diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9597 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.99835255354201,
+  "eval_steps": 500,
+  "global_step": 1365,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002196595277320154,
+      "grad_norm": 0.25721433758735657,
+      "learning_rate": 2.18978102189781e-06,
+      "loss": 1.3285,
+      "step": 1
+    },
+    {
+      "epoch": 0.004393190554640308,
+      "grad_norm": 0.2432270050048828,
+      "learning_rate": 4.37956204379562e-06,
+      "loss": 1.5143,
+      "step": 2
+    },
+    {
+      "epoch": 0.006589785831960461,
+      "grad_norm": 0.21148736774921417,
+      "learning_rate": 6.56934306569343e-06,
+      "loss": 1.2825,
+      "step": 3
+    },
+    {
+      "epoch": 0.008786381109280615,
+      "grad_norm": 0.27486374974250793,
+      "learning_rate": 8.75912408759124e-06,
+      "loss": 1.2207,
+      "step": 4
+    },
+    {
+      "epoch": 0.010982976386600769,
+      "grad_norm": 0.23223614692687988,
+      "learning_rate": 1.0948905109489049e-05,
+      "loss": 1.2187,
+      "step": 5
+    },
+    {
+      "epoch": 0.013179571663920923,
+      "grad_norm": 0.23989129066467285,
+      "learning_rate": 1.313868613138686e-05,
+      "loss": 1.0866,
+      "step": 6
+    },
+    {
+      "epoch": 0.015376166941241077,
+      "grad_norm": 0.20565643906593323,
+      "learning_rate": 1.532846715328467e-05,
+      "loss": 1.0803,
+      "step": 7
+    },
+    {
+      "epoch": 0.01757276221856123,
+      "grad_norm": 0.21996630728244781,
+      "learning_rate": 1.751824817518248e-05,
+      "loss": 1.1396,
+      "step": 8
+    },
+    {
+      "epoch": 0.019769357495881382,
+      "grad_norm": 0.34225115180015564,
+      "learning_rate": 1.9708029197080292e-05,
+      "loss": 1.585,
+      "step": 9
+    },
+    {
+      "epoch": 0.021965952773201538,
+      "grad_norm": 0.44036951661109924,
+      "learning_rate": 2.1897810218978098e-05,
+      "loss": 1.1784,
+      "step": 10
+    },
+    {
+      "epoch": 0.02416254805052169,
+      "grad_norm": 0.29524368047714233,
+      "learning_rate": 2.408759124087591e-05,
+      "loss": 1.4552,
+      "step": 11
+    },
+    {
+      "epoch": 0.026359143327841845,
+      "grad_norm": 0.27995917201042175,
+      "learning_rate": 2.627737226277372e-05,
+      "loss": 1.2969,
+      "step": 12
+    },
+    {
+      "epoch": 0.028555738605161998,
+      "grad_norm": 0.27276578545570374,
+      "learning_rate": 2.846715328467153e-05,
+      "loss": 1.1856,
+      "step": 13
+    },
+    {
+      "epoch": 0.030752333882482153,
+      "grad_norm": 0.32685738801956177,
+      "learning_rate": 3.065693430656934e-05,
+      "loss": 1.2993,
+      "step": 14
+    },
+    {
+      "epoch": 0.032948929159802305,
+      "grad_norm": 0.29171568155288696,
+      "learning_rate": 3.284671532846715e-05,
+      "loss": 1.351,
+      "step": 15
+    },
+    {
+      "epoch": 0.03514552443712246,
+      "grad_norm": 0.28125178813934326,
+      "learning_rate": 3.503649635036496e-05,
+      "loss": 1.0589,
+      "step": 16
+    },
+    {
+      "epoch": 0.037342119714442616,
+      "grad_norm": 0.2893100380897522,
+      "learning_rate": 3.722627737226277e-05,
+      "loss": 1.1692,
+      "step": 17
+    },
+    {
+      "epoch": 0.039538714991762765,
+      "grad_norm": 0.4304523169994354,
+      "learning_rate": 3.9416058394160584e-05,
+      "loss": 1.3758,
+      "step": 18
+    },
+    {
+      "epoch": 0.04173531026908292,
+      "grad_norm": 0.3006685674190521,
+      "learning_rate": 4.160583941605839e-05,
+      "loss": 1.282,
+      "step": 19
+    },
+    {
+      "epoch": 0.043931905546403076,
+      "grad_norm": 0.372256875038147,
+      "learning_rate": 4.3795620437956196e-05,
+      "loss": 1.2514,
+      "step": 20
+    },
+    {
+      "epoch": 0.04612850082372323,
+      "grad_norm": 0.3712996542453766,
+      "learning_rate": 4.5985401459854016e-05,
+      "loss": 1.2874,
+      "step": 21
+    },
+    {
+      "epoch": 0.04832509610104338,
+      "grad_norm": 0.42726194858551025,
+      "learning_rate": 4.817518248175182e-05,
+      "loss": 1.1858,
+      "step": 22
+    },
+    {
+      "epoch": 0.050521691378363535,
+      "grad_norm": 0.8086119294166565,
+      "learning_rate": 5.036496350364963e-05,
+      "loss": 1.4406,
+      "step": 23
+    },
+    {
+      "epoch": 0.05271828665568369,
+      "grad_norm": 0.6575807332992554,
+      "learning_rate": 5.255474452554744e-05,
+      "loss": 1.2778,
+      "step": 24
+    },
+    {
+      "epoch": 0.054914881933003847,
+      "grad_norm": 0.4381495416164398,
+      "learning_rate": 5.4744525547445253e-05,
+      "loss": 1.171,
+      "step": 25
+    },
+    {
+      "epoch": 0.057111477210323995,
+      "grad_norm": 0.5334455966949463,
+      "learning_rate": 5.693430656934306e-05,
+      "loss": 1.1864,
+      "step": 26
+    },
+    {
+      "epoch": 0.05930807248764415,
+      "grad_norm": 0.6762930154800415,
+      "learning_rate": 5.912408759124087e-05,
+      "loss": 1.1345,
+      "step": 27
+    },
+    {
+      "epoch": 0.061504667764964306,
+      "grad_norm": 0.5146059393882751,
+      "learning_rate": 6.131386861313868e-05,
+      "loss": 1.0422,
+      "step": 28
+    },
+    {
+      "epoch": 0.06370126304228446,
+      "grad_norm": 0.530326247215271,
+      "learning_rate": 6.350364963503648e-05,
+      "loss": 1.122,
+      "step": 29
+    },
+    {
+      "epoch": 0.06589785831960461,
+      "grad_norm": 0.48408037424087524,
+      "learning_rate": 6.56934306569343e-05,
+      "loss": 1.0279,
+      "step": 30
+    },
+    {
+      "epoch": 0.06809445359692477,
+      "grad_norm": 0.9486483335494995,
+      "learning_rate": 6.788321167883211e-05,
+      "loss": 1.1292,
+      "step": 31
+    },
+    {
+      "epoch": 0.07029104887424492,
+      "grad_norm": 0.6662981510162354,
+      "learning_rate": 7.007299270072992e-05,
+      "loss": 1.2851,
+      "step": 32
+    },
+    {
+      "epoch": 0.07248764415156507,
+      "grad_norm": 0.5754452347755432,
+      "learning_rate": 7.226277372262774e-05,
+      "loss": 1.441,
+      "step": 33
+    },
+    {
+      "epoch": 0.07468423942888523,
+      "grad_norm": 0.6604092121124268,
+      "learning_rate": 7.445255474452554e-05,
+      "loss": 1.0769,
+      "step": 34
+    },
+    {
+      "epoch": 0.07688083470620538,
+      "grad_norm": 0.7060566544532776,
+      "learning_rate": 7.664233576642336e-05,
+      "loss": 1.1145,
+      "step": 35
+    },
+    {
+      "epoch": 0.07907742998352553,
+      "grad_norm": 2.7896018028259277,
+      "learning_rate": 7.883211678832117e-05,
+      "loss": 1.1459,
+      "step": 36
+    },
+    {
+      "epoch": 0.08127402526084569,
+      "grad_norm": 0.6884010434150696,
+      "learning_rate": 8.102189781021897e-05,
+      "loss": 1.1314,
+      "step": 37
+    },
+    {
+      "epoch": 0.08347062053816584,
+      "grad_norm": 1.0191112756729126,
+      "learning_rate": 8.321167883211678e-05,
+      "loss": 1.1096,
+      "step": 38
+    },
+    {
+      "epoch": 0.085667215815486,
+      "grad_norm": 0.8716099262237549,
+      "learning_rate": 8.540145985401459e-05,
+      "loss": 1.227,
+      "step": 39
+    },
+    {
+      "epoch": 0.08786381109280615,
+      "grad_norm": 1.468738317489624,
+      "learning_rate": 8.759124087591239e-05,
+      "loss": 0.9615,
+      "step": 40
+    },
+    {
+      "epoch": 0.0900604063701263,
+      "grad_norm": 0.7689814567565918,
+      "learning_rate": 8.978102189781021e-05,
+      "loss": 1.0236,
+      "step": 41
+    },
+    {
+      "epoch": 0.09225700164744646,
+      "grad_norm": 1.7151082754135132,
+      "learning_rate": 9.197080291970803e-05,
+      "loss": 1.2397,
+      "step": 42
+    },
+    {
+      "epoch": 0.09445359692476661,
+      "grad_norm": 0.8910958766937256,
+      "learning_rate": 9.416058394160584e-05,
+      "loss": 0.9306,
+      "step": 43
+    },
+    {
+      "epoch": 0.09665019220208676,
+      "grad_norm": 0.8792837858200073,
+      "learning_rate": 9.635036496350364e-05,
+      "loss": 1.0464,
+      "step": 44
+    },
+    {
+      "epoch": 0.09884678747940692,
+      "grad_norm": 0.9079362154006958,
+      "learning_rate": 9.854014598540145e-05,
+      "loss": 1.1637,
+      "step": 45
+    },
+    {
+      "epoch": 0.10104338275672707,
+      "grad_norm": 0.7605229616165161,
+      "learning_rate": 0.00010072992700729926,
+      "loss": 0.9376,
+      "step": 46
+    },
+    {
+      "epoch": 0.10323997803404723,
+      "grad_norm": 0.8633959889411926,
+      "learning_rate": 0.00010291970802919708,
+      "loss": 1.0007,
+      "step": 47
+    },
+    {
+      "epoch": 0.10543657331136738,
+      "grad_norm": 0.9112871885299683,
+      "learning_rate": 0.00010510948905109488,
+      "loss": 1.0897,
+      "step": 48
+    },
+    {
+      "epoch": 0.10763316858868753,
+      "grad_norm": 1.229214072227478,
+      "learning_rate": 0.00010729927007299269,
+      "loss": 1.2099,
+      "step": 49
+    },
+    {
+      "epoch": 0.10982976386600769,
+      "grad_norm": 1.0759787559509277,
+      "learning_rate": 0.00010948905109489051,
+      "loss": 0.9922,
+      "step": 50
+    },
+    {
+      "epoch": 0.11202635914332784,
+      "grad_norm": 1.2643060684204102,
+      "learning_rate": 0.00011167883211678831,
+      "loss": 0.9486,
+      "step": 51
+    },
+    {
+      "epoch": 0.11422295442064799,
+      "grad_norm": 1.4596999883651733,
+      "learning_rate": 0.00011386861313868612,
+      "loss": 0.9502,
+      "step": 52
+    },
+    {
+      "epoch": 0.11641954969796815,
+      "grad_norm": 1.1586005687713623,
+      "learning_rate": 0.00011605839416058394,
+      "loss": 1.07,
+      "step": 53
+    },
+    {
+      "epoch": 0.1186161449752883,
+      "grad_norm": 1.063158392906189,
+      "learning_rate": 0.00011824817518248174,
+      "loss": 0.7428,
+      "step": 54
+    },
+    {
+      "epoch": 0.12081274025260846,
+      "grad_norm": 0.9586182832717896,
+      "learning_rate": 0.00012043795620437955,
+      "loss": 0.9323,
+      "step": 55
+    },
+    {
+      "epoch": 0.12300933552992861,
+      "grad_norm": 1.569503903388977,
+      "learning_rate": 0.00012262773722627736,
+      "loss": 1.0579,
+      "step": 56
+    },
+    {
+      "epoch": 0.12520593080724876,
+      "grad_norm": 0.9976353049278259,
+      "learning_rate": 0.00012481751824817516,
+      "loss": 0.7751,
+      "step": 57
+    },
+    {
+      "epoch": 0.12740252608456892,
+      "grad_norm": 1.0565471649169922,
+      "learning_rate": 0.00012700729927007297,
+      "loss": 1.0423,
+      "step": 58
+    },
+    {
+      "epoch": 0.12959912136188906,
+      "grad_norm": 1.2650275230407715,
+      "learning_rate": 0.00012919708029197077,
+      "loss": 0.811,
+      "step": 59
+    },
+    {
+      "epoch": 0.13179571663920922,
+      "grad_norm": 0.6920269131660461,
+      "learning_rate": 0.0001313868613138686,
+      "loss": 0.8496,
+      "step": 60
+    },
+    {
+      "epoch": 0.13399231191652938,
+      "grad_norm": 1.215423345565796,
+      "learning_rate": 0.00013357664233576641,
+      "loss": 1.01,
+      "step": 61
+    },
+    {
+      "epoch": 0.13618890719384955,
+      "grad_norm": 2.6585795879364014,
+      "learning_rate": 0.00013576642335766422,
+      "loss": 0.9036,
+      "step": 62
+    },
+    {
+      "epoch": 0.13838550247116968,
+      "grad_norm": 1.269023060798645,
+      "learning_rate": 0.00013795620437956203,
+      "loss": 0.7665,
+      "step": 63
+    },
+    {
+      "epoch": 0.14058209774848984,
+      "grad_norm": 0.9672825932502747,
+      "learning_rate": 0.00014014598540145983,
+      "loss": 0.8778,
+      "step": 64
+    },
+    {
+      "epoch": 0.14277869302581,
+      "grad_norm": 3.154684543609619,
+      "learning_rate": 0.00014233576642335764,
+      "loss": 0.8859,
+      "step": 65
+    },
+    {
+      "epoch": 0.14497528830313014,
+      "grad_norm": 0.8627920150756836,
+      "learning_rate": 0.00014452554744525547,
+      "loss": 0.6573,
+      "step": 66
+    },
+    {
+      "epoch": 0.1471718835804503,
+      "grad_norm": 1.2885806560516357,
+      "learning_rate": 0.00014671532846715328,
+      "loss": 0.9026,
+      "step": 67
+    },
+    {
+      "epoch": 0.14936847885777046,
+      "grad_norm": 1.5568703413009644,
+      "learning_rate": 0.00014890510948905108,
+      "loss": 0.8177,
+      "step": 68
+    },
+    {
+      "epoch": 0.1515650741350906,
+      "grad_norm": 1.4923875331878662,
+      "learning_rate": 0.00015109489051094892,
+      "loss": 0.9572,
+      "step": 69
+    },
+    {
+      "epoch": 0.15376166941241076,
+      "grad_norm": 3.543449640274048,
+      "learning_rate": 0.00015328467153284672,
+      "loss": 0.7153,
+      "step": 70
+    },
+    {
+      "epoch": 0.15595826468973092,
+      "grad_norm": 1.2630614042282104,
+      "learning_rate": 0.00015547445255474453,
+      "loss": 0.9614,
+      "step": 71
+    },
+    {
+      "epoch": 0.15815485996705106,
+      "grad_norm": 1.6532517671585083,
+      "learning_rate": 0.00015766423357664234,
+      "loss": 0.9317,
+      "step": 72
+    },
+    {
+      "epoch": 0.16035145524437122,
+      "grad_norm": 1.4471046924591064,
+      "learning_rate": 0.00015985401459854014,
+      "loss": 1.0455,
+      "step": 73
+    },
+    {
+      "epoch": 0.16254805052169138,
+      "grad_norm": 0.9957731366157532,
+      "learning_rate": 0.00016204379562043795,
+      "loss": 0.9827,
+      "step": 74
+    },
+    {
+      "epoch": 0.16474464579901152,
+      "grad_norm": 1.85066556930542,
+      "learning_rate": 0.00016423357664233575,
+      "loss": 0.7693,
+      "step": 75
+    },
+    {
+      "epoch": 0.16694124107633168,
+      "grad_norm": 1.0510252714157104,
+      "learning_rate": 0.00016642335766423356,
+      "loss": 0.8786,
+      "step": 76
+    },
+    {
+      "epoch": 0.16913783635365184,
+      "grad_norm": 1.0159761905670166,
+      "learning_rate": 0.00016861313868613137,
+      "loss": 0.8475,
+      "step": 77
+    },
+    {
+      "epoch": 0.171334431630972,
+      "grad_norm": 1.1185177564620972,
+      "learning_rate": 0.00017080291970802917,
+      "loss": 0.5964,
+      "step": 78
+    },
+    {
+      "epoch": 0.17353102690829214,
+      "grad_norm": 1.4104725122451782,
+      "learning_rate": 0.00017299270072992698,
+      "loss": 0.8459,
+      "step": 79
+    },
+    {
+      "epoch": 0.1757276221856123,
+      "grad_norm": 1.4093234539031982,
+      "learning_rate": 0.00017518248175182478,
+      "loss": 0.9071,
+      "step": 80
+    },
+    {
+      "epoch": 0.17792421746293247,
+      "grad_norm": 1.0570992231369019,
+      "learning_rate": 0.00017737226277372262,
+      "loss": 0.8152,
+      "step": 81
+    },
+    {
+      "epoch": 0.1801208127402526,
+      "grad_norm": 1.1137661933898926,
+      "learning_rate": 0.00017956204379562042,
+      "loss": 0.8412,
+      "step": 82
+    },
+    {
+      "epoch": 0.18231740801757276,
+      "grad_norm": 1.2393567562103271,
+      "learning_rate": 0.00018175182481751826,
+      "loss": 0.7963,
+      "step": 83
+    },
+    {
+      "epoch": 0.18451400329489293,
+      "grad_norm": 1.36848783493042,
+      "learning_rate": 0.00018394160583941606,
+      "loss": 0.8555,
+      "step": 84
+    },
+    {
+      "epoch": 0.18671059857221306,
+      "grad_norm": 1.4011822938919067,
+      "learning_rate": 0.00018613138686131387,
+      "loss": 0.9029,
+      "step": 85
+    },
+    {
+      "epoch": 0.18890719384953322,
+      "grad_norm": 1.1995725631713867,
+      "learning_rate": 0.00018832116788321167,
+      "loss": 1.114,
+      "step": 86
+    },
+    {
+      "epoch": 0.19110378912685339,
+      "grad_norm": 0.9653306603431702,
+      "learning_rate": 0.00019051094890510948,
+      "loss": 0.9104,
+      "step": 87
+    },
+    {
+      "epoch": 0.19330038440417352,
+      "grad_norm": 0.9534090161323547,
+      "learning_rate": 0.0001927007299270073,
+      "loss": 0.8769,
+      "step": 88
+    },
+    {
+      "epoch": 0.19549697968149368,
+      "grad_norm": 1.885206699371338,
+      "learning_rate": 0.0001948905109489051,
+      "loss": 0.6802,
+      "step": 89
+    },
+    {
+      "epoch": 0.19769357495881384,
+      "grad_norm": 1.0323516130447388,
+      "learning_rate": 0.0001970802919708029,
+      "loss": 1.0172,
+      "step": 90
+    },
+    {
+      "epoch": 0.19989017023613398,
+      "grad_norm": 1.2367773056030273,
+      "learning_rate": 0.0001992700729927007,
+      "loss": 0.8245,
+      "step": 91
+    },
+    {
+      "epoch": 0.20208676551345414,
+      "grad_norm": 1.0401471853256226,
+      "learning_rate": 0.0002014598540145985,
+      "loss": 0.8595,
+      "step": 92
+    },
+    {
+      "epoch": 0.2042833607907743,
+      "grad_norm": 1.1214776039123535,
+      "learning_rate": 0.00020364963503649632,
+      "loss": 0.7957,
+      "step": 93
+    },
+    {
+      "epoch": 0.20647995606809447,
+      "grad_norm": 1.2164690494537354,
+      "learning_rate": 0.00020583941605839415,
+      "loss": 1.0177,
+      "step": 94
+    },
+    {
+      "epoch": 0.2086765513454146,
+      "grad_norm": 0.9195399880409241,
+      "learning_rate": 0.00020802919708029196,
+      "loss": 0.7789,
+      "step": 95
+    },
+    {
+      "epoch": 0.21087314662273476,
+      "grad_norm": 1.6409378051757812,
+      "learning_rate": 0.00021021897810218976,
+      "loss": 0.9191,
+      "step": 96
+    },
+    {
+      "epoch": 0.21306974190005493,
+      "grad_norm": 1.3301479816436768,
+      "learning_rate": 0.00021240875912408757,
+      "loss": 0.942,
+      "step": 97
+    },
+    {
+      "epoch": 0.21526633717737506,
+      "grad_norm": 0.8969391584396362,
+      "learning_rate": 0.00021459854014598537,
+      "loss": 0.8748,
+      "step": 98
+    },
+    {
+      "epoch": 0.21746293245469522,
+      "grad_norm": 1.0673491954803467,
+      "learning_rate": 0.0002167883211678832,
+      "loss": 0.9409,
+      "step": 99
+    },
+    {
+      "epoch": 0.21965952773201539,
+      "grad_norm": 0.8517920970916748,
+      "learning_rate": 0.00021897810218978101,
+      "loss": 0.8311,
+      "step": 100
+    },
+    {
+      "epoch": 0.22185612300933552,
+      "grad_norm": 1.309915542602539,
+      "learning_rate": 0.00022116788321167882,
+      "loss": 0.9646,
+      "step": 101
+    },
+    {
+      "epoch": 0.22405271828665568,
+      "grad_norm": 1.012101173400879,
+      "learning_rate": 0.00022335766423357663,
+      "loss": 0.8336,
+      "step": 102
+    },
+    {
+      "epoch": 0.22624931356397585,
+      "grad_norm": 0.901973307132721,
+      "learning_rate": 0.00022554744525547443,
+      "loss": 0.9094,
+      "step": 103
+    },
+    {
+      "epoch": 0.22844590884129598,
+      "grad_norm": 1.3551793098449707,
+      "learning_rate": 0.00022773722627737224,
+      "loss": 0.7341,
+      "step": 104
+    },
+    {
+      "epoch": 0.23064250411861614,
+      "grad_norm": 0.8551684617996216,
+      "learning_rate": 0.00022992700729927004,
+      "loss": 0.903,
+      "step": 105
+    },
+    {
+      "epoch": 0.2328390993959363,
+      "grad_norm": 0.9541202187538147,
+      "learning_rate": 0.00023211678832116788,
+      "loss": 0.9764,
+      "step": 106
+    },
+    {
+      "epoch": 0.23503569467325644,
+      "grad_norm": 1.2370141744613647,
+      "learning_rate": 0.00023430656934306568,
+      "loss": 0.8687,
+      "step": 107
+    },
+    {
+      "epoch": 0.2372322899505766,
+      "grad_norm": 1.2177876234054565,
+      "learning_rate": 0.0002364963503649635,
+      "loss": 0.7983,
+      "step": 108
+    },
+    {
+      "epoch": 0.23942888522789676,
+      "grad_norm": 1.0416233539581299,
+      "learning_rate": 0.0002386861313868613,
+      "loss": 0.7401,
+      "step": 109
+    },
+    {
+      "epoch": 0.24162548050521693,
+      "grad_norm": 2.0632760524749756,
+      "learning_rate": 0.0002408759124087591,
+      "loss": 0.8392,
+      "step": 110
+    },
+    {
+      "epoch": 0.24382207578253706,
+      "grad_norm": 1.0319123268127441,
+      "learning_rate": 0.0002430656934306569,
+      "loss": 0.8413,
+      "step": 111
+    },
+    {
+      "epoch": 0.24601867105985722,
+      "grad_norm": 0.7957919239997864,
+      "learning_rate": 0.0002452554744525547,
+      "loss": 0.8145,
+      "step": 112
+    },
+    {
+      "epoch": 0.2482152663371774,
+      "grad_norm": 0.7251466512680054,
+      "learning_rate": 0.0002474452554744525,
+      "loss": 0.7947,
+      "step": 113
+    },
+    {
+      "epoch": 0.2504118616144975,
+      "grad_norm": 1.4797084331512451,
+      "learning_rate": 0.0002496350364963503,
+      "loss": 0.7174,
+      "step": 114
+    },
+    {
+      "epoch": 0.25260845689181766,
+      "grad_norm": 1.0762050151824951,
+      "learning_rate": 0.00025182481751824813,
+      "loss": 0.7884,
+      "step": 115
+    },
+    {
+      "epoch": 0.25480505216913785,
+      "grad_norm": 0.7813681960105896,
+      "learning_rate": 0.00025401459854014594,
+      "loss": 0.9394,
+      "step": 116
+    },
+    {
+      "epoch": 0.257001647446458,
+      "grad_norm": 0.8607606887817383,
+      "learning_rate": 0.00025620437956204374,
+      "loss": 0.8496,
+      "step": 117
+    },
+    {
+      "epoch": 0.2591982427237781,
+      "grad_norm": 0.7608792185783386,
+      "learning_rate": 0.00025839416058394155,
+      "loss": 0.8189,
+      "step": 118
+    },
+    {
+      "epoch": 0.2613948380010983,
+      "grad_norm": 0.8634129166603088,
+      "learning_rate": 0.0002605839416058394,
+      "loss": 0.7842,
+      "step": 119
+    },
+    {
+      "epoch": 0.26359143327841844,
+      "grad_norm": 0.7820402979850769,
+      "learning_rate": 0.0002627737226277372,
+      "loss": 0.674,
+      "step": 120
+    },
+    {
+      "epoch": 0.26578802855573863,
+      "grad_norm": 0.7390574812889099,
+      "learning_rate": 0.000264963503649635,
+      "loss": 0.775,
+      "step": 121
+    },
+    {
+      "epoch": 0.26798462383305877,
+      "grad_norm": 0.7152166366577148,
+      "learning_rate": 0.00026715328467153283,
+      "loss": 0.7093,
+      "step": 122
+    },
+    {
+      "epoch": 0.2701812191103789,
+      "grad_norm": 1.3190675973892212,
+      "learning_rate": 0.00026934306569343063,
+      "loss": 0.7733,
+      "step": 123
+    },
+    {
+      "epoch": 0.2723778143876991,
+      "grad_norm": 1.1428911685943604,
+      "learning_rate": 0.00027153284671532844,
+      "loss": 0.7289,
+      "step": 124
+    },
+    {
+      "epoch": 0.2745744096650192,
+      "grad_norm": 1.3784867525100708,
+      "learning_rate": 0.00027372262773722625,
+      "loss": 0.8646,
+      "step": 125
+    },
+    {
+      "epoch": 0.27677100494233936,
+      "grad_norm": 1.01291823387146,
+      "learning_rate": 0.00027591240875912405,
+      "loss": 0.5823,
+      "step": 126
+    },
+    {
+      "epoch": 0.27896760021965955,
+      "grad_norm": 0.7914912104606628,
+      "learning_rate": 0.00027810218978102186,
+      "loss": 0.7833,
+      "step": 127
+    },
+    {
+      "epoch": 0.2811641954969797,
+      "grad_norm": 1.2167949676513672,
+      "learning_rate": 0.00028029197080291966,
+      "loss": 1.0536,
+      "step": 128
+    },
+    {
+      "epoch": 0.2833607907742998,
+      "grad_norm": 0.6264751553535461,
+      "learning_rate": 0.00028248175182481747,
+      "loss": 0.9051,
+      "step": 129
+    },
+    {
+      "epoch": 0.28555738605162,
+      "grad_norm": 0.9899543523788452,
+      "learning_rate": 0.0002846715328467153,
+      "loss": 0.7668,
+      "step": 130
+    },
+    {
+      "epoch": 0.28775398132894014,
+      "grad_norm": 0.8331793546676636,
+      "learning_rate": 0.00028686131386861314,
+      "loss": 0.8112,
+      "step": 131
+    },
+    {
+      "epoch": 0.2899505766062603,
+      "grad_norm": 0.5522931218147278,
+      "learning_rate": 0.00028905109489051094,
+      "loss": 0.6626,
+      "step": 132
+    },
+    {
+      "epoch": 0.29214717188358047,
+      "grad_norm": 1.1690224409103394,
+      "learning_rate": 0.00029124087591240875,
+      "loss": 0.6982,
+      "step": 133
+    },
+    {
+      "epoch": 0.2943437671609006,
+      "grad_norm": 0.854404628276825,
+      "learning_rate": 0.00029343065693430656,
+      "loss": 0.7089,
+      "step": 134
+    },
+    {
+      "epoch": 0.29654036243822074,
+      "grad_norm": 0.6672958135604858,
+      "learning_rate": 0.00029562043795620436,
+      "loss": 0.8113,
+      "step": 135
+    },
+    {
+      "epoch": 0.29873695771554093,
+      "grad_norm": 1.0035560131072998,
+      "learning_rate": 0.00029781021897810217,
+      "loss": 0.938,
+      "step": 136
+    },
+    {
+      "epoch": 0.30093355299286106,
+      "grad_norm": 0.7653816342353821,
+      "learning_rate": 0.0003,
+      "loss": 0.7909,
+      "step": 137
+    },
+    {
+      "epoch": 0.3031301482701812,
+      "grad_norm": 0.5396126508712769,
+      "learning_rate": 0.00029975570032573286,
+      "loss": 0.8203,
+      "step": 138
+    },
+    {
+      "epoch": 0.3053267435475014,
+      "grad_norm": 0.9712949395179749,
+      "learning_rate": 0.00029951140065146575,
+      "loss": 0.7563,
+      "step": 139
+    },
+    {
+      "epoch": 0.3075233388248215,
+      "grad_norm": 0.7658219337463379,
+      "learning_rate": 0.0002992671009771987,
+      "loss": 0.8821,
+      "step": 140
+    },
+    {
+      "epoch": 0.30971993410214166,
+      "grad_norm": 2.0263867378234863,
+      "learning_rate": 0.0002990228013029316,
+      "loss": 0.8306,
+      "step": 141
+    },
+    {
+      "epoch": 0.31191652937946185,
+      "grad_norm": 1.16373610496521,
+      "learning_rate": 0.0002987785016286645,
+      "loss": 0.9306,
+      "step": 142
+    },
+    {
+      "epoch": 0.314113124656782,
+      "grad_norm": 0.710020899772644,
+      "learning_rate": 0.00029853420195439736,
+      "loss": 0.808,
+      "step": 143
+    },
+    {
+      "epoch": 0.3163097199341021,
+      "grad_norm": 0.7870954871177673,
+      "learning_rate": 0.00029828990228013025,
+      "loss": 0.9492,
+      "step": 144
+    },
+    {
+      "epoch": 0.3185063152114223,
+      "grad_norm": 0.8841456174850464,
+      "learning_rate": 0.00029804560260586314,
+      "loss": 0.8703,
+      "step": 145
+    },
+    {
+      "epoch": 0.32070291048874244,
+      "grad_norm": 0.6581629514694214,
+      "learning_rate": 0.0002978013029315961,
+      "loss": 0.8029,
+      "step": 146
+    },
+    {
+      "epoch": 0.3228995057660626,
+      "grad_norm": 1.1622114181518555,
+      "learning_rate": 0.000297557003257329,
+      "loss": 0.7137,
+      "step": 147
+    },
+    {
+      "epoch": 0.32509610104338277,
+      "grad_norm": 0.9114773869514465,
+      "learning_rate": 0.00029731270358306186,
+      "loss": 0.7112,
+      "step": 148
+    },
+    {
+      "epoch": 0.3272926963207029,
+      "grad_norm": 0.7541104555130005,
+      "learning_rate": 0.00029706840390879475,
+      "loss": 0.778,
+      "step": 149
+    },
+    {
+      "epoch": 0.32948929159802304,
+      "grad_norm": 0.7600201368331909,
+      "learning_rate": 0.00029682410423452764,
+      "loss": 1.1153,
+      "step": 150
+    },
+    {
+      "epoch": 0.3316858868753432,
+      "grad_norm": 0.6182475090026855,
+      "learning_rate": 0.0002965798045602606,
+      "loss": 0.922,
+      "step": 151
+    },
+    {
+      "epoch": 0.33388248215266336,
+      "grad_norm": 0.7557980418205261,
+      "learning_rate": 0.0002963355048859935,
+      "loss": 0.9118,
+      "step": 152
+    },
+    {
+      "epoch": 0.33607907742998355,
+      "grad_norm": 0.6967447996139526,
+      "learning_rate": 0.00029609120521172636,
+      "loss": 0.7084,
+      "step": 153
+    },
+    {
+      "epoch": 0.3382756727073037,
+      "grad_norm": 0.774140477180481,
+      "learning_rate": 0.00029584690553745925,
+      "loss": 0.7821,
+      "step": 154
+    },
+    {
+      "epoch": 0.3404722679846238,
+      "grad_norm": 0.9071000814437866,
+      "learning_rate": 0.00029560260586319214,
+      "loss": 0.7164,
+      "step": 155
+    },
+    {
+      "epoch": 0.342668863261944,
+      "grad_norm": 0.5448876023292542,
+      "learning_rate": 0.00029535830618892503,
+      "loss": 0.8082,
+      "step": 156
+    },
+    {
+      "epoch": 0.34486545853926415,
+      "grad_norm": 2.521521806716919,
+      "learning_rate": 0.000295114006514658,
+      "loss": 0.8154,
+      "step": 157
+    },
+    {
+      "epoch": 0.3470620538165843,
+      "grad_norm": 0.691748321056366,
+      "learning_rate": 0.00029486970684039086,
+      "loss": 0.7647,
+      "step": 158
+    },
+    {
+      "epoch": 0.34925864909390447,
+      "grad_norm": 0.6816907525062561,
+      "learning_rate": 0.00029462540716612375,
+      "loss": 0.9097,
+      "step": 159
+    },
+    {
+      "epoch": 0.3514552443712246,
+      "grad_norm": 0.7295282483100891,
+      "learning_rate": 0.00029438110749185664,
+      "loss": 0.7701,
+      "step": 160
+    },
+    {
+      "epoch": 0.35365183964854474,
+      "grad_norm": 1.8857238292694092,
+      "learning_rate": 0.00029413680781758953,
+      "loss": 0.7752,
+      "step": 161
+    },
+    {
+      "epoch": 0.35584843492586493,
+      "grad_norm": 0.4814364016056061,
+      "learning_rate": 0.0002938925081433224,
+      "loss": 0.6934,
+      "step": 162
+    },
+    {
+      "epoch": 0.35804503020318507,
+      "grad_norm": 0.6610971689224243,
+      "learning_rate": 0.00029364820846905536,
+      "loss": 0.5388,
+      "step": 163
+    },
+    {
+      "epoch": 0.3602416254805052,
+      "grad_norm": 0.8607411980628967,
+      "learning_rate": 0.00029340390879478825,
+      "loss": 0.7394,
+      "step": 164
+    },
+    {
+      "epoch": 0.3624382207578254,
+      "grad_norm": 0.772109866142273,
+      "learning_rate": 0.00029315960912052114,
+      "loss": 0.9169,
+      "step": 165
+    },
+    {
+      "epoch": 0.3646348160351455,
+      "grad_norm": 0.9444490671157837,
+      "learning_rate": 0.00029291530944625403,
+      "loss": 0.7791,
+      "step": 166
+    },
+    {
+      "epoch": 0.36683141131246566,
+      "grad_norm": 3.3315327167510986,
+      "learning_rate": 0.0002926710097719869,
+      "loss": 0.7821,
+      "step": 167
+    },
+    {
+      "epoch": 0.36902800658978585,
+      "grad_norm": 0.8838211894035339,
+      "learning_rate": 0.00029242671009771986,
+      "loss": 0.7836,
+      "step": 168
+    },
+    {
+      "epoch": 0.371224601867106,
+      "grad_norm": 0.8367225527763367,
+      "learning_rate": 0.00029218241042345275,
+      "loss": 0.8315,
+      "step": 169
+    },
+    {
+      "epoch": 0.3734211971444261,
+      "grad_norm": 0.997858464717865,
+      "learning_rate": 0.00029193811074918564,
+      "loss": 0.9123,
