diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5776 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1638,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003663003663003663,
+      "grad_norm": 1.5506155490875244,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.681140899658203,
+      "step": 2
+    },
+    {
+      "epoch": 0.007326007326007326,
+      "grad_norm": 0.4526568055152893,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.6674047708511353,
+      "step": 4
+    },
+    {
+      "epoch": 0.01098901098901099,
+      "grad_norm": 0.8889099359512329,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.8801467418670654,
+      "step": 6
+    },
+    {
+      "epoch": 0.014652014652014652,
+      "grad_norm": 0.19507154822349548,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.0659124851226807,
+      "step": 8
+    },
+    {
+      "epoch": 0.018315018315018316,
+      "grad_norm": 0.3095138967037201,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.2201435565948486,
+      "step": 10
+    },
+    {
+      "epoch": 0.02197802197802198,
+      "grad_norm": 0.7828930020332336,
+      "learning_rate": 4.4e-06,
+      "loss": 2.0232832431793213,
+      "step": 12
+    },
+    {
+      "epoch": 0.02564102564102564,
+      "grad_norm": 0.21790610253810883,
+      "learning_rate": 5.2e-06,
+      "loss": 1.7574424743652344,
+      "step": 14
+    },
+    {
+      "epoch": 0.029304029304029304,
+      "grad_norm": 0.25006550550460815,
+      "learning_rate": 6e-06,
+      "loss": 1.8244725465774536,
+      "step": 16
+    },
+    {
+      "epoch": 0.03296703296703297,
+      "grad_norm": 1.2375390529632568,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.7521573305130005,
+      "step": 18
+    },
+    {
+      "epoch": 0.03663003663003663,
+      "grad_norm": 0.22365400195121765,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.7442874908447266,
+      "step": 20
+    },
+    {
+      "epoch": 0.040293040293040296,
+      "grad_norm": 1.432734489440918,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.137043833732605,
+      "step": 22
+    },
+    {
+      "epoch": 0.04395604395604396,
+      "grad_norm": 0.23710986971855164,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.3199552297592163,
+      "step": 24
+    },
+    {
+      "epoch": 0.047619047619047616,
+      "grad_norm": 0.6795807480812073,
+      "learning_rate": 1e-05,
+      "loss": 1.4519306421279907,
+      "step": 26
+    },
+    {
+      "epoch": 0.05128205128205128,
+      "grad_norm": 1.1287591457366943,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 1.4781732559204102,
+      "step": 28
+    },
+    {
+      "epoch": 0.054945054945054944,
+      "grad_norm": 0.40312108397483826,
+      "learning_rate": 1.16e-05,
+      "loss": 1.1731195449829102,
+      "step": 30
+    },
+    {
+      "epoch": 0.05860805860805861,
+      "grad_norm": 0.3888772428035736,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 1.10336172580719,
+      "step": 32
+    },
+    {
+      "epoch": 0.06227106227106227,
+      "grad_norm": 0.057199470698833466,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 1.2399317026138306,
+      "step": 34
+    },
+    {
+      "epoch": 0.06593406593406594,
+      "grad_norm": 0.24481792747974396,
+      "learning_rate": 1.4e-05,
+      "loss": 1.61336088180542,
+      "step": 36
+    },
+    {
+      "epoch": 0.0695970695970696,
+      "grad_norm": 0.37236636877059937,
+      "learning_rate": 1.48e-05,
+      "loss": 1.372746229171753,
+      "step": 38
+    },
+    {
+      "epoch": 0.07326007326007326,
+      "grad_norm": 0.5762131810188293,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 1.4253513813018799,
+      "step": 40
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 0.17068041861057281,
+      "learning_rate": 1.64e-05,
+      "loss": 1.136276364326477,
+      "step": 42
+    },
+    {
+      "epoch": 0.08058608058608059,
+      "grad_norm": 0.19196651875972748,
+      "learning_rate": 1.72e-05,
+      "loss": 1.5315269231796265,
+      "step": 44
+    },
+    {
+      "epoch": 0.08424908424908426,
+      "grad_norm": 0.12281376868486404,
+      "learning_rate": 1.8e-05,
+      "loss": 1.6208034753799438,
+      "step": 46
+    },
+    {
+      "epoch": 0.08791208791208792,
+      "grad_norm": 0.40122199058532715,
+      "learning_rate": 1.88e-05,
+      "loss": 1.05559241771698,
+      "step": 48
+    },
+    {
+      "epoch": 0.09157509157509157,
+      "grad_norm": 0.687041163444519,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.5109608173370361,
+      "step": 50
+    },
+    {
+      "epoch": 0.09523809523809523,
+      "grad_norm": 0.15380625426769257,
+      "learning_rate": 1.999998238790087e-05,
+      "loss": 0.7463083863258362,
+      "step": 52
+    },
+    {
+      "epoch": 0.0989010989010989,
+      "grad_norm": 0.37560248374938965,
+      "learning_rate": 1.999984149152137e-05,
+      "loss": 0.9765978455543518,
+      "step": 54
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "grad_norm": 0.15966370701789856,
+      "learning_rate": 1.999955970096814e-05,
+      "loss": 1.3462445735931396,
+      "step": 56
+    },
+    {
+      "epoch": 0.10622710622710622,
+      "grad_norm": 0.11719954758882523,
+      "learning_rate": 1.9999137020652663e-05,
+      "loss": 1.197383999824524,
+      "step": 58
+    },
+    {
+      "epoch": 0.10989010989010989,
+      "grad_norm": 0.3115270137786865,
+      "learning_rate": 1.999857345719207e-05,
+      "loss": 1.4108028411865234,
+      "step": 60
+    },
+    {
+      "epoch": 0.11355311355311355,
+      "grad_norm": 0.31386664509773254,
+      "learning_rate": 1.9997869019409047e-05,
+      "loss": 1.4261771440505981,
+      "step": 62
+    },
+    {
+      "epoch": 0.11721611721611722,
+      "grad_norm": 0.25177937746047974,
+      "learning_rate": 1.9997023718331707e-05,
+      "loss": 1.3881282806396484,
+      "step": 64
+    },
+    {
+      "epoch": 0.12087912087912088,
+      "grad_norm": 0.14471372961997986,
+      "learning_rate": 1.9996037567193388e-05,
+      "loss": 1.3539735078811646,
+      "step": 66
+    },
+    {
+      "epoch": 0.12454212454212454,
+      "grad_norm": 0.14020830392837524,
+      "learning_rate": 1.9994910581432466e-05,
+      "loss": 1.3212106227874756,
+      "step": 68
+    },
+    {
+      "epoch": 0.1282051282051282,
+      "grad_norm": 0.6085403561592102,
+      "learning_rate": 1.9993642778692116e-05,
+      "loss": 1.0624397993087769,
+      "step": 70
+    },
+    {
+      "epoch": 0.13186813186813187,
+      "grad_norm": 0.11608725786209106,
+      "learning_rate": 1.999223417882002e-05,
+      "loss": 1.4300200939178467,
+      "step": 72
+    },
+    {
+      "epoch": 0.13553113553113552,
+      "grad_norm": 0.35819724202156067,
+      "learning_rate": 1.9990684803868068e-05,
+      "loss": 1.5944573879241943,
+      "step": 74
+    },
+    {
+      "epoch": 0.1391941391941392,
+      "grad_norm": 0.09426046907901764,
+      "learning_rate": 1.9988994678092007e-05,
+      "loss": 1.0820951461791992,
+      "step": 76
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 0.1737513393163681,
+      "learning_rate": 1.9987163827951077e-05,
+      "loss": 1.4329181909561157,
+      "step": 78
+    },
+    {
+      "epoch": 0.14652014652014653,
+      "grad_norm": 0.25684013962745667,
+      "learning_rate": 1.998519228210756e-05,
+      "loss": 1.5802102088928223,
+      "step": 80
+    },
+    {
+      "epoch": 0.15018315018315018,
+      "grad_norm": 0.6553322672843933,
+      "learning_rate": 1.998308007142638e-05,
+      "loss": 1.1970324516296387,
+      "step": 82
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.16790863871574402,
+      "learning_rate": 1.9980827228974575e-05,
+      "loss": 1.3608276844024658,
+      "step": 84
+    },
+    {
+      "epoch": 0.1575091575091575,
+      "grad_norm": 0.3864733874797821,
+      "learning_rate": 1.997843379002081e-05,
+      "loss": 1.4945706129074097,
+      "step": 86
+    },
+    {
+      "epoch": 0.16117216117216118,
+      "grad_norm": 0.12279342859983444,
+      "learning_rate": 1.9975899792034824e-05,
+      "loss": 0.700541615486145,
+      "step": 88
+    },
+    {
+      "epoch": 0.16483516483516483,
+      "grad_norm": 0.06569632887840271,
+      "learning_rate": 1.9973225274686804e-05,
+      "loss": 0.8554237484931946,
+      "step": 90
+    },
+    {
+      "epoch": 0.1684981684981685,
+      "grad_norm": 0.14802643656730652,
+      "learning_rate": 1.9970410279846816e-05,
+      "loss": 1.31403648853302,
+      "step": 92
+    },
+    {
+      "epoch": 0.17216117216117216,
+      "grad_norm": 0.13368292152881622,
+      "learning_rate": 1.9967454851584132e-05,
+      "loss": 1.3383275270462036,
+      "step": 94
+    },
+    {
+      "epoch": 0.17582417582417584,
+      "grad_norm": 0.22364209592342377,
+      "learning_rate": 1.996435903616651e-05,
+      "loss": 1.2345792055130005,
+      "step": 96
+    },
+    {
+      "epoch": 0.1794871794871795,
+      "grad_norm": 0.3644048571586609,
+      "learning_rate": 1.9961122882059523e-05,
+      "loss": 1.3272985219955444,
+      "step": 98
+    },
+    {
+      "epoch": 0.18315018315018314,
+      "grad_norm": 0.38108178973197937,
+      "learning_rate": 1.9957746439925748e-05,
+      "loss": 1.1760129928588867,
+      "step": 100
+    },
+    {
+      "epoch": 0.18681318681318682,
+      "grad_norm": 0.37011611461639404,
+      "learning_rate": 1.9954229762624016e-05,
+      "loss": 1.2455718517303467,
+      "step": 102
+    },
+    {
+      "epoch": 0.19047619047619047,
+      "grad_norm": 0.21642433106899261,
+      "learning_rate": 1.995057290520855e-05,
+      "loss": 0.897014319896698,
+      "step": 104
+    },
+    {
+      "epoch": 0.19413919413919414,
+      "grad_norm": 0.3669384717941284,
+      "learning_rate": 1.9946775924928132e-05,
+      "loss": 1.0838041305541992,
+      "step": 106
+    },
+    {
+      "epoch": 0.1978021978021978,
+      "grad_norm": 0.38939031958580017,
+      "learning_rate": 1.9942838881225183e-05,
+      "loss": 1.3144299983978271,
+      "step": 108
+    },
+    {
+      "epoch": 0.20146520146520147,
+      "grad_norm": 0.6144610643386841,
+      "learning_rate": 1.9938761835734842e-05,
+      "loss": 1.1635433435440063,
+      "step": 110
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 0.1933593600988388,
+      "learning_rate": 1.9934544852284013e-05,
+      "loss": 1.2244360446929932,
+      "step": 112
+    },
+    {
+      "epoch": 0.2087912087912088,
+      "grad_norm": 0.30885422229766846,
+      "learning_rate": 1.9930187996890347e-05,
+      "loss": 0.6363462209701538,
+      "step": 114
+    },
+    {
+      "epoch": 0.21245421245421245,
+      "grad_norm": 0.1966598927974701,
+      "learning_rate": 1.992569133776121e-05,
+      "loss": 1.300977349281311,
+      "step": 116
+    },
+    {
+      "epoch": 0.21611721611721613,
+      "grad_norm": 0.8058923482894897,
+      "learning_rate": 1.992105494529264e-05,
+      "loss": 1.2940763235092163,
+      "step": 118
+    },
+    {
+      "epoch": 0.21978021978021978,
+      "grad_norm": 0.2799862325191498,
+      "learning_rate": 1.99162788920682e-05,
+      "loss": 1.3638042211532593,
+      "step": 120
+    },
+    {
+      "epoch": 0.22344322344322345,
+      "grad_norm": 9.588260650634766,
+      "learning_rate": 1.9911363252857887e-05,
+      "loss": 1.2911320924758911,
+      "step": 122
+    },
+    {
+      "epoch": 0.2271062271062271,
+      "grad_norm": 0.3367455303668976,
+      "learning_rate": 1.990630810461694e-05,
+      "loss": 1.022411823272705,
+      "step": 124
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.18774043023586273,
+      "learning_rate": 1.990111352648463e-05,
+      "loss": 0.8959170579910278,
+      "step": 126
+    },
+    {
+      "epoch": 0.23443223443223443,
+      "grad_norm": 0.19097784161567688,
+      "learning_rate": 1.9895779599783033e-05,
+      "loss": 1.1906862258911133,
+      "step": 128
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 0.26235634088516235,
+      "learning_rate": 1.989030640801576e-05,
+      "loss": 1.2996340990066528,
+      "step": 130
+    },
+    {
+      "epoch": 0.24175824175824176,
+      "grad_norm": 0.22878225147724152,
+      "learning_rate": 1.9884694036866624e-05,
+      "loss": 1.391095757484436,
+      "step": 132
+    },
+    {
+      "epoch": 0.2454212454212454,
+      "grad_norm": 0.6518133878707886,
+      "learning_rate": 1.9878942574198334e-05,
+      "loss": 1.289358377456665,
+      "step": 134
+    },
+    {
+      "epoch": 0.2490842490842491,
+      "grad_norm": 0.16628128290176392,
+      "learning_rate": 1.9873052110051094e-05,
+      "loss": 1.273111343383789,
+      "step": 136
+    },
+    {
+      "epoch": 0.25274725274725274,
+      "grad_norm": 0.9112105369567871,
+      "learning_rate": 1.9867022736641205e-05,
+      "loss": 1.089441180229187,
+      "step": 138
+    },
+    {
+      "epoch": 0.2564102564102564,
+      "grad_norm": 0.1825258433818817,
+      "learning_rate": 1.9860854548359615e-05,
+      "loss": 1.2736470699310303,
+      "step": 140
+    },
+    {
+      "epoch": 0.2600732600732601,
+      "grad_norm": 0.40484118461608887,
+      "learning_rate": 1.9854547641770446e-05,
+      "loss": 1.2917908430099487,
+      "step": 142
+    },
+    {
+      "epoch": 0.26373626373626374,
+      "grad_norm": 0.1851384937763214,
+      "learning_rate": 1.9848102115609483e-05,
+      "loss": 1.2552529573440552,
+      "step": 144
+    },
+    {
+      "epoch": 0.2673992673992674,
+      "grad_norm": 0.3763132095336914,
+      "learning_rate": 1.9841518070782615e-05,
+      "loss": 1.4075181484222412,
+      "step": 146
+    },
+    {
+      "epoch": 0.27106227106227104,
+      "grad_norm": 0.19049830734729767,
+      "learning_rate": 1.983479561036429e-05,
+      "loss": 1.3139081001281738,
+      "step": 148
+    },
+    {
+      "epoch": 0.27472527472527475,
+      "grad_norm": 0.06231250986456871,
+      "learning_rate": 1.982793483959585e-05,
+      "loss": 0.9256948828697205,
+      "step": 150
+    },
+    {
+      "epoch": 0.2783882783882784,
+      "grad_norm": 0.09106171131134033,
+      "learning_rate": 1.9820935865883924e-05,
+      "loss": 0.635033369064331,
+      "step": 152
+    },
+    {
+      "epoch": 0.28205128205128205,
+      "grad_norm": 0.18586048483848572,
+      "learning_rate": 1.981379879879874e-05,
+      "loss": 1.1024783849716187,
+      "step": 154
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 0.18290774524211884,
+      "learning_rate": 1.9806523750072385e-05,
+      "loss": 1.3396060466766357,
+      "step": 156
+    },
+    {
+      "epoch": 0.2893772893772894,
+      "grad_norm": 0.2552820146083832,
+      "learning_rate": 1.9799110833597093e-05,
+      "loss": 1.2755292654037476,
+      "step": 158
+    },
+    {
+      "epoch": 0.29304029304029305,
+      "grad_norm": 0.21683865785598755,
+      "learning_rate": 1.9791560165423433e-05,
+      "loss": 0.9428300857543945,
+      "step": 160
+    },
+    {
+      "epoch": 0.2967032967032967,
+      "grad_norm": 0.16783960163593292,
+      "learning_rate": 1.9783871863758503e-05,
+      "loss": 1.5332800149917603,
+      "step": 162
+    },
+    {
+      "epoch": 0.30036630036630035,
+      "grad_norm": 0.4921853244304657,
+      "learning_rate": 1.9776046048964082e-05,
+      "loss": 1.0453159809112549,
+      "step": 164
+    },
+    {
+      "epoch": 0.304029304029304,
+      "grad_norm": 0.17284567654132843,
+      "learning_rate": 1.9768082843554737e-05,
+      "loss": 1.389228105545044,
+      "step": 166
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.16623765230178833,
+      "learning_rate": 1.9759982372195918e-05,
+      "loss": 1.129130244255066,
+      "step": 168
+    },
+    {
+      "epoch": 0.31135531135531136,
+      "grad_norm": 0.3981800079345703,
+      "learning_rate": 1.9751744761701984e-05,
+      "loss": 1.2560561895370483,
+      "step": 170
+    },
+    {
+      "epoch": 0.315018315018315,
+      "grad_norm": 0.17309850454330444,
+      "learning_rate": 1.9743370141034248e-05,
+      "loss": 1.0026013851165771,
+      "step": 172
+    },
+    {
+      "epoch": 0.31868131868131866,
+      "grad_norm": 0.12312160432338715,
+      "learning_rate": 1.973485864129894e-05,
+      "loss": 0.8527880311012268,
+      "step": 174
+    },
+    {
+      "epoch": 0.32234432234432236,
+      "grad_norm": 0.7391336560249329,
+      "learning_rate": 1.9726210395745148e-05,
+      "loss": 1.3922659158706665,
+      "step": 176
+    },
+    {
+      "epoch": 0.326007326007326,
+      "grad_norm": 0.16699042916297913,
+      "learning_rate": 1.971742553976275e-05,
+      "loss": 0.921902596950531,
+      "step": 178
+    },
+    {
+      "epoch": 0.32967032967032966,
+      "grad_norm": 0.22556649148464203,
+      "learning_rate": 1.9708504210880284e-05,
+      "loss": 1.4865270853042603,
+      "step": 180
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.2530522346496582,
+      "learning_rate": 1.969944654876279e-05,
+      "loss": 0.973817765712738,
+      "step": 182
+    },
+    {
+      "epoch": 0.336996336996337,
+      "grad_norm": 0.2421599179506302,
+      "learning_rate": 1.9690252695209636e-05,
+      "loss": 1.2389814853668213,
+      "step": 184
+    },
+    {
+      "epoch": 0.34065934065934067,
+      "grad_norm": 0.33265063166618347,
+      "learning_rate": 1.9680922794152294e-05,
+      "loss": 1.3093231916427612,
+      "step": 186
+    },
+    {
+      "epoch": 0.3443223443223443,
+      "grad_norm": 0.14455479383468628,
+      "learning_rate": 1.9671456991652072e-05,
+      "loss": 1.1712009906768799,
+      "step": 188
+    },
+    {
+      "epoch": 0.34798534798534797,
+      "grad_norm": 0.16971716284751892,
+      "learning_rate": 1.9661855435897858e-05,
+      "loss": 1.2694875001907349,
+      "step": 190
+    },
+    {
+      "epoch": 0.3516483516483517,
+      "grad_norm": 0.42642462253570557,
+      "learning_rate": 1.9652118277203767e-05,
+      "loss": 1.0975149869918823,
+      "step": 192
+    },
+    {
+      "epoch": 0.3553113553113553,
+      "grad_norm": 0.14698947966098785,
+      "learning_rate": 1.9642245668006814e-05,
+      "loss": 1.2505216598510742,
+      "step": 194
+    },
+    {
+      "epoch": 0.358974358974359,
+      "grad_norm": 0.16152219474315643,
+      "learning_rate": 1.963223776286451e-05,
+      "loss": 1.252524495124817,
+      "step": 196
+    },
+    {
+      "epoch": 0.3626373626373626,
+      "grad_norm": 0.11607574671506882,
+      "learning_rate": 1.9622094718452448e-05,
+      "loss": 0.8584736585617065,
+      "step": 198
+    },
+    {
+      "epoch": 0.3663003663003663,
+      "grad_norm": 0.24364766478538513,
+      "learning_rate": 1.9611816693561858e-05,
+      "loss": 1.0111479759216309,
+      "step": 200
+    },
+    {
+      "epoch": 0.36996336996337,
+      "grad_norm": 0.9354498386383057,
+      "learning_rate": 1.96014038490971e-05,
+      "loss": 1.4314929246902466,
+      "step": 202
+    },
+    {
+      "epoch": 0.37362637362637363,
+      "grad_norm": 0.20628754794597626,
+      "learning_rate": 1.9590856348073182e-05,
+      "loss": 1.205211877822876,
+      "step": 204
+    },
+    {
+      "epoch": 0.3772893772893773,
+      "grad_norm": 0.8566434979438782,
+      "learning_rate": 1.9580174355613168e-05,
+      "loss": 0.7619340419769287,
+      "step": 206
+    },
+    {
+      "epoch": 0.38095238095238093,
+      "grad_norm": 0.222511425614357,
