diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 6.3447867128256075,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0422,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.439654362976193,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0532,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.477007818284127,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0593,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.232817777454256,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0336,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.946554724139001,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0379,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.799196077524012,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0318,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.509936063081163,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.0113,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.140134430912968,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9956,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.784628253312094,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9895,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.532229315530329,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9777,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.516108519981893,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9727,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.818826561712318,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9435,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.135416226290209,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9287,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.043147382521033,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9388,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.7243291938719003,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9118,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.6805522458603757,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.888,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.4155376637929664,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8767,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.049702265795266,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8627,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.5216443989670203,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8569,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.0528422916856042,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8486,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.0568459851855145,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8391,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.199519755796892,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8278,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 0.9894788616894945,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.8021,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9485980117510749,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7939,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.967165894948629,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7963,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.8575001710264698,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.789,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.8907129407353669,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7916,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.8924298818065751,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7818,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.8510229177481562,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7791,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.8047241212510425,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7663,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7612825823404162,
+      "learning_rate": 1e-05,
+      "loss": 0.7599,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7993534340409865,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7518,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.787055684760684,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7466,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.8067977659962494,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7485,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6678650883916137,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.731,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.6979631984781679,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7364,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6187032947008843,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7364,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6212732951294923,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7581,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.7378023396899993,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7368,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.5320036179789444,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7165,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.701626796288855,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7326,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.7618917032698544,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7377,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5531689100118112,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7246,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.6323643127656188,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7279,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.6441947207752302,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7213,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5960887903201538,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.7179,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6246195897908293,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7048,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5262943719364639,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.7105,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.6771554802212756,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.7201,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.7299771719889979,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.7223,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6132447732653235,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6907,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5338775310606483,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.7112,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.7200474446935462,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.7115,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5899173307055836,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.7157,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6499362470255478,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.7079,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6275771930954436,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.694,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6787811054453996,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6949,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.4884253265837821,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6923,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.579136553380181,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.6871,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5963320827030353,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.694,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.6964662086339933,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6813,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.7075746909683307,
+      "learning_rate": 2e-05,
+      "loss": 0.6891,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.5730638871834602,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.7065,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6393922725763991,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6905,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.8580238873974992,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.7015,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 1.165772555505928,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6986,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 1.0393980664609377,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6831,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.7712021825796748,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6829,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.6461674560050142,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6947,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.9410301604130327,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6853,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 1.20707392348855,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6761,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 1.2084846006844252,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6896,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.8776821406627359,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6831,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 1.1870402911921232,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6802,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.2871233430299975,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6847,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.6505428407126033,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6732,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.9006801569496308,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.678,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.913286923980069,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6775,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.6374350869325311,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6771,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.8415222052682735,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6901,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 2.0410951910692354,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6655,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.3665874803926905,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6882,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 2.0896832133072447,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6892,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.9304525946167428,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6806,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.472886577747304,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6804,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.8195059669637372,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6719,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.1994236291241185,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6916,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 1.6457594326613658,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6788,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.048203014379773,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6791,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 2.0270080695262918,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6781,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.5982857490542295,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6802,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.8309641731103132,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6686,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.4508472082035073,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6641,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.5931127825888312,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6582,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.534185882976404,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6779,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.2276012330740855,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6803,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.5060326380783569,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6862,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.1143279389259821,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6599,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.5483401312134613,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6596,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.0150354250991827,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6696,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.2010162856565598,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6712,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.796467353179619,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6618,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.1509132396183175,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6635,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 2.3461573188373666,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6701,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.8920601986756629,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.68,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 2.19920698733967,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6723,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 2.1610902819996705,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6673,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.6833938558414785,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6598,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.562350875477834,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6881,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.4754740033679723,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6713,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.3072723193787898,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6597,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.715635271107145,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6641,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.3436483686837126,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6624,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.6299991458986127,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6814,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.4045215737594747,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6713,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 2.2027415688015175,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6755,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.6509148226550678,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6578,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 2.2740654702112444,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6429,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.7883431314936624,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6769,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.5638006855897784,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6595,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.4510804327051765,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6613,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.6118563780766548,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6633,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.631739304841748,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6612,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.4046611691828992,
+      "learning_rate": 4e-05,
+      "loss": 0.6601,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.3937469043882416,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6584,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.7795026922153292,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6526,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.8805672912725078,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6556,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.8399965875459114,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6594,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.2228691287107798,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.667,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 2.266721459537049,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6669,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.7608955401662982,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6534,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.8398606336274823,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6612,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.952911582412621,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6596,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.1898519316962413,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6647,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.7715582950880537,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.6539,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.4236736810410857,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6622,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.3779394863369623,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6603,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 1.115874897620166,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.663,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 1.952272158797266,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6632,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 1.3154743630662533,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6458,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 1.0016642134576585,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6605,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 1.646190806491737,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.659,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 1.126197280323483,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6534,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 1.3410188550795068,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6553,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 1.7785546309828615,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.651,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 1.3140732355935478,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6381,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 1.0064797210454623,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6551,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 1.5615221180714294,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6613,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 1.5463861148725349,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.656,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 1.1396432247413433,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6684,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 1.677001258260287,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6589,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 1.4805446583181046,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6544,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 1.3593654783206655,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6497,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 1.3102264200349476,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6475,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 1.384343328462875,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6567,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 