+      "step": 170
+    },
+    {
+      "epoch": 0.3756177924217463,
+      "grad_norm": 1.398758888244629,
+      "learning_rate": 0.00029169381107491853,
+      "loss": 0.7386,
+      "step": 171
+    },
+    {
+      "epoch": 0.37781438769906645,
+      "grad_norm": 0.6019458770751953,
+      "learning_rate": 0.0002914495114006514,
+      "loss": 0.646,
+      "step": 172
+    },
+    {
+      "epoch": 0.3800109829763866,
+      "grad_norm": 0.6285191178321838,
+      "learning_rate": 0.0002912052117263843,
+      "loss": 0.8025,
+      "step": 173
+    },
+    {
+      "epoch": 0.38220757825370677,
+      "grad_norm": 1.2815974950790405,
+      "learning_rate": 0.00029096091205211725,
+      "loss": 1.0532,
+      "step": 174
+    },
+    {
+      "epoch": 0.3844041735310269,
+      "grad_norm": 0.6931642293930054,
+      "learning_rate": 0.00029071661237785014,
+      "loss": 0.655,
+      "step": 175
+    },
+    {
+      "epoch": 0.38660076880834704,
+      "grad_norm": 0.7077628970146179,
+      "learning_rate": 0.00029047231270358303,
+      "loss": 0.9975,
+      "step": 176
+    },
+    {
+      "epoch": 0.38879736408566723,
+      "grad_norm": 0.7028080224990845,
+      "learning_rate": 0.0002902280130293159,
+      "loss": 0.5206,
+      "step": 177
+    },
+    {
+      "epoch": 0.39099395936298736,
+      "grad_norm": 0.6993598341941833,
+      "learning_rate": 0.0002899837133550488,
+      "loss": 0.8134,
+      "step": 178
+    },
+    {
+      "epoch": 0.3931905546403075,
+      "grad_norm": 0.9487060308456421,
+      "learning_rate": 0.00028973941368078175,
+      "loss": 0.8331,
+      "step": 179
+    },
+    {
+      "epoch": 0.3953871499176277,
+      "grad_norm": 0.6464899778366089,
+      "learning_rate": 0.00028949511400651464,
+      "loss": 0.6935,
+      "step": 180
+    },
+    {
+      "epoch": 0.3975837451949478,
+      "grad_norm": 1.02102792263031,
+      "learning_rate": 0.00028925081433224753,
+      "loss": 0.7109,
+      "step": 181
+    },
+    {
+      "epoch": 0.39978034047226796,
+      "grad_norm": 0.7637264132499695,
+      "learning_rate": 0.0002890065146579804,
+      "loss": 0.6217,
+      "step": 182
+    },
+    {
+      "epoch": 0.40197693574958815,
+      "grad_norm": 0.6201541423797607,
+      "learning_rate": 0.0002887622149837133,
+      "loss": 0.8404,
+      "step": 183
+    },
+    {
+      "epoch": 0.4041735310269083,
+      "grad_norm": 0.6842285990715027,
+      "learning_rate": 0.0002885179153094462,
+      "loss": 1.1365,
+      "step": 184
+    },
+    {
+      "epoch": 0.4063701263042284,
+      "grad_norm": 0.777034342288971,
+      "learning_rate": 0.00028827361563517914,
+      "loss": 0.8934,
+      "step": 185
+    },
+    {
+      "epoch": 0.4085667215815486,
+      "grad_norm": 0.7470583915710449,
+      "learning_rate": 0.00028802931596091203,
+      "loss": 0.8096,
+      "step": 186
+    },
+    {
+      "epoch": 0.41076331685886874,
+      "grad_norm": 1.3292444944381714,
+      "learning_rate": 0.0002877850162866449,
+      "loss": 0.6082,
+      "step": 187
+    },
+    {
+      "epoch": 0.41295991213618893,
+      "grad_norm": 0.6230322122573853,
+      "learning_rate": 0.0002875407166123778,
+      "loss": 0.715,
+      "step": 188
+    },
+    {
+      "epoch": 0.41515650741350907,
+      "grad_norm": 1.3206838369369507,
+      "learning_rate": 0.0002872964169381107,
+      "loss": 0.8536,
+      "step": 189
+    },
+    {
+      "epoch": 0.4173531026908292,
+      "grad_norm": 0.785211443901062,
+      "learning_rate": 0.00028705211726384364,
+      "loss": 0.777,
+      "step": 190
+    },
+    {
+      "epoch": 0.4195496979681494,
+      "grad_norm": 4.01908016204834,
+      "learning_rate": 0.00028680781758957653,
+      "loss": 0.847,
+      "step": 191
+    },
+    {
+      "epoch": 0.42174629324546953,
+      "grad_norm": 0.7183860540390015,
+      "learning_rate": 0.0002865635179153094,
+      "loss": 0.875,
+      "step": 192
+    },
+    {
+      "epoch": 0.42394288852278966,
+      "grad_norm": 1.0391347408294678,
+      "learning_rate": 0.0002863192182410423,
+      "loss": 0.8001,
+      "step": 193
+    },
+    {
+      "epoch": 0.42613948380010985,
+      "grad_norm": 0.8809317350387573,
+      "learning_rate": 0.0002860749185667752,
+      "loss": 0.8371,
+      "step": 194
+    },
+    {
+      "epoch": 0.42833607907743,
+      "grad_norm": 0.9177643060684204,
+      "learning_rate": 0.0002858306188925081,
+      "loss": 1.0427,
+      "step": 195
+    },
+    {
+      "epoch": 0.4305326743547501,
+      "grad_norm": 0.5758705139160156,
+      "learning_rate": 0.00028558631921824103,
+      "loss": 1.1949,
+      "step": 196
+    },
+    {
+      "epoch": 0.4327292696320703,
+      "grad_norm": 0.8856555223464966,
+      "learning_rate": 0.0002853420195439739,
+      "loss": 0.7758,
+      "step": 197
+    },
+    {
+      "epoch": 0.43492586490939045,
+      "grad_norm": 0.6175949573516846,
+      "learning_rate": 0.0002850977198697068,
+      "loss": 0.6945,
+      "step": 198
+    },
+    {
+      "epoch": 0.4371224601867106,
+      "grad_norm": 0.9313458800315857,
+      "learning_rate": 0.0002848534201954397,
+      "loss": 0.7812,
+      "step": 199
+    },
+    {
+      "epoch": 0.43931905546403077,
+      "grad_norm": 0.7142427563667297,
+      "learning_rate": 0.0002846091205211726,
+      "loss": 0.9095,
+      "step": 200
+    },
+    {
+      "epoch": 0.4415156507413509,
+      "grad_norm": 1.292242169380188,
+      "learning_rate": 0.0002843648208469055,
+      "loss": 0.6471,
+      "step": 201
+    },
+    {
+      "epoch": 0.44371224601867104,
+      "grad_norm": 0.8002964854240417,
+      "learning_rate": 0.0002841205211726384,
+      "loss": 0.6997,
+      "step": 202
+    },
+    {
+      "epoch": 0.44590884129599123,
+      "grad_norm": 0.5370644330978394,
+      "learning_rate": 0.0002838762214983713,
+      "loss": 0.792,
+      "step": 203
+    },
+    {
+      "epoch": 0.44810543657331137,
+      "grad_norm": 0.6608684062957764,
+      "learning_rate": 0.0002836319218241042,
+      "loss": 0.9066,
+      "step": 204
+    },
+    {
+      "epoch": 0.4503020318506315,
+      "grad_norm": 0.6515672206878662,
+      "learning_rate": 0.0002833876221498371,
+      "loss": 0.9699,
+      "step": 205
+    },
+    {
+      "epoch": 0.4524986271279517,
+      "grad_norm": 1.0029901266098022,
+      "learning_rate": 0.00028314332247557,
+      "loss": 1.0307,
+      "step": 206
+    },
+    {
+      "epoch": 0.4546952224052718,
+      "grad_norm": 0.9367266893386841,
+      "learning_rate": 0.0002828990228013029,
+      "loss": 0.8316,
+      "step": 207
+    },
+    {
+      "epoch": 0.45689181768259196,
+      "grad_norm": 0.5828459858894348,
+      "learning_rate": 0.0002826547231270358,
+      "loss": 0.7577,
+      "step": 208
+    },
+    {
+      "epoch": 0.45908841295991215,
+      "grad_norm": 1.0911840200424194,
+      "learning_rate": 0.0002824104234527687,
+      "loss": 0.8395,
+      "step": 209
+    },
+    {
+      "epoch": 0.4612850082372323,
+      "grad_norm": 1.2683651447296143,
+      "learning_rate": 0.0002821661237785016,
+      "loss": 0.9315,
+      "step": 210
+    },
+    {
+      "epoch": 0.4634816035145524,
+      "grad_norm": 0.7885321378707886,
+      "learning_rate": 0.0002819218241042345,
+      "loss": 0.9687,
+      "step": 211
+    },
+    {
+      "epoch": 0.4656781987918726,
+      "grad_norm": 0.6706778407096863,
+      "learning_rate": 0.00028167752442996737,
+      "loss": 0.8335,
+      "step": 212
+    },
+    {
+      "epoch": 0.46787479406919275,
+      "grad_norm": 0.5244065523147583,
+      "learning_rate": 0.0002814332247557003,
+      "loss": 0.5743,
+      "step": 213
+    },
+    {
+      "epoch": 0.4700713893465129,
+      "grad_norm": 0.47366344928741455,
+      "learning_rate": 0.0002811889250814332,
+      "loss": 0.7479,
+      "step": 214
+    },
+    {
+      "epoch": 0.47226798462383307,
+      "grad_norm": 0.8778125047683716,
+      "learning_rate": 0.0002809446254071661,
+      "loss": 0.9419,
+      "step": 215
+    },
+    {
+      "epoch": 0.4744645799011532,
+      "grad_norm": 0.7257303595542908,
+      "learning_rate": 0.000280700325732899,
+      "loss": 0.7803,
+      "step": 216
+    },
+    {
+      "epoch": 0.47666117517847334,
+      "grad_norm": 0.5721178650856018,
+      "learning_rate": 0.00028045602605863187,
+      "loss": 0.7003,
+      "step": 217
+    },
+    {
+      "epoch": 0.47885777045579353,
+      "grad_norm": 0.5440539121627808,
+      "learning_rate": 0.0002802117263843648,
+      "loss": 0.881,
+      "step": 218
+    },
+    {
+      "epoch": 0.48105436573311366,
+      "grad_norm": 0.7223842740058899,
+      "learning_rate": 0.0002799674267100977,
+      "loss": 0.7399,
+      "step": 219
+    },
+    {
+      "epoch": 0.48325096101043385,
+      "grad_norm": 1.251110553741455,
+      "learning_rate": 0.0002797231270358306,
+      "loss": 0.8416,
+      "step": 220
+    },
+    {
+      "epoch": 0.485447556287754,
+      "grad_norm": 0.7959457039833069,
+      "learning_rate": 0.0002794788273615635,
+      "loss": 0.7418,
+      "step": 221
+    },
+    {
+      "epoch": 0.4876441515650741,
+      "grad_norm": 0.6500054001808167,
+      "learning_rate": 0.00027923452768729637,
+      "loss": 0.643,
+      "step": 222
+    },
+    {
+      "epoch": 0.4898407468423943,
+      "grad_norm": 0.8789168000221252,
+      "learning_rate": 0.00027899022801302926,
+      "loss": 0.8576,
+      "step": 223
+    },
+    {
+      "epoch": 0.49203734211971445,
+      "grad_norm": 0.5097317695617676,
+      "learning_rate": 0.0002787459283387622,
+      "loss": 0.8547,
+      "step": 224
+    },
+    {
+      "epoch": 0.4942339373970346,
+      "grad_norm": 0.5988264083862305,
+      "learning_rate": 0.0002785016286644951,
+      "loss": 0.6227,
+      "step": 225
+    },
+    {
+      "epoch": 0.4964305326743548,
+      "grad_norm": 0.5047509670257568,
+      "learning_rate": 0.000278257328990228,
+      "loss": 1.0636,
+      "step": 226
+    },
+    {
+      "epoch": 0.4986271279516749,
+      "grad_norm": 0.3800826668739319,
+      "learning_rate": 0.00027801302931596087,
+      "loss": 0.7078,
+      "step": 227
+    },
+    {
+      "epoch": 0.500823723228995,
+      "grad_norm": 0.5140970945358276,
+      "learning_rate": 0.00027776872964169376,
+      "loss": 0.7652,
+      "step": 228
+    },
+    {
+      "epoch": 0.5030203185063152,
+      "grad_norm": 0.6369237899780273,
+      "learning_rate": 0.0002775244299674267,
+      "loss": 0.7997,
+      "step": 229
+    },
+    {
+      "epoch": 0.5052169137836353,
+      "grad_norm": 0.5654761791229248,
+      "learning_rate": 0.0002772801302931596,
+      "loss": 0.8956,
+      "step": 230
+    },
+    {
+      "epoch": 0.5074135090609555,
+      "grad_norm": 0.7557458281517029,
+      "learning_rate": 0.0002770358306188925,
+      "loss": 0.7907,
+      "step": 231
+    },
+    {
+      "epoch": 0.5096101043382757,
+      "grad_norm": 0.5533975958824158,
+      "learning_rate": 0.00027679153094462537,
+      "loss": 0.7527,
+      "step": 232
+    },
+    {
+      "epoch": 0.5118066996155958,
+      "grad_norm": 0.48601865768432617,
+      "learning_rate": 0.00027654723127035826,
+      "loss": 0.8416,
+      "step": 233
+    },
+    {
+      "epoch": 0.514003294892916,
+      "grad_norm": 0.5968899130821228,
+      "learning_rate": 0.00027630293159609115,
+      "loss": 0.8511,
+      "step": 234
+    },
+    {
+      "epoch": 0.5161998901702362,
+      "grad_norm": 0.884617805480957,
+      "learning_rate": 0.0002760586319218241,
+      "loss": 0.8657,
+      "step": 235
+    },
+    {
+      "epoch": 0.5183964854475562,
+      "grad_norm": 0.40955522656440735,
+      "learning_rate": 0.000275814332247557,
+      "loss": 0.6999,
+      "step": 236
+    },
+    {
+      "epoch": 0.5205930807248764,
+      "grad_norm": 0.5755212903022766,
+      "learning_rate": 0.00027557003257328987,
+      "loss": 1.0084,
+      "step": 237
+    },
+    {
+      "epoch": 0.5227896760021966,
+      "grad_norm": 0.5894397497177124,
+      "learning_rate": 0.00027532573289902276,
+      "loss": 0.8961,
+      "step": 238
+    },
+    {
+      "epoch": 0.5249862712795168,
+      "grad_norm": 0.4248576760292053,
+      "learning_rate": 0.00027508143322475565,
+      "loss": 0.7588,
+      "step": 239
+    },
+    {
+      "epoch": 0.5271828665568369,
+      "grad_norm": 0.655170738697052,
+      "learning_rate": 0.00027483713355048854,
+      "loss": 0.8639,
+      "step": 240
+    },
+    {
+      "epoch": 0.5293794618341571,
+      "grad_norm": 0.6119483113288879,
+      "learning_rate": 0.0002745928338762215,
+      "loss": 0.851,
+      "step": 241
+    },
+    {
+      "epoch": 0.5315760571114773,
+      "grad_norm": 0.5455263257026672,
+      "learning_rate": 0.00027434853420195437,
+      "loss": 0.7463,
+      "step": 242
+    },
+    {
+      "epoch": 0.5337726523887973,
+      "grad_norm": 0.7501972913742065,
+      "learning_rate": 0.00027410423452768726,
+      "loss": 0.6987,
+      "step": 243
+    },
+    {
+      "epoch": 0.5359692476661175,
+      "grad_norm": 0.917352557182312,
+      "learning_rate": 0.00027385993485342015,
+      "loss": 0.8947,
+      "step": 244
+    },
+    {
+      "epoch": 0.5381658429434377,
+      "grad_norm": 0.5471184849739075,
+      "learning_rate": 0.00027361563517915304,
+      "loss": 0.8118,
+      "step": 245
+    },
+    {
+      "epoch": 0.5403624382207578,
+      "grad_norm": 0.6769572496414185,
+      "learning_rate": 0.000273371335504886,
+      "loss": 0.8453,
+      "step": 246
+    },
+    {
+      "epoch": 0.542559033498078,
+      "grad_norm": 0.6192574501037598,
+      "learning_rate": 0.00027312703583061887,
+      "loss": 0.6973,
+      "step": 247
+    },
+    {
+      "epoch": 0.5447556287753982,
+      "grad_norm": 0.6607881188392639,
+      "learning_rate": 0.00027288273615635176,
+      "loss": 0.8128,
+      "step": 248
+    },
+    {
+      "epoch": 0.5469522240527183,
+      "grad_norm": 0.4594419002532959,
+      "learning_rate": 0.00027263843648208465,
+      "loss": 0.6797,
+      "step": 249
+    },
+    {
+      "epoch": 0.5491488193300385,
+      "grad_norm": 0.6312422156333923,
+      "learning_rate": 0.00027239413680781754,
+      "loss": 0.771,
+      "step": 250
+    },
+    {
+      "epoch": 0.5513454146073586,
+      "grad_norm": 0.6837294101715088,
+      "learning_rate": 0.0002721498371335504,
+      "loss": 0.8035,
+      "step": 251
+    },
+    {
+      "epoch": 0.5535420098846787,
+      "grad_norm": 0.7187260985374451,
+      "learning_rate": 0.00027190553745928337,
+      "loss": 0.8153,
+      "step": 252
+    },
+    {
+      "epoch": 0.5557386051619989,
+      "grad_norm": 0.5017946362495422,
+      "learning_rate": 0.00027166123778501626,
+      "loss": 0.8675,
+      "step": 253
+    },
+    {
+      "epoch": 0.5579352004393191,
+      "grad_norm": 0.6429175138473511,
+      "learning_rate": 0.00027141693811074915,
+      "loss": 0.7653,
+      "step": 254
+    },
+    {
+      "epoch": 0.5601317957166392,
+      "grad_norm": 0.6479099988937378,
+      "learning_rate": 0.00027117263843648204,
+      "loss": 0.8472,
+      "step": 255
+    },
+    {
+      "epoch": 0.5623283909939594,
+      "grad_norm": 0.7594976425170898,
+      "learning_rate": 0.0002709283387622149,
+      "loss": 0.807,
+      "step": 256
+    },
+    {
+      "epoch": 0.5645249862712796,
+      "grad_norm": 0.6511563062667847,
+      "learning_rate": 0.00027068403908794787,
+      "loss": 0.7198,
+      "step": 257
+    },
+    {
+      "epoch": 0.5667215815485996,
+      "grad_norm": 0.6936286091804504,
+      "learning_rate": 0.00027043973941368076,
+      "loss": 0.9117,
+      "step": 258
+    },
+    {
+      "epoch": 0.5689181768259198,
+      "grad_norm": 0.5176697373390198,
+      "learning_rate": 0.00027019543973941365,
+      "loss": 0.6935,
+      "step": 259
+    },
+    {
+      "epoch": 0.57111477210324,
+      "grad_norm": 1.2337859869003296,
+      "learning_rate": 0.00026995114006514654,
+      "loss": 0.8181,
+      "step": 260
+    },
+    {
+      "epoch": 0.5733113673805601,
+      "grad_norm": 0.4775691628456116,
+      "learning_rate": 0.00026970684039087943,
+      "loss": 0.831,
+      "step": 261
+    },
+    {
+      "epoch": 0.5755079626578803,
+      "grad_norm": 0.4145483374595642,
+      "learning_rate": 0.0002694625407166123,
+      "loss": 0.7984,
+      "step": 262
+    },
+    {
+      "epoch": 0.5777045579352005,
+      "grad_norm": 0.513195276260376,
+      "learning_rate": 0.00026921824104234526,
+      "loss": 0.6342,
+      "step": 263
+    },
+    {
+      "epoch": 0.5799011532125206,
+      "grad_norm": 0.7491148114204407,
+      "learning_rate": 0.00026897394136807815,
+      "loss": 0.8334,
+      "step": 264
+    },
+    {
+      "epoch": 0.5820977484898407,
+      "grad_norm": 0.7298436164855957,
+      "learning_rate": 0.00026872964169381104,
+      "loss": 1.1353,
+      "step": 265
+    },
+    {
+      "epoch": 0.5842943437671609,
+      "grad_norm": 0.5644757151603699,
+      "learning_rate": 0.00026848534201954393,
+      "loss": 0.9235,
+      "step": 266
+    },
+    {
+      "epoch": 0.586490939044481,
+      "grad_norm": 0.5992796421051025,
+      "learning_rate": 0.0002682410423452768,
+      "loss": 0.8871,
+      "step": 267
+    },
+    {
+      "epoch": 0.5886875343218012,
+      "grad_norm": 0.6124032139778137,
+      "learning_rate": 0.00026799674267100976,
+      "loss": 0.8538,
+      "step": 268
+    },
+    {
+      "epoch": 0.5908841295991214,
+      "grad_norm": 0.7897042632102966,
+      "learning_rate": 0.00026775244299674265,
+      "loss": 0.9903,
+      "step": 269
+    },
+    {
+      "epoch": 0.5930807248764415,
+      "grad_norm": 0.8144699335098267,
+      "learning_rate": 0.00026750814332247554,
+      "loss": 0.9419,
+      "step": 270
+    },
+    {
+      "epoch": 0.5952773201537617,
+      "grad_norm": 1.1205129623413086,
+      "learning_rate": 0.00026726384364820843,
+      "loss": 0.7545,
+      "step": 271
+    },
+    {
+      "epoch": 0.5974739154310819,
+      "grad_norm": 0.6697807312011719,
+      "learning_rate": 0.0002670195439739413,
+      "loss": 0.7732,
+      "step": 272
+    },
+    {
+      "epoch": 0.5996705107084019,
+      "grad_norm": 0.7082396149635315,
+      "learning_rate": 0.0002667752442996742,
+      "loss": 0.7227,
+      "step": 273
+    },
+    {
+      "epoch": 0.6018671059857221,
+      "grad_norm": 0.847800612449646,
+      "learning_rate": 0.00026653094462540715,
+      "loss": 0.7932,
+      "step": 274
+    },
+    {
+      "epoch": 0.6040637012630423,
+      "grad_norm": 0.7571720480918884,
+      "learning_rate": 0.00026628664495114004,
+      "loss": 0.8001,
+      "step": 275
+    },
+    {
+      "epoch": 0.6062602965403624,
+      "grad_norm": 0.6260753273963928,
+      "learning_rate": 0.00026604234527687293,
+      "loss": 0.5891,
+      "step": 276
+    },
+    {
+      "epoch": 0.6084568918176826,
+      "grad_norm": 0.48631101846694946,
+      "learning_rate": 0.0002657980456026058,
+      "loss": 0.8093,
+      "step": 277
+    },
+    {
+      "epoch": 0.6106534870950028,
+      "grad_norm": 0.9804533123970032,
+      "learning_rate": 0.0002655537459283387,
+      "loss": 0.9345,
+      "step": 278
+    },
+    {
+      "epoch": 0.6128500823723229,
+      "grad_norm": 0.5989837646484375,
+      "learning_rate": 0.00026530944625407165,
+      "loss": 0.7303,
+      "step": 279
+    },
+    {
+      "epoch": 0.615046677649643,
+      "grad_norm": 0.6332892775535583,
+      "learning_rate": 0.00026506514657980454,
+      "loss": 0.7189,
+      "step": 280
+    },
+    {
+      "epoch": 0.6172432729269632,
+      "grad_norm": 0.7891057729721069,
+      "learning_rate": 0.00026482084690553743,
+      "loss": 0.6574,
+      "step": 281
+    },
+    {
+      "epoch": 0.6194398682042833,
+      "grad_norm": 0.42643800377845764,
+      "learning_rate": 0.0002645765472312703,
+      "loss": 0.7113,
+      "step": 282
+    },
+    {
+      "epoch": 0.6216364634816035,
+      "grad_norm": 0.557184100151062,
+      "learning_rate": 0.0002643322475570032,
+      "loss": 0.8779,
+      "step": 283
+    },
+    {
+      "epoch": 0.6238330587589237,
+      "grad_norm": 0.5179101824760437,
+      "learning_rate": 0.0002640879478827361,
+      "loss": 0.7402,
+      "step": 284
+    },
+    {
+      "epoch": 0.6260296540362438,
+      "grad_norm": 0.5307636260986328,
+      "learning_rate": 0.00026384364820846904,
+      "loss": 0.7603,
+      "step": 285
+    },
+    {
+      "epoch": 0.628226249313564,
+      "grad_norm": 0.576750636100769,
+      "learning_rate": 0.00026359934853420193,
+      "loss": 0.737,
+      "step": 286
+    },
+    {
+      "epoch": 0.6304228445908842,
+      "grad_norm": 0.5686762928962708,
+      "learning_rate": 0.0002633550488599348,
+      "loss": 0.873,
+      "step": 287
+    },
+    {
+      "epoch": 0.6326194398682042,
+      "grad_norm": 0.4733280539512634,
+      "learning_rate": 0.0002631107491856677,
+      "loss": 0.7205,
+      "step": 288
+    },
+    {
+      "epoch": 0.6348160351455244,
+      "grad_norm": 0.49974897503852844,
+      "learning_rate": 0.0002628664495114006,
+      "loss": 0.6279,
+      "step": 289
+    },
+    {
+      "epoch": 0.6370126304228446,
+      "grad_norm": 0.7190808653831482,
+      "learning_rate": 0.0002626221498371335,
+      "loss": 0.6908,
+      "step": 290
+    },
+    {
+      "epoch": 0.6392092257001647,
+      "grad_norm": 1.0794044733047485,
+      "learning_rate": 0.00026237785016286643,
+      "loss": 0.838,
+      "step": 291
+    },
+    {
+      "epoch": 0.6414058209774849,
+      "grad_norm": 0.9557288289070129,
+      "learning_rate": 0.0002621335504885993,
+      "loss": 0.8268,
+      "step": 292
+    },
+    {
+      "epoch": 0.6436024162548051,
+      "grad_norm": 0.6780329346656799,
+      "learning_rate": 0.00026188925081433226,
+      "loss": 0.8891,
+      "step": 293
+    },
+    {
+      "epoch": 0.6457990115321252,
+      "grad_norm": 0.8087173700332642,
+      "learning_rate": 0.00026164495114006515,
+      "loss": 0.8201,
+      "step": 294
+    },
+    {
+      "epoch": 0.6479956068094453,
+      "grad_norm": 0.4673290252685547,
+      "learning_rate": 0.00026140065146579804,
+      "loss": 0.7768,
+      "step": 295
+    },
+    {
+      "epoch": 0.6501922020867655,
+      "grad_norm": 0.7236621975898743,
+      "learning_rate": 0.00026115635179153093,
+      "loss": 0.8103,
+      "step": 296
+    },
+    {
+      "epoch": 0.6523887973640856,
+      "grad_norm": 1.0686399936676025,
+      "learning_rate": 0.0002609120521172638,
+      "loss": 0.6794,
+      "step": 297
+    },
+    {
+      "epoch": 0.6545853926414058,
+      "grad_norm": 0.592258870601654,
+      "learning_rate": 0.00026066775244299676,
+      "loss": 0.7674,
+      "step": 298
+    },
+    {
+      "epoch": 0.656781987918726,
+      "grad_norm": 0.44138240814208984,
+      "learning_rate": 0.00026042345276872965,
+      "loss": 0.8456,
+      "step": 299
+    },
+    {
+      "epoch": 0.6589785831960461,
+      "grad_norm": 0.6155474781990051,
+      "learning_rate": 0.00026017915309446254,
+      "loss": 0.8285,
+      "step": 300
+    },
+    {
+      "epoch": 0.6611751784733663,
+      "grad_norm": 1.2270978689193726,
+      "learning_rate": 0.00025993485342019543,
+      "loss": 0.8813,
+      "step": 301
+    },
+    {
+      "epoch": 0.6633717737506865,
+      "grad_norm": 0.5877152681350708,
+      "learning_rate": 0.0002596905537459283,
+      "loss": 0.7936,
+      "step": 302
+    },
+    {
+      "epoch": 0.6655683690280065,
+      "grad_norm": 0.6085101366043091,
+      "learning_rate": 0.0002594462540716612,
+      "loss": 0.6771,
+      "step": 303
+    },
+    {
+      "epoch": 0.6677649643053267,
+      "grad_norm": 1.4141267538070679,
+      "learning_rate": 0.00025920195439739415,
+      "loss": 0.8748,
+      "step": 304
+    },
+    {
+      "epoch": 0.6699615595826469,
+      "grad_norm": 0.6005570292472839,
+      "learning_rate": 0.00025895765472312704,
+      "loss": 0.7937,
+      "step": 305
+    },
+    {
+      "epoch": 0.6721581548599671,
+      "grad_norm": 0.6758508682250977,
+      "learning_rate": 0.00025871335504885993,
+      "loss": 0.7669,
+      "step": 306
+    },
+    {
+      "epoch": 0.6743547501372872,
+      "grad_norm": 0.5862293243408203,
+      "learning_rate": 0.0002584690553745928,
+      "loss": 0.8344,
+      "step": 307
+    },
+    {
+      "epoch": 0.6765513454146074,
+      "grad_norm": 0.47516682744026184,
+      "learning_rate": 0.0002582247557003257,
+      "loss": 0.8147,
+      "step": 308
+    },
+    {
+      "epoch": 0.6787479406919276,
+      "grad_norm": 0.6955301761627197,
+      "learning_rate": 0.00025798045602605865,
+      "loss": 0.8716,
+      "step": 309
+    },
+    {
+      "epoch": 0.6809445359692476,
+      "grad_norm": 0.6095973253250122,
+      "learning_rate": 0.00025773615635179154,
+      "loss": 0.875,
+      "step": 310
+    },
+    {
+      "epoch": 0.6831411312465678,
+      "grad_norm": 0.7598492503166199,
+      "learning_rate": 0.00025749185667752443,
+      "loss": 0.7066,
+      "step": 311
+    },
+    {
+      "epoch": 0.685337726523888,
+      "grad_norm": 0.5361849665641785,
+      "learning_rate": 0.0002572475570032573,
+      "loss": 0.8712,
+      "step": 312
+    },
+    {
+      "epoch": 0.6875343218012081,
+      "grad_norm": 0.5668314695358276,
+      "learning_rate": 0.0002570032573289902,
+      "loss": 0.7839,
+      "step": 313
+    },
+    {
+      "epoch": 0.6897309170785283,
+      "grad_norm": 0.49800625443458557,
+      "learning_rate": 0.0002567589576547231,
+      "loss": 0.5503,
+      "step": 314
+    },
+    {
+      "epoch": 0.6919275123558485,
+      "grad_norm": 0.44881466031074524,
+      "learning_rate": 0.00025651465798045604,
+      "loss": 0.5841,
+      "step": 315
+    },
+    {
+      "epoch": 0.6941241076331686,
+      "grad_norm": 0.9642932415008545,
+      "learning_rate": 0.00025627035830618893,
+      "loss": 0.9029,
+      "step": 316
+    },
+    {
+      "epoch": 0.6963207029104888,
+      "grad_norm": 0.5519043803215027,
+      "learning_rate": 0.0002560260586319218,
+      "loss": 0.9195,
+      "step": 317
+    },
+    {
+      "epoch": 0.6985172981878089,
+      "grad_norm": 0.5313773155212402,
+      "learning_rate": 0.0002557817589576547,
+      "loss": 0.8969,
+      "step": 318
+    },
+    {
+      "epoch": 0.700713893465129,
+      "grad_norm": 0.5429991483688354,
+      "learning_rate": 0.0002555374592833876,
+      "loss": 0.7625,
+      "step": 319
+    },
+    {
+      "epoch": 0.7029104887424492,
+      "grad_norm": 0.4437033236026764,
+      "learning_rate": 0.0002552931596091205,
+      "loss": 0.7814,
+      "step": 320
+    },
+    {
+      "epoch": 0.7051070840197694,
+      "grad_norm": 0.7429347038269043,
+      "learning_rate": 0.00025504885993485343,
+      "loss": 0.8139,
+      "step": 321
+    },
+    {
+      "epoch": 0.7073036792970895,
+      "grad_norm": 0.47756335139274597,
+      "learning_rate": 0.0002548045602605863,
+      "loss": 0.8824,
+      "step": 322
+    },
+    {
+      "epoch": 0.7095002745744097,
+      "grad_norm": 0.5488697290420532,
+      "learning_rate": 0.0002545602605863192,
+      "loss": 0.691,
+      "step": 323
+    },
+    {
+      "epoch": 0.7116968698517299,
+      "grad_norm": 0.5253273844718933,
+      "learning_rate": 0.0002543159609120521,
+      "loss": 0.8753,
+      "step": 324
+    },
+    {
+      "epoch": 0.7138934651290499,
+      "grad_norm": 0.8657964468002319,
+      "learning_rate": 0.000254071661237785,
+      "loss": 0.7422,
+      "step": 325
+    },
+    {
+      "epoch": 0.7160900604063701,
+      "grad_norm": 0.5959983468055725,
+      "learning_rate": 0.00025382736156351793,
+      "loss": 0.7712,
+      "step": 326
+    },
+    {
+      "epoch": 0.7182866556836903,
+      "grad_norm": 0.4230397045612335,
+      "learning_rate": 0.0002535830618892508,
+      "loss": 0.826,
+      "step": 327
+    },
+    {
+      "epoch": 0.7204832509610104,
+      "grad_norm": 0.6883108615875244,
+      "learning_rate": 0.0002533387622149837,
+      "loss": 0.9158,
+      "step": 328
+    },
+    {
+      "epoch": 0.7226798462383306,
+      "grad_norm": 0.5036619901657104,
+      "learning_rate": 0.0002530944625407166,
+      "loss": 0.7856,
+      "step": 329
+    },
+    {
+      "epoch": 0.7248764415156508,
+      "grad_norm": 0.9503098726272583,
+      "learning_rate": 0.0002528501628664495,
+      "loss": 0.7635,
+      "step": 330
+    },
+    {
+      "epoch": 0.7270730367929709,
+      "grad_norm": 0.6271690726280212,
+      "learning_rate": 0.0002526058631921824,
+      "loss": 0.8704,
+      "step": 331
+    },
+    {
+      "epoch": 0.729269632070291,
+      "grad_norm": 1.6658426523208618,
+      "learning_rate": 0.0002523615635179153,
+      "loss": 0.7426,
+      "step": 332
+    },
+    {
+      "epoch": 0.7314662273476112,
+      "grad_norm": 0.5795270204544067,
+      "learning_rate": 0.0002521172638436482,
+      "loss": 0.8309,
+      "step": 333
+    },
+    {
+      "epoch": 0.7336628226249313,
+      "grad_norm": 0.6320841312408447,
+      "learning_rate": 0.0002518729641693811,
+      "loss": 0.8711,
+      "step": 334
+    },
+    {
+      "epoch": 0.7358594179022515,
+      "grad_norm": 0.46587979793548584,
+      "learning_rate": 0.000251628664495114,
+      "loss": 0.803,
+      "step": 335
+    },
+    {
+      "epoch": 0.7380560131795717,
+      "grad_norm": 0.5105106830596924,
+      "learning_rate": 0.0002513843648208469,
+      "loss": 0.7671,
+      "step": 336
+    },
+    {
+      "epoch": 0.7402526084568918,
+      "grad_norm": 0.4844941198825836,
+      "learning_rate": 0.0002511400651465798,
+      "loss": 0.9549,
+      "step": 337
+    },
+    {
+      "epoch": 0.742449203734212,
+      "grad_norm": 0.7415852546691895,
+      "learning_rate": 0.0002508957654723127,
+      "loss": 0.7612,
+      "step": 338
+    },
+    {
+      "epoch": 0.7446457990115322,
+      "grad_norm": 0.7414440512657166,
+      "learning_rate": 0.0002506514657980456,
+      "loss": 0.8088,
+      "step": 339
+    },
+    {
+      "epoch": 0.7468423942888522,
+      "grad_norm": 0.5264499187469482,
+      "learning_rate": 0.0002504071661237785,
+      "loss": 0.8413,
+      "step": 340
+    },
+    {
+      "epoch": 0.7490389895661724,
+      "grad_norm": 0.6200268864631653,
+      "learning_rate": 0.0002501628664495114,
+      "loss": 0.7636,
+      "step": 341
+    },
+    {
