diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17226 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.9973915878708834,
+  "eval_steps": 500,
+  "global_step": 2452,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0016302575806977503,
+      "grad_norm": 0.7840946912765503,
+      "learning_rate": 0.0,
+      "loss": 0.4647,
+      "step": 1
+    },
+    {
+      "epoch": 0.0032605151613955006,
+      "grad_norm": 0.30409395694732666,
+      "learning_rate": 2.1533827903669654e-05,
+      "loss": 0.4264,
+      "step": 2
+    },
+    {
+      "epoch": 0.00489077274209325,
+      "grad_norm": 0.2784154713153839,
+      "learning_rate": 3.413030972429927e-05,
+      "loss": 0.3859,
+      "step": 3
+    },
+    {
+      "epoch": 0.006521030322791001,
+      "grad_norm": 0.2332989126443863,
+      "learning_rate": 4.306765580733931e-05,
+      "loss": 0.3903,
+      "step": 4
+    },
+    {
+      "epoch": 0.008151287903488751,
+      "grad_norm": 0.18156054615974426,
+      "learning_rate": 5e-05,
+      "loss": 0.4113,
+      "step": 5
+    },
+    {
+      "epoch": 0.0097815454841865,
+      "grad_norm": 0.1587788164615631,
+      "learning_rate": 5.5664137627968925e-05,
+      "loss": 0.3769,
+      "step": 6
+    },
+    {
+      "epoch": 0.011411803064884252,
+      "grad_norm": 0.13205036520957947,
+      "learning_rate": 6.0453097756108376e-05,
+      "loss": 0.356,
+      "step": 7
+    },
+    {
+      "epoch": 0.013042060645582002,
+      "grad_norm": 0.131794735789299,
+      "learning_rate": 6.460148371100896e-05,
+      "loss": 0.3509,
+      "step": 8
+    },
+    {
+      "epoch": 0.014672318226279752,
+      "grad_norm": 0.13253478705883026,
+      "learning_rate": 6.826061944859854e-05,
+      "loss": 0.3566,
+      "step": 9
+    },
+    {
+      "epoch": 0.016302575806977502,
+      "grad_norm": 0.1417822539806366,
+      "learning_rate": 7.153382790366967e-05,
+      "loss": 0.3328,
+      "step": 10
+    },
+    {
+      "epoch": 0.01793283338767525,
+      "grad_norm": 0.1301659494638443,
+      "learning_rate": 7.449480512024892e-05,
+      "loss": 0.3558,
+      "step": 11
+    },
+    {
+      "epoch": 0.019563090968373,
+      "grad_norm": 0.1542751044034958,
+      "learning_rate": 7.719796553163858e-05,
+      "loss": 0.3789,
+      "step": 12
+    },
+    {
+      "epoch": 0.021193348549070755,
+      "grad_norm": 0.11910240352153778,
+      "learning_rate": 7.968463205835412e-05,
+      "loss": 0.3458,
+      "step": 13
+    },
+    {
+      "epoch": 0.022823606129768505,
+      "grad_norm": 0.128227099776268,
+      "learning_rate": 8.198692565977803e-05,
+      "loss": 0.3414,
+      "step": 14
+    },
+    {
+      "epoch": 0.024453863710466255,
+      "grad_norm": 0.1177757978439331,
+      "learning_rate": 8.413030972429928e-05,
+      "loss": 0.3145,
+      "step": 15
+    },
+    {
+      "epoch": 0.026084121291164004,
+      "grad_norm": 0.1591070592403412,
+      "learning_rate": 8.613531161467861e-05,
+      "loss": 0.335,
+      "step": 16
+    },
+    {
+      "epoch": 0.027714378871861754,
+      "grad_norm": 0.1489201784133911,
+      "learning_rate": 8.80187213861294e-05,
+      "loss": 0.2942,
+      "step": 17
+    },
+    {
+      "epoch": 0.029344636452559504,
+      "grad_norm": 0.1476200520992279,
+      "learning_rate": 8.979444735226819e-05,
+      "loss": 0.3173,
+      "step": 18
+    },
+    {
+      "epoch": 0.030974894033257254,
+      "grad_norm": 0.20016862452030182,
+      "learning_rate": 9.147414002175752e-05,
+      "loss": 0.3336,
+      "step": 19
+    },
+    {
+      "epoch": 0.032605151613955004,
+      "grad_norm": 0.15757253766059875,
+      "learning_rate": 9.306765580733931e-05,
+      "loss": 0.324,
+      "step": 20
+    },
+    {
+      "epoch": 0.034235409194652754,
+      "grad_norm": 0.19339101016521454,
+      "learning_rate": 9.458340748040766e-05,
+      "loss": 0.3186,
+      "step": 21
+    },
+    {
+      "epoch": 0.0358656667753505,
+      "grad_norm": 0.1541537493467331,
+      "learning_rate": 9.602863302391859e-05,
+      "loss": 0.3142,
+      "step": 22
+    },
+    {
+      "epoch": 0.03749592435604825,
+      "grad_norm": 0.1751069873571396,
+      "learning_rate": 9.740960467331899e-05,
+      "loss": 0.3163,
+      "step": 23
+    },
+    {
+      "epoch": 0.039126181936746,
+      "grad_norm": 0.13823041319847107,
+      "learning_rate": 9.873179343530825e-05,
+      "loss": 0.3033,
+      "step": 24
+    },
+    {
+      "epoch": 0.04075643951744375,
+      "grad_norm": 0.15399403870105743,
+      "learning_rate": 0.0001,
+      "loss": 0.2893,
+      "step": 25
+    },
+    {
+      "epoch": 0.04238669709814151,
+      "grad_norm": 0.16494381427764893,
+      "learning_rate": 0.0001,
+      "loss": 0.3078,
+      "step": 26
+    },
+    {
+      "epoch": 0.04401695467883926,
+      "grad_norm": 0.14704033732414246,
+      "learning_rate": 0.0001,
+      "loss": 0.2921,
+      "step": 27
+    },
+    {
+      "epoch": 0.04564721225953701,
+      "grad_norm": 0.15865178406238556,
+      "learning_rate": 0.0001,
+      "loss": 0.317,
+      "step": 28
+    },
+    {
+      "epoch": 0.04727746984023476,
+      "grad_norm": 0.12749037146568298,
+      "learning_rate": 0.0001,
+      "loss": 0.2851,
+      "step": 29
+    },
+    {
+      "epoch": 0.04890772742093251,
+      "grad_norm": 0.12464679777622223,
+      "learning_rate": 0.0001,
+      "loss": 0.2865,
+      "step": 30
+    },
+    {
+      "epoch": 0.05053798500163026,
+      "grad_norm": 0.14522074162960052,
+      "learning_rate": 0.0001,
+      "loss": 0.2914,
+      "step": 31
+    },
+    {
+      "epoch": 0.05216824258232801,
+      "grad_norm": 0.11201729625463486,
+      "learning_rate": 0.0001,
+      "loss": 0.2773,
+      "step": 32
+    },
+    {
+      "epoch": 0.05379850016302576,
+      "grad_norm": 0.1620592176914215,
+      "learning_rate": 0.0001,
+      "loss": 0.2885,
+      "step": 33
+    },
+    {
+      "epoch": 0.05542875774372351,
+      "grad_norm": 0.12515400350093842,
+      "learning_rate": 0.0001,
+      "loss": 0.2874,
+      "step": 34
+    },
+    {
+      "epoch": 0.05705901532442126,
+      "grad_norm": 0.11235728859901428,
+      "learning_rate": 0.0001,
+      "loss": 0.2552,
+      "step": 35
+    },
+    {
+      "epoch": 0.05868927290511901,
+      "grad_norm": 0.1376715451478958,
+      "learning_rate": 0.0001,
+      "loss": 0.2738,
+      "step": 36
+    },
+    {
+      "epoch": 0.06031953048581676,
+      "grad_norm": 0.12501214444637299,
+      "learning_rate": 0.0001,
+      "loss": 0.2698,
+      "step": 37
+    },
+    {
+      "epoch": 0.06194978806651451,
+      "grad_norm": 0.16538746654987335,
+      "learning_rate": 0.0001,
+      "loss": 0.2846,
+      "step": 38
+    },
+    {
+      "epoch": 0.06358004564721226,
+      "grad_norm": 0.11548929661512375,
+      "learning_rate": 0.0001,
+      "loss": 0.2848,
+      "step": 39
+    },
+    {
+      "epoch": 0.06521030322791001,
+      "grad_norm": 0.10166307538747787,
+      "learning_rate": 0.0001,
+      "loss": 0.2578,
+      "step": 40
+    },
+    {
+      "epoch": 0.06684056080860776,
+      "grad_norm": 0.10707216709852219,
+      "learning_rate": 0.0001,
+      "loss": 0.2724,
+      "step": 41
+    },
+    {
+      "epoch": 0.06847081838930551,
+      "grad_norm": 0.11877011507749557,
+      "learning_rate": 0.0001,
+      "loss": 0.2896,
+      "step": 42
+    },
+    {
+      "epoch": 0.07010107597000326,
+      "grad_norm": 0.10695178061723709,
+      "learning_rate": 0.0001,
+      "loss": 0.2517,
+      "step": 43
+    },
+    {
+      "epoch": 0.071731333550701,
+      "grad_norm": 0.1269809901714325,
+      "learning_rate": 0.0001,
+      "loss": 0.2928,
+      "step": 44
+    },
+    {
+      "epoch": 0.07336159113139876,
+      "grad_norm": 0.1300528198480606,
+      "learning_rate": 0.0001,
+      "loss": 0.2843,
+      "step": 45
+    },
+    {
+      "epoch": 0.0749918487120965,
+      "grad_norm": 0.10712793469429016,
+      "learning_rate": 0.0001,
+      "loss": 0.252,
+      "step": 46
+    },
+    {
+      "epoch": 0.07662210629279426,
+      "grad_norm": 0.11702849715948105,
+      "learning_rate": 0.0001,
+      "loss": 0.2711,
+      "step": 47
+    },
+    {
+      "epoch": 0.078252363873492,
+      "grad_norm": 0.11782944947481155,
+      "learning_rate": 0.0001,
+      "loss": 0.2752,
+      "step": 48
+    },
+    {
+      "epoch": 0.07988262145418976,
+      "grad_norm": 0.12026762962341309,
+      "learning_rate": 0.0001,
+      "loss": 0.2669,
+      "step": 49
+    },
+    {
+      "epoch": 0.0815128790348875,
+      "grad_norm": 0.11663699150085449,
+      "learning_rate": 0.0001,
+      "loss": 0.2618,
+      "step": 50
+    },
+    {
+      "epoch": 0.08314313661558526,
+      "grad_norm": 0.1347968578338623,
+      "learning_rate": 0.0001,
+      "loss": 0.2622,
+      "step": 51
+    },
+    {
+      "epoch": 0.08477339419628302,
+      "grad_norm": 0.11729194968938828,
+      "learning_rate": 0.0001,
+      "loss": 0.2718,
+      "step": 52
+    },
+    {
+      "epoch": 0.08640365177698077,
+      "grad_norm": 0.17483912408351898,
+      "learning_rate": 0.0001,
+      "loss": 0.2824,
+      "step": 53
+    },
+    {
+      "epoch": 0.08803390935767852,
+      "grad_norm": 0.13471879065036774,
+      "learning_rate": 0.0001,
+      "loss": 0.2767,
+      "step": 54
+    },
+    {
+      "epoch": 0.08966416693837627,
+      "grad_norm": 0.14546330273151398,
+      "learning_rate": 0.0001,
+      "loss": 0.2742,
+      "step": 55
+    },
+    {
+      "epoch": 0.09129442451907402,
+      "grad_norm": 0.1298438459634781,
+      "learning_rate": 0.0001,
+      "loss": 0.2542,
+      "step": 56
+    },
+    {
+      "epoch": 0.09292468209977177,
+      "grad_norm": 0.1432414948940277,
+      "learning_rate": 0.0001,
+      "loss": 0.2737,
+      "step": 57
+    },
+    {
+      "epoch": 0.09455493968046952,
+      "grad_norm": 0.12804590165615082,
+      "learning_rate": 0.0001,
+      "loss": 0.2737,
+      "step": 58
+    },
+    {
+      "epoch": 0.09618519726116727,
+      "grad_norm": 0.1591406762599945,
+      "learning_rate": 0.0001,
+      "loss": 0.2662,
+      "step": 59
+    },
+    {
+      "epoch": 0.09781545484186502,
+      "grad_norm": 0.11162862926721573,
+      "learning_rate": 0.0001,
+      "loss": 0.2519,
+      "step": 60
+    },
+    {
+      "epoch": 0.09944571242256277,
+      "grad_norm": 0.13492637872695923,
+      "learning_rate": 0.0001,
+      "loss": 0.2567,
+      "step": 61
+    },
+    {
+      "epoch": 0.10107597000326052,
+      "grad_norm": 0.14795710146427155,
+      "learning_rate": 0.0001,
+      "loss": 0.2624,
+      "step": 62
+    },
+    {
+      "epoch": 0.10270622758395827,
+      "grad_norm": 0.18151932954788208,
+      "learning_rate": 0.0001,
+      "loss": 0.2661,
+      "step": 63
+    },
+    {
+      "epoch": 0.10433648516465602,
+      "grad_norm": 0.1397087723016739,
+      "learning_rate": 0.0001,
+      "loss": 0.2668,
+      "step": 64
+    },
+    {
+      "epoch": 0.10596674274535377,
+      "grad_norm": 0.15592671930789948,
+      "learning_rate": 0.0001,
+      "loss": 0.2624,
+      "step": 65
+    },
+    {
+      "epoch": 0.10759700032605152,
+      "grad_norm": 0.13322919607162476,
+      "learning_rate": 0.0001,
+      "loss": 0.2588,
+      "step": 66
+    },
+    {
+      "epoch": 0.10922725790674927,
+      "grad_norm": 0.14810779690742493,
+      "learning_rate": 0.0001,
+      "loss": 0.2603,
+      "step": 67
+    },
+    {
+      "epoch": 0.11085751548744702,
+      "grad_norm": 0.15098345279693604,
+      "learning_rate": 0.0001,
+      "loss": 0.2648,
+      "step": 68
+    },
+    {
+      "epoch": 0.11248777306814477,
+      "grad_norm": 0.13700629770755768,
+      "learning_rate": 0.0001,
+      "loss": 0.2743,
+      "step": 69
+    },
+    {
+      "epoch": 0.11411803064884252,
+      "grad_norm": 0.1113063395023346,
+      "learning_rate": 0.0001,
+      "loss": 0.2355,
+      "step": 70
+    },
+    {
+      "epoch": 0.11574828822954027,
+      "grad_norm": 0.13374823331832886,
+      "learning_rate": 0.0001,
+      "loss": 0.2743,
+      "step": 71
+    },
+    {
+      "epoch": 0.11737854581023802,
+      "grad_norm": 0.12639856338500977,
+      "learning_rate": 0.0001,
+      "loss": 0.2543,
+      "step": 72
+    },
+    {
+      "epoch": 0.11900880339093577,
+      "grad_norm": 0.10666421055793762,
+      "learning_rate": 0.0001,
+      "loss": 0.2579,
+      "step": 73
+    },
+    {
+      "epoch": 0.12063906097163352,
+      "grad_norm": 0.1311688870191574,
+      "learning_rate": 0.0001,
+      "loss": 0.2582,
+      "step": 74
+    },
+    {
+      "epoch": 0.12226931855233127,
+      "grad_norm": 0.10417995601892471,
+      "learning_rate": 0.0001,
+      "loss": 0.2607,
+      "step": 75
+    },
+    {
+      "epoch": 0.12389957613302902,
+      "grad_norm": 0.12071429938077927,
+      "learning_rate": 0.0001,
+      "loss": 0.2383,
+      "step": 76
+    },
+    {
+      "epoch": 0.12552983371372678,
+      "grad_norm": 0.10600121319293976,
+      "learning_rate": 0.0001,
+      "loss": 0.2463,
+      "step": 77
+    },
+    {
+      "epoch": 0.12716009129442452,
+      "grad_norm": 0.12528207898139954,
+      "learning_rate": 0.0001,
+      "loss": 0.262,
+      "step": 78
+    },
+    {
+      "epoch": 0.12879034887512228,
+      "grad_norm": 0.10099775344133377,
+      "learning_rate": 0.0001,
+      "loss": 0.2524,
+      "step": 79
+    },
+    {
+      "epoch": 0.13042060645582002,
+      "grad_norm": 0.12773346900939941,
+      "learning_rate": 0.0001,
+      "loss": 0.2572,
+      "step": 80
+    },
+    {
+      "epoch": 0.13205086403651778,
+      "grad_norm": 0.09972451627254486,
+      "learning_rate": 0.0001,
+      "loss": 0.2503,
+      "step": 81
+    },
+    {
+      "epoch": 0.13368112161721551,
+      "grad_norm": 0.1333281397819519,
+      "learning_rate": 0.0001,
+      "loss": 0.2469,
+      "step": 82
+    },
+    {
+      "epoch": 0.13531137919791328,
+      "grad_norm": 0.12170375138521194,
+      "learning_rate": 0.0001,
+      "loss": 0.2603,
+      "step": 83
+    },
+    {
+      "epoch": 0.13694163677861101,
+      "grad_norm": 0.13836190104484558,
+      "learning_rate": 0.0001,
+      "loss": 0.2736,
+      "step": 84
+    },
+    {
+      "epoch": 0.13857189435930878,
+      "grad_norm": 0.11757717281579971,
+      "learning_rate": 0.0001,
+      "loss": 0.2738,
+      "step": 85
+    },
+    {
+      "epoch": 0.14020215194000651,
+      "grad_norm": 0.1507899910211563,
+      "learning_rate": 0.0001,
+      "loss": 0.2455,
+      "step": 86
+    },
+    {
+      "epoch": 0.14183240952070428,
+      "grad_norm": 0.12341590225696564,
+      "learning_rate": 0.0001,
+      "loss": 0.2572,
+      "step": 87
+    },
+    {
+      "epoch": 0.143462667101402,
+      "grad_norm": 0.1795368492603302,
+      "learning_rate": 0.0001,
+      "loss": 0.2788,
+      "step": 88
+    },
+    {
+      "epoch": 0.14509292468209978,
+      "grad_norm": 0.10201910138130188,
+      "learning_rate": 0.0001,
+      "loss": 0.2324,
+      "step": 89
+    },
+    {
+      "epoch": 0.1467231822627975,
+      "grad_norm": 0.12101958692073822,
+      "learning_rate": 0.0001,
+      "loss": 0.2427,
+      "step": 90
+    },
+    {
+      "epoch": 0.14835343984349528,
+      "grad_norm": 0.18389801681041718,
+      "learning_rate": 0.0001,
+      "loss": 0.2653,
+      "step": 91
+    },
+    {
+      "epoch": 0.149983697424193,
+      "grad_norm": 0.13933244347572327,
+      "learning_rate": 0.0001,
+      "loss": 0.2488,
+      "step": 92
+    },
+    {
+      "epoch": 0.15161395500489078,
+      "grad_norm": 0.18606036901474,
+      "learning_rate": 0.0001,
+      "loss": 0.2618,
+      "step": 93
+    },
+    {
+      "epoch": 0.1532442125855885,
+      "grad_norm": 0.14273059368133545,
+      "learning_rate": 0.0001,
+      "loss": 0.2437,
+      "step": 94
+    },
+    {
+      "epoch": 0.15487447016628628,
+      "grad_norm": 0.10715391486883163,
+      "learning_rate": 0.0001,
+      "loss": 0.2532,
+      "step": 95
+    },
+    {
+      "epoch": 0.156504727746984,
+      "grad_norm": 0.14086727797985077,
+      "learning_rate": 0.0001,
+      "loss": 0.2512,
+      "step": 96
+    },
+    {
+      "epoch": 0.15813498532768178,
+      "grad_norm": 0.12368986010551453,
+      "learning_rate": 0.0001,
+      "loss": 0.2468,
+      "step": 97
+    },
+    {
+      "epoch": 0.1597652429083795,
+      "grad_norm": 0.14032916724681854,
+      "learning_rate": 0.0001,
+      "loss": 0.2706,
+      "step": 98
+    },
+    {
+      "epoch": 0.16139550048907728,
+      "grad_norm": 0.12382455170154572,
+      "learning_rate": 0.0001,
+      "loss": 0.2432,
+      "step": 99
+    },
+    {
+      "epoch": 0.163025758069775,
+      "grad_norm": 0.11922313272953033,
+      "learning_rate": 0.0001,
+      "loss": 0.2584,
+      "step": 100
+    },
+    {
+      "epoch": 0.16465601565047278,
+      "grad_norm": 0.12890221178531647,
+      "learning_rate": 0.0001,
+      "loss": 0.2503,
+      "step": 101
+    },
+    {
+      "epoch": 0.1662862732311705,
+      "grad_norm": 0.1167726069688797,
+      "learning_rate": 0.0001,
+      "loss": 0.2513,
+      "step": 102
+    },
+    {
+      "epoch": 0.16791653081186828,
+      "grad_norm": 0.1356429159641266,
+      "learning_rate": 0.0001,
+      "loss": 0.2774,
+      "step": 103
+    },
+    {
+      "epoch": 0.16954678839256604,
+      "grad_norm": 0.13073065876960754,
+      "learning_rate": 0.0001,
+      "loss": 0.2497,
+      "step": 104
+    },
+    {
+      "epoch": 0.17117704597326378,
+      "grad_norm": 0.1371566206216812,
+      "learning_rate": 0.0001,
+      "loss": 0.2606,
+      "step": 105
+    },
+    {
+      "epoch": 0.17280730355396154,
+      "grad_norm": 0.13844291865825653,
+      "learning_rate": 0.0001,
+      "loss": 0.2461,
+      "step": 106
+    },
+    {
+      "epoch": 0.17443756113465927,
+      "grad_norm": 0.12421681731939316,
+      "learning_rate": 0.0001,
+      "loss": 0.2353,
+      "step": 107
+    },
+    {
+      "epoch": 0.17606781871535704,
+      "grad_norm": 0.11806869506835938,
+      "learning_rate": 0.0001,
+      "loss": 0.2458,
+      "step": 108
+    },
+    {
+      "epoch": 0.17769807629605477,
+      "grad_norm": 0.13172969222068787,
+      "learning_rate": 0.0001,
+      "loss": 0.2596,
+      "step": 109
+    },
+    {
+      "epoch": 0.17932833387675254,
+      "grad_norm": 0.10319314897060394,
+      "learning_rate": 0.0001,
+      "loss": 0.2294,
+      "step": 110
+    },
+    {
+      "epoch": 0.18095859145745027,
+      "grad_norm": 0.1400890052318573,
+      "learning_rate": 0.0001,
+      "loss": 0.243,
+      "step": 111
+    },
+    {
+      "epoch": 0.18258884903814804,
+      "grad_norm": 0.11478696018457413,
+      "learning_rate": 0.0001,
+      "loss": 0.2586,
+      "step": 112
+    },
+    {
+      "epoch": 0.18421910661884577,
+      "grad_norm": 0.11043928563594818,
+      "learning_rate": 0.0001,
+      "loss": 0.2481,
+      "step": 113
+    },
+    {
+      "epoch": 0.18584936419954354,
+      "grad_norm": 0.12820887565612793,
+      "learning_rate": 0.0001,
+      "loss": 0.2627,
+      "step": 114
+    },
+    {
+      "epoch": 0.18747962178024127,
+      "grad_norm": 0.14865200221538544,
+      "learning_rate": 0.0001,
+      "loss": 0.2664,
+      "step": 115
+    },
+    {
+      "epoch": 0.18910987936093904,
+      "grad_norm": 0.11971151828765869,
+      "learning_rate": 0.0001,
+      "loss": 0.2462,
+      "step": 116
+    },
+    {
+      "epoch": 0.19074013694163677,
+      "grad_norm": 0.11976753175258636,
+      "learning_rate": 0.0001,
+      "loss": 0.2364,
+      "step": 117
+    },
+    {
+      "epoch": 0.19237039452233454,
+      "grad_norm": 0.11274496465921402,
+      "learning_rate": 0.0001,
+      "loss": 0.2444,
+      "step": 118
+    },
+    {
+      "epoch": 0.19400065210303227,
+      "grad_norm": 0.13468791544437408,
+      "learning_rate": 0.0001,
+      "loss": 0.2449,
+      "step": 119
+    },
+    {
+      "epoch": 0.19563090968373004,
+      "grad_norm": 0.14791305363178253,
+      "learning_rate": 0.0001,
+      "loss": 0.2687,
+      "step": 120
+    },
+    {
+      "epoch": 0.19726116726442777,
+      "grad_norm": 0.11851610988378525,
+      "learning_rate": 0.0001,
+      "loss": 0.2395,
+      "step": 121
+    },
+    {
+      "epoch": 0.19889142484512554,
+      "grad_norm": 0.1273575872182846,
+      "learning_rate": 0.0001,
+      "loss": 0.2421,
+      "step": 122
+    },
+    {
+      "epoch": 0.20052168242582327,
+      "grad_norm": 0.12378577888011932,
+      "learning_rate": 0.0001,
+      "loss": 0.2394,
+      "step": 123
+    },
+    {
+      "epoch": 0.20215194000652104,
+      "grad_norm": 0.14497146010398865,
+      "learning_rate": 0.0001,
+      "loss": 0.2377,
+      "step": 124
+    },
+    {
+      "epoch": 0.20378219758721877,
+      "grad_norm": 0.15099285542964935,
+      "learning_rate": 0.0001,
+      "loss": 0.2437,
+      "step": 125
+    },
+    {
+      "epoch": 0.20541245516791654,
+      "grad_norm": 0.1162499412894249,
+      "learning_rate": 0.0001,
+      "loss": 0.2463,
+      "step": 126
+    },
+    {
+      "epoch": 0.20704271274861427,
+      "grad_norm": 0.13596996665000916,
+      "learning_rate": 0.0001,
+      "loss": 0.2251,
+      "step": 127
+    },
+    {
+      "epoch": 0.20867297032931204,
+      "grad_norm": 0.1420516073703766,
+      "learning_rate": 0.0001,
+      "loss": 0.2616,
+      "step": 128
+    },
+    {
+      "epoch": 0.21030322791000977,
+      "grad_norm": 0.1711655855178833,
+      "learning_rate": 0.0001,
+      "loss": 0.2599,
+      "step": 129
+    },
+    {
+      "epoch": 0.21193348549070753,
+      "grad_norm": 0.12194457650184631,
+      "learning_rate": 0.0001,
+      "loss": 0.2441,
+      "step": 130
+    },
+    {
+      "epoch": 0.21356374307140527,
+      "grad_norm": 0.11060874909162521,
+      "learning_rate": 0.0001,
+      "loss": 0.2207,
+      "step": 131
+    },
+    {
+      "epoch": 0.21519400065210303,
+      "grad_norm": 0.1700860857963562,
+      "learning_rate": 0.0001,
+      "loss": 0.2587,
+      "step": 132
+    },
+    {
+      "epoch": 0.21682425823280077,
+      "grad_norm": 0.12098715454339981,
+      "learning_rate": 0.0001,
+      "loss": 0.2407,
+      "step": 133
+    },
+    {
+      "epoch": 0.21845451581349853,
+      "grad_norm": 0.1312795877456665,
+      "learning_rate": 0.0001,
+      "loss": 0.2505,
+      "step": 134
+    },
+    {
+      "epoch": 0.22008477339419627,
+      "grad_norm": 0.12589812278747559,
+      "learning_rate": 0.0001,
+      "loss": 0.2416,
+      "step": 135
+    },
+    {
+      "epoch": 0.22171503097489403,
+      "grad_norm": 0.11886783689260483,
+      "learning_rate": 0.0001,
+      "loss": 0.2589,
+      "step": 136
+    },
+    {
+      "epoch": 0.22334528855559177,
+      "grad_norm": 0.12084617465734482,
+      "learning_rate": 0.0001,
+      "loss": 0.2443,
+      "step": 137
+    },
+    {
+      "epoch": 0.22497554613628953,
+      "grad_norm": 0.12979595363140106,
+      "learning_rate": 0.0001,
+      "loss": 0.2441,
+      "step": 138
+    },
+    {
+      "epoch": 0.2266058037169873,
+      "grad_norm": 0.1361086517572403,
+      "learning_rate": 0.0001,
+      "loss": 0.2304,
+      "step": 139
+    },
+    {
+      "epoch": 0.22823606129768503,
+      "grad_norm": 0.13962669670581818,
+      "learning_rate": 0.0001,
+      "loss": 0.2511,
+      "step": 140
+    },
+    {
+      "epoch": 0.2298663188783828,
+      "grad_norm": 0.13188579678535461,
+      "learning_rate": 0.0001,
+      "loss": 0.2478,
+      "step": 141
+    },
+    {
+      "epoch": 0.23149657645908053,
+      "grad_norm": 0.14598271250724792,
+      "learning_rate": 0.0001,
+      "loss": 0.2516,
+      "step": 142
+    },
+    {
+      "epoch": 0.2331268340397783,
+      "grad_norm": 0.14357416331768036,
+      "learning_rate": 0.0001,
+      "loss": 0.2414,
+      "step": 143
+    },
+    {
+      "epoch": 0.23475709162047603,
+      "grad_norm": 0.11369388550519943,
+      "learning_rate": 0.0001,
+      "loss": 0.2331,
+      "step": 144
+    },
+    {
+      "epoch": 0.2363873492011738,
+      "grad_norm": 0.14625732600688934,
+      "learning_rate": 0.0001,
+      "loss": 0.2177,
+      "step": 145
+    },
+    {
+      "epoch": 0.23801760678187153,
+      "grad_norm": 0.14222382009029388,
+      "learning_rate": 0.0001,
+      "loss": 0.2541,
+      "step": 146
+    },
+    {
+      "epoch": 0.2396478643625693,
+      "grad_norm": 0.11487408727407455,
+      "learning_rate": 0.0001,
+      "loss": 0.224,
+      "step": 147
+    },
+    {
+      "epoch": 0.24127812194326703,
+      "grad_norm": 0.1486762911081314,
+      "learning_rate": 0.0001,
+      "loss": 0.2235,
+      "step": 148
+    },
+    {
+      "epoch": 0.2429083795239648,
+      "grad_norm": 0.1313285082578659,
+      "learning_rate": 0.0001,
+      "loss": 0.2558,
+      "step": 149
+    },
+    {
+      "epoch": 0.24453863710466253,
+      "grad_norm": 0.10049161314964294,
+      "learning_rate": 0.0001,
+      "loss": 0.2342,
+      "step": 150
+    },
+    {
+      "epoch": 0.2461688946853603,
+      "grad_norm": 0.13103991746902466,
+      "learning_rate": 0.0001,
+      "loss": 0.2472,
+      "step": 151
+    },
+    {
+      "epoch": 0.24779915226605803,
+      "grad_norm": 0.1439107060432434,
+      "learning_rate": 0.0001,
+      "loss": 0.2536,
+      "step": 152
+    },
+    {
+      "epoch": 0.2494294098467558,
+      "grad_norm": 0.11082714796066284,
+      "learning_rate": 0.0001,
+      "loss": 0.2245,
+      "step": 153
+    },
+    {
+      "epoch": 0.25105966742745356,
+      "grad_norm": 0.15587423741817474,
+      "learning_rate": 0.0001,
+      "loss": 0.2539,
+      "step": 154
+    },
+    {
+      "epoch": 0.2526899250081513,
+      "grad_norm": 0.13212022185325623,
+      "learning_rate": 0.0001,
+      "loss": 0.2488,
+      "step": 155
+    },
+    {
+      "epoch": 0.25432018258884903,
+      "grad_norm": 0.1342269331216812,
+      "learning_rate": 0.0001,
+      "loss": 0.2414,
+      "step": 156
+    },
+    {
+      "epoch": 0.25595044016954677,
+      "grad_norm": 0.14503146708011627,
+      "learning_rate": 0.0001,
+      "loss": 0.2464,
+      "step": 157
+    },
+    {
+      "epoch": 0.25758069775024456,
+      "grad_norm": 0.1345171481370926,
+      "learning_rate": 0.0001,
+      "loss": 0.2528,
+      "step": 158
+    },
+    {
+      "epoch": 0.2592109553309423,
+      "grad_norm": 0.12593773007392883,
+      "learning_rate": 0.0001,
+      "loss": 0.2293,
+      "step": 159
+    },
+    {
+      "epoch": 0.26084121291164003,
+      "grad_norm": 0.11615213751792908,
+      "learning_rate": 0.0001,
+      "loss": 0.2103,
+      "step": 160
+    },
+    {
+      "epoch": 0.26247147049233777,
+      "grad_norm": 0.12266945838928223,
+      "learning_rate": 0.0001,
+      "loss": 0.2338,
+      "step": 161
+    },
+    {
+      "epoch": 0.26410172807303556,
+      "grad_norm": 0.13719286024570465,
+      "learning_rate": 0.0001,
+      "loss": 0.2526,
+      "step": 162
+    },
+    {
+      "epoch": 0.2657319856537333,
+      "grad_norm": 0.12018352746963501,
+      "learning_rate": 0.0001,
+      "loss": 0.2344,
+      "step": 163
+    },
+    {
+      "epoch": 0.26736224323443103,
+      "grad_norm": 0.12734843790531158,
+      "learning_rate": 0.0001,
+      "loss": 0.2228,
+      "step": 164
+    },
+    {
+      "epoch": 0.26899250081512877,
+      "grad_norm": 0.18533261120319366,
+      "learning_rate": 0.0001,
+      "loss": 0.2616,
+      "step": 165
+    },
+    {
+      "epoch": 0.27062275839582656,
+      "grad_norm": 0.12489785999059677,
+      "learning_rate": 0.0001,
+      "loss": 0.2481,
+      "step": 166
+    },
+    {
+      "epoch": 0.2722530159765243,
+      "grad_norm": 0.11786579340696335,
+      "learning_rate": 0.0001,
+      "loss": 0.2459,
+      "step": 167
+    },
+    {
+      "epoch": 0.27388327355722203,
+      "grad_norm": 0.11804036796092987,
+      "learning_rate": 0.0001,
+      "loss": 0.2507,
+      "step": 168
+    },
+    {
+      "epoch": 0.27551353113791976,
+      "grad_norm": 0.12300978600978851,
+      "learning_rate": 0.0001,
+      "loss": 0.2402,
+      "step": 169
+    },
+    {
+      "epoch": 0.27714378871861756,
+      "grad_norm": 0.123172327876091,
+      "learning_rate": 0.0001,
+      "loss": 0.2393,
+      "step": 170
+    },
+    {
+      "epoch": 0.2787740462993153,
+      "grad_norm": 0.12156080454587936,
+      "learning_rate": 0.0001,
+      "loss": 0.2366,
+      "step": 171
+    },
+    {
+      "epoch": 0.28040430388001303,
+      "grad_norm": 0.12459408491849899,
+      "learning_rate": 0.0001,
+      "loss": 0.2449,
+      "step": 172
+    },
+    {
+      "epoch": 0.2820345614607108,
+      "grad_norm": 0.10773272812366486,
+      "learning_rate": 0.0001,
+      "loss": 0.2327,
+      "step": 173
+    },
+    {
+      "epoch": 0.28366481904140856,
+      "grad_norm": 0.11818479001522064,
+      "learning_rate": 0.0001,
+      "loss": 0.2429,
+      "step": 174
+    },
+    {
+      "epoch": 0.2852950766221063,
+      "grad_norm": 0.11246544122695923,
+      "learning_rate": 0.0001,
+      "loss": 0.2264,
+      "step": 175
+    },
+    {
+      "epoch": 0.286925334202804,
+      "grad_norm": 0.1404242068529129,
+      "learning_rate": 0.0001,
+      "loss": 0.2523,
+      "step": 176
+    },
+    {
+      "epoch": 0.2885555917835018,
+      "grad_norm": 0.18125034868717194,
+      "learning_rate": 0.0001,
+      "loss": 0.2378,
+      "step": 177
+    },
+    {
+      "epoch": 0.29018584936419956,
+      "grad_norm": 0.14246758818626404,
+      "learning_rate": 0.0001,
+      "loss": 0.2454,
+      "step": 178
+    },
+    {
+      "epoch": 0.2918161069448973,
+      "grad_norm": 0.13600490987300873,
+      "learning_rate": 0.0001,
+      "loss": 0.2473,
+      "step": 179
+    },
+    {
+      "epoch": 0.293446364525595,
+      "grad_norm": 0.12414231896400452,
+      "learning_rate": 0.0001,
+      "loss": 0.2438,
+      "step": 180
+    },
+    {
+      "epoch": 0.2950766221062928,
+      "grad_norm": 0.12273810058832169,
+      "learning_rate": 0.0001,
+      "loss": 0.2311,
+      "step": 181
+    },
+    {
+      "epoch": 0.29670687968699055,
+      "grad_norm": 0.104936882853508,
+      "learning_rate": 0.0001,
+      "loss": 0.2223,
+      "step": 182
+    },
+    {
+      "epoch": 0.2983371372676883,
+      "grad_norm": 0.12966670095920563,
+      "learning_rate": 0.0001,
+      "loss": 0.2269,
+      "step": 183
+    },
+    {
+      "epoch": 0.299967394848386,
+      "grad_norm": 0.11775387078523636,
+      "learning_rate": 0.0001,
+      "loss": 0.2434,
+      "step": 184
+    },
+    {
+      "epoch": 0.3015976524290838,
+      "grad_norm": 0.16570591926574707,
+      "learning_rate": 0.0001,
+      "loss": 0.2497,
+      "step": 185
+    },
+    {
+      "epoch": 0.30322791000978155,
+      "grad_norm": 0.1248738244175911,
+      "learning_rate": 0.0001,
+      "loss": 0.2348,
+      "step": 186
+    },
+    {
+      "epoch": 0.3048581675904793,
+      "grad_norm": 0.17375342547893524,
+      "learning_rate": 0.0001,
+      "loss": 0.2368,
+      "step": 187
+    },
+    {
+      "epoch": 0.306488425171177,
+      "grad_norm": 0.1663849800825119,
+      "learning_rate": 0.0001,
+      "loss": 0.2289,
+      "step": 188
+    },
+    {
+      "epoch": 0.3081186827518748,
+      "grad_norm": 0.10794688761234283,
+      "learning_rate": 0.0001,
+      "loss": 0.2214,
+      "step": 189
+    },
+    {
+      "epoch": 0.30974894033257255,
+      "grad_norm": 0.14207707345485687,
+      "learning_rate": 0.0001,
+      "loss": 0.2278,
+      "step": 190
+    },
+    {
+      "epoch": 0.3113791979132703,
+      "grad_norm": 0.14883771538734436,
+      "learning_rate": 0.0001,
+      "loss": 0.2543,
+      "step": 191
+    },
+    {
+      "epoch": 0.313009455493968,
+      "grad_norm": 0.1482788771390915,
+      "learning_rate": 0.0001,
+      "loss": 0.2361,
+      "step": 192
+    },
+    {
+      "epoch": 0.3146397130746658,
+      "grad_norm": 0.16753090918064117,
+      "learning_rate": 0.0001,
+      "loss": 0.2441,
+      "step": 193
+    },
+    {
+      "epoch": 0.31626997065536355,
+      "grad_norm": 0.10236582159996033,
+      "learning_rate": 0.0001,
+      "loss": 0.2091,
+      "step": 194
+    },
+    {
+      "epoch": 0.3179002282360613,
+      "grad_norm": 0.11371749639511108,
+      "learning_rate": 0.0001,
+      "loss": 0.2303,
+      "step": 195
+    },
+    {
+      "epoch": 0.319530485816759,
+      "grad_norm": 0.12658603489398956,
+      "learning_rate": 0.0001,
+      "loss": 0.2361,
+      "step": 196
+    },
+    {
+      "epoch": 0.3211607433974568,
+      "grad_norm": 0.11970578879117966,
+      "learning_rate": 0.0001,
+      "loss": 0.2361,
+      "step": 197
+    },
+    {
+      "epoch": 0.32279100097815455,
+      "grad_norm": 0.13204766809940338,
+      "learning_rate": 0.0001,
+      "loss": 0.2475,
+      "step": 198
+    },
+    {
+      "epoch": 0.3244212585588523,
+      "grad_norm": 0.11528757214546204,
+      "learning_rate": 0.0001,
+      "loss": 0.2395,
+      "step": 199
+    },
+    {
+      "epoch": 0.32605151613955,
+      "grad_norm": 0.11398864537477493,
+      "learning_rate": 0.0001,
+      "loss": 0.2294,
+      "step": 200
+    },
+    {
+      "epoch": 0.3276817737202478,
+      "grad_norm": 0.13498589396476746,
+      "learning_rate": 0.0001,
+      "loss": 0.2461,
+      "step": 201
+    },
+    {
+      "epoch": 0.32931203130094555,
+      "grad_norm": 0.10650527477264404,
+      "learning_rate": 0.0001,
+      "loss": 0.2279,
+      "step": 202
+    },
+    {
+      "epoch": 0.3309422888816433,
+      "grad_norm": 0.10871734470129013,
+      "learning_rate": 0.0001,
+      "loss": 0.2379,
+      "step": 203
+    },
+    {
+      "epoch": 0.332572546462341,
+      "grad_norm": 0.13913440704345703,
+      "learning_rate": 0.0001,
+      "loss": 0.2374,
+      "step": 204
+    },
+    {
+      "epoch": 0.3342028040430388,
+      "grad_norm": 0.1410413384437561,
+      "learning_rate": 0.0001,
+      "loss": 0.2389,
+      "step": 205
+    },
+    {
+      "epoch": 0.33583306162373655,
+      "grad_norm": 0.1187388077378273,
+      "learning_rate": 0.0001,
+      "loss": 0.2173,
+      "step": 206
+    },
+    {
+      "epoch": 0.3374633192044343,
+      "grad_norm": 0.10804294794797897,
+      "learning_rate": 0.0001,
+      "loss": 0.2312,
+      "step": 207
+    },
+    {
+      "epoch": 0.3390935767851321,
+      "grad_norm": 0.11228567361831665,
+      "learning_rate": 0.0001,
+      "loss": 0.2287,
+      "step": 208
+    },
+    {
+      "epoch": 0.3407238343658298,
+      "grad_norm": 0.12240911275148392,
+      "learning_rate": 0.0001,
+      "loss": 0.2264,
+      "step": 209
+    },
+    {
+      "epoch": 0.34235409194652755,
+      "grad_norm": 0.11001206934452057,
+      "learning_rate": 0.0001,
+      "loss": 0.2465,
+      "step": 210
+    },
+    {
+      "epoch": 0.3439843495272253,
+      "grad_norm": 0.11600527912378311,
+      "learning_rate": 0.0001,
+      "loss": 0.2259,
+      "step": 211
+    },
+    {
+      "epoch": 0.3456146071079231,
+      "grad_norm": 0.11184750497341156,
+      "learning_rate": 0.0001,
+      "loss": 0.2134,
+      "step": 212
+    },
+    {
+      "epoch": 0.3472448646886208,
+      "grad_norm": 0.1267496645450592,
+      "learning_rate": 0.0001,
+      "loss": 0.2471,
+      "step": 213
+    },
+    {
+      "epoch": 0.34887512226931855,
+      "grad_norm": 0.12638495862483978,
+      "learning_rate": 0.0001,
+      "loss": 0.2486,
+      "step": 214
+    },
+    {
+      "epoch": 0.3505053798500163,
+      "grad_norm": 0.11810494214296341,
+      "learning_rate": 0.0001,
+      "loss": 0.2296,
+      "step": 215
+    },
+    {
+      "epoch": 0.3521356374307141,
+      "grad_norm": 0.1319112330675125,
+      "learning_rate": 0.0001,
+      "loss": 0.223,
+      "step": 216
+    },
+    {
+      "epoch": 0.3537658950114118,
+      "grad_norm": 0.13196417689323425,
+      "learning_rate": 0.0001,
+      "loss": 0.2273,
+      "step": 217
+    },
+    {
+      "epoch": 0.35539615259210955,
+      "grad_norm": 0.13283449411392212,
+      "learning_rate": 0.0001,
+      "loss": 0.2395,
+      "step": 218
+    },
+    {
+      "epoch": 0.3570264101728073,
+      "grad_norm": 0.11351513117551804,
+      "learning_rate": 0.0001,
+      "loss": 0.2305,
+      "step": 219
+    },
+    {
+      "epoch": 0.3586566677535051,
+      "grad_norm": 0.11622969061136246,
+      "learning_rate": 0.0001,
+      "loss": 0.2267,
+      "step": 220
+    },
+    {
+      "epoch": 0.3602869253342028,
+      "grad_norm": 0.12463167309761047,
+      "learning_rate": 0.0001,
+      "loss": 0.2345,
+      "step": 221
+    },
+    {
+      "epoch": 0.36191718291490055,
+      "grad_norm": 0.12457557767629623,
+      "learning_rate": 0.0001,
+      "loss": 0.2394,
+      "step": 222
+    },
+    {
+      "epoch": 0.3635474404955983,
+      "grad_norm": 0.11849746108055115,
+      "learning_rate": 0.0001,
+      "loss": 0.2183,
+      "step": 223
+    },
+    {
+      "epoch": 0.3651776980762961,
+      "grad_norm": 0.13964387774467468,
+      "learning_rate": 0.0001,
+      "loss": 0.2577,
+      "step": 224
+    },
+    {
+      "epoch": 0.3668079556569938,
+      "grad_norm": 0.11555647104978561,
+      "learning_rate": 0.0001,
+      "loss": 0.2154,
+      "step": 225
+    },
+    {
+      "epoch": 0.36843821323769155,
+      "grad_norm": 0.09998490661382675,
+      "learning_rate": 0.0001,
+      "loss": 0.2122,
+      "step": 226
+    },
+    {
+      "epoch": 0.3700684708183893,
+      "grad_norm": 0.12696896493434906,
+      "learning_rate": 0.0001,
+      "loss": 0.2353,
+      "step": 227
+    },
+    {
+      "epoch": 0.3716987283990871,
+      "grad_norm": 0.118056520819664,
+      "learning_rate": 0.0001,
+      "loss": 0.2229,
+      "step": 228
+    },
+    {
+      "epoch": 0.3733289859797848,
+      "grad_norm": 0.11977580189704895,
+      "learning_rate": 0.0001,
+      "loss": 0.2306,
+      "step": 229
+    },
+    {
+      "epoch": 0.37495924356048255,
+      "grad_norm": 0.13963200151920319,
+      "learning_rate": 0.0001,
+      "loss": 0.238,
+      "step": 230
+    },
+    {
+      "epoch": 0.3765895011411803,
+      "grad_norm": 0.1165039986371994,
+      "learning_rate": 0.0001,
+      "loss": 0.2298,
+      "step": 231
+    },
+    {
+      "epoch": 0.3782197587218781,
+      "grad_norm": 0.15564818680286407,
+      "learning_rate": 0.0001,
+      "loss": 0.2177,
+      "step": 232
+    },
+    {
+      "epoch": 0.3798500163025758,
+      "grad_norm": 0.12317235767841339,
+      "learning_rate": 0.0001,
+      "loss": 0.2229,
+      "step": 233
+    },
+    {
+      "epoch": 0.38148027388327355,
+      "grad_norm": 0.12770213186740875,
+      "learning_rate": 0.0001,
+      "loss": 0.239,
+      "step": 234
+    },
+    {
+      "epoch": 0.3831105314639713,
+      "grad_norm": 0.13377554714679718,
+      "learning_rate": 0.0001,
+      "loss": 0.2194,
+      "step": 235
+    },
+    {
+      "epoch": 0.3847407890446691,
+      "grad_norm": 0.2012566477060318,
+      "learning_rate": 0.0001,
+      "loss": 0.2432,
+      "step": 236
+    },
+    {
+      "epoch": 0.3863710466253668,
+      "grad_norm": 0.14003917574882507,
+      "learning_rate": 0.0001,
+      "loss": 0.2331,
+      "step": 237
+    },
+    {
+      "epoch": 0.38800130420606455,
+      "grad_norm": 0.13453909754753113,
+      "learning_rate": 0.0001,
+      "loss": 0.2255,
+      "step": 238
+    },
+    {
+      "epoch": 0.3896315617867623,
+      "grad_norm": 0.12276874482631683,
+      "learning_rate": 0.0001,
+      "loss": 0.2256,
+      "step": 239
+    },
+    {
+      "epoch": 0.3912618193674601,
+      "grad_norm": 0.12277834117412567,
+      "learning_rate": 0.0001,
+      "loss": 0.2126,
+      "step": 240
+    },
+    {
+      "epoch": 0.3928920769481578,
+      "grad_norm": 0.10638459771871567,
+      "learning_rate": 0.0001,
+      "loss": 0.2303,
+      "step": 241
+    },
+    {
+      "epoch": 0.39452233452885554,
+      "grad_norm": 0.13198097050189972,
+      "learning_rate": 0.0001,
+      "loss": 0.223,
+      "step": 242
+    },
+    {
+      "epoch": 0.39615259210955334,
+      "grad_norm": 0.11877847462892532,
+      "learning_rate": 0.0001,
+      "loss": 0.252,
+      "step": 243
+    },
+    {
+      "epoch": 0.3977828496902511,
+      "grad_norm": 0.1367732435464859,
+      "learning_rate": 0.0001,
+      "loss": 0.2388,
+      "step": 244
+    },
+    {
+      "epoch": 0.3994131072709488,
+      "grad_norm": 0.10984091460704803,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 245
+    },
+    {
+      "epoch": 0.40104336485164654,
+      "grad_norm": 0.12004056572914124,
+      "learning_rate": 0.0001,
+      "loss": 0.2145,
+      "step": 246
+    },
+    {
+      "epoch": 0.40267362243234434,
+      "grad_norm": 0.1057971715927124,
+      "learning_rate": 0.0001,
+      "loss": 0.2316,
+      "step": 247
+    },
+    {
+      "epoch": 0.40430388001304207,
+      "grad_norm": 0.11223854869604111,
+      "learning_rate": 0.0001,
+      "loss": 0.237,
+      "step": 248
+    },
+    {
+      "epoch": 0.4059341375937398,
+      "grad_norm": 0.11464496701955795,
+      "learning_rate": 0.0001,
+      "loss": 0.2384,
+      "step": 249
+    },
+    {
+      "epoch": 0.40756439517443754,
+      "grad_norm": 0.12331773340702057,
+      "learning_rate": 0.0001,
+      "loss": 0.2375,
+      "step": 250
+    },
+    {
+      "epoch": 0.40919465275513534,
+      "grad_norm": 0.1102685034275055,
+      "learning_rate": 0.0001,
+      "loss": 0.2361,
+      "step": 251
+    },
+    {
+      "epoch": 0.41082491033583307,
+      "grad_norm": 0.12823829054832458,
+      "learning_rate": 0.0001,
+      "loss": 0.23,
+      "step": 252
+    },
+    {
+      "epoch": 0.4124551679165308,
+      "grad_norm": 0.12393566220998764,
+      "learning_rate": 0.0001,
+      "loss": 0.2442,
+      "step": 253
+    },
+    {
+      "epoch": 0.41408542549722854,
+      "grad_norm": 0.10847245156764984,
+      "learning_rate": 0.0001,
+      "loss": 0.2253,
+      "step": 254
+    },
+    {
+      "epoch": 0.41571568307792633,
+      "grad_norm": 0.10520011931657791,
+      "learning_rate": 0.0001,
+      "loss": 0.238,
+      "step": 255
+    },
+    {
+      "epoch": 0.41734594065862407,
+      "grad_norm": 0.10316835343837738,
+      "learning_rate": 0.0001,
+      "loss": 0.2147,
+      "step": 256
+    },
+    {
+      "epoch": 0.4189761982393218,
+      "grad_norm": 0.09753026813268661,
+      "learning_rate": 0.0001,
+      "loss": 0.2212,
+      "step": 257
+    },
+    {
+      "epoch": 0.42060645582001954,
+      "grad_norm": 0.12001580744981766,
+      "learning_rate": 0.0001,
+      "loss": 0.2347,
+      "step": 258
+    },
+    {
+      "epoch": 0.42223671340071733,
+      "grad_norm": 0.1057935580611229,
+      "learning_rate": 0.0001,
+      "loss": 0.2184,
+      "step": 259
+    },
+    {
+      "epoch": 0.42386697098141507,
+      "grad_norm": 0.12323079258203506,
+      "learning_rate": 0.0001,
+      "loss": 0.2318,
+      "step": 260
+    },
+    {
+      "epoch": 0.4254972285621128,
+      "grad_norm": 0.11835148930549622,
+      "learning_rate": 0.0001,
+      "loss": 0.2373,
+      "step": 261
+    },
+    {
+      "epoch": 0.42712748614281054,
+      "grad_norm": 0.11595102399587631,
+      "learning_rate": 0.0001,
+      "loss": 0.2171,
+      "step": 262
+    },
+    {
+      "epoch": 0.42875774372350833,
+      "grad_norm": 0.11034155637025833,
+      "learning_rate": 0.0001,
+      "loss": 0.2209,
+      "step": 263
+    },
+    {
+      "epoch": 0.43038800130420607,
+      "grad_norm": 0.10659411549568176,
+      "learning_rate": 0.0001,
+      "loss": 0.2228,
+      "step": 264
+    },
+    {
+      "epoch": 0.4320182588849038,
+      "grad_norm": 0.15346385538578033,
+      "learning_rate": 0.0001,
+      "loss": 0.234,
+      "step": 265
+    },
+    {
+      "epoch": 0.43364851646560154,
+      "grad_norm": 0.11137716472148895,
+      "learning_rate": 0.0001,
+      "loss": 0.2297,
+      "step": 266
+    },
+    {
+      "epoch": 0.43527877404629933,
+      "grad_norm": 0.12295756489038467,
+      "learning_rate": 0.0001,
+      "loss": 0.24,
+      "step": 267
+    },
+    {
+      "epoch": 0.43690903162699707,
+      "grad_norm": 0.13455668091773987,
+      "learning_rate": 0.0001,
+      "loss": 0.2379,
+      "step": 268
+    },
+    {
+      "epoch": 0.4385392892076948,
+      "grad_norm": 0.12878896296024323,
+      "learning_rate": 0.0001,
+      "loss": 0.2406,
+      "step": 269
+    },
+    {
+      "epoch": 0.44016954678839254,
+      "grad_norm": 0.112666055560112,
+      "learning_rate": 0.0001,
+      "loss": 0.208,
+      "step": 270
+    },
+    {
+      "epoch": 0.44179980436909033,
+      "grad_norm": 0.13783974945545197,
+      "learning_rate": 0.0001,
+      "loss": 0.2497,
+      "step": 271
+    },
+    {
+      "epoch": 0.44343006194978807,
+      "grad_norm": 0.10861429572105408,
+      "learning_rate": 0.0001,
+      "loss": 0.217,
+      "step": 272
+    },
+    {
+      "epoch": 0.4450603195304858,
+      "grad_norm": 0.16053490340709686,
+      "learning_rate": 0.0001,
+      "loss": 0.2504,
+      "step": 273
+    },
+    {
+      "epoch": 0.44669057711118354,
+      "grad_norm": 0.12352034449577332,
+      "learning_rate": 0.0001,
+      "loss": 0.2389,
+      "step": 274
+    },
+    {
+      "epoch": 0.44832083469188133,
+      "grad_norm": 0.12012365460395813,
+      "learning_rate": 0.0001,
+      "loss": 0.2213,
+      "step": 275
+    },
+    {
+      "epoch": 0.44995109227257907,
+      "grad_norm": 0.10671349614858627,
+      "learning_rate": 0.0001,
+      "loss": 0.2286,
+      "step": 276
+    },
+    {
+      "epoch": 0.4515813498532768,
+      "grad_norm": 0.12723585963249207,
+      "learning_rate": 0.0001,
+      "loss": 0.2204,
+      "step": 277
+    },
+    {
+      "epoch": 0.4532116074339746,
+      "grad_norm": 0.13079780340194702,
+      "learning_rate": 0.0001,
+      "loss": 0.2383,
+      "step": 278
+    },
+    {
+      "epoch": 0.45484186501467233,
+      "grad_norm": 0.13152571022510529,
+      "learning_rate": 0.0001,
+      "loss": 0.2296,
+      "step": 279
+    },
+    {
+      "epoch": 0.45647212259537007,
+      "grad_norm": 0.17980937659740448,
+      "learning_rate": 0.0001,
+      "loss": 0.2227,
+      "step": 280
+    },
+    {
+      "epoch": 0.4581023801760678,
+      "grad_norm": 0.1296575367450714,
+      "learning_rate": 0.0001,
+      "loss": 0.233,
+      "step": 281
+    },
+    {
+      "epoch": 0.4597326377567656,
+      "grad_norm": 0.14362689852714539,
+      "learning_rate": 0.0001,
+      "loss": 0.2299,
+      "step": 282
+    },
+    {
+      "epoch": 0.46136289533746333,
+      "grad_norm": 0.13908450305461884,
+      "learning_rate": 0.0001,
+      "loss": 0.2261,
+      "step": 283
+    },
+    {
+      "epoch": 0.46299315291816107,
+      "grad_norm": 0.13326814770698547,
+      "learning_rate": 0.0001,
+      "loss": 0.2254,
+      "step": 284
+    },
+    {
+      "epoch": 0.4646234104988588,
+      "grad_norm": 0.1157006099820137,
+      "learning_rate": 0.0001,
+      "loss": 0.2255,
+      "step": 285
+    },
+    {
+      "epoch": 0.4662536680795566,
+      "grad_norm": 0.14651785790920258,
+      "learning_rate": 0.0001,
+      "loss": 0.2243,
+      "step": 286
+    },
+    {
+      "epoch": 0.46788392566025433,
+      "grad_norm": 0.12821364402770996,
+      "learning_rate": 0.0001,
+      "loss": 0.2324,
+      "step": 287
+    },
+    {
+      "epoch": 0.46951418324095207,
+      "grad_norm": 0.13196085393428802,
+      "learning_rate": 0.0001,
+      "loss": 0.2286,
+      "step": 288
