diff --git "a/checkpoint-3500/trainer_state.json" "b/checkpoint-3500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3500/trainer_state.json"
@@ -0,0 +1,4989 @@
+{
+  "best_metric": 0.07722621411085129,
+  "best_model_checkpoint": "/teamspace/studios/this_studio/output/Mistral_End/checkpoint-3500",
+  "epoch": 1.6376184349046672,
+  "eval_steps": 500,
+  "global_step": 3500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0023394549070066676,
+      "grad_norm": 1.2956504821777344,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.7204,
+      "step": 5
+    },
+    {
+      "epoch": 0.004678909814013335,
+      "grad_norm": 1.2448140382766724,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.7411,
+      "step": 10
+    },
+    {
+      "epoch": 0.007018364721020002,
+      "grad_norm": 1.3956960439682007,
+      "learning_rate": 1.5e-06,
+      "loss": 0.7235,
+      "step": 15
+    },
+    {
+      "epoch": 0.00935781962802667,
+      "grad_norm": 1.10906982421875,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7176,
+      "step": 20
+    },
+    {
+      "epoch": 0.011697274535033338,
+      "grad_norm": 1.1599094867706299,
+      "learning_rate": 2.5e-06,
+      "loss": 0.7439,
+      "step": 25
+    },
+    {
+      "epoch": 0.014036729442040005,
+      "grad_norm": 1.085955023765564,
+      "learning_rate": 3e-06,
+      "loss": 0.6894,
+      "step": 30
+    },
+    {
+      "epoch": 0.016376184349046672,
+      "grad_norm": 0.9708353877067566,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 0.6607,
+      "step": 35
+    },
+    {
+      "epoch": 0.01871563925605334,
+      "grad_norm": 0.8504016399383545,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6335,
+      "step": 40
+    },
+    {
+      "epoch": 0.021055094163060006,
+      "grad_norm": 0.7278764843940735,
+      "learning_rate": 4.5e-06,
+      "loss": 0.5814,
+      "step": 45
+    },
+    {
+      "epoch": 0.023394549070066675,
+      "grad_norm": 0.5420047044754028,
+      "learning_rate": 5e-06,
+      "loss": 0.5636,
+      "step": 50
+    },
+    {
+      "epoch": 0.02573400397707334,
+      "grad_norm": 0.5314227938652039,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.5845,
+      "step": 55
+    },
+    {
+      "epoch": 0.02807345888408001,
+      "grad_norm": 0.48584049940109253,
+      "learning_rate": 6e-06,
+      "loss": 0.494,
+      "step": 60
+    },
+    {
+      "epoch": 0.030412913791086675,
+      "grad_norm": 0.49247828125953674,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.4809,
+      "step": 65
+    },
+    {
+      "epoch": 0.032752368698093344,
+      "grad_norm": 0.470062255859375,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.446,
+      "step": 70
+    },
+    {
+      "epoch": 0.03509182360510001,
+      "grad_norm": 0.4149915874004364,
+      "learning_rate": 7.5e-06,
+      "loss": 0.3915,
+      "step": 75
+    },
+    {
+      "epoch": 0.03743127851210668,
+      "grad_norm": 0.41831153631210327,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.3696,
+      "step": 80
+    },
+    {
+      "epoch": 0.039770733419113344,
+      "grad_norm": 0.40496742725372314,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 0.3464,
+      "step": 85
+    },
+    {
+      "epoch": 0.04211018832612001,
+      "grad_norm": 0.41090327501296997,
+      "learning_rate": 9e-06,
+      "loss": 0.3097,
+      "step": 90
+    },
+    {
+      "epoch": 0.04444964323312668,
+      "grad_norm": 0.41252198815345764,
+      "learning_rate": 9.5e-06,
+      "loss": 0.3111,
+      "step": 95
+    },
+    {
+      "epoch": 0.04678909814013335,
+      "grad_norm": 0.44755125045776367,
+      "learning_rate": 1e-05,
+      "loss": 0.2696,
+      "step": 100
+    },
+    {
+      "epoch": 0.04912855304714002,
+      "grad_norm": 0.4423600137233734,
+      "learning_rate": 1.05e-05,
+      "loss": 0.2461,
+      "step": 105
+    },
+    {
+      "epoch": 0.05146800795414668,
+      "grad_norm": 0.44379884004592896,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.2139,
+      "step": 110
+    },
+    {
+      "epoch": 0.05380746286115335,
+      "grad_norm": 0.43690061569213867,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 0.2045,
+      "step": 115
+    },
+    {
+      "epoch": 0.05614691776816002,
+      "grad_norm": 0.4275108277797699,
+      "learning_rate": 1.2e-05,
+      "loss": 0.1987,
+      "step": 120
+    },
+    {
+      "epoch": 0.05848637267516669,
+      "grad_norm": 0.5754674077033997,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1899,
+      "step": 125
+    },
+    {
+      "epoch": 0.06082582758217335,
+      "grad_norm": 0.4689083397388458,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.171,
+      "step": 130
+    },
+    {
+      "epoch": 0.06316528248918002,
+      "grad_norm": 0.42702773213386536,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.1636,
+      "step": 135
+    },
+    {
+      "epoch": 0.06550473739618669,
+      "grad_norm": 0.4773492217063904,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.1579,
+      "step": 140
+    },
+    {
+      "epoch": 0.06784419230319336,
+      "grad_norm": 0.5190818309783936,
+      "learning_rate": 1.45e-05,
+      "loss": 0.1377,
+      "step": 145
+    },
+    {
+      "epoch": 0.07018364721020003,
+      "grad_norm": 0.4696808159351349,
+      "learning_rate": 1.5e-05,
+      "loss": 0.1325,
+      "step": 150
+    },
+    {
+      "epoch": 0.0725231021172067,
+      "grad_norm": 0.4878705143928528,
+      "learning_rate": 1.55e-05,
+      "loss": 0.1384,
+      "step": 155
+    },
+    {
+      "epoch": 0.07486255702421336,
+      "grad_norm": 0.48388373851776123,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.1198,
+      "step": 160
+    },
+    {
+      "epoch": 0.07720201193122003,
+      "grad_norm": 0.47326740622520447,
+      "learning_rate": 1.65e-05,
+      "loss": 0.1192,
+      "step": 165
+    },
+    {
+      "epoch": 0.07954146683822669,
+      "grad_norm": 0.4732869565486908,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.1208,
+      "step": 170
+    },
+    {
+      "epoch": 0.08188092174523336,
+      "grad_norm": 0.45708340406417847,
+      "learning_rate": 1.75e-05,
+      "loss": 0.1099,
+      "step": 175
+    },
+    {
+      "epoch": 0.08422037665224003,
+      "grad_norm": 0.5060539245605469,
+      "learning_rate": 1.8e-05,
+      "loss": 0.104,
+      "step": 180
+    },
+    {
+      "epoch": 0.0865598315592467,
+      "grad_norm": 0.45235252380371094,
+      "learning_rate": 1.85e-05,
+      "loss": 0.1489,
+      "step": 185
+    },
+    {
+      "epoch": 0.08889928646625336,
+      "grad_norm": 0.48685696721076965,
+      "learning_rate": 1.9e-05,
+      "loss": 0.1102,
+      "step": 190
+    },
+    {
+      "epoch": 0.09123874137326003,
+      "grad_norm": 0.45249509811401367,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 0.1172,
+      "step": 195
+    },
+    {
+      "epoch": 0.0935781962802667,
+      "grad_norm": 0.456840455532074,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 200
+    },
+    {
+      "epoch": 0.09591765118727337,
+      "grad_norm": 0.4239897131919861,
+      "learning_rate": 2.05e-05,
+      "loss": 0.1024,
+      "step": 205
+    },
+    {
+      "epoch": 0.09825710609428004,
+      "grad_norm": 0.3757131099700928,
+      "learning_rate": 2.1e-05,
+      "loss": 0.1117,
+      "step": 210
+    },
+    {
+      "epoch": 0.1005965610012867,
+      "grad_norm": 0.43743741512298584,
+      "learning_rate": 2.15e-05,
+      "loss": 0.1177,
+      "step": 215
+    },
+    {
+      "epoch": 0.10293601590829336,
+      "grad_norm": 0.39099183678627014,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.0988,
+      "step": 220
+    },
+    {
+      "epoch": 0.10527547081530003,
+      "grad_norm": 0.44669702649116516,
+      "learning_rate": 2.25e-05,
+      "loss": 0.0943,
+      "step": 225
+    },
+    {
+      "epoch": 0.1076149257223067,
+      "grad_norm": 0.4071020185947418,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.0989,
+      "step": 230
+    },
+    {
+      "epoch": 0.10995438062931337,
+      "grad_norm": 0.3775785565376282,
+      "learning_rate": 2.35e-05,
+      "loss": 0.0953,
+      "step": 235
+    },
+    {
+      "epoch": 0.11229383553632004,
+      "grad_norm": 0.417341947555542,
+      "learning_rate": 2.4e-05,
+      "loss": 0.0989,
+      "step": 240
+    },
+    {
+      "epoch": 0.11463329044332671,
+      "grad_norm": 0.43018296360969543,
+      "learning_rate": 2.45e-05,
+      "loss": 0.0933,
+      "step": 245
+    },
+    {
+      "epoch": 0.11697274535033338,
+      "grad_norm": 0.4740806818008423,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1041,
+      "step": 250
+    },
+    {
+      "epoch": 0.11931220025734005,
+      "grad_norm": 0.3639010488986969,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 0.0983,
+      "step": 255
+    },
+    {
+      "epoch": 0.1216516551643467,
+      "grad_norm": 0.46424400806427,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.1017,
+      "step": 260
+    },
+    {
+      "epoch": 0.12399111007135337,
+      "grad_norm": 0.5055582523345947,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 0.1486,
+      "step": 265
+    },
+    {
+      "epoch": 0.12633056497836004,
+      "grad_norm": 0.39390507340431213,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.0973,
+      "step": 270
+    },
+    {
+      "epoch": 0.12867001988536672,
+      "grad_norm": 0.35929396748542786,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.0957,
+      "step": 275
+    },
+    {
+      "epoch": 0.13100947479237338,
+      "grad_norm": 0.4053295850753784,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.1,
+      "step": 280
+    },
+    {
+      "epoch": 0.13334892969938003,
+      "grad_norm": 0.45273882150650024,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 0.0878,
+      "step": 285
+    },
+    {
+      "epoch": 0.1356883846063867,
+      "grad_norm": 0.3821883797645569,
+      "learning_rate": 2.9e-05,
+      "loss": 0.0881,
+      "step": 290
+    },
+    {
+      "epoch": 0.13802783951339337,
+      "grad_norm": 0.47926971316337585,
+      "learning_rate": 2.95e-05,
+      "loss": 0.1115,
+      "step": 295
+    },
+    {
+      "epoch": 0.14036729442040005,
+      "grad_norm": 0.37854063510894775,
+      "learning_rate": 3e-05,
+      "loss": 0.1025,
+      "step": 300
+    },
+    {
+      "epoch": 0.1427067493274067,
+      "grad_norm": 0.49879732728004456,
+      "learning_rate": 3.05e-05,
+      "loss": 0.0978,
+      "step": 305
+    },
+    {
+      "epoch": 0.1450462042344134,
+      "grad_norm": 0.372403085231781,
+      "learning_rate": 3.1e-05,
+      "loss": 0.0931,
+      "step": 310
+    },
+    {
+      "epoch": 0.14738565914142004,
+      "grad_norm": 0.3892665505409241,
+      "learning_rate": 3.15e-05,
+      "loss": 0.0947,
+      "step": 315
+    },
+    {
+      "epoch": 0.14972511404842673,
+      "grad_norm": 0.39579498767852783,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.0876,
+      "step": 320
+    },
+    {
+      "epoch": 0.15206456895543338,
+      "grad_norm": 0.33920520544052124,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.0941,
+      "step": 325
+    },
+    {
+      "epoch": 0.15440402386244007,
+      "grad_norm": 0.32931971549987793,
+      "learning_rate": 3.3e-05,
+      "loss": 0.0825,
+      "step": 330
+    },
+    {
+      "epoch": 0.15674347876944672,
+      "grad_norm": 0.34293580055236816,
+      "learning_rate": 3.35e-05,
+      "loss": 0.0803,
+      "step": 335
+    },
+    {
+      "epoch": 0.15908293367645338,
+      "grad_norm": 0.2816343903541565,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.081,
+      "step": 340
+    },
+    {
+      "epoch": 0.16142238858346006,
+      "grad_norm": 0.4197327792644501,
+      "learning_rate": 3.45e-05,
+      "loss": 0.0895,
+      "step": 345
+    },
+    {
+      "epoch": 0.1637618434904667,
+      "grad_norm": 0.33747074007987976,
+      "learning_rate": 3.5e-05,
+      "loss": 0.0882,
+      "step": 350
+    },
+    {
+      "epoch": 0.1661012983974734,
+      "grad_norm": 0.31165778636932373,
+      "learning_rate": 3.55e-05,
+      "loss": 0.0928,
+      "step": 355
+    },
+    {
+      "epoch": 0.16844075330448005,
+      "grad_norm": 0.3359187841415405,
+      "learning_rate": 3.6e-05,
+      "loss": 0.0939,
+      "step": 360
+    },
+    {
+      "epoch": 0.17078020821148673,
+      "grad_norm": 0.30003583431243896,
+      "learning_rate": 3.65e-05,
+      "loss": 0.0842,
+      "step": 365
+    },
+    {
+      "epoch": 0.1731196631184934,
+      "grad_norm": 0.35046064853668213,
+      "learning_rate": 3.7e-05,
+      "loss": 0.0954,
+      "step": 370
+    },
+    {
+      "epoch": 0.17545911802550007,
+      "grad_norm": 0.37043917179107666,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.0838,
+      "step": 375
+    },
+    {
+      "epoch": 0.17779857293250673,
+      "grad_norm": 0.30147868394851685,
+      "learning_rate": 3.8e-05,
+      "loss": 0.101,
+      "step": 380
+    },
+    {
+      "epoch": 0.18013802783951338,
+      "grad_norm": 0.3372386395931244,
+      "learning_rate": 3.85e-05,
+      "loss": 0.0873,
+      "step": 385
+    },
+    {
+      "epoch": 0.18247748274652006,
+      "grad_norm": 0.32464468479156494,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.09,
+      "step": 390
+    },
+    {
+      "epoch": 0.18481693765352672,
+      "grad_norm": 0.3182346224784851,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 0.0892,
+      "step": 395
+    },
+    {
+      "epoch": 0.1871563925605334,
+      "grad_norm": 0.2824024558067322,
+      "learning_rate": 4e-05,
+      "loss": 0.0832,
+      "step": 400
+    },
+    {
+      "epoch": 0.18949584746754006,
+      "grad_norm": 0.30845776200294495,
+      "learning_rate": 4.05e-05,
+      "loss": 0.093,
+      "step": 405
+    },
+    {
+      "epoch": 0.19183530237454674,
+      "grad_norm": 0.3025493323802948,
+      "learning_rate": 4.1e-05,
+      "loss": 0.0877,
+      "step": 410
+    },
+    {
+      "epoch": 0.1941747572815534,
+      "grad_norm": 0.35149937868118286,
+      "learning_rate": 4.15e-05,
+      "loss": 0.0841,
+      "step": 415
+    },
+    {
+      "epoch": 0.19651421218856008,
+      "grad_norm": 0.34748056530952454,
+      "learning_rate": 4.2e-05,
+      "loss": 0.0923,
+      "step": 420
+    },
+    {
+      "epoch": 0.19885366709556673,
+      "grad_norm": 0.3216325640678406,
+      "learning_rate": 4.25e-05,
+      "loss": 0.081,
+      "step": 425
+    },
+    {
+      "epoch": 0.2011931220025734,
+      "grad_norm": 0.2797127366065979,
+      "learning_rate": 4.3e-05,
+      "loss": 0.0808,
+      "step": 430
+    },
+    {
+      "epoch": 0.20353257690958007,
+      "grad_norm": 0.2927537262439728,
+      "learning_rate": 4.35e-05,
+      "loss": 0.1383,
+      "step": 435
+    },
+    {
+      "epoch": 0.20587203181658673,
+      "grad_norm": 0.2892308533191681,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.0836,
+      "step": 440
+    },
+    {
+      "epoch": 0.2082114867235934,
+      "grad_norm": 0.35608819127082825,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 0.0796,
+      "step": 445
+    },
+    {
+      "epoch": 0.21055094163060006,
+      "grad_norm": 0.29528915882110596,
+      "learning_rate": 4.5e-05,
+      "loss": 0.09,
+      "step": 450
+    },
+    {
+      "epoch": 0.21289039653760675,
+      "grad_norm": 0.3688681721687317,
+      "learning_rate": 4.55e-05,
+      "loss": 0.0928,
+      "step": 455
+    },
+    {
+      "epoch": 0.2152298514446134,
+      "grad_norm": 0.3129563629627228,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.0914,
+      "step": 460
+    },
+    {
+      "epoch": 0.21756930635162008,
+      "grad_norm": 0.34032294154167175,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 0.0978,