+      "epoch": 0.7512355848434926,
+      "grad_norm": 0.45897340774536133,
+      "learning_rate": 0.00024991856677524427,
+      "loss": 0.666,
+      "step": 342
+    },
+    {
+      "epoch": 0.7534321801208127,
+      "grad_norm": 0.47164544463157654,
+      "learning_rate": 0.0002496742671009772,
+      "loss": 0.7407,
+      "step": 343
+    },
+    {
+      "epoch": 0.7556287753981329,
+      "grad_norm": 0.6819187998771667,
+      "learning_rate": 0.0002494299674267101,
+      "loss": 0.7493,
+      "step": 344
+    },
+    {
+      "epoch": 0.7578253706754531,
+      "grad_norm": 6.232176303863525,
+      "learning_rate": 0.000249185667752443,
+      "loss": 0.6493,
+      "step": 345
+    },
+    {
+      "epoch": 0.7600219659527732,
+      "grad_norm": 0.8008776307106018,
+      "learning_rate": 0.0002489413680781759,
+      "loss": 0.7898,
+      "step": 346
+    },
+    {
+      "epoch": 0.7622185612300933,
+      "grad_norm": 0.5185206532478333,
+      "learning_rate": 0.00024869706840390877,
+      "loss": 0.8283,
+      "step": 347
+    },
+    {
+      "epoch": 0.7644151565074135,
+      "grad_norm": 0.5190396904945374,
+      "learning_rate": 0.0002484527687296417,
+      "loss": 0.9191,
+      "step": 348
+    },
+    {
+      "epoch": 0.7666117517847336,
+      "grad_norm": 0.6393092274665833,
+      "learning_rate": 0.0002482084690553746,
+      "loss": 0.5842,
+      "step": 349
+    },
+    {
+      "epoch": 0.7688083470620538,
+      "grad_norm": 0.6160006523132324,
+      "learning_rate": 0.0002479641693811075,
+      "loss": 0.938,
+      "step": 350
+    },
+    {
+      "epoch": 0.771004942339374,
+      "grad_norm": 0.8104996085166931,
+      "learning_rate": 0.0002477198697068404,
+      "loss": 0.9992,
+      "step": 351
+    },
+    {
+      "epoch": 0.7732015376166941,
+      "grad_norm": 0.6645625233650208,
+      "learning_rate": 0.00024747557003257327,
+      "loss": 0.832,
+      "step": 352
+    },
+    {
+      "epoch": 0.7753981328940143,
+      "grad_norm": 0.5135954022407532,
+      "learning_rate": 0.00024723127035830616,
+      "loss": 0.6186,
+      "step": 353
+    },
+    {
+      "epoch": 0.7775947281713345,
+      "grad_norm": 0.59861159324646,
+      "learning_rate": 0.0002469869706840391,
+      "loss": 0.8256,
+      "step": 354
+    },
+    {
+      "epoch": 0.7797913234486545,
+      "grad_norm": 0.7403861284255981,
+      "learning_rate": 0.000246742671009772,
+      "loss": 0.6046,
+      "step": 355
+    },
+    {
+      "epoch": 0.7819879187259747,
+      "grad_norm": 0.42505377531051636,
+      "learning_rate": 0.0002464983713355049,
+      "loss": 0.6309,
+      "step": 356
+    },
+    {
+      "epoch": 0.7841845140032949,
+      "grad_norm": 0.40494921803474426,
+      "learning_rate": 0.00024625407166123777,
+      "loss": 0.773,
+      "step": 357
+    },
+    {
+      "epoch": 0.786381109280615,
+      "grad_norm": 0.40009498596191406,
+      "learning_rate": 0.00024600977198697066,
+      "loss": 0.7652,
+      "step": 358
+    },
+    {
+      "epoch": 0.7885777045579352,
+      "grad_norm": 0.6804608106613159,
+      "learning_rate": 0.00024576547231270354,
+      "loss": 0.6911,
+      "step": 359
+    },
+    {
+      "epoch": 0.7907742998352554,
+      "grad_norm": 0.3816309869289398,
+      "learning_rate": 0.0002455211726384365,
+      "loss": 0.6565,
+      "step": 360
+    },
+    {
+      "epoch": 0.7929708951125755,
+      "grad_norm": 1.9469501972198486,
+      "learning_rate": 0.0002452768729641694,
+      "loss": 0.8808,
+      "step": 361
+    },
+    {
+      "epoch": 0.7951674903898956,
+      "grad_norm": 0.5728010535240173,
+      "learning_rate": 0.00024503257328990227,
+      "loss": 0.8507,
+      "step": 362
+    },
+    {
+      "epoch": 0.7973640856672158,
+      "grad_norm": 0.46572014689445496,
+      "learning_rate": 0.00024478827361563516,
+      "loss": 0.6526,
+      "step": 363
+    },
+    {
+      "epoch": 0.7995606809445359,
+      "grad_norm": 0.5280835628509521,
+      "learning_rate": 0.00024454397394136804,
+      "loss": 0.7206,
+      "step": 364
+    },
+    {
+      "epoch": 0.8017572762218561,
+      "grad_norm": 0.5204635262489319,
+      "learning_rate": 0.000244299674267101,
+      "loss": 0.5554,
+      "step": 365
+    },
+    {
+      "epoch": 0.8039538714991763,
+      "grad_norm": 0.6064555048942566,
+      "learning_rate": 0.00024405537459283385,
+      "loss": 0.8607,
+      "step": 366
+    },
+    {
+      "epoch": 0.8061504667764964,
+      "grad_norm": 1.166086196899414,
+      "learning_rate": 0.00024381107491856677,
+      "loss": 0.7672,
+      "step": 367
+    },
+    {
+      "epoch": 0.8083470620538166,
+      "grad_norm": 0.6400506496429443,
+      "learning_rate": 0.00024356677524429966,
+      "loss": 0.6063,
+      "step": 368
+    },
+    {
+      "epoch": 0.8105436573311368,
+      "grad_norm": 0.5417277216911316,
+      "learning_rate": 0.00024332247557003257,
+      "loss": 0.707,
+      "step": 369
+    },
+    {
+      "epoch": 0.8127402526084568,
+      "grad_norm": 0.3942015469074249,
+      "learning_rate": 0.00024307817589576546,
+      "loss": 0.5793,
+      "step": 370
+    },
+    {
+      "epoch": 0.814936847885777,
+      "grad_norm": 1.4134067296981812,
+      "learning_rate": 0.00024283387622149835,
+      "loss": 0.8865,
+      "step": 371
+    },
+    {
+      "epoch": 0.8171334431630972,
+      "grad_norm": 0.8390003442764282,
+      "learning_rate": 0.00024258957654723127,
+      "loss": 0.8035,
+      "step": 372
+    },
+    {
+      "epoch": 0.8193300384404174,
+      "grad_norm": 0.6087154746055603,
+      "learning_rate": 0.00024234527687296416,
+      "loss": 0.7737,
+      "step": 373
+    },
+    {
+      "epoch": 0.8215266337177375,
+      "grad_norm": 0.5439601540565491,
+      "learning_rate": 0.00024210097719869705,
+      "loss": 0.7504,
+      "step": 374
+    },
+    {
+      "epoch": 0.8237232289950577,
+      "grad_norm": 0.5487961173057556,
+      "learning_rate": 0.00024185667752442996,
+      "loss": 0.8824,
+      "step": 375
+    },
+    {
+      "epoch": 0.8259198242723779,
+      "grad_norm": 0.5056906938552856,
+      "learning_rate": 0.00024161237785016285,
+      "loss": 0.8443,
+      "step": 376
+    },
+    {
+      "epoch": 0.828116419549698,
+      "grad_norm": 0.7600564956665039,
+      "learning_rate": 0.00024136807817589574,
+      "loss": 0.8511,
+      "step": 377
+    },
+    {
+      "epoch": 0.8303130148270181,
+      "grad_norm": 0.5692034363746643,
+      "learning_rate": 0.00024112377850162866,
+      "loss": 0.7646,
+      "step": 378
+    },
+    {
+      "epoch": 0.8325096101043383,
+      "grad_norm": 0.5418962836265564,
+      "learning_rate": 0.00024087947882736155,
+      "loss": 0.5794,
+      "step": 379
+    },
+    {
+      "epoch": 0.8347062053816584,
+      "grad_norm": 0.7694405317306519,
+      "learning_rate": 0.00024063517915309446,
+      "loss": 0.8226,
+      "step": 380
+    },
+    {
+      "epoch": 0.8369028006589786,
+      "grad_norm": 0.7077735662460327,
+      "learning_rate": 0.00024039087947882735,
+      "loss": 0.7818,
+      "step": 381
+    },
+    {
+      "epoch": 0.8390993959362988,
+      "grad_norm": 0.49665164947509766,
+      "learning_rate": 0.00024014657980456024,
+      "loss": 0.8318,
+      "step": 382
+    },
+    {
+      "epoch": 0.8412959912136189,
+      "grad_norm": 0.4180140495300293,
+      "learning_rate": 0.00023990228013029316,
+      "loss": 0.8277,
+      "step": 383
+    },
+    {
+      "epoch": 0.8434925864909391,
+      "grad_norm": 1.4263370037078857,
+      "learning_rate": 0.00023965798045602605,
+      "loss": 0.8387,
+      "step": 384
+    },
+    {
+      "epoch": 0.8456891817682592,
+      "grad_norm": 0.7930571436882019,
+      "learning_rate": 0.00023941368078175893,
+      "loss": 0.9531,
+      "step": 385
+    },
+    {
+      "epoch": 0.8478857770455793,
+      "grad_norm": 0.6997365951538086,
+      "learning_rate": 0.00023916938110749185,
+      "loss": 0.8889,
+      "step": 386
+    },
+    {
+      "epoch": 0.8500823723228995,
+      "grad_norm": 0.522689700126648,
+      "learning_rate": 0.00023892508143322474,
+      "loss": 0.7695,
+      "step": 387
+    },
+    {
+      "epoch": 0.8522789676002197,
+      "grad_norm": 0.5728150606155396,
+      "learning_rate": 0.00023868078175895763,
+      "loss": 0.8406,
+      "step": 388
+    },
+    {
+      "epoch": 0.8544755628775398,
+      "grad_norm": 0.44824403524398804,
+      "learning_rate": 0.00023843648208469055,
+      "loss": 0.5965,
+      "step": 389
+    },
+    {
+      "epoch": 0.85667215815486,
+      "grad_norm": 0.5365848541259766,
+      "learning_rate": 0.00023819218241042343,
+      "loss": 0.7078,
+      "step": 390
+    },
+    {
+      "epoch": 0.8588687534321802,
+      "grad_norm": 0.8644400835037231,
+      "learning_rate": 0.00023794788273615632,
+      "loss": 0.7686,
+      "step": 391
+    },
+    {
+      "epoch": 0.8610653487095002,
+      "grad_norm": 0.5399034023284912,
+      "learning_rate": 0.00023770358306188924,
+      "loss": 1.0111,
+      "step": 392
+    },
+    {
+      "epoch": 0.8632619439868204,
+      "grad_norm": 1.0570623874664307,
+      "learning_rate": 0.00023745928338762213,
+      "loss": 1.0995,
+      "step": 393
+    },
+    {
+      "epoch": 0.8654585392641406,
+      "grad_norm": 0.4918040931224823,
+      "learning_rate": 0.00023721498371335505,
+      "loss": 0.7552,
+      "step": 394
+    },
+    {
+      "epoch": 0.8676551345414607,
+      "grad_norm": 0.6023840308189392,
+      "learning_rate": 0.00023697068403908794,
+      "loss": 0.7403,
+      "step": 395
+    },
+    {
+      "epoch": 0.8698517298187809,
+      "grad_norm": 0.45580849051475525,
+      "learning_rate": 0.00023672638436482082,
+      "loss": 0.7915,
+      "step": 396
+    },
+    {
+      "epoch": 0.8720483250961011,
+      "grad_norm": 0.9017300009727478,
+      "learning_rate": 0.00023648208469055374,
+      "loss": 0.7842,
+      "step": 397
+    },
+    {
+      "epoch": 0.8742449203734212,
+      "grad_norm": 0.4300403594970703,
+      "learning_rate": 0.00023623778501628663,
+      "loss": 0.7057,
+      "step": 398
+    },
+    {
+      "epoch": 0.8764415156507414,
+      "grad_norm": 0.8209181427955627,
+      "learning_rate": 0.00023599348534201952,
+      "loss": 0.7975,
+      "step": 399
+    },
+    {
+      "epoch": 0.8786381109280615,
+      "grad_norm": 0.5315681099891663,
+      "learning_rate": 0.00023574918566775244,
+      "loss": 0.8406,
+      "step": 400
+    },
+    {
+      "epoch": 0.8808347062053816,
+      "grad_norm": 0.5124934911727905,
+      "learning_rate": 0.00023550488599348532,
+      "loss": 0.6251,
+      "step": 401
+    },
+    {
+      "epoch": 0.8830313014827018,
+      "grad_norm": 0.6417288184165955,
+      "learning_rate": 0.00023526058631921821,
+      "loss": 0.7816,
+      "step": 402
+    },
+    {
+      "epoch": 0.885227896760022,
+      "grad_norm": 0.536501407623291,
+      "learning_rate": 0.00023501628664495113,
+      "loss": 0.9187,
+      "step": 403
+    },
+    {
+      "epoch": 0.8874244920373421,
+      "grad_norm": 0.3684980571269989,
+      "learning_rate": 0.00023477198697068402,
+      "loss": 0.9446,
+      "step": 404
+    },
+    {
+      "epoch": 0.8896210873146623,
+      "grad_norm": 0.6072811484336853,
+      "learning_rate": 0.0002345276872964169,
+      "loss": 0.6612,
+      "step": 405
+    },
+    {
+      "epoch": 0.8918176825919825,
+      "grad_norm": 0.4615521728992462,
+      "learning_rate": 0.00023428338762214982,
+      "loss": 0.7011,
+      "step": 406
+    },
+    {
+      "epoch": 0.8940142778693025,
+      "grad_norm": 0.5641044974327087,
+      "learning_rate": 0.00023403908794788271,
+      "loss": 0.6606,
+      "step": 407
+    },
+    {
+      "epoch": 0.8962108731466227,
+      "grad_norm": 0.5180623531341553,
+      "learning_rate": 0.00023379478827361563,
+      "loss": 0.6418,
+      "step": 408
+    },
+    {
+      "epoch": 0.8984074684239429,
+      "grad_norm": 0.3889411687850952,
+      "learning_rate": 0.00023355048859934852,
+      "loss": 0.7072,
+      "step": 409
+    },
+    {
+      "epoch": 0.900604063701263,
+      "grad_norm": 0.6386185884475708,
+      "learning_rate": 0.0002333061889250814,
+      "loss": 0.9109,
+      "step": 410
+    },
+    {
+      "epoch": 0.9028006589785832,
+      "grad_norm": 0.4829352796077728,
+      "learning_rate": 0.00023306188925081432,
+      "loss": 0.6574,
+      "step": 411
+    },
+    {
+      "epoch": 0.9049972542559034,
+      "grad_norm": 0.5586270689964294,
+      "learning_rate": 0.00023281758957654721,
+      "loss": 0.697,
+      "step": 412
+    },
+    {
+      "epoch": 0.9071938495332235,
+      "grad_norm": 0.48839375376701355,
+      "learning_rate": 0.0002325732899022801,
+      "loss": 0.7498,
+      "step": 413
+    },
+    {
+      "epoch": 0.9093904448105437,
+      "grad_norm": 1.1973849534988403,
+      "learning_rate": 0.00023232899022801302,
+      "loss": 0.8829,
+      "step": 414
+    },
+    {
+      "epoch": 0.9115870400878638,
+      "grad_norm": 0.6850463151931763,
+      "learning_rate": 0.0002320846905537459,
+      "loss": 0.7429,
+      "step": 415
+    },
+    {
+      "epoch": 0.9137836353651839,
+      "grad_norm": 1.0142226219177246,
+      "learning_rate": 0.0002318403908794788,
+      "loss": 0.8662,
+      "step": 416
+    },
+    {
+      "epoch": 0.9159802306425041,
+      "grad_norm": 0.5065467357635498,
+      "learning_rate": 0.00023159609120521171,
+      "loss": 0.7851,
+      "step": 417
+    },
+    {
+      "epoch": 0.9181768259198243,
+      "grad_norm": 0.973021388053894,
+      "learning_rate": 0.0002313517915309446,
+      "loss": 0.5521,
+      "step": 418
+    },
+    {
+      "epoch": 0.9203734211971444,
+      "grad_norm": 0.8830389976501465,
+      "learning_rate": 0.00023110749185667752,
+      "loss": 0.6607,
+      "step": 419
+    },
+    {
+      "epoch": 0.9225700164744646,
+      "grad_norm": 0.46159791946411133,
+      "learning_rate": 0.0002308631921824104,
+      "loss": 0.8189,
+      "step": 420
+    },
+    {
+      "epoch": 0.9247666117517848,
+      "grad_norm": 0.6138774752616882,
+      "learning_rate": 0.0002306188925081433,
+      "loss": 0.7716,
+      "step": 421
+    },
+    {
+      "epoch": 0.9269632070291048,
+      "grad_norm": 0.5318027138710022,
+      "learning_rate": 0.00023037459283387621,
+      "loss": 0.7454,
+      "step": 422
+    },
+    {
+      "epoch": 0.929159802306425,
+      "grad_norm": 0.6046528816223145,
+      "learning_rate": 0.0002301302931596091,
+      "loss": 0.9802,
+      "step": 423
+    },
+    {
+      "epoch": 0.9313563975837452,
+      "grad_norm": 0.531801700592041,
+      "learning_rate": 0.000229885993485342,
+      "loss": 0.763,
+      "step": 424
+    },
+    {
+      "epoch": 0.9335529928610653,
+      "grad_norm": 0.4280356466770172,
+      "learning_rate": 0.0002296416938110749,
+      "loss": 0.6956,
+      "step": 425
+    },
+    {
+      "epoch": 0.9357495881383855,
+      "grad_norm": 0.5393470525741577,
+      "learning_rate": 0.0002293973941368078,
+      "loss": 0.6899,
+      "step": 426
+    },
+    {
+      "epoch": 0.9379461834157057,
+      "grad_norm": 0.3521820604801178,
+      "learning_rate": 0.0002291530944625407,
+      "loss": 0.7282,
+      "step": 427
+    },
+    {
+      "epoch": 0.9401427786930258,
+      "grad_norm": 0.6252460479736328,
+      "learning_rate": 0.0002289087947882736,
+      "loss": 0.6041,
+      "step": 428
+    },
+    {
+      "epoch": 0.942339373970346,
+      "grad_norm": 0.46751296520233154,
+      "learning_rate": 0.0002286644951140065,
+      "loss": 0.5458,
+      "step": 429
+    },
+    {
+      "epoch": 0.9445359692476661,
+      "grad_norm": 0.4605407416820526,
+      "learning_rate": 0.00022842019543973938,
+      "loss": 0.8372,
+      "step": 430
+    },
+    {
+      "epoch": 0.9467325645249862,
+      "grad_norm": 0.618071436882019,
+      "learning_rate": 0.0002281758957654723,
+      "loss": 0.6297,
+      "step": 431
+    },
+    {
+      "epoch": 0.9489291598023064,
+      "grad_norm": 0.8186476826667786,
+      "learning_rate": 0.0002279315960912052,
+      "loss": 0.8972,
+      "step": 432
+    },
+    {
+      "epoch": 0.9511257550796266,
+      "grad_norm": 0.7547953128814697,
+      "learning_rate": 0.0002276872964169381,
+      "loss": 0.8109,
+      "step": 433
+    },
+    {
+      "epoch": 0.9533223503569467,
+      "grad_norm": 0.705167829990387,
+      "learning_rate": 0.000227442996742671,
+      "loss": 0.6578,
+      "step": 434
+    },
+    {
+      "epoch": 0.9555189456342669,
+      "grad_norm": 0.3530736565589905,
+      "learning_rate": 0.00022719869706840388,
+      "loss": 0.6983,
+      "step": 435
+    },
+    {
+      "epoch": 0.9577155409115871,
+      "grad_norm": 0.552609920501709,
+      "learning_rate": 0.0002269543973941368,
+      "loss": 0.6008,
+      "step": 436
+    },
+    {
+      "epoch": 0.9599121361889071,
+      "grad_norm": 0.6262020468711853,
+      "learning_rate": 0.0002267100977198697,
+      "loss": 0.778,
+      "step": 437
+    },
+    {
+      "epoch": 0.9621087314662273,
+      "grad_norm": 0.5063844323158264,
+      "learning_rate": 0.00022646579804560258,
+      "loss": 0.6075,
+      "step": 438
+    },
+    {
+      "epoch": 0.9643053267435475,
+      "grad_norm": 0.4671212434768677,
+      "learning_rate": 0.0002262214983713355,
+      "loss": 1.0387,
+      "step": 439
+    },
+    {
+      "epoch": 0.9665019220208677,
+      "grad_norm": 0.817815899848938,
+      "learning_rate": 0.00022597719869706838,
+      "loss": 0.8517,
+      "step": 440
+    },
+    {
+      "epoch": 0.9686985172981878,
+      "grad_norm": 0.5379441976547241,
+      "learning_rate": 0.00022573289902280127,
+      "loss": 0.6717,
+      "step": 441
+    },
+    {
+      "epoch": 0.970895112575508,
+      "grad_norm": 0.5827528834342957,
+      "learning_rate": 0.0002254885993485342,
+      "loss": 0.699,
+      "step": 442
+    },
+    {
+      "epoch": 0.9730917078528282,
+      "grad_norm": 0.5454041361808777,
+      "learning_rate": 0.00022524429967426708,
+      "loss": 0.7954,
+      "step": 443
+    },
+    {
+      "epoch": 0.9752883031301482,
+      "grad_norm": 0.5619916319847107,
+      "learning_rate": 0.000225,
+      "loss": 0.9038,
+      "step": 444
+    },
+    {
+      "epoch": 0.9774848984074684,
+      "grad_norm": 0.7569581270217896,
+      "learning_rate": 0.00022475570032573288,
+      "loss": 0.7545,
+      "step": 445
+    },
+    {
+      "epoch": 0.9796814936847886,
+      "grad_norm": 1.7369670867919922,
+      "learning_rate": 0.00022451140065146577,
+      "loss": 0.9441,
+      "step": 446
+    },
+    {
+      "epoch": 0.9818780889621087,
+      "grad_norm": 0.8345696330070496,
+      "learning_rate": 0.0002242671009771987,
+      "loss": 0.9165,
+      "step": 447
+    },
+    {
+      "epoch": 0.9840746842394289,
+      "grad_norm": 0.4759431481361389,
+      "learning_rate": 0.00022402280130293158,
+      "loss": 0.7465,
+      "step": 448
+    },
+    {
+      "epoch": 0.9862712795167491,
+      "grad_norm": 0.42961016297340393,
+      "learning_rate": 0.00022377850162866447,
+      "loss": 0.8212,
+      "step": 449
+    },
+    {
+      "epoch": 0.9884678747940692,
+      "grad_norm": 0.9614746570587158,
+      "learning_rate": 0.00022353420195439738,
+      "loss": 0.724,
+      "step": 450
+    },
+    {
+      "epoch": 0.9906644700713894,
+      "grad_norm": 0.6232669949531555,
+      "learning_rate": 0.00022328990228013027,
+      "loss": 0.756,
+      "step": 451
+    },
+    {
+      "epoch": 0.9928610653487095,
+      "grad_norm": 1.2444524765014648,
+      "learning_rate": 0.00022304560260586316,
+      "loss": 0.8468,
+      "step": 452
+    },
+    {
+      "epoch": 0.9950576606260296,
+      "grad_norm": 0.4964475929737091,
+      "learning_rate": 0.00022280130293159608,
+      "loss": 0.8454,
+      "step": 453
+    },
+    {
+      "epoch": 0.9972542559033498,
+      "grad_norm": 0.5804954171180725,
+      "learning_rate": 0.00022255700325732897,
+      "loss": 0.6596,
+      "step": 454
+    },
+    {
+      "epoch": 0.99945085118067,
+      "grad_norm": 0.3958580493927002,
+      "learning_rate": 0.00022231270358306186,
+      "loss": 0.7288,
+      "step": 455
+    },
+    {
+      "epoch": 1.00164744645799,
+      "grad_norm": 0.5487098693847656,
+      "learning_rate": 0.00022206840390879477,
+      "loss": 0.8433,
+      "step": 456
+    },
+    {
+      "epoch": 1.0038440417353103,
+      "grad_norm": 0.4553420841693878,
+      "learning_rate": 0.00022182410423452766,
+      "loss": 0.791,
+      "step": 457
+    },
+    {
+      "epoch": 1.0060406370126305,
+      "grad_norm": 0.8235368728637695,
+      "learning_rate": 0.00022157980456026058,
+      "loss": 0.7682,
+      "step": 458
+    },
+    {
+      "epoch": 1.0082372322899507,
+      "grad_norm": 0.7421154975891113,
+      "learning_rate": 0.00022133550488599347,
+      "loss": 0.6993,
+      "step": 459
+    },
+    {
+      "epoch": 1.0104338275672706,
+      "grad_norm": 0.42016327381134033,
+      "learning_rate": 0.00022109120521172636,
+      "loss": 0.7066,
+      "step": 460
+    },
+    {
+      "epoch": 1.0126304228445908,
+      "grad_norm": 0.8311179280281067,
+      "learning_rate": 0.00022084690553745927,
+      "loss": 0.6105,
+      "step": 461
+    },
+    {
+      "epoch": 1.014827018121911,
+      "grad_norm": 0.8407668471336365,
+      "learning_rate": 0.00022060260586319216,
+      "loss": 1.0297,
+      "step": 462
+    },
+    {
+      "epoch": 1.0170236133992312,
+      "grad_norm": 0.6933571696281433,
+      "learning_rate": 0.00022035830618892505,
+      "loss": 0.9122,
+      "step": 463
+    },
+    {
+      "epoch": 1.0192202086765514,
+      "grad_norm": 0.5876277089118958,
+      "learning_rate": 0.00022011400651465797,
+      "loss": 0.7067,
+      "step": 464
+    },
+    {
+      "epoch": 1.0214168039538716,
+      "grad_norm": 0.6828694939613342,
+      "learning_rate": 0.00021986970684039086,
+      "loss": 0.922,
+      "step": 465
+    },
+    {
+      "epoch": 1.0236133992311915,
+      "grad_norm": 0.4986603856086731,
+      "learning_rate": 0.00021962540716612375,
+      "loss": 0.9267,
+      "step": 466
+    },
+    {
+      "epoch": 1.0258099945085117,
+      "grad_norm": 0.4752190411090851,
+      "learning_rate": 0.00021938110749185666,
+      "loss": 0.7755,
+      "step": 467
+    },
+    {
+      "epoch": 1.028006589785832,
+      "grad_norm": 0.994474470615387,
+      "learning_rate": 0.00021913680781758955,
+      "loss": 0.662,
+      "step": 468
+    },
+    {
+      "epoch": 1.0302031850631521,
+      "grad_norm": 0.5271409749984741,
+      "learning_rate": 0.00021889250814332244,
+      "loss": 0.8835,
+      "step": 469
+    },
+    {
+      "epoch": 1.0323997803404723,
+      "grad_norm": 0.6310428380966187,
+      "learning_rate": 0.00021864820846905536,
+      "loss": 0.8573,
+      "step": 470
+    },
+    {
+      "epoch": 1.0345963756177925,
+      "grad_norm": 0.4058026373386383,
+      "learning_rate": 0.00021840390879478825,
+      "loss": 0.604,
+      "step": 471
+    },
+    {
+      "epoch": 1.0367929708951125,
+      "grad_norm": 0.5298481583595276,
+      "learning_rate": 0.00021815960912052116,
+      "loss": 0.722,
+      "step": 472
+    },
+    {
+      "epoch": 1.0389895661724327,
+      "grad_norm": 0.6016893982887268,
+      "learning_rate": 0.00021791530944625405,
+      "loss": 0.9334,
+      "step": 473
+    },
+    {
+      "epoch": 1.0411861614497528,
+      "grad_norm": 0.4841894209384918,
+      "learning_rate": 0.00021767100977198694,
+      "loss": 0.7503,
+      "step": 474
+    },
+    {
+      "epoch": 1.043382756727073,
+      "grad_norm": 0.42801499366760254,
+      "learning_rate": 0.00021742671009771986,
+      "loss": 0.7223,
+      "step": 475
+    },
+    {
+      "epoch": 1.0455793520043932,
+      "grad_norm": 0.43856412172317505,
+      "learning_rate": 0.00021718241042345275,
+      "loss": 1.0442,
+      "step": 476
+    },
+    {
+      "epoch": 1.0477759472817134,
+      "grad_norm": 0.6297312378883362,
+      "learning_rate": 0.00021693811074918564,
+      "loss": 0.6824,
+      "step": 477
+    },
+    {
+      "epoch": 1.0499725425590336,
+      "grad_norm": 0.42614036798477173,
+      "learning_rate": 0.00021669381107491855,
+      "loss": 0.7945,
+      "step": 478
+    },
+    {
+      "epoch": 1.0521691378363536,
+      "grad_norm": 0.5742759704589844,
+      "learning_rate": 0.00021644951140065144,
+      "loss": 0.7868,
+      "step": 479
+    },
+    {
+      "epoch": 1.0543657331136738,
+      "grad_norm": 0.540634274482727,
+      "learning_rate": 0.00021620521172638433,
+      "loss": 0.6786,
+      "step": 480
+    },
+    {
+      "epoch": 1.056562328390994,
+      "grad_norm": 0.35537514090538025,
+      "learning_rate": 0.00021596091205211725,
+      "loss": 0.7075,
+      "step": 481
+    },
+    {
+      "epoch": 1.0587589236683141,
+      "grad_norm": 0.4580304026603699,
+      "learning_rate": 0.00021571661237785014,
+      "loss": 0.6953,
+      "step": 482
+    },
+    {
+      "epoch": 1.0609555189456343,
+      "grad_norm": 0.9031398296356201,
+      "learning_rate": 0.00021547231270358305,
+      "loss": 0.6817,
+      "step": 483
+    },
+    {
+      "epoch": 1.0631521142229543,
+      "grad_norm": 0.3687251806259155,
+      "learning_rate": 0.00021522801302931594,
+      "loss": 0.7513,
+      "step": 484
+    },
+    {
+      "epoch": 1.0653487095002745,
+      "grad_norm": 0.3983041048049927,
+      "learning_rate": 0.00021498371335504883,
+      "loss": 0.7525,
+      "step": 485
+    },
+    {
+      "epoch": 1.0675453047775947,
+      "grad_norm": 0.8923289179801941,
+      "learning_rate": 0.00021473941368078175,
+      "loss": 0.647,
+      "step": 486
+    },
+    {
+      "epoch": 1.0697419000549149,
+      "grad_norm": 0.45228809118270874,
+      "learning_rate": 0.00021449511400651464,
+      "loss": 0.7473,
+      "step": 487
+    },
+    {
+      "epoch": 1.071938495332235,
+      "grad_norm": 0.49838489294052124,
+      "learning_rate": 0.00021425081433224753,
+      "loss": 0.8418,
+      "step": 488
+    },
+    {
+      "epoch": 1.0741350906095553,
+      "grad_norm": 1.0038297176361084,
+      "learning_rate": 0.00021400651465798044,
+      "loss": 0.6535,
+      "step": 489
+    },
+    {
+      "epoch": 1.0763316858868754,
+      "grad_norm": 0.3770967423915863,
+      "learning_rate": 0.00021376221498371333,
+      "loss": 0.6617,
+      "step": 490
+    },
+    {
+      "epoch": 1.0785282811641954,
+      "grad_norm": 0.5993874073028564,
+      "learning_rate": 0.00021351791530944622,
+      "loss": 0.973,
+      "step": 491
+    },
+    {
+      "epoch": 1.0807248764415156,
+      "grad_norm": 0.42255812883377075,
+      "learning_rate": 0.00021327361563517914,
+      "loss": 0.6458,
+      "step": 492
+    },
+    {
+      "epoch": 1.0829214717188358,
+      "grad_norm": 1.043294072151184,
+      "learning_rate": 0.00021302931596091203,
+      "loss": 0.8081,
+      "step": 493
+    },
+    {
+      "epoch": 1.085118066996156,
+      "grad_norm": 0.4316386878490448,
+      "learning_rate": 0.00021278501628664491,
+      "loss": 0.6502,
+      "step": 494
+    },
+    {
+      "epoch": 1.0873146622734762,
+      "grad_norm": 0.7171972990036011,
+      "learning_rate": 0.00021254071661237783,
+      "loss": 0.7644,
+      "step": 495
+    },
+    {
+      "epoch": 1.0895112575507964,
+      "grad_norm": 0.37957897782325745,
+      "learning_rate": 0.00021229641693811072,
+      "loss": 0.643,
+      "step": 496
+    },
+    {
+      "epoch": 1.0917078528281163,
+      "grad_norm": 0.38239768147468567,
+      "learning_rate": 0.00021205211726384364,
+      "loss": 0.6676,
+      "step": 497
+    },
+    {
+      "epoch": 1.0939044481054365,
+      "grad_norm": 0.7416626811027527,
+      "learning_rate": 0.00021180781758957653,
+      "loss": 0.7294,
+      "step": 498
+    },
+    {
+      "epoch": 1.0961010433827567,
+      "grad_norm": 0.753420889377594,
+      "learning_rate": 0.00021156351791530941,
+      "loss": 0.7764,
+      "step": 499
+    },
+    {
+      "epoch": 1.098297638660077,
+      "grad_norm": 0.5058073401451111,
+      "learning_rate": 0.00021131921824104233,
+      "loss": 0.8209,
+      "step": 500
+    },
+    {
+      "epoch": 1.100494233937397,
+      "grad_norm": 0.49321016669273376,
+      "learning_rate": 0.00021107491856677522,
+      "loss": 0.9045,
+      "step": 501
+    },
+    {
+      "epoch": 1.1026908292147173,
+      "grad_norm": 0.38129690289497375,
+      "learning_rate": 0.0002108306188925081,
+      "loss": 0.6446,
+      "step": 502
+    },
+    {
+      "epoch": 1.1048874244920373,
+      "grad_norm": 0.6222732663154602,
+      "learning_rate": 0.00021058631921824103,
+      "loss": 0.8108,
+      "step": 503
+    },
+    {
+      "epoch": 1.1070840197693574,
+      "grad_norm": 0.4250493049621582,
+      "learning_rate": 0.00021034201954397392,
+      "loss": 0.6714,
+      "step": 504
+    },
+    {
+      "epoch": 1.1092806150466776,
+      "grad_norm": 0.6469359993934631,
+      "learning_rate": 0.0002100977198697068,
+      "loss": 0.9572,
+      "step": 505
+    },
+    {
+      "epoch": 1.1114772103239978,
+      "grad_norm": 0.5700991153717041,
+      "learning_rate": 0.00020985342019543972,
+      "loss": 0.7858,
+      "step": 506
+    },
+    {
+      "epoch": 1.113673805601318,
+      "grad_norm": 0.5541654825210571,
+      "learning_rate": 0.0002096091205211726,
+      "loss": 0.7971,
+      "step": 507
+    },
+    {
+      "epoch": 1.1158704008786382,
+      "grad_norm": 0.7180041670799255,
+      "learning_rate": 0.0002093648208469055,
+      "loss": 0.8488,
+      "step": 508
+    },
+    {
+      "epoch": 1.1180669961559582,
+      "grad_norm": 0.4243312478065491,
+      "learning_rate": 0.00020912052117263842,
+      "loss": 0.7777,
+      "step": 509
+    },
+    {
+      "epoch": 1.1202635914332784,
+      "grad_norm": 0.5100587606430054,
+      "learning_rate": 0.0002088762214983713,
+      "loss": 0.8446,
+      "step": 510
+    },
+    {
+      "epoch": 1.1224601867105986,
+      "grad_norm": 0.5331737399101257,
+      "learning_rate": 0.00020863192182410422,
+      "loss": 0.7195,
+      "step": 511
+    },
+    {
+      "epoch": 1.1246567819879187,
+      "grad_norm": 0.522591769695282,
+      "learning_rate": 0.0002083876221498371,
+      "loss": 0.7428,
+      "step": 512
+    },
+    {