+    },
+    {
+      "epoch": 0.4711444408216498,
+      "grad_norm": 0.1328345388174057,
+      "learning_rate": 0.0001,
+      "loss": 0.2309,
+      "step": 289
+    },
+    {
+      "epoch": 0.4727746984023476,
+      "grad_norm": 0.12376872450113297,
+      "learning_rate": 0.0001,
+      "loss": 0.2434,
+      "step": 290
+    },
+    {
+      "epoch": 0.47440495598304533,
+      "grad_norm": 0.11294668167829514,
+      "learning_rate": 0.0001,
+      "loss": 0.1987,
+      "step": 291
+    },
+    {
+      "epoch": 0.47603521356374306,
+      "grad_norm": 0.11866224557161331,
+      "learning_rate": 0.0001,
+      "loss": 0.2086,
+      "step": 292
+    },
+    {
+      "epoch": 0.4776654711444408,
+      "grad_norm": 0.11296320706605911,
+      "learning_rate": 0.0001,
+      "loss": 0.2217,
+      "step": 293
+    },
+    {
+      "epoch": 0.4792957287251386,
+      "grad_norm": 0.1112779900431633,
+      "learning_rate": 0.0001,
+      "loss": 0.2112,
+      "step": 294
+    },
+    {
+      "epoch": 0.48092598630583633,
+      "grad_norm": 0.11955477297306061,
+      "learning_rate": 0.0001,
+      "loss": 0.2377,
+      "step": 295
+    },
+    {
+      "epoch": 0.48255624388653406,
+      "grad_norm": 0.11908381432294846,
+      "learning_rate": 0.0001,
+      "loss": 0.2353,
+      "step": 296
+    },
+    {
+      "epoch": 0.4841865014672318,
+      "grad_norm": 0.10947412997484207,
+      "learning_rate": 0.0001,
+      "loss": 0.2351,
+      "step": 297
+    },
+    {
+      "epoch": 0.4858167590479296,
+      "grad_norm": 0.19641439616680145,
+      "learning_rate": 0.0001,
+      "loss": 0.2293,
+      "step": 298
+    },
+    {
+      "epoch": 0.4874470166286273,
+      "grad_norm": 0.11824516952037811,
+      "learning_rate": 0.0001,
+      "loss": 0.2332,
+      "step": 299
+    },
+    {
+      "epoch": 0.48907727420932506,
+      "grad_norm": 0.11680760979652405,
+      "learning_rate": 0.0001,
+      "loss": 0.2469,
+      "step": 300
+    },
+    {
+      "epoch": 0.4907075317900228,
+      "grad_norm": 0.12153971940279007,
+      "learning_rate": 0.0001,
+      "loss": 0.2228,
+      "step": 301
+    },
+    {
+      "epoch": 0.4923377893707206,
+      "grad_norm": 0.12106509506702423,
+      "learning_rate": 0.0001,
+      "loss": 0.2375,
+      "step": 302
+    },
+    {
+      "epoch": 0.4939680469514183,
+      "grad_norm": 0.1114591732621193,
+      "learning_rate": 0.0001,
+      "loss": 0.2175,
+      "step": 303
+    },
+    {
+      "epoch": 0.49559830453211606,
+      "grad_norm": 0.10288353264331818,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 304
+    },
+    {
+      "epoch": 0.4972285621128138,
+      "grad_norm": 0.12334591895341873,
+      "learning_rate": 0.0001,
+      "loss": 0.2355,
+      "step": 305
+    },
+    {
+      "epoch": 0.4988588196935116,
+      "grad_norm": 0.12344373017549515,
+      "learning_rate": 0.0001,
+      "loss": 0.2278,
+      "step": 306
+    },
+    {
+      "epoch": 0.5004890772742093,
+      "grad_norm": 0.11695542931556702,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 307
+    },
+    {
+      "epoch": 0.5021193348549071,
+      "grad_norm": 0.1250382661819458,
+      "learning_rate": 0.0001,
+      "loss": 0.2299,
+      "step": 308
+    },
+    {
+      "epoch": 0.5037495924356048,
+      "grad_norm": 0.10904502868652344,
+      "learning_rate": 0.0001,
+      "loss": 0.2178,
+      "step": 309
+    },
+    {
+      "epoch": 0.5053798500163026,
+      "grad_norm": 0.12705446779727936,
+      "learning_rate": 0.0001,
+      "loss": 0.2278,
+      "step": 310
+    },
+    {
+      "epoch": 0.5070101075970004,
+      "grad_norm": 0.12775245308876038,
+      "learning_rate": 0.0001,
+      "loss": 0.2344,
+      "step": 311
+    },
+    {
+      "epoch": 0.5086403651776981,
+      "grad_norm": 0.1119925007224083,
+      "learning_rate": 0.0001,
+      "loss": 0.2176,
+      "step": 312
+    },
+    {
+      "epoch": 0.5102706227583959,
+      "grad_norm": 0.13966351747512817,
+      "learning_rate": 0.0001,
+      "loss": 0.2242,
+      "step": 313
+    },
+    {
+      "epoch": 0.5119008803390935,
+      "grad_norm": 0.11344596743583679,
+      "learning_rate": 0.0001,
+      "loss": 0.233,
+      "step": 314
+    },
+    {
+      "epoch": 0.5135311379197913,
+      "grad_norm": 0.12142585963010788,
+      "learning_rate": 0.0001,
+      "loss": 0.2279,
+      "step": 315
+    },
+    {
+      "epoch": 0.5151613955004891,
+      "grad_norm": 0.11385513842105865,
+      "learning_rate": 0.0001,
+      "loss": 0.2197,
+      "step": 316
+    },
+    {
+      "epoch": 0.5167916530811868,
+      "grad_norm": 0.10977032035589218,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 317
+    },
+    {
+      "epoch": 0.5184219106618846,
+      "grad_norm": 0.11549077928066254,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 318
+    },
+    {
+      "epoch": 0.5200521682425824,
+      "grad_norm": 0.13420367240905762,
+      "learning_rate": 0.0001,
+      "loss": 0.2214,
+      "step": 319
+    },
+    {
+      "epoch": 0.5216824258232801,
+      "grad_norm": 0.12192299962043762,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 320
+    },
+    {
+      "epoch": 0.5233126834039779,
+      "grad_norm": 0.1398283839225769,
+      "learning_rate": 0.0001,
+      "loss": 0.2213,
+      "step": 321
+    },
+    {
+      "epoch": 0.5249429409846755,
+      "grad_norm": 0.13691936433315277,
+      "learning_rate": 0.0001,
+      "loss": 0.2457,
+      "step": 322
+    },
+    {
+      "epoch": 0.5265731985653733,
+      "grad_norm": 0.14428359270095825,
+      "learning_rate": 0.0001,
+      "loss": 0.235,
+      "step": 323
+    },
+    {
+      "epoch": 0.5282034561460711,
+      "grad_norm": 0.10904989391565323,
+      "learning_rate": 0.0001,
+      "loss": 0.2124,
+      "step": 324
+    },
+    {
+      "epoch": 0.5298337137267688,
+      "grad_norm": 0.12136954814195633,
+      "learning_rate": 0.0001,
+      "loss": 0.2157,
+      "step": 325
+    },
+    {
+      "epoch": 0.5314639713074666,
+      "grad_norm": 0.11056441813707352,
+      "learning_rate": 0.0001,
+      "loss": 0.2304,
+      "step": 326
+    },
+    {
+      "epoch": 0.5330942288881644,
+      "grad_norm": 0.1466270238161087,
+      "learning_rate": 0.0001,
+      "loss": 0.2202,
+      "step": 327
+    },
+    {
+      "epoch": 0.5347244864688621,
+      "grad_norm": 0.11299032717943192,
+      "learning_rate": 0.0001,
+      "loss": 0.2241,
+      "step": 328
+    },
+    {
+      "epoch": 0.5363547440495599,
+      "grad_norm": 0.1448090374469757,
+      "learning_rate": 0.0001,
+      "loss": 0.2382,
+      "step": 329
+    },
+    {
+      "epoch": 0.5379850016302575,
+      "grad_norm": 0.146869957447052,
+      "learning_rate": 0.0001,
+      "loss": 0.2184,
+      "step": 330
+    },
+    {
+      "epoch": 0.5396152592109553,
+      "grad_norm": 0.12791307270526886,
+      "learning_rate": 0.0001,
+      "loss": 0.221,
+      "step": 331
+    },
+    {
+      "epoch": 0.5412455167916531,
+      "grad_norm": 0.13846737146377563,
+      "learning_rate": 0.0001,
+      "loss": 0.2247,
+      "step": 332
+    },
+    {
+      "epoch": 0.5428757743723508,
+      "grad_norm": 0.12737876176834106,
+      "learning_rate": 0.0001,
+      "loss": 0.2114,
+      "step": 333
+    },
+    {
+      "epoch": 0.5445060319530486,
+      "grad_norm": 0.11661113798618317,
+      "learning_rate": 0.0001,
+      "loss": 0.2291,
+      "step": 334
+    },
+    {
+      "epoch": 0.5461362895337464,
+      "grad_norm": 0.11759155988693237,
+      "learning_rate": 0.0001,
+      "loss": 0.2153,
+      "step": 335
+    },
+    {
+      "epoch": 0.5477665471144441,
+      "grad_norm": 0.11843594163656235,
+      "learning_rate": 0.0001,
+      "loss": 0.2205,
+      "step": 336
+    },
+    {
+      "epoch": 0.5493968046951418,
+      "grad_norm": 0.12133964151144028,
+      "learning_rate": 0.0001,
+      "loss": 0.2369,
+      "step": 337
+    },
+    {
+      "epoch": 0.5510270622758395,
+      "grad_norm": 0.1370978206396103,
+      "learning_rate": 0.0001,
+      "loss": 0.2233,
+      "step": 338
+    },
+    {
+      "epoch": 0.5526573198565373,
+      "grad_norm": 0.14518651366233826,
+      "learning_rate": 0.0001,
+      "loss": 0.2257,
+      "step": 339
+    },
+    {
+      "epoch": 0.5542875774372351,
+      "grad_norm": 0.12819088995456696,
+      "learning_rate": 0.0001,
+      "loss": 0.2387,
+      "step": 340
+    },
+    {
+      "epoch": 0.5559178350179328,
+      "grad_norm": 0.102646104991436,
+      "learning_rate": 0.0001,
+      "loss": 0.2262,
+      "step": 341
+    },
+    {
+      "epoch": 0.5575480925986306,
+      "grad_norm": 0.11202642321586609,
+      "learning_rate": 0.0001,
+      "loss": 0.24,
+      "step": 342
+    },
+    {
+      "epoch": 0.5591783501793284,
+      "grad_norm": 0.12525996565818787,
+      "learning_rate": 0.0001,
+      "loss": 0.2284,
+      "step": 343
+    },
+    {
+      "epoch": 0.5608086077600261,
+      "grad_norm": 0.09891889244318008,
+      "learning_rate": 0.0001,
+      "loss": 0.218,
+      "step": 344
+    },
+    {
+      "epoch": 0.5624388653407238,
+      "grad_norm": 0.11816040426492691,
+      "learning_rate": 0.0001,
+      "loss": 0.2204,
+      "step": 345
+    },
+    {
+      "epoch": 0.5640691229214216,
+      "grad_norm": 0.10832750797271729,
+      "learning_rate": 0.0001,
+      "loss": 0.2202,
+      "step": 346
+    },
+    {
+      "epoch": 0.5656993805021193,
+      "grad_norm": 0.10662806034088135,
+      "learning_rate": 0.0001,
+      "loss": 0.2239,
+      "step": 347
+    },
+    {
+      "epoch": 0.5673296380828171,
+      "grad_norm": 0.11352869868278503,
+      "learning_rate": 0.0001,
+      "loss": 0.2217,
+      "step": 348
+    },
+    {
+      "epoch": 0.5689598956635148,
+      "grad_norm": 0.10960965603590012,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 349
+    },
+    {
+      "epoch": 0.5705901532442126,
+      "grad_norm": 0.1158093810081482,
+      "learning_rate": 0.0001,
+      "loss": 0.2313,
+      "step": 350
+    },
+    {
+      "epoch": 0.5722204108249104,
+      "grad_norm": 0.09648110717535019,
+      "learning_rate": 0.0001,
+      "loss": 0.2136,
+      "step": 351
+    },
+    {
+      "epoch": 0.573850668405608,
+      "grad_norm": 0.10889653116464615,
+      "learning_rate": 0.0001,
+      "loss": 0.2079,
+      "step": 352
+    },
+    {
+      "epoch": 0.5754809259863058,
+      "grad_norm": 0.12388119101524353,
+      "learning_rate": 0.0001,
+      "loss": 0.2307,
+      "step": 353
+    },
+    {
+      "epoch": 0.5771111835670036,
+      "grad_norm": 0.11496174335479736,
+      "learning_rate": 0.0001,
+      "loss": 0.2232,
+      "step": 354
+    },
+    {
+      "epoch": 0.5787414411477013,
+      "grad_norm": 0.1000954732298851,
+      "learning_rate": 0.0001,
+      "loss": 0.2018,
+      "step": 355
+    },
+    {
+      "epoch": 0.5803716987283991,
+      "grad_norm": 0.11189039051532745,
+      "learning_rate": 0.0001,
+      "loss": 0.2044,
+      "step": 356
+    },
+    {
+      "epoch": 0.5820019563090968,
+      "grad_norm": 0.12651148438453674,
+      "learning_rate": 0.0001,
+      "loss": 0.2152,
+      "step": 357
+    },
+    {
+      "epoch": 0.5836322138897946,
+      "grad_norm": 0.12214814871549606,
+      "learning_rate": 0.0001,
+      "loss": 0.2174,
+      "step": 358
+    },
+    {
+      "epoch": 0.5852624714704924,
+      "grad_norm": 0.11027688533067703,
+      "learning_rate": 0.0001,
+      "loss": 0.219,
+      "step": 359
+    },
+    {
+      "epoch": 0.58689272905119,
+      "grad_norm": 0.12585967779159546,
+      "learning_rate": 0.0001,
+      "loss": 0.2198,
+      "step": 360
+    },
+    {
+      "epoch": 0.5885229866318878,
+      "grad_norm": 0.11782504618167877,
+      "learning_rate": 0.0001,
+      "loss": 0.2107,
+      "step": 361
+    },
+    {
+      "epoch": 0.5901532442125856,
+      "grad_norm": 0.12490631639957428,
+      "learning_rate": 0.0001,
+      "loss": 0.2342,
+      "step": 362
+    },
+    {
+      "epoch": 0.5917835017932833,
+      "grad_norm": 0.10052972286939621,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 363
+    },
+    {
+      "epoch": 0.5934137593739811,
+      "grad_norm": 0.11178889870643616,
+      "learning_rate": 0.0001,
+      "loss": 0.2152,
+      "step": 364
+    },
+    {
+      "epoch": 0.5950440169546788,
+      "grad_norm": 0.10644172877073288,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 365
+    },
+    {
+      "epoch": 0.5966742745353766,
+      "grad_norm": 0.1278517097234726,
+      "learning_rate": 0.0001,
+      "loss": 0.2059,
+      "step": 366
+    },
+    {
+      "epoch": 0.5983045321160744,
+      "grad_norm": 0.15825654566287994,
+      "learning_rate": 0.0001,
+      "loss": 0.2269,
+      "step": 367
+    },
+    {
+      "epoch": 0.599934789696772,
+      "grad_norm": 0.11004846543073654,
+      "learning_rate": 0.0001,
+      "loss": 0.2184,
+      "step": 368
+    },
+    {
+      "epoch": 0.6015650472774698,
+      "grad_norm": 0.10813707113265991,
+      "learning_rate": 0.0001,
+      "loss": 0.2287,
+      "step": 369
+    },
+    {
+      "epoch": 0.6031953048581676,
+      "grad_norm": 0.10354539006948471,
+      "learning_rate": 0.0001,
+      "loss": 0.2154,
+      "step": 370
+    },
+    {
+      "epoch": 0.6048255624388653,
+      "grad_norm": 0.11675768345594406,
+      "learning_rate": 0.0001,
+      "loss": 0.2341,
+      "step": 371
+    },
+    {
+      "epoch": 0.6064558200195631,
+      "grad_norm": 0.12332635372877121,
+      "learning_rate": 0.0001,
+      "loss": 0.2268,
+      "step": 372
+    },
+    {
+      "epoch": 0.6080860776002608,
+      "grad_norm": 0.11673764884471893,
+      "learning_rate": 0.0001,
+      "loss": 0.2371,
+      "step": 373
+    },
+    {
+      "epoch": 0.6097163351809586,
+      "grad_norm": 0.11986733227968216,
+      "learning_rate": 0.0001,
+      "loss": 0.2218,
+      "step": 374
+    },
+    {
+      "epoch": 0.6113465927616564,
+      "grad_norm": 0.12927494943141937,
+      "learning_rate": 0.0001,
+      "loss": 0.2271,
+      "step": 375
+    },
+    {
+      "epoch": 0.612976850342354,
+      "grad_norm": 0.0948595330119133,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 376
+    },
+    {
+      "epoch": 0.6146071079230518,
+      "grad_norm": 0.12510277330875397,
+      "learning_rate": 0.0001,
+      "loss": 0.226,
+      "step": 377
+    },
+    {
+      "epoch": 0.6162373655037496,
+      "grad_norm": 0.10695146024227142,
+      "learning_rate": 0.0001,
+      "loss": 0.2146,
+      "step": 378
+    },
+    {
+      "epoch": 0.6178676230844473,
+      "grad_norm": 0.1367439478635788,
+      "learning_rate": 0.0001,
+      "loss": 0.2397,
+      "step": 379
+    },
+    {
+      "epoch": 0.6194978806651451,
+      "grad_norm": 0.1275574266910553,
+      "learning_rate": 0.0001,
+      "loss": 0.2028,
+      "step": 380
+    },
+    {
+      "epoch": 0.6211281382458429,
+      "grad_norm": 0.11883818358182907,
+      "learning_rate": 0.0001,
+      "loss": 0.2277,
+      "step": 381
+    },
+    {
+      "epoch": 0.6227583958265406,
+      "grad_norm": 0.12322292476892471,
+      "learning_rate": 0.0001,
+      "loss": 0.2318,
+      "step": 382
+    },
+    {
+      "epoch": 0.6243886534072384,
+      "grad_norm": 0.11404566466808319,
+      "learning_rate": 0.0001,
+      "loss": 0.23,
+      "step": 383
+    },
+    {
+      "epoch": 0.626018910987936,
+      "grad_norm": 0.1219937726855278,
+      "learning_rate": 0.0001,
+      "loss": 0.2208,
+      "step": 384
+    },
+    {
+      "epoch": 0.6276491685686338,
+      "grad_norm": 0.11882983148097992,
+      "learning_rate": 0.0001,
+      "loss": 0.221,
+      "step": 385
+    },
+    {
+      "epoch": 0.6292794261493316,
+      "grad_norm": 0.1069592535495758,
+      "learning_rate": 0.0001,
+      "loss": 0.2239,
+      "step": 386
+    },
+    {
+      "epoch": 0.6309096837300293,
+      "grad_norm": 0.12435311079025269,
+      "learning_rate": 0.0001,
+      "loss": 0.2252,
+      "step": 387
+    },
+    {
+      "epoch": 0.6325399413107271,
+      "grad_norm": 0.11291278898715973,
+      "learning_rate": 0.0001,
+      "loss": 0.2187,
+      "step": 388
+    },
+    {
+      "epoch": 0.6341701988914249,
+      "grad_norm": 0.14565667510032654,
+      "learning_rate": 0.0001,
+      "loss": 0.2238,
+      "step": 389
+    },
+    {
+      "epoch": 0.6358004564721226,
+      "grad_norm": 0.11719778180122375,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 390
+    },
+    {
+      "epoch": 0.6374307140528204,
+      "grad_norm": 0.11908719688653946,
+      "learning_rate": 0.0001,
+      "loss": 0.2056,
+      "step": 391
+    },
+    {
+      "epoch": 0.639060971633518,
+      "grad_norm": 0.14423371851444244,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 392
+    },
+    {
+      "epoch": 0.6406912292142158,
+      "grad_norm": 0.1208101361989975,
+      "learning_rate": 0.0001,
+      "loss": 0.2229,
+      "step": 393
+    },
+    {
+      "epoch": 0.6423214867949136,
+      "grad_norm": 0.13141198456287384,
+      "learning_rate": 0.0001,
+      "loss": 0.2265,
+      "step": 394
+    },
+    {
+      "epoch": 0.6439517443756113,
+      "grad_norm": 0.1494779735803604,
+      "learning_rate": 0.0001,
+      "loss": 0.2154,
+      "step": 395
+    },
+    {
+      "epoch": 0.6455820019563091,
+      "grad_norm": 0.11805243790149689,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 396
+    },
+    {
+      "epoch": 0.6472122595370069,
+      "grad_norm": 0.13480523228645325,
+      "learning_rate": 0.0001,
+      "loss": 0.2172,
+      "step": 397
+    },
+    {
+      "epoch": 0.6488425171177046,
+      "grad_norm": 0.11627724766731262,
+      "learning_rate": 0.0001,
+      "loss": 0.2377,
+      "step": 398
+    },
+    {
+      "epoch": 0.6504727746984024,
+      "grad_norm": 0.1052049919962883,
+      "learning_rate": 0.0001,
+      "loss": 0.2203,
+      "step": 399
+    },
+    {
+      "epoch": 0.6521030322791,
+      "grad_norm": 0.12532803416252136,
+      "learning_rate": 0.0001,
+      "loss": 0.224,
+      "step": 400
+    },
+    {
+      "epoch": 0.6537332898597978,
+      "grad_norm": 0.11664199829101562,
+      "learning_rate": 0.0001,
+      "loss": 0.2354,
+      "step": 401
+    },
+    {
+      "epoch": 0.6553635474404956,
+      "grad_norm": 0.10853651911020279,
+      "learning_rate": 0.0001,
+      "loss": 0.2097,
+      "step": 402
+    },
+    {
+      "epoch": 0.6569938050211933,
+      "grad_norm": 0.10371599346399307,
+      "learning_rate": 0.0001,
+      "loss": 0.2058,
+      "step": 403
+    },
+    {
+      "epoch": 0.6586240626018911,
+      "grad_norm": 0.1397281438112259,
+      "learning_rate": 0.0001,
+      "loss": 0.2259,
+      "step": 404
+    },
+    {
+      "epoch": 0.6602543201825889,
+      "grad_norm": 0.11628320068120956,
+      "learning_rate": 0.0001,
+      "loss": 0.2183,
+      "step": 405
+    },
+    {
+      "epoch": 0.6618845777632866,
+      "grad_norm": 0.13096722960472107,
+      "learning_rate": 0.0001,
+      "loss": 0.2219,
+      "step": 406
+    },
+    {
+      "epoch": 0.6635148353439844,
+      "grad_norm": 0.11224295198917389,
+      "learning_rate": 0.0001,
+      "loss": 0.2249,
+      "step": 407
+    },
+    {
+      "epoch": 0.665145092924682,
+      "grad_norm": 0.1743100881576538,
+      "learning_rate": 0.0001,
+      "loss": 0.2356,
+      "step": 408
+    },
+    {
+      "epoch": 0.6667753505053798,
+      "grad_norm": 0.1063716933131218,
+      "learning_rate": 0.0001,
+      "loss": 0.2049,
+      "step": 409
+    },
+    {
+      "epoch": 0.6684056080860776,
+      "grad_norm": 0.12047860771417618,
+      "learning_rate": 0.0001,
+      "loss": 0.2176,
+      "step": 410
+    },
+    {
+      "epoch": 0.6700358656667753,
+      "grad_norm": 0.1088520810008049,
+      "learning_rate": 0.0001,
+      "loss": 0.2267,
+      "step": 411
+    },
+    {
+      "epoch": 0.6716661232474731,
+      "grad_norm": 0.12248346954584122,
+      "learning_rate": 0.0001,
+      "loss": 0.2098,
+      "step": 412
+    },
+    {
+      "epoch": 0.6732963808281709,
+      "grad_norm": 0.13586483895778656,
+      "learning_rate": 0.0001,
+      "loss": 0.222,
+      "step": 413
+    },
+    {
+      "epoch": 0.6749266384088686,
+      "grad_norm": 0.1342141330242157,
+      "learning_rate": 0.0001,
+      "loss": 0.2126,
+      "step": 414
+    },
+    {
+      "epoch": 0.6765568959895664,
+      "grad_norm": 0.11067284643650055,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 415
+    },
+    {
+      "epoch": 0.6781871535702642,
+      "grad_norm": 0.14228413999080658,
+      "learning_rate": 0.0001,
+      "loss": 0.2252,
+      "step": 416
+    },
+    {
+      "epoch": 0.6798174111509618,
+      "grad_norm": 0.12066151201725006,
+      "learning_rate": 0.0001,
+      "loss": 0.2163,
+      "step": 417
+    },
+    {
+      "epoch": 0.6814476687316596,
+      "grad_norm": 0.15157178044319153,
+      "learning_rate": 0.0001,
+      "loss": 0.2243,
+      "step": 418
+    },
+    {
+      "epoch": 0.6830779263123573,
+      "grad_norm": 0.12287676334381104,
+      "learning_rate": 0.0001,
+      "loss": 0.2185,
+      "step": 419
+    },
+    {
+      "epoch": 0.6847081838930551,
+      "grad_norm": 0.11439267545938492,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 420
+    },
+    {
+      "epoch": 0.6863384414737529,
+      "grad_norm": 0.10508660227060318,
+      "learning_rate": 0.0001,
+      "loss": 0.2143,
+      "step": 421
+    },
+    {
+      "epoch": 0.6879686990544506,
+      "grad_norm": 0.10114973783493042,
+      "learning_rate": 0.0001,
+      "loss": 0.2089,
+      "step": 422
+    },
+    {
+      "epoch": 0.6895989566351484,
+      "grad_norm": 0.11443648487329483,
+      "learning_rate": 0.0001,
+      "loss": 0.2205,
+      "step": 423
+    },
+    {
+      "epoch": 0.6912292142158462,
+      "grad_norm": 0.1153668612241745,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 424
+    },
+    {
+      "epoch": 0.6928594717965438,
+      "grad_norm": 0.10309400409460068,
+      "learning_rate": 0.0001,
+      "loss": 0.2197,
+      "step": 425
+    },
+    {
+      "epoch": 0.6944897293772416,
+      "grad_norm": 0.10628131031990051,
+      "learning_rate": 0.0001,
+      "loss": 0.2073,
+      "step": 426
+    },
+    {
+      "epoch": 0.6961199869579393,
+      "grad_norm": 0.1151309385895729,
+      "learning_rate": 0.0001,
+      "loss": 0.22,
+      "step": 427
+    },
+    {
+      "epoch": 0.6977502445386371,
+      "grad_norm": 0.13086111843585968,
+      "learning_rate": 0.0001,
+      "loss": 0.2192,
+      "step": 428
+    },
+    {
+      "epoch": 0.6993805021193349,
+      "grad_norm": 0.1018524318933487,
+      "learning_rate": 0.0001,
+      "loss": 0.2168,
+      "step": 429
+    },
+    {
+      "epoch": 0.7010107597000326,
+      "grad_norm": 0.1204724982380867,
+      "learning_rate": 0.0001,
+      "loss": 0.2226,
+      "step": 430
+    },
+    {
+      "epoch": 0.7026410172807304,
+      "grad_norm": 0.11043181270360947,
+      "learning_rate": 0.0001,
+      "loss": 0.2207,
+      "step": 431
+    },
+    {
+      "epoch": 0.7042712748614282,
+      "grad_norm": 0.10462189465761185,
+      "learning_rate": 0.0001,
+      "loss": 0.2173,
+      "step": 432
+    },
+    {
+      "epoch": 0.7059015324421258,
+      "grad_norm": 0.11756114661693573,
+      "learning_rate": 0.0001,
+      "loss": 0.2165,
+      "step": 433
+    },
+    {
+      "epoch": 0.7075317900228236,
+      "grad_norm": 0.11371811479330063,
+      "learning_rate": 0.0001,
+      "loss": 0.217,
+      "step": 434
+    },
+    {
+      "epoch": 0.7091620476035213,
+      "grad_norm": 0.11309879273176193,
+      "learning_rate": 0.0001,
+      "loss": 0.2046,
+      "step": 435
+    },
+    {
+      "epoch": 0.7107923051842191,
+      "grad_norm": 0.11104720830917358,
+      "learning_rate": 0.0001,
+      "loss": 0.1995,
+      "step": 436
+    },
+    {
+      "epoch": 0.7124225627649169,
+      "grad_norm": 0.10949276387691498,
+      "learning_rate": 0.0001,
+      "loss": 0.21,
+      "step": 437
+    },
+    {
+      "epoch": 0.7140528203456146,
+      "grad_norm": 0.13036222755908966,
+      "learning_rate": 0.0001,
+      "loss": 0.2187,
+      "step": 438
+    },
+    {
+      "epoch": 0.7156830779263124,
+      "grad_norm": 0.11818300932645798,
+      "learning_rate": 0.0001,
+      "loss": 0.2164,
+      "step": 439
+    },
+    {
+      "epoch": 0.7173133355070102,
+      "grad_norm": 0.11692320555448532,
+      "learning_rate": 0.0001,
+      "loss": 0.2113,
+      "step": 440
+    },
+    {
+      "epoch": 0.7189435930877078,
+      "grad_norm": 0.11360203474760056,
+      "learning_rate": 0.0001,
+      "loss": 0.2244,
+      "step": 441
+    },
+    {
+      "epoch": 0.7205738506684056,
+      "grad_norm": 0.11834365874528885,
+      "learning_rate": 0.0001,
+      "loss": 0.2303,
+      "step": 442
+    },
+    {
+      "epoch": 0.7222041082491033,
+      "grad_norm": 0.13222841918468475,
+      "learning_rate": 0.0001,
+      "loss": 0.2192,
+      "step": 443
+    },
+    {
+      "epoch": 0.7238343658298011,
+      "grad_norm": 0.12007411569356918,
+      "learning_rate": 0.0001,
+      "loss": 0.2158,
+      "step": 444
+    },
+    {
+      "epoch": 0.7254646234104989,
+      "grad_norm": 0.11357498168945312,
+      "learning_rate": 0.0001,
+      "loss": 0.2222,
+      "step": 445
+    },
+    {
+      "epoch": 0.7270948809911966,
+      "grad_norm": 0.14090843498706818,
+      "learning_rate": 0.0001,
+      "loss": 0.2173,
+      "step": 446
+    },
+    {
+      "epoch": 0.7287251385718944,
+      "grad_norm": 0.13058516383171082,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 447
+    },
+    {
+      "epoch": 0.7303553961525922,
+      "grad_norm": 0.12427127361297607,
+      "learning_rate": 0.0001,
+      "loss": 0.2235,
+      "step": 448
+    },
+    {
+      "epoch": 0.7319856537332898,
+      "grad_norm": 0.10180344432592392,
+      "learning_rate": 0.0001,
+      "loss": 0.2056,
+      "step": 449
+    },
+    {
+      "epoch": 0.7336159113139876,
+      "grad_norm": 0.11706260591745377,
+      "learning_rate": 0.0001,
+      "loss": 0.2178,
+      "step": 450
+    },
+    {
+      "epoch": 0.7352461688946854,
+      "grad_norm": 0.14062075316905975,
+      "learning_rate": 0.0001,
+      "loss": 0.2296,
+      "step": 451
+    },
+    {
+      "epoch": 0.7368764264753831,
+      "grad_norm": 0.1115693598985672,
+      "learning_rate": 0.0001,
+      "loss": 0.2348,
+      "step": 452
+    },
+    {
+      "epoch": 0.7385066840560809,
+      "grad_norm": 0.1121508777141571,
+      "learning_rate": 0.0001,
+      "loss": 0.217,
+      "step": 453
+    },
+    {
+      "epoch": 0.7401369416367786,
+      "grad_norm": 0.12056824564933777,
+      "learning_rate": 0.0001,
+      "loss": 0.2225,
+      "step": 454
+    },
+    {
+      "epoch": 0.7417671992174764,
+      "grad_norm": 0.12506930530071259,
+      "learning_rate": 0.0001,
+      "loss": 0.2215,
+      "step": 455
+    },
+    {
+      "epoch": 0.7433974567981741,
+      "grad_norm": 0.1259409487247467,
+      "learning_rate": 0.0001,
+      "loss": 0.2015,
+      "step": 456
+    },
+    {
+      "epoch": 0.7450277143788718,
+      "grad_norm": 0.13509246706962585,
+      "learning_rate": 0.0001,
+      "loss": 0.2271,
+      "step": 457
+    },
+    {
+      "epoch": 0.7466579719595696,
+      "grad_norm": 0.13348042964935303,
+      "learning_rate": 0.0001,
+      "loss": 0.2227,
+      "step": 458
+    },
+    {
+      "epoch": 0.7482882295402674,
+      "grad_norm": 0.12885409593582153,
+      "learning_rate": 0.0001,
+      "loss": 0.2144,
+      "step": 459
+    },
+    {
+      "epoch": 0.7499184871209651,
+      "grad_norm": 0.13527405261993408,
+      "learning_rate": 0.0001,
+      "loss": 0.2349,
+      "step": 460
+    },
+    {
+      "epoch": 0.7515487447016629,
+      "grad_norm": 0.15392392873764038,
+      "learning_rate": 0.0001,
+      "loss": 0.2258,
+      "step": 461
+    },
+    {
+      "epoch": 0.7531790022823606,
+      "grad_norm": 0.11966606229543686,
+      "learning_rate": 0.0001,
+      "loss": 0.2179,
+      "step": 462
+    },
+    {
+      "epoch": 0.7548092598630584,
+      "grad_norm": 0.10815181583166122,
+      "learning_rate": 0.0001,
+      "loss": 0.2222,
+      "step": 463
+    },
+    {
+      "epoch": 0.7564395174437561,
+      "grad_norm": 0.11875923722982407,
+      "learning_rate": 0.0001,
+      "loss": 0.2169,
+      "step": 464
+    },
+    {
+      "epoch": 0.7580697750244538,
+      "grad_norm": 0.11258353292942047,
+      "learning_rate": 0.0001,
+      "loss": 0.2024,
+      "step": 465
+    },
+    {
+      "epoch": 0.7597000326051516,
+      "grad_norm": 0.1222681850194931,
+      "learning_rate": 0.0001,
+      "loss": 0.2277,
+      "step": 466
+    },
+    {
+      "epoch": 0.7613302901858494,
+      "grad_norm": 0.11641412228345871,
+      "learning_rate": 0.0001,
+      "loss": 0.2178,
+      "step": 467
+    },
+    {
+      "epoch": 0.7629605477665471,
+      "grad_norm": 0.12195785343647003,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 468
+    },
+    {
+      "epoch": 0.7645908053472449,
+      "grad_norm": 0.10462360829114914,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 469
+    },
+    {
+      "epoch": 0.7662210629279426,
+      "grad_norm": 0.11324596405029297,
+      "learning_rate": 0.0001,
+      "loss": 0.2238,
+      "step": 470
+    },
+    {
+      "epoch": 0.7678513205086404,
+      "grad_norm": 0.12687797844409943,
+      "learning_rate": 0.0001,
+      "loss": 0.231,
+      "step": 471
+    },
+    {
+      "epoch": 0.7694815780893381,
+      "grad_norm": 0.12114600092172623,
+      "learning_rate": 0.0001,
+      "loss": 0.2217,
+      "step": 472
+    },
+    {
+      "epoch": 0.7711118356700358,
+      "grad_norm": 0.12222952395677567,
+      "learning_rate": 0.0001,
+      "loss": 0.2238,
+      "step": 473
+    },
+    {
+      "epoch": 0.7727420932507336,
+      "grad_norm": 0.11903360486030579,
+      "learning_rate": 0.0001,
+      "loss": 0.2079,
+      "step": 474
+    },
+    {
+      "epoch": 0.7743723508314314,
+      "grad_norm": 0.12019330263137817,
+      "learning_rate": 0.0001,
+      "loss": 0.2074,
+      "step": 475
+    },
+    {
+      "epoch": 0.7760026084121291,
+      "grad_norm": 0.1122792437672615,
+      "learning_rate": 0.0001,
+      "loss": 0.2086,
+      "step": 476
+    },
+    {
+      "epoch": 0.7776328659928269,
+      "grad_norm": 0.11570253223180771,
+      "learning_rate": 0.0001,
+      "loss": 0.2172,
+      "step": 477
+    },
+    {
+      "epoch": 0.7792631235735246,
+      "grad_norm": 0.14799444377422333,
+      "learning_rate": 0.0001,
+      "loss": 0.2163,
+      "step": 478
+    },
+    {
+      "epoch": 0.7808933811542224,
+      "grad_norm": 0.09483097493648529,
+      "learning_rate": 0.0001,
+      "loss": 0.1999,
+      "step": 479
+    },
+    {
+      "epoch": 0.7825236387349201,
+      "grad_norm": 0.11984442174434662,
+      "learning_rate": 0.0001,
+      "loss": 0.2017,
+      "step": 480
+    },
+    {
+      "epoch": 0.7841538963156178,
+      "grad_norm": 0.11621296405792236,
+      "learning_rate": 0.0001,
+      "loss": 0.2149,
+      "step": 481
+    },
+    {
+      "epoch": 0.7857841538963156,
+      "grad_norm": 0.0937175378203392,
+      "learning_rate": 0.0001,
+      "loss": 0.2032,
+      "step": 482
+    },
+    {
+      "epoch": 0.7874144114770134,
+      "grad_norm": 0.12285979092121124,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 483
+    },
+    {
+      "epoch": 0.7890446690577111,
+      "grad_norm": 0.13726738095283508,
+      "learning_rate": 0.0001,
+      "loss": 0.2189,
+      "step": 484
+    },
+    {
+      "epoch": 0.7906749266384089,
+      "grad_norm": 0.1135246530175209,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 485
+    },
+    {
+      "epoch": 0.7923051842191067,
+      "grad_norm": 0.11096853017807007,
+      "learning_rate": 0.0001,
+      "loss": 0.1984,
+      "step": 486
+    },
+    {
+      "epoch": 0.7939354417998044,
+      "grad_norm": 0.10587968677282333,
+      "learning_rate": 0.0001,
+      "loss": 0.2213,
+      "step": 487
+    },
+    {
+      "epoch": 0.7955656993805021,
+      "grad_norm": 0.11208285391330719,
+      "learning_rate": 0.0001,
+      "loss": 0.2104,
+      "step": 488
+    },
+    {
+      "epoch": 0.7971959569611998,
+      "grad_norm": 0.12265545129776001,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 489
+    },
+    {
+      "epoch": 0.7988262145418976,
+      "grad_norm": 0.09317664057016373,
+      "learning_rate": 0.0001,
+      "loss": 0.1927,
+      "step": 490
+    },
+    {
+      "epoch": 0.8004564721225954,
+      "grad_norm": 0.10856229066848755,
+      "learning_rate": 0.0001,
+      "loss": 0.2024,
+      "step": 491
+    },
+    {
+      "epoch": 0.8020867297032931,
+      "grad_norm": 0.09949705749750137,
+      "learning_rate": 0.0001,
+      "loss": 0.2015,
+      "step": 492
+    },
+    {
+      "epoch": 0.8037169872839909,
+      "grad_norm": 0.13871848583221436,
+      "learning_rate": 0.0001,
+      "loss": 0.2161,
+      "step": 493
+    },
+    {
+      "epoch": 0.8053472448646887,
+      "grad_norm": 0.11129366606473923,
+      "learning_rate": 0.0001,
+      "loss": 0.2064,
+      "step": 494
+    },
+    {
+      "epoch": 0.8069775024453864,
+      "grad_norm": 0.12229236960411072,
+      "learning_rate": 0.0001,
+      "loss": 0.2145,
+      "step": 495
+    },
+    {
+      "epoch": 0.8086077600260841,
+      "grad_norm": 0.12729904055595398,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 496
+    },
+    {
+      "epoch": 0.8102380176067818,
+      "grad_norm": 0.13496865332126617,
+      "learning_rate": 0.0001,
+      "loss": 0.2275,
+      "step": 497
+    },
+    {
+      "epoch": 0.8118682751874796,
+      "grad_norm": 0.11651238799095154,
+      "learning_rate": 0.0001,
+      "loss": 0.2113,
+      "step": 498
+    },
+    {
+      "epoch": 0.8134985327681774,
+      "grad_norm": 0.14498771727085114,
+      "learning_rate": 0.0001,
+      "loss": 0.2324,
+      "step": 499
+    },
+    {
+      "epoch": 0.8151287903488751,
+      "grad_norm": 0.13545864820480347,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 500
+    },
+    {
+      "epoch": 0.8151287903488751,
+      "eval_loss": 0.21210059523582458,
+      "eval_runtime": 469.38,
+      "eval_samples_per_second": 4.02,
+      "eval_steps_per_second": 1.006,
+      "step": 500
+    },
+    {
+      "epoch": 0.8167590479295729,
+      "grad_norm": 0.1212705671787262,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 501
+    },
+    {
+      "epoch": 0.8183893055102707,
+      "grad_norm": 0.12040545791387558,
+      "learning_rate": 0.0001,
+      "loss": 0.2149,
+      "step": 502
+    },
+    {
+      "epoch": 0.8200195630909684,
+      "grad_norm": 0.10691879689693451,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 503
+    },
+    {
+      "epoch": 0.8216498206716661,
+      "grad_norm": 0.11082441359758377,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 504
+    },
+    {
+      "epoch": 0.8232800782523638,
+      "grad_norm": 0.12334942817687988,
+      "learning_rate": 0.0001,
+      "loss": 0.202,
+      "step": 505
+    },
+    {
+      "epoch": 0.8249103358330616,
+      "grad_norm": 0.12807224690914154,
+      "learning_rate": 0.0001,
+      "loss": 0.1993,
+      "step": 506
+    },
+    {
+      "epoch": 0.8265405934137594,
+      "grad_norm": 0.11735547333955765,
+      "learning_rate": 0.0001,
+      "loss": 0.23,
+      "step": 507
+    },
+    {
+      "epoch": 0.8281708509944571,
+      "grad_norm": 0.12395918369293213,
+      "learning_rate": 0.0001,
+      "loss": 0.2266,
+      "step": 508
+    },
+    {
+      "epoch": 0.8298011085751549,
+      "grad_norm": 0.1415017992258072,
+      "learning_rate": 0.0001,
+      "loss": 0.2122,
+      "step": 509
+    },
+    {
+      "epoch": 0.8314313661558527,
+      "grad_norm": 0.11149226874113083,
+      "learning_rate": 0.0001,
+      "loss": 0.2073,
+      "step": 510
+    },
+    {
+      "epoch": 0.8330616237365503,
+      "grad_norm": 0.12386713922023773,
+      "learning_rate": 0.0001,
+      "loss": 0.2291,
+      "step": 511
+    },
+    {
+      "epoch": 0.8346918813172481,
+      "grad_norm": 0.10768789052963257,
+      "learning_rate": 0.0001,
+      "loss": 0.2047,
+      "step": 512
+    },
+    {
+      "epoch": 0.8363221388979458,
+      "grad_norm": 0.11108648031949997,
+      "learning_rate": 0.0001,
+      "loss": 0.2126,
+      "step": 513
+    },
+    {
+      "epoch": 0.8379523964786436,
+      "grad_norm": 0.11573906987905502,
+      "learning_rate": 0.0001,
+      "loss": 0.2045,
+      "step": 514
+    },
+    {
+      "epoch": 0.8395826540593414,
+      "grad_norm": 0.09793173521757126,
+      "learning_rate": 0.0001,
+      "loss": 0.1954,
+      "step": 515
+    },
+    {
+      "epoch": 0.8412129116400391,
+      "grad_norm": 0.12155526131391525,
+      "learning_rate": 0.0001,
+      "loss": 0.2151,
+      "step": 516
+    },
+    {
+      "epoch": 0.8428431692207369,
+      "grad_norm": 0.1365749090909958,
+      "learning_rate": 0.0001,
+      "loss": 0.2231,
+      "step": 517
+    },
+    {
+      "epoch": 0.8444734268014347,
+      "grad_norm": 0.10344486683607101,
+      "learning_rate": 0.0001,
+      "loss": 0.1829,
+      "step": 518
+    },
+    {
+      "epoch": 0.8461036843821323,
+      "grad_norm": 0.11081143468618393,
+      "learning_rate": 0.0001,
+      "loss": 0.2193,
+      "step": 519
+    },
+    {
+      "epoch": 0.8477339419628301,
+      "grad_norm": 0.11781128495931625,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 520
+    },
+    {
+      "epoch": 0.8493641995435279,
+      "grad_norm": 0.1616438925266266,
+      "learning_rate": 0.0001,
+      "loss": 0.2099,
+      "step": 521
+    },
+    {
+      "epoch": 0.8509944571242256,
+      "grad_norm": 0.12249930948019028,
+      "learning_rate": 0.0001,
+      "loss": 0.2336,
+      "step": 522
+    },
+    {
+      "epoch": 0.8526247147049234,
+      "grad_norm": 0.11882970482110977,
+      "learning_rate": 0.0001,
+      "loss": 0.2219,
+      "step": 523
+    },
+    {
+      "epoch": 0.8542549722856211,
+      "grad_norm": 0.10900752991437912,
+      "learning_rate": 0.0001,
+      "loss": 0.2135,
+      "step": 524
+    },
+    {
+      "epoch": 0.8558852298663189,
+      "grad_norm": 0.11639473587274551,
+      "learning_rate": 0.0001,
+      "loss": 0.2128,
+      "step": 525
+    },
+    {
+      "epoch": 0.8575154874470167,
+      "grad_norm": 0.1093585193157196,
+      "learning_rate": 0.0001,
+      "loss": 0.2132,
+      "step": 526
+    },
+    {
+      "epoch": 0.8591457450277143,
+      "grad_norm": 0.1086277961730957,
+      "learning_rate": 0.0001,
+      "loss": 0.228,
+      "step": 527
+    },
+    {
+      "epoch": 0.8607760026084121,
+      "grad_norm": 0.10523232817649841,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 528
+    },
+    {
+      "epoch": 0.8624062601891099,
+      "grad_norm": 0.10368810594081879,
+      "learning_rate": 0.0001,
+      "loss": 0.2233,
+      "step": 529
+    },
+    {
+      "epoch": 0.8640365177698076,
+      "grad_norm": 0.09829546511173248,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 530
+    },
+    {
+      "epoch": 0.8656667753505054,
+      "grad_norm": 0.12013105303049088,
+      "learning_rate": 0.0001,
+      "loss": 0.223,
+      "step": 531
+    },
+    {
+      "epoch": 0.8672970329312031,
+      "grad_norm": 0.13377144932746887,
+      "learning_rate": 0.0001,
+      "loss": 0.2159,
+      "step": 532
+    },
+    {
+      "epoch": 0.8689272905119009,
+      "grad_norm": 0.11236433684825897,
+      "learning_rate": 0.0001,
+      "loss": 0.2165,
+      "step": 533
+    },
+    {
+      "epoch": 0.8705575480925987,
+      "grad_norm": 0.11153242737054825,
+      "learning_rate": 0.0001,
+      "loss": 0.2063,
+      "step": 534
+    },
+    {
+      "epoch": 0.8721878056732963,
+      "grad_norm": 0.13448600471019745,
+      "learning_rate": 0.0001,
+      "loss": 0.2289,
+      "step": 535
+    },
+    {
+      "epoch": 0.8738180632539941,
+      "grad_norm": 0.11363071203231812,
+      "learning_rate": 0.0001,
+      "loss": 0.2285,
+      "step": 536
+    },
+    {
+      "epoch": 0.8754483208346919,
+      "grad_norm": 0.13538241386413574,
+      "learning_rate": 0.0001,
+      "loss": 0.2052,
+      "step": 537
+    },
+    {
+      "epoch": 0.8770785784153896,
+      "grad_norm": 0.11804122477769852,
+      "learning_rate": 0.0001,
+      "loss": 0.1973,
+      "step": 538
+    },
+    {
+      "epoch": 0.8787088359960874,
+      "grad_norm": 0.13027773797512054,
+      "learning_rate": 0.0001,
+      "loss": 0.2136,
+      "step": 539
+    },
+    {
+      "epoch": 0.8803390935767851,
+      "grad_norm": 0.10185594856739044,
+      "learning_rate": 0.0001,
+      "loss": 0.2155,
+      "step": 540
+    },
+    {
+      "epoch": 0.8819693511574829,
+      "grad_norm": 0.14260874688625336,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 541
+    },
+    {
+      "epoch": 0.8835996087381807,
+      "grad_norm": 0.11737281084060669,
+      "learning_rate": 0.0001,
+      "loss": 0.2172,
+      "step": 542
+    },
+    {
+      "epoch": 0.8852298663188783,
+      "grad_norm": 0.12430301308631897,
+      "learning_rate": 0.0001,
+      "loss": 0.2108,
+      "step": 543
+    },
+    {
+      "epoch": 0.8868601238995761,
+      "grad_norm": 0.1229623332619667,
+      "learning_rate": 0.0001,
+      "loss": 0.214,
+      "step": 544
+    },
+    {
+      "epoch": 0.8884903814802739,
+      "grad_norm": 0.1039818823337555,
+      "learning_rate": 0.0001,
+      "loss": 0.2094,
+      "step": 545
+    },
+    {
+      "epoch": 0.8901206390609716,
+      "grad_norm": 0.11727850139141083,
+      "learning_rate": 0.0001,
+      "loss": 0.1965,
+      "step": 546
+    },
+    {
+      "epoch": 0.8917508966416694,
+      "grad_norm": 0.11485180258750916,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 547
+    },
+    {
+      "epoch": 0.8933811542223671,
+      "grad_norm": 0.13651061058044434,
+      "learning_rate": 0.0001,
+      "loss": 0.224,
+      "step": 548
+    },
+    {
+      "epoch": 0.8950114118030649,
+      "grad_norm": 0.11545626074075699,
+      "learning_rate": 0.0001,
+      "loss": 0.219,
+      "step": 549
+    },
+    {
+      "epoch": 0.8966416693837627,
+      "grad_norm": 0.10819267481565475,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 550
+    },
+    {
+      "epoch": 0.8982719269644603,
+      "grad_norm": 0.12021040916442871,
+      "learning_rate": 0.0001,
+      "loss": 0.2109,
+      "step": 551
+    },
+    {
+      "epoch": 0.8999021845451581,
+      "grad_norm": 0.10751428455114365,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 552
+    },
+    {
+      "epoch": 0.9015324421258559,
+      "grad_norm": 0.11271633207798004,
+      "learning_rate": 0.0001,
+      "loss": 0.2074,
+      "step": 553
+    },
+    {
+      "epoch": 0.9031626997065536,
+      "grad_norm": 0.11875735968351364,
+      "learning_rate": 0.0001,
+      "loss": 0.2087,
+      "step": 554
+    },
+    {
+      "epoch": 0.9047929572872514,
+      "grad_norm": 0.1122482642531395,
+      "learning_rate": 0.0001,
+      "loss": 0.2114,
+      "step": 555
+    },
+    {
+      "epoch": 0.9064232148679492,
+      "grad_norm": 0.13074877858161926,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 556
+    },
+    {
+      "epoch": 0.9080534724486469,
+      "grad_norm": 0.10370821505784988,
+      "learning_rate": 0.0001,
+      "loss": 0.1987,
+      "step": 557
+    },
+    {
+      "epoch": 0.9096837300293447,
+      "grad_norm": 0.11413192003965378,
+      "learning_rate": 0.0001,
+      "loss": 0.1996,
+      "step": 558
+    },
+    {
+      "epoch": 0.9113139876100423,
+      "grad_norm": 0.11211336404085159,
+      "learning_rate": 0.0001,
+      "loss": 0.2042,
+      "step": 559
+    },
+    {
+      "epoch": 0.9129442451907401,
+      "grad_norm": 0.11445408314466476,
+      "learning_rate": 0.0001,
+      "loss": 0.2132,
+      "step": 560
+    },
+    {
+      "epoch": 0.9145745027714379,
+      "grad_norm": 0.11431146413087845,
+      "learning_rate": 0.0001,
+      "loss": 0.2208,
+      "step": 561
+    },
+    {
+      "epoch": 0.9162047603521356,
+      "grad_norm": 0.12157010287046432,
+      "learning_rate": 0.0001,
+      "loss": 0.1936,
+      "step": 562
+    },
+    {
+      "epoch": 0.9178350179328334,
+      "grad_norm": 0.10648233443498611,
+      "learning_rate": 0.0001,
+      "loss": 0.2021,
+      "step": 563
+    },
+    {
+      "epoch": 0.9194652755135312,
+      "grad_norm": 0.1113019660115242,
+      "learning_rate": 0.0001,
+      "loss": 0.1996,
+      "step": 564
+    },
+    {
+      "epoch": 0.9210955330942289,
+      "grad_norm": 0.1145205944776535,
+      "learning_rate": 0.0001,
+      "loss": 0.2106,
+      "step": 565
+    },
+    {
+      "epoch": 0.9227257906749267,
+      "grad_norm": 0.11848907917737961,
+      "learning_rate": 0.0001,
+      "loss": 0.2165,
+      "step": 566
+    },
+    {
+      "epoch": 0.9243560482556243,
+      "grad_norm": 0.10608423501253128,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 567
+    },
+    {
+      "epoch": 0.9259863058363221,
+      "grad_norm": 0.1123075932264328,
+      "learning_rate": 0.0001,
+      "loss": 0.2138,
+      "step": 568
+    },
+    {
+      "epoch": 0.9276165634170199,
+      "grad_norm": 0.11272522807121277,
+      "learning_rate": 0.0001,
+      "loss": 0.2124,
+      "step": 569
+    },
+    {
+      "epoch": 0.9292468209977176,
+      "grad_norm": 0.13107183575630188,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 570
+    },
+    {
+      "epoch": 0.9308770785784154,
+      "grad_norm": 0.11995590478181839,
+      "learning_rate": 0.0001,
+      "loss": 0.2188,
+      "step": 571
+    },
+    {
+      "epoch": 0.9325073361591132,
+      "grad_norm": 0.5125681161880493,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 572
+    },
+    {
+      "epoch": 0.9341375937398109,
+      "grad_norm": 0.15056641399860382,
+      "learning_rate": 0.0001,
+      "loss": 0.216,
+      "step": 573
+    },
+    {
+      "epoch": 0.9357678513205087,
+      "grad_norm": 0.13608971238136292,
+      "learning_rate": 0.0001,
+      "loss": 0.2072,
+      "step": 574
+    },
+    {
+      "epoch": 0.9373981089012063,
+      "grad_norm": 0.11466903239488602,
+      "learning_rate": 0.0001,
+      "loss": 0.2098,
+      "step": 575
+    },
+    {
+      "epoch": 0.9390283664819041,
+      "grad_norm": 0.12261076271533966,
+      "learning_rate": 0.0001,
+      "loss": 0.1975,
+      "step": 576
+    },
+    {
+      "epoch": 0.9406586240626019,
+      "grad_norm": 0.11684715747833252,
+      "learning_rate": 0.0001,
+      "loss": 0.2048,
+      "step": 577
+    },
+    {
+      "epoch": 0.9422888816432996,
+      "grad_norm": 0.11564768105745316,
+      "learning_rate": 0.0001,
+      "loss": 0.2082,
+      "step": 578
+    },
+    {
+      "epoch": 0.9439191392239974,
+      "grad_norm": 0.10801440477371216,
+      "learning_rate": 0.0001,
+      "loss": 0.2089,
+      "step": 579
+    },
+    {
+      "epoch": 0.9455493968046952,
+      "grad_norm": 0.11598918586969376,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 580
+    },
+    {
+      "epoch": 0.9471796543853929,
+      "grad_norm": 0.12915532290935516,
+      "learning_rate": 0.0001,
+      "loss": 0.2011,
+      "step": 581
+    },
+    {
+      "epoch": 0.9488099119660907,
+      "grad_norm": 0.14291346073150635,
+      "learning_rate": 0.0001,
+      "loss": 0.2236,
+      "step": 582
+    },
+    {
