss": 2.2745, "step": 41200 }, { "epoch": 1.4297590889000844e-13, "grad_norm": 0.6165634393692017, "learning_rate": 6.716201661986308e-05, "loss": 2.267, "step": 41210 }, { "epoch": 1.4301060335952798e-13, "grad_norm": 0.6227937936782837, "learning_rate": 6.713965672654381e-05, "loss": 2.2819, "step": 41220 }, { "epoch": 1.4304529782904751e-13, "grad_norm": 0.6130265593528748, "learning_rate": 6.711729708788652e-05, "loss": 2.2516, "step": 41230 }, { "epoch": 1.4307999229856705e-13, "grad_norm": 0.5949985384941101, "learning_rate": 6.709493770742709e-05, "loss": 2.2899, "step": 41240 }, { "epoch": 1.4311468676808659e-13, "grad_norm": 0.6485351324081421, "learning_rate": 6.707257858870154e-05, "loss": 2.2672, "step": 41250 }, { "epoch": 1.4314938123760612e-13, "grad_norm": 0.6416646838188171, "learning_rate": 6.705021973524575e-05, "loss": 2.2602, "step": 41260 }, { "epoch": 1.4318407570712566e-13, "grad_norm": 0.608563244342804, "learning_rate": 6.702786115059556e-05, "loss": 2.3017, "step": 41270 }, { "epoch": 1.432187701766452e-13, "grad_norm": 0.597795307636261, "learning_rate": 6.700550283828684e-05, "loss": 2.2844, "step": 41280 }, { "epoch": 1.4325346464616473e-13, "grad_norm": 0.6425365805625916, "learning_rate": 6.698314480185529e-05, "loss": 2.277, "step": 41290 }, { "epoch": 1.4328815911568427e-13, "grad_norm": 0.6221733093261719, "learning_rate": 6.696078704483669e-05, "loss": 2.2912, "step": 41300 }, { "epoch": 1.433228535852038e-13, "grad_norm": 0.6262274384498596, "learning_rate": 6.693842957076674e-05, "loss": 2.3003, "step": 41310 }, { "epoch": 1.4335754805472334e-13, "grad_norm": 0.6199135780334473, "learning_rate": 6.691607238318103e-05, "loss": 2.2747, "step": 41320 }, { "epoch": 1.4339224252424287e-13, "grad_norm": 0.65901118516922, "learning_rate": 6.68937154856152e-05, "loss": 2.2536, "step": 41330 }, { "epoch": 1.434269369937624e-13, "grad_norm": 0.611878514289856, "learning_rate": 6.687135888160476e-05, "loss": 2.2593, "step": 41340 }, { "epoch": 1.4346163146328195e-13, "grad_norm": 0.6068300008773804, "learning_rate": 6.68490025746853e-05, "loss": 2.2694, "step": 41350 }, { "epoch": 1.4349632593280148e-13, "grad_norm": 0.5994897484779358, "learning_rate": 6.682664656839217e-05, "loss": 2.2578, "step": 41360 }, { "epoch": 1.4353102040232102e-13, "grad_norm": 0.6071301698684692, "learning_rate": 6.680429086626084e-05, "loss": 2.2936, "step": 41370 }, { "epoch": 1.4356571487184056e-13, "grad_norm": 0.6376314759254456, "learning_rate": 6.678193547182668e-05, "loss": 2.2864, "step": 41380 }, { "epoch": 1.436004093413601e-13, "grad_norm": 0.5900683403015137, "learning_rate": 6.675958038862497e-05, "loss": 2.2786, "step": 41390 }, { "epoch": 1.4363510381087963e-13, "grad_norm": 0.6133044958114624, "learning_rate": 6.673722562019098e-05, "loss": 2.2947, "step": 41400 }, { "epoch": 1.4366979828039916e-13, "grad_norm": 0.6002002954483032, "learning_rate": 6.671487117005994e-05, "loss": 2.2674, "step": 41410 }, { "epoch": 1.437044927499187e-13, "grad_norm": 0.680560827255249, "learning_rate": 6.6692517041767e-05, "loss": 2.3042, "step": 41420 }, { "epoch": 1.4373918721943824e-13, "grad_norm": 0.6400943994522095, "learning_rate": 6.667016323884728e-05, "loss": 2.2606, "step": 41430 }, { "epoch": 1.4377388168895777e-13, "grad_norm": 0.6342461109161377, "learning_rate": 6.664780976483583e-05, "loss": 2.2966, "step": 41440 }, { "epoch": 1.438085761584773e-13, "grad_norm": 0.6213639974594116, "learning_rate": 6.662545662326771e-05, "loss": 2.2903, "step": 41450 }, { "epoch": 1.4384327062799684e-13, "grad_norm": 0.6282036304473877, "learning_rate": 6.660310381767782e-05, "loss": 2.311, "step": 41460 }, { "epoch": 1.4387796509751638e-13, "grad_norm": 0.6131877899169922, "learning_rate": 6.658075135160104e-05, "loss": 2.2708, "step": 41470 }, { "epoch": 1.4391265956703592e-13, "grad_norm": 0.6494730114936829, "learning_rate": 6.655839922857232e-05, "loss": 2.2866, "step": 41480 }, { "epoch": 1.4394735403655545e-13, "grad_norm": 0.6496139764785767, "learning_rate": 6.653604745212635e-05, "loss": 2.2817, "step": 41490 }, { "epoch": 1.43982048506075e-13, "grad_norm": 0.6472102403640747, "learning_rate": 6.651369602579794e-05, "loss": 2.2827, "step": 41500 }, { "epoch": 1.4401674297559452e-13, "grad_norm": 0.6271105408668518, "learning_rate": 6.649134495312174e-05, "loss": 2.2791, "step": 41510 }, { "epoch": 1.4405143744511406e-13, "grad_norm": 0.6233764886856079, "learning_rate": 6.646899423763241e-05, "loss": 2.2697, "step": 41520 }, { "epoch": 1.440861319146336e-13, "grad_norm": 0.6060241460800171, "learning_rate": 6.64466438828645e-05, "loss": 2.2632, "step": 41530 }, { "epoch": 1.4412082638415313e-13, "grad_norm": 0.6406155228614807, "learning_rate": 6.642429389235252e-05, "loss": 2.3252, "step": 41540 }, { "epoch": 1.4415552085367267e-13, "grad_norm": 0.5952444672584534, "learning_rate": 6.640194426963098e-05, "loss": 2.2923, "step": 41550 }, { "epoch": 1.441902153231922e-13, "grad_norm": 0.6399208307266235, "learning_rate": 6.637959501823421e-05, "loss": 2.2783, "step": 41560 }, { "epoch": 1.4422490979271174e-13, "grad_norm": 0.6175060272216797, "learning_rate": 6.635724614169657e-05, "loss": 2.2769, "step": 41570 }, { "epoch": 1.4425960426223128e-13, "grad_norm": 0.6302570104598999, "learning_rate": 6.633489764355239e-05, "loss": 2.2835, "step": 41580 }, { "epoch": 1.4429429873175081e-13, "grad_norm": 0.6180248856544495, "learning_rate": 6.631254952733581e-05, "loss": 2.2771, "step": 41590 }, { "epoch": 1.4432899320127035e-13, "grad_norm": 0.6384199857711792, "learning_rate": 6.62902017965811e-05, "loss": 2.276, "step": 41600 }, { "epoch": 1.4436368767078989e-13, "grad_norm": 0.619249701499939, "learning_rate": 6.626785445482227e-05, "loss": 2.3031, "step": 41610 }, { "epoch": 1.4439838214030942e-13, "grad_norm": 0.604288637638092, "learning_rate": 6.624550750559338e-05, "loss": 2.3269, "step": 41620 }, { "epoch": 1.4443307660982896e-13, "grad_norm": 0.6108273267745972, "learning_rate": 6.622316095242845e-05, "loss": 2.2866, "step": 41630 }, { "epoch": 1.444677710793485e-13, "grad_norm": 0.5914251804351807, "learning_rate": 6.620081479886134e-05, "loss": 2.2674, "step": 41640 }, { "epoch": 1.4450246554886803e-13, "grad_norm": 0.6228236556053162, "learning_rate": 6.617846904842593e-05, "loss": 2.2889, "step": 41650 }, { "epoch": 1.4453716001838757e-13, "grad_norm": 0.6239244937896729, "learning_rate": 6.615612370465601e-05, "loss": 2.2736, "step": 41660 }, { "epoch": 1.445718544879071e-13, "grad_norm": 0.6313282251358032, "learning_rate": 6.613377877108531e-05, "loss": 2.302, "step": 41670 }, { "epoch": 1.4460654895742664e-13, "grad_norm": 0.630631685256958, "learning_rate": 6.611143425124745e-05, "loss": 2.2981, "step": 41680 }, { "epoch": 1.4464124342694618e-13, "grad_norm": 0.6321432590484619, "learning_rate": 6.608909014867606e-05, "loss": 2.2832, "step": 41690 }, { "epoch": 1.446759378964657e-13, "grad_norm": 0.615907609462738, "learning_rate": 6.606674646690467e-05, "loss": 2.2728, "step": 41700 }, { "epoch": 1.4471063236598525e-13, "grad_norm": 0.6390239000320435, "learning_rate": 6.604440320946673e-05, "loss": 2.3004, "step": 41710 }, { "epoch": 1.4474532683550478e-13, "grad_norm": 0.6269553303718567, "learning_rate": 6.602206037989561e-05, "loss": 2.3245, "step": 41720 }, { "epoch": 1.4478002130502432e-13, "grad_norm": 0.6491886377334595, "learning_rate": 6.599971798172465e-05, "loss": 2.2999, "step": 41730 }, { "epoch": 1.4481471577454386e-13, "grad_norm": 0.6414786577224731, "learning_rate": 6.597737601848716e-05, "loss": 2.3013, "step": 41740 }, { "epoch": 1.448494102440634e-13, "grad_norm": 0.6429856419563293, "learning_rate": 6.595503449371627e-05, "loss": 2.2827, "step": 41750 }, { "epoch": 1.4488410471358293e-13, "grad_norm": 0.6523755788803101, "learning_rate": 6.59326934109451e-05, "loss": 2.2871, "step": 41760 }, { "epoch": 1.4491879918310246e-13, "grad_norm": 0.6268453001976013, "learning_rate": 6.591035277370673e-05, "loss": 2.2857, "step": 41770 }, { "epoch": 1.44953493652622e-13, "grad_norm": 0.6240869760513306, "learning_rate": 6.588801258553411e-05, "loss": 2.2899, "step": 41780 }, { "epoch": 1.4498818812214154e-13, "grad_norm": 0.6625106334686279, "learning_rate": 6.586567284996018e-05, "loss": 2.2872, "step": 41790 }, { "epoch": 1.4502288259166107e-13, "grad_norm": 0.6265219449996948, "learning_rate": 6.584333357051776e-05, "loss": 2.2885, "step": 41800 }, { "epoch": 1.450575770611806e-13, "grad_norm": 0.6335184574127197, "learning_rate": 6.58209947507396e-05, "loss": 2.2836, "step": 41810 }, { "epoch": 1.4509227153070015e-13, "grad_norm": 0.6582463383674622, "learning_rate": 6.579865639415841e-05, "loss": 2.293, "step": 41820 }, { "epoch": 1.4512696600021968e-13, "grad_norm": 0.6187800168991089, "learning_rate": 6.577631850430682e-05, "loss": 2.2901, "step": 41830 }, { "epoch": 1.4516166046973922e-13, "grad_norm": 0.6246480345726013, "learning_rate": 6.575398108471734e-05, "loss": 2.281, "step": 41840 }, { "epoch": 1.4519635493925875e-13, "grad_norm": 0.6084679961204529, "learning_rate": 6.57316441389225e-05, "loss": 2.303, "step": 41850 }, { "epoch": 1.452310494087783e-13, "grad_norm": 0.6179131865501404, "learning_rate": 6.57093076704546e-05, "loss": 2.3173, "step": 41860 }, { "epoch": 1.4526574387829783e-13, "grad_norm": 0.6460108757019043, "learning_rate": 6.568697168284606e-05, "loss": 2.2884, "step": 41870 }, { "epoch": 1.4530043834781736e-13, "grad_norm": 0.6343269348144531, "learning_rate": 6.566463617962908e-05, "loss": 2.2992, "step": 41880 }, { "epoch": 1.453351328173369e-13, "grad_norm": 0.6176756024360657, "learning_rate": 6.564230116433581e-05, "loss": 2.2725, "step": 41890 }, { "epoch": 1.4536982728685643e-13, "grad_norm": 0.6189109086990356, "learning_rate": 6.561996664049837e-05, "loss": 2.2557, "step": 41900 }, { "epoch": 1.4540452175637597e-13, "grad_norm": 0.6273884177207947, "learning_rate": 6.559763261164875e-05, "loss": 2.3025, "step": 41910 }, { "epoch": 1.454392162258955e-13, "grad_norm": 0.6282914280891418, "learning_rate": 6.557529908131892e-05, "loss": 2.258, "step": 41920 }, { "epoch": 1.4547391069541504e-13, "grad_norm": 0.6213845610618591, "learning_rate": 6.55529660530407e-05, "loss": 2.2592, "step": 41930 }, { "epoch": 1.4550860516493458e-13, "grad_norm": 0.7264476418495178, "learning_rate": 6.553063353034586e-05, "loss": 2.318, "step": 41940 }, { "epoch": 1.4554329963445412e-13, "grad_norm": 0.6367590427398682, "learning_rate": 6.550830151676617e-05, "loss": 2.2694, "step": 41950 }, { "epoch": 1.4557799410397365e-13, "grad_norm": 0.6383909583091736, "learning_rate": 6.548597001583313e-05, "loss": 2.2883, "step": 41960 }, { "epoch": 1.456126885734932e-13, "grad_norm": 0.6283553838729858, "learning_rate": 6.546363903107837e-05, "loss": 2.2808, "step": 41970 }, { "epoch": 1.4564738304301272e-13, "grad_norm": 0.6441916823387146, "learning_rate": 6.544130856603328e-05, "loss": 2.3191, "step": 41980 }, { "epoch": 1.4568207751253226e-13, "grad_norm": 0.6158684492111206, "learning_rate": 6.54189786242293e-05, "loss": 2.2984, "step": 41990 }, { "epoch": 1.457167719820518e-13, "grad_norm": 0.651002824306488, "learning_rate": 6.539664920919764e-05, "loss": 2.2312, "step": 42000 }, { "epoch": 1.4575146645157133e-13, "grad_norm": 0.6367666125297546, "learning_rate": 6.537432032446955e-05, "loss": 2.3147, "step": 42010 }, { "epoch": 1.4578616092109087e-13, "grad_norm": 0.6127921342849731, "learning_rate": 6.535199197357616e-05, "loss": 2.2666, "step": 42020 }, { "epoch": 1.458208553906104e-13, "grad_norm": 0.6111031174659729, "learning_rate": 6.532966416004846e-05, "loss": 2.2478, "step": 42030 }, { "epoch": 1.4585554986012994e-13, "grad_norm": 0.6275156140327454, "learning_rate": 6.530733688741745e-05, "loss": 2.2726, "step": 42040 }, { "epoch": 1.4589024432964948e-13, "grad_norm": 0.6135126352310181, "learning_rate": 6.528501015921396e-05, "loss": 2.2587, "step": 42050 }, { "epoch": 1.4592493879916901e-13, "grad_norm": 0.652019202709198, "learning_rate": 6.52626839789688e-05, "loss": 2.2756, "step": 42060 }, { "epoch": 1.4595963326868855e-13, "grad_norm": 0.6263571977615356, "learning_rate": 6.524035835021267e-05, "loss": 2.2756, "step": 42070 }, { "epoch": 1.4599432773820809e-13, "grad_norm": 0.6355133056640625, "learning_rate": 6.521803327647614e-05, "loss": 2.3076, "step": 42080 }, { "epoch": 1.4602902220772762e-13, "grad_norm": 0.6303281188011169, "learning_rate": 6.519570876128975e-05, "loss": 2.3118, "step": 42090 }, { "epoch": 1.4606371667724716e-13, "grad_norm": 0.6150932312011719, "learning_rate": 6.51733848081839e-05, "loss": 2.263, "step": 42100 }, { "epoch": 1.460984111467667e-13, "grad_norm": 0.6200718283653259, "learning_rate": 6.515106142068896e-05, "loss": 2.2589, "step": 42110 }, { "epoch": 1.4613310561628623e-13, "grad_norm": 0.6168888807296753, "learning_rate": 6.512873860233518e-05, "loss": 2.2833, "step": 42120 }, { "epoch": 1.4616780008580577e-13, "grad_norm": 0.6445562243461609, "learning_rate": 6.510641635665272e-05, "loss": 2.3098, "step": 42130 }, { "epoch": 1.462024945553253e-13, "grad_norm": 0.6450097560882568, "learning_rate": 6.508409468717167e-05, "loss": 2.31, "step": 42140 }, { "epoch": 1.4623718902484484e-13, "grad_norm": 0.620744526386261, "learning_rate": 6.506177359742195e-05, "loss": 2.2696, "step": 42150 }, { "epoch": 1.4627188349436437e-13, "grad_norm": 0.644432783126831, "learning_rate": 6.50394530909335e-05, "loss": 2.313, "step": 42160 }, { "epoch": 1.463065779638839e-13, "grad_norm": 0.6358450651168823, "learning_rate": 6.501713317123613e-05, "loss": 2.3334, "step": 42170 }, { "epoch": 1.4634127243340345e-13, "grad_norm": 0.6240207552909851, "learning_rate": 6.499481384185949e-05, "loss": 2.2733, "step": 42180 }, { "epoch": 1.4637596690292298e-13, "grad_norm": 0.6781615614891052, "learning_rate": 6.497249510633324e-05, "loss": 2.3046, "step": 42190 }, { "epoch": 1.4641066137244252e-13, "grad_norm": 0.6508637070655823, "learning_rate": 6.495017696818686e-05, "loss": 2.2836, "step": 42200 }, { "epoch": 1.4644535584196205e-13, "grad_norm": 0.6415733098983765, "learning_rate": 6.492785943094976e-05, "loss": 2.2937, "step": 42210 }, { "epoch": 1.464800503114816e-13, "grad_norm": 0.6434285640716553, "learning_rate": 6.49055424981513e-05, "loss": 2.2577, "step": 42220 }, { "epoch": 1.4651474478100113e-13, "grad_norm": 0.6359096169471741, "learning_rate": 6.488322617332071e-05, "loss": 2.2883, "step": 42230 }, { "epoch": 1.4654943925052066e-13, "grad_norm": 0.6213241219520569, "learning_rate": 6.486091045998713e-05, "loss": 2.3142, "step": 42240 }, { "epoch": 1.465841337200402e-13, "grad_norm": 0.6358934640884399, "learning_rate": 6.483859536167954e-05, "loss": 2.2679, "step": 42250 }, { "epoch": 1.4661882818955974e-13, "grad_norm": 0.6474254131317139, "learning_rate": 6.481628088192695e-05, "loss": 2.2572, "step": 42260 }, { "epoch": 1.4665352265907927e-13, "grad_norm": 0.6572834849357605, "learning_rate": 6.47939670242582e-05, "loss": 2.2965, "step": 42270 }, { "epoch": 1.466882171285988e-13, "grad_norm": 0.6820470690727234, "learning_rate": 6.477165379220198e-05, "loss": 2.2765, "step": 42280 }, { "epoch": 1.4672291159811834e-13, "grad_norm": 0.6223180294036865, "learning_rate": 6.474934118928697e-05, "loss": 2.2861, "step": 42290 }, { "epoch": 1.4675760606763788e-13, "grad_norm": 0.6144195795059204, "learning_rate": 6.472702921904172e-05, "loss": 2.2836, "step": 42300 }, { "epoch": 1.4679230053715742e-13, "grad_norm": 0.6373863816261292, "learning_rate": 6.470471788499465e-05, "loss": 2.3131, "step": 42310 }, { "epoch": 1.4682699500667695e-13, "grad_norm": 0.6107750535011292, "learning_rate": 6.468240719067417e-05, "loss": 2.2714, "step": 42320 }, { "epoch": 1.468616894761965e-13, "grad_norm": 0.6166872978210449, "learning_rate": 6.466009713960843e-05, "loss": 2.2747, "step": 42330 }, { "epoch": 1.4689638394571602e-13, "grad_norm": 0.6326593160629272, "learning_rate": 6.463778773532567e-05, "loss": 2.2835, "step": 42340 }, { "epoch": 1.4693107841523556e-13, "grad_norm": 0.6297916769981384, "learning_rate": 6.461547898135385e-05, "loss": 2.2716, "step": 42350 }, { "epoch": 1.469657728847551e-13, "grad_norm": 0.628437340259552, "learning_rate": 6.459317088122094e-05, "loss": 2.2893, "step": 42360 }, { "epoch": 1.4700046735427463e-13, "grad_norm": 0.651005744934082, "learning_rate": 6.457086343845477e-05, "loss": 2.2819, "step": 42370 }, { "epoch": 1.4703516182379417e-13, "grad_norm": 0.6287461519241333, "learning_rate": 6.454855665658307e-05, "loss": 2.3246, "step": 42380 }, { "epoch": 1.470698562933137e-13, "grad_norm": 0.635413646697998, "learning_rate": 6.452625053913349e-05, "loss": 2.298, "step": 42390 }, { "epoch": 1.4710455076283324e-13, "grad_norm": 0.6354314088821411, "learning_rate": 6.450394508963352e-05, "loss": 2.2878, "step": 42400 }, { "epoch": 1.4713924523235278e-13, "grad_norm": 0.6147415041923523, "learning_rate": 6.448164031161057e-05, "loss": 2.2585, "step": 42410 }, { "epoch": 1.4717393970187231e-13, "grad_norm": 0.6237214803695679, "learning_rate": 6.445933620859199e-05, "loss": 2.2804, "step": 42420 }, { "epoch": 1.4720863417139185e-13, "grad_norm": 0.6372138857841492, "learning_rate": 6.443703278410492e-05, "loss": 2.256, "step": 42430 }, { "epoch": 1.4724332864091139e-13, "grad_norm": 0.6223177909851074, "learning_rate": 6.44147300416765e-05, "loss": 2.2457, "step": 42440 }, { "epoch": 1.4727802311043092e-13, "grad_norm": 0.690325140953064, "learning_rate": 6.43924279848337e-05, "loss": 2.2855, "step": 42450 }, { "epoch": 1.4731271757995046e-13, "grad_norm": 0.6120038628578186, "learning_rate": 6.437012661710343e-05, "loss": 2.2619, "step": 42460 }, { "epoch": 1.4734741204947e-13, "grad_norm": 0.6035917401313782, "learning_rate": 6.434782594201243e-05, "loss": 2.295, "step": 42470 }, { "epoch": 1.4738210651898953e-13, "grad_norm": 0.6228966116905212, "learning_rate": 6.432552596308732e-05, "loss": 2.2592, "step": 42480 }, { "epoch": 1.4741680098850907e-13, "grad_norm": 0.6237044334411621, "learning_rate": 6.430322668385476e-05, "loss": 2.2633, "step": 42490 }, { "epoch": 1.474514954580286e-13, "grad_norm": 0.6350477933883667, "learning_rate": 6.428092810784107e-05, "loss": 2.2912, "step": 42500 }, { "epoch": 1.4748618992754814e-13, "grad_norm": 0.6348557472229004, "learning_rate": 6.425863023857264e-05, "loss": 2.2852, "step": 42510 }, { "epoch": 1.4752088439706768e-13, "grad_norm": 0.6818049550056458, "learning_rate": 6.42363330795757e-05, "loss": 2.2695, "step": 42520 }, { "epoch": 1.475555788665872e-13, "grad_norm": 0.6323081254959106, "learning_rate": 6.42140366343763e-05, "loss": 2.2748, "step": 42530 }, { "epoch": 1.4759027333610675e-13, "grad_norm": 0.6162043809890747, "learning_rate": 6.419174090650047e-05, "loss": 2.2439, "step": 42540 }, { "epoch": 1.4762496780562628e-13, "grad_norm": 0.656658411026001, "learning_rate": 6.416944589947407e-05, "loss": 2.2918, "step": 42550 }, { "epoch": 1.4765966227514582e-13, "grad_norm": 0.6105976104736328, "learning_rate": 6.41471516168229e-05, "loss": 2.2837, "step": 42560 }, { "epoch": 1.4769435674466536e-13, "grad_norm": 0.632189154624939, "learning_rate": 6.412485806207255e-05, "loss": 2.2712, "step": 42570 }, { "epoch": 1.477290512141849e-13, "grad_norm": 0.6174262166023254, "learning_rate": 6.410256523874859e-05, "loss": 2.31, "step": 42580 }, { "epoch": 1.4776374568370443e-13, "grad_norm": 0.6398562788963318, "learning_rate": 6.408027315037646e-05, "loss": 2.27, "step": 42590 }, { "epoch": 1.4779844015322396e-13, "grad_norm": 0.6094763278961182, "learning_rate": 6.405798180048141e-05, "loss": 2.2465, "step": 42600 }, { "epoch": 1.478331346227435e-13, "grad_norm": 0.6576292514801025, "learning_rate": 6.403569119258864e-05, "loss": 2.2348, "step": 42610 }, { "epoch": 1.4786782909226304e-13, "grad_norm": 0.6358358860015869, "learning_rate": 6.401340133022323e-05, "loss": 2.3103, "step": 42620 }, { "epoch": 1.4790252356178257e-13, "grad_norm": 0.6362991333007812, "learning_rate": 6.399111221691014e-05, "loss": 2.2765, "step": 42630 }, { "epoch": 1.479372180313021e-13, "grad_norm": 0.6684389710426331, "learning_rate": 6.396882385617418e-05, "loss": 2.2591, "step": 42640 }, { "epoch": 1.4797191250082165e-13, "grad_norm": 0.6303521990776062, "learning_rate": 6.394653625154007e-05, "loss": 2.3154, "step": 42650 }, { "epoch": 1.4800660697034118e-13, "grad_norm": 0.6112372279167175, "learning_rate": 6.392424940653242e-05, "loss": 2.2525, "step": 42660 }, { "epoch": 1.4804130143986072e-13, "grad_norm": 0.6418835520744324, "learning_rate": 6.390196332467567e-05, "loss": 2.2554, "step": 42670 }, { "epoch": 1.4807599590938025e-13, "grad_norm": 0.6422322988510132, "learning_rate": 6.387967800949419e-05, "loss": 2.2623, "step": 42680 }, { "epoch": 1.481106903788998e-13, "grad_norm": 0.6208407282829285, "learning_rate": 6.385739346451219e-05, "loss": 2.3002, "step": 42690 }, { "epoch": 1.4814538484841933e-13, "grad_norm": 0.6210253238677979, "learning_rate": 6.383510969325382e-05, "loss": 2.2696, "step": 42700 }, { "epoch": 1.4818007931793886e-13, "grad_norm": 0.6334348320960999, "learning_rate": 6.381282669924306e-05, "loss": 2.2598, "step": 42710 }, { "epoch": 1.482147737874584e-13, "grad_norm": 0.5954141020774841, "learning_rate": 6.379054448600371e-05, "loss": 2.2931, "step": 42720 }, { "epoch": 1.4824946825697793e-13, "grad_norm": 0.6161249279975891, "learning_rate": 6.376826305705962e-05, "loss": 2.225, "step": 42730 }, { "epoch": 1.4828416272649747e-13, "grad_norm": 0.6235619783401489, "learning_rate": 6.37459824159343e-05, "loss": 2.2925, "step": 42740 }, { "epoch": 1.48318857196017e-13, "grad_norm": 0.6517283916473389, "learning_rate": 6.372370256615129e-05, "loss": 2.3073, "step": 42750 }, { "epoch": 1.4835355166553654e-13, "grad_norm": 0.6268484592437744, "learning_rate": 6.370142351123395e-05, "loss": 2.3187, "step": 42760 }, { "epoch": 1.4838824613505608e-13, "grad_norm": 0.6337164044380188, "learning_rate": 6.367914525470553e-05, "loss": 2.2733, "step": 42770 }, { "epoch": 1.4842294060457561e-13, "grad_norm": 0.6310874223709106, "learning_rate": 6.365686780008915e-05, "loss": 2.2926, "step": 42780 }, { "epoch": 1.4845763507409515e-13, "grad_norm": 0.6283103823661804, "learning_rate": 6.363459115090777e-05, "loss": 2.2927, "step": 42790 }, { "epoch": 1.4849232954361469e-13, "grad_norm": 0.6243809461593628, "learning_rate": 6.361231531068427e-05, "loss": 2.2842, "step": 42800 }, { "epoch": 1.4852702401313422e-13, "grad_norm": 0.613784909248352, "learning_rate": 6.359004028294139e-05, "loss": 2.2566, "step": 42810 }, { "epoch": 1.4856171848265376e-13, "grad_norm": 0.6525654196739197, "learning_rate": 6.356776607120171e-05, "loss": 2.2711, "step": 42820 }, { "epoch": 1.485964129521733e-13, "grad_norm": 0.6272528767585754, "learning_rate": 6.354549267898772e-05, "loss": 2.291, "step": 42830 }, { "epoch": 1.4863110742169283e-13, "grad_norm": 0.620411217212677, "learning_rate": 6.352322010982178e-05, "loss": 2.2494, "step": 42840 }, { "epoch": 1.4866580189121237e-13, "grad_norm": 0.6520774364471436, "learning_rate": 6.350094836722604e-05, "loss": 2.253, "step": 42850 }, { "epoch": 1.487004963607319e-13, "grad_norm": 0.6071910262107849, "learning_rate": 6.347867745472267e-05, "loss": 2.2658, "step": 42860 }, { "epoch": 1.4873519083025144e-13, "grad_norm": 0.6669682860374451, "learning_rate": 6.345640737583357e-05, "loss": 2.2731, "step": 42870 }, { "epoch": 1.4876988529977098e-13, "grad_norm": 0.5838926434516907, "learning_rate": 6.343413813408059e-05, "loss": 2.2809, "step": 42880 }, { "epoch": 1.488045797692905e-13, "grad_norm": 0.6382725238800049, "learning_rate": 6.341186973298539e-05, "loss": 2.2917, "step": 42890 }, { "epoch": 1.4883927423881005e-13, "grad_norm": 0.6208001375198364, "learning_rate": 6.338960217606953e-05, "loss": 2.2944, "step": 42900 }, { "epoch": 1.4887396870832958e-13, "grad_norm": 0.6538322567939758, "learning_rate": 6.336733546685448e-05, "loss": 2.2738, "step": 42910 }, { "epoch": 1.4890866317784912e-13, "grad_norm": 0.6277846097946167, "learning_rate": 6.334506960886146e-05, "loss": 2.2653, "step": 42920 }, { "epoch": 1.4894335764736866e-13, "grad_norm": 0.6464409232139587, "learning_rate": 6.332280460561168e-05, "loss": 2.2563, "step": 42930 }, { "epoch": 1.489780521168882e-13, "grad_norm": 0.6420938372612, "learning_rate": 6.330054046062612e-05, "loss": 2.2636, "step": 42940 }, { "epoch": 1.4901274658640773e-13, "grad_norm": 0.6493202447891235, "learning_rate": 6.327827717742565e-05, "loss": 2.2971, "step": 42950 }, { "epoch": 1.4904744105592727e-13, "grad_norm": 0.6505615711212158, "learning_rate": 6.325601475953109e-05, "loss": 2.2805, "step": 42960 }, { "epoch": 1.490821355254468e-13, "grad_norm": 0.622382640838623, "learning_rate": 6.323375321046299e-05, "loss": 2.3048, "step": 42970 }, { "epoch": 1.4911682999496634e-13, "grad_norm": 0.6493998169898987, "learning_rate": 6.321149253374183e-05, "loss": 2.2734, "step": 42980 }, { "epoch": 1.4915152446448587e-13, "grad_norm": 0.5876531600952148, "learning_rate": 6.318923273288794e-05, "loss": 2.2606, "step": 42990 }, { "epoch": 1.491862189340054e-13, "grad_norm": 0.6196975111961365, "learning_rate": 6.316697381142154e-05, "loss": 2.2628, "step": 43000 }, { "epoch": 1.4922091340352495e-13, "grad_norm": 0.6433928608894348, "learning_rate": 6.314471577286267e-05, "loss": 2.2625, "step": 43010 }, { "epoch": 1.4925560787304448e-13, "grad_norm": 0.627196729183197, "learning_rate": 6.312245862073123e-05, "loss": 2.2914, "step": 43020 }, { "epoch": 1.4929030234256402e-13, "grad_norm": 0.6570674777030945, "learning_rate": 6.310020235854706e-05, "loss": 2.3124, "step": 43030 }, { "epoch": 1.4932499681208355e-13, "grad_norm": 0.6107063889503479, "learning_rate": 6.307794698982971e-05, "loss": 2.2952, "step": 43040 }, { "epoch": 1.493596912816031e-13, "grad_norm": 0.6329039931297302, "learning_rate": 6.305569251809876e-05, "loss": 2.2404, "step": 43050 }, { "epoch": 1.4939438575112263e-13, "grad_norm": 0.6434234380722046, "learning_rate": 6.303343894687348e-05, "loss": 2.2611, "step": 43060 }, { "epoch": 1.4942908022064216e-13, "grad_norm": 0.6097171902656555, "learning_rate": 6.301118627967315e-05, "loss": 2.2576, "step": 43070 }, { "epoch": 1.494637746901617e-13, "grad_norm": 0.6630396246910095, "learning_rate": 6.298893452001679e-05, "loss": 2.3136, "step": 43080 }, { "epoch": 1.4949846915968124e-13, "grad_norm": 0.6431439518928528, "learning_rate": 6.296668367142334e-05, "loss": 2.2488, "step": 43090 }, { "epoch": 1.4953316362920077e-13, "grad_norm": 0.6391198039054871, "learning_rate": 6.294443373741163e-05, "loss": 2.2878, "step": 43100 }, { "epoch": 1.495678580987203e-13, "grad_norm": 0.6224705576896667, "learning_rate": 6.29221847215002e-05, "loss": 2.2476, "step": 43110 }, { "epoch": 1.4960255256823984e-13, "grad_norm": 0.6171806454658508, "learning_rate": 6.289993662720759e-05, "loss": 2.2738, "step": 43120 }, { "epoch": 1.4963724703775938e-13, "grad_norm": 0.6213910579681396, "learning_rate": 6.28776894580522e-05, "loss": 2.2316, "step": 43130 }, { "epoch": 1.4967194150727892e-13, "grad_norm": 0.6484046578407288, "learning_rate": 6.285544321755212e-05, "loss": 2.276, "step": 43140 }, { "epoch": 1.4970663597679845e-13, "grad_norm": 0.613954484462738, "learning_rate": 6.283319790922547e-05, "loss": 2.2968, "step": 43150 }, { "epoch": 1.49741330446318e-13, "grad_norm": 0.6323534846305847, "learning_rate": 6.281095353659013e-05, "loss": 2.2872, "step": 43160 }, { "epoch": 1.4977602491583752e-13, "grad_norm": 0.6360291242599487, "learning_rate": 6.278871010316387e-05, "loss": 2.2874, "step": 43170 }, { "epoch": 1.4981071938535706e-13, "grad_norm": 0.633061408996582, "learning_rate": 6.276646761246431e-05, "loss": 2.3229, "step": 43180 }, { "epoch": 1.498454138548766e-13, "grad_norm": 0.6012416481971741, "learning_rate": 6.274422606800886e-05, "loss": 2.2618, "step": 43190 }, { "epoch": 1.4988010832439613e-13, "grad_norm": 0.6582069993019104, "learning_rate": 6.27219854733149e-05, "loss": 2.2548, "step": 43200 }, { "epoch": 1.4991480279391567e-13, "grad_norm": 0.6381293535232544, "learning_rate": 6.269974583189951e-05, "loss": 2.2638, "step": 43210 }, { "epoch": 1.499494972634352e-13, "grad_norm": 0.6182708144187927, "learning_rate": 6.267750714727973e-05, "loss": 2.2718, "step": 43220 }, { "epoch": 1.4998419173295474e-13, "grad_norm": 0.6321814656257629, "learning_rate": 6.265526942297246e-05, "loss": 2.2877, "step": 43230 }, { "epoch": 1.5001888620247428e-13, "grad_norm": 0.6319812536239624, "learning_rate": 6.263303266249434e-05, "loss": 2.2786, "step": 43240 }, { "epoch": 1.5005358067199381e-13, "grad_norm": 0.6441251635551453, "learning_rate": 6.261079686936197e-05, "loss": 2.2817, "step": 43250 }, { "epoch": 1.5008827514151335e-13, "grad_norm": 0.6372296214103699, "learning_rate": 6.25885620470917e-05, "loss": 2.2521, "step": 43260 }, { "epoch": 1.5012296961103289e-13, "grad_norm": 0.621122419834137, "learning_rate": 6.256632819919982e-05, "loss": 2.2487, "step": 43270 }, { "epoch": 1.5015766408055242e-13, "grad_norm": 0.6311206221580505, "learning_rate": 6.254409532920241e-05, "loss": 2.2655, "step": 43280 }, { "epoch": 1.5019235855007196e-13, "grad_norm": 0.6081211566925049, "learning_rate": 6.252186344061539e-05, "loss": 2.2859, "step": 43290 }, { "epoch": 1.502270530195915e-13, "grad_norm": 0.6408022046089172, "learning_rate": 6.249963253695458e-05, "loss": 2.2758, "step": 43300 }, { "epoch": 1.5026174748911103e-13, "grad_norm": 0.6372427940368652, "learning_rate": 6.247740262173556e-05, "loss": 2.2257, "step": 43310 }, { "epoch": 1.5029644195863057e-13, "grad_norm": 0.6311560869216919, "learning_rate": 6.245517369847383e-05, "loss": 2.2796, "step": 43320 }, { "epoch": 1.503311364281501e-13, "grad_norm": 0.6192646026611328, "learning_rate": 6.243294577068467e-05, "loss": 2.3045, "step": 43330 }, { "epoch": 1.5036583089766964e-13, "grad_norm": 0.6314637064933777, "learning_rate": 6.241071884188328e-05, "loss": 2.2941, "step": 43340 }, { "epoch": 1.5040052536718918e-13, "grad_norm": 0.6473142504692078, "learning_rate": 6.238849291558464e-05, "loss": 2.2395, "step": 43350 }, { "epoch": 1.504352198367087e-13, "grad_norm": 0.6214422583580017, "learning_rate": 6.236626799530357e-05, "loss": 2.2785, "step": 43360 }, { "epoch": 1.5046991430622825e-13, "grad_norm": 0.614575207233429, "learning_rate": 6.23440440845548e-05, "loss": 2.2806, "step": 43370 }, { "epoch": 1.5050460877574778e-13, "grad_norm": 0.6298142671585083, "learning_rate": 6.232182118685277e-05, "loss": 2.2378, "step": 43380 }, { "epoch": 1.5053930324526732e-13, "grad_norm": 0.6160654425621033, "learning_rate": 6.22995993057119e-05, "loss": 2.2675, "step": 43390 }, { "epoch": 1.5057399771478686e-13, "grad_norm": 0.6492564082145691, "learning_rate": 6.227737844464638e-05, "loss": 2.2455, "step": 43400 }, { "epoch": 1.506086921843064e-13, "grad_norm": 0.6275131106376648, "learning_rate": 6.225515860717022e-05, "loss": 2.2742, "step": 43410 }, { "epoch": 1.5064338665382593e-13, "grad_norm": 0.653770387172699, "learning_rate": 6.223293979679737e-05, "loss": 2.2966, "step": 43420 }, { "epoch": 1.5067808112334546e-13, "grad_norm": 0.6337292194366455, "learning_rate": 6.221072201704144e-05, "loss": 2.2683, "step": 43430 }, { "epoch": 1.50712775592865e-13, "grad_norm": 0.6829734444618225, "learning_rate": 6.218850527141605e-05, "loss": 2.2939, "step": 43440 }, { "epoch": 1.5074747006238454e-13, "grad_norm": 0.631050705909729, "learning_rate": 6.216628956343458e-05, "loss": 2.2975, "step": 43450 }, { "epoch": 1.5078216453190407e-13, "grad_norm": 0.6387974619865417, "learning_rate": 6.214407489661023e-05, "loss": 2.271, "step": 43460 }, { "epoch": 1.508168590014236e-13, "grad_norm": 0.6252230405807495, "learning_rate": 6.212186127445609e-05, "loss": 2.2463, "step": 43470 }, { "epoch": 1.5085155347094314e-13, "grad_norm": 0.6250976920127869, "learning_rate": 6.2099648700485e-05, "loss": 2.227, "step": 43480 }, { "epoch": 1.5088624794046268e-13, "grad_norm": 0.6172824501991272, "learning_rate": 6.207743717820974e-05, "loss": 2.2606, "step": 43490 }, { "epoch": 1.5092094240998222e-13, "grad_norm": 0.608857274055481, "learning_rate": 6.205522671114288e-05, "loss": 2.3, "step": 43500 }, { "epoch": 1.5095563687950175e-13, "grad_norm": 0.6451225876808167, "learning_rate": 6.203301730279674e-05, "loss": 2.2757, "step": 43510 }, { "epoch": 1.509903313490213e-13, "grad_norm": 0.6240336894989014, "learning_rate": 6.201080895668362e-05, "loss": 2.2808, "step": 43520 }, { "epoch": 1.5102502581854083e-13, "grad_norm": 0.6318941116333008, "learning_rate": 6.198860167631554e-05, "loss": 2.2771, "step": 43530 }, { "epoch": 1.5105972028806036e-13, "grad_norm": 0.6594717502593994, "learning_rate": 6.19663954652044e-05, "loss": 2.2963, "step": 43540 }, { "epoch": 1.510944147575799e-13, "grad_norm": 0.6482905745506287, "learning_rate": 6.194419032686194e-05, "loss": 2.2883, "step": 43550 }, { "epoch": 1.5112910922709943e-13, "grad_norm": 0.6213971376419067, "learning_rate": 