{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0177259543895294, "eval_steps": 500, "global_step": 29513, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0005889021391870206, "grad_norm": 0.30055442452430725, "learning_rate": 1.8e-07, "loss": 0.593233871459961, "step": 10 }, { "epoch": 0.0011778042783740412, "grad_norm": 0.25422927737236023, "learning_rate": 3.8e-07, "loss": 0.6708829402923584, "step": 20 }, { "epoch": 0.0017667064175610617, "grad_norm": 0.32899609208106995, "learning_rate": 5.800000000000001e-07, "loss": 0.5315020084381104, "step": 30 }, { "epoch": 0.0023556085567480825, "grad_norm": 0.3091534376144409, "learning_rate": 7.8e-07, "loss": 0.5945694923400879, "step": 40 }, { "epoch": 0.002944510695935103, "grad_norm": 0.45176923274993896, "learning_rate": 9.800000000000001e-07, "loss": 0.6226115703582764, "step": 50 }, { "epoch": 0.0035334128351221235, "grad_norm": 0.4495423436164856, "learning_rate": 1.1800000000000001e-06, "loss": 0.4949666500091553, "step": 60 }, { "epoch": 0.004122314974309144, "grad_norm": 0.3979267477989197, "learning_rate": 1.3800000000000001e-06, "loss": 0.5676890850067139, "step": 70 }, { "epoch": 0.004711217113496165, "grad_norm": 0.2633461654186249, "learning_rate": 1.5800000000000001e-06, "loss": 0.5029969692230225, "step": 80 }, { "epoch": 0.005300119252683185, "grad_norm": 0.3449569344520569, "learning_rate": 1.7800000000000001e-06, "loss": 0.5873546600341797, "step": 90 }, { "epoch": 0.005889021391870206, "grad_norm": 0.22054193913936615, "learning_rate": 1.98e-06, "loss": 0.555787181854248, "step": 100 }, { "epoch": 0.006477923531057227, "grad_norm": 0.2452852427959442, "learning_rate": 2.1800000000000003e-06, "loss": 0.5232597351074219, "step": 110 }, { "epoch": 0.007066825670244247, "grad_norm": 0.16745756566524506, "learning_rate": 2.38e-06, "loss": 0.5179643630981445, "step": 120 }, { "epoch": 0.007655727809431268, "grad_norm": 0.28422579169273376, "learning_rate": 2.5800000000000003e-06, "loss": 0.5884719848632812, "step": 130 }, { "epoch": 0.008244629948618288, "grad_norm": 0.165652796626091, "learning_rate": 2.7800000000000005e-06, "loss": 0.6129399299621582, "step": 140 }, { "epoch": 0.00883353208780531, "grad_norm": 0.16147030889987946, "learning_rate": 2.9800000000000003e-06, "loss": 0.549650001525879, "step": 150 }, { "epoch": 0.00942243422699233, "grad_norm": 0.2225613296031952, "learning_rate": 3.1800000000000005e-06, "loss": 0.573492956161499, "step": 160 }, { "epoch": 0.01001133636617935, "grad_norm": 0.2057325392961502, "learning_rate": 3.3800000000000007e-06, "loss": 0.49120163917541504, "step": 170 }, { "epoch": 0.01060023850536637, "grad_norm": 0.2145884782075882, "learning_rate": 3.58e-06, "loss": 0.4865445613861084, "step": 180 }, { "epoch": 0.011189140644553392, "grad_norm": 0.14130713045597076, "learning_rate": 3.7800000000000002e-06, "loss": 0.5594169139862061, "step": 190 }, { "epoch": 0.011778042783740412, "grad_norm": 0.1634829342365265, "learning_rate": 3.980000000000001e-06, "loss": 0.4612868785858154, "step": 200 }, { "epoch": 0.012366944922927432, "grad_norm": 0.16171133518218994, "learning_rate": 4.18e-06, "loss": 0.4875588893890381, "step": 210 }, { "epoch": 0.012955847062114454, "grad_norm": 0.18812493979930878, "learning_rate": 4.38e-06, "loss": 0.49331092834472656, "step": 220 }, { "epoch": 0.013544749201301474, "grad_norm": 0.1633215844631195, "learning_rate": 4.58e-06, "loss": 0.474062442779541, "step": 230 }, { "epoch": 0.014133651340488494, "grad_norm": 0.14991550147533417, "learning_rate": 4.78e-06, "loss": 0.4770942687988281, "step": 240 }, { "epoch": 0.014722553479675516, "grad_norm": 0.21784667670726776, "learning_rate": 4.980000000000001e-06, "loss": 0.4441679954528809, "step": 250 }, { "epoch": 0.015311455618862536, "grad_norm": 0.15078988671302795, "learning_rate": 5.18e-06, "loss": 0.4766075611114502, "step": 260 }, { "epoch": 0.015900357758049558, "grad_norm": 0.14167527854442596, "learning_rate": 5.380000000000001e-06, "loss": 0.43286657333374023, "step": 270 }, { "epoch": 0.016489259897236576, "grad_norm": 0.18567802011966705, "learning_rate": 5.580000000000001e-06, "loss": 0.4119394302368164, "step": 280 }, { "epoch": 0.017078162036423598, "grad_norm": 0.17662471532821655, "learning_rate": 5.78e-06, "loss": 0.4863849639892578, "step": 290 }, { "epoch": 0.01766706417561062, "grad_norm": 0.15525980293750763, "learning_rate": 5.98e-06, "loss": 0.4440967559814453, "step": 300 }, { "epoch": 0.018255966314797638, "grad_norm": 0.11325185745954514, "learning_rate": 6.18e-06, "loss": 0.4640179634094238, "step": 310 }, { "epoch": 0.01884486845398466, "grad_norm": 0.17372415959835052, "learning_rate": 6.380000000000001e-06, "loss": 0.3911218881607056, "step": 320 }, { "epoch": 0.01943377059317168, "grad_norm": 0.1281571388244629, "learning_rate": 6.5800000000000005e-06, "loss": 0.39266138076782225, "step": 330 }, { "epoch": 0.0200226727323587, "grad_norm": 0.12149865925312042, "learning_rate": 6.780000000000001e-06, "loss": 0.4214672088623047, "step": 340 }, { "epoch": 0.02061157487154572, "grad_norm": 0.1575547754764557, "learning_rate": 6.98e-06, "loss": 0.4417559623718262, "step": 350 }, { "epoch": 0.02120047701073274, "grad_norm": 0.11566848307847977, "learning_rate": 7.180000000000001e-06, "loss": 0.3897959470748901, "step": 360 }, { "epoch": 0.02178937914991976, "grad_norm": 0.18317797780036926, "learning_rate": 7.3800000000000005e-06, "loss": 0.38766677379608155, "step": 370 }, { "epoch": 0.022378281289106784, "grad_norm": 0.1508084237575531, "learning_rate": 7.58e-06, "loss": 0.4639895915985107, "step": 380 }, { "epoch": 0.022967183428293802, "grad_norm": 0.16696086525917053, "learning_rate": 7.78e-06, "loss": 0.3851975440979004, "step": 390 }, { "epoch": 0.023556085567480824, "grad_norm": 0.18619266152381897, "learning_rate": 7.980000000000002e-06, "loss": 0.3914435863494873, "step": 400 }, { "epoch": 0.024144987706667845, "grad_norm": 0.1889142096042633, "learning_rate": 8.18e-06, "loss": 0.3851407289505005, "step": 410 }, { "epoch": 0.024733889845854864, "grad_norm": 0.23976512253284454, "learning_rate": 8.380000000000001e-06, "loss": 0.4295720100402832, "step": 420 }, { "epoch": 0.025322791985041886, "grad_norm": 0.15691791474819183, "learning_rate": 8.580000000000001e-06, "loss": 0.3598081827163696, "step": 430 }, { "epoch": 0.025911694124228907, "grad_norm": 0.12324974685907364, "learning_rate": 8.78e-06, "loss": 0.39954218864440916, "step": 440 }, { "epoch": 0.026500596263415926, "grad_norm": 0.22567057609558105, "learning_rate": 8.98e-06, "loss": 0.4399960994720459, "step": 450 }, { "epoch": 0.027089498402602948, "grad_norm": 0.13568201661109924, "learning_rate": 9.180000000000002e-06, "loss": 0.39374141693115233, "step": 460 }, { "epoch": 0.02767840054178997, "grad_norm": 0.30869171023368835, "learning_rate": 9.38e-06, "loss": 0.46657428741455076, "step": 470 }, { "epoch": 0.028267302680976988, "grad_norm": 0.1977085918188095, "learning_rate": 9.58e-06, "loss": 0.38493053913116454, "step": 480 }, { "epoch": 0.02885620482016401, "grad_norm": 0.202088862657547, "learning_rate": 9.780000000000001e-06, "loss": 0.43812108039855957, "step": 490 }, { "epoch": 0.02944510695935103, "grad_norm": 0.2875714898109436, "learning_rate": 9.980000000000001e-06, "loss": 0.45606083869934083, "step": 500 }, { "epoch": 0.03003400909853805, "grad_norm": 0.1783527284860611, "learning_rate": 1.018e-05, "loss": 0.3775629043579102, "step": 510 }, { "epoch": 0.03062291123772507, "grad_norm": 0.2886123061180115, "learning_rate": 1.038e-05, "loss": 0.44734859466552734, "step": 520 }, { "epoch": 0.031211813376912093, "grad_norm": 0.24763664603233337, "learning_rate": 1.0580000000000002e-05, "loss": 0.39458038806915285, "step": 530 }, { "epoch": 0.031800715516099115, "grad_norm": 0.19266977906227112, "learning_rate": 1.0780000000000002e-05, "loss": 0.35241677761077883, "step": 540 }, { "epoch": 0.03238961765528613, "grad_norm": 0.20720165967941284, "learning_rate": 1.0980000000000002e-05, "loss": 0.383864426612854, "step": 550 }, { "epoch": 0.03297851979447315, "grad_norm": 0.21185018122196198, "learning_rate": 1.1180000000000001e-05, "loss": 0.37051801681518554, "step": 560 }, { "epoch": 0.03356742193366018, "grad_norm": 0.1749272495508194, "learning_rate": 1.138e-05, "loss": 0.3536161661148071, "step": 570 }, { "epoch": 0.034156324072847195, "grad_norm": 0.16397960484027863, "learning_rate": 1.1580000000000001e-05, "loss": 0.39590160846710204, "step": 580 }, { "epoch": 0.034745226212034214, "grad_norm": 0.8980286121368408, "learning_rate": 1.178e-05, "loss": 0.41092305183410643, "step": 590 }, { "epoch": 0.03533412835122124, "grad_norm": 0.30533626675605774, "learning_rate": 1.198e-05, "loss": 0.42215800285339355, "step": 600 }, { "epoch": 0.03592303049040826, "grad_norm": 0.1995972841978073, "learning_rate": 1.218e-05, "loss": 0.41655683517456055, "step": 610 }, { "epoch": 0.036511932629595276, "grad_norm": 0.18818414211273193, "learning_rate": 1.2380000000000002e-05, "loss": 0.4264305591583252, "step": 620 }, { "epoch": 0.0371008347687823, "grad_norm": 0.20818759500980377, "learning_rate": 1.2580000000000002e-05, "loss": 0.3974606037139893, "step": 630 }, { "epoch": 0.03768973690796932, "grad_norm": 0.26332953572273254, "learning_rate": 1.2780000000000001e-05, "loss": 0.39855568408966063, "step": 640 }, { "epoch": 0.03827863904715634, "grad_norm": 0.315928190946579, "learning_rate": 1.2980000000000001e-05, "loss": 0.4730992794036865, "step": 650 }, { "epoch": 0.03886754118634336, "grad_norm": 0.3363407850265503, "learning_rate": 1.3180000000000001e-05, "loss": 0.3759429454803467, "step": 660 }, { "epoch": 0.03945644332553038, "grad_norm": 0.29042065143585205, "learning_rate": 1.3380000000000002e-05, "loss": 0.43245887756347656, "step": 670 }, { "epoch": 0.0400453454647174, "grad_norm": 0.24075721204280853, "learning_rate": 1.3580000000000002e-05, "loss": 0.45029253959655763, "step": 680 }, { "epoch": 0.04063424760390442, "grad_norm": 0.26263678073883057, "learning_rate": 1.378e-05, "loss": 0.3834373474121094, "step": 690 }, { "epoch": 0.04122314974309144, "grad_norm": 0.2820330262184143, "learning_rate": 1.398e-05, "loss": 0.3908398151397705, "step": 700 }, { "epoch": 0.04181205188227846, "grad_norm": 0.26546624302864075, "learning_rate": 1.418e-05, "loss": 0.3614314079284668, "step": 710 }, { "epoch": 0.04240095402146548, "grad_norm": 0.26427915692329407, "learning_rate": 1.4380000000000001e-05, "loss": 0.3988569021224976, "step": 720 }, { "epoch": 0.042989856160652505, "grad_norm": 0.35622239112854004, "learning_rate": 1.4580000000000001e-05, "loss": 0.3719624996185303, "step": 730 }, { "epoch": 0.04357875829983952, "grad_norm": 0.20107628405094147, "learning_rate": 1.478e-05, "loss": 0.3983035087585449, "step": 740 }, { "epoch": 0.04416766043902654, "grad_norm": 0.2417820543050766, "learning_rate": 1.498e-05, "loss": 0.4147758483886719, "step": 750 }, { "epoch": 0.04475656257821357, "grad_norm": 0.31352657079696655, "learning_rate": 1.5180000000000002e-05, "loss": 0.4093564510345459, "step": 760 }, { "epoch": 0.045345464717400585, "grad_norm": 0.24956892430782318, "learning_rate": 1.5380000000000002e-05, "loss": 0.38494172096252444, "step": 770 }, { "epoch": 0.045934366856587604, "grad_norm": 0.26918911933898926, "learning_rate": 1.5580000000000003e-05, "loss": 0.35888545513153075, "step": 780 }, { "epoch": 0.04652326899577463, "grad_norm": 0.35767316818237305, "learning_rate": 1.578e-05, "loss": 0.42031660079956057, "step": 790 }, { "epoch": 0.04711217113496165, "grad_norm": 0.23957036435604095, "learning_rate": 1.5980000000000003e-05, "loss": 0.3810563325881958, "step": 800 }, { "epoch": 0.047701073274148666, "grad_norm": 0.3439910113811493, "learning_rate": 1.618e-05, "loss": 0.3902775287628174, "step": 810 }, { "epoch": 0.04828997541333569, "grad_norm": 0.37484967708587646, "learning_rate": 1.638e-05, "loss": 0.44611191749572754, "step": 820 }, { "epoch": 0.04887887755252271, "grad_norm": 0.41926881670951843, "learning_rate": 1.658e-05, "loss": 0.379734206199646, "step": 830 }, { "epoch": 0.04946777969170973, "grad_norm": 0.2716629207134247, "learning_rate": 1.6780000000000002e-05, "loss": 0.36273813247680664, "step": 840 }, { "epoch": 0.05005668183089675, "grad_norm": 0.25876402854919434, "learning_rate": 1.698e-05, "loss": 0.42391037940979004, "step": 850 }, { "epoch": 0.05064558397008377, "grad_norm": 0.32415834069252014, "learning_rate": 1.718e-05, "loss": 0.41620922088623047, "step": 860 }, { "epoch": 0.05123448610927079, "grad_norm": 0.3579925298690796, "learning_rate": 1.7380000000000003e-05, "loss": 0.3903594970703125, "step": 870 }, { "epoch": 0.051823388248457815, "grad_norm": 0.2536069452762604, "learning_rate": 1.758e-05, "loss": 0.33806312084198, "step": 880 }, { "epoch": 0.05241229038764483, "grad_norm": 0.2185448408126831, "learning_rate": 1.7780000000000003e-05, "loss": 0.35108771324157717, "step": 890 }, { "epoch": 0.05300119252683185, "grad_norm": 0.27737244963645935, "learning_rate": 1.798e-05, "loss": 0.34936952590942383, "step": 900 }, { "epoch": 0.05359009466601888, "grad_norm": 0.2531609833240509, "learning_rate": 1.8180000000000002e-05, "loss": 0.3414201259613037, "step": 910 }, { "epoch": 0.054178996805205895, "grad_norm": 0.25409701466560364, "learning_rate": 1.8380000000000004e-05, "loss": 0.36421785354614256, "step": 920 }, { "epoch": 0.05476789894439291, "grad_norm": 0.5448822379112244, "learning_rate": 1.858e-05, "loss": 0.3743453502655029, "step": 930 }, { "epoch": 0.05535680108357994, "grad_norm": 0.32449638843536377, "learning_rate": 1.878e-05, "loss": 0.4309371471405029, "step": 940 }, { "epoch": 0.05594570322276696, "grad_norm": 0.30742713809013367, "learning_rate": 1.898e-05, "loss": 0.3893160581588745, "step": 950 }, { "epoch": 0.056534605361953975, "grad_norm": 0.44779667258262634, "learning_rate": 1.918e-05, "loss": 0.3879957675933838, "step": 960 }, { "epoch": 0.057123507501141, "grad_norm": 0.34386008977890015, "learning_rate": 1.938e-05, "loss": 0.3768059968948364, "step": 970 }, { "epoch": 0.05771240964032802, "grad_norm": 0.20766432583332062, "learning_rate": 1.9580000000000002e-05, "loss": 0.3719948291778564, "step": 980 }, { "epoch": 0.05830131177951504, "grad_norm": 0.32065755128860474, "learning_rate": 1.978e-05, "loss": 0.3994438171386719, "step": 990 }, { "epoch": 0.05889021391870206, "grad_norm": 0.24154752492904663, "learning_rate": 1.9980000000000002e-05, "loss": 0.35357491970062255, "step": 1000 }, { "epoch": 0.05947911605788908, "grad_norm": 0.35957077145576477, "learning_rate": 1.9999996999670047e-05, "loss": 0.3719403028488159, "step": 1010 }, { "epoch": 0.0600680181970761, "grad_norm": 0.3171429932117462, "learning_rate": 1.9999986628161402e-05, "loss": 0.40247883796691897, "step": 1020 }, { "epoch": 0.060656920336263125, "grad_norm": 0.5313934683799744, "learning_rate": 1.9999968848440642e-05, "loss": 0.3755397081375122, "step": 1030 }, { "epoch": 0.06124582247545014, "grad_norm": 0.2482350915670395, "learning_rate": 1.9999943660520933e-05, "loss": 0.3522895097732544, "step": 1040 }, { "epoch": 0.06183472461463716, "grad_norm": 0.2722140848636627, "learning_rate": 1.9999911064420936e-05, "loss": 0.35248892307281493, "step": 1050 }, { "epoch": 0.062423626753824187, "grad_norm": 0.3374186158180237, "learning_rate": 1.9999871060164803e-05, "loss": 0.35298824310302734, "step": 1060 }, { "epoch": 0.0630125288930112, "grad_norm": 0.27511799335479736, "learning_rate": 1.9999823647782162e-05, "loss": 0.35006740093231203, "step": 1070 }, { "epoch": 0.06360143103219823, "grad_norm": 0.25159454345703125, "learning_rate": 1.999976882730815e-05, "loss": 0.3835632562637329, "step": 1080 }, { "epoch": 0.06419033317138524, "grad_norm": 0.41270872950553894, "learning_rate": 1.999970659878337e-05, "loss": 0.37142493724823, "step": 1090 }, { "epoch": 0.06477923531057227, "grad_norm": 0.2009297013282776, "learning_rate": 1.999963696225392e-05, "loss": 0.4300539016723633, "step": 1100 }, { "epoch": 0.06536813744975929, "grad_norm": 0.2561275362968445, "learning_rate": 1.9999559917771395e-05, "loss": 0.3584117412567139, "step": 1110 }, { "epoch": 0.0659570395889463, "grad_norm": 0.3519864082336426, "learning_rate": 1.9999475465392867e-05, "loss": 0.37747583389282224, "step": 1120 }, { "epoch": 0.06654594172813333, "grad_norm": 0.48844802379608154, "learning_rate": 1.99993836051809e-05, "loss": 0.3504490375518799, "step": 1130 }, { "epoch": 0.06713484386732035, "grad_norm": 0.2759878933429718, "learning_rate": 1.999928433720355e-05, "loss": 0.3575859308242798, "step": 1140 }, { "epoch": 0.06772374600650737, "grad_norm": 0.3054344058036804, "learning_rate": 1.999917766153436e-05, "loss": 0.39182143211364745, "step": 1150 }, { "epoch": 0.06831264814569439, "grad_norm": 0.2878853678703308, "learning_rate": 1.9999063578252342e-05, "loss": 0.39737012386322024, "step": 1160 }, { "epoch": 0.06890155028488142, "grad_norm": 0.28099119663238525, "learning_rate": 1.9998942087442027e-05, "loss": 0.39304533004760744, "step": 1170 }, { "epoch": 0.06949045242406843, "grad_norm": 0.30117881298065186, "learning_rate": 1.9998813189193416e-05, "loss": 0.40151357650756836, "step": 1180 }, { "epoch": 0.07007935456325545, "grad_norm": 0.35916316509246826, "learning_rate": 1.9998676883601992e-05, "loss": 0.35724358558654784, "step": 1190 }, { "epoch": 0.07066825670244248, "grad_norm": 0.2696571350097656, "learning_rate": 1.9998533170768736e-05, "loss": 0.35288252830505373, "step": 1200 }, { "epoch": 0.07125715884162949, "grad_norm": 0.21914415061473846, "learning_rate": 1.999838205080012e-05, "loss": 0.43004298210144043, "step": 1210 }, { "epoch": 0.07184606098081651, "grad_norm": 0.2637389302253723, "learning_rate": 1.9998223523808092e-05, "loss": 0.4122148990631104, "step": 1220 }, { "epoch": 0.07243496312000354, "grad_norm": 0.4856111705303192, "learning_rate": 1.9998057589910092e-05, "loss": 0.38489413261413574, "step": 1230 }, { "epoch": 0.07302386525919055, "grad_norm": 0.3211883008480072, "learning_rate": 1.9997884249229047e-05, "loss": 0.3539762496948242, "step": 1240 }, { "epoch": 0.07361276739837758, "grad_norm": 0.23093311488628387, "learning_rate": 1.9997703501893376e-05, "loss": 0.3433551788330078, "step": 1250 }, { "epoch": 0.0742016695375646, "grad_norm": 0.3929930329322815, "learning_rate": 1.9997515348036972e-05, "loss": 0.3683741807937622, "step": 1260 }, { "epoch": 0.07479057167675161, "grad_norm": 0.32220542430877686, "learning_rate": 1.9997319787799233e-05, "loss": 0.3976790189743042, "step": 1270 }, { "epoch": 0.07537947381593864, "grad_norm": 0.35373228788375854, "learning_rate": 1.9997116821325032e-05, "loss": 0.32846252918243407, "step": 1280 }, { "epoch": 0.07596837595512566, "grad_norm": 0.23148596286773682, "learning_rate": 1.9996906448764725e-05, "loss": 0.3550188779830933, "step": 1290 }, { "epoch": 0.07655727809431268, "grad_norm": 0.25427037477493286, "learning_rate": 1.9996688670274168e-05, "loss": 0.3619978904724121, "step": 1300 }, { "epoch": 0.0771461802334997, "grad_norm": 0.29191234707832336, "learning_rate": 1.9996463486014694e-05, "loss": 0.35933504104614256, "step": 1310 }, { "epoch": 0.07773508237268673, "grad_norm": 0.3222062587738037, "learning_rate": 1.9996230896153125e-05, "loss": 0.39723160266876223, "step": 1320 }, { "epoch": 0.07832398451187374, "grad_norm": 0.3470778465270996, "learning_rate": 1.9995990900861765e-05, "loss": 0.3437675952911377, "step": 1330 }, { "epoch": 0.07891288665106076, "grad_norm": 0.30392149090766907, "learning_rate": 1.999574350031841e-05, "loss": 0.3057088375091553, "step": 1340 }, { "epoch": 0.07950178879024777, "grad_norm": 0.26725855469703674, "learning_rate": 1.9995488694706343e-05, "loss": 0.3208949089050293, "step": 1350 }, { "epoch": 0.0800906909294348, "grad_norm": 0.3819669485092163, "learning_rate": 1.9995226484214322e-05, "loss": 0.34355878829956055, "step": 1360 }, { "epoch": 0.08067959306862182, "grad_norm": 0.2797771394252777, "learning_rate": 1.9994956869036607e-05, "loss": 0.36545472145080565, "step": 1370 }, { "epoch": 0.08126849520780884, "grad_norm": 0.3064350485801697, "learning_rate": 1.999467984937293e-05, "loss": 0.366259503364563, "step": 1380 }, { "epoch": 0.08185739734699586, "grad_norm": 0.24379532039165497, "learning_rate": 1.9994395425428514e-05, "loss": 0.34901211261749265, "step": 1390 }, { "epoch": 0.08244629948618289, "grad_norm": 0.2608974874019623, "learning_rate": 1.999410359741407e-05, "loss": 0.3621335983276367, "step": 1400 }, { "epoch": 0.0830352016253699, "grad_norm": 0.5054655075073242, "learning_rate": 1.9993804365545785e-05, "loss": 0.3735924005508423, "step": 1410 }, { "epoch": 0.08362410376455692, "grad_norm": 0.27388259768486023, "learning_rate": 1.999349773004534e-05, "loss": 0.3486018180847168, "step": 1420 }, { "epoch": 0.08421300590374395, "grad_norm": 0.30909374356269836, "learning_rate": 1.9993183691139898e-05, "loss": 0.34521534442901614, "step": 1430 }, { "epoch": 0.08480190804293096, "grad_norm": 0.4512241780757904, "learning_rate": 1.9992862249062102e-05, "loss": 0.36939611434936526, "step": 1440 }, { "epoch": 0.08539081018211798, "grad_norm": 0.3617790639400482, "learning_rate": 1.9992533404050087e-05, "loss": 0.36021573543548585, "step": 1450 }, { "epoch": 0.08597971232130501, "grad_norm": 0.34313061833381653, "learning_rate": 1.999219715634747e-05, "loss": 0.3720792531967163, "step": 1460 }, { "epoch": 0.08656861446049202, "grad_norm": 0.31782153248786926, "learning_rate": 1.9991853506203348e-05, "loss": 0.4202549457550049, "step": 1470 }, { "epoch": 0.08715751659967905, "grad_norm": 0.31494858860969543, "learning_rate": 1.9991502453872304e-05, "loss": 0.37131671905517577, "step": 1480 }, { "epoch": 0.08774641873886607, "grad_norm": 0.26755189895629883, "learning_rate": 1.9991143999614406e-05, "loss": 0.32838404178619385, "step": 1490 }, { "epoch": 0.08833532087805308, "grad_norm": 0.23882579803466797, "learning_rate": 1.9990778143695207e-05, "loss": 0.3767066955566406, "step": 1500 }, { "epoch": 0.08892422301724011, "grad_norm": 0.3205873668193817, "learning_rate": 1.9990404886385737e-05, "loss": 0.33512287139892577, "step": 1510 }, { "epoch": 0.08951312515642713, "grad_norm": 0.5275142788887024, "learning_rate": 1.9990024227962518e-05, "loss": 0.4129493713378906, "step": 1520 }, { "epoch": 0.09010202729561415, "grad_norm": 0.5536600351333618, "learning_rate": 1.9989636168707548e-05, "loss": 0.3915761947631836, "step": 1530 }, { "epoch": 0.09069092943480117, "grad_norm": 0.37033963203430176, "learning_rate": 1.9989240708908312e-05, "loss": 0.392348575592041, "step": 1540 }, { "epoch": 0.0912798315739882, "grad_norm": 0.23174281418323517, "learning_rate": 1.9988837848857773e-05, "loss": 0.3706318140029907, "step": 1550 }, { "epoch": 0.09186873371317521, "grad_norm": 0.3149441182613373, "learning_rate": 1.9988427588854375e-05, "loss": 0.3773585557937622, "step": 1560 }, { "epoch": 0.09245763585236223, "grad_norm": 0.3513354957103729, "learning_rate": 1.9988009929202055e-05, "loss": 0.3904543161392212, "step": 1570 }, { "epoch": 0.09304653799154926, "grad_norm": 0.3202134370803833, "learning_rate": 1.9987584870210223e-05, "loss": 0.34647886753082274, "step": 1580 }, { "epoch": 0.09363544013073627, "grad_norm": 0.25048819184303284, "learning_rate": 1.9987152412193767e-05, "loss": 0.32433013916015624, "step": 1590 }, { "epoch": 0.0942243422699233, "grad_norm": 0.35656121373176575, "learning_rate": 1.9986712555473066e-05, "loss": 0.37526726722717285, "step": 1600 }, { "epoch": 0.09481324440911032, "grad_norm": 0.2917329967021942, "learning_rate": 1.9986265300373976e-05, "loss": 0.3636118650436401, "step": 1610 }, { "epoch": 0.09540214654829733, "grad_norm": 0.34469714760780334, "learning_rate": 1.998581064722783e-05, "loss": 0.32952446937561036, "step": 1620 }, { "epoch": 0.09599104868748436, "grad_norm": 0.32917824387550354, "learning_rate": 1.998534859637145e-05, "loss": 0.33797760009765626, "step": 1630 }, { "epoch": 0.09657995082667138, "grad_norm": 0.3244582414627075, "learning_rate": 1.9984879148147128e-05, "loss": 0.3548794984817505, "step": 1640 }, { "epoch": 0.0971688529658584, "grad_norm": 0.34238481521606445, "learning_rate": 1.9984402302902648e-05, "loss": 0.3642404079437256, "step": 1650 }, { "epoch": 0.09775775510504542, "grad_norm": 0.3359019160270691, "learning_rate": 1.9983918060991264e-05, "loss": 0.3804023265838623, "step": 1660 }, { "epoch": 0.09834665724423244, "grad_norm": 0.34736567735671997, "learning_rate": 1.9983426422771712e-05, "loss": 0.3747722148895264, "step": 1670 }, { "epoch": 0.09893555938341946, "grad_norm": 0.4304959177970886, "learning_rate": 1.9982927388608208e-05, "loss": 0.35581934452056885, "step": 1680 }, { "epoch": 0.09952446152260648, "grad_norm": 0.2786387801170349, "learning_rate": 1.998242095887045e-05, "loss": 0.32633798122406005, "step": 1690 }, { "epoch": 0.1001133636617935, "grad_norm": 0.3190578818321228, "learning_rate": 1.9981907133933614e-05, "loss": 0.37211883068084717, "step": 1700 }, { "epoch": 0.10070226580098052, "grad_norm": 0.3161371350288391, "learning_rate": 1.9981385914178347e-05, "loss": 0.35942854881286623, "step": 1710 }, { "epoch": 0.10129116794016754, "grad_norm": 0.28922584652900696, "learning_rate": 1.998085729999079e-05, "loss": 0.3097316026687622, "step": 1720 }, { "epoch": 0.10188007007935457, "grad_norm": 0.27709895372390747, "learning_rate": 1.998032129176254e-05, "loss": 0.31153972148895265, "step": 1730 }, { "epoch": 0.10246897221854158, "grad_norm": 0.6579906344413757, "learning_rate": 1.9979777889890692e-05, "loss": 0.39062230587005614, "step": 1740 }, { "epoch": 0.1030578743577286, "grad_norm": 0.38017305731773376, "learning_rate": 1.9979227094777807e-05, "loss": 0.37226507663726804, "step": 1750 }, { "epoch": 0.10364677649691563, "grad_norm": 0.22411757707595825, "learning_rate": 1.997866890683193e-05, "loss": 0.41033344268798827, "step": 1760 }, { "epoch": 0.10423567863610264, "grad_norm": 0.43651142716407776, "learning_rate": 1.9978103326466572e-05, "loss": 0.38497300148010255, "step": 1770 }, { "epoch": 0.10482458077528967, "grad_norm": 0.3064855635166168, "learning_rate": 1.9977530354100733e-05, "loss": 0.3386725664138794, "step": 1780 }, { "epoch": 0.10541348291447669, "grad_norm": 0.30971527099609375, "learning_rate": 1.9976949990158882e-05, "loss": 0.32645952701568604, "step": 1790 }, { "epoch": 0.1060023850536637, "grad_norm": 0.24303875863552094, "learning_rate": 1.997636223507097e-05, "loss": 0.31678769588470457, "step": 1800 }, { "epoch": 0.10659128719285073, "grad_norm": 0.2994871437549591, "learning_rate": 1.997576708927241e-05, "loss": 0.3940234661102295, "step": 1810 }, { "epoch": 0.10718018933203775, "grad_norm": 0.26071974635124207, "learning_rate": 1.9975164553204106e-05, "loss": 0.3652027130126953, "step": 1820 }, { "epoch": 0.10776909147122476, "grad_norm": 0.36072787642478943, "learning_rate": 1.997455462731243e-05, "loss": 0.3681498527526855, "step": 1830 }, { "epoch": 0.10835799361041179, "grad_norm": 0.32390961050987244, "learning_rate": 1.997393731204922e-05, "loss": 0.34806056022644044, "step": 1840 }, { "epoch": 0.10894689574959882, "grad_norm": 0.24703992903232574, "learning_rate": 1.997331260787181e-05, "loss": 0.3153179168701172, "step": 1850 }, { "epoch": 0.10953579788878583, "grad_norm": 0.22500476241111755, "learning_rate": 1.9972680515242987e-05, "loss": 0.3393397331237793, "step": 1860 }, { "epoch": 0.11012470002797285, "grad_norm": 0.4042288362979889, "learning_rate": 1.9972041034631017e-05, "loss": 0.3514078617095947, "step": 1870 }, { "epoch": 0.11071360216715988, "grad_norm": 0.32466819882392883, "learning_rate": 1.9971394166509652e-05, "loss": 0.36292495727539065, "step": 1880 }, { "epoch": 0.11130250430634689, "grad_norm": 0.35624822974205017, "learning_rate": 1.9970739911358095e-05, "loss": 0.30636093616485593, "step": 1890 }, { "epoch": 0.11189140644553391, "grad_norm": 0.3025331497192383, "learning_rate": 1.9970078269661037e-05, "loss": 0.3539124965667725, "step": 1900 }, { "epoch": 0.11248030858472094, "grad_norm": 0.35953035950660706, "learning_rate": 1.9969409241908636e-05, "loss": 0.3196086406707764, "step": 1910 }, { "epoch": 0.11306921072390795, "grad_norm": 0.25831371545791626, "learning_rate": 1.9968732828596528e-05, "loss": 0.33699257373809816, "step": 1920 }, { "epoch": 0.11365811286309498, "grad_norm": 0.369141161441803, "learning_rate": 1.9968049030225808e-05, "loss": 0.3449979305267334, "step": 1930 }, { "epoch": 0.114247015002282, "grad_norm": 0.40488868951797485, "learning_rate": 1.996735784730305e-05, "loss": 0.3672555685043335, "step": 1940 }, { "epoch": 0.11483591714146901, "grad_norm": 0.19778484106063843, "learning_rate": 1.99666592803403e-05, "loss": 0.3580432891845703, "step": 1950 }, { "epoch": 0.11542481928065604, "grad_norm": 0.2903326451778412, "learning_rate": 1.9965953329855073e-05, "loss": 0.346855616569519, "step": 1960 }, { "epoch": 0.11601372141984306, "grad_norm": 0.24865485727787018, "learning_rate": 1.9965239996370348e-05, "loss": 0.3302056312561035, "step": 1970 }, { "epoch": 0.11660262355903007, "grad_norm": 0.2663765847682953, "learning_rate": 1.9964519280414584e-05, "loss": 0.357549262046814, "step": 1980 }, { "epoch": 0.1171915256982171, "grad_norm": 0.31425854563713074, "learning_rate": 1.9963791182521698e-05, "loss": 0.37770910263061525, "step": 1990 }, { "epoch": 0.11778042783740413, "grad_norm": 0.33793577551841736, "learning_rate": 1.9963055703231085e-05, "loss": 0.38348329067230225, "step": 2000 }, { "epoch": 0.0005889021391870206, "grad_norm": 0.3161006569862366, "learning_rate": 1.9962312843087606e-05, "loss": 0.36336259841918944, "step": 2010 }, { "epoch": 0.0011778042783740412, "grad_norm": 0.3583987057209015, "learning_rate": 1.996156260264158e-05, "loss": 0.37775363922119143, "step": 2020 }, { "epoch": 0.0017667064175610617, "grad_norm": 0.26518672704696655, "learning_rate": 1.996080498244881e-05, "loss": 0.3317399501800537, "step": 2030 }, { "epoch": 0.0023556085567480825, "grad_norm": 0.33767473697662354, "learning_rate": 1.9960039983070554e-05, "loss": 0.3751254081726074, "step": 2040 }, { "epoch": 0.002944510695935103, "grad_norm": 0.5078189969062805, "learning_rate": 1.9959267605073544e-05, "loss": 0.365812873840332, "step": 2050 }, { "epoch": 0.0035334128351221235, "grad_norm": 0.48954063653945923, "learning_rate": 1.995848784902997e-05, "loss": 0.29844534397125244, "step": 2060 }, { "epoch": 0.004122314974309144, "grad_norm": 0.4205349087715149, "learning_rate": 1.9957700715517493e-05, "loss": 0.35003373622894285, "step": 2070 }, { "epoch": 0.004711217113496165, "grad_norm": 0.2797721028327942, "learning_rate": 1.9956906205119245e-05, "loss": 0.3295466184616089, "step": 2080 }, { "epoch": 0.005300119252683185, "grad_norm": 0.3392319083213806, "learning_rate": 1.9956104318423813e-05, "loss": 0.3644935369491577, "step": 2090 }, { "epoch": 0.005889021391870206, "grad_norm": 0.26566851139068604, "learning_rate": 1.9955295056025248e-05, "loss": 0.3661762237548828, "step": 2100 }, { "epoch": 0.006477923531057227, "grad_norm": 0.38221612572669983, "learning_rate": 1.9954478418523075e-05, "loss": 0.3571451187133789, "step": 2110 }, { "epoch": 0.007066825670244247, "grad_norm": 0.24299713969230652, "learning_rate": 1.9953654406522278e-05, "loss": 0.36051723957061765, "step": 2120 }, { "epoch": 0.007655727809431268, "grad_norm": 0.4137168228626251, "learning_rate": 1.9952823020633302e-05, "loss": 0.3918939113616943, "step": 2130 }, { "epoch": 0.008244629948618288, "grad_norm": 0.2750120162963867, "learning_rate": 1.995198426147205e-05, "loss": 0.4137897491455078, "step": 2140 }, { "epoch": 0.00883353208780531, "grad_norm": 0.27459850907325745, "learning_rate": 1.9951138129659907e-05, "loss": 0.370199179649353, "step": 2150 }, { "epoch": 0.00942243422699233, "grad_norm": 0.39682823419570923, "learning_rate": 1.9950284625823694e-05, "loss": 0.39414496421813966, "step": 2160 }, { "epoch": 0.01001133636617935, "grad_norm": 0.3149157464504242, "learning_rate": 1.9949423750595707e-05, "loss": 0.35489416122436523, "step": 2170 }, { "epoch": 0.01060023850536637, "grad_norm": 0.2882961630821228, "learning_rate": 1.9948555504613706e-05, "loss": 0.3669857978820801, "step": 2180 }, { "epoch": 0.011189140644553392, "grad_norm": 0.2816716432571411, "learning_rate": 1.994767988852091e-05, "loss": 0.4075425624847412, "step": 2190 }, { "epoch": 0.011778042783740412, "grad_norm": 0.30024778842926025, "learning_rate": 1.9946796902965984e-05, "loss": 0.3382577657699585, "step": 2200 }, { "epoch": 0.012366944922927432, "grad_norm": 0.28759366273880005, "learning_rate": 1.9945906548603072e-05, "loss": 0.37319872379302976, "step": 2210 }, { "epoch": 0.012955847062114454, "grad_norm": 0.3235315680503845, "learning_rate": 1.9945008826091767e-05, "loss": 0.3805527687072754, "step": 2220 }, { "epoch": 0.013544749201301474, "grad_norm": 0.3207806646823883, "learning_rate": 1.9944103736097116e-05, "loss": 0.36231379508972167, "step": 2230 }, { "epoch": 0.014133651340488494, "grad_norm": 0.3226885497570038, "learning_rate": 1.9943191279289634e-05, "loss": 0.3652675151824951, "step": 2240 }, { "epoch": 0.014722553479675516, "grad_norm": 0.3766103982925415, "learning_rate": 1.9942271456345293e-05, "loss": 0.3255454063415527, "step": 2250 }, { "epoch": 0.015311455618862536, "grad_norm": 0.3062818944454193, "learning_rate": 1.9941344267945516e-05, "loss": 0.37049312591552735, "step": 2260 }, { "epoch": 0.015900357758049558, "grad_norm": 0.2686237394809723, "learning_rate": 1.994040971477718e-05, "loss": 0.34719486236572267, "step": 2270 }, { "epoch": 0.016489259897236576, "grad_norm": 0.3991428315639496, "learning_rate": 1.9939467797532627e-05, "loss": 0.33168759346008303, "step": 2280 }, { "epoch": 0.017078162036423598, "grad_norm": 0.5348249673843384, "learning_rate": 1.9938518516909647e-05, "loss": 0.3583676338195801, "step": 2290 }, { "epoch": 0.01766706417561062, "grad_norm": 0.32452839612960815, "learning_rate": 1.9937561873611496e-05, "loss": 0.3520155668258667, "step": 2300 }, { "epoch": 0.018255966314797638, "grad_norm": 0.26055750250816345, "learning_rate": 1.9936597868346864e-05, "loss": 0.3709836721420288, "step": 2310 }, { "epoch": 0.01884486845398466, "grad_norm": 0.3782638609409332, "learning_rate": 1.993562650182992e-05, "loss": 0.3011315107345581, "step": 2320 }, { "epoch": 0.01943377059317168, "grad_norm": 0.30157801508903503, "learning_rate": 1.993464777478027e-05, "loss": 0.31611127853393556, "step": 2330 }, { "epoch": 0.0200226727323587, "grad_norm": 0.22805851697921753, "learning_rate": 1.993366168792297e-05, "loss": 0.33050880432128904, "step": 2340 }, { "epoch": 0.02061157487154572, "grad_norm": 0.36990898847579956, "learning_rate": 1.9932668241988543e-05, "loss": 0.36366562843322753, "step": 2350 }, { "epoch": 0.02120047701073274, "grad_norm": 0.21041983366012573, "learning_rate": 1.9931667437712953e-05, "loss": 0.33885080814361573, "step": 2360 }, { "epoch": 0.02178937914991976, "grad_norm": 0.3204512298107147, "learning_rate": 1.9930659275837618e-05, "loss": 0.32399630546569824, "step": 2370 }, { "epoch": 0.022378281289106784, "grad_norm": 0.2835949659347534, "learning_rate": 1.9929643757109406e-05, "loss": 0.3722275972366333, "step": 2380 }, { "epoch": 0.022967183428293802, "grad_norm": 0.43311357498168945, "learning_rate": 1.9928620882280633e-05, "loss": 0.31155877113342284, "step": 2390 }, { "epoch": 0.023556085567480824, "grad_norm": 0.39346984028816223, "learning_rate": 1.9927590652109075e-05, "loss": 0.335731053352356, "step": 2400 }, { "epoch": 0.024144987706667845, "grad_norm": 0.3715560734272003, "learning_rate": 1.9926553067357945e-05, "loss": 0.3198944091796875, "step": 2410 }, { "epoch": 0.024733889845854864, "grad_norm": 0.5043146014213562, "learning_rate": 1.992550812879591e-05, "loss": 0.3521722316741943, "step": 2420 }, { "epoch": 0.025322791985041886, "grad_norm": 0.2830747663974762, "learning_rate": 1.9924455837197084e-05, "loss": 0.31241612434387206, "step": 2430 }, { "epoch": 0.025911694124228907, "grad_norm": 0.21121449768543243, "learning_rate": 1.9923396193341022e-05, "loss": 0.3274442434310913, "step": 2440 }, { "epoch": 0.026500596263415926, "grad_norm": 0.334094762802124, "learning_rate": 1.992232919801274e-05, "loss": 0.37161076068878174, "step": 2450 }, { "epoch": 0.027089498402602948, "grad_norm": 0.22716519236564636, "learning_rate": 1.992125485200269e-05, "loss": 0.3406389236450195, "step": 2460 }, { "epoch": 0.02767840054178997, "grad_norm": 0.5283288955688477, "learning_rate": 1.9920173156106765e-05, "loss": 0.3927865982055664, "step": 2470 }, { "epoch": 0.028267302680976988, "grad_norm": 0.3524184823036194, "learning_rate": 1.9919084111126318e-05, "loss": 0.3335184812545776, "step": 2480 }, { "epoch": 0.02885620482016401, "grad_norm": 0.3663606643676758, "learning_rate": 1.991798771786813e-05, "loss": 0.3778546333312988, "step": 2490 }, { "epoch": 0.02944510695935103, "grad_norm": 0.4033188819885254, "learning_rate": 1.991688397714444e-05, "loss": 0.3827038288116455, "step": 2500 }, { "epoch": 0.03003400909853805, "grad_norm": 0.2844197750091553, "learning_rate": 1.991577288977292e-05, "loss": 0.32161526679992675, "step": 2510 }, { "epoch": 0.03062291123772507, "grad_norm": 0.5696726441383362, "learning_rate": 1.991465445657669e-05, "loss": 0.36856200695037844, "step": 2520 }, { "epoch": 0.031211813376912093, "grad_norm": 0.3536056578159332, "learning_rate": 1.9913528678384308e-05, "loss": 0.3414334297180176, "step": 2530 }, { "epoch": 0.031800715516099115, "grad_norm": 0.367120623588562, "learning_rate": 1.9912395556029776e-05, "loss": 0.30569746494293215, "step": 2540 }, { "epoch": 0.03238961765528613, "grad_norm": 0.2956843078136444, "learning_rate": 1.991125509035254e-05, "loss": 0.33148267269134524, "step": 2550 }, { "epoch": 0.03297851979447315, "grad_norm": 0.30674588680267334, "learning_rate": 1.9910107282197477e-05, "loss": 0.3310613393783569, "step": 2560 }, { "epoch": 0.03356742193366018, "grad_norm": 0.2839662730693817, "learning_rate": 1.9908952132414913e-05, "loss": 0.3120028257369995, "step": 2570 }, { "epoch": 0.034156324072847195, "grad_norm": 0.24074120819568634, "learning_rate": 1.9907789641860604e-05, "loss": 0.33746721744537356, "step": 2580 }, { "epoch": 0.034745226212034214, "grad_norm": 0.2727612853050232, "learning_rate": 1.9906619811395753e-05, "loss": 0.35102965831756594, "step": 2590 }, { "epoch": 0.03533412835122124, "grad_norm": 0.4260251820087433, "learning_rate": 1.9905442641886994e-05, "loss": 0.3665417194366455, "step": 2600 }, { "epoch": 0.03592303049040826, "grad_norm": 0.31084173917770386, "learning_rate": 1.99042581342064e-05, "loss": 0.3595787286758423, "step": 2610 }, { "epoch": 0.036511932629595276, "grad_norm": 0.28408244252204895, "learning_rate": 1.990306628923148e-05, "loss": 0.3687930107116699, "step": 2620 }, { "epoch": 0.0371008347687823, "grad_norm": 0.3619328439235687, "learning_rate": 1.9901867107845187e-05, "loss": 0.34442615509033203, "step": 2630 }, { "epoch": 0.03768973690796932, "grad_norm": 0.35435163974761963, "learning_rate": 1.990066059093589e-05, "loss": 0.3432980298995972, "step": 2640 }, { "epoch": 0.03827863904715634, "grad_norm": 0.46718883514404297, "learning_rate": 1.989944673939741e-05, "loss": 0.40845460891723634, "step": 2650 }, { "epoch": 0.03886754118634336, "grad_norm": 0.4547567367553711, "learning_rate": 1.9898225554128995e-05, "loss": 0.33099899291992185, "step": 2660 }, { "epoch": 0.03945644332553038, "grad_norm": 0.39546436071395874, "learning_rate": 1.9896997036035324e-05, "loss": 0.3692225694656372, "step": 2670 }, { "epoch": 0.0400453454647174, "grad_norm": 0.3667880892753601, "learning_rate": 1.989576118602651e-05, "loss": 0.3930224418640137, "step": 2680 }, { "epoch": 0.04063424760390442, "grad_norm": 0.37184926867485046, "learning_rate": 1.98945180050181e-05, "loss": 0.3347638130187988, "step": 2690 }, { "epoch": 0.04122314974309144, "grad_norm": 0.3768233358860016, "learning_rate": 1.9893267493931072e-05, "loss": 0.33841514587402344, "step": 2700 }, { "epoch": 0.04181205188227846, "grad_norm": 0.3401988446712494, "learning_rate": 1.989200965369183e-05, "loss": 0.3239858150482178, "step": 2710 }, { "epoch": 0.04240095402146548, "grad_norm": 0.37386664748191833, "learning_rate": 1.989074448523221e-05, "loss": 0.3558147192001343, "step": 2720 }, { "epoch": 0.042989856160652505, "grad_norm": 0.4841950237751007, "learning_rate": 1.9889471989489473e-05, "loss": 0.32613201141357423, "step": 2730 }, { "epoch": 0.04357875829983952, "grad_norm": 0.3056938648223877, "learning_rate": 1.9888192167406316e-05, "loss": 0.35435411930084226, "step": 2740 }, { "epoch": 0.04416766043902654, "grad_norm": 0.3050881028175354, "learning_rate": 1.9886905019930863e-05, "loss": 0.3633758544921875, "step": 2750 }, { "epoch": 0.04475656257821357, "grad_norm": 0.4220479726791382, "learning_rate": 1.9885610548016655e-05, "loss": 0.36458942890167234, "step": 2760 }, { "epoch": 0.045345464717400585, "grad_norm": 0.3379204273223877, "learning_rate": 1.9884308752622672e-05, "loss": 0.34271576404571535, "step": 2770 }, { "epoch": 0.045934366856587604, "grad_norm": 0.40134039521217346, "learning_rate": 1.9882999634713307e-05, "loss": 0.31763601303100586, "step": 2780 }, { "epoch": 0.04652326899577463, "grad_norm": 0.4859100580215454, "learning_rate": 1.9881683195258387e-05, "loss": 0.3775187969207764, "step": 2790 }, { "epoch": 0.04711217113496165, "grad_norm": 0.34237682819366455, "learning_rate": 1.988035943523316e-05, "loss": 0.34224438667297363, "step": 2800 }, { "epoch": 0.047701073274148666, "grad_norm": 0.387445867061615, "learning_rate": 1.9879028355618293e-05, "loss": 0.35814676284790037, "step": 2810 }, { "epoch": 0.04828997541333569, "grad_norm": 0.4260205626487732, "learning_rate": 1.987768995739988e-05, "loss": 0.3971248626708984, "step": 2820 }, { "epoch": 0.04887887755252271, "grad_norm": 0.4546390771865845, "learning_rate": 1.9876344241569445e-05, "loss": 0.33820457458496095, "step": 2830 }, { "epoch": 0.04946777969170973, "grad_norm": 0.3829171061515808, "learning_rate": 1.987499120912391e-05, "loss": 0.3322794198989868, "step": 2840 }, { "epoch": 0.05005668183089675, "grad_norm": 0.34023571014404297, "learning_rate": 1.987363086106564e-05, "loss": 0.3759793281555176, "step": 2850 }, { "epoch": 0.05064558397008377, "grad_norm": 0.39458194375038147, "learning_rate": 1.9872263198402413e-05, "loss": 0.3718777418136597, "step": 2860 }, { "epoch": 0.05123448610927079, "grad_norm": 0.4321185350418091, "learning_rate": 1.9870888222147418e-05, "loss": 0.3566880702972412, "step": 2870 }, { "epoch": 0.051823388248457815, "grad_norm": 0.33078229427337646, "learning_rate": 1.986950593331927e-05, "loss": 0.30704669952392577, "step": 2880 }, { "epoch": 0.05241229038764483, "grad_norm": 0.30275803804397583, "learning_rate": 1.9868116332941998e-05, "loss": 0.3206380605697632, "step": 2890 }, { "epoch": 0.05300119252683185, "grad_norm": 0.3664777874946594, "learning_rate": 1.9866719422045055e-05, "loss": 0.31364681720733645, "step": 2900 }, { "epoch": 0.05359009466601888, "grad_norm": 0.36234548687934875, "learning_rate": 1.9865315201663297e-05, "loss": 0.31425225734710693, "step": 2910 }, { "epoch": 0.054178996805205895, "grad_norm": 0.3521561920642853, "learning_rate": 1.9863903672837003e-05, "loss": 0.32815525531768797, "step": 2920 }, { "epoch": 0.05476789894439291, "grad_norm": 0.6721714735031128, "learning_rate": 1.9862484836611865e-05, "loss": 0.33475346565246583, "step": 2930 }, { "epoch": 0.05535680108357994, "grad_norm": 0.40165120363235474, "learning_rate": 1.986105869403899e-05, "loss": 0.38275372982025146, "step": 2940 }, { "epoch": 0.05594570322276696, "grad_norm": 0.43605470657348633, "learning_rate": 1.9859625246174894e-05, "loss": 0.35219426155090333, "step": 2950 }, { "epoch": 0.056534605361953975, "grad_norm": 0.5244734883308411, "learning_rate": 1.985818449408151e-05, "loss": 0.34538085460662843, "step": 2960 }, { "epoch": 0.057123507501141, "grad_norm": 0.4803990125656128, "learning_rate": 1.9856736438826175e-05, "loss": 0.33985898494720457, "step": 2970 }, { "epoch": 0.05771240964032802, "grad_norm": 0.30203282833099365, "learning_rate": 1.9855281081481643e-05, "loss": 0.3386751651763916, "step": 2980 }, { "epoch": 0.05830131177951504, "grad_norm": 0.42464005947113037, "learning_rate": 1.9853818423126078e-05, "loss": 0.36141960620880126, "step": 2990 }, { "epoch": 0.05889021391870206, "grad_norm": 0.3404679298400879, "learning_rate": 1.9852348464843043e-05, "loss": 0.31191933155059814, "step": 3000 }, { "epoch": 0.05947911605788908, "grad_norm": 0.45890533924102783, "learning_rate": 1.985087120772152e-05, "loss": 0.3424859046936035, "step": 3010 }, { "epoch": 0.0600680181970761, "grad_norm": 0.3603091835975647, "learning_rate": 1.9849386652855894e-05, "loss": 0.36296787261962893, "step": 3020 }, { "epoch": 0.060656920336263125, "grad_norm": 0.6057437658309937, "learning_rate": 1.9847894801345956e-05, "loss": 0.3406908750534058, "step": 3030 }, { "epoch": 0.06124582247545014, "grad_norm": 0.31126701831817627, "learning_rate": 1.98463956542969e-05, "loss": 0.3193426370620728, "step": 3040 }, { "epoch": 0.06183472461463716, "grad_norm": 0.3801962435245514, "learning_rate": 1.984488921281933e-05, "loss": 0.32549176216125486, "step": 3050 }, { "epoch": 0.062423626753824187, "grad_norm": 0.4758055508136749, "learning_rate": 1.984337547802925e-05, "loss": 0.32301671504974366, "step": 3060 }, { "epoch": 0.0630125288930112, "grad_norm": 0.3774026334285736, "learning_rate": 1.984185445104807e-05, "loss": 0.3203294038772583, "step": 3070 }, { "epoch": 0.06360143103219823, "grad_norm": 0.362237811088562, "learning_rate": 1.9840326133002602e-05, "loss": 0.34834287166595457, "step": 3080 }, { "epoch": 0.06419033317138524, "grad_norm": 0.5041883587837219, "learning_rate": 1.983879052502505e-05, "loss": 0.33974030017852785, "step": 3090 }, { "epoch": 0.06477923531057227, "grad_norm": 0.29693466424942017, "learning_rate": 1.9837247628253037e-05, "loss": 0.3911701202392578, "step": 3100 }, { "epoch": 0.06536813744975929, "grad_norm": 0.352876216173172, "learning_rate": 1.983569744382957e-05, "loss": 0.32885549068450926, "step": 3110 }, { "epoch": 0.0659570395889463, "grad_norm": 0.4597175717353821, "learning_rate": 1.9834139972903057e-05, "loss": 0.33990418910980225, "step": 3120 }, { "epoch": 0.06654594172813333, "grad_norm": 0.5461044311523438, "learning_rate": 1.9832575216627313e-05, "loss": 0.322416090965271, "step": 3130 }, { "epoch": 0.06713484386732035, "grad_norm": 0.41199618577957153, "learning_rate": 1.983100317616154e-05, "loss": 0.3255443811416626, "step": 3140 }, { "epoch": 0.06772374600650737, "grad_norm": 0.4572743773460388, "learning_rate": 1.9829423852670344e-05, "loss": 0.35251660346984864, "step": 3150 }, { "epoch": 0.06831264814569439, "grad_norm": 0.3923286199569702, "learning_rate": 1.9827837247323722e-05, "loss": 0.35719263553619385, "step": 3160 }, { "epoch": 0.06890155028488142, "grad_norm": 0.38015827536582947, "learning_rate": 1.982624336129706e-05, "loss": 0.3620716094970703, "step": 3170 }, { "epoch": 0.06949045242406843, "grad_norm": 0.3968326151371002, "learning_rate": 1.9824642195771152e-05, "loss": 0.36899256706237793, "step": 3180 }, { "epoch": 0.07007935456325545, "grad_norm": 0.429862380027771, "learning_rate": 1.9823033751932174e-05, "loss": 0.3274944305419922, "step": 3190 }, { "epoch": 0.07066825670244248, "grad_norm": 0.3822976350784302, "learning_rate": 1.98214180309717e-05, "loss": 0.3226094961166382, "step": 3200 }, { "epoch": 0.07125715884162949, "grad_norm": 0.3318658471107483, "learning_rate": 1.981979503408669e-05, "loss": 0.3953235387802124, "step": 3210 }, { "epoch": 0.07184606098081651, "grad_norm": 0.3661399185657501, "learning_rate": 1.981816476247949e-05, "loss": 0.3763868808746338, "step": 3220 }, { "epoch": 0.07243496312000354, "grad_norm": 0.5271238684654236, "learning_rate": 1.9816527217357853e-05, "loss": 0.3509507179260254, "step": 3230 }, { "epoch": 0.07302386525919055, "grad_norm": 0.40106454491615295, "learning_rate": 1.98148823999349e-05, "loss": 0.32432315349578855, "step": 3240 }, { "epoch": 0.07361276739837758, "grad_norm": 0.320115327835083, "learning_rate": 1.981323031142915e-05, "loss": 0.3159781932830811, "step": 3250 }, { "epoch": 0.0742016695375646, "grad_norm": 0.44302898645401, "learning_rate": 1.981157095306451e-05, "loss": 0.3379368782043457, "step": 3260 }, { "epoch": 0.07479057167675161, "grad_norm": 0.445977121591568, "learning_rate": 1.9809904326070265e-05, "loss": 0.3663978576660156, "step": 3270 }, { "epoch": 0.07537947381593864, "grad_norm": 0.4700806736946106, "learning_rate": 1.9808230431681092e-05, "loss": 0.30336794853210447, "step": 3280 }, { "epoch": 0.07596837595512566, "grad_norm": 0.3289479911327362, "learning_rate": 1.980654927113705e-05, "loss": 0.32774853706359863, "step": 3290 }, { "epoch": 0.07655727809431268, "grad_norm": 0.375135213136673, "learning_rate": 1.9804860845683578e-05, "loss": 0.33609509468078613, "step": 3300 }, { "epoch": 0.0771461802334997, "grad_norm": 0.4059208631515503, "learning_rate": 1.98031651565715e-05, "loss": 0.3278010606765747, "step": 3310 }, { "epoch": 0.07773508237268673, "grad_norm": 0.4442974627017975, "learning_rate": 1.980146220505702e-05, "loss": 0.36862385272979736, "step": 3320 }, { "epoch": 0.07832398451187374, "grad_norm": 0.4624920189380646, "learning_rate": 1.979975199240172e-05, "loss": 0.31739590167999265, "step": 3330 }, { "epoch": 0.07891288665106076, "grad_norm": 0.4360860288143158, "learning_rate": 1.9798034519872572e-05, "loss": 0.2852126359939575, "step": 3340 }, { "epoch": 0.07950178879024777, "grad_norm": 0.3785429298877716, "learning_rate": 1.9796309788741905e-05, "loss": 0.2986567974090576, "step": 3350 }, { "epoch": 0.0800906909294348, "grad_norm": 0.4497954249382019, "learning_rate": 1.9794577800287448e-05, "loss": 0.3224040985107422, "step": 3360 }, { "epoch": 0.08067959306862182, "grad_norm": 0.39757978916168213, "learning_rate": 1.979283855579229e-05, "loss": 0.3367490291595459, "step": 3370 }, { "epoch": 0.08126849520780884, "grad_norm": 0.415041446685791, "learning_rate": 1.979109205654491e-05, "loss": 0.3380012273788452, "step": 3380 }, { "epoch": 0.08185739734699586, "grad_norm": 0.33320409059524536, "learning_rate": 1.978933830383915e-05, "loss": 0.32474546432495116, "step": 3390 }, { "epoch": 0.08244629948618289, "grad_norm": 0.3458709418773651, "learning_rate": 1.978757729897422e-05, "loss": 0.33915324211120607, "step": 3400 }, { "epoch": 0.0830352016253699, "grad_norm": 0.5591266751289368, "learning_rate": 1.978580904325472e-05, "loss": 0.3417789459228516, "step": 3410 }, { "epoch": 0.08362410376455692, "grad_norm": 0.39545151591300964, "learning_rate": 1.9784033537990618e-05, "loss": 0.31967964172363283, "step": 3420 }, { "epoch": 0.08421300590374395, "grad_norm": 0.4815097749233246, "learning_rate": 1.978225078449724e-05, "loss": 0.31837735176086424, "step": 3430 }, { "epoch": 0.08480190804293096, "grad_norm": 0.5563396215438843, "learning_rate": 1.9780460784095288e-05, "loss": 0.34514014720916747, "step": 3440 }, { "epoch": 0.08539081018211798, "grad_norm": 0.42059171199798584, "learning_rate": 1.977866353811084e-05, "loss": 0.3347254991531372, "step": 3450 }, { "epoch": 0.08597971232130501, "grad_norm": 0.4785255193710327, "learning_rate": 1.977685904787533e-05, "loss": 0.3490870714187622, "step": 3460 }, { "epoch": 0.08656861446049202, "grad_norm": 0.4209532141685486, "learning_rate": 1.977504731472557e-05, "loss": 0.39239494800567626, "step": 3470 }, { "epoch": 0.08715751659967905, "grad_norm": 0.41144394874572754, "learning_rate": 1.977322834000373e-05, "loss": 0.3419785022735596, "step": 3480 }, { "epoch": 0.08774641873886607, "grad_norm": 0.37077438831329346, "learning_rate": 1.9771402125057344e-05, "loss": 0.3031206846237183, "step": 3490 }, { "epoch": 0.08833532087805308, "grad_norm": 0.3205106258392334, "learning_rate": 1.9769568671239313e-05, "loss": 0.34897873401641843, "step": 3500 }, { "epoch": 0.08892422301724011, "grad_norm": 0.41304776072502136, "learning_rate": 1.9767727979907908e-05, "loss": 0.3083486080169678, "step": 3510 }, { "epoch": 0.08951312515642713, "grad_norm": 0.6389936804771423, "learning_rate": 1.9765880052426744e-05, "loss": 0.3805606603622437, "step": 3520 }, { "epoch": 0.09010202729561415, "grad_norm": 0.6531752347946167, "learning_rate": 1.9764024890164815e-05, "loss": 0.35923013687133787, "step": 3530 }, { "epoch": 0.09069092943480117, "grad_norm": 0.5180909633636475, "learning_rate": 1.9762162494496457e-05, "loss": 0.36575396060943605, "step": 3540 }, { "epoch": 0.0912798315739882, "grad_norm": 0.33680054545402527, "learning_rate": 1.9760292866801384e-05, "loss": 0.34167966842651365, "step": 3550 }, { "epoch": 0.09186873371317521, "grad_norm": 0.4303724765777588, "learning_rate": 1.9758416008464652e-05, "loss": 0.3527165412902832, "step": 3560 }, { "epoch": 0.09245763585236223, "grad_norm": 0.47436773777008057, "learning_rate": 1.9756531920876678e-05, "loss": 0.3608031988143921, "step": 3570 }, { "epoch": 0.09304653799154926, "grad_norm": 0.4718152582645416, "learning_rate": 1.975464060543324e-05, "loss": 0.3245852947235107, "step": 3580 }, { "epoch": 0.09363544013073627, "grad_norm": 0.35804763436317444, "learning_rate": 1.9752742063535467e-05, "loss": 0.30201632976531984, "step": 3590 }, { "epoch": 0.0942243422699233, "grad_norm": 0.4880467653274536, "learning_rate": 1.9750836296589836e-05, "loss": 0.3489097833633423, "step": 3600 }, { "epoch": 0.09481324440911032, "grad_norm": 0.430411159992218, "learning_rate": 1.9748923306008185e-05, "loss": 0.33640055656433104, "step": 3610 }, { "epoch": 0.09540214654829733, "grad_norm": 0.4741526246070862, "learning_rate": 1.97470030932077e-05, "loss": 0.3038966655731201, "step": 3620 }, { "epoch": 0.09599104868748436, "grad_norm": 0.4422605633735657, "learning_rate": 1.9745075659610916e-05, "loss": 0.3146333932876587, "step": 3630 }, { "epoch": 0.09657995082667138, "grad_norm": 0.4382513165473938, "learning_rate": 1.9743141006645715e-05, "loss": 0.3277940273284912, "step": 3640 }, { "epoch": 0.0971688529658584, "grad_norm": 0.49546757340431213, "learning_rate": 1.9741199135745338e-05, "loss": 0.33587558269500734, "step": 3650 }, { "epoch": 0.09775775510504542, "grad_norm": 0.46132901310920715, "learning_rate": 1.9739250048348365e-05, "loss": 0.3550309419631958, "step": 3660 }, { "epoch": 0.09834665724423244, "grad_norm": 0.42588311433792114, "learning_rate": 1.9737293745898724e-05, "loss": 0.34794442653656005, "step": 3670 }, { "epoch": 0.09893555938341946, "grad_norm": 0.5894307494163513, "learning_rate": 1.9735330229845676e-05, "loss": 0.33161396980285646, "step": 3680 }, { "epoch": 0.09952446152260648, "grad_norm": 0.42777156829833984, "learning_rate": 1.9733359501643853e-05, "loss": 0.30537378787994385, "step": 3690 }, { "epoch": 0.1001133636617935, "grad_norm": 0.46326184272766113, "learning_rate": 1.9731381562753203e-05, "loss": 0.34515957832336425, "step": 3700 }, { "epoch": 0.10070226580098052, "grad_norm": 0.42314091324806213, "learning_rate": 1.9729396414639033e-05, "loss": 0.3359150648117065, "step": 3710 }, { "epoch": 