diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8421 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.1765497760024717,
+  "eval_steps": 500,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 478.3545837402344,
+      "learning_rate": 5e-06,
+      "loss": 31.0705,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 434.907958984375,
+      "learning_rate": 1e-05,
+      "loss": 28.891,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 368.5296325683594,
+      "learning_rate": 1.5e-05,
+      "loss": 25.5933,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 503.1113586425781,
+      "learning_rate": 2e-05,
+      "loss": 20.8527,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 368.26031494140625,
+      "learning_rate": 2.5e-05,
+      "loss": 15.4835,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 74.01641082763672,
+      "learning_rate": 3e-05,
+      "loss": 11.9051,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 24.374834060668945,
+      "learning_rate": 3.5e-05,
+      "loss": 10.515,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.605514526367188,
+      "learning_rate": 4e-05,
+      "loss": 9.4836,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.240267753601074,
+      "learning_rate": 4.5e-05,
+      "loss": 8.9312,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.362184524536133,
+      "learning_rate": 5e-05,
+      "loss": 8.8889,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.066555023193359,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 8.8341,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.799922943115234,
+      "learning_rate": 6e-05,
+      "loss": 8.8406,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9685542583465576,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 8.8038,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.083511829376221,
+      "learning_rate": 7e-05,
+      "loss": 8.7779,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.640880107879639,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 8.767,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.425374984741211,
+      "learning_rate": 8e-05,
+      "loss": 8.6288,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.693755149841309,
+      "learning_rate": 8.5e-05,
+      "loss": 8.541,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.871931076049805,
+      "learning_rate": 9e-05,
+      "loss": 8.2984,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.55854320526123,
+      "learning_rate": 9.5e-05,
+      "loss": 8.2864,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.773725509643555,
+      "learning_rate": 0.0001,
+      "loss": 7.997,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.709348678588867,
+      "learning_rate": 9.999999462732701e-05,
+      "loss": 7.8218,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.505725383758545,
+      "learning_rate": 9.999997850930916e-05,
+      "loss": 7.7296,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.4634528160095215,
+      "learning_rate": 9.999995164594991e-05,
+      "loss": 7.5755,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.271254539489746,
+      "learning_rate": 9.999991403725507e-05,
+      "loss": 7.4039,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.218053340911865,
+      "learning_rate": 9.99998656832327e-05,
+      "loss": 7.2513,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.235344886779785,
+      "learning_rate": 9.999980658389317e-05,
+      "loss": 6.9604,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.064645767211914,
+      "learning_rate": 9.999973673924921e-05,
+      "loss": 7.0002,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.520571708679199,
+      "learning_rate": 9.999965614931583e-05,
+      "loss": 6.889,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.385054588317871,
+      "learning_rate": 9.999956481411035e-05,
+      "loss": 6.7944,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.702651023864746,
+      "learning_rate": 9.999946273365238e-05,
+      "loss": 6.6166,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.620483875274658,
+      "learning_rate": 9.999934990796387e-05,
+      "loss": 6.5376,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.1606404781341553,
+      "learning_rate": 9.999922633706906e-05,
+      "loss": 6.4425,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.3338799476623535,
+      "learning_rate": 9.999909202099453e-05,
+      "loss": 6.4766,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1104202270507812,
+      "learning_rate": 9.999894695976912e-05,
+      "loss": 6.3852,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.746636390686035,
+      "learning_rate": 9.9998791153424e-05,
+      "loss": 6.2562,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0738823413848877,
+      "learning_rate": 9.999862460199267e-05,
+      "loss": 6.2296,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.2022833824157715,
+      "learning_rate": 9.999844730551092e-05,
+      "loss": 6.0873,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.044377565383911,
+      "learning_rate": 9.999825926401686e-05,
+      "loss": 6.1572,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.589770555496216,
+      "learning_rate": 9.999806047755088e-05,
+      "loss": 6.1705,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.060673236846924,
+      "learning_rate": 9.999785094615573e-05,
+      "loss": 6.082,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.717599868774414,
+      "learning_rate": 9.99976306698764e-05,
+      "loss": 5.878,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8951687812805176,
+      "learning_rate": 9.999739964876026e-05,
+      "loss": 5.8594,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.317716598510742,
+      "learning_rate": 9.999715788285695e-05,
+      "loss": 5.8505,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.033480644226074,
+      "learning_rate": 9.999690537221844e-05,
+      "loss": 5.6895,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.143639087677002,
+      "learning_rate": 9.999664211689897e-05,
+      "loss": 5.6906,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3342349529266357,
+      "learning_rate": 9.999636811695513e-05,
+      "loss": 5.7538,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1114284992218018,
+      "learning_rate": 9.99960833724458e-05,
+      "loss": 5.6346,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2238693237304688,
+      "learning_rate": 9.999578788343218e-05,
+      "loss": 5.6705,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 17.501359939575195,
+      "learning_rate": 9.999548164997778e-05,
+      "loss": 5.5411,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1959753036499023,
+      "learning_rate": 9.999516467214839e-05,
+      "loss": 5.5846,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9218592643737793,
+      "learning_rate": 9.999483695001214e-05,
+      "loss": 5.5611,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.955921173095703,
+      "learning_rate": 9.999449848363946e-05,
+      "loss": 5.4486,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.894824743270874,
+      "learning_rate": 9.999414927310309e-05,
+      "loss": 5.4841,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.377030611038208,
+      "learning_rate": 9.99937893184781e-05,
+      "loss": 5.4896,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.277271032333374,
+      "learning_rate": 9.99934186198418e-05,
+      "loss": 5.4607,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7723071575164795,
+      "learning_rate": 9.999303717727389e-05,
+      "loss": 5.4045,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5409841537475586,
+      "learning_rate": 9.999264499085633e-05,
+      "loss": 5.524,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4675724506378174,
+      "learning_rate": 9.999224206067341e-05,
+      "loss": 5.4212,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.626497268676758,
+      "learning_rate": 9.999182838681171e-05,
+      "loss": 5.3312,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8383604288101196,
+      "learning_rate": 9.999140396936016e-05,
+      "loss": 5.2254,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.275822639465332,
+      "learning_rate": 9.999096880840994e-05,
+      "loss": 5.2645,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.082676887512207,
+      "learning_rate": 9.999052290405459e-05,
+      "loss": 5.2825,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.158535957336426,
+      "learning_rate": 9.999006625638994e-05,
+      "loss": 5.2639,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8159430027008057,
+      "learning_rate": 9.998959886551411e-05,
+      "loss": 5.2209,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6903239488601685,
+      "learning_rate": 9.998912073152755e-05,
+      "loss": 5.2021,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.360044240951538,
+      "learning_rate": 9.998863185453302e-05,
+      "loss": 5.1197,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9583823680877686,
+      "learning_rate": 9.998813223463557e-05,
+      "loss": 5.2302,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.666857123374939,
+      "learning_rate": 9.998762187194258e-05,
+      "loss": 5.2472,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.365285634994507,
+      "learning_rate": 9.998710076656373e-05,
+      "loss": 4.9995,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.383746862411499,
+      "learning_rate": 9.998656891861102e-05,
+      "loss": 5.1372,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.203564405441284,
+      "learning_rate": 9.998602632819874e-05,
+      "loss": 5.2314,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.817901849746704,
+      "learning_rate": 9.998547299544348e-05,
+      "loss": 5.1666,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7917828559875488,
+      "learning_rate": 9.99849089204642e-05,
+      "loss": 5.1041,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9972175359725952,
+      "learning_rate": 9.998433410338206e-05,
+      "loss": 5.056,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5718955993652344,
+      "learning_rate": 9.998374854432064e-05,
+      "loss": 5.087,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9581388235092163,
+      "learning_rate": 9.998315224340577e-05,
+      "loss": 5.0483,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5632574558258057,
+      "learning_rate": 9.99825452007656e-05,
+      "loss": 5.052,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6055289506912231,
+      "learning_rate": 9.998192741653057e-05,
+      "loss": 5.0023,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5558191537857056,
+      "learning_rate": 9.998129889083346e-05,
+      "loss": 4.9542,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1225342750549316,
+      "learning_rate": 9.998065962380936e-05,
+      "loss": 4.9237,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.750321388244629,
+      "learning_rate": 9.998000961559562e-05,
+      "loss": 4.8812,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6736820936203003,
+      "learning_rate": 9.997934886633195e-05,
+      "loss": 4.9762,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9538848400115967,
+      "learning_rate": 9.997867737616035e-05,
+      "loss": 4.921,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1453213691711426,
+      "learning_rate": 9.997799514522513e-05,
+      "loss": 4.9797,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.924251675605774,
+      "learning_rate": 9.99773021736729e-05,
+      "loss": 5.0381,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5894109010696411,
+      "learning_rate": 9.99765984616526e-05,
+      "loss": 4.9416,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.359071969985962,
+      "learning_rate": 9.997588400931542e-05,
+      "loss": 4.8131,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.794271469116211,
+      "learning_rate": 9.997515881681496e-05,
+      "loss": 4.9064,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8491193056106567,
+      "learning_rate": 9.9974422884307e-05,
+      "loss": 4.8096,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.226208448410034,
+      "learning_rate": 9.997367621194977e-05,
+      "loss": 4.8459,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7671284675598145,
+      "learning_rate": 9.99729187999037e-05,
+      "loss": 4.845,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7640037536621094,
+      "learning_rate": 9.997215064833155e-05,
+      "loss": 4.7431,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5617685317993164,
+      "learning_rate": 9.99713717573984e-05,
+      "loss": 4.7976,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4372621774673462,
+      "learning_rate": 9.997058212727169e-05,
+      "loss": 4.7013,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4049116373062134,
+      "learning_rate": 9.996978175812105e-05,
+      "loss": 4.8108,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.724307894706726,
+      "learning_rate": 9.996897065011852e-05,
+      "loss": 4.7491,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.133436679840088,
+      "learning_rate": 9.996814880343842e-05,
+      "loss": 4.6389,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8031115531921387,
+      "learning_rate": 9.996731621825736e-05,
+      "loss": 4.666,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0313258171081543,
+      "learning_rate": 9.996647289475426e-05,
+      "loss": 4.7464,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.927821636199951,
+      "learning_rate": 9.996561883311036e-05,
+      "loss": 4.8532,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.933828592300415,
+      "learning_rate": 9.996475403350921e-05,
+      "loss": 4.6774,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4829879999160767,
+      "learning_rate": 9.996387849613667e-05,
+      "loss": 4.6889,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7947884798049927,
+      "learning_rate": 9.996299222118086e-05,
+      "loss": 4.7092,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0747809410095215,
+      "learning_rate": 9.99620952088323e-05,
+      "loss": 4.6474,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5779904127120972,
+      "learning_rate": 9.996118745928373e-05,
+      "loss": 4.6678,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6540535688400269,
+      "learning_rate": 9.996026897273024e-05,
+      "loss": 4.7449,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4234607219696045,
+      "learning_rate": 9.995933974936922e-05,
+      "loss": 4.7696,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.427040696144104,
+      "learning_rate": 9.995839978940036e-05,
+      "loss": 4.7433,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.029886722564697,
+      "learning_rate": 9.995744909302567e-05,
+      "loss": 4.6948,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.719966173171997,
+      "learning_rate": 9.995648766044947e-05,
+      "loss": 4.7001,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.479351282119751,
+      "learning_rate": 9.995551549187836e-05,
+      "loss": 4.6948,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.962671160697937,
+      "learning_rate": 9.995453258752127e-05,
+      "loss": 4.7104,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4135764837265015,
+      "learning_rate": 9.995353894758945e-05,
+      "loss": 4.6672,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4754066467285156,
+      "learning_rate": 9.995253457229643e-05,
+      "loss": 4.7024,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.91579008102417,
+      "learning_rate": 9.995151946185804e-05,
+      "loss": 4.6623,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.609208106994629,
+      "learning_rate": 9.995049361649247e-05,
+      "loss": 4.5949,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2482264041900635,
+      "learning_rate": 9.994945703642013e-05,
+      "loss": 4.5112,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6363732814788818,
+      "learning_rate": 9.994840972186386e-05,
+      "loss": 4.6106,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3510633707046509,
+      "learning_rate": 9.994735167304866e-05,
+      "loss": 4.4927,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.564787745475769,
+      "learning_rate": 9.994628289020195e-05,
+      "loss": 4.6206,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.122218370437622,
+      "learning_rate": 9.994520337355344e-05,
+      "loss": 4.6646,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5749847888946533,
+      "learning_rate": 9.994411312333507e-05,
+      "loss": 4.5621,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8749823570251465,
+      "learning_rate": 9.99430121397812e-05,
+      "loss": 4.5092,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6864360570907593,
+      "learning_rate": 9.994190042312839e-05,
+      "loss": 4.6384,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8673864603042603,
+      "learning_rate": 9.994077797361559e-05,
+      "loss": 4.5146,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3098419904708862,
+      "learning_rate": 9.9939644791484e-05,
+      "loss": 4.5066,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3702442646026611,
+      "learning_rate": 9.993850087697716e-05,
+      "loss": 4.4408,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4416930675506592,
+      "learning_rate": 9.99373462303409e-05,
+      "loss": 4.6322,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1168832778930664,
+      "learning_rate": 9.993618085182337e-05,
+      "loss": 4.4886,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3242096900939941,
+      "learning_rate": 9.993500474167501e-05,
+      "loss": 4.5532,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3831658363342285,
+      "learning_rate": 9.99338179001486e-05,
+      "loss": 4.6082,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3776096105575562,
+      "learning_rate": 9.993262032749914e-05,
+      "loss": 4.5082,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.313111662864685,
+      "learning_rate": 9.993141202398407e-05,
+      "loss": 4.4713,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3964635133743286,
+      "learning_rate": 9.993019298986301e-05,
+      "loss": 4.4869,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.381272554397583,
+      "learning_rate": 9.992896322539796e-05,
+      "loss": 4.5183,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4879767894744873,
+      "learning_rate": 9.992772273085319e-05,
+      "loss": 4.4806,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.558326005935669,
+      "learning_rate": 9.99264715064953e-05,
+      "loss": 4.4681,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5182600021362305,
+      "learning_rate": 9.992520955259319e-05,
+      "loss": 4.3681,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3877164125442505,
+      "learning_rate": 9.992393686941808e-05,
+      "loss": 4.5626,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.470719575881958,
+      "learning_rate": 9.992265345724344e-05,
+      "loss": 4.3964,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1671606302261353,
+      "learning_rate": 9.99213593163451e-05,
+      "loss": 4.4275,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.224204659461975,
+      "learning_rate": 9.992005444700121e-05,
+      "loss": 4.3805,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3641725778579712,
+      "learning_rate": 9.991873884949214e-05,
+      "loss": 4.4337,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4890764951705933,
+      "learning_rate": 9.991741252410067e-05,
+      "loss": 4.4221,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3305805921554565,
+      "learning_rate": 9.99160754711118e-05,
+      "loss": 4.4163,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3834174871444702,
+      "learning_rate": 9.99147276908129e-05,
+      "loss": 4.5121,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4710280895233154,
+      "learning_rate": 9.991336918349361e-05,
+      "loss": 4.3974,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.387502908706665,
+      "learning_rate": 9.991199994944586e-05,
+      "loss": 4.5384,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.301598310470581,
+      "learning_rate": 9.991061998896393e-05,
+      "loss": 4.3673,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2739602327346802,
