diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,19064 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 100,
+  "global_step": 2630,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9174869060516357,
+      "learning_rate": 7.59493670886076e-08,
+      "loss": 0.1337,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.145163059234619,
+      "learning_rate": 1.518987341772152e-07,
+      "loss": 0.1635,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.648493766784668,
+      "learning_rate": 2.278481012658228e-07,
+      "loss": 0.1938,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.465250015258789,
+      "learning_rate": 3.037974683544304e-07,
+      "loss": 0.1841,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.006047248840332,
+      "learning_rate": 3.79746835443038e-07,
+      "loss": 0.1522,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.263247966766357,
+      "learning_rate": 4.556962025316456e-07,
+      "loss": 0.1758,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.174136161804199,
+      "learning_rate": 5.316455696202532e-07,
+      "loss": 0.1764,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.27267599105835,
+      "learning_rate": 6.075949367088608e-07,
+      "loss": 0.1204,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.41044282913208,
+      "learning_rate": 6.835443037974684e-07,
+      "loss": 0.1632,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.198723316192627,
+      "learning_rate": 7.59493670886076e-07,
+      "loss": 0.1841,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.47627067565918,
+      "learning_rate": 8.354430379746835e-07,
+      "loss": 0.133,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.962169647216797,
+      "learning_rate": 9.113924050632912e-07,
+      "loss": 0.1501,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.1871724128723145,
+      "learning_rate": 9.873417721518986e-07,
+      "loss": 0.1398,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.34702730178833,
+      "learning_rate": 1.0632911392405063e-06,
+      "loss": 0.1425,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.515460968017578,
+      "learning_rate": 1.139240506329114e-06,
+      "loss": 0.2264,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9229485988616943,
+      "learning_rate": 1.2151898734177215e-06,
+      "loss": 0.0807,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.312820911407471,
+      "learning_rate": 1.2911392405063292e-06,
+      "loss": 0.1463,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.201635837554932,
+      "learning_rate": 1.3670886075949367e-06,
+      "loss": 0.1908,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.418750762939453,
+      "learning_rate": 1.4430379746835445e-06,
+      "loss": 0.1752,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.2834320068359375,
+      "learning_rate": 1.518987341772152e-06,
+      "loss": 0.126,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5940651893615723,
+      "learning_rate": 1.5949367088607595e-06,
+      "loss": 0.1116,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.570346832275391,
+      "learning_rate": 1.670886075949367e-06,
+      "loss": 0.1701,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.715400218963623,
+      "learning_rate": 1.7468354430379747e-06,
+      "loss": 0.1396,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.73274564743042,
+      "learning_rate": 1.8227848101265824e-06,
+      "loss": 0.0864,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.468443870544434,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 0.1829,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.871713161468506,
+      "learning_rate": 1.974683544303797e-06,
+      "loss": 0.1563,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.424239635467529,
+      "learning_rate": 2.050632911392405e-06,
+      "loss": 0.1663,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7690980434417725,
+      "learning_rate": 2.1265822784810126e-06,
+      "loss": 0.1014,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.508544445037842,
+      "learning_rate": 2.2025316455696203e-06,
+      "loss": 0.1843,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.534231662750244,
+      "learning_rate": 2.278481012658228e-06,
+      "loss": 0.1424,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.032422065734863,
+      "learning_rate": 2.3544303797468353e-06,
+      "loss": 0.1301,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.542902946472168,
+      "learning_rate": 2.430379746835443e-06,
+      "loss": 0.1658,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.088016986846924,
+      "learning_rate": 2.5063291139240508e-06,
+      "loss": 0.1622,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.6727375984191895,
+      "learning_rate": 2.5822784810126585e-06,
+      "loss": 0.1836,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.456547260284424,
+      "learning_rate": 2.6582278481012658e-06,
+      "loss": 0.1711,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.620174884796143,
+      "learning_rate": 2.7341772151898735e-06,
+      "loss": 0.1578,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.64295768737793,
+      "learning_rate": 2.8101265822784812e-06,
+      "loss": 0.1777,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.874703407287598,
+      "learning_rate": 2.886075949367089e-06,
+      "loss": 0.1401,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.705727577209473,
+      "learning_rate": 2.9620253164556962e-06,
+      "loss": 0.13,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5065159797668457,
+      "learning_rate": 3.037974683544304e-06,
+      "loss": 0.1192,
+      "step": 40
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.04217529296875,
+      "learning_rate": 3.1139240506329112e-06,
+      "loss": 0.1467,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.761617660522461,
+      "learning_rate": 3.189873417721519e-06,
+      "loss": 0.1391,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.766785621643066,
+      "learning_rate": 3.2658227848101267e-06,
+      "loss": 0.1049,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.26468563079834,
+      "learning_rate": 3.341772151898734e-06,
+      "loss": 0.1555,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.346744060516357,
+      "learning_rate": 3.417721518987342e-06,
+      "loss": 0.1159,
+      "step": 45
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.06795072555542,
+      "learning_rate": 3.4936708860759494e-06,
+      "loss": 0.0822,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.018447399139404,
+      "learning_rate": 3.5696202531645567e-06,
+      "loss": 0.1567,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.176362991333008,
+      "learning_rate": 3.645569620253165e-06,
+      "loss": 0.1537,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.070743560791016,
+      "learning_rate": 3.721518987341772e-06,
+      "loss": 0.1285,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.114931106567383,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 0.1708,
+      "step": 50
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.514534950256348,
+      "learning_rate": 3.8734177215189875e-06,
+      "loss": 0.1807,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.852578639984131,
+      "learning_rate": 3.949367088607594e-06,
+      "loss": 0.2006,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.3057756423950195,
+      "learning_rate": 4.025316455696203e-06,
+      "loss": 0.1892,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.195265769958496,
+      "learning_rate": 4.10126582278481e-06,
+      "loss": 0.2516,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.599604606628418,
+      "learning_rate": 4.177215189873418e-06,
+      "loss": 0.1495,
+      "step": 55
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.8647260665893555,
+      "learning_rate": 4.253164556962025e-06,
+      "loss": 0.1684,
+      "step": 56
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.897496223449707,
+      "learning_rate": 4.329113924050633e-06,
+      "loss": 0.1439,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 8.042851448059082,
+      "learning_rate": 4.405063291139241e-06,
+      "loss": 0.2363,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.7889251708984375,
+      "learning_rate": 4.481012658227848e-06,
+      "loss": 0.1332,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.9873738288879395,
+      "learning_rate": 4.556962025316456e-06,
+      "loss": 0.2498,
+      "step": 60
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.110446453094482,
+      "learning_rate": 4.632911392405064e-06,
+      "loss": 0.1361,
+      "step": 61
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.05630874633789,
+      "learning_rate": 4.708860759493671e-06,
+      "loss": 0.2296,
+      "step": 62
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.772387504577637,
+      "learning_rate": 4.784810126582278e-06,
+      "loss": 0.1284,
+      "step": 63
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.846549987792969,
+      "learning_rate": 4.860759493670886e-06,
+      "loss": 0.1568,
+      "step": 64
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.86070442199707,
+      "learning_rate": 4.936708860759494e-06,
+      "loss": 0.1497,
+      "step": 65
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.157785892486572,
+      "learning_rate": 5.0126582278481016e-06,
+      "loss": 0.1508,
+      "step": 66
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.129785060882568,
+      "learning_rate": 5.088607594936708e-06,
+      "loss": 0.1749,
+      "step": 67
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.110858917236328,
+      "learning_rate": 5.164556962025317e-06,
+      "loss": 0.1394,
+      "step": 68
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.702938556671143,
+      "learning_rate": 5.240506329113924e-06,
+      "loss": 0.2261,
+      "step": 69
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.428067684173584,
+      "learning_rate": 5.3164556962025316e-06,
+      "loss": 0.1759,
+      "step": 70
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.251758575439453,
+      "learning_rate": 5.392405063291139e-06,
+      "loss": 0.2168,
+      "step": 71
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.9722771644592285,
+      "learning_rate": 5.468354430379747e-06,
+      "loss": 0.1844,
+      "step": 72
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.163035869598389,
+      "learning_rate": 5.544303797468355e-06,
+      "loss": 0.231,
+      "step": 73
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.07459020614624,
+      "learning_rate": 5.6202531645569624e-06,
+      "loss": 0.1883,
+      "step": 74
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.986507415771484,
+      "learning_rate": 5.696202531645569e-06,
+      "loss": 0.2229,
+      "step": 75
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.926273822784424,
+      "learning_rate": 5.772151898734178e-06,
+      "loss": 0.2071,
+      "step": 76
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 8.529656410217285,
+      "learning_rate": 5.848101265822785e-06,
+      "loss": 0.2312,
+      "step": 77
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.946063041687012,
+      "learning_rate": 5.9240506329113924e-06,
+      "loss": 0.2208,
+      "step": 78
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.68623685836792,
+      "learning_rate": 6e-06,
+      "loss": 0.2525,
+      "step": 79
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 8.017003059387207,
+      "learning_rate": 5.999997725059553e-06,
+      "loss": 0.2401,
+      "step": 80
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.697761058807373,
+      "learning_rate": 5.999990900241661e-06,
+      "loss": 0.3185,
+      "step": 81
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 8.340555191040039,
+      "learning_rate": 5.999979525556675e-06,
+      "loss": 0.2007,
+      "step": 82
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.945841789245605,
+      "learning_rate": 5.9999636010218455e-06,
+      "loss": 0.3304,
+      "step": 83
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 8.234783172607422,
+      "learning_rate": 5.999943126661326e-06,
+      "loss": 0.2644,
+      "step": 84
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.193201065063477,
+      "learning_rate": 5.999918102506167e-06,
+      "loss": 0.2521,
+      "step": 85
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.385964870452881,
+      "learning_rate": 5.9998885285943205e-06,
+      "loss": 0.2493,
+      "step": 86
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.631782531738281,
+      "learning_rate": 5.999854404970641e-06,
+      "loss": 0.3257,
+      "step": 87
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 9.276768684387207,
+      "learning_rate": 5.9998157316868795e-06,
+      "loss": 0.3099,
+      "step": 88
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 8.009159088134766,
+      "learning_rate": 5.999772508801689e-06,
+      "loss": 0.214,
+      "step": 89
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 8.737258911132812,
+      "learning_rate": 5.999724736380623e-06,
+      "loss": 0.3798,
+      "step": 90
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.847580909729004,
+      "learning_rate": 5.999672414496135e-06,
+      "loss": 0.2329,
+      "step": 91
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.296484470367432,
+      "learning_rate": 5.999615543227577e-06,
+      "loss": 0.2493,
+      "step": 92
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.4983601570129395,
+      "learning_rate": 5.999554122661201e-06,
+      "loss": 0.2677,
+      "step": 93
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 10.352579116821289,
+      "learning_rate": 5.99948815289016e-06,
+      "loss": 0.3505,
+      "step": 94
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 9.0335054397583,
+      "learning_rate": 5.999417634014505e-06,
+      "loss": 0.2514,
+      "step": 95
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.6608662605285645,
+      "learning_rate": 5.9993425661411885e-06,
+      "loss": 0.2696,
+      "step": 96
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.929989814758301,
+      "learning_rate": 5.999262949384058e-06,
+      "loss": 0.2581,
+      "step": 97
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.588248252868652,
+      "learning_rate": 5.999178783863863e-06,
+      "loss": 0.3297,
+      "step": 98
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 7.850109577178955,
+      "learning_rate": 5.999090069708253e-06,
+      "loss": 0.2931,
+      "step": 99
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.107547760009766,
+      "learning_rate": 5.998996807051772e-06,
+      "loss": 0.2895,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "eval_blimp_filtered_avg": 0.7288059701492537,
+      "eval_blimp_filtered_std": 0.0048950820348148,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.01742250245418861,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "eval_vqa_filtered_avg": 0.29,
+      "eval_vqa_filtered_std": 0.045604802157206845,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "eval_winoground_filtered_avg": 0.52,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.332022666931152,
+      "learning_rate": 5.9988989960358655e-06,
+      "loss": 0.192,
+      "step": 101
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.271563529968262,
+      "learning_rate": 5.998796636808878e-06,
+      "loss": 0.2517,
+      "step": 102
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 7.382165431976318,
+      "learning_rate": 5.998689729526047e-06,
+      "loss": 0.267,
+      "step": 103
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 8.096851348876953,
+      "learning_rate": 5.998578274349514e-06,
+      "loss": 0.3436,
+      "step": 104
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 8.90272045135498,
+      "learning_rate": 5.998462271448312e-06,
+      "loss": 0.3391,
+      "step": 105
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 8.719164848327637,
+      "learning_rate": 5.998341720998377e-06,
+      "loss": 0.316,
+      "step": 106
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.664676666259766,
+      "learning_rate": 5.998216623182538e-06,
+      "loss": 0.2428,
+      "step": 107
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.710020542144775,
+      "learning_rate": 5.99808697819052e-06,
+      "loss": 0.3452,
+      "step": 108
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.475564002990723,
+      "learning_rate": 5.997952786218949e-06,
+      "loss": 0.317,
+      "step": 109
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.706830978393555,
+      "learning_rate": 5.997814047471343e-06,
+      "loss": 0.2801,
+      "step": 110
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.506506443023682,
+      "learning_rate": 5.997670762158116e-06,
+      "loss": 0.2865,
+      "step": 111
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.888820648193359,
+      "learning_rate": 5.997522930496579e-06,
+      "loss": 0.2572,
+      "step": 112
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.447524070739746,
+      "learning_rate": 5.997370552710937e-06,
+      "loss": 0.2927,
+      "step": 113
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.928182125091553,
+      "learning_rate": 5.9972136290322924e-06,
+      "loss": 0.2606,
+      "step": 114
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.703876972198486,
+      "learning_rate": 5.997052159698637e-06,
+      "loss": 0.2656,
+      "step": 115
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 9.299442291259766,
+      "learning_rate": 5.996886144954862e-06,
+      "loss": 0.4017,
+      "step": 116
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.311100959777832,
+      "learning_rate": 5.996715585052748e-06,
+      "loss": 0.2351,
+      "step": 117
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.121333599090576,
+      "learning_rate": 5.996540480250971e-06,
+      "loss": 0.2314,
+      "step": 118
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.909224033355713,
+      "learning_rate": 5.996360830815101e-06,
+      "loss": 0.2791,
+      "step": 119
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.696131706237793,
+      "learning_rate": 5.996176637017597e-06,
+      "loss": 0.2669,
+      "step": 120
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.528383731842041,
+      "learning_rate": 5.995987899137815e-06,
+      "loss": 0.3315,
+      "step": 121
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.79910945892334,
+      "learning_rate": 5.995794617461998e-06,
+      "loss": 0.2786,
+      "step": 122
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.45144510269165,
+      "learning_rate": 5.9955967922832834e-06,
+      "loss": 0.3272,
+      "step": 123
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.872969150543213,
+      "learning_rate": 5.995394423901696e-06,
+      "loss": 0.3024,
+      "step": 124
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 8.758634567260742,
+      "learning_rate": 5.995187512624155e-06,
+      "loss": 0.3921,
+      "step": 125
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 9.4890718460083,
+      "learning_rate": 5.994976058764468e-06,
+      "loss": 0.4107,
+      "step": 126
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.974234580993652,
+      "learning_rate": 5.994760062643331e-06,
+      "loss": 0.3257,
+      "step": 127
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.807271957397461,
+      "learning_rate": 5.994539524588329e-06,
+      "loss": 0.3348,
+      "step": 128
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 7.318472862243652,
+      "learning_rate": 5.994314444933936e-06,
+      "loss": 0.3229,
+      "step": 129
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 8.531023979187012,
+      "learning_rate": 5.9940848240215146e-06,
+      "loss": 0.361,
+      "step": 130
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 7.488437175750732,
+      "learning_rate": 5.993850662199314e-06,
+      "loss": 0.2541,
+      "step": 131
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 7.932046890258789,
+      "learning_rate": 5.99361195982247e-06,
+      "loss": 0.3692,
+      "step": 132
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.904051780700684,
+      "learning_rate": 5.993368717253005e-06,
+      "loss": 0.2923,
+      "step": 133
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.949537754058838,
+      "learning_rate": 5.9931209348598276e-06,
+      "loss": 0.2593,
+      "step": 134
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.493448734283447,
+      "learning_rate": 5.9928686130187315e-06,
+      "loss": 0.304,
+      "step": 135
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.794533729553223,
+      "learning_rate": 5.992611752112394e-06,
+      "loss": 0.2531,
+      "step": 136
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.292413234710693,
+      "learning_rate": 5.992350352530377e-06,
+      "loss": 0.2819,
+      "step": 137
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 8.067985534667969,
+      "learning_rate": 5.9920844146691274e-06,
+      "loss": 0.3062,
+      "step": 138
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.349759101867676,
+      "learning_rate": 5.991813938931972e-06,
+      "loss": 0.2568,
+      "step": 139
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.0964131355285645,
+      "learning_rate": 5.991538925729124e-06,
+      "loss": 0.2621,
+      "step": 140
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.693470001220703,
+      "learning_rate": 5.991259375477674e-06,
+      "loss": 0.2866,
+      "step": 141
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.278418064117432,
+      "learning_rate": 5.990975288601597e-06,
+      "loss": 0.2199,
+      "step": 142
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.202484130859375,
+      "learning_rate": 5.990686665531745e-06,
+      "loss": 0.3284,
+      "step": 143
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.192249774932861,
+      "learning_rate": 5.990393506705852e-06,
+      "loss": 0.2369,
+      "step": 144
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.918360710144043,
+      "learning_rate": 5.990095812568532e-06,
+      "loss": 0.2477,
+      "step": 145
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.642050743103027,
+      "learning_rate": 5.9897935835712745e-06,
+      "loss": 0.3102,
+      "step": 146
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.785747051239014,
+      "learning_rate": 5.9894868201724475e-06,
+      "loss": 0.2123,
+      "step": 147
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.0005879402160645,
+      "learning_rate": 5.989175522837298e-06,
+      "loss": 0.2182,
+      "step": 148
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.708948135375977,
+      "learning_rate": 5.988859692037948e-06,
+      "loss": 0.2636,
+      "step": 149
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.683592796325684,
+      "learning_rate": 5.988539328253395e-06,
+      "loss": 0.3229,
+      "step": 150
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.434894561767578,
+      "learning_rate": 5.9882144319695106e-06,
+      "loss": 0.243,
+      "step": 151
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.594137668609619,
+      "learning_rate": 5.9878850036790415e-06,
+      "loss": 0.2946,
+      "step": 152
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.602251052856445,
+      "learning_rate": 5.987551043881608e-06,
+      "loss": 0.2599,
+      "step": 153
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.790675163269043,
+      "learning_rate": 5.987212553083702e-06,
+      "loss": 0.2441,
+      "step": 154
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.013497352600098,
+      "learning_rate": 5.986869531798688e-06,
+      "loss": 0.2743,
+      "step": 155
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.549381256103516,
+      "learning_rate": 5.986521980546801e-06,
+      "loss": 0.2497,
+      "step": 156
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.2744622230529785,
+      "learning_rate": 5.986169899855147e-06,
+      "loss": 0.2117,
+      "step": 157
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 7.700123310089111,
+      "learning_rate": 5.985813290257701e-06,
+      "loss": 0.269,
+      "step": 158
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 7.242618560791016,
+      "learning_rate": 5.985452152295307e-06,
+      "loss": 0.3893,
+      "step": 159
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.780466079711914,
+      "learning_rate": 5.985086486515676e-06,
+      "loss": 0.234,
+      "step": 160
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.663663387298584,
+      "learning_rate": 5.984716293473387e-06,
+      "loss": 0.1698,
+      "step": 161
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.299855709075928,
+      "learning_rate": 5.9843415737298844e-06,
+      "loss": 0.2479,
+      "step": 162
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.320824146270752,
+      "learning_rate": 5.983962327853479e-06,
+      "loss": 0.2117,
+      "step": 163
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.215819835662842,
+      "learning_rate": 5.983578556419344e-06,
+      "loss": 0.2126,
+      "step": 164
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 7.234714984893799,
+      "learning_rate": 5.9831902600095185e-06,
+      "loss": 0.2736,
+      "step": 165
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.790665149688721,
+      "learning_rate": 5.982797439212904e-06,
+      "loss": 0.3217,
+      "step": 166
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.784404754638672,
+      "learning_rate": 5.982400094625261e-06,
+      "loss": 0.2763,
+      "step": 167
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.761205673217773,
+      "learning_rate": 5.981998226849215e-06,
+      "loss": 0.2778,
+      "step": 168
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.808236598968506,
+      "learning_rate": 5.981591836494248e-06,
+      "loss": 0.241,
+      "step": 169
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.418996334075928,
+      "learning_rate": 5.981180924176704e-06,
+      "loss": 0.253,
+      "step": 170
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.545456409454346,
+      "learning_rate": 5.980765490519783e-06,
+      "loss": 0.2021,
+      "step": 171
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.346310138702393,
+      "learning_rate": 5.980345536153543e-06,
+      "loss": 0.2147,
+      "step": 172
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.4729413986206055,
+      "learning_rate": 5.979921061714898e-06,
+      "loss": 0.2421,
+      "step": 173
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 7.273982048034668,
+      "learning_rate": 5.979492067847616e-06,
+      "loss": 0.3092,
+      "step": 174
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.104562759399414,
+      "learning_rate": 5.979058555202324e-06,
+      "loss": 0.2462,
+      "step": 175
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 7.050571441650391,
+      "learning_rate": 5.978620524436496e-06,
+      "loss": 0.2889,
+      "step": 176
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.387786388397217,
+      "learning_rate": 5.978177976214462e-06,
+      "loss": 0.1832,
+      "step": 177
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.717920780181885,
+      "learning_rate": 5.977730911207404e-06,
+      "loss": 0.2878,
+      "step": 178
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.989630222320557,
+      "learning_rate": 5.97727933009335e-06,
+      "loss": 0.2949,
+      "step": 179
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.186075210571289,
+      "learning_rate": 5.976823233557183e-06,
+      "loss": 0.2501,
+      "step": 180
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.795561790466309,
+      "learning_rate": 5.976362622290631e-06,
+      "loss": 0.2998,
+      "step": 181
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.796092987060547,
+      "learning_rate": 5.975897496992267e-06,
+      "loss": 0.3189,
+      "step": 182
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.8384504318237305,
+      "learning_rate": 5.9754278583675145e-06,
+      "loss": 0.2284,
+      "step": 183
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.71648645401001,
+      "learning_rate": 5.97495370712864e-06,
+      "loss": 0.2517,
+      "step": 184
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.3072052001953125,
+      "learning_rate": 5.974475043994753e-06,
+      "loss": 0.1879,
+      "step": 185
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.999935626983643,
+      "learning_rate": 5.973991869691809e-06,
+      "loss": 0.2282,
+      "step": 186
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.950961589813232,
+      "learning_rate": 5.973504184952602e-06,
+      "loss": 0.3093,
+      "step": 187
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.138779163360596,
+      "learning_rate": 5.973011990516766e-06,
+      "loss": 0.235,
+      "step": 188
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.682694435119629,
+      "learning_rate": 5.972515287130779e-06,
+      "loss": 0.2699,
+      "step": 189
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.761726379394531,
+      "learning_rate": 5.972014075547954e-06,
+      "loss": 0.2064,
+      "step": 190
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.754604339599609,
+      "learning_rate": 5.971508356528442e-06,
+      "loss": 0.2216,
+      "step": 191
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.822874069213867,
+      "learning_rate": 5.97099813083923e-06,
+      "loss": 0.2049,
+      "step": 192
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.247862815856934,
+      "learning_rate": 5.97048339925414e-06,
+      "loss": 0.2213,
+      "step": 193
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.7987141609191895,
+      "learning_rate": 5.969964162553827e-06,
+      "loss": 0.2003,
+      "step": 194
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.975724220275879,
+      "learning_rate": 5.96944042152578e-06,
+      "loss": 0.1822,
+      "step": 195
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.213257312774658,
+      "learning_rate": 5.96891217696432e-06,
+      "loss": 0.1838,
+      "step": 196
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.594130039215088,
+      "learning_rate": 5.968379429670596e-06,
+      "loss": 0.2185,
+      "step": 197
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.589231491088867,
+      "learning_rate": 5.967842180452585e-06,
+      "loss": 0.2773,
+      "step": 198
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.134817600250244,
+      "learning_rate": 5.967300430125097e-06,
+      "loss": 0.2112,
+      "step": 199
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.7841691970825195,
+      "learning_rate": 5.966754179509763e-06,
+      "loss": 0.205,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "eval_blimp_filtered_avg": 0.7280597014925373,
+      "eval_blimp_filtered_std": 0.0048949211443126365,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.01759947366308548,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "eval_vqa_filtered_avg": 0.33,
+      "eval_vqa_filtered_std": 0.047258156262526045,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.621191024780273,
+      "learning_rate": 5.966203429435042e-06,
+      "loss": 0.2607,
+      "step": 201
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.640338897705078,
+      "learning_rate": 5.9656481807362164e-06,
+      "loss": 0.1499,
+      "step": 202
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.798372268676758,
+      "learning_rate": 5.965088434255392e-06,
+      "loss": 0.2435,
+      "step": 203
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.283669471740723,
+      "learning_rate": 5.964524190841494e-06,
+      "loss": 0.2258,
+      "step": 204
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 7.879794120788574,
+      "learning_rate": 5.9639554513502696e-06,
+      "loss": 0.2955,
+      "step": 205
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.847405910491943,
+      "learning_rate": 5.963382216644286e-06,
+      "loss": 0.2254,
+      "step": 206
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.591825485229492,
+      "learning_rate": 5.9628044875929245e-06,
+      "loss": 0.2075,
+      "step": 207
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.625509738922119,
+      "learning_rate": 5.962222265072385e-06,
+      "loss": 0.1663,
+      "step": 208
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.57966423034668,
+      "learning_rate": 5.961635549965683e-06,
+      "loss": 0.2579,
+      "step": 209
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.772419452667236,
+      "learning_rate": 5.9610443431626445e-06,
+      "loss": 0.234,
+      "step": 210
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.683804035186768,
+      "learning_rate": 5.9604486455599115e-06,
+      "loss": 0.2416,
+      "step": 211
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.956271648406982,
+      "learning_rate": 5.959848458060934e-06,
+      "loss": 0.1756,
+      "step": 212
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.463733196258545,
+      "learning_rate": 5.959243781575972e-06,
+      "loss": 0.1996,
+      "step": 213
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.295461177825928,
+      "learning_rate": 5.958634617022096e-06,
+      "loss": 0.2013,
+      "step": 214
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 7.800921440124512,
+      "learning_rate": 5.958020965323181e-06,
+      "loss": 0.2498,
+      "step": 215
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.127272605895996,
+      "learning_rate": 5.957402827409906e-06,
+      "loss": 0.1933,
+      "step": 216
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.579216957092285,
+      "learning_rate": 5.956780204219757e-06,
+      "loss": 0.1979,
+      "step": 217
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.690545558929443,
+      "learning_rate": 5.956153096697021e-06,
+      "loss": 0.2403,
+      "step": 218
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.022470951080322,
+      "learning_rate": 5.955521505792786e-06,
+      "loss": 0.1782,
+      "step": 219
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.724915504455566,
+      "learning_rate": 5.95488543246494e-06,
+      "loss": 0.1887,
+      "step": 220
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.674469947814941,
+      "learning_rate": 5.954244877678168e-06,
+      "loss": 0.1978,
+      "step": 221
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.8040852546691895,
+      "learning_rate": 5.953599842403954e-06,
+      "loss": 0.2146,
+      "step": 222
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.939209938049316,
+      "learning_rate": 5.9529503276205765e-06,
+      "loss": 0.2857,
+      "step": 223
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.101150035858154,
+      "learning_rate": 5.952296334313105e-06,
+      "loss": 0.2045,
+      "step": 224
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.389438629150391,
+      "learning_rate": 5.951637863473404e-06,
+      "loss": 0.1431,
+      "step": 225
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.203385353088379,
+      "learning_rate": 5.950974916100128e-06,
+      "loss": 0.2202,
+      "step": 226
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.5857343673706055,
+      "learning_rate": 5.950307493198723e-06,
+      "loss": 0.1803,
+      "step": 227
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.834022521972656,
+      "learning_rate": 5.949635595781418e-06,
+      "loss": 0.2049,
+      "step": 228
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.941519737243652,
+      "learning_rate": 5.948959224867232e-06,
+      "loss": 0.1826,
+      "step": 229
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.212564468383789,
+      "learning_rate": 5.9482783814819665e-06,
+      "loss": 0.2161,
+      "step": 230
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.441732883453369,
+      "learning_rate": 5.947593066658208e-06,
+      "loss": 0.202,
+      "step": 231
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.872804164886475,
+      "learning_rate": 5.9469032814353225e-06,
+      "loss": 0.1895,
+      "step": 232
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.955694675445557,
+      "learning_rate": 5.946209026859458e-06,
+      "loss": 0.2135,
+      "step": 233
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.044952869415283,
+      "learning_rate": 5.945510303983539e-06,
+      "loss": 0.1683,
+      "step": 234
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.198512077331543,
+      "learning_rate": 5.944807113867266e-06,
+      "loss": 0.1777,
+      "step": 235
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.2534003257751465,
+      "learning_rate": 5.944099457577119e-06,
+      "loss": 0.1798,
+      "step": 236
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.885759353637695,
+      "learning_rate": 5.943387336186346e-06,
+      "loss": 0.2029,
+      "step": 237
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.019042015075684,
+      "learning_rate": 5.9426707507749715e-06,
+      "loss": 0.2047,
+      "step": 238
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.782329559326172,
+      "learning_rate": 5.941949702429787e-06,
+      "loss": 0.2167,
+      "step": 239
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.999391555786133,
+      "learning_rate": 5.941224192244354e-06,
+      "loss": 0.1682,
+      "step": 240
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.875488758087158,
+      "learning_rate": 5.940494221319002e-06,
+      "loss": 0.2877,
+      "step": 241
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.213758945465088,
+      "learning_rate": 5.939759790760823e-06,
+      "loss": 0.2036,
+      "step": 242
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.73900032043457,
+      "learning_rate": 5.939020901683676e-06,
+      "loss": 0.2387,
+      "step": 243
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.371427536010742,
+      "learning_rate": 5.938277555208178e-06,
+      "loss": 0.2093,
+      "step": 244
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.622852802276611,
+      "learning_rate": 5.937529752461709e-06,
+      "loss": 0.1566,
+      "step": 245
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.873755931854248,
+      "learning_rate": 5.936777494578408e-06,
+      "loss": 0.2085,
+      "step": 246
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.685544013977051,
+      "learning_rate": 5.936020782699168e-06,
+      "loss": 0.1869,
+      "step": 247
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.591735363006592,
+      "learning_rate": 5.935259617971638e-06,
+      "loss": 0.2711,
+      "step": 248
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.620720386505127,
+      "learning_rate": 5.934494001550224e-06,
+      "loss": 0.2322,
+      "step": 249
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.292867183685303,
+      "learning_rate": 5.933723934596079e-06,
+      "loss": 0.1884,
+      "step": 250
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.179301738739014,
+      "learning_rate": 5.932949418277107e-06,
+      "loss": 0.1658,
+      "step": 251
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.223184108734131,
+      "learning_rate": 5.93217045376796e-06,
+      "loss": 0.1898,
+      "step": 252
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.383382797241211,
+      "learning_rate": 5.931387042250036e-06,
+      "loss": 0.1925,
+      "step": 253
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.917378902435303,
+      "learning_rate": 5.93059918491148e-06,
+      "loss": 0.2629,
+      "step": 254
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.101637840270996,
+      "learning_rate": 5.929806882947177e-06,
+      "loss": 0.1978,
+      "step": 255
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.144092559814453,
+      "learning_rate": 5.929010137558754e-06,
+      "loss": 0.1764,
+      "step": 256
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.171979904174805,
+      "learning_rate": 5.928208949954574e-06,
+      "loss": 0.2182,
+      "step": 257
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.625131130218506,
+      "learning_rate": 5.927403321349742e-06,
+      "loss": 0.189,
+      "step": 258
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.811361312866211,
+      "learning_rate": 5.926593252966096e-06,
+      "loss": 0.1701,
+      "step": 259
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.489573955535889,
+      "learning_rate": 5.925778746032207e-06,
+      "loss": 0.2214,
+      "step": 260
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.30504846572876,
+      "learning_rate": 5.924959801783379e-06,
+      "loss": 0.1836,
+      "step": 261
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.819775581359863,
+      "learning_rate": 5.924136421461643e-06,
+      "loss": 0.2209,
+      "step": 262
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.9493513107299805,
+      "learning_rate": 5.923308606315761e-06,
+      "loss": 0.1882,
+      "step": 263
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.387957572937012,
+      "learning_rate": 5.92247635760122e-06,
+      "loss": 0.1942,
+      "step": 264
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.911758899688721,
+      "learning_rate": 5.921639676580232e-06,
+      "loss": 0.221,
+      "step": 265
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.313304424285889,
+      "learning_rate": 5.920798564521727e-06,
+      "loss": 0.1849,
+      "step": 266
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.95141077041626,
+      "learning_rate": 5.91995302270136e-06,
+      "loss": 0.1965,
+      "step": 267
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.044646263122559,
+      "learning_rate": 5.919103052401502e-06,
+      "loss": 0.252,
+      "step": 268
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.235661506652832,
+      "learning_rate": 5.918248654911243e-06,
+      "loss": 0.1196,
+      "step": 269
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.777853965759277,
+      "learning_rate": 5.917389831526383e-06,
+      "loss": 0.2069,
+      "step": 270
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.293239593505859,
+      "learning_rate": 5.916526583549437e-06,
+      "loss": 0.1723,
+      "step": 271
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.071403503417969,
+      "learning_rate": 5.915658912289629e-06,
+      "loss": 0.1916,
+      "step": 272
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.521011829376221,
+      "learning_rate": 5.914786819062896e-06,
+      "loss": 0.2021,
+      "step": 273
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.806651592254639,
+      "learning_rate": 5.913910305191876e-06,
+      "loss": 0.2036,
+      "step": 274
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.70013952255249,
+      "learning_rate": 5.913029372005913e-06,
+      "loss": 0.1682,
+      "step": 275
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.494007110595703,
+      "learning_rate": 5.912144020841055e-06,
+      "loss": 0.2294,
+      "step": 276
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.935788154602051,
+      "learning_rate": 5.911254253040049e-06,
+      "loss": 0.184,
+      "step": 277
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.4013895988464355,
+      "learning_rate": 5.910360069952341e-06,
+      "loss": 0.1763,
+      "step": 278
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.569982528686523,
+      "learning_rate": 5.909461472934073e-06,
+      "loss": 0.1799,
+      "step": 279
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.674581050872803,
+      "learning_rate": 5.9085584633480825e-06,
+      "loss": 0.2062,
+      "step": 280
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.409502983093262,
+      "learning_rate": 5.907651042563896e-06,
+      "loss": 0.1823,
+      "step": 281
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.826064109802246,
+      "learning_rate": 5.906739211957735e-06,
+      "loss": 0.1869,
+      "step": 282
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.245818614959717,
+      "learning_rate": 5.9058229729125045e-06,
+      "loss": 0.1562,
+      "step": 283
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.995058536529541,
+      "learning_rate": 5.904902326817798e-06,
+      "loss": 0.2239,
+      "step": 284
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.317048072814941,
+      "learning_rate": 5.9039772750698925e-06,
+      "loss": 0.2031,
+      "step": 285
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.205016136169434,
+      "learning_rate": 5.903047819071745e-06,
+      "loss": 0.2126,
+      "step": 286
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.21030855178833,
+      "learning_rate": 5.9021139602329955e-06,
+      "loss": 0.203,
+      "step": 287
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.603773593902588,
+      "learning_rate": 5.901175699969959e-06,
+      "loss": 0.2081,
+      "step": 288
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.86403751373291,
+      "learning_rate": 5.900233039705625e-06,
+      "loss": 0.2295,
+      "step": 289
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.677400588989258,
+      "learning_rate": 5.899285980869658e-06,
+      "loss": 0.2291,
+      "step": 290
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.17527437210083,
+      "learning_rate": 5.898334524898394e-06,
+      "loss": 0.1408,
+      "step": 291
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.426589488983154,
+      "learning_rate": 5.897378673234836e-06,
+      "loss": 0.1494,
+      "step": 292
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.473422527313232,
+      "learning_rate": 5.896418427328655e-06,
+      "loss": 0.2263,
+      "step": 293
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.745096206665039,
+      "learning_rate": 5.895453788636185e-06,
+      "loss": 0.2803,
+      "step": 294
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.608872413635254,
+      "learning_rate": 5.894484758620423e-06,
+      "loss": 0.1724,
+      "step": 295
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.932277679443359,
+      "learning_rate": 5.893511338751027e-06,
+      "loss": 0.1923,
+      "step": 296
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.495649814605713,
+      "learning_rate": 5.89253353050431e-06,
+      "loss": 0.1992,
+      "step": 297
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.911552906036377,
+      "learning_rate": 5.891551335363245e-06,
+      "loss": 0.2035,
+      "step": 298
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.5787649154663086,
+      "learning_rate": 5.8905647548174535e-06,
+      "loss": 0.1121,
+      "step": 299
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.282078742980957,
+      "learning_rate": 5.8895737903632114e-06,
+      "loss": 0.2065,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "eval_blimp_filtered_avg": 0.7270149253731343,
+      "eval_blimp_filtered_std": 0.0049016051387686735,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.01753892488357453,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "eval_vqa_filtered_avg": 0.23,
+      "eval_vqa_filtered_std": 0.04229525846816505,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.802579402923584,
+      "learning_rate": 5.888578443503442e-06,
+      "loss": 0.1819,
+      "step": 301
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.195893287658691,
+      "learning_rate": 5.887578715747714e-06,
+      "loss": 0.2814,
+      "step": 302
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.042200088500977,
+      "learning_rate": 5.886574608612242e-06,
+      "loss": 0.2061,
+      "step": 303
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.922192573547363,
+      "learning_rate": 5.885566123619883e-06,
+      "loss": 0.1757,
+      "step": 304
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.534228801727295,
+      "learning_rate": 5.884553262300132e-06,
+      "loss": 0.1902,
+      "step": 305
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.769097328186035,
+      "learning_rate": 5.883536026189122e-06,
+      "loss": 0.1931,
+      "step": 306
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.802075386047363,
+      "learning_rate": 5.882514416829619e-06,
+      "loss": 0.1791,
+      "step": 307
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.773422718048096,
+      "learning_rate": 5.8814884357710245e-06,
+      "loss": 0.1634,
+      "step": 308
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.283807754516602,
+      "learning_rate": 5.88045808456937e-06,
+      "loss": 0.1726,
+      "step": 309
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.87108039855957,
+      "learning_rate": 5.879423364787313e-06,
+      "loss": 0.2769,
+      "step": 310
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.6586761474609375,
+      "learning_rate": 5.878384277994136e-06,
+      "loss": 0.1384,
+      "step": 311
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.785504341125488,
+      "learning_rate": 5.877340825765749e-06,
+      "loss": 0.1804,
+      "step": 312
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.149272918701172,
+      "learning_rate": 5.876293009684678e-06,
+      "loss": 0.1553,
+      "step": 313
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.6378302574157715,
+      "learning_rate": 5.87524083134007e-06,
+      "loss": 0.1523,
+      "step": 314
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.7601141929626465,
+      "learning_rate": 5.8741842923276855e-06,
+      "loss": 0.1229,
+      "step": 315
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.569176197052002,
+      "learning_rate": 5.8731233942499015e-06,
+      "loss": 0.1277,
+      "step": 316
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.780937433242798,
+      "learning_rate": 5.872058138715704e-06,
+      "loss": 0.1126,
+      "step": 317
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.161221027374268,
+      "learning_rate": 5.870988527340689e-06,
+      "loss": 0.1256,
+      "step": 318
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.344121932983398,
+      "learning_rate": 5.869914561747057e-06,
+      "loss": 0.1604,
+      "step": 319
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.143471717834473,
+      "learning_rate": 5.868836243563613e-06,
+      "loss": 0.2869,
+      "step": 320
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.164417266845703,
+      "learning_rate": 5.867753574425765e-06,
+      "loss": 0.2275,
+      "step": 321
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.326757431030273,
+      "learning_rate": 5.866666555975516e-06,
+      "loss": 0.1487,
+      "step": 322
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.372414588928223,
+      "learning_rate": 5.86557518986147e-06,
+      "loss": 0.2216,
+      "step": 323
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.256776809692383,
+      "learning_rate": 5.86447947773882e-06,
+      "loss": 0.0857,
+      "step": 324
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.767634868621826,
+      "learning_rate": 5.863379421269354e-06,
+      "loss": 0.2346,
+      "step": 325
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.65963077545166,
+      "learning_rate": 5.862275022121447e-06,
+      "loss": 0.1668,
+      "step": 326
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.1615777015686035,
+      "learning_rate": 5.86116628197006e-06,
+      "loss": 0.206,
+      "step": 327
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.6640944480896,
+      "learning_rate": 5.8600532024967385e-06,
+      "loss": 0.1845,
+      "step": 328
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.7616994380950928,
+      "learning_rate": 5.858935785389609e-06,
+      "loss": 0.1432,
+      "step": 329
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.3153233528137207,
+      "learning_rate": 5.857814032343376e-06,
+      "loss": 0.0938,
+      "step": 330
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.44951057434082,
+      "learning_rate": 5.856687945059322e-06,
+      "loss": 0.1186,
+      "step": 331
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.504004001617432,
+      "learning_rate": 5.8555575252452985e-06,
+      "loss": 0.184,
+      "step": 332
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.861044883728027,
+      "learning_rate": 5.854422774615734e-06,
+      "loss": 0.1805,
+      "step": 333
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.7136921882629395,
+      "learning_rate": 5.853283694891619e-06,
+      "loss": 0.1309,
+      "step": 334
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.808126449584961,
+      "learning_rate": 5.852140287800513e-06,
+      "loss": 0.164,
+      "step": 335
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.805078983306885,
+      "learning_rate": 5.85099255507654e-06,
+      "loss": 0.1881,
+      "step": 336
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.177211284637451,
+      "learning_rate": 5.849840498460381e-06,
+      "loss": 0.2132,
+      "step": 337
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.94837760925293,
+      "learning_rate": 5.848684119699275e-06,
+      "loss": 0.1547,
+      "step": 338
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.27651834487915,
+      "learning_rate": 5.847523420547019e-06,