+      "epoch": 0.9504401695467883,
+      "grad_norm": 0.11026699841022491,
+      "learning_rate": 0.0001,
+      "loss": 0.2129,
+      "step": 583
+    },
+    {
+      "epoch": 0.9520704271274861,
+      "grad_norm": 0.14368709921836853,
+      "learning_rate": 0.0001,
+      "loss": 0.2064,
+      "step": 584
+    },
+    {
+      "epoch": 0.9537006847081839,
+      "grad_norm": 0.10485952347517014,
+      "learning_rate": 0.0001,
+      "loss": 0.2016,
+      "step": 585
+    },
+    {
+      "epoch": 0.9553309422888816,
+      "grad_norm": 0.1238325983285904,
+      "learning_rate": 0.0001,
+      "loss": 0.2199,
+      "step": 586
+    },
+    {
+      "epoch": 0.9569611998695794,
+      "grad_norm": 0.11727755516767502,
+      "learning_rate": 0.0001,
+      "loss": 0.2069,
+      "step": 587
+    },
+    {
+      "epoch": 0.9585914574502772,
+      "grad_norm": 0.12809327244758606,
+      "learning_rate": 0.0001,
+      "loss": 0.2109,
+      "step": 588
+    },
+    {
+      "epoch": 0.9602217150309749,
+      "grad_norm": 0.1122494712471962,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 589
+    },
+    {
+      "epoch": 0.9618519726116727,
+      "grad_norm": 0.11698026210069656,
+      "learning_rate": 0.0001,
+      "loss": 0.2147,
+      "step": 590
+    },
+    {
+      "epoch": 0.9634822301923704,
+      "grad_norm": 0.15164814889431,
+      "learning_rate": 0.0001,
+      "loss": 0.2159,
+      "step": 591
+    },
+    {
+      "epoch": 0.9651124877730681,
+      "grad_norm": 0.18676526844501495,
+      "learning_rate": 0.0001,
+      "loss": 0.216,
+      "step": 592
+    },
+    {
+      "epoch": 0.9667427453537659,
+      "grad_norm": 0.13141092658042908,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 593
+    },
+    {
+      "epoch": 0.9683730029344636,
+      "grad_norm": 0.1369103491306305,
+      "learning_rate": 0.0001,
+      "loss": 0.2081,
+      "step": 594
+    },
+    {
+      "epoch": 0.9700032605151614,
+      "grad_norm": 0.127171128988266,
+      "learning_rate": 0.0001,
+      "loss": 0.2147,
+      "step": 595
+    },
+    {
+      "epoch": 0.9716335180958592,
+      "grad_norm": 0.11275147646665573,
+      "learning_rate": 0.0001,
+      "loss": 0.2078,
+      "step": 596
+    },
+    {
+      "epoch": 0.9732637756765569,
+      "grad_norm": 0.12019902467727661,
+      "learning_rate": 0.0001,
+      "loss": 0.2128,
+      "step": 597
+    },
+    {
+      "epoch": 0.9748940332572547,
+      "grad_norm": 0.11609843373298645,
+      "learning_rate": 0.0001,
+      "loss": 0.2217,
+      "step": 598
+    },
+    {
+      "epoch": 0.9765242908379524,
+      "grad_norm": 0.10565144568681717,
+      "learning_rate": 0.0001,
+      "loss": 0.2131,
+      "step": 599
+    },
+    {
+      "epoch": 0.9781545484186501,
+      "grad_norm": 0.1125059723854065,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 600
+    },
+    {
+      "epoch": 0.9797848059993479,
+      "grad_norm": 0.10678024590015411,
+      "learning_rate": 0.0001,
+      "loss": 0.2137,
+      "step": 601
+    },
+    {
+      "epoch": 0.9814150635800456,
+      "grad_norm": 0.11686601489782333,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 602
+    },
+    {
+      "epoch": 0.9830453211607434,
+      "grad_norm": 0.16189296543598175,
+      "learning_rate": 0.0001,
+      "loss": 0.202,
+      "step": 603
+    },
+    {
+      "epoch": 0.9846755787414412,
+      "grad_norm": 0.10957061499357224,
+      "learning_rate": 0.0001,
+      "loss": 0.2212,
+      "step": 604
+    },
+    {
+      "epoch": 0.9863058363221389,
+      "grad_norm": 0.1792004108428955,
+      "learning_rate": 0.0001,
+      "loss": 0.2027,
+      "step": 605
+    },
+    {
+      "epoch": 0.9879360939028367,
+      "grad_norm": 0.13383732736110687,
+      "learning_rate": 0.0001,
+      "loss": 0.2186,
+      "step": 606
+    },
+    {
+      "epoch": 0.9895663514835344,
+      "grad_norm": 0.1217399537563324,
+      "learning_rate": 0.0001,
+      "loss": 0.2169,
+      "step": 607
+    },
+    {
+      "epoch": 0.9911966090642321,
+      "grad_norm": 0.11006692051887512,
+      "learning_rate": 0.0001,
+      "loss": 0.2061,
+      "step": 608
+    },
+    {
+      "epoch": 0.9928268666449299,
+      "grad_norm": 0.1142544150352478,
+      "learning_rate": 0.0001,
+      "loss": 0.2202,
+      "step": 609
+    },
+    {
+      "epoch": 0.9944571242256276,
+      "grad_norm": 0.10690510272979736,
+      "learning_rate": 0.0001,
+      "loss": 0.2125,
+      "step": 610
+    },
+    {
+      "epoch": 0.9960873818063254,
+      "grad_norm": 0.11827582865953445,
+      "learning_rate": 0.0001,
+      "loss": 0.2109,
+      "step": 611
+    },
+    {
+      "epoch": 0.9977176393870232,
+      "grad_norm": 0.1035865843296051,
+      "learning_rate": 0.0001,
+      "loss": 0.2035,
+      "step": 612
+    },
+    {
+      "epoch": 0.9993478969677209,
+      "grad_norm": 0.10805841535329819,
+      "learning_rate": 0.0001,
+      "loss": 0.2071,
+      "step": 613
+    },
+    {
+      "epoch": 1.0009781545484187,
+      "grad_norm": 0.11635322868824005,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 614
+    },
+    {
+      "epoch": 1.0026084121291163,
+      "grad_norm": 0.1252707839012146,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 615
+    },
+    {
+      "epoch": 1.0042386697098142,
+      "grad_norm": 0.11579035967588425,
+      "learning_rate": 0.0001,
+      "loss": 0.2082,
+      "step": 616
+    },
+    {
+      "epoch": 1.005868927290512,
+      "grad_norm": 0.11776672303676605,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 617
+    },
+    {
+      "epoch": 1.0074991848712096,
+      "grad_norm": 0.12210860103368759,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 618
+    },
+    {
+      "epoch": 1.0091294424519075,
+      "grad_norm": 0.11382795125246048,
+      "learning_rate": 0.0001,
+      "loss": 0.2096,
+      "step": 619
+    },
+    {
+      "epoch": 1.0107597000326052,
+      "grad_norm": 0.12547257542610168,
+      "learning_rate": 0.0001,
+      "loss": 0.2095,
+      "step": 620
+    },
+    {
+      "epoch": 1.0123899576133029,
+      "grad_norm": 0.10857869684696198,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 621
+    },
+    {
+      "epoch": 1.0140202151940008,
+      "grad_norm": 0.11652690917253494,
+      "learning_rate": 0.0001,
+      "loss": 0.2101,
+      "step": 622
+    },
+    {
+      "epoch": 1.0156504727746984,
+      "grad_norm": 0.13851098716259003,
+      "learning_rate": 0.0001,
+      "loss": 0.2037,
+      "step": 623
+    },
+    {
+      "epoch": 1.0172807303553961,
+      "grad_norm": 0.10633445531129837,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "step": 624
+    },
+    {
+      "epoch": 1.0189109879360938,
+      "grad_norm": 0.13127733767032623,
+      "learning_rate": 0.0001,
+      "loss": 0.2161,
+      "step": 625
+    },
+    {
+      "epoch": 1.0205412455167917,
+      "grad_norm": 0.14254391193389893,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 626
+    },
+    {
+      "epoch": 1.0221715030974894,
+      "grad_norm": 0.12574298679828644,
+      "learning_rate": 0.0001,
+      "loss": 0.212,
+      "step": 627
+    },
+    {
+      "epoch": 1.023801760678187,
+      "grad_norm": 0.1356106996536255,
+      "learning_rate": 0.0001,
+      "loss": 0.2258,
+      "step": 628
+    },
+    {
+      "epoch": 1.025432018258885,
+      "grad_norm": 0.11071506142616272,
+      "learning_rate": 0.0001,
+      "loss": 0.2125,
+      "step": 629
+    },
+    {
+      "epoch": 1.0270622758395827,
+      "grad_norm": 0.12080221623182297,
+      "learning_rate": 0.0001,
+      "loss": 0.1954,
+      "step": 630
+    },
+    {
+      "epoch": 1.0286925334202803,
+      "grad_norm": 0.11128834635019302,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 631
+    },
+    {
+      "epoch": 1.0303227910009782,
+      "grad_norm": 0.13292762637138367,
+      "learning_rate": 0.0001,
+      "loss": 0.2087,
+      "step": 632
+    },
+    {
+      "epoch": 1.031953048581676,
+      "grad_norm": 0.11642225086688995,
+      "learning_rate": 0.0001,
+      "loss": 0.2161,
+      "step": 633
+    },
+    {
+      "epoch": 1.0335833061623736,
+      "grad_norm": 0.12899143993854523,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 634
+    },
+    {
+      "epoch": 1.0352135637430715,
+      "grad_norm": 0.11403586715459824,
+      "learning_rate": 0.0001,
+      "loss": 0.1926,
+      "step": 635
+    },
+    {
+      "epoch": 1.0368438213237692,
+      "grad_norm": 0.13663539290428162,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 636
+    },
+    {
+      "epoch": 1.0384740789044669,
+      "grad_norm": 0.11796414107084274,
+      "learning_rate": 0.0001,
+      "loss": 0.1918,
+      "step": 637
+    },
+    {
+      "epoch": 1.0401043364851648,
+      "grad_norm": 0.11379235237836838,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 638
+    },
+    {
+      "epoch": 1.0417345940658624,
+      "grad_norm": 0.11975998431444168,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 639
+    },
+    {
+      "epoch": 1.0433648516465601,
+      "grad_norm": 0.12381618469953537,
+      "learning_rate": 0.0001,
+      "loss": 0.202,
+      "step": 640
+    },
+    {
+      "epoch": 1.0449951092272578,
+      "grad_norm": 0.14013221859931946,
+      "learning_rate": 0.0001,
+      "loss": 0.2016,
+      "step": 641
+    },
+    {
+      "epoch": 1.0466253668079557,
+      "grad_norm": 0.14774538576602936,
+      "learning_rate": 0.0001,
+      "loss": 0.2068,
+      "step": 642
+    },
+    {
+      "epoch": 1.0482556243886534,
+      "grad_norm": 0.10894253849983215,
+      "learning_rate": 0.0001,
+      "loss": 0.1881,
+      "step": 643
+    },
+    {
+      "epoch": 1.049885881969351,
+      "grad_norm": 0.12854108214378357,
+      "learning_rate": 0.0001,
+      "loss": 0.2095,
+      "step": 644
+    },
+    {
+      "epoch": 1.051516139550049,
+      "grad_norm": 0.1026046872138977,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 645
+    },
+    {
+      "epoch": 1.0531463971307466,
+      "grad_norm": 0.11042814701795578,
+      "learning_rate": 0.0001,
+      "loss": 0.1953,
+      "step": 646
+    },
+    {
+      "epoch": 1.0547766547114443,
+      "grad_norm": 0.11693835258483887,
+      "learning_rate": 0.0001,
+      "loss": 0.2152,
+      "step": 647
+    },
+    {
+      "epoch": 1.0564069122921422,
+      "grad_norm": 0.11914242058992386,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 648
+    },
+    {
+      "epoch": 1.05803716987284,
+      "grad_norm": 0.11633922904729843,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 649
+    },
+    {
+      "epoch": 1.0596674274535376,
+      "grad_norm": 0.10438397526741028,
+      "learning_rate": 0.0001,
+      "loss": 0.1942,
+      "step": 650
+    },
+    {
+      "epoch": 1.0612976850342355,
+      "grad_norm": 0.14384230971336365,
+      "learning_rate": 0.0001,
+      "loss": 0.2175,
+      "step": 651
+    },
+    {
+      "epoch": 1.0629279426149332,
+      "grad_norm": 0.13774406909942627,
+      "learning_rate": 0.0001,
+      "loss": 0.208,
+      "step": 652
+    },
+    {
+      "epoch": 1.0645582001956309,
+      "grad_norm": 0.1285228431224823,
+      "learning_rate": 0.0001,
+      "loss": 0.1938,
+      "step": 653
+    },
+    {
+      "epoch": 1.0661884577763288,
+      "grad_norm": 0.1321713626384735,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 654
+    },
+    {
+      "epoch": 1.0678187153570264,
+      "grad_norm": 0.13438454270362854,
+      "learning_rate": 0.0001,
+      "loss": 0.2087,
+      "step": 655
+    },
+    {
+      "epoch": 1.0694489729377241,
+      "grad_norm": 0.12169782817363739,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 656
+    },
+    {
+      "epoch": 1.0710792305184218,
+      "grad_norm": 0.12447503954172134,
+      "learning_rate": 0.0001,
+      "loss": 0.2029,
+      "step": 657
+    },
+    {
+      "epoch": 1.0727094880991197,
+      "grad_norm": 0.135478213429451,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 658
+    },
+    {
+      "epoch": 1.0743397456798174,
+      "grad_norm": 0.11875299364328384,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 659
+    },
+    {
+      "epoch": 1.075970003260515,
+      "grad_norm": 0.13131263852119446,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 660
+    },
+    {
+      "epoch": 1.077600260841213,
+      "grad_norm": 0.10770665854215622,
+      "learning_rate": 0.0001,
+      "loss": 0.189,
+      "step": 661
+    },
+    {
+      "epoch": 1.0792305184219106,
+      "grad_norm": 0.11619222909212112,
+      "learning_rate": 0.0001,
+      "loss": 0.2015,
+      "step": 662
+    },
+    {
+      "epoch": 1.0808607760026083,
+      "grad_norm": 0.10690273344516754,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 663
+    },
+    {
+      "epoch": 1.0824910335833062,
+      "grad_norm": 0.13410069048404694,
+      "learning_rate": 0.0001,
+      "loss": 0.1988,
+      "step": 664
+    },
+    {
+      "epoch": 1.084121291164004,
+      "grad_norm": 0.12140218168497086,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 665
+    },
+    {
+      "epoch": 1.0857515487447016,
+      "grad_norm": 0.13700392842292786,
+      "learning_rate": 0.0001,
+      "loss": 0.2007,
+      "step": 666
+    },
+    {
+      "epoch": 1.0873818063253995,
+      "grad_norm": 0.12108683586120605,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 667
+    },
+    {
+      "epoch": 1.0890120639060972,
+      "grad_norm": 0.24911844730377197,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 668
+    },
+    {
+      "epoch": 1.0906423214867949,
+      "grad_norm": 0.1707676649093628,
+      "learning_rate": 0.0001,
+      "loss": 0.2091,
+      "step": 669
+    },
+    {
+      "epoch": 1.0922725790674928,
+      "grad_norm": 0.1396542340517044,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 670
+    },
+    {
+      "epoch": 1.0939028366481904,
+      "grad_norm": 0.12876304984092712,
+      "learning_rate": 0.0001,
+      "loss": 0.1981,
+      "step": 671
+    },
+    {
+      "epoch": 1.0955330942288881,
+      "grad_norm": 0.16617383062839508,
+      "learning_rate": 0.0001,
+      "loss": 0.2049,
+      "step": 672
+    },
+    {
+      "epoch": 1.097163351809586,
+      "grad_norm": 0.1533372402191162,
+      "learning_rate": 0.0001,
+      "loss": 0.2143,
+      "step": 673
+    },
+    {
+      "epoch": 1.0987936093902837,
+      "grad_norm": 0.16476184129714966,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 674
+    },
+    {
+      "epoch": 1.1004238669709814,
+      "grad_norm": 0.27153271436691284,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 675
+    },
+    {
+      "epoch": 1.102054124551679,
+      "grad_norm": 0.14181676506996155,
+      "learning_rate": 0.0001,
+      "loss": 0.2063,
+      "step": 676
+    },
+    {
+      "epoch": 1.103684382132377,
+      "grad_norm": 0.1463928073644638,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 677
+    },
+    {
+      "epoch": 1.1053146397130746,
+      "grad_norm": 0.12665878236293793,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 678
+    },
+    {
+      "epoch": 1.1069448972937723,
+      "grad_norm": 0.12541510164737701,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 679
+    },
+    {
+      "epoch": 1.1085751548744702,
+      "grad_norm": 0.11624184995889664,
+      "learning_rate": 0.0001,
+      "loss": 0.1958,
+      "step": 680
+    },
+    {
+      "epoch": 1.110205412455168,
+      "grad_norm": 0.11761149019002914,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 681
+    },
+    {
+      "epoch": 1.1118356700358656,
+      "grad_norm": 0.12375003099441528,
+      "learning_rate": 0.0001,
+      "loss": 0.2028,
+      "step": 682
+    },
+    {
+      "epoch": 1.1134659276165635,
+      "grad_norm": 0.12303782999515533,
+      "learning_rate": 0.0001,
+      "loss": 0.2031,
+      "step": 683
+    },
+    {
+      "epoch": 1.1150961851972612,
+      "grad_norm": 0.11100725084543228,
+      "learning_rate": 0.0001,
+      "loss": 0.2101,
+      "step": 684
+    },
+    {
+      "epoch": 1.1167264427779588,
+      "grad_norm": 0.09348457306623459,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 685
+    },
+    {
+      "epoch": 1.1183567003586568,
+      "grad_norm": 0.1131131500005722,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 686
+    },
+    {
+      "epoch": 1.1199869579393544,
+      "grad_norm": 0.15573875606060028,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 687
+    },
+    {
+      "epoch": 1.1216172155200521,
+      "grad_norm": 0.24191588163375854,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 688
+    },
+    {
+      "epoch": 1.12324747310075,
+      "grad_norm": 0.14230898022651672,
+      "learning_rate": 0.0001,
+      "loss": 0.2029,
+      "step": 689
+    },
+    {
+      "epoch": 1.1248777306814477,
+      "grad_norm": 0.1414661705493927,
+      "learning_rate": 0.0001,
+      "loss": 0.2048,
+      "step": 690
+    },
+    {
+      "epoch": 1.1265079882621454,
+      "grad_norm": 0.12486743181943893,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 691
+    },
+    {
+      "epoch": 1.1281382458428433,
+      "grad_norm": 0.12714742124080658,
+      "learning_rate": 0.0001,
+      "loss": 0.1963,
+      "step": 692
+    },
+    {
+      "epoch": 1.129768503423541,
+      "grad_norm": 0.12481556832790375,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 693
+    },
+    {
+      "epoch": 1.1313987610042386,
+      "grad_norm": 0.10976168513298035,
+      "learning_rate": 0.0001,
+      "loss": 0.2018,
+      "step": 694
+    },
+    {
+      "epoch": 1.1330290185849363,
+      "grad_norm": 0.11793534457683563,
+      "learning_rate": 0.0001,
+      "loss": 0.1985,
+      "step": 695
+    },
+    {
+      "epoch": 1.1346592761656342,
+      "grad_norm": 0.22251418232917786,
+      "learning_rate": 0.0001,
+      "loss": 0.2008,
+      "step": 696
+    },
+    {
+      "epoch": 1.136289533746332,
+      "grad_norm": 0.14538496732711792,
+      "learning_rate": 0.0001,
+      "loss": 0.192,
+      "step": 697
+    },
+    {
+      "epoch": 1.1379197913270296,
+      "grad_norm": 0.1503850817680359,
+      "learning_rate": 0.0001,
+      "loss": 0.2001,
+      "step": 698
+    },
+    {
+      "epoch": 1.1395500489077275,
+      "grad_norm": 0.1038767620921135,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 699
+    },
+    {
+      "epoch": 1.1411803064884252,
+      "grad_norm": 0.12644357979297638,
+      "learning_rate": 0.0001,
+      "loss": 0.1931,
+      "step": 700
+    },
+    {
+      "epoch": 1.1428105640691228,
+      "grad_norm": 0.13702809810638428,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 701
+    },
+    {
+      "epoch": 1.1444408216498207,
+      "grad_norm": 0.10647303611040115,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 702
+    },
+    {
+      "epoch": 1.1460710792305184,
+      "grad_norm": 0.12528447806835175,
+      "learning_rate": 0.0001,
+      "loss": 0.2013,
+      "step": 703
+    },
+    {
+      "epoch": 1.147701336811216,
+      "grad_norm": 0.21072790026664734,
+      "learning_rate": 0.0001,
+      "loss": 0.1979,
+      "step": 704
+    },
+    {
+      "epoch": 1.149331594391914,
+      "grad_norm": 0.14460940659046173,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 705
+    },
+    {
+      "epoch": 1.1509618519726117,
+      "grad_norm": 0.14423388242721558,
+      "learning_rate": 0.0001,
+      "loss": 0.2138,
+      "step": 706
+    },
+    {
+      "epoch": 1.1525921095533094,
+      "grad_norm": 0.13311851024627686,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 707
+    },
+    {
+      "epoch": 1.154222367134007,
+      "grad_norm": 0.2122778594493866,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 708
+    },
+    {
+      "epoch": 1.155852624714705,
+      "grad_norm": 0.12857376039028168,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 709
+    },
+    {
+      "epoch": 1.1574828822954026,
+      "grad_norm": 0.1168324202299118,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 710
+    },
+    {
+      "epoch": 1.1591131398761005,
+      "grad_norm": 0.1203528344631195,
+      "learning_rate": 0.0001,
+      "loss": 0.2015,
+      "step": 711
+    },
+    {
+      "epoch": 1.1607433974567982,
+      "grad_norm": 0.10224917531013489,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 712
+    },
+    {
+      "epoch": 1.162373655037496,
+      "grad_norm": 0.13524509966373444,
+      "learning_rate": 0.0001,
+      "loss": 0.1919,
+      "step": 713
+    },
+    {
+      "epoch": 1.1640039126181936,
+      "grad_norm": 0.20357482135295868,
+      "learning_rate": 0.0001,
+      "loss": 0.1938,
+      "step": 714
+    },
+    {
+      "epoch": 1.1656341701988915,
+      "grad_norm": 0.11235658079385757,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 715
+    },
+    {
+      "epoch": 1.1672644277795892,
+      "grad_norm": 0.12972696125507355,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 716
+    },
+    {
+      "epoch": 1.1688946853602868,
+      "grad_norm": 0.128658264875412,
+      "learning_rate": 0.0001,
+      "loss": 0.2,
+      "step": 717
+    },
+    {
+      "epoch": 1.1705249429409847,
+      "grad_norm": 0.13908959925174713,
+      "learning_rate": 0.0001,
+      "loss": 0.2102,
+      "step": 718
+    },
+    {
+      "epoch": 1.1721552005216824,
+      "grad_norm": 0.12704800069332123,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 719
+    },
+    {
+      "epoch": 1.17378545810238,
+      "grad_norm": 0.11856772005558014,
+      "learning_rate": 0.0001,
+      "loss": 0.1963,
+      "step": 720
+    },
+    {
+      "epoch": 1.175415715683078,
+      "grad_norm": 0.11944851279258728,
+      "learning_rate": 0.0001,
+      "loss": 0.203,
+      "step": 721
+    },
+    {
+      "epoch": 1.1770459732637757,
+      "grad_norm": 0.12153194844722748,
+      "learning_rate": 0.0001,
+      "loss": 0.2094,
+      "step": 722
+    },
+    {
+      "epoch": 1.1786762308444734,
+      "grad_norm": 0.09558034688234329,
+      "learning_rate": 0.0001,
+      "loss": 0.1688,
+      "step": 723
+    },
+    {
+      "epoch": 1.1803064884251713,
+      "grad_norm": 0.1317276656627655,
+      "learning_rate": 0.0001,
+      "loss": 0.2023,
+      "step": 724
+    },
+    {
+      "epoch": 1.181936746005869,
+      "grad_norm": 0.10799970477819443,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 725
+    },
+    {
+      "epoch": 1.1835670035865666,
+      "grad_norm": 0.11460768431425095,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 726
+    },
+    {
+      "epoch": 1.1851972611672643,
+      "grad_norm": 0.10214045643806458,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 727
+    },
+    {
+      "epoch": 1.1868275187479622,
+      "grad_norm": 0.11441674083471298,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 728
+    },
+    {
+      "epoch": 1.18845777632866,
+      "grad_norm": 0.1123800054192543,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 729
+    },
+    {
+      "epoch": 1.1900880339093578,
+      "grad_norm": 0.11834810674190521,
+      "learning_rate": 0.0001,
+      "loss": 0.1871,
+      "step": 730
+    },
+    {
+      "epoch": 1.1917182914900555,
+      "grad_norm": 0.12345534563064575,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 731
+    },
+    {
+      "epoch": 1.1933485490707532,
+      "grad_norm": 0.11334647983312607,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 732
+    },
+    {
+      "epoch": 1.1949788066514508,
+      "grad_norm": 0.13123269379138947,
+      "learning_rate": 0.0001,
+      "loss": 0.2042,
+      "step": 733
+    },
+    {
+      "epoch": 1.1966090642321487,
+      "grad_norm": 0.14825505018234253,
+      "learning_rate": 0.0001,
+      "loss": 0.1988,
+      "step": 734
+    },
+    {
+      "epoch": 1.1982393218128464,
+      "grad_norm": 0.17332376539707184,
+      "learning_rate": 0.0001,
+      "loss": 0.2023,
+      "step": 735
+    },
+    {
+      "epoch": 1.199869579393544,
+      "grad_norm": 0.1413118690252304,
+      "learning_rate": 0.0001,
+      "loss": 0.2003,
+      "step": 736
+    },
+    {
+      "epoch": 1.201499836974242,
+      "grad_norm": 0.14043155312538147,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 737
+    },
+    {
+      "epoch": 1.2031300945549397,
+      "grad_norm": 0.11321795731782913,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 738
+    },
+    {
+      "epoch": 1.2047603521356374,
+      "grad_norm": 0.11889634281396866,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 739
+    },
+    {
+      "epoch": 1.2063906097163353,
+      "grad_norm": 0.14727137982845306,
+      "learning_rate": 0.0001,
+      "loss": 0.2097,
+      "step": 740
+    },
+    {
+      "epoch": 1.208020867297033,
+      "grad_norm": 0.1467553675174713,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 741
+    },
+    {
+      "epoch": 1.2096511248777306,
+      "grad_norm": 0.12245435267686844,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 742
+    },
+    {
+      "epoch": 1.2112813824584285,
+      "grad_norm": 0.12549766898155212,
+      "learning_rate": 0.0001,
+      "loss": 0.1981,
+      "step": 743
+    },
+    {
+      "epoch": 1.2129116400391262,
+      "grad_norm": 0.11142990738153458,
+      "learning_rate": 0.0001,
+      "loss": 0.2054,
+      "step": 744
+    },
+    {
+      "epoch": 1.214541897619824,
+      "grad_norm": 0.11524354666471481,
+      "learning_rate": 0.0001,
+      "loss": 0.1892,
+      "step": 745
+    },
+    {
+      "epoch": 1.2161721552005216,
+      "grad_norm": 0.15625351667404175,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 746
+    },
+    {
+      "epoch": 1.2178024127812195,
+      "grad_norm": 0.10953667014837265,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 747
+    },
+    {
+      "epoch": 1.2194326703619172,
+      "grad_norm": 0.13718314468860626,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 748
+    },
+    {
+      "epoch": 1.2210629279426148,
+      "grad_norm": 0.13286104798316956,
+      "learning_rate": 0.0001,
+      "loss": 0.2169,
+      "step": 749
+    },
+    {
+      "epoch": 1.2226931855233127,
+      "grad_norm": 0.122381292283535,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 750
+    },
+    {
+      "epoch": 1.2243234431040104,
+      "grad_norm": 0.13149864971637726,
+      "learning_rate": 0.0001,
+      "loss": 0.1999,
+      "step": 751
+    },
+    {
+      "epoch": 1.225953700684708,
+      "grad_norm": 0.129011869430542,
+      "learning_rate": 0.0001,
+      "loss": 0.2067,
+      "step": 752
+    },
+    {
+      "epoch": 1.227583958265406,
+      "grad_norm": 0.12609954178333282,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 753
+    },
+    {
+      "epoch": 1.2292142158461037,
+      "grad_norm": 0.12401840835809708,
+      "learning_rate": 0.0001,
+      "loss": 0.2102,
+      "step": 754
+    },
+    {
+      "epoch": 1.2308444734268014,
+      "grad_norm": 0.1191437765955925,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 755
+    },
+    {
+      "epoch": 1.2324747310074993,
+      "grad_norm": 0.11774501204490662,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 756
+    },
+    {
+      "epoch": 1.234104988588197,
+      "grad_norm": 0.09502781182527542,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 757
+    },
+    {
+      "epoch": 1.2357352461688946,
+      "grad_norm": 0.12842731177806854,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 758
+    },
+    {
+      "epoch": 1.2373655037495925,
+      "grad_norm": 0.1238941103219986,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 759
+    },
+    {
+      "epoch": 1.2389957613302902,
+      "grad_norm": 0.120622418820858,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 760
+    },
+    {
+      "epoch": 1.240626018910988,
+      "grad_norm": 0.16394123435020447,
+      "learning_rate": 0.0001,
+      "loss": 0.2024,
+      "step": 761
+    },
+    {
+      "epoch": 1.2422562764916858,
+      "grad_norm": 0.13155362010002136,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 762
+    },
+    {
+      "epoch": 1.2438865340723835,
+      "grad_norm": 0.11317794770002365,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 763
+    },
+    {
+      "epoch": 1.2455167916530812,
+      "grad_norm": 0.1305047571659088,
+      "learning_rate": 0.0001,
+      "loss": 0.2106,
+      "step": 764
+    },
+    {
+      "epoch": 1.2471470492337788,
+      "grad_norm": 0.12853112816810608,
+      "learning_rate": 0.0001,
+      "loss": 0.1993,
+      "step": 765
+    },
+    {
+      "epoch": 1.2487773068144767,
+      "grad_norm": 0.12205260246992111,
+      "learning_rate": 0.0001,
+      "loss": 0.195,
+      "step": 766
+    },
+    {
+      "epoch": 1.2504075643951744,
+      "grad_norm": 0.12035010755062103,
+      "learning_rate": 0.0001,
+      "loss": 0.2054,
+      "step": 767
+    },
+    {
+      "epoch": 1.2520378219758723,
+      "grad_norm": 0.12386268377304077,
+      "learning_rate": 0.0001,
+      "loss": 0.2047,
+      "step": 768
+    },
+    {
+      "epoch": 1.25366807955657,
+      "grad_norm": 0.11850553750991821,
+      "learning_rate": 0.0001,
+      "loss": 0.2104,
+      "step": 769
+    },
+    {
+      "epoch": 1.2552983371372677,
+      "grad_norm": 0.12291909754276276,
+      "learning_rate": 0.0001,
+      "loss": 0.2073,
+      "step": 770
+    },
+    {
+      "epoch": 1.2569285947179654,
+      "grad_norm": 0.11463002860546112,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 771
+    },
+    {
+      "epoch": 1.2585588522986633,
+      "grad_norm": 0.11392191052436829,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 772
+    },
+    {
+      "epoch": 1.260189109879361,
+      "grad_norm": 0.11817040294408798,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 773
+    },
+    {
+      "epoch": 1.2618193674600586,
+      "grad_norm": 0.11032837629318237,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 774
+    },
+    {
+      "epoch": 1.2634496250407565,
+      "grad_norm": 0.12903887033462524,
+      "learning_rate": 0.0001,
+      "loss": 0.2054,
+      "step": 775
+    },
+    {
+      "epoch": 1.2650798826214542,
+      "grad_norm": 0.10510163009166718,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 776
+    },
+    {
+      "epoch": 1.266710140202152,
+      "grad_norm": 0.10939998179674149,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 777
+    },
+    {
+      "epoch": 1.2683403977828496,
+      "grad_norm": 0.13864178955554962,
+      "learning_rate": 0.0001,
+      "loss": 0.1995,
+      "step": 778
+    },
+    {
+      "epoch": 1.2699706553635475,
+      "grad_norm": 0.13182048499584198,
+      "learning_rate": 0.0001,
+      "loss": 0.1982,
+      "step": 779
+    },
+    {
+      "epoch": 1.2716009129442452,
+      "grad_norm": 0.13241499662399292,
+      "learning_rate": 0.0001,
+      "loss": 0.2119,
+      "step": 780
+    },
+    {
+      "epoch": 1.273231170524943,
+      "grad_norm": 0.10950156301259995,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 781
+    },
+    {
+      "epoch": 1.2748614281056407,
+      "grad_norm": 0.1152586042881012,
+      "learning_rate": 0.0001,
+      "loss": 0.2138,
+      "step": 782
+    },
+    {
+      "epoch": 1.2764916856863384,
+      "grad_norm": 0.09927522391080856,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 783
+    },
+    {
+      "epoch": 1.278121943267036,
+      "grad_norm": 0.12837468087673187,
+      "learning_rate": 0.0001,
+      "loss": 0.2013,
+      "step": 784
+    },
+    {
+      "epoch": 1.279752200847734,
+      "grad_norm": 0.11714199930429459,
+      "learning_rate": 0.0001,
+      "loss": 0.1913,
+      "step": 785
+    },
+    {
+      "epoch": 1.2813824584284317,
+      "grad_norm": 0.1406959891319275,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 786
+    },
+    {
+      "epoch": 1.2830127160091294,
+      "grad_norm": 0.11594724655151367,
+      "learning_rate": 0.0001,
+      "loss": 0.1758,
+      "step": 787
+    },
+    {
+      "epoch": 1.2846429735898273,
+      "grad_norm": 0.1368754804134369,
+      "learning_rate": 0.0001,
+      "loss": 0.2067,
+      "step": 788
+    },
+    {
+      "epoch": 1.286273231170525,
+      "grad_norm": 0.14420491456985474,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 789
+    },
+    {
+      "epoch": 1.2879034887512226,
+      "grad_norm": 0.11989482492208481,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 790
+    },
+    {
+      "epoch": 1.2895337463319203,
+      "grad_norm": 0.09911071509122849,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 791
+    },
+    {
+      "epoch": 1.2911640039126182,
+      "grad_norm": 0.10733790695667267,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 792
+    },
+    {
+      "epoch": 1.2927942614933159,
+      "grad_norm": 0.12739041447639465,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 793
+    },
+    {
+      "epoch": 1.2944245190740138,
+      "grad_norm": 0.12702591717243195,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 794
+    },
+    {
+      "epoch": 1.2960547766547115,
+      "grad_norm": 0.10774042457342148,
+      "learning_rate": 0.0001,
+      "loss": 0.2061,
+      "step": 795
+    },
+    {
+      "epoch": 1.2976850342354092,
+      "grad_norm": 0.14348004758358002,
+      "learning_rate": 0.0001,
+      "loss": 0.2007,
+      "step": 796
+    },
+    {
+      "epoch": 1.2993152918161068,
+      "grad_norm": 0.11154497414827347,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 797
+    },
+    {
+      "epoch": 1.3009455493968047,
+      "grad_norm": 0.11274070292711258,
+      "learning_rate": 0.0001,
+      "loss": 0.199,
+      "step": 798
+    },
+    {
+      "epoch": 1.3025758069775024,
+      "grad_norm": 0.12772290408611298,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 799
+    },
+    {
+      "epoch": 1.3042060645582003,
+      "grad_norm": 0.138117715716362,
+      "learning_rate": 0.0001,
+      "loss": 0.2096,
+      "step": 800
+    },
+    {
+      "epoch": 1.305836322138898,
+      "grad_norm": 0.11759902536869049,
+      "learning_rate": 0.0001,
+      "loss": 0.1935,
+      "step": 801
+    },
+    {
+      "epoch": 1.3074665797195957,
+      "grad_norm": 0.1468537598848343,
+      "learning_rate": 0.0001,
+      "loss": 0.1878,
+      "step": 802
+    },
+    {
+      "epoch": 1.3090968373002934,
+      "grad_norm": 0.19811546802520752,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 803
+    },
+    {
+      "epoch": 1.3107270948809913,
+      "grad_norm": 0.11782370507717133,
+      "learning_rate": 0.0001,
+      "loss": 0.1935,
+      "step": 804
+    },
+    {
+      "epoch": 1.312357352461689,
+      "grad_norm": 0.14963723719120026,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 805
+    },
+    {
+      "epoch": 1.3139876100423866,
+      "grad_norm": 0.12052904069423676,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 806
+    },
+    {
+      "epoch": 1.3156178676230845,
+      "grad_norm": 0.13625605404376984,
+      "learning_rate": 0.0001,
+      "loss": 0.2156,
+      "step": 807
+    },
+    {
+      "epoch": 1.3172481252037822,
+      "grad_norm": 0.13118813931941986,
+      "learning_rate": 0.0001,
+      "loss": 0.2115,
+      "step": 808
+    },
+    {
+      "epoch": 1.3188783827844799,
+      "grad_norm": 0.12743224203586578,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 809
+    },
+    {
+      "epoch": 1.3205086403651776,
+      "grad_norm": 0.10934583097696304,
+      "learning_rate": 0.0001,
+      "loss": 0.1803,
+      "step": 810
+    },
+    {
+      "epoch": 1.3221388979458755,
+      "grad_norm": 0.12755925953388214,
+      "learning_rate": 0.0001,
+      "loss": 0.195,
+      "step": 811
+    },
+    {
+      "epoch": 1.3237691555265731,
+      "grad_norm": 0.13710075616836548,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 812
+    },
+    {
+      "epoch": 1.325399413107271,
+      "grad_norm": 0.11187458783388138,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 813
+    },
+    {
+      "epoch": 1.3270296706879687,
+      "grad_norm": 0.12023862451314926,
+      "learning_rate": 0.0001,
+      "loss": 0.2037,
+      "step": 814
+    },
+    {
+      "epoch": 1.3286599282686664,
+      "grad_norm": 0.13946884870529175,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 815
+    },
+    {
+      "epoch": 1.330290185849364,
+      "grad_norm": 0.1312728077173233,
+      "learning_rate": 0.0001,
+      "loss": 0.2021,
+      "step": 816
+    },
+    {
+      "epoch": 1.331920443430062,
+      "grad_norm": 0.12684758007526398,
+      "learning_rate": 0.0001,
+      "loss": 0.2044,
+      "step": 817
+    },
+    {
+      "epoch": 1.3335507010107597,
+      "grad_norm": 0.11161263287067413,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 818
+    },
+    {
+      "epoch": 1.3351809585914576,
+      "grad_norm": 0.11830423027276993,
+      "learning_rate": 0.0001,
+      "loss": 0.2042,
+      "step": 819
+    },
+    {
+      "epoch": 1.3368112161721553,
+      "grad_norm": 0.13050754368305206,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 820
+    },
+    {
+      "epoch": 1.338441473752853,
+      "grad_norm": 0.11719314008951187,
+      "learning_rate": 0.0001,
+      "loss": 0.1935,
+      "step": 821
+    },
+    {
+      "epoch": 1.3400717313335506,
+      "grad_norm": 0.1417597383260727,
+      "learning_rate": 0.0001,
+      "loss": 0.1915,
+      "step": 822
+    },
+    {
+      "epoch": 1.3417019889142485,
+      "grad_norm": 0.10985542833805084,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 823
+    },
+    {
+      "epoch": 1.3433322464949462,
+      "grad_norm": 0.1182430163025856,
+      "learning_rate": 0.0001,
+      "loss": 0.2075,
+      "step": 824
+    },
+    {
+      "epoch": 1.3449625040756439,
+      "grad_norm": 0.11384329944849014,
+      "learning_rate": 0.0001,
+      "loss": 0.1973,
+      "step": 825
+    },
+    {
+      "epoch": 1.3465927616563418,
+      "grad_norm": 0.13239991664886475,
+      "learning_rate": 0.0001,
+      "loss": 0.2035,
+      "step": 826
+    },
+    {
+      "epoch": 1.3482230192370395,
+      "grad_norm": 0.11426941305398941,
+      "learning_rate": 0.0001,
+      "loss": 0.196,
+      "step": 827
+    },
+    {
+      "epoch": 1.3498532768177371,
+      "grad_norm": 0.11955778300762177,
+      "learning_rate": 0.0001,
+      "loss": 0.2019,
+      "step": 828
+    },
+    {
+      "epoch": 1.3514835343984348,
+      "grad_norm": 0.11563610285520554,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 829
+    },
+    {
+      "epoch": 1.3531137919791327,
+      "grad_norm": 0.09637649357318878,
+      "learning_rate": 0.0001,
+      "loss": 0.1875,
+      "step": 830
+    },
+    {
+      "epoch": 1.3547440495598304,
+      "grad_norm": 0.12103250622749329,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 831
+    },
+    {
+      "epoch": 1.3563743071405283,
+      "grad_norm": 0.11947387456893921,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 832
+    },
+    {
+      "epoch": 1.358004564721226,
+      "grad_norm": 0.12026092410087585,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 833
+    },
+    {
+      "epoch": 1.3596348223019237,
+      "grad_norm": 0.12217531353235245,
+      "learning_rate": 0.0001,
+      "loss": 0.2122,
+      "step": 834
+    },
+    {
+      "epoch": 1.3612650798826214,
+      "grad_norm": 0.11990203708410263,
+      "learning_rate": 0.0001,
+      "loss": 0.1959,
+      "step": 835
+    },
+    {
+      "epoch": 1.3628953374633193,
+      "grad_norm": 0.12255218625068665,
+      "learning_rate": 0.0001,
+      "loss": 0.2043,
+      "step": 836
+    },
+    {
+      "epoch": 1.364525595044017,
+      "grad_norm": 0.13003982603549957,
+      "learning_rate": 0.0001,
+      "loss": 0.1988,
+      "step": 837
+    },
+    {
+      "epoch": 1.3661558526247148,
+      "grad_norm": 0.12093237787485123,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 838
+    },
+    {
+      "epoch": 1.3677861102054125,
+      "grad_norm": 0.11946821212768555,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 839
+    },
+    {
+      "epoch": 1.3694163677861102,
+      "grad_norm": 0.18368889391422272,
+      "learning_rate": 0.0001,
+      "loss": 0.1706,
+      "step": 840
+    },
+    {
+      "epoch": 1.3710466253668079,
+      "grad_norm": 0.1539110541343689,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 841
+    },
+    {
+      "epoch": 1.3726768829475058,
+      "grad_norm": 0.11917300522327423,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 842
+    },
+    {
+      "epoch": 1.3743071405282035,
+      "grad_norm": 0.1315378099679947,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 843
+    },
+    {
+      "epoch": 1.3759373981089011,
+      "grad_norm": 0.13452459871768951,
+      "learning_rate": 0.0001,
+      "loss": 0.2124,
+      "step": 844
+    },
+    {
+      "epoch": 1.377567655689599,
+      "grad_norm": 0.11259663850069046,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 845
+    },
+    {
+      "epoch": 1.3791979132702967,
+      "grad_norm": 0.12540358304977417,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 846
+    },
+    {
+      "epoch": 1.3808281708509944,
+      "grad_norm": 0.1402936577796936,
+      "learning_rate": 0.0001,
+      "loss": 0.2006,
+      "step": 847
+    },
+    {
+      "epoch": 1.382458428431692,
+      "grad_norm": 0.13874374330043793,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 848
+    },
+    {
+      "epoch": 1.38408868601239,
+      "grad_norm": 0.12956838309764862,
+      "learning_rate": 0.0001,
+      "loss": 0.1886,
+      "step": 849
+    },
+    {
+      "epoch": 1.3857189435930877,
+      "grad_norm": 0.12310940027236938,
+      "learning_rate": 0.0001,
+      "loss": 0.1776,
+      "step": 850
+    },
+    {
+      "epoch": 1.3873492011737856,
+      "grad_norm": 0.13739284873008728,
+      "learning_rate": 0.0001,
+      "loss": 0.207,
+      "step": 851
+    },
+    {
+      "epoch": 1.3889794587544833,
+      "grad_norm": 0.1304449588060379,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 852
+    },
+    {
+      "epoch": 1.390609716335181,
+      "grad_norm": 0.12641535699367523,
+      "learning_rate": 0.0001,
+      "loss": 0.1935,
+      "step": 853
+    },
+    {
+      "epoch": 1.3922399739158786,
+      "grad_norm": 0.11408405750989914,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 854
+    },
+    {
+      "epoch": 1.3938702314965765,
+      "grad_norm": 0.1254139244556427,
+      "learning_rate": 0.0001,
+      "loss": 0.2097,
+      "step": 855
+    },
+    {
+      "epoch": 1.3955004890772742,
+      "grad_norm": 0.1231856718659401,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 856
+    },
+    {
+      "epoch": 1.3971307466579719,
+      "grad_norm": 0.1396886110305786,
+      "learning_rate": 0.0001,
+      "loss": 0.2226,
+      "step": 857
+    },
+    {
+      "epoch": 1.3987610042386698,
+      "grad_norm": 0.12817426025867462,
+      "learning_rate": 0.0001,
+      "loss": 0.2079,
+      "step": 858
+    },
+    {
+      "epoch": 1.4003912618193675,
+      "grad_norm": 0.13148726522922516,
+      "learning_rate": 0.0001,
+      "loss": 0.1935,
+      "step": 859
+    },
+    {
+      "epoch": 1.4020215194000651,
+      "grad_norm": 0.13093724846839905,
+      "learning_rate": 0.0001,
+      "loss": 0.2066,
+      "step": 860
+    },
+    {
+      "epoch": 1.4036517769807628,
+      "grad_norm": 0.12283019721508026,
+      "learning_rate": 0.0001,
+      "loss": 0.186,
+      "step": 861
+    },
+    {
+      "epoch": 1.4052820345614607,
+      "grad_norm": 0.1363324522972107,
+      "learning_rate": 0.0001,
+      "loss": 0.2062,
+      "step": 862
+    },
+    {
+      "epoch": 1.4069122921421584,
+      "grad_norm": 0.14044034481048584,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 863
+    },
+    {
+      "epoch": 1.4085425497228563,
+      "grad_norm": 0.11587396264076233,
+      "learning_rate": 0.0001,
+      "loss": 0.2011,
+      "step": 864
+    },
+    {
+      "epoch": 1.410172807303554,
+      "grad_norm": 0.12819702923297882,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 865
+    },
+    {
+      "epoch": 1.4118030648842517,
+      "grad_norm": 0.11986611038446426,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 866
+    },
+    {
+      "epoch": 1.4134333224649493,
+      "grad_norm": 0.12252791970968246,
+      "learning_rate": 0.0001,
+      "loss": 0.2095,
+      "step": 867
+    },
+    {
+      "epoch": 1.4150635800456473,
+      "grad_norm": 0.12779594957828522,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 868
+    },
+    {
+      "epoch": 1.416693837626345,
+      "grad_norm": 0.12142222374677658,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 869
+    },
+    {
+      "epoch": 1.4183240952070428,
+      "grad_norm": 0.12713171541690826,
+      "learning_rate": 0.0001,
+      "loss": 0.2122,
+      "step": 870
+    },
+    {
+      "epoch": 1.4199543527877405,
+      "grad_norm": 0.11739431321620941,
+      "learning_rate": 0.0001,
+      "loss": 0.1981,
+      "step": 871
+    },
+    {
+      "epoch": 1.4215846103684382,
+      "grad_norm": 0.11351068317890167,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 872
+    },
+    {
+      "epoch": 1.4232148679491359,
+      "grad_norm": 0.12316130101680756,
+      "learning_rate": 0.0001,
+      "loss": 0.1886,
+      "step": 873
+    },
+    {
+      "epoch": 1.4248451255298338,
+      "grad_norm": 0.11452283710241318,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 874
+    },
+    {
+      "epoch": 1.4264753831105315,
+      "grad_norm": 0.12895885109901428,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 875
+    },
+    {
+      "epoch": 1.4281056406912291,
+      "grad_norm": 0.11376369744539261,
+      "learning_rate": 0.0001,
+      "loss": 0.1972,
+      "step": 876
+    },
+    {
+      "epoch": 1.429735898271927,
+      "grad_norm": 0.0964648500084877,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 877
+    },
+    {
+      "epoch": 1.4313661558526247,
+      "grad_norm": 0.12030114978551865,
+      "learning_rate": 0.0001,
+      "loss": 0.2071,
+      "step": 878
+    },
+    {
+      "epoch": 1.4329964134333224,
+      "grad_norm": 0.1270144283771515,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 879
+    },
+    {
+      "epoch": 1.43462667101402,
+      "grad_norm": 0.11537966132164001,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 880
+    },
+    {
+      "epoch": 1.436256928594718,
+      "grad_norm": 0.10482136160135269,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 881
+    },
+    {
+      "epoch": 1.4378871861754157,
+      "grad_norm": 0.10979556292295456,
+      "learning_rate": 0.0001,
+      "loss": 0.1939,
+      "step": 882
+    },
+    {
+      "epoch": 1.4395174437561136,
+      "grad_norm": 0.1133527085185051,
+      "learning_rate": 0.0001,
+      "loss": 0.1985,
+      "step": 883
+    },
+    {
+      "epoch": 1.4411477013368112,
+      "grad_norm": 0.12253648787736893,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 884
+    },
+    {
+      "epoch": 1.442777958917509,
+      "grad_norm": 0.1556876003742218,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 885
+    },
+    {
+      "epoch": 1.4444082164982066,
+      "grad_norm": 0.13067957758903503,
+      "learning_rate": 0.0001,
+      "loss": 0.2008,
+      "step": 886
+    },
+    {
+      "epoch": 1.4460384740789045,
+      "grad_norm": 0.13614705204963684,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 887
+    },
+    {
+      "epoch": 1.4476687316596022,
+      "grad_norm": 0.13412481546401978,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 888