+      "learning_rate": 9.99092293023444e-05,
+      "loss": 4.3824,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3726129531860352,
+      "learning_rate": 9.990782788988611e-05,
+      "loss": 4.362,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2516895532608032,
+      "learning_rate": 9.990641575189022e-05,
+      "loss": 4.3577,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.503158450126648,
+      "learning_rate": 9.990499288866024e-05,
+      "loss": 4.4,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3117765188217163,
+      "learning_rate": 9.990355930050194e-05,
+      "loss": 4.3725,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5629382133483887,
+      "learning_rate": 9.990211498772341e-05,
+      "loss": 4.3329,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9236044883728027,
+      "learning_rate": 9.990065995063506e-05,
+      "loss": 4.407,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7203289270401,
+      "learning_rate": 9.989919418954956e-05,
+      "loss": 4.3945,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.876997947692871,
+      "learning_rate": 9.989771770478191e-05,
+      "loss": 4.3066,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3711439371109009,
+      "learning_rate": 9.989623049664944e-05,
+      "loss": 4.3752,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.544084906578064,
+      "learning_rate": 9.989473256547175e-05,
+      "loss": 4.282,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3692681789398193,
+      "learning_rate": 9.989322391157076e-05,
+      "loss": 4.289,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6250331401824951,
+      "learning_rate": 9.989170453527068e-05,
+      "loss": 4.3617,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3970587253570557,
+      "learning_rate": 9.989017443689805e-05,
+      "loss": 4.3184,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2188156843185425,
+      "learning_rate": 9.988863361678169e-05,
+      "loss": 4.2004,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.231415033340454,
+      "learning_rate": 9.988708207525272e-05,
+      "loss": 4.3254,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4238386154174805,
+      "learning_rate": 9.98855198126446e-05,
+      "loss": 4.3202,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4838637113571167,
+      "learning_rate": 9.988394682929305e-05,
+      "loss": 4.3413,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3006327152252197,
+      "learning_rate": 9.988236312553614e-05,
+      "loss": 4.233,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4355450868606567,
+      "learning_rate": 9.988076870171419e-05,
+      "loss": 4.3761,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6764552593231201,
+      "learning_rate": 9.987916355816988e-05,
+      "loss": 4.3285,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.336979627609253,
+      "learning_rate": 9.987754769524814e-05,
+      "loss": 4.2272,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1556148529052734,
+      "learning_rate": 9.987592111329626e-05,
+      "loss": 4.3399,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1569610834121704,
+      "learning_rate": 9.987428381266377e-05,
+      "loss": 4.215,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8817005157470703,
+      "learning_rate": 9.987263579370257e-05,
+      "loss": 4.3409,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1100988388061523,
+      "learning_rate": 9.987097705676681e-05,
+      "loss": 4.2823,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.451621174812317,
+      "learning_rate": 9.986930760221296e-05,
+      "loss": 4.2739,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6406223773956299,
+      "learning_rate": 9.986762743039983e-05,
+      "loss": 4.2309,
+      "step": 885
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4428579807281494,
+      "learning_rate": 9.986593654168846e-05,
+      "loss": 4.2729,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.344936490058899,
+      "learning_rate": 9.986423493644225e-05,
+      "loss": 4.3755,
+      "step": 895
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4583784341812134,
+      "learning_rate": 9.98625226150269e-05,
+      "loss": 4.2583,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.374634027481079,
+      "learning_rate": 9.986079957781038e-05,
+      "loss": 4.3451,
+      "step": 905
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3091877698898315,
+      "learning_rate": 9.9859065825163e-05,
+      "loss": 4.2927,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.142930269241333,
+      "learning_rate": 9.985732135745733e-05,
+      "loss": 4.2664,
+      "step": 915
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2377405166625977,
+      "learning_rate": 9.985556617506828e-05,
+      "loss": 4.2192,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3546297550201416,
+      "learning_rate": 9.985380027837308e-05,
+      "loss": 4.1331,
+      "step": 925
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7205119132995605,
+      "learning_rate": 9.98520236677512e-05,
+      "loss": 4.2702,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3901563882827759,
+      "learning_rate": 9.985023634358443e-05,
+      "loss": 4.2802,
+      "step": 935
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3179224729537964,
+      "learning_rate": 9.984843830625691e-05,
+      "loss": 4.2645,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3709591627120972,
+      "learning_rate": 9.984662955615507e-05,
+      "loss": 4.335,
+      "step": 945
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6397948265075684,
+      "learning_rate": 9.984481009366758e-05,
+      "loss": 4.2984,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.349935531616211,
+      "learning_rate": 9.984297991918548e-05,
+      "loss": 4.1684,
+      "step": 955
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2610434293746948,
+      "learning_rate": 9.984113903310206e-05,
+      "loss": 4.1968,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2904622554779053,
+      "learning_rate": 9.983928743581296e-05,
+      "loss": 4.2086,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1896486282348633,
+      "learning_rate": 9.983742512771611e-05,
+      "loss": 4.2377,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2967603206634521,
+      "learning_rate": 9.983555210921171e-05,
+      "loss": 4.2007,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2947237491607666,
+      "learning_rate": 9.983366838070231e-05,
+      "loss": 4.3303,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2442939281463623,
+      "learning_rate": 9.983177394259271e-05,
+      "loss": 4.2092,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1637173891067505,
+      "learning_rate": 9.982986879529006e-05,
+      "loss": 4.3049,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2222923040390015,
+      "learning_rate": 9.982795293920377e-05,
+      "loss": 4.2021,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.424842357635498,
+      "learning_rate": 9.982602637474558e-05,
+      "loss": 4.2842,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3271372318267822,
+      "learning_rate": 9.982408910232952e-05,
+      "loss": 4.1957,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7598788738250732,
+      "learning_rate": 9.982214112237193e-05,
+      "loss": 4.134,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6390225887298584,
+      "learning_rate": 9.982018243529146e-05,
+      "loss": 4.2221,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.27547287940979,
+      "learning_rate": 9.981821304150901e-05,
+      "loss": 4.2012,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3307771682739258,
+      "learning_rate": 9.981623294144782e-05,
+      "loss": 4.1475,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2902096509933472,
+      "learning_rate": 9.981424213553344e-05,
+      "loss": 4.0677,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4464447498321533,
+      "learning_rate": 9.981224062419372e-05,
+      "loss": 4.0157,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7057852745056152,
+      "learning_rate": 9.981022840785877e-05,
+      "loss": 4.1578,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2440887689590454,
+      "learning_rate": 9.980820548696106e-05,
+      "loss": 4.2148,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6559925079345703,
+      "learning_rate": 9.98061718619353e-05,
+      "loss": 4.1925,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4405591487884521,
+      "learning_rate": 9.980412753321856e-05,
+      "loss": 4.1253,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3089070320129395,
+      "learning_rate": 9.980207250125015e-05,
+      "loss": 4.0625,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3073471784591675,
+      "learning_rate": 9.980000676647174e-05,
+      "loss": 4.1465,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2623634338378906,
+      "learning_rate": 9.979793032932725e-05,
+      "loss": 4.1561,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1022056341171265,
+      "learning_rate": 9.979584319026292e-05,
+      "loss": 4.2139,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.447632908821106,
+      "learning_rate": 9.97937453497273e-05,
+      "loss": 4.0373,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.681929349899292,
+      "learning_rate": 9.979163680817124e-05,
+      "loss": 4.0781,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0082168579101562,
+      "learning_rate": 9.978951756604786e-05,
+      "loss": 4.0091,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.321807622909546,
+      "learning_rate": 9.978738762381262e-05,
+      "loss": 4.225,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3603324890136719,
+      "learning_rate": 9.978524698192322e-05,
+      "loss": 4.2713,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2551687955856323,
+      "learning_rate": 9.978309564083975e-05,
+      "loss": 4.1709,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.172796368598938,
+      "learning_rate": 9.97809336010245e-05,
+      "loss": 4.1452,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1891378164291382,
+      "learning_rate": 9.977876086294216e-05,
+      "loss": 4.0375,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3378764390945435,
+      "learning_rate": 9.977657742705962e-05,
+      "loss": 4.0972,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2819081544876099,
+      "learning_rate": 9.977438329384614e-05,
+      "loss": 4.1029,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5339614152908325,
+      "learning_rate": 9.977217846377324e-05,
+      "loss": 4.0272,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3084993362426758,
+      "learning_rate": 9.976996293731476e-05,
+      "loss": 4.0633,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1629005670547485,
+      "learning_rate": 9.976773671494684e-05,
+      "loss": 4.1783,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2501487731933594,
+      "learning_rate": 9.976549979714791e-05,
+      "loss": 4.1057,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3647338151931763,
+      "learning_rate": 9.976325218439868e-05,
+      "loss": 4.0625,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.187032699584961,
+      "learning_rate": 9.976099387718218e-05,
+      "loss": 4.0894,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2116146087646484,
+      "learning_rate": 9.975872487598377e-05,
+      "loss": 4.1672,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2495877742767334,
+      "learning_rate": 9.975644518129104e-05,
+      "loss": 4.0716,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.324497938156128,
+      "learning_rate": 9.975415479359394e-05,
+      "loss": 4.0947,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1987227201461792,
+      "learning_rate": 9.975185371338464e-05,
+      "loss": 3.9168,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.267565369606018,
+      "learning_rate": 9.97495419411577e-05,
+      "loss": 4.0161,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1384609937667847,
+      "learning_rate": 9.974721947740995e-05,
+      "loss": 4.1129,
+      "step": 1185
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1208268404006958,
+      "learning_rate": 9.974488632264046e-05,
+      "loss": 4.0352,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2156633138656616,
+      "learning_rate": 9.974254247735066e-05,
+      "loss": 4.196,
+      "step": 1195
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5510070323944092,
+      "learning_rate": 9.974018794204426e-05,
+      "loss": 3.9904,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1789484024047852,
+      "learning_rate": 9.973782271722727e-05,
+      "loss": 4.1237,
+      "step": 1205
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3516958951950073,
+      "learning_rate": 9.973544680340799e-05,
+      "loss": 4.0205,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.155287742614746,
+      "learning_rate": 9.973306020109701e-05,
+      "loss": 4.0374,
+      "step": 1215
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.330330491065979,
+      "learning_rate": 9.973066291080724e-05,
+      "loss": 3.9585,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2452833652496338,
+      "learning_rate": 9.972825493305387e-05,
+      "loss": 4.0135,
+      "step": 1225
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2431944608688354,
+      "learning_rate": 9.97258362683544e-05,
+      "loss": 4.0059,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3347609043121338,
+      "learning_rate": 9.972340691722858e-05,
+      "loss": 4.0928,
+      "step": 1235
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2513017654418945,
+      "learning_rate": 9.972096688019856e-05,
+      "loss": 3.9783,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.199873685836792,
+      "learning_rate": 9.971851615778868e-05,
+      "loss": 4.0104,
+      "step": 1245
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2518723011016846,
+      "learning_rate": 9.971605475052561e-05,
+      "loss": 4.0184,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2435201406478882,
+      "learning_rate": 9.971358265893834e-05,
+      "loss": 4.0217,
+      "step": 1255
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1471809148788452,
+      "learning_rate": 9.971109988355814e-05,
+      "loss": 3.9383,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.273890733718872,
+      "learning_rate": 9.970860642491854e-05,
+      "loss": 4.0101,
+      "step": 1265
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1873737573623657,
+      "learning_rate": 9.970610228355548e-05,
+      "loss": 4.0622,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4525707960128784,
+      "learning_rate": 9.970358746000704e-05,
+      "loss": 3.9631,
+      "step": 1275
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.557721495628357,
+      "learning_rate": 9.97010619548137e-05,
+      "loss": 4.0766,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.372824788093567,
+      "learning_rate": 9.969852576851822e-05,
+      "loss": 3.9415,
+      "step": 1285
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.37738835811615,
+      "learning_rate": 9.969597890166565e-05,
+      "loss": 4.1385,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1939241886138916,
+      "learning_rate": 9.969342135480328e-05,
+      "loss": 3.9445,
+      "step": 1295
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0946446657180786,
+      "learning_rate": 9.969085312848081e-05,
+      "loss": 4.1541,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0999705791473389,
+      "learning_rate": 9.968827422325012e-05,
+      "loss": 4.0257,
+      "step": 1305
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1700714826583862,
+      "learning_rate": 9.968568463966545e-05,
+      "loss": 3.9695,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1905884742736816,
+      "learning_rate": 9.968308437828333e-05,
+      "loss": 3.9971,
+      "step": 1315
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2550491094589233,
+      "learning_rate": 9.968047343966256e-05,
+      "loss": 3.9317,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1926301717758179,
+      "learning_rate": 9.967785182436426e-05,
+      "loss": 3.9405,
+      "step": 1325
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.047027587890625,
+      "learning_rate": 9.967521953295185e-05,
+      "loss": 3.9499,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2469595670700073,
+      "learning_rate": 9.967257656599099e-05,
+      "loss": 3.9804,
+      "step": 1335
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3009898662567139,
+      "learning_rate": 9.966992292404967e-05,
+      "loss": 4.0377,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.364970088005066,
+      "learning_rate": 9.966725860769821e-05,
+      "loss": 4.0886,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1067553758621216,
+      "learning_rate": 9.966458361750916e-05,
+      "loss": 4.0004,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.265783667564392,
+      "learning_rate": 9.966189795405743e-05,
+      "loss": 3.8869,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.215670108795166,
+      "learning_rate": 9.965920161792016e-05,
+      "loss": 4.0067,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1300758123397827,
+      "learning_rate": 9.96564946096768e-05,
+      "loss": 3.9226,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1739470958709717,
+      "learning_rate": 9.965377692990914e-05,
+      "loss": 3.9069,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2430639266967773,
+      "learning_rate": 9.965104857920121e-05,
+      "loss": 4.0824,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1683624982833862,
+      "learning_rate": 9.964830955813935e-05,
+      "loss": 3.8645,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2676290273666382,
+      "learning_rate": 9.964555986731219e-05,
+      "loss": 3.9638,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0805763006210327,
+      "learning_rate": 9.964279950731066e-05,
+      "loss": 3.8979,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2815303802490234,
+      "learning_rate": 9.9640028478728e-05,
+      "loss": 3.9317,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0681824684143066,
+      "learning_rate": 9.963724678215968e-05,
+      "loss": 3.9388,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.294988751411438,
+      "learning_rate": 9.963445441820356e-05,
+      "loss": 3.9797,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1649422645568848,
+      "learning_rate": 9.963165138745969e-05,
+      "loss": 3.9875,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.254568099975586,
+      "learning_rate": 9.962883769053051e-05,
+      "loss": 3.8479,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3542883396148682,
+      "learning_rate": 9.962601332802065e-05,
+      "loss": 3.9093,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3190867900848389,
+      "learning_rate": 9.962317830053711e-05,
+      "loss": 3.9775,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3115884065628052,
+      "learning_rate": 9.962033260868916e-05,
+      "loss": 3.986,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1862053871154785,
+      "learning_rate": 9.961747625308838e-05,
+      "loss": 3.8992,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1135691404342651,
+      "learning_rate": 9.961460923434858e-05,
+      "loss": 3.8649,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2167949676513672,
+      "learning_rate": 9.961173155308591e-05,
+      "loss": 3.9252,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2078405618667603,
+      "learning_rate": 9.960884320991881e-05,
+      "loss": 4.0237,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.168792486190796,
+      "learning_rate": 9.960594420546802e-05,
+      "loss": 3.9738,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2164114713668823,
+      "learning_rate": 9.960303454035654e-05,
+      "loss": 3.9563,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.370684027671814,
+      "learning_rate": 9.960011421520969e-05,
+      "loss": 3.9558,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3439457416534424,
+      "learning_rate": 9.959718323065503e-05,
