{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9998873493297286,
  "eval_steps": 500,
  "global_step": 2219,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004506026810859525,
      "grad_norm": 4.815864771779815,
      "learning_rate": 4.504504504504505e-08,
      "loss": 0.366,
      "step": 1
    },
    {
      "epoch": 0.000901205362171905,
      "grad_norm": 5.195074053996181,
      "learning_rate": 9.00900900900901e-08,
      "loss": 0.3833,
      "step": 2
    },
    {
      "epoch": 0.0013518080432578573,
      "grad_norm": 5.109775812530038,
      "learning_rate": 1.3513513513513515e-07,
      "loss": 0.3872,
      "step": 3
    },
    {
      "epoch": 0.00180241072434381,
      "grad_norm": 4.912948106635599,
      "learning_rate": 1.801801801801802e-07,
      "loss": 0.374,
      "step": 4
    },
    {
      "epoch": 0.002253013405429762,
      "grad_norm": 4.96137274114766,
      "learning_rate": 2.2522522522522524e-07,
      "loss": 0.3723,
      "step": 5
    },
    {
      "epoch": 0.0027036160865157146,
      "grad_norm": 4.898137124077649,
      "learning_rate": 2.702702702702703e-07,
      "loss": 0.3896,
      "step": 6
    },
    {
      "epoch": 0.0031542187676016674,
      "grad_norm": 4.794314493950962,
      "learning_rate": 3.153153153153153e-07,
      "loss": 0.4154,
      "step": 7
    },
    {
      "epoch": 0.00360482144868762,
      "grad_norm": 5.105069245772989,
      "learning_rate": 3.603603603603604e-07,
      "loss": 0.3943,
      "step": 8
    },
    {
      "epoch": 0.004055424129773572,
      "grad_norm": 4.437918536689456,
      "learning_rate": 4.0540540540540546e-07,
      "loss": 0.3531,
      "step": 9
    },
    {
      "epoch": 0.004506026810859524,
      "grad_norm": 4.369828896152025,
      "learning_rate": 4.504504504504505e-07,
      "loss": 0.3822,
      "step": 10
    },
    {
      "epoch": 0.004956629491945477,
      "grad_norm": 4.647490729965825,
      "learning_rate": 4.954954954954956e-07,
      "loss": 0.3663,
      "step": 11
    },
    {
      "epoch": 0.005407232173031429,
      "grad_norm": 4.816738574275561,
      "learning_rate": 5.405405405405406e-07,
      "loss": 0.3648,
      "step": 12
    },
    {
      "epoch": 0.005857834854117382,
      "grad_norm": 3.969135638580672,
      "learning_rate": 5.855855855855856e-07,
      "loss": 0.3598,
      "step": 13
    },
    {
      "epoch": 0.006308437535203335,
      "grad_norm": 3.6447362042128044,
      "learning_rate": 6.306306306306306e-07,
      "loss": 0.3544,
      "step": 14
    },
    {
      "epoch": 0.006759040216289287,
      "grad_norm": 2.8706345170518266,
      "learning_rate": 6.756756756756758e-07,
      "loss": 0.3397,
      "step": 15
    },
    {
      "epoch": 0.00720964289737524,
      "grad_norm": 2.778442530150813,
      "learning_rate": 7.207207207207208e-07,
      "loss": 0.3503,
      "step": 16
    },
    {
      "epoch": 0.007660245578461192,
      "grad_norm": 3.0191413167735655,
      "learning_rate": 7.657657657657658e-07,
      "loss": 0.3786,
      "step": 17
    },
    {
      "epoch": 0.008110848259547145,
      "grad_norm": 2.6863762331965004,
      "learning_rate": 8.108108108108109e-07,
      "loss": 0.3487,
      "step": 18
    },
    {
      "epoch": 0.008561450940633097,
      "grad_norm": 2.500154127846048,
      "learning_rate": 8.55855855855856e-07,
      "loss": 0.3555,
      "step": 19
    },
    {
      "epoch": 0.009012053621719049,
      "grad_norm": 2.3648275264818084,
      "learning_rate": 9.00900900900901e-07,
      "loss": 0.3476,
      "step": 20
    },
    {
      "epoch": 0.009462656302805001,
      "grad_norm": 2.6069567739452872,
      "learning_rate": 9.459459459459461e-07,
      "loss": 0.3431,
      "step": 21
    },
    {
      "epoch": 0.009913258983890954,
      "grad_norm": 2.6546117604315937,
      "learning_rate": 9.909909909909911e-07,
      "loss": 0.3011,
      "step": 22
    },
    {
      "epoch": 0.010363861664976907,
      "grad_norm": 2.429078931197282,
      "learning_rate": 1.0360360360360361e-06,
      "loss": 0.3117,
      "step": 23
    },
    {
      "epoch": 0.010814464346062858,
      "grad_norm": 1.9589677923898288,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 0.327,
      "step": 24
    },
    {
      "epoch": 0.011265067027148811,
      "grad_norm": 1.8522547214325198,
      "learning_rate": 1.1261261261261262e-06,
      "loss": 0.3097,
      "step": 25
    },
    {
      "epoch": 0.011715669708234764,
      "grad_norm": 1.5812319133542672,
      "learning_rate": 1.1711711711711712e-06,
      "loss": 0.2835,
      "step": 26
    },
    {
      "epoch": 0.012166272389320717,
      "grad_norm": 1.478551103830084,
      "learning_rate": 1.2162162162162164e-06,
      "loss": 0.3027,
      "step": 27
    },
    {
      "epoch": 0.01261687507040667,
      "grad_norm": 1.6047585590317004,
      "learning_rate": 1.2612612612612613e-06,
      "loss": 0.2851,
      "step": 28
    },
    {
      "epoch": 0.01306747775149262,
      "grad_norm": 1.8154984583588165,
      "learning_rate": 1.3063063063063065e-06,
      "loss": 0.2757,
      "step": 29
    },
    {
      "epoch": 0.013518080432578574,
      "grad_norm": 2.027167277213967,
      "learning_rate": 1.3513513513513515e-06,
      "loss": 0.3006,
      "step": 30
    },
    {
      "epoch": 0.013968683113664527,
      "grad_norm": 1.840440668341431,
      "learning_rate": 1.3963963963963963e-06,
      "loss": 0.2894,
      "step": 31
    },
    {
      "epoch": 0.01441928579475048,
      "grad_norm": 1.8547802317028712,
      "learning_rate": 1.4414414414414416e-06,
      "loss": 0.3104,
      "step": 32
    },
    {
      "epoch": 0.01486988847583643,
      "grad_norm": 1.6017226824079185,
      "learning_rate": 1.4864864864864868e-06,
      "loss": 0.2825,
      "step": 33
    },
    {
      "epoch": 0.015320491156922383,
      "grad_norm": 1.618932965986627,
      "learning_rate": 1.5315315315315316e-06,
      "loss": 0.2897,
      "step": 34
    },
    {
      "epoch": 0.015771093838008336,
      "grad_norm": 1.725889184071538,
      "learning_rate": 1.5765765765765766e-06,
      "loss": 0.2492,
      "step": 35
    },
    {
      "epoch": 0.01622169651909429,
      "grad_norm": 1.7545616100435062,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 0.2885,
      "step": 36
    },
    {
      "epoch": 0.016672299200180242,
      "grad_norm": 1.443938382974907,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.2847,
      "step": 37
    },
    {
      "epoch": 0.017122901881266195,
      "grad_norm": 1.3439469209062533,
      "learning_rate": 1.711711711711712e-06,
      "loss": 0.273,
      "step": 38
    },
    {
      "epoch": 0.017573504562352148,
      "grad_norm": 1.4919993368852724,
      "learning_rate": 1.756756756756757e-06,
      "loss": 0.2724,
      "step": 39
    },
    {
      "epoch": 0.018024107243438097,
      "grad_norm": 1.646749223197058,
      "learning_rate": 1.801801801801802e-06,
      "loss": 0.2597,
      "step": 40
    },
    {
      "epoch": 0.01847470992452405,
      "grad_norm": 1.280899828503881,
      "learning_rate": 1.846846846846847e-06,
      "loss": 0.2483,
      "step": 41
    },
    {
      "epoch": 0.018925312605610003,
      "grad_norm": 1.297992220461041,
      "learning_rate": 1.8918918918918922e-06,
      "loss": 0.2659,
      "step": 42
    },
    {
      "epoch": 0.019375915286695956,
      "grad_norm": 1.45364906642596,
      "learning_rate": 1.9369369369369372e-06,
      "loss": 0.2777,
      "step": 43
    },
    {
      "epoch": 0.01982651796778191,
      "grad_norm": 1.7350773523889338,
      "learning_rate": 1.9819819819819822e-06,
      "loss": 0.2699,
      "step": 44
    },
    {
      "epoch": 0.02027712064886786,
      "grad_norm": 1.3800904283438646,
      "learning_rate": 2.0270270270270273e-06,
      "loss": 0.2543,
      "step": 45
    },
    {
      "epoch": 0.020727723329953814,
      "grad_norm": 1.4181131950530033,
      "learning_rate": 2.0720720720720723e-06,
      "loss": 0.2576,
      "step": 46
    },
    {
      "epoch": 0.021178326011039767,
      "grad_norm": 1.535738819405899,
      "learning_rate": 2.1171171171171173e-06,
      "loss": 0.2432,
      "step": 47
    },
    {
      "epoch": 0.021628928692125717,
      "grad_norm": 1.2566148211560952,
      "learning_rate": 2.1621621621621623e-06,
      "loss": 0.2581,
      "step": 48
    },
    {
      "epoch": 0.02207953137321167,
      "grad_norm": 1.3215876440850718,
      "learning_rate": 2.2072072072072073e-06,
      "loss": 0.2544,
      "step": 49
    },
    {
      "epoch": 0.022530134054297622,
      "grad_norm": 1.3222957733533551,
      "learning_rate": 2.2522522522522524e-06,
      "loss": 0.2383,
      "step": 50
    },
    {
      "epoch": 0.022980736735383575,
      "grad_norm": 1.3558975651467386,
      "learning_rate": 2.297297297297298e-06,
      "loss": 0.2623,
      "step": 51
    },
    {
      "epoch": 0.023431339416469528,
      "grad_norm": 1.180665933720355,
      "learning_rate": 2.3423423423423424e-06,
      "loss": 0.2358,
      "step": 52
    },
    {
      "epoch": 0.02388194209755548,
      "grad_norm": 1.3377467432333017,
      "learning_rate": 2.3873873873873874e-06,
      "loss": 0.2674,
      "step": 53
    },
    {
      "epoch": 0.024332544778641434,
      "grad_norm": 1.429078367789538,
      "learning_rate": 2.432432432432433e-06,
      "loss": 0.2539,
      "step": 54
    },
    {
      "epoch": 0.024783147459727387,
      "grad_norm": 1.29497107657176,
      "learning_rate": 2.4774774774774775e-06,
      "loss": 0.2431,
      "step": 55
    },
    {
      "epoch": 0.02523375014081334,
      "grad_norm": 1.4338057761852927,
      "learning_rate": 2.5225225225225225e-06,
      "loss": 0.2477,
      "step": 56
    },
    {
      "epoch": 0.02568435282189929,
      "grad_norm": 1.4626114161947357,
      "learning_rate": 2.5675675675675675e-06,
      "loss": 0.2573,
      "step": 57
    },
    {
      "epoch": 0.02613495550298524,
      "grad_norm": 1.3908318133820947,
      "learning_rate": 2.612612612612613e-06,
      "loss": 0.2592,
      "step": 58
    },
    {
      "epoch": 0.026585558184071195,
      "grad_norm": 1.2880400988089797,
      "learning_rate": 2.657657657657658e-06,
      "loss": 0.2446,
      "step": 59
    },
    {
      "epoch": 0.027036160865157147,
      "grad_norm": 1.3643355524484657,
      "learning_rate": 2.702702702702703e-06,
      "loss": 0.2463,
      "step": 60
    },
    {
      "epoch": 0.0274867635462431,
      "grad_norm": 1.3930601643893419,
      "learning_rate": 2.747747747747748e-06,
      "loss": 0.247,
      "step": 61
    },
    {
      "epoch": 0.027937366227329053,
      "grad_norm": 1.3674157761140657,
      "learning_rate": 2.7927927927927926e-06,
      "loss": 0.239,
      "step": 62
    },
    {
      "epoch": 0.028387968908415006,
      "grad_norm": 1.2947244011903152,
      "learning_rate": 2.837837837837838e-06,
      "loss": 0.245,
      "step": 63
    },
    {
      "epoch": 0.02883857158950096,
      "grad_norm": 1.3182407851781994,
      "learning_rate": 2.882882882882883e-06,
      "loss": 0.2519,
      "step": 64
    },
    {
      "epoch": 0.02928917427058691,
      "grad_norm": 1.2817823833464046,
      "learning_rate": 2.927927927927928e-06,
      "loss": 0.254,
      "step": 65
    },
    {
      "epoch": 0.02973977695167286,
      "grad_norm": 1.3451925823541346,
      "learning_rate": 2.9729729729729736e-06,
      "loss": 0.2402,
      "step": 66
    },
    {
      "epoch": 0.030190379632758814,
      "grad_norm": 1.2943481824061323,
      "learning_rate": 3.0180180180180186e-06,
      "loss": 0.2359,
      "step": 67
    },
    {
      "epoch": 0.030640982313844767,
      "grad_norm": 1.3423425334342036,
      "learning_rate": 3.063063063063063e-06,
      "loss": 0.248,
      "step": 68
    },
    {
      "epoch": 0.03109158499493072,
      "grad_norm": 1.2049652231573809,
      "learning_rate": 3.1081081081081082e-06,
      "loss": 0.2301,
      "step": 69
    },
    {
      "epoch": 0.03154218767601667,
      "grad_norm": 1.3344500076487555,
      "learning_rate": 3.1531531531531532e-06,
      "loss": 0.2342,
      "step": 70
    },
    {
      "epoch": 0.031992790357102625,
      "grad_norm": 1.1874737435413407,
      "learning_rate": 3.1981981981981987e-06,
      "loss": 0.2209,
      "step": 71
    },
    {
      "epoch": 0.03244339303818858,
      "grad_norm": 1.3779776082559232,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 0.2361,
      "step": 72
    },
    {
      "epoch": 0.03289399571927453,
      "grad_norm": 1.2969763953943954,
      "learning_rate": 3.2882882882882887e-06,
      "loss": 0.253,
      "step": 73
    },
    {
      "epoch": 0.033344598400360484,
      "grad_norm": 1.3469669191031444,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.2411,
      "step": 74
    },
    {
      "epoch": 0.03379520108144644,
      "grad_norm": 1.3704530488229245,
      "learning_rate": 3.3783783783783788e-06,
      "loss": 0.234,
      "step": 75
    },
    {
      "epoch": 0.03424580376253239,
      "grad_norm": 1.223619143070724,
      "learning_rate": 3.423423423423424e-06,
      "loss": 0.2437,
      "step": 76
    },
    {
      "epoch": 0.03469640644361834,
      "grad_norm": 1.399539495833997,
      "learning_rate": 3.468468468468469e-06,
      "loss": 0.2564,
      "step": 77
    },
    {
      "epoch": 0.035147009124704295,
      "grad_norm": 1.2277545572758175,
      "learning_rate": 3.513513513513514e-06,
      "loss": 0.2293,
      "step": 78
    },
    {
      "epoch": 0.03559761180579024,
      "grad_norm": 1.3602232090766564,
      "learning_rate": 3.5585585585585584e-06,
      "loss": 0.2521,
      "step": 79
    },
    {
      "epoch": 0.036048214486876194,
      "grad_norm": 1.355790485541872,
      "learning_rate": 3.603603603603604e-06,
      "loss": 0.2429,
      "step": 80
    },
    {
      "epoch": 0.03649881716796215,
      "grad_norm": 1.1987258690046945,
      "learning_rate": 3.648648648648649e-06,
      "loss": 0.245,
      "step": 81
    },
    {
      "epoch": 0.0369494198490481,
      "grad_norm": 1.6169310703051463,
      "learning_rate": 3.693693693693694e-06,
      "loss": 0.2494,
      "step": 82
    },
    {
      "epoch": 0.03740002253013405,
      "grad_norm": 1.5187379828279406,
      "learning_rate": 3.7387387387387394e-06,
      "loss": 0.2346,
      "step": 83
    },
    {
      "epoch": 0.037850625211220006,
      "grad_norm": 1.3308952951835935,
      "learning_rate": 3.7837837837837844e-06,
      "loss": 0.2396,
      "step": 84
    },
    {
      "epoch": 0.03830122789230596,
      "grad_norm": 1.5888441541358347,
      "learning_rate": 3.828828828828829e-06,
      "loss": 0.249,
      "step": 85
    },
    {
      "epoch": 0.03875183057339191,
      "grad_norm": 1.2796516821327377,
      "learning_rate": 3.8738738738738744e-06,
      "loss": 0.2381,
      "step": 86
    },
    {
      "epoch": 0.039202433254477864,
      "grad_norm": 1.291791666163532,
      "learning_rate": 3.918918918918919e-06,
      "loss": 0.2422,
      "step": 87
    },
    {
      "epoch": 0.03965303593556382,
      "grad_norm": 1.381850806769349,
      "learning_rate": 3.9639639639639645e-06,
      "loss": 0.2558,
      "step": 88
    },
    {
      "epoch": 0.04010363861664977,
      "grad_norm": 1.3086943706864933,
      "learning_rate": 4.009009009009009e-06,
      "loss": 0.2497,
      "step": 89
    },
    {
      "epoch": 0.04055424129773572,
      "grad_norm": 1.2999519245744664,
      "learning_rate": 4.0540540540540545e-06,
      "loss": 0.2241,
      "step": 90
    },
    {
      "epoch": 0.041004843978821676,
      "grad_norm": 1.183224660888697,
      "learning_rate": 4.099099099099099e-06,
      "loss": 0.2327,
      "step": 91
    },
    {
      "epoch": 0.04145544665990763,
      "grad_norm": 1.2998101785174705,
      "learning_rate": 4.1441441441441446e-06,
      "loss": 0.2282,
      "step": 92
    },
    {
      "epoch": 0.04190604934099358,
      "grad_norm": 1.2410993409820663,
      "learning_rate": 4.189189189189189e-06,
      "loss": 0.2478,
      "step": 93
    },
    {
      "epoch": 0.042356652022079534,
      "grad_norm": 1.2645254985527676,
      "learning_rate": 4.234234234234235e-06,
      "loss": 0.2479,
      "step": 94
    },
    {
      "epoch": 0.04280725470316549,
      "grad_norm": 1.2459528417226233,
      "learning_rate": 4.27927927927928e-06,
      "loss": 0.2314,
      "step": 95
    },
    {
      "epoch": 0.04325785738425143,
      "grad_norm": 1.249112068397591,
      "learning_rate": 4.324324324324325e-06,
      "loss": 0.2255,
      "step": 96
    },
    {
      "epoch": 0.043708460065337386,
      "grad_norm": 1.2728802945422262,
      "learning_rate": 4.369369369369369e-06,
      "loss": 0.2387,
      "step": 97
    },
    {
      "epoch": 0.04415906274642334,
      "grad_norm": 1.4177091138057598,
      "learning_rate": 4.414414414414415e-06,
      "loss": 0.2218,
      "step": 98
    },
    {
      "epoch": 0.04460966542750929,
      "grad_norm": 1.296078758285749,
      "learning_rate": 4.45945945945946e-06,
      "loss": 0.2293,
      "step": 99
    },
    {
      "epoch": 0.045060268108595244,
      "grad_norm": 1.46261721387381,
      "learning_rate": 4.504504504504505e-06,
      "loss": 0.2279,
      "step": 100
    },
    {
      "epoch": 0.0455108707896812,
      "grad_norm": 1.3107218066814923,
      "learning_rate": 4.54954954954955e-06,
      "loss": 0.2289,
      "step": 101
    },
    {
      "epoch": 0.04596147347076715,
      "grad_norm": 1.275172277252069,
      "learning_rate": 4.594594594594596e-06,
      "loss": 0.2418,
      "step": 102
    },
    {
      "epoch": 0.0464120761518531,
      "grad_norm": 1.464017582981461,
      "learning_rate": 4.63963963963964e-06,
      "loss": 0.2424,
      "step": 103
    },
    {
      "epoch": 0.046862678832939056,
      "grad_norm": 1.3024312968634821,
      "learning_rate": 4.684684684684685e-06,
      "loss": 0.2242,
      "step": 104
    },
    {
      "epoch": 0.04731328151402501,
      "grad_norm": 1.363653722681725,
      "learning_rate": 4.72972972972973e-06,
      "loss": 0.2418,
      "step": 105
    },
    {
      "epoch": 0.04776388419511096,
      "grad_norm": 1.1701030694099703,
      "learning_rate": 4.774774774774775e-06,
      "loss": 0.2311,
      "step": 106
    },
    {
      "epoch": 0.048214486876196914,
      "grad_norm": 1.3797624934807715,
      "learning_rate": 4.81981981981982e-06,
      "loss": 0.2374,
      "step": 107
    },
    {
      "epoch": 0.04866508955728287,
      "grad_norm": 1.2136932937379246,
      "learning_rate": 4.864864864864866e-06,
      "loss": 0.2239,
      "step": 108
    },
    {
      "epoch": 0.04911569223836882,
      "grad_norm": 1.2629715016785896,
      "learning_rate": 4.90990990990991e-06,
      "loss": 0.2251,
      "step": 109
    },
    {
      "epoch": 0.04956629491945477,
      "grad_norm": 1.4317438257489863,
      "learning_rate": 4.954954954954955e-06,
      "loss": 0.2366,
      "step": 110
    },
    {
      "epoch": 0.050016897600540726,
      "grad_norm": 1.3241056157209627,
      "learning_rate": 5e-06,
      "loss": 0.233,
      "step": 111
    },
    {
      "epoch": 0.05046750028162668,
      "grad_norm": 1.290038853019236,
      "learning_rate": 4.999997223686756e-06,
      "loss": 0.2421,
      "step": 112
    },
    {
      "epoch": 0.050918102962712625,
      "grad_norm": 1.365409438767111,
      "learning_rate": 4.999988894753189e-06,
      "loss": 0.2393,
      "step": 113
    },
    {
      "epoch": 0.05136870564379858,
      "grad_norm": 1.2940027464844308,
      "learning_rate": 4.999975013217796e-06,
      "loss": 0.2351,
      "step": 114
    },
    {
      "epoch": 0.05181930832488453,
      "grad_norm": 1.5242296190272844,
      "learning_rate": 4.999955579111413e-06,
      "loss": 0.2224,
      "step": 115
    },
    {
      "epoch": 0.05226991100597048,
      "grad_norm": 1.2441974279995303,
      "learning_rate": 4.9999305924772e-06,
      "loss": 0.2346,
      "step": 116
    },
    {
      "epoch": 0.052720513687056436,
      "grad_norm": 1.2318049884586677,
      "learning_rate": 4.999900053370657e-06,
      "loss": 0.2197,
      "step": 117
    },
    {
      "epoch": 0.05317111636814239,
      "grad_norm": 1.378129834491324,
      "learning_rate": 4.99986396185961e-06,
      "loss": 0.2396,
      "step": 118
    },
    {
      "epoch": 0.05362171904922834,
      "grad_norm": 1.2120301657823338,
      "learning_rate": 4.999822318024222e-06,
      "loss": 0.2265,
      "step": 119
    },
    {
      "epoch": 0.054072321730314295,
      "grad_norm": 1.332984287268552,
      "learning_rate": 4.9997751219569844e-06,
      "loss": 0.2448,
      "step": 120
    },
    {
      "epoch": 0.05452292441140025,
      "grad_norm": 1.2389274937015193,
      "learning_rate": 4.999722373762725e-06,
      "loss": 0.224,
      "step": 121
    },
    {
      "epoch": 0.0549735270924862,
      "grad_norm": 1.204117312589835,
      "learning_rate": 4.999664073558596e-06,
      "loss": 0.2204,
      "step": 122
    },
    {
      "epoch": 0.05542412977357215,
      "grad_norm": 1.3101533213641092,
      "learning_rate": 4.999600221474089e-06,
      "loss": 0.2188,
      "step": 123
    },
    {
      "epoch": 0.055874732454658106,
      "grad_norm": 1.355729531514012,
      "learning_rate": 4.99953081765102e-06,
      "loss": 0.219,
      "step": 124
    },
    {
      "epoch": 0.05632533513574406,
      "grad_norm": 1.3145496397104341,
      "learning_rate": 4.999455862243539e-06,
      "loss": 0.2416,
      "step": 125
    },
    {
      "epoch": 0.05677593781683001,
      "grad_norm": 1.2549927498664672,
      "learning_rate": 4.999375355418128e-06,
      "loss": 0.2419,
      "step": 126
    },
    {
      "epoch": 0.057226540497915965,
      "grad_norm": 1.3163216207252515,
      "learning_rate": 4.999289297353593e-06,
      "loss": 0.2318,
      "step": 127
    },
    {
      "epoch": 0.05767714317900192,
      "grad_norm": 1.221692049141614,
      "learning_rate": 4.999197688241076e-06,
      "loss": 0.2329,
      "step": 128
    },
    {
      "epoch": 0.05812774586008787,
      "grad_norm": 1.2504114744133756,
      "learning_rate": 4.999100528284045e-06,
      "loss": 0.246,
      "step": 129
    },
    {
      "epoch": 0.05857834854117382,
      "grad_norm": 1.1644125285534819,
      "learning_rate": 4.998997817698298e-06,
      "loss": 0.2203,
      "step": 130
    },
    {
      "epoch": 0.05902895122225977,
      "grad_norm": 1.1744114893394246,
      "learning_rate": 4.998889556711958e-06,
      "loss": 0.2328,
      "step": 131
    },
    {
      "epoch": 0.05947955390334572,
      "grad_norm": 1.2414638105224192,
      "learning_rate": 4.998775745565479e-06,
      "loss": 0.2465,
      "step": 132
    },
    {
      "epoch": 0.059930156584431675,
      "grad_norm": 1.249926424886369,
      "learning_rate": 4.998656384511643e-06,
      "loss": 0.2264,
      "step": 133
    },
    {
      "epoch": 0.06038075926551763,
      "grad_norm": 1.2572203072207218,
      "learning_rate": 4.9985314738155545e-06,
      "loss": 0.2345,
      "step": 134
    },
    {
      "epoch": 0.06083136194660358,
      "grad_norm": 1.1950518724866406,
      "learning_rate": 4.9984010137546475e-06,
      "loss": 0.231,
      "step": 135
    },
    {
      "epoch": 0.061281964627689534,
      "grad_norm": 1.2768234243037746,
      "learning_rate": 4.998265004618682e-06,
      "loss": 0.2213,
      "step": 136
    },
    {
      "epoch": 0.061732567308775486,
      "grad_norm": 1.2224411308170113,
      "learning_rate": 4.998123446709739e-06,
      "loss": 0.228,
      "step": 137
    },
    {
      "epoch": 0.06218316998986144,
      "grad_norm": 1.282263129739265,
      "learning_rate": 4.997976340342226e-06,
      "loss": 0.2167,
      "step": 138
    },
    {
      "epoch": 0.06263377267094739,
      "grad_norm": 1.1557240536583024,
      "learning_rate": 4.997823685842875e-06,
      "loss": 0.2076,
      "step": 139
    },
    {
      "epoch": 0.06308437535203335,
      "grad_norm": 1.284456691992294,
      "learning_rate": 4.997665483550739e-06,
      "loss": 0.2416,
      "step": 140
    },
    {
      "epoch": 0.06353497803311929,
      "grad_norm": 1.1028932126513764,
      "learning_rate": 4.997501733817191e-06,
      "loss": 0.2031,
      "step": 141
    },
    {
      "epoch": 0.06398558071420525,
      "grad_norm": 1.1975966518750552,
      "learning_rate": 4.997332437005932e-06,
      "loss": 0.2201,
      "step": 142
    },
    {
      "epoch": 0.0644361833952912,
      "grad_norm": 1.3778575705019447,
      "learning_rate": 4.997157593492974e-06,
      "loss": 0.2289,
      "step": 143
    },
    {
      "epoch": 0.06488678607637716,
      "grad_norm": 1.212901982108371,
      "learning_rate": 4.996977203666657e-06,
      "loss": 0.2263,
      "step": 144
    },
    {
      "epoch": 0.0653373887574631,
      "grad_norm": 1.2524708726219558,
      "learning_rate": 4.996791267927632e-06,
      "loss": 0.2418,
      "step": 145
    },
    {
      "epoch": 0.06578799143854906,
      "grad_norm": 1.1251179828824371,
      "learning_rate": 4.996599786688876e-06,
      "loss": 0.2262,
      "step": 146
    },
    {
      "epoch": 0.06623859411963501,
      "grad_norm": 1.1644234192309482,
      "learning_rate": 4.996402760375676e-06,
      "loss": 0.2223,
      "step": 147
    },
    {
      "epoch": 0.06668919680072097,
      "grad_norm": 1.2518098382195715,
      "learning_rate": 4.996200189425638e-06,
      "loss": 0.2304,
      "step": 148
    },
    {
      "epoch": 0.06713979948180691,
      "grad_norm": 1.268646829769258,
      "learning_rate": 4.9959920742886815e-06,
      "loss": 0.2381,
      "step": 149
    },
    {
      "epoch": 0.06759040216289287,
      "grad_norm": 1.1373995313396938,
      "learning_rate": 4.995778415427042e-06,
      "loss": 0.2295,
      "step": 150
    },
    {
      "epoch": 0.06804100484397882,
      "grad_norm": 1.4124184544565843,
      "learning_rate": 4.995559213315267e-06,
      "loss": 0.2319,
      "step": 151
    },
    {
      "epoch": 0.06849160752506478,
      "grad_norm": 1.1988625116568261,
      "learning_rate": 4.995334468440213e-06,
      "loss": 0.2239,
      "step": 152
    },
    {
      "epoch": 0.06894221020615073,
      "grad_norm": 1.2400456487781812,
      "learning_rate": 4.995104181301052e-06,
      "loss": 0.2364,
      "step": 153
    },
    {
      "epoch": 0.06939281288723669,
      "grad_norm": 1.169373197319975,
      "learning_rate": 4.994868352409263e-06,
      "loss": 0.2209,
      "step": 154
    },
    {
      "epoch": 0.06984341556832263,
      "grad_norm": 1.2105436784965689,
      "learning_rate": 4.9946269822886335e-06,
      "loss": 0.2294,
      "step": 155
    },
    {
      "epoch": 0.07029401824940859,
      "grad_norm": 1.1146096056941384,
      "learning_rate": 4.9943800714752586e-06,
      "loss": 0.214,
      "step": 156
    },
    {
      "epoch": 0.07074462093049454,
      "grad_norm": 1.107431297406545,
      "learning_rate": 4.9941276205175405e-06,
      "loss": 0.2153,
      "step": 157
    },
    {
      "epoch": 0.07119522361158048,
      "grad_norm": 1.1493108111633201,
      "learning_rate": 4.9938696299761856e-06,
      "loss": 0.2127,
      "step": 158
    },
    {
      "epoch": 0.07164582629266644,
      "grad_norm": 1.1357095289262957,
      "learning_rate": 4.993606100424202e-06,
      "loss": 0.2089,
      "step": 159
    },
    {
      "epoch": 0.07209642897375239,
      "grad_norm": 1.2497593534516505,
      "learning_rate": 4.9933370324469045e-06,
      "loss": 0.2326,
      "step": 160
    },
    {
      "epoch": 0.07254703165483835,
      "grad_norm": 1.119865007378002,
      "learning_rate": 4.993062426641906e-06,
      "loss": 0.2084,
      "step": 161
    },
    {
      "epoch": 0.0729976343359243,
      "grad_norm": 1.1584209347845942,
      "learning_rate": 4.9927822836191185e-06,
      "loss": 0.2124,
      "step": 162
    },
    {
      "epoch": 0.07344823701701025,
      "grad_norm": 1.1154381951406347,
      "learning_rate": 4.992496604000756e-06,
      "loss": 0.2123,
      "step": 163
    },
    {
      "epoch": 0.0738988396980962,
      "grad_norm": 1.2442046545527665,
      "learning_rate": 4.992205388421326e-06,
      "loss": 0.2253,
      "step": 164
    },
    {
      "epoch": 0.07434944237918216,
      "grad_norm": 1.300606227018797,
      "learning_rate": 4.991908637527634e-06,
      "loss": 0.2472,
      "step": 165
    },
    {
      "epoch": 0.0748000450602681,
      "grad_norm": 1.2450291533554696,
      "learning_rate": 4.9916063519787775e-06,
      "loss": 0.2369,
      "step": 166
    },
    {
      "epoch": 0.07525064774135407,
      "grad_norm": 1.2066695787715223,
      "learning_rate": 4.991298532446149e-06,
      "loss": 0.2242,
      "step": 167
    },
    {
      "epoch": 0.07570125042244001,
      "grad_norm": 1.0901339065424078,
      "learning_rate": 4.990985179613431e-06,
      "loss": 0.2155,
      "step": 168
    },
    {
      "epoch": 0.07615185310352597,
      "grad_norm": 1.2221127283502966,
      "learning_rate": 4.990666294176596e-06,
      "loss": 0.2231,
      "step": 169
    },
    {
      "epoch": 0.07660245578461192,
      "grad_norm": 1.1620496889627006,
      "learning_rate": 4.990341876843904e-06,
      "loss": 0.2211,
      "step": 170
    },
    {
      "epoch": 0.07705305846569788,
      "grad_norm": 1.3071221084528932,
      "learning_rate": 4.9900119283359025e-06,
      "loss": 0.2164,
      "step": 171
    },
    {
      "epoch": 0.07750366114678382,
      "grad_norm": 1.206047751336447,
      "learning_rate": 4.989676449385426e-06,
      "loss": 0.2281,
      "step": 172
    },
    {
      "epoch": 0.07795426382786978,
      "grad_norm": 1.4597757942350271,
      "learning_rate": 4.989335440737587e-06,
      "loss": 0.2506,
      "step": 173
    },
    {
      "epoch": 0.07840486650895573,
      "grad_norm": 1.379958299076823,
      "learning_rate": 4.988988903149784e-06,
      "loss": 0.2189,
      "step": 174
    },
    {
      "epoch": 0.07885546919004167,
      "grad_norm": 1.20920858944242,
      "learning_rate": 4.988636837391696e-06,
      "loss": 0.2314,
      "step": 175
    },
    {
      "epoch": 0.07930607187112763,
      "grad_norm": 1.2313142738643095,
      "learning_rate": 4.988279244245278e-06,
      "loss": 0.2205,
      "step": 176
    },
    {
      "epoch": 0.07975667455221358,
      "grad_norm": 1.3661071015991848,
      "learning_rate": 4.987916124504761e-06,
      "loss": 0.2306,
      "step": 177
    },
    {
      "epoch": 0.08020727723329954,
      "grad_norm": 1.1046734494634796,
      "learning_rate": 4.987547478976655e-06,
      "loss": 0.2081,
      "step": 178
    },
    {
      "epoch": 0.08065787991438549,
      "grad_norm": 1.3121004896955752,
      "learning_rate": 4.987173308479738e-06,
      "loss": 0.2515,
      "step": 179
    },
    {
      "epoch": 0.08110848259547145,
      "grad_norm": 1.0664036431852173,
      "learning_rate": 4.9867936138450635e-06,
      "loss": 0.2124,
      "step": 180
    },
    {
      "epoch": 0.08155908527655739,
      "grad_norm": 1.1905058604489476,
      "learning_rate": 4.98640839591595e-06,
      "loss": 0.2087,
      "step": 181
    },
    {
      "epoch": 0.08200968795764335,
      "grad_norm": 1.1400198695770818,
      "learning_rate": 4.986017655547989e-06,
      "loss": 0.2155,
      "step": 182
    },
    {
      "epoch": 0.0824602906387293,
      "grad_norm": 1.1958747980717914,
      "learning_rate": 4.985621393609032e-06,
      "loss": 0.2299,
      "step": 183
    },
    {
      "epoch": 0.08291089331981526,
      "grad_norm": 1.1612767983501473,
      "learning_rate": 4.9852196109792e-06,
      "loss": 0.2128,
      "step": 184
    },
    {
      "epoch": 0.0833614960009012,
      "grad_norm": 1.09026246469268,
      "learning_rate": 4.984812308550869e-06,
      "loss": 0.2099,
      "step": 185
    },
    {
      "epoch": 0.08381209868198716,
      "grad_norm": 1.235643788294065,
      "learning_rate": 4.98439948722868e-06,
      "loss": 0.2101,
      "step": 186
    },
    {
      "epoch": 0.08426270136307311,
      "grad_norm": 1.0869085522856416,
      "learning_rate": 4.98398114792953e-06,
      "loss": 0.2193,
      "step": 187
    },
    {
      "epoch": 0.08471330404415907,
      "grad_norm": 1.1675783445811345,
      "learning_rate": 4.983557291582572e-06,
      "loss": 0.2398,
      "step": 188
    },
    {
      "epoch": 0.08516390672524501,
      "grad_norm": 1.2182948524762915,
      "learning_rate": 4.9831279191292114e-06,
      "loss": 0.2315,
      "step": 189
    },
    {
      "epoch": 0.08561450940633097,
      "grad_norm": 1.1259138226408614,
      "learning_rate": 4.982693031523107e-06,
      "loss": 0.2234,
      "step": 190
    },
    {
      "epoch": 0.08606511208741692,
      "grad_norm": 1.117678078505707,
      "learning_rate": 4.982252629730167e-06,
      "loss": 0.2308,
      "step": 191
    },
    {
      "epoch": 0.08651571476850287,
      "grad_norm": 1.0912133687527328,
      "learning_rate": 4.981806714728543e-06,
      "loss": 0.205,
      "step": 192
    },
    {
      "epoch": 0.08696631744958883,
      "grad_norm": 1.1608197726702825,
      "learning_rate": 4.981355287508638e-06,
      "loss": 0.2223,
      "step": 193
    },
    {
      "epoch": 0.08741692013067477,
      "grad_norm": 1.1867139699465552,
      "learning_rate": 4.980898349073094e-06,
      "loss": 0.2356,
      "step": 194
    },
    {
      "epoch": 0.08786752281176073,
      "grad_norm": 1.1699794880063066,
      "learning_rate": 4.980435900436793e-06,
      "loss": 0.212,
      "step": 195
    },
    {
      "epoch": 0.08831812549284668,
      "grad_norm": 1.162570000485188,
      "learning_rate": 4.9799679426268575e-06,
      "loss": 0.2385,
      "step": 196
    },
    {
      "epoch": 0.08876872817393264,
      "grad_norm": 1.1699257911021248,
      "learning_rate": 4.979494476682647e-06,
      "loss": 0.2215,
      "step": 197
    },
    {
      "epoch": 0.08921933085501858,
      "grad_norm": 1.142705953177672,
      "learning_rate": 4.979015503655751e-06,
      "loss": 0.2426,
      "step": 198
    },
    {
      "epoch": 0.08966993353610454,
      "grad_norm": 1.1079034543480484,
      "learning_rate": 4.978531024609994e-06,
      "loss": 0.1941,
      "step": 199
    },
    {
      "epoch": 0.09012053621719049,
      "grad_norm": 1.3549506501630537,
      "learning_rate": 4.978041040621428e-06,
      "loss": 0.2129,
      "step": 200
    },
    {
      "epoch": 0.09057113889827645,
      "grad_norm": 1.2443686889827277,
      "learning_rate": 4.977545552778333e-06,
      "loss": 0.2423,
      "step": 201
    },
    {
      "epoch": 0.0910217415793624,
      "grad_norm": 1.3448778534838146,
      "learning_rate": 4.977044562181212e-06,
      "loss": 0.2263,
      "step": 202
    },
    {
      "epoch": 0.09147234426044835,
      "grad_norm": 1.3809990244237884,
      "learning_rate": 4.9765380699427905e-06,
      "loss": 0.2332,
      "step": 203
    },
    {
      "epoch": 0.0919229469415343,
      "grad_norm": 1.194186222769067,
      "learning_rate": 4.976026077188013e-06,
      "loss": 0.2123,
      "step": 204
    },
    {
      "epoch": 0.09237354962262026,
      "grad_norm": 1.2171969561529559,
      "learning_rate": 4.9755085850540426e-06,
      "loss": 0.2176,
      "step": 205
    },
    {
      "epoch": 0.0928241523037062,
      "grad_norm": 1.2441216278856086,
      "learning_rate": 4.974985594690255e-06,
      "loss": 0.219,
      "step": 206
    },
    {
      "epoch": 0.09327475498479217,
      "grad_norm": 1.1897634301147553,
      "learning_rate": 4.9744571072582365e-06,
      "loss": 0.2118,
      "step": 207
    },
    {
      "epoch": 0.09372535766587811,
      "grad_norm": 1.2669159238364731,
      "learning_rate": 4.973923123931786e-06,
      "loss": 0.2289,
      "step": 208
    },
    {
      "epoch": 0.09417596034696406,
      "grad_norm": 1.0996102317785694,
      "learning_rate": 4.973383645896908e-06,
      "loss": 0.2184,
      "step": 209
    },
    {
      "epoch": 0.09462656302805002,
      "grad_norm": 1.2185753643269248,
      "learning_rate": 4.97283867435181e-06,
      "loss": 0.2207,
      "step": 210
    },
    {
      "epoch": 0.09507716570913596,
      "grad_norm": 1.138433632415245,
      "learning_rate": 4.972288210506902e-06,
      "loss": 0.212,
      "step": 211
    },
    {
      "epoch": 0.09552776839022192,
      "grad_norm": 1.132998283546536,
      "learning_rate": 4.971732255584789e-06,
      "loss": 0.2367,
      "step": 212
    },
    {
      "epoch": 0.09597837107130787,
      "grad_norm": 1.0780010142719922,
      "learning_rate": 4.971170810820279e-06,
      "loss": 0.2113,
      "step": 213
    },
    {
      "epoch": 0.09642897375239383,
      "grad_norm": 1.1546441401328669,
      "learning_rate": 4.970603877460367e-06,
      "loss": 0.2252,
      "step": 214
    },
    {
      "epoch": 0.09687957643347977,
      "grad_norm": 1.1087587222673596,
      "learning_rate": 4.970031456764242e-06,
      "loss": 0.2202,
      "step": 215
    },
    {
      "epoch": 0.09733017911456573,
      "grad_norm": 1.0767581515636935,
      "learning_rate": 4.969453550003277e-06,
      "loss": 0.2074,
      "step": 216
    },
    {
      "epoch": 0.09778078179565168,
      "grad_norm": 1.2355538300715063,
      "learning_rate": 4.9688701584610345e-06,
      "loss": 0.2331,
      "step": 217
    },
    {
      "epoch": 0.09823138447673764,
      "grad_norm": 1.0345630784935012,
      "learning_rate": 4.968281283433256e-06,
      "loss": 0.2089,
      "step": 218
    },
    {
      "epoch": 0.09868198715782359,
      "grad_norm": 1.1203797258878334,
      "learning_rate": 4.967686926227862e-06,
      "loss": 0.1997,
      "step": 219
    },
    {
      "epoch": 0.09913258983890955,
      "grad_norm": 1.1144022340348496,
      "learning_rate": 4.967087088164951e-06,
      "loss": 0.205,
      "step": 220
    },
    {
      "epoch": 0.09958319251999549,
      "grad_norm": 1.0263017472875062,
      "learning_rate": 4.966481770576793e-06,
      "loss": 0.1965,
      "step": 221
    },
    {
      "epoch": 0.10003379520108145,
      "grad_norm": 1.1372226739430344,
      "learning_rate": 4.965870974807829e-06,
      "loss": 0.2166,
      "step": 222
    },
    {
      "epoch": 0.1004843978821674,
      "grad_norm": 1.1239275991584654,
      "learning_rate": 4.965254702214668e-06,
      "loss": 0.2011,
      "step": 223
    },
    {
      "epoch": 0.10093500056325336,
      "grad_norm": 1.2744989782535152,
      "learning_rate": 4.964632954166081e-06,
      "loss": 0.2139,
      "step": 224
    },
    {
      "epoch": 0.1013856032443393,
      "grad_norm": 1.1648813557027615,
      "learning_rate": 4.964005732043003e-06,
      "loss": 0.2103,
      "step": 225
    },
    {
      "epoch": 0.10183620592542525,
      "grad_norm": 1.0781643061072272,
      "learning_rate": 4.963373037238527e-06,
      "loss": 0.2101,
      "step": 226
    },
    {
      "epoch": 0.10228680860651121,
      "grad_norm": 1.2248828989051597,
      "learning_rate": 4.9627348711578996e-06,
      "loss": 0.2174,
      "step": 227
    },
    {
      "epoch": 0.10273741128759716,
      "grad_norm": 1.1754380302621135,
      "learning_rate": 4.962091235218518e-06,
      "loss": 0.207,
      "step": 228
    },
    {
      "epoch": 0.10318801396868311,
      "grad_norm": 1.1695992674462652,
      "learning_rate": 4.961442130849933e-06,
      "loss": 0.2207,
      "step": 229
    },
    {
      "epoch": 0.10363861664976906,
      "grad_norm": 1.1412176861499799,
      "learning_rate": 4.960787559493836e-06,
      "loss": 0.2327,
      "step": 230
    },
    {
      "epoch": 0.10408921933085502,
      "grad_norm": 1.2884964764354325,
      "learning_rate": 4.960127522604065e-06,
      "loss": 0.2151,
      "step": 231
    },
    {
      "epoch": 0.10453982201194097,
      "grad_norm": 1.1575013412061594,
      "learning_rate": 4.959462021646593e-06,
      "loss": 0.2082,
      "step": 232
    },
    {
      "epoch": 0.10499042469302693,
      "grad_norm": 1.041034111903912,
      "learning_rate": 4.958791058099533e-06,
      "loss": 0.2302,
      "step": 233
    },
    {
      "epoch": 0.10544102737411287,
      "grad_norm": 1.2510411794297505,
      "learning_rate": 4.95811463345313e-06,
      "loss": 0.2197,
      "step": 234
    },
    {
      "epoch": 0.10589163005519883,
      "grad_norm": 1.1556663492126247,
      "learning_rate": 4.957432749209755e-06,
      "loss": 0.2086,
      "step": 235
    },
    {
      "epoch": 0.10634223273628478,
      "grad_norm": 1.2247058461119267,
      "learning_rate": 4.956745406883909e-06,
      "loss": 0.2032,
      "step": 236
    },
    {
      "epoch": 0.10679283541737074,
      "grad_norm": 1.192347886801293,
      "learning_rate": 4.956052608002212e-06,
      "loss": 0.2254,
      "step": 237
    },
    {
      "epoch": 0.10724343809845668,
      "grad_norm": 1.223611134683997,
      "learning_rate": 4.9553543541034086e-06,
      "loss": 0.2285,
      "step": 238
    },
    {
      "epoch": 0.10769404077954264,
      "grad_norm": 1.249263593149513,
      "learning_rate": 4.954650646738354e-06,
      "loss": 0.2073,
      "step": 239
    },
    {
      "epoch": 0.10814464346062859,
      "grad_norm": 1.2193624268365642,
      "learning_rate": 4.953941487470017e-06,
      "loss": 0.224,
      "step": 240
    },
    {
      "epoch": 0.10859524614171455,
      "grad_norm": 1.1481178521059008,
      "learning_rate": 4.953226877873479e-06,
      "loss": 0.2107,
      "step": 241
    },
    {
      "epoch": 0.1090458488228005,
      "grad_norm": 1.10926361952047,
      "learning_rate": 4.952506819535922e-06,
      "loss": 0.2134,
      "step": 242
    },
    {
      "epoch": 0.10949645150388645,
      "grad_norm": 1.0597226992856483,
      "learning_rate": 4.951781314056633e-06,
      "loss": 0.2056,
      "step": 243
    },
    {
      "epoch": 0.1099470541849724,
      "grad_norm": 1.1791047972354118,
      "learning_rate": 4.951050363046995e-06,
      "loss": 0.232,
      "step": 244
    },
    {
      "epoch": 0.11039765686605835,
      "grad_norm": 1.234202693540785,
      "learning_rate": 4.950313968130488e-06,
      "loss": 0.241,
      "step": 245
    },
    {
      "epoch": 0.1108482595471443,
      "grad_norm": 1.1783658842486509,
      "learning_rate": 4.949572130942683e-06,
      "loss": 0.2186,
      "step": 246
    },
    {
      "epoch": 0.11129886222823025,
      "grad_norm": 1.1081432331819334,
      "learning_rate": 4.948824853131237e-06,
      "loss": 0.1974,
      "step": 247
    },
    {
      "epoch": 0.11174946490931621,
      "grad_norm": 1.122692447305528,
      "learning_rate": 4.948072136355892e-06,
      "loss": 0.2176,
      "step": 248
    },
    {
      "epoch": 0.11220006759040216,
      "grad_norm": 1.1112537761570087,
      "learning_rate": 4.94731398228847e-06,
      "loss": 0.2141,
      "step": 249
    },
    {
      "epoch": 0.11265067027148812,
      "grad_norm": 1.2229695315887537,
      "learning_rate": 4.94655039261287e-06,
      "loss": 0.2324,
      "step": 250
    },
    {
      "epoch": 0.11310127295257406,
      "grad_norm": 1.1468824200816254,
      "learning_rate": 4.9457813690250635e-06,
      "loss": 0.2212,
      "step": 251
    },
    {
      "epoch": 0.11355187563366002,
      "grad_norm": 1.1102275136522628,
      "learning_rate": 4.94500691323309e-06,
      "loss": 0.218,
      "step": 252
    },
    {
      "epoch": 0.11400247831474597,
      "grad_norm": 1.0561632156543708,
      "learning_rate": 4.9442270269570545e-06,
      "loss": 0.1878,
      "step": 253
    },
    {
      "epoch": 0.11445308099583193,
      "grad_norm": 1.0005190577253862,
      "learning_rate": 4.943441711929126e-06,
      "loss": 0.1999,
      "step": 254
    },
    {
      "epoch": 0.11490368367691788,
      "grad_norm": 1.2093067612482298,
      "learning_rate": 4.942650969893527e-06,
      "loss": 0.2288,
      "step": 255
    },
    {
      "epoch": 0.11535428635800384,
      "grad_norm": 1.1937432006596116,
      "learning_rate": 4.941854802606537e-06,
      "loss": 0.2274,
      "step": 256
    },
    {
      "epoch": 0.11580488903908978,
      "grad_norm": 1.1330892183935246,
      "learning_rate": 4.941053211836482e-06,
      "loss": 0.2198,
      "step": 257
    },
    {
      "epoch": 0.11625549172017574,
      "grad_norm": 1.0548114873532413,
      "learning_rate": 4.940246199363737e-06,
      "loss": 0.2115,
      "step": 258
    },
    {
      "epoch": 0.11670609440126169,
      "grad_norm": 1.0442025208171648,
      "learning_rate": 4.939433766980717e-06,
      "loss": 0.1878,
      "step": 259
    },
    {
      "epoch": 0.11715669708234765,
      "grad_norm": 1.1337795310606045,
      "learning_rate": 4.9386159164918764e-06,
      "loss": 0.2234,
      "step": 260
    },
    {
      "epoch": 0.11760729976343359,
      "grad_norm": 1.14992228918911,
      "learning_rate": 4.937792649713701e-06,
      "loss": 0.2237,
      "step": 261
    },
    {
      "epoch": 0.11805790244451954,
      "grad_norm": 1.14839235156934,
      "learning_rate": 4.9369639684747095e-06,
      "loss": 0.2221,
      "step": 262
    },
    {
      "epoch": 0.1185085051256055,
      "grad_norm": 1.0767897369879735,
      "learning_rate": 4.936129874615443e-06,
      "loss": 0.2039,
      "step": 263
    },
    {
      "epoch": 0.11895910780669144,
      "grad_norm": 0.9979796809326557,
      "learning_rate": 4.935290369988468e-06,
      "loss": 0.2034,
      "step": 264
    },
    {
      "epoch": 0.1194097104877774,
      "grad_norm": 1.0530126787897642,
      "learning_rate": 4.934445456458366e-06,
      "loss": 0.2211,
      "step": 265
    },
    {
      "epoch": 0.11986031316886335,
      "grad_norm": 1.1032116938045,
      "learning_rate": 4.933595135901733e-06,
      "loss": 0.2063,
      "step": 266
    },
    {
      "epoch": 0.12031091584994931,
      "grad_norm": 1.168793383146328,
      "learning_rate": 4.932739410207172e-06,
      "loss": 0.2217,
      "step": 267
    },
    {
      "epoch": 0.12076151853103526,
      "grad_norm": 1.1618233220895118,
      "learning_rate": 4.931878281275296e-06,
      "loss": 0.195,
      "step": 268
    },
    {
      "epoch": 0.12121212121212122,
      "grad_norm": 1.1834676718267456,
      "learning_rate": 4.931011751018715e-06,
      "loss": 0.2276,
      "step": 269
    },
    {
      "epoch": 0.12166272389320716,
      "grad_norm": 1.1934211671010873,
      "learning_rate": 4.930139821362036e-06,
      "loss": 0.2232,
      "step": 270
    },
    {
      "epoch": 0.12211332657429312,
      "grad_norm": 1.0812911650522306,
      "learning_rate": 4.929262494241859e-06,
      "loss": 0.2235,
      "step": 271
    },
    {
      "epoch": 0.12256392925537907,
      "grad_norm": 1.0858861894225589,
      "learning_rate": 4.928379771606773e-06,
      "loss": 0.2024,
      "step": 272
    },
    {
      "epoch": 0.12301453193646503,
      "grad_norm": 0.9937957531157218,
      "learning_rate": 4.927491655417347e-06,
      "loss": 0.2064,
      "step": 273
    },
    {
      "epoch": 0.12346513461755097,
      "grad_norm": 1.02867764239494,
      "learning_rate": 4.926598147646134e-06,
      "loss": 0.1913,
      "step": 274
    },
    {
      "epoch": 0.12391573729863693,
      "grad_norm": 1.1906053232315303,
      "learning_rate": 4.9256992502776605e-06,
      "loss": 0.2348,
      "step": 275
    },
    {
      "epoch": 0.12436633997972288,
      "grad_norm": 1.1698084508619528,
      "learning_rate": 4.924794965308421e-06,
      "loss": 0.2167,
      "step": 276
    },
    {
      "epoch": 0.12481694266080884,
      "grad_norm": 1.0742190591584848,
      "learning_rate": 4.9238852947468796e-06,
      "loss": 0.2228,
      "step": 277
    },
    {
      "epoch": 0.12526754534189477,
      "grad_norm": 1.2705248393755768,
      "learning_rate": 4.922970240613461e-06,
      "loss": 0.218,
      "step": 278
    },
    {
      "epoch": 0.12571814802298073,