+      "learning_rate": 1.9569358038945617e-05,
+      "loss": 1.1318811178207397,
+      "step": 208
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.20434831082820892,
+      "learning_rate": 1.9558407567401945e-05,
+      "loss": 1.407196283340454,
+      "step": 210
+    },
+    {
+      "epoch": 0.3882783882783883,
+      "grad_norm": 0.14155927300453186,
+      "learning_rate": 1.9547323112413806e-05,
+      "loss": 1.0677672624588013,
+      "step": 212
+    },
+    {
+      "epoch": 0.39194139194139194,
+      "grad_norm": 0.20749327540397644,
+      "learning_rate": 1.9536104847510384e-05,
+      "loss": 1.1320164203643799,
+      "step": 214
+    },
+    {
+      "epoch": 0.3956043956043956,
+      "grad_norm": 0.24210180342197418,
+      "learning_rate": 1.9524752948315677e-05,
+      "loss": 1.2214279174804688,
+      "step": 216
+    },
+    {
+      "epoch": 0.3992673992673993,
+      "grad_norm": 0.20847296714782715,
+      "learning_rate": 1.9513267592545752e-05,
+      "loss": 1.2568475008010864,
+      "step": 218
+    },
+    {
+      "epoch": 0.40293040293040294,
+      "grad_norm": 0.14721128344535828,
+      "learning_rate": 1.9501648960005964e-05,
+      "loss": 0.6154600381851196,
+      "step": 220
+    },
+    {
+      "epoch": 0.4065934065934066,
+      "grad_norm": 0.6793058514595032,
+      "learning_rate": 1.948989723258815e-05,
+      "loss": 1.3387889862060547,
+      "step": 222
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 0.32100027799606323,
+      "learning_rate": 1.9478012594267757e-05,
+      "loss": 1.0697791576385498,
+      "step": 224
+    },
+    {
+      "epoch": 0.4139194139194139,
+      "grad_norm": 0.594149649143219,
+      "learning_rate": 1.946599523110099e-05,
+      "loss": 1.2391932010650635,
+      "step": 226
+    },
+    {
+      "epoch": 0.4175824175824176,
+      "grad_norm": 0.21164406836032867,
+      "learning_rate": 1.945384533122187e-05,
+      "loss": 1.3020524978637695,
+      "step": 228
+    },
+    {
+      "epoch": 0.42124542124542125,
+      "grad_norm": 0.4633813500404358,
+      "learning_rate": 1.9441563084839324e-05,
+      "loss": 1.2441799640655518,
+      "step": 230
+    },
+    {
+      "epoch": 0.4249084249084249,
+      "grad_norm": 0.3936166763305664,
+      "learning_rate": 1.942914868423417e-05,
+      "loss": 0.9889364838600159,
+      "step": 232
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 0.3139214515686035,
+      "learning_rate": 1.941660232375614e-05,
+      "loss": 1.4886717796325684,
+      "step": 234
+    },
+    {
+      "epoch": 0.43223443223443225,
+      "grad_norm": 0.3724687993526459,
+      "learning_rate": 1.9403924199820813e-05,
+      "loss": 1.016167163848877,
+      "step": 236
+    },
+    {
+      "epoch": 0.4358974358974359,
+      "grad_norm": 0.17974892258644104,
+      "learning_rate": 1.9391114510906546e-05,
+      "loss": 1.0665429830551147,
+      "step": 238
+    },
+    {
+      "epoch": 0.43956043956043955,
+      "grad_norm": 0.3367801010608673,
+      "learning_rate": 1.937817345755138e-05,
+      "loss": 0.8994011878967285,
+      "step": 240
+    },
+    {
+      "epoch": 0.4432234432234432,
+      "grad_norm": 0.4172503650188446,
+      "learning_rate": 1.9365101242349883e-05,
+      "loss": 0.8775147795677185,
+      "step": 242
+    },
+    {
+      "epoch": 0.4468864468864469,
+      "grad_norm": 0.09528416395187378,
+      "learning_rate": 1.9351898069949985e-05,
+      "loss": 0.5708340406417847,
+      "step": 244
+    },
+    {
+      "epoch": 0.45054945054945056,
+      "grad_norm": 0.16236096620559692,
+      "learning_rate": 1.9338564147049785e-05,
+      "loss": 1.2500817775726318,
+      "step": 246
+    },
+    {
+      "epoch": 0.4542124542124542,
+      "grad_norm": 2.8649377822875977,
+      "learning_rate": 1.9325099682394296e-05,
+      "loss": 0.8762341141700745,
+      "step": 248
+    },
+    {
+      "epoch": 0.45787545787545786,
+      "grad_norm": 0.45987099409103394,
+      "learning_rate": 1.9311504886772183e-05,
+      "loss": 1.2730751037597656,
+      "step": 250
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.11307661235332489,
+      "learning_rate": 1.929777997301248e-05,
+      "loss": 1.1749809980392456,
+      "step": 252
+    },
+    {
+      "epoch": 0.4652014652014652,
+      "grad_norm": 0.277836412191391,
+      "learning_rate": 1.9283925155981228e-05,
+      "loss": 0.9623442888259888,
+      "step": 254
+    },
+    {
+      "epoch": 0.46886446886446886,
+      "grad_norm": 0.15862256288528442,
+      "learning_rate": 1.9269940652578143e-05,
+      "loss": 1.2541102170944214,
+      "step": 256
+    },
+    {
+      "epoch": 0.4725274725274725,
+      "grad_norm": 0.20002801716327667,
+      "learning_rate": 1.9255826681733194e-05,
+      "loss": 1.2813372611999512,
+      "step": 258
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 0.0928114503622055,
+      "learning_rate": 1.924158346440319e-05,
+      "loss": 0.7289823293685913,
+      "step": 260
+    },
+    {
+      "epoch": 0.47985347985347987,
+      "grad_norm": 0.1095692440867424,
+      "learning_rate": 1.9227211223568317e-05,
+      "loss": 1.1414886713027954,
+      "step": 262
+    },
+    {
+      "epoch": 0.4835164835164835,
+      "grad_norm": 0.13363513350486755,
+      "learning_rate": 1.9212710184228654e-05,
+      "loss": 1.2244765758514404,
+      "step": 264
+    },
+    {
+      "epoch": 0.48717948717948717,
+      "grad_norm": 0.1490332931280136,
+      "learning_rate": 1.9198080573400634e-05,
+      "loss": 1.4965099096298218,
+      "step": 266
+    },
+    {
+      "epoch": 0.4908424908424908,
+      "grad_norm": 0.43857482075691223,
+      "learning_rate": 1.9183322620113505e-05,
+      "loss": 0.7830735445022583,
+      "step": 268
+    },
+    {
+      "epoch": 0.4945054945054945,
+      "grad_norm": 0.3251877427101135,
+      "learning_rate": 1.916843655540574e-05,
+      "loss": 1.1967202425003052,
+      "step": 270
+    },
+    {
+      "epoch": 0.4981684981684982,
+      "grad_norm": 0.27459704875946045,
+      "learning_rate": 1.915342261232142e-05,
+      "loss": 0.8744536638259888,
+      "step": 272
+    },
+    {
+      "epoch": 0.5018315018315018,
+      "grad_norm": 0.21172094345092773,
+      "learning_rate": 1.913828102590659e-05,
+      "loss": 1.248273253440857,
+      "step": 274
+    },
+    {
+      "epoch": 0.5054945054945055,
+      "grad_norm": 0.3026851713657379,
+      "learning_rate": 1.9123012033205564e-05,
+      "loss": 0.8064572215080261,
+      "step": 276
+    },
+    {
+      "epoch": 0.5091575091575091,
+      "grad_norm": 0.0853167474269867,
+      "learning_rate": 1.9107615873257234e-05,
+      "loss": 0.8765072226524353,
+      "step": 278
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 0.3027302026748657,
+      "learning_rate": 1.909209278709131e-05,
+      "loss": 1.2487154006958008,
+      "step": 280
+    },
+    {
+      "epoch": 0.5164835164835165,
+      "grad_norm": 0.40836283564567566,
+      "learning_rate": 1.9076443017724568e-05,
+      "loss": 1.2448886632919312,
+      "step": 282
+    },
+    {
+      "epoch": 0.5201465201465202,
+      "grad_norm": 0.2344697266817093,
+      "learning_rate": 1.9060666810157025e-05,
+      "loss": 1.2436648607254028,
+      "step": 284
+    },
+    {
+      "epoch": 0.5238095238095238,
+      "grad_norm": 1.9295133352279663,
+      "learning_rate": 1.9044764411368106e-05,
+      "loss": 1.0280476808547974,
+      "step": 286
+    },
+    {
+      "epoch": 0.5274725274725275,
+      "grad_norm": 0.2943372130393982,
+      "learning_rate": 1.9028736070312796e-05,
+      "loss": 1.2490639686584473,
+      "step": 288
+    },
+    {
+      "epoch": 0.5311355311355311,
+      "grad_norm": 0.20403721928596497,
+      "learning_rate": 1.9012582037917713e-05,
+      "loss": 1.2165021896362305,
+      "step": 290
+    },
+    {
+      "epoch": 0.5347985347985348,
+      "grad_norm": 0.07780837267637253,
+      "learning_rate": 1.8996302567077217e-05,
+      "loss": 0.7315054535865784,
+      "step": 292
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.02841615118086338,
+      "learning_rate": 1.897989791264941e-05,
+      "loss": 0.9443866610527039,
+      "step": 294
+    },
+    {
+      "epoch": 0.5421245421245421,
+      "grad_norm": 1.233228087425232,
+      "learning_rate": 1.8963368331452172e-05,
+      "loss": 1.0225800275802612,
+      "step": 296
+    },
+    {
+      "epoch": 0.5457875457875457,
+      "grad_norm": 0.4084673523902893,
+      "learning_rate": 1.8946714082259145e-05,
+      "loss": 1.2971231937408447,
+      "step": 298
+    },
+    {
+      "epoch": 0.5494505494505495,
+      "grad_norm": 0.10006222873926163,
+      "learning_rate": 1.8929935425795655e-05,
+      "loss": 1.1916959285736084,
+      "step": 300
+    },
+    {
+      "epoch": 0.5531135531135531,
+      "grad_norm": 0.09651292860507965,
+      "learning_rate": 1.8913032624734657e-05,
+      "loss": 1.1871459484100342,
+      "step": 302
+    },
+    {
+      "epoch": 0.5567765567765568,
+      "grad_norm": 1.0658628940582275,
+      "learning_rate": 1.8896005943692614e-05,
+      "loss": 0.9745575189590454,
+      "step": 304
+    },
+    {
+      "epoch": 0.5604395604395604,
+      "grad_norm": 1.3608713150024414,
+      "learning_rate": 1.8878855649225346e-05,
+      "loss": 0.9455310106277466,
+      "step": 306
+    },
+    {
+      "epoch": 0.5641025641025641,
+      "grad_norm": 0.2951708436012268,
+      "learning_rate": 1.8861582009823868e-05,
+      "loss": 1.4047762155532837,
+      "step": 308
+    },
+    {
+      "epoch": 0.5677655677655677,
+      "grad_norm": 0.21646811068058014,
+      "learning_rate": 1.884418529591018e-05,
+      "loss": 0.9867649078369141,
+      "step": 310
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.16589109599590302,
+      "learning_rate": 1.882666577983304e-05,
+      "loss": 1.2163362503051758,
+      "step": 312
+    },
+    {
+      "epoch": 0.575091575091575,
+      "grad_norm": 0.13253678381443024,
+      "learning_rate": 1.8809023735863693e-05,
+      "loss": 1.1416099071502686,
+      "step": 314
+    },
+    {
+      "epoch": 0.5787545787545788,
+      "grad_norm": 0.34199172258377075,
+      "learning_rate": 1.879125944019158e-05,
+      "loss": 1.2828210592269897,
+      "step": 316
+    },
+    {
+      "epoch": 0.5824175824175825,
+      "grad_norm": 0.16684100031852722,
+      "learning_rate": 1.8773373170920022e-05,
+      "loss": 1.1197882890701294,
+      "step": 318
+    },
+    {
+      "epoch": 0.5860805860805861,
+      "grad_norm": 0.1224118322134018,
+      "learning_rate": 1.875536520806185e-05,
+      "loss": 1.3375335931777954,
+      "step": 320
+    },
+    {
+      "epoch": 0.5897435897435898,
+      "grad_norm": 0.20716524124145508,
+      "learning_rate": 1.8737235833535033e-05,
+      "loss": 1.5252546072006226,
+      "step": 322
+    },
+    {
+      "epoch": 0.5934065934065934,
+      "grad_norm": 0.15489473938941956,
+      "learning_rate": 1.871898533115827e-05,
+      "loss": 1.2672544717788696,
+      "step": 324
+    },
+    {
+      "epoch": 0.5970695970695971,
+      "grad_norm": 0.10702221095561981,
+      "learning_rate": 1.870061398664653e-05,
+      "loss": 1.359837293624878,
+      "step": 326
+    },
+    {
+      "epoch": 0.6007326007326007,
+      "grad_norm": 0.22020572423934937,
+      "learning_rate": 1.868212208760658e-05,
+      "loss": 1.2261296510696411,
+      "step": 328
+    },
+    {
+      "epoch": 0.6043956043956044,
+      "grad_norm": 0.2609195113182068,
+      "learning_rate": 1.8663509923532514e-05,
+      "loss": 1.1154756546020508,
+      "step": 330
+    },
+    {
+      "epoch": 0.608058608058608,
+      "grad_norm": 0.37561291456222534,
+      "learning_rate": 1.8644777785801175e-05,
+      "loss": 1.1825737953186035,
+      "step": 332
+    },
+    {
+      "epoch": 0.6117216117216118,
+      "grad_norm": 0.32921549677848816,
+      "learning_rate": 1.862592596766763e-05,
+      "loss": 1.2822954654693604,
+      "step": 334
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.40565115213394165,
+      "learning_rate": 1.8606954764260556e-05,
+      "loss": 0.8941524028778076,
+      "step": 336
+    },
+    {
+      "epoch": 0.6190476190476191,
+      "grad_norm": 0.12613314390182495,
+      "learning_rate": 1.8587864472577632e-05,
+      "loss": 1.2352339029312134,
+      "step": 338
+    },
+    {
+      "epoch": 0.6227106227106227,
+      "grad_norm": 0.38321077823638916,
+      "learning_rate": 1.8568655391480882e-05,
+      "loss": 1.2283233404159546,
+      "step": 340
+    },
+    {
+      "epoch": 0.6263736263736264,
+      "grad_norm": 0.17666201293468475,
+      "learning_rate": 1.8549327821692008e-05,
+      "loss": 0.5744314193725586,
+      "step": 342
+    },
+    {
+      "epoch": 0.63003663003663,
+      "grad_norm": 0.18057258427143097,
+      "learning_rate": 1.852988206578767e-05,
+      "loss": 1.4381451606750488,
+      "step": 344
+    },
+    {
+      "epoch": 0.6336996336996337,
+      "grad_norm": 0.29170721769332886,
+      "learning_rate": 1.851031842819475e-05,
+      "loss": 0.6940987706184387,
+      "step": 346
+    },
+    {
+      "epoch": 0.6373626373626373,
+      "grad_norm": 0.3080287277698517,
+      "learning_rate": 1.849063721518559e-05,
+      "loss": 1.1522477865219116,
+      "step": 348
+    },
+    {
+      "epoch": 0.6410256410256411,
+      "grad_norm": 0.5422199964523315,
+      "learning_rate": 1.8470838734873205e-05,
+      "loss": 0.8789457082748413,
+      "step": 350
+    },
+    {
+      "epoch": 0.6446886446886447,
+      "grad_norm": 0.3600572347640991,
+      "learning_rate": 1.8450923297206446e-05,
+      "loss": 0.921578049659729,
+      "step": 352
+    },
+    {
+      "epoch": 0.6483516483516484,
+      "grad_norm": 0.2704165279865265,
+      "learning_rate": 1.8430891213965146e-05,
+      "loss": 0.9506340026855469,
+      "step": 354
+    },
+    {
+      "epoch": 0.652014652014652,
+      "grad_norm": 0.193269282579422,
+      "learning_rate": 1.8410742798755255e-05,
+      "loss": 1.1715575456619263,
+      "step": 356
+    },
+    {
+      "epoch": 0.6556776556776557,
+      "grad_norm": 0.27199605107307434,
+      "learning_rate": 1.8390478367003922e-05,
+      "loss": 1.151785135269165,
+      "step": 358
+    },
+    {
+      "epoch": 0.6593406593406593,
+      "grad_norm": 0.4195846915245056,
+      "learning_rate": 1.8370098235954553e-05,
+      "loss": 0.6752058267593384,
+      "step": 360
+    },
+    {
+      "epoch": 0.663003663003663,
+      "grad_norm": 0.23578767478466034,
+      "learning_rate": 1.834960272466184e-05,
+      "loss": 0.9451608657836914,
+      "step": 362
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.24091851711273193,
+      "learning_rate": 1.832899215398679e-05,
+      "loss": 0.9306972026824951,
+      "step": 364
+    },
+    {
+      "epoch": 0.6703296703296703,
+      "grad_norm": 0.2906991243362427,
+      "learning_rate": 1.8308266846591673e-05,
+      "loss": 1.1896872520446777,
+      "step": 366
+    },
+    {
+      "epoch": 0.673992673992674,
+      "grad_norm": 0.8038737177848816,
+      "learning_rate": 1.828742712693499e-05,
+      "loss": 1.0567468404769897,
+      "step": 368
+    },
+    {
+      "epoch": 0.6776556776556777,
+      "grad_norm": 0.3876494765281677,
+      "learning_rate": 1.8266473321266385e-05,
+      "loss": 1.0929261445999146,
+      "step": 370
+    },
+    {
+      "epoch": 0.6813186813186813,
+      "grad_norm": 0.34046557545661926,
+      "learning_rate": 1.824540575762154e-05,
+      "loss": 1.1846730709075928,
+      "step": 372
+    },
+    {
+      "epoch": 0.684981684981685,
+      "grad_norm": 0.17690998315811157,
+      "learning_rate": 1.8224224765817033e-05,
+      "loss": 1.2002733945846558,
+      "step": 374
+    },
+    {
+      "epoch": 0.6886446886446886,
+      "grad_norm": 0.9796584844589233,
+      "learning_rate": 1.820293067744519e-05,
+      "loss": 0.8656985759735107,
+      "step": 376
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.1290479600429535,
+      "learning_rate": 1.8181523825868882e-05,
+      "loss": 0.8376519680023193,
+      "step": 378
+    },
+    {
+      "epoch": 0.6959706959706959,
+      "grad_norm": 0.5467531085014343,
+      "learning_rate": 1.816000454621631e-05,
+      "loss": 1.0472257137298584,
+      "step": 380
+    },
+    {
+      "epoch": 0.6996336996336996,
+      "grad_norm": 0.1477574110031128,
+      "learning_rate": 1.8138373175375744e-05,
+      "loss": 0.9799332022666931,
+      "step": 382
+    },
+    {
+      "epoch": 0.7032967032967034,
+      "grad_norm": 0.2413100004196167,
+      "learning_rate": 1.8116630051990283e-05,
+      "loss": 1.1842981576919556,
+      "step": 384
+    },
+    {
+      "epoch": 0.706959706959707,
+      "grad_norm": 0.20904967188835144,
+      "learning_rate": 1.8094775516452522e-05,
+      "loss": 1.0900828838348389,
+      "step": 386
+    },
+    {
+      "epoch": 0.7106227106227107,
+      "grad_norm": 0.13387838006019592,
+      "learning_rate": 1.807280991089923e-05,
+      "loss": 0.8862177729606628,
+      "step": 388
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 0.2151884138584137,
+      "learning_rate": 1.8050733579206005e-05,
+      "loss": 1.094993233680725,
+      "step": 390
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 0.12963874638080597,
+      "learning_rate": 1.8028546866981875e-05,
+      "loss": 1.1782118082046509,
+      "step": 392
+    },
+    {
+      "epoch": 0.7216117216117216,
+      "grad_norm": 0.2768458127975464,
+      "learning_rate": 1.8006250121563903e-05,
+      "loss": 1.118064284324646,
+      "step": 394
+    },
+    {
+      "epoch": 0.7252747252747253,
+      "grad_norm": 0.1374271959066391,
+      "learning_rate": 1.798384369201174e-05,
+      "loss": 1.2445368766784668,
+      "step": 396
+    },
+    {
+      "epoch": 0.7289377289377289,
+      "grad_norm": 0.1631670594215393,
+      "learning_rate": 1.796132792910216e-05,
+      "loss": 0.921625554561615,
+      "step": 398
+    },
+    {
+      "epoch": 0.7326007326007326,
+      "grad_norm": 0.13680344820022583,
+      "learning_rate": 1.7938703185323575e-05,
+      "loss": 0.8475565910339355,
+      "step": 400
+    },
+    {
+      "epoch": 0.7362637362637363,
+      "grad_norm": 0.17989139258861542,
+      "learning_rate": 1.7915969814870508e-05,
+      "loss": 1.2503498792648315,
+      "step": 402
+    },
+    {
+      "epoch": 0.73992673992674,
+      "grad_norm": 0.10058735311031342,
+      "learning_rate": 1.789312817363805e-05,
+      "loss": 0.8367084264755249,
+      "step": 404
+    },
+    {
+      "epoch": 0.7435897435897436,
+      "grad_norm": 0.1782654970884323,
+      "learning_rate": 1.7870178619216304e-05,
+      "loss": 1.018811821937561,
+      "step": 406
+    },
+    {
+      "epoch": 0.7472527472527473,
+      "grad_norm": 0.18461261689662933,
+      "learning_rate": 1.784712151088476e-05,
+      "loss": 1.0193127393722534,
+      "step": 408
+    },
+    {
+      "epoch": 0.7509157509157509,
+      "grad_norm": 0.48014354705810547,
+      "learning_rate": 1.782395720960669e-05,
+      "loss": 0.8782365322113037,
+      "step": 410
+    },
+    {
+      "epoch": 0.7545787545787546,
+      "grad_norm": 0.20913685858249664,
+      "learning_rate": 1.780068607802349e-05,
+      "loss": 1.1719920635223389,
+      "step": 412
+    },
+    {
+      "epoch": 0.7582417582417582,