6.19219862647997e-05, "loss": 2.2649, "step": 43560 }, { "epoch": 1.5116380369661897e-13, "grad_norm": 0.6299808621406555, "learning_rate": 6.189978328252903e-05, "loss": 2.2684, "step": 43570 }, { "epoch": 1.511984981661385e-13, "grad_norm": 0.6338030099868774, "learning_rate": 6.187758138356117e-05, "loss": 2.2641, "step": 43580 }, { "epoch": 1.5123319263565804e-13, "grad_norm": 0.6390663385391235, "learning_rate": 6.185538057140716e-05, "loss": 2.2908, "step": 43590 }, { "epoch": 1.5126788710517758e-13, "grad_norm": 0.6417630314826965, "learning_rate": 6.18331808495779e-05, "loss": 2.2623, "step": 43600 }, { "epoch": 1.5130258157469711e-13, "grad_norm": 0.6221705079078674, "learning_rate": 6.1810982221584e-05, "loss": 2.2474, "step": 43610 }, { "epoch": 1.5133727604421665e-13, "grad_norm": 0.645975649356842, "learning_rate": 6.178878469093606e-05, "loss": 2.2768, "step": 43620 }, { "epoch": 1.5137197051373619e-13, "grad_norm": 0.6628739237785339, "learning_rate": 6.176658826114438e-05, "loss": 2.2934, "step": 43630 }, { "epoch": 1.5140666498325572e-13, "grad_norm": 0.6226702332496643, "learning_rate": 6.174439293571917e-05, "loss": 2.2898, "step": 43640 }, { "epoch": 1.5144135945277526e-13, "grad_norm": 0.6273365616798401, "learning_rate": 6.172219871817038e-05, "loss": 2.2529, "step": 43650 }, { "epoch": 1.514760539222948e-13, "grad_norm": 0.6297997236251831, "learning_rate": 6.170000561200788e-05, "loss": 2.2697, "step": 43660 }, { "epoch": 1.5151074839181433e-13, "grad_norm": 0.6684073209762573, "learning_rate": 6.167781362074134e-05, "loss": 2.2606, "step": 43670 }, { "epoch": 1.5154544286133387e-13, "grad_norm": 0.6374183297157288, "learning_rate": 6.165562274788015e-05, "loss": 2.313, "step": 43680 }, { "epoch": 1.515801373308534e-13, "grad_norm": 0.6333292126655579, "learning_rate": 6.16334329969337e-05, "loss": 2.2846, "step": 43690 }, { "epoch": 1.5161483180037294e-13, "grad_norm": 0.700228214263916, "learning_rate": 6.161124437141102e-05, "loss": 2.2892, "step": 43700 }, { "epoch": 1.5164952626989248e-13, "grad_norm": 0.6140533089637756, "learning_rate": 6.15890568748211e-05, "loss": 2.3096, "step": 43710 }, { "epoch": 1.51684220739412e-13, "grad_norm": 0.6523221731185913, "learning_rate": 6.156687051067271e-05, "loss": 2.2864, "step": 43720 }, { "epoch": 1.5171891520893155e-13, "grad_norm": 0.6310720443725586, "learning_rate": 6.154468528247441e-05, "loss": 2.2806, "step": 43730 }, { "epoch": 1.5175360967845108e-13, "grad_norm": 0.6196998953819275, "learning_rate": 6.152250119373462e-05, "loss": 2.2781, "step": 43740 }, { "epoch": 1.5178830414797062e-13, "grad_norm": 0.6303544044494629, "learning_rate": 6.150031824796153e-05, "loss": 2.2885, "step": 43750 }, { "epoch": 1.5182299861749016e-13, "grad_norm": 0.6176508069038391, "learning_rate": 6.147813644866321e-05, "loss": 2.2881, "step": 43760 }, { "epoch": 1.518576930870097e-13, "grad_norm": 0.6455751061439514, "learning_rate": 6.145595579934753e-05, "loss": 2.2964, "step": 43770 }, { "epoch": 1.5189238755652923e-13, "grad_norm": 0.6446955800056458, "learning_rate": 6.143377630352212e-05, "loss": 2.2634, "step": 43780 }, { "epoch": 1.5192708202604877e-13, "grad_norm": 0.6410716772079468, "learning_rate": 6.141159796469454e-05, "loss": 2.3204, "step": 43790 }, { "epoch": 1.519617764955683e-13, "grad_norm": 0.641045331954956, "learning_rate": 6.138942078637205e-05, "loss": 2.3038, "step": 43800 }, { "epoch": 1.5199647096508784e-13, "grad_norm": 0.6410529017448425, "learning_rate": 6.13672447720618e-05, "loss": 2.2614, "step": 43810 }, { "epoch": 1.5203116543460737e-13, "grad_norm": 0.5920453071594238, "learning_rate": 6.134506992527076e-05, "loss": 2.2474, "step": 43820 }, { "epoch": 1.520658599041269e-13, "grad_norm": 0.6783031225204468, "learning_rate": 6.132289624950568e-05, "loss": 2.2877, "step": 43830 }, { "epoch": 1.5210055437364645e-13, "grad_norm": 0.6617169976234436, "learning_rate": 6.130072374827312e-05, "loss": 2.2729, "step": 43840 }, { "epoch": 1.5213524884316598e-13, "grad_norm": 0.6466237306594849, "learning_rate": 6.127855242507943e-05, "loss": 2.2777, "step": 43850 }, { "epoch": 1.5216994331268552e-13, "grad_norm": 0.6397132873535156, "learning_rate": 6.125638228343089e-05, "loss": 2.27, "step": 43860 }, { "epoch": 1.5220463778220505e-13, "grad_norm": 0.647975742816925, "learning_rate": 6.123421332683348e-05, "loss": 2.2693, "step": 43870 }, { "epoch": 1.522393322517246e-13, "grad_norm": 0.6411163210868835, "learning_rate": 6.121204555879305e-05, "loss": 2.2832, "step": 43880 }, { "epoch": 1.5227402672124413e-13, "grad_norm": 0.6673391461372375, "learning_rate": 6.118987898281523e-05, "loss": 2.2757, "step": 43890 }, { "epoch": 1.5230872119076366e-13, "grad_norm": 0.6306431293487549, "learning_rate": 6.116771360240546e-05, "loss": 2.2528, "step": 43900 }, { "epoch": 1.523434156602832e-13, "grad_norm": 0.6430878043174744, "learning_rate": 6.1145549421069e-05, "loss": 2.2907, "step": 43910 }, { "epoch": 1.5237811012980274e-13, "grad_norm": 0.6253710389137268, "learning_rate": 6.112338644231096e-05, "loss": 2.2887, "step": 43920 }, { "epoch": 1.5241280459932227e-13, "grad_norm": 0.6325743198394775, "learning_rate": 6.11012246696362e-05, "loss": 2.3094, "step": 43930 }, { "epoch": 1.524474990688418e-13, "grad_norm": 0.6345682740211487, "learning_rate": 6.107906410654943e-05, "loss": 2.2513, "step": 43940 }, { "epoch": 1.5248219353836134e-13, "grad_norm": 0.6283734440803528, "learning_rate": 6.105690475655512e-05, "loss": 2.2674, "step": 43950 }, { "epoch": 1.5251688800788088e-13, "grad_norm": 0.6304333209991455, "learning_rate": 6.103474662315761e-05, "loss": 2.2489, "step": 43960 }, { "epoch": 1.5255158247740042e-13, "grad_norm": 0.6455303430557251, "learning_rate": 6.1012589709861e-05, "loss": 2.2884, "step": 43970 }, { "epoch": 1.5258627694691995e-13, "grad_norm": 0.6544297337532043, "learning_rate": 6.0990434020169214e-05, "loss": 2.2755, "step": 43980 }, { "epoch": 1.526209714164395e-13, "grad_norm": 0.686224102973938, "learning_rate": 6.0968279557586034e-05, "loss": 2.2767, "step": 43990 }, { "epoch": 1.5265566588595902e-13, "grad_norm": 0.6485445499420166, "learning_rate": 6.0946126325614915e-05, "loss": 2.2943, "step": 44000 }, { "epoch": 1.5269036035547856e-13, "grad_norm": 0.6134834885597229, "learning_rate": 6.0923974327759274e-05, "loss": 2.265, "step": 44010 }, { "epoch": 1.527250548249981e-13, "grad_norm": 0.6202679872512817, "learning_rate": 6.09018235675222e-05, "loss": 2.2674, "step": 44020 }, { "epoch": 1.5275974929451763e-13, "grad_norm": 0.6403876543045044, "learning_rate": 6.087967404840666e-05, "loss": 2.2461, "step": 44030 }, { "epoch": 1.5279444376403717e-13, "grad_norm": 0.6685853600502014, "learning_rate": 6.085752577391546e-05, "loss": 2.2971, "step": 44040 }, { "epoch": 1.528291382335567e-13, "grad_norm": 0.6506821513175964, "learning_rate": 6.0835378747551086e-05, "loss": 2.3187, "step": 44050 }, { "epoch": 1.5286383270307624e-13, "grad_norm": 0.6338192224502563, "learning_rate": 6.081323297281596e-05, "loss": 2.2426, "step": 44060 }, { "epoch": 1.5289852717259578e-13, "grad_norm": 0.6314855217933655, "learning_rate": 6.07910884532122e-05, "loss": 2.2883, "step": 44070 }, { "epoch": 1.5293322164211531e-13, "grad_norm": 0.6329444050788879, "learning_rate": 6.0768945192241795e-05, "loss": 2.289, "step": 44080 }, { "epoch": 1.5296791611163485e-13, "grad_norm": 0.646793007850647, "learning_rate": 6.0746803193406517e-05, "loss": 2.2577, "step": 44090 }, { "epoch": 1.5300261058115439e-13, "grad_norm": 0.6617595553398132, "learning_rate": 6.0724662460207916e-05, "loss": 2.247, "step": 44100 }, { "epoch": 1.5303730505067392e-13, "grad_norm": 0.6412273049354553, "learning_rate": 6.070252299614737e-05, "loss": 2.2992, "step": 44110 }, { "epoch": 1.5307199952019346e-13, "grad_norm": 0.6539624333381653, "learning_rate": 6.0680384804726015e-05, "loss": 2.2849, "step": 44120 }, { "epoch": 1.53106693989713e-13, "grad_norm": 0.6526220440864563, "learning_rate": 6.065824788944485e-05, "loss": 2.2754, "step": 44130 }, { "epoch": 1.5314138845923253e-13, "grad_norm": 0.6682146191596985, "learning_rate": 6.063611225380464e-05, "loss": 2.279, "step": 44140 }, { "epoch": 1.5317608292875207e-13, "grad_norm": 0.6260033249855042, "learning_rate": 6.061397790130592e-05, "loss": 2.3185, "step": 44150 }, { "epoch": 1.532107773982716e-13, "grad_norm": 0.6595573425292969, "learning_rate": 6.059184483544908e-05, "loss": 2.26, "step": 44160 }, { "epoch": 1.5324547186779114e-13, "grad_norm": 0.6310247182846069, "learning_rate": 6.056971305973423e-05, "loss": 2.2904, "step": 44170 }, { "epoch": 1.5328016633731067e-13, "grad_norm": 0.6518720388412476, "learning_rate": 6.054758257766134e-05, "loss": 2.2697, "step": 44180 }, { "epoch": 1.533148608068302e-13, "grad_norm": 0.6500868797302246, "learning_rate": 6.052545339273017e-05, "loss": 2.2756, "step": 44190 }, { "epoch": 1.5334955527634975e-13, "grad_norm": 0.6210752725601196, "learning_rate": 6.050332550844024e-05, "loss": 2.2489, "step": 44200 }, { "epoch": 1.5338424974586928e-13, "grad_norm": 0.6463794708251953, "learning_rate": 6.048119892829092e-05, "loss": 2.275, "step": 44210 }, { "epoch": 1.5341894421538882e-13, "grad_norm": 0.6216136813163757, "learning_rate": 6.045907365578128e-05, "loss": 2.3063, "step": 44220 }, { "epoch": 1.5345363868490836e-13, "grad_norm": 0.6587020754814148, "learning_rate": 6.0436949694410287e-05, "loss": 2.2531, "step": 44230 }, { "epoch": 1.534883331544279e-13, "grad_norm": 0.6299315094947815, "learning_rate": 6.041482704767665e-05, "loss": 2.2973, "step": 44240 }, { "epoch": 1.5352302762394743e-13, "grad_norm": 0.6423088312149048, "learning_rate": 6.0392705719078845e-05, "loss": 2.2785, "step": 44250 }, { "epoch": 1.5355772209346696e-13, "grad_norm": 0.6206254363059998, "learning_rate": 6.037058571211521e-05, "loss": 2.2773, "step": 44260 }, { "epoch": 1.535924165629865e-13, "grad_norm": 0.729271650314331, "learning_rate": 6.0348467030283796e-05, "loss": 2.2893, "step": 44270 }, { "epoch": 1.5362711103250604e-13, "grad_norm": 0.6610986590385437, "learning_rate": 6.03263496770825e-05, "loss": 2.2894, "step": 44280 }, { "epoch": 1.5366180550202557e-13, "grad_norm": 0.636085033416748, "learning_rate": 6.030423365600901e-05, "loss": 2.2669, "step": 44290 }, { "epoch": 1.536964999715451e-13, "grad_norm": 0.6315069198608398, "learning_rate": 6.028211897056074e-05, "loss": 2.2631, "step": 44300 }, { "epoch": 1.5373119444106464e-13, "grad_norm": 0.6609626412391663, "learning_rate": 6.0260005624235e-05, "loss": 2.2777, "step": 44310 }, { "epoch": 1.5376588891058418e-13, "grad_norm": 0.6276383399963379, "learning_rate": 6.023789362052876e-05, "loss": 2.2806, "step": 44320 }, { "epoch": 1.5380058338010372e-13, "grad_norm": 0.6158760190010071, "learning_rate": 6.0215782962938894e-05, "loss": 2.2363, "step": 44330 }, { "epoch": 1.5383527784962325e-13, "grad_norm": 0.6271770000457764, "learning_rate": 6.019367365496195e-05, "loss": 2.2793, "step": 44340 }, { "epoch": 1.538699723191428e-13, "grad_norm": 0.6299939155578613, "learning_rate": 6.0171565700094384e-05, "loss": 2.2803, "step": 44350 }, { "epoch": 1.5390466678866233e-13, "grad_norm": 0.6325493454933167, "learning_rate": 6.0149459101832366e-05, "loss": 2.2786, "step": 44360 }, { "epoch": 1.5393936125818186e-13, "grad_norm": 0.6251688003540039, "learning_rate": 6.012735386367184e-05, "loss": 2.2814, "step": 44370 }, { "epoch": 1.539740557277014e-13, "grad_norm": 0.6376455426216125, "learning_rate": 6.010524998910862e-05, "loss": 2.2844, "step": 44380 }, { "epoch": 1.5400875019722093e-13, "grad_norm": 0.6146237850189209, "learning_rate": 6.008314748163815e-05, "loss": 2.295, "step": 44390 }, { "epoch": 1.5404344466674047e-13, "grad_norm": 0.635073721408844, "learning_rate": 6.006104634475582e-05, "loss": 2.2513, "step": 44400 }, { "epoch": 1.5407813913626e-13, "grad_norm": 0.6288551092147827, "learning_rate": 6.0038946581956715e-05, "loss": 2.2559, "step": 44410 }, { "epoch": 1.5411283360577954e-13, "grad_norm": 0.6092137098312378, "learning_rate": 6.0016848196735724e-05, "loss": 2.2765, "step": 44420 }, { "epoch": 1.5414752807529908e-13, "grad_norm": 0.7244047522544861, "learning_rate": 5.999475119258751e-05, "loss": 2.2596, "step": 44430 }, { "epoch": 1.5418222254481861e-13, "grad_norm": 0.6425639986991882, "learning_rate": 5.997265557300653e-05, "loss": 2.2782, "step": 44440 }, { "epoch": 1.5421691701433815e-13, "grad_norm": 0.6253790855407715, "learning_rate": 5.9950561341487006e-05, "loss": 2.2417, "step": 44450 }, { "epoch": 1.5425161148385769e-13, "grad_norm": 0.684918224811554, "learning_rate": 5.992846850152299e-05, "loss": 2.2525, "step": 44460 }, { "epoch": 1.5428630595337722e-13, "grad_norm": 0.6359127759933472, "learning_rate": 5.990637705660822e-05, "loss": 2.2912, "step": 44470 }, { "epoch": 1.5432100042289676e-13, "grad_norm": 0.651907205581665, "learning_rate": 5.988428701023631e-05, "loss": 2.269, "step": 44480 }, { "epoch": 1.543556948924163e-13, "grad_norm": 0.6181255578994751, "learning_rate": 5.9862198365900566e-05, "loss": 2.2816, "step": 44490 }, { "epoch": 1.5439038936193583e-13, "grad_norm": 0.6385693550109863, "learning_rate": 5.9840111127094156e-05, "loss": 2.3033, "step": 44500 }, { "epoch": 1.5442508383145537e-13, "grad_norm": 0.6128095388412476, "learning_rate": 5.981802529730999e-05, "loss": 2.2764, "step": 44510 }, { "epoch": 1.544597783009749e-13, "grad_norm": 0.6269370317459106, "learning_rate": 5.97959408800407e-05, "loss": 2.2591, "step": 44520 }, { "epoch": 1.5449447277049444e-13, "grad_norm": 0.6568520069122314, "learning_rate": 5.977385787877883e-05, "loss": 2.2606, "step": 44530 }, { "epoch": 1.5452916724001398e-13, "grad_norm": 0.6126347184181213, "learning_rate": 5.975177629701654e-05, "loss": 2.2637, "step": 44540 }, { "epoch": 1.545638617095335e-13, "grad_norm": 0.6276071667671204, "learning_rate": 5.9729696138245874e-05, "loss": 2.2842, "step": 44550 }, { "epoch": 1.5459855617905305e-13, "grad_norm": 0.6390787959098816, "learning_rate": 5.9707617405958637e-05, "loss": 2.2684, "step": 44560 }, { "epoch": 1.5463325064857258e-13, "grad_norm": 0.6481234431266785, "learning_rate": 5.968554010364634e-05, "loss": 2.283, "step": 44570 }, { "epoch": 1.5466794511809212e-13, "grad_norm": 0.6392210721969604, "learning_rate": 5.966346423480036e-05, "loss": 2.2256, "step": 44580 }, { "epoch": 1.5470263958761166e-13, "grad_norm": 0.6447489261627197, "learning_rate": 5.9641389802911785e-05, "loss": 2.297, "step": 44590 }, { "epoch": 1.547373340571312e-13, "grad_norm": 0.6486096382141113, "learning_rate": 5.961931681147152e-05, "loss": 2.2844, "step": 44600 }, { "epoch": 1.5477202852665073e-13, "grad_norm": 0.6558685898780823, "learning_rate": 5.9597245263970204e-05, "loss": 2.2776, "step": 44610 }, { "epoch": 1.5480672299617027e-13, "grad_norm": 0.6515876650810242, "learning_rate": 5.957517516389825e-05, "loss": 2.2574, "step": 44620 }, { "epoch": 1.548414174656898e-13, "grad_norm": 0.6271746754646301, "learning_rate": 5.955310651474588e-05, "loss": 2.2683, "step": 44630 }, { "epoch": 1.5487611193520934e-13, "grad_norm": 0.6408298015594482, "learning_rate": 5.953103932000303e-05, "loss": 2.259, "step": 44640 }, { "epoch": 1.5491080640472887e-13, "grad_norm": 0.6388988494873047, "learning_rate": 5.950897358315947e-05, "loss": 2.2568, "step": 44650 }, { "epoch": 1.549455008742484e-13, "grad_norm": 0.6340315937995911, "learning_rate": 5.948690930770469e-05, "loss": 2.2915, "step": 44660 }, { "epoch": 1.5498019534376795e-13, "grad_norm": 0.6335310339927673, "learning_rate": 5.946484649712795e-05, "loss": 2.2571, "step": 44670 }, { "epoch": 1.5501488981328748e-13, "grad_norm": 0.6677462458610535, "learning_rate": 5.9442785154918326e-05, "loss": 2.2806, "step": 44680 }, { "epoch": 1.5504958428280702e-13, "grad_norm": 0.657321572303772, "learning_rate": 5.94207252845646e-05, "loss": 2.2629, "step": 44690 }, { "epoch": 1.5508427875232655e-13, "grad_norm": 0.6287903785705566, "learning_rate": 5.9398666889555375e-05, "loss": 2.2724, "step": 44700 }, { "epoch": 1.551189732218461e-13, "grad_norm": 0.6393409371376038, "learning_rate": 5.9376609973378985e-05, "loss": 2.2891, "step": 44710 }, { "epoch": 1.5515366769136563e-13, "grad_norm": 0.6496983766555786, "learning_rate": 5.935455453952353e-05, "loss": 2.2726, "step": 44720 }, { "epoch": 1.5518836216088516e-13, "grad_norm": 0.6253910064697266, "learning_rate": 5.933250059147691e-05, "loss": 2.2488, "step": 44730 }, { "epoch": 1.552230566304047e-13, "grad_norm": 0.6506061553955078, "learning_rate": 5.9310448132726734e-05, "loss": 2.2715, "step": 44740 }, { "epoch": 1.5525775109992423e-13, "grad_norm": 0.6646672487258911, "learning_rate": 5.928839716676045e-05, "loss": 2.2962, "step": 44750 }, { "epoch": 1.5529244556944377e-13, "grad_norm": 0.6602984666824341, "learning_rate": 5.9266347697065195e-05, "loss": 2.2617, "step": 44760 }, { "epoch": 1.553271400389633e-13, "grad_norm": 0.6616416573524475, "learning_rate": 5.9244299727127914e-05, "loss": 2.2774, "step": 44770 }, { "epoch": 1.5536183450848284e-13, "grad_norm": 0.6487297415733337, "learning_rate": 5.922225326043533e-05, "loss": 2.2506, "step": 44780 }, { "epoch": 1.5539652897800238e-13, "grad_norm": 0.6373663544654846, "learning_rate": 5.920020830047386e-05, "loss": 2.2928, "step": 44790 }, { "epoch": 1.5543122344752192e-13, "grad_norm": 0.6521641612052917, "learning_rate": 5.9178164850729755e-05, "loss": 2.2843, "step": 44800 }, { "epoch": 1.5546591791704145e-13, "grad_norm": 0.6198095083236694, "learning_rate": 5.915612291468897e-05, "loss": 2.2936, "step": 44810 }, { "epoch": 1.55500612386561e-13, "grad_norm": 0.6531361937522888, "learning_rate": 5.913408249583725e-05, "loss": 2.256, "step": 44820 }, { "epoch": 1.5553530685608052e-13, "grad_norm": 0.6318401098251343, "learning_rate": 5.9112043597660136e-05, "loss": 2.2372, "step": 44830 }, { "epoch": 1.5557000132560006e-13, "grad_norm": 0.6772351264953613, "learning_rate": 5.909000622364283e-05, "loss": 2.2633, "step": 44840 }, { "epoch": 1.556046957951196e-13, "grad_norm": 0.7113361954689026, "learning_rate": 5.9067970377270435e-05, "loss": 2.2707, "step": 44850 }, { "epoch": 1.5563939026463913e-13, "grad_norm": 0.6575614213943481, "learning_rate": 5.904593606202764e-05, "loss": 2.2997, "step": 44860 }, { "epoch": 1.5567408473415867e-13, "grad_norm": 0.6322242021560669, "learning_rate": 5.902390328139902e-05, "loss": 2.2282, "step": 44870 }, { "epoch": 1.557087792036782e-13, "grad_norm": 0.648428738117218, "learning_rate": 5.9001872038868904e-05, "loss": 2.2407, "step": 44880 }, { "epoch": 1.5574347367319774e-13, "grad_norm": 0.6432998180389404, "learning_rate": 5.8979842337921296e-05, "loss": 2.2916, "step": 44890 }, { "epoch": 1.5577816814271728e-13, "grad_norm": 0.6362506151199341, "learning_rate": 5.895781418204002e-05, "loss": 2.2632, "step": 44900 }, { "epoch": 1.5581286261223681e-13, "grad_norm": 0.6277554631233215, "learning_rate": 5.893578757470863e-05, "loss": 2.2689, "step": 44910 }, { "epoch": 1.5584755708175635e-13, "grad_norm": 0.6272978186607361, "learning_rate": 5.891376251941047e-05, "loss": 2.2961, "step": 44920 }, { "epoch": 1.5588225155127589e-13, "grad_norm": 0.6374770998954773, "learning_rate": 5.8891739019628603e-05, "loss": 2.2597, "step": 44930 }, { "epoch": 1.5591694602079542e-13, "grad_norm": 0.6397907733917236, "learning_rate": 5.886971707884582e-05, "loss": 2.2708, "step": 44940 }, { "epoch": 1.5595164049031496e-13, "grad_norm": 0.6373804807662964, "learning_rate": 5.884769670054477e-05, "loss": 2.295, "step": 44950 }, { "epoch": 1.559863349598345e-13, "grad_norm": 0.6416113376617432, "learning_rate": 5.8825677888207726e-05, "loss": 2.2459, "step": 44960 }, { "epoch": 1.5602102942935403e-13, "grad_norm": 0.6359094977378845, "learning_rate": 5.880366064531679e-05, "loss": 2.2412, "step": 44970 }, { "epoch": 1.5605572389887357e-13, "grad_norm": 0.657007098197937, "learning_rate": 5.8781644975353804e-05, "loss": 2.2267, "step": 44980 }, { "epoch": 1.560904183683931e-13, "grad_norm": 0.6215488314628601, "learning_rate": 5.8759630881800364e-05, "loss": 2.2705, "step": 44990 }, { "epoch": 1.5612511283791264e-13, "grad_norm": 0.6561613082885742, "learning_rate": 5.8737618368137806e-05, "loss": 2.2523, "step": 45000 }, { "epoch": 1.5615980730743217e-13, "grad_norm": 0.6818130612373352, "learning_rate": 5.871560743784721e-05, "loss": 2.259, "step": 45010 }, { "epoch": 1.561945017769517e-13, "grad_norm": 0.6317986249923706, "learning_rate": 5.869359809440942e-05, "loss": 2.2451, "step": 45020 }, { "epoch": 1.5622919624647125e-13, "grad_norm": 0.6473068594932556, "learning_rate": 5.867159034130502e-05, "loss": 2.2749, "step": 45030 }, { "epoch": 1.5626389071599078e-13, "grad_norm": 0.6687057018280029, "learning_rate": 5.864958418201437e-05, "loss": 2.2853, "step": 45040 }, { "epoch": 1.5629858518551032e-13, "grad_norm": 0.6647137999534607, "learning_rate": 5.8627579620017536e-05, "loss": 2.2682, "step": 45050 }, { "epoch": 1.5633327965502986e-13, "grad_norm": 0.6234599947929382, "learning_rate": 5.860557665879433e-05, "loss": 2.2562, "step": 45060 }, { "epoch": 1.563679741245494e-13, "grad_norm": 0.6421186327934265, "learning_rate": 5.858357530182437e-05, "loss": 2.3004, "step": 45070 }, { "epoch": 1.5640266859406893e-13, "grad_norm": 0.6590601205825806, "learning_rate": 5.856157555258694e-05, "loss": 2.2524, "step": 45080 }, { "epoch": 1.5643736306358846e-13, "grad_norm": 0.615868330001831, "learning_rate": 5.853957741456115e-05, "loss": 2.2338, "step": 45090 }, { "epoch": 1.56472057533108e-13, "grad_norm": 0.6743262410163879, "learning_rate": 5.85175808912258e-05, "loss": 2.2886, "step": 45100 }, { "epoch": 1.5650675200262754e-13, "grad_norm": 0.6345463991165161, "learning_rate": 5.849558598605945e-05, "loss": 2.2774, "step": 45110 }, { "epoch": 1.5654144647214707e-13, "grad_norm": 0.647571325302124, "learning_rate": 5.847359270254042e-05, "loss": 2.2898, "step": 45120 }, { "epoch": 1.565761409416666e-13, "grad_norm": 0.6327414512634277, "learning_rate": 5.8451601044146724e-05, "loss": 2.2805, "step": 45130 }, { "epoch": 1.5661083541118614e-13, "grad_norm": 0.9272290468215942, "learning_rate": 5.842961101435616e-05, "loss": 2.2496, "step": 45140 }, { "epoch": 1.5664552988070568e-13, "grad_norm": 0.6437693238258362, "learning_rate": 5.840762261664631e-05, "loss": 2.2675, "step": 45150 }, { "epoch": 1.5668022435022522e-13, "grad_norm": 0.6358346939086914, "learning_rate": 5.838563585449439e-05, "loss": 2.272, "step": 45160 }, { "epoch": 1.5671491881974475e-13, "grad_norm": 0.6506536602973938, "learning_rate": 5.8363650731377464e-05, "loss": 2.259, "step": 45170 }, { "epoch": 1.567496132892643e-13, "grad_norm": 0.645283579826355, "learning_rate": 5.834166725077225e-05, "loss": 2.2512, "step": 45180 }, { "epoch": 1.5678430775878383e-13, "grad_norm": 0.6331977248191833, "learning_rate": 5.831968541615526e-05, "loss": 2.2753, "step": 45190 }, { "epoch": 1.5681900222830336e-13, "grad_norm": 0.6620502471923828, "learning_rate": 5.829770523100277e-05, "loss": 2.2552, "step": 45200 }, { "epoch": 1.568536966978229e-13, "grad_norm": 0.6764684915542603, "learning_rate": 5.8275726698790695e-05, "loss": 2.2877, "step": 45210 }, { "epoch": 1.5688839116734243e-13, "grad_norm": 0.6655756831169128, "learning_rate": 5.825374982299479e-05, "loss": 2.257, "step": 45220 }, { "epoch": 1.5692308563686197e-13, "grad_norm": 0.6327849626541138, "learning_rate": 5.823177460709048e-05, "loss": 2.2648, "step": 45230 }, { "epoch": 1.569577801063815e-13, "grad_norm": 0.6365050673484802, "learning_rate": 5.820980105455298e-05, "loss": 2.2915, "step": 45240 }, { "epoch": 1.5699247457590104e-13, "grad_norm": 0.616590678691864, "learning_rate": 5.818782916885724e-05, "loss": 2.2291, "step": 45250 }, { "epoch": 1.5702716904542058e-13, "grad_norm": 0.64516282081604, "learning_rate": 5.816585895347787e-05, "loss": 2.2559, "step": 45260 }, { "epoch": 1.5706186351494011e-13, "grad_norm": 0.6646727919578552, "learning_rate": 5.814389041188932e-05, "loss": 2.3021, "step": 45270 }, { "epoch": 1.5709655798445965e-13, "grad_norm": 0.68141770362854, "learning_rate": 5.812192354756568e-05, "loss": 2.2428, "step": 45280 }, { "epoch": 1.5713125245397919e-13, "grad_norm": 0.6431976556777954, "learning_rate": 5.809995836398085e-05, "loss": 2.2746, "step": 45290 }, { "epoch": 1.5716594692349872e-13, "grad_norm": 0.6615820527076721, "learning_rate": 5.807799486460843e-05, "loss": 2.2475, "step": 45300 }, { "epoch": 1.5720064139301826e-13, "grad_norm": 0.6598491072654724, "learning_rate": 5.8056033052921756e-05, "loss": 2.2636, "step": 45310 }, { "epoch": 1.572353358625378e-13, "grad_norm": 0.6488229036331177, "learning_rate": 5.8034072932393916e-05, "loss": 2.297, "step": 45320 }, { "epoch": 1.5727003033205733e-13, "grad_norm": 0.685732364654541, "learning_rate": 5.801211450649768e-05, "loss": 2.2444, "step": 45330 }, { "epoch": 1.5730472480157687e-13, "grad_norm": 0.6551774740219116, "learning_rate": 5.799015777870563e-05, "loss": 2.2913, "step": 45340 }, { "epoch": 1.573394192710964e-13, "grad_norm": 0.6405503749847412, "learning_rate": 5.796820275248997e-05, "loss": 2.242, "step": 45350 }, { "epoch": 1.5737411374061594e-13, "grad_norm": 0.6241506338119507, "learning_rate": 5.7946249431322734e-05, "loss": 2.2186, "step": 45360 }, { "epoch": 1.5740880821013548e-13, "grad_norm": 0.6351121664047241, "learning_rate": 5.792429781867567e-05, "loss": 2.2605, "step": 45370 }, { "epoch": 1.57443502679655e-13, "grad_norm": 0.6389881372451782, "learning_rate": 5.790234791802022e-05, "loss": 2.2393, "step": 45380 }, { "epoch": 1.5747819714917455e-13, "grad_norm": 0.6306284070014954, "learning_rate": 5.788039973282756e-05, "loss": 2.2903, "step": 45390 }, { "epoch": 1.5751289161869408e-13, "grad_norm": 0.6648603081703186, "learning_rate": 5.785845326656862e-05, "loss": 2.2754, "step": 45400 }, { "epoch": 1.5754758608821362e-13, "grad_norm": 0.7138168811798096, "learning_rate": 5.7836508522714036e-05, "loss": 2.2645, "step": 45410 }, { "epoch": 1.5758228055773316e-13, "grad_norm": 0.6668126583099365, "learning_rate": 5.78145655047342e-05, "loss": 2.2688, "step": 45420 }, { "epoch": 1.576169750272527e-13, "grad_norm": 0.6329817175865173, "learning_rate": 5.7792624216099184e-05, "loss": 2.2784, "step": 45430 }, { "epoch": 1.5765166949677223e-13, "grad_norm": 0.6631296277046204, "learning_rate": 5.7770684660278834e-05, "loss": 2.2776, "step": 45440 }, { "epoch": 1.5768636396629176e-13, "grad_norm": 0.6577072739601135, "learning_rate": 5.7748746840742675e-05, "loss": 2.2554, "step": 45450 }, { "epoch": 1.577210584358113e-13, "grad_norm": 0.6657236218452454, "learning_rate": 5.772681076096002e-05, "loss": 2.2868, "step": 45460 }, { "epoch": 1.5775575290533084e-13, "grad_norm": 0.635830283164978, "learning_rate": 5.770487642439985e-05, "loss": 2.287, "step": 45470 }, { "epoch": 1.5779044737485037e-13, "grad_norm": 0.6379696130752563, "learning_rate": 5.768294383453089e-05, "loss": 2.2498, "step": 45480 }, { "epoch": 1.578251418443699e-13, "grad_norm": 0.6281797289848328, "learning_rate": 5.7661012994821625e-05, "loss": 2.2696, "step": 45490 }, { "epoch": 1.5785983631388945e-13, "grad_norm": 0.6396915316581726, "learning_rate": 5.763908390874016e-05, "loss": 2.2592, "step": 45500 }, { "epoch": 1.5789453078340898e-13, "grad_norm": 0.6427440047264099, "learning_rate": 5.761715657975446e-05, "loss": 2.2799, "step": 45510 }, { "epoch": 1.5792922525292852e-13, "grad_norm": 0.6500253081321716, "learning_rate": 5.7595231011332114e-05, "loss": 2.2598, "step": 45520 }, { "epoch": 1.5796391972244805e-13, "grad_norm": 0.6226224303245544, "learning_rate": 5.757330720694044e-05, "loss": 2.2614, "step": 45530 }, { "epoch": 1.579986141919676e-13, "grad_norm": 0.6442948579788208, "learning_rate": 5.7551385170046545e-05, "loss": 2.2671, "step": 45540 }, { "epoch": 1.5803330866148713e-13, "grad_norm": 0.6406270861625671, "learning_rate": 5.7529464904117174e-05, "loss": 2.2525, "step": 45550 }, { "epoch": 1.5806800313100666e-13, "grad_norm": 0.6645494699478149, "learning_rate": 5.7507546412618836e-05, "loss": 2.2545, "step": 45560 }, { "epoch": 1.581026976005262e-13, "grad_norm": 0.6608544588088989, "learning_rate": 5.7485629699017785e-05, "loss": 2.2737, "step": 45570 }, { "epoch": 1.5813739207004573e-13, "grad_norm": 0.636687695980072, "learning_rate": 5.7463714766779904e-05, "loss": 2.2834, "step": 45580 }, { "epoch": 1.5817208653956527e-13, "grad_norm": 0.6703835725784302, "learning_rate": 5.744180161937091e-05, "loss": 2.2397, "step": 45590 }, { "epoch": 1.582067810090848e-13, "grad_norm": 0.6742992401123047, "learning_rate": 5.741989026025611e-05, "loss": 2.264, "step": 45600 }, { "epoch": 1.5824147547860434e-13, "grad_norm": 0.6307300925254822, "learning_rate": 5.739798069290067e-05, "loss": 2.2592, "step": 45610 }, { "epoch": 1.5827616994812388e-13, "grad_norm": 0.6740711331367493, "learning_rate": 5.737607292076934e-05, "loss": 2.2709, "step": 45620 }, { "epoch": 1.5831086441764342e-13, "grad_norm": 0.6362267732620239, "learning_rate": 5.7354166947326664e-05, "loss": 2.2523, "step": 45630 }, { "epoch": 1.5834555888716295e-13, "grad_norm": 0.6672873497009277, "learning_rate": 5.7332262776036925e-05, "loss": 2.2623, "step": 45640 }, { "epoch": 1.583802533566825e-13, "grad_norm": 0.6595062017440796, "learning_rate": 5.7310360410364014e-05, "loss": 2.2867, "step": 45650 }, { "epoch": 1.5841494782620202e-13, "grad_norm": 0.6619556546211243, "learning_rate": 5.728845985377165e-05, "loss": 2.2609, "step": 45660 }, { "epoch": 1.5844964229572156e-13, "grad_norm": 0.6348480582237244, "learning_rate": 5.726656110972318e-05, "loss": 2.2825, "step": 45670 }, { "epoch": 1.584843367652411e-13, "grad_norm": 0.6451088786125183, "learning_rate": 5.7244664181681716e-05, "loss": 2.2583, "step": 45680 }, { "epoch": 1.5851903123476063e-13, "grad_norm": 0.6263609528541565, "learning_rate": 5.7222769073110096e-05, "loss": 2.2739, "step": 45690 }, { "epoch": 1.5855372570428017e-13, "grad_norm": 0.673694908618927, "learning_rate": 5.720087578747081e-05, "loss": 2.2639, "step": 45700 }, { "epoch": 1.585884201737997e-13, "grad_norm": 0.6678094267845154, "learning_rate": 5.71789843282261e-05, "loss": 2.2385, "step": 45710 }, { "epoch": 1.5862311464331924e-13, "grad_norm": 0.6366090178489685, "learning_rate": 5.715709469883791e-05, "loss": 2.242, "step": 45720 }, { "epoch": 1.5865780911283878e-13, "grad_norm": 0.665158212184906, "learning_rate": 5.7135206902767896e-05, "loss": 2.2504, "step": 45730 }, { "epoch": 1.5869250358235831e-13, "grad_norm": 0.6649131774902344, "learning_rate": 5.711332094347746e-05, "loss": 2.2796, "step": 45740 }, { "epoch": 1.5872719805187785e-13, "grad_norm": 0.6236298680305481, "learning_rate": 5.709143682442762e-05, "loss": 2.2525, "step": 45750 }, { "epoch": 1.5876189252139739e-13, "grad_norm": 0.6299776434898376, "learning_rate": 5.706955454907921e-05, "loss": 2.3153, "step": 45760 }, { "epoch": 1.5879658699091692e-13, "grad_norm": 0.6896002888679504, "learning_rate": 5.704767412089269e-05, "loss": 2.2612, "step": 45770 }, { "epoch": 1.5883128146043646e-13, "grad_norm": 0.6505682468414307, "learning_rate": 5.7025795543328266e-05, "loss": 2.2628, "step": 45780 }, { "epoch": 1.58865975929956e-13, "grad_norm": 0.6634906530380249, "learning_rate": 5.700391881984586e-05, "loss": 2.2158, "step": 45790 }, { "epoch": 1.5890067039947553e-13, "grad_norm": 0.6300268769264221, "learning_rate": 5.6982043953905075e-05, "loss": 2.245, "step": 45800 }, { "epoch": 1.5893536486899507e-13, "grad_norm": 0.6400492191314697, "learning_rate": 5.696017094896526e-05, "loss": 2.2843, "step": 45810 }, { "epoch": 1.589700593385146e-13, "grad_norm": 0.6509669423103333, "learning_rate": 5.69382998084854e-05, "loss": 2.2592, "step": 45820 }, { "epoch": 1.5900475380803414e-13, "grad_norm": 0.6620467901229858, "learning_rate": 5.6916430535924256e-05, "loss": 2.2754, "step": 45830 }, { "epoch": 1.5903944827755367e-13, "grad_norm": 0.6355090141296387, "learning_rate": 5.689456313474027e-05, "loss": 2.2864, "step": 45840 }, { "epoch": 1.590741427470732e-13, "grad_norm": 0.6609693765640259, "learning_rate": 5.687269760839156e-05, "loss": 2.279, "step": 45850 }, { "epoch": 1.5910883721659275e-13, "grad_norm": 0.657814085483551, "learning_rate": 5.685083396033598e-05, "loss": 2.241, "step": 45860 }, { "epoch": 1.5914353168611228e-13, "grad_norm": 0.6278262734413147, "learning_rate": 5.6828972194031075e-05, "loss": 2.304, "step": 45870 }, { "epoch": 1.5917822615563182e-13, "grad_norm": 0.6529012322425842, "learning_rate": 5.6807112312934106e-05, "loss": 2.2615, "step": 45880 }, { "epoch": 1.5921292062515136e-13, "grad_norm": 0.6600780487060547, "learning_rate": 5.678525432050205e-05, "loss": 2.2153, "step": 45890 }, { "epoch": 1.592476150946709e-13, "grad_norm": 0.659832775592804, "learning_rate": 5.67633982201915e-05, "loss": 2.2807, "step": 45900 }, { "epoch": 1.5928230956419043e-13, "grad_norm": 0.6633009910583496, "learning_rate": 5.674154401545885e-05, "loss": 