+      "loss": 0.1439,
+      "step": 339
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.234774112701416,
+      "learning_rate": 5.846358402763961e-06,
+      "loss": 0.1833,
+      "step": 340
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.461999416351318,
+      "learning_rate": 5.845189068116997e-06,
+      "loss": 0.178,
+      "step": 341
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.2992844581604,
+      "learning_rate": 5.844015418379572e-06,
+      "loss": 0.133,
+      "step": 342
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.415333271026611,
+      "learning_rate": 5.842837455331674e-06,
+      "loss": 0.1401,
+      "step": 343
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.321592807769775,
+      "learning_rate": 5.841655180759835e-06,
+      "loss": 0.154,
+      "step": 344
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.7492876052856445,
+      "learning_rate": 5.840468596457124e-06,
+      "loss": 0.1809,
+      "step": 345
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.936952590942383,
+      "learning_rate": 5.8392777042231455e-06,
+      "loss": 0.1082,
+      "step": 346
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.044267654418945,
+      "learning_rate": 5.838082505864041e-06,
+      "loss": 0.1746,
+      "step": 347
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.872096061706543,
+      "learning_rate": 5.836883003192478e-06,
+      "loss": 0.1727,
+      "step": 348
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.0386128425598145,
+      "learning_rate": 5.835679198027657e-06,
+      "loss": 0.1126,
+      "step": 349
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.224057197570801,
+      "learning_rate": 5.834471092195301e-06,
+      "loss": 0.1334,
+      "step": 350
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.778866767883301,
+      "learning_rate": 5.8332586875276535e-06,
+      "loss": 0.1038,
+      "step": 351
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.6813859939575195,
+      "learning_rate": 5.832041985863483e-06,
+      "loss": 0.1649,
+      "step": 352
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.387735843658447,
+      "learning_rate": 5.830820989048071e-06,
+      "loss": 0.1261,
+      "step": 353
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.038870811462402,
+      "learning_rate": 5.829595698933213e-06,
+      "loss": 0.1923,
+      "step": 354
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.366016387939453,
+      "learning_rate": 5.828366117377218e-06,
+      "loss": 0.202,
+      "step": 355
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.06331729888916,
+      "learning_rate": 5.8271322462449034e-06,
+      "loss": 0.136,
+      "step": 356
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.009934425354004,
+      "learning_rate": 5.825894087407591e-06,
+      "loss": 0.1023,
+      "step": 357
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.109599590301514,
+      "learning_rate": 5.824651642743105e-06,
+      "loss": 0.1543,
+      "step": 358
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.588776111602783,
+      "learning_rate": 5.823404914135772e-06,
+      "loss": 0.1698,
+      "step": 359
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.137748718261719,
+      "learning_rate": 5.822153903476412e-06,
+      "loss": 0.1236,
+      "step": 360
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.39997673034668,
+      "learning_rate": 5.8208986126623434e-06,
+      "loss": 0.1416,
+      "step": 361
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.199944019317627,
+      "learning_rate": 5.819639043597372e-06,
+      "loss": 0.29,
+      "step": 362
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.699921607971191,
+      "learning_rate": 5.818375198191798e-06,
+      "loss": 0.0942,
+      "step": 363
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.945669174194336,
+      "learning_rate": 5.8171070783624e-06,
+      "loss": 0.1715,
+      "step": 364
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.314772367477417,
+      "learning_rate": 5.815834686032444e-06,
+      "loss": 0.096,
+      "step": 365
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.8522868156433105,
+      "learning_rate": 5.814558023131674e-06,
+      "loss": 0.1827,
+      "step": 366
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.065451622009277,
+      "learning_rate": 5.813277091596311e-06,
+      "loss": 0.2297,
+      "step": 367
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.233007431030273,
+      "learning_rate": 5.811991893369053e-06,
+      "loss": 0.1799,
+      "step": 368
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.822303533554077,
+      "learning_rate": 5.810702430399063e-06,
+      "loss": 0.1231,
+      "step": 369
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.254659652709961,
+      "learning_rate": 5.8094087046419765e-06,
+      "loss": 0.2061,
+      "step": 370
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.60621976852417,
+      "learning_rate": 5.808110718059893e-06,
+      "loss": 0.2226,
+      "step": 371
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.88984489440918,
+      "learning_rate": 5.806808472621374e-06,
+      "loss": 0.1375,
+      "step": 372
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.166769027709961,
+      "learning_rate": 5.805501970301441e-06,
+      "loss": 0.1599,
+      "step": 373
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.204936981201172,
+      "learning_rate": 5.804191213081569e-06,
+      "loss": 0.2155,
+      "step": 374
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.352365970611572,
+      "learning_rate": 5.802876202949688e-06,
+      "loss": 0.1639,
+      "step": 375
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.661808967590332,
+      "learning_rate": 5.801556941900178e-06,
+      "loss": 0.1351,
+      "step": 376
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.030056476593018,
+      "learning_rate": 5.800233431933867e-06,
+      "loss": 0.1221,
+      "step": 377
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.73909854888916,
+      "learning_rate": 5.798905675058024e-06,
+      "loss": 0.1515,
+      "step": 378
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.265095233917236,
+      "learning_rate": 5.797573673286363e-06,
+      "loss": 0.1939,
+      "step": 379
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.3581647872924805,
+      "learning_rate": 5.796237428639032e-06,
+      "loss": 0.1097,
+      "step": 380
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.153365612030029,
+      "learning_rate": 5.7948969431426164e-06,
+      "loss": 0.1767,
+      "step": 381
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.248008728027344,
+      "learning_rate": 5.793552218830132e-06,
+      "loss": 0.2245,
+      "step": 382
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.567169189453125,
+      "learning_rate": 5.7922032577410256e-06,
+      "loss": 0.1347,
+      "step": 383
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.524489402770996,
+      "learning_rate": 5.790850061921166e-06,
+      "loss": 0.144,
+      "step": 384
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.236601829528809,
+      "learning_rate": 5.789492633422847e-06,
+      "loss": 0.1283,
+      "step": 385
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6281654834747314,
+      "learning_rate": 5.788130974304783e-06,
+      "loss": 0.0941,
+      "step": 386
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.6235711574554443,
+      "learning_rate": 5.786765086632102e-06,
+      "loss": 0.1201,
+      "step": 387
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.345760345458984,
+      "learning_rate": 5.785394972476344e-06,
+      "loss": 0.225,
+      "step": 388
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.20592737197876,
+      "learning_rate": 5.784020633915463e-06,
+      "loss": 0.1499,
+      "step": 389
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.902143478393555,
+      "learning_rate": 5.782642073033819e-06,
+      "loss": 0.2511,
+      "step": 390
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.782464504241943,
+      "learning_rate": 5.7812592919221705e-06,
+      "loss": 0.167,
+      "step": 391
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.78548526763916,
+      "learning_rate": 5.779872292677686e-06,
+      "loss": 0.1835,
+      "step": 392
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.649770736694336,
+      "learning_rate": 5.778481077403922e-06,
+      "loss": 0.1112,
+      "step": 393
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.852251052856445,
+      "learning_rate": 5.777085648210835e-06,
+      "loss": 0.1535,
+      "step": 394
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.420834064483643,
+      "learning_rate": 5.77568600721477e-06,
+      "loss": 0.1627,
+      "step": 395
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.201798915863037,
+      "learning_rate": 5.77428215653846e-06,
+      "loss": 0.1297,
+      "step": 396
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.464669227600098,
+      "learning_rate": 5.772874098311023e-06,
+      "loss": 0.1683,
+      "step": 397
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.31666374206543,
+      "learning_rate": 5.771461834667958e-06,
+      "loss": 0.2166,
+      "step": 398
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.333014965057373,
+      "learning_rate": 5.770045367751142e-06,
+      "loss": 0.204,
+      "step": 399
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.171229839324951,
+      "learning_rate": 5.768624699708827e-06,
+      "loss": 0.1626,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "eval_blimp_filtered_avg": 0.7291044776119403,
+      "eval_blimp_filtered_std": 0.004891633222931577,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "eval_blimp_supplement_avg": 0.7952586206896551,
+      "eval_blimp_supplement_std": 0.01768895435017441,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "eval_vqa_filtered_avg": 0.24,
+      "eval_vqa_filtered_std": 0.042923469599092816,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.950312376022339,
+      "learning_rate": 5.7671998326956375e-06,
+      "loss": 0.1296,
+      "step": 401
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.189338684082031,
+      "learning_rate": 5.765770768872563e-06,
+      "loss": 0.2145,
+      "step": 402
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.8659608364105225,
+      "learning_rate": 5.764337510406964e-06,
+      "loss": 0.1097,
+      "step": 403
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.442851543426514,
+      "learning_rate": 5.7629000594725536e-06,
+      "loss": 0.157,
+      "step": 404
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.575229167938232,
+      "learning_rate": 5.7614584182494135e-06,
+      "loss": 0.1425,
+      "step": 405
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.581119060516357,
+      "learning_rate": 5.7600125889239734e-06,
+      "loss": 0.1676,
+      "step": 406
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.397432804107666,
+      "learning_rate": 5.758562573689016e-06,
+      "loss": 0.1591,
+      "step": 407
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.398586750030518,
+      "learning_rate": 5.757108374743677e-06,
+      "loss": 0.1346,
+      "step": 408
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.74045991897583,
+      "learning_rate": 5.75564999429343e-06,
+      "loss": 0.1871,
+      "step": 409
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.79887056350708,
+      "learning_rate": 5.754187434550096e-06,
+      "loss": 0.1815,
+      "step": 410
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.850034713745117,
+      "learning_rate": 5.752720697731832e-06,
+      "loss": 0.1819,
+      "step": 411
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.531152725219727,
+      "learning_rate": 5.751249786063132e-06,
+      "loss": 0.1231,
+      "step": 412
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.814409017562866,
+      "learning_rate": 5.749774701774818e-06,
+      "loss": 0.094,
+      "step": 413
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6776301860809326,
+      "learning_rate": 5.748295447104044e-06,
+      "loss": 0.1197,
+      "step": 414
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.085291385650635,
+      "learning_rate": 5.746812024294288e-06,
+      "loss": 0.1143,
+      "step": 415
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.230500221252441,
+      "learning_rate": 5.745324435595348e-06,
+      "loss": 0.1452,
+      "step": 416
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.562592029571533,
+      "learning_rate": 5.743832683263342e-06,
+      "loss": 0.1413,
+      "step": 417
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.930176019668579,
+      "learning_rate": 5.7423367695607e-06,
+      "loss": 0.12,
+      "step": 418
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.076503753662109,
+      "learning_rate": 5.740836696756167e-06,
+      "loss": 0.2291,
+      "step": 419
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.78164529800415,
+      "learning_rate": 5.739332467124793e-06,
+      "loss": 0.1196,
+      "step": 420
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.002651214599609,
+      "learning_rate": 5.737824082947933e-06,
+      "loss": 0.1078,
+      "step": 421
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.15153169631958,
+      "learning_rate": 5.736311546513244e-06,
+      "loss": 0.076,
+      "step": 422
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.024011611938477,
+      "learning_rate": 5.734794860114679e-06,
+      "loss": 0.1517,
+      "step": 423
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.082853317260742,
+      "learning_rate": 5.733274026052486e-06,
+      "loss": 0.1244,
+      "step": 424
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.482311248779297,
+      "learning_rate": 5.731749046633201e-06,
+      "loss": 0.2449,
+      "step": 425
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.9700796604156494,
+      "learning_rate": 5.730219924169652e-06,
+      "loss": 0.1262,
+      "step": 426
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.051784992218018,
+      "learning_rate": 5.728686660980946e-06,
+      "loss": 0.1984,
+      "step": 427
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.3557305335998535,
+      "learning_rate": 5.727149259392469e-06,
+      "loss": 0.1492,
+      "step": 428
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.627366542816162,
+      "learning_rate": 5.725607721735889e-06,
+      "loss": 0.1759,
+      "step": 429
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.074859142303467,
+      "learning_rate": 5.724062050349143e-06,
+      "loss": 0.1528,
+      "step": 430
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.426521301269531,
+      "learning_rate": 5.722512247576436e-06,
+      "loss": 0.1361,
+      "step": 431
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.0030927658081055,
+      "learning_rate": 5.720958315768243e-06,
+      "loss": 0.1498,
+      "step": 432
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.144859313964844,
+      "learning_rate": 5.719400257281298e-06,
+      "loss": 0.1663,
+      "step": 433
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.290488243103027,
+      "learning_rate": 5.7178380744785935e-06,
+      "loss": 0.2085,
+      "step": 434
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4937922954559326,
+      "learning_rate": 5.716271769729379e-06,
+      "loss": 0.1001,
+      "step": 435
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.401832580566406,
+      "learning_rate": 5.714701345409155e-06,
+      "loss": 0.1387,
+      "step": 436
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6487908363342285,
+      "learning_rate": 5.713126803899668e-06,
+      "loss": 0.0918,
+      "step": 437
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.593252182006836,
+      "learning_rate": 5.711548147588911e-06,
+      "loss": 0.1862,
+      "step": 438
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.309147357940674,
+      "learning_rate": 5.7099653788711174e-06,
+      "loss": 0.1479,
+      "step": 439
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.18737268447876,
+      "learning_rate": 5.708378500146754e-06,
+      "loss": 0.1776,
+      "step": 440
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.764026641845703,
+      "learning_rate": 5.706787513822528e-06,
+      "loss": 0.1215,
+      "step": 441
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.649545669555664,
+      "learning_rate": 5.70519242231137e-06,
+      "loss": 0.0954,
+      "step": 442
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8633673191070557,
+      "learning_rate": 5.703593228032439e-06,
+      "loss": 0.1421,
+      "step": 443
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.142177104949951,
+      "learning_rate": 5.7019899334111155e-06,
+      "loss": 0.1115,
+      "step": 444
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.139797210693359,
+      "learning_rate": 5.700382540879e-06,
+      "loss": 0.125,
+      "step": 445
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.872627258300781,
+      "learning_rate": 5.698771052873908e-06,
+      "loss": 0.2354,
+      "step": 446
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.027861595153809,
+      "learning_rate": 5.697155471839865e-06,
+      "loss": 0.2227,
+      "step": 447
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.893920660018921,
+      "learning_rate": 5.695535800227104e-06,
+      "loss": 0.1188,
+      "step": 448
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.382118225097656,
+      "learning_rate": 5.693912040492063e-06,
+      "loss": 0.1772,
+      "step": 449
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.291519641876221,
+      "learning_rate": 5.692284195097382e-06,
+      "loss": 0.1863,
+      "step": 450
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4965579509735107,
+      "learning_rate": 5.6906522665118915e-06,
+      "loss": 0.0823,
+      "step": 451
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.500997066497803,
+      "learning_rate": 5.689016257210622e-06,
+      "loss": 0.16,
+      "step": 452
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.367881774902344,
+      "learning_rate": 5.6873761696747865e-06,
+      "loss": 0.1417,
+      "step": 453
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.258253574371338,
+      "learning_rate": 5.685732006391787e-06,
+      "loss": 0.1636,
+      "step": 454
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.6018218994140625,
+      "learning_rate": 5.684083769855206e-06,
+      "loss": 0.1278,
+      "step": 455
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5137486457824707,
+      "learning_rate": 5.682431462564805e-06,
+      "loss": 0.1034,
+      "step": 456
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.9920482635498047,
+      "learning_rate": 5.6807750870265145e-06,
+      "loss": 0.1639,
+      "step": 457
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.439868927001953,
+      "learning_rate": 5.679114645752441e-06,
+      "loss": 0.1304,
+      "step": 458
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.91691780090332,
+      "learning_rate": 5.677450141260852e-06,
+      "loss": 0.1529,
+      "step": 459
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.471147060394287,
+      "learning_rate": 5.675781576076183e-06,
+      "loss": 0.1837,
+      "step": 460
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.582545757293701,
+      "learning_rate": 5.674108952729023e-06,
+      "loss": 0.1952,
+      "step": 461
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7622666358947754,
+      "learning_rate": 5.672432273756118e-06,
+      "loss": 0.1369,
+      "step": 462
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.058557987213135,
+      "learning_rate": 5.670751541700365e-06,
+      "loss": 0.1718,
+      "step": 463
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.9158682823181152,
+      "learning_rate": 5.669066759110808e-06,
+      "loss": 0.1312,
+      "step": 464
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8238027095794678,
+      "learning_rate": 5.667377928542631e-06,
+      "loss": 0.1065,
+      "step": 465
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.093987941741943,
+      "learning_rate": 5.665685052557163e-06,
+      "loss": 0.1637,
+      "step": 466
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.187507152557373,
+      "learning_rate": 5.663988133721864e-06,
+      "loss": 0.1386,
+      "step": 467
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.436338424682617,
+      "learning_rate": 5.662287174610328e-06,
+      "loss": 0.1835,
+      "step": 468
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.633703708648682,
+      "learning_rate": 5.660582177802273e-06,
+      "loss": 0.1837,
+      "step": 469
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.094142913818359,
+      "learning_rate": 5.658873145883546e-06,
+      "loss": 0.2448,
+      "step": 470
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4598758220672607,
+      "learning_rate": 5.657160081446108e-06,
+      "loss": 0.0946,
+      "step": 471
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1474716663360596,
+      "learning_rate": 5.655442987088041e-06,
+      "loss": 0.0879,
+      "step": 472
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.290012836456299,
+      "learning_rate": 5.6537218654135365e-06,
+      "loss": 0.138,
+      "step": 473
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.8465044498443604,
+      "learning_rate": 5.651996719032893e-06,
+      "loss": 0.1436,
+      "step": 474
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.622616767883301,
+      "learning_rate": 5.650267550562514e-06,
+      "loss": 0.1301,
+      "step": 475
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.120974063873291,
+      "learning_rate": 5.6485343626249055e-06,
+      "loss": 0.1165,
+      "step": 476
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.7332305908203125,
+      "learning_rate": 5.646797157848663e-06,
+      "loss": 0.1925,
+      "step": 477
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.321225643157959,
+      "learning_rate": 5.64505593886848e-06,
+      "loss": 0.1657,
+      "step": 478
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.9361348152160645,
+      "learning_rate": 5.643310708325137e-06,
+      "loss": 0.1472,
+      "step": 479
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.339200019836426,
+      "learning_rate": 5.641561468865498e-06,
+      "loss": 0.1238,
+      "step": 480
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.1937947273254395,
+      "learning_rate": 5.6398082231425045e-06,
+      "loss": 0.1501,
+      "step": 481
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.7182111740112305,
+      "learning_rate": 5.638050973815178e-06,
+      "loss": 0.1521,
+      "step": 482
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.051305770874023,
+      "learning_rate": 5.636289723548609e-06,
+      "loss": 0.1097,
+      "step": 483
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.896054983139038,
+      "learning_rate": 5.634524475013958e-06,
+      "loss": 0.1012,
+      "step": 484
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7512948513031006,
+      "learning_rate": 5.63275523088845e-06,
+      "loss": 0.1308,
+      "step": 485
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.296629905700684,
+      "learning_rate": 5.6309819938553645e-06,
+      "loss": 0.1452,
+      "step": 486
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.228237628936768,
+      "learning_rate": 5.6292047666040446e-06,
+      "loss": 0.1362,
+      "step": 487
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.105273246765137,
+      "learning_rate": 5.627423551829878e-06,
+      "loss": 0.1293,
+      "step": 488
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.119396209716797,
+      "learning_rate": 5.625638352234305e-06,
+      "loss": 0.1443,
+      "step": 489
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7751595973968506,
+      "learning_rate": 5.623849170524807e-06,
+      "loss": 0.1239,
+      "step": 490
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.488837718963623,
+      "learning_rate": 5.622056009414905e-06,
+      "loss": 0.134,
+      "step": 491
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4268994331359863,
+      "learning_rate": 5.620258871624155e-06,
+      "loss": 0.1106,
+      "step": 492
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.150995254516602,
+      "learning_rate": 5.618457759878146e-06,
+      "loss": 0.1863,
+      "step": 493
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.818627119064331,
+      "learning_rate": 5.61665267690849e-06,
+      "loss": 0.1192,
+      "step": 494
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.185111045837402,
+      "learning_rate": 5.614843625452827e-06,
+      "loss": 0.1017,
+      "step": 495
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.9999732971191406,
+      "learning_rate": 5.613030608254813e-06,
+      "loss": 0.1184,
+      "step": 496
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5874311923980713,
+      "learning_rate": 5.611213628064119e-06,
+      "loss": 0.0898,
+      "step": 497
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.131513833999634,
+      "learning_rate": 5.609392687636424e-06,
+      "loss": 0.0994,
+      "step": 498
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.6310601234436035,
+      "learning_rate": 5.607567789733419e-06,
+      "loss": 0.182,
+      "step": 499
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8321306705474854,
+      "learning_rate": 5.605738937122789e-06,
+      "loss": 0.1142,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "eval_blimp_filtered_avg": 0.7295522388059702,
+      "eval_blimp_filtered_std": 0.004882217084715261,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.017697910699351765,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "eval_vqa_filtered_avg": 0.27,
+      "eval_vqa_filtered_std": 0.044619604333847394,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.860255718231201,
+      "learning_rate": 5.603906132578224e-06,
+      "loss": 0.096,
+      "step": 501
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8408665657043457,
+      "learning_rate": 5.602069378879404e-06,
+      "loss": 0.149,
+      "step": 502
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.65156626701355,
+      "learning_rate": 5.600228678812e-06,
+      "loss": 0.1264,
+      "step": 503
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.593369245529175,
+      "learning_rate": 5.598384035167666e-06,
+      "loss": 0.0912,
+      "step": 504
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.405891418457031,
+      "learning_rate": 5.596535450744039e-06,
+      "loss": 0.1704,
+      "step": 505
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.662251949310303,
+      "learning_rate": 5.594682928344732e-06,
+      "loss": 0.1652,
+      "step": 506
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.694492816925049,
+      "learning_rate": 5.592826470779331e-06,
+      "loss": 0.1408,
+      "step": 507
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.416996002197266,
+      "learning_rate": 5.590966080863388e-06,
+      "loss": 0.1246,
+      "step": 508
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0707900524139404,
+      "learning_rate": 5.589101761418423e-06,
+      "loss": 0.1479,
+      "step": 509
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.8668935298919678,
+      "learning_rate": 5.5872335152719105e-06,
+      "loss": 0.1353,
+      "step": 510
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.979825973510742,
+      "learning_rate": 5.585361345257284e-06,
+      "loss": 0.0773,
+      "step": 511
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.69364070892334,
+      "learning_rate": 5.5834852542139276e-06,
+      "loss": 0.0963,
+      "step": 512
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.492682456970215,
+      "learning_rate": 5.581605244987171e-06,
+      "loss": 0.2012,
+      "step": 513
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.42671012878418,
+      "learning_rate": 5.5797213204282865e-06,
+      "loss": 0.2056,
+      "step": 514
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8651270866394043,
+      "learning_rate": 5.577833483394487e-06,
+      "loss": 0.0952,
+      "step": 515
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4559667110443115,
+      "learning_rate": 5.5759417367489135e-06,
+      "loss": 0.1095,
+      "step": 516
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4004547595977783,
+      "learning_rate": 5.574046083360643e-06,
+      "loss": 0.1044,
+      "step": 517
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.0661115646362305,
+      "learning_rate": 5.572146526104673e-06,
+      "loss": 0.1935,
+      "step": 518
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.822204113006592,
+      "learning_rate": 5.570243067861924e-06,
+      "loss": 0.0963,
+      "step": 519
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.503787517547607,
+      "learning_rate": 5.568335711519231e-06,
+      "loss": 0.1517,
+      "step": 520
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.96858286857605,
+      "learning_rate": 5.566424459969344e-06,
+      "loss": 0.0891,
+      "step": 521
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9107344150543213,
+      "learning_rate": 5.564509316110917e-06,
+      "loss": 0.1196,
+      "step": 522
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9766218662261963,
+      "learning_rate": 5.562590282848509e-06,
+      "loss": 0.111,
+      "step": 523
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.729395627975464,
+      "learning_rate": 5.560667363092579e-06,
+      "loss": 0.1373,
+      "step": 524
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.698935031890869,
+      "learning_rate": 5.558740559759477e-06,
+      "loss": 0.116,
+      "step": 525
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.236121654510498,
+      "learning_rate": 5.5568098757714465e-06,
+      "loss": 0.0937,
+      "step": 526
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3916594982147217,
+      "learning_rate": 5.554875314056614e-06,
+      "loss": 0.0638,
+      "step": 527
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5698769092559814,
+      "learning_rate": 5.552936877548987e-06,
+      "loss": 0.0899,
+      "step": 528
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.778810739517212,
+      "learning_rate": 5.550994569188452e-06,
+      "loss": 0.0771,
+      "step": 529
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.8623769283294678,
+      "learning_rate": 5.549048391920767e-06,
+      "loss": 0.0714,
+      "step": 530
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.5749802589416504,
+      "learning_rate": 5.5470983486975555e-06,
+      "loss": 0.0774,
+      "step": 531
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.8800995349884033,
+      "learning_rate": 5.545144442476305e-06,
+      "loss": 0.0604,
+      "step": 532
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.098710775375366,
+      "learning_rate": 5.543186676220364e-06,
+      "loss": 0.0637,
+      "step": 533
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.800239086151123,
+      "learning_rate": 5.541225052898934e-06,
+      "loss": 0.0696,
+      "step": 534
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.033840656280518,
+      "learning_rate": 5.5392595754870635e-06,
+      "loss": 0.0896,
+      "step": 535
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.6125569343566895,
+      "learning_rate": 5.5372902469656505e-06,
+      "loss": 0.0829,
+      "step": 536
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.417607069015503,
+      "learning_rate": 5.53531707032143e-06,
+      "loss": 0.0549,
+      "step": 537
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.355532169342041,
+      "learning_rate": 5.533340048546977e-06,
+      "loss": 0.0461,
+      "step": 538
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.067363739013672,
+      "learning_rate": 5.531359184640695e-06,
+      "loss": 0.0782,
+      "step": 539
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.753206253051758,
+      "learning_rate": 5.5293744816068144e-06,
+      "loss": 0.0827,
+      "step": 540
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.7500948905944824,
+      "learning_rate": 5.527385942455392e-06,
+      "loss": 0.0658,
+      "step": 541
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.7404375076293945,
+      "learning_rate": 5.525393570202298e-06,
+      "loss": 0.0884,
+      "step": 542
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.700200080871582,
+      "learning_rate": 5.523397367869218e-06,
+      "loss": 0.0658,
+      "step": 543
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.8523411750793457,
+      "learning_rate": 5.521397338483647e-06,
+      "loss": 0.1066,
+      "step": 544
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.2977793216705322,
+      "learning_rate": 5.5193934850788825e-06,
+      "loss": 0.0741,
+      "step": 545
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.536318063735962,
+      "learning_rate": 5.517385810694025e-06,
+      "loss": 0.0921,
+      "step": 546
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.7307636737823486,
+      "learning_rate": 5.515374318373964e-06,
+      "loss": 0.0868,
+      "step": 547
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.518336296081543,
+      "learning_rate": 5.5133590111693855e-06,
+      "loss": 0.0729,
+      "step": 548
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.9823107719421387,
+      "learning_rate": 5.511339892136757e-06,
+      "loss": 0.0721,
+      "step": 549
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.480621576309204,
+      "learning_rate": 5.509316964338331e-06,
+      "loss": 0.0832,
+      "step": 550
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.876634359359741,
+      "learning_rate": 5.507290230842133e-06,
+      "loss": 0.0955,
+      "step": 551
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.8087801933288574,
+      "learning_rate": 5.505259694721962e-06,
+      "loss": 0.0716,
+      "step": 552
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.3994557857513428,
+      "learning_rate": 5.503225359057384e-06,
+      "loss": 0.0974,
+      "step": 553
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.1241281032562256,
+      "learning_rate": 5.501187226933727e-06,
+      "loss": 0.0601,
+      "step": 554
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.3551878929138184,
+      "learning_rate": 5.499145301442078e-06,
+      "loss": 0.0497,
+      "step": 555
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.755703926086426,
+      "learning_rate": 5.497099585679275e-06,
+      "loss": 0.1085,
+      "step": 556
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.884028911590576,
+      "learning_rate": 5.495050082747905e-06,
+      "loss": 0.0743,
+      "step": 557
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.239414930343628,
+      "learning_rate": 5.492996795756301e-06,
+      "loss": 0.0699,
+      "step": 558
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.7830045223236084,
+      "learning_rate": 5.4909397278185346e-06,
+      "loss": 0.0692,
+      "step": 559
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.8996188640594482,
+      "learning_rate": 5.488878882054408e-06,
+      "loss": 0.0737,
+      "step": 560
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.966162919998169,
+      "learning_rate": 5.486814261589456e-06,
+      "loss": 0.0674,
+      "step": 561
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.2157530784606934,
+      "learning_rate": 5.484745869554938e-06,
+      "loss": 0.034,
+      "step": 562
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.927938222885132,
+      "learning_rate": 5.482673709087833e-06,
+      "loss": 0.0668,
+      "step": 563
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.660696029663086,
+      "learning_rate": 5.480597783330837e-06,
+      "loss": 0.0858,
+      "step": 564
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.449079751968384,
+      "learning_rate": 5.478518095432352e-06,
+      "loss": 0.0508,
+      "step": 565
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.7392425537109375,
+      "learning_rate": 5.476434648546492e-06,
+      "loss": 0.1061,
+      "step": 566
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.404443740844727,
+      "learning_rate": 5.474347445833066e-06,
+      "loss": 0.0903,
+      "step": 567
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 3.237344980239868,
+      "learning_rate": 5.472256490457584e-06,
+      "loss": 0.0483,
+      "step": 568
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 3.680342197418213,
+      "learning_rate": 5.470161785591243e-06,
+      "loss": 0.0874,
+      "step": 569
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.4339213371276855,
+      "learning_rate": 5.468063334410931e-06,
+      "loss": 0.0535,
+      "step": 570
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.4135918617248535,
+      "learning_rate": 5.465961140099214e-06,
+      "loss": 0.082,
+      "step": 571
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.237689018249512,
+      "learning_rate": 5.463855205844338e-06,
+      "loss": 0.0813,
+      "step": 572
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.0851545333862305,
+      "learning_rate": 5.461745534840219e-06,
+      "loss": 0.0535,
+      "step": 573
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.692598819732666,
+      "learning_rate": 5.459632130286441e-06,
+      "loss": 0.0748,
+      "step": 574
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.349057197570801,
+      "learning_rate": 5.45751499538825e-06,
+      "loss": 0.1029,
+      "step": 575
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.8856046199798584,
+      "learning_rate": 5.455394133356552e-06,
+      "loss": 0.0479,
+      "step": 576
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 3.6796460151672363,
+      "learning_rate": 5.4532695474078994e-06,
+      "loss": 0.092,
+      "step": 577
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.295149803161621,
+      "learning_rate": 5.4511412407645005e-06,
+      "loss": 0.0804,
+      "step": 578
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.799071788787842,
+      "learning_rate": 5.449009216654201e-06,
+      "loss": 0.0574,
+      "step": 579
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 3.1664304733276367,
+      "learning_rate": 5.446873478310485e-06,
+      "loss": 0.0646,
+      "step": 580
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.697225093841553,
+      "learning_rate": 5.4447340289724725e-06,
+      "loss": 0.0803,
+      "step": 581
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.9874212741851807,
+      "learning_rate": 5.4425908718849095e-06,
+      "loss": 0.049,
+      "step": 582
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.170323133468628,
+      "learning_rate": 5.440444010298166e-06,
+      "loss": 0.0757,
+      "step": 583
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.637627601623535,
+      "learning_rate": 5.43829344746823e-06,
+      "loss": 0.0802,
+      "step": 584
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.3064918518066406,
+      "learning_rate": 5.436139186656703e-06,
+      "loss": 0.0891,
+      "step": 585
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.083530902862549,
+      "learning_rate": 5.433981231130796e-06,
+      "loss": 0.0665,
+      "step": 586
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.22542667388916,
+      "learning_rate": 5.431819584163321e-06,
+      "loss": 0.0688,
+      "step": 587
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.6223087310791016,
+      "learning_rate": 5.429654249032691e-06,
+      "loss": 0.0529,
+      "step": 588
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.714124917984009,
+      "learning_rate": 5.42748522902291e-06,
+      "loss": 0.08,
+      "step": 589
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.2308220863342285,
+      "learning_rate": 5.4253125274235755e-06,
+      "loss": 0.0702,
+      "step": 590
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.2309176921844482,
+      "learning_rate": 5.423136147529862e-06,
+      "loss": 0.071,
+      "step": 591
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.4496161937713623,
+      "learning_rate": 5.4209560926425285e-06,
+      "loss": 0.042,
+      "step": 592
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 3.8146402835845947,
+      "learning_rate": 5.4187723660679045e-06,
+      "loss": 0.0886,
+      "step": 593
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.023262023925781,
+      "learning_rate": 5.416584971117887e-06,
+      "loss": 0.106,
+      "step": 594
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 3.2619643211364746,
+      "learning_rate": 5.41439391110994e-06,
+      "loss": 0.0599,
+      "step": 595
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.5765275955200195,
+      "learning_rate": 5.412199189367082e-06,
+      "loss": 0.0537,
+      "step": 596
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 3.6795287132263184,
+      "learning_rate": 5.41000080921789e-06,
+      "loss": 0.0831,
+      "step": 597
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 5.061902046203613,
+      "learning_rate": 5.407798773996484e-06,
+      "loss": 0.1148,
+      "step": 598
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.920510768890381,
+      "learning_rate": 5.405593087042532e-06,
+      "loss": 0.0597,
+      "step": 599
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.1807656288146973,
+      "learning_rate": 5.403383751701236e-06,
+      "loss": 0.0555,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "eval_blimp_filtered_avg": 0.7292537313432835,
+      "eval_blimp_filtered_std": 0.004893817752662818,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "eval_blimp_supplement_avg": 0.7952586206896551,
+      "eval_blimp_supplement_std": 0.017722020555895352,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "eval_winoground_filtered_avg": 0.47,
+      "eval_winoground_filtered_std": 0.05016135580465919,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.231100559234619,
+      "learning_rate": 5.401170771323338e-06,
+      "loss": 0.0668,
+      "step": 601
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.0673625469207764,
+      "learning_rate": 5.398954149265099e-06,
+      "loss": 0.0615,
+      "step": 602
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.0846047401428223,
+      "learning_rate": 5.396733888888309e-06,
+      "loss": 0.0572,
+      "step": 603
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.7387914657592773,
+      "learning_rate": 5.394509993560276e-06,
+      "loss": 0.0517,
+      "step": 604
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 4.237431526184082,
+      "learning_rate": 5.392282466653819e-06,
+      "loss": 0.0955,
+      "step": 605
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.924257755279541,
+      "learning_rate": 5.390051311547265e-06,
+      "loss": 0.0629,
+      "step": 606
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.239067316055298,
+      "learning_rate": 5.387816531624445e-06,
+      "loss": 0.0816,
+      "step": 607
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.122785806655884,
+      "learning_rate": 5.385578130274685e-06,
+      "loss": 0.0497,
+      "step": 608
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.2336864471435547,
+      "learning_rate": 5.383336110892806e-06,
+      "loss": 0.0608,
+      "step": 609
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.889869451522827,
+      "learning_rate": 5.381090476879115e-06,
+      "loss": 0.0753,
+      "step": 610
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.2316601276397705,
+      "learning_rate": 5.378841231639401e-06,
+      "loss": 0.0545,
+      "step": 611
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.473172187805176,
+      "learning_rate": 5.376588378584928e-06,
+      "loss": 0.0568,
+      "step": 612
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.363525867462158,
+      "learning_rate": 5.374331921132437e-06,
+      "loss": 0.0661,
+      "step": 613
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.048120975494385,
+      "learning_rate": 5.3720718627041304e-06,
+      "loss": 0.0764,
+      "step": 614
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.9007833003997803,
+      "learning_rate": 5.369808206727674e-06,
+      "loss": 0.0588,
+      "step": 615
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.1275248527526855,
+      "learning_rate": 5.36754095663619e-06,
+      "loss": 0.0687,
+      "step": 616
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.3745083808898926,
+      "learning_rate": 5.36527011586825e-06,
+      "loss": 0.048,
+      "step": 617
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.4216058254241943,
+      "learning_rate": 5.362995687867873e-06,
+      "loss": 0.0565,
+      "step": 618
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.6002349853515625,
+      "learning_rate": 5.360717676084518e-06,
+      "loss": 0.0676,
+      "step": 619
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.95861554145813,
+      "learning_rate": 5.358436083973077e-06,
+      "loss": 0.0653,
+      "step": 620
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.010774612426758,
+      "learning_rate": 5.356150914993879e-06,
+      "loss": 0.0624,
+      "step": 621
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.677701473236084,
+      "learning_rate": 5.353862172612668e-06,
+      "loss": 0.0704,
+      "step": 622
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.704709053039551,
+      "learning_rate": 5.351569860300614e-06,
+      "loss": 0.0801,
+      "step": 623
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.5201194286346436,
+      "learning_rate": 5.349273981534301e-06,
+      "loss": 0.0647,
+      "step": 624
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.691957950592041,
+      "learning_rate": 5.3469745397957196e-06,
+      "loss": 0.0748,
+      "step": 625
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.878396511077881,
+      "learning_rate": 5.344671538572264e-06,
+      "loss": 0.0684,
+      "step": 626
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.697798728942871,
+      "learning_rate": 5.342364981356729e-06,
+      "loss": 0.0526,
+      "step": 627
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.9654245376586914,
+      "learning_rate": 5.340054871647303e-06,
+      "loss": 0.0618,
+      "step": 628
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.1462242603302,
+      "learning_rate": 5.337741212947558e-06,
+      "loss": 0.0682,
+      "step": 629
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.2505290508270264,
+      "learning_rate": 5.335424008766453e-06,
+      "loss": 0.0523,
+      "step": 630
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.2521116733551025,
+      "learning_rate": 5.3331032626183205e-06,
+      "loss": 0.0685,
+      "step": 631
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.4915645122528076,
+      "learning_rate": 5.3307789780228695e-06,
+      "loss": 0.0822,
+      "step": 632
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 4.367254734039307,
+      "learning_rate": 5.32845115850517e-06,
+      "loss": 0.0616,
+      "step": 633
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.406554937362671,
+      "learning_rate": 5.326119807595657e-06,
+      "loss": 0.0732,
+      "step": 634
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.4456567764282227,
+      "learning_rate": 5.323784928830121e-06,
+      "loss": 0.0621,
+      "step": 635
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.032163143157959,
+      "learning_rate": 5.321446525749701e-06,
+      "loss": 0.0548,
+      "step": 636
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.0345816612243652,
+      "learning_rate": 5.319104601900882e-06,
+      "loss": 0.0588,
+      "step": 637
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 4.1745285987854,
+      "learning_rate": 5.316759160835489e-06,
+      "loss": 0.0777,
+      "step": 638
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.12252140045166,
+      "learning_rate": 5.314410206110681e-06,
+      "loss": 0.0866,
+      "step": 639
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 3.454665184020996,
+      "learning_rate": 5.312057741288948e-06,
+      "loss": 0.0806,
+      "step": 640
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 4.29226541519165,
+      "learning_rate": 5.3097017699380984e-06,
+      "loss": 0.1048,
+      "step": 641
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 4.312617778778076,
+      "learning_rate": 5.3073422956312655e-06,
+      "loss": 0.0999,
+      "step": 642
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.742352247238159,
+      "learning_rate": 5.304979321946888e-06,
+      "loss": 0.0593,
+      "step": 643
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.474982976913452,
+      "learning_rate": 5.302612852468717e-06,
+      "loss": 0.0445,
+      "step": 644
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 3.145623207092285,
+      "learning_rate": 5.300242890785805e-06,
+      "loss": 0.0689,
+      "step": 645
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.1722779273986816,
+      "learning_rate": 5.297869440492499e-06,
+      "loss": 0.0443,
+      "step": 646
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.736063241958618,
+      "learning_rate": 5.2954925051884375e-06,
+      "loss": 0.0711,
+      "step": 647
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.872098922729492,
+      "learning_rate": 5.293112088478544e-06,
+      "loss": 0.0858,
+      "step": 648
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 3.408586263656616,
+      "learning_rate": 5.290728193973024e-06,
+      "loss": 0.1029,
+      "step": 649
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.3267316818237305,
+      "learning_rate": 5.288340825287356e-06,
+      "loss": 0.0809,
+      "step": 650
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.632853031158447,
+      "learning_rate": 5.285949986042288e-06,
+      "loss": 0.0941,
+      "step": 651
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.9303972721099854,
+      "learning_rate": 5.283555679863828e-06,
+      "loss": 0.0649,
+      "step": 652
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.9880752563476562,
+      "learning_rate": 5.28115791038325e-06,
+      "loss": 0.0693,
+      "step": 653
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 3.187875747680664,
+      "learning_rate": 5.278756681237074e-06,
+      "loss": 0.0609,
+      "step": 654
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 3.20235013961792,
+      "learning_rate": 5.276351996067069e-06,
+      "loss": 0.074,
+      "step": 655
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.702887773513794,
+      "learning_rate": 5.2739438585202435e-06,
+      "loss": 0.061,
+      "step": 656
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 3.799008369445801,
+      "learning_rate": 5.271532272248846e-06,
+      "loss": 0.0938,
+      "step": 657
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.965301036834717,
+      "learning_rate": 5.269117240910353e-06,
+      "loss": 0.0609,
+      "step": 658
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 3.880682945251465,
+      "learning_rate": 5.266698768167466e-06,
+      "loss": 0.0644,
+      "step": 659
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 3.0273046493530273,
+      "learning_rate": 5.264276857688105e-06,
+      "loss": 0.0436,
+      "step": 660
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.1428825855255127,
+      "learning_rate": 5.261851513145406e-06,
+      "loss": 0.0643,
+      "step": 661
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.027501344680786,
+      "learning_rate": 5.259422738217712e-06,
+      "loss": 0.0555,
+      "step": 662
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.8350963592529297,
+      "learning_rate": 5.256990536588567e-06,
+      "loss": 0.072,
+      "step": 663
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.1238343715667725,
+      "learning_rate": 5.254554911946714e-06,
+      "loss": 0.06,
+      "step": 664
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.2324893474578857,
+      "learning_rate": 5.25211586798609e-06,
+      "loss": 0.0707,
+      "step": 665
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.5057523250579834,
+      "learning_rate": 5.24967340840581e-06,
+      "loss": 0.0844,
+      "step": 666
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.5193774700164795,
+      "learning_rate": 5.247227536910178e-06,
+      "loss": 0.0412,
+      "step": 667
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.5131397247314453,
+      "learning_rate": 5.244778257208667e-06,
+      "loss": 0.071,
+      "step": 668
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.6189281940460205,
+      "learning_rate": 5.24232557301592e-06,
+      "loss": 0.0659,
+      "step": 669
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.021785020828247,
+      "learning_rate": 5.239869488051745e-06,
+      "loss": 0.0594,
+      "step": 670
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.6992058753967285,