1.85933862851565,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6474,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 1.1536395931252161,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6616,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 1.9727269967202583,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6527,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 1.4390884669185007,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6544,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 1.6401867436335007,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6437,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 1.403856822034845,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6441,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 1.2327531245938181,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6549,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 1.048742250548721,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6419,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 1.5582719986571003,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6392,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 1.053967565033768,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6383,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 1.8813090109194368,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.652,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 1.388006957306901,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6423,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 1.8146765718323339,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6478,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 1.436001788869168,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6562,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 1.4864655199250352,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6403,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 1.613775375918424,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6502,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 1.041185446539446,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.635,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 1.1996259403442844,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6465,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 1.1783307302311967,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6421,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 1.315395033220733,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6481,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 1.598450225104334,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6479,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.9126880078538218,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.6593,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 1.3363271573288649,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6484,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 1.9012154094464644,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6505,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 1.0748368918685243,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6574,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 1.8749510370674656,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6494,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 1.4458123689870916,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6268,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 1.8322477680854437,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6469,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 1.2679424614556574,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.6499,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 1.8874152534662174,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6475,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 1.1790497374528133,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6429,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 2.046444974663716,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.6522,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 1.589154075733338,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6383,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 1.674937648477168,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6411,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 1.5778503531557173,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6513,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 1.6127947562966893,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6479,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 1.2229299765747335,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6547,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 1.5477354466523527,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6437,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 1.1784382608813548,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.6353,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 1.0830521556598798,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6263,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 1.228302656777323,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6327,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 1.3057147802824736,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6401,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 1.482078141729799,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.6377,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.8033425779139995,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.6341,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 1.3944691053439606,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6283,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 2.6852413049226853,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6397,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.9811234289027273,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.642,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 1.4041862850851916,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6495,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 1.717710777613339,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6364,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6691146646039374,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.6376,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 1.4882845552749497,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6322,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 1.547017280046924,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6345,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 1.1736858663098875,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6223,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.9805974114639442,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6375,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 1.5372996202796547,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6428,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 1.489472832576732,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6425,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.7287503289100691,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6369,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 1.2404546247325332,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6331,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 1.5436133930612776,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6396,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.8706915213597608,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6182,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.8021371878325223,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6255,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 1.0469063184990284,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6298,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 1.6510971593533346,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.6379,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.8295214833804474,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.6283,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 1.3294592920261095,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.6276,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 1.4807763475815783,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6291,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.8592691318808567,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6285,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 1.3779009231015777,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.6321,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 1.196828478168595,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.6196,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 1.3078427264235912,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6232,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 1.3158808186679833,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6211,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 1.0187673942995952,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6415,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 1.608062294546663,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6499,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 1.0808447657818303,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.6334,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 1.1754018646210762,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.631,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 1.4241485817825614,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.6446,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 1.119454162725145,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.624,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 1.3925808153483956,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6337,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.9415249427773501,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.6197,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 2.0313957689628475,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.6223,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 1.3649805734971192,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.6244,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 1.794881807965697,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.643,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 1.7360773908098737,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.635,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 1.1750940010926882,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.6319,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 1.3582406740538595,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6176,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 1.0296967277923943,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.6154,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.9242598325764001,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.6213,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 1.2200843771880387,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.6242,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 1.0717424981354544,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.628,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 1.0599125746873794,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.6336,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 1.2573214379185482,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.622,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1887078523802308,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6226,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 1.097140731523833,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.599,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.7714521624293681,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5899,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 1.1610157502610974,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5908,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 1.1151909472128356,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5815,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 1.1398449892537668,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5865,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.7511919619400171,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5877,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.7022176269750608,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5811,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.8381508647074305,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5865,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.8551947097562298,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5917,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 30.03715770361503,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.6484,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 2.3067959434708905,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.6134,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 2.0739985375846297,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5993,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 1.603813853260374,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5957,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 1.4083152682028066,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5848,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 1.2289171656400342,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5937,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 1.7650102388338806,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5864,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 1.2581206394615199,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5887,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 2.2636228175118354,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.6003,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 2.0970818699773788,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.6167,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 1.5052807510617823,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5939,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 1.5093205971566526,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5952,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 1.6221374210953494,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5936,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 1.3425048795261114,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.599,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 1.5337108795520327,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5919,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 1.3582455730393523,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.6052,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 1.38276456453041,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5851,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 1.2155343647024819,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.586,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 1.221002948869727,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5761,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 1.2657522785556383,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5705,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 1.0862251684115174,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5755,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 1.4038614339825026,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5887,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 1.010917444222867,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5823,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 1.4324677372629981,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5758,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.9326701933519457,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5787,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 1.4918277276442156,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5918,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 1.0465804320624257,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5923,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 1.7775582645135515,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5898,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 1.6163544474849862,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5906,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 1.316274457287899,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5856,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 1.2668108891330028,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5867,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 1.1337386367722493,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5907,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 1.1620710208613885,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5905,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 1.218710342938741,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5997,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.9653335831659968,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5714,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 1.2278373149804647,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5881,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.929995846201122,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5728,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 1.2299534480781409,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5889,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.87638781894861,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5728,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 1.0507874718230534,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5891,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 1.0567955071803738,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5808,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.7090472176762397,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5717,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.8722139208484548,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.6021,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.7736968793295702,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5949,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.8344144033690646,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5792,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.7831257818259786,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5898,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 1.005691350211793,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5876,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.8057728286096866,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5928,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.7277629654868789,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5822,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.7365604558886565,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5874,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.8201102328492048,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5744,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 