+      "grad_norm": 0.18413150310516357,
+      "learning_rate": 1.7777308480449006e-05,
+      "loss": 0.9983189105987549,
+      "step": 414
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "grad_norm": 0.13885222375392914,
+      "learning_rate": 1.7753824782863827e-05,
+      "loss": 1.2759360074996948,
+      "step": 416
+    },
+    {
+      "epoch": 0.7655677655677655,
+      "grad_norm": 0.05709720030426979,
+      "learning_rate": 1.773023535290956e-05,
+      "loss": 0.6787292957305908,
+      "step": 418
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.18361720442771912,
+      "learning_rate": 1.7706540559883066e-05,
+      "loss": 1.2402327060699463,
+      "step": 420
+    },
+    {
+      "epoch": 0.7728937728937729,
+      "grad_norm": 0.1252235621213913,
+      "learning_rate": 1.7682740774730688e-05,
+      "loss": 0.9981698989868164,
+      "step": 422
+    },
+    {
+      "epoch": 0.7765567765567766,
+      "grad_norm": 0.4331834018230438,
+      "learning_rate": 1.7658836370042443e-05,
+      "loss": 0.4972750246524811,
+      "step": 424
+    },
+    {
+      "epoch": 0.7802197802197802,
+      "grad_norm": 0.24165484309196472,
+      "learning_rate": 1.7634827720046178e-05,
+      "loss": 0.7953031063079834,
+      "step": 426
+    },
+    {
+      "epoch": 0.7838827838827839,
+      "grad_norm": 0.28377044200897217,
+      "learning_rate": 1.7610715200601727e-05,
+      "loss": 1.0705211162567139,
+      "step": 428
+    },
+    {
+      "epoch": 0.7875457875457875,
+      "grad_norm": 0.14546895027160645,
+      "learning_rate": 1.7586499189195016e-05,
+      "loss": 1.2132048606872559,
+      "step": 430
+    },
+    {
+      "epoch": 0.7912087912087912,
+      "grad_norm": 0.12278863787651062,
+      "learning_rate": 1.7562180064932158e-05,
+      "loss": 1.2941319942474365,
+      "step": 432
+    },
+    {
+      "epoch": 0.7948717948717948,
+      "grad_norm": 0.3506087064743042,
+      "learning_rate": 1.7537758208533516e-05,
+      "loss": 0.8893874883651733,
+      "step": 434
+    },
+    {
+      "epoch": 0.7985347985347986,
+      "grad_norm": 0.1808759868144989,
+      "learning_rate": 1.7513234002327738e-05,
+      "loss": 0.963141918182373,
+      "step": 436
+    },
+    {
+      "epoch": 0.8021978021978022,
+      "grad_norm": 0.5765663385391235,
+      "learning_rate": 1.748860783024579e-05,
+      "loss": 0.8931108117103577,
+      "step": 438
+    },
+    {
+      "epoch": 0.8058608058608059,
+      "grad_norm": 0.16014474630355835,
+      "learning_rate": 1.746388007781492e-05,
+      "loss": 1.3087202310562134,
+      "step": 440
+    },
+    {
+      "epoch": 0.8095238095238095,
+      "grad_norm": 0.10628227144479752,
+      "learning_rate": 1.7439051132152644e-05,
+      "loss": 1.202932596206665,
+      "step": 442
+    },
+    {
+      "epoch": 0.8131868131868132,
+      "grad_norm": 0.1464785486459732,
+      "learning_rate": 1.741412138196067e-05,
+      "loss": 1.2049145698547363,
+      "step": 444
+    },
+    {
+      "epoch": 0.8168498168498168,
+      "grad_norm": 0.25733712315559387,
+      "learning_rate": 1.738909121751882e-05,
+      "loss": 1.2221276760101318,
+      "step": 446
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 0.35550013184547424,
+      "learning_rate": 1.736396103067893e-05,
+      "loss": 1.230087161064148,
+      "step": 448
+    },
+    {
+      "epoch": 0.8241758241758241,
+      "grad_norm": 0.18633843958377838,
+      "learning_rate": 1.7338731214858688e-05,
+      "loss": 1.3565971851348877,
+      "step": 450
+    },
+    {
+      "epoch": 0.8278388278388278,
+      "grad_norm": 0.34363844990730286,
+      "learning_rate": 1.7313402165035504e-05,
+      "loss": 0.9984432458877563,
+      "step": 452
+    },
+    {
+      "epoch": 0.8315018315018315,
+      "grad_norm": 0.05576507747173309,
+      "learning_rate": 1.728797427774031e-05,
+      "loss": 0.4852646291255951,
+      "step": 454
+    },
+    {
+      "epoch": 0.8351648351648352,
+      "grad_norm": 0.07310578972101212,
+      "learning_rate": 1.7262447951051366e-05,
+      "loss": 0.8963858485221863,
+      "step": 456
+    },
+    {
+      "epoch": 0.8388278388278388,
+      "grad_norm": 0.28023576736450195,
+      "learning_rate": 1.7236823584587995e-05,
+      "loss": 0.8434333801269531,
+      "step": 458
+    },
+    {
+      "epoch": 0.8424908424908425,
+      "grad_norm": 1.2096292972564697,
+      "learning_rate": 1.7211101579504382e-05,
+      "loss": 1.029900312423706,
+      "step": 460
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.3247793912887573,
+      "learning_rate": 1.7185282338483243e-05,
+      "loss": 1.2301326990127563,
+      "step": 462
+    },
+    {
+      "epoch": 0.8498168498168498,
+      "grad_norm": 0.11958979070186615,
+      "learning_rate": 1.7159366265729537e-05,
+      "loss": 1.1807194948196411,
+      "step": 464
+    },
+    {
+      "epoch": 0.8534798534798534,
+      "grad_norm": 0.04206932708621025,
+      "learning_rate": 1.713335376696416e-05,
+      "loss": 1.2045433521270752,
+      "step": 466
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.15847940742969513,
+      "learning_rate": 1.7107245249417556e-05,
+      "loss": 0.8860416412353516,
+      "step": 468
+    },
+    {
+      "epoch": 0.8608058608058609,
+      "grad_norm": 0.2803979516029358,
+      "learning_rate": 1.7081041121823375e-05,
+      "loss": 0.9149615168571472,
+      "step": 470
+    },
+    {
+      "epoch": 0.8644688644688645,
+      "grad_norm": 0.13727536797523499,
+      "learning_rate": 1.705474179441205e-05,
+      "loss": 1.1724745035171509,
+      "step": 472
+    },
+    {
+      "epoch": 0.8681318681318682,
+      "grad_norm": 0.09685394912958145,
+      "learning_rate": 1.7028347678904388e-05,
+      "loss": 0.863320529460907,
+      "step": 474
+    },
+    {
+      "epoch": 0.8717948717948718,
+      "grad_norm": 0.17197725176811218,
+      "learning_rate": 1.700185918850512e-05,
+      "loss": 1.096718192100525,
+      "step": 476
+    },
+    {
+      "epoch": 0.8754578754578755,
+      "grad_norm": 0.5562029480934143,
+      "learning_rate": 1.6975276737896443e-05,
+      "loss": 1.0467816591262817,
+      "step": 478
+    },
+    {
+      "epoch": 0.8791208791208791,
+      "grad_norm": 2.393152952194214,
+      "learning_rate": 1.69486007432315e-05,
+      "loss": 1.0700321197509766,
+      "step": 480
+    },
+    {
+      "epoch": 0.8827838827838828,
+      "grad_norm": 0.2319096475839615,
+      "learning_rate": 1.6921831622127905e-05,
+      "loss": 1.1908187866210938,
+      "step": 482
+    },
+    {
+      "epoch": 0.8864468864468864,
+      "grad_norm": 0.1952940672636032,
+      "learning_rate": 1.6894969793661163e-05,
+      "loss": 1.2682039737701416,
+      "step": 484
+    },
+    {
+      "epoch": 0.8901098901098901,
+      "grad_norm": 0.1970641314983368,
+      "learning_rate": 1.686801567835814e-05,
+      "loss": 0.9106331467628479,
+      "step": 486
+    },
+    {
+      "epoch": 0.8937728937728938,
+      "grad_norm": 0.3220268785953522,
+      "learning_rate": 1.6840969698190467e-05,
+      "loss": 1.1676743030548096,
+      "step": 488
+    },
+    {
+      "epoch": 0.8974358974358975,
+      "grad_norm": 0.2038663923740387,
+      "learning_rate": 1.6813832276567942e-05,
+      "loss": 1.1185270547866821,
+      "step": 490
+    },
+    {
+      "epoch": 0.9010989010989011,
+      "grad_norm": 0.17804059386253357,
+      "learning_rate": 1.6786603838331894e-05,
+      "loss": 1.0551954507827759,
+      "step": 492
+    },
+    {
+      "epoch": 0.9047619047619048,
+      "grad_norm": 0.4494330883026123,
+      "learning_rate": 1.6759284809748522e-05,
+      "loss": 0.5789248943328857,
+      "step": 494
+    },
+    {
+      "epoch": 0.9084249084249084,
+      "grad_norm": 0.37853649258613586,
+      "learning_rate": 1.673187561850225e-05,
+      "loss": 1.2827361822128296,
+      "step": 496
+    },
+    {
+      "epoch": 0.9120879120879121,
+      "grad_norm": 0.2299915850162506,
+      "learning_rate": 1.6704376693689003e-05,
+      "loss": 1.1320176124572754,
+      "step": 498
+    },
+    {
+      "epoch": 0.9157509157509157,
+      "grad_norm": 0.2586250603199005,
+      "learning_rate": 1.6676788465809506e-05,
+      "loss": 0.8153626322746277,
+      "step": 500
+    },
+    {
+      "epoch": 0.9194139194139194,
+      "grad_norm": 0.061475999653339386,
+      "learning_rate": 1.6649111366762552e-05,
+      "loss": 0.8643592596054077,
+      "step": 502
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.24921494722366333,
+      "learning_rate": 1.66213458298382e-05,
+      "loss": 0.9420300126075745,
+      "step": 504
+    },
+    {
+      "epoch": 0.9267399267399268,
+      "grad_norm": 0.0703081414103508,
+      "learning_rate": 1.659349228971105e-05,
+      "loss": 0.8538585305213928,
+      "step": 506
+    },
+    {
+      "epoch": 0.9304029304029304,
+      "grad_norm": 0.14408151805400848,
+      "learning_rate": 1.6565551182433382e-05,
+      "loss": 1.1675981283187866,
+      "step": 508
+    },
+    {
+      "epoch": 0.9340659340659341,
+      "grad_norm": 0.2426132708787918,
+      "learning_rate": 1.6537522945428386e-05,
+      "loss": 1.2252295017242432,
+      "step": 510
+    },
+    {
+      "epoch": 0.9377289377289377,
+      "grad_norm": 0.5424879789352417,
+      "learning_rate": 1.6509408017483258e-05,
+      "loss": 1.1778167486190796,
+      "step": 512
+    },
+    {
+      "epoch": 0.9413919413919414,
+      "grad_norm": 0.142863929271698,
+      "learning_rate": 1.6481206838742362e-05,
+      "loss": 0.9604276418685913,
+      "step": 514
+    },
+    {
+      "epoch": 0.945054945054945,
+      "grad_norm": 0.27180016040802,
+      "learning_rate": 1.645291985070034e-05,
+      "loss": 1.1854896545410156,
+      "step": 516
+    },
+    {
+      "epoch": 0.9487179487179487,
+      "grad_norm": 0.16782020032405853,
+      "learning_rate": 1.64245474961952e-05,
+      "loss": 1.2195581197738647,
+      "step": 518
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 0.15222570300102234,
+      "learning_rate": 1.639609021940136e-05,
+      "loss": 1.218988299369812,
+      "step": 520
+    },
+    {
+      "epoch": 0.9560439560439561,
+      "grad_norm": 0.2450752556324005,
+      "learning_rate": 1.6367548465822723e-05,
+      "loss": 0.8905650973320007,
+      "step": 522
+    },
+    {
+      "epoch": 0.9597069597069597,
+      "grad_norm": 0.5845417380332947,
+      "learning_rate": 1.6338922682285697e-05,
+      "loss": 1.0242419242858887,
+      "step": 524
+    },
+    {
+      "epoch": 0.9633699633699634,
+      "grad_norm": 2.221842050552368,
+      "learning_rate": 1.6310213316932187e-05,
+      "loss": 0.9667062759399414,
+      "step": 526
+    },
+    {
+      "epoch": 0.967032967032967,
+      "grad_norm": 0.15865476429462433,
+      "learning_rate": 1.6281420819212578e-05,
+      "loss": 0.6576095819473267,
+      "step": 528
+    },
+    {
+      "epoch": 0.9706959706959707,
+      "grad_norm": 0.10818128287792206,
+      "learning_rate": 1.6252545639878728e-05,
+      "loss": 0.907448947429657,
+      "step": 530
+    },
+    {
+      "epoch": 0.9743589743589743,
+      "grad_norm": 1.5423500537872314,
+      "learning_rate": 1.6223588230976874e-05,
+      "loss": 1.3604565858840942,
+      "step": 532
+    },
+    {
+      "epoch": 0.978021978021978,
+      "grad_norm": 0.38299986720085144,
+      "learning_rate": 1.6194549045840582e-05,
+      "loss": 0.604587733745575,
+      "step": 534
+    },
+    {
+      "epoch": 0.9816849816849816,
+      "grad_norm": 0.35183185338974,
+      "learning_rate": 1.616542853908363e-05,
+      "loss": 0.8585751056671143,
+      "step": 536
+    },
+    {
+      "epoch": 0.9853479853479854,
+      "grad_norm": 8.051664352416992,
+      "learning_rate": 1.6136227166592912e-05,
+      "loss": 0.8037823438644409,
+      "step": 538
+    },
+    {
+      "epoch": 0.989010989010989,
+      "grad_norm": 0.36428943276405334,
+      "learning_rate": 1.6106945385521286e-05,
+      "loss": 1.1241040229797363,
+      "step": 540
+    },
+    {
+      "epoch": 0.9926739926739927,
+      "grad_norm": 0.39597558975219727,
+      "learning_rate": 1.6077583654280416e-05,
+      "loss": 1.1745156049728394,
+      "step": 542
+    },
+    {
+      "epoch": 0.9963369963369964,
+      "grad_norm": 1.5844489336013794,
+      "learning_rate": 1.60481424325336e-05,
+      "loss": 1.186415195465088,
+      "step": 544
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.17938172817230225,
+      "learning_rate": 1.6018622181188594e-05,
+      "loss": 1.3618619441986084,
+      "step": 546
+    },
+    {
+      "epoch": 1.0036630036630036,
+      "grad_norm": 0.3727249503135681,
+      "learning_rate": 1.598902336239035e-05,
+      "loss": 0.9441794157028198,
+      "step": 548
+    },
+    {
+      "epoch": 1.0073260073260073,
+      "grad_norm": 0.24385389685630798,
+      "learning_rate": 1.595934643951382e-05,
+      "loss": 1.1801525354385376,
+      "step": 550
+    },
+    {
+      "epoch": 1.010989010989011,
+      "grad_norm": 0.1038205698132515,
+      "learning_rate": 1.5929591877156694e-05,
+      "loss": 0.6706070899963379,
+      "step": 552
+    },
+    {
+      "epoch": 1.0146520146520146,
+      "grad_norm": 0.2101455181837082,
+      "learning_rate": 1.5899760141132115e-05,
+      "loss": 1.1351317167282104,
+      "step": 554
+    },
+    {
+      "epoch": 1.0183150183150182,
+      "grad_norm": 0.2257443219423294,
+      "learning_rate": 1.58698516984614e-05,
+      "loss": 1.0271389484405518,
+      "step": 556
+    },
+    {
+      "epoch": 1.021978021978022,
+      "grad_norm": 0.1515074372291565,
+      "learning_rate": 1.583986701736672e-05,
+      "loss": 1.2197285890579224,
+      "step": 558
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "grad_norm": 0.11847762763500214,
+      "learning_rate": 1.5809806567263767e-05,
+      "loss": 0.9212762713432312,
+      "step": 560
+    },
+    {
+      "epoch": 1.0293040293040292,
+      "grad_norm": 0.40212902426719666,
+      "learning_rate": 1.577967081875442e-05,
+      "loss": 1.1707442998886108,
+      "step": 562
+    },
+    {
+      "epoch": 1.032967032967033,
+      "grad_norm": 0.2982182502746582,
+      "learning_rate": 1.574946024361936e-05,
+      "loss": 1.1547435522079468,
+      "step": 564
+    },
+    {
+      "epoch": 1.0366300366300367,
+      "grad_norm": 0.34150704741477966,
+      "learning_rate": 1.5719175314810706e-05,
+      "loss": 1.044006109237671,
+      "step": 566
+    },
+    {
+      "epoch": 1.0402930402930404,
+      "grad_norm": 0.42247244715690613,
+      "learning_rate": 1.568881650644458e-05,
+      "loss": 1.0475645065307617,
+      "step": 568
+    },
+    {
+      "epoch": 1.043956043956044,
+      "grad_norm": 0.2374022752046585,
+      "learning_rate": 1.565838429379371e-05,
+      "loss": 1.1118239164352417,
+      "step": 570
+    },
+    {
+      "epoch": 1.0476190476190477,
+      "grad_norm": 0.48579737544059753,
+      "learning_rate": 1.5627879153279986e-05,
+      "loss": 1.2503812313079834,
+      "step": 572
+    },
+    {
+      "epoch": 1.0512820512820513,
+      "grad_norm": 0.6810855269432068,
+      "learning_rate": 1.559730156246699e-05,
+      "loss": 1.0755858421325684,
+      "step": 574
+    },
+    {
+      "epoch": 1.054945054945055,
+      "grad_norm": 0.2626459002494812,
+      "learning_rate": 1.5566652000052533e-05,
+      "loss": 1.2320541143417358,
+      "step": 576
+    },
+    {
+      "epoch": 1.0586080586080586,
+      "grad_norm": 1.1261385679244995,
+      "learning_rate": 1.553593094586115e-05,
+      "loss": 1.2240521907806396,
+      "step": 578
+    },
+    {
+      "epoch": 1.0622710622710623,
+      "grad_norm": 0.27919474244117737,
+      "learning_rate": 1.5505138880836595e-05,
+      "loss": 1.2212425470352173,
+      "step": 580
+    },
+    {
+      "epoch": 1.065934065934066,
+      "grad_norm": 0.2261185646057129,
+      "learning_rate": 1.5474276287034305e-05,
+      "loss": 0.9912468194961548,
+      "step": 582
+    },
+    {
+      "epoch": 1.0695970695970696,
+      "grad_norm": 0.23183542490005493,
+      "learning_rate": 1.544334364761387e-05,
+      "loss": 1.147226333618164,
+      "step": 584
+    },
+    {
+      "epoch": 1.0732600732600732,
+      "grad_norm": 0.3248075246810913,
+      "learning_rate": 1.541234144683144e-05,
+      "loss": 1.0802392959594727,
+      "step": 586
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.23752079904079437,
+      "learning_rate": 1.5381270170032173e-05,
+      "loss": 0.821092426776886,
+      "step": 588
+    },
+    {
+      "epoch": 1.0805860805860805,
+      "grad_norm": 0.20078615844249725,
+      "learning_rate": 1.5350130303642625e-05,
+      "loss": 1.2033530473709106,
+      "step": 590
+    },
+    {
+      "epoch": 1.0842490842490842,
+      "grad_norm": 0.13361191749572754,
+      "learning_rate": 1.5318922335163128e-05,
+      "loss": 1.0244792699813843,
+      "step": 592
+    },
+    {
+      "epoch": 1.0879120879120878,
+      "grad_norm": 0.21560516953468323,
+      "learning_rate": 1.5287646753160174e-05,
+      "loss": 0.9856408834457397,
+      "step": 594
+    },
+    {
+      "epoch": 1.0915750915750915,
+      "grad_norm": 0.1309068351984024,
+      "learning_rate": 1.5256304047258739e-05,
+      "loss": 1.0107301473617554,
+      "step": 596
+    },
+    {
+      "epoch": 1.0952380952380953,
+      "grad_norm": 0.40338659286499023,
+      "learning_rate": 1.522489470813466e-05,
+      "loss": 1.0993225574493408,
+      "step": 598
+    },
+    {
+      "epoch": 1.098901098901099,
+      "grad_norm": 0.3209727108478546,
+      "learning_rate": 1.5193419227506913e-05,
+      "loss": 1.1531182527542114,
+      "step": 600
+    },
+    {
+      "epoch": 1.1025641025641026,
+      "grad_norm": 0.1865486353635788,
+      "learning_rate": 1.5161878098129937e-05,
+      "loss": 0.8158029317855835,
+      "step": 602
+    },
+    {
+      "epoch": 1.1062271062271063,
+      "grad_norm": 0.3306953012943268,
+      "learning_rate": 1.5130271813785908e-05,
+      "loss": 0.9586283564567566,
+      "step": 604
+    },
+    {
+      "epoch": 1.10989010989011,
+      "grad_norm": 0.1971583068370819,
+      "learning_rate": 1.509860086927703e-05,
+      "loss": 0.8959888815879822,
+      "step": 606
+    },
+    {
+      "epoch": 1.1135531135531136,
+      "grad_norm": 0.34381699562072754,
+      "learning_rate": 1.5066865760417757e-05,
+      "loss": 1.2862759828567505,
+      "step": 608
+    },
+    {
+      "epoch": 1.1172161172161172,
+      "grad_norm": 0.27361565828323364,
+      "learning_rate": 1.5035066984027053e-05,
+      "loss": 0.7218859195709229,
+      "step": 610
+    },
+    {
+      "epoch": 1.120879120879121,
+      "grad_norm": 0.3765149414539337,
+      "learning_rate": 1.5003205037920616e-05,
+      "loss": 1.1658059358596802,
+      "step": 612
+    },
+    {
+      "epoch": 1.1245421245421245,
+      "grad_norm": 0.21422307193279266,
+      "learning_rate": 1.497128042090307e-05,
+      "loss": 0.9905625581741333,
+      "step": 614
+    },
+    {
+      "epoch": 1.1282051282051282,
+      "grad_norm": 0.1535208374261856,
+      "learning_rate": 1.493929363276017e-05,
+      "loss": 1.1560765504837036,
+      "step": 616
+    },
+    {
+      "epoch": 1.1318681318681318,
+      "grad_norm": 0.22722600400447845,
+      "learning_rate": 1.4907245174250957e-05,
+      "loss": 0.41150620579719543,