2.2593, "step": 45910 }, { "epoch": 1.5931700403370996e-13, "grad_norm": 0.6331675052642822, "learning_rate": 5.671969170976013e-05, "loss": 2.264, "step": 45920 }, { "epoch": 1.593516985032295e-13, "grad_norm": 0.6225167512893677, "learning_rate": 5.669784130655112e-05, "loss": 2.2273, "step": 45930 }, { "epoch": 1.5938639297274904e-13, "grad_norm": 0.662460207939148, "learning_rate": 5.667599280928726e-05, "loss": 2.216, "step": 45940 }, { "epoch": 1.5942108744226857e-13, "grad_norm": 0.6491160988807678, "learning_rate": 5.665414622142369e-05, "loss": 2.2761, "step": 45950 }, { "epoch": 1.594557819117881e-13, "grad_norm": 0.6469171047210693, "learning_rate": 5.66323015464153e-05, "loss": 2.2591, "step": 45960 }, { "epoch": 1.5949047638130764e-13, "grad_norm": 0.652906596660614, "learning_rate": 5.661045878771659e-05, "loss": 2.2641, "step": 45970 }, { "epoch": 1.5952517085082718e-13, "grad_norm": 0.6579794883728027, "learning_rate": 5.6588617948781854e-05, "loss": 2.27, "step": 45980 }, { "epoch": 1.5955986532034672e-13, "grad_norm": 0.6566510200500488, "learning_rate": 5.6566779033064966e-05, "loss": 2.296, "step": 45990 }, { "epoch": 1.5959455978986625e-13, "grad_norm": 0.6266568303108215, "learning_rate": 5.654494204401961e-05, "loss": 2.2694, "step": 46000 }, { "epoch": 1.596292542593858e-13, "grad_norm": 0.672868013381958, "learning_rate": 5.652310698509912e-05, "loss": 2.3039, "step": 46010 }, { "epoch": 1.5966394872890532e-13, "grad_norm": 0.6611214876174927, "learning_rate": 5.6501273859756494e-05, "loss": 2.2593, "step": 46020 }, { "epoch": 1.5969864319842486e-13, "grad_norm": 0.653467059135437, "learning_rate": 5.64794426714445e-05, "loss": 2.265, "step": 46030 }, { "epoch": 1.597333376679444e-13, "grad_norm": 0.6491495966911316, "learning_rate": 5.64576134236155e-05, "loss": 2.2613, "step": 46040 }, { "epoch": 1.5976803213746393e-13, "grad_norm": 0.6595935821533203, "learning_rate": 5.643578611972161e-05, "loss": 2.2611, "step": 46050 }, { "epoch": 1.5980272660698347e-13, "grad_norm": 0.6159700155258179, "learning_rate": 5.6413960763214656e-05, "loss": 2.2696, "step": 46060 }, { "epoch": 1.59837421076503e-13, "grad_norm": 0.6429363489151001, "learning_rate": 5.6392137357546125e-05, "loss": 2.276, "step": 46070 }, { "epoch": 1.5987211554602254e-13, "grad_norm": 0.6197131872177124, "learning_rate": 5.637031590616719e-05, "loss": 2.2725, "step": 46080 }, { "epoch": 1.5990681001554208e-13, "grad_norm": 0.6423285603523254, "learning_rate": 5.6348496412528735e-05, "loss": 2.2397, "step": 46090 }, { "epoch": 1.5994150448506161e-13, "grad_norm": 0.6553585529327393, "learning_rate": 5.6326678880081304e-05, "loss": 2.2726, "step": 46100 }, { "epoch": 1.5997619895458115e-13, "grad_norm": 0.644662618637085, "learning_rate": 5.630486331227518e-05, "loss": 2.2804, "step": 46110 }, { "epoch": 1.6001089342410069e-13, "grad_norm": 0.6553017497062683, "learning_rate": 5.628304971256028e-05, "loss": 2.269, "step": 46120 }, { "epoch": 1.6004558789362022e-13, "grad_norm": 0.6549898386001587, "learning_rate": 5.626123808438628e-05, "loss": 2.2911, "step": 46130 }, { "epoch": 1.6008028236313976e-13, "grad_norm": 0.6293321251869202, "learning_rate": 5.623942843120244e-05, "loss": 2.2332, "step": 46140 }, { "epoch": 1.601149768326593e-13, "grad_norm": 0.6313555836677551, "learning_rate": 5.621762075645783e-05, "loss": 2.3014, "step": 46150 }, { "epoch": 1.6014967130217883e-13, "grad_norm": 0.649871826171875, "learning_rate": 5.6195815063601105e-05, "loss": 2.2537, "step": 46160 }, { "epoch": 1.6018436577169837e-13, "grad_norm": 0.6291935443878174, "learning_rate": 5.617401135608065e-05, "loss": 2.2766, "step": 46170 }, { "epoch": 1.602190602412179e-13, "grad_norm": 0.6430219411849976, "learning_rate": 5.615220963734456e-05, "loss": 2.2856, "step": 46180 }, { "epoch": 1.6025375471073744e-13, "grad_norm": 0.6461676955223083, "learning_rate": 5.613040991084056e-05, "loss": 2.2724, "step": 46190 }, { "epoch": 1.6028844918025698e-13, "grad_norm": 0.6470482349395752, "learning_rate": 5.6108612180016104e-05, "loss": 2.2305, "step": 46200 }, { "epoch": 1.603231436497765e-13, "grad_norm": 0.6422904133796692, "learning_rate": 5.608681644831834e-05, "loss": 2.2867, "step": 46210 }, { "epoch": 1.6035783811929605e-13, "grad_norm": 0.6672900319099426, "learning_rate": 5.606502271919402e-05, "loss": 2.2667, "step": 46220 }, { "epoch": 1.6039253258881558e-13, "grad_norm": 0.6275389790534973, "learning_rate": 5.6043230996089686e-05, "loss": 2.2203, "step": 46230 }, { "epoch": 1.6042722705833512e-13, "grad_norm": 0.6430591344833374, "learning_rate": 5.602144128245147e-05, "loss": 2.2698, "step": 46240 }, { "epoch": 1.6046192152785466e-13, "grad_norm": 0.6429505944252014, "learning_rate": 5.599965358172525e-05, "loss": 2.223, "step": 46250 }, { "epoch": 1.604966159973742e-13, "grad_norm": 0.6285407543182373, "learning_rate": 5.5977867897356565e-05, "loss": 2.2865, "step": 46260 }, { "epoch": 1.6053131046689373e-13, "grad_norm": 0.6503690481185913, "learning_rate": 5.595608423279062e-05, "loss": 2.257, "step": 46270 }, { "epoch": 1.6056600493641326e-13, "grad_norm": 0.6371756196022034, "learning_rate": 5.593430259147235e-05, "loss": 2.2516, "step": 46280 }, { "epoch": 1.606006994059328e-13, "grad_norm": 0.642767071723938, "learning_rate": 5.591252297684628e-05, "loss": 2.2766, "step": 46290 }, { "epoch": 1.6063539387545234e-13, "grad_norm": 0.6446628570556641, "learning_rate": 5.5890745392356706e-05, "loss": 2.2875, "step": 46300 }, { "epoch": 1.6067008834497187e-13, "grad_norm": 0.6313104033470154, "learning_rate": 5.5868969841447556e-05, "loss": 2.2716, "step": 46310 }, { "epoch": 1.607047828144914e-13, "grad_norm": 0.6604300737380981, "learning_rate": 5.5847196327562425e-05, "loss": 2.2584, "step": 46320 }, { "epoch": 1.6073947728401095e-13, "grad_norm": 0.6584978103637695, "learning_rate": 5.582542485414466e-05, "loss": 2.2428, "step": 46330 }, { "epoch": 1.6077417175353048e-13, "grad_norm": 0.6447314023971558, "learning_rate": 5.5803655424637165e-05, "loss": 2.2897, "step": 46340 }, { "epoch": 1.6080886622305002e-13, "grad_norm": 0.640203058719635, "learning_rate": 5.578188804248266e-05, "loss": 2.2586, "step": 46350 }, { "epoch": 1.6084356069256955e-13, "grad_norm": 0.6289672255516052, "learning_rate": 5.5760122711123396e-05, "loss": 2.2763, "step": 46360 }, { "epoch": 1.608782551620891e-13, "grad_norm": 0.6381742358207703, "learning_rate": 5.573835943400141e-05, "loss": 2.2731, "step": 46370 }, { "epoch": 1.6091294963160863e-13, "grad_norm": 0.6592421531677246, "learning_rate": 5.5716598214558395e-05, "loss": 2.2663, "step": 46380 }, { "epoch": 1.6094764410112816e-13, "grad_norm": 0.6601542830467224, "learning_rate": 5.569483905623566e-05, "loss": 2.2623, "step": 46390 }, { "epoch": 1.609823385706477e-13, "grad_norm": 0.6416941285133362, "learning_rate": 5.567308196247427e-05, "loss": 2.2576, "step": 46400 }, { "epoch": 1.6101703304016723e-13, "grad_norm": 0.6604258418083191, "learning_rate": 5.565132693671489e-05, "loss": 2.2669, "step": 46410 }, { "epoch": 1.6105172750968677e-13, "grad_norm": 0.622443675994873, "learning_rate": 5.56295739823979e-05, "loss": 2.2466, "step": 46420 }, { "epoch": 1.610864219792063e-13, "grad_norm": 0.6523269414901733, "learning_rate": 5.560782310296335e-05, "loss": 2.2443, "step": 46430 }, { "epoch": 1.6112111644872584e-13, "grad_norm": 0.6185179352760315, "learning_rate": 5.558607430185093e-05, "loss": 2.2738, "step": 46440 }, { "epoch": 1.6115581091824538e-13, "grad_norm": 0.6459250450134277, "learning_rate": 5.556432758250007e-05, "loss": 2.2738, "step": 46450 }, { "epoch": 1.6119050538776492e-13, "grad_norm": 0.6330058574676514, "learning_rate": 5.5542582948349767e-05, "loss": 2.2704, "step": 46460 }, { "epoch": 1.6122519985728445e-13, "grad_norm": 0.7403548359870911, "learning_rate": 5.5520840402838795e-05, "loss": 2.2799, "step": 46470 }, { "epoch": 1.61259894326804e-13, "grad_norm": 0.6108893752098083, "learning_rate": 5.549909994940555e-05, "loss": 2.2823, "step": 46480 }, { "epoch": 1.6129458879632352e-13, "grad_norm": 0.6367582678794861, "learning_rate": 5.547736159148807e-05, "loss": 2.2636, "step": 46490 }, { "epoch": 1.6132928326584306e-13, "grad_norm": 0.6442914009094238, "learning_rate": 5.5455625332524115e-05, "loss": 2.2701, "step": 46500 }, { "epoch": 1.613639777353626e-13, "grad_norm": 0.6496990323066711, "learning_rate": 5.5433891175951044e-05, "loss": 2.2606, "step": 46510 }, { "epoch": 1.6139867220488213e-13, "grad_norm": 0.6378779411315918, "learning_rate": 5.5412159125205977e-05, "loss": 2.2581, "step": 46520 }, { "epoch": 1.6143336667440167e-13, "grad_norm": 0.6757528781890869, "learning_rate": 5.539042918372565e-05, "loss": 2.2422, "step": 46530 }, { "epoch": 1.614680611439212e-13, "grad_norm": 0.6545336246490479, "learning_rate": 5.5368701354946426e-05, "loss": 2.2696, "step": 46540 }, { "epoch": 1.6150275561344074e-13, "grad_norm": 0.6605238318443298, "learning_rate": 5.5346975642304415e-05, "loss": 2.2563, "step": 46550 }, { "epoch": 1.6153745008296028e-13, "grad_norm": 0.6622722744941711, "learning_rate": 5.53252520492353e-05, "loss": 2.2966, "step": 46560 }, { "epoch": 1.6157214455247981e-13, "grad_norm": 0.6640079021453857, "learning_rate": 5.530353057917453e-05, "loss": 2.2827, "step": 46570 }, { "epoch": 1.6160683902199935e-13, "grad_norm": 0.6438538432121277, "learning_rate": 5.528181123555716e-05, "loss": 2.2584, "step": 46580 }, { "epoch": 1.6164153349151889e-13, "grad_norm": 0.6667578816413879, "learning_rate": 5.5260094021817895e-05, "loss": 2.2427, "step": 46590 }, { "epoch": 1.6167622796103842e-13, "grad_norm": 0.6468599438667297, "learning_rate": 5.523837894139115e-05, "loss": 2.2735, "step": 46600 }, { "epoch": 1.6171092243055796e-13, "grad_norm": 0.6381464600563049, "learning_rate": 5.5216665997710954e-05, "loss": 2.2446, "step": 46610 }, { "epoch": 1.617456169000775e-13, "grad_norm": 0.6169043183326721, "learning_rate": 5.5194955194211054e-05, "loss": 2.251, "step": 46620 }, { "epoch": 1.6178031136959703e-13, "grad_norm": 0.662562370300293, "learning_rate": 5.5173246534324776e-05, "loss": 2.2651, "step": 46630 }, { "epoch": 1.6181500583911657e-13, "grad_norm": 0.6331522464752197, "learning_rate": 5.5151540021485206e-05, "loss": 2.2804, "step": 46640 }, { "epoch": 1.618497003086361e-13, "grad_norm": 0.6437850594520569, "learning_rate": 5.512983565912502e-05, "loss": 2.2598, "step": 46650 }, { "epoch": 1.6188439477815564e-13, "grad_norm": 0.6397480368614197, "learning_rate": 5.5108133450676584e-05, "loss": 2.2831, "step": 46660 }, { "epoch": 1.6191908924767517e-13, "grad_norm": 0.6636770963668823, "learning_rate": 5.5086433399571935e-05, "loss": 2.27, "step": 46670 }, { "epoch": 1.619537837171947e-13, "grad_norm": 0.6574015617370605, "learning_rate": 5.50647355092427e-05, "loss": 2.2354, "step": 46680 }, { "epoch": 1.6198847818671425e-13, "grad_norm": 0.6528858542442322, "learning_rate": 5.504303978312024e-05, "loss": 2.3026, "step": 46690 }, { "epoch": 1.6202317265623378e-13, "grad_norm": 0.6418211460113525, "learning_rate": 5.502134622463557e-05, "loss": 2.2403, "step": 46700 }, { "epoch": 1.6205786712575332e-13, "grad_norm": 0.6723405718803406, "learning_rate": 5.4999654837219293e-05, "loss": 2.2579, "step": 46710 }, { "epoch": 1.6209256159527285e-13, "grad_norm": 0.6618378162384033, "learning_rate": 5.4977965624301756e-05, "loss": 2.2464, "step": 46720 }, { "epoch": 1.621272560647924e-13, "grad_norm": 0.6179137229919434, "learning_rate": 5.495627858931289e-05, "loss": 2.2235, "step": 46730 }, { "epoch": 1.6216195053431193e-13, "grad_norm": 0.6983278393745422, "learning_rate": 5.493459373568232e-05, "loss": 2.2404, "step": 46740 }, { "epoch": 1.6219664500383146e-13, "grad_norm": 0.6754205226898193, "learning_rate": 5.4912911066839374e-05, "loss": 2.265, "step": 46750 }, { "epoch": 1.62231339473351e-13, "grad_norm": 0.6517820358276367, "learning_rate": 5.48912305862129e-05, "loss": 2.2363, "step": 46760 }, { "epoch": 1.6226603394287054e-13, "grad_norm": 0.6579523682594299, "learning_rate": 5.486955229723152e-05, "loss": 2.2732, "step": 46770 }, { "epoch": 1.6230072841239007e-13, "grad_norm": 0.683289110660553, "learning_rate": 5.4847876203323444e-05, "loss": 2.2705, "step": 46780 }, { "epoch": 1.623354228819096e-13, "grad_norm": 0.6588258147239685, "learning_rate": 5.482620230791659e-05, "loss": 2.2809, "step": 46790 }, { "epoch": 1.6237011735142914e-13, "grad_norm": 0.6358712315559387, "learning_rate": 5.4804530614438475e-05, "loss": 2.2542, "step": 46800 }, { "epoch": 1.6240481182094868e-13, "grad_norm": 0.6675070524215698, "learning_rate": 5.478286112631631e-05, "loss": 2.2715, "step": 46810 }, { "epoch": 1.6243950629046822e-13, "grad_norm": 0.6285918354988098, "learning_rate": 5.4761193846976936e-05, "loss": 2.2617, "step": 46820 }, { "epoch": 1.6247420075998775e-13, "grad_norm": 0.6584048867225647, "learning_rate": 5.473952877984681e-05, "loss": 2.2418, "step": 46830 }, { "epoch": 1.625088952295073e-13, "grad_norm": 0.6248345971107483, "learning_rate": 5.4717865928352106e-05, "loss": 2.2755, "step": 46840 }, { "epoch": 1.6254358969902682e-13, "grad_norm": 0.6432167291641235, "learning_rate": 5.469620529591865e-05, "loss": 2.2322, "step": 46850 }, { "epoch": 1.6257828416854636e-13, "grad_norm": 0.6193022727966309, "learning_rate": 5.467454688597183e-05, "loss": 2.2669, "step": 46860 }, { "epoch": 1.626129786380659e-13, "grad_norm": 0.6421118974685669, "learning_rate": 5.4652890701936785e-05, "loss": 2.2918, "step": 46870 }, { "epoch": 1.6264767310758543e-13, "grad_norm": 0.6659696698188782, "learning_rate": 5.4631236747238187e-05, "loss": 2.2812, "step": 46880 }, { "epoch": 1.6268236757710497e-13, "grad_norm": 0.6286711692810059, "learning_rate": 5.460958502530047e-05, "loss": 2.2814, "step": 46890 }, { "epoch": 1.627170620466245e-13, "grad_norm": 0.6380379796028137, "learning_rate": 5.4587935539547663e-05, "loss": 2.2717, "step": 46900 }, { "epoch": 1.6275175651614404e-13, "grad_norm": 0.6296868920326233, "learning_rate": 5.456628829340343e-05, "loss": 2.2531, "step": 46910 }, { "epoch": 1.6278645098566358e-13, "grad_norm": 0.6568717956542969, "learning_rate": 5.4544643290291144e-05, "loss": 2.2909, "step": 46920 }, { "epoch": 1.6282114545518311e-13, "grad_norm": 0.6614868640899658, "learning_rate": 5.4523000533633724e-05, "loss": 2.2841, "step": 46930 }, { "epoch": 1.6285583992470265e-13, "grad_norm": 0.6480491757392883, "learning_rate": 5.450136002685381e-05, "loss": 2.2654, "step": 46940 }, { "epoch": 1.6289053439422219e-13, "grad_norm": 0.652580201625824, "learning_rate": 5.447972177337364e-05, "loss": 2.2418, "step": 46950 }, { "epoch": 1.6292522886374172e-13, "grad_norm": 0.644186794757843, "learning_rate": 5.4458085776615126e-05, "loss": 2.2704, "step": 46960 }, { "epoch": 1.6295992333326126e-13, "grad_norm": 0.6247049570083618, "learning_rate": 5.443645203999983e-05, "loss": 2.2439, "step": 46970 }, { "epoch": 1.629946178027808e-13, "grad_norm": 0.6614518761634827, "learning_rate": 5.4414820566948934e-05, "loss": 2.2572, "step": 46980 }, { "epoch": 1.6302931227230033e-13, "grad_norm": 0.6488226652145386, "learning_rate": 5.4393191360883275e-05, "loss": 2.2396, "step": 46990 }, { "epoch": 1.6306400674181987e-13, "grad_norm": 0.6243111491203308, "learning_rate": 5.43715644252233e-05, "loss": 2.2011, "step": 47000 }, { "epoch": 1.630987012113394e-13, "grad_norm": 0.6893453001976013, "learning_rate": 5.434993976338916e-05, "loss": 2.2332, "step": 47010 }, { "epoch": 1.6313339568085894e-13, "grad_norm": 0.6808966994285583, "learning_rate": 5.432831737880059e-05, "loss": 2.2853, "step": 47020 }, { "epoch": 1.6316809015037848e-13, "grad_norm": 0.6311924457550049, "learning_rate": 5.430669727487697e-05, "loss": 2.2934, "step": 47030 }, { "epoch": 1.63202784619898e-13, "grad_norm": 0.6762214303016663, "learning_rate": 5.428507945503736e-05, "loss": 2.2248, "step": 47040 }, { "epoch": 1.6323747908941755e-13, "grad_norm": 0.6526935696601868, "learning_rate": 5.4263463922700384e-05, "loss": 2.2357, "step": 47050 }, { "epoch": 1.6327217355893708e-13, "grad_norm": 0.6678684949874878, "learning_rate": 5.424185068128441e-05, "loss": 2.257, "step": 47060 }, { "epoch": 1.6330686802845662e-13, "grad_norm": 0.65354323387146, "learning_rate": 5.422023973420738e-05, "loss": 2.2699, "step": 47070 }, { "epoch": 1.6334156249797616e-13, "grad_norm": 0.652792751789093, "learning_rate": 5.419863108488686e-05, "loss": 2.2647, "step": 47080 }, { "epoch": 1.633762569674957e-13, "grad_norm": 0.6464436650276184, "learning_rate": 5.4177024736740074e-05, "loss": 2.283, "step": 47090 }, { "epoch": 1.6341095143701523e-13, "grad_norm": 0.667171061038971, "learning_rate": 5.415542069318387e-05, "loss": 2.2723, "step": 47100 }, { "epoch": 1.6344564590653476e-13, "grad_norm": 0.6563230156898499, "learning_rate": 5.413381895763474e-05, "loss": 2.2461, "step": 47110 }, { "epoch": 1.634803403760543e-13, "grad_norm": 0.6725370287895203, "learning_rate": 5.411221953350885e-05, "loss": 2.2151, "step": 47120 }, { "epoch": 1.6351503484557384e-13, "grad_norm": 0.691878080368042, "learning_rate": 5.409062242422193e-05, "loss": 2.2562, "step": 47130 }, { "epoch": 1.6354972931509337e-13, "grad_norm": 0.6631229519844055, "learning_rate": 5.4069027633189384e-05, "loss": 2.2094, "step": 47140 }, { "epoch": 1.635844237846129e-13, "grad_norm": 0.6448671221733093, "learning_rate": 5.404743516382624e-05, "loss": 2.2409, "step": 47150 }, { "epoch": 1.6361911825413245e-13, "grad_norm": 0.6400879621505737, "learning_rate": 5.402584501954716e-05, "loss": 2.2567, "step": 47160 }, { "epoch": 1.6365381272365198e-13, "grad_norm": 0.6609793901443481, "learning_rate": 5.4004257203766464e-05, "loss": 2.2656, "step": 47170 }, { "epoch": 1.6368850719317152e-13, "grad_norm": 0.6196011900901794, "learning_rate": 5.3982671719898026e-05, "loss": 2.2565, "step": 47180 }, { "epoch": 1.6372320166269105e-13, "grad_norm": 0.6467150449752808, "learning_rate": 5.3961088571355485e-05, "loss": 2.2341, "step": 47190 }, { "epoch": 1.637578961322106e-13, "grad_norm": 0.6619185209274292, "learning_rate": 5.393950776155195e-05, "loss": 2.234, "step": 47200 }, { "epoch": 1.6379259060173013e-13, "grad_norm": 0.6752145290374756, "learning_rate": 5.391792929390025e-05, "loss": 2.2614, "step": 47210 }, { "epoch": 1.6382728507124966e-13, "grad_norm": 0.653263509273529, "learning_rate": 5.3896353171812894e-05, "loss": 2.2764, "step": 47220 }, { "epoch": 1.638619795407692e-13, "grad_norm": 0.6623634099960327, "learning_rate": 5.38747793987019e-05, "loss": 2.2414, "step": 47230 }, { "epoch": 1.6389667401028873e-13, "grad_norm": 0.6825626492500305, "learning_rate": 5.385320797797903e-05, "loss": 2.2741, "step": 47240 }, { "epoch": 1.6393136847980827e-13, "grad_norm": 0.680928647518158, "learning_rate": 5.3831638913055566e-05, "loss": 2.2543, "step": 47250 }, { "epoch": 1.639660629493278e-13, "grad_norm": 0.6761045455932617, "learning_rate": 5.38100722073425e-05, "loss": 2.2894, "step": 47260 }, { "epoch": 1.6400075741884734e-13, "grad_norm": 0.6520535349845886, "learning_rate": 5.37885078642504e-05, "loss": 2.248, "step": 47270 }, { "epoch": 1.6403545188836688e-13, "grad_norm": 0.6909275650978088, "learning_rate": 5.376694588718948e-05, "loss": 2.2539, "step": 47280 }, { "epoch": 1.6407014635788641e-13, "grad_norm": 0.6701956391334534, "learning_rate": 5.374538627956961e-05, "loss": 2.2425, "step": 47290 }, { "epoch": 1.6410484082740595e-13, "grad_norm": 0.6581993103027344, "learning_rate": 5.372382904480022e-05, "loss": 2.2518, "step": 47300 }, { "epoch": 1.6413953529692549e-13, "grad_norm": 0.6469643712043762, "learning_rate": 5.370227418629045e-05, "loss": 2.226, "step": 47310 }, { "epoch": 1.6417422976644502e-13, "grad_norm": 0.683096706867218, "learning_rate": 5.368072170744898e-05, "loss": 2.2484, "step": 47320 }, { "epoch": 1.6420892423596456e-13, "grad_norm": 0.6638057231903076, "learning_rate": 5.365917161168413e-05, "loss": 2.2579, "step": 47330 }, { "epoch": 1.642436187054841e-13, "grad_norm": 0.6201068758964539, "learning_rate": 5.363762390240391e-05, "loss": 2.2567, "step": 47340 }, { "epoch": 1.6427831317500363e-13, "grad_norm": 0.6723567247390747, "learning_rate": 5.361607858301587e-05, "loss": 2.2995, "step": 47350 }, { "epoch": 1.6431300764452317e-13, "grad_norm": 0.6349596381187439, "learning_rate": 5.3594535656927236e-05, "loss": 2.2615, "step": 47360 }, { "epoch": 1.643477021140427e-13, "grad_norm": 0.6479366421699524, "learning_rate": 5.357299512754482e-05, "loss": 2.243, "step": 47370 }, { "epoch": 1.6438239658356224e-13, "grad_norm": 0.6193259954452515, "learning_rate": 5.3551456998275075e-05, "loss": 2.2744, "step": 47380 }, { "epoch": 1.6441709105308178e-13, "grad_norm": 0.6439872980117798, "learning_rate": 5.35299212725241e-05, "loss": 2.2734, "step": 47390 }, { "epoch": 1.644517855226013e-13, "grad_norm": 0.6389901041984558, "learning_rate": 5.3508387953697545e-05, "loss": 2.2446, "step": 47400 }, { "epoch": 1.6448647999212085e-13, "grad_norm": 0.6522786021232605, "learning_rate": 5.348685704520075e-05, "loss": 2.243, "step": 47410 }, { "epoch": 1.6452117446164038e-13, "grad_norm": 0.6444287300109863, "learning_rate": 5.34653285504386e-05, "loss": 2.2792, "step": 47420 }, { "epoch": 1.6455586893115992e-13, "grad_norm": 0.6371967792510986, "learning_rate": 5.344380247281567e-05, "loss": 2.25, "step": 47430 }, { "epoch": 1.6459056340067946e-13, "grad_norm": 0.6474576592445374, "learning_rate": 5.3422278815736135e-05, "loss": 2.286, "step": 47440 }, { "epoch": 1.64625257870199e-13, "grad_norm": 0.6618943810462952, "learning_rate": 5.3400757582603763e-05, "loss": 2.2563, "step": 47450 }, { "epoch": 1.6465995233971853e-13, "grad_norm": 0.6451180577278137, "learning_rate": 5.337923877682196e-05, "loss": 2.2662, "step": 47460 }, { "epoch": 1.6469464680923807e-13, "grad_norm": 0.6412723064422607, "learning_rate": 5.3357722401793723e-05, "loss": 2.2309, "step": 47470 }, { "epoch": 1.647293412787576e-13, "grad_norm": 0.6543615460395813, "learning_rate": 5.3336208460921673e-05, "loss": 2.2727, "step": 47480 }, { "epoch": 1.6476403574827714e-13, "grad_norm": 0.6361386179924011, "learning_rate": 5.331469695760811e-05, "loss": 2.2634, "step": 47490 }, { "epoch": 1.6479873021779667e-13, "grad_norm": 0.6536937355995178, "learning_rate": 5.329318789525482e-05, "loss": 2.2857, "step": 47500 }, { "epoch": 1.648334246873162e-13, "grad_norm": 0.65840083360672, "learning_rate": 5.327168127726334e-05, "loss": 2.2812, "step": 47510 }, { "epoch": 1.6486811915683575e-13, "grad_norm": 0.6637308597564697, "learning_rate": 5.325017710703471e-05, "loss": 2.2878, "step": 47520 }, { "epoch": 1.6490281362635528e-13, "grad_norm": 0.677973210811615, "learning_rate": 5.3228675387969644e-05, "loss": 2.2458, "step": 47530 }, { "epoch": 1.6493750809587482e-13, "grad_norm": 0.6441047191619873, "learning_rate": 5.3207176123468464e-05, "loss": 2.2613, "step": 47540 }, { "epoch": 1.6497220256539435e-13, "grad_norm": 0.67959064245224, "learning_rate": 5.3185679316931076e-05, "loss": 2.2372, "step": 47550 }, { "epoch": 1.650068970349139e-13, "grad_norm": 0.638171374797821, "learning_rate": 5.316418497175707e-05, "loss": 2.2707, "step": 47560 }, { "epoch": 1.6504159150443343e-13, "grad_norm": 0.650738000869751, "learning_rate": 5.314269309134549e-05, "loss": 2.2551, "step": 47570 }, { "epoch": 1.6507628597395296e-13, "grad_norm": 0.6525148749351501, "learning_rate": 5.3121203679095205e-05, "loss": 2.2456, "step": 47580 }, { "epoch": 1.651109804434725e-13, "grad_norm": 0.6548921465873718, "learning_rate": 5.309971673840448e-05, "loss": 2.2538, "step": 47590 }, { "epoch": 1.6514567491299204e-13, "grad_norm": 0.6702997088432312, "learning_rate": 5.3078232272671334e-05, "loss": 2.2616, "step": 47600 }, { "epoch": 1.6518036938251157e-13, "grad_norm": 0.6823610663414001, "learning_rate": 5.305675028529336e-05, "loss": 2.2849, "step": 47610 }, { "epoch": 1.652150638520311e-13, "grad_norm": 0.6303390264511108, "learning_rate": 5.303527077966774e-05, "loss": 2.266, "step": 47620 }, { "epoch": 1.6524975832155064e-13, "grad_norm": 0.6199856996536255, "learning_rate": 5.3013793759191285e-05, "loss": 2.2515, "step": 47630 }, { "epoch": 1.6528445279107018e-13, "grad_norm": 0.6594345569610596, "learning_rate": 5.299231922726037e-05, "loss": 2.2391, "step": 47640 }, { "epoch": 1.6531914726058972e-13, "grad_norm": 0.6606456637382507, "learning_rate": 5.2970847187271036e-05, "loss": 2.2679, "step": 47650 }, { "epoch": 1.6535384173010925e-13, "grad_norm": 0.6935097575187683, "learning_rate": 5.294937764261891e-05, "loss": 2.2248, "step": 47660 }, { "epoch": 1.653885361996288e-13, "grad_norm": 0.6607401371002197, "learning_rate": 5.2927910596699156e-05, "loss": 2.2341, "step": 47670 }, { "epoch": 1.6542323066914832e-13, "grad_norm": 0.6186218857765198, "learning_rate": 5.290644605290668e-05, "loss": 2.2534, "step": 47680 }, { "epoch": 1.6545792513866786e-13, "grad_norm": 0.6346797943115234, "learning_rate": 5.288498401463586e-05, "loss": 2.2484, "step": 47690 }, { "epoch": 1.654926196081874e-13, "grad_norm": 0.6608079671859741, "learning_rate": 5.2863524485280726e-05, "loss": 2.2796, "step": 47700 }, { "epoch": 1.6552731407770693e-13, "grad_norm": 0.684888482093811, "learning_rate": 5.2842067468235e-05, "loss": 2.2883, "step": 47710 }, { "epoch": 1.6556200854722647e-13, "grad_norm": 0.6608452200889587, "learning_rate": 5.2820612966891836e-05, "loss": 2.222, "step": 47720 }, { "epoch": 1.65596703016746e-13, "grad_norm": 0.6515388488769531, "learning_rate": 5.279916098464413e-05, "loss": 2.2398, "step": 47730 }, { "epoch": 1.6563139748626554e-13, "grad_norm": 0.6929025650024414, "learning_rate": 5.277771152488428e-05, "loss": 2.2246, "step": 47740 }, { "epoch": 1.6566609195578508e-13, "grad_norm": 0.6548754572868347, "learning_rate": 5.2756264591004354e-05, "loss": 2.267, "step": 47750 }, { "epoch": 1.6570078642530461e-13, "grad_norm": 0.6552505493164062, "learning_rate": 5.273482018639604e-05, "loss": 2.2745, "step": 47760 }, { "epoch": 1.6573548089482415e-13, "grad_norm": 0.6453737616539001, "learning_rate": 5.271337831445054e-05, "loss": 2.2856, "step": 47770 }, { "epoch": 1.6577017536434369e-13, "grad_norm": 0.636993944644928, "learning_rate": 5.2691938978558726e-05, "loss": 2.2188, "step": 47780 }, { "epoch": 1.6580486983386322e-13, "grad_norm": 0.673159658908844, "learning_rate": 5.2670502182111015e-05, "loss": 2.2458, "step": 47790 }, { "epoch": 1.6583956430338276e-13, "grad_norm": 0.6346321105957031, "learning_rate": 5.264906792849748e-05, "loss": 2.2774, "step": 47800 }, { "epoch": 1.658742587729023e-13, "grad_norm": 0.6660905480384827, "learning_rate": 5.262763622110776e-05, "loss": 2.2761, "step": 47810 }, { "epoch": 1.6590895324242183e-13, "grad_norm": 0.6557947993278503, "learning_rate": 5.260620706333108e-05, "loss": 2.2881, "step": 47820 }, { "epoch": 1.6594364771194137e-13, "grad_norm": 0.6542929410934448, "learning_rate": 5.2584780458556286e-05, "loss": 2.2592, "step": 47830 }, { "epoch": 1.659783421814609e-13, "grad_norm": 0.65299391746521, "learning_rate": 5.256335641017182e-05, "loss": 2.2545, "step": 47840 }, { "epoch": 1.6601303665098044e-13, "grad_norm": 0.6248157620429993, "learning_rate": 5.25419349215657e-05, "loss": 2.282, "step": 47850 }, { "epoch": 1.6604773112049998e-13, "grad_norm": 0.6384463906288147, "learning_rate": 5.252051599612556e-05, "loss": 2.2751, "step": 47860 }, { "epoch": 1.660824255900195e-13, "grad_norm": 0.6442371606826782, "learning_rate": 5.249909963723859e-05, "loss": 2.2708, "step": 47870 }, { "epoch": 1.6611712005953905e-13, "grad_norm": 0.6648536324501038, "learning_rate": 5.2477685848291645e-05, "loss": 2.2398, "step": 47880 }, { "epoch": 1.6615181452905858e-13, "grad_norm": 0.6637696027755737, "learning_rate": 5.24562746326711e-05, "loss": 2.2961, "step": 47890 }, { "epoch": 1.6618650899857812e-13, "grad_norm": 0.6377448439598083, "learning_rate": 5.2434865993762946e-05, "loss": 2.2525, "step": 47900 }, { "epoch": 1.6622120346809766e-13, "grad_norm": 0.662497878074646, "learning_rate": 5.241345993495282e-05, "loss": 2.2622, "step": 47910 }, { "epoch": 1.662558979376172e-13, "grad_norm": 0.6302546858787537, "learning_rate": 5.239205645962587e-05, "loss": 2.2548, "step": 47920 }, { "epoch": 1.6629059240713673e-13, "grad_norm": 0.7400845289230347, "learning_rate": 5.237065557116688e-05, "loss": 2.2567, "step": 47930 }, { "epoch": 1.6632528687665626e-13, "grad_norm": 0.6370478272438049, "learning_rate": 5.234925727296022e-05, "loss": 2.2748, "step": 47940 }, { "epoch": 1.663599813461758e-13, "grad_norm": 0.6797226667404175, "learning_rate": 5.2327861568389856e-05, "loss": 2.2452, "step": 47950 }, { "epoch": 1.6639467581569534e-13, "grad_norm": 0.6468557119369507, "learning_rate": 5.230646846083929e-05, "loss": 2.2363, "step": 47960 }, { "epoch": 1.6642937028521487e-13, "grad_norm": 0.6482199430465698, "learning_rate": 5.228507795369171e-05, "loss": 2.2381, "step": 47970 }, { "epoch": 1.664640647547344e-13, "grad_norm": 0.659304141998291, "learning_rate": 5.226369005032984e-05, "loss": 2.2691, "step": 47980 }, { "epoch": 1.6649875922425394e-13, "grad_norm": 0.6557511687278748, "learning_rate": 5.224230475413594e-05, "loss": 2.294, "step": 47990 }, { "epoch": 1.6653345369377348e-13, "grad_norm": 0.6313788890838623, "learning_rate": 5.2220922068491965e-05, "loss": 2.262, "step": 48000 }, { "epoch": 1.6656814816329302e-13, "grad_norm": 0.6565086245536804, "learning_rate": 5.219954199677936e-05, "loss": 2.2642, "step": 48010 }, { "epoch": 1.6660284263281255e-13, "grad_norm": 0.6585000157356262, "learning_rate": 5.217816454237923e-05, "loss": 2.2532, "step": 48020 }, { "epoch": 1.666375371023321e-13, "grad_norm": 0.6537624001502991, "learning_rate": 5.215678970867224e-05, "loss": 2.2129, "step": 48030 }, { "epoch": 1.6667223157185163e-13, "grad_norm": 0.6525982022285461, "learning_rate": 5.2135417499038604e-05, "loss": 2.2292, "step": 48040 }, { "epoch": 1.6670692604137116e-13, "grad_norm": 0.6579944491386414, "learning_rate": 5.21140479168582e-05, "loss": 2.2321, "step": 48050 }, { "epoch": 1.667416205108907e-13, "grad_norm": 0.6645757555961609, "learning_rate": 5.209268096551038e-05, "loss": 2.2437, "step": 48060 }, { "epoch": 1.6677631498041023e-13, "grad_norm": 0.6423951983451843, "learning_rate": 5.20713166483742e-05, "loss": 2.2716, "step": 48070 }, { "epoch": 1.6681100944992977e-13, "grad_norm": 0.6562032699584961, "learning_rate": 5.204995496882821e-05, "loss": 2.2391, "step": 48080 }, { "epoch": 1.668457039194493e-13, "grad_norm": 0.66448575258255, "learning_rate": 5.2028595930250585e-05, "loss": 2.2556, "step": 48090 }, { "epoch": 1.6688039838896884e-13, "grad_norm": 0.6194359064102173, "learning_rate": 5.200723953601909e-05, "loss": 2.2672, "step": 48100 }, { "epoch": 1.6691509285848838e-13, "grad_norm": 0.7086573243141174, "learning_rate": 5.198588578951102e-05, "loss": 2.258, "step": 48110 }, { "epoch": 1.6694978732800791e-13, "grad_norm": 0.6715900301933289, "learning_rate": 5.196453469410331e-05, "loss": 2.2464, "step": 48120 }, { "epoch": 1.6698448179752745e-13, "grad_norm": 0.6471863389015198, "learning_rate": 5.194318625317247e-05, "loss": 2.2415, "step": 48130 }, { "epoch": 1.6701917626704699e-13, "grad_norm": 0.6559344530105591, "learning_rate": 5.1921840470094515e-05, "loss": 2.2393, "step": 48140 }, { "epoch": 1.6705387073656652e-13, "grad_norm": 0.6764195561408997, "learning_rate": 5.190049734824516e-05, "loss": 2.2534, "step": 48150 }, { "epoch": 1.6708856520608606e-13, "grad_norm": 0.6709803938865662, "learning_rate": 5.187915689099959e-05, "loss": 2.2538, "step": 48160 }, { "epoch": 1.671232596756056e-13, "grad_norm": 0.6308369636535645, "learning_rate": 5.185781910173264e-05, "loss": 2.2419, "step": 48170 }, { "epoch": 1.6715795414512513e-13, "grad_norm": 0.6561791896820068, "learning_rate": 5.183648398381871e-05, "loss": 2.2487, "step": 48180 }, { "epoch": 1.6719264861464467e-13, "grad_norm": 0.6220047473907471, "learning_rate": 5.181515154063171e-05, "loss": 2.2281, "step": 48190 }, { "epoch": 1.672273430841642e-13, "grad_norm": 0.6670177578926086, "learning_rate": 5.1793821775545246e-05, "loss": 2.2437, "step": 48200 }, { "epoch": 1.6726203755368374e-13, "grad_norm": 0.6621571779251099, "learning_rate": 5.177249469193239e-05, "loss": 2.2431, "step": 48210 }, { "epoch": 1.6729673202320328e-13, "grad_norm": 0.6526253819465637, "learning_rate": 5.175117029316585e-05, "loss": 2.2353, "step": 48220 }, { "epoch": 1.673314264927228e-13, "grad_norm": 0.6406999826431274, "learning_rate": 5.172984858261792e-05, "loss": 2.2319, "step": 48230 }, { "epoch": 1.6736612096224235e-13, "grad_norm": 0.6268854737281799, "learning_rate": 5.170852956366043e-05, "loss": 2.2732, "step": 48240 }, { "epoch": 1.6740081543176188e-13, "grad_norm": 0.6582944989204407, "learning_rate": 5.1687213239664785e-05, "loss": 2.2755, "step": 48250 }, { "epoch": 1.6743550990128142e-13, "grad_norm": 0.6515049934387207, "learning_rate": 5.166589961400198e-05, "loss": 2.2217, "step": 