diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,7012 +2,3162 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.018306636155606407,
+  "epoch": 0.008237986270022883,
   "eval_steps": 50,
-  "global_step": 1000,
+  "global_step": 450,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 1.8306636155606407e-05,
-      "grad_norm": 6.385849475860596,
+      "grad_norm": 6.099522590637207,
       "learning_rate": 0.0,
-      "loss": 0.6987,
+      "loss": 0.6392,
       "step": 1
     },
     {
       "epoch": 3.6613272311212814e-05,
-      "grad_norm": 12.189742088317871,
+      "grad_norm": 8.014727592468262,
       "learning_rate": 2.0000000000000002e-07,
-      "loss": 0.9891,
+      "loss": 0.8375,
       "step": 2
     },
     {
       "epoch": 5.491990846681922e-05,
-      "grad_norm": 7.4394330978393555,
+      "grad_norm": 10.58983039855957,
       "learning_rate": 4.0000000000000003e-07,
-      "loss": 0.6886,
+      "loss": 0.5021,
       "step": 3
     },
     {
       "epoch": 7.322654462242563e-05,
-      "grad_norm": 19.46971321105957,
+      "grad_norm": 6.642515659332275,
       "learning_rate": 6.000000000000001e-07,
-      "loss": 1.0006,
+      "loss": 0.7624,
       "step": 4
     },
     {
       "epoch": 9.153318077803204e-05,
-      "grad_norm": 10.222038269042969,
+      "grad_norm": 8.307601928710938,
       "learning_rate": 8.000000000000001e-07,
-      "loss": 1.0917,
+      "loss": 1.005,
       "step": 5
     },
     {
       "epoch": 0.00010983981693363844,
-      "grad_norm": 15.960102081298828,
+      "grad_norm": 11.282310485839844,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 1.5058,
+      "loss": 0.939,
       "step": 6
     },
     {
       "epoch": 0.00012814645308924485,
-      "grad_norm": 9.664617538452148,
+      "grad_norm": 9.1802339553833,
       "learning_rate": 1.2000000000000002e-06,
-      "loss": 1.0587,
+      "loss": 1.0419,
       "step": 7
     },
     {
       "epoch": 0.00014645308924485126,
-      "grad_norm": 12.8255033493042,
+      "grad_norm": 2.651777744293213,
       "learning_rate": 1.4000000000000001e-06,
-      "loss": 1.0243,
+      "loss": 0.4708,
       "step": 8
     },
     {
       "epoch": 0.00016475972540045766,
-      "grad_norm": 9.836108207702637,
+      "grad_norm": 7.600271701812744,
       "learning_rate": 1.6000000000000001e-06,
-      "loss": 0.7258,
+      "loss": 0.6082,
       "step": 9
     },
     {
       "epoch": 0.00018306636155606407,
-      "grad_norm": 5.548786163330078,
+      "grad_norm": 6.370179653167725,
       "learning_rate": 1.8000000000000001e-06,
-      "loss": 0.6062,
+      "loss": 0.8817,
       "step": 10
     },
     {
       "epoch": 0.00020137299771167048,
-      "grad_norm": 8.437209129333496,
+      "grad_norm": 7.1832499504089355,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.7671,
+      "loss": 0.5443,
       "step": 11
     },
     {
       "epoch": 0.00021967963386727689,
-      "grad_norm": 9.735587120056152,
+      "grad_norm": 6.7331223487854,
       "learning_rate": 2.2e-06,
-      "loss": 1.0208,
+      "loss": 0.5602,
       "step": 12
     },
     {
       "epoch": 0.0002379862700228833,
-      "grad_norm": 11.217083930969238,
+      "grad_norm": 8.009416580200195,
       "learning_rate": 2.4000000000000003e-06,
-      "loss": 1.1293,
+      "loss": 0.741,
       "step": 13
     },
     {
       "epoch": 0.0002562929061784897,
-      "grad_norm": 16.09665298461914,
+      "grad_norm": 9.607633590698242,
       "learning_rate": 2.6e-06,
-      "loss": 1.4595,
+      "loss": 0.7022,
       "step": 14
     },
     {
       "epoch": 0.00027459954233409613,
-      "grad_norm": 9.518461227416992,
+      "grad_norm": 10.585896492004395,
       "learning_rate": 2.8000000000000003e-06,
-      "loss": 0.7914,
+      "loss": 1.1929,
       "step": 15
     },
     {
       "epoch": 0.0002929061784897025,
-      "grad_norm": 14.459754943847656,
+      "grad_norm": 3.4262940883636475,
       "learning_rate": 3e-06,
-      "loss": 1.0895,
+      "loss": 0.5217,
       "step": 16
     },
     {
       "epoch": 0.00031121281464530895,
-      "grad_norm": 8.780324935913086,
+      "grad_norm": 11.94228458404541,
       "learning_rate": 3.2000000000000003e-06,
-      "loss": 0.7839,
+      "loss": 1.0208,
       "step": 17
     },
     {
       "epoch": 0.00032951945080091533,
-      "grad_norm": 13.737929344177246,
+      "grad_norm": 8.004399299621582,
       "learning_rate": 3.4000000000000005e-06,
-      "loss": 1.2222,
+      "loss": 0.5694,
       "step": 18
     },
     {
       "epoch": 0.00034782608695652176,
-      "grad_norm": 5.206906318664551,
+      "grad_norm": 7.853118896484375,
       "learning_rate": 3.6000000000000003e-06,
-      "loss": 0.6575,
+      "loss": 0.5695,
       "step": 19
     },
     {
       "epoch": 0.00036613272311212814,
-      "grad_norm": 13.852402687072754,
+      "grad_norm": 6.011293411254883,
       "learning_rate": 3.8000000000000005e-06,
-      "loss": 1.2024,
+      "loss": 0.4659,
       "step": 20
     },
     {
       "epoch": 0.0003844393592677346,
-      "grad_norm": 15.006126403808594,
+      "grad_norm": 10.169012069702148,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 1.3288,
+      "loss": 0.8618,
       "step": 21
     },
     {
       "epoch": 0.00040274599542334096,
-      "grad_norm": 17.97928810119629,
+      "grad_norm": 3.0373430252075195,
       "learning_rate": 4.2000000000000004e-06,
-      "loss": 1.5079,
+      "loss": 0.2469,
       "step": 22
     },
     {
       "epoch": 0.0004210526315789474,
-      "grad_norm": 16.182106018066406,
+      "grad_norm": 7.847787380218506,
       "learning_rate": 4.4e-06,
-      "loss": 1.1849,
+      "loss": 0.738,
       "step": 23
     },
     {
       "epoch": 0.00043935926773455377,
-      "grad_norm": 18.808963775634766,
+      "grad_norm": 6.6885809898376465,
       "learning_rate": 4.600000000000001e-06,
-      "loss": 1.5441,
+      "loss": 0.6311,
       "step": 24
     },
     {
       "epoch": 0.0004576659038901602,
-      "grad_norm": 10.004517555236816,
+      "grad_norm": 7.829857349395752,
       "learning_rate": 4.800000000000001e-06,
-      "loss": 0.916,
+      "loss": 0.5967,
       "step": 25
     },
     {
       "epoch": 0.0004759725400457666,
-      "grad_norm": 2.7199032306671143,
+      "grad_norm": 15.481698989868164,
       "learning_rate": 5e-06,
-      "loss": 0.4172,
+      "loss": 0.9285,
       "step": 26
     },
     {
       "epoch": 0.000494279176201373,
-      "grad_norm": 10.674932479858398,
+      "grad_norm": 23.23982810974121,
       "learning_rate": 5.2e-06,
-      "loss": 0.7953,
+      "loss": 0.6175,
       "step": 27
     },
     {
       "epoch": 0.0005125858123569794,
-      "grad_norm": 11.257180213928223,
+      "grad_norm": 9.985934257507324,
       "learning_rate": 5.400000000000001e-06,
-      "loss": 0.5156,
+      "loss": 0.4765,
       "step": 28
     },
     {
       "epoch": 0.0005308924485125858,
-      "grad_norm": 5.36810302734375,
+      "grad_norm": 17.83190155029297,
       "learning_rate": 5.600000000000001e-06,
-      "loss": 0.5779,
+      "loss": 1.1398,
       "step": 29
     },
     {
       "epoch": 0.0005491990846681923,
-      "grad_norm": 7.723548412322998,
+      "grad_norm": 7.259341239929199,
       "learning_rate": 5.8e-06,
-      "loss": 0.7211,
+      "loss": 0.3864,
       "step": 30
     },
     {
       "epoch": 0.0005675057208237986,
-      "grad_norm": 10.500932693481445,
+      "grad_norm": 8.247320175170898,
       "learning_rate": 6e-06,
-      "loss": 0.8241,
+      "loss": 0.5853,
       "step": 31
     },
     {
       "epoch": 0.000585812356979405,
-      "grad_norm": 4.589667320251465,
+      "grad_norm": 8.801642417907715,
       "learning_rate": 6.200000000000001e-06,
-      "loss": 0.6312,
+      "loss": 1.197,
       "step": 32
     },
     {
       "epoch": 0.0006041189931350115,
-      "grad_norm": 5.961172580718994,
+      "grad_norm": 8.88306713104248,
       "learning_rate": 6.4000000000000006e-06,
-      "loss": 0.4912,
+      "loss": 0.3122,
       "step": 33
     },
     {
       "epoch": 0.0006224256292906179,
-      "grad_norm": 5.868521213531494,
+      "grad_norm": 10.055649757385254,
       "learning_rate": 6.600000000000001e-06,
-      "loss": 0.6721,
+      "loss": 0.5566,
       "step": 34
     },
     {
       "epoch": 0.0006407322654462242,
-      "grad_norm": 10.867189407348633,
+      "grad_norm": 12.692378044128418,
       "learning_rate": 6.800000000000001e-06,
-      "loss": 0.9532,
+      "loss": 1.0306,
       "step": 35
     },
     {
       "epoch": 0.0006590389016018307,
-      "grad_norm": 10.853588104248047,
+      "grad_norm": 6.696789741516113,
       "learning_rate": 7e-06,
-      "loss": 1.0711,
+      "loss": 0.3655,
       "step": 36
     },
     {
       "epoch": 0.0006773455377574371,
-      "grad_norm": 7.248025417327881,
+      "grad_norm": 5.720799922943115,
       "learning_rate": 7.2000000000000005e-06,
-      "loss": 0.6016,
+      "loss": 0.4338,
       "step": 37
     },
     {
       "epoch": 0.0006956521739130435,
-      "grad_norm": 2.20741868019104,
+      "grad_norm": 6.0295610427856445,
       "learning_rate": 7.4e-06,
-      "loss": 0.243,
+      "loss": 0.5826,
       "step": 38
     },
     {
       "epoch": 0.0007139588100686499,
-      "grad_norm": 3.712090253829956,
+      "grad_norm": 3.975026845932007,
       "learning_rate": 7.600000000000001e-06,
-      "loss": 0.6429,
+      "loss": 0.5669,
       "step": 39
     },
     {
       "epoch": 0.0007322654462242563,
-      "grad_norm": 4.5562825202941895,
+      "grad_norm": 8.472329139709473,
       "learning_rate": 7.800000000000002e-06,
-      "loss": 0.3965,
+      "loss": 0.8793,
       "step": 40
     },
     {
       "epoch": 0.0007505720823798627,
-      "grad_norm": 5.0433573722839355,
+      "grad_norm": 4.2498321533203125,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.4462,
+      "loss": 0.3409,
       "step": 41
     },
     {
       "epoch": 0.0007688787185354692,
-      "grad_norm": 5.870055675506592,
+      "grad_norm": 8.282734870910645,
       "learning_rate": 8.2e-06,
-      "loss": 0.5616,
+      "loss": 0.549,
       "step": 42
     },
     {
       "epoch": 0.0007871853546910755,
-      "grad_norm": 4.7254109382629395,
+      "grad_norm": 7.966742992401123,
       "learning_rate": 8.400000000000001e-06,
-      "loss": 0.5478,
+      "loss": 0.7406,
       "step": 43
     },
     {
       "epoch": 0.0008054919908466819,
-      "grad_norm": 13.798440933227539,
+      "grad_norm": 17.36241340637207,
       "learning_rate": 8.6e-06,
-      "loss": 0.5848,
+      "loss": 1.0215,
       "step": 44
     },
     {
       "epoch": 0.0008237986270022883,
-      "grad_norm": 4.205646514892578,
+      "grad_norm": 8.946808815002441,
       "learning_rate": 8.8e-06,
-      "loss": 0.4753,
+      "loss": 0.5102,
       "step": 45
     },
     {
       "epoch": 0.0008421052631578948,
-      "grad_norm": 8.086190223693848,
+      "grad_norm": 9.421518325805664,
       "learning_rate": 9e-06,
-      "loss": 0.4394,
+      "loss": 1.1237,
       "step": 46
     },
     {
       "epoch": 0.0008604118993135011,
-      "grad_norm": 3.1762030124664307,
+      "grad_norm": 11.310033798217773,
       "learning_rate": 9.200000000000002e-06,
-      "loss": 0.2779,
+      "loss": 0.7825,
       "step": 47
     },
     {
       "epoch": 0.0008787185354691075,
-      "grad_norm": 9.121731758117676,
+      "grad_norm": 11.267560005187988,
       "learning_rate": 9.4e-06,
-      "loss": 0.6942,
+      "loss": 1.0458,
       "step": 48
     },
     {
       "epoch": 0.000897025171624714,
-      "grad_norm": 4.997460842132568,
+      "grad_norm": 2.9605565071105957,
       "learning_rate": 9.600000000000001e-06,
-      "loss": 0.4247,
+      "loss": 0.2229,
       "step": 49
     },
     {
       "epoch": 0.0009153318077803204,
-      "grad_norm": 7.679621696472168,
+      "grad_norm": 5.507591724395752,
       "learning_rate": 9.800000000000001e-06,
-      "loss": 0.7129,
+      "loss": 0.3601,
       "step": 50
     },
     {
       "epoch": 0.0009336384439359267,
-      "grad_norm": 9.885171890258789,
+      "grad_norm": 7.83518123626709,
       "learning_rate": 1e-05,
-      "loss": 1.0787,
+      "loss": 0.5938,
       "step": 51
     },
     {
       "epoch": 0.0009519450800915332,
-      "grad_norm": 2.634310245513916,
+      "grad_norm": 6.110620975494385,
       "learning_rate": 1.02e-05,
-      "loss": 0.5534,
+      "loss": 0.8618,
       "step": 52
     },
     {
       "epoch": 0.0009702517162471396,
-      "grad_norm": 5.5200653076171875,
+      "grad_norm": 7.460414409637451,
       "learning_rate": 1.04e-05,
-      "loss": 0.4369,
+      "loss": 0.8692,
       "step": 53
     },
     {
       "epoch": 0.000988558352402746,
-      "grad_norm": 3.140720844268799,
+      "grad_norm": 7.539251804351807,
       "learning_rate": 1.0600000000000002e-05,
-      "loss": 0.5352,
+      "loss": 0.7236,
       "step": 54
     },
     {
       "epoch": 0.0010068649885583525,
-      "grad_norm": 2.700253963470459,
+      "grad_norm": 5.357115268707275,
       "learning_rate": 1.0800000000000002e-05,
-      "loss": 0.2498,
+      "loss": 0.3101,
       "step": 55
     },
     {
       "epoch": 0.0010251716247139588,
-      "grad_norm": 4.904640197753906,
+      "grad_norm": 3.6166787147521973,
       "learning_rate": 1.1000000000000001e-05,
-      "loss": 0.3029,
+      "loss": 0.4426,
       "step": 56
     },
     {
       "epoch": 0.0010434782608695651,
-      "grad_norm": 7.171940326690674,
+      "grad_norm": 5.8816680908203125,
       "learning_rate": 1.1200000000000001e-05,
-      "loss": 0.3727,
+      "loss": 0.4829,
       "step": 57
     },
     {
       "epoch": 0.0010617848970251717,
-      "grad_norm": 8.557679176330566,
+      "grad_norm": 8.535571098327637,
       "learning_rate": 1.14e-05,
-      "loss": 0.5864,
+      "loss": 0.9369,
       "step": 58
     },
     {
       "epoch": 0.001080091533180778,
-      "grad_norm": 2.2290961742401123,
+      "grad_norm": 5.3663482666015625,
       "learning_rate": 1.16e-05,
-      "loss": 0.2341,
+      "loss": 0.3664,
       "step": 59
     },
     {
       "epoch": 0.0010983981693363845,
-      "grad_norm": 2.422307252883911,
+      "grad_norm": 10.66060733795166,
       "learning_rate": 1.18e-05,
-      "loss": 0.2012,
+      "loss": 1.2902,
       "step": 60
     },
     {
       "epoch": 0.0011167048054919909,
-      "grad_norm": 3.0570621490478516,
+      "grad_norm": 2.3751587867736816,
       "learning_rate": 1.2e-05,
-      "loss": 0.447,
+      "loss": 0.2074,
       "step": 61
     },
     {
       "epoch": 0.0011350114416475972,
-      "grad_norm": 8.050760269165039,
+      "grad_norm": 6.494527816772461,
       "learning_rate": 1.22e-05,
-      "loss": 0.4355,
+      "loss": 0.5742,
       "step": 62
     },
     {
       "epoch": 0.0011533180778032037,
-      "grad_norm": 4.595751762390137,
+      "grad_norm": 1.9130464792251587,
       "learning_rate": 1.2400000000000002e-05,
-      "loss": 0.2931,
+      "loss": 0.1833,
       "step": 63
     },
     {
       "epoch": 0.00117162471395881,
-      "grad_norm": 3.4909543991088867,
+      "grad_norm": 7.697911262512207,
       "learning_rate": 1.2600000000000001e-05,
-      "loss": 0.2413,
+      "loss": 1.0019,
       "step": 64
     },
     {
       "epoch": 0.0011899313501144164,
-      "grad_norm": 15.33580493927002,
+      "grad_norm": 16.3773250579834,
       "learning_rate": 1.2800000000000001e-05,
-      "loss": 0.6939,
+      "loss": 1.5617,
       "step": 65
     },
     {
       "epoch": 0.001208237986270023,
-      "grad_norm": 1.1174702644348145,
+      "grad_norm": 6.1076836585998535,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 0.1509,
+      "loss": 0.5233,
       "step": 66
     },
     {
       "epoch": 0.0012265446224256293,
-      "grad_norm": 4.3092041015625,
+      "grad_norm": 1.6926074028015137,
       "learning_rate": 1.3200000000000002e-05,
-      "loss": 0.4519,
+      "loss": 0.2422,
       "step": 67
     },
     {
       "epoch": 0.0012448512585812358,
-      "grad_norm": 1.2255864143371582,
+      "grad_norm": 3.413689374923706,
       "learning_rate": 1.3400000000000002e-05,
-      "loss": 0.1941,
+      "loss": 0.2152,
       "step": 68
     },
     {
       "epoch": 0.0012631578947368421,
-      "grad_norm": 2.7325243949890137,
+      "grad_norm": 9.58433723449707,
       "learning_rate": 1.3600000000000002e-05,
-      "loss": 0.2606,
+      "loss": 1.0936,
       "step": 69
     },
     {
       "epoch": 0.0012814645308924484,
-      "grad_norm": 2.7921698093414307,
+      "grad_norm": 7.524669647216797,
       "learning_rate": 1.38e-05,
-      "loss": 0.395,
+      "loss": 1.1815,
       "step": 70
     },
     {
       "epoch": 0.001299771167048055,
-      "grad_norm": 2.5454518795013428,
+      "grad_norm": 6.692052841186523,
       "learning_rate": 1.4e-05,
-      "loss": 0.1967,
+      "loss": 0.5309,
       "step": 71
     },
     {
       "epoch": 0.0013180778032036613,
-      "grad_norm": 1.991735816001892,
+      "grad_norm": 14.243474006652832,
       "learning_rate": 1.4200000000000001e-05,
-      "loss": 0.196,
+      "loss": 0.8553,
       "step": 72
     },
     {
       "epoch": 0.0013363844393592676,
-      "grad_norm": 4.355623245239258,
+      "grad_norm": 10.684292793273926,
       "learning_rate": 1.4400000000000001e-05,
-      "loss": 0.3516,
+      "loss": 0.9399,
       "step": 73
     },
     {
       "epoch": 0.0013546910755148742,
-      "grad_norm": 1.876690149307251,
+      "grad_norm": 5.949723243713379,
       "learning_rate": 1.46e-05,
-      "loss": 0.1932,
+      "loss": 0.7573,
       "step": 74
     },
     {
       "epoch": 0.0013729977116704805,
-      "grad_norm": 0.9775153398513794,
+      "grad_norm": 6.119026184082031,
       "learning_rate": 1.48e-05,
-      "loss": 0.1331,
+      "loss": 0.5557,
       "step": 75
     },
     {
       "epoch": 0.001391304347826087,
-      "grad_norm": 1.4858100414276123,
+      "grad_norm": 6.306831359863281,
       "learning_rate": 1.5000000000000002e-05,
-      "loss": 0.1303,
+      "loss": 0.6165,
       "step": 76
     },
     {
       "epoch": 0.0014096109839816934,
-      "grad_norm": 3.9447224140167236,
+      "grad_norm": 5.815463066101074,
       "learning_rate": 1.5200000000000002e-05,
-      "loss": 0.25,
+      "loss": 0.6491,
       "step": 77
     },
     {
       "epoch": 0.0014279176201372997,
-      "grad_norm": 7.695821762084961,
+      "grad_norm": 4.443011283874512,
       "learning_rate": 1.54e-05,
-      "loss": 0.489,
+      "loss": 0.4984,
       "step": 78
     },
     {
       "epoch": 0.0014462242562929062,
-      "grad_norm": 1.9784696102142334,
+      "grad_norm": 4.615481376647949,
       "learning_rate": 1.5600000000000003e-05,
-      "loss": 0.1807,
+      "loss": 0.3598,
       "step": 79
     },
     {
       "epoch": 0.0014645308924485126,
-      "grad_norm": 2.0325303077697754,
+      "grad_norm": 6.703000545501709,
       "learning_rate": 1.58e-05,
-      "loss": 0.138,
+      "loss": 0.926,
       "step": 80
     },
     {
       "epoch": 0.001482837528604119,
-      "grad_norm": 7.010915756225586,
+      "grad_norm": 3.9417295455932617,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.4858,
+      "loss": 0.4523,
       "step": 81
     },
     {
       "epoch": 0.0015011441647597254,
-      "grad_norm": 3.553511142730713,
+      "grad_norm": 5.64249849319458,
       "learning_rate": 1.62e-05,
-      "loss": 0.2599,
+      "loss": 0.4319,
       "step": 82
     },
     {
       "epoch": 0.0015194508009153318,
-      "grad_norm": 11.670984268188477,
+      "grad_norm": 4.152187347412109,
       "learning_rate": 1.64e-05,
-      "loss": 0.4707,
+      "loss": 0.3675,
       "step": 83
     },
     {
       "epoch": 0.0015377574370709383,
-      "grad_norm": 0.805892825126648,
+      "grad_norm": 5.285708427429199,
       "learning_rate": 1.66e-05,
-      "loss": 0.0971,
+      "loss": 0.383,
       "step": 84
     },
     {
       "epoch": 0.0015560640732265446,
-      "grad_norm": 5.276065826416016,
+      "grad_norm": 9.757657051086426,
       "learning_rate": 1.6800000000000002e-05,
-      "loss": 0.2417,
+      "loss": 0.8236,
       "step": 85
     },
     {
       "epoch": 0.001574370709382151,
-      "grad_norm": 1.6359636783599854,
+      "grad_norm": 3.1572093963623047,
       "learning_rate": 1.7e-05,
-      "loss": 0.1095,
+      "loss": 0.556,
       "step": 86
     },
     {
       "epoch": 0.0015926773455377575,
-      "grad_norm": 7.448240756988525,
+      "grad_norm": 4.673479080200195,
       "learning_rate": 1.72e-05,
-      "loss": 0.2632,
+      "loss": 0.3168,
       "step": 87
     },
     {
       "epoch": 0.0016109839816933638,
-      "grad_norm": 1.0354855060577393,
+      "grad_norm": 4.761971950531006,
       "learning_rate": 1.7400000000000003e-05,
-      "loss": 0.1002,
+      "loss": 0.7314,
       "step": 88
     },
     {
       "epoch": 0.0016292906178489702,
-      "grad_norm": 11.703107833862305,
+      "grad_norm": 4.868711471557617,
       "learning_rate": 1.76e-05,
-      "loss": 0.5361,
+      "loss": 0.5399,
       "step": 89
     },
     {
       "epoch": 0.0016475972540045767,
-      "grad_norm": 2.071930408477783,
+      "grad_norm": 11.782761573791504,
       "learning_rate": 1.7800000000000002e-05,
-      "loss": 0.1399,
+      "loss": 0.7641,
       "step": 90
     },
     {
       "epoch": 0.001665903890160183,
-      "grad_norm": 6.908803462982178,
+      "grad_norm": 7.96957540512085,
       "learning_rate": 1.8e-05,
-      "loss": 0.2534,
+      "loss": 0.7079,
       "step": 91
     },
     {
       "epoch": 0.0016842105263157896,
-      "grad_norm": 5.774661540985107,
+      "grad_norm": 2.9694902896881104,
       "learning_rate": 1.8200000000000002e-05,
-      "loss": 0.3042,
+      "loss": 0.2075,
       "step": 92
     },
     {
       "epoch": 0.001702517162471396,
-      "grad_norm": 1.1571534872055054,
+      "grad_norm": 8.391504287719727,
       "learning_rate": 1.8400000000000003e-05,
-      "loss": 0.1797,
+      "loss": 0.7879,
       "step": 93
     },
     {
       "epoch": 0.0017208237986270022,
-      "grad_norm": 1.0660196542739868,
+      "grad_norm": 5.025172710418701,
       "learning_rate": 1.86e-05,
-      "loss": 0.1148,
+      "loss": 0.7385,
       "step": 94
     },
     {
       "epoch": 0.0017391304347826088,
-      "grad_norm": 3.5088953971862793,
+      "grad_norm": 9.963128089904785,
       "learning_rate": 1.88e-05,
-      "loss": 0.19,
+      "loss": 0.8,
       "step": 95
     },
     {
       "epoch": 0.001757437070938215,
-      "grad_norm": 0.866124153137207,
+      "grad_norm": 5.368072032928467,
       "learning_rate": 1.9e-05,
-      "loss": 0.1134,
+      "loss": 0.6446,
       "step": 96
     },
     {
       "epoch": 0.0017757437070938214,
-      "grad_norm": 6.6813883781433105,
+      "grad_norm": 3.758234739303589,
       "learning_rate": 1.9200000000000003e-05,
-      "loss": 0.5447,
+      "loss": 0.4276,
       "step": 97
     },
     {
       "epoch": 0.001794050343249428,
-      "grad_norm": 1.0230540037155151,
+      "grad_norm": 5.807685852050781,
       "learning_rate": 1.94e-05,
-      "loss": 0.1068,
+      "loss": 0.4856,
       "step": 98
     },
     {
       "epoch": 0.0018123569794050343,
-      "grad_norm": 3.497803211212158,
+      "grad_norm": 4.966123104095459,
       "learning_rate": 1.9600000000000002e-05,
-      "loss": 0.1331,
+      "loss": 0.3153,
       "step": 99
     },
     {
       "epoch": 0.0018306636155606408,
-      "grad_norm": 3.7190122604370117,
+      "grad_norm": 6.432656764984131,
       "learning_rate": 1.98e-05,
-      "loss": 0.1459,
+      "loss": 0.4553,
       "step": 100
     },
     {
       "epoch": 0.0018489702517162471,
-      "grad_norm": 0.7365463972091675,
+      "grad_norm": 6.9228410720825195,
       "learning_rate": 2e-05,
-      "loss": 0.1014,
+      "loss": 0.375,
       "step": 101
     },
     {
       "epoch": 0.0018672768878718535,
-      "grad_norm": 5.429503440856934,
+      "grad_norm": 5.237428665161133,
       "learning_rate": 1.9999939076577906e-05,
-      "loss": 0.4682,
+      "loss": 0.484,
       "step": 102
     },
     {
       "epoch": 0.00188558352402746,
-      "grad_norm": 1.353772521018982,
+      "grad_norm": 5.173490524291992,
       "learning_rate": 1.9999756307053947e-05,
-      "loss": 0.1411,
+      "loss": 0.4135,
       "step": 103
     },
     {
       "epoch": 0.0019038901601830663,
-      "grad_norm": 7.793995380401611,
+      "grad_norm": 4.331842422485352,
       "learning_rate": 1.9999451693655125e-05,
-      "loss": 0.2016,
+      "loss": 0.3857,
       "step": 104
     },
     {
       "epoch": 0.0019221967963386727,
-      "grad_norm": 9.21011734008789,
+      "grad_norm": 6.129421710968018,
       "learning_rate": 1.9999025240093045e-05,
-      "loss": 0.1819,
+      "loss": 0.5126,
       "step": 105
     },
     {
       "epoch": 0.0019405034324942792,
-      "grad_norm": 0.5532894134521484,
+      "grad_norm": 4.619616985321045,
       "learning_rate": 1.9998476951563914e-05,
-      "loss": 0.103,
+      "loss": 0.477,
       "step": 106
     },
     {
       "epoch": 0.0019588100686498858,
-      "grad_norm": 0.793460488319397,
+      "grad_norm": 3.936915397644043,
       "learning_rate": 1.9997806834748455e-05,
-      "loss": 0.0985,
+      "loss": 0.3516,
       "step": 107
     },
     {
       "epoch": 0.001977116704805492,
-      "grad_norm": 1.2488313913345337,
+      "grad_norm": 8.568585395812988,
       "learning_rate": 1.9997014897811834e-05,
-      "loss": 0.1235,
+      "loss": 0.7743,
       "step": 108
     },
     {
       "epoch": 0.0019954233409610984,
-      "grad_norm": 12.01301383972168,
+      "grad_norm": 4.892576694488525,
       "learning_rate": 1.9996101150403543e-05,
-      "loss": 0.2274,
+      "loss": 0.4471,
       "step": 109
     },
     {