+      "learning_rate": 5.237410006041107e-06,
+      "loss": 0.0401,
+      "step": 671
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 3.90799880027771,
+      "learning_rate": 5.234947130714122e-06,
+      "loss": 0.0743,
+      "step": 672
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 3.9819931983947754,
+      "learning_rate": 5.232480865806053e-06,
+      "loss": 0.0762,
+      "step": 673
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 3.1596999168395996,
+      "learning_rate": 5.230011215057304e-06,
+      "loss": 0.0602,
+      "step": 674
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 5.315537929534912,
+      "learning_rate": 5.227538182213414e-06,
+      "loss": 0.0772,
+      "step": 675
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.5624639987945557,
+      "learning_rate": 5.225061771025052e-06,
+      "loss": 0.0355,
+      "step": 676
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.7240264415740967,
+      "learning_rate": 5.22258198524801e-06,
+      "loss": 0.0544,
+      "step": 677
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 4.807624816894531,
+      "learning_rate": 5.220098828643197e-06,
+      "loss": 0.0721,
+      "step": 678
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 3.1555521488189697,
+      "learning_rate": 5.2176123049766345e-06,
+      "loss": 0.0726,
+      "step": 679
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.8005642890930176,
+      "learning_rate": 5.215122418019454e-06,
+      "loss": 0.0627,
+      "step": 680
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 3.1825637817382812,
+      "learning_rate": 5.212629171547883e-06,
+      "loss": 0.0767,
+      "step": 681
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.6690762042999268,
+      "learning_rate": 5.210132569343247e-06,
+      "loss": 0.0486,
+      "step": 682
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.1990206241607666,
+      "learning_rate": 5.20763261519196e-06,
+      "loss": 0.0438,
+      "step": 683
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.6149675846099854,
+      "learning_rate": 5.20512931288552e-06,
+      "loss": 0.0565,
+      "step": 684
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.3237783908843994,
+      "learning_rate": 5.202622666220503e-06,
+      "loss": 0.08,
+      "step": 685
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.6199722290039062,
+      "learning_rate": 5.200112678998557e-06,
+      "loss": 0.088,
+      "step": 686
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.495051383972168,
+      "learning_rate": 5.197599355026396e-06,
+      "loss": 0.0641,
+      "step": 687
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.131354808807373,
+      "learning_rate": 5.1950826981157955e-06,
+      "loss": 0.0596,
+      "step": 688
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.942925214767456,
+      "learning_rate": 5.192562712083584e-06,
+      "loss": 0.0482,
+      "step": 689
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.033991813659668,
+      "learning_rate": 5.190039400751641e-06,
+      "loss": 0.0743,
+      "step": 690
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.4703967571258545,
+      "learning_rate": 5.18751276794689e-06,
+      "loss": 0.098,
+      "step": 691
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.7910807132720947,
+      "learning_rate": 5.184982817501288e-06,
+      "loss": 0.081,
+      "step": 692
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.391430139541626,
+      "learning_rate": 5.1824495532518265e-06,
+      "loss": 0.067,
+      "step": 693
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.375948905944824,
+      "learning_rate": 5.179912979040523e-06,
+      "loss": 0.0468,
+      "step": 694
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.732227325439453,
+      "learning_rate": 5.177373098714414e-06,
+      "loss": 0.0552,
+      "step": 695
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.723480224609375,
+      "learning_rate": 5.174829916125552e-06,
+      "loss": 0.0409,
+      "step": 696
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.408365488052368,
+      "learning_rate": 5.172283435130994e-06,
+      "loss": 0.0636,
+      "step": 697
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.0623438358306885,
+      "learning_rate": 5.169733659592804e-06,
+      "loss": 0.0743,
+      "step": 698
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.511911153793335,
+      "learning_rate": 5.167180593378037e-06,
+      "loss": 0.0628,
+      "step": 699
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.2863922119140625,
+      "learning_rate": 5.164624240358746e-06,
+      "loss": 0.1008,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "eval_blimp_filtered_avg": 0.7292537313432835,
+      "eval_blimp_filtered_std": 0.0048922541884969676,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "eval_blimp_supplement_avg": 0.8017241379310345,
+      "eval_blimp_supplement_std": 0.017464477050379356,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "eval_vqa_filtered_avg": 0.28,
+      "eval_vqa_filtered_std": 0.045126085985421276,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "eval_winoground_filtered_avg": 0.45,
+      "eval_winoground_filtered_std": 0.05,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.417008876800537,
+      "learning_rate": 5.162064604411962e-06,
+      "loss": 0.0574,
+      "step": 701
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.310934543609619,
+      "learning_rate": 5.159501689419701e-06,
+      "loss": 0.0453,
+      "step": 702
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 4.085020542144775,
+      "learning_rate": 5.156935499268944e-06,
+      "loss": 0.0985,
+      "step": 703
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.5779457092285156,
+      "learning_rate": 5.154366037851649e-06,
+      "loss": 0.0874,
+      "step": 704
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.4925928115844727,
+      "learning_rate": 5.1517933090647274e-06,
+      "loss": 0.0594,
+      "step": 705
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.114696979522705,
+      "learning_rate": 5.149217316810052e-06,
+      "loss": 0.0875,
+      "step": 706
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.090488910675049,
+      "learning_rate": 5.1466380649944395e-06,
+      "loss": 0.0478,
+      "step": 707
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.9231483936309814,
+      "learning_rate": 5.144055557529654e-06,
+      "loss": 0.0722,
+      "step": 708
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 4.116023063659668,
+      "learning_rate": 5.141469798332395e-06,
+      "loss": 0.0845,
+      "step": 709
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.6202545166015625,
+      "learning_rate": 5.138880791324295e-06,
+      "loss": 0.0407,
+      "step": 710
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.550189971923828,
+      "learning_rate": 5.136288540431913e-06,
+      "loss": 0.0751,
+      "step": 711
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.3703055381774902,
+      "learning_rate": 5.133693049586725e-06,
+      "loss": 0.0613,
+      "step": 712
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.4189391136169434,
+      "learning_rate": 5.131094322725124e-06,
+      "loss": 0.0659,
+      "step": 713
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.6138412952423096,
+      "learning_rate": 5.128492363788407e-06,
+      "loss": 0.094,
+      "step": 714
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.911017417907715,
+      "learning_rate": 5.125887176722777e-06,
+      "loss": 0.0749,
+      "step": 715
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.5934674739837646,
+      "learning_rate": 5.1232787654793305e-06,
+      "loss": 0.0629,
+      "step": 716
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.3743412494659424,
+      "learning_rate": 5.1206671340140535e-06,
+      "loss": 0.0587,
+      "step": 717
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.8414058685302734,
+      "learning_rate": 5.118052286287818e-06,
+      "loss": 0.0544,
+      "step": 718
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 5.23115348815918,
+      "learning_rate": 5.115434226266372e-06,
+      "loss": 0.1176,
+      "step": 719
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.3688852787017822,
+      "learning_rate": 5.112812957920336e-06,
+      "loss": 0.0557,
+      "step": 720
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.833150625228882,
+      "learning_rate": 5.110188485225195e-06,
+      "loss": 0.081,
+      "step": 721
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.3594491481781006,
+      "learning_rate": 5.107560812161298e-06,
+      "loss": 0.0526,
+      "step": 722
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.5965960025787354,
+      "learning_rate": 5.104929942713842e-06,
+      "loss": 0.0779,
+      "step": 723
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.64615535736084,
+      "learning_rate": 5.102295880872876e-06,
+      "loss": 0.0804,
+      "step": 724
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.924147367477417,
+      "learning_rate": 5.099658630633288e-06,
+      "loss": 0.0803,
+      "step": 725
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.822766065597534,
+      "learning_rate": 5.097018195994804e-06,
+      "loss": 0.0628,
+      "step": 726
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.295947313308716,
+      "learning_rate": 5.0943745809619774e-06,
+      "loss": 0.0761,
+      "step": 727
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.2180023193359375,
+      "learning_rate": 5.091727789544188e-06,
+      "loss": 0.0623,
+      "step": 728
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.5102899074554443,
+      "learning_rate": 5.089077825755628e-06,
+      "loss": 0.0553,
+      "step": 729
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.3373754024505615,
+      "learning_rate": 5.0864246936153066e-06,
+      "loss": 0.0586,
+      "step": 730
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 4.313369274139404,
+      "learning_rate": 5.083768397147035e-06,
+      "loss": 0.1397,
+      "step": 731
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.9763259887695312,
+      "learning_rate": 5.081108940379423e-06,
+      "loss": 0.049,
+      "step": 732
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.7267589569091797,
+      "learning_rate": 5.0784463273458744e-06,
+      "loss": 0.0576,
+      "step": 733
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.279266834259033,
+      "learning_rate": 5.075780562084581e-06,
+      "loss": 0.0838,
+      "step": 734
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.5530104637145996,
+      "learning_rate": 5.073111648638514e-06,
+      "loss": 0.0609,
+      "step": 735
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.665799617767334,
+      "learning_rate": 5.07043959105542e-06,
+      "loss": 0.0648,
+      "step": 736
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.5091285705566406,
+      "learning_rate": 5.067764393387811e-06,
+      "loss": 0.0815,
+      "step": 737
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 4.548854351043701,
+      "learning_rate": 5.065086059692967e-06,
+      "loss": 0.0883,
+      "step": 738
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.0658884048461914,
+      "learning_rate": 5.06240459403292e-06,
+      "loss": 0.0793,
+      "step": 739
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.6845388412475586,
+      "learning_rate": 5.059720000474453e-06,
+      "loss": 0.0802,
+      "step": 740
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.0084714889526367,
+      "learning_rate": 5.057032283089094e-06,
+      "loss": 0.0779,
+      "step": 741
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 5.003321170806885,
+      "learning_rate": 5.0543414459531055e-06,
+      "loss": 0.091,
+      "step": 742
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.5993950366973877,
+      "learning_rate": 5.051647493147485e-06,
+      "loss": 0.0554,
+      "step": 743
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.830029249191284,
+      "learning_rate": 5.0489504287579535e-06,
+      "loss": 0.0819,
+      "step": 744
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.902651071548462,
+      "learning_rate": 5.046250256874953e-06,
+      "loss": 0.0805,
+      "step": 745
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.5792648792266846,
+      "learning_rate": 5.043546981593635e-06,
+      "loss": 0.0525,
+      "step": 746
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.467514753341675,
+      "learning_rate": 5.040840607013861e-06,
+      "loss": 0.0475,
+      "step": 747
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.0335798263549805,
+      "learning_rate": 5.0381311372401905e-06,
+      "loss": 0.0547,
+      "step": 748
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 4.355191707611084,
+      "learning_rate": 5.03541857638188e-06,
+      "loss": 0.1099,
+      "step": 749
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 3.922290802001953,
+      "learning_rate": 5.032702928552872e-06,
+      "loss": 0.0952,
+      "step": 750
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.965048313140869,
+      "learning_rate": 5.0299841978717895e-06,
+      "loss": 0.0657,
+      "step": 751
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.4267725944519043,
+      "learning_rate": 5.027262388461935e-06,
+      "loss": 0.0448,
+      "step": 752
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 3.281719207763672,
+      "learning_rate": 5.024537504451277e-06,
+      "loss": 0.0579,
+      "step": 753
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.6463265419006348,
+      "learning_rate": 5.021809549972448e-06,
+      "loss": 0.07,
+      "step": 754
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 3.150157928466797,
+      "learning_rate": 5.019078529162737e-06,
+      "loss": 0.0619,
+      "step": 755
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.822209596633911,
+      "learning_rate": 5.016344446164085e-06,
+      "loss": 0.077,
+      "step": 756
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 3.4186105728149414,
+      "learning_rate": 5.013607305123076e-06,
+      "loss": 0.0781,
+      "step": 757
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 4.415417194366455,
+      "learning_rate": 5.010867110190931e-06,
+      "loss": 0.0646,
+      "step": 758
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 3.8366706371307373,
+      "learning_rate": 5.008123865523503e-06,
+      "loss": 0.0792,
+      "step": 759
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 4.582269191741943,
+      "learning_rate": 5.005377575281272e-06,
+      "loss": 0.1036,
+      "step": 760
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.9546284675598145,
+      "learning_rate": 5.002628243629335e-06,
+      "loss": 0.0723,
+      "step": 761
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.871063470840454,
+      "learning_rate": 4.999875874737404e-06,
+      "loss": 0.0644,
+      "step": 762
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 4.189358711242676,
+      "learning_rate": 4.9971204727797935e-06,
+      "loss": 0.077,
+      "step": 763
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.6419239044189453,
+      "learning_rate": 4.994362041935422e-06,
+      "loss": 0.0489,
+      "step": 764
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.5423083305358887,
+      "learning_rate": 4.9916005863878005e-06,
+      "loss": 0.0702,
+      "step": 765
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.1722166538238525,
+      "learning_rate": 4.9888361103250255e-06,
+      "loss": 0.0688,
+      "step": 766
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.9100143909454346,
+      "learning_rate": 4.986068617939777e-06,
+      "loss": 0.0658,
+      "step": 767
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.4976348876953125,
+      "learning_rate": 4.983298113429308e-06,
+      "loss": 0.0711,
+      "step": 768
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.6911211013793945,
+      "learning_rate": 4.980524600995442e-06,
+      "loss": 0.058,
+      "step": 769
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 4.350445747375488,
+      "learning_rate": 4.977748084844561e-06,
+      "loss": 0.0874,
+      "step": 770
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.5535972118377686,
+      "learning_rate": 4.974968569187603e-06,
+      "loss": 0.0519,
+      "step": 771
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 4.250094413757324,
+      "learning_rate": 4.97218605824006e-06,
+      "loss": 0.1219,
+      "step": 772
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 3.5592308044433594,
+      "learning_rate": 4.969400556221959e-06,
+      "loss": 0.0491,
+      "step": 773
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.734199285507202,
+      "learning_rate": 4.966612067357871e-06,
+      "loss": 0.0453,
+      "step": 774
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 3.443498373031616,
+      "learning_rate": 4.963820595876893e-06,
+      "loss": 0.0874,
+      "step": 775
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.3944900035858154,
+      "learning_rate": 4.961026146012644e-06,
+      "loss": 0.0796,
+      "step": 776
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.7430405616760254,
+      "learning_rate": 4.958228722003263e-06,
+      "loss": 0.0721,
+      "step": 777
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.884028196334839,
+      "learning_rate": 4.9554283280913985e-06,
+      "loss": 0.0889,
+      "step": 778
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.374960422515869,
+      "learning_rate": 4.952624968524204e-06,
+      "loss": 0.0556,
+      "step": 779
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.5512306690216064,
+      "learning_rate": 4.949818647553329e-06,
+      "loss": 0.1132,
+      "step": 780
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.345276117324829,
+      "learning_rate": 4.947009369434917e-06,
+      "loss": 0.0396,
+      "step": 781
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.180399179458618,
+      "learning_rate": 4.944197138429595e-06,
+      "loss": 0.0666,
+      "step": 782
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.419724702835083,
+      "learning_rate": 4.941381958802467e-06,
+      "loss": 0.0892,
+      "step": 783
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.469177007675171,
+      "learning_rate": 4.9385638348231105e-06,
+      "loss": 0.0691,
+      "step": 784
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.8321540355682373,
+      "learning_rate": 4.93574277076557e-06,
+      "loss": 0.1219,
+      "step": 785
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 4.477923393249512,
+      "learning_rate": 4.9329187709083445e-06,
+      "loss": 0.1107,
+      "step": 786
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 4.694797992706299,
+      "learning_rate": 4.9300918395343916e-06,
+      "loss": 0.0978,
+      "step": 787
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.6760900020599365,
+      "learning_rate": 4.92726198093111e-06,
+      "loss": 0.0572,
+      "step": 788
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 3.0248184204101562,
+      "learning_rate": 4.9244291993903386e-06,
+      "loss": 0.0571,
+      "step": 789
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.699068307876587,
+      "learning_rate": 4.921593499208351e-06,
+      "loss": 0.0498,
+      "step": 790
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 3.457937240600586,
+      "learning_rate": 4.918754884685848e-06,
+      "loss": 0.0946,
+      "step": 791
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.744284152984619,
+      "learning_rate": 4.915913360127947e-06,
+      "loss": 0.062,
+      "step": 792
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.9463562965393066,
+      "learning_rate": 4.9130689298441825e-06,
+      "loss": 0.0766,
+      "step": 793
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.299647808074951,
+      "learning_rate": 4.910221598148492e-06,
+      "loss": 0.0759,
+      "step": 794
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.4989094734191895,
+      "learning_rate": 4.907371369359218e-06,
+      "loss": 0.0755,
+      "step": 795
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 4.583041191101074,
+      "learning_rate": 4.904518247799092e-06,
+      "loss": 0.117,
+      "step": 796
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.5209977626800537,
+      "learning_rate": 4.901662237795236e-06,
+      "loss": 0.0769,
+      "step": 797
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.7468550205230713,
+      "learning_rate": 4.898803343679152e-06,
+      "loss": 0.0572,
+      "step": 798
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.183668851852417,
+      "learning_rate": 4.895941569786717e-06,
+      "loss": 0.0732,
+      "step": 799
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.1500980854034424,
+      "learning_rate": 4.893076920458173e-06,
+      "loss": 0.0704,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "eval_blimp_filtered_avg": 0.7317910447761194,
+      "eval_blimp_filtered_std": 0.004875044963512702,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.017481943715127863,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "eval_vqa_filtered_avg": 0.27,
+      "eval_vqa_filtered_std": 0.044619604333847394,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.4926350116729736,
+      "learning_rate": 4.890209400038124e-06,
+      "loss": 0.0604,
+      "step": 801
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.508463144302368,
+      "learning_rate": 4.88733901287553e-06,
+      "loss": 0.08,
+      "step": 802
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.7854483127593994,
+      "learning_rate": 4.884465763323696e-06,
+      "loss": 0.0538,
+      "step": 803
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.138047456741333,
+      "learning_rate": 4.881589655740272e-06,
+      "loss": 0.0583,
+      "step": 804
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 4.414126873016357,
+      "learning_rate": 4.878710694487239e-06,
+      "loss": 0.0849,
+      "step": 805
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.9349567890167236,
+      "learning_rate": 4.875828883930907e-06,
+      "loss": 0.0676,
+      "step": 806
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.7681500911712646,
+      "learning_rate": 4.872944228441908e-06,
+      "loss": 0.0683,
+      "step": 807
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 3.0172042846679688,
+      "learning_rate": 4.870056732395189e-06,
+      "loss": 0.0724,
+      "step": 808
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 3.090986967086792,
+      "learning_rate": 4.867166400170003e-06,
+      "loss": 0.059,
+      "step": 809
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.7678325176239014,
+      "learning_rate": 4.864273236149907e-06,
+      "loss": 0.0487,
+      "step": 810
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 3.4007985591888428,
+      "learning_rate": 4.86137724472275e-06,
+      "loss": 0.0702,
+      "step": 811
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.768096446990967,
+      "learning_rate": 4.858478430280673e-06,
+      "loss": 0.0613,
+      "step": 812
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.812058448791504,
+      "learning_rate": 4.855576797220094e-06,
+      "loss": 0.0528,
+      "step": 813
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 3.0758378505706787,
+      "learning_rate": 4.85267234994171e-06,
+      "loss": 0.0614,
+      "step": 814
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 4.042909145355225,
+      "learning_rate": 4.849765092850481e-06,
+      "loss": 0.0651,
+      "step": 815
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.4650936126708984,
+      "learning_rate": 4.846855030355634e-06,
+      "loss": 0.0583,
+      "step": 816
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 3.453695774078369,
+      "learning_rate": 4.843942166870648e-06,
+      "loss": 0.061,
+      "step": 817
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.7488958835601807,
+      "learning_rate": 4.84102650681325e-06,
+      "loss": 0.0657,
+      "step": 818
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.3512628078460693,
+      "learning_rate": 4.8381080546054084e-06,
+      "loss": 0.0826,
+      "step": 819
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.5793585777282715,
+      "learning_rate": 4.8351868146733265e-06,
+      "loss": 0.0696,
+      "step": 820
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.982083320617676,
+      "learning_rate": 4.832262791447436e-06,
+      "loss": 0.0596,
+      "step": 821
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.9933652877807617,
+      "learning_rate": 4.829335989362389e-06,
+      "loss": 0.0558,
+      "step": 822
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.5046451091766357,
+      "learning_rate": 4.8264064128570525e-06,
+      "loss": 0.0955,
+      "step": 823
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 3.3016903400421143,
+      "learning_rate": 4.823474066374502e-06,
+      "loss": 0.0744,
+      "step": 824
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.8992178440093994,
+      "learning_rate": 4.820538954362012e-06,
+      "loss": 0.0477,
+      "step": 825
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.881643295288086,
+      "learning_rate": 4.817601081271054e-06,
+      "loss": 0.0846,
+      "step": 826
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.9012608528137207,
+      "learning_rate": 4.814660451557284e-06,
+      "loss": 0.0532,
+      "step": 827
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 3.2729530334472656,
+      "learning_rate": 4.811717069680539e-06,
+      "loss": 0.0648,
+      "step": 828
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 3.4763076305389404,
+      "learning_rate": 4.808770940104836e-06,
+      "loss": 0.0637,
+      "step": 829
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 3.4917445182800293,
+      "learning_rate": 4.805822067298351e-06,
+      "loss": 0.0742,
+      "step": 830
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.938612461090088,
+      "learning_rate": 4.802870455733425e-06,
+      "loss": 0.0435,
+      "step": 831
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.8640530109405518,
+      "learning_rate": 4.799916109886551e-06,
+      "loss": 0.0673,
+      "step": 832
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 3.4792532920837402,
+      "learning_rate": 4.79695903423837e-06,
+      "loss": 0.0682,
+      "step": 833
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.5372092723846436,
+      "learning_rate": 4.793999233273663e-06,
+      "loss": 0.0709,
+      "step": 834
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.998063564300537,
+      "learning_rate": 4.791036711481343e-06,
+      "loss": 0.0691,
+      "step": 835
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.0680971145629883,
+      "learning_rate": 4.788071473354453e-06,
+      "loss": 0.0612,
+      "step": 836
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.3042004108428955,
+      "learning_rate": 4.78510352339015e-06,
+      "loss": 0.0491,
+      "step": 837
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.7906112670898438,
+      "learning_rate": 4.782132866089708e-06,
+      "loss": 0.0887,
+      "step": 838
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 4.182991981506348,
+      "learning_rate": 4.779159505958507e-06,
+      "loss": 0.0852,
+      "step": 839
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.6279900074005127,
+      "learning_rate": 4.776183447506023e-06,
+      "loss": 0.0516,
+      "step": 840
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 4.046883583068848,
+      "learning_rate": 4.7732046952458285e-06,
+      "loss": 0.0936,
+      "step": 841
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 4.838845729827881,
+      "learning_rate": 4.770223253695579e-06,
+      "loss": 0.0753,
+      "step": 842
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 4.783663749694824,
+      "learning_rate": 4.767239127377009e-06,
+      "loss": 0.0546,
+      "step": 843
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.5956008434295654,
+      "learning_rate": 4.764252320815926e-06,
+      "loss": 0.082,
+      "step": 844
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.1544735431671143,
+      "learning_rate": 4.761262838542198e-06,
+      "loss": 0.049,
+      "step": 845
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.6342859268188477,
+      "learning_rate": 4.7582706850897585e-06,
+      "loss": 0.0505,
+      "step": 846
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.999417304992676,
+      "learning_rate": 4.7552758649965865e-06,
+      "loss": 0.0736,
+      "step": 847
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.6854333877563477,
+      "learning_rate": 4.752278382804705e-06,
+      "loss": 0.0749,
+      "step": 848
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.588259696960449,
+      "learning_rate": 4.7492782430601795e-06,
+      "loss": 0.0696,
+      "step": 849
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.741121768951416,
+      "learning_rate": 4.746275450313101e-06,
+      "loss": 0.0731,
+      "step": 850
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 3.206705331802368,
+      "learning_rate": 4.743270009117586e-06,
+      "loss": 0.0812,
+      "step": 851
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.795283555984497,
+      "learning_rate": 4.740261924031768e-06,
+      "loss": 0.036,
+      "step": 852
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 3.526078939437866,
+      "learning_rate": 4.73725119961779e-06,
+      "loss": 0.0742,
+      "step": 853
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.5192487239837646,
+      "learning_rate": 4.7342378404417975e-06,
+      "loss": 0.0529,
+      "step": 854
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.6468584537506104,
+      "learning_rate": 4.7312218510739336e-06,
+      "loss": 0.0616,
+      "step": 855
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.13260555267334,
+      "learning_rate": 4.728203236088327e-06,
+      "loss": 0.0598,
+      "step": 856
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.2082693576812744,
+      "learning_rate": 4.725182000063093e-06,
+      "loss": 0.062,
+      "step": 857
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 4.530020713806152,
+      "learning_rate": 4.722158147580316e-06,
+      "loss": 0.0779,
+      "step": 858
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.9379100799560547,
+      "learning_rate": 4.719131683226057e-06,
+      "loss": 0.0449,
+      "step": 859
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.6129441261291504,
+      "learning_rate": 4.716102611590329e-06,
+      "loss": 0.0814,
+      "step": 860
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.013540744781494,
+      "learning_rate": 4.7130709372671065e-06,
+      "loss": 0.0711,
+      "step": 861
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.8282206058502197,
+      "learning_rate": 4.710036664854308e-06,
+      "loss": 0.0615,
+      "step": 862
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.661595106124878,
+      "learning_rate": 4.706999798953792e-06,
+      "loss": 0.0729,
+      "step": 863
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.9159328937530518,
+      "learning_rate": 4.703960344171352e-06,
+      "loss": 0.1056,
+      "step": 864
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.9074695110321045,
+      "learning_rate": 4.700918305116707e-06,
+      "loss": 0.0475,
+      "step": 865
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.44800066947937,
+      "learning_rate": 4.697873686403494e-06,
+      "loss": 0.0501,
+      "step": 866
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.7500429153442383,
+      "learning_rate": 4.694826492649266e-06,
+      "loss": 0.0569,
+      "step": 867
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 3.242976427078247,
+      "learning_rate": 4.6917767284754785e-06,
+      "loss": 0.0548,
+      "step": 868
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.710387706756592,
+      "learning_rate": 4.688724398507486e-06,
+      "loss": 0.0422,
+      "step": 869
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.5250134468078613,
+      "learning_rate": 4.685669507374535e-06,
+      "loss": 0.0438,
+      "step": 870
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.4566807746887207,
+      "learning_rate": 4.682612059709753e-06,
+      "loss": 0.0532,
+      "step": 871
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.3873097896575928,
+      "learning_rate": 4.679552060150151e-06,
+      "loss": 0.0743,
+      "step": 872
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.369290590286255,
+      "learning_rate": 4.676489513336606e-06,
+      "loss": 0.0757,
+      "step": 873
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 4.575608730316162,
+      "learning_rate": 4.673424423913858e-06,
+      "loss": 0.0692,
+      "step": 874
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.6137380599975586,
+      "learning_rate": 4.670356796530506e-06,
+      "loss": 0.0611,
+      "step": 875
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.457421064376831,
+      "learning_rate": 4.667286635838994e-06,
+      "loss": 0.0623,
+      "step": 876
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.1409964561462402,
+      "learning_rate": 4.664213946495611e-06,
+      "loss": 0.0609,
+      "step": 877
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.406859874725342,
+      "learning_rate": 4.661138733160483e-06,
+      "loss": 0.0847,
+      "step": 878
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.08162522315979,
+      "learning_rate": 4.65806100049756e-06,
+      "loss": 0.0471,
+      "step": 879
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.1542272567749023,
+      "learning_rate": 4.654980753174613e-06,
+      "loss": 0.0652,
+      "step": 880
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.224440574645996,
+      "learning_rate": 4.65189799586323e-06,
+      "loss": 0.0669,
+      "step": 881
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.6096351146698,
+      "learning_rate": 4.648812733238802e-06,
+      "loss": 0.0462,
+      "step": 882
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.430137872695923,
+      "learning_rate": 4.645724969980524e-06,
+      "loss": 0.0625,
+      "step": 883
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.439643621444702,
+      "learning_rate": 4.642634710771379e-06,
+      "loss": 0.0386,
+      "step": 884
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.2073745727539062,
+      "learning_rate": 4.639541960298138e-06,
+      "loss": 0.061,
+      "step": 885
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.939011573791504,
+      "learning_rate": 4.63644672325135e-06,
+      "loss": 0.0703,
+      "step": 886
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 3.063945770263672,
+      "learning_rate": 4.633349004325334e-06,
+      "loss": 0.0833,
+      "step": 887
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 3.1042251586914062,
+      "learning_rate": 4.630248808218177e-06,
+      "loss": 0.0536,
+      "step": 888
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 3.494886875152588,
+      "learning_rate": 4.627146139631715e-06,
+      "loss": 0.0555,
+      "step": 889
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.579806089401245,
+      "learning_rate": 4.624041003271544e-06,
+      "loss": 0.0622,
+      "step": 890
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.295497417449951,
+      "learning_rate": 4.620933403846994e-06,
+      "loss": 0.0438,
+      "step": 891
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.6644203662872314,
+      "learning_rate": 4.617823346071136e-06,
+      "loss": 0.0505,
+      "step": 892
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.7021586894989014,
+      "learning_rate": 4.614710834660767e-06,
+      "loss": 0.0797,
+      "step": 893
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 4.569178104400635,
+      "learning_rate": 4.611595874336405e-06,
+      "loss": 0.0859,
+      "step": 894
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 3.2065868377685547,
+      "learning_rate": 4.608478469822284e-06,
+      "loss": 0.0756,
+      "step": 895
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 3.1257216930389404,
+      "learning_rate": 4.605358625846343e-06,
+      "loss": 0.0671,
+      "step": 896
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.304539203643799,
+      "learning_rate": 4.602236347140221e-06,
+      "loss": 0.0376,
+      "step": 897
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.3594095706939697,
+      "learning_rate": 4.599111638439252e-06,
+      "loss": 0.0639,
+      "step": 898
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.2329976558685303,
+      "learning_rate": 4.595984504482451e-06,
+      "loss": 0.0687,
+      "step": 899
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.9502577781677246,
+      "learning_rate": 4.592854950012515e-06,
+      "loss": 0.0553,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "eval_blimp_filtered_avg": 0.7304477611940299,
+      "eval_blimp_filtered_std": 0.004875068041904964,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.017444301474488378,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "eval_vqa_filtered_avg": 0.3,
+      "eval_vqa_filtered_std": 0.046056618647183814,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.2676796913146973,
+      "learning_rate": 4.58972297977581e-06,
+      "loss": 0.0506,
+      "step": 901
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.6399917602539062,
+      "learning_rate": 4.586588598522368e-06,
+      "loss": 0.076,
+      "step": 902
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.217087745666504,
+      "learning_rate": 4.583451811005875e-06,
+      "loss": 0.0337,
+      "step": 903
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.461214065551758,
+      "learning_rate": 4.580312621983668e-06,
+      "loss": 0.0476,
+      "step": 904
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 3.9499711990356445,
+      "learning_rate": 4.577171036216725e-06,
+      "loss": 0.0778,
+      "step": 905
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 3.358560085296631,
+      "learning_rate": 4.5740270584696606e-06,
+      "loss": 0.0668,
+      "step": 906
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.93816876411438,
+      "learning_rate": 4.570880693510716e-06,
+      "loss": 0.065,
+      "step": 907
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.7721872329711914,
+      "learning_rate": 4.5677319461117516e-06,
+      "loss": 0.0688,
+      "step": 908
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.0420219898223877,
+      "learning_rate": 4.5645808210482445e-06,
+      "loss": 0.0414,
+      "step": 909
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.7129127979278564,
+      "learning_rate": 4.561427323099276e-06,
+      "loss": 0.0532,
+      "step": 910
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.365619659423828,
+      "learning_rate": 4.558271457047525e-06,
+      "loss": 0.0524,
+      "step": 911
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.466017484664917,
+      "learning_rate": 4.555113227679264e-06,
+      "loss": 0.0858,
+      "step": 912
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.6452770233154297,
+      "learning_rate": 4.551952639784348e-06,
+      "loss": 0.0558,
+      "step": 913
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 3.3776464462280273,
+      "learning_rate": 4.54878969815621e-06,
+      "loss": 0.0685,
+      "step": 914
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.8052330017089844,
+      "learning_rate": 4.545624407591853e-06,
+      "loss": 0.0461,
+      "step": 915
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.696474075317383,
+      "learning_rate": 4.5424567728918414e-06,
+      "loss": 0.0558,
+      "step": 916
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 3.1220052242279053,
+      "learning_rate": 4.539286798860296e-06,
+      "loss": 0.0647,
+      "step": 917
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.9479033946990967,
+      "learning_rate": 4.536114490304885e-06,
+      "loss": 0.0846,
+      "step": 918
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.8405308723449707,
+      "learning_rate": 4.532939852036817e-06,
+      "loss": 0.0459,
+      "step": 919
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 3.046926736831665,
+      "learning_rate": 4.529762888870832e-06,
+      "loss": 0.0593,
+      "step": 920
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.361496925354004,
+      "learning_rate": 4.526583605625201e-06,
+      "loss": 0.0404,
+      "step": 921
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.940732002258301,
+      "learning_rate": 4.52340200712171e-06,
+      "loss": 0.0461,
+      "step": 922
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 3.1042771339416504,
+      "learning_rate": 4.520218098185656e-06,
+      "loss": 0.0732,
+      "step": 923
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 3.173741340637207,
+      "learning_rate": 4.517031883645842e-06,
+      "loss": 0.0612,
+      "step": 924
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 3.005009174346924,
+      "learning_rate": 4.513843368334566e-06,
+      "loss": 0.0695,
+      "step": 925
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 3.764728307723999,
+      "learning_rate": 4.510652557087617e-06,
+      "loss": 0.063,
+      "step": 926
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.984952688217163,
+      "learning_rate": 4.507459454744266e-06,
+      "loss": 0.0609,
+      "step": 927
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 3.4865636825561523,
+      "learning_rate": 4.504264066147256e-06,
+      "loss": 0.0776,
+      "step": 928
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.5523931980133057,
+      "learning_rate": 4.501066396142803e-06,
+      "loss": 0.0541,
+      "step": 929
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.5705180168151855,
+      "learning_rate": 4.497866449580575e-06,
+      "loss": 0.0774,
+      "step": 930
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 4.013316631317139,
+      "learning_rate": 4.4946642313137e-06,
+      "loss": 0.0621,
+      "step": 931
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.1012487411499023,
+      "learning_rate": 4.491459746198749e-06,
+      "loss": 0.067,
+      "step": 932
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.4574036598205566,
+      "learning_rate": 4.48825299909573e-06,
+      "loss": 0.0574,
+      "step": 933
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.5531845092773438,
+      "learning_rate": 4.485043994868082e-06,
+      "loss": 0.0412,
+      "step": 934
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.901017665863037,
+      "learning_rate": 4.481832738382667e-06,
+      "loss": 0.0765,
+      "step": 935
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.783984661102295,
+      "learning_rate": 4.478619234509764e-06,
+      "loss": 0.0424,
+      "step": 936
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.2479729652404785,
+      "learning_rate": 4.475403488123058e-06,
+      "loss": 0.077,
+      "step": 937
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.97456955909729,
+      "learning_rate": 4.472185504099637e-06,
+      "loss": 0.063,
+      "step": 938
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.892801523208618,
+      "learning_rate": 4.468965287319985e-06,
+      "loss": 0.0493,
+      "step": 939
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 3.923037528991699,
+      "learning_rate": 4.465742842667967e-06,
+      "loss": 0.1001,
+      "step": 940
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.8867766857147217,
+      "learning_rate": 4.462518175030831e-06,
+      "loss": 0.0589,
+      "step": 941
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.991340160369873,
+      "learning_rate": 4.459291289299192e-06,
+      "loss": 0.0581,
+      "step": 942
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.9316601753234863,
+      "learning_rate": 4.456062190367035e-06,
+      "loss": 0.0517,
+      "step": 943
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.7171854972839355,
+      "learning_rate": 4.452830883131697e-06,
+      "loss": 0.0684,
+      "step": 944
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 3.5220720767974854,
+      "learning_rate": 4.4495973724938676e-06,
+      "loss": 0.0574,
+      "step": 945
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.2584519386291504,
+      "learning_rate": 4.446361663357573e-06,
+      "loss": 0.0529,
+      "step": 946
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 3.2773149013519287,
+      "learning_rate": 4.44312376063018e-06,
+      "loss": 0.0686,
+      "step": 947
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.332120180130005,
+      "learning_rate": 4.439883669222377e-06,
+      "loss": 0.0455,
+      "step": 948
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.864718198776245,
+      "learning_rate": 4.436641394048175e-06,
+      "loss": 0.0596,
+      "step": 949
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.610189437866211,
+      "learning_rate": 4.433396940024897e-06,
+      "loss": 0.0528,
+      "step": 950
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.384280204772949,
+      "learning_rate": 4.430150312073166e-06,
+      "loss": 0.0937,
+      "step": 951
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.6756765842437744,
+      "learning_rate": 4.4269015151169105e-06,
+      "loss": 0.0775,
+      "step": 952
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 4.20609712600708,
+      "learning_rate": 4.423650554083338e-06,
+      "loss": 0.077,
+      "step": 953
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.455314874649048,
+      "learning_rate": 4.4203974339029485e-06,
+      "loss": 0.0392,
+      "step": 954
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.5252482891082764,
+      "learning_rate": 4.4171421595095096e-06,
+      "loss": 0.0607,
+      "step": 955
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 3.1200687885284424,
+      "learning_rate": 4.413884735840057e-06,
+      "loss": 0.0879,
+      "step": 956
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 3.3120663166046143,
+      "learning_rate": 4.410625167834891e-06,
+      "loss": 0.0498,
+      "step": 957
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 3.2222700119018555,
+      "learning_rate": 4.407363460437557e-06,
+      "loss": 0.0684,
+      "step": 958
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.3726274967193604,
+      "learning_rate": 4.404099618594849e-06,
+      "loss": 0.045,
+      "step": 959
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 3.321824789047241,
+      "learning_rate": 4.400833647256798e-06,
+      "loss": 0.07,
+      "step": 960
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.878753423690796,
+      "learning_rate": 4.397565551376664e-06,
+      "loss": 0.0593,
+      "step": 961
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.990257501602173,
+      "learning_rate": 4.39429533591093e-06,
+      "loss": 0.0602,
+      "step": 962
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.7450101375579834,
+      "learning_rate": 4.391023005819292e-06,
+      "loss": 0.0541,
+      "step": 963
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.4538486003875732,
+      "learning_rate": 4.3877485660646554e-06,
+      "loss": 0.0574,
+      "step": 964
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.0417442321777344,
+      "learning_rate": 4.384472021613122e-06,
+      "loss": 0.0422,
+      "step": 965
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.5068628787994385,
+      "learning_rate": 4.381193377433989e-06,
+      "loss": 0.0539,
+      "step": 966
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 4.005906105041504,
+      "learning_rate": 4.377912638499735e-06,
+      "loss": 0.0702,
+      "step": 967
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 3.372375965118408,
+      "learning_rate": 4.37462980978602e-06,
+      "loss": 0.061,
+      "step": 968
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.6943588256835938,
+      "learning_rate": 4.371344896271667e-06,
+      "loss": 0.0498,
+      "step": 969
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.97693133354187,
+      "learning_rate": 4.368057902938666e-06,
+      "loss": 0.0606,
+      "step": 970
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 2.3645269870758057,
+      "learning_rate": 4.3647688347721615e-06,
+      "loss": 0.042,
+      "step": 971
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 3.1078710556030273,
+      "learning_rate": 4.361477696760439e-06,
+      "loss": 0.0629,
+      "step": 972
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 3.1994948387145996,
+      "learning_rate": 4.358184493894931e-06,
+      "loss": 0.0594,
+      "step": 973
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 4.063432216644287,
+      "learning_rate": 4.354889231170196e-06,
+      "loss": 0.0643,
+      "step": 974
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 2.221763849258423,
+      "learning_rate": 4.351591913583918e-06,
+      "loss": 0.039,
+      "step": 975
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 2.271899461746216,
+      "learning_rate": 4.348292546136899e-06,
+      "loss": 0.0532,
+      "step": 976
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 3.3591384887695312,
+      "learning_rate": 4.344991133833047e-06,
+      "loss": 0.0398,
+      "step": 977
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.959396481513977,
+      "learning_rate": 4.341687681679375e-06,
+      "loss": 0.0402,
+      "step": 978
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 2.190448760986328,
+      "learning_rate": 4.338382194685985e-06,
+      "loss": 0.0579,
+      "step": 979
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 3.087825059890747,
+      "learning_rate": 4.335074677866071e-06,
+      "loss": 0.0651,
+      "step": 980
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.8548355102539062,
+      "learning_rate": 4.331765136235899e-06,
+      "loss": 0.0571,
+      "step": 981
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 3.7755696773529053,
+      "learning_rate": 4.32845357481481e-06,
+      "loss": 0.0668,
+      "step": 982
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.579369068145752,
+      "learning_rate": 4.325139998625208e-06,
+      "loss": 0.0631,
+      "step": 983
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.740443229675293,
+      "learning_rate": 4.321824412692552e-06,
+      "loss": 0.0532,
+      "step": 984
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 4.43401575088501,
+      "learning_rate": 4.318506822045349e-06,
+      "loss": 0.094,
+      "step": 985
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.784532308578491,
+      "learning_rate": 4.315187231715145e-06,
+      "loss": 0.0513,
+      "step": 986
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.685720682144165,
+      "learning_rate": 4.311865646736523e-06,
+      "loss": 0.0474,
+      "step": 987
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.6349680423736572,
+      "learning_rate": 4.308542072147087e-06,
+      "loss": 0.0611,
+      "step": 988
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 3.229809045791626,
+      "learning_rate": 4.305216512987459e-06,
+      "loss": 0.0558,
+      "step": 989
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 3.601703643798828,
+      "learning_rate": 4.301888974301271e-06,
+      "loss": 0.0905,
+      "step": 990
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.8302342891693115,
+      "learning_rate": 4.298559461135161e-06,
+      "loss": 0.0814,
+      "step": 991
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 3.0556039810180664,
+      "learning_rate": 4.295227978538757e-06,
+      "loss": 0.0465,
+      "step": 992
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 3.2168853282928467,
+      "learning_rate": 4.291894531564673e-06,
+      "loss": 0.0507,
+      "step": 993
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 4.481518268585205,
+      "learning_rate": 4.2885591252685075e-06,
+      "loss": 0.1071,
+      "step": 994
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 3.270374298095703,
+      "learning_rate": 4.285221764708827e-06,
+      "loss": 0.0506,
+      "step": 995
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 2.6043612957000732,
+      "learning_rate": 4.281882454947162e-06,
+      "loss": 0.066,
+      "step": 996
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 3.250666856765747,
+      "learning_rate": 4.278541201048e-06,
+      "loss": 0.0753,
+      "step": 997
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.5600380897521973,
+      "learning_rate": 4.275198008078777e-06,
+      "loss": 0.0479,
+      "step": 998
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 3.0057122707366943,
+      "learning_rate": 4.271852881109869e-06,
+      "loss": 0.0747,
+      "step": 999
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 3.306511640548706,
+      "learning_rate": 4.268505825214586e-06,
+      "loss": 0.0759,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "eval_blimp_filtered_avg": 0.7301492537313433,