+      "step": 618
+    },
+    {
+      "epoch": 1.1355311355311355,
+      "grad_norm": 0.1882268637418747,
+      "learning_rate": 1.4875135547099953e-05,
+      "loss": 1.0893880128860474,
+      "step": 620
+    },
+    {
+      "epoch": 1.1391941391941391,
+      "grad_norm": 1.5121841430664062,
+      "learning_rate": 1.484296525398927e-05,
+      "loss": 0.7574386596679688,
+      "step": 622
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.581392228603363,
+      "learning_rate": 1.4810734798550769e-05,
+      "loss": 1.0913819074630737,
+      "step": 624
+    },
+    {
+      "epoch": 1.1465201465201464,
+      "grad_norm": 0.2060534954071045,
+      "learning_rate": 1.4778444685358147e-05,
+      "loss": 1.3770023584365845,
+      "step": 626
+    },
+    {
+      "epoch": 1.15018315018315,
+      "grad_norm": 0.27674826979637146,
+      "learning_rate": 1.4746095419919075e-05,
+      "loss": 0.4543880820274353,
+      "step": 628
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.060255009680986404,
+      "learning_rate": 1.4713687508667251e-05,
+      "loss": 1.114593744277954,
+      "step": 630
+    },
+    {
+      "epoch": 1.1575091575091574,
+      "grad_norm": 1.074992060661316,
+      "learning_rate": 1.4681221458954484e-05,
+      "loss": 1.0868229866027832,
+      "step": 632
+    },
+    {
+      "epoch": 1.1611721611721613,
+      "grad_norm": 0.42347779870033264,
+      "learning_rate": 1.4648697779042754e-05,
+      "loss": 0.8624401092529297,
+      "step": 634
+    },
+    {
+      "epoch": 1.164835164835165,
+      "grad_norm": 0.3118894398212433,
+      "learning_rate": 1.461611697809625e-05,
+      "loss": 0.9895141124725342,
+      "step": 636
+    },
+    {
+      "epoch": 1.1684981684981686,
+      "grad_norm": 0.22496363520622253,
+      "learning_rate": 1.4583479566173401e-05,
+      "loss": 1.1844947338104248,
+      "step": 638
+    },
+    {
+      "epoch": 1.1721611721611722,
+      "grad_norm": 0.07173963636159897,
+      "learning_rate": 1.4550786054218902e-05,
+      "loss": 0.7458541989326477,
+      "step": 640
+    },
+    {
+      "epoch": 1.1758241758241759,
+      "grad_norm": 0.26074904203414917,
+      "learning_rate": 1.4518036954055685e-05,
+      "loss": 1.1217985153198242,
+      "step": 642
+    },
+    {
+      "epoch": 1.1794871794871795,
+      "grad_norm": 0.49440035223960876,
+      "learning_rate": 1.4485232778376945e-05,
+      "loss": 1.041925311088562,
+      "step": 644
+    },
+    {
+      "epoch": 1.1831501831501832,
+      "grad_norm": 0.214096337556839,
+      "learning_rate": 1.4452374040738078e-05,
+      "loss": 0.7592092156410217,
+      "step": 646
+    },
+    {
+      "epoch": 1.1868131868131868,
+      "grad_norm": 1.0578991174697876,
+      "learning_rate": 1.4419461255548666e-05,
+      "loss": 0.8962647914886475,
+      "step": 648
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 0.26899221539497375,
+      "learning_rate": 1.4386494938064417e-05,
+      "loss": 1.141674280166626,
+      "step": 650
+    },
+    {
+      "epoch": 1.1941391941391941,
+      "grad_norm": 0.21340245008468628,
+      "learning_rate": 1.4353475604379093e-05,
+      "loss": 0.6671714186668396,
+      "step": 652
+    },
+    {
+      "epoch": 1.1978021978021978,
+      "grad_norm": 0.21647410094738007,
+      "learning_rate": 1.4320403771416438e-05,
+      "loss": 1.2914996147155762,
+      "step": 654
+    },
+    {
+      "epoch": 1.2014652014652014,
+      "grad_norm": 0.2455645203590393,
+      "learning_rate": 1.4287279956922076e-05,
+      "loss": 0.8283839225769043,
+      "step": 656
+    },
+    {
+      "epoch": 1.205128205128205,
+      "grad_norm": 1.0074414014816284,
+      "learning_rate": 1.4254104679455416e-05,
+      "loss": 0.808253824710846,
+      "step": 658
+    },
+    {
+      "epoch": 1.2087912087912087,
+      "grad_norm": 0.19323746860027313,
+      "learning_rate": 1.4220878458381523e-05,
+      "loss": 1.1553109884262085,
+      "step": 660
+    },
+    {
+      "epoch": 1.2124542124542124,
+      "grad_norm": 0.48162999749183655,
+      "learning_rate": 1.418760181386301e-05,
+      "loss": 1.0250654220581055,
+      "step": 662
+    },
+    {
+      "epoch": 1.2161172161172162,
+      "grad_norm": 0.23541465401649475,
+      "learning_rate": 1.4154275266851856e-05,
+      "loss": 0.8853683471679688,
+      "step": 664
+    },
+    {
+      "epoch": 1.2197802197802199,
+      "grad_norm": 0.5028203129768372,
+      "learning_rate": 1.4120899339081291e-05,
+      "loss": 0.8249969482421875,
+      "step": 666
+    },
+    {
+      "epoch": 1.2234432234432235,
+      "grad_norm": 0.2646612823009491,
+      "learning_rate": 1.4087474553057599e-05,
+      "loss": 0.9055181741714478,
+      "step": 668
+    },
+    {
+      "epoch": 1.2271062271062272,
+      "grad_norm": 0.2554120123386383,
+      "learning_rate": 1.405400143205195e-05,
+      "loss": 0.6745082139968872,
+      "step": 670
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.25614455342292786,
+      "learning_rate": 1.4020480500092217e-05,
+      "loss": 1.1828240156173706,
+      "step": 672
+    },
+    {
+      "epoch": 1.2344322344322345,
+      "grad_norm": 0.5676877498626709,
+      "learning_rate": 1.3986912281954745e-05,
+      "loss": 1.1596636772155762,
+      "step": 674
+    },
+    {
+      "epoch": 1.2380952380952381,
+      "grad_norm": 0.25206154584884644,
+      "learning_rate": 1.3953297303156174e-05,
+      "loss": 1.1791561841964722,
+      "step": 676
+    },
+    {
+      "epoch": 1.2417582417582418,
+      "grad_norm": 0.3657374382019043,
+      "learning_rate": 1.391963608994517e-05,
+      "loss": 0.7706201076507568,
+      "step": 678
+    },
+    {
+      "epoch": 1.2454212454212454,
+      "grad_norm": 0.9007764458656311,
+      "learning_rate": 1.3885929169294218e-05,
+      "loss": 0.8264885544776917,
+      "step": 680
+    },
+    {
+      "epoch": 1.249084249084249,
+      "grad_norm": 0.6578343510627747,
+      "learning_rate": 1.3852177068891364e-05,
+      "loss": 1.192352294921875,
+      "step": 682
+    },
+    {
+      "epoch": 1.2527472527472527,
+      "grad_norm": 0.17964546382427216,
+      "learning_rate": 1.3818380317131946e-05,
+      "loss": 1.1608870029449463,
+      "step": 684
+    },
+    {
+      "epoch": 1.2564102564102564,
+      "grad_norm": 0.21571148931980133,
+      "learning_rate": 1.3784539443110323e-05,
+      "loss": 0.8176043629646301,
+      "step": 686
+    },
+    {
+      "epoch": 1.26007326007326,
+      "grad_norm": 0.2216089963912964,
+      "learning_rate": 1.375065497661161e-05,
+      "loss": 1.1090242862701416,
+      "step": 688
+    },
+    {
+      "epoch": 1.2637362637362637,
+      "grad_norm": 0.2962993085384369,
+      "learning_rate": 1.3716727448103356e-05,
+      "loss": 1.1419543027877808,
+      "step": 690
+    },
+    {
+      "epoch": 1.2673992673992673,
+      "grad_norm": 0.1734267771244049,
+      "learning_rate": 1.3682757388727261e-05,
+      "loss": 1.1804542541503906,
+      "step": 692
+    },
+    {
+      "epoch": 1.271062271062271,
+      "grad_norm": 0.4003927707672119,
+      "learning_rate": 1.3648745330290848e-05,
+      "loss": 0.9813081622123718,
+      "step": 694
+    },
+    {
+      "epoch": 1.2747252747252746,
+      "grad_norm": 0.7455025315284729,
+      "learning_rate": 1.361469180525916e-05,
+      "loss": 1.0020716190338135,
+      "step": 696
+    },
+    {
+      "epoch": 1.2783882783882783,
+      "grad_norm": 0.23015426099300385,
+      "learning_rate": 1.358059734674638e-05,
+      "loss": 0.9081999063491821,
+      "step": 698
+    },
+    {
+      "epoch": 1.282051282051282,
+      "grad_norm": 0.2477727234363556,
+      "learning_rate": 1.3546462488507532e-05,
+      "loss": 0.6512075662612915,
+      "step": 700
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 0.2636442184448242,
+      "learning_rate": 1.3512287764930102e-05,
+      "loss": 0.5245524644851685,
+      "step": 702
+    },
+    {
+      "epoch": 1.2893772893772895,
+      "grad_norm": 0.3451531231403351,
+      "learning_rate": 1.347807371102567e-05,
+      "loss": 1.2959914207458496,
+      "step": 704
+    },
+    {
+      "epoch": 1.293040293040293,
+      "grad_norm": 0.19401608407497406,
+      "learning_rate": 1.3443820862421542e-05,
+      "loss": 0.9994240403175354,
+      "step": 706
+    },
+    {
+      "epoch": 1.2967032967032968,
+      "grad_norm": 0.09005500376224518,
+      "learning_rate": 1.3409529755352361e-05,
+      "loss": 0.955507755279541,
+      "step": 708
+    },
+    {
+      "epoch": 1.3003663003663004,
+      "grad_norm": 0.14722907543182373,
+      "learning_rate": 1.3375200926651719e-05,
+      "loss": 0.5620253682136536,
+      "step": 710
+    },
+    {
+      "epoch": 1.304029304029304,
+      "grad_norm": 1.1694523096084595,
+      "learning_rate": 1.3340834913743742e-05,
+      "loss": 0.8808104991912842,
+      "step": 712
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.21442115306854248,
+      "learning_rate": 1.3306432254634676e-05,
+      "loss": 1.2082892656326294,
+      "step": 714
+    },
+    {
+      "epoch": 1.3113553113553114,
+      "grad_norm": 0.3201224207878113,
+      "learning_rate": 1.3271993487904485e-05,
+      "loss": 1.1394985914230347,
+      "step": 716
+    },
+    {
+      "epoch": 1.315018315018315,
+      "grad_norm": 0.18615205585956573,
+      "learning_rate": 1.3237519152698392e-05,
+      "loss": 0.9257374405860901,
+      "step": 718
+    },
+    {
+      "epoch": 1.3186813186813187,
+      "grad_norm": 0.23433499038219452,
+      "learning_rate": 1.3203009788718454e-05,
+      "loss": 0.92364901304245,
+      "step": 720
+    },
+    {
+      "epoch": 1.3223443223443223,
+      "grad_norm": 0.5179888606071472,
+      "learning_rate": 1.3168465936215114e-05,
+      "loss": 0.9131177067756653,
+      "step": 722
+    },
+    {
+      "epoch": 1.326007326007326,
+      "grad_norm": 0.27583321928977966,
+      "learning_rate": 1.3133888135978733e-05,
+      "loss": 1.2074042558670044,
+      "step": 724
+    },
+    {
+      "epoch": 1.3296703296703296,
+      "grad_norm": 0.43936124444007874,
+      "learning_rate": 1.3099276929331132e-05,
+      "loss": 1.0659313201904297,
+      "step": 726
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 1.1440553665161133,
+      "learning_rate": 1.3064632858117123e-05,
+      "loss": 1.1416211128234863,
+      "step": 728
+    },
+    {
+      "epoch": 1.3369963369963371,
+      "grad_norm": 0.4313775599002838,
+      "learning_rate": 1.3029956464696006e-05,
+      "loss": 0.7388544082641602,
+      "step": 730
+    },
+    {
+      "epoch": 1.3406593406593408,
+      "grad_norm": 0.38146156072616577,
+      "learning_rate": 1.2995248291933099e-05,
+      "loss": 0.8170838356018066,
+      "step": 732
+    },
+    {
+      "epoch": 1.3443223443223444,
+      "grad_norm": 0.32572418451309204,
+      "learning_rate": 1.296050888319123e-05,
+      "loss": 0.6172389984130859,
+      "step": 734
+    },
+    {
+      "epoch": 1.347985347985348,
+      "grad_norm": 0.29194140434265137,
+      "learning_rate": 1.2925738782322232e-05,
+      "loss": 1.132319450378418,
+      "step": 736
+    },
+    {
+      "epoch": 1.3516483516483517,
+      "grad_norm": 0.29850953817367554,
+      "learning_rate": 1.2890938533658429e-05,
+      "loss": 0.7506582736968994,
+      "step": 738
+    },
+    {
+      "epoch": 1.3553113553113554,
+      "grad_norm": 0.275314599275589,
+      "learning_rate": 1.2856108682004116e-05,
+      "loss": 1.0084264278411865,
+      "step": 740
+    },
+    {
+      "epoch": 1.358974358974359,
+      "grad_norm": 1.8079997301101685,
+      "learning_rate": 1.282124977262702e-05,
+      "loss": 0.6014983057975769,
+      "step": 742
+    },
+    {
+      "epoch": 1.3626373626373627,
+      "grad_norm": 0.39328068494796753,
+      "learning_rate": 1.2786362351249785e-05,
+      "loss": 1.23367178440094,
+      "step": 744
+    },
+    {
+      "epoch": 1.3663003663003663,
+      "grad_norm": 0.19623665511608124,
+      "learning_rate": 1.2751446964041405e-05,
+      "loss": 1.0585216283798218,
+      "step": 746
+    },
+    {
+      "epoch": 1.36996336996337,
+      "grad_norm": 0.2618620991706848,
+      "learning_rate": 1.2716504157608693e-05,
+      "loss": 1.06695556640625,
+      "step": 748
+    },
+    {
+      "epoch": 1.3736263736263736,
+      "grad_norm": 0.2631770968437195,
+      "learning_rate": 1.2681534478987703e-05,
+      "loss": 0.7815660834312439,
+      "step": 750
+    },
+    {
+      "epoch": 1.3772893772893773,
+      "grad_norm": 0.25882411003112793,
+      "learning_rate": 1.264653847563519e-05,
+      "loss": 1.120620608329773,
+      "step": 752
+    },
+    {
+      "epoch": 1.380952380952381,
+      "grad_norm": 0.09388554096221924,
+      "learning_rate": 1.2611516695420023e-05,
+      "loss": 0.9709092974662781,
+      "step": 754
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.1428227424621582,
+      "learning_rate": 1.2576469686614608e-05,
+      "loss": 1.3016420602798462,
+      "step": 756
+    },
+    {
+      "epoch": 1.3882783882783882,
+      "grad_norm": 0.22650009393692017,
+      "learning_rate": 1.2541397997886317e-05,
+      "loss": 1.2032549381256104,
+      "step": 758
+    },
+    {
+      "epoch": 1.3919413919413919,
+      "grad_norm": 2.37180233001709,
+      "learning_rate": 1.2506302178288887e-05,
+      "loss": 1.1462368965148926,
+      "step": 760
+    },
+    {
+      "epoch": 1.3956043956043955,
+      "grad_norm": 0.25279995799064636,
+      "learning_rate": 1.2471182777253832e-05,
+      "loss": 1.1458882093429565,
+      "step": 762
+    },
+    {
+      "epoch": 1.3992673992673992,
+      "grad_norm": 1.0492079257965088,
+      "learning_rate": 1.2436040344581824e-05,
+      "loss": 0.6942178606987,
+      "step": 764
+    },
+    {
+      "epoch": 1.4029304029304028,
+      "grad_norm": 0.27227652072906494,
+      "learning_rate": 1.2400875430434119e-05,
+      "loss": 0.8875712752342224,
+      "step": 766
+    },
+    {
+      "epoch": 1.4065934065934065,
+      "grad_norm": 0.7895340919494629,
+      "learning_rate": 1.236568858532391e-05,
+      "loss": 0.8964008688926697,
+      "step": 768
+    },
+    {
+      "epoch": 1.4102564102564101,
+      "grad_norm": 0.17773286998271942,
+      "learning_rate": 1.2330480360107728e-05,
+      "loss": 1.1805744171142578,
+      "step": 770
+    },
+    {
+      "epoch": 1.4139194139194138,
+      "grad_norm": 0.23372013866901398,
+      "learning_rate": 1.2295251305976818e-05,
+      "loss": 1.2107068300247192,
+      "step": 772
+    },
+    {
+      "epoch": 1.4175824175824177,
+      "grad_norm": 1.012639045715332,
+      "learning_rate": 1.2260001974448504e-05,
+      "loss": 1.019040584564209,
+      "step": 774
+    },
+    {
+      "epoch": 1.4212454212454213,
+      "grad_norm": 0.23378640413284302,
+      "learning_rate": 1.222473291735754e-05,
+      "loss": 1.1538594961166382,
+      "step": 776
+    },
+    {
+      "epoch": 1.424908424908425,
+      "grad_norm": 0.30686837434768677,
+      "learning_rate": 1.218944468684752e-05,
+      "loss": 0.8257187604904175,
+      "step": 778
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.1656091809272766,
+      "learning_rate": 1.215413783536217e-05,
+      "loss": 1.3644243478775024,
+      "step": 780
+    },
+    {
+      "epoch": 1.4322344322344323,
+      "grad_norm": 0.3846034109592438,
+      "learning_rate": 1.2118812915636744e-05,
+      "loss": 1.2287310361862183,
+      "step": 782
+    },
+    {
+      "epoch": 1.435897435897436,
+      "grad_norm": 0.1750563234090805,
+      "learning_rate": 1.2083470480689363e-05,
+      "loss": 1.1567542552947998,
+      "step": 784
+    },
+    {
+      "epoch": 1.4395604395604396,
+      "grad_norm": 3.485445499420166,
+      "learning_rate": 1.2048111083812342e-05,
+      "loss": 0.9774308800697327,
+      "step": 786
+    },
+    {
+      "epoch": 1.4432234432234432,
+      "grad_norm": 0.34514904022216797,
+      "learning_rate": 1.2012735278563546e-05,
+      "loss": 1.1295884847640991,
+      "step": 788
+    },
+    {
+      "epoch": 1.4468864468864469,
+      "grad_norm": 0.12276670336723328,
+      "learning_rate": 1.1977343618757702e-05,
+      "loss": 0.7207637429237366,
+      "step": 790
+    },
+    {
+      "epoch": 1.4505494505494505,
+      "grad_norm": 0.1892281472682953,
+      "learning_rate": 1.1941936658457769e-05,
+      "loss": 0.9219919443130493,
+      "step": 792
+    },
+    {
+      "epoch": 1.4542124542124542,
+      "grad_norm": 0.21250323951244354,
+      "learning_rate": 1.1906514951966208e-05,
+      "loss": 0.8157789707183838,
+      "step": 794
+    },
+    {
+      "epoch": 1.4578754578754578,
+      "grad_norm": 0.16933120787143707,
+      "learning_rate": 1.1871079053816357e-05,
+      "loss": 1.1426329612731934,
+      "step": 796
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.4070507884025574,
+      "learning_rate": 1.1835629518763714e-05,
+      "loss": 0.7938690781593323,
+      "step": 798
+    },
+    {
+      "epoch": 1.4652014652014653,
+      "grad_norm": 0.47800537943840027,
+      "learning_rate": 1.1800166901777272e-05,
+      "loss": 1.024507999420166,
+      "step": 800
+    },
+    {
+      "epoch": 1.468864468864469,
+      "grad_norm": 0.3107675313949585,
+      "learning_rate": 1.1764691758030825e-05,
+      "loss": 1.5597076416015625,
+      "step": 802
+    },
+    {
+      "epoch": 1.4725274725274726,
+      "grad_norm": 1.4179226160049438,
+      "learning_rate": 1.1729204642894265e-05,
+      "loss": 1.0233888626098633,
+      "step": 804
+    },
+    {
+      "epoch": 1.4761904761904763,
+      "grad_norm": 0.335848331451416,
+      "learning_rate": 1.1693706111924912e-05,
+      "loss": 1.1873747110366821,
+      "step": 806
+    },
+    {
+      "epoch": 1.47985347985348,
+      "grad_norm": 0.2710278630256653,
+      "learning_rate": 1.1658196720858794e-05,
+      "loss": 1.1727930307388306,
+      "step": 808
+    },
+    {
+      "epoch": 1.4835164835164836,
+      "grad_norm": 0.37410542368888855,
+      "learning_rate": 1.1622677025601966e-05,
+      "loss": 1.0092246532440186,
+      "step": 810
+    },
+    {
+      "epoch": 1.4871794871794872,
+      "grad_norm": 0.20915468037128448,
+      "learning_rate": 1.1587147582221776e-05,
+      "loss": 0.8401330709457397,
+      "step": 812
+    },
+    {
+      "epoch": 1.4908424908424909,
+      "grad_norm": 0.6524186134338379,
+      "learning_rate": 1.1551608946938208e-05,
+      "loss": 1.2012218236923218,
+      "step": 814
+    },
+    {
+      "epoch": 1.4945054945054945,
+      "grad_norm": 1.3064533472061157,
+      "learning_rate": 1.1516061676115124e-05,
+      "loss": 1.137012004852295,
+      "step": 816
+    },
+    {
+      "epoch": 1.4981684981684982,
+      "grad_norm": 0.08933006227016449,
+      "learning_rate": 1.1480506326251595e-05,
+      "loss": 0.4151400327682495,
+      "step": 818
+    },
+    {
+      "epoch": 1.5018315018315018,
+      "grad_norm": 0.3812463879585266,
+      "learning_rate": 1.1444943453973155e-05,
+      "loss": 1.1498603820800781,
+      "step": 820
+    },
+    {
+      "epoch": 1.5054945054945055,
+      "grad_norm": 0.16644573211669922,
+      "learning_rate": 1.1409373616023111e-05,
+      "loss": 0.9069132804870605,
+      "step": 822
+    },
+    {
+      "epoch": 1.5091575091575091,
+      "grad_norm": 0.9926286935806274,
+      "learning_rate": 1.1373797369253818e-05,