       "epoch": 0.002013729977116705,
-      "grad_norm": 6.007355213165283,
+      "grad_norm": 10.018097877502441,
       "learning_rate": 1.9995065603657317e-05,
-      "loss": 0.1955,
+      "loss": 1.1247,
       "step": 110
     },
     {
       "epoch": 0.002032036613272311,
-      "grad_norm": 2.662137985229492,
+      "grad_norm": 5.875444412231445,
       "learning_rate": 1.999390827019096e-05,
-      "loss": 0.1412,
+      "loss": 0.7445,
       "step": 111
     },
     {
       "epoch": 0.0020503432494279176,
-      "grad_norm": 0.6729631423950195,
+      "grad_norm": 7.044773101806641,
       "learning_rate": 1.999262916410621e-05,
-      "loss": 0.0943,
+      "loss": 0.4433,
       "step": 112
     },
     {
       "epoch": 0.002068649885583524,
-      "grad_norm": 9.703642845153809,
+      "grad_norm": 4.854395866394043,
       "learning_rate": 1.9991228300988586e-05,
-      "loss": 0.4103,
+      "loss": 0.4177,
       "step": 113
     },
     {
       "epoch": 0.0020869565217391303,
-      "grad_norm": 1.5533950328826904,
+      "grad_norm": 3.947049379348755,
       "learning_rate": 1.998970569790715e-05,
-      "loss": 0.1147,
+      "loss": 0.3204,
       "step": 114
     },
     {
       "epoch": 0.002105263157894737,
-      "grad_norm": 8.093120574951172,
+      "grad_norm": 2.9266178607940674,
       "learning_rate": 1.9988061373414342e-05,
-      "loss": 0.3097,
+      "loss": 0.3421,
       "step": 115
     },
     {
       "epoch": 0.0021235697940503433,
-      "grad_norm": 5.31638765335083,
+      "grad_norm": 14.068440437316895,
       "learning_rate": 1.9986295347545738e-05,
-      "loss": 0.2167,
+      "loss": 1.2803,
       "step": 116
     },
     {
       "epoch": 0.0021418764302059494,
-      "grad_norm": 12.155153274536133,
+      "grad_norm": 2.746870279312134,
       "learning_rate": 1.9984407641819812e-05,
-      "loss": 0.7526,
+      "loss": 0.1913,
       "step": 117
     },
     {
       "epoch": 0.002160183066361556,
-      "grad_norm": 5.362995624542236,
+      "grad_norm": 5.603977680206299,
       "learning_rate": 1.9982398279237657e-05,
-      "loss": 0.2234,
+      "loss": 0.535,
       "step": 118
     },
     {
       "epoch": 0.0021784897025171625,
-      "grad_norm": 3.9258711338043213,
+      "grad_norm": 4.298122882843018,
       "learning_rate": 1.9980267284282718e-05,
-      "loss": 0.1682,
+      "loss": 0.3521,
       "step": 119
     },
     {
       "epoch": 0.002196796338672769,
-      "grad_norm": 5.492271900177002,
+      "grad_norm": 3.4252305030822754,
       "learning_rate": 1.9978014682920503e-05,
-      "loss": 0.2237,
+      "loss": 0.3127,
       "step": 120
     },
     {
       "epoch": 0.002215102974828375,
-      "grad_norm": 7.385561466217041,
+      "grad_norm": 4.943688869476318,
       "learning_rate": 1.9975640502598243e-05,
-      "loss": 0.2285,
+      "loss": 0.2819,
       "step": 121
     },
     {
       "epoch": 0.0022334096109839817,
-      "grad_norm": 1.9505259990692139,
+      "grad_norm": 6.345489978790283,
       "learning_rate": 1.997314477224458e-05,
-      "loss": 0.1422,
+      "loss": 0.8686,
       "step": 122
     },
     {
       "epoch": 0.0022517162471395883,
-      "grad_norm": 1.234209418296814,
+      "grad_norm": 6.48477840423584,
       "learning_rate": 1.9970527522269204e-05,
-      "loss": 0.1491,
+      "loss": 0.4396,
       "step": 123
     },
     {
       "epoch": 0.0022700228832951944,
-      "grad_norm": 17.440046310424805,
+      "grad_norm": 12.649872779846191,
       "learning_rate": 1.9967788784562474e-05,
-      "loss": 1.079,
+      "loss": 0.4447,
       "step": 124
     },
     {
       "epoch": 0.002288329519450801,
-      "grad_norm": 3.154329538345337,
+      "grad_norm": 3.368851900100708,
       "learning_rate": 1.9964928592495046e-05,
-      "loss": 0.1235,
+      "loss": 0.213,
       "step": 125
     },
     {
       "epoch": 0.0023066361556064075,
-      "grad_norm": 0.5870721340179443,
+      "grad_norm": 5.273118019104004,
       "learning_rate": 1.9961946980917457e-05,
-      "loss": 0.0818,
+      "loss": 0.6374,
       "step": 126
     },
     {
       "epoch": 0.0023249427917620136,
-      "grad_norm": 1.0891448259353638,
+      "grad_norm": 8.816353797912598,
       "learning_rate": 1.9958843986159705e-05,
-      "loss": 0.1008,
+      "loss": 0.8738,
       "step": 127
     },
     {
       "epoch": 0.00234324942791762,
-      "grad_norm": 0.8425506353378296,
+      "grad_norm": 7.0079345703125,
       "learning_rate": 1.99556196460308e-05,
-      "loss": 0.1039,
+      "loss": 0.7057,
       "step": 128
     },
     {
       "epoch": 0.0023615560640732267,
-      "grad_norm": 0.9806830883026123,
+      "grad_norm": 2.5678298473358154,
       "learning_rate": 1.9952273999818312e-05,
-      "loss": 0.099,
+      "loss": 0.182,
       "step": 129
     },
     {
       "epoch": 0.0023798627002288328,
-      "grad_norm": 0.5745424032211304,
+      "grad_norm": 5.133856773376465,
       "learning_rate": 1.9948807088287884e-05,
-      "loss": 0.086,
+      "loss": 0.3598,
       "step": 130
     },
     {
       "epoch": 0.0023981693363844393,
-      "grad_norm": 0.7142914533615112,
+      "grad_norm": 9.948471069335938,
       "learning_rate": 1.9945218953682736e-05,
-      "loss": 0.0826,
+      "loss": 0.4483,
       "step": 131
     },
     {
       "epoch": 0.002416475972540046,
-      "grad_norm": 32.16014862060547,
+      "grad_norm": 5.9176459312438965,
       "learning_rate": 1.9941509639723155e-05,
-      "loss": 1.1946,
+      "loss": 0.8298,
       "step": 132
     },
     {
       "epoch": 0.002434782608695652,
-      "grad_norm": 0.8117746114730835,
+      "grad_norm": 2.4215924739837646,
       "learning_rate": 1.9937679191605964e-05,
-      "loss": 0.1023,
+      "loss": 0.1617,
       "step": 133
     },
     {
       "epoch": 0.0024530892448512585,
-      "grad_norm": 1.2140308618545532,
+      "grad_norm": 11.955750465393066,
       "learning_rate": 1.9933727656003964e-05,
-      "loss": 0.0924,
+      "loss": 0.609,
       "step": 134
     },
     {
       "epoch": 0.002471395881006865,
-      "grad_norm": 0.5409651398658752,
+      "grad_norm": 5.947329998016357,
       "learning_rate": 1.992965508106537e-05,
-      "loss": 0.0792,
+      "loss": 0.5718,
       "step": 135
     },
     {
       "epoch": 0.0024897025171624716,
-      "grad_norm": 0.8377587795257568,
+      "grad_norm": 5.794270038604736,
       "learning_rate": 1.9925461516413224e-05,
-      "loss": 0.1254,
+      "loss": 0.5392,
       "step": 136
     },
     {
       "epoch": 0.0025080091533180777,
-      "grad_norm": 0.7054142355918884,
+      "grad_norm": 5.274944305419922,
       "learning_rate": 1.9921147013144782e-05,
-      "loss": 0.0895,
+      "loss": 0.5278,
       "step": 137
     },
     {
       "epoch": 0.0025263157894736842,
-      "grad_norm": 4.393170356750488,
+      "grad_norm": 5.6011962890625,
       "learning_rate": 1.9916711623830904e-05,
-      "loss": 0.1658,
+      "loss": 0.4404,
       "step": 138
     },
     {
       "epoch": 0.0025446224256292908,
-      "grad_norm": 0.8794388771057129,
+      "grad_norm": 7.330393314361572,
       "learning_rate": 1.991215540251542e-05,
-      "loss": 0.1007,
+      "loss": 0.6554,
       "step": 139
     },
     {
       "epoch": 0.002562929061784897,
-      "grad_norm": 0.8057275414466858,
+      "grad_norm": 8.020462036132812,
       "learning_rate": 1.9907478404714438e-05,
-      "loss": 0.0946,
+      "loss": 0.7649,
       "step": 140
     },
     {
       "epoch": 0.0025812356979405034,
-      "grad_norm": 13.412788391113281,
+      "grad_norm": 3.3086674213409424,
       "learning_rate": 1.9902680687415704e-05,
-      "loss": 0.1297,
+      "loss": 0.3638,
       "step": 141
     },
     {
       "epoch": 0.00259954233409611,
-      "grad_norm": 0.5345849990844727,
+      "grad_norm": 7.1618781089782715,
       "learning_rate": 1.989776230907789e-05,
-      "loss": 0.0931,
+      "loss": 0.4597,
       "step": 142
     },
     {
       "epoch": 0.002617848970251716,
-      "grad_norm": 1.7970515489578247,
+      "grad_norm": 5.93783712387085,
       "learning_rate": 1.9892723329629885e-05,
-      "loss": 0.1706,
+      "loss": 0.4452,
       "step": 143
     },
     {
       "epoch": 0.0026361556064073226,
-      "grad_norm": 1.6152037382125854,
+      "grad_norm": 5.123695373535156,
       "learning_rate": 1.988756381047006e-05,
-      "loss": 0.1303,
+      "loss": 0.5444,
       "step": 144
     },
     {
       "epoch": 0.002654462242562929,
-      "grad_norm": 1.8181124925613403,
+      "grad_norm": 7.8915181159973145,
       "learning_rate": 1.988228381446553e-05,
-      "loss": 0.1497,
+      "loss": 0.758,
       "step": 145
     },
     {
       "epoch": 0.0026727688787185353,
-      "grad_norm": 0.9535139203071594,
+      "grad_norm": 7.822751998901367,
       "learning_rate": 1.9876883405951378e-05,
-      "loss": 0.0784,
+      "loss": 0.6956,
       "step": 146
     },
     {
       "epoch": 0.002691075514874142,
-      "grad_norm": 0.7281251549720764,
+      "grad_norm": 2.519596815109253,
       "learning_rate": 1.987136265072988e-05,
-      "loss": 0.0951,
+      "loss": 0.238,
       "step": 147
     },
     {
       "epoch": 0.0027093821510297484,
-      "grad_norm": 2.499201536178589,
+      "grad_norm": 3.658250093460083,
       "learning_rate": 1.9865721616069695e-05,
-      "loss": 0.1303,
+      "loss": 0.3844,
       "step": 148
     },
     {
       "epoch": 0.0027276887871853545,
-      "grad_norm": 0.7563035488128662,
+      "grad_norm": 5.0784430503845215,
       "learning_rate": 1.985996037070505e-05,
-      "loss": 0.079,
+      "loss": 0.5564,
       "step": 149
     },
     {
       "epoch": 0.002745995423340961,
-      "grad_norm": 19.374439239501953,
+      "grad_norm": 6.504764556884766,
       "learning_rate": 1.9854078984834904e-05,
-      "loss": 0.5997,
+      "loss": 0.4866,
       "step": 150
     },
     {
       "epoch": 0.0027643020594965676,
-      "grad_norm": 0.9670190811157227,
+      "grad_norm": 4.09471321105957,
       "learning_rate": 1.9848077530122083e-05,
-      "loss": 0.088,
+      "loss": 0.2998,
       "step": 151
     },
     {
       "epoch": 0.002782608695652174,
-      "grad_norm": 0.38407567143440247,
+      "grad_norm": 5.1176886558532715,
       "learning_rate": 1.984195607969242e-05,
-      "loss": 0.0744,
+      "loss": 0.2826,
       "step": 152
     },
     {
       "epoch": 0.00280091533180778,
-      "grad_norm": 0.6701276302337646,
+      "grad_norm": 12.725276947021484,
       "learning_rate": 1.983571470813386e-05,
-      "loss": 0.0908,
+      "loss": 0.7915,
       "step": 153
     },
     {
       "epoch": 0.0028192219679633868,
-      "grad_norm": 0.6903744339942932,
+      "grad_norm": 7.101748943328857,
       "learning_rate": 1.9829353491495545e-05,
-      "loss": 0.0976,
+      "loss": 0.7455,
       "step": 154
     },
     {
       "epoch": 0.0028375286041189933,
-      "grad_norm": 0.9913358092308044,
+      "grad_norm": 6.615306854248047,
       "learning_rate": 1.982287250728689e-05,
-      "loss": 0.0803,
+      "loss": 0.8178,
       "step": 155
     },
     {
       "epoch": 0.0028558352402745994,
-      "grad_norm": 1.005542278289795,
+      "grad_norm": 6.7726240158081055,
       "learning_rate": 1.9816271834476642e-05,
-      "loss": 0.1147,
+      "loss": 0.3764,
       "step": 156
     },
     {
       "epoch": 0.002874141876430206,
-      "grad_norm": 0.7195746898651123,
+      "grad_norm": 4.513782978057861,
       "learning_rate": 1.9809551553491918e-05,
-      "loss": 0.0974,
+      "loss": 0.3002,
       "step": 157
     },
     {
       "epoch": 0.0028924485125858125,
-      "grad_norm": 18.382600784301758,
+      "grad_norm": 6.409885883331299,
       "learning_rate": 1.9802711746217222e-05,
-      "loss": 0.5985,
+      "loss": 0.4299,
       "step": 158
     },
     {
       "epoch": 0.0029107551487414186,
-      "grad_norm": 0.4871819317340851,
+      "grad_norm": 1.4180457592010498,
       "learning_rate": 1.979575249599344e-05,
-      "loss": 0.0767,
+      "loss": 0.0436,
       "step": 159
     },
     {
       "epoch": 0.002929061784897025,
-      "grad_norm": 0.7498767375946045,
+      "grad_norm": 7.368013858795166,
       "learning_rate": 1.9788673887616852e-05,
-      "loss": 0.0701,
+      "loss": 0.4066,
       "step": 160
     },
     {
       "epoch": 0.0029473684210526317,
-      "grad_norm": 0.6817031502723694,
+      "grad_norm": 3.97849178314209,
       "learning_rate": 1.9781476007338058e-05,
-      "loss": 0.0799,
+      "loss": 0.2171,
       "step": 161
     },
     {
       "epoch": 0.002965675057208238,
-      "grad_norm": 0.8051683902740479,
+      "grad_norm": 8.33909797668457,
       "learning_rate": 1.9774158942860962e-05,
-      "loss": 0.0794,
+      "loss": 0.5762,
       "step": 162
     },
     {
       "epoch": 0.0029839816933638443,
-      "grad_norm": 0.9560118913650513,
+      "grad_norm": 1.4704389572143555,
       "learning_rate": 1.9766722783341682e-05,
-      "loss": 0.0703,
+      "loss": 0.1528,
       "step": 163
     },
     {
       "epoch": 0.003002288329519451,
-      "grad_norm": 0.9312361478805542,
+      "grad_norm": 6.321743965148926,
       "learning_rate": 1.9759167619387474e-05,
-      "loss": 0.0846,
+      "loss": 0.4852,
       "step": 164
     },
     {
       "epoch": 0.0030205949656750574,
-      "grad_norm": 1.9724496603012085,
+      "grad_norm": 8.366913795471191,
       "learning_rate": 1.9751493543055634e-05,
-      "loss": 0.0864,
+      "loss": 0.4964,
       "step": 165
     },
     {
       "epoch": 0.0030389016018306635,
-      "grad_norm": 0.4255484640598297,
+      "grad_norm": 6.21965217590332,
       "learning_rate": 1.9743700647852356e-05,
-      "loss": 0.0727,
+      "loss": 0.527,
       "step": 166
     },
     {
       "epoch": 0.00305720823798627,
-      "grad_norm": 0.600476086139679,
+      "grad_norm": 9.058350563049316,
       "learning_rate": 1.9735789028731603e-05,
-      "loss": 0.0714,
+      "loss": 0.9455,
       "step": 167
     },
     {
       "epoch": 0.0030755148741418766,
-      "grad_norm": 3.5824263095855713,
+      "grad_norm": 7.52754020690918,
       "learning_rate": 1.972775878209397e-05,
-      "loss": 0.1631,
+      "loss": 0.618,
       "step": 168
     },
     {
       "epoch": 0.0030938215102974827,
-      "grad_norm": 3.309394359588623,
+      "grad_norm": 8.560585975646973,
       "learning_rate": 1.9719610005785466e-05,
-      "loss": 0.2207,
+      "loss": 0.476,
       "step": 169
     },
     {
       "epoch": 0.0031121281464530893,
-      "grad_norm": 1.2910877466201782,
+      "grad_norm": 7.208701133728027,
       "learning_rate": 1.971134279909636e-05,
-      "loss": 0.1144,
+      "loss": 0.7781,
       "step": 170
     },
     {
       "epoch": 0.003130434782608696,
-      "grad_norm": 0.6535178422927856,
+      "grad_norm": 5.757314205169678,
       "learning_rate": 1.9702957262759964e-05,
-      "loss": 0.0736,
+      "loss": 0.518,
       "step": 171
     },
     {
       "epoch": 0.003148741418764302,
-      "grad_norm": 1.0068871974945068,
+      "grad_norm": 2.8194165229797363,
       "learning_rate": 1.9694453498951392e-05,
-      "loss": 0.0982,
+      "loss": 0.2128,
       "step": 172
     },
     {
       "epoch": 0.0031670480549199085,
-      "grad_norm": 2.130021810531616,
+      "grad_norm": 3.0178167819976807,
       "learning_rate": 1.9685831611286312e-05,
-      "loss": 0.0931,
+      "loss": 0.3055,
       "step": 173
     },
     {
       "epoch": 0.003185354691075515,
-      "grad_norm": 0.7793130278587341,
+      "grad_norm": 1.6432427167892456,
       "learning_rate": 1.9677091704819714e-05,
-      "loss": 0.1053,
+      "loss": 0.216,
       "step": 174
     },
     {
       "epoch": 0.003203661327231121,
-      "grad_norm": 0.5476394295692444,
+      "grad_norm": 8.396689414978027,
       "learning_rate": 1.9668233886044597e-05,
-      "loss": 0.0602,
+      "loss": 0.6197,
       "step": 175
     },
     {
       "epoch": 0.0032219679633867277,
-      "grad_norm": 5.817631244659424,
+      "grad_norm": 7.577538967132568,
       "learning_rate": 1.9659258262890683e-05,
-      "loss": 0.3113,
+      "loss": 0.7145,
       "step": 176
     },
     {
       "epoch": 0.003240274599542334,
-      "grad_norm": 2.4060657024383545,
+      "grad_norm": 4.660325527191162,
       "learning_rate": 1.9650164944723116e-05,
-      "loss": 0.1207,
+      "loss": 0.3487,
       "step": 177
     },
     {
       "epoch": 0.0032585812356979403,
-      "grad_norm": 3.327852249145508,
+      "grad_norm": 2.1696557998657227,
       "learning_rate": 1.96409540423411e-05,
-      "loss": 0.1366,
+      "loss": 0.3491,
       "step": 178
     },
     {
       "epoch": 0.003276887871853547,
-      "grad_norm": 0.4215916395187378,
+      "grad_norm": 3.31132435798645,
       "learning_rate": 1.9631625667976584e-05,
-      "loss": 0.0706,
+      "loss": 0.2511,
       "step": 179
     },
     {
       "epoch": 0.0032951945080091534,
-      "grad_norm": 3.387308120727539,
+      "grad_norm": 5.26754903793335,
       "learning_rate": 1.9622179935292855e-05,
-      "loss": 0.169,
+      "loss": 0.4235,
       "step": 180
     },
     {
       "epoch": 0.00331350114416476,
-      "grad_norm": 0.352518230676651,
+      "grad_norm": 4.9605865478515625,
       "learning_rate": 1.961261695938319e-05,
-      "loss": 0.063,
+      "loss": 0.2065,
       "step": 181
     },
     {
       "epoch": 0.003331807780320366,
-      "grad_norm": 2.262887954711914,
+      "grad_norm": 5.259721755981445,
       "learning_rate": 1.9602936856769432e-05,
-      "loss": 0.1339,
+      "loss": 0.3804,
       "step": 182
     },
     {
       "epoch": 0.0033501144164759726,
-      "grad_norm": 3.2673592567443848,
+      "grad_norm": 3.692326784133911,
       "learning_rate": 1.9593139745400575e-05,
-      "loss": 0.1469,
+      "loss": 0.3388,
       "step": 183
     },
     {
       "epoch": 0.003368421052631579,
-      "grad_norm": 3.5156006813049316,
+      "grad_norm": 4.228832721710205,
       "learning_rate": 1.9583225744651334e-05,
-      "loss": 0.1552,
+      "loss": 0.2778,
       "step": 184
     },
     {
       "epoch": 0.0033867276887871852,
-      "grad_norm": 1.561367392539978,
+      "grad_norm": 8.879986763000488,
       "learning_rate": 1.9573194975320672e-05,
-      "loss": 0.0796,
+      "loss": 0.7031,
       "step": 185
     },
     {
       "epoch": 0.003405034324942792,
-      "grad_norm": 1.4852555990219116,
+      "grad_norm": 7.6145853996276855,
       "learning_rate": 1.9563047559630356e-05,
-      "loss": 0.0816,
+      "loss": 0.6261,
       "step": 186
     },
     {
       "epoch": 0.0034233409610983983,
-      "grad_norm": 1.7594807147979736,
+      "grad_norm": 10.076028823852539,
       "learning_rate": 1.9552783621223437e-05,
-      "loss": 0.0937,
+      "loss": 0.6617,
       "step": 187
     },
     {
       "epoch": 0.0034416475972540044,
-      "grad_norm": 4.05676794052124,
+      "grad_norm": 4.281607151031494,
       "learning_rate": 1.954240328516277e-05,
-      "loss": 0.1537,
+      "loss": 0.2965,
       "step": 188
     },
     {
       "epoch": 0.003459954233409611,
-      "grad_norm": 0.46737217903137207,
+      "grad_norm": 3.990032911300659,
       "learning_rate": 1.9531906677929472e-05,
-      "loss": 0.0747,
+      "loss": 0.2232,
       "step": 189
     },
     {
       "epoch": 0.0034782608695652175,
-      "grad_norm": 0.5305610299110413,
+      "grad_norm": 10.36355209350586,
       "learning_rate": 1.9521293927421388e-05,
-      "loss": 0.0744,
+      "loss": 0.3606,
       "step": 190
     },
     {
       "epoch": 0.0034965675057208236,
-      "grad_norm": 0.6470791101455688,
+      "grad_norm": 9.396668434143066,
       "learning_rate": 1.9510565162951538e-05,
-      "loss": 0.0668,
+      "loss": 0.6809,
       "step": 191
     },
     {
       "epoch": 0.00351487414187643,
-      "grad_norm": 0.7232733368873596,
+      "grad_norm": 4.969096660614014,
       "learning_rate": 1.9499720515246524e-05,
-      "loss": 0.1023,
+      "loss": 0.3711,
       "step": 192
     },
     {
       "epoch": 0.0035331807780320367,
-      "grad_norm": 0.6899909973144531,
+      "grad_norm": 2.855027914047241,
       "learning_rate": 1.9488760116444966e-05,
-      "loss": 0.0806,
+      "loss": 0.3334,
       "step": 193
     },
     {
       "epoch": 0.003551487414187643,
-      "grad_norm": 0.626487135887146,
+      "grad_norm": 9.800653457641602,
       "learning_rate": 1.947768410009586e-05,
-      "loss": 0.0704,
+      "loss": 0.8046,
       "step": 194
     },
     {
       "epoch": 0.0035697940503432494,
-      "grad_norm": 17.721529006958008,
+      "grad_norm": 6.4518632888793945,
       "learning_rate": 1.9466492601156964e-05,
-      "loss": 0.2869,
+      "loss": 0.7623,
       "step": 195
     },
     {
       "epoch": 0.003588100686498856,
-      "grad_norm": 0.32237544655799866,
+      "grad_norm": 9.151044845581055,
       "learning_rate": 1.945518575599317e-05,
-      "loss": 0.0599,
+      "loss": 0.4479,
       "step": 196
     },
     {
       "epoch": 0.0036064073226544625,
-      "grad_norm": 0.6628690361976624,
+      "grad_norm": 6.074557304382324,
       "learning_rate": 1.944376370237481e-05,
-      "loss": 0.0759,
+      "loss": 0.3635,
       "step": 197
     },
     {
       "epoch": 0.0036247139588100686,
-      "grad_norm": 0.4883255660533905,
+      "grad_norm": 6.840311050415039,
       "learning_rate": 1.943222657947601e-05,
-      "loss": 0.065,
+      "loss": 0.3051,
       "step": 198
     },
     {
       "epoch": 0.003643020594965675,
-      "grad_norm": 0.5567423701286316,
+      "grad_norm": 3.7585532665252686,
       "learning_rate": 1.942057452787297e-05,
-      "loss": 0.056,
+      "loss": 0.3266,
       "step": 199
     },
     {
       "epoch": 0.0036613272311212816,
-      "grad_norm": 1.4766464233398438,
+      "grad_norm": 4.411327838897705,
       "learning_rate": 1.9408807689542257e-05,
-      "loss": 0.0866,
+      "loss": 0.2551,
       "step": 200
     },
     {
       "epoch": 0.0036796338672768878,
-      "grad_norm": 0.43290388584136963,
+      "grad_norm": 3.8587820529937744,
       "learning_rate": 1.9396926207859085e-05,
-      "loss": 0.0663,
+      "loss": 0.3105,
       "step": 201
     },
     {
       "epoch": 0.0036979405034324943,
-      "grad_norm": 0.8691856265068054,
+      "grad_norm": 4.376249313354492,
       "learning_rate": 1.938493022759556e-05,
-      "loss": 0.0768,
+      "loss": 0.2243,
       "step": 202
     },
     {
       "epoch": 0.003716247139588101,
-      "grad_norm": 2.7752292156219482,
+      "grad_norm": 5.54730224609375,
       "learning_rate": 1.937281989491892e-05,
-      "loss": 0.1332,
+      "loss": 0.4158,
       "step": 203
     },
     {
       "epoch": 0.003734553775743707,
-      "grad_norm": 0.6449118852615356,
+      "grad_norm": 2.6035118103027344,
       "learning_rate": 1.9360595357389735e-05,
-      "loss": 0.0826,
+      "loss": 0.1788,
       "step": 204
     },
     {
       "epoch": 0.0037528604118993135,
-      "grad_norm": 0.4738762676715851,
+      "grad_norm": 10.728706359863281,
       "learning_rate": 1.9348256763960146e-05,
-      "loss": 0.0666,
+      "loss": 1.0367,
       "step": 205
     },
     {
       "epoch": 0.00377116704805492,
-      "grad_norm": 0.7696526646614075,
+      "grad_norm": 5.158708572387695,
       "learning_rate": 1.9335804264972018e-05,
-      "loss": 0.0807,
+      "loss": 0.4394,
       "step": 206
     },
     {
       "epoch": 0.003789473684210526,
-      "grad_norm": 1.0092347860336304,
+      "grad_norm": 9.297528266906738,
       "learning_rate": 1.9323238012155125e-05,
-      "loss": 0.0745,
+      "loss": 0.4115,
       "step": 207
     },
     {
       "epoch": 0.0038077803203661327,
-      "grad_norm": 4.243185043334961,
+      "grad_norm": 11.042555809020996,
       "learning_rate": 1.9310558158625286e-05,
-      "loss": 0.0814,
+      "loss": 0.7565,
       "step": 208
     },
     {
       "epoch": 0.0038260869565217392,
-      "grad_norm": 0.6016342043876648,
+      "grad_norm": 6.377527236938477,
       "learning_rate": 1.9297764858882516e-05,
-      "loss": 0.0908,
+      "loss": 0.5614,
       "step": 209
     },
     {
       "epoch": 0.0038443935926773453,
-      "grad_norm": 1.8694993257522583,
+      "grad_norm": 3.4690988063812256,
       "learning_rate": 1.9284858268809135e-05,
-      "loss": 0.0923,
+      "loss": 0.1462,
       "step": 210
     },
     {
       "epoch": 0.003862700228832952,
-      "grad_norm": 0.5105456113815308,
+      "grad_norm": 8.21645450592041,
       "learning_rate": 1.9271838545667876e-05,
-      "loss": 0.0711,
+      "loss": 0.6482,
       "step": 211
     },
     {
       "epoch": 0.0038810068649885584,
-      "grad_norm": 1.095218300819397,
+      "grad_norm": 13.70019817352295,
       "learning_rate": 1.925870584809995e-05,
-      "loss": 0.0865,
+      "loss": 1.3146,
       "step": 212
     },
     {
       "epoch": 0.003899313501144165,
-      "grad_norm": 0.6418291926383972,
+      "grad_norm": 6.1762375831604,
       "learning_rate": 1.9245460336123136e-05,
-      "loss": 0.0791,
+      "loss": 0.2371,
       "step": 213
     },
     {
       "epoch": 0.0039176201372997715,
-      "grad_norm": 1.9073235988616943,
+      "grad_norm": 3.4190640449523926,
       "learning_rate": 1.923210217112981e-05,
-      "loss": 0.0983,
+      "loss": 0.1756,
       "step": 214
     },
     {
       "epoch": 0.003935926773455377,
-      "grad_norm": 0.6898337006568909,
+      "grad_norm": 4.4201250076293945,
       "learning_rate": 1.9218631515885007e-05,
-      "loss": 0.0727,
+      "loss": 0.3832,
       "step": 215
     },
     {
       "epoch": 0.003954233409610984,
-      "grad_norm": 0.39293479919433594,
+      "grad_norm": 4.916990756988525,
       "learning_rate": 1.9205048534524405e-05,
-      "loss": 0.071,
+      "loss": 0.55,
       "step": 216
     },
     {
       "epoch": 0.00397254004576659,