+      "eval_blimp_filtered_std": 0.004875714192537722,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.017444301474488378,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "eval_vqa_filtered_avg": 0.29,
+      "eval_vqa_filtered_std": 0.045604802157206845,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "eval_winoground_filtered_avg": 0.46,
+      "eval_winoground_filtered_std": 0.05009082659620333,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 3.4723522663116455,
+      "learning_rate": 4.2651568454691625e-06,
+      "loss": 0.0474,
+      "step": 1001
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 3.7219743728637695,
+      "learning_rate": 4.2618059469527534e-06,
+      "loss": 0.1028,
+      "step": 1002
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 3.145127534866333,
+      "learning_rate": 4.258453134747421e-06,
+      "loss": 0.0724,
+      "step": 1003
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 4.006146430969238,
+      "learning_rate": 4.255098413938129e-06,
+      "loss": 0.0856,
+      "step": 1004
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.08056378364563,
+      "learning_rate": 4.251741789612739e-06,
+      "loss": 0.0435,
+      "step": 1005
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 3.0988831520080566,
+      "learning_rate": 4.248383266861998e-06,
+      "loss": 0.0476,
+      "step": 1006
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 3.019416093826294,
+      "learning_rate": 4.245022850779532e-06,
+      "loss": 0.0489,
+      "step": 1007
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 3.3077499866485596,
+      "learning_rate": 4.241660546461838e-06,
+      "loss": 0.0603,
+      "step": 1008
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 3.610659599304199,
+      "learning_rate": 4.238296359008279e-06,
+      "loss": 0.08,
+      "step": 1009
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 2.46352219581604,
+      "learning_rate": 4.234930293521071e-06,
+      "loss": 0.0492,
+      "step": 1010
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 2.479257583618164,
+      "learning_rate": 4.23156235510528e-06,
+      "loss": 0.0567,
+      "step": 1011
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 2.658146381378174,
+      "learning_rate": 4.228192548868811e-06,
+      "loss": 0.0531,
+      "step": 1012
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 2.5000007152557373,
+      "learning_rate": 4.224820879922405e-06,
+      "loss": 0.0426,
+      "step": 1013
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 3.387082576751709,
+      "learning_rate": 4.221447353379627e-06,
+      "loss": 0.0686,
+      "step": 1014
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 4.312678813934326,
+      "learning_rate": 4.218071974356855e-06,
+      "loss": 0.0851,
+      "step": 1015
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 2.885866641998291,
+      "learning_rate": 4.214694747973281e-06,
+      "loss": 0.0667,
+      "step": 1016
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 2.7459475994110107,
+      "learning_rate": 4.211315679350899e-06,
+      "loss": 0.0718,
+      "step": 1017
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.946201801300049,
+      "learning_rate": 4.207934773614493e-06,
+      "loss": 0.0692,
+      "step": 1018
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.5044426918029785,
+      "learning_rate": 4.204552035891639e-06,
+      "loss": 0.0726,
+      "step": 1019
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.416746139526367,
+      "learning_rate": 4.201167471312686e-06,
+      "loss": 0.0406,
+      "step": 1020
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 3.3682641983032227,
+      "learning_rate": 4.197781085010758e-06,
+      "loss": 0.0605,
+      "step": 1021
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.2761874198913574,
+      "learning_rate": 4.1943928821217375e-06,
+      "loss": 0.046,
+      "step": 1022
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.4868669509887695,
+      "learning_rate": 4.191002867784265e-06,
+      "loss": 0.0519,
+      "step": 1023
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.1130192279815674,
+      "learning_rate": 4.187611047139728e-06,
+      "loss": 0.0408,
+      "step": 1024
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 3.1110267639160156,
+      "learning_rate": 4.184217425332253e-06,
+      "loss": 0.0509,
+      "step": 1025
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.195739269256592,
+      "learning_rate": 4.180822007508698e-06,
+      "loss": 0.0561,
+      "step": 1026
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.7927037477493286,
+      "learning_rate": 4.177424798818646e-06,
+      "loss": 0.0409,
+      "step": 1027
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 4.053756237030029,
+      "learning_rate": 4.174025804414395e-06,
+      "loss": 0.0861,
+      "step": 1028
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.665940999984741,
+      "learning_rate": 4.170625029450951e-06,
+      "loss": 0.0382,
+      "step": 1029
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.460106134414673,
+      "learning_rate": 4.167222479086022e-06,
+      "loss": 0.0495,
+      "step": 1030
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.917155385017395,
+      "learning_rate": 4.163818158480007e-06,
+      "loss": 0.0308,
+      "step": 1031
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 3.421854019165039,
+      "learning_rate": 4.1604120727959905e-06,
+      "loss": 0.0947,
+      "step": 1032
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 3.2238974571228027,
+      "learning_rate": 4.157004227199735e-06,
+      "loss": 0.0478,
+      "step": 1033
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 2.800593137741089,
+      "learning_rate": 4.153594626859668e-06,
+      "loss": 0.0378,
+      "step": 1034
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 2.222750425338745,
+      "learning_rate": 4.150183276946884e-06,
+      "loss": 0.0355,
+      "step": 1035
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 2.9419667720794678,
+      "learning_rate": 4.146770182635128e-06,
+      "loss": 0.0514,
+      "step": 1036
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 3.043267011642456,
+      "learning_rate": 4.143355349100791e-06,
+      "loss": 0.0695,
+      "step": 1037
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 2.915107250213623,
+      "learning_rate": 4.1399387815229005e-06,
+      "loss": 0.0523,
+      "step": 1038
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 3.37503719329834,
+      "learning_rate": 4.136520485083115e-06,
+      "loss": 0.0704,
+      "step": 1039
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.2963778972625732,
+      "learning_rate": 4.133100464965716e-06,
+      "loss": 0.0328,
+      "step": 1040
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.6175026893615723,
+      "learning_rate": 4.1296787263575976e-06,
+      "loss": 0.0664,
+      "step": 1041
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.9905483722686768,
+      "learning_rate": 4.126255274448261e-06,
+      "loss": 0.0748,
+      "step": 1042
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.951986312866211,
+      "learning_rate": 4.122830114429806e-06,
+      "loss": 0.0312,
+      "step": 1043
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 3.0539143085479736,
+      "learning_rate": 4.119403251496923e-06,
+      "loss": 0.0527,
+      "step": 1044
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 2.4951043128967285,
+      "learning_rate": 4.115974690846883e-06,
+      "loss": 0.0439,
+      "step": 1045
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 2.873356819152832,
+      "learning_rate": 4.112544437679535e-06,
+      "loss": 0.073,
+      "step": 1046
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 3.5707778930664062,
+      "learning_rate": 4.109112497197293e-06,
+      "loss": 0.0575,
+      "step": 1047
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 4.003203392028809,
+      "learning_rate": 4.105678874605131e-06,
+      "loss": 0.0472,
+      "step": 1048
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 2.783019781112671,
+      "learning_rate": 4.102243575110574e-06,
+      "loss": 0.0623,
+      "step": 1049
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 4.583861351013184,
+      "learning_rate": 4.098806603923688e-06,
+      "loss": 0.0855,
+      "step": 1050
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.169262409210205,
+      "learning_rate": 4.095367966257077e-06,
+      "loss": 0.0766,
+      "step": 1051
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.2409114837646484,
+      "learning_rate": 4.0919276673258725e-06,
+      "loss": 0.045,
+      "step": 1052
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.7422466278076172,
+      "learning_rate": 4.088485712347724e-06,
+      "loss": 0.0248,
+      "step": 1053
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.3879575729370117,
+      "learning_rate": 4.085042106542791e-06,
+      "loss": 0.037,
+      "step": 1054
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 4.255546569824219,
+      "learning_rate": 4.081596855133744e-06,
+      "loss": 0.0453,
+      "step": 1055
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 2.34485125541687,
+      "learning_rate": 4.07814996334574e-06,
+      "loss": 0.0419,
+      "step": 1056
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 2.542569398880005,
+      "learning_rate": 4.07470143640643e-06,
+      "loss": 0.0366,
+      "step": 1057
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 2.2456531524658203,
+      "learning_rate": 4.071251279545942e-06,
+      "loss": 0.0509,
+      "step": 1058
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 2.0784802436828613,
+      "learning_rate": 4.067799497996877e-06,
+      "loss": 0.0362,
+      "step": 1059
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 2.5523648262023926,
+      "learning_rate": 4.0643460969943e-06,
+      "loss": 0.029,
+      "step": 1060
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 2.9249513149261475,
+      "learning_rate": 4.060891081775734e-06,
+      "loss": 0.038,
+      "step": 1061
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 3.3008673191070557,
+      "learning_rate": 4.057434457581144e-06,
+      "loss": 0.0207,
+      "step": 1062
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 3.2306180000305176,
+      "learning_rate": 4.053976229652943e-06,
+      "loss": 0.0476,
+      "step": 1063
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 2.6013245582580566,
+      "learning_rate": 4.0505164032359714e-06,
+      "loss": 0.0457,
+      "step": 1064
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 2.1071321964263916,
+      "learning_rate": 4.047054983577496e-06,
+      "loss": 0.0271,
+      "step": 1065
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 2.2183568477630615,
+      "learning_rate": 4.043591975927199e-06,
+      "loss": 0.0323,
+      "step": 1066
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 2.201223850250244,
+      "learning_rate": 4.040127385537171e-06,
+      "loss": 0.031,
+      "step": 1067
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.603704571723938,
+      "learning_rate": 4.036661217661901e-06,
+      "loss": 0.0259,
+      "step": 1068
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 2.2893483638763428,
+      "learning_rate": 4.0331934775582774e-06,
+      "loss": 0.0404,
+      "step": 1069
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 2.693096399307251,
+      "learning_rate": 4.029724170485564e-06,
+      "loss": 0.0562,
+      "step": 1070
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.2193703651428223,
+      "learning_rate": 4.026253301705408e-06,
+      "loss": 0.0457,
+      "step": 1071
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.7405165433883667,
+      "learning_rate": 4.022780876481822e-06,
+      "loss": 0.02,
+      "step": 1072
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.5510613918304443,
+      "learning_rate": 4.0193069000811795e-06,
+      "loss": 0.034,
+      "step": 1073
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.5228335857391357,
+      "learning_rate": 4.0158313777722064e-06,
+      "loss": 0.0263,
+      "step": 1074
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.9858816862106323,
+      "learning_rate": 4.012354314825974e-06,
+      "loss": 0.0386,
+      "step": 1075
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 3.7941200733184814,
+      "learning_rate": 4.00887571651589e-06,
+      "loss": 0.0565,
+      "step": 1076
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.6925660371780396,
+      "learning_rate": 4.005395588117689e-06,
+      "loss": 0.0237,
+      "step": 1077
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.8314508199691772,
+      "learning_rate": 4.00191393490943e-06,
+      "loss": 0.0269,
+      "step": 1078
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.896227240562439,
+      "learning_rate": 3.99843076217148e-06,
+      "loss": 0.0289,
+      "step": 1079
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 2.922424554824829,
+      "learning_rate": 3.994946075186513e-06,
+      "loss": 0.0399,
+      "step": 1080
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.300778388977051,
+      "learning_rate": 3.9914598792395e-06,
+      "loss": 0.0302,
+      "step": 1081
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.7674708366394043,
+      "learning_rate": 3.9879721796177e-06,
+      "loss": 0.031,
+      "step": 1082
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.4154834747314453,
+      "learning_rate": 3.984482981610652e-06,
+      "loss": 0.0311,
+      "step": 1083
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 3.0201163291931152,
+      "learning_rate": 3.980992290510167e-06,
+      "loss": 0.0526,
+      "step": 1084
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.4157745838165283,
+      "learning_rate": 3.977500111610322e-06,
+      "loss": 0.0316,
+      "step": 1085
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.0900139808654785,
+      "learning_rate": 3.974006450207449e-06,
+      "loss": 0.0315,
+      "step": 1086
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.017199754714966,
+      "learning_rate": 3.97051131160013e-06,
+      "loss": 0.0237,
+      "step": 1087
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.4157301187515259,
+      "learning_rate": 3.967014701089185e-06,
+      "loss": 0.0223,
+      "step": 1088
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.4726345539093018,
+      "learning_rate": 3.96351662397767e-06,
+      "loss": 0.0192,
+      "step": 1089
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.733639717102051,
+      "learning_rate": 3.960017085570862e-06,
+      "loss": 0.0339,
+      "step": 1090
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.9150118827819824,
+      "learning_rate": 3.956516091176255e-06,
+      "loss": 0.0334,
+      "step": 1091
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.334935188293457,
+      "learning_rate": 3.953013646103552e-06,
+      "loss": 0.0429,
+      "step": 1092
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 3.028473377227783,
+      "learning_rate": 3.949509755664655e-06,
+      "loss": 0.0409,
+      "step": 1093
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 3.481776237487793,
+      "learning_rate": 3.946004425173659e-06,
+      "loss": 0.0242,
+      "step": 1094
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.7127203941345215,
+      "learning_rate": 3.942497659946843e-06,
+      "loss": 0.0261,
+      "step": 1095
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.6174989938735962,
+      "learning_rate": 3.938989465302661e-06,
+      "loss": 0.0237,
+      "step": 1096
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 2.9467060565948486,
+      "learning_rate": 3.935479846561737e-06,
+      "loss": 0.0381,
+      "step": 1097
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 2.425067901611328,
+      "learning_rate": 3.9319688090468516e-06,
+      "loss": 0.031,
+      "step": 1098
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 3.359729290008545,
+      "learning_rate": 3.928456358082941e-06,
+      "loss": 0.0409,
+      "step": 1099
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 2.199843406677246,
+      "learning_rate": 3.924942498997081e-06,
+      "loss": 0.0369,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "eval_blimp_filtered_avg": 0.7308955223880597,
+      "eval_blimp_filtered_std": 0.004880534543081456,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.017553057681552325,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 2.1037707328796387,
+      "learning_rate": 3.921427237118486e-06,
+      "loss": 0.0335,
+      "step": 1101
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.519431710243225,
+      "learning_rate": 3.917910577778498e-06,
+      "loss": 0.0154,
+      "step": 1102
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.3168420791625977,
+      "learning_rate": 3.914392526310575e-06,
+      "loss": 0.0382,
+      "step": 1103
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.1969997882843018,
+      "learning_rate": 3.91087308805029e-06,
+      "loss": 0.0276,
+      "step": 1104
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.2204389572143555,
+      "learning_rate": 3.907352268335319e-06,
+      "loss": 0.0192,
+      "step": 1105
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.109356164932251,
+      "learning_rate": 3.903830072505432e-06,
+      "loss": 0.0284,
+      "step": 1106
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.496445894241333,
+      "learning_rate": 3.900306505902484e-06,
+      "loss": 0.034,
+      "step": 1107
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.8398334980010986,
+      "learning_rate": 3.896781573870412e-06,
+      "loss": 0.0296,
+      "step": 1108
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.636084794998169,
+      "learning_rate": 3.893255281755223e-06,
+      "loss": 0.03,
+      "step": 1109
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.4047505855560303,
+      "learning_rate": 3.88972763490499e-06,
+      "loss": 0.0311,
+      "step": 1110
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.8697782754898071,
+      "learning_rate": 3.886198638669831e-06,
+      "loss": 0.0254,
+      "step": 1111
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.8394330739974976,
+      "learning_rate": 3.882668298401923e-06,
+      "loss": 0.0275,
+      "step": 1112
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.3102149963378906,
+      "learning_rate": 3.8791366194554714e-06,
+      "loss": 0.0319,
+      "step": 1113
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 3.7447218894958496,
+      "learning_rate": 3.875603607186716e-06,
+      "loss": 0.0254,
+      "step": 1114
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.6898393630981445,
+      "learning_rate": 3.872069266953921e-06,
+      "loss": 0.0335,
+      "step": 1115
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 3.0130579471588135,
+      "learning_rate": 3.868533604117359e-06,
+      "loss": 0.0415,
+      "step": 1116
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.545301914215088,
+      "learning_rate": 3.864996624039314e-06,
+      "loss": 0.0336,
+      "step": 1117
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.5601253509521484,
+      "learning_rate": 3.861458332084064e-06,
+      "loss": 0.0286,
+      "step": 1118
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.4592409133911133,
+      "learning_rate": 3.857918733617878e-06,
+      "loss": 0.0261,
+      "step": 1119
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.343907117843628,
+      "learning_rate": 3.854377834009008e-06,
+      "loss": 0.0329,
+      "step": 1120
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.3947246074676514,
+      "learning_rate": 3.8508356386276755e-06,
+      "loss": 0.0447,
+      "step": 1121
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.549394369125366,
+      "learning_rate": 3.847292152846071e-06,
+      "loss": 0.0344,
+      "step": 1122
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.3057541847229004,
+      "learning_rate": 3.84374738203834e-06,
+      "loss": 0.0329,
+      "step": 1123
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 2.4492993354797363,
+      "learning_rate": 3.840201331580579e-06,
+      "loss": 0.0267,
+      "step": 1124
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.582555890083313,
+      "learning_rate": 3.836654006850822e-06,
+      "loss": 0.017,
+      "step": 1125
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.7588337659835815,
+      "learning_rate": 3.8331054132290375e-06,
+      "loss": 0.0318,
+      "step": 1126
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.8603861331939697,
+      "learning_rate": 3.829555556097119e-06,
+      "loss": 0.0207,
+      "step": 1127
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 2.3144755363464355,
+      "learning_rate": 3.826004440838875e-06,
+      "loss": 0.0198,
+      "step": 1128
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 3.8551037311553955,
+      "learning_rate": 3.822452072840023e-06,
+      "loss": 0.04,
+      "step": 1129
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 2.5129449367523193,
+      "learning_rate": 3.81889845748818e-06,
+      "loss": 0.0379,
+      "step": 1130
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 2.312497138977051,
+      "learning_rate": 3.815343600172854e-06,
+      "loss": 0.0345,
+      "step": 1131
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 2.515662670135498,
+      "learning_rate": 3.811787506285439e-06,
+      "loss": 0.0441,
+      "step": 1132
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.6347442865371704,
+      "learning_rate": 3.8082301812192027e-06,
+      "loss": 0.0131,
+      "step": 1133
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.386735200881958,
+      "learning_rate": 3.804671630369279e-06,
+      "loss": 0.0274,
+      "step": 1134
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.267157554626465,
+      "learning_rate": 3.801111859132663e-06,
+      "loss": 0.0312,
+      "step": 1135
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.4939565658569336,
+      "learning_rate": 3.7975508729081997e-06,
+      "loss": 0.0273,
+      "step": 1136
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 3.614370107650757,
+      "learning_rate": 3.7939886770965756e-06,
+      "loss": 0.0484,
+      "step": 1137
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.673005223274231,
+      "learning_rate": 3.7904252771003152e-06,
+      "loss": 0.0256,
+      "step": 1138
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.108344316482544,
+      "learning_rate": 3.7868606783237644e-06,
+      "loss": 0.0268,
+      "step": 1139
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.2001359462738037,
+      "learning_rate": 3.7832948861730926e-06,
+      "loss": 0.0285,
+      "step": 1140
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 4.103482246398926,
+      "learning_rate": 3.7797279060562745e-06,
+      "loss": 0.0362,
+      "step": 1141
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.360914945602417,
+      "learning_rate": 3.7761597433830877e-06,
+      "loss": 0.0362,
+      "step": 1142
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 3.841494560241699,
+      "learning_rate": 3.772590403565107e-06,
+      "loss": 0.0545,
+      "step": 1143
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 3.156510353088379,
+      "learning_rate": 3.7690198920156858e-06,
+      "loss": 0.0231,
+      "step": 1144
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.3391034603118896,
+      "learning_rate": 3.7654482141499597e-06,
+      "loss": 0.0305,
+      "step": 1145
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.9625736474990845,
+      "learning_rate": 3.7618753753848325e-06,
+      "loss": 0.0279,
+      "step": 1146
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.9140450954437256,
+      "learning_rate": 3.758301381138967e-06,
+      "loss": 0.0359,
+      "step": 1147
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.6853861808776855,
+      "learning_rate": 3.754726236832779e-06,
+      "loss": 0.0251,
+      "step": 1148
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.3279006481170654,
+      "learning_rate": 3.75114994788843e-06,
+      "loss": 0.0336,
+      "step": 1149
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 2.346069574356079,
+      "learning_rate": 3.7475725197298136e-06,
+      "loss": 0.0331,
+      "step": 1150
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 2.0036544799804688,
+      "learning_rate": 3.743993957782557e-06,
+      "loss": 0.0334,
+      "step": 1151
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 2.3989925384521484,
+      "learning_rate": 3.740414267474002e-06,
+      "loss": 0.0379,
+      "step": 1152
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.8445849418640137,
+      "learning_rate": 3.7368334542332034e-06,
+      "loss": 0.0272,
+      "step": 1153
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 3.003511667251587,
+      "learning_rate": 3.73325152349092e-06,
+      "loss": 0.052,
+      "step": 1154
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 2.2605748176574707,
+      "learning_rate": 3.7296684806796025e-06,
+      "loss": 0.0314,
+      "step": 1155
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 4.271913528442383,
+      "learning_rate": 3.7260843312333934e-06,
+      "loss": 0.0548,
+      "step": 1156
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 2.337669610977173,
+      "learning_rate": 3.7224990805881077e-06,
+      "loss": 0.0394,
+      "step": 1157
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 2.0145599842071533,
+      "learning_rate": 3.718912734181235e-06,
+      "loss": 0.0289,
+      "step": 1158
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 3.1163170337677,
+      "learning_rate": 3.715325297451924e-06,
+      "loss": 0.0464,
+      "step": 1159
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.3496215343475342,
+      "learning_rate": 3.7117367758409776e-06,
+      "loss": 0.0223,
+      "step": 1160
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.046804428100586,
+      "learning_rate": 3.7081471747908445e-06,
+      "loss": 0.0262,
+      "step": 1161
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.754666805267334,
+      "learning_rate": 3.7045564997456125e-06,
+      "loss": 0.0412,
+      "step": 1162
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 3.0417022705078125,
+      "learning_rate": 3.700964756150994e-06,
+      "loss": 0.0341,
+      "step": 1163
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 3.140596628189087,
+      "learning_rate": 3.697371949454325e-06,
+      "loss": 0.0311,
+      "step": 1164
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.0838050842285156,
+      "learning_rate": 3.693778085104553e-06,
+      "loss": 0.0315,
+      "step": 1165
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.250493049621582,
+      "learning_rate": 3.6901831685522284e-06,
+      "loss": 0.0334,
+      "step": 1166
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.255739688873291,
+      "learning_rate": 3.686587205249501e-06,
+      "loss": 0.0343,
+      "step": 1167
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.6830729246139526,
+      "learning_rate": 3.6829902006501026e-06,
+      "loss": 0.018,
+      "step": 1168
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.8268733024597168,
+      "learning_rate": 3.679392160209351e-06,
+      "loss": 0.0261,
+      "step": 1169
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.903795003890991,
+      "learning_rate": 3.675793089384129e-06,
+      "loss": 0.0394,
+      "step": 1170
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.9097932577133179,
+      "learning_rate": 3.672192993632884e-06,
+      "loss": 0.0232,
+      "step": 1171
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 3.392961025238037,
+      "learning_rate": 3.66859187841562e-06,
+      "loss": 0.0432,
+      "step": 1172
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 2.7188127040863037,
+      "learning_rate": 3.664989749193884e-06,
+      "loss": 0.0365,
+      "step": 1173
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.9038841724395752,
+      "learning_rate": 3.6613866114307637e-06,
+      "loss": 0.0306,
+      "step": 1174
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 2.509185552597046,
+      "learning_rate": 3.657782470590874e-06,
+      "loss": 0.0331,
+      "step": 1175
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.123779058456421,
+      "learning_rate": 3.654177332140352e-06,
+      "loss": 0.0316,
+      "step": 1176
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.072915554046631,
+      "learning_rate": 3.650571201546849e-06,
+      "loss": 0.0266,
+      "step": 1177
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.855332374572754,
+      "learning_rate": 3.6469640842795176e-06,
+      "loss": 0.0336,
+      "step": 1178
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.5203237533569336,
+      "learning_rate": 3.6433559858090118e-06,
+      "loss": 0.0363,
+      "step": 1179
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.272364854812622,
+      "learning_rate": 3.63974691160747e-06,
+      "loss": 0.0179,
+      "step": 1180
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.6612459421157837,
+      "learning_rate": 3.6361368671485114e-06,
+      "loss": 0.0177,
+      "step": 1181
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.958754301071167,
+      "learning_rate": 3.6325258579072265e-06,
+      "loss": 0.045,
+      "step": 1182
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.9610812664031982,
+      "learning_rate": 3.628913889360169e-06,
+      "loss": 0.0324,
+      "step": 1183
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.6242554187774658,
+      "learning_rate": 3.6253009669853477e-06,
+      "loss": 0.021,
+      "step": 1184
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.453378677368164,
+      "learning_rate": 3.62168709626222e-06,
+      "loss": 0.0384,
+      "step": 1185
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.5371179580688477,
+      "learning_rate": 3.6180722826716782e-06,
+      "loss": 0.0247,
+      "step": 1186
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 2.5103209018707275,
+      "learning_rate": 3.614456531696046e-06,
+      "loss": 0.0354,
+      "step": 1187
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.873592734336853,
+      "learning_rate": 3.6108398488190685e-06,
+      "loss": 0.0235,
+      "step": 1188
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 2.4609694480895996,
+      "learning_rate": 3.6072222395259052e-06,
+      "loss": 0.0315,
+      "step": 1189
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 2.4446394443511963,
+      "learning_rate": 3.6036037093031204e-06,
+      "loss": 0.0356,
+      "step": 1190
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 2.9288768768310547,
+      "learning_rate": 3.5999842636386726e-06,
+      "loss": 0.0408,
+      "step": 1191
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.552250862121582,
+      "learning_rate": 3.5963639080219135e-06,
+      "loss": 0.0338,
+      "step": 1192
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.942089557647705,
+      "learning_rate": 3.5927426479435703e-06,
+      "loss": 0.0419,
+      "step": 1193
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.147165298461914,
+      "learning_rate": 3.589120488895743e-06,
+      "loss": 0.023,
+      "step": 1194
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.7635631561279297,
+      "learning_rate": 3.5854974363718977e-06,
+      "loss": 0.0354,
+      "step": 1195
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.5819435119628906,
+      "learning_rate": 3.581873495866852e-06,
+      "loss": 0.039,
+      "step": 1196
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.8446459770202637,
+      "learning_rate": 3.578248672876772e-06,
+      "loss": 0.0284,
+      "step": 1197
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.780993938446045,
+      "learning_rate": 3.574622972899163e-06,
+      "loss": 0.0205,
+      "step": 1198
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 2.9176411628723145,
+      "learning_rate": 3.570996401432858e-06,
+      "loss": 0.0367,
+      "step": 1199
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.895798683166504,
+      "learning_rate": 3.5673689639780144e-06,
+      "loss": 0.0243,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "eval_blimp_filtered_avg": 0.7311940298507462,
+      "eval_blimp_filtered_std": 0.004880626752164257,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.017476319339625877,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 2.6913645267486572,
+      "learning_rate": 3.5637406660361e-06,
+      "loss": 0.0336,
+      "step": 1201
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.467780113220215,
+      "learning_rate": 3.5601115131098904e-06,
+      "loss": 0.0236,
+      "step": 1202
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.4315142631530762,
+      "learning_rate": 3.556481510703457e-06,
+      "loss": 0.0297,
+      "step": 1203
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.4904978275299072,
+      "learning_rate": 3.5528506643221593e-06,
+      "loss": 0.032,
+      "step": 1204
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.8667006492614746,
+      "learning_rate": 3.5492189794726365e-06,
+      "loss": 0.0291,
+      "step": 1205
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.2573022842407227,
+      "learning_rate": 3.5455864616627994e-06,
+      "loss": 0.0328,
+      "step": 1206
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.6707193851470947,
+      "learning_rate": 3.541953116401822e-06,
+      "loss": 0.0409,
+      "step": 1207
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 3.2035815715789795,
+      "learning_rate": 3.5383189492001358e-06,
+      "loss": 0.0341,
+      "step": 1208
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.7268242835998535,
+      "learning_rate": 3.5346839655694153e-06,
+      "loss": 0.0196,
+      "step": 1209
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.8853451013565063,
+      "learning_rate": 3.5310481710225744e-06,
+      "loss": 0.0223,
+      "step": 1210
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.1230509281158447,
+      "learning_rate": 3.527411571073758e-06,
+      "loss": 0.0344,
+      "step": 1211
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.6838618516921997,
+      "learning_rate": 3.523774171238331e-06,
+      "loss": 0.0291,
+      "step": 1212
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.1772899627685547,
+      "learning_rate": 3.520135977032873e-06,
+      "loss": 0.0286,
+      "step": 1213
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.9106106758117676,
+      "learning_rate": 3.5164969939751654e-06,
+      "loss": 0.0345,
+      "step": 1214
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 3.0209720134735107,
+      "learning_rate": 3.5128572275841912e-06,
+      "loss": 0.0312,
+      "step": 1215
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.7701351642608643,
+      "learning_rate": 3.509216683380115e-06,
+      "loss": 0.0364,
+      "step": 1216
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.7230111360549927,
+      "learning_rate": 3.5055753668842865e-06,
+      "loss": 0.0283,
+      "step": 1217
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.6088255643844604,
+      "learning_rate": 3.5019332836192243e-06,
+      "loss": 0.0249,
+      "step": 1218
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.5892186164855957,
+      "learning_rate": 3.4982904391086084e-06,
+      "loss": 0.0189,
+      "step": 1219
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 3.2972137928009033,
+      "learning_rate": 3.494646838877277e-06,
+      "loss": 0.0398,
+      "step": 1220
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.8599462509155273,
+      "learning_rate": 3.4910024884512117e-06,
+      "loss": 0.0297,
+      "step": 1221
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.5254584550857544,
+      "learning_rate": 3.4873573933575324e-06,
+      "loss": 0.0217,
+      "step": 1222
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.3779330253601074,
+      "learning_rate": 3.4837115591244883e-06,
+      "loss": 0.0345,
+      "step": 1223
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.760943651199341,
+      "learning_rate": 3.4800649912814505e-06,
+      "loss": 0.0351,
+      "step": 1224
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.230483055114746,
+      "learning_rate": 3.4764176953589016e-06,
+      "loss": 0.0389,
+      "step": 1225
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.891994833946228,
+      "learning_rate": 3.47276967688843e-06,
+      "loss": 0.0294,
+      "step": 1226
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.8476885557174683,
+      "learning_rate": 3.4691209414027184e-06,
+      "loss": 0.0215,
+      "step": 1227
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.9772863388061523,
+      "learning_rate": 3.4654714944355365e-06,
+      "loss": 0.0182,
+      "step": 1228
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.5737272500991821,
+      "learning_rate": 3.4618213415217357e-06,
+      "loss": 0.0197,
+      "step": 1229
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.8838216066360474,
+      "learning_rate": 3.458170488197235e-06,
+      "loss": 0.0253,
+      "step": 1230
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 2.3223137855529785,
+      "learning_rate": 3.4545189399990172e-06,
+      "loss": 0.0263,
+      "step": 1231
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 2.715822696685791,
+      "learning_rate": 3.450866702465119e-06,
+      "loss": 0.0362,
+      "step": 1232
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 2.264216899871826,
+      "learning_rate": 3.4472137811346216e-06,
+      "loss": 0.027,
+      "step": 1233
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.7976338863372803,
+      "learning_rate": 3.443560181547646e-06,
+      "loss": 0.0429,
+      "step": 1234
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.111776351928711,
+      "learning_rate": 3.439905909245337e-06,
+      "loss": 0.0179,
+      "step": 1235
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.551102876663208,
+      "learning_rate": 3.4362509697698652e-06,
+      "loss": 0.0173,
+      "step": 1236
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 3.3495659828186035,
+      "learning_rate": 3.432595368664408e-06,
+      "loss": 0.0496,
+      "step": 1237
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 3.659907579421997,
+      "learning_rate": 3.428939111473151e-06,
+      "loss": 0.0373,
+      "step": 1238
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.6961734294891357,
+      "learning_rate": 3.425282203741271e-06,
+      "loss": 0.037,
+      "step": 1239
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.8081846237182617,
+      "learning_rate": 3.421624651014932e-06,
+      "loss": 0.0281,
+      "step": 1240
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.495588541030884,
+      "learning_rate": 3.417966458841279e-06,
+      "loss": 0.0412,
+      "step": 1241
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.7064850330352783,
+      "learning_rate": 3.4143076327684243e-06,
+      "loss": 0.0203,
+      "step": 1242
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.007570266723633,
+      "learning_rate": 3.410648178345442e-06,
+      "loss": 0.0325,
+      "step": 1243
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 3.4155590534210205,
+      "learning_rate": 3.4069881011223595e-06,
+      "loss": 0.0398,
+      "step": 1244
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.3245530128479004,
+      "learning_rate": 3.403327406650147e-06,
+      "loss": 0.0286,
+      "step": 1245
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.212587833404541,
+      "learning_rate": 3.399666100480714e-06,
+      "loss": 0.04,
+      "step": 1246
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.391765832901001,
+      "learning_rate": 3.396004188166898e-06,
+      "loss": 0.0309,
+      "step": 1247
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.2388553619384766,
+      "learning_rate": 3.39234167526245e-06,
+      "loss": 0.0395,
+      "step": 1248
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.7749336957931519,
+      "learning_rate": 3.3886785673220387e-06,
+      "loss": 0.0198,
+      "step": 1249
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.6987144947052002,
+      "learning_rate": 3.385014869901232e-06,
+      "loss": 0.0217,
+      "step": 1250
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.110356569290161,
+      "learning_rate": 3.3813505885564907e-06,
+      "loss": 0.0316,
+      "step": 1251
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.521801710128784,
+      "learning_rate": 3.3776857288451646e-06,
+      "loss": 0.0382,
+      "step": 1252
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.0271427631378174,
+      "learning_rate": 3.3740202963254772e-06,
+      "loss": 0.0291,
+      "step": 1253
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.2707107067108154,
+      "learning_rate": 3.3703542965565237e-06,
+      "loss": 0.0189,
+      "step": 1254
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.5131962299346924,
+      "learning_rate": 3.3666877350982566e-06,
+      "loss": 0.0285,
+      "step": 1255
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.301226854324341,
+      "learning_rate": 3.363020617511483e-06,
+      "loss": 0.0346,
+      "step": 1256
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.744776725769043,
+      "learning_rate": 3.3593529493578522e-06,
+      "loss": 0.0247,
+      "step": 1257
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.5976932048797607,
+      "learning_rate": 3.3556847361998477e-06,
+      "loss": 0.0497,
+      "step": 1258
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.4436731338500977,
+      "learning_rate": 3.3520159836007817e-06,
+      "loss": 0.0392,
+      "step": 1259
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.1173036098480225,
+      "learning_rate": 3.3483466971247833e-06,
+      "loss": 0.0288,
+      "step": 1260
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.0751702785491943,
+      "learning_rate": 3.344676882336791e-06,
+      "loss": 0.0264,
+      "step": 1261
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.9291141033172607,
+      "learning_rate": 3.341006544802545e-06,
+      "loss": 0.0294,
+      "step": 1262
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.6938070058822632,
+      "learning_rate": 3.3373356900885767e-06,
+      "loss": 0.0296,
+      "step": 1263
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.0150370597839355,
+      "learning_rate": 3.3336643237622054e-06,
+      "loss": 0.0262,
+      "step": 1264
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.478817105293274,
+      "learning_rate": 3.3299924513915233e-06,
+      "loss": 0.0252,
+      "step": 1265
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.69711434841156,
+      "learning_rate": 3.326320078545392e-06,
+      "loss": 0.0228,
+      "step": 1266
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.662680745124817,
+      "learning_rate": 3.322647210793429e-06,
+      "loss": 0.0285,
+      "step": 1267
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.8895726203918457,
+      "learning_rate": 3.3189738537060064e-06,
+      "loss": 0.0303,
+      "step": 1268
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.9535151720046997,
+      "learning_rate": 3.3153000128542356e-06,
+      "loss": 0.0354,
+      "step": 1269
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 2.2534191608428955,
+      "learning_rate": 3.3116256938099648e-06,
+      "loss": 0.0207,
+      "step": 1270
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.8271726369857788,
+      "learning_rate": 3.3079509021457624e-06,
+      "loss": 0.0274,
+      "step": 1271
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.150408983230591,
+      "learning_rate": 3.3042756434349185e-06,
+      "loss": 0.0263,
+      "step": 1272
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.9939157962799072,
+      "learning_rate": 3.3005999232514295e-06,
+      "loss": 0.0284,
+      "step": 1273
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.3205487728118896,
+      "learning_rate": 3.2969237471699906e-06,
+      "loss": 0.0331,
+      "step": 1274
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.1258716583251953,
+      "learning_rate": 3.29324712076599e-06,
+      "loss": 0.0342,
+      "step": 1275
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 2.927276134490967,
+      "learning_rate": 3.289570049615499e-06,
+      "loss": 0.0389,
+      "step": 1276
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.589333415031433,
+      "learning_rate": 3.2858925392952624e-06,
+      "loss": 0.0169,
+      "step": 1277
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 2.6529407501220703,
+      "learning_rate": 3.2822145953826922e-06,
+      "loss": 0.0387,
+      "step": 1278
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 2.3025693893432617,
+      "learning_rate": 3.2785362234558556e-06,
+      "loss": 0.0181,
+      "step": 1279
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 2.323138475418091,
+      "learning_rate": 3.2748574290934724e-06,
+      "loss": 0.0159,
+      "step": 1280
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 2.28291916847229,
+      "learning_rate": 3.2711782178749005e-06,
+      "loss": 0.0212,
+      "step": 1281
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.7579914331436157,
+      "learning_rate": 3.2674985953801307e-06,
+      "loss": 0.0174,
+      "step": 1282
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 2.316755533218384,
+      "learning_rate": 3.2638185671897795e-06,
+      "loss": 0.0216,
+      "step": 1283
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 2.3650054931640625,
+      "learning_rate": 3.260138138885075e-06,
+      "loss": 0.0333,
+      "step": 1284
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.9737592935562134,
+      "learning_rate": 3.2564573160478542e-06,
+      "loss": 0.017,
+      "step": 1285
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 2.7900679111480713,
+      "learning_rate": 3.2527761042605517e-06,
+      "loss": 0.0462,
+      "step": 1286
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.5225279331207275,
+      "learning_rate": 3.2490945091061936e-06,
+      "loss": 0.0267,
+      "step": 1287
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.5343048572540283,
+      "learning_rate": 3.2454125361683872e-06,
+      "loss": 0.0326,
+      "step": 1288
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.310825824737549,
+      "learning_rate": 3.2417301910313103e-06,
+      "loss": 0.0349,
+      "step": 1289
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.7550227642059326,
+      "learning_rate": 3.2380474792797078e-06,
+      "loss": 0.0257,
+      "step": 1290
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.7225751876831055,
+      "learning_rate": 3.2343644064988793e-06,
+      "loss": 0.0316,
+      "step": 1291
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 2.3662021160125732,
+      "learning_rate": 3.2306809782746713e-06,
+      "loss": 0.0301,
+      "step": 1292
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 2.3651530742645264,
+      "learning_rate": 3.2269972001934723e-06,
+      "loss": 0.0282,
+      "step": 1293
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 2.344818115234375,
+      "learning_rate": 3.223313077842198e-06,
+      "loss": 0.0362,
+      "step": 1294
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 2.5864577293395996,
+      "learning_rate": 3.219628616808288e-06,
+      "loss": 0.0401,
+      "step": 1295
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.993554711341858,
+      "learning_rate": 3.215943822679697e-06,
+      "loss": 0.0212,
+      "step": 1296
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 2.171398162841797,
+      "learning_rate": 3.2122587010448798e-06,
+      "loss": 0.03,
+      "step": 1297
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 2.290483236312866,
+      "learning_rate": 3.2085732574927944e-06,
+      "loss": 0.0233,
+      "step": 1298
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 2.4070208072662354,
+      "learning_rate": 3.204887497612881e-06,
+      "loss": 0.0261,
+      "step": 1299
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 2.913848876953125,
+      "learning_rate": 3.2012014269950653e-06,
+      "loss": 0.0357,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "eval_blimp_filtered_avg": 0.7295522388059702,
+      "eval_blimp_filtered_std": 0.004884060300291912,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.01758637940432669,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "eval_vqa_filtered_avg": 0.29,
+      "eval_vqa_filtered_std": 0.045604802157206845,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956911,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.5696524381637573,
+      "learning_rate": 3.1975150512297403e-06,
+      "loss": 0.0153,
+      "step": 1301
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.996586799621582,
+      "learning_rate": 3.193828375907761e-06,
+      "loss": 0.0291,
+      "step": 1302
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.8274586200714111,
+      "learning_rate": 3.1901414066204417e-06,
+      "loss": 0.0234,
+      "step": 1303
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.165045976638794,
+      "learning_rate": 3.186454148959537e-06,
+      "loss": 0.0292,
+      "step": 1304
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.4775798320770264,
+      "learning_rate": 3.1827666085172426e-06,
+      "loss": 0.0289,
+      "step": 1305
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.625504970550537,
+      "learning_rate": 3.1790787908861812e-06,
+      "loss": 0.023,
+      "step": 1306
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.9473897218704224,
+      "learning_rate": 3.175390701659397e-06,
+      "loss": 0.0229,
+      "step": 1307
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 2.007969379425049,
+      "learning_rate": 3.171702346430345e-06,
+      "loss": 0.0267,
+      "step": 1308
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 2.2145326137542725,
+      "learning_rate": 3.168013730792885e-06,
+      "loss": 0.0245,
+      "step": 1309
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.7361934185028076,
+      "learning_rate": 3.1643248603412695e-06,
+      "loss": 0.0339,
+      "step": 1310
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 2.399824857711792,
+      "learning_rate": 3.160635740670141e-06,
+      "loss": 0.0318,
+      "step": 1311
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.8164557218551636,
+      "learning_rate": 3.1569463773745162e-06,
+      "loss": 0.0348,
+      "step": 1312
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.9061485528945923,
+      "learning_rate": 3.153256776049783e-06,
+      "loss": 0.0277,
+      "step": 1313
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.476215124130249,
+      "learning_rate": 3.1495669422916915e-06,
+      "loss": 0.0281,
+      "step": 1314
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.397589921951294,
+      "learning_rate": 3.1458768816963413e-06,
+      "loss": 0.0296,
+      "step": 1315
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.6021326780319214,
+      "learning_rate": 3.1421865998601793e-06,
+      "loss": 0.018,
+      "step": 1316
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.5362943410873413,
+      "learning_rate": 3.1384961023799855e-06,
+      "loss": 0.0216,
+      "step": 1317
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.8499761819839478,
+      "learning_rate": 3.134805394852869e-06,
+      "loss": 0.0236,
+      "step": 1318
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.40010929107666,
+      "learning_rate": 3.1311144828762554e-06,
+      "loss": 0.0311,
+      "step": 1319
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.0220916271209717,
+      "learning_rate": 3.127423372047881e-06,
+      "loss": 0.0354,
+      "step": 1320
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.036777973175049,
+      "learning_rate": 3.123732067965785e-06,
+      "loss": 0.0236,
+      "step": 1321
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 3.1668882369995117,
+      "learning_rate": 3.1200405762282983e-06,
+      "loss": 0.0466,
+      "step": 1322
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.066866636276245,
+      "learning_rate": 3.116348902434037e-06,
+      "loss": 0.0228,
+      "step": 1323
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.426096200942993,
+      "learning_rate": 3.1126570521818926e-06,
+      "loss": 0.0372,
+      "step": 1324
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.0556159019470215,
+      "learning_rate": 3.108965031071025e-06,
+      "loss": 0.0307,
+      "step": 1325
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.7965686321258545,
+      "learning_rate": 3.1052728447008523e-06,
+      "loss": 0.0329,
+      "step": 1326
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.952653169631958,
+      "learning_rate": 3.1015804986710465e-06,