+    },
+    {
+      "epoch": 1.4492989892403,
+      "grad_norm": 0.12189559638500214,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 889
+    },
+    {
+      "epoch": 1.4509292468209978,
+      "grad_norm": 0.10976207256317139,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 890
+    },
+    {
+      "epoch": 1.4525595044016955,
+      "grad_norm": 0.13595980405807495,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 891
+    },
+    {
+      "epoch": 1.4541897619823931,
+      "grad_norm": 0.1207745224237442,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 892
+    },
+    {
+      "epoch": 1.455820019563091,
+      "grad_norm": 0.12685881555080414,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 893
+    },
+    {
+      "epoch": 1.4574502771437887,
+      "grad_norm": 0.1244385838508606,
+      "learning_rate": 0.0001,
+      "loss": 0.2037,
+      "step": 894
+    },
+    {
+      "epoch": 1.4590805347244864,
+      "grad_norm": 0.11779557913541794,
+      "learning_rate": 0.0001,
+      "loss": 0.2037,
+      "step": 895
+    },
+    {
+      "epoch": 1.4607107923051843,
+      "grad_norm": 0.13534864783287048,
+      "learning_rate": 0.0001,
+      "loss": 0.1915,
+      "step": 896
+    },
+    {
+      "epoch": 1.462341049885882,
+      "grad_norm": 0.1300112009048462,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 897
+    },
+    {
+      "epoch": 1.4639713074665797,
+      "grad_norm": 0.11177927255630493,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 898
+    },
+    {
+      "epoch": 1.4656015650472773,
+      "grad_norm": 0.12038631737232208,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 899
+    },
+    {
+      "epoch": 1.4672318226279752,
+      "grad_norm": 0.1431179791688919,
+      "learning_rate": 0.0001,
+      "loss": 0.2004,
+      "step": 900
+    },
+    {
+      "epoch": 1.468862080208673,
+      "grad_norm": 0.11475679278373718,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 901
+    },
+    {
+      "epoch": 1.4704923377893708,
+      "grad_norm": 0.12329886853694916,
+      "learning_rate": 0.0001,
+      "loss": 0.191,
+      "step": 902
+    },
+    {
+      "epoch": 1.4721225953700685,
+      "grad_norm": 0.12444540858268738,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 903
+    },
+    {
+      "epoch": 1.4737528529507662,
+      "grad_norm": 0.1236625537276268,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 904
+    },
+    {
+      "epoch": 1.4753831105314639,
+      "grad_norm": 0.11803305894136429,
+      "learning_rate": 0.0001,
+      "loss": 0.1939,
+      "step": 905
+    },
+    {
+      "epoch": 1.4770133681121618,
+      "grad_norm": 0.11509499698877335,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 906
+    },
+    {
+      "epoch": 1.4786436256928595,
+      "grad_norm": 0.13276603817939758,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 907
+    },
+    {
+      "epoch": 1.4802738832735574,
+      "grad_norm": 0.14081943035125732,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 908
+    },
+    {
+      "epoch": 1.481904140854255,
+      "grad_norm": 0.15057384967803955,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 909
+    },
+    {
+      "epoch": 1.4835343984349527,
+      "grad_norm": 0.15684540569782257,
+      "learning_rate": 0.0001,
+      "loss": 0.2001,
+      "step": 910
+    },
+    {
+      "epoch": 1.4851646560156504,
+      "grad_norm": 0.15486294031143188,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 911
+    },
+    {
+      "epoch": 1.4867949135963483,
+      "grad_norm": 0.13155148923397064,
+      "learning_rate": 0.0001,
+      "loss": 0.1966,
+      "step": 912
+    },
+    {
+      "epoch": 1.488425171177046,
+      "grad_norm": 0.11298996210098267,
+      "learning_rate": 0.0001,
+      "loss": 0.1944,
+      "step": 913
+    },
+    {
+      "epoch": 1.4900554287577437,
+      "grad_norm": 0.1128094419836998,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 914
+    },
+    {
+      "epoch": 1.4916856863384416,
+      "grad_norm": 0.12357712537050247,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 915
+    },
+    {
+      "epoch": 1.4933159439191392,
+      "grad_norm": 0.10895108431577682,
+      "learning_rate": 0.0001,
+      "loss": 0.193,
+      "step": 916
+    },
+    {
+      "epoch": 1.494946201499837,
+      "grad_norm": 0.10755927115678787,
+      "learning_rate": 0.0001,
+      "loss": 0.1935,
+      "step": 917
+    },
+    {
+      "epoch": 1.4965764590805346,
+      "grad_norm": 0.17024989426136017,
+      "learning_rate": 0.0001,
+      "loss": 0.196,
+      "step": 918
+    },
+    {
+      "epoch": 1.4982067166612325,
+      "grad_norm": 0.1299123913049698,
+      "learning_rate": 0.0001,
+      "loss": 0.2118,
+      "step": 919
+    },
+    {
+      "epoch": 1.4998369742419302,
+      "grad_norm": 0.12259229272603989,
+      "learning_rate": 0.0001,
+      "loss": 0.2014,
+      "step": 920
+    },
+    {
+      "epoch": 1.501467231822628,
+      "grad_norm": 0.12206512689590454,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 921
+    },
+    {
+      "epoch": 1.5030974894033258,
+      "grad_norm": 0.10726633667945862,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 922
+    },
+    {
+      "epoch": 1.5047277469840235,
+      "grad_norm": 0.14176525175571442,
+      "learning_rate": 0.0001,
+      "loss": 0.211,
+      "step": 923
+    },
+    {
+      "epoch": 1.5063580045647211,
+      "grad_norm": 0.11247315257787704,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 924
+    },
+    {
+      "epoch": 1.5079882621454188,
+      "grad_norm": 0.12529021501541138,
+      "learning_rate": 0.0001,
+      "loss": 0.1911,
+      "step": 925
+    },
+    {
+      "epoch": 1.5096185197261167,
+      "grad_norm": 0.15964820981025696,
+      "learning_rate": 0.0001,
+      "loss": 0.1881,
+      "step": 926
+    },
+    {
+      "epoch": 1.5112487773068146,
+      "grad_norm": 0.14119292795658112,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 927
+    },
+    {
+      "epoch": 1.5128790348875123,
+      "grad_norm": 0.14880679547786713,
+      "learning_rate": 0.0001,
+      "loss": 0.2091,
+      "step": 928
+    },
+    {
+      "epoch": 1.51450929246821,
+      "grad_norm": 0.11705358326435089,
+      "learning_rate": 0.0001,
+      "loss": 0.1885,
+      "step": 929
+    },
+    {
+      "epoch": 1.5161395500489077,
+      "grad_norm": 0.12521898746490479,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 930
+    },
+    {
+      "epoch": 1.5177698076296053,
+      "grad_norm": 0.1378403753042221,
+      "learning_rate": 0.0001,
+      "loss": 0.2104,
+      "step": 931
+    },
+    {
+      "epoch": 1.5194000652103032,
+      "grad_norm": 0.10180335491895676,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 932
+    },
+    {
+      "epoch": 1.5210303227910011,
+      "grad_norm": 0.12105914950370789,
+      "learning_rate": 0.0001,
+      "loss": 0.207,
+      "step": 933
+    },
+    {
+      "epoch": 1.5226605803716988,
+      "grad_norm": 0.12631486356258392,
+      "learning_rate": 0.0001,
+      "loss": 0.2082,
+      "step": 934
+    },
+    {
+      "epoch": 1.5242908379523965,
+      "grad_norm": 0.12594158947467804,
+      "learning_rate": 0.0001,
+      "loss": 0.1866,
+      "step": 935
+    },
+    {
+      "epoch": 1.5259210955330942,
+      "grad_norm": 0.10104666650295258,
+      "learning_rate": 0.0001,
+      "loss": 0.1938,
+      "step": 936
+    },
+    {
+      "epoch": 1.5275513531137919,
+      "grad_norm": 0.13341638445854187,
+      "learning_rate": 0.0001,
+      "loss": 0.2042,
+      "step": 937
+    },
+    {
+      "epoch": 1.5291816106944898,
+      "grad_norm": 0.15109512209892273,
+      "learning_rate": 0.0001,
+      "loss": 0.1878,
+      "step": 938
+    },
+    {
+      "epoch": 1.5308118682751874,
+      "grad_norm": 0.12633870542049408,
+      "learning_rate": 0.0001,
+      "loss": 0.21,
+      "step": 939
+    },
+    {
+      "epoch": 1.5324421258558854,
+      "grad_norm": 0.1257101148366928,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 940
+    },
+    {
+      "epoch": 1.534072383436583,
+      "grad_norm": 0.1757507622241974,
+      "learning_rate": 0.0001,
+      "loss": 0.2001,
+      "step": 941
+    },
+    {
+      "epoch": 1.5357026410172807,
+      "grad_norm": 0.12703821063041687,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 942
+    },
+    {
+      "epoch": 1.5373328985979784,
+      "grad_norm": 0.11847993731498718,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 943
+    },
+    {
+      "epoch": 1.538963156178676,
+      "grad_norm": 0.12398132681846619,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 944
+    },
+    {
+      "epoch": 1.540593413759374,
+      "grad_norm": 0.13036128878593445,
+      "learning_rate": 0.0001,
+      "loss": 0.1866,
+      "step": 945
+    },
+    {
+      "epoch": 1.5422236713400719,
+      "grad_norm": 0.10606612265110016,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 946
+    },
+    {
+      "epoch": 1.5438539289207696,
+      "grad_norm": 0.12699076533317566,
+      "learning_rate": 0.0001,
+      "loss": 0.1923,
+      "step": 947
+    },
+    {
+      "epoch": 1.5454841865014672,
+      "grad_norm": 0.13484394550323486,
+      "learning_rate": 0.0001,
+      "loss": 0.191,
+      "step": 948
+    },
+    {
+      "epoch": 1.547114444082165,
+      "grad_norm": 0.11915871500968933,
+      "learning_rate": 0.0001,
+      "loss": 0.1805,
+      "step": 949
+    },
+    {
+      "epoch": 1.5487447016628626,
+      "grad_norm": 0.09939049929380417,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 950
+    },
+    {
+      "epoch": 1.5503749592435605,
+      "grad_norm": 0.11767233163118362,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 951
+    },
+    {
+      "epoch": 1.5520052168242582,
+      "grad_norm": 0.12781041860580444,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 952
+    },
+    {
+      "epoch": 1.553635474404956,
+      "grad_norm": 0.1346462368965149,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 953
+    },
+    {
+      "epoch": 1.5552657319856538,
+      "grad_norm": 0.12352897226810455,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 954
+    },
+    {
+      "epoch": 1.5568959895663514,
+      "grad_norm": 0.12448864430189133,
+      "learning_rate": 0.0001,
+      "loss": 0.2005,
+      "step": 955
+    },
+    {
+      "epoch": 1.5585262471470491,
+      "grad_norm": 0.11041487753391266,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 956
+    },
+    {
+      "epoch": 1.560156504727747,
+      "grad_norm": 0.11932788044214249,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 957
+    },
+    {
+      "epoch": 1.5617867623084447,
+      "grad_norm": 0.11703822016716003,
+      "learning_rate": 0.0001,
+      "loss": 0.1866,
+      "step": 958
+    },
+    {
+      "epoch": 1.5634170198891426,
+      "grad_norm": 0.11386875063180923,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 959
+    },
+    {
+      "epoch": 1.5650472774698403,
+      "grad_norm": 0.1266459971666336,
+      "learning_rate": 0.0001,
+      "loss": 0.1847,
+      "step": 960
+    },
+    {
+      "epoch": 1.566677535050538,
+      "grad_norm": 0.10677938908338547,
+      "learning_rate": 0.0001,
+      "loss": 0.1955,
+      "step": 961
+    },
+    {
+      "epoch": 1.5683077926312357,
+      "grad_norm": 0.13318541646003723,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 962
+    },
+    {
+      "epoch": 1.5699380502119333,
+      "grad_norm": 0.11351260542869568,
+      "learning_rate": 0.0001,
+      "loss": 0.1935,
+      "step": 963
+    },
+    {
+      "epoch": 1.5715683077926312,
+      "grad_norm": 0.10484421998262405,
+      "learning_rate": 0.0001,
+      "loss": 0.1985,
+      "step": 964
+    },
+    {
+      "epoch": 1.5731985653733291,
+      "grad_norm": 0.12113457173109055,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 965
+    },
+    {
+      "epoch": 1.5748288229540268,
+      "grad_norm": 0.11888866871595383,
+      "learning_rate": 0.0001,
+      "loss": 0.1857,
+      "step": 966
+    },
+    {
+      "epoch": 1.5764590805347245,
+      "grad_norm": 0.1369660496711731,
+      "learning_rate": 0.0001,
+      "loss": 0.2087,
+      "step": 967
+    },
+    {
+      "epoch": 1.5780893381154222,
+      "grad_norm": 0.11224696040153503,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 968
+    },
+    {
+      "epoch": 1.5797195956961199,
+      "grad_norm": 0.12822385132312775,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 969
+    },
+    {
+      "epoch": 1.5813498532768178,
+      "grad_norm": 0.13289442658424377,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 970
+    },
+    {
+      "epoch": 1.5829801108575154,
+      "grad_norm": 0.10120034962892532,
+      "learning_rate": 0.0001,
+      "loss": 0.1809,
+      "step": 971
+    },
+    {
+      "epoch": 1.5846103684382133,
+      "grad_norm": 0.1271483451128006,
+      "learning_rate": 0.0001,
+      "loss": 0.1933,
+      "step": 972
+    },
+    {
+      "epoch": 1.586240626018911,
+      "grad_norm": 0.12362337112426758,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 973
+    },
+    {
+      "epoch": 1.5878708835996087,
+      "grad_norm": 0.12856562435626984,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 974
+    },
+    {
+      "epoch": 1.5895011411803064,
+      "grad_norm": 0.10737952589988708,
+      "learning_rate": 0.0001,
+      "loss": 0.1892,
+      "step": 975
+    },
+    {
+      "epoch": 1.5911313987610043,
+      "grad_norm": 0.1198209896683693,
+      "learning_rate": 0.0001,
+      "loss": 0.1995,
+      "step": 976
+    },
+    {
+      "epoch": 1.592761656341702,
+      "grad_norm": 0.1190398558974266,
+      "learning_rate": 0.0001,
+      "loss": 0.1832,
+      "step": 977
+    },
+    {
+      "epoch": 1.5943919139223999,
+      "grad_norm": 0.1241382509469986,
+      "learning_rate": 0.0001,
+      "loss": 0.1949,
+      "step": 978
+    },
+    {
+      "epoch": 1.5960221715030976,
+      "grad_norm": 0.10942984372377396,
+      "learning_rate": 0.0001,
+      "loss": 0.187,
+      "step": 979
+    },
+    {
+      "epoch": 1.5976524290837952,
+      "grad_norm": 0.16293250024318695,
+      "learning_rate": 0.0001,
+      "loss": 0.1952,
+      "step": 980
+    },
+    {
+      "epoch": 1.599282686664493,
+      "grad_norm": 0.1106327474117279,
+      "learning_rate": 0.0001,
+      "loss": 0.1839,
+      "step": 981
+    },
+    {
+      "epoch": 1.6009129442451906,
+      "grad_norm": 0.15175805985927582,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 982
+    },
+    {
+      "epoch": 1.6025432018258885,
+      "grad_norm": 0.13483884930610657,
+      "learning_rate": 0.0001,
+      "loss": 0.195,
+      "step": 983
+    },
+    {
+      "epoch": 1.6041734594065864,
+      "grad_norm": 0.1230054423213005,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 984
+    },
+    {
+      "epoch": 1.605803716987284,
+      "grad_norm": 0.12960274517536163,
+      "learning_rate": 0.0001,
+      "loss": 0.1969,
+      "step": 985
+    },
+    {
+      "epoch": 1.6074339745679818,
+      "grad_norm": 0.12781794369220734,
+      "learning_rate": 0.0001,
+      "loss": 0.1961,
+      "step": 986
+    },
+    {
+      "epoch": 1.6090642321486794,
+      "grad_norm": 0.12113626301288605,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 987
+    },
+    {
+      "epoch": 1.6106944897293771,
+      "grad_norm": 0.10890072584152222,
+      "learning_rate": 0.0001,
+      "loss": 0.1975,
+      "step": 988
+    },
+    {
+      "epoch": 1.612324747310075,
+      "grad_norm": 0.12309688329696655,
+      "learning_rate": 0.0001,
+      "loss": 0.1866,
+      "step": 989
+    },
+    {
+      "epoch": 1.6139550048907727,
+      "grad_norm": 0.12740392982959747,
+      "learning_rate": 0.0001,
+      "loss": 0.1993,
+      "step": 990
+    },
+    {
+      "epoch": 1.6155852624714706,
+      "grad_norm": 0.12641726434230804,
+      "learning_rate": 0.0001,
+      "loss": 0.2091,
+      "step": 991
+    },
+    {
+      "epoch": 1.6172155200521683,
+      "grad_norm": 0.13405801355838776,
+      "learning_rate": 0.0001,
+      "loss": 0.2103,
+      "step": 992
+    },
+    {
+      "epoch": 1.618845777632866,
+      "grad_norm": 0.10853724181652069,
+      "learning_rate": 0.0001,
+      "loss": 0.2,
+      "step": 993
+    },
+    {
+      "epoch": 1.6204760352135636,
+      "grad_norm": 0.13972024619579315,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 994
+    },
+    {
+      "epoch": 1.6221062927942613,
+      "grad_norm": 0.11462700366973877,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 995
+    },
+    {
+      "epoch": 1.6237365503749592,
+      "grad_norm": 0.11051095277070999,
+      "learning_rate": 0.0001,
+      "loss": 0.1897,
+      "step": 996
+    },
+    {
+      "epoch": 1.6253668079556571,
+      "grad_norm": 0.11667358130216599,
+      "learning_rate": 0.0001,
+      "loss": 0.1997,
+      "step": 997
+    },
+    {
+      "epoch": 1.6269970655363548,
+      "grad_norm": 0.11986871063709259,
+      "learning_rate": 0.0001,
+      "loss": 0.1857,
+      "step": 998
+    },
+    {
+      "epoch": 1.6286273231170525,
+      "grad_norm": 0.10935333371162415,
+      "learning_rate": 0.0001,
+      "loss": 0.2016,
+      "step": 999
+    },
+    {
+      "epoch": 1.6302575806977502,
+      "grad_norm": 0.12035983800888062,
+      "learning_rate": 0.0001,
+      "loss": 0.2016,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6302575806977502,
+      "eval_loss": 0.2019711285829544,
+      "eval_runtime": 461.1407,
+      "eval_samples_per_second": 4.092,
+      "eval_steps_per_second": 1.024,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6318878382784479,
+      "grad_norm": 0.165378138422966,
+      "learning_rate": 0.0001,
+      "loss": 0.1832,
+      "step": 1001
+    },
+    {
+      "epoch": 1.6335180958591458,
+      "grad_norm": 0.12353488057851791,
+      "learning_rate": 0.0001,
+      "loss": 0.1881,
+      "step": 1002
+    },
+    {
+      "epoch": 1.6351483534398437,
+      "grad_norm": 0.13291586935520172,
+      "learning_rate": 0.0001,
+      "loss": 0.1882,
+      "step": 1003
+    },
+    {
+      "epoch": 1.6367786110205413,
+      "grad_norm": 0.1183796152472496,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1004
+    },
+    {
+      "epoch": 1.638408868601239,
+      "grad_norm": 0.14914032816886902,
+      "learning_rate": 0.0001,
+      "loss": 0.1938,
+      "step": 1005
+    },
+    {
+      "epoch": 1.6400391261819367,
+      "grad_norm": 0.13322560489177704,
+      "learning_rate": 0.0001,
+      "loss": 0.193,
+      "step": 1006
+    },
+    {
+      "epoch": 1.6416693837626344,
+      "grad_norm": 0.14066238701343536,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 1007
+    },
+    {
+      "epoch": 1.6432996413433323,
+      "grad_norm": 0.11830136179924011,
+      "learning_rate": 0.0001,
+      "loss": 0.1775,
+      "step": 1008
+    },
+    {
+      "epoch": 1.64492989892403,
+      "grad_norm": 0.11095772683620453,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 1009
+    },
+    {
+      "epoch": 1.6465601565047279,
+      "grad_norm": 0.12288574874401093,
+      "learning_rate": 0.0001,
+      "loss": 0.1918,
+      "step": 1010
+    },
+    {
+      "epoch": 1.6481904140854255,
+      "grad_norm": 0.11359456181526184,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 1011
+    },
+    {
+      "epoch": 1.6498206716661232,
+      "grad_norm": 0.11324212700128555,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 1012
+    },
+    {
+      "epoch": 1.651450929246821,
+      "grad_norm": 0.11443588137626648,
+      "learning_rate": 0.0001,
+      "loss": 0.2122,
+      "step": 1013
+    },
+    {
+      "epoch": 1.6530811868275186,
+      "grad_norm": 0.14412713050842285,
+      "learning_rate": 0.0001,
+      "loss": 0.1851,
+      "step": 1014
+    },
+    {
+      "epoch": 1.6547114444082165,
+      "grad_norm": 0.10094897449016571,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 1015
+    },
+    {
+      "epoch": 1.6563417019889144,
+      "grad_norm": 0.12090327590703964,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1016
+    },
+    {
+      "epoch": 1.657971959569612,
+      "grad_norm": 0.13659027218818665,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 1017
+    },
+    {
+      "epoch": 1.6596022171503098,
+      "grad_norm": 0.12800638377666473,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 1018
+    },
+    {
+      "epoch": 1.6612324747310074,
+      "grad_norm": 0.11905695497989655,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 1019
+    },
+    {
+      "epoch": 1.6628627323117051,
+      "grad_norm": 0.1223512515425682,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 1020
+    },
+    {
+      "epoch": 1.664492989892403,
+      "grad_norm": 0.12449689954519272,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 1021
+    },
+    {
+      "epoch": 1.6661232474731007,
+      "grad_norm": 0.11523964256048203,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 1022
+    },
+    {
+      "epoch": 1.6677535050537986,
+      "grad_norm": 0.1144275814294815,
+      "learning_rate": 0.0001,
+      "loss": 0.1925,
+      "step": 1023
+    },
+    {
+      "epoch": 1.6693837626344963,
+      "grad_norm": 0.11656132340431213,
+      "learning_rate": 0.0001,
+      "loss": 0.1766,
+      "step": 1024
+    },
+    {
+      "epoch": 1.671014020215194,
+      "grad_norm": 0.10800933092832565,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1025
+    },
+    {
+      "epoch": 1.6726442777958916,
+      "grad_norm": 0.12445061653852463,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 1026
+    },
+    {
+      "epoch": 1.6742745353765895,
+      "grad_norm": 0.11647329479455948,
+      "learning_rate": 0.0001,
+      "loss": 0.1942,
+      "step": 1027
+    },
+    {
+      "epoch": 1.6759047929572872,
+      "grad_norm": 0.11205076426267624,
+      "learning_rate": 0.0001,
+      "loss": 0.1746,
+      "step": 1028
+    },
+    {
+      "epoch": 1.6775350505379851,
+      "grad_norm": 0.12594562768936157,
+      "learning_rate": 0.0001,
+      "loss": 0.1758,
+      "step": 1029
+    },
+    {
+      "epoch": 1.6791653081186828,
+      "grad_norm": 0.13459977507591248,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 1030
+    },
+    {
+      "epoch": 1.6807955656993805,
+      "grad_norm": 0.12494435161352158,
+      "learning_rate": 0.0001,
+      "loss": 0.2046,
+      "step": 1031
+    },
+    {
+      "epoch": 1.6824258232800782,
+      "grad_norm": 0.11778414994478226,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 1032
+    },
+    {
+      "epoch": 1.6840560808607759,
+      "grad_norm": 0.10984843969345093,
+      "learning_rate": 0.0001,
+      "loss": 0.1893,
+      "step": 1033
+    },
+    {
+      "epoch": 1.6856863384414738,
+      "grad_norm": 0.6984158754348755,
+      "learning_rate": 0.0001,
+      "loss": 0.2081,
+      "step": 1034
+    },
+    {
+      "epoch": 1.6873165960221717,
+      "grad_norm": 0.10825484246015549,
+      "learning_rate": 0.0001,
+      "loss": 0.1846,
+      "step": 1035
+    },
+    {
+      "epoch": 1.6889468536028693,
+      "grad_norm": 0.1061546802520752,
+      "learning_rate": 0.0001,
+      "loss": 0.1866,
+      "step": 1036
+    },
+    {
+      "epoch": 1.690577111183567,
+      "grad_norm": 0.11936734616756439,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 1037
+    },
+    {
+      "epoch": 1.6922073687642647,
+      "grad_norm": 0.13179874420166016,
+      "learning_rate": 0.0001,
+      "loss": 0.1872,
+      "step": 1038
+    },
+    {
+      "epoch": 1.6938376263449624,
+      "grad_norm": 0.10697422921657562,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 1039
+    },
+    {
+      "epoch": 1.6954678839256603,
+      "grad_norm": 0.1438966691493988,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 1040
+    },
+    {
+      "epoch": 1.697098141506358,
+      "grad_norm": 0.10862574726343155,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 1041
+    },
+    {
+      "epoch": 1.6987283990870559,
+      "grad_norm": 0.10851150006055832,
+      "learning_rate": 0.0001,
+      "loss": 0.1918,
+      "step": 1042
+    },
+    {
+      "epoch": 1.7003586566677535,
+      "grad_norm": 0.11880330741405487,
+      "learning_rate": 0.0001,
+      "loss": 0.1802,
+      "step": 1043
+    },
+    {
+      "epoch": 1.7019889142484512,
+      "grad_norm": 0.12955494225025177,
+      "learning_rate": 0.0001,
+      "loss": 0.1981,
+      "step": 1044
+    },
+    {
+      "epoch": 1.703619171829149,
+      "grad_norm": 0.1307879388332367,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 1045
+    },
+    {
+      "epoch": 1.7052494294098468,
+      "grad_norm": 0.12933744490146637,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 1046
+    },
+    {
+      "epoch": 1.7068796869905445,
+      "grad_norm": 0.11336127668619156,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1047
+    },
+    {
+      "epoch": 1.7085099445712424,
+      "grad_norm": 0.1487550288438797,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 1048
+    },
+    {
+      "epoch": 1.71014020215194,
+      "grad_norm": 0.1337575614452362,
+      "learning_rate": 0.0001,
+      "loss": 0.1934,
+      "step": 1049
+    },
+    {
+      "epoch": 1.7117704597326378,
+      "grad_norm": 0.13716711103916168,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 1050
+    },
+    {
+      "epoch": 1.7134007173133354,
+      "grad_norm": 0.14090760052204132,
+      "learning_rate": 0.0001,
+      "loss": 0.1993,
+      "step": 1051
+    },
+    {
+      "epoch": 1.7150309748940331,
+      "grad_norm": 0.11731863021850586,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 1052
+    },
+    {
+      "epoch": 1.716661232474731,
+      "grad_norm": 0.13154546916484833,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1053
+    },
+    {
+      "epoch": 1.718291490055429,
+      "grad_norm": 0.11974471062421799,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "step": 1054
+    },
+    {
+      "epoch": 1.7199217476361266,
+      "grad_norm": 0.11386546492576599,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 1055
+    },
+    {
+      "epoch": 1.7215520052168243,
+      "grad_norm": 0.09487878531217575,
+      "learning_rate": 0.0001,
+      "loss": 0.1891,
+      "step": 1056
+    },
+    {
+      "epoch": 1.723182262797522,
+      "grad_norm": 0.13436681032180786,
+      "learning_rate": 0.0001,
+      "loss": 0.2091,
+      "step": 1057
+    },
+    {
+      "epoch": 1.7248125203782196,
+      "grad_norm": 0.11488918960094452,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1058
+    },
+    {
+      "epoch": 1.7264427779589175,
+      "grad_norm": 0.1182805746793747,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 1059
+    },
+    {
+      "epoch": 1.7280730355396152,
+      "grad_norm": 0.11342515796422958,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 1060
+    },
+    {
+      "epoch": 1.7297032931203131,
+      "grad_norm": 0.14482362568378448,
+      "learning_rate": 0.0001,
+      "loss": 0.2023,
+      "step": 1061
+    },
+    {
+      "epoch": 1.7313335507010108,
+      "grad_norm": 0.12669838964939117,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 1062
+    },
+    {
+      "epoch": 1.7329638082817085,
+      "grad_norm": 0.12242146581411362,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1063
+    },
+    {
+      "epoch": 1.7345940658624062,
+      "grad_norm": 0.11942725628614426,
+      "learning_rate": 0.0001,
+      "loss": 0.1975,
+      "step": 1064
+    },
+    {
+      "epoch": 1.7362243234431038,
+      "grad_norm": 0.14859893918037415,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 1065
+    },
+    {
+      "epoch": 1.7378545810238017,
+      "grad_norm": 0.13270403444766998,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 1066
+    },
+    {
+      "epoch": 1.7394848386044997,
+      "grad_norm": 0.13862530887126923,
+      "learning_rate": 0.0001,
+      "loss": 0.1901,
+      "step": 1067
+    },
+    {
+      "epoch": 1.7411150961851973,
+      "grad_norm": 0.13995380699634552,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 1068
+    },
+    {
+      "epoch": 1.742745353765895,
+      "grad_norm": 0.11337511986494064,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 1069
+    },
+    {
+      "epoch": 1.7443756113465927,
+      "grad_norm": 0.12498489022254944,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 1070
+    },
+    {
+      "epoch": 1.7460058689272904,
+      "grad_norm": 0.127224400639534,
+      "learning_rate": 0.0001,
+      "loss": 0.1939,
+      "step": 1071
+    },
+    {
+      "epoch": 1.7476361265079883,
+      "grad_norm": 0.11792348325252533,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 1072
+    },
+    {
+      "epoch": 1.7492663840886862,
+      "grad_norm": 0.10551140457391739,
+      "learning_rate": 0.0001,
+      "loss": 0.1702,
+      "step": 1073
+    },
+    {
+      "epoch": 1.7508966416693839,
+      "grad_norm": 0.11569567769765854,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 1074
+    },
+    {
+      "epoch": 1.7525268992500815,
+      "grad_norm": 0.12134891003370285,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 1075
+    },
+    {
+      "epoch": 1.7541571568307792,
+      "grad_norm": 0.10986746102571487,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 1076
+    },
+    {
+      "epoch": 1.755787414411477,
+      "grad_norm": 0.12267926335334778,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 1077
+    },
+    {
+      "epoch": 1.7574176719921748,
+      "grad_norm": 0.11747080087661743,
+      "learning_rate": 0.0001,
+      "loss": 0.1897,
+      "step": 1078
+    },
+    {
+      "epoch": 1.7590479295728725,
+      "grad_norm": 0.12816251814365387,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1079
+    },
+    {
+      "epoch": 1.7606781871535704,
+      "grad_norm": 0.11122124642133713,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 1080
+    },
+    {
+      "epoch": 1.762308444734268,
+      "grad_norm": 0.18602591753005981,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1081
+    },
+    {
+      "epoch": 1.7639387023149657,
+      "grad_norm": 0.11141426861286163,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 1082
+    },
+    {
+      "epoch": 1.7655689598956634,
+      "grad_norm": 0.13159684836864471,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 1083
+    },
+    {
+      "epoch": 1.767199217476361,
+      "grad_norm": 0.12319090962409973,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 1084
+    },
+    {
+      "epoch": 1.768829475057059,
+      "grad_norm": 0.12106230854988098,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 1085
+    },
+    {
+      "epoch": 1.770459732637757,
+      "grad_norm": 0.10471806675195694,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 1086
+    },
+    {
+      "epoch": 1.7720899902184546,
+      "grad_norm": 0.12956561148166656,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 1087
+    },
+    {
+      "epoch": 1.7737202477991523,
+      "grad_norm": 0.12020842730998993,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 1088
+    },
+    {
+      "epoch": 1.77535050537985,
+      "grad_norm": 0.12040827423334122,
+      "learning_rate": 0.0001,
+      "loss": 0.194,
+      "step": 1089
+    },
+    {
+      "epoch": 1.7769807629605476,
+      "grad_norm": 0.11612360924482346,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1090
+    },
+    {
+      "epoch": 1.7786110205412455,
+      "grad_norm": 0.11711908876895905,
+      "learning_rate": 0.0001,
+      "loss": 0.1898,
+      "step": 1091
+    },
+    {
+      "epoch": 1.7802412781219432,
+      "grad_norm": 0.13765960931777954,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 1092
+    },
+    {
+      "epoch": 1.7818715357026411,
+      "grad_norm": 0.12794670462608337,
+      "learning_rate": 0.0001,
+      "loss": 0.1918,
+      "step": 1093
+    },
+    {
+      "epoch": 1.7835017932833388,
+      "grad_norm": 0.12181553989648819,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 1094
+    },
+    {
+      "epoch": 1.7851320508640365,
+      "grad_norm": 0.12774871289730072,
+      "learning_rate": 0.0001,
+      "loss": 0.2011,
+      "step": 1095
+    },
+    {
+      "epoch": 1.7867623084447342,
+      "grad_norm": 0.11980225145816803,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1096
+    },
+    {
+      "epoch": 1.788392566025432,
+      "grad_norm": 0.1250666081905365,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 1097
+    },
+    {
+      "epoch": 1.7900228236061297,
+      "grad_norm": 0.13532552123069763,
+      "learning_rate": 0.0001,
+      "loss": 0.1936,
+      "step": 1098
+    },
+    {
+      "epoch": 1.7916530811868276,
+      "grad_norm": 0.1301998496055603,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 1099
+    },
+    {
+      "epoch": 1.7932833387675253,
+      "grad_norm": 0.12082832306623459,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 1100
+    },
+    {
+      "epoch": 1.794913596348223,
+      "grad_norm": 0.12959355115890503,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 1101
+    },
+    {
+      "epoch": 1.7965438539289207,
+      "grad_norm": 0.1250586211681366,
+      "learning_rate": 0.0001,
+      "loss": 0.2007,
+      "step": 1102
+    },
+    {
+      "epoch": 1.7981741115096184,
+      "grad_norm": 0.1253538429737091,
+      "learning_rate": 0.0001,
+      "loss": 0.1903,
+      "step": 1103
+    },
+    {
+      "epoch": 1.7998043690903163,
+      "grad_norm": 0.09879365563392639,
+      "learning_rate": 0.0001,
+      "loss": 0.1746,
+      "step": 1104
+    },
+    {
+      "epoch": 1.8014346266710142,
+      "grad_norm": 0.10257220268249512,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 1105
+    },
+    {
+      "epoch": 1.8030648842517119,
+      "grad_norm": 0.10447902977466583,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 1106
+    },
+    {
+      "epoch": 1.8046951418324095,
+      "grad_norm": 0.10268338769674301,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 1107
+    },
+    {
+      "epoch": 1.8063253994131072,
+      "grad_norm": 0.12003128230571747,
+      "learning_rate": 0.0001,
+      "loss": 0.1872,
+      "step": 1108
+    },
+    {
+      "epoch": 1.807955656993805,
+      "grad_norm": 0.10243313759565353,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 1109
+    },
+    {
+      "epoch": 1.8095859145745028,
+      "grad_norm": 0.11695826798677444,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 1110
+    },
+    {
+      "epoch": 1.8112161721552005,
+      "grad_norm": 0.10703260451555252,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 1111
+    },
+    {
+      "epoch": 1.8128464297358984,
+      "grad_norm": 0.13467635214328766,
+      "learning_rate": 0.0001,
+      "loss": 0.1881,
+      "step": 1112
+    },
+    {
+      "epoch": 1.814476687316596,
+      "grad_norm": 0.11541393399238586,
+      "learning_rate": 0.0001,
+      "loss": 0.192,
+      "step": 1113
+    },
+    {
+      "epoch": 1.8161069448972937,
+      "grad_norm": 0.1386149823665619,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 1114
+    },
+    {
+      "epoch": 1.8177372024779914,
+      "grad_norm": 0.12145870178937912,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 1115
+    },
+    {
+      "epoch": 1.8193674600586893,
+      "grad_norm": 0.10783080011606216,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 1116
+    },
+    {
+      "epoch": 1.820997717639387,
+      "grad_norm": 0.10322029888629913,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 1117
+    },
+    {
+      "epoch": 1.822627975220085,
+      "grad_norm": 0.11130011826753616,
+      "learning_rate": 0.0001,
+      "loss": 0.1765,
+      "step": 1118
+    },
+    {
+      "epoch": 1.8242582328007826,
+      "grad_norm": 0.09618648141622543,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 1119
+    },
+    {
+      "epoch": 1.8258884903814803,
+      "grad_norm": 0.11504556983709335,
+      "learning_rate": 0.0001,
+      "loss": 0.1936,
+      "step": 1120
+    },
+    {
+      "epoch": 1.827518747962178,
+      "grad_norm": 0.11279092729091644,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 1121
+    },
+    {
+      "epoch": 1.8291490055428756,
+      "grad_norm": 0.14116674661636353,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 1122
+    },
+    {
+      "epoch": 1.8307792631235735,
+      "grad_norm": 0.11786951869726181,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "step": 1123
+    },
+    {
+      "epoch": 1.8324095207042714,
+      "grad_norm": 0.10708369314670563,
+      "learning_rate": 0.0001,
+      "loss": 0.195,
+      "step": 1124
+    },
+    {
+      "epoch": 1.8340397782849691,
+      "grad_norm": 0.18581970036029816,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 1125
+    },
+    {
+      "epoch": 1.8356700358656668,
+      "grad_norm": 0.11519227176904678,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1126
+    },
+    {
+      "epoch": 1.8373002934463645,
+      "grad_norm": 0.11876104772090912,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1127
+    },
+    {
+      "epoch": 1.8389305510270622,
+      "grad_norm": 0.10485244542360306,
+      "learning_rate": 0.0001,
+      "loss": 0.1813,
+      "step": 1128
+    },
+    {
+      "epoch": 1.84056080860776,
+      "grad_norm": 0.12142175436019897,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 1129
+    },
+    {
+      "epoch": 1.8421910661884577,
+      "grad_norm": 0.12452993541955948,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 1130
+    },
+    {
+      "epoch": 1.8438213237691556,
+      "grad_norm": 0.12156898528337479,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 1131
+    },
+    {
+      "epoch": 1.8454515813498533,
+      "grad_norm": 0.12071850895881653,
+      "learning_rate": 0.0001,
+      "loss": 0.1949,
+      "step": 1132
+    },
+    {
+      "epoch": 1.847081838930551,
+      "grad_norm": 0.12315471470355988,
+      "learning_rate": 0.0001,
+      "loss": 0.1965,
+      "step": 1133
+    },
+    {
+      "epoch": 1.8487120965112487,
+      "grad_norm": 0.11038068681955338,
+      "learning_rate": 0.0001,
+      "loss": 0.1868,
+      "step": 1134
+    },
+    {
+      "epoch": 1.8503423540919464,
+      "grad_norm": 0.10914380848407745,
+      "learning_rate": 0.0001,
+      "loss": 0.2023,
+      "step": 1135
+    },
+    {
+      "epoch": 1.8519726116726443,
+      "grad_norm": 0.12893681228160858,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 1136
+    },
+    {
+      "epoch": 1.8536028692533422,
+      "grad_norm": 0.10367900878190994,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 1137
+    },
+    {
+      "epoch": 1.8552331268340398,
+      "grad_norm": 0.11435539275407791,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 1138
+    },
+    {
+      "epoch": 1.8568633844147375,
+      "grad_norm": 0.11469294130802155,
+      "learning_rate": 0.0001,
+      "loss": 0.1936,
+      "step": 1139
+    },
+    {
+      "epoch": 1.8584936419954352,
+      "grad_norm": 0.1339050680398941,
+      "learning_rate": 0.0001,
+      "loss": 0.193,
+      "step": 1140
+    },
+    {
+      "epoch": 1.8601238995761329,
+      "grad_norm": 0.10412506759166718,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 1141
+    },
+    {
+      "epoch": 1.8617541571568308,
+      "grad_norm": 0.10916057229042053,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 1142
+    },
+    {
+      "epoch": 1.8633844147375287,
+      "grad_norm": 0.11035683751106262,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 1143
+    },
+    {
+      "epoch": 1.8650146723182264,
+      "grad_norm": 0.13064561784267426,
+      "learning_rate": 0.0001,
+      "loss": 0.1969,
+      "step": 1144
+    },
+    {
+      "epoch": 1.866644929898924,
+      "grad_norm": 0.11412317305803299,
+      "learning_rate": 0.0001,
+      "loss": 0.1915,
+      "step": 1145
+    },
+    {
+      "epoch": 1.8682751874796217,
+      "grad_norm": 0.11612699925899506,
+      "learning_rate": 0.0001,
+      "loss": 0.2052,
+      "step": 1146
+    },
+    {
+      "epoch": 1.8699054450603194,
+      "grad_norm": 0.1117459088563919,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 1147
+    },
+    {
+      "epoch": 1.8715357026410173,
+      "grad_norm": 0.1192610114812851,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 1148
+    },
+    {
+      "epoch": 1.873165960221715,
+      "grad_norm": 0.12196903675794601,
+      "learning_rate": 0.0001,
+      "loss": 0.2023,
+      "step": 1149
+    },
+    {
+      "epoch": 1.874796217802413,
+      "grad_norm": 0.11657217144966125,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 1150
+    },
+    {
+      "epoch": 1.8764264753831106,
+      "grad_norm": 0.14465437829494476,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1151
+    },
+    {
+      "epoch": 1.8780567329638083,
+      "grad_norm": 0.11741483211517334,
+      "learning_rate": 0.0001,
+      "loss": 0.1881,
+      "step": 1152
+    },
+    {
+      "epoch": 1.879686990544506,
+      "grad_norm": 0.12043508142232895,
+      "learning_rate": 0.0001,
+      "loss": 0.1929,
+      "step": 1153
+    },
+    {
+      "epoch": 1.8813172481252036,
+      "grad_norm": 0.14909182488918304,
+      "learning_rate": 0.0001,
+      "loss": 0.2239,
+      "step": 1154
+    },
+    {
+      "epoch": 1.8829475057059015,
+      "grad_norm": 0.12423070520162582,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 1155
+    },
+    {
+      "epoch": 1.8845777632865994,
+      "grad_norm": 0.11938756704330444,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 1156
+    },
+    {
+      "epoch": 1.886208020867297,
+      "grad_norm": 0.10287628322839737,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1157
+    },
+    {
+      "epoch": 1.8878382784479948,
+      "grad_norm": 0.09794101864099503,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1158
+    },
+    {
+      "epoch": 1.8894685360286925,
+      "grad_norm": 0.12463726103305817,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 1159
+    },
+    {
+      "epoch": 1.8910987936093901,
+      "grad_norm": 0.11366378515958786,
+      "learning_rate": 0.0001,
+      "loss": 0.1954,
+      "step": 1160
+    },
+    {
+      "epoch": 1.892729051190088,
+      "grad_norm": 0.1246451884508133,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 1161
+    },
+    {
+      "epoch": 1.8943593087707857,
+      "grad_norm": 0.13116668164730072,
+      "learning_rate": 0.0001,
+      "loss": 0.1821,
+      "step": 1162
+    },
+    {
+      "epoch": 1.8959895663514836,
+      "grad_norm": 0.104774110019207,
+      "learning_rate": 0.0001,
+      "loss": 0.1801,
+      "step": 1163
+    },
+    {
+      "epoch": 1.8976198239321813,
+      "grad_norm": 0.10643444210290909,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 1164
+    },
+    {
+      "epoch": 1.899250081512879,
+      "grad_norm": 0.11924157291650772,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1165
+    },
+    {
+      "epoch": 1.9008803390935767,
+      "grad_norm": 0.1133570596575737,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1166
+    },
+    {
+      "epoch": 1.9025105966742746,
+      "grad_norm": 0.11677032709121704,
+      "learning_rate": 0.0001,
+      "loss": 0.1876,
+      "step": 1167
+    },
+    {
+      "epoch": 1.9041408542549723,
+      "grad_norm": 0.11180409044027328,
+      "learning_rate": 0.0001,
+      "loss": 0.1787,
+      "step": 1168
+    },
+    {
+      "epoch": 1.9057711118356702,
+      "grad_norm": 0.10403387248516083,
+      "learning_rate": 0.0001,
+      "loss": 0.1816,
+      "step": 1169
+    },
+    {
+      "epoch": 1.9074013694163678,
+      "grad_norm": 0.1255159080028534,
+      "learning_rate": 0.0001,
+      "loss": 0.2094,
+      "step": 1170
+    },
+    {
+      "epoch": 1.9090316269970655,
+      "grad_norm": 0.1264365166425705,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 1171
+    },
+    {
+      "epoch": 1.9106618845777632,
+      "grad_norm": 0.12751294672489166,
+      "learning_rate": 0.0001,
+      "loss": 0.2049,
+      "step": 1172
+    },
+    {
+      "epoch": 1.9122921421584609,
+      "grad_norm": 0.10625582933425903,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1173
+    },
+    {
+      "epoch": 1.9139223997391588,
+      "grad_norm": 0.11829791963100433,
+      "learning_rate": 0.0001,
+      "loss": 0.1863,
+      "step": 1174
+    },
+    {
+      "epoch": 1.9155526573198567,
+      "grad_norm": 0.11165602505207062,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 1175
+    },
+    {
+      "epoch": 1.9171829149005544,
+      "grad_norm": 0.14051920175552368,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 1176
+    },
+    {
+      "epoch": 1.918813172481252,
+      "grad_norm": 0.11515091359615326,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1177
+    },
+    {
+      "epoch": 1.9204434300619497,
+      "grad_norm": 0.14358852803707123,
+      "learning_rate": 0.0001,
+      "loss": 0.1786,
+      "step": 1178
+    },
+    {
+      "epoch": 1.9220736876426474,
+      "grad_norm": 0.1013508066534996,
+      "learning_rate": 0.0001,
+      "loss": 0.1724,
+      "step": 1179
+    },
+    {
+      "epoch": 1.9237039452233453,
+      "grad_norm": 0.12144607305526733,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1180
+    },
+    {
+      "epoch": 1.925334202804043,
+      "grad_norm": 0.1413160115480423,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1181
+    },
+    {
+      "epoch": 1.926964460384741,
+      "grad_norm": 0.14951057732105255,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 1182
+    },
+    {
+      "epoch": 1.9285947179654386,
+      "grad_norm": 0.11757660657167435,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1183
+    },
+    {
+      "epoch": 1.9302249755461363,
+      "grad_norm": 0.1191357746720314,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 1184
+    },
+    {
+      "epoch": 1.931855233126834,
+      "grad_norm": 0.12074853479862213,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1185
+    },
+    {
+      "epoch": 1.9334854907075318,
+      "grad_norm": 0.11659263074398041,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 1186
+    },
+    {
+      "epoch": 1.9351157482882295,
+      "grad_norm": 0.10916323959827423,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 1187
+    },
+    {
+      "epoch": 1.9367460058689274,
+      "grad_norm": 0.10679040104150772,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1188
+    },
+    {
+      "epoch": 1.938376263449625,
+      "grad_norm": 0.13911648094654083,
+      "learning_rate": 0.0001,
+      "loss": 0.1896,
+      "step": 1189
+    },
+    {
+      "epoch": 1.9400065210303228,
+      "grad_norm": 0.09244771301746368,
+      "learning_rate": 0.0001,
+      "loss": 0.1823,
+      "step": 1190
+    },
+    {
+      "epoch": 1.9416367786110205,
+      "grad_norm": 0.14823158085346222,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 1191
+    },
+    {
+      "epoch": 1.9432670361917181,
+      "grad_norm": 0.1267862170934677,
+      "learning_rate": 0.0001,
+      "loss": 0.1908,
+      "step": 1192
+    },
+    {
+      "epoch": 1.944897293772416,
+      "grad_norm": 0.12872754037380219,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 1193
+    },
+    {
+      "epoch": 1.946527551353114,
+      "grad_norm": 0.15068793296813965,
+      "learning_rate": 0.0001,
+      "loss": 0.1919,
+      "step": 1194
+    },
+    {
+      "epoch": 1.9481578089338116,