+      "step": 465
+    },
+    {
+      "epoch": 0.21990876125862674,
+      "grad_norm": 0.283371239900589,
+      "learning_rate": 4.7e-05,
+      "loss": 0.1109,
+      "step": 470
+    },
+    {
+      "epoch": 0.2222482161656334,
+      "grad_norm": 0.25409555435180664,
+      "learning_rate": 4.75e-05,
+      "loss": 0.0804,
+      "step": 475
+    },
+    {
+      "epoch": 0.22458767107264008,
+      "grad_norm": 0.24365410208702087,
+      "learning_rate": 4.8e-05,
+      "loss": 0.0737,
+      "step": 480
+    },
+    {
+      "epoch": 0.22692712597964673,
+      "grad_norm": 0.28039586544036865,
+      "learning_rate": 4.85e-05,
+      "loss": 0.0801,
+      "step": 485
+    },
+    {
+      "epoch": 0.22926658088665341,
+      "grad_norm": 0.28205862641334534,
+      "learning_rate": 4.9e-05,
+      "loss": 0.0784,
+      "step": 490
+    },
+    {
+      "epoch": 0.23160603579366007,
+      "grad_norm": 0.2688550353050232,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 0.0838,
+      "step": 495
+    },
+    {
+      "epoch": 0.23394549070066675,
+      "grad_norm": 0.31778618693351746,
+      "learning_rate": 5e-05,
+      "loss": 0.0893,
+      "step": 500
+    },
+    {
+      "epoch": 0.23394549070066675,
+      "eval_loss": 0.09720832854509354,
+      "eval_runtime": 235.4562,
+      "eval_samples_per_second": 2.964,
+      "eval_steps_per_second": 0.743,
+      "step": 500
+    },
+    {
+      "epoch": 0.2362849456076734,
+      "grad_norm": 0.27649974822998047,
+      "learning_rate": 4.999978345640206e-05,
+      "loss": 0.0818,
+      "step": 505
+    },
+    {
+      "epoch": 0.2386244005146801,
+      "grad_norm": 0.24827256798744202,
+      "learning_rate": 4.9999133829359514e-05,
+      "loss": 0.0812,
+      "step": 510
+    },
+    {
+      "epoch": 0.24096385542168675,
+      "grad_norm": 0.24930593371391296,
+      "learning_rate": 4.999805113012618e-05,
+      "loss": 0.0732,
+      "step": 515
+    },
+    {
+      "epoch": 0.2433033103286934,
+      "grad_norm": 0.2697354555130005,
+      "learning_rate": 4.999653537745817e-05,
+      "loss": 0.0843,
+      "step": 520
+    },
+    {
+      "epoch": 0.24564276523570008,
+      "grad_norm": 0.258434534072876,
+      "learning_rate": 4.999458659761363e-05,
+      "loss": 0.07,
+      "step": 525
+    },
+    {
+      "epoch": 0.24798222014270674,
+      "grad_norm": 0.2943266034126282,
+      "learning_rate": 4.9992204824352213e-05,
+      "loss": 0.0852,
+      "step": 530
+    },
+    {
+      "epoch": 0.2503216750497134,
+      "grad_norm": 0.23471561074256897,
+      "learning_rate": 4.998939009893453e-05,
+      "loss": 0.0847,
+      "step": 535
+    },
+    {
+      "epoch": 0.2526611299567201,
+      "grad_norm": 0.27082759141921997,
+      "learning_rate": 4.9986142470121453e-05,
+      "loss": 0.0727,
+      "step": 540
+    },
+    {
+      "epoch": 0.25500058486372673,
+      "grad_norm": 0.26077497005462646,
+      "learning_rate": 4.9982461994173234e-05,
+      "loss": 0.0893,
+      "step": 545
+    },
+    {
+      "epoch": 0.25734003977073344,
+      "grad_norm": 0.2686660587787628,
+      "learning_rate": 4.9978348734848556e-05,
+      "loss": 0.0764,
+      "step": 550
+    },
+    {
+      "epoch": 0.2596794946777401,
+      "grad_norm": 0.2592894434928894,
+      "learning_rate": 4.9973802763403424e-05,
+      "loss": 0.1747,
+      "step": 555
+    },
+    {
+      "epoch": 0.26201894958474675,
+      "grad_norm": 0.26700106263160706,
+      "learning_rate": 4.996882415858991e-05,
+      "loss": 0.0808,
+      "step": 560
+    },
+    {
+      "epoch": 0.2643584044917534,
+      "grad_norm": 0.29679951071739197,
+      "learning_rate": 4.996341300665481e-05,
+      "loss": 0.0847,
+      "step": 565
+    },
+    {
+      "epoch": 0.26669785939876006,
+      "grad_norm": 0.29322996735572815,
+      "learning_rate": 4.995756940133817e-05,
+      "loss": 0.0884,
+      "step": 570
+    },
+    {
+      "epoch": 0.2690373143057668,
+      "grad_norm": 0.24166248738765717,
+      "learning_rate": 4.995129344387158e-05,
+      "loss": 0.0784,
+      "step": 575
+    },
+    {
+      "epoch": 0.2713767692127734,
+      "grad_norm": 0.23448926210403442,
+      "learning_rate": 4.9944585242976546e-05,
+      "loss": 0.146,
+      "step": 580
+    },
+    {
+      "epoch": 0.2737162241197801,
+      "grad_norm": 0.2622728645801544,
+      "learning_rate": 4.99374449148625e-05,
+      "loss": 0.0894,
+      "step": 585
+    },
+    {
+      "epoch": 0.27605567902678674,
+      "grad_norm": 0.23316816985607147,
+      "learning_rate": 4.992987258322481e-05,
+      "loss": 0.0821,
+      "step": 590
+    },
+    {
+      "epoch": 0.27839513393379345,
+      "grad_norm": 0.29072973132133484,
+      "learning_rate": 4.992186837924268e-05,
+      "loss": 0.098,
+      "step": 595
+    },
+    {
+      "epoch": 0.2807345888408001,
+      "grad_norm": 0.20675267279148102,
+      "learning_rate": 4.9913432441576845e-05,
+      "loss": 0.0776,
+      "step": 600
+    },
+    {
+      "epoch": 0.28307404374780676,
+      "grad_norm": 0.26899468898773193,
+      "learning_rate": 4.9904564916367166e-05,
+      "loss": 0.0862,
+      "step": 605
+    },
+    {
+      "epoch": 0.2854134986548134,
+      "grad_norm": 0.22569362819194794,
+      "learning_rate": 4.989526595723012e-05,
+      "loss": 0.0921,
+      "step": 610
+    },
+    {
+      "epoch": 0.28775295356182007,
+      "grad_norm": 0.2874622046947479,
+      "learning_rate": 4.988553572525609e-05,
+      "loss": 0.0811,
+      "step": 615
+    },
+    {
+      "epoch": 0.2900924084688268,
+      "grad_norm": 0.2119138389825821,
+      "learning_rate": 4.987537438900664e-05,
+      "loss": 0.0711,
+      "step": 620
+    },
+    {
+      "epoch": 0.29243186337583343,
+      "grad_norm": 0.2240922749042511,
+      "learning_rate": 4.986478212451157e-05,
+      "loss": 0.0802,
+      "step": 625
+    },
+    {
+      "epoch": 0.2947713182828401,
+      "grad_norm": 0.24686302244663239,
+      "learning_rate": 4.985375911526582e-05,
+      "loss": 0.0825,
+      "step": 630
+    },
+    {
+      "epoch": 0.29711077318984674,
+      "grad_norm": 0.22553415596485138,
+      "learning_rate": 4.9842305552226365e-05,
+      "loss": 0.0676,
+      "step": 635
+    },
+    {
+      "epoch": 0.29945022809685345,
+      "grad_norm": 0.24262481927871704,
+      "learning_rate": 4.9830421633808876e-05,
+      "loss": 0.0752,
+      "step": 640
+    },
+    {
+      "epoch": 0.3017896830038601,
+      "grad_norm": 0.24724553525447845,
+      "learning_rate": 4.981810756588426e-05,
+      "loss": 0.0823,
+      "step": 645
+    },
+    {
+      "epoch": 0.30412913791086676,
+      "grad_norm": 0.2152285873889923,
+      "learning_rate": 4.980536356177513e-05,
+      "loss": 0.0795,
+      "step": 650
+    },
+    {
+      "epoch": 0.3064685928178734,
+      "grad_norm": 0.2218402624130249,
+      "learning_rate": 4.9792189842252074e-05,
+      "loss": 0.0711,
+      "step": 655
+    },
+    {
+      "epoch": 0.30880804772488013,
+      "grad_norm": 0.24227651953697205,
+      "learning_rate": 4.977858663552987e-05,
+      "loss": 0.0756,
+      "step": 660
+    },
+    {
+      "epoch": 0.3111475026318868,
+      "grad_norm": 0.22776497900485992,
+      "learning_rate": 4.9764554177263507e-05,
+      "loss": 0.0742,
+      "step": 665
+    },
+    {
+      "epoch": 0.31348695753889344,
+      "grad_norm": 0.20029255747795105,
+      "learning_rate": 4.975009271054409e-05,
+      "loss": 0.0698,
+      "step": 670
+    },
+    {
+      "epoch": 0.3158264124459001,
+      "grad_norm": 0.21585966646671295,
+      "learning_rate": 4.973520248589469e-05,
+      "loss": 0.0858,
+      "step": 675
+    },
+    {
+      "epoch": 0.31816586735290675,
+      "grad_norm": 0.23440401256084442,
+      "learning_rate": 4.9719883761265906e-05,
+      "loss": 0.0818,
+      "step": 680
+    },
+    {
+      "epoch": 0.32050532225991346,
+      "grad_norm": 0.22107180953025818,
+      "learning_rate": 4.9704136802031485e-05,
+      "loss": 0.0671,
+      "step": 685
+    },
+    {
+      "epoch": 0.3228447771669201,
+      "grad_norm": 0.19328860938549042,
+      "learning_rate": 4.968796188098369e-05,
+      "loss": 0.0725,
+      "step": 690
+    },
+    {
+      "epoch": 0.32518423207392677,
+      "grad_norm": 0.2299336940050125,
+      "learning_rate": 4.967135927832856e-05,
+      "loss": 0.0852,
+      "step": 695
+    },
+    {
+      "epoch": 0.3275236869809334,
+      "grad_norm": 0.24605049192905426,
+      "learning_rate": 4.9654329281681094e-05,
+      "loss": 0.0754,
+      "step": 700
+    },
+    {
+      "epoch": 0.32986314188794014,
+      "grad_norm": 0.23175948858261108,
+      "learning_rate": 4.9636872186060215e-05,
+      "loss": 0.0961,
+      "step": 705
+    },
+    {
+      "epoch": 0.3322025967949468,
+      "grad_norm": 0.2416304647922516,
+      "learning_rate": 4.961898829388372e-05,
+      "loss": 0.1093,
+      "step": 710
+    },
+    {
+      "epoch": 0.33454205170195345,
+      "grad_norm": 0.20699331164360046,
+      "learning_rate": 4.960067791496299e-05,
+      "loss": 0.0793,
+      "step": 715
+    },
+    {
+      "epoch": 0.3368815066089601,
+      "grad_norm": 0.22395586967468262,
+      "learning_rate": 4.958194136649765e-05,
+      "loss": 0.0805,
+      "step": 720
+    },
+    {
+      "epoch": 0.33922096151596676,
+      "grad_norm": 0.21491140127182007,
+      "learning_rate": 4.956277897307008e-05,
+      "loss": 0.073,
+      "step": 725
+    },
+    {
+      "epoch": 0.34156041642297347,
+      "grad_norm": 0.23053660988807678,
+      "learning_rate": 4.954319106663976e-05,
+      "loss": 0.0769,
+      "step": 730
+    },
+    {
+      "epoch": 0.3438998713299801,
+      "grad_norm": 0.21306245028972626,
+      "learning_rate": 4.952317798653755e-05,
+      "loss": 0.075,
+      "step": 735
+    },
+    {
+      "epoch": 0.3462393262369868,
+      "grad_norm": 0.1973639279603958,
+      "learning_rate": 4.95027400794598e-05,
+      "loss": 0.0697,
+      "step": 740
+    },
+    {
+      "epoch": 0.34857878114399343,
+      "grad_norm": 0.2036924660205841,
+      "learning_rate": 4.948187769946234e-05,
+      "loss": 0.08,
+      "step": 745
+    },
+    {
+      "epoch": 0.35091823605100014,
+      "grad_norm": 0.2153988480567932,
+      "learning_rate": 4.946059120795437e-05,
+      "loss": 0.0963,
+      "step": 750
+    },
+    {
+      "epoch": 0.3532576909580068,
+      "grad_norm": 0.20555338263511658,
+      "learning_rate": 4.943888097369216e-05,
+      "loss": 0.0913,
+      "step": 755
+    },
+    {
+      "epoch": 0.35559714586501345,
+      "grad_norm": 0.26464715600013733,
+      "learning_rate": 4.941674737277268e-05,
+      "loss": 0.0845,
+      "step": 760
+    },
+    {
+      "epoch": 0.3579366007720201,
+      "grad_norm": 0.19657152891159058,
+      "learning_rate": 4.9394190788627106e-05,
+      "loss": 0.0756,
+      "step": 765
+    },
+    {
+      "epoch": 0.36027605567902676,
+      "grad_norm": 0.21668066084384918,
+      "learning_rate": 4.937121161201415e-05,
+      "loss": 0.0803,
+      "step": 770
+    },
+    {
+      "epoch": 0.3626155105860335,
+      "grad_norm": 0.24815738201141357,
+      "learning_rate": 4.9347810241013294e-05,
+      "loss": 0.0823,
+      "step": 775
+    },
+    {
+      "epoch": 0.36495496549304013,
+      "grad_norm": 0.20182116329669952,
+      "learning_rate": 4.932398708101791e-05,
+      "loss": 0.0703,
+      "step": 780
+    },
+    {
+      "epoch": 0.3672944204000468,
+      "grad_norm": 0.2374674379825592,
+      "learning_rate": 4.92997425447282e-05,
+      "loss": 0.084,
+      "step": 785
+    },
+    {
+      "epoch": 0.36963387530705344,
+      "grad_norm": 0.2246353179216385,
+      "learning_rate": 4.927507705214412e-05,
+      "loss": 0.0868,
+      "step": 790
+    },
+    {
+      "epoch": 0.37197333021406015,
+      "grad_norm": 0.21759703755378723,
+      "learning_rate": 4.924999103055802e-05,
+      "loss": 0.0754,
+      "step": 795
+    },
+    {
+      "epoch": 0.3743127851210668,
+      "grad_norm": 0.20431989431381226,
+      "learning_rate": 4.922448491454729e-05,
+      "loss": 0.0803,
+      "step": 800
+    },
+    {
+      "epoch": 0.37665224002807346,
+      "grad_norm": 0.20232437551021576,
+      "learning_rate": 4.9198559145966824e-05,
+      "loss": 0.0751,
+      "step": 805
+    },
+    {
+      "epoch": 0.3789916949350801,
+      "grad_norm": 0.1968168169260025,
+      "learning_rate": 4.917221417394134e-05,
+      "loss": 0.0798,
+      "step": 810
+    },
+    {
+      "epoch": 0.38133114984208677,
+      "grad_norm": 0.2001713663339615,
+      "learning_rate": 4.914545045485767e-05,
+      "loss": 0.0776,
+      "step": 815
+    },
+    {
+      "epoch": 0.3836706047490935,
+      "grad_norm": 0.19992566108703613,
+      "learning_rate": 4.911826845235676e-05,
+      "loss": 0.0767,
+      "step": 820
+    },
+    {
+      "epoch": 0.38601005965610014,
+      "grad_norm": 0.21567636728286743,
+      "learning_rate": 4.9090668637325696e-05,
+      "loss": 0.0752,
+      "step": 825
+    },
+    {
+      "epoch": 0.3883495145631068,
+      "grad_norm": 0.19452187418937683,
+      "learning_rate": 4.906265148788954e-05,
+      "loss": 0.0722,
+      "step": 830
+    },
+    {
+      "epoch": 0.39068896947011345,
+      "grad_norm": 0.23549924790859222,
+      "learning_rate": 4.9034217489403045e-05,
+      "loss": 0.0764,
+      "step": 835
+    },
+    {
+      "epoch": 0.39302842437712016,
+      "grad_norm": 0.2083604633808136,
+      "learning_rate": 4.9005367134442235e-05,
+      "loss": 0.0767,
+      "step": 840
+    },
+    {
+      "epoch": 0.3953678792841268,
+      "grad_norm": 0.2386019378900528,
+      "learning_rate": 4.8976100922795884e-05,
+      "loss": 0.0815,
+      "step": 845
+    },
+    {
+      "epoch": 0.39770733419113347,
+      "grad_norm": 0.19933900237083435,
+      "learning_rate": 4.894641936145686e-05,
+      "loss": 0.0935,
+      "step": 850
+    },
+    {
+      "epoch": 0.4000467890981401,
+      "grad_norm": 0.16899636387825012,
+      "learning_rate": 4.8916322964613316e-05,
+      "loss": 0.0767,
+      "step": 855
+    },
+    {
+      "epoch": 0.4023862440051468,
+      "grad_norm": 0.20377202332019806,
+      "learning_rate": 4.888581225363982e-05,
+      "loss": 0.0771,
+      "step": 860
+    },
+    {
+      "epoch": 0.4047256989121535,
+      "grad_norm": 0.2286689430475235,
+      "learning_rate": 4.885488775708831e-05,
+      "loss": 0.0798,
+      "step": 865
+    },
+    {
+      "epoch": 0.40706515381916014,
+      "grad_norm": 0.2132563591003418,
+      "learning_rate": 4.882355001067892e-05,
+      "loss": 0.0723,
+      "step": 870
+    },
+    {
+      "epoch": 0.4094046087261668,
+      "grad_norm": 0.21681547164916992,
+      "learning_rate": 4.879179955729072e-05,
+      "loss": 0.0777,
+      "step": 875
+    },
+    {
+      "epoch": 0.41174406363317345,
+      "grad_norm": 0.20257344841957092,
+      "learning_rate": 4.87596369469523e-05,
+      "loss": 0.0777,
+      "step": 880
+    },
+    {
+      "epoch": 0.41408351854018016,
+      "grad_norm": 0.18595652282238007,
+      "learning_rate": 4.872706273683225e-05,
+      "loss": 0.0648,
+      "step": 885
+    },
+    {
+      "epoch": 0.4164229734471868,
+      "grad_norm": 0.1963788866996765,
+      "learning_rate": 4.869407749122951e-05,
+      "loss": 0.0688,
+      "step": 890
+    },
+    {
+      "epoch": 0.41876242835419347,
+      "grad_norm": 0.22948507964611053,
+      "learning_rate": 4.8660681781563576e-05,
+      "loss": 0.0715,
+      "step": 895
+    },
+    {
+      "epoch": 0.4211018832612001,
+      "grad_norm": 0.16155906021595,
+      "learning_rate": 4.8626876186364624e-05,
+      "loss": 0.0651,
+      "step": 900
+    },
+    {