48260 }, { "epoch": 1.6747020437080096e-13, "grad_norm": 0.6678375601768494, "learning_rate": 5.164458869004261e-05, "loss": 2.2141, "step": 48270 }, { "epoch": 1.675048988403205e-13, "grad_norm": 0.6477675437927246, "learning_rate": 5.1623280471156774e-05, "loss": 2.2792, "step": 48280 }, { "epoch": 1.6753959330984003e-13, "grad_norm": 0.6527089476585388, "learning_rate": 5.16019749607142e-05, "loss": 2.2443, "step": 48290 }, { "epoch": 1.6757428777935957e-13, "grad_norm": 0.6976597905158997, "learning_rate": 5.15806721620842e-05, "loss": 2.2368, "step": 48300 }, { "epoch": 1.676089822488791e-13, "grad_norm": 0.681282639503479, "learning_rate": 5.1559372078635594e-05, "loss": 2.2583, "step": 48310 }, { "epoch": 1.6764367671839864e-13, "grad_norm": 0.9245108962059021, "learning_rate": 5.153807471373681e-05, "loss": 2.2855, "step": 48320 }, { "epoch": 1.6767837118791817e-13, "grad_norm": 0.6911811232566833, "learning_rate": 5.151678007075585e-05, "loss": 2.2579, "step": 48330 }, { "epoch": 1.677130656574377e-13, "grad_norm": 0.683565080165863, "learning_rate": 5.1495488153060265e-05, "loss": 2.2364, "step": 48340 }, { "epoch": 1.6774776012695725e-13, "grad_norm": 0.6613283753395081, "learning_rate": 5.147419896401723e-05, "loss": 2.2475, "step": 48350 }, { "epoch": 1.6778245459647678e-13, "grad_norm": 0.654361367225647, "learning_rate": 5.1452912506993396e-05, "loss": 2.2559, "step": 48360 }, { "epoch": 1.6781714906599632e-13, "grad_norm": 0.624901294708252, "learning_rate": 5.1431628785355065e-05, "loss": 2.2416, "step": 48370 }, { "epoch": 1.6785184353551585e-13, "grad_norm": 0.6792962551116943, "learning_rate": 5.1410347802468064e-05, "loss": 2.2641, "step": 48380 }, { "epoch": 1.678865380050354e-13, "grad_norm": 0.6980791687965393, "learning_rate": 5.1389069561697784e-05, "loss": 2.2398, "step": 48390 }, { "epoch": 1.6792123247455493e-13, "grad_norm": 0.6462530493736267, "learning_rate": 5.136779406640922e-05, "loss": 2.2275, "step": 48400 }, { "epoch": 1.6795592694407446e-13, "grad_norm": 0.6456608176231384, "learning_rate": 5.134652131996688e-05, "loss": 2.2491, "step": 48410 }, { "epoch": 1.67990621413594e-13, "grad_norm": 0.6401094198226929, "learning_rate": 5.1325251325734916e-05, "loss": 2.2251, "step": 48420 }, { "epoch": 1.6802531588311354e-13, "grad_norm": 0.6545324921607971, "learning_rate": 5.130398408707693e-05, "loss": 2.2527, "step": 48430 }, { "epoch": 1.6806001035263307e-13, "grad_norm": 0.631479024887085, "learning_rate": 5.128271960735619e-05, "loss": 2.272, "step": 48440 }, { "epoch": 1.680947048221526e-13, "grad_norm": 0.6307965517044067, "learning_rate": 5.126145788993551e-05, "loss": 2.2754, "step": 48450 }, { "epoch": 1.6812939929167214e-13, "grad_norm": 0.6371705532073975, "learning_rate": 5.124019893817719e-05, "loss": 2.2745, "step": 48460 }, { "epoch": 1.6816409376119168e-13, "grad_norm": 0.6564168334007263, "learning_rate": 5.12189427554432e-05, "loss": 2.278, "step": 48470 }, { "epoch": 1.6819878823071122e-13, "grad_norm": 0.6617359519004822, "learning_rate": 5.1197689345095e-05, "loss": 2.2305, "step": 48480 }, { "epoch": 1.6823348270023075e-13, "grad_norm": 0.6606935262680054, "learning_rate": 5.1176438710493626e-05, "loss": 2.2466, "step": 48490 }, { "epoch": 1.682681771697503e-13, "grad_norm": 0.6548057198524475, "learning_rate": 5.1155190854999726e-05, "loss": 2.2428, "step": 48500 }, { "epoch": 1.6830287163926982e-13, "grad_norm": 0.6352211236953735, "learning_rate": 5.1133945781973415e-05, "loss": 2.2428, "step": 48510 }, { "epoch": 1.6833756610878936e-13, "grad_norm": 0.6685473322868347, "learning_rate": 5.111270349477445e-05, "loss": 2.229, "step": 48520 }, { "epoch": 1.683722605783089e-13, "grad_norm": 0.6660799980163574, "learning_rate": 5.1091463996762096e-05, "loss": 2.2415, "step": 48530 }, { "epoch": 1.6840695504782843e-13, "grad_norm": 0.6511508226394653, "learning_rate": 5.1070227291295205e-05, "loss": 2.2711, "step": 48540 }, { "epoch": 1.6844164951734797e-13, "grad_norm": 0.6200708150863647, "learning_rate": 5.1048993381732175e-05, "loss": 2.2473, "step": 48550 }, { "epoch": 1.684763439868675e-13, "grad_norm": 0.6668808460235596, "learning_rate": 5.102776227143098e-05, "loss": 2.2343, "step": 48560 }, { "epoch": 1.6851103845638704e-13, "grad_norm": 0.6324388980865479, "learning_rate": 5.100653396374913e-05, "loss": 2.2232, "step": 48570 }, { "epoch": 1.6854573292590658e-13, "grad_norm": 0.6485615968704224, "learning_rate": 5.098530846204369e-05, "loss": 2.2362, "step": 48580 }, { "epoch": 1.6858042739542611e-13, "grad_norm": 0.6843249797821045, "learning_rate": 5.096408576967131e-05, "loss": 2.2657, "step": 48590 }, { "epoch": 1.6861512186494565e-13, "grad_norm": 0.6408401131629944, "learning_rate": 5.0942865889988144e-05, "loss": 2.2578, "step": 48600 }, { "epoch": 1.6864981633446519e-13, "grad_norm": 0.6869108080863953, "learning_rate": 5.092164882634995e-05, "loss": 2.2533, "step": 48610 }, { "epoch": 1.6868451080398472e-13, "grad_norm": 0.6572664976119995, "learning_rate": 5.090043458211205e-05, "loss": 2.2401, "step": 48620 }, { "epoch": 1.6871920527350426e-13, "grad_norm": 0.6469053030014038, "learning_rate": 5.087922316062925e-05, "loss": 2.2567, "step": 48630 }, { "epoch": 1.687538997430238e-13, "grad_norm": 0.6384056210517883, "learning_rate": 5.085801456525599e-05, "loss": 2.2538, "step": 48640 }, { "epoch": 1.6878859421254333e-13, "grad_norm": 0.6707901358604431, "learning_rate": 5.083680879934619e-05, "loss": 2.242, "step": 48650 }, { "epoch": 1.6882328868206287e-13, "grad_norm": 0.6522870063781738, "learning_rate": 5.081560586625339e-05, "loss": 2.2307, "step": 48660 }, { "epoch": 1.688579831515824e-13, "grad_norm": 0.7150393724441528, "learning_rate": 5.0794405769330654e-05, "loss": 2.2521, "step": 48670 }, { "epoch": 1.6889267762110194e-13, "grad_norm": 0.6612302660942078, "learning_rate": 5.077320851193057e-05, "loss": 2.2964, "step": 48680 }, { "epoch": 1.6892737209062147e-13, "grad_norm": 0.6326499581336975, "learning_rate": 5.0752014097405336e-05, "loss": 2.2201, "step": 48690 }, { "epoch": 1.68962066560141e-13, "grad_norm": 0.655125081539154, "learning_rate": 5.073082252910662e-05, "loss": 2.257, "step": 48700 }, { "epoch": 1.6899676102966055e-13, "grad_norm": 0.6416597366333008, "learning_rate": 5.070963381038571e-05, "loss": 2.2469, "step": 48710 }, { "epoch": 1.6903145549918008e-13, "grad_norm": 0.6726952195167542, "learning_rate": 5.068844794459343e-05, "loss": 2.2231, "step": 48720 }, { "epoch": 1.6906614996869962e-13, "grad_norm": 0.6810444593429565, "learning_rate": 5.066726493508013e-05, "loss": 2.2094, "step": 48730 }, { "epoch": 1.6910084443821916e-13, "grad_norm": 0.6443086266517639, "learning_rate": 5.064608478519574e-05, "loss": 2.2066, "step": 48740 }, { "epoch": 1.691355389077387e-13, "grad_norm": 0.6560356616973877, "learning_rate": 5.062490749828967e-05, "loss": 2.2395, "step": 48750 }, { "epoch": 1.6917023337725823e-13, "grad_norm": 0.6586894392967224, "learning_rate": 5.0603733077710965e-05, "loss": 2.2552, "step": 48760 }, { "epoch": 1.6920492784677776e-13, "grad_norm": 0.6515486240386963, "learning_rate": 5.058256152680819e-05, "loss": 2.2582, "step": 48770 }, { "epoch": 1.692396223162973e-13, "grad_norm": 0.6909579634666443, "learning_rate": 5.056139284892941e-05, "loss": 2.2574, "step": 48780 }, { "epoch": 1.6927431678581684e-13, "grad_norm": 0.6741933226585388, "learning_rate": 5.054022704742228e-05, "loss": 2.2727, "step": 48790 }, { "epoch": 1.6930901125533637e-13, "grad_norm": 0.6525344848632812, "learning_rate": 5.051906412563399e-05, "loss": 2.2564, "step": 48800 }, { "epoch": 1.693437057248559e-13, "grad_norm": 0.6499718427658081, "learning_rate": 5.049790408691128e-05, "loss": 2.2369, "step": 48810 }, { "epoch": 1.6937840019437544e-13, "grad_norm": 0.646273136138916, "learning_rate": 5.047674693460044e-05, "loss": 2.2421, "step": 48820 }, { "epoch": 1.6941309466389498e-13, "grad_norm": 0.6303622126579285, "learning_rate": 5.0455592672047247e-05, "loss": 2.2622, "step": 48830 }, { "epoch": 1.6944778913341452e-13, "grad_norm": 0.6321339011192322, "learning_rate": 5.043444130259711e-05, "loss": 2.2677, "step": 48840 }, { "epoch": 1.6948248360293405e-13, "grad_norm": 0.6630828380584717, "learning_rate": 5.041329282959492e-05, "loss": 2.2645, "step": 48850 }, { "epoch": 1.695171780724536e-13, "grad_norm": 0.6994386315345764, "learning_rate": 5.0392147256385106e-05, "loss": 2.251, "step": 48860 }, { "epoch": 1.6955187254197313e-13, "grad_norm": 0.6407799124717712, "learning_rate": 5.037100458631171e-05, "loss": 2.2411, "step": 48870 }, { "epoch": 1.6958656701149266e-13, "grad_norm": 0.6702898740768433, "learning_rate": 5.0349864822718194e-05, "loss": 2.2556, "step": 48880 }, { "epoch": 1.696212614810122e-13, "grad_norm": 0.6476818919181824, "learning_rate": 5.0328727968947706e-05, "loss": 2.2346, "step": 48890 }, { "epoch": 1.6965595595053173e-13, "grad_norm": 0.6417297720909119, "learning_rate": 5.030759402834278e-05, "loss": 2.2507, "step": 48900 }, { "epoch": 1.6969065042005127e-13, "grad_norm": 0.6783440113067627, "learning_rate": 5.028646300424565e-05, "loss": 2.2572, "step": 48910 }, { "epoch": 1.697253448895708e-13, "grad_norm": 0.655590832233429, "learning_rate": 5.026533489999793e-05, "loss": 2.2518, "step": 48920 }, { "epoch": 1.6976003935909034e-13, "grad_norm": 0.639215886592865, "learning_rate": 5.0244209718940865e-05, "loss": 2.2491, "step": 48930 }, { "epoch": 1.6979473382860988e-13, "grad_norm": 0.6514368653297424, "learning_rate": 5.022308746441527e-05, "loss": 2.2306, "step": 48940 }, { "epoch": 1.6982942829812941e-13, "grad_norm": 0.7151576280593872, "learning_rate": 5.020196813976138e-05, "loss": 2.2532, "step": 48950 }, { "epoch": 1.6986412276764895e-13, "grad_norm": 0.6934369802474976, "learning_rate": 5.01808517483191e-05, "loss": 2.261, "step": 48960 }, { "epoch": 1.6989881723716849e-13, "grad_norm": 0.6428804993629456, "learning_rate": 5.0159738293427735e-05, "loss": 2.2408, "step": 48970 }, { "epoch": 1.6993351170668802e-13, "grad_norm": 0.7013445496559143, "learning_rate": 5.013862777842625e-05, "loss": 2.2647, "step": 48980 }, { "epoch": 1.6996820617620756e-13, "grad_norm": 0.6338802576065063, "learning_rate": 5.0117520206653085e-05, "loss": 2.2455, "step": 48990 }, { "epoch": 1.700029006457271e-13, "grad_norm": 0.658200740814209, "learning_rate": 5.009641558144619e-05, "loss": 2.1972, "step": 49000 }, { "epoch": 1.7003759511524663e-13, "grad_norm": 0.6371662616729736, "learning_rate": 5.007531390614313e-05, "loss": 2.2499, "step": 49010 }, { "epoch": 1.7007228958476617e-13, "grad_norm": 0.6597043871879578, "learning_rate": 5.005421518408089e-05, "loss": 2.2337, "step": 49020 }, { "epoch": 1.701069840542857e-13, "grad_norm": 0.6634889245033264, "learning_rate": 5.003311941859609e-05, "loss": 2.2598, "step": 49030 }, { "epoch": 1.7014167852380524e-13, "grad_norm": 0.6260908246040344, "learning_rate": 5.001202661302484e-05, "loss": 2.2381, "step": 49040 }, { "epoch": 1.7017637299332478e-13, "grad_norm": 0.6611477732658386, "learning_rate": 4.999093677070278e-05, "loss": 2.242, "step": 49050 }, { "epoch": 1.702110674628443e-13, "grad_norm": 0.652044951915741, "learning_rate": 4.9969849894965117e-05, "loss": 2.2249, "step": 49060 }, { "epoch": 1.7024576193236385e-13, "grad_norm": 0.6660426259040833, "learning_rate": 4.9948765989146505e-05, "loss": 2.2546, "step": 49070 }, { "epoch": 1.7028045640188338e-13, "grad_norm": 0.682984471321106, "learning_rate": 4.992768505658121e-05, "loss": 2.247, "step": 49080 }, { "epoch": 1.7031515087140292e-13, "grad_norm": 0.6513186693191528, "learning_rate": 4.990660710060302e-05, "loss": 2.2454, "step": 49090 }, { "epoch": 1.7034984534092246e-13, "grad_norm": 0.6725996136665344, "learning_rate": 4.988553212454519e-05, "loss": 2.2603, "step": 49100 }, { "epoch": 1.70384539810442e-13, "grad_norm": 0.6651396751403809, "learning_rate": 4.9864460131740593e-05, "loss": 2.2514, "step": 49110 }, { "epoch": 1.7041923427996153e-13, "grad_norm": 0.6416012644767761, "learning_rate": 4.984339112552153e-05, "loss": 2.2579, "step": 49120 }, { "epoch": 1.7045392874948107e-13, "grad_norm": 0.6412383913993835, "learning_rate": 4.982232510921991e-05, "loss": 2.2606, "step": 49130 }, { "epoch": 1.704886232190006e-13, "grad_norm": 0.6706433296203613, "learning_rate": 4.980126208616717e-05, "loss": 2.261, "step": 49140 }, { "epoch": 1.7052331768852014e-13, "grad_norm": 0.6513493657112122, "learning_rate": 4.978020205969421e-05, "loss": 2.244, "step": 49150 }, { "epoch": 1.7055801215803967e-13, "grad_norm": 0.6345247626304626, "learning_rate": 4.97591450331315e-05, "loss": 2.2294, "step": 49160 }, { "epoch": 1.705927066275592e-13, "grad_norm": 0.6325739026069641, "learning_rate": 4.973809100980903e-05, "loss": 2.2081, "step": 49170 }, { "epoch": 1.7062740109707875e-13, "grad_norm": 0.6263296008110046, "learning_rate": 4.9717039993056296e-05, "loss": 2.2425, "step": 49180 }, { "epoch": 1.7066209556659828e-13, "grad_norm": 0.6507211923599243, "learning_rate": 4.9695991986202367e-05, "loss": 2.237, "step": 49190 }, { "epoch": 1.7069679003611782e-13, "grad_norm": 0.6763225197792053, "learning_rate": 4.967494699257577e-05, "loss": 2.2335, "step": 49200 }, { "epoch": 1.7073148450563735e-13, "grad_norm": 0.6716728806495667, "learning_rate": 4.9653905015504636e-05, "loss": 2.2773, "step": 49210 }, { "epoch": 1.707661789751569e-13, "grad_norm": 0.6629450917243958, "learning_rate": 4.9632866058316515e-05, "loss": 2.2404, "step": 49220 }, { "epoch": 1.7080087344467643e-13, "grad_norm": 0.6656882166862488, "learning_rate": 4.96118301243386e-05, "loss": 2.2272, "step": 49230 }, { "epoch": 1.7083556791419596e-13, "grad_norm": 0.6469679474830627, "learning_rate": 4.959079721689748e-05, "loss": 2.2092, "step": 49240 }, { "epoch": 1.708702623837155e-13, "grad_norm": 0.6292406320571899, "learning_rate": 4.956976733931935e-05, "loss": 2.2534, "step": 49250 }, { "epoch": 1.7090495685323503e-13, "grad_norm": 0.6580608487129211, "learning_rate": 4.954874049492993e-05, "loss": 2.2866, "step": 49260 }, { "epoch": 1.7093965132275457e-13, "grad_norm": 0.6470617651939392, "learning_rate": 4.95277166870544e-05, "loss": 2.2635, "step": 49270 }, { "epoch": 1.709743457922741e-13, "grad_norm": 0.6628009676933289, "learning_rate": 4.950669591901753e-05, "loss": 2.2529, "step": 49280 }, { "epoch": 1.7100904026179364e-13, "grad_norm": 0.6749653220176697, "learning_rate": 4.948567819414353e-05, "loss": 2.2421, "step": 49290 }, { "epoch": 1.7104373473131318e-13, "grad_norm": 0.6496937274932861, "learning_rate": 4.946466351575619e-05, "loss": 2.2369, "step": 49300 }, { "epoch": 1.7107842920083272e-13, "grad_norm": 0.6534180641174316, "learning_rate": 4.9443651887178834e-05, "loss": 2.2245, "step": 49310 }, { "epoch": 1.7111312367035225e-13, "grad_norm": 0.6378750801086426, "learning_rate": 4.942264331173423e-05, "loss": 2.2588, "step": 49320 }, { "epoch": 1.711478181398718e-13, "grad_norm": 0.6479583978652954, "learning_rate": 4.9401637792744714e-05, "loss": 2.2258, "step": 49330 }, { "epoch": 1.7118251260939132e-13, "grad_norm": 0.6680727005004883, "learning_rate": 4.938063533353214e-05, "loss": 2.2448, "step": 49340 }, { "epoch": 1.7121720707891086e-13, "grad_norm": 0.6476069688796997, "learning_rate": 4.9359635937417824e-05, "loss": 2.2754, "step": 49350 }, { "epoch": 1.712519015484304e-13, "grad_norm": 0.6912627220153809, "learning_rate": 4.9338639607722696e-05, "loss": 2.2555, "step": 49360 }, { "epoch": 1.7128659601794993e-13, "grad_norm": 0.6628328561782837, "learning_rate": 4.931764634776711e-05, "loss": 2.198, "step": 49370 }, { "epoch": 1.7132129048746947e-13, "grad_norm": 0.6807160973548889, "learning_rate": 4.929665616087101e-05, "loss": 2.2062, "step": 49380 }, { "epoch": 1.71355984956989e-13, "grad_norm": 0.6794020533561707, "learning_rate": 4.9275669050353746e-05, "loss": 2.2645, "step": 49390 }, { "epoch": 1.7139067942650854e-13, "grad_norm": 0.6483139395713806, "learning_rate": 4.925468501953429e-05, "loss": 2.2128, "step": 49400 }, { "epoch": 1.7142537389602808e-13, "grad_norm": 0.6863059401512146, "learning_rate": 4.9233704071731114e-05, "loss": 2.2325, "step": 49410 }, { "epoch": 1.7146006836554761e-13, "grad_norm": 0.6680251955986023, "learning_rate": 4.92127262102621e-05, "loss": 2.2526, "step": 49420 }, { "epoch": 1.7149476283506715e-13, "grad_norm": 0.6643838286399841, "learning_rate": 4.919175143844477e-05, "loss": 2.2341, "step": 49430 }, { "epoch": 1.7152945730458669e-13, "grad_norm": 0.6922069787979126, "learning_rate": 4.917077975959609e-05, "loss": 2.2414, "step": 49440 }, { "epoch": 1.7156415177410622e-13, "grad_norm": 0.6687880754470825, "learning_rate": 4.914981117703255e-05, "loss": 2.2479, "step": 49450 }, { "epoch": 1.7159884624362576e-13, "grad_norm": 0.6596695780754089, "learning_rate": 4.912884569407018e-05, "loss": 2.2781, "step": 49460 }, { "epoch": 1.716335407131453e-13, "grad_norm": 0.6663424968719482, "learning_rate": 4.910788331402444e-05, "loss": 2.2717, "step": 49470 }, { "epoch": 1.7166823518266483e-13, "grad_norm": 0.6553273797035217, "learning_rate": 4.908692404021038e-05, "loss": 2.241, "step": 49480 }, { "epoch": 1.7170292965218437e-13, "grad_norm": 0.6284639835357666, "learning_rate": 4.906596787594252e-05, "loss": 2.2097, "step": 49490 }, { "epoch": 1.717376241217039e-13, "grad_norm": 0.6397973299026489, "learning_rate": 4.9045014824534885e-05, "loss": 2.2371, "step": 49500 }, { "epoch": 1.7177231859122344e-13, "grad_norm": 0.6709812879562378, "learning_rate": 4.9024064889301055e-05, "loss": 2.2364, "step": 49510 }, { "epoch": 1.7180701306074297e-13, "grad_norm": 0.6602444052696228, "learning_rate": 4.900311807355404e-05, "loss": 2.247, "step": 49520 }, { "epoch": 1.718417075302625e-13, "grad_norm": 0.6867843270301819, "learning_rate": 4.898217438060645e-05, "loss": 2.239, "step": 49530 }, { "epoch": 1.7187640199978205e-13, "grad_norm": 0.6686187982559204, "learning_rate": 4.8961233813770294e-05, "loss": 2.2701, "step": 49540 }, { "epoch": 1.7191109646930158e-13, "grad_norm": 0.6483383178710938, "learning_rate": 4.894029637635716e-05, "loss": 2.2314, "step": 49550 }, { "epoch": 1.7194579093882112e-13, "grad_norm": 0.671918511390686, "learning_rate": 4.891936207167816e-05, "loss": 2.262, "step": 49560 }, { "epoch": 1.7198048540834066e-13, "grad_norm": 0.6469581723213196, "learning_rate": 4.889843090304383e-05, "loss": 2.2778, "step": 49570 }, { "epoch": 1.720151798778602e-13, "grad_norm": 0.6559305191040039, "learning_rate": 4.887750287376428e-05, "loss": 2.2454, "step": 49580 }, { "epoch": 1.7204987434737973e-13, "grad_norm": 0.6638306379318237, "learning_rate": 4.8856577987149056e-05, "loss": 2.2135, "step": 49590 }, { "epoch": 1.7208456881689926e-13, "grad_norm": 0.6407155990600586, "learning_rate": 4.883565624650732e-05, "loss": 2.2481, "step": 49600 }, { "epoch": 1.721192632864188e-13, "grad_norm": 0.6599859595298767, "learning_rate": 4.88147376551476e-05, "loss": 2.2637, "step": 49610 }, { "epoch": 1.7215395775593834e-13, "grad_norm": 0.6534910202026367, "learning_rate": 4.8793822216378e-05, "loss": 2.2334, "step": 49620 }, { "epoch": 1.7218865222545787e-13, "grad_norm": 0.6695348024368286, "learning_rate": 4.877290993350618e-05, "loss": 2.2658, "step": 49630 }, { "epoch": 1.722233466949774e-13, "grad_norm": 0.6529632806777954, "learning_rate": 4.8752000809839145e-05, "loss": 2.2622, "step": 49640 }, { "epoch": 1.7225804116449694e-13, "grad_norm": 0.6569838523864746, "learning_rate": 4.873109484868357e-05, "loss": 2.2395, "step": 49650 }, { "epoch": 1.7229273563401648e-13, "grad_norm": 0.646441638469696, "learning_rate": 4.8710192053345496e-05, "loss": 2.2181, "step": 49660 }, { "epoch": 1.7232743010353602e-13, "grad_norm": 0.6457195281982422, "learning_rate": 4.8689292427130566e-05, "loss": 2.2632, "step": 49670 }, { "epoch": 1.7236212457305555e-13, "grad_norm": 0.6517437696456909, "learning_rate": 4.866839597334384e-05, "loss": 2.2657, "step": 49680 }, { "epoch": 1.723968190425751e-13, "grad_norm": 0.658046543598175, "learning_rate": 4.864750269528993e-05, "loss": 2.2065, "step": 49690 }, { "epoch": 1.7243151351209463e-13, "grad_norm": 0.6385759711265564, "learning_rate": 4.862661259627295e-05, "loss": 2.2539, "step": 49700 }, { "epoch": 1.7246620798161416e-13, "grad_norm": 0.6696906685829163, "learning_rate": 4.8605725679596424e-05, "loss": 2.2237, "step": 49710 }, { "epoch": 1.725009024511337e-13, "grad_norm": 0.6551892161369324, "learning_rate": 4.8584841948563495e-05, "loss": 2.2374, "step": 49720 }, { "epoch": 1.7253559692065323e-13, "grad_norm": 0.639886200428009, "learning_rate": 4.8563961406476744e-05, "loss": 2.2374, "step": 49730 }, { "epoch": 1.7257029139017277e-13, "grad_norm": 0.6505104899406433, "learning_rate": 4.854308405663821e-05, "loss": 2.2504, "step": 49740 }, { "epoch": 1.726049858596923e-13, "grad_norm": 0.6843997836112976, "learning_rate": 4.852220990234951e-05, "loss": 2.2367, "step": 49750 }, { "epoch": 1.7263968032921184e-13, "grad_norm": 0.6925933957099915, "learning_rate": 4.8501338946911664e-05, "loss": 2.2482, "step": 49760 }, { "epoch": 1.7267437479873138e-13, "grad_norm": 0.677096426486969, "learning_rate": 4.8480471193625266e-05, "loss": 2.2373, "step": 49770 }, { "epoch": 1.7270906926825091e-13, "grad_norm": 0.667182207107544, "learning_rate": 4.845960664579041e-05, "loss": 2.226, "step": 49780 }, { "epoch": 1.7274376373777045e-13, "grad_norm": 0.6463890075683594, "learning_rate": 4.843874530670657e-05, "loss": 2.2521, "step": 49790 }, { "epoch": 1.7277845820728999e-13, "grad_norm": 0.6778801679611206, "learning_rate": 4.8417887179672826e-05, "loss": 2.2731, "step": 49800 }, { "epoch": 1.7281315267680952e-13, "grad_norm": 0.6417852640151978, "learning_rate": 4.839703226798771e-05, "loss": 2.2556, "step": 49810 }, { "epoch": 1.7284784714632906e-13, "grad_norm": 0.6505464911460876, "learning_rate": 4.837618057494922e-05, "loss": 2.2829, "step": 49820 }, { "epoch": 1.728825416158486e-13, "grad_norm": 0.6514515280723572, "learning_rate": 4.83553321038549e-05, "loss": 2.2405, "step": 49830 }, { "epoch": 1.7291723608536813e-13, "grad_norm": 0.6259245872497559, "learning_rate": 4.833448685800175e-05, "loss": 2.2467, "step": 49840 }, { "epoch": 1.7295193055488767e-13, "grad_norm": 0.6810902953147888, "learning_rate": 4.831364484068629e-05, "loss": 2.2574, "step": 49850 }, { "epoch": 1.729866250244072e-13, "grad_norm": 0.6422649025917053, "learning_rate": 4.829280605520446e-05, "loss": 2.2564, "step": 49860 }, { "epoch": 1.7302131949392674e-13, "grad_norm": 0.6213822364807129, "learning_rate": 4.827197050485174e-05, "loss": 2.2416, "step": 49870 }, { "epoch": 1.7305601396344628e-13, "grad_norm": 0.6546393036842346, "learning_rate": 4.825113819292315e-05, "loss": 2.2665, "step": 49880 }, { "epoch": 1.730907084329658e-13, "grad_norm": 0.6906052231788635, "learning_rate": 4.823030912271306e-05, "loss": 2.2526, "step": 49890 }, { "epoch": 1.7312540290248535e-13, "grad_norm": 0.6403911113739014, "learning_rate": 4.820948329751547e-05, "loss": 2.2012, "step": 49900 }, { "epoch": 1.7316009737200488e-13, "grad_norm": 0.7019725441932678, "learning_rate": 4.818866072062376e-05, "loss": 2.2474, "step": 49910 }, { "epoch": 1.7319479184152442e-13, "grad_norm": 0.6220070123672485, "learning_rate": 4.81678413953309e-05, "loss": 2.2734, "step": 49920 }, { "epoch": 1.7322948631104396e-13, "grad_norm": 0.6723545789718628, "learning_rate": 4.814702532492922e-05, "loss": 2.2315, "step": 49930 }, { "epoch": 1.732641807805635e-13, "grad_norm": 0.6397414207458496, "learning_rate": 4.812621251271064e-05, "loss": 2.2619, "step": 49940 }, { "epoch": 1.7329887525008303e-13, "grad_norm": 0.6924042701721191, "learning_rate": 4.810540296196653e-05, "loss": 2.2231, "step": 49950 }, { "epoch": 1.7333356971960256e-13, "grad_norm": 0.6579439640045166, "learning_rate": 4.8084596675987735e-05, "loss": 2.2541, "step": 49960 }, { "epoch": 1.733682641891221e-13, "grad_norm": 0.6742655634880066, "learning_rate": 4.806379365806457e-05, "loss": 2.2675, "step": 49970 }, { "epoch": 1.7340295865864164e-13, "grad_norm": 0.6816757917404175, "learning_rate": 4.8042993911486864e-05, "loss": 2.2983, "step": 49980 }, { "epoch": 1.7343765312816117e-13, "grad_norm": 0.666793167591095, "learning_rate": 4.8022197439543944e-05, "loss": 2.2351, "step": 49990 }, { "epoch": 1.734723475976807e-13, "grad_norm": 0.6611276268959045, "learning_rate": 4.800140424552457e-05, "loss": 2.2438, "step": 50000 }, { "epoch": 1.7350704206720025e-13, "grad_norm": 0.6401536464691162, "learning_rate": 4.7980614332717e-05, "loss": 2.1967, "step": 50010 }, { "epoch": 1.7354173653671978e-13, "grad_norm": 0.6650921106338501, "learning_rate": 4.7959827704409e-05, "loss": 2.2518, "step": 50020 }, { "epoch": 1.7357643100623932e-13, "grad_norm": 0.6831722855567932, "learning_rate": 4.7939044363887783e-05, "loss": 2.2517, "step": 50030 }, { "epoch": 1.7361112547575885e-13, "grad_norm": 0.6841179132461548, "learning_rate": 4.791826431444005e-05, "loss": 2.2351, "step": 50040 }, { "epoch": 1.736458199452784e-13, "grad_norm": 0.6566017866134644, "learning_rate": 4.7897487559352e-05, "loss": 2.2476, "step": 50050 }, { "epoch": 1.7368051441479793e-13, "grad_norm": 0.6486518979072571, "learning_rate": 4.7876714101909285e-05, "loss": 2.2378, "step": 50060 }, { "epoch": 1.7371520888431746e-13, "grad_norm": 0.6606009602546692, "learning_rate": 4.785594394539706e-05, "loss": 2.2284, "step": 50070 }, { "epoch": 1.73749903353837e-13, "grad_norm": 0.686075747013092, "learning_rate": 4.783517709309995e-05, "loss": 2.2691, "step": 50080 }, { "epoch": 1.7378459782335653e-13, "grad_norm": 0.6549529433250427, "learning_rate": 4.781441354830203e-05, "loss": 2.2402, "step": 50090 }, { "epoch": 1.7381929229287607e-13, "grad_norm": 0.6840599179267883, "learning_rate": 4.779365331428692e-05, "loss": 2.2245, "step": 50100 }, { "epoch": 1.738539867623956e-13, "grad_norm": 0.643988311290741, "learning_rate": 4.7772896394337626e-05, "loss": 2.2693, "step": 50110 }, { "epoch": 1.7388868123191514e-13, "grad_norm": 0.6760673522949219, "learning_rate": 4.775214279173671e-05, "loss": 2.2319, "step": 50120 }, { "epoch": 1.7392337570143468e-13, "grad_norm": 0.6298791766166687, "learning_rate": 4.7731392509766145e-05, "loss": 2.2371, "step": 50130 }, { "epoch": 1.7395807017095422e-13, "grad_norm": 0.6386145353317261, "learning_rate": 4.771064555170743e-05, "loss": 2.2331, "step": 50140 }, { "epoch": 1.7399276464047375e-13, "grad_norm": 0.6885201930999756, "learning_rate": 4.768990192084152e-05, "loss": 2.2279, "step": 50150 }, { "epoch": 1.740274591099933e-13, "grad_norm": 0.6702964305877686, "learning_rate": 4.766916162044881e-05, "loss": 2.2475, "step": 50160 }, { "epoch": 1.7406215357951282e-13, "grad_norm": 0.6993691921234131, "learning_rate": 4.764842465380925e-05, "loss": 2.2324, "step": 50170 }, { "epoch": 1.7409684804903236e-13, "grad_norm": 0.6675325036048889, "learning_rate": 4.7627691024202184e-05, "loss": 2.2244, "step": 50180 }, { "epoch": 1.741315425185519e-13, "grad_norm": 0.6855633854866028, "learning_rate": 4.760696073490644e-05, "loss": 2.2872, "step": 50190 }, { "epoch": 1.7416623698807143e-13, "grad_norm": 0.6740665435791016, "learning_rate": 4.758623378920039e-05, "loss": 2.2218, "step": 50200 }, { "epoch": 1.7420093145759097e-13, "grad_norm": 0.6623677611351013, "learning_rate": 4.756551019036176e-05, "loss": 2.232, "step": 50210 }, { "epoch": 1.742356259271105e-13, "grad_norm": 0.6539167165756226, "learning_rate": 4.754478994166786e-05, "loss": 2.2652, "step": 50220 }, { "epoch": 1.7427032039663004e-13, "grad_norm": 0.6762073636054993, "learning_rate": 4.7524073046395377e-05, "loss": 2.2046, "step": 50230 }, { "epoch": 1.7430501486614958e-13, "grad_norm": 0.6892687082290649, "learning_rate": 4.750335950782053e-05, "loss": 2.2628, "step": 50240 }, { "epoch": 1.7433970933566911e-13, "grad_norm": 0.6559104323387146, "learning_rate": 4.748264932921897e-05, "loss": 2.214, "step": 50250 }, { "epoch": 1.7437440380518865e-13, "grad_norm": 0.6592954993247986, "learning_rate": 4.746194251386586e-05, "loss": 2.2747, "step": 50260 }, { "epoch": 1.7440909827470819e-13, "grad_norm": 0.6644606590270996, "learning_rate": 4.744123906503582e-05, "loss": 2.2623, "step": 50270 }, { "epoch": 1.7444379274422772e-13, "grad_norm": 0.671148419380188, "learning_rate": 4.742053898600286e-05, "loss": 2.2252, "step": 50280 }, { "epoch": 1.7447848721374726e-13, "grad_norm": 0.6449558138847351, "learning_rate": 4.739984228004057e-05, "loss": 2.2527, "step": 50290 }, { "epoch": 1.745131816832668e-13, "grad_norm": 0.6617783904075623, "learning_rate": 4.737914895042192e-05, "loss": 2.2526, "step": 50300 }, { "epoch": 1.7454787615278633e-13, "grad_norm": 0.6666909456253052, "learning_rate": 4.735845900041941e-05, "loss": 2.2577, "step": 50310 }, { "epoch": 1.7458257062230587e-13, "grad_norm": 0.6817652583122253, "learning_rate": 4.733777243330499e-05, "loss": 2.2426, "step": 50320 }, { "epoch": 1.746172650918254e-13, "grad_norm": 0.6207377314567566, "learning_rate": 4.7317089252350025e-05, "loss": 2.2416, "step": 50330 }, { "epoch": 1.7465195956134494e-13, "grad_norm": 0.673552930355072, "learning_rate": 4.729640946082542e-05, "loss": 2.2364, "step": 50340 }, { "epoch": 1.7468665403086447e-13, "grad_norm": 0.6262770891189575, "learning_rate": 4.727573306200146e-05, "loss": 2.269, "step": 50350 }, { "epoch": 1.74721348500384e-13, "grad_norm": 0.6506192088127136, "learning_rate": 4.725506005914797e-05, "loss": 2.2743, "step": 50360 }, { "epoch": 1.7475604296990355e-13, "grad_norm": 0.7055709958076477, "learning_rate": 4.723439045553419e-05, "loss": 2.2755, "step": 50370 }, { "epoch": 1.7479073743942308e-13, "grad_norm": 0.6766008138656616, "learning_rate": 4.721372425442888e-05, "loss": 2.2601, "step": 50380 }, { "epoch": 1.7482543190894262e-13, "grad_norm": 0.6633342504501343, "learning_rate": 4.719306145910017e-05, "loss": 2.2799, "step": 50390 }, { "epoch": 1.7486012637846216e-13, "grad_norm": 0.6587151288986206, "learning_rate": 4.717240207281573e-05, "loss": 2.2454, "step": 50400 }, { "epoch": 1.748948208479817e-13, "grad_norm": 0.6629822850227356, "learning_rate": 4.7151746098842664e-05, "loss": 2.2628, "step": 50410 }, { "epoch": 1.7492951531750123e-13, "grad_norm": 0.6818222999572754, "learning_rate": 4.713109354044755e-05, "loss": 2.2326, "step": 50420 }, { "epoch": 1.7496420978702076e-13, "grad_norm": 0.638218343257904, "learning_rate": 4.711044440089637e-05, "loss": 2.2311, "step": 50430 }, { "epoch": 1.749989042565403e-13, "grad_norm": 0.6589730978012085, "learning_rate": 4.708979868345467e-05, "loss": 2.2365, "step": 50440 }, { "epoch": 1.7503359872605984e-13, "grad_norm": 0.6275207996368408, "learning_rate": 4.7069156391387314e-05, "loss": 2.2372, "step": 50450 }, { "epoch": 1.7506829319557937e-13, "grad_norm": 0.6578156352043152, "learning_rate": 4.704851752795876e-05, "loss": 2.1886, "step": 50460 }, { "epoch": 1.751029876650989e-13, "grad_norm": 0.661830484867096, "learning_rate": 4.702788209643285e-05, "loss": 2.2687, "step": 50470 }, { "epoch": 1.7513768213461844e-13, "grad_norm": 0.6504020094871521, "learning_rate": 4.7007250100072893e-05, "loss": 2.2733, "step": 50480 }, { "epoch": 1.7517237660413798e-13, "grad_norm": 0.6736395359039307, "learning_rate": 4.69866215421417e-05, "loss": 2.2298, "step": 50490 }, { "epoch": 1.7520707107365752e-13, "grad_norm": 0.6616833806037903, "learning_rate": 4.696599642590145e-05, "loss": 2.2518, "step": 50500 }, { "epoch": 1.7524176554317705e-13, "grad_norm": 0.6278871893882751, "learning_rate": 4.694537475461384e-05, "loss": 2.2251, "step": 50510 }, { "epoch": 1.752764600126966e-13, "grad_norm": 0.6672563552856445, "learning_rate": 4.692475653154005e-05, "loss": 2.2377, "step": 50520 }, { "epoch": 1.7531115448221612e-13, "grad_norm": 0.6513992547988892, "learning_rate": 4.6904141759940595e-05, "loss": 2.2741, "step": 50530 }, { "epoch": 1.7534584895173566e-13, "grad_norm": 0.6841776967048645, "learning_rate": 4.68835304430756e-05, "loss": 2.2088, "step": 50540 }, { "epoch": 1.753805434212552e-13, "grad_norm": 0.6696118116378784, "learning_rate": 4.686292258420452e-05, "loss": 2.239, "step": 50550 }, { "epoch": 1.7541523789077473e-13, "grad_norm": 0.6920375227928162, "learning_rate": 4.6842318186586355e-05, "loss": 2.267, "step": 50560 }, { "epoch": 1.7544993236029427e-13, "grad_norm": 0.632017195224762, "learning_rate": 4.682171725347946e-05, "loss": 2.2559, "step": 50570 }, { "epoch": 1.754846268298138e-13, "grad_norm": 0.6675474643707275, "learning_rate": 4.680111978814171e-05, "loss": 2.2271, "step": 50580 }, { "epoch": 1.7551932129933334e-13, "grad_norm": 0.6745187640190125, "learning_rate": 4.678052579383046e-05, "loss": 2.246, "step": 50590 }, { "epoch": 1.7555401576885288e-13, "grad_norm": 0.6869926452636719, "learning_rate": 4.675993527380244e-05, "loss": 2.2486, "step": 50600 }, { "epoch": 1.7558871023837241e-13, "grad_norm": 0.633073091506958, "learning_rate": 4.673934823131384e-05, "loss": 2.213, "step": 50610 }, { "epoch": 