-      "grad_norm": 8.08031940460205,
+      "grad_norm": 5.253897666931152,
       "learning_rate": 1.9191353392552346e-05,
-      "loss": 0.1062,
+      "loss": 0.4081,
       "step": 217
     },
     {
       "epoch": 0.003990846681922197,
-      "grad_norm": 11.346750259399414,
+      "grad_norm": 9.138065338134766,
       "learning_rate": 1.9177546256839814e-05,
-      "loss": 0.1452,
+      "loss": 0.4096,
       "step": 218
     },
     {
       "epoch": 0.004009153318077803,
-      "grad_norm": 0.4992099702358246,
+      "grad_norm": 5.5543437004089355,
       "learning_rate": 1.9163627295622397e-05,
-      "loss": 0.0832,
+      "loss": 0.3725,
       "step": 219
     },
     {
       "epoch": 0.00402745995423341,
-      "grad_norm": 0.46828693151474,
+      "grad_norm": 6.35120153427124,
       "learning_rate": 1.914959667849825e-05,
-      "loss": 0.0765,
+      "loss": 0.6631,
       "step": 220
     },
     {
       "epoch": 0.0040457665903890164,
-      "grad_norm": 1.182172179222107,
+      "grad_norm": 6.352338790893555,
       "learning_rate": 1.913545457642601e-05,
-      "loss": 0.0983,
+      "loss": 0.6608,
       "step": 221
     },
     {
       "epoch": 0.004064073226544622,
-      "grad_norm": 0.4203752875328064,
+      "grad_norm": 5.439535140991211,
       "learning_rate": 1.9121201161722732e-05,
-      "loss": 0.067,
+      "loss": 0.3461,
       "step": 222
     },
     {
       "epoch": 0.004082379862700229,
-      "grad_norm": 5.034867286682129,
+      "grad_norm": 6.05610990524292,
       "learning_rate": 1.910683660806177e-05,
-      "loss": 0.1064,
+      "loss": 0.7047,
       "step": 223
     },
     {
       "epoch": 0.004100686498855835,
-      "grad_norm": 0.5466944575309753,
+      "grad_norm": 16.13218879699707,
       "learning_rate": 1.9092361090470688e-05,
-      "loss": 0.0519,
+      "loss": 0.9417,
       "step": 224
     },
     {
       "epoch": 0.004118993135011442,
-      "grad_norm": 33.792999267578125,
+      "grad_norm": 11.939602851867676,
       "learning_rate": 1.907777478532909e-05,
-      "loss": 0.2122,
+      "loss": 0.4711,
       "step": 225
     },
     {
       "epoch": 0.004137299771167048,
-      "grad_norm": 1.5701090097427368,
+      "grad_norm": 9.328332901000977,
       "learning_rate": 1.9063077870366504e-05,
-      "loss": 0.1003,
+      "loss": 0.4825,
       "step": 226
     },
     {
       "epoch": 0.004155606407322655,
-      "grad_norm": 8.198397636413574,
+      "grad_norm": 3.4537312984466553,
       "learning_rate": 1.9048270524660197e-05,
-      "loss": 0.2533,
+      "loss": 0.2696,
       "step": 227
     },
     {
       "epoch": 0.0041739130434782605,
-      "grad_norm": 0.681303083896637,
+      "grad_norm": 8.465746879577637,
       "learning_rate": 1.903335292863301e-05,
-      "loss": 0.0715,
+      "loss": 0.3368,
       "step": 228
     },
     {
       "epoch": 0.004192219679633867,
-      "grad_norm": 0.4660561978816986,
+      "grad_norm": 2.80598783493042,
       "learning_rate": 1.901832526405114e-05,
-      "loss": 0.0546,
+      "loss": 0.2279,
       "step": 229
     },
     {
       "epoch": 0.004210526315789474,
-      "grad_norm": 0.4165865480899811,
+      "grad_norm": 9.187653541564941,
       "learning_rate": 1.9003187714021936e-05,
-      "loss": 0.0759,
+      "loss": 0.5421,
       "step": 230
     },
     {
       "epoch": 0.00422883295194508,
-      "grad_norm": 2.369324207305908,
+      "grad_norm": 2.9173386096954346,
       "learning_rate": 1.8987940462991673e-05,
-      "loss": 0.153,
+      "loss": 0.1865,
       "step": 231
     },
     {
       "epoch": 0.004247139588100687,
-      "grad_norm": 2.088148832321167,
+      "grad_norm": 8.808391571044922,
       "learning_rate": 1.8972583696743284e-05,
-      "loss": 0.1479,
+      "loss": 0.6127,
       "step": 232
     },
     {
       "epoch": 0.004265446224256293,
-      "grad_norm": 0.9253685474395752,
+      "grad_norm": 3.902742862701416,
       "learning_rate": 1.895711760239413e-05,
-      "loss": 0.0959,
+      "loss": 0.4918,
       "step": 233
     },
     {
       "epoch": 0.004283752860411899,
-      "grad_norm": 0.44856250286102295,
+      "grad_norm": 7.3988237380981445,
       "learning_rate": 1.8941542368393683e-05,
-      "loss": 0.0539,
+      "loss": 0.4754,
       "step": 234
     },
     {
       "epoch": 0.0043020594965675054,
-      "grad_norm": 0.4497087001800537,
+      "grad_norm": 3.434345006942749,
       "learning_rate": 1.892585818452126e-05,
-      "loss": 0.0591,
+      "loss": 0.2698,
       "step": 235
     },
     {
       "epoch": 0.004320366132723112,
-      "grad_norm": 0.5914357900619507,
+      "grad_norm": 5.183997631072998,
       "learning_rate": 1.891006524188368e-05,
-      "loss": 0.0615,
+      "loss": 0.2622,
       "step": 236
     },
     {
       "epoch": 0.0043386727688787185,
-      "grad_norm": 0.624913215637207,
+      "grad_norm": 4.6859235763549805,
       "learning_rate": 1.889416373291298e-05,
-      "loss": 0.0586,
+      "loss": 0.2881,
       "step": 237
     },
     {
       "epoch": 0.004356979405034325,
-      "grad_norm": 0.5881243348121643,
+      "grad_norm": 3.172283887863159,
       "learning_rate": 1.8878153851364013e-05,
-      "loss": 0.0673,
+      "loss": 0.2605,
       "step": 238
     },
     {
       "epoch": 0.004375286041189932,
-      "grad_norm": 2.140298366546631,
+      "grad_norm": 11.702736854553223,
       "learning_rate": 1.8862035792312148e-05,
-      "loss": 0.1119,
+      "loss": 0.6882,
       "step": 239
     },
     {
       "epoch": 0.004393592677345538,
-      "grad_norm": 0.5690272450447083,
+      "grad_norm": 6.005770683288574,
       "learning_rate": 1.884580975215084e-05,
-      "loss": 0.0576,
+      "loss": 0.4795,
       "step": 240
     },
     {
       "epoch": 0.004411899313501144,
-      "grad_norm": 0.9032683968544006,
+      "grad_norm": 3.117210865020752,
       "learning_rate": 1.8829475928589272e-05,
-      "loss": 0.0867,
+      "loss": 0.3978,
       "step": 241
     },
     {
       "epoch": 0.00443020594965675,
-      "grad_norm": 1.510884165763855,
+      "grad_norm": 6.215647220611572,
       "learning_rate": 1.8813034520649923e-05,
-      "loss": 0.1079,
+      "loss": 0.4373,
       "step": 242
     },
     {
       "epoch": 0.004448512585812357,
-      "grad_norm": 0.5766165852546692,
+      "grad_norm": 5.335672378540039,
       "learning_rate": 1.879648572866617e-05,
-      "loss": 0.0653,
+      "loss": 0.2903,
       "step": 243
     },
     {
       "epoch": 0.0044668192219679635,
-      "grad_norm": 6.860500812530518,
+      "grad_norm": 1.4431347846984863,
       "learning_rate": 1.8779829754279806e-05,
-      "loss": 0.1549,
+      "loss": 0.1387,
       "step": 244
     },
     {
       "epoch": 0.00448512585812357,
-      "grad_norm": 0.5166245102882385,
+      "grad_norm": 8.997931480407715,
       "learning_rate": 1.8763066800438638e-05,
-      "loss": 0.0819,
+      "loss": 0.7452,
       "step": 245
     },
     {
       "epoch": 0.0045034324942791765,
-      "grad_norm": 0.77106773853302,
+      "grad_norm": 7.0658721923828125,
       "learning_rate": 1.874619707139396e-05,
-      "loss": 0.0758,
+      "loss": 0.4082,
       "step": 246
     },
     {
       "epoch": 0.004521739130434782,
-      "grad_norm": 0.7701351046562195,
+      "grad_norm": 16.99602508544922,
       "learning_rate": 1.8729220772698096e-05,
-      "loss": 0.0607,
+      "loss": 0.5462,
       "step": 247
     },
     {
       "epoch": 0.004540045766590389,
-      "grad_norm": 0.4365614056587219,
+      "grad_norm": 4.291457653045654,
       "learning_rate": 1.8712138111201898e-05,
-      "loss": 0.0571,
+      "loss": 0.2849,
       "step": 248
     },
     {
       "epoch": 0.004558352402745995,
-      "grad_norm": 0.33743348717689514,
+      "grad_norm": 1.2436845302581787,
       "learning_rate": 1.869494929505219e-05,
-      "loss": 0.0545,
+      "loss": 0.1299,
       "step": 249
     },
     {
       "epoch": 0.004576659038901602,
-      "grad_norm": 10.381525993347168,
+      "grad_norm": 13.032325744628906,
       "learning_rate": 1.8677654533689287e-05,
-      "loss": 0.7511,
+      "loss": 1.1319,
       "step": 250
     },
     {
       "epoch": 0.004594965675057208,
-      "grad_norm": 1.2742400169372559,
+      "grad_norm": 4.819525718688965,
       "learning_rate": 1.866025403784439e-05,
-      "loss": 0.0939,
+      "loss": 0.2317,
       "step": 251
     },
     {
       "epoch": 0.004613272311212815,
-      "grad_norm": 30.52121353149414,
+      "grad_norm": 4.058289051055908,
       "learning_rate": 1.864274801953705e-05,
-      "loss": 0.3629,
+      "loss": 0.1868,
       "step": 252
     },
     {
       "epoch": 0.0046315789473684215,
-      "grad_norm": 0.7482168078422546,
+      "grad_norm": 5.123549461364746,
       "learning_rate": 1.8625136692072577e-05,
-      "loss": 0.0756,
+      "loss": 0.2682,
       "step": 253
     },
     {
       "epoch": 0.004649885583524027,
-      "grad_norm": 0.6211861371994019,
+      "grad_norm": 4.105663299560547,
       "learning_rate": 1.860742027003944e-05,
-      "loss": 0.0615,
+      "loss": 0.2852,
       "step": 254
     },
     {
       "epoch": 0.004668192219679634,
-      "grad_norm": 0.6214051246643066,
+      "grad_norm": 8.609756469726562,
       "learning_rate": 1.8589598969306646e-05,
-      "loss": 0.0928,
+      "loss": 0.638,
       "step": 255
     },
     {
       "epoch": 0.00468649885583524,
-      "grad_norm": 12.26028060913086,
+      "grad_norm": 10.366744041442871,
       "learning_rate": 1.8571673007021124e-05,
-      "loss": 0.1937,
+      "loss": 0.7038,
       "step": 256
     },
     {
       "epoch": 0.004704805491990847,
-      "grad_norm": 0.9524597525596619,
+      "grad_norm": 3.733368158340454,
       "learning_rate": 1.855364260160507e-05,
-      "loss": 0.1023,
+      "loss": 0.3285,
       "step": 257
     },
     {
       "epoch": 0.004723112128146453,
-      "grad_norm": 0.6243967413902283,
+      "grad_norm": 8.309182167053223,
       "learning_rate": 1.8535507972753275e-05,
-      "loss": 0.0811,
+      "loss": 0.9593,
       "step": 258
     },
     {
       "epoch": 0.00474141876430206,
-      "grad_norm": 0.44563159346580505,
+      "grad_norm": 9.302151679992676,
       "learning_rate": 1.851726934143048e-05,
-      "loss": 0.046,
+      "loss": 0.5591,
       "step": 259
     },
     {
       "epoch": 0.0047597254004576655,
-      "grad_norm": 1.6660269498825073,
+      "grad_norm": 1.1159979104995728,
       "learning_rate": 1.849892692986864e-05,
-      "loss": 0.1168,
+      "loss": 0.1096,
       "step": 260
     },
     {
       "epoch": 0.004778032036613272,
-      "grad_norm": 2.0700368881225586,
+      "grad_norm": 5.174903392791748,
       "learning_rate": 1.848048096156426e-05,
-      "loss": 0.154,
+      "loss": 0.2641,
       "step": 261
     },
     {
       "epoch": 0.004796338672768879,
-      "grad_norm": 0.3434784412384033,
+      "grad_norm": 4.628619194030762,
       "learning_rate": 1.8461931661275642e-05,
-      "loss": 0.0687,
+      "loss": 0.4448,
       "step": 262
     },
     {
       "epoch": 0.004814645308924485,
-      "grad_norm": 0.3164476454257965,
+      "grad_norm": 2.9997923374176025,
       "learning_rate": 1.8443279255020153e-05,
-      "loss": 0.0547,
+      "loss": 0.2183,
       "step": 263
     },
     {
       "epoch": 0.004832951945080092,
-      "grad_norm": 7.64108943939209,
+      "grad_norm": 9.500200271606445,
       "learning_rate": 1.842452397007148e-05,
-      "loss": 0.1533,
+      "loss": 0.4821,
       "step": 264
     },
     {
       "epoch": 0.004851258581235698,
-      "grad_norm": 0.3708806335926056,
+      "grad_norm": 7.443029403686523,
       "learning_rate": 1.8405666034956842e-05,
-      "loss": 0.0712,
+      "loss": 0.6391,
       "step": 265
     },
     {
       "epoch": 0.004869565217391304,
-      "grad_norm": 1.0373973846435547,
+      "grad_norm": 13.987396240234375,
       "learning_rate": 1.8386705679454243e-05,
-      "loss": 0.1111,
+      "loss": 0.8478,
       "step": 266
     },
     {
       "epoch": 0.0048878718535469105,
-      "grad_norm": 0.6912388205528259,
+      "grad_norm": 3.692591667175293,
       "learning_rate": 1.836764313458962e-05,
-      "loss": 0.0685,
+      "loss": 0.2808,
       "step": 267
     },
     {
       "epoch": 0.004906178489702517,
-      "grad_norm": 0.90571129322052,
+      "grad_norm": 6.317406177520752,
       "learning_rate": 1.8348478632634067e-05,
-      "loss": 0.0847,
+      "loss": 0.3462,
       "step": 268
     },
     {
       "epoch": 0.0049244851258581235,
-      "grad_norm": 0.39074182510375977,
+      "grad_norm": 10.023675918579102,
       "learning_rate": 1.8329212407100996e-05,
-      "loss": 0.0569,
+      "loss": 1.0451,
       "step": 269
     },
     {
       "epoch": 0.00494279176201373,
-      "grad_norm": 0.5868404507637024,
+      "grad_norm": 5.884584426879883,
       "learning_rate": 1.8309844692743283e-05,
-      "loss": 0.065,
+      "loss": 0.5844,
       "step": 270
     },
     {
       "epoch": 0.004961098398169337,
-      "grad_norm": 1.790574312210083,
+      "grad_norm": 2.740469217300415,
       "learning_rate": 1.8290375725550417e-05,
-      "loss": 0.1407,
+      "loss": 0.1969,
       "step": 271
     },
     {
       "epoch": 0.004979405034324943,
-      "grad_norm": 0.46091529726982117,
+      "grad_norm": 6.099292755126953,
       "learning_rate": 1.827080574274562e-05,
-      "loss": 0.0607,
+      "loss": 0.726,
       "step": 272
     },
     {
       "epoch": 0.004997711670480549,
-      "grad_norm": 0.5939885377883911,
+      "grad_norm": 6.810953140258789,
       "learning_rate": 1.8251134982782952e-05,
-      "loss": 0.0665,
+      "loss": 0.5525,
       "step": 273
     },
     {
       "epoch": 0.005016018306636155,
-      "grad_norm": 1.24867582321167,
+      "grad_norm": 17.33843231201172,
       "learning_rate": 1.8231363685344422e-05,
-      "loss": 0.1072,
+      "loss": 0.9222,
       "step": 274
     },
     {
       "epoch": 0.005034324942791762,
-      "grad_norm": 0.5410619378089905,
+      "grad_norm": 6.817041397094727,
       "learning_rate": 1.821149209133704e-05,
-      "loss": 0.05,
+      "loss": 0.4998,
       "step": 275
     },
     {
       "epoch": 0.0050526315789473685,
-      "grad_norm": 0.5314016938209534,
+      "grad_norm": 9.165572166442871,
       "learning_rate": 1.819152044288992e-05,
-      "loss": 0.0526,
+      "loss": 0.6976,
       "step": 276
     },
     {
       "epoch": 0.005070938215102975,
-      "grad_norm": 0.6637174487113953,
+      "grad_norm": 4.791163444519043,
       "learning_rate": 1.8171448983351284e-05,
-      "loss": 0.0632,
+      "loss": 0.4825,
       "step": 277
     },
     {
       "epoch": 0.0050892448512585816,
-      "grad_norm": 0.5429244637489319,
+      "grad_norm": 2.2928450107574463,
       "learning_rate": 1.815127795728554e-05,
-      "loss": 0.0622,
+      "loss": 0.1951,
       "step": 278
     },
     {
       "epoch": 0.005107551487414187,
-      "grad_norm": 0.8398737907409668,
+      "grad_norm": 4.821455955505371,
       "learning_rate": 1.8131007610470278e-05,
-      "loss": 0.0837,
+      "loss": 0.2852,
       "step": 279
     },
     {
       "epoch": 0.005125858123569794,
-      "grad_norm": 2.092486619949341,
+      "grad_norm": 6.22464656829834,
       "learning_rate": 1.8110638189893267e-05,
-      "loss": 0.1488,
+      "loss": 0.3738,
       "step": 280
     },
     {
       "epoch": 0.0051441647597254,
-      "grad_norm": 2.5059332847595215,
+      "grad_norm": 4.849123001098633,
       "learning_rate": 1.8090169943749477e-05,
-      "loss": 0.1148,
+      "loss": 0.296,
       "step": 281
     },
     {
       "epoch": 0.005162471395881007,
-      "grad_norm": 0.5902626514434814,
+      "grad_norm": 6.145839691162109,
       "learning_rate": 1.806960312143802e-05,
-      "loss": 0.0813,
+      "loss": 0.3723,
       "step": 282
     },
     {
       "epoch": 0.005180778032036613,
-      "grad_norm": 0.7806552052497864,
+      "grad_norm": 4.0611724853515625,
       "learning_rate": 1.804893797355914e-05,
-      "loss": 0.0717,
+      "loss": 0.4265,
       "step": 283
     },
     {
       "epoch": 0.00519908466819222,
-      "grad_norm": 0.7111049294471741,
+      "grad_norm": 3.252166271209717,
       "learning_rate": 1.8028174751911147e-05,
-      "loss": 0.0668,
+      "loss": 0.3467,
       "step": 284
     },
     {
       "epoch": 0.0052173913043478265,
-      "grad_norm": 0.71833735704422,
+      "grad_norm": 6.532531261444092,
       "learning_rate": 1.8007313709487334e-05,
-      "loss": 0.089,
+      "loss": 0.3968,
       "step": 285
     },
     {
       "epoch": 0.005235697940503432,
-      "grad_norm": 0.6862083673477173,
+      "grad_norm": 8.545121192932129,
       "learning_rate": 1.798635510047293e-05,
-      "loss": 0.0787,
+      "loss": 0.5772,
       "step": 286
     },
     {
       "epoch": 0.005254004576659039,
-      "grad_norm": 2.0314245223999023,
+      "grad_norm": 3.646812677383423,
       "learning_rate": 1.7965299180241963e-05,
-      "loss": 0.0918,
+      "loss": 0.1868,
       "step": 287
     },
     {
       "epoch": 0.005272311212814645,
-      "grad_norm": 8.271821975708008,
+      "grad_norm": 6.056759357452393,
       "learning_rate": 1.7944146205354182e-05,
-      "loss": 0.7081,
+      "loss": 0.615,
       "step": 288
     },
     {
       "epoch": 0.005290617848970252,
-      "grad_norm": 0.6371070146560669,
+      "grad_norm": 9.61752986907959,
       "learning_rate": 1.792289643355191e-05,
-      "loss": 0.0749,
+      "loss": 0.6434,
       "step": 289
     },
     {
       "epoch": 0.005308924485125858,
-      "grad_norm": 0.47990158200263977,
+      "grad_norm": 7.096751689910889,
       "learning_rate": 1.7901550123756906e-05,
-      "loss": 0.0791,
+      "loss": 0.3842,
       "step": 290
     },
     {
       "epoch": 0.005327231121281465,
-      "grad_norm": 0.5896520614624023,
+      "grad_norm": 3.8110721111297607,
       "learning_rate": 1.788010753606722e-05,
-      "loss": 0.0659,
+      "loss": 0.2328,
       "step": 291
     },
     {
       "epoch": 0.0053455377574370706,
-      "grad_norm": 0.6541833877563477,
+      "grad_norm": 5.858436584472656,
       "learning_rate": 1.785856893175402e-05,
-      "loss": 0.0697,
+      "loss": 0.431,
       "step": 292
     },
     {
       "epoch": 0.005363844393592677,
-      "grad_norm": 0.7276694178581238,
+      "grad_norm": 4.623099327087402,
       "learning_rate": 1.78369345732584e-05,
-      "loss": 0.067,
+      "loss": 0.337,
       "step": 293
     },
     {
       "epoch": 0.005382151029748284,
-      "grad_norm": 0.3601659834384918,
+      "grad_norm": 5.802640914916992,
       "learning_rate": 1.781520472418819e-05,
-      "loss": 0.0618,
+      "loss": 0.4879,
       "step": 294
     },
     {
       "epoch": 0.00540045766590389,
-      "grad_norm": 0.5127450227737427,
+      "grad_norm": 5.544589042663574,
       "learning_rate": 1.7793379649314743e-05,
-      "loss": 0.072,
+      "loss": 0.4911,
       "step": 295
     },
     {
       "epoch": 0.005418764302059497,
-      "grad_norm": 14.209413528442383,
+      "grad_norm": 6.768428802490234,
       "learning_rate": 1.777145961456971e-05,
-      "loss": 0.2248,
+      "loss": 0.5283,
       "step": 296
     },
     {
       "epoch": 0.005437070938215103,
-      "grad_norm": 0.7696670889854431,
+      "grad_norm": 4.4601731300354,
       "learning_rate": 1.7749444887041797e-05,
-      "loss": 0.0905,
+      "loss": 0.4052,
       "step": 297
     },
     {
       "epoch": 0.005455377574370709,
-      "grad_norm": 0.7976656556129456,
+      "grad_norm": 16.009952545166016,
       "learning_rate": 1.7727335734973512e-05,
-      "loss": 0.079,
+      "loss": 0.4814,
       "step": 298
     },
     {
       "epoch": 0.0054736842105263155,
-      "grad_norm": 0.8515931963920593,
+      "grad_norm": 10.445368766784668,
       "learning_rate": 1.7705132427757895e-05,
-      "loss": 0.0992,
+      "loss": 0.8704,
       "step": 299
     },
     {
       "epoch": 0.005491990846681922,
-      "grad_norm": 1.721758246421814,
+      "grad_norm": 3.6240663528442383,
       "learning_rate": 1.7682835235935236e-05,
-      "loss": 0.1088,
+      "loss": 0.1805,
       "step": 300
     },
     {
       "epoch": 0.005510297482837529,
-      "grad_norm": 1.963814377784729,
+      "grad_norm": 6.040288925170898,
       "learning_rate": 1.766044443118978e-05,
-      "loss": 0.0956,
+      "loss": 0.536,
       "step": 301
     },
     {
       "epoch": 0.005528604118993135,
-      "grad_norm": 0.5371356010437012,
+      "grad_norm": 5.494821548461914,
       "learning_rate": 1.7637960286346423e-05,
-      "loss": 0.0692,
+      "loss": 0.3142,
       "step": 302
     },
     {
       "epoch": 0.005546910755148742,
-      "grad_norm": 1.6635653972625732,
+      "grad_norm": 4.8304219245910645,
       "learning_rate": 1.761538307536737e-05,
-      "loss": 0.1238,
+      "loss": 0.444,
       "step": 303
     },
     {
       "epoch": 0.005565217391304348,
-      "grad_norm": 0.5697126984596252,
+      "grad_norm": 8.82863712310791,
       "learning_rate": 1.759271307334881e-05,
-      "loss": 0.0611,
+      "loss": 0.7043,
       "step": 304
     },
     {
       "epoch": 0.005583524027459954,
-      "grad_norm": 0.428237646818161,
+      "grad_norm": 11.274286270141602,
       "learning_rate": 1.7569950556517566e-05,
-      "loss": 0.0522,
+      "loss": 1.0306,
       "step": 305
     },
     {
       "epoch": 0.00560183066361556,
-      "grad_norm": 0.35257813334465027,
+      "grad_norm": 7.698188304901123,
       "learning_rate": 1.7547095802227723e-05,
-      "loss": 0.066,
+      "loss": 0.5291,
       "step": 306
     },
     {
       "epoch": 0.005620137299771167,
-      "grad_norm": 0.4639001488685608,
+      "grad_norm": 6.544892311096191,
       "learning_rate": 1.7524149088957244e-05,
-      "loss": 0.0679,
+      "loss": 0.679,
       "step": 307
     },
     {
       "epoch": 0.0056384439359267735,
-      "grad_norm": 0.4770897924900055,
+      "grad_norm": 13.220139503479004,
       "learning_rate": 1.7501110696304598e-05,
-      "loss": 0.0605,
+      "loss": 0.7442,
       "step": 308
     },
     {
       "epoch": 0.00565675057208238,
-      "grad_norm": 1.2645610570907593,
+      "grad_norm": 4.368415355682373,
       "learning_rate": 1.747798090498532e-05,
-      "loss": 0.115,
+      "loss": 0.3106,
       "step": 309
     },
     {
       "epoch": 0.005675057208237987,
-      "grad_norm": 1.2967489957809448,
+      "grad_norm": 5.565270900726318,
       "learning_rate": 1.7454759996828622e-05,
-      "loss": 0.0786,
+      "loss": 0.2814,
       "step": 310
     },
     {
       "epoch": 0.005693363844393592,
-      "grad_norm": 0.38287168741226196,
+      "grad_norm": 6.844931125640869,
       "learning_rate": 1.7431448254773943e-05,
-      "loss": 0.0706,
+      "loss": 0.5184,
       "step": 311
     },
     {
       "epoch": 0.005711670480549199,
-      "grad_norm": 35.53742218017578,
+      "grad_norm": 4.695760726928711,
       "learning_rate": 1.74080459628675e-05,
-      "loss": 0.4773,
+      "loss": 0.463,
       "step": 312
     },
     {
       "epoch": 0.005729977116704805,
-      "grad_norm": 0.30804502964019775,
+      "grad_norm": 7.724005222320557,
       "learning_rate": 1.7384553406258842e-05,
-      "loss": 0.0628,
+      "loss": 0.468,
       "step": 313
     },
     {
       "epoch": 0.005748283752860412,
-      "grad_norm": 25.294322967529297,
+      "grad_norm": 3.6156113147735596,
       "learning_rate": 1.7360970871197347e-05,
-      "loss": 0.4301,
+      "loss": 0.3026,
       "step": 314
     },
     {
       "epoch": 0.0057665903890160184,
-      "grad_norm": 13.664905548095703,
+      "grad_norm": 1.1042062044143677,
       "learning_rate": 1.7337298645028764e-05,
-      "loss": 0.9881,
+      "loss": 0.0863,
       "step": 315
     },
     {
       "epoch": 0.005784897025171625,
-      "grad_norm": 0.46416544914245605,
+      "grad_norm": 3.604750633239746,
       "learning_rate": 1.7313537016191706e-05,
-      "loss": 0.0684,
+      "loss": 0.1948,
       "step": 316
     },
     {
       "epoch": 0.0058032036613272315,
-      "grad_norm": 1.7975246906280518,
+      "grad_norm": 3.3758010864257812,
       "learning_rate": 1.7289686274214116e-05,
-      "loss": 0.1174,
+      "loss": 0.2802,
       "step": 317
     },
     {
       "epoch": 0.005821510297482837,
-      "grad_norm": 16.436054229736328,
+      "grad_norm": 12.799773216247559,
       "learning_rate": 1.7265746709709762e-05,
-      "loss": 0.4171,
+      "loss": 0.9552,
       "step": 318
     },
     {
       "epoch": 0.005839816933638444,
-      "grad_norm": 1.439911127090454,
+      "grad_norm": 2.934678077697754,
       "learning_rate": 1.7241718614374678e-05,
-      "loss": 0.1316,
+      "loss": 0.2117,
       "step": 319
     },
     {
       "epoch": 0.00585812356979405,
-      "grad_norm": 0.7375670671463013,
+      "grad_norm": 2.4731688499450684,
       "learning_rate": 1.7217602280983622e-05,
-      "loss": 0.074,
+      "loss": 0.1622,
       "step": 320
     },
     {
       "epoch": 0.005876430205949657,
-      "grad_norm": 10.360668182373047,
+      "grad_norm": 2.6618807315826416,
       "learning_rate": 1.7193398003386514e-05,
-      "loss": 0.2191,
+      "loss": 0.211,
       "step": 321
     },
     {
       "epoch": 0.005894736842105263,
-      "grad_norm": 1.8050432205200195,
+      "grad_norm": 5.7936906814575195,
       "learning_rate": 1.716910607650483e-05,
-      "loss": 0.1257,
+      "loss": 0.6325,
       "step": 322
     },
     {
       "epoch": 0.00591304347826087,
-      "grad_norm": 0.4870532155036926,
+      "grad_norm": 5.763343334197998,
       "learning_rate": 1.7144726796328034e-05,
-      "loss": 0.0665,
+      "loss": 0.6117,
       "step": 323
     },
     {
       "epoch": 0.005931350114416476,
-      "grad_norm": 1.0243052244186401,