+      "epoch": 0.4234413381682068,
+      "grad_norm": 0.193013533949852,
+      "learning_rate": 4.859266129126345e-05,
+      "loss": 0.0754,
+      "step": 905
+    },
+    {
+      "epoch": 0.4257807930752135,
+      "grad_norm": 0.21157501637935638,
+      "learning_rate": 4.85580376889814e-05,
+      "loss": 0.1229,
+      "step": 910
+    },
+    {
+      "epoch": 0.42812024798222015,
+      "grad_norm": 0.2193852812051773,
+      "learning_rate": 4.8523005979320015e-05,
+      "loss": 0.0755,
+      "step": 915
+    },
+    {
+      "epoch": 0.4304597028892268,
+      "grad_norm": 0.20411770045757294,
+      "learning_rate": 4.848756676915069e-05,
+      "loss": 0.0848,
+      "step": 920
+    },
+    {
+      "epoch": 0.43279915779623346,
+      "grad_norm": 0.23284126818180084,
+      "learning_rate": 4.845172067240415e-05,
+      "loss": 0.0764,
+      "step": 925
+    },
+    {
+      "epoch": 0.43513861270324017,
+      "grad_norm": 0.21067671477794647,
+      "learning_rate": 4.841546831005982e-05,
+      "loss": 0.0821,
+      "step": 930
+    },
+    {
+      "epoch": 0.4374780676102468,
+      "grad_norm": 0.1912618726491928,
+      "learning_rate": 4.837881031013506e-05,
+      "loss": 0.0755,
+      "step": 935
+    },
+    {
+      "epoch": 0.4398175225172535,
+      "grad_norm": 0.1677297055721283,
+      "learning_rate": 4.8341747307674276e-05,
+      "loss": 0.0695,
+      "step": 940
+    },
+    {
+      "epoch": 0.44215697742426013,
+      "grad_norm": 0.22734542191028595,
+      "learning_rate": 4.8304279944737954e-05,
+      "loss": 0.086,
+      "step": 945
+    },
+    {
+      "epoch": 0.4444964323312668,
+      "grad_norm": 0.1839355230331421,
+      "learning_rate": 4.8266408870391484e-05,
+      "loss": 0.0797,
+      "step": 950
+    },
+    {
+      "epoch": 0.4468358872382735,
+      "grad_norm": 0.20233304798603058,
+      "learning_rate": 4.822813474069398e-05,
+      "loss": 0.084,
+      "step": 955
+    },
+    {
+      "epoch": 0.44917534214528015,
+      "grad_norm": 0.20918181538581848,
+      "learning_rate": 4.8189458218686846e-05,
+      "loss": 0.0754,
+      "step": 960
+    },
+    {
+      "epoch": 0.4515147970522868,
+      "grad_norm": 0.2046424299478531,
+      "learning_rate": 4.815037997438234e-05,
+      "loss": 0.0762,
+      "step": 965
+    },
+    {
+      "epoch": 0.45385425195929346,
+      "grad_norm": 0.20791682600975037,
+      "learning_rate": 4.811090068475197e-05,
+      "loss": 0.0686,
+      "step": 970
+    },
+    {
+      "epoch": 0.4561937068663002,
+      "grad_norm": 0.1891777068376541,
+      "learning_rate": 4.807102103371472e-05,
+      "loss": 0.0818,
+      "step": 975
+    },
+    {
+      "epoch": 0.45853316177330683,
+      "grad_norm": 0.1632051318883896,
+      "learning_rate": 4.803074171212524e-05,
+      "loss": 0.0693,
+      "step": 980
+    },
+    {
+      "epoch": 0.4608726166803135,
+      "grad_norm": 0.18187515437602997,
+      "learning_rate": 4.799006341776185e-05,
+      "loss": 0.0642,
+      "step": 985
+    },
+    {
+      "epoch": 0.46321207158732014,
+      "grad_norm": 0.20138341188430786,
+      "learning_rate": 4.7948986855314523e-05,
+      "loss": 0.0759,
+      "step": 990
+    },
+    {
+      "epoch": 0.4655515264943268,
+      "grad_norm": 0.19479338824748993,
+      "learning_rate": 4.790751273637256e-05,
+      "loss": 0.0761,
+      "step": 995
+    },
+    {
+      "epoch": 0.4678909814013335,
+      "grad_norm": 0.18999288976192474,
+      "learning_rate": 4.7865641779412366e-05,
+      "loss": 0.0753,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4678909814013335,
+      "eval_loss": 0.08787659555673599,
+      "eval_runtime": 236.9363,
+      "eval_samples_per_second": 2.946,
+      "eval_steps_per_second": 0.739,
+      "step": 1000
+    },
+    {
+      "epoch": 0.47023043630834016,
+      "grad_norm": 0.22196418046951294,
+      "learning_rate": 4.7823374709784955e-05,
+      "loss": 0.0731,
+      "step": 1005
+    },
+    {
+      "epoch": 0.4725698912153468,
+      "grad_norm": 0.17047221958637238,
+      "learning_rate": 4.77807122597034e-05,
+      "loss": 0.082,
+      "step": 1010
+    },
+    {
+      "epoch": 0.47490934612235347,
+      "grad_norm": 0.23016561567783356,
+      "learning_rate": 4.7737655168230114e-05,
+      "loss": 0.0775,
+      "step": 1015
+    },
+    {
+      "epoch": 0.4772488010293602,
+      "grad_norm": 0.21696537733078003,
+      "learning_rate": 4.769420418126412e-05,
+      "loss": 0.0805,
+      "step": 1020
+    },
+    {
+      "epoch": 0.47958825593636684,
+      "grad_norm": 0.19352155923843384,
+      "learning_rate": 4.7650360051528054e-05,
+      "loss": 0.0754,
+      "step": 1025
+    },
+    {
+      "epoch": 0.4819277108433735,
+      "grad_norm": 0.1966516673564911,
+      "learning_rate": 4.760612353855517e-05,
+      "loss": 0.068,
+      "step": 1030
+    },
+    {
+      "epoch": 0.48426716575038015,
+      "grad_norm": 0.22759053111076355,
+      "learning_rate": 4.7561495408676146e-05,
+      "loss": 0.073,
+      "step": 1035
+    },
+    {
+      "epoch": 0.4866066206573868,
+      "grad_norm": 0.17850154638290405,
+      "learning_rate": 4.7516476435005865e-05,
+      "loss": 0.074,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4889460755643935,
+      "grad_norm": 0.19016915559768677,
+      "learning_rate": 4.7471067397429956e-05,
+      "loss": 0.0734,
+      "step": 1045
+    },
+    {
+      "epoch": 0.49128553047140017,
+      "grad_norm": 0.19601960480213165,
+      "learning_rate": 4.742526908259134e-05,
+      "loss": 0.0763,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4936249853784068,
+      "grad_norm": 0.17694705724716187,
+      "learning_rate": 4.7379082283876566e-05,
+      "loss": 0.0703,
+      "step": 1055
+    },
+    {
+      "epoch": 0.4959644402854135,
+      "grad_norm": 0.18579581379890442,
+      "learning_rate": 4.733250780140206e-05,
+      "loss": 0.0863,
+      "step": 1060
+    },
+    {
+      "epoch": 0.4983038951924202,
+      "grad_norm": 0.18680576980113983,
+      "learning_rate": 4.728554644200034e-05,
+      "loss": 0.1042,
+      "step": 1065
+    },
+    {
+      "epoch": 0.5006433500994268,
+      "grad_norm": 0.1797022670507431,
+      "learning_rate": 4.723819901920591e-05,
+      "loss": 0.0743,
+      "step": 1070
+    },
+    {
+      "epoch": 0.5029828050064336,
+      "grad_norm": 0.20039916038513184,
+      "learning_rate": 4.719046635324129e-05,
+      "loss": 0.0707,
+      "step": 1075
+    },
+    {
+      "epoch": 0.5053222599134402,
+      "grad_norm": 0.19813776016235352,
+      "learning_rate": 4.7142349271002735e-05,
+      "loss": 0.0738,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5076617148204469,
+      "grad_norm": 0.18406710028648376,
+      "learning_rate": 4.709384860604593e-05,
+      "loss": 0.0724,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5100011697274535,
+      "grad_norm": 0.20730999112129211,
+      "learning_rate": 4.704496519857156e-05,
+      "loss": 0.0738,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5123406246344602,
+      "grad_norm": 0.2113994061946869,
+      "learning_rate": 4.699569989541074e-05,
+      "loss": 0.0699,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5146800795414669,
+      "grad_norm": 0.1848883181810379,
+      "learning_rate": 4.694605355001034e-05,
+      "loss": 0.072,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5170195344484735,
+      "grad_norm": 0.19623607397079468,
+      "learning_rate": 4.689602702241823e-05,
+      "loss": 0.0761,
+      "step": 1105
+    },
+    {
+      "epoch": 0.5193589893554802,
+      "grad_norm": 0.18376454710960388,
+      "learning_rate": 4.684562117926835e-05,
+      "loss": 0.0793,
+      "step": 1110
+    },
+    {
+      "epoch": 0.5216984442624868,
+      "grad_norm": 0.18506892025470734,
+      "learning_rate": 4.679483689376571e-05,
+      "loss": 0.0731,
+      "step": 1115
+    },
+    {
+      "epoch": 0.5240378991694935,
+      "grad_norm": 0.1819106787443161,
+      "learning_rate": 4.674367504567127e-05,
+      "loss": 0.0795,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5263773540765002,
+      "grad_norm": 0.19489003717899323,
+      "learning_rate": 4.669213652128667e-05,
+      "loss": 0.0716,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5287168089835068,
+      "grad_norm": 0.16565723717212677,
+      "learning_rate": 4.664022221343892e-05,
+      "loss": 0.0717,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5310562638905135,
+      "grad_norm": 0.18749143183231354,
+      "learning_rate": 4.658793302146489e-05,
+      "loss": 0.0756,
+      "step": 1135
+    },
+    {
+      "epoch": 0.5333957187975201,
+      "grad_norm": 0.2000792771577835,
+      "learning_rate": 4.653526985119577e-05,
+      "loss": 0.0789,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5357351737045268,
+      "grad_norm": 0.17934930324554443,
+      "learning_rate": 4.648223361494135e-05,
+      "loss": 0.0823,
+      "step": 1145
+    },
+    {
+      "epoch": 0.5380746286115335,
+      "grad_norm": 0.204596146941185,
+      "learning_rate": 4.642882523147422e-05,
+      "loss": 0.0804,
+      "step": 1150
+    },
+    {
+      "epoch": 0.5404140835185401,
+      "grad_norm": 0.17674653232097626,
+      "learning_rate": 4.637504562601386e-05,
+      "loss": 0.0699,
+      "step": 1155
+    },
+    {
+      "epoch": 0.5427535384255469,
+      "grad_norm": 0.17949679493904114,
+      "learning_rate": 4.6320895730210616e-05,
+      "loss": 0.0717,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5450929933325536,
+      "grad_norm": 0.1828782856464386,
+      "learning_rate": 4.626637648212955e-05,
+      "loss": 0.0816,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5474324482395602,
+      "grad_norm": 0.1803893744945526,
+      "learning_rate": 4.6211488826234186e-05,
+      "loss": 0.0766,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5497719031465669,
+      "grad_norm": 0.1853674352169037,
+      "learning_rate": 4.615623371337016e-05,
+      "loss": 0.0728,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5521113580535735,
+      "grad_norm": 0.20650415122509003,
+      "learning_rate": 4.6100612100748765e-05,
+      "loss": 0.073,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5544508129605802,
+      "grad_norm": 0.18589161336421967,
+      "learning_rate": 4.604462495193031e-05,
+      "loss": 0.0803,
+      "step": 1185
+    },
+    {
+      "epoch": 0.5567902678675869,
+      "grad_norm": 0.16948671638965607,
+      "learning_rate": 4.59882732368075e-05,
+      "loss": 0.0728,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5591297227745935,
+      "grad_norm": 0.17474383115768433,
+      "learning_rate": 4.593155793158859e-05,
+      "loss": 0.0697,
+      "step": 1195
+    },
+    {
+      "epoch": 0.5614691776816002,
+      "grad_norm": 0.1744118481874466,
+      "learning_rate": 4.5874480018780456e-05,
+      "loss": 0.0771,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5638086325886068,
+      "grad_norm": 0.17954926192760468,
+      "learning_rate": 4.581704048717166e-05,
+      "loss": 0.0722,
+      "step": 1205
+    },
+    {
+      "epoch": 0.5661480874956135,
+      "grad_norm": 0.19267572462558746,
+      "learning_rate": 4.57592403318152e-05,
+      "loss": 0.0739,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5684875424026202,
+      "grad_norm": 0.1731371432542801,
+      "learning_rate": 4.570108055401138e-05,
+      "loss": 0.076,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5708269973096268,
+      "grad_norm": 0.18417227268218994,
+      "learning_rate": 4.5642562161290406e-05,
+      "loss": 0.0708,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5731664522166335,
+      "grad_norm": 0.19248449802398682,
+      "learning_rate": 4.558368616739493e-05,
+      "loss": 0.0775,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5755059071236401,
+      "grad_norm": 0.20043963193893433,
+      "learning_rate": 4.552445359226252e-05,
+      "loss": 0.0758,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5778453620306468,
+      "grad_norm": 0.20923319458961487,
+      "learning_rate": 4.546486546200798e-05,
+      "loss": 0.0648,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5801848169376536,
+      "grad_norm": 0.19847044348716736,
+      "learning_rate": 4.540492280890555e-05,
+      "loss": 0.0684,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5825242718446602,
+      "grad_norm": 0.17543412744998932,
+      "learning_rate": 4.534462667137105e-05,
+      "loss": 0.0746,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5848637267516669,
+      "grad_norm": 0.1842990517616272,
+      "learning_rate": 4.528397809394389e-05,
+      "loss": 0.0771,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5872031816586736,
+      "grad_norm": 0.15888415277004242,
+      "learning_rate": 4.522297812726897e-05,
+      "loss": 0.0654,
+      "step": 1255
+    },
+    {
+      "epoch": 0.5895426365656802,
+      "grad_norm": 0.20020891726016998,
+      "learning_rate": 4.516162782807845e-05,
+      "loss": 0.0759,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5918820914726869,
+      "grad_norm": 0.164358988404274,
+      "learning_rate": 4.5099928259173516e-05,
+      "loss": 0.0666,
+      "step": 1265
+    },
+    {
+      "epoch": 0.5942215463796935,
+      "grad_norm": 0.16114068031311035,
+      "learning_rate": 4.503788048940589e-05,
+      "loss": 0.0677,
+      "step": 1270
+    },
+    {
+      "epoch": 0.5965610012867002,
+      "grad_norm": 0.18885543942451477,
+      "learning_rate": 4.497548559365935e-05,
+      "loss": 0.0697,
+      "step": 1275
+    },
+    {
+      "epoch": 0.5989004561937069,
+      "grad_norm": 0.16143983602523804,
+      "learning_rate": 4.4912744652831116e-05,
+      "loss": 0.072,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6012399111007135,
+      "grad_norm": 0.16090819239616394,
+      "learning_rate": 4.4849658753813126e-05,
+      "loss": 0.068,
+      "step": 1285
+    },
+    {
+      "epoch": 0.6035793660077202,
+      "grad_norm": 0.19243429601192474,
+      "learning_rate": 4.4786228989473164e-05,
+      "loss": 0.0744,
+      "step": 1290
+    },
+    {
+      "epoch": 0.6059188209147268,
+      "grad_norm": 0.17546018958091736,
+      "learning_rate": 4.4722456458635995e-05,
+      "loss": 0.0745,
+      "step": 1295
+    },
+    {
+      "epoch": 0.6082582758217335,
+      "grad_norm": 0.1936197578907013,
+      "learning_rate": 4.465834226606428e-05,
+      "loss": 0.0769,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6105977307287402,
+      "grad_norm": 0.16884423792362213,
+      "learning_rate": 4.459388752243945e-05,
+      "loss": 0.0809,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6129371856357468,
+      "grad_norm": 0.20882828533649445,
+      "learning_rate": 4.452909334434247e-05,
+      "loss": 0.08,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6152766405427536,
+      "grad_norm": 0.1857665777206421,
+      "learning_rate": 4.4463960854234506e-05,
+      "loss": 0.0668,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6176160954497603,
+      "grad_norm": 0.17508915066719055,
+      "learning_rate": 4.4398491180437446e-05,
+      "loss": 0.0775,
+      "step": 1320
+    },
+    {
+      "epoch": 0.6199555503567669,
+      "grad_norm": 0.16894185543060303,
+      "learning_rate": 4.43326854571144e-05,
+      "loss": 0.0714,
+      "step": 1325
+    },
+    {
+      "epoch": 0.6222950052637736,
+      "grad_norm": 0.19014866650104523,
+      "learning_rate": 4.426654482425e-05,
+      "loss": 0.0789,
+      "step": 1330
+    },
+    {
+      "epoch": 0.6246344601707802,
+      "grad_norm": 0.15668885409832,
+      "learning_rate": 4.420007042763071e-05,
+      "loss": 0.0641,
+      "step": 1335
+    },
+    {