1.7562340470789195e-13, "grad_norm": 0.6786808371543884, "learning_rate": 4.671876466962036e-05, "loss": 2.2859, "step": 50620 }, { "epoch": 1.7565809917741149e-13, "grad_norm": 0.6773485541343689, "learning_rate": 4.66981845919771e-05, "loss": 2.2379, "step": 50630 }, { "epoch": 1.7569279364693102e-13, "grad_norm": 0.6599009037017822, "learning_rate": 4.6677608001638647e-05, "loss": 2.1842, "step": 50640 }, { "epoch": 1.7572748811645056e-13, "grad_norm": 0.6530455946922302, "learning_rate": 4.665703490185895e-05, "loss": 2.2304, "step": 50650 }, { "epoch": 1.757621825859701e-13, "grad_norm": 0.6576827764511108, "learning_rate": 4.663646529589154e-05, "loss": 2.2194, "step": 50660 }, { "epoch": 1.7579687705548963e-13, "grad_norm": 0.6469854116439819, "learning_rate": 4.661589918698927e-05, "loss": 2.2523, "step": 50670 }, { "epoch": 1.7583157152500917e-13, "grad_norm": 0.6567795276641846, "learning_rate": 4.659533657840451e-05, "loss": 2.2649, "step": 50680 }, { "epoch": 1.758662659945287e-13, "grad_norm": 0.6594148874282837, "learning_rate": 4.6574777473389066e-05, "loss": 2.2495, "step": 50690 }, { "epoch": 1.7590096046404824e-13, "grad_norm": 0.6703126430511475, "learning_rate": 4.655422187519418e-05, "loss": 2.2188, "step": 50700 }, { "epoch": 1.7593565493356778e-13, "grad_norm": 0.694254457950592, "learning_rate": 4.653366978707055e-05, "loss": 2.2226, "step": 50710 }, { "epoch": 1.759703494030873e-13, "grad_norm": 0.6587435007095337, "learning_rate": 4.651312121226828e-05, "loss": 2.2482, "step": 50720 }, { "epoch": 1.7600504387260685e-13, "grad_norm": 0.6495389342308044, "learning_rate": 4.6492576154036984e-05, "loss": 2.2667, "step": 50730 }, { "epoch": 1.7603973834212638e-13, "grad_norm": 0.6650920510292053, "learning_rate": 4.64720346156257e-05, "loss": 2.2654, "step": 50740 }, { "epoch": 1.7607443281164592e-13, "grad_norm": 0.7193239331245422, "learning_rate": 4.6451496600282834e-05, "loss": 2.2003, "step": 50750 }, { "epoch": 1.7610912728116546e-13, "grad_norm": 0.6869370341300964, "learning_rate": 4.643096211125636e-05, "loss": 2.2162, "step": 50760 }, { "epoch": 1.76143821750685e-13, "grad_norm": 0.6359992623329163, "learning_rate": 4.641043115179359e-05, "loss": 2.2571, "step": 50770 }, { "epoch": 1.7617851622020453e-13, "grad_norm": 0.6494545936584473, "learning_rate": 4.6389903725141345e-05, "loss": 2.2773, "step": 50780 }, { "epoch": 1.7621321068972406e-13, "grad_norm": 0.6623913645744324, "learning_rate": 4.636937983454584e-05, "loss": 2.2416, "step": 50790 }, { "epoch": 1.762479051592436e-13, "grad_norm": 0.7157592177391052, "learning_rate": 4.634885948325275e-05, "loss": 2.2806, "step": 50800 }, { "epoch": 1.7628259962876314e-13, "grad_norm": 0.6729205250740051, "learning_rate": 4.632834267450724e-05, "loss": 2.2058, "step": 50810 }, { "epoch": 1.7631729409828267e-13, "grad_norm": 0.6807459592819214, "learning_rate": 4.6307829411553804e-05, "loss": 2.2283, "step": 50820 }, { "epoch": 1.763519885678022e-13, "grad_norm": 0.6741043329238892, "learning_rate": 4.628731969763646e-05, "loss": 2.2415, "step": 50830 }, { "epoch": 1.7638668303732175e-13, "grad_norm": 0.6472215056419373, "learning_rate": 4.626681353599865e-05, "loss": 2.2569, "step": 50840 }, { "epoch": 1.7642137750684128e-13, "grad_norm": 0.6864660978317261, "learning_rate": 4.624631092988324e-05, "loss": 2.2689, "step": 50850 }, { "epoch": 1.7645607197636082e-13, "grad_norm": 0.6491309404373169, "learning_rate": 4.6225811882532556e-05, "loss": 2.24, "step": 50860 }, { "epoch": 1.7649076644588035e-13, "grad_norm": 0.6734909415245056, "learning_rate": 4.6205316397188304e-05, "loss": 2.2473, "step": 50870 }, { "epoch": 1.765254609153999e-13, "grad_norm": 0.6334789395332336, "learning_rate": 4.618482447709174e-05, "loss": 2.2575, "step": 50880 }, { "epoch": 1.7656015538491943e-13, "grad_norm": 0.6747210621833801, "learning_rate": 4.6164336125483404e-05, "loss": 2.2532, "step": 50890 }, { "epoch": 1.7659484985443896e-13, "grad_norm": 0.6646233797073364, "learning_rate": 4.6143851345603405e-05, "loss": 2.254, "step": 50900 }, { "epoch": 1.766295443239585e-13, "grad_norm": 0.6597729921340942, "learning_rate": 4.6123370140691235e-05, "loss": 2.2503, "step": 50910 }, { "epoch": 1.7666423879347803e-13, "grad_norm": 0.6683228015899658, "learning_rate": 4.610289251398579e-05, "loss": 2.234, "step": 50920 }, { "epoch": 1.7669893326299757e-13, "grad_norm": 0.6456986665725708, "learning_rate": 4.608241846872545e-05, "loss": 2.2144, "step": 50930 }, { "epoch": 1.767336277325171e-13, "grad_norm": 0.678687334060669, "learning_rate": 4.606194800814799e-05, "loss": 2.2553, "step": 50940 }, { "epoch": 1.7676832220203664e-13, "grad_norm": 0.6331659555435181, "learning_rate": 4.6041481135490656e-05, "loss": 2.2448, "step": 50950 }, { "epoch": 1.7680301667155618e-13, "grad_norm": 0.656473696231842, "learning_rate": 4.602101785399014e-05, "loss": 2.2371, "step": 50960 }, { "epoch": 1.7683771114107572e-13, "grad_norm": 0.6882248520851135, "learning_rate": 4.6000558166882465e-05, "loss": 2.2026, "step": 50970 }, { "epoch": 1.7687240561059525e-13, "grad_norm": 0.6298640966415405, "learning_rate": 4.59801020774032e-05, "loss": 2.204, "step": 50980 }, { "epoch": 1.769071000801148e-13, "grad_norm": 0.6612494587898254, "learning_rate": 4.5959649588787286e-05, "loss": 2.2223, "step": 50990 }, { "epoch": 1.7694179454963432e-13, "grad_norm": 0.6725093126296997, "learning_rate": 4.5939200704269104e-05, "loss": 2.238, "step": 51000 }, { "epoch": 1.7697648901915386e-13, "grad_norm": 0.6554820537567139, "learning_rate": 4.591875542708247e-05, "loss": 2.2199, "step": 51010 }, { "epoch": 1.770111834886734e-13, "grad_norm": 0.6537080407142639, "learning_rate": 4.589831376046063e-05, "loss": 2.2263, "step": 51020 }, { "epoch": 1.7704587795819293e-13, "grad_norm": 0.666305661201477, "learning_rate": 4.587787570763628e-05, "loss": 2.2334, "step": 51030 }, { "epoch": 1.7708057242771247e-13, "grad_norm": 0.6706642508506775, "learning_rate": 4.585744127184148e-05, "loss": 2.2872, "step": 51040 }, { "epoch": 1.77115266897232e-13, "grad_norm": 0.6497992277145386, "learning_rate": 4.5837010456307785e-05, "loss": 2.2517, "step": 51050 }, { "epoch": 1.7714996136675154e-13, "grad_norm": 0.6581203937530518, "learning_rate": 4.581658326426617e-05, "loss": 2.2695, "step": 51060 }, { "epoch": 1.7718465583627108e-13, "grad_norm": 0.6849382519721985, "learning_rate": 4.5796159698946985e-05, "loss": 2.2577, "step": 51070 }, { "epoch": 1.7721935030579061e-13, "grad_norm": 0.670357346534729, "learning_rate": 4.577573976358006e-05, "loss": 2.2859, "step": 51080 }, { "epoch": 1.7725404477531015e-13, "grad_norm": 0.6641718149185181, "learning_rate": 4.575532346139463e-05, "loss": 2.2839, "step": 51090 }, { "epoch": 1.7728873924482969e-13, "grad_norm": 0.6476278901100159, "learning_rate": 4.573491079561934e-05, "loss": 2.2497, "step": 51100 }, { "epoch": 1.7732343371434922e-13, "grad_norm": 0.6458255648612976, "learning_rate": 4.5714501769482304e-05, "loss": 2.21, "step": 51110 }, { "epoch": 1.7735812818386876e-13, "grad_norm": 0.6764106750488281, "learning_rate": 4.569409638621101e-05, "loss": 2.2218, "step": 51120 }, { "epoch": 1.773928226533883e-13, "grad_norm": 0.6746717691421509, "learning_rate": 4.567369464903244e-05, "loss": 2.2609, "step": 51130 }, { "epoch": 1.7742751712290783e-13, "grad_norm": 0.6695423722267151, "learning_rate": 4.565329656117288e-05, "loss": 2.2324, "step": 51140 }, { "epoch": 1.7746221159242737e-13, "grad_norm": 0.6610060334205627, "learning_rate": 4.563290212585816e-05, "loss": 2.2431, "step": 51150 }, { "epoch": 1.774969060619469e-13, "grad_norm": 0.649928092956543, "learning_rate": 4.56125113463135e-05, "loss": 2.2572, "step": 51160 }, { "epoch": 1.7753160053146644e-13, "grad_norm": 0.6612371206283569, "learning_rate": 4.559212422576348e-05, "loss": 2.2612, "step": 51170 }, { "epoch": 1.7756629500098597e-13, "grad_norm": 0.6472668051719666, "learning_rate": 4.557174076743217e-05, "loss": 2.261, "step": 51180 }, { "epoch": 1.776009894705055e-13, "grad_norm": 0.6732909679412842, "learning_rate": 4.555136097454304e-05, "loss": 2.1992, "step": 51190 }, { "epoch": 1.7763568394002505e-13, "grad_norm": 0.6405823230743408, "learning_rate": 4.553098485031899e-05, "loss": 2.2696, "step": 51200 }, { "epoch": 1.7767037840954458e-13, "grad_norm": 0.65529865026474, "learning_rate": 4.551061239798229e-05, "loss": 2.2417, "step": 51210 }, { "epoch": 1.7770507287906412e-13, "grad_norm": 0.6570575833320618, "learning_rate": 4.5490243620754714e-05, "loss": 2.2339, "step": 51220 }, { "epoch": 1.7773976734858365e-13, "grad_norm": 0.6600787043571472, "learning_rate": 4.546987852185738e-05, "loss": 2.2204, "step": 51230 }, { "epoch": 1.777744618181032e-13, "grad_norm": 0.6977851986885071, "learning_rate": 4.5449517104510854e-05, "loss": 2.2427, "step": 51240 }, { "epoch": 1.7780915628762273e-13, "grad_norm": 0.6708824038505554, "learning_rate": 4.5429159371935145e-05, "loss": 2.2164, "step": 51250 }, { "epoch": 1.7784385075714226e-13, "grad_norm": 0.6699910759925842, "learning_rate": 4.5408805327349616e-05, "loss": 2.2515, "step": 51260 }, { "epoch": 1.778785452266618e-13, "grad_norm": 0.6695120930671692, "learning_rate": 4.5388454973973094e-05, "loss": 2.2166, "step": 51270 }, { "epoch": 1.7791323969618134e-13, "grad_norm": 0.6568233370780945, "learning_rate": 4.5368108315023844e-05, "loss": 2.2107, "step": 51280 }, { "epoch": 1.7794793416570087e-13, "grad_norm": 0.6958299875259399, "learning_rate": 4.5347765353719486e-05, "loss": 2.2353, "step": 51290 }, { "epoch": 1.779826286352204e-13, "grad_norm": 0.6625078916549683, "learning_rate": 4.53274260932771e-05, "loss": 2.2346, "step": 51300 }, { "epoch": 1.7801732310473994e-13, "grad_norm": 0.6570340991020203, "learning_rate": 4.530709053691312e-05, "loss": 2.2377, "step": 51310 }, { "epoch": 1.7805201757425948e-13, "grad_norm": 0.6589665412902832, "learning_rate": 4.52867586878435e-05, "loss": 2.2182, "step": 51320 }, { "epoch": 1.7808671204377902e-13, "grad_norm": 0.6614060401916504, "learning_rate": 4.526643054928352e-05, "loss": 2.239, "step": 51330 }, { "epoch": 1.7812140651329855e-13, "grad_norm": 0.6706212162971497, "learning_rate": 4.524610612444789e-05, "loss": 2.2403, "step": 51340 }, { "epoch": 1.781561009828181e-13, "grad_norm": 0.6588356494903564, "learning_rate": 4.5225785416550755e-05, "loss": 2.2115, "step": 51350 }, { "epoch": 1.7819079545233762e-13, "grad_norm": 2.082343578338623, "learning_rate": 4.520546842880567e-05, "loss": 2.2304, "step": 51360 }, { "epoch": 1.7822548992185716e-13, "grad_norm": 0.640962541103363, "learning_rate": 4.518515516442556e-05, "loss": 2.2175, "step": 51370 }, { "epoch": 1.782601843913767e-13, "grad_norm": 0.6354601979255676, "learning_rate": 4.516484562662282e-05, "loss": 2.2486, "step": 51380 }, { "epoch": 1.7829487886089623e-13, "grad_norm": 0.685139536857605, "learning_rate": 4.514453981860923e-05, "loss": 2.2405, "step": 51390 }, { "epoch": 1.7832957333041577e-13, "grad_norm": 0.6753227114677429, "learning_rate": 4.512423774359597e-05, "loss": 2.2484, "step": 51400 }, { "epoch": 1.783642677999353e-13, "grad_norm": 0.6603354215621948, "learning_rate": 4.510393940479365e-05, "loss": 2.2688, "step": 51410 }, { "epoch": 1.7839896226945484e-13, "grad_norm": 0.6916431784629822, "learning_rate": 4.508364480541223e-05, "loss": 2.2598, "step": 51420 }, { "epoch": 1.7843365673897438e-13, "grad_norm": 0.6522465348243713, "learning_rate": 4.5063353948661205e-05, "loss": 2.2787, "step": 51430 }, { "epoch": 1.7846835120849391e-13, "grad_norm": 0.6616971492767334, "learning_rate": 4.5043066837749333e-05, "loss": 2.2154, "step": 51440 }, { "epoch": 1.7850304567801345e-13, "grad_norm": 0.6645694375038147, "learning_rate": 4.5022783475884874e-05, "loss": 2.225, "step": 51450 }, { "epoch": 1.7853774014753299e-13, "grad_norm": 0.7097288370132446, "learning_rate": 4.5002503866275465e-05, "loss": 2.2443, "step": 51460 }, { "epoch": 1.7857243461705252e-13, "grad_norm": 0.7174943685531616, "learning_rate": 4.498222801212813e-05, "loss": 2.2385, "step": 51470 }, { "epoch": 1.7860712908657206e-13, "grad_norm": 0.6662665009498596, "learning_rate": 4.4961955916649383e-05, "loss": 2.2258, "step": 51480 }, { "epoch": 1.786418235560916e-13, "grad_norm": 0.677483081817627, "learning_rate": 4.4941687583044993e-05, "loss": 2.2161, "step": 51490 }, { "epoch": 1.7867651802561113e-13, "grad_norm": 0.6726411581039429, "learning_rate": 4.4921423014520316e-05, "loss": 2.221, "step": 51500 }, { "epoch": 1.7871121249513067e-13, "grad_norm": 0.6688690781593323, "learning_rate": 4.490116221427992e-05, "loss": 2.2294, "step": 51510 }, { "epoch": 1.787459069646502e-13, "grad_norm": 0.673214852809906, "learning_rate": 4.488090518552797e-05, "loss": 2.1975, "step": 51520 }, { "epoch": 1.7878060143416974e-13, "grad_norm": 0.6816896200180054, "learning_rate": 4.4860651931467876e-05, "loss": 2.2431, "step": 51530 }, { "epoch": 1.7881529590368928e-13, "grad_norm": 0.6541166305541992, "learning_rate": 4.4840402455302534e-05, "loss": 2.2191, "step": 51540 }, { "epoch": 1.788499903732088e-13, "grad_norm": 0.6505644917488098, "learning_rate": 4.482015676023425e-05, "loss": 2.2476, "step": 51550 }, { "epoch": 1.7888468484272835e-13, "grad_norm": 0.6997380256652832, "learning_rate": 4.479991484946466e-05, "loss": 2.2445, "step": 51560 }, { "epoch": 1.7891937931224788e-13, "grad_norm": 0.6450188755989075, "learning_rate": 4.477967672619492e-05, "loss": 2.2517, "step": 51570 }, { "epoch": 1.7895407378176742e-13, "grad_norm": 0.6810334324836731, "learning_rate": 4.4759442393625405e-05, "loss": 2.2245, "step": 51580 }, { "epoch": 1.7898876825128696e-13, "grad_norm": 0.6489949226379395, "learning_rate": 4.47392118549561e-05, "loss": 2.1896, "step": 51590 }, { "epoch": 1.790234627208065e-13, "grad_norm": 0.6472588181495667, "learning_rate": 4.471898511338625e-05, "loss": 2.1916, "step": 51600 }, { "epoch": 1.7905815719032603e-13, "grad_norm": 0.6486843228340149, "learning_rate": 4.469876217211452e-05, "loss": 2.2491, "step": 51610 }, { "epoch": 1.7909285165984556e-13, "grad_norm": 0.6721324920654297, "learning_rate": 4.467854303433906e-05, "loss": 2.2594, "step": 51620 }, { "epoch": 1.791275461293651e-13, "grad_norm": 0.6858819127082825, "learning_rate": 4.465832770325727e-05, "loss": 2.216, "step": 51630 }, { "epoch": 1.7916224059888464e-13, "grad_norm": 0.6446341276168823, "learning_rate": 4.463811618206609e-05, "loss": 2.2297, "step": 51640 }, { "epoch": 1.7919693506840417e-13, "grad_norm": 0.6954633593559265, "learning_rate": 4.4617908473961785e-05, "loss": 2.2384, "step": 51650 }, { "epoch": 1.792316295379237e-13, "grad_norm": 0.6658587455749512, "learning_rate": 4.459770458214002e-05, "loss": 2.2463, "step": 51660 }, { "epoch": 1.7926632400744325e-13, "grad_norm": 0.6736853122711182, "learning_rate": 4.4577504509795875e-05, "loss": 2.2755, "step": 51670 }, { "epoch": 1.7930101847696278e-13, "grad_norm": 0.6703035831451416, "learning_rate": 4.4557308260123814e-05, "loss": 2.2326, "step": 51680 }, { "epoch": 1.7933571294648232e-13, "grad_norm": 0.6733819842338562, "learning_rate": 4.4537115836317674e-05, "loss": 2.2579, "step": 51690 }, { "epoch": 1.7937040741600185e-13, "grad_norm": 0.669181764125824, "learning_rate": 4.451692724157075e-05, "loss": 2.2633, "step": 51700 }, { "epoch": 1.794051018855214e-13, "grad_norm": 0.6907341480255127, "learning_rate": 4.44967424790757e-05, "loss": 2.228, "step": 51710 }, { "epoch": 1.7943979635504093e-13, "grad_norm": 0.6631664037704468, "learning_rate": 4.447656155202455e-05, "loss": 2.2082, "step": 51720 }, { "epoch": 1.7947449082456046e-13, "grad_norm": 0.6443657875061035, "learning_rate": 4.445638446360874e-05, "loss": 2.2417, "step": 51730 }, { "epoch": 1.7950918529408e-13, "grad_norm": 0.6683334112167358, "learning_rate": 4.4436211217019074e-05, "loss": 2.2405, "step": 51740 }, { "epoch": 1.7954387976359953e-13, "grad_norm": 0.6586707234382629, "learning_rate": 4.441604181544587e-05, "loss": 2.218, "step": 51750 }, { "epoch": 1.7957857423311907e-13, "grad_norm": 0.7044174671173096, "learning_rate": 4.4395876262078635e-05, "loss": 2.2114, "step": 51760 }, { "epoch": 1.796132687026386e-13, "grad_norm": 0.67087322473526, "learning_rate": 4.4375714560106446e-05, "loss": 2.2794, "step": 51770 }, { "epoch": 1.7964796317215814e-13, "grad_norm": 0.6853402853012085, "learning_rate": 4.43555567127177e-05, "loss": 2.2454, "step": 51780 }, { "epoch": 1.7968265764167768e-13, "grad_norm": 0.688803493976593, "learning_rate": 4.433540272310013e-05, "loss": 2.2427, "step": 51790 }, { "epoch": 1.7971735211119721e-13, "grad_norm": 0.69647216796875, "learning_rate": 4.431525259444101e-05, "loss": 2.2374, "step": 51800 }, { "epoch": 1.7975204658071675e-13, "grad_norm": 0.6697468161582947, "learning_rate": 4.429510632992681e-05, "loss": 2.2529, "step": 51810 }, { "epoch": 1.7978674105023629e-13, "grad_norm": 0.6736290454864502, "learning_rate": 4.427496393274358e-05, "loss": 2.2451, "step": 51820 }, { "epoch": 1.7982143551975582e-13, "grad_norm": 0.6633614301681519, "learning_rate": 4.425482540607657e-05, "loss": 2.2544, "step": 51830 }, { "epoch": 1.7985612998927536e-13, "grad_norm": 0.6817221641540527, "learning_rate": 4.42346907531106e-05, "loss": 2.2116, "step": 51840 }, { "epoch": 1.798908244587949e-13, "grad_norm": 0.6715324521064758, "learning_rate": 4.421455997702975e-05, "loss": 2.223, "step": 51850 }, { "epoch": 1.7992551892831443e-13, "grad_norm": 0.6499103903770447, "learning_rate": 4.419443308101752e-05, "loss": 2.2271, "step": 51860 }, { "epoch": 1.7996021339783397e-13, "grad_norm": 0.6393144726753235, "learning_rate": 4.417431006825685e-05, "loss": 2.2521, "step": 51870 }, { "epoch": 1.799949078673535e-13, "grad_norm": 0.6592061519622803, "learning_rate": 4.4154190941929976e-05, "loss": 2.2413, "step": 51880 }, { "epoch": 1.8002960233687304e-13, "grad_norm": 0.6552098393440247, "learning_rate": 4.413407570521862e-05, "loss": 2.2605, "step": 51890 }, { "epoch": 1.8006429680639258e-13, "grad_norm": 0.6583502888679504, "learning_rate": 4.4113964361303726e-05, "loss": 2.2217, "step": 51900 }, { "epoch": 1.800989912759121e-13, "grad_norm": 0.6797038316726685, "learning_rate": 4.4093856913365854e-05, "loss": 2.2295, "step": 51910 }, { "epoch": 1.8013368574543165e-13, "grad_norm": 0.6645620465278625, "learning_rate": 4.407375336458476e-05, "loss": 2.2702, "step": 51920 }, { "epoch": 1.8016838021495118e-13, "grad_norm": 0.6605492234230042, "learning_rate": 4.405365371813963e-05, "loss": 2.2553, "step": 51930 }, { "epoch": 1.8020307468447072e-13, "grad_norm": 0.6358739733695984, "learning_rate": 4.403355797720912e-05, "loss": 2.2433, "step": 51940 }, { "epoch": 1.8023776915399026e-13, "grad_norm": 0.6560929417610168, "learning_rate": 4.401346614497112e-05, "loss": 2.2459, "step": 51950 }, { "epoch": 1.802724636235098e-13, "grad_norm": 0.6958606243133545, "learning_rate": 4.399337822460303e-05, "loss": 2.231, "step": 51960 }, { "epoch": 1.8030715809302933e-13, "grad_norm": 0.6711402535438538, "learning_rate": 4.397329421928156e-05, "loss": 2.2173, "step": 51970 }, { "epoch": 1.8034185256254887e-13, "grad_norm": 0.6646532416343689, "learning_rate": 4.395321413218284e-05, "loss": 2.2478, "step": 51980 }, { "epoch": 1.803765470320684e-13, "grad_norm": 0.6518904566764832, "learning_rate": 4.393313796648235e-05, "loss": 2.203, "step": 51990 }, { "epoch": 1.8041124150158794e-13, "grad_norm": 0.6499040126800537, "learning_rate": 4.3913065725354986e-05, "loss": 2.2189, "step": 52000 }, { "epoch": 1.8044593597110747e-13, "grad_norm": 0.6587778925895691, "learning_rate": 4.389299741197493e-05, "loss": 2.2316, "step": 52010 }, { "epoch": 1.80480630440627e-13, "grad_norm": 0.6494327187538147, "learning_rate": 4.3872933029515895e-05, "loss": 2.2735, "step": 52020 }, { "epoch": 1.8051532491014655e-13, "grad_norm": 0.6748734712600708, "learning_rate": 4.385287258115085e-05, "loss": 2.2582, "step": 52030 }, { "epoch": 1.8055001937966608e-13, "grad_norm": 0.6549054980278015, "learning_rate": 4.3832816070052195e-05, "loss": 2.2636, "step": 52040 }, { "epoch": 1.8058471384918562e-13, "grad_norm": 0.6651930212974548, "learning_rate": 4.38127634993917e-05, "loss": 2.2369, "step": 52050 }, { "epoch": 1.8061940831870515e-13, "grad_norm": 0.6477914452552795, "learning_rate": 4.379271487234046e-05, "loss": 2.2343, "step": 52060 }, { "epoch": 1.806541027882247e-13, "grad_norm": 0.6414824724197388, "learning_rate": 4.3772670192069085e-05, "loss": 2.2053, "step": 52070 }, { "epoch": 1.8068879725774423e-13, "grad_norm": 0.6520712971687317, "learning_rate": 4.3752629461747364e-05, "loss": 2.2371, "step": 52080 }, { "epoch": 1.8072349172726376e-13, "grad_norm": 0.6691559553146362, "learning_rate": 4.3732592684544645e-05, "loss": 2.2258, "step": 52090 }, { "epoch": 1.807581861967833e-13, "grad_norm": 0.6567494869232178, "learning_rate": 4.371255986362953e-05, "loss": 2.2411, "step": 52100 }, { "epoch": 1.8079288066630284e-13, "grad_norm": 0.6748369932174683, "learning_rate": 4.3692531002170024e-05, "loss": 2.2323, "step": 52110 }, { "epoch": 1.8082757513582237e-13, "grad_norm": 0.6691351532936096, "learning_rate": 4.367250610333359e-05, "loss": 2.2678, "step": 52120 }, { "epoch": 1.808622696053419e-13, "grad_norm": 0.6770411133766174, "learning_rate": 4.3652485170286916e-05, "loss": 2.2314, "step": 52130 }, { "epoch": 1.8089696407486144e-13, "grad_norm": 0.6564658880233765, "learning_rate": 4.363246820619621e-05, "loss": 2.2482, "step": 52140 }, { "epoch": 1.8093165854438098e-13, "grad_norm": 0.6713173985481262, "learning_rate": 4.361245521422689e-05, "loss": 2.2469, "step": 52150 }, { "epoch": 1.8096635301390052e-13, "grad_norm": 0.71761155128479, "learning_rate": 4.359244619754392e-05, "loss": 2.2286, "step": 52160 }, { "epoch": 1.8100104748342005e-13, "grad_norm": 0.6923911571502686, "learning_rate": 4.3572441159311524e-05, "loss": 2.2367, "step": 52170 }, { "epoch": 1.810357419529396e-13, "grad_norm": 0.6280853152275085, "learning_rate": 4.355244010269331e-05, "loss": 2.2417, "step": 52180 }, { "epoch": 1.8107043642245912e-13, "grad_norm": 0.6730988025665283, "learning_rate": 4.353244303085234e-05, "loss": 2.2686, "step": 52190 }, { "epoch": 1.8110513089197866e-13, "grad_norm": 0.6737232804298401, "learning_rate": 4.351244994695087e-05, "loss": 2.2668, "step": 52200 }, { "epoch": 1.811398253614982e-13, "grad_norm": 0.6907827258110046, "learning_rate": 4.349246085415076e-05, "loss": 2.2433, "step": 52210 }, { "epoch": 1.8117451983101773e-13, "grad_norm": 0.6724910140037537, "learning_rate": 4.3472475755613e-05, "loss": 2.2642, "step": 52220 }, { "epoch": 1.8120921430053727e-13, "grad_norm": 0.695618212223053, "learning_rate": 4.3452494654498125e-05, "loss": 2.2501, "step": 52230 }, { "epoch": 1.812439087700568e-13, "grad_norm": 0.6691828966140747, "learning_rate": 4.343251755396595e-05, "loss": 2.2476, "step": 52240 }, { "epoch": 1.8127860323957634e-13, "grad_norm": 0.6381668448448181, "learning_rate": 4.3412544457175684e-05, "loss": 2.224, "step": 52250 }, { "epoch": 1.8131329770909588e-13, "grad_norm": 0.6541465520858765, "learning_rate": 4.3392575367285934e-05, "loss": 2.2441, "step": 52260 }, { "epoch": 1.8134799217861541e-13, "grad_norm": 0.6691240072250366, "learning_rate": 4.3372610287454576e-05, "loss": 2.2199, "step": 52270 }, { "epoch": 1.8138268664813495e-13, "grad_norm": 0.703231692314148, "learning_rate": 4.335264922083897e-05, "loss": 2.2327, "step": 52280 }, { "epoch": 1.8141738111765449e-13, "grad_norm": 0.6476724147796631, "learning_rate": 4.333269217059577e-05, "loss": 2.2436, "step": 52290 }, { "epoch": 1.8145207558717402e-13, "grad_norm": 0.6784743070602417, "learning_rate": 4.331273913988101e-05, "loss": 2.2594, "step": 52300 }, { "epoch": 1.8148677005669356e-13, "grad_norm": 0.6508558392524719, "learning_rate": 4.329279013185008e-05, "loss": 2.229, "step": 52310 }, { "epoch": 1.815214645262131e-13, "grad_norm": 0.6367679238319397, "learning_rate": 4.327284514965775e-05, "loss": 2.2381, "step": 52320 }, { "epoch": 1.8155615899573263e-13, "grad_norm": 0.6914564967155457, "learning_rate": 4.325290419645816e-05, "loss": 2.2218, "step": 52330 }, { "epoch": 1.8159085346525217e-13, "grad_norm": 0.6791236996650696, "learning_rate": 4.3232967275404795e-05, "loss": 2.2042, "step": 52340 }, { "epoch": 1.816255479347717e-13, "grad_norm": 0.6534780859947205, "learning_rate": 4.32130343896505e-05, "loss": 2.2245, "step": 52350 }, { "epoch": 1.8166024240429124e-13, "grad_norm": 0.6602740287780762, "learning_rate": 4.31931055423475e-05, "loss": 2.235, "step": 52360 }, { "epoch": 1.8169493687381078e-13, "grad_norm": 0.6656560301780701, "learning_rate": 4.317318073664735e-05, "loss": 2.2376, "step": 52370 }, { "epoch": 1.817296313433303e-13, "grad_norm": 0.6455004215240479, "learning_rate": 4.315325997570098e-05, "loss": 2.2225, "step": 52380 }, { "epoch": 1.8176432581284985e-13, "grad_norm": 0.7052327394485474, "learning_rate": 4.3133343262658764e-05, "loss": 2.2332, "step": 52390 }, { "epoch": 1.8179902028236938e-13, "grad_norm": 0.6520074605941772, "learning_rate": 4.3113430600670245e-05, "loss": 2.1997, "step": 52400 }, { "epoch": 1.8183371475188892e-13, "grad_norm": 0.6701760292053223, "learning_rate": 4.309352199288452e-05, "loss": 2.2464, "step": 52410 }, { "epoch": 1.8186840922140846e-13, "grad_norm": 0.6598680019378662, "learning_rate": 4.3073617442449944e-05, "loss": 2.2458, "step": 52420 }, { "epoch": 1.81903103690928e-13, "grad_norm": 0.6785238981246948, "learning_rate": 4.305371695251421e-05, "loss": 2.2234, "step": 52430 }, { "epoch": 1.8193779816044753e-13, "grad_norm": 0.6488957405090332, "learning_rate": 4.3033820526224507e-05, "loss": 2.222, "step": 52440 }, { "epoch": 1.8197249262996706e-13, "grad_norm": 0.6570152044296265, "learning_rate": 4.3013928166727175e-05, "loss": 2.2531, "step": 52450 }, { "epoch": 1.820071870994866e-13, "grad_norm": 0.6664747595787048, "learning_rate": 4.29940398771681e-05, "loss": 2.2267, "step": 52460 }, { "epoch": 1.8204188156900614e-13, "grad_norm": 0.6675178408622742, "learning_rate": 4.297415566069237e-05, "loss": 2.2019, "step": 52470 }, { "epoch": 1.8207657603852567e-13, "grad_norm": 0.800495445728302, "learning_rate": 4.295427552044457e-05, "loss": 2.2353, "step": 52480 }, { "epoch": 1.821112705080452e-13, "grad_norm": 0.669031023979187, "learning_rate": 4.293439945956854e-05, "loss": 2.2332, "step": 52490 }, { "epoch": 1.8214596497756474e-13, "grad_norm": 0.688978374004364, "learning_rate": 4.2914527481207486e-05, "loss": 2.2355, "step": 52500 }, { "epoch": 1.8218065944708428e-13, "grad_norm": 0.6755325794219971, "learning_rate": 4.289465958850407e-05, "loss": 2.2459, "step": 52510 }, { "epoch": 1.8221535391660382e-13, "grad_norm": 0.6865004897117615, "learning_rate": 4.287479578460013e-05, "loss": 2.2754, "step": 52520 }, { "epoch": 1.8225004838612335e-13, "grad_norm": 0.6863786578178406, "learning_rate": 4.285493607263704e-05, "loss": 2.2558, "step": 52530 }, { "epoch": 1.822847428556429e-13, "grad_norm": 0.7013972401618958, "learning_rate": 4.2835080455755365e-05, "loss": 2.228, "step": 52540 }, { "epoch": 1.8231943732516243e-13, "grad_norm": 0.6612114906311035, "learning_rate": 4.2815228937095155e-05, "loss": 2.2422, "step": 52550 }, { "epoch": 1.8235413179468196e-13, "grad_norm": 0.674712598323822, "learning_rate": 4.279538151979575e-05, "loss": 2.22, "step": 52560 }, { "epoch": 1.823888262642015e-13, "grad_norm": 0.6807072162628174, "learning_rate": 4.2775538206995806e-05, "loss": 2.2405, "step": 52570 }, { "epoch": 1.8242352073372103e-13, "grad_norm": 0.69111168384552, "learning_rate": 4.275569900183346e-05, "loss": 2.2049, "step": 52580 }, { "epoch": 1.8245821520324057e-13, "grad_norm": 0.6542432904243469, "learning_rate": 4.2735863907446004e-05, "loss": 2.1987, "step": 52590 }, { "epoch": 1.824929096727601e-13, "grad_norm": 0.6642695665359497, "learning_rate": 4.271603292697027e-05, "loss": 2.2175, "step": 52600 }, { "epoch": 1.8252760414227964e-13, "grad_norm": 0.652988076210022, "learning_rate": 4.2696206063542314e-05, "loss": 2.1982, "step": 52610 }, { "epoch": 1.8256229861179918e-13, "grad_norm": 0.6390336155891418, "learning_rate": 4.2676383320297605e-05, "loss": 2.2253, "step": 52620 }, { "epoch": 1.8259699308131871e-13, "grad_norm": 0.6523343920707703, "learning_rate": 4.265656470037092e-05, "loss": 2.2654, "step": 52630 }, { "epoch": 1.8263168755083825e-13, "grad_norm": 0.6482159495353699, "learning_rate": 4.2636750206896405e-05, "loss": 2.229, "step": 52640 }, { "epoch": 1.8266638202035779e-13, "grad_norm": 0.67522794008255, "learning_rate": 4.261693984300758e-05, "loss": 2.2429, "step": 52650 }, { "epoch": 1.8270107648987732e-13, "grad_norm": 0.6669532656669617, "learning_rate": 4.2597133611837264e-05, "loss": 2.2677, "step": 52660 }, { "epoch": 1.8273577095939686e-13, "grad_norm": 0.6594654321670532, "learning_rate": 4.2577331516517636e-05, "loss": 2.2488, "step": 52670 }, { "epoch": 1.827704654289164e-13, "grad_norm": 0.7072031497955322, "learning_rate": 4.255753356018024e-05, "loss": 2.2422, "step": 52680 }, { "epoch": 1.8280515989843593e-13, "grad_norm": 0.6857393383979797, "learning_rate": 4.253773974595594e-05, "loss": 2.2393, "step": 52690 }, { "epoch": 1.8283985436795547e-13, "grad_norm": 0.660468339920044, "learning_rate": 4.251795007697495e-05, "loss": 2.2464, "step": 52700 }, { "epoch": 1.82874548837475e-13, "grad_norm": 0.6486420631408691, "learning_rate": 4.249816455636686e-05, "loss": 2.2602, "step": 52710 }, { "epoch": 1.8290924330699454e-13, "grad_norm": 0.6525582075119019, "learning_rate": 4.247838318726058e-05, "loss": 2.2537, "step": 52720 }, { "epoch": 1.8294393777651408e-13, "grad_norm": 0.6706557273864746, "learning_rate": 4.2458605972784333e-05, "loss": 2.2627, "step": 52730 }, { "epoch": 1.829786322460336e-13, "grad_norm": 0.6663236618041992, "learning_rate": 4.2438832916065745e-05, "loss": 2.2661, "step": 52740 }, { "epoch": 1.8301332671555315e-13, "grad_norm": 0.6716512441635132, "learning_rate": 4.241906402023172e-05, "loss": 2.2561, "step": 52750 }, { "epoch": 1.8304802118507268e-13, "grad_norm": 0.6660996675491333, "learning_rate": 4.23992992884086e-05, "loss": 2.2476, "step": 52760 }, { "epoch": 1.8308271565459222e-13, "grad_norm": 0.6408829689025879, "learning_rate": 4.237953872372194e-05, "loss": 2.2795, "step": 52770 }, { "epoch": 1.8311741012411176e-13, "grad_norm": 0.8843540549278259, "learning_rate": 4.2359782329296755e-05, "loss": 2.2662, "step": 52780 }, { "epoch": 1.831521045936313e-13, "grad_norm": 0.6721517443656921, "learning_rate": 4.2340030108257295e-05, "loss": 2.2226, "step": 52790 }, { "epoch": 1.8318679906315083e-13, "grad_norm": 0.6906976699829102, "learning_rate": 4.2320282063727255e-05, "loss": 2.2831, "step": 52800 }, { "epoch": 1.8322149353267037e-13, "grad_norm": 0.6344401836395264, "learning_rate": 4.23005381988296e-05, "loss": 2.2441, "step": 52810 }, { "epoch": 1.832561880021899e-13, "grad_norm": 0.6796155571937561, "learning_rate": 4.2280798516686626e-05, "loss": 2.2356, "step": 52820 }, { "epoch": 1.8329088247170944e-13, "grad_norm": 0.6743686199188232, "learning_rate": 4.226106302042006e-05, "loss": 2.2104, "step": 52830 }, { "epoch": 1.8332557694122897e-13, "grad_norm": 0.6812524199485779, "learning_rate": 4.224133171315082e-05, "loss": 2.2148, "step": 52840 }, { "epoch": 1.833602714107485e-13, "grad_norm": 0.6461817026138306, "learning_rate": 4.2221604597999325e-05, "loss": 2.1833, "step": 52850 }, { "epoch": 1.8339496588026805e-13, "grad_norm": 0.6616369485855103, "learning_rate": 4.220188167808515e-05, "loss": 2.256, "step": 52860 }, { "epoch": 1.8342966034978758e-13, "grad_norm": 0.6691532731056213, "learning_rate": 4.218216295652739e-05, "loss": 2.2192, "step": 52870 }, { "epoch": 1.8346435481930712e-13, "grad_norm": 0.6872954964637756, "learning_rate": 4.216244843644437e-05, "loss": 2.2515, "step": 52880 }, { "epoch": 1.8349904928882665e-13, "grad_norm": 0.6814136505126953, "learning_rate": 4.214273812095373e-05, "loss": 2.2577, "step": 52890 }, { "epoch": 1.835337437583462e-13, "grad_norm": 0.6775414347648621, "learning_rate": 4.2123032013172566e-05, "loss": 2.2862, "step": 52900 }, { "epoch": 1.8356843822786573e-13, "grad_norm": 0.663301944732666, "learning_rate": 4.2103330116217145e-05, "loss": 2.2596, "step": 52910 }, { "epoch": 1.8360313269738526e-13, "grad_norm": 0.6408085227012634, "learning_rate": 4.20836324332032e-05, "loss": 2.2261, "step": 52920 }, { "epoch": 1.836378271669048e-13, "grad_norm": 0.6594605445861816, "learning_rate": 4.206393896724575e-05, "loss": 2.2351, "step": 52930 }, { "epoch": 1.8367252163642434e-13, "grad_norm": 0.6693123579025269, "learning_rate": 4.204424972145913e-05, "loss": 2.2432, "step": 52940 }, { "epoch": 1.8370721610594387e-13, "grad_norm": 0.6582949161529541, "learning_rate": 4.202456469895703e-05, "loss": 2.2432, "step": 52950 }, { "epoch": 1.837419105754634e-13, "grad_norm": 0.6855507493019104, "learning_rate": 4.2004883902852454e-05, "loss": 2.2352, "step": 52960 }, { "epoch": 1.8377660504498294e-13, "grad_norm": 