+      "grad_norm": 6.462403774261475,
       "learning_rate": 1.712026045990997e-05,
-      "loss": 0.0885,
+      "loss": 0.3717,
       "step": 324
     },
     {
       "epoch": 0.005949656750572082,
-      "grad_norm": 6.752919673919678,
+      "grad_norm": 2.8275930881500244,
       "learning_rate": 1.709570736536521e-05,
-      "loss": 0.505,
+      "loss": 0.1894,
       "step": 325
     },
     {
       "epoch": 0.005967963386727689,
-      "grad_norm": 0.6106039881706238,
+      "grad_norm": 7.836248874664307,
       "learning_rate": 1.7071067811865477e-05,
-      "loss": 0.0692,
+      "loss": 0.4134,
       "step": 326
     },
     {
       "epoch": 0.005986270022883295,
-      "grad_norm": 0.3898545801639557,
+      "grad_norm": 4.337584972381592,
       "learning_rate": 1.7046342099635948e-05,
-      "loss": 0.0684,
+      "loss": 0.3016,
       "step": 327
     },
     {
       "epoch": 0.006004576659038902,
-      "grad_norm": 1.2045406103134155,
+      "grad_norm": 4.750609874725342,
       "learning_rate": 1.7021530529951627e-05,
-      "loss": 0.0866,
+      "loss": 0.7295,
       "step": 328
     },
     {
       "epoch": 0.006022883295194508,
-      "grad_norm": 16.497941970825195,
+      "grad_norm": 4.368948459625244,
       "learning_rate": 1.6996633405133656e-05,
-      "loss": 0.5174,
+      "loss": 0.4981,
       "step": 329
     },
     {
       "epoch": 0.006041189931350115,
-      "grad_norm": 0.5971490740776062,
+      "grad_norm": 5.648835182189941,
       "learning_rate": 1.697165102854565e-05,
-      "loss": 0.0804,
+      "loss": 0.3509,
       "step": 330
     },
     {
       "epoch": 0.0060594965675057205,
-      "grad_norm": 4.933838367462158,
+      "grad_norm": 6.33278226852417,
       "learning_rate": 1.6946583704589973e-05,
-      "loss": 0.2847,
+      "loss": 0.553,
       "step": 331
     },
     {
       "epoch": 0.006077803203661327,
-      "grad_norm": 2.9864003658294678,
+      "grad_norm": 3.5381555557250977,
       "learning_rate": 1.692143173870407e-05,
-      "loss": 0.1657,
+      "loss": 0.3945,
       "step": 332
     },
     {
       "epoch": 0.006096109839816934,
-      "grad_norm": 9.935009002685547,
+      "grad_norm": 6.314048767089844,
       "learning_rate": 1.68961954373567e-05,
-      "loss": 0.3233,
+      "loss": 0.4074,
       "step": 333
     },
     {
       "epoch": 0.00611441647597254,
-      "grad_norm": 2.5702402591705322,
+      "grad_norm": 10.963640213012695,
       "learning_rate": 1.6870875108044233e-05,
-      "loss": 0.127,
+      "loss": 0.4685,
       "step": 334
     },
     {
       "epoch": 0.006132723112128147,
-      "grad_norm": 0.6971304416656494,
+      "grad_norm": 4.838613510131836,
       "learning_rate": 1.684547105928689e-05,
-      "loss": 0.0849,
+      "loss": 0.3112,
       "step": 335
     },
     {
       "epoch": 0.006151029748283753,
-      "grad_norm": 1.7663006782531738,
+      "grad_norm": 6.644429683685303,
       "learning_rate": 1.6819983600624986e-05,
-      "loss": 0.0945,
+      "loss": 0.4721,
       "step": 336
     },
     {
       "epoch": 0.006169336384439359,
-      "grad_norm": 1.617133378982544,
+      "grad_norm": 7.574588775634766,
       "learning_rate": 1.6794413042615168e-05,
-      "loss": 0.1298,
+      "loss": 0.4311,
       "step": 337
     },
     {
       "epoch": 0.0061876430205949655,
-      "grad_norm": 1.397450566291809,
+      "grad_norm": 16.03948402404785,
       "learning_rate": 1.6768759696826608e-05,
-      "loss": 0.1045,
+      "loss": 1.1899,
       "step": 338
     },
     {
       "epoch": 0.006205949656750572,
-      "grad_norm": 3.111521005630493,
+      "grad_norm": 8.45483112335205,
       "learning_rate": 1.6743023875837233e-05,
-      "loss": 0.1684,
+      "loss": 0.4542,
       "step": 339
     },
     {
       "epoch": 0.0062242562929061785,
-      "grad_norm": 0.8679645657539368,
+      "grad_norm": 5.345456600189209,
       "learning_rate": 1.6717205893229904e-05,
-      "loss": 0.0745,
+      "loss": 0.514,
       "step": 340
     },
     {
       "epoch": 0.006242562929061785,
-      "grad_norm": 0.3660913109779358,
+      "grad_norm": 6.069270610809326,
       "learning_rate": 1.6691306063588583e-05,
-      "loss": 0.0606,
+      "loss": 0.675,
       "step": 341
     },
     {
       "epoch": 0.006260869565217392,
-      "grad_norm": 0.6691651344299316,
+      "grad_norm": 8.289877891540527,
       "learning_rate": 1.6665324702494524e-05,
-      "loss": 0.0809,
+      "loss": 0.4616,
       "step": 342
     },
     {
       "epoch": 0.006279176201372997,
-      "grad_norm": 0.5350494980812073,
+      "grad_norm": 8.072705268859863,
       "learning_rate": 1.6639262126522417e-05,
-      "loss": 0.0678,
+      "loss": 0.4349,
       "step": 343
     },
     {
       "epoch": 0.006297482837528604,
-      "grad_norm": 0.8005491495132446,
+      "grad_norm": 8.641850471496582,
       "learning_rate": 1.661311865323652e-05,
-      "loss": 0.0683,
+      "loss": 0.4676,
       "step": 344
     },
     {
       "epoch": 0.00631578947368421,
-      "grad_norm": 0.46259623765945435,
+      "grad_norm": 4.457225799560547,
       "learning_rate": 1.6586894601186804e-05,
-      "loss": 0.0662,
+      "loss": 0.4408,
       "step": 345
     },
     {
       "epoch": 0.006334096109839817,
-      "grad_norm": 0.4334816038608551,
+      "grad_norm": 4.834903240203857,
       "learning_rate": 1.6560590289905074e-05,
-      "loss": 0.0582,
+      "loss": 0.3358,
       "step": 346
     },
     {
       "epoch": 0.0063524027459954235,
-      "grad_norm": 0.5770576000213623,
+      "grad_norm": 11.663195610046387,
       "learning_rate": 1.6534206039901057e-05,
-      "loss": 0.0673,
+      "loss": 0.6072,
       "step": 347
     },
     {
       "epoch": 0.00637070938215103,
-      "grad_norm": 0.5860402584075928,
+      "grad_norm": 3.625476360321045,
       "learning_rate": 1.650774217265851e-05,
-      "loss": 0.0783,
+      "loss": 0.2633,
       "step": 348
     },
     {
       "epoch": 0.0063890160183066366,
-      "grad_norm": 0.8329232335090637,
+      "grad_norm": 6.559308052062988,
       "learning_rate": 1.6481199010631312e-05,
-      "loss": 0.0668,
+      "loss": 0.2867,
       "step": 349
     },
     {
       "epoch": 0.006407322654462242,
-      "grad_norm": 1.0413185358047485,
+      "grad_norm": 11.27363395690918,
       "learning_rate": 1.645457687723951e-05,
-      "loss": 0.0931,
+      "loss": 0.8009,
       "step": 350
     },
     {
       "epoch": 0.006425629290617849,
-      "grad_norm": 0.36341050267219543,
+      "grad_norm": 6.976966381072998,
       "learning_rate": 1.6427876096865394e-05,
-      "loss": 0.045,
+      "loss": 0.5221,
       "step": 351
     },
     {
       "epoch": 0.006443935926773455,
-      "grad_norm": 0.30069461464881897,
+      "grad_norm": 5.93279504776001,
       "learning_rate": 1.6401096994849558e-05,
-      "loss": 0.0589,
+      "loss": 0.3511,
       "step": 352
     },
     {
       "epoch": 0.006462242562929062,
-      "grad_norm": 0.7618889808654785,
+      "grad_norm": 5.895767688751221,
       "learning_rate": 1.63742398974869e-05,
-      "loss": 0.067,
+      "loss": 0.3957,
       "step": 353
     },
     {
       "epoch": 0.006480549199084668,
-      "grad_norm": 0.44705328345298767,
+      "grad_norm": 4.896230220794678,
       "learning_rate": 1.6347305132022677e-05,
-      "loss": 0.0638,
+      "loss": 0.4095,
       "step": 354
     },
     {
       "epoch": 0.006498855835240275,
-      "grad_norm": 1.4741106033325195,
+      "grad_norm": 4.811487674713135,
       "learning_rate": 1.632029302664851e-05,
-      "loss": 0.0735,
+      "loss": 0.3359,
       "step": 355
     },
     {
       "epoch": 0.006517162471395881,
-      "grad_norm": 36.54021453857422,
+      "grad_norm": 6.440311908721924,
       "learning_rate": 1.6293203910498375e-05,
-      "loss": 0.49,
+      "loss": 0.4722,
       "step": 356
     },
     {
       "epoch": 0.006535469107551487,
-      "grad_norm": 0.5257440209388733,
+      "grad_norm": 4.4008965492248535,
       "learning_rate": 1.6266038113644605e-05,
-      "loss": 0.0799,
+      "loss": 0.325,
       "step": 357
     },
     {
       "epoch": 0.006553775743707094,
-      "grad_norm": 0.5939889550209045,
+      "grad_norm": 5.9258713722229,
       "learning_rate": 1.6238795967093865e-05,
-      "loss": 0.0752,
+      "loss": 0.4317,
       "step": 358
     },
     {
       "epoch": 0.0065720823798627,
-      "grad_norm": 0.43999776244163513,
+      "grad_norm": 9.05823040008545,
       "learning_rate": 1.6211477802783105e-05,
-      "loss": 0.0571,
+      "loss": 0.6237,
       "step": 359
     },
     {
       "epoch": 0.006590389016018307,
-      "grad_norm": 0.3336666524410248,
+      "grad_norm": 12.546273231506348,
       "learning_rate": 1.6184083953575543e-05,
-      "loss": 0.0571,
+      "loss": 0.4178,
       "step": 360
     },
     {
       "epoch": 0.006608695652173913,
-      "grad_norm": 0.5457674860954285,
+      "grad_norm": 12.221634864807129,
       "learning_rate": 1.6156614753256583e-05,
-      "loss": 0.0758,
+      "loss": 0.6527,
       "step": 361
     },
     {
       "epoch": 0.00662700228832952,
-      "grad_norm": 0.4836837351322174,
+      "grad_norm": 8.335714340209961,
       "learning_rate": 1.6129070536529767e-05,
-      "loss": 0.0764,
+      "loss": 0.5695,
       "step": 362
     },
     {
       "epoch": 0.0066453089244851255,
-      "grad_norm": 0.3922410011291504,
+      "grad_norm": 9.451141357421875,
       "learning_rate": 1.610145163901268e-05,
-      "loss": 0.0591,
+      "loss": 0.4343,
       "step": 363
     },
     {
       "epoch": 0.006663615560640732,
-      "grad_norm": 0.7826827764511108,
+      "grad_norm": 6.33415412902832,
       "learning_rate": 1.607375839723287e-05,
-      "loss": 0.0814,
+      "loss": 0.3475,
       "step": 364
     },
     {
       "epoch": 0.006681922196796339,
-      "grad_norm": 0.6293019652366638,
+      "grad_norm": 8.798182487487793,
       "learning_rate": 1.6045991148623752e-05,
-      "loss": 0.0843,
+      "loss": 0.4564,
       "step": 365
     },
     {
       "epoch": 0.006700228832951945,
-      "grad_norm": 0.46461156010627747,
+      "grad_norm": 7.0115647315979,
       "learning_rate": 1.6018150231520486e-05,
-      "loss": 0.0617,
+      "loss": 0.4848,
       "step": 366
     },
     {
       "epoch": 0.006718535469107552,
-      "grad_norm": 0.35606852173805237,
+      "grad_norm": 11.819734573364258,
       "learning_rate": 1.599023598515586e-05,
-      "loss": 0.0506,
+      "loss": 0.9038,
       "step": 367
     },
     {
       "epoch": 0.006736842105263158,
-      "grad_norm": 5.784524440765381,
+      "grad_norm": 7.8828325271606445,
       "learning_rate": 1.5962248749656158e-05,
-      "loss": 0.5141,
+      "loss": 0.252,
       "step": 368
     },
     {
       "epoch": 0.006755148741418764,
-      "grad_norm": 4.421563148498535,
+      "grad_norm": 1.1683067083358765,
       "learning_rate": 1.5934188866037017e-05,
-      "loss": 0.2153,
+      "loss": 0.0875,
       "step": 369
     },
     {
       "epoch": 0.0067734553775743705,
-      "grad_norm": 0.5804718136787415,
+      "grad_norm": 4.919974327087402,
       "learning_rate": 1.5906056676199256e-05,
-      "loss": 0.0503,
+      "loss": 0.1626,
       "step": 370
     },
     {
       "epoch": 0.006791762013729977,
-      "grad_norm": 0.9003938436508179,
+      "grad_norm": 5.247345447540283,
       "learning_rate": 1.5877852522924733e-05,
-      "loss": 0.0698,
+      "loss": 0.2482,
       "step": 371
     },
     {
       "epoch": 0.006810068649885584,
-      "grad_norm": 0.9294069409370422,
+      "grad_norm": 2.816721200942993,
       "learning_rate": 1.584957674987216e-05,
-      "loss": 0.0888,
+      "loss": 0.1829,
       "step": 372
     },
     {
       "epoch": 0.00682837528604119,
-      "grad_norm": 1.867724895477295,
+      "grad_norm": 15.067553520202637,
       "learning_rate": 1.5821229701572897e-05,
-      "loss": 0.0894,
+      "loss": 0.9891,
       "step": 373
     },
     {
       "epoch": 0.006846681922196797,
-      "grad_norm": 1.3612340688705444,
+      "grad_norm": 12.05018424987793,
       "learning_rate": 1.5792811723426787e-05,
-      "loss": 0.0908,
+      "loss": 0.5695,
       "step": 374
     },
     {
       "epoch": 0.006864988558352402,
-      "grad_norm": 0.744966447353363,
+      "grad_norm": 7.035960674285889,
       "learning_rate": 1.5764323161697933e-05,
-      "loss": 0.0712,
+      "loss": 0.362,
       "step": 375
     },
     {
       "epoch": 0.006883295194508009,
-      "grad_norm": 4.086651802062988,
+      "grad_norm": 10.163368225097656,
       "learning_rate": 1.573576436351046e-05,
-      "loss": 0.1219,
+      "loss": 0.5263,
       "step": 376
     },
     {
       "epoch": 0.006901601830663615,
-      "grad_norm": 1.0539871454238892,
+      "grad_norm": 8.28557014465332,
       "learning_rate": 1.570713567684432e-05,
-      "loss": 0.0832,
+      "loss": 0.4129,
       "step": 377
     },
     {
       "epoch": 0.006919908466819222,
-      "grad_norm": 0.9177609086036682,
+      "grad_norm": 12.570793151855469,
       "learning_rate": 1.5678437450531014e-05,
-      "loss": 0.0768,
+      "loss": 0.7611,
       "step": 378
     },
     {
       "epoch": 0.0069382151029748285,
-      "grad_norm": 0.2821899950504303,
+      "grad_norm": 6.368137836456299,
       "learning_rate": 1.564967003424938e-05,
-      "loss": 0.0521,
+      "loss": 0.5362,
       "step": 379
     },
     {
       "epoch": 0.006956521739130435,
-      "grad_norm": 1.0433887243270874,
+      "grad_norm": 3.3329522609710693,
       "learning_rate": 1.5620833778521306e-05,
-      "loss": 0.0759,
+      "loss": 0.3633,
       "step": 380
     },
     {
       "epoch": 0.006974828375286042,
-      "grad_norm": 0.9624257683753967,
+      "grad_norm": 5.921566009521484,
       "learning_rate": 1.5591929034707468e-05,
-      "loss": 0.0857,
+      "loss": 0.45,
       "step": 381
     },
     {
       "epoch": 0.006993135011441647,
-      "grad_norm": 5.068337917327881,
+      "grad_norm": 4.826172828674316,
       "learning_rate": 1.556295615500305e-05,
-      "loss": 0.1328,
+      "loss": 0.3278,
       "step": 382
     },
     {
       "epoch": 0.007011441647597254,
-      "grad_norm": 0.38359129428863525,
+      "grad_norm": 6.323006629943848,
       "learning_rate": 1.553391549243344e-05,
-      "loss": 0.055,
+      "loss": 0.3469,
       "step": 383
     },
     {
       "epoch": 0.00702974828375286,
-      "grad_norm": 1.4515821933746338,
+      "grad_norm": 5.0740203857421875,
       "learning_rate": 1.5504807400849957e-05,
-      "loss": 0.0887,
+      "loss": 0.3713,
       "step": 384
     },
     {
       "epoch": 0.007048054919908467,
-      "grad_norm": 0.8682299852371216,
+      "grad_norm": 4.266338348388672,
       "learning_rate": 1.5475632234925505e-05,
-      "loss": 0.0895,
+      "loss": 0.2165,
       "step": 385
     },
     {
       "epoch": 0.007066361556064073,
-      "grad_norm": 1.6979293823242188,
+      "grad_norm": 4.6646504402160645,
       "learning_rate": 1.5446390350150272e-05,
-      "loss": 0.1248,
+      "loss": 0.4328,
       "step": 386
     },
     {
       "epoch": 0.00708466819221968,
-      "grad_norm": 0.48389941453933716,
+      "grad_norm": 4.261294841766357,
       "learning_rate": 1.54170821028274e-05,
-      "loss": 0.0715,
+      "loss": 0.4952,
       "step": 387
     },
     {
       "epoch": 0.007102974828375286,
-      "grad_norm": 0.45502373576164246,
+      "grad_norm": 4.2962846755981445,
       "learning_rate": 1.5387707850068633e-05,
-      "loss": 0.0521,
+      "loss": 0.3523,
       "step": 388
     },
     {
       "epoch": 0.007121281464530892,
-      "grad_norm": 1.2198498249053955,
+      "grad_norm": 3.559885263442993,
       "learning_rate": 1.5358267949789968e-05,
-      "loss": 0.0835,
+      "loss": 0.1836,
       "step": 389
     },
     {
       "epoch": 0.007139588100686499,
-      "grad_norm": 0.34176769852638245,
+      "grad_norm": 4.17678689956665,
       "learning_rate": 1.53287627607073e-05,
-      "loss": 0.0571,
+      "loss": 0.1978,
       "step": 390
     },
     {
       "epoch": 0.007157894736842105,
-      "grad_norm": 1.1584203243255615,
+      "grad_norm": 9.370595932006836,
       "learning_rate": 1.529919264233205e-05,
-      "loss": 0.0787,
+      "loss": 0.5643,
       "step": 391
     },
     {
       "epoch": 0.007176201372997712,
-      "grad_norm": 0.8584113121032715,
+      "grad_norm": 4.1542463302612305,
       "learning_rate": 1.5269557954966777e-05,
-      "loss": 0.0708,
+      "loss": 0.3046,
       "step": 392
     },
     {
       "epoch": 0.007194508009153318,
-      "grad_norm": 0.5073244571685791,
+      "grad_norm": 7.581055164337158,
       "learning_rate": 1.5239859059700794e-05,
-      "loss": 0.071,
+      "loss": 0.4282,
       "step": 393
     },
     {
       "epoch": 0.007212814645308925,
-      "grad_norm": 15.320381164550781,
+      "grad_norm": 2.4097397327423096,
       "learning_rate": 1.5210096318405768e-05,
-      "loss": 0.4549,
+      "loss": 0.154,
       "step": 394
     },
     {
       "epoch": 0.007231121281464531,
-      "grad_norm": 0.4920025169849396,
+      "grad_norm": 3.8569397926330566,
       "learning_rate": 1.5180270093731305e-05,
-      "loss": 0.0681,
+      "loss": 0.0843,
       "step": 395
     },
     {
       "epoch": 0.007249427917620137,
-      "grad_norm": 0.37975895404815674,
+      "grad_norm": 7.704616069793701,
       "learning_rate": 1.5150380749100545e-05,
-      "loss": 0.0685,
+      "loss": 0.5627,
       "step": 396
     },
     {
       "epoch": 0.007267734553775744,
-      "grad_norm": 0.4086936414241791,
+      "grad_norm": 4.946331024169922,
       "learning_rate": 1.5120428648705716e-05,
-      "loss": 0.0495,
+      "loss": 0.2496,
       "step": 397
     },
     {
       "epoch": 0.00728604118993135,
-      "grad_norm": 0.3639077842235565,
+      "grad_norm": 3.8404288291931152,
       "learning_rate": 1.5090414157503715e-05,
-      "loss": 0.0541,
+      "loss": 0.1916,
       "step": 398
     },
     {
       "epoch": 0.007304347826086957,
-      "grad_norm": 0.7182059288024902,
+      "grad_norm": 1.226649522781372,
       "learning_rate": 1.5060337641211637e-05,
-      "loss": 0.0636,
+      "loss": 0.1359,
       "step": 399
     },
     {
       "epoch": 0.007322654462242563,
-      "grad_norm": 0.6276530623435974,
+      "grad_norm": 10.208996772766113,
       "learning_rate": 1.5030199466302354e-05,
-      "loss": 0.0636,
+      "loss": 0.3555,
       "step": 400
     },
     {
       "epoch": 0.007340961098398169,
-      "grad_norm": 1.9585328102111816,
+      "grad_norm": 7.164267539978027,
       "learning_rate": 1.5000000000000002e-05,
-      "loss": 0.1138,
+      "loss": 0.4576,
       "step": 401
     },
     {
       "epoch": 0.0073592677345537755,
-      "grad_norm": 0.5767062902450562,
+      "grad_norm": 6.956635475158691,
       "learning_rate": 1.4969739610275556e-05,
-      "loss": 0.0876,
+      "loss": 0.3394,
       "step": 402
     },
     {
       "epoch": 0.007377574370709382,
-      "grad_norm": 0.5097327828407288,
+      "grad_norm": 11.293649673461914,
       "learning_rate": 1.493941866584231e-05,
-      "loss": 0.0497,
+      "loss": 0.6647,
       "step": 403
     },
     {
       "epoch": 0.007395881006864989,
-      "grad_norm": 1.5687953233718872,
+      "grad_norm": 9.281755447387695,
       "learning_rate": 1.490903753615141e-05,
-      "loss": 0.1114,
+      "loss": 0.4833,
       "step": 404
     },
     {
       "epoch": 0.007414187643020595,
-      "grad_norm": 0.510195255279541,
+      "grad_norm": 3.4506609439849854,
       "learning_rate": 1.4878596591387329e-05,
-      "loss": 0.0685,
+      "loss": 0.239,
       "step": 405
     },
     {
       "epoch": 0.007432494279176202,
-      "grad_norm": 0.5343835353851318,
+      "grad_norm": 1.7865498065948486,
       "learning_rate": 1.4848096202463373e-05,
-      "loss": 0.0652,
+      "loss": 0.135,
       "step": 406
     },
     {
       "epoch": 0.007450800915331807,
-      "grad_norm": 0.7216667532920837,
+      "grad_norm": 2.3500795364379883,
       "learning_rate": 1.4817536741017153e-05,
-      "loss": 0.0973,
+      "loss": 0.2015,
       "step": 407
     },
     {
       "epoch": 0.007469107551487414,
-      "grad_norm": 0.9994480609893799,
+      "grad_norm": 8.29547119140625,
       "learning_rate": 1.478691857940607e-05,
-      "loss": 0.1074,
+      "loss": 0.4149,
       "step": 408
     },
     {
       "epoch": 0.0074874141876430204,
-      "grad_norm": 0.6079360842704773,
+      "grad_norm": 2.8842177391052246,
       "learning_rate": 1.4756242090702756e-05,
-      "loss": 0.0601,
+      "loss": 0.1729,
       "step": 409
     },
     {
       "epoch": 0.007505720823798627,
-      "grad_norm": 0.6980652213096619,
+      "grad_norm": 8.263029098510742,
       "learning_rate": 1.4725507648690542e-05,
-      "loss": 0.0749,
+      "loss": 0.5903,
       "step": 410
     },
     {
       "epoch": 0.0075240274599542335,
-      "grad_norm": 17.103120803833008,
+      "grad_norm": 4.133641719818115,
       "learning_rate": 1.469471562785891e-05,
-      "loss": 0.9587,
+      "loss": 0.3041,
       "step": 411
     },
     {
       "epoch": 0.00754233409610984,
-      "grad_norm": 0.425077885389328,
+      "grad_norm": 9.040873527526855,
       "learning_rate": 1.4663866403398915e-05,
-      "loss": 0.0642,
+      "loss": 0.6155,
       "step": 412
     },
     {
       "epoch": 0.007560640732265447,
-      "grad_norm": 0.3487469553947449,
+      "grad_norm": 5.259375095367432,
       "learning_rate": 1.463296035119862e-05,
-      "loss": 0.0668,
+      "loss": 0.357,
       "step": 413
     },
     {
       "epoch": 0.007578947368421052,
-      "grad_norm": 0.3773779571056366,
+      "grad_norm": 5.846144676208496,
       "learning_rate": 1.4601997847838518e-05,
-      "loss": 0.0521,
+      "loss": 0.2454,
       "step": 414
     },
     {
       "epoch": 0.007597254004576659,
-      "grad_norm": 0.37802889943122864,
+      "grad_norm": 5.604888439178467,
       "learning_rate": 1.4570979270586944e-05,
-      "loss": 0.0459,
+      "loss": 0.4757,
       "step": 415
     },
     {
       "epoch": 0.007615560640732265,
-      "grad_norm": 4.217767715454102,
+      "grad_norm": 6.676849365234375,
       "learning_rate": 1.4539904997395468e-05,
-      "loss": 0.0959,
+      "loss": 0.2546,
       "step": 416
     },
     {
       "epoch": 0.007633867276887872,
-      "grad_norm": 0.541473388671875,
+      "grad_norm": 7.3254523277282715,
       "learning_rate": 1.4508775406894308e-05,
-      "loss": 0.0765,
+      "loss": 0.4365,
       "step": 417
     },
     {
       "epoch": 0.0076521739130434785,
-      "grad_norm": 0.7299866080284119,
+      "grad_norm": 7.06563663482666,
       "learning_rate": 1.4477590878387697e-05,
-      "loss": 0.1009,
+      "loss": 0.3348,
       "step": 418
     },
     {
       "epoch": 0.007670480549199085,
-      "grad_norm": 2.0802814960479736,
+      "grad_norm": 4.191524982452393,
       "learning_rate": 1.4446351791849276e-05,
-      "loss": 0.1292,
+      "loss": 0.2775,
       "step": 419
     },
     {
       "epoch": 0.007688787185354691,
-      "grad_norm": 0.5377467274665833,
+      "grad_norm": 6.883635520935059,
       "learning_rate": 1.4415058527917454e-05,
-      "loss": 0.0708,
+      "loss": 0.336,
       "step": 420
     },
     {
       "epoch": 0.007707093821510297,
-      "grad_norm": 0.37011802196502686,
+      "grad_norm": 2.3200838565826416,
       "learning_rate": 1.4383711467890776e-05,
-      "loss": 0.0606,
+      "loss": 0.1396,
       "step": 421
     },
     {
       "epoch": 0.007725400457665904,
-      "grad_norm": 0.5028330683708191,
+      "grad_norm": 4.974113464355469,
       "learning_rate": 1.4352310993723277e-05,
-      "loss": 0.0686,
+      "loss": 0.2149,
       "step": 422
     },
     {
       "epoch": 0.00774370709382151,
-      "grad_norm": 0.4025152623653412,
+      "grad_norm": 7.065567970275879,
       "learning_rate": 1.4320857488019826e-05,
-      "loss": 0.0527,
+      "loss": 0.3439,
       "step": 423
     },
     {
       "epoch": 0.007762013729977117,
-      "grad_norm": 0.746654748916626,
+      "grad_norm": 6.593064308166504,
       "learning_rate": 1.4289351334031461e-05,
-      "loss": 0.0543,
+      "loss": 0.327,
       "step": 424
     },
     {
       "epoch": 0.007780320366132723,
-      "grad_norm": 0.46465712785720825,
+      "grad_norm": 4.018970012664795,
       "learning_rate": 1.4257792915650728e-05,
-      "loss": 0.0622,
+      "loss": 0.2449,
       "step": 425
     },
     {
       "epoch": 0.00779862700228833,
-      "grad_norm": 1.0541694164276123,
+      "grad_norm": 1.6839569807052612,
       "learning_rate": 1.4226182617406996e-05,
-      "loss": 0.0699,
+      "loss": 0.1382,
       "step": 426
     },
     {
       "epoch": 0.007816933638443936,
-      "grad_norm": 2.2482950687408447,
+      "grad_norm": 3.03248929977417,
       "learning_rate": 1.4194520824461773e-05,
-      "loss": 0.1408,
+      "loss": 0.1527,
       "step": 427
     },
     {
       "epoch": 0.007835240274599543,
-      "grad_norm": 0.4945947825908661,
+      "grad_norm": 10.530253410339355,
       "learning_rate": 1.4162807922604014e-05,
-      "loss": 0.0689,
+      "loss": 0.4916,
       "step": 428
     },
     {
       "epoch": 0.00785354691075515,
-      "grad_norm": 0.45765721797943115,
+      "grad_norm": 9.175287246704102,
       "learning_rate": 1.413104429824542e-05,
-      "loss": 0.0596,
+      "loss": 0.5146,
       "step": 429
     },
     {
       "epoch": 0.007871853546910754,
-      "grad_norm": 2.1746416091918945,
+      "grad_norm": 9.110077857971191,
       "learning_rate": 1.4099230338415728e-05,
-      "loss": 0.1263,
+      "loss": 0.2713,
       "step": 430
     },
     {
       "epoch": 0.007890160183066361,
-      "grad_norm": 0.41592302918434143,
+      "grad_norm": 6.078294277191162,