      "grad_norm": 1.056558181907751,
      "learning_rate": 4.922049804940546e-06,
      "loss": 0.2061,
      "step": 279
    },
    {
      "epoch": 0.1261687507040667,
      "grad_norm": 1.1196465892653233,
      "learning_rate": 4.9211239897724685e-06,
      "loss": 0.2195,
      "step": 280
    },
    {
      "epoch": 0.12661935338515265,
      "grad_norm": 1.1621109193661827,
      "learning_rate": 4.920192797165511e-06,
      "loss": 0.2256,
      "step": 281
    },
    {
      "epoch": 0.12706995606623858,
      "grad_norm": 1.300028969200385,
      "learning_rate": 4.9192562291879e-06,
      "loss": 0.2344,
      "step": 282
    },
    {
      "epoch": 0.12752055874732454,
      "grad_norm": 1.1955520034272398,
      "learning_rate": 4.9183142879198e-06,
      "loss": 0.211,
      "step": 283
    },
    {
      "epoch": 0.1279711614284105,
      "grad_norm": 1.155931030252852,
      "learning_rate": 4.917366975453311e-06,
      "loss": 0.216,
      "step": 284
    },
    {
      "epoch": 0.12842176410949646,
      "grad_norm": 1.083861187612339,
      "learning_rate": 4.9164142938924595e-06,
      "loss": 0.2222,
      "step": 285
    },
    {
      "epoch": 0.1288723667905824,
      "grad_norm": 1.1248471153486164,
      "learning_rate": 4.915456245353202e-06,
      "loss": 0.2093,
      "step": 286
    },
    {
      "epoch": 0.12932296947166835,
      "grad_norm": 1.221633030384653,
      "learning_rate": 4.914492831963411e-06,
      "loss": 0.2086,
      "step": 287
    },
    {
      "epoch": 0.1297735721527543,
      "grad_norm": 1.2342848534915776,
      "learning_rate": 4.9135240558628786e-06,
      "loss": 0.224,
      "step": 288
    },
    {
      "epoch": 0.13022417483384027,
      "grad_norm": 1.0704411714834268,
      "learning_rate": 4.9125499192033035e-06,
      "loss": 0.1953,
      "step": 289
    },
    {
      "epoch": 0.1306747775149262,
      "grad_norm": 1.1195374340985238,
      "learning_rate": 4.911570424148293e-06,
      "loss": 0.2052,
      "step": 290
    },
    {
      "epoch": 0.13112538019601216,
      "grad_norm": 1.2051951166308854,
      "learning_rate": 4.910585572873355e-06,
      "loss": 0.2335,
      "step": 291
    },
    {
      "epoch": 0.13157598287709812,
      "grad_norm": 1.1643319404848254,
      "learning_rate": 4.9095953675658945e-06,
      "loss": 0.2146,
      "step": 292
    },
    {
      "epoch": 0.13202658555818408,
      "grad_norm": 1.0772311574842537,
      "learning_rate": 4.908599810425208e-06,
      "loss": 0.2192,
      "step": 293
    },
    {
      "epoch": 0.13247718823927002,
      "grad_norm": 1.0844966317512954,
      "learning_rate": 4.907598903662477e-06,
      "loss": 0.2222,
      "step": 294
    },
    {
      "epoch": 0.13292779092035598,
      "grad_norm": 1.0524330356877105,
      "learning_rate": 4.906592649500767e-06,
      "loss": 0.2306,
      "step": 295
    },
    {
      "epoch": 0.13337839360144194,
      "grad_norm": 1.1527168046897904,
      "learning_rate": 4.9055810501750205e-06,
      "loss": 0.2085,
      "step": 296
    },
    {
      "epoch": 0.13382899628252787,
      "grad_norm": 1.093657074566182,
      "learning_rate": 4.904564107932048e-06,
      "loss": 0.2107,
      "step": 297
    },
    {
      "epoch": 0.13427959896361383,
      "grad_norm": 1.0682596809770557,
      "learning_rate": 4.9035418250305314e-06,
      "loss": 0.2092,
      "step": 298
    },
    {
      "epoch": 0.1347302016446998,
      "grad_norm": 1.0844093558079646,
      "learning_rate": 4.902514203741013e-06,
      "loss": 0.2102,
      "step": 299
    },
    {
      "epoch": 0.13518080432578575,
      "grad_norm": 1.1018401031463165,
      "learning_rate": 4.9014812463458905e-06,
      "loss": 0.2192,
      "step": 300
    },
    {
      "epoch": 0.13563140700687168,
      "grad_norm": 1.1860232064810416,
      "learning_rate": 4.9004429551394155e-06,
      "loss": 0.2195,
      "step": 301
    },
    {
      "epoch": 0.13608200968795764,
      "grad_norm": 1.0961452643370815,
      "learning_rate": 4.899399332427685e-06,
      "loss": 0.214,
      "step": 302
    },
    {
      "epoch": 0.1365326123690436,
      "grad_norm": 1.1694339378011251,
      "learning_rate": 4.898350380528638e-06,
      "loss": 0.2242,
      "step": 303
    },
    {
      "epoch": 0.13698321505012956,
      "grad_norm": 1.1381527321950415,
      "learning_rate": 4.89729610177205e-06,
      "loss": 0.2129,
      "step": 304
    },
    {
      "epoch": 0.1374338177312155,
      "grad_norm": 1.0391770844547668,
      "learning_rate": 4.896236498499526e-06,
      "loss": 0.2175,
      "step": 305
    },
    {
      "epoch": 0.13788442041230145,
      "grad_norm": 1.195827567002921,
      "learning_rate": 4.8951715730645e-06,
      "loss": 0.2103,
      "step": 306
    },
    {
      "epoch": 0.1383350230933874,
      "grad_norm": 1.0423117479504835,
      "learning_rate": 4.894101327832225e-06,
      "loss": 0.2097,
      "step": 307
    },
    {
      "epoch": 0.13878562577447337,
      "grad_norm": 1.0259954075618385,
      "learning_rate": 4.89302576517977e-06,
      "loss": 0.1999,
      "step": 308
    },
    {
      "epoch": 0.1392362284555593,
      "grad_norm": 1.0589967702762215,
      "learning_rate": 4.891944887496013e-06,
      "loss": 0.22,
      "step": 309
    },
    {
      "epoch": 0.13968683113664526,
      "grad_norm": 1.1349746682270179,
      "learning_rate": 4.890858697181638e-06,
      "loss": 0.2167,
      "step": 310
    },
    {
      "epoch": 0.14013743381773122,
      "grad_norm": 1.1205630914176257,
      "learning_rate": 4.8897671966491315e-06,
      "loss": 0.2113,
      "step": 311
    },
    {
      "epoch": 0.14058803649881718,
      "grad_norm": 1.106503889764021,
      "learning_rate": 4.888670388322768e-06,
      "loss": 0.2199,
      "step": 312
    },
    {
      "epoch": 0.1410386391799031,
      "grad_norm": 1.1486711288247042,
      "learning_rate": 4.887568274638616e-06,
      "loss": 0.2196,
      "step": 313
    },
    {
      "epoch": 0.14148924186098907,
      "grad_norm": 1.0386088355319454,
      "learning_rate": 4.886460858044524e-06,
      "loss": 0.2107,
      "step": 314
    },
    {
      "epoch": 0.14193984454207503,
      "grad_norm": 1.1205610283336147,
      "learning_rate": 4.8853481410001225e-06,
      "loss": 0.2114,
      "step": 315
    },
    {
      "epoch": 0.14239044722316097,
      "grad_norm": 1.0635378718798172,
      "learning_rate": 4.884230125976812e-06,
      "loss": 0.215,
      "step": 316
    },
    {
      "epoch": 0.14284104990424693,
      "grad_norm": 1.1079718736074389,
      "learning_rate": 4.883106815457758e-06,
      "loss": 0.2083,
      "step": 317
    },
    {
      "epoch": 0.14329165258533288,
      "grad_norm": 1.093694928005843,
      "learning_rate": 4.881978211937895e-06,
      "loss": 0.208,
      "step": 318
    },
    {
      "epoch": 0.14374225526641884,
      "grad_norm": 1.189116563826118,
      "learning_rate": 4.8808443179239025e-06,
      "loss": 0.2353,
      "step": 319
    },
    {
      "epoch": 0.14419285794750478,
      "grad_norm": 1.0418971408472701,
      "learning_rate": 4.87970513593422e-06,
      "loss": 0.194,
      "step": 320
    },
    {
      "epoch": 0.14464346062859074,
      "grad_norm": 1.2566564381502143,
      "learning_rate": 4.878560668499029e-06,
      "loss": 0.2026,
      "step": 321
    },
    {
      "epoch": 0.1450940633096767,
      "grad_norm": 1.1717876183709541,
      "learning_rate": 4.877410918160247e-06,
      "loss": 0.2263,
      "step": 322
    },
    {
      "epoch": 0.14554466599076266,
      "grad_norm": 1.0480720367468714,
      "learning_rate": 4.87625588747153e-06,
      "loss": 0.1973,
      "step": 323
    },
    {
      "epoch": 0.1459952686718486,
      "grad_norm": 1.124387310287164,
      "learning_rate": 4.875095578998258e-06,
      "loss": 0.2079,
      "step": 324
    },
    {
      "epoch": 0.14644587135293455,
      "grad_norm": 1.0591920566737216,
      "learning_rate": 4.873929995317535e-06,
      "loss": 0.2112,
      "step": 325
    },
    {
      "epoch": 0.1468964740340205,
      "grad_norm": 1.0816650484251327,
      "learning_rate": 4.872759139018183e-06,
      "loss": 0.1993,
      "step": 326
    },
    {
      "epoch": 0.14734707671510647,
      "grad_norm": 1.048432719751396,
      "learning_rate": 4.87158301270073e-06,
      "loss": 0.2012,
      "step": 327
    },
    {
      "epoch": 0.1477976793961924,
      "grad_norm": 1.0772249510553387,
      "learning_rate": 4.870401618977415e-06,
      "loss": 0.2111,
      "step": 328
    },
    {
      "epoch": 0.14824828207727836,
      "grad_norm": 1.10268941823853,
      "learning_rate": 4.869214960472172e-06,
      "loss": 0.1897,
      "step": 329
    },
    {
      "epoch": 0.14869888475836432,
      "grad_norm": 1.1182951938865848,
      "learning_rate": 4.868023039820629e-06,
      "loss": 0.2033,
      "step": 330
    },
    {
      "epoch": 0.14914948743945025,
      "grad_norm": 1.1372072508746598,
      "learning_rate": 4.8668258596701035e-06,
      "loss": 0.22,
      "step": 331
    },
    {
      "epoch": 0.1496000901205362,
      "grad_norm": 1.2455087803794016,
      "learning_rate": 4.865623422679593e-06,
      "loss": 0.2272,
      "step": 332
    },
    {
      "epoch": 0.15005069280162217,
      "grad_norm": 1.0882360090058665,
      "learning_rate": 4.864415731519769e-06,
      "loss": 0.2232,
      "step": 333
    },
    {
      "epoch": 0.15050129548270813,
      "grad_norm": 1.1280177553587423,
      "learning_rate": 4.8632027888729765e-06,
      "loss": 0.2105,
      "step": 334
    },
    {
      "epoch": 0.15095189816379406,
      "grad_norm": 1.112279326112284,
      "learning_rate": 4.861984597433223e-06,
      "loss": 0.2189,
      "step": 335
    },
    {
      "epoch": 0.15140250084488002,
      "grad_norm": 0.9537982452845898,
      "learning_rate": 4.860761159906171e-06,
      "loss": 0.1827,
      "step": 336
    },
    {
      "epoch": 0.15185310352596598,
      "grad_norm": 1.0957786431292713,
      "learning_rate": 4.859532479009138e-06,
      "loss": 0.2112,
      "step": 337
    },
    {
      "epoch": 0.15230370620705194,
      "grad_norm": 1.0489404604521035,
      "learning_rate": 4.858298557471089e-06,
      "loss": 0.2092,
      "step": 338
    },
    {
      "epoch": 0.15275430888813787,
      "grad_norm": 1.0606372053262079,
      "learning_rate": 4.857059398032622e-06,
      "loss": 0.2042,
      "step": 339
    },
    {
      "epoch": 0.15320491156922383,
      "grad_norm": 1.0656352397918851,
      "learning_rate": 4.855815003445975e-06,
      "loss": 0.2111,
      "step": 340
    },
    {
      "epoch": 0.1536555142503098,
      "grad_norm": 1.1368796379748434,
      "learning_rate": 4.8545653764750125e-06,
      "loss": 0.2239,
      "step": 341
    },
    {
      "epoch": 0.15410611693139575,
      "grad_norm": 1.1472947549450268,
      "learning_rate": 4.853310519895217e-06,
      "loss": 0.2022,
      "step": 342
    },
    {
      "epoch": 0.15455671961248169,
      "grad_norm": 1.0772677119769347,
      "learning_rate": 4.85205043649369e-06,
      "loss": 0.2068,
      "step": 343
    },
    {
      "epoch": 0.15500732229356765,
      "grad_norm": 1.0646966878370703,
      "learning_rate": 4.850785129069139e-06,
      "loss": 0.2005,
      "step": 344
    },
    {
      "epoch": 0.1554579249746536,
      "grad_norm": 1.1191305904999096,
      "learning_rate": 4.849514600431877e-06,
      "loss": 0.2006,
      "step": 345
    },
    {
      "epoch": 0.15590852765573956,
      "grad_norm": 1.1166711817286015,
      "learning_rate": 4.848238853403813e-06,
      "loss": 0.2368,
      "step": 346
    },
    {
      "epoch": 0.1563591303368255,
      "grad_norm": 1.0823358127817442,
      "learning_rate": 4.846957890818444e-06,
      "loss": 0.2174,
      "step": 347
    },
    {
      "epoch": 0.15680973301791146,
      "grad_norm": 1.1055726640147139,
      "learning_rate": 4.845671715520853e-06,
      "loss": 0.1982,
      "step": 348
    },
    {
      "epoch": 0.15726033569899742,
      "grad_norm": 1.2381768961440425,
      "learning_rate": 4.844380330367701e-06,
      "loss": 0.2163,
      "step": 349
    },
    {
      "epoch": 0.15771093838008335,
      "grad_norm": 1.150113639278982,
      "learning_rate": 4.84308373822722e-06,
      "loss": 0.2211,
      "step": 350
    },
    {
      "epoch": 0.1581615410611693,
      "grad_norm": 1.0537141133450023,
      "learning_rate": 4.841781941979207e-06,
      "loss": 0.2055,
      "step": 351
    },
    {
      "epoch": 0.15861214374225527,
      "grad_norm": 1.213212670501889,
      "learning_rate": 4.840474944515017e-06,
      "loss": 0.2377,
      "step": 352
    },
    {
      "epoch": 0.15906274642334123,
      "grad_norm": 1.1778409627654554,
      "learning_rate": 4.839162748737556e-06,
      "loss": 0.2274,
      "step": 353
    },
    {
      "epoch": 0.15951334910442716,
      "grad_norm": 1.1285072146679263,
      "learning_rate": 4.8378453575612785e-06,
      "loss": 0.2235,
      "step": 354
    },
    {
      "epoch": 0.15996395178551312,
      "grad_norm": 1.0263423089549462,
      "learning_rate": 4.836522773912178e-06,
      "loss": 0.2088,
      "step": 355
    },
    {
      "epoch": 0.16041455446659908,
      "grad_norm": 1.1673268220763229,
      "learning_rate": 4.835195000727778e-06,
      "loss": 0.2186,
      "step": 356
    },
    {
      "epoch": 0.16086515714768504,
      "grad_norm": 1.010037043801117,
      "learning_rate": 4.83386204095713e-06,
      "loss": 0.2196,
      "step": 357
    },
    {
      "epoch": 0.16131575982877097,
      "grad_norm": 1.038204264055667,
      "learning_rate": 4.832523897560806e-06,
      "loss": 0.2085,
      "step": 358
    },
    {
      "epoch": 0.16176636250985693,
      "grad_norm": 1.014200195964716,
      "learning_rate": 4.83118057351089e-06,
      "loss": 0.2022,
      "step": 359
    },
    {
      "epoch": 0.1622169651909429,
      "grad_norm": 1.044154989517685,
      "learning_rate": 4.829832071790972e-06,
      "loss": 0.2096,
      "step": 360
    },
    {
      "epoch": 0.16266756787202885,
      "grad_norm": 1.0855706592511223,
      "learning_rate": 4.828478395396143e-06,
      "loss": 0.2167,
      "step": 361
    },
    {
      "epoch": 0.16311817055311478,
      "grad_norm": 1.144346091551533,
      "learning_rate": 4.827119547332988e-06,
      "loss": 0.2092,
      "step": 362
    },
    {
      "epoch": 0.16356877323420074,
      "grad_norm": 1.0894953489918704,
      "learning_rate": 4.825755530619576e-06,
      "loss": 0.1952,
      "step": 363
    },
    {
      "epoch": 0.1640193759152867,
      "grad_norm": 1.0974545315300446,
      "learning_rate": 4.824386348285456e-06,
      "loss": 0.2099,
      "step": 364
    },
    {
      "epoch": 0.16446997859637263,
      "grad_norm": 1.1097692923627165,
      "learning_rate": 4.8230120033716525e-06,
      "loss": 0.2006,
      "step": 365
    },
    {
      "epoch": 0.1649205812774586,
      "grad_norm": 1.065810253027864,
      "learning_rate": 4.821632498930656e-06,
      "loss": 0.2029,
      "step": 366
    },
    {
      "epoch": 0.16537118395854455,
      "grad_norm": 1.1383355108751643,
      "learning_rate": 4.820247838026414e-06,
      "loss": 0.2212,
      "step": 367
    },
    {
      "epoch": 0.16582178663963051,
      "grad_norm": 1.0915403947722182,
      "learning_rate": 4.81885802373433e-06,
      "loss": 0.2087,
      "step": 368
    },
    {
      "epoch": 0.16627238932071645,
      "grad_norm": 1.1135779035259634,
      "learning_rate": 4.8174630591412495e-06,
      "loss": 0.2038,
      "step": 369
    },
    {
      "epoch": 0.1667229920018024,
      "grad_norm": 1.0641064873717077,
      "learning_rate": 4.816062947345462e-06,
      "loss": 0.1974,
      "step": 370
    },
    {
      "epoch": 0.16717359468288837,
      "grad_norm": 1.101177412661662,
      "learning_rate": 4.814657691456685e-06,
      "loss": 0.2086,
      "step": 371
    },
    {
      "epoch": 0.16762419736397433,
      "grad_norm": 1.025754827090159,
      "learning_rate": 4.813247294596065e-06,
      "loss": 0.1984,
      "step": 372
    },
    {
      "epoch": 0.16807480004506026,
      "grad_norm": 1.0645510570530907,
      "learning_rate": 4.8118317598961625e-06,
      "loss": 0.2132,
      "step": 373
    },
    {
      "epoch": 0.16852540272614622,
      "grad_norm": 1.1139680760636317,
      "learning_rate": 4.810411090500952e-06,
      "loss": 0.2032,
      "step": 374
    },
    {
      "epoch": 0.16897600540723218,
      "grad_norm": 1.0717001593600692,
      "learning_rate": 4.808985289565813e-06,
      "loss": 0.1972,
      "step": 375
    },
    {
      "epoch": 0.16942660808831814,
      "grad_norm": 1.1069937243229389,
      "learning_rate": 4.807554360257522e-06,
      "loss": 0.2119,
      "step": 376
    },
    {
      "epoch": 0.16987721076940407,
      "grad_norm": 1.1217259717457204,
      "learning_rate": 4.8061183057542424e-06,
      "loss": 0.2078,
      "step": 377
    },
    {
      "epoch": 0.17032781345049003,
      "grad_norm": 1.0468196713561457,
      "learning_rate": 4.804677129245527e-06,
      "loss": 0.2038,
      "step": 378
    },
    {
      "epoch": 0.170778416131576,
      "grad_norm": 1.034643233619461,
      "learning_rate": 4.803230833932302e-06,
      "loss": 0.2051,
      "step": 379
    },
    {
      "epoch": 0.17122901881266195,
      "grad_norm": 1.0967312312186066,
      "learning_rate": 4.80177942302686e-06,
      "loss": 0.2101,
      "step": 380
    },
    {
      "epoch": 0.17167962149374788,
      "grad_norm": 1.1600308475491419,
      "learning_rate": 4.800322899752859e-06,
      "loss": 0.2215,
      "step": 381
    },
    {
      "epoch": 0.17213022417483384,
      "grad_norm": 1.0126108557101248,
      "learning_rate": 4.798861267345312e-06,
      "loss": 0.1999,
      "step": 382
    },
    {
      "epoch": 0.1725808268559198,
      "grad_norm": 0.9515557584883886,
      "learning_rate": 4.797394529050577e-06,
      "loss": 0.1771,
      "step": 383
    },
    {
      "epoch": 0.17303142953700573,
      "grad_norm": 1.1155117108927086,
      "learning_rate": 4.795922688126355e-06,
      "loss": 0.2343,
      "step": 384
    },
    {
      "epoch": 0.1734820322180917,
      "grad_norm": 0.9933539774509211,
      "learning_rate": 4.794445747841679e-06,
      "loss": 0.2018,
      "step": 385
    },
    {
      "epoch": 0.17393263489917765,
      "grad_norm": 1.0253085493855625,
      "learning_rate": 4.792963711476908e-06,
      "loss": 0.2109,
      "step": 386
    },
    {
      "epoch": 0.1743832375802636,
      "grad_norm": 1.0937444150984799,
      "learning_rate": 4.791476582323719e-06,
      "loss": 0.203,
      "step": 387
    },
    {
      "epoch": 0.17483384026134954,
      "grad_norm": 1.0994215912796548,
      "learning_rate": 4.7899843636851014e-06,
      "loss": 0.2066,
      "step": 388
    },
    {
      "epoch": 0.1752844429424355,
      "grad_norm": 1.0391589566973922,
      "learning_rate": 4.78848705887535e-06,
      "loss": 0.1987,
      "step": 389
    },
    {
      "epoch": 0.17573504562352146,
      "grad_norm": 1.025549842249766,
      "learning_rate": 4.786984671220053e-06,
      "loss": 0.2023,
      "step": 390
    },
    {
      "epoch": 0.17618564830460742,
      "grad_norm": 1.1952860424909204,
      "learning_rate": 4.785477204056089e-06,
      "loss": 0.2218,
      "step": 391
    },
    {
      "epoch": 0.17663625098569335,
      "grad_norm": 1.0996497512784118,
      "learning_rate": 4.78396466073162e-06,
      "loss": 0.2119,
      "step": 392
    },
    {
      "epoch": 0.17708685366677931,
      "grad_norm": 1.225758035064243,
      "learning_rate": 4.78244704460608e-06,
      "loss": 0.22,
      "step": 393
    },
    {
      "epoch": 0.17753745634786527,
      "grad_norm": 1.1385526357533455,
      "learning_rate": 4.7809243590501725e-06,
      "loss": 0.2062,
      "step": 394
    },
    {
      "epoch": 0.17798805902895123,
      "grad_norm": 1.0395990820077765,
      "learning_rate": 4.779396607445858e-06,
      "loss": 0.2056,
      "step": 395
    },
    {
      "epoch": 0.17843866171003717,
      "grad_norm": 1.1634899736709214,
      "learning_rate": 4.777863793186351e-06,
      "loss": 0.2025,
      "step": 396
    },
    {
      "epoch": 0.17888926439112313,
      "grad_norm": 1.2017617562113168,
      "learning_rate": 4.776325919676109e-06,
      "loss": 0.2005,
      "step": 397
    },
    {
      "epoch": 0.17933986707220909,
      "grad_norm": 1.0989936285862003,
      "learning_rate": 4.774782990330828e-06,
      "loss": 0.2109,
      "step": 398
    },
    {
      "epoch": 0.17979046975329505,
      "grad_norm": 1.0609884089685264,
      "learning_rate": 4.77323500857743e-06,
      "loss": 0.2122,
      "step": 399
    },
    {
      "epoch": 0.18024107243438098,
      "grad_norm": 0.9956576638871627,
      "learning_rate": 4.771681977854062e-06,
      "loss": 0.2017,
      "step": 400
    },
    {
      "epoch": 0.18069167511546694,
      "grad_norm": 1.0886524252129588,
      "learning_rate": 4.770123901610085e-06,
      "loss": 0.2165,
      "step": 401
    },
    {
      "epoch": 0.1811422777965529,
      "grad_norm": 1.049117446897747,
      "learning_rate": 4.768560783306064e-06,
      "loss": 0.2129,
      "step": 402
    },
    {
      "epoch": 0.18159288047763883,
      "grad_norm": 1.1728204295468192,
      "learning_rate": 4.7669926264137625e-06,
      "loss": 0.2196,
      "step": 403
    },
    {
      "epoch": 0.1820434831587248,
      "grad_norm": 1.0530113188774437,
      "learning_rate": 4.765419434416138e-06,
      "loss": 0.1988,
      "step": 404
    },
    {
      "epoch": 0.18249408583981075,
      "grad_norm": 1.1006552707986152,
      "learning_rate": 4.763841210807329e-06,
      "loss": 0.1969,
      "step": 405
    },
    {
      "epoch": 0.1829446885208967,
      "grad_norm": 1.1056839598789527,
      "learning_rate": 4.762257959092651e-06,
      "loss": 0.2165,
      "step": 406
    },
    {
      "epoch": 0.18339529120198264,
      "grad_norm": 1.166339387634052,
      "learning_rate": 4.760669682788584e-06,
      "loss": 0.2025,
      "step": 407
    },
    {
      "epoch": 0.1838458938830686,
      "grad_norm": 1.1845695122450088,
      "learning_rate": 4.759076385422773e-06,
      "loss": 0.2288,
      "step": 408
    },
    {
      "epoch": 0.18429649656415456,
      "grad_norm": 1.0314361289522471,
      "learning_rate": 4.7574780705340094e-06,
      "loss": 0.212,
      "step": 409
    },
    {
      "epoch": 0.18474709924524052,
      "grad_norm": 1.0060908332453342,
      "learning_rate": 4.755874741672233e-06,
      "loss": 0.1893,
      "step": 410
    },
    {
      "epoch": 0.18519770192632645,
      "grad_norm": 0.9813435104532242,
      "learning_rate": 4.754266402398517e-06,
      "loss": 0.2055,
      "step": 411
    },
    {
      "epoch": 0.1856483046074124,
      "grad_norm": 1.0197480260218619,
      "learning_rate": 4.752653056285066e-06,
      "loss": 0.2134,
      "step": 412
    },
    {
      "epoch": 0.18609890728849837,
      "grad_norm": 1.0464587885926853,
      "learning_rate": 4.7510347069152015e-06,
      "loss": 0.2224,
      "step": 413
    },
    {
      "epoch": 0.18654950996958433,
      "grad_norm": 1.1570694502589038,
      "learning_rate": 4.74941135788336e-06,
      "loss": 0.2171,
      "step": 414
    },
    {
      "epoch": 0.18700011265067026,
      "grad_norm": 1.0691121161245825,
      "learning_rate": 4.747783012795083e-06,
      "loss": 0.1931,
      "step": 415
    },
    {
      "epoch": 0.18745071533175622,
      "grad_norm": 1.0451364800748664,
      "learning_rate": 4.746149675267005e-06,
      "loss": 0.193,
      "step": 416
    },
    {
      "epoch": 0.18790131801284218,
      "grad_norm": 1.135060265323339,
      "learning_rate": 4.744511348926855e-06,
      "loss": 0.2188,
      "step": 417
    },
    {
      "epoch": 0.18835192069392812,
      "grad_norm": 1.0453942068357704,
      "learning_rate": 4.742868037413435e-06,
      "loss": 0.1882,
      "step": 418
    },
    {
      "epoch": 0.18880252337501408,
      "grad_norm": 1.1036284009807769,
      "learning_rate": 4.741219744376624e-06,
      "loss": 0.2077,
      "step": 419
    },
    {
      "epoch": 0.18925312605610004,
      "grad_norm": 1.0896714127730052,
      "learning_rate": 4.739566473477365e-06,
      "loss": 0.1957,
      "step": 420
    },
    {
      "epoch": 0.189703728737186,
      "grad_norm": 1.093929205439995,
      "learning_rate": 4.737908228387656e-06,
      "loss": 0.2204,
      "step": 421
    },
    {
      "epoch": 0.19015433141827193,
      "grad_norm": 0.957604866376255,
      "learning_rate": 4.736245012790543e-06,
      "loss": 0.1875,
      "step": 422
    },
    {
      "epoch": 0.1906049340993579,
      "grad_norm": 1.1117554592845362,
      "learning_rate": 4.734576830380113e-06,
      "loss": 0.2054,
      "step": 423
    },
    {
      "epoch": 0.19105553678044385,
      "grad_norm": 1.0285882348865756,
      "learning_rate": 4.732903684861482e-06,
      "loss": 0.1886,
      "step": 424
    },
    {
      "epoch": 0.1915061394615298,
      "grad_norm": 1.0396359377134285,
      "learning_rate": 4.731225579950791e-06,
      "loss": 0.2,
      "step": 425
    },
    {
      "epoch": 0.19195674214261574,
      "grad_norm": 1.0661040136052777,
      "learning_rate": 4.7295425193751974e-06,
      "loss": 0.2043,
      "step": 426
    },
    {
      "epoch": 0.1924073448237017,
      "grad_norm": 1.1689089296608768,
      "learning_rate": 4.727854506872863e-06,
      "loss": 0.2203,
      "step": 427
    },
    {
      "epoch": 0.19285794750478766,
      "grad_norm": 1.0981202998096562,
      "learning_rate": 4.726161546192949e-06,
      "loss": 0.2138,
      "step": 428
    },
    {
      "epoch": 0.19330855018587362,
      "grad_norm": 1.0895700511789579,
      "learning_rate": 4.724463641095606e-06,
      "loss": 0.2092,
      "step": 429
    },
    {
      "epoch": 0.19375915286695955,
      "grad_norm": 1.1454103133342515,
      "learning_rate": 4.7227607953519686e-06,
      "loss": 0.2123,
      "step": 430
    },
    {
      "epoch": 0.1942097555480455,
      "grad_norm": 1.081115059551756,
      "learning_rate": 4.721053012744142e-06,
      "loss": 0.2004,
      "step": 431
    },
    {
      "epoch": 0.19466035822913147,
      "grad_norm": 1.047592461395092,
      "learning_rate": 4.719340297065198e-06,
      "loss": 0.2012,
      "step": 432
    },
    {
      "epoch": 0.19511096091021743,
      "grad_norm": 1.0496672476197972,
      "learning_rate": 4.717622652119166e-06,
      "loss": 0.2096,
      "step": 433
    },
    {
      "epoch": 0.19556156359130336,
      "grad_norm": 1.1483209659869804,
      "learning_rate": 4.715900081721021e-06,
      "loss": 0.2035,
      "step": 434
    },
    {
      "epoch": 0.19601216627238932,
      "grad_norm": 1.0567737050713122,
      "learning_rate": 4.71417258969668e-06,
      "loss": 0.2093,
      "step": 435
    },
    {
      "epoch": 0.19646276895347528,
      "grad_norm": 1.006874897763615,
      "learning_rate": 4.712440179882989e-06,
      "loss": 0.1982,
      "step": 436
    },
    {
      "epoch": 0.1969133716345612,
      "grad_norm": 1.0512955804872948,
      "learning_rate": 4.710702856127718e-06,
      "loss": 0.2128,
      "step": 437
    },
    {
      "epoch": 0.19736397431564717,
      "grad_norm": 1.0421413033250877,
      "learning_rate": 4.708960622289552e-06,
      "loss": 0.196,
      "step": 438
    },
    {
      "epoch": 0.19781457699673313,
      "grad_norm": 1.0393078215848977,
      "learning_rate": 4.70721348223808e-06,
      "loss": 0.2003,
      "step": 439
    },
    {
      "epoch": 0.1982651796778191,
      "grad_norm": 1.0652170410294124,
      "learning_rate": 4.705461439853789e-06,
      "loss": 0.2038,
      "step": 440
    },
    {
      "epoch": 0.19871578235890502,
      "grad_norm": 1.148884011184797,
      "learning_rate": 4.703704499028052e-06,
      "loss": 0.2105,
      "step": 441
    },
    {
      "epoch": 0.19916638503999098,
      "grad_norm": 1.1182902562395605,
      "learning_rate": 4.701942663663126e-06,
      "loss": 0.2115,
      "step": 442
    },
    {
      "epoch": 0.19961698772107694,
      "grad_norm": 1.0273097092583952,
      "learning_rate": 4.700175937672134e-06,
      "loss": 0.2096,
      "step": 443
    },
    {
      "epoch": 0.2000675904021629,
      "grad_norm": 1.0524293589683538,
      "learning_rate": 4.698404324979066e-06,
      "loss": 0.2152,
      "step": 444
    },
    {
      "epoch": 0.20051819308324884,
      "grad_norm": 1.0461725590043855,
      "learning_rate": 4.696627829518761e-06,
      "loss": 0.2095,
      "step": 445
    },
    {
      "epoch": 0.2009687957643348,
      "grad_norm": 1.1052123985360112,
      "learning_rate": 4.6948464552369075e-06,
      "loss": 0.2008,
      "step": 446
    },
    {
      "epoch": 0.20141939844542076,
      "grad_norm": 1.0488786380047408,
      "learning_rate": 4.693060206090028e-06,
      "loss": 0.203,
      "step": 447
    },
    {
      "epoch": 0.20187000112650672,
      "grad_norm": 1.1015751617793508,
      "learning_rate": 4.69126908604547e-06,
      "loss": 0.2022,
      "step": 448
    },
    {
      "epoch": 0.20232060380759265,
      "grad_norm": 1.0223628260749347,
      "learning_rate": 4.689473099081403e-06,
      "loss": 0.2077,
      "step": 449
    },
    {
      "epoch": 0.2027712064886786,
      "grad_norm": 1.1696606518556552,
      "learning_rate": 4.687672249186805e-06,
      "loss": 0.2236,
      "step": 450
    },
    {
      "epoch": 0.20322180916976457,
      "grad_norm": 1.0789601171063559,
      "learning_rate": 4.685866540361456e-06,
      "loss": 0.1996,
      "step": 451
    },
    {
      "epoch": 0.2036724118508505,
      "grad_norm": 1.1208502934238764,
      "learning_rate": 4.684055976615924e-06,
      "loss": 0.2077,
      "step": 452
    },
    {
      "epoch": 0.20412301453193646,
      "grad_norm": 1.1152284615506332,
      "learning_rate": 4.682240561971565e-06,
      "loss": 0.2188,
      "step": 453
    },
    {
      "epoch": 0.20457361721302242,
      "grad_norm": 1.06037564984265,
      "learning_rate": 4.680420300460505e-06,
      "loss": 0.2118,
      "step": 454
    },
    {
      "epoch": 0.20502421989410838,
      "grad_norm": 1.0336715129306815,
      "learning_rate": 4.678595196125638e-06,
      "loss": 0.2088,
      "step": 455
    },
    {
      "epoch": 0.2054748225751943,
      "grad_norm": 1.0917414927010969,
      "learning_rate": 4.676765253020613e-06,
      "loss": 0.2027,
      "step": 456
    },
    {
      "epoch": 0.20592542525628027,
      "grad_norm": 1.115251383762241,
      "learning_rate": 4.674930475209827e-06,
      "loss": 0.2171,
      "step": 457
    },
    {
      "epoch": 0.20637602793736623,
      "grad_norm": 1.0157487847463096,
      "learning_rate": 4.673090866768412e-06,
      "loss": 0.2004,
      "step": 458
    },
    {
      "epoch": 0.2068266306184522,
      "grad_norm": 1.0532015695939707,
      "learning_rate": 4.671246431782234e-06,
      "loss": 0.2017,
      "step": 459
    },
    {
      "epoch": 0.20727723329953812,
      "grad_norm": 1.1298334121350209,
      "learning_rate": 4.669397174347874e-06,
      "loss": 0.2153,
      "step": 460
    },
    {
      "epoch": 0.20772783598062408,
      "grad_norm": 1.1027909606697068,
      "learning_rate": 4.667543098572627e-06,
      "loss": 0.206,
      "step": 461
    },
    {
      "epoch": 0.20817843866171004,
      "grad_norm": 1.088777757241232,
      "learning_rate": 4.665684208574492e-06,
      "loss": 0.2272,
      "step": 462
    },
    {
      "epoch": 0.208629041342796,
      "grad_norm": 1.189137049814458,
      "learning_rate": 4.6638205084821544e-06,
      "loss": 0.2178,
      "step": 463
    },
    {
      "epoch": 0.20907964402388193,
      "grad_norm": 1.0758757066228637,
      "learning_rate": 4.661952002434988e-06,
      "loss": 0.2166,
      "step": 464
    },
    {
      "epoch": 0.2095302467049679,
      "grad_norm": 1.0313498928838842,
      "learning_rate": 4.660078694583037e-06,
      "loss": 0.2157,
      "step": 465
    },
    {
      "epoch": 0.20998084938605385,
      "grad_norm": 1.023803389318251,
      "learning_rate": 4.658200589087016e-06,
      "loss": 0.2086,
      "step": 466
    },
    {
      "epoch": 0.2104314520671398,
      "grad_norm": 0.9904629162792291,
      "learning_rate": 4.656317690118291e-06,
      "loss": 0.188,
      "step": 467
    },
    {
      "epoch": 0.21088205474822574,
      "grad_norm": 1.0719361560549883,
      "learning_rate": 4.654430001858874e-06,
      "loss": 0.1975,
      "step": 468
    },
    {
      "epoch": 0.2113326574293117,
      "grad_norm": 1.0159140827322095,
      "learning_rate": 4.6525375285014195e-06,
      "loss": 0.2041,
      "step": 469
    },
    {
      "epoch": 0.21178326011039766,
      "grad_norm": 1.057379808185924,
      "learning_rate": 4.650640274249205e-06,
      "loss": 0.1936,
      "step": 470
    },
    {
      "epoch": 0.2122338627914836,
      "grad_norm": 1.0517236533065415,
      "learning_rate": 4.648738243316128e-06,
      "loss": 0.2069,
      "step": 471
    },
    {
      "epoch": 0.21268446547256956,
      "grad_norm": 1.283567142637064,
      "learning_rate": 4.646831439926696e-06,
      "loss": 0.2162,
      "step": 472
    },
    {
      "epoch": 0.21313506815365552,
      "grad_norm": 1.0710202659430694,
      "learning_rate": 4.644919868316014e-06,
      "loss": 0.2083,
      "step": 473
    },
    {
      "epoch": 0.21358567083474148,
      "grad_norm": 1.082095821155301,
      "learning_rate": 4.643003532729783e-06,
      "loss": 0.2109,
      "step": 474
    },
    {
      "epoch": 0.2140362735158274,
      "grad_norm": 1.0884595291687782,
      "learning_rate": 4.641082437424277e-06,
      "loss": 0.205,
      "step": 475
    },
    {
      "epoch": 0.21448687619691337,
      "grad_norm": 1.0536622056317448,
      "learning_rate": 4.639156586666349e-06,
      "loss": 0.2103,
      "step": 476
    },
    {
      "epoch": 0.21493747887799933,
      "grad_norm": 1.0232598149017575,
      "learning_rate": 4.63722598473341e-06,
      "loss": 0.2066,
      "step": 477
    },
    {
      "epoch": 0.2153880815590853,
      "grad_norm": 1.033764557474931,
      "learning_rate": 4.635290635913425e-06,
      "loss": 0.1858,
      "step": 478
    },
    {
      "epoch": 0.21583868424017122,
      "grad_norm": 1.0772507958771667,
      "learning_rate": 4.633350544504899e-06,
      "loss": 0.2181,
      "step": 479
    },
    {
      "epoch": 0.21628928692125718,
      "grad_norm": 0.970929042699419,
      "learning_rate": 4.6314057148168765e-06,
      "loss": 0.2022,
      "step": 480
    },
    {
      "epoch": 0.21673988960234314,
      "grad_norm": 1.0328732636920106,
      "learning_rate": 4.629456151168921e-06,
      "loss": 0.2078,
      "step": 481
    },
    {
      "epoch": 0.2171904922834291,
      "grad_norm": 0.9803373687650507,
      "learning_rate": 4.627501857891113e-06,
      "loss": 0.1961,
      "step": 482
    },
    {
      "epoch": 0.21764109496451503,
      "grad_norm": 0.9604150843019789,
      "learning_rate": 4.625542839324036e-06,
      "loss": 0.1949,
      "step": 483
    },
    {
      "epoch": 0.218091697645601,
      "grad_norm": 1.075511254310141,
      "learning_rate": 4.623579099818769e-06,
      "loss": 0.2203,
      "step": 484
    },
    {
      "epoch": 0.21854230032668695,
      "grad_norm": 1.0782563669494776,
      "learning_rate": 4.621610643736878e-06,
      "loss": 0.2148,
      "step": 485
    },
    {
      "epoch": 0.2189929030077729,
      "grad_norm": 1.03653667368632,
      "learning_rate": 4.6196374754504024e-06,
      "loss": 0.1953,
      "step": 486
    },
    {
      "epoch": 0.21944350568885884,
      "grad_norm": 1.0098473894676974,
      "learning_rate": 4.617659599341849e-06,
      "loss": 0.1939,
      "step": 487
    },
    {
      "epoch": 0.2198941083699448,
      "grad_norm": 1.130106809561545,
      "learning_rate": 4.615677019804182e-06,
      "loss": 0.2112,
      "step": 488
    },
    {
      "epoch": 0.22034471105103076,
      "grad_norm": 0.9678884435582474,
      "learning_rate": 4.6136897412408084e-06,
      "loss": 0.1814,
      "step": 489
    },
    {
      "epoch": 0.2207953137321167,
      "grad_norm": 1.013409423056004,
      "learning_rate": 4.611697768065577e-06,
      "loss": 0.1911,
      "step": 490
    },
    {
      "epoch": 0.22124591641320265,
      "grad_norm": 1.075379836070552,
      "learning_rate": 4.609701104702759e-06,
      "loss": 0.202,
      "step": 491
    },
    {
      "epoch": 0.2216965190942886,
      "grad_norm": 1.0091142238608348,
      "learning_rate": 4.607699755587046e-06,
      "loss": 0.2074,
      "step": 492
    },
    {
      "epoch": 0.22214712177537457,
      "grad_norm": 1.0663643831417986,
      "learning_rate": 4.605693725163536e-06,
      "loss": 0.1986,
      "step": 493
    },
    {
      "epoch": 0.2225977244564605,
      "grad_norm": 1.145112521709734,
      "learning_rate": 4.603683017887722e-06,
      "loss": 0.2005,
      "step": 494
    },
    {
      "epoch": 0.22304832713754646,
      "grad_norm": 1.0339947461099595,
      "learning_rate": 4.6016676382254895e-06,
      "loss": 0.2077,
      "step": 495
    },
    {
      "epoch": 0.22349892981863242,
      "grad_norm": 1.0179207841068676,
      "learning_rate": 4.5996475906530955e-06,
      "loss": 0.2023,
      "step": 496
    },
    {
      "epoch": 0.22394953249971838,
      "grad_norm": 1.0543173957144043,
      "learning_rate": 4.597622879657171e-06,
      "loss": 0.1885,
      "step": 497
    },
    {
      "epoch": 0.22440013518080432,
      "grad_norm": 1.052136332232682,
      "learning_rate": 4.595593509734699e-06,
      "loss": 0.2052,
      "step": 498
    },
    {
      "epoch": 0.22485073786189028,
      "grad_norm": 1.0295283490102973,
      "learning_rate": 4.593559485393015e-06,
      "loss": 0.199,
      "step": 499
    },
    {
      "epoch": 0.22530134054297624,
      "grad_norm": 1.0855825905904153,
      "learning_rate": 4.591520811149787e-06,
      "loss": 0.2103,
      "step": 500
    },
    {
      "epoch": 0.22530134054297624,
      "eval_loss": 0.20520994067192078,
      "eval_runtime": 59.4657,
      "eval_samples_per_second": 24.132,
      "eval_steps_per_second": 3.027,
      "step": 500
    },
    {
      "epoch": 0.2257519432240622,
      "grad_norm": 1.0302681861841023,
      "learning_rate": 4.589477491533016e-06,
      "loss": 0.186,
      "step": 501
    },
    {
      "epoch": 0.22620254590514813,
      "grad_norm": 1.061738831547382,
      "learning_rate": 4.587429531081019e-06,
      "loss": 0.2081,
      "step": 502
    },
    {
      "epoch": 0.2266531485862341,
      "grad_norm": 0.9794242056013339,
      "learning_rate": 4.585376934342418e-06,
      "loss": 0.1995,
      "step": 503
    },
    {
      "epoch": 0.22710375126732005,
      "grad_norm": 0.9686601035730674,
      "learning_rate": 4.583319705876133e-06,
      "loss": 0.1873,
      "step": 504
    },
    {
      "epoch": 0.22755435394840598,
      "grad_norm": 0.9664239720091494,
      "learning_rate": 4.581257850251376e-06,
      "loss": 0.1876,
      "step": 505
    },
    {
      "epoch": 0.22800495662949194,
      "grad_norm": 1.0649310131480245,
      "learning_rate": 4.579191372047631e-06,
      "loss": 0.1981,
      "step": 506
    },
    {
      "epoch": 0.2284555593105779,
      "grad_norm": 1.054654976580904,
      "learning_rate": 4.577120275854649e-06,
      "loss": 0.2,
      "step": 507
    },
    {
      "epoch": 0.22890616199166386,
      "grad_norm": 1.0938236596107724,
      "learning_rate": 4.5750445662724426e-06,
      "loss": 0.2133,
      "step": 508
    },
    {
      "epoch": 0.2293567646727498,
      "grad_norm": 0.9920852942754904,
      "learning_rate": 4.572964247911265e-06,
      "loss": 0.1971,
      "step": 509
    },
    {
      "epoch": 0.22980736735383575,
      "grad_norm": 1.044583342143946,
      "learning_rate": 4.5708793253916104e-06,
      "loss": 0.1967,
      "step": 510
    },
    {
      "epoch": 0.2302579700349217,
      "grad_norm": 1.1506941684890593,
      "learning_rate": 4.568789803344196e-06,
      "loss": 0.2156,
      "step": 511
    },
    {
      "epoch": 0.23070857271600767,
      "grad_norm": 0.9861261159798321,
      "learning_rate": 4.566695686409957e-06,
      "loss": 0.1931,
      "step": 512
    },
    {
      "epoch": 0.2311591753970936,
      "grad_norm": 1.0039176366458205,
      "learning_rate": 4.564596979240031e-06,
      "loss": 0.1922,
      "step": 513
    },
    {
      "epoch": 0.23160977807817956,
      "grad_norm": 1.0205337354605422,
      "learning_rate": 4.562493686495756e-06,
      "loss": 0.2001,
      "step": 514
    },
    {
      "epoch": 0.23206038075926552,
      "grad_norm": 0.9917793017218205,
      "learning_rate": 4.56038581284865e-06,
      "loss": 0.2012,
      "step": 515
    },
    {
      "epoch": 0.23251098344035148,
      "grad_norm": 1.0321116954854799,
      "learning_rate": 4.558273362980406e-06,
      "loss": 0.2173,
      "step": 516
    },
    {
      "epoch": 0.23296158612143741,
      "grad_norm": 1.0540135978614726,
      "learning_rate": 4.556156341582884e-06,
      "loss": 0.2176,
      "step": 517
    },
    {
      "epoch": 0.23341218880252337,
      "grad_norm": 1.026187489177222,
      "learning_rate": 4.5540347533580935e-06,
      "loss": 0.1982,
      "step": 518
    },
    {
      "epoch": 0.23386279148360933,
      "grad_norm": 1.0683916408326868,
      "learning_rate": 4.551908603018191e-06,
      "loss": 0.1916,
      "step": 519
    },
    {
      "epoch": 0.2343133941646953,
      "grad_norm": 0.9800651964726533,
      "learning_rate": 4.549777895285464e-06,
      "loss": 0.1928,
      "step": 520
    },
    {
      "epoch": 0.23476399684578123,
      "grad_norm": 1.0102074038951585,
      "learning_rate": 4.547642634892321e-06,
      "loss": 0.2039,
      "step": 521
    },
    {
      "epoch": 0.23521459952686719,
      "grad_norm": 1.1222993821068836,
      "learning_rate": 4.545502826581284e-06,
      "loss": 0.2234,
      "step": 522
    },
    {
      "epoch": 0.23566520220795314,
      "grad_norm": 1.084335390939449,
      "learning_rate": 4.543358475104975e-06,
      "loss": 0.2147,
      "step": 523
    },
    {
      "epoch": 0.23611580488903908,
      "grad_norm": 1.1110169688938027,
      "learning_rate": 4.541209585226109e-06,
      "loss": 0.1997,
      "step": 524
    },
    {
      "epoch": 0.23656640757012504,
      "grad_norm": 0.9454783484088987,
      "learning_rate": 4.539056161717477e-06,
      "loss": 0.1906,
      "step": 525
    },
    {
      "epoch": 0.237017010251211,
      "grad_norm": 1.0979245101280635,
      "learning_rate": 4.536898209361942e-06,
      "loss": 0.2097,
      "step": 526
    },
    {
      "epoch": 0.23746761293229696,
      "grad_norm": 1.0218511577120122,
      "learning_rate": 4.5347357329524254e-06,
      "loss": 0.1942,
      "step": 527
    },
    {
      "epoch": 0.2379182156133829,
      "grad_norm": 1.1530219699070952,
      "learning_rate": 4.532568737291898e-06,
      "loss": 0.2013,
      "step": 528
    },
    {
      "epoch": 0.23836881829446885,
      "grad_norm": 1.0873274217971285,
      "learning_rate": 4.530397227193365e-06,
      "loss": 0.207,
      "step": 529
    },
    {
      "epoch": 0.2388194209755548,
      "grad_norm": 1.0476737638086648,
      "learning_rate": 4.528221207479862e-06,
      "loss": 0.1988,
      "step": 530
    },
    {
      "epoch": 0.23927002365664077,
      "grad_norm": 1.096263795148243,
      "learning_rate": 4.5260406829844364e-06,
      "loss": 0.2012,
      "step": 531
    },
    {
      "epoch": 0.2397206263377267,
      "grad_norm": 1.030880422031526,
      "learning_rate": 4.523855658550146e-06,
      "loss": 0.2027,
      "step": 532
    },
    {
      "epoch": 0.24017122901881266,
      "grad_norm": 1.0067117254618483,
      "learning_rate": 4.521666139030039e-06,
      "loss": 0.1951,
      "step": 533
    },
    {
      "epoch": 0.24062183169989862,
      "grad_norm": 1.0747100498903996,
      "learning_rate": 4.51947212928715e-06,
      "loss": 0.1859,
      "step": 534
    },
    {
      "epoch": 0.24107243438098458,
      "grad_norm": 1.2309094629093367,
      "learning_rate": 4.5172736341944845e-06,
      "loss": 0.222,
      "step": 535
    },
    {
      "epoch": 0.2415230370620705,
      "grad_norm": 1.103494575925489,
      "learning_rate": 4.515070658635013e-06,
      "loss": 0.2129,
      "step": 536
    },
    {
      "epoch": 0.24197363974315647,
      "grad_norm": 1.0397715812091397,
      "learning_rate": 4.512863207501654e-06,
      "loss": 0.1839,
      "step": 537
    },
    {
      "epoch": 0.24242424242424243,
      "grad_norm": 1.0560302158424433,
      "learning_rate": 4.510651285697269e-06,
      "loss": 0.1952,
      "step": 538
    },
    {
      "epoch": 0.24287484510532836,
      "grad_norm": 1.1120154366274742,
      "learning_rate": 4.5084348981346495e-06,
      "loss": 0.2071,
      "step": 539
    },
    {
      "epoch": 0.24332544778641432,
      "grad_norm": 1.0545051119809872,
      "learning_rate": 4.506214049736502e-06,
      "loss": 0.2219,
      "step": 540
    },
    {
      "epoch": 0.24377605046750028,
      "grad_norm": 1.0352049466055129,
      "learning_rate": 4.503988745435443e-06,
      "loss": 0.1836,
      "step": 541
    },
    {
      "epoch": 0.24422665314858624,
      "grad_norm": 1.0809072576838936,
      "learning_rate": 4.5017589901739885e-06,
      "loss": 0.2045,
      "step": 542
    },
    {
      "epoch": 0.24467725582967217,
      "grad_norm": 1.110072290591729,
      "learning_rate": 4.499524788904537e-06,
      "loss": 0.2059,
      "step": 543
    },
    {
      "epoch": 0.24512785851075813,
      "grad_norm": 0.9665808924396558,
      "learning_rate": 4.497286146589361e-06,
      "loss": 0.2011,
      "step": 544
    },
    {
      "epoch": 0.2455784611918441,
      "grad_norm": 1.0639169166435112,
      "learning_rate": 4.4950430682005995e-06,
      "loss": 0.1952,
      "step": 545
    },
    {
      "epoch": 0.24602906387293005,
      "grad_norm": 1.1497485140437143,
      "learning_rate": 4.492795558720242e-06,
      "loss": 0.2067,
      "step": 546
    },
    {
      "epoch": 0.24647966655401599,
      "grad_norm": 1.0118960522373834,
      "learning_rate": 4.490543623140123e-06,
      "loss": 0.1972,
      "step": 547
    },
    {
      "epoch": 0.24693026923510195,
      "grad_norm": 1.100344333423072,
      "learning_rate": 4.488287266461904e-06,
      "loss": 0.2132,
      "step": 548
    },
    {
      "epoch": 0.2473808719161879,
      "grad_norm": 1.059434492217794,
      "learning_rate": 4.486026493697067e-06,
      "loss": 0.1949,
      "step": 549
    },
    {
      "epoch": 0.24783147459727387,
      "grad_norm": 1.1281625090194913,
      "learning_rate": 4.483761309866902e-06,
      "loss": 0.2168,
      "step": 550
    },
    {
      "epoch": 0.2482820772783598,
      "grad_norm": 1.0960219014600239,
      "learning_rate": 4.481491720002499e-06,
      "loss": 0.1988,
      "step": 551
    },
    {
      "epoch": 0.24873267995944576,
      "grad_norm": 1.0403322611009544,
      "learning_rate": 4.479217729144731e-06,
      "loss": 0.2003,
      "step": 552
    },
    {
      "epoch": 0.24918328264053172,
      "grad_norm": 1.0318746166370159,
      "learning_rate": 4.476939342344246e-06,
      "loss": 0.1929,
      "step": 553
    },
    {
      "epoch": 0.24963388532161768,
      "grad_norm": 1.0679148861170464,
      "learning_rate": 4.474656564661458e-06,
      "loss": 0.183,
      "step": 554
    },
    {
      "epoch": 0.2500844880027036,
      "grad_norm": 1.0258442730168609,
      "learning_rate": 4.472369401166531e-06,
      "loss": 0.1892,
      "step": 555
    },
    {
      "epoch": 0.25053509068378954,
      "grad_norm": 1.065168544541267,
      "learning_rate": 4.47007785693937e-06,
      "loss": 0.1855,
      "step": 556
    },
    {