0.6637006402015686, "learning_rate": 4.198520733625778e-05, "loss": 2.2503, "step": 52970 }, { "epoch": 1.8381129951450248e-13, "grad_norm": 0.6558538675308228, "learning_rate": 4.1965535002284684e-05, "loss": 2.2524, "step": 52980 }, { "epoch": 1.8384599398402202e-13, "grad_norm": 0.681670069694519, "learning_rate": 4.194586690404416e-05, "loss": 2.2289, "step": 52990 }, { "epoch": 1.8388068845354155e-13, "grad_norm": 0.6867555379867554, "learning_rate": 4.1926203044646544e-05, "loss": 2.2363, "step": 53000 }, { "epoch": 1.839153829230611e-13, "grad_norm": 0.683379590511322, "learning_rate": 4.190654342720153e-05, "loss": 2.2392, "step": 53010 }, { "epoch": 1.8395007739258062e-13, "grad_norm": 0.6799705624580383, "learning_rate": 4.188688805481807e-05, "loss": 2.2674, "step": 53020 }, { "epoch": 1.8398477186210016e-13, "grad_norm": 0.6961343288421631, "learning_rate": 4.1867236930604555e-05, "loss": 2.2497, "step": 53030 }, { "epoch": 1.840194663316197e-13, "grad_norm": 0.6913780570030212, "learning_rate": 4.184759005766861e-05, "loss": 2.2377, "step": 53040 }, { "epoch": 1.8405416080113923e-13, "grad_norm": 0.6805413961410522, "learning_rate": 4.1827947439117216e-05, "loss": 2.2163, "step": 53050 }, { "epoch": 1.8408885527065877e-13, "grad_norm": 0.6526440382003784, "learning_rate": 4.1808309078056684e-05, "loss": 2.2268, "step": 53060 }, { "epoch": 1.841235497401783e-13, "grad_norm": 0.6619260907173157, "learning_rate": 4.1788674977592645e-05, "loss": 2.2356, "step": 53070 }, { "epoch": 1.8415824420969784e-13, "grad_norm": 0.6905510425567627, "learning_rate": 4.1769045140830106e-05, "loss": 2.1999, "step": 53080 }, { "epoch": 1.8419293867921738e-13, "grad_norm": 0.6547953486442566, "learning_rate": 4.174941957087329e-05, "loss": 2.2468, "step": 53090 }, { "epoch": 1.8422763314873691e-13, "grad_norm": 0.6584991812705994, "learning_rate": 4.172979827082586e-05, "loss": 2.2414, "step": 53100 }, { "epoch": 1.8426232761825645e-13, "grad_norm": 0.6683449149131775, "learning_rate": 4.171018124379074e-05, "loss": 2.2597, "step": 53110 }, { "epoch": 1.8429702208777599e-13, "grad_norm": 0.6804712414741516, "learning_rate": 4.16905684928702e-05, "loss": 2.2492, "step": 53120 }, { "epoch": 1.8433171655729552e-13, "grad_norm": 0.6732161045074463, "learning_rate": 4.167096002116583e-05, "loss": 2.2058, "step": 53130 }, { "epoch": 1.8436641102681506e-13, "grad_norm": 0.682375431060791, "learning_rate": 4.165135583177851e-05, "loss": 2.2076, "step": 53140 }, { "epoch": 1.844011054963346e-13, "grad_norm": 0.6725254654884338, "learning_rate": 4.1631755927808544e-05, "loss": 2.208, "step": 53150 }, { "epoch": 1.8443579996585413e-13, "grad_norm": 0.7339206337928772, "learning_rate": 4.1612160312355385e-05, "loss": 2.2286, "step": 53160 }, { "epoch": 1.8447049443537367e-13, "grad_norm": 0.6848053932189941, "learning_rate": 4.159256898851801e-05, "loss": 2.2623, "step": 53170 }, { "epoch": 1.845051889048932e-13, "grad_norm": 0.6579086780548096, "learning_rate": 4.157298195939456e-05, "loss": 2.2288, "step": 53180 }, { "epoch": 1.8453988337441274e-13, "grad_norm": 0.6636441349983215, "learning_rate": 4.155339922808258e-05, "loss": 2.2533, "step": 53190 }, { "epoch": 1.8457457784393227e-13, "grad_norm": 0.6742754578590393, "learning_rate": 4.1533820797678906e-05, "loss": 2.2038, "step": 53200 }, { "epoch": 1.846092723134518e-13, "grad_norm": 0.6775166392326355, "learning_rate": 4.1514246671279664e-05, "loss": 2.239, "step": 53210 }, { "epoch": 1.8464396678297135e-13, "grad_norm": 0.6990178227424622, "learning_rate": 4.1494676851980415e-05, "loss": 2.2418, "step": 53220 }, { "epoch": 1.8467866125249088e-13, "grad_norm": 0.6707760691642761, "learning_rate": 4.1475111342875856e-05, "loss": 2.2132, "step": 53230 }, { "epoch": 1.8471335572201042e-13, "grad_norm": 0.7394474148750305, "learning_rate": 4.145555014706018e-05, "loss": 2.2418, "step": 53240 }, { "epoch": 1.8474805019152996e-13, "grad_norm": 0.6690255999565125, "learning_rate": 4.1435993267626804e-05, "loss": 2.2446, "step": 53250 }, { "epoch": 1.847827446610495e-13, "grad_norm": 0.6677428483963013, "learning_rate": 4.141644070766845e-05, "loss": 2.2281, "step": 53260 }, { "epoch": 1.8481743913056903e-13, "grad_norm": 0.684267520904541, "learning_rate": 4.139689247027723e-05, "loss": 2.2429, "step": 53270 }, { "epoch": 1.8485213360008856e-13, "grad_norm": 0.6615341305732727, "learning_rate": 4.137734855854447e-05, "loss": 2.2249, "step": 53280 }, { "epoch": 1.848868280696081e-13, "grad_norm": 0.6463817954063416, "learning_rate": 4.135780897556094e-05, "loss": 2.2068, "step": 53290 }, { "epoch": 1.8492152253912764e-13, "grad_norm": 0.6666473746299744, "learning_rate": 4.133827372441663e-05, "loss": 2.293, "step": 53300 }, { "epoch": 1.8495621700864717e-13, "grad_norm": 0.6830403208732605, "learning_rate": 4.1318742808200874e-05, "loss": 2.207, "step": 53310 }, { "epoch": 1.849909114781667e-13, "grad_norm": 0.659822404384613, "learning_rate": 4.129921623000229e-05, "loss": 2.2275, "step": 53320 }, { "epoch": 1.8502560594768624e-13, "grad_norm": 0.668125569820404, "learning_rate": 4.127969399290889e-05, "loss": 2.2421, "step": 53330 }, { "epoch": 1.8506030041720578e-13, "grad_norm": 0.675563633441925, "learning_rate": 4.126017610000788e-05, "loss": 2.203, "step": 53340 }, { "epoch": 1.8509499488672532e-13, "grad_norm": 0.6547757387161255, "learning_rate": 4.124066255438591e-05, "loss": 2.2138, "step": 53350 }, { "epoch": 1.8512968935624485e-13, "grad_norm": 0.6494236588478088, "learning_rate": 4.122115335912886e-05, "loss": 2.2362, "step": 53360 }, { "epoch": 1.851643838257644e-13, "grad_norm": 0.6666968464851379, "learning_rate": 4.120164851732194e-05, "loss": 2.2433, "step": 53370 }, { "epoch": 1.8519907829528393e-13, "grad_norm": 0.6806256771087646, "learning_rate": 4.118214803204966e-05, "loss": 2.2363, "step": 53380 }, { "epoch": 1.8523377276480346e-13, "grad_norm": 0.6941131353378296, "learning_rate": 4.1162651906395846e-05, "loss": 2.2543, "step": 53390 }, { "epoch": 1.85268467234323e-13, "grad_norm": 0.6878867149353027, "learning_rate": 4.1143160143443695e-05, "loss": 2.2582, "step": 53400 }, { "epoch": 1.8530316170384253e-13, "grad_norm": 0.6790107488632202, "learning_rate": 4.112367274627559e-05, "loss": 2.2212, "step": 53410 }, { "epoch": 1.8533785617336207e-13, "grad_norm": 0.6775779724121094, "learning_rate": 4.1104189717973344e-05, "loss": 2.208, "step": 53420 }, { "epoch": 1.853725506428816e-13, "grad_norm": 0.6672387719154358, "learning_rate": 4.108471106161803e-05, "loss": 2.3018, "step": 53430 }, { "epoch": 1.8540724511240114e-13, "grad_norm": 0.6413131952285767, "learning_rate": 4.106523678029001e-05, "loss": 2.2048, "step": 53440 }, { "epoch": 1.8544193958192068e-13, "grad_norm": 0.6456266045570374, "learning_rate": 4.104576687706898e-05, "loss": 2.2392, "step": 53450 }, { "epoch": 1.8547663405144021e-13, "grad_norm": 0.6569863557815552, "learning_rate": 4.102630135503392e-05, "loss": 2.2157, "step": 53460 }, { "epoch": 1.8551132852095975e-13, "grad_norm": 0.6529606580734253, "learning_rate": 4.1006840217263203e-05, "loss": 2.2124, "step": 53470 }, { "epoch": 1.8554602299047929e-13, "grad_norm": 0.699000358581543, "learning_rate": 4.0987383466834335e-05, "loss": 2.2237, "step": 53480 }, { "epoch": 1.8558071745999882e-13, "grad_norm": 0.6972898840904236, "learning_rate": 4.0967931106824315e-05, "loss": 2.2508, "step": 53490 }, { "epoch": 1.8561541192951836e-13, "grad_norm": 0.668679416179657, "learning_rate": 4.0948483140309344e-05, "loss": 2.2117, "step": 53500 }, { "epoch": 1.856501063990379e-13, "grad_norm": 0.6603649258613586, "learning_rate": 4.092903957036495e-05, "loss": 2.2256, "step": 53510 }, { "epoch": 1.8568480086855743e-13, "grad_norm": 0.6724984049797058, "learning_rate": 4.090960040006596e-05, "loss": 2.2445, "step": 53520 }, { "epoch": 1.8571949533807697e-13, "grad_norm": 0.6637265682220459, "learning_rate": 4.089016563248649e-05, "loss": 2.2713, "step": 53530 }, { "epoch": 1.857541898075965e-13, "grad_norm": 0.6875990033149719, "learning_rate": 4.087073527070005e-05, "loss": 2.2342, "step": 53540 }, { "epoch": 1.8578888427711604e-13, "grad_norm": 0.6663734316825867, "learning_rate": 4.08513093177793e-05, "loss": 2.2224, "step": 53550 }, { "epoch": 1.8582357874663558e-13, "grad_norm": 0.6859173774719238, "learning_rate": 4.083188777679635e-05, "loss": 2.2486, "step": 53560 }, { "epoch": 1.858582732161551e-13, "grad_norm": 0.6870951652526855, "learning_rate": 4.081247065082252e-05, "loss": 2.2333, "step": 53570 }, { "epoch": 1.8589296768567465e-13, "grad_norm": 0.7484785914421082, "learning_rate": 4.079305794292848e-05, "loss": 2.2456, "step": 53580 }, { "epoch": 1.8592766215519418e-13, "grad_norm": 0.6802883148193359, "learning_rate": 4.077364965618417e-05, "loss": 2.2572, "step": 53590 }, { "epoch": 1.8596235662471372e-13, "grad_norm": 0.6897130012512207, "learning_rate": 4.075424579365881e-05, "loss": 2.2274, "step": 53600 }, { "epoch": 1.8599705109423326e-13, "grad_norm": 0.6887518763542175, "learning_rate": 4.073484635842103e-05, "loss": 2.2236, "step": 53610 }, { "epoch": 1.860317455637528e-13, "grad_norm": 0.6649071574211121, "learning_rate": 4.071545135353862e-05, "loss": 2.2522, "step": 53620 }, { "epoch": 1.8606644003327233e-13, "grad_norm": 0.6657181978225708, "learning_rate": 4.069606078207878e-05, "loss": 2.2483, "step": 53630 }, { "epoch": 1.8610113450279187e-13, "grad_norm": 0.6690980195999146, "learning_rate": 4.067667464710792e-05, "loss": 2.2004, "step": 53640 }, { "epoch": 1.861358289723114e-13, "grad_norm": 0.6650794148445129, "learning_rate": 4.065729295169181e-05, "loss": 2.21, "step": 53650 }, { "epoch": 1.8617052344183094e-13, "grad_norm": 0.6997435092926025, "learning_rate": 4.063791569889548e-05, "loss": 2.2196, "step": 53660 }, { "epoch": 1.8620521791135047e-13, "grad_norm": 0.6581323146820068, "learning_rate": 4.061854289178331e-05, "loss": 2.2322, "step": 53670 }, { "epoch": 1.8623991238087e-13, "grad_norm": 0.6998741030693054, "learning_rate": 4.059917453341892e-05, "loss": 2.2204, "step": 53680 }, { "epoch": 1.8627460685038955e-13, "grad_norm": 0.704800009727478, "learning_rate": 4.057981062686526e-05, "loss": 2.2459, "step": 53690 }, { "epoch": 1.8630930131990908e-13, "grad_norm": 0.6928750872612, "learning_rate": 4.056045117518456e-05, "loss": 2.2577, "step": 53700 }, { "epoch": 1.8634399578942862e-13, "grad_norm": 0.6920477747917175, "learning_rate": 4.054109618143832e-05, "loss": 2.199, "step": 53710 }, { "epoch": 1.8637869025894815e-13, "grad_norm": 0.6596131920814514, "learning_rate": 4.052174564868745e-05, "loss": 2.1977, "step": 53720 }, { "epoch": 1.864133847284677e-13, "grad_norm": 0.6948193311691284, "learning_rate": 4.050239957999196e-05, "loss": 2.2748, "step": 53730 }, { "epoch": 1.8644807919798723e-13, "grad_norm": 0.7349778413772583, "learning_rate": 4.048305797841134e-05, "loss": 2.2377, "step": 53740 }, { "epoch": 1.8648277366750676e-13, "grad_norm": 0.7100182771682739, "learning_rate": 4.046372084700429e-05, "loss": 2.2052, "step": 53750 }, { "epoch": 1.865174681370263e-13, "grad_norm": 0.6498998999595642, "learning_rate": 4.044438818882879e-05, "loss": 2.241, "step": 53760 }, { "epoch": 1.8655216260654583e-13, "grad_norm": 0.6804719567298889, "learning_rate": 4.0425060006942144e-05, "loss": 2.2526, "step": 53770 }, { "epoch": 1.8658685707606537e-13, "grad_norm": 0.6874762177467346, "learning_rate": 4.0405736304400904e-05, "loss": 2.2318, "step": 53780 }, { "epoch": 1.866215515455849e-13, "grad_norm": 0.6268221735954285, "learning_rate": 4.038641708426104e-05, "loss": 2.2426, "step": 53790 }, { "epoch": 1.8665624601510444e-13, "grad_norm": 0.6612706184387207, "learning_rate": 4.03671023495776e-05, "loss": 2.2309, "step": 53800 }, { "epoch": 1.8669094048462398e-13, "grad_norm": 0.6997544765472412, "learning_rate": 4.034779210340512e-05, "loss": 2.2488, "step": 53810 }, { "epoch": 1.8672563495414352e-13, "grad_norm": 0.7011980414390564, "learning_rate": 4.032848634879735e-05, "loss": 2.2391, "step": 53820 }, { "epoch": 1.8676032942366305e-13, "grad_norm": 0.6804572343826294, "learning_rate": 4.03091850888073e-05, "loss": 2.1993, "step": 53830 }, { "epoch": 1.867950238931826e-13, "grad_norm": 0.7133411169052124, "learning_rate": 4.028988832648731e-05, "loss": 2.2311, "step": 53840 }, { "epoch": 1.8682971836270212e-13, "grad_norm": 0.6570286154747009, "learning_rate": 4.027059606488896e-05, "loss": 2.2689, "step": 53850 }, { "epoch": 1.8686441283222166e-13, "grad_norm": 0.6658950448036194, "learning_rate": 4.025130830706325e-05, "loss": 2.235, "step": 53860 }, { "epoch": 1.868991073017412e-13, "grad_norm": 0.6551535129547119, "learning_rate": 4.0232025056060264e-05, "loss": 2.2367, "step": 53870 }, { "epoch": 1.8693380177126073e-13, "grad_norm": 0.6769562363624573, "learning_rate": 4.021274631492955e-05, "loss": 2.2394, "step": 53880 }, { "epoch": 1.8696849624078027e-13, "grad_norm": 0.635049045085907, "learning_rate": 4.0193472086719854e-05, "loss": 2.2182, "step": 53890 }, { "epoch": 1.870031907102998e-13, "grad_norm": 0.6888550519943237, "learning_rate": 4.017420237447924e-05, "loss": 2.2861, "step": 53900 }, { "epoch": 1.8703788517981934e-13, "grad_norm": 0.7357144951820374, "learning_rate": 4.0154937181255034e-05, "loss": 2.2031, "step": 53910 }, { "epoch": 1.8707257964933888e-13, "grad_norm": 0.6855191588401794, "learning_rate": 4.0135676510093836e-05, "loss": 2.247, "step": 53920 }, { "epoch": 1.8710727411885841e-13, "grad_norm": 0.7095708847045898, "learning_rate": 4.011642036404162e-05, "loss": 2.2201, "step": 53930 }, { "epoch": 1.8714196858837795e-13, "grad_norm": 0.6655850410461426, "learning_rate": 4.009716874614352e-05, "loss": 2.2249, "step": 53940 }, { "epoch": 1.8717666305789749e-13, "grad_norm": 0.6758877635002136, "learning_rate": 4.007792165944407e-05, "loss": 2.1935, "step": 53950 }, { "epoch": 1.8721135752741702e-13, "grad_norm": 0.6658623814582825, "learning_rate": 4.0058679106986976e-05, "loss": 2.1958, "step": 53960 }, { "epoch": 1.8724605199693656e-13, "grad_norm": 0.6915726065635681, "learning_rate": 4.0039441091815316e-05, "loss": 2.2256, "step": 53970 }, { "epoch": 1.872807464664561e-13, "grad_norm": 0.6911454200744629, "learning_rate": 4.002020761697138e-05, "loss": 2.2655, "step": 53980 }, { "epoch": 1.8731544093597563e-13, "grad_norm": 0.6806970834732056, "learning_rate": 4.000097868549682e-05, "loss": 2.2388, "step": 53990 }, { "epoch": 1.8735013540549517e-13, "grad_norm": 0.6900203824043274, "learning_rate": 3.998175430043251e-05, "loss": 2.2438, "step": 54000 }, { "epoch": 1.873848298750147e-13, "grad_norm": 0.691048264503479, "learning_rate": 3.996253446481861e-05, "loss": 2.227, "step": 54010 }, { "epoch": 1.8741952434453424e-13, "grad_norm": 0.6828586459159851, "learning_rate": 3.99433191816946e-05, "loss": 2.272, "step": 54020 }, { "epoch": 1.8745421881405377e-13, "grad_norm": 0.6876731514930725, "learning_rate": 3.992410845409915e-05, "loss": 2.2635, "step": 54030 }, { "epoch": 1.874889132835733e-13, "grad_norm": 0.6913195252418518, "learning_rate": 3.9904902285070365e-05, "loss": 2.2128, "step": 54040 }, { "epoch": 1.8752360775309285e-13, "grad_norm": 0.6832589507102966, "learning_rate": 3.988570067764544e-05, "loss": 2.2585, "step": 54050 }, { "epoch": 1.8755830222261238e-13, "grad_norm": 0.7184513211250305, "learning_rate": 3.986650363486101e-05, "loss": 2.2221, "step": 54060 }, { "epoch": 1.8759299669213192e-13, "grad_norm": 0.6988794207572937, "learning_rate": 3.98473111597529e-05, "loss": 2.264, "step": 54070 }, { "epoch": 1.8762769116165146e-13, "grad_norm": 0.6800140142440796, "learning_rate": 3.9828123255356204e-05, "loss": 2.2237, "step": 54080 }, { "epoch": 1.87662385631171e-13, "grad_norm": 0.6869401931762695, "learning_rate": 3.980893992470541e-05, "loss": 2.2466, "step": 54090 }, { "epoch": 1.8769708010069053e-13, "grad_norm": 0.6755515336990356, "learning_rate": 3.978976117083409e-05, "loss": 2.2275, "step": 54100 }, { "epoch": 1.8773177457021006e-13, "grad_norm": 0.730562150478363, "learning_rate": 3.9770586996775284e-05, "loss": 2.263, "step": 54110 }, { "epoch": 1.877664690397296e-13, "grad_norm": 0.6805828809738159, "learning_rate": 3.975141740556115e-05, "loss": 2.2328, "step": 54120 }, { "epoch": 1.8780116350924914e-13, "grad_norm": 0.6658015251159668, "learning_rate": 3.9732252400223265e-05, "loss": 2.2418, "step": 54130 }, { "epoch": 1.8783585797876867e-13, "grad_norm": 0.6988531947135925, "learning_rate": 3.9713091983792366e-05, "loss": 2.258, "step": 54140 }, { "epoch": 1.878705524482882e-13, "grad_norm": 0.7102271318435669, "learning_rate": 3.969393615929852e-05, "loss": 2.248, "step": 54150 }, { "epoch": 1.8790524691780774e-13, "grad_norm": 0.6842500567436218, "learning_rate": 3.9674784929771056e-05, "loss": 2.2024, "step": 54160 }, { "epoch": 1.8793994138732728e-13, "grad_norm": 0.6973617076873779, "learning_rate": 3.965563829823855e-05, "loss": 2.2127, "step": 54170 }, { "epoch": 1.8797463585684682e-13, "grad_norm": 0.6811538338661194, "learning_rate": 3.9636496267728956e-05, "loss": 2.2253, "step": 54180 }, { "epoch": 1.8800933032636635e-13, "grad_norm": 0.6912691593170166, "learning_rate": 3.961735884126931e-05, "loss": 2.1856, "step": 54190 }, { "epoch": 1.880440247958859e-13, "grad_norm": 0.6983669400215149, "learning_rate": 3.959822602188611e-05, "loss": 2.2277, "step": 54200 }, { "epoch": 1.8807871926540543e-13, "grad_norm": 0.6533454060554504, "learning_rate": 3.9579097812605036e-05, "loss": 2.2092, "step": 54210 }, { "epoch": 1.8811341373492496e-13, "grad_norm": 0.6453465819358826, "learning_rate": 3.955997421645102e-05, "loss": 2.1983, "step": 54220 }, { "epoch": 1.881481082044445e-13, "grad_norm": 0.6427932977676392, "learning_rate": 3.954085523644832e-05, "loss": 2.2489, "step": 54230 }, { "epoch": 1.8818280267396403e-13, "grad_norm": 0.6938084959983826, "learning_rate": 3.952174087562041e-05, "loss": 2.2242, "step": 54240 }, { "epoch": 1.8821749714348357e-13, "grad_norm": 0.6561230421066284, "learning_rate": 3.950263113699011e-05, "loss": 2.2602, "step": 54250 }, { "epoch": 1.882521916130031e-13, "grad_norm": 0.6826820969581604, "learning_rate": 3.948352602357942e-05, "loss": 2.2188, "step": 54260 }, { "epoch": 1.8828688608252264e-13, "grad_norm": 0.6713894605636597, "learning_rate": 3.946442553840967e-05, "loss": 2.2236, "step": 54270 }, { "epoch": 1.8832158055204218e-13, "grad_norm": 0.6855889558792114, "learning_rate": 3.944532968450143e-05, "loss": 2.2479, "step": 54280 }, { "epoch": 1.8835627502156171e-13, "grad_norm": 0.6865391731262207, "learning_rate": 3.942623846487454e-05, "loss": 2.2409, "step": 54290 }, { "epoch": 1.8839096949108125e-13, "grad_norm": 0.7017579078674316, "learning_rate": 3.94071518825481e-05, "loss": 2.2381, "step": 54300 }, { "epoch": 1.8842566396060079e-13, "grad_norm": 0.7023707032203674, "learning_rate": 3.938806994054052e-05, "loss": 2.2171, "step": 54310 }, { "epoch": 1.8846035843012032e-13, "grad_norm": 0.6683211326599121, "learning_rate": 3.936899264186943e-05, "loss": 2.2272, "step": 54320 }, { "epoch": 1.8849505289963986e-13, "grad_norm": 0.6610502004623413, "learning_rate": 3.934991998955174e-05, "loss": 2.2439, "step": 54330 }, { "epoch": 1.885297473691594e-13, "grad_norm": 0.6826403141021729, "learning_rate": 3.9330851986603634e-05, "loss": 2.2136, "step": 54340 }, { "epoch": 1.8856444183867893e-13, "grad_norm": 0.6905397772789001, "learning_rate": 3.9311788636040515e-05, "loss": 2.2232, "step": 54350 }, { "epoch": 1.8859913630819847e-13, "grad_norm": 0.6722270846366882, "learning_rate": 3.929272994087716e-05, "loss": 2.2332, "step": 54360 }, { "epoch": 1.88633830777718e-13, "grad_norm": 0.6504330039024353, "learning_rate": 3.9273675904127455e-05, "loss": 2.2151, "step": 54370 }, { "epoch": 1.8866852524723754e-13, "grad_norm": 0.6707262992858887, "learning_rate": 3.9254626528804695e-05, "loss": 2.2341, "step": 54380 }, { "epoch": 1.8870321971675708e-13, "grad_norm": 0.6691398024559021, "learning_rate": 3.923558181792135e-05, "loss": 2.1784, "step": 54390 }, { "epoch": 1.887379141862766e-13, "grad_norm": 0.6917273998260498, "learning_rate": 3.9216541774489155e-05, "loss": 2.2563, "step": 54400 }, { "epoch": 1.8877260865579615e-13, "grad_norm": 0.6944751143455505, "learning_rate": 3.919750640151919e-05, "loss": 2.2265, "step": 54410 }, { "epoch": 1.8880730312531568e-13, "grad_norm": 0.6798033118247986, "learning_rate": 3.917847570202167e-05, "loss": 2.2576, "step": 54420 }, { "epoch": 1.8884199759483522e-13, "grad_norm": 0.6981064081192017, "learning_rate": 3.91594496790062e-05, "loss": 2.2476, "step": 54430 }, { "epoch": 1.8887669206435476e-13, "grad_norm": 0.6621366739273071, "learning_rate": 3.914042833548149e-05, "loss": 2.1949, "step": 54440 }, { "epoch": 1.889113865338743e-13, "grad_norm": 0.7056812644004822, "learning_rate": 3.91214116744557e-05, "loss": 2.1959, "step": 54450 }, { "epoch": 1.8894608100339383e-13, "grad_norm": 0.6803156733512878, "learning_rate": 3.91023996989361e-05, "loss": 2.2519, "step": 54460 }, { "epoch": 1.8898077547291336e-13, "grad_norm": 0.6627170443534851, "learning_rate": 3.9083392411929263e-05, "loss": 2.2398, "step": 54470 }, { "epoch": 1.890154699424329e-13, "grad_norm": 0.6774154305458069, "learning_rate": 3.906438981644108e-05, "loss": 2.2657, "step": 54480 }, { "epoch": 1.8905016441195244e-13, "grad_norm": 0.6686736345291138, "learning_rate": 3.904539191547657e-05, "loss": 2.2072, "step": 54490 }, { "epoch": 1.8908485888147197e-13, "grad_norm": 0.6739482283592224, "learning_rate": 3.902639871204018e-05, "loss": 2.2664, "step": 54500 }, { "epoch": 1.891195533509915e-13, "grad_norm": 0.6401817798614502, "learning_rate": 3.900741020913542e-05, "loss": 2.2165, "step": 54510 }, { "epoch": 1.8915424782051105e-13, "grad_norm": 0.6651803255081177, "learning_rate": 3.898842640976523e-05, "loss": 2.2657, "step": 54520 }, { "epoch": 1.8918894229003058e-13, "grad_norm": 0.693337619304657, "learning_rate": 3.8969447316931724e-05, "loss": 2.2201, "step": 54530 }, { "epoch": 1.8922363675955012e-13, "grad_norm": 0.6979964375495911, "learning_rate": 3.895047293363626e-05, "loss": 2.2056, "step": 54540 }, { "epoch": 1.8925833122906965e-13, "grad_norm": 0.7216173410415649, "learning_rate": 3.893150326287949e-05, "loss": 2.2365, "step": 54550 }, { "epoch": 1.892930256985892e-13, "grad_norm": 0.6855332851409912, "learning_rate": 3.891253830766128e-05, "loss": 2.2205, "step": 54560 }, { "epoch": 1.8932772016810873e-13, "grad_norm": 0.6599966883659363, "learning_rate": 3.8893578070980815e-05, "loss": 2.2259, "step": 54570 }, { "epoch": 1.8936241463762826e-13, "grad_norm": 0.668992280960083, "learning_rate": 3.887462255583647e-05, "loss": 2.2361, "step": 54580 }, { "epoch": 1.893971091071478e-13, "grad_norm": 0.6688276529312134, "learning_rate": 3.88556717652259e-05, "loss": 2.269, "step": 54590 }, { "epoch": 1.8943180357666733e-13, "grad_norm": 0.6702283620834351, "learning_rate": 3.883672570214601e-05, "loss": 2.2378, "step": 54600 }, { "epoch": 1.8946649804618687e-13, "grad_norm": 0.66355961561203, "learning_rate": 3.8817784369592956e-05, "loss": 2.2282, "step": 54610 }, { "epoch": 1.895011925157064e-13, "grad_norm": 0.6786127090454102, "learning_rate": 3.879884777056212e-05, "loss": 2.2309, "step": 54620 }, { "epoch": 1.8953588698522594e-13, "grad_norm": 0.6682628989219666, "learning_rate": 3.8779915908048206e-05, "loss": 2.2274, "step": 54630 }, { "epoch": 1.8957058145474548e-13, "grad_norm": 0.6826069951057434, "learning_rate": 3.8760988785045115e-05, "loss": 2.2536, "step": 54640 }, { "epoch": 1.8960527592426502e-13, "grad_norm": 0.6779120564460754, "learning_rate": 3.8742066404546e-05, "loss": 2.2469, "step": 54650 }, { "epoch": 1.8963997039378455e-13, "grad_norm": 0.6847909688949585, "learning_rate": 3.872314876954326e-05, "loss": 2.2325, "step": 54660 }, { "epoch": 1.896746648633041e-13, "grad_norm": 0.7028515338897705, "learning_rate": 3.870423588302857e-05, "loss": 2.2308, "step": 54670 }, { "epoch": 1.8970935933282362e-13, "grad_norm": 0.6758776903152466, "learning_rate": 3.868532774799287e-05, "loss": 2.2109, "step": 54680 }, { "epoch": 1.8974405380234316e-13, "grad_norm": 0.6523134708404541, "learning_rate": 3.866642436742625e-05, "loss": 2.2333, "step": 54690 }, { "epoch": 1.897787482718627e-13, "grad_norm": 0.6914126873016357, "learning_rate": 3.864752574431817e-05, "loss": 2.2326, "step": 54700 }, { "epoch": 1.8981344274138223e-13, "grad_norm": 0.6673750877380371, "learning_rate": 3.862863188165727e-05, "loss": 2.245, "step": 54710 }, { "epoch": 1.8984813721090177e-13, "grad_norm": 0.7268213033676147, "learning_rate": 3.860974278243143e-05, "loss": 2.1853, "step": 54720 }, { "epoch": 1.898828316804213e-13, "grad_norm": 0.6664856672286987, "learning_rate": 3.8590858449627865e-05, "loss": 2.2417, "step": 54730 }, { "epoch": 1.8991752614994084e-13, "grad_norm": 0.6651062965393066, "learning_rate": 3.857197888623289e-05, "loss": 2.2313, "step": 54740 }, { "epoch": 1.8995222061946038e-13, "grad_norm": 0.6597408056259155, "learning_rate": 3.8553104095232205e-05, "loss": 2.248, "step": 54750 }, { "epoch": 1.8998691508897991e-13, "grad_norm": 0.6736698746681213, "learning_rate": 3.853423407961064e-05, "loss": 2.2346, "step": 54760 }, { "epoch": 1.9002160955849945e-13, "grad_norm": 0.6879186034202576, "learning_rate": 3.851536884235238e-05, "loss": 2.2109, "step": 54770 }, { "epoch": 1.9005630402801899e-13, "grad_norm": 0.6827185750007629, "learning_rate": 3.8496508386440766e-05, "loss": 2.2126, "step": 54780 }, { "epoch": 1.9009099849753852e-13, "grad_norm": 0.660140872001648, "learning_rate": 3.847765271485841e-05, "loss": 2.234, "step": 54790 }, { "epoch": 1.9012569296705806e-13, "grad_norm": 0.682762086391449, "learning_rate": 3.845880183058723e-05, "loss": 2.2147, "step": 54800 }, { "epoch": 1.901603874365776e-13, "grad_norm": 0.7029463052749634, "learning_rate": 3.8439955736608245e-05, "loss": 2.2144, "step": 54810 }, { "epoch": 1.9019508190609713e-13, "grad_norm": 0.689655065536499, "learning_rate": 3.842111443590191e-05, "loss": 2.2291, "step": 54820 }, { "epoch": 1.9022977637561667e-13, "grad_norm": 0.6905373930931091, "learning_rate": 3.8402277931447695e-05, "loss": 2.2191, "step": 54830 }, { "epoch": 1.902644708451362e-13, "grad_norm": 0.6781824827194214, "learning_rate": 3.8383446226224524e-05, "loss": 2.2416, "step": 54840 }, { "epoch": 1.9029916531465574e-13, "grad_norm": 0.6831398010253906, "learning_rate": 3.836461932321042e-05, "loss": 2.2625, "step": 54850 }, { "epoch": 1.9033385978417527e-13, "grad_norm": 0.6737621426582336, "learning_rate": 3.834579722538273e-05, "loss": 2.2247, "step": 54860 }, { "epoch": 1.903685542536948e-13, "grad_norm": 0.6816014051437378, "learning_rate": 3.832697993571797e-05, "loss": 2.2335, "step": 54870 }, { "epoch": 1.9040324872321435e-13, "grad_norm": 0.7351601123809814, "learning_rate": 3.830816745719194e-05, "loss": 2.2324, "step": 54880 }, { "epoch": 1.9043794319273388e-13, "grad_norm": 0.6823431849479675, "learning_rate": 3.82893597927797e-05, "loss": 2.2467, "step": 54890 }, { "epoch": 1.9047263766225342e-13, "grad_norm": 0.6838575601577759, "learning_rate": 3.8270556945455505e-05, "loss": 2.2386, "step": 54900 }, { "epoch": 1.9050733213177296e-13, "grad_norm": 0.6789243817329407, "learning_rate": 3.825175891819285e-05, "loss": 2.2129, "step": 54910 }, { "epoch": 1.905420266012925e-13, "grad_norm": 0.6695847511291504, "learning_rate": 3.823296571396451e-05, "loss": 2.2134, "step": 54920 }, { "epoch": 1.9057672107081203e-13, "grad_norm": 0.6725510358810425, "learning_rate": 3.821417733574243e-05, "loss": 2.2129, "step": 54930 }, { "epoch": 1.9061141554033156e-13, "grad_norm": 0.7058640122413635, "learning_rate": 3.819539378649783e-05, "loss": 2.2491, "step": 54940 }, { "epoch": 1.906461100098511e-13, "grad_norm": 0.6755629777908325, "learning_rate": 3.817661506920123e-05, "loss": 2.2606, "step": 54950 }, { "epoch": 1.9068080447937064e-13, "grad_norm": 0.690696656703949, "learning_rate": 3.815784118682226e-05, "loss": 2.2233, "step": 54960 }, { "epoch": 1.9071549894889017e-13, "grad_norm": 0.6531834602355957, "learning_rate": 3.8139072142329866e-05, "loss": 2.2188, "step": 54970 }, { "epoch": 1.907501934184097e-13, "grad_norm": 0.686489999294281, "learning_rate": 3.812030793869221e-05, "loss": 2.2417, "step": 54980 }, { "epoch": 1.9078488788792924e-13, "grad_norm": 0.6516639590263367, "learning_rate": 3.810154857887668e-05, "loss": 2.2121, "step": 54990 }, { "epoch": 1.9081958235744878e-13, "grad_norm": 0.6701606512069702, "learning_rate": 3.8082794065849965e-05, "loss": 2.2088, "step": 55000 }, { "epoch": 1.9085427682696832e-13, "grad_norm": 0.6619659662246704, "learning_rate": 3.806404440257783e-05, "loss": 2.2114, "step": 55010 }, { "epoch": 1.9088897129648785e-13, "grad_norm": 0.685746967792511, "learning_rate": 3.8045299592025456e-05, "loss": 2.2494, "step": 55020 }, { "epoch": 1.909236657660074e-13, "grad_norm": 0.6591439843177795, "learning_rate": 3.8026559637157144e-05, "loss": 2.2416, "step": 55030 }, { "epoch": 1.9095836023552692e-13, "grad_norm": 0.6904867887496948, "learning_rate": 3.8007824540936434e-05, "loss": 2.2104, "step": 55040 }, { "epoch": 1.9099305470504646e-13, "grad_norm": 0.6601822972297668, "learning_rate": 3.7989094306326174e-05, "loss": 2.2553, "step": 55050 }, { "epoch": 1.91027749174566e-13, "grad_norm": 0.7028012275695801, "learning_rate": 3.7970368936288316e-05, "loss": 2.2167, "step": 55060 }, { "epoch": 1.9106244364408553e-13, "grad_norm": 0.6863738298416138, "learning_rate": 3.795164843378421e-05, "loss": 2.2486, "step": 55070 }, { "epoch": 1.9109713811360507e-13, "grad_norm": 0.6639660000801086, "learning_rate": 3.793293280177425e-05, "loss": 2.2492, "step": 55080 }, { "epoch": 1.911318325831246e-13, "grad_norm": 0.6527723073959351, "learning_rate": 3.7914222043218207e-05, "loss": 2.2263, "step": 55090 }, { "epoch": 1.9116652705264414e-13, "grad_norm": 0.66157466173172, "learning_rate": 3.7895516161075004e-05, "loss": 2.2246, "step": 55100 }, { "epoch": 1.9120122152216368e-13, "grad_norm": 0.6808109283447266, "learning_rate": 3.787681515830282e-05, "loss": 2.2515, "step": 55110 }, { "epoch": 1.9123591599168321e-13, "grad_norm": 0.6780797839164734, "learning_rate": 3.7858119037859094e-05, "loss": 2.1954, "step": 55120 }, { "epoch": 1.9127061046120275e-13, "grad_norm": 0.695754885673523, "learning_rate": 3.783942780270038e-05, "loss": 2.2009, "step": 55130 }, { "epoch": 1.9130530493072229e-13, "grad_norm": 0.6614218354225159, "learning_rate": 3.7820741455782595e-05, "loss": 2.2179, "step": 55140 }, { "epoch": 1.9133999940024182e-13, "grad_norm": 0.6865163445472717, "learning_rate": 3.780206000006081e-05, "loss": 2.2423, "step": 55150 }, { "epoch": 1.9137469386976136e-13, "grad_norm": 0.6763449311256409, "learning_rate": 3.778338343848934e-05, "loss": 2.2144, "step": 55160 }, { "epoch": 1.914093883392809e-13, "grad_norm": 0.6696048974990845, "learning_rate": 3.7764711774021714e-05, "loss": 2.2278, "step": 55170 }, { "epoch": 1.9144408280880043e-13, "grad_norm": 0.7182686924934387, "learning_rate": 3.774604500961068e-05, "loss": 2.1931, "step": 55180 }, { "epoch": 1.9147877727831997e-13, "grad_norm": 0.720092236995697, "learning_rate": 3.772738314820828e-05, "loss": 2.2438, "step": 55190 }, { "epoch": 1.915134717478395e-13, "grad_norm": 0.6932389140129089, "learning_rate": 3.770872619276565e-05, "loss": 2.2152, "step": 55200 }, { "epoch": 1.9154816621735904e-13, "grad_norm": 0.6770545840263367, "learning_rate": 3.7690074146233295e-05, "loss": 2.2295, "step": 55210 }, { "epoch": 1.9158286068687858e-13, "grad_norm": 0.6751441359519958, "learning_rate": 3.767142701156084e-05, "loss": 2.2311, "step": 55220 }, { "epoch": 1.916175551563981e-13, "grad_norm": 0.6667707562446594, "learning_rate": 3.765278479169718e-05, "loss": 2.2206, "step": 55230 }, { "epoch": 1.9165224962591765e-13, "grad_norm": 0.671670138835907, "learning_rate": 3.763414748959042e-05, "loss": 2.2432, "step": 55240 }, { "epoch": 1.9168694409543718e-13, "grad_norm": 0.6498752236366272, "learning_rate": 3.7615515108187887e-05, "loss": 2.2457, "step": 55250 }, { "epoch": 1.9172163856495672e-13, "grad_norm": 0.6825661659240723, "learning_rate": 3.75968876504361e-05, "loss": 2.2256, "step": 55260 }, { "epoch": 1.9175633303447626e-13, "grad_norm": 0.7013789415359497, "learning_rate": 3.75782651192809e-05, "loss": 2.2537, "step": 55270 }, { "epoch": 1.917910275039958e-13, "grad_norm": 0.6515285968780518, "learning_rate": 3.755964751766723e-05, "loss": 2.2058, "step": 55280 }, { "epoch": 1.9182572197351533e-13, "grad_norm": 0.6900692582130432, "learning_rate": 3.754103484853933e-05, "loss": 2.2145, "step": 55290 }, { "epoch": 1.9186041644303486e-13, "grad_norm": 0.6937234997749329, "learning_rate": 3.752242711484061e-05, "loss": 2.2077, "step": 55300 }, { "epoch": 1.918951109125544e-13, "grad_norm": 0.6713269352912903, "learning_rate": 3.750382431951372e-05, "loss": 2.2179, "step": 55310 }, { "epoch": 1.9192980538207394e-13, "grad_norm": 