+      "loss": 3.759,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5467573404312134,
+      "learning_rate": 9.959424158732252e-05,
+      "loss": 3.9233,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4238743782043457,
+      "learning_rate": 9.959128928584426e-05,
+      "loss": 3.7612,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1702549457550049,
+      "learning_rate": 9.958832632685478e-05,
+      "loss": 3.9554,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1538137197494507,
+      "learning_rate": 9.95853527109908e-05,
+      "loss": 3.8025,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1372348070144653,
+      "learning_rate": 9.95823684388914e-05,
+      "loss": 3.8801,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1551882028579712,
+      "learning_rate": 9.95793735111979e-05,
+      "loss": 3.9507,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2327194213867188,
+      "learning_rate": 9.957636792855394e-05,
+      "loss": 3.9354,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3141382932662964,
+      "learning_rate": 9.957335169160545e-05,
+      "loss": 4.0113,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0926406383514404,
+      "learning_rate": 9.957032480100061e-05,
+      "loss": 3.9201,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1369736194610596,
+      "learning_rate": 9.956728725738996e-05,
+      "loss": 3.9451,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0971754789352417,
+      "learning_rate": 9.956423906142626e-05,
+      "loss": 3.8602,
+      "step": 1525
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.182023048400879,
+      "learning_rate": 9.956118021376461e-05,
+      "loss": 3.8667,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1283915042877197,
+      "learning_rate": 9.955811071506236e-05,
+      "loss": 3.9306,
+      "step": 1535
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1004523038864136,
+      "learning_rate": 9.955503056597917e-05,
+      "loss": 3.8622,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0940377712249756,
+      "learning_rate": 9.955193976717698e-05,
+      "loss": 3.923,
+      "step": 1545
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.166231393814087,
+      "learning_rate": 9.954883831932003e-05,
+      "loss": 3.9338,
+      "step": 1550
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1110129356384277,
+      "learning_rate": 9.954572622307485e-05,
+      "loss": 3.8293,
+      "step": 1555
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1789952516555786,
+      "learning_rate": 9.954260347911025e-05,
+      "loss": 3.7873,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2525103092193604,
+      "learning_rate": 9.953947008809731e-05,
+      "loss": 3.8053,
+      "step": 1565
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2002336978912354,
+      "learning_rate": 9.953632605070944e-05,
+      "loss": 3.9112,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1078345775604248,
+      "learning_rate": 9.95331713676223e-05,
+      "loss": 3.757,
+      "step": 1575
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.218383550643921,
+      "learning_rate": 9.953000603951386e-05,
+      "loss": 3.978,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1606932878494263,
+      "learning_rate": 9.952683006706438e-05,
+      "loss": 3.9921,
+      "step": 1585
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2580599784851074,
+      "learning_rate": 9.952364345095638e-05,
+      "loss": 3.8183,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.031925082206726,
+      "learning_rate": 9.95204461918747e-05,
+      "loss": 3.8068,
+      "step": 1595
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.379530906677246,
+      "learning_rate": 9.951723829050646e-05,
+      "loss": 3.8227,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1440260410308838,
+      "learning_rate": 9.951401974754101e-05,
+      "loss": 3.8989,
+      "step": 1605
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2267813682556152,
+      "learning_rate": 9.951079056367013e-05,
+      "loss": 3.9197,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.145469307899475,
+      "learning_rate": 9.95075507395877e-05,
+      "loss": 3.8872,
+      "step": 1615
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1627545356750488,
+      "learning_rate": 9.950430027599003e-05,
+      "loss": 3.9378,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.152345895767212,
+      "learning_rate": 9.950103917357566e-05,
+      "loss": 3.7308,
+      "step": 1625
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.4001072645187378,
+      "learning_rate": 9.949776743304542e-05,
+      "loss": 3.8376,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3782297372817993,
+      "learning_rate": 9.949448505510243e-05,
+      "loss": 3.7022,
+      "step": 1635
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1278905868530273,
+      "learning_rate": 9.94911920404521e-05,
+      "loss": 3.8527,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1769365072250366,
+      "learning_rate": 9.94878883898021e-05,
+      "loss": 3.8674,
+      "step": 1645
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2792556285858154,
+      "learning_rate": 9.948457410386244e-05,
+      "loss": 3.8299,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2607660293579102,
+      "learning_rate": 9.948124918334538e-05,
+      "loss": 3.8099,
+      "step": 1655
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0751128196716309,
+      "learning_rate": 9.947791362896543e-05,
+      "loss": 3.8948,
+      "step": 1660
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2059290409088135,
+      "learning_rate": 9.947456744143947e-05,
+      "loss": 3.8965,
+      "step": 1665
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.159613847732544,
+      "learning_rate": 9.947121062148658e-05,
+      "loss": 3.8894,
+      "step": 1670
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1447834968566895,
+      "learning_rate": 9.94678431698282e-05,
+      "loss": 3.8217,
+      "step": 1675
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1052196025848389,
+      "learning_rate": 9.946446508718801e-05,
+      "loss": 3.7493,
+      "step": 1680
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.112877368927002,
+      "learning_rate": 9.946107637429196e-05,
+      "loss": 3.862,
+      "step": 1685
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1055601835250854,
+      "learning_rate": 9.945767703186832e-05,
+      "loss": 3.9033,
+      "step": 1690
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.248663306236267,
+      "learning_rate": 9.945426706064764e-05,
+      "loss": 3.8359,
+      "step": 1695
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2156147956848145,
+      "learning_rate": 9.945084646136275e-05,
+      "loss": 3.866,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0845564603805542,
+      "learning_rate": 9.944741523474874e-05,
+      "loss": 3.7272,
+      "step": 1705
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1792004108428955,
+      "learning_rate": 9.944397338154301e-05,
+      "loss": 3.8402,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2441108226776123,
+      "learning_rate": 9.944052090248526e-05,
+      "loss": 3.8127,
+      "step": 1715
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1368649005889893,
+      "learning_rate": 9.943705779831742e-05,
+      "loss": 3.7741,
+      "step": 1720
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0540251731872559,
+      "learning_rate": 9.943358406978377e-05,
+      "loss": 3.8522,
+      "step": 1725
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0849189758300781,
+      "learning_rate": 9.943009971763082e-05,
+      "loss": 3.8221,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1149324178695679,
+      "learning_rate": 9.942660474260737e-05,
+      "loss": 3.9317,
+      "step": 1735
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1307498216629028,
+      "learning_rate": 9.942309914546453e-05,
+      "loss": 3.7584,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1501778364181519,
+      "learning_rate": 9.941958292695565e-05,
+      "loss": 3.8572,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0906379222869873,
+      "learning_rate": 9.941605608783644e-05,
+      "loss": 3.8587,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0634100437164307,
+      "learning_rate": 9.94125186288648e-05,
+      "loss": 3.7806,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3154226541519165,
+      "learning_rate": 9.940897055080098e-05,
+      "loss": 3.7054,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2821316719055176,
+      "learning_rate": 9.940541185440746e-05,
+      "loss": 3.86,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.132352590560913,
+      "learning_rate": 9.940184254044904e-05,
+      "loss": 3.7827,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.159180760383606,
+      "learning_rate": 9.939826260969282e-05,
+      "loss": 3.7915,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1498976945877075,
+      "learning_rate": 9.93946720629081e-05,
+      "loss": 3.7884,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3698921203613281,
+      "learning_rate": 9.939107090086654e-05,
+      "loss": 3.6416,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0793654918670654,
+      "learning_rate": 9.938745912434208e-05,
+      "loss": 3.785,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.221152663230896,
+      "learning_rate": 9.938383673411087e-05,
+      "loss": 3.7182,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1927577257156372,
+      "learning_rate": 9.938020373095141e-05,
+      "loss": 3.775,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2865766286849976,
+      "learning_rate": 9.937656011564445e-05,
+      "loss": 3.7954,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1164186000823975,
+      "learning_rate": 9.937290588897302e-05,
+      "loss": 3.7728,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1258209943771362,
+      "learning_rate": 9.936924105172248e-05,
+      "loss": 3.8328,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.101777195930481,
+      "learning_rate": 9.936556560468037e-05,
+      "loss": 3.747,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0887569189071655,
+      "learning_rate": 9.93618795486366e-05,
+      "loss": 3.8125,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.108844518661499,
+      "learning_rate": 9.935818288438334e-05,
+      "loss": 3.8101,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1649794578552246,
+      "learning_rate": 9.935447561271501e-05,
+      "loss": 3.8001,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1505036354064941,
+      "learning_rate": 9.935075773442835e-05,
+      "loss": 3.7714,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1773943901062012,
+      "learning_rate": 9.934702925032232e-05,
+      "loss": 3.7502,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.055796504020691,
+      "learning_rate": 9.934329016119824e-05,
+      "loss": 3.8664,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.062416672706604,
+      "learning_rate": 9.933954046785963e-05,
+      "loss": 3.8561,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.085184931755066,
+      "learning_rate": 9.933578017111233e-05,
+      "loss": 3.7201,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1576060056686401,
+      "learning_rate": 9.933200927176449e-05,
+      "loss": 3.7706,
+      "step": 1865
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2417963743209839,
+      "learning_rate": 9.932822777062646e-05,
+      "loss": 3.6465,
+      "step": 1870
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1891114711761475,
+      "learning_rate": 9.932443566851092e-05,
+      "loss": 3.8632,
+      "step": 1875
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1396006345748901,
+      "learning_rate": 9.932063296623284e-05,
+      "loss": 3.8757,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1973085403442383,
+      "learning_rate": 9.931681966460944e-05,
+      "loss": 3.7976,
+      "step": 1885
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0854040384292603,
+      "learning_rate": 9.93129957644602e-05,
+      "loss": 3.7478,
+      "step": 1890
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3773126602172852,
+      "learning_rate": 9.930916126660691e-05,
+      "loss": 3.7924,
+      "step": 1895
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2364964485168457,
+      "learning_rate": 9.930531617187367e-05,
+      "loss": 3.6941,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.121944785118103,
+      "learning_rate": 9.930146048108678e-05,
+      "loss": 3.7555,
+      "step": 1905
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1721174716949463,
+      "learning_rate": 9.929759419507487e-05,
+      "loss": 3.7838,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0079779624938965,
+      "learning_rate": 9.92937173146688e-05,
+      "loss": 3.7461,
+      "step": 1915
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0269569158554077,
+      "learning_rate": 9.928982984070179e-05,
+      "loss": 3.8159,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0920534133911133,
+      "learning_rate": 9.928593177400925e-05,
+      "loss": 3.7634,
+      "step": 1925
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1636321544647217,
+      "learning_rate": 9.928202311542889e-05,
+      "loss": 3.7264,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0848201513290405,
+      "learning_rate": 9.927810386580075e-05,
+      "loss": 3.7209,
+      "step": 1935
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.084503412246704,
+      "learning_rate": 9.927417402596707e-05,
+      "loss": 3.6589,
+      "step": 1940
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0684373378753662,
+      "learning_rate": 9.927023359677241e-05,
+      "loss": 3.7524,
+      "step": 1945
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1085054874420166,
+      "learning_rate": 9.92662825790636e-05,
+      "loss": 3.8118,
+      "step": 1950
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1981648206710815,
+      "learning_rate": 9.926232097368974e-05,
+      "loss": 3.7354,
+      "step": 1955
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0137853622436523,
+      "learning_rate": 9.925834878150219e-05,
+      "loss": 3.7957,
+      "step": 1960
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1091909408569336,
+      "learning_rate": 9.925436600335464e-05,
+      "loss": 3.6707,
+      "step": 1965
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1084091663360596,
+      "learning_rate": 9.925037264010298e-05,
+      "loss": 3.648,
+      "step": 1970
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1454684734344482,
+      "learning_rate": 9.924636869260542e-05,
+      "loss": 3.7992,
+      "step": 1975
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.438757061958313,
+      "learning_rate": 9.924235416172243e-05,
+      "loss": 3.6862,
+      "step": 1980
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2715901136398315,
+      "learning_rate": 9.923832904831679e-05,
+      "loss": 3.6412,
+      "step": 1985
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1406395435333252,
+      "learning_rate": 9.923429335325349e-05,
+      "loss": 3.7224,
+      "step": 1990
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1130586862564087,
+      "learning_rate": 9.923024707739985e-05,
+      "loss": 3.7454,
+      "step": 1995
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1147019863128662,
+      "learning_rate": 9.922619022162545e-05,
+      "loss": 3.8271,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1101478338241577,
+      "learning_rate": 9.92221227868021e-05,
+      "loss": 3.6998,
+      "step": 2005
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0648882389068604,
+      "learning_rate": 9.921804477380394e-05,
+      "loss": 3.6909,
+      "step": 2010
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1238460540771484,
+      "learning_rate": 9.921395618350739e-05,
+      "loss": 3.7286,
+      "step": 2015
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0422731637954712,
+      "learning_rate": 9.920985701679106e-05,
+      "loss": 3.6522,
+      "step": 2020
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1619477272033691,
+      "learning_rate": 9.920574727453595e-05,
+      "loss": 3.6499,
+      "step": 2025
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1016838550567627,
+      "learning_rate": 9.920162695762522e-05,
+      "loss": 3.757,
+      "step": 2030
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.070639729499817,
+      "learning_rate": 9.919749606694439e-05,
+      "loss": 3.7667,
+      "step": 2035
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0555191040039062,
+      "learning_rate": 9.919335460338122e-05,
+      "loss": 3.5839,
+      "step": 2040
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1802692413330078,
+      "learning_rate": 9.918920256782572e-05,
+      "loss": 3.7389,
+      "step": 2045
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1555787324905396,
+      "learning_rate": 9.918503996117016e-05,
+      "loss": 3.734,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1251564025878906,
+      "learning_rate": 9.918086678430917e-05,
+      "loss": 3.6589,
+      "step": 2055
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.044123649597168,
+      "learning_rate": 9.917668303813959e-05,
+      "loss": 3.7634,
+      "step": 2060
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0908089876174927,
+      "learning_rate": 9.917248872356051e-05,
+      "loss": 3.7281,
+      "step": 2065
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1887187957763672,
+      "learning_rate": 9.916828384147331e-05,
+      "loss": 3.786,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.087288498878479,
+      "learning_rate": 9.916406839278167e-05,
+      "loss": 3.6687,
+      "step": 2075
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0338019132614136,
+      "learning_rate": 9.915984237839152e-05,
+      "loss": 3.8061,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0947060585021973,
+      "learning_rate": 9.915560579921106e-05,
+      "loss": 3.7306,
+      "step": 2085
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1791576147079468,
+      "learning_rate": 9.915135865615074e-05,
+      "loss": 3.7199,
+      "step": 2090
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2083804607391357,
+      "learning_rate": 9.914710095012332e-05,
+      "loss": 3.4363,
+      "step": 2095
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.154111385345459,
+      "learning_rate": 9.91428326820438e-05,
+      "loss": 3.6123,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2971912622451782,
+      "learning_rate": 9.913855385282947e-05,
+      "loss": 3.6416,
+      "step": 2105
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0566489696502686,
+      "learning_rate": 9.913426446339986e-05,
+      "loss": 3.8028,
+      "step": 2110
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2255592346191406,
+      "learning_rate": 9.912996451467682e-05,
+      "loss": 3.7183,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1278361082077026,
+      "learning_rate": 9.912565400758443e-05,
+      "loss": 3.734,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2824383974075317,
+      "learning_rate": 9.912133294304901e-05,
+      "loss": 3.7336,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0863618850708008,
+      "learning_rate": 9.911700132199924e-05,
+      "loss": 3.7979,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.079748272895813,
+      "learning_rate": 9.911265914536598e-05,
+      "loss": 3.5865,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.078432321548462,
+      "learning_rate": 9.910830641408242e-05,
+      "loss": 3.6965,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.154131293296814,
+      "learning_rate": 9.910394312908397e-05,
+      "loss": 3.6382,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0361031293869019,
+      "learning_rate": 9.909956929130834e-05,
+      "loss": 3.6642,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0190991163253784,
+      "learning_rate": 9.90951849016955e-05,
+      "loss": 3.6078,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0850460529327393,
+      "learning_rate": 9.909078996118768e-05,
+      "loss": 3.6513,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0098525285720825,