      "epoch": 0.25098569336487553,
      "grad_norm": 1.0375402257867379,
      "learning_rate": 4.467781937069611e-06,
      "loss": 0.1934,
      "step": 557
    },
    {
      "epoch": 0.25143629604596146,
      "grad_norm": 1.0182166840713762,
      "learning_rate": 4.465481646656608e-06,
      "loss": 0.2082,
      "step": 558
    },
    {
      "epoch": 0.25188689872704745,
      "grad_norm": 0.9643240424365269,
      "learning_rate": 4.463176990809423e-06,
      "loss": 0.1913,
      "step": 559
    },
    {
      "epoch": 0.2523375014081334,
      "grad_norm": 1.049581595009854,
      "learning_rate": 4.460867974646814e-06,
      "loss": 0.2022,
      "step": 560
    },
    {
      "epoch": 0.2527881040892193,
      "grad_norm": 0.9918991445137725,
      "learning_rate": 4.45855460329722e-06,
      "loss": 0.1845,
      "step": 561
    },
    {
      "epoch": 0.2532387067703053,
      "grad_norm": 1.0873121965600965,
      "learning_rate": 4.45623688189876e-06,
      "loss": 0.2085,
      "step": 562
    },
    {
      "epoch": 0.25368930945139123,
      "grad_norm": 0.987846557405897,
      "learning_rate": 4.453914815599206e-06,
      "loss": 0.1953,
      "step": 563
    },
    {
      "epoch": 0.25413991213247716,
      "grad_norm": 1.0298405968391158,
      "learning_rate": 4.451588409555988e-06,
      "loss": 0.1962,
      "step": 564
    },
    {
      "epoch": 0.25459051481356315,
      "grad_norm": 1.0326677560841386,
      "learning_rate": 4.4492576689361705e-06,
      "loss": 0.1948,
      "step": 565
    },
    {
      "epoch": 0.2550411174946491,
      "grad_norm": 1.0519711817977115,
      "learning_rate": 4.446922598916445e-06,
      "loss": 0.204,
      "step": 566
    },
    {
      "epoch": 0.25549172017573507,
      "grad_norm": 1.027262799140467,
      "learning_rate": 4.444583204683123e-06,
      "loss": 0.1938,
      "step": 567
    },
    {
      "epoch": 0.255942322856821,
      "grad_norm": 1.0965809840354055,
      "learning_rate": 4.4422394914321145e-06,
      "loss": 0.2047,
      "step": 568
    },
    {
      "epoch": 0.25639292553790693,
      "grad_norm": 1.049829610661597,
      "learning_rate": 4.439891464368927e-06,
      "loss": 0.2064,
      "step": 569
    },
    {
      "epoch": 0.2568435282189929,
      "grad_norm": 0.9765801119075459,
      "learning_rate": 4.437539128708647e-06,
      "loss": 0.183,
      "step": 570
    },
    {
      "epoch": 0.25729413090007885,
      "grad_norm": 1.058406240013727,
      "learning_rate": 4.435182489675931e-06,
      "loss": 0.2043,
      "step": 571
    },
    {
      "epoch": 0.2577447335811648,
      "grad_norm": 0.9556853971932254,
      "learning_rate": 4.432821552504994e-06,
      "loss": 0.1949,
      "step": 572
    },
    {
      "epoch": 0.2581953362622508,
      "grad_norm": 1.0499296701467336,
      "learning_rate": 4.430456322439596e-06,
      "loss": 0.2013,
      "step": 573
    },
    {
      "epoch": 0.2586459389433367,
      "grad_norm": 1.0824220937552163,
      "learning_rate": 4.4280868047330325e-06,
      "loss": 0.2068,
      "step": 574
    },
    {
      "epoch": 0.25909654162442264,
      "grad_norm": 1.060056337801298,
      "learning_rate": 4.425713004648123e-06,
      "loss": 0.2064,
      "step": 575
    },
    {
      "epoch": 0.2595471443055086,
      "grad_norm": 1.0606451489538105,
      "learning_rate": 4.423334927457198e-06,
      "loss": 0.2046,
      "step": 576
    },
    {
      "epoch": 0.25999774698659456,
      "grad_norm": 1.091280452882288,
      "learning_rate": 4.420952578442086e-06,
      "loss": 0.2156,
      "step": 577
    },
    {
      "epoch": 0.26044834966768055,
      "grad_norm": 0.9696606076862494,
      "learning_rate": 4.4185659628941054e-06,
      "loss": 0.1856,
      "step": 578
    },
    {
      "epoch": 0.2608989523487665,
      "grad_norm": 1.0364418339411665,
      "learning_rate": 4.416175086114049e-06,
      "loss": 0.2086,
      "step": 579
    },
    {
      "epoch": 0.2613495550298524,
      "grad_norm": 1.0784544410291654,
      "learning_rate": 4.4137799534121785e-06,
      "loss": 0.1933,
      "step": 580
    },
    {
      "epoch": 0.2618001577109384,
      "grad_norm": 1.01841478080086,
      "learning_rate": 4.4113805701082e-06,
      "loss": 0.1914,
      "step": 581
    },
    {
      "epoch": 0.26225076039202433,
      "grad_norm": 1.165117484518737,
      "learning_rate": 4.408976941531269e-06,
      "loss": 0.2226,
      "step": 582
    },
    {
      "epoch": 0.26270136307311026,
      "grad_norm": 0.952663941356781,
      "learning_rate": 4.406569073019965e-06,
      "loss": 0.2004,
      "step": 583
    },
    {
      "epoch": 0.26315196575419625,
      "grad_norm": 1.0687364462500708,
      "learning_rate": 4.404156969922284e-06,
      "loss": 0.2009,
      "step": 584
    },
    {
      "epoch": 0.2636025684352822,
      "grad_norm": 0.9840008293163256,
      "learning_rate": 4.401740637595633e-06,
      "loss": 0.1878,
      "step": 585
    },
    {
      "epoch": 0.26405317111636817,
      "grad_norm": 0.9605772910240837,
      "learning_rate": 4.3993200814068035e-06,
      "loss": 0.1893,
      "step": 586
    },
    {
      "epoch": 0.2645037737974541,
      "grad_norm": 0.9941565744960191,
      "learning_rate": 4.396895306731978e-06,
      "loss": 0.193,
      "step": 587
    },
    {
      "epoch": 0.26495437647854003,
      "grad_norm": 0.9451043951281098,
      "learning_rate": 4.394466318956701e-06,
      "loss": 0.1861,
      "step": 588
    },
    {
      "epoch": 0.265404979159626,
      "grad_norm": 0.9357557601128812,
      "learning_rate": 4.392033123475876e-06,
      "loss": 0.1928,
      "step": 589
    },
    {
      "epoch": 0.26585558184071195,
      "grad_norm": 1.211949872669835,
      "learning_rate": 4.389595725693756e-06,
      "loss": 0.1953,
      "step": 590
    },
    {
      "epoch": 0.2663061845217979,
      "grad_norm": 0.9744058315459057,
      "learning_rate": 4.387154131023924e-06,
      "loss": 0.1953,
      "step": 591
    },
    {
      "epoch": 0.26675678720288387,
      "grad_norm": 0.9240075427887562,
      "learning_rate": 4.384708344889285e-06,
      "loss": 0.1762,
      "step": 592
    },
    {
      "epoch": 0.2672073898839698,
      "grad_norm": 0.9646315497774892,
      "learning_rate": 4.382258372722054e-06,
      "loss": 0.1925,
      "step": 593
    },
    {
      "epoch": 0.26765799256505574,
      "grad_norm": 1.1246869317376789,
      "learning_rate": 4.379804219963742e-06,
      "loss": 0.2179,
      "step": 594
    },
    {
      "epoch": 0.2681085952461417,
      "grad_norm": 0.9989771573424515,
      "learning_rate": 4.377345892065149e-06,
      "loss": 0.1947,
      "step": 595
    },
    {
      "epoch": 0.26855919792722766,
      "grad_norm": 1.0192448988756184,
      "learning_rate": 4.374883394486343e-06,
      "loss": 0.183,
      "step": 596
    },
    {
      "epoch": 0.26900980060831364,
      "grad_norm": 1.1840938286800815,
      "learning_rate": 4.3724167326966575e-06,
      "loss": 0.2031,
      "step": 597
    },
    {
      "epoch": 0.2694604032893996,
      "grad_norm": 0.9756787429230731,
      "learning_rate": 4.3699459121746726e-06,
      "loss": 0.206,
      "step": 598
    },
    {
      "epoch": 0.2699110059704855,
      "grad_norm": 1.1107694091005171,
      "learning_rate": 4.367470938408204e-06,
      "loss": 0.1869,
      "step": 599
    },
    {
      "epoch": 0.2703616086515715,
      "grad_norm": 0.9232888391273938,
      "learning_rate": 4.364991816894296e-06,
      "loss": 0.1788,
      "step": 600
    },
    {
      "epoch": 0.2708122113326574,
      "grad_norm": 1.0025886336195018,
      "learning_rate": 4.362508553139203e-06,
      "loss": 0.2023,
      "step": 601
    },
    {
      "epoch": 0.27126281401374336,
      "grad_norm": 1.1433010440724132,
      "learning_rate": 4.360021152658378e-06,
      "loss": 0.1994,
      "step": 602
    },
    {
      "epoch": 0.27171341669482935,
      "grad_norm": 0.9895897645038598,
      "learning_rate": 4.357529620976463e-06,
      "loss": 0.1883,
      "step": 603
    },
    {
      "epoch": 0.2721640193759153,
      "grad_norm": 1.0296556364889053,
      "learning_rate": 4.355033963627277e-06,
      "loss": 0.2105,
      "step": 604
    },
    {
      "epoch": 0.27261462205700127,
      "grad_norm": 1.1099119086432157,
      "learning_rate": 4.352534186153802e-06,
      "loss": 0.1953,
      "step": 605
    },
    {
      "epoch": 0.2730652247380872,
      "grad_norm": 1.0549408039081614,
      "learning_rate": 4.3500302941081685e-06,
      "loss": 0.1977,
      "step": 606
    },
    {
      "epoch": 0.27351582741917313,
      "grad_norm": 1.039314334394156,
      "learning_rate": 4.3475222930516484e-06,
      "loss": 0.2092,
      "step": 607
    },
    {
      "epoch": 0.2739664301002591,
      "grad_norm": 1.1203701706687121,
      "learning_rate": 4.345010188554638e-06,
      "loss": 0.1932,
      "step": 608
    },
    {
      "epoch": 0.27441703278134505,
      "grad_norm": 1.0628458411287045,
      "learning_rate": 4.34249398619665e-06,
      "loss": 0.1841,
      "step": 609
    },
    {
      "epoch": 0.274867635462431,
      "grad_norm": 1.1398229296350408,
      "learning_rate": 4.339973691566297e-06,
      "loss": 0.2031,
      "step": 610
    },
    {
      "epoch": 0.27531823814351697,
      "grad_norm": 0.9514446506895067,
      "learning_rate": 4.337449310261279e-06,
      "loss": 0.1942,
      "step": 611
    },
    {
      "epoch": 0.2757688408246029,
      "grad_norm": 1.061471271498229,
      "learning_rate": 4.334920847888376e-06,
      "loss": 0.217,
      "step": 612
    },
    {
      "epoch": 0.27621944350568883,
      "grad_norm": 1.0214906979222622,
      "learning_rate": 4.332388310063431e-06,
      "loss": 0.1981,
      "step": 613
    },
    {
      "epoch": 0.2766700461867748,
      "grad_norm": 1.0346404733819128,
      "learning_rate": 4.329851702411339e-06,
      "loss": 0.1946,
      "step": 614
    },
    {
      "epoch": 0.27712064886786075,
      "grad_norm": 1.002298040924044,
      "learning_rate": 4.327311030566033e-06,
      "loss": 0.1962,
      "step": 615
    },
    {
      "epoch": 0.27757125154894674,
      "grad_norm": 1.0160693312542004,
      "learning_rate": 4.324766300170473e-06,
      "loss": 0.2045,
      "step": 616
    },
    {
      "epoch": 0.2780218542300327,
      "grad_norm": 1.0621190326896313,
      "learning_rate": 4.322217516876635e-06,
      "loss": 0.1997,
      "step": 617
    },
    {
      "epoch": 0.2784724569111186,
      "grad_norm": 1.044710760150364,
      "learning_rate": 4.3196646863454975e-06,
      "loss": 0.1975,
      "step": 618
    },
    {
      "epoch": 0.2789230595922046,
      "grad_norm": 1.0565566812202085,
      "learning_rate": 4.317107814247022e-06,
      "loss": 0.2179,
      "step": 619
    },
    {
      "epoch": 0.2793736622732905,
      "grad_norm": 1.0822662595100792,
      "learning_rate": 4.314546906260156e-06,
      "loss": 0.2171,
      "step": 620
    },
    {
      "epoch": 0.27982426495437646,
      "grad_norm": 0.9847692683483855,
      "learning_rate": 4.3119819680728e-06,
      "loss": 0.1789,
      "step": 621
    },
    {
      "epoch": 0.28027486763546244,
      "grad_norm": 1.0377605692374061,
      "learning_rate": 4.3094130053818164e-06,
      "loss": 0.2069,
      "step": 622
    },
    {
      "epoch": 0.2807254703165484,
      "grad_norm": 1.064720287531545,
      "learning_rate": 4.306840023892998e-06,
      "loss": 0.1854,
      "step": 623
    },
    {
      "epoch": 0.28117607299763436,
      "grad_norm": 1.033767676273761,
      "learning_rate": 4.304263029321069e-06,
      "loss": 0.1931,
      "step": 624
    },
    {
      "epoch": 0.2816266756787203,
      "grad_norm": 0.8815106297337543,
      "learning_rate": 4.301682027389663e-06,
      "loss": 0.1676,
      "step": 625
    },
    {
      "epoch": 0.2820772783598062,
      "grad_norm": 1.014281098690602,
      "learning_rate": 4.299097023831318e-06,
      "loss": 0.1952,
      "step": 626
    },
    {
      "epoch": 0.2825278810408922,
      "grad_norm": 1.0085581349510135,
      "learning_rate": 4.2965080243874555e-06,
      "loss": 0.1873,
      "step": 627
    },
    {
      "epoch": 0.28297848372197815,
      "grad_norm": 1.030623931702184,
      "learning_rate": 4.293915034808376e-06,
      "loss": 0.1922,
      "step": 628
    },
    {
      "epoch": 0.2834290864030641,
      "grad_norm": 1.0188262260920022,
      "learning_rate": 4.29131806085324e-06,
      "loss": 0.1966,
      "step": 629
    },
    {
      "epoch": 0.28387968908415007,
      "grad_norm": 1.0037022727298737,
      "learning_rate": 4.288717108290056e-06,
      "loss": 0.2009,
      "step": 630
    },
    {
      "epoch": 0.284330291765236,
      "grad_norm": 1.0271567668441708,
      "learning_rate": 4.2861121828956745e-06,
      "loss": 0.2024,
      "step": 631
    },
    {
      "epoch": 0.28478089444632193,
      "grad_norm": 1.012705939797216,
      "learning_rate": 4.283503290455765e-06,
      "loss": 0.1863,
      "step": 632
    },
    {
      "epoch": 0.2852314971274079,
      "grad_norm": 1.0628715517829384,
      "learning_rate": 4.28089043676481e-06,
      "loss": 0.2175,
      "step": 633
    },
    {
      "epoch": 0.28568209980849385,
      "grad_norm": 1.0106381943649927,
      "learning_rate": 4.27827362762609e-06,
      "loss": 0.2087,
      "step": 634
    },
    {
      "epoch": 0.28613270248957984,
      "grad_norm": 1.0630146592405614,
      "learning_rate": 4.275652868851669e-06,
      "loss": 0.1938,
      "step": 635
    },
    {
      "epoch": 0.28658330517066577,
      "grad_norm": 1.0499418639693274,
      "learning_rate": 4.2730281662623866e-06,
      "loss": 0.1864,
      "step": 636
    },
    {
      "epoch": 0.2870339078517517,
      "grad_norm": 1.0016998697053148,
      "learning_rate": 4.270399525687839e-06,
      "loss": 0.1993,
      "step": 637
    },
    {
      "epoch": 0.2874845105328377,
      "grad_norm": 1.0127517858972737,
      "learning_rate": 4.267766952966369e-06,
      "loss": 0.1863,
      "step": 638
    },
    {
      "epoch": 0.2879351132139236,
      "grad_norm": 1.0939363436908969,
      "learning_rate": 4.265130453945056e-06,
      "loss": 0.1922,
      "step": 639
    },
    {
      "epoch": 0.28838571589500955,
      "grad_norm": 1.0282674433305248,
      "learning_rate": 4.262490034479697e-06,
      "loss": 0.1974,
      "step": 640
    },
    {
      "epoch": 0.28883631857609554,
      "grad_norm": 1.0713994069774757,
      "learning_rate": 4.259845700434797e-06,
      "loss": 0.2178,
      "step": 641
    },
    {
      "epoch": 0.2892869212571815,
      "grad_norm": 1.0095365158524388,
      "learning_rate": 4.257197457683556e-06,
      "loss": 0.1984,
      "step": 642
    },
    {
      "epoch": 0.2897375239382674,
      "grad_norm": 0.9517358327424785,
      "learning_rate": 4.254545312107854e-06,
      "loss": 0.1995,
      "step": 643
    },
    {
      "epoch": 0.2901881266193534,
      "grad_norm": 0.9988737112663147,
      "learning_rate": 4.251889269598241e-06,
      "loss": 0.201,
      "step": 644
    },
    {
      "epoch": 0.2906387293004393,
      "grad_norm": 0.9491077816555287,
      "learning_rate": 4.249229336053924e-06,
      "loss": 0.1948,
      "step": 645
    },
    {
      "epoch": 0.2910893319815253,
      "grad_norm": 0.9177074871325192,
      "learning_rate": 4.2465655173827465e-06,
      "loss": 0.1875,
      "step": 646
    },
    {
      "epoch": 0.29153993466261124,
      "grad_norm": 1.0676678310448093,
      "learning_rate": 4.243897819501187e-06,
      "loss": 0.2117,
      "step": 647
    },
    {
      "epoch": 0.2919905373436972,
      "grad_norm": 0.9838253108409748,
      "learning_rate": 4.241226248334335e-06,
      "loss": 0.1998,
      "step": 648
    },
    {
      "epoch": 0.29244114002478316,
      "grad_norm": 0.9928941143354909,
      "learning_rate": 4.238550809815889e-06,
      "loss": 0.1895,
      "step": 649
    },
    {
      "epoch": 0.2928917427058691,
      "grad_norm": 1.0128875324516162,
      "learning_rate": 4.23587150988813e-06,
      "loss": 0.2203,
      "step": 650
    },
    {
      "epoch": 0.29334234538695503,
      "grad_norm": 1.028303567499585,
      "learning_rate": 4.233188354501921e-06,
      "loss": 0.206,
      "step": 651
    },
    {
      "epoch": 0.293792948068041,
      "grad_norm": 1.1145425080945328,
      "learning_rate": 4.230501349616683e-06,
      "loss": 0.2088,
      "step": 652
    },
    {
      "epoch": 0.29424355074912695,
      "grad_norm": 1.1234390517892725,
      "learning_rate": 4.227810501200393e-06,
      "loss": 0.1889,
      "step": 653
    },
    {
      "epoch": 0.29469415343021294,
      "grad_norm": 0.9793683475808044,
      "learning_rate": 4.225115815229559e-06,
      "loss": 0.1878,
      "step": 654
    },
    {
      "epoch": 0.29514475611129887,
      "grad_norm": 1.0115361523498685,
      "learning_rate": 4.222417297689217e-06,
      "loss": 0.1919,
      "step": 655
    },
    {
      "epoch": 0.2955953587923848,
      "grad_norm": 1.0416286762299183,
      "learning_rate": 4.219714954572909e-06,
      "loss": 0.1994,
      "step": 656
    },
    {
      "epoch": 0.2960459614734708,
      "grad_norm": 0.9552050906912731,
      "learning_rate": 4.217008791882678e-06,
      "loss": 0.1893,
      "step": 657
    },
    {
      "epoch": 0.2964965641545567,
      "grad_norm": 1.0577253087294156,
      "learning_rate": 4.214298815629046e-06,
      "loss": 0.1884,
      "step": 658
    },
    {
      "epoch": 0.29694716683564265,
      "grad_norm": 0.9647606522962644,
      "learning_rate": 4.211585031831007e-06,
      "loss": 0.1953,
      "step": 659
    },
    {
      "epoch": 0.29739776951672864,
      "grad_norm": 0.9562447824803063,
      "learning_rate": 4.208867446516015e-06,
      "loss": 0.1855,
      "step": 660
    },
    {
      "epoch": 0.29784837219781457,
      "grad_norm": 0.972853425024187,
      "learning_rate": 4.206146065719963e-06,
      "loss": 0.1819,
      "step": 661
    },
    {
      "epoch": 0.2982989748789005,
      "grad_norm": 0.9787638447089178,
      "learning_rate": 4.203420895487175e-06,
      "loss": 0.2029,
      "step": 662
    },
    {
      "epoch": 0.2987495775599865,
      "grad_norm": 0.9759659661884763,
      "learning_rate": 4.200691941870392e-06,
      "loss": 0.2023,
      "step": 663
    },
    {
      "epoch": 0.2992001802410724,
      "grad_norm": 0.9943286482597673,
      "learning_rate": 4.197959210930759e-06,
      "loss": 0.1978,
      "step": 664
    },
    {
      "epoch": 0.2996507829221584,
      "grad_norm": 0.9801408645055518,
      "learning_rate": 4.195222708737809e-06,
      "loss": 0.2013,
      "step": 665
    },
    {
      "epoch": 0.30010138560324434,
      "grad_norm": 1.0539478116036578,
      "learning_rate": 4.192482441369451e-06,
      "loss": 0.2035,
      "step": 666
    },
    {
      "epoch": 0.3005519882843303,
      "grad_norm": 1.0997940239624653,
      "learning_rate": 4.189738414911959e-06,
      "loss": 0.22,
      "step": 667
    },
    {
      "epoch": 0.30100259096541626,
      "grad_norm": 0.9853856855339789,
      "learning_rate": 4.186990635459954e-06,
      "loss": 0.2016,
      "step": 668
    },
    {
      "epoch": 0.3014531936465022,
      "grad_norm": 1.1197830097140093,
      "learning_rate": 4.184239109116393e-06,
      "loss": 0.2102,
      "step": 669
    },
    {
      "epoch": 0.3019037963275881,
      "grad_norm": 1.0300352624715958,
      "learning_rate": 4.181483841992556e-06,
      "loss": 0.2018,
      "step": 670
    },
    {
      "epoch": 0.3023543990086741,
      "grad_norm": 1.0137327368224471,
      "learning_rate": 4.178724840208029e-06,
      "loss": 0.1986,
      "step": 671
    },
    {
      "epoch": 0.30280500168976004,
      "grad_norm": 0.9204806350681561,
      "learning_rate": 4.175962109890697e-06,
      "loss": 0.1813,
      "step": 672
    },
    {
      "epoch": 0.30325560437084603,
      "grad_norm": 1.0673745916166968,
      "learning_rate": 4.1731956571767215e-06,
      "loss": 0.1998,
      "step": 673
    },
    {
      "epoch": 0.30370620705193196,
      "grad_norm": 1.0160656530915944,
      "learning_rate": 4.170425488210534e-06,
      "loss": 0.207,
      "step": 674
    },
    {
      "epoch": 0.3041568097330179,
      "grad_norm": 1.054838095733939,
      "learning_rate": 4.167651609144822e-06,
      "loss": 0.1942,
      "step": 675
    },
    {
      "epoch": 0.3046074124141039,
      "grad_norm": 1.077041886917572,
      "learning_rate": 4.164874026140511e-06,
      "loss": 0.2014,
      "step": 676
    },
    {
      "epoch": 0.3050580150951898,
      "grad_norm": 1.0857669925430111,
      "learning_rate": 4.1620927453667515e-06,
      "loss": 0.2099,
      "step": 677
    },
    {
      "epoch": 0.30550861777627575,
      "grad_norm": 1.0608069365987411,
      "learning_rate": 4.159307773000909e-06,
      "loss": 0.1873,
      "step": 678
    },
    {
      "epoch": 0.30595922045736174,
      "grad_norm": 0.957307574538134,
      "learning_rate": 4.15651911522855e-06,
      "loss": 0.1796,
      "step": 679
    },
    {
      "epoch": 0.30640982313844767,
      "grad_norm": 1.0234422261068838,
      "learning_rate": 4.153726778243422e-06,
      "loss": 0.1903,
      "step": 680
    },
    {
      "epoch": 0.3068604258195336,
      "grad_norm": 1.03627065798202,
      "learning_rate": 4.150930768247449e-06,
      "loss": 0.1859,
      "step": 681
    },
    {
      "epoch": 0.3073110285006196,
      "grad_norm": 1.0456496616827005,
      "learning_rate": 4.148131091450709e-06,
      "loss": 0.1975,
      "step": 682
    },
    {
      "epoch": 0.3077616311817055,
      "grad_norm": 1.0630130936439768,
      "learning_rate": 4.145327754071427e-06,
      "loss": 0.2077,
      "step": 683
    },
    {
      "epoch": 0.3082122338627915,
      "grad_norm": 1.065688708892473,
      "learning_rate": 4.142520762335957e-06,
      "loss": 0.1849,
      "step": 684
    },
    {
      "epoch": 0.30866283654387744,
      "grad_norm": 1.1263818513906516,
      "learning_rate": 4.13971012247877e-06,
      "loss": 0.2117,
      "step": 685
    },
    {
      "epoch": 0.30911343922496337,
      "grad_norm": 1.0666870069787027,
      "learning_rate": 4.136895840742437e-06,
      "loss": 0.2082,
      "step": 686
    },
    {
      "epoch": 0.30956404190604936,
      "grad_norm": 1.1195428726866035,
      "learning_rate": 4.134077923377622e-06,
      "loss": 0.2128,
      "step": 687
    },
    {
      "epoch": 0.3100146445871353,
      "grad_norm": 0.9706588710375661,
      "learning_rate": 4.131256376643062e-06,
      "loss": 0.194,
      "step": 688
    },
    {
      "epoch": 0.3104652472682212,
      "grad_norm": 0.9841129879832418,
      "learning_rate": 4.128431206805556e-06,
      "loss": 0.191,
      "step": 689
    },
    {
      "epoch": 0.3109158499493072,
      "grad_norm": 0.964224745997007,
      "learning_rate": 4.125602420139947e-06,
      "loss": 0.1909,
      "step": 690
    },
    {
      "epoch": 0.31136645263039314,
      "grad_norm": 0.9875736538745279,
      "learning_rate": 4.122770022929114e-06,
      "loss": 0.1862,
      "step": 691
    },
    {
      "epoch": 0.31181705531147913,
      "grad_norm": 1.0739578378289165,
      "learning_rate": 4.119934021463956e-06,
      "loss": 0.2165,
      "step": 692
    },
    {
      "epoch": 0.31226765799256506,
      "grad_norm": 0.9672865623771372,
      "learning_rate": 4.117094422043374e-06,
      "loss": 0.1858,
      "step": 693
    },
    {
      "epoch": 0.312718260673651,
      "grad_norm": 1.0671701635171398,
      "learning_rate": 4.114251230974263e-06,
      "loss": 0.2117,
      "step": 694
    },
    {
      "epoch": 0.313168863354737,
      "grad_norm": 1.0091868850822507,
      "learning_rate": 4.1114044545714935e-06,
      "loss": 0.2034,
      "step": 695
    },
    {
      "epoch": 0.3136194660358229,
      "grad_norm": 1.025473722449347,
      "learning_rate": 4.1085540991579e-06,
      "loss": 0.2114,
      "step": 696
    },
    {
      "epoch": 0.31407006871690885,
      "grad_norm": 0.9791997644517443,
      "learning_rate": 4.105700171064267e-06,
      "loss": 0.1877,
      "step": 697
    },
    {
      "epoch": 0.31452067139799483,
      "grad_norm": 1.0485648239190921,
      "learning_rate": 4.102842676629313e-06,
      "loss": 0.2098,
      "step": 698
    },
    {
      "epoch": 0.31497127407908077,
      "grad_norm": 0.9932204868367267,
      "learning_rate": 4.0999816221996755e-06,
      "loss": 0.2016,
      "step": 699
    },
    {
      "epoch": 0.3154218767601667,
      "grad_norm": 1.0052350903793839,
      "learning_rate": 4.097117014129903e-06,
      "loss": 0.2031,
      "step": 700
    },
    {
      "epoch": 0.3158724794412527,
      "grad_norm": 1.0588148231889312,
      "learning_rate": 4.094248858782436e-06,
      "loss": 0.2059,
      "step": 701
    },
    {
      "epoch": 0.3163230821223386,
      "grad_norm": 0.9062525913899364,
      "learning_rate": 4.091377162527592e-06,
      "loss": 0.1751,
      "step": 702
    },
    {
      "epoch": 0.3167736848034246,
      "grad_norm": 1.033952410424911,
      "learning_rate": 4.088501931743551e-06,
      "loss": 0.1969,
      "step": 703
    },
    {
      "epoch": 0.31722428748451054,
      "grad_norm": 1.0173227691023725,
      "learning_rate": 4.08562317281635e-06,
      "loss": 0.1939,
      "step": 704
    },
    {
      "epoch": 0.31767489016559647,
      "grad_norm": 0.9377764057356622,
      "learning_rate": 4.082740892139856e-06,
      "loss": 0.1808,
      "step": 705
    },
    {
      "epoch": 0.31812549284668246,
      "grad_norm": 1.0885942813084413,
      "learning_rate": 4.07985509611576e-06,
      "loss": 0.1933,
      "step": 706
    },
    {
      "epoch": 0.3185760955277684,
      "grad_norm": 1.0460613454763035,
      "learning_rate": 4.076965791153562e-06,
      "loss": 0.1853,
      "step": 707
    },
    {
      "epoch": 0.3190266982088543,
      "grad_norm": 1.0112840919837836,
      "learning_rate": 4.074072983670555e-06,
      "loss": 0.1992,
      "step": 708
    },
    {
      "epoch": 0.3194773008899403,
      "grad_norm": 1.0035685646513803,
      "learning_rate": 4.071176680091809e-06,
      "loss": 0.1914,
      "step": 709
    },
    {
      "epoch": 0.31992790357102624,
      "grad_norm": 0.9455942563033818,
      "learning_rate": 4.068276886850162e-06,
      "loss": 0.1977,
      "step": 710
    },
    {
      "epoch": 0.3203785062521122,
      "grad_norm": 1.0248608488543998,
      "learning_rate": 4.065373610386201e-06,
      "loss": 0.2098,
      "step": 711
    },
    {
      "epoch": 0.32082910893319816,
      "grad_norm": 1.0833442868468053,
      "learning_rate": 4.06246685714825e-06,
      "loss": 0.1971,
      "step": 712
    },
    {
      "epoch": 0.3212797116142841,
      "grad_norm": 1.0376988954502842,
      "learning_rate": 4.059556633592356e-06,
      "loss": 0.2036,
      "step": 713
    },
    {
      "epoch": 0.3217303142953701,
      "grad_norm": 0.9076161254666845,
      "learning_rate": 4.056642946182271e-06,
      "loss": 0.181,
      "step": 714
    },
    {
      "epoch": 0.322180916976456,
      "grad_norm": 1.4291617734098425,
      "learning_rate": 4.0537258013894434e-06,
      "loss": 0.1913,
      "step": 715
    },
    {
      "epoch": 0.32263151965754194,
      "grad_norm": 1.0678912008243973,
      "learning_rate": 4.0508052056929995e-06,
      "loss": 0.2181,
      "step": 716
    },
    {
      "epoch": 0.32308212233862793,
      "grad_norm": 1.007985475433121,
      "learning_rate": 4.047881165579729e-06,
      "loss": 0.1888,
      "step": 717
    },
    {
      "epoch": 0.32353272501971386,
      "grad_norm": 0.9951732450674088,
      "learning_rate": 4.044953687544074e-06,
      "loss": 0.1886,
      "step": 718
    },
    {
      "epoch": 0.3239833277007998,
      "grad_norm": 1.0522626864012918,
      "learning_rate": 4.042022778088111e-06,
      "loss": 0.1925,
      "step": 719
    },
    {
      "epoch": 0.3244339303818858,
      "grad_norm": 1.1114385816164356,
      "learning_rate": 4.039088443721538e-06,
      "loss": 0.2056,
      "step": 720
    },
    {
      "epoch": 0.3248845330629717,
      "grad_norm": 1.1086070930098275,
      "learning_rate": 4.03615069096166e-06,
      "loss": 0.201,
      "step": 721
    },
    {
      "epoch": 0.3253351357440577,
      "grad_norm": 1.1126787479550855,
      "learning_rate": 4.033209526333375e-06,
      "loss": 0.1989,
      "step": 722
    },
    {
      "epoch": 0.32578573842514363,
      "grad_norm": 1.1134234887380865,
      "learning_rate": 4.030264956369158e-06,
      "loss": 0.1872,
      "step": 723
    },
    {
      "epoch": 0.32623634110622957,
      "grad_norm": 0.9526105310356239,
      "learning_rate": 4.0273169876090475e-06,
      "loss": 0.1684,
      "step": 724
    },
    {
      "epoch": 0.32668694378731555,
      "grad_norm": 1.0535295444698698,
      "learning_rate": 4.024365626600632e-06,
      "loss": 0.1918,
      "step": 725
    },
    {
      "epoch": 0.3271375464684015,
      "grad_norm": 1.0045054252904533,
      "learning_rate": 4.021410879899035e-06,
      "loss": 0.1906,
      "step": 726
    },
    {
      "epoch": 0.3275881491494874,
      "grad_norm": 0.9547446035752759,
      "learning_rate": 4.018452754066895e-06,
      "loss": 0.1938,
      "step": 727
    },
    {
      "epoch": 0.3280387518305734,
      "grad_norm": 0.9450343638165539,
      "learning_rate": 4.015491255674362e-06,
      "loss": 0.1848,
      "step": 728
    },
    {
      "epoch": 0.32848935451165934,
      "grad_norm": 0.9879269597269819,
      "learning_rate": 4.012526391299073e-06,
      "loss": 0.1902,
      "step": 729
    },
    {
      "epoch": 0.32893995719274527,
      "grad_norm": 1.038235237988296,
      "learning_rate": 4.0095581675261405e-06,
      "loss": 0.184,
      "step": 730
    },
    {
      "epoch": 0.32939055987383126,
      "grad_norm": 1.0063041190990036,
      "learning_rate": 4.006586590948141e-06,
      "loss": 0.1944,
      "step": 731
    },
    {
      "epoch": 0.3298411625549172,
      "grad_norm": 1.0159279699242378,
      "learning_rate": 4.003611668165097e-06,
      "loss": 0.2028,
      "step": 732
    },
    {
      "epoch": 0.3302917652360032,
      "grad_norm": 0.963739146223804,
      "learning_rate": 4.000633405784461e-06,
      "loss": 0.1932,
      "step": 733
    },
    {
      "epoch": 0.3307423679170891,
      "grad_norm": 1.1048402537588942,
      "learning_rate": 3.997651810421106e-06,
      "loss": 0.1941,
      "step": 734
    },
    {
      "epoch": 0.33119297059817504,
      "grad_norm": 1.159578027342481,
      "learning_rate": 3.994666888697304e-06,
      "loss": 0.2027,
      "step": 735
    },
    {
      "epoch": 0.33164357327926103,
      "grad_norm": 1.1470168711955169,
      "learning_rate": 3.991678647242719e-06,
      "loss": 0.1942,
      "step": 736
    },
    {
      "epoch": 0.33209417596034696,
      "grad_norm": 0.9693885269072713,
      "learning_rate": 3.988687092694386e-06,
      "loss": 0.1848,
      "step": 737
    },
    {
      "epoch": 0.3325447786414329,
      "grad_norm": 1.0257260940401318,
      "learning_rate": 3.985692231696699e-06,
      "loss": 0.1778,
      "step": 738
    },
    {
      "epoch": 0.3329953813225189,
      "grad_norm": 1.0951165232433586,
      "learning_rate": 3.982694070901396e-06,
      "loss": 0.1892,
      "step": 739
    },
    {
      "epoch": 0.3334459840036048,
      "grad_norm": 1.0589651968870157,
      "learning_rate": 3.979692616967543e-06,
      "loss": 0.2101,
      "step": 740
    },
    {
      "epoch": 0.3338965866846908,
      "grad_norm": 1.0310320306160434,
      "learning_rate": 3.976687876561523e-06,
      "loss": 0.1834,
      "step": 741
    },
    {
      "epoch": 0.33434718936577673,
      "grad_norm": 1.0199776554567461,
      "learning_rate": 3.973679856357014e-06,
      "loss": 0.2089,
      "step": 742
    },
    {
      "epoch": 0.33479779204686266,
      "grad_norm": 1.0407462383516777,
      "learning_rate": 3.970668563034982e-06,
      "loss": 0.206,
      "step": 743
    },
    {
      "epoch": 0.33524839472794865,
      "grad_norm": 0.9809522973667985,
      "learning_rate": 3.967654003283662e-06,
      "loss": 0.1841,
      "step": 744
    },
    {
      "epoch": 0.3356989974090346,
      "grad_norm": 1.0533085891523293,
      "learning_rate": 3.9646361837985435e-06,
      "loss": 0.1955,
      "step": 745
    },
    {
      "epoch": 0.3361496000901205,
      "grad_norm": 0.9753893521548701,
      "learning_rate": 3.961615111282357e-06,
      "loss": 0.2019,
      "step": 746
    },
    {
      "epoch": 0.3366002027712065,
      "grad_norm": 0.9561918974139727,
      "learning_rate": 3.958590792445057e-06,
      "loss": 0.1904,
      "step": 747
    },
    {
      "epoch": 0.33705080545229243,
      "grad_norm": 0.9983077525053788,
      "learning_rate": 3.9555632340038075e-06,
      "loss": 0.2079,
      "step": 748
    },
    {
      "epoch": 0.33750140813337837,
      "grad_norm": 0.9907836859881152,
      "learning_rate": 3.9525324426829716e-06,
      "loss": 0.2009,
      "step": 749
    },
    {
      "epoch": 0.33795201081446435,
      "grad_norm": 1.02928274413019,
      "learning_rate": 3.949498425214088e-06,
      "loss": 0.1992,
      "step": 750
    },
    {
      "epoch": 0.3384026134955503,
      "grad_norm": 0.971530439209323,
      "learning_rate": 3.946461188335863e-06,
      "loss": 0.2045,
      "step": 751
    },
    {
      "epoch": 0.3388532161766363,
      "grad_norm": 0.9787331384874378,
      "learning_rate": 3.943420738794153e-06,
      "loss": 0.1934,
      "step": 752
    },
    {
      "epoch": 0.3393038188577222,
      "grad_norm": 1.0152487571688742,
      "learning_rate": 3.9403770833419535e-06,
      "loss": 0.1958,
      "step": 753
    },
    {
      "epoch": 0.33975442153880814,
      "grad_norm": 1.0198547090516574,
      "learning_rate": 3.937330228739374e-06,
      "loss": 0.1917,
      "step": 754
    },
    {
      "epoch": 0.3402050242198941,
      "grad_norm": 0.9536730400178354,
      "learning_rate": 3.934280181753634e-06,
      "loss": 0.1938,
      "step": 755
    },
    {
      "epoch": 0.34065562690098006,
      "grad_norm": 0.9630484450286109,
      "learning_rate": 3.931226949159041e-06,
      "loss": 0.2046,
      "step": 756
    },
    {
      "epoch": 0.341106229582066,
      "grad_norm": 0.9695068944165888,
      "learning_rate": 3.9281705377369814e-06,
      "loss": 0.1848,
      "step": 757
    },
    {
      "epoch": 0.341556832263152,
      "grad_norm": 1.0305403392456742,
      "learning_rate": 3.925110954275897e-06,
      "loss": 0.2004,
      "step": 758
    },
    {
      "epoch": 0.3420074349442379,
      "grad_norm": 1.0173098492440267,
      "learning_rate": 3.922048205571279e-06,
      "loss": 0.2033,
      "step": 759
    },
    {
      "epoch": 0.3424580376253239,
      "grad_norm": 1.005538120788323,
      "learning_rate": 3.918982298425647e-06,
      "loss": 0.1996,
      "step": 760
    },
    {
      "epoch": 0.34290864030640983,
      "grad_norm": 0.9893565295690564,
      "learning_rate": 3.915913239648535e-06,
      "loss": 0.1974,
      "step": 761
    },
    {
      "epoch": 0.34335924298749576,
      "grad_norm": 0.9822976359231225,
      "learning_rate": 3.91284103605648e-06,
      "loss": 0.195,
      "step": 762
    },
    {
      "epoch": 0.34380984566858175,
      "grad_norm": 0.9662221221778032,
      "learning_rate": 3.909765694473e-06,
      "loss": 0.2002,
      "step": 763
    },
    {
      "epoch": 0.3442604483496677,
      "grad_norm": 0.9521020154940343,
      "learning_rate": 3.906687221728583e-06,
      "loss": 0.2094,
      "step": 764
    },
    {
      "epoch": 0.3447110510307536,
      "grad_norm": 1.0427912181383632,
      "learning_rate": 3.903605624660676e-06,
      "loss": 0.2075,
      "step": 765
    },
    {
      "epoch": 0.3451616537118396,
      "grad_norm": 1.0258973989098399,
      "learning_rate": 3.900520910113659e-06,
      "loss": 0.1995,
      "step": 766
    },
    {
      "epoch": 0.34561225639292553,
      "grad_norm": 0.9627094669670759,
      "learning_rate": 3.897433084938841e-06,
      "loss": 0.1969,
      "step": 767
    },
    {
      "epoch": 0.34606285907401146,
      "grad_norm": 1.013620020316575,
      "learning_rate": 3.894342155994437e-06,
      "loss": 0.2025,
      "step": 768
    },
    {
      "epoch": 0.34651346175509745,
      "grad_norm": 1.0600168839850888,
      "learning_rate": 3.891248130145556e-06,
      "loss": 0.1916,
      "step": 769
    },
    {
      "epoch": 0.3469640644361834,
      "grad_norm": 1.0029113853836964,
      "learning_rate": 3.888151014264189e-06,
      "loss": 0.2109,
      "step": 770
    },
    {
      "epoch": 0.34741466711726937,
      "grad_norm": 0.940613635155343,
      "learning_rate": 3.885050815229182e-06,
      "loss": 0.1868,
      "step": 771
    },
    {
      "epoch": 0.3478652697983553,
      "grad_norm": 1.032773679790588,
      "learning_rate": 3.881947539926239e-06,
      "loss": 0.1952,
      "step": 772
    },
    {
      "epoch": 0.34831587247944124,
      "grad_norm": 1.0247046836218952,
      "learning_rate": 3.878841195247888e-06,
      "loss": 0.2031,
      "step": 773
    },
    {
      "epoch": 0.3487664751605272,
      "grad_norm": 1.0044205882937856,
      "learning_rate": 3.875731788093478e-06,
      "loss": 0.2025,
      "step": 774
    },
    {
      "epoch": 0.34921707784161315,
      "grad_norm": 1.0192088758913846,
      "learning_rate": 3.872619325369162e-06,
      "loss": 0.2009,
      "step": 775
    },
    {
      "epoch": 0.3496676805226991,
      "grad_norm": 1.041923818288308,
      "learning_rate": 3.869503813987876e-06,
      "loss": 0.2059,
      "step": 776
    },
    {
      "epoch": 0.3501182832037851,
      "grad_norm": 0.9724353251647057,
      "learning_rate": 3.866385260869327e-06,
      "loss": 0.1917,
      "step": 777
    },
    {
      "epoch": 0.350568885884871,
      "grad_norm": 1.0820177279797962,
      "learning_rate": 3.8632636729399815e-06,
      "loss": 0.1858,
      "step": 778
    },
    {
      "epoch": 0.351019488565957,
      "grad_norm": 0.9825284921649515,
      "learning_rate": 3.860139057133042e-06,
      "loss": 0.1897,
      "step": 779
    },
    {
      "epoch": 0.3514700912470429,
      "grad_norm": 1.0056511814944353,
      "learning_rate": 3.85701142038844e-06,
      "loss": 0.1978,
      "step": 780
    },
    {
      "epoch": 0.35192069392812886,
      "grad_norm": 0.9430992622095392,
      "learning_rate": 3.853880769652815e-06,
      "loss": 0.1803,
      "step": 781
    },
    {
      "epoch": 0.35237129660921485,
      "grad_norm": 1.0122134922297075,
      "learning_rate": 3.850747111879499e-06,
      "loss": 0.2087,
      "step": 782
    },
    {
      "epoch": 0.3528218992903008,
      "grad_norm": 0.9881164713411805,
      "learning_rate": 3.8476104540285054e-06,
      "loss": 0.1887,
      "step": 783
    },
    {
      "epoch": 0.3532725019713867,
      "grad_norm": 0.9507298728508743,
      "learning_rate": 3.84447080306651e-06,
      "loss": 0.1912,
      "step": 784
    },
    {
      "epoch": 0.3537231046524727,
      "grad_norm": 1.022312185699959,
      "learning_rate": 3.841328165966837e-06,
      "loss": 0.188,
      "step": 785
    },
    {
      "epoch": 0.35417370733355863,
      "grad_norm": 0.9611389218860413,
      "learning_rate": 3.838182549709442e-06,
      "loss": 0.1833,
      "step": 786
    },
    {
      "epoch": 0.35462431001464456,
      "grad_norm": 0.8927961208592826,
      "learning_rate": 3.835033961280898e-06,
      "loss": 0.1792,
      "step": 787
    },
    {
      "epoch": 0.35507491269573055,
      "grad_norm": 0.9756392838597563,
      "learning_rate": 3.831882407674379e-06,
      "loss": 0.2036,
      "step": 788
    },
    {
      "epoch": 0.3555255153768165,
      "grad_norm": 1.057959342523254,
      "learning_rate": 3.828727895889644e-06,
      "loss": 0.2099,
      "step": 789
    },
    {
      "epoch": 0.35597611805790247,
      "grad_norm": 1.0103033156169718,
      "learning_rate": 3.825570432933026e-06,
      "loss": 0.2006,
      "step": 790
    },
    {
      "epoch": 0.3564267207389884,
      "grad_norm": 1.0889593061950777,
      "learning_rate": 3.8224100258174066e-06,
      "loss": 0.2034,
      "step": 791
    },
    {
      "epoch": 0.35687732342007433,
      "grad_norm": 1.0371619150509108,
      "learning_rate": 3.819246681562212e-06,
      "loss": 0.1879,
      "step": 792
    },
    {
      "epoch": 0.3573279261011603,
      "grad_norm": 1.0877887937676791,
      "learning_rate": 3.81608040719339e-06,
      "loss": 0.2106,
      "step": 793
    },
    {
      "epoch": 0.35777852878224625,
      "grad_norm": 1.1276852463298828,
      "learning_rate": 3.812911209743395e-06,
      "loss": 0.1981,
      "step": 794
    },
    {
      "epoch": 0.3582291314633322,
      "grad_norm": 1.0746615401972357,
      "learning_rate": 3.809739096251176e-06,
      "loss": 0.207,
      "step": 795
    },
    {
      "epoch": 0.35867973414441817,
      "grad_norm": 1.1233647909244424,
      "learning_rate": 3.8065640737621566e-06,
      "loss": 0.2118,
      "step": 796
    },
    {
      "epoch": 0.3591303368255041,
      "grad_norm": 0.932081242806347,
      "learning_rate": 3.803386149328223e-06,
      "loss": 0.1762,
      "step": 797
    },
    {
      "epoch": 0.3595809395065901,
      "grad_norm": 0.9944199466037139,
      "learning_rate": 3.8002053300077056e-06,
      "loss": 0.2005,
      "step": 798
    },
    {
      "epoch": 0.360031542187676,
      "grad_norm": 0.9892400702131467,
      "learning_rate": 3.7970216228653667e-06,
      "loss": 0.2013,
      "step": 799
    },
    {
      "epoch": 0.36048214486876196,
      "grad_norm": 1.0418347804871562,
      "learning_rate": 3.7938350349723784e-06,
      "loss": 0.2153,
      "step": 800
    },
    {
      "epoch": 0.36093274754984794,
      "grad_norm": 0.967000729688822,
      "learning_rate": 3.7906455734063156e-06,
      "loss": 0.1836,
      "step": 801
    },
    {
      "epoch": 0.3613833502309339,
      "grad_norm": 1.0134046794785294,
      "learning_rate": 3.7874532452511324e-06,
      "loss": 0.194,
      "step": 802
    },
    {
      "epoch": 0.3618339529120198,
      "grad_norm": 0.963474210883665,
      "learning_rate": 3.7842580575971533e-06,
      "loss": 0.1878,
      "step": 803
    },
    {
      "epoch": 0.3622845555931058,
      "grad_norm": 0.9863618310097502,
      "learning_rate": 3.7810600175410493e-06,
      "loss": 0.2038,
      "step": 804
    },
    {
      "epoch": 0.3627351582741917,
      "grad_norm": 1.0560263462424802,
      "learning_rate": 3.77785913218583e-06,
      "loss": 0.2048,
      "step": 805
    },
    {
      "epoch": 0.36318576095527766,
      "grad_norm": 0.9669821875912328,
      "learning_rate": 3.7746554086408245e-06,
      "loss": 0.1921,
      "step": 806
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 1.0005888075873781,
      "learning_rate": 3.7714488540216637e-06,
      "loss": 0.1807,
      "step": 807
    },
    {
      "epoch": 0.3640869663174496,
      "grad_norm": 1.0506705011418698,
      "learning_rate": 3.7682394754502687e-06,
      "loss": 0.2082,
      "step": 808
    },
    {
      "epoch": 0.36453756899853557,
      "grad_norm": 1.0058206144540291,
      "learning_rate": 3.7650272800548316e-06,
      "loss": 0.1972,
      "step": 809
    },
    {
      "epoch": 0.3649881716796215,
      "grad_norm": 1.015804371446931,
      "learning_rate": 3.7618122749697993e-06,
      "loss": 0.1856,
      "step": 810
    },
    {
      "epoch": 0.36543877436070743,
      "grad_norm": 1.00510305109926,
      "learning_rate": 3.7585944673358632e-06,
      "loss": 0.1879,
      "step": 811
    },
    {
      "epoch": 0.3658893770417934,
      "grad_norm": 0.9593373407976435,
      "learning_rate": 3.7553738642999354e-06,
      "loss": 0.2011,
      "step": 812
    },
    {
      "epoch": 0.36633997972287935,
      "grad_norm": 1.0581373680068094,
      "learning_rate": 3.7521504730151382e-06,
      "loss": 0.1945,
      "step": 813
    },
    {
      "epoch": 0.3667905824039653,
      "grad_norm": 1.0159773971934891,
      "learning_rate": 3.748924300640787e-06,
      "loss": 0.2079,
      "step": 814
    },
    {
      "epoch": 0.36724118508505127,
      "grad_norm": 0.9934744368057637,
      "learning_rate": 3.745695354342374e-06,
      "loss": 0.1884,
      "step": 815
    },
    {
      "epoch": 0.3676917877661372,
      "grad_norm": 0.966842175173914,
      "learning_rate": 3.742463641291552e-06,
      "loss": 0.1866,
      "step": 816
    },
    {
      "epoch": 0.36814239044722313,
      "grad_norm": 0.9893481101549668,
      "learning_rate": 3.73922916866612e-06,
      "loss": 0.1812,
      "step": 817
    },
    {
      "epoch": 0.3685929931283091,
      "grad_norm": 0.9005100733960006,
      "learning_rate": 3.7359919436500038e-06,
      "loss": 0.1809,
      "step": 818
    },
    {
      "epoch": 0.36904359580939505,
      "grad_norm": 1.0286666188930826,
      "learning_rate": 3.7327519734332453e-06,
      "loss": 0.2104,
      "step": 819
    },
    {
      "epoch": 0.36949419849048104,
      "grad_norm": 1.0335040765709167,
      "learning_rate": 3.7295092652119815e-06,
      "loss": 0.1893,
      "step": 820
    },
    {
      "epoch": 0.369944801171567,
      "grad_norm": 0.9711351753344023,
      "learning_rate": 3.726263826188432e-06,
      "loss": 0.181,
      "step": 821
    },
    {
      "epoch": 0.3703954038526529,
      "grad_norm": 1.0233958256343774,
      "learning_rate": 3.7230156635708815e-06,
      "loss": 0.2092,
      "step": 822
    },
    {
      "epoch": 0.3708460065337389,
      "grad_norm": 1.0021813440100957,
      "learning_rate": 3.7197647845736616e-06,
      "loss": 0.1829,
      "step": 823
    },
    {
      "epoch": 0.3712966092148248,
      "grad_norm": 1.0745683687708758,
      "learning_rate": 3.7165111964171407e-06,
      "loss": 0.1905,
      "step": 824
    },
    {
      "epoch": 0.37174721189591076,
      "grad_norm": 1.0130321630619799,
      "learning_rate": 3.7132549063277033e-06,
      "loss": 0.1907,
      "step": 825
    },
    {
      "epoch": 0.37219781457699674,
      "grad_norm": 0.9754258549582402,
      "learning_rate": 3.7099959215377325e-06,
      "loss": 0.1881,
      "step": 826
    },
    {
      "epoch": 0.3726484172580827,
      "grad_norm": 1.0538670773156578,
      "learning_rate": 3.7067342492855997e-06,
      "loss": 0.1743,
      "step": 827
    },
    {
      "epoch": 0.37309901993916866,
      "grad_norm": 0.9937882758720834,
      "learning_rate": 3.7034698968156434e-06,
      "loss": 0.1768,
      "step": 828
    },
    {
      "epoch": 0.3735496226202546,
      "grad_norm": 1.0726219314072167,
      "learning_rate": 3.700202871378156e-06,
      "loss": 0.2057,
      "step": 829
    },
    {
      "epoch": 0.3740002253013405,
      "grad_norm": 1.0479897737139336,
      "learning_rate": 3.696933180229366e-06,
      "loss": 0.1936,
      "step": 830
    },
    {
      "epoch": 0.3744508279824265,
      "grad_norm": 1.0252496809768523,
      "learning_rate": 3.6936608306314227e-06,
      "loss": 0.2142,
      "step": 831
    },
    {
      "epoch": 0.37490143066351245,
      "grad_norm": 1.0155015291771505,
      "learning_rate": 3.690385829852381e-06,
      "loss": 0.2065,
      "step": 832
    },
    {
      "epoch": 0.3753520333445984,
      "grad_norm": 1.0121077503776583,
      "learning_rate": 3.6871081851661825e-06,
      "loss": 0.2129,
      "step": 833
    },
    {
      "epoch": 0.37580263602568437,
      "grad_norm": 1.001133609252554,
      "learning_rate": 3.6838279038526427e-06,
      "loss": 0.1742,
      "step": 834
    },
    {
      "epoch": 0.3762532387067703,
      "grad_norm": 1.024247819419461,
      "learning_rate": 3.6805449931974313e-06,
      "loss": 0.1944,
      "step": 835
    },
    {
      "epoch": 0.37670384138785623,
      "grad_norm": 0.9805878653120293,