0.10129116794016754, "grad_norm": 0.4275287091732025, "learning_rate": 1.9727404058771982e-05, "loss": 0.29092471599578856, "step": 3720 }, { "epoch": 0.10188007007935457, "grad_norm": 0.3545110821723938, "learning_rate": 1.972540449662803e-05, "loss": 0.2875332832336426, "step": 3730 }, { "epoch": 0.10246897221854158, "grad_norm": 0.6371763944625854, "learning_rate": 1.9723397729688503e-05, "loss": 0.35977559089660643, "step": 3740 }, { "epoch": 0.1030578743577286, "grad_norm": 0.534267783164978, "learning_rate": 1.9721383759440055e-05, "loss": 0.3457848787307739, "step": 3750 }, { "epoch": 0.10364677649691563, "grad_norm": 0.31745147705078125, "learning_rate": 1.971936258737468e-05, "loss": 0.3847079038619995, "step": 3760 }, { "epoch": 0.10423567863610264, "grad_norm": 0.5258601903915405, "learning_rate": 1.9717334214989706e-05, "loss": 0.3541384220123291, "step": 3770 }, { "epoch": 0.10482458077528967, "grad_norm": 0.41624611616134644, "learning_rate": 1.9715298643787793e-05, "loss": 0.3153546333312988, "step": 3780 }, { "epoch": 0.10541348291447669, "grad_norm": 0.4447533190250397, "learning_rate": 1.9713255875276946e-05, "loss": 0.3028708457946777, "step": 3790 }, { "epoch": 0.1060023850536637, "grad_norm": 0.3623751401901245, "learning_rate": 1.9711205910970486e-05, "loss": 0.2964569091796875, "step": 3800 }, { "epoch": 0.10659128719285073, "grad_norm": 0.41511157155036926, "learning_rate": 1.970914875238708e-05, "loss": 0.36160356998443605, "step": 3810 }, { "epoch": 0.10718018933203775, "grad_norm": 0.3735142946243286, "learning_rate": 1.9707084401050706e-05, "loss": 0.34250948429107664, "step": 3820 }, { "epoch": 0.10776909147122476, "grad_norm": 0.5246334075927734, "learning_rate": 1.9705012858490692e-05, "loss": 0.34458045959472655, "step": 3830 }, { "epoch": 0.10835799361041179, "grad_norm": 0.43436336517333984, "learning_rate": 1.9702934126241672e-05, "loss": 0.32449707984924314, "step": 3840 }, { "epoch": 0.10894689574959882, "grad_norm": 0.3659230172634125, "learning_rate": 1.9700848205843627e-05, "loss": 0.2934209585189819, "step": 3850 }, { "epoch": 0.10953579788878583, "grad_norm": 0.3394876718521118, "learning_rate": 1.9698755098841845e-05, "loss": 0.3164581537246704, "step": 3860 }, { "epoch": 0.11012470002797285, "grad_norm": 0.5550746917724609, "learning_rate": 1.969665480678695e-05, "loss": 0.3302888870239258, "step": 3870 }, { "epoch": 0.11071360216715988, "grad_norm": 0.4233824610710144, "learning_rate": 1.969454733123489e-05, "loss": 0.339612078666687, "step": 3880 }, { "epoch": 0.11130250430634689, "grad_norm": 0.3699467182159424, "learning_rate": 1.9692432673746922e-05, "loss": 0.2867459535598755, "step": 3890 }, { "epoch": 0.11189140644553391, "grad_norm": 0.38969963788986206, "learning_rate": 1.9690310835889633e-05, "loss": 0.33392531871795655, "step": 3900 }, { "epoch": 0.11248030858472094, "grad_norm": 0.45890840888023376, "learning_rate": 1.968818181923493e-05, "loss": 0.29840831756591796, "step": 3910 }, { "epoch": 0.11306921072390795, "grad_norm": 0.36322155594825745, "learning_rate": 1.9686045625360035e-05, "loss": 0.316160249710083, "step": 3920 }, { "epoch": 0.11365811286309498, "grad_norm": 0.5006183385848999, "learning_rate": 1.9683902255847483e-05, "loss": 0.3214092493057251, "step": 3930 }, { "epoch": 0.114247015002282, "grad_norm": 0.5320672392845154, "learning_rate": 1.9681751712285135e-05, "loss": 0.3435194969177246, "step": 3940 }, { "epoch": 0.11483591714146901, "grad_norm": 0.2928580343723297, "learning_rate": 1.967959399626616e-05, "loss": 0.33108718395233155, "step": 3950 }, { "epoch": 0.11542481928065604, "grad_norm": 0.3775174021720886, "learning_rate": 1.9677429109389045e-05, "loss": 0.32305495738983153, "step": 3960 }, { "epoch": 0.11601372141984306, "grad_norm": 0.3407578766345978, "learning_rate": 1.967525705325758e-05, "loss": 0.3111711502075195, "step": 3970 }, { "epoch": 0.11660262355903007, "grad_norm": 0.3857874870300293, "learning_rate": 1.9673077829480876e-05, "loss": 0.3332832813262939, "step": 3980 }, { "epoch": 0.1171915256982171, "grad_norm": 0.46411848068237305, "learning_rate": 1.9670891439673353e-05, "loss": 0.36013948917388916, "step": 3990 }, { "epoch": 0.11778042783740413, "grad_norm": 0.4782850444316864, "learning_rate": 1.966869788545473e-05, "loss": 0.3700621843338013, "step": 4000 }, { "epoch": 0.11836932997659114, "grad_norm": 0.3819936513900757, "learning_rate": 1.966649716845005e-05, "loss": 0.322025728225708, "step": 4010 }, { "epoch": 0.11895823211577816, "grad_norm": 0.3533529043197632, "learning_rate": 1.9664289290289644e-05, "loss": 0.3418720245361328, "step": 4020 }, { "epoch": 0.11954713425496519, "grad_norm": 0.5174623131752014, "learning_rate": 1.9662074252609164e-05, "loss": 0.33044984340667727, "step": 4030 }, { "epoch": 0.1201360363941522, "grad_norm": 0.704045832157135, "learning_rate": 1.9659852057049553e-05, "loss": 0.36021876335144043, "step": 4040 }, { "epoch": 0.12072493853333922, "grad_norm": 0.49357664585113525, "learning_rate": 1.965762270525707e-05, "loss": 0.3556889295578003, "step": 4050 }, { "epoch": 0.12131384067252625, "grad_norm": 0.45066148042678833, "learning_rate": 1.965538619888326e-05, "loss": 0.3576862573623657, "step": 4060 }, { "epoch": 0.12190274281171326, "grad_norm": 0.4556524455547333, "learning_rate": 1.9653142539584983e-05, "loss": 0.3831826448440552, "step": 4070 }, { "epoch": 0.12249164495090029, "grad_norm": 0.4837978184223175, "learning_rate": 1.9650891729024392e-05, "loss": 0.35622103214263917, "step": 4080 }, { "epoch": 0.12308054709008731, "grad_norm": 0.42440637946128845, "learning_rate": 1.9648633768868934e-05, "loss": 0.36050245761871336, "step": 4090 }, { "epoch": 0.12366944922927432, "grad_norm": 0.2993285655975342, "learning_rate": 1.9646368660791352e-05, "loss": 0.38050825595855714, "step": 4100 }, { "epoch": 0.12425835136846135, "grad_norm": 0.42855608463287354, "learning_rate": 1.9644096406469698e-05, "loss": 0.3538414716720581, "step": 4110 }, { "epoch": 0.12484725350764837, "grad_norm": 0.3527754843235016, "learning_rate": 1.9641817007587303e-05, "loss": 0.3601287841796875, "step": 4120 }, { "epoch": 0.12543615564683538, "grad_norm": 0.4725300073623657, "learning_rate": 1.9639530465832796e-05, "loss": 0.3628203868865967, "step": 4130 }, { "epoch": 0.1260250577860224, "grad_norm": 0.5760388970375061, "learning_rate": 1.96372367829001e-05, "loss": 0.38325626850128175, "step": 4140 }, { "epoch": 0.12661395992520943, "grad_norm": 0.366617888212204, "learning_rate": 1.9634935960488423e-05, "loss": 0.3365812301635742, "step": 4150 }, { "epoch": 0.12720286206439646, "grad_norm": 0.4136699140071869, "learning_rate": 1.963262800030227e-05, "loss": 0.3742504596710205, "step": 4160 }, { "epoch": 0.12779176420358346, "grad_norm": 0.586840033531189, "learning_rate": 1.9630312904051428e-05, "loss": 0.4371016025543213, "step": 4170 }, { "epoch": 0.12838066634277048, "grad_norm": 0.46486392617225647, "learning_rate": 1.9627990673450968e-05, "loss": 0.3324949502944946, "step": 4180 }, { "epoch": 0.1289695684819575, "grad_norm": 0.45803558826446533, "learning_rate": 1.962566131022125e-05, "loss": 0.3715265989303589, "step": 4190 }, { "epoch": 0.12955847062114453, "grad_norm": 0.38504135608673096, "learning_rate": 1.9623324816087925e-05, "loss": 0.32581610679626466, "step": 4200 }, { "epoch": 0.13014737276033156, "grad_norm": 0.3219816982746124, "learning_rate": 1.9620981192781912e-05, "loss": 0.36291491985321045, "step": 4210 }, { "epoch": 0.13073627489951858, "grad_norm": 0.40102025866508484, "learning_rate": 1.9618630442039425e-05, "loss": 0.36337971687316895, "step": 4220 }, { "epoch": 0.13132517703870558, "grad_norm": 0.3246297836303711, "learning_rate": 1.9616272565601945e-05, "loss": 0.3574495553970337, "step": 4230 }, { "epoch": 0.1319140791778926, "grad_norm": 0.39966171979904175, "learning_rate": 1.9613907565216247e-05, "loss": 0.37541120052337645, "step": 4240 }, { "epoch": 0.13250298131707963, "grad_norm": 0.4184902310371399, "learning_rate": 1.9611535442634376e-05, "loss": 0.3572368860244751, "step": 4250 }, { "epoch": 0.13309188345626666, "grad_norm": 0.3778114318847656, "learning_rate": 1.9609156199613646e-05, "loss": 0.38230893611907957, "step": 4260 }, { "epoch": 0.13368078559545368, "grad_norm": 0.45292168855667114, "learning_rate": 1.9606769837916655e-05, "loss": 0.30334997177124023, "step": 4270 }, { "epoch": 0.1342696877346407, "grad_norm": 0.29183900356292725, "learning_rate": 1.9604376359311277e-05, "loss": 0.3188220262527466, "step": 4280 }, { "epoch": 0.1348585898738277, "grad_norm": 0.3132287859916687, "learning_rate": 1.9601975765570653e-05, "loss": 0.3439900875091553, "step": 4290 }, { "epoch": 0.13544749201301473, "grad_norm": 0.37252914905548096, "learning_rate": 1.9599568058473192e-05, "loss": 0.3484548568725586, "step": 4300 }, { "epoch": 0.13603639415220176, "grad_norm": 0.3305906653404236, "learning_rate": 1.959715323980258e-05, "loss": 0.33546617031097414, "step": 4310 }, { "epoch": 0.13662529629138878, "grad_norm": 0.4443693459033966, "learning_rate": 1.959473131134777e-05, "loss": 0.3481668710708618, "step": 4320 }, { "epoch": 0.1372141984305758, "grad_norm": 0.3441902995109558, "learning_rate": 1.9592302274902975e-05, "loss": 0.3362309217453003, "step": 4330 }, { "epoch": 0.13780310056976283, "grad_norm": 0.3859376013278961, "learning_rate": 1.9589866132267685e-05, "loss": 0.33944904804229736, "step": 4340 }, { "epoch": 0.13839200270894983, "grad_norm": 0.636002242565155, "learning_rate": 1.9587422885246647e-05, "loss": 0.3531444787979126, "step": 4350 }, { "epoch": 0.13898090484813685, "grad_norm": 0.49579060077667236, "learning_rate": 1.958497253564987e-05, "loss": 0.34980833530426025, "step": 4360 }, { "epoch": 0.13956980698732388, "grad_norm": 0.34414634108543396, "learning_rate": 1.9582515085292625e-05, "loss": 0.3600553035736084, "step": 4370 }, { "epoch": 0.1401587091265109, "grad_norm": 0.28972095251083374, "learning_rate": 1.9580050535995455e-05, "loss": 0.34074323177337645, "step": 4380 }, { "epoch": 0.14074761126569793, "grad_norm": 0.33913278579711914, "learning_rate": 1.9577578889584146e-05, "loss": 0.36076414585113525, "step": 4390 }, { "epoch": 0.14133651340488496, "grad_norm": 0.4062053859233856, "learning_rate": 1.9575100147889752e-05, "loss": 0.3597541093826294, "step": 4400 }, { "epoch": 0.14192541554407195, "grad_norm": 0.36677029728889465, "learning_rate": 1.9572614312748574e-05, "loss": 0.35508482456207274, "step": 4410 }, { "epoch": 0.14251431768325898, "grad_norm": 0.400703489780426, "learning_rate": 1.9570121386002182e-05, "loss": 0.31672699451446534, "step": 4420 }, { "epoch": 0.143103219822446, "grad_norm": 0.41097626090049744, "learning_rate": 1.9567621369497384e-05, "loss": 0.34503655433654784, "step": 4430 }, { "epoch": 0.14369212196163303, "grad_norm": 0.4233058989048004, "learning_rate": 1.9565114265086254e-05, "loss": 0.34176433086395264, "step": 4440 }, { "epoch": 0.14428102410082005, "grad_norm": 0.47049835324287415, "learning_rate": 1.9562600074626104e-05, "loss": 0.3550875663757324, "step": 4450 }, { "epoch": 0.14486992624000708, "grad_norm": 0.365561842918396, "learning_rate": 1.956007879997951e-05, "loss": 0.37898266315460205, "step": 4460 }, { "epoch": 0.14545882837919408, "grad_norm": 0.3830397129058838, "learning_rate": 1.955755044301428e-05, "loss": 0.3137513160705566, "step": 4470 }, { "epoch": 0.1460477305183811, "grad_norm": 0.41378650069236755, "learning_rate": 1.9555015005603485e-05, "loss": 0.35283448696136477, "step": 4480 }, { "epoch": 0.14663663265756813, "grad_norm": 0.45452654361724854, "learning_rate": 1.9552472489625426e-05, "loss": 0.369162917137146, "step": 4490 }, { "epoch": 0.14722553479675515, "grad_norm": 0.36307984590530396, "learning_rate": 1.954992289696366e-05, "loss": 0.32602477073669434, "step": 4500 }, { "epoch": 0.14781443693594218, "grad_norm": 0.33574187755584717, "learning_rate": 1.9547366229506984e-05, "loss": 0.35888123512268066, "step": 4510 }, { "epoch": 0.1484033390751292, "grad_norm": 0.350274920463562, "learning_rate": 1.9544802489149423e-05, "loss": 0.33026506900787356, "step": 4520 }, { "epoch": 0.1489922412143162, "grad_norm": 0.37705662846565247, "learning_rate": 1.9542231677790268e-05, "loss": 0.3376604080200195, "step": 4530 }, { "epoch": 0.14958114335350323, "grad_norm": 0.34810078144073486, "learning_rate": 1.953965379733402e-05, "loss": 0.36969723701477053, "step": 4540 }, { "epoch": 0.15017004549269025, "grad_norm": 0.4486561119556427, "learning_rate": 1.9537068849690437e-05, "loss": 0.3893007755279541, "step": 4550 }, { "epoch": 0.15075894763187728, "grad_norm": 0.3682912588119507, "learning_rate": 1.9534476836774506e-05, "loss": 0.36459085941314695, "step": 4560 }, { "epoch": 0.1513478497710643, "grad_norm": 0.4376179277896881, "learning_rate": 1.9531877760506443e-05, "loss": 0.3600914478302002, "step": 4570 }, { "epoch": 0.15193675191025133, "grad_norm": 0.3780360221862793, "learning_rate": 1.952927162281171e-05, "loss": 0.3614177703857422, "step": 4580 }, { "epoch": 0.15252565404943833, "grad_norm": 0.3256906270980835, "learning_rate": 1.9526658425620977e-05, "loss": 0.33696417808532714, "step": 4590 }, { "epoch": 0.15311455618862535, "grad_norm": 0.4643029272556305, "learning_rate": 1.9524038170870173e-05, "loss": 0.31309847831726073, "step": 4600 }, { "epoch": 0.15370345832781238, "grad_norm": 0.323414146900177, "learning_rate": 1.9521410860500432e-05, "loss": 0.36131308078765867, "step": 4610 }, { "epoch": 0.1542923604669994, "grad_norm": 0.4564089775085449, "learning_rate": 1.951877649645813e-05, "loss": 0.3289318323135376, "step": 4620 }, { "epoch": 0.15488126260618643, "grad_norm": 0.4721972942352295, "learning_rate": 1.9516135080694857e-05, "loss": 0.4065412521362305, "step": 4630 }, { "epoch": 0.15547016474537345, "grad_norm": 0.46390628814697266, "learning_rate": 1.951348661516744e-05, "loss": 0.3615289211273193, "step": 4640 }, { "epoch": 0.15605906688456045, "grad_norm": 0.2660098373889923, "learning_rate": 1.9510831101837912e-05, "loss": 0.3549086093902588, "step": 4650 }, { "epoch": 0.15664796902374747, "grad_norm": 0.4660618007183075, "learning_rate": 1.950816854267354e-05, "loss": 0.35413825511932373, "step": 4660 }, { "epoch": 0.1572368711629345, "grad_norm": 0.28706806898117065, "learning_rate": 1.9505498939646807e-05, "loss": 0.3686026334762573, "step": 4670 }, { "epoch": 0.15782577330212152, "grad_norm": 0.35940805077552795, "learning_rate": 1.9502822294735416e-05, "loss": 0.3380800008773804, "step": 4680 }, { "epoch": 0.15841467544130855, "grad_norm": 0.3398219048976898, "learning_rate": 1.9500138609922286e-05, "loss": 0.3259181022644043, "step": 4690 }, { "epoch": 0.15900357758049555, "grad_norm": 0.4000222086906433, "learning_rate": 1.9497447887195542e-05, "loss": 0.34556257724761963, "step": 4700 }, { "epoch": 0.15959247971968257, "grad_norm": 0.39331474900245667, "learning_rate": 1.949475012854854e-05, "loss": 0.32714698314666746, "step": 4710 }, { "epoch": 0.1601813818588696, "grad_norm": 0.4767175614833832, "learning_rate": 1.9492045335979836e-05, "loss": 0.3834760904312134, "step": 4720 }, { "epoch": 0.16077028399805662, "grad_norm": 0.3517037034034729, "learning_rate": 1.9489333511493202e-05, "loss": 0.3433971643447876, "step": 4730 }, { "epoch": 0.16135918613724365, "grad_norm": 0.2833130955696106, "learning_rate": 1.9486614657097612e-05, "loss": 0.39803826808929443, "step": 4740 }, { "epoch": 0.16194808827643067, "grad_norm": 0.2875945270061493, "learning_rate": 1.9483888774807263e-05, "loss": 0.30677592754364014, "step": 4750 }, { "epoch": 0.16253699041561767, "grad_norm": 0.34686827659606934, "learning_rate": 1.9481155866641542e-05, "loss": 0.3381738424301147, "step": 4760 }, { "epoch": 0.1631258925548047, "grad_norm": 0.5017157793045044, "learning_rate": 1.947841593462505e-05, "loss": 0.3478159666061401, "step": 4770 }, { "epoch": 0.16371479469399172, "grad_norm": 0.3217681646347046, "learning_rate": 1.947566898078759e-05, "loss": 0.38387269973754884, "step": 4780 }, { "epoch": 0.16430369683317875, "grad_norm": 0.281624972820282, "learning_rate": 1.947291500716417e-05, "loss": 0.3533159732818604, "step": 4790 }, { "epoch": 0.16489259897236577, "grad_norm": 0.44366517663002014, "learning_rate": 1.9470154015794985e-05, "loss": 0.3490705966949463, "step": 4800 }, { "epoch": 0.1654815011115528, "grad_norm": 0.29895246028900146, "learning_rate": 1.946738600872545e-05, "loss": 0.3250757932662964, "step": 4810 }, { "epoch": 0.1660704032507398, "grad_norm": 0.43396562337875366, "learning_rate": 1.9464610988006156e-05, "loss": 0.3375032663345337, "step": 4820 }, { "epoch": 0.16665930538992682, "grad_norm": 0.33238139748573303, "learning_rate": 1.9461828955692906e-05, "loss": 0.35816154479980467, "step": 4830 }, { "epoch": 0.16724820752911385, "grad_norm": 0.34477075934410095, "learning_rate": 1.9459039913846694e-05, "loss": 0.42202444076538087, "step": 4840 }, { "epoch": 0.16783710966830087, "grad_norm": 0.42859041690826416, "learning_rate": 1.9456243864533698e-05, "loss": 0.3423308372497559, "step": 4850 }, { "epoch": 0.1684260118074879, "grad_norm": 0.4235915541648865, "learning_rate": 1.9453440809825292e-05, "loss": 0.3233405828475952, "step": 4860 }, { "epoch": 0.16901491394667492, "grad_norm": 0.2769205868244171, "learning_rate": 1.945063075179805e-05, "loss": 0.3434084177017212, "step": 4870 }, { "epoch": 0.16960381608586192, "grad_norm": 0.444341242313385, "learning_rate": 1.944781369253372e-05, "loss": 0.39083764553070066, "step": 4880 }, { "epoch": 0.17019271822504894, "grad_norm": 0.36332398653030396, "learning_rate": 1.9444989634119236e-05, "loss": 0.3892350435256958, "step": 4890 }, { "epoch": 0.17078162036423597, "grad_norm": 0.2785060703754425, "learning_rate": 1.9442158578646736e-05, "loss": 0.31912741661071775, "step": 4900 }, { "epoch": 0.171370522503423, "grad_norm": 0.3680092692375183, "learning_rate": 1.943932052821352e-05, "loss": 0.34790334701538084, "step": 4910 }, { "epoch": 0.17195942464261002, "grad_norm": 0.3061811327934265, "learning_rate": 1.943647548492208e-05, "loss": 0.32892475128173826, "step": 4920 }, { "epoch": 0.17254832678179705, "grad_norm": 0.43211883306503296, "learning_rate": 1.943362345088009e-05, "loss": 0.33555381298065184, "step": 4930 }, { "epoch": 0.17313722892098404, "grad_norm": 0.4171203374862671, "learning_rate": 1.9430764428200395e-05, "loss": 0.3575333595275879, "step": 4940 }, { "epoch": 0.17372613106017107, "grad_norm": 0.5524247884750366, "learning_rate": 1.942789841900103e-05, "loss": 0.31523220539093016, "step": 4950 }, { "epoch": 0.1743150331993581, "grad_norm": 0.49512261152267456, "learning_rate": 1.942502542540519e-05, "loss": 0.34351847171783445, "step": 4960 }, { "epoch": 0.17490393533854512, "grad_norm": 0.3623511493206024, "learning_rate": 1.9422145449541257e-05, "loss": 0.3225362777709961, "step": 4970 }, { "epoch": 0.17549283747773214, "grad_norm": 0.3999245762825012, "learning_rate": 1.9419258493542785e-05, "loss": 0.3162348985671997, "step": 4980 }, { "epoch": 0.17608173961691917, "grad_norm": 0.38360288739204407, "learning_rate": 1.9416364559548485e-05, "loss": 0.3464597940444946, "step": 4990 }, { "epoch": 0.17667064175610617, "grad_norm": 0.3222116231918335, "learning_rate": 1.941346364970226e-05, "loss": 0.41750454902648926, "step": 5000 }, { "epoch": 0.1772595438952932, "grad_norm": 0.319277822971344, "learning_rate": 1.9410555766153153e-05, "loss": 0.3399773836135864, "step": 5010 }, { "epoch": 0.17784844603448022, "grad_norm": 0.5658179521560669, "learning_rate": 1.9407640911055404e-05, "loss": 0.37204837799072266, "step": 5020 }, { "epoch": 0.17843734817366724, "grad_norm": 0.3561462163925171, "learning_rate": 1.9404719086568393e-05, "loss": 0.353580117225647, "step": 5030 }, { "epoch": 0.17902625031285427, "grad_norm": 0.3087005615234375, "learning_rate": 1.9401790294856676e-05, "loss": 0.3811809539794922, "step": 5040 }, { "epoch": 0.1796151524520413, "grad_norm": 0.35339170694351196, "learning_rate": 1.9398854538089964e-05, "loss": 0.3410301446914673, "step": 5050 }, { "epoch": 0.1802040545912283, "grad_norm": 0.38856926560401917, "learning_rate": 1.9395911818443134e-05, "loss": 0.35084168910980223, "step": 5060 }, { "epoch": 0.18079295673041532, "grad_norm": 0.5336427092552185, "learning_rate": 1.9392962138096218e-05, "loss": 0.37982590198516847, "step": 5070 }, { "epoch": 0.18138185886960234, "grad_norm": 0.2694058120250702, "learning_rate": 1.9390005499234405e-05, "loss": 0.3243702411651611, "step": 5080 }, { "epoch": 0.18197076100878937, "grad_norm": 0.624387800693512, "learning_rate": 1.938704190404803e-05, "loss": 0.36205310821533204, "step": 5090 }, { "epoch": 0.1825596631479764, "grad_norm": 0.4540765881538391, "learning_rate": 1.9384071354732604e-05, "loss": 0.32845985889434814, "step": 5100 }, { "epoch": 0.18314856528716342, "grad_norm": 0.5112797021865845, "learning_rate": 1.938109385348877e-05, "loss": 0.33001596927642823, "step": 5110 }, { "epoch": 0.18373746742635041, "grad_norm": 0.2632962465286255, "learning_rate": 1.937810940252232e-05, "loss": 0.35652382373809816, "step": 5120 }, { "epoch": 0.18432636956553744, "grad_norm": 0.39654380083084106, "learning_rate": 1.9375118004044212e-05, "loss": 0.3480530738830566, "step": 5130 }, { "epoch": 0.18491527170472447, "grad_norm": 0.2995850443840027, "learning_rate": 1.9372119660270537e-05, "loss": 0.35982651710510255, "step": 5140 }, { "epoch": 0.1855041738439115, "grad_norm": 0.4218921959400177, "learning_rate": 1.9369114373422534e-05, "loss": 0.38222179412841795, "step": 5150 }, { "epoch": 0.18609307598309852, "grad_norm": 0.311526894569397, "learning_rate": 1.9366102145726584e-05, "loss": 0.3595756769180298, "step": 5160 }, { "epoch": 0.18668197812228554, "grad_norm": 0.36336952447891235, "learning_rate": 1.9363082979414217e-05, "loss": 0.36009218692779543, "step": 5170 }, { "epoch": 0.18727088026147254, "grad_norm": 0.4175403118133545, "learning_rate": 1.9360056876722093e-05, "loss": 0.32066547870635986, "step": 5180 }, { "epoch": 0.18785978240065956, "grad_norm": 0.33498746156692505, "learning_rate": 1.935702383989202e-05, "loss": 0.3413220405578613, "step": 5190 }, { "epoch": 0.1884486845398466, "grad_norm": 0.3336363136768341, "learning_rate": 1.9353983871170938e-05, "loss": 0.3706406354904175, "step": 5200 }, { "epoch": 0.18903758667903361, "grad_norm": 0.28715190291404724, "learning_rate": 1.9350936972810923e-05, "loss": 0.31749184131622316, "step": 5210 }, { "epoch": 0.18962648881822064, "grad_norm": 0.33433932065963745, "learning_rate": 1.9347883147069184e-05, "loss": 0.3236706256866455, "step": 5220 }, { "epoch": 0.19021539095740767, "grad_norm": 0.4574109613895416, "learning_rate": 1.9344822396208065e-05, "loss": 0.2998748779296875, "step": 5230 }, { "epoch": 0.19080429309659466, "grad_norm": 0.4006776511669159, "learning_rate": 1.9341754722495037e-05, "loss": 0.33463616371154786, "step": 5240 }, { "epoch": 0.1913931952357817, "grad_norm": 0.3217872977256775, "learning_rate": 1.93386801282027e-05, "loss": 0.3178912878036499, "step": 5250 }, { "epoch": 0.1919820973749687, "grad_norm": 0.41075778007507324, "learning_rate": 1.9335598615608787e-05, "loss": 0.31856513023376465, "step": 5260 }, { "epoch": 0.19257099951415574, "grad_norm": 0.43858978152275085, "learning_rate": 1.933251018699614e-05, "loss": 0.3461496829986572, "step": 5270 }, { "epoch": 0.19315990165334276, "grad_norm": 0.3249024450778961, "learning_rate": 1.9329414844652747e-05, "loss": 0.32003169059753417, "step": 5280 }, { "epoch": 0.1937488037925298, "grad_norm": 0.30814671516418457, "learning_rate": 1.93263125908717e-05, "loss": 0.35569875240325927, "step": 5290 }, { "epoch": 0.1943377059317168, "grad_norm": 0.34764233231544495, "learning_rate": 1.932320342795122e-05, "loss": 0.3104766845703125, "step": 5300 }, { "epoch": 0.1949266080709038, "grad_norm": 0.3945125937461853, "learning_rate": 1.932008735819464e-05, "loss": 0.39055213928222654, "step": 5310 }, { "epoch": 0.19551551021009084, "grad_norm": 0.37002015113830566, "learning_rate": 1.931696438391042e-05, "loss": 0.3784043550491333, "step": 5320 }, { "epoch": 0.19610441234927786, "grad_norm": 0.3255281448364258, "learning_rate": 1.931383450741212e-05, "loss": 0.3594492197036743, "step": 5330 }, { "epoch": 0.1966933144884649, "grad_norm": 0.3012765645980835, "learning_rate": 1.9310697731018432e-05, "loss": 0.3819704294204712, "step": 5340 }, { "epoch": 0.1972822166276519, "grad_norm": 0.32620561122894287, "learning_rate": 1.9307554057053143e-05, "loss": 0.2987483263015747, "step": 5350 }, { "epoch": 0.1978711187668389, "grad_norm": 0.5583878755569458, "learning_rate": 1.930440348784516e-05, "loss": 0.33310518264770506, "step": 5360 }, { "epoch": 0.19846002090602594, "grad_norm": 0.34953203797340393, "learning_rate": 1.9301246025728494e-05, "loss": 0.35765149593353274, "step": 5370 }, { "epoch": 0.19904892304521296, "grad_norm": 0.34738239645957947, "learning_rate": 1.929808167304226e-05, "loss": 0.3176546812057495, "step": 5380 }, { "epoch": 0.1996378251844, "grad_norm": 0.3600005805492401, "learning_rate": 1.929491043213068e-05, "loss": 0.31638121604919434, "step": 5390 }, { "epoch": 0.200226727323587, "grad_norm": 0.3891764283180237, "learning_rate": 1.9291732305343083e-05, "loss": 0.3286361932754517, "step": 5400 }, { "epoch": 0.20081562946277404, "grad_norm": 0.40190720558166504, "learning_rate": 1.9288547295033897e-05, "loss": 0.3743762969970703, "step": 5410 }, { "epoch": 0.20140453160196103, "grad_norm": 0.38915151357650757, "learning_rate": 1.9285355403562648e-05, "loss": 0.34850468635559084, "step": 5420 }, { "epoch": 0.20199343374114806, "grad_norm": 0.32579168677330017, "learning_rate": 1.9282156633293957e-05, "loss": 0.33887591361999514, "step": 5430 }, { "epoch": 0.20258233588033508, "grad_norm": 0.30491894483566284, "learning_rate": 1.9278950986597546e-05, "loss": 0.3300605535507202, "step": 5440 }, { "epoch": 0.2031712380195221, "grad_norm": 0.4268820881843567, "learning_rate": 1.9275738465848226e-05, "loss": 0.36262080669403074, "step": 5450 }, { "epoch": 0.20376014015870914, "grad_norm": 0.39529848098754883, "learning_rate": 1.927251907342591e-05, "loss": 0.3232415676116943, "step": 5460 }, { "epoch": 0.20434904229789616, "grad_norm": 0.43254342675209045, "learning_rate": 1.9269292811715594e-05, "loss": 0.341535758972168, "step": 5470 }, { "epoch": 0.20493794443708316, "grad_norm": 0.4622741639614105, "learning_rate": 1.9266059683107355e-05, "loss": 0.38448162078857423, "step": 5480 }, { "epoch": 0.20552684657627018, "grad_norm": 0.405993789434433, "learning_rate": 1.9262819689996383e-05, "loss": 0.34380488395690917, "step": 5490 }, { "epoch": 0.2061157487154572, "grad_norm": 0.3986377716064453, "learning_rate": 1.925957283478292e-05, "loss": 0.3515615463256836, "step": 5500 }, { "epoch": 0.20670465085464423, "grad_norm": 0.4691663384437561, "learning_rate": 1.925631911987232e-05, "loss": 0.32768921852111815, "step": 5510 }, { "epoch": 0.20729355299383126, "grad_norm": 0.41800087690353394, "learning_rate": 1.9253058547675002e-05, "loss": 0.4168867111206055, "step": 5520 }, { "epoch": 0.20788245513301826, "grad_norm": 0.32558324933052063, "learning_rate": 1.9249791120606472e-05, "loss": 0.3407518625259399, "step": 5530 }, { "epoch": 0.20847135727220528, "grad_norm": 0.4589636027812958, "learning_rate": 1.924651684108731e-05, "loss": 0.3495429039001465, "step": 5540 }, { "epoch": 0.2090602594113923, "grad_norm": 0.38787806034088135, "learning_rate": 1.924323571154318e-05, "loss": 0.3741002559661865, "step": 5550 }, { "epoch": 0.20964916155057933, "grad_norm": 0.4227662980556488, "learning_rate": 1.923994773440481e-05, "loss": 0.3285939931869507, "step": 5560 }, { "epoch": 0.21023806368976636, "grad_norm": 0.30622485280036926, "learning_rate": 1.9236652912108012e-05, "loss": 0.35839707851409913, "step": 5570 }, { "epoch": 0.21082696582895338, "grad_norm": 0.2941175401210785, "learning_rate": 1.923335124709366e-05, "loss": 0.3683901786804199, "step": 5580 }, { "epoch": 0.21141586796814038, "grad_norm": 0.40459367632865906, "learning_rate": 1.9230042741807703e-05, "loss": 0.3703796863555908, "step": 5590 }, { "epoch": 0.2120047701073274, "grad_norm": 0.47950279712677, "learning_rate": 1.922672739870115e-05, "loss": 0.32418456077575686, "step": 5600 }, { "epoch": 0.21259367224651443, "grad_norm": 0.6299868226051331, "learning_rate": 1.9223405220230088e-05, "loss": 0.34185354709625243, "step": 5610 }, { "epoch": 0.21318257438570146, "grad_norm": 0.28527894616127014, "learning_rate": 1.922007620885565e-05, "loss": 0.3136728048324585, "step": 5620 }, { "epoch": 0.21377147652488848, "grad_norm": 0.42076173424720764, "learning_rate": 1.9216740367044056e-05, "loss": 0.3722181797027588, "step": 5630 }, { "epoch": 0.2143603786640755, "grad_norm": 0.4104028344154358, "learning_rate": 1.921339769726656e-05, "loss": 0.32635447978973386, "step": 5640 }, { "epoch": 0.2149492808032625, "grad_norm": 0.3604673743247986, "learning_rate": 1.921004820199949e-05, "loss": 0.31961090564727784, "step": 5650 }, { "epoch": 0.21553818294244953, "grad_norm": 0.4634048640727997, "learning_rate": 1.9206691883724228e-05, "loss": 0.3572373628616333, "step": 5660 }, { "epoch": 0.21612708508163656, "grad_norm": 0.34200552105903625, "learning_rate": 1.9203328744927208e-05, "loss": 0.29799709320068357, "step": 5670 }, { "epoch": 0.21671598722082358, "grad_norm": 0.41354504227638245, "learning_rate": 1.9199958788099916e-05, "loss": 0.33680174350738523, "step": 5680 }, { "epoch": 0.2173048893600106, "grad_norm": 0.4618787467479706, "learning_rate": 1.9196582015738895e-05, "loss": 0.305244517326355, "step": 5690 }, { "epoch": 0.21789379149919763, "grad_norm": 0.33308058977127075, "learning_rate": 1.9193198430345727e-05, "loss": 0.3595688581466675, "step": 5700 }, { "epoch": 0.21848269363838463, "grad_norm": 0.32838737964630127, "learning_rate": 1.9189808034427053e-05, "loss": 0.34522619247436526, "step": 5710 }, { "epoch": 0.21907159577757165, "grad_norm": 0.32584068179130554, "learning_rate": 1.918641083049455e-05, "loss": 0.33600773811340334, "step": 5720 }, { "epoch": 0.21966049791675868, "grad_norm": 0.4814643859863281, "learning_rate": 1.9183006821064945e-05, "loss": 0.37623484134674073, "step": 5730 }, { "epoch": 0.2202494000559457, "grad_norm": 0.4097294807434082, "learning_rate": 1.917959600866e-05, "loss": 0.33853731155395506, "step": 5740 }, { "epoch": 0.22083830219513273, "grad_norm": 0.3052961230278015, "learning_rate": 1.9176178395806524e-05, "loss": 0.3484332084655762, "step": 5750 }, { "epoch": 0.22142720433431975, "grad_norm": 0.3587331473827362, "learning_rate": 1.9172753985036358e-05, "loss": 0.3808445453643799, "step": 5760 }, { "epoch": 0.22201610647350675, "grad_norm": 0.3678300082683563, "learning_rate": 1.916932277888638e-05, "loss": 0.33621342182159425, "step": 5770 }, { "epoch": 0.22260500861269378, "grad_norm": 0.26930516958236694, "learning_rate": 1.916588477989851e-05, "loss": 0.40053744316101075, "step": 5780 }, { "epoch": 0.2231939107518808, "grad_norm": 0.4368574917316437, "learning_rate": 1.916243999061969e-05, "loss": 0.3005595445632935, "step": 5790 }, { "epoch": 0.22378281289106783, "grad_norm": 0.42014551162719727, "learning_rate": 1.9158988413601897e-05, "loss": 0.35243213176727295, "step": 5800 }, { "epoch": 0.22437171503025485, "grad_norm": 0.32240304350852966, "learning_rate": 1.9155530051402132e-05, "loss": 0.36419217586517333, "step": 5810 }, { "epoch": 0.22496061716944188, "grad_norm": 0.3341010510921478, "learning_rate": 1.915206490658243e-05, "loss": 0.3367243766784668, "step": 5820 }, { "epoch": 0.22554951930862888, "grad_norm": 0.3256852924823761, "learning_rate": 1.9148592981709854e-05, "loss": 0.34184701442718507, "step": 5830 }, { "epoch": 0.2261384214478159, "grad_norm": 0.6045019030570984, "learning_rate": 1.9145114279356466e-05, "loss": 0.3376529455184937, "step": 5840 }, { "epoch": 0.22672732358700293, "grad_norm": 0.3571658432483673, "learning_rate": 1.9141628802099384e-05, "loss": 0.3290455102920532, "step": 5850 }, { "epoch": 0.22731622572618995, "grad_norm": 0.5257264971733093, "learning_rate": 1.9138136552520717e-05, "loss": 0.3663639068603516, "step": 5860 }, { "epoch": 0.22790512786537698, "grad_norm": 0.3706505000591278, "learning_rate": 1.9134637533207602e-05, "loss": 0.3178016185760498, "step": 5870 }, { "epoch": 0.228494030004564, "grad_norm": 0.3579416573047638, "learning_rate": 1.913113174675219e-05, "loss": 0.3323383331298828, "step": 5880 }, { "epoch": 0.229082932143751, "grad_norm": 0.29987281560897827, "learning_rate": 1.912761919575165e-05, "loss": 0.3567740201950073, "step": 5890 }, { "epoch": 0.22967183428293803, "grad_norm": 0.32137593626976013, "learning_rate": 1.9124099882808152e-05, "loss": 0.3611839771270752, "step": 5900 }, { "epoch": 0.23026073642212505, "grad_norm": 0.3326427638530731, "learning_rate": 1.9120573810528887e-05, "loss": 0.3729482889175415, "step": 5910 }, { "epoch": 0.23084963856131208, "grad_norm": 0.26235708594322205, "learning_rate": 1.9117040981526044e-05, "loss": 0.36013343334198, "step": 5920 }, { "epoch": 0.2314385407004991, "grad_norm": 0.4158004820346832, "learning_rate": 1.9113501398416816e-05, "loss": 0.3498688697814941, "step": 5930 }, { "epoch": 0.23202744283968613, "grad_norm": 0.3621995151042938, "learning_rate": 1.9109955063823416e-05, "loss": 0.3480418920516968, "step": 5940 }, { "epoch": 0.23261634497887312, "grad_norm": 0.5027883648872375, "learning_rate": 1.910640198037304e-05, "loss": 0.3425896644592285, "step": 5950 }, { "epoch": 0.23320524711806015, "grad_norm": 0.32797208428382874, "learning_rate": 1.9102842150697893e-05, "loss": 0.361010217666626, "step": 5960 }, { "epoch": 0.23379414925724717, "grad_norm": 0.39303940534591675, "learning_rate": 1.9099275577435174e-05, "loss": 0.3370736360549927, "step": 5970 }, { "epoch": 0.2343830513964342, "grad_norm": 0.5471805334091187, "learning_rate": 1.9095702263227082e-05, "loss": 0.34532463550567627, "step": 5980 }, { "epoch": 0.23497195353562123, "grad_norm": 0.34963223338127136, "learning_rate": 1.9092122210720804e-05, "loss": 0.3439953804016113, "step": 5990 }, { "epoch": 0.23556085567480825, "grad_norm": 0.34326741099357605, "learning_rate": 1.908853542256853e-05, "loss": 0.3019896984100342, "step": 6000 }, { "epoch": 0.0005889021391870206, "grad_norm": 0.33294183015823364, "learning_rate": 1.9084941901427428e-05, "loss": 0.33980751037597656, "step": 6010 }, { "epoch": 0.0011778042783740412, "grad_norm": 0.38579535484313965, "learning_rate": 1.9081341649959655e-05, "loss": 0.34964427947998045, "step": 6020 }, { "epoch": 0.0017667064175610617, "grad_norm": 0.3382600247859955, "learning_rate": 1.907773467083236e-05, "loss": 0.3123203992843628, "step": 6030 }, { "epoch": 0.0023556085567480825, "grad_norm": 0.40939387679100037, "learning_rate": 1.907412096671767e-05, "loss": 0.34806864261627196, "step": 6040 }, { "epoch": 0.002944510695935103, "grad_norm": 0.5862222909927368, "learning_rate": 1.9070500540292705e-05, "loss": 0.3373312711715698, "step": 6050 }, { "epoch": 0.0035334128351221235, "grad_norm": 0.44338226318359375, "learning_rate": 1.906687339423955e-05, "loss": 0.2760844469070435, "step": 6060 }, { "epoch": 0.004122314974309144, "grad_norm": 0.4359859228134155, "learning_rate": 1.9063239531245282e-05, "loss": 0.32126970291137696, "step": 6070 }, { "epoch": 0.004711217113496165, "grad_norm": 0.3573221266269684, "learning_rate": 1.905959895400194e-05, "loss": 0.30707354545593263, "step": 6080 }, { "epoch": 0.005300119252683185, "grad_norm": 0.40812551975250244, "learning_rate": 1.905595166520655e-05, "loss": 0.3413190126419067, "step": 6090 }, { "epoch": 0.005889021391870206, "grad_norm": 0.3227921426296234, "learning_rate": 1.9052297667561098e-05, "loss": 0.34063377380371096, "step": 6100 }, { "epoch": 0.006477923531057227, "grad_norm": 0.44783005118370056, "learning_rate": 1.9048636963772557e-05, "loss": 0.3298830270767212, "step": 6110 }, { "epoch": 0.007066825670244247, "grad_norm": 0.30037808418273926, "learning_rate": 1.9044969556552847e-05, "loss": 0.33820807933807373, "step": 6120 }, { "epoch": 0.007655727809431268, "grad_norm": 0.47846779227256775, "learning_rate": 1.904129544861887e-05, "loss": 0.3618520975112915, "step": 6130 }, { "epoch": 0.008244629948618288, "grad_norm": 0.3711289167404175, "learning_rate": 1.9037614642692487e-05, "loss": 0.3845088243484497, "step": 6140 }, { "epoch": 0.00883353208780531, "grad_norm": 0.3398926556110382, "learning_rate": 1.9033927141500518e-05, "loss": 0.3438566207885742, "step": 6150 }, { "epoch": 0.00942243422699233, "grad_norm": 0.5356273651123047, "learning_rate": 1.903023294777475e-05, "loss": 0.36286215782165526, "step": 6160 }, { "epoch": 0.01001133636617935, "grad_norm": 0.4085761308670044, "learning_rate": 1.902653206425192e-05, "loss": 0.33037471771240234, "step": 6170 }, { "epoch": 0.01060023850536637, "grad_norm": 0.3843350112438202, "learning_rate": 1.9022824493673725e-05, "loss": 0.3440325975418091, "step": 6180 }, { "epoch": 0.011189140644553392, "grad_norm": 0.33052459359169006, "learning_rate": 1.901911023878682e-05, "loss": 0.3776798486709595, "step": 6190 }, { "epoch": 0.011778042783740412, "grad_norm": 0.36497923731803894, "learning_rate": 1.90153893023428e-05, "loss": 0.3137929916381836, "step": 6200 }, { "epoch": 0.012366944922927432, "grad_norm": 0.36098185181617737, "learning_rate": 1.9011661687098222e-05, "loss": 0.3452603816986084, "step": 6210 }, { "epoch": 0.012955847062114454, "grad_norm": 0.4105343818664551, "learning_rate": 1.900792739581459e-05, "loss": 0.35447494983673095, "step": 6220 }, { "epoch": 0.013544749201301474, "grad_norm": 0.40811142325401306, "learning_rate": 1.900418643125834e-05, "loss": 0.33590824604034425, "step": 6230 }, { "epoch": 0.014133651340488494, "grad_norm": 0.37506169080734253, "learning_rate": 1.900043879620087e-05, "loss": 0.3398254871368408, "step": 6240 }, { "epoch": 0.014722553479675516, "grad_norm": 0.43585440516471863, "learning_rate": 1.8996684493418505e-05, "loss": 0.3014153242111206, "step": 6250 }, { "epoch": 0.015311455618862536, "grad_norm": 0.3972862958908081, "learning_rate": 1.8992923525692522e-05, "loss": 0.3400855302810669, "step": 6260 }, { "epoch": 0.015900357758049558, "grad_norm": 0.33220595121383667, "learning_rate": 1.8989155895809124e-05, "loss": 0.3219140529632568, "step": 6270 }, { "epoch": 0.016489259897236576, "grad_norm": 0.5383298993110657, "learning_rate": 1.8985381606559458e-05, "loss": 0.3069839239120483, "step": 6280 }, { "epoch": 0.017078162036423598, "grad_norm": 0.5083141922950745, "learning_rate": 1.8981600660739602e-05, "loss": 0.3292830467224121, "step": 6290 }, { "epoch": 0.01766706417561062, "grad_norm": 0.4113636612892151, "learning_rate": 1.8977813061150563e-05, "loss": 0.3235149621963501, "step": 6300 }, { "epoch": 0.018255966314797638, "grad_norm": 0.2899695634841919, "learning_rate": 1.8974018810598277e-05, "loss": 0.3429668426513672, "step": 6310 }, { "epoch": 0.01884486845398466, "grad_norm": 0.47155115008354187, "learning_rate": 1.8970217911893614e-05, "loss": 0.2798586368560791, "step": 6320 }, { "epoch": 0.01943377059317168, "grad_norm": 0.3762354850769043, "learning_rate": 1.896641036785236e-05, "loss": 0.29409730434417725, "step": 6330 }, { "epoch": 0.0200226727323587, "grad_norm": 0.3030949532985687, "learning_rate": 1.8962596181295228e-05, "loss": 0.30261027812957764, "step": 6340 }, { "epoch": 0.02061157487154572, "grad_norm": 0.5158665776252747, "learning_rate": 1.8958775355047857e-05, "loss": 0.33803558349609375, "step": 6350 }, { "epoch": 0.02120047701073274, "grad_norm": 0.27912232279777527, "learning_rate": 1.8954947891940793e-05, "loss": 0.3150915622711182, "step": 6360 }, { "epoch": 0.02178937914991976, "grad_norm": 0.4139896333217621, "learning_rate": 1.895111379480951e-05, "loss": 0.29991438388824465, "step": 6370 }, { "epoch": 0.022378281289106784, "grad_norm": 0.37455320358276367, "learning_rate": 1.8947273066494394e-05, "loss": 0.3454668760299683, "step": 6380 }, { "epoch": 0.022967183428293802, "grad_norm": 0.45149052143096924, "learning_rate": 1.8943425709840736e-05, "loss": 0.28916804790496825, "step": 6390 }, { "epoch": 0.023556085567480824, "grad_norm": 0.4145452678203583, "learning_rate": 1.893957172769875e-05, "loss": 0.3082667350769043, "step": 6400 }, { "epoch": 0.024144987706667845, "grad_norm": 0.48954659700393677, "learning_rate": 1.8935711122923545e-05, "loss": 0.29708328247070315, "step": 6410 }, { "epoch": 0.024733889845854864, "grad_norm": 0.6098476052284241, "learning_rate": 1.8931843898375146e-05, "loss": 0.32892308235168455, "step": 6420 }, { "epoch": 0.025322791985041886, "grad_norm": 0.3716776371002197, "learning_rate": 1.892797005691848e-05, "loss": 0.2925910234451294, "step": 6430 }, { "epoch": 0.025911694124228907, "grad_norm": 0.2832426130771637, "learning_rate": 1.892408960142337e-05, "loss": 0.30242950916290284, "step": 6440 }, { "epoch": 0.026500596263415926, "grad_norm": 0.4275444746017456, "learning_rate": 1.8920202534764552e-05, "loss": 0.35083608627319335, "step": 6450 }, { "epoch": 0.027089498402602948, "grad_norm": 0.2936806380748749, "learning_rate": 1.8916308859821643e-05, "loss": 0.3184484958648682, "step": 6460 }, { "epoch": 0.02767840054178997, "grad_norm": 0.5813065767288208, "learning_rate": 1.8912408579479167e-05, "loss": 0.3627588748931885, "step": 6470 }, { "epoch": 0.028267302680976988, "grad_norm": 0.45354995131492615, "learning_rate": 1.8908501696626536e-05, "loss": 0.3104326486587524, "step": 6480 }, { "epoch": 0.02885620482016401, "grad_norm": 0.4790821969509125, "learning_rate": 1.890458821415806e-05, "loss": 0.35127475261688235, "step": 6490 }, { "epoch": 0.02944510695935103, "grad_norm": 0.49345073103904724, "learning_rate": 1.8900668134972934e-05, "loss": 0.3545774221420288, "step": 6500 }, { "epoch": 0.03003400909853805, "grad_norm": 0.37492114305496216, "learning_rate": 1.8896741461975236e-05, "loss": 0.3005502223968506, "step": 6510 }, { "epoch": 0.03062291123772507, "grad_norm": 0.6516212821006775, "learning_rate": 1.889280819807393e-05, "loss": 0.3346642255783081, "step": 6520 }, { "epoch": 0.031211813376912093, "grad_norm": 0.4631853401660919, "learning_rate": 1.888886834618287e-05, "loss": 0.31685624122619627, "step": 6530 }, { "epoch": 0.031800715516099115, "grad_norm": 0.4188757538795471, "learning_rate": 1.888492190922079e-05, "loss": 0.2862930059432983, "step": 6540 }, { "epoch": 0.03238961765528613, "grad_norm": 0.384168803691864, "learning_rate": 1.8880968890111288e-05, "loss": 0.3061197757720947, "step": 6550 }, { "epoch": 0.03297851979447315, "grad_norm": 0.4417096972465515, "learning_rate": 1.8877009291782854e-05, "loss": 0.3119416475296021, "step": 6560 }, { "epoch": 0.03356742193366018, "grad_norm": 0.3827332854270935, "learning_rate": 1.8873043117168842e-05, "loss": 0.2922665119171143, "step": 6570 }, { "epoch": 0.034156324072847195, "grad_norm": 0.32135483622550964, "learning_rate": 1.8869070369207492e-05, "loss": 0.31396567821502686, "step": 6580 }, { "epoch": 0.034745226212034214, "grad_norm": 0.3784911334514618, "learning_rate": 1.8865091050841898e-05, "loss": 0.3256594896316528, "step": 6590 }, { "epoch": 0.03533412835122124, "grad_norm": 0.5342302322387695, "learning_rate": 1.886110516502002e-05, "loss": 0.34003937244415283, "step": 6600 }, { "epoch": 0.03592303049040826, "grad_norm": 0.39827096462249756, "learning_rate": 1.8857112714694704e-05, "loss": 0.3316729784011841, "step": 6610 }, { "epoch": 0.036511932629595276, "grad_norm": 0.37929797172546387, "learning_rate": 1.8853113702823636e-05, "loss": 0.34579896926879883, "step": 6620 }, { "epoch": 0.0371008347687823, "grad_norm": 0.36576342582702637, "learning_rate": 1.884910813236938e-05, "loss": 0.3195429563522339, "step": 6630 }, { "epoch": 0.03768973690796932, "grad_norm": 0.4533837139606476, "learning_rate": 1.8845096006299343e-05, "loss": 0.3219160556793213, "step": 6640 }, { "epoch": 0.03827863904715634, "grad_norm": 0.5741034150123596, "learning_rate": 1.8841077327585807e-05, "loss": 0.37753958702087403, "step": 6650 }, { "epoch": 0.03886754118634336, "grad_norm": 0.551394522190094, "learning_rate": 1.883705209920589e-05, "loss": 0.3080314874649048, "step": 6660 }, { "epoch": 0.03945644332553038, "grad_norm": 0.48944413661956787, "learning_rate": 1.8833020324141578e-05, "loss": 0.3413158178329468, "step": 6670 }, { "epoch": 0.0400453454647174, "grad_norm": 0.4582391381263733, "learning_rate": 1.882898200537969e-05, "loss": 0.3642479181289673, "step": 6680 }, { "epoch": 0.04063424760390442, "grad_norm": 0.4074981212615967, "learning_rate": 1.882493714591191e-05, "loss": 0.3119884729385376, "step": 6690 }, { "epoch": 0.04122314974309144, "grad_norm": 0.4694328010082245, "learning_rate": 1.8820885748734757e-05, "loss": 0.30898730754852294, "step": 6700 }, { "epoch": 0.04181205188227846, "grad_norm": 0.41390666365623474, "learning_rate": 1.88168278168496e-05, "loss": 0.3018419504165649, "step": 6710 }, { "epoch": 0.04240095402146548, "grad_norm": 0.503276526927948, "learning_rate": 1.8812763353262635e-05, "loss": 0.3303172826766968, "step": 6720 }, { "epoch": 0.042989856160652505, "grad_norm": 0.5782155990600586, "learning_rate": 1.8808692360984917e-05, "loss": 0.3018216609954834, "step": 6730 }, { "epoch": 0.04357875829983952, "grad_norm": 0.40546154975891113, "learning_rate": 1.8804614843032324e-05, "loss": 0.3280763626098633, "step": 6740 }, { "epoch": 0.04416766043902654, "grad_norm": 0.39916181564331055, "learning_rate": 1.880053080242557e-05, "loss": 0.34003138542175293, "step": 6750 }, { "epoch": 0.04475656257821357, "grad_norm": 0.4973351061344147, "learning_rate": 1.8796440242190207e-05, "loss": 0.3417187213897705, "step": 6760 }, { "epoch": 0.045345464717400585, "grad_norm": 0.4385925829410553, "learning_rate": 1.8792343165356607e-05, "loss": 0.31786537170410156, "step": 6770 }, { "epoch": 0.045934366856587604, "grad_norm": 0.44145429134368896, "learning_rate": 1.878823957495998e-05, "loss": 0.2931888818740845, "step": 6780 }, { "epoch": 0.04652326899577463, "grad_norm": 0.5780170559883118, "learning_rate": 1.8784129474040354e-05, "loss": 0.348368501663208, "step": 6790 }, { "epoch": 0.04711217113496165, "grad_norm": 0.42527514696121216, "learning_rate": 1.8780012865642586e-05, "loss": 0.319173002243042, "step": 6800 }, { "epoch": 0.047701073274148666, "grad_norm": 0.4721429944038391, "learning_rate": 1.877588975281635e-05, "loss": 0.33538064956665037, "step": 6810 }, { "epoch": 0.04828997541333569, "grad_norm": 0.49376386404037476, "learning_rate": 1.8771760138616138e-05, "loss": 0.37085413932800293, "step": 6820 }, { "epoch": 0.04887887755252271, "grad_norm": 0.5191429853439331, "learning_rate": 1.8767624026101265e-05, "loss": 0.31192061901092527, "step": 6830 }, { "epoch": 0.04946777969170973, "grad_norm": 0.449126660823822, "learning_rate": 1.8763481418335846e-05, "loss": 0.3095136880874634, "step": 6840 }, { "epoch": 0.05005668183089675, "grad_norm": 0.42593568563461304, "learning_rate": 1.8759332318388822e-05, "loss": 0.3530463218688965, "step": 6850 }, { "epoch": 0.05064558397008377, "grad_norm": 0.4863432049751282, "learning_rate": 1.8755176729333942e-05, "loss": 0.3451318502426147, "step": 6860 }, { "epoch": 0.05123448610927079, "grad_norm": 0.4638890326023102, "learning_rate": 1.875101465424975e-05, "loss": 0.33043909072875977, "step": 6870 }, { "epoch": 0.051823388248457815, "grad_norm": 0.44819706678390503, "learning_rate": 1.8746846096219613e-05, "loss": 0.2879673480987549, "step": 6880 }, { "epoch": 0.05241229038764483, "grad_norm": 0.38105639815330505, "learning_rate": 1.8742671058331686e-05, "loss": 0.29657397270202634, "step": 6890 }, { "epoch": 0.05300119252683185, "grad_norm": 0.4615638554096222, "learning_rate": 1.873848954367893e-05, "loss": 0.28893115520477297, "step": 6900 }, { "epoch": 0.05359009466601888, "grad_norm": 0.4831949472427368, "learning_rate": 1.8734301555359104e-05, "loss": 0.29406445026397704, "step": 6910 }, { "epoch": 0.054178996805205895, "grad_norm": 0.4373294413089752, "learning_rate": 1.873010709647476e-05, "loss": 0.3043278455734253, "step": 6920 }, { "epoch": 0.05476789894439291, "grad_norm": 0.7469714283943176, "learning_rate": 1.8725906170133252e-05, "loss": 0.3056368589401245, "step": 6930 }, { "epoch": 0.05535680108357994, "grad_norm": 0.49857208132743835, "learning_rate": 1.8721698779446716e-05, "loss": 0.35424964427947997, "step": 6940 }, { "epoch": 0.05594570322276696, "grad_norm": 0.5402007102966309, "learning_rate": 1.8717484927532083e-05, "loss": 0.3272466421127319, "step": 6950 }, { "epoch": 0.056534605361953975, "grad_norm": 0.6509560346603394, "learning_rate": 1.8713264617511068e-05, "loss": 0.3221544027328491, "step": 6960 }, { "epoch": 0.057123507501141, "grad_norm": 0.5780757665634155, "learning_rate": 1.8709037852510165e-05, "loss": 0.31449203491210936, "step": 6970 }, { "epoch": 0.05771240964032802, "grad_norm": 0.3802502155303955, "learning_rate": 1.870480463566066e-05, "loss": 0.3145084619522095, "step": 6980 }, { "epoch": 0.05830131177951504, "grad_norm": 0.5190765261650085, "learning_rate": 1.8700564970098615e-05, "loss": 0.3384179353713989, "step": 6990 }, { "epoch": 0.05889021391870206, "grad_norm": 0.42626041173934937, "learning_rate": 1.8696318858964865e-05, "loss": 0.2922704219818115, "step": 7000 }, { "epoch": 0.0005889021391870206, "grad_norm": 0.4670228660106659, "learning_rate": 1.8692066305405024e-05, "loss": 0.32336034774780276, "step": 7010 }, { "epoch": 0.0011778042783740412, "grad_norm": 0.5152823328971863, "learning_rate": 1.8687807312569483e-05, "loss": 0.3304766654968262, "step": 7020 }, { "epoch": 0.0017667064175610617, "grad_norm": 0.48382753133773804, "learning_rate": 1.868354188361339e-05, "loss": 0.29819140434265134, "step": 7030 }, { "epoch": 0.0023556085567480825, "grad_norm": 0.5226377844810486, "learning_rate": 1.8679270021696675e-05, "loss": 0.32792413234710693, "step": 7040 }, { "epoch": 0.002944510695935103, "grad_norm": 0.7019636034965515, "learning_rate": 1.8674991729984028e-05, "loss": 0.31871623992919923, "step": 7050 }, { "epoch": 0.0035334128351221235, "grad_norm": 0.5537834763526917, "learning_rate": 1.86707070116449e-05, "loss": 0.2593334674835205, "step": 7060 }, { "epoch": 0.004122314974309144, "grad_norm": 0.5452659726142883, "learning_rate": 1.866641586985351e-05, "loss": 0.3006539583206177, "step": 7070 }, { "epoch": 0.004711217113496165, "grad_norm": 0.4653977155685425, "learning_rate": 1.8662118307788827e-05, "loss": 0.2906734704971313, "step": 7080 }, { "epoch": 0.005300119252683185, "grad_norm": 0.5211588740348816, "learning_rate": 1.865781432863458e-05, "loss": 0.3227615118026733, "step": 7090 }, { "epoch": 0.005889021391870206, "grad_norm": 0.42735543847084045, "learning_rate": 1.8653503935579256e-05, "loss": 0.324461030960083, "step": 7100 }, { "epoch": 0.006477923531057227, "grad_norm": 0.5588556528091431, "learning_rate": 1.864918713181609e-05, "loss": 0.3091172456741333, "step": 7110 }, { "epoch": 0.007066825670244247, "grad_norm": 0.40968185663223267, "learning_rate": 1.8644863920543058e-05, "loss": 0.32048099040985106, "step": 7120 }, { "epoch": 0.007655727809431268, "grad_norm": 0.5716606974601746, "learning_rate": 1.8640534304962898e-05, "loss": 0.33917195796966554, "step": 7130 }, { "epoch": 0.008244629948618288, "grad_norm": 0.5098016858100891, "learning_rate": 1.8636198288283087e-05, "loss": 0.3608222961425781, "step": 7140 }, { "epoch": 0.00883353208780531, "grad_norm": 0.44653716683387756, "learning_rate": 1.8631855873715837e-05, "loss": 0.3257593154907227, "step": 7150 }, { "epoch": 0.00942243422699233, "grad_norm": 0.5824905037879944, "learning_rate": 1.862750706447811e-05, "loss": 0.34039530754089353, "step": 7160 }, { "epoch": 0.01001133636617935, "grad_norm": 0.5256181359291077, "learning_rate": 1.8623151863791597e-05, "loss": 0.3106685638427734, "step": 7170 }, { "epoch": 0.01060023850536637, "grad_norm": 0.5326651930809021, "learning_rate": 1.8618790274882728e-05, "loss": 0.32616212368011477, "step": 7180 }, { "epoch": 0.011189140644553392, "grad_norm": 0.43696507811546326, "learning_rate": 1.8614422300982664e-05, "loss": 0.35401144027709963, "step": 7190 }, { "epoch": 0.011778042783740412, "grad_norm": 0.4762822389602661, "learning_rate": 1.86100479453273e-05, "loss": 0.2947083473205566, "step": 7200 }, { "epoch": 0.012366944922927432, "grad_norm": 0.46605056524276733, "learning_rate": 1.8605667211157254e-05, "loss": 0.32283482551574705, "step": 7210 }, { "epoch": 0.012955847062114454, "grad_norm": 0.49954235553741455, "learning_rate": 1.8601280101717872e-05, "loss": 0.3323812961578369, "step": 7220 }, { "epoch": 0.013544749201301474, "grad_norm": 0.5314835906028748, "learning_rate": 1.859688662025922e-05, "loss": 0.3149606466293335, "step": 7230 }, { "epoch": 0.014133651340488494, "grad_norm": 0.48421570658683777, "learning_rate": 1.8592486770036087e-05, "loss": 0.3186631202697754, "step": 7240 }, { "epoch": 0.014722553479675516, "grad_norm": 0.5565971732139587, "learning_rate": 1.8588080554307984e-05, "loss": 0.28166680335998534, "step": 7250 }, { "epoch": 0.015311455618862536, "grad_norm": 0.5287695527076721, "learning_rate": 1.858366797633913e-05, "loss": 0.31664609909057617, "step": 7260 }, { "epoch": 0.015900357758049558, "grad_norm": 0.4422435164451599, "learning_rate": 1.857924903939846e-05, "loss": 0.3016538619995117, "step": 7270 }, { "epoch": 0.016489259897236576, "grad_norm": 0.6503362655639648, "learning_rate": 1.857482374675962e-05, "loss": 0.2866894006729126, "step": 7280 }, { "epoch": 0.017078162036423598, "grad_norm": 0.6423026919364929, "learning_rate": 1.8570392101700968e-05, "loss": 0.3062369585037231, "step": 7290 }, { "epoch": 0.01766706417561062, "grad_norm": 0.5810728073120117, "learning_rate": 1.8565954107505563e-05, "loss": 0.2995160579681396, "step": 7300 }, { "epoch": 0.018255966314797638, "grad_norm": 0.39179208874702454, "learning_rate": 1.856150976746117e-05, "loss": 0.3214749813079834, "step": 7310 }, { "epoch": 0.01884486845398466, "grad_norm": 0.6131115555763245, "learning_rate": 1.8557059084860253e-05, "loss": 0.26139225959777834, "step": 7320 }, { "epoch": 0.01943377059317168, "grad_norm": 0.5276258587837219, "learning_rate": 1.855260206299998e-05, "loss": 0.2753570795059204, "step": 7330 }, { "epoch": 0.0200226727323587, "grad_norm": 0.4048614799976349, "learning_rate": 1.8548138705182213e-05, "loss": 0.2805802822113037, "step": 7340 }, { "epoch": 0.02061157487154572, "grad_norm": 0.5778293609619141, "learning_rate": 1.8543669014713503e-05, "loss": 0.3178767919540405, "step": 7350 }, { "epoch": 0.02120047701073274, "grad_norm": 0.36737576127052307, "learning_rate": 1.8539192994905095e-05, "loss": 0.29636483192443847, "step": 7360 }, { "epoch": 0.02178937914991976, "grad_norm": 0.5123438239097595, "learning_rate": 1.8534710649072926e-05, "loss": 0.2797044515609741, "step": 7370 }, { "epoch": 0.022378281289106784, "grad_norm": 0.48234421014785767, "learning_rate": 1.8530221980537617e-05, "loss": 0.3216846466064453, "step": 7380 }, { "epoch": 0.022967183428293802, "grad_norm": 0.6405563950538635, "learning_rate": 1.8525726992624473e-05, "loss": 0.26858437061309814, "step": 7390 }, { "epoch": 