       "learning_rate": 1.4067366430758004e-05,
-      "loss": 0.0491,
+      "loss": 0.297,
       "step": 431
     },
     {
       "epoch": 0.007908466819221967,
-      "grad_norm": 0.36904650926589966,
+      "grad_norm": 9.00320816040039,
       "learning_rate": 1.4035452963523903e-05,
-      "loss": 0.0375,
+      "loss": 0.219,
       "step": 432
     },
     {
       "epoch": 0.007926773455377574,
-      "grad_norm": 0.49035295844078064,
+      "grad_norm": 11.155050277709961,
       "learning_rate": 1.4003490325568953e-05,
-      "loss": 0.0725,
+      "loss": 0.5544,
       "step": 433
     },
     {
       "epoch": 0.00794508009153318,
-      "grad_norm": 0.7628765106201172,
+      "grad_norm": 9.077802658081055,
       "learning_rate": 1.3971478906347806e-05,
-      "loss": 0.066,
+      "loss": 0.3477,
       "step": 434
     },
     {
       "epoch": 0.007963386727688787,
-      "grad_norm": 0.701587975025177,
+      "grad_norm": 5.8080668449401855,
       "learning_rate": 1.3939419095909513e-05,
-      "loss": 0.0642,
+      "loss": 0.275,
       "step": 435
     },
     {
       "epoch": 0.007981693363844394,
-      "grad_norm": 0.4282008707523346,
+      "grad_norm": 19.085351943969727,
       "learning_rate": 1.3907311284892737e-05,
-      "loss": 0.0566,
+      "loss": 0.3296,
       "step": 436
     },
     {
       "epoch": 0.008,
-      "grad_norm": 0.44497933983802795,
+      "grad_norm": 12.508645057678223,
       "learning_rate": 1.3875155864521031e-05,
-      "loss": 0.0673,
+      "loss": 0.5693,
       "step": 437
     },
     {
       "epoch": 0.008018306636155607,
-      "grad_norm": 0.3473840355873108,
+      "grad_norm": 5.50254487991333,
       "learning_rate": 1.3842953226598036e-05,
-      "loss": 0.0412,
+      "loss": 0.1062,
       "step": 438
     },
     {
       "epoch": 0.008036613272311213,
-      "grad_norm": 0.6821406483650208,
+      "grad_norm": 14.789398193359375,
       "learning_rate": 1.3810703763502744e-05,
-      "loss": 0.0567,
+      "loss": 0.9844,
       "step": 439
     },
     {
       "epoch": 0.00805491990846682,
-      "grad_norm": 1.5390894412994385,
+      "grad_norm": 19.38977813720703,
       "learning_rate": 1.3778407868184674e-05,
-      "loss": 0.08,
+      "loss": 0.6566,
       "step": 440
     },
     {
       "epoch": 0.008073226544622426,
-      "grad_norm": 0.5233414173126221,
+      "grad_norm": 13.897294998168945,
       "learning_rate": 1.3746065934159123e-05,
-      "loss": 0.0637,
+      "loss": 0.7898,
       "step": 441
     },
     {
       "epoch": 0.008091533180778033,
-      "grad_norm": 0.39452603459358215,
+      "grad_norm": 9.941490173339844,
       "learning_rate": 1.371367835550235e-05,
-      "loss": 0.0517,
+      "loss": 0.4879,
       "step": 442
     },
     {
       "epoch": 0.008109839816933638,
-      "grad_norm": 15.558564186096191,
+      "grad_norm": 2.901772975921631,
       "learning_rate": 1.3681245526846782e-05,
-      "loss": 0.5675,
+      "loss": 0.2148,
       "step": 443
     },
     {
       "epoch": 0.008128146453089244,
-      "grad_norm": 0.5562123656272888,
+      "grad_norm": 4.679471492767334,
       "learning_rate": 1.3648767843376196e-05,
-      "loss": 0.0491,
+      "loss": 0.2317,
       "step": 444
     },
     {
       "epoch": 0.00814645308924485,
-      "grad_norm": 0.6518932580947876,
+      "grad_norm": 3.4032607078552246,
       "learning_rate": 1.3616245700820922e-05,
-      "loss": 0.0544,
+      "loss": 0.2151,
       "step": 445
     },
     {
       "epoch": 0.008164759725400457,
-      "grad_norm": 0.426328182220459,
+      "grad_norm": 2.5919456481933594,
       "learning_rate": 1.3583679495453e-05,
-      "loss": 0.0607,
+      "loss": 0.1284,
       "step": 446
     },
     {
       "epoch": 0.008183066361556064,
-      "grad_norm": 0.8472924828529358,
+      "grad_norm": 6.18490743637085,
       "learning_rate": 1.3551069624081372e-05,
-      "loss": 0.0531,
+      "loss": 0.3029,
       "step": 447
     },
     {
       "epoch": 0.00820137299771167,
-      "grad_norm": 0.6323686838150024,
+      "grad_norm": 10.599230766296387,
       "learning_rate": 1.3518416484047018e-05,
-      "loss": 0.0754,
+      "loss": 0.5019,
       "step": 448
     },
     {
       "epoch": 0.008219679633867277,
-      "grad_norm": 0.48989608883857727,
+      "grad_norm": 8.500578880310059,
       "learning_rate": 1.3485720473218153e-05,
-      "loss": 0.069,
+      "loss": 0.4647,
       "step": 449
     },
     {
       "epoch": 0.008237986270022883,
-      "grad_norm": 0.4766976237297058,
+      "grad_norm": 2.737617254257202,
       "learning_rate": 1.3452981989985347e-05,
-      "loss": 0.0655,
+      "loss": 0.1593,
       "step": 450
-    },
-    {
-      "epoch": 0.00825629290617849,
-      "grad_norm": 0.1481637805700302,
-      "learning_rate": 1.342020143325669e-05,
-      "loss": 0.0302,
-      "step": 451
-    },
-    {
-      "epoch": 0.008274599542334097,
-      "grad_norm": 10.94448184967041,
-      "learning_rate": 1.3387379202452917e-05,
-      "loss": 0.6379,
-      "step": 452
-    },
-    {
-      "epoch": 0.008292906178489703,
-      "grad_norm": 0.31550925970077515,
-      "learning_rate": 1.3354515697502552e-05,
-      "loss": 0.0537,
-      "step": 453
-    },
-    {
-      "epoch": 0.00831121281464531,
-      "grad_norm": 8.416996002197266,
-      "learning_rate": 1.3321611318837033e-05,
-      "loss": 0.2144,
-      "step": 454
-    },
-    {
-      "epoch": 0.008329519450800914,
-      "grad_norm": 1.015204668045044,
-      "learning_rate": 1.3288666467385834e-05,
-      "loss": 0.0648,
-      "step": 455
-    },
-    {
-      "epoch": 0.008347826086956521,
-      "grad_norm": 0.900848388671875,
-      "learning_rate": 1.3255681544571568e-05,
-      "loss": 0.0713,
-      "step": 456
-    },
-    {
-      "epoch": 0.008366132723112128,
-      "grad_norm": 0.567472517490387,
-      "learning_rate": 1.3222656952305113e-05,
-      "loss": 0.0579,
-      "step": 457
-    },
-    {
-      "epoch": 0.008384439359267734,
-      "grad_norm": 0.8376880288124084,
-      "learning_rate": 1.3189593092980701e-05,
-      "loss": 0.0761,
-      "step": 458
-    },
-    {
-      "epoch": 0.00840274599542334,
-      "grad_norm": 0.780007541179657,
-      "learning_rate": 1.3156490369471026e-05,
-      "loss": 0.0549,
-      "step": 459
-    },
-    {
-      "epoch": 0.008421052631578947,
-      "grad_norm": 1.0251318216323853,
-      "learning_rate": 1.3123349185122328e-05,
-      "loss": 0.0991,
-      "step": 460
-    },
-    {
-      "epoch": 0.008439359267734554,
-      "grad_norm": 0.8971799612045288,
-      "learning_rate": 1.3090169943749475e-05,
-      "loss": 0.0664,
-      "step": 461
-    },
-    {
-      "epoch": 0.00845766590389016,
-      "grad_norm": 1.0607576370239258,
-      "learning_rate": 1.3056953049631059e-05,
-      "loss": 0.0738,
-      "step": 462
-    },
-    {
-      "epoch": 0.008475972540045767,
-      "grad_norm": 6.68612003326416,
-      "learning_rate": 1.3023698907504447e-05,
-      "loss": 0.2481,
-      "step": 463
-    },
-    {
-      "epoch": 0.008494279176201373,
-      "grad_norm": 1.331336498260498,
-      "learning_rate": 1.2990407922560869e-05,
-      "loss": 0.0785,
-      "step": 464
-    },
-    {
-      "epoch": 0.00851258581235698,
-      "grad_norm": 0.6772834062576294,
-      "learning_rate": 1.2957080500440469e-05,
-      "loss": 0.0536,
-      "step": 465
-    },
-    {
-      "epoch": 0.008530892448512586,
-      "grad_norm": 1.3969557285308838,
-      "learning_rate": 1.2923717047227368e-05,
-      "loss": 0.1515,
-      "step": 466
-    },
-    {
-      "epoch": 0.008549199084668193,
-      "grad_norm": 0.30725976824760437,
-      "learning_rate": 1.2890317969444716e-05,
-      "loss": 0.0387,
-      "step": 467
-    },
-    {
-      "epoch": 0.008567505720823798,
-      "grad_norm": 2.5167860984802246,
-      "learning_rate": 1.2856883674049736e-05,
-      "loss": 0.1035,
-      "step": 468
-    },
-    {
-      "epoch": 0.008585812356979404,
-      "grad_norm": 0.8405506014823914,
-      "learning_rate": 1.2823414568428767e-05,
-      "loss": 0.059,
-      "step": 469
-    },
-    {
-      "epoch": 0.008604118993135011,
-      "grad_norm": 6.711755275726318,
-      "learning_rate": 1.2789911060392295e-05,
-      "loss": 0.6134,
-      "step": 470
-    },
-    {
-      "epoch": 0.008622425629290617,
-      "grad_norm": 6.742506504058838,
-      "learning_rate": 1.2756373558169992e-05,
-      "loss": 0.2486,
-      "step": 471
-    },
-    {
-      "epoch": 0.008640732265446224,
-      "grad_norm": 0.5245745182037354,
-      "learning_rate": 1.2722802470405744e-05,
-      "loss": 0.0654,
-      "step": 472
-    },
-    {
-      "epoch": 0.00865903890160183,
-      "grad_norm": 0.22681592404842377,
-      "learning_rate": 1.2689198206152657e-05,
-      "loss": 0.0429,
-      "step": 473
-    },
-    {
-      "epoch": 0.008677345537757437,
-      "grad_norm": 1.1675441265106201,
-      "learning_rate": 1.265556117486809e-05,
-      "loss": 0.0694,
-      "step": 474
-    },
-    {
-      "epoch": 0.008695652173913044,
-      "grad_norm": 0.9049202799797058,
-      "learning_rate": 1.2621891786408648e-05,
-      "loss": 0.0648,
-      "step": 475
-    },
-    {
-      "epoch": 0.00871395881006865,
-      "grad_norm": 2.439650535583496,
-      "learning_rate": 1.2588190451025209e-05,
-      "loss": 0.0709,
-      "step": 476
-    },
-    {
-      "epoch": 0.008732265446224257,
-      "grad_norm": 0.4854981005191803,
-      "learning_rate": 1.2554457579357906e-05,
-      "loss": 0.0582,
-      "step": 477
-    },
-    {
-      "epoch": 0.008750572082379863,
-      "grad_norm": 0.5455697774887085,
-      "learning_rate": 1.252069358243114e-05,
-      "loss": 0.0604,
-      "step": 478
-    },
-    {
-      "epoch": 0.00876887871853547,
-      "grad_norm": 0.2915131449699402,
-      "learning_rate": 1.2486898871648552e-05,
-      "loss": 0.0439,
-      "step": 479
-    },
-    {
-      "epoch": 0.008787185354691076,
-      "grad_norm": 0.6063660383224487,
-      "learning_rate": 1.2453073858788027e-05,
-      "loss": 0.0596,
-      "step": 480
-    },
-    {
-      "epoch": 0.008805491990846681,
-      "grad_norm": 0.7432474493980408,
-      "learning_rate": 1.2419218955996677e-05,
-      "loss": 0.0536,
-      "step": 481
-    },
-    {
-      "epoch": 0.008823798627002288,
-      "grad_norm": 0.2047077864408493,
-      "learning_rate": 1.238533457578581e-05,
-      "loss": 0.0436,
-      "step": 482
-    },
-    {
-      "epoch": 0.008842105263157894,
-      "grad_norm": 0.8452476263046265,
-      "learning_rate": 1.23514211310259e-05,
-      "loss": 0.0676,
-      "step": 483
-    },
-    {
-      "epoch": 0.0088604118993135,
-      "grad_norm": 8.1668119430542,
-      "learning_rate": 1.2317479034941572e-05,
-      "loss": 0.4087,
-      "step": 484
-    },
-    {
-      "epoch": 0.008878718535469107,
-      "grad_norm": 0.4986136853694916,
-      "learning_rate": 1.2283508701106559e-05,
-      "loss": 0.0548,
-      "step": 485
-    },
-    {
-      "epoch": 0.008897025171624714,
-      "grad_norm": 1.895058035850525,
-      "learning_rate": 1.2249510543438652e-05,
-      "loss": 0.0677,
-      "step": 486
-    },
-    {
-      "epoch": 0.00891533180778032,
-      "grad_norm": 0.5336563587188721,
-      "learning_rate": 1.2215484976194675e-05,
-      "loss": 0.0765,
-      "step": 487
-    },
-    {
-      "epoch": 0.008933638443935927,
-      "grad_norm": 2.029003620147705,
-      "learning_rate": 1.2181432413965428e-05,
-      "loss": 0.081,
-      "step": 488
-    },
-    {
-      "epoch": 0.008951945080091533,
-      "grad_norm": 1.112306833267212,
-      "learning_rate": 1.2147353271670634e-05,
-      "loss": 0.1074,
-      "step": 489
-    },
-    {
-      "epoch": 0.00897025171624714,
-      "grad_norm": 0.5803303122520447,
-      "learning_rate": 1.211324796455389e-05,
-      "loss": 0.0645,
-      "step": 490
-    },
-    {
-      "epoch": 0.008988558352402747,
-      "grad_norm": 0.5115679502487183,
-      "learning_rate": 1.2079116908177592e-05,
-      "loss": 0.0694,
-      "step": 491
-    },
-    {
-      "epoch": 0.009006864988558353,
-      "grad_norm": 0.5267685651779175,
-      "learning_rate": 1.2044960518417902e-05,
-      "loss": 0.0565,
-      "step": 492
-    },
-    {
-      "epoch": 0.00902517162471396,
-      "grad_norm": 0.3912874460220337,
-      "learning_rate": 1.2010779211459649e-05,
-      "loss": 0.0491,
-      "step": 493
-    },
-    {
-      "epoch": 0.009043478260869564,
-      "grad_norm": 4.084512233734131,
-      "learning_rate": 1.1976573403791263e-05,
-      "loss": 0.2377,
-      "step": 494
-    },
-    {
-      "epoch": 0.009061784897025171,
-      "grad_norm": 1.038677453994751,
-      "learning_rate": 1.194234351219972e-05,
-      "loss": 0.0788,
-      "step": 495
-    },
-    {
-      "epoch": 0.009080091533180778,
-      "grad_norm": 1.0220599174499512,
-      "learning_rate": 1.190808995376545e-05,
-      "loss": 0.0578,
-      "step": 496
-    },
-    {
-      "epoch": 0.009098398169336384,
-      "grad_norm": 4.992150783538818,
-      "learning_rate": 1.187381314585725e-05,
-      "loss": 0.1877,
-      "step": 497
-    },
-    {
-      "epoch": 0.00911670480549199,
-      "grad_norm": 2.0098202228546143,
-      "learning_rate": 1.1839513506127202e-05,
-      "loss": 0.0963,
-      "step": 498
-    },
-    {
-      "epoch": 0.009135011441647597,
-      "grad_norm": 0.3368827998638153,
-      "learning_rate": 1.1805191452505602e-05,
-      "loss": 0.0398,
-      "step": 499
-    },
-    {
-      "epoch": 0.009153318077803204,
-      "grad_norm": 0.6661342978477478,
-      "learning_rate": 1.1770847403195836e-05,
-      "loss": 0.0782,
-      "step": 500
-    },
-    {
-      "epoch": 0.00917162471395881,
-      "grad_norm": 0.57005774974823,
-      "learning_rate": 1.1736481776669307e-05,
-      "loss": 0.0528,
-      "step": 501
-    },
-    {
-      "epoch": 0.009189931350114417,
-      "grad_norm": 1.5390543937683105,
-      "learning_rate": 1.1702094991660326e-05,
-      "loss": 0.1156,
-      "step": 502
-    },
-    {
-      "epoch": 0.009208237986270023,
-      "grad_norm": 0.8325178027153015,
-      "learning_rate": 1.1667687467161025e-05,
-      "loss": 0.0696,
-      "step": 503
-    },
-    {
-      "epoch": 0.00922654462242563,
-      "grad_norm": 0.7934985160827637,
-      "learning_rate": 1.1633259622416224e-05,
-      "loss": 0.0659,
-      "step": 504
-    },
-    {
-      "epoch": 0.009244851258581236,
-      "grad_norm": 0.5219599604606628,
-      "learning_rate": 1.159881187691835e-05,
-      "loss": 0.0655,
-      "step": 505
-    },
-    {
-      "epoch": 0.009263157894736843,
-      "grad_norm": 0.834323525428772,
-      "learning_rate": 1.156434465040231e-05,
-      "loss": 0.0821,
-      "step": 506
-    },
-    {
-      "epoch": 0.009281464530892448,
-      "grad_norm": 1.001497507095337,
-      "learning_rate": 1.1529858362840383e-05,
-      "loss": 0.0567,
-      "step": 507
-    },
-    {
-      "epoch": 0.009299771167048054,
-      "grad_norm": 0.5971242189407349,
-      "learning_rate": 1.1495353434437098e-05,
-      "loss": 0.0572,
-      "step": 508
-    },
-    {
-      "epoch": 0.00931807780320366,
-      "grad_norm": 0.3258911371231079,
-      "learning_rate": 1.1460830285624119e-05,
-      "loss": 0.0509,
-      "step": 509
-    },
-    {
-      "epoch": 0.009336384439359267,
-      "grad_norm": 0.4368310272693634,
-      "learning_rate": 1.1426289337055119e-05,
-      "loss": 0.0527,
-      "step": 510
-    },
-    {
-      "epoch": 0.009354691075514874,
-      "grad_norm": 0.36562836170196533,
-      "learning_rate": 1.1391731009600655e-05,
-      "loss": 0.0537,
-      "step": 511
-    },
-    {
-      "epoch": 0.00937299771167048,
-      "grad_norm": 0.5770419836044312,
-      "learning_rate": 1.1357155724343046e-05,
-      "loss": 0.0617,
-      "step": 512
-    },
-    {
-      "epoch": 0.009391304347826087,
-      "grad_norm": 0.4801030158996582,
-      "learning_rate": 1.1322563902571227e-05,
-      "loss": 0.0753,
-      "step": 513
-    },
-    {
-      "epoch": 0.009409610983981694,
-      "grad_norm": 0.47884055972099304,
-      "learning_rate": 1.128795596577563e-05,
-      "loss": 0.0608,
-      "step": 514
-    },
-    {
-      "epoch": 0.0094279176201373,
-      "grad_norm": 2.7863001823425293,
-      "learning_rate": 1.1253332335643043e-05,
-      "loss": 0.0826,
-      "step": 515
-    },
-    {
-      "epoch": 0.009446224256292907,
-      "grad_norm": 0.3365868628025055,
-      "learning_rate": 1.1218693434051475e-05,
-      "loss": 0.0615,
-      "step": 516
-    },
-    {
-      "epoch": 0.009464530892448513,
-      "grad_norm": 0.8875742554664612,
-      "learning_rate": 1.1184039683065014e-05,
-      "loss": 0.0868,
-      "step": 517
-    },
-    {
-      "epoch": 0.00948283752860412,
-      "grad_norm": 0.47412580251693726,
-      "learning_rate": 1.1149371504928667e-05,
-      "loss": 0.0655,
-      "step": 518
-    },
-    {
-      "epoch": 0.009501144164759726,
-      "grad_norm": 0.6756238341331482,
-      "learning_rate": 1.1114689322063255e-05,
-      "loss": 0.0697,
-      "step": 519
-    },
-    {
-      "epoch": 0.009519450800915331,
-      "grad_norm": 0.4974288046360016,
-      "learning_rate": 1.1079993557060228e-05,
-      "loss": 0.0522,
-      "step": 520
-    },
-    {
-      "epoch": 0.009537757437070938,
-      "grad_norm": 0.4407835304737091,
-      "learning_rate": 1.1045284632676535e-05,
-      "loss": 0.0565,
-      "step": 521
-    },
-    {
-      "epoch": 0.009556064073226544,
-      "grad_norm": 0.5031054019927979,
-      "learning_rate": 1.1010562971829464e-05,
-      "loss": 0.0811,
-      "step": 522
-    },
-    {
-      "epoch": 0.00957437070938215,
-      "grad_norm": 0.49856647849082947,
-      "learning_rate": 1.0975828997591496e-05,
-      "loss": 0.0467,
-      "step": 523
-    },
-    {
-      "epoch": 0.009592677345537757,
-      "grad_norm": 0.4997098743915558,
-      "learning_rate": 1.0941083133185146e-05,
-      "loss": 0.0544,
-      "step": 524
-    },
-    {
-      "epoch": 0.009610983981693364,
-      "grad_norm": 0.3242589831352234,
-      "learning_rate": 1.0906325801977804e-05,
-      "loss": 0.0505,
-      "step": 525
-    },
-    {
-      "epoch": 0.00962929061784897,
-      "grad_norm": 8.494279861450195,
-      "learning_rate": 1.0871557427476585e-05,
-      "loss": 0.332,
-      "step": 526
-    },
-    {
-      "epoch": 0.009647597254004577,
-      "grad_norm": 0.3497224450111389,
-      "learning_rate": 1.083677843332316e-05,
-      "loss": 0.0575,
-      "step": 527
-    },
-    {
-      "epoch": 0.009665903890160183,
-      "grad_norm": 0.3623608648777008,
-      "learning_rate": 1.0801989243288588e-05,
-      "loss": 0.0587,
-      "step": 528
-    },
-    {
-      "epoch": 0.00968421052631579,
-      "grad_norm": 0.6769117116928101,
-      "learning_rate": 1.0767190281268187e-05,
-      "loss": 0.0613,
-      "step": 529
-    },
-    {
-      "epoch": 0.009702517162471396,
-      "grad_norm": 0.73320072889328,
-      "learning_rate": 1.0732381971276318e-05,
-      "loss": 0.0826,
-      "step": 530
-    },
-    {
-      "epoch": 0.009720823798627003,
-      "grad_norm": 0.2625424563884735,
-      "learning_rate": 1.0697564737441254e-05,
-      "loss": 0.0312,
-      "step": 531
-    },
-    {
-      "epoch": 0.009739130434782608,
-      "grad_norm": 0.3892664313316345,
-      "learning_rate": 1.0662739004000005e-05,
-      "loss": 0.0515,
-      "step": 532
-    },
-    {
-      "epoch": 0.009757437070938214,
-      "grad_norm": 0.4974634647369385,
-      "learning_rate": 1.0627905195293135e-05,
-      "loss": 0.0575,
-      "step": 533
-    },
-    {
-      "epoch": 0.009775743707093821,
-      "grad_norm": 0.4967100918292999,
-      "learning_rate": 1.0593063735759619e-05,
-      "loss": 0.0528,
-      "step": 534
-    },
-    {
-      "epoch": 0.009794050343249427,
-      "grad_norm": 0.34812474250793457,
-      "learning_rate": 1.055821504993164e-05,
-      "loss": 0.0541,
-      "step": 535
-    },
-    {
-      "epoch": 0.009812356979405034,
-      "grad_norm": 0.26066046953201294,
-      "learning_rate": 1.0523359562429441e-05,
-      "loss": 0.0465,
-      "step": 536
-    },
-    {
-      "epoch": 0.00983066361556064,
-      "grad_norm": 0.44758743047714233,
-      "learning_rate": 1.0488497697956134e-05,
-      "loss": 0.0549,
-      "step": 537
-    },
-    {
-      "epoch": 0.009848970251716247,
-      "grad_norm": 3.4418416023254395,
-      "learning_rate": 1.0453629881292537e-05,
-      "loss": 0.1751,
-      "step": 538
-    },
-    {
-      "epoch": 0.009867276887871854,
-      "grad_norm": 0.7012713551521301,
-      "learning_rate": 1.0418756537291996e-05,
-      "loss": 0.0829,
-      "step": 539
-    },
-    {
-      "epoch": 0.00988558352402746,
-      "grad_norm": 0.38024449348449707,
-      "learning_rate": 1.03838780908752e-05,
-      "loss": 0.0515,
-      "step": 540
-    },
-    {
-      "epoch": 0.009903890160183067,
-      "grad_norm": 0.35983434319496155,
-      "learning_rate": 1.0348994967025012e-05,
-      "loss": 0.0537,
-      "step": 541
-    },
-    {
-      "epoch": 0.009922196796338673,
-      "grad_norm": 0.712012767791748,
-      "learning_rate": 1.0314107590781284e-05,
-      "loss": 0.0663,
-      "step": 542
-    },
-    {
-      "epoch": 0.00994050343249428,
-      "grad_norm": 0.4738052487373352,
-      "learning_rate": 1.0279216387235691e-05,
-      "loss": 0.0654,
-      "step": 543
-    },
-    {
-      "epoch": 0.009958810068649886,
-      "grad_norm": 3.883636236190796,
-      "learning_rate": 1.0244321781526533e-05,
-      "loss": 0.1111,
-      "step": 544
-    },
-    {
-      "epoch": 0.009977116704805491,
-      "grad_norm": 0.5673316717147827,
-      "learning_rate": 1.0209424198833571e-05,
-      "loss": 0.0535,
-      "step": 545
-    },
-    {
-      "epoch": 0.009995423340961098,
-      "grad_norm": 0.40713363885879517,
-      "learning_rate": 1.0174524064372837e-05,
-      "loss": 0.0494,
-      "step": 546
-    },
-    {
-      "epoch": 0.010013729977116704,
-      "grad_norm": 0.5395656824111938,
-      "learning_rate": 1.0139621803391454e-05,
-      "loss": 0.0562,
-      "step": 547
-    },
-    {
-      "epoch": 0.01003203661327231,
-      "grad_norm": 0.560680091381073,
-      "learning_rate": 1.010471784116246e-05,
-      "loss": 0.0643,
-      "step": 548
-    },
-    {
-      "epoch": 0.010050343249427917,
-      "grad_norm": 0.565285861492157,
-      "learning_rate": 1.0069812602979617e-05,
-      "loss": 0.0469,
-      "step": 549
-    },
-    {
-      "epoch": 0.010068649885583524,
-      "grad_norm": 1.6419192552566528,
-      "learning_rate": 1.0034906514152239e-05,
-      "loss": 0.0941,
-      "step": 550
-    },
-    {
-      "epoch": 0.01008695652173913,
-      "grad_norm": 0.4307943880558014,
-      "learning_rate": 1e-05,
-      "loss": 0.0467,
-      "step": 551
-    },
-    {
-      "epoch": 0.010105263157894737,
-      "grad_norm": 0.508974552154541,
-      "learning_rate": 9.965093485847766e-06,
-      "loss": 0.0676,
-      "step": 552
-    },
-    {
-      "epoch": 0.010123569794050344,
-      "grad_norm": 0.5743252635002136,
-      "learning_rate": 9.930187397020385e-06,
-      "loss": 0.0698,
-      "step": 553
-    },
-    {
-      "epoch": 0.01014187643020595,
-      "grad_norm": 7.205303192138672,
-      "learning_rate": 9.895282158837545e-06,
-      "loss": 0.1653,
-      "step": 554
-    },
-    {
-      "epoch": 0.010160183066361557,
-      "grad_norm": 0.5084084272384644,
-      "learning_rate": 9.860378196608549e-06,
-      "loss": 0.068,
-      "step": 555
-    },
-    {
-      "epoch": 0.010178489702517163,
-      "grad_norm": 0.6783789396286011,
-      "learning_rate": 9.825475935627165e-06,
-      "loss": 0.0745,
-      "step": 556
-    },
-    {
-      "epoch": 0.01019679633867277,
-      "grad_norm": 0.2569085955619812,
-      "learning_rate": 9.790575801166432e-06,
-      "loss": 0.0318,
-      "step": 557
-    },
-    {
-      "epoch": 0.010215102974828374,
-      "grad_norm": 0.4879227578639984,
-      "learning_rate": 9.75567821847347e-06,
-      "loss": 0.0693,
-      "step": 558
-    },
-    {
-      "epoch": 0.010233409610983981,
-      "grad_norm": 0.3231469392776489,
-      "learning_rate": 9.720783612764314e-06,
-      "loss": 0.0431,
-      "step": 559
-    },
-    {
-      "epoch": 0.010251716247139588,
-      "grad_norm": 1.1669337749481201,
-      "learning_rate": 9.685892409218718e-06,
-      "loss": 0.1176,
-      "step": 560
-    },
-    {
-      "epoch": 0.010270022883295194,
-      "grad_norm": 0.4549786448478699,
-      "learning_rate": 9.651005032974994e-06,
-      "loss": 0.0569,
-      "step": 561
-    },
-    {
-      "epoch": 0.0102883295194508,
-      "grad_norm": 2.1591649055480957,
-      "learning_rate": 9.616121909124801e-06,
-      "loss": 0.078,
-      "step": 562
-    },
-    {
-      "epoch": 0.010306636155606407,
-      "grad_norm": 0.486910343170166,
-      "learning_rate": 9.581243462708007e-06,
-      "loss": 0.0468,
-      "step": 563
-    },
-    {
-      "epoch": 0.010324942791762014,
-      "grad_norm": 5.886855602264404,
-      "learning_rate": 9.546370118707463e-06,
-      "loss": 0.237,
-      "step": 564
-    },
-    {
-      "epoch": 0.01034324942791762,
-      "grad_norm": 0.4082443416118622,
-      "learning_rate": 9.511502302043867e-06,
-      "loss": 0.0457,
-      "step": 565
-    },
-    {
-      "epoch": 0.010361556064073227,
-      "grad_norm": 0.4189354479312897,
-      "learning_rate": 9.476640437570562e-06,
-      "loss": 0.0574,
-      "step": 566
-    },
-    {
-      "epoch": 0.010379862700228833,
-      "grad_norm": 0.3079005181789398,
-      "learning_rate": 9.441784950068362e-06,
-      "loss": 0.0507,
-      "step": 567
-    },
-    {
-      "epoch": 0.01039816933638444,
-      "grad_norm": 0.6179872751235962,
-      "learning_rate": 9.406936264240386e-06,
-      "loss": 0.077,