+      "grad_norm": 0.1112789437174797,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 1195
+    },
+    {
+      "epoch": 1.9497880665145093,
+      "grad_norm": 0.12267830222845078,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 1196
+    },
+    {
+      "epoch": 1.951418324095207,
+      "grad_norm": 0.13730347156524658,
+      "learning_rate": 0.0001,
+      "loss": 0.1858,
+      "step": 1197
+    },
+    {
+      "epoch": 1.9530485816759047,
+      "grad_norm": 0.11384674906730652,
+      "learning_rate": 0.0001,
+      "loss": 0.1929,
+      "step": 1198
+    },
+    {
+      "epoch": 1.9546788392566026,
+      "grad_norm": 0.15709474682807922,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 1199
+    },
+    {
+      "epoch": 1.9563090968373003,
+      "grad_norm": 0.15006817877292633,
+      "learning_rate": 0.0001,
+      "loss": 0.1886,
+      "step": 1200
+    },
+    {
+      "epoch": 1.9579393544179982,
+      "grad_norm": 0.11046171188354492,
+      "learning_rate": 0.0001,
+      "loss": 0.1863,
+      "step": 1201
+    },
+    {
+      "epoch": 1.9595696119986958,
+      "grad_norm": 0.10871347039937973,
+      "learning_rate": 0.0001,
+      "loss": 0.194,
+      "step": 1202
+    },
+    {
+      "epoch": 1.9611998695793935,
+      "grad_norm": 0.12221578508615494,
+      "learning_rate": 0.0001,
+      "loss": 0.1885,
+      "step": 1203
+    },
+    {
+      "epoch": 1.9628301271600912,
+      "grad_norm": 0.11434303224086761,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 1204
+    },
+    {
+      "epoch": 1.9644603847407889,
+      "grad_norm": 0.12363463640213013,
+      "learning_rate": 0.0001,
+      "loss": 0.1999,
+      "step": 1205
+    },
+    {
+      "epoch": 1.9660906423214868,
+      "grad_norm": 0.1334676444530487,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 1206
+    },
+    {
+      "epoch": 1.9677208999021847,
+      "grad_norm": 0.1140584945678711,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1207
+    },
+    {
+      "epoch": 1.9693511574828824,
+      "grad_norm": 0.11625649034976959,
+      "learning_rate": 0.0001,
+      "loss": 0.212,
+      "step": 1208
+    },
+    {
+      "epoch": 1.97098141506358,
+      "grad_norm": 0.10492118448019028,
+      "learning_rate": 0.0001,
+      "loss": 0.1825,
+      "step": 1209
+    },
+    {
+      "epoch": 1.9726116726442777,
+      "grad_norm": 0.11723576486110687,
+      "learning_rate": 0.0001,
+      "loss": 0.1995,
+      "step": 1210
+    },
+    {
+      "epoch": 1.9742419302249754,
+      "grad_norm": 0.10941202193498611,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 1211
+    },
+    {
+      "epoch": 1.9758721878056733,
+      "grad_norm": 0.11242581903934479,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 1212
+    },
+    {
+      "epoch": 1.9775024453863712,
+      "grad_norm": 0.1318778693675995,
+      "learning_rate": 0.0001,
+      "loss": 0.1958,
+      "step": 1213
+    },
+    {
+      "epoch": 1.979132702967069,
+      "grad_norm": 0.12168639153242111,
+      "learning_rate": 0.0001,
+      "loss": 0.1882,
+      "step": 1214
+    },
+    {
+      "epoch": 1.9807629605477666,
+      "grad_norm": 0.1154153048992157,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 1215
+    },
+    {
+      "epoch": 1.9823932181284643,
+      "grad_norm": 0.12487414479255676,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 1216
+    },
+    {
+      "epoch": 1.984023475709162,
+      "grad_norm": 0.14309844374656677,
+      "learning_rate": 0.0001,
+      "loss": 0.199,
+      "step": 1217
+    },
+    {
+      "epoch": 1.9856537332898598,
+      "grad_norm": 0.127132847905159,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1218
+    },
+    {
+      "epoch": 1.9872839908705575,
+      "grad_norm": 0.1264181286096573,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 1219
+    },
+    {
+      "epoch": 1.9889142484512554,
+      "grad_norm": 0.12706303596496582,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 1220
+    },
+    {
+      "epoch": 1.990544506031953,
+      "grad_norm": 0.12314368784427643,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 1221
+    },
+    {
+      "epoch": 1.9921747636126508,
+      "grad_norm": 0.11701434850692749,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 1222
+    },
+    {
+      "epoch": 1.9938050211933485,
+      "grad_norm": 0.1571284830570221,
+      "learning_rate": 0.0001,
+      "loss": 0.1791,
+      "step": 1223
+    },
+    {
+      "epoch": 1.9954352787740461,
+      "grad_norm": 0.12066090106964111,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 1224
+    },
+    {
+      "epoch": 1.997065536354744,
+      "grad_norm": 0.1215023398399353,
+      "learning_rate": 0.0001,
+      "loss": 0.2005,
+      "step": 1225
+    },
+    {
+      "epoch": 1.998695793935442,
+      "grad_norm": 0.11120277643203735,
+      "learning_rate": 0.0001,
+      "loss": 0.1825,
+      "step": 1226
+    },
+    {
+      "epoch": 2.0003260515161396,
+      "grad_norm": 0.14585649967193604,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "step": 1227
+    },
+    {
+      "epoch": 2.0019563090968373,
+      "grad_norm": 0.10910741984844208,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 1228
+    },
+    {
+      "epoch": 2.003586566677535,
+      "grad_norm": 0.1540917009115219,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1229
+    },
+    {
+      "epoch": 2.0052168242582327,
+      "grad_norm": 0.12682808935642242,
+      "learning_rate": 0.0001,
+      "loss": 0.1778,
+      "step": 1230
+    },
+    {
+      "epoch": 2.0068470818389303,
+      "grad_norm": 0.12131527066230774,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1231
+    },
+    {
+      "epoch": 2.0084773394196285,
+      "grad_norm": 0.13246504962444305,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 1232
+    },
+    {
+      "epoch": 2.010107597000326,
+      "grad_norm": 0.13079266250133514,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1233
+    },
+    {
+      "epoch": 2.011737854581024,
+      "grad_norm": 0.13317003846168518,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 1234
+    },
+    {
+      "epoch": 2.0133681121617215,
+      "grad_norm": 0.15676195919513702,
+      "learning_rate": 0.0001,
+      "loss": 0.1738,
+      "step": 1235
+    },
+    {
+      "epoch": 2.014998369742419,
+      "grad_norm": 0.12478923052549362,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1236
+    },
+    {
+      "epoch": 2.016628627323117,
+      "grad_norm": 0.15440379083156586,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 1237
+    },
+    {
+      "epoch": 2.018258884903815,
+      "grad_norm": 0.13177958130836487,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 1238
+    },
+    {
+      "epoch": 2.0198891424845127,
+      "grad_norm": 0.1117987260222435,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 1239
+    },
+    {
+      "epoch": 2.0215194000652104,
+      "grad_norm": 0.15712931752204895,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 1240
+    },
+    {
+      "epoch": 2.023149657645908,
+      "grad_norm": 0.11659948527812958,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 1241
+    },
+    {
+      "epoch": 2.0247799152266057,
+      "grad_norm": 0.08816540241241455,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 1242
+    },
+    {
+      "epoch": 2.0264101728073034,
+      "grad_norm": 0.13327382504940033,
+      "learning_rate": 0.0001,
+      "loss": 0.1814,
+      "step": 1243
+    },
+    {
+      "epoch": 2.0280404303880015,
+      "grad_norm": 0.11885062605142593,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 1244
+    },
+    {
+      "epoch": 2.029670687968699,
+      "grad_norm": 0.15437649190425873,
+      "learning_rate": 0.0001,
+      "loss": 0.1726,
+      "step": 1245
+    },
+    {
+      "epoch": 2.031300945549397,
+      "grad_norm": 0.13453203439712524,
+      "learning_rate": 0.0001,
+      "loss": 0.1823,
+      "step": 1246
+    },
+    {
+      "epoch": 2.0329312031300946,
+      "grad_norm": 0.12537498772144318,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 1247
+    },
+    {
+      "epoch": 2.0345614607107922,
+      "grad_norm": 0.12894268333911896,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 1248
+    },
+    {
+      "epoch": 2.03619171829149,
+      "grad_norm": 0.1385597288608551,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 1249
+    },
+    {
+      "epoch": 2.0378219758721876,
+      "grad_norm": 0.15800078213214874,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1250
+    },
+    {
+      "epoch": 2.0394522334528857,
+      "grad_norm": 0.09816145896911621,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 1251
+    },
+    {
+      "epoch": 2.0410824910335834,
+      "grad_norm": 0.1402331292629242,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1252
+    },
+    {
+      "epoch": 2.042712748614281,
+      "grad_norm": 0.13870537281036377,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1253
+    },
+    {
+      "epoch": 2.0443430061949788,
+      "grad_norm": 0.1497325599193573,
+      "learning_rate": 0.0001,
+      "loss": 0.1896,
+      "step": 1254
+    },
+    {
+      "epoch": 2.0459732637756765,
+      "grad_norm": 0.1422789841890335,
+      "learning_rate": 0.0001,
+      "loss": 0.1776,
+      "step": 1255
+    },
+    {
+      "epoch": 2.047603521356374,
+      "grad_norm": 0.12478403002023697,
+      "learning_rate": 0.0001,
+      "loss": 0.1802,
+      "step": 1256
+    },
+    {
+      "epoch": 2.0492337789370723,
+      "grad_norm": 0.12358732521533966,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 1257
+    },
+    {
+      "epoch": 2.05086403651777,
+      "grad_norm": 0.11842041462659836,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1258
+    },
+    {
+      "epoch": 2.0524942940984676,
+      "grad_norm": 0.11563142389059067,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 1259
+    },
+    {
+      "epoch": 2.0541245516791653,
+      "grad_norm": 0.12804879248142242,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 1260
+    },
+    {
+      "epoch": 2.055754809259863,
+      "grad_norm": 0.13581717014312744,
+      "learning_rate": 0.0001,
+      "loss": 0.1779,
+      "step": 1261
+    },
+    {
+      "epoch": 2.0573850668405607,
+      "grad_norm": 0.1498471200466156,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 1262
+    },
+    {
+      "epoch": 2.0590153244212583,
+      "grad_norm": 0.11999404430389404,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 1263
+    },
+    {
+      "epoch": 2.0606455820019565,
+      "grad_norm": 0.15887455642223358,
+      "learning_rate": 0.0001,
+      "loss": 0.1803,
+      "step": 1264
+    },
+    {
+      "epoch": 2.062275839582654,
+      "grad_norm": 0.139693021774292,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 1265
+    },
+    {
+      "epoch": 2.063906097163352,
+      "grad_norm": 0.15679016709327698,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 1266
+    },
+    {
+      "epoch": 2.0655363547440495,
+      "grad_norm": 0.1563042551279068,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 1267
+    },
+    {
+      "epoch": 2.067166612324747,
+      "grad_norm": 0.13360831141471863,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 1268
+    },
+    {
+      "epoch": 2.068796869905445,
+      "grad_norm": 0.123869888484478,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 1269
+    },
+    {
+      "epoch": 2.070427127486143,
+      "grad_norm": 0.1273512840270996,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 1270
+    },
+    {
+      "epoch": 2.0720573850668407,
+      "grad_norm": 0.1433614194393158,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1271
+    },
+    {
+      "epoch": 2.0736876426475384,
+      "grad_norm": 0.1275768131017685,
+      "learning_rate": 0.0001,
+      "loss": 0.1742,
+      "step": 1272
+    },
+    {
+      "epoch": 2.075317900228236,
+      "grad_norm": 0.113472118973732,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 1273
+    },
+    {
+      "epoch": 2.0769481578089337,
+      "grad_norm": 0.14065606892108917,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 1274
+    },
+    {
+      "epoch": 2.0785784153896314,
+      "grad_norm": 0.10859358310699463,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 1275
+    },
+    {
+      "epoch": 2.0802086729703295,
+      "grad_norm": 0.12903261184692383,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 1276
+    },
+    {
+      "epoch": 2.081838930551027,
+      "grad_norm": 0.1369621455669403,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 1277
+    },
+    {
+      "epoch": 2.083469188131725,
+      "grad_norm": 0.13974396884441376,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 1278
+    },
+    {
+      "epoch": 2.0850994457124226,
+      "grad_norm": 0.1181628480553627,
+      "learning_rate": 0.0001,
+      "loss": 0.1778,
+      "step": 1279
+    },
+    {
+      "epoch": 2.0867297032931202,
+      "grad_norm": 0.14868682622909546,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1280
+    },
+    {
+      "epoch": 2.088359960873818,
+      "grad_norm": 0.1333058476448059,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 1281
+    },
+    {
+      "epoch": 2.0899902184545156,
+      "grad_norm": 0.14653992652893066,
+      "learning_rate": 0.0001,
+      "loss": 0.1805,
+      "step": 1282
+    },
+    {
+      "epoch": 2.0916204760352137,
+      "grad_norm": 0.15411017835140228,
+      "learning_rate": 0.0001,
+      "loss": 0.1601,
+      "step": 1283
+    },
+    {
+      "epoch": 2.0932507336159114,
+      "grad_norm": 0.1162109524011612,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 1284
+    },
+    {
+      "epoch": 2.094880991196609,
+      "grad_norm": 0.14571483433246613,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1285
+    },
+    {
+      "epoch": 2.0965112487773068,
+      "grad_norm": 0.15159551799297333,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 1286
+    },
+    {
+      "epoch": 2.0981415063580044,
+      "grad_norm": 0.1452435553073883,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 1287
+    },
+    {
+      "epoch": 2.099771763938702,
+      "grad_norm": 0.1417376548051834,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1288
+    },
+    {
+      "epoch": 2.1014020215194003,
+      "grad_norm": 0.15437503159046173,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1289
+    },
+    {
+      "epoch": 2.103032279100098,
+      "grad_norm": 0.13350600004196167,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 1290
+    },
+    {
+      "epoch": 2.1046625366807956,
+      "grad_norm": 0.12360616028308868,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 1291
+    },
+    {
+      "epoch": 2.1062927942614933,
+      "grad_norm": 0.12257912009954453,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 1292
+    },
+    {
+      "epoch": 2.107923051842191,
+      "grad_norm": 0.1374562382698059,
+      "learning_rate": 0.0001,
+      "loss": 0.1786,
+      "step": 1293
+    },
+    {
+      "epoch": 2.1095533094228887,
+      "grad_norm": 0.17024345695972443,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 1294
+    },
+    {
+      "epoch": 2.111183567003587,
+      "grad_norm": 0.13668283820152283,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 1295
+    },
+    {
+      "epoch": 2.1128138245842845,
+      "grad_norm": 0.12531708180904388,
+      "learning_rate": 0.0001,
+      "loss": 0.1693,
+      "step": 1296
+    },
+    {
+      "epoch": 2.114444082164982,
+      "grad_norm": 0.12956015765666962,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 1297
+    },
+    {
+      "epoch": 2.11607433974568,
+      "grad_norm": 0.11305827647447586,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 1298
+    },
+    {
+      "epoch": 2.1177045973263775,
+      "grad_norm": 0.1285475343465805,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1299
+    },
+    {
+      "epoch": 2.119334854907075,
+      "grad_norm": 0.10780393332242966,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 1300
+    },
+    {
+      "epoch": 2.120965112487773,
+      "grad_norm": 0.14362122118473053,
+      "learning_rate": 0.0001,
+      "loss": 0.1818,
+      "step": 1301
+    },
+    {
+      "epoch": 2.122595370068471,
+      "grad_norm": 0.13147400319576263,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 1302
+    },
+    {
+      "epoch": 2.1242256276491687,
+      "grad_norm": 0.11345292627811432,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 1303
+    },
+    {
+      "epoch": 2.1258558852298663,
+      "grad_norm": 0.12018296867609024,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 1304
+    },
+    {
+      "epoch": 2.127486142810564,
+      "grad_norm": 0.12988026440143585,
+      "learning_rate": 0.0001,
+      "loss": 0.1911,
+      "step": 1305
+    },
+    {
+      "epoch": 2.1291164003912617,
+      "grad_norm": 0.12335322052240372,
+      "learning_rate": 0.0001,
+      "loss": 0.1599,
+      "step": 1306
+    },
+    {
+      "epoch": 2.1307466579719594,
+      "grad_norm": 0.13894210755825043,
+      "learning_rate": 0.0001,
+      "loss": 0.189,
+      "step": 1307
+    },
+    {
+      "epoch": 2.1323769155526575,
+      "grad_norm": 0.13918493688106537,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1308
+    },
+    {
+      "epoch": 2.134007173133355,
+      "grad_norm": 0.12675361335277557,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 1309
+    },
+    {
+      "epoch": 2.135637430714053,
+      "grad_norm": 0.13385078310966492,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1310
+    },
+    {
+      "epoch": 2.1372676882947506,
+      "grad_norm": 0.17780406773090363,
+      "learning_rate": 0.0001,
+      "loss": 0.1674,
+      "step": 1311
+    },
+    {
+      "epoch": 2.1388979458754482,
+      "grad_norm": 0.16155046224594116,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1312
+    },
+    {
+      "epoch": 2.140528203456146,
+      "grad_norm": 0.177040234208107,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 1313
+    },
+    {
+      "epoch": 2.1421584610368436,
+      "grad_norm": 0.14751054346561432,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 1314
+    },
+    {
+      "epoch": 2.1437887186175417,
+      "grad_norm": 0.11751122027635574,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 1315
+    },
+    {
+      "epoch": 2.1454189761982394,
+      "grad_norm": 0.12306412309408188,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1316
+    },
+    {
+      "epoch": 2.147049233778937,
+      "grad_norm": 0.10852048546075821,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 1317
+    },
+    {
+      "epoch": 2.1486794913596348,
+      "grad_norm": 0.11529412865638733,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 1318
+    },
+    {
+      "epoch": 2.1503097489403324,
+      "grad_norm": 0.11132712662220001,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 1319
+    },
+    {
+      "epoch": 2.15194000652103,
+      "grad_norm": 0.1375824362039566,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1320
+    },
+    {
+      "epoch": 2.1535702641017282,
+      "grad_norm": 0.1289290189743042,
+      "learning_rate": 0.0001,
+      "loss": 0.1883,
+      "step": 1321
+    },
+    {
+      "epoch": 2.155200521682426,
+      "grad_norm": 0.1294177919626236,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 1322
+    },
+    {
+      "epoch": 2.1568307792631236,
+      "grad_norm": 0.13281673192977905,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 1323
+    },
+    {
+      "epoch": 2.1584610368438213,
+      "grad_norm": 0.1543353796005249,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 1324
+    },
+    {
+      "epoch": 2.160091294424519,
+      "grad_norm": 0.13094060122966766,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1325
+    },
+    {
+      "epoch": 2.1617215520052167,
+      "grad_norm": 0.1372898817062378,
+      "learning_rate": 0.0001,
+      "loss": 0.1693,
+      "step": 1326
+    },
+    {
+      "epoch": 2.1633518095859148,
+      "grad_norm": 0.1310727298259735,
+      "learning_rate": 0.0001,
+      "loss": 0.1934,
+      "step": 1327
+    },
+    {
+      "epoch": 2.1649820671666125,
+      "grad_norm": 0.13207446038722992,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 1328
+    },
+    {
+      "epoch": 2.16661232474731,
+      "grad_norm": 0.13898542523384094,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1329
+    },
+    {
+      "epoch": 2.168242582328008,
+      "grad_norm": 0.1400642991065979,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1330
+    },
+    {
+      "epoch": 2.1698728399087055,
+      "grad_norm": 0.12981773912906647,
+      "learning_rate": 0.0001,
+      "loss": 0.1772,
+      "step": 1331
+    },
+    {
+      "epoch": 2.171503097489403,
+      "grad_norm": 0.15546952188014984,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1332
+    },
+    {
+      "epoch": 2.1731333550701013,
+      "grad_norm": 0.11434978246688843,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 1333
+    },
+    {
+      "epoch": 2.174763612650799,
+      "grad_norm": 0.13666783273220062,
+      "learning_rate": 0.0001,
+      "loss": 0.1712,
+      "step": 1334
+    },
+    {
+      "epoch": 2.1763938702314967,
+      "grad_norm": 0.12208392471075058,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 1335
+    },
+    {
+      "epoch": 2.1780241278121943,
+      "grad_norm": 0.13914120197296143,
+      "learning_rate": 0.0001,
+      "loss": 0.1871,
+      "step": 1336
+    },
+    {
+      "epoch": 2.179654385392892,
+      "grad_norm": 0.13285982608795166,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 1337
+    },
+    {
+      "epoch": 2.1812846429735897,
+      "grad_norm": 0.1425054520368576,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 1338
+    },
+    {
+      "epoch": 2.1829149005542874,
+      "grad_norm": 0.12354496121406555,
+      "learning_rate": 0.0001,
+      "loss": 0.1912,
+      "step": 1339
+    },
+    {
+      "epoch": 2.1845451581349855,
+      "grad_norm": 0.12002517282962799,
+      "learning_rate": 0.0001,
+      "loss": 0.1915,
+      "step": 1340
+    },
+    {
+      "epoch": 2.186175415715683,
+      "grad_norm": 0.1574515998363495,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 1341
+    },
+    {
+      "epoch": 2.187805673296381,
+      "grad_norm": 0.12115081399679184,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 1342
+    },
+    {
+      "epoch": 2.1894359308770786,
+      "grad_norm": 0.1330757737159729,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 1343
+    },
+    {
+      "epoch": 2.1910661884577762,
+      "grad_norm": 0.11849580705165863,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 1344
+    },
+    {
+      "epoch": 2.192696446038474,
+      "grad_norm": 0.13124977052211761,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1345
+    },
+    {
+      "epoch": 2.194326703619172,
+      "grad_norm": 0.15525256097316742,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1346
+    },
+    {
+      "epoch": 2.1959569611998697,
+      "grad_norm": 0.13997358083724976,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1347
+    },
+    {
+      "epoch": 2.1975872187805674,
+      "grad_norm": 0.12414538115262985,
+      "learning_rate": 0.0001,
+      "loss": 0.1803,
+      "step": 1348
+    },
+    {
+      "epoch": 2.199217476361265,
+      "grad_norm": 0.12983959913253784,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 1349
+    },
+    {
+      "epoch": 2.2008477339419628,
+      "grad_norm": 0.10924775898456573,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 1350
+    },
+    {
+      "epoch": 2.2024779915226604,
+      "grad_norm": 0.12970325350761414,
+      "learning_rate": 0.0001,
+      "loss": 0.1997,
+      "step": 1351
+    },
+    {
+      "epoch": 2.204108249103358,
+      "grad_norm": 0.13614393770694733,
+      "learning_rate": 0.0001,
+      "loss": 0.1772,
+      "step": 1352
+    },
+    {
+      "epoch": 2.2057385066840562,
+      "grad_norm": 0.16756513714790344,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 1353
+    },
+    {
+      "epoch": 2.207368764264754,
+      "grad_norm": 0.12025320529937744,
+      "learning_rate": 0.0001,
+      "loss": 0.1729,
+      "step": 1354
+    },
+    {
+      "epoch": 2.2089990218454516,
+      "grad_norm": 0.13931293785572052,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 1355
+    },
+    {
+      "epoch": 2.2106292794261493,
+      "grad_norm": 0.14095361530780792,
+      "learning_rate": 0.0001,
+      "loss": 0.1794,
+      "step": 1356
+    },
+    {
+      "epoch": 2.212259537006847,
+      "grad_norm": 0.14122013747692108,
+      "learning_rate": 0.0001,
+      "loss": 0.1732,
+      "step": 1357
+    },
+    {
+      "epoch": 2.2138897945875446,
+      "grad_norm": 0.14121028780937195,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 1358
+    },
+    {
+      "epoch": 2.2155200521682428,
+      "grad_norm": 0.1547023057937622,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 1359
+    },
+    {
+      "epoch": 2.2171503097489405,
+      "grad_norm": 0.12952187657356262,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 1360
+    },
+    {
+      "epoch": 2.218780567329638,
+      "grad_norm": 0.17359581589698792,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 1361
+    },
+    {
+      "epoch": 2.220410824910336,
+      "grad_norm": 0.12868142127990723,
+      "learning_rate": 0.0001,
+      "loss": 0.1834,
+      "step": 1362
+    },
+    {
+      "epoch": 2.2220410824910335,
+      "grad_norm": 0.15855832397937775,
+      "learning_rate": 0.0001,
+      "loss": 0.1985,
+      "step": 1363
+    },
+    {
+      "epoch": 2.223671340071731,
+      "grad_norm": 0.14433158934116364,
+      "learning_rate": 0.0001,
+      "loss": 0.1943,
+      "step": 1364
+    },
+    {
+      "epoch": 2.225301597652429,
+      "grad_norm": 0.13805247843265533,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 1365
+    },
+    {
+      "epoch": 2.226931855233127,
+      "grad_norm": 0.12476252019405365,
+      "learning_rate": 0.0001,
+      "loss": 0.1835,
+      "step": 1366
+    },
+    {
+      "epoch": 2.2285621128138247,
+      "grad_norm": 0.14098286628723145,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1367
+    },
+    {
+      "epoch": 2.2301923703945223,
+      "grad_norm": 0.10581690073013306,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 1368
+    },
+    {
+      "epoch": 2.23182262797522,
+      "grad_norm": 0.13662414252758026,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 1369
+    },
+    {
+      "epoch": 2.2334528855559177,
+      "grad_norm": 0.13440090417861938,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 1370
+    },
+    {
+      "epoch": 2.2350831431366154,
+      "grad_norm": 0.17255081236362457,
+      "learning_rate": 0.0001,
+      "loss": 0.1885,
+      "step": 1371
+    },
+    {
+      "epoch": 2.2367134007173135,
+      "grad_norm": 0.13008393347263336,
+      "learning_rate": 0.0001,
+      "loss": 0.1834,
+      "step": 1372
+    },
+    {
+      "epoch": 2.238343658298011,
+      "grad_norm": 0.19530728459358215,
+      "learning_rate": 0.0001,
+      "loss": 0.1802,
+      "step": 1373
+    },
+    {
+      "epoch": 2.239973915878709,
+      "grad_norm": 0.12233468890190125,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 1374
+    },
+    {
+      "epoch": 2.2416041734594065,
+      "grad_norm": 0.12828123569488525,
+      "learning_rate": 0.0001,
+      "loss": 0.1693,
+      "step": 1375
+    },
+    {
+      "epoch": 2.2432344310401042,
+      "grad_norm": 0.1313515156507492,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 1376
+    },
+    {
+      "epoch": 2.244864688620802,
+      "grad_norm": 0.1494060903787613,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1377
+    },
+    {
+      "epoch": 2.2464949462015,
+      "grad_norm": 0.13343721628189087,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 1378
+    },
+    {
+      "epoch": 2.2481252037821977,
+      "grad_norm": 0.14259859919548035,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 1379
+    },
+    {
+      "epoch": 2.2497554613628954,
+      "grad_norm": 0.14472530782222748,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 1380
+    },
+    {
+      "epoch": 2.251385718943593,
+      "grad_norm": 0.1287330836057663,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 1381
+    },
+    {
+      "epoch": 2.2530159765242908,
+      "grad_norm": 0.11863413453102112,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "step": 1382
+    },
+    {
+      "epoch": 2.2546462341049884,
+      "grad_norm": 0.10795796662569046,
+      "learning_rate": 0.0001,
+      "loss": 0.1687,
+      "step": 1383
+    },
+    {
+      "epoch": 2.2562764916856866,
+      "grad_norm": 0.10524063557386398,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 1384
+    },
+    {
+      "epoch": 2.2579067492663842,
+      "grad_norm": 0.1203174889087677,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 1385
+    },
+    {
+      "epoch": 2.259537006847082,
+      "grad_norm": 0.1155901700258255,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 1386
+    },
+    {
+      "epoch": 2.2611672644277796,
+      "grad_norm": 0.13262033462524414,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1387
+    },
+    {
+      "epoch": 2.2627975220084773,
+      "grad_norm": 0.13157077133655548,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1388
+    },
+    {
+      "epoch": 2.264427779589175,
+      "grad_norm": 0.12929612398147583,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1389
+    },
+    {
+      "epoch": 2.2660580371698726,
+      "grad_norm": 0.12153502553701401,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 1390
+    },
+    {
+      "epoch": 2.2676882947505708,
+      "grad_norm": 0.15442398190498352,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 1391
+    },
+    {
+      "epoch": 2.2693185523312684,
+      "grad_norm": 0.15606394410133362,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 1392
+    },
+    {
+      "epoch": 2.270948809911966,
+      "grad_norm": 0.13780561089515686,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 1393
+    },
+    {
+      "epoch": 2.272579067492664,
+      "grad_norm": 0.4755675792694092,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 1394
+    },
+    {
+      "epoch": 2.2742093250733615,
+      "grad_norm": 0.1509324312210083,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 1395
+    },
+    {
+      "epoch": 2.275839582654059,
+      "grad_norm": 0.18392273783683777,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1396
+    },
+    {
+      "epoch": 2.2774698402347573,
+      "grad_norm": 0.12845247983932495,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 1397
+    },
+    {
+      "epoch": 2.279100097815455,
+      "grad_norm": 0.13720661401748657,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 1398
+    },
+    {
+      "epoch": 2.2807303553961527,
+      "grad_norm": 0.13770131766796112,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 1399
+    },
+    {
+      "epoch": 2.2823606129768503,
+      "grad_norm": 0.11446554213762283,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 1400
+    },
+    {
+      "epoch": 2.283990870557548,
+      "grad_norm": 0.12670235335826874,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 1401
+    },
+    {
+      "epoch": 2.2856211281382457,
+      "grad_norm": 0.14371562004089355,
+      "learning_rate": 0.0001,
+      "loss": 0.1895,
+      "step": 1402
+    },
+    {
+      "epoch": 2.2872513857189434,
+      "grad_norm": 0.12471149116754532,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 1403
+    },
+    {
+      "epoch": 2.2888816432996415,
+      "grad_norm": 0.13023298978805542,
+      "learning_rate": 0.0001,
+      "loss": 0.1715,
+      "step": 1404
+    },
+    {
+      "epoch": 2.290511900880339,
+      "grad_norm": 0.1387406587600708,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1405
+    },
+    {
+      "epoch": 2.292142158461037,
+      "grad_norm": 0.127483531832695,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1406
+    },
+    {
+      "epoch": 2.2937724160417345,
+      "grad_norm": 0.14067308604717255,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1407
+    },
+    {
+      "epoch": 2.295402673622432,
+      "grad_norm": 0.12968185544013977,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 1408
+    },
+    {
+      "epoch": 2.29703293120313,
+      "grad_norm": 0.23598945140838623,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 1409
+    },
+    {
+      "epoch": 2.298663188783828,
+      "grad_norm": 0.1529974788427353,
+      "learning_rate": 0.0001,
+      "loss": 0.1738,
+      "step": 1410
+    },
+    {
+      "epoch": 2.3002934463645257,
+      "grad_norm": 0.1288285255432129,
+      "learning_rate": 0.0001,
+      "loss": 0.1828,
+      "step": 1411
+    },
+    {
+      "epoch": 2.3019237039452234,
+      "grad_norm": 0.12833067774772644,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 1412
+    },
+    {
+      "epoch": 2.303553961525921,
+      "grad_norm": 0.11376657336950302,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 1413
+    },
+    {
+      "epoch": 2.3051842191066187,
+      "grad_norm": 0.23640869557857513,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 1414
+    },
+    {
+      "epoch": 2.3068144766873164,
+      "grad_norm": 0.11846059560775757,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 1415
+    },
+    {
+      "epoch": 2.308444734268014,
+      "grad_norm": 0.1381588727235794,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 1416
+    },
+    {
+      "epoch": 2.3100749918487122,
+      "grad_norm": 0.28972718119621277,
+      "learning_rate": 0.0001,
+      "loss": 0.1956,
+      "step": 1417
+    },
+    {
+      "epoch": 2.31170524942941,
+      "grad_norm": 0.1268901377916336,
+      "learning_rate": 0.0001,
+      "loss": 0.1725,
+      "step": 1418
+    },
+    {
+      "epoch": 2.3133355070101076,
+      "grad_norm": 0.13899771869182587,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "step": 1419
+    },
+    {
+      "epoch": 2.3149657645908053,
+      "grad_norm": 0.16093842685222626,
+      "learning_rate": 0.0001,
+      "loss": 0.1987,
+      "step": 1420
+    },
+    {
+      "epoch": 2.316596022171503,
+      "grad_norm": 0.13014528155326843,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 1421
+    },
+    {
+      "epoch": 2.318226279752201,
+      "grad_norm": 0.13447855412960052,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 1422
+    },
+    {
+      "epoch": 2.3198565373328988,
+      "grad_norm": 0.1403612345457077,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 1423
+    },
+    {
+      "epoch": 2.3214867949135964,
+      "grad_norm": 0.12157420068979263,
+      "learning_rate": 0.0001,
+      "loss": 0.1851,
+      "step": 1424
+    },
+    {
+      "epoch": 2.323117052494294,
+      "grad_norm": 0.11106284707784653,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 1425
+    },
+    {
+      "epoch": 2.324747310074992,
+      "grad_norm": 0.1440584808588028,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 1426
+    },
+    {
+      "epoch": 2.3263775676556895,
+      "grad_norm": 0.143922358751297,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 1427
+    },
+    {
+      "epoch": 2.328007825236387,
+      "grad_norm": 0.13588687777519226,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 1428
+    },
+    {
+      "epoch": 2.3296380828170853,
+      "grad_norm": 0.12500828504562378,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 1429
+    },
+    {
+      "epoch": 2.331268340397783,
+      "grad_norm": 0.13726584613323212,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 1430
+    },
+    {
+      "epoch": 2.3328985979784806,
+      "grad_norm": 0.12741105258464813,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 1431
+    },
+    {
+      "epoch": 2.3345288555591783,
+      "grad_norm": 0.15056177973747253,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 1432
+    },
+    {
+      "epoch": 2.336159113139876,
+      "grad_norm": 0.12522774934768677,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 1433
+    },
+    {
+      "epoch": 2.3377893707205737,
+      "grad_norm": 0.3474383056163788,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 1434
+    },
+    {
+      "epoch": 2.339419628301272,
+      "grad_norm": 0.12175151705741882,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 1435
+    },
+    {
+      "epoch": 2.3410498858819695,
+      "grad_norm": 0.12453535199165344,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 1436
+    },
+    {
+      "epoch": 2.342680143462667,
+      "grad_norm": 0.12441691756248474,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "step": 1437
+    },
+    {
+      "epoch": 2.344310401043365,
+      "grad_norm": 0.14301401376724243,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1438
+    },
+    {
+      "epoch": 2.3459406586240625,
+      "grad_norm": 0.11441715806722641,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1439
+    },
+    {
+      "epoch": 2.34757091620476,
+      "grad_norm": 0.12235062569379807,
+      "learning_rate": 0.0001,
+      "loss": 0.1738,
+      "step": 1440
+    },
+    {
+      "epoch": 2.349201173785458,
+      "grad_norm": 0.1444825977087021,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 1441
+    },
+    {
+      "epoch": 2.350831431366156,
+      "grad_norm": 0.12760686874389648,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 1442
+    },
+    {
+      "epoch": 2.3524616889468537,
+      "grad_norm": 0.12846995890140533,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 1443
+    },
+    {
+      "epoch": 2.3540919465275514,
+      "grad_norm": 0.12887585163116455,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 1444
+    },
+    {
+      "epoch": 2.355722204108249,
+      "grad_norm": 0.15019284188747406,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 1445
+    },
+    {
+      "epoch": 2.3573524616889467,
+      "grad_norm": 0.13723956048488617,
+      "learning_rate": 0.0001,
+      "loss": 0.1918,
+      "step": 1446
+    },
+    {
+      "epoch": 2.3589827192696444,
+      "grad_norm": 0.13037189841270447,
+      "learning_rate": 0.0001,
+      "loss": 0.1725,
+      "step": 1447
+    },
+    {
+      "epoch": 2.3606129768503425,
+      "grad_norm": 0.12533016502857208,
+      "learning_rate": 0.0001,
+      "loss": 0.1726,
+      "step": 1448
+    },
+    {
+      "epoch": 2.3622432344310402,
+      "grad_norm": 0.11401855945587158,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 1449
+    },
+    {
+      "epoch": 2.363873492011738,
+      "grad_norm": 0.12723423540592194,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 1450
+    },
+    {
+      "epoch": 2.3655037495924356,
+      "grad_norm": 0.12735813856124878,
+      "learning_rate": 0.0001,
+      "loss": 0.176,
+      "step": 1451
+    },
+    {
+      "epoch": 2.3671340071731333,
+      "grad_norm": 0.12557192146778107,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 1452
+    },
+    {
+      "epoch": 2.368764264753831,
+      "grad_norm": 0.1219494640827179,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1453
+    },
+    {
+      "epoch": 2.3703945223345286,
+      "grad_norm": 0.1263827085494995,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 1454
+    },
+    {
+      "epoch": 2.3720247799152268,
+      "grad_norm": 0.13217493891716003,
+      "learning_rate": 0.0001,
+      "loss": 0.1833,
+      "step": 1455
+    },
+    {
+      "epoch": 2.3736550374959244,
+      "grad_norm": 0.1541173756122589,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 1456
+    },
+    {
+      "epoch": 2.375285295076622,
+      "grad_norm": 0.13489516079425812,
+      "learning_rate": 0.0001,
+      "loss": 0.1801,
+      "step": 1457
+    },
+    {
+      "epoch": 2.37691555265732,
+      "grad_norm": 0.13315707445144653,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 1458
+    },
+    {
+      "epoch": 2.3785458102380175,
+      "grad_norm": 0.14040572941303253,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 1459
+    },
+    {
+      "epoch": 2.3801760678187156,
+      "grad_norm": 0.1226130798459053,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 1460
+    },
+    {
+      "epoch": 2.3818063253994133,
+      "grad_norm": 0.14467410743236542,
+      "learning_rate": 0.0001,
+      "loss": 0.193,
+      "step": 1461
+    },
+    {
+      "epoch": 2.383436582980111,
+      "grad_norm": 0.1495625525712967,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 1462
+    },
+    {
+      "epoch": 2.3850668405608086,
+      "grad_norm": 0.17007926106452942,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1463
+    },
+    {
+      "epoch": 2.3866970981415063,
+      "grad_norm": 0.09908633679151535,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 1464
+    },
+    {
+      "epoch": 2.388327355722204,
+      "grad_norm": 0.1391552835702896,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 1465
+    },
+    {
+      "epoch": 2.3899576133029017,
+      "grad_norm": 0.12647175788879395,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1466
+    },
+    {
+      "epoch": 2.3915878708835994,
+      "grad_norm": 0.13236671686172485,
+      "learning_rate": 0.0001,
+      "loss": 0.1911,
+      "step": 1467
+    },
+    {
+      "epoch": 2.3932181284642975,
+      "grad_norm": 0.11911871284246445,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 1468
+    },
+    {
+      "epoch": 2.394848386044995,
+      "grad_norm": 0.11741984635591507,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 1469
+    },
+    {
+      "epoch": 2.396478643625693,
+      "grad_norm": 0.13254094123840332,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 1470
+    },
+    {
+      "epoch": 2.3981089012063905,
+      "grad_norm": 0.12681660056114197,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 1471
+    },
+    {
+      "epoch": 2.399739158787088,
+      "grad_norm": 0.14695464074611664,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1472
+    },
+    {
+      "epoch": 2.4013694163677863,
+      "grad_norm": 0.15564168989658356,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 1473
+    },
+    {
+      "epoch": 2.402999673948484,
+      "grad_norm": 0.12581582367420197,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 1474
+    },
+    {
+      "epoch": 2.4046299315291817,
+      "grad_norm": 0.13274727761745453,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 1475
+    },
+    {
+      "epoch": 2.4062601891098794,
+      "grad_norm": 0.12894096970558167,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 1476
+    },
+    {
+      "epoch": 2.407890446690577,
+      "grad_norm": 0.11936718225479126,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 1477
+    },
+    {
+      "epoch": 2.4095207042712747,
+      "grad_norm": 0.16726456582546234,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 1478
+    },
+    {
+      "epoch": 2.4111509618519724,
+      "grad_norm": 0.12068488448858261,
+      "learning_rate": 0.0001,
+      "loss": 0.1823,
+      "step": 1479
+    },
+    {
+      "epoch": 2.4127812194326705,
+      "grad_norm": 0.11540605872869492,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1480
+    },
+    {
+      "epoch": 2.4144114770133682,
+      "grad_norm": 0.1343366652727127,
+      "learning_rate": 0.0001,
+      "loss": 0.1693,
+      "step": 1481
+    },
+    {
+      "epoch": 2.416041734594066,
+      "grad_norm": 0.11893413960933685,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 1482
+    },
+    {
+      "epoch": 2.4176719921747636,
+      "grad_norm": 0.124918632209301,
+      "learning_rate": 0.0001,
+      "loss": 0.1742,
+      "step": 1483
+    },
+    {
+      "epoch": 2.4193022497554613,
+      "grad_norm": 0.13789139688014984,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 1484
+    },
+    {
+      "epoch": 2.420932507336159,
+      "grad_norm": 0.13478578627109528,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 1485
+    },
+    {
+      "epoch": 2.422562764916857,
+      "grad_norm": 0.13447391986846924,
+      "learning_rate": 0.0001,
+      "loss": 0.1851,
+      "step": 1486
+    },
+    {
+      "epoch": 2.4241930224975548,
+      "grad_norm": 0.12459027767181396,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 1487
+    },
+    {
+      "epoch": 2.4258232800782524,
+      "grad_norm": 0.12010418623685837,
+      "learning_rate": 0.0001,
+      "loss": 0.1724,
+      "step": 1488
+    },
+    {
+      "epoch": 2.42745353765895,
+      "grad_norm": 0.14664210379123688,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1489
+    },
+    {
+      "epoch": 2.429083795239648,
+      "grad_norm": 0.12377094477415085,
+      "learning_rate": 0.0001,
+      "loss": 0.1693,
+      "step": 1490
+    },
+    {
+      "epoch": 2.4307140528203455,
+      "grad_norm": 0.12321583181619644,
+      "learning_rate": 0.0001,
+      "loss": 0.1778,
+      "step": 1491
+    },
+    {
+      "epoch": 2.432344310401043,
+      "grad_norm": 0.14363206923007965,
+      "learning_rate": 0.0001,
+      "loss": 0.1738,
+      "step": 1492
+    },
+    {
+      "epoch": 2.4339745679817413,
+      "grad_norm": 0.1732596904039383,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 1493
+    },
+    {
+      "epoch": 2.435604825562439,
+      "grad_norm": 0.12334398180246353,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 1494
+    },
+    {
+      "epoch": 2.4372350831431366,
+      "grad_norm": 0.13248227536678314,
+      "learning_rate": 0.0001,
+      "loss": 0.183,
+      "step": 1495
+    },
+    {
+      "epoch": 2.4388653407238343,
+      "grad_norm": 0.13810336589813232,
+      "learning_rate": 0.0001,
+      "loss": 0.1833,
+      "step": 1496
+    },
+    {
+      "epoch": 2.440495598304532,
+      "grad_norm": 0.1454666256904602,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 1497
+    },
+    {
+      "epoch": 2.4421258558852297,
+      "grad_norm": 0.14574459195137024,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 1498
+    },
+    {
+      "epoch": 2.443756113465928,
+      "grad_norm": 0.13363990187644958,
+      "learning_rate": 0.0001,
+      "loss": 0.1881,
+      "step": 1499
+    },
+    {
+      "epoch": 2.4453863710466255,
+      "grad_norm": 0.14776456356048584,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4453863710466255,
+      "eval_loss": 0.1997021734714508,
+      "eval_runtime": 460.2618,
+      "eval_samples_per_second": 4.1,
+      "eval_steps_per_second": 1.026,
+      "step": 1500
+    },
+    {
+      "epoch": 2.447016628627323,
+      "grad_norm": 0.10768138617277145,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 1501
+    },
+    {
+      "epoch": 2.448646886208021,
+      "grad_norm": 0.13541437685489655,
+      "learning_rate": 0.0001,
+      "loss": 0.1735,
+      "step": 1502
+    },
+    {
+      "epoch": 2.4502771437887185,
+      "grad_norm": 0.12808628380298615,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 1503
+    },
+    {
+      "epoch": 2.451907401369416,
+      "grad_norm": 0.11601930111646652,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 1504
+    },
+    {
+      "epoch": 2.453537658950114,
+      "grad_norm": 0.12584732472896576,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 1505
+    },
+    {
+      "epoch": 2.455167916530812,
+      "grad_norm": 0.12000913172960281,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 1506
+    },
+    {
+      "epoch": 2.4567981741115097,
+      "grad_norm": 0.1270546168088913,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 1507