0.023556085567480824, "grad_norm": 0.5551750063896179, "learning_rate": 1.852122568866348e-05, "loss": 0.2876331090927124, "step": 7400 }, { "epoch": 0.024144987706667845, "grad_norm": 0.6201905608177185, "learning_rate": 1.8516718071989307e-05, "loss": 0.27775614261627196, "step": 7410 }, { "epoch": 0.024733889845854864, "grad_norm": 0.7327443361282349, "learning_rate": 1.8512204145941296e-05, "loss": 0.3094367027282715, "step": 7420 }, { "epoch": 0.025322791985041886, "grad_norm": 0.46031653881073, "learning_rate": 1.8507683913863456e-05, "loss": 0.27566373348236084, "step": 7430 }, { "epoch": 0.025911694124228907, "grad_norm": 0.38727280497550964, "learning_rate": 1.8503157379104485e-05, "loss": 0.281298565864563, "step": 7440 }, { "epoch": 0.026500596263415926, "grad_norm": 0.549418032169342, "learning_rate": 1.849862454501774e-05, "loss": 0.33087332248687745, "step": 7450 }, { "epoch": 0.027089498402602948, "grad_norm": 0.39114809036254883, "learning_rate": 1.849408541496124e-05, "loss": 0.2979250907897949, "step": 7460 }, { "epoch": 0.02767840054178997, "grad_norm": 0.7451658844947815, "learning_rate": 1.8489539992297674e-05, "loss": 0.33554198741912844, "step": 7470 }, { "epoch": 0.028267302680976988, "grad_norm": 0.5872669219970703, "learning_rate": 1.848498828039439e-05, "loss": 0.28809494972229005, "step": 7480 }, { "epoch": 0.02885620482016401, "grad_norm": 0.5828214883804321, "learning_rate": 1.8480430282623407e-05, "loss": 0.32876129150390626, "step": 7490 }, { "epoch": 0.02944510695935103, "grad_norm": 0.6419627070426941, "learning_rate": 1.8475866002361383e-05, "loss": 0.33149688243865966, "step": 7500 }, { "epoch": 0.03003400909853805, "grad_norm": 0.49241793155670166, "learning_rate": 1.8471295442989636e-05, "loss": 0.2826056003570557, "step": 7510 }, { "epoch": 0.03062291123772507, "grad_norm": 0.8149327635765076, "learning_rate": 1.8466718607894147e-05, "loss": 0.310204815864563, "step": 7520 }, { "epoch": 0.031211813376912093, "grad_norm": 0.6114128828048706, "learning_rate": 1.8462135500465527e-05, "loss": 0.2954772233963013, "step": 7530 }, { "epoch": 0.031800715516099115, "grad_norm": 0.5291472673416138, "learning_rate": 1.845754612409905e-05, "loss": 0.2691232681274414, "step": 7540 }, { "epoch": 0.03238961765528613, "grad_norm": 0.48778867721557617, "learning_rate": 1.8452950482194628e-05, "loss": 0.28377976417541506, "step": 7550 }, { "epoch": 0.03297851979447315, "grad_norm": 0.5150596499443054, "learning_rate": 1.844834857815681e-05, "loss": 0.2952188730239868, "step": 7560 }, { "epoch": 0.03356742193366018, "grad_norm": 0.49722176790237427, "learning_rate": 1.8443740415394793e-05, "loss": 0.2729278326034546, "step": 7570 }, { "epoch": 0.034156324072847195, "grad_norm": 0.44024261832237244, "learning_rate": 1.8439125997322406e-05, "loss": 0.29323663711547854, "step": 7580 }, { "epoch": 0.034745226212034214, "grad_norm": 0.461325079202652, "learning_rate": 1.8434505327358107e-05, "loss": 0.3031508207321167, "step": 7590 }, { "epoch": 0.03533412835122124, "grad_norm": 0.667325496673584, "learning_rate": 1.8429878408924997e-05, "loss": 0.3174131870269775, "step": 7600 }, { "epoch": 0.03592303049040826, "grad_norm": 0.5020397305488586, "learning_rate": 1.84252452454508e-05, "loss": 0.3072798728942871, "step": 7610 }, { "epoch": 0.036511932629595276, "grad_norm": 0.4997575581073761, "learning_rate": 1.8420605840367863e-05, "loss": 0.32510809898376464, "step": 7620 }, { "epoch": 0.0371008347687823, "grad_norm": 0.4914683401584625, "learning_rate": 1.841596019711316e-05, "loss": 0.29755799770355223, "step": 7630 }, { "epoch": 0.03768973690796932, "grad_norm": 0.5340143442153931, "learning_rate": 1.8411308319128292e-05, "loss": 0.302083683013916, "step": 7640 }, { "epoch": 0.03827863904715634, "grad_norm": 0.7549685835838318, "learning_rate": 1.8406650209859468e-05, "loss": 0.3505316972732544, "step": 7650 }, { "epoch": 0.03886754118634336, "grad_norm": 0.6807670593261719, "learning_rate": 1.8401985872757517e-05, "loss": 0.2864945650100708, "step": 7660 }, { "epoch": 0.03945644332553038, "grad_norm": 0.6540709137916565, "learning_rate": 1.839731531127789e-05, "loss": 0.3159742832183838, "step": 7670 }, { "epoch": 0.0400453454647174, "grad_norm": 0.5888230204582214, "learning_rate": 1.839263852888064e-05, "loss": 0.3363761901855469, "step": 7680 }, { "epoch": 0.04063424760390442, "grad_norm": 0.5552172064781189, "learning_rate": 1.8387955529030434e-05, "loss": 0.29251296520233155, "step": 7690 }, { "epoch": 0.04122314974309144, "grad_norm": 0.5863175988197327, "learning_rate": 1.8383266315196533e-05, "loss": 0.2849805593490601, "step": 7700 }, { "epoch": 0.04181205188227846, "grad_norm": 0.5256586074829102, "learning_rate": 1.837857089085282e-05, "loss": 0.2827972650527954, "step": 7710 }, { "epoch": 0.04240095402146548, "grad_norm": 0.6052111387252808, "learning_rate": 1.8373869259477765e-05, "loss": 0.3042726755142212, "step": 7720 }, { "epoch": 0.042989856160652505, "grad_norm": 0.7223770022392273, "learning_rate": 1.8369161424554445e-05, "loss": 0.2804349660873413, "step": 7730 }, { "epoch": 0.04357875829983952, "grad_norm": 0.5192850232124329, "learning_rate": 1.8364447389570522e-05, "loss": 0.30552082061767577, "step": 7740 }, { "epoch": 0.04416766043902654, "grad_norm": 0.5204986929893494, "learning_rate": 1.8359727158018266e-05, "loss": 0.3142412185668945, "step": 7750 }, { "epoch": 0.04475656257821357, "grad_norm": 0.6546810865402222, "learning_rate": 1.8355000733394522e-05, "loss": 0.31967737674713137, "step": 7760 }, { "epoch": 0.045345464717400585, "grad_norm": 0.599091112613678, "learning_rate": 1.8350268119200736e-05, "loss": 0.295152473449707, "step": 7770 }, { "epoch": 0.045934366856587604, "grad_norm": 0.5906034111976624, "learning_rate": 1.834552931894293e-05, "loss": 0.26719496250152586, "step": 7780 }, { "epoch": 0.04652326899577463, "grad_norm": 0.7184613347053528, "learning_rate": 1.8340784336131715e-05, "loss": 0.3206007719039917, "step": 7790 }, { "epoch": 0.04711217113496165, "grad_norm": 0.5639662146568298, "learning_rate": 1.8336033174282276e-05, "loss": 0.29834699630737305, "step": 7800 }, { "epoch": 0.047701073274148666, "grad_norm": 0.599759578704834, "learning_rate": 1.8331275836914383e-05, "loss": 0.31459650993347166, "step": 7810 }, { "epoch": 0.04828997541333569, "grad_norm": 0.6120832562446594, "learning_rate": 1.832651232755237e-05, "loss": 0.3445911169052124, "step": 7820 }, { "epoch": 0.04887887755252271, "grad_norm": 0.6756870746612549, "learning_rate": 1.8321742649725163e-05, "loss": 0.287298583984375, "step": 7830 }, { "epoch": 0.04946777969170973, "grad_norm": 0.5867186188697815, "learning_rate": 1.8316966806966237e-05, "loss": 0.2891498327255249, "step": 7840 }, { "epoch": 0.05005668183089675, "grad_norm": 0.5390121340751648, "learning_rate": 1.831218480281364e-05, "loss": 0.32871150970458984, "step": 7850 }, { "epoch": 0.05064558397008377, "grad_norm": 0.6184263229370117, "learning_rate": 1.8307396640809992e-05, "loss": 0.3199973821640015, "step": 7860 }, { "epoch": 0.05123448610927079, "grad_norm": 0.5995836853981018, "learning_rate": 1.830260232450247e-05, "loss": 0.3047621250152588, "step": 7870 }, { "epoch": 0.051823388248457815, "grad_norm": 0.6007784008979797, "learning_rate": 1.829780185744281e-05, "loss": 0.26698405742645265, "step": 7880 }, { "epoch": 0.05241229038764483, "grad_norm": 0.5096645355224609, "learning_rate": 1.8292995243187298e-05, "loss": 0.27388753890991213, "step": 7890 }, { "epoch": 0.05300119252683185, "grad_norm": 0.5879145860671997, "learning_rate": 1.8288182485296786e-05, "loss": 0.2656082630157471, "step": 7900 }, { "epoch": 0.05359009466601888, "grad_norm": 0.6038352847099304, "learning_rate": 1.828336358733667e-05, "loss": 0.2756047248840332, "step": 7910 }, { "epoch": 0.054178996805205895, "grad_norm": 0.5859820246696472, "learning_rate": 1.8278538552876902e-05, "loss": 0.2797574758529663, "step": 7920 }, { "epoch": 0.05476789894439291, "grad_norm": 0.9440249800682068, "learning_rate": 1.827370738549197e-05, "loss": 0.2747712850570679, "step": 7930 }, { "epoch": 0.05535680108357994, "grad_norm": 0.69752037525177, "learning_rate": 1.826887008876091e-05, "loss": 0.3240897417068481, "step": 7940 }, { "epoch": 0.05594570322276696, "grad_norm": 0.6957165002822876, "learning_rate": 1.8264026666267295e-05, "loss": 0.30165762901306153, "step": 7950 }, { "epoch": 0.056534605361953975, "grad_norm": 0.8371090888977051, "learning_rate": 1.8259177121599247e-05, "loss": 0.2995856523513794, "step": 7960 }, { "epoch": 0.057123507501141, "grad_norm": 0.7484301924705505, "learning_rate": 1.8254321458349413e-05, "loss": 0.2882041931152344, "step": 7970 }, { "epoch": 0.05771240964032802, "grad_norm": 0.5141607522964478, "learning_rate": 1.8249459680114978e-05, "loss": 0.291724419593811, "step": 7980 }, { "epoch": 0.05830131177951504, "grad_norm": 0.6654802560806274, "learning_rate": 1.8244591790497655e-05, "loss": 0.31154632568359375, "step": 7990 }, { "epoch": 0.05889021391870206, "grad_norm": 0.5754157900810242, "learning_rate": 1.823971779310368e-05, "loss": 0.2685930013656616, "step": 8000 }, { "epoch": 0.05947911605788908, "grad_norm": 0.7153839468955994, "learning_rate": 1.8234837691543826e-05, "loss": 0.31938810348510743, "step": 8010 }, { "epoch": 0.0600680181970761, "grad_norm": 0.5771943926811218, "learning_rate": 1.8229951489433376e-05, "loss": 0.3407033681869507, "step": 8020 }, { "epoch": 0.060656920336263125, "grad_norm": 0.7616878151893616, "learning_rate": 1.8225059190392136e-05, "loss": 0.3160083770751953, "step": 8030 }, { "epoch": 0.06124582247545014, "grad_norm": 0.4769116938114166, "learning_rate": 1.8220160798044434e-05, "loss": 0.30030920505523684, "step": 8040 }, { "epoch": 0.06183472461463716, "grad_norm": 0.5229161977767944, "learning_rate": 1.8215256316019103e-05, "loss": 0.30566179752349854, "step": 8050 }, { "epoch": 0.062423626753824187, "grad_norm": 0.6544932723045349, "learning_rate": 1.8210345747949493e-05, "loss": 0.3026392459869385, "step": 8060 }, { "epoch": 0.0630125288930112, "grad_norm": 0.5496081113815308, "learning_rate": 1.8205429097473465e-05, "loss": 0.29754712581634524, "step": 8070 }, { "epoch": 0.06360143103219823, "grad_norm": 0.5344751477241516, "learning_rate": 1.8200506368233384e-05, "loss": 0.32501375675201416, "step": 8080 }, { "epoch": 0.06419033317138524, "grad_norm": 0.6754211783409119, "learning_rate": 1.819557756387611e-05, "loss": 0.32000908851623533, "step": 8090 }, { "epoch": 0.06477923531057227, "grad_norm": 0.44688841700553894, "learning_rate": 1.8190642688053018e-05, "loss": 0.36702795028686525, "step": 8100 }, { "epoch": 0.06536813744975929, "grad_norm": 0.5084608197212219, "learning_rate": 1.8185701744419968e-05, "loss": 0.3060584783554077, "step": 8110 }, { "epoch": 0.0659570395889463, "grad_norm": 0.6092038154602051, "learning_rate": 1.818075473663733e-05, "loss": 0.31388583183288576, "step": 8120 }, { "epoch": 0.06654594172813333, "grad_norm": 0.6749376058578491, "learning_rate": 1.817580166836994e-05, "loss": 0.3029475688934326, "step": 8130 }, { "epoch": 0.06713484386732035, "grad_norm": 0.5725331902503967, "learning_rate": 1.817084254328716e-05, "loss": 0.3033494710922241, "step": 8140 }, { "epoch": 0.06772374600650737, "grad_norm": 0.649511992931366, "learning_rate": 1.8165877365062807e-05, "loss": 0.3292105674743652, "step": 8150 }, { "epoch": 0.06831264814569439, "grad_norm": 0.5173346996307373, "learning_rate": 1.81609061373752e-05, "loss": 0.3359072208404541, "step": 8160 }, { "epoch": 0.06890155028488142, "grad_norm": 0.5079783797264099, "learning_rate": 1.8155928863907134e-05, "loss": 0.34200263023376465, "step": 8170 }, { "epoch": 0.06949045242406843, "grad_norm": 0.522568941116333, "learning_rate": 1.815094554834589e-05, "loss": 0.34421303272247317, "step": 8180 }, { "epoch": 0.07007935456325545, "grad_norm": 0.5756241083145142, "learning_rate": 1.8145956194383208e-05, "loss": 0.30488877296447753, "step": 8190 }, { "epoch": 0.07066825670244248, "grad_norm": 0.5565503835678101, "learning_rate": 1.8140960805715316e-05, "loss": 0.2986006259918213, "step": 8200 }, { "epoch": 0.07125715884162949, "grad_norm": 0.44994962215423584, "learning_rate": 1.813595938604291e-05, "loss": 0.3702967166900635, "step": 8210 }, { "epoch": 0.07184606098081651, "grad_norm": 0.523077666759491, "learning_rate": 1.813095193907116e-05, "loss": 0.3510415077209473, "step": 8220 }, { "epoch": 0.07243496312000354, "grad_norm": 0.6263032555580139, "learning_rate": 1.8125938468509677e-05, "loss": 0.32593045234680174, "step": 8230 }, { "epoch": 0.07302386525919055, "grad_norm": 0.5222161412239075, "learning_rate": 1.8120918978072567e-05, "loss": 0.30152652263641355, "step": 8240 }, { "epoch": 0.07361276739837758, "grad_norm": 0.43882477283477783, "learning_rate": 1.8115893471478374e-05, "loss": 0.2928597450256348, "step": 8250 }, { "epoch": 0.0742016695375646, "grad_norm": 0.5649147629737854, "learning_rate": 1.8110861952450104e-05, "loss": 0.3169670581817627, "step": 8260 }, { "epoch": 0.07479057167675161, "grad_norm": 0.5733256340026855, "learning_rate": 1.810582442471522e-05, "loss": 0.34437668323516846, "step": 8270 }, { "epoch": 0.07537947381593864, "grad_norm": 0.6135724186897278, "learning_rate": 1.810078089200563e-05, "loss": 0.28531076908111574, "step": 8280 }, { "epoch": 0.07596837595512566, "grad_norm": 0.44871217012405396, "learning_rate": 1.8095731358057698e-05, "loss": 0.3045381784439087, "step": 8290 }, { "epoch": 0.07655727809431268, "grad_norm": 0.5072545409202576, "learning_rate": 1.809067582661223e-05, "loss": 0.31702208518981934, "step": 8300 }, { "epoch": 0.0771461802334997, "grad_norm": 0.5509281754493713, "learning_rate": 1.8085614301414486e-05, "loss": 0.30524184703826907, "step": 8310 }, { "epoch": 0.07773508237268673, "grad_norm": 0.5679279565811157, "learning_rate": 1.808054678621414e-05, "loss": 0.3460038185119629, "step": 8320 }, { "epoch": 0.07832398451187374, "grad_norm": 0.6263118982315063, "learning_rate": 1.8075473284765327e-05, "loss": 0.29924683570861815, "step": 8330 }, { "epoch": 0.07891288665106076, "grad_norm": 0.5915454030036926, "learning_rate": 1.8070393800826612e-05, "loss": 0.26807918548583987, "step": 8340 }, { "epoch": 0.07950178879024777, "grad_norm": 0.4844604730606079, "learning_rate": 1.8065308338160984e-05, "loss": 0.2812062740325928, "step": 8350 }, { "epoch": 0.0800906909294348, "grad_norm": 0.5705661773681641, "learning_rate": 1.8060216900535872e-05, "loss": 0.30386080741882326, "step": 8360 }, { "epoch": 0.08067959306862182, "grad_norm": 0.529895007610321, "learning_rate": 1.8055119491723122e-05, "loss": 0.3108320951461792, "step": 8370 }, { "epoch": 0.08126849520780884, "grad_norm": 0.5491604804992676, "learning_rate": 1.805001611549901e-05, "loss": 0.31636178493499756, "step": 8380 }, { "epoch": 0.08185739734699586, "grad_norm": 0.4477883577346802, "learning_rate": 1.804490677564423e-05, "loss": 0.30482504367828367, "step": 8390 }, { "epoch": 0.08244629948618289, "grad_norm": 0.44800901412963867, "learning_rate": 1.8039791475943895e-05, "loss": 0.32025971412658694, "step": 8400 }, { "epoch": 0.0830352016253699, "grad_norm": 0.6632330417633057, "learning_rate": 1.8034670220187533e-05, "loss": 0.3191234111785889, "step": 8410 }, { "epoch": 0.08362410376455692, "grad_norm": 0.5309356451034546, "learning_rate": 1.802954301216908e-05, "loss": 0.29991374015808103, "step": 8420 }, { "epoch": 0.08421300590374395, "grad_norm": 0.6280932426452637, "learning_rate": 1.8024409855686887e-05, "loss": 0.29702742099761964, "step": 8430 }, { "epoch": 0.08480190804293096, "grad_norm": 0.6494479775428772, "learning_rate": 1.8019270754543716e-05, "loss": 0.32411842346191405, "step": 8440 }, { "epoch": 0.08539081018211798, "grad_norm": 0.560813844203949, "learning_rate": 1.801412571254672e-05, "loss": 0.31367785930633546, "step": 8450 }, { "epoch": 0.08597971232130501, "grad_norm": 0.57318514585495, "learning_rate": 1.800897473350746e-05, "loss": 0.328279447555542, "step": 8460 }, { "epoch": 0.08656861446049202, "grad_norm": 0.5380470752716064, "learning_rate": 1.8003817821241907e-05, "loss": 0.36828625202178955, "step": 8470 }, { "epoch": 0.08715751659967905, "grad_norm": 0.532735288143158, "learning_rate": 1.7998654979570403e-05, "loss": 0.3197162628173828, "step": 8480 }, { "epoch": 0.08774641873886607, "grad_norm": 0.46673163771629333, "learning_rate": 1.79934862123177e-05, "loss": 0.2847506046295166, "step": 8490 }, { "epoch": 0.08833532087805308, "grad_norm": 0.4169611632823944, "learning_rate": 1.798831152331294e-05, "loss": 0.33082387447357176, "step": 8500 }, { "epoch": 0.08892422301724011, "grad_norm": 0.5211071372032166, "learning_rate": 1.798313091638964e-05, "loss": 0.2868521213531494, "step": 8510 }, { "epoch": 0.08951312515642713, "grad_norm": 0.7379778027534485, "learning_rate": 1.7977944395385713e-05, "loss": 0.35359015464782717, "step": 8520 }, { "epoch": 0.09010202729561415, "grad_norm": 0.8079105019569397, "learning_rate": 1.7972751964143446e-05, "loss": 0.3335242748260498, "step": 8530 }, { "epoch": 0.09069092943480117, "grad_norm": 0.6004855036735535, "learning_rate": 1.7967553626509514e-05, "loss": 0.3412461280822754, "step": 8540 }, { "epoch": 0.0912798315739882, "grad_norm": 0.4581132233142853, "learning_rate": 1.7962349386334953e-05, "loss": 0.31906723976135254, "step": 8550 }, { "epoch": 0.09186873371317521, "grad_norm": 0.5638097524642944, "learning_rate": 1.7957139247475185e-05, "loss": 0.32842292785644533, "step": 8560 }, { "epoch": 0.09245763585236223, "grad_norm": 0.5912101864814758, "learning_rate": 1.7951923213789994e-05, "loss": 0.33472537994384766, "step": 8570 }, { "epoch": 0.09304653799154926, "grad_norm": 0.6408345699310303, "learning_rate": 1.7946701289143534e-05, "loss": 0.3040100336074829, "step": 8580 }, { "epoch": 0.09363544013073627, "grad_norm": 0.474580317735672, "learning_rate": 1.7941473477404324e-05, "loss": 0.2843565702438354, "step": 8590 }, { "epoch": 0.0942243422699233, "grad_norm": 0.5969856381416321, "learning_rate": 1.7936239782445238e-05, "loss": 0.3271888494491577, "step": 8600 }, { "epoch": 0.09481324440911032, "grad_norm": 0.56269770860672, "learning_rate": 1.7931000208143522e-05, "loss": 0.31423585414886473, "step": 8610 }, { "epoch": 0.09540214654829733, "grad_norm": 0.5924084186553955, "learning_rate": 1.7925754758380766e-05, "loss": 0.28243083953857423, "step": 8620 }, { "epoch": 0.09599104868748436, "grad_norm": 0.5676245093345642, "learning_rate": 1.792050343704291e-05, "loss": 0.2934816837310791, "step": 8630 }, { "epoch": 0.09657995082667138, "grad_norm": 0.567372739315033, "learning_rate": 1.7915246248020256e-05, "loss": 0.30671114921569825, "step": 8640 }, { "epoch": 0.0971688529658584, "grad_norm": 0.6174682974815369, "learning_rate": 1.790998319520744e-05, "loss": 0.31080405712127684, "step": 8650 }, { "epoch": 0.09775775510504542, "grad_norm": 0.5565900206565857, "learning_rate": 1.7904714282503455e-05, "loss": 0.33416457176208497, "step": 8660 }, { "epoch": 0.09834665724423244, "grad_norm": 0.5403124690055847, "learning_rate": 1.7899439513811626e-05, "loss": 0.3245969295501709, "step": 8670 }, { "epoch": 0.09893555938341946, "grad_norm": 0.7215847969055176, "learning_rate": 1.7894158893039617e-05, "loss": 0.3096026420593262, "step": 8680 }, { "epoch": 0.09952446152260648, "grad_norm": 0.5219393968582153, "learning_rate": 1.7888872424099433e-05, "loss": 0.28542404174804686, "step": 8690 }, { "epoch": 0.1001133636617935, "grad_norm": 0.6145344972610474, "learning_rate": 1.7883580110907402e-05, "loss": 0.32107884883880616, "step": 8700 }, { "epoch": 0.10070226580098052, "grad_norm": 0.5188111662864685, "learning_rate": 1.7878281957384192e-05, "loss": 0.3141686201095581, "step": 8710 }, { "epoch": 0.10129116794016754, "grad_norm": 0.5420286059379578, "learning_rate": 1.787297796745479e-05, "loss": 0.2729212999343872, "step": 8720 }, { "epoch": 0.10188007007935457, "grad_norm": 0.44228431582450867, "learning_rate": 1.7867668145048515e-05, "loss": 0.26797459125518797, "step": 8730 }, { "epoch": 0.10246897221854158, "grad_norm": 0.6908876895904541, "learning_rate": 1.7862352494098993e-05, "loss": 0.3344642400741577, "step": 8740 }, { "epoch": 0.1030578743577286, "grad_norm": 0.6478210091590881, "learning_rate": 1.7857031018544177e-05, "loss": 0.3215817451477051, "step": 8750 }, { "epoch": 0.10364677649691563, "grad_norm": 0.42139357328414917, "learning_rate": 1.7851703722326342e-05, "loss": 0.3613389015197754, "step": 8760 }, { "epoch": 0.10423567863610264, "grad_norm": 0.6784501075744629, "learning_rate": 1.784637060939206e-05, "loss": 0.3252772569656372, "step": 8770 }, { "epoch": 0.10482458077528967, "grad_norm": 0.5360516309738159, "learning_rate": 1.7841031683692224e-05, "loss": 0.29561748504638674, "step": 8780 }, { "epoch": 0.10541348291447669, "grad_norm": 0.5412061214447021, "learning_rate": 1.7835686949182025e-05, "loss": 0.2827857255935669, "step": 8790 }, { "epoch": 0.1060023850536637, "grad_norm": 0.46323704719543457, "learning_rate": 1.7830336409820967e-05, "loss": 0.28080737590789795, "step": 8800 }, { "epoch": 0.10659128719285073, "grad_norm": 0.5422869920730591, "learning_rate": 1.7824980069572838e-05, "loss": 0.3356359004974365, "step": 8810 }, { "epoch": 0.10718018933203775, "grad_norm": 0.4769618511199951, "learning_rate": 1.7819617932405745e-05, "loss": 0.3220425844192505, "step": 8820 }, { "epoch": 0.10776909147122476, "grad_norm": 0.6571393609046936, "learning_rate": 1.781425000229207e-05, "loss": 0.3209482431411743, "step": 8830 }, { "epoch": 0.10835799361041179, "grad_norm": 0.5600336194038391, "learning_rate": 1.78088762832085e-05, "loss": 0.30030393600463867, "step": 8840 }, { "epoch": 0.10894689574959882, "grad_norm": 0.48409608006477356, "learning_rate": 1.7803496779136e-05, "loss": 0.2724201440811157, "step": 8850 }, { "epoch": 0.10953579788878583, "grad_norm": 0.4519372284412384, "learning_rate": 1.7798111494059832e-05, "loss": 0.2949885606765747, "step": 8860 }, { "epoch": 0.11012470002797285, "grad_norm": 0.691805899143219, "learning_rate": 1.7792720431969535e-05, "loss": 0.30992741584777833, "step": 8870 }, { "epoch": 0.11071360216715988, "grad_norm": 0.5523871779441833, "learning_rate": 1.778732359685892e-05, "loss": 0.31872284412384033, "step": 8880 }, { "epoch": 0.11130250430634689, "grad_norm": 0.47554463148117065, "learning_rate": 1.7781920992726092e-05, "loss": 0.26727263927459716, "step": 8890 }, { "epoch": 0.11189140644553391, "grad_norm": 0.5204135179519653, "learning_rate": 1.7776512623573412e-05, "loss": 0.312812876701355, "step": 8900 }, { "epoch": 0.11248030858472094, "grad_norm": 0.5961715579032898, "learning_rate": 1.7771098493407525e-05, "loss": 0.28026399612426756, "step": 8910 }, { "epoch": 0.11306921072390795, "grad_norm": 0.44586679339408875, "learning_rate": 1.776567860623933e-05, "loss": 0.296142053604126, "step": 8920 }, { "epoch": 0.11365811286309498, "grad_norm": 0.5810695886611938, "learning_rate": 1.7760252966084013e-05, "loss": 0.30072593688964844, "step": 8930 }, { "epoch": 0.114247015002282, "grad_norm": 0.6419770121574402, "learning_rate": 1.7754821576961e-05, "loss": 0.3205129384994507, "step": 8940 }, { "epoch": 0.11483591714146901, "grad_norm": 0.40178006887435913, "learning_rate": 1.7749384442893988e-05, "loss": 0.30826168060302733, "step": 8950 }, { "epoch": 0.11542481928065604, "grad_norm": 0.46402880549430847, "learning_rate": 1.7743941567910926e-05, "loss": 0.30106043815612793, "step": 8960 }, { "epoch": 0.11601372141984306, "grad_norm": 0.445298969745636, "learning_rate": 1.773849295604402e-05, "loss": 0.291152286529541, "step": 8970 }, { "epoch": 0.11660262355903007, "grad_norm": 0.48656588792800903, "learning_rate": 1.773303861132971e-05, "loss": 0.3111910343170166, "step": 8980 }, { "epoch": 0.1171915256982171, "grad_norm": 0.5927103161811829, "learning_rate": 1.7727578537808712e-05, "loss": 0.3387373208999634, "step": 8990 }, { "epoch": 0.11778042783740413, "grad_norm": 0.568318247795105, "learning_rate": 1.7722112739525957e-05, "loss": 0.3451366901397705, "step": 9000 }, { "epoch": 0.11836932997659114, "grad_norm": 0.47397559881210327, "learning_rate": 1.771664122053064e-05, "loss": 0.3002439022064209, "step": 9010 }, { "epoch": 0.11895823211577816, "grad_norm": 0.4477004110813141, "learning_rate": 1.7711163984876175e-05, "loss": 0.31965763568878175, "step": 9020 }, { "epoch": 0.11954713425496519, "grad_norm": 0.6503050923347473, "learning_rate": 1.7705681036620226e-05, "loss": 0.3081531524658203, "step": 9030 }, { "epoch": 0.1201360363941522, "grad_norm": 0.7502251267433167, "learning_rate": 1.7700192379824676e-05, "loss": 0.33728666305541993, "step": 9040 }, { "epoch": 0.12072493853333922, "grad_norm": 0.58191978931427, "learning_rate": 1.7694698018555652e-05, "loss": 0.33299062252044676, "step": 9050 }, { "epoch": 0.12131384067252625, "grad_norm": 0.5400508642196655, "learning_rate": 1.7689197956883493e-05, "loss": 0.3296528339385986, "step": 9060 }, { "epoch": 0.12190274281171326, "grad_norm": 0.4875931441783905, "learning_rate": 1.768369219888277e-05, "loss": 0.3599841833114624, "step": 9070 }, { "epoch": 0.12249164495090029, "grad_norm": 0.5647627711296082, "learning_rate": 1.767818074863227e-05, "loss": 0.33332417011260984, "step": 9080 }, { "epoch": 0.12308054709008731, "grad_norm": 0.5367016792297363, "learning_rate": 1.7672663610214997e-05, "loss": 0.33733866214752195, "step": 9090 }, { "epoch": 0.12366944922927432, "grad_norm": 0.4060792624950409, "learning_rate": 1.7667140787718166e-05, "loss": 0.3575096607208252, "step": 9100 }, { "epoch": 0.12425835136846135, "grad_norm": 0.5416434407234192, "learning_rate": 1.7661612285233215e-05, "loss": 0.33249285221099856, "step": 9110 }, { "epoch": 0.12484725350764837, "grad_norm": 0.48278871178627014, "learning_rate": 1.7656078106855776e-05, "loss": 0.33504889011383054, "step": 9120 }, { "epoch": 0.12543615564683538, "grad_norm": 0.5664723515510559, "learning_rate": 1.7650538256685688e-05, "loss": 0.34177384376525877, "step": 9130 }, { "epoch": 0.1260250577860224, "grad_norm": 0.5942503809928894, "learning_rate": 1.7644992738827005e-05, "loss": 0.36244730949401854, "step": 9140 }, { "epoch": 0.12661395992520943, "grad_norm": 0.6301754117012024, "learning_rate": 1.763944155738796e-05, "loss": 0.3131944417953491, "step": 9150 }, { "epoch": 0.12720286206439646, "grad_norm": 0.5387119054794312, "learning_rate": 1.7633884716480996e-05, "loss": 0.35011563301086424, "step": 9160 }, { "epoch": 0.12779176420358346, "grad_norm": 0.7234879732131958, "learning_rate": 1.7628322220222744e-05, "loss": 0.4088891983032227, "step": 9170 }, { "epoch": 0.12838066634277048, "grad_norm": 0.5819865465164185, "learning_rate": 1.7622754072734022e-05, "loss": 0.3133880376815796, "step": 9180 }, { "epoch": 0.1289695684819575, "grad_norm": 0.5872946381568909, "learning_rate": 1.761718027813984e-05, "loss": 0.35115299224853513, "step": 9190 }, { "epoch": 0.12955847062114453, "grad_norm": 0.48805347084999084, "learning_rate": 1.7611600840569387e-05, "loss": 0.3060199737548828, "step": 9200 }, { "epoch": 0.13014737276033156, "grad_norm": 0.432907372713089, "learning_rate": 1.760601576415604e-05, "loss": 0.34428246021270753, "step": 9210 }, { "epoch": 0.13073627489951858, "grad_norm": 0.515501856803894, "learning_rate": 1.760042505303734e-05, "loss": 0.34322757720947267, "step": 9220 }, { "epoch": 0.13132517703870558, "grad_norm": 0.4413253962993622, "learning_rate": 1.759482871135501e-05, "loss": 0.33609833717346194, "step": 9230 }, { "epoch": 0.1319140791778926, "grad_norm": 0.5109516382217407, "learning_rate": 1.7589226743254954e-05, "loss": 0.35489342212677, "step": 9240 }, { "epoch": 0.13250298131707963, "grad_norm": 0.5526895523071289, "learning_rate": 1.7583619152887222e-05, "loss": 0.33297603130340575, "step": 9250 }, { "epoch": 0.13309188345626666, "grad_norm": 0.5462705492973328, "learning_rate": 1.7578005944406046e-05, "loss": 0.35762484073638917, "step": 9260 }, { "epoch": 0.13368078559545368, "grad_norm": 0.5664397478103638, "learning_rate": 1.757238712196982e-05, "loss": 0.28384287357330323, "step": 9270 }, { "epoch": 0.1342696877346407, "grad_norm": 0.4124225378036499, "learning_rate": 1.7566762689741086e-05, "loss": 0.3034247636795044, "step": 9280 }, { "epoch": 0.1348585898738277, "grad_norm": 0.41483551263809204, "learning_rate": 1.7561132651886552e-05, "loss": 0.3216688632965088, "step": 9290 }, { "epoch": 0.13544749201301473, "grad_norm": 0.4457356035709381, "learning_rate": 1.7555497012577074e-05, "loss": 0.3284491777420044, "step": 9300 }, { "epoch": 0.13603639415220176, "grad_norm": 0.45647573471069336, "learning_rate": 1.7549855775987655e-05, "loss": 0.31407151222229, "step": 9310 }, { "epoch": 0.13662529629138878, "grad_norm": 0.5767122507095337, "learning_rate": 1.754420894629746e-05, "loss": 0.32571580410003664, "step": 9320 }, { "epoch": 0.1372141984305758, "grad_norm": 0.46255171298980713, "learning_rate": 1.753855652768977e-05, "loss": 0.31461324691772463, "step": 9330 }, { "epoch": 0.13780310056976283, "grad_norm": 0.5105121731758118, "learning_rate": 1.7532898524352037e-05, "loss": 0.32081334590911864, "step": 9340 }, { "epoch": 0.13839200270894983, "grad_norm": 0.7357951998710632, "learning_rate": 1.7527234940475823e-05, "loss": 0.3309557199478149, "step": 9350 }, { "epoch": 0.13898090484813685, "grad_norm": 0.626787543296814, "learning_rate": 1.7521565780256843e-05, "loss": 0.32755241394042967, "step": 9360 }, { "epoch": 0.13956980698732388, "grad_norm": 0.45639967918395996, "learning_rate": 1.7515891047894937e-05, "loss": 0.3391565799713135, "step": 9370 }, { "epoch": 0.1401587091265109, "grad_norm": 0.3734910190105438, "learning_rate": 1.7510210747594076e-05, "loss": 0.3201342344284058, "step": 9380 }, { "epoch": 0.14074761126569793, "grad_norm": 0.4519542157649994, "learning_rate": 1.750452488356234e-05, "loss": 0.3397360324859619, "step": 9390 }, { "epoch": 0.14133651340488496, "grad_norm": 0.5369659662246704, "learning_rate": 1.749883346001196e-05, "loss": 0.34076268672943116, "step": 9400 }, { "epoch": 0.14192541554407195, "grad_norm": 0.4556417763233185, "learning_rate": 1.7493136481159253e-05, "loss": 0.3351482629776001, "step": 9410 }, { "epoch": 0.14251431768325898, "grad_norm": 0.5269331336021423, "learning_rate": 1.748743395122468e-05, "loss": 0.2994524955749512, "step": 9420 }, { "epoch": 0.143103219822446, "grad_norm": 0.5399971008300781, "learning_rate": 1.7481725874432797e-05, "loss": 0.32305335998535156, "step": 9430 }, { "epoch": 0.14369212196163303, "grad_norm": 0.5371549725532532, "learning_rate": 1.7476012255012272e-05, "loss": 0.323539137840271, "step": 9440 }, { "epoch": 0.14428102410082005, "grad_norm": 0.5847804546356201, "learning_rate": 1.747029309719588e-05, "loss": 0.3369602680206299, "step": 9450 }, { "epoch": 0.14486992624000708, "grad_norm": 0.4882728159427643, "learning_rate": 1.746456840522051e-05, "loss": 0.3561502456665039, "step": 9460 }, { "epoch": 0.14545882837919408, "grad_norm": 0.49506500363349915, "learning_rate": 1.745883818332713e-05, "loss": 0.29680519104003905, "step": 9470 }, { "epoch": 0.1460477305183811, "grad_norm": 0.5349467396736145, "learning_rate": 1.7453102435760815e-05, "loss": 0.33216631412506104, "step": 9480 }, { "epoch": 0.14663663265756813, "grad_norm": 0.5766996145248413, "learning_rate": 1.7447361166770743e-05, "loss": 0.35026843547821046, "step": 9490 }, { "epoch": 0.14722553479675515, "grad_norm": 0.4670804440975189, "learning_rate": 1.744161438061017e-05, "loss": 0.3063899278640747, "step": 9500 }, { "epoch": 0.14781443693594218, "grad_norm": 0.4678296148777008, "learning_rate": 1.743586208153644e-05, "loss": 0.3369497299194336, "step": 9510 }, { "epoch": 0.1484033390751292, "grad_norm": 0.4559159278869629, "learning_rate": 1.743010427381099e-05, "loss": 0.3108888387680054, "step": 9520 }, { "epoch": 0.1489922412143162, "grad_norm": 0.5224471092224121, "learning_rate": 1.7424340961699326e-05, "loss": 0.3169156789779663, "step": 9530 }, { "epoch": 0.14958114335350323, "grad_norm": 0.4845999777317047, "learning_rate": 1.7418572149471035e-05, "loss": 0.34582109451293946, "step": 9540 }, { "epoch": 0.15017004549269025, "grad_norm": 0.5872278809547424, "learning_rate": 1.7412797841399787e-05, "loss": 0.3665048360824585, "step": 9550 }, { "epoch": 0.15075894763187728, "grad_norm": 0.46173569560050964, "learning_rate": 1.740701804176332e-05, "loss": 0.3433325529098511, "step": 9560 }, { "epoch": 0.1513478497710643, "grad_norm": 0.5579553246498108, "learning_rate": 1.7401232754843433e-05, "loss": 0.33804056644439695, "step": 9570 }, { "epoch": 0.15193675191025133, "grad_norm": 0.4774749279022217, "learning_rate": 1.739544198492599e-05, "loss": 0.3423000335693359, "step": 9580 }, { "epoch": 0.15252565404943833, "grad_norm": 0.45394524931907654, "learning_rate": 1.7389645736300933e-05, "loss": 0.31943957805633544, "step": 9590 }, { "epoch": 0.15311455618862535, "grad_norm": 0.5898488163948059, "learning_rate": 1.7383844013262247e-05, "loss": 0.2940314054489136, "step": 9600 }, { "epoch": 0.15370345832781238, "grad_norm": 0.43330103158950806, "learning_rate": 1.7378036820107973e-05, "loss": 0.33776302337646485, "step": 9610 }, { "epoch": 0.1542923604669994, "grad_norm": 0.5427607297897339, "learning_rate": 1.7372224161140214e-05, "loss": 0.3093045473098755, "step": 9620 }, { "epoch": 0.15488126260618643, "grad_norm": 0.6369903087615967, "learning_rate": 1.7366406040665117e-05, "loss": 0.38350977897644045, "step": 9630 }, { "epoch": 0.15547016474537345, "grad_norm": 0.5941272974014282, "learning_rate": 1.7360582462992873e-05, "loss": 0.34077892303466795, "step": 9640 }, { "epoch": 0.15605906688456045, "grad_norm": 0.34357699751853943, "learning_rate": 1.7354753432437713e-05, "loss": 0.3327446937561035, "step": 9650 }, { "epoch": 0.15664796902374747, "grad_norm": 0.6025687456130981, "learning_rate": 1.734891895331792e-05, "loss": 0.33445355892181394, "step": 9660 }, { "epoch": 0.1572368711629345, "grad_norm": 0.3877888023853302, "learning_rate": 1.7343079029955804e-05, "loss": 0.34593877792358396, "step": 9670 }, { "epoch": 0.15782577330212152, "grad_norm": 0.4718611240386963, "learning_rate": 1.733723366667771e-05, "loss": 0.3194020986557007, "step": 9680 }, { "epoch": 0.15841467544130855, "grad_norm": 0.4512499272823334, "learning_rate": 1.7331382867814005e-05, "loss": 0.30712323188781737, "step": 9690 }, { "epoch": 0.15900357758049555, "grad_norm": 0.5423719882965088, "learning_rate": 1.73255266376991e-05, "loss": 0.32567930221557617, "step": 9700 }, { "epoch": 0.15959247971968257, "grad_norm": 0.4962805509567261, "learning_rate": 1.731966498067142e-05, "loss": 0.306526255607605, "step": 9710 }, { "epoch": 0.1601813818588696, "grad_norm": 0.6139407157897949, "learning_rate": 1.7313797901073407e-05, "loss": 0.36092400550842285, "step": 9720 }, { "epoch": 0.16077028399805662, "grad_norm": 0.47277799248695374, "learning_rate": 1.7307925403251525e-05, "loss": 0.32727727890014646, "step": 9730 }, { "epoch": 0.16135918613724365, "grad_norm": 0.38605040311813354, "learning_rate": 1.730204749155625e-05, "loss": 0.37814507484436033, "step": 9740 }, { "epoch": 0.16194808827643067, "grad_norm": 0.3757716417312622, "learning_rate": 1.7296164170342067e-05, "loss": 0.29011001586914065, "step": 9750 }, { "epoch": 0.16253699041561767, "grad_norm": 0.4504150152206421, "learning_rate": 1.7290275443967475e-05, "loss": 0.3184079885482788, "step": 9760 }, { "epoch": 0.1631258925548047, "grad_norm": 0.6415475606918335, "learning_rate": 1.728438131679498e-05, "loss": 0.32665295600891114, "step": 9770 }, { "epoch": 0.16371479469399172, "grad_norm": 0.43516111373901367, "learning_rate": 1.7278481793191063e-05, "loss": 0.3622438430786133, "step": 9780 }, { "epoch": 0.16430369683317875, "grad_norm": 0.35421666502952576, "learning_rate": 1.727257687752624e-05, "loss": 0.33107523918151854, "step": 9790 }, { "epoch": 0.16489259897236577, "grad_norm": 0.5788500905036926, "learning_rate": 1.7266666574174997e-05, "loss": 0.33078441619873045, "step": 9800 }, { "epoch": 0.1654815011115528, "grad_norm": 0.41172733902931213, "learning_rate": 1.726075088751582e-05, "loss": 0.3059119701385498, "step": 9810 }, { "epoch": 0.1660704032507398, "grad_norm": 0.5776916146278381, "learning_rate": 1.7254829821931176e-05, "loss": 0.31921911239624023, "step": 9820 }, { "epoch": 0.16665930538992682, "grad_norm": 0.46796953678131104, "learning_rate": 1.7248903381807526e-05, "loss": 0.334405779838562, "step": 9830 }, { "epoch": 0.16724820752911385, "grad_norm": 0.47263407707214355, "learning_rate": 1.7242971571535306e-05, "loss": 0.40093770027160647, "step": 9840 }, { "epoch": 0.16783710966830087, "grad_norm": 0.5774385929107666, "learning_rate": 1.7237034395508937e-05, "loss": 0.3211785554885864, "step": 9850 }, { "epoch": 0.1684260118074879, "grad_norm": 0.4690518081188202, "learning_rate": 1.7231091858126806e-05, "loss": 0.30664017200469973, "step": 9860 }, { "epoch": 0.16901491394667492, "grad_norm": 0.37115564942359924, "learning_rate": 1.7225143963791282e-05, "loss": 0.32364890575408933, "step": 9870 }, { "epoch": 0.16960381608586192, "grad_norm": 0.5883516073226929, "learning_rate": 1.7219190716908692e-05, "loss": 0.36805052757263185, "step": 9880 }, { "epoch": 0.17019271822504894, "grad_norm": 0.46171772480010986, "learning_rate": 1.7213232121889333e-05, "loss": 0.36825149059295653, "step": 9890 }, { "epoch": 0.17078162036423597, "grad_norm": 0.39534974098205566, "learning_rate": 1.7207268183147473e-05, "loss": 0.3024407625198364, "step": 9900 }, { "epoch": 0.171370522503423, "grad_norm": 0.4856685996055603, "learning_rate": 1.7201298905101317e-05, "loss": 0.3307697534561157, "step": 9910 }, { "epoch": 0.17195942464261002, "grad_norm": 0.41740620136260986, "learning_rate": 1.719532429217305e-05, "loss": 0.3118201732635498, "step": 9920 }, { "epoch": 0.17254832678179705, "grad_norm": 0.5734211206436157, "learning_rate": 1.7189344348788793e-05, "loss": 0.31644062995910643, "step": 9930 }, { "epoch": 0.17313722892098404, "grad_norm": 0.559532105922699, "learning_rate": 1.7183359079378622e-05, "loss": 0.33839993476867675, "step": 9940 }, { "epoch": 0.17372613106017107, "grad_norm": 0.4984675943851471, "learning_rate": 1.7177368488376553e-05, "loss": 0.29887216091156005, "step": 9950 }, { "epoch": 0.1743150331993581, "grad_norm": 0.6226099133491516, "learning_rate": 1.7171372580220557e-05, "loss": 0.3188284635543823, "step": 9960 }, { "epoch": 0.17490393533854512, "grad_norm": 0.4868571162223816, "learning_rate": 1.7165371359352533e-05, "loss": 0.30336196422576905, "step": 9970 }, { "epoch": 0.17549283747773214, "grad_norm": 0.5465160608291626, "learning_rate": 1.7159364830218312e-05, "loss": 0.29795525074005125, "step": 9980 }, { "epoch": 0.17608173961691917, "grad_norm": 0.5348778963088989, "learning_rate": 1.7153352997267674e-05, "loss": 0.3269251108169556, "step": 9990 }, { "epoch": 0.17667064175610617, "grad_norm": 0.44998589158058167, "learning_rate": 1.7147335864954315e-05, "loss": 0.39244470596313474, "step": 10000 }, { "epoch": 0.0005889021391870206, "grad_norm": 0.46115952730178833, "learning_rate": 1.7141313437735853e-05, "loss": 0.30862295627593994, "step": 10010 }, { "epoch": 0.0011778042783740412, "grad_norm": 0.5199177265167236, "learning_rate": 1.713528572007385e-05, "loss": 0.31673879623413087, "step": 10020 }, { "epoch": 0.0017667064175610617, "grad_norm": 0.5028446316719055, "learning_rate": 1.7129252716433756e-05, "loss": 0.284792160987854, "step": 10030 }, { "epoch": 0.0023556085567480825, "grad_norm": 0.6014943718910217, "learning_rate": 1.712321443128497e-05, "loss": 0.3121055603027344, "step": 10040 }, { "epoch": 0.002944510695935103, "grad_norm": 0.7373807430267334, "learning_rate": 1.7117170869100775e-05, "loss": 0.30137109756469727, "step": 10050 }, { "epoch": 0.0035334128351221235, "grad_norm": 0.6074202656745911, "learning_rate": 1.7111122034358384e-05, "loss": 0.24542243480682374, "step": 10060 }, { "epoch": 0.004122314974309144, "grad_norm": 0.6550632119178772, "learning_rate": 1.7105067931538907e-05, "loss": 0.2821102857589722, "step": 10070 }, { "epoch": 0.004711217113496165, "grad_norm": 0.5452698469161987, "learning_rate": 1.7099008565127357e-05, "loss": 0.27634763717651367, "step": 10080 }, { "epoch": 0.005300119252683185, "grad_norm": 0.5902866721153259, "learning_rate": 1.7092943939612647e-05, "loss": 0.3062936782836914, "step": 10090 }, { "epoch": 0.005889021391870206, "grad_norm": 0.4794277548789978, "learning_rate": 1.7086874059487588e-05, "loss": 0.3070604562759399, "step": 10100 }, { "epoch": 0.006477923531057227, "grad_norm": 0.6252293586730957, "learning_rate": 1.708079892924887e-05, "loss": 0.29135522842407224, "step": 10110 }, { "epoch": 0.007066825670244247, "grad_norm": 0.4769746661186218, "learning_rate": 1.7074718553397104e-05, "loss": 0.3046879768371582, "step": 10120 }, { "epoch": 0.007655727809431268, "grad_norm": 0.6325712203979492, "learning_rate": 1.7068632936436755e-05, "loss": 0.3210947275161743, "step": 10130 }, { "epoch": 0.008244629948618288, "grad_norm": 0.5207534432411194, "learning_rate": 1.7062542082876187e-05, "loss": 0.3416034460067749, "step": 10140 }, { "epoch": 0.00883353208780531, "grad_norm": 0.5305103659629822, "learning_rate": 1.7056445997227638e-05, "loss": 0.3088728904724121, "step": 10150 }, { "epoch": 0.00942243422699233, "grad_norm": 0.7219800353050232, "learning_rate": 1.7050344684007222e-05, "loss": 0.3214219808578491, "step": 10160 }, { "epoch": 0.01001133636617935, "grad_norm": 0.5919177532196045, "learning_rate": 1.7044238147734932e-05, "loss": 0.2937323570251465, "step": 10170 }, { "epoch": 0.01060023850536637, "grad_norm": 0.610954225063324, "learning_rate": 1.703812639293462e-05, "loss": 0.3083319902420044, "step": 10180 }, { "epoch": 0.011189140644553392, "grad_norm": 0.5556350946426392, "learning_rate": 1.7032009424134015e-05, "loss": 0.33402838706970217, "step": 10190 }, { "epoch": 0.011778042783740412, "grad_norm": 0.5359517335891724, "learning_rate": 1.70258872458647e-05, "loss": 0.2768877029418945, "step": 10200 }, { "epoch": 0.012366944922927432, "grad_norm": 0.5280352234840393, "learning_rate": 1.7019759862662122e-05, "loss": 0.3029083967208862, "step": 10210 }, { "epoch": 0.012955847062114454, "grad_norm": 0.7022429704666138, "learning_rate": 1.7013627279065582e-05, "loss": 0.31228060722351075, "step": 10220 }, { "epoch": 0.013544749201301474, "grad_norm": 0.6143155694007874, "learning_rate": 1.7007489499618233e-05, "loss": 0.2940253019332886, "step": 10230 }, { "epoch": 0.014133651340488494, "grad_norm": 0.5733249187469482, "learning_rate": 1.7001346528867083e-05, "loss": 0.300798225402832, "step": 10240 }, { "epoch": 0.014722553479675516, "grad_norm": 0.6247124075889587, "learning_rate": 1.6995198371362972e-05, "loss": 0.2660230159759521, "step": 10250 }, { "epoch": 0.015311455618862536, "grad_norm": 0.5972670316696167, "learning_rate": 1.6989045031660603e-05, "loss": 0.2991211414337158, "step": 10260 }, { "epoch": 0.015900357758049558, "grad_norm": 0.5129410624504089, "learning_rate": 1.69828865143185e-05, "loss": 0.28291056156158445, "step": 10270 }, { "epoch": 0.016489259897236576, "grad_norm": 0.7213442325592041, "learning_rate": 1.6976722823899027e-05, "loss": 0.26688284873962403, "step": 10280 }, { "epoch": 0.017078162036423598, "grad_norm": 0.8889928460121155, "learning_rate": 1.6970553964968385e-05, "loss": 0.28685085773468016, "step": 10290 }, { "epoch": 0.01766706417561062, "grad_norm": 0.7347835302352905, "learning_rate": 1.6964379942096606e-05, "loss": 0.27613487243652346, "step": 10300 }, { "epoch": 0.018255966314797638, "grad_norm": 0.4574574828147888, "learning_rate": 1.6958200759857536e-05, "loss": 0.3017237424850464, "step": 10310 }, { "epoch": 0.01884486845398466, "grad_norm": 0.7133186459541321, "learning_rate": 1.695201642282886e-05, "loss": 0.24433934688568115, "step": 10320 }, { "epoch": 0.01943377059317168, "grad_norm": 0.5874927639961243, "learning_rate": 1.694582693559206e-05, "loss": 0.2585667610168457, "step": 10330 }, { "epoch": 0.0200226727323587, "grad_norm": 0.4877403676509857, "learning_rate": 1.6939632302732456e-05, "loss": 0.2625157117843628, "step": 10340 }, { "epoch": 0.02061157487154572, "grad_norm": 0.7177917957305908, "learning_rate": 1.6933432528839163e-05, "loss": 0.300957179069519, "step": 10350 }, { "epoch": 0.02120047701073274, "grad_norm": 0.44749322533607483, "learning_rate": 1.6927227618505115e-05, "loss": 0.2773359060287476, "step": 10360 }, { "epoch": 0.02178937914991976, "grad_norm": 0.6647253036499023, "learning_rate": 1.6921017576327048e-05, "loss": 0.26094100475311277, "step": 10370 }, { "epoch": 0.022378281289106784, "grad_norm": 0.5644856095314026, "learning_rate": 1.691480240690549e-05, "loss": 0.3003634691238403, "step": 10380 }, { "epoch": 0.022967183428293802, "grad_norm": 0.6572015285491943, "learning_rate": 1.6908582114844793e-05, "loss": 0.247390079498291, "step": 10390 }, { "epoch": 0.023556085567480824, "grad_norm": 0.6398977637290955, "learning_rate": 1.6902356704753075e-05, "loss": 0.26758203506469724, "step": 10400 }, { "epoch": 0.024144987706667845, "grad_norm": 0.7201460003852844, "learning_rate": 1.6896126181242256e-05, "loss": 0.26046898365020754, "step": 10410 }, { "epoch": 0.024733889845854864, "grad_norm": 0.8687049746513367, "learning_rate": 1.6889890548928055e-05, "loss": 0.29185583591461184, "step": 10420 }, { "epoch": 0.025322791985041886, "grad_norm": 0.5508840680122375, "learning_rate": 1.6883649812429964e-05, "loss": 0.2602362155914307, "step": 10430 }, { "epoch": 0.025911694124228907, "grad_norm": 0.48455679416656494, "learning_rate": 1.6877403976371257e-05, "loss": 0.26238512992858887, "step": 10440 }, { "epoch": 0.026500596263415926, "grad_norm": 0.6405127644538879, "learning_rate": 1.687115304537899e-05, "loss": 0.3135169744491577, "step": 10450 }, { "epoch": 0.027089498402602948, "grad_norm": 0.4634402394294739, "learning_rate": 1.686489702408399e-05, "loss": 0.2801950216293335, "step": 10460 }, { "epoch": 0.02767840054178997, "grad_norm": 0.838387131690979, "learning_rate": 1.685863591712086e-05, "loss": 0.3122610569000244, "step": 10470 }, { "epoch": 0.028267302680976988, "grad_norm": 0.690018355846405, "learning_rate": 1.685236972912796e-05, "loss": 0.26887073516845705, "step": 10480 }, { "epoch": 0.02885620482016401, "grad_norm": 0.6898316144943237, "learning_rate": 1.6846098464747428e-05, "loss": 0.3082263469696045, "step": 10490 }, { "epoch": 0.02944510695935103, "grad_norm": 0.7773162722587585, "learning_rate": 1.6839822128625153e-05, "loss": 0.30830371379852295, "step": 10500 }, { "epoch": 0.03003400909853805, "grad_norm": 0.6032030582427979, "learning_rate": 1.6833540725410783e-05, "loss": 0.26619548797607423, "step": 10510 }, { "epoch": 0.03062291123772507, "grad_norm": 0.8550388813018799, "learning_rate": 1.6827254259757722e-05, "loss": 0.2872445106506348, "step": 10520 }, { "epoch": 0.031211813376912093, "grad_norm": 0.7546632885932922, "learning_rate": 1.6820962736323125e-05, "loss": 0.27695188522338865, "step": 10530 }, { "epoch": 0.031800715516099115, "grad_norm": 0.6351567506790161, "learning_rate": 1.681466615976789e-05, "loss": 0.25202672481536864, "step": 10540 }, { "epoch": 0.03238961765528613, "grad_norm": 0.6071174740791321, "learning_rate": 1.6808364534756668e-05, "loss": 0.2630735397338867, "step": 10550 }, { "epoch": 0.03297851979447315, "grad_norm": 0.6072619557380676, "learning_rate": 1.6802057865957827e-05, "loss": 0.279038143157959, "step": 10560 }, { "epoch": 0.03356742193366018, "grad_norm": 0.5999947786331177, "learning_rate": 1.6795746158043505e-05, "loss": 0.2567354917526245, "step": 10570 }, { "epoch": 0.034156324072847195, "grad_norm": 0.5543020963668823, "learning_rate": 1.6789429415689542e-05, "loss": 0.2742922782897949, "step": 10580 }, { "epoch": 0.034745226212034214, "grad_norm": 0.5775655508041382, "learning_rate": 1.6783107643575526e-05, "loss": 0.2827472686767578, "step": 10590 }, { "epoch": 0.03533412835122124, "grad_norm": 0.7712039351463318, "learning_rate": 1.677678084638477e-05, "loss": 0.29876835346221925, "step": 10600 }, { "epoch": 0.03592303049040826, "grad_norm": 0.6124355792999268, "learning_rate": 1.67704490288043e-05, "loss": 0.2873201608657837, "step": 10610 }, { "epoch": 0.036511932629595276, "grad_norm": 0.590207576751709, "learning_rate": 1.6764112195524868e-05, "loss": 0.3054657459259033, "step": 10620 }, { "epoch": 0.0371008347687823, "grad_norm": 0.5760242342948914, "learning_rate": 1.6757770351240943e-05, "loss": 0.27776589393615725, "step": 10630 }, { "epoch": 0.03768973690796932, "grad_norm": 0.6507149338722229, "learning_rate": 1.6751423500650702e-05, "loss": 0.2846635580062866, "step": 10640 }, { "epoch": 0.03827863904715634, "grad_norm": 0.8435494899749756, "learning_rate": 1.674507164845603e-05, "loss": 0.32600107192993166, "step": 10650 }, { "epoch": 0.03886754118634336, "grad_norm": 0.8265935778617859, "learning_rate": 1.673871479936253e-05, "loss": 0.26729545593261717, "step": 10660 }, { "epoch": 0.03945644332553038, "grad_norm": 0.75574791431427, "learning_rate": 1.673235295807949e-05, "loss": 0.2944334506988525, "step": 10670 }, { "epoch": 0.0400453454647174, "grad_norm": 0.708920419216156, "learning_rate": 1.6725986129319905e-05, "loss": 0.31214470863342286, "step": 10680 }, { "epoch": 0.04063424760390442, "grad_norm": 0.7520795464515686, "learning_rate": 1.671961431780046e-05, "loss": 0.27503433227539065, "step": 10690 }, { "epoch": 0.04122314974309144, "grad_norm": 0.6811811923980713, "learning_rate": 1.6713237528241536e-05, "loss": 0.2651083469390869, "step": 10700 }, { "epoch": 0.04181205188227846, "grad_norm": 0.6107529997825623, "learning_rate": 1.6706855765367202e-05, "loss": 0.26419174671173096, "step": 10710 }, { "epoch": 0.04240095402146548, "grad_norm": 0.7589848041534424, "learning_rate": 1.670046903390521e-05, "loss": 0.2819946765899658, "step": 10720 }, { "epoch": 0.042989856160652505, "grad_norm": 0.8486412167549133, "learning_rate": 1.6694077338586988e-05, "loss": 0.2627174615859985, "step": 10730 }, { "epoch": 0.04357875829983952, "grad_norm": 0.6179655194282532, "learning_rate": 1.6687680684147652e-05, "loss": 0.2855645418167114, "step": 10740 }, { "epoch": 0.04416766043902654, "grad_norm": 0.6270648837089539, "learning_rate": 1.668127907532598e-05, "loss": 0.2923930883407593, "step": 10750 }, { "epoch": 0.04475656257821357, "grad_norm": 0.7796370387077332, "learning_rate": 1.667487251686443e-05, "loss": 0.29988322257995603, "step": 10760 }, { "epoch": 0.045345464717400585, "grad_norm": 0.7513877153396606, "learning_rate": 1.666846101350912e-05, "loss": 0.2739178895950317, "step": 10770 }, { "epoch": 0.045934366856587604, "grad_norm": 0.6559301018714905, "learning_rate": 1.6662044570009834e-05, "loss": 0.2452981948852539, "step": 10780 }, { "epoch": 0.04652326899577463, "grad_norm": 0.8454421162605286, "learning_rate": 1.6655623191120018e-05, "loss": 0.2962732553482056, "step": 10790 }, { "epoch": 0.04711217113496165, "grad_norm": 0.6726862192153931, "learning_rate": 1.6649196881596773e-05, "loss": 0.2784935474395752, "step": 10800 }, { "epoch": 0.047701073274148666, "grad_norm": 0.7130102515220642, "learning_rate": 1.6642765646200846e-05, "loss": 0.2939387083053589, "step": 10810 }, { "epoch": 0.04828997541333569, "grad_norm": 0.7259536981582642, "learning_rate": 1.6636329489696646e-05, "loss": 0.32226707935333254, "step": 10820 }, { "epoch": 0.04887887755252271, "grad_norm": 0.8115344643592834, "learning_rate": 1.6629888416852217e-05, "loss": 0.2643917083740234, "step": 10830 }, { "epoch": 0.04946777969170973, "grad_norm": 0.6777119636535645, "learning_rate": 1.662344243243925e-05, "loss": 0.2711274862289429, "step": 10840 }, { "epoch": 0.05005668183089675, "grad_norm": 0.6506727337837219, "learning_rate": 1.6616991541233073e-05, "loss": 0.3081958293914795, "step": 10850 }, { "epoch": 0.05064558397008377, "grad_norm": 0.7310405373573303, "learning_rate": 1.6610535748012643e-05, "loss": 0.3000586986541748, "step": 10860 }, { "epoch": 0.05123448610927079, "grad_norm": 0.7205409407615662, "learning_rate": 1.660407505756057e-05, "loss": 0.28290371894836425, "step": 10870 }, { "epoch": 0.051823388248457815, "grad_norm": 0.7629085779190063, "learning_rate": 1.659760947466306e-05, "loss": 0.24788575172424315, "step": 10880 }, { "epoch": 0.05241229038764483, "grad_norm": 0.6187683939933777, "learning_rate": 1.6591139004109974e-05, "loss": 0.25477232933044436, "step": 10890 }, { "epoch": 0.05300119252683185, "grad_norm": 0.7072340250015259, "learning_rate": 1.658466365069477e-05, "loss": 0.2450669527053833, "step": 10900 }, { "epoch": 0.05359009466601888, "grad_norm": 0.7151548862457275, "learning_rate": 1.6578183419214542e-05, "loss": 0.25872464179992677, "step": 10910 }, { "epoch": 0.054178996805205895, "grad_norm": 0.6893681287765503, "learning_rate": 1.657169831446998e-05, "loss": 0.2585454940795898, "step": 10920 }, { "epoch": 0.05476789894439291, "grad_norm": 1.1241185665130615, "learning_rate": 1.6565208341265404e-05, "loss": 0.24937915802001953, "step": 10930 }, { "epoch": 0.05535680108357994, "grad_norm": 0.8676333427429199, "learning_rate": 1.6558713504408723e-05, "loss": 0.29741320610046384, "step": 10940 }, { "epoch": 0.05594570322276696, "grad_norm": 0.7981345057487488, "learning_rate": 1.655221380871146e-05, "loss": 0.2795196294784546, "step": 10950 }, { "epoch": 0.056534605361953975, "grad_norm": 0.9824395179748535, "learning_rate": 1.6545709258988734e-05, "loss": 0.2791579723358154, "step": 10960 }, { "epoch": 0.057123507501141, "grad_norm": 0.8808866739273071, "learning_rate": 1.6539199860059257e-05, "loss": 0.2661163806915283, "step": 10970 }, { "epoch": 0.05771240964032802, "grad_norm": 0.6264880299568176, "learning_rate": 1.6532685616745334e-05, "loss": 0.27133903503417967, "step": 10980 }, { "epoch": 0.05830131177951504, "grad_norm": 0.843337893486023, "learning_rate": 1.6526166533872867e-05, "loss": 0.288726282119751, "step": 10990 }, { "epoch": 0.05889021391870206, "grad_norm": 0.7111064791679382, "learning_rate": 1.6519642616271335e-05, "loss": 0.24761765003204345, "step": 11000 }, { "epoch": 0.05947911605788908, "grad_norm": 0.760053277015686, "learning_rate": 1.65131138687738e-05, "loss": 0.291316819190979, "step": 11010 }, { "epoch": 0.0600680181970761, "grad_norm": 0.6777175068855286, "learning_rate": 1.650658029621691e-05, "loss": 0.3140115737915039, "step": 11020 }, { "epoch": 0.060656920336263125, "grad_norm": 1.1089611053466797, "learning_rate": 1.650004190344087e-05, "loss": 0.2917371988296509, "step": 11030 }, { "epoch": 0.06124582247545014, "grad_norm": 0.5550308227539062, "learning_rate": 1.649349869528947e-05, "loss": 0.28005859851837156, "step": 11040 }, { "epoch": 0.06183472461463716, "grad_norm": 0.6270084977149963, "learning_rate": 1.6486950676610067e-05, "loss": 0.2838952302932739, "step": 11050 }, { "epoch": 0.062423626753824187, "grad_norm": 0.7439520359039307, "learning_rate": 1.6480397852253575e-05, "loss": 0.2814916133880615, "step": 11060 }, { "epoch": 0.0630125288930112, "grad_norm": 0.6540569067001343, "learning_rate": 1.6473840227074477e-05, "loss": 0.2761623620986938, "step": 