+      "epoch": 1.126853377265239,
+      "grad_norm": 0.567529022693634,
+      "learning_rate": 0.00020814332247557,
+      "loss": 0.8969,
+      "step": 513
+    },
+    {
+      "epoch": 1.1290499725425591,
+      "grad_norm": 0.39562782645225525,
+      "learning_rate": 0.00020789902280130292,
+      "loss": 0.6761,
+      "step": 514
+    },
+    {
+      "epoch": 1.131246567819879,
+      "grad_norm": 0.5828169584274292,
+      "learning_rate": 0.0002076547231270358,
+      "loss": 0.7259,
+      "step": 515
+    },
+    {
+      "epoch": 1.1334431630971993,
+      "grad_norm": 0.5210989117622375,
+      "learning_rate": 0.0002074104234527687,
+      "loss": 0.7666,
+      "step": 516
+    },
+    {
+      "epoch": 1.1356397583745195,
+      "grad_norm": 0.5703113079071045,
+      "learning_rate": 0.0002071661237785016,
+      "loss": 0.7984,
+      "step": 517
+    },
+    {
+      "epoch": 1.1378363536518397,
+      "grad_norm": 0.6043124794960022,
+      "learning_rate": 0.0002069218241042345,
+      "loss": 0.583,
+      "step": 518
+    },
+    {
+      "epoch": 1.1400329489291599,
+      "grad_norm": 0.6745511889457703,
+      "learning_rate": 0.0002066775244299674,
+      "loss": 0.536,
+      "step": 519
+    },
+    {
+      "epoch": 1.14222954420648,
+      "grad_norm": 0.5378568768501282,
+      "learning_rate": 0.0002064332247557003,
+      "loss": 0.7626,
+      "step": 520
+    },
+    {
+      "epoch": 1.1444261394838002,
+      "grad_norm": 0.5234605669975281,
+      "learning_rate": 0.0002061889250814332,
+      "loss": 0.6721,
+      "step": 521
+    },
+    {
+      "epoch": 1.1466227347611202,
+      "grad_norm": 0.45719388127326965,
+      "learning_rate": 0.0002059446254071661,
+      "loss": 0.656,
+      "step": 522
+    },
+    {
+      "epoch": 1.1488193300384404,
+      "grad_norm": 0.39946669340133667,
+      "learning_rate": 0.000205700325732899,
+      "loss": 0.5573,
+      "step": 523
+    },
+    {
+      "epoch": 1.1510159253157606,
+      "grad_norm": 0.4715358316898346,
+      "learning_rate": 0.0002054560260586319,
+      "loss": 0.8552,
+      "step": 524
+    },
+    {
+      "epoch": 1.1532125205930808,
+      "grad_norm": 0.4580067992210388,
+      "learning_rate": 0.0002052117263843648,
+      "loss": 0.7438,
+      "step": 525
+    },
+    {
+      "epoch": 1.155409115870401,
+      "grad_norm": 0.6139888763427734,
+      "learning_rate": 0.0002049674267100977,
+      "loss": 0.9503,
+      "step": 526
+    },
+    {
+      "epoch": 1.157605711147721,
+      "grad_norm": 0.5124037861824036,
+      "learning_rate": 0.00020472312703583058,
+      "loss": 0.6875,
+      "step": 527
+    },
+    {
+      "epoch": 1.1598023064250411,
+      "grad_norm": 0.8412673473358154,
+      "learning_rate": 0.0002044788273615635,
+      "loss": 0.8689,
+      "step": 528
+    },
+    {
+      "epoch": 1.1619989017023613,
+      "grad_norm": 0.45381826162338257,
+      "learning_rate": 0.0002042345276872964,
+      "loss": 0.5982,
+      "step": 529
+    },
+    {
+      "epoch": 1.1641954969796815,
+      "grad_norm": 0.5494648814201355,
+      "learning_rate": 0.00020399022801302928,
+      "loss": 0.548,
+      "step": 530
+    },
+    {
+      "epoch": 1.1663920922570017,
+      "grad_norm": 0.4707123935222626,
+      "learning_rate": 0.0002037459283387622,
+      "loss": 0.6784,
+      "step": 531
+    },
+    {
+      "epoch": 1.1685886875343219,
+      "grad_norm": 0.467019259929657,
+      "learning_rate": 0.00020350162866449508,
+      "loss": 0.824,
+      "step": 532
+    },
+    {
+      "epoch": 1.170785282811642,
+      "grad_norm": 0.517022967338562,
+      "learning_rate": 0.00020325732899022797,
+      "loss": 0.8558,
+      "step": 533
+    },
+    {
+      "epoch": 1.172981878088962,
+      "grad_norm": 0.5522506833076477,
+      "learning_rate": 0.0002030130293159609,
+      "loss": 0.8262,
+      "step": 534
+    },
+    {
+      "epoch": 1.1751784733662822,
+      "grad_norm": 0.4813742935657501,
+      "learning_rate": 0.00020276872964169378,
+      "loss": 0.703,
+      "step": 535
+    },
+    {
+      "epoch": 1.1773750686436024,
+      "grad_norm": 0.5663685202598572,
+      "learning_rate": 0.0002025244299674267,
+      "loss": 0.6641,
+      "step": 536
+    },
+    {
+      "epoch": 1.1795716639209226,
+      "grad_norm": 0.5498743653297424,
+      "learning_rate": 0.00020228013029315958,
+      "loss": 0.7404,
+      "step": 537
+    },
+    {
+      "epoch": 1.1817682591982428,
+      "grad_norm": 0.5944446325302124,
+      "learning_rate": 0.00020203583061889247,
+      "loss": 0.6542,
+      "step": 538
+    },
+    {
+      "epoch": 1.1839648544755628,
+      "grad_norm": 0.5669150948524475,
+      "learning_rate": 0.0002017915309446254,
+      "loss": 0.9698,
+      "step": 539
+    },
+    {
+      "epoch": 1.186161449752883,
+      "grad_norm": 0.4735161066055298,
+      "learning_rate": 0.00020154723127035828,
+      "loss": 0.6446,
+      "step": 540
+    },
+    {
+      "epoch": 1.1883580450302031,
+      "grad_norm": 0.8892743587493896,
+      "learning_rate": 0.00020130293159609117,
+      "loss": 0.7482,
+      "step": 541
+    },
+    {
+      "epoch": 1.1905546403075233,
+      "grad_norm": 0.4790038466453552,
+      "learning_rate": 0.00020105863192182408,
+      "loss": 0.6423,
+      "step": 542
+    },
+    {
+      "epoch": 1.1927512355848435,
+      "grad_norm": 0.5276747345924377,
+      "learning_rate": 0.00020081433224755697,
+      "loss": 0.7769,
+      "step": 543
+    },
+    {
+      "epoch": 1.1949478308621637,
+      "grad_norm": 0.4293566346168518,
+      "learning_rate": 0.00020057003257328986,
+      "loss": 0.7157,
+      "step": 544
+    },
+    {
+      "epoch": 1.197144426139484,
+      "grad_norm": 0.7193005681037903,
+      "learning_rate": 0.00020032573289902278,
+      "loss": 0.8155,
+      "step": 545
+    },
+    {
+      "epoch": 1.1993410214168039,
+      "grad_norm": 1.9112204313278198,
+      "learning_rate": 0.00020008143322475567,
+      "loss": 0.7719,
+      "step": 546
+    },
+    {
+      "epoch": 1.201537616694124,
+      "grad_norm": 0.5370739698410034,
+      "learning_rate": 0.00019983713355048856,
+      "loss": 1.0388,
+      "step": 547
+    },
+    {
+      "epoch": 1.2037342119714443,
+      "grad_norm": 0.39147570729255676,
+      "learning_rate": 0.00019959283387622147,
+      "loss": 0.7383,
+      "step": 548
+    },
+    {
+      "epoch": 1.2059308072487644,
+      "grad_norm": 0.46493563055992126,
+      "learning_rate": 0.00019934853420195436,
+      "loss": 0.9891,
+      "step": 549
+    },
+    {
+      "epoch": 1.2081274025260846,
+      "grad_norm": 0.6021866202354431,
+      "learning_rate": 0.00019910423452768728,
+      "loss": 0.8369,
+      "step": 550
+    },
+    {
+      "epoch": 1.2103239978034046,
+      "grad_norm": 0.5282336473464966,
+      "learning_rate": 0.00019885993485342017,
+      "loss": 0.6927,
+      "step": 551
+    },
+    {
+      "epoch": 1.2125205930807248,
+      "grad_norm": 0.6158313751220703,
+      "learning_rate": 0.00019861563517915306,
+      "loss": 0.8462,
+      "step": 552
+    },
+    {
+      "epoch": 1.214717188358045,
+      "grad_norm": 0.49901267886161804,
+      "learning_rate": 0.00019837133550488597,
+      "loss": 0.8062,
+      "step": 553
+    },
+    {
+      "epoch": 1.2169137836353652,
+      "grad_norm": 0.4741625189781189,
+      "learning_rate": 0.00019812703583061886,
+      "loss": 0.7728,
+      "step": 554
+    },
+    {
+      "epoch": 1.2191103789126854,
+      "grad_norm": 0.5865007638931274,
+      "learning_rate": 0.00019788273615635175,
+      "loss": 0.8021,
+      "step": 555
+    },
+    {
+      "epoch": 1.2213069741900056,
+      "grad_norm": 0.5611051321029663,
+      "learning_rate": 0.00019763843648208467,
+      "loss": 0.6282,
+      "step": 556
+    },
+    {
+      "epoch": 1.2235035694673257,
+      "grad_norm": 0.4477647840976715,
+      "learning_rate": 0.00019739413680781756,
+      "loss": 0.7718,
+      "step": 557
+    },
+    {
+      "epoch": 1.2257001647446457,
+      "grad_norm": 0.5085916519165039,
+      "learning_rate": 0.00019714983713355045,
+      "loss": 0.6999,
+      "step": 558
+    },
+    {
+      "epoch": 1.227896760021966,
+      "grad_norm": 0.9920333027839661,
+      "learning_rate": 0.00019690553745928336,
+      "loss": 0.9375,
+      "step": 559
+    },
+    {
+      "epoch": 1.230093355299286,
+      "grad_norm": 0.41038957238197327,
+      "learning_rate": 0.00019666123778501625,
+      "loss": 0.8176,
+      "step": 560
+    },
+    {
+      "epoch": 1.2322899505766063,
+      "grad_norm": 0.7359328866004944,
+      "learning_rate": 0.00019641693811074917,
+      "loss": 0.7106,
+      "step": 561
+    },
+    {
+      "epoch": 1.2344865458539265,
+      "grad_norm": 0.5618932247161865,
+      "learning_rate": 0.00019617263843648206,
+      "loss": 0.7484,
+      "step": 562
+    },
+    {
+      "epoch": 1.2366831411312464,
+      "grad_norm": 0.611487627029419,
+      "learning_rate": 0.00019592833876221495,
+      "loss": 0.8827,
+      "step": 563
+    },
+    {
+      "epoch": 1.2388797364085666,
+      "grad_norm": 0.5960579514503479,
+      "learning_rate": 0.00019568403908794786,
+      "loss": 0.6847,
+      "step": 564
+    },
+    {
+      "epoch": 1.2410763316858868,
+      "grad_norm": 0.5108380913734436,
+      "learning_rate": 0.00019543973941368075,
+      "loss": 0.82,
+      "step": 565
+    },
+    {
+      "epoch": 1.243272926963207,
+      "grad_norm": 0.6912088394165039,
+      "learning_rate": 0.00019519543973941364,
+      "loss": 0.7286,
+      "step": 566
+    },
+    {
+      "epoch": 1.2454695222405272,
+      "grad_norm": 0.8419110774993896,
+      "learning_rate": 0.00019495114006514656,
+      "loss": 0.7063,
+      "step": 567
+    },
+    {
+      "epoch": 1.2476661175178474,
+      "grad_norm": 0.6587998867034912,
+      "learning_rate": 0.00019470684039087945,
+      "loss": 0.8201,
+      "step": 568
+    },
+    {
+      "epoch": 1.2498627127951676,
+      "grad_norm": 0.5778517127037048,
+      "learning_rate": 0.00019446254071661234,
+      "loss": 0.6399,
+      "step": 569
+    },
+    {
+      "epoch": 1.2520593080724876,
+      "grad_norm": 0.48817864060401917,
+      "learning_rate": 0.00019421824104234525,
+      "loss": 0.5237,
+      "step": 570
+    },
+    {
+      "epoch": 1.2542559033498077,
+      "grad_norm": 0.45017483830451965,
+      "learning_rate": 0.00019397394136807814,
+      "loss": 0.8457,
+      "step": 571
+    },
+    {
+      "epoch": 1.256452498627128,
+      "grad_norm": 0.5680342316627502,
+      "learning_rate": 0.00019372964169381103,
+      "loss": 0.739,
+      "step": 572
+    },
+    {
+      "epoch": 1.2586490939044481,
+      "grad_norm": 0.4731301963329315,
+      "learning_rate": 0.00019348534201954395,
+      "loss": 0.7185,
+      "step": 573
+    },
+    {
+      "epoch": 1.2608456891817683,
+      "grad_norm": 0.7031453847885132,
+      "learning_rate": 0.00019324104234527684,
+      "loss": 0.8136,
+      "step": 574
+    },
+    {
+      "epoch": 1.2630422844590883,
+      "grad_norm": 0.816160261631012,
+      "learning_rate": 0.00019299674267100975,
+      "loss": 0.876,
+      "step": 575
+    },
+    {
+      "epoch": 1.2652388797364087,
+      "grad_norm": 1.737231969833374,
+      "learning_rate": 0.00019275244299674264,
+      "loss": 0.6882,
+      "step": 576
+    },
+    {
+      "epoch": 1.2674354750137287,
+      "grad_norm": 0.5609646439552307,
+      "learning_rate": 0.00019250814332247553,
+      "loss": 0.9445,
+      "step": 577
+    },
+    {
+      "epoch": 1.2696320702910489,
+      "grad_norm": 0.5112996697425842,
+      "learning_rate": 0.00019226384364820845,
+      "loss": 0.5572,
+      "step": 578
+    },
+    {
+      "epoch": 1.271828665568369,
+      "grad_norm": 0.5386213064193726,
+      "learning_rate": 0.00019201954397394134,
+      "loss": 0.5684,
+      "step": 579
+    },
+    {
+      "epoch": 1.2740252608456892,
+      "grad_norm": 0.3530223071575165,
+      "learning_rate": 0.00019177524429967423,
+      "loss": 0.6032,
+      "step": 580
+    },
+    {
+      "epoch": 1.2762218561230094,
+      "grad_norm": 0.5544403195381165,
+      "learning_rate": 0.00019153094462540714,
+      "loss": 0.7504,
+      "step": 581
+    },
+    {
+      "epoch": 1.2784184514003294,
+      "grad_norm": 0.6056380867958069,
+      "learning_rate": 0.00019128664495114003,
+      "loss": 0.6519,
+      "step": 582
+    },
+    {
+      "epoch": 1.2806150466776496,
+      "grad_norm": 0.38104236125946045,
+      "learning_rate": 0.00019104234527687292,
+      "loss": 0.6601,
+      "step": 583
+    },
+    {
+      "epoch": 1.2828116419549698,
+      "grad_norm": 0.6476126313209534,
+      "learning_rate": 0.00019079804560260584,
+      "loss": 0.6364,
+      "step": 584
+    },
+    {
+      "epoch": 1.28500823723229,
+      "grad_norm": 0.7528172731399536,
+      "learning_rate": 0.00019055374592833873,
+      "loss": 0.7879,
+      "step": 585
+    },
+    {
+      "epoch": 1.2872048325096102,
+      "grad_norm": 0.37613144516944885,
+      "learning_rate": 0.00019030944625407164,
+      "loss": 0.744,
+      "step": 586
+    },
+    {
+      "epoch": 1.2894014277869301,
+      "grad_norm": 0.4509885013103485,
+      "learning_rate": 0.00019006514657980453,
+      "loss": 0.7107,
+      "step": 587
+    },
+    {
+      "epoch": 1.2915980230642505,
+      "grad_norm": 0.5426962971687317,
+      "learning_rate": 0.00018982084690553742,
+      "loss": 0.7519,
+      "step": 588
+    },
+    {
+      "epoch": 1.2937946183415705,
+      "grad_norm": 0.5657976865768433,
+      "learning_rate": 0.00018957654723127034,
+      "loss": 0.6982,
+      "step": 589
+    },
+    {
+      "epoch": 1.2959912136188907,
+      "grad_norm": 0.7673377394676208,
+      "learning_rate": 0.00018933224755700323,
+      "loss": 0.7778,
+      "step": 590
+    },
+    {
+      "epoch": 1.2981878088962109,
+      "grad_norm": 0.3374074101448059,
+      "learning_rate": 0.00018908794788273612,
+      "loss": 0.5816,
+      "step": 591
+    },
+    {
+      "epoch": 1.300384404173531,
+      "grad_norm": 0.6337448954582214,
+      "learning_rate": 0.00018884364820846903,
+      "loss": 0.732,
+      "step": 592
+    },
+    {
+      "epoch": 1.3025809994508513,
+      "grad_norm": 0.38166627287864685,
+      "learning_rate": 0.00018859934853420192,
+      "loss": 0.6208,
+      "step": 593
+    },
+    {
+      "epoch": 1.3047775947281712,
+      "grad_norm": 0.5037196278572083,
+      "learning_rate": 0.0001883550488599348,
+      "loss": 0.7758,
+      "step": 594
+    },
+    {
+      "epoch": 1.3069741900054914,
+      "grad_norm": 0.694405734539032,
+      "learning_rate": 0.00018811074918566773,
+      "loss": 0.8537,
+      "step": 595
+    },
+    {
+      "epoch": 1.3091707852828116,
+      "grad_norm": 0.5364735126495361,
+      "learning_rate": 0.00018786644951140062,
+      "loss": 0.6846,
+      "step": 596
+    },
+    {
+      "epoch": 1.3113673805601318,
+      "grad_norm": 0.48998206853866577,
+      "learning_rate": 0.0001876221498371335,
+      "loss": 0.7236,
+      "step": 597
+    },
+    {
+      "epoch": 1.313563975837452,
+      "grad_norm": 0.41311854124069214,
+      "learning_rate": 0.00018737785016286645,
+      "loss": 0.6356,
+      "step": 598
+    },
+    {
+      "epoch": 1.3157605711147722,
+      "grad_norm": 0.4746629297733307,
+      "learning_rate": 0.00018713355048859934,
+      "loss": 0.7034,
+      "step": 599
+    },
+    {
+      "epoch": 1.3179571663920924,
+      "grad_norm": 0.4528580605983734,
+      "learning_rate": 0.00018688925081433225,
+      "loss": 0.9231,
+      "step": 600
+    },
+    {
+      "epoch": 1.3201537616694123,
+      "grad_norm": 0.5020402669906616,
+      "learning_rate": 0.00018664495114006514,
+      "loss": 0.7613,
+      "step": 601
+    },
+    {
+      "epoch": 1.3223503569467325,
+      "grad_norm": 0.5897614359855652,
+      "learning_rate": 0.00018640065146579803,
+      "loss": 0.7449,
+      "step": 602
+    },
+    {
+      "epoch": 1.3245469522240527,
+      "grad_norm": 0.5130749940872192,
+      "learning_rate": 0.00018615635179153095,
+      "loss": 0.8434,
+      "step": 603
+    },
+    {
+      "epoch": 1.326743547501373,
+      "grad_norm": 0.4698207974433899,
+      "learning_rate": 0.00018591205211726384,
+      "loss": 0.7717,
+      "step": 604
+    },
+    {
+      "epoch": 1.328940142778693,
+      "grad_norm": 0.48740091919898987,
+      "learning_rate": 0.00018566775244299675,
+      "loss": 0.8922,
+      "step": 605
+    },
+    {
+      "epoch": 1.331136738056013,
+      "grad_norm": 0.48060208559036255,
+      "learning_rate": 0.00018542345276872964,
+      "loss": 0.6542,
+      "step": 606
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.40898892283439636,
+      "learning_rate": 0.00018517915309446253,
+      "loss": 0.558,
+      "step": 607
+    },
+    {
+      "epoch": 1.3355299286106534,
+      "grad_norm": 0.5028170347213745,
+      "learning_rate": 0.00018493485342019545,
+      "loss": 0.7165,
+      "step": 608
+    },
+    {
+      "epoch": 1.3377265238879736,
+      "grad_norm": 0.5232133269309998,
+      "learning_rate": 0.00018469055374592834,
+      "loss": 0.8026,
+      "step": 609
+    },
+    {
+      "epoch": 1.3399231191652938,
+      "grad_norm": 0.5108156800270081,
+      "learning_rate": 0.00018444625407166123,
+      "loss": 0.8031,
+      "step": 610
+    },
+    {
+      "epoch": 1.342119714442614,
+      "grad_norm": 0.5589872002601624,
+      "learning_rate": 0.00018420195439739414,
+      "loss": 0.7812,
+      "step": 611
+    },
+    {
+      "epoch": 1.3443163097199342,
+      "grad_norm": 0.4798857569694519,
+      "learning_rate": 0.00018395765472312703,
+      "loss": 0.6756,
+      "step": 612
+    },
+    {
+      "epoch": 1.3465129049972542,
+      "grad_norm": 0.4463402032852173,
+      "learning_rate": 0.00018371335504885992,
+      "loss": 0.8132,
+      "step": 613
+    },
+    {
+      "epoch": 1.3487095002745744,
+      "grad_norm": 0.37967008352279663,
+      "learning_rate": 0.00018346905537459284,
+      "loss": 0.6944,
+      "step": 614
+    },
+    {
+      "epoch": 1.3509060955518946,
+      "grad_norm": 0.8725545406341553,
+      "learning_rate": 0.00018322475570032573,
+      "loss": 0.91,
+      "step": 615
+    },
+    {
+      "epoch": 1.3531026908292147,
+      "grad_norm": 0.4622761309146881,
+      "learning_rate": 0.00018298045602605864,
+      "loss": 0.8033,
+      "step": 616
+    },
+    {
+      "epoch": 1.355299286106535,
+      "grad_norm": 0.38005897402763367,
+      "learning_rate": 0.00018273615635179153,
+      "loss": 0.7221,
+      "step": 617
+    },
+    {
+      "epoch": 1.357495881383855,
+      "grad_norm": 0.5848243832588196,
+      "learning_rate": 0.00018249185667752442,
+      "loss": 1.0093,
+      "step": 618
+    },
+    {
+      "epoch": 1.359692476661175,
+      "grad_norm": 0.4712541699409485,
+      "learning_rate": 0.00018224755700325734,
+      "loss": 0.5921,
+      "step": 619
+    },
+    {
+      "epoch": 1.3618890719384953,
+      "grad_norm": 0.602802038192749,
+      "learning_rate": 0.00018200325732899023,
+      "loss": 0.6777,
+      "step": 620
+    },
+    {
+      "epoch": 1.3640856672158155,
+      "grad_norm": 0.5420047640800476,
+      "learning_rate": 0.00018175895765472312,
+      "loss": 0.9305,
+      "step": 621
+    },
+    {
+      "epoch": 1.3662822624931357,
+      "grad_norm": 0.5281029939651489,
+      "learning_rate": 0.00018151465798045603,
+      "loss": 0.6984,
+      "step": 622
+    },
+    {
+      "epoch": 1.3684788577704559,
+      "grad_norm": 0.4343905448913574,
+      "learning_rate": 0.00018127035830618892,
+      "loss": 0.6246,
+      "step": 623
+    },
+    {
+      "epoch": 1.370675453047776,
+      "grad_norm": 0.5427189469337463,
+      "learning_rate": 0.0001810260586319218,
+      "loss": 0.7849,
+      "step": 624
+    },
+    {
+      "epoch": 1.372872048325096,
+      "grad_norm": 0.6234015226364136,
+      "learning_rate": 0.00018078175895765473,
+      "loss": 0.8165,
+      "step": 625
+    },
+    {
+      "epoch": 1.3750686436024162,
+      "grad_norm": 0.39222970604896545,
+      "learning_rate": 0.00018053745928338762,
+      "loss": 0.658,
+      "step": 626
+    },
+    {
+      "epoch": 1.3772652388797364,
+      "grad_norm": 0.4553707540035248,
+      "learning_rate": 0.0001802931596091205,
+      "loss": 0.6861,
+      "step": 627
+    },
+    {
+      "epoch": 1.3794618341570566,
+      "grad_norm": 0.5022857785224915,
+      "learning_rate": 0.00018004885993485342,
+      "loss": 0.5369,
+      "step": 628
+    },
+    {
+      "epoch": 1.3816584294343768,
+      "grad_norm": 0.4557149112224579,
+      "learning_rate": 0.0001798045602605863,
+      "loss": 0.7248,
+      "step": 629
+    },
+    {
+      "epoch": 1.3838550247116967,
+      "grad_norm": 0.47795894742012024,
+      "learning_rate": 0.00017956026058631923,
+      "loss": 0.6932,
+      "step": 630
+    },
+    {
+      "epoch": 1.3860516199890172,
+      "grad_norm": 0.44724732637405396,
+      "learning_rate": 0.00017931596091205212,
+      "loss": 0.6997,
+      "step": 631
+    },
+    {
+      "epoch": 1.3882482152663371,
+      "grad_norm": 0.4972147047519684,
+      "learning_rate": 0.000179071661237785,
+      "loss": 0.8966,
+      "step": 632
+    },
+    {
+      "epoch": 1.3904448105436573,
+      "grad_norm": 0.8879517912864685,
+      "learning_rate": 0.00017882736156351792,
+      "loss": 0.7682,
+      "step": 633
+    },
+    {
+      "epoch": 1.3926414058209775,
+      "grad_norm": 0.4316074252128601,
+      "learning_rate": 0.0001785830618892508,
+      "loss": 0.6888,
+      "step": 634
+    },
+    {
+      "epoch": 1.3948380010982977,
+      "grad_norm": 0.5394413471221924,
+      "learning_rate": 0.0001783387622149837,
+      "loss": 0.7468,
+      "step": 635
+    },
+    {
+      "epoch": 1.3970345963756179,
+      "grad_norm": 0.500372052192688,
+      "learning_rate": 0.00017809446254071662,
+      "loss": 0.8419,
+      "step": 636
+    },
+    {
+      "epoch": 1.3992311916529379,
+      "grad_norm": 0.6006109714508057,
+      "learning_rate": 0.0001778501628664495,
+      "loss": 0.7054,
+      "step": 637
+    },
+    {
+      "epoch": 1.401427786930258,
+      "grad_norm": 0.48186105489730835,
+      "learning_rate": 0.0001776058631921824,
+      "loss": 0.6745,
+      "step": 638
+    },
+    {
+      "epoch": 1.4036243822075782,
+      "grad_norm": 1.2341715097427368,
+      "learning_rate": 0.0001773615635179153,
+      "loss": 0.6893,
+      "step": 639
+    },
+    {
+      "epoch": 1.4058209774848984,
+      "grad_norm": 0.409532755613327,
+      "learning_rate": 0.0001771172638436482,
+      "loss": 0.9609,
+      "step": 640
+    },
+    {
+      "epoch": 1.4080175727622186,
+      "grad_norm": 0.9368273615837097,
+      "learning_rate": 0.0001768729641693811,
+      "loss": 0.7174,
+      "step": 641
+    },
+    {
+      "epoch": 1.4102141680395386,
+      "grad_norm": 0.4923757314682007,
+      "learning_rate": 0.000176628664495114,
+      "loss": 0.8577,
+      "step": 642
+    },
+    {
+      "epoch": 1.412410763316859,
+      "grad_norm": 0.8126463890075684,
+      "learning_rate": 0.0001763843648208469,
+      "loss": 0.6764,
+      "step": 643
+    },
+    {
+      "epoch": 1.414607358594179,
+      "grad_norm": 0.4702051281929016,
+      "learning_rate": 0.0001761400651465798,
+      "loss": 0.649,
+      "step": 644
+    },
+    {
+      "epoch": 1.4168039538714992,
+      "grad_norm": 0.772032618522644,
+      "learning_rate": 0.0001758957654723127,
+      "loss": 0.7129,
+      "step": 645
+    },
+    {
+      "epoch": 1.4190005491488193,
+      "grad_norm": 0.626891016960144,
+      "learning_rate": 0.0001756514657980456,
+      "loss": 0.7304,
+      "step": 646
+    },
+    {
+      "epoch": 1.4211971444261395,
+      "grad_norm": 0.675136148929596,
+      "learning_rate": 0.0001754071661237785,
+      "loss": 0.7425,
+      "step": 647
+    },
+    {
+      "epoch": 1.4233937397034597,
+      "grad_norm": 0.46823063492774963,
+      "learning_rate": 0.0001751628664495114,
+      "loss": 0.6993,
+      "step": 648
+    },
+    {
+      "epoch": 1.4255903349807797,
+      "grad_norm": 0.5550483465194702,
+      "learning_rate": 0.00017491856677524429,
+      "loss": 0.8161,
+      "step": 649
+    },
+    {
+      "epoch": 1.4277869302580999,
+      "grad_norm": 0.4957829713821411,
+      "learning_rate": 0.0001746742671009772,
+      "loss": 0.6271,
+      "step": 650
+    },
+    {
+      "epoch": 1.42998352553542,
+      "grad_norm": 0.40619269013404846,
+      "learning_rate": 0.0001744299674267101,
+      "loss": 0.6704,
+      "step": 651
+    },
+    {
+      "epoch": 1.4321801208127403,
+      "grad_norm": 0.43223240971565247,
+      "learning_rate": 0.00017418566775244298,
+      "loss": 0.7227,
+      "step": 652
+    },
+    {
+      "epoch": 1.4343767160900605,
+      "grad_norm": 0.5181798338890076,
+      "learning_rate": 0.0001739413680781759,
+      "loss": 0.7575,
+      "step": 653
+    },
+    {
+      "epoch": 1.4365733113673804,
+      "grad_norm": 0.5141262412071228,
+      "learning_rate": 0.00017369706840390879,
+      "loss": 0.7071,
+      "step": 654
+    },
+    {
+      "epoch": 1.4387699066447008,
+      "grad_norm": 0.7030331492424011,
+      "learning_rate": 0.0001734527687296417,
+      "loss": 0.7337,
+      "step": 655
+    },
+    {
+      "epoch": 1.4409665019220208,
+      "grad_norm": 0.545166015625,
+      "learning_rate": 0.0001732084690553746,
+      "loss": 0.7484,
+      "step": 656
+    },
+    {
+      "epoch": 1.443163097199341,
+      "grad_norm": 0.4787271022796631,
+      "learning_rate": 0.00017296416938110748,
+      "loss": 0.9041,
+      "step": 657
+    },
+    {
+      "epoch": 1.4453596924766612,
+      "grad_norm": 0.5908335447311401,
+      "learning_rate": 0.0001727198697068404,
+      "loss": 0.8319,
+      "step": 658
+    },
+    {
+      "epoch": 1.4475562877539814,
+      "grad_norm": 0.4967615604400635,
+      "learning_rate": 0.00017247557003257329,
+      "loss": 0.9126,
+      "step": 659
+    },
+    {
+      "epoch": 1.4497528830313016,
+      "grad_norm": 1.550089955329895,
+      "learning_rate": 0.00017223127035830618,
+      "loss": 0.676,
+      "step": 660
+    },
+    {
+      "epoch": 1.4519494783086215,
+      "grad_norm": 0.42429283261299133,
+      "learning_rate": 0.0001719869706840391,
+      "loss": 0.6154,
+      "step": 661
+    },
+    {
+      "epoch": 1.4541460735859417,
+      "grad_norm": 0.617673397064209,
+      "learning_rate": 0.00017174267100977198,
+      "loss": 0.7819,
+      "step": 662
+    },
+    {
+      "epoch": 1.456342668863262,
+      "grad_norm": 0.4414830505847931,
+      "learning_rate": 0.00017149837133550487,
+      "loss": 0.7599,
+      "step": 663
+    },
+    {
+      "epoch": 1.458539264140582,
+      "grad_norm": 0.4655902087688446,
+      "learning_rate": 0.00017125407166123779,
+      "loss": 0.7029,
+      "step": 664
+    },
+    {
+      "epoch": 1.4607358594179023,
+      "grad_norm": 0.5203214287757874,
+      "learning_rate": 0.00017100977198697068,
+      "loss": 0.9012,
+      "step": 665
+    },
+    {
+      "epoch": 1.4629324546952225,
+      "grad_norm": 0.621769905090332,
+      "learning_rate": 0.00017076547231270356,
+      "loss": 0.8995,
+      "step": 666
+    },
+    {
+      "epoch": 1.4651290499725427,
+      "grad_norm": 0.7759563326835632,
+      "learning_rate": 0.00017052117263843648,
+      "loss": 0.6897,
+      "step": 667
+    },
+    {
+      "epoch": 1.4673256452498626,
+      "grad_norm": 0.5186193585395813,
+      "learning_rate": 0.00017027687296416937,
+      "loss": 0.8785,
+      "step": 668
+    },
+    {
+      "epoch": 1.4695222405271828,
+      "grad_norm": 0.5344983339309692,
+      "learning_rate": 0.00017003257328990229,
+      "loss": 0.7036,
+      "step": 669
+    },
+    {
+      "epoch": 1.471718835804503,
+      "grad_norm": 0.5185381770133972,
+      "learning_rate": 0.00016978827361563518,
+      "loss": 0.7372,
+      "step": 670
+    },
+    {
+      "epoch": 1.4739154310818232,
+      "grad_norm": 0.5804538726806641,
+      "learning_rate": 0.00016954397394136806,
+      "loss": 0.8488,
+      "step": 671
+    },
+    {
+      "epoch": 1.4761120263591434,
+      "grad_norm": 0.42855504155158997,
+      "learning_rate": 0.00016929967426710098,
+      "loss": 0.7758,
+      "step": 672
+    },
+    {
+      "epoch": 1.4783086216364634,
+      "grad_norm": 0.5313897132873535,
+      "learning_rate": 0.00016905537459283387,
+      "loss": 0.7149,
+      "step": 673
+    },
+    {
+      "epoch": 1.4805052169137836,
+      "grad_norm": 0.5259653329849243,
+      "learning_rate": 0.00016881107491856676,
+      "loss": 0.84,
+      "step": 674
+    },
+    {
+      "epoch": 1.4827018121911038,
+      "grad_norm": 0.42905616760253906,
+      "learning_rate": 0.00016856677524429968,
+      "loss": 0.631,
+      "step": 675
+    },
+    {
+      "epoch": 1.484898407468424,
+      "grad_norm": 0.660906970500946,
+      "learning_rate": 0.00016832247557003257,
+      "loss": 0.7452,
+      "step": 676
+    },
+    {
+      "epoch": 1.4870950027457441,
+      "grad_norm": 0.39352715015411377,
+      "learning_rate": 0.00016807817589576545,
+      "loss": 0.6992,
+      "step": 677
+    },
+    {
+      "epoch": 1.4892915980230643,
+      "grad_norm": 0.5723733901977539,
+      "learning_rate": 0.00016783387622149837,
+      "loss": 0.7973,
+      "step": 678
+    },
+    {
+      "epoch": 1.4914881933003845,
+      "grad_norm": 0.43104228377342224,
+      "learning_rate": 0.00016758957654723126,
+      "loss": 0.7955,
+      "step": 679
+    },
+    {
+      "epoch": 1.4936847885777045,
+      "grad_norm": 1.0099215507507324,
+      "learning_rate": 0.00016734527687296415,
+      "loss": 0.9055,
+      "step": 680
+    },
+    {
+      "epoch": 1.4958813838550247,
+      "grad_norm": 0.4106708765029907,
+      "learning_rate": 0.00016710097719869707,
+      "loss": 0.7082,
+      "step": 681
+    },
+    {
+      "epoch": 1.4980779791323449,
+      "grad_norm": 0.5595798492431641,
+      "learning_rate": 0.00016685667752442995,
+      "loss": 0.9041,
+      "step": 682
+    },
+    {
+      "epoch": 1.500274574409665,
+      "grad_norm": 0.5724670886993408,
+      "learning_rate": 0.00016661237785016287,
+      "loss": 0.5003,
+      "step": 683
+    },
+    {
+      "epoch": 1.5024711696869852,