+      "loss": 0.828985869884491,
+      "step": 824
+    },
+    {
+      "epoch": 1.5128205128205128,
+      "grad_norm": 0.27975189685821533,
+      "learning_rate": 1.1338215270617967e-05,
+      "loss": 1.0592471361160278,
+      "step": 826
+    },
+    {
+      "epoch": 1.5164835164835164,
+      "grad_norm": 1.065266728401184,
+      "learning_rate": 1.130262787715985e-05,
+      "loss": 1.0117770433425903,
+      "step": 828
+    },
+    {
+      "epoch": 1.52014652014652,
+      "grad_norm": 0.755531907081604,
+      "learning_rate": 1.1267035746006658e-05,
+      "loss": 1.012010931968689,
+      "step": 830
+    },
+    {
+      "epoch": 1.5238095238095237,
+      "grad_norm": 0.2652962803840637,
+      "learning_rate": 1.1231439434359755e-05,
+      "loss": 1.2559067010879517,
+      "step": 832
+    },
+    {
+      "epoch": 1.5274725274725274,
+      "grad_norm": 1.055659294128418,
+      "learning_rate": 1.119583949948594e-05,
+      "loss": 0.8118237853050232,
+      "step": 834
+    },
+    {
+      "epoch": 1.531135531135531,
+      "grad_norm": 0.16545364260673523,
+      "learning_rate": 1.1160236498708742e-05,
+      "loss": 1.156209111213684,
+      "step": 836
+    },
+    {
+      "epoch": 1.5347985347985347,
+      "grad_norm": 0.29566454887390137,
+      "learning_rate": 1.112463098939969e-05,
+      "loss": 1.207381248474121,
+      "step": 838
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.3882679343223572,
+      "learning_rate": 1.1089023528969576e-05,
+      "loss": 1.0491501092910767,
+      "step": 840
+    },
+    {
+      "epoch": 1.542124542124542,
+      "grad_norm": 0.14187085628509521,
+      "learning_rate": 1.1053414674859741e-05,
+      "loss": 0.872413694858551,
+      "step": 842
+    },
+    {
+      "epoch": 1.5457875457875456,
+      "grad_norm": 0.3208055794239044,
+      "learning_rate": 1.1017804984533351e-05,
+      "loss": 1.038373589515686,
+      "step": 844
+    },
+    {
+      "epoch": 1.5494505494505495,
+      "grad_norm": 0.274668425321579,
+      "learning_rate": 1.0982195015466652e-05,
+      "loss": 1.182910680770874,
+      "step": 846
+    },
+    {
+      "epoch": 1.5531135531135531,
+      "grad_norm": 1.7933740615844727,
+      "learning_rate": 1.0946585325140261e-05,
+      "loss": 0.5585577487945557,
+      "step": 848
+    },
+    {
+      "epoch": 1.5567765567765568,
+      "grad_norm": 0.17488670349121094,
+      "learning_rate": 1.0910976471030428e-05,
+      "loss": 1.0341525077819824,
+      "step": 850
+    },
+    {
+      "epoch": 1.5604395604395604,
+      "grad_norm": 0.38739433884620667,
+      "learning_rate": 1.0875369010600317e-05,
+      "loss": 1.1761656999588013,
+      "step": 852
+    },
+    {
+      "epoch": 1.564102564102564,
+      "grad_norm": 0.4399110674858093,
+      "learning_rate": 1.083976350129126e-05,
+      "loss": 0.8638155460357666,
+      "step": 854
+    },
+    {
+      "epoch": 1.5677655677655677,
+      "grad_norm": 0.34226399660110474,
+      "learning_rate": 1.0804160500514062e-05,
+      "loss": 0.8107349872589111,
+      "step": 856
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 1.9198914766311646,
+      "learning_rate": 1.0768560565640252e-05,
+      "loss": 0.9437478184700012,
+      "step": 858
+    },
+    {
+      "epoch": 1.575091575091575,
+      "grad_norm": 0.15224924683570862,
+      "learning_rate": 1.0732964253993343e-05,
+      "loss": 0.7484935522079468,
+      "step": 860
+    },
+    {
+      "epoch": 1.578754578754579,
+      "grad_norm": 0.26914143562316895,
+      "learning_rate": 1.0697372122840156e-05,
+      "loss": 1.1590977907180786,
+      "step": 862
+    },
+    {
+      "epoch": 1.5824175824175826,
+      "grad_norm": 1.1250046491622925,
+      "learning_rate": 1.0661784729382036e-05,
+      "loss": 0.9178829789161682,
+      "step": 864
+    },
+    {
+      "epoch": 1.5860805860805862,
+      "grad_norm": 0.2830469310283661,
+      "learning_rate": 1.0626202630746183e-05,
+      "loss": 1.1874239444732666,
+      "step": 866
+    },
+    {
+      "epoch": 1.5897435897435899,
+      "grad_norm": 0.4739990830421448,
+      "learning_rate": 1.0590626383976894e-05,
+      "loss": 1.1853182315826416,
+      "step": 868
+    },
+    {
+      "epoch": 1.5934065934065935,
+      "grad_norm": 0.2765795886516571,
+      "learning_rate": 1.055505654602685e-05,
+      "loss": 0.5874127149581909,
+      "step": 870
+    },
+    {
+      "epoch": 1.5970695970695972,
+      "grad_norm": 0.18368150293827057,
+      "learning_rate": 1.0519493673748406e-05,
+      "loss": 1.1814969778060913,
+      "step": 872
+    },
+    {
+      "epoch": 1.6007326007326008,
+      "grad_norm": 0.1637224406003952,
+      "learning_rate": 1.0483938323884879e-05,
+      "loss": 1.0709137916564941,
+      "step": 874
+    },
+    {
+      "epoch": 1.6043956043956045,
+      "grad_norm": 1.2013723850250244,
+      "learning_rate": 1.0448391053061795e-05,
+      "loss": 0.7793064117431641,
+      "step": 876
+    },
+    {
+      "epoch": 1.6080586080586081,
+      "grad_norm": 0.38753247261047363,
+      "learning_rate": 1.0412852417778225e-05,
+      "loss": 0.9944717884063721,
+      "step": 878
+    },
+    {
+      "epoch": 1.6117216117216118,
+      "grad_norm": 0.24931028485298157,
+      "learning_rate": 1.037732297439804e-05,
+      "loss": 0.9063097834587097,
+      "step": 880
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.24992220103740692,
+      "learning_rate": 1.034180327914121e-05,
+      "loss": 1.1387708187103271,
+      "step": 882
+    },
+    {
+      "epoch": 1.619047619047619,
+      "grad_norm": 1.5768555402755737,
+      "learning_rate": 1.030629388807509e-05,
+      "loss": 0.9485580921173096,
+      "step": 884
+    },
+    {
+      "epoch": 1.6227106227106227,
+      "grad_norm": 0.31033453345298767,
+      "learning_rate": 1.0270795357105738e-05,
+      "loss": 1.145193099975586,
+      "step": 886
+    },
+    {
+      "epoch": 1.6263736263736264,
+      "grad_norm": 1.4770396947860718,
+      "learning_rate": 1.023530824196918e-05,
+      "loss": 0.8466076254844666,
+      "step": 888
+    },
+    {
+      "epoch": 1.63003663003663,
+      "grad_norm": 0.24302436411380768,
+      "learning_rate": 1.019983309822273e-05,
+      "loss": 1.2269943952560425,
+      "step": 890
+    },
+    {
+      "epoch": 1.6336996336996337,
+      "grad_norm": 2.4740350246429443,
+      "learning_rate": 1.0164370481236292e-05,
+      "loss": 0.9066356420516968,
+      "step": 892
+    },
+    {
+      "epoch": 1.6373626373626373,
+      "grad_norm": 0.26794663071632385,
+      "learning_rate": 1.0128920946183646e-05,
+      "loss": 1.1919015645980835,
+      "step": 894
+    },
+    {
+      "epoch": 1.641025641025641,
+      "grad_norm": 0.5033538937568665,
+      "learning_rate": 1.0093485048033798e-05,
+      "loss": 0.7559452652931213,
+      "step": 896
+    },
+    {
+      "epoch": 1.6446886446886446,
+      "grad_norm": 0.4728987514972687,
+      "learning_rate": 1.0058063341542238e-05,
+      "loss": 1.141265869140625,
+      "step": 898
+    },
+    {
+      "epoch": 1.6483516483516483,
+      "grad_norm": 0.0506194643676281,
+      "learning_rate": 1.0022656381242297e-05,
+      "loss": 0.835241973400116,
+      "step": 900
+    },
+    {
+      "epoch": 1.652014652014652,
+      "grad_norm": 0.20770415663719177,
+      "learning_rate": 9.98726472143646e-06,
+      "loss": 0.8866770267486572,
+      "step": 902
+    },
+    {
+      "epoch": 1.6556776556776556,
+      "grad_norm": 0.3993974030017853,
+      "learning_rate": 9.951888916187662e-06,
+      "loss": 1.027202844619751,
+      "step": 904
+    },
+    {
+      "epoch": 1.6593406593406592,
+      "grad_norm": 0.19648966193199158,
+      "learning_rate": 9.916529519310638e-06,
+      "loss": 1.1409128904342651,
+      "step": 906
+    },
+    {
+      "epoch": 1.6630036630036629,
+      "grad_norm": 0.18085846304893494,
+      "learning_rate": 9.881187084363257e-06,
+      "loss": 0.7516112923622131,
+      "step": 908
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.21360376477241516,
+      "learning_rate": 9.845862164637834e-06,
+      "loss": 1.2104125022888184,
+      "step": 910
+    },
+    {
+      "epoch": 1.6703296703296702,
+      "grad_norm": 0.18259240686893463,
+      "learning_rate": 9.810555313152486e-06,
+      "loss": 1.2229918241500854,
+      "step": 912
+    },
+    {
+      "epoch": 1.673992673992674,
+      "grad_norm": 0.20374548435211182,
+      "learning_rate": 9.775267082642461e-06,
+      "loss": 1.1026571989059448,
+      "step": 914
+    },
+    {
+      "epoch": 1.6776556776556777,
+      "grad_norm": 0.21295559406280518,
+      "learning_rate": 9.7399980255515e-06,
+      "loss": 1.2173269987106323,
+      "step": 916
+    },
+    {
+      "epoch": 1.6813186813186813,
+      "grad_norm": 0.2531648278236389,
+      "learning_rate": 9.704748694023183e-06,
+      "loss": 0.8437496423721313,
+      "step": 918
+    },
+    {
+      "epoch": 1.684981684981685,
+      "grad_norm": 0.16702847182750702,
+      "learning_rate": 9.669519639892275e-06,
+      "loss": 1.2237019538879395,
+      "step": 920
+    },
+    {
+      "epoch": 1.6886446886446886,
+      "grad_norm": 0.2449825406074524,
+      "learning_rate": 9.634311414676096e-06,
+      "loss": 1.0298762321472168,
+      "step": 922
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.43799275159835815,
+      "learning_rate": 9.599124569565887e-06,
+      "loss": 0.8457880616188049,
+      "step": 924
+    },
+    {
+      "epoch": 1.695970695970696,
+      "grad_norm": 0.462398886680603,
+      "learning_rate": 9.56395965541818e-06,
+      "loss": 0.9856649041175842,
+      "step": 926
+    },
+    {
+      "epoch": 1.6996336996336996,
+      "grad_norm": 0.22268880903720856,
+      "learning_rate": 9.528817222746171e-06,
+      "loss": 0.8122522234916687,
+      "step": 928
+    },
+    {
+      "epoch": 1.7032967032967035,
+      "grad_norm": 0.12412460893392563,
+      "learning_rate": 9.493697821711116e-06,
+      "loss": 0.9117051362991333,
+      "step": 930
+    },
+    {
+      "epoch": 1.7069597069597071,
+      "grad_norm": 0.3573848605155945,
+      "learning_rate": 9.458602002113684e-06,
+      "loss": 0.9280415773391724,
+      "step": 932
+    },
+    {
+      "epoch": 1.7106227106227108,
+      "grad_norm": 0.19333380460739136,
+      "learning_rate": 9.423530313385395e-06,
+      "loss": 1.3979382514953613,
+      "step": 934
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.28463131189346313,
+      "learning_rate": 9.388483304579983e-06,
+      "loss": 1.1958733797073364,
+      "step": 936
+    },
+    {
+      "epoch": 1.717948717948718,
+      "grad_norm": 0.5562649965286255,
+      "learning_rate": 9.353461524364814e-06,
+      "loss": 0.48058995604515076,
+      "step": 938
+    },
+    {
+      "epoch": 1.7216117216117217,
+      "grad_norm": 0.2503082752227783,
+      "learning_rate": 9.318465521012298e-06,
+      "loss": 0.556159257888794,
+      "step": 940
+    },
+    {
+      "epoch": 1.7252747252747254,
+      "grad_norm": 0.28722119331359863,
+      "learning_rate": 9.283495842391313e-06,
+      "loss": 1.130286693572998,
+      "step": 942
+    },
+    {
+      "epoch": 1.728937728937729,
+      "grad_norm": 1.835166573524475,
+      "learning_rate": 9.248553035958596e-06,
+      "loss": 0.9355916380882263,
+      "step": 944
+    },
+    {
+      "epoch": 1.7326007326007327,
+      "grad_norm": 0.5465191006660461,
+      "learning_rate": 9.213637648750217e-06,
+      "loss": 1.1549943685531616,
+      "step": 946
+    },
+    {
+      "epoch": 1.7362637362637363,
+      "grad_norm": 0.2970430850982666,
+      "learning_rate": 9.178750227372983e-06,
+      "loss": 1.100421667098999,
+      "step": 948
+    },
+    {
+      "epoch": 1.73992673992674,
+      "grad_norm": 0.36570602655410767,
+      "learning_rate": 9.143891317995888e-06,
+      "loss": 0.9944745302200317,
+      "step": 950
+    },
+    {
+      "epoch": 1.7435897435897436,
+      "grad_norm": 0.16801096498966217,
+      "learning_rate": 9.109061466341576e-06,
+      "loss": 0.9477764368057251,
+      "step": 952
+    },
+    {
+      "epoch": 1.7472527472527473,
+      "grad_norm": 0.3478679358959198,
+      "learning_rate": 9.074261217677771e-06,
+      "loss": 1.2121840715408325,
+      "step": 954
+    },
+    {
+      "epoch": 1.750915750915751,
+      "grad_norm": 0.25078827142715454,
+      "learning_rate": 9.039491116808773e-06,
+      "loss": 0.7902039885520935,
+      "step": 956
+    },
+    {
+      "epoch": 1.7545787545787546,
+      "grad_norm": 0.48620137572288513,
+      "learning_rate": 9.004751708066906e-06,
+      "loss": 1.1979793310165405,
+      "step": 958
+    },
+    {
+      "epoch": 1.7582417582417582,
+      "grad_norm": 0.12998303771018982,
+      "learning_rate": 8.970043535303999e-06,
+      "loss": 0.5474309325218201,
+      "step": 960
+    },
+    {
+      "epoch": 1.7619047619047619,
+      "grad_norm": 0.28665241599082947,
+      "learning_rate": 8.93536714188288e-06,
+      "loss": 0.9992084503173828,
+      "step": 962
+    },
+    {
+      "epoch": 1.7655677655677655,
+      "grad_norm": 0.17530608177185059,
+      "learning_rate": 8.900723070668869e-06,
+      "loss": 1.0385854244232178,
+      "step": 964
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.5553234815597534,
+      "learning_rate": 8.86611186402127e-06,
+      "loss": 1.1807163953781128,
+      "step": 966
+    },
+    {
+      "epoch": 1.7728937728937728,
+      "grad_norm": 2.357882499694824,
+      "learning_rate": 8.831534063784891e-06,
+      "loss": 0.5750354528427124,
+      "step": 968
+    },
+    {
+      "epoch": 1.7765567765567765,
+      "grad_norm": 0.8607165813446045,
+      "learning_rate": 8.796990211281549e-06,
+      "loss": 0.8479418158531189,
+      "step": 970
+    },
+    {
+      "epoch": 1.7802197802197801,
+      "grad_norm": 0.3071545362472534,
+      "learning_rate": 8.76248084730161e-06,
+      "loss": 0.9116554856300354,
+      "step": 972
+    },
+    {
+      "epoch": 1.7838827838827838,
+      "grad_norm": 0.39947524666786194,
+      "learning_rate": 8.728006512095517e-06,
+      "loss": 1.2301732301712036,
+      "step": 974
+    },
+    {
+      "epoch": 1.7875457875457874,
+      "grad_norm": 0.37182819843292236,
+      "learning_rate": 8.693567745365325e-06,
+      "loss": 1.1915199756622314,
+      "step": 976
+    },
+    {
+      "epoch": 1.791208791208791,
+      "grad_norm": 0.155500128865242,
+      "learning_rate": 8.659165086256263e-06,
+      "loss": 0.9201015830039978,
+      "step": 978
+    },
+    {
+      "epoch": 1.7948717948717947,
+      "grad_norm": 0.21511544287204742,
+      "learning_rate": 8.624799073348282e-06,
+      "loss": 0.9540326595306396,
+      "step": 980
+    },
+    {
+      "epoch": 1.7985347985347986,
+      "grad_norm": 0.26243695616722107,
+      "learning_rate": 8.590470244647643e-06,
+      "loss": 1.1440948247909546,
+      "step": 982
+    },
+    {
+      "epoch": 1.8021978021978022,
+      "grad_norm": 1.7372164726257324,
+      "learning_rate": 8.556179137578461e-06,
+      "loss": 1.100319504737854,
+      "step": 984
+    },
+    {
+      "epoch": 1.8058608058608059,
+      "grad_norm": 0.1159660816192627,
+      "learning_rate": 8.521926288974336e-06,
+      "loss": 0.6495481133460999,
+      "step": 986
+    },
+    {
+      "epoch": 1.8095238095238095,
+      "grad_norm": 0.2123069316148758,
+      "learning_rate": 8.487712235069901e-06,
+      "loss": 0.8149735927581787,
+      "step": 988
+    },
+    {
+      "epoch": 1.8131868131868132,
+      "grad_norm": 0.42196395993232727,
+      "learning_rate": 8.453537511492469e-06,
+      "loss": 0.7469933032989502,
+      "step": 990
+    },
+    {
+      "epoch": 1.8168498168498168,
+      "grad_norm": 0.2649979293346405,
+      "learning_rate": 8.419402653253623e-06,
+      "loss": 0.7891843914985657,
+      "step": 992
+    },
+    {
+      "epoch": 1.8205128205128205,
+      "grad_norm": 0.21673156321048737,
+      "learning_rate": 8.385308194740846e-06,
+      "loss": 0.8571860790252686,
+      "step": 994
+    },
+    {
+      "epoch": 1.8241758241758241,
+      "grad_norm": 0.35649216175079346,
+      "learning_rate": 8.35125466970915e-06,
+      "loss": 1.1353570222854614,
+      "step": 996
+    },
+    {
+      "epoch": 1.8278388278388278,
+      "grad_norm": 0.2921370565891266,
+      "learning_rate": 8.317242611272745e-06,
+      "loss": 0.8522858023643494,
+      "step": 998
+    },
+    {
+      "epoch": 1.8315018315018317,
+      "grad_norm": 0.577377438545227,
+      "learning_rate": 8.283272551896649e-06,
+      "loss": 1.1177325248718262,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8351648351648353,
+      "grad_norm": 0.3225407898426056,
+      "learning_rate": 8.249345023388393e-06,
+      "loss": 1.145124912261963,
+      "step": 1002
+    },
+    {
+      "epoch": 1.838827838827839,
+      "grad_norm": 0.34550943970680237,
+      "learning_rate": 8.21546055688968e-06,
+      "loss": 1.1394211053848267,
+      "step": 1004
+    },
+    {
+      "epoch": 1.8424908424908426,
+      "grad_norm": 0.22849220037460327,
+      "learning_rate": 8.181619682868059e-06,
+      "loss": 1.1420109272003174,
+      "step": 1006
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.3721167743206024,
+      "learning_rate": 8.147822931108638e-06,
+      "loss": 0.7952710390090942,
+      "step": 1008
+    },
+    {
+      "epoch": 1.84981684981685,
+      "grad_norm": 0.17802205681800842,
+      "learning_rate": 8.114070830705785e-06,
+      "loss": 1.1106821298599243,
+      "step": 1010
+    },
+    {
+      "epoch": 1.8534798534798536,
+      "grad_norm": 0.15619701147079468,
+      "learning_rate": 8.080363910054833e-06,
+      "loss": 0.7631734609603882,
+      "step": 1012
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 0.3461393117904663,
+      "learning_rate": 8.04670269684383e-06,
+      "loss": 1.162695288658142,
+      "step": 1014
+    },
+    {
+      "epoch": 1.8608058608058609,
+      "grad_norm": 0.18843670189380646,
+      "learning_rate": 8.013087718045256e-06,
+      "loss": 1.1480703353881836,
+      "step": 1016
+    },
+    {
+      "epoch": 1.8644688644688645,
+      "grad_norm": 0.30292752385139465,
+      "learning_rate": 7.979519499907786e-06,
+      "loss": 1.2293591499328613,
+      "step": 1018
+    },
+    {
+      "epoch": 1.8681318681318682,
+      "grad_norm": 0.09791062027215958,
+      "learning_rate": 7.945998567948052e-06,
+      "loss": 0.9643331170082092,
+      "step": 1020
+    },
+    {
+      "epoch": 1.8717948717948718,
+      "grad_norm": 0.20030497014522552,
+      "learning_rate": 7.912525446942406e-06,
+      "loss": 1.1193994283676147,
+      "step": 1022
+    },
+    {
+      "epoch": 1.8754578754578755,
+      "grad_norm": 0.3839741051197052,
+      "learning_rate": 7.879100660918713e-06,
+      "loss": 0.5604696869850159,
+      "step": 1024
+    },
+    {
+      "epoch": 1.879120879120879,
+      "grad_norm": 0.15717224776744843,
+      "learning_rate": 7.845724733148149e-06,
+      "loss": 1.1571592092514038,
+      "step": 1026
+    },
+    {
+      "epoch": 1.8827838827838828,
+      "grad_norm": 0.27877768874168396,
+      "learning_rate": 7.812398186136994e-06,
+      "loss": 0.954494059085846,
+      "step": 1028
+    },
+    {
+      "epoch": 1.8864468864468864,
+      "grad_norm": 0.16563951969146729,