+    },
+    {
+      "epoch": 2.4584284316922074,
+      "grad_norm": 0.12946002185344696,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1508
+    },
+    {
+      "epoch": 2.460058689272905,
+      "grad_norm": 0.14367902278900146,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1509
+    },
+    {
+      "epoch": 2.4616889468536027,
+      "grad_norm": 0.14034403860569,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 1510
+    },
+    {
+      "epoch": 2.463319204434301,
+      "grad_norm": 0.11473418027162552,
+      "learning_rate": 0.0001,
+      "loss": 0.1522,
+      "step": 1511
+    },
+    {
+      "epoch": 2.4649494620149985,
+      "grad_norm": 0.12510861456394196,
+      "learning_rate": 0.0001,
+      "loss": 0.1695,
+      "step": 1512
+    },
+    {
+      "epoch": 2.466579719595696,
+      "grad_norm": 0.12651176750659943,
+      "learning_rate": 0.0001,
+      "loss": 0.1877,
+      "step": 1513
+    },
+    {
+      "epoch": 2.468209977176394,
+      "grad_norm": 0.13196395337581635,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 1514
+    },
+    {
+      "epoch": 2.4698402347570916,
+      "grad_norm": 0.13804642856121063,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 1515
+    },
+    {
+      "epoch": 2.4714704923377893,
+      "grad_norm": 0.14535880088806152,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 1516
+    },
+    {
+      "epoch": 2.473100749918487,
+      "grad_norm": 0.12995412945747375,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 1517
+    },
+    {
+      "epoch": 2.474731007499185,
+      "grad_norm": 0.1427994966506958,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 1518
+    },
+    {
+      "epoch": 2.4763612650798827,
+      "grad_norm": 0.11478241533041,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 1519
+    },
+    {
+      "epoch": 2.4779915226605804,
+      "grad_norm": 0.1354149729013443,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 1520
+    },
+    {
+      "epoch": 2.479621780241278,
+      "grad_norm": 0.11658424139022827,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 1521
+    },
+    {
+      "epoch": 2.481252037821976,
+      "grad_norm": 0.12623891234397888,
+      "learning_rate": 0.0001,
+      "loss": 0.1809,
+      "step": 1522
+    },
+    {
+      "epoch": 2.4828822954026735,
+      "grad_norm": 0.12294737994670868,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 1523
+    },
+    {
+      "epoch": 2.4845125529833716,
+      "grad_norm": 0.1138739064335823,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 1524
+    },
+    {
+      "epoch": 2.4861428105640693,
+      "grad_norm": 0.13236074149608612,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1525
+    },
+    {
+      "epoch": 2.487773068144767,
+      "grad_norm": 0.11817993223667145,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1526
+    },
+    {
+      "epoch": 2.4894033257254646,
+      "grad_norm": 0.10676276683807373,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 1527
+    },
+    {
+      "epoch": 2.4910335833061623,
+      "grad_norm": 0.13718129694461823,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 1528
+    },
+    {
+      "epoch": 2.49266384088686,
+      "grad_norm": 0.13279683887958527,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 1529
+    },
+    {
+      "epoch": 2.4942940984675577,
+      "grad_norm": 0.13865776360034943,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 1530
+    },
+    {
+      "epoch": 2.495924356048256,
+      "grad_norm": 0.13090378046035767,
+      "learning_rate": 0.0001,
+      "loss": 0.1777,
+      "step": 1531
+    },
+    {
+      "epoch": 2.4975546136289535,
+      "grad_norm": 0.12446148693561554,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1532
+    },
+    {
+      "epoch": 2.499184871209651,
+      "grad_norm": 0.1395845115184784,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 1533
+    },
+    {
+      "epoch": 2.500815128790349,
+      "grad_norm": 0.11929094046354294,
+      "learning_rate": 0.0001,
+      "loss": 0.1686,
+      "step": 1534
+    },
+    {
+      "epoch": 2.5024453863710465,
+      "grad_norm": 0.14516448974609375,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1535
+    },
+    {
+      "epoch": 2.5040756439517446,
+      "grad_norm": 0.11706582456827164,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 1536
+    },
+    {
+      "epoch": 2.5057059015324423,
+      "grad_norm": 0.13064873218536377,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 1537
+    },
+    {
+      "epoch": 2.50733615911314,
+      "grad_norm": 0.12183361500501633,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 1538
+    },
+    {
+      "epoch": 2.5089664166938377,
+      "grad_norm": 0.11860349029302597,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1539
+    },
+    {
+      "epoch": 2.5105966742745354,
+      "grad_norm": 0.11384200304746628,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1540
+    },
+    {
+      "epoch": 2.512226931855233,
+      "grad_norm": 0.13300910592079163,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 1541
+    },
+    {
+      "epoch": 2.5138571894359307,
+      "grad_norm": 0.11307025700807571,
+      "learning_rate": 0.0001,
+      "loss": 0.1778,
+      "step": 1542
+    },
+    {
+      "epoch": 2.5154874470166284,
+      "grad_norm": 0.11402298510074615,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 1543
+    },
+    {
+      "epoch": 2.5171177045973265,
+      "grad_norm": 0.13868348300457,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1544
+    },
+    {
+      "epoch": 2.518747962178024,
+      "grad_norm": 0.14691469073295593,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 1545
+    },
+    {
+      "epoch": 2.520378219758722,
+      "grad_norm": 0.1223243921995163,
+      "learning_rate": 0.0001,
+      "loss": 0.1828,
+      "step": 1546
+    },
+    {
+      "epoch": 2.5220084773394196,
+      "grad_norm": 0.1313992589712143,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1547
+    },
+    {
+      "epoch": 2.5236387349201173,
+      "grad_norm": 0.12788835167884827,
+      "learning_rate": 0.0001,
+      "loss": 0.1786,
+      "step": 1548
+    },
+    {
+      "epoch": 2.5252689925008154,
+      "grad_norm": 0.11976204812526703,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 1549
+    },
+    {
+      "epoch": 2.526899250081513,
+      "grad_norm": 0.11993979662656784,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 1550
+    },
+    {
+      "epoch": 2.5285295076622107,
+      "grad_norm": 0.1298224925994873,
+      "learning_rate": 0.0001,
+      "loss": 0.1801,
+      "step": 1551
+    },
+    {
+      "epoch": 2.5301597652429084,
+      "grad_norm": 0.12879474461078644,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 1552
+    },
+    {
+      "epoch": 2.531790022823606,
+      "grad_norm": 0.11655417829751968,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 1553
+    },
+    {
+      "epoch": 2.533420280404304,
+      "grad_norm": 0.14049752056598663,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 1554
+    },
+    {
+      "epoch": 2.5350505379850015,
+      "grad_norm": 0.13857971131801605,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 1555
+    },
+    {
+      "epoch": 2.536680795565699,
+      "grad_norm": 0.12514682114124298,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 1556
+    },
+    {
+      "epoch": 2.5383110531463973,
+      "grad_norm": 0.12018663436174393,
+      "learning_rate": 0.0001,
+      "loss": 0.1788,
+      "step": 1557
+    },
+    {
+      "epoch": 2.539941310727095,
+      "grad_norm": 0.14534510672092438,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 1558
+    },
+    {
+      "epoch": 2.5415715683077926,
+      "grad_norm": 0.1236572340130806,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 1559
+    },
+    {
+      "epoch": 2.5432018258884903,
+      "grad_norm": 0.13306783139705658,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1560
+    },
+    {
+      "epoch": 2.544832083469188,
+      "grad_norm": 0.10747732222080231,
+      "learning_rate": 0.0001,
+      "loss": 0.1766,
+      "step": 1561
+    },
+    {
+      "epoch": 2.546462341049886,
+      "grad_norm": 0.12410397827625275,
+      "learning_rate": 0.0001,
+      "loss": 0.1829,
+      "step": 1562
+    },
+    {
+      "epoch": 2.548092598630584,
+      "grad_norm": 0.14367325603961945,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1563
+    },
+    {
+      "epoch": 2.5497228562112815,
+      "grad_norm": 0.12617376446723938,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 1564
+    },
+    {
+      "epoch": 2.551353113791979,
+      "grad_norm": 0.13464583456516266,
+      "learning_rate": 0.0001,
+      "loss": 0.1825,
+      "step": 1565
+    },
+    {
+      "epoch": 2.552983371372677,
+      "grad_norm": 0.15118366479873657,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1566
+    },
+    {
+      "epoch": 2.5546136289533745,
+      "grad_norm": 0.13260018825531006,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 1567
+    },
+    {
+      "epoch": 2.556243886534072,
+      "grad_norm": 0.1262848824262619,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 1568
+    },
+    {
+      "epoch": 2.55787414411477,
+      "grad_norm": 0.1359010487794876,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 1569
+    },
+    {
+      "epoch": 2.559504401695468,
+      "grad_norm": 0.12151306122541428,
+      "learning_rate": 0.0001,
+      "loss": 0.1809,
+      "step": 1570
+    },
+    {
+      "epoch": 2.5611346592761657,
+      "grad_norm": 0.144693985581398,
+      "learning_rate": 0.0001,
+      "loss": 0.1637,
+      "step": 1571
+    },
+    {
+      "epoch": 2.5627649168568634,
+      "grad_norm": 0.13245713710784912,
+      "learning_rate": 0.0001,
+      "loss": 0.1741,
+      "step": 1572
+    },
+    {
+      "epoch": 2.564395174437561,
+      "grad_norm": 0.14154277741909027,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1573
+    },
+    {
+      "epoch": 2.5660254320182587,
+      "grad_norm": 0.14346501231193542,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1574
+    },
+    {
+      "epoch": 2.567655689598957,
+      "grad_norm": 0.12976010143756866,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 1575
+    },
+    {
+      "epoch": 2.5692859471796545,
+      "grad_norm": 0.11773863434791565,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 1576
+    },
+    {
+      "epoch": 2.570916204760352,
+      "grad_norm": 0.12860426306724548,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1577
+    },
+    {
+      "epoch": 2.57254646234105,
+      "grad_norm": 0.12525953352451324,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1578
+    },
+    {
+      "epoch": 2.5741767199217476,
+      "grad_norm": 0.13797353208065033,
+      "learning_rate": 0.0001,
+      "loss": 0.1886,
+      "step": 1579
+    },
+    {
+      "epoch": 2.5758069775024452,
+      "grad_norm": 0.13514308631420135,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 1580
+    },
+    {
+      "epoch": 2.577437235083143,
+      "grad_norm": 0.1197233498096466,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1581
+    },
+    {
+      "epoch": 2.5790674926638406,
+      "grad_norm": 0.1371515542268753,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 1582
+    },
+    {
+      "epoch": 2.5806977502445387,
+      "grad_norm": 0.12182509154081345,
+      "learning_rate": 0.0001,
+      "loss": 0.1808,
+      "step": 1583
+    },
+    {
+      "epoch": 2.5823280078252364,
+      "grad_norm": 0.12575587630271912,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 1584
+    },
+    {
+      "epoch": 2.583958265405934,
+      "grad_norm": 0.11890911310911179,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 1585
+    },
+    {
+      "epoch": 2.5855885229866318,
+      "grad_norm": 0.12488003820180893,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 1586
+    },
+    {
+      "epoch": 2.58721878056733,
+      "grad_norm": 0.13378724455833435,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 1587
+    },
+    {
+      "epoch": 2.5888490381480276,
+      "grad_norm": 0.14022649824619293,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 1588
+    },
+    {
+      "epoch": 2.5904792957287253,
+      "grad_norm": 0.12191809713840485,
+      "learning_rate": 0.0001,
+      "loss": 0.1657,
+      "step": 1589
+    },
+    {
+      "epoch": 2.592109553309423,
+      "grad_norm": 0.11590269953012466,
+      "learning_rate": 0.0001,
+      "loss": 0.1559,
+      "step": 1590
+    },
+    {
+      "epoch": 2.5937398108901206,
+      "grad_norm": 0.18004612624645233,
+      "learning_rate": 0.0001,
+      "loss": 0.1821,
+      "step": 1591
+    },
+    {
+      "epoch": 2.5953700684708183,
+      "grad_norm": 0.11519701033830643,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 1592
+    },
+    {
+      "epoch": 2.597000326051516,
+      "grad_norm": 0.110859714448452,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 1593
+    },
+    {
+      "epoch": 2.5986305836322137,
+      "grad_norm": 0.11771699786186218,
+      "learning_rate": 0.0001,
+      "loss": 0.1929,
+      "step": 1594
+    },
+    {
+      "epoch": 2.600260841212912,
+      "grad_norm": 0.1058066114783287,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 1595
+    },
+    {
+      "epoch": 2.6018910987936095,
+      "grad_norm": 0.13187898695468903,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 1596
+    },
+    {
+      "epoch": 2.603521356374307,
+      "grad_norm": 0.14017753303050995,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1597
+    },
+    {
+      "epoch": 2.605151613955005,
+      "grad_norm": 0.13309310376644135,
+      "learning_rate": 0.0001,
+      "loss": 0.1684,
+      "step": 1598
+    },
+    {
+      "epoch": 2.6067818715357025,
+      "grad_norm": 0.1273549348115921,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 1599
+    },
+    {
+      "epoch": 2.6084121291164006,
+      "grad_norm": 0.16523495316505432,
+      "learning_rate": 0.0001,
+      "loss": 0.1751,
+      "step": 1600
+    },
+    {
+      "epoch": 2.6100423866970983,
+      "grad_norm": 0.1349019855260849,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1601
+    },
+    {
+      "epoch": 2.611672644277796,
+      "grad_norm": 0.1419772207736969,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 1602
+    },
+    {
+      "epoch": 2.6133029018584937,
+      "grad_norm": 0.12186914682388306,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 1603
+    },
+    {
+      "epoch": 2.6149331594391914,
+      "grad_norm": 0.13109511137008667,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 1604
+    },
+    {
+      "epoch": 2.616563417019889,
+      "grad_norm": 0.14104005694389343,
+      "learning_rate": 0.0001,
+      "loss": 0.1839,
+      "step": 1605
+    },
+    {
+      "epoch": 2.6181936746005867,
+      "grad_norm": 0.13870073854923248,
+      "learning_rate": 0.0001,
+      "loss": 0.1758,
+      "step": 1606
+    },
+    {
+      "epoch": 2.6198239321812844,
+      "grad_norm": 0.1275268942117691,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 1607
+    },
+    {
+      "epoch": 2.6214541897619825,
+      "grad_norm": 0.14321641623973846,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 1608
+    },
+    {
+      "epoch": 2.62308444734268,
+      "grad_norm": 0.13970732688903809,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 1609
+    },
+    {
+      "epoch": 2.624714704923378,
+      "grad_norm": 0.13474468886852264,
+      "learning_rate": 0.0001,
+      "loss": 0.1903,
+      "step": 1610
+    },
+    {
+      "epoch": 2.6263449625040756,
+      "grad_norm": 0.12260457128286362,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 1611
+    },
+    {
+      "epoch": 2.6279752200847732,
+      "grad_norm": 0.112511046230793,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 1612
+    },
+    {
+      "epoch": 2.6296054776654714,
+      "grad_norm": 0.13217821717262268,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 1613
+    },
+    {
+      "epoch": 2.631235735246169,
+      "grad_norm": 0.13116757571697235,
+      "learning_rate": 0.0001,
+      "loss": 0.1777,
+      "step": 1614
+    },
+    {
+      "epoch": 2.6328659928268667,
+      "grad_norm": 0.1306123584508896,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 1615
+    },
+    {
+      "epoch": 2.6344962504075644,
+      "grad_norm": 0.13786140084266663,
+      "learning_rate": 0.0001,
+      "loss": 0.187,
+      "step": 1616
+    },
+    {
+      "epoch": 2.636126507988262,
+      "grad_norm": 0.14164811372756958,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 1617
+    },
+    {
+      "epoch": 2.6377567655689598,
+      "grad_norm": 0.1309959590435028,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 1618
+    },
+    {
+      "epoch": 2.6393870231496575,
+      "grad_norm": 0.12645256519317627,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 1619
+    },
+    {
+      "epoch": 2.641017280730355,
+      "grad_norm": 0.13473668694496155,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 1620
+    },
+    {
+      "epoch": 2.6426475383110533,
+      "grad_norm": 0.11596634984016418,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1621
+    },
+    {
+      "epoch": 2.644277795891751,
+      "grad_norm": 0.12235169112682343,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1622
+    },
+    {
+      "epoch": 2.6459080534724486,
+      "grad_norm": 0.13206972181797028,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 1623
+    },
+    {
+      "epoch": 2.6475383110531463,
+      "grad_norm": 0.12648923695087433,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 1624
+    },
+    {
+      "epoch": 2.6491685686338444,
+      "grad_norm": 0.11728816479444504,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1625
+    },
+    {
+      "epoch": 2.650798826214542,
+      "grad_norm": 0.13588158786296844,
+      "learning_rate": 0.0001,
+      "loss": 0.1724,
+      "step": 1626
+    },
+    {
+      "epoch": 2.65242908379524,
+      "grad_norm": 0.25838354229927063,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 1627
+    },
+    {
+      "epoch": 2.6540593413759375,
+      "grad_norm": 0.13654319941997528,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 1628
+    },
+    {
+      "epoch": 2.655689598956635,
+      "grad_norm": 0.11923132836818695,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1629
+    },
+    {
+      "epoch": 2.657319856537333,
+      "grad_norm": 0.13744457066059113,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 1630
+    },
+    {
+      "epoch": 2.6589501141180305,
+      "grad_norm": 0.14435473084449768,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1631
+    },
+    {
+      "epoch": 2.660580371698728,
+      "grad_norm": 0.1060493215918541,
+      "learning_rate": 0.0001,
+      "loss": 0.1448,
+      "step": 1632
+    },
+    {
+      "epoch": 2.6622106292794263,
+      "grad_norm": 0.12072440981864929,
+      "learning_rate": 0.0001,
+      "loss": 0.1766,
+      "step": 1633
+    },
+    {
+      "epoch": 2.663840886860124,
+      "grad_norm": 0.11674067378044128,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 1634
+    },
+    {
+      "epoch": 2.6654711444408217,
+      "grad_norm": 0.12532764673233032,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1635
+    },
+    {
+      "epoch": 2.6671014020215194,
+      "grad_norm": 0.10460726171731949,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 1636
+    },
+    {
+      "epoch": 2.668731659602217,
+      "grad_norm": 0.1254202276468277,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1637
+    },
+    {
+      "epoch": 2.670361917182915,
+      "grad_norm": 0.1245632991194725,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1638
+    },
+    {
+      "epoch": 2.671992174763613,
+      "grad_norm": 0.1212785616517067,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 1639
+    },
+    {
+      "epoch": 2.6736224323443105,
+      "grad_norm": 0.1357593685388565,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 1640
+    },
+    {
+      "epoch": 2.675252689925008,
+      "grad_norm": 0.13232427835464478,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 1641
+    },
+    {
+      "epoch": 2.676882947505706,
+      "grad_norm": 0.12210696190595627,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 1642
+    },
+    {
+      "epoch": 2.6785132050864036,
+      "grad_norm": 0.12548168003559113,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 1643
+    },
+    {
+      "epoch": 2.6801434626671012,
+      "grad_norm": 0.13502956926822662,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 1644
+    },
+    {
+      "epoch": 2.681773720247799,
+      "grad_norm": 0.12499292939901352,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 1645
+    },
+    {
+      "epoch": 2.683403977828497,
+      "grad_norm": 0.12501081824302673,
+      "learning_rate": 0.0001,
+      "loss": 0.1741,
+      "step": 1646
+    },
+    {
+      "epoch": 2.6850342354091947,
+      "grad_norm": 0.13593406975269318,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 1647
+    },
+    {
+      "epoch": 2.6866644929898924,
+      "grad_norm": 0.17511261999607086,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 1648
+    },
+    {
+      "epoch": 2.68829475057059,
+      "grad_norm": 0.14490318298339844,
+      "learning_rate": 0.0001,
+      "loss": 0.1684,
+      "step": 1649
+    },
+    {
+      "epoch": 2.6899250081512878,
+      "grad_norm": 0.22610324621200562,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 1650
+    },
+    {
+      "epoch": 2.691555265731986,
+      "grad_norm": 0.15227797627449036,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 1651
+    },
+    {
+      "epoch": 2.6931855233126836,
+      "grad_norm": 0.13343138992786407,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1652
+    },
+    {
+      "epoch": 2.6948157808933813,
+      "grad_norm": 0.1372375637292862,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 1653
+    },
+    {
+      "epoch": 2.696446038474079,
+      "grad_norm": 0.10308275371789932,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 1654
+    },
+    {
+      "epoch": 2.6980762960547766,
+      "grad_norm": 0.11079935729503632,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 1655
+    },
+    {
+      "epoch": 2.6997065536354743,
+      "grad_norm": 0.12217281013727188,
+      "learning_rate": 0.0001,
+      "loss": 0.1782,
+      "step": 1656
+    },
+    {
+      "epoch": 2.701336811216172,
+      "grad_norm": 0.10372958332300186,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 1657
+    },
+    {
+      "epoch": 2.7029670687968697,
+      "grad_norm": 0.10747478157281876,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 1658
+    },
+    {
+      "epoch": 2.704597326377568,
+      "grad_norm": 0.12394733726978302,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 1659
+    },
+    {
+      "epoch": 2.7062275839582655,
+      "grad_norm": 0.11882739514112473,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 1660
+    },
+    {
+      "epoch": 2.707857841538963,
+      "grad_norm": 0.11708527058362961,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1661
+    },
+    {
+      "epoch": 2.709488099119661,
+      "grad_norm": 0.1281583458185196,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 1662
+    },
+    {
+      "epoch": 2.7111183567003585,
+      "grad_norm": 0.11309745907783508,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1663
+    },
+    {
+      "epoch": 2.7127486142810566,
+      "grad_norm": 0.1279565989971161,
+      "learning_rate": 0.0001,
+      "loss": 0.1788,
+      "step": 1664
+    },
+    {
+      "epoch": 2.7143788718617543,
+      "grad_norm": 0.11305457353591919,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 1665
+    },
+    {
+      "epoch": 2.716009129442452,
+      "grad_norm": 0.18214936554431915,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 1666
+    },
+    {
+      "epoch": 2.7176393870231497,
+      "grad_norm": 0.15045790374279022,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 1667
+    },
+    {
+      "epoch": 2.7192696446038473,
+      "grad_norm": 0.1176130399107933,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 1668
+    },
+    {
+      "epoch": 2.720899902184545,
+      "grad_norm": 0.12828783690929413,
+      "learning_rate": 0.0001,
+      "loss": 0.176,
+      "step": 1669
+    },
+    {
+      "epoch": 2.7225301597652427,
+      "grad_norm": 0.1470414698123932,
+      "learning_rate": 0.0001,
+      "loss": 0.1911,
+      "step": 1670
+    },
+    {
+      "epoch": 2.7241604173459404,
+      "grad_norm": 0.13573643565177917,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 1671
+    },
+    {
+      "epoch": 2.7257906749266385,
+      "grad_norm": 0.13036182522773743,
+      "learning_rate": 0.0001,
+      "loss": 0.1868,
+      "step": 1672
+    },
+    {
+      "epoch": 2.727420932507336,
+      "grad_norm": 0.13708871603012085,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 1673
+    },
+    {
+      "epoch": 2.729051190088034,
+      "grad_norm": 0.1127118393778801,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 1674
+    },
+    {
+      "epoch": 2.7306814476687316,
+      "grad_norm": 0.19948649406433105,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1675
+    },
+    {
+      "epoch": 2.7323117052494297,
+      "grad_norm": 0.12707367539405823,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 1676
+    },
+    {
+      "epoch": 2.7339419628301274,
+      "grad_norm": 0.11897812783718109,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 1677
+    },
+    {
+      "epoch": 2.735572220410825,
+      "grad_norm": 0.11845947057008743,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 1678
+    },
+    {
+      "epoch": 2.7372024779915227,
+      "grad_norm": 0.11247242242097855,
+      "learning_rate": 0.0001,
+      "loss": 0.1795,
+      "step": 1679
+    },
+    {
+      "epoch": 2.7388327355722204,
+      "grad_norm": 0.1320265382528305,
+      "learning_rate": 0.0001,
+      "loss": 0.1791,
+      "step": 1680
+    },
+    {
+      "epoch": 2.740462993152918,
+      "grad_norm": 0.13398633897304535,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1681
+    },
+    {
+      "epoch": 2.7420932507336158,
+      "grad_norm": 0.13365308940410614,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 1682
+    },
+    {
+      "epoch": 2.7437235083143134,
+      "grad_norm": 0.15530569851398468,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 1683
+    },
+    {
+      "epoch": 2.7453537658950116,
+      "grad_norm": 0.16007980704307556,
+      "learning_rate": 0.0001,
+      "loss": 0.1847,
+      "step": 1684
+    },
+    {
+      "epoch": 2.7469840234757092,
+      "grad_norm": 0.13407476246356964,
+      "learning_rate": 0.0001,
+      "loss": 0.1818,
+      "step": 1685
+    },
+    {
+      "epoch": 2.748614281056407,
+      "grad_norm": 0.1119256541132927,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 1686
+    },
+    {
+      "epoch": 2.7502445386371046,
+      "grad_norm": 0.1142873466014862,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1687
+    },
+    {
+      "epoch": 2.7518747962178023,
+      "grad_norm": 0.3455221951007843,
+      "learning_rate": 0.0001,
+      "loss": 0.1765,
+      "step": 1688
+    },
+    {
+      "epoch": 2.7535050537985004,
+      "grad_norm": 0.11033251881599426,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 1689
+    },
+    {
+      "epoch": 2.755135311379198,
+      "grad_norm": 0.10998693853616714,
+      "learning_rate": 0.0001,
+      "loss": 0.1795,
+      "step": 1690
+    },
+    {
+      "epoch": 2.7567655689598958,
+      "grad_norm": 0.12498677521944046,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 1691
+    },
+    {
+      "epoch": 2.7583958265405935,
+      "grad_norm": 0.10666051506996155,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 1692
+    },
+    {
+      "epoch": 2.760026084121291,
+      "grad_norm": 0.11262010782957077,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 1693
+    },
+    {
+      "epoch": 2.761656341701989,
+      "grad_norm": 0.12692642211914062,
+      "learning_rate": 0.0001,
+      "loss": 0.1619,
+      "step": 1694
+    },
+    {
+      "epoch": 2.7632865992826865,
+      "grad_norm": 0.13128408789634705,
+      "learning_rate": 0.0001,
+      "loss": 0.1657,
+      "step": 1695
+    },
+    {
+      "epoch": 2.764916856863384,
+      "grad_norm": 0.12888456881046295,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 1696
+    },
+    {
+      "epoch": 2.7665471144440823,
+      "grad_norm": 0.15095765888690948,
+      "learning_rate": 0.0001,
+      "loss": 0.1847,
+      "step": 1697
+    },
+    {
+      "epoch": 2.76817737202478,
+      "grad_norm": 0.14810307323932648,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 1698
+    },
+    {
+      "epoch": 2.7698076296054777,
+      "grad_norm": 0.1464967578649521,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 1699
+    },
+    {
+      "epoch": 2.7714378871861753,
+      "grad_norm": 0.11904679238796234,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 1700
+    },
+    {
+      "epoch": 2.773068144766873,
+      "grad_norm": 0.1260305941104889,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 1701
+    },
+    {
+      "epoch": 2.774698402347571,
+      "grad_norm": 0.12736454606056213,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 1702
+    },
+    {
+      "epoch": 2.776328659928269,
+      "grad_norm": 0.14235186576843262,
+      "learning_rate": 0.0001,
+      "loss": 0.1567,
+      "step": 1703
+    },
+    {
+      "epoch": 2.7779589175089665,
+      "grad_norm": 0.13851489126682281,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 1704
+    },
+    {
+      "epoch": 2.779589175089664,
+      "grad_norm": 0.15208859741687775,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1705
+    },
+    {
+      "epoch": 2.781219432670362,
+      "grad_norm": 0.12781186401844025,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 1706
+    },
+    {
+      "epoch": 2.7828496902510595,
+      "grad_norm": 0.16855067014694214,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 1707
+    },
+    {
+      "epoch": 2.7844799478317572,
+      "grad_norm": 0.10379106551408768,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 1708
+    },
+    {
+      "epoch": 2.786110205412455,
+      "grad_norm": 0.13109523057937622,
+      "learning_rate": 0.0001,
+      "loss": 0.1858,
+      "step": 1709
+    },
+    {
+      "epoch": 2.787740462993153,
+      "grad_norm": 0.11424817889928818,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1710
+    },
+    {
+      "epoch": 2.7893707205738507,
+      "grad_norm": 0.1378762423992157,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 1711
+    },
+    {
+      "epoch": 2.7910009781545484,
+      "grad_norm": 0.13009703159332275,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 1712
+    },
+    {
+      "epoch": 2.792631235735246,
+      "grad_norm": 0.1202034205198288,
+      "learning_rate": 0.0001,
+      "loss": 0.1674,
+      "step": 1713
+    },
+    {
+      "epoch": 2.7942614933159438,
+      "grad_norm": 0.11197958886623383,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 1714
+    },
+    {
+      "epoch": 2.795891750896642,
+      "grad_norm": 0.1207757517695427,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 1715
+    },
+    {
+      "epoch": 2.7975220084773396,
+      "grad_norm": 0.13465166091918945,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1716
+    },
+    {
+      "epoch": 2.7991522660580372,
+      "grad_norm": 0.12799595296382904,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 1717
+    },
+    {
+      "epoch": 2.800782523638735,
+      "grad_norm": 0.10873240232467651,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 1718
+    },
+    {
+      "epoch": 2.8024127812194326,
+      "grad_norm": 0.204450324177742,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 1719
+    },
+    {
+      "epoch": 2.8040430388001303,
+      "grad_norm": 0.12960658967494965,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 1720
+    },
+    {
+      "epoch": 2.805673296380828,
+      "grad_norm": 0.14518047869205475,
+      "learning_rate": 0.0001,
+      "loss": 0.1732,
+      "step": 1721
+    },
+    {
+      "epoch": 2.8073035539615256,
+      "grad_norm": 0.13319043815135956,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 1722
+    },
+    {
+      "epoch": 2.8089338115422238,
+      "grad_norm": 0.13059182465076447,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 1723
+    },
+    {
+      "epoch": 2.8105640691229214,
+      "grad_norm": 0.13546039164066315,
+      "learning_rate": 0.0001,
+      "loss": 0.1754,
+      "step": 1724
+    },
+    {
+      "epoch": 2.812194326703619,
+      "grad_norm": 0.10475914925336838,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 1725
+    },
+    {
+      "epoch": 2.813824584284317,
+      "grad_norm": 0.11786188930273056,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 1726
+    },
+    {
+      "epoch": 2.815454841865015,
+      "grad_norm": 0.12777145206928253,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 1727
+    },
+    {
+      "epoch": 2.8170850994457126,
+      "grad_norm": 0.11510585993528366,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 1728
+    },
+    {
+      "epoch": 2.8187153570264103,
+      "grad_norm": 0.13350418210029602,
+      "learning_rate": 0.0001,
+      "loss": 0.1758,
+      "step": 1729
+    },
+    {
+      "epoch": 2.820345614607108,
+      "grad_norm": 0.12786592543125153,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1730
+    },
+    {
+      "epoch": 2.8219758721878057,
+      "grad_norm": 0.1287817358970642,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 1731
+    },
+    {
+      "epoch": 2.8236061297685033,
+      "grad_norm": 0.12238069623708725,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 1732
+    },
+    {
+      "epoch": 2.825236387349201,
+      "grad_norm": 0.12411892414093018,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 1733
+    },
+    {
+      "epoch": 2.8268666449298987,
+      "grad_norm": 0.12480080127716064,
+      "learning_rate": 0.0001,
+      "loss": 0.1814,
+      "step": 1734
+    },
+    {
+      "epoch": 2.828496902510597,
+      "grad_norm": 0.13746987283229828,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 1735
+    },
+    {
+      "epoch": 2.8301271600912945,
+      "grad_norm": 0.14211837947368622,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 1736
+    },
+    {
+      "epoch": 2.831757417671992,
+      "grad_norm": 0.13357684016227722,
+      "learning_rate": 0.0001,
+      "loss": 0.1751,
+      "step": 1737
+    },
+    {
+      "epoch": 2.83338767525269,
+      "grad_norm": 0.13745687901973724,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 1738
+    },
+    {
+      "epoch": 2.8350179328333875,
+      "grad_norm": 0.1132083386182785,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 1739
+    },
+    {
+      "epoch": 2.8366481904140857,
+      "grad_norm": 0.13176746666431427,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 1740
+    },
+    {
+      "epoch": 2.8382784479947833,
+      "grad_norm": 0.12045666575431824,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 1741
+    },
+    {
+      "epoch": 2.839908705575481,
+      "grad_norm": 0.1171451285481453,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 1742
+    },
+    {
+      "epoch": 2.8415389631561787,
+      "grad_norm": 0.13992704451084137,
+      "learning_rate": 0.0001,
+      "loss": 0.1853,
+      "step": 1743
+    },
+    {
+      "epoch": 2.8431692207368764,
+      "grad_norm": 0.1277761161327362,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 1744
+    },
+    {
+      "epoch": 2.844799478317574,
+      "grad_norm": 0.14471198618412018,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1745
+    },
+    {
+      "epoch": 2.8464297358982718,
+      "grad_norm": 0.1245482787489891,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1746
+    },
+    {
+      "epoch": 2.8480599934789694,
+      "grad_norm": 0.1273997575044632,
+      "learning_rate": 0.0001,
+      "loss": 0.1832,
+      "step": 1747
+    },
+    {
+      "epoch": 2.8496902510596676,
+      "grad_norm": 0.1431506723165512,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1748
+    },
+    {
+      "epoch": 2.8513205086403652,
+      "grad_norm": 0.12924441695213318,
+      "learning_rate": 0.0001,
+      "loss": 0.1686,
+      "step": 1749
+    },
+    {
+      "epoch": 2.852950766221063,
+      "grad_norm": 0.12894988059997559,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1750
+    },
+    {
+      "epoch": 2.8545810238017606,
+      "grad_norm": 0.1198873519897461,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 1751
+    },
+    {
+      "epoch": 2.8562112813824583,
+      "grad_norm": 0.12302883714437485,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 1752
+    },
+    {
+      "epoch": 2.8578415389631564,
+      "grad_norm": 0.14283844828605652,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 1753
+    },
+    {
+      "epoch": 2.859471796543854,
+      "grad_norm": 0.1237570121884346,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 1754
+    },
+    {
+      "epoch": 2.8611020541245518,
+      "grad_norm": 0.12284982204437256,
+      "learning_rate": 0.0001,
+      "loss": 0.1813,
+      "step": 1755
+    },
+    {
+      "epoch": 2.8627323117052494,
+      "grad_norm": 0.16714490950107574,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 1756
+    },
+    {
+      "epoch": 2.864362569285947,
+      "grad_norm": 0.12366341799497604,
+      "learning_rate": 0.0001,
+      "loss": 0.1715,
+      "step": 1757
+    },
+    {
+      "epoch": 2.865992826866645,
+      "grad_norm": 0.13020524382591248,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "step": 1758
+    },
+    {
+      "epoch": 2.8676230844473425,
+      "grad_norm": 0.11508645117282867,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "step": 1759
+    },
+    {
+      "epoch": 2.86925334202804,
+      "grad_norm": 0.13335858285427094,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 1760
+    },
+    {
+      "epoch": 2.8708835996087383,
+      "grad_norm": 0.12083142250776291,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 1761
+    },
+    {
+      "epoch": 2.872513857189436,
+      "grad_norm": 0.12425024807453156,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1762
+    },
+    {
+      "epoch": 2.8741441147701337,
+      "grad_norm": 0.11999020725488663,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 1763
+    },
+    {
+      "epoch": 2.8757743723508313,
+      "grad_norm": 0.12801139056682587,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 1764
+    },
+    {
+      "epoch": 2.8774046299315295,
+      "grad_norm": 0.12506726384162903,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 1765
+    },
+    {
+      "epoch": 2.879034887512227,
+      "grad_norm": 0.12049156427383423,
+      "learning_rate": 0.0001,
+      "loss": 0.1794,
+      "step": 1766
+    },
+    {
+      "epoch": 2.880665145092925,
+      "grad_norm": 0.11800320446491241,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 1767
+    },
+    {
+      "epoch": 2.8822954026736225,
+      "grad_norm": 0.13116998970508575,
+      "learning_rate": 0.0001,
+      "loss": 0.1742,
+      "step": 1768
+    },
+    {
+      "epoch": 2.88392566025432,
+      "grad_norm": 0.11840593814849854,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1769
+    },
+    {
+      "epoch": 2.885555917835018,
+      "grad_norm": 0.11977692693471909,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 1770
+    },
+    {
+      "epoch": 2.8871861754157155,
+      "grad_norm": 0.13329051434993744,
+      "learning_rate": 0.0001,
+      "loss": 0.1786,
+      "step": 1771
+    },
+    {
+      "epoch": 2.888816432996413,
+      "grad_norm": 0.19213466346263885,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 1772
+    },
+    {
+      "epoch": 2.8904466905771113,
+      "grad_norm": 0.1305350363254547,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 1773
+    },
+    {
+      "epoch": 2.892076948157809,
+      "grad_norm": 0.14388902485370636,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 1774
+    },
+    {
+      "epoch": 2.8937072057385067,
+      "grad_norm": 0.1325220912694931,
+      "learning_rate": 0.0001,
+      "loss": 0.1674,
+      "step": 1775
+    },
+    {
+      "epoch": 2.8953374633192044,
+      "grad_norm": 0.14095838367938995,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 1776
+    },
+    {
+      "epoch": 2.896967720899902,
+      "grad_norm": 0.12072951346635818,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 1777
+    },
+    {
+      "epoch": 2.8985979784806,
+      "grad_norm": 0.15178942680358887,
+      "learning_rate": 0.0001,
+      "loss": 0.1891,
+      "step": 1778
+    },
+    {
+      "epoch": 2.900228236061298,
+      "grad_norm": 0.19420669972896576,
+      "learning_rate": 0.0001,
+      "loss": 0.1833,
+      "step": 1779
+    },
+    {
+      "epoch": 2.9018584936419956,
+      "grad_norm": 0.125785693526268,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 1780
+    },
+    {
+      "epoch": 2.9034887512226932,
+      "grad_norm": 0.13364505767822266,
+      "learning_rate": 0.0001,
+      "loss": 0.1561,
+      "step": 1781
+    },
+    {
+      "epoch": 2.905119008803391,
+      "grad_norm": 0.12486680597066879,
+      "learning_rate": 0.0001,
+      "loss": 0.1758,
+      "step": 1782
+    },
+    {
+      "epoch": 2.9067492663840886,
+      "grad_norm": 0.13013507425785065,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 1783
+    },
+    {
+      "epoch": 2.9083795239647863,
+      "grad_norm": 0.1089881956577301,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 1784
+    },
+    {
+      "epoch": 2.910009781545484,
+      "grad_norm": 0.145898699760437,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 1785
+    },
+    {
+      "epoch": 2.911640039126182,
+      "grad_norm": 0.12093661725521088,
+      "learning_rate": 0.0001,
+      "loss": 0.1829,
+      "step": 1786
+    },
+    {
+      "epoch": 2.9132702967068798,
+      "grad_norm": 0.12347777187824249,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 1787
+    },
+    {
+      "epoch": 2.9149005542875774,
+      "grad_norm": 0.10860280692577362,
+      "learning_rate": 0.0001,
+      "loss": 0.1674,
+      "step": 1788
+    },
+    {
+      "epoch": 2.916530811868275,
+      "grad_norm": 0.12860281765460968,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 1789
+    },
+    {
+      "epoch": 2.918161069448973,
+      "grad_norm": 0.12744852900505066,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 1790
+    },
+    {
+      "epoch": 2.919791327029671,
+      "grad_norm": 0.13266021013259888,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 1791
+    },
+    {
+      "epoch": 2.9214215846103686,
+      "grad_norm": 0.1323481798171997,
+      "learning_rate": 0.0001,
+      "loss": 0.1786,
+      "step": 1792
+    },
+    {
+      "epoch": 2.9230518421910663,
+      "grad_norm": 0.12896117568016052,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1793
+    },
+    {
+      "epoch": 2.924682099771764,
+      "grad_norm": 0.12157560884952545,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 1794
+    },
+    {
+      "epoch": 2.9263123573524616,
+      "grad_norm": 0.12362449616193771,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 1795
+    },
+    {
+      "epoch": 2.9279426149331593,
+      "grad_norm": 0.1297728419303894,
+      "learning_rate": 0.0001,
+      "loss": 0.1897,
+      "step": 1796
+    },
+    {
+      "epoch": 2.929572872513857,
+      "grad_norm": 0.10805771499872208,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 1797
+    },
+    {
+      "epoch": 2.9312031300945547,
+      "grad_norm": 0.1192641630768776,
+      "learning_rate": 0.0001,
+      "loss": 0.176,
+      "step": 1798
+    },
+    {
+      "epoch": 2.932833387675253,
+      "grad_norm": 0.14953552186489105,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 1799
+    },
+    {
+      "epoch": 2.9344636452559505,
+      "grad_norm": 0.12457464635372162,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 1800
+    },
+    {
+      "epoch": 2.936093902836648,
+      "grad_norm": 0.12195872515439987,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 1801
+    },
+    {
+      "epoch": 2.937724160417346,
+      "grad_norm": 0.1217537596821785,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 1802
+    },
+    {
+      "epoch": 2.9393544179980435,
+      "grad_norm": 0.10911030322313309,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 1803
+    },
+    {
+      "epoch": 2.9409846755787417,
+      "grad_norm": 0.15903405845165253,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1804
+    },
+    {
+      "epoch": 2.9426149331594393,
+      "grad_norm": 0.14546822011470795,
+      "learning_rate": 0.0001,
+      "loss": 0.1735,
+      "step": 1805
+    },
+    {
+      "epoch": 2.944245190740137,
+      "grad_norm": 0.13845416903495789,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 1806
+    },
+    {
+      "epoch": 2.9458754483208347,
+      "grad_norm": 0.12575270235538483,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 1807
+    },
+    {
+      "epoch": 2.9475057059015324,
+      "grad_norm": 0.11531825363636017,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 1808
+    },
+    {
+      "epoch": 2.94913596348223,
+      "grad_norm": 0.12678653001785278,
+      "learning_rate": 0.0001,
+      "loss": 0.1782,
+      "step": 1809
+    },
+    {
+      "epoch": 2.9507662210629277,
+      "grad_norm": 0.11455704271793365,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 1810
+    },
+    {
+      "epoch": 2.9523964786436254,
+      "grad_norm": 0.12815020978450775,
+      "learning_rate": 0.0001,
+      "loss": 0.1821,
+      "step": 1811
+    },
+    {
+      "epoch": 2.9540267362243235,
+      "grad_norm": 0.14364176988601685,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1812
+    },
+    {
+      "epoch": 2.9556569938050212,
+      "grad_norm": 0.1432567834854126,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 1813
+    },
+    {
+      "epoch": 2.957287251385719,
+      "grad_norm": 0.1255834549665451,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 1814
+    },
+    {
+      "epoch": 2.9589175089664166,
+      "grad_norm": 0.13830745220184326,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 1815
+    },
+    {
+      "epoch": 2.9605477665471147,
+      "grad_norm": 0.1475990116596222,
+      "learning_rate": 0.0001,
+      "loss": 0.1938,
+      "step": 1816
+    },
+    {
+      "epoch": 2.9621780241278124,
+      "grad_norm": 0.13100960850715637,
+      "learning_rate": 0.0001,
+      "loss": 0.1877,
+      "step": 1817
+    },
+    {
+      "epoch": 2.96380828170851,
+      "grad_norm": 0.11911682784557343,
+      "learning_rate": 0.0001,
+      "loss": 0.1668,
+      "step": 1818
+    },
+    {
+      "epoch": 2.9654385392892078,
+      "grad_norm": 0.13728202879428864,
+      "learning_rate": 0.0001,
+      "loss": 0.181,
+      "step": 1819
+    },
+    {
+      "epoch": 2.9670687968699054,
+      "grad_norm": 0.1287861168384552,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 1820
+    },
+    {
+      "epoch": 2.968699054450603,
+      "grad_norm": 0.1377815157175064,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1821
+    },
+    {
+      "epoch": 2.970329312031301,