+      "learning_rate": 9.908638447072939e-05,
+      "loss": 3.772,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0971590280532837,
+      "learning_rate": 9.90819684312674e-05,
+      "loss": 3.7151,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1846588850021362,
+      "learning_rate": 9.907754184375074e-05,
+      "loss": 3.6172,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.571771264076233,
+      "learning_rate": 9.90731047091307e-05,
+      "loss": 3.7266,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2662121057510376,
+      "learning_rate": 9.906865702836088e-05,
+      "loss": 3.6457,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0804154872894287,
+      "learning_rate": 9.90641988023971e-05,
+      "loss": 3.7323,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1559226512908936,
+      "learning_rate": 9.905973003219748e-05,
+      "loss": 3.6846,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.232035756111145,
+      "learning_rate": 9.905525071872237e-05,
+      "loss": 3.7547,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0110341310501099,
+      "learning_rate": 9.905076086293443e-05,
+      "loss": 3.7347,
+      "step": 2205
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.128730297088623,
+      "learning_rate": 9.904626046579852e-05,
+      "loss": 3.5852,
+      "step": 2210
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1578927040100098,
+      "learning_rate": 9.904174952828186e-05,
+      "loss": 3.6928,
+      "step": 2215
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0288796424865723,
+      "learning_rate": 9.903722805135385e-05,
+      "loss": 3.6973,
+      "step": 2220
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5191551446914673,
+      "learning_rate": 9.903269603598617e-05,
+      "loss": 3.6236,
+      "step": 2225
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.275098204612732,
+      "learning_rate": 9.902815348315282e-05,
+      "loss": 3.647,
+      "step": 2230
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0530247688293457,
+      "learning_rate": 9.902360039383e-05,
+      "loss": 3.6238,
+      "step": 2235
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1691607236862183,
+      "learning_rate": 9.901903676899621e-05,
+      "loss": 3.6836,
+      "step": 2240
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0681238174438477,
+      "learning_rate": 9.901446260963221e-05,
+      "loss": 3.4766,
+      "step": 2245
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1328433752059937,
+      "learning_rate": 9.9009877916721e-05,
+      "loss": 3.7453,
+      "step": 2250
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1731886863708496,
+      "learning_rate": 9.900528269124787e-05,
+      "loss": 3.5549,
+      "step": 2255
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.261199712753296,
+      "learning_rate": 9.900067693420039e-05,
+      "loss": 3.634,
+      "step": 2260
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2321131229400635,
+      "learning_rate": 9.899606064656832e-05,
+      "loss": 3.694,
+      "step": 2265
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1463013887405396,
+      "learning_rate": 9.899143382934379e-05,
+      "loss": 3.6321,
+      "step": 2270
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0411920547485352,
+      "learning_rate": 9.898679648352108e-05,
+      "loss": 3.571,
+      "step": 2275
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1323692798614502,
+      "learning_rate": 9.898214861009683e-05,
+      "loss": 3.5781,
+      "step": 2280
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9899020791053772,
+      "learning_rate": 9.897749021006988e-05,
+      "loss": 3.6665,
+      "step": 2285
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1613025665283203,
+      "learning_rate": 9.897282128444135e-05,
+      "loss": 3.4767,
+      "step": 2290
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1543056964874268,
+      "learning_rate": 9.896814183421464e-05,
+      "loss": 3.5864,
+      "step": 2295
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0871474742889404,
+      "learning_rate": 9.896345186039538e-05,
+      "loss": 3.6191,
+      "step": 2300
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0214766263961792,
+      "learning_rate": 9.895875136399149e-05,
+      "loss": 3.532,
+      "step": 2305
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2820186614990234,
+      "learning_rate": 9.895404034601313e-05,
+      "loss": 3.715,
+      "step": 2310
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.305051565170288,
+      "learning_rate": 9.894931880747273e-05,
+      "loss": 3.6176,
+      "step": 2315
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.152801275253296,
+      "learning_rate": 9.894458674938499e-05,
+      "loss": 3.6984,
+      "step": 2320
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0803343057632446,
+      "learning_rate": 9.893984417276686e-05,
+      "loss": 3.7141,
+      "step": 2325
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0708693265914917,
+      "learning_rate": 9.893509107863754e-05,
+      "loss": 3.4484,
+      "step": 2330
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1257356405258179,
+      "learning_rate": 9.893032746801854e-05,
+      "loss": 3.5978,
+      "step": 2335
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1155155897140503,
+      "learning_rate": 9.892555334193353e-05,
+      "loss": 3.6202,
+      "step": 2340
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.38144850730896,
+      "learning_rate": 9.892076870140857e-05,
+      "loss": 3.6877,
+      "step": 2345
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.070850133895874,
+      "learning_rate": 9.891597354747186e-05,
+      "loss": 3.647,
+      "step": 2350
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.188733696937561,
+      "learning_rate": 9.891116788115393e-05,
+      "loss": 3.6138,
+      "step": 2355
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1980376243591309,
+      "learning_rate": 9.890635170348756e-05,
+      "loss": 3.5316,
+      "step": 2360
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4594496488571167,
+      "learning_rate": 9.890152501550777e-05,
+      "loss": 3.5698,
+      "step": 2365
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2152003049850464,
+      "learning_rate": 9.889668781825185e-05,
+      "loss": 3.5372,
+      "step": 2370
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.08065664768219,
+      "learning_rate": 9.889184011275936e-05,
+      "loss": 3.6214,
+      "step": 2375
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1151206493377686,
+      "learning_rate": 9.888698190007209e-05,
+      "loss": 3.4842,
+      "step": 2380
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0255228281021118,
+      "learning_rate": 9.88821131812341e-05,
+      "loss": 3.5903,
+      "step": 2385
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.326302170753479,
+      "learning_rate": 9.887723395729174e-05,
+      "loss": 3.6795,
+      "step": 2390
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2403773069381714,
+      "learning_rate": 9.887234422929355e-05,
+      "loss": 3.7132,
+      "step": 2395
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.34036386013031,
+      "learning_rate": 9.886744399829039e-05,
+      "loss": 3.5179,
+      "step": 2400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0320769548416138,
+      "learning_rate": 9.886253326533535e-05,
+      "loss": 3.5983,
+      "step": 2405
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2648769617080688,
+      "learning_rate": 9.885761203148378e-05,
+      "loss": 3.624,
+      "step": 2410
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0475295782089233,
+      "learning_rate": 9.885268029779327e-05,
+      "loss": 3.5959,
+      "step": 2415
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1629871129989624,
+      "learning_rate": 9.884773806532373e-05,
+      "loss": 3.5531,
+      "step": 2420
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0903927087783813,
+      "learning_rate": 9.884278533513723e-05,
+      "loss": 3.6235,
+      "step": 2425
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2667114734649658,
+      "learning_rate": 9.883782210829816e-05,
+      "loss": 3.6523,
+      "step": 2430
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0616856813430786,
+      "learning_rate": 9.883284838587317e-05,
+      "loss": 3.6279,
+      "step": 2435
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9978397488594055,
+      "learning_rate": 9.882786416893112e-05,
+      "loss": 3.5887,
+      "step": 2440
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.751414179801941,
+      "learning_rate": 9.882286945854319e-05,
+      "loss": 3.5608,
+      "step": 2445
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1269872188568115,
+      "learning_rate": 9.881786425578274e-05,
+      "loss": 3.6644,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0960004329681396,
+      "learning_rate": 9.881284856172545e-05,
+      "loss": 3.5293,
+      "step": 2455
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1556843519210815,
+      "learning_rate": 9.880782237744921e-05,
+      "loss": 3.5626,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.169901967048645,
+      "learning_rate": 9.880278570403419e-05,
+      "loss": 3.5271,
+      "step": 2465
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1093649864196777,
+      "learning_rate": 9.87977385425628e-05,
+      "loss": 3.6219,
+      "step": 2470
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.014595627784729,
+      "learning_rate": 9.87926808941197e-05,
+      "loss": 3.4912,
+      "step": 2475
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9790247082710266,
+      "learning_rate": 9.878761275979184e-05,
+      "loss": 3.5369,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0990004539489746,
+      "learning_rate": 9.878253414066839e-05,
+      "loss": 3.5177,
+      "step": 2485
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0641359090805054,
+      "learning_rate": 9.877744503784076e-05,
+      "loss": 3.6073,
+      "step": 2490
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1036216020584106,
+      "learning_rate": 9.877234545240265e-05,
+      "loss": 3.5044,
+      "step": 2495
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1042238473892212,
+      "learning_rate": 9.876723538545001e-05,
+      "loss": 3.6595,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1421784162521362,
+      "learning_rate": 9.8762114838081e-05,
+      "loss": 3.5531,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2461539506912231,
+      "learning_rate": 9.875698381139607e-05,
+      "loss": 3.5563,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.342589259147644,
+      "learning_rate": 9.875184230649792e-05,
+      "loss": 3.4942,
+      "step": 2515
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.050052523612976,
+      "learning_rate": 9.874669032449149e-05,
+      "loss": 3.4938,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9648129940032959,
+      "learning_rate": 9.874152786648399e-05,
+      "loss": 3.5123,
+      "step": 2525
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0168482065200806,
+      "learning_rate": 9.873635493358484e-05,
+      "loss": 3.6236,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1292569637298584,
+      "learning_rate": 9.873117152690576e-05,
+      "loss": 3.5438,
+      "step": 2535
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1720370054244995,
+      "learning_rate": 9.872597764756071e-05,
+      "loss": 3.6319,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0277713537216187,
+      "learning_rate": 9.872077329666587e-05,
+      "loss": 3.5847,
+      "step": 2545
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0888758897781372,
+      "learning_rate": 9.871555847533968e-05,
+      "loss": 3.5598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1157809495925903,
+      "learning_rate": 9.871033318470289e-05,
+      "loss": 3.6757,
+      "step": 2555
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.015991449356079,
+      "learning_rate": 9.870509742587838e-05,
+      "loss": 3.5076,
+      "step": 2560
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0340766906738281,
+      "learning_rate": 9.869985119999143e-05,
+      "loss": 3.6289,
+      "step": 2565
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0396736860275269,
+      "learning_rate": 9.869459450816942e-05,
+      "loss": 3.559,
+      "step": 2570
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2371712923049927,
+      "learning_rate": 9.86893273515421e-05,
+      "loss": 3.6202,
+      "step": 2575
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.158126950263977,
+      "learning_rate": 9.868404973124139e-05,
+      "loss": 3.6196,
+      "step": 2580
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9961780309677124,
+      "learning_rate": 9.86787616484015e-05,
+      "loss": 3.6127,
+      "step": 2585
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.125521183013916,
+      "learning_rate": 9.867346310415886e-05,
+      "loss": 3.6531,
+      "step": 2590
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1232786178588867,
+      "learning_rate": 9.866815409965219e-05,
+      "loss": 3.5345,
+      "step": 2595
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1041135787963867,
+      "learning_rate": 9.86628346360224e-05,
+      "loss": 3.5586,
+      "step": 2600
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9918065071105957,
+      "learning_rate": 9.865750471441271e-05,
+      "loss": 3.5837,
+      "step": 2605
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.051330327987671,
+      "learning_rate": 9.865216433596853e-05,
+      "loss": 3.5309,
+      "step": 2610
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0439777374267578,
+      "learning_rate": 9.864681350183758e-05,
+      "loss": 3.5586,
+      "step": 2615
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1238760948181152,
+      "learning_rate": 9.864145221316975e-05,
+      "loss": 3.6062,
+      "step": 2620
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0839402675628662,
+      "learning_rate": 9.863608047111725e-05,
+      "loss": 3.5278,
+      "step": 2625
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0253914594650269,
+      "learning_rate": 9.863069827683446e-05,
+      "loss": 3.6927,
+      "step": 2630
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2269690036773682,
+      "learning_rate": 9.862530563147812e-05,
+      "loss": 3.5033,
+      "step": 2635
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0736929178237915,
+      "learning_rate": 9.861990253620709e-05,
+      "loss": 3.5033,
+      "step": 2640
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0313290357589722,
+      "learning_rate": 9.861448899218255e-05,
+      "loss": 3.5388,
+      "step": 2645
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1959092617034912,
+      "learning_rate": 9.860906500056792e-05,
+      "loss": 3.6057,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0732204914093018,
+      "learning_rate": 9.860363056252883e-05,
+      "loss": 3.5842,
+      "step": 2655
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1429522037506104,
+      "learning_rate": 9.859818567923321e-05,
+      "loss": 3.5485,
+      "step": 2660
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.094766616821289,
+      "learning_rate": 9.859273035185118e-05,
+      "loss": 3.6418,
+      "step": 2665
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0736944675445557,
+      "learning_rate": 9.858726458155512e-05,
+      "loss": 3.5989,
+      "step": 2670
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0924025774002075,
+      "learning_rate": 9.858178836951967e-05,
+      "loss": 3.6088,
+      "step": 2675
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.981315553188324,
+      "learning_rate": 9.857630171692174e-05,
+      "loss": 3.5713,
+      "step": 2680
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9852755665779114,
+      "learning_rate": 9.857080462494039e-05,
+      "loss": 3.4977,
+      "step": 2685
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0812565088272095,
+      "learning_rate": 9.856529709475703e-05,
+      "loss": 3.594,
+      "step": 2690
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0880519151687622,
+      "learning_rate": 9.855977912755524e-05,
+      "loss": 3.5914,
+      "step": 2695
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9606928825378418,
+      "learning_rate": 9.855425072452086e-05,
+      "loss": 3.5997,
+      "step": 2700
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0760209560394287,
+      "learning_rate": 9.854871188684204e-05,
+      "loss": 3.5271,
+      "step": 2705
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9680733680725098,
+      "learning_rate": 9.854316261570905e-05,
+      "loss": 3.533,
+      "step": 2710
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.285902738571167,
+      "learning_rate": 9.85376029123145e-05,
+      "loss": 3.5647,
+      "step": 2715
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1772093772888184,
+      "learning_rate": 9.853203277785318e-05,
+      "loss": 3.519,
+      "step": 2720
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.997449517250061,
+      "learning_rate": 9.85264522135222e-05,
+      "loss": 3.4344,
+      "step": 2725
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.030928373336792,
+      "learning_rate": 9.852086122052083e-05,
+      "loss": 3.4852,
+      "step": 2730
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.07460355758667,
+      "learning_rate": 9.85152598000506e-05,
+      "loss": 3.5319,
+      "step": 2735
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0324660539627075,
+      "learning_rate": 9.850964795331531e-05,
+      "loss": 3.6045,
+      "step": 2740
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3597207069396973,
+      "learning_rate": 9.8504025681521e-05,
+      "loss": 3.6306,
+      "step": 2745
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0797796249389648,
+      "learning_rate": 9.84983929858759e-05,
+      "loss": 3.5318,
+      "step": 2750
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1963589191436768,
+      "learning_rate": 9.849274986759054e-05,
+      "loss": 3.4703,
+      "step": 2755
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.268329381942749,
+      "learning_rate": 9.848709632787768e-05,
+      "loss": 3.5278,
+      "step": 2760
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2875295877456665,
+      "learning_rate": 9.848143236795225e-05,
+      "loss": 3.5672,
+      "step": 2765
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.847575798903152e-05,
+      "loss": 3.5612,
+      "step": 2770
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.110255241394043,
+      "learning_rate": 9.847007319233496e-05,
+      "loss": 3.5305,
+      "step": 2775
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.103633165359497,
+      "learning_rate": 9.846437797908424e-05,
+      "loss": 3.5624,
+      "step": 2780
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9643424153327942,
+      "learning_rate": 9.845867235050333e-05,
+      "loss": 3.4022,
+      "step": 2785
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0237890481948853,
+      "learning_rate": 9.845295630781837e-05,
+      "loss": 3.5468,
+      "step": 2790
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0827010869979858,
+      "learning_rate": 9.844722985225783e-05,
+      "loss": 3.5086,
+      "step": 2795
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.05714750289917,
+      "learning_rate": 9.844149298505232e-05,
+      "loss": 3.5865,
+      "step": 2800
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.048947811126709,
+      "learning_rate": 9.843574570743474e-05,
+      "loss": 3.4404,
+      "step": 2805
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0050597190856934,
+      "learning_rate": 9.842998802064025e-05,
+      "loss": 3.4457,
+      "step": 2810
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1087723970413208,
+      "learning_rate": 9.842421992590619e-05,
+      "loss": 3.5265,
+      "step": 2815
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0290030241012573,
+      "learning_rate": 9.841844142447215e-05,
+      "loss": 3.5128,
+      "step": 2820