+      "grad_norm": 0.5939717888832092,
+      "learning_rate": 0.00016636807817589576,
+      "loss": 0.6438,
+      "step": 684
+    },
+    {
+      "epoch": 1.5046677649643052,
+      "grad_norm": 0.628688633441925,
+      "learning_rate": 0.00016612377850162865,
+      "loss": 0.8332,
+      "step": 685
+    },
+    {
+      "epoch": 1.5068643602416256,
+      "grad_norm": 0.6859570145606995,
+      "learning_rate": 0.00016587947882736157,
+      "loss": 0.8868,
+      "step": 686
+    },
+    {
+      "epoch": 1.5090609555189456,
+      "grad_norm": 0.5950968861579895,
+      "learning_rate": 0.00016563517915309445,
+      "loss": 0.7917,
+      "step": 687
+    },
+    {
+      "epoch": 1.5112575507962658,
+      "grad_norm": 0.8112242817878723,
+      "learning_rate": 0.00016539087947882734,
+      "loss": 0.8346,
+      "step": 688
+    },
+    {
+      "epoch": 1.513454146073586,
+      "grad_norm": 0.5476120710372925,
+      "learning_rate": 0.00016514657980456026,
+      "loss": 0.6472,
+      "step": 689
+    },
+    {
+      "epoch": 1.515650741350906,
+      "grad_norm": 0.4667632579803467,
+      "learning_rate": 0.00016490228013029315,
+      "loss": 0.6922,
+      "step": 690
+    },
+    {
+      "epoch": 1.5178473366282264,
+      "grad_norm": 0.7376199960708618,
+      "learning_rate": 0.00016465798045602604,
+      "loss": 0.7966,
+      "step": 691
+    },
+    {
+      "epoch": 1.5200439319055463,
+      "grad_norm": 0.7364751100540161,
+      "learning_rate": 0.00016441368078175895,
+      "loss": 0.7955,
+      "step": 692
+    },
+    {
+      "epoch": 1.5222405271828665,
+      "grad_norm": 0.48585665225982666,
+      "learning_rate": 0.00016416938110749184,
+      "loss": 0.7716,
+      "step": 693
+    },
+    {
+      "epoch": 1.5244371224601867,
+      "grad_norm": 0.8947611451148987,
+      "learning_rate": 0.00016392508143322476,
+      "loss": 0.6573,
+      "step": 694
+    },
+    {
+      "epoch": 1.526633717737507,
+      "grad_norm": 0.38048291206359863,
+      "learning_rate": 0.00016368078175895765,
+      "loss": 0.7572,
+      "step": 695
+    },
+    {
+      "epoch": 1.528830313014827,
+      "grad_norm": 0.4531625211238861,
+      "learning_rate": 0.00016343648208469054,
+      "loss": 0.6794,
+      "step": 696
+    },
+    {
+      "epoch": 1.531026908292147,
+      "grad_norm": 0.39911630749702454,
+      "learning_rate": 0.00016319218241042346,
+      "loss": 0.7322,
+      "step": 697
+    },
+    {
+      "epoch": 1.5332235035694675,
+      "grad_norm": 0.4888548254966736,
+      "learning_rate": 0.00016294788273615634,
+      "loss": 0.7082,
+      "step": 698
+    },
+    {
+      "epoch": 1.5354200988467874,
+      "grad_norm": 0.7266231179237366,
+      "learning_rate": 0.00016270358306188923,
+      "loss": 0.7734,
+      "step": 699
+    },
+    {
+      "epoch": 1.5376166941241076,
+      "grad_norm": 0.6152387857437134,
+      "learning_rate": 0.00016245928338762215,
+      "loss": 0.8997,
+      "step": 700
+    },
+    {
+      "epoch": 1.5398132894014278,
+      "grad_norm": 0.5999181866645813,
+      "learning_rate": 0.00016221498371335504,
+      "loss": 0.7377,
+      "step": 701
+    },
+    {
+      "epoch": 1.5420098846787478,
+      "grad_norm": 0.4883151650428772,
+      "learning_rate": 0.00016197068403908793,
+      "loss": 0.7121,
+      "step": 702
+    },
+    {
+      "epoch": 1.5442064799560682,
+      "grad_norm": 0.3487742245197296,
+      "learning_rate": 0.00016172638436482084,
+      "loss": 0.616,
+      "step": 703
+    },
+    {
+      "epoch": 1.5464030752333882,
+      "grad_norm": 0.4792243242263794,
+      "learning_rate": 0.00016148208469055373,
+      "loss": 0.583,
+      "step": 704
+    },
+    {
+      "epoch": 1.5485996705107083,
+      "grad_norm": 0.5166910886764526,
+      "learning_rate": 0.00016123778501628662,
+      "loss": 0.7606,
+      "step": 705
+    },
+    {
+      "epoch": 1.5507962657880285,
+      "grad_norm": 0.38755539059638977,
+      "learning_rate": 0.00016099348534201954,
+      "loss": 0.6985,
+      "step": 706
+    },
+    {
+      "epoch": 1.5529928610653487,
+      "grad_norm": 0.6558068990707397,
+      "learning_rate": 0.00016074918566775243,
+      "loss": 0.766,
+      "step": 707
+    },
+    {
+      "epoch": 1.555189456342669,
+      "grad_norm": 0.7644460797309875,
+      "learning_rate": 0.00016050488599348534,
+      "loss": 0.7598,
+      "step": 708
+    },
+    {
+      "epoch": 1.5573860516199889,
+      "grad_norm": 0.7870489954948425,
+      "learning_rate": 0.00016026058631921823,
+      "loss": 0.9012,
+      "step": 709
+    },
+    {
+      "epoch": 1.5595826468973093,
+      "grad_norm": 0.5084778666496277,
+      "learning_rate": 0.00016001628664495112,
+      "loss": 0.4973,
+      "step": 710
+    },
+    {
+      "epoch": 1.5617792421746293,
+      "grad_norm": 0.5386030673980713,
+      "learning_rate": 0.00015977198697068404,
+      "loss": 0.8371,
+      "step": 711
+    },
+    {
+      "epoch": 1.5639758374519495,
+      "grad_norm": 0.5492763519287109,
+      "learning_rate": 0.00015952768729641693,
+      "loss": 0.5789,
+      "step": 712
+    },
+    {
+      "epoch": 1.5661724327292696,
+      "grad_norm": 1.2466025352478027,
+      "learning_rate": 0.00015928338762214982,
+      "loss": 0.6546,
+      "step": 713
+    },
+    {
+      "epoch": 1.5683690280065898,
+      "grad_norm": 0.4138627350330353,
+      "learning_rate": 0.00015903908794788273,
+      "loss": 0.6385,
+      "step": 714
+    },
+    {
+      "epoch": 1.57056562328391,
+      "grad_norm": 0.583316445350647,
+      "learning_rate": 0.00015879478827361562,
+      "loss": 0.825,
+      "step": 715
+    },
+    {
+      "epoch": 1.57276221856123,
+      "grad_norm": 0.6477898359298706,
+      "learning_rate": 0.0001585504885993485,
+      "loss": 0.9284,
+      "step": 716
+    },
+    {
+      "epoch": 1.5749588138385504,
+      "grad_norm": 0.544056236743927,
+      "learning_rate": 0.00015830618892508143,
+      "loss": 0.7584,
+      "step": 717
+    },
+    {
+      "epoch": 1.5771554091158704,
+      "grad_norm": 0.6870483756065369,
+      "learning_rate": 0.00015806188925081432,
+      "loss": 0.6525,
+      "step": 718
+    },
+    {
+      "epoch": 1.5793520043931906,
+      "grad_norm": 0.5568920969963074,
+      "learning_rate": 0.00015781758957654723,
+      "loss": 0.7269,
+      "step": 719
+    },
+    {
+      "epoch": 1.5815485996705108,
+      "grad_norm": 0.771216094493866,
+      "learning_rate": 0.00015757328990228012,
+      "loss": 0.6953,
+      "step": 720
+    },
+    {
+      "epoch": 1.5837451949478307,
+      "grad_norm": 0.4723263084888458,
+      "learning_rate": 0.000157328990228013,
+      "loss": 0.8732,
+      "step": 721
+    },
+    {
+      "epoch": 1.5859417902251511,
+      "grad_norm": 0.6392165422439575,
+      "learning_rate": 0.00015708469055374593,
+      "loss": 0.874,
+      "step": 722
+    },
+    {
+      "epoch": 1.588138385502471,
+      "grad_norm": 0.42938390374183655,
+      "learning_rate": 0.00015684039087947882,
+      "loss": 0.7169,
+      "step": 723
+    },
+    {
+      "epoch": 1.5903349807797913,
+      "grad_norm": 0.5666797161102295,
+      "learning_rate": 0.0001565960912052117,
+      "loss": 0.623,
+      "step": 724
+    },
+    {
+      "epoch": 1.5925315760571115,
+      "grad_norm": 0.774155855178833,
+      "learning_rate": 0.00015635179153094462,
+      "loss": 0.8099,
+      "step": 725
+    },
+    {
+      "epoch": 1.5947281713344317,
+      "grad_norm": 0.7930957078933716,
+      "learning_rate": 0.0001561074918566775,
+      "loss": 0.731,
+      "step": 726
+    },
+    {
+      "epoch": 1.5969247666117519,
+      "grad_norm": 0.7446067929267883,
+      "learning_rate": 0.0001558631921824104,
+      "loss": 0.8547,
+      "step": 727
+    },
+    {
+      "epoch": 1.5991213618890718,
+      "grad_norm": 0.4947164058685303,
+      "learning_rate": 0.00015561889250814332,
+      "loss": 0.8023,
+      "step": 728
+    },
+    {
+      "epoch": 1.6013179571663922,
+      "grad_norm": 0.7591903209686279,
+      "learning_rate": 0.0001553745928338762,
+      "loss": 0.7555,
+      "step": 729
+    },
+    {
+      "epoch": 1.6035145524437122,
+      "grad_norm": 0.5278122425079346,
+      "learning_rate": 0.0001551302931596091,
+      "loss": 0.74,
+      "step": 730
+    },
+    {
+      "epoch": 1.6057111477210324,
+      "grad_norm": 0.40393689274787903,
+      "learning_rate": 0.000154885993485342,
+      "loss": 0.4805,
+      "step": 731
+    },
+    {
+      "epoch": 1.6079077429983526,
+      "grad_norm": 0.535178005695343,
+      "learning_rate": 0.0001546416938110749,
+      "loss": 0.7462,
+      "step": 732
+    },
+    {
+      "epoch": 1.6101043382756726,
+      "grad_norm": 0.5995372533798218,
+      "learning_rate": 0.00015439739413680782,
+      "loss": 0.8889,
+      "step": 733
+    },
+    {
+      "epoch": 1.612300933552993,
+      "grad_norm": 0.45476219058036804,
+      "learning_rate": 0.0001541530944625407,
+      "loss": 0.8031,
+      "step": 734
+    },
+    {
+      "epoch": 1.614497528830313,
+      "grad_norm": 0.6498055458068848,
+      "learning_rate": 0.0001539087947882736,
+      "loss": 0.8291,
+      "step": 735
+    },
+    {
+      "epoch": 1.6166941241076331,
+      "grad_norm": 0.5195559859275818,
+      "learning_rate": 0.0001536644951140065,
+      "loss": 0.7869,
+      "step": 736
+    },
+    {
+      "epoch": 1.6188907193849533,
+      "grad_norm": 0.5386220812797546,
+      "learning_rate": 0.0001534201954397394,
+      "loss": 0.8004,
+      "step": 737
+    },
+    {
+      "epoch": 1.6210873146622735,
+      "grad_norm": 0.7313625812530518,
+      "learning_rate": 0.0001531758957654723,
+      "loss": 0.7099,
+      "step": 738
+    },
+    {
+      "epoch": 1.6232839099395937,
+      "grad_norm": 0.46535149216651917,
+      "learning_rate": 0.0001529315960912052,
+      "loss": 0.6892,
+      "step": 739
+    },
+    {
+      "epoch": 1.6254805052169137,
+      "grad_norm": 0.5318624377250671,
+      "learning_rate": 0.0001526872964169381,
+      "loss": 0.7843,
+      "step": 740
+    },
+    {
+      "epoch": 1.627677100494234,
+      "grad_norm": 0.45150521397590637,
+      "learning_rate": 0.00015244299674267099,
+      "loss": 0.8363,
+      "step": 741
+    },
+    {
+      "epoch": 1.629873695771554,
+      "grad_norm": 0.5278283953666687,
+      "learning_rate": 0.0001521986970684039,
+      "loss": 0.7198,
+      "step": 742
+    },
+    {
+      "epoch": 1.6320702910488742,
+      "grad_norm": 0.7059549689292908,
+      "learning_rate": 0.0001519543973941368,
+      "loss": 0.6903,
+      "step": 743
+    },
+    {
+      "epoch": 1.6342668863261944,
+      "grad_norm": 0.5246755480766296,
+      "learning_rate": 0.00015171009771986968,
+      "loss": 0.5945,
+      "step": 744
+    },
+    {
+      "epoch": 1.6364634816035144,
+      "grad_norm": 0.5703069567680359,
+      "learning_rate": 0.0001514657980456026,
+      "loss": 0.6404,
+      "step": 745
+    },
+    {
+      "epoch": 1.6386600768808348,
+      "grad_norm": 0.5598427057266235,
+      "learning_rate": 0.0001512214983713355,
+      "loss": 0.4746,
+      "step": 746
+    },
+    {
+      "epoch": 1.6408566721581548,
+      "grad_norm": 0.5548290610313416,
+      "learning_rate": 0.0001509771986970684,
+      "loss": 0.7605,
+      "step": 747
+    },
+    {
+      "epoch": 1.643053267435475,
+      "grad_norm": 0.6005709171295166,
+      "learning_rate": 0.0001507328990228013,
+      "loss": 0.6344,
+      "step": 748
+    },
+    {
+      "epoch": 1.6452498627127952,
+      "grad_norm": 0.4307134449481964,
+      "learning_rate": 0.00015048859934853418,
+      "loss": 0.6385,
+      "step": 749
+    },
+    {
+      "epoch": 1.6474464579901154,
+      "grad_norm": 0.4333352744579315,
+      "learning_rate": 0.0001502442996742671,
+      "loss": 0.6341,
+      "step": 750
+    },
+    {
+      "epoch": 1.6496430532674355,
+      "grad_norm": 0.6455225348472595,
+      "learning_rate": 0.00015,
+      "loss": 0.7465,
+      "step": 751
+    },
+    {
+      "epoch": 1.6518396485447555,
+      "grad_norm": 0.6464992761611938,
+      "learning_rate": 0.00014975570032573288,
+      "loss": 0.8436,
+      "step": 752
+    },
+    {
+      "epoch": 1.654036243822076,
+      "grad_norm": 0.7667239904403687,
+      "learning_rate": 0.0001495114006514658,
+      "loss": 0.9745,
+      "step": 753
+    },
+    {
+      "epoch": 1.656232839099396,
+      "grad_norm": 0.5757808685302734,
+      "learning_rate": 0.00014926710097719868,
+      "loss": 0.6076,
+      "step": 754
+    },
+    {
+      "epoch": 1.658429434376716,
+      "grad_norm": 0.4610781669616699,
+      "learning_rate": 0.00014902280130293157,
+      "loss": 0.6967,
+      "step": 755
+    },
+    {
+      "epoch": 1.6606260296540363,
+      "grad_norm": 0.35578879714012146,
+      "learning_rate": 0.0001487785016286645,
+      "loss": 0.7178,
+      "step": 756
+    },
+    {
+      "epoch": 1.6628226249313562,
+      "grad_norm": 0.7207755446434021,
+      "learning_rate": 0.00014853420195439738,
+      "loss": 0.8331,
+      "step": 757
+    },
+    {
+      "epoch": 1.6650192202086767,
+      "grad_norm": 0.6547420024871826,
+      "learning_rate": 0.0001482899022801303,
+      "loss": 0.7983,
+      "step": 758
+    },
+    {
+      "epoch": 1.6672158154859966,
+      "grad_norm": 0.5377845764160156,
+      "learning_rate": 0.00014804560260586318,
+      "loss": 0.8322,
+      "step": 759
+    },
+    {
+      "epoch": 1.6694124107633168,
+      "grad_norm": 0.8457434773445129,
+      "learning_rate": 0.00014780130293159607,
+      "loss": 0.7739,
+      "step": 760
+    },
+    {
+      "epoch": 1.671609006040637,
+      "grad_norm": 0.5992394685745239,
+      "learning_rate": 0.000147557003257329,
+      "loss": 0.7495,
+      "step": 761
+    },
+    {
+      "epoch": 1.6738056013179572,
+      "grad_norm": 0.4401768445968628,
+      "learning_rate": 0.00014731270358306188,
+      "loss": 0.774,
+      "step": 762
+    },
+    {
+      "epoch": 1.6760021965952774,
+      "grad_norm": 0.47606122493743896,
+      "learning_rate": 0.00014706840390879477,
+      "loss": 0.7386,
+      "step": 763
+    },
+    {
+      "epoch": 1.6781987918725974,
+      "grad_norm": 1.0029031038284302,
+      "learning_rate": 0.00014682410423452768,
+      "loss": 0.826,
+      "step": 764
+    },
+    {
+      "epoch": 1.6803953871499178,
+      "grad_norm": 0.802718460559845,
+      "learning_rate": 0.00014657980456026057,
+      "loss": 0.7973,
+      "step": 765
+    },
+    {
+      "epoch": 1.6825919824272377,
+      "grad_norm": 0.5505915880203247,
+      "learning_rate": 0.00014633550488599346,
+      "loss": 0.6694,
+      "step": 766
+    },
+    {
+      "epoch": 1.684788577704558,
+      "grad_norm": 0.5079130530357361,
+      "learning_rate": 0.00014609120521172638,
+      "loss": 0.7702,
+      "step": 767
+    },
+    {
+      "epoch": 1.6869851729818781,
+      "grad_norm": 0.43140462040901184,
+      "learning_rate": 0.00014584690553745927,
+      "loss": 0.638,
+      "step": 768
+    },
+    {
+      "epoch": 1.689181768259198,
+      "grad_norm": 0.5110360383987427,
+      "learning_rate": 0.00014560260586319216,
+      "loss": 0.7396,
+      "step": 769
+    },
+    {
+      "epoch": 1.6913783635365185,
+      "grad_norm": 1.2050848007202148,
+      "learning_rate": 0.00014535830618892507,
+      "loss": 0.8622,
+      "step": 770
+    },
+    {
+      "epoch": 1.6935749588138385,
+      "grad_norm": 0.6973075270652771,
+      "learning_rate": 0.00014511400651465796,
+      "loss": 0.8945,
+      "step": 771
+    },
+    {
+      "epoch": 1.6957715540911587,
+      "grad_norm": 1.1463433504104614,
+      "learning_rate": 0.00014486970684039088,
+      "loss": 0.8421,
+      "step": 772
+    },
+    {
+      "epoch": 1.6979681493684788,
+      "grad_norm": 0.5088452696800232,
+      "learning_rate": 0.00014462540716612377,
+      "loss": 0.6982,
+      "step": 773
+    },
+    {
+      "epoch": 1.700164744645799,
+      "grad_norm": 0.5248615741729736,
+      "learning_rate": 0.00014438110749185666,
+      "loss": 0.6233,
+      "step": 774
+    },
+    {
+      "epoch": 1.7023613399231192,
+      "grad_norm": 0.514994204044342,
+      "learning_rate": 0.00014413680781758957,
+      "loss": 0.7752,
+      "step": 775
+    },
+    {
+      "epoch": 1.7045579352004392,
+      "grad_norm": 0.5187984704971313,
+      "learning_rate": 0.00014389250814332246,
+      "loss": 0.5669,
+      "step": 776
+    },
+    {
+      "epoch": 1.7067545304777596,
+      "grad_norm": 0.48899856209754944,
+      "learning_rate": 0.00014364820846905535,
+      "loss": 0.8046,
+      "step": 777
+    },
+    {
+      "epoch": 1.7089511257550796,
+      "grad_norm": 0.5431028604507446,
+      "learning_rate": 0.00014340390879478827,
+      "loss": 0.8145,
+      "step": 778
+    },
+    {
+      "epoch": 1.7111477210323998,
+      "grad_norm": 0.5340853929519653,
+      "learning_rate": 0.00014315960912052116,
+      "loss": 0.7382,
+      "step": 779
+    },
+    {
+      "epoch": 1.71334431630972,
+      "grad_norm": 0.6564583778381348,
+      "learning_rate": 0.00014291530944625404,
+      "loss": 0.7887,
+      "step": 780
+    },
+    {
+      "epoch": 1.7155409115870401,
+      "grad_norm": 0.5532479286193848,
+      "learning_rate": 0.00014267100977198696,
+      "loss": 0.7707,
+      "step": 781
+    },
+    {
+      "epoch": 1.7177375068643603,
+      "grad_norm": 0.6629275679588318,
+      "learning_rate": 0.00014242671009771985,
+      "loss": 0.666,
+      "step": 782
+    },
+    {
+      "epoch": 1.7199341021416803,
+      "grad_norm": 0.4812797009944916,
+      "learning_rate": 0.00014218241042345274,
+      "loss": 0.7633,
+      "step": 783
+    },
+    {
+      "epoch": 1.7221306974190007,
+      "grad_norm": 0.4363406002521515,
+      "learning_rate": 0.00014193811074918566,
+      "loss": 0.6764,
+      "step": 784
+    },
+    {
+      "epoch": 1.7243272926963207,
+      "grad_norm": 0.5929400324821472,
+      "learning_rate": 0.00014169381107491854,
+      "loss": 0.9671,
+      "step": 785
+    },
+    {
+      "epoch": 1.7265238879736409,
+      "grad_norm": 0.39160266518592834,
+      "learning_rate": 0.00014144951140065146,
+      "loss": 0.6441,
+      "step": 786
+    },
+    {
+      "epoch": 1.728720483250961,
+      "grad_norm": 0.7725654244422913,
+      "learning_rate": 0.00014120521172638435,
+      "loss": 0.6956,
+      "step": 787
+    },
+    {
+      "epoch": 1.730917078528281,
+      "grad_norm": 0.4521915912628174,
+      "learning_rate": 0.00014096091205211724,
+      "loss": 0.7159,
+      "step": 788
+    },
+    {
+      "epoch": 1.7331136738056014,
+      "grad_norm": 0.683752715587616,
+      "learning_rate": 0.00014071661237785016,
+      "loss": 0.8236,
+      "step": 789
+    },
+    {
+      "epoch": 1.7353102690829214,
+      "grad_norm": 0.5970603823661804,
+      "learning_rate": 0.00014047231270358305,
+      "loss": 0.9332,
+      "step": 790
+    },
+    {
+      "epoch": 1.7375068643602416,
+      "grad_norm": 0.6367862224578857,
+      "learning_rate": 0.00014022801302931593,
+      "loss": 0.7012,
+      "step": 791
+    },
+    {
+      "epoch": 1.7397034596375618,
+      "grad_norm": 0.4243377149105072,
+      "learning_rate": 0.00013998371335504885,
+      "loss": 0.6885,
+      "step": 792
+    },
+    {
+      "epoch": 1.741900054914882,
+      "grad_norm": 0.49693796038627625,
+      "learning_rate": 0.00013973941368078174,
+      "loss": 0.6716,
+      "step": 793
+    },
+    {
+      "epoch": 1.7440966501922022,
+      "grad_norm": 0.5494533777236938,
+      "learning_rate": 0.00013949511400651463,
+      "loss": 0.882,
+      "step": 794
+    },
+    {
+      "epoch": 1.7462932454695221,
+      "grad_norm": 0.49662622809410095,
+      "learning_rate": 0.00013925081433224755,
+      "loss": 0.7041,
+      "step": 795
+    },
+    {
+      "epoch": 1.7484898407468425,
+      "grad_norm": 0.7410906553268433,
+      "learning_rate": 0.00013900651465798043,
+      "loss": 0.6673,
+      "step": 796
+    },
+    {
+      "epoch": 1.7506864360241625,
+      "grad_norm": 0.5417400002479553,
+      "learning_rate": 0.00013876221498371335,
+      "loss": 0.8614,
+      "step": 797
+    },
+    {
+      "epoch": 1.7528830313014827,
+      "grad_norm": 0.5067940950393677,
+      "learning_rate": 0.00013851791530944624,
+      "loss": 0.7569,
+      "step": 798
+    },
+    {
+      "epoch": 1.755079626578803,
+      "grad_norm": 0.6262591481208801,
+      "learning_rate": 0.00013827361563517913,
+      "loss": 0.8846,
+      "step": 799
+    },
+    {
+      "epoch": 1.7572762218561229,
+      "grad_norm": 0.4682822823524475,
+      "learning_rate": 0.00013802931596091205,
+      "loss": 0.926,
+      "step": 800
+    },
+    {
+      "epoch": 1.7594728171334433,
+      "grad_norm": 0.49318933486938477,
+      "learning_rate": 0.00013778501628664493,
+      "loss": 0.8228,
+      "step": 801
+    },
+    {
+      "epoch": 1.7616694124107632,
+      "grad_norm": 0.5845035910606384,
+      "learning_rate": 0.00013754071661237782,
+      "loss": 0.9787,
+      "step": 802
+    },
+    {
+      "epoch": 1.7638660076880834,
+      "grad_norm": 0.6149041652679443,
+      "learning_rate": 0.00013729641693811074,
+      "loss": 1.0297,
+      "step": 803
+    },
+    {
+      "epoch": 1.7660626029654036,
+      "grad_norm": 0.6234105229377747,
+      "learning_rate": 0.00013705211726384363,
+      "loss": 1.002,
+      "step": 804
+    },
+    {
+      "epoch": 1.7682591982427238,
+      "grad_norm": 0.5687574148178101,
+      "learning_rate": 0.00013680781758957652,
+      "loss": 0.7784,
+      "step": 805
+    },
+    {
+      "epoch": 1.770455793520044,
+      "grad_norm": 0.5291481614112854,
+      "learning_rate": 0.00013656351791530943,
+      "loss": 0.6964,
+      "step": 806
+    },
+    {
+      "epoch": 1.772652388797364,
+      "grad_norm": 0.5180975794792175,
+      "learning_rate": 0.00013631921824104232,
+      "loss": 0.7522,
+      "step": 807
+    },
+    {
+      "epoch": 1.7748489840746844,
+      "grad_norm": 1.086072325706482,
+      "learning_rate": 0.0001360749185667752,
+      "loss": 0.8329,
+      "step": 808
+    },
+    {
+      "epoch": 1.7770455793520044,
+      "grad_norm": 0.49188101291656494,
+      "learning_rate": 0.00013583061889250813,
+      "loss": 0.8629,
+      "step": 809
+    },
+    {
+      "epoch": 1.7792421746293245,
+      "grad_norm": 0.5753602981567383,
+      "learning_rate": 0.00013558631921824102,
+      "loss": 0.5485,
+      "step": 810
+    },
+    {
+      "epoch": 1.7814387699066447,
+      "grad_norm": 0.7131554484367371,
+      "learning_rate": 0.00013534201954397394,
+      "loss": 0.6188,
+      "step": 811
+    },
+    {
+      "epoch": 1.7836353651839647,
+      "grad_norm": 0.5149171352386475,
+      "learning_rate": 0.00013509771986970682,
+      "loss": 0.7357,
+      "step": 812
+    },
+    {
+      "epoch": 1.7858319604612851,
+      "grad_norm": 0.5610647797584534,
+      "learning_rate": 0.00013485342019543971,
+      "loss": 0.8912,
+      "step": 813
+    },
+    {
+      "epoch": 1.788028555738605,
+      "grad_norm": 0.5049970149993896,
+      "learning_rate": 0.00013460912052117263,
+      "loss": 0.8,
+      "step": 814
+    },
+    {
+      "epoch": 1.7902251510159253,
+      "grad_norm": 0.684876024723053,
+      "learning_rate": 0.00013436482084690552,
+      "loss": 0.8236,
+      "step": 815
+    },
+    {
+      "epoch": 1.7924217462932455,
+      "grad_norm": 0.42195042967796326,
+      "learning_rate": 0.0001341205211726384,
+      "loss": 0.7635,
+      "step": 816
+    },
+    {
+      "epoch": 1.7946183415705657,
+      "grad_norm": 0.5523778796195984,
+      "learning_rate": 0.00013387622149837132,
+      "loss": 0.6633,
+      "step": 817
+    },
+    {
+      "epoch": 1.7968149368478858,
+      "grad_norm": 0.5226792693138123,
+      "learning_rate": 0.00013363192182410421,
+      "loss": 0.7962,
+      "step": 818
+    },
+    {
+      "epoch": 1.7990115321252058,
+      "grad_norm": 1.8618338108062744,
+      "learning_rate": 0.0001333876221498371,
+      "loss": 0.7461,
+      "step": 819
+    },
+    {
+      "epoch": 1.8012081274025262,
+      "grad_norm": 0.8910929560661316,
+      "learning_rate": 0.00013314332247557002,
+      "loss": 0.796,
+      "step": 820
+    },
+    {
+      "epoch": 1.8034047226798462,
+      "grad_norm": 0.4908788800239563,
+      "learning_rate": 0.0001328990228013029,
+      "loss": 0.79,
+      "step": 821
+    },
+    {
+      "epoch": 1.8056013179571664,
+      "grad_norm": 0.5825263261795044,
+      "learning_rate": 0.00013265472312703582,
+      "loss": 0.8567,
+      "step": 822
+    },
+    {
+      "epoch": 1.8077979132344866,
+      "grad_norm": 1.334252119064331,
+      "learning_rate": 0.00013241042345276871,
+      "loss": 0.7816,
+      "step": 823
+    },
+    {
+      "epoch": 1.8099945085118065,
+      "grad_norm": 0.5359366536140442,
+      "learning_rate": 0.0001321661237785016,
+      "loss": 0.7432,
+      "step": 824
+    },
+    {
+      "epoch": 1.812191103789127,
+      "grad_norm": 0.4927462339401245,
+      "learning_rate": 0.00013192182410423452,
+      "loss": 0.6469,
+      "step": 825
+    },
+    {
+      "epoch": 1.814387699066447,
+      "grad_norm": 1.1213749647140503,
+      "learning_rate": 0.0001316775244299674,
+      "loss": 0.6334,
+      "step": 826
+    },
+    {
+      "epoch": 1.8165842943437671,
+      "grad_norm": 0.4652058482170105,
+      "learning_rate": 0.0001314332247557003,
+      "loss": 0.8189,
+      "step": 827
+    },
+    {
+      "epoch": 1.8187808896210873,
+      "grad_norm": 0.585376501083374,
+      "learning_rate": 0.00013118892508143321,
+      "loss": 0.659,
+      "step": 828
+    },
+    {
+      "epoch": 1.8209774848984075,
+      "grad_norm": 0.5371707677841187,
+      "learning_rate": 0.00013094462540716613,
+      "loss": 0.6342,
+      "step": 829
+    },
+    {
+      "epoch": 1.8231740801757277,
+      "grad_norm": 0.5396832227706909,
+      "learning_rate": 0.00013070032573289902,
+      "loss": 0.5716,
+      "step": 830
+    },
+    {
+      "epoch": 1.8253706754530477,
+      "grad_norm": 0.5688884258270264,
+      "learning_rate": 0.0001304560260586319,
+      "loss": 0.5058,
+      "step": 831
+    },
+    {
+      "epoch": 1.827567270730368,
+      "grad_norm": 0.4664338231086731,
+      "learning_rate": 0.00013021172638436483,
+      "loss": 0.7766,
+      "step": 832
+    },
+    {
+      "epoch": 1.829763866007688,
+      "grad_norm": 0.5277000069618225,
+      "learning_rate": 0.00012996742671009771,
+      "loss": 0.7853,
+      "step": 833
+    },
+    {
+      "epoch": 1.8319604612850082,
+      "grad_norm": 0.4625236988067627,
+      "learning_rate": 0.0001297231270358306,
+      "loss": 0.9124,
+      "step": 834
+    },
+    {
+      "epoch": 1.8341570565623284,
+      "grad_norm": 0.5138170719146729,
+      "learning_rate": 0.00012947882736156352,
+      "loss": 0.8158,
+      "step": 835
+    },
+    {
+      "epoch": 1.8363536518396484,
+      "grad_norm": 0.5177038311958313,
+      "learning_rate": 0.0001292345276872964,
+      "loss": 0.5932,
+      "step": 836
+    },
+    {
+      "epoch": 1.8385502471169688,
+      "grad_norm": 0.47715386748313904,
+      "learning_rate": 0.00012899022801302933,
+      "loss": 0.6999,
+      "step": 837
+    },
+    {
+      "epoch": 1.8407468423942888,
+      "grad_norm": 0.7109763026237488,
+      "learning_rate": 0.00012874592833876221,
+      "loss": 0.7681,
+      "step": 838
+    },
+    {
+      "epoch": 1.842943437671609,
+      "grad_norm": 0.5324888229370117,
+      "learning_rate": 0.0001285016286644951,
+      "loss": 0.6549,
+      "step": 839
+    },
+    {
+      "epoch": 1.8451400329489291,
+      "grad_norm": 0.762935221195221,
+      "learning_rate": 0.00012825732899022802,
+      "loss": 0.7465,
+      "step": 840
+    },
+    {
+      "epoch": 1.8473366282262493,
+      "grad_norm": 0.40015485882759094,
+      "learning_rate": 0.0001280130293159609,
+      "loss": 0.7168,
+      "step": 841
+    },
+    {
+      "epoch": 1.8495332235035695,
+      "grad_norm": 0.7393567562103271,
+      "learning_rate": 0.0001277687296416938,
+      "loss": 0.7195,
+      "step": 842
+    },
+    {
+      "epoch": 1.8517298187808895,
+      "grad_norm": 0.45974859595298767,
+      "learning_rate": 0.00012752442996742671,
+      "loss": 0.9483,
+      "step": 843
+    },
+    {
+      "epoch": 1.85392641405821,
+      "grad_norm": 0.5409133434295654,
+      "learning_rate": 0.0001272801302931596,
+      "loss": 0.7699,
+      "step": 844
+    },
+    {
+      "epoch": 1.8561230093355299,
+      "grad_norm": 0.42055976390838623,
+      "learning_rate": 0.0001270358306188925,
+      "loss": 0.7252,
+      "step": 845
+    },
+    {
+      "epoch": 1.85831960461285,
+      "grad_norm": 0.44278857111930847,
+      "learning_rate": 0.0001267915309446254,
+      "loss": 0.6539,
+      "step": 846
+    },
+    {
+      "epoch": 1.8605161998901703,
+      "grad_norm": 0.7070727348327637,
+      "learning_rate": 0.0001265472312703583,
+      "loss": 0.9178,
+      "step": 847
+    },
+    {
+      "epoch": 1.8627127951674904,
+      "grad_norm": 0.4646122455596924,
+      "learning_rate": 0.0001263029315960912,
+      "loss": 0.6509,
+      "step": 848
+    },
+    {
+      "epoch": 1.8649093904448106,
+      "grad_norm": 0.43875470757484436,
+      "learning_rate": 0.0001260586319218241,
+      "loss": 0.6558,
+      "step": 849
+    },
+    {
+      "epoch": 1.8671059857221306,
+      "grad_norm": 0.41009125113487244,
+      "learning_rate": 0.000125814332247557,
+      "loss": 0.6553,
+      "step": 850
+    },
+    {
+      "epoch": 1.869302580999451,
+      "grad_norm": 0.4891856610774994,
+      "learning_rate": 0.0001255700325732899,
+      "loss": 0.5859,
+      "step": 851
+    },
+    {
+      "epoch": 1.871499176276771,
+      "grad_norm": 0.5452266335487366,
+      "learning_rate": 0.0001253257328990228,
+      "loss": 0.7749,
+      "step": 852
+    },
+    {
+      "epoch": 1.8736957715540912,
+      "grad_norm": 0.48001888394355774,
+      "learning_rate": 0.0001250814332247557,
+      "loss": 0.9401,
+      "step": 853
+    },
+    {
+      "epoch": 1.8758923668314114,
+      "grad_norm": 0.7396098375320435,
+      "learning_rate": 0.0001248371335504886,
+      "loss": 0.9172,
+      "step": 854
+    },
+    {
+      "epoch": 1.8780889621087313,
+      "grad_norm": 0.34491559863090515,