+      "epoch": 0.6269739150777869,
+      "grad_norm": 0.15578554570674896,
+      "learning_rate": 4.413326341882492e-05,
+      "loss": 0.0665,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6293133699847936,
+      "grad_norm": 0.20437653362751007,
+      "learning_rate": 4.4066124955163046e-05,
+      "loss": 0.0784,
+      "step": 1345
+    },
+    {
+      "epoch": 0.6316528248918002,
+      "grad_norm": 0.16404947638511658,
+      "learning_rate": 4.3998656199717435e-05,
+      "loss": 0.0675,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6339922797988069,
+      "grad_norm": 0.147497296333313,
+      "learning_rate": 4.393085832128226e-05,
+      "loss": 0.0711,
+      "step": 1355
+    },
+    {
+      "epoch": 0.6363317347058135,
+      "grad_norm": 0.17336557805538177,
+      "learning_rate": 4.386273249435324e-05,
+      "loss": 0.0766,
+      "step": 1360
+    },
+    {
+      "epoch": 0.6386711896128202,
+      "grad_norm": 0.18283897638320923,
+      "learning_rate": 4.3794279899107304e-05,
+      "loss": 0.0746,
+      "step": 1365
+    },
+    {
+      "epoch": 0.6410106445198269,
+      "grad_norm": 0.16502517461776733,
+      "learning_rate": 4.3725501721382165e-05,
+      "loss": 0.0638,
+      "step": 1370
+    },
+    {
+      "epoch": 0.6433500994268335,
+      "grad_norm": 0.16824820637702942,
+      "learning_rate": 4.3656399152655735e-05,
+      "loss": 0.0698,
+      "step": 1375
+    },
+    {
+      "epoch": 0.6456895543338402,
+      "grad_norm": 0.17634496092796326,
+      "learning_rate": 4.358697339002553e-05,
+      "loss": 0.0952,
+      "step": 1380
+    },
+    {
+      "epoch": 0.6480290092408468,
+      "grad_norm": 0.1962941437959671,
+      "learning_rate": 4.3517225636187906e-05,
+      "loss": 0.0728,
+      "step": 1385
+    },
+    {
+      "epoch": 0.6503684641478535,
+      "grad_norm": 0.1725791096687317,
+      "learning_rate": 4.344715709941722e-05,
+      "loss": 0.0721,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6527079190548603,
+      "grad_norm": 0.1703636348247528,
+      "learning_rate": 4.337676899354493e-05,
+      "loss": 0.072,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6550473739618669,
+      "grad_norm": 0.16488994657993317,
+      "learning_rate": 4.330606253793851e-05,
+      "loss": 0.072,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6573868288688736,
+      "grad_norm": 0.16995395720005035,
+      "learning_rate": 4.32350389574804e-05,
+      "loss": 0.0747,
+      "step": 1405
+    },
+    {
+      "epoch": 0.6597262837758803,
+      "grad_norm": 0.17242836952209473,
+      "learning_rate": 4.316369948254674e-05,
+      "loss": 0.0698,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6620657386828869,
+      "grad_norm": 0.16680659353733063,
+      "learning_rate": 4.3092045348986034e-05,
+      "loss": 0.0683,
+      "step": 1415
+    },
+    {
+      "epoch": 0.6644051935898936,
+      "grad_norm": 0.18627899885177612,
+      "learning_rate": 4.302007779809781e-05,
+      "loss": 0.0672,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6667446484969002,
+      "grad_norm": 0.14819131791591644,
+      "learning_rate": 4.294779807661105e-05,
+      "loss": 0.0683,
+      "step": 1425
+    },
+    {
+      "epoch": 0.6690841034039069,
+      "grad_norm": 0.18117517232894897,
+      "learning_rate": 4.287520743666263e-05,
+      "loss": 0.0728,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6714235583109136,
+      "grad_norm": 0.17058628797531128,
+      "learning_rate": 4.280230713577564e-05,
+      "loss": 0.077,
+      "step": 1435
+    },
+    {
+      "epoch": 0.6737630132179202,
+      "grad_norm": 0.15503695607185364,
+      "learning_rate": 4.2729098436837536e-05,
+      "loss": 0.0739,
+      "step": 1440
+    },
+    {
+      "epoch": 0.6761024681249269,
+      "grad_norm": 0.1836390346288681,
+      "learning_rate": 4.2655582608078315e-05,
+      "loss": 0.079,
+      "step": 1445
+    },
+    {
+      "epoch": 0.6784419230319335,
+      "grad_norm": 0.15766695141792297,
+      "learning_rate": 4.2581760923048554e-05,
+      "loss": 0.0653,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6807813779389402,
+      "grad_norm": 0.18035438656806946,
+      "learning_rate": 4.2507634660597315e-05,
+      "loss": 0.0793,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6831208328459469,
+      "grad_norm": 0.1778097152709961,
+      "learning_rate": 4.243320510485001e-05,
+      "loss": 0.0689,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6854602877529535,
+      "grad_norm": 0.1773742139339447,
+      "learning_rate": 4.235847354518614e-05,
+      "loss": 0.0726,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6877997426599602,
+      "grad_norm": 0.16845591366291046,
+      "learning_rate": 4.228344127621695e-05,
+      "loss": 0.0691,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6901391975669668,
+      "grad_norm": 0.14897610247135162,
+      "learning_rate": 4.220810959776307e-05,
+      "loss": 0.0684,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6924786524739736,
+      "grad_norm": 0.17343005537986755,
+      "learning_rate": 4.213247981483189e-05,
+      "loss": 0.0624,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6948181073809803,
+      "grad_norm": 0.19513456523418427,
+      "learning_rate": 4.205655323759505e-05,
+      "loss": 0.0751,
+      "step": 1485
+    },
+    {
+      "epoch": 0.6971575622879869,
+      "grad_norm": 0.17424218356609344,
+      "learning_rate": 4.1980331181365685e-05,
+      "loss": 0.0704,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6994970171949936,
+      "grad_norm": 0.1491064429283142,
+      "learning_rate": 4.190381496657565e-05,
+      "loss": 0.0717,
+      "step": 1495
+    },
+    {
+      "epoch": 0.7018364721020003,
+      "grad_norm": 0.1655229777097702,
+      "learning_rate": 4.182700591875267e-05,
+      "loss": 0.071,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7018364721020003,
+      "eval_loss": 0.0841016098856926,
+      "eval_runtime": 237.2254,
+      "eval_samples_per_second": 2.942,
+      "eval_steps_per_second": 0.738,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7041759270090069,
+      "grad_norm": 0.18065688014030457,
+      "learning_rate": 4.1749905368497345e-05,
+      "loss": 0.0733,
+      "step": 1505
+    },
+    {
+      "epoch": 0.7065153819160136,
+      "grad_norm": 0.16063852608203888,
+      "learning_rate": 4.1672514651460124e-05,
+      "loss": 0.0646,
+      "step": 1510
+    },
+    {
+      "epoch": 0.7088548368230202,
+      "grad_norm": 0.18853044509887695,
+      "learning_rate": 4.159483510831814e-05,
+      "loss": 0.0623,
+      "step": 1515
+    },
+    {
+      "epoch": 0.7111942917300269,
+      "grad_norm": 0.15105347335338593,
+      "learning_rate": 4.151686808475204e-05,
+      "loss": 0.0638,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7135337466370336,
+      "grad_norm": 0.17402097582817078,
+      "learning_rate": 4.143861493142258e-05,
+      "loss": 0.0714,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7158732015440402,
+      "grad_norm": 0.17532867193222046,
+      "learning_rate": 4.136007700394733e-05,
+      "loss": 0.0732,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7182126564510469,
+      "grad_norm": 0.16897249221801758,
+      "learning_rate": 4.128125566287711e-05,
+      "loss": 0.0705,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7205521113580535,
+      "grad_norm": 0.19206973910331726,
+      "learning_rate": 4.120215227367247e-05,
+      "loss": 0.0722,
+      "step": 1540
+    },
+    {
+      "epoch": 0.7228915662650602,
+      "grad_norm": 0.1461183875799179,
+      "learning_rate": 4.112276820668001e-05,
+      "loss": 0.0654,
+      "step": 1545
+    },
+    {
+      "epoch": 0.725231021172067,
+      "grad_norm": 0.14818954467773438,
+      "learning_rate": 4.104310483710864e-05,
+      "loss": 0.0664,
+      "step": 1550
+    },
+    {
+      "epoch": 0.7275704760790735,
+      "grad_norm": 0.1520642787218094,
+      "learning_rate": 4.096316354500578e-05,
+      "loss": 0.0647,
+      "step": 1555
+    },
+    {
+      "epoch": 0.7299099309860803,
+      "grad_norm": 0.18049833178520203,
+      "learning_rate": 4.0882945715233426e-05,
+      "loss": 0.0647,
+      "step": 1560
+    },
+    {
+      "epoch": 0.7322493858930869,
+      "grad_norm": 0.14968477189540863,
+      "learning_rate": 4.080245273744419e-05,
+      "loss": 0.0681,
+      "step": 1565
+    },
+    {
+      "epoch": 0.7345888408000936,
+      "grad_norm": 0.17562703788280487,
+      "learning_rate": 4.072168600605718e-05,
+      "loss": 0.0682,
+      "step": 1570
+    },
+    {
+      "epoch": 0.7369282957071003,
+      "grad_norm": 0.14080215990543365,
+      "learning_rate": 4.064064692023389e-05,
+      "loss": 0.0626,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7392677506141069,
+      "grad_norm": 0.16380788385868073,
+      "learning_rate": 4.055933688385394e-05,
+      "loss": 0.1316,
+      "step": 1580
+    },
+    {
+      "epoch": 0.7416072055211136,
+      "grad_norm": 0.15547692775726318,
+      "learning_rate": 4.047775730549075e-05,
+      "loss": 0.08,
+      "step": 1585
+    },
+    {
+      "epoch": 0.7439466604281203,
+      "grad_norm": 0.15663617849349976,
+      "learning_rate": 4.039590959838715e-05,
+      "loss": 0.0722,
+      "step": 1590
+    },
+    {
+      "epoch": 0.7462861153351269,
+      "grad_norm": 0.16741621494293213,
+      "learning_rate": 4.031379518043091e-05,
+      "loss": 0.0708,
+      "step": 1595
+    },
+    {
+      "epoch": 0.7486255702421336,
+      "grad_norm": 0.15329977869987488,
+      "learning_rate": 4.023141547413015e-05,
+      "loss": 0.063,
+      "step": 1600
+    },
+    {
+      "epoch": 0.7509650251491402,
+      "grad_norm": 0.16843506693840027,
+      "learning_rate": 4.0148771906588706e-05,
+      "loss": 0.0766,
+      "step": 1605
+    },
+    {
+      "epoch": 0.7533044800561469,
+      "grad_norm": 0.17443476617336273,
+      "learning_rate": 4.0065865909481417e-05,
+      "loss": 0.0663,
+      "step": 1610
+    },
+    {
+      "epoch": 0.7556439349631536,
+      "grad_norm": 0.14947757124900818,
+      "learning_rate": 3.9982698919029305e-05,
+      "loss": 0.0804,
+      "step": 1615
+    },
+    {
+      "epoch": 0.7579833898701602,
+      "grad_norm": 0.16239669919013977,
+      "learning_rate": 3.9899272375974726e-05,
+      "loss": 0.076,
+      "step": 1620
+    },
+    {
+      "epoch": 0.7603228447771669,
+      "grad_norm": 0.15543022751808167,
+      "learning_rate": 3.981558772555638e-05,
+      "loss": 0.0697,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7626622996841735,
+      "grad_norm": 0.16913045942783356,
+      "learning_rate": 3.97316464174843e-05,
+      "loss": 0.0752,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7650017545911802,
+      "grad_norm": 0.16426704823970795,
+      "learning_rate": 3.964744990591471e-05,
+      "loss": 0.0827,
+      "step": 1635
+    },
+    {
+      "epoch": 0.767341209498187,
+      "grad_norm": 0.1630501002073288,
+      "learning_rate": 3.956299964942485e-05,
+      "loss": 0.0716,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7696806644051936,
+      "grad_norm": 0.15198907256126404,
+      "learning_rate": 3.947829711098772e-05,
+      "loss": 0.0763,
+      "step": 1645
+    },
+    {
+      "epoch": 0.7720201193122003,
+      "grad_norm": 0.17114469408988953,
+      "learning_rate": 3.9393343757946696e-05,
+      "loss": 0.0662,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7743595742192069,
+      "grad_norm": 0.14625594019889832,
+      "learning_rate": 3.930814106199018e-05,
+      "loss": 0.0721,
+      "step": 1655
+    },
+    {
+      "epoch": 0.7766990291262136,
+      "grad_norm": 0.16365419328212738,
+      "learning_rate": 3.922269049912602e-05,
+      "loss": 0.0755,
+      "step": 1660
+    },
+    {
+      "epoch": 0.7790384840332203,
+      "grad_norm": 0.19672849774360657,
+      "learning_rate": 3.9136993549656006e-05,
+      "loss": 0.0721,
+      "step": 1665
+    },
+    {
+      "epoch": 0.7813779389402269,
+      "grad_norm": 0.17917795479297638,
+      "learning_rate": 3.905105169815021e-05,
+      "loss": 0.0638,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7837173938472336,
+      "grad_norm": 0.15261490643024445,
+      "learning_rate": 3.896486643342124e-05,
+      "loss": 0.0779,
+      "step": 1675
+    },
+    {
+      "epoch": 0.7860568487542403,
+      "grad_norm": 0.1785430610179901,
+      "learning_rate": 3.887843924849849e-05,
+      "loss": 0.0715,
+      "step": 1680
+    },
+    {
+      "epoch": 0.7883963036612469,
+      "grad_norm": 0.15738226473331451,
+      "learning_rate": 3.879177164060225e-05,
+      "loss": 0.0719,
+      "step": 1685
+    },
+    {
+      "epoch": 0.7907357585682536,
+      "grad_norm": 0.17169606685638428,
+      "learning_rate": 3.8704865111117746e-05,
+      "loss": 0.0661,
+      "step": 1690
+    },
+    {
+      "epoch": 0.7930752134752602,
+      "grad_norm": 0.15554682910442352,
+      "learning_rate": 3.861772116556921e-05,
+      "loss": 0.0715,
+      "step": 1695
+    },
+    {
+      "epoch": 0.7954146683822669,
+      "grad_norm": 0.2050098329782486,
+      "learning_rate": 3.853034131359371e-05,
+      "loss": 0.0726,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7977541232892736,
+      "grad_norm": 0.14537398517131805,
+      "learning_rate": 3.8442727068915066e-05,
+      "loss": 0.0736,
+      "step": 1705
+    },
+    {
+      "epoch": 0.8000935781962802,
+      "grad_norm": 0.16626664996147156,
+      "learning_rate": 3.8354879949317546e-05,
+      "loss": 0.0641,
+      "step": 1710
+    },
+    {
+      "epoch": 0.802433033103287,
+      "grad_norm": 0.17925380170345306,
+      "learning_rate": 3.8266801476619694e-05,
+      "loss": 0.07,
+      "step": 1715
+    },
+    {
+      "epoch": 0.8047724880102936,
+      "grad_norm": 0.1543145477771759,
+      "learning_rate": 3.817849317664784e-05,
+      "loss": 0.069,
+      "step": 1720
+    },
+    {
+      "epoch": 0.8071119429173003,
+      "grad_norm": 0.15055689215660095,
+      "learning_rate": 3.808995657920975e-05,
+      "loss": 0.0686,
+      "step": 1725
+    },
+    {
+      "epoch": 0.809451397824307,
+      "grad_norm": 0.13166533410549164,
+      "learning_rate": 3.800119321806809e-05,
+      "loss": 0.0643,
+      "step": 1730
+    },
+    {
+      "epoch": 0.8117908527313136,
+      "grad_norm": 0.20518459379673004,
+      "learning_rate": 3.791220463091387e-05,
+      "loss": 0.0846,
+      "step": 1735
+    },
+    {
+      "epoch": 0.8141303076383203,
+      "grad_norm": 0.13176603615283966,
+      "learning_rate": 3.7822992359339794e-05,
+      "loss": 0.0697,
+      "step": 1740
+    },
+    {
+      "epoch": 0.816469762545327,
+      "grad_norm": 0.16138117015361786,
+      "learning_rate": 3.7733557948813575e-05,
+      "loss": 0.0679,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8188092174523336,
+      "grad_norm": 0.20597168803215027,
+      "learning_rate": 3.764390294865112e-05,
+      "loss": 0.0779,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8211486723593403,
+      "grad_norm": 0.18723969161510468,
+      "learning_rate": 3.755402891198974e-05,
+      "loss": 0.141,
+      "step": 1755
+    },
+    {
+      "epoch": 0.8234881272663469,
+      "grad_norm": 0.16847847402095795,
+      "learning_rate": 3.746393739576121e-05,
+      "loss": 0.0675,
+      "step": 1760
+    },
+    {
+      "epoch": 0.8258275821733536,
+      "grad_norm": 0.1512673944234848,
+      "learning_rate": 3.737362996066483e-05,
+      "loss": 0.0708,
+      "step": 1765
+    },
+    {
+      "epoch": 0.8281670370803603,
+      "grad_norm": 0.17795756459236145,
+      "learning_rate": 3.728310817114034e-05,
+      "loss": 0.0675,
+      "step": 1770