      "learning_rate": 3.67725946049206e-06,
      "loss": 0.193,
      "step": 836
    },
    {
      "epoch": 0.3771544440689422,
      "grad_norm": 0.9405085224910799,
      "learning_rate": 3.6739713130338617e-06,
      "loss": 0.1848,
      "step": 837
    },
    {
      "epoch": 0.37760504675002815,
      "grad_norm": 0.9600181235407473,
      "learning_rate": 3.6706805581259807e-06,
      "loss": 0.1664,
      "step": 838
    },
    {
      "epoch": 0.37805564943111414,
      "grad_norm": 1.089790825383409,
      "learning_rate": 3.6673872030773473e-06,
      "loss": 0.1956,
      "step": 839
    },
    {
      "epoch": 0.37850625211220007,
      "grad_norm": 1.0607218450180482,
      "learning_rate": 3.664091255202672e-06,
      "loss": 0.19,
      "step": 840
    },
    {
      "epoch": 0.378956854793286,
      "grad_norm": 0.9635783809020961,
      "learning_rate": 3.66079272182242e-06,
      "loss": 0.1844,
      "step": 841
    },
    {
      "epoch": 0.379407457474372,
      "grad_norm": 0.9742041910839031,
      "learning_rate": 3.657491610262802e-06,
      "loss": 0.1826,
      "step": 842
    },
    {
      "epoch": 0.3798580601554579,
      "grad_norm": 1.0457625000926623,
      "learning_rate": 3.654187927855754e-06,
      "loss": 0.1853,
      "step": 843
    },
    {
      "epoch": 0.38030866283654385,
      "grad_norm": 0.9692868387633684,
      "learning_rate": 3.6508816819389216e-06,
      "loss": 0.1834,
      "step": 844
    },
    {
      "epoch": 0.38075926551762984,
      "grad_norm": 1.0098777109203647,
      "learning_rate": 3.6475728798556426e-06,
      "loss": 0.1971,
      "step": 845
    },
    {
      "epoch": 0.3812098681987158,
      "grad_norm": 1.0007350741530037,
      "learning_rate": 3.6442615289549354e-06,
      "loss": 0.2078,
      "step": 846
    },
    {
      "epoch": 0.38166047087980176,
      "grad_norm": 1.0018669288063613,
      "learning_rate": 3.6409476365914786e-06,
      "loss": 0.1927,
      "step": 847
    },
    {
      "epoch": 0.3821110735608877,
      "grad_norm": 0.9336398545199618,
      "learning_rate": 3.6376312101255934e-06,
      "loss": 0.1829,
      "step": 848
    },
    {
      "epoch": 0.3825616762419736,
      "grad_norm": 0.9529685621807624,
      "learning_rate": 3.6343122569232313e-06,
      "loss": 0.182,
      "step": 849
    },
    {
      "epoch": 0.3830122789230596,
      "grad_norm": 0.9503579977057526,
      "learning_rate": 3.6309907843559542e-06,
      "loss": 0.1889,
      "step": 850
    },
    {
      "epoch": 0.38346288160414554,
      "grad_norm": 1.0280068475319317,
      "learning_rate": 3.6276667998009242e-06,
      "loss": 0.1921,
      "step": 851
    },
    {
      "epoch": 0.3839134842852315,
      "grad_norm": 1.095055906870189,
      "learning_rate": 3.624340310640875e-06,
      "loss": 0.1965,
      "step": 852
    },
    {
      "epoch": 0.38436408696631746,
      "grad_norm": 0.9701378921242254,
      "learning_rate": 3.621011324264109e-06,
      "loss": 0.1992,
      "step": 853
    },
    {
      "epoch": 0.3848146896474034,
      "grad_norm": 0.9833259317334301,
      "learning_rate": 3.617679848064474e-06,
      "loss": 0.1856,
      "step": 854
    },
    {
      "epoch": 0.38526529232848933,
      "grad_norm": 1.0079759165427837,
      "learning_rate": 3.6143458894413463e-06,
      "loss": 0.194,
      "step": 855
    },
    {
      "epoch": 0.3857158950095753,
      "grad_norm": 1.0679285294794556,
      "learning_rate": 3.611009455799617e-06,
      "loss": 0.1969,
      "step": 856
    },
    {
      "epoch": 0.38616649769066125,
      "grad_norm": 1.0490902535539468,
      "learning_rate": 3.6076705545496743e-06,
      "loss": 0.1909,
      "step": 857
    },
    {
      "epoch": 0.38661710037174724,
      "grad_norm": 0.9517268188133337,
      "learning_rate": 3.604329193107386e-06,
      "loss": 0.1799,
      "step": 858
    },
    {
      "epoch": 0.38706770305283317,
      "grad_norm": 0.9713171651660856,
      "learning_rate": 3.600985378894086e-06,
      "loss": 0.1796,
      "step": 859
    },
    {
      "epoch": 0.3875183057339191,
      "grad_norm": 1.0165864377870517,
      "learning_rate": 3.5976391193365544e-06,
      "loss": 0.2003,
      "step": 860
    },
    {
      "epoch": 0.3879689084150051,
      "grad_norm": 0.9665093569343551,
      "learning_rate": 3.5942904218670025e-06,
      "loss": 0.168,
      "step": 861
    },
    {
      "epoch": 0.388419511096091,
      "grad_norm": 0.9587625653826859,
      "learning_rate": 3.590939293923058e-06,
      "loss": 0.1758,
      "step": 862
    },
    {
      "epoch": 0.38887011377717695,
      "grad_norm": 0.9738142413965079,
      "learning_rate": 3.5875857429477447e-06,
      "loss": 0.1818,
      "step": 863
    },
    {
      "epoch": 0.38932071645826294,
      "grad_norm": 0.9544055095909694,
      "learning_rate": 3.584229776389468e-06,
      "loss": 0.1776,
      "step": 864
    },
    {
      "epoch": 0.38977131913934887,
      "grad_norm": 1.0326363275288095,
      "learning_rate": 3.580871401702002e-06,
      "loss": 0.1977,
      "step": 865
    },
    {
      "epoch": 0.39022192182043486,
      "grad_norm": 1.0214661988145164,
      "learning_rate": 3.5775106263444644e-06,
      "loss": 0.1899,
      "step": 866
    },
    {
      "epoch": 0.3906725245015208,
      "grad_norm": 0.9328284154481535,
      "learning_rate": 3.5741474577813086e-06,
      "loss": 0.1908,
      "step": 867
    },
    {
      "epoch": 0.3911231271826067,
      "grad_norm": 0.9729363140620655,
      "learning_rate": 3.570781903482302e-06,
      "loss": 0.1794,
      "step": 868
    },
    {
      "epoch": 0.3915737298636927,
      "grad_norm": 1.0659796965374109,
      "learning_rate": 3.5674139709225104e-06,
      "loss": 0.2057,
      "step": 869
    },
    {
      "epoch": 0.39202433254477864,
      "grad_norm": 0.9969607922349554,
      "learning_rate": 3.5640436675822833e-06,
      "loss": 0.1924,
      "step": 870
    },
    {
      "epoch": 0.3924749352258646,
      "grad_norm": 0.9635216162411704,
      "learning_rate": 3.5606710009472335e-06,
      "loss": 0.1955,
      "step": 871
    },
    {
      "epoch": 0.39292553790695056,
      "grad_norm": 1.0526216128071109,
      "learning_rate": 3.5572959785082264e-06,
      "loss": 0.1956,
      "step": 872
    },
    {
      "epoch": 0.3933761405880365,
      "grad_norm": 0.9289014978788578,
      "learning_rate": 3.5539186077613562e-06,
      "loss": 0.1821,
      "step": 873
    },
    {
      "epoch": 0.3938267432691224,
      "grad_norm": 1.0324049051398465,
      "learning_rate": 3.5505388962079337e-06,
      "loss": 0.1986,
      "step": 874
    },
    {
      "epoch": 0.3942773459502084,
      "grad_norm": 1.02273364347944,
      "learning_rate": 3.54715685135447e-06,
      "loss": 0.1918,
      "step": 875
    },
    {
      "epoch": 0.39472794863129435,
      "grad_norm": 0.9961316154921143,
      "learning_rate": 3.5437724807126583e-06,
      "loss": 0.1851,
      "step": 876
    },
    {
      "epoch": 0.39517855131238033,
      "grad_norm": 0.9794174446801692,
      "learning_rate": 3.5403857917993554e-06,
      "loss": 0.186,
      "step": 877
    },
    {
      "epoch": 0.39562915399346626,
      "grad_norm": 1.0200190502541928,
      "learning_rate": 3.5369967921365718e-06,
      "loss": 0.1769,
      "step": 878
    },
    {
      "epoch": 0.3960797566745522,
      "grad_norm": 0.9950100030363491,
      "learning_rate": 3.5336054892514437e-06,
      "loss": 0.1856,
      "step": 879
    },
    {
      "epoch": 0.3965303593556382,
      "grad_norm": 1.0393570190944585,
      "learning_rate": 3.530211890676229e-06,
      "loss": 0.1926,
      "step": 880
    },
    {
      "epoch": 0.3969809620367241,
      "grad_norm": 0.998482884761975,
      "learning_rate": 3.52681600394828e-06,
      "loss": 0.1846,
      "step": 881
    },
    {
      "epoch": 0.39743156471781005,
      "grad_norm": 1.0263931600593703,
      "learning_rate": 3.5234178366100343e-06,
      "loss": 0.1866,
      "step": 882
    },
    {
      "epoch": 0.39788216739889604,
      "grad_norm": 0.9802252111597417,
      "learning_rate": 3.520017396208993e-06,
      "loss": 0.1735,
      "step": 883
    },
    {
      "epoch": 0.39833277007998197,
      "grad_norm": 1.0025582938431925,
      "learning_rate": 3.5166146902977055e-06,
      "loss": 0.1762,
      "step": 884
    },
    {
      "epoch": 0.39878337276106796,
      "grad_norm": 1.0495303756517684,
      "learning_rate": 3.5132097264337546e-06,
      "loss": 0.19,
      "step": 885
    },
    {
      "epoch": 0.3992339754421539,
      "grad_norm": 0.9685699335532568,
      "learning_rate": 3.5098025121797375e-06,
      "loss": 0.1731,
      "step": 886
    },
    {
      "epoch": 0.3996845781232398,
      "grad_norm": 0.9810351192503403,
      "learning_rate": 3.5063930551032494e-06,
      "loss": 0.1903,
      "step": 887
    },
    {
      "epoch": 0.4001351808043258,
      "grad_norm": 1.1162636062796865,
      "learning_rate": 3.5029813627768665e-06,
      "loss": 0.2147,
      "step": 888
    },
    {
      "epoch": 0.40058578348541174,
      "grad_norm": 1.0342032669702461,
      "learning_rate": 3.499567442778131e-06,
      "loss": 0.1809,
      "step": 889
    },
    {
      "epoch": 0.40103638616649767,
      "grad_norm": 0.9844265403732025,
      "learning_rate": 3.49615130268953e-06,
      "loss": 0.2001,
      "step": 890
    },
    {
      "epoch": 0.40148698884758366,
      "grad_norm": 0.9606329542973896,
      "learning_rate": 3.4927329500984857e-06,
      "loss": 0.1743,
      "step": 891
    },
    {
      "epoch": 0.4019375915286696,
      "grad_norm": 1.0172110372775587,
      "learning_rate": 3.489312392597331e-06,
      "loss": 0.1924,
      "step": 892
    },
    {
      "epoch": 0.4023881942097555,
      "grad_norm": 1.0223046255744688,
      "learning_rate": 3.4858896377832966e-06,
      "loss": 0.2013,
      "step": 893
    },
    {
      "epoch": 0.4028387968908415,
      "grad_norm": 1.0198574264570306,
      "learning_rate": 3.482464693258496e-06,
      "loss": 0.1931,
      "step": 894
    },
    {
      "epoch": 0.40328939957192744,
      "grad_norm": 0.9771282166871073,
      "learning_rate": 3.4790375666299026e-06,
      "loss": 0.1896,
      "step": 895
    },
    {
      "epoch": 0.40374000225301343,
      "grad_norm": 0.9854569207671147,
      "learning_rate": 3.4756082655093387e-06,
      "loss": 0.1974,
      "step": 896
    },
    {
      "epoch": 0.40419060493409936,
      "grad_norm": 0.9855416122374189,
      "learning_rate": 3.4721767975134557e-06,
      "loss": 0.1961,
      "step": 897
    },
    {
      "epoch": 0.4046412076151853,
      "grad_norm": 1.108058477150135,
      "learning_rate": 3.4687431702637165e-06,
      "loss": 0.1965,
      "step": 898
    },
    {
      "epoch": 0.4050918102962713,
      "grad_norm": 0.9481181010928448,
      "learning_rate": 3.465307391386383e-06,
      "loss": 0.1869,
      "step": 899
    },
    {
      "epoch": 0.4055424129773572,
      "grad_norm": 1.0237695974057641,
      "learning_rate": 3.4618694685124927e-06,
      "loss": 0.2043,
      "step": 900
    },
    {
      "epoch": 0.40599301565844315,
      "grad_norm": 1.0884923581804866,
      "learning_rate": 3.458429409277846e-06,
      "loss": 0.1971,
      "step": 901
    },
    {
      "epoch": 0.40644361833952913,
      "grad_norm": 1.0271173038047998,
      "learning_rate": 3.454987221322989e-06,
      "loss": 0.1933,
      "step": 902
    },
    {
      "epoch": 0.40689422102061507,
      "grad_norm": 0.9107687670560624,
      "learning_rate": 3.4515429122931955e-06,
      "loss": 0.1685,
      "step": 903
    },
    {
      "epoch": 0.407344823701701,
      "grad_norm": 0.9680902887861618,
      "learning_rate": 3.4480964898384495e-06,
      "loss": 0.1869,
      "step": 904
    },
    {
      "epoch": 0.407795426382787,
      "grad_norm": 1.037144807969508,
      "learning_rate": 3.44464796161343e-06,
      "loss": 0.1981,
      "step": 905
    },
    {
      "epoch": 0.4082460290638729,
      "grad_norm": 0.974989893773787,
      "learning_rate": 3.4411973352774917e-06,
      "loss": 0.1862,
      "step": 906
    },
    {
      "epoch": 0.4086966317449589,
      "grad_norm": 1.0336123109595516,
      "learning_rate": 3.437744618494653e-06,
      "loss": 0.2073,
      "step": 907
    },
    {
      "epoch": 0.40914723442604484,
      "grad_norm": 0.95994206370728,
      "learning_rate": 3.4342898189335692e-06,
      "loss": 0.1902,
      "step": 908
    },
    {
      "epoch": 0.40959783710713077,
      "grad_norm": 1.0055854015029335,
      "learning_rate": 3.4308329442675276e-06,
      "loss": 0.1962,
      "step": 909
    },
    {
      "epoch": 0.41004843978821676,
      "grad_norm": 0.9612475566675484,
      "learning_rate": 3.42737400217442e-06,
      "loss": 0.1923,
      "step": 910
    },
    {
      "epoch": 0.4104990424693027,
      "grad_norm": 0.9549762987559874,
      "learning_rate": 3.423913000336732e-06,
      "loss": 0.1839,
      "step": 911
    },
    {
      "epoch": 0.4109496451503886,
      "grad_norm": 0.9058879282248115,
      "learning_rate": 3.4204499464415253e-06,
      "loss": 0.1776,
      "step": 912
    },
    {
      "epoch": 0.4114002478314746,
      "grad_norm": 0.9557686317220591,
      "learning_rate": 3.4169848481804165e-06,
      "loss": 0.1753,
      "step": 913
    },
    {
      "epoch": 0.41185085051256054,
      "grad_norm": 1.0083873515901405,
      "learning_rate": 3.4135177132495632e-06,
      "loss": 0.1961,
      "step": 914
    },
    {
      "epoch": 0.4123014531936465,
      "grad_norm": 0.9948199841105111,
      "learning_rate": 3.41004854934965e-06,
      "loss": 0.1921,
      "step": 915
    },
    {
      "epoch": 0.41275205587473246,
      "grad_norm": 1.064213580358891,
      "learning_rate": 3.406577364185864e-06,
      "loss": 0.1931,
      "step": 916
    },
    {
      "epoch": 0.4132026585558184,
      "grad_norm": 0.9679940604781346,
      "learning_rate": 3.403104165467883e-06,
      "loss": 0.1998,
      "step": 917
    },
    {
      "epoch": 0.4136532612369044,
      "grad_norm": 0.9808223597712648,
      "learning_rate": 3.399628960909857e-06,
      "loss": 0.19,
      "step": 918
    },
    {
      "epoch": 0.4141038639179903,
      "grad_norm": 0.9916725836748305,
      "learning_rate": 3.3961517582303916e-06,
      "loss": 0.1904,
      "step": 919
    },
    {
      "epoch": 0.41455446659907624,
      "grad_norm": 0.970381514220033,
      "learning_rate": 3.39267256515253e-06,
      "loss": 0.1871,
      "step": 920
    },
    {
      "epoch": 0.41500506928016223,
      "grad_norm": 0.9112522410429553,
      "learning_rate": 3.3891913894037354e-06,
      "loss": 0.18,
      "step": 921
    },
    {
      "epoch": 0.41545567196124816,
      "grad_norm": 0.9689677718277616,
      "learning_rate": 3.385708238715876e-06,
      "loss": 0.1822,
      "step": 922
    },
    {
      "epoch": 0.4159062746423341,
      "grad_norm": 0.9579319965915192,
      "learning_rate": 3.3822231208252053e-06,
      "loss": 0.1839,
      "step": 923
    },
    {
      "epoch": 0.4163568773234201,
      "grad_norm": 1.0343627135702063,
      "learning_rate": 3.3787360434723466e-06,
      "loss": 0.1905,
      "step": 924
    },
    {
      "epoch": 0.416807480004506,
      "grad_norm": 0.9307878163914602,
      "learning_rate": 3.3752470144022745e-06,
      "loss": 0.1672,
      "step": 925
    },
    {
      "epoch": 0.417258082685592,
      "grad_norm": 0.9749987708402977,
      "learning_rate": 3.371756041364301e-06,
      "loss": 0.1977,
      "step": 926
    },
    {
      "epoch": 0.41770868536667793,
      "grad_norm": 1.0281325905465875,
      "learning_rate": 3.3682631321120507e-06,
      "loss": 0.2033,
      "step": 927
    },
    {
      "epoch": 0.41815928804776387,
      "grad_norm": 0.9271935858859663,
      "learning_rate": 3.364768294403455e-06,
      "loss": 0.1642,
      "step": 928
    },
    {
      "epoch": 0.41860989072884985,
      "grad_norm": 0.9740269407929583,
      "learning_rate": 3.361271536000723e-06,
      "loss": 0.1839,
      "step": 929
    },
    {
      "epoch": 0.4190604934099358,
      "grad_norm": 0.9661329367186603,
      "learning_rate": 3.3577728646703335e-06,
      "loss": 0.1789,
      "step": 930
    },
    {
      "epoch": 0.4195110960910217,
      "grad_norm": 1.0041765071251951,
      "learning_rate": 3.354272288183012e-06,
      "loss": 0.1957,
      "step": 931
    },
    {
      "epoch": 0.4199616987721077,
      "grad_norm": 0.984166508205924,
      "learning_rate": 3.3507698143137157e-06,
      "loss": 0.1852,
      "step": 932
    },
    {
      "epoch": 0.42041230145319364,
      "grad_norm": 1.0382196872498093,
      "learning_rate": 3.3472654508416157e-06,
      "loss": 0.1881,
      "step": 933
    },
    {
      "epoch": 0.4208629041342796,
      "grad_norm": 1.0415012014327485,
      "learning_rate": 3.3437592055500825e-06,
      "loss": 0.1798,
      "step": 934
    },
    {
      "epoch": 0.42131350681536556,
      "grad_norm": 1.0009036129479045,
      "learning_rate": 3.340251086226663e-06,
      "loss": 0.189,
      "step": 935
    },
    {
      "epoch": 0.4217641094964515,
      "grad_norm": 0.9821502289558824,
      "learning_rate": 3.3367411006630677e-06,
      "loss": 0.2014,
      "step": 936
    },
    {
      "epoch": 0.4222147121775375,
      "grad_norm": 0.8914293408127837,
      "learning_rate": 3.333229256655153e-06,
      "loss": 0.1873,
      "step": 937
    },
    {
      "epoch": 0.4226653148586234,
      "grad_norm": 1.0219408028076342,
      "learning_rate": 3.3297155620029e-06,
      "loss": 0.1913,
      "step": 938
    },
    {
      "epoch": 0.42311591753970934,
      "grad_norm": 1.0374615443742041,
      "learning_rate": 3.326200024510405e-06,
      "loss": 0.2085,
      "step": 939
    },
    {
      "epoch": 0.42356652022079533,
      "grad_norm": 1.0704603560066228,
      "learning_rate": 3.3226826519858526e-06,
      "loss": 0.1789,
      "step": 940
    },
    {
      "epoch": 0.42401712290188126,
      "grad_norm": 0.9484752054686993,
      "learning_rate": 3.3191634522415064e-06,
      "loss": 0.2042,
      "step": 941
    },
    {
      "epoch": 0.4244677255829672,
      "grad_norm": 0.9627020313640821,
      "learning_rate": 3.315642433093686e-06,
      "loss": 0.2018,
      "step": 942
    },
    {
      "epoch": 0.4249183282640532,
      "grad_norm": 1.0542705175273657,
      "learning_rate": 3.3121196023627543e-06,
      "loss": 0.2026,
      "step": 943
    },
    {
      "epoch": 0.4253689309451391,
      "grad_norm": 1.0358385566299095,
      "learning_rate": 3.3085949678730953e-06,
      "loss": 0.1942,
      "step": 944
    },
    {
      "epoch": 0.4258195336262251,
      "grad_norm": 0.9995141966693947,
      "learning_rate": 3.305068537453102e-06,
      "loss": 0.188,
      "step": 945
    },
    {
      "epoch": 0.42627013630731103,
      "grad_norm": 0.9300383955846249,
      "learning_rate": 3.3015403189351536e-06,
      "loss": 0.1782,
      "step": 946
    },
    {
      "epoch": 0.42672073898839696,
      "grad_norm": 0.9899587938284999,
      "learning_rate": 3.2980103201556023e-06,
      "loss": 0.1884,
      "step": 947
    },
    {
      "epoch": 0.42717134166948295,
      "grad_norm": 0.9620887217479097,
      "learning_rate": 3.2944785489547544e-06,
      "loss": 0.1795,
      "step": 948
    },
    {
      "epoch": 0.4276219443505689,
      "grad_norm": 0.9312121794699554,
      "learning_rate": 3.290945013176852e-06,
      "loss": 0.1852,
      "step": 949
    },
    {
      "epoch": 0.4280725470316548,
      "grad_norm": 1.118845733713994,
      "learning_rate": 3.2874097206700566e-06,
      "loss": 0.2082,
      "step": 950
    },
    {
      "epoch": 0.4285231497127408,
      "grad_norm": 1.0087779710378257,
      "learning_rate": 3.2838726792864315e-06,
      "loss": 0.1922,
      "step": 951
    },
    {
      "epoch": 0.42897375239382673,
      "grad_norm": 0.9053213847907385,
      "learning_rate": 3.2803338968819264e-06,
      "loss": 0.1706,
      "step": 952
    },
    {
      "epoch": 0.4294243550749127,
      "grad_norm": 0.94550246667197,
      "learning_rate": 3.2767933813163542e-06,
      "loss": 0.1878,
      "step": 953
    },
    {
      "epoch": 0.42987495775599865,
      "grad_norm": 1.0667255593304628,
      "learning_rate": 3.2732511404533797e-06,
      "loss": 0.1851,
      "step": 954
    },
    {
      "epoch": 0.4303255604370846,
      "grad_norm": 0.9771340881855369,
      "learning_rate": 3.2697071821604986e-06,
      "loss": 0.1882,
      "step": 955
    },
    {
      "epoch": 0.4307761631181706,
      "grad_norm": 0.9550554995820587,
      "learning_rate": 3.266161514309023e-06,
      "loss": 0.1739,
      "step": 956
    },
    {
      "epoch": 0.4312267657992565,
      "grad_norm": 1.0298010176964945,
      "learning_rate": 3.262614144774059e-06,
      "loss": 0.1776,
      "step": 957
    },
    {
      "epoch": 0.43167736848034244,
      "grad_norm": 1.0111021112344831,
      "learning_rate": 3.259065081434495e-06,
      "loss": 0.1827,
      "step": 958
    },
    {
      "epoch": 0.4321279711614284,
      "grad_norm": 1.0110533511752027,
      "learning_rate": 3.255514332172979e-06,
      "loss": 0.1921,
      "step": 959
    },
    {
      "epoch": 0.43257857384251436,
      "grad_norm": 1.0569579356257515,
      "learning_rate": 3.2519619048759056e-06,
      "loss": 0.199,
      "step": 960
    },
    {
      "epoch": 0.4330291765236003,
      "grad_norm": 1.0231127453347497,
      "learning_rate": 3.248407807433396e-06,
      "loss": 0.1816,
      "step": 961
    },
    {
      "epoch": 0.4334797792046863,
      "grad_norm": 1.0261507249808781,
      "learning_rate": 3.2448520477392788e-06,
      "loss": 0.1998,
      "step": 962
    },
    {
      "epoch": 0.4339303818857722,
      "grad_norm": 1.0499398268465796,
      "learning_rate": 3.2412946336910778e-06,
      "loss": 0.1879,
      "step": 963
    },
    {
      "epoch": 0.4343809845668582,
      "grad_norm": 0.9178380998321612,
      "learning_rate": 3.237735573189989e-06,
      "loss": 0.1749,
      "step": 964
    },
    {
      "epoch": 0.43483158724794413,
      "grad_norm": 1.0622394272309097,
      "learning_rate": 3.234174874140866e-06,
      "loss": 0.1816,
      "step": 965
    },
    {
      "epoch": 0.43528218992903006,
      "grad_norm": 1.0657095360369357,
      "learning_rate": 3.230612544452202e-06,
      "loss": 0.1993,
      "step": 966
    },
    {
      "epoch": 0.43573279261011605,
      "grad_norm": 0.9392726541767252,
      "learning_rate": 3.2270485920361093e-06,
      "loss": 0.1816,
      "step": 967
    },
    {
      "epoch": 0.436183395291202,
      "grad_norm": 1.024166929104129,
      "learning_rate": 3.2234830248083095e-06,
      "loss": 0.1845,
      "step": 968
    },
    {
      "epoch": 0.4366339979722879,
      "grad_norm": 1.0403201840799952,
      "learning_rate": 3.219915850688106e-06,
      "loss": 0.1871,
      "step": 969
    },
    {
      "epoch": 0.4370846006533739,
      "grad_norm": 1.0511669741921026,
      "learning_rate": 3.2163470775983733e-06,
      "loss": 0.2009,
      "step": 970
    },
    {
      "epoch": 0.43753520333445983,
      "grad_norm": 0.9891281098274592,
      "learning_rate": 3.2127767134655374e-06,
      "loss": 0.1888,
      "step": 971
    },
    {
      "epoch": 0.4379858060155458,
      "grad_norm": 1.0288193967999293,
      "learning_rate": 3.209204766219558e-06,
      "loss": 0.1897,
      "step": 972
    },
    {
      "epoch": 0.43843640869663175,
      "grad_norm": 0.9114869638159263,
      "learning_rate": 3.205631243793909e-06,
      "loss": 0.1746,
      "step": 973
    },
    {
      "epoch": 0.4388870113777177,
      "grad_norm": 0.9351530196812792,
      "learning_rate": 3.202056154125567e-06,
      "loss": 0.1821,
      "step": 974
    },
    {
      "epoch": 0.43933761405880367,
      "grad_norm": 0.8879072088338338,
      "learning_rate": 3.198479505154984e-06,
      "loss": 0.1681,
      "step": 975
    },
    {
      "epoch": 0.4397882167398896,
      "grad_norm": 0.9869879230018886,
      "learning_rate": 3.1949013048260813e-06,
      "loss": 0.1813,
      "step": 976
    },
    {
      "epoch": 0.44023881942097554,
      "grad_norm": 0.9639695729445393,
      "learning_rate": 3.1913215610862208e-06,
      "loss": 0.1739,
      "step": 977
    },
    {
      "epoch": 0.4406894221020615,
      "grad_norm": 0.9647122784620339,
      "learning_rate": 3.1877402818861954e-06,
      "loss": 0.1771,
      "step": 978
    },
    {
      "epoch": 0.44114002478314746,
      "grad_norm": 0.9353403144241794,
      "learning_rate": 3.184157475180208e-06,
      "loss": 0.182,
      "step": 979
    },
    {
      "epoch": 0.4415906274642334,
      "grad_norm": 1.0466040073099763,
      "learning_rate": 3.1805731489258516e-06,
      "loss": 0.2019,
      "step": 980
    },
    {
      "epoch": 0.4420412301453194,
      "grad_norm": 0.9855274074373859,
      "learning_rate": 3.1769873110840977e-06,
      "loss": 0.195,
      "step": 981
    },
    {
      "epoch": 0.4424918328264053,
      "grad_norm": 0.9571646517475396,
      "learning_rate": 3.1733999696192736e-06,
      "loss": 0.1923,
      "step": 982
    },
    {
      "epoch": 0.4429424355074913,
      "grad_norm": 0.9290347394228827,
      "learning_rate": 3.1698111324990454e-06,
      "loss": 0.1716,
      "step": 983
    },
    {
      "epoch": 0.4433930381885772,
      "grad_norm": 0.9607777203317588,
      "learning_rate": 3.1662208076944027e-06,
      "loss": 0.1693,
      "step": 984
    },
    {
      "epoch": 0.44384364086966316,
      "grad_norm": 0.9973605790758735,
      "learning_rate": 3.162629003179638e-06,
      "loss": 0.2023,
      "step": 985
    },
    {
      "epoch": 0.44429424355074915,
      "grad_norm": 1.0225259641771225,
      "learning_rate": 3.1590357269323312e-06,
      "loss": 0.2018,
      "step": 986
    },
    {
      "epoch": 0.4447448462318351,
      "grad_norm": 0.9810464652010766,
      "learning_rate": 3.15544098693333e-06,
      "loss": 0.19,
      "step": 987
    },
    {
      "epoch": 0.445195448912921,
      "grad_norm": 1.032890935446246,
      "learning_rate": 3.151844791166735e-06,
      "loss": 0.1913,
      "step": 988
    },
    {
      "epoch": 0.445646051594007,
      "grad_norm": 1.0218822651561286,
      "learning_rate": 3.1482471476198784e-06,
      "loss": 0.1998,
      "step": 989
    },
    {
      "epoch": 0.44609665427509293,
      "grad_norm": 0.9838750989611218,
      "learning_rate": 3.1446480642833077e-06,
      "loss": 0.1879,
      "step": 990
    },
    {
      "epoch": 0.44654725695617886,
      "grad_norm": 0.9533897607350283,
      "learning_rate": 3.14104754915077e-06,
      "loss": 0.1718,
      "step": 991
    },
    {
      "epoch": 0.44699785963726485,
      "grad_norm": 1.0097739680223856,
      "learning_rate": 3.137445610219192e-06,
      "loss": 0.1856,
      "step": 992
    },
    {
      "epoch": 0.4474484623183508,
      "grad_norm": 0.975630130381292,
      "learning_rate": 3.133842255488661e-06,
      "loss": 0.2084,
      "step": 993
    },
    {
      "epoch": 0.44789906499943677,
      "grad_norm": 0.9277306719461578,
      "learning_rate": 3.130237492962411e-06,
      "loss": 0.1828,
      "step": 994
    },
    {
      "epoch": 0.4483496676805227,
      "grad_norm": 0.9568578595773181,
      "learning_rate": 3.1266313306468018e-06,
      "loss": 0.193,
      "step": 995
    },
    {
      "epoch": 0.44880027036160863,
      "grad_norm": 1.0000320220613919,
      "learning_rate": 3.1230237765513023e-06,
      "loss": 0.1813,
      "step": 996
    },
    {
      "epoch": 0.4492508730426946,
      "grad_norm": 0.9982728552087634,
      "learning_rate": 3.119414838688473e-06,
      "loss": 0.1962,
      "step": 997
    },
    {
      "epoch": 0.44970147572378055,
      "grad_norm": 0.9273058649051045,
      "learning_rate": 3.1158045250739473e-06,
      "loss": 0.1845,
      "step": 998
    },
    {
      "epoch": 0.4501520784048665,
      "grad_norm": 0.9096138394901124,
      "learning_rate": 3.1121928437264138e-06,
      "loss": 0.1715,
      "step": 999
    },
    {
      "epoch": 0.4506026810859525,
      "grad_norm": 1.0183371982803404,
      "learning_rate": 3.1085798026676e-06,
      "loss": 0.1811,
      "step": 1000
    },
    {
      "epoch": 0.4506026810859525,
      "eval_loss": 0.189128577709198,
      "eval_runtime": 59.4414,
      "eval_samples_per_second": 24.141,
      "eval_steps_per_second": 3.028,
      "step": 1000
    },
    {
      "epoch": 0.4510532837670384,
      "grad_norm": 0.9671572024242282,
      "learning_rate": 3.1049654099222542e-06,
      "loss": 0.1912,
      "step": 1001
    },
    {
      "epoch": 0.4515038864481244,
      "grad_norm": 0.9162211305759308,
      "learning_rate": 3.1013496735181232e-06,
      "loss": 0.176,
      "step": 1002
    },
    {
      "epoch": 0.4519544891292103,
      "grad_norm": 1.0090932364505272,
      "learning_rate": 3.0977326014859415e-06,
      "loss": 0.2013,
      "step": 1003
    },
    {
      "epoch": 0.45240509181029626,
      "grad_norm": 0.8951383914420861,
      "learning_rate": 3.09411420185941e-06,
      "loss": 0.1749,
      "step": 1004
    },
    {
      "epoch": 0.45285569449138224,
      "grad_norm": 0.9397408568438251,
      "learning_rate": 3.090494482675176e-06,
      "loss": 0.1804,
      "step": 1005
    },
    {
      "epoch": 0.4533062971724682,
      "grad_norm": 0.9022667632823632,
      "learning_rate": 3.0868734519728194e-06,
      "loss": 0.1788,
      "step": 1006
    },
    {
      "epoch": 0.4537568998535541,
      "grad_norm": 1.0031529885463817,
      "learning_rate": 3.0832511177948326e-06,
      "loss": 0.2066,
      "step": 1007
    },
    {
      "epoch": 0.4542075025346401,
      "grad_norm": 1.0303330103237227,
      "learning_rate": 3.0796274881866034e-06,
      "loss": 0.1992,
      "step": 1008
    },
    {
      "epoch": 0.454658105215726,
      "grad_norm": 1.0239442642225518,
      "learning_rate": 3.0760025711963964e-06,
      "loss": 0.1913,
      "step": 1009
    },
    {
      "epoch": 0.45510870789681196,
      "grad_norm": 1.0115757759674917,
      "learning_rate": 3.0723763748753354e-06,
      "loss": 0.181,
      "step": 1010
    },
    {
      "epoch": 0.45555931057789795,
      "grad_norm": 1.024472357217328,
      "learning_rate": 3.0687489072773864e-06,
      "loss": 0.1949,
      "step": 1011
    },
    {
      "epoch": 0.4560099132589839,
      "grad_norm": 1.0663046028032832,
      "learning_rate": 3.0651201764593375e-06,
      "loss": 0.2073,
      "step": 1012
    },
    {
      "epoch": 0.45646051594006987,
      "grad_norm": 1.0420504549269418,
      "learning_rate": 3.0614901904807836e-06,
      "loss": 0.1842,
      "step": 1013
    },
    {
      "epoch": 0.4569111186211558,
      "grad_norm": 1.1242567350514105,
      "learning_rate": 3.0578589574041097e-06,
      "loss": 0.2077,
      "step": 1014
    },
    {
      "epoch": 0.45736172130224173,
      "grad_norm": 0.9648102387148972,
      "learning_rate": 3.0542264852944635e-06,
      "loss": 0.182,
      "step": 1015
    },
    {
      "epoch": 0.4578123239833277,
      "grad_norm": 0.9493708361147652,
      "learning_rate": 3.0505927822197533e-06,
      "loss": 0.1685,
      "step": 1016
    },
    {
      "epoch": 0.45826292666441365,
      "grad_norm": 0.9934079363031364,
      "learning_rate": 3.0469578562506165e-06,
      "loss": 0.1935,
      "step": 1017
    },
    {
      "epoch": 0.4587135293454996,
      "grad_norm": 0.9805921173422116,
      "learning_rate": 3.0433217154604067e-06,
      "loss": 0.1908,
      "step": 1018
    },
    {
      "epoch": 0.45916413202658557,
      "grad_norm": 1.065514609639169,
      "learning_rate": 3.0396843679251777e-06,
      "loss": 0.2006,
      "step": 1019
    },
    {
      "epoch": 0.4596147347076715,
      "grad_norm": 0.9809771886572722,
      "learning_rate": 3.0360458217236604e-06,
      "loss": 0.1815,
      "step": 1020
    },
    {
      "epoch": 0.4600653373887575,
      "grad_norm": 1.023903500208084,
      "learning_rate": 3.0324060849372526e-06,
      "loss": 0.1962,
      "step": 1021
    },
    {
      "epoch": 0.4605159400698434,
      "grad_norm": 1.0217469739032734,
      "learning_rate": 3.028765165649992e-06,
      "loss": 0.1985,
      "step": 1022
    },
    {
      "epoch": 0.46096654275092935,
      "grad_norm": 1.0221731312641227,
      "learning_rate": 3.0251230719485465e-06,
      "loss": 0.1992,
      "step": 1023
    },
    {
      "epoch": 0.46141714543201534,
      "grad_norm": 1.0164802191447049,
      "learning_rate": 3.0214798119221884e-06,
      "loss": 0.1994,
      "step": 1024
    },
    {
      "epoch": 0.4618677481131013,
      "grad_norm": 0.9419980545476748,
      "learning_rate": 3.0178353936627835e-06,
      "loss": 0.1812,
      "step": 1025
    },
    {
      "epoch": 0.4623183507941872,
      "grad_norm": 0.956154108614536,
      "learning_rate": 3.0141898252647682e-06,
      "loss": 0.1887,
      "step": 1026
    },
    {
      "epoch": 0.4627689534752732,
      "grad_norm": 1.0955572002986789,
      "learning_rate": 3.0105431148251364e-06,
      "loss": 0.1871,
      "step": 1027
    },
    {
      "epoch": 0.4632195561563591,
      "grad_norm": 1.0245926161342613,
      "learning_rate": 3.0068952704434145e-06,
      "loss": 0.186,
      "step": 1028
    },
    {
      "epoch": 0.46367015883744506,
      "grad_norm": 0.9618126188953687,
      "learning_rate": 3.0032463002216504e-06,
      "loss": 0.1867,
      "step": 1029
    },
    {
      "epoch": 0.46412076151853104,
      "grad_norm": 1.1893777161990908,
      "learning_rate": 2.999596212264392e-06,
      "loss": 0.2028,
      "step": 1030
    },
    {
      "epoch": 0.464571364199617,
      "grad_norm": 0.9948596042457155,
      "learning_rate": 2.9959450146786674e-06,
      "loss": 0.1914,
      "step": 1031
    },
    {
      "epoch": 0.46502196688070296,
      "grad_norm": 1.0487962174424645,
      "learning_rate": 2.9922927155739737e-06,
      "loss": 0.2008,
      "step": 1032
    },
    {
      "epoch": 0.4654725695617889,
      "grad_norm": 1.065771087796345,
      "learning_rate": 2.9886393230622507e-06,
      "loss": 0.1951,
      "step": 1033
    },
    {
      "epoch": 0.46592317224287483,
      "grad_norm": 1.0549580408530124,
      "learning_rate": 2.984984845257868e-06,
      "loss": 0.2018,
      "step": 1034
    },
    {
      "epoch": 0.4663737749239608,
      "grad_norm": 0.9777277229138798,
      "learning_rate": 2.981329290277605e-06,
      "loss": 0.1827,
      "step": 1035
    },
    {
      "epoch": 0.46682437760504675,
      "grad_norm": 0.9528614427706873,
      "learning_rate": 2.977672666240636e-06,
      "loss": 0.1826,
      "step": 1036
    },
    {
      "epoch": 0.4672749802861327,
      "grad_norm": 0.893984600025219,
      "learning_rate": 2.974014981268507e-06,
      "loss": 0.1708,
      "step": 1037
    },
    {
      "epoch": 0.46772558296721867,
      "grad_norm": 0.94876165184311,
      "learning_rate": 2.9703562434851218e-06,
      "loss": 0.1789,
      "step": 1038
    },
    {
      "epoch": 0.4681761856483046,
      "grad_norm": 1.0414212143930934,
      "learning_rate": 2.966696461016721e-06,
      "loss": 0.2034,
      "step": 1039
    },
    {
      "epoch": 0.4686267883293906,
      "grad_norm": 0.9462011872528657,
      "learning_rate": 2.9630356419918682e-06,
      "loss": 0.1719,
      "step": 1040
    },
    {
      "epoch": 0.4690773910104765,
      "grad_norm": 0.9266863274496223,
      "learning_rate": 2.9593737945414264e-06,
      "loss": 0.1813,
      "step": 1041
    },
    {
      "epoch": 0.46952799369156245,
      "grad_norm": 1.0780107256167346,
      "learning_rate": 2.9557109267985445e-06,
      "loss": 0.1891,
      "step": 1042
    },
    {
      "epoch": 0.46997859637264844,
      "grad_norm": 0.9718365354652926,
      "learning_rate": 2.952047046898637e-06,
      "loss": 0.1778,
      "step": 1043
    },
    {
      "epoch": 0.47042919905373437,
      "grad_norm": 0.9854537626498936,
      "learning_rate": 2.9483821629793673e-06,
      "loss": 0.1855,
      "step": 1044
    },
    {
      "epoch": 0.4708798017348203,
      "grad_norm": 0.9509490677167141,
      "learning_rate": 2.9447162831806275e-06,
      "loss": 0.1871,
      "step": 1045
    },
    {
      "epoch": 0.4713304044159063,
      "grad_norm": 0.9835264138568843,
      "learning_rate": 2.941049415644522e-06,
      "loss": 0.181,
      "step": 1046
    },
    {
      "epoch": 0.4717810070969922,
      "grad_norm": 0.976177751134452,
      "learning_rate": 2.9373815685153485e-06,
      "loss": 0.1745,
      "step": 1047
    },
    {
      "epoch": 0.47223160977807815,
      "grad_norm": 1.0395516553789195,
      "learning_rate": 2.933712749939582e-06,
      "loss": 0.2009,
      "step": 1048
    },
    {
      "epoch": 0.47268221245916414,
      "grad_norm": 0.9508188955276239,
      "learning_rate": 2.9300429680658538e-06,
      "loss": 0.192,
      "step": 1049
    },
    {
      "epoch": 0.4731328151402501,
      "grad_norm": 1.030095865559414,
      "learning_rate": 2.9263722310449353e-06,
      "loss": 0.2087,
      "step": 1050
    },
    {
      "epoch": 0.47358341782133606,
      "grad_norm": 1.0090146660033465,
      "learning_rate": 2.9227005470297194e-06,
      "loss": 0.198,
      "step": 1051
    },
    {
      "epoch": 0.474034020502422,
      "grad_norm": 1.01886158123234,
      "learning_rate": 2.919027924175201e-06,
      "loss": 0.1982,
      "step": 1052
    },
    {
      "epoch": 0.4744846231835079,
      "grad_norm": 0.9858334972569566,
      "learning_rate": 2.915354370638462e-06,
      "loss": 0.1833,
      "step": 1053
    },
    {
      "epoch": 0.4749352258645939,
      "grad_norm": 0.9198114480645141,
      "learning_rate": 2.9116798945786515e-06,
      "loss": 0.1779,
      "step": 1054
    },
    {
      "epoch": 0.47538582854567984,
      "grad_norm": 1.0202969087409115,
      "learning_rate": 2.9080045041569647e-06,
      "loss": 0.1925,
      "step": 1055
    },
    {
      "epoch": 0.4758364312267658,
      "grad_norm": 0.9323881824017032,
      "learning_rate": 2.904328207536632e-06,
      "loss": 0.1896,
      "step": 1056
    },
    {
      "epoch": 0.47628703390785176,
      "grad_norm": 0.9957756137102491,
      "learning_rate": 2.900651012882893e-06,
      "loss": 0.1846,
      "step": 1057
    },
    {
      "epoch": 0.4767376365889377,
      "grad_norm": 0.921082253021773,
      "learning_rate": 2.896972928362983e-06,
      "loss": 0.1858,
      "step": 1058
    },
    {
      "epoch": 0.4771882392700237,
      "grad_norm": 0.9220354600553122,
      "learning_rate": 2.893293962146114e-06,
      "loss": 0.1695,
      "step": 1059
    },
    {
      "epoch": 0.4776388419511096,
      "grad_norm": 1.045451312341842,
      "learning_rate": 2.8896141224034554e-06,
      "loss": 0.199,
      "step": 1060
    },
    {
      "epoch": 0.47808944463219555,
      "grad_norm": 0.9166173855182119,
      "learning_rate": 2.885933417308118e-06,
      "loss": 0.1632,
      "step": 1061
    },
    {
      "epoch": 0.47854004731328154,
      "grad_norm": 0.9995930807668175,
      "learning_rate": 2.8822518550351356e-06,
      "loss": 0.1839,
      "step": 1062
    },
    {
      "epoch": 0.47899064999436747,
      "grad_norm": 0.9823258706934731,
      "learning_rate": 2.878569443761442e-06,
      "loss": 0.1845,
      "step": 1063
    },
    {
      "epoch": 0.4794412526754534,
      "grad_norm": 1.0462859714933759,
      "learning_rate": 2.87488619166586e-06,
      "loss": 0.2096,
      "step": 1064
    },
    {
      "epoch": 0.4798918553565394,
      "grad_norm": 0.9484837330036915,
      "learning_rate": 2.8712021069290786e-06,
      "loss": 0.1748,
      "step": 1065
    },
    {
      "epoch": 0.4803424580376253,
      "grad_norm": 0.943429390904282,
      "learning_rate": 2.8675171977336357e-06,
      "loss": 0.1777,
      "step": 1066
    },
    {
      "epoch": 0.48079306071871125,
      "grad_norm": 0.965224271549074,
      "learning_rate": 2.863831472263904e-06,
      "loss": 0.1924,
      "step": 1067
    },
    {
      "epoch": 0.48124366339979724,
      "grad_norm": 1.0724951204071391,
      "learning_rate": 2.8601449387060622e-06,
      "loss": 0.2004,
      "step": 1068
    },
    {
      "epoch": 0.48169426608088317,
      "grad_norm": 1.0527358404221765,
      "learning_rate": 2.8564576052480895e-06,
      "loss": 0.1992,
      "step": 1069
    },
    {
      "epoch": 0.48214486876196916,
      "grad_norm": 0.9835497307888469,
      "learning_rate": 2.8527694800797417e-06,
      "loss": 0.1818,
      "step": 1070
    },
    {
      "epoch": 0.4825954714430551,
      "grad_norm": 0.9246506213040395,
      "learning_rate": 2.8490805713925298e-06,
      "loss": 0.1791,
      "step": 1071
    },
    {
      "epoch": 0.483046074124141,
      "grad_norm": 0.9701524857121642,
      "learning_rate": 2.845390887379706e-06,
      "loss": 0.1793,
      "step": 1072
    },
    {
      "epoch": 0.483496676805227,
      "grad_norm": 1.01141868910757,
      "learning_rate": 2.8417004362362465e-06,
      "loss": 0.1831,
      "step": 1073
    },
    {
      "epoch": 0.48394727948631294,
      "grad_norm": 0.9620211674176621,
      "learning_rate": 2.838009226158829e-06,
      "loss": 0.1934,
      "step": 1074
    },
    {
      "epoch": 0.4843978821673989,
      "grad_norm": 0.9797701458800275,
      "learning_rate": 2.8343172653458194e-06,
      "loss": 0.1898,
      "step": 1075
    },
    {
      "epoch": 0.48484848484848486,
      "grad_norm": 0.9517023202475261,
      "learning_rate": 2.8306245619972476e-06,
      "loss": 0.1795,
      "step": 1076
    },
    {
      "epoch": 0.4852990875295708,
      "grad_norm": 0.9797022376101314,
      "learning_rate": 2.826931124314796e-06,
      "loss": 0.181,
      "step": 1077
    },
    {
      "epoch": 0.4857496902106567,
      "grad_norm": 0.9979686434950376,
      "learning_rate": 2.8232369605017757e-06,
      "loss": 0.1806,
      "step": 1078
    },
    {
      "epoch": 0.4862002928917427,
      "grad_norm": 1.022734670572374,
      "learning_rate": 2.8195420787631113e-06,
      "loss": 0.2038,
      "step": 1079
    },
    {
      "epoch": 0.48665089557282865,
      "grad_norm": 1.0508347141889778,
      "learning_rate": 2.8158464873053236e-06,
      "loss": 0.1871,
      "step": 1080
    },
    {
      "epoch": 0.48710149825391463,
      "grad_norm": 1.0117363738105383,
      "learning_rate": 2.8121501943365066e-06,
      "loss": 0.1844,
      "step": 1081
    },
    {
      "epoch": 0.48755210093500057,
      "grad_norm": 0.9731494514961763,
      "learning_rate": 2.808453208066314e-06,
      "loss": 0.1796,
      "step": 1082
    },
    {
      "epoch": 0.4880027036160865,
      "grad_norm": 0.9779594049370961,
      "learning_rate": 2.8047555367059404e-06,
      "loss": 0.1842,
      "step": 1083
    },
    {
      "epoch": 0.4884533062971725,
      "grad_norm": 1.0236799385043172,
      "learning_rate": 2.8010571884681004e-06,
      "loss": 0.1994,
      "step": 1084
    },
    {
      "epoch": 0.4889039089782584,
      "grad_norm": 0.9975674985465199,
      "learning_rate": 2.7973581715670124e-06,
      "loss": 0.1939,
      "step": 1085
    },
    {
      "epoch": 0.48935451165934435,
      "grad_norm": 0.9722700577518006,
      "learning_rate": 2.7936584942183804e-06,
      "loss": 0.1864,
      "step": 1086
    },
    {
      "epoch": 0.48980511434043034,
      "grad_norm": 0.9792037443912968,
      "learning_rate": 2.7899581646393746e-06,
      "loss": 0.1741,
      "step": 1087
    },
    {
      "epoch": 0.49025571702151627,
      "grad_norm": 0.925245177190625,
      "learning_rate": 2.7862571910486148e-06,
      "loss": 0.1793,
      "step": 1088
    },
    {
      "epoch": 0.49070631970260226,
      "grad_norm": 0.9677656832455358,
      "learning_rate": 2.7825555816661503e-06,
      "loss": 0.1824,
      "step": 1089
    },
    {
      "epoch": 0.4911569223836882,
      "grad_norm": 0.9340259731910094,
      "learning_rate": 2.778853344713443e-06,
      "loss": 0.1846,
      "step": 1090
    },
    {
      "epoch": 0.4916075250647741,
      "grad_norm": 0.9199222980083757,
      "learning_rate": 2.7751504884133484e-06,
      "loss": 0.1725,
      "step": 1091
    },
    {
      "epoch": 0.4920581277458601,
      "grad_norm": 1.0357457170145339,
      "learning_rate": 2.7714470209900974e-06,
      "loss": 0.2018,
      "step": 1092
    },
    {
      "epoch": 0.49250873042694604,
      "grad_norm": 1.0034531580348318,
      "learning_rate": 2.7677429506692788e-06,
      "loss": 0.1843,
      "step": 1093
    },
    {
      "epoch": 0.49295933310803197,
      "grad_norm": 0.9227250756376808,
      "learning_rate": 2.76403828567782e-06,
      "loss": 0.1703,
      "step": 1094
    },
    {
      "epoch": 0.49340993578911796,
      "grad_norm": 1.1108052954539598,
      "learning_rate": 2.7603330342439686e-06,
      "loss": 0.1943,
      "step": 1095
    },
    {
      "epoch": 0.4938605384702039,
      "grad_norm": 1.0136992592552376,
      "learning_rate": 2.7566272045972777e-06,
      "loss": 0.2009,
      "step": 1096
    },
    {
      "epoch": 0.4943111411512898,
      "grad_norm": 1.0125863121371006,
      "learning_rate": 2.752920804968581e-06,
      "loss": 0.1933,
      "step": 1097
    },
    {
      "epoch": 0.4947617438323758,
      "grad_norm": 0.9430089748065448,
      "learning_rate": 2.7492138435899794e-06,
      "loss": 0.1862,
      "step": 1098
    },
    {
      "epoch": 0.49521234651346174,
      "grad_norm": 0.9233452523390339,
      "learning_rate": 2.745506328694822e-06,
      "loss": 0.1728,
      "step": 1099
    },
    {
      "epoch": 0.49566294919454773,
      "grad_norm": 0.9918009786809843,
      "learning_rate": 2.741798268517687e-06,
      "loss": 0.182,
      "step": 1100
    },
    {
      "epoch": 0.49611355187563366,
      "grad_norm": 0.9232139062649801,
      "learning_rate": 2.738089671294364e-06,
      "loss": 0.1838,
      "step": 1101
    },
    {
      "epoch": 0.4965641545567196,
      "grad_norm": 0.9492452596399719,
      "learning_rate": 2.734380545261835e-06,
      "loss": 0.1889,
      "step": 1102
    },
    {
      "epoch": 0.4970147572378056,
      "grad_norm": 0.9370507529188418,
      "learning_rate": 2.730670898658255e-06,
      "loss": 0.1852,
      "step": 1103
    },
    {
      "epoch": 0.4974653599188915,
      "grad_norm": 0.9675402055012017,
      "learning_rate": 2.726960739722939e-06,
      "loss": 0.1858,
      "step": 1104
    },
    {
      "epoch": 0.49791596259997745,
      "grad_norm": 0.9610111983594906,
      "learning_rate": 2.7232500766963373e-06,
      "loss": 0.1664,
      "step": 1105
    },
    {
      "epoch": 0.49836656528106343,
      "grad_norm": 0.9763726053294081,
      "learning_rate": 2.7195389178200194e-06,
      "loss": 0.1883,
      "step": 1106
    },
    {
      "epoch": 0.49881716796214937,
      "grad_norm": 0.8905104609381886,
      "learning_rate": 2.7158272713366573e-06,
      "loss": 0.1648,
      "step": 1107
    },
    {
      "epoch": 0.49926777064323535,
      "grad_norm": 0.9757044444911063,
      "learning_rate": 2.7121151454900048e-06,
      "loss": 0.1829,
      "step": 1108
    },
    {
      "epoch": 0.4997183733243213,
      "grad_norm": 0.9268012060057049,
      "learning_rate": 2.7084025485248827e-06,
      "loss": 0.1772,
      "step": 1109
    },
    {
      "epoch": 0.5001689760054072,
      "grad_norm": 0.96130114765164,
      "learning_rate": 2.7046894886871564e-06,
      "loss": 0.1813,
      "step": 1110
    },
    {
      "epoch": 0.5006195786864932,
      "grad_norm": 0.9083792140560981,
      "learning_rate": 2.700975974223719e-06,
      "loss": 0.1688,
      "step": 1111
    },
    {
      "epoch": 0.5010701813675791,
      "grad_norm": 0.9850647912366456,
      "learning_rate": 2.6972620133824745e-06,
      "loss": 0.1824,
      "step": 1112
    },