1.1594406375800466,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5754,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.6257555256632278,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5863,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.7498284687107053,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.578,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6822015562311042,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5853,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.7787940847673933,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5991,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.6664094260452307,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5754,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.8876909575941285,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5747,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.9503728036423639,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5815,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 1.1935510662601023,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.586,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.6545853979434895,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5923,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.722207198604051,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5745,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 1.1787405762552152,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5919,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.9983245122398122,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5869,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.7895921619410664,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.574,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 1.0447317664330589,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5855,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.6893667194575682,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5672,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.9243105147509393,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5886,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 1.315373915656225,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5817,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.5217890592862913,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5748,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 1.0120661461345262,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5861,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 1.1738452179647179,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5787,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.7200980242592514,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5834,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.9977250372720496,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5853,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 1.1360016543475802,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5867,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5933743020857358,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5803,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.9882712645066826,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5849,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 1.0441160149851862,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5941,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.860102469761664,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.577,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 1.1852179439575705,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5878,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.7130338702216505,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5853,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.7213110324440192,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5886,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.8634693672674271,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5698,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.8946454301304606,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5855,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.7670892073264993,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5853,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.808364531308017,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5984,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.9937694778216899,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5988,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.7126910378199309,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5828,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.7583317842387667,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5938,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.764714444819354,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5802,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.6559356412699964,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5897,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.6660782205658407,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5864,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.6252376990870971,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.59,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.6076805964032446,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5885,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 3.0806030792913877,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5976,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 1.0257438749844077,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.582,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 1.358768256548057,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5927,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.9594731658048932,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5802,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 1.425777703925013,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5961,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 1.172523040673255,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5797,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 1.2269054115161975,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5801,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 1.2835519505512416,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5937,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 1.1087005330457111,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5809,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 1.134899389311504,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5799,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 1.11558967097291,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5783,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 1.0731769363291594,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5789,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.7506193396896935,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5838,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.9531802887791891,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5903,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.8340759020607735,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5717,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.8850481035978446,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5815,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 3.8728652676806643,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.6046,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 1.8594190654920033,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5933,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.678281062699284,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5913,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 1.6185609960528493,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5791,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 1.3700110456417058,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5892,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.9830470479063641,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.583,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 1.2524343564166052,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5938,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.7062685536442775,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5857,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 1.0160229520193589,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5872,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 1.0803725220862928,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5824,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.9764624231157973,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5881,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.743620897996315,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5947,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.9152711559895897,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5806,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.7801538862034105,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5838,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.7490257494696708,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.58,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.6465140338059667,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.579,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.7328571890362952,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5851,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.8257028844118471,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5758,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.7807681940474502,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.6035,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.5298249834100283,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5987,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.7225109643005442,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5786,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.6101654766612085,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5787,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.7499822248300196,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5704,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.5030781950744067,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5602,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.7220772343584037,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5737,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.45722433393320694,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5754,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.6501265465264776,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5877,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.6685130014474889,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5956,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.9045841674523535,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5814,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.4719325677053337,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5975,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.7178993228742651,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5842,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.7010086020548963,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5893,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.525367701237517,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5789,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.5336179568482654,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5811,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.5538835393951939,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5783,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.5612548936477374,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5832,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.6107017003029162,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5821,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.5421361690421385,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5998,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.6961716766498794,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5784,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.6357536738518704,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5851,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.5481431174130532,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5676,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.6452704822242573,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5785,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.5046590857399498,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5894,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.7787085939629875,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5853,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.615738322535074,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5791,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.6098710090665205,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5959,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.6464293191663651,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5867,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.667004471388334,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5778,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.42274472763181004,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.573,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5412706207077673,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.591,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.5477310443480006,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5704,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.528489628405472,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5683,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.5840480953712247,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.595,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.4961859092881989,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5776,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.762272761604723,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5644,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.6150219998554074,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5771,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.5097300704660893,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5817,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.6772497024706594,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5707,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.6472040071566341,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5863,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5055083443858408,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5795,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.41805950540383446,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.571,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.8692206865167788,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5798,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.47724458290171484,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.57,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4609386086476279,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5723,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.5283293736398331,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.581,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.6963596465091149,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5882,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 1.5334270247794046,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5716,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.6943376836320574,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5731,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.8275415420611829,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5979,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.6804629006502328,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5719,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.6511775468467607,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5961,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4998628734576159,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5845,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5193876578217718,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5829,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.561731772882176,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5853,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.5326268926544255,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5665,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.6110895415589708,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5796,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.6801614892747327,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5894,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.8263865902051055,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5816,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.6459945672107686,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.5874,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 1.7052385252786952,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.6248,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.6719351393803251,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5823,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.9286193826773385,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5833,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.7869225450229067,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.572,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 5.6197341907178044,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5962,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 1.2163436251521411,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5937,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 1.13802099639949,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5836,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 1.2251983683657022,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.6682,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 