11070 }, { "epoch": 0.06360143103219823, "grad_norm": 0.6433505415916443, "learning_rate": 1.64672778059308e-05, "loss": 0.30278191566467283, "step": 11080 }, { "epoch": 0.06419033317138524, "grad_norm": 0.7987252473831177, "learning_rate": 1.646071059368414e-05, "loss": 0.2999276161193848, "step": 11090 }, { "epoch": 0.06477923531057227, "grad_norm": 0.5620887279510498, "learning_rate": 1.6454138595199626e-05, "loss": 0.3434076547622681, "step": 11100 }, { "epoch": 0.06536813744975929, "grad_norm": 0.6241165399551392, "learning_rate": 1.6447561815345938e-05, "loss": 0.28463082313537597, "step": 11110 }, { "epoch": 0.0659570395889463, "grad_norm": 0.7484923005104065, "learning_rate": 1.6440980258995312e-05, "loss": 0.2897254228591919, "step": 11120 }, { "epoch": 0.06654594172813333, "grad_norm": 0.7894267439842224, "learning_rate": 1.6434393931023505e-05, "loss": 0.2823662757873535, "step": 11130 }, { "epoch": 0.06713484386732035, "grad_norm": 0.689630389213562, "learning_rate": 1.6427802836309813e-05, "loss": 0.28103973865509035, "step": 11140 }, { "epoch": 0.06772374600650737, "grad_norm": 0.7255921959877014, "learning_rate": 1.6421206979737068e-05, "loss": 0.30650007724761963, "step": 11150 }, { "epoch": 0.06831264814569439, "grad_norm": 0.6353716254234314, "learning_rate": 1.641460636619162e-05, "loss": 0.3172675848007202, "step": 11160 }, { "epoch": 0.06890155028488142, "grad_norm": 0.6860131621360779, "learning_rate": 1.6408001000563364e-05, "loss": 0.3225912809371948, "step": 11170 }, { "epoch": 0.06949045242406843, "grad_norm": 0.6357450485229492, "learning_rate": 1.6401390887745692e-05, "loss": 0.32307448387146, "step": 11180 }, { "epoch": 0.07007935456325545, "grad_norm": 0.7554405927658081, "learning_rate": 1.639477603263552e-05, "loss": 0.28299312591552733, "step": 11190 }, { "epoch": 0.07066825670244248, "grad_norm": 0.6660791635513306, "learning_rate": 1.6388156440133288e-05, "loss": 0.2775435447692871, "step": 11200 }, { "epoch": 0.07125715884162949, "grad_norm": 0.5638531446456909, "learning_rate": 1.638153211514293e-05, "loss": 0.3467556953430176, "step": 11210 }, { "epoch": 0.07184606098081651, "grad_norm": 0.6460939645767212, "learning_rate": 1.63749030625719e-05, "loss": 0.3279921770095825, "step": 11220 }, { "epoch": 0.07243496312000354, "grad_norm": 0.7482138872146606, "learning_rate": 1.636826928733114e-05, "loss": 0.3037510633468628, "step": 11230 }, { "epoch": 0.07302386525919055, "grad_norm": 0.6293128728866577, "learning_rate": 1.63616307943351e-05, "loss": 0.2807432174682617, "step": 11240 }, { "epoch": 0.07361276739837758, "grad_norm": 0.5591038465499878, "learning_rate": 1.6354987588501723e-05, "loss": 0.27299673557281495, "step": 11250 }, { "epoch": 0.0742016695375646, "grad_norm": 0.6497546434402466, "learning_rate": 1.6348339674752443e-05, "loss": 0.2968681812286377, "step": 11260 }, { "epoch": 0.07479057167675161, "grad_norm": 0.6669679880142212, "learning_rate": 1.6341687058012186e-05, "loss": 0.32380967140197753, "step": 11270 }, { "epoch": 0.07537947381593864, "grad_norm": 0.7237711548805237, "learning_rate": 1.6335029743209356e-05, "loss": 0.26773786544799805, "step": 11280 }, { "epoch": 0.07596837595512566, "grad_norm": 0.5363181233406067, "learning_rate": 1.632836773527584e-05, "loss": 0.28376939296722414, "step": 11290 }, { "epoch": 0.07655727809431268, "grad_norm": 0.6236751079559326, "learning_rate": 1.6321701039147e-05, "loss": 0.2998849630355835, "step": 11300 }, { "epoch": 0.0004711217113496165, "grad_norm": 0.5796450972557068, "learning_rate": 1.6315029659761672e-05, "loss": 0.2876850962638855, "step": 11310 }, { "epoch": 0.001060023850536637, "grad_norm": 0.620351254940033, "learning_rate": 1.6308353602062165e-05, "loss": 0.27274460792541505, "step": 11320 }, { "epoch": 0.0016489259897236577, "grad_norm": 0.7074437737464905, "learning_rate": 1.6301672870994246e-05, "loss": 0.2942031383514404, "step": 11330 }, { "epoch": 0.0022378281289106784, "grad_norm": 0.6389545798301697, "learning_rate": 1.6294987471507155e-05, "loss": 0.27881178855895994, "step": 11340 }, { "epoch": 0.002826730268097699, "grad_norm": 0.6629053354263306, "learning_rate": 1.628829740855358e-05, "loss": 0.28976993560791015, "step": 11350 }, { "epoch": 0.0034156324072847194, "grad_norm": 0.4685785174369812, "learning_rate": 1.6281602687089666e-05, "loss": 0.2501103162765503, "step": 11360 }, { "epoch": 0.00400453454647174, "grad_norm": 0.7621864080429077, "learning_rate": 1.627490331207502e-05, "loss": 0.2541220664978027, "step": 11370 }, { "epoch": 0.004593436685658761, "grad_norm": 0.6377124190330505, "learning_rate": 1.6268199288472678e-05, "loss": 0.23815083503723145, "step": 11380 }, { "epoch": 0.005182338824845781, "grad_norm": 0.6526587605476379, "learning_rate": 1.6261490621249135e-05, "loss": 0.27955379486083987, "step": 11390 }, { "epoch": 0.005771240964032802, "grad_norm": 0.671714723110199, "learning_rate": 1.6254777315374323e-05, "loss": 0.2845767974853516, "step": 11400 }, { "epoch": 0.006360143103219823, "grad_norm": 0.7389091849327087, "learning_rate": 1.62480593758216e-05, "loss": 0.282643985748291, "step": 11410 }, { "epoch": 0.006949045242406843, "grad_norm": 0.628589928150177, "learning_rate": 1.624133680756777e-05, "loss": 0.2944262742996216, "step": 11420 }, { "epoch": 0.007537947381593864, "grad_norm": 0.6892663836479187, "learning_rate": 1.6234609615593063e-05, "loss": 0.2841221570968628, "step": 11430 }, { "epoch": 0.008126849520780884, "grad_norm": 0.7156624794006348, "learning_rate": 1.6227877804881126e-05, "loss": 0.32349486351013185, "step": 11440 }, { "epoch": 0.008715751659967904, "grad_norm": 0.6145578622817993, "learning_rate": 1.622114138041904e-05, "loss": 0.28476226329803467, "step": 11450 }, { "epoch": 0.009304653799154926, "grad_norm": 0.6926509737968445, "learning_rate": 1.6214400347197292e-05, "loss": 0.269835901260376, "step": 11460 }, { "epoch": 0.009893555938341946, "grad_norm": 0.684954047203064, "learning_rate": 1.6207654710209796e-05, "loss": 0.28847203254699705, "step": 11470 }, { "epoch": 0.010482458077528966, "grad_norm": 0.5501247644424438, "learning_rate": 1.6200904474453864e-05, "loss": 0.27903892993927004, "step": 11480 }, { "epoch": 0.011071360216715988, "grad_norm": 0.8063901662826538, "learning_rate": 1.619414964493022e-05, "loss": 0.30527563095092775, "step": 11490 }, { "epoch": 0.011660262355903008, "grad_norm": 0.7595698237419128, "learning_rate": 1.6187390226642997e-05, "loss": 0.2774637460708618, "step": 11500 }, { "epoch": 0.012249164495090028, "grad_norm": 0.6345116496086121, "learning_rate": 1.6180626224599717e-05, "loss": 0.25841965675354006, "step": 11510 }, { "epoch": 0.01283806663427705, "grad_norm": 0.8438473343849182, "learning_rate": 1.6173857643811305e-05, "loss": 0.2947725772857666, "step": 11520 }, { "epoch": 0.01342696877346407, "grad_norm": 1.092965006828308, "learning_rate": 1.6167084489292074e-05, "loss": 0.2798546552658081, "step": 11530 }, { "epoch": 0.01401587091265109, "grad_norm": 0.7148837447166443, "learning_rate": 1.6160306766059733e-05, "loss": 0.28312077522277834, "step": 11540 }, { "epoch": 0.014604773051838112, "grad_norm": 0.9082318544387817, "learning_rate": 1.6153524479135363e-05, "loss": 0.24763193130493164, "step": 11550 }, { "epoch": 0.015193675191025132, "grad_norm": 0.8716150522232056, "learning_rate": 1.614673763354344e-05, "loss": 0.2693489074707031, "step": 11560 }, { "epoch": 0.015782577330212152, "grad_norm": 1.0516949892044067, "learning_rate": 1.61399462343118e-05, "loss": 0.22180335521697997, "step": 11570 }, { "epoch": 0.016371479469399174, "grad_norm": 0.8204823136329651, "learning_rate": 1.613315028647167e-05, "loss": 0.28469181060791016, "step": 11580 }, { "epoch": 0.016960381608586192, "grad_norm": 0.8387027978897095, "learning_rate": 1.612634979505764e-05, "loss": 0.24824028015136718, "step": 11590 }, { "epoch": 0.017549283747773214, "grad_norm": 0.7905215620994568, "learning_rate": 1.611954476510766e-05, "loss": 0.24791302680969238, "step": 11600 }, { "epoch": 0.018138185886960236, "grad_norm": 0.8026900291442871, "learning_rate": 1.6112735201663046e-05, "loss": 0.29231479167938235, "step": 11610 }, { "epoch": 0.018727088026147254, "grad_norm": 0.851542592048645, "learning_rate": 1.6105921109768482e-05, "loss": 0.20736305713653563, "step": 11620 }, { "epoch": 0.019315990165334276, "grad_norm": 0.9491105675697327, "learning_rate": 1.6099102494471992e-05, "loss": 0.23800692558288575, "step": 11630 }, { "epoch": 0.0005889021391870206, "grad_norm": 0.7384411692619324, "learning_rate": 1.6092279360824957e-05, "loss": 0.2663137435913086, "step": 11640 }, { "epoch": 0.0011778042783740412, "grad_norm": 0.8486633896827698, "learning_rate": 1.6085451713882112e-05, "loss": 0.2623342275619507, "step": 11650 }, { "epoch": 0.0017667064175610617, "grad_norm": 0.7810729742050171, "learning_rate": 1.6078619558701523e-05, "loss": 0.24170422554016113, "step": 11660 }, { "epoch": 0.0023556085567480825, "grad_norm": 0.8094569444656372, "learning_rate": 1.6071782900344606e-05, "loss": 0.2610111474990845, "step": 11670 }, { "epoch": 0.002944510695935103, "grad_norm": 0.993614912033081, "learning_rate": 1.6064941743876106e-05, "loss": 0.2525993585586548, "step": 11680 }, { "epoch": 0.0035334128351221235, "grad_norm": 0.8261240124702454, "learning_rate": 1.605809609436411e-05, "loss": 0.2036445379257202, "step": 11690 }, { "epoch": 0.004122314974309144, "grad_norm": 0.8767113089561462, "learning_rate": 1.6051245956880018e-05, "loss": 0.22631304264068602, "step": 11700 }, { "epoch": 0.004711217113496165, "grad_norm": 0.8173907399177551, "learning_rate": 1.604439133649857e-05, "loss": 0.23083627223968506, "step": 11710 }, { "epoch": 0.005300119252683185, "grad_norm": 0.875468373298645, "learning_rate": 1.6037532238297823e-05, "loss": 0.25173628330230713, "step": 11720 }, { "epoch": 0.005889021391870206, "grad_norm": 0.740479052066803, "learning_rate": 1.603066866735914e-05, "loss": 0.2605211973190308, "step": 11730 }, { "epoch": 0.006477923531057227, "grad_norm": 0.9701038599014282, "learning_rate": 1.6023800628767218e-05, "loss": 0.23768560886383056, "step": 11740 }, { "epoch": 0.007066825670244247, "grad_norm": 0.7320893406867981, "learning_rate": 1.6016928127610046e-05, "loss": 0.2563259363174438, "step": 11750 }, { "epoch": 0.007655727809431268, "grad_norm": 0.9099878668785095, "learning_rate": 1.6010051168978926e-05, "loss": 0.2617072582244873, "step": 11760 }, { "epoch": 0.008244629948618288, "grad_norm": 0.806411623954773, "learning_rate": 1.6003169757968464e-05, "loss": 0.28125367164611814, "step": 11770 }, { "epoch": 0.00883353208780531, "grad_norm": 0.768580973148346, "learning_rate": 1.5996283899676558e-05, "loss": 0.2596011161804199, "step": 11780 }, { "epoch": 0.00942243422699233, "grad_norm": 0.9919658303260803, "learning_rate": 1.598939359920441e-05, "loss": 0.26127941608428956, "step": 11790 }, { "epoch": 0.01001133636617935, "grad_norm": 0.8792932033538818, "learning_rate": 1.5982498861656503e-05, "loss": 0.2395712375640869, "step": 11800 }, { "epoch": 0.01060023850536637, "grad_norm": 0.8684144020080566, "learning_rate": 1.597559969214061e-05, "loss": 0.25873074531555174, "step": 11810 }, { "epoch": 0.011189140644553392, "grad_norm": 0.7897999882698059, "learning_rate": 1.5968696095767796e-05, "loss": 0.2745176315307617, "step": 11820 }, { "epoch": 0.011778042783740412, "grad_norm": 0.8397480845451355, "learning_rate": 1.5961788077652393e-05, "loss": 0.22602765560150145, "step": 11830 }, { "epoch": 0.012366944922927432, "grad_norm": 0.8243021368980408, "learning_rate": 1.595487564291202e-05, "loss": 0.24685113430023192, "step": 11840 }, { "epoch": 0.012955847062114454, "grad_norm": 0.8467777967453003, "learning_rate": 1.5947958796667554e-05, "loss": 0.249222993850708, "step": 11850 }, { "epoch": 0.013544749201301474, "grad_norm": 0.9049258828163147, "learning_rate": 1.5941037544043153e-05, "loss": 0.23682787418365478, "step": 11860 }, { "epoch": 0.014133651340488494, "grad_norm": 0.7718921303749084, "learning_rate": 1.5934111890166234e-05, "loss": 0.2469231367111206, "step": 11870 }, { "epoch": 0.014722553479675516, "grad_norm": 0.8714404106140137, "learning_rate": 1.5927181840167478e-05, "loss": 0.21399922370910646, "step": 11880 }, { "epoch": 0.015311455618862536, "grad_norm": 0.8980563282966614, "learning_rate": 1.5920247399180816e-05, "loss": 0.2340184450149536, "step": 11890 }, { "epoch": 0.015900357758049558, "grad_norm": 0.8788493871688843, "learning_rate": 1.5913308572343433e-05, "loss": 0.2304084539413452, "step": 11900 }, { "epoch": 0.016489259897236576, "grad_norm": 0.9436264038085938, "learning_rate": 1.5906365364795774e-05, "loss": 0.21188149452209473, "step": 11910 }, { "epoch": 0.017078162036423598, "grad_norm": 0.9586700201034546, "learning_rate": 1.5899417781681515e-05, "loss": 0.2231370210647583, "step": 11920 }, { "epoch": 0.01766706417561062, "grad_norm": 0.9777308702468872, "learning_rate": 1.589246582814758e-05, "loss": 0.20736236572265626, "step": 11930 }, { "epoch": 0.018255966314797638, "grad_norm": 0.7282084822654724, "learning_rate": 1.5885509509344137e-05, "loss": 0.24200258255004883, "step": 11940 }, { "epoch": 0.01884486845398466, "grad_norm": 0.9893869757652283, "learning_rate": 1.5878548830424574e-05, "loss": 0.19200587272644043, "step": 11950 }, { "epoch": 0.01943377059317168, "grad_norm": 1.1981679201126099, "learning_rate": 1.587158379654552e-05, "loss": 0.21378543376922607, "step": 11960 }, { "epoch": 0.0200226727323587, "grad_norm": 0.8052239418029785, "learning_rate": 1.5864614412866825e-05, "loss": 0.2439023733139038, "step": 11970 }, { "epoch": 0.02061157487154572, "grad_norm": 0.879245936870575, "learning_rate": 1.5857640684551565e-05, "loss": 0.28030521869659425, "step": 11980 }, { "epoch": 0.02120047701073274, "grad_norm": 0.6030606627464294, "learning_rate": 1.5850662616766034e-05, "loss": 0.2596741676330566, "step": 11990 }, { "epoch": 0.02178937914991976, "grad_norm": 0.7907504439353943, "learning_rate": 1.584368021467974e-05, "loss": 0.24229321479797364, "step": 12000 }, { "epoch": 0.022378281289106784, "grad_norm": 0.7820566296577454, "learning_rate": 1.5836693483465394e-05, "loss": 0.2770668983459473, "step": 12010 }, { "epoch": 0.022967183428293802, "grad_norm": 0.8558528423309326, "learning_rate": 1.5829702428298924e-05, "loss": 0.22346346378326415, "step": 12020 }, { "epoch": 0.023556085567480824, "grad_norm": 0.928383469581604, "learning_rate": 1.5822707054359467e-05, "loss": 0.2461709976196289, "step": 12030 }, { "epoch": 0.024144987706667845, "grad_norm": 0.9304463267326355, "learning_rate": 1.5815707366829343e-05, "loss": 0.24130589962005616, "step": 12040 }, { "epoch": 0.024733889845854864, "grad_norm": 1.0313597917556763, "learning_rate": 1.5808703370894082e-05, "loss": 0.2730099201202393, "step": 12050 }, { "epoch": 0.025322791985041886, "grad_norm": 0.7051250338554382, "learning_rate": 1.5801695071742394e-05, "loss": 0.24365234375, "step": 12060 }, { "epoch": 0.025911694124228907, "grad_norm": 0.6559680104255676, "learning_rate": 1.5794682474566188e-05, "loss": 0.23986153602600097, "step": 12070 }, { "epoch": 0.026500596263415926, "grad_norm": 0.9303649067878723, "learning_rate": 1.5787665584560546e-05, "loss": 0.2942129850387573, "step": 12080 }, { "epoch": 0.027089498402602948, "grad_norm": 0.6016737818717957, "learning_rate": 1.5780644406923738e-05, "loss": 0.25920240879058837, "step": 12090 }, { "epoch": 0.02767840054178997, "grad_norm": 1.0208640098571777, "learning_rate": 1.5773618946857213e-05, "loss": 0.28609237670898435, "step": 12100 }, { "epoch": 0.028267302680976988, "grad_norm": 0.9377001523971558, "learning_rate": 1.576658920956558e-05, "loss": 0.24636492729187012, "step": 12110 }, { "epoch": 0.02885620482016401, "grad_norm": 0.9141872525215149, "learning_rate": 1.575955520025663e-05, "loss": 0.2834151744842529, "step": 12120 }, { "epoch": 0.02944510695935103, "grad_norm": 1.0406179428100586, "learning_rate": 1.5752516924141313e-05, "loss": 0.28334996700286863, "step": 12130 }, { "epoch": 0.03003400909853805, "grad_norm": 0.8159959316253662, "learning_rate": 1.5745474386433736e-05, "loss": 0.2469841480255127, "step": 12140 }, { "epoch": 0.03062291123772507, "grad_norm": 1.087107539176941, "learning_rate": 1.5738427592351172e-05, "loss": 0.26403958797454835, "step": 12150 }, { "epoch": 0.031211813376912093, "grad_norm": 0.9794124960899353, "learning_rate": 1.5731376547114038e-05, "loss": 0.2549163579940796, "step": 12160 }, { "epoch": 0.031800715516099115, "grad_norm": 0.7795324325561523, "learning_rate": 1.572432125594591e-05, "loss": 0.2329784870147705, "step": 12170 }, { "epoch": 0.03238961765528613, "grad_norm": 0.7577768564224243, "learning_rate": 1.57172617240735e-05, "loss": 0.24108314514160156, "step": 12180 }, { "epoch": 0.03297851979447315, "grad_norm": 0.7712846994400024, "learning_rate": 1.571019795672667e-05, "loss": 0.2616602420806885, "step": 12190 }, { "epoch": 0.03356742193366018, "grad_norm": 0.7499082684516907, "learning_rate": 1.5703129959138413e-05, "loss": 0.2374483823776245, "step": 12200 }, { "epoch": 0.034156324072847195, "grad_norm": 0.7410082221031189, "learning_rate": 1.5696057736544858e-05, "loss": 0.25235106945037844, "step": 12210 }, { "epoch": 0.034745226212034214, "grad_norm": 0.7372048497200012, "learning_rate": 1.5688981294185266e-05, "loss": 0.2599467754364014, "step": 12220 }, { "epoch": 0.03533412835122124, "grad_norm": 0.9567103981971741, "learning_rate": 1.5681900637302023e-05, "loss": 0.2778938293457031, "step": 12230 }, { "epoch": 0.03592303049040826, "grad_norm": 0.7484291791915894, "learning_rate": 1.5674815771140634e-05, "loss": 0.26285884380340574, "step": 12240 }, { "epoch": 0.036511932629595276, "grad_norm": 0.7345116138458252, "learning_rate": 1.566772670094973e-05, "loss": 0.2842077732086182, "step": 12250 }, { "epoch": 0.0371008347687823, "grad_norm": 0.7571231722831726, "learning_rate": 1.5660633431981044e-05, "loss": 0.25629239082336425, "step": 12260 }, { "epoch": 0.03768973690796932, "grad_norm": 0.8892847299575806, "learning_rate": 1.5653535969489433e-05, "loss": 0.264446496963501, "step": 12270 }, { "epoch": 0.03827863904715634, "grad_norm": 1.1645851135253906, "learning_rate": 1.5646434318732856e-05, "loss": 0.2995718479156494, "step": 12280 }, { "epoch": 0.03886754118634336, "grad_norm": 1.024957299232483, "learning_rate": 1.563932848497237e-05, "loss": 0.24670801162719727, "step": 12290 }, { "epoch": 0.03945644332553038, "grad_norm": 0.9087234735488892, "learning_rate": 1.5632218473472134e-05, "loss": 0.26925506591796877, "step": 12300 }, { "epoch": 0.0400453454647174, "grad_norm": 0.909965455532074, "learning_rate": 1.5625104289499407e-05, "loss": 0.28475401401519773, "step": 12310 }, { "epoch": 0.04063424760390442, "grad_norm": 0.8738179802894592, "learning_rate": 1.561798593832453e-05, "loss": 0.25357880592346194, "step": 12320 }, { "epoch": 0.04122314974309144, "grad_norm": 0.8214993476867676, "learning_rate": 1.5610863425220933e-05, "loss": 0.24240617752075194, "step": 12330 }, { "epoch": 0.04181205188227846, "grad_norm": 0.7530680298805237, "learning_rate": 1.560373675546514e-05, "loss": 0.2454620599746704, "step": 12340 }, { "epoch": 0.04240095402146548, "grad_norm": 0.8500219583511353, "learning_rate": 1.5596605934336744e-05, "loss": 0.25774548053741453, "step": 12350 }, { "epoch": 0.042989856160652505, "grad_norm": 1.0240579843521118, "learning_rate": 1.5589470967118412e-05, "loss": 0.2415456771850586, "step": 12360 }, { "epoch": 0.04357875829983952, "grad_norm": 0.7986820340156555, "learning_rate": 1.558233185909589e-05, "loss": 0.26571853160858155, "step": 12370 }, { "epoch": 0.04416766043902654, "grad_norm": 0.7938029766082764, "learning_rate": 1.557518861555798e-05, "loss": 0.2660914659500122, "step": 12380 }, { "epoch": 0.04475656257821357, "grad_norm": 0.9721712470054626, "learning_rate": 1.5568041241796566e-05, "loss": 0.2774193286895752, "step": 12390 }, { "epoch": 0.045345464717400585, "grad_norm": 0.8742009401321411, "learning_rate": 1.556088974310657e-05, "loss": 0.25099222660064696, "step": 12400 }, { "epoch": 0.045934366856587604, "grad_norm": 0.8509458303451538, "learning_rate": 1.5553734124785994e-05, "loss": 0.22030246257781982, "step": 12410 }, { "epoch": 0.04652326899577463, "grad_norm": 0.9471731185913086, "learning_rate": 1.5546574392135868e-05, "loss": 0.26781535148620605, "step": 12420 }, { "epoch": 0.04711217113496165, "grad_norm": 0.8678634762763977, "learning_rate": 1.5539410550460285e-05, "loss": 0.25667169094085696, "step": 12430 }, { "epoch": 0.047701073274148666, "grad_norm": 0.8831449747085571, "learning_rate": 1.553224260506638e-05, "loss": 0.2719725608825684, "step": 12440 }, { "epoch": 0.04828997541333569, "grad_norm": 0.8634048104286194, "learning_rate": 1.552507056126432e-05, "loss": 0.29556522369384763, "step": 12450 }, { "epoch": 0.04887887755252271, "grad_norm": 1.0266213417053223, "learning_rate": 1.551789442436732e-05, "loss": 0.2401069164276123, "step": 12460 }, { "epoch": 0.04946777969170973, "grad_norm": 0.827022910118103, "learning_rate": 1.551071419969162e-05, "loss": 0.25113873481750487, "step": 12470 }, { "epoch": 0.05005668183089675, "grad_norm": 0.8410949110984802, "learning_rate": 1.5503529892556493e-05, "loss": 0.28369357585906985, "step": 12480 }, { "epoch": 0.05064558397008377, "grad_norm": 0.8468912243843079, "learning_rate": 1.5496341508284226e-05, "loss": 0.27729268074035646, "step": 12490 }, { "epoch": 0.05123448610927079, "grad_norm": 0.9013922214508057, "learning_rate": 1.5489149052200138e-05, "loss": 0.25900471210479736, "step": 12500 }, { "epoch": 0.051823388248457815, "grad_norm": 0.9383749961853027, "learning_rate": 1.548195252963256e-05, "loss": 0.22554328441619872, "step": 12510 }, { "epoch": 0.05241229038764483, "grad_norm": 0.7380775809288025, "learning_rate": 1.5474751945912835e-05, "loss": 0.23168020248413085, "step": 12520 }, { "epoch": 0.05300119252683185, "grad_norm": 0.8592778444290161, "learning_rate": 1.546754730637532e-05, "loss": 0.22238407135009766, "step": 12530 }, { "epoch": 0.05359009466601888, "grad_norm": 0.8518626689910889, "learning_rate": 1.5460338616357364e-05, "loss": 0.23937733173370362, "step": 12540 }, { "epoch": 0.054178996805205895, "grad_norm": 0.847419798374176, "learning_rate": 1.5453125881199327e-05, "loss": 0.23455655574798584, "step": 12550 }, { "epoch": 0.05476789894439291, "grad_norm": 1.4479928016662598, "learning_rate": 1.5445909106244564e-05, "loss": 0.22133772373199462, "step": 12560 }, { "epoch": 0.05535680108357994, "grad_norm": 1.0371724367141724, "learning_rate": 1.5438688296839426e-05, "loss": 0.2700294256210327, "step": 12570 }, { "epoch": 0.05594570322276696, "grad_norm": 1.0472230911254883, "learning_rate": 1.5431463458333243e-05, "loss": 0.25331385135650636, "step": 12580 }, { "epoch": 0.056534605361953975, "grad_norm": 1.1498336791992188, "learning_rate": 1.542423459607834e-05, "loss": 0.25460312366485593, "step": 12590 }, { "epoch": 0.057123507501141, "grad_norm": 1.1439204216003418, "learning_rate": 1.541700171543002e-05, "loss": 0.23962085247039794, "step": 12600 }, { "epoch": 0.05771240964032802, "grad_norm": 0.8316977024078369, "learning_rate": 1.540976482174655e-05, "loss": 0.2479788541793823, "step": 12610 }, { "epoch": 0.05830131177951504, "grad_norm": 0.9963715672492981, "learning_rate": 1.54025239203892e-05, "loss": 0.26311383247375486, "step": 12620 }, { "epoch": 0.05889021391870206, "grad_norm": 0.8688972592353821, "learning_rate": 1.5395279016722176e-05, "loss": 0.22406220436096191, "step": 12630 }, { "epoch": 0.05947911605788908, "grad_norm": 0.9333828091621399, "learning_rate": 1.538803011611267e-05, "loss": 0.2658046007156372, "step": 12640 }, { "epoch": 0.0600680181970761, "grad_norm": 0.795620322227478, "learning_rate": 1.5380777223930823e-05, "loss": 0.2865647077560425, "step": 12650 }, { "epoch": 0.060656920336263125, "grad_norm": 1.0222352743148804, "learning_rate": 1.5373520345549743e-05, "loss": 0.26830549240112306, "step": 12660 }, { "epoch": 0.06124582247545014, "grad_norm": 0.7058101296424866, "learning_rate": 1.5366259486345488e-05, "loss": 0.26091799736022947, "step": 12670 }, { "epoch": 0.06183472461463716, "grad_norm": 0.7747994065284729, "learning_rate": 1.535899465169706e-05, "loss": 0.26307215690612795, "step": 12680 }, { "epoch": 0.062423626753824187, "grad_norm": 0.8910185098648071, "learning_rate": 1.535172584698641e-05, "loss": 0.2615605592727661, "step": 12690 }, { "epoch": 0.0630125288930112, "grad_norm": 0.8106526732444763, "learning_rate": 1.5344453077598437e-05, "loss": 0.2539846658706665, "step": 12700 }, { "epoch": 0.06360143103219823, "grad_norm": 0.7884617447853088, "learning_rate": 1.533717634892096e-05, "loss": 0.2816328763961792, "step": 12710 }, { "epoch": 0.06419033317138524, "grad_norm": 0.9555450081825256, "learning_rate": 1.5329895666344746e-05, "loss": 0.28100976943969724, "step": 12720 }, { "epoch": 0.06477923531057227, "grad_norm": 0.7128174901008606, "learning_rate": 1.5322611035263484e-05, "loss": 0.3244236946105957, "step": 12730 }, { "epoch": 0.06536813744975929, "grad_norm": 0.7825138568878174, "learning_rate": 1.5315322461073795e-05, "loss": 0.2631181240081787, "step": 12740 }, { "epoch": 0.0659570395889463, "grad_norm": 0.8691178560256958, "learning_rate": 1.5308029949175213e-05, "loss": 0.266397500038147, "step": 12750 }, { "epoch": 0.06654594172813333, "grad_norm": 0.955094575881958, "learning_rate": 1.5300733504970192e-05, "loss": 0.26110742092132566, "step": 12760 }, { "epoch": 0.06713484386732035, "grad_norm": 0.8442791104316711, "learning_rate": 1.5293433133864106e-05, "loss": 0.25921380519866943, "step": 12770 }, { "epoch": 0.06772374600650737, "grad_norm": 0.8636476397514343, "learning_rate": 1.528612884126522e-05, "loss": 0.2836390733718872, "step": 12780 }, { "epoch": 0.06831264814569439, "grad_norm": 0.7830700874328613, "learning_rate": 1.527882063258473e-05, "loss": 0.30038459300994874, "step": 12790 }, { "epoch": 0.06890155028488142, "grad_norm": 0.8087319135665894, "learning_rate": 1.527150851323671e-05, "loss": 0.3024989128112793, "step": 12800 }, { "epoch": 0.06949045242406843, "grad_norm": 0.8315728306770325, "learning_rate": 1.5264192488638142e-05, "loss": 0.3018558263778687, "step": 12810 }, { "epoch": 0.07007935456325545, "grad_norm": 0.8211753964424133, "learning_rate": 1.5256872564208905e-05, "loss": 0.26163496971130373, "step": 12820 }, { "epoch": 0.07066825670244248, "grad_norm": 0.8267154693603516, "learning_rate": 1.5249548745371755e-05, "loss": 0.25537106990814207, "step": 12830 }, { "epoch": 0.07125715884162949, "grad_norm": 0.6814873814582825, "learning_rate": 1.524222103755234e-05, "loss": 0.3218426465988159, "step": 12840 }, { "epoch": 0.07184606098081651, "grad_norm": 0.8436844348907471, "learning_rate": 1.5234889446179188e-05, "loss": 0.3057889461517334, "step": 12850 }, { "epoch": 0.07243496312000354, "grad_norm": 0.9201265573501587, "learning_rate": 1.5227553976683713e-05, "loss": 0.2833522081375122, "step": 12860 }, { "epoch": 0.07302386525919055, "grad_norm": 0.7376656532287598, "learning_rate": 1.5220214634500186e-05, "loss": 0.25899648666381836, "step": 12870 }, { "epoch": 0.07361276739837758, "grad_norm": 0.6988154053688049, "learning_rate": 1.5212871425065757e-05, "loss": 0.2536582946777344, "step": 12880 }, { "epoch": 0.0742016695375646, "grad_norm": 0.7980970144271851, "learning_rate": 1.5205524353820434e-05, "loss": 0.2777430057525635, "step": 12890 }, { "epoch": 0.07479057167675161, "grad_norm": 0.8025479912757874, "learning_rate": 1.5198173426207095e-05, "loss": 0.3030802488327026, "step": 12900 }, { "epoch": 0.07537947381593864, "grad_norm": 0.8725082278251648, "learning_rate": 1.5190818647671473e-05, "loss": 0.2494436025619507, "step": 12910 }, { "epoch": 0.07596837595512566, "grad_norm": 0.6699666976928711, "learning_rate": 1.5183460023662147e-05, "loss": 0.26429452896118166, "step": 12920 }, { "epoch": 0.07655727809431268, "grad_norm": 0.7624750733375549, "learning_rate": 1.5176097559630549e-05, "loss": 0.2841654300689697, "step": 12930 }, { "epoch": 0.0771461802334997, "grad_norm": 0.8341474533081055, "learning_rate": 1.516873126103096e-05, "loss": 0.2863186836242676, "step": 12940 }, { "epoch": 0.07773508237268673, "grad_norm": 0.8438423275947571, "learning_rate": 1.5161361133320489e-05, "loss": 0.33439548015594484, "step": 12950 }, { "epoch": 0.07832398451187374, "grad_norm": 0.8691487908363342, "learning_rate": 1.51539871819591e-05, "loss": 0.28634071350097656, "step": 12960 }, { "epoch": 0.07891288665106076, "grad_norm": 0.7721070647239685, "learning_rate": 1.5146609412409574e-05, "loss": 0.2562690258026123, "step": 12970 }, { "epoch": 0.07950178879024777, "grad_norm": 0.6665549278259277, "learning_rate": 1.5139227830137528e-05, "loss": 0.2701383590698242, "step": 12980 }, { "epoch": 0.0800906909294348, "grad_norm": 0.7902573943138123, "learning_rate": 1.5131842440611402e-05, "loss": 0.29200100898742676, "step": 12990 }, { "epoch": 0.08067959306862182, "grad_norm": 0.7318968772888184, "learning_rate": 1.512445324930246e-05, "loss": 0.29139723777771, "step": 13000 }, { "epoch": 0.08126849520780884, "grad_norm": 0.7768885493278503, "learning_rate": 1.5117060261684775e-05, "loss": 0.29955482482910156, "step": 13010 }, { "epoch": 0.08185739734699586, "grad_norm": 0.648906946182251, "learning_rate": 1.5109663483235231e-05, "loss": 0.29016039371490476, "step": 13020 }, { "epoch": 0.08244629948618289, "grad_norm": 0.6252663731575012, "learning_rate": 1.5102262919433536e-05, "loss": 0.30808005332946775, "step": 13030 }, { "epoch": 0.0830352016253699, "grad_norm": 0.9363887906074524, "learning_rate": 1.5094858575762188e-05, "loss": 0.30365605354309083, "step": 13040 }, { "epoch": 0.08362410376455692, "grad_norm": 0.7193140387535095, "learning_rate": 1.5087450457706487e-05, "loss": 0.28358473777771, "step": 13050 }, { "epoch": 0.08421300590374395, "grad_norm": 0.812024712562561, "learning_rate": 1.5080038570754534e-05, "loss": 0.2798715114593506, "step": 13060 }, { "epoch": 0.08480190804293096, "grad_norm": 0.8600712418556213, "learning_rate": 1.5072622920397222e-05, "loss": 0.30823321342468263, "step": 13070 }, { "epoch": 0.08539081018211798, "grad_norm": 0.7338094115257263, "learning_rate": 1.5065203512128222e-05, "loss": 0.2982280969619751, "step": 13080 }, { "epoch": 0.08597971232130501, "grad_norm": 0.7348097562789917, "learning_rate": 1.5057780351444e-05, "loss": 0.3127293109893799, "step": 13090 }, { "epoch": 0.08656861446049202, "grad_norm": 0.7174997925758362, "learning_rate": 1.5050353443843801e-05, "loss": 0.3525747060775757, "step": 13100 }, { "epoch": 0.08715751659967905, "grad_norm": 0.7118021249771118, "learning_rate": 1.5042922794829639e-05, "loss": 0.30384957790374756, "step": 13110 }, { "epoch": 0.08774641873886607, "grad_norm": 0.6335732936859131, "learning_rate": 1.5035488409906307e-05, "loss": 0.2726653814315796, "step": 13120 }, { "epoch": 0.08833532087805308, "grad_norm": 0.5588036775588989, "learning_rate": 1.5028050294581362e-05, "loss": 0.3171590805053711, "step": 13130 }, { "epoch": 0.08892422301724011, "grad_norm": 0.6822083592414856, "learning_rate": 1.5020608454365122e-05, "loss": 0.2713651180267334, "step": 13140 }, { "epoch": 0.08951312515642713, "grad_norm": 0.9313845634460449, "learning_rate": 1.501316289477067e-05, "loss": 0.3333893775939941, "step": 13150 }, { "epoch": 0.09010202729561415, "grad_norm": 1.0424259901046753, "learning_rate": 1.500571362131384e-05, "loss": 0.31451025009155276, "step": 13160 }, { "epoch": 0.09069092943480117, "grad_norm": 0.8081899285316467, "learning_rate": 1.499826063951322e-05, "loss": 0.3215681552886963, "step": 13170 }, { "epoch": 0.0912798315739882, "grad_norm": 0.6237261891365051, "learning_rate": 1.4990803954890147e-05, "loss": 0.302283787727356, "step": 13180 }, { "epoch": 0.09186873371317521, "grad_norm": 0.7115081548690796, "learning_rate": 1.4983343572968693e-05, "loss": 0.30943524837493896, "step": 13190 }, { "epoch": 0.09245763585236223, "grad_norm": 0.7499514222145081, "learning_rate": 1.497587949927568e-05, "loss": 0.31459941864013674, "step": 13200 }, { "epoch": 0.09304653799154926, "grad_norm": 0.8456103205680847, "learning_rate": 1.4968411739340659e-05, "loss": 0.2884716510772705, "step": 13210 }, { "epoch": 0.09363544013073627, "grad_norm": 0.6372024416923523, "learning_rate": 1.4960940298695907e-05, "loss": 0.27087113857269285, "step": 13220 }, { "epoch": 0.0942243422699233, "grad_norm": 0.7735040187835693, "learning_rate": 1.4953465182876439e-05, "loss": 0.31072239875793456, "step": 13230 }, { "epoch": 0.09481324440911032, "grad_norm": 0.7542491555213928, "learning_rate": 1.4945986397419985e-05, "loss": 0.2965421676635742, "step": 13240 }, { "epoch": 0.09540214654829733, "grad_norm": 0.7682355046272278, "learning_rate": 1.4938503947866996e-05, "loss": 0.26643354892730714, "step": 13250 }, { "epoch": 0.09599104868748436, "grad_norm": 0.7570721507072449, "learning_rate": 1.493101783976063e-05, "loss": 0.2764775514602661, "step": 13260 }, { "epoch": 0.09657995082667138, "grad_norm": 0.7590603828430176, "learning_rate": 1.4923528078646774e-05, "loss": 0.2913054943084717, "step": 13270 }, { "epoch": 0.0971688529658584, "grad_norm": 0.8102444410324097, "learning_rate": 1.4916034670074e-05, "loss": 0.2930643320083618, "step": 13280 }, { "epoch": 0.09775775510504542, "grad_norm": 0.7198961973190308, "learning_rate": 1.4908537619593597e-05, "loss": 0.3171946287155151, "step": 13290 }, { "epoch": 0.09834665724423244, "grad_norm": 0.6881096959114075, "learning_rate": 1.4901036932759541e-05, "loss": 0.30657267570495605, "step": 13300 }, { "epoch": 0.09893555938341946, "grad_norm": 0.8637014031410217, "learning_rate": 1.4893532615128513e-05, "loss": 0.2922516822814941, "step": 13310 }, { "epoch": 0.09952446152260648, "grad_norm": 0.6620384454727173, "learning_rate": 1.4886024672259874e-05, "loss": 0.2702995777130127, "step": 13320 }, { "epoch": 0.1001133636617935, "grad_norm": 0.8244351744651794, "learning_rate": 1.4878513109715675e-05, "loss": 0.3023902654647827, "step": 13330 }, { "epoch": 0.10070226580098052, "grad_norm": 0.66795814037323, "learning_rate": 1.4870997933060651e-05, "loss": 0.2981382131576538, "step": 13340 }, { "epoch": 0.10129116794016754, "grad_norm": 0.7063947916030884, "learning_rate": 1.486347914786221e-05, "loss": 0.25862534046173097, "step": 13350 }, { "epoch": 0.10188007007935457, "grad_norm": 0.5699521899223328, "learning_rate": 1.4855956759690434e-05, "loss": 0.25439021587371824, "step": 13360 }, { "epoch": 0.10246897221854158, "grad_norm": 0.8804308176040649, "learning_rate": 1.4848430774118078e-05, "loss": 0.316045880317688, "step": 13370 }, { "epoch": 0.1030578743577286, "grad_norm": 0.8362777233123779, "learning_rate": 1.484090119672056e-05, "loss": 0.3030997753143311, "step": 13380 }, { "epoch": 0.10364677649691563, "grad_norm": 0.5357567667961121, "learning_rate": 1.4833368033075952e-05, "loss": 0.3442386150360107, "step": 13390 }, { "epoch": 0.10423567863610264, "grad_norm": 0.824743390083313, "learning_rate": 1.4825831288764993e-05, "loss": 0.30246012210845946, "step": 13400 }, { "epoch": 0.10482458077528967, "grad_norm": 0.7215975522994995, "learning_rate": 1.4818290969371072e-05, "loss": 0.27987103462219237, "step": 13410 }, { "epoch": 0.10541348291447669, "grad_norm": 0.7034637331962585, "learning_rate": 1.481074708048022e-05, "loss": 0.26503539085388184, "step": 13420 }, { "epoch": 0.1060023850536637, "grad_norm": 0.6101664900779724, "learning_rate": 1.4803199627681123e-05, "loss": 0.26772398948669435, "step": 13430 }, { "epoch": 0.10659128719285073, "grad_norm": 0.6687732338905334, "learning_rate": 1.4795648616565098e-05, "loss": 0.3161060094833374, "step": 13440 }, { "epoch": 0.10718018933203775, "grad_norm": 0.6353771090507507, "learning_rate": 1.4788094052726104e-05, "loss": 0.3050650358200073, "step": 13450 }, { "epoch": 0.10776909147122476, "grad_norm": 0.860487699508667, "learning_rate": 1.4780535941760727e-05, "loss": 0.3034648418426514, "step": 13460 }, { "epoch": 0.10835799361041179, "grad_norm": 0.7225093245506287, "learning_rate": 1.4772974289268185e-05, "loss": 0.2835752248764038, "step": 13470 }, { "epoch": 0.10894689574959882, "grad_norm": 0.6576554775238037, "learning_rate": 1.4765409100850318e-05, "loss": 0.2566915273666382, "step": 13480 }, { "epoch": 0.10953579788878583, "grad_norm": 0.5869423747062683, "learning_rate": 1.4757840382111586e-05, "loss": 0.2783390283584595, "step": 13490 }, { "epoch": 0.11012470002797285, "grad_norm": 0.8595415949821472, "learning_rate": 1.4750268138659066e-05, "loss": 0.2933320999145508, "step": 13500 }, { "epoch": 0.11071360216715988, "grad_norm": 0.7233706116676331, "learning_rate": 1.4742692376102436e-05, "loss": 0.30211615562438965, "step": 13510 }, { "epoch": 0.11130250430634689, "grad_norm": 0.6554062366485596, "learning_rate": 1.4735113100054e-05, "loss": 0.25289771556854246, "step": 13520 }, { "epoch": 0.11189140644553391, "grad_norm": 0.6764119863510132, "learning_rate": 1.4727530316128643e-05, "loss": 0.29608750343322754, "step": 13530 }, { "epoch": 0.11248030858472094, "grad_norm": 0.7845160961151123, "learning_rate": 1.4719944029943868e-05, "loss": 0.2663578510284424, "step": 13540 }, { "epoch": 0.11306921072390795, "grad_norm": 0.5667412281036377, "learning_rate": 1.4712354247119763e-05, "loss": 0.28085601329803467, "step": 13550 }, { "epoch": 0.11365811286309498, "grad_norm": 0.7553285360336304, "learning_rate": 1.4704760973279004e-05, "loss": 0.2845356225967407, "step": 13560 }, { "epoch": 0.114247015002282, "grad_norm": 0.8191734552383423, "learning_rate": 1.469716421404686e-05, "loss": 0.30377218723297117, "step": 13570 }, { "epoch": 0.11483591714146901, "grad_norm": 0.5517035126686096, "learning_rate": 1.4689563975051177e-05, "loss": 0.2927053928375244, "step": 13580 }, { "epoch": 0.11542481928065604, "grad_norm": 0.6463730335235596, "learning_rate": 1.468196026192238e-05, "loss": 0.2838677167892456, "step": 13590 }, { "epoch": 0.11601372141984306, "grad_norm": 0.586831271648407, "learning_rate": 1.4674353080293475e-05, "loss": 0.27516562938690187, "step": 13600 }, { "epoch": 0.11660262355903007, "grad_norm": 0.7714823484420776, "learning_rate": 1.4666742435800021e-05, "loss": 0.2953354835510254, "step": 13610 }, { "epoch": 0.1171915256982171, "grad_norm": 0.7124198079109192, "learning_rate": 1.4659128334080159e-05, "loss": 0.3215376138687134, "step": 13620 }, { "epoch": 0.11778042783740413, "grad_norm": 0.696985125541687, "learning_rate": 1.4651510780774585e-05, "loss": 0.3277964353561401, "step": 13630 }, { "epoch": 0.11836932997659114, "grad_norm": 0.591172993183136, "learning_rate": 1.464388978152655e-05, "loss": 0.28361508846282957, "step": 13640 }, { "epoch": 0.11895823211577816, "grad_norm": 0.5818543434143066, "learning_rate": 1.4636265341981856e-05, "loss": 0.3027889013290405, "step": 13650 }, { "epoch": 0.11954713425496519, "grad_norm": 0.8337877988815308, "learning_rate": 1.4628637467788862e-05, "loss": 0.2921525239944458, "step": 13660 }, { "epoch": 0.1201360363941522, "grad_norm": 0.9060330986976624, "learning_rate": 1.4621006164598465e-05, "loss": 0.31959483623504636, "step": 13670 }, { "epoch": 0.12072493853333922, "grad_norm": 0.7409210205078125, "learning_rate": 1.4613371438064104e-05, "loss": 0.31727797985076905, "step": 13680 }, { "epoch": 0.12131384067252625, "grad_norm": 0.6660519242286682, "learning_rate": 1.4605733293841754e-05, "loss": 0.30997018814086913, "step": 13690 }, { "epoch": 0.12190274281171326, "grad_norm": 0.6209553480148315, "learning_rate": 1.4598091737589921e-05, "loss": 0.34272012710571287, "step": 13700 }, { "epoch": 0.12249164495090029, "grad_norm": 0.7438091039657593, "learning_rate": 1.459044677496964e-05, "loss": 0.3172158241271973, "step": 13710 }, { "epoch": 0.12308054709008731, "grad_norm": 0.7077498435974121, "learning_rate": 1.4582798411644468e-05, "loss": 0.31994214057922366, "step": 13720 }, { "epoch": 0.12366944922927432, "grad_norm": 0.5506160855293274, "learning_rate": 1.4575146653280486e-05, "loss": 0.34152414798736574, "step": 13730 }, { "epoch": 0.12425835136846135, "grad_norm": 0.7787634134292603, "learning_rate": 1.4567491505546283e-05, "loss": 0.31667494773864746, "step": 13740 }, { "epoch": 0.12484725350764837, "grad_norm": 0.6417492628097534, "learning_rate": 1.455983297411296e-05, "loss": 0.31601998805999754, "step": 13750 }, { "epoch": 0.12543615564683538, "grad_norm": 0.703084409236908, "learning_rate": 1.4552171064654133e-05, "loss": 0.3274413824081421, "step": 13760 }, { "epoch": 0.1260250577860224, "grad_norm": 0.6935283541679382, "learning_rate": 1.4544505782845915e-05, "loss": 0.34646332263946533, "step": 13770 }, { "epoch": 0.12661395992520943, "grad_norm": 0.6519109010696411, "learning_rate": 1.453683713436691e-05, "loss": 0.29553234577178955, "step": 13780 }, { "epoch": 0.12720286206439646, "grad_norm": 0.6781842708587646, "learning_rate": 1.4529165124898227e-05, "loss": 0.33284592628479004, "step": 13790 }, { "epoch": 0.12779176420358346, "grad_norm": 0.8738517761230469, "learning_rate": 1.4521489760123462e-05, "loss": 0.3854241371154785, "step": 13800 }, { "epoch": 0.12838066634277048, "grad_norm": 0.7508092522621155, "learning_rate": 1.4513811045728697e-05, "loss": 0.2982900619506836, "step": 13810 }, { "epoch": 0.1289695684819575, "grad_norm": 0.7469606995582581, "learning_rate": 1.4506128987402488e-05, "loss": 0.33575100898742677, "step": 13820 }, { "epoch": 0.12955847062114453, "grad_norm": 0.6228040456771851, "learning_rate": 1.449844359083588e-05, "loss": 0.2894963264465332, "step": 13830 }, { "epoch": 0.13014737276033156, "grad_norm": 0.626414954662323, "learning_rate": 1.4490754861722385e-05, "loss": 0.32844531536102295, "step": 13840 }, { "epoch": 0.13073627489951858, "grad_norm": 0.6477864980697632, "learning_rate": 1.4483062805757978e-05, "loss": 0.3281565189361572, "step": 13850 }, { "epoch": 0.13132517703870558, "grad_norm": 0.5560583472251892, "learning_rate": 1.4475367428641115e-05, "loss": 0.32048187255859373, "step": 13860 }, { "epoch": 0.1319140791778926, "grad_norm": 0.6434531807899475, "learning_rate": 1.4467668736072693e-05, "loss": 0.34002792835235596, "step": 13870 }, { "epoch": 0.13250298131707963, "grad_norm": 0.6729580760002136, "learning_rate": 1.445996673375608e-05, "loss": 0.3162633657455444, "step": 13880 }, { "epoch": 0.13309188345626666, "grad_norm": 0.6491057872772217, "learning_rate": 1.445226142739709e-05, "loss": 0.3363695859909058, "step": 13890 }, { "epoch": 0.13368078559545368, "grad_norm": 0.7860457897186279, "learning_rate": 1.4444552822703982e-05, "loss": 0.26846554279327395, "step": 13900 }, { "epoch": 0.1342696877346407, "grad_norm": 0.5288339257240295, "learning_rate": 1.4436840925387464e-05, "loss": 0.2898768424987793, "step": 13910 }, { "epoch": 0.1348585898738277, "grad_norm": 0.5519075989723206, "learning_rate": 1.4429125741160677e-05, "loss": 0.30516173839569094, "step": 13920 }, { "epoch": 0.13544749201301473, "grad_norm": 0.5976701974868774, "learning_rate": 1.4421407275739206e-05, "loss": 0.31471436023712157, "step": 13930 }, { "epoch": 0.13603639415220176, "grad_norm": 0.6020264029502869, "learning_rate": 1.441368553484106e-05, "loss": 0.30030026435852053, "step": 13940 }, { "epoch": 0.13662529629138878, "grad_norm": 0.7422778606414795, "learning_rate": 1.4405960524186673e-05, "loss": 0.3085629463195801, "step": 13950 }, { "epoch": 0.1372141984305758, "grad_norm": 0.5411365628242493, "learning_rate": 1.439823224949891e-05, "loss": 0.29759917259216306, "step": 13960 }, { "epoch": 0.13780310056976283, "grad_norm": 0.6690809726715088, "learning_rate": 1.439050071650304e-05, "loss": 0.30604236125946044, "step": 13970 }, { "epoch": 0.13839200270894983, "grad_norm": 0.8886741399765015, "learning_rate": 1.4382765930926763e-05, "loss": 0.31396629810333254, "step": 13980 }, { "epoch": 0.13898090484813685, "grad_norm": 0.8224364519119263, "learning_rate": 1.4375027898500176e-05, "loss": 0.3112340927124023, "step": 13990 }, { "epoch": 0.13956980698732388, "grad_norm": 0.5930447578430176, "learning_rate": 1.4367286624955785e-05, "loss": 0.3218132972717285, "step": 14000 }, { "epoch": 0.1401587091265109, "grad_norm": 0.5149272084236145, "learning_rate": 1.4359542116028498e-05, "loss": 0.30363523960113525, "step": 14010 }, { "epoch": 0.14074761126569793, "grad_norm": 0.5948032736778259, "learning_rate": 1.4351794377455619e-05, "loss": 0.3229851722717285, "step": 14020 }, { "epoch": 0.14133651340488496, "grad_norm": 0.7040603756904602, "learning_rate": 1.4344043414976848e-05, "loss": 0.326879096031189, "step": 14030 }, { "epoch": 0.14192541554407195, "grad_norm": 0.6131165623664856, "learning_rate": 1.4336289234334262e-05, "loss": 0.32100982666015626, "step": 14040 }, { "epoch": 0.14251431768325898, "grad_norm": 0.6748123168945312, "learning_rate": 1.4328531841272337e-05, "loss": 0.2846100330352783, "step": 14050 }, { "epoch": 0.143103219822446, "grad_norm": 0.663483202457428, "learning_rate": 1.4320771241537919e-05, "loss": 0.30771186351776125, "step": 14060 }, { "epoch": 0.14369212196163303, "grad_norm": 0.707445502281189, "learning_rate": 1.4313007440880235e-05, "loss": 0.3091113805770874, "step": 14070 }, { "epoch": 0.14428102410082005, "grad_norm": 0.7439794540405273, "learning_rate": 1.4305240445050878e-05, "loss": 0.3232275485992432, "step": 14080 }, { "epoch": 0.14486992624000708, "grad_norm": 0.6431906223297119, "learning_rate": 1.4297470259803814e-05, "loss": 0.3384516716003418, "step": 14090 }, { "epoch": 0.14545882837919408, "grad_norm": 0.6182807683944702, "learning_rate": 1.428969689089537e-05, "loss": 0.28228154182434084, "step": 14100 }, { "epoch": 0.1460477305183811, "grad_norm": 0.6816801428794861, "learning_rate": 1.4281920344084224e-05, "loss": 0.3162877082824707, "step": 14110 }, { "epoch": 0.14663663265756813, "grad_norm": 0.6848419308662415, "learning_rate": 1.4274140625131419e-05, "loss": 0.3347774028778076, "step": 14120 }, { "epoch": 0.14722553479675515, "grad_norm": 0.5635632276535034, "learning_rate": 1.4266357739800343e-05, "loss": 0.2926140308380127, "step": 14130 }, { "epoch": 0.14781443693594218, "grad_norm": 0.6192018985748291, "learning_rate": 1.4258571693856729e-05, "loss": 0.3188788652420044, "step": 14140 }, { "epoch": 0.1484033390751292, "grad_norm": 0.5717487931251526, "learning_rate": 1.4250782493068653e-05, "loss": 0.29560601711273193, "step": 14150 }, { "epoch": 0.1489922412143162, "grad_norm": 0.6981801390647888, "learning_rate": 1.4242990143206527e-05, "loss": 0.29928054809570315, "step": 14160 }, { "epoch": 0.14958114335350323, "grad_norm": 0.6472418904304504, "learning_rate": 1.4235194650043105e-05, "loss": 0.3263159513473511, "step": 14170 }, { "epoch": 0.15017004549269025, "grad_norm": 0.7637180685997009, "learning_rate": 1.4227396019353447e-05, "loss": 0.3473826885223389, "step": 14180 }, { "epoch": 0.15075894763187728, "grad_norm": 0.599142849445343, "learning_rate": 1.4219594256914963e-05, "loss": 0.323313045501709, "step": 14190 }, { "epoch": 0.1513478497710643, "grad_norm": 0.6933709979057312, "learning_rate": 1.4211789368507369e-05, "loss": 0.32069756984710696, "step": 14200 }, { "epoch": 0.15193675191025133, "grad_norm": 0.6267322897911072, "learning_rate": 1.42039813599127e-05, "loss": 0.3259279727935791, "step": 14210 }, { "epoch": 0.15252565404943833, "grad_norm": 0.6046898365020752, "learning_rate": 1.4196170236915296e-05, "loss": 0.3063561677932739, "step": 14220 }, { "epoch": 0.15311455618862535, "grad_norm": 0.7506183385848999, "learning_rate": 1.4188356005301817e-05, "loss": 0.27788991928100587, "step": 14230 }, { "epoch": 0.15370345832781238, "grad_norm": 0.5560901165008545, "learning_rate": 1.418053867086122e-05, "loss": 0.3199488878250122, "step": 14240 }, { "epoch": 0.1542923604669994, "grad_norm": 0.8307417631149292, "learning_rate": 1.4172718239384755e-05, "loss": 0.29327077865600587, "step": 14250 }, { "epoch": 0.15488126260618643, "grad_norm": 0.8435890674591064, "learning_rate": 1.4164894716665978e-05, "loss": 0.3653285264968872, "step": 14260 }, { "epoch": 0.15547016474537345, "grad_norm": 0.7695178985595703, "learning_rate": 1.4157068108500722e-05, "loss": 0.3231559991836548, "step": 14270 }, { "epoch": 0.15605906688456045, "grad_norm": 0.45125705003738403, "learning_rate": 1.4149238420687113e-05, "loss": 0.3153369426727295, "step": 14280 }, { "epoch": 0.15664796902374747, "grad_norm": 0.7607532143592834, "learning_rate": 1.4141405659025559e-05, "loss": 0.3205312967300415, "step": 14290 }, { "epoch": 0.1572368711629345, "grad_norm": 0.5081688165664673, "learning_rate": 1.4133569829318742e-05, "loss": 0.32796311378479004, "step": 14300 }, { "epoch": 0.15782577330212152, "grad_norm": 0.6292189359664917, "learning_rate": 1.4125730937371625e-05, "loss": 0.3034865140914917, "step": 14310 }, { "epoch": 0.15841467544130855, "grad_norm": 0.6047752499580383, "learning_rate": 1.4117888988991424e-05, "loss": 0.29184932708740235, "step": 14320 }, { "epoch": 0.15900357758049555, "grad_norm": 0.7075582146644592, "learning_rate": 1.4110043989987634e-05, "loss": 0.30865304470062255, "step": 14330 }, { "epoch": 0.15959247971968257, "grad_norm": 0.5587031841278076, "learning_rate": 1.4102195946172006e-05, "loss": 0.288725471496582, "step": 14340 }, { "epoch": 0.1601813818588696, "grad_norm": 0.7812707424163818, "learning_rate": 1.4094344863358535e-05, "loss": 0.3417935848236084, "step": 14350 }, { "epoch": 0.16077028399805662, "grad_norm": 0.6260077953338623, "learning_rate": 1.4086490747363492e-05, "loss": 0.3146245002746582, "step": 14360 }, { "epoch": 0.16135918613724365, "grad_norm": 0.5032168626785278, "learning_rate": 1.4078633604005369e-05, "loss": 0.3620777130126953, "step": 14370 }, { "epoch": 0.16194808827643067, "grad_norm": 0.4904204308986664, "learning_rate": 1.407077343910492e-05, "loss": 0.2775260448455811, "step": 14380 }, { "epoch": 0.16253699041561767, "grad_norm": 0.5815445780754089, "learning_rate": 1.4062910258485127e-05, "loss": 0.3015543699264526, "step": 14390 }, { "epoch": 0.1631258925548047, "grad_norm": 0.8119376301765442, "learning_rate": 1.4055044067971205e-05, "loss": 0.30892186164855956, "step": 14400 }, { "epoch": 0.16371479469399172, "grad_norm": 0.5785220265388489, "learning_rate": 1.4047174873390607e-05, "loss": 0.34497082233428955, "step": 14410 }, { "epoch": 0.16430369683317875, "grad_norm": 0.4791807532310486, "learning_rate": 1.4039302680573004e-05, "loss": 0.3143503665924072, "step": 14420 }, { "epoch": 0.16489259897236577, "grad_norm": 0.7553791999816895, "learning_rate": 1.4031427495350297e-05, "loss": 0.3146370410919189, "step": 14430 }, { "epoch": 0.1654815011115528, "grad_norm": 0.5333262085914612, "learning_rate": 1.4023549323556589e-05, "loss": 0.2912472724914551, "step": 14440 }, { "epoch": 0.1660704032507398, "grad_norm": 0.7319854497909546, "learning_rate": 1.4015668171028211e-05, "loss": 0.3031296968460083, "step": 14450 }, { "epoch": 0.16665930538992682, "grad_norm": 0.6194757223129272, "learning_rate": 1.4007784043603695e-05, "loss": 0.315138840675354, "step": 14460 }, { "epoch": 0.16724820752911385, "grad_norm": 0.6372362375259399, "learning_rate": 1.3999896947123778e-05, "loss": 0.3831063985824585, "step": 14470 }, { "epoch": 0.16783710966830087, "grad_norm": 0.7551998496055603, "learning_rate": 1.3992006887431395e-05, "loss": 0.3029595375061035, "step": 14480 }, { "epoch": 0.1684260118074879, "grad_norm": 0.6119477152824402, "learning_rate": 1.3984113870371676e-05, "loss": 0.29344136714935304, "step": 14490 }, { "epoch": 0.16901491394667492, "grad_norm": 0.4946797490119934, "learning_rate": 1.3976217901791947e-05, "loss": 0.3061702251434326, "step": 14500 }, { "epoch": 0.16960381608586192, "grad_norm": 0.7585649490356445, "learning_rate": 1.3968318987541714e-05, "loss": 0.34977221488952637, "step": 14510 }, { "epoch": 0.17019271822504894, "grad_norm": 0.6132660508155823, "learning_rate": 1.3960417133472666e-05, "loss": 0.35099096298217775, "step": 14520 }, { "epoch": 0.17078162036423597, "grad_norm": 0.5135638117790222, "learning_rate": 1.3952512345438681e-05, "loss": 0.28878734111785886, "step": 14530 }, { "epoch": 0.171370522503423, "grad_norm": 0.6353283524513245, "learning_rate": 1.3944604629295789e-05, "loss": 0.31544952392578124, "step": 14540 }, { "epoch": 0.17195942464261002, "grad_norm": 0.5589971542358398, "learning_rate": 1.3936693990902207e-05, "loss": 0.29819293022155763, "step": 14550 }, { "epoch": 0.17254832678179705, "grad_norm": 0.7383033633232117, "learning_rate": 1.3928780436118316e-05, "loss": 0.3017134666442871, "step": 14560 }, { "epoch": 0.17313722892098404, "grad_norm": 0.6932368874549866, "learning_rate": 1.3920863970806651e-05, "loss": 0.3218738079071045, "step": 14570 }, { "epoch": 0.17372613106017107, "grad_norm": 0.6316989660263062, "learning_rate": 1.3912944600831902e-05, "loss": 0.2877125024795532, "step": 14580 }, { "epoch": 0.1743150331993581, "grad_norm": 0.9135677218437195, "learning_rate": 1.3905022332060913e-05, "loss": 0.2999564647674561, "step": 14590 }, { "epoch": 0.17490393533854512, "grad_norm": 0.598606526851654, "learning_rate": 1.3897097170362683e-05, "loss": 0.2875926733016968, "step": 14600 }, { "epoch": 0.17549283747773214, "grad_norm": 0.6995345950126648, "learning_rate": 1.3889169121608342e-05, "loss": 0.28288590908050537, "step": 14610 }, { "epoch": 0.17608173961691917, "grad_norm": 0.7077192664146423, "learning_rate": 1.3881238191671166e-05, "loss": 0.30970277786254885, "step": 14620 }, { "epoch": 0.17667064175610617, "grad_norm": 0.5926223993301392, "learning_rate": 1.3873304386426569e-05, "loss": 0.3735161781311035, "step": 14630 }, { "epoch": 0.1772595438952932, "grad_norm": 0.571921706199646, "learning_rate": 1.3865367711752083e-05, "loss": 0.3219525575637817, "step": 14640 }, { "epoch": 0.17784844603448022, "grad_norm": 0.8765428066253662, "learning_rate": 1.3857428173527376e-05, "loss": 0.35245072841644287, "step": 14650 }, { "epoch": 0.17843734817366724, "grad_norm": 0.6428269743919373, "learning_rate": 1.3849485777634233e-05, "loss": 0.33241469860076905, "step": 14660 }, { "epoch": 0.17902625031285427, "grad_norm": 0.5542425513267517, "learning_rate": 1.3841540529956559e-05, "loss": 0.35813159942626954, "step": 14670 }, { "epoch": 0.1796151524520413, "grad_norm": 0.5967099070549011, "learning_rate": 1.3833592436380367e-05, "loss": 0.3234807014465332, "step": 14680 }, { "epoch": 0.1802040545912283, "grad_norm": 0.6039279699325562, "learning_rate": 1.3825641502793784e-05, "loss": 0.331186318397522, "step": 14690 }, { "epoch": 0.18079295673041532, "grad_norm": 0.8451281189918518, "learning_rate": 1.381768773508704e-05, "loss": 0.35905206203460693, "step": 14700 }, { "epoch": 0.18138185886960234, "grad_norm": 0.47460833191871643, "learning_rate": 1.3809731139152459e-05, "loss": 0.30673439502716066, "step": 14710 }, { "epoch": 0.18197076100878937, "grad_norm": 0.9559748768806458, "learning_rate": 1.380177172088446e-05, "loss": 0.34105839729309084, "step": 14720 }, { "epoch": 0.1825596631479764, "grad_norm": 0.7317312955856323, "learning_rate": 1.3793809486179566e-05, "loss": 0.3115206718444824, "step": 14730 }, { "epoch": 