    {
      "epoch": 0.5015207840486651,
      "grad_norm": 0.9354381030521021,
      "learning_rate": 2.6935476144123173e-06,
      "loss": 0.1722,
      "step": 1113
    },
    {
      "epoch": 0.5019713867297511,
      "grad_norm": 0.9619636972861941,
      "learning_rate": 2.689832785563116e-06,
      "loss": 0.1833,
      "step": 1114
    },
    {
      "epoch": 0.502421989410837,
      "grad_norm": 1.0638894566317683,
      "learning_rate": 2.6861175350856937e-06,
      "loss": 0.1972,
      "step": 1115
    },
    {
      "epoch": 0.5028725920919229,
      "grad_norm": 1.0561930996704842,
      "learning_rate": 2.6824018712318084e-06,
      "loss": 0.1976,
      "step": 1116
    },
    {
      "epoch": 0.5033231947730089,
      "grad_norm": 0.9432639175030324,
      "learning_rate": 2.6786858022541385e-06,
      "loss": 0.1802,
      "step": 1117
    },
    {
      "epoch": 0.5037737974540949,
      "grad_norm": 1.0134235760379509,
      "learning_rate": 2.674969336406262e-06,
      "loss": 0.1984,
      "step": 1118
    },
    {
      "epoch": 0.5042244001351808,
      "grad_norm": 1.000797354765843,
      "learning_rate": 2.6712524819426355e-06,
      "loss": 0.2042,
      "step": 1119
    },
    {
      "epoch": 0.5046750028162668,
      "grad_norm": 0.9847284359031943,
      "learning_rate": 2.6675352471185824e-06,
      "loss": 0.1993,
      "step": 1120
    },
    {
      "epoch": 0.5051256054973527,
      "grad_norm": 0.9365176593909826,
      "learning_rate": 2.6638176401902693e-06,
      "loss": 0.1624,
      "step": 1121
    },
    {
      "epoch": 0.5055762081784386,
      "grad_norm": 0.9596545364462479,
      "learning_rate": 2.6600996694146876e-06,
      "loss": 0.1835,
      "step": 1122
    },
    {
      "epoch": 0.5060268108595246,
      "grad_norm": 0.9683771795353776,
      "learning_rate": 2.656381343049641e-06,
      "loss": 0.1707,
      "step": 1123
    },
    {
      "epoch": 0.5064774135406106,
      "grad_norm": 0.916376052612141,
      "learning_rate": 2.652662669353719e-06,
      "loss": 0.1723,
      "step": 1124
    },
    {
      "epoch": 0.5069280162216965,
      "grad_norm": 1.027462308260936,
      "learning_rate": 2.648943656586284e-06,
      "loss": 0.1814,
      "step": 1125
    },
    {
      "epoch": 0.5073786189027825,
      "grad_norm": 0.9776859126929065,
      "learning_rate": 2.6452243130074523e-06,
      "loss": 0.193,
      "step": 1126
    },
    {
      "epoch": 0.5078292215838685,
      "grad_norm": 0.9894419131466585,
      "learning_rate": 2.6415046468780726e-06,
      "loss": 0.173,
      "step": 1127
    },
    {
      "epoch": 0.5082798242649543,
      "grad_norm": 1.1271817158890742,
      "learning_rate": 2.637784666459714e-06,
      "loss": 0.1994,
      "step": 1128
    },
    {
      "epoch": 0.5087304269460403,
      "grad_norm": 1.1120470123407524,
      "learning_rate": 2.6340643800146387e-06,
      "loss": 0.2024,
      "step": 1129
    },
    {
      "epoch": 0.5091810296271263,
      "grad_norm": 0.9944937657896967,
      "learning_rate": 2.6303437958057932e-06,
      "loss": 0.1974,
      "step": 1130
    },
    {
      "epoch": 0.5096316323082122,
      "grad_norm": 1.0567641371999308,
      "learning_rate": 2.626622922096782e-06,
      "loss": 0.1935,
      "step": 1131
    },
    {
      "epoch": 0.5100822349892982,
      "grad_norm": 0.9953924398661689,
      "learning_rate": 2.622901767151855e-06,
      "loss": 0.1708,
      "step": 1132
    },
    {
      "epoch": 0.5105328376703842,
      "grad_norm": 0.9867187399054106,
      "learning_rate": 2.619180339235884e-06,
      "loss": 0.1738,
      "step": 1133
    },
    {
      "epoch": 0.5109834403514701,
      "grad_norm": 0.9951517194658875,
      "learning_rate": 2.6154586466143495e-06,
      "loss": 0.1813,
      "step": 1134
    },
    {
      "epoch": 0.511434043032556,
      "grad_norm": 1.0909408491594175,
      "learning_rate": 2.6117366975533187e-06,
      "loss": 0.1938,
      "step": 1135
    },
    {
      "epoch": 0.511884645713642,
      "grad_norm": 0.9980885855116574,
      "learning_rate": 2.60801450031943e-06,
      "loss": 0.1783,
      "step": 1136
    },
    {
      "epoch": 0.512335248394728,
      "grad_norm": 0.9713276180645412,
      "learning_rate": 2.604292063179871e-06,
      "loss": 0.187,
      "step": 1137
    },
    {
      "epoch": 0.5127858510758139,
      "grad_norm": 0.9996504568137589,
      "learning_rate": 2.600569394402363e-06,
      "loss": 0.1817,
      "step": 1138
    },
    {
      "epoch": 0.5132364537568999,
      "grad_norm": 1.079413432325811,
      "learning_rate": 2.596846502255142e-06,
      "loss": 0.1814,
      "step": 1139
    },
    {
      "epoch": 0.5136870564379858,
      "grad_norm": 0.9937278770929203,
      "learning_rate": 2.5931233950069385e-06,
      "loss": 0.1817,
      "step": 1140
    },
    {
      "epoch": 0.5141376591190717,
      "grad_norm": 0.9741045869421943,
      "learning_rate": 2.589400080926964e-06,
      "loss": 0.1811,
      "step": 1141
    },
    {
      "epoch": 0.5145882618001577,
      "grad_norm": 0.9975149063522775,
      "learning_rate": 2.585676568284886e-06,
      "loss": 0.1828,
      "step": 1142
    },
    {
      "epoch": 0.5150388644812437,
      "grad_norm": 0.942867421002799,
      "learning_rate": 2.581952865350815e-06,
      "loss": 0.1763,
      "step": 1143
    },
    {
      "epoch": 0.5154894671623296,
      "grad_norm": 0.9346474421337058,
      "learning_rate": 2.578228980395283e-06,
      "loss": 0.1752,
      "step": 1144
    },
    {
      "epoch": 0.5159400698434156,
      "grad_norm": 0.9362968071159,
      "learning_rate": 2.5745049216892286e-06,
      "loss": 0.1844,
      "step": 1145
    },
    {
      "epoch": 0.5163906725245015,
      "grad_norm": 0.9218340206816221,
      "learning_rate": 2.570780697503973e-06,
      "loss": 0.1731,
      "step": 1146
    },
    {
      "epoch": 0.5168412752055874,
      "grad_norm": 0.9672691329516838,
      "learning_rate": 2.5670563161112073e-06,
      "loss": 0.175,
      "step": 1147
    },
    {
      "epoch": 0.5172918778866734,
      "grad_norm": 1.0414538127693016,
      "learning_rate": 2.56333178578297e-06,
      "loss": 0.1918,
      "step": 1148
    },
    {
      "epoch": 0.5177424805677594,
      "grad_norm": 0.9228840524787749,
      "learning_rate": 2.5596071147916325e-06,
      "loss": 0.1855,
      "step": 1149
    },
    {
      "epoch": 0.5181930832488453,
      "grad_norm": 0.9483825423269993,
      "learning_rate": 2.555882311409878e-06,
      "loss": 0.1808,
      "step": 1150
    },
    {
      "epoch": 0.5186436859299313,
      "grad_norm": 0.943622736088532,
      "learning_rate": 2.5521573839106815e-06,
      "loss": 0.1624,
      "step": 1151
    },
    {
      "epoch": 0.5190942886110173,
      "grad_norm": 0.9228808632463539,
      "learning_rate": 2.5484323405672965e-06,
      "loss": 0.1651,
      "step": 1152
    },
    {
      "epoch": 0.5195448912921032,
      "grad_norm": 0.9450047308942011,
      "learning_rate": 2.544707189653233e-06,
      "loss": 0.1828,
      "step": 1153
    },
    {
      "epoch": 0.5199954939731891,
      "grad_norm": 1.0551519234183033,
      "learning_rate": 2.5409819394422386e-06,
      "loss": 0.1832,
      "step": 1154
    },
    {
      "epoch": 0.5204460966542751,
      "grad_norm": 0.9628349705620093,
      "learning_rate": 2.5372565982082843e-06,
      "loss": 0.1779,
      "step": 1155
    },
    {
      "epoch": 0.5208966993353611,
      "grad_norm": 0.9919441111964074,
      "learning_rate": 2.5335311742255392e-06,
      "loss": 0.176,
      "step": 1156
    },
    {
      "epoch": 0.521347302016447,
      "grad_norm": 1.000419350534717,
      "learning_rate": 2.5298056757683604e-06,
      "loss": 0.1931,
      "step": 1157
    },
    {
      "epoch": 0.521797904697533,
      "grad_norm": 0.9758752044237246,
      "learning_rate": 2.5260801111112677e-06,
      "loss": 0.1807,
      "step": 1158
    },
    {
      "epoch": 0.5222485073786189,
      "grad_norm": 1.0378540918761032,
      "learning_rate": 2.5223544885289287e-06,
      "loss": 0.1716,
      "step": 1159
    },
    {
      "epoch": 0.5226991100597048,
      "grad_norm": 0.9893420988758019,
      "learning_rate": 2.518628816296139e-06,
      "loss": 0.1878,
      "step": 1160
    },
    {
      "epoch": 0.5231497127407908,
      "grad_norm": 1.0141282352076857,
      "learning_rate": 2.5149031026878063e-06,
      "loss": 0.1768,
      "step": 1161
    },
    {
      "epoch": 0.5236003154218768,
      "grad_norm": 1.0045365971161448,
      "learning_rate": 2.5111773559789277e-06,
      "loss": 0.1845,
      "step": 1162
    },
    {
      "epoch": 0.5240509181029627,
      "grad_norm": 0.9600097817814024,
      "learning_rate": 2.5074515844445774e-06,
      "loss": 0.1658,
      "step": 1163
    },
    {
      "epoch": 0.5245015207840487,
      "grad_norm": 0.9917555687212044,
      "learning_rate": 2.50372579635988e-06,
      "loss": 0.1892,
      "step": 1164
    },
    {
      "epoch": 0.5249521234651346,
      "grad_norm": 0.932460946654997,
      "learning_rate": 2.5e-06,
      "loss": 0.1808,
      "step": 1165
    },
    {
      "epoch": 0.5254027261462205,
      "grad_norm": 0.9802496388899974,
      "learning_rate": 2.4962742036401213e-06,
      "loss": 0.1967,
      "step": 1166
    },
    {
      "epoch": 0.5258533288273065,
      "grad_norm": 0.963276152099357,
      "learning_rate": 2.4925484155554235e-06,
      "loss": 0.1861,
      "step": 1167
    },
    {
      "epoch": 0.5263039315083925,
      "grad_norm": 1.007968690345994,
      "learning_rate": 2.4888226440210723e-06,
      "loss": 0.1948,
      "step": 1168
    },
    {
      "epoch": 0.5267545341894784,
      "grad_norm": 1.0312240246625133,
      "learning_rate": 2.4850968973121945e-06,
      "loss": 0.1878,
      "step": 1169
    },
    {
      "epoch": 0.5272051368705644,
      "grad_norm": 0.969518162593365,
      "learning_rate": 2.481371183703862e-06,
      "loss": 0.1786,
      "step": 1170
    },
    {
      "epoch": 0.5276557395516503,
      "grad_norm": 0.9951428379491346,
      "learning_rate": 2.477645511471073e-06,
      "loss": 0.1922,
      "step": 1171
    },
    {
      "epoch": 0.5281063422327363,
      "grad_norm": 0.9458709445109998,
      "learning_rate": 2.473919888888733e-06,
      "loss": 0.1909,
      "step": 1172
    },
    {
      "epoch": 0.5285569449138222,
      "grad_norm": 0.9502320144205059,
      "learning_rate": 2.4701943242316405e-06,
      "loss": 0.1845,
      "step": 1173
    },
    {
      "epoch": 0.5290075475949082,
      "grad_norm": 1.0147978070284416,
      "learning_rate": 2.466468825774461e-06,
      "loss": 0.1909,
      "step": 1174
    },
    {
      "epoch": 0.5294581502759942,
      "grad_norm": 0.9538740775918518,
      "learning_rate": 2.462743401791716e-06,
      "loss": 0.1719,
      "step": 1175
    },
    {
      "epoch": 0.5299087529570801,
      "grad_norm": 1.0356645892445344,
      "learning_rate": 2.4590180605577614e-06,
      "loss": 0.1772,
      "step": 1176
    },
    {
      "epoch": 0.530359355638166,
      "grad_norm": 0.9667337701658915,
      "learning_rate": 2.4552928103467677e-06,
      "loss": 0.1851,
      "step": 1177
    },
    {
      "epoch": 0.530809958319252,
      "grad_norm": 1.0215511619289341,
      "learning_rate": 2.4515676594327035e-06,
      "loss": 0.1862,
      "step": 1178
    },
    {
      "epoch": 0.5312605610003379,
      "grad_norm": 1.080239159714636,
      "learning_rate": 2.4478426160893197e-06,
      "loss": 0.1938,
      "step": 1179
    },
    {
      "epoch": 0.5317111636814239,
      "grad_norm": 0.978810820359603,
      "learning_rate": 2.4441176885901234e-06,
      "loss": 0.1929,
      "step": 1180
    },
    {
      "epoch": 0.5321617663625099,
      "grad_norm": 1.0027349214556345,
      "learning_rate": 2.440392885208368e-06,
      "loss": 0.1827,
      "step": 1181
    },
    {
      "epoch": 0.5326123690435958,
      "grad_norm": 0.9746023894623473,
      "learning_rate": 2.436668214217031e-06,
      "loss": 0.1741,
      "step": 1182
    },
    {
      "epoch": 0.5330629717246818,
      "grad_norm": 0.9550186212617013,
      "learning_rate": 2.4329436838887936e-06,
      "loss": 0.1636,
      "step": 1183
    },
    {
      "epoch": 0.5335135744057677,
      "grad_norm": 1.052266512449798,
      "learning_rate": 2.4292193024960275e-06,
      "loss": 0.193,
      "step": 1184
    },
    {
      "epoch": 0.5339641770868536,
      "grad_norm": 1.0365594108241232,
      "learning_rate": 2.425495078310772e-06,
      "loss": 0.1916,
      "step": 1185
    },
    {
      "epoch": 0.5344147797679396,
      "grad_norm": 1.0742016996862758,
      "learning_rate": 2.4217710196047166e-06,
      "loss": 0.1945,
      "step": 1186
    },
    {
      "epoch": 0.5348653824490256,
      "grad_norm": 0.9998451715585931,
      "learning_rate": 2.4180471346491864e-06,
      "loss": 0.1799,
      "step": 1187
    },
    {
      "epoch": 0.5353159851301115,
      "grad_norm": 1.0481115869339797,
      "learning_rate": 2.414323431715115e-06,
      "loss": 0.2012,
      "step": 1188
    },
    {
      "epoch": 0.5357665878111975,
      "grad_norm": 0.9740939934965689,
      "learning_rate": 2.410599919073037e-06,
      "loss": 0.1907,
      "step": 1189
    },
    {
      "epoch": 0.5362171904922834,
      "grad_norm": 0.9782525842641908,
      "learning_rate": 2.4068766049930623e-06,
      "loss": 0.1829,
      "step": 1190
    },
    {
      "epoch": 0.5366677931733694,
      "grad_norm": 0.926823673464016,
      "learning_rate": 2.403153497744859e-06,
      "loss": 0.1731,
      "step": 1191
    },
    {
      "epoch": 0.5371183958544553,
      "grad_norm": 0.9317322933086251,
      "learning_rate": 2.3994306055976374e-06,
      "loss": 0.1766,
      "step": 1192
    },
    {
      "epoch": 0.5375689985355413,
      "grad_norm": 0.906705256516563,
      "learning_rate": 2.3957079368201293e-06,
      "loss": 0.1637,
      "step": 1193
    },
    {
      "epoch": 0.5380196012166273,
      "grad_norm": 0.9111380070644578,
      "learning_rate": 2.39198549968057e-06,
      "loss": 0.1797,
      "step": 1194
    },
    {
      "epoch": 0.5384702038977132,
      "grad_norm": 1.0235451500244883,
      "learning_rate": 2.3882633024466813e-06,
      "loss": 0.1902,
      "step": 1195
    },
    {
      "epoch": 0.5389208065787991,
      "grad_norm": 0.919670741033484,
      "learning_rate": 2.3845413533856517e-06,
      "loss": 0.1713,
      "step": 1196
    },
    {
      "epoch": 0.5393714092598851,
      "grad_norm": 1.0088308399238464,
      "learning_rate": 2.3808196607641176e-06,
      "loss": 0.194,
      "step": 1197
    },
    {
      "epoch": 0.539822011940971,
      "grad_norm": 0.92840775528469,
      "learning_rate": 2.3770982328481464e-06,
      "loss": 0.1841,
      "step": 1198
    },
    {
      "epoch": 0.540272614622057,
      "grad_norm": 0.9848018352185613,
      "learning_rate": 2.3733770779032185e-06,
      "loss": 0.1635,
      "step": 1199
    },
    {
      "epoch": 0.540723217303143,
      "grad_norm": 0.9327976302812879,
      "learning_rate": 2.3696562041942076e-06,
      "loss": 0.1804,
      "step": 1200
    },
    {
      "epoch": 0.5411738199842289,
      "grad_norm": 0.9655142952949034,
      "learning_rate": 2.3659356199853617e-06,
      "loss": 0.1825,
      "step": 1201
    },
    {
      "epoch": 0.5416244226653149,
      "grad_norm": 1.0219936814274773,
      "learning_rate": 2.362215333540287e-06,
      "loss": 0.1968,
      "step": 1202
    },
    {
      "epoch": 0.5420750253464008,
      "grad_norm": 0.8904345856244417,
      "learning_rate": 2.3584953531219278e-06,
      "loss": 0.1662,
      "step": 1203
    },
    {
      "epoch": 0.5425256280274867,
      "grad_norm": 0.9514585972491365,
      "learning_rate": 2.3547756869925485e-06,
      "loss": 0.1766,
      "step": 1204
    },
    {
      "epoch": 0.5429762307085727,
      "grad_norm": 1.0241101215031847,
      "learning_rate": 2.3510563434137175e-06,
      "loss": 0.1767,
      "step": 1205
    },
    {
      "epoch": 0.5434268333896587,
      "grad_norm": 0.9578253213796663,
      "learning_rate": 2.347337330646282e-06,
      "loss": 0.1772,
      "step": 1206
    },
    {
      "epoch": 0.5438774360707446,
      "grad_norm": 0.9817433827617003,
      "learning_rate": 2.3436186569503598e-06,
      "loss": 0.1936,
      "step": 1207
    },
    {
      "epoch": 0.5443280387518306,
      "grad_norm": 0.9864355407353985,
      "learning_rate": 2.339900330585313e-06,
      "loss": 0.1932,
      "step": 1208
    },
    {
      "epoch": 0.5447786414329165,
      "grad_norm": 0.9722800161647528,
      "learning_rate": 2.3361823598097316e-06,
      "loss": 0.1852,
      "step": 1209
    },
    {
      "epoch": 0.5452292441140025,
      "grad_norm": 1.0342230857446506,
      "learning_rate": 2.332464752881418e-06,
      "loss": 0.1955,
      "step": 1210
    },
    {
      "epoch": 0.5456798467950884,
      "grad_norm": 0.9414693753247567,
      "learning_rate": 2.3287475180573653e-06,
      "loss": 0.182,
      "step": 1211
    },
    {
      "epoch": 0.5461304494761744,
      "grad_norm": 0.8860482726242138,
      "learning_rate": 2.3250306635937385e-06,
      "loss": 0.1634,
      "step": 1212
    },
    {
      "epoch": 0.5465810521572604,
      "grad_norm": 0.9469723822144591,
      "learning_rate": 2.3213141977458615e-06,
      "loss": 0.184,
      "step": 1213
    },
    {
      "epoch": 0.5470316548383463,
      "grad_norm": 0.8631153012390324,
      "learning_rate": 2.3175981287681924e-06,
      "loss": 0.1573,
      "step": 1214
    },
    {
      "epoch": 0.5474822575194322,
      "grad_norm": 0.9624737172159589,
      "learning_rate": 2.3138824649143076e-06,
      "loss": 0.1709,
      "step": 1215
    },
    {
      "epoch": 0.5479328602005182,
      "grad_norm": 0.981564862048761,
      "learning_rate": 2.310167214436885e-06,
      "loss": 0.1905,
      "step": 1216
    },
    {
      "epoch": 0.5483834628816041,
      "grad_norm": 0.9860990048581212,
      "learning_rate": 2.306452385587683e-06,
      "loss": 0.1893,
      "step": 1217
    },
    {
      "epoch": 0.5488340655626901,
      "grad_norm": 0.9997846222056105,
      "learning_rate": 2.3027379866175263e-06,
      "loss": 0.1816,
      "step": 1218
    },
    {
      "epoch": 0.5492846682437761,
      "grad_norm": 0.9839070512185661,
      "learning_rate": 2.2990240257762817e-06,
      "loss": 0.1789,
      "step": 1219
    },
    {
      "epoch": 0.549735270924862,
      "grad_norm": 1.010734431680887,
      "learning_rate": 2.295310511312844e-06,
      "loss": 0.1811,
      "step": 1220
    },
    {
      "epoch": 0.550185873605948,
      "grad_norm": 0.9636102309687704,
      "learning_rate": 2.2915974514751173e-06,
      "loss": 0.1811,
      "step": 1221
    },
    {
      "epoch": 0.5506364762870339,
      "grad_norm": 0.9112518930451056,
      "learning_rate": 2.287884854509995e-06,
      "loss": 0.17,
      "step": 1222
    },
    {
      "epoch": 0.5510870789681198,
      "grad_norm": 0.9724041065846214,
      "learning_rate": 2.2841727286633444e-06,
      "loss": 0.1809,
      "step": 1223
    },
    {
      "epoch": 0.5515376816492058,
      "grad_norm": 0.8976637885767892,
      "learning_rate": 2.280461082179982e-06,
      "loss": 0.1531,
      "step": 1224
    },
    {
      "epoch": 0.5519882843302918,
      "grad_norm": 0.9710139357320663,
      "learning_rate": 2.2767499233036635e-06,
      "loss": 0.1715,
      "step": 1225
    },
    {
      "epoch": 0.5524388870113777,
      "grad_norm": 1.022286767055151,
      "learning_rate": 2.2730392602770617e-06,
      "loss": 0.1831,
      "step": 1226
    },
    {
      "epoch": 0.5528894896924637,
      "grad_norm": 0.9629754668562253,
      "learning_rate": 2.269329101341745e-06,
      "loss": 0.1729,
      "step": 1227
    },
    {
      "epoch": 0.5533400923735496,
      "grad_norm": 0.9665296459469704,
      "learning_rate": 2.265619454738166e-06,
      "loss": 0.1732,
      "step": 1228
    },
    {
      "epoch": 0.5537906950546356,
      "grad_norm": 1.0583946774908262,
      "learning_rate": 2.2619103287056366e-06,
      "loss": 0.1897,
      "step": 1229
    },
    {
      "epoch": 0.5542412977357215,
      "grad_norm": 1.0533800740583372,
      "learning_rate": 2.2582017314823135e-06,
      "loss": 0.1825,
      "step": 1230
    },
    {
      "epoch": 0.5546919004168075,
      "grad_norm": 0.9304386750010083,
      "learning_rate": 2.254493671305179e-06,
      "loss": 0.1746,
      "step": 1231
    },
    {
      "epoch": 0.5551425030978935,
      "grad_norm": 0.9594946709485886,
      "learning_rate": 2.250786156410022e-06,
      "loss": 0.1714,
      "step": 1232
    },
    {
      "epoch": 0.5555931057789794,
      "grad_norm": 1.0124409711112767,
      "learning_rate": 2.24707919503142e-06,
      "loss": 0.1887,
      "step": 1233
    },
    {
      "epoch": 0.5560437084600653,
      "grad_norm": 1.110284743291757,
      "learning_rate": 2.2433727954027227e-06,
      "loss": 0.1724,
      "step": 1234
    },
    {
      "epoch": 0.5564943111411513,
      "grad_norm": 0.9968654870523563,
      "learning_rate": 2.239666965756032e-06,
      "loss": 0.1838,
      "step": 1235
    },
    {
      "epoch": 0.5569449138222372,
      "grad_norm": 1.0236544356271942,
      "learning_rate": 2.2359617143221805e-06,
      "loss": 0.1798,
      "step": 1236
    },
    {
      "epoch": 0.5573955165033232,
      "grad_norm": 1.0263790451012669,
      "learning_rate": 2.232257049330722e-06,
      "loss": 0.1988,
      "step": 1237
    },
    {
      "epoch": 0.5578461191844092,
      "grad_norm": 0.9338162135659887,
      "learning_rate": 2.2285529790099034e-06,
      "loss": 0.1749,
      "step": 1238
    },
    {
      "epoch": 0.5582967218654951,
      "grad_norm": 1.0169422018470666,
      "learning_rate": 2.224849511586652e-06,
      "loss": 0.1837,
      "step": 1239
    },
    {
      "epoch": 0.558747324546581,
      "grad_norm": 1.045923343357873,
      "learning_rate": 2.221146655286558e-06,
      "loss": 0.1913,
      "step": 1240
    },
    {
      "epoch": 0.559197927227667,
      "grad_norm": 0.9631444964816668,
      "learning_rate": 2.21744441833385e-06,
      "loss": 0.1861,
      "step": 1241
    },
    {
      "epoch": 0.5596485299087529,
      "grad_norm": 0.9529062092810353,
      "learning_rate": 2.2137428089513857e-06,
      "loss": 0.1811,
      "step": 1242
    },
    {
      "epoch": 0.5600991325898389,
      "grad_norm": 1.0154141341033607,
      "learning_rate": 2.2100418353606262e-06,
      "loss": 0.1794,
      "step": 1243
    },
    {
      "epoch": 0.5605497352709249,
      "grad_norm": 0.9766059599452342,
      "learning_rate": 2.20634150578162e-06,
      "loss": 0.1819,
      "step": 1244
    },
    {
      "epoch": 0.5610003379520108,
      "grad_norm": 0.932965228493953,
      "learning_rate": 2.202641828432988e-06,
      "loss": 0.1748,
      "step": 1245
    },
    {
      "epoch": 0.5614509406330968,
      "grad_norm": 0.9702059483136485,
      "learning_rate": 2.1989428115319005e-06,
      "loss": 0.1839,
      "step": 1246
    },
    {
      "epoch": 0.5619015433141827,
      "grad_norm": 1.0785932501489395,
      "learning_rate": 2.19524446329406e-06,
      "loss": 0.187,
      "step": 1247
    },
    {
      "epoch": 0.5623521459952687,
      "grad_norm": 0.9990242640449595,
      "learning_rate": 2.1915467919336862e-06,
      "loss": 0.1842,
      "step": 1248
    },
    {
      "epoch": 0.5628027486763546,
      "grad_norm": 1.0202625100556078,
      "learning_rate": 2.1878498056634946e-06,
      "loss": 0.1795,
      "step": 1249
    },
    {
      "epoch": 0.5632533513574406,
      "grad_norm": 1.0275123282984282,
      "learning_rate": 2.1841535126946777e-06,
      "loss": 0.1832,
      "step": 1250
    },
    {
      "epoch": 0.5637039540385266,
      "grad_norm": 1.0132575380436037,
      "learning_rate": 2.180457921236889e-06,
      "loss": 0.1765,
      "step": 1251
    },
    {
      "epoch": 0.5641545567196125,
      "grad_norm": 1.003409258685893,
      "learning_rate": 2.176763039498225e-06,
      "loss": 0.1879,
      "step": 1252
    },
    {
      "epoch": 0.5646051594006984,
      "grad_norm": 0.9874249442484657,
      "learning_rate": 2.1730688756852046e-06,
      "loss": 0.195,
      "step": 1253
    },
    {
      "epoch": 0.5650557620817844,
      "grad_norm": 0.9868641897122264,
      "learning_rate": 2.1693754380027533e-06,
      "loss": 0.177,
      "step": 1254
    },
    {
      "epoch": 0.5655063647628703,
      "grad_norm": 1.0504175835437466,
      "learning_rate": 2.165682734654181e-06,
      "loss": 0.1825,
      "step": 1255
    },
    {
      "epoch": 0.5659569674439563,
      "grad_norm": 1.0619516029800902,
      "learning_rate": 2.161990773841171e-06,
      "loss": 0.1922,
      "step": 1256
    },
    {
      "epoch": 0.5664075701250423,
      "grad_norm": 1.0344653262001975,
      "learning_rate": 2.1582995637637543e-06,
      "loss": 0.1783,
      "step": 1257
    },
    {
      "epoch": 0.5668581728061282,
      "grad_norm": 0.9973482494131128,
      "learning_rate": 2.1546091126202955e-06,
      "loss": 0.1778,
      "step": 1258
    },
    {
      "epoch": 0.5673087754872141,
      "grad_norm": 0.9561514880489893,
      "learning_rate": 2.150919428607472e-06,
      "loss": 0.1726,
      "step": 1259
    },
    {
      "epoch": 0.5677593781683001,
      "grad_norm": 1.0191580883847822,
      "learning_rate": 2.147230519920259e-06,
      "loss": 0.1741,
      "step": 1260
    },
    {
      "epoch": 0.568209980849386,
      "grad_norm": 1.0366037060072417,
      "learning_rate": 2.143542394751911e-06,
      "loss": 0.1833,
      "step": 1261
    },
    {
      "epoch": 0.568660583530472,
      "grad_norm": 1.0454064471673492,
      "learning_rate": 2.139855061293939e-06,
      "loss": 0.1896,
      "step": 1262
    },
    {
      "epoch": 0.569111186211558,
      "grad_norm": 1.00899648660538,
      "learning_rate": 2.1361685277360973e-06,
      "loss": 0.1747,
      "step": 1263
    },
    {
      "epoch": 0.5695617888926439,
      "grad_norm": 1.0075600486399183,
      "learning_rate": 2.132482802266364e-06,
      "loss": 0.1892,
      "step": 1264
    },
    {
      "epoch": 0.5700123915737298,
      "grad_norm": 0.9291801587696538,
      "learning_rate": 2.128797893070922e-06,
      "loss": 0.1563,
      "step": 1265
    },
    {
      "epoch": 0.5704629942548158,
      "grad_norm": 0.9408535292316663,
      "learning_rate": 2.1251138083341404e-06,
      "loss": 0.1759,
      "step": 1266
    },
    {
      "epoch": 0.5709135969359018,
      "grad_norm": 0.9543537844054255,
      "learning_rate": 2.1214305562385592e-06,
      "loss": 0.1833,
      "step": 1267
    },
    {
      "epoch": 0.5713641996169877,
      "grad_norm": 1.0435966860876906,
      "learning_rate": 2.117748144964865e-06,
      "loss": 0.1857,
      "step": 1268
    },
    {
      "epoch": 0.5718148022980737,
      "grad_norm": 1.0360293503534796,
      "learning_rate": 2.1140665826918823e-06,
      "loss": 0.1922,
      "step": 1269
    },
    {
      "epoch": 0.5722654049791597,
      "grad_norm": 0.9475430842301149,
      "learning_rate": 2.1103858775965455e-06,
      "loss": 0.1717,
      "step": 1270
    },
    {
      "epoch": 0.5727160076602456,
      "grad_norm": 0.9873360818255287,
      "learning_rate": 2.106706037853887e-06,
      "loss": 0.1766,
      "step": 1271
    },
    {
      "epoch": 0.5731666103413315,
      "grad_norm": 0.9988576739952401,
      "learning_rate": 2.103027071637018e-06,
      "loss": 0.1917,
      "step": 1272
    },
    {
      "epoch": 0.5736172130224175,
      "grad_norm": 0.9794579225845605,
      "learning_rate": 2.099348987117108e-06,
      "loss": 0.1652,
      "step": 1273
    },
    {
      "epoch": 0.5740678157035034,
      "grad_norm": 0.9049004556537857,
      "learning_rate": 2.095671792463368e-06,
      "loss": 0.1735,
      "step": 1274
    },
    {
      "epoch": 0.5745184183845894,
      "grad_norm": 0.9161327176570143,
      "learning_rate": 2.0919954958430357e-06,
      "loss": 0.1733,
      "step": 1275
    },
    {
      "epoch": 0.5749690210656754,
      "grad_norm": 0.9504311110927137,
      "learning_rate": 2.0883201054213493e-06,
      "loss": 0.1822,
      "step": 1276
    },
    {
      "epoch": 0.5754196237467613,
      "grad_norm": 1.0141813177166452,
      "learning_rate": 2.0846456293615384e-06,
      "loss": 0.1872,
      "step": 1277
    },
    {
      "epoch": 0.5758702264278472,
      "grad_norm": 0.9481128040139777,
      "learning_rate": 2.0809720758247997e-06,
      "loss": 0.1764,
      "step": 1278
    },
    {
      "epoch": 0.5763208291089332,
      "grad_norm": 0.9956268281053715,
      "learning_rate": 2.077299452970282e-06,
      "loss": 0.1839,
      "step": 1279
    },
    {
      "epoch": 0.5767714317900191,
      "grad_norm": 0.9221100295953925,
      "learning_rate": 2.0736277689550655e-06,
      "loss": 0.1696,
      "step": 1280
    },
    {
      "epoch": 0.5772220344711051,
      "grad_norm": 0.9992405251980395,
      "learning_rate": 2.069957031934147e-06,
      "loss": 0.174,
      "step": 1281
    },
    {
      "epoch": 0.5776726371521911,
      "grad_norm": 0.975709935305381,
      "learning_rate": 2.066287250060418e-06,
      "loss": 0.1685,
      "step": 1282
    },
    {
      "epoch": 0.578123239833277,
      "grad_norm": 1.0276498478372256,
      "learning_rate": 2.062618431484652e-06,
      "loss": 0.168,
      "step": 1283
    },
    {
      "epoch": 0.578573842514363,
      "grad_norm": 0.9649717673625273,
      "learning_rate": 2.05895058435548e-06,
      "loss": 0.1735,
      "step": 1284
    },
    {
      "epoch": 0.5790244451954489,
      "grad_norm": 1.0183890478781983,
      "learning_rate": 2.0552837168193738e-06,
      "loss": 0.1793,
      "step": 1285
    },
    {
      "epoch": 0.5794750478765348,
      "grad_norm": 0.9693380530197512,
      "learning_rate": 2.051617837020633e-06,
      "loss": 0.175,
      "step": 1286
    },
    {
      "epoch": 0.5799256505576208,
      "grad_norm": 1.0035692278148098,
      "learning_rate": 2.047952953101363e-06,
      "loss": 0.1716,
      "step": 1287
    },
    {
      "epoch": 0.5803762532387068,
      "grad_norm": 0.9543283088936171,
      "learning_rate": 2.0442890732014563e-06,
      "loss": 0.1782,
      "step": 1288
    },
    {
      "epoch": 0.5808268559197928,
      "grad_norm": 0.9856491679341663,
      "learning_rate": 2.040626205458574e-06,
      "loss": 0.1926,
      "step": 1289
    },
    {
      "epoch": 0.5812774586008786,
      "grad_norm": 0.9368718149030635,
      "learning_rate": 2.0369643580081326e-06,
      "loss": 0.1745,
      "step": 1290
    },
    {
      "epoch": 0.5817280612819646,
      "grad_norm": 1.054142375462708,
      "learning_rate": 2.0333035389832795e-06,
      "loss": 0.1881,
      "step": 1291
    },
    {
      "epoch": 0.5821786639630506,
      "grad_norm": 1.0454295548530987,
      "learning_rate": 2.0296437565148786e-06,
      "loss": 0.1927,
      "step": 1292
    },
    {
      "epoch": 0.5826292666441365,
      "grad_norm": 0.9451479577628555,
      "learning_rate": 2.025985018731494e-06,
      "loss": 0.1698,
      "step": 1293
    },
    {
      "epoch": 0.5830798693252225,
      "grad_norm": 1.0436155713832889,
      "learning_rate": 2.0223273337593647e-06,
      "loss": 0.1911,
      "step": 1294
    },
    {
      "epoch": 0.5835304720063085,
      "grad_norm": 0.9725657216642605,
      "learning_rate": 2.0186707097223952e-06,
      "loss": 0.1693,
      "step": 1295
    },
    {
      "epoch": 0.5839810746873944,
      "grad_norm": 0.9906222847408853,
      "learning_rate": 2.0150151547421333e-06,
      "loss": 0.1912,
      "step": 1296
    },
    {
      "epoch": 0.5844316773684803,
      "grad_norm": 0.9044813618430019,
      "learning_rate": 2.0113606769377497e-06,
      "loss": 0.1733,
      "step": 1297
    },
    {
      "epoch": 0.5848822800495663,
      "grad_norm": 0.9667967965279892,
      "learning_rate": 2.0077072844260267e-06,
      "loss": 0.1761,
      "step": 1298
    },
    {
      "epoch": 0.5853328827306522,
      "grad_norm": 0.9802438293326249,
      "learning_rate": 2.0040549853213326e-06,
      "loss": 0.1676,
      "step": 1299
    },
    {
      "epoch": 0.5857834854117382,
      "grad_norm": 0.972410342837984,
      "learning_rate": 2.0004037877356085e-06,
      "loss": 0.1904,
      "step": 1300
    },
    {
      "epoch": 0.5862340880928242,
      "grad_norm": 0.9597522742672756,
      "learning_rate": 1.9967536997783495e-06,
      "loss": 0.1834,
      "step": 1301
    },
    {
      "epoch": 0.5866846907739101,
      "grad_norm": 0.9814692842912235,
      "learning_rate": 1.9931047295565863e-06,
      "loss": 0.1732,
      "step": 1302
    },
    {
      "epoch": 0.587135293454996,
      "grad_norm": 0.9533163314110267,
      "learning_rate": 1.989456885174865e-06,
      "loss": 0.1723,
      "step": 1303
    },
    {
      "epoch": 0.587585896136082,
      "grad_norm": 1.0013117897122807,
      "learning_rate": 1.9858101747352326e-06,
      "loss": 0.1831,
      "step": 1304
    },
    {
      "epoch": 0.5880364988171679,
      "grad_norm": 1.0237539849809265,
      "learning_rate": 1.9821646063372174e-06,
      "loss": 0.1924,
      "step": 1305
    },
    {
      "epoch": 0.5884871014982539,
      "grad_norm": 1.0430023568792837,
      "learning_rate": 1.978520188077813e-06,
      "loss": 0.1798,
      "step": 1306
    },
    {
      "epoch": 0.5889377041793399,
      "grad_norm": 0.9744176987681425,
      "learning_rate": 1.9748769280514544e-06,
      "loss": 0.1777,
      "step": 1307
    },
    {
      "epoch": 0.5893883068604259,
      "grad_norm": 0.9218200088489699,
      "learning_rate": 1.971234834350008e-06,
      "loss": 0.1732,
      "step": 1308
    },
    {
      "epoch": 0.5898389095415117,
      "grad_norm": 0.9739010233717921,
      "learning_rate": 1.967593915062748e-06,
      "loss": 0.1836,
      "step": 1309
    },
    {
      "epoch": 0.5902895122225977,
      "grad_norm": 0.9373857688708352,
      "learning_rate": 1.96395417827634e-06,
      "loss": 0.172,
      "step": 1310
    },
    {
      "epoch": 0.5907401149036837,
      "grad_norm": 1.0305890099141217,
      "learning_rate": 1.960315632074824e-06,
      "loss": 0.1891,
      "step": 1311
    },
    {
      "epoch": 0.5911907175847696,
      "grad_norm": 0.9677534466523053,
      "learning_rate": 1.9566782845395945e-06,
      "loss": 0.1771,
      "step": 1312
    },
    {
      "epoch": 0.5916413202658556,
      "grad_norm": 0.8941128931209235,
      "learning_rate": 1.9530421437493843e-06,
      "loss": 0.1695,
      "step": 1313
    },
    {
      "epoch": 0.5920919229469416,
      "grad_norm": 0.9555412056280095,
      "learning_rate": 1.949407217780247e-06,
      "loss": 0.1672,
      "step": 1314
    },
    {
      "epoch": 0.5925425256280274,
      "grad_norm": 0.9689284207826656,
      "learning_rate": 1.945773514705537e-06,
      "loss": 0.1622,
      "step": 1315
    },
    {
      "epoch": 0.5929931283091134,
      "grad_norm": 0.9930091192518737,
      "learning_rate": 1.9421410425958915e-06,
      "loss": 0.1744,
      "step": 1316
    },
    {
      "epoch": 0.5934437309901994,
      "grad_norm": 0.9963914708795829,
      "learning_rate": 1.938509809519216e-06,
      "loss": 0.1822,
      "step": 1317
    },
    {
      "epoch": 0.5938943336712853,
      "grad_norm": 1.0169006281779729,
      "learning_rate": 1.934879823540663e-06,
      "loss": 0.1743,
      "step": 1318
    },
    {
      "epoch": 0.5943449363523713,
      "grad_norm": 0.9630659155754449,
      "learning_rate": 1.931251092722615e-06,
      "loss": 0.1734,
      "step": 1319
    },
    {
      "epoch": 0.5947955390334573,
      "grad_norm": 1.038229967298548,
      "learning_rate": 1.9276236251246655e-06,
      "loss": 0.1952,
      "step": 1320
    },
    {
      "epoch": 0.5952461417145432,
      "grad_norm": 1.0316333623222642,
      "learning_rate": 1.9239974288036044e-06,
      "loss": 0.1758,
      "step": 1321
    },
    {
      "epoch": 0.5956967443956291,
      "grad_norm": 0.9971216934509304,
      "learning_rate": 1.920372511813397e-06,
      "loss": 0.1866,
      "step": 1322
    },
    {
      "epoch": 0.5961473470767151,
      "grad_norm": 0.9896930008148693,
      "learning_rate": 1.916748882205168e-06,
      "loss": 0.1864,
      "step": 1323
    },
    {
      "epoch": 0.596597949757801,
      "grad_norm": 0.9268799065373754,
      "learning_rate": 1.913126548027181e-06,
      "loss": 0.1754,
      "step": 1324
    },
    {
      "epoch": 0.597048552438887,
      "grad_norm": 0.9682132912782482,
      "learning_rate": 1.909505517324825e-06,
      "loss": 0.1792,
      "step": 1325
    },
    {
      "epoch": 0.597499155119973,
      "grad_norm": 0.9930817836872204,
      "learning_rate": 1.905885798140591e-06,
      "loss": 0.1713,
      "step": 1326
    },
    {
      "epoch": 0.597949757801059,
      "grad_norm": 0.9429726025467023,
      "learning_rate": 1.9022673985140585e-06,
      "loss": 0.1676,
      "step": 1327
    },
    {
      "epoch": 0.5984003604821448,
      "grad_norm": 0.9461399995953571,
      "learning_rate": 1.8986503264818785e-06,
      "loss": 0.1687,
      "step": 1328
    },
    {
      "epoch": 0.5988509631632308,
      "grad_norm": 0.9023598770727506,
      "learning_rate": 1.895034590077747e-06,
      "loss": 0.1723,
      "step": 1329
    },
    {
      "epoch": 0.5993015658443168,
      "grad_norm": 0.9853066110059597,
      "learning_rate": 1.8914201973324004e-06,
      "loss": 0.1651,
      "step": 1330
    },
    {
      "epoch": 0.5997521685254027,
      "grad_norm": 1.0013857184348935,
      "learning_rate": 1.8878071562735873e-06,
      "loss": 0.1876,
      "step": 1331
    },
    {
      "epoch": 0.6002027712064887,
      "grad_norm": 1.0148699485876223,
      "learning_rate": 1.8841954749260535e-06,
      "loss": 0.1723,
      "step": 1332
    },
    {
      "epoch": 0.6006533738875747,
      "grad_norm": 0.9383847419000086,
      "learning_rate": 1.8805851613115278e-06,
      "loss": 0.1698,
      "step": 1333
    },
    {
      "epoch": 0.6011039765686605,
      "grad_norm": 0.9219094194233376,
      "learning_rate": 1.8769762234486982e-06,
      "loss": 0.1752,
      "step": 1334
    },
    {
      "epoch": 0.6015545792497465,
      "grad_norm": 0.9551193816669125,
      "learning_rate": 1.8733686693531986e-06,
      "loss": 0.1876,
      "step": 1335
    },
    {
      "epoch": 0.6020051819308325,
      "grad_norm": 0.9149029805316468,
      "learning_rate": 1.8697625070375893e-06,
      "loss": 0.1771,
      "step": 1336
    },
    {
      "epoch": 0.6024557846119184,
      "grad_norm": 0.965925307004654,
      "learning_rate": 1.8661577445113399e-06,
      "loss": 0.1779,
      "step": 1337
    },
    {
      "epoch": 0.6029063872930044,
      "grad_norm": 0.9551150150198514,
      "learning_rate": 1.8625543897808094e-06,
      "loss": 0.181,
      "step": 1338
    },
    {
      "epoch": 0.6033569899740904,
      "grad_norm": 0.898041242799122,
      "learning_rate": 1.8589524508492308e-06,
      "loss": 0.1739,
      "step": 1339
    },
    {
      "epoch": 0.6038075926551763,
      "grad_norm": 0.9870301856111156,
      "learning_rate": 1.8553519357166927e-06,
      "loss": 0.1762,
      "step": 1340
    },
    {
      "epoch": 0.6042581953362622,
      "grad_norm": 0.9238496565541336,
      "learning_rate": 1.8517528523801226e-06,
      "loss": 0.1631,
      "step": 1341
    },
    {
      "epoch": 0.6047087980173482,
      "grad_norm": 0.9021602741648262,
      "learning_rate": 1.8481552088332656e-06,
      "loss": 0.1776,
      "step": 1342
    },
    {
      "epoch": 0.6051594006984341,
      "grad_norm": 0.9457868006313538,
      "learning_rate": 1.84455901306667e-06,
      "loss": 0.1815,
      "step": 1343
    },
    {
      "epoch": 0.6056100033795201,
      "grad_norm": 0.9644972060860808,
      "learning_rate": 1.8409642730676692e-06,
      "loss": 0.1953,
      "step": 1344
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 0.9569847068003844,
      "learning_rate": 1.8373709968203624e-06,
      "loss": 0.1861,
      "step": 1345
    },
    {
      "epoch": 0.6065112087416921,
      "grad_norm": 1.0258866036102268,
      "learning_rate": 1.8337791923055983e-06,
      "loss": 0.1874,
      "step": 1346
    },
    {
      "epoch": 0.6069618114227779,
      "grad_norm": 0.9898107412694851,
      "learning_rate": 1.8301888675009554e-06,
      "loss": 0.1753,
      "step": 1347
    },
    {
      "epoch": 0.6074124141038639,
      "grad_norm": 0.9746476143849012,
      "learning_rate": 1.8266000303807272e-06,
      "loss": 0.1873,
      "step": 1348
    },
    {
      "epoch": 0.6078630167849499,
      "grad_norm": 0.9905149297692676,
      "learning_rate": 1.8230126889159027e-06,
      "loss": 0.1897,
      "step": 1349
    },
    {
      "epoch": 0.6083136194660358,
      "grad_norm": 0.9566033353505723,
      "learning_rate": 1.8194268510741493e-06,
      "loss": 0.1896,
      "step": 1350
    },
    {
      "epoch": 0.6087642221471218,
      "grad_norm": 0.9952340005510767,
      "learning_rate": 1.8158425248197931e-06,
      "loss": 0.1772,
      "step": 1351
    },
    {
      "epoch": 0.6092148248282078,
      "grad_norm": 0.9738307656928912,
      "learning_rate": 1.812259718113805e-06,
      "loss": 0.1736,
      "step": 1352
    },
    {
      "epoch": 0.6096654275092936,
      "grad_norm": 0.9621906341557562,
      "learning_rate": 1.8086784389137796e-06,
      "loss": 0.1754,
      "step": 1353
    },
    {
      "epoch": 0.6101160301903796,
      "grad_norm": 0.891429745139326,
      "learning_rate": 1.8050986951739201e-06,
      "loss": 0.1486,
      "step": 1354
    },
    {
      "epoch": 0.6105666328714656,
      "grad_norm": 1.0068295759803783,
      "learning_rate": 1.8015204948450166e-06,
      "loss": 0.1941,
      "step": 1355
    },
    {
      "epoch": 0.6110172355525515,
      "grad_norm": 0.9585690106550426,
      "learning_rate": 1.7979438458744343e-06,
      "loss": 0.1841,
      "step": 1356
    },
    {
      "epoch": 0.6114678382336375,
      "grad_norm": 0.9641696373258147,
      "learning_rate": 1.7943687562060919e-06,
      "loss": 0.1789,
      "step": 1357
    },
    {
      "epoch": 0.6119184409147235,
      "grad_norm": 1.0316316029702115,
      "learning_rate": 1.7907952337804429e-06,
      "loss": 0.1873,
      "step": 1358
    },
    {
      "epoch": 0.6123690435958093,
      "grad_norm": 0.9616240258140892,
      "learning_rate": 1.787223286534463e-06,
      "loss": 0.1641,
      "step": 1359
    },
    {
      "epoch": 0.6128196462768953,
      "grad_norm": 0.9791093566130695,
      "learning_rate": 1.783652922401627e-06,
      "loss": 0.1685,
      "step": 1360
    },
    {
      "epoch": 0.6132702489579813,
      "grad_norm": 0.9810794464604121,
      "learning_rate": 1.7800841493118942e-06,
      "loss": 0.1765,
      "step": 1361
    },
    {
      "epoch": 0.6137208516390672,
      "grad_norm": 0.9590742488463256,
      "learning_rate": 1.776516975191691e-06,
      "loss": 0.1679,
      "step": 1362
    },
    {
      "epoch": 0.6141714543201532,
      "grad_norm": 0.988237012807688,
      "learning_rate": 1.7729514079638915e-06,
      "loss": 0.1876,
      "step": 1363
    },
    {
      "epoch": 0.6146220570012392,
      "grad_norm": 1.0202288172318874,
      "learning_rate": 1.7693874555477996e-06,
      "loss": 0.1981,
      "step": 1364
    },
    {
      "epoch": 0.6150726596823252,
      "grad_norm": 0.9592437603221992,
      "learning_rate": 1.7658251258591352e-06,
      "loss": 0.1809,
      "step": 1365
    },
    {
      "epoch": 0.615523262363411,
      "grad_norm": 0.9248493524713802,
      "learning_rate": 1.7622644268100116e-06,
      "loss": 0.1781,
      "step": 1366
    },
    {
      "epoch": 0.615973865044497,
      "grad_norm": 1.0127059943444199,
      "learning_rate": 1.7587053663089233e-06,
      "loss": 0.1826,
      "step": 1367
    },
    {
      "epoch": 0.616424467725583,
      "grad_norm": 1.0484692118625194,
      "learning_rate": 1.755147952260722e-06,
      "loss": 0.1981,
      "step": 1368
    },
    {
      "epoch": 0.6168750704066689,
      "grad_norm": 1.0462986733395399,
      "learning_rate": 1.7515921925666053e-06,
      "loss": 0.189,
      "step": 1369
    },
    {
      "epoch": 0.6173256730877549,
      "grad_norm": 0.9569914429667652,
      "learning_rate": 1.748038095124095e-06,
      "loss": 0.1797,
      "step": 1370
    },
    {
      "epoch": 0.6177762757688409,
      "grad_norm": 0.8934663502251415,
      "learning_rate": 1.7444856678270218e-06,
      "loss": 0.167,
      "step": 1371
    },
    {
      "epoch": 0.6182268784499267,
      "grad_norm": 0.9411554465371383,
      "learning_rate": 1.7409349185655067e-06,
      "loss": 0.1857,
      "step": 1372
    },
    {
      "epoch": 0.6186774811310127,
      "grad_norm": 0.9232735692254691,
      "learning_rate": 1.7373858552259421e-06,
      "loss": 0.1827,
      "step": 1373
    },
    {
      "epoch": 0.6191280838120987,
      "grad_norm": 0.9723262002837848,
      "learning_rate": 1.733838485690978e-06,
      "loss": 0.1818,
      "step": 1374
    },
    {
      "epoch": 0.6195786864931846,
      "grad_norm": 0.9639076957534941,
      "learning_rate": 1.7302928178395018e-06,
      "loss": 0.1914,
      "step": 1375
    },
    {
      "epoch": 0.6200292891742706,
      "grad_norm": 0.9430657640039586,
      "learning_rate": 1.726748859546621e-06,
      "loss": 0.1733,
      "step": 1376
    },
    {
      "epoch": 0.6204798918553566,
      "grad_norm": 0.9399938223372823,
      "learning_rate": 1.723206618683646e-06,
      "loss": 0.1815,
      "step": 1377
    },
    {
      "epoch": 0.6209304945364424,
      "grad_norm": 0.950436274242313,
      "learning_rate": 1.7196661031180738e-06,
      "loss": 0.1771,
      "step": 1378
    },
    {
      "epoch": 0.6213810972175284,
      "grad_norm": 0.9333762226180593,
      "learning_rate": 1.716127320713568e-06,
      "loss": 0.1779,
      "step": 1379
    },
    {
      "epoch": 0.6218316998986144,
      "grad_norm": 0.9819914710831309,
      "learning_rate": 1.7125902793299434e-06,
      "loss": 0.1833,
      "step": 1380
    },
    {
      "epoch": 0.6222823025797003,
      "grad_norm": 0.9524398256739168,
      "learning_rate": 1.7090549868231492e-06,
      "loss": 0.1712,
      "step": 1381
    },
    {
      "epoch": 0.6227329052607863,
      "grad_norm": 0.9914782024254578,
      "learning_rate": 1.7055214510452462e-06,
      "loss": 0.1699,
      "step": 1382
    },
    {
      "epoch": 0.6231835079418723,
      "grad_norm": 0.9496482485318583,
      "learning_rate": 1.7019896798443984e-06,
      "loss": 0.1608,
      "step": 1383
    },
    {
      "epoch": 0.6236341106229583,
      "grad_norm": 1.001098465203453,
      "learning_rate": 1.6984596810648475e-06,
      "loss": 0.1825,
      "step": 1384
    },
    {
      "epoch": 0.6240847133040441,
      "grad_norm": 0.9402250107639619,
      "learning_rate": 1.6949314625468985e-06,
      "loss": 0.1608,
      "step": 1385
    },
    {
      "epoch": 0.6245353159851301,
      "grad_norm": 1.0011483352758226,
      "learning_rate": 1.6914050321269049e-06,
      "loss": 0.1806,
      "step": 1386
    },
    {
      "epoch": 0.6249859186662161,
      "grad_norm": 1.0503798519563663,
      "learning_rate": 1.6878803976372465e-06,
      "loss": 0.1964,
      "step": 1387
    },
    {
      "epoch": 0.625436521347302,
      "grad_norm": 0.9440187921912822,
      "learning_rate": 1.6843575669063142e-06,
      "loss": 0.1738,
      "step": 1388
    },
    {
      "epoch": 0.625887124028388,
      "grad_norm": 0.9326821497482316,
      "learning_rate": 1.6808365477584953e-06,
      "loss": 0.1648,
      "step": 1389
    },
    {
      "epoch": 0.626337726709474,
      "grad_norm": 0.9424691557698677,
      "learning_rate": 1.6773173480141487e-06,
      "loss": 0.1652,