1.086777479891643,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5879,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.8249273848644826,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5765,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.518054690324678,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5737,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.7948372286237714,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5875,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.8233466070685586,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.598,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.5551123736570134,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5641,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.5140492077692678,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.58,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.7194880880200427,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.575,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.7992140333043878,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.583,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.5994891119583597,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5777,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.766661490398812,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.576,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.7105799686380936,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5834,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.6761103118822163,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5828,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.7705304365106496,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5896,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.5856362456306747,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5815,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.5706016223945973,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5708,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.6385290100826337,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5774,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.5671195577193272,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5788,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.523154875873285,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5843,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.7161357730101257,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5829,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.840116674387191,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5777,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.6425961131160857,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5747,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.6091256738245012,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5779,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.898300896069647,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5815,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.4676392105523348,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5717,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.7088481973541988,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5842,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.6427991192687617,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5858,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.5192861816794802,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5823,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.543882684351673,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5805,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.595096204731217,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5684,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.5837840486068413,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5742,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.64201840251952,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5704,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.7279927449800089,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5747,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.8543823904124836,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5856,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.5739795420183283,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5659,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.7897497003836669,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5776,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.6551518488983435,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5631,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.6323273922630065,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5841,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.8842614757080792,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5745,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.6903819037774254,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5699,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.46571592786444793,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.5654,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.7240877672468907,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.524,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.7156963646955468,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.515,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.7500371172905458,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.5152,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.6735709927467212,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.5195,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.7945716198752965,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.5136,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.7219794056157205,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.5092,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.7224244541708376,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.5282,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.6587060420715729,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.5072,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.7940955681286936,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.5224,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.6320384826337261,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.516,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 1.5407291837703374,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.5721,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.6101071719622037,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.5086,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.6976642714421898,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.5189,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.6210254305193782,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.5141,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.6878335197277363,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.513,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.6026090423221591,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.5198,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 1.8265850719384622,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.5597,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.5274447094760762,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.5166,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.5190270028122966,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.5099,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 3.598844612157652,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.5354,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.6597829522656884,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.5042,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.6378898478427508,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.5285,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.6277755371016039,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.53,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.7623407914388239,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.5237,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.833100923647852,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.5123,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.45679642093106476,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.5177,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.772761900230102,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.5242,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 1.3305686017898533,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.5259,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.5560813444930961,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.5111,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.5129264939982549,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.5206,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.5150325598325256,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.5252,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.555939931747421,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.5232,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.5203679542503983,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.5119,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4684511370431327,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.5104,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.5495232381911188,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.5188,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.41574271530529106,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.5069,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.5498184636945983,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.5159,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.5485787835218067,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.5087,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.4745167950739232,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.5038,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.565250882371374,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.5275,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.5403984728770808,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.5146,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.4365106847534293,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.5282,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.5059557764614245,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.5294,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.42983355930118466,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.5117,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.4484419953442025,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.5274,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.6258237647919184,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.5092,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.4185085062555919,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.5225,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 1.7494725001857727,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.5406,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.49507184831512246,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.5077,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.7402674154049783,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.5186,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.3927429900250646,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.5239,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.8389066925320666,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.536,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.4719627597039572,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.5288,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.7885121882015999,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.5104,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.639092812015225,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.5247,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.5008391808893752,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.5132,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.6084731928377707,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.5079,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.6773313592215321,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.5344,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.4961654240837391,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.5249,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.407577259488453,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.5158,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.5299353818617272,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.5319,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.474835249454219,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.5275,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.54321345761091,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.5294,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3454757414016062,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.506,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.445957219337664,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.5269,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.36137169503177535,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.5106,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.39660317107959636,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.5144,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.34750096533153485,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.5159,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.39927167304841304,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.5123,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.36866946869801087,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.5194,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.40531704645528904,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.5306,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.3897515875022129,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.5102,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.3961455130899523,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.5138,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.33241338476012733,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.5124,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.402273103634971,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.5205,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.33010292985038336,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.5272,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.36015774061828854,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.5096,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.35842691621167777,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.5256,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.35046559858747867,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.5116,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.3472574384936036,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.5135,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.3668383163252556,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.5163,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.3684607482044998,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.5115,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.40738192564480824,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.5247,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.38933221644302785,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.5138,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.39409884426529435,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.5101,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.41830536304191,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.5346,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.4270582721777864,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.5212,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.4365886706011099,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.5129,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.41836530763456503,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.5042,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.49259166477736116,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.5268,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3833574199838685,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.5205,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.43300261910132765,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.5074,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3662543992521878,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.5197,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.5139827972573235,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.5154,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.4050173684003349,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.5074,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.4558744235068469,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.5114,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.3749480346736794,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.513,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.37997885873833814,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.5119,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.35534087718955526,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.504,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.8073802311789213,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.5335,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.35061558928966086,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.5062,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3689291318713997,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.5132,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3818464525278109,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.5148,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3656889231365447,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.5037,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.4305911466754329,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.5077,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3340064906930322,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.523,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 