+      "learning_rate": 0.0001245928338762215,
+      "loss": 0.6467,
+      "step": 855
+    },
+    {
+      "epoch": 1.8802855573860517,
+      "grad_norm": 0.5994675755500793,
+      "learning_rate": 0.00012434853420195438,
+      "loss": 0.9318,
+      "step": 856
+    },
+    {
+      "epoch": 1.8824821526633717,
+      "grad_norm": 0.7101975679397583,
+      "learning_rate": 0.0001241042345276873,
+      "loss": 0.6987,
+      "step": 857
+    },
+    {
+      "epoch": 1.884678747940692,
+      "grad_norm": 0.4602426588535309,
+      "learning_rate": 0.0001238599348534202,
+      "loss": 0.8382,
+      "step": 858
+    },
+    {
+      "epoch": 1.886875343218012,
+      "grad_norm": 0.5606499314308167,
+      "learning_rate": 0.00012361563517915308,
+      "loss": 0.6476,
+      "step": 859
+    },
+    {
+      "epoch": 1.8890719384953323,
+      "grad_norm": 0.33069920539855957,
+      "learning_rate": 0.000123371335504886,
+      "loss": 0.6667,
+      "step": 860
+    },
+    {
+      "epoch": 1.8912685337726525,
+      "grad_norm": 0.5912771821022034,
+      "learning_rate": 0.00012312703583061888,
+      "loss": 0.8978,
+      "step": 861
+    },
+    {
+      "epoch": 1.8934651290499724,
+      "grad_norm": 0.5003272891044617,
+      "learning_rate": 0.00012288273615635177,
+      "loss": 0.7651,
+      "step": 862
+    },
+    {
+      "epoch": 1.8956617243272929,
+      "grad_norm": 0.5458565950393677,
+      "learning_rate": 0.0001226384364820847,
+      "loss": 1.0196,
+      "step": 863
+    },
+    {
+      "epoch": 1.8978583196046128,
+      "grad_norm": 0.4693315923213959,
+      "learning_rate": 0.00012239413680781758,
+      "loss": 0.6398,
+      "step": 864
+    },
+    {
+      "epoch": 1.900054914881933,
+      "grad_norm": 0.6220865249633789,
+      "learning_rate": 0.0001221498371335505,
+      "loss": 0.8108,
+      "step": 865
+    },
+    {
+      "epoch": 1.9022515101592532,
+      "grad_norm": 0.6001383066177368,
+      "learning_rate": 0.00012190553745928338,
+      "loss": 0.7301,
+      "step": 866
+    },
+    {
+      "epoch": 1.9044481054365732,
+      "grad_norm": 0.578180730342865,
+      "learning_rate": 0.00012166123778501629,
+      "loss": 0.7002,
+      "step": 867
+    },
+    {
+      "epoch": 1.9066447007138936,
+      "grad_norm": 0.6893333792686462,
+      "learning_rate": 0.00012141693811074918,
+      "loss": 0.9559,
+      "step": 868
+    },
+    {
+      "epoch": 1.9088412959912135,
+      "grad_norm": 0.6146922707557678,
+      "learning_rate": 0.00012117263843648208,
+      "loss": 0.9618,
+      "step": 869
+    },
+    {
+      "epoch": 1.9110378912685337,
+      "grad_norm": 0.3893856108188629,
+      "learning_rate": 0.00012092833876221498,
+      "loss": 0.6906,
+      "step": 870
+    },
+    {
+      "epoch": 1.913234486545854,
+      "grad_norm": 0.6333613395690918,
+      "learning_rate": 0.00012068403908794787,
+      "loss": 0.7067,
+      "step": 871
+    },
+    {
+      "epoch": 1.9154310818231741,
+      "grad_norm": 0.4291081130504608,
+      "learning_rate": 0.00012043973941368077,
+      "loss": 0.8637,
+      "step": 872
+    },
+    {
+      "epoch": 1.9176276771004943,
+      "grad_norm": 0.5443022847175598,
+      "learning_rate": 0.00012019543973941368,
+      "loss": 0.8832,
+      "step": 873
+    },
+    {
+      "epoch": 1.9198242723778143,
+      "grad_norm": 0.4588859975337982,
+      "learning_rate": 0.00011995114006514658,
+      "loss": 1.0252,
+      "step": 874
+    },
+    {
+      "epoch": 1.9220208676551347,
+      "grad_norm": 0.6883907914161682,
+      "learning_rate": 0.00011970684039087947,
+      "loss": 0.8991,
+      "step": 875
+    },
+    {
+      "epoch": 1.9242174629324547,
+      "grad_norm": 0.7925276160240173,
+      "learning_rate": 0.00011946254071661237,
+      "loss": 0.835,
+      "step": 876
+    },
+    {
+      "epoch": 1.9264140582097748,
+      "grad_norm": 0.43276742100715637,
+      "learning_rate": 0.00011921824104234527,
+      "loss": 0.6583,
+      "step": 877
+    },
+    {
+      "epoch": 1.928610653487095,
+      "grad_norm": 0.8298956751823425,
+      "learning_rate": 0.00011897394136807816,
+      "loss": 0.8107,
+      "step": 878
+    },
+    {
+      "epoch": 1.930807248764415,
+      "grad_norm": 0.4175255596637726,
+      "learning_rate": 0.00011872964169381106,
+      "loss": 0.7718,
+      "step": 879
+    },
+    {
+      "epoch": 1.9330038440417354,
+      "grad_norm": 0.6015777587890625,
+      "learning_rate": 0.00011848534201954397,
+      "loss": 0.6683,
+      "step": 880
+    },
+    {
+      "epoch": 1.9352004393190554,
+      "grad_norm": 0.6181275248527527,
+      "learning_rate": 0.00011824104234527687,
+      "loss": 0.8333,
+      "step": 881
+    },
+    {
+      "epoch": 1.9373970345963756,
+      "grad_norm": 0.7126802206039429,
+      "learning_rate": 0.00011799674267100976,
+      "loss": 0.6129,
+      "step": 882
+    },
+    {
+      "epoch": 1.9395936298736958,
+      "grad_norm": 0.6737213730812073,
+      "learning_rate": 0.00011775244299674266,
+      "loss": 0.8774,
+      "step": 883
+    },
+    {
+      "epoch": 1.941790225151016,
+      "grad_norm": 0.5473393797874451,
+      "learning_rate": 0.00011750814332247557,
+      "loss": 0.7971,
+      "step": 884
+    },
+    {
+      "epoch": 1.9439868204283361,
+      "grad_norm": 0.4686277210712433,
+      "learning_rate": 0.00011726384364820845,
+      "loss": 1.1632,
+      "step": 885
+    },
+    {
+      "epoch": 1.9461834157056561,
+      "grad_norm": 0.5200642943382263,
+      "learning_rate": 0.00011701954397394136,
+      "loss": 0.7601,
+      "step": 886
+    },
+    {
+      "epoch": 1.9483800109829765,
+      "grad_norm": 0.3913096487522125,
+      "learning_rate": 0.00011677524429967426,
+      "loss": 0.6688,
+      "step": 887
+    },
+    {
+      "epoch": 1.9505766062602965,
+      "grad_norm": 0.4563502371311188,
+      "learning_rate": 0.00011653094462540716,
+      "loss": 0.6344,
+      "step": 888
+    },
+    {
+      "epoch": 1.9527732015376167,
+      "grad_norm": 0.51047283411026,
+      "learning_rate": 0.00011628664495114005,
+      "loss": 0.7262,
+      "step": 889
+    },
+    {
+      "epoch": 1.9549697968149369,
+      "grad_norm": 0.4879443943500519,
+      "learning_rate": 0.00011604234527687295,
+      "loss": 0.7738,
+      "step": 890
+    },
+    {
+      "epoch": 1.9571663920922568,
+      "grad_norm": 0.3569847345352173,
+      "learning_rate": 0.00011579804560260586,
+      "loss": 0.8864,
+      "step": 891
+    },
+    {
+      "epoch": 1.9593629873695773,
+      "grad_norm": 0.7129375338554382,
+      "learning_rate": 0.00011555374592833876,
+      "loss": 0.6866,
+      "step": 892
+    },
+    {
+      "epoch": 1.9615595826468972,
+      "grad_norm": 0.5206934809684753,
+      "learning_rate": 0.00011530944625407165,
+      "loss": 0.7309,
+      "step": 893
+    },
+    {
+      "epoch": 1.9637561779242174,
+      "grad_norm": 0.8318841457366943,
+      "learning_rate": 0.00011506514657980455,
+      "loss": 0.8257,
+      "step": 894
+    },
+    {
+      "epoch": 1.9659527732015376,
+      "grad_norm": 0.5583496689796448,
+      "learning_rate": 0.00011482084690553745,
+      "loss": 0.5842,
+      "step": 895
+    },
+    {
+      "epoch": 1.9681493684788578,
+      "grad_norm": 0.5960033535957336,
+      "learning_rate": 0.00011457654723127034,
+      "loss": 0.9168,
+      "step": 896
+    },
+    {
+      "epoch": 1.970345963756178,
+      "grad_norm": 0.3846653401851654,
+      "learning_rate": 0.00011433224755700325,
+      "loss": 0.6421,
+      "step": 897
+    },
+    {
+      "epoch": 1.972542559033498,
+      "grad_norm": 0.5396634936332703,
+      "learning_rate": 0.00011408794788273615,
+      "loss": 0.9012,
+      "step": 898
+    },
+    {
+      "epoch": 1.9747391543108184,
+      "grad_norm": 0.7258114218711853,
+      "learning_rate": 0.00011384364820846905,
+      "loss": 0.8932,
+      "step": 899
+    },
+    {
+      "epoch": 1.9769357495881383,
+      "grad_norm": 0.5657538771629333,
+      "learning_rate": 0.00011359934853420194,
+      "loss": 0.6597,
+      "step": 900
+    },
+    {
+      "epoch": 1.9791323448654585,
+      "grad_norm": 0.528349757194519,
+      "learning_rate": 0.00011335504885993484,
+      "loss": 0.8229,
+      "step": 901
+    },
+    {
+      "epoch": 1.9813289401427787,
+      "grad_norm": 0.578503429889679,
+      "learning_rate": 0.00011311074918566775,
+      "loss": 0.8544,
+      "step": 902
+    },
+    {
+      "epoch": 1.9835255354200987,
+      "grad_norm": 0.4433348476886749,
+      "learning_rate": 0.00011286644951140064,
+      "loss": 0.7496,
+      "step": 903
+    },
+    {
+      "epoch": 1.985722130697419,
+      "grad_norm": 1.9704525470733643,
+      "learning_rate": 0.00011262214983713354,
+      "loss": 0.6258,
+      "step": 904
+    },
+    {
+      "epoch": 1.987918725974739,
+      "grad_norm": 0.6201616525650024,
+      "learning_rate": 0.00011237785016286644,
+      "loss": 0.6199,
+      "step": 905
+    },
+    {
+      "epoch": 1.9901153212520593,
+      "grad_norm": 0.44435518980026245,
+      "learning_rate": 0.00011213355048859934,
+      "loss": 0.8849,
+      "step": 906
+    },
+    {
+      "epoch": 1.9923119165293794,
+      "grad_norm": 0.715987503528595,
+      "learning_rate": 0.00011188925081433223,
+      "loss": 0.5931,
+      "step": 907
+    },
+    {
+      "epoch": 1.9945085118066996,
+      "grad_norm": 0.36977577209472656,
+      "learning_rate": 0.00011164495114006514,
+      "loss": 0.6175,
+      "step": 908
+    },
+    {
+      "epoch": 1.9967051070840198,
+      "grad_norm": 0.8635120391845703,
+      "learning_rate": 0.00011140065146579804,
+      "loss": 0.554,
+      "step": 909
+    },
+    {
+      "epoch": 1.9989017023613398,
+      "grad_norm": 0.8785513043403625,
+      "learning_rate": 0.00011115635179153093,
+      "loss": 0.7709,
+      "step": 910
+    },
+    {
+      "epoch": 2.00109829763866,
+      "grad_norm": 0.5983577966690063,
+      "learning_rate": 0.00011091205211726383,
+      "loss": 0.8763,
+      "step": 911
+    },
+    {
+      "epoch": 2.00329489291598,
+      "grad_norm": 0.532983660697937,
+      "learning_rate": 0.00011066775244299673,
+      "loss": 0.6721,
+      "step": 912
+    },
+    {
+      "epoch": 2.0054914881933006,
+      "grad_norm": 0.42840269207954407,
+      "learning_rate": 0.00011042345276872964,
+      "loss": 0.6033,
+      "step": 913
+    },
+    {
+      "epoch": 2.0076880834706206,
+      "grad_norm": 2.3540680408477783,
+      "learning_rate": 0.00011017915309446253,
+      "loss": 0.7574,
+      "step": 914
+    },
+    {
+      "epoch": 2.0098846787479405,
+      "grad_norm": 0.4376530647277832,
+      "learning_rate": 0.00010993485342019543,
+      "loss": 0.6788,
+      "step": 915
+    },
+    {
+      "epoch": 2.012081274025261,
+      "grad_norm": 0.48186030983924866,
+      "learning_rate": 0.00010969055374592833,
+      "loss": 0.9389,
+      "step": 916
+    },
+    {
+      "epoch": 2.014277869302581,
+      "grad_norm": 0.6513840556144714,
+      "learning_rate": 0.00010944625407166122,
+      "loss": 0.9325,
+      "step": 917
+    },
+    {
+      "epoch": 2.0164744645799013,
+      "grad_norm": 0.6486092805862427,
+      "learning_rate": 0.00010920195439739412,
+      "loss": 0.8937,
+      "step": 918
+    },
+    {
+      "epoch": 2.0186710598572213,
+      "grad_norm": 0.8259550333023071,
+      "learning_rate": 0.00010895765472312703,
+      "loss": 0.6786,
+      "step": 919
+    },
+    {
+      "epoch": 2.0208676551345413,
+      "grad_norm": 0.40635544061660767,
+      "learning_rate": 0.00010871335504885993,
+      "loss": 0.6621,
+      "step": 920
+    },
+    {
+      "epoch": 2.0230642504118617,
+      "grad_norm": 0.5063907504081726,
+      "learning_rate": 0.00010846905537459282,
+      "loss": 0.6345,
+      "step": 921
+    },
+    {
+      "epoch": 2.0252608456891816,
+      "grad_norm": 0.5356837511062622,
+      "learning_rate": 0.00010822475570032572,
+      "loss": 0.7682,
+      "step": 922
+    },
+    {
+      "epoch": 2.027457440966502,
+      "grad_norm": 0.6577078700065613,
+      "learning_rate": 0.00010798045602605862,
+      "loss": 0.7868,
+      "step": 923
+    },
+    {
+      "epoch": 2.029654036243822,
+      "grad_norm": 0.6087715029716492,
+      "learning_rate": 0.00010773615635179153,
+      "loss": 0.8738,
+      "step": 924
+    },
+    {
+      "epoch": 2.0318506315211424,
+      "grad_norm": 0.5042778253555298,
+      "learning_rate": 0.00010749185667752442,
+      "loss": 0.6842,
+      "step": 925
+    },
+    {
+      "epoch": 2.0340472267984624,
+      "grad_norm": 0.47653740644454956,
+      "learning_rate": 0.00010724755700325732,
+      "loss": 0.7682,
+      "step": 926
+    },
+    {
+      "epoch": 2.0362438220757824,
+      "grad_norm": 0.5474477410316467,
+      "learning_rate": 0.00010700325732899022,
+      "loss": 0.8883,
+      "step": 927
+    },
+    {
+      "epoch": 2.0384404173531028,
+      "grad_norm": 0.3539595901966095,
+      "learning_rate": 0.00010675895765472311,
+      "loss": 0.7013,
+      "step": 928
+    },
+    {
+      "epoch": 2.0406370126304227,
+      "grad_norm": 0.47084885835647583,
+      "learning_rate": 0.00010651465798045601,
+      "loss": 0.6776,
+      "step": 929
+    },
+    {
+      "epoch": 2.042833607907743,
+      "grad_norm": 0.3642069697380066,
+      "learning_rate": 0.00010627035830618892,
+      "loss": 0.5999,
+      "step": 930
+    },
+    {
+      "epoch": 2.045030203185063,
+      "grad_norm": 0.4179600179195404,
+      "learning_rate": 0.00010602605863192182,
+      "loss": 0.6904,
+      "step": 931
+    },
+    {
+      "epoch": 2.047226798462383,
+      "grad_norm": 0.5462180376052856,
+      "learning_rate": 0.00010578175895765471,
+      "loss": 0.7425,
+      "step": 932
+    },
+    {
+      "epoch": 2.0494233937397035,
+      "grad_norm": 0.5287339091300964,
+      "learning_rate": 0.00010553745928338761,
+      "loss": 0.7302,
+      "step": 933
+    },
+    {
+      "epoch": 2.0516199890170235,
+      "grad_norm": 0.5358802676200867,
+      "learning_rate": 0.00010529315960912051,
+      "loss": 0.6654,
+      "step": 934
+    },
+    {
+      "epoch": 2.053816584294344,
+      "grad_norm": 0.4570070505142212,
+      "learning_rate": 0.0001050488599348534,
+      "loss": 0.6943,
+      "step": 935
+    },
+    {
+      "epoch": 2.056013179571664,
+      "grad_norm": 0.6630257368087769,
+      "learning_rate": 0.0001048045602605863,
+      "loss": 0.8802,
+      "step": 936
+    },
+    {
+      "epoch": 2.0582097748489843,
+      "grad_norm": 0.7332631349563599,
+      "learning_rate": 0.00010456026058631921,
+      "loss": 0.7899,
+      "step": 937
+    },
+    {
+      "epoch": 2.0604063701263042,
+      "grad_norm": 0.44699570536613464,
+      "learning_rate": 0.00010431596091205211,
+      "loss": 0.5469,
+      "step": 938
+    },
+    {
+      "epoch": 2.062602965403624,
+      "grad_norm": 0.5481554865837097,
+      "learning_rate": 0.000104071661237785,
+      "loss": 0.8774,
+      "step": 939
+    },
+    {
+      "epoch": 2.0647995606809446,
+      "grad_norm": 0.5326241254806519,
+      "learning_rate": 0.0001038273615635179,
+      "loss": 0.7054,
+      "step": 940
+    },
+    {
+      "epoch": 2.0669961559582646,
+      "grad_norm": 0.46254974603652954,
+      "learning_rate": 0.0001035830618892508,
+      "loss": 0.744,
+      "step": 941
+    },
+    {
+      "epoch": 2.069192751235585,
+      "grad_norm": 0.48701396584510803,
+      "learning_rate": 0.0001033387622149837,
+      "loss": 0.6039,
+      "step": 942
+    },
+    {
+      "epoch": 2.071389346512905,
+      "grad_norm": 0.44532889127731323,
+      "learning_rate": 0.0001030944625407166,
+      "loss": 0.7572,
+      "step": 943
+    },
+    {
+      "epoch": 2.073585941790225,
+      "grad_norm": 0.5286267995834351,
+      "learning_rate": 0.0001028501628664495,
+      "loss": 0.6644,
+      "step": 944
+    },
+    {
+      "epoch": 2.0757825370675453,
+      "grad_norm": 0.4091227948665619,
+      "learning_rate": 0.0001026058631921824,
+      "loss": 0.7045,
+      "step": 945
+    },
+    {
+      "epoch": 2.0779791323448653,
+      "grad_norm": 0.39404329657554626,
+      "learning_rate": 0.00010236156351791529,
+      "loss": 0.7216,
+      "step": 946
+    },
+    {
+      "epoch": 2.0801757276221857,
+      "grad_norm": 0.44896674156188965,
+      "learning_rate": 0.0001021172638436482,
+      "loss": 0.5797,
+      "step": 947
+    },
+    {
+      "epoch": 2.0823723228995057,
+      "grad_norm": 0.7126267552375793,
+      "learning_rate": 0.0001018729641693811,
+      "loss": 0.6604,
+      "step": 948
+    },
+    {
+      "epoch": 2.084568918176826,
+      "grad_norm": 0.4855166971683502,
+      "learning_rate": 0.00010162866449511399,
+      "loss": 0.8652,
+      "step": 949
+    },
+    {
+      "epoch": 2.086765513454146,
+      "grad_norm": 0.43222448229789734,
+      "learning_rate": 0.00010138436482084689,
+      "loss": 0.7487,
+      "step": 950
+    },
+    {
+      "epoch": 2.088962108731466,
+      "grad_norm": 1.3848443031311035,
+      "learning_rate": 0.00010114006514657979,
+      "loss": 0.7458,
+      "step": 951
+    },
+    {
+      "epoch": 2.0911587040087865,
+      "grad_norm": 0.890978991985321,
+      "learning_rate": 0.0001008957654723127,
+      "loss": 0.8127,
+      "step": 952
+    },
+    {
+      "epoch": 2.0933552992861064,
+      "grad_norm": 0.986899197101593,
+      "learning_rate": 0.00010065146579804558,
+      "loss": 0.8401,
+      "step": 953
+    },
+    {
+      "epoch": 2.095551894563427,
+      "grad_norm": 0.48363620042800903,
+      "learning_rate": 0.00010040716612377849,
+      "loss": 0.6759,
+      "step": 954
+    },
+    {
+      "epoch": 2.097748489840747,
+      "grad_norm": 0.4698428511619568,
+      "learning_rate": 0.00010016286644951139,
+      "loss": 0.8038,
+      "step": 955
+    },
+    {
+      "epoch": 2.099945085118067,
+      "grad_norm": 0.5993090271949768,
+      "learning_rate": 9.991856677524428e-05,
+      "loss": 0.7457,
+      "step": 956
+    },
+    {
+      "epoch": 2.102141680395387,
+      "grad_norm": 0.6166810393333435,
+      "learning_rate": 9.967426710097718e-05,
+      "loss": 0.7691,
+      "step": 957
+    },
+    {
+      "epoch": 2.104338275672707,
+      "grad_norm": 0.47788071632385254,
+      "learning_rate": 9.942996742671008e-05,
+      "loss": 0.6685,
+      "step": 958
+    },
+    {
+      "epoch": 2.1065348709500276,
+      "grad_norm": 0.47276636958122253,
+      "learning_rate": 9.918566775244299e-05,
+      "loss": 0.7159,
+      "step": 959
+    },
+    {
+      "epoch": 2.1087314662273475,
+      "grad_norm": 0.43806830048561096,
+      "learning_rate": 9.894136807817588e-05,
+      "loss": 0.7239,
+      "step": 960
+    },
+    {
+      "epoch": 2.110928061504668,
+      "grad_norm": 0.5375027656555176,
+      "learning_rate": 9.869706840390878e-05,
+      "loss": 0.8618,
+      "step": 961
+    },
+    {
+      "epoch": 2.113124656781988,
+      "grad_norm": 0.6547253131866455,
+      "learning_rate": 9.845276872964168e-05,
+      "loss": 0.6979,
+      "step": 962
+    },
+    {
+      "epoch": 2.115321252059308,
+      "grad_norm": 0.619748055934906,
+      "learning_rate": 9.820846905537458e-05,
+      "loss": 0.6,
+      "step": 963
+    },
+    {
+      "epoch": 2.1175178473366283,
+      "grad_norm": 0.44008174538612366,
+      "learning_rate": 9.796416938110747e-05,
+      "loss": 0.7251,
+      "step": 964
+    },
+    {
+      "epoch": 2.1197144426139483,
+      "grad_norm": 0.4453674554824829,
+      "learning_rate": 9.771986970684038e-05,
+      "loss": 0.7009,
+      "step": 965
+    },
+    {
+      "epoch": 2.1219110378912687,
+      "grad_norm": 2.2442803382873535,
+      "learning_rate": 9.747557003257328e-05,
+      "loss": 0.6744,
+      "step": 966
+    },
+    {
+      "epoch": 2.1241076331685886,
+      "grad_norm": 0.4414275884628296,
+      "learning_rate": 9.723127035830617e-05,
+      "loss": 0.6801,
+      "step": 967
+    },
+    {
+      "epoch": 2.1263042284459086,
+      "grad_norm": 0.4093416631221771,
+      "learning_rate": 9.698697068403907e-05,
+      "loss": 0.6807,
+      "step": 968
+    },
+    {
+      "epoch": 2.128500823723229,
+      "grad_norm": 0.49372339248657227,
+      "learning_rate": 9.674267100977197e-05,
+      "loss": 0.6325,
+      "step": 969
+    },
+    {
+      "epoch": 2.130697419000549,
+      "grad_norm": 0.6541997790336609,
+      "learning_rate": 9.649837133550488e-05,
+      "loss": 0.8546,
+      "step": 970
+    },
+    {
+      "epoch": 2.1328940142778694,
+      "grad_norm": 0.7643210291862488,
+      "learning_rate": 9.625407166123777e-05,
+      "loss": 0.6952,
+      "step": 971
+    },
+    {
+      "epoch": 2.1350906095551894,
+      "grad_norm": 0.5390772223472595,
+      "learning_rate": 9.600977198697067e-05,
+      "loss": 0.8211,
+      "step": 972
+    },
+    {
+      "epoch": 2.13728720483251,
+      "grad_norm": 0.451479971408844,
+      "learning_rate": 9.576547231270357e-05,
+      "loss": 0.7138,
+      "step": 973
+    },
+    {
+      "epoch": 2.1394838001098297,
+      "grad_norm": 0.5709292888641357,
+      "learning_rate": 9.552117263843646e-05,
+      "loss": 0.77,
+      "step": 974
+    },
+    {
+      "epoch": 2.1416803953871497,
+      "grad_norm": 0.5563143491744995,
+      "learning_rate": 9.527687296416936e-05,
+      "loss": 0.8098,
+      "step": 975
+    },
+    {
+      "epoch": 2.14387699066447,
+      "grad_norm": 0.4982394874095917,
+      "learning_rate": 9.503257328990227e-05,
+      "loss": 0.7924,
+      "step": 976
+    },
+    {
+      "epoch": 2.14607358594179,
+      "grad_norm": 0.35699889063835144,
+      "learning_rate": 9.478827361563517e-05,
+      "loss": 0.7496,
+      "step": 977
+    },
+    {
+      "epoch": 2.1482701812191105,
+      "grad_norm": 0.4535363018512726,
+      "learning_rate": 9.454397394136806e-05,
+      "loss": 0.6395,
+      "step": 978
+    },
+    {
+      "epoch": 2.1504667764964305,
+      "grad_norm": 0.36046692728996277,
+      "learning_rate": 9.429967426710096e-05,
+      "loss": 0.7138,
+      "step": 979
+    },
+    {
+      "epoch": 2.152663371773751,
+      "grad_norm": 1.0834375619888306,
+      "learning_rate": 9.405537459283386e-05,
+      "loss": 0.8217,
+      "step": 980
+    },
+    {
+      "epoch": 2.154859967051071,
+      "grad_norm": 0.7800245881080627,
+      "learning_rate": 9.381107491856675e-05,
+      "loss": 0.5754,
+      "step": 981
+    },
+    {
+      "epoch": 2.157056562328391,
+      "grad_norm": 0.4859004616737366,
+      "learning_rate": 9.356677524429967e-05,
+      "loss": 0.608,
+      "step": 982
+    },
+    {
+      "epoch": 2.1592531576057112,
+      "grad_norm": 0.678435206413269,
+      "learning_rate": 9.332247557003257e-05,
+      "loss": 0.8361,
+      "step": 983
+    },
+    {
+      "epoch": 2.161449752883031,
+      "grad_norm": 0.5359833240509033,
+      "learning_rate": 9.307817589576547e-05,
+      "loss": 0.6269,
+      "step": 984
+    },
+    {
+      "epoch": 2.1636463481603516,
+      "grad_norm": 0.4197586178779602,
+      "learning_rate": 9.283387622149838e-05,
+      "loss": 0.63,
+      "step": 985
+    },
+    {
+      "epoch": 2.1658429434376716,
+      "grad_norm": 0.5479422211647034,
+      "learning_rate": 9.258957654723127e-05,
+      "loss": 0.65,
+      "step": 986
+    },
+    {
+      "epoch": 2.168039538714992,
+      "grad_norm": 0.5608152747154236,
+      "learning_rate": 9.234527687296417e-05,
+      "loss": 0.8261,
+      "step": 987
+    },
+    {
+      "epoch": 2.170236133992312,
+      "grad_norm": 0.5804690718650818,
+      "learning_rate": 9.210097719869707e-05,
+      "loss": 0.6296,
+      "step": 988
+    },
+    {
+      "epoch": 2.172432729269632,
+      "grad_norm": 0.4730856120586395,
+      "learning_rate": 9.185667752442996e-05,
+      "loss": 0.7183,
+      "step": 989
+    },
+    {
+      "epoch": 2.1746293245469523,
+      "grad_norm": 0.44884636998176575,
+      "learning_rate": 9.161237785016286e-05,
+      "loss": 0.8656,
+      "step": 990
+    },
+    {
+      "epoch": 2.1768259198242723,
+      "grad_norm": 0.8014638423919678,
+      "learning_rate": 9.136807817589577e-05,
+      "loss": 0.726,
+      "step": 991
+    },
+    {
+      "epoch": 2.1790225151015927,
+      "grad_norm": 0.36201271414756775,
+      "learning_rate": 9.112377850162867e-05,
+      "loss": 0.6398,
+      "step": 992
+    },
+    {
+      "epoch": 2.1812191103789127,
+      "grad_norm": 0.39075160026550293,
+      "learning_rate": 9.087947882736156e-05,
+      "loss": 0.7947,
+      "step": 993
+    },
+    {
+      "epoch": 2.1834157056562327,
+      "grad_norm": 1.2395297288894653,
+      "learning_rate": 9.063517915309446e-05,
+      "loss": 0.6698,
+      "step": 994
+    },
+    {
+      "epoch": 2.185612300933553,
+      "grad_norm": 0.44856181740760803,
+      "learning_rate": 9.039087947882736e-05,
+      "loss": 0.6901,
+      "step": 995
+    },
+    {
+      "epoch": 2.187808896210873,
+      "grad_norm": 0.8215703964233398,
+      "learning_rate": 9.014657980456025e-05,
+      "loss": 0.8317,
+      "step": 996
+    },
+    {
+      "epoch": 2.1900054914881935,
+      "grad_norm": 0.49343183636665344,
+      "learning_rate": 8.990228013029316e-05,
+      "loss": 0.7871,
+      "step": 997
+    },
+    {
+      "epoch": 2.1922020867655134,
+      "grad_norm": 0.4945894181728363,
+      "learning_rate": 8.965798045602606e-05,
+      "loss": 0.7399,
+      "step": 998
+    },
+    {
+      "epoch": 2.1943986820428334,
+      "grad_norm": 0.4777913987636566,
+      "learning_rate": 8.941368078175896e-05,
+      "loss": 0.8757,
+      "step": 999
+    },
+    {
+      "epoch": 2.196595277320154,
+      "grad_norm": 0.38988450169563293,
+      "learning_rate": 8.916938110749185e-05,
+      "loss": 0.6927,
+      "step": 1000
+    },
+    {
+      "epoch": 2.1987918725974738,
+      "grad_norm": 0.44612404704093933,
+      "learning_rate": 8.892508143322475e-05,
+      "loss": 0.4858,
+      "step": 1001
+    },
+    {
+      "epoch": 2.200988467874794,
+      "grad_norm": 0.44903135299682617,
+      "learning_rate": 8.868078175895766e-05,
+      "loss": 0.6434,
+      "step": 1002
+    },
+    {
+      "epoch": 2.203185063152114,
+      "grad_norm": 0.6219984889030457,
+      "learning_rate": 8.843648208469055e-05,
+      "loss": 0.6863,
+      "step": 1003
+    },
+    {
+      "epoch": 2.2053816584294346,
+      "grad_norm": 0.6907383799552917,
+      "learning_rate": 8.819218241042345e-05,
+      "loss": 0.6245,
+      "step": 1004
+    },
+    {
+      "epoch": 2.2075782537067545,
+      "grad_norm": 0.8279937505722046,
+      "learning_rate": 8.794788273615635e-05,
+      "loss": 0.6778,
+      "step": 1005
+    },
+    {
+      "epoch": 2.2097748489840745,
+      "grad_norm": 0.6698524355888367,
+      "learning_rate": 8.770358306188925e-05,
+      "loss": 0.7507,
+      "step": 1006
+    },
+    {
+      "epoch": 2.211971444261395,
+      "grad_norm": 0.451610267162323,
+      "learning_rate": 8.745928338762214e-05,
+      "loss": 0.61,
+      "step": 1007
+    },
+    {
+      "epoch": 2.214168039538715,
+      "grad_norm": 0.7460264563560486,
+      "learning_rate": 8.721498371335505e-05,
+      "loss": 0.9192,
+      "step": 1008
+    },
+    {
+      "epoch": 2.2163646348160353,
+      "grad_norm": 1.2460687160491943,
+      "learning_rate": 8.697068403908795e-05,
+      "loss": 0.7664,
+      "step": 1009
+    },
+    {
+      "epoch": 2.2185612300933553,
+      "grad_norm": 1.1255820989608765,
+      "learning_rate": 8.672638436482085e-05,
+      "loss": 0.711,
+      "step": 1010
+    },
+    {
+      "epoch": 2.2207578253706757,
+      "grad_norm": 0.41432857513427734,
+      "learning_rate": 8.648208469055374e-05,
+      "loss": 0.7211,
+      "step": 1011
+    },
+    {
+      "epoch": 2.2229544206479956,
+      "grad_norm": 0.5486708879470825,
+      "learning_rate": 8.623778501628664e-05,
+      "loss": 0.9237,
+      "step": 1012
+    },
+    {
+      "epoch": 2.2251510159253156,
+      "grad_norm": 1.0036720037460327,
+      "learning_rate": 8.599348534201955e-05,
+      "loss": 0.7353,
+      "step": 1013
+    },
+    {
+      "epoch": 2.227347611202636,
+      "grad_norm": 0.5861718058586121,
+      "learning_rate": 8.574918566775243e-05,
+      "loss": 0.8733,
+      "step": 1014
+    },
+    {
+      "epoch": 2.229544206479956,
+      "grad_norm": 0.5786802768707275,
+      "learning_rate": 8.550488599348534e-05,
+      "loss": 0.8344,
+      "step": 1015
+    },
+    {
+      "epoch": 2.2317408017572764,
+      "grad_norm": 0.48991167545318604,
+      "learning_rate": 8.526058631921824e-05,
+      "loss": 0.7554,
+      "step": 1016
+    },
+    {
+      "epoch": 2.2339373970345964,
+      "grad_norm": 0.5740125775337219,
+      "learning_rate": 8.501628664495114e-05,
+      "loss": 0.8727,
+      "step": 1017
+    },
+    {
+      "epoch": 2.2361339923119163,
+      "grad_norm": 0.6309413313865662,
+      "learning_rate": 8.477198697068403e-05,
+      "loss": 0.8942,
+      "step": 1018
+    },
+    {
+      "epoch": 2.2383305875892368,
+      "grad_norm": 0.4485383629798889,
+      "learning_rate": 8.452768729641694e-05,
+      "loss": 0.8229,
+      "step": 1019
+    },
+    {
+      "epoch": 2.2405271828665567,
+      "grad_norm": 0.5865241885185242,
+      "learning_rate": 8.428338762214984e-05,
+      "loss": 0.6924,
+      "step": 1020
+    },
+    {
+      "epoch": 2.242723778143877,
+      "grad_norm": 0.41497427225112915,
+      "learning_rate": 8.403908794788273e-05,
+      "loss": 0.7168,
+      "step": 1021
+    },
+    {
+      "epoch": 2.244920373421197,
+      "grad_norm": 0.4187897741794586,
+      "learning_rate": 8.379478827361563e-05,
+      "loss": 0.5747,
+      "step": 1022
+    },
+    {
+      "epoch": 2.247116968698517,
+      "grad_norm": 0.512890636920929,
+      "learning_rate": 8.355048859934853e-05,
+      "loss": 0.8112,
+      "step": 1023
+    },
+    {
+      "epoch": 2.2493135639758375,
+      "grad_norm": 0.585892379283905,
+      "learning_rate": 8.330618892508144e-05,
+      "loss": 0.6756,
+      "step": 1024
+    },
+    {
+      "epoch": 2.2515101592531575,
+      "grad_norm": 0.36007410287857056,
+      "learning_rate": 8.306188925081432e-05,
+      "loss": 0.6859,
+      "step": 1025
+    },
+    {
+      "epoch": 2.253706754530478,
+      "grad_norm": 0.43299710750579834,