      "step": 1390
    },
    {
      "epoch": 0.6267883293905598,
      "grad_norm": 0.9281771608858566,
      "learning_rate": 1.6737999754895965e-06,
      "loss": 0.1777,
      "step": 1391
    },
    {
      "epoch": 0.6272389320716458,
      "grad_norm": 0.9752991219227313,
      "learning_rate": 1.6702844379971012e-06,
      "loss": 0.1697,
      "step": 1392
    },
    {
      "epoch": 0.6276895347527318,
      "grad_norm": 0.960553601801517,
      "learning_rate": 1.6667707433448482e-06,
      "loss": 0.1752,
      "step": 1393
    },
    {
      "epoch": 0.6281401374338177,
      "grad_norm": 0.9489921824793127,
      "learning_rate": 1.663258899336933e-06,
      "loss": 0.1745,
      "step": 1394
    },
    {
      "epoch": 0.6285907401149037,
      "grad_norm": 0.9941610654543575,
      "learning_rate": 1.6597489137733377e-06,
      "loss": 0.1692,
      "step": 1395
    },
    {
      "epoch": 0.6290413427959897,
      "grad_norm": 0.9797241069847795,
      "learning_rate": 1.6562407944499175e-06,
      "loss": 0.1707,
      "step": 1396
    },
    {
      "epoch": 0.6294919454770755,
      "grad_norm": 0.9293994462914644,
      "learning_rate": 1.652734549158384e-06,
      "loss": 0.1665,
      "step": 1397
    },
    {
      "epoch": 0.6299425481581615,
      "grad_norm": 1.03038629799781,
      "learning_rate": 1.6492301856862855e-06,
      "loss": 0.1695,
      "step": 1398
    },
    {
      "epoch": 0.6303931508392475,
      "grad_norm": 0.9622243992637415,
      "learning_rate": 1.6457277118169893e-06,
      "loss": 0.1781,
      "step": 1399
    },
    {
      "epoch": 0.6308437535203334,
      "grad_norm": 1.002772156035792,
      "learning_rate": 1.6422271353296675e-06,
      "loss": 0.1891,
      "step": 1400
    },
    {
      "epoch": 0.6312943562014194,
      "grad_norm": 1.0526800841741413,
      "learning_rate": 1.6387284639992773e-06,
      "loss": 0.1777,
      "step": 1401
    },
    {
      "epoch": 0.6317449588825054,
      "grad_norm": 1.0387407303203702,
      "learning_rate": 1.6352317055965458e-06,
      "loss": 0.1875,
      "step": 1402
    },
    {
      "epoch": 0.6321955615635914,
      "grad_norm": 1.0162282724569995,
      "learning_rate": 1.6317368678879497e-06,
      "loss": 0.1841,
      "step": 1403
    },
    {
      "epoch": 0.6326461642446772,
      "grad_norm": 0.9823874974583439,
      "learning_rate": 1.6282439586356999e-06,
      "loss": 0.1769,
      "step": 1404
    },
    {
      "epoch": 0.6330967669257632,
      "grad_norm": 0.9863386832219366,
      "learning_rate": 1.6247529855977256e-06,
      "loss": 0.1704,
      "step": 1405
    },
    {
      "epoch": 0.6335473696068492,
      "grad_norm": 1.0528860491393666,
      "learning_rate": 1.6212639565276538e-06,
      "loss": 0.174,
      "step": 1406
    },
    {
      "epoch": 0.6339979722879351,
      "grad_norm": 0.9737102543598695,
      "learning_rate": 1.6177768791747957e-06,
      "loss": 0.184,
      "step": 1407
    },
    {
      "epoch": 0.6344485749690211,
      "grad_norm": 1.0153204654045858,
      "learning_rate": 1.6142917612841252e-06,
      "loss": 0.1849,
      "step": 1408
    },
    {
      "epoch": 0.6348991776501071,
      "grad_norm": 0.8991885233662639,
      "learning_rate": 1.610808610596265e-06,
      "loss": 0.1632,
      "step": 1409
    },
    {
      "epoch": 0.6353497803311929,
      "grad_norm": 0.9713420971163291,
      "learning_rate": 1.607327434847471e-06,
      "loss": 0.173,
      "step": 1410
    },
    {
      "epoch": 0.6358003830122789,
      "grad_norm": 0.9842018617155793,
      "learning_rate": 1.6038482417696095e-06,
      "loss": 0.1681,
      "step": 1411
    },
    {
      "epoch": 0.6362509856933649,
      "grad_norm": 1.0444753107781046,
      "learning_rate": 1.6003710390901434e-06,
      "loss": 0.2003,
      "step": 1412
    },
    {
      "epoch": 0.6367015883744508,
      "grad_norm": 1.009201486730144,
      "learning_rate": 1.5968958345321178e-06,
      "loss": 0.1866,
      "step": 1413
    },
    {
      "epoch": 0.6371521910555368,
      "grad_norm": 0.9793931412926407,
      "learning_rate": 1.5934226358141368e-06,
      "loss": 0.1865,
      "step": 1414
    },
    {
      "epoch": 0.6376027937366228,
      "grad_norm": 1.0086558938363293,
      "learning_rate": 1.5899514506503499e-06,
      "loss": 0.188,
      "step": 1415
    },
    {
      "epoch": 0.6380533964177086,
      "grad_norm": 1.012525844933204,
      "learning_rate": 1.5864822867504376e-06,
      "loss": 0.1644,
      "step": 1416
    },
    {
      "epoch": 0.6385039990987946,
      "grad_norm": 0.9497102281955796,
      "learning_rate": 1.5830151518195846e-06,
      "loss": 0.1638,
      "step": 1417
    },
    {
      "epoch": 0.6389546017798806,
      "grad_norm": 1.0713360411193678,
      "learning_rate": 1.5795500535584758e-06,
      "loss": 0.1864,
      "step": 1418
    },
    {
      "epoch": 0.6394052044609665,
      "grad_norm": 0.9669969719861,
      "learning_rate": 1.5760869996632685e-06,
      "loss": 0.1667,
      "step": 1419
    },
    {
      "epoch": 0.6398558071420525,
      "grad_norm": 0.9016232986127477,
      "learning_rate": 1.572625997825581e-06,
      "loss": 0.1694,
      "step": 1420
    },
    {
      "epoch": 0.6403064098231385,
      "grad_norm": 1.0014722298530099,
      "learning_rate": 1.5691670557324734e-06,
      "loss": 0.1896,
      "step": 1421
    },
    {
      "epoch": 0.6407570125042245,
      "grad_norm": 0.9885424898642345,
      "learning_rate": 1.5657101810664314e-06,
      "loss": 0.1759,
      "step": 1422
    },
    {
      "epoch": 0.6412076151853103,
      "grad_norm": 1.0163722000157043,
      "learning_rate": 1.5622553815053476e-06,
      "loss": 0.1814,
      "step": 1423
    },
    {
      "epoch": 0.6416582178663963,
      "grad_norm": 1.0470342847013532,
      "learning_rate": 1.558802664722508e-06,
      "loss": 0.1803,
      "step": 1424
    },
    {
      "epoch": 0.6421088205474823,
      "grad_norm": 0.9986140861027836,
      "learning_rate": 1.555352038386571e-06,
      "loss": 0.166,
      "step": 1425
    },
    {
      "epoch": 0.6425594232285682,
      "grad_norm": 0.962847713888351,
      "learning_rate": 1.5519035101615518e-06,
      "loss": 0.186,
      "step": 1426
    },
    {
      "epoch": 0.6430100259096542,
      "grad_norm": 0.93191438242337,
      "learning_rate": 1.5484570877068055e-06,
      "loss": 0.1728,
      "step": 1427
    },
    {
      "epoch": 0.6434606285907402,
      "grad_norm": 0.9810361262173285,
      "learning_rate": 1.5450127786770116e-06,
      "loss": 0.1704,
      "step": 1428
    },
    {
      "epoch": 0.643911231271826,
      "grad_norm": 0.9952257906091154,
      "learning_rate": 1.5415705907221545e-06,
      "loss": 0.178,
      "step": 1429
    },
    {
      "epoch": 0.644361833952912,
      "grad_norm": 0.9429605864762918,
      "learning_rate": 1.5381305314875084e-06,
      "loss": 0.1738,
      "step": 1430
    },
    {
      "epoch": 0.644812436633998,
      "grad_norm": 1.0299359367615164,
      "learning_rate": 1.5346926086136171e-06,
      "loss": 0.1779,
      "step": 1431
    },
    {
      "epoch": 0.6452630393150839,
      "grad_norm": 1.0049274983457543,
      "learning_rate": 1.5312568297362834e-06,
      "loss": 0.1746,
      "step": 1432
    },
    {
      "epoch": 0.6457136419961699,
      "grad_norm": 0.9967559026461587,
      "learning_rate": 1.5278232024865458e-06,
      "loss": 0.1818,
      "step": 1433
    },
    {
      "epoch": 0.6461642446772559,
      "grad_norm": 0.97845506443779,
      "learning_rate": 1.5243917344906625e-06,
      "loss": 0.1889,
      "step": 1434
    },
    {
      "epoch": 0.6466148473583417,
      "grad_norm": 0.9698446326236689,
      "learning_rate": 1.5209624333700985e-06,
      "loss": 0.1698,
      "step": 1435
    },
    {
      "epoch": 0.6470654500394277,
      "grad_norm": 0.9755719784670406,
      "learning_rate": 1.517535306741505e-06,
      "loss": 0.1772,
      "step": 1436
    },
    {
      "epoch": 0.6475160527205137,
      "grad_norm": 0.9745589060736123,
      "learning_rate": 1.5141103622167042e-06,
      "loss": 0.1709,
      "step": 1437
    },
    {
      "epoch": 0.6479666554015996,
      "grad_norm": 0.9965046320724716,
      "learning_rate": 1.51068760740267e-06,
      "loss": 0.1813,
      "step": 1438
    },
    {
      "epoch": 0.6484172580826856,
      "grad_norm": 0.9464651783958671,
      "learning_rate": 1.5072670499015151e-06,
      "loss": 0.177,
      "step": 1439
    },
    {
      "epoch": 0.6488678607637716,
      "grad_norm": 0.9722466387532496,
      "learning_rate": 1.5038486973104704e-06,
      "loss": 0.1521,
      "step": 1440
    },
    {
      "epoch": 0.6493184634448576,
      "grad_norm": 1.012247646444772,
      "learning_rate": 1.5004325572218698e-06,
      "loss": 0.1747,
      "step": 1441
    },
    {
      "epoch": 0.6497690661259434,
      "grad_norm": 0.9999964959383242,
      "learning_rate": 1.4970186372231347e-06,
      "loss": 0.1601,
      "step": 1442
    },
    {
      "epoch": 0.6502196688070294,
      "grad_norm": 0.9028737511479311,
      "learning_rate": 1.493606944896751e-06,
      "loss": 0.1688,
      "step": 1443
    },
    {
      "epoch": 0.6506702714881154,
      "grad_norm": 1.0231645677089356,
      "learning_rate": 1.490197487820263e-06,
      "loss": 0.185,
      "step": 1444
    },
    {
      "epoch": 0.6511208741692013,
      "grad_norm": 0.9358664932252706,
      "learning_rate": 1.486790273566246e-06,
      "loss": 0.1744,
      "step": 1445
    },
    {
      "epoch": 0.6515714768502873,
      "grad_norm": 0.9300778756974287,
      "learning_rate": 1.483385309702295e-06,
      "loss": 0.1703,
      "step": 1446
    },
    {
      "epoch": 0.6520220795313733,
      "grad_norm": 0.9169957045442526,
      "learning_rate": 1.4799826037910082e-06,
      "loss": 0.1662,
      "step": 1447
    },
    {
      "epoch": 0.6524726822124591,
      "grad_norm": 1.0311445049370658,
      "learning_rate": 1.4765821633899663e-06,
      "loss": 0.1766,
      "step": 1448
    },
    {
      "epoch": 0.6529232848935451,
      "grad_norm": 1.0370975219569198,
      "learning_rate": 1.4731839960517202e-06,
      "loss": 0.1952,
      "step": 1449
    },
    {
      "epoch": 0.6533738875746311,
      "grad_norm": 1.0233401948492444,
      "learning_rate": 1.4697881093237714e-06,
      "loss": 0.1923,
      "step": 1450
    },
    {
      "epoch": 0.653824490255717,
      "grad_norm": 0.9461622243859847,
      "learning_rate": 1.4663945107485567e-06,
      "loss": 0.1782,
      "step": 1451
    },
    {
      "epoch": 0.654275092936803,
      "grad_norm": 1.055504555070659,
      "learning_rate": 1.4630032078634293e-06,
      "loss": 0.1815,
      "step": 1452
    },
    {
      "epoch": 0.654725695617889,
      "grad_norm": 0.9234637697624012,
      "learning_rate": 1.4596142082006448e-06,
      "loss": 0.1752,
      "step": 1453
    },
    {
      "epoch": 0.6551762982989748,
      "grad_norm": 0.9507817258448786,
      "learning_rate": 1.456227519287343e-06,
      "loss": 0.1827,
      "step": 1454
    },
    {
      "epoch": 0.6556269009800608,
      "grad_norm": 0.9759081998164142,
      "learning_rate": 1.4528431486455311e-06,
      "loss": 0.1773,
      "step": 1455
    },
    {
      "epoch": 0.6560775036611468,
      "grad_norm": 0.9777184016717716,
      "learning_rate": 1.4494611037920667e-06,
      "loss": 0.1747,
      "step": 1456
    },
    {
      "epoch": 0.6565281063422327,
      "grad_norm": 0.9271892542931569,
      "learning_rate": 1.4460813922386446e-06,
      "loss": 0.1652,
      "step": 1457
    },
    {
      "epoch": 0.6569787090233187,
      "grad_norm": 0.9337910757457696,
      "learning_rate": 1.4427040214917742e-06,
      "loss": 0.1689,
      "step": 1458
    },
    {
      "epoch": 0.6574293117044047,
      "grad_norm": 0.9282004178300622,
      "learning_rate": 1.4393289990527665e-06,
      "loss": 0.1751,
      "step": 1459
    },
    {
      "epoch": 0.6578799143854905,
      "grad_norm": 0.8970983420937423,
      "learning_rate": 1.4359563324177176e-06,
      "loss": 0.1638,
      "step": 1460
    },
    {
      "epoch": 0.6583305170665765,
      "grad_norm": 1.0074464827163818,
      "learning_rate": 1.43258602907749e-06,
      "loss": 0.1826,
      "step": 1461
    },
    {
      "epoch": 0.6587811197476625,
      "grad_norm": 0.990811575091218,
      "learning_rate": 1.429218096517699e-06,
      "loss": 0.1814,
      "step": 1462
    },
    {
      "epoch": 0.6592317224287485,
      "grad_norm": 0.9480008770331717,
      "learning_rate": 1.425852542218692e-06,
      "loss": 0.1772,
      "step": 1463
    },
    {
      "epoch": 0.6596823251098344,
      "grad_norm": 0.9752983222946117,
      "learning_rate": 1.4224893736555364e-06,
      "loss": 0.1749,
      "step": 1464
    },
    {
      "epoch": 0.6601329277909204,
      "grad_norm": 0.91785989621098,
      "learning_rate": 1.4191285982979992e-06,
      "loss": 0.1677,
      "step": 1465
    },
    {
      "epoch": 0.6605835304720064,
      "grad_norm": 0.9753765831245645,
      "learning_rate": 1.4157702236105326e-06,
      "loss": 0.1647,
      "step": 1466
    },
    {
      "epoch": 0.6610341331530922,
      "grad_norm": 1.001355344074237,
      "learning_rate": 1.412414257052256e-06,
      "loss": 0.1877,
      "step": 1467
    },
    {
      "epoch": 0.6614847358341782,
      "grad_norm": 0.9502520083816269,
      "learning_rate": 1.4090607060769423e-06,
      "loss": 0.1702,
      "step": 1468
    },
    {
      "epoch": 0.6619353385152642,
      "grad_norm": 0.9227004266546573,
      "learning_rate": 1.4057095781329983e-06,
      "loss": 0.1779,
      "step": 1469
    },
    {
      "epoch": 0.6623859411963501,
      "grad_norm": 1.0407222566367818,
      "learning_rate": 1.402360880663447e-06,
      "loss": 0.1956,
      "step": 1470
    },
    {
      "epoch": 0.6628365438774361,
      "grad_norm": 0.9579131430837633,
      "learning_rate": 1.3990146211059141e-06,
      "loss": 0.1688,
      "step": 1471
    },
    {
      "epoch": 0.6632871465585221,
      "grad_norm": 0.9511338591956713,
      "learning_rate": 1.3956708068926141e-06,
      "loss": 0.1818,
      "step": 1472
    },
    {
      "epoch": 0.6637377492396079,
      "grad_norm": 0.8779448931101682,
      "learning_rate": 1.3923294454503263e-06,
      "loss": 0.1546,
      "step": 1473
    },
    {
      "epoch": 0.6641883519206939,
      "grad_norm": 0.9808281780180107,
      "learning_rate": 1.3889905442003836e-06,
      "loss": 0.1874,
      "step": 1474
    },
    {
      "epoch": 0.6646389546017799,
      "grad_norm": 0.9098712770252587,
      "learning_rate": 1.3856541105586545e-06,
      "loss": 0.1512,
      "step": 1475
    },
    {
      "epoch": 0.6650895572828658,
      "grad_norm": 0.9049853200687668,
      "learning_rate": 1.382320151935527e-06,
      "loss": 0.1561,
      "step": 1476
    },
    {
      "epoch": 0.6655401599639518,
      "grad_norm": 0.9485229447896064,
      "learning_rate": 1.3789886757358916e-06,
      "loss": 0.1728,
      "step": 1477
    },
    {
      "epoch": 0.6659907626450378,
      "grad_norm": 0.9490990019665922,
      "learning_rate": 1.375659689359126e-06,
      "loss": 0.1599,
      "step": 1478
    },
    {
      "epoch": 0.6664413653261236,
      "grad_norm": 0.9711185150006841,
      "learning_rate": 1.3723332001990774e-06,
      "loss": 0.1786,
      "step": 1479
    },
    {
      "epoch": 0.6668919680072096,
      "grad_norm": 0.9973765149760946,
      "learning_rate": 1.369009215644046e-06,
      "loss": 0.1818,
      "step": 1480
    },
    {
      "epoch": 0.6673425706882956,
      "grad_norm": 0.9644391158864567,
      "learning_rate": 1.36568774307677e-06,
      "loss": 0.18,
      "step": 1481
    },
    {
      "epoch": 0.6677931733693816,
      "grad_norm": 0.8888734345308115,
      "learning_rate": 1.362368789874407e-06,
      "loss": 0.1647,
      "step": 1482
    },
    {
      "epoch": 0.6682437760504675,
      "grad_norm": 0.9481091827652459,
      "learning_rate": 1.3590523634085218e-06,
      "loss": 0.1747,
      "step": 1483
    },
    {
      "epoch": 0.6686943787315535,
      "grad_norm": 1.0330690689616744,
      "learning_rate": 1.3557384710450644e-06,
      "loss": 0.1889,
      "step": 1484
    },
    {
      "epoch": 0.6691449814126395,
      "grad_norm": 0.960954850888663,
      "learning_rate": 1.3524271201443578e-06,
      "loss": 0.1719,
      "step": 1485
    },
    {
      "epoch": 0.6695955840937253,
      "grad_norm": 1.0055793871460201,
      "learning_rate": 1.3491183180610807e-06,
      "loss": 0.1871,
      "step": 1486
    },
    {
      "epoch": 0.6700461867748113,
      "grad_norm": 1.0545932802846039,
      "learning_rate": 1.3458120721442464e-06,
      "loss": 0.1799,
      "step": 1487
    },
    {
      "epoch": 0.6704967894558973,
      "grad_norm": 0.962979895049587,
      "learning_rate": 1.3425083897371983e-06,
      "loss": 0.1765,
      "step": 1488
    },
    {
      "epoch": 0.6709473921369832,
      "grad_norm": 0.9431647526868862,
      "learning_rate": 1.3392072781775806e-06,
      "loss": 0.167,
      "step": 1489
    },
    {
      "epoch": 0.6713979948180692,
      "grad_norm": 0.9559904240284464,
      "learning_rate": 1.335908744797329e-06,
      "loss": 0.1769,
      "step": 1490
    },
    {
      "epoch": 0.6718485974991552,
      "grad_norm": 1.048775633545965,
      "learning_rate": 1.3326127969226535e-06,
      "loss": 0.1846,
      "step": 1491
    },
    {
      "epoch": 0.672299200180241,
      "grad_norm": 0.9744426542397633,
      "learning_rate": 1.3293194418740207e-06,
      "loss": 0.1817,
      "step": 1492
    },
    {
      "epoch": 0.672749802861327,
      "grad_norm": 0.9623650677514825,
      "learning_rate": 1.3260286869661378e-06,
      "loss": 0.17,
      "step": 1493
    },
    {
      "epoch": 0.673200405542413,
      "grad_norm": 0.9659168187588352,
      "learning_rate": 1.32274053950794e-06,
      "loss": 0.1821,
      "step": 1494
    },
    {
      "epoch": 0.6736510082234989,
      "grad_norm": 0.9368891207747331,
      "learning_rate": 1.3194550068025697e-06,
      "loss": 0.1672,
      "step": 1495
    },
    {
      "epoch": 0.6741016109045849,
      "grad_norm": 0.938184877171042,
      "learning_rate": 1.3161720961473583e-06,
      "loss": 0.1706,
      "step": 1496
    },
    {
      "epoch": 0.6745522135856709,
      "grad_norm": 0.9467721060732873,
      "learning_rate": 1.3128918148338183e-06,
      "loss": 0.177,
      "step": 1497
    },
    {
      "epoch": 0.6750028162667567,
      "grad_norm": 0.9552196388275247,
      "learning_rate": 1.3096141701476189e-06,
      "loss": 0.1658,
      "step": 1498
    },
    {
      "epoch": 0.6754534189478427,
      "grad_norm": 0.9321703673762389,
      "learning_rate": 1.3063391693685773e-06,
      "loss": 0.1758,
      "step": 1499
    },
    {
      "epoch": 0.6759040216289287,
      "grad_norm": 0.967964781154864,
      "learning_rate": 1.3030668197706347e-06,
      "loss": 0.1809,
      "step": 1500
    },
    {
      "epoch": 0.6759040216289287,
      "eval_loss": 0.1767047941684723,
      "eval_runtime": 59.4127,
      "eval_samples_per_second": 24.153,
      "eval_steps_per_second": 3.03,
      "step": 1500
    },
    {
      "epoch": 0.6763546243100147,
      "grad_norm": 0.9101930612729155,
      "learning_rate": 1.2997971286218448e-06,
      "loss": 0.1727,
      "step": 1501
    },
    {
      "epoch": 0.6768052269911006,
      "grad_norm": 1.0473837945881148,
      "learning_rate": 1.2965301031843574e-06,
      "loss": 0.1747,
      "step": 1502
    },
    {
      "epoch": 0.6772558296721866,
      "grad_norm": 0.9865409373557444,
      "learning_rate": 1.2932657507144014e-06,
      "loss": 0.1817,
      "step": 1503
    },
    {
      "epoch": 0.6777064323532725,
      "grad_norm": 0.9575822375259619,
      "learning_rate": 1.2900040784622686e-06,
      "loss": 0.1861,
      "step": 1504
    },
    {
      "epoch": 0.6781570350343584,
      "grad_norm": 0.9333033180300786,
      "learning_rate": 1.286745093672298e-06,
      "loss": 0.1716,
      "step": 1505
    },
    {
      "epoch": 0.6786076377154444,
      "grad_norm": 0.9589865720767485,
      "learning_rate": 1.2834888035828597e-06,
      "loss": 0.1697,
      "step": 1506
    },
    {
      "epoch": 0.6790582403965304,
      "grad_norm": 0.9599746633693138,
      "learning_rate": 1.2802352154263392e-06,
      "loss": 0.1649,
      "step": 1507
    },
    {
      "epoch": 0.6795088430776163,
      "grad_norm": 0.9459104304885603,
      "learning_rate": 1.2769843364291202e-06,
      "loss": 0.1758,
      "step": 1508
    },
    {
      "epoch": 0.6799594457587023,
      "grad_norm": 0.9237537511212347,
      "learning_rate": 1.2737361738115681e-06,
      "loss": 0.1712,
      "step": 1509
    },
    {
      "epoch": 0.6804100484397883,
      "grad_norm": 0.9363027728902537,
      "learning_rate": 1.2704907347880185e-06,
      "loss": 0.1672,
      "step": 1510
    },
    {
      "epoch": 0.6808606511208741,
      "grad_norm": 0.9624914550683219,
      "learning_rate": 1.2672480265667553e-06,
      "loss": 0.1701,
      "step": 1511
    },
    {
      "epoch": 0.6813112538019601,
      "grad_norm": 1.0049184213593736,
      "learning_rate": 1.2640080563499977e-06,
      "loss": 0.1684,
      "step": 1512
    },
    {
      "epoch": 0.6817618564830461,
      "grad_norm": 0.9386797820514148,
      "learning_rate": 1.2607708313338818e-06,
      "loss": 0.1669,
      "step": 1513
    },
    {
      "epoch": 0.682212459164132,
      "grad_norm": 0.9281750049626383,
      "learning_rate": 1.2575363587084486e-06,
      "loss": 0.163,
      "step": 1514
    },
    {
      "epoch": 0.682663061845218,
      "grad_norm": 0.9465979108054715,
      "learning_rate": 1.2543046456576267e-06,
      "loss": 0.1835,
      "step": 1515
    },
    {
      "epoch": 0.683113664526304,
      "grad_norm": 1.0080017997547543,
      "learning_rate": 1.2510756993592138e-06,
      "loss": 0.1846,
      "step": 1516
    },
    {
      "epoch": 0.6835642672073898,
      "grad_norm": 0.9580231845014326,
      "learning_rate": 1.2478495269848626e-06,
      "loss": 0.1726,
      "step": 1517
    },
    {
      "epoch": 0.6840148698884758,
      "grad_norm": 0.943782011616573,
      "learning_rate": 1.2446261357000655e-06,
      "loss": 0.1757,
      "step": 1518
    },
    {
      "epoch": 0.6844654725695618,
      "grad_norm": 0.9322388978426438,
      "learning_rate": 1.2414055326641378e-06,
      "loss": 0.1729,
      "step": 1519
    },
    {
      "epoch": 0.6849160752506478,
      "grad_norm": 0.9737181786806623,
      "learning_rate": 1.2381877250302002e-06,
      "loss": 0.1711,
      "step": 1520
    },
    {
      "epoch": 0.6853666779317337,
      "grad_norm": 0.9187221744081951,
      "learning_rate": 1.2349727199451696e-06,
      "loss": 0.1546,
      "step": 1521
    },
    {
      "epoch": 0.6858172806128197,
      "grad_norm": 0.9164566553499499,
      "learning_rate": 1.2317605245497324e-06,
      "loss": 0.161,
      "step": 1522
    },
    {
      "epoch": 0.6862678832939056,
      "grad_norm": 0.9373259647717872,
      "learning_rate": 1.2285511459783373e-06,
      "loss": 0.1666,
      "step": 1523
    },
    {
      "epoch": 0.6867184859749915,
      "grad_norm": 0.9556482155420707,
      "learning_rate": 1.225344591359177e-06,
      "loss": 0.1769,
      "step": 1524
    },
    {
      "epoch": 0.6871690886560775,
      "grad_norm": 0.9826183613099548,
      "learning_rate": 1.2221408678141702e-06,
      "loss": 0.1678,
      "step": 1525
    },
    {
      "epoch": 0.6876196913371635,
      "grad_norm": 0.9866834247082366,
      "learning_rate": 1.2189399824589513e-06,
      "loss": 0.1692,
      "step": 1526
    },
    {
      "epoch": 0.6880702940182494,
      "grad_norm": 0.9346005849361148,
      "learning_rate": 1.2157419424028473e-06,
      "loss": 0.1713,
      "step": 1527
    },
    {
      "epoch": 0.6885208966993354,
      "grad_norm": 0.9740865834584703,
      "learning_rate": 1.2125467547488676e-06,
      "loss": 0.1719,
      "step": 1528
    },
    {
      "epoch": 0.6889714993804213,
      "grad_norm": 0.9326077413958622,
      "learning_rate": 1.2093544265936848e-06,
      "loss": 0.1695,
      "step": 1529
    },
    {
      "epoch": 0.6894221020615072,
      "grad_norm": 0.9579890239329361,
      "learning_rate": 1.206164965027622e-06,
      "loss": 0.1631,
      "step": 1530
    },
    {
      "epoch": 0.6898727047425932,
      "grad_norm": 0.9193027867511092,
      "learning_rate": 1.2029783771346344e-06,
      "loss": 0.1513,
      "step": 1531
    },
    {
      "epoch": 0.6903233074236792,
      "grad_norm": 1.0313184201825583,
      "learning_rate": 1.1997946699922946e-06,
      "loss": 0.1861,
      "step": 1532
    },
    {
      "epoch": 0.6907739101047651,
      "grad_norm": 1.0699790891833825,
      "learning_rate": 1.1966138506717776e-06,
      "loss": 0.198,
      "step": 1533
    },
    {
      "epoch": 0.6912245127858511,
      "grad_norm": 0.9909657945142227,
      "learning_rate": 1.1934359262378443e-06,
      "loss": 0.1824,
      "step": 1534
    },
    {
      "epoch": 0.691675115466937,
      "grad_norm": 1.0552297480751036,
      "learning_rate": 1.190260903748825e-06,
      "loss": 0.1876,
      "step": 1535
    },
    {
      "epoch": 0.6921257181480229,
      "grad_norm": 0.9639349014155145,
      "learning_rate": 1.187088790256605e-06,
      "loss": 0.1847,
      "step": 1536
    },
    {
      "epoch": 0.6925763208291089,
      "grad_norm": 0.9933982369645189,
      "learning_rate": 1.1839195928066101e-06,
      "loss": 0.178,
      "step": 1537
    },
    {
      "epoch": 0.6930269235101949,
      "grad_norm": 0.9881179962964661,
      "learning_rate": 1.1807533184377882e-06,
      "loss": 0.1867,
      "step": 1538
    },
    {
      "epoch": 0.6934775261912809,
      "grad_norm": 0.9497248106372381,
      "learning_rate": 1.1775899741825947e-06,
      "loss": 0.1656,
      "step": 1539
    },
    {
      "epoch": 0.6939281288723668,
      "grad_norm": 0.9751954371033619,
      "learning_rate": 1.1744295670669752e-06,
      "loss": 0.1732,
      "step": 1540
    },
    {
      "epoch": 0.6943787315534528,
      "grad_norm": 0.9850014562760682,
      "learning_rate": 1.171272104110356e-06,
      "loss": 0.1709,
      "step": 1541
    },
    {
      "epoch": 0.6948293342345387,
      "grad_norm": 0.9356887222349297,
      "learning_rate": 1.168117592325622e-06,
      "loss": 0.1709,
      "step": 1542
    },
    {
      "epoch": 0.6952799369156246,
      "grad_norm": 0.9789522658468988,
      "learning_rate": 1.1649660387191027e-06,
      "loss": 0.1763,
      "step": 1543
    },
    {
      "epoch": 0.6957305395967106,
      "grad_norm": 0.97631716881953,
      "learning_rate": 1.1618174502905586e-06,
      "loss": 0.1798,
      "step": 1544
    },
    {
      "epoch": 0.6961811422777966,
      "grad_norm": 0.958842559879847,
      "learning_rate": 1.1586718340331634e-06,
      "loss": 0.1659,
      "step": 1545
    },
    {
      "epoch": 0.6966317449588825,
      "grad_norm": 0.9418656914716783,
      "learning_rate": 1.1555291969334907e-06,
      "loss": 0.1764,
      "step": 1546
    },
    {
      "epoch": 0.6970823476399685,
      "grad_norm": 0.9320216201816975,
      "learning_rate": 1.1523895459714948e-06,
      "loss": 0.1666,
      "step": 1547
    },
    {
      "epoch": 0.6975329503210544,
      "grad_norm": 0.9956915927235741,
      "learning_rate": 1.1492528881205027e-06,
      "loss": 0.1702,
      "step": 1548
    },
    {
      "epoch": 0.6979835530021403,
      "grad_norm": 0.9832844212602442,
      "learning_rate": 1.146119230347187e-06,
      "loss": 0.1921,
      "step": 1549
    },
    {
      "epoch": 0.6984341556832263,
      "grad_norm": 0.9762699484093021,
      "learning_rate": 1.142988579611561e-06,
      "loss": 0.1702,
      "step": 1550
    },
    {
      "epoch": 0.6988847583643123,
      "grad_norm": 0.9540162501795935,
      "learning_rate": 1.1398609428669582e-06,
      "loss": 0.1734,
      "step": 1551
    },
    {
      "epoch": 0.6993353610453982,
      "grad_norm": 0.9471062331433902,
      "learning_rate": 1.136736327060019e-06,
      "loss": 0.1631,
      "step": 1552
    },
    {
      "epoch": 0.6997859637264842,
      "grad_norm": 0.960608892965562,
      "learning_rate": 1.133614739130673e-06,
      "loss": 0.1811,
      "step": 1553
    },
    {
      "epoch": 0.7002365664075701,
      "grad_norm": 1.0008247062038094,
      "learning_rate": 1.1304961860121246e-06,
      "loss": 0.1847,
      "step": 1554
    },
    {
      "epoch": 0.700687169088656,
      "grad_norm": 0.9546410817103299,
      "learning_rate": 1.127380674630838e-06,
      "loss": 0.189,
      "step": 1555
    },
    {
      "epoch": 0.701137771769742,
      "grad_norm": 0.9665656588679624,
      "learning_rate": 1.1242682119065217e-06,
      "loss": 0.1801,
      "step": 1556
    },
    {
      "epoch": 0.701588374450828,
      "grad_norm": 0.9687500702022401,
      "learning_rate": 1.121158804752113e-06,
      "loss": 0.1778,
      "step": 1557
    },
    {
      "epoch": 0.702038977131914,
      "grad_norm": 0.9262435022683353,
      "learning_rate": 1.1180524600737624e-06,
      "loss": 0.1672,
      "step": 1558
    },
    {
      "epoch": 0.7024895798129999,
      "grad_norm": 0.9802604827010679,
      "learning_rate": 1.1149491847708186e-06,
      "loss": 0.1797,
      "step": 1559
    },
    {
      "epoch": 0.7029401824940859,
      "grad_norm": 0.9429318887072621,
      "learning_rate": 1.1118489857358129e-06,
      "loss": 0.1686,
      "step": 1560
    },
    {
      "epoch": 0.7033907851751718,
      "grad_norm": 0.9530986895583247,
      "learning_rate": 1.1087518698544444e-06,
      "loss": 0.1515,
      "step": 1561
    },
    {
      "epoch": 0.7038413878562577,
      "grad_norm": 0.9803536176709677,
      "learning_rate": 1.1056578440055631e-06,
      "loss": 0.1829,
      "step": 1562
    },
    {
      "epoch": 0.7042919905373437,
      "grad_norm": 1.0034683246802105,
      "learning_rate": 1.1025669150611594e-06,
      "loss": 0.1861,
      "step": 1563
    },
    {
      "epoch": 0.7047425932184297,
      "grad_norm": 0.9442434238308353,
      "learning_rate": 1.0994790898863409e-06,
      "loss": 0.164,
      "step": 1564
    },
    {
      "epoch": 0.7051931958995156,
      "grad_norm": 0.9803487037575894,
      "learning_rate": 1.0963943753393252e-06,
      "loss": 0.1745,
      "step": 1565
    },
    {
      "epoch": 0.7056437985806016,
      "grad_norm": 1.0057243604822155,
      "learning_rate": 1.0933127782714175e-06,
      "loss": 0.1892,
      "step": 1566
    },
    {
      "epoch": 0.7060944012616875,
      "grad_norm": 0.9489071650027512,
      "learning_rate": 1.0902343055270006e-06,
      "loss": 0.1769,
      "step": 1567
    },
    {
      "epoch": 0.7065450039427734,
      "grad_norm": 0.9722597903746137,
      "learning_rate": 1.0871589639435204e-06,
      "loss": 0.173,
      "step": 1568
    },
    {
      "epoch": 0.7069956066238594,
      "grad_norm": 0.9839533611583738,
      "learning_rate": 1.0840867603514648e-06,
      "loss": 0.1784,
      "step": 1569
    },
    {
      "epoch": 0.7074462093049454,
      "grad_norm": 0.9817529379115976,
      "learning_rate": 1.0810177015743536e-06,
      "loss": 0.185,
      "step": 1570
    },
    {
      "epoch": 0.7078968119860313,
      "grad_norm": 0.9029027680731931,
      "learning_rate": 1.0779517944287216e-06,
      "loss": 0.1555,
      "step": 1571
    },
    {
      "epoch": 0.7083474146671173,
      "grad_norm": 0.9949792727511889,
      "learning_rate": 1.0748890457241037e-06,
      "loss": 0.1941,
      "step": 1572
    },
    {
      "epoch": 0.7087980173482032,
      "grad_norm": 0.9906999802115285,
      "learning_rate": 1.0718294622630188e-06,
      "loss": 0.169,
      "step": 1573
    },
    {
      "epoch": 0.7092486200292891,
      "grad_norm": 0.9126582765867913,
      "learning_rate": 1.0687730508409594e-06,
      "loss": 0.1614,
      "step": 1574
    },
    {
      "epoch": 0.7096992227103751,
      "grad_norm": 1.0667875398406488,
      "learning_rate": 1.065719818246367e-06,
      "loss": 0.1809,
      "step": 1575
    },
    {
      "epoch": 0.7101498253914611,
      "grad_norm": 0.9828684797908929,
      "learning_rate": 1.062669771260627e-06,
      "loss": 0.175,
      "step": 1576
    },
    {
      "epoch": 0.7106004280725471,
      "grad_norm": 1.0258993060364723,
      "learning_rate": 1.0596229166580477e-06,
      "loss": 0.1743,
      "step": 1577
    },
    {
      "epoch": 0.711051030753633,
      "grad_norm": 0.8864350336490167,
      "learning_rate": 1.0565792612058462e-06,
      "loss": 0.1545,
      "step": 1578
    },
    {
      "epoch": 0.711501633434719,
      "grad_norm": 1.014170295448755,
      "learning_rate": 1.0535388116641376e-06,
      "loss": 0.1828,
      "step": 1579
    },
    {
      "epoch": 0.7119522361158049,
      "grad_norm": 0.9755717840362919,
      "learning_rate": 1.050501574785913e-06,
      "loss": 0.1764,
      "step": 1580
    },
    {
      "epoch": 0.7124028387968908,
      "grad_norm": 0.9763870999531946,
      "learning_rate": 1.0474675573170293e-06,
      "loss": 0.1722,
      "step": 1581
    },
    {
      "epoch": 0.7128534414779768,
      "grad_norm": 1.01133859060459,
      "learning_rate": 1.0444367659961927e-06,
      "loss": 0.1807,
      "step": 1582
    },
    {
      "epoch": 0.7133040441590628,
      "grad_norm": 0.976304679761672,
      "learning_rate": 1.041409207554944e-06,
      "loss": 0.1633,
      "step": 1583
    },
    {
      "epoch": 0.7137546468401487,
      "grad_norm": 1.0054771418264985,
      "learning_rate": 1.0383848887176437e-06,
      "loss": 0.1789,
      "step": 1584
    },
    {
      "epoch": 0.7142052495212347,
      "grad_norm": 0.9916973069684641,
      "learning_rate": 1.035363816201457e-06,
      "loss": 0.1589,
      "step": 1585
    },
    {
      "epoch": 0.7146558522023206,
      "grad_norm": 0.9446406731337711,
      "learning_rate": 1.032345996716339e-06,
      "loss": 0.1724,
      "step": 1586
    },
    {
      "epoch": 0.7151064548834065,
      "grad_norm": 0.9517791019473585,
      "learning_rate": 1.0293314369650193e-06,
      "loss": 0.1657,
      "step": 1587
    },
    {
      "epoch": 0.7155570575644925,
      "grad_norm": 0.9264839629163255,
      "learning_rate": 1.0263201436429873e-06,
      "loss": 0.1623,
      "step": 1588
    },
    {
      "epoch": 0.7160076602455785,
      "grad_norm": 1.0208241763086796,
      "learning_rate": 1.0233121234384777e-06,
      "loss": 0.1776,
      "step": 1589
    },
    {
      "epoch": 0.7164582629266644,
      "grad_norm": 0.962412964362255,
      "learning_rate": 1.0203073830324566e-06,
      "loss": 0.1753,
      "step": 1590
    },
    {
      "epoch": 0.7169088656077504,
      "grad_norm": 0.9392689838924189,
      "learning_rate": 1.0173059290986048e-06,
      "loss": 0.1731,
      "step": 1591
    },
    {
      "epoch": 0.7173594682888363,
      "grad_norm": 0.9674414745912845,
      "learning_rate": 1.0143077683033017e-06,
      "loss": 0.1853,
      "step": 1592
    },
    {
      "epoch": 0.7178100709699222,
      "grad_norm": 0.9463688929691734,
      "learning_rate": 1.0113129073056149e-06,
      "loss": 0.1771,
      "step": 1593
    },
    {
      "epoch": 0.7182606736510082,
      "grad_norm": 1.0074875241387953,
      "learning_rate": 1.008321352757281e-06,
      "loss": 0.1954,
      "step": 1594
    },
    {
      "epoch": 0.7187112763320942,
      "grad_norm": 0.9871282729852052,
      "learning_rate": 1.0053331113026962e-06,
      "loss": 0.1695,
      "step": 1595
    },
    {
      "epoch": 0.7191618790131802,
      "grad_norm": 0.9518387475350171,
      "learning_rate": 1.002348189578895e-06,
      "loss": 0.1811,
      "step": 1596
    },
    {
      "epoch": 0.7196124816942661,
      "grad_norm": 0.9170341529828044,
      "learning_rate": 9.993665942155395e-07,
      "loss": 0.1687,
      "step": 1597
    },
    {
      "epoch": 0.720063084375352,
      "grad_norm": 0.9957667942869256,
      "learning_rate": 9.963883318349039e-07,
      "loss": 0.1792,
      "step": 1598
    },
    {
      "epoch": 0.720513687056438,
      "grad_norm": 0.9548282910026167,
      "learning_rate": 9.934134090518593e-07,
      "loss": 0.171,
      "step": 1599
    },
    {
      "epoch": 0.7209642897375239,
      "grad_norm": 0.9649719991717198,
      "learning_rate": 9.904418324738605e-07,
      "loss": 0.1859,
      "step": 1600
    },
    {
      "epoch": 0.7214148924186099,
      "grad_norm": 0.9461436353544868,
      "learning_rate": 9.874736087009285e-07,
      "loss": 0.1708,
      "step": 1601
    },
    {
      "epoch": 0.7218654950996959,
      "grad_norm": 1.0458030537435585,
      "learning_rate": 9.84508744325639e-07,
      "loss": 0.1967,
      "step": 1602
    },
    {
      "epoch": 0.7223160977807818,
      "grad_norm": 0.9635061887687748,
      "learning_rate": 9.815472459331061e-07,
      "loss": 0.1655,
      "step": 1603
    },
    {
      "epoch": 0.7227667004618678,
      "grad_norm": 0.9821847851496524,
      "learning_rate": 9.785891201009667e-07,
      "loss": 0.1729,
      "step": 1604
    },
    {
      "epoch": 0.7232173031429537,
      "grad_norm": 0.9389984597940151,
      "learning_rate": 9.756343733993679e-07,
      "loss": 0.1713,
      "step": 1605
    },
    {
      "epoch": 0.7236679058240396,
      "grad_norm": 0.9598950848804205,
      "learning_rate": 9.726830123909527e-07,
      "loss": 0.1719,
      "step": 1606
    },
    {
      "epoch": 0.7241185085051256,
      "grad_norm": 0.9662287057802987,
      "learning_rate": 9.697350436308428e-07,
      "loss": 0.1884,
      "step": 1607
    },
    {
      "epoch": 0.7245691111862116,
      "grad_norm": 0.8983592840074917,
      "learning_rate": 9.667904736666258e-07,
      "loss": 0.159,
      "step": 1608
    },
    {
      "epoch": 0.7250197138672975,
      "grad_norm": 1.0072194197806417,
      "learning_rate": 9.638493090383408e-07,
      "loss": 0.1801,
      "step": 1609
    },
    {
      "epoch": 0.7254703165483835,
      "grad_norm": 0.9665513942121345,
      "learning_rate": 9.60911556278463e-07,
      "loss": 0.1607,
      "step": 1610
    },
    {
      "epoch": 0.7259209192294694,
      "grad_norm": 0.8889613238958063,
      "learning_rate": 9.579772219118899e-07,
      "loss": 0.1582,
      "step": 1611
    },
    {
      "epoch": 0.7263715219105553,
      "grad_norm": 0.9675942815591192,
      "learning_rate": 9.550463124559267e-07,
      "loss": 0.1889,
      "step": 1612
    },
    {
      "epoch": 0.7268221245916413,
      "grad_norm": 0.9684044848937015,
      "learning_rate": 9.521188344202717e-07,
      "loss": 0.1828,
      "step": 1613
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.9711203932408818,
      "learning_rate": 9.491947943070015e-07,
      "loss": 0.1636,
      "step": 1614
    },
    {
      "epoch": 0.7277233299538133,
      "grad_norm": 0.9589346438716095,
      "learning_rate": 9.462741986105573e-07,
      "loss": 0.1728,
      "step": 1615
    },
    {
      "epoch": 0.7281739326348992,
      "grad_norm": 0.9509324960345422,
      "learning_rate": 9.433570538177289e-07,
      "loss": 0.1808,
      "step": 1616
    },
    {
      "epoch": 0.7286245353159851,
      "grad_norm": 0.9226311502358819,
      "learning_rate": 9.404433664076442e-07,
      "loss": 0.15,
      "step": 1617
    },
    {
      "epoch": 0.7290751379970711,
      "grad_norm": 0.9459518608458831,
      "learning_rate": 9.375331428517506e-07,
      "loss": 0.1619,
      "step": 1618
    },
    {
      "epoch": 0.729525740678157,
      "grad_norm": 0.9204979082818124,
      "learning_rate": 9.346263896138e-07,
      "loss": 0.1636,
      "step": 1619
    },
    {
      "epoch": 0.729976343359243,
      "grad_norm": 0.9540585406156679,
      "learning_rate": 9.317231131498383e-07,
      "loss": 0.1745,
      "step": 1620
    },
    {
      "epoch": 0.730426946040329,
      "grad_norm": 0.9859865659794163,
      "learning_rate": 9.288233199081914e-07,
      "loss": 0.1924,
      "step": 1621
    },
    {
      "epoch": 0.7308775487214149,
      "grad_norm": 1.0188292649434882,
      "learning_rate": 9.259270163294457e-07,
      "loss": 0.1768,
      "step": 1622
    },
    {
      "epoch": 0.7313281514025008,
      "grad_norm": 0.9572221089304336,
      "learning_rate": 9.230342088464381e-07,
      "loss": 0.1639,
      "step": 1623
    },
    {
      "epoch": 0.7317787540835868,
      "grad_norm": 1.0402117668520487,
      "learning_rate": 9.201449038842403e-07,
      "loss": 0.1764,
      "step": 1624
    },
    {
      "epoch": 0.7322293567646727,
      "grad_norm": 0.9431140245076933,
      "learning_rate": 9.172591078601448e-07,
      "loss": 0.1771,
      "step": 1625
    },
    {
      "epoch": 0.7326799594457587,
      "grad_norm": 0.96634616013845,
      "learning_rate": 9.143768271836506e-07,
      "loss": 0.1783,
      "step": 1626
    },
    {
      "epoch": 0.7331305621268447,
      "grad_norm": 0.9303321472969261,
      "learning_rate": 9.114980682564492e-07,
      "loss": 0.1751,
      "step": 1627
    },
    {
      "epoch": 0.7335811648079306,
      "grad_norm": 0.9357908307152987,
      "learning_rate": 9.086228374724096e-07,
      "loss": 0.171,
      "step": 1628
    },
    {
      "epoch": 0.7340317674890166,
      "grad_norm": 0.9427016806703898,
      "learning_rate": 9.057511412175646e-07,
      "loss": 0.1683,
      "step": 1629
    },
    {
      "epoch": 0.7344823701701025,
      "grad_norm": 1.0535091981818019,
      "learning_rate": 9.028829858700974e-07,
      "loss": 0.1728,
      "step": 1630
    },
    {
      "epoch": 0.7349329728511884,
      "grad_norm": 1.015879638374581,
      "learning_rate": 9.000183778003246e-07,
      "loss": 0.1815,
      "step": 1631
    },
    {
      "epoch": 0.7353835755322744,
      "grad_norm": 0.9704576831219224,
      "learning_rate": 8.971573233706881e-07,
      "loss": 0.1719,
      "step": 1632
    },
    {
      "epoch": 0.7358341782133604,
      "grad_norm": 0.9348539870408025,
      "learning_rate": 8.942998289357333e-07,
      "loss": 0.1669,
      "step": 1633
    },
    {
      "epoch": 0.7362847808944463,
      "grad_norm": 1.0153054184600172,
      "learning_rate": 8.914459008421e-07,
      "loss": 0.1905,
      "step": 1634
    },
    {
      "epoch": 0.7367353835755323,
      "grad_norm": 1.022740942947041,
      "learning_rate": 8.885955454285078e-07,
      "loss": 0.178,
      "step": 1635
    },
    {
      "epoch": 0.7371859862566182,
      "grad_norm": 0.9811667045250413,
      "learning_rate": 8.857487690257374e-07,
      "loss": 0.1666,
      "step": 1636
    },
    {
      "epoch": 0.7376365889377042,
      "grad_norm": 0.9817560207242689,
      "learning_rate": 8.829055779566262e-07,
      "loss": 0.1717,
      "step": 1637
    },
    {
      "epoch": 0.7380871916187901,
      "grad_norm": 0.9468531084095793,
      "learning_rate": 8.800659785360444e-07,
      "loss": 0.1699,
      "step": 1638
    },
    {
      "epoch": 0.7385377942998761,
      "grad_norm": 0.9261482499229096,
      "learning_rate": 8.772299770708859e-07,
      "loss": 0.1703,
      "step": 1639
    },
    {
      "epoch": 0.7389883969809621,
      "grad_norm": 0.9865460071470947,
      "learning_rate": 8.743975798600535e-07,
      "loss": 0.1649,
      "step": 1640
    },
    {
      "epoch": 0.739438999662048,
      "grad_norm": 0.9327954533317464,
      "learning_rate": 8.71568793194445e-07,
      "loss": 0.1595,
      "step": 1641
    },
    {
      "epoch": 0.739889602343134,
      "grad_norm": 0.9404578144721737,
      "learning_rate": 8.687436233569375e-07,
      "loss": 0.1518,
      "step": 1642
    },
    {
      "epoch": 0.7403402050242199,
      "grad_norm": 0.9621250147866677,
      "learning_rate": 8.659220766223778e-07,
      "loss": 0.1818,
      "step": 1643
    },
    {
      "epoch": 0.7407908077053058,
      "grad_norm": 0.914458397158177,
      "learning_rate": 8.631041592575643e-07,
      "loss": 0.1702,
      "step": 1644
    },
    {
      "epoch": 0.7412414103863918,
      "grad_norm": 0.9466787090846438,
      "learning_rate": 8.602898775212317e-07,
      "loss": 0.1731,
      "step": 1645
    },
    {
      "epoch": 0.7416920130674778,
      "grad_norm": 0.9196682383255834,
      "learning_rate": 8.57479237664044e-07,
      "loss": 0.1659,
      "step": 1646
    },
    {
      "epoch": 0.7421426157485637,
      "grad_norm": 0.9387110760030573,
      "learning_rate": 8.546722459285727e-07,
      "loss": 0.1699,
      "step": 1647
    },
    {
      "epoch": 0.7425932184296496,
      "grad_norm": 0.9798617561875471,
      "learning_rate": 8.518689085492909e-07,
      "loss": 0.1697,
      "step": 1648
    },
    {
      "epoch": 0.7430438211107356,
      "grad_norm": 0.8746481602186569,
      "learning_rate": 8.490692317525514e-07,
      "loss": 0.1519,
      "step": 1649
    },
    {
      "epoch": 0.7434944237918215,
      "grad_norm": 0.9623227355213652,
      "learning_rate": 8.462732217565783e-07,
      "loss": 0.162,
      "step": 1650
    },
    {
      "epoch": 0.7439450264729075,
      "grad_norm": 0.8927891324002865,
      "learning_rate": 8.434808847714512e-07,
      "loss": 0.1702,
      "step": 1651
    },
    {
      "epoch": 0.7443956291539935,
      "grad_norm": 0.9417444618831278,
      "learning_rate": 8.406922269990917e-07,
      "loss": 0.1644,
      "step": 1652
    },
    {
      "epoch": 0.7448462318350794,
      "grad_norm": 0.9919400797788069,
      "learning_rate": 8.379072546332498e-07,
      "loss": 0.1844,
      "step": 1653
    },
    {
      "epoch": 0.7452968345161654,
      "grad_norm": 1.0314878215406893,
      "learning_rate": 8.351259738594902e-07,
      "loss": 0.1797,
      "step": 1654
    },
    {
      "epoch": 0.7457474371972513,
      "grad_norm": 0.974844209179654,
      "learning_rate": 8.323483908551783e-07,
      "loss": 0.1814,
      "step": 1655
    },
    {
      "epoch": 0.7461980398783373,
      "grad_norm": 0.9434763418229525,
      "learning_rate": 8.29574511789466e-07,
      "loss": 0.1765,
      "step": 1656
    },
    {
      "epoch": 0.7466486425594232,
      "grad_norm": 0.9667369016142698,
      "learning_rate": 8.268043428232798e-07,
      "loss": 0.1686,
      "step": 1657
    },
    {
      "epoch": 0.7470992452405092,
      "grad_norm": 0.9442150411912407,
      "learning_rate": 8.240378901093035e-07,
      "loss": 0.1703,
      "step": 1658
    },
    {
      "epoch": 0.7475498479215952,
      "grad_norm": 0.9185798422832856,
      "learning_rate": 8.212751597919708e-07,
      "loss": 0.1804,
      "step": 1659
    },
    {
      "epoch": 0.748000450602681,
      "grad_norm": 0.9150873838854694,
      "learning_rate": 8.185161580074444e-07,
      "loss": 0.1588,
      "step": 1660
    },
    {
      "epoch": 0.748451053283767,
      "grad_norm": 0.9401507535925011,
      "learning_rate": 8.157608908836071e-07,
      "loss": 0.1583,
      "step": 1661
    },
    {
      "epoch": 0.748901655964853,
      "grad_norm": 0.901302230754303,
      "learning_rate": 8.130093645400469e-07,
      "loss": 0.1677,
      "step": 1662
    },
    {
      "epoch": 0.7493522586459389,
      "grad_norm": 0.9197142291521521,
      "learning_rate": 8.102615850880413e-07,
      "loss": 0.1724,
      "step": 1663
    },
    {
      "epoch": 0.7498028613270249,
      "grad_norm": 0.9382328996600817,
      "learning_rate": 8.075175586305492e-07,
      "loss": 0.1726,
      "step": 1664
    },
    {
      "epoch": 0.7502534640081109,
      "grad_norm": 0.90462702708837,
      "learning_rate": 8.047772912621921e-07,
      "loss": 0.1636,
      "step": 1665
    },
    {
      "epoch": 0.7507040666891968,
      "grad_norm": 0.9711524591154302,
      "learning_rate": 8.020407890692419e-07,
      "loss": 0.1716,
      "step": 1666
    },
    {
      "epoch": 0.7511546693702827,
      "grad_norm": 0.9434890487318909,