0.6722569465637207, "learning_rate": 3.7485226465500586e-05, "loss": 2.2281, "step": 55320 }, { "epoch": 1.9196449985159347e-13, "grad_norm": 0.7039327621459961, "learning_rate": 3.746663355574221e-05, "loss": 2.2114, "step": 55330 }, { "epoch": 1.91999194321113e-13, "grad_norm": 0.7098692655563354, "learning_rate": 3.7448045593178976e-05, "loss": 2.2604, "step": 55340 }, { "epoch": 1.9203388879063255e-13, "grad_norm": 0.6752566695213318, "learning_rate": 3.7429462580750374e-05, "loss": 2.2061, "step": 55350 }, { "epoch": 1.9206858326015208e-13, "grad_norm": 0.6793072819709778, "learning_rate": 3.741088452139514e-05, "loss": 2.2455, "step": 55360 }, { "epoch": 1.9210327772967162e-13, "grad_norm": 0.6850199699401855, "learning_rate": 3.739231141805129e-05, "loss": 2.2278, "step": 55370 }, { "epoch": 1.9213797219919115e-13, "grad_norm": 0.6880426406860352, "learning_rate": 3.737374327365592e-05, "loss": 2.2086, "step": 55380 }, { "epoch": 1.921726666687107e-13, "grad_norm": 0.6772320866584778, "learning_rate": 3.7355180091145505e-05, "loss": 2.2215, "step": 55390 }, { "epoch": 1.9220736113823023e-13, "grad_norm": 0.6555379033088684, "learning_rate": 3.733662187345556e-05, "loss": 2.2366, "step": 55400 }, { "epoch": 1.9224205560774976e-13, "grad_norm": 0.7007245421409607, "learning_rate": 3.731806862352098e-05, "loss": 2.2376, "step": 55410 }, { "epoch": 1.922767500772693e-13, "grad_norm": 0.6798804402351379, "learning_rate": 3.7299520344275795e-05, "loss": 2.2248, "step": 55420 }, { "epoch": 1.9231144454678883e-13, "grad_norm": 0.6837450265884399, "learning_rate": 3.72809770386532e-05, "loss": 2.2208, "step": 55430 }, { "epoch": 1.9234613901630837e-13, "grad_norm": 0.6677340269088745, "learning_rate": 3.726243870958576e-05, "loss": 2.2526, "step": 55440 }, { "epoch": 1.923808334858279e-13, "grad_norm": 0.690360963344574, "learning_rate": 3.724390536000503e-05, "loss": 2.2274, "step": 55450 }, { "epoch": 1.9241552795534744e-13, "grad_norm": 0.7055402994155884, "learning_rate": 3.7225376992842e-05, "loss": 2.2407, "step": 55460 }, { "epoch": 1.9245022242486698e-13, "grad_norm": 0.6836403012275696, "learning_rate": 3.7206853611026714e-05, "loss": 2.2257, "step": 55470 }, { "epoch": 1.9248491689438652e-13, "grad_norm": 0.6921740174293518, "learning_rate": 3.7188335217488516e-05, "loss": 2.257, "step": 55480 }, { "epoch": 1.9251961136390605e-13, "grad_norm": 0.7149255275726318, "learning_rate": 3.716982181515592e-05, "loss": 2.2126, "step": 55490 }, { "epoch": 1.925543058334256e-13, "grad_norm": 0.6802980899810791, "learning_rate": 3.715131340695664e-05, "loss": 2.2296, "step": 55500 }, { "epoch": 1.9258900030294512e-13, "grad_norm": 0.6746107339859009, "learning_rate": 3.713280999581767e-05, "loss": 2.2326, "step": 55510 }, { "epoch": 1.9262369477246466e-13, "grad_norm": 0.6731981635093689, "learning_rate": 3.71143115846651e-05, "loss": 2.2108, "step": 55520 }, { "epoch": 1.926583892419842e-13, "grad_norm": 0.6924176216125488, "learning_rate": 3.7095818176424367e-05, "loss": 2.2089, "step": 55530 }, { "epoch": 1.9269308371150373e-13, "grad_norm": 0.6800044775009155, "learning_rate": 3.7077329774019985e-05, "loss": 2.2276, "step": 55540 }, { "epoch": 1.9272777818102327e-13, "grad_norm": 0.6945277452468872, "learning_rate": 3.705884638037578e-05, "loss": 2.2362, "step": 55550 }, { "epoch": 1.927624726505428e-13, "grad_norm": 0.6923208832740784, "learning_rate": 3.7040367998414704e-05, "loss": 2.2497, "step": 55560 }, { "epoch": 1.9279716712006234e-13, "grad_norm": 0.7028375864028931, "learning_rate": 3.7021894631058955e-05, "loss": 2.2374, "step": 55570 }, { "epoch": 1.9283186158958188e-13, "grad_norm": 0.6897575855255127, "learning_rate": 3.7003426281229985e-05, "loss": 2.1993, "step": 55580 }, { "epoch": 1.9286655605910141e-13, "grad_norm": 0.6952674388885498, "learning_rate": 3.6984962951848364e-05, "loss": 2.222, "step": 55590 }, { "epoch": 1.9290125052862095e-13, "grad_norm": 0.7112200856208801, "learning_rate": 3.6966504645833915e-05, "loss": 2.2257, "step": 55600 }, { "epoch": 1.9293594499814049e-13, "grad_norm": 0.7726920247077942, "learning_rate": 3.694805136610566e-05, "loss": 2.2065, "step": 55610 }, { "epoch": 1.9297063946766002e-13, "grad_norm": 0.6873964071273804, "learning_rate": 3.692960311558183e-05, "loss": 2.2394, "step": 55620 }, { "epoch": 1.9300533393717956e-13, "grad_norm": 0.6654849648475647, "learning_rate": 3.691115989717984e-05, "loss": 2.2027, "step": 55630 }, { "epoch": 1.930400284066991e-13, "grad_norm": 0.6881231665611267, "learning_rate": 3.689272171381637e-05, "loss": 2.2196, "step": 55640 }, { "epoch": 1.9307472287621863e-13, "grad_norm": 0.668402373790741, "learning_rate": 3.68742885684072e-05, "loss": 2.2465, "step": 55650 }, { "epoch": 1.9310941734573817e-13, "grad_norm": 0.6969978213310242, "learning_rate": 3.685586046386742e-05, "loss": 2.2323, "step": 55660 }, { "epoch": 1.931441118152577e-13, "grad_norm": 0.7328675389289856, "learning_rate": 3.683743740311127e-05, "loss": 2.2013, "step": 55670 }, { "epoch": 1.9317880628477724e-13, "grad_norm": 0.6689388751983643, "learning_rate": 3.681901938905216e-05, "loss": 2.2425, "step": 55680 }, { "epoch": 1.9321350075429677e-13, "grad_norm": 0.7037913203239441, "learning_rate": 3.680060642460282e-05, "loss": 2.2205, "step": 55690 }, { "epoch": 1.932481952238163e-13, "grad_norm": 0.6823405027389526, "learning_rate": 3.6782198512675e-05, "loss": 2.206, "step": 55700 }, { "epoch": 1.9328288969333585e-13, "grad_norm": 0.6727436780929565, "learning_rate": 3.6763795656179865e-05, "loss": 2.2162, "step": 55710 }, { "epoch": 1.9331758416285538e-13, "grad_norm": 0.6826083064079285, "learning_rate": 3.674539785802755e-05, "loss": 2.2, "step": 55720 }, { "epoch": 1.9335227863237492e-13, "grad_norm": 0.6764669418334961, "learning_rate": 3.6727005121127585e-05, "loss": 2.2337, "step": 55730 }, { "epoch": 1.9338697310189445e-13, "grad_norm": 0.6913479566574097, "learning_rate": 3.6708617448388616e-05, "loss": 2.2423, "step": 55740 }, { "epoch": 1.93421667571414e-13, "grad_norm": 0.6934717893600464, "learning_rate": 3.6690234842718456e-05, "loss": 2.2351, "step": 55750 }, { "epoch": 1.9345636204093353e-13, "grad_norm": 0.6885758638381958, "learning_rate": 3.6671857307024215e-05, "loss": 2.2512, "step": 55760 }, { "epoch": 1.9349105651045306e-13, "grad_norm": 0.7418928742408752, "learning_rate": 3.6653484844212084e-05, "loss": 2.2198, "step": 55770 }, { "epoch": 1.935257509799726e-13, "grad_norm": 0.6654466390609741, "learning_rate": 3.663511745718755e-05, "loss": 2.1962, "step": 55780 }, { "epoch": 1.9356044544949214e-13, "grad_norm": 0.7070629000663757, "learning_rate": 3.661675514885525e-05, "loss": 2.2087, "step": 55790 }, { "epoch": 1.9359513991901167e-13, "grad_norm": 0.6918935775756836, "learning_rate": 3.659839792211901e-05, "loss": 2.2229, "step": 55800 }, { "epoch": 1.936298343885312e-13, "grad_norm": 0.6750578284263611, "learning_rate": 3.658004577988188e-05, "loss": 2.2314, "step": 55810 }, { "epoch": 1.9366452885805074e-13, "grad_norm": 0.7055685520172119, "learning_rate": 3.656169872504607e-05, "loss": 2.1835, "step": 55820 }, { "epoch": 1.9369922332757028e-13, "grad_norm": 0.6926283836364746, "learning_rate": 3.654335676051306e-05, "loss": 2.2537, "step": 55830 }, { "epoch": 1.9373391779708982e-13, "grad_norm": 0.7037910223007202, "learning_rate": 3.6525019889183406e-05, "loss": 2.2356, "step": 55840 }, { "epoch": 1.9376861226660935e-13, "grad_norm": 0.6638105511665344, "learning_rate": 3.650668811395699e-05, "loss": 2.2303, "step": 55850 }, { "epoch": 1.938033067361289e-13, "grad_norm": 0.6951574087142944, "learning_rate": 3.6488361437732784e-05, "loss": 2.2185, "step": 55860 }, { "epoch": 1.9383800120564842e-13, "grad_norm": 0.6666550040245056, "learning_rate": 3.6470039863409016e-05, "loss": 2.2117, "step": 55870 }, { "epoch": 1.9387269567516796e-13, "grad_norm": 0.6733582615852356, "learning_rate": 3.645172339388307e-05, "loss": 2.2175, "step": 55880 }, { "epoch": 1.939073901446875e-13, "grad_norm": 0.7271906137466431, "learning_rate": 3.6433412032051525e-05, "loss": 2.1924, "step": 55890 }, { "epoch": 1.9394208461420703e-13, "grad_norm": 0.6867315769195557, "learning_rate": 3.641510578081021e-05, "loss": 2.2362, "step": 55900 }, { "epoch": 1.9397677908372657e-13, "grad_norm": 0.6789223551750183, "learning_rate": 3.6396804643054066e-05, "loss": 2.2035, "step": 55910 }, { "epoch": 1.940114735532461e-13, "grad_norm": 0.6933147311210632, "learning_rate": 3.637850862167727e-05, "loss": 2.2211, "step": 55920 }, { "epoch": 1.9404616802276564e-13, "grad_norm": 0.6734651923179626, "learning_rate": 3.636021771957318e-05, "loss": 2.2431, "step": 55930 }, { "epoch": 1.9408086249228518e-13, "grad_norm": 0.6897377967834473, "learning_rate": 3.634193193963435e-05, "loss": 2.2464, "step": 55940 }, { "epoch": 1.9411555696180471e-13, "grad_norm": 0.6935487985610962, "learning_rate": 3.63236512847525e-05, "loss": 2.2267, "step": 55950 }, { "epoch": 1.9415025143132425e-13, "grad_norm": 0.6898427605628967, "learning_rate": 3.630537575781858e-05, "loss": 2.2043, "step": 55960 }, { "epoch": 1.9418494590084379e-13, "grad_norm": 0.6860925555229187, "learning_rate": 3.62871053617227e-05, "loss": 2.2208, "step": 55970 }, { "epoch": 1.9421964037036332e-13, "grad_norm": 0.6665491461753845, "learning_rate": 3.626884009935416e-05, "loss": 2.2038, "step": 55980 }, { "epoch": 1.9425433483988286e-13, "grad_norm": 0.6764766573905945, "learning_rate": 3.6250579973601466e-05, "loss": 2.221, "step": 55990 }, { "epoch": 1.942890293094024e-13, "grad_norm": 0.7007741332054138, "learning_rate": 3.623232498735227e-05, "loss": 2.2381, "step": 56000 }, { "epoch": 1.9432372377892193e-13, "grad_norm": 0.7233401536941528, "learning_rate": 3.621407514349351e-05, "loss": 2.2378, "step": 56010 }, { "epoch": 1.9435841824844147e-13, "grad_norm": 0.6928693056106567, "learning_rate": 3.619583044491115e-05, "loss": 2.2188, "step": 56020 }, { "epoch": 1.94393112717961e-13, "grad_norm": 2.091756582260132, "learning_rate": 3.6177590894490506e-05, "loss": 2.2344, "step": 56030 }, { "epoch": 1.9442780718748054e-13, "grad_norm": 0.6868274807929993, "learning_rate": 3.615935649511595e-05, "loss": 2.2235, "step": 56040 }, { "epoch": 1.9446250165700008e-13, "grad_norm": 0.6834449172019958, "learning_rate": 3.614112724967113e-05, "loss": 2.227, "step": 56050 }, { "epoch": 1.944971961265196e-13, "grad_norm": 0.6735183596611023, "learning_rate": 3.612290316103885e-05, "loss": 2.2133, "step": 56060 }, { "epoch": 1.9453189059603915e-13, "grad_norm": 0.6634794473648071, "learning_rate": 3.610468423210105e-05, "loss": 2.2093, "step": 56070 }, { "epoch": 1.9456658506555868e-13, "grad_norm": 0.6848280429840088, "learning_rate": 3.608647046573897e-05, "loss": 2.2149, "step": 56080 }, { "epoch": 1.9460127953507822e-13, "grad_norm": 0.6999218463897705, "learning_rate": 3.606826186483288e-05, "loss": 2.2056, "step": 56090 }, { "epoch": 1.9463597400459776e-13, "grad_norm": 0.6784867644309998, "learning_rate": 3.605005843226236e-05, "loss": 2.2357, "step": 56100 }, { "epoch": 1.946706684741173e-13, "grad_norm": 0.6868731379508972, "learning_rate": 3.603186017090612e-05, "loss": 2.1751, "step": 56110 }, { "epoch": 1.9470536294363683e-13, "grad_norm": 0.6900941133499146, "learning_rate": 3.601366708364206e-05, "loss": 2.2015, "step": 56120 }, { "epoch": 1.9474005741315636e-13, "grad_norm": 0.6834601759910583, "learning_rate": 3.599547917334725e-05, "loss": 2.2118, "step": 56130 }, { "epoch": 1.947747518826759e-13, "grad_norm": 0.6848406195640564, "learning_rate": 3.597729644289793e-05, "loss": 2.1993, "step": 56140 }, { "epoch": 1.9480944635219544e-13, "grad_norm": 0.6729844212532043, "learning_rate": 3.5959118895169615e-05, "loss": 2.2174, "step": 56150 }, { "epoch": 1.9484414082171497e-13, "grad_norm": 0.6660759449005127, "learning_rate": 3.5940946533036844e-05, "loss": 2.2183, "step": 56160 }, { "epoch": 1.948788352912345e-13, "grad_norm": 0.7031444311141968, "learning_rate": 3.592277935937348e-05, "loss": 2.203, "step": 56170 }, { "epoch": 1.9491352976075405e-13, "grad_norm": 0.6811057329177856, "learning_rate": 3.590461737705247e-05, "loss": 2.2321, "step": 56180 }, { "epoch": 1.9494822423027358e-13, "grad_norm": 0.7084023356437683, "learning_rate": 3.5886460588945986e-05, "loss": 2.1736, "step": 56190 }, { "epoch": 1.9498291869979312e-13, "grad_norm": 0.6810675263404846, "learning_rate": 3.586830899792538e-05, "loss": 2.2064, "step": 56200 }, { "epoch": 1.9501761316931265e-13, "grad_norm": 0.7086247801780701, "learning_rate": 3.585016260686112e-05, "loss": 2.2526, "step": 56210 }, { "epoch": 1.950523076388322e-13, "grad_norm": 0.6729748249053955, "learning_rate": 3.5832021418622984e-05, "loss": 2.2455, "step": 56220 }, { "epoch": 1.9508700210835173e-13, "grad_norm": 0.6769990921020508, "learning_rate": 3.581388543607979e-05, "loss": 2.2012, "step": 56230 }, { "epoch": 1.9512169657787126e-13, "grad_norm": 0.6918368339538574, "learning_rate": 3.579575466209959e-05, "loss": 2.2093, "step": 56240 }, { "epoch": 1.951563910473908e-13, "grad_norm": 0.7085497975349426, "learning_rate": 3.577762909954963e-05, "loss": 2.2357, "step": 56250 }, { "epoch": 1.9519108551691033e-13, "grad_norm": 0.6607381105422974, "learning_rate": 3.575950875129629e-05, "loss": 2.2086, "step": 56260 }, { "epoch": 1.9522577998642987e-13, "grad_norm": 0.7221829295158386, "learning_rate": 3.5741393620205144e-05, "loss": 2.2489, "step": 56270 }, { "epoch": 1.952604744559494e-13, "grad_norm": 0.6797705888748169, "learning_rate": 3.5723283709140975e-05, "loss": 2.2181, "step": 56280 }, { "epoch": 1.9529516892546894e-13, "grad_norm": 0.6669319272041321, "learning_rate": 3.570517902096769e-05, "loss": 2.2177, "step": 56290 }, { "epoch": 1.9532986339498848e-13, "grad_norm": 0.6563889980316162, "learning_rate": 3.56870795585484e-05, "loss": 2.2013, "step": 56300 }, { "epoch": 1.9536455786450801e-13, "grad_norm": 0.7103325724601746, "learning_rate": 3.566898532474536e-05, "loss": 2.1841, "step": 56310 }, { "epoch": 1.9539925233402755e-13, "grad_norm": 0.6967499852180481, "learning_rate": 3.565089632242001e-05, "loss": 2.2251, "step": 56320 }, { "epoch": 1.9543394680354709e-13, "grad_norm": 0.678066611289978, "learning_rate": 3.563281255443303e-05, "loss": 2.2251, "step": 56330 }, { "epoch": 1.9546864127306662e-13, "grad_norm": 0.6573480367660522, "learning_rate": 3.561473402364412e-05, "loss": 2.2514, "step": 56340 }, { "epoch": 1.9550333574258616e-13, "grad_norm": 0.667327880859375, "learning_rate": 3.5596660732912335e-05, "loss": 2.2032, "step": 56350 }, { "epoch": 1.955380302121057e-13, "grad_norm": 0.7204027771949768, "learning_rate": 3.5578592685095735e-05, "loss": 2.2116, "step": 56360 }, { "epoch": 1.9557272468162523e-13, "grad_norm": 0.658541738986969, "learning_rate": 3.556052988305168e-05, "loss": 2.2051, "step": 56370 }, { "epoch": 1.9560741915114477e-13, "grad_norm": 0.659784197807312, "learning_rate": 3.554247232963663e-05, "loss": 2.1999, "step": 56380 }, { "epoch": 1.956421136206643e-13, "grad_norm": 0.6890407800674438, "learning_rate": 3.5524420027706214e-05, "loss": 2.2535, "step": 56390 }, { "epoch": 1.9567680809018384e-13, "grad_norm": 0.6726247668266296, "learning_rate": 3.5506372980115294e-05, "loss": 2.192, "step": 56400 }, { "epoch": 1.9571150255970338e-13, "grad_norm": 0.7111452221870422, "learning_rate": 3.5488331189717784e-05, "loss": 2.2615, "step": 56410 }, { "epoch": 1.957461970292229e-13, "grad_norm": 0.6357385516166687, "learning_rate": 3.54702946593669e-05, "loss": 2.1872, "step": 56420 }, { "epoch": 1.9578089149874245e-13, "grad_norm": 0.664034903049469, "learning_rate": 3.5452263391914946e-05, "loss": 2.2053, "step": 56430 }, { "epoch": 1.9581558596826198e-13, "grad_norm": 0.6536839008331299, "learning_rate": 3.5434237390213396e-05, "loss": 2.2142, "step": 56440 }, { "epoch": 1.9585028043778152e-13, "grad_norm": 0.7111732959747314, "learning_rate": 3.541621665711292e-05, "loss": 2.1919, "step": 56450 }, { "epoch": 1.9588497490730106e-13, "grad_norm": 0.6886693835258484, "learning_rate": 3.5398201195463334e-05, "loss": 2.2267, "step": 56460 }, { "epoch": 1.959196693768206e-13, "grad_norm": 0.6717545390129089, "learning_rate": 3.538019100811367e-05, "loss": 2.2122, "step": 56470 }, { "epoch": 1.9595436384634013e-13, "grad_norm": 0.6840323805809021, "learning_rate": 3.5362186097912e-05, "loss": 2.2222, "step": 56480 }, { "epoch": 1.9598905831585967e-13, "grad_norm": 0.7064836025238037, "learning_rate": 3.534418646770572e-05, "loss": 2.1971, "step": 56490 }, { "epoch": 1.960237527853792e-13, "grad_norm": 0.6862282156944275, "learning_rate": 3.53261921203413e-05, "loss": 2.1947, "step": 56500 }, { "epoch": 1.9605844725489874e-13, "grad_norm": 0.6821401715278625, "learning_rate": 3.530820305866437e-05, "loss": 2.2568, "step": 56510 }, { "epoch": 1.9609314172441827e-13, "grad_norm": 0.6870879530906677, "learning_rate": 3.529021928551975e-05, "loss": 2.2131, "step": 56520 }, { "epoch": 1.961278361939378e-13, "grad_norm": 0.6852291226387024, "learning_rate": 3.527224080375142e-05, "loss": 2.21, "step": 56530 }, { "epoch": 1.9616253066345735e-13, "grad_norm": 0.6651721596717834, "learning_rate": 3.5254267616202546e-05, "loss": 2.2486, "step": 56540 }, { "epoch": 1.9619722513297688e-13, "grad_norm": 0.6890032291412354, "learning_rate": 3.523629972571542e-05, "loss": 2.2143, "step": 56550 }, { "epoch": 1.9623191960249642e-13, "grad_norm": 0.680349588394165, "learning_rate": 3.521833713513149e-05, "loss": 2.226, "step": 56560 }, { "epoch": 1.9626661407201595e-13, "grad_norm": 0.7016533017158508, "learning_rate": 3.5200379847291406e-05, "loss": 2.2211, "step": 56570 }, { "epoch": 1.963013085415355e-13, "grad_norm": 0.6790884733200073, "learning_rate": 3.518242786503496e-05, "loss": 2.2142, "step": 56580 }, { "epoch": 1.9633600301105503e-13, "grad_norm": 0.6732655763626099, "learning_rate": 3.5164481191201066e-05, "loss": 2.2292, "step": 56590 }, { "epoch": 1.9637069748057456e-13, "grad_norm": 0.7228893041610718, "learning_rate": 3.51465398286279e-05, "loss": 2.2214, "step": 56600 }, { "epoch": 1.964053919500941e-13, "grad_norm": 0.6838578581809998, "learning_rate": 3.512860378015268e-05, "loss": 2.2518, "step": 56610 }, { "epoch": 1.9644008641961364e-13, "grad_norm": 0.6795223355293274, "learning_rate": 3.5110673048611867e-05, "loss": 2.2142, "step": 56620 }, { "epoch": 1.9647478088913317e-13, "grad_norm": 0.6823692321777344, "learning_rate": 3.509274763684105e-05, "loss": 2.248, "step": 56630 }, { "epoch": 1.965094753586527e-13, "grad_norm": 0.6915234923362732, "learning_rate": 3.507482754767493e-05, "loss": 2.1964, "step": 56640 }, { "epoch": 1.9654416982817224e-13, "grad_norm": 0.7000309824943542, "learning_rate": 3.505691278394751e-05, "loss": 2.2282, "step": 56650 }, { "epoch": 1.9657886429769178e-13, "grad_norm": 0.6830637454986572, "learning_rate": 3.503900334849175e-05, "loss": 2.2261, "step": 56660 }, { "epoch": 1.9661355876721132e-13, "grad_norm": 0.7008503675460815, "learning_rate": 3.502109924413996e-05, "loss": 2.2223, "step": 56670 }, { "epoch": 1.9664825323673085e-13, "grad_norm": 0.6954385042190552, "learning_rate": 3.5003200473723475e-05, "loss": 2.2251, "step": 56680 }, { "epoch": 1.966829477062504e-13, "grad_norm": 0.7049707770347595, "learning_rate": 3.498530704007284e-05, "loss": 2.2033, "step": 56690 }, { "epoch": 1.9671764217576992e-13, "grad_norm": 0.7110181450843811, "learning_rate": 3.496741894601775e-05, "loss": 2.262, "step": 56700 }, { "epoch": 1.9675233664528946e-13, "grad_norm": 0.693149983882904, "learning_rate": 3.494953619438704e-05, "loss": 2.2437, "step": 56710 }, { "epoch": 1.96787031114809e-13, "grad_norm": 0.7242529392242432, "learning_rate": 3.493165878800875e-05, "loss": 2.2152, "step": 56720 }, { "epoch": 1.9682172558432853e-13, "grad_norm": 0.6860178709030151, "learning_rate": 3.491378672970997e-05, "loss": 2.2383, "step": 56730 }, { "epoch": 1.9685642005384807e-13, "grad_norm": 0.6881654262542725, "learning_rate": 3.489592002231709e-05, "loss": 2.1999, "step": 56740 }, { "epoch": 1.968911145233676e-13, "grad_norm": 0.6922546625137329, "learning_rate": 3.487805866865554e-05, "loss": 2.2517, "step": 56750 }, { "epoch": 1.9692580899288714e-13, "grad_norm": 0.6911097168922424, "learning_rate": 3.486020267154993e-05, "loss": 2.2405, "step": 56760 }, { "epoch": 1.9696050346240668e-13, "grad_norm": 0.6826547980308533, "learning_rate": 3.4842352033824044e-05, "loss": 2.2576, "step": 56770 }, { "epoch": 1.9699519793192621e-13, "grad_norm": 0.7031705379486084, "learning_rate": 3.4824506758300786e-05, "loss": 2.2009, "step": 56780 }, { "epoch": 1.9702989240144575e-13, "grad_norm": 0.6843180060386658, "learning_rate": 3.480666684780227e-05, "loss": 2.2261, "step": 56790 }, { "epoch": 1.9706458687096529e-13, "grad_norm": 0.6674593687057495, "learning_rate": 3.4788832305149707e-05, "loss": 2.1994, "step": 56800 }, { "epoch": 1.9709928134048482e-13, "grad_norm": 0.688152551651001, "learning_rate": 3.477100313316348e-05, "loss": 2.1901, "step": 56810 }, { "epoch": 1.9713397581000436e-13, "grad_norm": 0.7152511477470398, "learning_rate": 3.475317933466311e-05, "loss": 2.224, "step": 56820 }, { "epoch": 1.971686702795239e-13, "grad_norm": 0.6852162480354309, "learning_rate": 3.4735360912467276e-05, "loss": 2.2344, "step": 56830 }, { "epoch": 1.9720336474904343e-13, "grad_norm": 0.6760079860687256, "learning_rate": 3.471754786939382e-05, "loss": 2.2023, "step": 56840 }, { "epoch": 1.9723805921856297e-13, "grad_norm": 0.6742992997169495, "learning_rate": 3.46997402082597e-05, "loss": 2.2073, "step": 56850 }, { "epoch": 1.972727536880825e-13, "grad_norm": 0.6769511699676514, "learning_rate": 3.468193793188107e-05, "loss": 2.2372, "step": 56860 }, { "epoch": 1.9730744815760204e-13, "grad_norm": 0.6826435923576355, "learning_rate": 3.4664141043073195e-05, "loss": 2.2224, "step": 56870 }, { "epoch": 1.9734214262712158e-13, "grad_norm": 0.6958000659942627, "learning_rate": 3.4646349544650514e-05, "loss": 2.2596, "step": 56880 }, { "epoch": 1.973768370966411e-13, "grad_norm": 0.6643345355987549, "learning_rate": 3.462856343942657e-05, "loss": 2.2114, "step": 56890 }, { "epoch": 1.9741153156616065e-13, "grad_norm": 0.7324702143669128, "learning_rate": 3.46107827302141e-05, "loss": 2.2021, "step": 56900 }, { "epoch": 1.9744622603568018e-13, "grad_norm": 0.6909886598587036, "learning_rate": 3.459300741982496e-05, "loss": 2.2205, "step": 56910 }, { "epoch": 1.9748092050519972e-13, "grad_norm": 0.6759948134422302, "learning_rate": 3.45752375110702e-05, "loss": 2.2359, "step": 56920 }, { "epoch": 1.9751561497471926e-13, "grad_norm": 0.6506768465042114, "learning_rate": 3.4557473006759935e-05, "loss": 2.2324, "step": 56930 }, { "epoch": 1.975503094442388e-13, "grad_norm": 0.6765527725219727, "learning_rate": 3.453971390970348e-05, "loss": 2.2254, "step": 56940 }, { "epoch": 1.9758500391375833e-13, "grad_norm": 0.6912445425987244, "learning_rate": 3.4521960222709285e-05, "loss": 2.222, "step": 56950 }, { "epoch": 1.9761969838327786e-13, "grad_norm": 0.6864023804664612, "learning_rate": 3.4504211948584924e-05, "loss": 2.199, "step": 56960 }, { "epoch": 1.976543928527974e-13, "grad_norm": 0.6963790655136108, "learning_rate": 3.448646909013721e-05, "loss": 2.2232, "step": 56970 }, { "epoch": 1.9768908732231694e-13, "grad_norm": 0.7311121821403503, "learning_rate": 3.446873165017191e-05, "loss": 2.2198, "step": 56980 }, { "epoch": 1.9772378179183647e-13, "grad_norm": 0.6754865646362305, "learning_rate": 3.4450999631494135e-05, "loss": 2.2151, "step": 56990 }, { "epoch": 1.97758476261356e-13, "grad_norm": 0.7094979286193848, "learning_rate": 3.4433273036908004e-05, "loss": 2.2472, "step": 57000 }, { "epoch": 1.9779317073087554e-13, "grad_norm": 0.7325478196144104, "learning_rate": 3.441555186921686e-05, "loss": 2.2044, "step": 57010 }, { "epoch": 1.9782786520039508e-13, "grad_norm": 0.6764374375343323, "learning_rate": 3.439783613122313e-05, "loss": 2.2241, "step": 57020 }, { "epoch": 1.9786255966991462e-13, "grad_norm": 0.6790602207183838, "learning_rate": 3.438012582572838e-05, "loss": 2.1925, "step": 57030 }, { "epoch": 1.9789725413943415e-13, "grad_norm": 0.6904220581054688, "learning_rate": 3.436242095553342e-05, "loss": 2.2344, "step": 57040 }, { "epoch": 1.979319486089537e-13, "grad_norm": 0.6879763007164001, "learning_rate": 3.434472152343804e-05, "loss": 2.2417, "step": 57050 }, { "epoch": 1.9796664307847323e-13, "grad_norm": 0.693697452545166, "learning_rate": 3.43270275322413e-05, "loss": 2.1998, "step": 57060 }, { "epoch": 1.9800133754799276e-13, "grad_norm": 0.6731449365615845, "learning_rate": 3.430933898474135e-05, "loss": 2.2255, "step": 57070 }, { "epoch": 1.980360320175123e-13, "grad_norm": 0.6917685270309448, "learning_rate": 3.429165588373547e-05, "loss": 2.2371, "step": 57080 }, { "epoch": 1.9807072648703183e-13, "grad_norm": 0.7331467270851135, "learning_rate": 3.427397823202009e-05, "loss": 2.2215, "step": 57090 }, { "epoch": 1.9810542095655137e-13, "grad_norm": 0.6843092441558838, "learning_rate": 3.425630603239076e-05, "loss": 2.1851, "step": 57100 }, { "epoch": 1.981401154260709e-13, "grad_norm": 0.6689584255218506, "learning_rate": 3.4238639287642226e-05, "loss": 2.2262, "step": 57110 }, { "epoch": 1.9817480989559044e-13, "grad_norm": 0.7507203221321106, "learning_rate": 3.422097800056831e-05, "loss": 2.244, "step": 57120 }, { "epoch": 1.9820950436510998e-13, "grad_norm": 0.7193174362182617, "learning_rate": 3.4203322173962006e-05, "loss": 2.2325, "step": 57130 }, { "epoch": 1.9824419883462951e-13, "grad_norm": 0.6945371031761169, "learning_rate": 3.418567181061542e-05, "loss": 2.1994, "step": 57140 }, { "epoch": 1.9827889330414905e-13, "grad_norm": 0.6742719411849976, "learning_rate": 3.41680269133198e-05, "loss": 2.234, "step": 57150 }, { "epoch": 1.9831358777366859e-13, "grad_norm": 0.6976245641708374, "learning_rate": 3.415038748486554e-05, "loss": 2.2235, "step": 57160 }, { "epoch": 1.9834828224318812e-13, "grad_norm": 0.7040976285934448, "learning_rate": 3.413275352804214e-05, "loss": 2.2505, "step": 57170 }, { "epoch": 1.9838297671270766e-13, "grad_norm": 0.6987464427947998, "learning_rate": 3.411512504563831e-05, "loss": 2.2461, "step": 57180 }, { "epoch": 1.984176711822272e-13, "grad_norm": 0.6992568373680115, "learning_rate": 3.409750204044181e-05, "loss": 2.2378, "step": 57190 }, { "epoch": 1.9845236565174673e-13, "grad_norm": 0.7094656825065613, "learning_rate": 3.4079884515239566e-05, "loss": 2.2503, "step": 57200 }, { "epoch": 1.9848706012126627e-13, "grad_norm": 0.7030715346336365, "learning_rate": 3.406227247281765e-05, "loss": 2.2388, "step": 57210 }, { "epoch": 1.985217545907858e-13, "grad_norm": 0.6705163717269897, "learning_rate": 3.404466591596126e-05, "loss": 2.2295, "step": 57220 }, { "epoch": 1.9855644906030534e-13, "grad_norm": 0.6717033982276917, "learning_rate": 3.4027064847454675e-05, "loss": 2.2356, "step": 57230 }, { "epoch": 1.9859114352982488e-13, "grad_norm": 0.6768667101860046, "learning_rate": 3.4009469270081404e-05, "loss": 2.2015, "step": 57240 }, { "epoch": 1.986258379993444e-13, "grad_norm": 0.6854925155639648, "learning_rate": 3.399187918662402e-05, "loss": 2.2009, "step": 57250 }, { "epoch": 1.9866053246886395e-13, "grad_norm": 0.6880122423171997, "learning_rate": 3.3974294599864255e-05, "loss": 2.2156, "step": 57260 }, { "epoch": 1.9869522693838348e-13, "grad_norm": 0.7094545364379883, "learning_rate": 3.395671551258294e-05, "loss": 2.2355, "step": 57270 }, { "epoch": 1.9872992140790302e-13, "grad_norm": 0.6826455593109131, "learning_rate": 3.393914192756005e-05, "loss": 2.2416, "step": 57280 }, { "epoch": 1.9876461587742256e-13, "grad_norm": 0.668476939201355, "learning_rate": 3.392157384757475e-05, "loss": 2.1966, "step": 57290 }, { "epoch": 1.987993103469421e-13, "grad_norm": 0.7034956216812134, "learning_rate": 3.390401127540519e-05, "loss": 2.2068, "step": 57300 }, { "epoch": 1.9883400481646163e-13, "grad_norm": 0.6725413799285889, "learning_rate": 3.3886454213828815e-05, "loss": 2.2608, "step": 57310 }, { "epoch": 1.9886869928598117e-13, "grad_norm": 0.7032600045204163, "learning_rate": 3.386890266562211e-05, "loss": 2.2174, "step": 57320 }, { "epoch": 1.989033937555007e-13, "grad_norm": 0.6920255422592163, "learning_rate": 3.3851356633560647e-05, "loss": 2.1805, "step": 57330 }, { "epoch": 1.9893808822502024e-13, "grad_norm": 0.7154740691184998, "learning_rate": 3.383381612041929e-05, "loss": 2.2002, "step": 57340 }, { "epoch": 1.9897278269453977e-13, "grad_norm": 0.719352662563324, "learning_rate": 3.3816281128971796e-05, "loss": 2.2319, "step": 57350 }, { "epoch": 1.990074771640593e-13, "grad_norm": 0.6863684058189392, "learning_rate": 3.379875166199127e-05, "loss": 2.2065, "step": 57360 }, { "epoch": 1.9904217163357885e-13, "grad_norm": 0.6937190294265747, "learning_rate": 3.378122772224977e-05, "loss": 2.1858, "step": 57370 }, { "epoch": 1.9907686610309838e-13, "grad_norm": 0.6926230192184448, "learning_rate": 3.3763709312518614e-05, "loss": 2.2339, "step": 57380 }, { "epoch": 1.9911156057261792e-13, "grad_norm": 0.6941501498222351, "learning_rate": 3.3746196435568155e-05, "loss": 2.1835, "step": 57390 }, { "epoch": 1.9914625504213745e-13, "grad_norm": 0.7045302391052246, "learning_rate": 3.372868909416793e-05, "loss": 2.1985, "step": 57400 }, { "epoch": 1.99180949511657e-13, "grad_norm": 0.6915671229362488, "learning_rate": 3.371118729108655e-05, "loss": 2.2424, "step": 57410 }, { "epoch": 1.9921564398117653e-13, "grad_norm": 0.719722330570221, "learning_rate": 3.369369102909177e-05, "loss": 2.2174, "step": 57420 }, { "epoch": 1.9925033845069606e-13, "grad_norm": 0.6746155619621277, "learning_rate": 3.3676200310950506e-05, "loss": 2.2172, "step": 57430 }, { "epoch": 1.992850329202156e-13, "grad_norm": 0.6612242460250854, "learning_rate": 3.3658715139428736e-05, "loss": 2.2363, "step": 57440 }, { "epoch": 1.9931972738973514e-13, "grad_norm": 0.7015637159347534, "learning_rate": 3.3641235517291615e-05, "loss": 2.2514, "step": 57450 }, { "epoch": 1.9935442185925467e-13, "grad_norm": 0.6826054453849792, "learning_rate": 3.362376144730336e-05, "loss": 2.219, "step": 57460 }, { "epoch": 1.993891163287742e-13, "grad_norm": 0.7055874466896057, "learning_rate": 3.360629293222738e-05, "loss": 2.2142, "step": 57470 }, { "epoch": 1.9942381079829374e-13, "grad_norm": 0.6902127861976624, "learning_rate": 3.358882997482614e-05, "loss": 2.2023, "step": 57480 }, { "epoch": 1.9945850526781328e-13, "grad_norm": 0.6946866512298584, "learning_rate": 3.3571372577861264e-05, "loss": 2.2149, "step": 57490 }, { "epoch": 1.9949319973733282e-13, "grad_norm": 0.6820455193519592, "learning_rate": 3.355392074409351e-05, "loss": 2.2236, "step": 57500 }, { "epoch": 1.9952789420685235e-13, "grad_norm": 0.6881712079048157, "learning_rate": 3.353647447628272e-05, "loss": 2.2261, "step": 57510 }, { "epoch": 1.995625886763719e-13, "grad_norm": 0.7004384994506836, "learning_rate": 3.351903377718786e-05, "loss": 2.2202, "step": 57520 }, { "epoch": 1.9959728314589142e-13, "grad_norm": 0.6851642727851868, "learning_rate": 3.350159864956706e-05, "loss": 2.2336, "step": 57530 }, { "epoch": 1.9963197761541096e-13, "grad_norm": 0.6986420750617981, "learning_rate": 3.348416909617749e-05, "loss": 2.2164, "step": 57540 }, { "epoch": 1.996666720849305e-13, "grad_norm": 0.6653397083282471, "learning_rate": 3.34667451197755e-05, "loss": 2.2051, "step": 57550 }, { "epoch": 1.9970136655445003e-13, "grad_norm": 0.7188720107078552, "learning_rate": 3.344932672311657e-05, "loss": 2.2248, "step": 57560 }, { "epoch": 1.9973606102396957e-13, "grad_norm": 0.6813328862190247, "learning_rate": 3.343191390895525e-05, "loss": 2.2127, "step": 57570 }, { "epoch": 1.997707554934891e-13, "grad_norm": 0.6814850568771362, "learning_rate": 3.341450668004522e-05, "loss": 2.1935, "step": 57580 }, { "epoch": 1.9980544996300864e-13, "grad_norm": 0.700226902961731, "learning_rate": 3.3397105039139285e-05, "loss": 2.2214, "step": 57590 }, { "epoch": 1.9984014443252818e-13, "grad_norm": 0.719834566116333, "learning_rate": 3.337970898898936e-05, "loss": 2.196, "step": 57600 }, { "epoch": 1.9987483890204771e-13, "grad_norm": 0.6895444989204407, "learning_rate": 3.336231853234652e-05, "loss": 2.1716, "step": 57610 }, { "epoch": 1.9990953337156725e-13, "grad_norm": 0.6895988583564758, "learning_rate": 3.334493367196084e-05, "loss": 2.1988, "step": 57620 }, { "epoch": 1.9994422784108679e-13, "grad_norm": 0.68779057264328, "learning_rate": 3.332755441058166e-05, "loss": 2.2237, "step": 57630 }, { "epoch": 1.9997892231060632e-13, "grad_norm": 0.7078351974487305, "learning_rate": 3.331018075095732e-05, "loss": 2.2373, "step": 57640 }, { "epoch": 2.0001361678012586e-13, "grad_norm": 0.6864888072013855, "learning_rate": 3.3292812695835324e-05, "loss": 2.2275, "step": 57650 }, { "epoch": 2.000483112496454e-13, "grad_norm": 0.698531448841095, "learning_rate": 3.32754502479623e-05, "loss": 2.2236, "step": 57660 }, { "epoch": 2.0008300571916493e-13, "grad_norm": 