-      "step": 568
-    },
-    {
-      "epoch": 0.010416475972540046,
-      "grad_norm": 1.722801923751831,
-      "learning_rate": 9.372094804706867e-06,
-      "loss": 0.1001,
-      "step": 569
-    },
-    {
-      "epoch": 0.010434782608695653,
-      "grad_norm": 0.2567720115184784,
-      "learning_rate": 9.337260996000002e-06,
-      "loss": 0.0475,
-      "step": 570
-    },
-    {
-      "epoch": 0.010453089244851258,
-      "grad_norm": 1.292287826538086,
-      "learning_rate": 9.302435262558748e-06,
-      "loss": 0.0709,
-      "step": 571
-    },
-    {
-      "epoch": 0.010471395881006864,
-      "grad_norm": 0.3817519247531891,
-      "learning_rate": 9.267618028723687e-06,
-      "loss": 0.0677,
-      "step": 572
-    },
-    {
-      "epoch": 0.010489702517162471,
-      "grad_norm": 0.5979799628257751,
-      "learning_rate": 9.232809718731815e-06,
-      "loss": 0.0593,
-      "step": 573
-    },
-    {
-      "epoch": 0.010508009153318077,
-      "grad_norm": 0.37470927834510803,
-      "learning_rate": 9.198010756711413e-06,
-      "loss": 0.0628,
-      "step": 574
-    },
-    {
-      "epoch": 0.010526315789473684,
-      "grad_norm": 0.9199326038360596,
-      "learning_rate": 9.163221566676847e-06,
-      "loss": 0.0851,
-      "step": 575
-    },
-    {
-      "epoch": 0.01054462242562929,
-      "grad_norm": 0.48233625292778015,
-      "learning_rate": 9.128442572523418e-06,
-      "loss": 0.065,
-      "step": 576
-    },
-    {
-      "epoch": 0.010562929061784897,
-      "grad_norm": 3.3563363552093506,
-      "learning_rate": 9.093674198022201e-06,
-      "loss": 0.1219,
-      "step": 577
-    },
-    {
-      "epoch": 0.010581235697940504,
-      "grad_norm": 0.5125579833984375,
-      "learning_rate": 9.058916866814857e-06,
-      "loss": 0.0725,
-      "step": 578
-    },
-    {
-      "epoch": 0.01059954233409611,
-      "grad_norm": 0.6339969038963318,
-      "learning_rate": 9.024171002408507e-06,
-      "loss": 0.0783,
-      "step": 579
-    },
-    {
-      "epoch": 0.010617848970251717,
-      "grad_norm": 0.40467336773872375,
-      "learning_rate": 8.989437028170537e-06,
-      "loss": 0.0551,
-      "step": 580
-    },
-    {
-      "epoch": 0.010636155606407323,
-      "grad_norm": 0.45042675733566284,
-      "learning_rate": 8.954715367323468e-06,
-      "loss": 0.0681,
-      "step": 581
-    },
-    {
-      "epoch": 0.01065446224256293,
-      "grad_norm": 0.6091125011444092,
-      "learning_rate": 8.920006442939772e-06,
-      "loss": 0.0717,
-      "step": 582
-    },
-    {
-      "epoch": 0.010672768878718536,
-      "grad_norm": 0.5294237732887268,
-      "learning_rate": 8.885310677936746e-06,
-      "loss": 0.0484,
-      "step": 583
-    },
-    {
-      "epoch": 0.010691075514874141,
-      "grad_norm": 0.6482405662536621,
-      "learning_rate": 8.850628495071336e-06,
-      "loss": 0.0546,
-      "step": 584
-    },
-    {
-      "epoch": 0.010709382151029748,
-      "grad_norm": 0.5270109176635742,
-      "learning_rate": 8.815960316934991e-06,
-      "loss": 0.0631,
-      "step": 585
-    },
-    {
-      "epoch": 0.010727688787185354,
-      "grad_norm": 0.29208895564079285,
-      "learning_rate": 8.781306565948528e-06,
-      "loss": 0.0466,
-      "step": 586
-    },
-    {
-      "epoch": 0.01074599542334096,
-      "grad_norm": 0.8654493093490601,
-      "learning_rate": 8.746667664356957e-06,
-      "loss": 0.0887,
-      "step": 587
-    },
-    {
-      "epoch": 0.010764302059496567,
-      "grad_norm": 0.586911141872406,
-      "learning_rate": 8.712044034224374e-06,
-      "loss": 0.0626,
-      "step": 588
-    },
-    {
-      "epoch": 0.010782608695652174,
-      "grad_norm": 0.5669876933097839,
-      "learning_rate": 8.677436097428775e-06,
-      "loss": 0.054,
-      "step": 589
-    },
-    {
-      "epoch": 0.01080091533180778,
-      "grad_norm": 0.38175487518310547,
-      "learning_rate": 8.642844275656957e-06,
-      "loss": 0.0451,
-      "step": 590
-    },
-    {
-      "epoch": 0.010819221967963387,
-      "grad_norm": 1.8506982326507568,
-      "learning_rate": 8.60826899039935e-06,
-      "loss": 0.094,
-      "step": 591
-    },
-    {
-      "epoch": 0.010837528604118993,
-      "grad_norm": 0.5754877328872681,
-      "learning_rate": 8.573710662944884e-06,
-      "loss": 0.0618,
-      "step": 592
-    },
-    {
-      "epoch": 0.0108558352402746,
-      "grad_norm": 0.5001334547996521,
-      "learning_rate": 8.539169714375885e-06,
-      "loss": 0.0552,
-      "step": 593
-    },
-    {
-      "epoch": 0.010874141876430207,
-      "grad_norm": 0.33197906613349915,
-      "learning_rate": 8.504646565562907e-06,
-      "loss": 0.0511,
-      "step": 594
-    },
-    {
-      "epoch": 0.010892448512585813,
-      "grad_norm": 0.39409497380256653,
-      "learning_rate": 8.47014163715962e-06,
-      "loss": 0.0738,
-      "step": 595
-    },
-    {
-      "epoch": 0.010910755148741418,
-      "grad_norm": 27.13417625427246,
-      "learning_rate": 8.43565534959769e-06,
-      "loss": 0.2403,
-      "step": 596
-    },
-    {
-      "epoch": 0.010929061784897024,
-      "grad_norm": 0.39403894543647766,
-      "learning_rate": 8.401188123081653e-06,
-      "loss": 0.0464,
-      "step": 597
-    },
-    {
-      "epoch": 0.010947368421052631,
-      "grad_norm": 0.32075920701026917,
-      "learning_rate": 8.366740377583781e-06,
-      "loss": 0.0483,
-      "step": 598
-    },
-    {
-      "epoch": 0.010965675057208238,
-      "grad_norm": 0.3784872889518738,
-      "learning_rate": 8.332312532838978e-06,
-      "loss": 0.0533,
-      "step": 599
-    },
-    {
-      "epoch": 0.010983981693363844,
-      "grad_norm": 0.2961476147174835,
-      "learning_rate": 8.297905008339677e-06,
-      "loss": 0.0442,
-      "step": 600
-    },
-    {
-      "epoch": 0.01100228832951945,
-      "grad_norm": 0.3700519800186157,
-      "learning_rate": 8.263518223330698e-06,
-      "loss": 0.0432,
-      "step": 601
-    },
-    {
-      "epoch": 0.011020594965675057,
-      "grad_norm": 8.323330879211426,
-      "learning_rate": 8.22915259680417e-06,
-      "loss": 0.2797,
-      "step": 602
-    },
-    {
-      "epoch": 0.011038901601830664,
-      "grad_norm": 0.4720452129840851,
-      "learning_rate": 8.194808547494401e-06,
-      "loss": 0.0548,
-      "step": 603
-    },
-    {
-      "epoch": 0.01105720823798627,
-      "grad_norm": 0.4294449985027313,
-      "learning_rate": 8.1604864938728e-06,
-      "loss": 0.0674,
-      "step": 604
-    },
-    {
-      "epoch": 0.011075514874141877,
-      "grad_norm": 0.4345700740814209,
-      "learning_rate": 8.126186854142752e-06,
-      "loss": 0.0613,
-      "step": 605
-    },
-    {
-      "epoch": 0.011093821510297483,
-      "grad_norm": 0.5550083518028259,
-      "learning_rate": 8.091910046234552e-06,
-      "loss": 0.0632,
-      "step": 606
-    },
-    {
-      "epoch": 0.01111212814645309,
-      "grad_norm": 0.4370512366294861,
-      "learning_rate": 8.057656487800283e-06,
-      "loss": 0.0453,
-      "step": 607
-    },
-    {
-      "epoch": 0.011130434782608696,
-      "grad_norm": 0.44062379002571106,
-      "learning_rate": 8.023426596208739e-06,
-      "loss": 0.0757,
-      "step": 608
-    },
-    {
-      "epoch": 0.011148741418764301,
-      "grad_norm": 0.346194863319397,
-      "learning_rate": 7.989220788540356e-06,
-      "loss": 0.043,
-      "step": 609
-    },
-    {
-      "epoch": 0.011167048054919908,
-      "grad_norm": 1.020596981048584,
-      "learning_rate": 7.955039481582098e-06,
-      "loss": 0.0678,
-      "step": 610
-    },
-    {
-      "epoch": 0.011185354691075514,
-      "grad_norm": 0.4142068326473236,
-      "learning_rate": 7.92088309182241e-06,
-      "loss": 0.0574,
-      "step": 611
-    },
-    {
-      "epoch": 0.01120366132723112,
-      "grad_norm": 0.4963121712207794,
-      "learning_rate": 7.886752035446116e-06,
-      "loss": 0.0515,
-      "step": 612
-    },
-    {
-      "epoch": 0.011221967963386727,
-      "grad_norm": 1.4897319078445435,
-      "learning_rate": 7.852646728329368e-06,
-      "loss": 0.069,
-      "step": 613
-    },
-    {
-      "epoch": 0.011240274599542334,
-      "grad_norm": 0.6016538143157959,
-      "learning_rate": 7.818567586034578e-06,
-      "loss": 0.055,
-      "step": 614
-    },
-    {
-      "epoch": 0.01125858123569794,
-      "grad_norm": 0.5622033476829529,
-      "learning_rate": 7.784515023805328e-06,
-      "loss": 0.0597,
-      "step": 615
-    },
-    {
-      "epoch": 0.011276887871853547,
-      "grad_norm": 0.3546046316623688,
-      "learning_rate": 7.750489456561351e-06,
-      "loss": 0.0417,
-      "step": 616
-    },
-    {
-      "epoch": 0.011295194508009154,
-      "grad_norm": 0.6203312277793884,
-      "learning_rate": 7.716491298893443e-06,
-      "loss": 0.0715,
-      "step": 617
-    },
-    {
-      "epoch": 0.01131350114416476,
-      "grad_norm": 0.3368282616138458,
-      "learning_rate": 7.68252096505843e-06,
-      "loss": 0.0535,
-      "step": 618
-    },
-    {
-      "epoch": 0.011331807780320367,
-      "grad_norm": 0.57204270362854,
-      "learning_rate": 7.6485788689741e-06,
-      "loss": 0.0636,
-      "step": 619
-    },
-    {
-      "epoch": 0.011350114416475973,
-      "grad_norm": 1.707099437713623,
-      "learning_rate": 7.6146654242141935e-06,
-      "loss": 0.1125,
-      "step": 620
-    },
-    {
-      "epoch": 0.01136842105263158,
-      "grad_norm": 21.12303352355957,
-      "learning_rate": 7.580781044003324e-06,
-      "loss": 0.6887,
-      "step": 621
-    },
-    {
-      "epoch": 0.011386727688787185,
-      "grad_norm": 0.47708362340927124,
-      "learning_rate": 7.546926141211975e-06,
-      "loss": 0.0532,
-      "step": 622
-    },
-    {
-      "epoch": 0.011405034324942791,
-      "grad_norm": 0.3427441716194153,
-      "learning_rate": 7.513101128351454e-06,
-      "loss": 0.0457,
-      "step": 623
-    },
-    {
-      "epoch": 0.011423340961098398,
-      "grad_norm": 0.5798665285110474,
-      "learning_rate": 7.4793064175688635e-06,
-      "loss": 0.0585,
-      "step": 624
-    },
-    {
-      "epoch": 0.011441647597254004,
-      "grad_norm": 0.4988366365432739,
-      "learning_rate": 7.445542420642097e-06,
-      "loss": 0.0527,
-      "step": 625
-    },
-    {
-      "epoch": 0.01145995423340961,
-      "grad_norm": 0.9486873149871826,
-      "learning_rate": 7.411809548974792e-06,
-      "loss": 0.0881,
-      "step": 626
-    },
-    {
-      "epoch": 0.011478260869565217,
-      "grad_norm": 2.5744500160217285,
-      "learning_rate": 7.378108213591355e-06,
-      "loss": 0.0574,
-      "step": 627
-    },
-    {
-      "epoch": 0.011496567505720824,
-      "grad_norm": 0.3006826341152191,
-      "learning_rate": 7.344438825131912e-06,
-      "loss": 0.0392,
-      "step": 628
-    },
-    {
-      "epoch": 0.01151487414187643,
-      "grad_norm": 0.5133454203605652,
-      "learning_rate": 7.310801793847344e-06,
-      "loss": 0.0509,
-      "step": 629
-    },
-    {
-      "epoch": 0.011533180778032037,
-      "grad_norm": 0.3883040249347687,
-      "learning_rate": 7.277197529594257e-06,
-      "loss": 0.057,
-      "step": 630
-    },
-    {
-      "epoch": 0.011551487414187643,
-      "grad_norm": 0.521661102771759,
-      "learning_rate": 7.243626441830009e-06,
-      "loss": 0.0618,
-      "step": 631
-    },
-    {
-      "epoch": 0.01156979405034325,
-      "grad_norm": 0.3240382671356201,
-      "learning_rate": 7.210088939607709e-06,
-      "loss": 0.0516,
-      "step": 632
-    },
-    {
-      "epoch": 0.011588100686498857,
-      "grad_norm": 0.5525559782981873,
-      "learning_rate": 7.176585431571235e-06,
-      "loss": 0.0583,
-      "step": 633
-    },
-    {
-      "epoch": 0.011606407322654463,
-      "grad_norm": 0.5469704866409302,
-      "learning_rate": 7.143116325950266e-06,
-      "loss": 0.0598,
-      "step": 634
-    },
-    {
-      "epoch": 0.011624713958810068,
-      "grad_norm": 9.659605026245117,
-      "learning_rate": 7.109682030555283e-06,
-      "loss": 0.1597,
-      "step": 635
-    },
-    {
-      "epoch": 0.011643020594965674,
-      "grad_norm": 0.330433189868927,
-      "learning_rate": 7.076282952772634e-06,
-      "loss": 0.0546,
-      "step": 636
-    },
-    {
-      "epoch": 0.011661327231121281,
-      "grad_norm": 1.3363778591156006,
-      "learning_rate": 7.042919499559538e-06,
-      "loss": 0.0653,
-      "step": 637
-    },
-    {
-      "epoch": 0.011679633867276887,
-      "grad_norm": 0.5439719557762146,
-      "learning_rate": 7.009592077439135e-06,
-      "loss": 0.0661,
-      "step": 638
-    },
-    {
-      "epoch": 0.011697940503432494,
-      "grad_norm": 0.4256076514720917,
-      "learning_rate": 6.976301092495556e-06,
-      "loss": 0.05,
-      "step": 639
-    },
-    {
-      "epoch": 0.0117162471395881,
-      "grad_norm": 0.5813340544700623,
-      "learning_rate": 6.943046950368944e-06,
-      "loss": 0.0646,
-      "step": 640
-    },
-    {
-      "epoch": 0.011734553775743707,
-      "grad_norm": 0.39457425475120544,
-      "learning_rate": 6.909830056250527e-06,
-      "loss": 0.0571,
-      "step": 641
-    },
-    {
-      "epoch": 0.011752860411899314,
-      "grad_norm": 0.3928937315940857,
-      "learning_rate": 6.876650814877675e-06,
-      "loss": 0.0661,
-      "step": 642
-    },
-    {
-      "epoch": 0.01177116704805492,
-      "grad_norm": 1.2465202808380127,
-      "learning_rate": 6.843509630528977e-06,
-      "loss": 0.0714,
-      "step": 643
-    },
-    {
-      "epoch": 0.011789473684210527,
-      "grad_norm": 0.8547323942184448,
-      "learning_rate": 6.8104069070193e-06,
-      "loss": 0.054,
-      "step": 644
-    },
-    {
-      "epoch": 0.011807780320366133,
-      "grad_norm": 1.043526530265808,
-      "learning_rate": 6.777343047694891e-06,
-      "loss": 0.1093,
-      "step": 645
-    },
-    {
-      "epoch": 0.01182608695652174,
-      "grad_norm": 1.0612878799438477,
-      "learning_rate": 6.744318455428436e-06,
-      "loss": 0.0723,
-      "step": 646
-    },
-    {
-      "epoch": 0.011844393592677346,
-      "grad_norm": 0.41253185272216797,
-      "learning_rate": 6.711333532614168e-06,
-      "loss": 0.0681,
-      "step": 647
-    },
-    {
-      "epoch": 0.011862700228832951,
-      "grad_norm": 0.5279849171638489,
-      "learning_rate": 6.67838868116297e-06,
-      "loss": 0.0758,
-      "step": 648
-    },
-    {
-      "epoch": 0.011881006864988558,
-      "grad_norm": 0.5534442067146301,
-      "learning_rate": 6.645484302497452e-06,
-      "loss": 0.0484,
-      "step": 649
-    },
-    {
-      "epoch": 0.011899313501144164,
-      "grad_norm": 0.9848259091377258,
-      "learning_rate": 6.612620797547087e-06,
-      "loss": 0.0715,
-      "step": 650
-    },
-    {
-      "epoch": 0.01191762013729977,
-      "grad_norm": 0.512266218662262,
-      "learning_rate": 6.579798566743314e-06,
-      "loss": 0.0561,
-      "step": 651
-    },
-    {
-      "epoch": 0.011935926773455377,
-      "grad_norm": 0.34552109241485596,
-      "learning_rate": 6.547018010014654e-06,
-      "loss": 0.0577,
-      "step": 652
-    },
-    {
-      "epoch": 0.011954233409610984,
-      "grad_norm": 0.191960409283638,
-      "learning_rate": 6.5142795267818505e-06,
-      "loss": 0.039,
-      "step": 653
-    },
-    {
-      "epoch": 0.01197254004576659,
-      "grad_norm": 0.7968975901603699,
-      "learning_rate": 6.481583515952983e-06,
-      "loss": 0.0539,
-      "step": 654
-    },
-    {
-      "epoch": 0.011990846681922197,
-      "grad_norm": 2.4363250732421875,
-      "learning_rate": 6.448930375918632e-06,
-      "loss": 0.1162,
-      "step": 655
-    },
-    {
-      "epoch": 0.012009153318077804,
-      "grad_norm": 1.4093053340911865,
-      "learning_rate": 6.4163205045469975e-06,
-      "loss": 0.0697,
-      "step": 656
-    },
-    {
-      "epoch": 0.01202745995423341,
-      "grad_norm": 0.5579004287719727,
-      "learning_rate": 6.383754299179079e-06,
-      "loss": 0.0637,
-      "step": 657
-    },
-    {
-      "epoch": 0.012045766590389017,
-      "grad_norm": 0.4740244448184967,
-      "learning_rate": 6.351232156623803e-06,
-      "loss": 0.0456,
-      "step": 658
-    },
-    {
-      "epoch": 0.012064073226544623,
-      "grad_norm": 0.5255889892578125,
-      "learning_rate": 6.318754473153221e-06,
-      "loss": 0.065,
-      "step": 659
-    },
-    {
-      "epoch": 0.01208237986270023,
-      "grad_norm": 0.8857507109642029,
-      "learning_rate": 6.286321644497655e-06,
-      "loss": 0.0694,
-      "step": 660
-    },
-    {
-      "epoch": 0.012100686498855834,
-      "grad_norm": 0.26788559556007385,
-      "learning_rate": 6.25393406584088e-06,
-      "loss": 0.0551,
-      "step": 661
-    },
-    {
-      "epoch": 0.012118993135011441,
-      "grad_norm": 2.883422374725342,
-      "learning_rate": 6.22159213181533e-06,
-      "loss": 0.0914,
-      "step": 662
-    },
-    {
-      "epoch": 0.012137299771167048,
-      "grad_norm": 1.6845279932022095,
-      "learning_rate": 6.18929623649726e-06,
-      "loss": 0.1316,
-      "step": 663
-    },
-    {
-      "epoch": 0.012155606407322654,
-      "grad_norm": 1.7865440845489502,
-      "learning_rate": 6.157046773401964e-06,
-      "loss": 0.0873,
-      "step": 664
-    },
-    {
-      "epoch": 0.01217391304347826,
-      "grad_norm": 0.4027663469314575,
-      "learning_rate": 6.124844135478971e-06,
-      "loss": 0.0556,
-      "step": 665
-    },
-    {
-      "epoch": 0.012192219679633867,
-      "grad_norm": 0.5781440734863281,
-      "learning_rate": 6.092688715107265e-06,
-      "loss": 0.0599,
-      "step": 666
-    },
-    {
-      "epoch": 0.012210526315789474,
-      "grad_norm": 0.5784943699836731,
-      "learning_rate": 6.06058090409049e-06,
-      "loss": 0.0479,
-      "step": 667
-    },
-    {
-      "epoch": 0.01222883295194508,
-      "grad_norm": 0.4784940779209137,
-      "learning_rate": 6.028521093652195e-06,
-      "loss": 0.0474,
-      "step": 668
-    },
-    {
-      "epoch": 0.012247139588100687,
-      "grad_norm": 0.9010287523269653,
-      "learning_rate": 5.996509674431053e-06,
-      "loss": 0.0567,
-      "step": 669
-    },
-    {
-      "epoch": 0.012265446224256293,
-      "grad_norm": 0.3100776970386505,
-      "learning_rate": 5.9645470364761e-06,
-      "loss": 0.0452,
-      "step": 670
-    },
-    {
-      "epoch": 0.0122837528604119,
-      "grad_norm": 5.5637288093566895,
-      "learning_rate": 5.932633569242e-06,
-      "loss": 0.1377,
-      "step": 671
-    },
-    {
-      "epoch": 0.012302059496567506,
-      "grad_norm": 0.7807855606079102,
-      "learning_rate": 5.900769661584273e-06,
-      "loss": 0.0643,
-      "step": 672
-    },
-    {
-      "epoch": 0.012320366132723111,
-      "grad_norm": 2.135533094406128,
-      "learning_rate": 5.868955701754584e-06,
-      "loss": 0.1135,
-      "step": 673
-    },
-    {
-      "epoch": 0.012338672768878718,
-      "grad_norm": 0.4125937521457672,
-      "learning_rate": 5.83719207739599e-06,
-      "loss": 0.0521,
-      "step": 674
-    },
-    {
-      "epoch": 0.012356979405034324,
-      "grad_norm": 0.37697210907936096,
-      "learning_rate": 5.8054791755382286e-06,
-      "loss": 0.0544,
-      "step": 675
-    },
-    {
-      "epoch": 0.012375286041189931,
-      "grad_norm": 13.010862350463867,
-      "learning_rate": 5.773817382593008e-06,
-      "loss": 0.2921,
-      "step": 676
-    },
-    {
-      "epoch": 0.012393592677345537,
-      "grad_norm": 0.5036109089851379,
-      "learning_rate": 5.742207084349274e-06,
-      "loss": 0.0498,
-      "step": 677
-    },
-    {
-      "epoch": 0.012411899313501144,
-      "grad_norm": 0.3997417688369751,
-      "learning_rate": 5.710648665968543e-06,
-      "loss": 0.0409,
-      "step": 678
-    },
-    {
-      "epoch": 0.01243020594965675,
-      "grad_norm": 0.43998438119888306,
-      "learning_rate": 5.679142511980176e-06,
-      "loss": 0.0523,
-      "step": 679
-    },
-    {
-      "epoch": 0.012448512585812357,
-      "grad_norm": 0.3062395751476288,
-      "learning_rate": 5.647689006276727e-06,
-      "loss": 0.041,
-      "step": 680
-    },
-    {
-      "epoch": 0.012466819221967964,
-      "grad_norm": 0.713205873966217,
-      "learning_rate": 5.616288532109225e-06,
-      "loss": 0.0648,
-      "step": 681
-    },
-    {
-      "epoch": 0.01248512585812357,
-      "grad_norm": 0.6406696438789368,
-      "learning_rate": 5.584941472082549e-06,
-      "loss": 0.0646,
-      "step": 682
-    },
-    {
-      "epoch": 0.012503432494279177,
-      "grad_norm": 0.35565221309661865,
-      "learning_rate": 5.553648208150728e-06,
-      "loss": 0.0474,
-      "step": 683
-    },
-    {
-      "epoch": 0.012521739130434783,
-      "grad_norm": 0.7003399133682251,
-      "learning_rate": 5.522409121612304e-06,
-      "loss": 0.0632,
-      "step": 684
-    },
-    {
-      "epoch": 0.01254004576659039,
-      "grad_norm": 0.46325060725212097,
-      "learning_rate": 5.491224593105695e-06,
-      "loss": 0.0585,
-      "step": 685
-    },
-    {
-      "epoch": 0.012558352402745995,
-      "grad_norm": 0.9421787858009338,
-      "learning_rate": 5.460095002604533e-06,
-      "loss": 0.0537,
-      "step": 686
-    },
-    {
-      "epoch": 0.012576659038901601,
-      "grad_norm": 1.844661831855774,
-      "learning_rate": 5.429020729413062e-06,
-      "loss": 0.0738,
-      "step": 687
-    },
-    {
-      "epoch": 0.012594965675057208,
-      "grad_norm": 1.890405535697937,
-      "learning_rate": 5.398002152161484e-06,
-      "loss": 0.115,
-      "step": 688
-    },
-    {
-      "epoch": 0.012613272311212814,
-      "grad_norm": 0.2724441587924957,
-      "learning_rate": 5.367039648801386e-06,
-      "loss": 0.0401,
-      "step": 689
-    },
-    {
-      "epoch": 0.01263157894736842,
-      "grad_norm": 0.22583965957164764,
-      "learning_rate": 5.336133596601089e-06,
-      "loss": 0.0314,
-      "step": 690
-    },
-    {
-      "epoch": 0.012649885583524027,
-      "grad_norm": 0.6007018089294434,
-      "learning_rate": 5.305284372141095e-06,
-      "loss": 0.0595,
-      "step": 691
-    },
-    {
-      "epoch": 0.012668192219679634,
-      "grad_norm": 0.46658557653427124,
-      "learning_rate": 5.274492351309462e-06,
-      "loss": 0.0554,
-      "step": 692
-    },
-    {
-      "epoch": 0.01268649885583524,
-      "grad_norm": 0.30909547209739685,
-      "learning_rate": 5.243757909297247e-06,
-      "loss": 0.0403,
-      "step": 693
-    },
-    {
-      "epoch": 0.012704805491990847,
-      "grad_norm": 2.325719118118286,
-      "learning_rate": 5.213081420593933e-06,
-      "loss": 0.068,
-      "step": 694
-    },
-    {
-      "epoch": 0.012723112128146453,
-      "grad_norm": 0.6708512902259827,
-      "learning_rate": 5.1824632589828465e-06,
-      "loss": 0.0502,
-      "step": 695
-    },
-    {
-      "epoch": 0.01274141876430206,
-      "grad_norm": 3.4225754737854004,
-      "learning_rate": 5.151903797536631e-06,
-      "loss": 0.1286,
-      "step": 696
-    },
-    {
-      "epoch": 0.012759725400457667,
-      "grad_norm": 0.41358789801597595,
-      "learning_rate": 5.121403408612672e-06,
-      "loss": 0.0535,
-      "step": 697
-    },
-    {
-      "epoch": 0.012778032036613273,
-      "grad_norm": 0.7607048153877258,
-      "learning_rate": 5.090962463848592e-06,
-      "loss": 0.0507,
-      "step": 698
-    },
-    {
-      "epoch": 0.012796338672768878,
-      "grad_norm": 4.125516891479492,
-      "learning_rate": 5.060581334157693e-06,
-      "loss": 0.1272,
-      "step": 699
-    },
-    {
-      "epoch": 0.012814645308924484,
-      "grad_norm": 0.5100997090339661,
-      "learning_rate": 5.030260389724447e-06,
-      "loss": 0.0798,
-      "step": 700
-    },
-    {
-      "epoch": 0.012832951945080091,
-      "grad_norm": 0.5078553557395935,
-      "learning_rate": 5.000000000000003e-06,
-      "loss": 0.0509,
-      "step": 701
-    },
-    {
-      "epoch": 0.012851258581235698,
-      "grad_norm": 0.31738731265068054,
-      "learning_rate": 4.96980053369765e-06,
-      "loss": 0.0455,
-      "step": 702
-    },
-    {
-      "epoch": 0.012869565217391304,
-      "grad_norm": 1.0013046264648438,
-      "learning_rate": 4.939662358788364e-06,
-      "loss": 0.084,
-      "step": 703
-    },
-    {
-      "epoch": 0.01288787185354691,
-      "grad_norm": 0.48317721486091614,
-      "learning_rate": 4.909585842496287e-06,
-      "loss": 0.0554,
-      "step": 704
-    },
-    {
-      "epoch": 0.012906178489702517,
-      "grad_norm": 0.5006605982780457,
-      "learning_rate": 4.879571351294287e-06,
-      "loss": 0.0587,
-      "step": 705
-    },
-    {
-      "epoch": 0.012924485125858124,
-      "grad_norm": 1.0299829244613647,
-      "learning_rate": 4.849619250899458e-06,
-      "loss": 0.1092,
-      "step": 706
-    },
-    {
-      "epoch": 0.01294279176201373,
-      "grad_norm": 0.43030059337615967,
-      "learning_rate": 4.8197299062687e-06,
-      "loss": 0.0701,
-      "step": 707
-    },
-    {
-      "epoch": 0.012961098398169337,
-      "grad_norm": 0.35754355788230896,
-      "learning_rate": 4.78990368159424e-06,
-      "loss": 0.0529,
-      "step": 708
-    },
-    {
-      "epoch": 0.012979405034324943,
-      "grad_norm": 0.45466580986976624,
-      "learning_rate": 4.76014094029921e-06,
-      "loss": 0.0561,
-      "step": 709
-    },
-    {
-      "epoch": 0.01299771167048055,
-      "grad_norm": 4.951099872589111,
-      "learning_rate": 4.7304420450332244e-06,
-      "loss": 0.1075,
-      "step": 710
-    },
-    {
-      "epoch": 0.013016018306636156,
-      "grad_norm": 0.9091598391532898,
-      "learning_rate": 4.700807357667953e-06,
-      "loss": 0.0723,
-      "step": 711
-    },
-    {
-      "epoch": 0.013034324942791761,
-      "grad_norm": 0.27953800559043884,
-      "learning_rate": 4.671237239292699e-06,
-      "loss": 0.0473,
-      "step": 712
-    },
-    {
-      "epoch": 0.013052631578947368,