+      "learning_rate": 7.779121541618478e-06,
+      "loss": 1.153045892715454,
+      "step": 1030
+    },
+    {
+      "epoch": 1.89010989010989,
+      "grad_norm": 0.13534830510616302,
+      "learning_rate": 7.74589532054459e-06,
+      "loss": 0.9504430890083313,
+      "step": 1032
+    },
+    {
+      "epoch": 1.8937728937728937,
+      "grad_norm": 0.3428526222705841,
+      "learning_rate": 7.712720043077929e-06,
+      "loss": 0.8016744256019592,
+      "step": 1034
+    },
+    {
+      "epoch": 1.8974358974358974,
+      "grad_norm": 0.3059753477573395,
+      "learning_rate": 7.679596228583563e-06,
+      "loss": 1.1903191804885864,
+      "step": 1036
+    },
+    {
+      "epoch": 1.901098901098901,
+      "grad_norm": 0.17783567309379578,
+      "learning_rate": 7.646524395620908e-06,
+      "loss": 1.157071828842163,
+      "step": 1038
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 0.7491412162780762,
+      "learning_rate": 7.613505061935584e-06,
+      "loss": 1.2254270315170288,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9084249084249083,
+      "grad_norm": 0.44477730989456177,
+      "learning_rate": 7.580538744451336e-06,
+      "loss": 0.6230685710906982,
+      "step": 1042
+    },
+    {
+      "epoch": 1.912087912087912,
+      "grad_norm": 0.14885884523391724,
+      "learning_rate": 7.547625959261928e-06,
+      "loss": 0.8747984766960144,
+      "step": 1044
+    },
+    {
+      "epoch": 1.9157509157509156,
+      "grad_norm": 0.1716078519821167,
+      "learning_rate": 7.5147672216230605e-06,
+      "loss": 1.117228388786316,
+      "step": 1046
+    },
+    {
+      "epoch": 1.9194139194139193,
+      "grad_norm": 0.20717094838619232,
+      "learning_rate": 7.481963045944318e-06,
+      "loss": 0.45573827624320984,
+      "step": 1048
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.5965511798858643,
+      "learning_rate": 7.449213945781102e-06,
+      "loss": 0.8882296085357666,
+      "step": 1050
+    },
+    {
+      "epoch": 1.9267399267399268,
+      "grad_norm": 0.2176082581281662,
+      "learning_rate": 7.416520433826599e-06,
+      "loss": 0.8158991932868958,
+      "step": 1052
+    },
+    {
+      "epoch": 1.9304029304029304,
+      "grad_norm": 0.31369099020957947,
+      "learning_rate": 7.383883021903755e-06,
+      "loss": 1.1231168508529663,
+      "step": 1054
+    },
+    {
+      "epoch": 1.934065934065934,
+      "grad_norm": 0.2577856183052063,
+      "learning_rate": 7.351302220957251e-06,
+      "loss": 0.748049259185791,
+      "step": 1056
+    },
+    {
+      "epoch": 1.9377289377289377,
+      "grad_norm": 0.31049013137817383,
+      "learning_rate": 7.318778541045517e-06,
+      "loss": 0.9868760704994202,
+      "step": 1058
+    },
+    {
+      "epoch": 1.9413919413919414,
+      "grad_norm": 0.3126990795135498,
+      "learning_rate": 7.286312491332754e-06,
+      "loss": 1.0847361087799072,
+      "step": 1060
+    },
+    {
+      "epoch": 1.945054945054945,
+      "grad_norm": 0.235934779047966,
+      "learning_rate": 7.253904580080926e-06,
+      "loss": 0.84217369556427,
+      "step": 1062
+    },
+    {
+      "epoch": 1.9487179487179487,
+      "grad_norm": 0.2684265375137329,
+      "learning_rate": 7.221555314641853e-06,
+      "loss": 0.8080073595046997,
+      "step": 1064
+    },
+    {
+      "epoch": 1.9523809523809523,
+      "grad_norm": 0.28575268387794495,
+      "learning_rate": 7.18926520144924e-06,
+      "loss": 1.1642062664031982,
+      "step": 1066
+    },
+    {
+      "epoch": 1.9560439560439562,
+      "grad_norm": 0.25401219725608826,
+      "learning_rate": 7.1570347460107335e-06,
+      "loss": 1.1827412843704224,
+      "step": 1068
+    },
+    {
+      "epoch": 1.9597069597069599,
+      "grad_norm": 0.11555124074220657,
+      "learning_rate": 7.124864452900049e-06,
+      "loss": 0.706343412399292,
+      "step": 1070
+    },
+    {
+      "epoch": 1.9633699633699635,
+      "grad_norm": 0.2669273912906647,
+      "learning_rate": 7.0927548257490465e-06,
+      "loss": 0.8516585230827332,
+      "step": 1072
+    },
+    {
+      "epoch": 1.9670329670329672,
+      "grad_norm": 0.31186506152153015,
+      "learning_rate": 7.060706367239836e-06,
+      "loss": 1.1490978002548218,
+      "step": 1074
+    },
+    {
+      "epoch": 1.9706959706959708,
+      "grad_norm": 0.19497060775756836,
+      "learning_rate": 7.028719579096932e-06,
+      "loss": 1.1234122514724731,
+      "step": 1076
+    },
+    {
+      "epoch": 1.9743589743589745,
+      "grad_norm": 0.21999667584896088,
+      "learning_rate": 6.9967949620793854e-06,
+      "loss": 0.9970806241035461,
+      "step": 1078
+    },
+    {
+      "epoch": 1.978021978021978,
+      "grad_norm": 0.5335208177566528,
+      "learning_rate": 6.964933015972947e-06,
+      "loss": 0.9913358688354492,
+      "step": 1080
+    },
+    {
+      "epoch": 1.9816849816849818,
+      "grad_norm": 0.2547147572040558,
+      "learning_rate": 6.933134239582246e-06,
+      "loss": 1.1181188821792603,
+      "step": 1082
+    },
+    {
+      "epoch": 1.9853479853479854,
+      "grad_norm": 0.07284919172525406,
+      "learning_rate": 6.9013991307229745e-06,
+      "loss": 0.7213895916938782,
+      "step": 1084
+    },
+    {
+      "epoch": 1.989010989010989,
+      "grad_norm": 1.2812455892562866,
+      "learning_rate": 6.869728186214093e-06,
+      "loss": 0.9767944812774658,
+      "step": 1086
+    },
+    {
+      "epoch": 1.9926739926739927,
+      "grad_norm": 0.2086431384086609,
+      "learning_rate": 6.8381219018700675e-06,
+      "loss": 0.9789687395095825,
+      "step": 1088
+    },
+    {
+      "epoch": 1.9963369963369964,
+      "grad_norm": 0.11151088029146194,
+      "learning_rate": 6.806580772493088e-06,
+      "loss": 0.930722713470459,
+      "step": 1090
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6231138706207275,
+      "learning_rate": 6.775105291865343e-06,
+      "loss": 1.042896032333374,
+      "step": 1092
+    },
+    {
+      "epoch": 2.0036630036630036,
+      "grad_norm": 0.1892917901277542,
+      "learning_rate": 6.743695952741265e-06,
+      "loss": 1.0761141777038574,
+      "step": 1094
+    },
+    {
+      "epoch": 2.0073260073260073,
+      "grad_norm": 0.39348432421684265,
+      "learning_rate": 6.71235324683983e-06,
+      "loss": 1.1419814825057983,
+      "step": 1096
+    },
+    {
+      "epoch": 2.010989010989011,
+      "grad_norm": 0.16642175614833832,
+      "learning_rate": 6.681077664836872e-06,
+      "loss": 1.088594675064087,
+      "step": 1098
+    },
+    {
+      "epoch": 2.0146520146520146,
+      "grad_norm": 0.3197147846221924,
+      "learning_rate": 6.649869696357381e-06,
+      "loss": 1.1661202907562256,
+      "step": 1100
+    },
+    {
+      "epoch": 2.0183150183150182,
+      "grad_norm": 0.43613532185554504,
+      "learning_rate": 6.6187298299678295e-06,
+      "loss": 0.8467984795570374,
+      "step": 1102
+    },
+    {
+      "epoch": 2.021978021978022,
+      "grad_norm": 0.24174755811691284,
+      "learning_rate": 6.587658553168563e-06,
+      "loss": 1.142805576324463,
+      "step": 1104
+    },
+    {
+      "epoch": 2.0256410256410255,
+      "grad_norm": 0.1317332535982132,
+      "learning_rate": 6.556656352386135e-06,
+      "loss": 0.7679157853126526,
+      "step": 1106
+    },
+    {
+      "epoch": 2.029304029304029,
+      "grad_norm": 0.18263843655586243,
+      "learning_rate": 6.525723712965698e-06,
+      "loss": 1.1841180324554443,
+      "step": 1108
+    },
+    {
+      "epoch": 2.032967032967033,
+      "grad_norm": 0.3168501853942871,
+      "learning_rate": 6.494861119163412e-06,
+      "loss": 0.9058336019515991,
+      "step": 1110
+    },
+    {
+      "epoch": 2.0366300366300365,
+      "grad_norm": 0.2593517303466797,
+      "learning_rate": 6.464069054138853e-06,
+      "loss": 0.6124511957168579,
+      "step": 1112
+    },
+    {
+      "epoch": 2.04029304029304,
+      "grad_norm": 0.245356947183609,
+      "learning_rate": 6.433347999947468e-06,
+      "loss": 0.845076322555542,
+      "step": 1114
+    },
+    {
+      "epoch": 2.043956043956044,
+      "grad_norm": 0.3254038095474243,
+      "learning_rate": 6.402698437533012e-06,
+      "loss": 1.1547578573226929,
+      "step": 1116
+    },
+    {
+      "epoch": 2.0476190476190474,
+      "grad_norm": 0.5246717929840088,
+      "learning_rate": 6.372120846720018e-06,
+      "loss": 1.1199109554290771,
+      "step": 1118
+    },
+    {
+      "epoch": 2.051282051282051,
+      "grad_norm": 0.6534336805343628,
+      "learning_rate": 6.341615706206292e-06,
+      "loss": 0.8209899067878723,
+      "step": 1120
+    },
+    {
+      "epoch": 2.0549450549450547,
+      "grad_norm": 0.8080871105194092,
+      "learning_rate": 6.311183493555426e-06,
+      "loss": 1.3262028694152832,
+      "step": 1122
+    },
+    {
+      "epoch": 2.0586080586080584,
+      "grad_norm": 0.23124472796916962,
+      "learning_rate": 6.280824685189296e-06,
+      "loss": 1.1404337882995605,
+      "step": 1124
+    },
+    {
+      "epoch": 2.062271062271062,
+      "grad_norm": 0.2290889322757721,
+      "learning_rate": 6.25053975638064e-06,
+      "loss": 0.7613569498062134,
+      "step": 1126
+    },
+    {
+      "epoch": 2.065934065934066,
+      "grad_norm": 0.18532392382621765,
+      "learning_rate": 6.220329181245585e-06,
+      "loss": 1.1030632257461548,
+      "step": 1128
+    },
+    {
+      "epoch": 2.06959706959707,
+      "grad_norm": 0.3047994375228882,
+      "learning_rate": 6.1901934327362355e-06,
+      "loss": 1.098275899887085,
+      "step": 1130
+    },
+    {
+      "epoch": 2.0732600732600734,
+      "grad_norm": 0.24303042888641357,
+      "learning_rate": 6.16013298263328e-06,
+      "loss": 1.130845069885254,
+      "step": 1132
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4075912535190582,
+      "learning_rate": 6.130148301538601e-06,
+      "loss": 1.1122570037841797,
+      "step": 1134
+    },
+    {
+      "epoch": 2.0805860805860807,
+      "grad_norm": 0.635519802570343,
+      "learning_rate": 6.100239858867887e-06,
+      "loss": 0.7130240201950073,
+      "step": 1136
+    },
+    {
+      "epoch": 2.0842490842490844,
+      "grad_norm": 0.19734634459018707,
+      "learning_rate": 6.070408122843311e-06,
+      "loss": 1.1177456378936768,
+      "step": 1138
+    },
+    {
+      "epoch": 2.087912087912088,
+      "grad_norm": 0.28150826692581177,
+      "learning_rate": 6.040653560486183e-06,
+      "loss": 1.0220168828964233,
+      "step": 1140
+    },
+    {
+      "epoch": 2.0915750915750917,
+      "grad_norm": 0.17299877107143402,
+      "learning_rate": 6.010976637609653e-06,
+      "loss": 1.106982707977295,
+      "step": 1142
+    },
+    {
+      "epoch": 2.0952380952380953,
+      "grad_norm": 0.15779122710227966,
+      "learning_rate": 5.9813778188114125e-06,
+      "loss": 0.3823546767234802,
+      "step": 1144
+    },
+    {
+      "epoch": 2.098901098901099,
+      "grad_norm": 0.36108678579330444,
+      "learning_rate": 5.951857567466401e-06,
+      "loss": 0.9157997369766235,
+      "step": 1146
+    },
+    {
+      "epoch": 2.1025641025641026,
+      "grad_norm": 0.177387997508049,
+      "learning_rate": 5.922416345719588e-06,
+      "loss": 0.8090324997901917,
+      "step": 1148
+    },
+    {
+      "epoch": 2.1062271062271063,
+      "grad_norm": 0.3467074930667877,
+      "learning_rate": 5.893054614478718e-06,
+      "loss": 0.8260840773582458,
+      "step": 1150
+    },
+    {
+      "epoch": 2.10989010989011,
+      "grad_norm": 0.2603940963745117,
+      "learning_rate": 5.8637728334070905e-06,
+      "loss": 0.9580783843994141,
+      "step": 1152
+    },
+    {
+      "epoch": 2.1135531135531136,
+      "grad_norm": 0.3268923759460449,
+      "learning_rate": 5.834571460916371e-06,
+      "loss": 0.7878354787826538,
+      "step": 1154
+    },
+    {
+      "epoch": 2.1172161172161172,
+      "grad_norm": 0.4938827455043793,
+      "learning_rate": 5.805450954159422e-06,
+      "loss": 1.0745412111282349,
+      "step": 1156
+    },
+    {
+      "epoch": 2.120879120879121,
+      "grad_norm": 0.2504938244819641,
+      "learning_rate": 5.776411769023127e-06,
+      "loss": 1.0261658430099487,
+      "step": 1158
+    },
+    {
+      "epoch": 2.1245421245421245,
+      "grad_norm": 0.24915434420108795,
+      "learning_rate": 5.747454360121274e-06,
+      "loss": 0.870224118232727,
+      "step": 1160
+    },
+    {
+      "epoch": 2.128205128205128,
+      "grad_norm": 0.4833395183086395,
+      "learning_rate": 5.718579180787425e-06,
+      "loss": 0.7795557379722595,
+      "step": 1162
+    },
+    {
+      "epoch": 2.131868131868132,
+      "grad_norm": 0.9497178196907043,
+      "learning_rate": 5.689786683067817e-06,
+      "loss": 0.918286144733429,
+      "step": 1164
+    },
+    {
+      "epoch": 2.1355311355311355,
+      "grad_norm": 0.6461044549942017,
+      "learning_rate": 5.661077317714303e-06,
+      "loss": 0.42682868242263794,
+      "step": 1166
+    },
+    {
+      "epoch": 2.139194139194139,
+      "grad_norm": 0.6395738124847412,
+      "learning_rate": 5.632451534177276e-06,
+      "loss": 0.4123497009277344,
+      "step": 1168
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 0.4208049178123474,
+      "learning_rate": 5.603909780598644e-06,
+      "loss": 0.8927979469299316,
+      "step": 1170
+    },
+    {
+      "epoch": 2.1465201465201464,
+      "grad_norm": 0.2959461808204651,
+      "learning_rate": 5.575452503804805e-06,
+      "loss": 1.1349587440490723,
+      "step": 1172
+    },
+    {
+      "epoch": 2.15018315018315,
+      "grad_norm": 0.8988657593727112,
+      "learning_rate": 5.5470801492996605e-06,
+      "loss": 1.2815641164779663,
+      "step": 1174
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.27057546377182007,
+      "learning_rate": 5.518793161257641e-06,
+      "loss": 0.7825716137886047,
+      "step": 1176
+    },
+    {
+      "epoch": 2.1575091575091574,
+      "grad_norm": 0.17563986778259277,
+      "learning_rate": 5.490591982516749e-06,
+      "loss": 1.142047643661499,
+      "step": 1178
+    },
+    {
+      "epoch": 2.161172161172161,
+      "grad_norm": 0.09482431411743164,
+      "learning_rate": 5.462477054571617e-06,
+      "loss": 1.127107858657837,
+      "step": 1180
+    },
+    {
+      "epoch": 2.1648351648351647,
+      "grad_norm": 0.17411017417907715,
+      "learning_rate": 5.4344488175666154e-06,
+      "loss": 1.1234813928604126,
+      "step": 1182
+    },
+    {
+      "epoch": 2.1684981684981683,
+      "grad_norm": 0.13109169900417328,
+      "learning_rate": 5.406507710288955e-06,
+      "loss": 1.1429002285003662,
+      "step": 1184
+    },
+    {
+      "epoch": 2.172161172161172,
+      "grad_norm": 0.16202552616596222,
+      "learning_rate": 5.378654170161805e-06,
+      "loss": 0.318652480840683,
+      "step": 1186
+    },
+    {
+      "epoch": 2.1758241758241756,
+      "grad_norm": 0.44094666838645935,
+      "learning_rate": 5.3508886332374534e-06,
+      "loss": 1.2102477550506592,
+      "step": 1188
+    },
+    {
+      "epoch": 2.1794871794871793,
+      "grad_norm": 0.8763576745986938,
+      "learning_rate": 5.323211534190496e-06,
+      "loss": 0.7540980577468872,
+      "step": 1190
+    },
+    {
+      "epoch": 2.183150183150183,
+      "grad_norm": 0.1877157837152481,
+      "learning_rate": 5.295623306310999e-06,
+      "loss": 0.9759551286697388,
+      "step": 1192
+    },
+    {
+      "epoch": 2.186813186813187,
+      "grad_norm": 0.120526023209095,
+      "learning_rate": 5.268124381497755e-06,
+      "loss": 0.802385687828064,
+      "step": 1194
+    },
+    {
+      "epoch": 2.1904761904761907,
+      "grad_norm": 0.27409690618515015,
+      "learning_rate": 5.240715190251484e-06,
+      "loss": 0.9069569110870361,
+      "step": 1196
+    },
+    {
+      "epoch": 2.1941391941391943,
+      "grad_norm": 0.20103105902671814,
+      "learning_rate": 5.213396161668111e-06,
+      "loss": 1.1158097982406616,
+      "step": 1198
+    },
+    {
+      "epoch": 2.197802197802198,
+      "grad_norm": 0.3620758652687073,
+      "learning_rate": 5.186167723432061e-06,
+      "loss": 0.7542502284049988,
+      "step": 1200
+    },
+    {
+      "epoch": 2.2014652014652016,
+      "grad_norm": 0.3984181582927704,
+      "learning_rate": 5.159030301809534e-06,
+      "loss": 1.1818705797195435,
+      "step": 1202
+    },
+    {
+      "epoch": 2.2051282051282053,
+      "grad_norm": 0.17239412665367126,
+      "learning_rate": 5.131984321641865e-06,
+      "loss": 1.1314865350723267,
+      "step": 1204
+    },
+    {
+      "epoch": 2.208791208791209,
+      "grad_norm": 0.23334990441799164,
+      "learning_rate": 5.105030206338843e-06,
+      "loss": 0.4661150276660919,
+      "step": 1206
+    },
+    {
+      "epoch": 2.2124542124542126,
+      "grad_norm": 3.1555933952331543,
+      "learning_rate": 5.0781683778720965e-06,
+      "loss": 0.7386508584022522,
+      "step": 1208
+    },
+    {
+      "epoch": 2.2161172161172162,
+      "grad_norm": 0.14592549204826355,
+      "learning_rate": 5.051399256768498e-06,
+      "loss": 0.6242368817329407,
+      "step": 1210
+    },
+    {
+      "epoch": 2.21978021978022,
+      "grad_norm": 0.23044905066490173,
+      "learning_rate": 5.024723262103559e-06,
+      "loss": 1.0724856853485107,
+      "step": 1212
+    },
+    {
+      "epoch": 2.2234432234432235,
+      "grad_norm": 0.2958521842956543,
+      "learning_rate": 4.998140811494881e-06,
+      "loss": 0.8155311942100525,
+      "step": 1214
+    },
+    {
+      "epoch": 2.227106227106227,
+      "grad_norm": 0.3939032256603241,
+      "learning_rate": 4.971652321095614e-06,
+      "loss": 0.8991819024085999,
+      "step": 1216
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.3258177638053894,
+      "learning_rate": 4.945258205587955e-06,
+      "loss": 1.1150569915771484,
+      "step": 1218
+    },
+    {
+      "epoch": 2.2344322344322345,
+      "grad_norm": 0.2855719029903412,
+      "learning_rate": 4.918958878176628e-06,
+      "loss": 1.1184298992156982,
+      "step": 1220
+    },
+    {
+      "epoch": 2.238095238095238,
+      "grad_norm": 0.2457854449748993,
+      "learning_rate": 4.8927547505824465e-06,
+      "loss": 1.1249486207962036,
+      "step": 1222
+    },
+    {
+      "epoch": 2.241758241758242,
+      "grad_norm": 0.31648901104927063,
+      "learning_rate": 4.866646233035845e-06,
+      "loss": 0.49614107608795166,
+      "step": 1224
+    },
+    {
+      "epoch": 2.2454212454212454,
+      "grad_norm": 0.19430601596832275,
+      "learning_rate": 4.840633734270464e-06,
+      "loss": 1.0849982500076294,
+      "step": 1226
+    },
+    {
+      "epoch": 2.249084249084249,
+      "grad_norm": 0.7305367588996887,
+      "learning_rate": 4.814717661516762e-06,
+      "loss": 0.6019390821456909,
+      "step": 1228
+    },
+    {
+      "epoch": 2.2527472527472527,
+      "grad_norm": 0.03867306187748909,
+      "learning_rate": 4.788898420495622e-06,
+      "loss": 0.5977374911308289,
+      "step": 1230
+    },
+    {
+      "epoch": 2.2564102564102564,
+      "grad_norm": 0.1346009522676468,
+      "learning_rate": 4.763176415412006e-06,
+      "loss": 0.6154077649116516,
+      "step": 1232
+    },
+    {
+      "epoch": 2.26007326007326,
+      "grad_norm": 0.7410935759544373,
+      "learning_rate": 4.7375520489486395e-06,
+      "loss": 1.1026445627212524,