5.011096270428012,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.5608,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.5179486848184476,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.5259,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.48431823652147926,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.5183,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.44921878481325345,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.5174,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.39254167164527004,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.5125,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.3886552604822217,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.5166,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.39110132259849095,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.521,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.46037489506174634,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.5182,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3803018603401352,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.5169,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.6170654565082991,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.5392,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.386889088712913,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.5203,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.4081241126007586,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.5112,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.4668039707196572,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.5171,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.4230812570131022,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.5189,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.4274137760893998,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.5083,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.41243638679020905,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.5078,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.4245728305871517,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.5083,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.33936942020782607,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.5151,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.37194240408615986,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.5067,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.3608126291160404,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.5142,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3174600854799729,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.5061,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.3319355172314559,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.5198,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.36894814485162647,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.5214,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3899505918080332,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.5146,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.3336383242215536,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.5089,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.40991299925330843,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.5239,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.37532571217688204,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.5053,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.38810415346730304,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.5207,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3663197201906485,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.5072,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3931694513474196,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.5255,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.31960763229002687,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.5033,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3967675444537336,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.5211,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.30581639671548694,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.5153,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.42110480444372766,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.5177,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3145384495990334,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.5227,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.4514703487211874,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.5212,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3507518293668063,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.5286,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3949185441728674,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.5244,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.36238051630881374,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.514,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.40580479200264213,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.5205,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.34153063011983076,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.5093,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.37605095218970824,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.5103,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.34370943208162813,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.5298,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3592049573447851,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.5011,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3131125111734124,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.512,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.3819975033334303,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.5058,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3325046934658268,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.5062,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.3558307067636251,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.507,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3542500767799057,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.5132,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.35545922157106524,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.5111,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3638056371688518,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.5166,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3732513942738441,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.5229,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.357438047905563,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.5077,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.36315291863377075,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.5199,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.36427685343715993,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.5205,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.37342295397882286,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.5119,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3640099208906316,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.5202,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.38265506144687156,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.5095,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.3409787926541473,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.5177,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.42782559031247624,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.5241,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.5049348942746082,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.5102,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.4435261373066045,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.5111,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.3487239420166624,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.506,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.40950411389726277,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.4958,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.3819736426978709,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.5097,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3511906029821774,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.5093,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.40706784383902295,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.5228,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.3122444517410818,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.5167,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.3831260220722175,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.5154,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.353608257951315,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.5147,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3641783137202425,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.5259,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.41171592295039,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.5091,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3221721364062219,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.516,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.42571188387735875,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.5163,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3257252651197605,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.522,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.43096900961928264,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.5161,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.3261668089067638,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.5079,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.46220421141006335,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.5162,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.31844936113632666,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.5144,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.47188583517052257,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.5197,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.30640598049597934,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.5082,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.40691785826414917,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.5128,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.35939706736206506,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.514,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3653096934425528,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.5153,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3848307790528547,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.5048,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.4727011648640664,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.5167,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.3422707907096817,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.5261,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.43704810711907144,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.5102,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.3602715524488223,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.5084,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.31320842891364437,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.5105,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.39057733865375427,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.51,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.3170204638150244,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.5222,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.42958204355955315,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.5157,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3788192473537807,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.5108,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.32759617897564036,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.509,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.38859568166907077,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.5026,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.3185326442898824,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.5165,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.4412753533475448,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.5046,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.35318268450816076,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.5095,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.369766983512423,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.5186,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.402965691898166,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.5231,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.33390012184620077,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.5146,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.35923818896391163,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.5048,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.33346548423384675,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.512,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.36331921126164574,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.5202,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.33148653757008717,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.5013,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.3876583107211026,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.5229,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.29030060040835987,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.5015,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.34403979887283354,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.5103,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3424909102733998,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.5117,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.3236145356202448,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.5195,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.3138419613871304,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.5145,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.31181246712174215,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.507,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.36008269901480894,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.5155,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.32460611892542224,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.5139,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3362943446882975,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.5167,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3443571210940731,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.5217,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.277937626191764,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.5053,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.384318725383178,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.4995,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.3703376471417818,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.5129,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.37473731012074507,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.505,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.39051641690849326,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.5091,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.31772555425799187,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.5165,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.37917094147956415,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.5112,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.3529174425666584,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.5052,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3109866899308044,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4992,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.33357038380647414,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.5209,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3599697126970573,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.5244,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3396132090857071,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.5069,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.32484262250222873,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.5062,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.31165028463820876,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.5161,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.30058731518541404,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.5048,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.3353272899751361,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.5036,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.3192277087920308,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.5101,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.3352976519938811,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.5107,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.317005672481687,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.5113,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3675355973546065,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.513,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.32204101075385944,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.5159,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3355625868568321,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.5122,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.32050092393371443,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.5123,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.7768130382222824,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.5162,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.4551145290158465,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.4667,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.33655808139023,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.4418,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.5551357430235673,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.461,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.3967578746152388,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.4526,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.4427873751031868,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.4716,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4418506831648277,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.4569,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.4189864899779121,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.4592,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.39084841931684966,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.4561,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.4467638506201691,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.4597,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.44161476346248146,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.4654,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.4356823817169035,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.4586,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.45547573256517865,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.4581,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.33690027337647926,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.4494,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.37679210338448665,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.4471,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3734446722220294,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.4457,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.33914850270609903,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.4603,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.35561659808080365,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.4581,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.31771499316883717,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.4544,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.4647146690097234,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.4606,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3477616690612885,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.459,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.36663748757572334,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.4625,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.35827735185295057,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.456,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.32279681911959535,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.4462,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.34348937576798844,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.4415,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.36780468424634294,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.4523,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.34738139675023316,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.4621,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3383725862754025,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.4614,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.3870810613921846,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.4548,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3205403601902443,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.4447,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.6210204567346861,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.4658,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.36382403221056714,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.4547,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3690624241985451,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.4492,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.35733982869332304,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.4465,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.34723910045136447,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