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0268406867980957,
+      "learning_rate": 9.841265251758e-05,
+      "loss": 3.4185,
+      "step": 2825
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1047559976577759,
+      "learning_rate": 9.840685320647381e-05,
+      "loss": 3.4578,
+      "step": 2830
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0163120031356812,
+      "learning_rate": 9.840104349239988e-05,
+      "loss": 3.4337,
+      "step": 2835
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1573519706726074,
+      "learning_rate": 9.839522337660677e-05,
+      "loss": 3.5795,
+      "step": 2840
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0155491828918457,
+      "learning_rate": 9.838939286034526e-05,
+      "loss": 3.4738,
+      "step": 2845
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9622009992599487,
+      "learning_rate": 9.838355194486836e-05,
+      "loss": 3.4871,
+      "step": 2850
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0036994218826294,
+      "learning_rate": 9.837770063143133e-05,
+      "loss": 3.5053,
+      "step": 2855
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9812444448471069,
+      "learning_rate": 9.837183892129164e-05,
+      "loss": 3.5922,
+      "step": 2860
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.199960470199585,
+      "learning_rate": 9.836596681570906e-05,
+      "loss": 3.4763,
+      "step": 2865
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0549086332321167,
+      "learning_rate": 9.83600843159455e-05,
+      "loss": 3.556,
+      "step": 2870
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.001253604888916,
+      "learning_rate": 9.835419142326517e-05,
+      "loss": 3.4683,
+      "step": 2875
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.001755714416504,
+      "learning_rate": 9.834828813893448e-05,
+      "loss": 3.5815,
+      "step": 2880
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2254658937454224,
+      "learning_rate": 9.83423744642221e-05,
+      "loss": 3.4933,
+      "step": 2885
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0125194787979126,
+      "learning_rate": 9.833645040039892e-05,
+      "loss": 3.4891,
+      "step": 2890
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9885510206222534,
+      "learning_rate": 9.833051594873805e-05,
+      "loss": 3.6201,
+      "step": 2895
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.100462794303894,
+      "learning_rate": 9.832457111051485e-05,
+      "loss": 3.6135,
+      "step": 2900
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9827430248260498,
+      "learning_rate": 9.831861588700692e-05,
+      "loss": 3.5163,
+      "step": 2905
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0402806997299194,
+      "learning_rate": 9.831265027949406e-05,
+      "loss": 3.5208,
+      "step": 2910
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.955693244934082,
+      "learning_rate": 9.830667428925833e-05,
+      "loss": 3.5386,
+      "step": 2915
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0769044160842896,
+      "learning_rate": 9.8300687917584e-05,
+      "loss": 3.4891,
+      "step": 2920
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0844204425811768,
+      "learning_rate": 9.82946911657576e-05,
+      "loss": 3.5198,
+      "step": 2925
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9972931146621704,
+      "learning_rate": 9.828868403506788e-05,
+      "loss": 3.4106,
+      "step": 2930
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9864761829376221,
+      "learning_rate": 9.828266652680577e-05,
+      "loss": 3.4301,
+      "step": 2935
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0817521810531616,
+      "learning_rate": 9.827663864226453e-05,
+      "loss": 3.5186,
+      "step": 2940
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0134162902832031,
+      "learning_rate": 9.827060038273956e-05,
+      "loss": 3.4016,
+      "step": 2945
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9517087936401367,
+      "learning_rate": 9.826455174952852e-05,
+      "loss": 3.5089,
+      "step": 2950
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1131198406219482,
+      "learning_rate": 9.825849274393133e-05,
+      "loss": 3.4629,
+      "step": 2955
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0658578872680664,
+      "learning_rate": 9.825242336725009e-05,
+      "loss": 3.4165,
+      "step": 2960
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0165268182754517,
+      "learning_rate": 9.824634362078915e-05,
+      "loss": 3.4778,
+      "step": 2965
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.119536280632019,
+      "learning_rate": 9.824025350585511e-05,
+      "loss": 3.452,
+      "step": 2970
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0860726833343506,
+      "learning_rate": 9.823415302375676e-05,
+      "loss": 3.4869,
+      "step": 2975
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0865418910980225,
+      "learning_rate": 9.822804217580515e-05,
+      "loss": 3.5318,
+      "step": 2980
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.017012357711792,
+      "learning_rate": 9.822192096331351e-05,
+      "loss": 3.5245,
+      "step": 2985
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9632712602615356,
+      "learning_rate": 9.82157893875974e-05,
+      "loss": 3.5834,
+      "step": 2990
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0306297540664673,
+      "learning_rate": 9.820964744997445e-05,
+      "loss": 3.4255,
+      "step": 2995
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0064184665679932,
+      "learning_rate": 9.820349515176466e-05,
+      "loss": 3.4985,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9767744541168213,
+      "learning_rate": 9.819733249429019e-05,
+      "loss": 3.5188,
+      "step": 3005
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7096657752990723,
+      "learning_rate": 9.819115947887545e-05,
+      "loss": 3.4306,
+      "step": 3010
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0770705938339233,
+      "learning_rate": 9.818497610684705e-05,
+      "loss": 3.3749,
+      "step": 3015
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9567260146141052,
+      "learning_rate": 9.817878237953383e-05,
+      "loss": 3.4588,
+      "step": 3020
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9681187272071838,
+      "learning_rate": 9.817257829826689e-05,
+      "loss": 3.4639,
+      "step": 3025
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0774757862091064,
+      "learning_rate": 9.81663638643795e-05,
+      "loss": 3.4055,
+      "step": 3030
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.970565676689148,
+      "learning_rate": 9.81601390792072e-05,
+      "loss": 3.6158,
+      "step": 3035
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0367475748062134,
+      "learning_rate": 9.815390394408776e-05,
+      "loss": 3.4249,
+      "step": 3040
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0215468406677246,
+      "learning_rate": 9.814765846036111e-05,
+      "loss": 3.4899,
+      "step": 3045
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0185517072677612,
+      "learning_rate": 9.81414026293695e-05,
+      "loss": 3.5214,
+      "step": 3050
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9981986284255981,
+      "learning_rate": 9.813513645245729e-05,
+      "loss": 3.5339,
+      "step": 3055
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9764263033866882,
+      "learning_rate": 9.812885993097118e-05,
+      "loss": 3.4869,
+      "step": 3060
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9552075862884521,
+      "learning_rate": 9.812257306626002e-05,
+      "loss": 3.444,
+      "step": 3065
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0467313528060913,
+      "learning_rate": 9.811627585967487e-05,
+      "loss": 3.4513,
+      "step": 3070
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0830802917480469,
+      "learning_rate": 9.810996831256909e-05,
+      "loss": 3.4145,
+      "step": 3075
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.22933030128479,
+      "learning_rate": 9.81036504262982e-05,
+      "loss": 3.4897,
+      "step": 3080
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1829131841659546,
+      "learning_rate": 9.809732220221995e-05,
+      "loss": 3.348,
+      "step": 3085
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0107975006103516,
+      "learning_rate": 9.80909836416943e-05,
+      "loss": 3.5119,
+      "step": 3090
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.043864369392395,
+      "learning_rate": 9.80846347460835e-05,
+      "loss": 3.5445,
+      "step": 3095
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9549990892410278,
+      "learning_rate": 9.807827551675192e-05,
+      "loss": 3.4151,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0937469005584717,
+      "learning_rate": 9.807190595506623e-05,
+      "loss": 3.3941,
+      "step": 3105
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0855787992477417,
+      "learning_rate": 9.80655260623953e-05,
+      "loss": 3.5123,
+      "step": 3110
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1411908864974976,
+      "learning_rate": 9.805913584011019e-05,
+      "loss": 3.4843,
+      "step": 3115
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.083005666732788,
+      "learning_rate": 9.805273528958423e-05,
+      "loss": 3.408,
+      "step": 3120
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9654300808906555,
+      "learning_rate": 9.804632441219293e-05,
+      "loss": 3.5365,
+      "step": 3125
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1123825311660767,
+      "learning_rate": 9.803990320931402e-05,
+      "loss": 3.467,
+      "step": 3130
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0562355518341064,
+      "learning_rate": 9.803347168232746e-05,
+      "loss": 3.4257,
+      "step": 3135
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0266400575637817,
+      "learning_rate": 9.802702983261545e-05,
+      "loss": 3.4928,
+      "step": 3140
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9600659608840942,
+      "learning_rate": 9.802057766156238e-05,
+      "loss": 3.4841,
+      "step": 3145
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.019291639328003,
+      "learning_rate": 9.801411517055488e-05,
+      "loss": 3.4745,
+      "step": 3150
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9733086824417114,
+      "learning_rate": 9.800764236098175e-05,
+      "loss": 3.4859,
+      "step": 3155
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0090728998184204,
+      "learning_rate": 9.800115923423405e-05,
+      "loss": 3.4667,
+      "step": 3160
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.013838529586792,
+      "learning_rate": 9.799466579170508e-05,
+      "loss": 3.4559,
+      "step": 3165
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1018142700195312,
+      "learning_rate": 9.798816203479029e-05,
+      "loss": 3.4378,
+      "step": 3170
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9640586972236633,
+      "learning_rate": 9.79816479648874e-05,
+      "loss": 3.4211,
+      "step": 3175
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.195827603340149,
+      "learning_rate": 9.797512358339633e-05,
+      "loss": 3.4077,
+      "step": 3180
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0667580366134644,
+      "learning_rate": 9.796858889171921e-05,
+      "loss": 3.473,
+      "step": 3185
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2441514730453491,
+      "learning_rate": 9.79620438912604e-05,
+      "loss": 3.4334,
+      "step": 3190
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9612923264503479,
+      "learning_rate": 9.795548858342645e-05,
+      "loss": 3.4773,
+      "step": 3195
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.137223482131958,
+      "learning_rate": 9.794892296962615e-05,
+      "loss": 3.4067,
+      "step": 3200
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.36946439743042,
+      "learning_rate": 9.794234705127047e-05,
+      "loss": 3.5078,
+      "step": 3205
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0428686141967773,
+      "learning_rate": 9.793576082977267e-05,
+      "loss": 3.4682,
+      "step": 3210
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1735200881958008,
+      "learning_rate": 9.792916430654815e-05,
+      "loss": 3.398,
+      "step": 3215
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0603055953979492,
+      "learning_rate": 9.792255748301453e-05,
+      "loss": 3.4711,
+      "step": 3220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9589853882789612,
+      "learning_rate": 9.791594036059168e-05,
+      "loss": 3.543,
+      "step": 3225
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9458810091018677,
+      "learning_rate": 9.790931294070169e-05,
+      "loss": 3.3583,
+      "step": 3230
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0348060131072998,
+      "learning_rate": 9.79026752247688e-05,
+      "loss": 3.512,
+      "step": 3235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0105575323104858,
+      "learning_rate": 9.78960272142195e-05,
+      "loss": 3.333,
+      "step": 3240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.192763328552246,
+      "learning_rate": 9.788936891048251e-05,
+      "loss": 3.3534,
+      "step": 3245
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9881122708320618,
+      "learning_rate": 9.788270031498876e-05,
+      "loss": 3.4441,
+      "step": 3250
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9550538659095764,
+      "learning_rate": 9.787602142917135e-05,
+      "loss": 3.3497,
+      "step": 3255
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9749652743339539,
+      "learning_rate": 9.786933225446563e-05,
+      "loss": 3.4057,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0314573049545288,
+      "learning_rate": 9.786263279230915e-05,
+      "loss": 3.535,
+      "step": 3265
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0040273666381836,
+      "learning_rate": 9.785592304414167e-05,
+      "loss": 3.5443,
+      "step": 3270
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1736184358596802,
+      "learning_rate": 9.784920301140516e-05,
+      "loss": 3.3826,
+      "step": 3275
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0267746448516846,
+      "learning_rate": 9.78424726955438e-05,
+      "loss": 3.3367,
+      "step": 3280
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9793338179588318,
+      "learning_rate": 9.783573209800398e-05,
+      "loss": 3.4609,
+      "step": 3285
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9645666480064392,
+      "learning_rate": 9.782898122023432e-05,
+      "loss": 3.3982,
+      "step": 3290
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1025176048278809,
+      "learning_rate": 9.78222200636856e-05,
+      "loss": 3.458,
+      "step": 3295
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0191633701324463,
+      "learning_rate": 9.781544862981087e-05,
+      "loss": 3.4591,
+      "step": 3300
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9507684111595154,
+      "learning_rate": 9.780866692006534e-05,
+      "loss": 3.4371,
+      "step": 3305
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.22274649143219,
+      "learning_rate": 9.780187493590645e-05,
+      "loss": 3.4014,
+      "step": 3310
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.047214150428772,
+      "learning_rate": 9.779507267879385e-05,
+      "loss": 3.4863,
+      "step": 3315
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0347914695739746,
+      "learning_rate": 9.778826015018936e-05,
+      "loss": 3.4497,
+      "step": 3320
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.005861759185791,
+      "learning_rate": 9.778143735155709e-05,
+      "loss": 3.4249,
+      "step": 3325
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9964653849601746,
+      "learning_rate": 9.777460428436327e-05,
+      "loss": 3.3549,
+      "step": 3330
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1029973030090332,
+      "learning_rate": 9.776776095007638e-05,
+      "loss": 3.3695,
+      "step": 3335
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9763995409011841,
+      "learning_rate": 9.776090735016711e-05,
+      "loss": 3.3994,
+      "step": 3340
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9883373975753784,
+      "learning_rate": 9.775404348610835e-05,
+      "loss": 3.4716,
+      "step": 3345
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9866198897361755,
+      "learning_rate": 9.774716935937518e-05,
+      "loss": 3.4467,
+      "step": 3350
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9254525303840637,
+      "learning_rate": 9.77402849714449e-05,
+      "loss": 3.4269,
+      "step": 3355
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0661156177520752,
+      "learning_rate": 9.773339032379701e-05,
+      "loss": 3.4768,
+      "step": 3360
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.016548991203308,
+      "learning_rate": 9.772648541791323e-05,
+      "loss": 3.4698,
+      "step": 3365
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.077796459197998,
+      "learning_rate": 9.771957025527745e-05,
+      "loss": 3.4289,
+      "step": 3370
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9630600214004517,
+      "learning_rate": 9.771264483737581e-05,
+      "loss": 3.4101,
+      "step": 3375
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9673587083816528,
+      "learning_rate": 9.770570916569661e-05,
+      "loss": 3.447,
+      "step": 3380
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0290669202804565,
+      "learning_rate": 9.76987632417304e-05,
+      "loss": 3.4171,
+      "step": 3385
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1489824056625366,
+      "learning_rate": 9.769180706696988e-05,
+      "loss": 3.4212,
+      "step": 3390
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0669625997543335,
+      "learning_rate": 9.768484064290997e-05,
+      "loss": 3.5411,
+      "step": 3395
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0255268812179565,
+      "learning_rate": 9.767786397104786e-05,
+      "loss": 3.3542,
+      "step": 3400
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0306657552719116,
+      "learning_rate": 9.767087705288283e-05,
+      "loss": 3.2761,
+      "step": 3405
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9816969037055969,
+      "learning_rate": 9.766387988991642e-05,
+      "loss": 3.4849,
+      "step": 3410
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9604994654655457,
+      "learning_rate": 9.765687248365241e-05,
+      "loss": 3.386,
+      "step": 3415
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.231518030166626,
+      "learning_rate": 9.76498548355967e-05,
+      "loss": 3.5045,
+      "step": 3420
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9708835482597351,
+      "learning_rate": 9.764282694725744e-05,
+      "loss": 3.3812,
+      "step": 3425
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0475298166275024,
+      "learning_rate": 9.7635788820145e-05,
+      "loss": 3.2903,
+      "step": 3430
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1823440790176392,
+      "learning_rate": 9.762874045577189e-05,
+      "loss": 3.3421,
+      "step": 3435
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0619900226593018,
+      "learning_rate": 9.762168185565284e-05,
+      "loss": 3.3866,
+      "step": 3440
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9650983810424805,
+      "learning_rate": 9.761461302130484e-05,
+      "loss": 3.5177,
+      "step": 3445
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9646357893943787,
+      "learning_rate": 9.760753395424699e-05,
+      "loss": 3.4513,
+      "step": 3450
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0089373588562012,
+      "learning_rate": 9.760044465600065e-05,
+      "loss": 3.4115,
+      "step": 3455
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0304383039474487,
+      "learning_rate": 9.759334512808935e-05,
+      "loss": 3.4592,
+      "step": 3460
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.6191725730895996,
+      "learning_rate": 9.758623537203883e-05,
+      "loss": 3.4785,
+      "step": 3465
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.154813289642334,
+      "learning_rate": 9.757911538937706e-05,