0.18314856528716342, "grad_norm": 0.8452470898628235, "learning_rate": 1.3785844440936373e-05, "loss": 0.30895228385925294, "step": 14740 }, { "epoch": 0.18373746742635041, "grad_norm": 0.4709519147872925, "learning_rate": 1.377787659105556e-05, "loss": 0.33412559032440187, "step": 14750 }, { "epoch": 0.18432636956553744, "grad_norm": 0.7033078670501709, "learning_rate": 1.3769905942439895e-05, "loss": 0.3245211362838745, "step": 14760 }, { "epoch": 0.18491527170472447, "grad_norm": 0.49945709109306335, "learning_rate": 1.3761932500994207e-05, "loss": 0.3372659683227539, "step": 14770 }, { "epoch": 0.1855041738439115, "grad_norm": 0.775993287563324, "learning_rate": 1.3753956272625391e-05, "loss": 0.3612861633300781, "step": 14780 }, { "epoch": 0.18609307598309852, "grad_norm": 0.522375762462616, "learning_rate": 1.3745977263242426e-05, "loss": 0.3417446851730347, "step": 14790 }, { "epoch": 0.18668197812228554, "grad_norm": 0.5898303389549255, "learning_rate": 1.3737995478756336e-05, "loss": 0.3399900674819946, "step": 14800 }, { "epoch": 0.18727088026147254, "grad_norm": 0.6491979956626892, "learning_rate": 1.3730010925080201e-05, "loss": 0.30159411430358884, "step": 14810 }, { "epoch": 0.18785978240065956, "grad_norm": 0.5290111303329468, "learning_rate": 1.3722023608129157e-05, "loss": 0.32310709953308103, "step": 14820 }, { "epoch": 0.1884486845398466, "grad_norm": 0.5479434728622437, "learning_rate": 1.3714033533820389e-05, "loss": 0.3528861999511719, "step": 14830 }, { "epoch": 0.18903758667903361, "grad_norm": 0.5175791382789612, "learning_rate": 1.3706040708073116e-05, "loss": 0.3028521776199341, "step": 14840 }, { "epoch": 0.18962648881822064, "grad_norm": 0.571006715297699, "learning_rate": 1.3698045136808602e-05, "loss": 0.30765423774719236, "step": 14850 }, { "epoch": 0.19021539095740767, "grad_norm": 0.6754320859909058, "learning_rate": 1.3690046825950147e-05, "loss": 0.28174893856048583, "step": 14860 }, { "epoch": 0.19080429309659466, "grad_norm": 0.6431025266647339, "learning_rate": 1.3682045781423076e-05, "loss": 0.317968487739563, "step": 14870 }, { "epoch": 0.1913931952357817, "grad_norm": 0.5244066715240479, "learning_rate": 1.3674042009154737e-05, "loss": 0.3008192300796509, "step": 14880 }, { "epoch": 0.1919820973749687, "grad_norm": 0.6614280343055725, "learning_rate": 1.366603551507451e-05, "loss": 0.3011381387710571, "step": 14890 }, { "epoch": 0.19257099951415574, "grad_norm": 0.7215622067451477, "learning_rate": 1.3658026305113776e-05, "loss": 0.32603633403778076, "step": 14900 }, { "epoch": 0.19315990165334276, "grad_norm": 0.5607608556747437, "learning_rate": 1.3650014385205937e-05, "loss": 0.30038487911224365, "step": 14910 }, { "epoch": 0.1937488037925298, "grad_norm": 0.5085550546646118, "learning_rate": 1.3641999761286402e-05, "loss": 0.3422208786010742, "step": 14920 }, { "epoch": 0.1943377059317168, "grad_norm": 0.5688707232475281, "learning_rate": 1.3633982439292585e-05, "loss": 0.2923550844192505, "step": 14930 }, { "epoch": 0.1949266080709038, "grad_norm": 0.6356753706932068, "learning_rate": 1.3625962425163892e-05, "loss": 0.3725019693374634, "step": 14940 }, { "epoch": 0.19551551021009084, "grad_norm": 0.6180427670478821, "learning_rate": 1.3617939724841728e-05, "loss": 0.3605834722518921, "step": 14950 }, { "epoch": 0.19610441234927786, "grad_norm": 0.5667227506637573, "learning_rate": 1.3609914344269494e-05, "loss": 0.3376338958740234, "step": 14960 }, { "epoch": 0.1966933144884649, "grad_norm": 0.4982045888900757, "learning_rate": 1.3601886289392562e-05, "loss": 0.3615573883056641, "step": 14970 }, { "epoch": 0.1972822166276519, "grad_norm": 0.5246447324752808, "learning_rate": 1.3593855566158293e-05, "loss": 0.2814157247543335, "step": 14980 }, { "epoch": 0.1978711187668389, "grad_norm": 0.7139537334442139, "learning_rate": 1.3585822180516031e-05, "loss": 0.315099310874939, "step": 14990 }, { "epoch": 0.19846002090602594, "grad_norm": 0.5879722833633423, "learning_rate": 1.3577786138417083e-05, "loss": 0.33630435466766356, "step": 15000 }, { "epoch": 0.19904892304521296, "grad_norm": 0.5788151025772095, "learning_rate": 1.356974744581473e-05, "loss": 0.30466248989105227, "step": 15010 }, { "epoch": 0.1996378251844, "grad_norm": 0.5564064383506775, "learning_rate": 1.356170610866421e-05, "loss": 0.2993898391723633, "step": 15020 }, { "epoch": 0.200226727323587, "grad_norm": 0.6146450638771057, "learning_rate": 1.3553662132922729e-05, "loss": 0.30703551769256593, "step": 15030 }, { "epoch": 0.20081562946277404, "grad_norm": 0.6837103962898254, "learning_rate": 1.3545615524549438e-05, "loss": 0.34827046394348143, "step": 15040 }, { "epoch": 0.20140453160196103, "grad_norm": 0.6376634240150452, "learning_rate": 1.3537566289505444e-05, "loss": 0.3279841661453247, "step": 15050 }, { "epoch": 0.20199343374114806, "grad_norm": 0.5109546780586243, "learning_rate": 1.3529514433753802e-05, "loss": 0.3150515079498291, "step": 15060 }, { "epoch": 0.20258233588033508, "grad_norm": 0.5345185399055481, "learning_rate": 1.3521459963259505e-05, "loss": 0.3078217267990112, "step": 15070 }, { "epoch": 0.2031712380195221, "grad_norm": 0.6460902094841003, "learning_rate": 1.351340288398948e-05, "loss": 0.33833112716674807, "step": 15080 }, { "epoch": 0.20376014015870914, "grad_norm": 0.631264865398407, "learning_rate": 1.350534320191259e-05, "loss": 0.3100293636322021, "step": 15090 }, { "epoch": 0.20434904229789616, "grad_norm": 0.645926296710968, "learning_rate": 1.3497280922999633e-05, "loss": 0.32422258853912356, "step": 15100 }, { "epoch": 0.20493794443708316, "grad_norm": 0.7345241904258728, "learning_rate": 1.348921605322332e-05, "loss": 0.35910515785217284, "step": 15110 }, { "epoch": 0.20552684657627018, "grad_norm": 0.6160382032394409, "learning_rate": 1.3481148598558282e-05, "loss": 0.3256230354309082, "step": 15120 }, { "epoch": 0.2061157487154572, "grad_norm": 0.6363213658332825, "learning_rate": 1.3473078564981077e-05, "loss": 0.3336115837097168, "step": 15130 }, { "epoch": 0.20670465085464423, "grad_norm": 0.7026886343955994, "learning_rate": 1.3465005958470158e-05, "loss": 0.3084343671798706, "step": 15140 }, { "epoch": 0.20729355299383126, "grad_norm": 0.645653247833252, "learning_rate": 1.3456930785005895e-05, "loss": 0.39668023586273193, "step": 15150 }, { "epoch": 0.20788245513301826, "grad_norm": 0.5513941645622253, "learning_rate": 1.3448853050570551e-05, "loss": 0.32083632946014407, "step": 15160 }, { "epoch": 0.20847135727220528, "grad_norm": 0.605783224105835, "learning_rate": 1.3440772761148298e-05, "loss": 0.3324552536010742, "step": 15170 }, { "epoch": 0.2090602594113923, "grad_norm": 0.6465606093406677, "learning_rate": 1.3432689922725186e-05, "loss": 0.35427165031433105, "step": 15180 }, { "epoch": 0.20964916155057933, "grad_norm": 0.6873339414596558, "learning_rate": 1.3424604541289167e-05, "loss": 0.3117310285568237, "step": 15190 }, { "epoch": 0.21023806368976636, "grad_norm": 0.6451834440231323, "learning_rate": 1.3416516622830073e-05, "loss": 0.34201071262359617, "step": 15200 }, { "epoch": 0.21082696582895338, "grad_norm": 0.5077345967292786, "learning_rate": 1.3408426173339605e-05, "loss": 0.34638681411743166, "step": 15210 }, { "epoch": 0.21141586796814038, "grad_norm": 0.6072590351104736, "learning_rate": 1.3400333198811355e-05, "loss": 0.3497732639312744, "step": 15220 }, { "epoch": 0.2120047701073274, "grad_norm": 0.7291898131370544, "learning_rate": 1.3392237705240774e-05, "loss": 0.3037578105926514, "step": 15230 }, { "epoch": 0.21259367224651443, "grad_norm": 0.7613815665245056, "learning_rate": 1.3384139698625185e-05, "loss": 0.32613070011138917, "step": 15240 }, { "epoch": 0.21318257438570146, "grad_norm": 0.4734254479408264, "learning_rate": 1.337603918496377e-05, "loss": 0.2969614267349243, "step": 15250 }, { "epoch": 0.21377147652488848, "grad_norm": 0.6860185861587524, "learning_rate": 1.3367936170257574e-05, "loss": 0.35388247966766356, "step": 15260 }, { "epoch": 0.2143603786640755, "grad_norm": 0.6701116561889648, "learning_rate": 1.3359830660509488e-05, "loss": 0.3040478229522705, "step": 15270 }, { "epoch": 0.2149492808032625, "grad_norm": 0.5193243622779846, "learning_rate": 1.3351722661724247e-05, "loss": 0.3024127960205078, "step": 15280 }, { "epoch": 0.21553818294244953, "grad_norm": 0.6716238260269165, "learning_rate": 1.3343612179908444e-05, "loss": 0.33761584758758545, "step": 15290 }, { "epoch": 0.21612708508163656, "grad_norm": 0.562988817691803, "learning_rate": 1.3335499221070502e-05, "loss": 0.28348584175109864, "step": 15300 }, { "epoch": 0.21671598722082358, "grad_norm": 0.69050532579422, "learning_rate": 1.3327383791220683e-05, "loss": 0.3170329570770264, "step": 15310 }, { "epoch": 0.2173048893600106, "grad_norm": 0.6139616966247559, "learning_rate": 1.3319265896371072e-05, "loss": 0.2887885093688965, "step": 15320 }, { "epoch": 0.21789379149919763, "grad_norm": 0.5025825500488281, "learning_rate": 1.331114554253559e-05, "loss": 0.3402873039245605, "step": 15330 }, { "epoch": 0.21848269363838463, "grad_norm": 0.532096803188324, "learning_rate": 1.3303022735729978e-05, "loss": 0.3245500087738037, "step": 15340 }, { "epoch": 0.21907159577757165, "grad_norm": 0.4834073483943939, "learning_rate": 1.3294897481971783e-05, "loss": 0.31464223861694335, "step": 15350 }, { "epoch": 0.21966049791675868, "grad_norm": 0.6754441261291504, "learning_rate": 1.3286769787280382e-05, "loss": 0.3563635587692261, "step": 15360 }, { "epoch": 0.2202494000559457, "grad_norm": 0.6337766051292419, "learning_rate": 1.3278639657676947e-05, "loss": 0.31983964443206786, "step": 15370 }, { "epoch": 0.22083830219513273, "grad_norm": 0.4947168827056885, "learning_rate": 1.327050709918446e-05, "loss": 0.3301743984222412, "step": 15380 }, { "epoch": 0.22142720433431975, "grad_norm": 0.5565149784088135, "learning_rate": 1.3262372117827703e-05, "loss": 0.3595422744750977, "step": 15390 }, { "epoch": 0.22201610647350675, "grad_norm": 0.6157445311546326, "learning_rate": 1.3254234719633245e-05, "loss": 0.3202836990356445, "step": 15400 }, { "epoch": 0.22260500861269378, "grad_norm": 0.4324851632118225, "learning_rate": 1.3246094910629457e-05, "loss": 0.383900785446167, "step": 15410 }, { "epoch": 0.2231939107518808, "grad_norm": 0.6441844701766968, "learning_rate": 1.3237952696846485e-05, "loss": 0.27836673259735106, "step": 15420 }, { "epoch": 0.22378281289106783, "grad_norm": 0.6665327548980713, "learning_rate": 1.3229808084316269e-05, "loss": 0.32993507385253906, "step": 15430 }, { "epoch": 0.22437171503025485, "grad_norm": 0.5131520628929138, "learning_rate": 1.3221661079072514e-05, "loss": 0.3440540313720703, "step": 15440 }, { "epoch": 0.22496061716944188, "grad_norm": 0.48847508430480957, "learning_rate": 1.32135116871507e-05, "loss": 0.313602089881897, "step": 15450 }, { "epoch": 0.22554951930862888, "grad_norm": 0.5346674919128418, "learning_rate": 1.3205359914588086e-05, "loss": 0.3202857494354248, "step": 15460 }, { "epoch": 0.2261384214478159, "grad_norm": 0.8720782995223999, "learning_rate": 1.3197205767423675e-05, "loss": 0.31865777969360354, "step": 15470 }, { "epoch": 0.22672732358700293, "grad_norm": 0.6408674120903015, "learning_rate": 1.3189049251698247e-05, "loss": 0.31161961555480955, "step": 15480 }, { "epoch": 0.22731622572618995, "grad_norm": 0.7599413990974426, "learning_rate": 1.318089037345433e-05, "loss": 0.34831762313842773, "step": 15490 }, { "epoch": 0.22790512786537698, "grad_norm": 0.5906134843826294, "learning_rate": 1.31727291387362e-05, "loss": 0.3009833097457886, "step": 15500 }, { "epoch": 0.228494030004564, "grad_norm": 0.5671749711036682, "learning_rate": 1.316456555358988e-05, "loss": 0.31406407356262206, "step": 15510 }, { "epoch": 0.229082932143751, "grad_norm": 0.5074796676635742, "learning_rate": 1.3156399624063139e-05, "loss": 0.33066582679748535, "step": 15520 }, { "epoch": 0.22967183428293803, "grad_norm": 0.5187433958053589, "learning_rate": 1.3148231356205478e-05, "loss": 0.3399573802947998, "step": 15530 }, { "epoch": 0.23026073642212505, "grad_norm": 0.5600962042808533, "learning_rate": 1.314006075606813e-05, "loss": 0.3507091045379639, "step": 15540 }, { "epoch": 0.23084963856131208, "grad_norm": 0.41015973687171936, "learning_rate": 1.3131887829704052e-05, "loss": 0.34493675231933596, "step": 15550 }, { "epoch": 0.2314385407004991, "grad_norm": 0.632327139377594, "learning_rate": 1.3123712583167938e-05, "loss": 0.3291771411895752, "step": 15560 }, { "epoch": 0.23202744283968613, "grad_norm": 0.6004683375358582, "learning_rate": 1.311553502251619e-05, "loss": 0.3275893688201904, "step": 15570 }, { "epoch": 0.23261634497887312, "grad_norm": 0.7258276343345642, "learning_rate": 1.3107355153806922e-05, "loss": 0.32151844501495364, "step": 15580 }, { "epoch": 0.23320524711806015, "grad_norm": 0.5072625279426575, "learning_rate": 1.3099172983099966e-05, "loss": 0.34006102085113527, "step": 15590 }, { "epoch": 0.23379414925724717, "grad_norm": 0.6049047112464905, "learning_rate": 1.3090988516456856e-05, "loss": 0.31402289867401123, "step": 15600 }, { "epoch": 0.2343830513964342, "grad_norm": 0.8434813022613525, "learning_rate": 1.3082801759940824e-05, "loss": 0.32865850925445556, "step": 15610 }, { "epoch": 0.23497195353562123, "grad_norm": 0.4994066655635834, "learning_rate": 1.3074612719616805e-05, "loss": 0.32472968101501465, "step": 15620 }, { "epoch": 0.23556085567480825, "grad_norm": 0.519798219203949, "learning_rate": 1.3066421401551415e-05, "loss": 0.2838024139404297, "step": 15630 }, { "epoch": 0.23614975781399525, "grad_norm": 0.506880521774292, "learning_rate": 1.3058227811812974e-05, "loss": 0.3639389514923096, "step": 15640 }, { "epoch": 0.23673865995318227, "grad_norm": 0.5432206988334656, "learning_rate": 1.3050031956471468e-05, "loss": 0.3011049509048462, "step": 15650 }, { "epoch": 0.2373275620923693, "grad_norm": 0.5722271800041199, "learning_rate": 1.3041833841598568e-05, "loss": 0.2993990421295166, "step": 15660 }, { "epoch": 0.23791646423155632, "grad_norm": 0.540938138961792, "learning_rate": 1.3033633473267627e-05, "loss": 0.34339611530303954, "step": 15670 }, { "epoch": 0.23850536637074335, "grad_norm": 0.6538553833961487, "learning_rate": 1.302543085755365e-05, "loss": 0.3585189342498779, "step": 15680 }, { "epoch": 0.23909426850993037, "grad_norm": 0.5708355903625488, "learning_rate": 1.3017226000533323e-05, "loss": 0.34968080520629885, "step": 15690 }, { "epoch": 0.23968317064911737, "grad_norm": 0.5555328726768494, "learning_rate": 1.3009018908284987e-05, "loss": 0.3280498743057251, "step": 15700 }, { "epoch": 0.2402720727883044, "grad_norm": 0.5042557716369629, "learning_rate": 1.3000809586888635e-05, "loss": 0.34505133628845214, "step": 15710 }, { "epoch": 0.24086097492749142, "grad_norm": 0.6462178230285645, "learning_rate": 1.2992598042425915e-05, "loss": 0.3351136207580566, "step": 15720 }, { "epoch": 0.24144987706667845, "grad_norm": 0.5288575887680054, "learning_rate": 1.2984384280980122e-05, "loss": 0.36388065814971926, "step": 15730 }, { "epoch": 0.24203877920586547, "grad_norm": 0.8081417679786682, "learning_rate": 1.2976168308636197e-05, "loss": 0.3234818935394287, "step": 15740 }, { "epoch": 0.2426276813450525, "grad_norm": 0.5891144871711731, "learning_rate": 1.2967950131480706e-05, "loss": 0.32728726863861085, "step": 15750 }, { "epoch": 0.2432165834842395, "grad_norm": 0.6240312457084656, "learning_rate": 1.2959729755601865e-05, "loss": 0.3483387231826782, "step": 15760 }, { "epoch": 0.24380548562342652, "grad_norm": 0.47667357325553894, "learning_rate": 1.2951507187089507e-05, "loss": 0.3625577688217163, "step": 15770 }, { "epoch": 0.24439438776261355, "grad_norm": 0.6927791833877563, "learning_rate": 1.2943282432035092e-05, "loss": 0.33784897327423097, "step": 15780 }, { "epoch": 0.24498328990180057, "grad_norm": 0.7186523079872131, "learning_rate": 1.2935055496531708e-05, "loss": 0.32825446128845215, "step": 15790 }, { "epoch": 0.2455721920409876, "grad_norm": 0.8077899813652039, "learning_rate": 1.2926826386674043e-05, "loss": 0.3642729759216309, "step": 15800 }, { "epoch": 0.24616109418017462, "grad_norm": 0.38159313797950745, "learning_rate": 1.2918595108558412e-05, "loss": 0.3254523754119873, "step": 15810 }, { "epoch": 0.24674999631936162, "grad_norm": 0.4513606131076813, "learning_rate": 1.2910361668282718e-05, "loss": 0.34708523750305176, "step": 15820 }, { "epoch": 0.24733889845854864, "grad_norm": 0.38990795612335205, "learning_rate": 1.2902126071946488e-05, "loss": 0.3266537427902222, "step": 15830 }, { "epoch": 0.24792780059773567, "grad_norm": 0.5906530022621155, "learning_rate": 1.2893888325650832e-05, "loss": 0.3588040590286255, "step": 15840 }, { "epoch": 0.2485167027369227, "grad_norm": 0.5619174838066101, "learning_rate": 1.2885648435498446e-05, "loss": 0.33747618198394774, "step": 15850 }, { "epoch": 0.24910560487610972, "grad_norm": 0.6912819147109985, "learning_rate": 1.2877406407593636e-05, "loss": 0.3294116497039795, "step": 15860 }, { "epoch": 0.24969450701529675, "grad_norm": 0.6145522594451904, "learning_rate": 1.2869162248042274e-05, "loss": 0.29547278881072997, "step": 15870 }, { "epoch": 0.25028340915448377, "grad_norm": 0.44174790382385254, "learning_rate": 1.2860915962951812e-05, "loss": 0.3487529993057251, "step": 15880 }, { "epoch": 0.25087231129367077, "grad_norm": 0.48937201499938965, "learning_rate": 1.2852667558431288e-05, "loss": 0.312595796585083, "step": 15890 }, { "epoch": 0.2514612134328578, "grad_norm": 0.5736889839172363, "learning_rate": 1.2844417040591301e-05, "loss": 0.32662181854248046, "step": 15900 }, { "epoch": 0.2520501155720448, "grad_norm": 0.5081770420074463, "learning_rate": 1.2836164415544017e-05, "loss": 0.36627347469329835, "step": 15910 }, { "epoch": 0.2526390177112318, "grad_norm": 0.5055372714996338, "learning_rate": 1.2827909689403164e-05, "loss": 0.3287820339202881, "step": 15920 }, { "epoch": 0.25322791985041887, "grad_norm": 0.5581505298614502, "learning_rate": 1.2819652868284027e-05, "loss": 0.3149190187454224, "step": 15930 }, { "epoch": 0.25381682198960587, "grad_norm": 0.5078492760658264, "learning_rate": 1.2811393958303442e-05, "loss": 0.3264275074005127, "step": 15940 }, { "epoch": 0.2544057241287929, "grad_norm": 0.6578114628791809, "learning_rate": 1.2803132965579793e-05, "loss": 0.35132904052734376, "step": 15950 }, { "epoch": 0.2549946262679799, "grad_norm": 0.4701768457889557, "learning_rate": 1.2794869896233007e-05, "loss": 0.3109602689743042, "step": 15960 }, { "epoch": 0.2555835284071669, "grad_norm": 0.4528977870941162, "learning_rate": 1.2786604756384548e-05, "loss": 0.31858205795288086, "step": 15970 }, { "epoch": 0.25617243054635397, "grad_norm": 0.41068124771118164, "learning_rate": 1.2778337552157416e-05, "loss": 0.35628154277801516, "step": 15980 }, { "epoch": 0.25676133268554097, "grad_norm": 0.4542391300201416, "learning_rate": 1.277006828967614e-05, "loss": 0.35854520797729494, "step": 15990 }, { "epoch": 0.257350234824728, "grad_norm": 0.44235438108444214, "learning_rate": 1.2761796975066774e-05, "loss": 0.3069314956665039, "step": 16000 }, { "epoch": 0.257939136963915, "grad_norm": 0.5040526986122131, "learning_rate": 1.2753523614456889e-05, "loss": 0.2986379861831665, "step": 16010 }, { "epoch": 0.25852803910310207, "grad_norm": 0.6338397264480591, "learning_rate": 1.2745248213975573e-05, "loss": 0.320386266708374, "step": 16020 }, { "epoch": 0.25911694124228907, "grad_norm": 0.5334436297416687, "learning_rate": 1.2736970779753435e-05, "loss": 0.33429827690124514, "step": 16030 }, { "epoch": 0.25970584338147606, "grad_norm": 0.6764389872550964, "learning_rate": 1.2728691317922573e-05, "loss": 0.3266124725341797, "step": 16040 }, { "epoch": 0.2602947455206631, "grad_norm": 0.526999294757843, "learning_rate": 1.2720409834616597e-05, "loss": 0.31841602325439455, "step": 16050 }, { "epoch": 0.2608836476598501, "grad_norm": 0.5762807130813599, "learning_rate": 1.2712126335970615e-05, "loss": 0.3842127323150635, "step": 16060 }, { "epoch": 0.26147254979903717, "grad_norm": 0.657174289226532, "learning_rate": 1.2703840828121234e-05, "loss": 0.41879024505615237, "step": 16070 }, { "epoch": 0.26206145193822417, "grad_norm": 0.5585598349571228, "learning_rate": 1.2695553317206531e-05, "loss": 0.3304882526397705, "step": 16080 }, { "epoch": 0.26265035407741116, "grad_norm": 0.4516439437866211, "learning_rate": 1.2687263809366082e-05, "loss": 0.3190563678741455, "step": 16090 }, { "epoch": 0.2632392562165982, "grad_norm": 0.5465164184570312, "learning_rate": 1.2678972310740942e-05, "loss": 0.31944942474365234, "step": 16100 }, { "epoch": 0.2638281583557852, "grad_norm": 0.4860529601573944, "learning_rate": 1.2670678827473632e-05, "loss": 0.3163974046707153, "step": 16110 }, { "epoch": 0.26441706049497227, "grad_norm": 0.5473213195800781, "learning_rate": 1.2662383365708153e-05, "loss": 0.350046706199646, "step": 16120 }, { "epoch": 0.26500596263415926, "grad_norm": 0.9127334356307983, "learning_rate": 1.2654085931589965e-05, "loss": 0.32546653747558596, "step": 16130 }, { "epoch": 0.26559486477334626, "grad_norm": 0.78261798620224, "learning_rate": 1.2645786531265993e-05, "loss": 0.3359572172164917, "step": 16140 }, { "epoch": 0.2661837669125333, "grad_norm": 0.6200259923934937, "learning_rate": 1.2637485170884614e-05, "loss": 0.3693458318710327, "step": 16150 }, { "epoch": 0.2667726690517203, "grad_norm": 0.5161575675010681, "learning_rate": 1.262918185659566e-05, "loss": 0.3222073554992676, "step": 16160 }, { "epoch": 0.26736157119090737, "grad_norm": 0.6196975708007812, "learning_rate": 1.2620876594550417e-05, "loss": 0.36136040687561033, "step": 16170 }, { "epoch": 0.26795047333009436, "grad_norm": 0.5832588076591492, "learning_rate": 1.2612569390901597e-05, "loss": 0.3326419353485107, "step": 16180 }, { "epoch": 0.2685393754692814, "grad_norm": 0.4247804284095764, "learning_rate": 1.2604260251803372e-05, "loss": 0.3233131647109985, "step": 16190 }, { "epoch": 0.2691282776084684, "grad_norm": 0.841849684715271, "learning_rate": 1.259594918341133e-05, "loss": 0.35884091854095457, "step": 16200 }, { "epoch": 0.2697171797476554, "grad_norm": 0.5580234527587891, "learning_rate": 1.2587636191882497e-05, "loss": 0.36801135540008545, "step": 16210 }, { "epoch": 0.27030608188684246, "grad_norm": 0.5727989077568054, "learning_rate": 1.257932128337532e-05, "loss": 0.3088529586791992, "step": 16220 }, { "epoch": 0.27089498402602946, "grad_norm": 0.5924689769744873, "learning_rate": 1.257100446404967e-05, "loss": 0.3326411724090576, "step": 16230 }, { "epoch": 0.2714838861652165, "grad_norm": 0.46880319714546204, "learning_rate": 1.2562685740066831e-05, "loss": 0.33724079132080076, "step": 16240 }, { "epoch": 0.2720727883044035, "grad_norm": 0.6654885411262512, "learning_rate": 1.2554365117589495e-05, "loss": 0.35315115451812745, "step": 16250 }, { "epoch": 0.2726616904435905, "grad_norm": 0.5040119290351868, "learning_rate": 1.2546042602781769e-05, "loss": 0.34661355018615725, "step": 16260 }, { "epoch": 0.27325059258277756, "grad_norm": 0.5504894852638245, "learning_rate": 1.2537718201809157e-05, "loss": 0.33851089477539065, "step": 16270 }, { "epoch": 0.27383949472196456, "grad_norm": 0.490853488445282, "learning_rate": 1.2529391920838554e-05, "loss": 0.35587081909179685, "step": 16280 }, { "epoch": 0.2744283968611516, "grad_norm": 0.4678669273853302, "learning_rate": 1.2521063766038259e-05, "loss": 0.3260726690292358, "step": 16290 }, { "epoch": 0.2750172990003386, "grad_norm": 0.6089847683906555, "learning_rate": 1.2512733743577952e-05, "loss": 0.34575154781341555, "step": 16300 }, { "epoch": 0.27560620113952566, "grad_norm": 0.4146714210510254, "learning_rate": 1.25044018596287e-05, "loss": 0.32469797134399414, "step": 16310 }, { "epoch": 0.27619510327871266, "grad_norm": 0.5187833905220032, "learning_rate": 1.2496068120362943e-05, "loss": 0.3550767183303833, "step": 16320 }, { "epoch": 0.27678400541789966, "grad_norm": 0.4485473334789276, "learning_rate": 1.2487732531954509e-05, "loss": 0.3391078233718872, "step": 16330 }, { "epoch": 0.2773729075570867, "grad_norm": 0.5659913420677185, "learning_rate": 1.247939510057858e-05, "loss": 0.3140303134918213, "step": 16340 }, { "epoch": 0.2779618096962737, "grad_norm": 0.6110917925834656, "learning_rate": 1.2471055832411707e-05, "loss": 0.336813759803772, "step": 16350 }, { "epoch": 0.27855071183546076, "grad_norm": 0.6013612151145935, "learning_rate": 1.2462714733631816e-05, "loss": 0.37469313144683836, "step": 16360 }, { "epoch": 0.27913961397464776, "grad_norm": 0.45579367876052856, "learning_rate": 1.2454371810418169e-05, "loss": 0.3188345909118652, "step": 16370 }, { "epoch": 0.27972851611383476, "grad_norm": 0.41055530309677124, "learning_rate": 1.2446027068951392e-05, "loss": 0.2757137298583984, "step": 16380 }, { "epoch": 0.2803174182530218, "grad_norm": 0.549924373626709, "learning_rate": 1.2437680515413458e-05, "loss": 0.3451666831970215, "step": 16390 }, { "epoch": 0.2809063203922088, "grad_norm": 0.47450706362724304, "learning_rate": 1.242933215598767e-05, "loss": 0.2893789529800415, "step": 16400 }, { "epoch": 0.28149522253139586, "grad_norm": 0.611369252204895, "learning_rate": 1.2420981996858688e-05, "loss": 0.3471691608428955, "step": 16410 }, { "epoch": 0.28208412467058286, "grad_norm": 0.6187147498130798, "learning_rate": 1.2412630044212488e-05, "loss": 0.36530489921569825, "step": 16420 }, { "epoch": 0.2826730268097699, "grad_norm": 0.4711047112941742, "learning_rate": 1.240427630423639e-05, "loss": 0.3298162937164307, "step": 16430 }, { "epoch": 0.2832619289489569, "grad_norm": 0.4564912021160126, "learning_rate": 1.239592078311902e-05, "loss": 0.3189046859741211, "step": 16440 }, { "epoch": 0.2838508310881439, "grad_norm": 0.4747132658958435, "learning_rate": 1.238756348705034e-05, "loss": 0.3266399621963501, "step": 16450 }, { "epoch": 0.28443973322733096, "grad_norm": 0.5201255083084106, "learning_rate": 1.237920442222162e-05, "loss": 0.3469388008117676, "step": 16460 }, { "epoch": 0.28502863536651796, "grad_norm": 0.5912818312644958, "learning_rate": 1.237084359482544e-05, "loss": 0.32522852420806886, "step": 16470 }, { "epoch": 0.285617537505705, "grad_norm": 0.4593675136566162, "learning_rate": 1.2362481011055684e-05, "loss": 0.33578855991363527, "step": 16480 }, { "epoch": 0.286206439644892, "grad_norm": 0.4641585648059845, "learning_rate": 1.2354116677107545e-05, "loss": 0.3735449552536011, "step": 16490 }, { "epoch": 0.286795341784079, "grad_norm": 0.558846652507782, "learning_rate": 1.2345750599177505e-05, "loss": 0.3338179111480713, "step": 16500 }, { "epoch": 0.28738424392326606, "grad_norm": 0.41969624161720276, "learning_rate": 1.2337382783463339e-05, "loss": 0.3418455123901367, "step": 16510 }, { "epoch": 0.28797314606245306, "grad_norm": 0.7156524062156677, "learning_rate": 1.232901323616411e-05, "loss": 0.3216254234313965, "step": 16520 }, { "epoch": 0.2885620482016401, "grad_norm": 0.5536055564880371, "learning_rate": 1.232064196348017e-05, "loss": 0.3830761194229126, "step": 16530 }, { "epoch": 0.2891509503408271, "grad_norm": 0.5426992177963257, "learning_rate": 1.2312268971613133e-05, "loss": 0.3287364959716797, "step": 16540 }, { "epoch": 0.28973985248001416, "grad_norm": 0.5782346129417419, "learning_rate": 1.2303894266765908e-05, "loss": 0.3426827907562256, "step": 16550 }, { "epoch": 0.29032875461920116, "grad_norm": 0.44015419483184814, "learning_rate": 1.2295517855142655e-05, "loss": 0.3044412851333618, "step": 16560 }, { "epoch": 0.29091765675838815, "grad_norm": 0.509355902671814, "learning_rate": 1.2287139742948813e-05, "loss": 0.3584832429885864, "step": 16570 }, { "epoch": 0.2915065588975752, "grad_norm": 0.5864397287368774, "learning_rate": 1.2278759936391067e-05, "loss": 0.3147047758102417, "step": 16580 }, { "epoch": 0.2920954610367622, "grad_norm": 0.42626631259918213, "learning_rate": 1.2270378441677367e-05, "loss": 0.3563106060028076, "step": 16590 }, { "epoch": 0.29268436317594926, "grad_norm": 0.4560808837413788, "learning_rate": 1.226199526501691e-05, "loss": 0.3468355178833008, "step": 16600 }, { "epoch": 0.29327326531513626, "grad_norm": 0.5296052694320679, "learning_rate": 1.2253610412620138e-05, "loss": 0.34153497219085693, "step": 16610 }, { "epoch": 0.29386216745432325, "grad_norm": 0.5143585801124573, "learning_rate": 1.2245223890698736e-05, "loss": 0.35652337074279783, "step": 16620 }, { "epoch": 0.2944510695935103, "grad_norm": 0.5356563329696655, "learning_rate": 1.2236835705465629e-05, "loss": 0.3054612159729004, "step": 16630 }, { "epoch": 0.2950399717326973, "grad_norm": 0.39650893211364746, "learning_rate": 1.2228445863134965e-05, "loss": 0.3245067596435547, "step": 16640 }, { "epoch": 0.29562887387188436, "grad_norm": 0.521914541721344, "learning_rate": 1.2220054369922132e-05, "loss": 0.3535057544708252, "step": 16650 }, { "epoch": 0.29621777601107135, "grad_norm": 0.5392968058586121, "learning_rate": 1.2211661232043728e-05, "loss": 0.32908480167388915, "step": 16660 }, { "epoch": 0.2968066781502584, "grad_norm": 0.5883451700210571, "learning_rate": 1.2203266455717582e-05, "loss": 0.32812039852142333, "step": 16670 }, { "epoch": 0.2973955802894454, "grad_norm": 0.4582393765449524, "learning_rate": 1.2194870047162724e-05, "loss": 0.31550276279449463, "step": 16680 }, { "epoch": 0.2979844824286324, "grad_norm": 0.5869547128677368, "learning_rate": 1.2186472012599406e-05, "loss": 0.3641475200653076, "step": 16690 }, { "epoch": 0.29857338456781946, "grad_norm": 0.4726334810256958, "learning_rate": 1.2178072358249078e-05, "loss": 0.33348019123077394, "step": 16700 }, { "epoch": 0.29916228670700645, "grad_norm": 0.5300145745277405, "learning_rate": 1.2169671090334385e-05, "loss": 0.323223614692688, "step": 16710 }, { "epoch": 0.2997511888461935, "grad_norm": 0.6471881866455078, "learning_rate": 1.2161268215079178e-05, "loss": 0.32628259658813474, "step": 16720 }, { "epoch": 0.3003400909853805, "grad_norm": 0.501270055770874, "learning_rate": 1.2152863738708491e-05, "loss": 0.3274534225463867, "step": 16730 }, { "epoch": 0.3009289931245675, "grad_norm": 0.41018274426460266, "learning_rate": 1.214445766744855e-05, "loss": 0.3258399963378906, "step": 16740 }, { "epoch": 0.30151789526375455, "grad_norm": 0.514765739440918, "learning_rate": 1.2136050007526752e-05, "loss": 0.31792287826538085, "step": 16750 }, { "epoch": 0.30210679740294155, "grad_norm": 0.6449528932571411, "learning_rate": 1.2127640765171687e-05, "loss": 0.34872872829437257, "step": 16760 }, { "epoch": 0.3026956995421286, "grad_norm": 0.5639067888259888, "learning_rate": 1.2119229946613105e-05, "loss": 0.2976230621337891, "step": 16770 }, { "epoch": 0.3032846016813156, "grad_norm": 0.5355367064476013, "learning_rate": 1.2110817558081922e-05, "loss": 0.3014484405517578, "step": 16780 }, { "epoch": 0.30387350382050266, "grad_norm": 0.43285930156707764, "learning_rate": 1.2102403605810232e-05, "loss": 0.31289350986480713, "step": 16790 }, { "epoch": 0.30446240595968965, "grad_norm": 0.4627956449985504, "learning_rate": 1.2093988096031268e-05, "loss": 0.32175915241241454, "step": 16800 }, { "epoch": 0.30505130809887665, "grad_norm": 0.6765831708908081, "learning_rate": 1.2085571034979432e-05, "loss": 0.31900882720947266, "step": 16810 }, { "epoch": 0.3056402102380637, "grad_norm": 0.6810938715934753, "learning_rate": 1.2077152428890268e-05, "loss": 0.3285127401351929, "step": 16820 }, { "epoch": 0.3062291123772507, "grad_norm": 0.536017894744873, "learning_rate": 1.206873228400047e-05, "loss": 0.31753971576690676, "step": 16830 }, { "epoch": 0.30681801451643775, "grad_norm": 0.3981994688510895, "learning_rate": 1.2060310606547865e-05, "loss": 0.3056586265563965, "step": 16840 }, { "epoch": 0.30740691665562475, "grad_norm": 0.4280725419521332, "learning_rate": 1.2051887402771414e-05, "loss": 0.3237426996231079, "step": 16850 }, { "epoch": 0.30799581879481175, "grad_norm": 0.5651392936706543, "learning_rate": 1.2043462678911224e-05, "loss": 0.3826097011566162, "step": 16860 }, { "epoch": 0.3085847209339988, "grad_norm": 0.5492408871650696, "learning_rate": 1.2035036441208508e-05, "loss": 0.35923752784729, "step": 16870 }, { "epoch": 0.3091736230731858, "grad_norm": 0.5015583634376526, "learning_rate": 1.2026608695905613e-05, "loss": 0.3188094854354858, "step": 16880 }, { "epoch": 0.30976252521237285, "grad_norm": 0.4915013611316681, "learning_rate": 1.2018179449246003e-05, "loss": 0.319187331199646, "step": 16890 }, { "epoch": 0.31035142735155985, "grad_norm": 0.5805712342262268, "learning_rate": 1.200974870747425e-05, "loss": 0.34655778408050536, "step": 16900 }, { "epoch": 0.3109403294907469, "grad_norm": 0.47708770632743835, "learning_rate": 1.2001316476836027e-05, "loss": 0.3589674472808838, "step": 16910 }, { "epoch": 0.3115292316299339, "grad_norm": 0.5847877860069275, "learning_rate": 1.1992882763578129e-05, "loss": 0.3358797788619995, "step": 16920 }, { "epoch": 0.3121181337691209, "grad_norm": 0.4165298044681549, "learning_rate": 1.1984447573948433e-05, "loss": 0.2909075736999512, "step": 16930 }, { "epoch": 0.31270703590830795, "grad_norm": 0.5033472776412964, "learning_rate": 1.1976010914195912e-05, "loss": 0.3396435737609863, "step": 16940 }, { "epoch": 0.31329593804749495, "grad_norm": 0.481092631816864, "learning_rate": 1.1967572790570634e-05, "loss": 0.349725341796875, "step": 16950 }, { "epoch": 0.313884840186682, "grad_norm": 0.6420412063598633, "learning_rate": 1.195913320932375e-05, "loss": 0.3309704542160034, "step": 16960 }, { "epoch": 0.314473742325869, "grad_norm": 0.4160732626914978, "learning_rate": 1.1950692176707484e-05, "loss": 0.3088299036026001, "step": 16970 }, { "epoch": 0.315062644465056, "grad_norm": 0.7493724822998047, "learning_rate": 1.1942249698975145e-05, "loss": 0.34613499641418455, "step": 16980 }, { "epoch": 0.31565154660424305, "grad_norm": 0.560613214969635, "learning_rate": 1.1933805782381106e-05, "loss": 0.3416209936141968, "step": 16990 }, { "epoch": 0.31624044874343005, "grad_norm": 0.39069998264312744, "learning_rate": 1.1925360433180811e-05, "loss": 0.3088143110275269, "step": 17000 }, { "epoch": 0.3168293508826171, "grad_norm": 0.5377970933914185, "learning_rate": 1.1916913657630756e-05, "loss": 0.34341752529144287, "step": 17010 }, { "epoch": 0.3174182530218041, "grad_norm": 0.7232832908630371, "learning_rate": 1.1908465461988504e-05, "loss": 0.36845357418060304, "step": 17020 }, { "epoch": 0.3180071551609911, "grad_norm": 0.6158499121665955, "learning_rate": 1.1900015852512665e-05, "loss": 0.3136057615280151, "step": 17030 }, { "epoch": 0.31859605730017815, "grad_norm": 0.6798778772354126, "learning_rate": 1.1891564835462899e-05, "loss": 0.3498586654663086, "step": 17040 }, { "epoch": 0.31918495943936515, "grad_norm": 0.4793880581855774, "learning_rate": 1.1883112417099902e-05, "loss": 0.403387451171875, "step": 17050 }, { "epoch": 0.3197738615785522, "grad_norm": 0.5634281635284424, "learning_rate": 1.1874658603685418e-05, "loss": 0.33276207447052003, "step": 17060 }, { "epoch": 0.3203627637177392, "grad_norm": 0.4227217137813568, "learning_rate": 1.1866203401482221e-05, "loss": 0.28128530979156496, "step": 17070 }, { "epoch": 0.32095166585692625, "grad_norm": 0.5922778844833374, "learning_rate": 1.185774681675411e-05, "loss": 0.37380969524383545, "step": 17080 }, { "epoch": 0.32154056799611325, "grad_norm": 0.5340359210968018, "learning_rate": 1.1849288855765908e-05, "loss": 0.3319640398025513, "step": 17090 }, { "epoch": 0.32212947013530024, "grad_norm": 0.5884603261947632, "learning_rate": 1.1840829524783467e-05, "loss": 0.3478586196899414, "step": 17100 }, { "epoch": 0.3227183722744873, "grad_norm": 0.541718602180481, "learning_rate": 1.183236883007364e-05, "loss": 0.3291804552078247, "step": 17110 }, { "epoch": 0.3233072744136743, "grad_norm": 0.45287781953811646, "learning_rate": 1.18239067779043e-05, "loss": 0.3369623899459839, "step": 17120 }, { "epoch": 0.32389617655286135, "grad_norm": 0.3728661835193634, "learning_rate": 1.1815443374544325e-05, "loss": 0.2932825326919556, "step": 17130 }, { "epoch": 0.32448507869204835, "grad_norm": 0.33350685238838196, "learning_rate": 1.1806978626263592e-05, "loss": 0.304435133934021, "step": 17140 }, { "epoch": 0.32507398083123534, "grad_norm": 0.5900481343269348, "learning_rate": 1.179851253933297e-05, "loss": 0.34465625286102297, "step": 17150 }, { "epoch": 0.3256628829704224, "grad_norm": 0.6594054102897644, "learning_rate": 1.179004512002433e-05, "loss": 0.3773390293121338, "step": 17160 }, { "epoch": 0.3262517851096094, "grad_norm": 0.43208757042884827, "learning_rate": 1.1781576374610521e-05, "loss": 0.2994920969009399, "step": 17170 }, { "epoch": 0.32684068724879645, "grad_norm": 0.6656627655029297, "learning_rate": 1.1773106309365376e-05, "loss": 0.35241796970367434, "step": 17180 }, { "epoch": 0.32742958938798344, "grad_norm": 0.5158475637435913, "learning_rate": 1.1764634930563713e-05, "loss": 0.37847516536712644, "step": 17190 }, { "epoch": 0.3280184915271705, "grad_norm": 0.6011676788330078, "learning_rate": 1.1756162244481315e-05, "loss": 0.3319359302520752, "step": 17200 }, { "epoch": 0.3286073936663575, "grad_norm": 0.6169087886810303, "learning_rate": 1.1747688257394935e-05, "loss": 0.3013732433319092, "step": 17210 }, { "epoch": 0.3291962958055445, "grad_norm": 0.5080479383468628, "learning_rate": 1.173921297558229e-05, "loss": 0.35474860668182373, "step": 17220 }, { "epoch": 0.32978519794473155, "grad_norm": 0.41231274604797363, "learning_rate": 1.1730736405322058e-05, "loss": 0.3309755802154541, "step": 17230 }, { "epoch": 0.33037410008391854, "grad_norm": 0.48149237036705017, "learning_rate": 1.1722258552893873e-05, "loss": 0.3156363725662231, "step": 17240 }, { "epoch": 0.3309630022231056, "grad_norm": 0.46051520109176636, "learning_rate": 1.1713779424578306e-05, "loss": 0.34392461776733396, "step": 17250 }, { "epoch": 0.3315519043622926, "grad_norm": 0.6114155650138855, "learning_rate": 1.17052990266569e-05, "loss": 0.35074615478515625, "step": 17260 }, { "epoch": 0.3321408065014796, "grad_norm": 0.3631711006164551, "learning_rate": 1.169681736541211e-05, "loss": 0.29846644401550293, "step": 17270 }, { "epoch": 0.33272970864066664, "grad_norm": 0.6994815468788147, "learning_rate": 1.1688334447127338e-05, "loss": 0.3308518648147583, "step": 17280 }, { "epoch": 0.33331861077985364, "grad_norm": 0.5173314213752747, "learning_rate": 1.1679850278086928e-05, "loss": 0.3892942428588867, "step": 17290 }, { "epoch": 0.3339075129190407, "grad_norm": 0.5165625810623169, "learning_rate": 1.1671364864576134e-05, "loss": 0.33359315395355227, "step": 17300 }, { "epoch": 0.3344964150582277, "grad_norm": 0.5825207829475403, "learning_rate": 1.166287821288114e-05, "loss": 0.30957281589508057, "step": 17310 }, { "epoch": 0.33508531719741474, "grad_norm": 0.46212196350097656, "learning_rate": 1.1654390329289043e-05, "loss": 0.33646080493927, "step": 17320 }, { "epoch": 0.33567421933660174, "grad_norm": 0.48147836327552795, "learning_rate": 1.1645901220087863e-05, "loss": 0.3066797494888306, "step": 17330 }, { "epoch": 0.33626312147578874, "grad_norm": 0.38833460211753845, "learning_rate": 1.1637410891566515e-05, "loss": 0.33685765266418455, "step": 17340 }, { "epoch": 0.3368520236149758, "grad_norm": 0.4933481514453888, "learning_rate": 1.1628919350014822e-05, "loss": 0.3211720705032349, "step": 17350 }, { "epoch": 0.3374409257541628, "grad_norm": 0.5086328983306885, "learning_rate": 1.1620426601723511e-05, "loss": 0.3363058090209961, "step": 17360 }, { "epoch": 0.33802982789334984, "grad_norm": 0.4331451654434204, "learning_rate": 1.1611932652984195e-05, "loss": 0.2997302055358887, "step": 17370 }, { "epoch": 0.33861873003253684, "grad_norm": 0.5514392256736755, "learning_rate": 1.160343751008938e-05, "loss": 0.3621697187423706, "step": 17380 }, { "epoch": 0.33920763217172384, "grad_norm": 0.4400123059749603, "learning_rate": 1.1594941179332459e-05, "loss": 0.3319763898849487, "step": 17390 }, { "epoch": 0.3397965343109109, "grad_norm": 0.4573168158531189, "learning_rate": 1.1586443667007698e-05, "loss": 0.3003279447555542, "step": 17400 }, { "epoch": 0.3403854364500979, "grad_norm": 0.4325551688671112, "learning_rate": 1.1577944979410246e-05, "loss": 0.3285285234451294, "step": 17410 }, { "epoch": 0.34097433858928494, "grad_norm": 0.5320870876312256, "learning_rate": 1.1569445122836121e-05, "loss": 0.3759173393249512, "step": 17420 }, { "epoch": 0.34156324072847194, "grad_norm": 0.7251632213592529, "learning_rate": 1.1560944103582206e-05, "loss": 0.33751797676086426, "step": 17430 }, { "epoch": 0.342152142867659, "grad_norm": 0.6429475545883179, "learning_rate": 1.1552441927946236e-05, "loss": 0.32382500171661377, "step": 17440 }, { "epoch": 0.342741045006846, "grad_norm": 0.4663575291633606, "learning_rate": 1.1543938602226822e-05, "loss": 0.3573769569396973, "step": 17450 }, { "epoch": 0.343329947146033, "grad_norm": 0.5212494134902954, "learning_rate": 1.1535434132723413e-05, "loss": 0.3427527904510498, "step": 17460 }, { "epoch": 0.34391884928522004, "grad_norm": 0.651847779750824, "learning_rate": 1.1526928525736304e-05, "loss": 0.38909182548522947, "step": 17470 }, { "epoch": 0.34450775142440704, "grad_norm": 0.3842572271823883, "learning_rate": 1.1518421787566646e-05, "loss": 0.30885374546051025, "step": 17480 }, { "epoch": 0.3450966535635941, "grad_norm": 0.6847534775733948, "learning_rate": 1.1509913924516411e-05, "loss": 0.3318326950073242, "step": 17490 }, { "epoch": 0.3456855557027811, "grad_norm": 0.47426706552505493, "learning_rate": 1.150140494288842e-05, "loss": 0.32754690647125245, "step": 17500 }, { "epoch": 0.3462744578419681, "grad_norm": 0.5355350375175476, "learning_rate": 1.1492894848986314e-05, "loss": 0.3402050733566284, "step": 17510 }, { "epoch": 0.34686335998115514, "grad_norm": 0.5681959986686707, "learning_rate": 1.1484383649114554e-05, "loss": 0.3161533117294312, "step": 17520 }, { "epoch": 0.34745226212034214, "grad_norm": 0.41901901364326477, "learning_rate": 1.1475871349578433e-05, "loss": 0.359666633605957, "step": 17530 }, { "epoch": 0.3480411642595292, "grad_norm": 0.5150687098503113, "learning_rate": 1.1467357956684046e-05, "loss": 0.2730682134628296, "step": 17540 }, { "epoch": 0.3486300663987162, "grad_norm": 0.6274443864822388, "learning_rate": 1.1458843476738305e-05, "loss": 0.3443694829940796, "step": 17550 }, { "epoch": 0.34921896853790324, "grad_norm": 0.5376511812210083, "learning_rate": 1.145032791604893e-05, "loss": 0.36712985038757323, "step": 17560 }, { "epoch": 0.34980787067709024, "grad_norm": 0.5284339189529419, "learning_rate": 1.1441811280924433e-05, "loss": 0.32198951244354246, "step": 17570 }, { "epoch": 0.35039677281627724, "grad_norm": 0.47663235664367676, "learning_rate": 1.1433293577674126e-05, "loss": 0.3281977653503418, "step": 17580 }, { "epoch": 0.3509856749554643, "grad_norm": 0.5215449333190918, "learning_rate": 1.1424774812608112e-05, "loss": 0.31394305229187014, "step": 17590 }, { "epoch": 0.3515745770946513, "grad_norm": 0.3889318108558655, "learning_rate": 1.141625499203729e-05, "loss": 0.30355846881866455, "step": 17600 }, { "epoch": 0.35216347923383834, "grad_norm": 0.501768171787262, "learning_rate": 1.140773412227332e-05, "loss": 0.3033878803253174, "step": 17610 }, { "epoch": 0.35275238137302534, "grad_norm": 0.5210403800010681, "learning_rate": 1.139921220962866e-05, "loss": 0.3389367818832397, "step": 17620 }, { "epoch": 0.35334128351221233, "grad_norm": 0.43747633695602417, "learning_rate": 1.1390689260416525e-05, "loss": 0.31826698780059814, "step": 17630 }, { "epoch": 0.3539301856513994, "grad_norm": 0.4340144693851471, "learning_rate": 1.1382165280950913e-05, "loss": 0.2706383943557739, "step": 17640 }, { "epoch": 0.3545190877905864, "grad_norm": 0.6043835878372192, "learning_rate": 1.1373640277546574e-05, "loss": 0.3420606851577759, "step": 17650 }, { "epoch": 0.35510798992977344, "grad_norm": 0.5960053205490112, "learning_rate": 1.1365114256519016e-05, "loss": 0.35730054378509524, "step": 17660 }, { "epoch": 0.35569689206896044, "grad_norm": 0.5361205339431763, "learning_rate": 1.1356587224184512e-05, "loss": 0.3293517351150513, "step": 17670 }, { "epoch": 0.3562857942081475, "grad_norm": 0.5568861961364746, "learning_rate": 1.1348059186860068e-05, "loss": 0.29970970153808596, "step": 17680 }, { "epoch": 0.3568746963473345, "grad_norm": 0.49755361676216125, "learning_rate": 1.1339530150863452e-05, "loss": 0.3156813383102417, "step": 17690 }, { "epoch": 0.3574635984865215, "grad_norm": 0.5404262542724609, "learning_rate": 1.1331000122513159e-05, "loss": 0.3754487276077271, "step": 17700 }, { "epoch": 0.35805250062570854, "grad_norm": 0.5739096403121948, "learning_rate": 1.132246910812842e-05, "loss": 0.36471569538116455, "step": 17710 }, { "epoch": 0.35864140276489553, "grad_norm": 0.42071059346199036, "learning_rate": 1.1313937114029208e-05, "loss": 0.36149144172668457, "step": 17720 }, { "epoch": 0.3592303049040826, "grad_norm": 0.3845641016960144, "learning_rate": 1.1305404146536208e-05, "loss": 0.3335672616958618, "step": 17730 }, { "epoch": 0.3598192070432696, "grad_norm": 0.5560808777809143, "learning_rate": 1.1296870211970833e-05, "loss": 0.3155803680419922, "step": 17740 }, { "epoch": 0.3604081091824566, "grad_norm": 0.45310431718826294, "learning_rate": 1.128833531665521e-05, "loss": 0.3067103385925293, "step": 17750 }, { "epoch": 0.36099701132164363, "grad_norm": 0.5308676958084106, "learning_rate": 1.1279799466912185e-05, "loss": 0.31024744510650637, "step": 17760 }, { "epoch": 0.36158591346083063, "grad_norm": 0.5986112952232361, "learning_rate": 1.1271262669065297e-05, "loss": 0.3418689012527466, "step": 17770 }, { "epoch": 0.3621748156000177, "grad_norm": 0.45526283979415894, "learning_rate": 1.1262724929438796e-05, "loss": 0.38426711559295657, "step": 17780 }, { "epoch": 0.3627637177392047, "grad_norm": 0.4339786767959595, "learning_rate": 1.1254186254357636e-05, "loss": 0.3331540584564209, "step": 17790 }, { "epoch": 0.3633526198783917, "grad_norm": 0.4134124517440796, "learning_rate": 1.124564665014745e-05, "loss": 0.3567472219467163, "step": 17800 }, { "epoch": 0.36394152201757873, "grad_norm": 0.40413060784339905, "learning_rate": 1.123710612313457e-05, "loss": 0.3055985689163208, "step": 17810 }, { "epoch": 0.36453042415676573, "grad_norm": 0.63724285364151, "learning_rate": 1.1228564679646007e-05, "loss": 0.3513849973678589, "step": 17820 }, { "epoch": 0.3651193262959528, "grad_norm": 0.4874816834926605, "learning_rate": 1.1220022326009454e-05, "loss": 0.34631502628326416, "step": 17830 }, { "epoch": 0.3657082284351398, "grad_norm": 0.5168736577033997, "learning_rate": 1.1211479068553272e-05, "loss": 0.3155064105987549, "step": 17840 }, { "epoch": 0.36629713057432683, "grad_norm": 0.5911180973052979, "learning_rate": 1.12029349136065e-05, "loss": 0.3136622905731201, "step": 17850 }, { "epoch": 0.36688603271351383, "grad_norm": 0.5655479431152344, "learning_rate": 1.1194389867498836e-05, "loss": 0.3297907590866089, "step": 17860 }, { "epoch": 0.36747493485270083, "grad_norm": 0.5870112180709839, "learning_rate": 1.1185843936560638e-05, "loss": 0.32051782608032225, "step": 17870 }, { "epoch": 0.3680638369918879, "grad_norm": 0.6464397311210632, "learning_rate": 1.1177297127122926e-05, "loss": 0.3478139877319336, "step": 17880 }, { "epoch": 0.3686527391310749, "grad_norm": 0.49403151869773865, "learning_rate": 1.116874944551736e-05, "loss": 0.3372506618499756, "step": 17890 }, { "epoch": 0.36924164127026193, "grad_norm": 0.45079585909843445, "learning_rate": 1.116020089807626e-05, "loss": 0.3355391025543213, "step": 17900 }, { "epoch": 0.36983054340944893, "grad_norm": 0.5029586553573608, "learning_rate": 1.115165149113257e-05, "loss": 0.3362027406692505, "step": 17910 }, { "epoch": 0.37041944554863593, "grad_norm": 0.5190810561180115, "learning_rate": 1.1143101231019887e-05, "loss": 0.3461099147796631, "step": 17920 }, { "epoch": 0.371008347687823, "grad_norm": 0.4179106056690216, "learning_rate": 1.1134550124072439e-05, "loss": 0.37421979904174807, "step": 17930 }, { "epoch": 0.37159724982701, "grad_norm": 0.45843154191970825, "learning_rate": 1.1125998176625063e-05, "loss": 0.34323863983154296, "step": 17940 }, { "epoch": 0.37218615196619703, "grad_norm": 0.5717971324920654, "learning_rate": 1.1117445395013238e-05, "loss": 0.3692241430282593, "step": 17950 }, { "epoch": 0.37277505410538403, "grad_norm": 0.5733130574226379, "learning_rate": 1.1108891785573057e-05, "loss": 0.30516059398651124, "step": 17960 }, { "epoch": 0.3733639562445711, "grad_norm": 0.6004065275192261, "learning_rate": 1.110033735464122e-05, "loss": 0.33603453636169434, "step": 17970 }, { "epoch": 0.3739528583837581, "grad_norm": 0.6029496192932129, "learning_rate": 1.109178210855504e-05, "loss": 0.3479745864868164, "step": 17980 }, { "epoch": 0.3745417605229451, "grad_norm": 0.46819090843200684, "learning_rate": 1.1083226053652434e-05, "loss": 0.3580667018890381, "step": 17990 }, { "epoch": 0.37513066266213213, "grad_norm": 0.4529450237751007, "learning_rate": 1.107466919627192e-05, "loss": 0.32753064632415774, "step": 18000 }, { "epoch": 0.37571956480131913, "grad_norm": 0.5019456148147583, "learning_rate": 1.10661115427526e-05, "loss": 0.3372443914413452, "step": 18010 }, { "epoch": 0.3763084669405062, "grad_norm": 0.4366326630115509, "learning_rate": 1.1057553099434182e-05, "loss": 0.34183778762817385, "step": 18020 }, { "epoch": 0.3768973690796932, "grad_norm": 0.5830532908439636, "learning_rate": 1.1048993872656947e-05, "loss": 0.3256777286529541, "step": 18030 }, { "epoch": 0.3774862712188802, "grad_norm": 0.5253162384033203, "learning_rate": 1.1040433868761762e-05, "loss": 0.36705794334411623, "step": 18040 }, { "epoch": 0.37807517335806723, "grad_norm": 0.5203476548194885, "learning_rate": 1.1031873094090064e-05, "loss": 0.32613799571990965, "step": 18050 }, { "epoch": 0.3786640754972542, "grad_norm": 0.4582841694355011, "learning_rate": 1.1023311554983871e-05, "loss": 0.3309525966644287, "step": 18060 }, { "epoch": 0.3792529776364413, "grad_norm": 0.6108660101890564, "learning_rate": 1.1014749257785757e-05, "loss": 0.2909572601318359, "step": 18070 }, { "epoch": 0.3798418797756283, "grad_norm": 0.431814968585968, "learning_rate": 1.1006186208838863e-05, "loss": 0.3219835042953491, "step": 18080 }, { "epoch": 0.38043078191481533, "grad_norm": 0.6464344263076782, "learning_rate": 1.0997622414486884e-05, "loss": 0.3050966262817383, "step": 18090 }, { "epoch": 0.3810196840540023, "grad_norm": 0.572990894317627, "learning_rate": 1.0989057881074075e-05, "loss": 0.3326403141021729, "step": 18100 }, { "epoch": 0.3816085861931893, "grad_norm": 0.6203936338424683, "learning_rate": 1.0980492614945224e-05, "loss": 0.3633547067642212, "step": 18110 }, { "epoch": 0.3821974883323764, "grad_norm": 0.4551924467086792, "learning_rate": 1.0971926622445677e-05, "loss": 0.3445324182510376, "step": 18120 }, { "epoch": 0.3827863904715634, "grad_norm": 0.4240310490131378, "learning_rate": 1.0963359909921305e-05, "loss": 0.3114022254943848, "step": 18130 }, { "epoch": 0.38337529261075043, "grad_norm": 0.6288055181503296, "learning_rate": 1.0954792483718528e-05, "loss": 0.3289827823638916, "step": 18140 }, { "epoch": 0.3839641947499374, "grad_norm": 0.9547879695892334, "learning_rate": 1.0946224350184279e-05, "loss": 0.3606033086776733, "step": 18150 }, { "epoch": 0.3845530968891244, "grad_norm": 0.461417555809021, "learning_rate": 1.0937655515666019e-05, "loss": 0.3582575798034668, "step": 18160 }, { "epoch": 0.3851419990283115, "grad_norm": 0.5724083781242371, "learning_rate": 1.092908598651174e-05, "loss": 0.34556844234466555, "step": 18170 }, { "epoch": 0.3857309011674985, "grad_norm": 0.5256638526916504, "learning_rate": 1.092051576906993e-05, "loss": 0.36998064517974855, "step": 18180 }, { "epoch": 0.3863198033066855, "grad_norm": 0.4151547849178314, "learning_rate": 1.0911944869689603e-05, "loss": 0.3025613307952881, "step": 18190 }, { "epoch": 0.3869087054458725, "grad_norm": 0.5589913129806519, "learning_rate": 1.090337329472027e-05, "loss": 0.3296489238739014, "step": 18200 }, { "epoch": 0.3874976075850596, "grad_norm": 0.5420318841934204, "learning_rate": 1.0894801050511941e-05, "loss": 0.33123204708099363, "step": 18210 }, { "epoch": 0.3880865097242466, "grad_norm": 0.4741280674934387, "learning_rate": 1.088622814341513e-05, "loss": 0.3957021236419678, "step": 18220 }, { "epoch": 0.3886754118634336, "grad_norm": 0.37447261810302734, "learning_rate": 1.087765457978083e-05, "loss": 0.3682077646255493, "step": 18230 }, { "epoch": 0.3892643140026206, "grad_norm": 0.565157949924469, "learning_rate": 1.0869080365960533e-05, "loss": 0.328628396987915, "step": 18240 }, { "epoch": 0.3898532161418076, "grad_norm": 0.6177994012832642, "learning_rate": 1.0860505508306205e-05, "loss": 0.31444711685180665, "step": 18250 }, { "epoch": 0.3904421182809947, "grad_norm": 0.4709830582141876, "learning_rate": 1.0851930013170292e-05, "loss": 0.38211669921875, "step": 18260 }, { "epoch": 0.3910310204201817, "grad_norm": 0.4428868591785431, "learning_rate": 1.0843353886905712e-05, "loss": 0.33335080146789553, "step": 18270 }, { "epoch": 0.39161992255936867, "grad_norm": 0.410986065864563, "learning_rate": 1.0834777135865844e-05, "loss": 0.3107865571975708, "step": 18280 }, { "epoch": 0.3922088246985557, "grad_norm": 0.5837542414665222, "learning_rate": 1.0826199766404542e-05, "loss": 0.3371870994567871, "step": 18290 }, { "epoch": 0.3927977268377427, "grad_norm": 0.5210637450218201, "learning_rate": 1.081762178487611e-05, "loss": 0.3847757339477539, "step": 18300 }, { "epoch": 0.3933866289769298, "grad_norm": 0.5802277326583862, "learning_rate": 1.0809043197635303e-05, "loss": 0.2958406686782837, "step": 18310 }, { "epoch": 0.3939755311161168, "grad_norm": 0.5560092329978943, "learning_rate": 1.0800464011037335e-05, "loss": 0.35242881774902346, "step": 18320 }, { "epoch": 0.3945644332553038, "grad_norm": 0.4784901440143585, "learning_rate": 1.0791884231437857e-05, "loss": 0.374701714515686, "step": 18330 }, { "epoch": 0.3951533353944908, "grad_norm": 0.42804038524627686, "learning_rate": 1.0783303865192956e-05, "loss": 0.3139012098312378, "step": 18340 }, { "epoch": 0.3957422375336778, "grad_norm": 0.487164705991745, "learning_rate": 1.0774722918659162e-05, "loss": 0.30117111206054686, "step": 18350 }, { "epoch": 0.3963311396728649, "grad_norm": 0.5271754264831543, "learning_rate": 1.0766141398193428e-05, "loss": 0.2747725248336792, "step": 18360 }, { "epoch": 0.39692004181205187, "grad_norm": 0.554478108882904, "learning_rate": 1.0757559310153136e-05, "loss": 0.3433955192565918, "step": 18370 }, { "epoch": 0.3975089439512389, "grad_norm": 0.5232164263725281, "learning_rate": 1.074897666089609e-05, "loss": 0.32048993110656737, "step": 18380 }, { "epoch": 0.3980978460904259, "grad_norm": 0.5831859707832336, "learning_rate": 1.07403934567805e-05, "loss": 0.34932880401611327, "step": 18390 }, { "epoch": 0.3986867482296129, "grad_norm": 0.517654538154602, "learning_rate": 1.0731809704165004e-05, "loss": 0.3019094705581665, "step": 18400 }, { "epoch": 0.3992756503688, "grad_norm": 0.42593055963516235, "learning_rate": 