+      "learning_rate": 8.281758957654723e-05,
+      "loss": 0.6505,
+      "step": 1026
+    },
+    {
+      "epoch": 2.255903349807798,
+      "grad_norm": 0.8015580177307129,
+      "learning_rate": 8.257328990228013e-05,
+      "loss": 0.7275,
+      "step": 1027
+    },
+    {
+      "epoch": 2.2580999450851182,
+      "grad_norm": 0.37666407227516174,
+      "learning_rate": 8.232899022801302e-05,
+      "loss": 0.8342,
+      "step": 1028
+    },
+    {
+      "epoch": 2.260296540362438,
+      "grad_norm": 0.5178048014640808,
+      "learning_rate": 8.208469055374592e-05,
+      "loss": 0.7893,
+      "step": 1029
+    },
+    {
+      "epoch": 2.262493135639758,
+      "grad_norm": 0.7453083395957947,
+      "learning_rate": 8.184039087947882e-05,
+      "loss": 0.6552,
+      "step": 1030
+    },
+    {
+      "epoch": 2.2646897309170786,
+      "grad_norm": 0.6306135058403015,
+      "learning_rate": 8.159609120521173e-05,
+      "loss": 0.6917,
+      "step": 1031
+    },
+    {
+      "epoch": 2.2668863261943986,
+      "grad_norm": 1.4548498392105103,
+      "learning_rate": 8.135179153094462e-05,
+      "loss": 0.6097,
+      "step": 1032
+    },
+    {
+      "epoch": 2.269082921471719,
+      "grad_norm": 0.5184520483016968,
+      "learning_rate": 8.110749185667752e-05,
+      "loss": 0.9145,
+      "step": 1033
+    },
+    {
+      "epoch": 2.271279516749039,
+      "grad_norm": 0.5624165534973145,
+      "learning_rate": 8.086319218241042e-05,
+      "loss": 0.7066,
+      "step": 1034
+    },
+    {
+      "epoch": 2.2734761120263594,
+      "grad_norm": 0.732881486415863,
+      "learning_rate": 8.061889250814331e-05,
+      "loss": 0.6515,
+      "step": 1035
+    },
+    {
+      "epoch": 2.2756727073036793,
+      "grad_norm": 0.4438563883304596,
+      "learning_rate": 8.037459283387621e-05,
+      "loss": 0.6712,
+      "step": 1036
+    },
+    {
+      "epoch": 2.2778693025809993,
+      "grad_norm": 0.5677716732025146,
+      "learning_rate": 8.013029315960912e-05,
+      "loss": 0.637,
+      "step": 1037
+    },
+    {
+      "epoch": 2.2800658978583197,
+      "grad_norm": 0.37578827142715454,
+      "learning_rate": 7.988599348534202e-05,
+      "loss": 0.6479,
+      "step": 1038
+    },
+    {
+      "epoch": 2.2822624931356397,
+      "grad_norm": 0.4222707450389862,
+      "learning_rate": 7.964169381107491e-05,
+      "loss": 0.6441,
+      "step": 1039
+    },
+    {
+      "epoch": 2.28445908841296,
+      "grad_norm": 0.5215080976486206,
+      "learning_rate": 7.939739413680781e-05,
+      "loss": 0.7466,
+      "step": 1040
+    },
+    {
+      "epoch": 2.28665568369028,
+      "grad_norm": 0.5852795243263245,
+      "learning_rate": 7.915309446254071e-05,
+      "loss": 0.72,
+      "step": 1041
+    },
+    {
+      "epoch": 2.2888522789676005,
+      "grad_norm": 0.500820517539978,
+      "learning_rate": 7.890879478827362e-05,
+      "loss": 0.7964,
+      "step": 1042
+    },
+    {
+      "epoch": 2.2910488742449204,
+      "grad_norm": 0.3418845534324646,
+      "learning_rate": 7.86644951140065e-05,
+      "loss": 0.5554,
+      "step": 1043
+    },
+    {
+      "epoch": 2.2932454695222404,
+      "grad_norm": 0.529072105884552,
+      "learning_rate": 7.842019543973941e-05,
+      "loss": 0.7306,
+      "step": 1044
+    },
+    {
+      "epoch": 2.295442064799561,
+      "grad_norm": 0.4056017994880676,
+      "learning_rate": 7.817589576547231e-05,
+      "loss": 0.8041,
+      "step": 1045
+    },
+    {
+      "epoch": 2.297638660076881,
+      "grad_norm": 0.48954829573631287,
+      "learning_rate": 7.79315960912052e-05,
+      "loss": 0.6671,
+      "step": 1046
+    },
+    {
+      "epoch": 2.2998352553542007,
+      "grad_norm": 0.49989765882492065,
+      "learning_rate": 7.76872964169381e-05,
+      "loss": 0.555,
+      "step": 1047
+    },
+    {
+      "epoch": 2.302031850631521,
+      "grad_norm": 0.6114453077316284,
+      "learning_rate": 7.7442996742671e-05,
+      "loss": 0.714,
+      "step": 1048
+    },
+    {
+      "epoch": 2.304228445908841,
+      "grad_norm": 0.4889865517616272,
+      "learning_rate": 7.719869706840391e-05,
+      "loss": 0.7135,
+      "step": 1049
+    },
+    {
+      "epoch": 2.3064250411861615,
+      "grad_norm": 0.3559395968914032,
+      "learning_rate": 7.69543973941368e-05,
+      "loss": 0.7366,
+      "step": 1050
+    },
+    {
+      "epoch": 2.3086216364634815,
+      "grad_norm": 0.5073962807655334,
+      "learning_rate": 7.67100977198697e-05,
+      "loss": 0.5576,
+      "step": 1051
+    },
+    {
+      "epoch": 2.310818231740802,
+      "grad_norm": 0.40690386295318604,
+      "learning_rate": 7.64657980456026e-05,
+      "loss": 0.6504,
+      "step": 1052
+    },
+    {
+      "epoch": 2.313014827018122,
+      "grad_norm": 0.46325579285621643,
+      "learning_rate": 7.622149837133549e-05,
+      "loss": 0.7456,
+      "step": 1053
+    },
+    {
+      "epoch": 2.315211422295442,
+      "grad_norm": 0.5068777203559875,
+      "learning_rate": 7.59771986970684e-05,
+      "loss": 0.7093,
+      "step": 1054
+    },
+    {
+      "epoch": 2.3174080175727623,
+      "grad_norm": 0.4510020315647125,
+      "learning_rate": 7.57328990228013e-05,
+      "loss": 0.8939,
+      "step": 1055
+    },
+    {
+      "epoch": 2.3196046128500822,
+      "grad_norm": 0.6199316382408142,
+      "learning_rate": 7.54885993485342e-05,
+      "loss": 0.6979,
+      "step": 1056
+    },
+    {
+      "epoch": 2.3218012081274026,
+      "grad_norm": 0.5785857439041138,
+      "learning_rate": 7.524429967426709e-05,
+      "loss": 0.9219,
+      "step": 1057
+    },
+    {
+      "epoch": 2.3239978034047226,
+      "grad_norm": 0.42037156224250793,
+      "learning_rate": 7.5e-05,
+      "loss": 0.5192,
+      "step": 1058
+    },
+    {
+      "epoch": 2.326194398682043,
+      "grad_norm": 0.4723362326622009,
+      "learning_rate": 7.47557003257329e-05,
+      "loss": 0.6653,
+      "step": 1059
+    },
+    {
+      "epoch": 2.328390993959363,
+      "grad_norm": 0.46067509055137634,
+      "learning_rate": 7.451140065146579e-05,
+      "loss": 0.6881,
+      "step": 1060
+    },
+    {
+      "epoch": 2.330587589236683,
+      "grad_norm": 0.4634696841239929,
+      "learning_rate": 7.426710097719869e-05,
+      "loss": 0.7007,
+      "step": 1061
+    },
+    {
+      "epoch": 2.3327841845140034,
+      "grad_norm": 0.4090059697628021,
+      "learning_rate": 7.402280130293159e-05,
+      "loss": 0.7251,
+      "step": 1062
+    },
+    {
+      "epoch": 2.3349807797913233,
+      "grad_norm": 0.42291703820228577,
+      "learning_rate": 7.37785016286645e-05,
+      "loss": 0.8174,
+      "step": 1063
+    },
+    {
+      "epoch": 2.3371773750686438,
+      "grad_norm": 0.5655874013900757,
+      "learning_rate": 7.353420195439738e-05,
+      "loss": 0.7782,
+      "step": 1064
+    },
+    {
+      "epoch": 2.3393739703459637,
+      "grad_norm": 0.8801429271697998,
+      "learning_rate": 7.328990228013029e-05,
+      "loss": 0.7202,
+      "step": 1065
+    },
+    {
+      "epoch": 2.341570565623284,
+      "grad_norm": 0.39593833684921265,
+      "learning_rate": 7.304560260586319e-05,
+      "loss": 0.8527,
+      "step": 1066
+    },
+    {
+      "epoch": 2.343767160900604,
+      "grad_norm": 0.5451957583427429,
+      "learning_rate": 7.280130293159608e-05,
+      "loss": 0.6096,
+      "step": 1067
+    },
+    {
+      "epoch": 2.345963756177924,
+      "grad_norm": 0.6000787615776062,
+      "learning_rate": 7.255700325732898e-05,
+      "loss": 0.7681,
+      "step": 1068
+    },
+    {
+      "epoch": 2.3481603514552445,
+      "grad_norm": 0.8677886128425598,
+      "learning_rate": 7.231270358306188e-05,
+      "loss": 0.8924,
+      "step": 1069
+    },
+    {
+      "epoch": 2.3503569467325645,
+      "grad_norm": 0.5110235214233398,
+      "learning_rate": 7.206840390879479e-05,
+      "loss": 0.6775,
+      "step": 1070
+    },
+    {
+      "epoch": 2.352553542009885,
+      "grad_norm": 0.5173139572143555,
+      "learning_rate": 7.182410423452768e-05,
+      "loss": 0.7653,
+      "step": 1071
+    },
+    {
+      "epoch": 2.354750137287205,
+      "grad_norm": 0.5598890781402588,
+      "learning_rate": 7.157980456026058e-05,
+      "loss": 0.6057,
+      "step": 1072
+    },
+    {
+      "epoch": 2.356946732564525,
+      "grad_norm": 1.2639806270599365,
+      "learning_rate": 7.133550488599348e-05,
+      "loss": 0.6118,
+      "step": 1073
+    },
+    {
+      "epoch": 2.359143327841845,
+      "grad_norm": 0.5161561369895935,
+      "learning_rate": 7.109120521172637e-05,
+      "loss": 0.6645,
+      "step": 1074
+    },
+    {
+      "epoch": 2.361339923119165,
+      "grad_norm": 0.5148932337760925,
+      "learning_rate": 7.084690553745927e-05,
+      "loss": 0.8637,
+      "step": 1075
+    },
+    {
+      "epoch": 2.3635365183964856,
+      "grad_norm": 0.6173725128173828,
+      "learning_rate": 7.060260586319218e-05,
+      "loss": 0.756,
+      "step": 1076
+    },
+    {
+      "epoch": 2.3657331136738056,
+      "grad_norm": 0.48136505484580994,
+      "learning_rate": 7.035830618892508e-05,
+      "loss": 0.683,
+      "step": 1077
+    },
+    {
+      "epoch": 2.3679297089511255,
+      "grad_norm": 0.5340093374252319,
+      "learning_rate": 7.011400651465797e-05,
+      "loss": 0.7476,
+      "step": 1078
+    },
+    {
+      "epoch": 2.370126304228446,
+      "grad_norm": 0.4871104955673218,
+      "learning_rate": 6.986970684039087e-05,
+      "loss": 0.5676,
+      "step": 1079
+    },
+    {
+      "epoch": 2.372322899505766,
+      "grad_norm": 0.4526054561138153,
+      "learning_rate": 6.962540716612377e-05,
+      "loss": 0.7103,
+      "step": 1080
+    },
+    {
+      "epoch": 2.3745194947830863,
+      "grad_norm": 0.694290041923523,
+      "learning_rate": 6.938110749185668e-05,
+      "loss": 0.9509,
+      "step": 1081
+    },
+    {
+      "epoch": 2.3767160900604063,
+      "grad_norm": 0.531172513961792,
+      "learning_rate": 6.913680781758956e-05,
+      "loss": 0.6804,
+      "step": 1082
+    },
+    {
+      "epoch": 2.3789126853377267,
+      "grad_norm": 0.4855298399925232,
+      "learning_rate": 6.889250814332247e-05,
+      "loss": 0.7128,
+      "step": 1083
+    },
+    {
+      "epoch": 2.3811092806150467,
+      "grad_norm": 0.5872212648391724,
+      "learning_rate": 6.864820846905537e-05,
+      "loss": 0.8631,
+      "step": 1084
+    },
+    {
+      "epoch": 2.3833058758923666,
+      "grad_norm": 0.6667178273200989,
+      "learning_rate": 6.840390879478826e-05,
+      "loss": 0.9732,
+      "step": 1085
+    },
+    {
+      "epoch": 2.385502471169687,
+      "grad_norm": 0.4494982659816742,
+      "learning_rate": 6.815960912052116e-05,
+      "loss": 0.6362,
+      "step": 1086
+    },
+    {
+      "epoch": 2.387699066447007,
+      "grad_norm": 0.8962666988372803,
+      "learning_rate": 6.791530944625406e-05,
+      "loss": 0.9684,
+      "step": 1087
+    },
+    {
+      "epoch": 2.3898956617243274,
+      "grad_norm": 0.6731237769126892,
+      "learning_rate": 6.767100977198697e-05,
+      "loss": 0.6354,
+      "step": 1088
+    },
+    {
+      "epoch": 2.3920922570016474,
+      "grad_norm": 0.6469890475273132,
+      "learning_rate": 6.742671009771986e-05,
+      "loss": 0.8498,
+      "step": 1089
+    },
+    {
+      "epoch": 2.394288852278968,
+      "grad_norm": 0.44095316529273987,
+      "learning_rate": 6.718241042345276e-05,
+      "loss": 0.6857,
+      "step": 1090
+    },
+    {
+      "epoch": 2.396485447556288,
+      "grad_norm": 0.48514145612716675,
+      "learning_rate": 6.693811074918566e-05,
+      "loss": 0.6839,
+      "step": 1091
+    },
+    {
+      "epoch": 2.3986820428336078,
+      "grad_norm": 0.5867401957511902,
+      "learning_rate": 6.669381107491855e-05,
+      "loss": 0.7365,
+      "step": 1092
+    },
+    {
+      "epoch": 2.400878638110928,
+      "grad_norm": 0.4710720181465149,
+      "learning_rate": 6.644951140065145e-05,
+      "loss": 0.7635,
+      "step": 1093
+    },
+    {
+      "epoch": 2.403075233388248,
+      "grad_norm": 0.3798518776893616,
+      "learning_rate": 6.620521172638436e-05,
+      "loss": 0.784,
+      "step": 1094
+    },
+    {
+      "epoch": 2.4052718286655685,
+      "grad_norm": 0.6328463554382324,
+      "learning_rate": 6.596091205211726e-05,
+      "loss": 0.6407,
+      "step": 1095
+    },
+    {
+      "epoch": 2.4074684239428885,
+      "grad_norm": 0.38195326924324036,
+      "learning_rate": 6.571661237785015e-05,
+      "loss": 0.5881,
+      "step": 1096
+    },
+    {
+      "epoch": 2.409665019220209,
+      "grad_norm": 0.5408303737640381,
+      "learning_rate": 6.547231270358307e-05,
+      "loss": 0.7769,
+      "step": 1097
+    },
+    {
+      "epoch": 2.411861614497529,
+      "grad_norm": 0.34367892146110535,
+      "learning_rate": 6.522801302931595e-05,
+      "loss": 0.757,
+      "step": 1098
+    },
+    {
+      "epoch": 2.414058209774849,
+      "grad_norm": 0.5159469246864319,
+      "learning_rate": 6.498371335504886e-05,
+      "loss": 0.801,
+      "step": 1099
+    },
+    {
+      "epoch": 2.4162548050521693,
+      "grad_norm": 0.536939263343811,
+      "learning_rate": 6.473941368078176e-05,
+      "loss": 0.7673,
+      "step": 1100
+    },
+    {
+      "epoch": 2.4184514003294892,
+      "grad_norm": 0.5042858719825745,
+      "learning_rate": 6.449511400651466e-05,
+      "loss": 0.6754,
+      "step": 1101
+    },
+    {
+      "epoch": 2.420647995606809,
+      "grad_norm": 0.4692958891391754,
+      "learning_rate": 6.425081433224755e-05,
+      "loss": 0.7609,
+      "step": 1102
+    },
+    {
+      "epoch": 2.4228445908841296,
+      "grad_norm": 0.6472638249397278,
+      "learning_rate": 6.400651465798045e-05,
+      "loss": 0.7223,
+      "step": 1103
+    },
+    {
+      "epoch": 2.4250411861614496,
+      "grad_norm": 0.645112156867981,
+      "learning_rate": 6.376221498371336e-05,
+      "loss": 0.7536,
+      "step": 1104
+    },
+    {
+      "epoch": 2.42723778143877,
+      "grad_norm": 0.6586037874221802,
+      "learning_rate": 6.351791530944625e-05,
+      "loss": 0.7716,
+      "step": 1105
+    },
+    {
+      "epoch": 2.42943437671609,
+      "grad_norm": 0.5551490187644958,
+      "learning_rate": 6.327361563517915e-05,
+      "loss": 0.6639,
+      "step": 1106
+    },
+    {
+      "epoch": 2.4316309719934104,
+      "grad_norm": 0.5462861061096191,
+      "learning_rate": 6.302931596091205e-05,
+      "loss": 0.7733,
+      "step": 1107
+    },
+    {
+      "epoch": 2.4338275672707304,
+      "grad_norm": 0.5068262815475464,
+      "learning_rate": 6.278501628664495e-05,
+      "loss": 0.8402,
+      "step": 1108
+    },
+    {
+      "epoch": 2.4360241625480503,
+      "grad_norm": 0.5244752764701843,
+      "learning_rate": 6.254071661237784e-05,
+      "loss": 0.7486,
+      "step": 1109
+    },
+    {
+      "epoch": 2.4382207578253707,
+      "grad_norm": 0.45535188913345337,
+      "learning_rate": 6.229641693811075e-05,
+      "loss": 0.7731,
+      "step": 1110
+    },
+    {
+      "epoch": 2.4404173531026907,
+      "grad_norm": 0.38719743490219116,
+      "learning_rate": 6.205211726384365e-05,
+      "loss": 0.7288,
+      "step": 1111
+    },
+    {
+      "epoch": 2.442613948380011,
+      "grad_norm": 0.46649402379989624,
+      "learning_rate": 6.180781758957654e-05,
+      "loss": 0.5059,
+      "step": 1112
+    },
+    {
+      "epoch": 2.444810543657331,
+      "grad_norm": 0.6002528071403503,
+      "learning_rate": 6.156351791530944e-05,
+      "loss": 0.9118,
+      "step": 1113
+    },
+    {
+      "epoch": 2.4470071389346515,
+      "grad_norm": 0.5003305673599243,
+      "learning_rate": 6.131921824104234e-05,
+      "loss": 0.9265,
+      "step": 1114
+    },
+    {
+      "epoch": 2.4492037342119715,
+      "grad_norm": 0.422230988740921,
+      "learning_rate": 6.107491856677525e-05,
+      "loss": 0.679,
+      "step": 1115
+    },
+    {
+      "epoch": 2.4514003294892914,
+      "grad_norm": 1.058239221572876,
+      "learning_rate": 6.083061889250814e-05,
+      "loss": 0.6453,
+      "step": 1116
+    },
+    {
+      "epoch": 2.453596924766612,
+      "grad_norm": 0.556039571762085,
+      "learning_rate": 6.058631921824104e-05,
+      "loss": 0.6956,
+      "step": 1117
+    },
+    {
+      "epoch": 2.455793520043932,
+      "grad_norm": 0.43044599890708923,
+      "learning_rate": 6.0342019543973935e-05,
+      "loss": 0.7238,
+      "step": 1118
+    },
+    {
+      "epoch": 2.4579901153212522,
+      "grad_norm": 0.457256555557251,
+      "learning_rate": 6.009771986970684e-05,
+      "loss": 0.7706,
+      "step": 1119
+    },
+    {
+      "epoch": 2.460186710598572,
+      "grad_norm": 0.4917759895324707,
+      "learning_rate": 5.9853420195439734e-05,
+      "loss": 0.6947,
+      "step": 1120
+    },
+    {
+      "epoch": 2.4623833058758926,
+      "grad_norm": 1.000158429145813,
+      "learning_rate": 5.9609120521172636e-05,
+      "loss": 0.7743,
+      "step": 1121
+    },
+    {
+      "epoch": 2.4645799011532126,
+      "grad_norm": 1.7269330024719238,
+      "learning_rate": 5.936482084690553e-05,
+      "loss": 1.1369,
+      "step": 1122
+    },
+    {
+      "epoch": 2.4667764964305325,
+      "grad_norm": 0.6047888398170471,
+      "learning_rate": 5.9120521172638435e-05,
+      "loss": 0.7416,
+      "step": 1123
+    },
+    {
+      "epoch": 2.468973091707853,
+      "grad_norm": 0.5752591490745544,
+      "learning_rate": 5.887622149837133e-05,
+      "loss": 0.7787,
+      "step": 1124
+    },
+    {
+      "epoch": 2.471169686985173,
+      "grad_norm": 0.46982571482658386,
+      "learning_rate": 5.863192182410423e-05,
+      "loss": 0.6303,
+      "step": 1125
+    },
+    {
+      "epoch": 2.473366282262493,
+      "grad_norm": 0.5997644066810608,
+      "learning_rate": 5.838762214983713e-05,
+      "loss": 0.855,
+      "step": 1126
+    },
+    {
+      "epoch": 2.4755628775398133,
+      "grad_norm": 0.5510478615760803,
+      "learning_rate": 5.8143322475570026e-05,
+      "loss": 0.6854,
+      "step": 1127
+    },
+    {
+      "epoch": 2.4777594728171333,
+      "grad_norm": 0.404849648475647,
+      "learning_rate": 5.789902280130293e-05,
+      "loss": 0.7209,
+      "step": 1128
+    },
+    {
+      "epoch": 2.4799560680944537,
+      "grad_norm": 0.4667147696018219,
+      "learning_rate": 5.7654723127035825e-05,
+      "loss": 0.8619,
+      "step": 1129
+    },
+    {
+      "epoch": 2.4821526633717736,
+      "grad_norm": 0.7835572957992554,
+      "learning_rate": 5.741042345276873e-05,
+      "loss": 0.7333,
+      "step": 1130
+    },
+    {
+      "epoch": 2.484349258649094,
+      "grad_norm": 0.4661206901073456,
+      "learning_rate": 5.716612377850162e-05,
+      "loss": 0.7103,
+      "step": 1131
+    },
+    {
+      "epoch": 2.486545853926414,
+      "grad_norm": 0.48164454102516174,
+      "learning_rate": 5.6921824104234526e-05,
+      "loss": 0.7366,
+      "step": 1132
+    },
+    {
+      "epoch": 2.488742449203734,
+      "grad_norm": 0.46449264883995056,
+      "learning_rate": 5.667752442996742e-05,
+      "loss": 0.6455,
+      "step": 1133
+    },
+    {
+      "epoch": 2.4909390444810544,
+      "grad_norm": 0.4726642668247223,
+      "learning_rate": 5.643322475570032e-05,
+      "loss": 0.7568,
+      "step": 1134
+    },
+    {
+      "epoch": 2.4931356397583744,
+      "grad_norm": 3.123643159866333,
+      "learning_rate": 5.618892508143322e-05,
+      "loss": 0.7537,
+      "step": 1135
+    },
+    {
+      "epoch": 2.495332235035695,
+      "grad_norm": 0.5379687547683716,
+      "learning_rate": 5.594462540716612e-05,
+      "loss": 0.6447,
+      "step": 1136
+    },
+    {
+      "epoch": 2.4975288303130148,
+      "grad_norm": 0.4394984543323517,
+      "learning_rate": 5.570032573289902e-05,
+      "loss": 0.5101,
+      "step": 1137
+    },
+    {
+      "epoch": 2.499725425590335,
+      "grad_norm": 0.5947285890579224,
+      "learning_rate": 5.5456026058631915e-05,
+      "loss": 0.7502,
+      "step": 1138
+    },
+    {
+      "epoch": 2.501922020867655,
+      "grad_norm": 2.268911838531494,
+      "learning_rate": 5.521172638436482e-05,
+      "loss": 0.8953,
+      "step": 1139
+    },
+    {
+      "epoch": 2.504118616144975,
+      "grad_norm": 1.0242410898208618,
+      "learning_rate": 5.4967426710097714e-05,
+      "loss": 0.6583,
+      "step": 1140
+    },
+    {
+      "epoch": 2.5063152114222955,
+      "grad_norm": 0.3080623149871826,
+      "learning_rate": 5.472312703583061e-05,
+      "loss": 0.6191,
+      "step": 1141
+    },
+    {
+      "epoch": 2.5085118066996155,
+      "grad_norm": 0.5389127135276794,
+      "learning_rate": 5.447882736156351e-05,
+      "loss": 0.7974,
+      "step": 1142
+    },
+    {
+      "epoch": 2.510708401976936,
+      "grad_norm": 0.5144618153572083,
+      "learning_rate": 5.423452768729641e-05,
+      "loss": 0.8575,
+      "step": 1143
+    },
+    {
+      "epoch": 2.512904997254256,
+      "grad_norm": 0.43258488178253174,
+      "learning_rate": 5.399022801302931e-05,
+      "loss": 0.7017,
+      "step": 1144
+    },
+    {
+      "epoch": 2.5151015925315763,
+      "grad_norm": 0.588394820690155,
+      "learning_rate": 5.374592833876221e-05,
+      "loss": 0.6938,
+      "step": 1145
+    },
+    {
+      "epoch": 2.5172981878088962,
+      "grad_norm": 0.5307460427284241,
+      "learning_rate": 5.350162866449511e-05,
+      "loss": 0.7091,
+      "step": 1146
+    },
+    {
+      "epoch": 2.519494783086216,
+      "grad_norm": 0.5652251243591309,
+      "learning_rate": 5.3257328990228006e-05,
+      "loss": 0.6801,
+      "step": 1147
+    },
+    {
+      "epoch": 2.5216913783635366,
+      "grad_norm": 0.526828408241272,
+      "learning_rate": 5.301302931596091e-05,
+      "loss": 0.6523,
+      "step": 1148
+    },
+    {
+      "epoch": 2.5238879736408566,
+      "grad_norm": 1.197115421295166,
+      "learning_rate": 5.2768729641693805e-05,
+      "loss": 0.8034,
+      "step": 1149
+    },
+    {
+      "epoch": 2.5260845689181766,
+      "grad_norm": 0.4833904206752777,
+      "learning_rate": 5.25244299674267e-05,
+      "loss": 0.9559,
+      "step": 1150
+    },
+    {
+      "epoch": 2.528281164195497,
+      "grad_norm": 0.7229256629943848,
+      "learning_rate": 5.2280130293159604e-05,
+      "loss": 0.78,
+      "step": 1151
+    },
+    {
+      "epoch": 2.5304777594728174,
+      "grad_norm": 0.4525471031665802,
+      "learning_rate": 5.20358306188925e-05,
+      "loss": 0.7016,
+      "step": 1152
+    },
+    {
+      "epoch": 2.5326743547501374,
+      "grad_norm": 0.5236949324607849,
+      "learning_rate": 5.17915309446254e-05,
+      "loss": 0.7273,
+      "step": 1153
+    },
+    {
+      "epoch": 2.5348709500274573,
+      "grad_norm": 0.47247302532196045,
+      "learning_rate": 5.15472312703583e-05,
+      "loss": 0.7046,
+      "step": 1154
+    },
+    {
+      "epoch": 2.5370675453047777,
+      "grad_norm": 0.37208980321884155,
+      "learning_rate": 5.13029315960912e-05,
+      "loss": 0.6981,
+      "step": 1155
+    },
+    {
+      "epoch": 2.5392641405820977,
+      "grad_norm": 0.6562976241111755,
+      "learning_rate": 5.10586319218241e-05,
+      "loss": 0.879,
+      "step": 1156
+    },
+    {
+      "epoch": 2.5414607358594177,
+      "grad_norm": 0.5763184428215027,
+      "learning_rate": 5.081433224755699e-05,
+      "loss": 0.7986,
+      "step": 1157
+    },
+    {
+      "epoch": 2.543657331136738,
+      "grad_norm": 0.47017866373062134,
+      "learning_rate": 5.0570032573289896e-05,
+      "loss": 0.6521,
+      "step": 1158
+    },
+    {
+      "epoch": 2.545853926414058,
+      "grad_norm": 0.4337298274040222,
+      "learning_rate": 5.032573289902279e-05,
+      "loss": 0.6295,
+      "step": 1159
+    },
+    {
+      "epoch": 2.5480505216913785,
+      "grad_norm": 0.46856433153152466,
+      "learning_rate": 5.0081433224755695e-05,
+      "loss": 0.7138,
+      "step": 1160
+    },
+    {
+      "epoch": 2.5502471169686984,
+      "grad_norm": 0.5523170232772827,
+      "learning_rate": 4.983713355048859e-05,
+      "loss": 0.5777,
+      "step": 1161
+    },
+    {
+      "epoch": 2.552443712246019,
+      "grad_norm": 3.289175510406494,
+      "learning_rate": 4.9592833876221493e-05,
+      "loss": 0.7541,
+      "step": 1162
+    },
+    {
+      "epoch": 2.554640307523339,
+      "grad_norm": 0.6942198276519775,
+      "learning_rate": 4.934853420195439e-05,
+      "loss": 0.6252,
+      "step": 1163
+    },
+    {
+      "epoch": 2.556836902800659,
+      "grad_norm": 0.4597557783126831,
+      "learning_rate": 4.910423452768729e-05,
+      "loss": 0.7861,
+      "step": 1164
+    },
+    {
+      "epoch": 2.559033498077979,
+      "grad_norm": 0.4139542281627655,
+      "learning_rate": 4.885993485342019e-05,
+      "loss": 0.7832,
+      "step": 1165
+    },
+    {
+      "epoch": 2.561230093355299,
+      "grad_norm": 0.531743586063385,
+      "learning_rate": 4.8615635179153084e-05,
+      "loss": 0.6837,
+      "step": 1166
+    },
+    {
+      "epoch": 2.5634266886326196,
+      "grad_norm": 0.4594990015029907,
+      "learning_rate": 4.837133550488599e-05,
+      "loss": 0.6277,
+      "step": 1167
+    },
+    {
+      "epoch": 2.5656232839099395,
+      "grad_norm": 0.47610026597976685,
+      "learning_rate": 4.812703583061888e-05,
+      "loss": 0.7605,
+      "step": 1168
+    },
+    {
+      "epoch": 2.56781987918726,
+      "grad_norm": 0.4345327317714691,
+      "learning_rate": 4.7882736156351786e-05,
+      "loss": 0.6874,
+      "step": 1169
+    },
+    {
+      "epoch": 2.57001647446458,
+      "grad_norm": 0.5750065445899963,
+      "learning_rate": 4.763843648208468e-05,
+      "loss": 0.6963,
+      "step": 1170
+    },
+    {
+      "epoch": 2.5722130697419,
+      "grad_norm": 0.4556567072868347,
+      "learning_rate": 4.7394136807817584e-05,
+      "loss": 0.6409,
+      "step": 1171
+    },
+    {
+      "epoch": 2.5744096650192203,
+      "grad_norm": 0.8773305416107178,
+      "learning_rate": 4.714983713355048e-05,
+      "loss": 0.6379,
+      "step": 1172
+    },
+    {
+      "epoch": 2.5766062602965403,
+      "grad_norm": 0.4118146598339081,
+      "learning_rate": 4.6905537459283376e-05,
+      "loss": 0.6564,
+      "step": 1173
+    },
+    {
+      "epoch": 2.5788028555738602,
+      "grad_norm": 0.5570434331893921,
+      "learning_rate": 4.6661237785016286e-05,
+      "loss": 0.7924,
+      "step": 1174
+    },
+    {
+      "epoch": 2.5809994508511807,
+      "grad_norm": 0.4674397110939026,
+      "learning_rate": 4.641693811074919e-05,
+      "loss": 0.7492,
+      "step": 1175
+    },
+    {
+      "epoch": 2.583196046128501,
+      "grad_norm": 0.5306969285011292,
+      "learning_rate": 4.6172638436482085e-05,
+      "loss": 0.6719,
+      "step": 1176
+    },
+    {
+      "epoch": 2.585392641405821,
+      "grad_norm": 0.5056284070014954,
+      "learning_rate": 4.592833876221498e-05,
+      "loss": 0.8847,
+      "step": 1177
+    },
+    {
+      "epoch": 2.587589236683141,
+      "grad_norm": 0.665483832359314,
+      "learning_rate": 4.568403908794788e-05,
+      "loss": 0.6194,
+      "step": 1178
+    },
+    {
+      "epoch": 2.5897858319604614,
+      "grad_norm": 0.6785182952880859,
+      "learning_rate": 4.543973941368078e-05,
+      "loss": 0.7992,
+      "step": 1179
+    },
+    {
+      "epoch": 2.5919824272377814,
+      "grad_norm": 0.37807726860046387,
+      "learning_rate": 4.519543973941368e-05,
+      "loss": 0.6303,
+      "step": 1180
+    },
+    {
+      "epoch": 2.5941790225151014,
+      "grad_norm": 0.6085902452468872,
+      "learning_rate": 4.495114006514658e-05,
+      "loss": 0.8179,
+      "step": 1181
+    },
+    {
+      "epoch": 2.5963756177924218,
+      "grad_norm": 0.7031683325767517,
+      "learning_rate": 4.470684039087948e-05,
+      "loss": 0.7601,
+      "step": 1182
+    },
+    {
+      "epoch": 2.598572213069742,
+      "grad_norm": 0.3876109719276428,
+      "learning_rate": 4.446254071661238e-05,
+      "loss": 0.7194,
+      "step": 1183
+    },
+    {
+      "epoch": 2.600768808347062,
+      "grad_norm": 0.5286533832550049,
+      "learning_rate": 4.421824104234527e-05,
+      "loss": 0.7409,
+      "step": 1184
+    },
+    {
+      "epoch": 2.602965403624382,
+      "grad_norm": 0.4775243103504181,
+      "learning_rate": 4.3973941368078175e-05,
+      "loss": 0.775,
+      "step": 1185
+    },
+    {
+      "epoch": 2.6051619989017025,
+      "grad_norm": 0.44586068391799927,
+      "learning_rate": 4.372964169381107e-05,
+      "loss": 0.7704,
+      "step": 1186
+    },
+    {
+      "epoch": 2.6073585941790225,
+      "grad_norm": 0.9450865983963013,
+      "learning_rate": 4.3485342019543974e-05,
+      "loss": 0.6381,
+      "step": 1187
+    },
+    {
+      "epoch": 2.6095551894563425,
+      "grad_norm": 0.4812171459197998,
+      "learning_rate": 4.324104234527687e-05,
+      "loss": 0.7171,
+      "step": 1188
+    },
+    {
+      "epoch": 2.611751784733663,
+      "grad_norm": 0.6057891249656677,
+      "learning_rate": 4.299674267100977e-05,
+      "loss": 0.6574,
+      "step": 1189
+    },
+    {
+      "epoch": 2.613948380010983,
+      "grad_norm": 1.0423542261123657,
+      "learning_rate": 4.275244299674267e-05,
+      "loss": 0.7518,
+      "step": 1190
+    },
+    {
+      "epoch": 2.6161449752883033,
+      "grad_norm": 0.5942821502685547,
+      "learning_rate": 4.250814332247557e-05,
+      "loss": 0.7053,
+      "step": 1191
+    },
+    {
+      "epoch": 2.618341570565623,
+      "grad_norm": 0.5034943222999573,
+      "learning_rate": 4.226384364820847e-05,
+      "loss": 0.5508,
+      "step": 1192
+    },
+    {
+      "epoch": 2.6205381658429436,
+      "grad_norm": 0.9159315228462219,
+      "learning_rate": 4.2019543973941364e-05,
+      "loss": 1.1516,
+      "step": 1193
+    },
+    {
+      "epoch": 2.6227347611202636,
+      "grad_norm": 0.6301320195198059,
+      "learning_rate": 4.1775244299674266e-05,
+      "loss": 0.8349,
+      "step": 1194
+    },
+    {
+      "epoch": 2.6249313563975836,
+      "grad_norm": 0.36837509274482727,
+      "learning_rate": 4.153094462540716e-05,
+      "loss": 0.4788,
+      "step": 1195
+    },
+    {
+      "epoch": 2.627127951674904,
+      "grad_norm": 0.4848581552505493,
+      "learning_rate": 4.1286644951140065e-05,
+      "loss": 0.6891,
+      "step": 1196
+    },
+    {
+      "epoch": 2.629324546952224,
+      "grad_norm": 0.8092116117477417,
+      "learning_rate": 4.104234527687296e-05,