+    },
+    {
+      "epoch": 0.8305064919873669,
+      "grad_norm": 0.18552307784557343,
+      "learning_rate": 3.719237359534087e-05,
+      "loss": 0.0758,
+      "step": 1775
+    },
+    {
+      "epoch": 0.8328459468943736,
+      "grad_norm": 0.16442035138607025,
+      "learning_rate": 3.710142780510573e-05,
+      "loss": 0.0728,
+      "step": 1780
+    },
+    {
+      "epoch": 0.8351854018013802,
+      "grad_norm": 0.16192203760147095,
+      "learning_rate": 3.7010272375933216e-05,
+      "loss": 0.0651,
+      "step": 1785
+    },
+    {
+      "epoch": 0.8375248567083869,
+      "grad_norm": 0.15759891271591187,
+      "learning_rate": 3.691890888695329e-05,
+      "loss": 0.0659,
+      "step": 1790
+    },
+    {
+      "epoch": 0.8398643116153937,
+      "grad_norm": 0.16758672893047333,
+      "learning_rate": 3.6827338920900254e-05,
+      "loss": 0.065,
+      "step": 1795
+    },
+    {
+      "epoch": 0.8422037665224003,
+      "grad_norm": 0.1629868447780609,
+      "learning_rate": 3.6735564064085296e-05,
+      "loss": 0.0748,
+      "step": 1800
+    },
+    {
+      "epoch": 0.844543221429407,
+      "grad_norm": 0.1754574477672577,
+      "learning_rate": 3.664358590636903e-05,
+      "loss": 0.0857,
+      "step": 1805
+    },
+    {
+      "epoch": 0.8468826763364136,
+      "grad_norm": 0.16642498970031738,
+      "learning_rate": 3.655140604113395e-05,
+      "loss": 0.0721,
+      "step": 1810
+    },
+    {
+      "epoch": 0.8492221312434203,
+      "grad_norm": 0.15343014895915985,
+      "learning_rate": 3.645902606525683e-05,
+      "loss": 0.0653,
+      "step": 1815
+    },
+    {
+      "epoch": 0.851561586150427,
+      "grad_norm": 0.15997187793254852,
+      "learning_rate": 3.636644757908107e-05,
+      "loss": 0.0752,
+      "step": 1820
+    },
+    {
+      "epoch": 0.8539010410574336,
+      "grad_norm": 0.1555909514427185,
+      "learning_rate": 3.627367218638893e-05,
+      "loss": 0.0692,
+      "step": 1825
+    },
+    {
+      "epoch": 0.8562404959644403,
+      "grad_norm": 0.1649237424135208,
+      "learning_rate": 3.618070149437381e-05,
+      "loss": 0.0706,
+      "step": 1830
+    },
+    {
+      "epoch": 0.858579950871447,
+      "grad_norm": 0.14074915647506714,
+      "learning_rate": 3.6087537113612364e-05,
+      "loss": 0.0612,
+      "step": 1835
+    },
+    {
+      "epoch": 0.8609194057784536,
+      "grad_norm": 0.15589754283428192,
+      "learning_rate": 3.59941806580366e-05,
+      "loss": 0.0724,
+      "step": 1840
+    },
+    {
+      "epoch": 0.8632588606854603,
+      "grad_norm": 0.1727776825428009,
+      "learning_rate": 3.590063374490595e-05,
+      "loss": 0.0851,
+      "step": 1845
+    },
+    {
+      "epoch": 0.8655983155924669,
+      "grad_norm": 0.14394600689411163,
+      "learning_rate": 3.580689799477921e-05,
+      "loss": 0.0669,
+      "step": 1850
+    },
+    {
+      "epoch": 0.8679377704994736,
+      "grad_norm": 0.15605291724205017,
+      "learning_rate": 3.5712975031486525e-05,
+      "loss": 0.0716,
+      "step": 1855
+    },
+    {
+      "epoch": 0.8702772254064803,
+      "grad_norm": 0.19002071022987366,
+      "learning_rate": 3.56188664821012e-05,
+      "loss": 0.0748,
+      "step": 1860
+    },
+    {
+      "epoch": 0.8726166803134869,
+      "grad_norm": 0.16365236043930054,
+      "learning_rate": 3.5524573976911546e-05,
+      "loss": 0.079,
+      "step": 1865
+    },
+    {
+      "epoch": 0.8749561352204936,
+      "grad_norm": 0.19456696510314941,
+      "learning_rate": 3.543009914939261e-05,
+      "loss": 0.0747,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8772955901275002,
+      "grad_norm": 0.15589672327041626,
+      "learning_rate": 3.533544363617796e-05,
+      "loss": 0.0657,
+      "step": 1875
+    },
+    {
+      "epoch": 0.879635045034507,
+      "grad_norm": 0.14598876237869263,
+      "learning_rate": 3.5240609077031196e-05,
+      "loss": 0.0642,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8819744999415137,
+      "grad_norm": 0.15636853873729706,
+      "learning_rate": 3.514559711481765e-05,
+      "loss": 0.0608,
+      "step": 1885
+    },
+    {
+      "epoch": 0.8843139548485203,
+      "grad_norm": 0.1746116727590561,
+      "learning_rate": 3.505040939547591e-05,
+      "loss": 0.077,
+      "step": 1890
+    },
+    {
+      "epoch": 0.886653409755527,
+      "grad_norm": 0.15822593867778778,
+      "learning_rate": 3.495504756798926e-05,
+      "loss": 0.0788,
+      "step": 1895
+    },
+    {
+      "epoch": 0.8889928646625336,
+      "grad_norm": 0.16369085013866425,
+      "learning_rate": 3.485951328435716e-05,
+      "loss": 0.0764,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8913323195695403,
+      "grad_norm": 0.1527598798274994,
+      "learning_rate": 3.476380819956662e-05,
+      "loss": 0.0722,
+      "step": 1905
+    },
+    {
+      "epoch": 0.893671774476547,
+      "grad_norm": 0.13147583603858948,
+      "learning_rate": 3.466793397156351e-05,
+      "loss": 0.0694,
+      "step": 1910
+    },
+    {
+      "epoch": 0.8960112293835536,
+      "grad_norm": 0.15432552993297577,
+      "learning_rate": 3.457189226122384e-05,
+      "loss": 0.0652,
+      "step": 1915
+    },
+    {
+      "epoch": 0.8983506842905603,
+      "grad_norm": 0.15792711079120636,
+      "learning_rate": 3.4475684732325024e-05,
+      "loss": 0.0763,
+      "step": 1920
+    },
+    {
+      "epoch": 0.900690139197567,
+      "grad_norm": 0.14269225299358368,
+      "learning_rate": 3.437931305151701e-05,
+      "loss": 0.0739,
+      "step": 1925
+    },
+    {
+      "epoch": 0.9030295941045736,
+      "grad_norm": 0.18912151455879211,
+      "learning_rate": 3.428277888829344e-05,
+      "loss": 0.0839,
+      "step": 1930
+    },
+    {
+      "epoch": 0.9053690490115803,
+      "grad_norm": 0.16357168555259705,
+      "learning_rate": 3.418608391496273e-05,
+      "loss": 0.0695,
+      "step": 1935
+    },
+    {
+      "epoch": 0.9077085039185869,
+      "grad_norm": 0.1561209261417389,
+      "learning_rate": 3.408922980661905e-05,
+      "loss": 0.0681,
+      "step": 1940
+    },
+    {
+      "epoch": 0.9100479588255936,
+      "grad_norm": 0.17604918777942657,
+      "learning_rate": 3.39922182411134e-05,
+      "loss": 0.0694,
+      "step": 1945
+    },
+    {
+      "epoch": 0.9123874137326003,
+      "grad_norm": 0.16690443456172943,
+      "learning_rate": 3.3895050899024416e-05,
+      "loss": 0.1077,
+      "step": 1950
+    },
+    {
+      "epoch": 0.914726868639607,
+      "grad_norm": 0.15058398246765137,
+      "learning_rate": 3.379772946362939e-05,
+      "loss": 0.0604,
+      "step": 1955
+    },
+    {
+      "epoch": 0.9170663235466137,
+      "grad_norm": 0.12406554818153381,
+      "learning_rate": 3.3700255620875025e-05,
+      "loss": 0.061,
+      "step": 1960
+    },
+    {
+      "epoch": 0.9194057784536203,
+      "grad_norm": 0.15860305726528168,
+      "learning_rate": 3.3602631059348245e-05,
+      "loss": 0.0714,
+      "step": 1965
+    },
+    {
+      "epoch": 0.921745233360627,
+      "grad_norm": 0.15389299392700195,
+      "learning_rate": 3.350485747024695e-05,
+      "loss": 0.0724,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9240846882676337,
+      "grad_norm": 0.16053031384944916,
+      "learning_rate": 3.340693654735073e-05,
+      "loss": 0.064,
+      "step": 1975
+    },
+    {
+      "epoch": 0.9264241431746403,
+      "grad_norm": 0.14700767397880554,
+      "learning_rate": 3.330886998699149e-05,
+      "loss": 0.0639,
+      "step": 1980
+    },
+    {
+      "epoch": 0.928763598081647,
+      "grad_norm": 0.15613354742527008,
+      "learning_rate": 3.321065948802411e-05,
+      "loss": 0.0667,
+      "step": 1985
+    },
+    {
+      "epoch": 0.9311030529886536,
+      "grad_norm": 0.15523682534694672,
+      "learning_rate": 3.311230675179697e-05,
+      "loss": 0.0634,
+      "step": 1990
+    },
+    {
+      "epoch": 0.9334425078956603,
+      "grad_norm": 0.14495091140270233,
+      "learning_rate": 3.301381348212249e-05,
+      "loss": 0.0632,
+      "step": 1995
+    },
+    {
+      "epoch": 0.935781962802667,
+      "grad_norm": 0.17717325687408447,
+      "learning_rate": 3.291518138524764e-05,
+      "loss": 0.0678,
+      "step": 2000
+    },
+    {
+      "epoch": 0.935781962802667,
+      "eval_loss": 0.08131425082683563,
+      "eval_runtime": 232.0604,
+      "eval_samples_per_second": 3.008,
+      "eval_steps_per_second": 0.754,
+      "step": 2000
+    },
+    {
+      "epoch": 0.9381214177096736,
+      "grad_norm": 0.15504471957683563,
+      "learning_rate": 3.281641216982435e-05,
+      "loss": 0.1141,
+      "step": 2005
+    },
+    {
+      "epoch": 0.9404608726166803,
+      "grad_norm": 0.15814577043056488,
+      "learning_rate": 3.271750754687991e-05,
+      "loss": 0.0658,
+      "step": 2010
+    },
+    {
+      "epoch": 0.942800327523687,
+      "grad_norm": 0.18318745493888855,
+      "learning_rate": 3.261846922978736e-05,
+      "loss": 0.0727,
+      "step": 2015
+    },
+    {
+      "epoch": 0.9451397824306936,
+      "grad_norm": 0.16373957693576813,
+      "learning_rate": 3.2519298934235796e-05,
+      "loss": 0.0683,
+      "step": 2020
+    },
+    {
+      "epoch": 0.9474792373377003,
+      "grad_norm": 0.1619662046432495,
+      "learning_rate": 3.24199983782006e-05,
+      "loss": 0.0663,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9498186922447069,
+      "grad_norm": 0.15787597000598907,
+      "learning_rate": 3.232056928191376e-05,
+      "loss": 0.0743,
+      "step": 2030
+    },
+    {
+      "epoch": 0.9521581471517137,
+      "grad_norm": 0.18520085513591766,
+      "learning_rate": 3.2221013367834016e-05,
+      "loss": 0.0994,
+      "step": 2035
+    },
+    {
+      "epoch": 0.9544976020587204,
+      "grad_norm": 0.1397067755460739,
+      "learning_rate": 3.212133236061704e-05,
+      "loss": 0.0835,
+      "step": 2040
+    },
+    {
+      "epoch": 0.956837056965727,
+      "grad_norm": 0.1584145724773407,
+      "learning_rate": 3.202152798708554e-05,
+      "loss": 0.0843,
+      "step": 2045
+    },
+    {
+      "epoch": 0.9591765118727337,
+      "grad_norm": 0.14914660155773163,
+      "learning_rate": 3.1921601976199365e-05,
+      "loss": 0.0744,
+      "step": 2050
+    },
+    {
+      "epoch": 0.9615159667797403,
+      "grad_norm": 0.14191728830337524,
+      "learning_rate": 3.182155605902556e-05,
+      "loss": 0.0654,
+      "step": 2055
+    },
+    {
+      "epoch": 0.963855421686747,
+      "grad_norm": 0.15701289474964142,
+      "learning_rate": 3.1721391968708335e-05,
+      "loss": 0.0665,
+      "step": 2060
+    },
+    {
+      "epoch": 0.9661948765937537,
+      "grad_norm": 0.14742569625377655,
+      "learning_rate": 3.162111144043911e-05,
+      "loss": 0.072,
+      "step": 2065
+    },
+    {
+      "epoch": 0.9685343315007603,
+      "grad_norm": 0.15007531642913818,
+      "learning_rate": 3.1520716211426385e-05,
+      "loss": 0.0754,
+      "step": 2070
+    },
+    {
+      "epoch": 0.970873786407767,
+      "grad_norm": 0.1743524670600891,
+      "learning_rate": 3.142020802086569e-05,
+      "loss": 0.0739,
+      "step": 2075
+    },
+    {
+      "epoch": 0.9732132413147736,
+      "grad_norm": 0.12957176566123962,
+      "learning_rate": 3.131958860990945e-05,
+      "loss": 0.0617,
+      "step": 2080
+    },
+    {
+      "epoch": 0.9755526962217803,
+      "grad_norm": 0.15021301805973053,
+      "learning_rate": 3.121885972163681e-05,
+      "loss": 0.0785,
+      "step": 2085
+    },
+    {
+      "epoch": 0.977892151128787,
+      "grad_norm": 0.18265359103679657,
+      "learning_rate": 3.111802310102341e-05,
+      "loss": 0.0675,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9802316060357936,
+      "grad_norm": 0.14601291716098785,
+      "learning_rate": 3.101708049491125e-05,
+      "loss": 0.0694,
+      "step": 2095
+    },
+    {
+      "epoch": 0.9825710609428003,
+      "grad_norm": 0.1612151712179184,
+      "learning_rate": 3.091603365197833e-05,
+      "loss": 0.0743,
+      "step": 2100
+    },
+    {
+      "epoch": 0.984910515849807,
+      "grad_norm": 0.14992724359035492,
+      "learning_rate": 3.081488432270841e-05,
+      "loss": 0.0664,
+      "step": 2105
+    },
+    {
+      "epoch": 0.9872499707568136,
+      "grad_norm": 0.14685116708278656,
+      "learning_rate": 3.071363425936066e-05,
+      "loss": 0.0762,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9895894256638204,
+      "grad_norm": 0.14769916236400604,
+      "learning_rate": 3.061228521593931e-05,
+      "loss": 0.0636,
+      "step": 2115
+    },
+    {
+      "epoch": 0.991928880570827,
+      "grad_norm": 0.14650078117847443,
+      "learning_rate": 3.0510838948163308e-05,
+      "loss": 0.0836,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9942683354778337,
+      "grad_norm": 0.1642710417509079,
+      "learning_rate": 3.0409297213435818e-05,
+      "loss": 0.0723,
+      "step": 2125
+    },
+    {
+      "epoch": 0.9966077903848404,
+      "grad_norm": 0.13404929637908936,
+      "learning_rate": 3.030766177081385e-05,
+      "loss": 0.0766,
+      "step": 2130
+    },
+    {
+      "epoch": 0.998947245291847,
+      "grad_norm": 0.1478378027677536,
+      "learning_rate": 3.020593438097776e-05,
+      "loss": 0.0619,
+      "step": 2135
+    },
+    {
+      "epoch": 1.0012867001988537,
+      "grad_norm": 0.1335192173719406,
+      "learning_rate": 3.010411680620075e-05,
+      "loss": 0.0641,
+      "step": 2140
+    },
+    {
+      "epoch": 1.0036261551058603,
+      "grad_norm": 0.1439552754163742,
+      "learning_rate": 3.0002210810318338e-05,
+      "loss": 0.0735,
+      "step": 2145
+    },
+    {
+      "epoch": 1.005965610012867,
+      "grad_norm": 0.14612635970115662,
+      "learning_rate": 2.9900218158697797e-05,
+      "loss": 0.0576,
+      "step": 2150
+    },
+    {
+      "epoch": 1.0083050649198737,
+      "grad_norm": 0.17644114792346954,
+      "learning_rate": 2.9798140618207598e-05,
+      "loss": 0.0721,
+      "step": 2155
+    },
+    {
+      "epoch": 1.0106445198268803,
+      "grad_norm": 0.15315324068069458,
+      "learning_rate": 2.9695979957186764e-05,
+      "loss": 0.0642,
+      "step": 2160
+    },
+    {
+      "epoch": 1.012983974733887,
+      "grad_norm": 0.15840911865234375,
+      "learning_rate": 2.9593737945414264e-05,
+      "loss": 0.0751,
+      "step": 2165
+    },
+    {
+      "epoch": 1.0153234296408937,
+      "grad_norm": 0.14203301072120667,
+      "learning_rate": 2.9491416354078343e-05,
+      "loss": 0.0619,
+      "step": 2170
+    },
+    {
+      "epoch": 1.0176628845479003,
+      "grad_norm": 0.17222335934638977,
+      "learning_rate": 2.938901695574585e-05,
+      "loss": 0.0719,
+      "step": 2175
+    },
+    {
+      "epoch": 1.020002339454907,
+      "grad_norm": 0.16653592884540558,
+      "learning_rate": 2.9286541524331525e-05,
+      "loss": 0.0597,
+      "step": 2180
+    },
+    {
+      "epoch": 1.0223417943619137,
+      "grad_norm": 0.12632368505001068,
+      "learning_rate": 2.9183991835067237e-05,
+      "loss": 0.0585,
+      "step": 2185
+    },
+    {
+      "epoch": 1.0246812492689203,
+      "grad_norm": 0.15063339471817017,
+      "learning_rate": 2.9081369664471293e-05,
+      "loss": 0.0712,
+      "step": 2190
+    },
+    {
+      "epoch": 1.027020704175927,
+      "grad_norm": 0.15402260422706604,
+      "learning_rate": 2.8978676790317605e-05,
+      "loss": 0.0631,
+      "step": 2195
+    },
+    {
+      "epoch": 1.0293601590829338,
+      "grad_norm": 0.17674040794372559,
+      "learning_rate": 2.8875914991604948e-05,
+      "loss": 0.0611,
+      "step": 2200
+    },
+    {
+      "epoch": 1.0316996139899404,
+      "grad_norm": 0.15764425694942474,
+      "learning_rate": 2.8773086048526076e-05,