1.0723225409408626e-05, "loss": 0.3196821451187134, "step": 18410 }, { "epoch": 0.39986455250798697, "grad_norm": 0.5698215961456299, "learning_rate": 1.071464057887081e-05, "loss": 0.39989051818847654, "step": 18420 }, { "epoch": 0.400453454647174, "grad_norm": 0.5515074729919434, "learning_rate": 1.0706055218911389e-05, "loss": 0.30324015617370603, "step": 18430 }, { "epoch": 0.401042356786361, "grad_norm": 0.5419061779975891, "learning_rate": 1.069746933589058e-05, "loss": 0.3269846200942993, "step": 18440 }, { "epoch": 0.4016312589255481, "grad_norm": 0.4402956962585449, "learning_rate": 1.0688882936169005e-05, "loss": 0.35354883670806886, "step": 18450 }, { "epoch": 0.40222016106473507, "grad_norm": 0.48040151596069336, "learning_rate": 1.0680296026107657e-05, "loss": 0.33436222076416017, "step": 18460 }, { "epoch": 0.40280906320392207, "grad_norm": 0.5155353546142578, "learning_rate": 1.0671708612067905e-05, "loss": 0.3132216691970825, "step": 18470 }, { "epoch": 0.4033979653431091, "grad_norm": 0.3809160888195038, "learning_rate": 1.0663120700411503e-05, "loss": 0.3218281030654907, "step": 18480 }, { "epoch": 0.4039868674822961, "grad_norm": 0.7503964900970459, "learning_rate": 1.065453229750056e-05, "loss": 0.3034623384475708, "step": 18490 }, { "epoch": 0.40457576962148317, "grad_norm": 0.5207904577255249, "learning_rate": 1.0645943409697562e-05, "loss": 0.385773229598999, "step": 18500 }, { "epoch": 0.40516467176067017, "grad_norm": 0.5078195929527283, "learning_rate": 1.0637354043365343e-05, "loss": 0.34061288833618164, "step": 18510 }, { "epoch": 0.40575357389985717, "grad_norm": 0.6765367984771729, "learning_rate": 1.0628764204867097e-05, "loss": 0.3652578592300415, "step": 18520 }, { "epoch": 0.4063424760390442, "grad_norm": 0.5735986828804016, "learning_rate": 1.0620173900566371e-05, "loss": 0.31646969318389895, "step": 18530 }, { "epoch": 0.4069313781782312, "grad_norm": 0.4917837977409363, "learning_rate": 1.0611583136827048e-05, "loss": 0.3018482208251953, "step": 18540 }, { "epoch": 0.40752028031741827, "grad_norm": 0.6735865473747253, "learning_rate": 1.0602991920013357e-05, "loss": 0.3174836874008179, "step": 18550 }, { "epoch": 0.40810918245660527, "grad_norm": 0.5308519601821899, "learning_rate": 1.0594400256489865e-05, "loss": 0.3270721912384033, "step": 18560 }, { "epoch": 0.4086980845957923, "grad_norm": 0.7150763273239136, "learning_rate": 1.058580815262147e-05, "loss": 0.32981464862823484, "step": 18570 }, { "epoch": 0.4092869867349793, "grad_norm": 0.43131059408187866, "learning_rate": 1.0577215614773385e-05, "loss": 0.3384051561355591, "step": 18580 }, { "epoch": 0.4098758888741663, "grad_norm": 0.5200600028038025, "learning_rate": 1.0568622649311155e-05, "loss": 0.3384431838989258, "step": 18590 }, { "epoch": 0.41046479101335337, "grad_norm": 0.6033081412315369, "learning_rate": 1.0560029262600646e-05, "loss": 0.35957648754119875, "step": 18600 }, { "epoch": 0.41105369315254037, "grad_norm": 0.5229355096817017, "learning_rate": 1.0551435461008021e-05, "loss": 0.29958195686340333, "step": 18610 }, { "epoch": 0.4116425952917274, "grad_norm": 0.4558381736278534, "learning_rate": 1.0542841250899767e-05, "loss": 0.33932530879974365, "step": 18620 }, { "epoch": 0.4122314974309144, "grad_norm": 0.5608645081520081, "learning_rate": 1.053424663864266e-05, "loss": 0.3608505964279175, "step": 18630 }, { "epoch": 0.4128203995701014, "grad_norm": 0.5770053267478943, "learning_rate": 1.0525651630603784e-05, "loss": 0.324606728553772, "step": 18640 }, { "epoch": 0.41340930170928847, "grad_norm": 0.5501401424407959, "learning_rate": 1.0517056233150507e-05, "loss": 0.33813939094543455, "step": 18650 }, { "epoch": 0.41399820384847547, "grad_norm": 0.5083633661270142, "learning_rate": 1.0508460452650494e-05, "loss": 0.3613657712936401, "step": 18660 }, { "epoch": 0.4145871059876625, "grad_norm": 0.36879709362983704, "learning_rate": 1.0499864295471686e-05, "loss": 0.3206190586090088, "step": 18670 }, { "epoch": 0.4151760081268495, "grad_norm": 0.41787195205688477, "learning_rate": 1.0491267767982309e-05, "loss": 0.2932650804519653, "step": 18680 }, { "epoch": 0.4157649102660365, "grad_norm": 0.49490445852279663, "learning_rate": 1.0482670876550866e-05, "loss": 0.33705289363861085, "step": 18690 }, { "epoch": 0.41635381240522357, "grad_norm": 0.6565790772438049, "learning_rate": 1.0474073627546119e-05, "loss": 0.35194010734558107, "step": 18700 }, { "epoch": 0.41694271454441056, "grad_norm": 0.3916196823120117, "learning_rate": 1.0465476027337105e-05, "loss": 0.3205439805984497, "step": 18710 }, { "epoch": 0.4175316166835976, "grad_norm": 0.42626434564590454, "learning_rate": 1.0456878082293116e-05, "loss": 0.3181691884994507, "step": 18720 }, { "epoch": 0.4181205188227846, "grad_norm": 0.44844311475753784, "learning_rate": 1.0448279798783696e-05, "loss": 0.3551438570022583, "step": 18730 }, { "epoch": 0.41870942096197167, "grad_norm": 0.5858026742935181, "learning_rate": 1.043968118317865e-05, "loss": 0.34233784675598145, "step": 18740 }, { "epoch": 0.41929832310115867, "grad_norm": 0.37613603472709656, "learning_rate": 1.043108224184802e-05, "loss": 0.2954718589782715, "step": 18750 }, { "epoch": 0.41988722524034566, "grad_norm": 0.5048325657844543, "learning_rate": 1.0422482981162098e-05, "loss": 0.3393975019454956, "step": 18760 }, { "epoch": 0.4204761273795327, "grad_norm": 0.3939732313156128, "learning_rate": 1.0413883407491404e-05, "loss": 0.3354003429412842, "step": 18770 }, { "epoch": 0.4210650295187197, "grad_norm": 0.4722183048725128, "learning_rate": 1.0405283527206686e-05, "loss": 0.3202767610549927, "step": 18780 }, { "epoch": 0.42165393165790677, "grad_norm": 0.3288310468196869, "learning_rate": 1.0396683346678937e-05, "loss": 0.3230001926422119, "step": 18790 }, { "epoch": 0.42224283379709376, "grad_norm": 0.47190192341804504, "learning_rate": 1.0388082872279358e-05, "loss": 0.391486382484436, "step": 18800 }, { "epoch": 0.42283173593628076, "grad_norm": 0.3880299925804138, "learning_rate": 1.0379482110379369e-05, "loss": 0.3536626100540161, "step": 18810 }, { "epoch": 0.4234206380754678, "grad_norm": 0.6372632384300232, "learning_rate": 1.0370881067350607e-05, "loss": 0.32936174869537355, "step": 18820 }, { "epoch": 0.4240095402146548, "grad_norm": 0.45869895815849304, "learning_rate": 1.0362279749564918e-05, "loss": 0.31795122623443606, "step": 18830 }, { "epoch": 0.42459844235384187, "grad_norm": 0.519741415977478, "learning_rate": 1.0353678163394345e-05, "loss": 0.31838126182556153, "step": 18840 }, { "epoch": 0.42518734449302886, "grad_norm": 0.48522117733955383, "learning_rate": 1.0345076315211136e-05, "loss": 0.3214590549468994, "step": 18850 }, { "epoch": 0.4257762466322159, "grad_norm": 0.6484357714653015, "learning_rate": 1.0336474211387735e-05, "loss": 0.329789400100708, "step": 18860 }, { "epoch": 0.4263651487714029, "grad_norm": 0.4920342266559601, "learning_rate": 1.0327871858296763e-05, "loss": 0.31730482578277586, "step": 18870 }, { "epoch": 0.4269540509105899, "grad_norm": 0.6230435371398926, "learning_rate": 1.031926926231104e-05, "loss": 0.30693764686584474, "step": 18880 }, { "epoch": 0.42754295304977696, "grad_norm": 0.42832139134407043, "learning_rate": 1.031066642980356e-05, "loss": 0.34644081592559817, "step": 18890 }, { "epoch": 0.42813185518896396, "grad_norm": 0.4056282341480255, "learning_rate": 1.0302063367147493e-05, "loss": 0.3430598258972168, "step": 18900 }, { "epoch": 0.428720757328151, "grad_norm": 0.6032532453536987, "learning_rate": 1.0293460080716173e-05, "loss": 0.31088368892669677, "step": 18910 }, { "epoch": 0.429309659467338, "grad_norm": 0.4424247741699219, "learning_rate": 1.028485657688311e-05, "loss": 0.3318124294281006, "step": 18920 }, { "epoch": 0.429898561606525, "grad_norm": 0.5976609587669373, "learning_rate": 1.0276252862021973e-05, "loss": 0.3603160858154297, "step": 18930 }, { "epoch": 0.43048746374571206, "grad_norm": 0.734911322593689, "learning_rate": 1.026764894250658e-05, "loss": 0.3058892250061035, "step": 18940 }, { "epoch": 0.43107636588489906, "grad_norm": 0.5912443995475769, "learning_rate": 1.0259044824710908e-05, "loss": 0.35925891399383547, "step": 18950 }, { "epoch": 0.4316652680240861, "grad_norm": 0.6138342022895813, "learning_rate": 1.0250440515009083e-05, "loss": 0.3689428329467773, "step": 18960 }, { "epoch": 0.4322541701632731, "grad_norm": 0.4771357476711273, "learning_rate": 1.0241836019775359e-05, "loss": 0.28993663787841795, "step": 18970 }, { "epoch": 0.43284307230246016, "grad_norm": 0.4162738621234894, "learning_rate": 1.0233231345384144e-05, "loss": 0.33173356056213377, "step": 18980 }, { "epoch": 0.43343197444164716, "grad_norm": 0.523639440536499, "learning_rate": 1.0224626498209968e-05, "loss": 0.3231334686279297, "step": 18990 }, { "epoch": 0.43402087658083416, "grad_norm": 0.5286206007003784, "learning_rate": 1.0216021484627498e-05, "loss": 0.2987439393997192, "step": 19000 }, { "epoch": 0.4346097787200212, "grad_norm": 0.7788223624229431, "learning_rate": 1.0207416311011515e-05, "loss": 0.34293618202209475, "step": 19010 }, { "epoch": 0.4351986808592082, "grad_norm": 0.5508906245231628, "learning_rate": 1.0198810983736925e-05, "loss": 0.2925238132476807, "step": 19020 }, { "epoch": 0.43578758299839526, "grad_norm": 0.4864313304424286, "learning_rate": 1.0190205509178743e-05, "loss": 0.3213895559310913, "step": 19030 }, { "epoch": 0.43637648513758226, "grad_norm": 0.49183255434036255, "learning_rate": 1.0181599893712096e-05, "loss": 0.2851630449295044, "step": 19040 }, { "epoch": 0.43696538727676926, "grad_norm": 0.5265741348266602, "learning_rate": 1.0172994143712214e-05, "loss": 0.3224570512771606, "step": 19050 }, { "epoch": 0.4375542894159563, "grad_norm": 0.40781253576278687, "learning_rate": 1.0164388265554432e-05, "loss": 0.3142603158950806, "step": 19060 }, { "epoch": 0.4381431915551433, "grad_norm": 0.7541432976722717, "learning_rate": 1.0155782265614173e-05, "loss": 0.3369288921356201, "step": 19070 }, { "epoch": 0.43873209369433036, "grad_norm": 0.6035271883010864, "learning_rate": 1.0147176150266951e-05, "loss": 0.3188556432723999, "step": 19080 }, { "epoch": 0.43932099583351736, "grad_norm": 0.5085703730583191, "learning_rate": 1.013856992588837e-05, "loss": 0.303800106048584, "step": 19090 }, { "epoch": 0.4399098979727044, "grad_norm": 0.5276831388473511, "learning_rate": 1.0129963598854115e-05, "loss": 0.3448601245880127, "step": 19100 }, { "epoch": 0.4404988001118914, "grad_norm": 0.46354660391807556, "learning_rate": 1.0121357175539935e-05, "loss": 0.3307918071746826, "step": 19110 }, { "epoch": 0.4410877022510784, "grad_norm": 0.4951556622982025, "learning_rate": 1.011275066232167e-05, "loss": 0.3401324987411499, "step": 19120 }, { "epoch": 0.44167660439026546, "grad_norm": 0.4232742488384247, "learning_rate": 1.0104144065575208e-05, "loss": 0.31592321395874023, "step": 19130 }, { "epoch": 0.44226550652945246, "grad_norm": 0.5328933596611023, "learning_rate": 1.0095537391676512e-05, "loss": 0.3405735015869141, "step": 19140 }, { "epoch": 0.4428544086686395, "grad_norm": 0.3479550778865814, "learning_rate": 1.0086930647001592e-05, "loss": 0.2949350833892822, "step": 19150 }, { "epoch": 0.4434433108078265, "grad_norm": 0.4959895610809326, "learning_rate": 1.007832383792652e-05, "loss": 0.32050161361694335, "step": 19160 }, { "epoch": 0.4440322129470135, "grad_norm": 0.49866873025894165, "learning_rate": 1.006971697082741e-05, "loss": 0.355045223236084, "step": 19170 }, { "epoch": 0.44462111508620056, "grad_norm": 0.451447993516922, "learning_rate": 1.0061110052080419e-05, "loss": 0.336004900932312, "step": 19180 }, { "epoch": 0.44521001722538756, "grad_norm": 0.33737385272979736, "learning_rate": 1.0052503088061747e-05, "loss": 0.29807679653167723, "step": 19190 }, { "epoch": 0.4457989193645746, "grad_norm": 0.7805156111717224, "learning_rate": 1.0043896085147619e-05, "loss": 0.3311940908432007, "step": 19200 }, { "epoch": 0.4463878215037616, "grad_norm": 0.6914463639259338, "learning_rate": 1.0035289049714298e-05, "loss": 0.4065889835357666, "step": 19210 }, { "epoch": 0.44697672364294866, "grad_norm": 0.45067766308784485, "learning_rate": 1.0026681988138067e-05, "loss": 0.3361151695251465, "step": 19220 }, { "epoch": 0.44756562578213566, "grad_norm": 0.6864141821861267, "learning_rate": 1.0018074906795223e-05, "loss": 0.3601547718048096, "step": 19230 }, { "epoch": 0.44815452792132265, "grad_norm": 0.6408707499504089, "learning_rate": 1.000946781206209e-05, "loss": 0.3232536792755127, "step": 19240 }, { "epoch": 0.4487434300605097, "grad_norm": 0.5016939640045166, "learning_rate": 1.000086071031499e-05, "loss": 0.32203757762908936, "step": 19250 }, { "epoch": 0.4493323321996967, "grad_norm": 0.6501980423927307, "learning_rate": 9.992253607930258e-06, "loss": 0.3689067602157593, "step": 19260 }, { "epoch": 0.44992123433888376, "grad_norm": 0.6264510750770569, "learning_rate": 9.983646511284225e-06, "loss": 0.30927553176879885, "step": 19270 }, { "epoch": 0.45051013647807076, "grad_norm": 0.6092177629470825, "learning_rate": 9.97503942675322e-06, "loss": 0.31059560775756834, "step": 19280 }, { "epoch": 0.45109903861725775, "grad_norm": 0.5700091123580933, "learning_rate": 9.96643236071356e-06, "loss": 0.3796665668487549, "step": 19290 }, { "epoch": 0.4516879407564448, "grad_norm": 0.5979943871498108, "learning_rate": 9.957825319541553e-06, "loss": 0.30678281784057615, "step": 19300 }, { "epoch": 0.4522768428956318, "grad_norm": 0.5555444955825806, "learning_rate": 9.949218309613485e-06, "loss": 0.3552163362503052, "step": 19310 }, { "epoch": 0.45286574503481886, "grad_norm": 0.6973819136619568, "learning_rate": 9.940611337305624e-06, "loss": 0.3331474542617798, "step": 19320 }, { "epoch": 0.45345464717400585, "grad_norm": 0.4870843291282654, "learning_rate": 9.932004408994199e-06, "loss": 0.3123453140258789, "step": 19330 }, { "epoch": 0.4540435493131929, "grad_norm": 0.5498847365379333, "learning_rate": 9.923397531055418e-06, "loss": 0.3552001953125, "step": 19340 }, { "epoch": 0.4546324514523799, "grad_norm": 0.5565585494041443, "learning_rate": 9.914790709865451e-06, "loss": 0.35761404037475586, "step": 19350 }, { "epoch": 0.4552213535915669, "grad_norm": 0.6108387112617493, "learning_rate": 9.906183951800413e-06, "loss": 0.3010367155075073, "step": 19360 }, { "epoch": 0.45581025573075395, "grad_norm": 0.43073299527168274, "learning_rate": 9.897577263236388e-06, "loss": 0.3692013740539551, "step": 19370 }, { "epoch": 0.45639915786994095, "grad_norm": 0.7079335451126099, "learning_rate": 9.8889706505494e-06, "loss": 0.34194226264953614, "step": 19380 }, { "epoch": 0.456988060009128, "grad_norm": 0.42447835206985474, "learning_rate": 9.88036412011542e-06, "loss": 0.322648286819458, "step": 19390 }, { "epoch": 0.457576962148315, "grad_norm": 0.5273091197013855, "learning_rate": 9.871757678310357e-06, "loss": 0.32131361961364746, "step": 19400 }, { "epoch": 0.458165864287502, "grad_norm": 0.7731435894966125, "learning_rate": 9.86315133151005e-06, "loss": 0.3342172384262085, "step": 19410 }, { "epoch": 0.45875476642668905, "grad_norm": 0.6305910348892212, "learning_rate": 9.85454508609028e-06, "loss": 0.31827573776245116, "step": 19420 }, { "epoch": 0.45934366856587605, "grad_norm": 0.5356095433235168, "learning_rate": 9.845938948426737e-06, "loss": 0.32098846435546874, "step": 19430 }, { "epoch": 0.4599325707050631, "grad_norm": 0.5123095512390137, "learning_rate": 9.83733292489504e-06, "loss": 0.32648205757141113, "step": 19440 }, { "epoch": 0.4605214728442501, "grad_norm": 0.5732564330101013, "learning_rate": 9.828727021870723e-06, "loss": 0.3755709171295166, "step": 19450 }, { "epoch": 0.46111037498343715, "grad_norm": 0.47076311707496643, "learning_rate": 9.820121245729235e-06, "loss": 0.34211254119873047, "step": 19460 }, { "epoch": 0.46169927712262415, "grad_norm": 0.6839869022369385, "learning_rate": 9.81151560284592e-06, "loss": 0.3498252868652344, "step": 19470 }, { "epoch": 0.46228817926181115, "grad_norm": 0.5186501741409302, "learning_rate": 9.802910099596028e-06, "loss": 0.36026391983032224, "step": 19480 }, { "epoch": 0.4628770814009982, "grad_norm": 0.4770297408103943, "learning_rate": 9.794304742354718e-06, "loss": 0.37056689262390136, "step": 19490 }, { "epoch": 0.4634659835401852, "grad_norm": 0.627440333366394, "learning_rate": 9.785699537497016e-06, "loss": 0.30895543098449707, "step": 19500 }, { "epoch": 0.46405488567937225, "grad_norm": 0.46718519926071167, "learning_rate": 9.777094491397855e-06, "loss": 0.3476523160934448, "step": 19510 }, { "epoch": 0.46464378781855925, "grad_norm": 0.49364498257637024, "learning_rate": 9.768489610432042e-06, "loss": 0.31025097370147703, "step": 19520 }, { "epoch": 0.46523268995774625, "grad_norm": 0.47269266843795776, "learning_rate": 9.759884900974271e-06, "loss": 0.34619460105895994, "step": 19530 }, { "epoch": 0.4658215920969333, "grad_norm": 0.4028869867324829, "learning_rate": 9.751280369399094e-06, "loss": 0.3347482681274414, "step": 19540 }, { "epoch": 0.4664104942361203, "grad_norm": 0.5064608454704285, "learning_rate": 9.742676022080938e-06, "loss": 0.2837033271789551, "step": 19550 }, { "epoch": 0.46699939637530735, "grad_norm": 0.6683738231658936, "learning_rate": 9.734071865394105e-06, "loss": 0.35189387798309324, "step": 19560 }, { "epoch": 0.46758829851449435, "grad_norm": 0.5355792045593262, "learning_rate": 9.725467905712734e-06, "loss": 0.33201959133148196, "step": 19570 }, { "epoch": 0.46817720065368135, "grad_norm": 0.46873006224632263, "learning_rate": 9.716864149410832e-06, "loss": 0.32122485637664794, "step": 19580 }, { "epoch": 0.4687661027928684, "grad_norm": 0.5289217829704285, "learning_rate": 9.70826060286226e-06, "loss": 0.30953078269958495, "step": 19590 }, { "epoch": 0.4693550049320554, "grad_norm": 0.514378547668457, "learning_rate": 9.699657272440708e-06, "loss": 0.34544012546539304, "step": 19600 }, { "epoch": 0.46994390707124245, "grad_norm": 0.48041027784347534, "learning_rate": 9.691054164519713e-06, "loss": 0.3328669786453247, "step": 19610 }, { "epoch": 0.47053280921042945, "grad_norm": 0.6179592609405518, "learning_rate": 9.682451285472655e-06, "loss": 0.32280745506286623, "step": 19620 }, { "epoch": 0.4711217113496165, "grad_norm": 0.46111759543418884, "learning_rate": 9.67384864167274e-06, "loss": 0.34469308853149416, "step": 19630 }, { "epoch": 0.4717106134888035, "grad_norm": 0.5347710847854614, "learning_rate": 9.665246239492987e-06, "loss": 0.3322284460067749, "step": 19640 }, { "epoch": 0.4722995156279905, "grad_norm": 0.495753675699234, "learning_rate": 9.65664408530625e-06, "loss": 0.33630900382995604, "step": 19650 }, { "epoch": 0.47288841776717755, "grad_norm": 0.4357936680316925, "learning_rate": 9.648042185485206e-06, "loss": 0.35305893421173096, "step": 19660 }, { "epoch": 0.47347731990636455, "grad_norm": 0.3818766474723816, "learning_rate": 9.63944054640232e-06, "loss": 0.3047273397445679, "step": 19670 }, { "epoch": 0.4740662220455516, "grad_norm": 0.5023157000541687, "learning_rate": 9.63083917442988e-06, "loss": 0.31944894790649414, "step": 19680 }, { "epoch": 0.4746551241847386, "grad_norm": 0.5345441699028015, "learning_rate": 9.622238075939978e-06, "loss": 0.3723097562789917, "step": 19690 }, { "epoch": 0.4752440263239256, "grad_norm": 0.44630956649780273, "learning_rate": 9.613637257304503e-06, "loss": 0.34868273735046384, "step": 19700 }, { "epoch": 0.47583292846311265, "grad_norm": 0.533795952796936, "learning_rate": 9.605036724895117e-06, "loss": 0.31860296726226806, "step": 19710 }, { "epoch": 0.47642183060229965, "grad_norm": 0.582995593547821, "learning_rate": 9.596436485083297e-06, "loss": 0.3060085535049438, "step": 19720 }, { "epoch": 0.4770107327414867, "grad_norm": 0.530475914478302, "learning_rate": 9.587836544240293e-06, "loss": 0.29827830791473386, "step": 19730 }, { "epoch": 0.4775996348806737, "grad_norm": 0.4991210103034973, "learning_rate": 9.579236908737122e-06, "loss": 0.309691047668457, "step": 19740 }, { "epoch": 0.47818853701986075, "grad_norm": 0.6017098426818848, "learning_rate": 9.570637584944594e-06, "loss": 0.3190932273864746, "step": 19750 }, { "epoch": 0.47877743915904775, "grad_norm": 0.5965593457221985, "learning_rate": 9.562038579233274e-06, "loss": 0.3408595323562622, "step": 19760 }, { "epoch": 0.47936634129823474, "grad_norm": 0.4765942692756653, "learning_rate": 9.553439897973506e-06, "loss": 0.2977576494216919, "step": 19770 }, { "epoch": 0.4799552434374218, "grad_norm": 0.4798029959201813, "learning_rate": 9.544841547535368e-06, "loss": 0.292481803894043, "step": 19780 }, { "epoch": 0.4805441455766088, "grad_norm": 0.5262750387191772, "learning_rate": 9.536243534288718e-06, "loss": 0.3158060312271118, "step": 19790 }, { "epoch": 0.48113304771579585, "grad_norm": 0.48585793375968933, "learning_rate": 9.52764586460316e-06, "loss": 0.36906185150146487, "step": 19800 }, { "epoch": 0.48172194985498284, "grad_norm": 0.4842257499694824, "learning_rate": 9.519048544848027e-06, "loss": 0.34153103828430176, "step": 19810 }, { "epoch": 0.48231085199416984, "grad_norm": 0.6354231238365173, "learning_rate": 9.510451581392411e-06, "loss": 0.3649547338485718, "step": 19820 }, { "epoch": 0.4828997541333569, "grad_norm": 0.4403260052204132, "learning_rate": 9.501854980605132e-06, "loss": 0.34307677745819093, "step": 19830 }, { "epoch": 0.4834886562725439, "grad_norm": 0.47408294677734375, "learning_rate": 9.493258748854749e-06, "loss": 0.3197556257247925, "step": 19840 }, { "epoch": 0.48407755841173095, "grad_norm": 0.5365132093429565, "learning_rate": 9.48466289250953e-06, "loss": 0.3353714942932129, "step": 19850 }, { "epoch": 0.48466646055091794, "grad_norm": 0.45704543590545654, "learning_rate": 9.476067417937483e-06, "loss": 0.3221976518630981, "step": 19860 }, { "epoch": 0.485255362690105, "grad_norm": 0.5673915147781372, "learning_rate": 9.467472331506328e-06, "loss": 0.3349665641784668, "step": 19870 }, { "epoch": 0.485844264829292, "grad_norm": 0.5818110704421997, "learning_rate": 9.458877639583491e-06, "loss": 0.3239950895309448, "step": 19880 }, { "epoch": 0.486433166968479, "grad_norm": 0.4815223217010498, "learning_rate": 9.450283348536113e-06, "loss": 0.35516743659973143, "step": 19890 }, { "epoch": 0.48702206910766604, "grad_norm": 0.3858400583267212, "learning_rate": 9.44168946473104e-06, "loss": 0.30508553981781006, "step": 19900 }, { "epoch": 0.48761097124685304, "grad_norm": 0.40322813391685486, "learning_rate": 9.4330959945348e-06, "loss": 0.34970314502716066, "step": 19910 }, { "epoch": 0.4881998733860401, "grad_norm": 0.4464055895805359, "learning_rate": 9.424502944313636e-06, "loss": 0.3294321775436401, "step": 19920 }, { "epoch": 0.4887887755252271, "grad_norm": 0.6183503270149231, "learning_rate": 9.415910320433468e-06, "loss": 0.3455226421356201, "step": 19930 }, { "epoch": 0.4893776776644141, "grad_norm": 0.40142229199409485, "learning_rate": 9.407318129259908e-06, "loss": 0.336969780921936, "step": 19940 }, { "epoch": 0.48996657980360114, "grad_norm": 0.4948447346687317, "learning_rate": 9.39872637715823e-06, "loss": 0.32353670597076417, "step": 19950 }, { "epoch": 0.49055548194278814, "grad_norm": 0.5817307829856873, "learning_rate": 9.390135070493405e-06, "loss": 0.3054076910018921, "step": 19960 }, { "epoch": 0.4911443840819752, "grad_norm": 0.5277529358863831, "learning_rate": 9.38154421563006e-06, "loss": 0.3014042615890503, "step": 19970 }, { "epoch": 0.4917332862211622, "grad_norm": 0.5169093012809753, "learning_rate": 9.372953818932492e-06, "loss": 0.3508548021316528, "step": 19980 }, { "epoch": 0.49232218836034924, "grad_norm": 0.9008108377456665, "learning_rate": 9.364363886764654e-06, "loss": 0.3293674230575562, "step": 19990 }, { "epoch": 0.49291109049953624, "grad_norm": 0.39181557297706604, "learning_rate": 9.355774425490164e-06, "loss": 0.3517799377441406, "step": 20000 }, { "epoch": 0.49349999263872324, "grad_norm": 0.6753020882606506, "learning_rate": 9.347185441472284e-06, "loss": 0.3150752067565918, "step": 20010 }, { "epoch": 0.4940888947779103, "grad_norm": 0.7037740349769592, "learning_rate": 9.33859694107392e-06, "loss": 0.3307088136672974, "step": 20020 }, { "epoch": 0.4946777969170973, "grad_norm": 0.44468972086906433, "learning_rate": 9.330008930657628e-06, "loss": 0.2880657196044922, "step": 20030 }, { "epoch": 0.49526669905628434, "grad_norm": 0.4342098832130432, "learning_rate": 9.321421416585597e-06, "loss": 0.2980025768280029, "step": 20040 }, { "epoch": 0.49585560119547134, "grad_norm": 0.4263576865196228, "learning_rate": 9.312834405219643e-06, "loss": 0.34778828620910646, "step": 20050 }, { "epoch": 0.49644450333465834, "grad_norm": 0.5838814377784729, "learning_rate": 9.304247902921221e-06, "loss": 0.3058819055557251, "step": 20060 }, { "epoch": 0.4970334054738454, "grad_norm": 0.594804584980011, "learning_rate": 9.295661916051399e-06, "loss": 0.32688255310058595, "step": 20070 }, { "epoch": 0.4976223076130324, "grad_norm": 0.5693280696868896, "learning_rate": 9.28707645097087e-06, "loss": 0.35323657989501955, "step": 20080 }, { "epoch": 0.49821120975221944, "grad_norm": 0.48369190096855164, "learning_rate": 9.278491514039933e-06, "loss": 0.28449625968933107, "step": 20090 }, { "epoch": 0.49880011189140644, "grad_norm": 0.49779272079467773, "learning_rate": 9.2699071116185e-06, "loss": 0.3553641319274902, "step": 20100 }, { "epoch": 0.4993890140305935, "grad_norm": 0.5822665691375732, "learning_rate": 9.261323250066091e-06, "loss": 0.3432048559188843, "step": 20110 }, { "epoch": 0.4999779161697805, "grad_norm": 0.46141040325164795, "learning_rate": 9.252739935741819e-06, "loss": 0.33115482330322266, "step": 20120 }, { "epoch": 0.5005668183089675, "grad_norm": 0.4029105305671692, "learning_rate": 9.24415717500439e-06, "loss": 0.2930819034576416, "step": 20130 }, { "epoch": 0.5011557204481545, "grad_norm": 0.3630213439464569, "learning_rate": 9.23557497421211e-06, "loss": 0.3301170587539673, "step": 20140 }, { "epoch": 0.5017446225873415, "grad_norm": 0.693270742893219, "learning_rate": 9.226993339722863e-06, "loss": 0.3494812726974487, "step": 20150 }, { "epoch": 0.5023335247265286, "grad_norm": 0.4895801842212677, "learning_rate": 9.21841227789411e-06, "loss": 0.35516374111175536, "step": 20160 }, { "epoch": 0.5029224268657156, "grad_norm": 0.5890467762947083, "learning_rate": 9.209831795082895e-06, "loss": 0.3570082664489746, "step": 20170 }, { "epoch": 0.5035113290049026, "grad_norm": 0.5062775611877441, "learning_rate": 9.201251897645831e-06, "loss": 0.3441716909408569, "step": 20180 }, { "epoch": 0.5041002311440896, "grad_norm": 0.4861823320388794, "learning_rate": 9.192672591939095e-06, "loss": 0.35355212688446047, "step": 20190 }, { "epoch": 0.5046891332832767, "grad_norm": 0.5408733487129211, "learning_rate": 9.184093884318426e-06, "loss": 0.31600022315979004, "step": 20200 }, { "epoch": 0.5052780354224636, "grad_norm": 0.4040115475654602, "learning_rate": 9.175515781139124e-06, "loss": 0.3064298391342163, "step": 20210 }, { "epoch": 0.5058669375616507, "grad_norm": 0.47857865691185, "learning_rate": 9.166938288756036e-06, "loss": 0.35083608627319335, "step": 20220 }, { "epoch": 0.5064558397008377, "grad_norm": 0.4509667158126831, "learning_rate": 9.15836141352356e-06, "loss": 0.3015639066696167, "step": 20230 }, { "epoch": 0.5070447418400247, "grad_norm": 0.503539502620697, "learning_rate": 9.149785161795635e-06, "loss": 0.314848518371582, "step": 20240 }, { "epoch": 0.5076336439792117, "grad_norm": 0.6507366895675659, "learning_rate": 9.14120953992574e-06, "loss": 0.3434900283813477, "step": 20250 }, { "epoch": 0.5082225461183988, "grad_norm": 0.4945518374443054, "learning_rate": 9.132634554266884e-06, "loss": 0.28918888568878176, "step": 20260 }, { "epoch": 0.5088114482575858, "grad_norm": 0.5628319382667542, "learning_rate": 9.124060211171607e-06, "loss": 0.3289486408233643, "step": 20270 }, { "epoch": 0.5094003503967728, "grad_norm": 0.612379252910614, "learning_rate": 9.115486516991976e-06, "loss": 0.3660238027572632, "step": 20280 }, { "epoch": 0.5099892525359598, "grad_norm": 0.48985615372657776, "learning_rate": 9.106913478079571e-06, "loss": 0.34131522178649903, "step": 20290 }, { "epoch": 0.5105781546751469, "grad_norm": 0.5824920535087585, "learning_rate": 9.098341100785489e-06, "loss": 0.35565876960754395, "step": 20300 }, { "epoch": 0.5111670568143338, "grad_norm": 0.556921660900116, "learning_rate": 9.089769391460338e-06, "loss": 0.3186305284500122, "step": 20310 }, { "epoch": 0.5117559589535209, "grad_norm": 0.5126838088035583, "learning_rate": 9.081198356454234e-06, "loss": 0.33225743770599364, "step": 20320 }, { "epoch": 0.5123448610927079, "grad_norm": 0.4831154942512512, "learning_rate": 9.072628002116785e-06, "loss": 0.34442825317382814, "step": 20330 }, { "epoch": 0.512933763231895, "grad_norm": 0.5144264101982117, "learning_rate": 9.064058334797101e-06, "loss": 0.30769901275634765, "step": 20340 }, { "epoch": 0.5135226653710819, "grad_norm": 0.3744441866874695, "learning_rate": 9.055489360843786e-06, "loss": 0.3281486988067627, "step": 20350 }, { "epoch": 0.514111567510269, "grad_norm": 0.6178848743438721, "learning_rate": 9.04692108660492e-06, "loss": 0.29697275161743164, "step": 20360 }, { "epoch": 0.514700469649456, "grad_norm": 0.5015506148338318, "learning_rate": 9.038353518428073e-06, "loss": 0.3397674560546875, "step": 20370 }, { "epoch": 0.515289371788643, "grad_norm": 0.48760369420051575, "learning_rate": 9.02978666266029e-06, "loss": 0.34339008331298826, "step": 20380 }, { "epoch": 0.51587827392783, "grad_norm": 0.6680124998092651, "learning_rate": 9.021220525648094e-06, "loss": 0.36819231510162354, "step": 20390 }, { "epoch": 0.5164671760670171, "grad_norm": 0.5157464742660522, "learning_rate": 9.012655113737461e-06, "loss": 0.31729907989501954, "step": 20400 }, { "epoch": 0.5170560782062041, "grad_norm": 0.549658477306366, "learning_rate": 9.00409043327384e-06, "loss": 0.31063170433044435, "step": 20410 }, { "epoch": 0.5176449803453911, "grad_norm": 0.37188032269477844, "learning_rate": 8.995526490602139e-06, "loss": 0.34298512935638426, "step": 20420 }, { "epoch": 0.5182338824845781, "grad_norm": 0.4365472197532654, "learning_rate": 8.986963292066714e-06, "loss": 0.3617975950241089, "step": 20430 }, { "epoch": 0.5188227846237652, "grad_norm": 0.4558432698249817, "learning_rate": 8.978400844011373e-06, "loss": 0.3443664312362671, "step": 20440 }, { "epoch": 0.5194116867629521, "grad_norm": 0.5337691903114319, "learning_rate": 8.969839152779368e-06, "loss": 0.3298299551010132, "step": 20450 }, { "epoch": 0.5200005889021392, "grad_norm": 0.403022825717926, "learning_rate": 8.961278224713397e-06, "loss": 0.32500669956207273, "step": 20460 }, { "epoch": 0.5205894910413262, "grad_norm": 0.617641031742096, "learning_rate": 8.952718066155576e-06, "loss": 0.30732457637786864, "step": 20470 }, { "epoch": 0.5211783931805132, "grad_norm": 0.49487122893333435, "learning_rate": 8.94415868344746e-06, "loss": 0.3351806640625, "step": 20480 }, { "epoch": 0.5217672953197002, "grad_norm": 0.49184170365333557, "learning_rate": 8.935600082930041e-06, "loss": 0.3267770051956177, "step": 20490 }, { "epoch": 0.5223561974588873, "grad_norm": 0.5031239986419678, "learning_rate": 8.927042270943708e-06, "loss": 0.32576675415039064, "step": 20500 }, { "epoch": 0.5229450995980743, "grad_norm": 0.6650512218475342, "learning_rate": 8.91848525382828e-06, "loss": 0.30576224327087403, "step": 20510 }, { "epoch": 0.5235340017372613, "grad_norm": 0.5808973908424377, "learning_rate": 8.90992903792299e-06, "loss": 0.3091950654983521, "step": 20520 }, { "epoch": 0.5241229038764483, "grad_norm": 0.8440518975257874, "learning_rate": 8.901373629566478e-06, "loss": 0.37259554862976074, "step": 20530 }, { "epoch": 0.5247118060156354, "grad_norm": 0.5153144598007202, "learning_rate": 8.892819035096767e-06, "loss": 0.30569963455200194, "step": 20540 }, { "epoch": 0.5253007081548223, "grad_norm": 0.5548160076141357, "learning_rate": 8.884265260851294e-06, "loss": 0.3229044437408447, "step": 20550 }, { "epoch": 0.5258896102940094, "grad_norm": 0.39233437180519104, "learning_rate": 8.875712313166896e-06, "loss": 0.30423948764801023, "step": 20560 }, { "epoch": 0.5264785124331964, "grad_norm": 0.432068794965744, "learning_rate": 8.867160198379774e-06, "loss": 0.3726584672927856, "step": 20570 }, { "epoch": 0.5270674145723835, "grad_norm": 0.4923165440559387, "learning_rate": 8.858608922825525e-06, "loss": 0.3356772899627686, "step": 20580 }, { "epoch": 0.5276563167115704, "grad_norm": 0.49191373586654663, "learning_rate": 8.850058492839136e-06, "loss": 0.3749016284942627, "step": 20590 }, { "epoch": 0.5282452188507575, "grad_norm": 0.4214537739753723, "learning_rate": 8.841508914754943e-06, "loss": 0.2916389226913452, "step": 20600 }, { "epoch": 0.5288341209899445, "grad_norm": 0.36057358980178833, "learning_rate": 8.832960194906664e-06, "loss": 0.33347761631011963, "step": 20610 }, { "epoch": 0.5294230231291315, "grad_norm": 0.4407459795475006, "learning_rate": 8.824412339627385e-06, "loss": 0.32480225563049314, "step": 20620 }, { "epoch": 0.5300119252683185, "grad_norm": 0.4679315984249115, "learning_rate": 8.81586535524955e-06, "loss": 0.34778380393981934, "step": 20630 }, { "epoch": 0.5306008274075056, "grad_norm": 0.4996224045753479, "learning_rate": 8.807319248104946e-06, "loss": 0.3080305576324463, "step": 20640 }, { "epoch": 0.5311897295466925, "grad_norm": 0.659571647644043, "learning_rate": 8.798774024524724e-06, "loss": 0.3540723562240601, "step": 20650 }, { "epoch": 0.5317786316858796, "grad_norm": 0.4343690574169159, "learning_rate": 8.790229690839381e-06, "loss": 0.3309625148773193, "step": 20660 }, { "epoch": 1.0001177804278374, "grad_norm": 0.5431579351425171, "learning_rate": 8.781686253378737e-06, "loss": 0.2577112913131714, "step": 20670 }, { "epoch": 1.0007066825670243, "grad_norm": 0.5656217932701111, "learning_rate": 8.773143718471962e-06, "loss": 0.3304065942764282, "step": 20680 }, { "epoch": 1.0012955847062115, "grad_norm": 0.5017174482345581, "learning_rate": 8.764602092447566e-06, "loss": 0.3006248474121094, "step": 20690 }, { "epoch": 1.0018844868453984, "grad_norm": 0.6396520733833313, "learning_rate": 8.75606138163337e-06, "loss": 0.3022319316864014, "step": 20700 }, { "epoch": 1.0024733889845854, "grad_norm": 0.6054333448410034, "learning_rate": 8.747521592356515e-06, "loss": 0.3206266641616821, "step": 20710 }, { "epoch": 1.0030622911237725, "grad_norm": 0.5231119394302368, "learning_rate": 8.738982730943473e-06, "loss": 0.31122736930847167, "step": 20720 }, { "epoch": 1.0036511932629595, "grad_norm": 0.42934882640838623, "learning_rate": 8.730444803720028e-06, "loss": 0.3232825994491577, "step": 20730 }, { "epoch": 1.0042400954021466, "grad_norm": 0.6007156372070312, "learning_rate": 8.721907817011253e-06, "loss": 0.30849595069885255, "step": 20740 }, { "epoch": 1.0048289975413336, "grad_norm": 0.5063642859458923, "learning_rate": 8.713371777141546e-06, "loss": 0.3662745237350464, "step": 20750 }, { "epoch": 1.0054178996805205, "grad_norm": 0.3994095027446747, "learning_rate": 8.704836690434592e-06, "loss": 0.33065094947814944, "step": 20760 }, { "epoch": 1.0060068018197077, "grad_norm": 0.5173213481903076, "learning_rate": 8.696302563213376e-06, "loss": 0.31879141330718996, "step": 20770 }, { "epoch": 1.0065957039588946, "grad_norm": 0.5755904912948608, "learning_rate": 8.687769401800165e-06, "loss": 0.3134969711303711, "step": 20780 }, { "epoch": 1.0071846060980816, "grad_norm": 0.5068026185035706, "learning_rate": 8.679237212516515e-06, "loss": 0.312576961517334, "step": 20790 }, { "epoch": 1.0077735082372687, "grad_norm": 0.5917866826057434, "learning_rate": 8.670706001683268e-06, "loss": 0.33536975383758544, "step": 20800 }, { "epoch": 1.0083624103764557, "grad_norm": 0.46439191699028015, "learning_rate": 8.66217577562052e-06, "loss": 0.30885701179504393, "step": 20810 }, { "epoch": 1.0089513125156426, "grad_norm": 0.6849414110183716, "learning_rate": 8.653646540647665e-06, "loss": 0.3611874580383301, "step": 20820 }, { "epoch": 1.0095402146548298, "grad_norm": 0.4316321909427643, "learning_rate": 8.645118303083343e-06, "loss": 0.33206474781036377, "step": 20830 }, { "epoch": 1.0101291167940167, "grad_norm": 0.5199546813964844, "learning_rate": 8.636591069245465e-06, "loss": 0.3244402647018433, "step": 20840 }, { "epoch": 1.0107180189332037, "grad_norm": 0.37254786491394043, "learning_rate": 8.628064845451192e-06, "loss": 0.3099872350692749, "step": 20850 }, { "epoch": 1.0113069210723908, "grad_norm": 0.447578102350235, "learning_rate": 8.619539638016943e-06, "loss": 0.302624249458313, "step": 20860 }, { "epoch": 1.0118958232115778, "grad_norm": 0.7044022083282471, "learning_rate": 8.611015453258383e-06, "loss": 0.35456070899963377, "step": 20870 }, { "epoch": 1.012484725350765, "grad_norm": 0.40635499358177185, "learning_rate": 8.60249229749041e-06, "loss": 0.315609073638916, "step": 20880 }, { "epoch": 1.0130736274899519, "grad_norm": 0.507087767124176, "learning_rate": 8.593970177027171e-06, "loss": 0.34087791442871096, "step": 20890 }, { "epoch": 1.0136625296291388, "grad_norm": 0.5087314248085022, "learning_rate": 8.585449098182043e-06, "loss": 0.323476767539978, "step": 20900 }, { "epoch": 1.014251431768326, "grad_norm": 0.7349691390991211, "learning_rate": 8.576929067267632e-06, "loss": 0.3375177621841431, "step": 20910 }, { "epoch": 1.014840333907513, "grad_norm": 0.474084734916687, "learning_rate": 8.56841009059576e-06, "loss": 0.33080832958221434, "step": 20920 }, { "epoch": 1.0154292360466999, "grad_norm": 0.5240881443023682, "learning_rate": 8.559892174477478e-06, "loss": 0.3074531555175781, "step": 20930 }, { "epoch": 1.016018138185887, "grad_norm": 0.4130280017852783, "learning_rate": 8.551375325223051e-06, "loss": 0.309173059463501, "step": 20940 }, { "epoch": 1.016607040325074, "grad_norm": 0.566071093082428, "learning_rate": 8.542859549141941e-06, "loss": 0.30594844818115235, "step": 20950 }, { "epoch": 1.017195942464261, "grad_norm": 0.4964791238307953, "learning_rate": 8.534344852542831e-06, "loss": 0.3217644214630127, "step": 20960 }, { "epoch": 1.017784844603448, "grad_norm": 0.5716405510902405, "learning_rate": 8.525831241733598e-06, "loss": 0.2752408742904663, "step": 20970 }, { "epoch": 1.018373746742635, "grad_norm": 0.5027134418487549, "learning_rate": 8.517318723021308e-06, "loss": 0.3137612819671631, "step": 20980 }, { "epoch": 1.018962648881822, "grad_norm": 0.4373280107975006, "learning_rate": 8.508807302712226e-06, "loss": 0.29294919967651367, "step": 20990 }, { "epoch": 1.0195515510210091, "grad_norm": 0.7312750816345215, "learning_rate": 8.500296987111805e-06, "loss": 0.3183205842971802, "step": 21000 }, { "epoch": 1.020140453160196, "grad_norm": 1.0019419193267822, "learning_rate": 8.491787782524673e-06, "loss": 0.3434912919998169, "step": 21010 }, { "epoch": 1.020729355299383, "grad_norm": 0.5943272709846497, "learning_rate": 8.483279695254636e-06, "loss": 0.2949512004852295, "step": 21020 }, { "epoch": 1.0213182574385702, "grad_norm": 0.5467690825462341, "learning_rate": 8.474772731604677e-06, "loss": 0.31776883602142336, "step": 21030 }, { "epoch": 1.0219071595777571, "grad_norm": 0.5589897036552429, "learning_rate": 8.466266897876943e-06, "loss": 0.3297996759414673, "step": 21040 }, { "epoch": 1.0224960617169443, "grad_norm": 0.4891567528247833, "learning_rate": 8.457762200372742e-06, "loss": 0.3188640832901001, "step": 21050 }, { "epoch": 1.0230849638561312, "grad_norm": 0.5349894762039185, "learning_rate": 8.449258645392542e-06, "loss": 0.3374501705169678, "step": 21060 }, { "epoch": 1.0236738659953182, "grad_norm": 0.48298966884613037, "learning_rate": 8.44075623923597e-06, "loss": 0.3040606021881104, "step": 21070 }, { "epoch": 1.0242627681345053, "grad_norm": 0.43995797634124756, "learning_rate": 8.432254988201797e-06, "loss": 0.3040143013000488, "step": 21080 }, { "epoch": 1.0248516702736923, "grad_norm": 0.5162304639816284, "learning_rate": 8.423754898587934e-06, "loss": 0.30702035427093505, "step": 21090 }, { "epoch": 1.0254405724128792, "grad_norm": 0.5936978459358215, "learning_rate": 8.415255976691439e-06, "loss": 0.3250797510147095, "step": 21100 }, { "epoch": 1.0260294745520664, "grad_norm": 0.7360086441040039, "learning_rate": 8.406758228808502e-06, "loss": 0.3072027683258057, "step": 21110 }, { "epoch": 1.0266183766912533, "grad_norm": 0.5226261615753174, "learning_rate": 8.398261661234443e-06, "loss": 0.3185266494750977, "step": 21120 }, { "epoch": 1.0272072788304403, "grad_norm": 0.4685438573360443, "learning_rate": 8.389766280263708e-06, "loss": 0.30079426765441897, "step": 21130 }, { "epoch": 1.0277961809696274, "grad_norm": 0.5932722687721252, "learning_rate": 8.381272092189862e-06, "loss": 0.3036510705947876, "step": 21140 }, { "epoch": 1.0283850831088144, "grad_norm": 0.6751628518104553, "learning_rate": 8.372779103305593e-06, "loss": 0.35922856330871583, "step": 21150 }, { "epoch": 1.0289739852480013, "grad_norm": 0.6838120222091675, "learning_rate": 8.364287319902689e-06, "loss": 0.31440584659576415, "step": 21160 }, { "epoch": 1.0295628873871885, "grad_norm": 0.5243028998374939, "learning_rate": 8.355796748272058e-06, "loss": 0.3055161952972412, "step": 21170 }, { "epoch": 1.0301517895263754, "grad_norm": 0.5890933871269226, "learning_rate": 8.3473073947037e-06, "loss": 0.3315397024154663, "step": 21180 }, { "epoch": 1.0307406916655624, "grad_norm": 0.5039840340614319, "learning_rate": 8.338819265486718e-06, "loss": 0.3099033832550049, "step": 21190 }, { "epoch": 1.0313295938047495, "grad_norm": 0.5723875164985657, "learning_rate": 8.330332366909306e-06, "loss": 0.33810627460479736, "step": 21200 }, { "epoch": 1.0319184959439365, "grad_norm": 0.4698494076728821, "learning_rate": 8.321846705258747e-06, "loss": 0.2978017568588257, "step": 21210 }, { "epoch": 1.0325073980831236, "grad_norm": 0.5213358402252197, "learning_rate": 8.313362286821409e-06, "loss": 0.33399035930633547, "step": 21220 }, { "epoch": 1.0330963002223106, "grad_norm": 0.5122473239898682, "learning_rate": 8.304879117882736e-06, "loss": 0.3406623125076294, "step": 21230 }, { "epoch": 1.0336852023614975, "grad_norm": 0.54488605260849, "learning_rate": 8.296397204727248e-06, "loss": 0.30620951652526857, "step": 21240 }, { "epoch": 1.0342741045006847, "grad_norm": 0.4760464131832123, "learning_rate": 8.287916553638535e-06, "loss": 0.2944849729537964, "step": 21250 }, { "epoch": 1.0348630066398716, "grad_norm": 0.42696648836135864, "learning_rate": 8.27943717089925e-06, "loss": 0.2778064012527466, "step": 21260 }, { "epoch": 1.0354519087790586, "grad_norm": 0.5648515820503235, "learning_rate": 8.27095906279111e-06, "loss": 0.32552597522735593, "step": 21270 }, { "epoch": 1.0360408109182457, "grad_norm": 0.4743388593196869, "learning_rate": 8.26248223559489e-06, "loss": 0.302048397064209, "step": 21280 }, { "epoch": 1.0366297130574327, "grad_norm": 0.5903350114822388, "learning_rate": 8.2540066955904e-06, "loss": 0.2942698001861572, "step": 21290 }, { "epoch": 1.0372186151966196, "grad_norm": 0.6139523983001709, "learning_rate": 8.245532449056518e-06, "loss": 0.3640142917633057, "step": 21300 }, { "epoch": 1.0378075173358068, "grad_norm": 0.6249023675918579, "learning_rate": 8.23705950227115e-06, "loss": 0.3366565465927124, "step": 21310 }, { "epoch": 1.0383964194749937, "grad_norm": 0.6966297626495361, "learning_rate": 8.228587861511248e-06, "loss": 0.31494531631469724, "step": 21320 }, { "epoch": 1.0389853216141807, "grad_norm": 0.6325928568840027, "learning_rate": 8.220117533052783e-06, "loss": 0.31868901252746584, "step": 21330 }, { "epoch": 1.0395742237533678, "grad_norm": 0.5605037808418274, "learning_rate": 8.21164852317077e-06, "loss": 0.28186750411987305, "step": 21340 }, { "epoch": 1.0401631258925548, "grad_norm": 0.6019505858421326, "learning_rate": 8.203180838139237e-06, "loss": 0.3601059436798096, "step": 21350 }, { "epoch": 1.0407520280317417, "grad_norm": 0.5883786082267761, "learning_rate": 8.194714484231232e-06, "loss": 0.28165948390960693, "step": 21360 }, { "epoch": 1.0413409301709289, "grad_norm": 0.5144480466842651, "learning_rate": 8.186249467718819e-06, "loss": 0.3021122217178345, "step": 21370 }, { "epoch": 1.0419298323101158, "grad_norm": 0.5759645700454712, "learning_rate": 8.177785794873067e-06, "loss": 0.30824234485626223, "step": 21380 }, { "epoch": 1.042518734449303, "grad_norm": 0.5930377840995789, "learning_rate": 8.169323471964062e-06, "loss": 0.33448286056518556, "step": 21390 }, { "epoch": 1.04310763658849, "grad_norm": 0.5330750346183777, "learning_rate": 8.16086250526087e-06, "loss": 0.34727308750152586, "step": 21400 }, { "epoch": 1.0436965387276769, "grad_norm": 0.4752162992954254, "learning_rate": 8.152402901031566e-06, "loss": 0.2913142204284668, "step": 21410 }, { "epoch": 1.044285440866864, "grad_norm": 0.5919825434684753, "learning_rate": 8.143944665543221e-06, "loss": 0.30481126308441164, "step": 21420 }, { "epoch": 1.044874343006051, "grad_norm": 0.5822833180427551, "learning_rate": 8.135487805061872e-06, "loss": 0.32891271114349363, "step": 21430 }, { "epoch": 1.045463245145238, "grad_norm": 0.5448771715164185, "learning_rate": 8.127032325852553e-06, "loss": 0.3372995138168335, "step": 21440 }, { "epoch": 1.046052147284425, "grad_norm": 0.6203205585479736, "learning_rate": 8.118578234179269e-06, "loss": 0.31540205478668215, "step": 21450 }, { "epoch": 1.046641049423612, "grad_norm": 0.5247964262962341, "learning_rate": 8.110125536305008e-06, "loss": 0.3504541635513306, "step": 21460 }, { "epoch": 1.047229951562799, "grad_norm": 0.5298458933830261, "learning_rate": 8.101674238491706e-06, "loss": 0.3117208480834961, "step": 21470 }, { "epoch": 1.0478188537019861, "grad_norm": 0.5793840289115906, "learning_rate": 8.093224347000272e-06, "loss": 0.277477765083313, "step": 21480 }, { "epoch": 1.048407755841173, "grad_norm": 0.573676347732544, "learning_rate": 8.084775868090585e-06, "loss": 0.31302475929260254, "step": 21490 }, { "epoch": 1.04899665798036, "grad_norm": 0.5949481725692749, "learning_rate": 8.076328808021452e-06, "loss": 0.348280143737793, "step": 21500 }, { "epoch": 1.0495855601195472, "grad_norm": 0.6175553202629089, "learning_rate": 8.067883173050647e-06, "loss": 0.3280457019805908, "step": 21510 }, { "epoch": 1.0501744622587341, "grad_norm": 0.47868600487709045, "learning_rate": 8.059438969434887e-06, "loss": 0.2914414882659912, "step": 21520 }, { "epoch": 1.050763364397921, "grad_norm": 0.5610666275024414, "learning_rate": 8.050996203429826e-06, "loss": 0.29220898151397706, "step": 21530 }, { "epoch": 1.0513522665371082, "grad_norm": 0.5156832337379456, "learning_rate": 8.042554881290045e-06, "loss": 0.32132012844085694, "step": 21540 }, { "epoch": 1.0519411686762952, "grad_norm": 0.8837072253227234, "learning_rate": 8.03411500926907e-06, "loss": 0.34771440029144285, "step": 21550 }, { "epoch": 1.0525300708154823, "grad_norm": 0.5226482152938843, "learning_rate": 8.025676593619349e-06, "loss": 0.35601909160614015, "step": 21560 }, { "epoch": 1.0531189729546693, "grad_norm": 0.41165435314178467, "learning_rate": 8.017239640592238e-06, "loss": 0.31964716911315916, "step": 21570 }, { "epoch": 1.000588902139187, "grad_norm": 0.4360155165195465, "learning_rate": 8.00880415643802e-06, "loss": 0.3045898675918579, "step": 21580 }, { "epoch": 1.001177804278374, "grad_norm": 0.45523330569267273, "learning_rate": 8.000370147405893e-06, "loss": 0.2939732074737549, "step": 21590 }, { "epoch": 1.001766706417561, "grad_norm": 0.46996915340423584, "learning_rate": 7.991937619743965e-06, "loss": 0.2905446767807007, "step": 21600 }, { "epoch": 1.002355608556748, "grad_norm": 0.5596964955329895, "learning_rate": 7.983506579699224e-06, "loss": 0.3024577617645264, "step": 21610 }, { "epoch": 1.0029445106959352, "grad_norm": 0.4323621094226837, "learning_rate": 7.975077033517581e-06, "loss": 0.3049370527267456, "step": 21620 }, { "epoch": 1.003533412835122, "grad_norm": 0.4908789396286011, "learning_rate": 7.966648987443836e-06, "loss": 0.2969056606292725, "step": 21630 }, { "epoch": 1.004122314974309, "grad_norm": 0.5171236991882324, "learning_rate": 7.95822244772166e-06, "loss": 0.31091558933258057, "step": 21640 }, { "epoch": 1.0047112171134962, "grad_norm": 0.6011161208152771, "learning_rate": 7.949797420593629e-06, "loss": 0.33983659744262695, "step": 21650 }, { "epoch": 1.0053001192526831, "grad_norm": 0.5088688135147095, "learning_rate": 7.94137391230119e-06, "loss": 0.30883886814117434, "step": 21660 }, { "epoch": 1.0058890213918703, "grad_norm": 0.43027040362358093, "learning_rate": 7.932951929084659e-06, "loss": 0.31936476230621336, "step": 21670 }, { "epoch": 1.0064779235310573, "grad_norm": 0.6836991310119629, "learning_rate": 7.924531477183234e-06, "loss": 0.2996341705322266, "step": 21680 }, { "epoch": 1.0070668256702442, "grad_norm": 0.48718008399009705, "learning_rate": 7.916112562834972e-06, "loss": 0.29176647663116456, "step": 21690 }, { "epoch": 1.0076557278094314, "grad_norm": 0.4564577341079712, "learning_rate": 7.907695192276794e-06, "loss": 0.32588858604431153, "step": 21700 }, { "epoch": 1.0082446299486183, "grad_norm": 0.6729231476783752, "learning_rate": 7.899279371744466e-06, "loss": 0.30163233280181884, "step": 21710 }, { "epoch": 1.0088335320878052, "grad_norm": 0.6937153935432434, "learning_rate": 7.890865107472623e-06, "loss": 0.3308787107467651, "step": 21720 }, { "epoch": 1.0094224342269924, "grad_norm": 0.6318997144699097, "learning_rate": 7.882452405694741e-06, "loss": 0.33919191360473633, "step": 21730 }, { "epoch": 1.0100113363661793, "grad_norm": 0.6127704977989197, "learning_rate": 7.874041272643124e-06, "loss": 0.30438416004180907, "step": 21740 }, { "epoch": 1.0106002385053663, "grad_norm": 0.7637942433357239, "learning_rate": 7.865631714548935e-06, "loss": 0.2895756721496582, "step": 21750 }, { "epoch": 1.0111891406445535, "grad_norm": 0.5038984417915344, "learning_rate": 7.857223737642162e-06, "loss": 0.3039198160171509, "step": 21760 }, { "epoch": 1.0117780427837404, "grad_norm": 0.5057458281517029, "learning_rate": 7.848817348151619e-06, "loss": 0.3313896179199219, "step": 21770 }, { "epoch": 1.0123669449229273, "grad_norm": 0.8855213522911072, "learning_rate": 7.840412552304946e-06, "loss": 0.3168225049972534, "step": 21780 }, { "epoch": 1.0129558470621145, "grad_norm": 0.5678660273551941, "learning_rate": 7.8320093563286e-06, "loss": 0.32615911960601807, "step": 21790 }, { "epoch": 1.0135447492013014, "grad_norm": 0.45175665616989136, "learning_rate": 7.823607766447862e-06, "loss": 0.2981481313705444, "step": 21800 }, { "epoch": 1.0141336513404884, "grad_norm": 0.7206020951271057, "learning_rate": 7.815207788886809e-06, "loss": 0.31194424629211426, "step": 21810 }, { "epoch": 1.0147225534796755, "grad_norm": 0.503878653049469, "learning_rate": 7.806809429868332e-06, "loss": 0.3337538242340088, "step": 21820 }, { "epoch": 1.0153114556188625, "grad_norm": 0.6249765157699585, "learning_rate": 7.798412695614127e-06, "loss": 0.29600656032562256, "step": 21830 }, { "epoch": 1.0159003577580497, "grad_norm": 0.5455305576324463, "learning_rate": 7.790017592344677e-06, "loss": 0.29645500183105467, "step": 21840 }, { "epoch": 1.0164892598972366, "grad_norm": 0.6103838086128235, "learning_rate": 7.781624126279259e-06, "loss": 0.2836963176727295, "step": 21850 }, { "epoch": 1.0170781620364235, "grad_norm": 0.5378738641738892, "learning_rate": 7.773232303635943e-06, "loss": 0.306247353553772, "step": 21860 }, { "epoch": 1.0176670641756107, "grad_norm": 0.5331019759178162, "learning_rate": 7.764842130631578e-06, "loss": 0.2789117097854614, "step": 21870 }, { "epoch": 1.0182559663147976, "grad_norm": 0.628761887550354, "learning_rate": 7.756453613481785e-06, "loss": 0.2968428373336792, "step": 21880 }, { "epoch": 1.0188448684539846, "grad_norm": 0.6370389461517334, "learning_rate": 7.74806675840097e-06, "loss": 0.2768991708755493, "step": 21890 }, { "epoch": 1.0194337705931718, "grad_norm": 0.4939384162425995, "learning_rate": 7.739681571602296e-06, "loss": 0.2903782606124878, "step": 21900 }, { "epoch": 1.0200226727323587, "grad_norm": 0.5532512068748474, "learning_rate": 7.731298059297703e-06, "loss": 0.3225206136703491, "step": 21910 }, { "epoch": 1.0206115748715456, "grad_norm": 0.4954109489917755, "learning_rate": 7.722916227697875e-06, "loss": 0.2948393106460571, "step": 21920 }, { "epoch": 1.0212004770107328, "grad_norm": 0.5349054932594299, "learning_rate": 7.714536083012262e-06, "loss": 0.3126988410949707, "step": 21930 }, { "epoch": 1.0217893791499197, "grad_norm": 0.6231203675270081, "learning_rate": 7.706157631449064e-06, "loss": 0.3034493446350098, "step": 21940 }, { "epoch": 1.0223782812891067, "grad_norm": 0.3781105875968933, "learning_rate": 7.69778087921522e-06, "loss": 0.3286472797393799, "step": 21950 }, { "epoch": 1.0229671834282938, "grad_norm": 0.6762375235557556, "learning_rate": 7.689405832516415e-06, "loss": 0.29863533973693845, "step": 21960 }, { "epoch": 1.0235560855674808, "grad_norm": 0.5935095548629761, "learning_rate": 7.681032497557071e-06, "loss": 0.29398813247680666, "step": 21970 }, { "epoch": 1.024144987706668, "grad_norm": 0.6417388319969177, "learning_rate": 7.672660880540338e-06, "loss": 0.29917452335357664, "step": 21980 }, { "epoch": 1.024733889845855, "grad_norm": 0.5046927332878113, "learning_rate": 7.664290987668099e-06, "loss": 0.28663339614868166, "step": 21990 }, { "epoch": 1.0253227919850418, "grad_norm": 0.5907155275344849, "learning_rate": 7.655922825140953e-06, "loss": 0.30097687244415283, "step": 22000 }, { "epoch": 1.000588902139187, "grad_norm": 0.48425525426864624, "learning_rate": 7.647556399158224e-06, "loss": 0.29228758811950684, "step": 22010 }, { "epoch": 1.001177804278374, "grad_norm": 0.48323771357536316, "learning_rate": 7.639191715917942e-06, "loss": 0.28048245906829833, "step": 22020 }, { "epoch": 1.001766706417561, "grad_norm": 0.5064662098884583, "learning_rate": 7.630828781616855e-06, "loss": 0.278092098236084, "step": 22030 }, { "epoch": 1.002355608556748, "grad_norm": 0.6046943664550781, "learning_rate": 7.622467602450408e-06, "loss": 0.28984732627868653, "step": 22040 }, { "epoch": 1.0029445106959352, "grad_norm": 0.47363725304603577, "learning_rate": 7.6141081846127476e-06, "loss": 0.29199268817901614, "step": 22050 }, { "epoch": 1.003533412835122, "grad_norm": 0.5396824479103088, "learning_rate": 7.605750534296715e-06, "loss": 0.28434274196624754, "step": 22060 }, { "epoch": 1.004122314974309, "grad_norm": 0.5625888705253601, "learning_rate": 7.597394657693846e-06, "loss": 0.29864068031311036, "step": 22070 }, { "epoch": 1.0047112171134962, "grad_norm": 0.663952112197876, "learning_rate": 7.589040560994359e-06, "loss": 0.3257955551147461, "step": 22080 }, { "epoch": 1.0053001192526831, "grad_norm": 0.5438215136528015, "learning_rate": 7.580688250387153e-06, "loss": 0.2925358057022095, "step": 22090 }, { "epoch": 1.0058890213918703, "grad_norm": 0.47057896852493286, "learning_rate": 7.572337732059804e-06, "loss": 0.30653402805328367, "step": 22100 }, { "epoch": 1.0064779235310573, "grad_norm": 0.7431734204292297, "learning_rate": 7.563989012198567e-06, "loss": 