.4507,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.32666014845916336,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.4705,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.3221262027420856,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.4471,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.35108675057005834,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.4547,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.37254540740725983,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.458,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.29275306301613474,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.4501,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3625385475947517,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.456,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.33064184252149526,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.4708,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.3107435690331078,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.4547,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.2945632273650488,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.464,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.2863358681314189,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.4572,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.28723647038587,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.4606,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.2749146352082215,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.4617,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.30658527323874285,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.4614,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.28344933861582755,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.4533,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.2861990740669545,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.4541,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3108316045314733,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.4561,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.2616684395527922,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.4663,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.2914591499085374,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.4493,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.8932532979745136,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.457,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.30842220479923205,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.4477,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.2822081880375794,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.4463,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2952442477199901,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.4606,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.293655350296024,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.4643,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.28062941458734675,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.45,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.33895627501601955,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.4579,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.3053337363595896,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.4591,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2979317661822809,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.4507,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.3127001456457014,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.4546,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.3164787521092631,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.461,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.3177007217575066,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.452,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.2894530681447828,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.4565,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.3485228832299339,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.4474,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2984167890479255,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.4674,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.33157265754566706,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.4554,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.3253102851385777,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.464,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.30454720553421755,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.4444,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.34394898633458615,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.4577,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.30946247230563023,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.4544,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.2989405635809137,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.4552,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.7326863502886879,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.4997,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.2852265500877927,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.4544,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.2602683924957928,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.467,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.3005148457099474,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.466,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.2608931659400924,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.4583,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.2990179149352068,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.4583,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.24840719942364517,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.4475,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2813625451856179,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.446,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.2847211218620605,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.4552,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.3052115857200969,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.4574,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2755718248512894,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.4657,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.28820253571472076,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.4525,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.27221728693215097,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.4499,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.3550331612313213,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.4557,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.27368520028850274,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.457,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.28020443205082207,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.4569,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.2555687134359365,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.4554,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2662048855074794,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.451,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.2728096287494088,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.4626,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.27014284345753276,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.4619,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.7047810211952485,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.4631,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2728540754370837,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.4615,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2718154750164643,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.4634,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.25444348743082584,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.4633,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.28101147442687796,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.457,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.2622748229836592,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.4464,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.31622978447671485,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.4551,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.29067129426513477,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.4489,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.2954502578821633,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.4592,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.261433243623921,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.4443,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.4766047017137072,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.4697,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.27212123218224804,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.4636,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.28462768527184124,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.4542,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.3412835086159602,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.4578,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2700957340085363,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.4555,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.3153146469818201,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.4654,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.2580838343298787,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.4629,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.32504676901604646,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.4551,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.30939028433020344,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.4789,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.28525205188357183,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.4533,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.3498912538074462,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.4641,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.31373155417303134,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.4591,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.28312414151536225,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.4525,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.3434469897843956,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.4581,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.2768824311771146,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.4532,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.3020963759880393,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.4562,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.3038749660376874,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.4559,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.29769496120275585,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.4492,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.2979726665684896,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.4548,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.3426833545537637,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.4592,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.30932909718987955,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.4555,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.3623198731364283,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.4525,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.31002237080126516,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.4638,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.2861509369881975,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.4571,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.2615837784307001,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.4606,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.3076508151492591,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.4435,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.28463015711477346,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.4528,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.26468843076008786,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.4571,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.2865238561288642,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.4515,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.28611059816125195,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.4539,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.25847370076260384,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.4452,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.292493295129515,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.45,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.5263317568781951,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.4544,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.2523173626623413,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.4614,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.25603123044211157,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.465,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.27967651258871656,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.4515,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.28366881880152145,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.4547,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.23908382148054416,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.4607,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.25986736582962355,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.4641,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.3415489995513468,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.4591,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.25343665476716964,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.4579,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.27366981553036823,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.4611,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.28718043616154515,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.4665,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2752610559207511,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.4543,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.23978625021672634,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.4651,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.3629273294182671,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.4646,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2697720372418395,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.4572,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.2589668366364851,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.4569,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.2639016110671341,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.4532,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2852368471905463,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.4519,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.279839795213761,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.4519,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.2737316560836765,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.4578,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.27873810998575577,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.4513,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.27141507024899436,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.4696,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.2732828654542953,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.4515,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.26990181783773276,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.4504,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2456849854005575,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.4504,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 3.7541482518227105,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.5583,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.3207405361351643,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.4698,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.26168490314935383,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.4664,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.29326020564554056,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.4494,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.29369607335221637,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.4534,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.24080948500604293,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.4528,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.2889500292036165,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.4589,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.2801080995576619,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.4536,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.2595219315704969,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.4584,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.25008207640282104,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.4584,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 4.080789798778054,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.4819,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.2861723243107422,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.4541,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.2437235964987764,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.4632,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2681192245388665,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.4586,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.28994553300243914,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.4645,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.24877272528718586,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.4586,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.26010518595845666,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.4568,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.27929961638847917,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.4626,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.3110218420185799,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.4621,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.24215172640202617,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.4618,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.269483693459621,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.4638,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.3066059462777288,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.4593,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.26431378779805986,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.439,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.2576100576868515,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.4533,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2651388130983313,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.4759,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.2531458420762247,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.4493,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.2392328583495915,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.4542,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2676628707051077,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.4574,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.23758474656199285,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.4645,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.245016452501867,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.4548,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2617782227801827,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.4594,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2478588144536031,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.4563,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.27625130013152355,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.4572,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.24671369715881536,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.4649,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2405371958085802,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.4382,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.25635194814614665,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.4566,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.26687694612390217,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.4653,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.2663655370790154,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.4536,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2514211777995869,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.4583,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.2742783622223541,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.447,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 