+      "loss": 0.0315,
+      "step": 1327
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.534970283508301,
+      "learning_rate": 3.0978879985815155e-06,
+      "loss": 0.0379,
+      "step": 1328
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.9477201700210571,
+      "learning_rate": 3.0941953500324073e-06,
+      "loss": 0.0265,
+      "step": 1329
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 2.6667840480804443,
+      "learning_rate": 3.0905025586240922e-06,
+      "loss": 0.0266,
+      "step": 1330
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 2.0272104740142822,
+      "learning_rate": 3.0868096299571543e-06,
+      "loss": 0.0186,
+      "step": 1331
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 2.069148302078247,
+      "learning_rate": 3.0831165696323923e-06,
+      "loss": 0.0248,
+      "step": 1332
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 3.0360147953033447,
+      "learning_rate": 3.0794233832507983e-06,
+      "loss": 0.0343,
+      "step": 1333
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.464059829711914,
+      "learning_rate": 3.0757300764135608e-06,
+      "loss": 0.0314,
+      "step": 1334
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.568377137184143,
+      "learning_rate": 3.072036654722047e-06,
+      "loss": 0.0218,
+      "step": 1335
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.1846024990081787,
+      "learning_rate": 3.0683431237778e-06,
+      "loss": 0.0187,
+      "step": 1336
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.253002882003784,
+      "learning_rate": 3.064649489182528e-06,
+      "loss": 0.0388,
+      "step": 1337
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.457123279571533,
+      "learning_rate": 3.0609557565380985e-06,
+      "loss": 0.0347,
+      "step": 1338
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.2878243923187256,
+      "learning_rate": 3.057261931446524e-06,
+      "loss": 0.038,
+      "step": 1339
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 4.011155605316162,
+      "learning_rate": 3.053568019509961e-06,
+      "loss": 0.0431,
+      "step": 1340
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.7179391384124756,
+      "learning_rate": 3.049874026330695e-06,
+      "loss": 0.0383,
+      "step": 1341
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.6705360412597656,
+      "learning_rate": 3.0461799575111363e-06,
+      "loss": 0.0308,
+      "step": 1342
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.940155506134033,
+      "learning_rate": 3.042485818653808e-06,
+      "loss": 0.042,
+      "step": 1343
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.971364736557007,
+      "learning_rate": 3.038791615361341e-06,
+      "loss": 0.0332,
+      "step": 1344
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.429924726486206,
+      "learning_rate": 3.0350973532364653e-06,
+      "loss": 0.0349,
+      "step": 1345
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.252393960952759,
+      "learning_rate": 3.0314030378819974e-06,
+      "loss": 0.0212,
+      "step": 1346
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.559187173843384,
+      "learning_rate": 3.0277086749008345e-06,
+      "loss": 0.0165,
+      "step": 1347
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.0490100383758545,
+      "learning_rate": 3.0240142698959484e-06,
+      "loss": 0.032,
+      "step": 1348
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.9401170015335083,
+      "learning_rate": 3.0203198284703727e-06,
+      "loss": 0.0308,
+      "step": 1349
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.6361333131790161,
+      "learning_rate": 3.016625356227198e-06,
+      "loss": 0.026,
+      "step": 1350
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 2.661823034286499,
+      "learning_rate": 3.012930858769559e-06,
+      "loss": 0.0373,
+      "step": 1351
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.4378269910812378,
+      "learning_rate": 3.0092363417006313e-06,
+      "loss": 0.0178,
+      "step": 1352
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 2.407885789871216,
+      "learning_rate": 3.0055418106236192e-06,
+      "loss": 0.0181,
+      "step": 1353
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 2.1674129962921143,
+      "learning_rate": 3.001847271141747e-06,
+      "loss": 0.0305,
+      "step": 1354
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.315542697906494,
+      "learning_rate": 2.998152728858253e-06,
+      "loss": 0.0316,
+      "step": 1355
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.274592399597168,
+      "learning_rate": 2.9944581893763813e-06,
+      "loss": 0.0195,
+      "step": 1356
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.0330843925476074,
+      "learning_rate": 2.990763658299369e-06,
+      "loss": 0.0256,
+      "step": 1357
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.4307626485824585,
+      "learning_rate": 2.9870691412304407e-06,
+      "loss": 0.0215,
+      "step": 1358
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.7875858545303345,
+      "learning_rate": 2.9833746437728025e-06,
+      "loss": 0.0225,
+      "step": 1359
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.4306483268737793,
+      "learning_rate": 2.979680171529628e-06,
+      "loss": 0.0373,
+      "step": 1360
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 3.135990619659424,
+      "learning_rate": 2.975985730104053e-06,
+      "loss": 0.0444,
+      "step": 1361
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.6599440574645996,
+      "learning_rate": 2.9722913250991656e-06,
+      "loss": 0.0329,
+      "step": 1362
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.6130428314208984,
+      "learning_rate": 2.9685969621180036e-06,
+      "loss": 0.0342,
+      "step": 1363
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.042341947555542,
+      "learning_rate": 2.9649026467635352e-06,
+      "loss": 0.0327,
+      "step": 1364
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.4063141345977783,
+      "learning_rate": 2.9612083846386587e-06,
+      "loss": 0.0234,
+      "step": 1365
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.4121804237365723,
+      "learning_rate": 2.9575141813461927e-06,
+      "loss": 0.0322,
+      "step": 1366
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.4911248683929443,
+      "learning_rate": 2.953820042488865e-06,
+      "loss": 0.0441,
+      "step": 1367
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.4976234436035156,
+      "learning_rate": 2.950125973669305e-06,
+      "loss": 0.0333,
+      "step": 1368
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.2136433124542236,
+      "learning_rate": 2.9464319804900393e-06,
+      "loss": 0.0303,
+      "step": 1369
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.283283233642578,
+      "learning_rate": 2.9427380685534766e-06,
+      "loss": 0.0343,
+      "step": 1370
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.5114150047302246,
+      "learning_rate": 2.9390442434619017e-06,
+      "loss": 0.0331,
+      "step": 1371
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.685104250907898,
+      "learning_rate": 2.935350510817472e-06,
+      "loss": 0.0343,
+      "step": 1372
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.1591382026672363,
+      "learning_rate": 2.931656876222201e-06,
+      "loss": 0.0256,
+      "step": 1373
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.493062138557434,
+      "learning_rate": 2.927963345277954e-06,
+      "loss": 0.0181,
+      "step": 1374
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.6076879501342773,
+      "learning_rate": 2.92426992358644e-06,
+      "loss": 0.0393,
+      "step": 1375
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.618229627609253,
+      "learning_rate": 2.9205766167492023e-06,
+      "loss": 0.0186,
+      "step": 1376
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.188441753387451,
+      "learning_rate": 2.916883430367609e-06,
+      "loss": 0.039,
+      "step": 1377
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.861607313156128,
+      "learning_rate": 2.9131903700428454e-06,
+      "loss": 0.0334,
+      "step": 1378
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.706430673599243,
+      "learning_rate": 2.9094974413759088e-06,
+      "loss": 0.0367,
+      "step": 1379
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.672380208969116,
+      "learning_rate": 2.905804649967593e-06,
+      "loss": 0.0233,
+      "step": 1380
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.9241265058517456,
+      "learning_rate": 2.902112001418484e-06,
+      "loss": 0.03,
+      "step": 1381
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.822526216506958,
+      "learning_rate": 2.898419501328954e-06,
+      "loss": 0.0327,
+      "step": 1382
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 2.4114909172058105,
+      "learning_rate": 2.894727155299148e-06,
+      "loss": 0.0288,
+      "step": 1383
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 2.5974793434143066,
+      "learning_rate": 2.891034968928975e-06,
+      "loss": 0.0315,
+      "step": 1384
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.1774488687515259,
+      "learning_rate": 2.8873429478181075e-06,
+      "loss": 0.0177,
+      "step": 1385
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.1663141250610352,
+      "learning_rate": 2.8836510975659637e-06,
+      "loss": 0.0208,
+      "step": 1386
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.8016681671142578,
+      "learning_rate": 2.8799594237717027e-06,
+      "loss": 0.0214,
+      "step": 1387
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 2.0615158081054688,
+      "learning_rate": 2.8762679320342155e-06,
+      "loss": 0.0237,
+      "step": 1388
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 2.319617986679077,
+      "learning_rate": 2.87257662795212e-06,
+      "loss": 0.0289,
+      "step": 1389
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 3.3268957138061523,
+      "learning_rate": 2.868885517123746e-06,
+      "loss": 0.0394,
+      "step": 1390
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 2.795973062515259,
+      "learning_rate": 2.8651946051471317e-06,
+      "loss": 0.0344,
+      "step": 1391
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 2.0799996852874756,
+      "learning_rate": 2.861503897620015e-06,
+      "loss": 0.0335,
+      "step": 1392
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.8418042659759521,
+      "learning_rate": 2.8578134001398217e-06,
+      "loss": 0.0171,
+      "step": 1393
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.5691992044448853,
+      "learning_rate": 2.8541231183036585e-06,
+      "loss": 0.0233,
+      "step": 1394
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 2.493579864501953,
+      "learning_rate": 2.8504330577083095e-06,
+      "loss": 0.0455,
+      "step": 1395
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 2.0936388969421387,
+      "learning_rate": 2.846743223950217e-06,
+      "loss": 0.0196,
+      "step": 1396
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 2.1416659355163574,
+      "learning_rate": 2.843053622625484e-06,
+      "loss": 0.0287,
+      "step": 1397
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.5485296249389648,
+      "learning_rate": 2.839364259329859e-06,
+      "loss": 0.0202,
+      "step": 1398
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.6313397884368896,
+      "learning_rate": 2.8356751396587307e-06,
+      "loss": 0.0259,
+      "step": 1399
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 2.4532973766326904,
+      "learning_rate": 2.831986269207116e-06,
+      "loss": 0.0342,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "eval_blimp_filtered_avg": 0.7304477611940299,
+      "eval_blimp_filtered_std": 0.004876729399130613,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.017444301474488378,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "eval_vqa_filtered_avg": 0.29,
+      "eval_vqa_filtered_std": 0.045604802157206845,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 3.0425612926483154,
+      "learning_rate": 2.8282976535696546e-06,
+      "loss": 0.0383,
+      "step": 1401
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.2351620197296143,
+      "learning_rate": 2.824609298340603e-06,
+      "loss": 0.0297,
+      "step": 1402
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.3294379711151123,
+      "learning_rate": 2.8209212091138193e-06,
+      "loss": 0.0186,
+      "step": 1403
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.571092128753662,
+      "learning_rate": 2.8172333914827575e-06,
+      "loss": 0.0178,
+      "step": 1404
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.007080554962158,
+      "learning_rate": 2.813545851040464e-06,
+      "loss": 0.0216,
+      "step": 1405
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.263432264328003,
+      "learning_rate": 2.8098585933795597e-06,
+      "loss": 0.0256,
+      "step": 1406
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.406308889389038,
+      "learning_rate": 2.8061716240922384e-06,
+      "loss": 0.031,
+      "step": 1407
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 2.354466199874878,
+      "learning_rate": 2.8024849487702607e-06,
+      "loss": 0.0427,
+      "step": 1408
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 2.0996108055114746,
+      "learning_rate": 2.7987985730049352e-06,
+      "loss": 0.0271,
+      "step": 1409
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 2.51289701461792,
+      "learning_rate": 2.7951125023871194e-06,
+      "loss": 0.0285,
+      "step": 1410
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.9438964128494263,
+      "learning_rate": 2.7914267425072066e-06,
+      "loss": 0.0198,
+      "step": 1411
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.4516046047210693,
+      "learning_rate": 2.7877412989551204e-06,
+      "loss": 0.0231,
+      "step": 1412
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 2.3531408309936523,
+      "learning_rate": 2.7840561773203046e-06,
+      "loss": 0.0374,
+      "step": 1413
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 2.4776668548583984,
+      "learning_rate": 2.780371383191712e-06,
+      "loss": 0.0335,
+      "step": 1414
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.781406044960022,
+      "learning_rate": 2.776686922157803e-06,
+      "loss": 0.0328,
+      "step": 1415
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 2.1665971279144287,
+      "learning_rate": 2.773002799806529e-06,
+      "loss": 0.0329,
+      "step": 1416
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 2.040463447570801,
+      "learning_rate": 2.769319021725329e-06,
+      "loss": 0.0283,
+      "step": 1417
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.862982988357544,
+      "learning_rate": 2.7656355935011217e-06,
+      "loss": 0.0298,
+      "step": 1418
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 2.255699634552002,
+      "learning_rate": 2.7619525207202932e-06,
+      "loss": 0.0305,
+      "step": 1419
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 2.0203378200531006,
+      "learning_rate": 2.75826980896869e-06,
+      "loss": 0.0201,
+      "step": 1420
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.6333909034729004,
+      "learning_rate": 2.7545874638316134e-06,
+      "loss": 0.015,
+      "step": 1421
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 4.09716272354126,
+      "learning_rate": 2.7509054908938065e-06,
+      "loss": 0.0362,
+      "step": 1422
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 3.3731162548065186,
+      "learning_rate": 2.7472238957394493e-06,
+      "loss": 0.0265,
+      "step": 1423
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.0378575325012207,
+      "learning_rate": 2.7435426839521464e-06,
+      "loss": 0.0203,
+      "step": 1424
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.9738328456878662,
+      "learning_rate": 2.7398618611149257e-06,
+      "loss": 0.0378,
+      "step": 1425
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.6407220363616943,
+      "learning_rate": 2.736181432810221e-06,
+      "loss": 0.0261,
+      "step": 1426
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.5810182094573975,
+      "learning_rate": 2.7325014046198686e-06,
+      "loss": 0.0374,
+      "step": 1427
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.7155685424804688,
+      "learning_rate": 2.7288217821251e-06,
+      "loss": 0.0384,
+      "step": 1428
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 2.101759433746338,
+      "learning_rate": 2.725142570906528e-06,
+      "loss": 0.0265,
+      "step": 1429
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.404998779296875,
+      "learning_rate": 2.721463776544144e-06,
+      "loss": 0.0205,
+      "step": 1430
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 2.158501148223877,
+      "learning_rate": 2.7177854046173083e-06,
+      "loss": 0.0261,
+      "step": 1431
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 2.305915355682373,
+      "learning_rate": 2.714107460704738e-06,
+      "loss": 0.0456,
+      "step": 1432
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.6299773454666138,
+      "learning_rate": 2.7104299503845012e-06,
+      "loss": 0.0212,
+      "step": 1433
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.4394813776016235,
+      "learning_rate": 2.70675287923401e-06,
+      "loss": 0.0196,
+      "step": 1434
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.8260653018951416,
+      "learning_rate": 2.7030762528300104e-06,
+      "loss": 0.0295,
+      "step": 1435
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.252345323562622,
+      "learning_rate": 2.6994000767485715e-06,
+      "loss": 0.03,
+      "step": 1436
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.9318733215332031,
+      "learning_rate": 2.6957243565650816e-06,
+      "loss": 0.0292,
+      "step": 1437
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.3213067054748535,
+      "learning_rate": 2.692049097854238e-06,
+      "loss": 0.0296,
+      "step": 1438
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 4.6879730224609375,
+      "learning_rate": 2.6883743061900366e-06,
+      "loss": 0.0396,
+      "step": 1439
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 2.6184113025665283,
+      "learning_rate": 2.6846999871457637e-06,
+      "loss": 0.0226,
+      "step": 1440
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 2.027442216873169,
+      "learning_rate": 2.681026146293994e-06,
+      "loss": 0.0231,
+      "step": 1441
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.6485204696655273,
+      "learning_rate": 2.6773527892065716e-06,
+      "loss": 0.0296,
+      "step": 1442
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.9452016353607178,
+      "learning_rate": 2.673679921454609e-06,
+      "loss": 0.018,
+      "step": 1443
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.3754701614379883,
+      "learning_rate": 2.6700075486084773e-06,
+      "loss": 0.0218,
+      "step": 1444
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 3.1341874599456787,
+      "learning_rate": 2.6663356762377956e-06,
+      "loss": 0.0388,
+      "step": 1445
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.8304383754730225,
+      "learning_rate": 2.6626643099114227e-06,
+      "loss": 0.0182,
+      "step": 1446
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.611701011657715,
+      "learning_rate": 2.6589934551974556e-06,
+      "loss": 0.0245,
+      "step": 1447
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.902561664581299,
+      "learning_rate": 2.655323117663209e-06,
+      "loss": 0.0293,
+      "step": 1448
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.7010092735290527,
+      "learning_rate": 2.651653302875217e-06,
+      "loss": 0.0401,
+      "step": 1449
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.272813320159912,
+      "learning_rate": 2.647984016399218e-06,
+      "loss": 0.0326,
+      "step": 1450
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.045222043991089,
+      "learning_rate": 2.6443152638001525e-06,
+      "loss": 0.0332,
+      "step": 1451
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.6979987621307373,
+      "learning_rate": 2.640647050642149e-06,
+      "loss": 0.0374,
+      "step": 1452
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.1752326488494873,
+      "learning_rate": 2.636979382488517e-06,
+      "loss": 0.0204,
+      "step": 1453
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.9497692584991455,
+      "learning_rate": 2.6333122649017435e-06,
+      "loss": 0.0206,
+      "step": 1454
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.2767921686172485,
+      "learning_rate": 2.6296457034434773e-06,
+      "loss": 0.0181,
+      "step": 1455
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 2.1974453926086426,
+      "learning_rate": 2.625979703674523e-06,
+      "loss": 0.0325,
+      "step": 1456
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.9323058128356934,
+      "learning_rate": 2.622314271154836e-06,
+      "loss": 0.0282,
+      "step": 1457
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.8675453662872314,
+      "learning_rate": 2.6186494114435094e-06,
+      "loss": 0.0252,
+      "step": 1458
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 2.5531086921691895,
+      "learning_rate": 2.614985130098769e-06,
+      "loss": 0.0274,
+      "step": 1459
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.5791561603546143,
+      "learning_rate": 2.6113214326779615e-06,
+      "loss": 0.0247,
+      "step": 1460
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.7262338399887085,
+      "learning_rate": 2.60765832473755e-06,
+      "loss": 0.022,
+      "step": 1461
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.8429864645004272,
+      "learning_rate": 2.6039958118331036e-06,
+      "loss": 0.0229,
+      "step": 1462
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.6144708395004272,
+      "learning_rate": 2.600333899519285e-06,
+      "loss": 0.0178,
+      "step": 1463
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 2.0176966190338135,
+      "learning_rate": 2.596672593349853e-06,
+      "loss": 0.0329,
+      "step": 1464
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.9623013734817505,
+      "learning_rate": 2.593011898877642e-06,
+      "loss": 0.0206,
+      "step": 1465
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.524536371231079,
+      "learning_rate": 2.589351821654559e-06,
+      "loss": 0.019,
+      "step": 1466
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.480729341506958,
+      "learning_rate": 2.5856923672315763e-06,
+      "loss": 0.0191,
+      "step": 1467
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.8096470832824707,
+      "learning_rate": 2.5820335411587217e-06,
+      "loss": 0.0269,
+      "step": 1468
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 3.9931509494781494,
+      "learning_rate": 2.5783753489850677e-06,
+      "loss": 0.0432,
+      "step": 1469
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 2.0325093269348145,
+      "learning_rate": 2.5747177962587295e-06,
+      "loss": 0.0263,
+      "step": 1470
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.201650619506836,
+      "learning_rate": 2.5710608885268497e-06,
+      "loss": 0.0317,
+      "step": 1471
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.5067620277404785,
+      "learning_rate": 2.567404631335592e-06,
+      "loss": 0.0337,
+      "step": 1472
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.1018080711364746,
+      "learning_rate": 2.5637490302301354e-06,
+      "loss": 0.0152,
+      "step": 1473
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.3227744102478027,
+      "learning_rate": 2.5600940907546634e-06,
+      "loss": 0.0119,
+      "step": 1474
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.1729917526245117,
+      "learning_rate": 2.5564398184523554e-06,
+      "loss": 0.0346,
+      "step": 1475
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 3.027639150619507,
+      "learning_rate": 2.5527862188653777e-06,
+      "loss": 0.0338,
+      "step": 1476
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.9144630432128906,
+      "learning_rate": 2.5491332975348812e-06,
+      "loss": 0.0141,
+      "step": 1477
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.960033655166626,
+      "learning_rate": 2.5454810600009833e-06,
+      "loss": 0.0218,
+      "step": 1478
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.4149283170700073,
+      "learning_rate": 2.5418295118027653e-06,
+      "loss": 0.0165,
+      "step": 1479
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.962011456489563,
+      "learning_rate": 2.538178658478265e-06,
+      "loss": 0.0271,
+      "step": 1480
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 2.129974842071533,
+      "learning_rate": 2.5345285055644636e-06,
+      "loss": 0.0222,
+      "step": 1481
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.7542123794555664,
+      "learning_rate": 2.5308790585972813e-06,
+      "loss": 0.0223,
+      "step": 1482
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.3654638528823853,
+      "learning_rate": 2.52723032311157e-06,
+      "loss": 0.0184,
+      "step": 1483
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.7264395952224731,
+      "learning_rate": 2.5235823046410985e-06,
+      "loss": 0.0173,
+      "step": 1484
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 2.1574065685272217,
+      "learning_rate": 2.519935008718551e-06,
+      "loss": 0.015,
+      "step": 1485
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.2185943126678467,
+      "learning_rate": 2.5162884408755123e-06,
+      "loss": 0.0218,
+      "step": 1486
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.8119635581970215,
+      "learning_rate": 2.5126426066424686e-06,
+      "loss": 0.0504,
+      "step": 1487
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.375873565673828,
+      "learning_rate": 2.5089975115487893e-06,
+      "loss": 0.0334,
+      "step": 1488
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.7450482845306396,
+      "learning_rate": 2.5053531611227236e-06,
+      "loss": 0.0356,
+      "step": 1489
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.225628614425659,
+      "learning_rate": 2.501709560891392e-06,
+      "loss": 0.0247,
+      "step": 1490
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.6890528202056885,
+      "learning_rate": 2.498066716380777e-06,
+      "loss": 0.019,
+      "step": 1491
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.3264089822769165,
+      "learning_rate": 2.494424633115713e-06,
+      "loss": 0.0135,
+      "step": 1492
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 2.230882167816162,
+      "learning_rate": 2.490783316619885e-06,
+      "loss": 0.0312,
+      "step": 1493
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.955730676651001,
+      "learning_rate": 2.4871427724158098e-06,
+      "loss": 0.0254,
+      "step": 1494
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.5270215272903442,
+      "learning_rate": 2.483503006024834e-06,
+      "loss": 0.0232,
+      "step": 1495
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.2805649042129517,
+      "learning_rate": 2.4798640229671277e-06,
+      "loss": 0.0125,
+      "step": 1496
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.5182636976242065,
+      "learning_rate": 2.4762258287616696e-06,
+      "loss": 0.0231,
+      "step": 1497
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 2.383779287338257,
+      "learning_rate": 2.472588428926243e-06,
+      "loss": 0.0325,
+      "step": 1498
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.981991171836853,
+      "learning_rate": 2.4689518289774258e-06,
+      "loss": 0.0161,
+      "step": 1499
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.6175023317337036,
+      "learning_rate": 2.4653160344305852e-06,
+      "loss": 0.0313,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "eval_blimp_filtered_avg": 0.731044776119403,
+      "eval_blimp_filtered_std": 0.004871281724173744,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.017513892765077744,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "eval_vqa_filtered_avg": 0.33,
+      "eval_vqa_filtered_std": 0.047258156262526045,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 2.061825752258301,
+      "learning_rate": 2.461681050799865e-06,
+      "loss": 0.0221,
+      "step": 1501
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.8649793863296509,
+      "learning_rate": 2.458046883598178e-06,
+      "loss": 0.0192,
+      "step": 1502
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 2.0011789798736572,
+      "learning_rate": 2.4544135383372016e-06,
+      "loss": 0.029,
+      "step": 1503
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 2.215034246444702,
+      "learning_rate": 2.450781020527365e-06,
+      "loss": 0.0223,
+      "step": 1504
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.8533267974853516,
+      "learning_rate": 2.447149335677841e-06,
+      "loss": 0.029,
+      "step": 1505
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 3.841268539428711,
+      "learning_rate": 2.4435184892965437e-06,
+      "loss": 0.0529,
+      "step": 1506
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 2.184337615966797,
+      "learning_rate": 2.43988848689011e-06,
+      "loss": 0.0272,
+      "step": 1507
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.6988147497177124,
+      "learning_rate": 2.4362593339639015e-06,
+      "loss": 0.0176,
+      "step": 1508
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.3001044988632202,
+      "learning_rate": 2.432631036021986e-06,
+      "loss": 0.0191,
+      "step": 1509
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 2.0413200855255127,
+      "learning_rate": 2.4290035985671424e-06,
+      "loss": 0.0236,
+      "step": 1510
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.4490123987197876,
+      "learning_rate": 2.425377027100838e-06,
+      "loss": 0.0173,
+      "step": 1511
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 2.248600482940674,
+      "learning_rate": 2.421751327123228e-06,
+      "loss": 0.0302,
+      "step": 1512
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.6515231132507324,
+      "learning_rate": 2.418126504133149e-06,
+      "loss": 0.0273,
+      "step": 1513
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.276062250137329,
+      "learning_rate": 2.4145025636281033e-06,
+      "loss": 0.0275,
+      "step": 1514
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.38073992729187,
+      "learning_rate": 2.4108795111042566e-06,
+      "loss": 0.0246,
+      "step": 1515
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.10103178024292,
+      "learning_rate": 2.4072573520564303e-06,
+      "loss": 0.0297,
+      "step": 1516
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.21614933013916,
+      "learning_rate": 2.403636091978087e-06,
+      "loss": 0.0283,
+      "step": 1517
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.4115705490112305,
+      "learning_rate": 2.4000157363613267e-06,
+      "loss": 0.0359,
+      "step": 1518
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.6232080459594727,
+      "learning_rate": 2.39639629069688e-06,
+      "loss": 0.0347,
+      "step": 1519
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.2947866916656494,
+      "learning_rate": 2.3927777604740953e-06,
+      "loss": 0.0243,
+      "step": 1520
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.779845952987671,
+      "learning_rate": 2.3891601511809325e-06,
+      "loss": 0.0175,
+      "step": 1521
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.7913329601287842,
+      "learning_rate": 2.3855434683039544e-06,
+      "loss": 0.0269,
+      "step": 1522
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.9485626220703125,
+      "learning_rate": 2.3819277173283223e-06,
+      "loss": 0.0211,
+      "step": 1523
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.8861695528030396,
+      "learning_rate": 2.3783129037377806e-06,
+      "loss": 0.0229,
+      "step": 1524
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 3.7487967014312744,
+      "learning_rate": 2.3746990330146516e-06,
+      "loss": 0.0245,
+      "step": 1525
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.7885574102401733,
+      "learning_rate": 2.3710861106398316e-06,
+      "loss": 0.0172,
+      "step": 1526
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 3.387913227081299,
+      "learning_rate": 2.367474142092775e-06,
+      "loss": 0.0379,
+      "step": 1527
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.7591413259506226,
+      "learning_rate": 2.363863132851489e-06,
+      "loss": 0.0245,
+      "step": 1528
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.8683927059173584,
+      "learning_rate": 2.3602530883925305e-06,
+      "loss": 0.0318,
+      "step": 1529
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 3.016078472137451,
+      "learning_rate": 2.356644014190989e-06,
+      "loss": 0.0614,
+      "step": 1530
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.5766273736953735,
+      "learning_rate": 2.3530359157204817e-06,
+      "loss": 0.017,
+      "step": 1531
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.6205812692642212,
+      "learning_rate": 2.3494287984531516e-06,
+      "loss": 0.0207,
+      "step": 1532
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.8584187030792236,
+      "learning_rate": 2.3458226678596485e-06,
+      "loss": 0.0223,
+      "step": 1533
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.161754846572876,
+      "learning_rate": 2.342217529409127e-06,
+      "loss": 0.0238,
+      "step": 1534
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.803743839263916,
+      "learning_rate": 2.338613388569237e-06,
+      "loss": 0.0338,
+      "step": 1535
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.884228229522705,
+      "learning_rate": 2.335010250806117e-06,
+      "loss": 0.0207,
+      "step": 1536
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.182255744934082,
+      "learning_rate": 2.331408121584381e-06,
+      "loss": 0.0314,
+      "step": 1537
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.9817016124725342,
+      "learning_rate": 2.327807006367116e-06,
+      "loss": 0.0241,
+      "step": 1538
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.9135236740112305,
+      "learning_rate": 2.3242069106158717e-06,
+      "loss": 0.0235,
+      "step": 1539
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.9788758754730225,
+      "learning_rate": 2.32060783979065e-06,
+      "loss": 0.0248,
+      "step": 1540
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 2.5612924098968506,
+      "learning_rate": 2.3170097993498967e-06,
+      "loss": 0.0417,
+      "step": 1541
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 2.4611902236938477,
+      "learning_rate": 2.3134127947504998e-06,
+      "loss": 0.0319,
+      "step": 1542
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.732075810432434,
+      "learning_rate": 2.309816831447772e-06,
+      "loss": 0.0218,
+      "step": 1543
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.4583700895309448,
+      "learning_rate": 2.3062219148954474e-06,
+      "loss": 0.0253,
+      "step": 1544
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 2.1532647609710693,
+      "learning_rate": 2.3026280505456752e-06,
+      "loss": 0.0345,
+      "step": 1545
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.6332167387008667,
+      "learning_rate": 2.2990352438490065e-06,
+      "loss": 0.0277,
+      "step": 1546
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.9777950048446655,
+      "learning_rate": 2.2954435002543885e-06,
+      "loss": 0.0235,
+      "step": 1547
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 2.7736923694610596,
+      "learning_rate": 2.291852825209155e-06,
+      "loss": 0.024,
+      "step": 1548
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 2.8943586349487305,
+      "learning_rate": 2.2882632241590234e-06,
+      "loss": 0.0435,
+      "step": 1549
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.6013497114181519,
+      "learning_rate": 2.284674702548077e-06,
+      "loss": 0.0272,
+      "step": 1550
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.495845913887024,
+      "learning_rate": 2.2810872658187657e-06,
+      "loss": 0.0227,
+      "step": 1551
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 2.4803192615509033,
+      "learning_rate": 2.277500919411893e-06,
+      "loss": 0.0353,
+      "step": 1552
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.4565989971160889,
+      "learning_rate": 2.273915668766608e-06,
+      "loss": 0.0145,
+      "step": 1553
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 3.3853442668914795,
+      "learning_rate": 2.270331519320397e-06,
+      "loss": 0.0248,
+      "step": 1554
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.3093631267547607,
+      "learning_rate": 2.2667484765090807e-06,
+      "loss": 0.0427,
+      "step": 1555
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.1562089920043945,
+      "learning_rate": 2.263166545766797e-06,
+      "loss": 0.0286,
+      "step": 1556
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.3147945404052734,
+      "learning_rate": 2.2595857325259993e-06,
+      "loss": 0.0171,
+      "step": 1557
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.1565611362457275,
+      "learning_rate": 2.2560060422174433e-06,
+      "loss": 0.0234,
+      "step": 1558
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.6333812475204468,
+      "learning_rate": 2.2524274802701865e-06,
+      "loss": 0.015,
+      "step": 1559
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 2.7551915645599365,
+      "learning_rate": 2.2488500521115714e-06,
+      "loss": 0.024,
+      "step": 1560
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.9077558517456055,
+      "learning_rate": 2.245273763167221e-06,
+      "loss": 0.0296,
+      "step": 1561
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.7142243385314941,
+      "learning_rate": 2.241698618861033e-06,
+      "loss": 0.0232,
+      "step": 1562
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 3.2503559589385986,
+      "learning_rate": 2.2381246246151676e-06,
+      "loss": 0.0345,
+      "step": 1563
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 2.4535558223724365,
+      "learning_rate": 2.23455178585004e-06,
+      "loss": 0.0256,
+      "step": 1564
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 2.760897397994995,
+      "learning_rate": 2.230980107984315e-06,
+      "loss": 0.0372,
+      "step": 1565
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 2.0208683013916016,
+      "learning_rate": 2.227409596434895e-06,
+      "loss": 0.0173,
+      "step": 1566
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.6440926790237427,
+      "learning_rate": 2.2238402566169116e-06,
+      "loss": 0.0283,
+      "step": 1567
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.9552116394042969,
+      "learning_rate": 2.220272093943726e-06,
+      "loss": 0.0329,
+      "step": 1568
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 2.271404504776001,
+      "learning_rate": 2.216705113826908e-06,
+      "loss": 0.0242,
+      "step": 1569
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.7705224752426147,
+      "learning_rate": 2.213139321676236e-06,
+      "loss": 0.0237,
+      "step": 1570
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 2.0369856357574463,
+      "learning_rate": 2.2095747228996858e-06,
+      "loss": 0.0196,
+      "step": 1571
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 2.470810651779175,
+      "learning_rate": 2.206011322903425e-06,
+      "loss": 0.0184,
+      "step": 1572
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 2.904381275177002,
+      "learning_rate": 2.2024491270918017e-06,
+      "loss": 0.0374,
+      "step": 1573
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.255245327949524,
+      "learning_rate": 2.198888140867338e-06,
+      "loss": 0.0186,
+      "step": 1574
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.848855972290039,
+      "learning_rate": 2.195328369630722e-06,
+      "loss": 0.0222,
+      "step": 1575
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 2.6878788471221924,
+      "learning_rate": 2.1917698187807987e-06,
+      "loss": 0.0277,
+      "step": 1576
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.6416897773742676,
+      "learning_rate": 2.1882124937145608e-06,
+      "loss": 0.0248,
+      "step": 1577
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.5208277702331543,
+      "learning_rate": 2.184656399827146e-06,
+      "loss": 0.0192,
+      "step": 1578
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.9335892200469971,
+      "learning_rate": 2.181101542511821e-06,
+      "loss": 0.0079,
+      "step": 1579
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.246731162071228,
+      "learning_rate": 2.1775479271599772e-06,
+      "loss": 0.0118,
+      "step": 1580
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 1.2990490198135376,
+      "learning_rate": 2.173995559161126e-06,
+      "loss": 0.0135,
+      "step": 1581
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 2.0331027507781982,
+      "learning_rate": 2.170444443902882e-06,
+      "loss": 0.0155,
+      "step": 1582
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 2.806368589401245,
+      "learning_rate": 2.1668945867709635e-06,
+      "loss": 0.024,
+      "step": 1583
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 1.303980827331543,
+      "learning_rate": 2.163345993149178e-06,
+      "loss": 0.0122,
+      "step": 1584
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 2.132803201675415,
+      "learning_rate": 2.1597986684194214e-06,
+      "loss": 0.019,
+      "step": 1585
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.1847878694534302,
+      "learning_rate": 2.15625261796166e-06,
+      "loss": 0.0103,
+      "step": 1586
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.6737439632415771,
+      "learning_rate": 2.152707847153929e-06,
+      "loss": 0.0155,
+      "step": 1587
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.6405411958694458,
+      "learning_rate": 2.1491643613723247e-06,
+      "loss": 0.0202,
+      "step": 1588
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.5736236572265625,
+      "learning_rate": 2.145622165990993e-06,
+      "loss": 0.0217,
+      "step": 1589
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.4893486499786377,
+      "learning_rate": 2.1420812663821218e-06,
+      "loss": 0.0129,
+      "step": 1590
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 2.827496290206909,
+      "learning_rate": 2.138541667915936e-06,
+      "loss": 0.0327,
+      "step": 1591
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.1139668226242065,
+      "learning_rate": 2.1350033759606865e-06,
+      "loss": 0.009,
+      "step": 1592
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.8998172283172607,
+      "learning_rate": 2.131466395882641e-06,
+      "loss": 0.0162,
+      "step": 1593
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.477875828742981,
+      "learning_rate": 2.1279307330460794e-06,
+      "loss": 0.0175,
+      "step": 1594
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.3113806247711182,
+      "learning_rate": 2.124396392813284e-06,
+      "loss": 0.0093,
+      "step": 1595
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.2860724925994873,
+      "learning_rate": 2.1208633805445296e-06,
+      "loss": 0.0169,
+      "step": 1596
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 2.421346664428711,
+      "learning_rate": 2.117331701598078e-06,
+      "loss": 0.0262,
+      "step": 1597
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 2.1626765727996826,
+      "learning_rate": 2.113801361330169e-06,
+      "loss": 0.0102,
+      "step": 1598
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 1.333755612373352,
+      "learning_rate": 2.1102723650950116e-06,
+      "loss": 0.0226,
+      "step": 1599
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 1.2760261297225952,
+      "learning_rate": 2.1067447182447755e-06,
+      "loss": 0.0144,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "eval_blimp_filtered_avg": 0.7313432835820896,
+      "eval_blimp_filtered_std": 0.004875714192537722,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.017513892765077744,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 1.7273930311203003,
+      "learning_rate": 2.103218426129588e-06,
+      "loss": 0.0171,
+      "step": 1601
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 1.130737543106079,
+      "learning_rate": 2.099693494097517e-06,
+      "loss": 0.0089,
+      "step": 1602
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 2.469149112701416,
+      "learning_rate": 2.0961699274945688e-06,
+      "loss": 0.0214,
+      "step": 1603
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 1.3174607753753662,
+      "learning_rate": 2.092647731664681e-06,
+      "loss": 0.0126,
+      "step": 1604
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.9995936751365662,
+      "learning_rate": 2.08912691194971e-06,
+      "loss": 0.0066,
+      "step": 1605
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 1.2984211444854736,
+      "learning_rate": 2.085607473689426e-06,
+      "loss": 0.0139,
+      "step": 1606
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 2.939004898071289,
+      "learning_rate": 2.0820894222215024e-06,
+      "loss": 0.015,
+      "step": 1607
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 1.8445813655853271,
+      "learning_rate": 2.078572762881514e-06,
+      "loss": 0.0217,
+      "step": 1608
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 1.748602271080017,
+      "learning_rate": 2.07505750100292e-06,
+      "loss": 0.0126,
+      "step": 1609
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 1.2420772314071655,
+      "learning_rate": 2.0715436419170595e-06,
+      "loss": 0.0144,
+      "step": 1610
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 1.2730077505111694,
+      "learning_rate": 2.0680311909531486e-06,
+      "loss": 0.0133,
+      "step": 1611
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 1.1965616941452026,
+      "learning_rate": 2.064520153438264e-06,
+      "loss": 0.0162,
+      "step": 1612
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 1.8696595430374146,
+      "learning_rate": 2.0610105346973386e-06,
+      "loss": 0.0147,
+      "step": 1613
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.9974368214607239,
+      "learning_rate": 2.057502340053157e-06,
+      "loss": 0.0137,
+      "step": 1614
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 1.4981776475906372,
+      "learning_rate": 2.0539955748263416e-06,
+      "loss": 0.0182,
+      "step": 1615
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 1.3763387203216553,
+      "learning_rate": 2.0504902443353453e-06,
+      "loss": 0.0142,
+      "step": 1616
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 1.3074193000793457,
+      "learning_rate": 2.0469863538964484e-06,
+      "loss": 0.0113,
+      "step": 1617
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 1.687665343284607,
+      "learning_rate": 2.0434839088237455e-06,
+      "loss": 0.0113,
+      "step": 1618
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 1.9023373126983643,
+      "learning_rate": 2.0399829144291382e-06,
+      "loss": 0.0184,
+      "step": 1619
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 1.5201495885849,
+      "learning_rate": 2.0364833760223296e-06,
+      "loss": 0.0098,
+      "step": 1620
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 1.2860575914382935,
+      "learning_rate": 2.032985298910815e-06,
+      "loss": 0.011,
+      "step": 1621
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 1.2321438789367676,
+      "learning_rate": 2.029488688399871e-06,
+      "loss": 0.0122,
+      "step": 1622
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 1.079641580581665,
+      "learning_rate": 2.0259935497925513e-06,
+      "loss": 0.0149,
+      "step": 1623
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 2.016981363296509,
+      "learning_rate": 2.022499888389679e-06,
+      "loss": 0.0143,
+      "step": 1624
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.9185414910316467,
+      "learning_rate": 2.019007709489834e-06,
+      "loss": 0.0108,
+      "step": 1625
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 1.3953144550323486,
+      "learning_rate": 2.0155170183893486e-06,
+      "loss": 0.011,
+      "step": 1626
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 1.4591954946517944,
+      "learning_rate": 2.0120278203823004e-06,
+      "loss": 0.0114,
+      "step": 1627
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 2.453949451446533,
+      "learning_rate": 2.0085401207605007e-06,
+      "loss": 0.0186,
+      "step": 1628
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 2.103576898574829,
+      "learning_rate": 2.0050539248134864e-06,
+      "loss": 0.0172,
+      "step": 1629
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 1.2003625631332397,
+      "learning_rate": 2.0015692378285208e-06,
+      "loss": 0.0101,
+      "step": 1630
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 1.3601410388946533,
+      "learning_rate": 1.998086065090571e-06,
+      "loss": 0.0171,
+      "step": 1631
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 2.139050006866455,
+      "learning_rate": 1.9946044118823114e-06,
+      "loss": 0.0152,
+      "step": 1632
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 2.038116216659546,
+      "learning_rate": 1.9911242834841106e-06,
+      "loss": 0.0162,
+      "step": 1633
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 1.3036565780639648,
+      "learning_rate": 1.9876456851740266e-06,
+      "loss": 0.0066,
+      "step": 1634
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 2.530449867248535,
+      "learning_rate": 1.9841686222277946e-06,
+      "loss": 0.0164,
+      "step": 1635
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 1.530340552330017,
+      "learning_rate": 1.9806930999188207e-06,
+      "loss": 0.0262,
+      "step": 1636
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 1.6057711839675903,
+      "learning_rate": 1.977219123518178e-06,
+      "loss": 0.0126,
+      "step": 1637
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 1.5105196237564087,
+      "learning_rate": 1.9737466982945917e-06,
+      "loss": 0.0116,
+      "step": 1638
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 1.6331992149353027,
+      "learning_rate": 1.970275829514435e-06,
+      "loss": 0.0149,
+      "step": 1639
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 1.4868899583816528,
+      "learning_rate": 1.966806522441723e-06,
+      "loss": 0.0067,
+      "step": 1640
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 1.3155184984207153,
+      "learning_rate": 1.9633387823380985e-06,
+      "loss": 0.0156,
+      "step": 1641
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 1.9279378652572632,
+      "learning_rate": 1.9598726144628294e-06,
+      "loss": 0.022,
+      "step": 1642
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 1.255454182624817,
+      "learning_rate": 1.9564080240728015e-06,
+      "loss": 0.0113,
+      "step": 1643
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.1962611675262451,
+      "learning_rate": 1.9529450164225045e-06,
+      "loss": 0.0122,
+      "step": 1644
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.025037407875061,
+      "learning_rate": 1.949483596764029e-06,
+      "loss": 0.0073,
+      "step": 1645
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.8792548179626465,
+      "learning_rate": 1.946023770347057e-06,
+      "loss": 0.0156,
+      "step": 1646
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.7499372959136963,
+      "learning_rate": 1.942565542418856e-06,
+      "loss": 0.0139,
+      "step": 1647
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.7629427909851074,
+      "learning_rate": 1.939108918224268e-06,
+      "loss": 0.0193,
+      "step": 1648
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.9979041814804077,
+      "learning_rate": 1.9356539030057e-06,
+      "loss": 0.0246,
+      "step": 1649
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 1.307125449180603,
+      "learning_rate": 1.9322005020031237e-06,
+      "loss": 0.0128,
+      "step": 1650
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 2.075721025466919,
+      "learning_rate": 1.9287487204540588e-06,
+      "loss": 0.0182,
+      "step": 1651
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 1.639565348625183,
+      "learning_rate": 1.92529856359357e-06,
+      "loss": 0.0166,
+      "step": 1652
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 1.2296085357666016,
+      "learning_rate": 1.9218500366542595e-06,
+      "loss": 0.0094,
+      "step": 1653
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 1.290384292602539,
+      "learning_rate": 1.9184031448662565e-06,
+      "loss": 0.0131,
+      "step": 1654
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 1.6350033283233643,
+      "learning_rate": 1.9149578934572084e-06,
+      "loss": 0.0122,
+      "step": 1655