+      "step": 1234
+    },
+    {
+      "epoch": 2.2637362637362637,
+      "grad_norm": 0.2963329553604126,
+      "learning_rate": 4.71202572225969e-06,
+      "loss": 0.7703139185905457,
+      "step": 1236
+    },
+    {
+      "epoch": 2.2673992673992673,
+      "grad_norm": 0.18143776059150696,
+      "learning_rate": 4.686597834964499e-06,
+      "loss": 0.9660443067550659,
+      "step": 1238
+    },
+    {
+      "epoch": 2.271062271062271,
+      "grad_norm": 0.14209656417369843,
+      "learning_rate": 4.661268785141316e-06,
+      "loss": 0.8756354451179504,
+      "step": 1240
+    },
+    {
+      "epoch": 2.2747252747252746,
+      "grad_norm": 0.33028721809387207,
+      "learning_rate": 4.636038969321073e-06,
+      "loss": 0.9074218273162842,
+      "step": 1242
+    },
+    {
+      "epoch": 2.2783882783882783,
+      "grad_norm": 0.263667494058609,
+      "learning_rate": 4.610908782481179e-06,
+      "loss": 1.1067166328430176,
+      "step": 1244
+    },
+    {
+      "epoch": 2.282051282051282,
+      "grad_norm": 0.07068176567554474,
+      "learning_rate": 4.5858786180393326e-06,
+      "loss": 0.730255126953125,
+      "step": 1246
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.4124971628189087,
+      "learning_rate": 4.560948867847359e-06,
+      "loss": 0.714017391204834,
+      "step": 1248
+    },
+    {
+      "epoch": 2.2893772893772892,
+      "grad_norm": 0.3742057681083679,
+      "learning_rate": 4.536119922185082e-06,
+      "loss": 1.1044319868087769,
+      "step": 1250
+    },
+    {
+      "epoch": 2.293040293040293,
+      "grad_norm": 0.6701974272727966,
+      "learning_rate": 4.511392169754214e-06,
+      "loss": 0.839243471622467,
+      "step": 1252
+    },
+    {
+      "epoch": 2.2967032967032965,
+      "grad_norm": 0.33513882756233215,
+      "learning_rate": 4.486765997672263e-06,
+      "loss": 1.1593670845031738,
+      "step": 1254
+    },
+    {
+      "epoch": 2.3003663003663,
+      "grad_norm": 0.20222492516040802,
+      "learning_rate": 4.46224179146649e-06,
+      "loss": 1.1839113235473633,
+      "step": 1256
+    },
+    {
+      "epoch": 2.304029304029304,
+      "grad_norm": 0.21135662496089935,
+      "learning_rate": 4.437819935067847e-06,
+      "loss": 0.7787933945655823,
+      "step": 1258
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.8632537126541138,
+      "learning_rate": 4.413500810804986e-06,
+      "loss": 1.1648868322372437,
+      "step": 1260
+    },
+    {
+      "epoch": 2.311355311355311,
+      "grad_norm": 0.31939440965652466,
+      "learning_rate": 4.389284799398276e-06,
+      "loss": 1.0294675827026367,
+      "step": 1262
+    },
+    {
+      "epoch": 2.315018315018315,
+      "grad_norm": 0.2278747856616974,
+      "learning_rate": 4.365172279953825e-06,
+      "loss": 0.8878557085990906,
+      "step": 1264
+    },
+    {
+      "epoch": 2.3186813186813184,
+      "grad_norm": 0.6909482479095459,
+      "learning_rate": 4.34116362995756e-06,
+      "loss": 0.7896403670310974,
+      "step": 1266
+    },
+    {
+      "epoch": 2.3223443223443225,
+      "grad_norm": 0.32261016964912415,
+      "learning_rate": 4.317259225269313e-06,
+      "loss": 0.8231415748596191,
+      "step": 1268
+    },
+    {
+      "epoch": 2.326007326007326,
+      "grad_norm": 0.6928750872612,
+      "learning_rate": 4.293459440116935e-06,
+      "loss": 1.1856201887130737,
+      "step": 1270
+    },
+    {
+      "epoch": 2.32967032967033,
+      "grad_norm": 0.5950198769569397,
+      "learning_rate": 4.269764647090442e-06,
+      "loss": 1.0062392950057983,
+      "step": 1272
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.4709881544113159,
+      "learning_rate": 4.246175217136176e-06,
+      "loss": 0.6258052587509155,
+      "step": 1274
+    },
+    {
+      "epoch": 2.336996336996337,
+      "grad_norm": 1.007083773612976,
+      "learning_rate": 4.2226915195509954e-06,
+      "loss": 1.0182530879974365,
+      "step": 1276
+    },
+    {
+      "epoch": 2.340659340659341,
+      "grad_norm": 0.23191671073436737,
+      "learning_rate": 4.199313921976511e-06,
+      "loss": 0.8344160318374634,
+      "step": 1278
+    },
+    {
+      "epoch": 2.3443223443223444,
+      "grad_norm": 0.3454512059688568,
+      "learning_rate": 4.176042790393313e-06,
+      "loss": 0.7739337682723999,
+      "step": 1280
+    },
+    {
+      "epoch": 2.347985347985348,
+      "grad_norm": 0.49713563919067383,
+      "learning_rate": 4.152878489115244e-06,
+      "loss": 0.6238831877708435,
+      "step": 1282
+    },
+    {
+      "epoch": 2.3516483516483517,
+      "grad_norm": 0.21321338415145874,
+      "learning_rate": 4.129821380783698e-06,
+      "loss": 1.091771125793457,
+      "step": 1284
+    },
+    {
+      "epoch": 2.3553113553113554,
+      "grad_norm": 0.9252800941467285,
+      "learning_rate": 4.106871826361952e-06,
+      "loss": 0.6089442372322083,
+      "step": 1286
+    },
+    {
+      "epoch": 2.358974358974359,
+      "grad_norm": 0.38212814927101135,
+      "learning_rate": 4.084030185129495e-06,
+      "loss": 0.9349772334098816,
+      "step": 1288
+    },
+    {
+      "epoch": 2.3626373626373627,
+      "grad_norm": 0.16641326248645782,
+      "learning_rate": 4.061296814676429e-06,
+      "loss": 0.896765947341919,
+      "step": 1290
+    },
+    {
+      "epoch": 2.3663003663003663,
+      "grad_norm": 0.3341747522354126,
+      "learning_rate": 4.038672070897844e-06,
+      "loss": 0.7659744024276733,
+      "step": 1292
+    },
+    {
+      "epoch": 2.36996336996337,
+      "grad_norm": 0.5208961963653564,
+      "learning_rate": 4.016156307988262e-06,
+      "loss": 0.8882443308830261,
+      "step": 1294
+    },
+    {
+      "epoch": 2.3736263736263736,
+      "grad_norm": 0.2249266356229782,
+      "learning_rate": 3.9937498784361e-06,
+      "loss": 1.2214046716690063,
+      "step": 1296
+    },
+    {
+      "epoch": 2.3772893772893773,
+      "grad_norm": 0.21020947396755219,
+      "learning_rate": 3.9714531330181275e-06,
+      "loss": 1.1040786504745483,
+      "step": 1298
+    },
+    {
+      "epoch": 2.380952380952381,
+      "grad_norm": 0.19527573883533478,
+      "learning_rate": 3.949266420793999e-06,
+      "loss": 1.1567643880844116,
+      "step": 1300
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.44200703501701355,
+      "learning_rate": 3.9271900891007734e-06,
+      "loss": 0.8775418996810913,
+      "step": 1302
+    },
+    {
+      "epoch": 2.3882783882783882,
+      "grad_norm": 0.41954803466796875,
+      "learning_rate": 3.905224483547479e-06,
+      "loss": 0.47357720136642456,
+      "step": 1304
+    },
+    {
+      "epoch": 2.391941391941392,
+      "grad_norm": 0.7417261600494385,
+      "learning_rate": 3.883369948009714e-06,
+      "loss": 0.9344196915626526,
+      "step": 1306
+    },
+    {
+      "epoch": 2.3956043956043955,
+      "grad_norm": 0.6811460256576538,
+      "learning_rate": 3.861626824624258e-06,
+      "loss": 1.1155997514724731,
+      "step": 1308
+    },
+    {
+      "epoch": 2.399267399267399,
+      "grad_norm": 0.42083945870399475,
+      "learning_rate": 3.839995453783694e-06,
+      "loss": 0.5117136836051941,
+      "step": 1310
+    },
+    {
+      "epoch": 2.402930402930403,
+      "grad_norm": 0.18507151305675507,
+      "learning_rate": 3.818476174131118e-06,
+      "loss": 1.11769437789917,
+      "step": 1312
+    },
+    {
+      "epoch": 2.4065934065934065,
+      "grad_norm": 0.1388321965932846,
+      "learning_rate": 3.7970693225548116e-06,
+      "loss": 0.8328091502189636,
+      "step": 1314
+    },
+    {
+      "epoch": 2.41025641025641,
+      "grad_norm": 0.31094399094581604,
+      "learning_rate": 3.7757752341829723e-06,
+      "loss": 1.115455985069275,
+      "step": 1316
+    },
+    {
+      "epoch": 2.413919413919414,
+      "grad_norm": 0.845039963722229,
+      "learning_rate": 3.754594242378466e-06,
+      "loss": 0.7884094715118408,
+      "step": 1318
+    },
+    {
+      "epoch": 2.4175824175824174,
+      "grad_norm": 0.31491175293922424,
+      "learning_rate": 3.7335266787336194e-06,
+      "loss": 0.7719835042953491,
+      "step": 1320
+    },
+    {
+      "epoch": 2.421245421245421,
+      "grad_norm": 0.5341482758522034,
+      "learning_rate": 3.712572873065012e-06,
+      "loss": 0.6111771464347839,
+      "step": 1322
+    },
+    {
+      "epoch": 2.4249084249084247,
+      "grad_norm": 1.4491015672683716,
+      "learning_rate": 3.69173315340833e-06,
+      "loss": 0.7621108889579773,
+      "step": 1324
+    },
+    {
+      "epoch": 2.4285714285714284,
+      "grad_norm": 1.1738007068634033,
+      "learning_rate": 3.6710078460132137e-06,
+      "loss": 0.7748345732688904,
+      "step": 1326
+    },
+    {
+      "epoch": 2.4322344322344325,
+      "grad_norm": 0.13677427172660828,
+      "learning_rate": 3.650397275338161e-06,
+      "loss": 0.6837164163589478,
+      "step": 1328
+    },
+    {
+      "epoch": 2.435897435897436,
+      "grad_norm": 1.0551925897598267,
+      "learning_rate": 3.6299017640454516e-06,
+      "loss": 0.8087068796157837,
+      "step": 1330
+    },
+    {
+      "epoch": 2.4395604395604398,
+      "grad_norm": 0.1782245635986328,
+      "learning_rate": 3.6095216329960786e-06,
+      "loss": 1.0505911111831665,
+      "step": 1332
+    },
+    {
+      "epoch": 2.4432234432234434,
+      "grad_norm": 0.20206193625926971,
+      "learning_rate": 3.5892572012447457e-06,
+      "loss": 0.7000587582588196,
+      "step": 1334
+    },
+    {
+      "epoch": 2.446886446886447,
+      "grad_norm": 0.2818435728549957,
+      "learning_rate": 3.5691087860348577e-06,
+      "loss": 0.6738724708557129,
+      "step": 1336
+    },
+    {
+      "epoch": 2.4505494505494507,
+      "grad_norm": 0.43494996428489685,
+      "learning_rate": 3.549076702793557e-06,
+      "loss": 1.0319753885269165,
+      "step": 1338
+    },
+    {
+      "epoch": 2.4542124542124544,
+      "grad_norm": 0.25145280361175537,
+      "learning_rate": 3.529161265126795e-06,
+      "loss": 0.8778097033500671,
+      "step": 1340
+    },
+    {
+      "epoch": 2.457875457875458,
+      "grad_norm": 0.08903591334819794,
+      "learning_rate": 3.5093627848144128e-06,
+      "loss": 0.6379270553588867,
+      "step": 1342
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.2862868309020996,
+      "learning_rate": 3.4896815718052534e-06,
+      "loss": 0.8156915903091431,
+      "step": 1344
+    },
+    {
+      "epoch": 2.4652014652014653,
+      "grad_norm": 0.14661253988742828,
+      "learning_rate": 3.4701179342123313e-06,
+      "loss": 1.0697602033615112,
+      "step": 1346
+    },
+    {
+      "epoch": 2.468864468864469,
+      "grad_norm": 0.1793731302022934,
+      "learning_rate": 3.4506721783079925e-06,
+      "loss": 1.1031157970428467,
+      "step": 1348
+    },
+    {
+      "epoch": 2.4725274725274726,
+      "grad_norm": 0.24754835665225983,
+      "learning_rate": 3.4313446085191203e-06,
+      "loss": 0.7289459705352783,
+      "step": 1350
+    },
+    {
+      "epoch": 2.4761904761904763,
+      "grad_norm": 0.159286767244339,
+      "learning_rate": 3.4121355274223727e-06,
+      "loss": 0.8309732675552368,
+      "step": 1352
+    },
+    {
+      "epoch": 2.47985347985348,
+      "grad_norm": 1.1426700353622437,
+      "learning_rate": 3.3930452357394473e-06,
+      "loss": 0.9143206477165222,
+      "step": 1354
+    },
+    {
+      "epoch": 2.4835164835164836,
+      "grad_norm": 0.05411989986896515,
+      "learning_rate": 3.3740740323323705e-06,
+      "loss": 0.9112240672111511,
+      "step": 1356
+    },
+    {
+      "epoch": 2.4871794871794872,
+      "grad_norm": 0.1567273885011673,
+      "learning_rate": 3.3552222141988257e-06,
+      "loss": 1.0814073085784912,
+      "step": 1358
+    },
+    {
+      "epoch": 2.490842490842491,
+      "grad_norm": 0.3032623827457428,
+      "learning_rate": 3.336490076467489e-06,
+      "loss": 0.8779569268226624,
+      "step": 1360
+    },
+    {
+      "epoch": 2.4945054945054945,
+      "grad_norm": 0.2695983350276947,
+      "learning_rate": 3.31787791239342e-06,
+      "loss": 0.8037658929824829,
+      "step": 1362
+    },
+    {
+      "epoch": 2.498168498168498,
+      "grad_norm": 0.2464946210384369,
+      "learning_rate": 3.2993860133534763e-06,
+      "loss": 0.8425225615501404,
+      "step": 1364
+    },
+    {
+      "epoch": 2.501831501831502,
+      "grad_norm": 0.16769325733184814,
+      "learning_rate": 3.2810146688417304e-06,
+      "loss": 1.1095960140228271,
+      "step": 1366
+    },
+    {
+      "epoch": 2.5054945054945055,
+      "grad_norm": 0.16507335007190704,
+      "learning_rate": 3.2627641664649666e-06,
+      "loss": 0.7693407535552979,
+      "step": 1368
+    },
+    {
+      "epoch": 2.509157509157509,
+      "grad_norm": 0.9319798350334167,
+      "learning_rate": 3.2446347919381533e-06,
+      "loss": 0.9375527501106262,
+      "step": 1370
+    },
+    {
+      "epoch": 2.5128205128205128,
+      "grad_norm": 0.4757522940635681,
+      "learning_rate": 3.226626829079979e-06,
+      "loss": 0.6353393197059631,
+      "step": 1372
+    },
+    {
+      "epoch": 2.5164835164835164,
+      "grad_norm": 0.20639510452747345,
+      "learning_rate": 3.2087405598084194e-06,
+      "loss": 0.7712477445602417,
+      "step": 1374
+    },
+    {
+      "epoch": 2.52014652014652,
+      "grad_norm": 0.2182164192199707,
+      "learning_rate": 3.1909762641363083e-06,
+      "loss": 0.9633672833442688,
+      "step": 1376
+    },
+    {
+      "epoch": 2.5238095238095237,
+      "grad_norm": 0.335929274559021,
+      "learning_rate": 3.173334220166962e-06,
+      "loss": 0.7830007076263428,
+      "step": 1378
+    },
+    {
+      "epoch": 2.5274725274725274,
+      "grad_norm": 0.23919770121574402,
+      "learning_rate": 3.155814704089823e-06,
+      "loss": 0.8970922827720642,
+      "step": 1380
+    },
+    {
+      "epoch": 2.531135531135531,
+      "grad_norm": 0.034020014107227325,
+      "learning_rate": 3.1384179901761343e-06,
+      "loss": 0.8635251522064209,
+      "step": 1382
+    },
+    {
+      "epoch": 2.5347985347985347,
+      "grad_norm": 0.215301051735878,
+      "learning_rate": 3.1211443507746546e-06,
+      "loss": 0.7926411628723145,
+      "step": 1384
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.14359386265277863,
+      "learning_rate": 3.1039940563073894e-06,
+      "loss": 1.1008884906768799,
+      "step": 1386
+    },
+    {
+      "epoch": 2.542124542124542,
+      "grad_norm": 0.46346187591552734,
+      "learning_rate": 3.0869673752653447e-06,
+      "loss": 0.7606490254402161,
+      "step": 1388
+    },
+    {
+      "epoch": 2.5457875457875456,
+      "grad_norm": 0.342520147562027,
+      "learning_rate": 3.0700645742043476e-06,
+      "loss": 0.8070803880691528,
+      "step": 1390
+    },
+    {
+      "epoch": 2.5494505494505493,
+      "grad_norm": 0.3418232202529907,
+      "learning_rate": 3.0532859177408587e-06,
+      "loss": 0.983840823173523,
+      "step": 1392
+    },
+    {
+      "epoch": 2.553113553113553,
+      "grad_norm": 0.1977006494998932,
+      "learning_rate": 3.03663166854783e-06,
+      "loss": 0.7210864424705505,
+      "step": 1394
+    },
+    {
+      "epoch": 2.5567765567765566,
+      "grad_norm": 0.16176871955394745,
+      "learning_rate": 3.020102087350594e-06,
+      "loss": 0.34565648436546326,
+      "step": 1396
+    },
+    {
+      "epoch": 2.5604395604395602,
+      "grad_norm": 0.2752530574798584,
+      "learning_rate": 3.0036974329227862e-06,
+      "loss": 1.1138232946395874,
+      "step": 1398
+    },
+    {
+      "epoch": 2.564102564102564,
+      "grad_norm": 0.2671622931957245,
+      "learning_rate": 2.9874179620822856e-06,
+      "loss": 1.1061241626739502,
+      "step": 1400
+    },
+    {
+      "epoch": 2.5677655677655675,
+      "grad_norm": 0.5395840406417847,
+      "learning_rate": 2.971263929687207e-06,
+      "loss": 0.8952687978744507,
+      "step": 1402
+    },
+    {
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.2224486917257309,
+      "learning_rate": 2.9552355886318968e-06,
+      "loss": 1.105149269104004,
+      "step": 1404
+    },
+    {
+      "epoch": 2.575091575091575,
+      "grad_norm": 1.0579370260238647,
+      "learning_rate": 2.9393331898429777e-06,
+      "loss": 0.5482126474380493,
+      "step": 1406
+    },
+    {
+      "epoch": 2.578754578754579,
+      "grad_norm": 0.2807334065437317,
+      "learning_rate": 2.9235569822754317e-06,
+      "loss": 1.1531801223754883,
+      "step": 1408
+    },
+    {
+      "epoch": 2.5824175824175826,
+      "grad_norm": 0.360220730304718,
+      "learning_rate": 2.9079072129086906e-06,
+      "loss": 0.8873265981674194,
+      "step": 1410
+    },
+    {
+      "epoch": 2.586080586080586,
+      "grad_norm": 0.6387040019035339,
+      "learning_rate": 2.89238412674277e-06,
+      "loss": 0.8655451536178589,
+      "step": 1412
+    },
+    {
+      "epoch": 2.58974358974359,
+      "grad_norm": 0.5153301358222961,
+      "learning_rate": 2.8769879667944393e-06,
+      "loss": 0.953106164932251,
+      "step": 1414
+    },
+    {
+      "epoch": 2.5934065934065935,
+      "grad_norm": 0.44717395305633545,
+      "learning_rate": 2.8617189740934113e-06,
+      "loss": 1.0878214836120605,
+      "step": 1416
+    },
+    {
+      "epoch": 2.597069597069597,
+      "grad_norm": 0.2903793454170227,
+      "learning_rate": 2.8465773876785786e-06,
+      "loss": 0.7813395261764526,
+      "step": 1418
+    },
+    {
+      "epoch": 2.600732600732601,
+      "grad_norm": 0.13546399772167206,
+      "learning_rate": 2.8315634445942623e-06,
+      "loss": 1.085912823677063,
+      "step": 1420
+    },
+    {
+      "epoch": 2.6043956043956045,
+      "grad_norm": 0.4397055208683014,
+      "learning_rate": 2.8166773798864978e-06,
+      "loss": 1.0346283912658691,
+      "step": 1422
+    },
+    {
+      "epoch": 2.608058608058608,
+      "grad_norm": 0.25535109639167786,
+      "learning_rate": 2.8019194265993683e-06,
+      "loss": 1.1859427690505981,
+      "step": 1424
+    },
+    {
+      "epoch": 2.6117216117216118,
+      "grad_norm": 0.4347076416015625,
+      "learning_rate": 2.787289815771348e-06,
+      "loss": 1.2661553621292114,
+      "step": 1426
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.15085160732269287,
+      "learning_rate": 2.7727887764316835e-06,
+      "loss": 1.1397687196731567,
+      "step": 1428
+    },
+    {
+      "epoch": 2.619047619047619,
+      "grad_norm": 0.3919295370578766,
+      "learning_rate": 2.758416535596812e-06,
+      "loss": 1.078932285308838,
+      "step": 1430
+    },
+    {
+      "epoch": 2.6227106227106227,
+      "grad_norm": 0.67396479845047,
+      "learning_rate": 2.744173318266809e-06,
+      "loss": 0.9404812455177307,
+      "step": 1432
+    },
+    {
+      "epoch": 2.6263736263736264,
+      "grad_norm": 0.6748982667922974,
+      "learning_rate": 2.7300593474218583e-06,
+      "loss": 0.944557785987854,
+      "step": 1434
+    },
+    {
+      "epoch": 2.63003663003663,
+      "grad_norm": 0.33066844940185547,
+      "learning_rate": 2.7160748440187736e-06,
+      "loss": 1.3000890016555786,
+      "step": 1436
+    },
+    {
+      "epoch": 2.6336996336996337,
+      "grad_norm": 0.7671491503715515,
+      "learning_rate": 2.702220026987525e-06,
+      "loss": 1.1358200311660767,
+      "step": 1438
+    },
+    {
+      "epoch": 2.6373626373626373,