+      "loss": 3.3797,
+      "step": 3470
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.997660756111145,
+      "learning_rate": 9.757198518163411e-05,
+      "loss": 3.3218,
+      "step": 3475
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0254180431365967,
+      "learning_rate": 9.756484475034235e-05,
+      "loss": 3.4015,
+      "step": 3480
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.972675621509552,
+      "learning_rate": 9.75576940970363e-05,
+      "loss": 3.4615,
+      "step": 3485
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9724268317222595,
+      "learning_rate": 9.755053322325269e-05,
+      "loss": 3.3979,
+      "step": 3490
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0327606201171875,
+      "learning_rate": 9.754336213053043e-05,
+      "loss": 3.3581,
+      "step": 3495
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0239059925079346,
+      "learning_rate": 9.753618082041066e-05,
+      "loss": 3.3654,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.05912184715271,
+      "learning_rate": 9.752898929443666e-05,
+      "loss": 3.3737,
+      "step": 3505
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9625441431999207,
+      "learning_rate": 9.752178755415396e-05,
+      "loss": 3.4784,
+      "step": 3510
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0818109512329102,
+      "learning_rate": 9.751457560111026e-05,
+      "loss": 3.3692,
+      "step": 3515
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0533093214035034,
+      "learning_rate": 9.750735343685547e-05,
+      "loss": 3.4183,
+      "step": 3520
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9250955581665039,
+      "learning_rate": 9.750012106294167e-05,
+      "loss": 3.3006,
+      "step": 3525
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8984771966934204,
+      "learning_rate": 9.749287848092315e-05,
+      "loss": 3.4206,
+      "step": 3530
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9278005361557007,
+      "learning_rate": 9.748562569235637e-05,
+      "loss": 3.5074,
+      "step": 3535
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9562070965766907,
+      "learning_rate": 9.747836269880003e-05,
+      "loss": 3.3382,
+      "step": 3540
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1674562692642212,
+      "learning_rate": 9.7471089501815e-05,
+      "loss": 3.4096,
+      "step": 3545
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4003989696502686,
+      "learning_rate": 9.746380610296435e-05,
+      "loss": 3.3318,
+      "step": 3550
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0634565353393555,
+      "learning_rate": 9.74565125038133e-05,
+      "loss": 3.3427,
+      "step": 3555
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.051472544670105,
+      "learning_rate": 9.744920870592931e-05,
+      "loss": 3.2847,
+      "step": 3560
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.028030514717102,
+      "learning_rate": 9.7441894710882e-05,
+      "loss": 3.3494,
+      "step": 3565
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9588027596473694,
+      "learning_rate": 9.743457052024323e-05,
+      "loss": 3.3754,
+      "step": 3570
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9801216721534729,
+      "learning_rate": 9.742723613558702e-05,
+      "loss": 3.3995,
+      "step": 3575
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0190894603729248,
+      "learning_rate": 9.741989155848953e-05,
+      "loss": 3.4107,
+      "step": 3580
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9818918704986572,
+      "learning_rate": 9.741253679052923e-05,
+      "loss": 3.4015,
+      "step": 3585
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1060587167739868,
+      "learning_rate": 9.740517183328666e-05,
+      "loss": 3.3161,
+      "step": 3590
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0445955991744995,
+      "learning_rate": 9.739779668834461e-05,
+      "loss": 3.4367,
+      "step": 3595
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9551312923431396,
+      "learning_rate": 9.739041135728806e-05,
+      "loss": 3.3891,
+      "step": 3600
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0832080841064453,
+      "learning_rate": 9.738301584170417e-05,
+      "loss": 3.2104,
+      "step": 3605
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5283780097961426,
+      "learning_rate": 9.737561014318228e-05,
+      "loss": 3.4104,
+      "step": 3610
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9377800226211548,
+      "learning_rate": 9.736819426331392e-05,
+      "loss": 3.4287,
+      "step": 3615
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.061420202255249,
+      "learning_rate": 9.736076820369283e-05,
+      "loss": 3.4821,
+      "step": 3620
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0048224925994873,
+      "learning_rate": 9.735333196591492e-05,
+      "loss": 3.3758,
+      "step": 3625
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9804865717887878,
+      "learning_rate": 9.734588555157826e-05,
+      "loss": 3.3474,
+      "step": 3630
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.008792757987976,
+      "learning_rate": 9.733842896228318e-05,
+      "loss": 3.4133,
+      "step": 3635
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9542476534843445,
+      "learning_rate": 9.733096219963211e-05,
+      "loss": 3.2909,
+      "step": 3640
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0257976055145264,
+      "learning_rate": 9.732348526522974e-05,
+      "loss": 3.4855,
+      "step": 3645
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9659914970397949,
+      "learning_rate": 9.73159981606829e-05,
+      "loss": 3.2425,
+      "step": 3650
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1229193210601807,
+      "learning_rate": 9.730850088760064e-05,
+      "loss": 3.3719,
+      "step": 3655
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1302958726882935,
+      "learning_rate": 9.730099344759413e-05,
+      "loss": 3.383,
+      "step": 3660
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0938242673873901,
+      "learning_rate": 9.729347584227684e-05,
+      "loss": 3.2922,
+      "step": 3665
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2072852849960327,
+      "learning_rate": 9.72859480732643e-05,
+      "loss": 3.3482,
+      "step": 3670
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1010493040084839,
+      "learning_rate": 9.727841014217429e-05,
+      "loss": 3.4482,
+      "step": 3675
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9669253826141357,
+      "learning_rate": 9.727086205062676e-05,
+      "loss": 3.3758,
+      "step": 3680
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0073599815368652,
+      "learning_rate": 9.72633038002439e-05,
+      "loss": 3.5038,
+      "step": 3685
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0733524560928345,
+      "learning_rate": 9.725573539264994e-05,
+      "loss": 3.4714,
+      "step": 3690
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9898948073387146,
+      "learning_rate": 9.724815682947146e-05,
+      "loss": 3.3193,
+      "step": 3695
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0156329870224,
+      "learning_rate": 9.72405681123371e-05,
+      "loss": 3.1834,
+      "step": 3700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9766227006912231,
+      "learning_rate": 9.723296924287775e-05,
+      "loss": 3.4284,
+      "step": 3705
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9082093834877014,
+      "learning_rate": 9.722536022272646e-05,
+      "loss": 3.3316,
+      "step": 3710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0939425230026245,
+      "learning_rate": 9.721774105351845e-05,
+      "loss": 3.3012,
+      "step": 3715
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1713863611221313,
+      "learning_rate": 9.721011173689114e-05,
+      "loss": 3.4531,
+      "step": 3720
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0378527641296387,
+      "learning_rate": 9.720247227448411e-05,
+      "loss": 3.3855,
+      "step": 3725
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.001621127128601,
+      "learning_rate": 9.719482266793916e-05,
+      "loss": 3.4471,
+      "step": 3730
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9554819464683533,
+      "learning_rate": 9.718716291890022e-05,
+      "loss": 3.2702,
+      "step": 3735
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.052608847618103,
+      "learning_rate": 9.717949302901343e-05,
+      "loss": 3.2425,
+      "step": 3740
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1667004823684692,
+      "learning_rate": 9.71718129999271e-05,
+      "loss": 3.3268,
+      "step": 3745
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9683908820152283,
+      "learning_rate": 9.716412283329172e-05,
+      "loss": 3.3756,
+      "step": 3750
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9849249124526978,
+      "learning_rate": 9.715642253075998e-05,
+      "loss": 3.3449,
+      "step": 3755
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9330701231956482,
+      "learning_rate": 9.714871209398671e-05,
+      "loss": 3.3884,
+      "step": 3760
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9724490642547607,
+      "learning_rate": 9.714099152462892e-05,
+      "loss": 3.297,
+      "step": 3765
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.977551281452179,
+      "learning_rate": 9.713326082434584e-05,
+      "loss": 3.4879,
+      "step": 3770
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.970198929309845,
+      "learning_rate": 9.712551999479884e-05,
+      "loss": 3.3953,
+      "step": 3775
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0191463232040405,
+      "learning_rate": 9.71177690376515e-05,
+      "loss": 3.3165,
+      "step": 3780
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.023660659790039,
+      "learning_rate": 9.711000795456951e-05,
+      "loss": 3.4669,
+      "step": 3785
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0177133083343506,
+      "learning_rate": 9.710223674722083e-05,
+      "loss": 3.3702,
+      "step": 3790
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0053876638412476,
+      "learning_rate": 9.709445541727551e-05,
+      "loss": 3.4406,
+      "step": 3795
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0708812475204468,
+      "learning_rate": 9.708666396640582e-05,
+      "loss": 3.3645,
+      "step": 3800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1253771781921387,
+      "learning_rate": 9.707886239628621e-05,
+      "loss": 3.3278,
+      "step": 3805
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0104162693023682,
+      "learning_rate": 9.707105070859326e-05,
+      "loss": 3.4004,
+      "step": 3810
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9553461670875549,
+      "learning_rate": 9.706322890500579e-05,
+      "loss": 3.3199,
+      "step": 3815
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.034374475479126,
+      "learning_rate": 9.705539698720476e-05,
+      "loss": 3.4594,
+      "step": 3820
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9739671945571899,
+      "learning_rate": 9.704755495687328e-05,
+      "loss": 3.3512,
+      "step": 3825
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.009758710861206,
+      "learning_rate": 9.703970281569668e-05,
+      "loss": 3.2401,
+      "step": 3830
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9840760827064514,
+      "learning_rate": 9.70318405653624e-05,
+      "loss": 3.4166,
+      "step": 3835
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9864522814750671,
+      "learning_rate": 9.702396820756014e-05,
+      "loss": 3.4362,
+      "step": 3840
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0947898626327515,
+      "learning_rate": 9.701608574398169e-05,
+      "loss": 3.4017,
+      "step": 3845
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9597834348678589,
+      "learning_rate": 9.700819317632107e-05,
+      "loss": 3.2519,
+      "step": 3850
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9407006502151489,
+      "learning_rate": 9.700029050627443e-05,
+      "loss": 3.3667,
+      "step": 3855
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9151723980903625,
+      "learning_rate": 9.699237773554012e-05,
+      "loss": 3.3498,
+      "step": 3860
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0483143329620361,
+      "learning_rate": 9.698445486581864e-05,
+      "loss": 3.4008,
+      "step": 3865
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9937460422515869,
+      "learning_rate": 9.697652189881266e-05,
+      "loss": 3.3307,
+      "step": 3870
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0062798261642456,
+      "learning_rate": 9.696857883622706e-05,
+      "loss": 3.3802,
+      "step": 3875
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0111775398254395,
+      "learning_rate": 9.696062567976885e-05,
+      "loss": 3.3731,
+      "step": 3880
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9286255240440369,
+      "learning_rate": 9.695266243114718e-05,
+      "loss": 3.4021,
+      "step": 3885
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9738205075263977,
+      "learning_rate": 9.694468909207345e-05,
+      "loss": 3.3104,
+      "step": 3890
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0184532403945923,
+      "learning_rate": 9.693670566426119e-05,
+      "loss": 3.3773,
+      "step": 3895
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0427223443984985,
+      "learning_rate": 9.692871214942608e-05,
+      "loss": 3.3314,
+      "step": 3900
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0255345106124878,
+      "learning_rate": 9.692070854928596e-05,
+      "loss": 3.4183,
+      "step": 3905
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0055878162384033,
+      "learning_rate": 9.691269486556088e-05,
+      "loss": 3.3138,
+      "step": 3910
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.033221960067749,
+      "learning_rate": 9.690467109997303e-05,
+      "loss": 3.3549,
+      "step": 3915
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.946124792098999,
+      "learning_rate": 9.68966372542468e-05,
+      "loss": 3.4005,
+      "step": 3920
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9950759410858154,
+      "learning_rate": 9.688859333010866e-05,
+      "loss": 3.3286,
+      "step": 3925
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.982597291469574,
+      "learning_rate": 9.688053932928737e-05,
+      "loss": 3.397,
+      "step": 3930
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9308192133903503,
+      "learning_rate": 9.687247525351375e-05,
+      "loss": 3.2862,
+      "step": 3935
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8803331851959229,
+      "learning_rate": 9.686440110452084e-05,
+      "loss": 3.4402,
+      "step": 3940
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0718430280685425,
+      "learning_rate": 9.685631688404384e-05,
+      "loss": 3.3474,
+      "step": 3945
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.915513813495636,
+      "learning_rate": 9.684822259382008e-05,
+      "loss": 3.3184,
+      "step": 3950
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9154273867607117,
+      "learning_rate": 9.684011823558909e-05,
+      "loss": 3.3381,
+      "step": 3955
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.108784556388855,
+      "learning_rate": 9.683200381109254e-05,
+      "loss": 3.3704,
+      "step": 3960
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9693109393119812,
+      "learning_rate": 9.68238793220743e-05,
+      "loss": 3.317,
+      "step": 3965
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.981911301612854,
+      "learning_rate": 9.681574477028039e-05,
+      "loss": 3.3189,
+      "step": 3970
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0235978364944458,
+      "learning_rate": 9.680760015745894e-05,
+      "loss": 3.2741,
+      "step": 3975
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9810819029808044,
+      "learning_rate": 9.67994454853603e-05,
+      "loss": 3.3369,
+      "step": 3980
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9484671354293823,
+      "learning_rate": 9.679128075573698e-05,
+      "loss": 3.3599,
+      "step": 3985
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0232363939285278,
+      "learning_rate": 9.678310597034364e-05,
+      "loss": 3.2565,
+      "step": 3990
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9815237522125244,
+      "learning_rate": 9.677492113093706e-05,
+      "loss": 3.2631,
+      "step": 3995
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9326673746109009,
+      "learning_rate": 9.676672623927625e-05,
+      "loss": 3.3662,
+      "step": 4000
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0044876337051392,
+      "learning_rate": 9.675852129712234e-05,
+      "loss": 3.3183,
+      "step": 4005
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9291512966156006,
+      "learning_rate": 9.675030630623864e-05,
+      "loss": 3.29,
+      "step": 4010
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0073751211166382,
+      "learning_rate": 9.67420812683906e-05,
+      "loss": 3.3051,
+      "step": 4015
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2806570529937744,
+      "learning_rate": 9.673384618534582e-05,
+      "loss": 3.3709,
+      "step": 4020
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9408910870552063,
+      "learning_rate": 9.672560105887409e-05,
+      "loss": 3.3796,
+      "step": 4025
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9625310897827148,
+      "learning_rate": 9.671734589074736e-05,
+      "loss": 3.341,
+      "step": 4030
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9210394024848938,
+      "learning_rate": 9.670908068273972e-05,
+      "loss": 3.2817,
+      "step": 4035
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0172799825668335,
+      "learning_rate": 9.67008054366274e-05,
+      "loss": 3.3526,
+      "step": 4040
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4452269077301025,
+      "learning_rate": 9.669252015418882e-05,
+      "loss": 3.3396,
+      "step": 4045
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9641324281692505,
+      "learning_rate": 9.668422483720456e-05,
+      "loss": 3.3023,
+      "step": 4050
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0998221635818481,
+      "learning_rate": 9.66759194874573e-05,
+      "loss": 3.3543,
+      "step": 4055
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9620511531829834,
+      "learning_rate": 9.666760410673195e-05,
+      "loss": 3.3018,
+      "step": 4060
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.008937120437622,
+      "learning_rate": 9.665927869681553e-05,
+      "loss": 3.3874,
+      "step": 4065
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8938576579093933,
+      "learning_rate": 9.665094325949723e-05,
+      "loss": 3.2679,
+      "step": 4070
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0615402460098267,
+      "learning_rate": 9.664259779656842e-05,
+      "loss": 3.3711,
+      "step": 4075
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9851990342140198,
+      "learning_rate": 9.663424230982254e-05,
+      "loss": 3.2756,
+      "step": 4080
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9351763725280762,
+      "learning_rate": 9.66258768010553e-05,
+      "loss": 3.2513,
+      "step": 4085
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.992143988609314,
+      "learning_rate": 9.661750127206445e-05,
+      "loss": 3.3392,
+      "step": 4090
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9083009362220764,
+      "learning_rate": 9.660911572465e-05,
+      "loss": 3.2852,
+      "step": 4095
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.998117983341217,
+      "learning_rate": 9.660072016061406e-05,
+      "loss": 3.3775,
+      "step": 4100
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0358247756958008,
+      "learning_rate": 9.659231458176085e-05,
+      "loss": 3.3137,
+      "step": 4105
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9363430142402649,
+      "learning_rate": 9.658389898989684e-05,
+      "loss": 3.3208,
+      "step": 4110
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9514473080635071,
+      "learning_rate": 9.657547338683055e-05,
+      "loss": 3.3266,
+      "step": 4115