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 1.219002604484558,
+      "learning_rate": 1.911514287652277e-06,
+      "loss": 0.014,
+      "step": 1656
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 1.1566717624664307,
+      "learning_rate": 1.9080723326741285e-06,
+      "loss": 0.0093,
+      "step": 1657
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 2.934467077255249,
+      "learning_rate": 1.9046320337429241e-06,
+      "loss": 0.0214,
+      "step": 1658
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 1.0122382640838623,
+      "learning_rate": 1.9011933960763127e-06,
+      "loss": 0.0087,
+      "step": 1659
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 1.4913413524627686,
+      "learning_rate": 1.8977564248894271e-06,
+      "loss": 0.0158,
+      "step": 1660
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 1.5406959056854248,
+      "learning_rate": 1.8943211253948693e-06,
+      "loss": 0.0104,
+      "step": 1661
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.9944964647293091,
+      "learning_rate": 1.8908875028027066e-06,
+      "loss": 0.0062,
+      "step": 1662
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 1.6341869831085205,
+      "learning_rate": 1.8874555623204653e-06,
+      "loss": 0.0135,
+      "step": 1663
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 1.3632527589797974,
+      "learning_rate": 1.8840253091531176e-06,
+      "loss": 0.0115,
+      "step": 1664
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.7795816659927368,
+      "learning_rate": 1.8805967485030773e-06,
+      "loss": 0.009,
+      "step": 1665
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 2.353595018386841,
+      "learning_rate": 1.8771698855701945e-06,
+      "loss": 0.0101,
+      "step": 1666
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.4596761465072632,
+      "learning_rate": 1.87374472555174e-06,
+      "loss": 0.017,
+      "step": 1667
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.4096475839614868,
+      "learning_rate": 1.8703212736424039e-06,
+      "loss": 0.0146,
+      "step": 1668
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.8360106945037842,
+      "learning_rate": 1.8668995350342847e-06,
+      "loss": 0.0146,
+      "step": 1669
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.3896836042404175,
+      "learning_rate": 1.863479514916886e-06,
+      "loss": 0.0229,
+      "step": 1670
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 1.5895816087722778,
+      "learning_rate": 1.860061218477101e-06,
+      "loss": 0.0082,
+      "step": 1671
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 2.0713579654693604,
+      "learning_rate": 1.85664465089921e-06,
+      "loss": 0.0121,
+      "step": 1672
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 1.563232660293579,
+      "learning_rate": 1.8532298173648723e-06,
+      "loss": 0.0153,
+      "step": 1673
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 1.5700125694274902,
+      "learning_rate": 1.8498167230531164e-06,
+      "loss": 0.0133,
+      "step": 1674
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 1.737511396408081,
+      "learning_rate": 1.8464053731403319e-06,
+      "loss": 0.0212,
+      "step": 1675
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 1.540568232536316,
+      "learning_rate": 1.842995772800266e-06,
+      "loss": 0.0081,
+      "step": 1676
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 1.693328857421875,
+      "learning_rate": 1.8395879272040097e-06,
+      "loss": 0.0156,
+      "step": 1677
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 2.1555137634277344,
+      "learning_rate": 1.836181841519993e-06,
+      "loss": 0.0207,
+      "step": 1678
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 1.5485413074493408,
+      "learning_rate": 1.832777520913978e-06,
+      "loss": 0.0107,
+      "step": 1679
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 1.6018801927566528,
+      "learning_rate": 1.8293749705490491e-06,
+      "loss": 0.0172,
+      "step": 1680
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 2.0386672019958496,
+      "learning_rate": 1.8259741955856062e-06,
+      "loss": 0.0159,
+      "step": 1681
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 2.905478000640869,
+      "learning_rate": 1.8225752011813538e-06,
+      "loss": 0.0178,
+      "step": 1682
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 1.2259677648544312,
+      "learning_rate": 1.819177992491302e-06,
+      "loss": 0.014,
+      "step": 1683
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 1.437726378440857,
+      "learning_rate": 1.8157825746677482e-06,
+      "loss": 0.0106,
+      "step": 1684
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 1.028694748878479,
+      "learning_rate": 1.8123889528602727e-06,
+      "loss": 0.0095,
+      "step": 1685
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 1.7536890506744385,
+      "learning_rate": 1.8089971322157358e-06,
+      "loss": 0.0124,
+      "step": 1686
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 1.4571354389190674,
+      "learning_rate": 1.805607117878264e-06,
+      "loss": 0.0172,
+      "step": 1687
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 1.4393140077590942,
+      "learning_rate": 1.802218914989242e-06,
+      "loss": 0.0094,
+      "step": 1688
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 1.7913202047348022,
+      "learning_rate": 1.7988325286873137e-06,
+      "loss": 0.015,
+      "step": 1689
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 1.9483704566955566,
+      "learning_rate": 1.7954479641083613e-06,
+      "loss": 0.0236,
+      "step": 1690
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 2.1679446697235107,
+      "learning_rate": 1.792065226385506e-06,
+      "loss": 0.0128,
+      "step": 1691
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 1.4608047008514404,
+      "learning_rate": 1.7886843206491019e-06,
+      "loss": 0.02,
+      "step": 1692
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 1.2416274547576904,
+      "learning_rate": 1.7853052520267195e-06,
+      "loss": 0.0148,
+      "step": 1693
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 1.158479928970337,
+      "learning_rate": 1.7819280256431466e-06,
+      "loss": 0.0126,
+      "step": 1694
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 2.3266148567199707,
+      "learning_rate": 1.7785526466203743e-06,
+      "loss": 0.0172,
+      "step": 1695
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 1.6359390020370483,
+      "learning_rate": 1.775179120077595e-06,
+      "loss": 0.0162,
+      "step": 1696
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 1.1562213897705078,
+      "learning_rate": 1.7718074511311895e-06,
+      "loss": 0.0087,
+      "step": 1697
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 1.7280696630477905,
+      "learning_rate": 1.7684376448947205e-06,
+      "loss": 0.0133,
+      "step": 1698
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 4.5138020515441895,
+      "learning_rate": 1.7650697064789294e-06,
+      "loss": 0.0204,
+      "step": 1699
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 1.274200677871704,
+      "learning_rate": 1.7617036409917213e-06,
+      "loss": 0.0114,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "eval_blimp_filtered_avg": 0.7292537313432835,
+      "eval_blimp_filtered_std": 0.0048860870343199626,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.017476319339625877,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 2.028714895248413,
+      "learning_rate": 1.7583394535381614e-06,
+      "loss": 0.0117,
+      "step": 1701
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.8119041919708252,
+      "learning_rate": 1.7549771492204684e-06,
+      "loss": 0.0164,
+      "step": 1702
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.687760353088379,
+      "learning_rate": 1.7516167331380025e-06,
+      "loss": 0.0207,
+      "step": 1703
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.7550314664840698,
+      "learning_rate": 1.7482582103872608e-06,
+      "loss": 0.0162,
+      "step": 1704
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.6663331985473633,
+      "learning_rate": 1.7449015860618716e-06,
+      "loss": 0.0143,
+      "step": 1705
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.4410921335220337,
+      "learning_rate": 1.7415468652525802e-06,
+      "loss": 0.0109,
+      "step": 1706
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 1.4207170009613037,
+      "learning_rate": 1.7381940530472476e-06,
+      "loss": 0.0131,
+      "step": 1707
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 1.9346827268600464,
+      "learning_rate": 1.734843154530837e-06,
+      "loss": 0.0146,
+      "step": 1708
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 1.5037176609039307,
+      "learning_rate": 1.7314941747854146e-06,
+      "loss": 0.0119,
+      "step": 1709
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 2.7576916217803955,
+      "learning_rate": 1.7281471188901328e-06,
+      "loss": 0.0196,
+      "step": 1710
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 1.1254243850708008,
+      "learning_rate": 1.7248019919212238e-06,
+      "loss": 0.0107,
+      "step": 1711
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 2.4174911975860596,
+      "learning_rate": 1.721458798952001e-06,
+      "loss": 0.0231,
+      "step": 1712
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 1.4880990982055664,
+      "learning_rate": 1.7181175450528389e-06,
+      "loss": 0.0146,
+      "step": 1713
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.918998122215271,
+      "learning_rate": 1.714778235291173e-06,
+      "loss": 0.0079,
+      "step": 1714
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.9897366762161255,
+      "learning_rate": 1.7114408747314926e-06,
+      "loss": 0.0061,
+      "step": 1715
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 1.1539252996444702,
+      "learning_rate": 1.7081054684353274e-06,
+      "loss": 0.0119,
+      "step": 1716
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 1.283732533454895,
+      "learning_rate": 1.7047720214612446e-06,
+      "loss": 0.0122,
+      "step": 1717
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.444400429725647,
+      "learning_rate": 1.7014405388648389e-06,
+      "loss": 0.0163,
+      "step": 1718
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.116744875907898,
+      "learning_rate": 1.6981110256987282e-06,
+      "loss": 0.0091,
+      "step": 1719
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.9108668565750122,
+      "learning_rate": 1.6947834870125418e-06,
+      "loss": 0.0118,
+      "step": 1720
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.8220466375350952,
+      "learning_rate": 1.691457927852914e-06,
+      "loss": 0.0144,
+      "step": 1721
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.643065333366394,
+      "learning_rate": 1.6881343532634772e-06,
+      "loss": 0.0156,
+      "step": 1722
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 1.163825273513794,
+      "learning_rate": 1.6848127682848552e-06,
+      "loss": 0.0076,
+      "step": 1723
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 1.8429875373840332,
+      "learning_rate": 1.6814931779546515e-06,
+      "loss": 0.013,
+      "step": 1724
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 2.4421889781951904,
+      "learning_rate": 1.6781755873074483e-06,
+      "loss": 0.016,
+      "step": 1725
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 1.3453940153121948,
+      "learning_rate": 1.6748600013747927e-06,
+      "loss": 0.014,
+      "step": 1726
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 1.2802871465682983,
+      "learning_rate": 1.67154642518519e-06,
+      "loss": 0.0127,
+      "step": 1727
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 1.5133326053619385,
+      "learning_rate": 1.6682348637641018e-06,
+      "loss": 0.0149,
+      "step": 1728
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 1.315102219581604,
+      "learning_rate": 1.6649253221339302e-06,
+      "loss": 0.0178,
+      "step": 1729
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 1.8380656242370605,
+      "learning_rate": 1.6616178053140154e-06,
+      "loss": 0.0138,
+      "step": 1730
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 2.404301881790161,
+      "learning_rate": 1.6583123183206247e-06,
+      "loss": 0.0164,
+      "step": 1731
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 1.6433770656585693,
+      "learning_rate": 1.6550088661669523e-06,
+      "loss": 0.0122,
+      "step": 1732
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 1.8122162818908691,
+      "learning_rate": 1.651707453863102e-06,
+      "loss": 0.014,
+      "step": 1733
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 1.4014800786972046,
+      "learning_rate": 1.6484080864160817e-06,
+      "loss": 0.0161,
+      "step": 1734
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 1.1212174892425537,
+      "learning_rate": 1.6451107688298045e-06,
+      "loss": 0.0117,
+      "step": 1735
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 1.6570181846618652,
+      "learning_rate": 1.6418155061050694e-06,
+      "loss": 0.0174,
+      "step": 1736
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 1.0117789506912231,
+      "learning_rate": 1.63852230323956e-06,
+      "loss": 0.0082,
+      "step": 1737
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.9598602056503296,
+      "learning_rate": 1.6352311652278388e-06,
+      "loss": 0.014,
+      "step": 1738
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 1.7547260522842407,
+      "learning_rate": 1.6319420970613338e-06,
+      "loss": 0.0101,
+      "step": 1739
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 1.2679924964904785,
+      "learning_rate": 1.6286551037283321e-06,
+      "loss": 0.0143,
+      "step": 1740
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 1.7287044525146484,
+      "learning_rate": 1.62537019021398e-06,
+      "loss": 0.0202,
+      "step": 1741
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.8793157935142517,
+      "learning_rate": 1.6220873615002645e-06,
+      "loss": 0.0082,
+      "step": 1742
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 1.3953230381011963,
+      "learning_rate": 1.6188066225660113e-06,
+      "loss": 0.0127,
+      "step": 1743
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.430336356163025,
+      "learning_rate": 1.6155279783868784e-06,
+      "loss": 0.0118,
+      "step": 1744
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.1960943937301636,
+      "learning_rate": 1.6122514339353458e-06,
+      "loss": 0.0066,
+      "step": 1745
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.501110553741455,
+      "learning_rate": 1.6089769941807085e-06,
+      "loss": 0.0085,
+      "step": 1746
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.378609299659729,
+      "learning_rate": 1.60570466408907e-06,
+      "loss": 0.0122,
+      "step": 1747
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.63499915599823,
+      "learning_rate": 1.602434448623336e-06,
+      "loss": 0.0147,
+      "step": 1748
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 1.3505918979644775,
+      "learning_rate": 1.599166352743203e-06,
+      "loss": 0.0093,
+      "step": 1749
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 1.3896654844284058,
+      "learning_rate": 1.595900381405151e-06,
+      "loss": 0.013,
+      "step": 1750
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 2.45300030708313,
+      "learning_rate": 1.5926365395624435e-06,
+      "loss": 0.0165,
+      "step": 1751
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 1.0396778583526611,
+      "learning_rate": 1.5893748321651098e-06,
+      "loss": 0.0071,
+      "step": 1752
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 1.6008306741714478,
+      "learning_rate": 1.5861152641599413e-06,
+      "loss": 0.0177,
+      "step": 1753
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 1.111643671989441,
+      "learning_rate": 1.5828578404904902e-06,
+      "loss": 0.0153,
+      "step": 1754
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.965140700340271,
+      "learning_rate": 1.5796025660970514e-06,
+      "loss": 0.0122,
+      "step": 1755
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 2.704420804977417,
+      "learning_rate": 1.5763494459166628e-06,
+      "loss": 0.0139,
+      "step": 1756
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 1.4204522371292114,
+      "learning_rate": 1.5730984848830907e-06,
+      "loss": 0.0081,
+      "step": 1757
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 2.4666922092437744,
+      "learning_rate": 1.569849687926834e-06,
+      "loss": 0.0199,
+      "step": 1758
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.8272417783737183,
+      "learning_rate": 1.5666030599751047e-06,
+      "loss": 0.007,
+      "step": 1759
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 1.8478165864944458,
+      "learning_rate": 1.5633586059518245e-06,
+      "loss": 0.014,
+      "step": 1760
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.8978012204170227,
+      "learning_rate": 1.5601163307776229e-06,
+      "loss": 0.008,
+      "step": 1761
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 1.2463209629058838,
+      "learning_rate": 1.5568762393698207e-06,
+      "loss": 0.0117,
+      "step": 1762
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 1.8828572034835815,
+      "learning_rate": 1.553638336642426e-06,
+      "loss": 0.0139,
+      "step": 1763
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 1.12079656124115,
+      "learning_rate": 1.5504026275061326e-06,
+      "loss": 0.0047,
+      "step": 1764
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 2.593034029006958,
+      "learning_rate": 1.5471691168683026e-06,
+      "loss": 0.0243,
+      "step": 1765
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.5877578258514404,
+      "learning_rate": 1.543937809632965e-06,
+      "loss": 0.0114,
+      "step": 1766
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.4614720344543457,
+      "learning_rate": 1.5407087107008086e-06,
+      "loss": 0.017,
+      "step": 1767
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.270501971244812,
+      "learning_rate": 1.5374818249691709e-06,
+      "loss": 0.019,
+      "step": 1768
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.6790293455123901,
+      "learning_rate": 1.5342571573320344e-06,
+      "loss": 0.0146,
+      "step": 1769
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.4556646347045898,
+      "learning_rate": 1.531034712680015e-06,
+      "loss": 0.0114,
+      "step": 1770
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.71495521068573,
+      "learning_rate": 1.5278144959003623e-06,
+      "loss": 0.0112,
+      "step": 1771
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.9057985544204712,
+      "learning_rate": 1.5245965118769429e-06,
+      "loss": 0.0102,
+      "step": 1772
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 2.0564494132995605,
+      "learning_rate": 1.5213807654902368e-06,
+      "loss": 0.0179,
+      "step": 1773
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.2085963487625122,
+      "learning_rate": 1.5181672616173331e-06,
+      "loss": 0.0113,
+      "step": 1774
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.4164783954620361,
+      "learning_rate": 1.5149560051319185e-06,
+      "loss": 0.0098,
+      "step": 1775
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 1.849463939666748,
+      "learning_rate": 1.5117470009042694e-06,
+      "loss": 0.0097,
+      "step": 1776
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.911065399646759,
+      "learning_rate": 1.50854025380125e-06,
+      "loss": 0.0082,
+      "step": 1777
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 1.3469946384429932,
+      "learning_rate": 1.5053357686862997e-06,
+      "loss": 0.0084,
+      "step": 1778
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 1.9119868278503418,
+      "learning_rate": 1.5021335504194265e-06,
+      "loss": 0.0257,
+      "step": 1779
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 1.4449726343154907,
+      "learning_rate": 1.4989336038571986e-06,
+      "loss": 0.012,
+      "step": 1780
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 1.3791338205337524,
+      "learning_rate": 1.4957359338527446e-06,
+      "loss": 0.0146,
+      "step": 1781
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 1.1868882179260254,
+      "learning_rate": 1.4925405452557357e-06,
+      "loss": 0.0126,
+      "step": 1782
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 1.4297133684158325,
+      "learning_rate": 1.4893474429123833e-06,
+      "loss": 0.0141,
+      "step": 1783
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 1.3513811826705933,
+      "learning_rate": 1.4861566316654348e-06,
+      "loss": 0.0115,
+      "step": 1784
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 2.0936295986175537,
+      "learning_rate": 1.482968116354159e-06,
+      "loss": 0.0175,
+      "step": 1785
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 1.4919151067733765,
+      "learning_rate": 1.4797819018143437e-06,
+      "loss": 0.0138,
+      "step": 1786
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 2.62192964553833,
+      "learning_rate": 1.47659799287829e-06,
+      "loss": 0.0205,
+      "step": 1787
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 1.3241380453109741,
+      "learning_rate": 1.4734163943747986e-06,
+      "loss": 0.0113,
+      "step": 1788
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 1.4754565954208374,
+      "learning_rate": 1.470237111129168e-06,
+      "loss": 0.0165,
+      "step": 1789
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 1.0342705249786377,
+      "learning_rate": 1.4670601479631845e-06,
+      "loss": 0.0061,
+      "step": 1790
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 1.332696557044983,
+      "learning_rate": 1.463885509695116e-06,
+      "loss": 0.015,
+      "step": 1791
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 2.0401675701141357,
+      "learning_rate": 1.4607132011397051e-06,
+      "loss": 0.0159,
+      "step": 1792
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 1.4252026081085205,
+      "learning_rate": 1.4575432271081587e-06,
+      "loss": 0.0094,
+      "step": 1793
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 1.7764272689819336,
+      "learning_rate": 1.4543755924081477e-06,
+      "loss": 0.013,
+      "step": 1794
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 1.418892741203308,
+      "learning_rate": 1.4512103018437907e-06,
+      "loss": 0.0105,
+      "step": 1795
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 2.442657470703125,
+      "learning_rate": 1.4480473602156523e-06,
+      "loss": 0.0233,
+      "step": 1796
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 2.1668405532836914,
+      "learning_rate": 1.4448867723207364e-06,
+      "loss": 0.0131,
+      "step": 1797
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 2.119786024093628,
+      "learning_rate": 1.4417285429524753e-06,
+      "loss": 0.0252,
+      "step": 1798
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 1.6841340065002441,
+      "learning_rate": 1.4385726769007233e-06,
+      "loss": 0.0131,
+      "step": 1799
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 1.1075589656829834,
+      "learning_rate": 1.435419178951755e-06,
+      "loss": 0.0094,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "eval_blimp_filtered_avg": 0.731044776119403,
+      "eval_blimp_filtered_std": 0.004873359945683911,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.01737392472820325,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "eval_vqa_filtered_avg": 0.29,
+      "eval_vqa_filtered_std": 0.045604802157206845,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 1.5981600284576416,
+      "learning_rate": 1.4322680538882482e-06,
+      "loss": 0.0184,
+      "step": 1801
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 1.5429378747940063,
+      "learning_rate": 1.4291193064892847e-06,
+      "loss": 0.0117,
+      "step": 1802
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 1.528489112854004,
+      "learning_rate": 1.42597294153034e-06,
+      "loss": 0.0069,
+      "step": 1803
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 1.4107662439346313,
+      "learning_rate": 1.4228289637832756e-06,
+      "loss": 0.0108,
+      "step": 1804
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 3.0161921977996826,
+      "learning_rate": 1.419687378016333e-06,
+      "loss": 0.0164,
+      "step": 1805
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 1.8486896753311157,
+      "learning_rate": 1.4165481889941249e-06,
+      "loss": 0.0153,
+      "step": 1806
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.2184429168701172,
+      "learning_rate": 1.4134114014776322e-06,
+      "loss": 0.0102,
+      "step": 1807
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 2.674408197402954,
+      "learning_rate": 1.4102770202241903e-06,
+      "loss": 0.0114,
+      "step": 1808
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.8841915130615234,
+      "learning_rate": 1.407145049987485e-06,
+      "loss": 0.0141,
+      "step": 1809
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.6998053789138794,
+      "learning_rate": 1.4040154955175494e-06,
+      "loss": 0.017,
+      "step": 1810
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.9993551969528198,
+      "learning_rate": 1.400888361560749e-06,
+      "loss": 0.0128,
+      "step": 1811
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 2.2432641983032227,
+      "learning_rate": 1.3977636528597794e-06,
+      "loss": 0.0177,
+      "step": 1812
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 1.8021825551986694,
+      "learning_rate": 1.394641374153658e-06,
+      "loss": 0.0125,
+      "step": 1813
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 2.1882472038269043,
+      "learning_rate": 1.391521530177717e-06,
+      "loss": 0.0205,
+      "step": 1814
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 1.8746211528778076,
+      "learning_rate": 1.388404125663596e-06,
+      "loss": 0.0244,
+      "step": 1815
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 1.7752881050109863,
+      "learning_rate": 1.3852891653392334e-06,
+      "loss": 0.0144,
+      "step": 1816
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 1.3766459226608276,
+      "learning_rate": 1.3821766539288644e-06,
+      "loss": 0.0104,
+      "step": 1817
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 1.784822702407837,
+      "learning_rate": 1.3790665961530061e-06,
+      "loss": 0.0175,
+      "step": 1818
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 1.2186046838760376,
+      "learning_rate": 1.375958996728456e-06,
+      "loss": 0.0133,
+      "step": 1819
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 2.186237096786499,
+      "learning_rate": 1.3728538603682844e-06,
+      "loss": 0.0103,
+      "step": 1820
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 1.3949315547943115,
+      "learning_rate": 1.3697511917818242e-06,
+      "loss": 0.0146,
+      "step": 1821
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 1.2624883651733398,
+      "learning_rate": 1.3666509956746651e-06,
+      "loss": 0.0166,
+      "step": 1822
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 1.6862766742706299,
+      "learning_rate": 1.3635532767486502e-06,
+      "loss": 0.0132,
+      "step": 1823
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 2.1594653129577637,
+      "learning_rate": 1.3604580397018624e-06,
+      "loss": 0.0221,
+      "step": 1824
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 2.378411293029785,
+      "learning_rate": 1.3573652892286215e-06,
+      "loss": 0.0215,
+      "step": 1825
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.9825218915939331,
+      "learning_rate": 1.3542750300194773e-06,
+      "loss": 0.0068,
+      "step": 1826
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 1.5997729301452637,
+      "learning_rate": 1.3511872667611987e-06,
+      "loss": 0.0123,
+      "step": 1827
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 1.6004849672317505,
+      "learning_rate": 1.3481020041367716e-06,
+      "loss": 0.0153,
+      "step": 1828
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 1.8620418310165405,
+      "learning_rate": 1.3450192468253876e-06,
+      "loss": 0.0127,
+      "step": 1829
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 1.4110547304153442,
+      "learning_rate": 1.341938999502441e-06,
+      "loss": 0.019,
+      "step": 1830
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 1.0071134567260742,
+      "learning_rate": 1.3388612668395175e-06,
+      "loss": 0.0043,
+      "step": 1831
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 1.337604284286499,
+      "learning_rate": 1.3357860535043881e-06,
+      "loss": 0.0155,
+      "step": 1832
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 1.6100351810455322,
+      "learning_rate": 1.3327133641610067e-06,
+      "loss": 0.0154,
+      "step": 1833
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 1.743006944656372,
+      "learning_rate": 1.3296432034694952e-06,
+      "loss": 0.012,
+      "step": 1834
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 1.6599669456481934,
+      "learning_rate": 1.3265755760861425e-06,
+      "loss": 0.0162,
+      "step": 1835
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 1.483189344406128,
+      "learning_rate": 1.3235104866633948e-06,
+      "loss": 0.0122,
+      "step": 1836
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 1.2852782011032104,
+      "learning_rate": 1.3204479398498496e-06,
+      "loss": 0.0089,
+      "step": 1837
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 2.171510696411133,
+      "learning_rate": 1.317387940290247e-06,
+      "loss": 0.0201,
+      "step": 1838
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 1.3694868087768555,
+      "learning_rate": 1.3143304926254664e-06,
+      "loss": 0.0086,
+      "step": 1839
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 1.8318085670471191,
+      "learning_rate": 1.3112756014925149e-06,
+      "loss": 0.0196,
+      "step": 1840
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 1.5885000228881836,
+      "learning_rate": 1.3082232715245227e-06,
+      "loss": 0.0127,
+      "step": 1841
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 1.0744603872299194,
+      "learning_rate": 1.305173507350734e-06,
+      "loss": 0.0079,
+      "step": 1842
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 1.551952600479126,
+      "learning_rate": 1.3021263135965063e-06,
+      "loss": 0.0109,
+      "step": 1843
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 1.1353726387023926,
+      "learning_rate": 1.2990816948832945e-06,
+      "loss": 0.0098,
+      "step": 1844
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 1.2844582796096802,
+      "learning_rate": 1.2960396558286484e-06,
+      "loss": 0.0105,
+      "step": 1845
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 1.9816162586212158,
+      "learning_rate": 1.2930002010462081e-06,
+      "loss": 0.0174,
+      "step": 1846
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 2.712876319885254,
+      "learning_rate": 1.2899633351456924e-06,
+      "loss": 0.0265,
+      "step": 1847
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 1.2672145366668701,
+      "learning_rate": 1.2869290627328939e-06,
+      "loss": 0.0122,
+      "step": 1848
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.7153728008270264,
+      "learning_rate": 1.2838973884096715e-06,
+      "loss": 0.009,
+      "step": 1849
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 1.4034442901611328,
+      "learning_rate": 1.2808683167739445e-06,
+      "loss": 0.0126,
+      "step": 1850
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 1.0858688354492188,
+      "learning_rate": 1.2778418524196835e-06,
+      "loss": 0.0106,
+      "step": 1851
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.9318276047706604,
+      "learning_rate": 1.2748179999369078e-06,
+      "loss": 0.0101,
+      "step": 1852
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.6633744239807129,
+      "learning_rate": 1.2717967639116733e-06,
+      "loss": 0.0049,
+      "step": 1853
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 1.8636674880981445,
+      "learning_rate": 1.2687781489260672e-06,
+      "loss": 0.0137,
+      "step": 1854
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 1.2148879766464233,
+      "learning_rate": 1.2657621595582018e-06,
+      "loss": 0.0114,
+      "step": 1855
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 1.8772797584533691,
+      "learning_rate": 1.26274880038221e-06,
+      "loss": 0.012,
+      "step": 1856
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 1.1231482028961182,
+      "learning_rate": 1.2597380759682325e-06,
+      "loss": 0.0134,
+      "step": 1857
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 2.5541090965270996,
+      "learning_rate": 1.2567299908824147e-06,
+      "loss": 0.0219,
+      "step": 1858
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 1.3695601224899292,
+      "learning_rate": 1.2537245496869003e-06,
+      "loss": 0.0102,
+      "step": 1859
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 1.8158103227615356,
+      "learning_rate": 1.250721756939822e-06,
+      "loss": 0.0132,
+      "step": 1860
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 2.249490261077881,
+      "learning_rate": 1.247721617195295e-06,
+      "loss": 0.0149,
+      "step": 1861
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 2.080064296722412,
+      "learning_rate": 1.2447241350034147e-06,
+      "loss": 0.0182,
+      "step": 1862
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.7345661520957947,
+      "learning_rate": 1.241729314910242e-06,
+      "loss": 0.0076,
+      "step": 1863
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 1.9775524139404297,
+      "learning_rate": 1.2387371614578022e-06,
+      "loss": 0.0148,
+      "step": 1864
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 2.0832226276397705,
+      "learning_rate": 1.2357476791840748e-06,
+      "loss": 0.0168,
+      "step": 1865
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 2.3076744079589844,
+      "learning_rate": 1.2327608726229908e-06,
+      "loss": 0.0142,
+      "step": 1866
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 1.7572283744812012,
+      "learning_rate": 1.229776746304421e-06,
+      "loss": 0.0193,
+      "step": 1867
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 1.2063733339309692,
+      "learning_rate": 1.2267953047541708e-06,
+      "loss": 0.0098,
+      "step": 1868
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 2.2456748485565186,
+      "learning_rate": 1.2238165524939766e-06,
+      "loss": 0.0177,
+      "step": 1869
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.8801554441452026,
+      "learning_rate": 1.2208404940414938e-06,
+      "loss": 0.0057,
+      "step": 1870
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.6722488403320312,
+      "learning_rate": 1.2178671339102925e-06,
+      "loss": 0.0124,
+      "step": 1871
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 2.0348072052001953,
+      "learning_rate": 1.2148964766098512e-06,
+      "loss": 0.0181,
+      "step": 1872
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.350773572921753,
+      "learning_rate": 1.2119285266455485e-06,
+      "loss": 0.0128,
+      "step": 1873
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.9413763284683228,
+      "learning_rate": 1.2089632885186565e-06,
+      "loss": 0.007,
+      "step": 1874
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.6112159490585327,
+      "learning_rate": 1.2060007667263375e-06,
+      "loss": 0.0166,
+      "step": 1875
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 1.0498020648956299,
+      "learning_rate": 1.2030409657616306e-06,
+      "loss": 0.0097,
+      "step": 1876
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 1.2480828762054443,
+      "learning_rate": 1.2000838901134503e-06,
+      "loss": 0.0078,
+      "step": 1877
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 1.661507248878479,
+      "learning_rate": 1.1971295442665756e-06,
+      "loss": 0.0165,
+      "step": 1878
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 2.152618885040283,
+      "learning_rate": 1.1941779327016494e-06,
+      "loss": 0.0269,
+      "step": 1879
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 1.1724984645843506,
+      "learning_rate": 1.1912290598951645e-06,
+      "loss": 0.0079,
+      "step": 1880
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 1.4129153490066528,
+      "learning_rate": 1.1882829303194606e-06,
+      "loss": 0.0132,
+      "step": 1881
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 1.293118953704834,
+      "learning_rate": 1.1853395484427177e-06,
+      "loss": 0.0124,
+      "step": 1882
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.9022077918052673,
+      "learning_rate": 1.1823989187289477e-06,
+      "loss": 0.0073,
+      "step": 1883
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 2.508563995361328,
+      "learning_rate": 1.179461045637988e-06,
+      "loss": 0.0219,
+      "step": 1884
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 1.272202730178833,
+      "learning_rate": 1.1765259336254985e-06,
+      "loss": 0.0098,
+      "step": 1885
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 1.4476107358932495,
+      "learning_rate": 1.1735935871429476e-06,
+      "loss": 0.0164,
+      "step": 1886
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 1.8265297412872314,
+      "learning_rate": 1.1706640106376104e-06,
+      "loss": 0.024,
+      "step": 1887
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 1.0646157264709473,
+      "learning_rate": 1.1677372085525638e-06,
+      "loss": 0.008,
+      "step": 1888
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 1.3727554082870483,
+      "learning_rate": 1.1648131853266734e-06,
+      "loss": 0.0092,
+      "step": 1889
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 1.1975858211517334,
+      "learning_rate": 1.161891945394592e-06,
+      "loss": 0.0099,
+      "step": 1890
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.4503860473632812,
+      "learning_rate": 1.1589734931867495e-06,
+      "loss": 0.008,
+      "step": 1891
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.1885504722595215,
+      "learning_rate": 1.1560578331293517e-06,
+      "loss": 0.0107,
+      "step": 1892
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.2859430313110352,
+      "learning_rate": 1.153144969644366e-06,
+      "loss": 0.0095,
+      "step": 1893
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.9196044206619263,
+      "learning_rate": 1.1502349071495192e-06,
+      "loss": 0.0114,
+      "step": 1894
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.3283616304397583,
+      "learning_rate": 1.1473276500582915e-06,
+      "loss": 0.015,
+      "step": 1895
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.5998096466064453,
+      "learning_rate": 1.1444232027799067e-06,
+      "loss": 0.0112,
+      "step": 1896
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 1.808823823928833,
+      "learning_rate": 1.1415215697193273e-06,
+      "loss": 0.0165,
+      "step": 1897
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 1.2320530414581299,
+      "learning_rate": 1.13862275527725e-06,
+      "loss": 0.0102,
+      "step": 1898
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 1.5036625862121582,
+      "learning_rate": 1.1357267638500937e-06,
+      "loss": 0.0103,
+      "step": 1899
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 1.3107577562332153,
+      "learning_rate": 1.1328335998299964e-06,
+      "loss": 0.0096,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "eval_blimp_filtered_avg": 0.7307462686567164,
+      "eval_blimp_filtered_std": 0.004871096950462897,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "eval_blimp_supplement_avg": 0.7974137931034483,
+      "eval_blimp_supplement_std": 0.017603725116253018,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 1.4927703142166138,
+      "learning_rate": 1.129943267604811e-06,
+      "loss": 0.0071,
+      "step": 1901
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 1.5945600271224976,
+      "learning_rate": 1.127055771558092e-06,
+      "loss": 0.0174,
+      "step": 1902
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 1.5043667554855347,
+      "learning_rate": 1.1241711160690933e-06,
+      "loss": 0.0072,
+      "step": 1903
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 1.5722846984863281,
+      "learning_rate": 1.1212893055127616e-06,
+      "loss": 0.018,
+      "step": 1904
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 1.504595160484314,
+      "learning_rate": 1.1184103442597289e-06,
+      "loss": 0.0139,
+      "step": 1905
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.7446656823158264,
+      "learning_rate": 1.1155342366763048e-06,
+      "loss": 0.0062,
+      "step": 1906
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 2.6022486686706543,
+      "learning_rate": 1.1126609871244706e-06,
+      "loss": 0.008,
+      "step": 1907
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 1.0146623849868774,
+      "learning_rate": 1.1097905999618765e-06,
+      "loss": 0.0131,
+      "step": 1908
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 1.168131947517395,
+      "learning_rate": 1.106923079541828e-06,
+      "loss": 0.0095,
+      "step": 1909
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 1.1857450008392334,
+      "learning_rate": 1.1040584302132828e-06,
+      "loss": 0.0113,
+      "step": 1910
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.8058726191520691,
+      "learning_rate": 1.1011966563208475e-06,
+      "loss": 0.0074,
+      "step": 1911
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 2.019279956817627,
+      "learning_rate": 1.0983377622047642e-06,
+      "loss": 0.0157,
+      "step": 1912
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 1.7267680168151855,
+      "learning_rate": 1.0954817522009086e-06,
+      "loss": 0.016,
+      "step": 1913
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 1.6040393114089966,
+      "learning_rate": 1.0926286306407821e-06,
+      "loss": 0.0149,
+      "step": 1914
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 1.1337083578109741,
+      "learning_rate": 1.0897784018515076e-06,
+      "loss": 0.0111,
+      "step": 1915
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 1.743375539779663,
+      "learning_rate": 1.086931070155818e-06,
+      "loss": 0.0113,
+      "step": 1916
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 2.2871227264404297,
+      "learning_rate": 1.0840866398720529e-06,
+      "loss": 0.0157,
+      "step": 1917
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 1.2900002002716064,
+      "learning_rate": 1.0812451153141524e-06,
+      "loss": 0.015,
+      "step": 1918
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 1.2271316051483154,
+      "learning_rate": 1.078406500791649e-06,
+      "loss": 0.0066,
+      "step": 1919
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.6867360472679138,
+      "learning_rate": 1.0755708006096616e-06,
+      "loss": 0.0055,
+      "step": 1920
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 1.169236183166504,
+      "learning_rate": 1.072738019068891e-06,
+      "loss": 0.0137,
+      "step": 1921
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 1.626917839050293,
+      "learning_rate": 1.0699081604656086e-06,
+      "loss": 0.0078,
+      "step": 1922
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.7707490921020508,
+      "learning_rate": 1.0670812290916544e-06,
+      "loss": 0.0102,
+      "step": 1923
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 1.1932252645492554,
+      "learning_rate": 1.0642572292344302e-06,
+      "loss": 0.0117,
+      "step": 1924
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 1.3485758304595947,
+      "learning_rate": 1.0614361651768892e-06,
+      "loss": 0.0092,
+      "step": 1925
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 1.2627525329589844,
+      "learning_rate": 1.0586180411975336e-06,
+      "loss": 0.0132,
+      "step": 1926
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 2.363339424133301,
+      "learning_rate": 1.0558028615704056e-06,
+      "loss": 0.0158,
+      "step": 1927
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 1.7450156211853027,
+      "learning_rate": 1.0529906305650834e-06,
+      "loss": 0.0169,
+      "step": 1928
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 1.2049115896224976,
+      "learning_rate": 1.0501813524466714e-06,
+      "loss": 0.0097,
+      "step": 1929
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 1.2596123218536377,
+      "learning_rate": 1.0473750314757965e-06,
+      "loss": 0.0118,
+      "step": 1930
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.8746306896209717,
+      "learning_rate": 1.0445716719086018e-06,
+      "loss": 0.0073,
+      "step": 1931
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 1.77753746509552,
+      "learning_rate": 1.0417712779967378e-06,
+      "loss": 0.0099,
+      "step": 1932
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 1.7045754194259644,
+      "learning_rate": 1.0389738539873559e-06,
+      "loss": 0.0118,
+      "step": 1933
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 2.7210423946380615,
+      "learning_rate": 1.036179404123107e-06,
+      "loss": 0.0251,
+      "step": 1934
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.8575809001922607,
+      "learning_rate": 1.0333879326421285e-06,
+      "loss": 0.0065,
+      "step": 1935
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 1.9449830055236816,
+      "learning_rate": 1.03059944377804e-06,
+      "loss": 0.0187,
+      "step": 1936
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 1.0673350095748901,
+      "learning_rate": 1.0278139417599402e-06,
+      "loss": 0.0091,
+      "step": 1937
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 1.6474530696868896,
+      "learning_rate": 1.0250314308123967e-06,
+      "loss": 0.0144,
+      "step": 1938
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 1.172705888748169,
+      "learning_rate": 1.0222519151554402e-06,
+      "loss": 0.0116,
+      "step": 1939
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 1.2624338865280151,
+      "learning_rate": 1.019475399004559e-06,
+      "loss": 0.0126,
+      "step": 1940
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 1.7820625305175781,
+      "learning_rate": 1.0167018865706923e-06,
+      "loss": 0.0128,
+      "step": 1941
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 2.0735831260681152,
+      "learning_rate": 1.0139313820602237e-06,
+      "loss": 0.0201,
+      "step": 1942
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 1.282099962234497,
+      "learning_rate": 1.0111638896749747e-06,
+      "loss": 0.0064,
+      "step": 1943
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 1.1267735958099365,
+      "learning_rate": 1.0083994136122003e-06,
+      "loss": 0.0097,
+      "step": 1944
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 3.1521356105804443,
+      "learning_rate": 1.0056379580645786e-06,
+      "loss": 0.0191,
+      "step": 1945
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 1.324111819267273,
+      "learning_rate": 1.0028795272202067e-06,
+      "loss": 0.0106,
+      "step": 1946
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.920366108417511,
+      "learning_rate": 1.000124125262597e-06,
+      "loss": 0.0064,
+      "step": 1947
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 1.910099744796753,
+      "learning_rate": 9.973717563706654e-07,
+      "loss": 0.0194,
+      "step": 1948
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 1.2067381143569946,
+      "learning_rate": 9.94622424718729e-07,
+      "loss": 0.0103,
+      "step": 1949
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 1.4103349447250366,
+      "learning_rate": 9.91876134476498e-07,
+      "loss": 0.0126,
+      "step": 1950
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.9455344676971436,
+      "learning_rate": 9.891328898090706e-07,
+      "loss": 0.0082,
+      "step": 1951
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 1.592507004737854,
+      "learning_rate": 9.863926948769251e-07,
+      "loss": 0.0099,
+      "step": 1952
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 1.1439069509506226,
+      "learning_rate": 9.836555538359148e-07,
+      "loss": 0.0058,
+      "step": 1953
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 1.5229073762893677,
+      "learning_rate": 9.80921470837263e-07,
+      "loss": 0.0089,
+      "step": 1954
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 1.3269532918930054,
+      "learning_rate": 9.781904500275526e-07,
+      "loss": 0.0168,
+      "step": 1955
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 1.148062825202942,
+      "learning_rate": 9.754624955487232e-07,
+      "loss": 0.0098,
+      "step": 1956
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 1.8106926679611206,
+      "learning_rate": 9.727376115380652e-07,
+      "loss": 0.0172,
+      "step": 1957
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 1.140676498413086,
+      "learning_rate": 9.700158021282107e-07,
+      "loss": 0.0086,
+      "step": 1958
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 1.265950083732605,
+      "learning_rate": 9.672970714471283e-07,
+      "loss": 0.0107,
+      "step": 1959
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 1.216072678565979,
+      "learning_rate": 9.645814236181196e-07,
+      "loss": 0.0087,
+      "step": 1960
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 1.2524449825286865,
+      "learning_rate": 9.61868862759809e-07,
+      "loss": 0.0073,
+      "step": 1961
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 1.360849380493164,
+      "learning_rate": 9.591593929861405e-07,
+      "loss": 0.0128,
+      "step": 1962
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 1.588875412940979,
+      "learning_rate": 9.564530184063654e-07,
+      "loss": 0.0135,
+      "step": 1963
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.9504747986793518,
+      "learning_rate": 9.537497431250479e-07,
+      "loss": 0.012,
+      "step": 1964
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 2.2380545139312744,
+      "learning_rate": 9.510495712420471e-07,
+      "loss": 0.0187,
+      "step": 1965
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 1.7045855522155762,
+      "learning_rate": 9.48352506852515e-07,
+      "loss": 0.0071,
+      "step": 1966
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 1.3318544626235962,
+      "learning_rate": 9.456585540468949e-07,
+      "loss": 0.0091,
+      "step": 1967
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.9779008030891418,
+      "learning_rate": 9.429677169109071e-07,
+      "loss": 0.0085,
+      "step": 1968
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 1.4517345428466797,
+      "learning_rate": 9.402799995255461e-07,
+      "loss": 0.0229,
+      "step": 1969
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 1.0749115943908691,
+      "learning_rate": 9.375954059670796e-07,
+      "loss": 0.0101,
+      "step": 1970
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 1.304335594177246,
+      "learning_rate": 9.349139403070327e-07,
+      "loss": 0.0101,
+      "step": 1971
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 1.8731235265731812,
+      "learning_rate": 9.32235606612189e-07,
+      "loss": 0.012,
+      "step": 1972
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 2.1465225219726562,
+      "learning_rate": 9.295604089445812e-07,
+      "loss": 0.021,
+      "step": 1973
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.7805631160736084,
+      "learning_rate": 9.268883513614867e-07,
+      "loss": 0.0059,
+      "step": 1974
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 1.0019477605819702,
+      "learning_rate": 9.2421943791542e-07,
+      "loss": 0.0097,
+      "step": 1975
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 1.5195963382720947,
+      "learning_rate": 9.215536726541259e-07,
+      "loss": 0.0209,
+      "step": 1976
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 1.8028939962387085,
+      "learning_rate": 9.188910596205779e-07,
+      "loss": 0.0208,
+      "step": 1977
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 2.336245059967041,
+      "learning_rate": 9.162316028529663e-07,
+      "loss": 0.0121,
+      "step": 1978
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 1.7370715141296387,
+      "learning_rate": 9.135753063846931e-07,
+      "loss": 0.0217,
+      "step": 1979
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 2.510857343673706,
+      "learning_rate": 9.10922174244372e-07,
+      "loss": 0.0195,
+      "step": 1980
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 1.4509906768798828,
+      "learning_rate": 9.082722104558127e-07,
+      "loss": 0.0083,
+      "step": 1981
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 1.0969922542572021,
+      "learning_rate": 9.05625419038022e-07,
+      "loss": 0.0093,