+      "grad_norm": 0.13828015327453613,
+      "learning_rate": 0.0001,
+      "loss": 0.1751,
+      "step": 1822
+    },
+    {
+      "epoch": 2.9719595696119985,
+      "grad_norm": 0.13699176907539368,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 1823
+    },
+    {
+      "epoch": 2.9735898271926966,
+      "grad_norm": 0.13141410052776337,
+      "learning_rate": 0.0001,
+      "loss": 0.1686,
+      "step": 1824
+    },
+    {
+      "epoch": 2.9752200847733943,
+      "grad_norm": 0.11644890159368515,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 1825
+    },
+    {
+      "epoch": 2.976850342354092,
+      "grad_norm": 0.12482846528291702,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 1826
+    },
+    {
+      "epoch": 2.9784805999347896,
+      "grad_norm": 0.1434062272310257,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 1827
+    },
+    {
+      "epoch": 2.9801108575154873,
+      "grad_norm": 0.1083054170012474,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1828
+    },
+    {
+      "epoch": 2.9817411150961854,
+      "grad_norm": 0.10897631198167801,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 1829
+    },
+    {
+      "epoch": 2.983371372676883,
+      "grad_norm": 0.12798282504081726,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 1830
+    },
+    {
+      "epoch": 2.985001630257581,
+      "grad_norm": 0.13066311180591583,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 1831
+    },
+    {
+      "epoch": 2.9866318878382785,
+      "grad_norm": 0.12644776701927185,
+      "learning_rate": 0.0001,
+      "loss": 0.1772,
+      "step": 1832
+    },
+    {
+      "epoch": 2.988262145418976,
+      "grad_norm": 0.14311029016971588,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 1833
+    },
+    {
+      "epoch": 2.989892402999674,
+      "grad_norm": 0.12313991785049438,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 1834
+    },
+    {
+      "epoch": 2.9915226605803715,
+      "grad_norm": 0.13823242485523224,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1835
+    },
+    {
+      "epoch": 2.993152918161069,
+      "grad_norm": 0.12686224281787872,
+      "learning_rate": 0.0001,
+      "loss": 0.1841,
+      "step": 1836
+    },
+    {
+      "epoch": 2.9947831757417673,
+      "grad_norm": 0.17587055265903473,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 1837
+    },
+    {
+      "epoch": 2.996413433322465,
+      "grad_norm": 0.11246396601200104,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 1838
+    },
+    {
+      "epoch": 2.9980436909031627,
+      "grad_norm": 0.13493315875530243,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 1839
+    },
+    {
+      "epoch": 2.9996739484838604,
+      "grad_norm": 0.12286293506622314,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 1840
+    },
+    {
+      "epoch": 3.001304206064558,
+      "grad_norm": 0.1269206702709198,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 1841
+    },
+    {
+      "epoch": 3.0029344636452557,
+      "grad_norm": 0.1355823576450348,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 1842
+    },
+    {
+      "epoch": 3.004564721225954,
+      "grad_norm": 0.11628458648920059,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 1843
+    },
+    {
+      "epoch": 3.0061949788066515,
+      "grad_norm": 0.11154302209615707,
+      "learning_rate": 0.0001,
+      "loss": 0.1444,
+      "step": 1844
+    },
+    {
+      "epoch": 3.007825236387349,
+      "grad_norm": 0.14044183492660522,
+      "learning_rate": 0.0001,
+      "loss": 0.1484,
+      "step": 1845
+    },
+    {
+      "epoch": 3.009455493968047,
+      "grad_norm": 0.11825685203075409,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 1846
+    },
+    {
+      "epoch": 3.0110857515487446,
+      "grad_norm": 0.17750664055347443,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 1847
+    },
+    {
+      "epoch": 3.0127160091294423,
+      "grad_norm": 0.16660550236701965,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 1848
+    },
+    {
+      "epoch": 3.0143462667101404,
+      "grad_norm": 0.16394905745983124,
+      "learning_rate": 0.0001,
+      "loss": 0.1738,
+      "step": 1849
+    },
+    {
+      "epoch": 3.015976524290838,
+      "grad_norm": 0.1530742645263672,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 1850
+    },
+    {
+      "epoch": 3.0176067818715357,
+      "grad_norm": 0.13188475370407104,
+      "learning_rate": 0.0001,
+      "loss": 0.1564,
+      "step": 1851
+    },
+    {
+      "epoch": 3.0192370394522334,
+      "grad_norm": 0.13797006011009216,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 1852
+    },
+    {
+      "epoch": 3.020867297032931,
+      "grad_norm": 0.16686192154884338,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 1853
+    },
+    {
+      "epoch": 3.022497554613629,
+      "grad_norm": 0.13372887670993805,
+      "learning_rate": 0.0001,
+      "loss": 0.1522,
+      "step": 1854
+    },
+    {
+      "epoch": 3.024127812194327,
+      "grad_norm": 0.1713315099477768,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 1855
+    },
+    {
+      "epoch": 3.0257580697750246,
+      "grad_norm": 0.17664030194282532,
+      "learning_rate": 0.0001,
+      "loss": 0.1706,
+      "step": 1856
+    },
+    {
+      "epoch": 3.0273883273557223,
+      "grad_norm": 0.15478695929050446,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 1857
+    },
+    {
+      "epoch": 3.02901858493642,
+      "grad_norm": 0.12312593311071396,
+      "learning_rate": 0.0001,
+      "loss": 0.1444,
+      "step": 1858
+    },
+    {
+      "epoch": 3.0306488425171176,
+      "grad_norm": 0.13076375424861908,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 1859
+    },
+    {
+      "epoch": 3.0322791000978153,
+      "grad_norm": 0.17195327579975128,
+      "learning_rate": 0.0001,
+      "loss": 0.1568,
+      "step": 1860
+    },
+    {
+      "epoch": 3.033909357678513,
+      "grad_norm": 0.15723173320293427,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 1861
+    },
+    {
+      "epoch": 3.035539615259211,
+      "grad_norm": 0.13662908971309662,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 1862
+    },
+    {
+      "epoch": 3.037169872839909,
+      "grad_norm": 0.14388419687747955,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 1863
+    },
+    {
+      "epoch": 3.0388001304206065,
+      "grad_norm": 0.1430109143257141,
+      "learning_rate": 0.0001,
+      "loss": 0.1448,
+      "step": 1864
+    },
+    {
+      "epoch": 3.040430388001304,
+      "grad_norm": 0.18873119354248047,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1865
+    },
+    {
+      "epoch": 3.042060645582002,
+      "grad_norm": 0.13666655123233795,
+      "learning_rate": 0.0001,
+      "loss": 0.1489,
+      "step": 1866
+    },
+    {
+      "epoch": 3.0436909031626995,
+      "grad_norm": 0.12187852710485458,
+      "learning_rate": 0.0001,
+      "loss": 0.1486,
+      "step": 1867
+    },
+    {
+      "epoch": 3.0453211607433976,
+      "grad_norm": 0.14884653687477112,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 1868
+    },
+    {
+      "epoch": 3.0469514183240953,
+      "grad_norm": 0.1332685947418213,
+      "learning_rate": 0.0001,
+      "loss": 0.1504,
+      "step": 1869
+    },
+    {
+      "epoch": 3.048581675904793,
+      "grad_norm": 0.15968628227710724,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 1870
+    },
+    {
+      "epoch": 3.0502119334854907,
+      "grad_norm": 0.13830626010894775,
+      "learning_rate": 0.0001,
+      "loss": 0.1556,
+      "step": 1871
+    },
+    {
+      "epoch": 3.0518421910661884,
+      "grad_norm": 0.14742417633533478,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 1872
+    },
+    {
+      "epoch": 3.053472448646886,
+      "grad_norm": 0.1588188111782074,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 1873
+    },
+    {
+      "epoch": 3.055102706227584,
+      "grad_norm": 0.14762498438358307,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 1874
+    },
+    {
+      "epoch": 3.056732963808282,
+      "grad_norm": 0.13328132033348083,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 1875
+    },
+    {
+      "epoch": 3.0583632213889795,
+      "grad_norm": 0.12717348337173462,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 1876
+    },
+    {
+      "epoch": 3.059993478969677,
+      "grad_norm": 0.15513722598552704,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 1877
+    },
+    {
+      "epoch": 3.061623736550375,
+      "grad_norm": 0.12475986033678055,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 1878
+    },
+    {
+      "epoch": 3.0632539941310726,
+      "grad_norm": 0.1064193844795227,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 1879
+    },
+    {
+      "epoch": 3.0648842517117703,
+      "grad_norm": 0.14102140069007874,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 1880
+    },
+    {
+      "epoch": 3.0665145092924684,
+      "grad_norm": 0.1496179848909378,
+      "learning_rate": 0.0001,
+      "loss": 0.157,
+      "step": 1881
+    },
+    {
+      "epoch": 3.068144766873166,
+      "grad_norm": 0.13995377719402313,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 1882
+    },
+    {
+      "epoch": 3.0697750244538637,
+      "grad_norm": 0.1414022445678711,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 1883
+    },
+    {
+      "epoch": 3.0714052820345614,
+      "grad_norm": 0.15052585303783417,
+      "learning_rate": 0.0001,
+      "loss": 0.1732,
+      "step": 1884
+    },
+    {
+      "epoch": 3.073035539615259,
+      "grad_norm": 0.1524229496717453,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1885
+    },
+    {
+      "epoch": 3.074665797195957,
+      "grad_norm": 0.13956795632839203,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1886
+    },
+    {
+      "epoch": 3.076296054776655,
+      "grad_norm": 0.13106916844844818,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 1887
+    },
+    {
+      "epoch": 3.0779263123573526,
+      "grad_norm": 0.1438571810722351,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1888
+    },
+    {
+      "epoch": 3.0795565699380503,
+      "grad_norm": 0.12486065924167633,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 1889
+    },
+    {
+      "epoch": 3.081186827518748,
+      "grad_norm": 0.15477533638477325,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1890
+    },
+    {
+      "epoch": 3.0828170850994456,
+      "grad_norm": 0.14669184386730194,
+      "learning_rate": 0.0001,
+      "loss": 0.1688,
+      "step": 1891
+    },
+    {
+      "epoch": 3.0844473426801433,
+      "grad_norm": 0.12497131526470184,
+      "learning_rate": 0.0001,
+      "loss": 0.1452,
+      "step": 1892
+    },
+    {
+      "epoch": 3.0860776002608414,
+      "grad_norm": 0.1318637579679489,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 1893
+    },
+    {
+      "epoch": 3.087707857841539,
+      "grad_norm": 0.1398518979549408,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 1894
+    },
+    {
+      "epoch": 3.089338115422237,
+      "grad_norm": 0.13344673812389374,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 1895
+    },
+    {
+      "epoch": 3.0909683730029345,
+      "grad_norm": 0.13853240013122559,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 1896
+    },
+    {
+      "epoch": 3.092598630583632,
+      "grad_norm": 0.15498271584510803,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 1897
+    },
+    {
+      "epoch": 3.09422888816433,
+      "grad_norm": 0.1456194967031479,
+      "learning_rate": 0.0001,
+      "loss": 0.1535,
+      "step": 1898
+    },
+    {
+      "epoch": 3.0958591457450275,
+      "grad_norm": 0.15903611481189728,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 1899
+    },
+    {
+      "epoch": 3.0974894033257256,
+      "grad_norm": 0.1665191799402237,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1900
+    },
+    {
+      "epoch": 3.0991196609064233,
+      "grad_norm": 0.1596616506576538,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 1901
+    },
+    {
+      "epoch": 3.100749918487121,
+      "grad_norm": 0.14981094002723694,
+      "learning_rate": 0.0001,
+      "loss": 0.1555,
+      "step": 1902
+    },
+    {
+      "epoch": 3.1023801760678187,
+      "grad_norm": 0.1317712813615799,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 1903
+    },
+    {
+      "epoch": 3.1040104336485164,
+      "grad_norm": 0.12632454931735992,
+      "learning_rate": 0.0001,
+      "loss": 0.1364,
+      "step": 1904
+    },
+    {
+      "epoch": 3.105640691229214,
+      "grad_norm": 0.16619497537612915,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 1905
+    },
+    {
+      "epoch": 3.107270948809912,
+      "grad_norm": 0.12969553470611572,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 1906
+    },
+    {
+      "epoch": 3.10890120639061,
+      "grad_norm": 0.1580219268798828,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 1907
+    },
+    {
+      "epoch": 3.1105314639713075,
+      "grad_norm": 0.14800585806369781,
+      "learning_rate": 0.0001,
+      "loss": 0.1775,
+      "step": 1908
+    },
+    {
+      "epoch": 3.112161721552005,
+      "grad_norm": 0.13492430746555328,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 1909
+    },
+    {
+      "epoch": 3.113791979132703,
+      "grad_norm": 0.13859906792640686,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1910
+    },
+    {
+      "epoch": 3.1154222367134006,
+      "grad_norm": 0.1361161321401596,
+      "learning_rate": 0.0001,
+      "loss": 0.1537,
+      "step": 1911
+    },
+    {
+      "epoch": 3.1170524942940983,
+      "grad_norm": 0.17924432456493378,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 1912
+    },
+    {
+      "epoch": 3.1186827518747964,
+      "grad_norm": 0.14079366624355316,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 1913
+    },
+    {
+      "epoch": 3.120313009455494,
+      "grad_norm": 0.13240069150924683,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 1914
+    },
+    {
+      "epoch": 3.1219432670361917,
+      "grad_norm": 0.1292378306388855,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 1915
+    },
+    {
+      "epoch": 3.1235735246168894,
+      "grad_norm": 0.1428757756948471,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 1916
+    },
+    {
+      "epoch": 3.125203782197587,
+      "grad_norm": 0.1274784803390503,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 1917
+    },
+    {
+      "epoch": 3.126834039778285,
+      "grad_norm": 0.13806386291980743,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 1918
+    },
+    {
+      "epoch": 3.128464297358983,
+      "grad_norm": 0.12677744030952454,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 1919
+    },
+    {
+      "epoch": 3.1300945549396806,
+      "grad_norm": 0.13189762830734253,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 1920
+    },
+    {
+      "epoch": 3.1317248125203783,
+      "grad_norm": 0.18951299786567688,
+      "learning_rate": 0.0001,
+      "loss": 0.1674,
+      "step": 1921
+    },
+    {
+      "epoch": 3.133355070101076,
+      "grad_norm": 0.1415640413761139,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 1922
+    },
+    {
+      "epoch": 3.1349853276817736,
+      "grad_norm": 0.1354334056377411,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 1923
+    },
+    {
+      "epoch": 3.1366155852624713,
+      "grad_norm": 0.15447008609771729,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 1924
+    },
+    {
+      "epoch": 3.138245842843169,
+      "grad_norm": 0.14421270787715912,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 1925
+    },
+    {
+      "epoch": 3.139876100423867,
+      "grad_norm": 0.13568207621574402,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 1926
+    },
+    {
+      "epoch": 3.141506358004565,
+      "grad_norm": 0.140906423330307,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 1927
+    },
+    {
+      "epoch": 3.1431366155852625,
+      "grad_norm": 0.11631222814321518,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 1928
+    },
+    {
+      "epoch": 3.14476687316596,
+      "grad_norm": 0.15498870611190796,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 1929
+    },
+    {
+      "epoch": 3.146397130746658,
+      "grad_norm": 0.16634507477283478,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 1930
+    },
+    {
+      "epoch": 3.148027388327356,
+      "grad_norm": 0.1286221295595169,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 1931
+    },
+    {
+      "epoch": 3.1496576459080536,
+      "grad_norm": 0.14258523285388947,
+      "learning_rate": 0.0001,
+      "loss": 0.1556,
+      "step": 1932
+    },
+    {
+      "epoch": 3.1512879034887513,
+      "grad_norm": 0.1418234407901764,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 1933
+    },
+    {
+      "epoch": 3.152918161069449,
+      "grad_norm": 0.13352856040000916,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 1934
+    },
+    {
+      "epoch": 3.1545484186501467,
+      "grad_norm": 0.12372912466526031,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 1935
+    },
+    {
+      "epoch": 3.1561786762308444,
+      "grad_norm": 0.15148986876010895,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 1936
+    },
+    {
+      "epoch": 3.157808933811542,
+      "grad_norm": 0.13402307033538818,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 1937
+    },
+    {
+      "epoch": 3.15943919139224,
+      "grad_norm": 0.15002478659152985,
+      "learning_rate": 0.0001,
+      "loss": 0.1552,
+      "step": 1938
+    },
+    {
+      "epoch": 3.161069448972938,
+      "grad_norm": 0.14147819578647614,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 1939
+    },
+    {
+      "epoch": 3.1626997065536355,
+      "grad_norm": 0.16846194863319397,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 1940
+    },
+    {
+      "epoch": 3.164329964134333,
+      "grad_norm": 0.17385046184062958,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 1941
+    },
+    {
+      "epoch": 3.165960221715031,
+      "grad_norm": 0.14873671531677246,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 1942
+    },
+    {
+      "epoch": 3.1675904792957286,
+      "grad_norm": 0.13968585431575775,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 1943
+    },
+    {
+      "epoch": 3.1692207368764267,
+      "grad_norm": 0.15636034309864044,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 1944
+    },
+    {
+      "epoch": 3.1708509944571244,
+      "grad_norm": 0.17083799839019775,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 1945
+    },
+    {
+      "epoch": 3.172481252037822,
+      "grad_norm": 0.13680125772953033,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 1946
+    },
+    {
+      "epoch": 3.1741115096185197,
+      "grad_norm": 0.14011834561824799,
+      "learning_rate": 0.0001,
+      "loss": 0.1407,
+      "step": 1947
+    },
+    {
+      "epoch": 3.1757417671992174,
+      "grad_norm": 0.14125655591487885,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 1948
+    },
+    {
+      "epoch": 3.177372024779915,
+      "grad_norm": 0.13686692714691162,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 1949
+    },
+    {
+      "epoch": 3.1790022823606128,
+      "grad_norm": 0.14849920570850372,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 1950
+    },
+    {
+      "epoch": 3.180632539941311,
+      "grad_norm": 0.2647811770439148,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 1951
+    },
+    {
+      "epoch": 3.1822627975220086,
+      "grad_norm": 0.1454608142375946,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 1952
+    },
+    {
+      "epoch": 3.1838930551027063,
+      "grad_norm": 0.15348565578460693,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 1953
+    },
+    {
+      "epoch": 3.185523312683404,
+      "grad_norm": 0.14148078858852386,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 1954
+    },
+    {
+      "epoch": 3.1871535702641016,
+      "grad_norm": 0.16094723343849182,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 1955
+    },
+    {
+      "epoch": 3.1887838278447993,
+      "grad_norm": 0.6947292685508728,
+      "learning_rate": 0.0001,
+      "loss": 0.1537,
+      "step": 1956
+    },
+    {
+      "epoch": 3.1904140854254974,
+      "grad_norm": 0.138764888048172,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 1957
+    },
+    {
+      "epoch": 3.192044343006195,
+      "grad_norm": 0.19771906733512878,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 1958
+    },
+    {
+      "epoch": 3.193674600586893,
+      "grad_norm": 0.15238133072853088,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 1959
+    },
+    {
+      "epoch": 3.1953048581675905,
+      "grad_norm": 0.14184920489788055,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 1960
+    },
+    {
+      "epoch": 3.196935115748288,
+      "grad_norm": 0.13385626673698425,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 1961
+    },
+    {
+      "epoch": 3.198565373328986,
+      "grad_norm": 0.1413835883140564,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 1962
+    },
+    {
+      "epoch": 3.2001956309096835,
+      "grad_norm": 0.14667753875255585,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 1963
+    },
+    {
+      "epoch": 3.2018258884903816,
+      "grad_norm": 0.16244405508041382,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 1964
+    },
+    {
+      "epoch": 3.2034561460710793,
+      "grad_norm": 0.157258540391922,
+      "learning_rate": 0.0001,
+      "loss": 0.1567,
+      "step": 1965
+    },
+    {
+      "epoch": 3.205086403651777,
+      "grad_norm": 0.16143973171710968,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 1966
+    },
+    {
+      "epoch": 3.2067166612324747,
+      "grad_norm": 0.15137411653995514,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 1967
+    },
+    {
+      "epoch": 3.2083469188131724,
+      "grad_norm": 0.13595400750637054,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 1968
+    },
+    {
+      "epoch": 3.20997717639387,
+      "grad_norm": 0.1366918534040451,
+      "learning_rate": 0.0001,
+      "loss": 0.1479,
+      "step": 1969
+    },
+    {
+      "epoch": 3.211607433974568,
+      "grad_norm": 0.12027698010206223,
+      "learning_rate": 0.0001,
+      "loss": 0.1475,
+      "step": 1970
+    },
+    {
+      "epoch": 3.213237691555266,
+      "grad_norm": 0.11735312640666962,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 1971
+    },
+    {
+      "epoch": 3.2148679491359635,
+      "grad_norm": 0.13770481944084167,
+      "learning_rate": 0.0001,
+      "loss": 0.1463,
+      "step": 1972
+    },
+    {
+      "epoch": 3.216498206716661,
+      "grad_norm": 0.14083652198314667,
+      "learning_rate": 0.0001,
+      "loss": 0.1596,
+      "step": 1973
+    },
+    {
+      "epoch": 3.218128464297359,
+      "grad_norm": 0.13524475693702698,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 1974
+    },
+    {
+      "epoch": 3.2197587218780566,
+      "grad_norm": 0.13120479881763458,
+      "learning_rate": 0.0001,
+      "loss": 0.1596,
+      "step": 1975
+    },
+    {
+      "epoch": 3.2213889794587547,
+      "grad_norm": 0.17046691477298737,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 1976
+    },
+    {
+      "epoch": 3.2230192370394524,
+      "grad_norm": 0.13418637216091156,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 1977
+    },
+    {
+      "epoch": 3.22464949462015,
+      "grad_norm": 0.11807394027709961,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 1978
+    },
+    {
+      "epoch": 3.2262797522008477,
+      "grad_norm": 0.15155768394470215,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 1979
+    },
+    {
+      "epoch": 3.2279100097815454,
+      "grad_norm": 0.1263338178396225,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 1980
+    },
+    {
+      "epoch": 3.229540267362243,
+      "grad_norm": 0.1663949191570282,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 1981
+    },
+    {
+      "epoch": 3.231170524942941,
+      "grad_norm": 0.1378049999475479,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 1982
+    },
+    {
+      "epoch": 3.232800782523639,
+      "grad_norm": 0.1389368176460266,
+      "learning_rate": 0.0001,
+      "loss": 0.1487,
+      "step": 1983
+    },
+    {
+      "epoch": 3.2344310401043366,
+      "grad_norm": 0.15180104970932007,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 1984
+    },
+    {
+      "epoch": 3.2360612976850343,
+      "grad_norm": 0.14475220441818237,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 1985
+    },
+    {
+      "epoch": 3.237691555265732,
+      "grad_norm": 0.14919638633728027,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 1986
+    },
+    {
+      "epoch": 3.2393218128464296,
+      "grad_norm": 0.14045490324497223,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 1987
+    },
+    {
+      "epoch": 3.2409520704271273,
+      "grad_norm": 0.1450393944978714,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 1988
+    },
+    {
+      "epoch": 3.2425823280078254,
+      "grad_norm": 0.12991821765899658,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 1989
+    },
+    {
+      "epoch": 3.244212585588523,
+      "grad_norm": 0.1459466964006424,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1990
+    },
+    {
+      "epoch": 3.245842843169221,
+      "grad_norm": 0.1518615484237671,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 1991
+    },
+    {
+      "epoch": 3.2474731007499185,
+      "grad_norm": 0.14151298999786377,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 1992
+    },
+    {
+      "epoch": 3.249103358330616,
+      "grad_norm": 0.13873840868473053,
+      "learning_rate": 0.0001,
+      "loss": 0.1434,
+      "step": 1993
+    },
+    {
+      "epoch": 3.250733615911314,
+      "grad_norm": 0.15724745392799377,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 1994
+    },
+    {
+      "epoch": 3.252363873492012,
+      "grad_norm": 0.12847794592380524,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 1995
+    },
+    {
+      "epoch": 3.2539941310727096,
+      "grad_norm": 0.1553923785686493,
+      "learning_rate": 0.0001,
+      "loss": 0.1673,
+      "step": 1996
+    },
+    {
+      "epoch": 3.2556243886534073,
+      "grad_norm": 0.17050054669380188,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1997
+    },
+    {
+      "epoch": 3.257254646234105,
+      "grad_norm": 0.13154077529907227,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 1998
+    },
+    {
+      "epoch": 3.2588849038148027,
+      "grad_norm": 0.1389462798833847,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 1999
+    },
+    {
+      "epoch": 3.2605151613955003,
+      "grad_norm": 0.12226948142051697,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2605151613955003,
+      "eval_loss": 0.2004072368144989,
+      "eval_runtime": 463.7076,
+      "eval_samples_per_second": 4.069,
+      "eval_steps_per_second": 1.018,
+      "step": 2000
+    },
+    {
+      "epoch": 3.262145418976198,
+      "grad_norm": 0.13234886527061462,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 2001
+    },
+    {
+      "epoch": 3.263775676556896,
+      "grad_norm": 0.12151322513818741,
+      "learning_rate": 0.0001,
+      "loss": 0.1433,
+      "step": 2002
+    },
+    {
+      "epoch": 3.265405934137594,
+      "grad_norm": 0.16095128655433655,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 2003
+    },
+    {
+      "epoch": 3.2670361917182915,
+      "grad_norm": 0.1361403614282608,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2004
+    },
+    {
+      "epoch": 3.268666449298989,
+      "grad_norm": 0.13729725778102875,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 2005
+    },
+    {
+      "epoch": 3.270296706879687,
+      "grad_norm": 0.16737613081932068,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 2006
+    },
+    {
+      "epoch": 3.2719269644603846,
+      "grad_norm": 0.13619449734687805,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 2007
+    },
+    {
+      "epoch": 3.2735572220410827,
+      "grad_norm": 0.1604672521352768,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 2008
+    },
+    {
+      "epoch": 3.2751874796217804,
+      "grad_norm": 0.14666776359081268,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 2009
+    },
+    {
+      "epoch": 3.276817737202478,
+      "grad_norm": 0.1353665441274643,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 2010
+    },
+    {
+      "epoch": 3.2784479947831757,
+      "grad_norm": 0.1277618557214737,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 2011
+    },
+    {
+      "epoch": 3.2800782523638734,
+      "grad_norm": 0.16270460188388824,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2012
+    },
+    {
+      "epoch": 3.281708509944571,
+      "grad_norm": 0.15774473547935486,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2013
+    },
+    {
+      "epoch": 3.2833387675252688,
+      "grad_norm": 0.13945020735263824,
+      "learning_rate": 0.0001,
+      "loss": 0.1484,
+      "step": 2014
+    },
+    {
+      "epoch": 3.284969025105967,
+      "grad_norm": 0.14141727983951569,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 2015
+    },
+    {
+      "epoch": 3.2865992826866646,
+      "grad_norm": 0.12424743920564651,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 2016
+    },
+    {
+      "epoch": 3.2882295402673622,
+      "grad_norm": 0.12121134996414185,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 2017
+    },
+    {
+      "epoch": 3.28985979784806,
+      "grad_norm": 0.1380610466003418,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 2018
+    },
+    {
+      "epoch": 3.2914900554287576,
+      "grad_norm": 0.10836822539567947,
+      "learning_rate": 0.0001,
+      "loss": 0.1328,
+      "step": 2019
+    },
+    {
+      "epoch": 3.2931203130094557,
+      "grad_norm": 0.12385184317827225,
+      "learning_rate": 0.0001,
+      "loss": 0.1446,
+      "step": 2020
+    },
+    {
+      "epoch": 3.2947505705901534,
+      "grad_norm": 0.13645856082439423,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 2021
+    },
+    {
+      "epoch": 3.296380828170851,
+      "grad_norm": 0.14024437963962555,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 2022
+    },
+    {
+      "epoch": 3.2980110857515488,
+      "grad_norm": 0.13787047564983368,
+      "learning_rate": 0.0001,
+      "loss": 0.1601,
+      "step": 2023
+    },
+    {
+      "epoch": 3.2996413433322465,
+      "grad_norm": 0.14804314076900482,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 2024
+    },
+    {
+      "epoch": 3.301271600912944,
+      "grad_norm": 0.1506766825914383,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 2025
+    },
+    {
+      "epoch": 3.302901858493642,
+      "grad_norm": 0.1486261785030365,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 2026
+    },
+    {
+      "epoch": 3.3045321160743395,
+      "grad_norm": 0.16172239184379578,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 2027
+    },
+    {
+      "epoch": 3.3061623736550376,
+      "grad_norm": 0.1388084590435028,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 2028
+    },
+    {
+      "epoch": 3.3077926312357353,
+      "grad_norm": 0.15074113011360168,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 2029
+    },
+    {
+      "epoch": 3.309422888816433,
+      "grad_norm": 0.14456892013549805,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 2030
+    },
+    {
+      "epoch": 3.3110531463971307,
+      "grad_norm": 0.1312306523323059,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 2031
+    },
+    {
+      "epoch": 3.3126834039778283,
+      "grad_norm": 0.17245103418827057,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 2032
+    },
+    {
+      "epoch": 3.3143136615585265,
+      "grad_norm": 0.16342630982398987,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 2033
+    },
+    {
+      "epoch": 3.315943919139224,
+      "grad_norm": 0.13496838510036469,
+      "learning_rate": 0.0001,
+      "loss": 0.1599,
+      "step": 2034
+    },
+    {
+      "epoch": 3.317574176719922,
+      "grad_norm": 0.1370801329612732,
+      "learning_rate": 0.0001,
+      "loss": 0.1556,
+      "step": 2035
+    },
+    {
+      "epoch": 3.3192044343006195,
+      "grad_norm": 0.12318190932273865,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 2036
+    },
+    {
+      "epoch": 3.320834691881317,
+      "grad_norm": 0.13986453413963318,
+      "learning_rate": 0.0001,
+      "loss": 0.1564,
+      "step": 2037
+    },
+    {
+      "epoch": 3.322464949462015,
+      "grad_norm": 0.14522337913513184,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 2038
+    },
+    {
+      "epoch": 3.3240952070427126,
+      "grad_norm": 0.1366148144006729,
+      "learning_rate": 0.0001,
+      "loss": 0.1391,
+      "step": 2039
+    },
+    {
+      "epoch": 3.3257254646234107,
+      "grad_norm": 0.1322958618402481,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 2040
+    },
+    {
+      "epoch": 3.3273557222041084,
+      "grad_norm": 0.1638648360967636,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 2041
+    },
+    {
+      "epoch": 3.328985979784806,
+      "grad_norm": 0.15232039988040924,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 2042
+    },
+    {
+      "epoch": 3.3306162373655037,
+      "grad_norm": 0.13481836020946503,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 2043
+    },
+    {
+      "epoch": 3.3322464949462014,
+      "grad_norm": 0.1377725750207901,
+      "learning_rate": 0.0001,
+      "loss": 0.1552,
+      "step": 2044
+    },
+    {
+      "epoch": 3.333876752526899,
+      "grad_norm": 0.1482841521501541,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 2045
+    },
+    {
+      "epoch": 3.335507010107597,
+      "grad_norm": 0.15259544551372528,
+      "learning_rate": 0.0001,
+      "loss": 0.1479,
+      "step": 2046
+    },
+    {
+      "epoch": 3.337137267688295,
+      "grad_norm": 0.14434483647346497,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 2047
+    },
+    {
+      "epoch": 3.3387675252689926,
+      "grad_norm": 0.14351776242256165,
+      "learning_rate": 0.0001,
+      "loss": 0.1605,
+      "step": 2048
+    },
+    {
+      "epoch": 3.3403977828496902,
+      "grad_norm": 0.1376352459192276,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 2049
+    },
+    {
+      "epoch": 3.342028040430388,
+      "grad_norm": 0.15512721240520477,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 2050
+    },
+    {
+      "epoch": 3.3436582980110856,
+      "grad_norm": 0.1421147584915161,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 2051
+    },
+    {
+      "epoch": 3.3452885555917833,
+      "grad_norm": 0.12368530035018921,
+      "learning_rate": 0.0001,
+      "loss": 0.1382,
+      "step": 2052
+    },
+    {
+      "epoch": 3.3469188131724814,
+      "grad_norm": 0.14277908205986023,
+      "learning_rate": 0.0001,
+      "loss": 0.1407,
+      "step": 2053
+    },
+    {
+      "epoch": 3.348549070753179,
+      "grad_norm": 0.15306757390499115,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 2054
+    },
+    {
+      "epoch": 3.3501793283338768,
+      "grad_norm": 0.28750717639923096,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 2055
+    },
+    {
+      "epoch": 3.3518095859145745,
+      "grad_norm": 0.33853110671043396,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 2056
+    },
+    {
+      "epoch": 3.353439843495272,
+      "grad_norm": 0.1322954297065735,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 2057
+    },
+    {
+      "epoch": 3.35507010107597,
+      "grad_norm": 0.15321014821529388,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 2058
+    },
+    {
+      "epoch": 3.356700358656668,
+      "grad_norm": 0.15461274981498718,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 2059
+    },
+    {
+      "epoch": 3.3583306162373656,
+      "grad_norm": 0.14355158805847168,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 2060
+    },
+    {
+      "epoch": 3.3599608738180633,
+      "grad_norm": 0.14021973311901093,
+      "learning_rate": 0.0001,
+      "loss": 0.1393,
+      "step": 2061
+    },
+    {
+      "epoch": 3.361591131398761,
+      "grad_norm": 0.13298651576042175,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 2062
+    },
+    {
+      "epoch": 3.3632213889794587,
+      "grad_norm": 0.12698578834533691,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 2063
+    },
+    {
+      "epoch": 3.3648516465601563,
+      "grad_norm": 0.13323427736759186,
+      "learning_rate": 0.0001,
+      "loss": 0.147,
+      "step": 2064
+    },
+    {
+      "epoch": 3.366481904140854,
+      "grad_norm": 0.14920176565647125,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 2065
+    },
+    {
+      "epoch": 3.368112161721552,
+      "grad_norm": 0.13470523059368134,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 2066
+    },
+    {
+      "epoch": 3.36974241930225,
+      "grad_norm": 0.13527804613113403,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 2067
+    },
+    {
+      "epoch": 3.3713726768829475,
+      "grad_norm": 0.14219944179058075,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 2068
+    },
+    {
+      "epoch": 3.373002934463645,
+      "grad_norm": 0.1605711281299591,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 2069
+    },
+    {
+      "epoch": 3.374633192044343,
+      "grad_norm": 0.12055234611034393,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 2070
+    },
+    {
+      "epoch": 3.376263449625041,
+      "grad_norm": 0.13583442568778992,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 2071
+    },
+    {
+      "epoch": 3.3778937072057387,
+      "grad_norm": 0.16419711709022522,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 2072
+    },
+    {
+      "epoch": 3.3795239647864364,
+      "grad_norm": 0.16373597085475922,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 2073
+    },
+    {
+      "epoch": 3.381154222367134,
+      "grad_norm": 0.15173253417015076,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 2074
+    },
+    {
+      "epoch": 3.3827844799478317,
+      "grad_norm": 0.13307532668113708,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2075
+    },
+    {
+      "epoch": 3.3844147375285294,
+      "grad_norm": 0.13389620184898376,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 2076
+    },
+    {
+      "epoch": 3.386044995109227,
+      "grad_norm": 0.1596551537513733,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 2077
+    },
+    {
+      "epoch": 3.387675252689925,
+      "grad_norm": 0.13653887808322906,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 2078
+    },
+    {
+      "epoch": 3.389305510270623,
+      "grad_norm": 0.1583099216222763,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 2079
+    },
+    {
+      "epoch": 3.3909357678513206,
+      "grad_norm": 0.1411380171775818,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 2080
+    },
+    {
+      "epoch": 3.3925660254320182,
+      "grad_norm": 0.1537819504737854,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 2081
+    },
+    {
+      "epoch": 3.394196283012716,
+      "grad_norm": 0.13339395821094513,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 2082
+    },
+    {
+      "epoch": 3.3958265405934136,
+      "grad_norm": 0.1422421634197235,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 2083
+    },
+    {
+      "epoch": 3.3974567981741117,
+      "grad_norm": 0.21194976568222046,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 2084
+    },
+    {
+      "epoch": 3.3990870557548094,
+      "grad_norm": 0.21863634884357452,
+      "learning_rate": 0.0001,
+      "loss": 0.1538,
+      "step": 2085
+    },
+    {
+      "epoch": 3.400717313335507,
+      "grad_norm": 0.13825783133506775,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2086
+    },
+    {
+      "epoch": 3.4023475709162048,
+      "grad_norm": 0.154763326048851,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 2087
+    },
+    {
+      "epoch": 3.4039778284969024,
+      "grad_norm": 0.14275521039962769,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 2088
+    },
+    {
+      "epoch": 3.4056080860776,
+      "grad_norm": 0.14550544321537018,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 2089
+    },
+    {
+      "epoch": 3.407238343658298,
+      "grad_norm": 0.1512393355369568,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 2090
+    },
+    {
+      "epoch": 3.408868601238996,
+      "grad_norm": 0.14268140494823456,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 2091
+    },
+    {
+      "epoch": 3.4104988588196936,
+      "grad_norm": 0.14155443012714386,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 2092
+    },
+    {
+      "epoch": 3.4121291164003913,
+      "grad_norm": 0.1586080640554428,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 2093
+    },
+    {
+      "epoch": 3.413759373981089,
+      "grad_norm": 0.14126574993133545,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 2094
+    },
+    {
+      "epoch": 3.4153896315617867,
+      "grad_norm": 0.12679395079612732,
+      "learning_rate": 0.0001,
+      "loss": 0.148,
+      "step": 2095
+    },
+    {
+      "epoch": 3.4170198891424843,
+      "grad_norm": 0.17171818017959595,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 2096
+    },
+    {
+      "epoch": 3.4186501467231825,
+      "grad_norm": 0.1353333592414856,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 2097
+    },
+    {
+      "epoch": 3.42028040430388,
+      "grad_norm": 0.20119301974773407,
+      "learning_rate": 0.0001,
+      "loss": 0.1555,
+      "step": 2098
+    },
+    {
+      "epoch": 3.421910661884578,
+      "grad_norm": 0.1484251767396927,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 2099
+    },
+    {
+      "epoch": 3.4235409194652755,
+      "grad_norm": 0.1589980572462082,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 2100
+    },
+    {
+      "epoch": 3.425171177045973,
+      "grad_norm": 0.13816691935062408,
+      "learning_rate": 0.0001,
+      "loss": 0.1556,
+      "step": 2101
+    },
+    {
+      "epoch": 3.426801434626671,
+      "grad_norm": 0.12436775863170624,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 2102
+    },
+    {
+      "epoch": 3.4284316922073685,
+      "grad_norm": 0.12955281138420105,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 2103
+    },
+    {
+      "epoch": 3.4300619497880667,
+      "grad_norm": 0.17406727373600006,
+      "learning_rate": 0.0001,
+      "loss": 0.1765,
+      "step": 2104
+    },
+    {
+      "epoch": 3.4316922073687643,
+      "grad_norm": 0.17298480868339539,
+      "learning_rate": 0.0001,
+      "loss": 0.1766,
+      "step": 2105
+    },
+    {
+      "epoch": 3.433322464949462,
+      "grad_norm": 0.15015456080436707,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 2106
+    },
+    {
+      "epoch": 3.4349527225301597,
+      "grad_norm": 0.13401339948177338,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 2107
+    },
+    {
+      "epoch": 3.4365829801108574,
+      "grad_norm": 0.15644237399101257,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 2108
+    },
+    {
+      "epoch": 3.4382132376915555,
+      "grad_norm": 0.1544109582901001,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 2109
+    },
+    {
+      "epoch": 3.439843495272253,
+      "grad_norm": 0.1289900243282318,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2110
+    },
+    {
+      "epoch": 3.441473752852951,
+      "grad_norm": 0.12115547060966492,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 2111
+    },
+    {
+      "epoch": 3.4431040104336486,
+      "grad_norm": 0.16109564900398254,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 2112
+    },
+    {
+      "epoch": 3.4447342680143462,
+      "grad_norm": 0.12790006399154663,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 2113
+    },
+    {
+      "epoch": 3.446364525595044,
+      "grad_norm": 0.12892436981201172,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 2114
+    },
+    {
+      "epoch": 3.4479947831757416,
+      "grad_norm": 0.1363893300294876,
+      "learning_rate": 0.0001,
+      "loss": 0.1535,
+      "step": 2115
+    },
+    {
+      "epoch": 3.4496250407564393,
+      "grad_norm": 0.1328737437725067,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 2116
+    },
+    {
+      "epoch": 3.4512552983371374,
+      "grad_norm": 0.14212438464164734,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2117
+    },
+    {
+      "epoch": 3.452885555917835,
+      "grad_norm": 0.16319912672042847,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 2118
+    },
+    {
+      "epoch": 3.4545158134985328,
+      "grad_norm": 0.14669449627399445,
+      "learning_rate": 0.0001,
+      "loss": 0.1726,
+      "step": 2119
+    },
+    {
+      "epoch": 3.4561460710792304,
+      "grad_norm": 0.12865161895751953,
+      "learning_rate": 0.0001,
+      "loss": 0.1431,
+      "step": 2120
+    },
+    {
+      "epoch": 3.457776328659928,
+      "grad_norm": 0.14727066457271576,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 2121
+    },
+    {
+      "epoch": 3.4594065862406262,
+      "grad_norm": 0.12576113641262054,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 2122
+    },
+    {
+      "epoch": 3.461036843821324,
+      "grad_norm": 0.14870908856391907,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 2123
+    },
+    {
+      "epoch": 3.4626671014020216,
+      "grad_norm": 0.14071956276893616,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 2124
+    },
+    {
+      "epoch": 3.4642973589827193,
+      "grad_norm": 0.13211530447006226,
+      "learning_rate": 0.0001,
+      "loss": 0.1549,
+      "step": 2125
+    },
+    {
+      "epoch": 3.465927616563417,
+      "grad_norm": 0.14687129855155945,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 2126
+    },
+    {
+      "epoch": 3.4675578741441146,
+      "grad_norm": 0.1233501061797142,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 2127
+    },
+    {
+      "epoch": 3.4691881317248123,
+      "grad_norm": 0.1451001763343811,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 2128
+    },
+    {
+      "epoch": 3.4708183893055105,
+      "grad_norm": 0.14870783686637878,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 2129
+    },
+    {
+      "epoch": 3.472448646886208,
+      "grad_norm": 0.1319887787103653,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 2130
+    },
+    {
+      "epoch": 3.474078904466906,
+      "grad_norm": 0.12232866883277893,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2131
+    },
+    {
+      "epoch": 3.4757091620476035,
+      "grad_norm": 0.12974146008491516,
+      "learning_rate": 0.0001,
+      "loss": 0.1554,
+      "step": 2132
+    },
+    {
+      "epoch": 3.477339419628301,
+      "grad_norm": 0.14855380356311798,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 2133
+    },
+    {
+      "epoch": 3.478969677208999,
+      "grad_norm": 0.15491904318332672,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 2134
+    },
+    {
+      "epoch": 3.480599934789697,
+      "grad_norm": 0.12838788330554962,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 2135
+    },
+    {
+      "epoch": 3.4822301923703947,
+      "grad_norm": 0.13425928354263306,
+      "learning_rate": 0.0001,
+      "loss": 0.1462,
+      "step": 2136
+    },
+    {
+      "epoch": 3.4838604499510923,