      "learning_rate": 7.993080581296087e-07,
      "loss": 0.1784,
      "step": 1667
    },
    {
      "epoch": 0.7516052720513687,
      "grad_norm": 0.9312101653818623,
      "learning_rate": 7.96579104512826e-07,
      "loss": 0.1644,
      "step": 1668
    },
    {
      "epoch": 0.7520558747324546,
      "grad_norm": 0.9848096739117619,
      "learning_rate": 7.938539342800373e-07,
      "loss": 0.1783,
      "step": 1669
    },
    {
      "epoch": 0.7525064774135406,
      "grad_norm": 0.9327157230766141,
      "learning_rate": 7.911325534839851e-07,
      "loss": 0.1664,
      "step": 1670
    },
    {
      "epoch": 0.7529570800946266,
      "grad_norm": 0.9750163171110814,
      "learning_rate": 7.884149681689937e-07,
      "loss": 0.1858,
      "step": 1671
    },
    {
      "epoch": 0.7534076827757125,
      "grad_norm": 0.9094999286751292,
      "learning_rate": 7.857011843709559e-07,
      "loss": 0.1582,
      "step": 1672
    },
    {
      "epoch": 0.7538582854567984,
      "grad_norm": 0.9650020997875528,
      "learning_rate": 7.829912081173238e-07,
      "loss": 0.1689,
      "step": 1673
    },
    {
      "epoch": 0.7543088881378844,
      "grad_norm": 0.9982763186279041,
      "learning_rate": 7.802850454270913e-07,
      "loss": 0.1758,
      "step": 1674
    },
    {
      "epoch": 0.7547594908189704,
      "grad_norm": 0.9579696747409793,
      "learning_rate": 7.775827023107835e-07,
      "loss": 0.1679,
      "step": 1675
    },
    {
      "epoch": 0.7552100935000563,
      "grad_norm": 1.0668093753827521,
      "learning_rate": 7.74884184770441e-07,
      "loss": 0.1969,
      "step": 1676
    },
    {
      "epoch": 0.7556606961811423,
      "grad_norm": 0.9626572194119604,
      "learning_rate": 7.721894987996076e-07,
      "loss": 0.1733,
      "step": 1677
    },
    {
      "epoch": 0.7561112988622283,
      "grad_norm": 0.9744882640905289,
      "learning_rate": 7.694986503833171e-07,
      "loss": 0.1801,
      "step": 1678
    },
    {
      "epoch": 0.7565619015433142,
      "grad_norm": 0.9403313127374262,
      "learning_rate": 7.668116454980804e-07,
      "loss": 0.1734,
      "step": 1679
    },
    {
      "epoch": 0.7570125042244001,
      "grad_norm": 0.9897631979152582,
      "learning_rate": 7.641284901118703e-07,
      "loss": 0.1855,
      "step": 1680
    },
    {
      "epoch": 0.7574631069054861,
      "grad_norm": 0.917940041749129,
      "learning_rate": 7.614491901841118e-07,
      "loss": 0.1649,
      "step": 1681
    },
    {
      "epoch": 0.757913709586572,
      "grad_norm": 0.9028036690911325,
      "learning_rate": 7.587737516656651e-07,
      "loss": 0.164,
      "step": 1682
    },
    {
      "epoch": 0.758364312267658,
      "grad_norm": 0.8959138677827185,
      "learning_rate": 7.561021804988141e-07,
      "loss": 0.1613,
      "step": 1683
    },
    {
      "epoch": 0.758814914948744,
      "grad_norm": 0.9453945367159504,
      "learning_rate": 7.534344826172546e-07,
      "loss": 0.1812,
      "step": 1684
    },
    {
      "epoch": 0.7592655176298299,
      "grad_norm": 0.9475739534763087,
      "learning_rate": 7.507706639460768e-07,
      "loss": 0.1774,
      "step": 1685
    },
    {
      "epoch": 0.7597161203109158,
      "grad_norm": 0.9284456020009094,
      "learning_rate": 7.481107304017588e-07,
      "loss": 0.1757,
      "step": 1686
    },
    {
      "epoch": 0.7601667229920018,
      "grad_norm": 0.9481356302559381,
      "learning_rate": 7.454546878921465e-07,
      "loss": 0.1717,
      "step": 1687
    },
    {
      "epoch": 0.7606173256730877,
      "grad_norm": 1.0275463149896271,
      "learning_rate": 7.428025423164456e-07,
      "loss": 0.1844,
      "step": 1688
    },
    {
      "epoch": 0.7610679283541737,
      "grad_norm": 0.9794885029503064,
      "learning_rate": 7.401542995652033e-07,
      "loss": 0.1798,
      "step": 1689
    },
    {
      "epoch": 0.7615185310352597,
      "grad_norm": 0.9444897520966486,
      "learning_rate": 7.375099655203033e-07,
      "loss": 0.1651,
      "step": 1690
    },
    {
      "epoch": 0.7619691337163456,
      "grad_norm": 0.9188622310083528,
      "learning_rate": 7.348695460549443e-07,
      "loss": 0.168,
      "step": 1691
    },
    {
      "epoch": 0.7624197363974315,
      "grad_norm": 0.9874990105238635,
      "learning_rate": 7.322330470336314e-07,
      "loss": 0.1838,
      "step": 1692
    },
    {
      "epoch": 0.7628703390785175,
      "grad_norm": 0.9549127415217946,
      "learning_rate": 7.296004743121627e-07,
      "loss": 0.1685,
      "step": 1693
    },
    {
      "epoch": 0.7633209417596035,
      "grad_norm": 0.9834011052029592,
      "learning_rate": 7.26971833737615e-07,
      "loss": 0.1752,
      "step": 1694
    },
    {
      "epoch": 0.7637715444406894,
      "grad_norm": 0.9292459980595047,
      "learning_rate": 7.243471311483322e-07,
      "loss": 0.1686,
      "step": 1695
    },
    {
      "epoch": 0.7642221471217754,
      "grad_norm": 0.9744168782441898,
      "learning_rate": 7.217263723739107e-07,
      "loss": 0.1864,
      "step": 1696
    },
    {
      "epoch": 0.7646727498028614,
      "grad_norm": 0.9696008836681465,
      "learning_rate": 7.191095632351908e-07,
      "loss": 0.1776,
      "step": 1697
    },
    {
      "epoch": 0.7651233524839472,
      "grad_norm": 0.96774867165191,
      "learning_rate": 7.164967095442357e-07,
      "loss": 0.1759,
      "step": 1698
    },
    {
      "epoch": 0.7655739551650332,
      "grad_norm": 0.9884684198285855,
      "learning_rate": 7.138878171043262e-07,
      "loss": 0.1732,
      "step": 1699
    },
    {
      "epoch": 0.7660245578461192,
      "grad_norm": 1.0068448942639325,
      "learning_rate": 7.112828917099438e-07,
      "loss": 0.1739,
      "step": 1700
    },
    {
      "epoch": 0.7664751605272051,
      "grad_norm": 0.9987811192949849,
      "learning_rate": 7.086819391467612e-07,
      "loss": 0.1831,
      "step": 1701
    },
    {
      "epoch": 0.7669257632082911,
      "grad_norm": 1.0017086323571975,
      "learning_rate": 7.060849651916244e-07,
      "loss": 0.1709,
      "step": 1702
    },
    {
      "epoch": 0.7673763658893771,
      "grad_norm": 0.9373080593200287,
      "learning_rate": 7.034919756125447e-07,
      "loss": 0.1696,
      "step": 1703
    },
    {
      "epoch": 0.767826968570463,
      "grad_norm": 0.9490094333958343,
      "learning_rate": 7.009029761686825e-07,
      "loss": 0.167,
      "step": 1704
    },
    {
      "epoch": 0.7682775712515489,
      "grad_norm": 0.9750208874260144,
      "learning_rate": 6.98317972610337e-07,
      "loss": 0.1604,
      "step": 1705
    },
    {
      "epoch": 0.7687281739326349,
      "grad_norm": 0.9624993279869102,
      "learning_rate": 6.957369706789319e-07,
      "loss": 0.1608,
      "step": 1706
    },
    {
      "epoch": 0.7691787766137208,
      "grad_norm": 0.9787624995991694,
      "learning_rate": 6.931599761070027e-07,
      "loss": 0.1691,
      "step": 1707
    },
    {
      "epoch": 0.7696293792948068,
      "grad_norm": 1.007340459210081,
      "learning_rate": 6.905869946181848e-07,
      "loss": 0.1812,
      "step": 1708
    },
    {
      "epoch": 0.7700799819758928,
      "grad_norm": 0.9808544623527553,
      "learning_rate": 6.880180319272006e-07,
      "loss": 0.1764,
      "step": 1709
    },
    {
      "epoch": 0.7705305846569787,
      "grad_norm": 0.9657443098275049,
      "learning_rate": 6.854530937398459e-07,
      "loss": 0.1787,
      "step": 1710
    },
    {
      "epoch": 0.7709811873380646,
      "grad_norm": 0.9448605482477895,
      "learning_rate": 6.828921857529774e-07,
      "loss": 0.1602,
      "step": 1711
    },
    {
      "epoch": 0.7714317900191506,
      "grad_norm": 0.9353366127853433,
      "learning_rate": 6.803353136545033e-07,
      "loss": 0.1814,
      "step": 1712
    },
    {
      "epoch": 0.7718823927002366,
      "grad_norm": 1.0082830443026907,
      "learning_rate": 6.777824831233645e-07,
      "loss": 0.1843,
      "step": 1713
    },
    {
      "epoch": 0.7723329953813225,
      "grad_norm": 1.0012175555664218,
      "learning_rate": 6.752336998295281e-07,
      "loss": 0.1656,
      "step": 1714
    },
    {
      "epoch": 0.7727835980624085,
      "grad_norm": 0.9510344959155705,
      "learning_rate": 6.726889694339689e-07,
      "loss": 0.1735,
      "step": 1715
    },
    {
      "epoch": 0.7732342007434945,
      "grad_norm": 0.9607739674740182,
      "learning_rate": 6.701482975886617e-07,
      "loss": 0.1761,
      "step": 1716
    },
    {
      "epoch": 0.7736848034245803,
      "grad_norm": 0.9266629906847302,
      "learning_rate": 6.676116899365692e-07,
      "loss": 0.1649,
      "step": 1717
    },
    {
      "epoch": 0.7741354061056663,
      "grad_norm": 0.9105923017777625,
      "learning_rate": 6.650791521116243e-07,
      "loss": 0.1663,
      "step": 1718
    },
    {
      "epoch": 0.7745860087867523,
      "grad_norm": 0.9094686688201417,
      "learning_rate": 6.625506897387215e-07,
      "loss": 0.1626,
      "step": 1719
    },
    {
      "epoch": 0.7750366114678382,
      "grad_norm": 1.0543916388352368,
      "learning_rate": 6.600263084337041e-07,
      "loss": 0.1732,
      "step": 1720
    },
    {
      "epoch": 0.7754872141489242,
      "grad_norm": 0.9678813090969567,
      "learning_rate": 6.575060138033504e-07,
      "loss": 0.1778,
      "step": 1721
    },
    {
      "epoch": 0.7759378168300102,
      "grad_norm": 0.9860219233564725,
      "learning_rate": 6.549898114453615e-07,
      "loss": 0.1761,
      "step": 1722
    },
    {
      "epoch": 0.776388419511096,
      "grad_norm": 0.9359894577473049,
      "learning_rate": 6.524777069483526e-07,
      "loss": 0.1663,
      "step": 1723
    },
    {
      "epoch": 0.776839022192182,
      "grad_norm": 1.0095972151252641,
      "learning_rate": 6.499697058918326e-07,
      "loss": 0.1748,
      "step": 1724
    },
    {
      "epoch": 0.777289624873268,
      "grad_norm": 0.9426089610465914,
      "learning_rate": 6.474658138461992e-07,
      "loss": 0.173,
      "step": 1725
    },
    {
      "epoch": 0.7777402275543539,
      "grad_norm": 0.9234669569269677,
      "learning_rate": 6.449660363727236e-07,
      "loss": 0.1663,
      "step": 1726
    },
    {
      "epoch": 0.7781908302354399,
      "grad_norm": 0.9405863138790215,
      "learning_rate": 6.424703790235374e-07,
      "loss": 0.1692,
      "step": 1727
    },
    {
      "epoch": 0.7786414329165259,
      "grad_norm": 0.9335835838722892,
      "learning_rate": 6.399788473416229e-07,
      "loss": 0.166,
      "step": 1728
    },
    {
      "epoch": 0.7790920355976118,
      "grad_norm": 0.9511706860844333,
      "learning_rate": 6.374914468607976e-07,
      "loss": 0.1791,
      "step": 1729
    },
    {
      "epoch": 0.7795426382786977,
      "grad_norm": 0.9852564601610345,
      "learning_rate": 6.35008183105704e-07,
      "loss": 0.1628,
      "step": 1730
    },
    {
      "epoch": 0.7799932409597837,
      "grad_norm": 1.0096784064849857,
      "learning_rate": 6.325290615917961e-07,
      "loss": 0.187,
      "step": 1731
    },
    {
      "epoch": 0.7804438436408697,
      "grad_norm": 0.9287036327371591,
      "learning_rate": 6.300540878253286e-07,
      "loss": 0.1689,
      "step": 1732
    },
    {
      "epoch": 0.7808944463219556,
      "grad_norm": 0.9645312989670097,
      "learning_rate": 6.27583267303343e-07,
      "loss": 0.1629,
      "step": 1733
    },
    {
      "epoch": 0.7813450490030416,
      "grad_norm": 0.9098747848669801,
      "learning_rate": 6.251166055136573e-07,
      "loss": 0.1663,
      "step": 1734
    },
    {
      "epoch": 0.7817956516841276,
      "grad_norm": 0.9672682550620934,
      "learning_rate": 6.226541079348517e-07,
      "loss": 0.1824,
      "step": 1735
    },
    {
      "epoch": 0.7822462543652134,
      "grad_norm": 0.9033211880815116,
      "learning_rate": 6.201957800362579e-07,
      "loss": 0.1572,
      "step": 1736
    },
    {
      "epoch": 0.7826968570462994,
      "grad_norm": 0.9729803882075092,
      "learning_rate": 6.177416272779468e-07,
      "loss": 0.1789,
      "step": 1737
    },
    {
      "epoch": 0.7831474597273854,
      "grad_norm": 1.0067811549313692,
      "learning_rate": 6.152916551107149e-07,
      "loss": 0.1843,
      "step": 1738
    },
    {
      "epoch": 0.7835980624084713,
      "grad_norm": 0.9130863572610125,
      "learning_rate": 6.12845868976076e-07,
      "loss": 0.1524,
      "step": 1739
    },
    {
      "epoch": 0.7840486650895573,
      "grad_norm": 0.9814494497101688,
      "learning_rate": 6.104042743062439e-07,
      "loss": 0.175,
      "step": 1740
    },
    {
      "epoch": 0.7844992677706433,
      "grad_norm": 0.9493598919925331,
      "learning_rate": 6.079668765241248e-07,
      "loss": 0.1685,
      "step": 1741
    },
    {
      "epoch": 0.7849498704517291,
      "grad_norm": 0.9674465998413315,
      "learning_rate": 6.05533681043301e-07,
      "loss": 0.1731,
      "step": 1742
    },
    {
      "epoch": 0.7854004731328151,
      "grad_norm": 0.9144085528498724,
      "learning_rate": 6.031046932680229e-07,
      "loss": 0.1577,
      "step": 1743
    },
    {
      "epoch": 0.7858510758139011,
      "grad_norm": 0.9306591229731381,
      "learning_rate": 6.006799185931964e-07,
      "loss": 0.1738,
      "step": 1744
    },
    {
      "epoch": 0.786301678494987,
      "grad_norm": 0.9666823036771994,
      "learning_rate": 5.982593624043682e-07,
      "loss": 0.1656,
      "step": 1745
    },
    {
      "epoch": 0.786752281176073,
      "grad_norm": 1.0216256666717884,
      "learning_rate": 5.958430300777157e-07,
      "loss": 0.1815,
      "step": 1746
    },
    {
      "epoch": 0.787202883857159,
      "grad_norm": 0.9414221341205914,
      "learning_rate": 5.934309269800359e-07,
      "loss": 0.1635,
      "step": 1747
    },
    {
      "epoch": 0.7876534865382449,
      "grad_norm": 1.073218436568168,
      "learning_rate": 5.910230584687316e-07,
      "loss": 0.1747,
      "step": 1748
    },
    {
      "epoch": 0.7881040892193308,
      "grad_norm": 1.022405588004278,
      "learning_rate": 5.886194298917994e-07,
      "loss": 0.1834,
      "step": 1749
    },
    {
      "epoch": 0.7885546919004168,
      "grad_norm": 0.9334120421015962,
      "learning_rate": 5.862200465878228e-07,
      "loss": 0.1603,
      "step": 1750
    },
    {
      "epoch": 0.7890052945815028,
      "grad_norm": 0.9383559818428457,
      "learning_rate": 5.838249138859509e-07,
      "loss": 0.1651,
      "step": 1751
    },
    {
      "epoch": 0.7894558972625887,
      "grad_norm": 0.8996153804259718,
      "learning_rate": 5.814340371058957e-07,
      "loss": 0.1596,
      "step": 1752
    },
    {
      "epoch": 0.7899064999436747,
      "grad_norm": 0.9662819628561679,
      "learning_rate": 5.79047421557915e-07,
      "loss": 0.1701,
      "step": 1753
    },
    {
      "epoch": 0.7903571026247607,
      "grad_norm": 0.9821115878303758,
      "learning_rate": 5.766650725428027e-07,
      "loss": 0.1721,
      "step": 1754
    },
    {
      "epoch": 0.7908077053058465,
      "grad_norm": 0.9530481652480088,
      "learning_rate": 5.742869953518773e-07,
      "loss": 0.1698,
      "step": 1755
    },
    {
      "epoch": 0.7912583079869325,
      "grad_norm": 0.9347238724778862,
      "learning_rate": 5.719131952669679e-07,
      "loss": 0.1746,
      "step": 1756
    },
    {
      "epoch": 0.7917089106680185,
      "grad_norm": 0.9974845092899297,
      "learning_rate": 5.695436775604049e-07,
      "loss": 0.1872,
      "step": 1757
    },
    {
      "epoch": 0.7921595133491044,
      "grad_norm": 0.9823704981517646,
      "learning_rate": 5.671784474950068e-07,
      "loss": 0.1646,
      "step": 1758
    },
    {
      "epoch": 0.7926101160301904,
      "grad_norm": 0.9528761781150609,
      "learning_rate": 5.648175103240694e-07,
      "loss": 0.1678,
      "step": 1759
    },
    {
      "epoch": 0.7930607187112764,
      "grad_norm": 0.9137049047059816,
      "learning_rate": 5.624608712913531e-07,
      "loss": 0.1702,
      "step": 1760
    },
    {
      "epoch": 0.7935113213923622,
      "grad_norm": 0.9307972970588964,
      "learning_rate": 5.601085356310734e-07,
      "loss": 0.176,
      "step": 1761
    },
    {
      "epoch": 0.7939619240734482,
      "grad_norm": 0.9444469133716082,
      "learning_rate": 5.577605085678858e-07,
      "loss": 0.1694,
      "step": 1762
    },
    {
      "epoch": 0.7944125267545342,
      "grad_norm": 0.9333704389867957,
      "learning_rate": 5.554167953168779e-07,
      "loss": 0.1675,
      "step": 1763
    },
    {
      "epoch": 0.7948631294356201,
      "grad_norm": 0.993750845065701,
      "learning_rate": 5.530774010835552e-07,
      "loss": 0.1819,
      "step": 1764
    },
    {
      "epoch": 0.7953137321167061,
      "grad_norm": 0.9348594108562784,
      "learning_rate": 5.507423310638299e-07,
      "loss": 0.1681,
      "step": 1765
    },
    {
      "epoch": 0.7957643347977921,
      "grad_norm": 0.9708757301258297,
      "learning_rate": 5.48411590444012e-07,
      "loss": 0.164,
      "step": 1766
    },
    {
      "epoch": 0.796214937478878,
      "grad_norm": 0.9431276887221065,
      "learning_rate": 5.460851844007945e-07,
      "loss": 0.1701,
      "step": 1767
    },
    {
      "epoch": 0.7966655401599639,
      "grad_norm": 0.9369828225416942,
      "learning_rate": 5.437631181012415e-07,
      "loss": 0.1725,
      "step": 1768
    },
    {
      "epoch": 0.7971161428410499,
      "grad_norm": 0.9742268350847076,
      "learning_rate": 5.414453967027797e-07,
      "loss": 0.1763,
      "step": 1769
    },
    {
      "epoch": 0.7975667455221359,
      "grad_norm": 0.9691529149171444,
      "learning_rate": 5.391320253531868e-07,
      "loss": 0.1813,
      "step": 1770
    },
    {
      "epoch": 0.7980173482032218,
      "grad_norm": 0.8867961718282267,
      "learning_rate": 5.368230091905774e-07,
      "loss": 0.1616,
      "step": 1771
    },
    {
      "epoch": 0.7984679508843078,
      "grad_norm": 0.9198921470166019,
      "learning_rate": 5.345183533433926e-07,
      "loss": 0.1619,
      "step": 1772
    },
    {
      "epoch": 0.7989185535653938,
      "grad_norm": 0.9747081060897254,
      "learning_rate": 5.322180629303902e-07,
      "loss": 0.1806,
      "step": 1773
    },
    {
      "epoch": 0.7993691562464796,
      "grad_norm": 0.9760127339119377,
      "learning_rate": 5.299221430606313e-07,
      "loss": 0.1758,
      "step": 1774
    },
    {
      "epoch": 0.7998197589275656,
      "grad_norm": 0.9701776719091275,
      "learning_rate": 5.276305988334701e-07,
      "loss": 0.1806,
      "step": 1775
    },
    {
      "epoch": 0.8002703616086516,
      "grad_norm": 0.988371999259503,
      "learning_rate": 5.253434353385422e-07,
      "loss": 0.1687,
      "step": 1776
    },
    {
      "epoch": 0.8007209642897375,
      "grad_norm": 0.8598044634557204,
      "learning_rate": 5.23060657655754e-07,
      "loss": 0.1523,
      "step": 1777
    },
    {
      "epoch": 0.8011715669708235,
      "grad_norm": 0.9071673557398994,
      "learning_rate": 5.207822708552695e-07,
      "loss": 0.1655,
      "step": 1778
    },
    {
      "epoch": 0.8016221696519095,
      "grad_norm": 0.9513483134209044,
      "learning_rate": 5.185082799975013e-07,
      "loss": 0.1738,
      "step": 1779
    },
    {
      "epoch": 0.8020727723329953,
      "grad_norm": 1.0075511177548173,
      "learning_rate": 5.162386901330977e-07,
      "loss": 0.161,
      "step": 1780
    },
    {
      "epoch": 0.8025233750140813,
      "grad_norm": 0.9359270752654021,
      "learning_rate": 5.139735063029338e-07,
      "loss": 0.1771,
      "step": 1781
    },
    {
      "epoch": 0.8029739776951673,
      "grad_norm": 0.9189830873891193,
      "learning_rate": 5.117127335380967e-07,
      "loss": 0.1623,
      "step": 1782
    },
    {
      "epoch": 0.8034245803762532,
      "grad_norm": 0.9536212390806852,
      "learning_rate": 5.094563768598773e-07,
      "loss": 0.1765,
      "step": 1783
    },
    {
      "epoch": 0.8038751830573392,
      "grad_norm": 0.892559123324922,
      "learning_rate": 5.07204441279758e-07,
      "loss": 0.1595,
      "step": 1784
    },
    {
      "epoch": 0.8043257857384252,
      "grad_norm": 1.0071090882494602,
      "learning_rate": 5.049569317994013e-07,
      "loss": 0.1775,
      "step": 1785
    },
    {
      "epoch": 0.804776388419511,
      "grad_norm": 0.908277153855911,
      "learning_rate": 5.027138534106399e-07,
      "loss": 0.1548,
      "step": 1786
    },
    {
      "epoch": 0.805226991100597,
      "grad_norm": 0.9223046738550524,
      "learning_rate": 5.004752110954642e-07,
      "loss": 0.1693,
      "step": 1787
    },
    {
      "epoch": 0.805677593781683,
      "grad_norm": 0.9814309602126301,
      "learning_rate": 4.982410098260118e-07,
      "loss": 0.173,
      "step": 1788
    },
    {
      "epoch": 0.806128196462769,
      "grad_norm": 0.949733318801411,
      "learning_rate": 4.96011254564557e-07,
      "loss": 0.1552,
      "step": 1789
    },
    {
      "epoch": 0.8065787991438549,
      "grad_norm": 0.9276630676010313,
      "learning_rate": 4.937859502634992e-07,
      "loss": 0.1684,
      "step": 1790
    },
    {
      "epoch": 0.8070294018249409,
      "grad_norm": 0.9493296870395656,
      "learning_rate": 4.915651018653511e-07,
      "loss": 0.1708,
      "step": 1791
    },
    {
      "epoch": 0.8074800045060269,
      "grad_norm": 0.9831480782450759,
      "learning_rate": 4.893487143027307e-07,
      "loss": 0.1716,
      "step": 1792
    },
    {
      "epoch": 0.8079306071871127,
      "grad_norm": 0.9536954872523767,
      "learning_rate": 4.871367924983458e-07,
      "loss": 0.1657,
      "step": 1793
    },
    {
      "epoch": 0.8083812098681987,
      "grad_norm": 0.9577400855458258,
      "learning_rate": 4.84929341364988e-07,
      "loss": 0.1695,
      "step": 1794
    },
    {
      "epoch": 0.8088318125492847,
      "grad_norm": 1.0109618072957722,
      "learning_rate": 4.827263658055161e-07,
      "loss": 0.176,
      "step": 1795
    },
    {
      "epoch": 0.8092824152303706,
      "grad_norm": 0.9240733937478894,
      "learning_rate": 4.805278707128505e-07,
      "loss": 0.1633,
      "step": 1796
    },
    {
      "epoch": 0.8097330179114566,
      "grad_norm": 0.9175205272410301,
      "learning_rate": 4.783338609699614e-07,
      "loss": 0.1639,
      "step": 1797
    },
    {
      "epoch": 0.8101836205925426,
      "grad_norm": 0.9832145863528448,
      "learning_rate": 4.7614434144985486e-07,
      "loss": 0.1839,
      "step": 1798
    },
    {
      "epoch": 0.8106342232736284,
      "grad_norm": 0.9250138705305986,
      "learning_rate": 4.73959317015564e-07,
      "loss": 0.1607,
      "step": 1799
    },
    {
      "epoch": 0.8110848259547144,
      "grad_norm": 0.9988990989007607,
      "learning_rate": 4.7177879252013945e-07,
      "loss": 0.1963,
      "step": 1800
    },
    {
      "epoch": 0.8115354286358004,
      "grad_norm": 0.9884011429992995,
      "learning_rate": 4.6960277280663574e-07,
      "loss": 0.1747,
      "step": 1801
    },
    {
      "epoch": 0.8119860313168863,
      "grad_norm": 0.9687937064883984,
      "learning_rate": 4.674312627081032e-07,
      "loss": 0.1746,
      "step": 1802
    },
    {
      "epoch": 0.8124366339979723,
      "grad_norm": 0.9680030187172702,
      "learning_rate": 4.6526426704757545e-07,
      "loss": 0.1683,
      "step": 1803
    },
    {
      "epoch": 0.8128872366790583,
      "grad_norm": 0.9128681785542643,
      "learning_rate": 4.6310179063805916e-07,
      "loss": 0.1583,
      "step": 1804
    },
    {
      "epoch": 0.8133378393601441,
      "grad_norm": 0.9503424077025497,
      "learning_rate": 4.609438382825246e-07,
      "loss": 0.168,
      "step": 1805
    },
    {
      "epoch": 0.8137884420412301,
      "grad_norm": 0.9854433552645877,
      "learning_rate": 4.587904147738925e-07,
      "loss": 0.1722,
      "step": 1806
    },
    {
      "epoch": 0.8142390447223161,
      "grad_norm": 0.9535187246316041,
      "learning_rate": 4.566415248950251e-07,
      "loss": 0.1714,
      "step": 1807
    },
    {
      "epoch": 0.814689647403402,
      "grad_norm": 0.9730755908109525,
      "learning_rate": 4.5449717341871646e-07,
      "loss": 0.1631,
      "step": 1808
    },
    {
      "epoch": 0.815140250084488,
      "grad_norm": 0.9486061901445837,
      "learning_rate": 4.5235736510767957e-07,
      "loss": 0.1631,
      "step": 1809
    },
    {
      "epoch": 0.815590852765574,
      "grad_norm": 0.9516956179529896,
      "learning_rate": 4.5022210471453664e-07,
      "loss": 0.177,
      "step": 1810
    },
    {
      "epoch": 0.81604145544666,
      "grad_norm": 0.9337467064350875,
      "learning_rate": 4.480913969818099e-07,
      "loss": 0.1738,
      "step": 1811
    },
    {
      "epoch": 0.8164920581277458,
      "grad_norm": 0.9557630287598873,
      "learning_rate": 4.4596524664190674e-07,
      "loss": 0.1564,
      "step": 1812
    },
    {
      "epoch": 0.8169426608088318,
      "grad_norm": 0.9106928433789534,
      "learning_rate": 4.4384365841711684e-07,
      "loss": 0.1682,
      "step": 1813
    },
    {
      "epoch": 0.8173932634899178,
      "grad_norm": 0.8929215982521911,
      "learning_rate": 4.417266370195944e-07,
      "loss": 0.1635,
      "step": 1814
    },
    {
      "epoch": 0.8178438661710037,
      "grad_norm": 0.9601587758950517,
      "learning_rate": 4.3961418715135097e-07,
      "loss": 0.1759,
      "step": 1815
    },
    {
      "epoch": 0.8182944688520897,
      "grad_norm": 0.9622107099235466,
      "learning_rate": 4.3750631350424456e-07,
      "loss": 0.1727,
      "step": 1816
    },
    {
      "epoch": 0.8187450715331757,
      "grad_norm": 0.9675048072605508,
      "learning_rate": 4.354030207599691e-07,
      "loss": 0.1744,
      "step": 1817
    },
    {
      "epoch": 0.8191956742142615,
      "grad_norm": 0.9853236739076215,
      "learning_rate": 4.333043135900436e-07,
      "loss": 0.1686,
      "step": 1818
    },
    {
      "epoch": 0.8196462768953475,
      "grad_norm": 0.9781020890619168,
      "learning_rate": 4.312101966558044e-07,
      "loss": 0.1729,
      "step": 1819
    },
    {
      "epoch": 0.8200968795764335,
      "grad_norm": 0.9115554590765232,
      "learning_rate": 4.2912067460839066e-07,
      "loss": 0.1581,
      "step": 1820
    },
    {
      "epoch": 0.8205474822575194,
      "grad_norm": 0.9616917821289626,
      "learning_rate": 4.2703575208873585e-07,
      "loss": 0.1565,
      "step": 1821
    },
    {
      "epoch": 0.8209980849386054,
      "grad_norm": 0.922130671794463,
      "learning_rate": 4.2495543372755854e-07,
      "loss": 0.1595,
      "step": 1822
    },
    {
      "epoch": 0.8214486876196914,
      "grad_norm": 0.9683579853431844,
      "learning_rate": 4.2287972414535084e-07,
      "loss": 0.1673,
      "step": 1823
    },
    {
      "epoch": 0.8218992903007772,
      "grad_norm": 0.961905725310261,
      "learning_rate": 4.208086279523699e-07,
      "loss": 0.1741,
      "step": 1824
    },
    {
      "epoch": 0.8223498929818632,
      "grad_norm": 0.9119598816657689,
      "learning_rate": 4.1874214974862436e-07,
      "loss": 0.1695,
      "step": 1825
    },
    {
      "epoch": 0.8228004956629492,
      "grad_norm": 0.973526812761145,
      "learning_rate": 4.1668029412386677e-07,
      "loss": 0.1733,
      "step": 1826
    },
    {
      "epoch": 0.8232510983440351,
      "grad_norm": 0.9599913535559322,
      "learning_rate": 4.146230656575831e-07,
      "loss": 0.1792,
      "step": 1827
    },
    {
      "epoch": 0.8237017010251211,
      "grad_norm": 0.9636632216252867,
      "learning_rate": 4.125704689189819e-07,
      "loss": 0.1761,
      "step": 1828
    },
    {
      "epoch": 0.8241523037062071,
      "grad_norm": 0.9087918090205587,
      "learning_rate": 4.105225084669839e-07,
      "loss": 0.1551,
      "step": 1829
    },
    {
      "epoch": 0.824602906387293,
      "grad_norm": 0.9532950453959921,
      "learning_rate": 4.084791888502135e-07,
      "loss": 0.1562,
      "step": 1830
    },
    {
      "epoch": 0.8250535090683789,
      "grad_norm": 0.9171866851813432,
      "learning_rate": 4.0644051460698634e-07,
      "loss": 0.1711,
      "step": 1831
    },
    {
      "epoch": 0.8255041117494649,
      "grad_norm": 0.9644212760081424,
      "learning_rate": 4.0440649026530166e-07,
      "loss": 0.1763,
      "step": 1832
    },
    {
      "epoch": 0.8259547144305509,
      "grad_norm": 0.9135624183632747,
      "learning_rate": 4.0237712034283004e-07,
      "loss": 0.1699,
      "step": 1833
    },
    {
      "epoch": 0.8264053171116368,
      "grad_norm": 0.9700605328629466,
      "learning_rate": 4.003524093469041e-07,
      "loss": 0.1795,
      "step": 1834
    },
    {
      "epoch": 0.8268559197927228,
      "grad_norm": 0.9494632861193675,
      "learning_rate": 3.983323617745111e-07,
      "loss": 0.166,
      "step": 1835
    },
    {
      "epoch": 0.8273065224738088,
      "grad_norm": 0.9689443716041818,
      "learning_rate": 3.963169821122778e-07,
      "loss": 0.1667,
      "step": 1836
    },
    {
      "epoch": 0.8277571251548946,
      "grad_norm": 0.9494816847933051,
      "learning_rate": 3.943062748364651e-07,
      "loss": 0.1864,
      "step": 1837
    },
    {
      "epoch": 0.8282077278359806,
      "grad_norm": 0.961626291646709,
      "learning_rate": 3.9230024441295394e-07,
      "loss": 0.1713,
      "step": 1838
    },
    {
      "epoch": 0.8286583305170666,
      "grad_norm": 0.8826493854231543,
      "learning_rate": 3.9029889529724113e-07,
      "loss": 0.1637,
      "step": 1839
    },
    {
      "epoch": 0.8291089331981525,
      "grad_norm": 0.91167006225646,
      "learning_rate": 3.8830223193442345e-07,
      "loss": 0.1588,
      "step": 1840
    },
    {
      "epoch": 0.8295595358792385,
      "grad_norm": 0.9264347983159924,
      "learning_rate": 3.863102587591919e-07,
      "loss": 0.16,
      "step": 1841
    },
    {
      "epoch": 0.8300101385603245,
      "grad_norm": 0.951334590876306,
      "learning_rate": 3.84322980195819e-07,
      "loss": 0.1683,
      "step": 1842
    },
    {
      "epoch": 0.8304607412414103,
      "grad_norm": 0.9659033816073591,
      "learning_rate": 3.823404006581513e-07,
      "loss": 0.1703,
      "step": 1843
    },
    {
      "epoch": 0.8309113439224963,
      "grad_norm": 0.9209942273315058,
      "learning_rate": 3.8036252454959844e-07,
      "loss": 0.1662,
      "step": 1844
    },
    {
      "epoch": 0.8313619466035823,
      "grad_norm": 0.9281131852365946,
      "learning_rate": 3.7838935626312246e-07,
      "loss": 0.1646,
      "step": 1845
    },
    {
      "epoch": 0.8318125492846682,
      "grad_norm": 0.9439773737834088,
      "learning_rate": 3.764209001812316e-07,
      "loss": 0.1748,
      "step": 1846
    },
    {
      "epoch": 0.8322631519657542,
      "grad_norm": 0.9266237935276338,
      "learning_rate": 3.7445716067596506e-07,
      "loss": 0.1731,
      "step": 1847
    },
    {
      "epoch": 0.8327137546468402,
      "grad_norm": 0.924011205749328,
      "learning_rate": 3.72498142108888e-07,
      "loss": 0.166,
      "step": 1848
    },
    {
      "epoch": 0.8331643573279262,
      "grad_norm": 0.9975906771664418,
      "learning_rate": 3.705438488310792e-07,
      "loss": 0.175,
      "step": 1849
    },
    {
      "epoch": 0.833614960009012,
      "grad_norm": 0.9870663130220605,
      "learning_rate": 3.6859428518312394e-07,
      "loss": 0.177,
      "step": 1850
    },
    {
      "epoch": 0.834065562690098,
      "grad_norm": 1.0589620302719522,
      "learning_rate": 3.666494554951014e-07,
      "loss": 0.1731,
      "step": 1851
    },
    {
      "epoch": 0.834516165371184,
      "grad_norm": 0.9792308554611852,
      "learning_rate": 3.6470936408657647e-07,
      "loss": 0.1544,
      "step": 1852
    },
    {
      "epoch": 0.8349667680522699,
      "grad_norm": 1.0004784542843967,
      "learning_rate": 3.6277401526659067e-07,
      "loss": 0.1729,
      "step": 1853
    },
    {
      "epoch": 0.8354173707333559,
      "grad_norm": 1.1253575928698794,
      "learning_rate": 3.6084341333365135e-07,
      "loss": 0.1718,
      "step": 1854
    },
    {
      "epoch": 0.8358679734144419,
      "grad_norm": 0.9624239214809235,
      "learning_rate": 3.58917562575723e-07,
      "loss": 0.1791,
      "step": 1855
    },
    {
      "epoch": 0.8363185760955277,
      "grad_norm": 0.9606501614996009,
      "learning_rate": 3.569964672702178e-07,
      "loss": 0.1714,
      "step": 1856
    },
    {
      "epoch": 0.8367691787766137,
      "grad_norm": 0.9759429493736111,
      "learning_rate": 3.550801316839858e-07,
      "loss": 0.1805,
      "step": 1857
    },
    {
      "epoch": 0.8372197814576997,
      "grad_norm": 0.9783279630837232,
      "learning_rate": 3.531685600733051e-07,
      "loss": 0.1681,
      "step": 1858
    },
    {
      "epoch": 0.8376703841387856,
      "grad_norm": 0.9918620830896449,
      "learning_rate": 3.5126175668387275e-07,
      "loss": 0.1843,
      "step": 1859
    },
    {
      "epoch": 0.8381209868198716,
      "grad_norm": 1.0541404809751138,
      "learning_rate": 3.4935972575079524e-07,
      "loss": 0.1868,
      "step": 1860
    },
    {
      "epoch": 0.8385715895009576,
      "grad_norm": 0.8875615125272613,
      "learning_rate": 3.474624714985805e-07,
      "loss": 0.1542,
      "step": 1861
    },
    {
      "epoch": 0.8390221921820434,
      "grad_norm": 0.947000239574121,
      "learning_rate": 3.455699981411259e-07,
      "loss": 0.1639,
      "step": 1862
    },
    {
      "epoch": 0.8394727948631294,
      "grad_norm": 0.9871624452171542,
      "learning_rate": 3.436823098817102e-07,
      "loss": 0.1717,
      "step": 1863
    },
    {
      "epoch": 0.8399233975442154,
      "grad_norm": 0.9771759407766166,
      "learning_rate": 3.417994109129852e-07,
      "loss": 0.1602,
      "step": 1864
    },
    {
      "epoch": 0.8403740002253013,
      "grad_norm": 0.9195044006987618,
      "learning_rate": 3.3992130541696336e-07,
      "loss": 0.1652,
      "step": 1865
    },
    {
      "epoch": 0.8408246029063873,
      "grad_norm": 0.9445742965717538,
      "learning_rate": 3.3804799756501335e-07,
      "loss": 0.1721,
      "step": 1866
    },
    {
      "epoch": 0.8412752055874733,
      "grad_norm": 0.9362275436972677,
      "learning_rate": 3.3617949151784623e-07,
      "loss": 0.1703,
      "step": 1867
    },
    {
      "epoch": 0.8417258082685593,
      "grad_norm": 1.014145441841758,
      "learning_rate": 3.343157914255085e-07,
      "loss": 0.1819,
      "step": 1868
    },
    {
      "epoch": 0.8421764109496451,
      "grad_norm": 0.9314737099222552,
      "learning_rate": 3.3245690142737236e-07,
      "loss": 0.1756,
      "step": 1869
    },
    {
      "epoch": 0.8426270136307311,
      "grad_norm": 0.9722306559540486,
      "learning_rate": 3.306028256521265e-07,
      "loss": 0.1699,
      "step": 1870
    },
    {
      "epoch": 0.8430776163118171,
      "grad_norm": 0.9672581166734934,
      "learning_rate": 3.287535682177667e-07,
      "loss": 0.1799,
      "step": 1871
    },
    {
      "epoch": 0.843528218992903,
      "grad_norm": 0.9858629820179523,
      "learning_rate": 3.2690913323158795e-07,
      "loss": 0.167,
      "step": 1872
    },
    {
      "epoch": 0.843978821673989,
      "grad_norm": 0.9771711710226948,
      "learning_rate": 3.2506952479017417e-07,
      "loss": 0.1755,
      "step": 1873
    },
    {
      "epoch": 0.844429424355075,
      "grad_norm": 0.9702797005479236,
      "learning_rate": 3.2323474697938727e-07,
      "loss": 0.1807,
      "step": 1874
    },
    {
      "epoch": 0.8448800270361608,
      "grad_norm": 0.9728669390673415,
      "learning_rate": 3.214048038743622e-07,
      "loss": 0.1735,
      "step": 1875
    },
    {
      "epoch": 0.8453306297172468,
      "grad_norm": 0.9032972459456295,
      "learning_rate": 3.1957969953949506e-07,
      "loss": 0.1514,
      "step": 1876
    },
    {
      "epoch": 0.8457812323983328,
      "grad_norm": 1.0259745438344021,
      "learning_rate": 3.1775943802843546e-07,
      "loss": 0.1889,
      "step": 1877
    },
    {
      "epoch": 0.8462318350794187,
      "grad_norm": 0.9278308411260581,
      "learning_rate": 3.1594402338407633e-07,
      "loss": 0.1619,
      "step": 1878
    },
    {
      "epoch": 0.8466824377605047,
      "grad_norm": 0.9486461271581993,
      "learning_rate": 3.141334596385448e-07,
      "loss": 0.1701,
      "step": 1879
    },
    {
      "epoch": 0.8471330404415907,
      "grad_norm": 0.9264937573351875,
      "learning_rate": 3.12327750813195e-07,
      "loss": 0.16,
      "step": 1880
    },
    {
      "epoch": 0.8475836431226765,
      "grad_norm": 0.9063889096458639,
      "learning_rate": 3.105269009185974e-07,
      "loss": 0.153,
      "step": 1881
    },
    {
      "epoch": 0.8480342458037625,
      "grad_norm": 0.9345172159415808,
      "learning_rate": 3.087309139545311e-07,
      "loss": 0.1706,
      "step": 1882
    },
    {
      "epoch": 0.8484848484848485,
      "grad_norm": 0.884600902211049,
      "learning_rate": 3.0693979390997333e-07,
      "loss": 0.1579,
      "step": 1883
    },
    {
      "epoch": 0.8489354511659344,
      "grad_norm": 0.9278539998900202,
      "learning_rate": 3.0515354476309293e-07,
      "loss": 0.1617,
      "step": 1884
    },
    {
      "epoch": 0.8493860538470204,
      "grad_norm": 0.9492452968446602,
      "learning_rate": 3.033721704812395e-07,
      "loss": 0.1676,
      "step": 1885
    },
    {
      "epoch": 0.8498366565281064,
      "grad_norm": 0.9832023339183169,
      "learning_rate": 3.0159567502093535e-07,
      "loss": 0.179,
      "step": 1886
    },
    {
      "epoch": 0.8502872592091923,
      "grad_norm": 0.9782668678333437,
      "learning_rate": 2.9982406232786614e-07,
      "loss": 0.1742,
      "step": 1887
    },
    {
      "epoch": 0.8507378618902782,
      "grad_norm": 0.9701650943366985,
      "learning_rate": 2.9805733633687467e-07,
      "loss": 0.175,
      "step": 1888
    },
    {
      "epoch": 0.8511884645713642,
      "grad_norm": 0.9651583163077686,
      "learning_rate": 2.9629550097194787e-07,
      "loss": 0.1782,
      "step": 1889
    },
    {
      "epoch": 0.8516390672524502,
      "grad_norm": 0.9024669829464759,
      "learning_rate": 2.9453856014621224e-07,
      "loss": 0.1623,
      "step": 1890
    },
    {
      "epoch": 0.8520896699335361,
      "grad_norm": 1.0192621240776767,
      "learning_rate": 2.9278651776192073e-07,
      "loss": 0.1936,
      "step": 1891
    },
    {
      "epoch": 0.8525402726146221,
      "grad_norm": 0.9918084930237235,
      "learning_rate": 2.910393777104481e-07,
      "loss": 0.1751,
      "step": 1892
    },
    {
      "epoch": 0.852990875295708,
      "grad_norm": 0.9650070719897159,
      "learning_rate": 2.892971438722822e-07,
      "loss": 0.1772,
      "step": 1893
    },
    {
      "epoch": 0.8534414779767939,
      "grad_norm": 0.9685100244625674,
      "learning_rate": 2.8755982011701183e-07,
      "loss": 0.1767,
      "step": 1894
    },
    {
      "epoch": 0.8538920806578799,
      "grad_norm": 0.9556958216529681,
      "learning_rate": 2.8582741030332095e-07,
      "loss": 0.1779,
      "step": 1895
    },
    {
      "epoch": 0.8543426833389659,
      "grad_norm": 0.9757391441636464,
      "learning_rate": 2.840999182789797e-07,
      "loss": 0.1679,
      "step": 1896
    },
    {
      "epoch": 0.8547932860200518,
      "grad_norm": 0.9672040065820859,
      "learning_rate": 2.823773478808348e-07,
      "loss": 0.1762,
      "step": 1897
    },
    {
      "epoch": 0.8552438887011378,
      "grad_norm": 0.9500345661356947,
      "learning_rate": 2.806597029348018e-07,
      "loss": 0.1616,
      "step": 1898
    },
    {
      "epoch": 0.8556944913822238,
      "grad_norm": 0.8833689073169025,
      "learning_rate": 2.7894698725585866e-07,
      "loss": 0.1587,
      "step": 1899
    },
    {
      "epoch": 0.8561450940633096,
      "grad_norm": 0.9467995781424163,
      "learning_rate": 2.772392046480324e-07,
      "loss": 0.1674,
      "step": 1900
    },
    {
      "epoch": 0.8565956967443956,
      "grad_norm": 0.9205995331832161,
      "learning_rate": 2.755363589043944e-07,
      "loss": 0.1634,
      "step": 1901
    },
    {
      "epoch": 0.8570462994254816,
      "grad_norm": 0.9563426840948828,
      "learning_rate": 2.738384538070518e-07,
      "loss": 0.1741,
      "step": 1902
    },
    {
      "epoch": 0.8574969021065675,
      "grad_norm": 0.9669901785480284,
      "learning_rate": 2.7214549312713723e-07,
      "loss": 0.1659,
      "step": 1903
    },
    {
      "epoch": 0.8579475047876535,
      "grad_norm": 0.9743277211364876,
      "learning_rate": 2.7045748062480254e-07,
      "loss": 0.1732,
      "step": 1904
    },
    {
      "epoch": 0.8583981074687395,
      "grad_norm": 0.9411003828673956,
      "learning_rate": 2.6877442004920873e-07,
      "loss": 0.1645,
      "step": 1905
    },
    {
      "epoch": 0.8588487101498254,
      "grad_norm": 0.9734653439920148,
      "learning_rate": 2.6709631513851834e-07,
      "loss": 0.1778,
      "step": 1906
    },
    {
      "epoch": 0.8592993128309113,
      "grad_norm": 0.9312433368537456,
      "learning_rate": 2.654231696198878e-07,
      "loss": 0.164,
      "step": 1907
    },
    {
      "epoch": 0.8597499155119973,
      "grad_norm": 0.9692174115782383,
      "learning_rate": 2.6375498720945717e-07,
      "loss": 0.1718,
      "step": 1908
    },
    {
      "epoch": 0.8602005181930833,
      "grad_norm": 0.8951997876415702,
      "learning_rate": 2.620917716123444e-07,
      "loss": 0.1693,
      "step": 1909
    },
    {
      "epoch": 0.8606511208741692,
      "grad_norm": 0.9778603788324381,
      "learning_rate": 2.604335265226354e-07,
      "loss": 0.1827,
      "step": 1910
    },
    {
      "epoch": 0.8611017235552552,
      "grad_norm": 1.0493504549340391,
      "learning_rate": 2.587802556233765e-07,
      "loss": 0.1845,
      "step": 1911
    },
    {
      "epoch": 0.8615523262363411,
      "grad_norm": 0.9252258432234339,
      "learning_rate": 2.571319625865662e-07,
      "loss": 0.1636,
      "step": 1912
    },
    {
      "epoch": 0.862002928917427,
      "grad_norm": 1.015522402651877,
      "learning_rate": 2.5548865107314606e-07,
      "loss": 0.1794,
      "step": 1913
    },
    {
      "epoch": 0.862453531598513,
      "grad_norm": 0.9275781007378919,
      "learning_rate": 2.5385032473299433e-07,
      "loss": 0.1751,
      "step": 1914
    },
    {
      "epoch": 0.862904134279599,
      "grad_norm": 0.9128078845993659,
      "learning_rate": 2.522169872049174e-07,
      "loss": 0.1589,
      "step": 1915
    },
    {
      "epoch": 0.8633547369606849,
      "grad_norm": 0.8879718361876989,
      "learning_rate": 2.5058864211664064e-07,
      "loss": 0.1706,
      "step": 1916
    },
    {
      "epoch": 0.8638053396417709,
      "grad_norm": 0.9372693909164671,
      "learning_rate": 2.4896529308479966e-07,
      "loss": 0.1698,
      "step": 1917
    },
    {
      "epoch": 0.8642559423228569,
      "grad_norm": 1.0087108719862292,
      "learning_rate": 2.4734694371493507e-07,
      "loss": 0.1661,
      "step": 1918
    },
    {
      "epoch": 0.8647065450039427,
      "grad_norm": 0.9844600152081066,
      "learning_rate": 2.4573359760148354e-07,
      "loss": 0.1842,
      "step": 1919
    },
    {
      "epoch": 0.8651571476850287,
      "grad_norm": 1.0349326975424789,
      "learning_rate": 2.441252583277678e-07,
      "loss": 0.1811,
      "step": 1920
    },
    {
      "epoch": 0.8656077503661147,
      "grad_norm": 0.9821007975011048,
      "learning_rate": 2.425219294659908e-07,
      "loss": 0.1825,
      "step": 1921
    },
    {
      "epoch": 0.8660583530472006,
      "grad_norm": 0.9446142750611491,
      "learning_rate": 2.409236145772276e-07,
      "loss": 0.1714,
      "step": 1922
    },
    {
      "epoch": 0.8665089557282866,
      "grad_norm": 0.9596367414516151,
      "learning_rate": 2.393303172114159e-07,
      "loss": 0.1732,
      "step": 1923
    },
    {
      "epoch": 0.8669595584093726,
      "grad_norm": 0.9551816035996794,
      "learning_rate": 2.377420409073497e-07,
      "loss": 0.1671,
      "step": 1924
    },
    {
      "epoch": 0.8674101610904585,
      "grad_norm": 0.8993318779686909,
      "learning_rate": 2.3615878919267116e-07,
      "loss": 0.171,
      "step": 1925
    },
    {
      "epoch": 0.8678607637715444,
      "grad_norm": 1.021525400453317,
      "learning_rate": 2.345805655838626e-07,
      "loss": 0.1914,
      "step": 1926
    },
    {
      "epoch": 0.8683113664526304,
      "grad_norm": 0.9314956686766991,
      "learning_rate": 2.3300737358623843e-07,
      "loss": 0.1746,
      "step": 1927
    },
    {
      "epoch": 0.8687619691337164,
      "grad_norm": 0.9964051166260204,
      "learning_rate": 2.314392166939375e-07,
      "loss": 0.1773,
      "step": 1928
    },
    {
      "epoch": 0.8692125718148023,
      "grad_norm": 1.0162505058504328,
      "learning_rate": 2.2987609838991536e-07,
      "loss": 0.1867,
      "step": 1929
    },
    {
      "epoch": 0.8696631744958883,
      "grad_norm": 0.9257283800837275,
      "learning_rate": 2.2831802214593774e-07,
      "loss": 0.1731,
      "step": 1930
    },
    {
      "epoch": 0.8701137771769742,
      "grad_norm": 0.9677329540924063,
      "learning_rate": 2.2676499142257002e-07,
      "loss": 0.1735,
      "step": 1931
    },
    {
      "epoch": 0.8705643798580601,
      "grad_norm": 1.000711690360383,
      "learning_rate": 2.2521700966917276e-07,
      "loss": 0.1754,
      "step": 1932
    },
    {
      "epoch": 0.8710149825391461,
      "grad_norm": 0.987391302910459,
      "learning_rate": 2.23674080323891e-07,
      "loss": 0.1743,
      "step": 1933
    },
    {
      "epoch": 0.8714655852202321,
      "grad_norm": 0.9747308865230088,
      "learning_rate": 2.221362068136493e-07,
      "loss": 0.1608,
      "step": 1934
    },
    {
      "epoch": 0.871916187901318,
      "grad_norm": 0.9376304181926186,
      "learning_rate": 2.2060339255414232e-07,
      "loss": 0.1628,
      "step": 1935
    },
    {
      "epoch": 0.872366790582404,
      "grad_norm": 0.9334000321457883,
      "learning_rate": 2.190756409498282e-07,
      "loss": 0.1597,
      "step": 1936
    },
    {
      "epoch": 0.87281739326349,
      "grad_norm": 0.9008641202830003,
      "learning_rate": 2.175529553939204e-07,
      "loss": 0.1596,
      "step": 1937
    },
    {
      "epoch": 0.8732679959445758,
      "grad_norm": 0.9367208891666635,
      "learning_rate": 2.1603533926838088e-07,
      "loss": 0.1693,
      "step": 1938
    },
    {
      "epoch": 0.8737185986256618,
      "grad_norm": 0.939496478640009,
      "learning_rate": 2.1452279594391167e-07,
      "loss": 0.1668,
      "step": 1939
    },
    {
      "epoch": 0.8741692013067478,
      "grad_norm": 0.9209952395233147,
      "learning_rate": 2.1301532877994747e-07,
      "loss": 0.1676,
      "step": 1940
    },
    {
      "epoch": 0.8746198039878337,
      "grad_norm": 0.9103353045426509,
      "learning_rate": 2.1151294112464997e-07,
      "loss": 0.1691,
      "step": 1941
    },
    {
      "epoch": 0.8750704066689197,
      "grad_norm": 0.8665460511350714,
      "learning_rate": 2.1001563631489807e-07,
      "loss": 0.1487,
      "step": 1942
    },
    {
      "epoch": 0.8755210093500057,
      "grad_norm": 0.9679891948748914,
      "learning_rate": 2.0852341767628182e-07,
      "loss": 0.169,
      "step": 1943
    },
    {
      "epoch": 0.8759716120310916,
      "grad_norm": 0.9496329960942475,
      "learning_rate": 2.0703628852309336e-07,
      "loss": 0.1699,
      "step": 1944
    },
    {
      "epoch": 0.8764222147121775,
      "grad_norm": 0.9084124393445112,
      "learning_rate": 2.0555425215832176e-07,