+      "loss": 0.1481,
+      "step": 2205
+    },
+    {
+      "epoch": 1.034039068896947,
+      "grad_norm": 0.17565783858299255,
+      "learning_rate": 2.8670191742436946e-05,
+      "loss": 0.0623,
+      "step": 2210
+    },
+    {
+      "epoch": 1.0363785238039536,
+      "grad_norm": 0.1751687377691269,
+      "learning_rate": 2.8567233855825813e-05,
+      "loss": 0.0599,
+      "step": 2215
+    },
+    {
+      "epoch": 1.0387179787109604,
+      "grad_norm": 0.15617790818214417,
+      "learning_rate": 2.8464214172282367e-05,
+      "loss": 0.0593,
+      "step": 2220
+    },
+    {
+      "epoch": 1.041057433617967,
+      "grad_norm": 0.1573198437690735,
+      "learning_rate": 2.8361134476466843e-05,
+      "loss": 0.0643,
+      "step": 2225
+    },
+    {
+      "epoch": 1.0433968885249736,
+      "grad_norm": 0.15318118035793304,
+      "learning_rate": 2.8257996554079113e-05,
+      "loss": 0.0582,
+      "step": 2230
+    },
+    {
+      "epoch": 1.0457363434319804,
+      "grad_norm": 0.14526399970054626,
+      "learning_rate": 2.8154802191827706e-05,
+      "loss": 0.0577,
+      "step": 2235
+    },
+    {
+      "epoch": 1.048075798338987,
+      "grad_norm": 0.1734457015991211,
+      "learning_rate": 2.805155317739891e-05,
+      "loss": 0.0618,
+      "step": 2240
+    },
+    {
+      "epoch": 1.0504152532459936,
+      "grad_norm": 0.16416044533252716,
+      "learning_rate": 2.7948251299425758e-05,
+      "loss": 0.067,
+      "step": 2245
+    },
+    {
+      "epoch": 1.0527547081530004,
+      "grad_norm": 0.14763504266738892,
+      "learning_rate": 2.784489834745709e-05,
+      "loss": 0.0626,
+      "step": 2250
+    },
+    {
+      "epoch": 1.055094163060007,
+      "grad_norm": 0.15287157893180847,
+      "learning_rate": 2.77414961119265e-05,
+      "loss": 0.0627,
+      "step": 2255
+    },
+    {
+      "epoch": 1.0574336179670136,
+      "grad_norm": 0.17158065736293793,
+      "learning_rate": 2.763804638412137e-05,
+      "loss": 0.0653,
+      "step": 2260
+    },
+    {
+      "epoch": 1.0597730728740204,
+      "grad_norm": 0.1766432225704193,
+      "learning_rate": 2.753455095615179e-05,
+      "loss": 0.0667,
+      "step": 2265
+    },
+    {
+      "epoch": 1.062112527781027,
+      "grad_norm": 0.16371361911296844,
+      "learning_rate": 2.7431011620919554e-05,
+      "loss": 0.0653,
+      "step": 2270
+    },
+    {
+      "epoch": 1.0644519826880336,
+      "grad_norm": 0.26466837525367737,
+      "learning_rate": 2.7327430172087077e-05,
+      "loss": 0.0605,
+      "step": 2275
+    },
+    {
+      "epoch": 1.0667914375950405,
+      "grad_norm": 0.1517406851053238,
+      "learning_rate": 2.722380840404632e-05,
+      "loss": 0.0595,
+      "step": 2280
+    },
+    {
+      "epoch": 1.069130892502047,
+      "grad_norm": 0.16470152139663696,
+      "learning_rate": 2.7120148111887732e-05,
+      "loss": 0.0706,
+      "step": 2285
+    },
+    {
+      "epoch": 1.0714703474090537,
+      "grad_norm": 0.15149088203907013,
+      "learning_rate": 2.7016451091369116e-05,
+      "loss": 0.0646,
+      "step": 2290
+    },
+    {
+      "epoch": 1.0738098023160603,
+      "grad_norm": 0.14142438769340515,
+      "learning_rate": 2.6912719138884553e-05,
+      "loss": 0.0612,
+      "step": 2295
+    },
+    {
+      "epoch": 1.076149257223067,
+      "grad_norm": 0.17522796988487244,
+      "learning_rate": 2.6808954051433248e-05,
+      "loss": 0.0657,
+      "step": 2300
+    },
+    {
+      "epoch": 1.0784887121300737,
+      "grad_norm": 0.1683465838432312,
+      "learning_rate": 2.670515762658844e-05,
+      "loss": 0.0649,
+      "step": 2305
+    },
+    {
+      "epoch": 1.0808281670370803,
+      "grad_norm": 0.1598363220691681,
+      "learning_rate": 2.6601331662466227e-05,
+      "loss": 0.0523,
+      "step": 2310
+    },
+    {
+      "epoch": 1.0831676219440871,
+      "grad_norm": 0.14117397367954254,
+      "learning_rate": 2.6497477957694443e-05,
+      "loss": 0.0565,
+      "step": 2315
+    },
+    {
+      "epoch": 1.0855070768510937,
+      "grad_norm": 0.16452661156654358,
+      "learning_rate": 2.6393598311381473e-05,
+      "loss": 0.0629,
+      "step": 2320
+    },
+    {
+      "epoch": 1.0878465317581003,
+      "grad_norm": 0.14130878448486328,
+      "learning_rate": 2.6289694523085112e-05,
+      "loss": 0.0685,
+      "step": 2325
+    },
+    {
+      "epoch": 1.0901859866651071,
+      "grad_norm": 0.1380264163017273,
+      "learning_rate": 2.6185768392781366e-05,
+      "loss": 0.056,
+      "step": 2330
+    },
+    {
+      "epoch": 1.0925254415721137,
+      "grad_norm": 0.16482488811016083,
+      "learning_rate": 2.6081821720833287e-05,
+      "loss": 0.0569,
+      "step": 2335
+    },
+    {
+      "epoch": 1.0948648964791203,
+      "grad_norm": 0.1513761430978775,
+      "learning_rate": 2.5977856307959798e-05,
+      "loss": 0.0622,
+      "step": 2340
+    },
+    {
+      "epoch": 1.097204351386127,
+      "grad_norm": 0.15945203602313995,
+      "learning_rate": 2.5873873955204448e-05,
+      "loss": 0.0574,
+      "step": 2345
+    },
+    {
+      "epoch": 1.0995438062931338,
+      "grad_norm": 0.16861629486083984,
+      "learning_rate": 2.5769876463904265e-05,
+      "loss": 0.0575,
+      "step": 2350
+    },
+    {
+      "epoch": 1.1018832612001404,
+      "grad_norm": 0.15520039200782776,
+      "learning_rate": 2.5665865635658527e-05,
+      "loss": 0.063,
+      "step": 2355
+    },
+    {
+      "epoch": 1.104222716107147,
+      "grad_norm": 0.1708049774169922,
+      "learning_rate": 2.5561843272297536e-05,
+      "loss": 0.0625,
+      "step": 2360
+    },
+    {
+      "epoch": 1.1065621710141538,
+      "grad_norm": 0.1442188173532486,
+      "learning_rate": 2.5457811175851465e-05,
+      "loss": 0.0539,
+      "step": 2365
+    },
+    {
+      "epoch": 1.1089016259211604,
+      "grad_norm": 0.15366148948669434,
+      "learning_rate": 2.5353771148519057e-05,
+      "loss": 0.0639,
+      "step": 2370
+    },
+    {
+      "epoch": 1.111241080828167,
+      "grad_norm": 0.1677359789609909,
+      "learning_rate": 2.524972499263646e-05,
+      "loss": 0.0749,
+      "step": 2375
+    },
+    {
+      "epoch": 1.1135805357351738,
+      "grad_norm": 0.16887301206588745,
+      "learning_rate": 2.514567451064599e-05,
+      "loss": 0.0632,
+      "step": 2380
+    },
+    {
+      "epoch": 1.1159199906421804,
+      "grad_norm": 0.16333596408367157,
+      "learning_rate": 2.50416215050649e-05,
+      "loss": 0.0578,
+      "step": 2385
+    },
+    {
+      "epoch": 1.118259445549187,
+      "grad_norm": 0.1616666167974472,
+      "learning_rate": 2.4937567778454188e-05,
+      "loss": 0.0603,
+      "step": 2390
+    },
+    {
+      "epoch": 1.1205989004561938,
+      "grad_norm": 0.15231609344482422,
+      "learning_rate": 2.4833515133387296e-05,
+      "loss": 0.0577,
+      "step": 2395
+    },
+    {
+      "epoch": 1.1229383553632004,
+      "grad_norm": 0.16632795333862305,
+      "learning_rate": 2.4729465372418972e-05,
+      "loss": 0.0715,
+      "step": 2400
+    },
+    {
+      "epoch": 1.125277810270207,
+      "grad_norm": 0.16579392552375793,
+      "learning_rate": 2.4625420298053968e-05,
+      "loss": 0.0696,
+      "step": 2405
+    },
+    {
+      "epoch": 1.1276172651772136,
+      "grad_norm": 0.1642322987318039,
+      "learning_rate": 2.4521381712715884e-05,
+      "loss": 0.0631,
+      "step": 2410
+    },
+    {
+      "epoch": 1.1299567200842204,
+      "grad_norm": 0.1781720668077469,
+      "learning_rate": 2.441735141871587e-05,
+      "loss": 0.0584,
+      "step": 2415
+    },
+    {
+      "epoch": 1.132296174991227,
+      "grad_norm": 0.16532465815544128,
+      "learning_rate": 2.4313331218221476e-05,
+      "loss": 0.0658,
+      "step": 2420
+    },
+    {
+      "epoch": 1.1346356298982336,
+      "grad_norm": 0.1574028581380844,
+      "learning_rate": 2.420932291322536e-05,
+      "loss": 0.0636,
+      "step": 2425
+    },
+    {
+      "epoch": 1.1369750848052405,
+      "grad_norm": 0.1727520376443863,
+      "learning_rate": 2.410532830551415e-05,
+      "loss": 0.0687,
+      "step": 2430
+    },
+    {
+      "epoch": 1.139314539712247,
+      "grad_norm": 0.17281392216682434,
+      "learning_rate": 2.4001349196637144e-05,
+      "loss": 0.065,
+      "step": 2435
+    },
+    {
+      "epoch": 1.1416539946192537,
+      "grad_norm": 0.16799040138721466,
+      "learning_rate": 2.3897387387875188e-05,
+      "loss": 0.0597,
+      "step": 2440
+    },
+    {
+      "epoch": 1.1439934495262605,
+      "grad_norm": 0.1646750122308731,
+      "learning_rate": 2.3793444680209397e-05,
+      "loss": 0.0617,
+      "step": 2445
+    },
+    {
+      "epoch": 1.146332904433267,
+      "grad_norm": 0.15122398734092712,
+      "learning_rate": 2.3689522874290028e-05,
+      "loss": 0.0646,
+      "step": 2450
+    },
+    {
+      "epoch": 1.1486723593402737,
+      "grad_norm": 0.15441857278347015,
+      "learning_rate": 2.358562377040519e-05,
+      "loss": 0.0734,
+      "step": 2455
+    },
+    {
+      "epoch": 1.1510118142472803,
+      "grad_norm": 0.1397976279258728,
+      "learning_rate": 2.3481749168449774e-05,
+      "loss": 0.0599,
+      "step": 2460
+    },
+    {
+      "epoch": 1.153351269154287,
+      "grad_norm": 0.1505647897720337,
+      "learning_rate": 2.3377900867894158e-05,
+      "loss": 0.0587,
+      "step": 2465
+    },
+    {
+      "epoch": 1.1556907240612937,
+      "grad_norm": 0.17228984832763672,
+      "learning_rate": 2.3274080667753128e-05,
+      "loss": 0.0713,
+      "step": 2470
+    },
+    {
+      "epoch": 1.1580301789683003,
+      "grad_norm": 0.1795542687177658,
+      "learning_rate": 2.3170290366554653e-05,
+      "loss": 0.0681,
+      "step": 2475
+    },
+    {
+      "epoch": 1.1603696338753071,
+      "grad_norm": 0.16465342044830322,
+      "learning_rate": 2.3066531762308766e-05,
+      "loss": 0.0598,
+      "step": 2480
+    },
+    {
+      "epoch": 1.1627090887823137,
+      "grad_norm": 0.18022869527339935,
+      "learning_rate": 2.2962806652476363e-05,
+      "loss": 0.0553,
+      "step": 2485
+    },
+    {
+      "epoch": 1.1650485436893203,
+      "grad_norm": 0.17633071541786194,
+      "learning_rate": 2.2859116833938146e-05,
+      "loss": 0.0694,
+      "step": 2490
+    },
+    {
+      "epoch": 1.1673879985963271,
+      "grad_norm": 0.1658298820257187,
+      "learning_rate": 2.2755464102963408e-05,
+      "loss": 0.0576,
+      "step": 2495
+    },
+    {
+      "epoch": 1.1697274535033337,
+      "grad_norm": 0.1415170431137085,
+      "learning_rate": 2.2651850255178974e-05,
+      "loss": 0.0697,
+      "step": 2500
+    },
+    {
+      "epoch": 1.1697274535033337,
+      "eval_loss": 0.07991591095924377,
+      "eval_runtime": 236.928,
+      "eval_samples_per_second": 2.946,
+      "eval_steps_per_second": 0.739,
+      "step": 2500
+    },
+    {
+      "epoch": 1.1720669084103403,
+      "grad_norm": 0.17509862780570984,
+      "learning_rate": 2.2548277085538092e-05,
+      "loss": 0.0669,
+      "step": 2505
+    },
+    {
+      "epoch": 1.1744063633173472,
+      "grad_norm": 0.17153185606002808,
+      "learning_rate": 2.2444746388289283e-05,
+      "loss": 0.0853,
+      "step": 2510
+    },
+    {
+      "epoch": 1.1767458182243538,
+      "grad_norm": 0.1431114226579666,
+      "learning_rate": 2.2341259956945342e-05,
+      "loss": 0.0595,
+      "step": 2515
+    },
+    {
+      "epoch": 1.1790852731313604,
+      "grad_norm": 0.14989079535007477,
+      "learning_rate": 2.2237819584252188e-05,
+      "loss": 0.0544,
+      "step": 2520
+    },
+    {
+      "epoch": 1.1814247280383672,
+      "grad_norm": 0.13587923347949982,
+      "learning_rate": 2.2134427062157866e-05,
+      "loss": 0.0652,
+      "step": 2525
+    },
+    {
+      "epoch": 1.1837641829453738,
+      "grad_norm": 0.1634293496608734,
+      "learning_rate": 2.2031084181781476e-05,
+      "loss": 0.0637,
+      "step": 2530
+    },
+    {
+      "epoch": 1.1861036378523804,
+      "grad_norm": 0.16133365035057068,
+      "learning_rate": 2.192779273338215e-05,
+      "loss": 0.079,
+      "step": 2535
+    },
+    {
+      "epoch": 1.188443092759387,
+      "grad_norm": 0.12266691774129868,
+      "learning_rate": 2.182455450632803e-05,
+      "loss": 0.0569,
+      "step": 2540
+    },
+    {
+      "epoch": 1.1907825476663938,
+      "grad_norm": 0.15535689890384674,
+      "learning_rate": 2.17213712890653e-05,
+      "loss": 0.0569,
+      "step": 2545
+    },
+    {
+      "epoch": 1.1931220025734004,
+      "grad_norm": 0.16880477964878082,
+      "learning_rate": 2.1618244869087157e-05,
+      "loss": 0.0651,
+      "step": 2550
+    },
+    {
+      "epoch": 1.195461457480407,
+      "grad_norm": 0.13848161697387695,
+      "learning_rate": 2.151517703290289e-05,
+      "loss": 0.0693,
+      "step": 2555
+    },
+    {
+      "epoch": 1.1978009123874138,
+      "grad_norm": 0.16127823293209076,
+      "learning_rate": 2.1412169566006897e-05,
+      "loss": 0.0598,
+      "step": 2560
+    },
+    {
+      "epoch": 1.2001403672944204,
+      "grad_norm": 0.11767289787530899,
+      "learning_rate": 2.1309224252847793e-05,
+      "loss": 0.0593,
+      "step": 2565
+    },
+    {
+      "epoch": 1.202479822201427,
+      "grad_norm": 0.17509087920188904,
+      "learning_rate": 2.1206342876797436e-05,
+      "loss": 0.0604,
+      "step": 2570
+    },
+    {
+      "epoch": 1.2048192771084336,
+      "grad_norm": 0.16853219270706177,
+      "learning_rate": 2.110352722012011e-05,
+      "loss": 0.063,
+      "step": 2575
+    },
+    {
+      "epoch": 1.2071587320154404,
+      "grad_norm": 0.13158155977725983,
+      "learning_rate": 2.1000779063941585e-05,
+      "loss": 0.0568,
+      "step": 2580
+    },
+    {
+      "epoch": 1.209498186922447,
+      "grad_norm": 0.15704578161239624,
+      "learning_rate": 2.0898100188218313e-05,
+      "loss": 0.0634,
+      "step": 2585
+    },
+    {
+      "epoch": 1.2118376418294536,
+      "grad_norm": 0.157634437084198,
+      "learning_rate": 2.0795492371706527e-05,
+      "loss": 0.0597,
+      "step": 2590
+    },
+    {
+      "epoch": 1.2141770967364605,
+      "grad_norm": 0.1686418503522873,
+      "learning_rate": 2.0692957391931485e-05,
+      "loss": 0.0654,
+      "step": 2595
+    },
+    {
+      "epoch": 1.216516551643467,
+      "grad_norm": 0.15651817619800568,
+      "learning_rate": 2.0590497025156676e-05,
+      "loss": 0.0591,
+      "step": 2600
+    },
+    {
+      "epoch": 1.2188560065504737,
+      "grad_norm": 0.16979578137397766,
+      "learning_rate": 2.0488113046353004e-05,
+      "loss": 0.0619,
+      "step": 2605
+    },
+    {
+      "epoch": 1.2211954614574805,
+      "grad_norm": 0.16162872314453125,
+      "learning_rate": 2.0385807229168104e-05,
+      "loss": 0.0653,
+      "step": 2610
+    },
+    {
+      "epoch": 1.223534916364487,
+      "grad_norm": 0.17247694730758667,
+      "learning_rate": 2.028358134589553e-05,
+      "loss": 0.0647,
+      "step": 2615
+    },
+    {
+      "epoch": 1.2258743712714937,
+      "grad_norm": 0.16605614125728607,
+      "learning_rate": 2.0181437167444144e-05,
+      "loss": 0.0632,
+      "step": 2620
+    },
+    {
+      "epoch": 1.2282138261785005,
+      "grad_norm": 0.14373953640460968,
+      "learning_rate": 2.0079376463307368e-05,
+      "loss": 0.0619,
+      "step": 2625
+    },
+    {
+      "epoch": 1.230553281085507,
+      "grad_norm": 0.1507708579301834,
+      "learning_rate": 1.9977401001532576e-05,
+      "loss": 0.0588,
+      "step": 2630
+    },
+    {
+      "epoch": 1.2328927359925137,
+      "grad_norm": 0.16803239285945892,
+      "learning_rate": 1.9875512548690433e-05,
+      "loss": 0.0591,
+      "step": 2635
+    },
+    {
+      "epoch": 1.2352321908995205,
+      "grad_norm": 0.16970671713352203,