-      "grad_norm": 0.3366512656211853,
-      "learning_rate": 4.641732050210032e-06,
-      "loss": 0.0497,
-      "step": 713
-    },
-    {
-      "epoch": 0.013070938215102974,
-      "grad_norm": 0.24474714696407318,
-      "learning_rate": 4.612292149931369e-06,
-      "loss": 0.0402,
-      "step": 714
-    },
-    {
-      "epoch": 0.01308924485125858,
-      "grad_norm": 0.34032437205314636,
-      "learning_rate": 4.582917897172603e-06,
-      "loss": 0.0437,
-      "step": 715
-    },
-    {
-      "epoch": 0.013107551487414187,
-      "grad_norm": 6.40517520904541,
-      "learning_rate": 4.5536096498497295e-06,
-      "loss": 0.1535,
-      "step": 716
-    },
-    {
-      "epoch": 0.013125858123569794,
-      "grad_norm": 0.496954083442688,
-      "learning_rate": 4.524367765074499e-06,
-      "loss": 0.072,
-      "step": 717
-    },
-    {
-      "epoch": 0.0131441647597254,
-      "grad_norm": 0.26708924770355225,
-      "learning_rate": 4.495192599150045e-06,
-      "loss": 0.0318,
-      "step": 718
-    },
-    {
-      "epoch": 0.013162471395881007,
-      "grad_norm": 0.2881370484828949,
-      "learning_rate": 4.46608450756656e-06,
-      "loss": 0.0446,
-      "step": 719
-    },
-    {
-      "epoch": 0.013180778032036614,
-      "grad_norm": 7.275075912475586,
-      "learning_rate": 4.437043844996952e-06,
-      "loss": 0.3213,
-      "step": 720
-    },
-    {
-      "epoch": 0.01319908466819222,
-      "grad_norm": 0.4543837010860443,
-      "learning_rate": 4.408070965292534e-06,
-      "loss": 0.0407,
-      "step": 721
-    },
-    {
-      "epoch": 0.013217391304347827,
-      "grad_norm": 0.4229177236557007,
-      "learning_rate": 4.379166221478697e-06,
-      "loss": 0.0535,
-      "step": 722
-    },
-    {
-      "epoch": 0.013235697940503433,
-      "grad_norm": 0.30908605456352234,
-      "learning_rate": 4.350329965750622e-06,
-      "loss": 0.0502,
-      "step": 723
-    },
-    {
-      "epoch": 0.01325400457665904,
-      "grad_norm": 0.43046000599861145,
-      "learning_rate": 4.321562549468991e-06,
-      "loss": 0.0627,
-      "step": 724
-    },
-    {
-      "epoch": 0.013272311212814645,
-      "grad_norm": 17.89219093322754,
-      "learning_rate": 4.292864323155684e-06,
-      "loss": 0.1194,
-      "step": 725
-    },
-    {
-      "epoch": 0.013290617848970251,
-      "grad_norm": 3.3594727516174316,
-      "learning_rate": 4.264235636489542e-06,
-      "loss": 0.1791,
-      "step": 726
-    },
-    {
-      "epoch": 0.013308924485125858,
-      "grad_norm": 0.23360133171081543,
-      "learning_rate": 4.235676838302069e-06,
-      "loss": 0.0433,
-      "step": 727
-    },
-    {
-      "epoch": 0.013327231121281464,
-      "grad_norm": 0.3113909363746643,
-      "learning_rate": 4.207188276573214e-06,
-      "loss": 0.0466,
-      "step": 728
-    },
-    {
-      "epoch": 0.01334553775743707,
-      "grad_norm": 0.9100260734558105,
-      "learning_rate": 4.178770298427107e-06,
-      "loss": 0.0898,
-      "step": 729
-    },
-    {
-      "epoch": 0.013363844393592677,
-      "grad_norm": 0.4789685606956482,
-      "learning_rate": 4.150423250127846e-06,
-      "loss": 0.0711,
-      "step": 730
-    },
-    {
-      "epoch": 0.013382151029748284,
-      "grad_norm": 0.2928090989589691,
-      "learning_rate": 4.12214747707527e-06,
-      "loss": 0.0388,
-      "step": 731
-    },
-    {
-      "epoch": 0.01340045766590389,
-      "grad_norm": 0.5105188488960266,
-      "learning_rate": 4.093943323800746e-06,
-      "loss": 0.0533,
-      "step": 732
-    },
-    {
-      "epoch": 0.013418764302059497,
-      "grad_norm": 0.537334144115448,
-      "learning_rate": 4.065811133962987e-06,
-      "loss": 0.0536,
-      "step": 733
-    },
-    {
-      "epoch": 0.013437070938215103,
-      "grad_norm": 0.31225720047950745,
-      "learning_rate": 4.037751250343841e-06,
-      "loss": 0.0405,
-      "step": 734
-    },
-    {
-      "epoch": 0.01345537757437071,
-      "grad_norm": 0.3561006188392639,
-      "learning_rate": 4.009764014844143e-06,
-      "loss": 0.0539,
-      "step": 735
-    },
-    {
-      "epoch": 0.013473684210526317,
-      "grad_norm": 0.3621096611022949,
-      "learning_rate": 3.981849768479516e-06,
-      "loss": 0.0421,
-      "step": 736
-    },
-    {
-      "epoch": 0.013491990846681921,
-      "grad_norm": 0.6794489622116089,
-      "learning_rate": 3.954008851376252e-06,
-      "loss": 0.0744,
-      "step": 737
-    },
-    {
-      "epoch": 0.013510297482837528,
-      "grad_norm": 29.703771591186523,
-      "learning_rate": 3.9262416027671354e-06,
-      "loss": 0.6054,
-      "step": 738
-    },
-    {
-      "epoch": 0.013528604118993134,
-      "grad_norm": 0.3844221532344818,
-      "learning_rate": 3.898548360987325e-06,
-      "loss": 0.0423,
-      "step": 739
-    },
-    {
-      "epoch": 0.013546910755148741,
-      "grad_norm": 0.4020056128501892,
-      "learning_rate": 3.8709294634702374e-06,
-      "loss": 0.0597,
-      "step": 740
-    },
-    {
-      "epoch": 0.013565217391304348,
-      "grad_norm": 0.4070514738559723,
-      "learning_rate": 3.8433852467434175e-06,
-      "loss": 0.0453,
-      "step": 741
-    },
-    {
-      "epoch": 0.013583524027459954,
-      "grad_norm": 1.147960901260376,
-      "learning_rate": 3.81591604642446e-06,
-      "loss": 0.0605,
-      "step": 742
-    },
-    {
-      "epoch": 0.01360183066361556,
-      "grad_norm": 0.3141006529331207,
-      "learning_rate": 3.7885221972168974e-06,
-      "loss": 0.0498,
-      "step": 743
-    },
-    {
-      "epoch": 0.013620137299771167,
-      "grad_norm": 3.179751396179199,
-      "learning_rate": 3.7612040329061405e-06,
-      "loss": 0.0948,
-      "step": 744
-    },
-    {
-      "epoch": 0.013638443935926774,
-      "grad_norm": 0.33194786310195923,
-      "learning_rate": 3.7339618863553983e-06,
-      "loss": 0.0557,
-      "step": 745
-    },
-    {
-      "epoch": 0.01365675057208238,
-      "grad_norm": 0.27089881896972656,
-      "learning_rate": 3.7067960895016277e-06,
-      "loss": 0.0491,
-      "step": 746
-    },
-    {
-      "epoch": 0.013675057208237987,
-      "grad_norm": 0.40646010637283325,
-      "learning_rate": 3.679706973351491e-06,
-      "loss": 0.0669,
-      "step": 747
-    },
-    {
-      "epoch": 0.013693363844393593,
-      "grad_norm": 0.680184543132782,
-      "learning_rate": 3.6526948679773256e-06,
-      "loss": 0.0602,
-      "step": 748
-    },
-    {
-      "epoch": 0.0137116704805492,
-      "grad_norm": 5.954007148742676,
-      "learning_rate": 3.625760102513103e-06,
-      "loss": 0.1002,
-      "step": 749
-    },
-    {
-      "epoch": 0.013729977116704805,
-      "grad_norm": 0.3961992561817169,
-      "learning_rate": 3.598903005150444e-06,
-      "loss": 0.0502,
-      "step": 750
-    },
-    {
-      "epoch": 0.013748283752860411,
-      "grad_norm": 0.4490237236022949,
-      "learning_rate": 3.5721239031346067e-06,
-      "loss": 0.0386,
-      "step": 751
-    },
-    {
-      "epoch": 0.013766590389016018,
-      "grad_norm": 0.4439714550971985,
-      "learning_rate": 3.545423122760493e-06,
-      "loss": 0.0695,
-      "step": 752
-    },
-    {
-      "epoch": 0.013784897025171624,
-      "grad_norm": 0.33902743458747864,
-      "learning_rate": 3.5188009893686916e-06,
-      "loss": 0.0461,
-      "step": 753
-    },
-    {
-      "epoch": 0.01380320366132723,
-      "grad_norm": 0.45062753558158875,
-      "learning_rate": 3.492257827341492e-06,
-      "loss": 0.0551,
-      "step": 754
-    },
-    {
-      "epoch": 0.013821510297482837,
-      "grad_norm": 1.196286916732788,
-      "learning_rate": 3.4657939600989453e-06,
-      "loss": 0.0668,
-      "step": 755
-    },
-    {
-      "epoch": 0.013839816933638444,
-      "grad_norm": 0.5484462380409241,
-      "learning_rate": 3.4394097100949286e-06,
-      "loss": 0.0768,
-      "step": 756
-    },
-    {
-      "epoch": 0.01385812356979405,
-      "grad_norm": 0.46227994561195374,
-      "learning_rate": 3.4131053988131947e-06,
-      "loss": 0.0431,
-      "step": 757
-    },
-    {
-      "epoch": 0.013876430205949657,
-      "grad_norm": 0.36154651641845703,
-      "learning_rate": 3.3868813467634833e-06,
-      "loss": 0.0515,
-      "step": 758
-    },
-    {
-      "epoch": 0.013894736842105264,
-      "grad_norm": 1.5799791812896729,
-      "learning_rate": 3.360737873477584e-06,
-      "loss": 0.0812,
-      "step": 759
-    },
-    {
-      "epoch": 0.01391304347826087,
-      "grad_norm": 0.4843911826610565,
-      "learning_rate": 3.3346752975054763e-06,
-      "loss": 0.0487,
-      "step": 760
-    },
-    {
-      "epoch": 0.013931350114416477,
-      "grad_norm": 3.2478013038635254,
-      "learning_rate": 3.308693936411421e-06,
-      "loss": 0.1092,
-      "step": 761
-    },
-    {
-      "epoch": 0.013949656750572083,
-      "grad_norm": 1.8524138927459717,
-      "learning_rate": 3.2827941067700996e-06,
-      "loss": 0.0816,
-      "step": 762
-    },
-    {
-      "epoch": 0.013967963386727688,
-      "grad_norm": 0.5480202436447144,
-      "learning_rate": 3.2569761241627694e-06,
-      "loss": 0.0647,
-      "step": 763
-    },
-    {
-      "epoch": 0.013986270022883295,
-      "grad_norm": 0.4044114947319031,
-      "learning_rate": 3.2312403031733943e-06,
-      "loss": 0.0608,
-      "step": 764
-    },
-    {
-      "epoch": 0.014004576659038901,
-      "grad_norm": 0.7909109592437744,
-      "learning_rate": 3.2055869573848374e-06,
-      "loss": 0.0711,
-      "step": 765
-    },
-    {
-      "epoch": 0.014022883295194508,
-      "grad_norm": 0.6618480086326599,
-      "learning_rate": 3.1800163993750166e-06,
-      "loss": 0.0652,
-      "step": 766
-    },
-    {
-      "epoch": 0.014041189931350114,
-      "grad_norm": 0.5210991501808167,
-      "learning_rate": 3.1545289407131128e-06,
-      "loss": 0.0548,
-      "step": 767
-    },
-    {
-      "epoch": 0.01405949656750572,
-      "grad_norm": 10.014850616455078,
-      "learning_rate": 3.1291248919557717e-06,
-      "loss": 0.3494,
-      "step": 768
-    },
-    {
-      "epoch": 0.014077803203661327,
-      "grad_norm": 0.34373170137405396,
-      "learning_rate": 3.103804562643302e-06,
-      "loss": 0.0475,
-      "step": 769
-    },
-    {
-      "epoch": 0.014096109839816934,
-      "grad_norm": 0.5122294425964355,
-      "learning_rate": 3.0785682612959334e-06,
-      "loss": 0.0445,
-      "step": 770
-    },
-    {
-      "epoch": 0.01411441647597254,
-      "grad_norm": 1.615704894065857,
-      "learning_rate": 3.0534162954100264e-06,
-      "loss": 0.0763,
-      "step": 771
-    },
-    {
-      "epoch": 0.014132723112128147,
-      "grad_norm": 0.5864883065223694,
-      "learning_rate": 3.028348971454356e-06,
-      "loss": 0.0623,
-      "step": 772
-    },
-    {
-      "epoch": 0.014151029748283753,
-      "grad_norm": 0.39968401193618774,
-      "learning_rate": 3.003366594866345e-06,
-      "loss": 0.0595,
-      "step": 773
-    },
-    {
-      "epoch": 0.01416933638443936,
-      "grad_norm": 0.27913814783096313,
-      "learning_rate": 2.978469470048376e-06,
-      "loss": 0.0416,
-      "step": 774
-    },
-    {
-      "epoch": 0.014187643020594966,
-      "grad_norm": 0.4395524263381958,
-      "learning_rate": 2.953657900364053e-06,
-      "loss": 0.0519,
-      "step": 775
-    },
-    {
-      "epoch": 0.014205949656750571,
-      "grad_norm": 1.1202012300491333,
-      "learning_rate": 2.9289321881345257e-06,
-      "loss": 0.0824,
-      "step": 776
-    },
-    {
-      "epoch": 0.014224256292906178,
-      "grad_norm": 0.6589459776878357,
-      "learning_rate": 2.9042926346347932e-06,
-      "loss": 0.0361,
-      "step": 777
-    },
-    {
-      "epoch": 0.014242562929061784,
-      "grad_norm": 0.3817829489707947,
-      "learning_rate": 2.8797395400900362e-06,
-      "loss": 0.0496,
-      "step": 778
-    },
-    {
-      "epoch": 0.014260869565217391,
-      "grad_norm": 0.9931367039680481,
-      "learning_rate": 2.855273203671969e-06,
-      "loss": 0.0595,
-      "step": 779
-    },
-    {
-      "epoch": 0.014279176201372997,
-      "grad_norm": 0.4699876308441162,
-      "learning_rate": 2.830893923495173e-06,
-      "loss": 0.0544,
-      "step": 780
-    },
-    {
-      "epoch": 0.014297482837528604,
-      "grad_norm": 0.780307412147522,
-      "learning_rate": 2.8066019966134907e-06,
-      "loss": 0.0814,
-      "step": 781
-    },
-    {
-      "epoch": 0.01431578947368421,
-      "grad_norm": 1.2149178981781006,
-      "learning_rate": 2.7823977190163788e-06,
-      "loss": 0.0714,
-      "step": 782
-    },
-    {
-      "epoch": 0.014334096109839817,
-      "grad_norm": 0.6094745397567749,
-      "learning_rate": 2.7582813856253276e-06,
-      "loss": 0.0652,
-      "step": 783
-    },
-    {
-      "epoch": 0.014352402745995424,
-      "grad_norm": 0.5055007338523865,
-      "learning_rate": 2.7342532902902418e-06,
-      "loss": 0.0343,
-      "step": 784
-    },
-    {
-      "epoch": 0.01437070938215103,
-      "grad_norm": 0.2958291471004486,
-      "learning_rate": 2.7103137257858867e-06,
-      "loss": 0.0457,
-      "step": 785
-    },
-    {
-      "epoch": 0.014389016018306637,
-      "grad_norm": 1.5310842990875244,
-      "learning_rate": 2.6864629838082957e-06,
-      "loss": 0.0542,
-      "step": 786
-    },
-    {
-      "epoch": 0.014407322654462243,
-      "grad_norm": 0.6714181900024414,
-      "learning_rate": 2.6627013549712355e-06,
-      "loss": 0.0738,
-      "step": 787
-    },
-    {
-      "epoch": 0.01442562929061785,
-      "grad_norm": 6.793590545654297,
-      "learning_rate": 2.639029128802657e-06,
-      "loss": 0.124,
-      "step": 788
-    },
-    {
-      "epoch": 0.014443935926773455,
-      "grad_norm": 0.4314299523830414,
-      "learning_rate": 2.615446593741161e-06,
-      "loss": 0.0609,
-      "step": 789
-    },
-    {
-      "epoch": 0.014462242562929061,
-      "grad_norm": 0.5197092294692993,
-      "learning_rate": 2.5919540371325005e-06,
-      "loss": 0.044,
-      "step": 790
-    },
-    {
-      "epoch": 0.014480549199084668,
-      "grad_norm": 1.95447838306427,
-      "learning_rate": 2.5685517452260566e-06,
-      "loss": 0.1181,
-      "step": 791
-    },
-    {
-      "epoch": 0.014498855835240274,
-      "grad_norm": 1.0716373920440674,
-      "learning_rate": 2.5452400031713786e-06,
-      "loss": 0.0596,
-      "step": 792
-    },
-    {
-      "epoch": 0.01451716247139588,
-      "grad_norm": 22.929214477539062,
-      "learning_rate": 2.522019095014683e-06,
-      "loss": 0.2369,
-      "step": 793
-    },
-    {
-      "epoch": 0.014535469107551487,
-      "grad_norm": 0.3595317304134369,
-      "learning_rate": 2.4988893036954045e-06,
-      "loss": 0.0412,
-      "step": 794
-    },
-    {
-      "epoch": 0.014553775743707094,
-      "grad_norm": 0.8079186081886292,
-      "learning_rate": 2.4758509110427576e-06,
-      "loss": 0.0639,
-      "step": 795
-    },
-    {
-      "epoch": 0.0145720823798627,
-      "grad_norm": 0.6569044589996338,
-      "learning_rate": 2.45290419777228e-06,
-      "loss": 0.0678,
-      "step": 796
-    },
-    {
-      "epoch": 0.014590389016018307,
-      "grad_norm": 0.6705801486968994,
-      "learning_rate": 2.4300494434824373e-06,
-      "loss": 0.0625,
-      "step": 797
-    },
-    {
-      "epoch": 0.014608695652173913,
-      "grad_norm": 0.2948321998119354,
-      "learning_rate": 2.407286926651192e-06,
-      "loss": 0.0454,
-      "step": 798
-    },
-    {
-      "epoch": 0.01462700228832952,
-      "grad_norm": 0.4905836582183838,
-      "learning_rate": 2.3846169246326345e-06,
-      "loss": 0.0505,
-      "step": 799
-    },
-    {
-      "epoch": 0.014645308924485127,
-      "grad_norm": 1.419656753540039,
-      "learning_rate": 2.362039713653581e-06,
-      "loss": 0.1469,
-      "step": 800
-    },
-    {
-      "epoch": 0.014663615560640733,
-      "grad_norm": 1.4188086986541748,
-      "learning_rate": 2.339555568810221e-06,
-      "loss": 0.1024,
-      "step": 801
-    },
-    {
-      "epoch": 0.014681922196796338,
-      "grad_norm": 0.4140421450138092,
-      "learning_rate": 2.317164764064769e-06,
-      "loss": 0.0425,
-      "step": 802
-    },
-    {
-      "epoch": 0.014700228832951944,
-      "grad_norm": 0.33173710107803345,
-      "learning_rate": 2.2948675722421086e-06,
-      "loss": 0.0472,
-      "step": 803
-    },
-    {
-      "epoch": 0.014718535469107551,
-      "grad_norm": 0.41335979104042053,
-      "learning_rate": 2.27266426502649e-06,
-      "loss": 0.0507,
-      "step": 804
-    },
-    {
-      "epoch": 0.014736842105263158,
-      "grad_norm": 2.6610915660858154,
-      "learning_rate": 2.2505551129582047e-06,
-      "loss": 0.116,
-      "step": 805
-    },
-    {
-      "epoch": 0.014755148741418764,
-      "grad_norm": 0.2616541385650635,
-      "learning_rate": 2.2285403854302912e-06,
-      "loss": 0.0591,
-      "step": 806
-    },
-    {
-      "epoch": 0.01477345537757437,
-      "grad_norm": 0.37494024634361267,
-      "learning_rate": 2.206620350685257e-06,
-      "loss": 0.0436,
-      "step": 807
-    },
-    {
-      "epoch": 0.014791762013729977,
-      "grad_norm": 0.9180958867073059,
-      "learning_rate": 2.1847952758118118e-06,
-      "loss": 0.0657,
-      "step": 808
-    },
-    {
-      "epoch": 0.014810068649885584,
-      "grad_norm": 0.5147186517715454,
-      "learning_rate": 2.163065426741603e-06,
-      "loss": 0.0621,
-      "step": 809
-    },
-    {
-      "epoch": 0.01482837528604119,
-      "grad_norm": 0.9221901297569275,
-      "learning_rate": 2.1414310682459805e-06,
-      "loss": 0.0646,
-      "step": 810
-    },
-    {
-      "epoch": 0.014846681922196797,
-      "grad_norm": 1.0287933349609375,
-      "learning_rate": 2.119892463932781e-06,
-      "loss": 0.0742,
-      "step": 811
-    },
-    {
-      "epoch": 0.014864988558352403,
-      "grad_norm": 0.5293546915054321,
-      "learning_rate": 2.098449876243096e-06,
-      "loss": 0.0336,
-      "step": 812
-    },
-    {
-      "epoch": 0.01488329519450801,
-      "grad_norm": 0.28643798828125,
-      "learning_rate": 2.0771035664480944e-06,
-      "loss": 0.0447,
-      "step": 813
-    },
-    {
-      "epoch": 0.014901601830663615,
-      "grad_norm": 0.537362277507782,
-      "learning_rate": 2.0558537946458177e-06,
-      "loss": 0.0578,
-      "step": 814
-    },
-    {
-      "epoch": 0.014919908466819221,
-      "grad_norm": 0.8356306552886963,
-      "learning_rate": 2.0347008197580376e-06,
-      "loss": 0.0641,
-      "step": 815
-    },
-    {
-      "epoch": 0.014938215102974828,
-      "grad_norm": 0.4210380017757416,
-      "learning_rate": 2.013644899527074e-06,
-      "loss": 0.0547,
-      "step": 816
-    },
-    {
-      "epoch": 0.014956521739130434,
-      "grad_norm": 0.6965814828872681,
-      "learning_rate": 1.9926862905126663e-06,
-      "loss": 0.1033,
-      "step": 817
-    },
-    {
-      "epoch": 0.014974828375286041,
-      "grad_norm": 0.3132105767726898,
-      "learning_rate": 1.9718252480888567e-06,
-      "loss": 0.0551,
-      "step": 818
-    },
-    {
-      "epoch": 0.014993135011441647,
-      "grad_norm": 1.153281569480896,
-      "learning_rate": 1.95106202644086e-06,
-      "loss": 0.0813,
-      "step": 819
-    },
-    {
-      "epoch": 0.015011441647597254,
-      "grad_norm": 0.23739849030971527,
-      "learning_rate": 1.930396878561983e-06,
-      "loss": 0.0411,
-      "step": 820
-    },
-    {
-      "epoch": 0.01502974828375286,
-      "grad_norm": 0.4633159935474396,
-      "learning_rate": 1.9098300562505266e-06,
-      "loss": 0.063,
-      "step": 821
-    },
-    {
-      "epoch": 0.015048054919908467,
-      "grad_norm": 2.9505999088287354,
-      "learning_rate": 1.8893618101067357e-06,
-      "loss": 0.0791,
-      "step": 822
-    },
-    {
-      "epoch": 0.015066361556064074,
-      "grad_norm": 0.5407328009605408,
-      "learning_rate": 1.8689923895297247e-06,
-      "loss": 0.0611,
-      "step": 823
-    },
-    {
-      "epoch": 0.01508466819221968,
-      "grad_norm": 0.4537065327167511,
-      "learning_rate": 1.848722042714457e-06,
-      "loss": 0.0445,
-      "step": 824
-    },
-    {
-      "epoch": 0.015102974828375287,
-      "grad_norm": 0.1787199229001999,
-      "learning_rate": 1.8285510166487154e-06,
-      "loss": 0.0414,
-      "step": 825
-    },
-    {
-      "epoch": 0.015121281464530893,
-      "grad_norm": 0.42959287762641907,
-      "learning_rate": 1.808479557110081e-06,
-      "loss": 0.0649,
-      "step": 826
-    },
-    {
-      "epoch": 0.015139588100686498,
-      "grad_norm": 0.6489471197128296,
-      "learning_rate": 1.7885079086629598e-06,
-      "loss": 0.0424,
-      "step": 827
-    },
-    {
-      "epoch": 0.015157894736842105,
-      "grad_norm": 0.36114031076431274,
-      "learning_rate": 1.7686363146555807e-06,
-      "loss": 0.0584,
-      "step": 828
-    },
-    {
-      "epoch": 0.015176201372997711,
-      "grad_norm": 0.4285096824169159,
-      "learning_rate": 1.7488650172170496e-06,
-      "loss": 0.0601,
-      "step": 829
-    },
-    {
-      "epoch": 0.015194508009153318,
-      "grad_norm": 0.6704092621803284,
-      "learning_rate": 1.7291942572543806e-06,
-      "loss": 0.0621,
-      "step": 830
-    },
-    {
-      "epoch": 0.015212814645308924,
-      "grad_norm": 13.114935874938965,
-      "learning_rate": 1.709624274449584e-06,
-      "loss": 0.1256,
-      "step": 831
-    },
-    {
-      "epoch": 0.01523112128146453,
-      "grad_norm": 0.5110636353492737,
-      "learning_rate": 1.6901553072567189e-06,
-      "loss": 0.0651,
-      "step": 832
-    },
-    {
-      "epoch": 0.015249427917620137,
-      "grad_norm": 15.618675231933594,
-      "learning_rate": 1.6707875928990059e-06,
-      "loss": 0.3269,
-      "step": 833
-    },
-    {
-      "epoch": 0.015267734553775744,
-      "grad_norm": 0.5213605761528015,
-      "learning_rate": 1.651521367365936e-06,
-      "loss": 0.0663,
-      "step": 834
-    },
-    {
-      "epoch": 0.01528604118993135,
-      "grad_norm": 0.297579288482666,
-      "learning_rate": 1.6323568654103838e-06,
-      "loss": 0.0484,
-      "step": 835
-    },
-    {
-      "epoch": 0.015304347826086957,
-      "grad_norm": 0.6647035479545593,
-      "learning_rate": 1.6132943205457607e-06,
-      "loss": 0.0657,
-      "step": 836
-    },
-    {
-      "epoch": 0.015322654462242563,
-      "grad_norm": 0.5179057717323303,
-      "learning_rate": 1.5943339650431578e-06,
-      "loss": 0.045,
-      "step": 837
-    },
-    {
-      "epoch": 0.01534096109839817,
-      "grad_norm": 1.4650276899337769,
-      "learning_rate": 1.5754760299285255e-06,
-      "loss": 0.086,
-      "step": 838
-    },
-    {
-      "epoch": 0.015359267734553777,
-      "grad_norm": 3.6571202278137207,
-      "learning_rate": 1.5567207449798517e-06,
-      "loss": 0.1774,
-      "step": 839
-    },
-    {
-      "epoch": 0.015377574370709381,
-      "grad_norm": 0.651657223701477,
-      "learning_rate": 1.538068338724361e-06,
-      "loss": 0.0681,
-      "step": 840
-    },
-    {
-      "epoch": 0.015395881006864988,
-      "grad_norm": 0.5654986500740051,
-      "learning_rate": 1.5195190384357405e-06,
-      "loss": 0.0769,
-      "step": 841
-    },
-    {
-      "epoch": 0.015414187643020594,
-      "grad_norm": 0.289632648229599,
-      "learning_rate": 1.5010730701313626e-06,
-      "loss": 0.0387,
-      "step": 842
-    },
-    {
-      "epoch": 0.015432494279176201,
-      "grad_norm": 0.7785899043083191,
-      "learning_rate": 1.4827306585695234e-06,
-      "loss": 0.0684,
-      "step": 843
-    },
-    {
-      "epoch": 0.015450800915331808,
-      "grad_norm": 0.4201623499393463,
-      "learning_rate": 1.4644920272467245e-06,
-      "loss": 0.062,
-      "step": 844
-    },
-    {
-      "epoch": 0.015469107551487414,
-      "grad_norm": 0.6181161403656006,
-      "learning_rate": 1.446357398394934e-06,
-      "loss": 0.0816,
-      "step": 845
-    },
-    {
-      "epoch": 0.01548741418764302,
-      "grad_norm": 4.406761169433594,
-      "learning_rate": 1.4283269929788779e-06,
-      "loss": 0.1588,
-      "step": 846
-    },
-    {
-      "epoch": 0.015505720823798627,
-      "grad_norm": 0.44824373722076416,
-      "learning_rate": 1.4104010306933558e-06,
-      "loss": 0.0445,
-      "step": 847
-    },
-    {
-      "epoch": 0.015524027459954234,
-      "grad_norm": 0.5433444976806641,
-      "learning_rate": 1.3925797299605649e-06,
-      "loss": 0.0548,
-      "step": 848
-    },
-    {
-      "epoch": 0.01554233409610984,
-      "grad_norm": 0.2696511149406433,
-      "learning_rate": 1.3748633079274254e-06,
-      "loss": 0.0485,
-      "step": 849
-    },
-    {
-      "epoch": 0.015560640732265447,
-      "grad_norm": 3.1581499576568604,
-      "learning_rate": 1.3572519804629537e-06,
-      "loss": 0.068,
-      "step": 850
-    },
-    {
-      "epoch": 0.015578947368421053,
-      "grad_norm": 1.2140463590621948,
-      "learning_rate": 1.339745962155613e-06,
-      "loss": 0.0521,
-      "step": 851
-    },
-    {
-      "epoch": 0.01559725400457666,
-      "grad_norm": 0.432435005903244,
-      "learning_rate": 1.322345466310717e-06,
-      "loss": 0.049,
-      "step": 852
-    },
-    {
-      "epoch": 0.015615560640732265,
-      "grad_norm": 0.4496971368789673,
-      "learning_rate": 1.30505070494781e-06,
-      "loss": 0.046,
-      "step": 853
-    },
-    {
-      "epoch": 0.015633867276887873,
-      "grad_norm": 0.638956606388092,
-      "learning_rate": 1.2878618887981064e-06,
-      "loss": 0.0744,
-      "step": 854
-    },
-    {
-      "epoch": 0.01565217391304348,
-      "grad_norm": 0.5040666460990906,
-      "learning_rate": 1.2707792273019049e-06,
-      "loss": 0.0462,
-      "step": 855
-    },
-    {
-      "epoch": 0.015670480549199086,
-      "grad_norm": 1.3714250326156616,
-      "learning_rate": 1.2538029286060428e-06,
-      "loss": 0.0626,
-      "step": 856
-    },
-    {