+      "grad_norm": 0.3073434829711914,
+      "learning_rate": 2.6884951132278185e-06,
+      "loss": 1.0983126163482666,
+      "step": 1440
+    },
+    {
+      "epoch": 2.641025641025641,
+      "grad_norm": 0.21354691684246063,
+      "learning_rate": 2.6749003176057092e-06,
+      "loss": 1.1217743158340454,
+      "step": 1442
+    },
+    {
+      "epoch": 2.6446886446886446,
+      "grad_norm": 0.8426246047019958,
+      "learning_rate": 2.6614358529502165e-06,
+      "loss": 1.0780593156814575,
+      "step": 1444
+    },
+    {
+      "epoch": 2.6483516483516483,
+      "grad_norm": 0.23787999153137207,
+      "learning_rate": 2.6481019300500166e-06,
+      "loss": 0.7865286469459534,
+      "step": 1446
+    },
+    {
+      "epoch": 2.652014652014652,
+      "grad_norm": 0.1286686360836029,
+      "learning_rate": 2.634898757650121e-06,
+      "loss": 0.8131098747253418,
+      "step": 1448
+    },
+    {
+      "epoch": 2.6556776556776556,
+      "grad_norm": 0.1445346623659134,
+      "learning_rate": 2.6218265424486233e-06,
+      "loss": 1.0926539897918701,
+      "step": 1450
+    },
+    {
+      "epoch": 2.659340659340659,
+      "grad_norm": 0.7427401542663574,
+      "learning_rate": 2.608885489093455e-06,
+      "loss": 0.6913706660270691,
+      "step": 1452
+    },
+    {
+      "epoch": 2.663003663003663,
+      "grad_norm": 0.06115258112549782,
+      "learning_rate": 2.5960758001791893e-06,
+      "loss": 0.6983692646026611,
+      "step": 1454
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.2783794701099396,
+      "learning_rate": 2.5833976762438605e-06,
+      "loss": 1.070138692855835,
+      "step": 1456
+    },
+    {
+      "epoch": 2.67032967032967,
+      "grad_norm": 0.2989695966243744,
+      "learning_rate": 2.5708513157658295e-06,
+      "loss": 0.7558898329734802,
+      "step": 1458
+    },
+    {
+      "epoch": 2.6739926739926743,
+      "grad_norm": 0.06930552423000336,
+      "learning_rate": 2.5584369151606785e-06,
+      "loss": 0.7775593400001526,
+      "step": 1460
+    },
+    {
+      "epoch": 2.677655677655678,
+      "grad_norm": 0.25155073404312134,
+      "learning_rate": 2.5461546687781325e-06,
+      "loss": 0.7822065949440002,
+      "step": 1462
+    },
+    {
+      "epoch": 2.6813186813186816,
+      "grad_norm": 0.24393150210380554,
+      "learning_rate": 2.5340047688990142e-06,
+      "loss": 1.0643365383148193,
+      "step": 1464
+    },
+    {
+      "epoch": 2.684981684981685,
+      "grad_norm": 0.23728282749652863,
+      "learning_rate": 2.5219874057322453e-06,
+      "loss": 0.8780837655067444,
+      "step": 1466
+    },
+    {
+      "epoch": 2.688644688644689,
+      "grad_norm": 0.19571846723556519,
+      "learning_rate": 2.5101027674118523e-06,
+      "loss": 1.0327012538909912,
+      "step": 1468
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.4742324650287628,
+      "learning_rate": 2.4983510399940377e-06,
+      "loss": 0.6416606903076172,
+      "step": 1470
+    },
+    {
+      "epoch": 2.695970695970696,
+      "grad_norm": 0.19169728457927704,
+      "learning_rate": 2.4867324074542525e-06,
+      "loss": 0.7808413505554199,
+      "step": 1472
+    },
+    {
+      "epoch": 2.6996336996337,
+      "grad_norm": 0.18114076554775238,
+      "learning_rate": 2.4752470516843257e-06,
+      "loss": 0.4357774257659912,
+      "step": 1474
+    },
+    {
+      "epoch": 2.7032967032967035,
+      "grad_norm": 0.2890799343585968,
+      "learning_rate": 2.463895152489617e-06,
+      "loss": 1.1387816667556763,
+      "step": 1476
+    },
+    {
+      "epoch": 2.706959706959707,
+      "grad_norm": 0.32990363240242004,
+      "learning_rate": 2.4526768875861938e-06,
+      "loss": 0.6536518335342407,
+      "step": 1478
+    },
+    {
+      "epoch": 2.7106227106227108,
+      "grad_norm": 0.3412167727947235,
+      "learning_rate": 2.4415924325980575e-06,
+      "loss": 1.0885263681411743,
+      "step": 1480
+    },
+    {
+      "epoch": 2.7142857142857144,
+      "grad_norm": 0.17220936715602875,
+      "learning_rate": 2.4306419610543885e-06,
+      "loss": 1.111185073852539,
+      "step": 1482
+    },
+    {
+      "epoch": 2.717948717948718,
+      "grad_norm": 0.3688224256038666,
+      "learning_rate": 2.4198256443868327e-06,
+      "loss": 1.0014547109603882,
+      "step": 1484
+    },
+    {
+      "epoch": 2.7216117216117217,
+      "grad_norm": 0.23104700446128845,
+      "learning_rate": 2.4091436519268167e-06,
+      "loss": 1.1393983364105225,
+      "step": 1486
+    },
+    {
+      "epoch": 2.7252747252747254,
+      "grad_norm": 0.22667832672595978,
+      "learning_rate": 2.3985961509028994e-06,
+      "loss": 1.1307849884033203,
+      "step": 1488
+    },
+    {
+      "epoch": 2.728937728937729,
+      "grad_norm": 0.7048260569572449,
+      "learning_rate": 2.3881833064381478e-06,
+      "loss": 0.5263049006462097,
+      "step": 1490
+    },
+    {
+      "epoch": 2.7326007326007327,
+      "grad_norm": 0.1926041543483734,
+      "learning_rate": 2.3779052815475553e-06,
+      "loss": 1.1333121061325073,
+      "step": 1492
+    },
+    {
+      "epoch": 2.7362637362637363,
+      "grad_norm": 0.42971038818359375,
+      "learning_rate": 2.3677622371354932e-06,
+      "loss": 0.6400864124298096,
+      "step": 1494
+    },
+    {
+      "epoch": 2.73992673992674,
+      "grad_norm": 0.2434394806623459,
+      "learning_rate": 2.357754331993187e-06,
+      "loss": 1.201290249824524,
+      "step": 1496
+    },
+    {
+      "epoch": 2.7435897435897436,
+      "grad_norm": 0.16831986606121063,
+      "learning_rate": 2.347881722796234e-06,
+      "loss": 1.0851210355758667,
+      "step": 1498
+    },
+    {
+      "epoch": 2.7472527472527473,
+      "grad_norm": 0.3970390856266022,
+      "learning_rate": 2.3381445641021445e-06,
+      "loss": 0.7578325271606445,
+      "step": 1500
+    },
+    {
+      "epoch": 2.750915750915751,
+      "grad_norm": 0.4430757761001587,
+      "learning_rate": 2.328543008347928e-06,
+      "loss": 1.035079002380371,
+      "step": 1502
+    },
+    {
+      "epoch": 2.7545787545787546,
+      "grad_norm": 0.23971140384674072,
+      "learning_rate": 2.31907720584771e-06,
+      "loss": 1.1079963445663452,
+      "step": 1504
+    },
+    {
+      "epoch": 2.758241758241758,
+      "grad_norm": 0.37300434708595276,
+      "learning_rate": 2.3097473047903645e-06,
+      "loss": 1.1286638975143433,
+      "step": 1506
+    },
+    {
+      "epoch": 2.761904761904762,
+      "grad_norm": 0.0422448106110096,
+      "learning_rate": 2.3005534512372106e-06,
+      "loss": 0.9173861145973206,
+      "step": 1508
+    },
+    {
+      "epoch": 2.7655677655677655,
+      "grad_norm": 0.25997260212898254,
+      "learning_rate": 2.2914957891197182e-06,
+      "loss": 1.0080313682556152,
+      "step": 1510
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.30770131945610046,
+      "learning_rate": 2.2825744602372506e-06,
+      "loss": 0.9287357330322266,
+      "step": 1512
+    },
+    {
+      "epoch": 2.772893772893773,
+      "grad_norm": 0.2603771686553955,
+      "learning_rate": 2.2737896042548537e-06,
+      "loss": 1.0911868810653687,
+      "step": 1514
+    },
+    {
+      "epoch": 2.7765567765567765,
+      "grad_norm": 0.38838595151901245,
+      "learning_rate": 2.2651413587010634e-06,
+      "loss": 1.0172020196914673,
+      "step": 1516
+    },
+    {
+      "epoch": 2.78021978021978,
+      "grad_norm": 0.04338948056101799,
+      "learning_rate": 2.2566298589657546e-06,
+      "loss": 0.9788475036621094,
+      "step": 1518
+    },
+    {
+      "epoch": 2.7838827838827838,
+      "grad_norm": 0.3132210373878479,
+      "learning_rate": 2.2482552382980194e-06,
+      "loss": 0.5113797187805176,
+      "step": 1520
+    },
+    {
+      "epoch": 2.7875457875457874,
+      "grad_norm": 0.41038262844085693,
+      "learning_rate": 2.240017627804088e-06,
+      "loss": 0.7732734084129333,
+      "step": 1522
+    },
+    {
+      "epoch": 2.791208791208791,
+      "grad_norm": 0.23137636482715607,
+      "learning_rate": 2.231917156445265e-06,
+      "loss": 0.8295901417732239,
+      "step": 1524
+    },
+    {
+      "epoch": 2.7948717948717947,
+      "grad_norm": 0.3408252000808716,
+      "learning_rate": 2.223953951035919e-06,
+      "loss": 1.143431544303894,
+      "step": 1526
+    },
+    {
+      "epoch": 2.7985347985347984,
+      "grad_norm": 0.2596583664417267,
+      "learning_rate": 2.216128136241497e-06,
+      "loss": 1.099791407585144,
+      "step": 1528
+    },
+    {
+      "epoch": 2.802197802197802,
+      "grad_norm": 0.18814221024513245,
+      "learning_rate": 2.208439834576568e-06,
+      "loss": 1.0971970558166504,
+      "step": 1530
+    },
+    {
+      "epoch": 2.8058608058608057,
+      "grad_norm": 0.20894794166088104,
+      "learning_rate": 2.200889166402908e-06,
+      "loss": 0.9817790389060974,
+      "step": 1532
+    },
+    {
+      "epoch": 2.8095238095238093,
+      "grad_norm": 0.3537425696849823,
+      "learning_rate": 2.193476249927617e-06,
+      "loss": 0.7601557970046997,
+      "step": 1534
+    },
+    {
+      "epoch": 2.813186813186813,
+      "grad_norm": 0.22301346063613892,
+      "learning_rate": 2.1862012012012647e-06,
+      "loss": 1.2858803272247314,
+      "step": 1536
+    },
+    {
+      "epoch": 2.8168498168498166,
+      "grad_norm": 0.5590947270393372,
+      "learning_rate": 2.179064134116078e-06,
+      "loss": 0.9057199954986572,
+      "step": 1538
+    },
+    {
+      "epoch": 2.8205128205128203,
+      "grad_norm": 0.38495972752571106,
+      "learning_rate": 2.1720651604041543e-06,
+      "loss": 0.7820447683334351,
+      "step": 1540
+    },
+    {
+      "epoch": 2.824175824175824,
+      "grad_norm": 0.41492217779159546,
+      "learning_rate": 2.1652043896357132e-06,
+      "loss": 0.8878074288368225,
+      "step": 1542
+    },
+    {
+      "epoch": 2.8278388278388276,
+      "grad_norm": 0.45532068610191345,
+      "learning_rate": 2.1584819292173844e-06,
+      "loss": 1.0764187574386597,
+      "step": 1544
+    },
+    {
+      "epoch": 2.8315018315018317,
+      "grad_norm": 0.2862100899219513,
+      "learning_rate": 2.1518978843905204e-06,
+      "loss": 1.1286197900772095,
+      "step": 1546
+    },
+    {
+      "epoch": 2.8351648351648353,
+      "grad_norm": 0.18794697523117065,
+      "learning_rate": 2.1454523582295567e-06,
+      "loss": 1.1888434886932373,
+      "step": 1548
+    },
+    {
+      "epoch": 2.838827838827839,
+      "grad_norm": 0.4953882396221161,
+      "learning_rate": 2.1391454516403876e-06,
+      "loss": 0.8518368601799011,
+      "step": 1550
+    },
+    {
+      "epoch": 2.8424908424908426,
+      "grad_norm": 0.4500260353088379,
+      "learning_rate": 2.1329772633587976e-06,
+      "loss": 0.5600649118423462,
+      "step": 1552
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.21223555505275726,
+      "learning_rate": 2.1269478899489068e-06,
+      "loss": 1.1281017065048218,
+      "step": 1554
+    },
+    {
+      "epoch": 2.84981684981685,
+      "grad_norm": 0.17230379581451416,
+      "learning_rate": 2.1210574258016675e-06,
+      "loss": 0.9187098741531372,
+      "step": 1556
+    },
+    {
+      "epoch": 2.8534798534798536,
+      "grad_norm": 0.13780535757541656,
+      "learning_rate": 2.1153059631333785e-06,
+      "loss": 1.0671217441558838,
+      "step": 1558
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.29513898491859436,
+      "learning_rate": 2.1096935919842434e-06,
+      "loss": 0.5967673063278198,
+      "step": 1560
+    },
+    {
+      "epoch": 2.860805860805861,
+      "grad_norm": 0.2297191023826599,
+      "learning_rate": 2.104220400216967e-06,
+      "loss": 0.705269992351532,
+      "step": 1562
+    },
+    {
+      "epoch": 2.8644688644688645,
+      "grad_norm": 0.23687344789505005,
+      "learning_rate": 2.0988864735153724e-06,
+      "loss": 0.8660311102867126,
+      "step": 1564
+    },
+    {
+      "epoch": 2.868131868131868,
+      "grad_norm": 0.4433053433895111,
+      "learning_rate": 2.0936918953830633e-06,
+      "loss": 0.6954091787338257,
+      "step": 1566
+    },
+    {
+      "epoch": 2.871794871794872,
+      "grad_norm": 0.30360910296440125,
+      "learning_rate": 2.088636747142114e-06,
+      "loss": 0.7069787979125977,
+      "step": 1568
+    },
+    {
+      "epoch": 2.8754578754578755,
+      "grad_norm": 0.06758692115545273,
+      "learning_rate": 2.083721107931803e-06,
+      "loss": 0.7010313868522644,
+      "step": 1570
+    },
+    {
+      "epoch": 2.879120879120879,
+      "grad_norm": 0.30862146615982056,
+      "learning_rate": 2.0789450547073634e-06,
+      "loss": 0.66036057472229,
+      "step": 1572
+    },
+    {
+      "epoch": 2.8827838827838828,
+      "grad_norm": 0.2272968739271164,
+      "learning_rate": 2.074308662238789e-06,
+      "loss": 0.9868662357330322,
+      "step": 1574
+    },
+    {
+      "epoch": 2.8864468864468864,
+      "grad_norm": 0.4267461895942688,
+      "learning_rate": 2.069812003109654e-06,
+      "loss": 0.9745535254478455,
+      "step": 1576
+    },
+    {
+      "epoch": 2.89010989010989,
+      "grad_norm": 0.08947110921144485,
+      "learning_rate": 2.0654551477159868e-06,
+      "loss": 0.8887320160865784,
+      "step": 1578
+    },
+    {
+      "epoch": 2.8937728937728937,
+      "grad_norm": 0.20402641594409943,
+      "learning_rate": 2.0612381642651584e-06,
+      "loss": 1.170695185661316,
+      "step": 1580
+    },
+    {
+      "epoch": 2.8974358974358974,
+      "grad_norm": 0.3134024441242218,
+      "learning_rate": 2.057161118774821e-06,
+      "loss": 0.8388773798942566,
+      "step": 1582
+    },
+    {
+      "epoch": 2.901098901098901,
+      "grad_norm": 0.04043447971343994,
+      "learning_rate": 2.05322407507187e-06,
+      "loss": 0.7783088684082031,
+      "step": 1584
+    },
+    {
+      "epoch": 2.9047619047619047,
+      "grad_norm": 0.30711326003074646,
+      "learning_rate": 2.0494270947914507e-06,
+      "loss": 0.788719892501831,
+      "step": 1586
+    },
+    {
+      "epoch": 2.9084249084249083,
+      "grad_norm": 0.3433006703853607,
+      "learning_rate": 2.0457702373759864e-06,
+      "loss": 1.1323091983795166,
+      "step": 1588
+    },
+    {
+      "epoch": 2.912087912087912,
+      "grad_norm": 0.2119935005903244,
+      "learning_rate": 2.0422535600742526e-06,
+      "loss": 1.1913877725601196,
+      "step": 1590
+    },
+    {
+      "epoch": 2.9157509157509156,
+      "grad_norm": 0.4435896575450897,
+      "learning_rate": 2.03887711794048e-06,
+      "loss": 0.5050473809242249,
+      "step": 1592
+    },
+    {
+      "epoch": 2.9194139194139193,
+      "grad_norm": 0.5109768509864807,
+      "learning_rate": 2.0356409638334902e-06,
+      "loss": 1.1414501667022705,
+      "step": 1594
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.4784262776374817,
+      "learning_rate": 2.032545148415871e-06,
+      "loss": 0.747880756855011,
+      "step": 1596
+    },
+    {
+      "epoch": 2.926739926739927,
+      "grad_norm": 0.19986164569854736,
+      "learning_rate": 2.0295897201531838e-06,
+      "loss": 1.1563737392425537,
+      "step": 1598
+    },
+    {
+      "epoch": 2.9304029304029307,
+      "grad_norm": 0.2983030378818512,
+      "learning_rate": 2.026774725313199e-06,
+      "loss": 0.7789967060089111,
+      "step": 1600
+    },
+    {
+      "epoch": 2.9340659340659343,
+      "grad_norm": 0.21115927398204803,
+      "learning_rate": 2.0241002079651803e-06,
+      "loss": 1.1291173696517944,
+      "step": 1602
+    },
+    {
+      "epoch": 2.937728937728938,
+      "grad_norm": 0.08551673591136932,
+      "learning_rate": 2.0215662099791874e-06,
+      "loss": 0.8520828485488892,
+      "step": 1604
+    },
+    {
+      "epoch": 2.9413919413919416,
+      "grad_norm": 0.17809857428073883,
+      "learning_rate": 2.019172771025426e-06,
+      "loss": 1.1100882291793823,
+      "step": 1606
+    },
+    {
+      "epoch": 2.9450549450549453,
+      "grad_norm": 0.18868334591388702,
+      "learning_rate": 2.0169199285736234e-06,
+      "loss": 0.7092351317405701,
+      "step": 1608
+    },
+    {
+      "epoch": 2.948717948717949,
+      "grad_norm": 0.32582998275756836,
+      "learning_rate": 2.0148077178924412e-06,
+      "loss": 1.0054452419281006,
+      "step": 1610
+    },
+    {
+      "epoch": 2.9523809523809526,
+      "grad_norm": 1.538607120513916,
+      "learning_rate": 2.0128361720489263e-06,
+      "loss": 0.8747723698616028,
+      "step": 1612
+    },
+    {
+      "epoch": 2.956043956043956,
+      "grad_norm": 1.1463967561721802,
+      "learning_rate": 2.0110053219079927e-06,
+      "loss": 0.6871626377105713,
+      "step": 1614
+    },
+    {
+      "epoch": 2.95970695970696,
+      "grad_norm": 0.29785850644111633,
+      "learning_rate": 2.009315196131934e-06,
+      "loss": 0.8241419792175293,
+      "step": 1616
+    },
+    {
+      "epoch": 2.9633699633699635,
+      "grad_norm": 0.2533627152442932,
+      "learning_rate": 2.0077658211799823e-06,
+      "loss": 1.3680229187011719,
+      "step": 1618
+    },
+    {
+      "epoch": 2.967032967032967,
+      "grad_norm": 0.41148898005485535,
+      "learning_rate": 2.0063572213078856e-06,
+      "loss": 1.2273290157318115,
+      "step": 1620
+    },
+    {
+      "epoch": 2.970695970695971,
+      "grad_norm": 0.2996610701084137,
+      "learning_rate": 2.0050894185675354e-06,
+      "loss": 0.9176530838012695,
+      "step": 1622
+    },
+    {
+      "epoch": 2.9743589743589745,
+      "grad_norm": 0.6570471525192261,
+      "learning_rate": 2.0039624328066154e-06,
+      "loss": 0.780877411365509,
+      "step": 1624
+    },
+    {
+      "epoch": 2.978021978021978,
+      "grad_norm": 0.5498982667922974,
+      "learning_rate": 2.0029762816682963e-06,
+      "loss": 0.8718687295913696,
+      "step": 1626
+    },
+    {
+      "epoch": 2.9816849816849818,
+      "grad_norm": 0.1393526792526245,
+      "learning_rate": 2.0021309805909546e-06,
+      "loss": 0.9456213116645813,
+      "step": 1628
+    },
+    {
+      "epoch": 2.9853479853479854,
+      "grad_norm": 1.554042935371399,
+      "learning_rate": 2.001426542807935e-06,
+      "loss": 1.4017192125320435,
+      "step": 1630
+    },
+    {
+      "epoch": 2.989010989010989,
+      "grad_norm": 0.4462849199771881,
+      "learning_rate": 2.000862979347339e-06,
+      "loss": 1.0045907497406006,
+      "step": 1632
+    },
+    {
+      "epoch": 2.9926739926739927,
+      "grad_norm": 0.33116626739501953,
+      "learning_rate": 2.0004402990318574e-06,
+      "loss": 0.7610074281692505,
+      "step": 1634
+    },
+    {
+      "epoch": 2.9963369963369964,
+      "grad_norm": 0.1885322779417038,
+      "learning_rate": 2.000158508478629e-06,
+      "loss": 1.2684826850891113,
+      "step": 1636
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.30161595344543457,
+      "learning_rate": 2.0000176120991345e-06,
+      "loss": 1.06321382522583,
+      "step": 1638
+    },
+    {
+      "epoch": 3.0,
+      "step": 1638,
+      "total_flos": 8.4482141520606e+18,
+      "train_loss": 1.0233670397410317,
+      "train_runtime": 54643.5024,
+      "train_samples_per_second": 0.719,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 1638,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 99999,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.4482141520606e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}