0.2856238842010498, "step": 22110 }, { "epoch": 1.0070668256702442, "grad_norm": 0.530850887298584, "learning_rate": 7.5556420969883514e-06, "loss": 0.2797433853149414, "step": 22120 }, { "epoch": 1.0076557278094314, "grad_norm": 0.4939509928226471, "learning_rate": 7.547296992612743e-06, "loss": 0.31263027191162107, "step": 22130 }, { "epoch": 1.0082446299486183, "grad_norm": 0.7328402996063232, "learning_rate": 7.538953705253976e-06, "loss": 0.2868973731994629, "step": 22140 }, { "epoch": 1.0088335320878052, "grad_norm": 0.7524517774581909, "learning_rate": 7.530612241092949e-06, "loss": 0.31657512187957765, "step": 22150 }, { "epoch": 1.0094224342269924, "grad_norm": 0.6675053834915161, "learning_rate": 7.522272606309197e-06, "loss": 0.32521488666534426, "step": 22160 }, { "epoch": 1.0100113363661793, "grad_norm": 0.6449823379516602, "learning_rate": 7.513934807080912e-06, "loss": 0.2921640872955322, "step": 22170 }, { "epoch": 1.0106002385053663, "grad_norm": 0.7885655164718628, "learning_rate": 7.505598849584919e-06, "loss": 0.2757026433944702, "step": 22180 }, { "epoch": 1.0111891406445535, "grad_norm": 0.5649700164794922, "learning_rate": 7.497264739996677e-06, "loss": 0.2918077945709229, "step": 22190 }, { "epoch": 1.0117780427837404, "grad_norm": 0.5488483309745789, "learning_rate": 7.488932484490285e-06, "loss": 0.3179105997085571, "step": 22200 }, { "epoch": 1.0123669449229273, "grad_norm": 0.9580336213111877, "learning_rate": 7.480602089238458e-06, "loss": 0.3026809453964233, "step": 22210 }, { "epoch": 1.0129558470621145, "grad_norm": 0.6262626051902771, "learning_rate": 7.4722735604125426e-06, "loss": 0.31271002292633054, "step": 22220 }, { "epoch": 1.0135447492013014, "grad_norm": 0.5093092918395996, "learning_rate": 7.463946904182493e-06, "loss": 0.2824387550354004, "step": 22230 }, { "epoch": 1.0141336513404884, "grad_norm": 0.7273538112640381, "learning_rate": 7.455622126716884e-06, "loss": 0.29818389415740965, "step": 22240 }, { "epoch": 1.0147225534796755, "grad_norm": 0.5526953935623169, "learning_rate": 7.447299234182897e-06, "loss": 0.3176830530166626, "step": 22250 }, { "epoch": 1.0153114556188625, "grad_norm": 0.6923249363899231, "learning_rate": 7.4389782327463115e-06, "loss": 0.28188557624816896, "step": 22260 }, { "epoch": 1.0159003577580497, "grad_norm": 0.5926777124404907, "learning_rate": 7.430659128571512e-06, "loss": 0.2830108642578125, "step": 22270 }, { "epoch": 1.0164892598972366, "grad_norm": 0.6492775678634644, "learning_rate": 7.422341927821472e-06, "loss": 0.26895556449890134, "step": 22280 }, { "epoch": 1.0170781620364235, "grad_norm": 0.6075423359870911, "learning_rate": 7.41402663665777e-06, "loss": 0.2936882495880127, "step": 22290 }, { "epoch": 1.0176670641756107, "grad_norm": 0.5558634996414185, "learning_rate": 7.405713261240546e-06, "loss": 0.2674739122390747, "step": 22300 }, { "epoch": 1.0182559663147976, "grad_norm": 0.6704955101013184, "learning_rate": 7.397401807728537e-06, "loss": 0.28234109878540037, "step": 22310 }, { "epoch": 1.0188448684539846, "grad_norm": 0.7046987414360046, "learning_rate": 7.389092282279056e-06, "loss": 0.26440184116363524, "step": 22320 }, { "epoch": 1.0194337705931718, "grad_norm": 0.544464111328125, "learning_rate": 7.380784691047978e-06, "loss": 0.27717225551605223, "step": 22330 }, { "epoch": 1.0200226727323587, "grad_norm": 0.6577533483505249, "learning_rate": 7.3724790401897535e-06, "loss": 0.30646774768829343, "step": 22340 }, { "epoch": 1.0206115748715456, "grad_norm": 0.5517728328704834, "learning_rate": 7.364175335857396e-06, "loss": 0.27901554107666016, "step": 22350 }, { "epoch": 1.0212004770107328, "grad_norm": 0.6010990142822266, "learning_rate": 7.355873584202469e-06, "loss": 0.2974077701568604, "step": 22360 }, { "epoch": 1.0217893791499197, "grad_norm": 0.666664183139801, "learning_rate": 7.347573791375098e-06, "loss": 0.29017441272735595, "step": 22370 }, { "epoch": 1.0223782812891067, "grad_norm": 0.42047885060310364, "learning_rate": 7.339275963523949e-06, "loss": 0.31345391273498535, "step": 22380 }, { "epoch": 1.0229671834282938, "grad_norm": 0.771675169467926, "learning_rate": 7.330980106796247e-06, "loss": 0.28289506435394285, "step": 22390 }, { "epoch": 1.0235560855674808, "grad_norm": 0.6523538827896118, "learning_rate": 7.322686227337736e-06, "loss": 0.2807029962539673, "step": 22400 }, { "epoch": 1.024144987706668, "grad_norm": 0.7218649387359619, "learning_rate": 7.3143943312927105e-06, "loss": 0.2857635736465454, "step": 22410 }, { "epoch": 1.024733889845855, "grad_norm": 0.5726776719093323, "learning_rate": 7.306104424803997e-06, "loss": 0.2752896785736084, "step": 22420 }, { "epoch": 1.0253227919850418, "grad_norm": 0.7518470883369446, "learning_rate": 7.297816514012932e-06, "loss": 0.2875993728637695, "step": 22430 }, { "epoch": 1.025911694124229, "grad_norm": 0.6067575216293335, "learning_rate": 7.289530605059387e-06, "loss": 0.3137255430221558, "step": 22440 }, { "epoch": 1.026500596263416, "grad_norm": 0.5945624709129333, "learning_rate": 7.2812467040817495e-06, "loss": 0.3012516498565674, "step": 22450 }, { "epoch": 1.0270894984026029, "grad_norm": 0.6497918367385864, "learning_rate": 7.272964817216921e-06, "loss": 0.2930523633956909, "step": 22460 }, { "epoch": 1.02767840054179, "grad_norm": 0.7743053436279297, "learning_rate": 7.264684950600298e-06, "loss": 0.2972320556640625, "step": 22470 }, { "epoch": 1.028267302680977, "grad_norm": 0.5921115875244141, "learning_rate": 7.256407110365793e-06, "loss": 0.33031015396118163, "step": 22480 }, { "epoch": 1.028856204820164, "grad_norm": 0.5421119928359985, "learning_rate": 7.248131302645822e-06, "loss": 0.2864166259765625, "step": 22490 }, { "epoch": 1.029445106959351, "grad_norm": 0.6747573614120483, "learning_rate": 7.239857533571275e-06, "loss": 0.32904815673828125, "step": 22500 }, { "epoch": 1.030034009098538, "grad_norm": 0.6753649711608887, "learning_rate": 7.231585809271548e-06, "loss": 0.32037887573242185, "step": 22510 }, { "epoch": 1.030622911237725, "grad_norm": 0.6607484817504883, "learning_rate": 7.2233161358745206e-06, "loss": 0.28330106735229493, "step": 22520 }, { "epoch": 1.0312118133769121, "grad_norm": 0.6534244418144226, "learning_rate": 7.215048519506553e-06, "loss": 0.319904088973999, "step": 22530 }, { "epoch": 1.031800715516099, "grad_norm": 0.5070501565933228, "learning_rate": 7.206782966292467e-06, "loss": 0.29261629581451415, "step": 22540 }, { "epoch": 1.032389617655286, "grad_norm": 0.8052894473075867, "learning_rate": 7.198519482355579e-06, "loss": 0.3149136543273926, "step": 22550 }, { "epoch": 1.0329785197944732, "grad_norm": 0.7005195021629333, "learning_rate": 7.190258073817663e-06, "loss": 0.3459923505783081, "step": 22560 }, { "epoch": 1.0335674219336601, "grad_norm": 0.6370934844017029, "learning_rate": 7.181998746798941e-06, "loss": 0.29389522075653074, "step": 22570 }, { "epoch": 1.0341563240728473, "grad_norm": 0.7414294481277466, "learning_rate": 7.173741507418118e-06, "loss": 0.28257572650909424, "step": 22580 }, { "epoch": 1.0347452262120342, "grad_norm": 0.48353108763694763, "learning_rate": 7.165486361792335e-06, "loss": 0.2590343713760376, "step": 22590 }, { "epoch": 1.0353341283512212, "grad_norm": 0.42479702830314636, "learning_rate": 7.157233316037193e-06, "loss": 0.31477794647216795, "step": 22600 }, { "epoch": 1.0359230304904083, "grad_norm": 0.5863731503486633, "learning_rate": 7.14898237626672e-06, "loss": 0.29864251613616943, "step": 22610 }, { "epoch": 1.0365119326295953, "grad_norm": 0.6245896220207214, "learning_rate": 7.140733548593405e-06, "loss": 0.2855991840362549, "step": 22620 }, { "epoch": 1.0371008347687822, "grad_norm": 0.5756831765174866, "learning_rate": 7.132486839128163e-06, "loss": 0.2991547346115112, "step": 22630 }, { "epoch": 1.0376897369079694, "grad_norm": 0.5729307532310486, "learning_rate": 7.1242422539803316e-06, "loss": 0.34747326374053955, "step": 22640 }, { "epoch": 1.0382786390471563, "grad_norm": 0.5047797560691833, "learning_rate": 7.115999799257687e-06, "loss": 0.30151143074035647, "step": 22650 }, { "epoch": 1.0388675411863433, "grad_norm": 0.5321477651596069, "learning_rate": 7.107759481066428e-06, "loss": 0.3011895418167114, "step": 22660 }, { "epoch": 1.0394564433255304, "grad_norm": 0.5457594990730286, "learning_rate": 7.099521305511155e-06, "loss": 0.2998720645904541, "step": 22670 }, { "epoch": 1.0400453454647174, "grad_norm": 0.519544243812561, "learning_rate": 7.091285278694897e-06, "loss": 0.32325191497802735, "step": 22680 }, { "epoch": 1.0406342476039043, "grad_norm": 0.48021751642227173, "learning_rate": 7.083051406719086e-06, "loss": 0.2925208806991577, "step": 22690 }, { "epoch": 1.0412231497430915, "grad_norm": 0.5858035087585449, "learning_rate": 7.074819695683559e-06, "loss": 0.288367486000061, "step": 22700 }, { "epoch": 1.0418120518822784, "grad_norm": 0.729720413684845, "learning_rate": 7.0665901516865445e-06, "loss": 0.29185311794281005, "step": 22710 }, { "epoch": 1.0424009540214654, "grad_norm": 0.6120954751968384, "learning_rate": 7.058362780824675e-06, "loss": 0.3332364559173584, "step": 22720 }, { "epoch": 1.0429898561606525, "grad_norm": 0.7623544335365295, "learning_rate": 7.050137589192971e-06, "loss": 0.32504799365997317, "step": 22730 }, { "epoch": 1.0435787582998395, "grad_norm": 0.748421847820282, "learning_rate": 7.041914582884831e-06, "loss": 0.2918930292129517, "step": 22740 }, { "epoch": 1.0441676604390266, "grad_norm": 0.6219499111175537, "learning_rate": 7.033693767992047e-06, "loss": 0.2775969743728638, "step": 22750 }, { "epoch": 1.0447565625782136, "grad_norm": 0.5952306389808655, "learning_rate": 7.025475150604777e-06, "loss": 0.30975172519683836, "step": 22760 }, { "epoch": 1.0453454647174005, "grad_norm": 0.5512995719909668, "learning_rate": 7.017258736811561e-06, "loss": 0.3375941514968872, "step": 22770 }, { "epoch": 1.0459343668565877, "grad_norm": 0.7291205525398254, "learning_rate": 7.009044532699294e-06, "loss": 0.2921276569366455, "step": 22780 }, { "epoch": 1.0465232689957746, "grad_norm": 0.5254185199737549, "learning_rate": 7.000832544353244e-06, "loss": 0.34774208068847656, "step": 22790 }, { "epoch": 1.0471121711349616, "grad_norm": 0.5430044531822205, "learning_rate": 6.992622777857035e-06, "loss": 0.27592105865478517, "step": 22800 }, { "epoch": 1.0477010732741487, "grad_norm": 0.6605882048606873, "learning_rate": 6.984415239292641e-06, "loss": 0.2969550848007202, "step": 22810 }, { "epoch": 1.0482899754133357, "grad_norm": 0.47405460476875305, "learning_rate": 6.976209934740392e-06, "loss": 0.258130407333374, "step": 22820 }, { "epoch": 1.0488788775525226, "grad_norm": 0.6499015092849731, "learning_rate": 6.968006870278959e-06, "loss": 0.3546860456466675, "step": 22830 }, { "epoch": 1.0494677796917098, "grad_norm": 0.6990050673484802, "learning_rate": 6.959806051985356e-06, "loss": 0.33871686458587646, "step": 22840 }, { "epoch": 1.0500566818308967, "grad_norm": 0.563175618648529, "learning_rate": 6.951607485934928e-06, "loss": 0.28433918952941895, "step": 22850 }, { "epoch": 1.0506455839700837, "grad_norm": 0.7286181449890137, "learning_rate": 6.943411178201355e-06, "loss": 0.2649394035339355, "step": 22860 }, { "epoch": 1.0512344861092708, "grad_norm": 0.5614933371543884, "learning_rate": 6.9352171348566466e-06, "loss": 0.31743242740631106, "step": 22870 }, { "epoch": 1.0518233882484578, "grad_norm": 1.1032007932662964, "learning_rate": 6.927025361971129e-06, "loss": 0.30576021671295167, "step": 22880 }, { "epoch": 1.052412290387645, "grad_norm": 0.6931600570678711, "learning_rate": 6.918835865613451e-06, "loss": 0.35455799102783203, "step": 22890 }, { "epoch": 1.0530011925268319, "grad_norm": 0.582574188709259, "learning_rate": 6.910648651850572e-06, "loss": 0.3298401594161987, "step": 22900 }, { "epoch": 1.0535900946660188, "grad_norm": 0.8043477535247803, "learning_rate": 6.902463726747765e-06, "loss": 0.2736107587814331, "step": 22910 }, { "epoch": 1.054178996805206, "grad_norm": 0.6644011735916138, "learning_rate": 6.8942810963686e-06, "loss": 0.3235773086547852, "step": 22920 }, { "epoch": 1.054767898944393, "grad_norm": 0.7287437915802002, "learning_rate": 6.886100766774954e-06, "loss": 0.33364484310150144, "step": 22930 }, { "epoch": 1.0553568010835799, "grad_norm": 0.762526273727417, "learning_rate": 6.877922744026999e-06, "loss": 0.3157224655151367, "step": 22940 }, { "epoch": 1.055945703222767, "grad_norm": 0.7220120429992676, "learning_rate": 6.869747034183189e-06, "loss": 0.3184772729873657, "step": 22950 }, { "epoch": 1.056534605361954, "grad_norm": 0.4332221746444702, "learning_rate": 6.8615736433002764e-06, "loss": 0.28147196769714355, "step": 22960 }, { "epoch": 1.057123507501141, "grad_norm": 0.778871476650238, "learning_rate": 6.853402577433291e-06, "loss": 0.3379077434539795, "step": 22970 }, { "epoch": 1.057712409640328, "grad_norm": 0.5467225909233093, "learning_rate": 6.845233842635537e-06, "loss": 0.28559274673461915, "step": 22980 }, { "epoch": 1.058301311779515, "grad_norm": 0.530658483505249, "learning_rate": 6.837067444958594e-06, "loss": 0.34005084037780764, "step": 22990 }, { "epoch": 1.058890213918702, "grad_norm": 0.5465559363365173, "learning_rate": 6.828903390452315e-06, "loss": 0.29611692428588865, "step": 23000 }, { "epoch": 1.0594791160578891, "grad_norm": 0.6153437495231628, "learning_rate": 6.8207416851648115e-06, "loss": 0.3055370092391968, "step": 23010 }, { "epoch": 1.060068018197076, "grad_norm": 0.7168208360671997, "learning_rate": 6.812582335142452e-06, "loss": 0.33273470401763916, "step": 23020 }, { "epoch": 1.060656920336263, "grad_norm": 0.7740054130554199, "learning_rate": 6.804425346429868e-06, "loss": 0.3207608461380005, "step": 23030 }, { "epoch": 1.0612458224754502, "grad_norm": 0.5297225117683411, "learning_rate": 6.796270725069938e-06, "loss": 0.30506162643432616, "step": 23040 }, { "epoch": 1.0618347246146371, "grad_norm": 0.6607385277748108, "learning_rate": 6.788118477103785e-06, "loss": 0.3192168712615967, "step": 23050 }, { "epoch": 1.0624236267538243, "grad_norm": 0.5932167768478394, "learning_rate": 6.779968608570773e-06, "loss": 0.3197173118591309, "step": 23060 }, { "epoch": 1.0630125288930112, "grad_norm": 0.5733965635299683, "learning_rate": 6.771821125508511e-06, "loss": 0.29614906311035155, "step": 23070 }, { "epoch": 1.0636014310321982, "grad_norm": 0.5505480766296387, "learning_rate": 6.763676033952833e-06, "loss": 0.306199049949646, "step": 23080 }, { "epoch": 1.0641903331713853, "grad_norm": 0.6366915106773376, "learning_rate": 6.755533339937801e-06, "loss": 0.30461385250091555, "step": 23090 }, { "epoch": 1.0647792353105723, "grad_norm": 0.46326422691345215, "learning_rate": 6.747393049495707e-06, "loss": 0.3038759231567383, "step": 23100 }, { "epoch": 1.0653681374497592, "grad_norm": 0.8641098141670227, "learning_rate": 6.7392551686570596e-06, "loss": 0.3320101022720337, "step": 23110 }, { "epoch": 1.0659570395889464, "grad_norm": 0.4770423173904419, "learning_rate": 6.731119703450577e-06, "loss": 0.3134412050247192, "step": 23120 }, { "epoch": 1.0665459417281333, "grad_norm": 0.4957725405693054, "learning_rate": 6.722986659903195e-06, "loss": 0.298116135597229, "step": 23130 }, { "epoch": 1.0671348438673203, "grad_norm": 0.5062552690505981, "learning_rate": 6.7148560440400546e-06, "loss": 0.3441824197769165, "step": 23140 }, { "epoch": 1.0677237460065074, "grad_norm": 0.5297567248344421, "learning_rate": 6.706727861884495e-06, "loss": 0.3127516984939575, "step": 23150 }, { "epoch": 1.0683126481456944, "grad_norm": 0.7307797074317932, "learning_rate": 6.698602119458053e-06, "loss": 0.32805323600769043, "step": 23160 }, { "epoch": 1.0689015502848813, "grad_norm": 0.6634035110473633, "learning_rate": 6.69047882278046e-06, "loss": 0.3173995018005371, "step": 23170 }, { "epoch": 1.0694904524240685, "grad_norm": 0.6504567265510559, "learning_rate": 6.682357977869635e-06, "loss": 0.3196385383605957, "step": 23180 }, { "epoch": 1.0700793545632554, "grad_norm": 0.4160931408405304, "learning_rate": 6.67423959074168e-06, "loss": 0.29849145412445066, "step": 23190 }, { "epoch": 1.0706682567024424, "grad_norm": 0.4582298696041107, "learning_rate": 6.666123667410875e-06, "loss": 0.30394299030303956, "step": 23200 }, { "epoch": 1.0712571588416295, "grad_norm": 0.5413811802864075, "learning_rate": 6.658010213889676e-06, "loss": 0.3259459972381592, "step": 23210 }, { "epoch": 1.0718460609808165, "grad_norm": 0.5159405469894409, "learning_rate": 6.6498992361887145e-06, "loss": 0.26974782943725584, "step": 23220 }, { "epoch": 1.0724349631200036, "grad_norm": 0.7687853574752808, "learning_rate": 6.641790740316778e-06, "loss": 0.32931056022644045, "step": 23230 }, { "epoch": 1.0730238652591906, "grad_norm": 0.5712857246398926, "learning_rate": 6.633684732280819e-06, "loss": 0.3145079851150513, "step": 23240 }, { "epoch": 1.0736127673983775, "grad_norm": 0.4860692620277405, "learning_rate": 6.625581218085954e-06, "loss": 0.31914145946502687, "step": 23250 }, { "epoch": 1.0742016695375647, "grad_norm": 0.6026865839958191, "learning_rate": 6.6174802037354415e-06, "loss": 0.3013422966003418, "step": 23260 }, { "epoch": 1.0747905716767516, "grad_norm": 0.8742293119430542, "learning_rate": 6.6093816952306945e-06, "loss": 0.3193085193634033, "step": 23270 }, { "epoch": 1.0753794738159386, "grad_norm": 0.5327329635620117, "learning_rate": 6.601285698571264e-06, "loss": 0.3271243333816528, "step": 23280 }, { "epoch": 1.0759683759551257, "grad_norm": 0.6938502192497253, "learning_rate": 6.593192219754853e-06, "loss": 0.32914984226226807, "step": 23290 }, { "epoch": 1.0765572780943127, "grad_norm": 0.4455547332763672, "learning_rate": 6.585101264777284e-06, "loss": 0.3265450239181519, "step": 23300 }, { "epoch": 1.0771461802334996, "grad_norm": 0.6295290589332581, "learning_rate": 6.577012839632513e-06, "loss": 0.3394952297210693, "step": 23310 }, { "epoch": 1.0777350823726868, "grad_norm": 0.5837503671646118, "learning_rate": 6.568926950312635e-06, "loss": 0.36922154426574705, "step": 23320 }, { "epoch": 1.0783239845118737, "grad_norm": 0.6192336678504944, "learning_rate": 6.560843602807847e-06, "loss": 0.3199045419692993, "step": 23330 }, { "epoch": 1.0789128866510607, "grad_norm": 0.6863105297088623, "learning_rate": 6.552762803106476e-06, "loss": 0.3045051097869873, "step": 23340 }, { "epoch": 1.0795017887902478, "grad_norm": 0.5284352898597717, "learning_rate": 6.544684557194959e-06, "loss": 0.30935277938842776, "step": 23350 }, { "epoch": 1.0800906909294348, "grad_norm": 0.7276309728622437, "learning_rate": 6.536608871057837e-06, "loss": 0.3371006488800049, "step": 23360 }, { "epoch": 1.0806795930686217, "grad_norm": 0.7430592179298401, "learning_rate": 6.528535750677758e-06, "loss": 0.32787470817565917, "step": 23370 }, { "epoch": 1.0812684952078089, "grad_norm": 0.5811182260513306, "learning_rate": 6.520465202035467e-06, "loss": 0.3073343515396118, "step": 23380 }, { "epoch": 1.0818573973469958, "grad_norm": 0.5074166059494019, "learning_rate": 6.512397231109816e-06, "loss": 0.36093127727508545, "step": 23390 }, { "epoch": 1.082446299486183, "grad_norm": 0.4022682309150696, "learning_rate": 6.5043318438777245e-06, "loss": 0.376980185508728, "step": 23400 }, { "epoch": 1.08303520162537, "grad_norm": 0.7705010175704956, "learning_rate": 6.496269046314214e-06, "loss": 0.3331691980361938, "step": 23410 }, { "epoch": 1.0836241037645569, "grad_norm": 0.537232518196106, "learning_rate": 6.488208844392392e-06, "loss": 0.36956846714019775, "step": 23420 }, { "epoch": 1.084213005903744, "grad_norm": 0.8245198726654053, "learning_rate": 6.480151244083423e-06, "loss": 0.33208930492401123, "step": 23430 }, { "epoch": 1.084801908042931, "grad_norm": 0.5468084216117859, "learning_rate": 6.472096251356561e-06, "loss": 0.34121453762054443, "step": 23440 }, { "epoch": 1.085390810182118, "grad_norm": 0.6673253774642944, "learning_rate": 6.464043872179124e-06, "loss": 0.3067692518234253, "step": 23450 }, { "epoch": 1.085979712321305, "grad_norm": 0.5695551037788391, "learning_rate": 6.455994112516499e-06, "loss": 0.2934034585952759, "step": 23460 }, { "epoch": 1.086568614460492, "grad_norm": 0.768214225769043, "learning_rate": 6.447946978332116e-06, "loss": 0.335164999961853, "step": 23470 }, { "epoch": 1.087157516599679, "grad_norm": 0.5016343593597412, "learning_rate": 6.439902475587474e-06, "loss": 0.34427461624145506, "step": 23480 }, { "epoch": 1.0877464187388661, "grad_norm": 0.6387108564376831, "learning_rate": 6.431860610242126e-06, "loss": 0.33103463649749754, "step": 23490 }, { "epoch": 1.088335320878053, "grad_norm": 0.5152999758720398, "learning_rate": 6.423821388253657e-06, "loss": 0.2968053579330444, "step": 23500 }, { "epoch": 1.08892422301724, "grad_norm": 0.6752516031265259, "learning_rate": 6.415784815577701e-06, "loss": 0.3069227457046509, "step": 23510 }, { "epoch": 1.0895131251564272, "grad_norm": 0.5172001123428345, "learning_rate": 6.407750898167933e-06, "loss": 0.2959017753601074, "step": 23520 }, { "epoch": 1.0901020272956141, "grad_norm": 0.7103534936904907, "learning_rate": 6.399719641976061e-06, "loss": 0.2916011571884155, "step": 23530 }, { "epoch": 1.090690929434801, "grad_norm": 0.45050686597824097, "learning_rate": 6.391691052951807e-06, "loss": 0.34113569259643556, "step": 23540 }, { "epoch": 1.0912798315739882, "grad_norm": 0.5195909142494202, "learning_rate": 6.383665137042936e-06, "loss": 0.321578311920166, "step": 23550 }, { "epoch": 1.0918687337131752, "grad_norm": 0.7008116245269775, "learning_rate": 6.375641900195226e-06, "loss": 0.30966293811798096, "step": 23560 }, { "epoch": 1.0924576358523623, "grad_norm": 0.4886181056499481, "learning_rate": 6.367621348352462e-06, "loss": 0.3510021686553955, "step": 23570 }, { "epoch": 1.0930465379915493, "grad_norm": 0.6226303577423096, "learning_rate": 6.359603487456451e-06, "loss": 0.323840594291687, "step": 23580 }, { "epoch": 1.0936354401307362, "grad_norm": 0.547340452671051, "learning_rate": 6.351588323447003e-06, "loss": 0.33883626461029054, "step": 23590 }, { "epoch": 1.0942243422699234, "grad_norm": 0.5152878165245056, "learning_rate": 6.343575862261933e-06, "loss": 0.36823582649230957, "step": 23600 }, { "epoch": 1.0948132444091103, "grad_norm": 0.4960673153400421, "learning_rate": 6.335566109837039e-06, "loss": 0.30619547367095945, "step": 23610 }, { "epoch": 1.0954021465482973, "grad_norm": 0.5171365141868591, "learning_rate": 6.327559072106131e-06, "loss": 0.3083407640457153, "step": 23620 }, { "epoch": 1.0959910486874844, "grad_norm": 0.5278074741363525, "learning_rate": 6.319554755001005e-06, "loss": 0.31797735691070556, "step": 23630 }, { "epoch": 1.0965799508266714, "grad_norm": 0.8184627294540405, "learning_rate": 6.311553164451425e-06, "loss": 0.3406806945800781, "step": 23640 }, { "epoch": 1.0971688529658583, "grad_norm": 0.7974048256874084, "learning_rate": 6.303554306385155e-06, "loss": 0.3064161777496338, "step": 23650 }, { "epoch": 1.0977577551050455, "grad_norm": 0.6364878416061401, "learning_rate": 6.295558186727929e-06, "loss": 0.31452116966247556, "step": 23660 }, { "epoch": 1.0983466572442324, "grad_norm": 0.8359082937240601, "learning_rate": 6.287564811403439e-06, "loss": 0.3450472831726074, "step": 23670 }, { "epoch": 1.0989355593834194, "grad_norm": 0.48772189021110535, "learning_rate": 6.279574186333366e-06, "loss": 0.3223247528076172, "step": 23680 }, { "epoch": 1.0995244615226065, "grad_norm": 0.5035157799720764, "learning_rate": 6.271586317437337e-06, "loss": 0.30733499526977537, "step": 23690 }, { "epoch": 1.1001133636617935, "grad_norm": 0.5382260680198669, "learning_rate": 6.263601210632946e-06, "loss": 0.3335845470428467, "step": 23700 }, { "epoch": 1.1007022658009804, "grad_norm": 0.6187599301338196, "learning_rate": 6.255618871835732e-06, "loss": 0.3489458322525024, "step": 23710 }, { "epoch": 1.1012911679401676, "grad_norm": 0.6221293807029724, "learning_rate": 6.247639306959193e-06, "loss": 0.31763813495635984, "step": 23720 }, { "epoch": 1.1018800700793545, "grad_norm": 0.502839982509613, "learning_rate": 6.239662521914771e-06, "loss": 0.2900780916213989, "step": 23730 }, { "epoch": 1.1024689722185417, "grad_norm": 0.47573405504226685, "learning_rate": 6.231688522611836e-06, "loss": 0.30693113803863525, "step": 23740 }, { "epoch": 1.1030578743577286, "grad_norm": 0.41235724091529846, "learning_rate": 6.22371731495771e-06, "loss": 0.31060075759887695, "step": 23750 }, { "epoch": 1.1036467764969156, "grad_norm": 0.7471031546592712, "learning_rate": 6.21574890485764e-06, "loss": 0.2891847133636475, "step": 23760 }, { "epoch": 1.1042356786361027, "grad_norm": 0.6671739220619202, "learning_rate": 6.207783298214804e-06, "loss": 0.28991239070892333, "step": 23770 }, { "epoch": 1.1048245807752897, "grad_norm": 0.6189566850662231, "learning_rate": 6.1998205009302935e-06, "loss": 0.32953386306762694, "step": 23780 }, { "epoch": 1.1054134829144766, "grad_norm": 0.675321102142334, "learning_rate": 6.19186051890313e-06, "loss": 0.27882912158966067, "step": 23790 }, { "epoch": 1.1060023850536638, "grad_norm": 0.5257890820503235, "learning_rate": 6.183903358030246e-06, "loss": 0.2885099411010742, "step": 23800 }, { "epoch": 1.1065912871928507, "grad_norm": 0.5491438508033752, "learning_rate": 6.175949024206478e-06, "loss": 0.3264081239700317, "step": 23810 }, { "epoch": 1.1071801893320377, "grad_norm": 0.5816881060600281, "learning_rate": 6.167997523324576e-06, "loss": 0.31781492233276365, "step": 23820 }, { "epoch": 1.1077690914712248, "grad_norm": 0.7377673387527466, "learning_rate": 6.1600488612751896e-06, "loss": 0.3776537418365479, "step": 23830 }, { "epoch": 1.1083579936104118, "grad_norm": 0.7539938688278198, "learning_rate": 6.152103043946864e-06, "loss": 0.26430230140686034, "step": 23840 }, { "epoch": 1.108946895749599, "grad_norm": 0.48345136642456055, "learning_rate": 6.144160077226035e-06, "loss": 0.3002410650253296, "step": 23850 }, { "epoch": 1.1095357978887859, "grad_norm": 0.4646276533603668, "learning_rate": 6.13621996699703e-06, "loss": 0.3147240877151489, "step": 23860 }, { "epoch": 1.1101247000279728, "grad_norm": 0.5888724327087402, "learning_rate": 6.12828271914206e-06, "loss": 0.33621978759765625, "step": 23870 }, { "epoch": 1.11071360216716, "grad_norm": 0.5602987408638, "learning_rate": 6.1203483395412115e-06, "loss": 0.2844818353652954, "step": 23880 }, { "epoch": 1.111302504306347, "grad_norm": 0.5588415265083313, "learning_rate": 6.11241683407245e-06, "loss": 0.3206324100494385, "step": 23890 }, { "epoch": 1.1118914064455339, "grad_norm": 0.5790587663650513, "learning_rate": 6.104488208611612e-06, "loss": 0.3176084518432617, "step": 23900 }, { "epoch": 1.112480308584721, "grad_norm": 0.6644494533538818, "learning_rate": 6.096562469032401e-06, "loss": 0.33560168743133545, "step": 23910 }, { "epoch": 1.113069210723908, "grad_norm": 0.5242338180541992, "learning_rate": 6.088639621206378e-06, "loss": 0.35024428367614746, "step": 23920 }, { "epoch": 1.113658112863095, "grad_norm": 0.5539028644561768, "learning_rate": 6.080719671002964e-06, "loss": 0.32883081436157224, "step": 23930 }, { "epoch": 1.114247015002282, "grad_norm": 0.454265832901001, "learning_rate": 6.072802624289437e-06, "loss": 0.3506674528121948, "step": 23940 }, { "epoch": 1.114835917141469, "grad_norm": 0.7922318577766418, "learning_rate": 6.064888486930918e-06, "loss": 0.30450680255889895, "step": 23950 }, { "epoch": 1.115424819280656, "grad_norm": 0.5599796772003174, "learning_rate": 6.056977264790376e-06, "loss": 0.3272010564804077, "step": 23960 }, { "epoch": 1.1160137214198431, "grad_norm": 0.6084268689155579, "learning_rate": 6.0490689637286215e-06, "loss": 0.3696732759475708, "step": 23970 }, { "epoch": 1.11660262355903, "grad_norm": 0.5000318288803101, "learning_rate": 6.041163589604299e-06, "loss": 0.3221790313720703, "step": 23980 }, { "epoch": 1.117191525698217, "grad_norm": 0.5816786885261536, "learning_rate": 6.033261148273886e-06, "loss": 0.2948575973510742, "step": 23990 }, { "epoch": 1.1177804278374042, "grad_norm": 0.5634832382202148, "learning_rate": 6.025361645591685e-06, "loss": 0.35864522457122805, "step": 24000 }, { "epoch": 1.118369329976591, "grad_norm": 0.6078681945800781, "learning_rate": 6.017465087409824e-06, "loss": 0.2866753816604614, "step": 24010 }, { "epoch": 1.1189582321157783, "grad_norm": 0.43836572766304016, "learning_rate": 6.009571479578247e-06, "loss": 0.30131232738494873, "step": 24020 }, { "epoch": 1.1195471342549652, "grad_norm": 0.5762682557106018, "learning_rate": 6.001680827944716e-06, "loss": 0.29266412258148194, "step": 24030 }, { "epoch": 1.1201360363941522, "grad_norm": 0.7147495746612549, "learning_rate": 5.993793138354801e-06, "loss": 0.35268733501434324, "step": 24040 }, { "epoch": 1.1207249385333393, "grad_norm": 0.5654435157775879, "learning_rate": 5.9859084166518755e-06, "loss": 0.3233843564987183, "step": 24050 }, { "epoch": 1.1213138406725263, "grad_norm": 0.6877713799476624, "learning_rate": 5.978026668677116e-06, "loss": 0.3154020071029663, "step": 24060 }, { "epoch": 1.1219027428117132, "grad_norm": 0.46290862560272217, "learning_rate": 5.970147900269498e-06, "loss": 0.3544290542602539, "step": 24070 }, { "epoch": 1.1224916449509004, "grad_norm": 0.6464409232139587, "learning_rate": 5.962272117265791e-06, "loss": 0.32313287258148193, "step": 24080 }, { "epoch": 1.1230805470900873, "grad_norm": 0.6726027131080627, "learning_rate": 5.954399325500545e-06, "loss": 0.32731194496154786, "step": 24090 }, { "epoch": 1.1236694492292743, "grad_norm": 0.6554591059684753, "learning_rate": 5.9465295308061e-06, "loss": 0.3153937101364136, "step": 24100 }, { "epoch": 1.1242583513684614, "grad_norm": 0.518043577671051, "learning_rate": 5.9386627390125785e-06, "loss": 0.28771989345550536, "step": 24110 }, { "epoch": 1.1248472535076484, "grad_norm": 0.4569878876209259, "learning_rate": 5.930798955947869e-06, "loss": 0.2728990316390991, "step": 24120 }, { "epoch": 1.1254361556468353, "grad_norm": 0.5845687389373779, "learning_rate": 5.922938187437639e-06, "loss": 0.33318471908569336, "step": 24130 }, { "epoch": 1.1260250577860225, "grad_norm": 0.6253288984298706, "learning_rate": 5.915080439305321e-06, "loss": 0.35851454734802246, "step": 24140 }, { "epoch": 1.1266139599252094, "grad_norm": 0.49539613723754883, "learning_rate": 5.907225717372109e-06, "loss": 0.27847607135772706, "step": 24150 }, { "epoch": 1.1272028620643963, "grad_norm": 0.7695813179016113, "learning_rate": 5.8993740274569545e-06, "loss": 0.350625205039978, "step": 24160 }, { "epoch": 1.1277917642035835, "grad_norm": 0.5830562710762024, "learning_rate": 5.891525375376563e-06, "loss": 0.31332595348358155, "step": 24170 }, { "epoch": 1.1283806663427705, "grad_norm": 0.5865841507911682, "learning_rate": 5.883679766945395e-06, "loss": 0.30213155746459963, "step": 24180 }, { "epoch": 1.1289695684819576, "grad_norm": 0.5040271282196045, "learning_rate": 5.875837207975645e-06, "loss": 0.3144997119903564, "step": 24190 }, { "epoch": 1.1295584706211446, "grad_norm": 0.5717757344245911, "learning_rate": 5.867997704277257e-06, "loss": 0.3481971025466919, "step": 24200 }, { "epoch": 1.1301473727603315, "grad_norm": 0.5343095064163208, "learning_rate": 5.860161261657911e-06, "loss": 0.34264750480651857, "step": 24210 }, { "epoch": 1.1307362748995187, "grad_norm": 0.7755244970321655, "learning_rate": 5.85232788592302e-06, "loss": 0.29542994499206543, "step": 24220 }, { "epoch": 1.1313251770387056, "grad_norm": 0.5947098135948181, "learning_rate": 5.8444975828757165e-06, "loss": 0.35787880420684814, "step": 24230 }, { "epoch": 1.1319140791778926, "grad_norm": 0.657024621963501, "learning_rate": 5.836670358316866e-06, "loss": 0.29777653217315675, "step": 24240 }, { "epoch": 1.1325029813170797, "grad_norm": 0.437287837266922, "learning_rate": 5.82884621804505e-06, "loss": 0.2766749143600464, "step": 24250 }, { "epoch": 1.1330918834562667, "grad_norm": 0.46983659267425537, "learning_rate": 5.821025167856564e-06, "loss": 0.349766206741333, "step": 24260 }, { "epoch": 1.1336807855954536, "grad_norm": 0.4579712450504303, "learning_rate": 5.81320721354542e-06, "loss": 0.2789180278778076, "step": 24270 }, { "epoch": 1.1342696877346408, "grad_norm": 0.6412621140480042, "learning_rate": 5.805392360903326e-06, "loss": 0.29716594219207765, "step": 24280 }, { "epoch": 1.1348585898738277, "grad_norm": 0.698083758354187, "learning_rate": 5.797580615719707e-06, "loss": 0.3372213840484619, "step": 24290 }, { "epoch": 1.1354474920130146, "grad_norm": 0.5666669607162476, "learning_rate": 5.789771983781669e-06, "loss": 0.33044726848602296, "step": 24300 }, { "epoch": 1.1360363941522018, "grad_norm": 0.6083332300186157, "learning_rate": 5.781966470874021e-06, "loss": 0.3456524610519409, "step": 24310 }, { "epoch": 1.1366252962913888, "grad_norm": 0.42675620317459106, "learning_rate": 5.7741640827792615e-06, "loss": 0.3084989070892334, "step": 24320 }, { "epoch": 1.1372141984305757, "grad_norm": 0.46465232968330383, "learning_rate": 5.766364825277575e-06, "loss": 0.3651388645172119, "step": 24330 }, { "epoch": 1.1378031005697629, "grad_norm": 0.5625426173210144, "learning_rate": 5.758568704146821e-06, "loss": 0.2985321044921875, "step": 24340 }, { "epoch": 1.1383920027089498, "grad_norm": 0.458469033241272, "learning_rate": 5.7507757251625465e-06, "loss": 0.29856698513031005, "step": 24350 }, { "epoch": 1.138980904848137, "grad_norm": 0.5663913488388062, "learning_rate": 5.742985894097953e-06, "loss": 0.3003609418869019, "step": 24360 }, { "epoch": 1.139569806987324, "grad_norm": 0.7425129413604736, "learning_rate": 5.7351992167239255e-06, "loss": 0.2998152256011963, "step": 24370 }, { "epoch": 1.1401587091265108, "grad_norm": 0.7187145948410034, "learning_rate": 5.727415698809008e-06, "loss": 0.3324389934539795, "step": 24380 }, { "epoch": 1.140747611265698, "grad_norm": 0.6350521445274353, "learning_rate": 5.719635346119401e-06, "loss": 0.33151707649230955, "step": 24390 }, { "epoch": 1.141336513404885, "grad_norm": 0.5882973670959473, "learning_rate": 5.711858164418965e-06, "loss": 0.3232302665710449, "step": 24400 }, { "epoch": 1.141925415544072, "grad_norm": 0.6505268216133118, "learning_rate": 5.704084159469208e-06, "loss": 0.38568212985992434, "step": 24410 }, { "epoch": 1.142514317683259, "grad_norm": 0.8239624500274658, "learning_rate": 5.696313337029289e-06, "loss": 0.34449918270111085, "step": 24420 }, { "epoch": 1.143103219822446, "grad_norm": 0.7531340718269348, "learning_rate": 5.688545702855999e-06, "loss": 0.3024947166442871, "step": 24430 }, { "epoch": 1.143692121961633, "grad_norm": 0.8131429553031921, "learning_rate": 5.680781262703776e-06, "loss": 0.3305517673492432, "step": 24440 }, { "epoch": 1.0002355608556748, "grad_norm": 0.5622875094413757, "learning_rate": 5.673020022324691e-06, "loss": 0.28127357363700867, "step": 24450 }, { "epoch": 1.0008244629948617, "grad_norm": 0.4450533390045166, "learning_rate": 5.665261987468443e-06, "loss": 0.27763540744781495, "step": 24460 }, { "epoch": 1.0014133651340489, "grad_norm": 0.4910314381122589, "learning_rate": 5.657507163882356e-06, "loss": 0.2885479688644409, "step": 24470 }, { "epoch": 1.0020022672732358, "grad_norm": 0.5840458869934082, "learning_rate": 5.649755557311377e-06, "loss": 0.25922451019287107, "step": 24480 }, { "epoch": 1.002591169412423, "grad_norm": 0.550125241279602, "learning_rate": 5.64200717349807e-06, "loss": 0.2895354747772217, "step": 24490 }, { "epoch": 1.00318007155161, "grad_norm": 0.580295205116272, "learning_rate": 5.634262018182604e-06, "loss": 0.27144062519073486, "step": 24500 }, { "epoch": 1.0037689736907969, "grad_norm": 0.7009913921356201, "learning_rate": 5.6265200971027635e-06, "loss": 0.3003673076629639, "step": 24510 }, { "epoch": 1.004357875829984, "grad_norm": 0.622144877910614, "learning_rate": 5.618781415993938e-06, "loss": 0.2740347385406494, "step": 24520 }, { "epoch": 1.004946777969171, "grad_norm": 0.7182214856147766, "learning_rate": 5.611045980589113e-06, "loss": 0.3216723918914795, "step": 24530 }, { "epoch": 1.005535680108358, "grad_norm": 0.5258731245994568, "learning_rate": 5.60331379661887e-06, "loss": 0.29696221351623536, "step": 24540 }, { "epoch": 1.006124582247545, "grad_norm": 0.6741461753845215, "learning_rate": 5.595584869811386e-06, "loss": 0.28220767974853517, "step": 24550 }, { "epoch": 1.006713484386732, "grad_norm": 0.5931569337844849, "learning_rate": 5.587859205892421e-06, "loss": 0.2665401458740234, "step": 24560 }, { "epoch": 1.007302386525919, "grad_norm": 0.7489706873893738, "learning_rate": 5.580136810585313e-06, "loss": 0.2929504156112671, "step": 24570 }, { "epoch": 1.0078912886651061, "grad_norm": 0.6351780891418457, "learning_rate": 5.572417689610987e-06, "loss": 0.2937645673751831, "step": 24580 }, { "epoch": 1.008480190804293, "grad_norm": 0.6483200788497925, "learning_rate": 5.56470184868794e-06, "loss": 0.2729827404022217, "step": 24590 }, { "epoch": 1.00906909294348, "grad_norm": 0.5925009846687317, "learning_rate": 5.556989293532238e-06, "loss": 0.3344477653503418, "step": 24600 }, { "epoch": 1.0096579950826672, "grad_norm": 0.5225246548652649, "learning_rate": 5.549280029857512e-06, "loss": 0.3000021457672119, "step": 24610 }, { "epoch": 1.0102468972218541, "grad_norm": 0.4749326705932617, "learning_rate": 5.541574063374957e-06, "loss": 0.2631241321563721, "step": 24620 }, { "epoch": 1.0108357993610413, "grad_norm": 0.6020548939704895, "learning_rate": 5.533871399793329e-06, "loss": 0.2831702947616577, "step": 24630 }, { "epoch": 1.0114247015002282, "grad_norm": 0.6574215888977051, "learning_rate": 5.526172044818923e-06, "loss": 0.2873628377914429, "step": 24640 }, { "epoch": 1.0120136036394152, "grad_norm": 0.5984800457954407, "learning_rate": 5.518476004155594e-06, "loss": 0.29934566020965575, "step": 24650 }, { "epoch": 1.0126025057786023, "grad_norm": 0.6633330583572388, "learning_rate": 5.51078328350474e-06, "loss": 0.31857099533081057, "step": 24660 }, { "epoch": 1.0131914079177893, "grad_norm": 0.6876775622367859, "learning_rate": 5.503093888565307e-06, "loss": 0.25982487201690674, "step": 24670 }, { "epoch": 1.0137803100569762, "grad_norm": 0.5286888480186462, "learning_rate": 5.495407825033761e-06, "loss": 0.27721240520477297, "step": 24680 }, { "epoch": 1.0143692121961634, "grad_norm": 0.7913944721221924, "learning_rate": 5.487725098604111e-06, "loss": 0.3153454065322876, "step": 24690 }, { "epoch": 1.0149581143353503, "grad_norm": 0.7203794121742249, "learning_rate": 5.480045714967894e-06, "loss": 0.27749617099761964, "step": 24700 }, { "epoch": 1.0155470164745373, "grad_norm": 0.5612097382545471, "learning_rate": 5.472369679814162e-06, "loss": 0.27747182846069335, "step": 24710 }, { "epoch": 1.0161359186137244, "grad_norm": 0.9052048325538635, "learning_rate": 5.464696998829496e-06, "loss": 0.252689790725708, "step": 24720 }, { "epoch": 1.0167248207529114, "grad_norm": 0.7302964925765991, "learning_rate": 5.457027677697988e-06, "loss": 0.29844748973846436, "step": 24730 }, { "epoch": 1.0173137228920983, "grad_norm": 0.5507508516311646, "learning_rate": 5.4493617221012405e-06, "loss": 0.2719395637512207, "step": 24740 }, { "epoch": 1.0179026250312855, "grad_norm": 0.6117476224899292, "learning_rate": 5.441699137718366e-06, "loss": 0.2446704626083374, "step": 24750 }, { "epoch": 1.0184915271704724, "grad_norm": 0.7470371723175049, "learning_rate": 5.434039930225975e-06, "loss": 0.25804843902587893, "step": 24760 }, { "epoch": 1.0190804293096594, "grad_norm": 0.5853258967399597, "learning_rate": 5.426384105298182e-06, "loss": 0.28151228427886965, "step": 24770 }, { "epoch": 1.0196693314488465, "grad_norm": 0.6286914944648743, "learning_rate": 5.418731668606587e-06, "loss": 0.2747290849685669, "step": 24780 }, { "epoch": 1.0202582335880335, "grad_norm": 0.7078003287315369, "learning_rate": 5.411082625820282e-06, "loss": 0.29234473705291747, "step": 24790 }, { "epoch": 1.0208471357272206, "grad_norm": 0.783917248249054, "learning_rate": 5.403436982605861e-06, "loss": 0.2684525728225708, "step": 24800 }, { "epoch": 1.0214360378664076, "grad_norm": 0.776978611946106, "learning_rate": 5.395794744627374e-06, "loss": 0.27434210777282714, "step": 24810 }, { "epoch": 1.0220249400055945, "grad_norm": 0.8138482570648193, "learning_rate": 5.388155917546365e-06, "loss": 0.2954927206039429, "step": 24820 }, { "epoch": 1.0226138421447817, "grad_norm": 0.7461369037628174, "learning_rate": 5.380520507021846e-06, "loss": 0.2799389123916626, "step": 24830 }, { "epoch": 1.0232027442839686, "grad_norm": 0.7602202892303467, "learning_rate": 5.372888518710303e-06, "loss": 0.2757270812988281, "step": 24840 }, { "epoch": 1.0237916464231556, "grad_norm": 0.5674859285354614, "learning_rate": 5.365259958265675e-06, "loss": 0.26570017337799073, "step": 24850 }, { "epoch": 1.0243805485623427, "grad_norm": 0.7815716862678528, "learning_rate": 5.357634831339369e-06, "loss": 0.27975509166717527, "step": 24860 }, { "epoch": 1.0249694507015297, "grad_norm": 0.6659267544746399, "learning_rate": 5.350013143580257e-06, "loss": 0.25829529762268066, "step": 24870 }, { "epoch": 1.0255583528407166, "grad_norm": 0.7585891485214233, "learning_rate": 5.3423949006346475e-06, "loss": 0.30030744075775145, "step": 24880 }, { "epoch": 1.0261472549799038, "grad_norm": 0.7286350727081299, "learning_rate": 5.334780108146306e-06, "loss": 0.2896328687667847, "step": 24890 }, { "epoch": 1.0267361571190907, "grad_norm": 0.7362012267112732, "learning_rate": 5.327168771756439e-06, "loss": 0.28926262855529783, "step": 24900 }, { "epoch": 1.0273250592582777, "grad_norm": 0.7092838287353516, "learning_rate": 5.319560897103697e-06, "loss": 0.26573054790496825, "step": 24910 }, { "epoch": 1.0279139613974648, "grad_norm": 0.804631233215332, "learning_rate": 5.311956489824153e-06, "loss": 0.29704172611236573, "step": 24920 }, { "epoch": 1.0285028635366518, "grad_norm": 0.5877637267112732, "learning_rate": 5.304355555551331e-06, "loss": 0.31557354927062986, "step": 24930 }, { "epoch": 1.0290917656758387, "grad_norm": 0.7902169227600098, "learning_rate": 5.296758099916171e-06, "loss": 0.31343631744384765, "step": 24940 }, { "epoch": 1.0296806678150259, "grad_norm": 0.47552689909935, "learning_rate": 5.28916412854703e-06, "loss": 0.27247645854949953, "step": 24950 }, { "epoch": 1.0302695699542128, "grad_norm": 0.6154338717460632, "learning_rate": 5.281573647069696e-06, "loss": 0.29145045280456544, "step": 24960 }, { "epoch": 1.0308584720934, "grad_norm": 0.5146239399909973, "learning_rate": 5.273986661107364e-06, "loss": 0.3055712699890137, "step": 24970 }, { "epoch": 1.031447374232587, "grad_norm": 0.610141932964325, "learning_rate": 5.266403176280646e-06, "loss": 0.2911428928375244, "step": 24980 }, { "epoch": 1.0320362763717739, "grad_norm": 0.6818845272064209, "learning_rate": 5.258823198207545e-06, "loss": 0.27402291297912595, "step": 24990 }, { "epoch": 1.032625178510961, "grad_norm": 0.9366025328636169, "learning_rate": 5.251246732503488e-06, "loss": 0.3217109203338623, "step": 25000 }, { "epoch": 1.033214080650148, "grad_norm": 0.725893497467041, "learning_rate": 5.2436737847812905e-06, "loss": 0.3119535207748413, "step": 25010 }, { "epoch": 1.033802982789335, "grad_norm": 0.6808173656463623, "learning_rate": 5.23610436065115e-06, "loss": 0.28316316604614256, "step": 25020 }, { "epoch": 1.034391884928522, "grad_norm": 0.7595789432525635, "learning_rate": 5.228538465720669e-06, "loss": 0.2617402791976929, "step": 25030 }, { "epoch": 1.034980787067709, "grad_norm": 0.549301266670227, "learning_rate": 5.2209761055948345e-06, "loss": 0.26310298442840574, "step": 25040 }, { "epoch": 1.035569689206896, "grad_norm": 0.6491830945014954, "learning_rate": 5.2134172858760005e-06, "loss": 0.3069706678390503, "step": 25050 }, { "epoch": 1.0361585913460831, "grad_norm": 0.9604558944702148, "learning_rate": 5.205862012163918e-06, "loss": 0.2713642597198486, "step": 25060 }, { "epoch": 1.03674749348527, "grad_norm": 0.7952796816825867, "learning_rate": 5.198310290055697e-06, "loss": 0.2822077512741089, "step": 25070 }, { "epoch": 1.037336395624457, "grad_norm": 0.926181435585022, "learning_rate": 5.190762125145827e-06, "loss": 0.33256509304046633, "step": 25080 }, { "epoch": 1.0379252977636442, "grad_norm": 0.6181250214576721, "learning_rate": 5.1832175230261495e-06, "loss": 0.3095921516418457, "step": 25090 }, { "epoch": 1.0385141999028311, "grad_norm": 0.6605007648468018, "learning_rate": 5.175676489285873e-06, "loss": 0.2769622802734375, "step": 25100 }, { "epoch": 1.0391031020420183, "grad_norm": 0.6269806027412415, "learning_rate": 5.1681390295115695e-06, "loss": 0.2971742391586304, "step": 25110 }, { "epoch": 1.0396920041812052, "grad_norm": 0.5881390571594238, "learning_rate": 5.1606051492871435e-06, "loss": 0.2731332302093506, "step": 25120 }, { "epoch": 1.0402809063203922, "grad_norm": 0.5107318758964539, "learning_rate": 5.1530748541938735e-06, "loss": 0.33083722591400144, "step": 25130 }, { "epoch": 1.0408698084595793, "grad_norm": 0.6016108989715576, "learning_rate": 5.145548149810365e-06, "loss": 0.2569439888000488, "step": 25140 }, { "epoch": 1.0414587105987663, "grad_norm": 0.81636643409729, "learning_rate": 5.138025041712572e-06, "loss": 0.2886439561843872, "step": 25150 }, { "epoch": 1.0420476127379532, "grad_norm": 0.5898492932319641, "learning_rate": 5.130505535473771e-06, "loss": 0.27481253147125245, "step": 25160 }, { "epoch": 1.0426365148771404, "grad_norm": 0.5956066846847534, "learning_rate": 5.122989636664585e-06, "loss": 0.33618409633636476, "step": 25170 }, { "epoch": 1.0432254170163273, "grad_norm": 0.6235300302505493, "learning_rate": 5.115477350852958e-06, "loss": 0.2945657253265381, "step": 25180 }, { "epoch": 1.0438143191555143, "grad_norm": 0.7785945534706116, "learning_rate": 5.1079686836041565e-06, "loss": 0.2678880453109741, "step": 25190 }, { "epoch": 1.0444032212947014, "grad_norm": 0.8227717280387878, "learning_rate": 5.100463640480768e-06, "loss": 0.3145641326904297, "step": 25200 }, { "epoch": 1.0449921234338884, "grad_norm": 0.7003596425056458, "learning_rate": 5.092962227042698e-06, "loss": 0.29204862117767333, "step": 25210 }, { "epoch": 1.0455810255730753, "grad_norm": 1.0225220918655396, "learning_rate": 5.085464448847161e-06, "loss": 0.30548458099365233, "step": 25220 }, { "epoch": 1.0461699277122625, "grad_norm": 0.8822780847549438, "learning_rate": 5.077970311448672e-06, "loss": 0.306138277053833, "step": 25230 }, { "epoch": 1.0467588298514494, "grad_norm": 0.8627164959907532, "learning_rate": 5.070479820399058e-06, "loss": 0.29824793338775635, "step": 25240 }, { "epoch": 1.0473477319906364, "grad_norm": 0.7551096081733704, "learning_rate": 5.06299298124744e-06, "loss": 0.2898961305618286, "step": 25250 }, { "epoch": 1.0479366341298235, "grad_norm": 0.5994674563407898, "learning_rate": 5.055509799540235e-06, "loss": 0.2494213819503784, "step": 25260 }, { "epoch": 1.0485255362690105, "grad_norm": 0.6916902661323547, "learning_rate": 5.048030280821149e-06, "loss": 0.3084700107574463, "step": 25270 }, { "epoch": 1.0491144384081976, "grad_norm": 0.5381240248680115, "learning_rate": 5.040554430631177e-06, "loss": 0.3123356342315674, "step": 25280 }, { "epoch": 1.0497033405473846, "grad_norm": 0.8144158124923706, "learning_rate": 5.033082254508599e-06, "loss": 0.30425314903259276, "step": 25290 }, { "epoch": 1.0502922426865715, "grad_norm": 0.7381141781806946, "learning_rate": 5.0256137579889606e-06, "loss": 0.2641162633895874, "step": 25300 }, { "epoch": 1.0508811448257587, "grad_norm": 0.7120150923728943, "learning_rate": 5.018148946605092e-06, "loss": 0.27290823459625246, "step": 25310 }, { "epoch": 1.0514700469649456, "grad_norm": 0.9150641560554504, "learning_rate": 5.010687825887095e-06, "loss": 0.3184216022491455, "step": 25320 }, { "epoch": 1.0520589491041326, "grad_norm": 0.7267884612083435, "learning_rate": 5.003230401362331e-06, "loss": 0.30739941596984866, "step": 25330 }, { "epoch": 1.0526478512433197, "grad_norm": 0.6238434314727783, "learning_rate": 4.995776678555427e-06, "loss": 0.3177881479263306, "step": 25340 }, { "epoch": 1.0532367533825067, "grad_norm": 0.6591129302978516, "learning_rate": 4.9883266629882635e-06, "loss": 0.29667866230010986, "step": 25350 }, { "epoch": 1.0538256555216936, "grad_norm": 0.8044952154159546, "learning_rate": 4.9808803601799834e-06, "loss": 0.3090502262115479, "step": 25360 }, { "epoch": 1.0544145576608808, "grad_norm": 0.5121888518333435, "learning_rate": 4.9734377756469655e-06, "loss": 0.2730462312698364, "step": 25370 }, { "epoch": 1.0550034598000677, "grad_norm": 0.6995940208435059, "learning_rate": 4.965998914902846e-06, "loss": 0.3240444898605347, "step": 25380 }, { "epoch": 1.0555923619392547, "grad_norm": 0.8209897875785828, "learning_rate": 4.958563783458496e-06, "loss": 0.3367152214050293, "step": 25390 }, { "epoch": 1.0561812640784418, "grad_norm": 0.5037044286727905, "learning_rate": 4.95113238682203e-06, "loss": 0.26214070320129396, "step": 25400 }, { "epoch": 1.0567701662176288, "grad_norm": 0.5413188338279724, "learning_rate": 4.943704730498786e-06, "loss": 0.2898594617843628, "step": 25410 }, { "epoch": 1.0573590683568157, "grad_norm": 0.5830031037330627, "learning_rate": 4.936280819991345e-06, "loss": 0.30105431079864503, "step": 25420 }, { "epoch": 1.0579479704960029, "grad_norm": 0.574720025062561, "learning_rate": 4.928860660799495e-06, "loss": 0.33500096797943113, "step": 25430 }, { "epoch": 1.0585368726351898, "grad_norm": 0.5556455850601196, "learning_rate": 4.921444258420259e-06, "loss": 0.306019926071167, "step": 25440 }, { "epoch": 1.059125774774377, "grad_norm": 0.922431230545044, "learning_rate": 4.914031618347873e-06, "loss": 0.2747491359710693, "step": 25450 }, { "epoch": 1.059714676913564, "grad_norm": 0.8595427870750427, "learning_rate": 4.906622746073785e-06, "loss": 0.2945817470550537, "step": 25460 }, { "epoch": 1.0603035790527509, "grad_norm": 0.5613863468170166, "learning_rate": 4.899217647086653e-06, "loss": 0.3142146348953247, "step": 25470 }, { "epoch": 1.060892481191938, "grad_norm": 0.6796558499336243, "learning_rate": 4.891816326872336e-06, "loss": 0.32175440788269044, "step": 25480 }, { "epoch": 1.061481383331125, "grad_norm": 0.7640790343284607, "learning_rate": 4.884418790913903e-06, "loss": 0.28779916763305663, "step": 25490 }, { "epoch": 1.062070285470312, "grad_norm": 0.7906290888786316, "learning_rate": 4.877025044691603e-06, "loss": 0.3044893741607666, "step": 25500 }, { "epoch": 1.062659187609499, "grad_norm": 0.4684810936450958, "learning_rate": 4.869635093682893e-06, "loss": 0.282194709777832, "step": 25510 }, { "epoch": 1.063248089748686, "grad_norm": 0.5431104898452759, "learning_rate": 4.862248943362412e-06, "loss": 0.31275110244750975, "step": 25520 }, { "epoch": 1.063836991887873, "grad_norm": 0.7206342220306396, "learning_rate": 4.854866599201985e-06, "loss": 0.2876091241836548, "step": 25530 }, { "epoch": 1.0644258940270601, "grad_norm": 0.5904025435447693, "learning_rate": 4.847488066670615e-06, "loss": 0.3051558256149292, "step": 25540 }, { "epoch": 1.065014796166247, "grad_norm": 0.46469229459762573, "learning_rate": 4.8401133512344844e-06, "loss": 0.27616205215454104, "step": 25550 }, { "epoch": 1.065603698305434, "grad_norm": 0.7174690365791321, "learning_rate": 4.832742458356948e-06, "loss": 0.33702080249786376, "step": 25560 }, { "epoch": 1.0661926004446212, "grad_norm": 0.8078405857086182, "learning_rate": 4.82537539349852e-06, "loss": 0.2909738779067993, "step": 25570 }, { "epoch": 1.066781502583808, "grad_norm": 0.5620206594467163, "learning_rate": 4.818012162116891e-06, "loss": 0.307749605178833, "step": 25580 }, { "epoch": 1.067370404722995, "grad_norm": 0.7037268280982971, "learning_rate": 4.810652769666903e-06, "loss": 0.318674373626709, "step": 25590 }, { "epoch": 1.0679593068621822, "grad_norm": 0.7540095448493958, "learning_rate": 4.803297221600561e-06, "loss": 0.3023289680480957, "step": 25600 }, { "epoch": 1.0685482090013692, "grad_norm": 0.5980381369590759, "learning_rate": 4.795945523367017e-06, "loss": 0.3087567090988159, "step": 25610 }, { "epoch": 1.0691371111405563, "grad_norm": 0.566725492477417, "learning_rate": 4.788597680412571e-06, "loss": 0.3205443859100342, "step": 25620 }, { "epoch": 1.0697260132797433, "grad_norm": 0.6397883892059326, "learning_rate": 4.781253698180674e-06, "loss": 0.31144301891326903, "step": 25630 }, { "epoch": 1.0703149154189302, "grad_norm": 0.538801372051239, "learning_rate": 4.773913582111901e-06, "loss": 0.27565417289733884, "step": 25640 }, { "epoch": 1.0709038175581174, "grad_norm": 0.5870261192321777, "learning_rate": 4.7665773376439785e-06, "loss": 0.30629069805145265, "step": 25650 }, { "epoch": 1.0714927196973043, "grad_norm": 0.537397027015686, "learning_rate": 4.759244970211755e-06, "loss": 0.2940568685531616, "step": 25660 }, { "epoch": 1.0720816218364912, "grad_norm": 0.7419536709785461, "learning_rate": 4.751916485247223e-06, "loss": 0.2710886001586914, "step": 25670 }, { "epoch": 1.0726705239756784, "grad_norm": 0.6621375679969788, "learning_rate": 4.744591888179475e-06, "loss": 0.3283437490463257, "step": 25680 }, { "epoch": 1.0732594261148654, "grad_norm": 0.5875841379165649, "learning_rate": 4.737271184434739e-06, "loss": 0.30051455497741697, "step": 25690 }, { "epoch": 1.0738483282540523, "grad_norm": 0.5750090479850769, "learning_rate": 4.729954379436359e-06, "loss": 0.30396370887756347, "step": 25700 }, { "epoch": 1.0744372303932395, "grad_norm": 0.5428311824798584, "learning_rate": 4.722641478604778e-06, "loss": 0.2787656784057617, "step": 25710 }, { "epoch": 1.0750261325324264, "grad_norm": 0.9091718196868896, "learning_rate": 4.71533248735756e-06, "loss": 0.33878886699676514, "step": 25720 }, { "epoch": 1.0756150346716133, "grad_norm": 0.6672152876853943, "learning_rate": 4.708027411109368e-06, "loss": 0.30420937538146975, "step": 25730 }, { "epoch": 1.0762039368108005, "grad_norm": 0.5834934711456299, "learning_rate": 4.700726255271964e-06, "loss": 0.31481308937072755, "step": 25740 }, { "epoch": 1.0767928389499875, "grad_norm": 0.6631849408149719, "learning_rate": 4.693429025254208e-06, "loss": 0.3298177242279053, "step": 25750 }, { "epoch": 1.0773817410891744, "grad_norm": 0.6906470060348511, "learning_rate": 4.686135726462049e-06, "loss": 0.32838120460510256, "step": 25760 }, { "epoch": 1.0779706432283616, "grad_norm": 0.5412715077400208, "learning_rate": 4.6788463642985275e-06, "loss": 0.33440361022949217, "step": 25770 }, { "epoch": 1.0785595453675485, "grad_norm": 0.6683573126792908, "learning_rate": 4.671560944163758e-06, "loss": 0.2955255746841431, "step": 25780 }, { "epoch": 1.0791484475067357, "grad_norm": 0.6698243021965027, "learning_rate": 4.664279471454941e-06, "loss": 0.31176769733428955, "step": 25790 }, { "epoch": 1.0797373496459226, "grad_norm": 0.5835884809494019, "learning_rate": 4.657001951566366e-06, "loss": 0.3118464946746826, "step": 25800 }, { "epoch": 1.0803262517851095, "grad_norm": 0.6649198532104492, "learning_rate": 4.649728389889368e-06, "loss": 0.3227553844451904, "step": 25810 }, { "epoch": 1.0809151539242967, "grad_norm": 0.672205924987793, "learning_rate": 4.642458791812369e-06, "loss": 