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9932163953781128,
+      "learning_rate": 9.656703777437273e-05,
+      "loss": 3.3424,
+      "step": 4120
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.988873302936554,
+      "learning_rate": 9.655859215433625e-05,
+      "loss": 3.3639,
+      "step": 4125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9485275745391846,
+      "learning_rate": 9.655013652853612e-05,
+      "loss": 3.2619,
+      "step": 4130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9783000946044922,
+      "learning_rate": 9.654167089878952e-05,
+      "loss": 3.3322,
+      "step": 4135
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9896408319473267,
+      "learning_rate": 9.653319526691577e-05,
+      "loss": 3.2799,
+      "step": 4140
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9538887739181519,
+      "learning_rate": 9.652470963473635e-05,
+      "loss": 3.3928,
+      "step": 4145
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9736451506614685,
+      "learning_rate": 9.651621400407487e-05,
+      "loss": 3.3543,
+      "step": 4150
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9088242650032043,
+      "learning_rate": 9.650770837675712e-05,
+      "loss": 3.2645,
+      "step": 4155
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0296894311904907,
+      "learning_rate": 9.649919275461099e-05,
+      "loss": 3.3664,
+      "step": 4160
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.293387532234192,
+      "learning_rate": 9.649066713946655e-05,
+      "loss": 3.2645,
+      "step": 4165
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9239462614059448,
+      "learning_rate": 9.648213153315603e-05,
+      "loss": 3.3839,
+      "step": 4170
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9994229674339294,
+      "learning_rate": 9.647358593751378e-05,
+      "loss": 3.3607,
+      "step": 4175
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0165865421295166,
+      "learning_rate": 9.646503035437633e-05,
+      "loss": 3.3363,
+      "step": 4180
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.002223253250122,
+      "learning_rate": 9.64564647855823e-05,
+      "loss": 3.2739,
+      "step": 4185
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9883744120597839,
+      "learning_rate": 9.644788923297253e-05,
+      "loss": 3.3435,
+      "step": 4190
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0210434198379517,
+      "learning_rate": 9.64393036983899e-05,
+      "loss": 3.2525,
+      "step": 4195
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9432671070098877,
+      "learning_rate": 9.643070818367957e-05,
+      "loss": 3.2584,
+      "step": 4200
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9658745527267456,
+      "learning_rate": 9.642210269068874e-05,
+      "loss": 3.3626,
+      "step": 4205
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0684314966201782,
+      "learning_rate": 9.641348722126679e-05,
+      "loss": 3.3866,
+      "step": 4210
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0435491800308228,
+      "learning_rate": 9.640486177726527e-05,
+      "loss": 3.3233,
+      "step": 4215
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0810564756393433,
+      "learning_rate": 9.639622636053781e-05,
+      "loss": 3.2266,
+      "step": 4220
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0151698589324951,
+      "learning_rate": 9.638758097294024e-05,
+      "loss": 3.3221,
+      "step": 4225
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0016512870788574,
+      "learning_rate": 9.637892561633054e-05,
+      "loss": 3.2629,
+      "step": 4230
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0402942895889282,
+      "learning_rate": 9.637026029256874e-05,
+      "loss": 3.3806,
+      "step": 4235
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.982180118560791,
+      "learning_rate": 9.636158500351713e-05,
+      "loss": 3.2225,
+      "step": 4240
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9304065704345703,
+      "learning_rate": 9.635289975104009e-05,
+      "loss": 3.4021,
+      "step": 4245
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9731111526489258,
+      "learning_rate": 9.634420453700413e-05,
+      "loss": 3.225,
+      "step": 4250
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9344516396522522,
+      "learning_rate": 9.633549936327789e-05,
+      "loss": 3.3955,
+      "step": 4255
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9811860918998718,
+      "learning_rate": 9.63267842317322e-05,
+      "loss": 3.2622,
+      "step": 4260
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9226004481315613,
+      "learning_rate": 9.631805914424e-05,
+      "loss": 3.4234,
+      "step": 4265
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8953272104263306,
+      "learning_rate": 9.630932410267637e-05,
+      "loss": 3.4141,
+      "step": 4270
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9837151765823364,
+      "learning_rate": 9.630057910891852e-05,
+      "loss": 3.353,
+      "step": 4275
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9232631921768188,
+      "learning_rate": 9.629182416484582e-05,
+      "loss": 3.3205,
+      "step": 4280
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9516783952713013,
+      "learning_rate": 9.628305927233975e-05,
+      "loss": 3.3311,
+      "step": 4285
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9195518493652344,
+      "learning_rate": 9.627428443328397e-05,
+      "loss": 3.3167,
+      "step": 4290
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9772219061851501,
+      "learning_rate": 9.626549964956424e-05,
+      "loss": 3.2701,
+      "step": 4295
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.946670651435852,
+      "learning_rate": 9.625670492306848e-05,
+      "loss": 3.2348,
+      "step": 4300
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9469735026359558,
+      "learning_rate": 9.624790025568673e-05,
+      "loss": 3.2484,
+      "step": 4305
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9314988255500793,
+      "learning_rate": 9.623908564931117e-05,
+      "loss": 3.2992,
+      "step": 4310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0034656524658203,
+      "learning_rate": 9.623026110583612e-05,
+      "loss": 3.3541,
+      "step": 4315
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9636813402175903,
+      "learning_rate": 9.622142662715808e-05,
+      "loss": 3.4254,
+      "step": 4320
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9690975546836853,
+      "learning_rate": 9.621258221517556e-05,
+      "loss": 3.3455,
+      "step": 4325
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9194464087486267,
+      "learning_rate": 9.620372787178934e-05,
+      "loss": 3.3135,
+      "step": 4330
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1044739484786987,
+      "learning_rate": 9.619486359890228e-05,
+      "loss": 3.2867,
+      "step": 4335
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9759377241134644,
+      "learning_rate": 9.618598939841934e-05,
+      "loss": 3.218,
+      "step": 4340
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9032315611839294,
+      "learning_rate": 9.617710527224768e-05,
+      "loss": 3.3451,
+      "step": 4345
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9108011722564697,
+      "learning_rate": 9.616821122229656e-05,
+      "loss": 3.3557,
+      "step": 4350
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9596870541572571,
+      "learning_rate": 9.615930725047732e-05,
+      "loss": 3.2776,
+      "step": 4355
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0220997333526611,
+      "learning_rate": 9.615039335870357e-05,
+      "loss": 3.306,
+      "step": 4360
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9353731274604797,
+      "learning_rate": 9.614146954889091e-05,
+      "loss": 3.3143,
+      "step": 4365
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9189670085906982,
+      "learning_rate": 9.613253582295714e-05,
+      "loss": 3.342,
+      "step": 4370
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0087100267410278,
+      "learning_rate": 9.612359218282216e-05,
+      "loss": 3.2956,
+      "step": 4375
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9949848651885986,
+      "learning_rate": 9.611463863040807e-05,
+      "loss": 3.3247,
+      "step": 4380
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9730173349380493,
+      "learning_rate": 9.610567516763902e-05,
+      "loss": 3.2551,
+      "step": 4385
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9703141450881958,
+      "learning_rate": 9.60967017964413e-05,
+      "loss": 3.4074,
+      "step": 4390
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9773962497711182,
+      "learning_rate": 9.608771851874338e-05,
+      "loss": 3.2408,
+      "step": 4395
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9969172477722168,
+      "learning_rate": 9.607872533647584e-05,
+      "loss": 3.2761,
+      "step": 4400
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.989789605140686,
+      "learning_rate": 9.606972225157134e-05,
+      "loss": 3.3356,
+      "step": 4405
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9738848805427551,
+      "learning_rate": 9.606070926596471e-05,
+      "loss": 3.3719,
+      "step": 4410
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9570205211639404,
+      "learning_rate": 9.605168638159294e-05,
+      "loss": 3.1724,
+      "step": 4415
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9249124526977539,
+      "learning_rate": 9.604265360039507e-05,
+      "loss": 3.3109,
+      "step": 4420
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.903891384601593,
+      "learning_rate": 9.603361092431234e-05,
+      "loss": 3.3958,
+      "step": 4425
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9579628705978394,
+      "learning_rate": 9.602455835528804e-05,
+      "loss": 3.3154,
+      "step": 4430
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9452762007713318,
+      "learning_rate": 9.601549589526767e-05,
+      "loss": 3.2447,
+      "step": 4435
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9880881309509277,
+      "learning_rate": 9.60064235461988e-05,
+      "loss": 3.3402,
+      "step": 4440
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9981437921524048,
+      "learning_rate": 9.599734131003114e-05,
+      "loss": 3.2998,
+      "step": 4445
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9262902736663818,
+      "learning_rate": 9.598824918871654e-05,
+      "loss": 3.3807,
+      "step": 4450
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0477426052093506,
+      "learning_rate": 9.597914718420892e-05,
+      "loss": 3.2679,
+      "step": 4455
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9843297004699707,
+      "learning_rate": 9.597003529846442e-05,
+      "loss": 3.2544,
+      "step": 4460
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.031115174293518,
+      "learning_rate": 9.59609135334412e-05,
+      "loss": 3.3535,
+      "step": 4465
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9233928322792053,
+      "learning_rate": 9.595178189109963e-05,
+      "loss": 3.2918,
+      "step": 4470
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0340205430984497,
+      "learning_rate": 9.594264037340212e-05,
+      "loss": 3.3102,
+      "step": 4475
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9994635581970215,
+      "learning_rate": 9.593348898231326e-05,
+      "loss": 3.2229,
+      "step": 4480
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9974543452262878,
+      "learning_rate": 9.592432771979978e-05,
+      "loss": 3.2645,
+      "step": 4485
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9204493165016174,
+      "learning_rate": 9.591515658783046e-05,
+      "loss": 3.3015,
+      "step": 4490
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.951542317867279,
+      "learning_rate": 9.590597558837626e-05,
+      "loss": 3.3347,
+      "step": 4495
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8938332200050354,
+      "learning_rate": 9.589678472341023e-05,
+      "loss": 3.2879,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9858137965202332,
+      "learning_rate": 9.588758399490755e-05,
+      "loss": 3.3051,
+      "step": 4505
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9379751682281494,
+      "learning_rate": 9.587837340484553e-05,
+      "loss": 3.2561,
+      "step": 4510
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8969516158103943,
+      "learning_rate": 9.586915295520358e-05,
+      "loss": 3.1581,
+      "step": 4515
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1102604866027832,
+      "learning_rate": 9.585992264796324e-05,
+      "loss": 3.2899,
+      "step": 4520
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9544146656990051,
+      "learning_rate": 9.585068248510817e-05,
+      "loss": 3.3764,
+      "step": 4525
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9384311437606812,
+      "learning_rate": 9.584143246862414e-05,
+      "loss": 3.2636,
+      "step": 4530
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9009018540382385,
+      "learning_rate": 9.583217260049907e-05,
+      "loss": 3.3268,
+      "step": 4535
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9352027773857117,
+      "learning_rate": 9.582290288272291e-05,
+      "loss": 3.2829,
+      "step": 4540
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9379473328590393,
+      "learning_rate": 9.581362331728784e-05,
+      "loss": 3.3758,
+      "step": 4545
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8939939737319946,
+      "learning_rate": 9.580433390618808e-05,
+      "loss": 3.296,
+      "step": 4550
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9407467246055603,
+      "learning_rate": 9.579503465142e-05,
+      "loss": 3.2827,
+      "step": 4555
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9775425791740417,
+      "learning_rate": 9.578572555498207e-05,
+      "loss": 3.3246,
+      "step": 4560
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9288820028305054,
+      "learning_rate": 9.577640661887488e-05,
+      "loss": 3.3197,
+      "step": 4565
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9894906282424927,
+      "learning_rate": 9.576707784510113e-05,
+      "loss": 3.3548,
+      "step": 4570
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9209339618682861,
+      "learning_rate": 9.575773923566563e-05,
+      "loss": 3.3439,
+      "step": 4575
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9034512042999268,
+      "learning_rate": 9.574839079257532e-05,
+      "loss": 3.355,
+      "step": 4580
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8943424820899963,
+      "learning_rate": 9.573903251783925e-05,
+      "loss": 3.2754,
+      "step": 4585
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.939596951007843,
+      "learning_rate": 9.572966441346858e-05,
+      "loss": 3.2734,
+      "step": 4590
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9413349628448486,
+      "learning_rate": 9.572028648147657e-05,
+      "loss": 3.2386,
+      "step": 4595
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1349595785140991,
+      "learning_rate": 9.571089872387862e-05,
+      "loss": 3.2828,
+      "step": 4600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9806763529777527,
+      "learning_rate": 9.57015011426922e-05,
+      "loss": 3.2637,
+      "step": 4605
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9659852385520935,
+      "learning_rate": 9.569209373993693e-05,
+      "loss": 3.3311,
+      "step": 4610
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0590827465057373,
+      "learning_rate": 9.568267651763451e-05,
+      "loss": 3.2763,
+      "step": 4615
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9480798244476318,
+      "learning_rate": 9.56732494778088e-05,
+      "loss": 3.3207,
+      "step": 4620
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2531214952468872,
+      "learning_rate": 9.56638126224857e-05,
+      "loss": 3.3673,
+      "step": 4625
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.162964940071106,
+      "learning_rate": 9.565436595369328e-05,
+      "loss": 3.2236,
+      "step": 4630
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9878843426704407,
+      "learning_rate": 9.564490947346168e-05,
+      "loss": 3.2894,
+      "step": 4635
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0094283819198608,
+      "learning_rate": 9.563544318382316e-05,
+      "loss": 3.2716,
+      "step": 4640
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.008020043373108,
+      "learning_rate": 9.56259670868121e-05,
+      "loss": 3.2199,
+      "step": 4645
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9279943108558655,
+      "learning_rate": 9.5616481184465e-05,
+      "loss": 3.3133,
+      "step": 4650
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9741253852844238,
+      "learning_rate": 9.560698547882041e-05,
+      "loss": 3.2429,
+      "step": 4655
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0694022178649902,
+      "learning_rate": 9.559747997191903e-05,
+      "loss": 3.2291,
+      "step": 4660
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9383769035339355,
+      "learning_rate": 9.558796466580366e-05,
+      "loss": 3.2954,
+      "step": 4665
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.981046736240387,
+      "learning_rate": 9.557843956251924e-05,
+      "loss": 3.2377,
+      "step": 4670
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9283106923103333,
+      "learning_rate": 9.556890466411272e-05,
+      "loss": 3.258,
+      "step": 4675
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.020955204963684,
+      "learning_rate": 9.555935997263328e-05,
+      "loss": 3.2848,
+      "step": 4680
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9723761677742004,
+      "learning_rate": 9.554980549013209e-05,
+      "loss": 3.1668,
+      "step": 4685
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9943270683288574,
+      "learning_rate": 9.554024121866249e-05,
+      "loss": 3.3771,
+      "step": 4690
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9659664630889893,
+      "learning_rate": 9.55306671602799e-05,
+      "loss": 3.3089,
+      "step": 4695
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9081193804740906,
+      "learning_rate": 9.552108331704189e-05,
+      "loss": 3.2909,
+      "step": 4700
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1502587795257568,
+      "learning_rate": 9.551148969100804e-05,
+      "loss": 3.3051,
+      "step": 4705
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0840284824371338,
+      "learning_rate": 9.550188628424012e-05,
+      "loss": 3.2805,
+      "step": 4710
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0257043838500977,
+      "learning_rate": 9.549227309880197e-05,
+      "loss": 3.2796,
+      "step": 4715
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9422656893730164,
+      "learning_rate": 9.548265013675951e-05,
+      "loss": 3.3049,
+      "step": 4720
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9991198778152466,
+      "learning_rate": 9.54730174001808e-05,
+      "loss": 3.2948,
+      "step": 4725
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2447710037231445,
+      "learning_rate": 9.546337489113597e-05,
+      "loss": 3.3122,
+      "step": 4730
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1291483640670776,
+      "learning_rate": 9.545372261169725e-05,
+      "loss": 3.2721,
+      "step": 4735
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.06948983669281,
+      "learning_rate": 9.544406056393903e-05,
+      "loss": 3.2378,
+      "step": 4740
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9939368367195129,
+      "learning_rate": 9.54343887499377e-05,
+      "loss": 3.2383,
+      "step": 4745
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9922817349433899,
+      "learning_rate": 9.542470717177182e-05,
+      "loss": 3.2991,
+      "step": 4750
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0056777000427246,
+      "learning_rate": 9.541501583152204e-05,
+      "loss": 3.2695,
+      "step": 4755
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1011972427368164,
+      "learning_rate": 9.540531473127106e-05,
+      "loss": 3.2661,
+      "step": 4760