+      "step": 1982
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 1.802712321281433,
+      "learning_rate": 9.02981804005196e-07,
+      "loss": 0.0171,
+      "step": 1983
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.9224696159362793,
+      "learning_rate": 9.003413693667124e-07,
+      "loss": 0.0107,
+      "step": 1984
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 1.128373384475708,
+      "learning_rate": 8.977041191271248e-07,
+      "loss": 0.0091,
+      "step": 1985
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 1.3442199230194092,
+      "learning_rate": 8.950700572861587e-07,
+      "loss": 0.0118,
+      "step": 1986
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 1.7752063274383545,
+      "learning_rate": 8.92439187838703e-07,
+      "loss": 0.0139,
+      "step": 1987
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 1.0391522645950317,
+      "learning_rate": 8.898115147748053e-07,
+      "loss": 0.0115,
+      "step": 1988
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 2.349747896194458,
+      "learning_rate": 8.871870420796645e-07,
+      "loss": 0.0329,
+      "step": 1989
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 1.3308055400848389,
+      "learning_rate": 8.845657737336285e-07,
+      "loss": 0.013,
+      "step": 1990
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 1.990352988243103,
+      "learning_rate": 8.819477137121824e-07,
+      "loss": 0.015,
+      "step": 1991
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.9006649255752563,
+      "learning_rate": 8.793328659859462e-07,
+      "loss": 0.0081,
+      "step": 1992
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.9341894388198853,
+      "learning_rate": 8.767212345206698e-07,
+      "loss": 0.0051,
+      "step": 1993
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 1.2526183128356934,
+      "learning_rate": 8.741128232772235e-07,
+      "loss": 0.0158,
+      "step": 1994
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 1.6916507482528687,
+      "learning_rate": 8.715076362115937e-07,
+      "loss": 0.0108,
+      "step": 1995
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 1.4565813541412354,
+      "learning_rate": 8.689056772748774e-07,
+      "loss": 0.0118,
+      "step": 1996
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 2.056687593460083,
+      "learning_rate": 8.663069504132758e-07,
+      "loss": 0.0106,
+      "step": 1997
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 1.2701400518417358,
+      "learning_rate": 8.637114595680876e-07,
+      "loss": 0.0141,
+      "step": 1998
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.8035250306129456,
+      "learning_rate": 8.611192086757052e-07,
+      "loss": 0.0067,
+      "step": 1999
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.7013640999794006,
+      "learning_rate": 8.585302016676058e-07,
+      "loss": 0.0069,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "eval_blimp_filtered_avg": 0.7313432835820896,
+      "eval_blimp_filtered_std": 0.004868139617291788,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "eval_blimp_supplement_avg": 0.8017241379310345,
+      "eval_blimp_supplement_std": 0.01749645799500491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "eval_winoground_filtered_avg": 0.47,
+      "eval_winoground_filtered_std": 0.0501613558046592,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.8819791674613953,
+      "learning_rate": 8.559444424703472e-07,
+      "loss": 0.0086,
+      "step": 2001
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 1.6181796789169312,
+      "learning_rate": 8.533619350055608e-07,
+      "loss": 0.0185,
+      "step": 2002
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 1.5702379941940308,
+      "learning_rate": 8.507826831899483e-07,
+      "loss": 0.0158,
+      "step": 2003
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.9878043532371521,
+      "learning_rate": 8.482066909352724e-07,
+      "loss": 0.0063,
+      "step": 2004
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 1.2024034261703491,
+      "learning_rate": 8.456339621483508e-07,
+      "loss": 0.0166,
+      "step": 2005
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 2.4302847385406494,
+      "learning_rate": 8.430645007310557e-07,
+      "loss": 0.0112,
+      "step": 2006
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 2.2916269302368164,
+      "learning_rate": 8.404983105803002e-07,
+      "loss": 0.0173,
+      "step": 2007
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 1.6374431848526,
+      "learning_rate": 8.379353955880379e-07,
+      "loss": 0.013,
+      "step": 2008
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.8495174646377563,
+      "learning_rate": 8.353757596412547e-07,
+      "loss": 0.0098,
+      "step": 2009
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 1.0661263465881348,
+      "learning_rate": 8.328194066219634e-07,
+      "loss": 0.0084,
+      "step": 2010
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 2.2611260414123535,
+      "learning_rate": 8.302663404071975e-07,
+      "loss": 0.0221,
+      "step": 2011
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 1.2480194568634033,
+      "learning_rate": 8.277165648690059e-07,
+      "loss": 0.0079,
+      "step": 2012
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.6549211144447327,
+      "learning_rate": 8.251700838744486e-07,
+      "loss": 0.0038,
+      "step": 2013
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 1.2810754776000977,
+      "learning_rate": 8.226269012855862e-07,
+      "loss": 0.009,
+      "step": 2014
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 2.2266530990600586,
+      "learning_rate": 8.200870209594769e-07,
+      "loss": 0.0186,
+      "step": 2015
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 1.87469482421875,
+      "learning_rate": 8.175504467481737e-07,
+      "loss": 0.0129,
+      "step": 2016
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 1.5738635063171387,
+      "learning_rate": 8.150171824987131e-07,
+      "loss": 0.0133,
+      "step": 2017
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.596197783946991,
+      "learning_rate": 8.124872320531111e-07,
+      "loss": 0.0059,
+      "step": 2018
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.9299052357673645,
+      "learning_rate": 8.09960599248359e-07,
+      "loss": 0.0097,
+      "step": 2019
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 1.5410739183425903,
+      "learning_rate": 8.074372879164165e-07,
+      "loss": 0.0108,
+      "step": 2020
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 1.4591364860534668,
+      "learning_rate": 8.049173018842047e-07,
+      "loss": 0.0184,
+      "step": 2021
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 1.8295645713806152,
+      "learning_rate": 8.024006449736041e-07,
+      "loss": 0.0126,
+      "step": 2022
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.9002048969268799,
+      "learning_rate": 7.998873210014435e-07,
+      "loss": 0.0094,
+      "step": 2023
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 2.543215036392212,
+      "learning_rate": 7.973773337794977e-07,
+      "loss": 0.0145,
+      "step": 2024
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 1.425246238708496,
+      "learning_rate": 7.948706871144799e-07,
+      "loss": 0.0173,
+      "step": 2025
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 1.1699798107147217,
+      "learning_rate": 7.923673848080405e-07,
+      "loss": 0.0118,
+      "step": 2026
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 1.0060489177703857,
+      "learning_rate": 7.898674306567539e-07,
+      "loss": 0.0156,
+      "step": 2027
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 1.2602635622024536,
+      "learning_rate": 7.87370828452117e-07,
+      "loss": 0.0104,
+      "step": 2028
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 2.092163324356079,
+      "learning_rate": 7.848775819805464e-07,
+      "loss": 0.0106,
+      "step": 2029
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.8655326962471008,
+      "learning_rate": 7.823876950233651e-07,
+      "loss": 0.0119,
+      "step": 2030
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 1.81997549533844,
+      "learning_rate": 7.799011713568036e-07,
+      "loss": 0.0133,
+      "step": 2031
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.9601938128471375,
+      "learning_rate": 7.774180147519904e-07,
+      "loss": 0.0065,
+      "step": 2032
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 1.8192497491836548,
+      "learning_rate": 7.749382289749479e-07,
+      "loss": 0.0211,
+      "step": 2033
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 1.666316270828247,
+      "learning_rate": 7.724618177865852e-07,
+      "loss": 0.0114,
+      "step": 2034
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 1.7068549394607544,
+      "learning_rate": 7.69988784942696e-07,
+      "loss": 0.0115,
+      "step": 2035
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 1.3636983633041382,
+      "learning_rate": 7.675191341939474e-07,
+      "loss": 0.0143,
+      "step": 2036
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 2.537733316421509,
+      "learning_rate": 7.650528692858789e-07,
+      "loss": 0.0158,
+      "step": 2037
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 1.2003999948501587,
+      "learning_rate": 7.62589993958893e-07,
+      "loss": 0.0081,
+      "step": 2038
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.4230929613113403,
+      "learning_rate": 7.601305119482549e-07,
+      "loss": 0.0159,
+      "step": 2039
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.9969327449798584,
+      "learning_rate": 7.576744269840803e-07,
+      "loss": 0.0106,
+      "step": 2040
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.6633245944976807,
+      "learning_rate": 7.552217427913338e-07,
+      "loss": 0.0111,
+      "step": 2041
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.3165638446807861,
+      "learning_rate": 7.527724630898229e-07,
+      "loss": 0.0168,
+      "step": 2042
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.7115468978881836,
+      "learning_rate": 7.503265915941906e-07,
+      "loss": 0.0054,
+      "step": 2043
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 2.3356001377105713,
+      "learning_rate": 7.478841320139108e-07,
+      "loss": 0.0148,
+      "step": 2044
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 2.3560001850128174,
+      "learning_rate": 7.454450880532855e-07,
+      "loss": 0.0208,
+      "step": 2045
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 2.3037476539611816,
+      "learning_rate": 7.430094634114336e-07,
+      "loss": 0.017,
+      "step": 2046
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 1.4733995199203491,
+      "learning_rate": 7.405772617822883e-07,
+      "loss": 0.013,
+      "step": 2047
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 1.321306824684143,
+      "learning_rate": 7.38148486854594e-07,
+      "loss": 0.0109,
+      "step": 2048
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 1.3658934831619263,
+      "learning_rate": 7.357231423118951e-07,
+      "loss": 0.0116,
+      "step": 2049
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 1.3277403116226196,
+      "learning_rate": 7.333012318325349e-07,
+      "loss": 0.0164,
+      "step": 2050
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 1.1482038497924805,
+      "learning_rate": 7.308827590896468e-07,
+      "loss": 0.0115,
+      "step": 2051
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 1.1053804159164429,
+      "learning_rate": 7.284677277511538e-07,
+      "loss": 0.0091,
+      "step": 2052
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.9881179332733154,
+      "learning_rate": 7.260561414797568e-07,
+      "loss": 0.01,
+      "step": 2053
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.7786324620246887,
+      "learning_rate": 7.236480039329319e-07,
+      "loss": 0.0058,
+      "step": 2054
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 1.2092219591140747,
+      "learning_rate": 7.212433187629266e-07,
+      "loss": 0.0147,
+      "step": 2055
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 1.357296109199524,
+      "learning_rate": 7.188420896167503e-07,
+      "loss": 0.0052,
+      "step": 2056
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 1.143980860710144,
+      "learning_rate": 7.164443201361711e-07,
+      "loss": 0.0078,
+      "step": 2057
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 2.5519163608551025,
+      "learning_rate": 7.140500139577133e-07,
+      "loss": 0.0218,
+      "step": 2058
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 1.8525258302688599,
+      "learning_rate": 7.116591747126445e-07,
+      "loss": 0.0111,
+      "step": 2059
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.8707548975944519,
+      "learning_rate": 7.092718060269762e-07,
+      "loss": 0.0049,
+      "step": 2060
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 2.1339242458343506,
+      "learning_rate": 7.06887911521456e-07,
+      "loss": 0.0167,
+      "step": 2061
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 1.6078965663909912,
+      "learning_rate": 7.045074948115631e-07,
+      "loss": 0.0122,
+      "step": 2062
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 1.3988463878631592,
+      "learning_rate": 7.021305595075015e-07,
+      "loss": 0.0123,
+      "step": 2063
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.9652006030082703,
+      "learning_rate": 6.997571092141956e-07,
+      "loss": 0.0104,
+      "step": 2064
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 1.4588285684585571,
+      "learning_rate": 6.973871475312835e-07,
+      "loss": 0.0084,
+      "step": 2065
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 1.9471288919448853,
+      "learning_rate": 6.950206780531132e-07,
+      "loss": 0.0144,
+      "step": 2066
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 1.0451194047927856,
+      "learning_rate": 6.926577043687356e-07,
+      "loss": 0.0071,
+      "step": 2067
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 1.2846094369888306,
+      "learning_rate": 6.902982300619015e-07,
+      "loss": 0.0097,
+      "step": 2068
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 1.385936975479126,
+      "learning_rate": 6.87942258711053e-07,
+      "loss": 0.0096,
+      "step": 2069
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 1.4081095457077026,
+      "learning_rate": 6.855897938893185e-07,
+      "loss": 0.0151,
+      "step": 2070
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 2.048920154571533,
+      "learning_rate": 6.832408391645112e-07,
+      "loss": 0.0163,
+      "step": 2071
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.5026159286499023,
+      "learning_rate": 6.808953980991186e-07,
+      "loss": 0.0041,
+      "step": 2072
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 1.1133794784545898,
+      "learning_rate": 6.785534742503e-07,
+      "loss": 0.013,
+      "step": 2073
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 1.4542337656021118,
+      "learning_rate": 6.762150711698789e-07,
+      "loss": 0.0109,
+      "step": 2074
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 1.9688831567764282,
+      "learning_rate": 6.738801924043424e-07,
+      "loss": 0.0148,
+      "step": 2075
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 1.2438271045684814,
+      "learning_rate": 6.715488414948302e-07,
+      "loss": 0.0123,
+      "step": 2076
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.9726313948631287,
+      "learning_rate": 6.692210219771313e-07,
+      "loss": 0.0083,
+      "step": 2077
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 1.2543562650680542,
+      "learning_rate": 6.668967373816797e-07,
+      "loss": 0.011,
+      "step": 2078
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 1.1004064083099365,
+      "learning_rate": 6.645759912335482e-07,
+      "loss": 0.0115,
+      "step": 2079
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 1.5934683084487915,
+      "learning_rate": 6.622587870524423e-07,
+      "loss": 0.014,
+      "step": 2080
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 1.4840106964111328,
+      "learning_rate": 6.599451283526977e-07,
+      "loss": 0.0163,
+      "step": 2081
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 2.356950044631958,
+      "learning_rate": 6.576350186432711e-07,
+      "loss": 0.0174,
+      "step": 2082
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 1.141806721687317,
+      "learning_rate": 6.553284614277361e-07,
+      "loss": 0.0172,
+      "step": 2083
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 1.998517394065857,
+      "learning_rate": 6.530254602042812e-07,
+      "loss": 0.0063,
+      "step": 2084
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 2.07460355758667,
+      "learning_rate": 6.507260184656993e-07,
+      "loss": 0.0175,
+      "step": 2085
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.2850923538208008,
+      "learning_rate": 6.484301396993858e-07,
+      "loss": 0.0139,
+      "step": 2086
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.330918550491333,
+      "learning_rate": 6.461378273873326e-07,
+      "loss": 0.0118,
+      "step": 2087
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.5412259101867676,
+      "learning_rate": 6.438490850061221e-07,
+      "loss": 0.0102,
+      "step": 2088
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.4215147495269775,
+      "learning_rate": 6.415639160269227e-07,
+      "loss": 0.0072,
+      "step": 2089
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.258819580078125,
+      "learning_rate": 6.392823239154824e-07,
+      "loss": 0.0117,
+      "step": 2090
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 2.0174872875213623,
+      "learning_rate": 6.370043121321275e-07,
+      "loss": 0.0109,
+      "step": 2091
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 2.113384962081909,
+      "learning_rate": 6.347298841317506e-07,
+      "loss": 0.0134,
+      "step": 2092
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.8860541582107544,
+      "learning_rate": 6.324590433638101e-07,
+      "loss": 0.0068,
+      "step": 2093
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 1.213399887084961,
+      "learning_rate": 6.301917932723258e-07,
+      "loss": 0.0076,
+      "step": 2094
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 1.6774463653564453,
+      "learning_rate": 6.279281372958694e-07,
+      "loss": 0.0169,
+      "step": 2095
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 1.360514760017395,
+      "learning_rate": 6.256680788675624e-07,
+      "loss": 0.0069,
+      "step": 2096
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 1.4548945426940918,
+      "learning_rate": 6.234116214150713e-07,
+      "loss": 0.0059,
+      "step": 2097
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.7219619750976562,
+      "learning_rate": 6.211587683605995e-07,
+      "loss": 0.0059,
+      "step": 2098
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 2.4164271354675293,
+      "learning_rate": 6.18909523120885e-07,
+      "loss": 0.0118,
+      "step": 2099
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 2.2620348930358887,
+      "learning_rate": 6.166638891071941e-07,
+      "loss": 0.0143,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "eval_blimp_filtered_avg": 0.7308955223880597,
+      "eval_blimp_filtered_std": 0.00487499880640042,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.017513892765077744,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 1.5732755661010742,
+      "learning_rate": 6.14421869725315e-07,
+      "loss": 0.0132,
+      "step": 2101
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.7027840614318848,
+      "learning_rate": 6.121834683755559e-07,
+      "loss": 0.0139,
+      "step": 2102
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.3724693059921265,
+      "learning_rate": 6.099486884527346e-07,
+      "loss": 0.0099,
+      "step": 2103
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.8518579602241516,
+      "learning_rate": 6.07717533346181e-07,
+      "loss": 0.0062,
+      "step": 2104
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.2080609798431396,
+      "learning_rate": 6.054900064397243e-07,
+      "loss": 0.0103,
+      "step": 2105
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.6411821842193604,
+      "learning_rate": 6.032661111116906e-07,
+      "loss": 0.0077,
+      "step": 2106
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.9128040075302124,
+      "learning_rate": 6.010458507349013e-07,
+      "loss": 0.0045,
+      "step": 2107
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.7447566390037537,
+      "learning_rate": 5.988292286766627e-07,
+      "loss": 0.0097,
+      "step": 2108
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 1.1188596487045288,
+      "learning_rate": 5.966162482987633e-07,
+      "loss": 0.0098,
+      "step": 2109
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.8513658046722412,
+      "learning_rate": 5.944069129574686e-07,
+      "loss": 0.0069,
+      "step": 2110
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 1.2933748960494995,
+      "learning_rate": 5.922012260035168e-07,
+      "loss": 0.0094,
+      "step": 2111
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.2002919912338257,
+      "learning_rate": 5.899991907821116e-07,
+      "loss": 0.0065,
+      "step": 2112
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.1026842594146729,
+      "learning_rate": 5.878008106329184e-07,
+      "loss": 0.0072,
+      "step": 2113
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.035070538520813,
+      "learning_rate": 5.856060888900613e-07,
+      "loss": 0.0069,
+      "step": 2114
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.2509002685546875,
+      "learning_rate": 5.834150288821142e-07,
+      "loss": 0.01,
+      "step": 2115
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.516931176185608,
+      "learning_rate": 5.812276339320962e-07,
+      "loss": 0.008,
+      "step": 2116
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.770128846168518,
+      "learning_rate": 5.790439073574713e-07,
+      "loss": 0.0115,
+      "step": 2117
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.9295368194580078,
+      "learning_rate": 5.768638524701377e-07,
+      "loss": 0.0113,
+      "step": 2118
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.9900155067443848,
+      "learning_rate": 5.746874725764245e-07,
+      "loss": 0.0061,
+      "step": 2119
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.5973102450370789,
+      "learning_rate": 5.725147709770897e-07,
+      "loss": 0.0036,
+      "step": 2120
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 1.1530249118804932,
+      "learning_rate": 5.703457509673098e-07,
+      "loss": 0.0062,
+      "step": 2121
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.971772313117981,
+      "learning_rate": 5.681804158366797e-07,
+      "loss": 0.0095,
+      "step": 2122
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 1.0491307973861694,
+      "learning_rate": 5.660187688692047e-07,
+      "loss": 0.0083,
+      "step": 2123
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 1.4142680168151855,
+      "learning_rate": 5.638608133432973e-07,
+      "loss": 0.0127,
+      "step": 2124
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.7620598673820496,
+      "learning_rate": 5.617065525317706e-07,
+      "loss": 0.0085,
+      "step": 2125
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.6016783118247986,
+      "learning_rate": 5.595559897018342e-07,
+      "loss": 0.0059,
+      "step": 2126
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.4560791552066803,
+      "learning_rate": 5.574091281150907e-07,
+      "loss": 0.0041,
+      "step": 2127
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 1.6730092763900757,
+      "learning_rate": 5.55265971027528e-07,
+      "loss": 0.0146,
+      "step": 2128
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.9316773414611816,
+      "learning_rate": 5.531265216895149e-07,
+      "loss": 0.0073,
+      "step": 2129
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 1.8690471649169922,
+      "learning_rate": 5.509907833457997e-07,
+      "loss": 0.0154,
+      "step": 2130
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 1.3836270570755005,
+      "learning_rate": 5.488587592354998e-07,
+      "loss": 0.0086,
+      "step": 2131
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.6450074911117554,
+      "learning_rate": 5.467304525921007e-07,
+      "loss": 0.0044,
+      "step": 2132
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.7457317113876343,
+      "learning_rate": 5.446058666434494e-07,
+      "loss": 0.0057,
+      "step": 2133
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.5213141441345215,
+      "learning_rate": 5.424850046117501e-07,
+      "loss": 0.0046,
+      "step": 2134
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.814653754234314,
+      "learning_rate": 5.403678697135597e-07,
+      "loss": 0.0066,
+      "step": 2135
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.7370361685752869,
+      "learning_rate": 5.382544651597813e-07,
+      "loss": 0.0083,
+      "step": 2136
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.7197270393371582,
+      "learning_rate": 5.361447941556626e-07,
+      "loss": 0.0057,
+      "step": 2137
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 1.6175410747528076,
+      "learning_rate": 5.340388599007867e-07,
+      "loss": 0.008,
+      "step": 2138
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 1.460473656654358,
+      "learning_rate": 5.319366655890697e-07,
+      "loss": 0.0087,
+      "step": 2139
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 1.52675461769104,
+      "learning_rate": 5.298382144087577e-07,
+      "loss": 0.0126,
+      "step": 2140
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.8698464035987854,
+      "learning_rate": 5.277435095424171e-07,
+      "loss": 0.0062,
+      "step": 2141
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.6765862107276917,
+      "learning_rate": 5.25652554166934e-07,
+      "loss": 0.0047,
+      "step": 2142
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 1.8694639205932617,
+      "learning_rate": 5.235653514535084e-07,
+      "loss": 0.0079,
+      "step": 2143
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.7510217428207397,
+      "learning_rate": 5.214819045676477e-07,
+      "loss": 0.0092,
+      "step": 2144
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 1.1484562158584595,
+      "learning_rate": 5.194022166691637e-07,
+      "loss": 0.0109,
+      "step": 2145
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 1.1169986724853516,
+      "learning_rate": 5.17326290912167e-07,
+      "loss": 0.0081,
+      "step": 2146
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 1.0156010389328003,
+      "learning_rate": 5.152541304450626e-07,
+      "loss": 0.0052,
+      "step": 2147
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 1.1187801361083984,
+      "learning_rate": 5.131857384105451e-07,
+      "loss": 0.0059,
+      "step": 2148
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 1.4852416515350342,
+      "learning_rate": 5.111211179455929e-07,
+      "loss": 0.013,
+      "step": 2149
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.6208791732788086,
+      "learning_rate": 5.090602721814663e-07,
+      "loss": 0.0042,
+      "step": 2150
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 1.208947777748108,
+      "learning_rate": 5.070032042436992e-07,
+      "loss": 0.0085,
+      "step": 2151
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.7779290676116943,
+      "learning_rate": 5.049499172520952e-07,
+      "loss": 0.004,
+      "step": 2152
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.8111461997032166,
+      "learning_rate": 5.02900414320726e-07,
+      "loss": 0.0089,
+      "step": 2153
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.2990548610687256,
+      "learning_rate": 5.008546985579229e-07,
+      "loss": 0.0092,
+      "step": 2154
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.0858360528945923,
+      "learning_rate": 4.988127730662734e-07,
+      "loss": 0.0057,
+      "step": 2155
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.8601486682891846,
+      "learning_rate": 4.967746409426166e-07,
+      "loss": 0.0038,
+      "step": 2156
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.5764520764350891,
+      "learning_rate": 4.947403052780386e-07,
+      "loss": 0.0037,
+      "step": 2157
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.1170367002487183,
+      "learning_rate": 4.927097691578677e-07,
+      "loss": 0.0093,
+      "step": 2158
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.6167320609092712,
+      "learning_rate": 4.906830356616692e-07,
+      "loss": 0.0038,
+      "step": 2159
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 2.0455071926116943,
+      "learning_rate": 4.886601078632433e-07,
+      "loss": 0.0131,
+      "step": 2160
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.6909149885177612,
+      "learning_rate": 4.866409888306157e-07,
+      "loss": 0.0084,
+      "step": 2161
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.5104364156723022,
+      "learning_rate": 4.846256816260364e-07,
+      "loss": 0.0112,
+      "step": 2162
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.2840027809143066,
+      "learning_rate": 4.82614189305976e-07,
+      "loss": 0.0102,
+      "step": 2163
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.0956279039382935,
+      "learning_rate": 4.806065149211173e-07,
+      "loss": 0.0089,
+      "step": 2164
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 1.2555795907974243,
+      "learning_rate": 4.78602661516353e-07,
+      "loss": 0.0117,
+      "step": 2165
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.6928210258483887,
+      "learning_rate": 4.766026321307818e-07,
+      "loss": 0.0052,
+      "step": 2166
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 1.1673662662506104,
+      "learning_rate": 4.7460642979770196e-07,
+      "loss": 0.0049,
+      "step": 2167
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.8430702686309814,
+      "learning_rate": 4.726140575446083e-07,
+      "loss": 0.006,
+      "step": 2168
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.638598620891571,
+      "learning_rate": 4.7062551839318536e-07,
+      "loss": 0.0032,
+      "step": 2169
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.2875529527664185,
+      "learning_rate": 4.686408153593057e-07,
+      "loss": 0.0098,
+      "step": 2170
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.8127309083938599,
+      "learning_rate": 4.6665995145302356e-07,
+      "loss": 0.004,
+      "step": 2171
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.494848608970642,
+      "learning_rate": 4.6468292967856973e-07,
+      "loss": 0.0099,
+      "step": 2172
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.2655764818191528,
+      "learning_rate": 4.6270975303435015e-07,
+      "loss": 0.0118,
+      "step": 2173
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.2465636730194092,
+      "learning_rate": 4.60740424512937e-07,
+      "loss": 0.006,
+      "step": 2174
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.0596250295639038,
+      "learning_rate": 4.5877494710106635e-07,
+      "loss": 0.0095,
+      "step": 2175
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 1.004158854484558,
+      "learning_rate": 4.568133237796356e-07,
+      "loss": 0.0103,
+      "step": 2176
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 1.3770273923873901,
+      "learning_rate": 4.548555575236948e-07,
+      "loss": 0.011,
+      "step": 2177
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 1.977569341659546,
+      "learning_rate": 4.5290165130244523e-07,
+      "loss": 0.0106,
+      "step": 2178
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 1.6919872760772705,
+      "learning_rate": 4.5095160807923375e-07,
+      "loss": 0.0072,
+      "step": 2179
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 1.9308655261993408,
+      "learning_rate": 4.490054308115483e-07,
+      "loss": 0.0119,
+      "step": 2180
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.9821391105651855,
+      "learning_rate": 4.470631224510132e-07,
+      "loss": 0.006,
+      "step": 2181
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 1.4615651369094849,
+      "learning_rate": 4.451246859433871e-07,
+      "loss": 0.0056,
+      "step": 2182
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 1.062390923500061,
+      "learning_rate": 4.431901242285542e-07,
+      "loss": 0.0096,
+      "step": 2183
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 1.0503475666046143,
+      "learning_rate": 4.412594402405233e-07,
+      "loss": 0.0102,
+      "step": 2184
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 1.1703752279281616,
+      "learning_rate": 4.393326369074212e-07,
+      "loss": 0.0071,
+      "step": 2185
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.0950582027435303,
+      "learning_rate": 4.374097171514906e-07,
+      "loss": 0.0074,
+      "step": 2186
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.7434263229370117,
+      "learning_rate": 4.354906838890833e-07,
+      "loss": 0.0051,
+      "step": 2187
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.509973406791687,
+      "learning_rate": 4.3357554003065594e-07,
+      "loss": 0.0095,
+      "step": 2188
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 2.005493402481079,
+      "learning_rate": 4.316642884807687e-07,
+      "loss": 0.0083,
+      "step": 2189
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.301384687423706,
+      "learning_rate": 4.2975693213807653e-07,
+      "loss": 0.0106,
+      "step": 2190
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 2.0791378021240234,
+      "learning_rate": 4.278534738953276e-07,
+      "loss": 0.0107,
+      "step": 2191
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 1.8680495023727417,
+      "learning_rate": 4.259539166393579e-07,
+      "loss": 0.0115,
+      "step": 2192
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.9698794484138489,
+      "learning_rate": 4.24058263251087e-07,
+      "loss": 0.0112,
+      "step": 2193
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.8297860622406006,
+      "learning_rate": 4.2216651660551377e-07,
+      "loss": 0.0055,
+      "step": 2194
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 1.1352063417434692,
+      "learning_rate": 4.202786795717135e-07,
+      "loss": 0.0056,
+      "step": 2195
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 1.0710866451263428,
+      "learning_rate": 4.1839475501282966e-07,
+      "loss": 0.0062,
+      "step": 2196
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 1.1035598516464233,
+      "learning_rate": 4.1651474578607296e-07,
+      "loss": 0.0055,
+      "step": 2197
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.9676154255867004,
+      "learning_rate": 4.14638654742716e-07,
+      "loss": 0.0088,
+      "step": 2198
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 1.0264902114868164,
+      "learning_rate": 4.1276648472808974e-07,
+      "loss": 0.0059,
+      "step": 2199
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 1.2106053829193115,
+      "learning_rate": 4.108982385815776e-07,
+      "loss": 0.0075,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "eval_blimp_filtered_avg": 0.731044776119403,
+      "eval_blimp_filtered_std": 0.004869803338183953,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.01737392472820325,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 1.1924365758895874,
+      "learning_rate": 4.0903391913661205e-07,
+      "loss": 0.0095,
+      "step": 2201
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 1.4727685451507568,
+      "learning_rate": 4.0717352922066966e-07,
+      "loss": 0.0081,
+      "step": 2202
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.970632791519165,
+      "learning_rate": 4.053170716552681e-07,
+      "loss": 0.0065,
+      "step": 2203
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.5361541509628296,
+      "learning_rate": 4.034645492559609e-07,
+      "loss": 0.0035,
+      "step": 2204
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 1.0572460889816284,
+      "learning_rate": 4.016159648323342e-07,
+      "loss": 0.0096,
+      "step": 2205
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.46283969283103943,
+      "learning_rate": 3.997713211880007e-07,
+      "loss": 0.0044,
+      "step": 2206
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.9670405983924866,
+      "learning_rate": 3.979306211205963e-07,
+      "loss": 0.0081,
+      "step": 2207
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.3277919590473175,
+      "learning_rate": 3.9609386742177576e-07,
+      "loss": 0.0019,
+      "step": 2208
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 1.2476757764816284,
+      "learning_rate": 3.942610628772112e-07,
+      "loss": 0.0045,
+      "step": 2209
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.9918078780174255,
+      "learning_rate": 3.9243221026658207e-07,
+      "loss": 0.0083,
+      "step": 2210
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.9351747035980225,
+      "learning_rate": 3.9060731236357493e-07,
+      "loss": 0.0084,
+      "step": 2211
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.699523389339447,
+      "learning_rate": 3.88786371935881e-07,
+      "loss": 0.0048,
+      "step": 2212
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 1.984757423400879,
+      "learning_rate": 3.8696939174518633e-07,
+      "loss": 0.0112,
+      "step": 2213
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 1.2210612297058105,
+      "learning_rate": 3.8515637454717266e-07,
+      "loss": 0.0095,
+      "step": 2214
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 1.842618465423584,
+      "learning_rate": 3.8334732309151024e-07,
+      "loss": 0.0146,
+      "step": 2215
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 1.3000807762145996,
+      "learning_rate": 3.815422401218556e-07,
+      "loss": 0.0099,
+      "step": 2216
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 1.0679538249969482,
+      "learning_rate": 3.797411283758455e-07,
+      "loss": 0.0085,
+      "step": 2217
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 1.1001348495483398,
+      "learning_rate": 3.7794399058509566e-07,
+      "loss": 0.0061,
+      "step": 2218
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.7503768801689148,
+      "learning_rate": 3.761508294751937e-07,
+      "loss": 0.0058,
+      "step": 2219
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 1.377563714981079,
+      "learning_rate": 3.7436164776569536e-07,
+      "loss": 0.0084,
+      "step": 2220
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 1.3701099157333374,
+      "learning_rate": 3.725764481701219e-07,
+      "loss": 0.0081,
+      "step": 2221
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.7862921953201294,
+      "learning_rate": 3.70795233395956e-07,
+      "loss": 0.0058,
+      "step": 2222
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.997042179107666,
+      "learning_rate": 3.6901800614463535e-07,
+      "loss": 0.0069,
+      "step": 2223
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.7635900378227234,
+      "learning_rate": 3.6724476911155116e-07,
+      "loss": 0.0047,
+      "step": 2224
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 1.7808359861373901,
+      "learning_rate": 3.654755249860419e-07,
+      "loss": 0.0097,
+      "step": 2225
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.785383403301239,
+      "learning_rate": 3.6371027645139144e-07,
+      "loss": 0.0054,
+      "step": 2226
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.8276168704032898,
+      "learning_rate": 3.6194902618482264e-07,
+      "loss": 0.0077,
+      "step": 2227
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.3346703052520752,
+      "learning_rate": 3.6019177685749606e-07,
+      "loss": 0.0111,
+      "step": 2228
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.4893268048763275,
+      "learning_rate": 3.584385311345029e-07,
+      "loss": 0.0029,
+      "step": 2229
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.533899188041687,
+      "learning_rate": 3.566892916748628e-07,
+      "loss": 0.0119,
+      "step": 2230
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.143508791923523,
+      "learning_rate": 3.549440611315199e-07,
+      "loss": 0.0112,
+      "step": 2231
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.1292240619659424,
+      "learning_rate": 3.532028421513377e-07,
+      "loss": 0.0078,
+      "step": 2232
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.8483757376670837,
+      "learning_rate": 3.5146563737509574e-07,
+      "loss": 0.0051,
+      "step": 2233
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.9757391810417175,
+      "learning_rate": 3.497324494374855e-07,
+      "loss": 0.0073,
+      "step": 2234
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 1.562127947807312,
+      "learning_rate": 3.48003280967107e-07,
+      "loss": 0.0128,
+      "step": 2235
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 1.7601343393325806,
+      "learning_rate": 3.4627813458646376e-07,
+      "loss": 0.0108,
+      "step": 2236
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 1.4311801195144653,
+      "learning_rate": 3.4455701291195884e-07,
+      "loss": 0.0151,
+      "step": 2237
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 1.0324448347091675,
+      "learning_rate": 3.428399185538923e-07,
+      "loss": 0.0053,
+      "step": 2238
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 1.254957675933838,
+      "learning_rate": 3.411268541164553e-07,
+      "loss": 0.008,
+      "step": 2239
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 1.2495008707046509,
+      "learning_rate": 3.3941782219772723e-07,
+      "loss": 0.0093,
+      "step": 2240
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 1.225306510925293,
+      "learning_rate": 3.3771282538967306e-07,
+      "loss": 0.0076,
+      "step": 2241
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.5283024907112122,
+      "learning_rate": 3.360118662781364e-07,
+      "loss": 0.0046,
+      "step": 2242
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.6233196258544922,
+      "learning_rate": 3.3431494744283664e-07,
+      "loss": 0.0029,
+      "step": 2243
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.750397801399231,
+      "learning_rate": 3.326220714573688e-07,
+      "loss": 0.0079,
+      "step": 2244
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 1.2939488887786865,
+      "learning_rate": 3.3093324088919275e-07,
+      "loss": 0.0066,
+      "step": 2245
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 2.1973257064819336,
+      "learning_rate": 3.2924845829963465e-07,
+      "loss": 0.0149,
+      "step": 2246
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 1.1178309917449951,
+      "learning_rate": 3.2756772624388167e-07,
+      "loss": 0.0061,
+      "step": 2247
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 1.8645424842834473,
+      "learning_rate": 3.2589104727097696e-07,
+      "loss": 0.0123,
+      "step": 2248
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 1.2148367166519165,
+      "learning_rate": 3.242184239238172e-07,
+      "loss": 0.0117,
+      "step": 2249
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 2.0877652168273926,
+      "learning_rate": 3.225498587391474e-07,
+      "loss": 0.0123,
+      "step": 2250
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.7729255557060242,
+      "learning_rate": 3.208853542475598e-07,
+      "loss": 0.0035,
+      "step": 2251
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 1.5989127159118652,
+      "learning_rate": 3.19224912973486e-07,
+      "loss": 0.0083,
+      "step": 2252
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 1.6901705265045166,
+      "learning_rate": 3.175685374351959e-07,
+      "loss": 0.009,
+      "step": 2253
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.9556978344917297,
+      "learning_rate": 3.1591623014479363e-07,
+      "loss": 0.0072,
+      "step": 2254
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 1.3902548551559448,
+      "learning_rate": 3.142679936082131e-07,
+      "loss": 0.0092,
+      "step": 2255
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 1.4000489711761475,
+      "learning_rate": 3.1262383032521436e-07,
+      "loss": 0.0111,
+      "step": 2256
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.8928489685058594,
+      "learning_rate": 3.109837427893787e-07,
+      "loss": 0.0063,
+      "step": 2257
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.7924060225486755,
+      "learning_rate": 3.093477334881084e-07,
+      "loss": 0.0059,
+      "step": 2258
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.708108127117157,
+      "learning_rate": 3.077158049026187e-07,
+      "loss": 0.0089,
+      "step": 2259
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.9139454364776611,
+      "learning_rate": 3.060879595079369e-07,
+      "loss": 0.007,
+      "step": 2260
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.9102306365966797,
+      "learning_rate": 3.0446419977289665e-07,
+      "loss": 0.0068,
+      "step": 2261
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.8086994290351868,
+      "learning_rate": 3.0284452816013607e-07,
+      "loss": 0.0046,
+      "step": 2262
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.9092122316360474,
+      "learning_rate": 3.012289471260923e-07,
+      "loss": 0.0071,
+      "step": 2263
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.6211750507354736,
+      "learning_rate": 2.996174591209999e-07,
+      "loss": 0.0035,
+      "step": 2264
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.6814675331115723,
+      "learning_rate": 2.9801006658888466e-07,
+      "loss": 0.0043,
+      "step": 2265
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.8002520203590393,
+      "learning_rate": 2.9640677196756106e-07,
+      "loss": 0.0061,
+      "step": 2266
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.8536789417266846,
+      "learning_rate": 2.9480757768862977e-07,
+      "loss": 0.0065,
+      "step": 2267
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 1.4087966680526733,
+      "learning_rate": 2.9321248617747165e-07,
+      "loss": 0.0074,
+      "step": 2268
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 1.4193106889724731,
+      "learning_rate": 2.9162149985324524e-07,
+      "loss": 0.016,
+      "step": 2269
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.094619631767273,
+      "learning_rate": 2.9003462112888347e-07,
+      "loss": 0.0062,
+      "step": 2270
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.3441765308380127,
+      "learning_rate": 2.884518524110894e-07,
+      "loss": 0.0092,
+      "step": 2271
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.7214688062667847,
+      "learning_rate": 2.8687319610033293e-07,
+      "loss": 0.0099,
+      "step": 2272
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.439801812171936,
+      "learning_rate": 2.852986545908458e-07,
+      "loss": 0.0072,
+      "step": 2273
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.8635445833206177,
+      "learning_rate": 2.837282302706212e-07,
+      "loss": 0.0058,
+      "step": 2274
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.872896671295166,
+      "learning_rate": 2.8216192552140693e-07,
+      "loss": 0.0126,
+      "step": 2275
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.6414180397987366,
+      "learning_rate": 2.8059974271870245e-07,
+      "loss": 0.0045,
+      "step": 2276
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.3696050643920898,
+      "learning_rate": 2.790416842317569e-07,
+      "loss": 0.0092,
+      "step": 2277
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.7142107486724854,
+      "learning_rate": 2.774877524235637e-07,
+      "loss": 0.0032,
+      "step": 2278
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.1741163730621338,
+      "learning_rate": 2.759379496508567e-07,
+      "loss": 0.0066,
+      "step": 2279
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.0115845203399658,
+      "learning_rate": 2.743922782641104e-07,
+      "loss": 0.0048,
+      "step": 2280
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.7260040640830994,
+      "learning_rate": 2.7285074060753057e-07,
+      "loss": 0.0033,
+      "step": 2281
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.9344484210014343,
+      "learning_rate": 2.7131333901905507e-07,
+      "loss": 0.0059,
+      "step": 2282
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 1.1766636371612549,
+      "learning_rate": 2.697800758303485e-07,
+      "loss": 0.0083,
+      "step": 2283
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.7113633751869202,
+      "learning_rate": 2.6825095336679894e-07,
+      "loss": 0.006,
+      "step": 2284
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.328793466091156,
+      "learning_rate": 2.667259739475151e-07,
+      "loss": 0.0027,
+      "step": 2285
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.9964905381202698,
+      "learning_rate": 2.6520513988532114e-07,
+      "loss": 0.0084,
+      "step": 2286
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 1.539788842201233,
+      "learning_rate": 2.6368845348675643e-07,
+      "loss": 0.0074,
+      "step": 2287
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 1.1505931615829468,
+      "learning_rate": 2.6217591705206753e-07,
+      "loss": 0.007,
+      "step": 2288
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.6105601787567139,
+      "learning_rate": 2.606675328752073e-07,
+      "loss": 0.0058,
+      "step": 2289
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 1.0669714212417603,
+      "learning_rate": 2.5916330324383364e-07,
+      "loss": 0.006,
+      "step": 2290
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 1.2686563730239868,
+      "learning_rate": 2.576632304393004e-07,
+      "loss": 0.0101,
+      "step": 2291
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 1.1767021417617798,
+      "learning_rate": 2.5616731673665905e-07,
+      "loss": 0.004,
+      "step": 2292
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 1.2006715536117554,
+      "learning_rate": 2.5467556440465247e-07,
+      "loss": 0.0061,
+      "step": 2293
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.6601417660713196,
+      "learning_rate": 2.531879757057125e-07,
+      "loss": 0.0046,
+      "step": 2294
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 1.281562328338623,
+      "learning_rate": 2.5170455289595583e-07,
+      "loss": 0.0044,
+      "step": 2295
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.9906952977180481,
+      "learning_rate": 2.5022529822518183e-07,
+      "loss": 0.0078,
+      "step": 2296
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.7911376953125,
+      "learning_rate": 2.487502139368685e-07,
+      "loss": 0.007,
+      "step": 2297
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.6763374209403992,
+      "learning_rate": 2.4727930226816765e-07,
+      "loss": 0.0053,
+      "step": 2298
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 1.1383639574050903,
+      "learning_rate": 2.458125654499038e-07,
+      "loss": 0.0048,
+      "step": 2299
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.6809481978416443,
+      "learning_rate": 2.4435000570656996e-07,
+      "loss": 0.0071,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "eval_blimp_filtered_avg": 0.7308955223880597,
+      "eval_blimp_filtered_std": 0.004874491049318892,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "eval_blimp_supplement_avg": 0.8017241379310345,
+      "eval_blimp_supplement_std": 0.017566343838219044,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "eval_vqa_filtered_avg": 0.29,
+      "eval_vqa_filtered_std": 0.045604802157206845,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.8581156730651855,
+      "learning_rate": 2.428916252563237e-07,
+      "loss": 0.0054,
+      "step": 2301
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 1.0434715747833252,
+      "learning_rate": 2.4143742631098307e-07,
+      "loss": 0.0061,
+      "step": 2302
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 1.1453216075897217,
+      "learning_rate": 2.39987411076027e-07,
+      "loss": 0.0083,
+      "step": 2303
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.7413375973701477,
+      "learning_rate": 2.3854158175058685e-07,
+      "loss": 0.0048,
+      "step": 2304
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 1.0638315677642822,
+      "learning_rate": 2.3709994052744598e-07,
+      "loss": 0.0051,
+      "step": 2305
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 1.4309489727020264,
+      "learning_rate": 2.3566248959303716e-07,
+      "loss": 0.0064,
+      "step": 2306
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 2.5271048545837402,
+      "learning_rate": 2.3422923112743676e-07,
+      "loss": 0.0138,
+      "step": 2307
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 1.4661903381347656,
+      "learning_rate": 2.3280016730436339e-07,