+      "loss": 0.6606,
+      "step": 1197
+    },
+    {
+      "epoch": 2.6315211422295444,
+      "grad_norm": 0.46400120854377747,
+      "learning_rate": 4.0798045602605864e-05,
+      "loss": 0.7563,
+      "step": 1198
+    },
+    {
+      "epoch": 2.6337177375068643,
+      "grad_norm": 0.47451305389404297,
+      "learning_rate": 4.055374592833876e-05,
+      "loss": 0.5976,
+      "step": 1199
+    },
+    {
+      "epoch": 2.6359143327841847,
+      "grad_norm": 0.49082696437835693,
+      "learning_rate": 4.0309446254071656e-05,
+      "loss": 0.7163,
+      "step": 1200
+    },
+    {
+      "epoch": 2.6381109280615047,
+      "grad_norm": 0.5413679480552673,
+      "learning_rate": 4.006514657980456e-05,
+      "loss": 0.6193,
+      "step": 1201
+    },
+    {
+      "epoch": 2.6403075233388247,
+      "grad_norm": 0.4185996651649475,
+      "learning_rate": 3.9820846905537454e-05,
+      "loss": 0.5389,
+      "step": 1202
+    },
+    {
+      "epoch": 2.642504118616145,
+      "grad_norm": 0.5521745681762695,
+      "learning_rate": 3.957654723127036e-05,
+      "loss": 0.914,
+      "step": 1203
+    },
+    {
+      "epoch": 2.644700713893465,
+      "grad_norm": 0.5430476665496826,
+      "learning_rate": 3.933224755700325e-05,
+      "loss": 0.6883,
+      "step": 1204
+    },
+    {
+      "epoch": 2.646897309170785,
+      "grad_norm": 0.4742693603038788,
+      "learning_rate": 3.9087947882736156e-05,
+      "loss": 0.7757,
+      "step": 1205
+    },
+    {
+      "epoch": 2.6490939044481054,
+      "grad_norm": 0.5665991306304932,
+      "learning_rate": 3.884364820846905e-05,
+      "loss": 0.844,
+      "step": 1206
+    },
+    {
+      "epoch": 2.651290499725426,
+      "grad_norm": 0.4997795820236206,
+      "learning_rate": 3.8599348534201955e-05,
+      "loss": 0.8338,
+      "step": 1207
+    },
+    {
+      "epoch": 2.653487095002746,
+      "grad_norm": 0.6155491471290588,
+      "learning_rate": 3.835504885993485e-05,
+      "loss": 0.7317,
+      "step": 1208
+    },
+    {
+      "epoch": 2.655683690280066,
+      "grad_norm": 0.5159393548965454,
+      "learning_rate": 3.8110749185667747e-05,
+      "loss": 0.5967,
+      "step": 1209
+    },
+    {
+      "epoch": 2.657880285557386,
+      "grad_norm": 0.5114976763725281,
+      "learning_rate": 3.786644951140065e-05,
+      "loss": 0.9115,
+      "step": 1210
+    },
+    {
+      "epoch": 2.660076880834706,
+      "grad_norm": 0.6197386980056763,
+      "learning_rate": 3.7622149837133545e-05,
+      "loss": 0.6937,
+      "step": 1211
+    },
+    {
+      "epoch": 2.662273476112026,
+      "grad_norm": 0.49142980575561523,
+      "learning_rate": 3.737785016286645e-05,
+      "loss": 0.6526,
+      "step": 1212
+    },
+    {
+      "epoch": 2.6644700713893466,
+      "grad_norm": 0.7175025939941406,
+      "learning_rate": 3.7133550488599344e-05,
+      "loss": 0.783,
+      "step": 1213
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.6020824909210205,
+      "learning_rate": 3.688925081433225e-05,
+      "loss": 0.6524,
+      "step": 1214
+    },
+    {
+      "epoch": 2.668863261943987,
+      "grad_norm": 1.0529444217681885,
+      "learning_rate": 3.664495114006514e-05,
+      "loss": 0.7009,
+      "step": 1215
+    },
+    {
+      "epoch": 2.671059857221307,
+      "grad_norm": 0.5170251727104187,
+      "learning_rate": 3.640065146579804e-05,
+      "loss": 0.7693,
+      "step": 1216
+    },
+    {
+      "epoch": 2.6732564524986273,
+      "grad_norm": 0.6865411996841431,
+      "learning_rate": 3.615635179153094e-05,
+      "loss": 0.772,
+      "step": 1217
+    },
+    {
+      "epoch": 2.6754530477759473,
+      "grad_norm": 0.39495792984962463,
+      "learning_rate": 3.591205211726384e-05,
+      "loss": 0.6929,
+      "step": 1218
+    },
+    {
+      "epoch": 2.6776496430532672,
+      "grad_norm": 1.0190775394439697,
+      "learning_rate": 3.566775244299674e-05,
+      "loss": 0.6486,
+      "step": 1219
+    },
+    {
+      "epoch": 2.6798462383305877,
+      "grad_norm": 0.5837611556053162,
+      "learning_rate": 3.5423452768729636e-05,
+      "loss": 0.6781,
+      "step": 1220
+    },
+    {
+      "epoch": 2.6820428336079076,
+      "grad_norm": 0.6491107940673828,
+      "learning_rate": 3.517915309446254e-05,
+      "loss": 0.8908,
+      "step": 1221
+    },
+    {
+      "epoch": 2.684239428885228,
+      "grad_norm": 0.7074655890464783,
+      "learning_rate": 3.4934853420195435e-05,
+      "loss": 0.5527,
+      "step": 1222
+    },
+    {
+      "epoch": 2.686436024162548,
+      "grad_norm": 0.4448198974132538,
+      "learning_rate": 3.469055374592834e-05,
+      "loss": 0.7951,
+      "step": 1223
+    },
+    {
+      "epoch": 2.6886326194398684,
+      "grad_norm": 0.8564078211784363,
+      "learning_rate": 3.4446254071661234e-05,
+      "loss": 0.7456,
+      "step": 1224
+    },
+    {
+      "epoch": 2.6908292147171884,
+      "grad_norm": 0.5537450909614563,
+      "learning_rate": 3.420195439739413e-05,
+      "loss": 0.8213,
+      "step": 1225
+    },
+    {
+      "epoch": 2.6930258099945084,
+      "grad_norm": 0.534600555896759,
+      "learning_rate": 3.395765472312703e-05,
+      "loss": 0.62,
+      "step": 1226
+    },
+    {
+      "epoch": 2.6952224052718288,
+      "grad_norm": 0.44154056906700134,
+      "learning_rate": 3.371335504885993e-05,
+      "loss": 0.7852,
+      "step": 1227
+    },
+    {
+      "epoch": 2.6974190005491487,
+      "grad_norm": 0.4525333046913147,
+      "learning_rate": 3.346905537459283e-05,
+      "loss": 0.6761,
+      "step": 1228
+    },
+    {
+      "epoch": 2.6996155958264687,
+      "grad_norm": 0.4910513758659363,
+      "learning_rate": 3.322475570032573e-05,
+      "loss": 0.7986,
+      "step": 1229
+    },
+    {
+      "epoch": 2.701812191103789,
+      "grad_norm": 0.4798869788646698,
+      "learning_rate": 3.298045602605863e-05,
+      "loss": 0.6122,
+      "step": 1230
+    },
+    {
+      "epoch": 2.7040087863811095,
+      "grad_norm": 0.358011394739151,
+      "learning_rate": 3.273615635179153e-05,
+      "loss": 0.6304,
+      "step": 1231
+    },
+    {
+      "epoch": 2.7062053816584295,
+      "grad_norm": 0.5586791038513184,
+      "learning_rate": 3.249185667752443e-05,
+      "loss": 0.8322,
+      "step": 1232
+    },
+    {
+      "epoch": 2.7084019769357495,
+      "grad_norm": 0.5098724961280823,
+      "learning_rate": 3.224755700325733e-05,
+      "loss": 0.6717,
+      "step": 1233
+    },
+    {
+      "epoch": 2.71059857221307,
+      "grad_norm": 0.5464269518852234,
+      "learning_rate": 3.200325732899023e-05,
+      "loss": 0.7552,
+      "step": 1234
+    },
+    {
+      "epoch": 2.71279516749039,
+      "grad_norm": 0.6027079820632935,
+      "learning_rate": 3.175895765472312e-05,
+      "loss": 0.8241,
+      "step": 1235
+    },
+    {
+      "epoch": 2.71499176276771,
+      "grad_norm": 0.5655648708343506,
+      "learning_rate": 3.1514657980456026e-05,
+      "loss": 0.9164,
+      "step": 1236
+    },
+    {
+      "epoch": 2.7171883580450302,
+      "grad_norm": 0.5104654431343079,
+      "learning_rate": 3.127035830618892e-05,
+      "loss": 0.7392,
+      "step": 1237
+    },
+    {
+      "epoch": 2.71938495332235,
+      "grad_norm": 0.8253754377365112,
+      "learning_rate": 3.1026058631921825e-05,
+      "loss": 0.7458,
+      "step": 1238
+    },
+    {
+      "epoch": 2.7215815485996706,
+      "grad_norm": 0.629065215587616,
+      "learning_rate": 3.078175895765472e-05,
+      "loss": 0.7403,
+      "step": 1239
+    },
+    {
+      "epoch": 2.7237781438769906,
+      "grad_norm": 0.44151559472084045,
+      "learning_rate": 3.0537459283387624e-05,
+      "loss": 0.7365,
+      "step": 1240
+    },
+    {
+      "epoch": 2.725974739154311,
+      "grad_norm": 0.43887874484062195,
+      "learning_rate": 3.029315960912052e-05,
+      "loss": 0.5476,
+      "step": 1241
+    },
+    {
+      "epoch": 2.728171334431631,
+      "grad_norm": 0.5551141500473022,
+      "learning_rate": 3.004885993485342e-05,
+      "loss": 0.6571,
+      "step": 1242
+    },
+    {
+      "epoch": 2.730367929708951,
+      "grad_norm": 0.4901895225048065,
+      "learning_rate": 2.9804560260586318e-05,
+      "loss": 0.8075,
+      "step": 1243
+    },
+    {
+      "epoch": 2.7325645249862713,
+      "grad_norm": 0.6756422519683838,
+      "learning_rate": 2.9560260586319218e-05,
+      "loss": 0.801,
+      "step": 1244
+    },
+    {
+      "epoch": 2.7347611202635913,
+      "grad_norm": 0.6377543807029724,
+      "learning_rate": 2.9315960912052114e-05,
+      "loss": 0.7223,
+      "step": 1245
+    },
+    {
+      "epoch": 2.7369577155409117,
+      "grad_norm": 1.1810672283172607,
+      "learning_rate": 2.9071661237785013e-05,
+      "loss": 0.7188,
+      "step": 1246
+    },
+    {
+      "epoch": 2.7391543108182317,
+      "grad_norm": 0.6899003982543945,
+      "learning_rate": 2.8827361563517912e-05,
+      "loss": 1.233,
+      "step": 1247
+    },
+    {
+      "epoch": 2.741350906095552,
+      "grad_norm": 0.4581334590911865,
+      "learning_rate": 2.858306188925081e-05,
+      "loss": 0.7326,
+      "step": 1248
+    },
+    {
+      "epoch": 2.743547501372872,
+      "grad_norm": 0.41814476251602173,
+      "learning_rate": 2.833876221498371e-05,
+      "loss": 0.8313,
+      "step": 1249
+    },
+    {
+      "epoch": 2.745744096650192,
+      "grad_norm": 0.4597308933734894,
+      "learning_rate": 2.809446254071661e-05,
+      "loss": 0.7606,
+      "step": 1250
+    },
+    {
+      "epoch": 2.7479406919275124,
+      "grad_norm": 0.849672257900238,
+      "learning_rate": 2.785016286644951e-05,
+      "loss": 0.8635,
+      "step": 1251
+    },
+    {
+      "epoch": 2.7501372872048324,
+      "grad_norm": 0.5546273589134216,
+      "learning_rate": 2.760586319218241e-05,
+      "loss": 0.7811,
+      "step": 1252
+    },
+    {
+      "epoch": 2.752333882482153,
+      "grad_norm": 0.640661358833313,
+      "learning_rate": 2.7361563517915305e-05,
+      "loss": 0.6045,
+      "step": 1253
+    },
+    {
+      "epoch": 2.754530477759473,
+      "grad_norm": 0.4509889781475067,
+      "learning_rate": 2.7117263843648204e-05,
+      "loss": 0.6529,
+      "step": 1254
+    },
+    {
+      "epoch": 2.756727073036793,
+      "grad_norm": 0.41180872917175293,
+      "learning_rate": 2.6872964169381104e-05,
+      "loss": 0.8657,
+      "step": 1255
+    },
+    {
+      "epoch": 2.758923668314113,
+      "grad_norm": 0.48240530490875244,
+      "learning_rate": 2.6628664495114003e-05,
+      "loss": 0.55,
+      "step": 1256
+    },
+    {
+      "epoch": 2.761120263591433,
+      "grad_norm": 0.6201155185699463,
+      "learning_rate": 2.6384364820846903e-05,
+      "loss": 0.7233,
+      "step": 1257
+    },
+    {
+      "epoch": 2.7633168588687536,
+      "grad_norm": 0.44931790232658386,
+      "learning_rate": 2.6140065146579802e-05,
+      "loss": 0.5943,
+      "step": 1258
+    },
+    {
+      "epoch": 2.7655134541460735,
+      "grad_norm": 0.6736621260643005,
+      "learning_rate": 2.58957654723127e-05,
+      "loss": 0.8165,
+      "step": 1259
+    },
+    {
+      "epoch": 2.7677100494233935,
+      "grad_norm": 0.505808413028717,
+      "learning_rate": 2.56514657980456e-05,
+      "loss": 0.7903,
+      "step": 1260
+    },
+    {
+      "epoch": 2.769906644700714,
+      "grad_norm": 0.512080729007721,
+      "learning_rate": 2.5407166123778497e-05,
+      "loss": 0.7157,
+      "step": 1261
+    },
+    {
+      "epoch": 2.7721032399780343,
+      "grad_norm": 0.514725923538208,
+      "learning_rate": 2.5162866449511396e-05,
+      "loss": 0.7336,
+      "step": 1262
+    },
+    {
+      "epoch": 2.7742998352553543,
+      "grad_norm": 0.637206494808197,
+      "learning_rate": 2.4918566775244295e-05,
+      "loss": 0.7951,
+      "step": 1263
+    },
+    {
+      "epoch": 2.7764964305326743,
+      "grad_norm": 0.43634292483329773,
+      "learning_rate": 2.4674267100977195e-05,
+      "loss": 0.8981,
+      "step": 1264
+    },
+    {
+      "epoch": 2.7786930258099947,
+      "grad_norm": 0.4418467581272125,
+      "learning_rate": 2.4429967426710094e-05,
+      "loss": 0.8288,
+      "step": 1265
+    },
+    {
+      "epoch": 2.7808896210873146,
+      "grad_norm": 0.3626846373081207,
+      "learning_rate": 2.4185667752442993e-05,
+      "loss": 0.6052,
+      "step": 1266
+    },
+    {
+      "epoch": 2.7830862163646346,
+      "grad_norm": 0.4869728982448578,
+      "learning_rate": 2.3941368078175893e-05,
+      "loss": 0.6506,
+      "step": 1267
+    },
+    {
+      "epoch": 2.785282811641955,
+      "grad_norm": 0.8676609992980957,
+      "learning_rate": 2.3697068403908792e-05,
+      "loss": 0.7181,
+      "step": 1268
+    },
+    {
+      "epoch": 2.787479406919275,
+      "grad_norm": 0.5998664498329163,
+      "learning_rate": 2.3452768729641688e-05,
+      "loss": 0.6367,
+      "step": 1269
+    },
+    {
+      "epoch": 2.7896760021965954,
+      "grad_norm": 0.5200008153915405,
+      "learning_rate": 2.3208469055374594e-05,
+      "loss": 0.6459,
+      "step": 1270
+    },
+    {
+      "epoch": 2.7918725974739154,
+      "grad_norm": 0.4447275400161743,
+      "learning_rate": 2.296416938110749e-05,
+      "loss": 0.6273,
+      "step": 1271
+    },
+    {
+      "epoch": 2.7940691927512358,
+      "grad_norm": 0.4883853793144226,
+      "learning_rate": 2.271986970684039e-05,
+      "loss": 0.7158,
+      "step": 1272
+    },
+    {
+      "epoch": 2.7962657880285557,
+      "grad_norm": 0.3995269238948822,
+      "learning_rate": 2.247557003257329e-05,
+      "loss": 0.7349,
+      "step": 1273
+    },
+    {
+      "epoch": 2.7984623833058757,
+      "grad_norm": 0.4498269557952881,
+      "learning_rate": 2.223127035830619e-05,
+      "loss": 0.9126,
+      "step": 1274
+    },
+    {
+      "epoch": 2.800658978583196,
+      "grad_norm": 0.5987016558647156,
+      "learning_rate": 2.1986970684039088e-05,
+      "loss": 0.7106,
+      "step": 1275
+    },
+    {
+      "epoch": 2.802855573860516,
+      "grad_norm": 0.6620035171508789,
+      "learning_rate": 2.1742671009771987e-05,
+      "loss": 0.8606,
+      "step": 1276
+    },
+    {
+      "epoch": 2.8050521691378365,
+      "grad_norm": 2.1700897216796875,
+      "learning_rate": 2.1498371335504886e-05,
+      "loss": 0.9735,
+      "step": 1277
+    },
+    {
+      "epoch": 2.8072487644151565,
+      "grad_norm": 0.5042754411697388,
+      "learning_rate": 2.1254071661237786e-05,
+      "loss": 0.7672,
+      "step": 1278
+    },
+    {
+      "epoch": 2.809445359692477,
+      "grad_norm": 0.46072670817375183,
+      "learning_rate": 2.1009771986970682e-05,
+      "loss": 0.6677,
+      "step": 1279
+    },
+    {
+      "epoch": 2.811641954969797,
+      "grad_norm": 0.4872454106807709,
+      "learning_rate": 2.076547231270358e-05,
+      "loss": 0.6817,
+      "step": 1280
+    },
+    {
+      "epoch": 2.813838550247117,
+      "grad_norm": 0.6175678968429565,
+      "learning_rate": 2.052117263843648e-05,
+      "loss": 0.8266,
+      "step": 1281
+    },
+    {
+      "epoch": 2.8160351455244372,
+      "grad_norm": 0.4629005193710327,
+      "learning_rate": 2.027687296416938e-05,
+      "loss": 0.7101,
+      "step": 1282
+    },
+    {
+      "epoch": 2.818231740801757,
+      "grad_norm": 0.5369590520858765,
+      "learning_rate": 2.003257328990228e-05,
+      "loss": 0.8781,
+      "step": 1283
+    },
+    {
+      "epoch": 2.820428336079077,
+      "grad_norm": 0.4451259672641754,
+      "learning_rate": 1.978827361563518e-05,
+      "loss": 0.8365,
+      "step": 1284
+    },
+    {
+      "epoch": 2.8226249313563976,
+      "grad_norm": 0.5523382425308228,
+      "learning_rate": 1.9543973941368078e-05,
+      "loss": 0.8101,
+      "step": 1285
+    },
+    {
+      "epoch": 2.824821526633718,
+      "grad_norm": 0.7317632436752319,
+      "learning_rate": 1.9299674267100977e-05,
+      "loss": 0.987,
+      "step": 1286
+    },
+    {
+      "epoch": 2.827018121911038,
+      "grad_norm": 0.6328700184822083,
+      "learning_rate": 1.9055374592833873e-05,
+      "loss": 0.6458,
+      "step": 1287
+    },
+    {
+      "epoch": 2.829214717188358,
+      "grad_norm": 0.5861051678657532,
+      "learning_rate": 1.8811074918566773e-05,
+      "loss": 0.7532,
+      "step": 1288
+    },
+    {
+      "epoch": 2.8314113124656783,
+      "grad_norm": 0.6192939281463623,
+      "learning_rate": 1.8566775244299672e-05,
+      "loss": 0.6951,
+      "step": 1289
+    },
+    {
+      "epoch": 2.8336079077429983,
+      "grad_norm": 0.57658451795578,
+      "learning_rate": 1.832247557003257e-05,
+      "loss": 0.7974,
+      "step": 1290
+    },
+    {
+      "epoch": 2.8358045030203183,
+      "grad_norm": 0.5019885301589966,
+      "learning_rate": 1.807817589576547e-05,
+      "loss": 0.6196,
+      "step": 1291
+    },
+    {
+      "epoch": 2.8380010982976387,
+      "grad_norm": 0.5013266205787659,
+      "learning_rate": 1.783387622149837e-05,
+      "loss": 0.7016,
+      "step": 1292
+    },
+    {
+      "epoch": 2.8401976935749587,
+      "grad_norm": 0.4292304217815399,
+      "learning_rate": 1.758957654723127e-05,
+      "loss": 0.5401,
+      "step": 1293
+    },
+    {
+      "epoch": 2.842394288852279,
+      "grad_norm": 0.6416001319885254,
+      "learning_rate": 1.734527687296417e-05,
+      "loss": 0.8176,
+      "step": 1294
+    },
+    {
+      "epoch": 2.844590884129599,
+      "grad_norm": 0.6593545079231262,
+      "learning_rate": 1.7100977198697065e-05,
+      "loss": 0.7976,
+      "step": 1295
+    },
+    {
+      "epoch": 2.8467874794069195,
+      "grad_norm": 0.47218093276023865,
+      "learning_rate": 1.6856677524429964e-05,
+      "loss": 0.8412,
+      "step": 1296
+    },
+    {
+      "epoch": 2.8489840746842394,
+      "grad_norm": 0.4412826895713806,
+      "learning_rate": 1.6612377850162864e-05,
+      "loss": 0.6351,
+      "step": 1297
+    },
+    {
+      "epoch": 2.8511806699615594,
+      "grad_norm": 0.5436484217643738,
+      "learning_rate": 1.6368078175895766e-05,
+      "loss": 0.9519,
+      "step": 1298
+    },
+    {
+      "epoch": 2.85337726523888,
+      "grad_norm": 0.8250836133956909,
+      "learning_rate": 1.6123778501628666e-05,
+      "loss": 0.7833,
+      "step": 1299
+    },
+    {
+      "epoch": 2.8555738605161998,
+      "grad_norm": 0.6315727829933167,
+      "learning_rate": 1.587947882736156e-05,
+      "loss": 0.9153,
+      "step": 1300
+    },
+    {
+      "epoch": 2.85777045579352,
+      "grad_norm": 0.5417777895927429,
+      "learning_rate": 1.563517915309446e-05,
+      "loss": 0.7535,
+      "step": 1301
+    },
+    {
+      "epoch": 2.85996705107084,
+      "grad_norm": 0.7258586287498474,
+      "learning_rate": 1.539087947882736e-05,
+      "loss": 0.64,
+      "step": 1302
+    },
+    {
+      "epoch": 2.8621636463481606,
+      "grad_norm": 0.47286394238471985,
+      "learning_rate": 1.514657980456026e-05,
+      "loss": 0.9278,
+      "step": 1303
+    },
+    {
+      "epoch": 2.8643602416254805,
+      "grad_norm": 0.4970191419124603,
+      "learning_rate": 1.4902280130293159e-05,
+      "loss": 0.6084,
+      "step": 1304
+    },
+    {
+      "epoch": 2.8665568369028005,
+      "grad_norm": 0.44317418336868286,
+      "learning_rate": 1.4657980456026057e-05,
+      "loss": 0.6955,
+      "step": 1305
+    },
+    {
+      "epoch": 2.868753432180121,
+      "grad_norm": 0.4458111822605133,
+      "learning_rate": 1.4413680781758956e-05,
+      "loss": 0.7311,
+      "step": 1306
+    },
+    {
+      "epoch": 2.870950027457441,
+      "grad_norm": 0.5451056361198425,
+      "learning_rate": 1.4169381107491856e-05,
+      "loss": 0.7979,
+      "step": 1307
+    },
+    {
+      "epoch": 2.873146622734761,
+      "grad_norm": 0.4472848176956177,
+      "learning_rate": 1.3925081433224755e-05,
+      "loss": 0.6204,
+      "step": 1308
+    },
+    {
+      "epoch": 2.8753432180120813,
+      "grad_norm": 0.6725212931632996,
+      "learning_rate": 1.3680781758957653e-05,
+      "loss": 0.7265,
+      "step": 1309
+    },
+    {
+      "epoch": 2.8775398132894017,
+      "grad_norm": 0.708122730255127,
+      "learning_rate": 1.3436482084690552e-05,
+      "loss": 0.8112,
+      "step": 1310
+    },
+    {
+      "epoch": 2.8797364085667216,
+      "grad_norm": 0.5034201741218567,
+      "learning_rate": 1.3192182410423451e-05,
+      "loss": 0.7278,
+      "step": 1311
+    },
+    {
+      "epoch": 2.8819330038440416,
+      "grad_norm": 0.6512319445610046,
+      "learning_rate": 1.294788273615635e-05,
+      "loss": 0.7767,
+      "step": 1312
+    },
+    {
+      "epoch": 2.884129599121362,
+      "grad_norm": 0.48872819542884827,
+      "learning_rate": 1.2703583061889248e-05,
+      "loss": 0.7275,
+      "step": 1313
+    },
+    {
+      "epoch": 2.886326194398682,
+      "grad_norm": 0.5452210903167725,
+      "learning_rate": 1.2459283387622148e-05,
+      "loss": 0.7383,
+      "step": 1314
+    },
+    {
+      "epoch": 2.888522789676002,
+      "grad_norm": 1.1614961624145508,
+      "learning_rate": 1.2214983713355047e-05,
+      "loss": 0.6579,
+      "step": 1315
+    },
+    {
+      "epoch": 2.8907193849533224,
+      "grad_norm": 0.6417173743247986,
+      "learning_rate": 1.1970684039087946e-05,
+      "loss": 0.7266,
+      "step": 1316
+    },
+    {
+      "epoch": 2.892915980230643,
+      "grad_norm": 0.5201627016067505,
+      "learning_rate": 1.1726384364820844e-05,
+      "loss": 0.6218,
+      "step": 1317
+    },
+    {
+      "epoch": 2.8951125755079627,
+      "grad_norm": 0.49721887707710266,
+      "learning_rate": 1.1482084690553745e-05,
+      "loss": 0.5777,
+      "step": 1318
+    },
+    {
+      "epoch": 2.8973091707852827,
+      "grad_norm": 0.4739466607570648,
+      "learning_rate": 1.1237785016286644e-05,
+      "loss": 0.5898,
+      "step": 1319
+    },
+    {
+      "epoch": 2.899505766062603,
+      "grad_norm": 0.6560915112495422,
+      "learning_rate": 1.0993485342019544e-05,
+      "loss": 0.8049,
+      "step": 1320
+    },
+    {
+      "epoch": 2.901702361339923,
+      "grad_norm": 0.544268786907196,
+      "learning_rate": 1.0749185667752443e-05,
+      "loss": 0.5575,
+      "step": 1321
+    },
+    {
+      "epoch": 2.903898956617243,
+      "grad_norm": 0.7952961921691895,
+      "learning_rate": 1.0504885993485341e-05,
+      "loss": 0.743,
+      "step": 1322
+    },
+    {
+      "epoch": 2.9060955518945635,
+      "grad_norm": 0.5153457522392273,
+      "learning_rate": 1.026058631921824e-05,
+      "loss": 0.6234,
+      "step": 1323
+    },
+    {
+      "epoch": 2.9082921471718834,
+      "grad_norm": 0.5607213377952576,
+      "learning_rate": 1.001628664495114e-05,
+      "loss": 0.7031,
+      "step": 1324
+    },
+    {
+      "epoch": 2.910488742449204,
+      "grad_norm": 0.5462546348571777,
+      "learning_rate": 9.771986970684039e-06,
+      "loss": 0.6374,
+      "step": 1325
+    },
+    {
+      "epoch": 2.912685337726524,
+      "grad_norm": 0.5088673233985901,
+      "learning_rate": 9.527687296416937e-06,
+      "loss": 0.7418,
+      "step": 1326
+    },
+    {
+      "epoch": 2.9148819330038442,
+      "grad_norm": 0.4698350727558136,
+      "learning_rate": 9.283387622149836e-06,
+      "loss": 0.7746,
+      "step": 1327
+    },
+    {
+      "epoch": 2.917078528281164,
+      "grad_norm": 0.5693159103393555,
+      "learning_rate": 9.039087947882735e-06,
+      "loss": 0.5745,
+      "step": 1328
+    },
+    {
+      "epoch": 2.919275123558484,
+      "grad_norm": 0.9007793068885803,
+      "learning_rate": 8.794788273615635e-06,
+      "loss": 0.5333,
+      "step": 1329
+    },
+    {
+      "epoch": 2.9214717188358046,
+      "grad_norm": 0.5462734699249268,
+      "learning_rate": 8.550488599348532e-06,
+      "loss": 0.6539,
+      "step": 1330
+    },
+    {
+      "epoch": 2.9236683141131246,
+      "grad_norm": 0.673041045665741,
+      "learning_rate": 8.306188925081432e-06,
+      "loss": 0.8152,
+      "step": 1331
+    },
+    {
+      "epoch": 2.925864909390445,
+      "grad_norm": 0.5064035058021545,
+      "learning_rate": 8.061889250814333e-06,
+      "loss": 0.404,
+      "step": 1332
+    },
+    {
+      "epoch": 2.928061504667765,
+      "grad_norm": 0.6657114624977112,
+      "learning_rate": 7.81758957654723e-06,
+      "loss": 0.6704,
+      "step": 1333
+    },
+    {
+      "epoch": 2.9302580999450853,
+      "grad_norm": 0.4478769898414612,
+      "learning_rate": 7.57328990228013e-06,
+      "loss": 0.6366,
+      "step": 1334
+    },
+    {
+      "epoch": 2.9324546952224053,
+      "grad_norm": 0.6483010053634644,
+      "learning_rate": 7.328990228013028e-06,
+      "loss": 0.6189,
+      "step": 1335
+    },
+    {
+      "epoch": 2.9346512904997253,
+      "grad_norm": 0.6626474261283875,
+      "learning_rate": 7.084690553745928e-06,
+      "loss": 0.6292,
+      "step": 1336
+    },
+    {
+      "epoch": 2.9368478857770457,
+      "grad_norm": 0.46029290556907654,
+      "learning_rate": 6.840390879478826e-06,
+      "loss": 0.613,
+      "step": 1337
+    },
+    {
+      "epoch": 2.9390444810543657,
+      "grad_norm": 0.5039319396018982,
+      "learning_rate": 6.596091205211726e-06,
+      "loss": 0.658,
+      "step": 1338
+    },
+    {
+      "epoch": 2.9412410763316856,
+      "grad_norm": 0.4084259271621704,
+      "learning_rate": 6.351791530944624e-06,
+      "loss": 0.659,
+      "step": 1339
+    },
+    {
+      "epoch": 2.943437671609006,
+      "grad_norm": 0.4596741199493408,
+      "learning_rate": 6.1074918566775235e-06,
+      "loss": 0.5019,
+      "step": 1340
+    },
+    {
+      "epoch": 2.9456342668863265,
+      "grad_norm": 0.4353286027908325,
+      "learning_rate": 5.863192182410422e-06,
+      "loss": 0.7048,
+      "step": 1341
+    },
+    {
+      "epoch": 2.9478308621636464,
+      "grad_norm": 0.4436233937740326,
+      "learning_rate": 5.618892508143322e-06,
+      "loss": 0.7606,
+      "step": 1342
+    },
+    {
+      "epoch": 2.9500274574409664,
+      "grad_norm": 0.5930688977241516,
+      "learning_rate": 5.374592833876222e-06,
+      "loss": 0.8574,
+      "step": 1343
+    },
+    {
+      "epoch": 2.952224052718287,
+      "grad_norm": 0.49891534447669983,
+      "learning_rate": 5.13029315960912e-06,
+      "loss": 0.8351,
+      "step": 1344
+    },
+    {
+      "epoch": 2.9544206479956068,
+      "grad_norm": 0.48590177297592163,
+      "learning_rate": 4.8859934853420195e-06,
+      "loss": 0.7691,
+      "step": 1345
+    },
+    {
+      "epoch": 2.9566172432729267,
+      "grad_norm": 0.5856354236602783,
+      "learning_rate": 4.641693811074918e-06,
+      "loss": 0.6358,
+      "step": 1346
+    },
+    {
+      "epoch": 2.958813838550247,
+      "grad_norm": 0.4865894317626953,
+      "learning_rate": 4.397394136807817e-06,
+      "loss": 0.6245,
+      "step": 1347
+    },
+    {
+      "epoch": 2.961010433827567,
+      "grad_norm": 0.5668537020683289,
+      "learning_rate": 4.153094462540716e-06,
+      "loss": 0.6052,
+      "step": 1348
+    },
+    {
+      "epoch": 2.9632070291048875,
+      "grad_norm": 0.5946842432022095,
+      "learning_rate": 3.908794788273615e-06,
+      "loss": 0.8955,
+      "step": 1349
+    },
+    {
+      "epoch": 2.9654036243822075,
+      "grad_norm": 0.568600594997406,
+      "learning_rate": 3.664495114006514e-06,
+      "loss": 0.6851,
+      "step": 1350
+    },
+    {
+      "epoch": 2.967600219659528,
+      "grad_norm": 0.4441044330596924,
+      "learning_rate": 3.420195439739413e-06,
+      "loss": 0.8337,
+      "step": 1351
+    },
+    {
+      "epoch": 2.969796814936848,
+      "grad_norm": 0.7190451622009277,
+      "learning_rate": 3.175895765472312e-06,
+      "loss": 0.6609,
+      "step": 1352
+    },
+    {
+      "epoch": 2.971993410214168,
+      "grad_norm": 0.4812300503253937,
+      "learning_rate": 2.931596091205211e-06,
+      "loss": 0.7508,
+      "step": 1353
+    },
+    {
+      "epoch": 2.9741900054914883,
+      "grad_norm": 0.4684293866157532,
+      "learning_rate": 2.687296416938111e-06,
+      "loss": 0.8381,
+      "step": 1354
+    },
+    {
+      "epoch": 2.9763866007688082,
+      "grad_norm": 0.5247290134429932,
+      "learning_rate": 2.4429967426710097e-06,
+      "loss": 0.6112,
+      "step": 1355
+    },
+    {
+      "epoch": 2.9785831960461286,
+      "grad_norm": 0.9329695701599121,
+      "learning_rate": 2.1986970684039087e-06,
+      "loss": 0.7751,
+      "step": 1356
+    },
+    {
+      "epoch": 2.9807797913234486,
+      "grad_norm": 0.7722187042236328,
+      "learning_rate": 1.9543973941368076e-06,
+      "loss": 0.7769,
+      "step": 1357
+    },
+    {
+      "epoch": 2.982976386600769,
+      "grad_norm": 0.56857830286026,
+      "learning_rate": 1.7100977198697066e-06,
+      "loss": 0.7727,
+      "step": 1358
+    },
+    {
+      "epoch": 2.985172981878089,
+      "grad_norm": 0.5151993036270142,
+      "learning_rate": 1.4657980456026055e-06,
+      "loss": 0.8697,
+      "step": 1359
+    },
+    {
+      "epoch": 2.987369577155409,
+      "grad_norm": 0.4162243604660034,
+      "learning_rate": 1.2214983713355049e-06,
+      "loss": 0.6017,
+      "step": 1360
+    },
+    {
+      "epoch": 2.9895661724327294,
+      "grad_norm": 0.4609653949737549,
+      "learning_rate": 9.771986970684038e-07,
+      "loss": 0.7389,
+      "step": 1361
+    },
+    {
+      "epoch": 2.9917627677100493,
+      "grad_norm": 0.60605388879776,
+      "learning_rate": 7.328990228013028e-07,
+      "loss": 0.8284,
+      "step": 1362
+    },
+    {
+      "epoch": 2.9939593629873693,
+      "grad_norm": 0.5641438364982605,
+      "learning_rate": 4.885993485342019e-07,
+      "loss": 0.7095,
+      "step": 1363
+    },
+    {
+      "epoch": 2.9961559582646897,
+      "grad_norm": 0.47373566031455994,
+      "learning_rate": 2.4429967426710095e-07,
+      "loss": 0.7898,
+      "step": 1364
+    },
+    {
+      "epoch": 2.99835255354201,
+      "grad_norm": 0.5009136199951172,
+      "learning_rate": 0.0,
+      "loss": 0.824,
+      "step": 1365
+    },
+    {
+      "epoch": 2.99835255354201,
+      "step": 1365,
+      "total_flos": 4.603349598658888e+17,
+      "train_loss": 0.7779110742575956,
+      "train_runtime": 2984.3873,
+      "train_samples_per_second": 7.321,
+      "train_steps_per_second": 0.457
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1365,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.603349598658888e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}