+      "grad_norm": 0.1410258710384369,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "step": 2137
+    },
+    {
+      "epoch": 3.48549070753179,
+      "grad_norm": 0.1354324221611023,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 2138
+    },
+    {
+      "epoch": 3.4871209651124877,
+      "grad_norm": 0.13679568469524384,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 2139
+    },
+    {
+      "epoch": 3.4887512226931854,
+      "grad_norm": 0.1500730812549591,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 2140
+    },
+    {
+      "epoch": 3.490381480273883,
+      "grad_norm": 0.13870522379875183,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2141
+    },
+    {
+      "epoch": 3.492011737854581,
+      "grad_norm": 0.13639062643051147,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 2142
+    },
+    {
+      "epoch": 3.493641995435279,
+      "grad_norm": 0.16066312789916992,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 2143
+    },
+    {
+      "epoch": 3.4952722530159765,
+      "grad_norm": 0.12547770142555237,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 2144
+    },
+    {
+      "epoch": 3.4969025105966742,
+      "grad_norm": 0.14830149710178375,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 2145
+    },
+    {
+      "epoch": 3.498532768177372,
+      "grad_norm": 0.13417202234268188,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2146
+    },
+    {
+      "epoch": 3.50016302575807,
+      "grad_norm": 0.12787555158138275,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 2147
+    },
+    {
+      "epoch": 3.5017932833387677,
+      "grad_norm": 0.1516534686088562,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 2148
+    },
+    {
+      "epoch": 3.5034235409194654,
+      "grad_norm": 0.11404507607221603,
+      "learning_rate": 0.0001,
+      "loss": 0.1433,
+      "step": 2149
+    },
+    {
+      "epoch": 3.505053798500163,
+      "grad_norm": 0.1648181825876236,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 2150
+    },
+    {
+      "epoch": 3.5066840560808608,
+      "grad_norm": 0.15040086209774017,
+      "learning_rate": 0.0001,
+      "loss": 0.1555,
+      "step": 2151
+    },
+    {
+      "epoch": 3.5083143136615584,
+      "grad_norm": 0.17668725550174713,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 2152
+    },
+    {
+      "epoch": 3.509944571242256,
+      "grad_norm": 0.13613073527812958,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2153
+    },
+    {
+      "epoch": 3.511574828822954,
+      "grad_norm": 0.18130412697792053,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 2154
+    },
+    {
+      "epoch": 3.513205086403652,
+      "grad_norm": 0.1680803745985031,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 2155
+    },
+    {
+      "epoch": 3.5148353439843496,
+      "grad_norm": 0.13636229932308197,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 2156
+    },
+    {
+      "epoch": 3.5164656015650473,
+      "grad_norm": 0.16443794965744019,
+      "learning_rate": 0.0001,
+      "loss": 0.1668,
+      "step": 2157
+    },
+    {
+      "epoch": 3.518095859145745,
+      "grad_norm": 0.15313531458377838,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 2158
+    },
+    {
+      "epoch": 3.5197261167264426,
+      "grad_norm": 0.12920008599758148,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 2159
+    },
+    {
+      "epoch": 3.5213563743071408,
+      "grad_norm": 0.14635585248470306,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 2160
+    },
+    {
+      "epoch": 3.5229866318878384,
+      "grad_norm": 0.15921540558338165,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 2161
+    },
+    {
+      "epoch": 3.524616889468536,
+      "grad_norm": 0.14027996361255646,
+      "learning_rate": 0.0001,
+      "loss": 0.1637,
+      "step": 2162
+    },
+    {
+      "epoch": 3.526247147049234,
+      "grad_norm": 0.17607349157333374,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 2163
+    },
+    {
+      "epoch": 3.5278774046299315,
+      "grad_norm": 0.14340530335903168,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 2164
+    },
+    {
+      "epoch": 3.529507662210629,
+      "grad_norm": 0.13496717810630798,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 2165
+    },
+    {
+      "epoch": 3.531137919791327,
+      "grad_norm": 0.13388283550739288,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 2166
+    },
+    {
+      "epoch": 3.5327681773720245,
+      "grad_norm": 0.12602393329143524,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 2167
+    },
+    {
+      "epoch": 3.5343984349527227,
+      "grad_norm": 0.13006533682346344,
+      "learning_rate": 0.0001,
+      "loss": 0.1483,
+      "step": 2168
+    },
+    {
+      "epoch": 3.5360286925334203,
+      "grad_norm": 0.14154121279716492,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 2169
+    },
+    {
+      "epoch": 3.537658950114118,
+      "grad_norm": 0.16337460279464722,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 2170
+    },
+    {
+      "epoch": 3.5392892076948157,
+      "grad_norm": 0.13401515781879425,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 2171
+    },
+    {
+      "epoch": 3.5409194652755134,
+      "grad_norm": 0.16181372106075287,
+      "learning_rate": 0.0001,
+      "loss": 0.1657,
+      "step": 2172
+    },
+    {
+      "epoch": 3.5425497228562115,
+      "grad_norm": 0.14788036048412323,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 2173
+    },
+    {
+      "epoch": 3.544179980436909,
+      "grad_norm": 0.16251921653747559,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 2174
+    },
+    {
+      "epoch": 3.545810238017607,
+      "grad_norm": 0.15175074338912964,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 2175
+    },
+    {
+      "epoch": 3.5474404955983045,
+      "grad_norm": 0.1437595635652542,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 2176
+    },
+    {
+      "epoch": 3.5490707531790022,
+      "grad_norm": 0.13593018054962158,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 2177
+    },
+    {
+      "epoch": 3.5507010107597,
+      "grad_norm": 0.1520191729068756,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 2178
+    },
+    {
+      "epoch": 3.5523312683403976,
+      "grad_norm": 0.13462144136428833,
+      "learning_rate": 0.0001,
+      "loss": 0.1537,
+      "step": 2179
+    },
+    {
+      "epoch": 3.5539615259210953,
+      "grad_norm": 0.12073640525341034,
+      "learning_rate": 0.0001,
+      "loss": 0.1544,
+      "step": 2180
+    },
+    {
+      "epoch": 3.5555917835017934,
+      "grad_norm": 0.15898512303829193,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 2181
+    },
+    {
+      "epoch": 3.557222041082491,
+      "grad_norm": 0.15629935264587402,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 2182
+    },
+    {
+      "epoch": 3.5588522986631888,
+      "grad_norm": 0.12848952412605286,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 2183
+    },
+    {
+      "epoch": 3.5604825562438864,
+      "grad_norm": 0.13705319166183472,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 2184
+    },
+    {
+      "epoch": 3.5621128138245846,
+      "grad_norm": 0.1567762941122055,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 2185
+    },
+    {
+      "epoch": 3.5637430714052822,
+      "grad_norm": 0.13675236701965332,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 2186
+    },
+    {
+      "epoch": 3.56537332898598,
+      "grad_norm": 0.14318861067295074,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2187
+    },
+    {
+      "epoch": 3.5670035865666776,
+      "grad_norm": 0.16439467668533325,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 2188
+    },
+    {
+      "epoch": 3.5686338441473753,
+      "grad_norm": 0.15110914409160614,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 2189
+    },
+    {
+      "epoch": 3.570264101728073,
+      "grad_norm": 0.1775410771369934,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 2190
+    },
+    {
+      "epoch": 3.5718943593087706,
+      "grad_norm": 0.14808377623558044,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 2191
+    },
+    {
+      "epoch": 3.5735246168894683,
+      "grad_norm": 0.14409534633159637,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 2192
+    },
+    {
+      "epoch": 3.5751548744701664,
+      "grad_norm": 0.14742523431777954,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 2193
+    },
+    {
+      "epoch": 3.576785132050864,
+      "grad_norm": 0.15724779665470123,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 2194
+    },
+    {
+      "epoch": 3.578415389631562,
+      "grad_norm": 0.1302788108587265,
+      "learning_rate": 0.0001,
+      "loss": 0.1561,
+      "step": 2195
+    },
+    {
+      "epoch": 3.5800456472122595,
+      "grad_norm": 0.1408357322216034,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 2196
+    },
+    {
+      "epoch": 3.581675904792957,
+      "grad_norm": 0.13979199528694153,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 2197
+    },
+    {
+      "epoch": 3.5833061623736553,
+      "grad_norm": 0.12203294783830643,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 2198
+    },
+    {
+      "epoch": 3.584936419954353,
+      "grad_norm": 0.12141190469264984,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 2199
+    },
+    {
+      "epoch": 3.5865666775350507,
+      "grad_norm": 0.13076384365558624,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 2200
+    },
+    {
+      "epoch": 3.5881969351157483,
+      "grad_norm": 0.14992623031139374,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 2201
+    },
+    {
+      "epoch": 3.589827192696446,
+      "grad_norm": 0.1519753336906433,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 2202
+    },
+    {
+      "epoch": 3.5914574502771437,
+      "grad_norm": 0.1315367966890335,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2203
+    },
+    {
+      "epoch": 3.5930877078578414,
+      "grad_norm": 0.15619060397148132,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 2204
+    },
+    {
+      "epoch": 3.594717965438539,
+      "grad_norm": 0.13551609218120575,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 2205
+    },
+    {
+      "epoch": 3.596348223019237,
+      "grad_norm": 0.12931670248508453,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 2206
+    },
+    {
+      "epoch": 3.597978480599935,
+      "grad_norm": 0.1995784491300583,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 2207
+    },
+    {
+      "epoch": 3.5996087381806325,
+      "grad_norm": 0.1508382260799408,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2208
+    },
+    {
+      "epoch": 3.60123899576133,
+      "grad_norm": 0.15891651809215546,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 2209
+    },
+    {
+      "epoch": 3.602869253342028,
+      "grad_norm": 0.14910927414894104,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 2210
+    },
+    {
+      "epoch": 3.604499510922726,
+      "grad_norm": 0.1462751179933548,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 2211
+    },
+    {
+      "epoch": 3.6061297685034237,
+      "grad_norm": 0.1275903880596161,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 2212
+    },
+    {
+      "epoch": 3.6077600260841214,
+      "grad_norm": 0.13851912319660187,
+      "learning_rate": 0.0001,
+      "loss": 0.1464,
+      "step": 2213
+    },
+    {
+      "epoch": 3.609390283664819,
+      "grad_norm": 0.121219202876091,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 2214
+    },
+    {
+      "epoch": 3.6110205412455167,
+      "grad_norm": 0.14287428557872772,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 2215
+    },
+    {
+      "epoch": 3.6126507988262144,
+      "grad_norm": 0.14093217253684998,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 2216
+    },
+    {
+      "epoch": 3.614281056406912,
+      "grad_norm": 0.12914489209651947,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 2217
+    },
+    {
+      "epoch": 3.61591131398761,
+      "grad_norm": 0.14418743550777435,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 2218
+    },
+    {
+      "epoch": 3.617541571568308,
+      "grad_norm": 0.13824236392974854,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2219
+    },
+    {
+      "epoch": 3.6191718291490056,
+      "grad_norm": 0.14314696192741394,
+      "learning_rate": 0.0001,
+      "loss": 0.1695,
+      "step": 2220
+    },
+    {
+      "epoch": 3.6208020867297033,
+      "grad_norm": 0.14805221557617188,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 2221
+    },
+    {
+      "epoch": 3.622432344310401,
+      "grad_norm": 0.1371402144432068,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 2222
+    },
+    {
+      "epoch": 3.6240626018910986,
+      "grad_norm": 0.1396588832139969,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 2223
+    },
+    {
+      "epoch": 3.6256928594717968,
+      "grad_norm": 0.15313009917736053,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "step": 2224
+    },
+    {
+      "epoch": 3.6273231170524944,
+      "grad_norm": 0.12829270958900452,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 2225
+    },
+    {
+      "epoch": 3.628953374633192,
+      "grad_norm": 0.14311690628528595,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2226
+    },
+    {
+      "epoch": 3.63058363221389,
+      "grad_norm": 0.1597692370414734,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 2227
+    },
+    {
+      "epoch": 3.6322138897945875,
+      "grad_norm": 0.16670234501361847,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 2228
+    },
+    {
+      "epoch": 3.633844147375285,
+      "grad_norm": 0.13492721319198608,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2229
+    },
+    {
+      "epoch": 3.635474404955983,
+      "grad_norm": 0.17079739272594452,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 2230
+    },
+    {
+      "epoch": 3.6371046625366805,
+      "grad_norm": 0.1510683000087738,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 2231
+    },
+    {
+      "epoch": 3.6387349201173786,
+      "grad_norm": 0.16345694661140442,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 2232
+    },
+    {
+      "epoch": 3.6403651776980763,
+      "grad_norm": 0.16322366893291473,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 2233
+    },
+    {
+      "epoch": 3.641995435278774,
+      "grad_norm": 0.14221617579460144,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2234
+    },
+    {
+      "epoch": 3.6436256928594717,
+      "grad_norm": 0.1472931206226349,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 2235
+    },
+    {
+      "epoch": 3.64525595044017,
+      "grad_norm": 0.15251535177230835,
+      "learning_rate": 0.0001,
+      "loss": 0.1814,
+      "step": 2236
+    },
+    {
+      "epoch": 3.6468862080208675,
+      "grad_norm": 0.12713898718357086,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 2237
+    },
+    {
+      "epoch": 3.648516465601565,
+      "grad_norm": 0.13150040805339813,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 2238
+    },
+    {
+      "epoch": 3.650146723182263,
+      "grad_norm": 0.1565570831298828,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 2239
+    },
+    {
+      "epoch": 3.6517769807629605,
+      "grad_norm": 0.13739293813705444,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 2240
+    },
+    {
+      "epoch": 3.653407238343658,
+      "grad_norm": 0.1197710707783699,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 2241
+    },
+    {
+      "epoch": 3.655037495924356,
+      "grad_norm": 0.1372688263654709,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 2242
+    },
+    {
+      "epoch": 3.6566677535050536,
+      "grad_norm": 0.14185823500156403,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 2243
+    },
+    {
+      "epoch": 3.6582980110857517,
+      "grad_norm": 0.18424132466316223,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 2244
+    },
+    {
+      "epoch": 3.6599282686664494,
+      "grad_norm": 0.15011794865131378,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 2245
+    },
+    {
+      "epoch": 3.661558526247147,
+      "grad_norm": 0.15816602110862732,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 2246
+    },
+    {
+      "epoch": 3.6631887838278447,
+      "grad_norm": 0.14764584600925446,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 2247
+    },
+    {
+      "epoch": 3.6648190414085424,
+      "grad_norm": 0.1543309986591339,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 2248
+    },
+    {
+      "epoch": 3.6664492989892405,
+      "grad_norm": 0.16506819427013397,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 2249
+    },
+    {
+      "epoch": 3.6680795565699382,
+      "grad_norm": 0.18144920468330383,
+      "learning_rate": 0.0001,
+      "loss": 0.1795,
+      "step": 2250
+    },
+    {
+      "epoch": 3.669709814150636,
+      "grad_norm": 0.1368333399295807,
+      "learning_rate": 0.0001,
+      "loss": 0.1537,
+      "step": 2251
+    },
+    {
+      "epoch": 3.6713400717313336,
+      "grad_norm": 0.14611852169036865,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 2252
+    },
+    {
+      "epoch": 3.6729703293120313,
+      "grad_norm": 0.14789550006389618,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 2253
+    },
+    {
+      "epoch": 3.674600586892729,
+      "grad_norm": 0.14431506395339966,
+      "learning_rate": 0.0001,
+      "loss": 0.1535,
+      "step": 2254
+    },
+    {
+      "epoch": 3.6762308444734266,
+      "grad_norm": 0.14077717065811157,
+      "learning_rate": 0.0001,
+      "loss": 0.1596,
+      "step": 2255
+    },
+    {
+      "epoch": 3.6778611020541243,
+      "grad_norm": 0.1420920193195343,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2256
+    },
+    {
+      "epoch": 3.6794913596348224,
+      "grad_norm": 0.13903354108333588,
+      "learning_rate": 0.0001,
+      "loss": 0.157,
+      "step": 2257
+    },
+    {
+      "epoch": 3.68112161721552,
+      "grad_norm": 0.12706349790096283,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 2258
+    },
+    {
+      "epoch": 3.682751874796218,
+      "grad_norm": 0.134929820895195,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2259
+    },
+    {
+      "epoch": 3.6843821323769155,
+      "grad_norm": 0.13152171671390533,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 2260
+    },
+    {
+      "epoch": 3.686012389957613,
+      "grad_norm": 0.12819035351276398,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 2261
+    },
+    {
+      "epoch": 3.6876426475383113,
+      "grad_norm": 0.14969497919082642,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 2262
+    },
+    {
+      "epoch": 3.689272905119009,
+      "grad_norm": 0.1479291170835495,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 2263
+    },
+    {
+      "epoch": 3.6909031626997066,
+      "grad_norm": 0.13185615837574005,
+      "learning_rate": 0.0001,
+      "loss": 0.145,
+      "step": 2264
+    },
+    {
+      "epoch": 3.6925334202804043,
+      "grad_norm": 0.1268220990896225,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 2265
+    },
+    {
+      "epoch": 3.694163677861102,
+      "grad_norm": 0.15159299969673157,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 2266
+    },
+    {
+      "epoch": 3.6957939354417997,
+      "grad_norm": 0.16078874468803406,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 2267
+    },
+    {
+      "epoch": 3.6974241930224974,
+      "grad_norm": 0.13626697659492493,
+      "learning_rate": 0.0001,
+      "loss": 0.1447,
+      "step": 2268
+    },
+    {
+      "epoch": 3.699054450603195,
+      "grad_norm": 0.14417946338653564,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 2269
+    },
+    {
+      "epoch": 3.700684708183893,
+      "grad_norm": 0.13131751120090485,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 2270
+    },
+    {
+      "epoch": 3.702314965764591,
+      "grad_norm": 0.142496719956398,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 2271
+    },
+    {
+      "epoch": 3.7039452233452885,
+      "grad_norm": 0.14883071184158325,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2272
+    },
+    {
+      "epoch": 3.705575480925986,
+      "grad_norm": 0.1475469022989273,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2273
+    },
+    {
+      "epoch": 3.707205738506684,
+      "grad_norm": 0.14427977800369263,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 2274
+    },
+    {
+      "epoch": 3.708835996087382,
+      "grad_norm": 0.13665615022182465,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 2275
+    },
+    {
+      "epoch": 3.7104662536680797,
+      "grad_norm": 0.1295589655637741,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 2276
+    },
+    {
+      "epoch": 3.7120965112487774,
+      "grad_norm": 0.139891117811203,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 2277
+    },
+    {
+      "epoch": 3.713726768829475,
+      "grad_norm": 0.14485996961593628,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 2278
+    },
+    {
+      "epoch": 3.7153570264101727,
+      "grad_norm": 0.13603661954402924,
+      "learning_rate": 0.0001,
+      "loss": 0.1559,
+      "step": 2279
+    },
+    {
+      "epoch": 3.7169872839908704,
+      "grad_norm": 0.13382630050182343,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 2280
+    },
+    {
+      "epoch": 3.718617541571568,
+      "grad_norm": 0.12754695117473602,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 2281
+    },
+    {
+      "epoch": 3.720247799152266,
+      "grad_norm": 0.13873539865016937,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2282
+    },
+    {
+      "epoch": 3.721878056732964,
+      "grad_norm": 0.1300293654203415,
+      "learning_rate": 0.0001,
+      "loss": 0.1596,
+      "step": 2283
+    },
+    {
+      "epoch": 3.7235083143136616,
+      "grad_norm": 0.1459476351737976,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 2284
+    },
+    {
+      "epoch": 3.7251385718943593,
+      "grad_norm": 0.12731602787971497,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 2285
+    },
+    {
+      "epoch": 3.726768829475057,
+      "grad_norm": 0.15073290467262268,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 2286
+    },
+    {
+      "epoch": 3.728399087055755,
+      "grad_norm": 0.13490375876426697,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 2287
+    },
+    {
+      "epoch": 3.7300293446364527,
+      "grad_norm": 0.15296167135238647,
+      "learning_rate": 0.0001,
+      "loss": 0.1528,
+      "step": 2288
+    },
+    {
+      "epoch": 3.7316596022171504,
+      "grad_norm": 0.18614895641803741,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 2289
+    },
+    {
+      "epoch": 3.733289859797848,
+      "grad_norm": 0.14630430936813354,
+      "learning_rate": 0.0001,
+      "loss": 0.1499,
+      "step": 2290
+    },
+    {
+      "epoch": 3.734920117378546,
+      "grad_norm": 0.13443906605243683,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 2291
+    },
+    {
+      "epoch": 3.7365503749592435,
+      "grad_norm": 0.14118340611457825,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 2292
+    },
+    {
+      "epoch": 3.738180632539941,
+      "grad_norm": 0.13624626398086548,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 2293
+    },
+    {
+      "epoch": 3.739810890120639,
+      "grad_norm": 0.13312043249607086,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 2294
+    },
+    {
+      "epoch": 3.741441147701337,
+      "grad_norm": 0.13018035888671875,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 2295
+    },
+    {
+      "epoch": 3.7430714052820346,
+      "grad_norm": 0.178181990981102,
+      "learning_rate": 0.0001,
+      "loss": 0.1538,
+      "step": 2296
+    },
+    {
+      "epoch": 3.7447016628627323,
+      "grad_norm": 0.13893936574459076,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 2297
+    },
+    {
+      "epoch": 3.74633192044343,
+      "grad_norm": 0.13853946328163147,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2298
+    },
+    {
+      "epoch": 3.7479621780241277,
+      "grad_norm": 0.13181748986244202,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 2299
+    },
+    {
+      "epoch": 3.749592435604826,
+      "grad_norm": 0.16607768833637238,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 2300
+    },
+    {
+      "epoch": 3.7512226931855235,
+      "grad_norm": 0.1501704752445221,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 2301
+    },
+    {
+      "epoch": 3.752852950766221,
+      "grad_norm": 0.15002481639385223,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 2302
+    },
+    {
+      "epoch": 3.754483208346919,
+      "grad_norm": 0.14106477797031403,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 2303
+    },
+    {
+      "epoch": 3.7561134659276165,
+      "grad_norm": 0.16274330019950867,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2304
+    },
+    {
+      "epoch": 3.757743723508314,
+      "grad_norm": 0.15609531104564667,
+      "learning_rate": 0.0001,
+      "loss": 0.151,
+      "step": 2305
+    },
+    {
+      "epoch": 3.759373981089012,
+      "grad_norm": 0.1472497135400772,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 2306
+    },
+    {
+      "epoch": 3.7610042386697096,
+      "grad_norm": 0.1381707787513733,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 2307
+    },
+    {
+      "epoch": 3.7626344962504077,
+      "grad_norm": 0.14111392199993134,
+      "learning_rate": 0.0001,
+      "loss": 0.1484,
+      "step": 2308
+    },
+    {
+      "epoch": 3.7642647538311054,
+      "grad_norm": 0.14470559358596802,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 2309
+    },
+    {
+      "epoch": 3.765895011411803,
+      "grad_norm": 0.13996532559394836,
+      "learning_rate": 0.0001,
+      "loss": 0.1552,
+      "step": 2310
+    },
+    {
+      "epoch": 3.7675252689925007,
+      "grad_norm": 0.14922195672988892,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 2311
+    },
+    {
+      "epoch": 3.7691555265731984,
+      "grad_norm": 0.15004834532737732,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 2312
+    },
+    {
+      "epoch": 3.7707857841538965,
+      "grad_norm": 0.13584741950035095,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 2313
+    },
+    {
+      "epoch": 3.772416041734594,
+      "grad_norm": 0.14363713562488556,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2314
+    },
+    {
+      "epoch": 3.774046299315292,
+      "grad_norm": 0.24087132513523102,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 2315
+    },
+    {
+      "epoch": 3.7756765568959896,
+      "grad_norm": 0.15411627292633057,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 2316
+    },
+    {
+      "epoch": 3.7773068144766873,
+      "grad_norm": 0.12080325186252594,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 2317
+    },
+    {
+      "epoch": 3.778937072057385,
+      "grad_norm": 0.14731939136981964,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 2318
+    },
+    {
+      "epoch": 3.7805673296380826,
+      "grad_norm": 0.13772052526474,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 2319
+    },
+    {
+      "epoch": 3.7821975872187803,
+      "grad_norm": 0.14156171679496765,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 2320
+    },
+    {
+      "epoch": 3.7838278447994784,
+      "grad_norm": 0.14452658593654633,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 2321
+    },
+    {
+      "epoch": 3.785458102380176,
+      "grad_norm": 0.12419889867305756,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 2322
+    },
+    {
+      "epoch": 3.787088359960874,
+      "grad_norm": 0.11582159996032715,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 2323
+    },
+    {
+      "epoch": 3.7887186175415715,
+      "grad_norm": 0.13737718760967255,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 2324
+    },
+    {
+      "epoch": 3.7903488751222696,
+      "grad_norm": 0.1628282368183136,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 2325
+    },
+    {
+      "epoch": 3.7919791327029673,
+      "grad_norm": 0.13625358045101166,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 2326
+    },
+    {
+      "epoch": 3.793609390283665,
+      "grad_norm": 0.1334877908229828,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 2327
+    },
+    {
+      "epoch": 3.7952396478643626,
+      "grad_norm": 0.13337381184101105,
+      "learning_rate": 0.0001,
+      "loss": 0.153,
+      "step": 2328
+    },
+    {
+      "epoch": 3.7968699054450603,
+      "grad_norm": 0.1452982872724533,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 2329
+    },
+    {
+      "epoch": 3.798500163025758,
+      "grad_norm": 0.14199711382389069,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 2330
+    },
+    {
+      "epoch": 3.8001304206064557,
+      "grad_norm": 0.14179185032844543,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 2331
+    },
+    {
+      "epoch": 3.8017606781871534,
+      "grad_norm": 0.14089013636112213,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 2332
+    },
+    {
+      "epoch": 3.8033909357678515,
+      "grad_norm": 0.1408660113811493,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 2333
+    },
+    {
+      "epoch": 3.805021193348549,
+      "grad_norm": 0.21150575578212738,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 2334
+    },
+    {
+      "epoch": 3.806651450929247,
+      "grad_norm": 0.1231960654258728,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 2335
+    },
+    {
+      "epoch": 3.8082817085099445,
+      "grad_norm": 0.16461803019046783,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 2336
+    },
+    {
+      "epoch": 3.809911966090642,
+      "grad_norm": 0.14631450176239014,
+      "learning_rate": 0.0001,
+      "loss": 0.1788,
+      "step": 2337
+    },
+    {
+      "epoch": 3.8115422236713403,
+      "grad_norm": 0.1487826257944107,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2338
+    },
+    {
+      "epoch": 3.813172481252038,
+      "grad_norm": 0.13970328867435455,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2339
+    },
+    {
+      "epoch": 3.8148027388327357,
+      "grad_norm": 0.13937072455883026,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 2340
+    },
+    {
+      "epoch": 3.8164329964134334,
+      "grad_norm": 0.13930930197238922,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 2341
+    },
+    {
+      "epoch": 3.818063253994131,
+      "grad_norm": 0.13589458167552948,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 2342
+    },
+    {
+      "epoch": 3.8196935115748287,
+      "grad_norm": 0.1309502124786377,
+      "learning_rate": 0.0001,
+      "loss": 0.1772,
+      "step": 2343
+    },
+    {
+      "epoch": 3.8213237691555264,
+      "grad_norm": 0.13713008165359497,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 2344
+    },
+    {
+      "epoch": 3.822954026736224,
+      "grad_norm": 0.1149279922246933,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 2345
+    },
+    {
+      "epoch": 3.824584284316922,
+      "grad_norm": 0.14622916281223297,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 2346
+    },
+    {
+      "epoch": 3.82621454189762,
+      "grad_norm": 0.1591426283121109,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 2347
+    },
+    {
+      "epoch": 3.8278447994783176,
+      "grad_norm": 0.15121379494667053,
+      "learning_rate": 0.0001,
+      "loss": 0.1487,
+      "step": 2348
+    },
+    {
+      "epoch": 3.8294750570590153,
+      "grad_norm": 0.14661197364330292,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 2349
+    },
+    {
+      "epoch": 3.831105314639713,
+      "grad_norm": 0.21391494572162628,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 2350
+    },
+    {
+      "epoch": 3.832735572220411,
+      "grad_norm": 0.14501704275608063,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 2351
+    },
+    {
+      "epoch": 3.8343658298011087,
+      "grad_norm": 0.17319004237651825,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 2352
+    },
+    {
+      "epoch": 3.8359960873818064,
+      "grad_norm": 0.15633517503738403,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 2353
+    },
+    {
+      "epoch": 3.837626344962504,
+      "grad_norm": 0.15051516890525818,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 2354
+    },
+    {
+      "epoch": 3.839256602543202,
+      "grad_norm": 0.1435638666152954,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 2355
+    },
+    {
+      "epoch": 3.8408868601238995,
+      "grad_norm": 0.14581768214702606,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2356
+    },
+    {
+      "epoch": 3.842517117704597,
+      "grad_norm": 0.1498011201620102,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 2357
+    },
+    {
+      "epoch": 3.844147375285295,
+      "grad_norm": 0.14038564264774323,
+      "learning_rate": 0.0001,
+      "loss": 0.1378,
+      "step": 2358
+    },
+    {
+      "epoch": 3.845777632865993,
+      "grad_norm": 0.14458616077899933,
+      "learning_rate": 0.0001,
+      "loss": 0.1535,
+      "step": 2359
+    },
+    {
+      "epoch": 3.8474078904466906,
+      "grad_norm": 0.17851635813713074,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2360
+    },
+    {
+      "epoch": 3.8490381480273883,
+      "grad_norm": 0.12941935658454895,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 2361
+    },
+    {
+      "epoch": 3.850668405608086,
+      "grad_norm": 0.1391555517911911,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 2362
+    },
+    {
+      "epoch": 3.8522986631887837,
+      "grad_norm": 0.13640303909778595,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 2363
+    },
+    {
+      "epoch": 3.853928920769482,
+      "grad_norm": 0.1430894136428833,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 2364
+    },
+    {
+      "epoch": 3.8555591783501795,
+      "grad_norm": 0.14540916681289673,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 2365
+    },
+    {
+      "epoch": 3.857189435930877,
+      "grad_norm": 0.1320265531539917,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 2366
+    },
+    {
+      "epoch": 3.858819693511575,
+      "grad_norm": 0.12641210854053497,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 2367
+    },
+    {
+      "epoch": 3.8604499510922725,
+      "grad_norm": 0.1293664276599884,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 2368
+    },
+    {
+      "epoch": 3.86208020867297,
+      "grad_norm": 0.13504886627197266,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 2369
+    },
+    {
+      "epoch": 3.863710466253668,
+      "grad_norm": 0.13120906054973602,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2370
+    },
+    {
+      "epoch": 3.8653407238343656,
+      "grad_norm": 0.1444963663816452,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 2371
+    },
+    {
+      "epoch": 3.8669709814150637,
+      "grad_norm": 0.13688752055168152,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 2372
+    },
+    {
+      "epoch": 3.8686012389957614,
+      "grad_norm": 0.13113614916801453,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 2373
+    },
+    {
+      "epoch": 3.870231496576459,
+      "grad_norm": 0.1431058645248413,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 2374
+    },
+    {
+      "epoch": 3.8718617541571567,
+      "grad_norm": 0.16322541236877441,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 2375
+    },
+    {
+      "epoch": 3.873492011737855,
+      "grad_norm": 0.15727584064006805,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 2376
+    },
+    {
+      "epoch": 3.8751222693185525,
+      "grad_norm": 0.13210931420326233,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 2377
+    },
+    {
+      "epoch": 3.87675252689925,
+      "grad_norm": 0.12558962404727936,
+      "learning_rate": 0.0001,
+      "loss": 0.1461,
+      "step": 2378
+    },
+    {
+      "epoch": 3.878382784479948,
+      "grad_norm": 0.14045056700706482,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 2379
+    },
+    {
+      "epoch": 3.8800130420606456,
+      "grad_norm": 0.1418190896511078,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 2380
+    },
+    {
+      "epoch": 3.8816432996413432,
+      "grad_norm": 0.12244697660207748,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 2381
+    },
+    {
+      "epoch": 3.883273557222041,
+      "grad_norm": 0.164667546749115,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 2382
+    },
+    {
+      "epoch": 3.8849038148027386,
+      "grad_norm": 0.15115118026733398,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 2383
+    },
+    {
+      "epoch": 3.8865340723834367,
+      "grad_norm": 0.13539472222328186,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2384
+    },
+    {
+      "epoch": 3.8881643299641344,
+      "grad_norm": 0.13099908828735352,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 2385
+    },
+    {
+      "epoch": 3.889794587544832,
+      "grad_norm": 0.12721744179725647,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 2386
+    },
+    {
+      "epoch": 3.8914248451255298,
+      "grad_norm": 0.1410721391439438,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2387
+    },
+    {
+      "epoch": 3.8930551027062275,
+      "grad_norm": 0.1388472616672516,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 2388
+    },
+    {
+      "epoch": 3.8946853602869256,
+      "grad_norm": 0.157010018825531,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 2389
+    },
+    {
+      "epoch": 3.8963156178676233,
+      "grad_norm": 0.14445838332176208,
+      "learning_rate": 0.0001,
+      "loss": 0.153,
+      "step": 2390
+    },
+    {
+      "epoch": 3.897945875448321,
+      "grad_norm": 0.14753592014312744,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 2391
+    },
+    {
+      "epoch": 3.8995761330290186,
+      "grad_norm": 0.13478566706180573,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 2392
+    },
+    {
+      "epoch": 3.9012063906097163,
+      "grad_norm": 0.13900446891784668,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 2393
+    },
+    {
+      "epoch": 3.902836648190414,
+      "grad_norm": 0.13876253366470337,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 2394
+    },
+    {
+      "epoch": 3.9044669057711117,
+      "grad_norm": 0.14419294893741608,
+      "learning_rate": 0.0001,
+      "loss": 0.1593,
+      "step": 2395
+    },
+    {
+      "epoch": 3.9060971633518093,
+      "grad_norm": 0.14167584478855133,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 2396
+    },
+    {
+      "epoch": 3.9077274209325075,
+      "grad_norm": 0.12826170027256012,
+      "learning_rate": 0.0001,
+      "loss": 0.147,
+      "step": 2397
+    },
+    {
+      "epoch": 3.909357678513205,
+      "grad_norm": 0.14393731951713562,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 2398
+    },
+    {
+      "epoch": 3.910987936093903,
+      "grad_norm": 0.15396815538406372,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 2399
+    },
+    {
+      "epoch": 3.9126181936746005,
+      "grad_norm": 0.11657989025115967,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 2400
+    },
+    {
+      "epoch": 3.914248451255298,
+      "grad_norm": 0.1413910686969757,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 2401
+    },
+    {
+      "epoch": 3.9158787088359963,
+      "grad_norm": 0.12488115578889847,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 2402
+    },
+    {
+      "epoch": 3.917508966416694,
+      "grad_norm": 0.1491551250219345,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 2403
+    },
+    {
+      "epoch": 3.9191392239973917,
+      "grad_norm": 0.13683412969112396,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 2404
+    },
+    {
+      "epoch": 3.9207694815780894,
+      "grad_norm": 0.15317296981811523,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 2405
+    },
+    {
+      "epoch": 3.922399739158787,
+      "grad_norm": 0.12439368665218353,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 2406
+    },
+    {
+      "epoch": 3.9240299967394847,
+      "grad_norm": 0.12986606359481812,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 2407
+    },
+    {
+      "epoch": 3.9256602543201824,
+      "grad_norm": 0.1569644659757614,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 2408
+    },
+    {
+      "epoch": 3.92729051190088,
+      "grad_norm": 0.14531882107257843,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 2409
+    },
+    {
+      "epoch": 3.928920769481578,
+      "grad_norm": 0.1490756869316101,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 2410
+    },
+    {
+      "epoch": 3.930551027062276,
+      "grad_norm": 0.11785148829221725,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 2411
+    },
+    {
+      "epoch": 3.9321812846429736,
+      "grad_norm": 0.13614097237586975,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 2412
+    },
+    {
+      "epoch": 3.9338115422236712,
+      "grad_norm": 0.12401864677667618,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 2413
+    },
+    {
+      "epoch": 3.9354417998043694,
+      "grad_norm": 0.12099269777536392,
+      "learning_rate": 0.0001,
+      "loss": 0.1305,
+      "step": 2414
+    },
+    {
+      "epoch": 3.937072057385067,
+      "grad_norm": 0.14183704555034637,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 2415
+    },
+    {
+      "epoch": 3.9387023149657647,
+      "grad_norm": 0.12612907588481903,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 2416
+    },
+    {
+      "epoch": 3.9403325725464624,
+      "grad_norm": 0.15672796964645386,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 2417
+    },
+    {
+      "epoch": 3.94196283012716,
+      "grad_norm": 0.1381552368402481,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 2418
+    },
+    {
+      "epoch": 3.9435930877078578,
+      "grad_norm": 0.2042001485824585,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 2419
+    },
+    {
+      "epoch": 3.9452233452885554,
+      "grad_norm": 0.13539600372314453,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 2420
+    },
+    {
+      "epoch": 3.946853602869253,
+      "grad_norm": 0.15428173542022705,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 2421
+    },
+    {
+      "epoch": 3.9484838604499513,
+      "grad_norm": 0.1534593552350998,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 2422
+    },
+    {
+      "epoch": 3.950114118030649,
+      "grad_norm": 0.15443755686283112,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 2423
+    },
+    {
+      "epoch": 3.9517443756113466,
+      "grad_norm": 0.1520153284072876,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 2424
+    },
+    {
+      "epoch": 3.9533746331920443,
+      "grad_norm": 0.14066120982170105,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 2425
+    },
+    {
+      "epoch": 3.955004890772742,
+      "grad_norm": 0.14006389677524567,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 2426
+    },
+    {
+      "epoch": 3.95663514835344,
+      "grad_norm": 0.1307193785905838,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 2427
+    },
+    {
+      "epoch": 3.958265405934138,
+      "grad_norm": 0.1285211592912674,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 2428
+    },
+    {
+      "epoch": 3.9598956635148355,
+      "grad_norm": 0.14015939831733704,
+      "learning_rate": 0.0001,
+      "loss": 0.1544,
+      "step": 2429
+    },
+    {
+      "epoch": 3.961525921095533,
+      "grad_norm": 0.13942451775074005,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2430
+    },
+    {
+      "epoch": 3.963156178676231,
+      "grad_norm": 0.12587985396385193,
+      "learning_rate": 0.0001,
+      "loss": 0.1578,
+      "step": 2431
+    },
+    {
+      "epoch": 3.9647864362569285,
+      "grad_norm": 0.14944875240325928,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 2432
+    },
+    {
+      "epoch": 3.966416693837626,
+      "grad_norm": 0.13582269847393036,
+      "learning_rate": 0.0001,
+      "loss": 0.1502,
+      "step": 2433
+    },
+    {
+      "epoch": 3.968046951418324,
+      "grad_norm": 0.15742120146751404,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 2434
+    },
+    {
+      "epoch": 3.969677208999022,
+      "grad_norm": 0.14265507459640503,
+      "learning_rate": 0.0001,
+      "loss": 0.1395,
+      "step": 2435
+    },
+    {
+      "epoch": 3.9713074665797197,
+      "grad_norm": 0.14941944181919098,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 2436
+    },
+    {
+      "epoch": 3.9729377241604173,
+      "grad_norm": 0.13401104509830475,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 2437
+    },
+    {
+      "epoch": 3.974567981741115,
+      "grad_norm": 0.13355430960655212,
+      "learning_rate": 0.0001,
+      "loss": 0.1732,
+      "step": 2438
+    },
+    {
+      "epoch": 3.9761982393218127,
+      "grad_norm": 0.14246457815170288,
+      "learning_rate": 0.0001,
+      "loss": 0.154,
+      "step": 2439
+    },
+    {
+      "epoch": 3.977828496902511,
+      "grad_norm": 0.1443282514810562,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 2440
+    },
+    {
+      "epoch": 3.9794587544832085,
+      "grad_norm": 0.13674692809581757,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 2441
+    },
+    {
+      "epoch": 3.981089012063906,
+      "grad_norm": 0.14663022756576538,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 2442
+    },
+    {
+      "epoch": 3.982719269644604,
+      "grad_norm": 0.23058007657527924,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 2443
+    },
+    {
+      "epoch": 3.9843495272253016,
+      "grad_norm": 0.12496799230575562,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 2444
+    },
+    {
+      "epoch": 3.9859797848059992,
+      "grad_norm": 0.14396335184574127,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 2445
+    },
+    {
+      "epoch": 3.987610042386697,
+      "grad_norm": 0.11941339075565338,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 2446
+    },
+    {
+      "epoch": 3.9892402999673946,
+      "grad_norm": 0.15236441791057587,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 2447
+    },
+    {
+      "epoch": 3.9908705575480927,
+      "grad_norm": 0.134572833776474,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 2448
+    },
+    {
+      "epoch": 3.9925008151287904,
+      "grad_norm": 0.1253572255373001,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 2449
+    },
+    {
+      "epoch": 3.994131072709488,
+      "grad_norm": 0.1377570629119873,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 2450
+    },
+    {
+      "epoch": 3.9957613302901858,
+      "grad_norm": 0.1263006627559662,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 2451
+    },
+    {
+      "epoch": 3.9973915878708834,
+      "grad_norm": 0.1317623406648636,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 2452
+    },
+    {
+      "epoch": 3.9973915878708834,
+      "step": 2452,
+      "total_flos": 5.923561418853974e+18,
+      "train_loss": 0.19042991894201586,
+      "train_runtime": 49765.6162,
+      "train_samples_per_second": 1.479,
+      "train_steps_per_second": 0.049
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2452,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "total_flos": 5.923561418853974e+18,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}