      "loss": 0.1448,
      "step": 1945
    },
    {
      "epoch": 0.8768728173932635,
      "grad_norm": 0.9325294190123458,
      "learning_rate": 2.0407731187364556e-07,
      "loss": 0.1611,
      "step": 1946
    },
    {
      "epoch": 0.8773234200743495,
      "grad_norm": 0.9608168170515962,
      "learning_rate": 2.026054709494235e-07,
      "loss": 0.1637,
      "step": 1947
    },
    {
      "epoch": 0.8777740227554354,
      "grad_norm": 0.9508933509137032,
      "learning_rate": 2.0113873265468875e-07,
      "loss": 0.1637,
      "step": 1948
    },
    {
      "epoch": 0.8782246254365214,
      "grad_norm": 0.9627415360121703,
      "learning_rate": 1.996771002471415e-07,
      "loss": 0.171,
      "step": 1949
    },
    {
      "epoch": 0.8786752281176073,
      "grad_norm": 0.9269279051004062,
      "learning_rate": 1.9822057697314102e-07,
      "loss": 0.158,
      "step": 1950
    },
    {
      "epoch": 0.8791258307986932,
      "grad_norm": 0.9975744188618978,
      "learning_rate": 1.9676916606769874e-07,
      "loss": 0.1733,
      "step": 1951
    },
    {
      "epoch": 0.8795764334797792,
      "grad_norm": 0.9907918900931711,
      "learning_rate": 1.9532287075447325e-07,
      "loss": 0.1717,
      "step": 1952
    },
    {
      "epoch": 0.8800270361608652,
      "grad_norm": 0.9029717491407483,
      "learning_rate": 1.9388169424575802e-07,
      "loss": 0.1526,
      "step": 1953
    },
    {
      "epoch": 0.8804776388419511,
      "grad_norm": 0.9924216757155009,
      "learning_rate": 1.9244563974247953e-07,
      "loss": 0.1871,
      "step": 1954
    },
    {
      "epoch": 0.8809282415230371,
      "grad_norm": 1.0046395234515613,
      "learning_rate": 1.910147104341875e-07,
      "loss": 0.1677,
      "step": 1955
    },
    {
      "epoch": 0.881378844204123,
      "grad_norm": 0.8428374340697592,
      "learning_rate": 1.8958890949904802e-07,
      "loss": 0.1428,
      "step": 1956
    },
    {
      "epoch": 0.8818294468852089,
      "grad_norm": 0.8916062368929274,
      "learning_rate": 1.881682401038379e-07,
      "loss": 0.1627,
      "step": 1957
    },
    {
      "epoch": 0.8822800495662949,
      "grad_norm": 0.9133925219285579,
      "learning_rate": 1.8675270540393532e-07,
      "loss": 0.1634,
      "step": 1958
    },
    {
      "epoch": 0.8827306522473809,
      "grad_norm": 0.9603725651655006,
      "learning_rate": 1.8534230854331454e-07,
      "loss": 0.1589,
      "step": 1959
    },
    {
      "epoch": 0.8831812549284668,
      "grad_norm": 0.8832900144509541,
      "learning_rate": 1.8393705265453838e-07,
      "loss": 0.1526,
      "step": 1960
    },
    {
      "epoch": 0.8836318576095528,
      "grad_norm": 0.9656796733108457,
      "learning_rate": 1.8253694085875047e-07,
      "loss": 0.1632,
      "step": 1961
    },
    {
      "epoch": 0.8840824602906387,
      "grad_norm": 1.0103093827826655,
      "learning_rate": 1.8114197626567105e-07,
      "loss": 0.1801,
      "step": 1962
    },
    {
      "epoch": 0.8845330629717247,
      "grad_norm": 0.9605960303059552,
      "learning_rate": 1.7975216197358648e-07,
      "loss": 0.174,
      "step": 1963
    },
    {
      "epoch": 0.8849836656528106,
      "grad_norm": 1.0141937286324803,
      "learning_rate": 1.7836750106934475e-07,
      "loss": 0.1691,
      "step": 1964
    },
    {
      "epoch": 0.8854342683338966,
      "grad_norm": 0.9434775531410926,
      "learning_rate": 1.7698799662834776e-07,
      "loss": 0.1602,
      "step": 1965
    },
    {
      "epoch": 0.8858848710149826,
      "grad_norm": 0.9267902111682871,
      "learning_rate": 1.7561365171454488e-07,
      "loss": 0.1601,
      "step": 1966
    },
    {
      "epoch": 0.8863354736960685,
      "grad_norm": 0.9224140721738877,
      "learning_rate": 1.7424446938042517e-07,
      "loss": 0.1638,
      "step": 1967
    },
    {
      "epoch": 0.8867860763771545,
      "grad_norm": 0.979554116800001,
      "learning_rate": 1.7288045266701247e-07,
      "loss": 0.178,
      "step": 1968
    },
    {
      "epoch": 0.8872366790582404,
      "grad_norm": 0.9415277153144294,
      "learning_rate": 1.7152160460385703e-07,
      "loss": 0.1578,
      "step": 1969
    },
    {
      "epoch": 0.8876872817393263,
      "grad_norm": 0.9751633874844109,
      "learning_rate": 1.701679282090285e-07,
      "loss": 0.1648,
      "step": 1970
    },
    {
      "epoch": 0.8881378844204123,
      "grad_norm": 0.9433218610454983,
      "learning_rate": 1.6881942648911077e-07,
      "loss": 0.1517,
      "step": 1971
    },
    {
      "epoch": 0.8885884871014983,
      "grad_norm": 0.9790183372335409,
      "learning_rate": 1.6747610243919437e-07,
      "loss": 0.1855,
      "step": 1972
    },
    {
      "epoch": 0.8890390897825842,
      "grad_norm": 0.9729615081723881,
      "learning_rate": 1.661379590428705e-07,
      "loss": 0.1692,
      "step": 1973
    },
    {
      "epoch": 0.8894896924636702,
      "grad_norm": 1.0050367822916513,
      "learning_rate": 1.6480499927222283e-07,
      "loss": 0.1759,
      "step": 1974
    },
    {
      "epoch": 0.8899402951447561,
      "grad_norm": 0.9976121317605783,
      "learning_rate": 1.6347722608782284e-07,
      "loss": 0.1761,
      "step": 1975
    },
    {
      "epoch": 0.890390897825842,
      "grad_norm": 0.9559950177435562,
      "learning_rate": 1.6215464243872186e-07,
      "loss": 0.1637,
      "step": 1976
    },
    {
      "epoch": 0.890841500506928,
      "grad_norm": 0.9723990375383502,
      "learning_rate": 1.608372512624448e-07,
      "loss": 0.1687,
      "step": 1977
    },
    {
      "epoch": 0.891292103188014,
      "grad_norm": 0.9655332499571794,
      "learning_rate": 1.595250554849842e-07,
      "loss": 0.1677,
      "step": 1978
    },
    {
      "epoch": 0.8917427058690999,
      "grad_norm": 0.9214762769979025,
      "learning_rate": 1.5821805802079343e-07,
      "loss": 0.1608,
      "step": 1979
    },
    {
      "epoch": 0.8921933085501859,
      "grad_norm": 0.9136099662881566,
      "learning_rate": 1.5691626177277986e-07,
      "loss": 0.1781,
      "step": 1980
    },
    {
      "epoch": 0.8926439112312718,
      "grad_norm": 0.9262022249621232,
      "learning_rate": 1.5561966963229925e-07,
      "loss": 0.1755,
      "step": 1981
    },
    {
      "epoch": 0.8930945139123577,
      "grad_norm": 0.9415300421061215,
      "learning_rate": 1.5432828447914743e-07,
      "loss": 0.1671,
      "step": 1982
    },
    {
      "epoch": 0.8935451165934437,
      "grad_norm": 0.9750913944440511,
      "learning_rate": 1.5304210918155677e-07,
      "loss": 0.1818,
      "step": 1983
    },
    {
      "epoch": 0.8939957192745297,
      "grad_norm": 0.916736823632913,
      "learning_rate": 1.5176114659618796e-07,
      "loss": 0.1506,
      "step": 1984
    },
    {
      "epoch": 0.8944463219556157,
      "grad_norm": 0.9672707724576233,
      "learning_rate": 1.5048539956812324e-07,
      "loss": 0.1663,
      "step": 1985
    },
    {
      "epoch": 0.8948969246367016,
      "grad_norm": 0.9182721543696385,
      "learning_rate": 1.4921487093086134e-07,
      "loss": 0.1699,
      "step": 1986
    },
    {
      "epoch": 0.8953475273177876,
      "grad_norm": 0.8891077554280398,
      "learning_rate": 1.4794956350631106e-07,
      "loss": 0.1663,
      "step": 1987
    },
    {
      "epoch": 0.8957981299988735,
      "grad_norm": 0.9034991158889414,
      "learning_rate": 1.4668948010478358e-07,
      "loss": 0.168,
      "step": 1988
    },
    {
      "epoch": 0.8962487326799594,
      "grad_norm": 1.0096026558664806,
      "learning_rate": 1.4543462352498844e-07,
      "loss": 0.171,
      "step": 1989
    },
    {
      "epoch": 0.8966993353610454,
      "grad_norm": 0.9336033398750972,
      "learning_rate": 1.4418499655402512e-07,
      "loss": 0.1723,
      "step": 1990
    },
    {
      "epoch": 0.8971499380421314,
      "grad_norm": 0.9866388504700837,
      "learning_rate": 1.4294060196737874e-07,
      "loss": 0.1649,
      "step": 1991
    },
    {
      "epoch": 0.8976005407232173,
      "grad_norm": 0.9743774072891516,
      "learning_rate": 1.417014425289126e-07,
      "loss": 0.1625,
      "step": 1992
    },
    {
      "epoch": 0.8980511434043033,
      "grad_norm": 0.90538670491052,
      "learning_rate": 1.4046752099086236e-07,
      "loss": 0.1538,
      "step": 1993
    },
    {
      "epoch": 0.8985017460853892,
      "grad_norm": 0.9378839919331537,
      "learning_rate": 1.3923884009382994e-07,
      "loss": 0.1532,
      "step": 1994
    },
    {
      "epoch": 0.8989523487664751,
      "grad_norm": 0.961745818647385,
      "learning_rate": 1.380154025667782e-07,
      "loss": 0.1649,
      "step": 1995
    },
    {
      "epoch": 0.8994029514475611,
      "grad_norm": 0.9689054481835896,
      "learning_rate": 1.367972111270241e-07,
      "loss": 0.1724,
      "step": 1996
    },
    {
      "epoch": 0.8998535541286471,
      "grad_norm": 0.9601026525696907,
      "learning_rate": 1.3558426848023165e-07,
      "loss": 0.1674,
      "step": 1997
    },
    {
      "epoch": 0.900304156809733,
      "grad_norm": 0.9116251944923741,
      "learning_rate": 1.3437657732040783e-07,
      "loss": 0.1675,
      "step": 1998
    },
    {
      "epoch": 0.900754759490819,
      "grad_norm": 0.9712578312105717,
      "learning_rate": 1.3317414032989668e-07,
      "loss": 0.1795,
      "step": 1999
    },
    {
      "epoch": 0.901205362171905,
      "grad_norm": 0.9513325556706582,
      "learning_rate": 1.3197696017937106e-07,
      "loss": 0.1841,
      "step": 2000
    },
    {
      "epoch": 0.901205362171905,
      "eval_loss": 0.17061425745487213,
      "eval_runtime": 59.2934,
      "eval_samples_per_second": 24.202,
      "eval_steps_per_second": 3.036,
      "step": 2000
    },
    {
      "epoch": 0.9016559648529908,
      "grad_norm": 0.9117132984829037,
      "learning_rate": 1.3078503952782845e-07,
      "loss": 0.1618,
      "step": 2001
    },
    {
      "epoch": 0.9021065675340768,
      "grad_norm": 0.9472154091042211,
      "learning_rate": 1.2959838102258537e-07,
      "loss": 0.1862,
      "step": 2002
    },
    {
      "epoch": 0.9025571702151628,
      "grad_norm": 0.9494247550660043,
      "learning_rate": 1.2841698729927022e-07,
      "loss": 0.1557,
      "step": 2003
    },
    {
      "epoch": 0.9030077728962488,
      "grad_norm": 0.9157731618730593,
      "learning_rate": 1.272408609818182e-07,
      "loss": 0.1603,
      "step": 2004
    },
    {
      "epoch": 0.9034583755773347,
      "grad_norm": 0.950573888351368,
      "learning_rate": 1.2607000468246533e-07,
      "loss": 0.1678,
      "step": 2005
    },
    {
      "epoch": 0.9039089782584206,
      "grad_norm": 0.961300360072569,
      "learning_rate": 1.2490442100174278e-07,
      "loss": 0.1777,
      "step": 2006
    },
    {
      "epoch": 0.9043595809395066,
      "grad_norm": 0.8913003543296754,
      "learning_rate": 1.237441125284708e-07,
      "loss": 0.1678,
      "step": 2007
    },
    {
      "epoch": 0.9048101836205925,
      "grad_norm": 0.9608134144134601,
      "learning_rate": 1.2258908183975322e-07,
      "loss": 0.1775,
      "step": 2008
    },
    {
      "epoch": 0.9052607863016785,
      "grad_norm": 0.9596706472769803,
      "learning_rate": 1.2143933150097154e-07,
      "loss": 0.165,
      "step": 2009
    },
    {
      "epoch": 0.9057113889827645,
      "grad_norm": 0.9328106883096551,
      "learning_rate": 1.2029486406577972e-07,
      "loss": 0.1687,
      "step": 2010
    },
    {
      "epoch": 0.9061619916638504,
      "grad_norm": 0.9509999156498098,
      "learning_rate": 1.191556820760978e-07,
      "loss": 0.1635,
      "step": 2011
    },
    {
      "epoch": 0.9066125943449364,
      "grad_norm": 0.986143322145186,
      "learning_rate": 1.1802178806210624e-07,
      "loss": 0.1646,
      "step": 2012
    },
    {
      "epoch": 0.9070631970260223,
      "grad_norm": 0.9252282348305862,
      "learning_rate": 1.1689318454224191e-07,
      "loss": 0.1594,
      "step": 2013
    },
    {
      "epoch": 0.9075137997071082,
      "grad_norm": 0.9665747400156657,
      "learning_rate": 1.1576987402318884e-07,
      "loss": 0.1686,
      "step": 2014
    },
    {
      "epoch": 0.9079644023881942,
      "grad_norm": 0.9242552281663089,
      "learning_rate": 1.1465185899987797e-07,
      "loss": 0.1653,
      "step": 2015
    },
    {
      "epoch": 0.9084150050692802,
      "grad_norm": 0.9338454099056986,
      "learning_rate": 1.1353914195547655e-07,
      "loss": 0.1664,
      "step": 2016
    },
    {
      "epoch": 0.9088656077503661,
      "grad_norm": 0.9488715062833334,
      "learning_rate": 1.1243172536138547e-07,
      "loss": 0.167,
      "step": 2017
    },
    {
      "epoch": 0.909316210431452,
      "grad_norm": 0.9192528840903248,
      "learning_rate": 1.1132961167723305e-07,
      "loss": 0.1602,
      "step": 2018
    },
    {
      "epoch": 0.909766813112538,
      "grad_norm": 0.9358983791063447,
      "learning_rate": 1.1023280335086956e-07,
      "loss": 0.1786,
      "step": 2019
    },
    {
      "epoch": 0.9102174157936239,
      "grad_norm": 0.8800096012948293,
      "learning_rate": 1.091413028183616e-07,
      "loss": 0.1551,
      "step": 2020
    },
    {
      "epoch": 0.9106680184747099,
      "grad_norm": 0.9819759586485249,
      "learning_rate": 1.0805511250398748e-07,
      "loss": 0.178,
      "step": 2021
    },
    {
      "epoch": 0.9111186211557959,
      "grad_norm": 0.9300351124591425,
      "learning_rate": 1.06974234820231e-07,
      "loss": 0.1677,
      "step": 2022
    },
    {
      "epoch": 0.9115692238368819,
      "grad_norm": 0.9574289668603256,
      "learning_rate": 1.0589867216777544e-07,
      "loss": 0.1783,
      "step": 2023
    },
    {
      "epoch": 0.9120198265179678,
      "grad_norm": 0.9480950786557066,
      "learning_rate": 1.0482842693550044e-07,
      "loss": 0.1598,
      "step": 2024
    },
    {
      "epoch": 0.9124704291990537,
      "grad_norm": 0.9389594010578164,
      "learning_rate": 1.0376350150047427e-07,
      "loss": 0.1748,
      "step": 2025
    },
    {
      "epoch": 0.9129210318801397,
      "grad_norm": 0.9902783035113683,
      "learning_rate": 1.0270389822795073e-07,
      "loss": 0.1759,
      "step": 2026
    },
    {
      "epoch": 0.9133716345612256,
      "grad_norm": 0.9010340950200544,
      "learning_rate": 1.0164961947136232e-07,
      "loss": 0.1666,
      "step": 2027
    },
    {
      "epoch": 0.9138222372423116,
      "grad_norm": 0.9640758949806947,
      "learning_rate": 1.0060066757231535e-07,
      "loss": 0.1617,
      "step": 2028
    },
    {
      "epoch": 0.9142728399233976,
      "grad_norm": 0.9466267617525821,
      "learning_rate": 9.955704486058482e-08,
      "loss": 0.1728,
      "step": 2029
    },
    {
      "epoch": 0.9147234426044835,
      "grad_norm": 0.9468720807805653,
      "learning_rate": 9.85187536541099e-08,
      "loss": 0.1762,
      "step": 2030
    },
    {
      "epoch": 0.9151740452855694,
      "grad_norm": 0.9519779843009848,
      "learning_rate": 9.748579625898758e-08,
      "loss": 0.159,
      "step": 2031
    },
    {
      "epoch": 0.9156246479666554,
      "grad_norm": 0.935520409067934,
      "learning_rate": 9.645817496946902e-08,
      "loss": 0.1534,
      "step": 2032
    },
    {
      "epoch": 0.9160752506477413,
      "grad_norm": 0.9398750660029521,
      "learning_rate": 9.54358920679524e-08,
      "loss": 0.1585,
      "step": 2033
    },
    {
      "epoch": 0.9165258533288273,
      "grad_norm": 0.97514422818835,
      "learning_rate": 9.441894982498035e-08,
      "loss": 0.19,
      "step": 2034
    },
    {
      "epoch": 0.9169764560099133,
      "grad_norm": 0.901524756172699,
      "learning_rate": 9.340735049923277e-08,
      "loss": 0.1549,
      "step": 2035
    },
    {
      "epoch": 0.9174270586909992,
      "grad_norm": 0.8978882505519804,
      "learning_rate": 9.24010963375227e-08,
      "loss": 0.1613,
      "step": 2036
    },
    {
      "epoch": 0.9178776613720852,
      "grad_norm": 0.943693740183377,
      "learning_rate": 9.140018957479236e-08,
      "loss": 0.1756,
      "step": 2037
    },
    {
      "epoch": 0.9183282640531711,
      "grad_norm": 0.9409050633896883,
      "learning_rate": 9.040463243410541e-08,
      "loss": 0.1705,
      "step": 2038
    },
    {
      "epoch": 0.918778866734257,
      "grad_norm": 0.9264646847635797,
      "learning_rate": 8.941442712664561e-08,
      "loss": 0.1578,
      "step": 2039
    },
    {
      "epoch": 0.919229469415343,
      "grad_norm": 0.9371489557681785,
      "learning_rate": 8.842957585170814e-08,
      "loss": 0.1686,
      "step": 2040
    },
    {
      "epoch": 0.919680072096429,
      "grad_norm": 0.9836891585497494,
      "learning_rate": 8.745008079669742e-08,
      "loss": 0.1756,
      "step": 2041
    },
    {
      "epoch": 0.920130674777515,
      "grad_norm": 0.9805837221537438,
      "learning_rate": 8.647594413712212e-08,
      "loss": 0.1618,
      "step": 2042
    },
    {
      "epoch": 0.9205812774586009,
      "grad_norm": 0.9261986023438442,
      "learning_rate": 8.550716803658904e-08,
      "loss": 0.1686,
      "step": 2043
    },
    {
      "epoch": 0.9210318801396868,
      "grad_norm": 0.9734722095132714,
      "learning_rate": 8.454375464679865e-08,
      "loss": 0.1695,
      "step": 2044
    },
    {
      "epoch": 0.9214824828207728,
      "grad_norm": 0.9188884072317375,
      "learning_rate": 8.358570610754097e-08,
      "loss": 0.1517,
      "step": 2045
    },
    {
      "epoch": 0.9219330855018587,
      "grad_norm": 1.004333840199922,
      "learning_rate": 8.263302454669025e-08,
      "loss": 0.1833,
      "step": 2046
    },
    {
      "epoch": 0.9223836881829447,
      "grad_norm": 0.9829396979168501,
      "learning_rate": 8.168571208020032e-08,
      "loss": 0.1644,
      "step": 2047
    },
    {
      "epoch": 0.9228342908640307,
      "grad_norm": 0.9260687624875473,
      "learning_rate": 8.074377081210033e-08,
      "loss": 0.1681,
      "step": 2048
    },
    {
      "epoch": 0.9232848935451166,
      "grad_norm": 0.9609051945697021,
      "learning_rate": 7.980720283448957e-08,
      "loss": 0.1677,
      "step": 2049
    },
    {
      "epoch": 0.9237354962262025,
      "grad_norm": 1.0120669225710806,
      "learning_rate": 7.887601022753238e-08,
      "loss": 0.1834,
      "step": 2050
    },
    {
      "epoch": 0.9241860989072885,
      "grad_norm": 0.9511276526179403,
      "learning_rate": 7.795019505945495e-08,
      "loss": 0.177,
      "step": 2051
    },
    {
      "epoch": 0.9246367015883744,
      "grad_norm": 0.9227281855415796,
      "learning_rate": 7.702975938653934e-08,
      "loss": 0.1644,
      "step": 2052
    },
    {
      "epoch": 0.9250873042694604,
      "grad_norm": 1.0033217399635819,
      "learning_rate": 7.611470525312054e-08,
      "loss": 0.1726,
      "step": 2053
    },
    {
      "epoch": 0.9255379069505464,
      "grad_norm": 0.9349013994174448,
      "learning_rate": 7.520503469157947e-08,
      "loss": 0.1631,
      "step": 2054
    },
    {
      "epoch": 0.9259885096316323,
      "grad_norm": 1.0181422908202642,
      "learning_rate": 7.430074972234053e-08,
      "loss": 0.1814,
      "step": 2055
    },
    {
      "epoch": 0.9264391123127182,
      "grad_norm": 0.9334383021799829,
      "learning_rate": 7.340185235386627e-08,
      "loss": 0.176,
      "step": 2056
    },
    {
      "epoch": 0.9268897149938042,
      "grad_norm": 1.0492598395611263,
      "learning_rate": 7.250834458265355e-08,
      "loss": 0.1847,
      "step": 2057
    },
    {
      "epoch": 0.9273403176748901,
      "grad_norm": 0.9613018614069104,
      "learning_rate": 7.162022839322824e-08,
      "loss": 0.1794,
      "step": 2058
    },
    {
      "epoch": 0.9277909203559761,
      "grad_norm": 0.91255311271636,
      "learning_rate": 7.073750575814136e-08,
      "loss": 0.1607,
      "step": 2059
    },
    {
      "epoch": 0.9282415230370621,
      "grad_norm": 0.9397419477936786,
      "learning_rate": 6.986017863796435e-08,
      "loss": 0.1695,
      "step": 2060
    },
    {
      "epoch": 0.9286921257181481,
      "grad_norm": 0.9247421158046627,
      "learning_rate": 6.898824898128515e-08,
      "loss": 0.16,
      "step": 2061
    },
    {
      "epoch": 0.929142728399234,
      "grad_norm": 0.9261537295041204,
      "learning_rate": 6.81217187247038e-08,
      "loss": 0.1561,
      "step": 2062
    },
    {
      "epoch": 0.9295933310803199,
      "grad_norm": 0.9764265206005728,
      "learning_rate": 6.726058979282774e-08,
      "loss": 0.1677,
      "step": 2063
    },
    {
      "epoch": 0.9300439337614059,
      "grad_norm": 0.902753626414874,
      "learning_rate": 6.640486409826785e-08,
      "loss": 0.1512,
      "step": 2064
    },
    {
      "epoch": 0.9304945364424918,
      "grad_norm": 0.8803283338825151,
      "learning_rate": 6.555454354163437e-08,
      "loss": 0.1474,
      "step": 2065
    },
    {
      "epoch": 0.9309451391235778,
      "grad_norm": 0.9548943830388827,
      "learning_rate": 6.470963001153268e-08,
      "loss": 0.1755,
      "step": 2066
    },
    {
      "epoch": 0.9313957418046638,
      "grad_norm": 0.941676196449646,
      "learning_rate": 6.387012538455723e-08,
      "loss": 0.1706,
      "step": 2067
    },
    {
      "epoch": 0.9318463444857497,
      "grad_norm": 0.9066120881953356,
      "learning_rate": 6.303603152529119e-08,
      "loss": 0.1603,
      "step": 2068
    },
    {
      "epoch": 0.9322969471668356,
      "grad_norm": 0.8975701997942634,
      "learning_rate": 6.220735028629937e-08,
      "loss": 0.1762,
      "step": 2069
    },
    {
      "epoch": 0.9327475498479216,
      "grad_norm": 0.937967144997872,
      "learning_rate": 6.13840835081242e-08,
      "loss": 0.1662,
      "step": 2070
    },
    {
      "epoch": 0.9331981525290075,
      "grad_norm": 0.9603182403140018,
      "learning_rate": 6.056623301928327e-08,
      "loss": 0.1778,
      "step": 2071
    },
    {
      "epoch": 0.9336487552100935,
      "grad_norm": 0.9593460237620045,
      "learning_rate": 5.975380063626356e-08,
      "loss": 0.1628,
      "step": 2072
    },
    {
      "epoch": 0.9340993578911795,
      "grad_norm": 0.9396155709094206,
      "learning_rate": 5.894678816351862e-08,
      "loss": 0.1759,
      "step": 2073
    },
    {
      "epoch": 0.9345499605722654,
      "grad_norm": 0.9063173270817928,
      "learning_rate": 5.8145197393463806e-08,
      "loss": 0.1632,
      "step": 2074
    },
    {
      "epoch": 0.9350005632533513,
      "grad_norm": 0.9198235629026493,
      "learning_rate": 5.73490301064733e-08,
      "loss": 0.1589,
      "step": 2075
    },
    {
      "epoch": 0.9354511659344373,
      "grad_norm": 0.9202608244954159,
      "learning_rate": 5.6558288070874544e-08,
      "loss": 0.1597,
      "step": 2076
    },
    {
      "epoch": 0.9359017686155232,
      "grad_norm": 0.9398303538897654,
      "learning_rate": 5.577297304294543e-08,
      "loss": 0.1715,
      "step": 2077
    },
    {
      "epoch": 0.9363523712966092,
      "grad_norm": 0.91695514632765,
      "learning_rate": 5.4993086766910733e-08,
      "loss": 0.1502,
      "step": 2078
    },
    {
      "epoch": 0.9368029739776952,
      "grad_norm": 0.9207099117368464,
      "learning_rate": 5.421863097493707e-08,
      "loss": 0.1644,
      "step": 2079
    },
    {
      "epoch": 0.9372535766587812,
      "grad_norm": 0.9549078634315331,
      "learning_rate": 5.344960738713018e-08,
      "loss": 0.1555,
      "step": 2080
    },
    {
      "epoch": 0.937704179339867,
      "grad_norm": 0.9115254832418993,
      "learning_rate": 5.268601771153042e-08,
      "loss": 0.1665,
      "step": 2081
    },
    {
      "epoch": 0.938154782020953,
      "grad_norm": 0.9002114451702619,
      "learning_rate": 5.192786364410868e-08,
      "loss": 0.1578,
      "step": 2082
    },
    {
      "epoch": 0.938605384702039,
      "grad_norm": 1.0061200857124815,
      "learning_rate": 5.117514686876379e-08,
      "loss": 0.1755,
      "step": 2083
    },
    {
      "epoch": 0.9390559873831249,
      "grad_norm": 0.8938090934337861,
      "learning_rate": 5.0427869057317894e-08,
      "loss": 0.1538,
      "step": 2084
    },
    {
      "epoch": 0.9395065900642109,
      "grad_norm": 0.977029876716849,
      "learning_rate": 4.9686031869512486e-08,
      "loss": 0.1787,
      "step": 2085
    },
    {
      "epoch": 0.9399571927452969,
      "grad_norm": 0.9161657090501041,
      "learning_rate": 4.89496369530057e-08,
      "loss": 0.1528,
      "step": 2086
    },
    {
      "epoch": 0.9404077954263828,
      "grad_norm": 0.9095643914608285,
      "learning_rate": 4.8218685943368094e-08,
      "loss": 0.1727,
      "step": 2087
    },
    {
      "epoch": 0.9408583981074687,
      "grad_norm": 0.9587057997902817,
      "learning_rate": 4.7493180464078246e-08,
      "loss": 0.1644,
      "step": 2088
    },
    {
      "epoch": 0.9413090007885547,
      "grad_norm": 0.9629604422897137,
      "learning_rate": 4.677312212652108e-08,
      "loss": 0.1703,
      "step": 2089
    },
    {
      "epoch": 0.9417596034696406,
      "grad_norm": 1.0003390614928875,
      "learning_rate": 4.605851252998256e-08,
      "loss": 0.1744,
      "step": 2090
    },
    {
      "epoch": 0.9422102061507266,
      "grad_norm": 0.9286037501169877,
      "learning_rate": 4.5349353261646414e-08,
      "loss": 0.1596,
      "step": 2091
    },
    {
      "epoch": 0.9426608088318126,
      "grad_norm": 1.0118829837532903,
      "learning_rate": 4.464564589659187e-08,
      "loss": 0.1692,
      "step": 2092
    },
    {
      "epoch": 0.9431114115128985,
      "grad_norm": 0.9425679544226208,
      "learning_rate": 4.3947391997787857e-08,
      "loss": 0.1627,
      "step": 2093
    },
    {
      "epoch": 0.9435620141939844,
      "grad_norm": 0.9436041869876578,
      "learning_rate": 4.325459311609187e-08,
      "loss": 0.1603,
      "step": 2094
    },
    {
      "epoch": 0.9440126168750704,
      "grad_norm": 0.8886903189454783,
      "learning_rate": 4.256725079024554e-08,
      "loss": 0.1574,
      "step": 2095
    },
    {
      "epoch": 0.9444632195561563,
      "grad_norm": 0.9329423795551214,
      "learning_rate": 4.1885366546870754e-08,
      "loss": 0.1596,
      "step": 2096
    },
    {
      "epoch": 0.9449138222372423,
      "grad_norm": 0.9293096433954279,
      "learning_rate": 4.120894190046687e-08,
      "loss": 0.1559,
      "step": 2097
    },
    {
      "epoch": 0.9453644249183283,
      "grad_norm": 0.911463989327341,
      "learning_rate": 4.053797835340739e-08,
      "loss": 0.1593,
      "step": 2098
    },
    {
      "epoch": 0.9458150275994143,
      "grad_norm": 0.9566973300551312,
      "learning_rate": 3.987247739593636e-08,
      "loss": 0.1778,
      "step": 2099
    },
    {
      "epoch": 0.9462656302805001,
      "grad_norm": 0.8613712769085644,
      "learning_rate": 3.9212440506164465e-08,
      "loss": 0.1573,
      "step": 2100
    },
    {
      "epoch": 0.9467162329615861,
      "grad_norm": 0.9437761617537445,
      "learning_rate": 3.855786915006793e-08,
      "loss": 0.1709,
      "step": 2101
    },
    {
      "epoch": 0.9471668356426721,
      "grad_norm": 0.9124133143204082,
      "learning_rate": 3.790876478148242e-08,
      "loss": 0.1643,
      "step": 2102
    },
    {
      "epoch": 0.947617438323758,
      "grad_norm": 0.9108542684114203,
      "learning_rate": 3.726512884210165e-08,
      "loss": 0.1609,
      "step": 2103
    },
    {
      "epoch": 0.948068041004844,
      "grad_norm": 0.9720712434506643,
      "learning_rate": 3.6626962761473205e-08,
      "loss": 0.1733,
      "step": 2104
    },
    {
      "epoch": 0.94851864368593,
      "grad_norm": 1.0000646581857153,
      "learning_rate": 3.599426795699662e-08,
      "loss": 0.1889,
      "step": 2105
    },
    {
      "epoch": 0.9489692463670159,
      "grad_norm": 0.9370680977666221,
      "learning_rate": 3.53670458339192e-08,
      "loss": 0.1576,
      "step": 2106
    },
    {
      "epoch": 0.9494198490481018,
      "grad_norm": 0.9063791405664849,
      "learning_rate": 3.474529778533298e-08,
      "loss": 0.1577,
      "step": 2107
    },
    {
      "epoch": 0.9498704517291878,
      "grad_norm": 0.9593311899211125,
      "learning_rate": 3.412902519217137e-08,
      "loss": 0.1713,
      "step": 2108
    },
    {
      "epoch": 0.9503210544102737,
      "grad_norm": 1.0327217247594411,
      "learning_rate": 3.351822942320754e-08,
      "loss": 0.1905,
      "step": 2109
    },
    {
      "epoch": 0.9507716570913597,
      "grad_norm": 0.9764468754314656,
      "learning_rate": 3.2912911835049634e-08,
      "loss": 0.1704,
      "step": 2110
    },
    {
      "epoch": 0.9512222597724457,
      "grad_norm": 0.8945964138702499,
      "learning_rate": 3.231307377213833e-08,
      "loss": 0.1689,
      "step": 2111
    },
    {
      "epoch": 0.9516728624535316,
      "grad_norm": 0.8908152247664284,
      "learning_rate": 3.171871656674458e-08,
      "loss": 0.1544,
      "step": 2112
    },
    {
      "epoch": 0.9521234651346175,
      "grad_norm": 0.9630683493999175,
      "learning_rate": 3.112984153896603e-08,
      "loss": 0.1631,
      "step": 2113
    },
    {
      "epoch": 0.9525740678157035,
      "grad_norm": 0.9599669484647724,
      "learning_rate": 3.0546449996723404e-08,
      "loss": 0.1672,
      "step": 2114
    },
    {
      "epoch": 0.9530246704967894,
      "grad_norm": 0.9241733201124437,
      "learning_rate": 2.996854323575937e-08,
      "loss": 0.1605,
      "step": 2115
    },
    {
      "epoch": 0.9534752731778754,
      "grad_norm": 0.9710764623086404,
      "learning_rate": 2.939612253963331e-08,
      "loss": 0.1812,
      "step": 2116
    },
    {
      "epoch": 0.9539258758589614,
      "grad_norm": 0.908121707533848,
      "learning_rate": 2.8829189179721552e-08,
      "loss": 0.1649,
      "step": 2117
    },
    {
      "epoch": 0.9543764785400474,
      "grad_norm": 0.9083640662527406,
      "learning_rate": 2.8267744415211296e-08,
      "loss": 0.1656,
      "step": 2118
    },
    {
      "epoch": 0.9548270812211332,
      "grad_norm": 0.8951491282974102,
      "learning_rate": 2.7711789493099495e-08,
      "loss": 0.1528,
      "step": 2119
    },
    {
      "epoch": 0.9552776839022192,
      "grad_norm": 0.9383039096921281,
      "learning_rate": 2.716132564819035e-08,
      "loss": 0.1665,
      "step": 2120
    },
    {
      "epoch": 0.9557282865833052,
      "grad_norm": 0.9518623239811919,
      "learning_rate": 2.661635410309199e-08,
      "loss": 0.1708,
      "step": 2121
    },
    {
      "epoch": 0.9561788892643911,
      "grad_norm": 0.9378862678041917,
      "learning_rate": 2.6076876068213965e-08,
      "loss": 0.1741,
      "step": 2122
    },
    {
      "epoch": 0.9566294919454771,
      "grad_norm": 0.9278206810862306,
      "learning_rate": 2.554289274176419e-08,
      "loss": 0.159,
      "step": 2123
    },
    {
      "epoch": 0.9570800946265631,
      "grad_norm": 0.9577788372059376,
      "learning_rate": 2.5014405309746193e-08,
      "loss": 0.1698,
      "step": 2124
    },
    {
      "epoch": 0.957530697307649,
      "grad_norm": 1.025839380977837,
      "learning_rate": 2.449141494595797e-08,
      "loss": 0.1655,
      "step": 2125
    },
    {
      "epoch": 0.9579812999887349,
      "grad_norm": 0.9433126152166391,
      "learning_rate": 2.3973922811987295e-08,
      "loss": 0.1747,
      "step": 2126
    },
    {
      "epoch": 0.9584319026698209,
      "grad_norm": 0.903691513339492,
      "learning_rate": 2.3461930057210037e-08,
      "loss": 0.167,
      "step": 2127
    },
    {
      "epoch": 0.9588825053509068,
      "grad_norm": 1.0058632683573403,
      "learning_rate": 2.2955437818788508e-08,
      "loss": 0.1899,
      "step": 2128
    },
    {
      "epoch": 0.9593331080319928,
      "grad_norm": 0.9399743408942693,
      "learning_rate": 2.2454447221667563e-08,
      "loss": 0.1595,
      "step": 2129
    },
    {
      "epoch": 0.9597837107130788,
      "grad_norm": 0.9594553246397292,
      "learning_rate": 2.1958959378572398e-08,
      "loss": 0.177,
      "step": 2130
    },
    {
      "epoch": 0.9602343133941647,
      "grad_norm": 0.8830875917954497,
      "learning_rate": 2.1468975390006587e-08,
      "loss": 0.1569,
      "step": 2131
    },
    {
      "epoch": 0.9606849160752506,
      "grad_norm": 0.9507601854325585,
      "learning_rate": 2.0984496344249596e-08,
      "loss": 0.16,
      "step": 2132
    },
    {
      "epoch": 0.9611355187563366,
      "grad_norm": 0.9559829111178594,
      "learning_rate": 2.0505523317353727e-08,
      "loss": 0.1765,
      "step": 2133
    },
    {
      "epoch": 0.9615861214374225,
      "grad_norm": 0.9352989883382946,
      "learning_rate": 2.0032057373142453e-08,
      "loss": 0.1586,
      "step": 2134
    },
    {
      "epoch": 0.9620367241185085,
      "grad_norm": 0.9464894185394025,
      "learning_rate": 1.956409956320737e-08,
      "loss": 0.1724,
      "step": 2135
    },
    {
      "epoch": 0.9624873267995945,
      "grad_norm": 0.9775886909433427,
      "learning_rate": 1.91016509269068e-08,
      "loss": 0.1733,
      "step": 2136
    },
    {
      "epoch": 0.9629379294806804,
      "grad_norm": 0.9707150276281535,
      "learning_rate": 1.864471249136218e-08,
      "loss": 0.1702,
      "step": 2137
    },
    {
      "epoch": 0.9633885321617663,
      "grad_norm": 0.9579207813847991,
      "learning_rate": 1.819328527145725e-08,
      "loss": 0.1612,
      "step": 2138
    },
    {
      "epoch": 0.9638391348428523,
      "grad_norm": 0.9901610730307316,
      "learning_rate": 1.774737026983414e-08,
      "loss": 0.1693,
      "step": 2139
    },
    {
      "epoch": 0.9642897375239383,
      "grad_norm": 0.9252177839544279,
      "learning_rate": 1.7306968476893393e-08,
      "loss": 0.1632,
      "step": 2140
    },
    {
      "epoch": 0.9647403402050242,
      "grad_norm": 0.9930268546277804,
      "learning_rate": 1.6872080870788955e-08,
      "loss": 0.1708,
      "step": 2141
    },
    {
      "epoch": 0.9651909428861102,
      "grad_norm": 0.9284837696552151,
      "learning_rate": 1.6442708417428732e-08,
      "loss": 0.1619,
      "step": 2142
    },
    {
      "epoch": 0.9656415455671962,
      "grad_norm": 0.9169153943890256,
      "learning_rate": 1.6018852070470437e-08,
      "loss": 0.1686,
      "step": 2143
    },
    {
      "epoch": 0.966092148248282,
      "grad_norm": 0.9161979724962043,
      "learning_rate": 1.5600512771320462e-08,
      "loss": 0.1611,
      "step": 2144
    },
    {
      "epoch": 0.966542750929368,
      "grad_norm": 0.9529401144797984,
      "learning_rate": 1.518769144913168e-08,
      "loss": 0.1756,
      "step": 2145
    },
    {
      "epoch": 0.966993353610454,
      "grad_norm": 0.9619057195005012,
      "learning_rate": 1.4780389020800923e-08,
      "loss": 0.1588,
      "step": 2146
    },
    {
      "epoch": 0.9674439562915399,
      "grad_norm": 0.9314450454764756,
      "learning_rate": 1.4378606390967609e-08,
      "loss": 0.1735,
      "step": 2147
    },
    {
      "epoch": 0.9678945589726259,
      "grad_norm": 1.047548802018552,
      "learning_rate": 1.3982344452011242e-08,
      "loss": 0.1904,
      "step": 2148
    },
    {
      "epoch": 0.9683451616537119,
      "grad_norm": 0.9194561078192867,
      "learning_rate": 1.3591604084049747e-08,
      "loss": 0.1647,
      "step": 2149
    },
    {
      "epoch": 0.9687957643347977,
      "grad_norm": 0.9575900898195624,
      "learning_rate": 1.3206386154937245e-08,
      "loss": 0.1735,
      "step": 2150
    },
    {
      "epoch": 0.9692463670158837,
      "grad_norm": 0.9411881280016939,
      "learning_rate": 1.2826691520262114e-08,
      "loss": 0.1534,
      "step": 2151
    },
    {
      "epoch": 0.9696969696969697,
      "grad_norm": 0.9482488682440647,
      "learning_rate": 1.2452521023345598e-08,
      "loss": 0.1721,
      "step": 2152
    },
    {
      "epoch": 0.9701475723780556,
      "grad_norm": 0.9573010716579112,
      "learning_rate": 1.2083875495238761e-08,
      "loss": 0.1812,
      "step": 2153
    },
    {
      "epoch": 0.9705981750591416,
      "grad_norm": 0.9727335771556579,
      "learning_rate": 1.1720755754722757e-08,
      "loss": 0.1724,
      "step": 2154
    },
    {
      "epoch": 0.9710487777402276,
      "grad_norm": 0.9349543815438219,
      "learning_rate": 1.1363162608304112e-08,
      "loss": 0.1732,
      "step": 2155
    },
    {
      "epoch": 0.9714993804213135,
      "grad_norm": 0.9566156847154865,
      "learning_rate": 1.1011096850215842e-08,
      "loss": 0.1753,
      "step": 2156
    },
    {
      "epoch": 0.9719499831023994,
      "grad_norm": 0.9612547178015083,
      "learning_rate": 1.0664559262413831e-08,
      "loss": 0.1676,
      "step": 2157
    },
    {
      "epoch": 0.9724005857834854,
      "grad_norm": 0.8969285709349176,
      "learning_rate": 1.0323550614574907e-08,
      "loss": 0.1726,
      "step": 2158
    },
    {
      "epoch": 0.9728511884645714,
      "grad_norm": 0.92332833314932,
      "learning_rate": 9.988071664097376e-09,
      "loss": 0.1712,
      "step": 2159
    },
    {
      "epoch": 0.9733017911456573,
      "grad_norm": 0.9574569632858295,
      "learning_rate": 9.658123156096599e-09,
      "loss": 0.1739,
      "step": 2160
    },
    {
      "epoch": 0.9737523938267433,
      "grad_norm": 1.017729151344142,
      "learning_rate": 9.333705823404981e-09,
      "loss": 0.1643,
      "step": 2161
    },
    {
      "epoch": 0.9742029965078293,
      "grad_norm": 0.9393637507775909,
      "learning_rate": 9.014820386569756e-09,
      "loss": 0.1517,
      "step": 2162
    },
    {
      "epoch": 0.9746535991889151,
      "grad_norm": 0.961731341583023,
      "learning_rate": 8.701467553851317e-09,
      "loss": 0.1694,
      "step": 2163
    },
    {
      "epoch": 0.9751042018700011,
      "grad_norm": 0.9826839522611126,
      "learning_rate": 8.393648021222666e-09,
      "loss": 0.1748,
      "step": 2164
    },
    {
      "epoch": 0.9755548045510871,
      "grad_norm": 0.907644786835214,
      "learning_rate": 8.09136247236636e-09,
      "loss": 0.1717,
      "step": 2165
    },
    {
      "epoch": 0.976005407232173,
      "grad_norm": 1.0061970372794153,
      "learning_rate": 7.79461157867395e-09,
      "loss": 0.1799,
      "step": 2166
    },
    {
      "epoch": 0.976456009913259,
      "grad_norm": 0.9357599268796304,
      "learning_rate": 7.503395999244045e-09,
      "loss": 0.1647,
      "step": 2167
    },
    {
      "epoch": 0.976906612594345,
      "grad_norm": 0.9357608624110657,
      "learning_rate": 7.217716380881479e-09,
      "loss": 0.1711,
      "step": 2168
    },
    {
      "epoch": 0.9773572152754308,
      "grad_norm": 0.962898995983811,
      "learning_rate": 6.937573358094529e-09,
      "loss": 0.1752,
      "step": 2169
    },
    {
      "epoch": 0.9778078179565168,
      "grad_norm": 0.9687006598048918,
      "learning_rate": 6.662967553095756e-09,
      "loss": 0.1674,
      "step": 2170
    },
    {
      "epoch": 0.9782584206376028,
      "grad_norm": 0.9104077045031495,
      "learning_rate": 6.3938995757981125e-09,
      "loss": 0.1609,
      "step": 2171
    },
    {
      "epoch": 0.9787090233186887,
      "grad_norm": 0.9437695260385421,
      "learning_rate": 6.1303700238152245e-09,
      "loss": 0.1705,
      "step": 2172
    },
    {
      "epoch": 0.9791596259997747,
      "grad_norm": 1.0098665718813014,
      "learning_rate": 5.8723794824597226e-09,
      "loss": 0.169,
      "step": 2173
    },
    {
      "epoch": 0.9796102286808607,
      "grad_norm": 0.9436186302476126,
      "learning_rate": 5.6199285247415805e-09,
      "loss": 0.1725,
      "step": 2174
    },
    {
      "epoch": 0.9800608313619465,
      "grad_norm": 0.9388694365891636,
      "learning_rate": 5.373017711367001e-09,
      "loss": 0.1674,
      "step": 2175
    },
    {
      "epoch": 0.9805114340430325,
      "grad_norm": 0.937859556500245,
      "learning_rate": 5.131647590737587e-09,
      "loss": 0.1694,
      "step": 2176
    },
    {
      "epoch": 0.9809620367241185,
      "grad_norm": 1.0038011437378807,
      "learning_rate": 4.895818698948396e-09,
      "loss": 0.1753,
      "step": 2177
    },
    {
      "epoch": 0.9814126394052045,
      "grad_norm": 0.934190879507068,
      "learning_rate": 4.6655315597876615e-09,
      "loss": 0.1548,
      "step": 2178
    },
    {
      "epoch": 0.9818632420862904,
      "grad_norm": 0.9931696688123586,
      "learning_rate": 4.440786684734577e-09,
      "loss": 0.1596,
      "step": 2179
    },
    {
      "epoch": 0.9823138447673764,
      "grad_norm": 0.9113361170344727,
      "learning_rate": 4.221584572958737e-09,
      "loss": 0.1569,
      "step": 2180
    },
    {
      "epoch": 0.9827644474484624,
      "grad_norm": 0.9184095650294223,
      "learning_rate": 4.0079257113190275e-09,
      "loss": 0.1563,
      "step": 2181
    },
    {
      "epoch": 0.9832150501295482,
      "grad_norm": 0.9809381433522103,
      "learning_rate": 3.799810574363072e-09,
      "loss": 0.18,
      "step": 2182
    },
    {
      "epoch": 0.9836656528106342,
      "grad_norm": 0.9200844447422045,
      "learning_rate": 3.597239624325011e-09,
      "loss": 0.1636,
      "step": 2183
    },
    {
      "epoch": 0.9841162554917202,
      "grad_norm": 0.9477262368387944,
      "learning_rate": 3.4002133111246673e-09,
      "loss": 0.163,
      "step": 2184
    },
    {
      "epoch": 0.9845668581728061,
      "grad_norm": 0.9829302587131702,
      "learning_rate": 3.208732072368104e-09,
      "loss": 0.1828,
      "step": 2185
    },
    {
      "epoch": 0.9850174608538921,
      "grad_norm": 0.9323177390240451,
      "learning_rate": 3.022796333344291e-09,
      "loss": 0.1686,
      "step": 2186
    },
    {
      "epoch": 0.9854680635349781,
      "grad_norm": 0.97959980859569,
      "learning_rate": 2.8424065070262186e-09,
      "loss": 0.1721,
      "step": 2187
    },
    {
      "epoch": 0.9859186662160639,
      "grad_norm": 0.9360035720240707,
      "learning_rate": 2.6675629940689508e-09,
      "loss": 0.1661,
      "step": 2188
    },
    {
      "epoch": 0.9863692688971499,
      "grad_norm": 0.9829560482882546,
      "learning_rate": 2.4982661828085175e-09,
      "loss": 0.1745,
      "step": 2189
    },
    {
      "epoch": 0.9868198715782359,
      "grad_norm": 0.8791817829948829,
      "learning_rate": 2.3345164492616367e-09,
      "loss": 0.1481,
      "step": 2190
    },
    {
      "epoch": 0.9872704742593218,
      "grad_norm": 0.9195574805142371,
      "learning_rate": 2.1763141571248813e-09,
      "loss": 0.173,
      "step": 2191
    },
    {
      "epoch": 0.9877210769404078,
      "grad_norm": 0.9689298058051112,
      "learning_rate": 2.0236596577738466e-09,
      "loss": 0.1746,
      "step": 2192
    },
    {
      "epoch": 0.9881716796214938,
      "grad_norm": 0.9370644186173055,
      "learning_rate": 1.876553290261207e-09,
      "loss": 0.1744,
      "step": 2193
    },
    {
      "epoch": 0.9886222823025796,
      "grad_norm": 0.905510873781387,
      "learning_rate": 1.7349953813183828e-09,
      "loss": 0.1543,
      "step": 2194
    },
    {
      "epoch": 0.9890728849836656,
      "grad_norm": 0.9101236065698919,
      "learning_rate": 1.5989862453522075e-09,
      "loss": 0.1578,
      "step": 2195
    },
    {
      "epoch": 0.9895234876647516,
      "grad_norm": 0.8847140026088114,
      "learning_rate": 1.468526184445762e-09,
      "loss": 0.1527,
      "step": 2196
    },
    {
      "epoch": 0.9899740903458376,
      "grad_norm": 0.9682107112247759,
      "learning_rate": 1.343615488357819e-09,
      "loss": 0.1662,
      "step": 2197
    },
    {
      "epoch": 0.9904246930269235,
      "grad_norm": 0.9471219244863741,
      "learning_rate": 1.2242544345211772e-09,
      "loss": 0.1595,
      "step": 2198
    },
    {
      "epoch": 0.9908752957080095,
      "grad_norm": 0.9464719657257396,
      "learning_rate": 1.1104432880429394e-09,
      "loss": 0.1605,
      "step": 2199
    },
    {
      "epoch": 0.9913258983890955,
      "grad_norm": 0.952522471678984,
      "learning_rate": 1.0021823017028475e-09,
      "loss": 0.1706,
      "step": 2200
    },
    {
      "epoch": 0.9917765010701813,
      "grad_norm": 0.9166081061347666,
      "learning_rate": 8.994717159546695e-10,
      "loss": 0.1585,
      "step": 2201
    },
    {
      "epoch": 0.9922271037512673,
      "grad_norm": 0.894856070306113,
      "learning_rate": 8.023117589237017e-10,
      "loss": 0.1594,
      "step": 2202
    },
    {
      "epoch": 0.9926777064323533,
      "grad_norm": 0.8818944106764158,
      "learning_rate": 7.10702646406769e-10,
      "loss": 0.156,
      "step": 2203
    },
    {
      "epoch": 0.9931283091134392,
      "grad_norm": 0.9621081747527447,
      "learning_rate": 6.246445818727798e-10,
      "loss": 0.1742,
      "step": 2204
    },
    {
      "epoch": 0.9935789117945252,
      "grad_norm": 0.9263306240100847,
      "learning_rate": 5.44137756460783e-10,
      "loss": 0.1645,
      "step": 2205
    },
    {
      "epoch": 0.9940295144756112,
      "grad_norm": 1.0081667902258236,
      "learning_rate": 4.691823489805236e-10,
      "loss": 0.1873,
      "step": 2206
    },
    {
      "epoch": 0.994480117156697,
      "grad_norm": 0.8830293356060712,
      "learning_rate": 3.9977852591188694e-10,
      "loss": 0.1563,
      "step": 2207
    },
    {
      "epoch": 0.994930719837783,
      "grad_norm": 0.9731367845524682,
      "learning_rate": 3.3592644140434393e-10,
      "loss": 0.1622,
      "step": 2208
    },
    {
      "epoch": 0.995381322518869,
      "grad_norm": 0.9199241966535081,
      "learning_rate": 2.776262372761185e-10,
      "loss": 0.1696,
      "step": 2209
    },
    {
      "epoch": 0.9958319251999549,
      "grad_norm": 0.960822046253473,
      "learning_rate": 2.2487804301557503e-10,
      "loss": 0.1634,
      "step": 2210
    },
    {
      "epoch": 0.9962825278810409,
      "grad_norm": 0.9356621441742621,
      "learning_rate": 1.776819757787207e-10,
      "loss": 0.164,
      "step": 2211
    },
    {
      "epoch": 0.9967331305621269,
      "grad_norm": 1.0160315804220252,
      "learning_rate": 1.3603814039031547e-10,
      "loss": 0.1824,
      "step": 2212
    },
    {
      "epoch": 0.9971837332432127,
      "grad_norm": 0.9952641254925227,
      "learning_rate": 9.994662934387223e-11,
      "loss": 0.1648,
      "step": 2213
    },
    {
      "epoch": 0.9976343359242987,
      "grad_norm": 0.9693799712923902,
      "learning_rate": 6.94075227999913e-11,
      "loss": 0.1677,
      "step": 2214
    },
    {
      "epoch": 0.9980849386053847,
      "grad_norm": 1.0026671968102894,
      "learning_rate": 4.44208885877484e-11,
      "loss": 0.1815,
      "step": 2215
    },
    {
      "epoch": 0.9985355412864707,
      "grad_norm": 0.9105412050504266,
      "learning_rate": 2.498678220386186e-11,
      "loss": 0.1594,
      "step": 2216
    },
    {
      "epoch": 0.9989861439675566,
      "grad_norm": 0.9882824663570642,
      "learning_rate": 1.1105246812137538e-11,
      "loss": 0.1816,
      "step": 2217
    },
    {
      "epoch": 0.9994367466486426,
      "grad_norm": 0.9497467838011637,
      "learning_rate": 2.7763132445790543e-12,
      "loss": 0.1736,
      "step": 2218
    },
    {
      "epoch": 0.9998873493297286,
      "grad_norm": 1.0189201478595424,
      "learning_rate": 0.0,
      "loss": 0.1765,
      "step": 2219
    },
    {
      "epoch": 0.9998873493297286,
      "step": 2219,
      "total_flos": 369431437639680.0,
      "train_loss": 0.19088607980840752,
      "train_runtime": 22854.2492,
      "train_samples_per_second": 6.214,
      "train_steps_per_second": 0.097
    }
  ],
  "logging_steps": 1,
  "max_steps": 2219,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 369431437639680.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}