1.006664159506192,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.465,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.23489066181662452,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.4528,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.25548264953405114,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.4492,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.23556431262670322,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.4421,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.2621964646758373,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.4506,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.23595774238074688,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.4497,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.24964727764875874,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.4805,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2766799944945222,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.4467,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.26975218570969206,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.4557,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.24659064320208673,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.466,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2634597303098371,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.4433,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.25547397095171803,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.457,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.2373563045704075,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.4674,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.24350597967761795,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.4524,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.23808640317234364,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.4672,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.2501385209234845,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.4595,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.24456563986073257,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.4581,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.2466202990506224,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.4525,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.26417437858667536,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.4541,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2530373028563514,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.4614,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.23902737685851214,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.4644,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.261623557369397,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.4525,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.25767050510654693,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.4677,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.22872113942500308,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.4578,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.23216212775106934,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.4521,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.2656607861796452,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.4632,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.24965877089643296,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.4586,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.23273507806862925,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.4699,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.22948717317658562,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.4524,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.23659044268432317,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.4538,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.2434310987564401,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.4571,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.21929857426671623,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.4434,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.23225060252789287,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.4608,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.23434192310934254,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.4567,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.24731382351652714,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.4572,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.22075212076305625,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.4573,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.25227891055632307,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.4596,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2725629553167632,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.4608,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.22532513838456633,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.4538,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.5101696816505671,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.4873,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.24758198007444304,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.4571,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.23643022518977758,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.4608,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.22519672442770997,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.4542,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.25807973316529814,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.4672,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.21405325352610197,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.4552,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.23512933739912745,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.4644,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.22485522494390314,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.4409,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.38891001453683205,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.4197,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.2926067860556826,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.4106,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.2590109134208376,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.4073,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.36009343044467673,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.4093,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3843714507800591,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.4214,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.282051175564027,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.4228,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3220489653609086,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.4244,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.33182823205393286,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.4214,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.33654700367890156,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.4199,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.27777647365066366,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.4131,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.27784480603279993,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.4231,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.29909437522338833,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.4234,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2943737400441974,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.4149,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2594065351375315,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.4095,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.26545710708438747,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.4085,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2811207000521566,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.4176,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.24617079391062513,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.4127,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.25772549466191574,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.4125,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.25184560964767416,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.4172,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.25165808856065675,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.4289,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.26444567180101963,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.4172,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.2593759856833136,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.4171,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.2329914942942321,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.4031,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.25019794857992117,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.4248,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.26740964820234725,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.437,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.24722283354311667,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.4147,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.24322753732562405,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.4223,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.26029747234695744,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.4288,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.24120471922737433,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.429,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.24319058404353414,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.4309,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.22862139103233145,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.4164,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2256181010718263,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.4237,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.23243458106708717,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.4153,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.36563696411779084,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.4384,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.22670121659337125,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.4215,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2299595625158766,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.4226,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.24137963359920372,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.3982,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.24432941921509263,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.4325,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2252950821062403,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.4006,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.2198137130449495,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.4266,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.21673597245427845,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.4308,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.2308135404421714,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.4237,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.22267111156276367,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.4264,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.23556505273827308,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.4258,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2345367070908846,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.4113,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.2256093552127892,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.4221,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2327284531555884,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.4123,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.23737407134981786,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.4194,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.22985121147420662,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.412,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.22853217350008626,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.4157,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.22831780816102115,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.415,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2385123348952365,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.4008,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.21600329390840603,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.4148,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.21502645217260094,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.419,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.22737620692184393,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.4196,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.22097087319342743,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.4244,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2151772445716174,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.4097,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.2265699584986458,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.4164,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.22134696688547234,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.4202,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.21582660562719583,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.4174,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.2220176910133415,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.412,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.21068000689348151,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.4183,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.20745189758415783,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.4155,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.21026024958411252,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.4088,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.2173402690438337,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.421,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.8118893090158115,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.4265,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.21320207862977275,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.4129,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.21074001761409666,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.4238,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.22111547544086313,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.4206,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.21262977996502438,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.4177,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.21776397452638563,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.4126,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.2188985568398712,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.4107,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.20175580693943568,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.4178,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.20989524165145534,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.4207,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.21919575830221602,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.4174,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.21474596157908632,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.4075,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.21642519863834514,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.4245,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.21215757992845763,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.4222,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.21180923038372815,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.4047,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.21710567915437207,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.4256,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.21532090786261193,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.4242,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.21028644737512778,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.4174,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.22138699979411833,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.4163,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.23776405652091445,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.4198,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2230816176076829,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.4143,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.20862686387404686,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.4208,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.21890107486896215,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.4212,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.45466941924406823,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.4287,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.2098147123744813,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.4122,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.21095251343455634,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.4151,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.20677780323014183,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.4126,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.22737807299952104,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.4074,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.20975145540638207,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.4283,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.21165185108403822,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.4094,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.21184499714155866,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.4166,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.20754919768363742,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.4164,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.20673634151586848,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.4143,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.20455981181849783,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.4167,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 