+      "learning_rate": 1.9773712869844323e-05,
+      "loss": 0.0708,
+      "step": 2640
+    },
+    {
+      "epoch": 1.2375716458065271,
+      "grad_norm": 0.14553260803222656,
+      "learning_rate": 1.9672003728519728e-05,
+      "loss": 0.0609,
+      "step": 2645
+    },
+    {
+      "epoch": 1.2399111007135337,
+      "grad_norm": 0.1582130342721939,
+      "learning_rate": 1.957038688667374e-05,
+      "loss": 0.0642,
+      "step": 2650
+    },
+    {
+      "epoch": 1.2422505556205403,
+      "grad_norm": 0.1633974015712738,
+      "learning_rate": 1.9468864104664464e-05,
+      "loss": 0.0638,
+      "step": 2655
+    },
+    {
+      "epoch": 1.2445900105275471,
+      "grad_norm": 0.15454533696174622,
+      "learning_rate": 1.9367437141220594e-05,
+      "loss": 0.0643,
+      "step": 2660
+    },
+    {
+      "epoch": 1.2469294654345537,
+      "grad_norm": 0.14684420824050903,
+      "learning_rate": 1.9266107753410884e-05,
+      "loss": 0.0611,
+      "step": 2665
+    },
+    {
+      "epoch": 1.2492689203415603,
+      "grad_norm": 0.13646100461483002,
+      "learning_rate": 1.916487769661377e-05,
+      "loss": 0.055,
+      "step": 2670
+    },
+    {
+      "epoch": 1.2516083752485672,
+      "grad_norm": 0.15355093777179718,
+      "learning_rate": 1.9063748724486887e-05,
+      "loss": 0.0518,
+      "step": 2675
+    },
+    {
+      "epoch": 1.2539478301555738,
+      "grad_norm": 0.16275903582572937,
+      "learning_rate": 1.896272258893677e-05,
+      "loss": 0.0618,
+      "step": 2680
+    },
+    {
+      "epoch": 1.2562872850625804,
+      "grad_norm": 0.14904741942882538,
+      "learning_rate": 1.8861801040088445e-05,
+      "loss": 0.0553,
+      "step": 2685
+    },
+    {
+      "epoch": 1.258626739969587,
+      "grad_norm": 0.16873393952846527,
+      "learning_rate": 1.8760985826255125e-05,
+      "loss": 0.063,
+      "step": 2690
+    },
+    {
+      "epoch": 1.2609661948765938,
+      "grad_norm": 0.1706465184688568,
+      "learning_rate": 1.8660278693907962e-05,
+      "loss": 0.0645,
+      "step": 2695
+    },
+    {
+      "epoch": 1.2633056497836004,
+      "grad_norm": 0.14628972113132477,
+      "learning_rate": 1.8559681387645715e-05,
+      "loss": 0.0579,
+      "step": 2700
+    },
+    {
+      "epoch": 1.265645104690607,
+      "grad_norm": 0.1530769318342209,
+      "learning_rate": 1.8459195650164612e-05,
+      "loss": 0.0629,
+      "step": 2705
+    },
+    {
+      "epoch": 1.2679845595976138,
+      "grad_norm": 0.16960960626602173,
+      "learning_rate": 1.8358823222228097e-05,
+      "loss": 0.0627,
+      "step": 2710
+    },
+    {
+      "epoch": 1.2703240145046204,
+      "grad_norm": 0.1773470938205719,
+      "learning_rate": 1.8258565842636703e-05,
+      "loss": 0.0582,
+      "step": 2715
+    },
+    {
+      "epoch": 1.272663469411627,
+      "grad_norm": 0.162547767162323,
+      "learning_rate": 1.815842524819793e-05,
+      "loss": 0.0606,
+      "step": 2720
+    },
+    {
+      "epoch": 1.2750029243186338,
+      "grad_norm": 0.15616440773010254,
+      "learning_rate": 1.8058403173696152e-05,
+      "loss": 0.0967,
+      "step": 2725
+    },
+    {
+      "epoch": 1.2773423792256404,
+      "grad_norm": 0.1639740914106369,
+      "learning_rate": 1.7958501351862545e-05,
+      "loss": 0.0585,
+      "step": 2730
+    },
+    {
+      "epoch": 1.279681834132647,
+      "grad_norm": 0.16226714849472046,
+      "learning_rate": 1.7858721513345117e-05,
+      "loss": 0.0664,
+      "step": 2735
+    },
+    {
+      "epoch": 1.2820212890396538,
+      "grad_norm": 0.1616683155298233,
+      "learning_rate": 1.7759065386678674e-05,
+      "loss": 0.0548,
+      "step": 2740
+    },
+    {
+      "epoch": 1.2843607439466604,
+      "grad_norm": 0.15211808681488037,
+      "learning_rate": 1.765953469825494e-05,
+      "loss": 0.0577,
+      "step": 2745
+    },
+    {
+      "epoch": 1.286700198853667,
+      "grad_norm": 0.15566258132457733,
+      "learning_rate": 1.7560131172292556e-05,
+      "loss": 0.0632,
+      "step": 2750
+    },
+    {
+      "epoch": 1.2890396537606739,
+      "grad_norm": 0.14558292925357819,
+      "learning_rate": 1.7460856530807315e-05,
+      "loss": 0.0592,
+      "step": 2755
+    },
+    {
+      "epoch": 1.2913791086676805,
+      "grad_norm": 0.16073749959468842,
+      "learning_rate": 1.7361712493582242e-05,
+      "loss": 0.057,
+      "step": 2760
+    },
+    {
+      "epoch": 1.293718563574687,
+      "grad_norm": 0.14757321774959564,
+      "learning_rate": 1.726270077813788e-05,
+      "loss": 0.0603,
+      "step": 2765
+    },
+    {
+      "epoch": 1.2960580184816939,
+      "grad_norm": 0.17697452008724213,
+      "learning_rate": 1.7163823099702466e-05,
+      "loss": 0.0625,
+      "step": 2770
+    },
+    {
+      "epoch": 1.2983974733887005,
+      "grad_norm": 0.17188061773777008,
+      "learning_rate": 1.7065081171182264e-05,
+      "loss": 0.0585,
+      "step": 2775
+    },
+    {
+      "epoch": 1.300736928295707,
+      "grad_norm": 0.1670866757631302,
+      "learning_rate": 1.696647670313186e-05,
+      "loss": 0.0569,
+      "step": 2780
+    },
+    {
+      "epoch": 1.303076383202714,
+      "grad_norm": 0.15840460360050201,
+      "learning_rate": 1.6868011403724582e-05,
+      "loss": 0.0616,
+      "step": 2785
+    },
+    {
+      "epoch": 1.3054158381097205,
+      "grad_norm": 0.16217677295207977,
+      "learning_rate": 1.676968697872282e-05,
+      "loss": 0.063,
+      "step": 2790
+    },
+    {
+      "epoch": 1.307755293016727,
+      "grad_norm": 0.15185709297657013,
+      "learning_rate": 1.667150513144856e-05,
+      "loss": 0.0517,
+      "step": 2795
+    },
+    {
+      "epoch": 1.3100947479237337,
+      "grad_norm": 0.1776837259531021,
+      "learning_rate": 1.657346756275386e-05,
+      "loss": 0.0614,
+      "step": 2800
+    },
+    {
+      "epoch": 1.3124342028307403,
+      "grad_norm": 0.17348352074623108,
+      "learning_rate": 1.6475575970991313e-05,
+      "loss": 0.0593,
+      "step": 2805
+    },
+    {
+      "epoch": 1.3147736577377471,
+      "grad_norm": 0.1484275907278061,
+      "learning_rate": 1.6377832051984755e-05,
+      "loss": 0.0558,
+      "step": 2810
+    },
+    {
+      "epoch": 1.3171131126447537,
+      "grad_norm": 0.15603071451187134,
+      "learning_rate": 1.628023749899975e-05,
+      "loss": 0.0544,
+      "step": 2815
+    },
+    {
+      "epoch": 1.3194525675517603,
+      "grad_norm": 0.16645818948745728,
+      "learning_rate": 1.618279400271436e-05,
+      "loss": 0.0635,
+      "step": 2820
+    },
+    {
+      "epoch": 1.3217920224587671,
+      "grad_norm": 0.1469196379184723,
+      "learning_rate": 1.608550325118981e-05,
+      "loss": 0.0559,
+      "step": 2825
+    },
+    {
+      "epoch": 1.3241314773657737,
+      "grad_norm": 0.19546544551849365,
+      "learning_rate": 1.5988366929841254e-05,
+      "loss": 0.07,
+      "step": 2830
+    },
+    {
+      "epoch": 1.3264709322727803,
+      "grad_norm": 0.1820397824048996,
+      "learning_rate": 1.589138672140856e-05,
+      "loss": 0.0718,
+      "step": 2835
+    },
+    {
+      "epoch": 1.3288103871797872,
+      "grad_norm": 0.14199690520763397,
+      "learning_rate": 1.57945643059272e-05,
+      "loss": 0.0566,
+      "step": 2840
+    },
+    {
+      "epoch": 1.3311498420867938,
+      "grad_norm": 0.1681860238313675,
+      "learning_rate": 1.56979013606991e-05,
+      "loss": 0.0628,
+      "step": 2845
+    },
+    {
+      "epoch": 1.3334892969938004,
+      "grad_norm": 0.16973961889743805,
+      "learning_rate": 1.560139956026363e-05,
+      "loss": 0.058,
+      "step": 2850
+    },
+    {
+      "epoch": 1.3358287519008072,
+      "grad_norm": 0.15116538107395172,
+      "learning_rate": 1.5505060576368545e-05,
+      "loss": 0.059,
+      "step": 2855
+    },
+    {
+      "epoch": 1.3381682068078138,
+      "grad_norm": 0.1909974217414856,
+      "learning_rate": 1.5408886077941074e-05,
+      "loss": 0.066,
+      "step": 2860
+    },
+    {
+      "epoch": 1.3405076617148204,
+      "grad_norm": 0.17362594604492188,
+      "learning_rate": 1.531287773105895e-05,
+      "loss": 0.0602,
+      "step": 2865
+    },
+    {
+      "epoch": 1.3428471166218272,
+      "grad_norm": 0.16387321054935455,
+      "learning_rate": 1.5217037198921625e-05,
+      "loss": 0.0633,
+      "step": 2870
+    },
+    {
+      "epoch": 1.3451865715288338,
+      "grad_norm": 0.15900340676307678,
+      "learning_rate": 1.512136614182137e-05,
+      "loss": 0.0694,
+      "step": 2875
+    },
+    {
+      "epoch": 1.3475260264358404,
+      "grad_norm": 0.15678687393665314,
+      "learning_rate": 1.5025866217114592e-05,
+      "loss": 0.0531,
+      "step": 2880
+    },
+    {
+      "epoch": 1.3498654813428472,
+      "grad_norm": 0.17534701526165009,
+      "learning_rate": 1.4930539079193078e-05,
+      "loss": 0.0648,
+      "step": 2885
+    },
+    {
+      "epoch": 1.3522049362498538,
+      "grad_norm": 0.15762171149253845,
+      "learning_rate": 1.483538637945533e-05,
+      "loss": 0.0684,
+      "step": 2890
+    },
+    {
+      "epoch": 1.3545443911568604,
+      "grad_norm": 0.13863767683506012,
+      "learning_rate": 1.474040976627799e-05,
+      "loss": 0.055,
+      "step": 2895
+    },
+    {
+      "epoch": 1.3568838460638672,
+      "grad_norm": 0.1766417771577835,
+      "learning_rate": 1.4645610884987265e-05,
+      "loss": 0.1258,
+      "step": 2900
+    },
+    {
+      "epoch": 1.3592233009708738,
+      "grad_norm": 0.17303304374217987,
+      "learning_rate": 1.4550991377830426e-05,
+      "loss": 0.0607,
+      "step": 2905
+    },
+    {
+      "epoch": 1.3615627558778804,
+      "grad_norm": 0.17528830468654633,
+      "learning_rate": 1.4456552883947333e-05,
+      "loss": 0.0618,
+      "step": 2910
+    },
+    {
+      "epoch": 1.363902210784887,
+      "grad_norm": 0.16436804831027985,
+      "learning_rate": 1.4362297039342098e-05,
+      "loss": 0.0648,
+      "step": 2915
+    },
+    {
+      "epoch": 1.3662416656918939,
+      "grad_norm": 0.1755647212266922,
+      "learning_rate": 1.4268225476854704e-05,
+      "loss": 0.0681,
+      "step": 2920
+    },
+    {
+      "epoch": 1.3685811205989005,
+      "grad_norm": 0.15491195023059845,
+      "learning_rate": 1.4174339826132727e-05,
+      "loss": 0.0587,
+      "step": 2925
+    },
+    {
+      "epoch": 1.370920575505907,
+      "grad_norm": 0.1469022035598755,
+      "learning_rate": 1.4080641713603072e-05,
+      "loss": 0.0669,
+      "step": 2930
+    },
+    {
+      "epoch": 1.3732600304129137,
+      "grad_norm": 0.18021321296691895,
+      "learning_rate": 1.3987132762443873e-05,
+      "loss": 0.0601,
+      "step": 2935
+    },
+    {
+      "epoch": 1.3755994853199205,
+      "grad_norm": 0.1528056561946869,
+      "learning_rate": 1.3893814592556287e-05,
+      "loss": 0.0599,
+      "step": 2940
+    },
+    {
+      "epoch": 1.377938940226927,
+      "grad_norm": 0.15603193640708923,
+      "learning_rate": 1.3800688820536517e-05,
+      "loss": 0.0539,
+      "step": 2945
+    },
+    {
+      "epoch": 1.3802783951339337,
+      "grad_norm": 0.15558800101280212,
+      "learning_rate": 1.3707757059647725e-05,
+      "loss": 0.0671,
+      "step": 2950
+    },
+    {
+      "epoch": 1.3826178500409405,
+      "grad_norm": 0.156876340508461,
+      "learning_rate": 1.3615020919792151e-05,
+      "loss": 0.0586,
+      "step": 2955
+    },
+    {
+      "epoch": 1.384957304947947,
+      "grad_norm": 0.1524912714958191,
+      "learning_rate": 1.3522482007483172e-05,
+      "loss": 0.0662,
+      "step": 2960
+    },
+    {
+      "epoch": 1.3872967598549537,
+      "grad_norm": 0.15926173329353333,
+      "learning_rate": 1.3430141925817532e-05,
+      "loss": 0.0637,
+      "step": 2965
+    },
+    {
+      "epoch": 1.3896362147619605,
+      "grad_norm": 0.16642306745052338,
+      "learning_rate": 1.333800227444749e-05,
+      "loss": 0.067,
+      "step": 2970
+    },
+    {
+      "epoch": 1.3919756696689671,
+      "grad_norm": 0.13672704994678497,
+      "learning_rate": 1.324606464955318e-05,
+      "loss": 0.0618,
+      "step": 2975
+    },
+    {
+      "epoch": 1.3943151245759737,
+      "grad_norm": 0.16178594529628754,
+      "learning_rate": 1.3154330643814938e-05,
+      "loss": 0.0564,
+      "step": 2980
+    },
+    {
+      "epoch": 1.3966545794829806,
+      "grad_norm": 0.16218972206115723,
+      "learning_rate": 1.30628018463857e-05,
+      "loss": 0.0592,
+      "step": 2985
+    },
+    {
+      "epoch": 1.3989940343899872,
+      "grad_norm": 0.16459548473358154,
+      "learning_rate": 1.2971479842863465e-05,
+      "loss": 0.0533,
+      "step": 2990
+    },
+    {
+      "epoch": 1.4013334892969938,
+      "grad_norm": 0.14604820311069489,
+      "learning_rate": 1.2880366215263845e-05,
+      "loss": 0.0629,
+      "step": 2995
+    },
+    {
+      "epoch": 1.4036729442040006,
+      "grad_norm": 0.16797035932540894,
+      "learning_rate": 1.2789462541992676e-05,
+      "loss": 0.0553,
+      "step": 3000
+    },
+    {
+      "epoch": 1.4036729442040006,
+      "eval_loss": 0.07826597988605499,
+      "eval_runtime": 234.1597,
+      "eval_samples_per_second": 2.981,
+      "eval_steps_per_second": 0.747,
+      "step": 3000
+    },
+    {
+      "epoch": 1.4060123991110072,
+      "grad_norm": 0.1841832548379898,
+      "learning_rate": 1.2698770397818616e-05,
+      "loss": 0.13,
+      "step": 3005
+    },
+    {
+      "epoch": 1.4083518540180138,
+      "grad_norm": 0.1567966490983963,
+      "learning_rate": 1.2608291353845941e-05,
+      "loss": 0.063,
+      "step": 3010
+    },
+    {
+      "epoch": 1.4106913089250206,
+      "grad_norm": 0.14550812542438507,
+      "learning_rate": 1.2518026977487252e-05,
+      "loss": 0.0598,
+      "step": 3015
+    },
+    {
+      "epoch": 1.4130307638320272,
+      "grad_norm": 0.15681859850883484,
+      "learning_rate": 1.242797883243638e-05,
+      "loss": 0.063,
+      "step": 3020
+    },
+    {
+      "epoch": 1.4153702187390338,
+      "grad_norm": 0.16205298900604248,
+      "learning_rate": 1.233814847864127e-05,
+      "loss": 0.0691,
+      "step": 3025
+    },
+    {
+      "epoch": 1.4177096736460404,
+      "grad_norm": 0.16490066051483154,
+      "learning_rate": 1.224853747227698e-05,
+      "loss": 0.0628,
+      "step": 3030
+    },
+    {
+      "epoch": 1.4200491285530472,
+      "grad_norm": 0.19161145389080048,
+      "learning_rate": 1.2159147365718664e-05,
+      "loss": 0.0646,
+      "step": 3035
+    },
+    {
+      "epoch": 1.4223885834600538,
+      "grad_norm": 0.16144663095474243,
+      "learning_rate": 1.2069979707514766e-05,
+      "loss": 0.0542,
+      "step": 3040
+    },
+    {
+      "epoch": 1.4247280383670604,
+      "grad_norm": 0.15800189971923828,
+      "learning_rate": 1.1981036042360109e-05,
+      "loss": 0.059,
+      "step": 3045
+    },
+    {
+      "epoch": 1.427067493274067,
+      "grad_norm": 0.15965986251831055,
+      "learning_rate": 1.1892317911069212e-05,
+      "loss": 0.0575,
+      "step": 3050
+    },
+    {
+      "epoch": 1.4294069481810738,
+      "grad_norm": 0.15217727422714233,
+      "learning_rate": 1.1803826850549526e-05,
+      "loss": 0.0604,
+      "step": 3055
+    },
+    {
+      "epoch": 1.4317464030880804,
+      "grad_norm": 0.15865880250930786,
+      "learning_rate": 1.171556439377488e-05,
+      "loss": 0.0659,
+      "step": 3060
+    },
+    {
+      "epoch": 1.434085857995087,
+      "grad_norm": 0.18922170996665955,
+      "learning_rate": 1.1627532069758851e-05,
+      "loss": 0.0605,
+      "step": 3065
+    },
+    {
+      "epoch": 1.4364253129020939,
+      "grad_norm": 0.17972765862941742,
+      "learning_rate": 1.1539731403528353e-05,
+      "loss": 0.0754,
+      "step": 3070
+    },
+    {