0.6818727850914001, "learning_rate": 3.325809341008394e-05, "loss": 2.1964, "step": 57670 }, { "epoch": 2.0011770018868447e-13, "grad_norm": 0.6956501007080078, "learning_rate": 3.324074218494511e-05, "loss": 2.2351, "step": 57680 }, { "epoch": 2.00152394658204e-13, "grad_norm": 0.7242567539215088, "learning_rate": 3.322339657528971e-05, "loss": 2.2223, "step": 57690 }, { "epoch": 2.0018708912772354e-13, "grad_norm": 0.6827629804611206, "learning_rate": 3.3206056583860834e-05, "loss": 2.2216, "step": 57700 }, { "epoch": 2.0022178359724307e-13, "grad_norm": 0.6701092720031738, "learning_rate": 3.318872221340065e-05, "loss": 2.2328, "step": 57710 }, { "epoch": 2.002564780667626e-13, "grad_norm": 0.724159836769104, "learning_rate": 3.3171393466650434e-05, "loss": 2.2218, "step": 57720 }, { "epoch": 2.0029117253628215e-13, "grad_norm": 0.6939992904663086, "learning_rate": 3.315407034635058e-05, "loss": 2.209, "step": 57730 }, { "epoch": 2.0032586700580168e-13, "grad_norm": 0.7049523591995239, "learning_rate": 3.313675285524058e-05, "loss": 2.2025, "step": 57740 }, { "epoch": 2.0036056147532122e-13, "grad_norm": 0.709958553314209, "learning_rate": 3.311944099605905e-05, "loss": 2.2278, "step": 57750 }, { "epoch": 2.0039525594484076e-13, "grad_norm": 0.6808635592460632, "learning_rate": 3.310213477154374e-05, "loss": 2.2425, "step": 57760 }, { "epoch": 2.004299504143603e-13, "grad_norm": 0.6906872987747192, "learning_rate": 3.308483418443145e-05, "loss": 2.2327, "step": 57770 }, { "epoch": 2.0046464488387983e-13, "grad_norm": 0.6896679997444153, "learning_rate": 3.306753923745812e-05, "loss": 2.2053, "step": 57780 }, { "epoch": 2.0049933935339936e-13, "grad_norm": 0.6759358048439026, "learning_rate": 3.3050249933358796e-05, "loss": 2.2089, "step": 57790 }, { "epoch": 2.005340338229189e-13, "grad_norm": 0.6671485304832458, "learning_rate": 3.3032966274867654e-05, "loss": 2.199, "step": 57800 }, { "epoch": 2.0056872829243844e-13, "grad_norm": 0.6772640347480774, "learning_rate": 3.301568826471792e-05, "loss": 2.2117, "step": 57810 }, { "epoch": 2.0060342276195797e-13, "grad_norm": 0.667956531047821, "learning_rate": 3.2998415905642e-05, "loss": 2.2502, "step": 57820 }, { "epoch": 2.006381172314775e-13, "grad_norm": 0.6905168890953064, "learning_rate": 3.298114920037135e-05, "loss": 2.1932, "step": 57830 }, { "epoch": 2.0067281170099704e-13, "grad_norm": 0.6978221535682678, "learning_rate": 3.296388815163656e-05, "loss": 2.2127, "step": 57840 }, { "epoch": 2.0070750617051658e-13, "grad_norm": 0.7429217100143433, "learning_rate": 3.2946632762167315e-05, "loss": 2.2557, "step": 57850 }, { "epoch": 2.0074220064003612e-13, "grad_norm": 0.7071837186813354, "learning_rate": 3.2929383034692386e-05, "loss": 2.2499, "step": 57860 }, { "epoch": 2.0077689510955565e-13, "grad_norm": 0.6730387210845947, "learning_rate": 3.2912138971939675e-05, "loss": 2.2252, "step": 57870 }, { "epoch": 2.008115895790752e-13, "grad_norm": 0.6911669969558716, "learning_rate": 3.28949005766362e-05, "loss": 2.2289, "step": 57880 }, { "epoch": 2.0084628404859473e-13, "grad_norm": 0.6853705644607544, "learning_rate": 3.2877667851508075e-05, "loss": 2.2406, "step": 57890 }, { "epoch": 2.0088097851811426e-13, "grad_norm": 0.6770988702774048, "learning_rate": 3.2860440799280466e-05, "loss": 2.2252, "step": 57900 }, { "epoch": 2.009156729876338e-13, "grad_norm": 0.6923528909683228, "learning_rate": 3.284321942267772e-05, "loss": 2.2378, "step": 57910 }, { "epoch": 2.0095036745715333e-13, "grad_norm": 0.691408634185791, "learning_rate": 3.282600372442321e-05, "loss": 2.2225, "step": 57920 }, { "epoch": 2.0098506192667287e-13, "grad_norm": 0.6811424493789673, "learning_rate": 3.2808793707239515e-05, "loss": 2.2155, "step": 57930 }, { "epoch": 2.010197563961924e-13, "grad_norm": 0.6985055804252625, "learning_rate": 3.279158937384817e-05, "loss": 2.2286, "step": 57940 }, { "epoch": 2.0105445086571194e-13, "grad_norm": 0.6760210394859314, "learning_rate": 3.277439072696996e-05, "loss": 2.221, "step": 57950 }, { "epoch": 2.0108914533523148e-13, "grad_norm": 0.6840370297431946, "learning_rate": 3.275719776932467e-05, "loss": 2.2537, "step": 57960 }, { "epoch": 2.0112383980475101e-13, "grad_norm": 0.6718596816062927, "learning_rate": 3.2740010503631203e-05, "loss": 2.181, "step": 57970 }, { "epoch": 2.0115853427427055e-13, "grad_norm": 0.6900506615638733, "learning_rate": 3.272282893260764e-05, "loss": 2.2336, "step": 57980 }, { "epoch": 2.0119322874379009e-13, "grad_norm": 0.6776990294456482, "learning_rate": 3.270565305897102e-05, "loss": 2.2098, "step": 57990 }, { "epoch": 2.0122792321330962e-13, "grad_norm": 0.6962273716926575, "learning_rate": 3.268848288543764e-05, "loss": 2.2657, "step": 58000 }, { "epoch": 2.0126261768282916e-13, "grad_norm": 0.6806641817092896, "learning_rate": 3.267131841472273e-05, "loss": 2.2438, "step": 58010 }, { "epoch": 2.012973121523487e-13, "grad_norm": 0.6974822282791138, "learning_rate": 3.2654159649540755e-05, "loss": 2.2039, "step": 58020 }, { "epoch": 2.0133200662186823e-13, "grad_norm": 0.7041734457015991, "learning_rate": 3.2637006592605225e-05, "loss": 2.1907, "step": 58030 }, { "epoch": 2.0136670109138777e-13, "grad_norm": 0.6784113049507141, "learning_rate": 3.261985924662872e-05, "loss": 2.221, "step": 58040 }, { "epoch": 2.014013955609073e-13, "grad_norm": 0.6753281354904175, "learning_rate": 3.2602717614323e-05, "loss": 2.1978, "step": 58050 }, { "epoch": 2.0143609003042684e-13, "grad_norm": 0.6644443273544312, "learning_rate": 3.25855816983988e-05, "loss": 2.2042, "step": 58060 }, { "epoch": 2.0147078449994638e-13, "grad_norm": 0.754115641117096, "learning_rate": 3.2568451501566065e-05, "loss": 2.1963, "step": 58070 }, { "epoch": 2.015054789694659e-13, "grad_norm": 0.7369775176048279, "learning_rate": 3.255132702653377e-05, "loss": 2.2416, "step": 58080 }, { "epoch": 2.0154017343898545e-13, "grad_norm": 0.6941677927970886, "learning_rate": 3.253420827601e-05, "loss": 2.2283, "step": 58090 }, { "epoch": 2.0157486790850498e-13, "grad_norm": 0.671677827835083, "learning_rate": 3.251709525270195e-05, "loss": 2.2349, "step": 58100 }, { "epoch": 2.0160956237802452e-13, "grad_norm": 0.7150155901908875, "learning_rate": 3.24999879593159e-05, "loss": 2.2073, "step": 58110 }, { "epoch": 2.0164425684754406e-13, "grad_norm": 0.6783721446990967, "learning_rate": 3.24828863985572e-05, "loss": 2.2294, "step": 58120 }, { "epoch": 2.016789513170636e-13, "grad_norm": 0.7031263709068298, "learning_rate": 3.246579057313032e-05, "loss": 2.1771, "step": 58130 }, { "epoch": 2.0171364578658313e-13, "grad_norm": 0.6875216364860535, "learning_rate": 3.244870048573882e-05, "loss": 2.1879, "step": 58140 }, { "epoch": 2.0174834025610267e-13, "grad_norm": 0.7253540754318237, "learning_rate": 3.2431616139085365e-05, "loss": 2.2352, "step": 58150 }, { "epoch": 2.017830347256222e-13, "grad_norm": 0.7039329409599304, "learning_rate": 3.2414537535871695e-05, "loss": 2.2289, "step": 58160 }, { "epoch": 2.0181772919514174e-13, "grad_norm": 0.7041508555412292, "learning_rate": 3.239746467879863e-05, "loss": 2.2273, "step": 58170 }, { "epoch": 2.0185242366466127e-13, "grad_norm": 0.6798490881919861, "learning_rate": 3.23803975705661e-05, "loss": 2.2128, "step": 58180 }, { "epoch": 2.018871181341808e-13, "grad_norm": 0.709338903427124, "learning_rate": 3.236333621387311e-05, "loss": 2.2841, "step": 58190 }, { "epoch": 2.0192181260370035e-13, "grad_norm": 0.6964086294174194, "learning_rate": 3.2346280611417795e-05, "loss": 2.2269, "step": 58200 }, { "epoch": 2.0195650707321988e-13, "grad_norm": 0.700760543346405, "learning_rate": 3.232923076589733e-05, "loss": 2.2079, "step": 58210 }, { "epoch": 2.0199120154273942e-13, "grad_norm": 0.6927037835121155, "learning_rate": 3.231218668000802e-05, "loss": 2.182, "step": 58220 }, { "epoch": 2.0202589601225895e-13, "grad_norm": 0.6749820113182068, "learning_rate": 3.229514835644522e-05, "loss": 2.2107, "step": 58230 }, { "epoch": 2.020605904817785e-13, "grad_norm": 0.6978556513786316, "learning_rate": 3.227811579790337e-05, "loss": 2.2271, "step": 58240 }, { "epoch": 2.0209528495129803e-13, "grad_norm": 0.7042207717895508, "learning_rate": 3.226108900707611e-05, "loss": 2.2124, "step": 58250 }, { "epoch": 2.0212997942081756e-13, "grad_norm": 0.6905052661895752, "learning_rate": 3.224406798665598e-05, "loss": 2.2576, "step": 58260 }, { "epoch": 2.021646738903371e-13, "grad_norm": 0.6919605731964111, "learning_rate": 3.2227052739334765e-05, "loss": 2.2532, "step": 58270 }, { "epoch": 2.0219936835985663e-13, "grad_norm": 0.7407299280166626, "learning_rate": 3.2210043267803286e-05, "loss": 2.227, "step": 58280 }, { "epoch": 2.0223406282937617e-13, "grad_norm": 0.7166368961334229, "learning_rate": 3.2193039574751396e-05, "loss": 2.2219, "step": 58290 }, { "epoch": 2.022687572988957e-13, "grad_norm": 0.6987365484237671, "learning_rate": 3.2176041662868154e-05, "loss": 2.1983, "step": 58300 }, { "epoch": 2.0230345176841524e-13, "grad_norm": 0.6973633766174316, "learning_rate": 3.215904953484155e-05, "loss": 2.2435, "step": 58310 }, { "epoch": 2.0233814623793478e-13, "grad_norm": 0.6826244592666626, "learning_rate": 3.214206319335883e-05, "loss": 2.1979, "step": 58320 }, { "epoch": 2.0237284070745432e-13, "grad_norm": 0.6877018213272095, "learning_rate": 3.212508264110614e-05, "loss": 2.2391, "step": 58330 }, { "epoch": 2.0240753517697385e-13, "grad_norm": 0.6992455124855042, "learning_rate": 3.210810788076889e-05, "loss": 2.1814, "step": 58340 }, { "epoch": 2.024422296464934e-13, "grad_norm": 0.71050626039505, "learning_rate": 3.2091138915031464e-05, "loss": 2.2519, "step": 58350 }, { "epoch": 2.0247692411601292e-13, "grad_norm": 0.6745152473449707, "learning_rate": 3.207417574657734e-05, "loss": 2.2347, "step": 58360 }, { "epoch": 2.0251161858553246e-13, "grad_norm": 0.7129634618759155, "learning_rate": 3.205721837808915e-05, "loss": 2.2566, "step": 58370 }, { "epoch": 2.02546313055052e-13, "grad_norm": 0.6890061497688293, "learning_rate": 3.2040266812248475e-05, "loss": 2.2306, "step": 58380 }, { "epoch": 2.0258100752457153e-13, "grad_norm": 0.6971957087516785, "learning_rate": 3.202332105173613e-05, "loss": 2.2631, "step": 58390 }, { "epoch": 2.0261570199409107e-13, "grad_norm": 0.6980923414230347, "learning_rate": 3.2006381099231916e-05, "loss": 2.2258, "step": 58400 }, { "epoch": 2.026503964636106e-13, "grad_norm": 0.7102230191230774, "learning_rate": 3.198944695741474e-05, "loss": 2.2466, "step": 58410 }, { "epoch": 2.0268509093313014e-13, "grad_norm": 0.7260563373565674, "learning_rate": 3.197251862896258e-05, "loss": 2.2511, "step": 58420 }, { "epoch": 2.0271978540264968e-13, "grad_norm": 0.6737874746322632, "learning_rate": 3.1955596116552505e-05, "loss": 2.2115, "step": 58430 }, { "epoch": 2.0275447987216921e-13, "grad_norm": 0.6787833571434021, "learning_rate": 3.1938679422860717e-05, "loss": 2.2178, "step": 58440 }, { "epoch": 2.0278917434168875e-13, "grad_norm": 0.7104066014289856, "learning_rate": 3.1921768550562365e-05, "loss": 2.2169, "step": 58450 }, { "epoch": 2.0282386881120829e-13, "grad_norm": 0.6827510595321655, "learning_rate": 3.1904863502331824e-05, "loss": 2.2058, "step": 58460 }, { "epoch": 2.0285856328072782e-13, "grad_norm": 0.7128645777702332, "learning_rate": 3.188796428084244e-05, "loss": 2.2255, "step": 58470 }, { "epoch": 2.0289325775024736e-13, "grad_norm": 0.6667765378952026, "learning_rate": 3.187107088876669e-05, "loss": 2.2218, "step": 58480 }, { "epoch": 2.029279522197669e-13, "grad_norm": 0.7030245661735535, "learning_rate": 3.185418332877614e-05, "loss": 2.2484, "step": 58490 }, { "epoch": 2.0296264668928643e-13, "grad_norm": 0.6583842039108276, "learning_rate": 3.183730160354139e-05, "loss": 2.2479, "step": 58500 }, { "epoch": 2.0299734115880597e-13, "grad_norm": 0.6922379732131958, "learning_rate": 3.182042571573212e-05, "loss": 2.2035, "step": 58510 }, { "epoch": 2.030320356283255e-13, "grad_norm": 0.7293688058853149, "learning_rate": 3.1803555668017154e-05, "loss": 2.2506, "step": 58520 }, { "epoch": 2.0306673009784504e-13, "grad_norm": 0.6891331076622009, "learning_rate": 3.178669146306433e-05, "loss": 2.2411, "step": 58530 }, { "epoch": 2.0310142456736457e-13, "grad_norm": 0.7089356184005737, "learning_rate": 3.1769833103540554e-05, "loss": 2.2195, "step": 58540 }, { "epoch": 2.031361190368841e-13, "grad_norm": 0.7006564736366272, "learning_rate": 3.175298059211185e-05, "loss": 2.2026, "step": 58550 }, { "epoch": 2.0317081350640365e-13, "grad_norm": 0.7441924214363098, "learning_rate": 3.1736133931443277e-05, "loss": 2.2003, "step": 58560 }, { "epoch": 2.0320550797592318e-13, "grad_norm": 0.6773059964179993, "learning_rate": 3.1719293124199046e-05, "loss": 2.2198, "step": 58570 }, { "epoch": 2.0324020244544272e-13, "grad_norm": 0.6968755722045898, "learning_rate": 3.17024581730423e-05, "loss": 2.237, "step": 58580 }, { "epoch": 2.0327489691496226e-13, "grad_norm": 0.689990758895874, "learning_rate": 3.168562908063541e-05, "loss": 2.242, "step": 58590 }, { "epoch": 2.033095913844818e-13, "grad_norm": 0.6784997582435608, "learning_rate": 3.1668805849639726e-05, "loss": 2.2298, "step": 58600 }, { "epoch": 2.0334428585400133e-13, "grad_norm": 0.7214030623435974, "learning_rate": 3.165198848271568e-05, "loss": 2.206, "step": 58610 }, { "epoch": 2.0337898032352086e-13, "grad_norm": 0.7107515335083008, "learning_rate": 3.163517698252286e-05, "loss": 2.2129, "step": 58620 }, { "epoch": 2.034136747930404e-13, "grad_norm": 0.6821419596672058, "learning_rate": 3.1618371351719755e-05, "loss": 2.2173, "step": 58630 }, { "epoch": 2.0344836926255994e-13, "grad_norm": 0.6824473738670349, "learning_rate": 3.160157159296415e-05, "loss": 2.2, "step": 58640 }, { "epoch": 2.0348306373207947e-13, "grad_norm": 0.6782498359680176, "learning_rate": 3.158477770891267e-05, "loss": 2.207, "step": 58650 }, { "epoch": 2.03517758201599e-13, "grad_norm": 0.6749300360679626, "learning_rate": 3.1567989702221196e-05, "loss": 2.2126, "step": 58660 }, { "epoch": 2.0355245267111854e-13, "grad_norm": 0.7308996915817261, "learning_rate": 3.15512075755446e-05, "loss": 2.2413, "step": 58670 }, { "epoch": 2.0358714714063808e-13, "grad_norm": 0.6661681532859802, "learning_rate": 3.153443133153678e-05, "loss": 2.2279, "step": 58680 }, { "epoch": 2.0362184161015762e-13, "grad_norm": 0.7010729312896729, "learning_rate": 3.151766097285083e-05, "loss": 2.2303, "step": 58690 }, { "epoch": 2.0365653607967715e-13, "grad_norm": 0.691994309425354, "learning_rate": 3.150089650213877e-05, "loss": 2.1904, "step": 58700 }, { "epoch": 2.036912305491967e-13, "grad_norm": 0.6876073479652405, "learning_rate": 3.1484137922051796e-05, "loss": 2.2118, "step": 58710 }, { "epoch": 2.0372592501871623e-13, "grad_norm": 0.7186281085014343, "learning_rate": 3.1467385235240127e-05, "loss": 2.2067, "step": 58720 }, { "epoch": 2.0376061948823576e-13, "grad_norm": 0.775360643863678, "learning_rate": 3.145063844435306e-05, "loss": 2.2248, "step": 58730 }, { "epoch": 2.037953139577553e-13, "grad_norm": 0.6902432441711426, "learning_rate": 3.143389755203893e-05, "loss": 2.2388, "step": 58740 }, { "epoch": 2.0383000842727483e-13, "grad_norm": 0.700825572013855, "learning_rate": 3.141716256094518e-05, "loss": 2.2376, "step": 58750 }, { "epoch": 2.0386470289679437e-13, "grad_norm": 0.6703195571899414, "learning_rate": 3.140043347371832e-05, "loss": 2.1887, "step": 58760 }, { "epoch": 2.038993973663139e-13, "grad_norm": 0.6940174102783203, "learning_rate": 3.13837102930039e-05, "loss": 2.2091, "step": 58770 }, { "epoch": 2.0393409183583344e-13, "grad_norm": 0.6772077679634094, "learning_rate": 3.136699302144655e-05, "loss": 2.2096, "step": 58780 }, { "epoch": 2.0396878630535298e-13, "grad_norm": 0.6905181407928467, "learning_rate": 3.1350281661689946e-05, "loss": 2.2257, "step": 58790 }, { "epoch": 2.0400348077487251e-13, "grad_norm": 0.7190039157867432, "learning_rate": 3.133357621637688e-05, "loss": 2.232, "step": 58800 }, { "epoch": 2.0403817524439205e-13, "grad_norm": 0.7259248495101929, "learning_rate": 3.131687668814913e-05, "loss": 2.2105, "step": 58810 }, { "epoch": 2.0407286971391159e-13, "grad_norm": 0.7141028046607971, "learning_rate": 3.13001830796476e-05, "loss": 2.2415, "step": 58820 }, { "epoch": 2.0410756418343112e-13, "grad_norm": 0.6928132772445679, "learning_rate": 3.128349539351226e-05, "loss": 2.2591, "step": 58830 }, { "epoch": 2.0414225865295066e-13, "grad_norm": 0.7004120349884033, "learning_rate": 3.126681363238211e-05, "loss": 2.1809, "step": 58840 }, { "epoch": 2.041769531224702e-13, "grad_norm": 0.7154905796051025, "learning_rate": 3.125013779889523e-05, "loss": 2.1885, "step": 58850 }, { "epoch": 2.0421164759198973e-13, "grad_norm": 0.7065553665161133, "learning_rate": 3.123346789568875e-05, "loss": 2.2229, "step": 58860 }, { "epoch": 2.0424634206150927e-13, "grad_norm": 0.6810075640678406, "learning_rate": 3.1216803925398874e-05, "loss": 2.2038, "step": 58870 }, { "epoch": 2.042810365310288e-13, "grad_norm": 0.6915969848632812, "learning_rate": 3.120014589066086e-05, "loss": 2.2382, "step": 58880 }, { "epoch": 2.0431573100054834e-13, "grad_norm": 0.7070938348770142, "learning_rate": 3.118349379410907e-05, "loss": 2.2206, "step": 58890 }, { "epoch": 2.0435042547006788e-13, "grad_norm": 0.6955183744430542, "learning_rate": 3.116684763837683e-05, "loss": 2.2123, "step": 58900 }, { "epoch": 2.043851199395874e-13, "grad_norm": 0.664555013179779, "learning_rate": 3.115020742609664e-05, "loss": 2.2201, "step": 58910 }, { "epoch": 2.0441981440910695e-13, "grad_norm": 0.7283259034156799, "learning_rate": 3.113357315989998e-05, "loss": 2.2104, "step": 58920 }, { "epoch": 2.0445450887862648e-13, "grad_norm": 0.7228061556816101, "learning_rate": 3.1116944842417404e-05, "loss": 2.2155, "step": 58930 }, { "epoch": 2.0448920334814602e-13, "grad_norm": 0.6870076060295105, "learning_rate": 3.1100322476278594e-05, "loss": 2.195, "step": 58940 }, { "epoch": 2.0452389781766556e-13, "grad_norm": 0.6823621392250061, "learning_rate": 3.108370606411216e-05, "loss": 2.2019, "step": 58950 }, { "epoch": 2.045585922871851e-13, "grad_norm": 0.7057481408119202, "learning_rate": 3.106709560854591e-05, "loss": 2.1883, "step": 58960 }, { "epoch": 2.0459328675670463e-13, "grad_norm": 0.6758392453193665, "learning_rate": 3.105049111220656e-05, "loss": 2.2398, "step": 58970 }, { "epoch": 2.0462798122622416e-13, "grad_norm": 0.6990610361099243, "learning_rate": 3.103389257772006e-05, "loss": 2.2508, "step": 58980 }, { "epoch": 2.046626756957437e-13, "grad_norm": 0.6919876337051392, "learning_rate": 3.101730000771127e-05, "loss": 2.2351, "step": 58990 }, { "epoch": 2.0469737016526324e-13, "grad_norm": 0.6783318519592285, "learning_rate": 3.100071340480418e-05, "loss": 2.1827, "step": 59000 }, { "epoch": 2.0473206463478277e-13, "grad_norm": 0.683006227016449, "learning_rate": 3.0984132771621836e-05, "loss": 2.2185, "step": 59010 }, { "epoch": 2.047667591043023e-13, "grad_norm": 0.6870086789131165, "learning_rate": 3.096755811078628e-05, "loss": 2.2275, "step": 59020 }, { "epoch": 2.0480145357382185e-13, "grad_norm": 0.6994991898536682, "learning_rate": 3.0950989424918685e-05, "loss": 2.1634, "step": 59030 }, { "epoch": 2.0483614804334138e-13, "grad_norm": 0.6948642134666443, "learning_rate": 3.0934426716639236e-05, "loss": 2.2407, "step": 59040 }, { "epoch": 2.0487084251286092e-13, "grad_norm": 0.6986746788024902, "learning_rate": 3.0917869988567185e-05, "loss": 2.2183, "step": 59050 }, { "epoch": 2.0490553698238045e-13, "grad_norm": 0.7276952862739563, "learning_rate": 3.090131924332084e-05, "loss": 2.1979, "step": 59060 }, { "epoch": 2.049402314519e-13, "grad_norm": 0.6690629720687866, "learning_rate": 3.088477448351753e-05, "loss": 2.2092, "step": 59070 }, { "epoch": 2.0497492592141953e-13, "grad_norm": 0.7615671157836914, "learning_rate": 3.0868235711773704e-05, "loss": 2.1957, "step": 59080 }, { "epoch": 2.0500962039093906e-13, "grad_norm": 0.7134905457496643, "learning_rate": 3.085170293070482e-05, "loss": 2.203, "step": 59090 }, { "epoch": 2.050443148604586e-13, "grad_norm": 0.6968287825584412, "learning_rate": 3.083517614292538e-05, "loss": 2.2028, "step": 59100 }, { "epoch": 2.0507900932997813e-13, "grad_norm": 0.7269591093063354, "learning_rate": 3.081865535104896e-05, "loss": 2.1976, "step": 59110 }, { "epoch": 2.0511370379949767e-13, "grad_norm": 0.6820884346961975, "learning_rate": 3.080214055768818e-05, "loss": 2.2224, "step": 59120 }, { "epoch": 2.051483982690172e-13, "grad_norm": 0.6916916966438293, "learning_rate": 3.078563176545472e-05, "loss": 2.2486, "step": 59130 }, { "epoch": 2.0518309273853674e-13, "grad_norm": 0.6983402967453003, "learning_rate": 3.076912897695928e-05, "loss": 2.1917, "step": 59140 }, { "epoch": 2.0521778720805628e-13, "grad_norm": 0.6924554705619812, "learning_rate": 3.075263219481166e-05, "loss": 2.1872, "step": 59150 }, { "epoch": 2.0525248167757582e-13, "grad_norm": 0.6856098175048828, "learning_rate": 3.0736141421620676e-05, "loss": 2.1952, "step": 59160 }, { "epoch": 2.0528717614709535e-13, "grad_norm": 0.6850850582122803, "learning_rate": 3.0719656659994204e-05, "loss": 2.2089, "step": 59170 }, { "epoch": 2.053218706166149e-13, "grad_norm": 0.6864318251609802, "learning_rate": 3.070317791253916e-05, "loss": 2.156, "step": 59180 }, { "epoch": 2.0535656508613442e-13, "grad_norm": 0.7246891856193542, "learning_rate": 3.0686705181861524e-05, "loss": 2.2125, "step": 59190 }, { "epoch": 2.0539125955565396e-13, "grad_norm": 0.7036147713661194, "learning_rate": 3.0670238470566295e-05, "loss": 2.2193, "step": 59200 }, { "epoch": 2.054259540251735e-13, "grad_norm": 0.6893668174743652, "learning_rate": 3.065377778125759e-05, "loss": 2.1962, "step": 59210 }, { "epoch": 2.0546064849469303e-13, "grad_norm": 0.6779186129570007, "learning_rate": 3.063732311653846e-05, "loss": 2.2354, "step": 59220 }, { "epoch": 2.0549534296421257e-13, "grad_norm": 0.6816940307617188, "learning_rate": 3.0620874479011124e-05, "loss": 2.1827, "step": 59230 }, { "epoch": 2.055300374337321e-13, "grad_norm": 0.690398633480072, "learning_rate": 3.060443187127677e-05, "loss": 2.1998, "step": 59240 }, { "epoch": 2.0556473190325164e-13, "grad_norm": 0.7058286070823669, "learning_rate": 3.058799529593564e-05, "loss": 2.2072, "step": 59250 }, { "epoch": 2.0559942637277118e-13, "grad_norm": 0.6871728897094727, "learning_rate": 3.0571564755587086e-05, "loss": 2.1955, "step": 59260 }, { "epoch": 2.0563412084229071e-13, "grad_norm": 0.651593029499054, "learning_rate": 3.0555140252829385e-05, "loss": 2.1964, "step": 59270 }, { "epoch": 2.0566881531181025e-13, "grad_norm": 0.6814644932746887, "learning_rate": 3.053872179026001e-05, "loss": 2.2371, "step": 59280 }, { "epoch": 2.0570350978132979e-13, "grad_norm": 0.6974759101867676, "learning_rate": 3.052230937047533e-05, "loss": 2.2134, "step": 59290 }, { "epoch": 2.0573820425084932e-13, "grad_norm": 0.6554661989212036, "learning_rate": 3.0505902996070862e-05, "loss": 2.2107, "step": 59300 }, { "epoch": 2.0577289872036886e-13, "grad_norm": 0.67425537109375, "learning_rate": 3.0489502669641133e-05, "loss": 2.2227, "step": 59310 }, { "epoch": 2.058075931898884e-13, "grad_norm": 0.6935015916824341, "learning_rate": 3.0473108393779684e-05, "loss": 2.2006, "step": 59320 }, { "epoch": 2.0584228765940793e-13, "grad_norm": 0.6763274073600769, "learning_rate": 3.04567201710792e-05, "loss": 2.2165, "step": 59330 }, { "epoch": 2.0587698212892747e-13, "grad_norm": 0.696061909198761, "learning_rate": 3.0440338004131245e-05, "loss": 2.2338, "step": 59340 }, { "epoch": 2.05911676598447e-13, "grad_norm": 0.7295688390731812, "learning_rate": 3.042396189552659e-05, "loss": 2.2362, "step": 59350 }, { "epoch": 2.0594637106796654e-13, "grad_norm": 0.6909222602844238, "learning_rate": 3.040759184785494e-05, "loss": 2.1725, "step": 59360 }, { "epoch": 2.0598106553748607e-13, "grad_norm": 0.6917080283164978, "learning_rate": 3.0391227863705092e-05, "loss": 2.2214, "step": 59370 }, { "epoch": 2.060157600070056e-13, "grad_norm": 0.7240514159202576, "learning_rate": 3.0374869945664857e-05, "loss": 2.235, "step": 59380 }, { "epoch": 2.0605045447652515e-13, "grad_norm": 0.6796057224273682, "learning_rate": 3.0358518096321087e-05, "loss": 2.2274, "step": 59390 }, { "epoch": 2.0608514894604468e-13, "grad_norm": 0.6871556639671326, "learning_rate": 3.0342172318259727e-05, "loss": 2.2236, "step": 59400 }, { "epoch": 2.0611984341556422e-13, "grad_norm": 0.7028726935386658, "learning_rate": 3.0325832614065688e-05, "loss": 2.2022, "step": 59410 }, { "epoch": 2.0615453788508376e-13, "grad_norm": 0.7124908566474915, "learning_rate": 3.0309498986322968e-05, "loss": 2.239, "step": 59420 }, { "epoch": 2.061892323546033e-13, "grad_norm": 0.7026295065879822, "learning_rate": 3.0293171437614592e-05, "loss": 2.2078, "step": 59430 }, { "epoch": 2.0622392682412283e-13, "grad_norm": 0.7244258522987366, "learning_rate": 3.0276849970522608e-05, "loss": 2.2193, "step": 59440 }, { "epoch": 2.0625862129364236e-13, "grad_norm": 0.6993584036827087, "learning_rate": 3.026053458762813e-05, "loss": 2.277, "step": 59450 }, { "epoch": 2.062933157631619e-13, "grad_norm": 0.7365899085998535, "learning_rate": 3.024422529151126e-05, "loss": 2.2034, "step": 59460 }, { "epoch": 2.0632801023268144e-13, "grad_norm": 0.7132790684700012, "learning_rate": 3.022792208475122e-05, "loss": 2.2549, "step": 59470 }, { "epoch": 2.0636270470220097e-13, "grad_norm": 0.7138282656669617, "learning_rate": 3.021162496992621e-05, "loss": 2.2248, "step": 59480 }, { "epoch": 2.063973991717205e-13, "grad_norm": 0.7041279673576355, "learning_rate": 3.0195333949613464e-05, "loss": 2.2445, "step": 59490 }, { "epoch": 2.0643209364124004e-13, "grad_norm": 0.6842527389526367, "learning_rate": 3.017904902638927e-05, "loss": 2.2051, "step": 59500 }, { "epoch": 2.0646678811075958e-13, "grad_norm": 0.6996809244155884, "learning_rate": 3.0162770202828954e-05, "loss": 2.1892, "step": 59510 }, { "epoch": 2.0650148258027912e-13, "grad_norm": 0.7173580527305603, "learning_rate": 3.014649748150685e-05, "loss": 2.244, "step": 59520 }, { "epoch": 2.0653617704979865e-13, "grad_norm": 0.7261619567871094, "learning_rate": 3.013023086499639e-05, "loss": 2.2088, "step": 59530 }, { "epoch": 2.065708715193182e-13, "grad_norm": 0.6811491250991821, "learning_rate": 3.0113970355869974e-05, "loss": 2.2265, "step": 59540 }, { "epoch": 2.0660556598883772e-13, "grad_norm": 0.71058589220047, "learning_rate": 3.0097715956699064e-05, "loss": 2.2291, "step": 59550 }, { "epoch": 2.0664026045835726e-13, "grad_norm": 0.7121955752372742, "learning_rate": 3.008146767005416e-05, "loss": 2.19, "step": 59560 }, { "epoch": 2.066749549278768e-13, "grad_norm": 0.6730297803878784, "learning_rate": 3.006522549850476e-05, "loss": 2.2258, "step": 59570 }, { "epoch": 2.0670964939739633e-13, "grad_norm": 0.6941480040550232, "learning_rate": 3.0048989444619486e-05, "loss": 2.2284, "step": 59580 }, { "epoch": 2.0674434386691587e-13, "grad_norm": 0.7246027588844299, "learning_rate": 3.003275951096585e-05, "loss": 2.2151, "step": 59590 }, { "epoch": 2.067790383364354e-13, "grad_norm": 0.7606185078620911, "learning_rate": 3.001653570011056e-05, "loss": 2.2364, "step": 59600 }, { "epoch": 2.0681373280595494e-13, "grad_norm": 0.7066753506660461, "learning_rate": 3.0000318014619183e-05, "loss": 2.226, "step": 59610 }, { "epoch": 2.0684842727547448e-13, "grad_norm": 0.7057150602340698, "learning_rate": 2.998410645705648e-05, "loss": 2.2269, "step": 59620 }, { "epoch": 2.0688312174499401e-13, "grad_norm": 0.6794722080230713, "learning_rate": 2.996790102998614e-05, "loss": 2.2256, "step": 59630 }, { "epoch": 2.0691781621451355e-13, "grad_norm": 0.6995885968208313, "learning_rate": 2.9951701735970888e-05, "loss": 2.2215, "step": 59640 }, { "epoch": 2.0695251068403309e-13, "grad_norm": 0.708159327507019, "learning_rate": 2.993550857757257e-05, "loss": 2.2094, "step": 59650 }, { "epoch": 2.0698720515355262e-13, "grad_norm": 0.7382416129112244, "learning_rate": 2.991932155735192e-05, "loss": 2.2333, "step": 59660 }, { "epoch": 2.0702189962307216e-13, "grad_norm": 0.714377760887146, "learning_rate": 2.990314067786882e-05, "loss": 2.2417, "step": 59670 }, { "epoch": 2.070565940925917e-13, "grad_norm": 0.6732675433158875, "learning_rate": 2.9886965941682135e-05, "loss": 2.1818, "step": 59680 }, { "epoch": 2.0709128856211123e-13, "grad_norm": 0.6936337351799011, "learning_rate": 2.9870797351349745e-05, "loss": 2.1987, "step": 59690 }, { "epoch": 2.0712598303163077e-13, "grad_norm": 0.6862123608589172, "learning_rate": 2.9854634909428576e-05, "loss": 2.2056, "step": 59700 }, { "epoch": 2.071606775011503e-13, "grad_norm": 0.7148523926734924, "learning_rate": 2.983847861847457e-05, "loss": 2.2359, "step": 59710 }, { "epoch": 2.0719537197066984e-13, "grad_norm": 0.7202795147895813, "learning_rate": 2.9822328481042734e-05, "loss": 2.1904, "step": 59720 }, { "epoch": 2.0723006644018938e-13, "grad_norm": 0.6942813396453857, "learning_rate": 2.980618449968706e-05, "loss": 2.2308, "step": 59730 }, { "epoch": 2.072647609097089e-13, "grad_norm": 0.7125425338745117, "learning_rate": 2.9790046676960573e-05, "loss": 2.189, "step": 59740 }, { "epoch": 2.0729945537922845e-13, "grad_norm": 0.7101068496704102, "learning_rate": 2.977391501541533e-05, "loss": 2.2272, "step": 59750 }, { "epoch": 2.0733414984874798e-13, "grad_norm": 0.707349419593811, "learning_rate": 2.975778951760243e-05, "loss": 2.204, "step": 59760 }, { "epoch": 2.0736884431826752e-13, "grad_norm": 0.6962711811065674, "learning_rate": 2.974167018607196e-05, "loss": 2.2466, "step": 59770 }, { "epoch": 2.0740353878778706e-13, "grad_norm": 0.6986240148544312, "learning_rate": 2.972555702337306e-05, "loss": 2.2161, "step": 59780 }, { "epoch": 2.074382332573066e-13, "grad_norm": 0.6951040625572205, "learning_rate": 2.9709450032053896e-05, "loss": 2.1888, "step": 59790 }, { "epoch": 2.0747292772682613e-13, "grad_norm": 0.7193922400474548, "learning_rate": 2.9693349214661652e-05, "loss": 2.2354, "step": 59800 }, { "epoch": 2.0750762219634566e-13, "grad_norm": 0.7158114314079285, "learning_rate": 2.9677254573742526e-05, "loss": 2.246, "step": 59810 }, { "epoch": 2.075423166658652e-13, "grad_norm": 0.7136943340301514, "learning_rate": 2.9661166111841746e-05, "loss": 2.2054, "step": 59820 }, { "epoch": 2.0757701113538474e-13, "grad_norm": 0.691861093044281, "learning_rate": 2.964508383150357e-05, "loss": 2.2285, "step": 59830 }, { "epoch": 2.0761170560490427e-13, "grad_norm": 0.6968216300010681, "learning_rate": 2.9629007735271256e-05, "loss": 2.2045, "step": 59840 }, { "epoch": 2.076464000744238e-13, "grad_norm": 0.6968995332717896, "learning_rate": 2.961293782568712e-05, "loss": 2.2209, "step": 59850 }, { "epoch": 2.0768109454394335e-13, "grad_norm": 0.6604534387588501, "learning_rate": 2.9596874105292484e-05, "loss": 2.2156, "step": 59860 }, { "epoch": 2.0771578901346288e-13, "grad_norm": 0.6763586401939392, "learning_rate": 2.9580816576627667e-05, "loss": 2.1845, "step": 59870 }, { "epoch": 2.0775048348298242e-13, "grad_norm": 0.7193185091018677, "learning_rate": 2.9564765242232036e-05, "loss": 2.2308, "step": 59880 }, { "epoch": 2.0778517795250195e-13, "grad_norm": 0.7303974032402039, "learning_rate": 2.9548720104643963e-05, "loss": 2.1759, "step": 59890 }, { "epoch": 2.078198724220215e-13, "grad_norm": 0.7189143300056458, "learning_rate": 2.953268116640089e-05, "loss": 2.222, "step": 59900 }, { "epoch": 2.0785456689154103e-13, "grad_norm": 0.7440142035484314, "learning_rate": 2.9516648430039175e-05, "loss": 2.2446, "step": 59910 }, { "epoch": 2.0788926136106056e-13, "grad_norm": 0.7006040811538696, "learning_rate": 2.9500621898094303e-05, "loss": 2.2226, "step": 59920 }, { "epoch": 2.079239558305801e-13, "grad_norm": 0.6784969568252563, "learning_rate": 2.9484601573100708e-05, "loss": 2.1705, "step": 59930 }, { "epoch": 2.0795865030009963e-13, "grad_norm": 0.6787081956863403, "learning_rate": 2.9468587457591886e-05, "loss": 2.1659, "step": 59940 }, { "epoch": 2.0799334476961917e-13, "grad_norm": 0.7283097505569458, "learning_rate": 2.945257955410032e-05, "loss": 2.2248, "step": 59950 }, { "epoch": 2.080280392391387e-13, "grad_norm": 0.6857704520225525, "learning_rate": 2.9436577865157496e-05, "loss": 2.2001, "step": 59960 }, { "epoch": 2.0806273370865824e-13, "grad_norm": 0.7324347496032715, "learning_rate": 2.9420582393294018e-05, "loss": 2.2178, "step": 59970 }, { "epoch": 2.0809742817817778e-13, "grad_norm": 0.7189435958862305, "learning_rate": 2.940459314103934e-05, "loss": 2.2034, "step": 59980 }, { "epoch": 2.0813212264769732e-13, "grad_norm": 0.6626642942428589, "learning_rate": 2.938861011092209e-05, "loss": 2.1987, "step": 59990 }, { "epoch": 2.0816681711721685e-13, "grad_norm": 0.7461784482002258, "learning_rate": 2.9372633305469833e-05, "loss": 2.2139, "step": 60000 } ], "logging_steps": 10, "max_steps": 80000, "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": 3.528027468201984e+18, "train_batch_size": 8, "trial_name": null, "trial_params": null }