1.304308406876891,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.4228,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.21776896494028625,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.4114,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.20527236283829953,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.4084,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.21779711431506263,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.4099,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.21084445033682453,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.4318,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.2105075108435762,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.4156,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.2127348942115985,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.4189,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.21041981745415647,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.4195,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.21166177821950008,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.4245,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.20923605421516733,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.4119,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.21492387534357263,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.415,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.2057012280292932,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.4181,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.23629234324533968,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.402,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.20839867064494555,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.4014,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.19792964812108546,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.4078,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.20191539256953578,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.4068,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.20937691995855565,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.426,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.21931019933294113,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.4187,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2145916138990697,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.409,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.2136897678737142,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.427,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.210353367285756,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.4177,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.256134624400985,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.414,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.21094535476607693,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.4199,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.2142676517504532,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.4212,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.20246984949915336,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.3961,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.5814764728553038,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.4143,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.2087056987516525,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.4102,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.2081186669294,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.4162,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.20412307897591037,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.4191,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.21343912549371452,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.4227,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.20533587156763322,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.412,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2954058978109314,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.4263,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.2104814015582867,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.4252,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.20472462883901862,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.405,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.21459229634980873,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.414,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2082019537877697,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.4176,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.21701762191317336,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.417,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.2047026949016458,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.4206,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.2111583001394666,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.4192,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.20874201354165642,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.409,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.20065898465451715,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.4132,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.20470939664764135,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.4126,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2060416631035903,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.4126,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.19653116497479936,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.4132,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.20169148242801754,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.4317,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.19677221303550008,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.4186,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.21184447674643517,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.4118,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2104890993993323,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.4218,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2103053282652665,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.4222,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.20866269513199576,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.4145,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.2078625029056544,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.4225,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.20424313694987417,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.4124,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.19913154945695852,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.4139,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.21489346302871143,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.4154,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.20267101632102044,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.4072,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.20492007333003998,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.4205,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.21572764998475402,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.4161,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 8.058427946560233,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.5109,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.24249094902106344,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.4198,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.20437067133342848,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.4192,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.20380547559828904,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.4128,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.20811978623975444,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.4204,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.20146577419111322,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.4214,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.2582592706370146,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.4248,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.19920334686154992,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.4167,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.1990342166741561,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.4319,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.208729939304804,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.4207,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.21253681912679553,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.4084,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.20028423393893235,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.4208,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.21640752768053467,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.4325,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.20106545936643383,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.4231,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.19968780860805463,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.42,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.20194050038301198,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.426,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.20537716536365255,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.4244,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.20847008288039295,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.4057,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.21313520260001728,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.4216,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.20486393861481111,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.4056,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.19631202951901164,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.4259,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.20094282273226866,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.4198,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.20177594704108967,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.4173,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.19956150827280814,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.4103,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2505240007000213,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.4162,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.19589318857787,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.4205,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.1964331914959973,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.4248,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.2004695521326146,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.422,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.19372662192714293,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.4172,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.20103416340955593,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.4191,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.20839494778614132,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.425,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2013740349362046,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.4245,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.20343663248564797,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.4045,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.20128299491581317,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.4169,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.20125732829993462,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.41,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.20832805149324007,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.417,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.20866381361439984,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.4218,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.21064041412391257,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.4234,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2988029833379755,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.4168,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.19978199339791947,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.4317,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.1984497040739864,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.4323,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.1961111872995065,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.4335,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.2000755712209826,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.4266,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.19922213746162135,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.4258,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.20706977464046306,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.4203,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.20757221360693245,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.4216,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.20199979973073134,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.4189,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.19931152238304328,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.4128,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.21485550286016988,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.416,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.20821331165932264,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.4143,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2134377281493586,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.4128,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.21787577831307908,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.4207,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.1936553919996971,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.4165,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.19667269250780606,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.4101,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.19982236031952197,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.4147,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.20258468259759804,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.4205,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 1.8173288146524307,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.4686,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.20002513819218581,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.4172,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.2022195803546575,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.4227,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.20361176984783358,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.4146,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.19243497606123225,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.4145,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.19613871670351307,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.4157,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.2097931614296291,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.4032,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.20059703126816328,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.4058,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.2091601333888993,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.4166,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2047246711926801,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.4171,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.20327674602941587,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.4224,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.1979923805289832,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.4166,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.9419196051543163,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.4318,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.21051711277081722,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.4218,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.1970147897732513,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.4121,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.2192634486660308,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.419,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.20593982818231538,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.4292,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.2027935451123194,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.4122,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.2018992623799976,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.4026,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.1983029437120332,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.4061,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.2032238666436107,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.4098,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.19432921028405514,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.4107,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.19722716343350086,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.4193,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.20522985540999072,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.4198,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.2063703253077824,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.4148,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.20204200062521693,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.4135,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.20772822098729488,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.4177,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.20529372390751863,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.4053,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.19802417512456402,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.4161,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.20513199365426607,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.4154,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.1956318917166491,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.4149,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.19924747845628346,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.4081,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.20292251032031722,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.4071,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.20328248433830562,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.4134,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.19910699862194042,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.4304,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2012825303091843,
+      "learning_rate": 0.0,
+      "loss": 0.4189,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 3.513876417471316e+18,
+      "train_loss": 0.5338705628748365,
+      "train_runtime": 31789.0566,
+      "train_samples_per_second": 4.97,
+      "train_steps_per_second": 0.039
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.513876417471316e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}