0.3082414627075195, "step": 25820 }, { "epoch": 1.0815040560634837, "grad_norm": 0.6540539860725403, "learning_rate": 4.635193162720846e-06, "loss": 0.3438351392745972, "step": 25830 }, { "epoch": 1.0820929582026706, "grad_norm": 0.6704062819480896, "learning_rate": 4.627931507997344e-06, "loss": 0.3234559059143066, "step": 25840 }, { "epoch": 1.0826818603418578, "grad_norm": 0.532505452632904, "learning_rate": 4.620673833021451e-06, "loss": 0.3714778900146484, "step": 25850 }, { "epoch": 1.0832707624810447, "grad_norm": 0.7438023090362549, "learning_rate": 4.613420143169812e-06, "loss": 0.33481950759887696, "step": 25860 }, { "epoch": 1.0838596646202316, "grad_norm": 0.6871994733810425, "learning_rate": 4.606170443816134e-06, "loss": 0.3256321907043457, "step": 25870 }, { "epoch": 1.0844485667594188, "grad_norm": 0.7335498929023743, "learning_rate": 4.598924740331143e-06, "loss": 0.34684317111968993, "step": 25880 }, { "epoch": 1.0850374688986058, "grad_norm": 0.5655535459518433, "learning_rate": 4.591683038082621e-06, "loss": 0.3127690315246582, "step": 25890 }, { "epoch": 1.0856263710377927, "grad_norm": 0.7050876617431641, "learning_rate": 4.584445342435383e-06, "loss": 0.30012192726135256, "step": 25900 }, { "epoch": 1.0862152731769799, "grad_norm": 0.6334879398345947, "learning_rate": 4.577211658751277e-06, "loss": 0.2787178039550781, "step": 25910 }, { "epoch": 1.0868041753161668, "grad_norm": 0.6019802689552307, "learning_rate": 4.569981992389166e-06, "loss": 0.3336029052734375, "step": 25920 }, { "epoch": 1.0873930774553537, "grad_norm": 0.8068843483924866, "learning_rate": 4.562756348704959e-06, "loss": 0.3307683229446411, "step": 25930 }, { "epoch": 1.087981979594541, "grad_norm": 0.6089287996292114, "learning_rate": 4.555534733051572e-06, "loss": 0.28925504684448244, "step": 25940 }, { "epoch": 1.0885708817337278, "grad_norm": 0.703535258769989, "learning_rate": 4.548317150778931e-06, "loss": 0.3197308540344238, "step": 25950 }, { "epoch": 1.089159783872915, "grad_norm": 0.7335026264190674, "learning_rate": 4.541103607233986e-06, "loss": 0.27911858558654784, "step": 25960 }, { "epoch": 1.089748686012102, "grad_norm": 0.615825891494751, "learning_rate": 4.533894107760689e-06, "loss": 0.2822470188140869, "step": 25970 }, { "epoch": 1.090337588151289, "grad_norm": 0.8438145518302917, "learning_rate": 4.526688657700002e-06, "loss": 0.2851257801055908, "step": 25980 }, { "epoch": 1.090926490290476, "grad_norm": 0.698829174041748, "learning_rate": 4.51948726238987e-06, "loss": 0.3399609088897705, "step": 25990 }, { "epoch": 1.091515392429663, "grad_norm": 0.6200672388076782, "learning_rate": 4.512289927165259e-06, "loss": 0.2993948459625244, "step": 26000 }, { "epoch": 1.09210429456885, "grad_norm": 0.9168295860290527, "learning_rate": 4.505096657358113e-06, "loss": 0.3318209171295166, "step": 26010 }, { "epoch": 1.092693196708037, "grad_norm": 0.7209838628768921, "learning_rate": 4.4979074582973606e-06, "loss": 0.31190714836120603, "step": 26020 }, { "epoch": 1.093282098847224, "grad_norm": 0.8221926093101501, "learning_rate": 4.4907223353089225e-06, "loss": 0.3182440042495728, "step": 26030 }, { "epoch": 1.093871000986411, "grad_norm": 0.7001331448554993, "learning_rate": 4.483541293715699e-06, "loss": 0.34880547523498534, "step": 26040 }, { "epoch": 1.0944599031255982, "grad_norm": 0.7397459149360657, "learning_rate": 4.476364338837562e-06, "loss": 0.34119155406951907, "step": 26050 }, { "epoch": 1.095048805264785, "grad_norm": 0.5605281591415405, "learning_rate": 4.4691914759913614e-06, "loss": 0.2838308095932007, "step": 26060 }, { "epoch": 1.0956377074039723, "grad_norm": 0.8539541959762573, "learning_rate": 4.462022710490913e-06, "loss": 0.3153409481048584, "step": 26070 }, { "epoch": 1.0962266095431592, "grad_norm": 0.7119800448417664, "learning_rate": 4.454858047647001e-06, "loss": 0.3185281753540039, "step": 26080 }, { "epoch": 1.0968155116823461, "grad_norm": 0.4694675803184509, "learning_rate": 4.44769749276736e-06, "loss": 0.2974011659622192, "step": 26090 }, { "epoch": 1.0974044138215333, "grad_norm": 0.5466023683547974, "learning_rate": 4.440541051156689e-06, "loss": 0.29101314544677737, "step": 26100 }, { "epoch": 1.0979933159607203, "grad_norm": 0.7692428231239319, "learning_rate": 4.433388728116646e-06, "loss": 0.33603904247283933, "step": 26110 }, { "epoch": 1.0985822180999072, "grad_norm": 0.7249442338943481, "learning_rate": 4.426240528945818e-06, "loss": 0.31178836822509765, "step": 26120 }, { "epoch": 1.0991711202390944, "grad_norm": 1.0524152517318726, "learning_rate": 4.4190964589397604e-06, "loss": 0.32381210327148435, "step": 26130 }, { "epoch": 1.0997600223782813, "grad_norm": 0.695786714553833, "learning_rate": 4.411956523390956e-06, "loss": 0.30167274475097655, "step": 26140 }, { "epoch": 1.1003489245174682, "grad_norm": 0.6747229695320129, "learning_rate": 4.40482072758883e-06, "loss": 0.3232776165008545, "step": 26150 }, { "epoch": 1.1009378266566554, "grad_norm": 0.5771536231040955, "learning_rate": 4.397689076819732e-06, "loss": 0.3347535848617554, "step": 26160 }, { "epoch": 1.1015267287958423, "grad_norm": 0.6531325578689575, "learning_rate": 4.390561576366952e-06, "loss": 0.28511476516723633, "step": 26170 }, { "epoch": 1.1021156309350293, "grad_norm": 0.5513671040534973, "learning_rate": 4.383438231510698e-06, "loss": 0.2861856698989868, "step": 26180 }, { "epoch": 1.1027045330742165, "grad_norm": 0.6338734030723572, "learning_rate": 4.376319047528104e-06, "loss": 0.3249156713485718, "step": 26190 }, { "epoch": 1.1032934352134034, "grad_norm": 0.6746898293495178, "learning_rate": 4.369204029693219e-06, "loss": 0.264251184463501, "step": 26200 }, { "epoch": 1.1038823373525903, "grad_norm": 0.7121074199676514, "learning_rate": 4.3620931832770065e-06, "loss": 0.27768650054931643, "step": 26210 }, { "epoch": 1.1044712394917775, "grad_norm": 0.7930808067321777, "learning_rate": 4.354986513547341e-06, "loss": 0.2860716819763184, "step": 26220 }, { "epoch": 1.1050601416309644, "grad_norm": 0.6648890376091003, "learning_rate": 4.347884025768997e-06, "loss": 0.30508558750152587, "step": 26230 }, { "epoch": 1.1056490437701516, "grad_norm": 0.6324370503425598, "learning_rate": 4.340785725203658e-06, "loss": 0.2597169876098633, "step": 26240 }, { "epoch": 1.1062379459093385, "grad_norm": 0.6880820393562317, "learning_rate": 4.333691617109901e-06, "loss": 0.3092649459838867, "step": 26250 }, { "epoch": 1.1068268480485255, "grad_norm": 0.7014530301094055, "learning_rate": 4.3266017067432e-06, "loss": 0.30547988414764404, "step": 26260 }, { "epoch": 1.1074157501877127, "grad_norm": 0.5722478032112122, "learning_rate": 4.319515999355917e-06, "loss": 0.338121223449707, "step": 26270 }, { "epoch": 1.1080046523268996, "grad_norm": 0.5766258835792542, "learning_rate": 4.312434500197303e-06, "loss": 0.31493582725524905, "step": 26280 }, { "epoch": 1.1085935544660865, "grad_norm": 0.7003030776977539, "learning_rate": 4.305357214513493e-06, "loss": 0.2769856214523315, "step": 26290 }, { "epoch": 1.1091824566052737, "grad_norm": 0.6668327450752258, "learning_rate": 4.298284147547489e-06, "loss": 0.3052297353744507, "step": 26300 }, { "epoch": 1.1097713587444606, "grad_norm": 0.5238191485404968, "learning_rate": 4.2912153045391815e-06, "loss": 0.27200756072998045, "step": 26310 }, { "epoch": 1.1103602608836476, "grad_norm": 0.5877994894981384, "learning_rate": 4.284150690725324e-06, "loss": 0.33698043823242185, "step": 26320 }, { "epoch": 1.1109491630228348, "grad_norm": 0.5911795496940613, "learning_rate": 4.277090311339541e-06, "loss": 0.3167292833328247, "step": 26330 }, { "epoch": 1.1115380651620217, "grad_norm": 0.8861486315727234, "learning_rate": 4.270034171612319e-06, "loss": 0.3124183177947998, "step": 26340 }, { "epoch": 1.1121269673012086, "grad_norm": 0.4790663719177246, "learning_rate": 4.262982276770999e-06, "loss": 0.27609455585479736, "step": 26350 }, { "epoch": 1.1127158694403958, "grad_norm": 0.5968602895736694, "learning_rate": 4.25593463203979e-06, "loss": 0.3590679168701172, "step": 26360 }, { "epoch": 1.1133047715795827, "grad_norm": 0.7167752981185913, "learning_rate": 4.248891242639733e-06, "loss": 0.31891517639160155, "step": 26370 }, { "epoch": 1.1138936737187697, "grad_norm": 0.6181983351707458, "learning_rate": 4.241852113788733e-06, "loss": 0.3617612600326538, "step": 26380 }, { "epoch": 1.1144825758579568, "grad_norm": 0.726188600063324, "learning_rate": 4.2348172507015315e-06, "loss": 0.2740046977996826, "step": 26390 }, { "epoch": 1.1150714779971438, "grad_norm": 0.6762018799781799, "learning_rate": 4.227786658589712e-06, "loss": 0.32732791900634767, "step": 26400 }, { "epoch": 1.115660380136331, "grad_norm": 0.6611469388008118, "learning_rate": 4.220760342661692e-06, "loss": 0.307441258430481, "step": 26410 }, { "epoch": 1.116249282275518, "grad_norm": 0.5329198837280273, "learning_rate": 4.213738308122728e-06, "loss": 0.3374851942062378, "step": 26420 }, { "epoch": 1.1168381844147048, "grad_norm": 0.5626828074455261, "learning_rate": 4.206720560174889e-06, "loss": 0.334534740447998, "step": 26430 }, { "epoch": 1.117427086553892, "grad_norm": 0.8568176627159119, "learning_rate": 4.199707104017086e-06, "loss": 0.28077595233917235, "step": 26440 }, { "epoch": 1.118015988693079, "grad_norm": 0.589466392993927, "learning_rate": 4.19269794484504e-06, "loss": 0.3369838953018188, "step": 26450 }, { "epoch": 1.1186048908322659, "grad_norm": 0.5603137612342834, "learning_rate": 4.185693087851292e-06, "loss": 0.27100162506103515, "step": 26460 }, { "epoch": 1.119193792971453, "grad_norm": 0.5019323825836182, "learning_rate": 4.178692538225197e-06, "loss": 0.29223344326019285, "step": 26470 }, { "epoch": 1.11978269511064, "grad_norm": 0.8067333102226257, "learning_rate": 4.171696301152917e-06, "loss": 0.3071276187896729, "step": 26480 }, { "epoch": 1.120371597249827, "grad_norm": 0.6087183356285095, "learning_rate": 4.164704381817423e-06, "loss": 0.34236392974853513, "step": 26490 }, { "epoch": 1.120960499389014, "grad_norm": 0.8120219111442566, "learning_rate": 4.157716785398477e-06, "loss": 0.31238837242126466, "step": 26500 }, { "epoch": 1.121549401528201, "grad_norm": 0.5562658905982971, "learning_rate": 4.1507335170726494e-06, "loss": 0.3191776990890503, "step": 26510 }, { "epoch": 1.122138303667388, "grad_norm": 0.6812038421630859, "learning_rate": 4.1437545820133e-06, "loss": 0.31237552165985105, "step": 26520 }, { "epoch": 1.1227272058065751, "grad_norm": 0.7029443383216858, "learning_rate": 4.136779985390578e-06, "loss": 0.31951777935028075, "step": 26530 }, { "epoch": 1.123316107945762, "grad_norm": 0.5407879948616028, "learning_rate": 4.129809732371422e-06, "loss": 0.3288877487182617, "step": 26540 }, { "epoch": 1.123905010084949, "grad_norm": 0.7316825985908508, "learning_rate": 4.1228438281195484e-06, "loss": 0.2884044647216797, "step": 26550 }, { "epoch": 1.1244939122241362, "grad_norm": 0.5215107798576355, "learning_rate": 4.115882277795458e-06, "loss": 0.2680527925491333, "step": 26560 }, { "epoch": 1.1250828143633231, "grad_norm": 0.6792302131652832, "learning_rate": 4.108925086556414e-06, "loss": 0.28067288398742674, "step": 26570 }, { "epoch": 1.1256717165025103, "grad_norm": 0.6256664395332336, "learning_rate": 4.101972259556462e-06, "loss": 0.37816543579101564, "step": 26580 }, { "epoch": 1.1262606186416972, "grad_norm": 0.6737295389175415, "learning_rate": 4.095023801946409e-06, "loss": 0.30282595157623293, "step": 26590 }, { "epoch": 1.1268495207808842, "grad_norm": 0.8297234177589417, "learning_rate": 4.088079718873831e-06, "loss": 0.27921864986419676, "step": 26600 }, { "epoch": 1.1274384229200713, "grad_norm": 0.6850873827934265, "learning_rate": 4.081140015483055e-06, "loss": 0.33476934432983396, "step": 26610 }, { "epoch": 1.1280273250592583, "grad_norm": 0.6007938385009766, "learning_rate": 4.074204696915168e-06, "loss": 0.27222981452941897, "step": 26620 }, { "epoch": 1.1286162271984452, "grad_norm": 0.6595368385314941, "learning_rate": 4.067273768308013e-06, "loss": 0.3059885025024414, "step": 26630 }, { "epoch": 1.1292051293376324, "grad_norm": 0.6208631992340088, "learning_rate": 4.060347234796167e-06, "loss": 0.3171453714370728, "step": 26640 }, { "epoch": 1.1297940314768193, "grad_norm": 0.660276472568512, "learning_rate": 4.0534251015109645e-06, "loss": 0.34055938720703127, "step": 26650 }, { "epoch": 1.1303829336160063, "grad_norm": 0.6752126812934875, "learning_rate": 4.046507373580471e-06, "loss": 0.31766514778137206, "step": 26660 }, { "epoch": 1.1309718357551934, "grad_norm": 0.8771119117736816, "learning_rate": 4.039594056129505e-06, "loss": 0.3189789056777954, "step": 26670 }, { "epoch": 1.1315607378943804, "grad_norm": 0.6773837208747864, "learning_rate": 4.032685154279595e-06, "loss": 0.3093159914016724, "step": 26680 }, { "epoch": 1.1321496400335673, "grad_norm": 0.5318689942359924, "learning_rate": 4.02578067314901e-06, "loss": 0.2748838424682617, "step": 26690 }, { "epoch": 1.1327385421727545, "grad_norm": 0.632626473903656, "learning_rate": 4.01888061785275e-06, "loss": 0.30108516216278075, "step": 26700 }, { "epoch": 1.1333274443119414, "grad_norm": 0.5742058157920837, "learning_rate": 4.0119849935025175e-06, "loss": 0.3046391248703003, "step": 26710 }, { "epoch": 1.1339163464511284, "grad_norm": 0.6403510570526123, "learning_rate": 4.0050938052067526e-06, "loss": 0.2891948938369751, "step": 26720 }, { "epoch": 1.1345052485903155, "grad_norm": 0.6658622622489929, "learning_rate": 3.9982070580705935e-06, "loss": 0.31296353340148925, "step": 26730 }, { "epoch": 1.1350941507295025, "grad_norm": 0.5680436491966248, "learning_rate": 3.991324757195907e-06, "loss": 0.305817985534668, "step": 26740 }, { "epoch": 1.1356830528686896, "grad_norm": 0.6361669898033142, "learning_rate": 3.984446907681244e-06, "loss": 0.32267470359802247, "step": 26750 }, { "epoch": 1.1362719550078766, "grad_norm": 0.7127870321273804, "learning_rate": 3.977573514621872e-06, "loss": 0.3326195001602173, "step": 26760 }, { "epoch": 1.1368608571470635, "grad_norm": 0.7166005373001099, "learning_rate": 3.970704583109755e-06, "loss": 0.32363865375518797, "step": 26770 }, { "epoch": 1.1374497592862507, "grad_norm": 0.6365798115730286, "learning_rate": 3.963840118233547e-06, "loss": 0.3234691143035889, "step": 26780 }, { "epoch": 1.1380386614254376, "grad_norm": 0.7349017858505249, "learning_rate": 3.956980125078591e-06, "loss": 0.2993540525436401, "step": 26790 }, { "epoch": 1.1386275635646246, "grad_norm": 0.5747594237327576, "learning_rate": 3.950124608726936e-06, "loss": 0.2867851734161377, "step": 26800 }, { "epoch": 1.1392164657038117, "grad_norm": 0.9707556366920471, "learning_rate": 3.943273574257289e-06, "loss": 0.289974308013916, "step": 26810 }, { "epoch": 1.1398053678429987, "grad_norm": 0.6085373163223267, "learning_rate": 3.936427026745053e-06, "loss": 0.2971749782562256, "step": 26820 }, { "epoch": 1.1403942699821856, "grad_norm": 0.5515556931495667, "learning_rate": 3.929584971262301e-06, "loss": 0.33025825023651123, "step": 26830 }, { "epoch": 1.1409831721213728, "grad_norm": 0.6236960887908936, "learning_rate": 3.922747412877783e-06, "loss": 0.3248487949371338, "step": 26840 }, { "epoch": 1.1415720742605597, "grad_norm": 0.7806437015533447, "learning_rate": 3.915914356656909e-06, "loss": 0.34444098472595214, "step": 26850 }, { "epoch": 1.142160976399747, "grad_norm": 0.7305387854576111, "learning_rate": 3.909085807661757e-06, "loss": 0.34127647876739503, "step": 26860 }, { "epoch": 1.1427498785389338, "grad_norm": 0.6881123185157776, "learning_rate": 3.902261770951079e-06, "loss": 0.3287925720214844, "step": 26870 }, { "epoch": 1.1433387806781208, "grad_norm": 0.6076834201812744, "learning_rate": 3.895442251580262e-06, "loss": 0.2787143707275391, "step": 26880 }, { "epoch": 1.1439276828173077, "grad_norm": 0.6119790077209473, "learning_rate": 3.888627254601361e-06, "loss": 0.3082557439804077, "step": 26890 }, { "epoch": 1.1445165849564949, "grad_norm": 0.46914830803871155, "learning_rate": 3.881816785063076e-06, "loss": 0.28409082889556886, "step": 26900 }, { "epoch": 1.1451054870956818, "grad_norm": 0.6486103534698486, "learning_rate": 3.8750108480107596e-06, "loss": 0.36963138580322263, "step": 26910 }, { "epoch": 1.145694389234869, "grad_norm": 0.49818629026412964, "learning_rate": 3.86820944848639e-06, "loss": 0.34208219051361083, "step": 26920 }, { "epoch": 1.146283291374056, "grad_norm": 0.651638388633728, "learning_rate": 3.861412591528605e-06, "loss": 0.2964688062667847, "step": 26930 }, { "epoch": 1.1468721935132429, "grad_norm": 0.7344010472297668, "learning_rate": 3.854620282172666e-06, "loss": 0.34241247177124023, "step": 26940 }, { "epoch": 1.14746109565243, "grad_norm": 0.5281785726547241, "learning_rate": 3.847832525450462e-06, "loss": 0.3126488924026489, "step": 26950 }, { "epoch": 1.148049997791617, "grad_norm": 0.5781857967376709, "learning_rate": 3.841049326390516e-06, "loss": 0.31527388095855713, "step": 26960 }, { "epoch": 1.148638899930804, "grad_norm": 0.5721484422683716, "learning_rate": 3.834270690017972e-06, "loss": 0.32687599658966066, "step": 26970 }, { "epoch": 1.149227802069991, "grad_norm": 0.8626046776771545, "learning_rate": 3.827496621354599e-06, "loss": 0.3376193046569824, "step": 26980 }, { "epoch": 1.149816704209178, "grad_norm": 0.7687717080116272, "learning_rate": 3.820727125418766e-06, "loss": 0.3190311431884766, "step": 26990 }, { "epoch": 1.150405606348365, "grad_norm": 0.641394853591919, "learning_rate": 3.8139622072254777e-06, "loss": 0.2834547281265259, "step": 27000 }, { "epoch": 1.1509945084875521, "grad_norm": 0.4911433458328247, "learning_rate": 3.8072018717863344e-06, "loss": 0.33786160945892335, "step": 27010 }, { "epoch": 1.151583410626739, "grad_norm": 0.5497865080833435, "learning_rate": 3.800446124109536e-06, "loss": 0.2840303897857666, "step": 27020 }, { "epoch": 1.1521723127659262, "grad_norm": 0.5705615282058716, "learning_rate": 3.7936949691998935e-06, "loss": 0.3448507785797119, "step": 27030 }, { "epoch": 1.1527612149051132, "grad_norm": 0.810813844203949, "learning_rate": 3.786948412058814e-06, "loss": 0.32787346839904785, "step": 27040 }, { "epoch": 1.1533501170443001, "grad_norm": 0.6585190296173096, "learning_rate": 3.7802064576842945e-06, "loss": 0.34873712062835693, "step": 27050 }, { "epoch": 1.153939019183487, "grad_norm": 0.7414364814758301, "learning_rate": 3.7734691110709265e-06, "loss": 0.31992521286010744, "step": 27060 }, { "epoch": 1.1545279213226742, "grad_norm": 0.7960109710693359, "learning_rate": 3.766736377209884e-06, "loss": 0.31913089752197266, "step": 27070 }, { "epoch": 1.1551168234618612, "grad_norm": 0.7899366617202759, "learning_rate": 3.7600082610889323e-06, "loss": 0.3507037878036499, "step": 27080 }, { "epoch": 1.1557057256010483, "grad_norm": 0.5075811743736267, "learning_rate": 3.753284767692398e-06, "loss": 0.31864151954650877, "step": 27090 }, { "epoch": 1.1562946277402353, "grad_norm": 0.5999122262001038, "learning_rate": 3.746565902001201e-06, "loss": 0.28962907791137693, "step": 27100 }, { "epoch": 1.1568835298794222, "grad_norm": 0.5132626295089722, "learning_rate": 3.739851668992829e-06, "loss": 0.30971264839172363, "step": 27110 }, { "epoch": 1.1574724320186094, "grad_norm": 0.5660801529884338, "learning_rate": 3.7331420736413237e-06, "loss": 0.3197577238082886, "step": 27120 }, { "epoch": 1.1580613341577963, "grad_norm": 0.6729997396469116, "learning_rate": 3.7264371209173134e-06, "loss": 0.3216303110122681, "step": 27130 }, { "epoch": 1.1586502362969833, "grad_norm": 0.8890305161476135, "learning_rate": 3.7197368157879723e-06, "loss": 0.3366010427474976, "step": 27140 }, { "epoch": 1.1592391384361704, "grad_norm": 0.6642903685569763, "learning_rate": 3.7130411632170384e-06, "loss": 0.3430779218673706, "step": 27150 }, { "epoch": 1.1598280405753574, "grad_norm": 0.5953430533409119, "learning_rate": 3.706350168164792e-06, "loss": 0.2733257293701172, "step": 27160 }, { "epoch": 1.1604169427145443, "grad_norm": 0.5659530758857727, "learning_rate": 3.6996638355880765e-06, "loss": 0.32862677574157717, "step": 27170 }, { "epoch": 1.1610058448537315, "grad_norm": 0.580048680305481, "learning_rate": 3.6929821704402746e-06, "loss": 0.34980101585388185, "step": 27180 }, { "epoch": 1.1615947469929184, "grad_norm": 0.7235574126243591, "learning_rate": 3.686305177671311e-06, "loss": 0.3101292371749878, "step": 27190 }, { "epoch": 1.1621836491321056, "grad_norm": 0.554358184337616, "learning_rate": 3.679632862227651e-06, "loss": 0.2799635887145996, "step": 27200 }, { "epoch": 1.1627725512712925, "grad_norm": 0.7577160596847534, "learning_rate": 3.6729652290522956e-06, "loss": 0.2949601888656616, "step": 27210 }, { "epoch": 1.1633614534104795, "grad_norm": 0.47828516364097595, "learning_rate": 3.666302283084776e-06, "loss": 0.2721676826477051, "step": 27220 }, { "epoch": 1.1639503555496664, "grad_norm": 0.6781826019287109, "learning_rate": 3.6596440292611458e-06, "loss": 0.34446425437927247, "step": 27230 }, { "epoch": 1.1645392576888536, "grad_norm": 0.5782594084739685, "learning_rate": 3.65299047251399e-06, "loss": 0.297525429725647, "step": 27240 }, { "epoch": 1.1651281598280405, "grad_norm": 0.5911884903907776, "learning_rate": 3.646341617772411e-06, "loss": 0.3320277452468872, "step": 27250 }, { "epoch": 1.1657170619672277, "grad_norm": 0.4564284086227417, "learning_rate": 3.639697469962028e-06, "loss": 0.29341826438903806, "step": 27260 }, { "epoch": 1.1663059641064146, "grad_norm": 0.7168318033218384, "learning_rate": 3.633058034004974e-06, "loss": 0.3244031429290771, "step": 27270 }, { "epoch": 1.1668948662456016, "grad_norm": 0.6190764904022217, "learning_rate": 3.626423314819889e-06, "loss": 0.30729169845581056, "step": 27280 }, { "epoch": 1.1674837683847887, "grad_norm": 0.47839635610580444, "learning_rate": 3.6197933173219245e-06, "loss": 0.28266737461090086, "step": 27290 }, { "epoch": 1.1680726705239757, "grad_norm": 0.9325916767120361, "learning_rate": 3.6131680464227236e-06, "loss": 0.3126579523086548, "step": 27300 }, { "epoch": 1.1686615726631626, "grad_norm": 0.5719999074935913, "learning_rate": 3.606547507030438e-06, "loss": 0.3391671657562256, "step": 27310 }, { "epoch": 1.1692504748023498, "grad_norm": 0.5883305072784424, "learning_rate": 3.5999317040497084e-06, "loss": 0.3124718189239502, "step": 27320 }, { "epoch": 1.1698393769415367, "grad_norm": 0.5833786129951477, "learning_rate": 3.593320642381669e-06, "loss": 0.3259094476699829, "step": 27330 }, { "epoch": 1.1704282790807237, "grad_norm": 0.7595736980438232, "learning_rate": 3.5867143269239423e-06, "loss": 0.2999396800994873, "step": 27340 }, { "epoch": 1.1710171812199108, "grad_norm": 0.5982180833816528, "learning_rate": 3.580112762570633e-06, "loss": 0.33542232513427733, "step": 27350 }, { "epoch": 1.1716060833590978, "grad_norm": 0.5541450381278992, "learning_rate": 3.573515954212329e-06, "loss": 0.2866466522216797, "step": 27360 }, { "epoch": 1.172194985498285, "grad_norm": 0.5894819498062134, "learning_rate": 3.566923906736087e-06, "loss": 0.3343859910964966, "step": 27370 }, { "epoch": 1.1727838876374719, "grad_norm": 0.6631375551223755, "learning_rate": 3.5603366250254447e-06, "loss": 0.36668701171875, "step": 27380 }, { "epoch": 1.1733727897766588, "grad_norm": 0.6598944067955017, "learning_rate": 3.553754113960406e-06, "loss": 0.2992446184158325, "step": 27390 }, { "epoch": 1.1739616919158458, "grad_norm": 0.6899659037590027, "learning_rate": 3.5471763784174417e-06, "loss": 0.29158403873443606, "step": 27400 }, { "epoch": 1.174550594055033, "grad_norm": 0.7143812775611877, "learning_rate": 3.540603423269485e-06, "loss": 0.40007505416870115, "step": 27410 }, { "epoch": 1.1751394961942199, "grad_norm": 0.6883352398872375, "learning_rate": 3.5340352533859267e-06, "loss": 0.3096306562423706, "step": 27420 }, { "epoch": 1.175728398333407, "grad_norm": 0.6358159184455872, "learning_rate": 3.5274718736326163e-06, "loss": 0.27739121913909914, "step": 27430 }, { "epoch": 1.176317300472594, "grad_norm": 0.6383141279220581, "learning_rate": 3.5209132888718446e-06, "loss": 0.3265470266342163, "step": 27440 }, { "epoch": 1.176906202611781, "grad_norm": 0.6109476089477539, "learning_rate": 3.514359503962359e-06, "loss": 0.28966987133026123, "step": 27450 }, { "epoch": 1.177495104750968, "grad_norm": 0.5153137445449829, "learning_rate": 3.507810523759351e-06, "loss": 0.31290085315704347, "step": 27460 }, { "epoch": 1.178084006890155, "grad_norm": 0.5001354217529297, "learning_rate": 3.5012663531144473e-06, "loss": 0.27049684524536133, "step": 27470 }, { "epoch": 1.178672909029342, "grad_norm": 0.7726515531539917, "learning_rate": 3.4947269968757193e-06, "loss": 0.3098325490951538, "step": 27480 }, { "epoch": 1.1792618111685291, "grad_norm": 0.5916563272476196, "learning_rate": 3.488192459887667e-06, "loss": 0.28933300971984866, "step": 27490 }, { "epoch": 1.179850713307716, "grad_norm": 0.6921252012252808, "learning_rate": 3.4816627469912147e-06, "loss": 0.31070187091827395, "step": 27500 }, { "epoch": 1.180439615446903, "grad_norm": 0.8219258189201355, "learning_rate": 3.475137863023723e-06, "loss": 0.36709530353546144, "step": 27510 }, { "epoch": 1.1810285175860902, "grad_norm": 0.729004979133606, "learning_rate": 3.4686178128189694e-06, "loss": 0.3167080879211426, "step": 27520 }, { "epoch": 1.1816174197252771, "grad_norm": 0.7125288248062134, "learning_rate": 3.4621026012071536e-06, "loss": 0.3163491725921631, "step": 27530 }, { "epoch": 1.1822063218644643, "grad_norm": 0.7269718050956726, "learning_rate": 3.455592233014887e-06, "loss": 0.3152503490447998, "step": 27540 }, { "epoch": 1.1827952240036512, "grad_norm": 0.48789188265800476, "learning_rate": 3.4490867130651962e-06, "loss": 0.31578314304351807, "step": 27550 }, { "epoch": 1.1833841261428382, "grad_norm": 0.690241277217865, "learning_rate": 3.4425860461775185e-06, "loss": 0.2932302474975586, "step": 27560 }, { "epoch": 1.1839730282820253, "grad_norm": 0.6941596865653992, "learning_rate": 3.4360902371676842e-06, "loss": 0.31160056591033936, "step": 27570 }, { "epoch": 1.1845619304212123, "grad_norm": 0.7184134721755981, "learning_rate": 3.429599290847937e-06, "loss": 0.35072038173675535, "step": 27580 }, { "epoch": 1.1851508325603992, "grad_norm": 0.5496609807014465, "learning_rate": 3.4231132120269106e-06, "loss": 0.3090402364730835, "step": 27590 }, { "epoch": 1.1857397346995864, "grad_norm": 0.5173737406730652, "learning_rate": 3.4166320055096457e-06, "loss": 0.3027217864990234, "step": 27600 }, { "epoch": 1.1863286368387733, "grad_norm": 0.5972543954849243, "learning_rate": 3.410155676097555e-06, "loss": 0.315409255027771, "step": 27610 }, { "epoch": 1.1869175389779603, "grad_norm": 0.5575653910636902, "learning_rate": 3.403684228588451e-06, "loss": 0.30985302925109864, "step": 27620 }, { "epoch": 1.1875064411171474, "grad_norm": 0.5498601794242859, "learning_rate": 3.397217667776527e-06, "loss": 0.29956443309783937, "step": 27630 }, { "epoch": 1.1880953432563344, "grad_norm": 0.7835243344306946, "learning_rate": 3.3907559984523487e-06, "loss": 0.2873212814331055, "step": 27640 }, { "epoch": 1.1886842453955213, "grad_norm": 0.5482907891273499, "learning_rate": 3.3842992254028696e-06, "loss": 0.29457950592041016, "step": 27650 }, { "epoch": 1.1892731475347085, "grad_norm": 0.6760403513908386, "learning_rate": 3.377847353411404e-06, "loss": 0.3576001405715942, "step": 27660 }, { "epoch": 1.1898620496738954, "grad_norm": 0.5523601770401001, "learning_rate": 3.371400387257654e-06, "loss": 0.3157463312149048, "step": 27670 }, { "epoch": 1.1904509518130824, "grad_norm": 0.558448851108551, "learning_rate": 3.364958331717665e-06, "loss": 0.2741700649261475, "step": 27680 }, { "epoch": 1.1910398539522695, "grad_norm": 0.690247118473053, "learning_rate": 3.358521191563857e-06, "loss": 0.3063296556472778, "step": 27690 }, { "epoch": 1.1916287560914565, "grad_norm": 0.4874234199523926, "learning_rate": 3.3520889715650116e-06, "loss": 0.29288692474365235, "step": 27700 }, { "epoch": 1.1922176582306436, "grad_norm": 0.5688715577125549, "learning_rate": 3.3456616764862514e-06, "loss": 0.3062859058380127, "step": 27710 }, { "epoch": 1.1928065603698306, "grad_norm": 0.6150898337364197, "learning_rate": 3.339239311089062e-06, "loss": 0.33794426918029785, "step": 27720 }, { "epoch": 1.1933954625090175, "grad_norm": 0.7115926742553711, "learning_rate": 3.3328218801312785e-06, "loss": 0.31481566429138186, "step": 27730 }, { "epoch": 1.1939843646482047, "grad_norm": 0.6787228584289551, "learning_rate": 3.3264093883670768e-06, "loss": 0.3223384618759155, "step": 27740 }, { "epoch": 1.1945732667873916, "grad_norm": 0.4922619163990021, "learning_rate": 3.320001840546966e-06, "loss": 0.3047023296356201, "step": 27750 }, { "epoch": 1.1951621689265786, "grad_norm": 0.7278366684913635, "learning_rate": 3.3135992414178054e-06, "loss": 0.349764084815979, "step": 27760 }, { "epoch": 1.1957510710657657, "grad_norm": 0.7394539713859558, "learning_rate": 3.307201595722783e-06, "loss": 0.3033350229263306, "step": 27770 }, { "epoch": 1.1963399732049527, "grad_norm": 0.5834295153617859, "learning_rate": 3.300808908201413e-06, "loss": 0.29743916988372804, "step": 27780 }, { "epoch": 1.1969288753441396, "grad_norm": 0.7297729849815369, "learning_rate": 3.294421183589538e-06, "loss": 0.3258350372314453, "step": 27790 }, { "epoch": 1.1975177774833268, "grad_norm": 0.5558218955993652, "learning_rate": 3.2880384266193365e-06, "loss": 0.29312591552734374, "step": 27800 }, { "epoch": 1.1981066796225137, "grad_norm": 0.5262384414672852, "learning_rate": 3.2816606420192874e-06, "loss": 0.3142503023147583, "step": 27810 }, { "epoch": 1.1986955817617007, "grad_norm": 0.5400353074073792, "learning_rate": 3.275287834514198e-06, "loss": 0.3062605381011963, "step": 27820 }, { "epoch": 1.1992844839008878, "grad_norm": 0.4843229055404663, "learning_rate": 3.268920008825186e-06, "loss": 0.3101103067398071, "step": 27830 }, { "epoch": 1.1998733860400748, "grad_norm": 0.5123708248138428, "learning_rate": 3.26255716966968e-06, "loss": 0.2904341459274292, "step": 27840 }, { "epoch": 1.2004622881792617, "grad_norm": 0.5715467929840088, "learning_rate": 3.2561993217614054e-06, "loss": 0.29806606769561766, "step": 27850 }, { "epoch": 1.2010511903184489, "grad_norm": 0.5823580026626587, "learning_rate": 3.249846469810405e-06, "loss": 0.32080349922180174, "step": 27860 }, { "epoch": 1.2016400924576358, "grad_norm": 0.6628195643424988, "learning_rate": 3.243498618523011e-06, "loss": 0.3118938684463501, "step": 27870 }, { "epoch": 1.202228994596823, "grad_norm": 0.642971932888031, "learning_rate": 3.2371557726018497e-06, "loss": 0.27577919960021974, "step": 27880 }, { "epoch": 1.20281789673601, "grad_norm": 0.7625564932823181, "learning_rate": 3.2308179367458425e-06, "loss": 0.32851932048797605, "step": 27890 }, { "epoch": 1.2034067988751969, "grad_norm": 0.5563864707946777, "learning_rate": 3.2244851156502012e-06, "loss": 0.3304588317871094, "step": 27900 }, { "epoch": 1.203995701014384, "grad_norm": 0.5686132311820984, "learning_rate": 3.2181573140064217e-06, "loss": 0.3248683214187622, "step": 27910 }, { "epoch": 1.204584603153571, "grad_norm": 0.8546581864356995, "learning_rate": 3.2118345365022696e-06, "loss": 0.31060817241668703, "step": 27920 }, { "epoch": 1.205173505292758, "grad_norm": 0.6397444605827332, "learning_rate": 3.2055167878218118e-06, "loss": 0.3356278419494629, "step": 27930 }, { "epoch": 1.205762407431945, "grad_norm": 0.8423532247543335, "learning_rate": 3.199204072645373e-06, "loss": 0.30249083042144775, "step": 27940 }, { "epoch": 1.206351309571132, "grad_norm": 0.6425405144691467, "learning_rate": 3.1928963956495494e-06, "loss": 0.3169689655303955, "step": 27950 }, { "epoch": 1.206940211710319, "grad_norm": 0.5669530630111694, "learning_rate": 3.1865937615072108e-06, "loss": 0.3553989171981812, "step": 27960 }, { "epoch": 1.2075291138495061, "grad_norm": 0.6348987221717834, "learning_rate": 3.180296174887486e-06, "loss": 0.312061071395874, "step": 27970 }, { "epoch": 1.208118015988693, "grad_norm": 0.5718738436698914, "learning_rate": 3.1740036404557695e-06, "loss": 0.29372096061706543, "step": 27980 }, { "epoch": 1.20870691812788, "grad_norm": 0.4990319013595581, "learning_rate": 3.1677161628737096e-06, "loss": 0.29748947620391847, "step": 27990 }, { "epoch": 1.2092958202670672, "grad_norm": 0.5153680443763733, "learning_rate": 3.1614337467992105e-06, "loss": 0.32971036434173584, "step": 28000 }, { "epoch": 1.000588902139187, "grad_norm": 0.485740602016449, "learning_rate": 3.1551563968864273e-06, "loss": 0.28117187023162843, "step": 28010 }, { "epoch": 1.001177804278374, "grad_norm": 0.5083101987838745, "learning_rate": 3.148884117785753e-06, "loss": 0.2692221403121948, "step": 28020 }, { "epoch": 1.001766706417561, "grad_norm": 0.5193440914154053, "learning_rate": 3.1426169141438367e-06, "loss": 0.2660349369049072, "step": 28030 }, { "epoch": 1.002355608556748, "grad_norm": 0.6036290526390076, "learning_rate": 3.136354790603561e-06, "loss": 0.2760800361633301, "step": 28040 }, { "epoch": 1.0029445106959352, "grad_norm": 0.49894896149635315, "learning_rate": 3.1300977518040455e-06, "loss": 0.2782724142074585, "step": 28050 }, { "epoch": 1.003533412835122, "grad_norm": 0.5713911056518555, "learning_rate": 3.123845802380644e-06, "loss": 0.26750295162200927, "step": 28060 }, { "epoch": 1.004122314974309, "grad_norm": 0.6022392511367798, "learning_rate": 3.1175989469649406e-06, "loss": 0.28304166793823243, "step": 28070 }, { "epoch": 1.0047112171134962, "grad_norm": 0.7185125946998596, "learning_rate": 3.111357190184745e-06, "loss": 0.30764083862304686, "step": 28080 }, { "epoch": 1.0053001192526831, "grad_norm": 0.6078447103500366, "learning_rate": 3.1051205366640845e-06, "loss": 0.2753716468811035, "step": 28090 }, { "epoch": 1.0058890213918703, "grad_norm": 0.5313395857810974, "learning_rate": 3.0988889910232134e-06, "loss": 0.2916360855102539, "step": 28100 }, { "epoch": 1.0064779235310573, "grad_norm": 0.7900572419166565, "learning_rate": 3.0926625578785994e-06, "loss": 0.26853921413421633, "step": 28110 }, { "epoch": 1.0070668256702442, "grad_norm": 0.5913740396499634, "learning_rate": 3.0864412418429225e-06, "loss": 0.26485111713409426, "step": 28120 }, { "epoch": 1.0076557278094314, "grad_norm": 0.5664137601852417, "learning_rate": 3.0802250475250707e-06, "loss": 0.2967384338378906, "step": 28130 }, { "epoch": 1.0082446299486183, "grad_norm": 0.8311530947685242, "learning_rate": 3.0740139795301407e-06, "loss": 0.2688852310180664, "step": 28140 }, { "epoch": 1.0088335320878052, "grad_norm": 0.8490670919418335, "learning_rate": 3.0678080424594305e-06, "loss": 0.298148512840271, "step": 28150 }, { "epoch": 1.0094224342269924, "grad_norm": 0.7639098167419434, "learning_rate": 3.0616072409104304e-06, "loss": 0.3073081970214844, "step": 28160 }, { "epoch": 1.0100113363661793, "grad_norm": 0.7341865301132202, "learning_rate": 3.0554115794768357e-06, "loss": 0.274745512008667, "step": 28170 }, { "epoch": 1.0106002385053663, "grad_norm": 0.8703059554100037, "learning_rate": 3.0492210627485297e-06, "loss": 0.25915825366973877, "step": 28180 }, { "epoch": 1.0111891406445535, "grad_norm": 0.6460238099098206, "learning_rate": 3.043035695311585e-06, "loss": 0.275437331199646, "step": 28190 }, { "epoch": 1.0117780427837404, "grad_norm": 0.628901481628418, "learning_rate": 3.0368554817482578e-06, "loss": 0.29956860542297364, "step": 28200 }, { "epoch": 1.0123669449229273, "grad_norm": 1.1061761379241943, "learning_rate": 3.0306804266369895e-06, "loss": 0.2843048572540283, "step": 28210 }, { "epoch": 1.0129558470621145, "grad_norm": 0.7598164081573486, "learning_rate": 3.0245105345523996e-06, "loss": 0.2958217620849609, "step": 28220 }, { "epoch": 1.0135447492013014, "grad_norm": 0.5487189888954163, "learning_rate": 3.0183458100652752e-06, "loss": 0.2598471164703369, "step": 28230 }, { "epoch": 1.0141336513404884, "grad_norm": 0.7974061369895935, "learning_rate": 3.012186257742584e-06, "loss": 0.27976415157318113, "step": 28240 }, { "epoch": 1.0147225534796755, "grad_norm": 0.6536929607391357, "learning_rate": 3.00603188214746e-06, "loss": 0.2970242738723755, "step": 28250 }, { "epoch": 1.0153114556188625, "grad_norm": 0.7494970560073853, "learning_rate": 2.9998826878392007e-06, "loss": 0.2612969160079956, "step": 28260 }, { "epoch": 1.0159003577580497, "grad_norm": 0.6491073966026306, "learning_rate": 2.993738679373266e-06, "loss": 0.2642483711242676, "step": 28270 }, { "epoch": 1.0164892598972366, "grad_norm": 0.7164976596832275, "learning_rate": 2.9875998613012725e-06, "loss": 0.2488785982131958, "step": 28280 }, { "epoch": 1.0170781620364235, "grad_norm": 0.723341166973114, "learning_rate": 2.981466238170998e-06, "loss": 0.27623016834259034, "step": 28290 }, { "epoch": 1.0176670641756107, "grad_norm": 0.659443736076355, "learning_rate": 2.97533781452636e-06, "loss": 0.251968240737915, "step": 28300 }, { "epoch": 1.0182559663147976, "grad_norm": 0.7834764122962952, "learning_rate": 2.9692145949074324e-06, "loss": 0.262114429473877, "step": 28310 }, { "epoch": 1.0188448684539846, "grad_norm": 0.8033127188682556, "learning_rate": 2.9630965838504343e-06, "loss": 0.24692878723144532, "step": 28320 }, { "epoch": 1.0194337705931718, "grad_norm": 0.6477122902870178, "learning_rate": 2.9569837858877237e-06, "loss": 0.25867395401000975, "step": 28330 }, { "epoch": 1.0200226727323587, "grad_norm": 0.7513903975486755, "learning_rate": 2.950876205547795e-06, "loss": 0.2839254140853882, "step": 28340 }, { "epoch": 1.0206115748715456, "grad_norm": 0.677004337310791, "learning_rate": 2.9447738473552823e-06, "loss": 0.2584822177886963, "step": 28350 }, { "epoch": 1.0212004770107328, "grad_norm": 0.7139666080474854, "learning_rate": 2.938676715830948e-06, "loss": 0.27616894245147705, "step": 28360 }, { "epoch": 1.0217893791499197, "grad_norm": 0.9151942729949951, "learning_rate": 2.932584815491679e-06, "loss": 0.27055964469909666, "step": 28370 }, { "epoch": 1.0223782812891067, "grad_norm": 0.5210399031639099, "learning_rate": 2.926498150850493e-06, "loss": 0.29176104068756104, "step": 28380 }, { "epoch": 1.0229671834282938, "grad_norm": 0.9257033467292786, "learning_rate": 2.920416726416525e-06, "loss": 0.2605153560638428, "step": 28390 }, { "epoch": 1.0235560855674808, "grad_norm": 0.760962188243866, "learning_rate": 2.9143405466950305e-06, "loss": 0.26238629817962644, "step": 28400 }, { "epoch": 1.024144987706668, "grad_norm": 0.8416774272918701, "learning_rate": 2.908269616187378e-06, "loss": 0.26667325496673583, "step": 28410 }, { "epoch": 1.024733889845855, "grad_norm": 0.6791412234306335, "learning_rate": 2.9022039393910462e-06, "loss": 0.2585523843765259, "step": 28420 }, { "epoch": 1.0253227919850418, "grad_norm": 0.8251725435256958, "learning_rate": 2.896143520799628e-06, "loss": 0.26902859210968016, "step": 28430 }, { "epoch": 1.025911694124229, "grad_norm": 0.7182223796844482, "learning_rate": 2.89008836490281e-06, "loss": 0.2902132749557495, "step": 28440 }, { "epoch": 1.026500596263416, "grad_norm": 0.704882800579071, "learning_rate": 2.8840384761863884e-06, "loss": 0.2771092414855957, "step": 28450 }, { "epoch": 1.0270894984026029, "grad_norm": 0.7823062539100647, "learning_rate": 2.8779938591322554e-06, "loss": 0.26958913803100587, "step": 28460 }, { "epoch": 1.02767840054179, "grad_norm": 0.9480551481246948, "learning_rate": 2.871954518218396e-06, "loss": 0.2758470058441162, "step": 28470 }, { "epoch": 1.028267302680977, "grad_norm": 0.7327569127082825, "learning_rate": 2.8659204579188905e-06, "loss": 0.30711913108825684, "step": 28480 }, { "epoch": 1.028856204820164, "grad_norm": 0.6298837661743164, "learning_rate": 2.859891682703907e-06, "loss": 0.26581223011016847, "step": 28490 }, { "epoch": 1.029445106959351, "grad_norm": 0.834546685218811, "learning_rate": 2.8538681970396886e-06, "loss": 0.3081381320953369, "step": 28500 }, { "epoch": 1.030034009098538, "grad_norm": 0.7993612885475159, "learning_rate": 2.847850005388572e-06, "loss": 0.2953542470932007, "step": 28510 }, { "epoch": 1.030622911237725, "grad_norm": 0.7847455739974976, "learning_rate": 2.841837112208966e-06, "loss": 0.2650317192077637, "step": 28520 }, { "epoch": 1.0312118133769121, "grad_norm": 0.8032693266868591, "learning_rate": 2.8358295219553556e-06, "loss": 0.29963929653167726, "step": 28530 }, { "epoch": 1.031800715516099, "grad_norm": 0.6124821901321411, "learning_rate": 2.829827239078298e-06, "loss": 0.27150189876556396, "step": 28540 }, { "epoch": 1.032389617655286, "grad_norm": 0.9862180352210999, "learning_rate": 2.8238302680244166e-06, "loss": 0.29371869564056396, "step": 28550 }, { "epoch": 1.0329785197944732, "grad_norm": 0.8537105917930603, "learning_rate": 2.8178386132364045e-06, "loss": 0.3221353769302368, "step": 28560 }, { "epoch": 1.0335674219336601, "grad_norm": 0.7776139974594116, "learning_rate": 2.8118522791530054e-06, "loss": 0.27172348499298093, "step": 28570 }, { "epoch": 1.0341563240728473, "grad_norm": 0.8639663457870483, "learning_rate": 2.8058712702090328e-06, "loss": 0.2627058744430542, "step": 28580 }, { "epoch": 1.0347452262120342, "grad_norm": 0.5535899996757507, "learning_rate": 2.7998955908353464e-06, "loss": 0.2417090892791748, "step": 28590 }, { "epoch": 1.0353341283512212, "grad_norm": 0.5221068263053894, "learning_rate": 2.793925245458873e-06, "loss": 0.29647390842437743, "step": 28600 }, { "epoch": 1.0359230304904083, "grad_norm": 0.7021687030792236, "learning_rate": 2.787960238502566e-06, "loss": 0.27668776512146, "step": 28610 }, { "epoch": 1.0365119326295953, "grad_norm": 0.7262839078903198, "learning_rate": 2.7820005743854406e-06, "loss": 0.2664266347885132, "step": 28620 }, { "epoch": 1.0371008347687822, "grad_norm": 0.7559595704078674, "learning_rate": 2.7760462575225488e-06, "loss": 0.27735397815704343, "step": 28630 }, { "epoch": 1.0376897369079694, "grad_norm": 0.6918007731437683, "learning_rate": 2.770097292324976e-06, "loss": 0.32111756801605223, "step": 28640 }, { "epoch": 1.0382786390471563, "grad_norm": 0.6178730130195618, "learning_rate": 2.7641536831998506e-06, "loss": 0.28166248798370364, "step": 28650 }, { "epoch": 1.0388675411863433, "grad_norm": 0.6685996055603027, "learning_rate": 2.7582154345503266e-06, "loss": 0.2783522367477417, "step": 28660 }, { "epoch": 1.0394564433255304, "grad_norm": 0.6685901880264282, "learning_rate": 2.7522825507756e-06, "loss": 0.28017702102661135, "step": 28670 }, { "epoch": 1.0400453454647174, "grad_norm": 0.6157472729682922, "learning_rate": 2.7463550362708735e-06, "loss": 0.3031559705734253, "step": 28680 }, { "epoch": 1.0406342476039043, "grad_norm": 0.5730321407318115, "learning_rate": 2.740432895427385e-06, "loss": 0.2743178367614746, "step": 28690 }, { "epoch": 1.0412231497430915, "grad_norm": 0.6820615530014038, "learning_rate": 2.73451613263239e-06, "loss": 0.26905460357666017, "step": 28700 }, { "epoch": 1.0418120518822784, "grad_norm": 0.8740739226341248, "learning_rate": 2.7286047522691527e-06, "loss": 0.27138118743896483, "step": 28710 }, { "epoch": 1.0424009540214654, "grad_norm": 0.7315871715545654, "learning_rate": 2.722698758716954e-06, "loss": 0.3138789892196655, "step": 28720 }, { "epoch": 1.0429898561606525, "grad_norm": 0.878474235534668, "learning_rate": 2.7167981563510927e-06, "loss": 0.3007857322692871, "step": 28730 }, { "epoch": 1.0435787582998395, "grad_norm": 0.8973017930984497, "learning_rate": 2.710902949542864e-06, "loss": 0.27273247241973875, "step": 28740 }, { "epoch": 1.0441676604390266, "grad_norm": 0.7324721217155457, "learning_rate": 2.705013142659563e-06, "loss": 0.25832743644714357, "step": 28750 }, { "epoch": 1.0447565625782136, "grad_norm": 0.6339487433433533, "learning_rate": 2.6991287400644916e-06, "loss": 0.28924145698547366, "step": 28760 }, { "epoch": 1.0453454647174005, "grad_norm": 0.6646597385406494, "learning_rate": 2.6932497461169517e-06, "loss": 0.31656527519226074, "step": 28770 }, { "epoch": 1.0459343668565877, "grad_norm": 0.7524189352989197, "learning_rate": 2.687376165172223e-06, "loss": 0.2708402633666992, "step": 28780 }, { "epoch": 1.0465232689957746, "grad_norm": 0.6263360977172852, "learning_rate": 2.6815080015815853e-06, "loss": 0.3211935758590698, "step": 28790 }, { "epoch": 1.0471121711349616, "grad_norm": 0.6591571569442749, "learning_rate": 2.6756452596923133e-06, "loss": 0.25768651962280276, "step": 28800 }, { "epoch": 1.0477010732741487, "grad_norm": 0.7696152925491333, "learning_rate": 2.669787943847653e-06, "loss": 0.27923829555511476, "step": 28810 }, { "epoch": 1.0482899754133357, "grad_norm": 0.5631375908851624, "learning_rate": 2.663936058386829e-06, "loss": 0.24238290786743164, "step": 28820 }, { "epoch": 1.0488788775525226, "grad_norm": 0.7319104671478271, "learning_rate": 2.6580896076450525e-06, "loss": 0.332311224937439, "step": 28830 }, { "epoch": 1.0494677796917098, "grad_norm": 0.8082656860351562, "learning_rate": 2.652248595953505e-06, "loss": 0.31509554386138916, "step": 28840 }, { "epoch": 1.0500566818308967, "grad_norm": 0.6703341603279114, "learning_rate": 2.646413027639331e-06, "loss": 0.2649134635925293, "step": 28850 }, { "epoch": 1.0506455839700837, "grad_norm": 0.8643590807914734, "learning_rate": 2.640582907025655e-06, "loss": 0.2445751667022705, "step": 28860 }, { "epoch": 1.0512344861092708, "grad_norm": 0.6826237440109253, "learning_rate": 2.6347582384315617e-06, "loss": 0.29661974906921384, "step": 28870 }, { "epoch": 1.0518233882484578, "grad_norm": 1.2298322916030884, "learning_rate": 2.628939026172088e-06, "loss": 0.2827969789505005, "step": 28880 }, { "epoch": 1.052412290387645, "grad_norm": 0.8363764882087708, "learning_rate": 2.6231252745582404e-06, "loss": 0.33168683052062986, "step": 28890 }, { "epoch": 1.0530011925268319, "grad_norm": 0.6877185106277466, "learning_rate": 2.617316987896973e-06, "loss": 0.3111942529678345, "step": 28900 }, { "epoch": 1.0535900946660188, "grad_norm": 0.9166999459266663, "learning_rate": 2.6115141704911985e-06, "loss": 0.25333831310272215, "step": 28910 }, { "epoch": 1.054178996805206, "grad_norm": 0.7758864164352417, "learning_rate": 2.6057168266397616e-06, "loss": 0.3013713836669922, "step": 28920 }, { "epoch": 1.054767898944393, "grad_norm": 0.9442702531814575, "learning_rate": 2.5999249606374742e-06, "loss": 0.31105518341064453, "step": 28930 }, { "epoch": 1.0553568010835799, "grad_norm": 0.8953559398651123, "learning_rate": 2.5941385767750784e-06, "loss": 0.2932743072509766, "step": 28940 }, { "epoch": 1.055945703222767, "grad_norm": 0.8710790872573853, "learning_rate": 2.5883576793392494e-06, "loss": 0.2959449768066406, "step": 28950 }, { "epoch": 1.056534605361954, "grad_norm": 0.5169467926025391, "learning_rate": 2.5825822726126082e-06, "loss": 0.2617682218551636, "step": 28960 }, { "epoch": 1.057123507501141, "grad_norm": 0.9348832368850708, "learning_rate": 2.576812360873705e-06, "loss": 0.31488785743713377, "step": 28970 }, { "epoch": 1.057712409640328, "grad_norm": 0.6582676768302917, "learning_rate": 2.5710479483970174e-06, "loss": 0.266589879989624, "step": 28980 }, { "epoch": 1.058301311779515, "grad_norm": 0.6413953900337219, "learning_rate": 2.5652890394529496e-06, "loss": 0.32013750076293945, "step": 28990 }, { "epoch": 1.058890213918702, "grad_norm": 0.6583389043807983, "learning_rate": 2.5595356383078305e-06, "loss": 0.27780919075012206, "step": 29000 }, { "epoch": 1.0594791160578891, "grad_norm": 0.8131144642829895, "learning_rate": 2.5537877492239093e-06, "loss": 0.283194899559021, "step": 29010 }, { "epoch": 1.060068018197076, "grad_norm": 0.910049319267273, "learning_rate": 2.548045376459345e-06, "loss": 0.3104546070098877, "step": 29020 }, { "epoch": 1.060656920336263, "grad_norm": 0.9563156962394714, "learning_rate": 2.542308524268218e-06, "loss": 0.2997922897338867, "step": 29030 }, { "epoch": 1.0612458224754502, "grad_norm": 0.6670770645141602, "learning_rate": 2.5365771969005136e-06, "loss": 0.2805549144744873, "step": 29040 }, { "epoch": 1.0618347246146371, "grad_norm": 0.8473305702209473, "learning_rate": 2.530851398602129e-06, "loss": 0.29952800273895264, "step": 29050 }, { "epoch": 1.0624236267538243, "grad_norm": 0.732576310634613, "learning_rate": 2.525131133614861e-06, "loss": 0.29922807216644287, "step": 29060 }, { "epoch": 1.0630125288930112, "grad_norm": 0.6896460652351379, "learning_rate": 2.5194164061764094e-06, "loss": 0.2771118640899658, "step": 29070 }, { "epoch": 1.0636014310321982, "grad_norm": 0.6631947159767151, "learning_rate": 2.513707220520374e-06, "loss": 0.28859360218048097, "step": 29080 }, { "epoch": 1.0641903331713853, "grad_norm": 0.7276545166969299, "learning_rate": 2.5080035808762406e-06, "loss": 0.2866827487945557, "step": 29090 }, { "epoch": 1.0647792353105723, "grad_norm": 0.5554602146148682, "learning_rate": 2.5023054914693946e-06, "loss": 0.2854539632797241, "step": 29100 }, { "epoch": 1.0653681374497592, "grad_norm": 0.9282269477844238, "learning_rate": 2.496612956521108e-06, "loss": 0.3130819320678711, "step": 29110 }, { "epoch": 1.0659570395889464, "grad_norm": 0.5553298592567444, "learning_rate": 2.4909259802485364e-06, "loss": 0.2917849779129028, "step": 29120 }, { "epoch": 1.0665459417281333, "grad_norm": 0.5885695219039917, "learning_rate": 2.4852445668647186e-06, "loss": 0.2790310621261597, "step": 29130 }, { "epoch": 1.0671348438673203, "grad_norm": 0.6164746880531311, "learning_rate": 2.4795687205785713e-06, "loss": 0.32441585063934325, "step": 29140 }, { "epoch": 1.0677237460065074, "grad_norm": 0.6399831175804138, "learning_rate": 2.47389844559489e-06, "loss": 0.2941070318222046, "step": 29150 }, { "epoch": 1.0683126481456944, "grad_norm": 0.9646731019020081, "learning_rate": 2.4682337461143356e-06, "loss": 0.3082376718521118, "step": 29160 }, { "epoch": 1.0689015502848813, "grad_norm": 0.787506103515625, "learning_rate": 2.462574626333446e-06, "loss": 0.299594521522522, "step": 29170 }, { "epoch": 1.0694904524240685, "grad_norm": 0.7882640957832336, "learning_rate": 2.4569210904446218e-06, "loss": 0.3026951551437378, "step": 29180 }, { "epoch": 1.0700793545632554, "grad_norm": 0.5071229338645935, "learning_rate": 2.451273142636128e-06, "loss": 0.2813900947570801, "step": 29190 }, { "epoch": 1.0706682567024424, "grad_norm": 0.5648660659790039, "learning_rate": 2.44563078709209e-06, "loss": 0.28492305278778074, "step": 29200 }, { "epoch": 1.0712571588416295, "grad_norm": 0.6572549343109131, "learning_rate": 2.4399940279924905e-06, "loss": 0.31006789207458496, "step": 29210 }, { "epoch": 1.0004711217113496, "grad_norm": 0.6735312938690186, "learning_rate": 2.4343628695131692e-06, "loss": 0.27653592824935913, "step": 29220 }, { "epoch": 1.0010600238505367, "grad_norm": 0.645003080368042, "learning_rate": 2.4287373158258055e-06, "loss": 0.26594996452331543, "step": 29230 }, { "epoch": 1.0016489259897237, "grad_norm": 0.4896929860115051, "learning_rate": 2.4231173710979396e-06, "loss": 0.25569305419921873, "step": 29240 }, { "epoch": 1.0022378281289106, "grad_norm": 0.7442900538444519, "learning_rate": 2.41750303949295e-06, "loss": 0.2633437395095825, "step": 29250 }, { "epoch": 1.0028267302680978, "grad_norm": 0.6730748414993286, "learning_rate": 2.411894325170059e-06, "loss": 0.2705569267272949, "step": 29260 }, { "epoch": 1.0034156324072847, "grad_norm": 0.6061217784881592, "learning_rate": 2.4062912322843258e-06, "loss": 0.25478348731994627, "step": 29270 }, { "epoch": 1.0040045345464716, "grad_norm": 0.6705806851387024, "learning_rate": 2.400693764986647e-06, "loss": 0.2800071954727173, "step": 29280 }, { "epoch": 1.0045934366856588, "grad_norm": 0.6596165299415588, "learning_rate": 2.395101927423752e-06, "loss": 0.26584227085113527, "step": 29290 }, { "epoch": 1.0051823388248458, "grad_norm": 0.9522062540054321, "learning_rate": 2.389515723738194e-06, "loss": 0.3052809476852417, "step": 29300 }, { "epoch": 1.0057712409640327, "grad_norm": 0.8020858764648438, "learning_rate": 2.383935158068359e-06, "loss": 0.2914546251296997, "step": 29310 }, { "epoch": 1.0063601431032199, "grad_norm": 0.8702990412712097, "learning_rate": 2.3783602345484538e-06, "loss": 0.255890417098999, "step": 29320 }, { "epoch": 1.0069490452424068, "grad_norm": 0.7657840847969055, "learning_rate": 2.372790957308506e-06, "loss": 0.2531628370285034, "step": 29330 }, { "epoch": 1.007537947381594, "grad_norm": 0.7143756747245789, "learning_rate": 2.36722733047436e-06, "loss": 0.29654741287231445, "step": 29340 }, { "epoch": 1.008126849520781, "grad_norm": 0.6144415736198425, "learning_rate": 2.361669358167673e-06, "loss": 0.244527006149292, "step": 29350 }, { "epoch": 1.0087157516599679, "grad_norm": 0.7765334844589233, "learning_rate": 2.3561170445059177e-06, "loss": 0.2790250778198242, "step": 29360 }, { "epoch": 1.009304653799155, "grad_norm": 0.7923306226730347, "learning_rate": 2.350570393602366e-06, "loss": 0.3214289426803589, "step": 29370 }, { "epoch": 1.009893555938342, "grad_norm": 0.790327250957489, "learning_rate": 2.345029409566102e-06, "loss": 0.27222745418548583, "step": 29380 }, { "epoch": 1.010482458077529, "grad_norm": 0.672583281993866, "learning_rate": 2.3394940965020106e-06, "loss": 0.2433936595916748, "step": 29390 }, { "epoch": 1.011071360216716, "grad_norm": 0.8394589424133301, "learning_rate": 2.3339644585107745e-06, "loss": 0.2755703926086426, "step": 29400 }, { "epoch": 1.011660262355903, "grad_norm": 0.5919690728187561, "learning_rate": 2.328440499688872e-06, "loss": 0.2799059391021729, "step": 29410 }, { "epoch": 1.01224916449509, "grad_norm": 0.7916842699050903, "learning_rate": 2.3229222241285753e-06, "loss": 0.2750527381896973, "step": 29420 }, { "epoch": 1.0128380666342771, "grad_norm": 0.858955442905426, "learning_rate": 2.3174096359179465e-06, "loss": 0.29750628471374513, "step": 29430 }, { "epoch": 1.013426968773464, "grad_norm": 0.6444618701934814, "learning_rate": 2.3119027391408287e-06, "loss": 0.2295013189315796, "step": 29440 }, { "epoch": 1.014015870912651, "grad_norm": 1.0950666666030884, "learning_rate": 2.3064015378768566e-06, "loss": 0.29247820377349854, "step": 29450 }, { "epoch": 1.0146047730518382, "grad_norm": 0.8245145678520203, "learning_rate": 2.3009060362014424e-06, "loss": 0.2996661901473999, "step": 29460 }, { "epoch": 1.015193675191025, "grad_norm": 0.7296099066734314, "learning_rate": 2.2954162381857737e-06, "loss": 0.2350468158721924, "step": 29470 }, { "epoch": 1.015782577330212, "grad_norm": 0.7119907736778259, "learning_rate": 2.2899321478968174e-06, "loss": 0.27033681869506837, "step": 29480 }, { "epoch": 1.0163714794693992, "grad_norm": 1.0497878789901733, "learning_rate": 2.284453769397308e-06, "loss": 0.23436946868896485, "step": 29490 }, { "epoch": 1.0169603816085862, "grad_norm": 0.683103621006012, "learning_rate": 2.2789811067457523e-06, "loss": 0.27512598037719727, "step": 29500 }, { "epoch": 1.0175492837477733, "grad_norm": 0.6718295216560364, "learning_rate": 2.2735141639964154e-06, "loss": 0.24754588603973388, "step": 29510 } ], "logging_steps": 10, "max_steps": 37500, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 3.0946093434232898e+19, "train_batch_size": 4, "trial_name": null, "trial_params": null }