+      "loss": 0.0055,
+      "step": 2308
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.9549493789672852,
+      "learning_rate": 2.3137530029117292e-07,
+      "loss": 0.0055,
+      "step": 2309
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 1.0982089042663574,
+      "learning_rate": 2.2995463224885848e-07,
+      "loss": 0.0067,
+      "step": 2310
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.8267478942871094,
+      "learning_rate": 2.2853816533204287e-07,
+      "loss": 0.0045,
+      "step": 2311
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.2912914752960205,
+      "learning_rate": 2.2712590168897763e-07,
+      "loss": 0.006,
+      "step": 2312
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.0430282354354858,
+      "learning_rate": 2.2571784346154044e-07,
+      "loss": 0.0039,
+      "step": 2313
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.6607396602630615,
+      "learning_rate": 2.2431399278523078e-07,
+      "loss": 0.011,
+      "step": 2314
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.1184636354446411,
+      "learning_rate": 2.2291435178916552e-07,
+      "loss": 0.0052,
+      "step": 2315
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.4285485744476318,
+      "learning_rate": 2.215189225960783e-07,
+      "loss": 0.0045,
+      "step": 2316
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.1582567691802979,
+      "learning_rate": 2.201277073223149e-07,
+      "loss": 0.0085,
+      "step": 2317
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 1.3505151271820068,
+      "learning_rate": 2.1874070807782953e-07,
+      "loss": 0.0115,
+      "step": 2318
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 1.5787919759750366,
+      "learning_rate": 2.1735792696618216e-07,
+      "loss": 0.0125,
+      "step": 2319
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 1.3569775819778442,
+      "learning_rate": 2.1597936608453728e-07,
+      "loss": 0.011,
+      "step": 2320
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.6764390468597412,
+      "learning_rate": 2.1460502752365673e-07,
+      "loss": 0.0042,
+      "step": 2321
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.7819216847419739,
+      "learning_rate": 2.132349133678989e-07,
+      "loss": 0.0077,
+      "step": 2322
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.9990053772926331,
+      "learning_rate": 2.1186902569521694e-07,
+      "loss": 0.0065,
+      "step": 2323
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.7323740720748901,
+      "learning_rate": 2.1050736657715242e-07,
+      "loss": 0.0045,
+      "step": 2324
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 1.799077033996582,
+      "learning_rate": 2.0914993807883376e-07,
+      "loss": 0.0091,
+      "step": 2325
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 1.1923933029174805,
+      "learning_rate": 2.0779674225897448e-07,
+      "loss": 0.0112,
+      "step": 2326
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 1.588623285293579,
+      "learning_rate": 2.0644778116986795e-07,
+      "loss": 0.0139,
+      "step": 2327
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 1.2804484367370605,
+      "learning_rate": 2.0510305685738395e-07,
+      "loss": 0.0099,
+      "step": 2328
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 1.8879567384719849,
+      "learning_rate": 2.0376257136096844e-07,
+      "loss": 0.0075,
+      "step": 2329
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 1.1125404834747314,
+      "learning_rate": 2.0242632671363747e-07,
+      "loss": 0.0077,
+      "step": 2330
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 1.5998907089233398,
+      "learning_rate": 2.01094324941976e-07,
+      "loss": 0.0087,
+      "step": 2331
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.5560637712478638,
+      "learning_rate": 1.9976656806613335e-07,
+      "loss": 0.0035,
+      "step": 2332
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.5213460922241211,
+      "learning_rate": 1.9844305809982177e-07,
+      "loss": 0.0027,
+      "step": 2333
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 1.1130348443984985,
+      "learning_rate": 1.9712379705031224e-07,
+      "loss": 0.0132,
+      "step": 2334
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.9435350894927979,
+      "learning_rate": 1.9580878691843128e-07,
+      "loss": 0.0082,
+      "step": 2335
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 1.478018045425415,
+      "learning_rate": 1.9449802969855924e-07,
+      "loss": 0.0095,
+      "step": 2336
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.9114845991134644,
+      "learning_rate": 1.9319152737862556e-07,
+      "loss": 0.0081,
+      "step": 2337
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 1.839444875717163,
+      "learning_rate": 1.918892819401069e-07,
+      "loss": 0.0118,
+      "step": 2338
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 1.042659044265747,
+      "learning_rate": 1.9059129535802378e-07,
+      "loss": 0.0069,
+      "step": 2339
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 1.6109708547592163,
+      "learning_rate": 1.8929756960093748e-07,
+      "loss": 0.0103,
+      "step": 2340
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.8770699501037598,
+      "learning_rate": 1.8800810663094747e-07,
+      "loss": 0.0066,
+      "step": 2341
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 1.475842833518982,
+      "learning_rate": 1.8672290840368845e-07,
+      "loss": 0.0103,
+      "step": 2342
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.39688166975975037,
+      "learning_rate": 1.8544197686832653e-07,
+      "loss": 0.0038,
+      "step": 2343
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 1.118986964225769,
+      "learning_rate": 1.8416531396755677e-07,
+      "loss": 0.0077,
+      "step": 2344
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 1.1198034286499023,
+      "learning_rate": 1.828929216376003e-07,
+      "loss": 0.0052,
+      "step": 2345
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.855065107345581,
+      "learning_rate": 1.8162480180820218e-07,
+      "loss": 0.0057,
+      "step": 2346
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.6895895600318909,
+      "learning_rate": 1.8036095640262727e-07,
+      "loss": 0.0054,
+      "step": 2347
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 1.8183060884475708,
+      "learning_rate": 1.791013873376567e-07,
+      "loss": 0.0084,
+      "step": 2348
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 1.3529072999954224,
+      "learning_rate": 1.778460965235883e-07,
+      "loss": 0.0083,
+      "step": 2349
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.7072827816009521,
+      "learning_rate": 1.7659508586422855e-07,
+      "loss": 0.0072,
+      "step": 2350
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.8957288265228271,
+      "learning_rate": 1.7534835725689513e-07,
+      "loss": 0.0034,
+      "step": 2351
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.5701649785041809,
+      "learning_rate": 1.7410591259240916e-07,
+      "loss": 0.0085,
+      "step": 2352
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.6708860993385315,
+      "learning_rate": 1.7286775375509667e-07,
+      "loss": 0.0089,
+      "step": 2353
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.2356315851211548,
+      "learning_rate": 1.7163388262278157e-07,
+      "loss": 0.0096,
+      "step": 2354
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 2.284837245941162,
+      "learning_rate": 1.7040430106678728e-07,
+      "loss": 0.011,
+      "step": 2355
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.0990779399871826,
+      "learning_rate": 1.691790109519301e-07,
+      "loss": 0.0061,
+      "step": 2356
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.6486763954162598,
+      "learning_rate": 1.6795801413651757e-07,
+      "loss": 0.0109,
+      "step": 2357
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.4316133260726929,
+      "learning_rate": 1.6674131247234635e-07,
+      "loss": 0.0152,
+      "step": 2358
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.6377578973770142,
+      "learning_rate": 1.6552890780469976e-07,
+      "loss": 0.0111,
+      "step": 2359
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 1.0370407104492188,
+      "learning_rate": 1.643208019723429e-07,
+      "loss": 0.0065,
+      "step": 2360
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 1.0653284788131714,
+      "learning_rate": 1.631169968075218e-07,
+      "loss": 0.0062,
+      "step": 2361
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.7859501242637634,
+      "learning_rate": 1.619174941359597e-07,
+      "loss": 0.0065,
+      "step": 2362
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.41982877254486084,
+      "learning_rate": 1.6072229577685472e-07,
+      "loss": 0.0027,
+      "step": 2363
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 1.1950647830963135,
+      "learning_rate": 1.5953140354287655e-07,
+      "loss": 0.0067,
+      "step": 2364
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 1.3048546314239502,
+      "learning_rate": 1.5834481924016543e-07,
+      "loss": 0.0083,
+      "step": 2365
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 1.1106764078140259,
+      "learning_rate": 1.5716254466832612e-07,
+      "loss": 0.0067,
+      "step": 2366
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 1.1692255735397339,
+      "learning_rate": 1.5598458162042873e-07,
+      "loss": 0.0063,
+      "step": 2367
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 2.009127140045166,
+      "learning_rate": 1.5481093188300343e-07,
+      "loss": 0.0147,
+      "step": 2368
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.6906253695487976,
+      "learning_rate": 1.5364159723603943e-07,
+      "loss": 0.0045,
+      "step": 2369
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 1.1262880563735962,
+      "learning_rate": 1.5247657945298076e-07,
+      "loss": 0.0096,
+      "step": 2370
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 1.0773130655288696,
+      "learning_rate": 1.5131588030072475e-07,
+      "loss": 0.0094,
+      "step": 2371
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 1.099623680114746,
+      "learning_rate": 1.501595015396199e-07,
+      "loss": 0.0049,
+      "step": 2372
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 1.3845664262771606,
+      "learning_rate": 1.490074449234603e-07,
+      "loss": 0.0083,
+      "step": 2373
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 1.2987943887710571,
+      "learning_rate": 1.4785971219948702e-07,
+      "loss": 0.0066,
+      "step": 2374
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.9373829960823059,
+      "learning_rate": 1.4671630510838196e-07,
+      "loss": 0.0063,
+      "step": 2375
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 1.3361494541168213,
+      "learning_rate": 1.4557722538426698e-07,
+      "loss": 0.0082,
+      "step": 2376
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.7322679162025452,
+      "learning_rate": 1.4444247475470118e-07,
+      "loss": 0.0063,
+      "step": 2377
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.8004128932952881,
+      "learning_rate": 1.4331205494067857e-07,
+      "loss": 0.0092,
+      "step": 2378
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 1.1367136240005493,
+      "learning_rate": 1.4218596765662374e-07,
+      "loss": 0.0071,
+      "step": 2379
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 1.342947006225586,
+      "learning_rate": 1.4106421461039122e-07,
+      "loss": 0.009,
+      "step": 2380
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 1.3259834051132202,
+      "learning_rate": 1.3994679750326146e-07,
+      "loss": 0.0092,
+      "step": 2381
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.8977048397064209,
+      "learning_rate": 1.3883371802994015e-07,
+      "loss": 0.0038,
+      "step": 2382
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 1.1481903791427612,
+      "learning_rate": 1.377249778785533e-07,
+      "loss": 0.0071,
+      "step": 2383
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.9288477301597595,
+      "learning_rate": 1.3662057873064614e-07,
+      "loss": 0.0069,
+      "step": 2384
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.6409772634506226,
+      "learning_rate": 1.3552052226118017e-07,
+      "loss": 0.0029,
+      "step": 2385
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.7437184453010559,
+      "learning_rate": 1.3442481013853048e-07,
+      "loss": 0.0055,
+      "step": 2386
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.47614043951034546,
+      "learning_rate": 1.3333344402448377e-07,
+      "loss": 0.0037,
+      "step": 2387
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 2.537379264831543,
+      "learning_rate": 1.3224642557423538e-07,
+      "loss": 0.014,
+      "step": 2388
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.6708997488021851,
+      "learning_rate": 1.3116375643638689e-07,
+      "loss": 0.0032,
+      "step": 2389
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 1.2695043087005615,
+      "learning_rate": 1.3008543825294317e-07,
+      "loss": 0.012,
+      "step": 2390
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 1.4075605869293213,
+      "learning_rate": 1.2901147265931103e-07,
+      "loss": 0.0089,
+      "step": 2391
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.9872757196426392,
+      "learning_rate": 1.2794186128429586e-07,
+      "loss": 0.0107,
+      "step": 2392
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.42113324999809265,
+      "learning_rate": 1.2687660575009842e-07,
+      "loss": 0.0035,
+      "step": 2393
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 1.5577365159988403,
+      "learning_rate": 1.25815707672314e-07,
+      "loss": 0.0092,
+      "step": 2394
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 1.060302495956421,
+      "learning_rate": 1.247591686599302e-07,
+      "loss": 0.0094,
+      "step": 2395
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 1.8661613464355469,
+      "learning_rate": 1.2370699031532162e-07,
+      "loss": 0.0083,
+      "step": 2396
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.31512585282325745,
+      "learning_rate": 1.2265917423425078e-07,
+      "loss": 0.0019,
+      "step": 2397
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 2.011923313140869,
+      "learning_rate": 1.2161572200586347e-07,
+      "loss": 0.0142,
+      "step": 2398
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 1.081145167350769,
+      "learning_rate": 1.205766352126878e-07,
+      "loss": 0.0077,
+      "step": 2399
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 1.5997120141983032,
+      "learning_rate": 1.195419154306302e-07,
+      "loss": 0.0101,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "eval_blimp_filtered_avg": 0.7308955223880597,
+      "eval_blimp_filtered_std": 0.0048762448950767255,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "eval_blimp_supplement_avg": 0.7995689655172413,
+      "eval_blimp_supplement_std": 0.01758637940432669,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "eval_vqa_filtered_avg": 0.29,
+      "eval_vqa_filtered_std": 0.045604802157206845,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 1.310752034187317,
+      "learning_rate": 1.1851156422897569e-07,
+      "loss": 0.0094,
+      "step": 2401
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.9254988431930542,
+      "learning_rate": 1.1748558317038161e-07,
+      "loss": 0.0091,
+      "step": 2402
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 1.086929202079773,
+      "learning_rate": 1.1646397381087892e-07,
+      "loss": 0.0073,
+      "step": 2403
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.6503797173500061,
+      "learning_rate": 1.1544673769986824e-07,
+      "loss": 0.004,
+      "step": 2404
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 2.2718751430511475,
+      "learning_rate": 1.1443387638011683e-07,
+      "loss": 0.0052,
+      "step": 2405
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 1.4930957555770874,
+      "learning_rate": 1.1342539138775787e-07,
+      "loss": 0.0114,
+      "step": 2406
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.6761605143547058,
+      "learning_rate": 1.1242128425228659e-07,
+      "loss": 0.0072,
+      "step": 2407
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 1.3330060243606567,
+      "learning_rate": 1.1142155649655883e-07,
+      "loss": 0.0078,
+      "step": 2408
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.8317005634307861,
+      "learning_rate": 1.1042620963678906e-07,
+      "loss": 0.0062,
+      "step": 2409
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 1.2150142192840576,
+      "learning_rate": 1.0943524518254644e-07,
+      "loss": 0.0054,
+      "step": 2410
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 1.1283634901046753,
+      "learning_rate": 1.084486646367554e-07,
+      "loss": 0.0059,
+      "step": 2411
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.6332213282585144,
+      "learning_rate": 1.0746646949569006e-07,
+      "loss": 0.0041,
+      "step": 2412
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.4226652681827545,
+      "learning_rate": 1.0648866124897383e-07,
+      "loss": 0.0026,
+      "step": 2413
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 1.2892417907714844,
+      "learning_rate": 1.0551524137957746e-07,
+      "loss": 0.0071,
+      "step": 2414
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 1.109286904335022,
+      "learning_rate": 1.045462113638157e-07,
+      "loss": 0.0071,
+      "step": 2415
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.8964385986328125,
+      "learning_rate": 1.035815726713456e-07,
+      "loss": 0.0084,
+      "step": 2416
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 1.623815655708313,
+      "learning_rate": 1.026213267651639e-07,
+      "loss": 0.0146,
+      "step": 2417
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.864382266998291,
+      "learning_rate": 1.0166547510160595e-07,
+      "loss": 0.0061,
+      "step": 2418
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 1.8746938705444336,
+      "learning_rate": 1.0071401913034184e-07,
+      "loss": 0.0142,
+      "step": 2419
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.5448727607727051,
+      "learning_rate": 9.97669602943756e-08,
+      "loss": 0.0036,
+      "step": 2420
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.5503562092781067,
+      "learning_rate": 9.882430003004161e-08,
+      "loss": 0.006,
+      "step": 2421
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 1.5032259225845337,
+      "learning_rate": 9.788603976700462e-08,
+      "loss": 0.0115,
+      "step": 2422
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 1.0028787851333618,
+      "learning_rate": 9.695218092825464e-08,
+      "loss": 0.0077,
+      "step": 2423
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 1.4789237976074219,
+      "learning_rate": 9.602272493010777e-08,
+      "loss": 0.0117,
+      "step": 2424
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.8586037755012512,
+      "learning_rate": 9.509767318220208e-08,
+      "loss": 0.0057,
+      "step": 2425
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.7459132075309753,
+      "learning_rate": 9.417702708749532e-08,
+      "loss": 0.0062,
+      "step": 2426
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 1.2474572658538818,
+      "learning_rate": 9.32607880422649e-08,
+      "loss": 0.0076,
+      "step": 2427
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 1.5835881233215332,
+      "learning_rate": 9.234895743610361e-08,
+      "loss": 0.0087,
+      "step": 2428
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 1.2471505403518677,
+      "learning_rate": 9.14415366519179e-08,
+      "loss": 0.0103,
+      "step": 2429
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.8866732716560364,
+      "learning_rate": 9.053852706592724e-08,
+      "loss": 0.0042,
+      "step": 2430
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.9714205861091614,
+      "learning_rate": 8.963993004765947e-08,
+      "loss": 0.006,
+      "step": 2431
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 1.3096208572387695,
+      "learning_rate": 8.874574695995174e-08,
+      "loss": 0.0094,
+      "step": 2432
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 1.3077837228775024,
+      "learning_rate": 8.78559791589456e-08,
+      "loss": 0.0046,
+      "step": 2433
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 1.2823196649551392,
+      "learning_rate": 8.697062799408761e-08,
+      "loss": 0.0048,
+      "step": 2434
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.7847549915313721,
+      "learning_rate": 8.608969480812467e-08,
+      "loss": 0.0071,
+      "step": 2435
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 1.5399121046066284,
+      "learning_rate": 8.521318093710373e-08,
+      "loss": 0.0129,
+      "step": 2436
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.6710610389709473,
+      "learning_rate": 8.434108771037041e-08,
+      "loss": 0.004,
+      "step": 2437
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 1.0376766920089722,
+      "learning_rate": 8.347341645056373e-08,
+      "loss": 0.0082,
+      "step": 2438
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.743603527545929,
+      "learning_rate": 8.261016847361736e-08,
+      "loss": 0.0077,
+      "step": 2439
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.7403782606124878,
+      "learning_rate": 8.175134508875703e-08,
+      "loss": 0.0066,
+      "step": 2440
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.7941461205482483,
+      "learning_rate": 8.08969475984972e-08,
+      "loss": 0.0086,
+      "step": 2441
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 1.6419627666473389,
+      "learning_rate": 8.004697729863997e-08,
+      "loss": 0.0095,
+      "step": 2442
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 1.453057050704956,
+      "learning_rate": 7.920143547827353e-08,
+      "loss": 0.0109,
+      "step": 2443
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.534957766532898,
+      "learning_rate": 7.836032341976907e-08,
+      "loss": 0.0026,
+      "step": 2444
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.4371516704559326,
+      "learning_rate": 7.752364239878018e-08,
+      "loss": 0.0027,
+      "step": 2445
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 1.3402639627456665,
+      "learning_rate": 7.669139368423916e-08,
+      "loss": 0.0051,
+      "step": 2446
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.8942230343818665,
+      "learning_rate": 7.586357853835768e-08,
+      "loss": 0.0056,
+      "step": 2447
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 1.3338133096694946,
+      "learning_rate": 7.504019821662212e-08,
+      "loss": 0.0054,
+      "step": 2448
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 1.4859192371368408,
+      "learning_rate": 7.422125396779322e-08,
+      "loss": 0.0099,
+      "step": 2449
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.7895346283912659,
+      "learning_rate": 7.340674703390415e-08,
+      "loss": 0.0037,
+      "step": 2450
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 1.1835581064224243,
+      "learning_rate": 7.259667865025777e-08,
+      "loss": 0.0089,
+      "step": 2451
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.6588142514228821,
+      "learning_rate": 7.17910500454263e-08,
+      "loss": 0.0039,
+      "step": 2452
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 2.1774704456329346,
+      "learning_rate": 7.098986244124706e-08,
+      "loss": 0.0079,
+      "step": 2453
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.5687236785888672,
+      "learning_rate": 7.019311705282305e-08,
+      "loss": 0.0046,
+      "step": 2454
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 1.4591701030731201,
+      "learning_rate": 6.94008150885197e-08,
+      "loss": 0.0068,
+      "step": 2455
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.7248699069023132,
+      "learning_rate": 6.861295774996346e-08,
+      "loss": 0.0041,
+      "step": 2456
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.7135469317436218,
+      "learning_rate": 6.782954623204052e-08,
+      "loss": 0.0044,
+      "step": 2457
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.7809768319129944,
+      "learning_rate": 6.705058172289346e-08,
+      "loss": 0.0065,
+      "step": 2458
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 2.0189013481140137,
+      "learning_rate": 6.627606540392095e-08,
+      "loss": 0.0066,
+      "step": 2459
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.3775126338005066,
+      "learning_rate": 6.550599844977534e-08,
+      "loss": 0.0017,
+      "step": 2460
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 2.530108690261841,
+      "learning_rate": 6.474038202836108e-08,
+      "loss": 0.0168,
+      "step": 2461
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 1.098971962928772,
+      "learning_rate": 6.397921730083234e-08,
+      "loss": 0.0036,
+      "step": 2462
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.8090375065803528,
+      "learning_rate": 6.322250542159269e-08,
+      "loss": 0.0049,
+      "step": 2463
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 4.484795570373535,
+      "learning_rate": 6.247024753829112e-08,
+      "loss": 0.0184,
+      "step": 2464
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 1.5359996557235718,
+      "learning_rate": 6.172244479182232e-08,
+      "loss": 0.01,
+      "step": 2465
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 1.8459322452545166,
+      "learning_rate": 6.097909831632475e-08,
+      "loss": 0.0135,
+      "step": 2466
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.9357349872589111,
+      "learning_rate": 6.024020923917695e-08,
+      "loss": 0.0052,
+      "step": 2467
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 1.2014546394348145,
+      "learning_rate": 5.950577868099816e-08,
+      "loss": 0.0083,
+      "step": 2468
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.5856483578681946,
+      "learning_rate": 5.8775807755645396e-08,
+      "loss": 0.003,
+      "step": 2469
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 1.2968889474868774,
+      "learning_rate": 5.8050297570213077e-08,
+      "loss": 0.0069,
+      "step": 2470
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 1.0443476438522339,
+      "learning_rate": 5.732924922502869e-08,
+      "loss": 0.0067,
+      "step": 2471
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 1.5003111362457275,
+      "learning_rate": 5.661266381365382e-08,
+      "loss": 0.0088,
+      "step": 2472
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 1.0909587144851685,
+      "learning_rate": 5.590054242288145e-08,
+      "loss": 0.006,
+      "step": 2473
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 1.198867678642273,
+      "learning_rate": 5.519288613273399e-08,
+      "loss": 0.0065,
+      "step": 2474
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 1.1948719024658203,
+      "learning_rate": 5.4489696016462253e-08,
+      "loss": 0.0088,
+      "step": 2475
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 1.1651867628097534,
+      "learning_rate": 5.379097314054249e-08,
+      "loss": 0.0051,
+      "step": 2476
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 1.2601349353790283,
+      "learning_rate": 5.309671856467768e-08,
+      "loss": 0.009,
+      "step": 2477
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 1.1437535285949707,
+      "learning_rate": 5.240693334179259e-08,
+      "loss": 0.0068,
+      "step": 2478
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 1.4795550107955933,
+      "learning_rate": 5.17216185180337e-08,
+      "loss": 0.0069,
+      "step": 2479
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 1.6751185655593872,
+      "learning_rate": 5.104077513276828e-08,
+      "loss": 0.0143,
+      "step": 2480
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.8270640969276428,
+      "learning_rate": 5.036440421858235e-08,
+      "loss": 0.0057,
+      "step": 2481
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.8076775074005127,
+      "learning_rate": 4.9692506801277017e-08,
+      "loss": 0.0064,
+      "step": 2482
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 1.4476237297058105,
+      "learning_rate": 4.902508389987148e-08,
+      "loss": 0.0046,
+      "step": 2483
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 1.4485738277435303,
+      "learning_rate": 4.8362136526596377e-08,
+      "loss": 0.0079,
+      "step": 2484
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 1.004165530204773,
+      "learning_rate": 4.7703665686895437e-08,
+      "loss": 0.0065,
+      "step": 2485
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 1.5622228384017944,
+      "learning_rate": 4.704967237942381e-08,
+      "loss": 0.0104,
+      "step": 2486
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 1.4013727903366089,
+      "learning_rate": 4.640015759604543e-08,
+      "loss": 0.007,
+      "step": 2487
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 1.5682252645492554,
+      "learning_rate": 4.575512232183166e-08,
+      "loss": 0.0083,
+      "step": 2488
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 1.3642469644546509,
+      "learning_rate": 4.511456753506027e-08,
+      "loss": 0.0138,
+      "step": 2489
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.8058323264122009,
+      "learning_rate": 4.447849420721417e-08,
+      "loss": 0.0057,
+      "step": 2490
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 1.4838799238204956,
+      "learning_rate": 4.384690330297936e-08,
+      "loss": 0.0108,
+      "step": 2491
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 1.16139817237854,
+      "learning_rate": 4.3219795780242934e-08,
+      "loss": 0.0029,
+      "step": 2492
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 1.0274087190628052,
+      "learning_rate": 4.25971725900941e-08,
+      "loss": 0.0107,
+      "step": 2493
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 1.5862236022949219,
+      "learning_rate": 4.197903467681985e-08,
+      "loss": 0.0119,
+      "step": 2494
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 1.2274856567382812,
+      "learning_rate": 4.1365382977903934e-08,
+      "loss": 0.0076,
+      "step": 2495
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 1.0131663084030151,
+      "learning_rate": 4.0756218424027547e-08,
+      "loss": 0.0075,
+      "step": 2496
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 1.1914234161376953,
+      "learning_rate": 4.0151541939066674e-08,
+      "loss": 0.0076,
+      "step": 2497
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 1.8618018627166748,
+      "learning_rate": 3.955135444008906e-08,
+      "loss": 0.0109,
+      "step": 2498
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 1.3499330282211304,
+      "learning_rate": 3.895565683735591e-08,
+      "loss": 0.0118,
+      "step": 2499
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 1.207308053970337,
+      "learning_rate": 3.836445003431788e-08,
+      "loss": 0.0059,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "eval_blimp_filtered_avg": 0.7308955223880597,
+      "eval_blimp_filtered_std": 0.004872690393320008,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "eval_blimp_supplement_avg": 0.8017241379310345,
+      "eval_blimp_supplement_std": 0.01749645799500491,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "eval_vqa_filtered_avg": 0.31,
+      "eval_vqa_filtered_std": 0.04648231987117316,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.9853691458702087,
+      "learning_rate": 3.7777734927615047e-08,
+      "loss": 0.0056,
+      "step": 2501
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.7459747791290283,
+      "learning_rate": 3.719551240707597e-08,
+      "loss": 0.0061,
+      "step": 2502
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 2.394340753555298,
+      "learning_rate": 3.661778335571464e-08,
+      "loss": 0.0085,
+      "step": 2503
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 1.5546541213989258,
+      "learning_rate": 3.6044548649730166e-08,
+      "loss": 0.0062,
+      "step": 2504
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.9458232522010803,
+      "learning_rate": 3.5475809158506124e-08,
+      "loss": 0.0066,
+      "step": 2505
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 1.3263143301010132,
+      "learning_rate": 3.491156574460852e-08,
+      "loss": 0.0073,
+      "step": 2506
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 1.2544054985046387,
+      "learning_rate": 3.4351819263783504e-08,
+      "loss": 0.0069,
+      "step": 2507
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.9352834224700928,
+      "learning_rate": 3.3796570564957996e-08,
+      "loss": 0.0051,
+      "step": 2508
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 1.5647450685501099,
+      "learning_rate": 3.3245820490237055e-08,
+      "loss": 0.0056,
+      "step": 2509
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.8107264637947083,
+      "learning_rate": 3.269956987490319e-08,
+      "loss": 0.0076,
+      "step": 2510
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 1.1367651224136353,
+      "learning_rate": 3.2157819547414704e-08,
+      "loss": 0.0058,
+      "step": 2511
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 1.1264668703079224,
+      "learning_rate": 3.1620570329404685e-08,
+      "loss": 0.0136,
+      "step": 2512
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 1.1125972270965576,
+      "learning_rate": 3.108782303568003e-08,
+      "loss": 0.0069,
+      "step": 2513
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.9868554472923279,
+      "learning_rate": 3.055957847421975e-08,
+      "loss": 0.0131,
+      "step": 2514
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 1.375131607055664,
+      "learning_rate": 3.0035837446173e-08,
+      "loss": 0.0086,
+      "step": 2515
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 1.8216625452041626,
+      "learning_rate": 2.9516600745860712e-08,
+      "loss": 0.0067,
+      "step": 2516
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 1.279541254043579,
+      "learning_rate": 2.90018691607703e-08,
+      "loss": 0.0082,
+      "step": 2517
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.6871073246002197,
+      "learning_rate": 2.8491643471557638e-08,
+      "loss": 0.0058,
+      "step": 2518
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 1.1138793230056763,
+      "learning_rate": 2.798592445204573e-08,
+      "loss": 0.0056,
+      "step": 2519
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 2.3589553833007812,
+      "learning_rate": 2.748471286922072e-08,
+      "loss": 0.0135,
+      "step": 2520
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 1.7307255268096924,
+      "learning_rate": 2.6988009483233877e-08,
+      "loss": 0.0087,
+      "step": 2521
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.8302087783813477,
+      "learning_rate": 2.6495815047398953e-08,
+      "loss": 0.0063,
+      "step": 2522
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 1.2758510112762451,
+      "learning_rate": 2.6008130308191157e-08,
+      "loss": 0.0082,
+      "step": 2523
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 2.7510476112365723,
+      "learning_rate": 2.5524956005246513e-08,
+      "loss": 0.0105,
+      "step": 2524
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 4.035161018371582,
+      "learning_rate": 2.5046292871360175e-08,
+      "loss": 0.0293,
+      "step": 2525
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.8683897256851196,
+      "learning_rate": 2.4572141632485777e-08,
+      "loss": 0.0051,
+      "step": 2526
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 1.320032000541687,
+      "learning_rate": 2.4102503007733422e-08,
+      "loss": 0.0079,
+      "step": 2527
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 1.105722188949585,
+      "learning_rate": 2.363737770936969e-08,
+      "loss": 0.0073,
+      "step": 2528
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.5613688230514526,
+      "learning_rate": 2.3176766442816633e-08,
+      "loss": 0.0047,
+      "step": 2529
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.8370179533958435,
+      "learning_rate": 2.2720669906649782e-08,
+      "loss": 0.0096,
+      "step": 2530
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 2.692255973815918,
+      "learning_rate": 2.2269088792596483e-08,
+      "loss": 0.0155,
+      "step": 2531
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.6379058361053467,
+      "learning_rate": 2.182202378553788e-08,
+      "loss": 0.0063,
+      "step": 2532
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 1.728951334953308,
+      "learning_rate": 2.137947556350428e-08,
+      "loss": 0.0148,
+      "step": 2533
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 1.2368959188461304,
+      "learning_rate": 2.0941444797676458e-08,
+      "loss": 0.0056,
+      "step": 2534
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 1.0770119428634644,
+      "learning_rate": 2.0507932152383668e-08,
+      "loss": 0.0059,
+      "step": 2535
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.8048055171966553,
+      "learning_rate": 2.007893828510299e-08,
+      "loss": 0.009,
+      "step": 2536
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.8457075953483582,
+      "learning_rate": 1.9654463846457637e-08,
+      "loss": 0.0055,
+      "step": 2537
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.9206335544586182,
+      "learning_rate": 1.923450948021732e-08,
+      "loss": 0.0067,
+      "step": 2538
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 1.1211744546890259,
+      "learning_rate": 1.8819075823295893e-08,
+      "loss": 0.0077,
+      "step": 2539
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.805101215839386,
+      "learning_rate": 1.8408163505751697e-08,
+      "loss": 0.005,
+      "step": 2540
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 1.109699010848999,
+      "learning_rate": 1.8001773150785217e-08,
+      "loss": 0.006,
+      "step": 2541
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.8636054992675781,
+      "learning_rate": 1.75999053747391e-08,
+      "loss": 0.0056,
+      "step": 2542
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 1.4183051586151123,
+      "learning_rate": 1.7202560787096476e-08,
+      "loss": 0.0134,
+      "step": 2543
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.7765252590179443,
+      "learning_rate": 1.6809739990481298e-08,
+      "loss": 0.0049,
+      "step": 2544
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 1.2009655237197876,
+      "learning_rate": 1.6421443580656005e-08,
+      "loss": 0.0078,
+      "step": 2545
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.6363665461540222,
+      "learning_rate": 1.6037672146521522e-08,
+      "loss": 0.0058,
+      "step": 2546
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.5663934946060181,
+      "learning_rate": 1.56584262701156e-08,
+      "loss": 0.0031,
+      "step": 2547
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 1.1623133420944214,
+      "learning_rate": 1.528370652661315e-08,
+      "loss": 0.0062,
+      "step": 2548
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 1.207197666168213,
+      "learning_rate": 1.4913513484323904e-08,
+      "loss": 0.008,
+      "step": 2549
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 1.1079109907150269,
+      "learning_rate": 1.4547847704693085e-08,
+      "loss": 0.0048,
+      "step": 2550
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 1.2629890441894531,
+      "learning_rate": 1.4186709742298743e-08,
+      "loss": 0.0079,
+      "step": 2551
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 1.597174048423767,
+      "learning_rate": 1.3830100144853085e-08,
+      "loss": 0.0083,
+      "step": 2552
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 1.205565094947815,
+      "learning_rate": 1.3478019453199153e-08,
+      "loss": 0.0057,
+      "step": 2553
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 1.043047547340393,
+      "learning_rate": 1.313046820131214e-08,
+      "loss": 0.0061,
+      "step": 2554
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 1.7523130178451538,
+      "learning_rate": 1.2787446916298407e-08,
+      "loss": 0.0173,
+      "step": 2555
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.8838502764701843,
+      "learning_rate": 1.2448956118392141e-08,
+      "loss": 0.0047,
+      "step": 2556
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 1.4478026628494263,
+      "learning_rate": 1.2114996320958694e-08,
+      "loss": 0.0182,
+      "step": 2557
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 1.4791135787963867,
+      "learning_rate": 1.178556803048958e-08,
+      "loss": 0.0127,
+      "step": 2558
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 2.129828453063965,
+      "learning_rate": 1.1460671746605144e-08,
+      "loss": 0.009,
+      "step": 2559
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 1.5449845790863037,
+      "learning_rate": 1.1140307962051565e-08,
+      "loss": 0.0107,
+      "step": 2560
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 1.4811952114105225,
+      "learning_rate": 1.0824477162701852e-08,
+      "loss": 0.0116,
+      "step": 2561
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 1.3483707904815674,
+      "learning_rate": 1.0513179827552844e-08,
+      "loss": 0.0101,
+      "step": 2562
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 1.19635009765625,
+      "learning_rate": 1.0206416428726551e-08,
+      "loss": 0.0073,
+      "step": 2563
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.7338892221450806,
+      "learning_rate": 9.90418743146848e-09,
+      "loss": 0.005,
+      "step": 2564
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 1.745922565460205,
+      "learning_rate": 9.606493294147977e-09,
+      "loss": 0.0111,
+      "step": 2565
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 1.6444079875946045,
+      "learning_rate": 9.31333446825522e-09,
+      "loss": 0.0094,
+      "step": 2566
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.8486001491546631,
+      "learning_rate": 9.024711398403218e-09,
+      "loss": 0.0036,
+      "step": 2567
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 1.0704160928726196,
+      "learning_rate": 8.74062452232549e-09,
+      "loss": 0.0076,
+      "step": 2568
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 1.3725554943084717,
+      "learning_rate": 8.461074270875724e-09,
+      "loss": 0.0099,
+      "step": 2569
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.913497805595398,
+      "learning_rate": 8.186061068027439e-09,
+      "loss": 0.0048,
+      "step": 2570
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.7770643830299377,
+      "learning_rate": 7.915585330872999e-09,
+      "loss": 0.0043,
+      "step": 2571
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 1.229243278503418,
+      "learning_rate": 7.649647469623267e-09,
+      "loss": 0.0098,
+      "step": 2572
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 1.3132737874984741,
+      "learning_rate": 7.388247887606614e-09,
+      "loss": 0.0061,
+      "step": 2573
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.7429608106613159,
+      "learning_rate": 7.131386981268917e-09,
+      "loss": 0.0031,
+      "step": 2574
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 1.8152470588684082,
+      "learning_rate": 6.879065140172225e-09,
+      "loss": 0.0122,
+      "step": 2575
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.7731820940971375,
+      "learning_rate": 6.63128274699476e-09,
+      "loss": 0.0056,
+      "step": 2576
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 1.9885917901992798,
+      "learning_rate": 6.388040177529919e-09,
+      "loss": 0.0128,
+      "step": 2577
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.524071991443634,
+      "learning_rate": 6.149337800685939e-09,
+      "loss": 0.0033,
+      "step": 2578
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.8258724808692932,
+      "learning_rate": 5.915175978485232e-09,
+      "loss": 0.006,
+      "step": 2579
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 1.7036328315734863,
+      "learning_rate": 5.685555066064052e-09,
+      "loss": 0.012,
+      "step": 2580
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 1.7967863082885742,
+      "learning_rate": 5.460475411671495e-09,
+      "loss": 0.0134,
+      "step": 2581
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 1.0922369956970215,
+      "learning_rate": 5.239937356669499e-09,
+      "loss": 0.009,
+      "step": 2582
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 1.0909658670425415,
+      "learning_rate": 5.023941235532181e-09,
+      "loss": 0.0093,
+      "step": 2583
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.7803958654403687,
+      "learning_rate": 4.812487375844832e-09,
+      "loss": 0.0069,
+      "step": 2584
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.7198560237884521,
+      "learning_rate": 4.605576098304254e-09,
+      "loss": 0.0035,
+      "step": 2585
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.6589062809944153,
+      "learning_rate": 4.4032077167174275e-09,
+      "loss": 0.004,
+      "step": 2586
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.884629487991333,
+      "learning_rate": 4.2053825380021784e-09,
+      "loss": 0.0098,
+      "step": 2587
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.6752747893333435,
+      "learning_rate": 4.012100862185175e-09,
+      "loss": 0.0046,
+      "step": 2588
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 1.9068794250488281,
+      "learning_rate": 3.823362982402601e-09,
+      "loss": 0.0076,
+      "step": 2589
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 1.1848429441452026,
+      "learning_rate": 3.639169184899482e-09,
+      "loss": 0.0059,
+      "step": 2590
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 1.8662199974060059,
+      "learning_rate": 3.4595197490290277e-09,
+      "loss": 0.0134,
+      "step": 2591
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 1.1020363569259644,
+      "learning_rate": 3.28441494725229e-09,
+      "loss": 0.0067,
+      "step": 2592
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.6803072094917297,
+      "learning_rate": 3.1138550451385027e-09,
+      "loss": 0.005,
+      "step": 2593
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 1.6054714918136597,
+      "learning_rate": 2.9478403013627475e-09,
+      "loss": 0.0083,
+      "step": 2594
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.6631161570549011,
+      "learning_rate": 2.7863709677079517e-09,
+      "loss": 0.0049,
+      "step": 2595
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 1.0527620315551758,
+      "learning_rate": 2.6294472890625587e-09,
+      "loss": 0.0108,
+      "step": 2596
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 1.2381750345230103,
+      "learning_rate": 2.477069503421192e-09,
+      "loss": 0.0082,
+      "step": 2597
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.9821559190750122,
+      "learning_rate": 2.329237841884657e-09,
+      "loss": 0.0071,
+      "step": 2598
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 1.0178099870681763,
+      "learning_rate": 2.185952528657609e-09,
+      "loss": 0.0061,
+      "step": 2599
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.6051089763641357,
+      "learning_rate": 2.047213781050883e-09,
+      "loss": 0.0035,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "eval_blimp_filtered_avg": 0.7304477611940299,
+      "eval_blimp_filtered_std": 0.004876729399130613,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "eval_blimp_supplement_avg": 0.8038793103448276,
+      "eval_blimp_supplement_std": 0.01737392472820325,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "eval_vqa_filtered_avg": 0.3,
+      "eval_vqa_filtered_std": 0.046056618647183814,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.9752629399299622,
+      "learning_rate": 1.9130218094798313e-09,
+      "loss": 0.0074,
+      "step": 2601
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 1.0484426021575928,
+      "learning_rate": 1.7833768174626563e-09,
+      "loss": 0.0035,
+      "step": 2602
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 1.1803113222122192,
+      "learning_rate": 1.6582790016234084e-09,
+      "loss": 0.0071,
+      "step": 2603
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 1.2522472143173218,
+      "learning_rate": 1.5377285516879892e-09,
+      "loss": 0.0085,
+      "step": 2604
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.6305513978004456,
+      "learning_rate": 1.421725650486816e-09,
+      "loss": 0.0063,
+      "step": 2605
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 1.4788143634796143,
+      "learning_rate": 1.3102704739531568e-09,
+      "loss": 0.0098,
+      "step": 2606
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 1.4118164777755737,
+      "learning_rate": 1.2033631911227972e-09,
+      "loss": 0.0076,
+      "step": 2607
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 1.5848385095596313,
+      "learning_rate": 1.1010039641343729e-09,
+      "loss": 0.0101,
+      "step": 2608
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 1.2394896745681763,
+      "learning_rate": 1.0031929482280377e-09,
+      "loss": 0.0064,
+      "step": 2609
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.9956507086753845,
+      "learning_rate": 9.099302917474628e-10,
+      "loss": 0.006,
+      "step": 2610
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.9133124947547913,
+      "learning_rate": 8.212161361368376e-10,
+      "loss": 0.006,
+      "step": 2611
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 2.3552138805389404,
+      "learning_rate": 7.370506159425361e-10,
+      "loss": 0.0129,
+      "step": 2612
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.8443816900253296,
+      "learning_rate": 6.574338588121176e-10,
+      "loss": 0.0057,
+      "step": 2613
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 1.3868720531463623,
+      "learning_rate": 5.823659854946595e-10,
+      "loss": 0.0152,
+      "step": 2614
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 1.4647369384765625,
+      "learning_rate": 5.118471098400912e-10,
+      "loss": 0.0079,
+      "step": 2615
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 1.2423732280731201,
+      "learning_rate": 4.4587733879919437e-10,
+      "loss": 0.0106,
+      "step": 2616
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 1.8461129665374756,
+      "learning_rate": 3.8445677242360256e-10,
+      "loss": 0.0106,
+      "step": 2617
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 1.1438802480697632,
+      "learning_rate": 3.275855038654685e-10,
+      "loss": 0.0117,
+      "step": 2618
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.7271893620491028,
+      "learning_rate": 2.7526361937679766e-10,
+      "loss": 0.005,
+      "step": 2619
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.9953367114067078,
+      "learning_rate": 2.2749119831111386e-10,
+      "loss": 0.0062,
+      "step": 2620
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.7768956422805786,
+      "learning_rate": 1.842683131207945e-10,
+      "loss": 0.0098,
+      "step": 2621
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.5906195044517517,
+      "learning_rate": 1.4559502935906909e-10,
+      "loss": 0.0034,
+      "step": 2622
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 1.2652437686920166,
+      "learning_rate": 1.1147140567868697e-10,
+      "loss": 0.0154,
+      "step": 2623
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.5899912118911743,
+      "learning_rate": 8.189749383291645e-11,
+      "loss": 0.0032,
+      "step": 2624
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 1.08551824092865,
+      "learning_rate": 5.687333867387956e-11,
+      "loss": 0.0087,
+      "step": 2625
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 1.398841381072998,
+      "learning_rate": 3.6398978153884264e-11,
+      "loss": 0.0067,
+      "step": 2626
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 1.4685105085372925,
+      "learning_rate": 2.0474443325424473e-11,
+      "loss": 0.0097,
+      "step": 2627
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.327163577079773,
+      "learning_rate": 9.099758339514708e-12,
+      "loss": 0.0083,
+      "step": 2628
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.5812302827835083,
+      "learning_rate": 2.2749404473554336e-12,
+      "loss": 0.0045,
+      "step": 2629
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.947285532951355,
+      "learning_rate": 0.0,
+      "loss": 0.0068,
+      "step": 2630
+    },
+    {
+      "epoch": 5.0,
+      "step": 2630,
+      "total_flos": 6.633565184222822e+17,
+      "train_loss": 0.06158972647445659,
+      "train_runtime": 6393.1333,
+      "train_samples_per_second": 52.619,
+      "train_steps_per_second": 0.411
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2630,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "total_flos": 6.633565184222822e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}