+      "epoch": 1.4387647678091005,
+      "grad_norm": 0.18478457629680634,
+      "learning_rate": 1.1452163916097141e-05,
+      "loss": 0.0689,
+      "step": 3075
+    },
+    {
+      "epoch": 1.441104222716107,
+      "grad_norm": 0.16333162784576416,
+      "learning_rate": 1.1364831124439526e-05,
+      "loss": 0.0669,
+      "step": 3080
+    },
+    {
+      "epoch": 1.4434436776231139,
+      "grad_norm": 0.1580970138311386,
+      "learning_rate": 1.1277734541464072e-05,
+      "loss": 0.0596,
+      "step": 3085
+    },
+    {
+      "epoch": 1.4457831325301205,
+      "grad_norm": 0.17736057937145233,
+      "learning_rate": 1.1190875675987356e-05,
+      "loss": 0.057,
+      "step": 3090
+    },
+    {
+      "epoch": 1.448122587437127,
+      "grad_norm": 0.17096631228923798,
+      "learning_rate": 1.1104256032707894e-05,
+      "loss": 0.0671,
+      "step": 3095
+    },
+    {
+      "epoch": 1.450462042344134,
+      "grad_norm": 0.17101797461509705,
+      "learning_rate": 1.1017877112180009e-05,
+      "loss": 0.0583,
+      "step": 3100
+    },
+    {
+      "epoch": 1.4528014972511405,
+      "grad_norm": 0.1477900892496109,
+      "learning_rate": 1.0931740410787895e-05,
+      "loss": 0.0617,
+      "step": 3105
+    },
+    {
+      "epoch": 1.455140952158147,
+      "grad_norm": 0.16209881007671356,
+      "learning_rate": 1.0845847420719632e-05,
+      "loss": 0.0619,
+      "step": 3110
+    },
+    {
+      "epoch": 1.457480407065154,
+      "grad_norm": 0.16828244924545288,
+      "learning_rate": 1.076019962994141e-05,
+      "loss": 0.0572,
+      "step": 3115
+    },
+    {
+      "epoch": 1.4598198619721605,
+      "grad_norm": 0.1566639393568039,
+      "learning_rate": 1.0674798522171668e-05,
+      "loss": 0.0632,
+      "step": 3120
+    },
+    {
+      "epoch": 1.4621593168791671,
+      "grad_norm": 0.1641826629638672,
+      "learning_rate": 1.0589645576855477e-05,
+      "loss": 0.055,
+      "step": 3125
+    },
+    {
+      "epoch": 1.464498771786174,
+      "grad_norm": 0.1548743098974228,
+      "learning_rate": 1.0504742269138835e-05,
+      "loss": 0.0631,
+      "step": 3130
+    },
+    {
+      "epoch": 1.4668382266931805,
+      "grad_norm": 0.16996270418167114,
+      "learning_rate": 1.0420090069843167e-05,
+      "loss": 0.0557,
+      "step": 3135
+    },
+    {
+      "epoch": 1.4691776816001871,
+      "grad_norm": 0.18014395236968994,
+      "learning_rate": 1.0335690445439817e-05,
+      "loss": 0.0623,
+      "step": 3140
+    },
+    {
+      "epoch": 1.471517136507194,
+      "grad_norm": 0.17947927117347717,
+      "learning_rate": 1.0251544858024662e-05,
+      "loss": 0.056,
+      "step": 3145
+    },
+    {
+      "epoch": 1.4738565914142006,
+      "grad_norm": 0.19152309000492096,
+      "learning_rate": 1.0167654765292742e-05,
+      "loss": 0.0581,
+      "step": 3150
+    },
+    {
+      "epoch": 1.4761960463212072,
+      "grad_norm": 0.16634751856327057,
+      "learning_rate": 1.0084021620513079e-05,
+      "loss": 0.0619,
+      "step": 3155
+    },
+    {
+      "epoch": 1.4785355012282138,
+      "grad_norm": 0.19897738099098206,
+      "learning_rate": 1.0000646872503419e-05,
+      "loss": 0.0784,
+      "step": 3160
+    },
+    {
+      "epoch": 1.4808749561352204,
+      "grad_norm": 0.1585209220647812,
+      "learning_rate": 9.917531965605211e-06,
+      "loss": 0.062,
+      "step": 3165
+    },
+    {
+      "epoch": 1.4832144110422272,
+      "grad_norm": 0.18893523514270782,
+      "learning_rate": 9.83467833965852e-06,
+      "loss": 0.057,
+      "step": 3170
+    },
+    {
+      "epoch": 1.4855538659492338,
+      "grad_norm": 0.17129787802696228,
+      "learning_rate": 9.752087429977143e-06,
+      "loss": 0.0674,
+      "step": 3175
+    },
+    {
+      "epoch": 1.4878933208562404,
+      "grad_norm": 0.1496947705745697,
+      "learning_rate": 9.669760667323685e-06,
+      "loss": 0.0842,
+      "step": 3180
+    },
+    {
+      "epoch": 1.4902327757632472,
+      "grad_norm": 0.17729486525058746,
+      "learning_rate": 9.58769947788482e-06,
+      "loss": 0.0736,
+      "step": 3185
+    },
+    {
+      "epoch": 1.4925722306702538,
+      "grad_norm": 0.17603404819965363,
+      "learning_rate": 9.505905283246578e-06,
+      "loss": 0.0638,
+      "step": 3190
+    },
+    {
+      "epoch": 1.4949116855772604,
+      "grad_norm": 0.17529185116291046,
+      "learning_rate": 9.424379500369674e-06,
+      "loss": 0.0672,
+      "step": 3195
+    },
+    {
+      "epoch": 1.4972511404842672,
+      "grad_norm": 0.1493815779685974,
+      "learning_rate": 9.343123541565035e-06,
+      "loss": 0.058,
+      "step": 3200
+    },
+    {
+      "epoch": 1.4995905953912738,
+      "grad_norm": 0.15651191771030426,
+      "learning_rate": 9.26213881446926e-06,
+      "loss": 0.0588,
+      "step": 3205
+    },
+    {
+      "epoch": 1.5019300502982804,
+      "grad_norm": 0.15472684800624847,
+      "learning_rate": 9.181426722020297e-06,
+      "loss": 0.0626,
+      "step": 3210
+    },
+    {
+      "epoch": 1.5042695052052872,
+      "grad_norm": 0.1653946340084076,
+      "learning_rate": 9.100988662433082e-06,
+      "loss": 0.0824,
+      "step": 3215
+    },
+    {
+      "epoch": 1.5066089601122938,
+      "grad_norm": 0.1525607407093048,
+      "learning_rate": 9.020826029175384e-06,
+      "loss": 0.0621,
+      "step": 3220
+    },
+    {
+      "epoch": 1.5089484150193004,
+      "grad_norm": 0.18066176772117615,
+      "learning_rate": 8.94094021094358e-06,
+      "loss": 0.0632,
+      "step": 3225
+    },
+    {
+      "epoch": 1.5112878699263073,
+      "grad_norm": 0.13805319368839264,
+      "learning_rate": 8.86133259163869e-06,
+      "loss": 0.0638,
+      "step": 3230
+    },
+    {
+      "epoch": 1.5136273248333139,
+      "grad_norm": 0.1525852382183075,
+      "learning_rate": 8.782004550342323e-06,
+      "loss": 0.0586,
+      "step": 3235
+    },
+    {
+      "epoch": 1.5159667797403205,
+      "grad_norm": 0.16709508001804352,
+      "learning_rate": 8.702957461292846e-06,
+      "loss": 0.0632,
+      "step": 3240
+    },
+    {
+      "epoch": 1.5183062346473273,
+      "grad_norm": 0.1565026491880417,
+      "learning_rate": 8.624192693861544e-06,
+      "loss": 0.0578,
+      "step": 3245
+    },
+    {
+      "epoch": 1.5206456895543339,
+      "grad_norm": 0.16678424179553986,
+      "learning_rate": 8.545711612528915e-06,
+      "loss": 0.063,
+      "step": 3250
+    },
+    {
+      "epoch": 1.5229851444613405,
+      "grad_norm": 0.19322334229946136,
+      "learning_rate": 8.467515576861002e-06,
+      "loss": 0.0657,
+      "step": 3255
+    },
+    {
+      "epoch": 1.5253245993683473,
+      "grad_norm": 0.1959463208913803,
+      "learning_rate": 8.389605941485893e-06,
+      "loss": 0.1184,
+      "step": 3260
+    },
+    {
+      "epoch": 1.5276640542753537,
+      "grad_norm": 0.1491149216890335,
+      "learning_rate": 8.31198405607019e-06,
+      "loss": 0.055,
+      "step": 3265
+    },
+    {
+      "epoch": 1.5300035091823605,
+      "grad_norm": 0.1650906205177307,
+      "learning_rate": 8.234651265295702e-06,
+      "loss": 0.0669,
+      "step": 3270
+    },
+    {
+      "epoch": 1.5323429640893673,
+      "grad_norm": 0.13482898473739624,
+      "learning_rate": 8.15760890883607e-06,
+      "loss": 0.0577,
+      "step": 3275
+    },
+    {
+      "epoch": 1.5346824189963737,
+      "grad_norm": 0.15607015788555145,
+      "learning_rate": 8.080858321333633e-06,
+      "loss": 0.063,
+      "step": 3280
+    },
+    {
+      "epoch": 1.5370218739033805,
+      "grad_norm": 0.15770550072193146,
+      "learning_rate": 8.004400832376258e-06,
+      "loss": 0.0617,
+      "step": 3285
+    },
+    {
+      "epoch": 1.5393613288103873,
+      "grad_norm": 0.15652437508106232,
+      "learning_rate": 7.928237766474314e-06,
+      "loss": 0.0668,
+      "step": 3290
+    },
+    {
+      "epoch": 1.5417007837173937,
+      "grad_norm": 0.174765482544899,
+      "learning_rate": 7.852370443037751e-06,
+      "loss": 0.0604,
+      "step": 3295
+    },
+    {
+      "epoch": 1.5440402386244005,
+      "grad_norm": 0.14293302595615387,
+      "learning_rate": 7.776800176353227e-06,
+      "loss": 0.0852,
+      "step": 3300
+    },
+    {
+      "epoch": 1.5463796935314071,
+      "grad_norm": 0.16163261234760284,
+      "learning_rate": 7.701528275561348e-06,
+      "loss": 0.0571,
+      "step": 3305
+    },
+    {
+      "epoch": 1.5487191484384137,
+      "grad_norm": 0.13872607052326202,
+      "learning_rate": 7.626556044633951e-06,
+      "loss": 0.0606,
+      "step": 3310
+    },
+    {
+      "epoch": 1.5510586033454206,
+      "grad_norm": 0.16931146383285522,
+      "learning_rate": 7.551884782351587e-06,
+      "loss": 0.0582,
+      "step": 3315
+    },
+    {
+      "epoch": 1.5533980582524272,
+      "grad_norm": 0.16759054362773895,
+      "learning_rate": 7.477515782280939e-06,
+      "loss": 0.0656,
+      "step": 3320
+    },
+    {
+      "epoch": 1.5557375131594338,
+      "grad_norm": 0.15919411182403564,
+      "learning_rate": 7.403450332752493e-06,
+      "loss": 0.0685,
+      "step": 3325
+    },
+    {
+      "epoch": 1.5580769680664406,
+      "grad_norm": 0.1733466535806656,
+      "learning_rate": 7.329689716838148e-06,
+      "loss": 0.06,
+      "step": 3330
+    },
+    {
+      "epoch": 1.5604164229734472,
+      "grad_norm": 0.15883946418762207,
+      "learning_rate": 7.256235212329055e-06,
+      "loss": 0.0537,
+      "step": 3335
+    },
+    {
+      "epoch": 1.5627558778804538,
+      "grad_norm": 0.20026150345802307,
+      "learning_rate": 7.183088091713405e-06,
+      "loss": 0.0676,
+      "step": 3340
+    },
+    {
+      "epoch": 1.5650953327874606,
+      "grad_norm": 0.14357200264930725,
+      "learning_rate": 7.110249622154469e-06,
+      "loss": 0.0596,
+      "step": 3345
+    },
+    {
+      "epoch": 1.5674347876944672,
+      "grad_norm": 0.15919791162014008,
+      "learning_rate": 7.037721065468591e-06,
+      "loss": 0.0504,
+      "step": 3350
+    },
+    {
+      "epoch": 1.5697742426014738,
+      "grad_norm": 0.16171149909496307,
+      "learning_rate": 6.96550367810333e-06,
+      "loss": 0.0688,
+      "step": 3355
+    },
+    {
+      "epoch": 1.5721136975084806,
+      "grad_norm": 0.17881818115711212,
+      "learning_rate": 6.89359871111572e-06,
+      "loss": 0.0545,
+      "step": 3360
+    },
+    {
+      "epoch": 1.5744531524154872,
+      "grad_norm": 0.1703099012374878,
+      "learning_rate": 6.822007410150591e-06,
+      "loss": 0.0624,
+      "step": 3365
+    },
+    {
+      "epoch": 1.5767926073224938,
+      "grad_norm": 0.17320412397384644,
+      "learning_rate": 6.750731015418959e-06,
+      "loss": 0.0659,
+      "step": 3370
+    },
+    {
+      "epoch": 1.5791320622295006,
+      "grad_norm": 0.15360009670257568,
+      "learning_rate": 6.679770761676596e-06,
+      "loss": 0.062,
+      "step": 3375
+    },
+    {
+      "epoch": 1.5814715171365072,
+      "grad_norm": 0.15514342486858368,
+      "learning_rate": 6.609127878202581e-06,
+      "loss": 0.0522,
+      "step": 3380
+    },
+    {
+      "epoch": 1.5838109720435138,
+      "grad_norm": 0.14531250298023224,
+      "learning_rate": 6.538803588778064e-06,
+      "loss": 0.0588,
+      "step": 3385
+    },
+    {
+      "epoch": 1.5861504269505207,
+      "grad_norm": 0.16243524849414825,
+      "learning_rate": 6.468799111665003e-06,
+      "loss": 0.0641,
+      "step": 3390
+    },
+    {
+      "epoch": 1.588489881857527,
+      "grad_norm": 0.19768266379833221,
+      "learning_rate": 6.399115659585103e-06,
+      "loss": 0.0668,
+      "step": 3395
+    },
+    {
+      "epoch": 1.5908293367645339,
+      "grad_norm": 0.1622917205095291,
+      "learning_rate": 6.329754439698804e-06,
+      "loss": 0.0531,
+      "step": 3400
+    },
+    {
+      "epoch": 1.5931687916715407,
+      "grad_norm": 0.18452420830726624,
+      "learning_rate": 6.260716653584359e-06,
+      "loss": 0.0992,
+      "step": 3405
+    },
+    {
+      "epoch": 1.595508246578547,
+      "grad_norm": 0.16582529246807098,
+      "learning_rate": 6.192003497217016e-06,
+      "loss": 0.0574,
+      "step": 3410
+    },
+    {
+      "epoch": 1.5978477014855539,
+      "grad_norm": 0.1616390198469162,
+      "learning_rate": 6.123616160948295e-06,
+      "loss": 0.0616,
+      "step": 3415
+    },
+    {
+      "epoch": 1.6001871563925605,
+      "grad_norm": 0.16133980453014374,
+      "learning_rate": 6.055555829485396e-06,
+      "loss": 0.0584,
+      "step": 3420
+    },
+    {
+      "epoch": 1.602526611299567,
+      "grad_norm": 0.14773264527320862,
+      "learning_rate": 5.98782368187063e-06,
+      "loss": 0.0557,
+      "step": 3425
+    },
+    {
+      "epoch": 1.604866066206574,
+      "grad_norm": 0.1554953008890152,
+      "learning_rate": 5.920420891461046e-06,
+      "loss": 0.0618,
+      "step": 3430
+    },
+    {
+      "epoch": 1.6072055211135805,
+      "grad_norm": 0.1828761100769043,
+      "learning_rate": 5.853348625908048e-06,
+      "loss": 0.0623,
+      "step": 3435
+    },
+    {
+      "epoch": 1.609544976020587,
+      "grad_norm": 0.1556580513715744,
+      "learning_rate": 5.786608047137229e-06,
+      "loss": 0.0564,
+      "step": 3440
+    },
+    {
+      "epoch": 1.611884430927594,
+      "grad_norm": 0.1514003425836563,
+      "learning_rate": 5.720200311328178e-06,
+      "loss": 0.0595,
+      "step": 3445
+    },
+    {
+      "epoch": 1.6142238858346005,
+      "grad_norm": 0.14597107470035553,
+      "learning_rate": 5.654126568894513e-06,
+      "loss": 0.0595,
+      "step": 3450
+    },
+    {
+      "epoch": 1.6165633407416071,
+      "grad_norm": 0.17368368804454803,
+      "learning_rate": 5.5883879644638925e-06,
+      "loss": 0.0606,
+      "step": 3455
+    },
+    {
+      "epoch": 1.618902795648614,
+      "grad_norm": 0.16808243095874786,
+      "learning_rate": 5.522985636858239e-06,
+      "loss": 0.058,
+      "step": 3460
+    },
+    {
+      "epoch": 1.6212422505556205,
+      "grad_norm": 0.1682329773902893,
+      "learning_rate": 5.4579207190739775e-06,
+      "loss": 0.0578,
+      "step": 3465
+    },
+    {
+      "epoch": 1.6235817054626271,
+      "grad_norm": 0.1642475128173828,
+      "learning_rate": 5.393194338262428e-06,
+      "loss": 0.0651,
+      "step": 3470
+    },
+    {
+      "epoch": 1.625921160369634,
+      "grad_norm": 0.18125009536743164,
+      "learning_rate": 5.328807615710246e-06,
+      "loss": 0.0586,
+      "step": 3475
+    },
+    {
+      "epoch": 1.6282606152766406,
+      "grad_norm": 0.14442673325538635,
+      "learning_rate": 5.264761666820054e-06,
+      "loss": 0.059,
+      "step": 3480
+    },
+    {
+      "epoch": 1.6306000701836472,
+      "grad_norm": 0.1675989329814911,
+      "learning_rate": 5.201057601091056e-06,
+      "loss": 0.0574,
+      "step": 3485
+    },
+    {
+      "epoch": 1.632939525090654,
+      "grad_norm": 0.17742133140563965,
+      "learning_rate": 5.137696522099858e-06,
+      "loss": 0.0577,
+      "step": 3490
+    },
+    {
+      "epoch": 1.6352789799976606,
+      "grad_norm": 0.15731069445610046,
+      "learning_rate": 5.074679527481349e-06,
+      "loss": 0.055,
+      "step": 3495
+    },
+    {
+      "epoch": 1.6376184349046672,
+      "grad_norm": 0.17156293988227844,
+      "learning_rate": 5.012007708909661e-06,
+      "loss": 0.057,
+      "step": 3500
+    },
+    {
+      "epoch": 1.6376184349046672,
+      "eval_loss": 0.07722621411085129,
+      "eval_runtime": 235.8253,
+      "eval_samples_per_second": 2.96,
+      "eval_steps_per_second": 0.742,
+      "step": 3500
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4274,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.4609092997910364e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}