-      "epoch": 0.015688787185354693,
-      "grad_norm": 0.3389531075954437,
-      "learning_rate": 1.2369331995613664e-06,
-      "loss": 0.0445,
-      "step": 857
-    },
-    {
-      "epoch": 0.0157070938215103,
-      "grad_norm": 0.30655309557914734,
-      "learning_rate": 1.2201702457201948e-06,
-      "loss": 0.0393,
-      "step": 858
-    },
-    {
-      "epoch": 0.015725400457665902,
-      "grad_norm": 0.6064903736114502,
-      "learning_rate": 1.2035142713338366e-06,
-      "loss": 0.0607,
-      "step": 859
-    },
-    {
-      "epoch": 0.01574370709382151,
-      "grad_norm": 0.3486827611923218,
-      "learning_rate": 1.1869654793500784e-06,
-      "loss": 0.0496,
-      "step": 860
-    },
-    {
-      "epoch": 0.015762013729977115,
-      "grad_norm": 0.5446186065673828,
-      "learning_rate": 1.1705240714107301e-06,
-      "loss": 0.0653,
-      "step": 861
-    },
-    {
-      "epoch": 0.015780320366132722,
-      "grad_norm": 0.2975693643093109,
-      "learning_rate": 1.1541902478491607e-06,
-      "loss": 0.0457,
-      "step": 862
-    },
-    {
-      "epoch": 0.01579862700228833,
-      "grad_norm": 0.48236411809921265,
-      "learning_rate": 1.1379642076878528e-06,
-      "loss": 0.0475,
-      "step": 863
-    },
-    {
-      "epoch": 0.015816933638443935,
-      "grad_norm": 0.7526642084121704,
-      "learning_rate": 1.1218461486359878e-06,
-      "loss": 0.0606,
-      "step": 864
-    },
-    {
-      "epoch": 0.01583524027459954,
-      "grad_norm": 0.2373325526714325,
-      "learning_rate": 1.1058362670870248e-06,
-      "loss": 0.0407,
-      "step": 865
-    },
-    {
-      "epoch": 0.015853546910755148,
-      "grad_norm": 0.5036309957504272,
-      "learning_rate": 1.0899347581163222e-06,
-      "loss": 0.0475,
-      "step": 866
-    },
-    {
-      "epoch": 0.015871853546910755,
-      "grad_norm": 0.3572957515716553,
-      "learning_rate": 1.0741418154787443e-06,
-      "loss": 0.0545,
-      "step": 867
-    },
-    {
-      "epoch": 0.01589016018306636,
-      "grad_norm": 0.5382295250892639,
-      "learning_rate": 1.058457631606319e-06,
-      "loss": 0.0514,
-      "step": 868
-    },
-    {
-      "epoch": 0.015908466819221968,
-      "grad_norm": 0.3161383867263794,
-      "learning_rate": 1.042882397605871e-06,
-      "loss": 0.059,
-      "step": 869
-    },
-    {
-      "epoch": 0.015926773455377574,
-      "grad_norm": 1.2964520454406738,
-      "learning_rate": 1.0274163032567165e-06,
-      "loss": 0.0727,
-      "step": 870
-    },
-    {
-      "epoch": 0.01594508009153318,
-      "grad_norm": 0.36181846261024475,
-      "learning_rate": 1.012059537008332e-06,
-      "loss": 0.0427,
-      "step": 871
-    },
-    {
-      "epoch": 0.015963386727688787,
-      "grad_norm": 0.8135401606559753,
-      "learning_rate": 9.968122859780648e-07,
-      "loss": 0.0729,
-      "step": 872
-    },
-    {
-      "epoch": 0.015981693363844394,
-      "grad_norm": 0.3147331178188324,
-      "learning_rate": 9.816747359488632e-07,
-      "loss": 0.0431,
-      "step": 873
-    },
-    {
-      "epoch": 0.016,
-      "grad_norm": 0.3021445870399475,
-      "learning_rate": 9.666470713669918e-07,
-      "loss": 0.044,
-      "step": 874
-    },
-    {
-      "epoch": 0.016018306636155607,
-      "grad_norm": 0.559781551361084,
-      "learning_rate": 9.517294753398066e-07,
-      "loss": 0.0495,
-      "step": 875
-    },
-    {
-      "epoch": 0.016036613272311213,
-      "grad_norm": 15.282647132873535,
-      "learning_rate": 9.369221296335007e-07,
-      "loss": 0.8154,
-      "step": 876
-    },
-    {
-      "epoch": 0.01605491990846682,
-      "grad_norm": 0.7999398708343506,
-      "learning_rate": 9.222252146709143e-07,
-      "loss": 0.0463,
-      "step": 877
-    },
-    {
-      "epoch": 0.016073226544622427,
-      "grad_norm": 0.23778311908245087,
-      "learning_rate": 9.076389095293148e-07,
-      "loss": 0.0428,
-      "step": 878
-    },
-    {
-      "epoch": 0.016091533180778033,
-      "grad_norm": 0.5569605231285095,
-      "learning_rate": 8.931633919382299e-07,
-      "loss": 0.0587,
-      "step": 879
-    },
-    {
-      "epoch": 0.01610983981693364,
-      "grad_norm": 0.22248607873916626,
-      "learning_rate": 8.787988382772705e-07,
-      "loss": 0.0359,
-      "step": 880
-    },
-    {
-      "epoch": 0.016128146453089246,
-      "grad_norm": 0.5969467163085938,
-      "learning_rate": 8.645454235739903e-07,
-      "loss": 0.0556,
-      "step": 881
-    },
-    {
-      "epoch": 0.016146453089244853,
-      "grad_norm": 0.8068287372589111,
-      "learning_rate": 8.504033215017527e-07,
-      "loss": 0.0759,
-      "step": 882
-    },
-    {
-      "epoch": 0.01616475972540046,
-      "grad_norm": 0.7625039219856262,
-      "learning_rate": 8.363727043776037e-07,
-      "loss": 0.0655,
-      "step": 883
-    },
-    {
-      "epoch": 0.016183066361556066,
-      "grad_norm": 0.3912850618362427,
-      "learning_rate": 8.224537431601886e-07,
-      "loss": 0.0501,
-      "step": 884
-    },
-    {
-      "epoch": 0.01620137299771167,
-      "grad_norm": 0.40553969144821167,
-      "learning_rate": 8.086466074476562e-07,
-      "loss": 0.0666,
-      "step": 885
-    },
-    {
-      "epoch": 0.016219679633867275,
-      "grad_norm": 0.4059741199016571,
-      "learning_rate": 7.949514654755963e-07,
-      "loss": 0.0581,
-      "step": 886
-    },
-    {
-      "epoch": 0.016237986270022882,
-      "grad_norm": 0.4925658702850342,
-      "learning_rate": 7.81368484114996e-07,
-      "loss": 0.0618,
-      "step": 887
-    },
-    {
-      "epoch": 0.01625629290617849,
-      "grad_norm": 0.6061264276504517,
-      "learning_rate": 7.678978288701911e-07,
-      "loss": 0.0675,
-      "step": 888
-    },
-    {
-      "epoch": 0.016274599542334095,
-      "grad_norm": 0.43979814648628235,
-      "learning_rate": 7.545396638768698e-07,
-      "loss": 0.0645,
-      "step": 889
-    },
-    {
-      "epoch": 0.0162929061784897,
-      "grad_norm": 0.40078485012054443,
-      "learning_rate": 7.412941519000527e-07,
-      "loss": 0.0528,
-      "step": 890
-    },
-    {
-      "epoch": 0.016311212814645308,
-      "grad_norm": 0.6207734942436218,
-      "learning_rate": 7.281614543321269e-07,
-      "loss": 0.0694,
-      "step": 891
-    },
-    {
-      "epoch": 0.016329519450800915,
-      "grad_norm": 0.32644596695899963,
-      "learning_rate": 7.151417311908648e-07,
-      "loss": 0.0429,
-      "step": 892
-    },
-    {
-      "epoch": 0.01634782608695652,
-      "grad_norm": 0.3338133692741394,
-      "learning_rate": 7.022351411174866e-07,
-      "loss": 0.0383,
-      "step": 893
-    },
-    {
-      "epoch": 0.016366132723112128,
-      "grad_norm": 0.41346168518066406,
-      "learning_rate": 6.894418413747183e-07,
-      "loss": 0.0638,
-      "step": 894
-    },
-    {
-      "epoch": 0.016384439359267734,
-      "grad_norm": 0.9189172983169556,
-      "learning_rate": 6.767619878448783e-07,
-      "loss": 0.0734,
-      "step": 895
-    },
-    {
-      "epoch": 0.01640274599542334,
-      "grad_norm": 0.35561859607696533,
-      "learning_rate": 6.641957350279838e-07,
-      "loss": 0.0484,
-      "step": 896
-    },
-    {
-      "epoch": 0.016421052631578947,
-      "grad_norm": 0.27437397837638855,
-      "learning_rate": 6.517432360398556e-07,
-      "loss": 0.0434,
-      "step": 897
-    },
-    {
-      "epoch": 0.016439359267734554,
-      "grad_norm": 0.29220953583717346,
-      "learning_rate": 6.394046426102673e-07,
-      "loss": 0.0464,
-      "step": 898
-    },
-    {
-      "epoch": 0.01645766590389016,
-      "grad_norm": 0.29715684056282043,
-      "learning_rate": 6.271801050810856e-07,
-      "loss": 0.0492,
-      "step": 899
-    },
-    {
-      "epoch": 0.016475972540045767,
-      "grad_norm": 0.39436575770378113,
-      "learning_rate": 6.150697724044407e-07,
-      "loss": 0.043,
-      "step": 900
-    },
-    {
-      "epoch": 0.016494279176201374,
-      "grad_norm": 0.4789891242980957,
-      "learning_rate": 6.030737921409169e-07,
-      "loss": 0.06,
-      "step": 901
-    },
-    {
-      "epoch": 0.01651258581235698,
-      "grad_norm": 0.46932849287986755,
-      "learning_rate": 5.911923104577455e-07,
-      "loss": 0.0529,
-      "step": 902
-    },
-    {
-      "epoch": 0.016530892448512587,
-      "grad_norm": 0.18611183762550354,
-      "learning_rate": 5.794254721270331e-07,
-      "loss": 0.0481,
-      "step": 903
-    },
-    {
-      "epoch": 0.016549199084668193,
-      "grad_norm": 0.5403739809989929,
-      "learning_rate": 5.677734205239904e-07,
-      "loss": 0.0521,
-      "step": 904
-    },
-    {
-      "epoch": 0.0165675057208238,
-      "grad_norm": 1.4559577703475952,
-      "learning_rate": 5.562362976251901e-07,
-      "loss": 0.0817,
-      "step": 905
-    },
-    {
-      "epoch": 0.016585812356979406,
-      "grad_norm": 0.48333314061164856,
-      "learning_rate": 5.448142440068316e-07,
-      "loss": 0.0531,
-      "step": 906
-    },
-    {
-      "epoch": 0.016604118993135013,
-      "grad_norm": 1.0288628339767456,
-      "learning_rate": 5.335073988430373e-07,
-      "loss": 0.0696,
-      "step": 907
-    },
-    {
-      "epoch": 0.01662242562929062,
-      "grad_norm": 0.41427233815193176,
-      "learning_rate": 5.223158999041444e-07,
-      "loss": 0.0527,
-      "step": 908
-    },
-    {
-      "epoch": 0.016640732265446226,
-      "grad_norm": 6.503000736236572,
-      "learning_rate": 5.112398835550348e-07,
-      "loss": 0.1117,
-      "step": 909
-    },
-    {
-      "epoch": 0.01665903890160183,
-      "grad_norm": 1.3399178981781006,
-      "learning_rate": 5.002794847534765e-07,
-      "loss": 0.081,
-      "step": 910
-    },
-    {
-      "epoch": 0.016677345537757435,
-      "grad_norm": 0.4355531632900238,
-      "learning_rate": 4.894348370484648e-07,
-      "loss": 0.0527,
-      "step": 911
-    },
-    {
-      "epoch": 0.016695652173913042,
-      "grad_norm": 0.5211818218231201,
-      "learning_rate": 4.787060725786141e-07,
-      "loss": 0.0596,
-      "step": 912
-    },
-    {
-      "epoch": 0.01671395881006865,
-      "grad_norm": 0.3041263222694397,
-      "learning_rate": 4.6809332207053083e-07,
-      "loss": 0.0462,
-      "step": 913
-    },
-    {
-      "epoch": 0.016732265446224255,
-      "grad_norm": 0.4235643744468689,
-      "learning_rate": 4.575967148372318e-07,
-      "loss": 0.0479,
-      "step": 914
-    },
-    {
-      "epoch": 0.01675057208237986,
-      "grad_norm": 0.3265625238418579,
-      "learning_rate": 4.4721637877656377e-07,
-      "loss": 0.059,
-      "step": 915
-    },
-    {
-      "epoch": 0.016768878718535468,
-      "grad_norm": 1.938266634941101,
-      "learning_rate": 4.3695244036964567e-07,
-      "loss": 0.1443,
-      "step": 916
-    },
-    {
-      "epoch": 0.016787185354691075,
-      "grad_norm": 0.4388870298862457,
-      "learning_rate": 4.268050246793276e-07,
-      "loss": 0.0476,
-      "step": 917
-    },
-    {
-      "epoch": 0.01680549199084668,
-      "grad_norm": 0.6746500134468079,
-      "learning_rate": 4.167742553486676e-07,
-      "loss": 0.0589,
-      "step": 918
-    },
-    {
-      "epoch": 0.016823798627002288,
-      "grad_norm": 0.4180094301700592,
-      "learning_rate": 4.068602545994249e-07,
-      "loss": 0.0572,
-      "step": 919
-    },
-    {
-      "epoch": 0.016842105263157894,
-      "grad_norm": 0.504507839679718,
-      "learning_rate": 3.9706314323056936e-07,
-      "loss": 0.0632,
-      "step": 920
-    },
-    {
-      "epoch": 0.0168604118993135,
-      "grad_norm": 0.44157758355140686,
-      "learning_rate": 3.8738304061681107e-07,
-      "loss": 0.0518,
-      "step": 921
-    },
-    {
-      "epoch": 0.016878718535469107,
-      "grad_norm": 0.42003756761550903,
-      "learning_rate": 3.7782006470714614e-07,
-      "loss": 0.0546,
-      "step": 922
-    },
-    {
-      "epoch": 0.016897025171624714,
-      "grad_norm": 0.4466019868850708,
-      "learning_rate": 3.68374332023419e-07,
-      "loss": 0.0409,
-      "step": 923
-    },
-    {
-      "epoch": 0.01691533180778032,
-      "grad_norm": 1.4708187580108643,
-      "learning_rate": 3.590459576589e-07,
-      "loss": 0.0949,
-      "step": 924
-    },
-    {
-      "epoch": 0.016933638443935927,
-      "grad_norm": 0.5779743790626526,
-      "learning_rate": 3.498350552768859e-07,
-      "loss": 0.0738,
-      "step": 925
-    },
-    {
-      "epoch": 0.016951945080091534,
-      "grad_norm": 0.42047813534736633,
-      "learning_rate": 3.4074173710931804e-07,
-      "loss": 0.061,
-      "step": 926
-    },
-    {
-      "epoch": 0.01697025171624714,
-      "grad_norm": 0.4672850966453552,
-      "learning_rate": 3.3176611395540625e-07,
-      "loss": 0.0483,
-      "step": 927
-    },
-    {
-      "epoch": 0.016988558352402747,
-      "grad_norm": 0.2697749137878418,
-      "learning_rate": 3.2290829518028867e-07,
-      "loss": 0.0528,
-      "step": 928
-    },
-    {
-      "epoch": 0.017006864988558353,
-      "grad_norm": 0.3218449354171753,
-      "learning_rate": 3.1416838871368925e-07,
-      "loss": 0.0463,
-      "step": 929
-    },
-    {
-      "epoch": 0.01702517162471396,
-      "grad_norm": 0.8932589292526245,
-      "learning_rate": 3.0554650104861137e-07,
-      "loss": 0.0657,
-      "step": 930
-    },
-    {
-      "epoch": 0.017043478260869566,
-      "grad_norm": 0.497283935546875,
-      "learning_rate": 2.970427372400353e-07,
-      "loss": 0.0557,
-      "step": 931
-    },
-    {
-      "epoch": 0.017061784897025173,
-      "grad_norm": 1.2191481590270996,
-      "learning_rate": 2.8865720090364037e-07,
-      "loss": 0.075,
-      "step": 932
-    },
-    {
-      "epoch": 0.01708009153318078,
-      "grad_norm": 0.5607751607894897,
-      "learning_rate": 2.8038999421453827e-07,
-      "loss": 0.0673,
-      "step": 933
-    },
-    {
-      "epoch": 0.017098398169336386,
-      "grad_norm": 0.5440273880958557,
-      "learning_rate": 2.7224121790603517e-07,
-      "loss": 0.0711,
-      "step": 934
-    },
-    {
-      "epoch": 0.017116704805491992,
-      "grad_norm": 0.32124122977256775,
-      "learning_rate": 2.6421097126839714e-07,
-      "loss": 0.0409,
-      "step": 935
-    },
-    {
-      "epoch": 0.017135011441647596,
-      "grad_norm": 0.5808600783348083,
-      "learning_rate": 2.5629935214764866e-07,
-      "loss": 0.0814,
-      "step": 936
-    },
-    {
-      "epoch": 0.017153318077803202,
-      "grad_norm": 0.4235082268714905,
-      "learning_rate": 2.4850645694436736e-07,
-      "loss": 0.0561,
-      "step": 937
-    },
-    {
-      "epoch": 0.01717162471395881,
-      "grad_norm": 0.6006746292114258,
-      "learning_rate": 2.4083238061252565e-07,
-      "loss": 0.0633,
-      "step": 938
-    },
-    {
-      "epoch": 0.017189931350114415,
-      "grad_norm": 0.45710572600364685,
-      "learning_rate": 2.332772166583208e-07,
-      "loss": 0.0625,
-      "step": 939
-    },
-    {
-      "epoch": 0.017208237986270022,
-      "grad_norm": 0.3344504237174988,
-      "learning_rate": 2.2584105713904126e-07,
-      "loss": 0.0549,
-      "step": 940
-    },
-    {
-      "epoch": 0.01722654462242563,
-      "grad_norm": 0.4938940405845642,
-      "learning_rate": 2.1852399266194312e-07,
-      "loss": 0.0501,
-      "step": 941
-    },
-    {
-      "epoch": 0.017244851258581235,
-      "grad_norm": 0.6497545838356018,
-      "learning_rate": 2.1132611238315004e-07,
-      "loss": 0.0528,
-      "step": 942
-    },
-    {
-      "epoch": 0.01726315789473684,
-      "grad_norm": 0.3821479380130768,
-      "learning_rate": 2.0424750400655947e-07,
-      "loss": 0.0537,
-      "step": 943
-    },
-    {
-      "epoch": 0.017281464530892448,
-      "grad_norm": 0.29370197653770447,
-      "learning_rate": 1.9728825378278248e-07,
-      "loss": 0.0527,
-      "step": 944
-    },
-    {
-      "epoch": 0.017299771167048054,
-      "grad_norm": 3.8220925331115723,
-      "learning_rate": 1.9044844650808468e-07,
-      "loss": 0.1688,
-      "step": 945
-    },
-    {
-      "epoch": 0.01731807780320366,
-      "grad_norm": 0.798316240310669,
-      "learning_rate": 1.8372816552336025e-07,
-      "loss": 0.0579,
-      "step": 946
-    },
-    {
-      "epoch": 0.017336384439359268,
-      "grad_norm": 0.3989717662334442,
-      "learning_rate": 1.7712749271311392e-07,
-      "loss": 0.0406,
-      "step": 947
-    },
-    {
-      "epoch": 0.017354691075514874,
-      "grad_norm": 2.389549493789673,
-      "learning_rate": 1.706465085044584e-07,
-      "loss": 0.0838,
-      "step": 948
-    },
-    {
-      "epoch": 0.01737299771167048,
-      "grad_norm": 0.9958433508872986,
-      "learning_rate": 1.6428529186614195e-07,
-      "loss": 0.0765,
-      "step": 949
-    },
-    {
-      "epoch": 0.017391304347826087,
-      "grad_norm": 0.2962484657764435,
-      "learning_rate": 1.580439203075812e-07,
-      "loss": 0.0511,
-      "step": 950
-    },
-    {
-      "epoch": 0.017409610983981694,
-      "grad_norm": 0.43318304419517517,
-      "learning_rate": 1.519224698779198e-07,
-      "loss": 0.0656,
-      "step": 951
-    },
-    {
-      "epoch": 0.0174279176201373,
-      "grad_norm": 0.5135586261749268,
-      "learning_rate": 1.4592101516509916e-07,
-      "loss": 0.0377,
-      "step": 952
-    },
-    {
-      "epoch": 0.017446224256292907,
-      "grad_norm": 0.552833080291748,
-      "learning_rate": 1.400396292949513e-07,
-      "loss": 0.0644,
-      "step": 953
-    },
-    {
-      "epoch": 0.017464530892448513,
-      "grad_norm": 0.32389238476753235,
-      "learning_rate": 1.3427838393030634e-07,
-      "loss": 0.0622,
-      "step": 954
-    },
-    {
-      "epoch": 0.01748283752860412,
-      "grad_norm": 1.5331782102584839,
-      "learning_rate": 1.2863734927012094e-07,
-      "loss": 0.1138,
-      "step": 955
-    },
-    {
-      "epoch": 0.017501144164759726,
-      "grad_norm": 0.3708762228488922,
-      "learning_rate": 1.231165940486234e-07,
-      "loss": 0.0415,
-      "step": 956
-    },
-    {
-      "epoch": 0.017519450800915333,
-      "grad_norm": 2.240487813949585,
-      "learning_rate": 1.1771618553447217e-07,
-      "loss": 0.134,
-      "step": 957
-    },
-    {
-      "epoch": 0.01753775743707094,
-      "grad_norm": 0.6137703657150269,
-      "learning_rate": 1.1243618952994195e-07,
-      "loss": 0.0712,
-      "step": 958
-    },
-    {
-      "epoch": 0.017556064073226546,
-      "grad_norm": 0.5043675303459167,
-      "learning_rate": 1.0727667037011668e-07,
-      "loss": 0.0493,
-      "step": 959
-    },
-    {
-      "epoch": 0.017574370709382153,
-      "grad_norm": 0.2610255777835846,
-      "learning_rate": 1.0223769092211012e-07,
-      "loss": 0.0375,
-      "step": 960
-    },
-    {
-      "epoch": 0.01759267734553776,
-      "grad_norm": 0.5602722764015198,
-      "learning_rate": 9.731931258429638e-08,
-      "loss": 0.0573,
-      "step": 961
-    },
-    {
-      "epoch": 0.017610983981693362,
-      "grad_norm": 0.6553979516029358,
-      "learning_rate": 9.252159528556404e-08,
-      "loss": 0.0851,
-      "step": 962
-    },
-    {
-      "epoch": 0.01762929061784897,
-      "grad_norm": 0.5403525829315186,
-      "learning_rate": 8.784459748458318e-08,
-      "loss": 0.0501,
-      "step": 963
-    },
-    {
-      "epoch": 0.017647597254004575,
-      "grad_norm": 0.37584608793258667,
-      "learning_rate": 8.328837616909612e-08,
-      "loss": 0.0371,
-      "step": 964
-    },
-    {
-      "epoch": 0.017665903890160182,
-      "grad_norm": 0.553325355052948,
-      "learning_rate": 7.885298685522235e-08,
-      "loss": 0.0524,
-      "step": 965
-    },
-    {
-      "epoch": 0.01768421052631579,
-      "grad_norm": 0.4244106411933899,
-      "learning_rate": 7.453848358678018e-08,
-      "loss": 0.0522,
-      "step": 966
-    },
-    {
-      "epoch": 0.017702517162471395,
-      "grad_norm": 0.6560888290405273,
-      "learning_rate": 7.034491893463059e-08,
-      "loss": 0.0587,
-      "step": 967
-    },
-    {
-      "epoch": 0.017720823798627,
-      "grad_norm": 0.5182058811187744,
-      "learning_rate": 6.627234399603554e-08,
-      "loss": 0.0656,
-      "step": 968
-    },
-    {
-      "epoch": 0.017739130434782608,
-      "grad_norm": 0.40440016984939575,
-      "learning_rate": 6.232080839403631e-08,
-      "loss": 0.0445,
-      "step": 969
-    },
-    {
-      "epoch": 0.017757437070938215,
-      "grad_norm": 0.41118109226226807,
-      "learning_rate": 5.849036027684607e-08,
-      "loss": 0.0526,
-      "step": 970
-    },
-    {
-      "epoch": 0.01777574370709382,
-      "grad_norm": 0.5555936098098755,
-      "learning_rate": 5.4781046317267103e-08,
-      "loss": 0.0518,
-      "step": 971
-    },
-    {
-      "epoch": 0.017794050343249428,
-      "grad_norm": 0.43469175696372986,
-      "learning_rate": 5.119291171211793e-08,
-      "loss": 0.0552,
-      "step": 972
-    },
-    {
-      "epoch": 0.017812356979405034,
-      "grad_norm": 17.94597053527832,
-      "learning_rate": 4.772600018168816e-08,
-      "loss": 0.335,
-      "step": 973
-    },
-    {
-      "epoch": 0.01783066361556064,
-      "grad_norm": 3.6954152584075928,
-      "learning_rate": 4.438035396920004e-08,
-      "loss": 0.1851,
-      "step": 974
-    },
-    {
-      "epoch": 0.017848970251716247,
-      "grad_norm": 0.9003121852874756,
-      "learning_rate": 4.115601384029666e-08,
-      "loss": 0.0821,
-      "step": 975
-    },
-    {
-      "epoch": 0.017867276887871854,
-      "grad_norm": 0.4816860556602478,
-      "learning_rate": 3.805301908254455e-08,
-      "loss": 0.056,
-      "step": 976
-    },
-    {
-      "epoch": 0.01788558352402746,
-      "grad_norm": 0.9474262595176697,
-      "learning_rate": 3.50714075049563e-08,
-      "loss": 0.0696,
-      "step": 977
-    },
-    {
-      "epoch": 0.017903890160183067,
-      "grad_norm": 3.800398349761963,
-      "learning_rate": 3.22112154375287e-08,
-      "loss": 0.1348,
-      "step": 978
-    },
-    {
-      "epoch": 0.017922196796338673,
-      "grad_norm": 0.3675464689731598,
-      "learning_rate": 2.947247773079753e-08,
-      "loss": 0.037,
-      "step": 979
-    },
-    {
-      "epoch": 0.01794050343249428,
-      "grad_norm": 1.0026291608810425,
-      "learning_rate": 2.6855227755419046e-08,
-      "loss": 0.0749,
-      "step": 980
-    },
-    {
-      "epoch": 0.017958810068649887,
-      "grad_norm": 0.6791110038757324,
-      "learning_rate": 2.4359497401758026e-08,
-      "loss": 0.0608,
-      "step": 981
-    },
-    {
-      "epoch": 0.017977116704805493,
-      "grad_norm": 0.5940614938735962,
-      "learning_rate": 2.1985317079500358e-08,
-      "loss": 0.0594,
-      "step": 982
-    },
-    {
-      "epoch": 0.0179954233409611,
-      "grad_norm": 0.45861196517944336,
-      "learning_rate": 1.973271571728441e-08,
-      "loss": 0.071,
-      "step": 983
-    },
-    {
-      "epoch": 0.018013729977116706,
-      "grad_norm": 0.32090625166893005,
-      "learning_rate": 1.7601720762346895e-08,
-      "loss": 0.0508,
-      "step": 984
-    },
-    {
-      "epoch": 0.018032036613272313,
-      "grad_norm": 0.4292711913585663,
-      "learning_rate": 1.5592358180189782e-08,
-      "loss": 0.0743,
-      "step": 985
-    },
-    {
-      "epoch": 0.01805034324942792,
-      "grad_norm": 0.9052971005439758,
-      "learning_rate": 1.370465245426167e-08,
-      "loss": 0.0801,
-      "step": 986
-    },
-    {
-      "epoch": 0.018068649885583522,
-      "grad_norm": 0.41669031977653503,
-      "learning_rate": 1.1938626585660252e-08,
-      "loss": 0.0591,
-      "step": 987
-    },
-    {
-      "epoch": 0.01808695652173913,
-      "grad_norm": 2.400463342666626,
-      "learning_rate": 1.0294302092853647e-08,
-      "loss": 0.191,
-      "step": 988
-    },
-    {
-      "epoch": 0.018105263157894735,
-      "grad_norm": 0.5987340211868286,
-      "learning_rate": 8.771699011416169e-09,
-      "loss": 0.0694,
-      "step": 989
-    },
-    {
-      "epoch": 0.018123569794050342,
-      "grad_norm": 0.3238181471824646,
-      "learning_rate": 7.370835893788508e-09,
-      "loss": 0.0532,
-      "step": 990
-    },
-    {
-      "epoch": 0.01814187643020595,
-      "grad_norm": 0.3499971926212311,
-      "learning_rate": 6.091729809042379e-09,
-      "loss": 0.0484,
-      "step": 991
-    },
-    {
-      "epoch": 0.018160183066361555,
-      "grad_norm": 0.5497595071792603,
-      "learning_rate": 4.9343963426840006e-09,
-      "loss": 0.0677,
-      "step": 992
-    },
-    {
-      "epoch": 0.01817848970251716,
-      "grad_norm": 0.41813722252845764,
-      "learning_rate": 3.898849596456477e-09,
-      "loss": 0.0472,
-      "step": 993
-    },
-    {
-      "epoch": 0.018196796338672768,
-      "grad_norm": 0.4581577777862549,
-      "learning_rate": 2.9851021881688314e-09,
-      "loss": 0.0529,
-      "step": 994
-    },
-    {
-      "epoch": 0.018215102974828375,
-      "grad_norm": 0.3413904309272766,
-      "learning_rate": 2.193165251545004e-09,
-      "loss": 0.0429,
-      "step": 995
-    },
-    {
-      "epoch": 0.01823340961098398,
-      "grad_norm": 0.37208008766174316,
-      "learning_rate": 1.5230484360873043e-09,
-      "loss": 0.0413,
-      "step": 996
-    },
-    {
-      "epoch": 0.018251716247139588,
-      "grad_norm": 1.2220373153686523,
-      "learning_rate": 9.74759906957612e-10,
-      "loss": 0.0827,
-      "step": 997
-    },
-    {
-      "epoch": 0.018270022883295194,
-      "grad_norm": 0.45487770438194275,
-      "learning_rate": 5.483063448785686e-10,
-      "loss": 0.052,
-      "step": 998
-    },
-    {
-      "epoch": 0.0182883295194508,
-      "grad_norm": 0.4343292713165283,
-      "learning_rate": 2.436929460525317e-10,
-      "loss": 0.0502,
-      "step": 999
-    },
-    {
-      "epoch": 0.018306636155606407,
-      "grad_norm": 0.4072776734828949,
-      "learning_rate": 6.092342209607083e-11,
-      "loss": 0.0507,
-      "step": 1000
     }
   ],
   "logging_steps": 1,