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9922934770584106,
+      "learning_rate": 9.539560387310375e-05,
+      "loss": 3.2391,
+      "step": 4765
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0375553369522095,
+      "learning_rate": 9.538588325910702e-05,
+      "loss": 3.1638,
+      "step": 4770
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9696425199508667,
+      "learning_rate": 9.53761528913699e-05,
+      "loss": 3.338,
+      "step": 4775
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9250494837760925,
+      "learning_rate": 9.536641277198353e-05,
+      "loss": 3.2108,
+      "step": 4780
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.978912353515625,
+      "learning_rate": 9.535666290304111e-05,
+      "loss": 3.1868,
+      "step": 4785
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9513201117515564,
+      "learning_rate": 9.534690328663796e-05,
+      "loss": 3.3478,
+      "step": 4790
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9592811465263367,
+      "learning_rate": 9.533713392487149e-05,
+      "loss": 3.1839,
+      "step": 4795
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0029288530349731,
+      "learning_rate": 9.532735481984121e-05,
+      "loss": 3.2838,
+      "step": 4800
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8990973234176636,
+      "learning_rate": 9.53175659736487e-05,
+      "loss": 3.244,
+      "step": 4805
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9940216541290283,
+      "learning_rate": 9.530776738839766e-05,
+      "loss": 3.2323,
+      "step": 4810
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9352378249168396,
+      "learning_rate": 9.529795906619387e-05,
+      "loss": 3.216,
+      "step": 4815
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9326065182685852,
+      "learning_rate": 9.528814100914522e-05,
+      "loss": 3.3473,
+      "step": 4820
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9078368544578552,
+      "learning_rate": 9.52783132193617e-05,
+      "loss": 3.2468,
+      "step": 4825
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.972861647605896,
+      "learning_rate": 9.52684756989553e-05,
+      "loss": 3.2496,
+      "step": 4830
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9495263695716858,
+      "learning_rate": 9.525862845004022e-05,
+      "loss": 3.2301,
+      "step": 4835
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9877751469612122,
+      "learning_rate": 9.52487714747327e-05,
+      "loss": 3.2754,
+      "step": 4840
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9694097638130188,
+      "learning_rate": 9.523890477515105e-05,
+      "loss": 3.3253,
+      "step": 4845
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9513000845909119,
+      "learning_rate": 9.522902835341573e-05,
+      "loss": 3.2491,
+      "step": 4850
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2141855955123901,
+      "learning_rate": 9.521914221164922e-05,
+      "loss": 3.2354,
+      "step": 4855
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.165208339691162,
+      "learning_rate": 9.520924635197612e-05,
+      "loss": 3.2223,
+      "step": 4860
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9424464106559753,
+      "learning_rate": 9.519934077652312e-05,
+      "loss": 3.1908,
+      "step": 4865
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9519426226615906,
+      "learning_rate": 9.518942548741902e-05,
+      "loss": 3.2904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9222538471221924,
+      "learning_rate": 9.517950048679467e-05,
+      "loss": 3.2389,
+      "step": 4875
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9294736981391907,
+      "learning_rate": 9.516956577678302e-05,
+      "loss": 3.2001,
+      "step": 4880
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.908400297164917,
+      "learning_rate": 9.51596213595191e-05,
+      "loss": 3.1927,
+      "step": 4885
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9579713344573975,
+      "learning_rate": 9.514966723714004e-05,
+      "loss": 3.3126,
+      "step": 4890
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9537719488143921,
+      "learning_rate": 9.513970341178506e-05,
+      "loss": 3.2829,
+      "step": 4895
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9682953953742981,
+      "learning_rate": 9.512972988559545e-05,
+      "loss": 3.3125,
+      "step": 4900
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0025780200958252,
+      "learning_rate": 9.511974666071458e-05,
+      "loss": 3.2675,
+      "step": 4905
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9651876091957092,
+      "learning_rate": 9.510975373928792e-05,
+      "loss": 3.2889,
+      "step": 4910
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9805078506469727,
+      "learning_rate": 9.509975112346303e-05,
+      "loss": 3.2521,
+      "step": 4915
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.143365502357483,
+      "learning_rate": 9.508973881538952e-05,
+      "loss": 3.2073,
+      "step": 4920
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9856778979301453,
+      "learning_rate": 9.507971681721911e-05,
+      "loss": 3.2185,
+      "step": 4925
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8946225643157959,
+      "learning_rate": 9.50696851311056e-05,
+      "loss": 3.3035,
+      "step": 4930
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9325090646743774,
+      "learning_rate": 9.505964375920489e-05,
+      "loss": 3.2821,
+      "step": 4935
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9487035870552063,
+      "learning_rate": 9.50495927036749e-05,
+      "loss": 3.1613,
+      "step": 4940
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9773942232131958,
+      "learning_rate": 9.503953196667569e-05,
+      "loss": 3.1835,
+      "step": 4945
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9351685047149658,
+      "learning_rate": 9.502946155036938e-05,
+      "loss": 3.2866,
+      "step": 4950
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9533562660217285,
+      "learning_rate": 9.50193814569202e-05,
+      "loss": 3.2759,
+      "step": 4955
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9264758229255676,
+      "learning_rate": 9.500929168849439e-05,
+      "loss": 3.3178,
+      "step": 4960
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8942647576332092,
+      "learning_rate": 9.499919224726032e-05,
+      "loss": 3.2695,
+      "step": 4965
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9583644270896912,
+      "learning_rate": 9.498908313538846e-05,
+      "loss": 3.1433,
+      "step": 4970
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.064148187637329,
+      "learning_rate": 9.49789643550513e-05,
+      "loss": 3.1914,
+      "step": 4975
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.96927410364151,
+      "learning_rate": 9.496883590842343e-05,
+      "loss": 3.2648,
+      "step": 4980
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8901596069335938,
+      "learning_rate": 9.495869779768154e-05,
+      "loss": 3.3044,
+      "step": 4985
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9301221370697021,
+      "learning_rate": 9.494855002500438e-05,
+      "loss": 3.1097,
+      "step": 4990
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9628689289093018,
+      "learning_rate": 9.493839259257278e-05,
+      "loss": 3.2502,
+      "step": 4995
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8811522722244263,
+      "learning_rate": 9.492822550256962e-05,
+      "loss": 3.3036,
+      "step": 5000
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.938870906829834,
+      "learning_rate": 9.49180487571799e-05,
+      "loss": 3.2656,
+      "step": 5005
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8680905699729919,
+      "learning_rate": 9.490786235859066e-05,
+      "loss": 3.2104,
+      "step": 5010
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.028796672821045,
+      "learning_rate": 9.489766630899104e-05,
+      "loss": 3.1976,
+      "step": 5015
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9754784107208252,
+      "learning_rate": 9.488746061057223e-05,
+      "loss": 3.1626,
+      "step": 5020
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9626196622848511,
+      "learning_rate": 9.487724526552753e-05,
+      "loss": 3.2342,
+      "step": 5025
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9482839107513428,
+      "learning_rate": 9.486702027605224e-05,
+      "loss": 3.2396,
+      "step": 5030
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9194731116294861,
+      "learning_rate": 9.485678564434381e-05,
+      "loss": 3.1722,
+      "step": 5035
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9679526686668396,
+      "learning_rate": 9.484654137260176e-05,
+      "loss": 3.3129,
+      "step": 5040
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9044382572174072,
+      "learning_rate": 9.483628746302761e-05,
+      "loss": 3.0986,
+      "step": 5045
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9199784994125366,
+      "learning_rate": 9.482602391782501e-05,
+      "loss": 3.1951,
+      "step": 5050
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9498360753059387,
+      "learning_rate": 9.481575073919968e-05,
+      "loss": 3.2598,
+      "step": 5055
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.038529634475708,
+      "learning_rate": 9.480546792935939e-05,
+      "loss": 3.235,
+      "step": 5060
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9499696493148804,
+      "learning_rate": 9.479517549051398e-05,
+      "loss": 3.1941,
+      "step": 5065
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9869998097419739,
+      "learning_rate": 9.478487342487536e-05,
+      "loss": 3.2315,
+      "step": 5070
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8991618752479553,
+      "learning_rate": 9.477456173465754e-05,
+      "loss": 3.2345,
+      "step": 5075
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9022316336631775,
+      "learning_rate": 9.476424042207655e-05,
+      "loss": 3.1896,
+      "step": 5080
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8768622279167175,
+      "learning_rate": 9.475390948935054e-05,
+      "loss": 3.2613,
+      "step": 5085
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8919715881347656,
+      "learning_rate": 9.474356893869967e-05,
+      "loss": 3.2667,
+      "step": 5090
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8786368370056152,
+      "learning_rate": 9.47332187723462e-05,
+      "loss": 3.2044,
+      "step": 5095
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9413724541664124,
+      "learning_rate": 9.472285899251445e-05,
+      "loss": 3.2619,
+      "step": 5100
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0225499868392944,
+      "learning_rate": 9.471248960143085e-05,
+      "loss": 3.1675,
+      "step": 5105
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.470983028411865,
+      "learning_rate": 9.47021106013238e-05,
+      "loss": 3.224,
+      "step": 5110
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9497906565666199,
+      "learning_rate": 9.469172199442383e-05,
+      "loss": 3.2005,
+      "step": 5115
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.894568920135498,
+      "learning_rate": 9.468132378296355e-05,
+      "loss": 3.3054,
+      "step": 5120
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0519976615905762,
+      "learning_rate": 9.467091596917758e-05,
+      "loss": 3.2722,
+      "step": 5125
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9790306687355042,
+      "learning_rate": 9.466049855530265e-05,
+      "loss": 3.1735,
+      "step": 5130
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9697803258895874,
+      "learning_rate": 9.465007154357752e-05,
+      "loss": 3.1679,
+      "step": 5135
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9840431213378906,
+      "learning_rate": 9.463963493624304e-05,
+      "loss": 3.1796,
+      "step": 5140
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0849897861480713,
+      "learning_rate": 9.462918873554209e-05,
+      "loss": 3.2028,
+      "step": 5145
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9015127420425415,
+      "learning_rate": 9.461873294371965e-05,
+      "loss": 3.154,
+      "step": 5150
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8814483880996704,
+      "learning_rate": 9.460826756302275e-05,
+      "loss": 3.1707,
+      "step": 5155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8743109107017517,
+      "learning_rate": 9.459779259570045e-05,
+      "loss": 3.2114,
+      "step": 5160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.989276111125946,
+      "learning_rate": 9.458730804400389e-05,
+      "loss": 3.161,
+      "step": 5165
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9580052495002747,
+      "learning_rate": 9.457681391018629e-05,
+      "loss": 3.2211,
+      "step": 5170
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9198920726776123,
+      "learning_rate": 9.45663101965029e-05,
+      "loss": 3.195,
+      "step": 5175
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9162493348121643,
+      "learning_rate": 9.455579690521106e-05,
+      "loss": 3.2536,
+      "step": 5180
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9283279776573181,
+      "learning_rate": 9.454527403857013e-05,
+      "loss": 3.2808,
+      "step": 5185
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9870271682739258,
+      "learning_rate": 9.453474159884155e-05,
+      "loss": 3.1987,
+      "step": 5190
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0627527236938477,
+      "learning_rate": 9.452419958828883e-05,
+      "loss": 3.2313,
+      "step": 5195
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9659613370895386,
+      "learning_rate": 9.45136480091775e-05,
+      "loss": 3.17,
+      "step": 5200
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2546665668487549,
+      "learning_rate": 9.450308686377516e-05,
+      "loss": 3.3152,
+      "step": 5205
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9374608993530273,
+      "learning_rate": 9.449251615435151e-05,
+      "loss": 3.2145,
+      "step": 5210
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9286046028137207,
+      "learning_rate": 9.448193588317822e-05,
+      "loss": 3.2241,
+      "step": 5215
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9130064845085144,
+      "learning_rate": 9.447134605252912e-05,
+      "loss": 3.2071,
+      "step": 5220
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0997915267944336,
+      "learning_rate": 9.446074666468e-05,
+      "loss": 3.2318,
+      "step": 5225
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0295606851577759,
+      "learning_rate": 9.445013772190874e-05,
+      "loss": 3.2951,
+      "step": 5230
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9639924764633179,
+      "learning_rate": 9.443951922649529e-05,
+      "loss": 3.2808,
+      "step": 5235
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9150679707527161,
+      "learning_rate": 9.442889118072162e-05,
+      "loss": 3.2824,
+      "step": 5240
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.901676595211029,
+      "learning_rate": 9.441825358687181e-05,
+      "loss": 3.1468,
+      "step": 5245
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9107379913330078,
+      "learning_rate": 9.440760644723192e-05,
+      "loss": 3.1305,
+      "step": 5250
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0108879804611206,
+      "learning_rate": 9.439694976409008e-05,
+      "loss": 3.2723,
+      "step": 5255
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.954930305480957,
+      "learning_rate": 9.438628353973653e-05,
+      "loss": 3.2332,
+      "step": 5260
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9375127553939819,
+      "learning_rate": 9.437560777646348e-05,
+      "loss": 3.2482,
+      "step": 5265
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0516828298568726,
+      "learning_rate": 9.436492247656523e-05,
+      "loss": 3.3517,
+      "step": 5270
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9580940008163452,
+      "learning_rate": 9.435422764233814e-05,
+      "loss": 3.3,
+      "step": 5275
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9752517342567444,
+      "learning_rate": 9.43435232760806e-05,
+      "loss": 3.2235,
+      "step": 5280
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9282833337783813,
+      "learning_rate": 9.433280938009304e-05,
+      "loss": 3.1718,
+      "step": 5285
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9630972146987915,
+      "learning_rate": 9.432208595667797e-05,
+      "loss": 3.1955,
+      "step": 5290
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8627937436103821,
+      "learning_rate": 9.431135300813989e-05,
+      "loss": 3.1968,
+      "step": 5295
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9036207795143127,
+      "learning_rate": 9.430061053678542e-05,
+      "loss": 3.167,
+      "step": 5300
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9075141549110413,
+      "learning_rate": 9.42898585449232e-05,
+      "loss": 3.2008,
+      "step": 5305
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9554218649864197,
+      "learning_rate": 9.427909703486388e-05,
+      "loss": 3.3015,
+      "step": 5310
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9110196232795715,
+      "learning_rate": 9.426832600892019e-05,
+      "loss": 3.1462,
+      "step": 5315
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9299665093421936,
+      "learning_rate": 9.425754546940691e-05,
+      "loss": 3.3425,
+      "step": 5320
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9102502465248108,
+      "learning_rate": 9.424675541864081e-05,
+      "loss": 3.3197,
+      "step": 5325
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9179167747497559,
+      "learning_rate": 9.423595585894082e-05,
+      "loss": 3.2412,
+      "step": 5330
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.078345537185669,
+      "learning_rate": 9.422514679262778e-05,
+      "loss": 3.0983,
+      "step": 5335
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9529665112495422,
+      "learning_rate": 9.421432822202465e-05,
+      "loss": 3.156,
+      "step": 5340
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9041405916213989,
+      "learning_rate": 9.420350014945643e-05,
+      "loss": 3.2691,
+      "step": 5345
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.937498927116394,
+      "learning_rate": 9.419266257725013e-05,
+      "loss": 3.2171,
+      "step": 5350
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9096983671188354,
+      "learning_rate": 9.418181550773483e-05,
+      "loss": 3.1662,
+      "step": 5355
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9198148250579834,
+      "learning_rate": 9.417095894324163e-05,
+      "loss": 3.3028,
+      "step": 5360
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.005050539970398,
+      "learning_rate": 9.41600928861037e-05,
+      "loss": 3.2348,
+      "step": 5365
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9785653352737427,
+      "learning_rate": 9.41492173386562e-05,
+      "loss": 3.1288,
+      "step": 5370
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.031983733177185,
+      "learning_rate": 9.413833230323639e-05,
+      "loss": 3.2015,
+      "step": 5375
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9726940393447876,
+      "learning_rate": 9.412743778218351e-05,
+      "loss": 3.167,
+      "step": 5380
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8828642964363098,
+      "learning_rate": 9.41165337778389e-05,
+      "loss": 3.1446,
+      "step": 5385
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9154539108276367,
+      "learning_rate": 9.410562029254589e-05,
+      "loss": 3.2449,
+      "step": 5390
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8780710101127625,
+      "learning_rate": 9.409469732864986e-05,
+      "loss": 3.1989,
+      "step": 5395
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9317775368690491,
+      "learning_rate": 9.408376488849824e-05,
+      "loss": 3.1379,
+      "step": 5400
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8550082445144653,
+      "learning_rate": 9.407282297444048e-05,
+      "loss": 3.2324,
+      "step": 5405
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9069817662239075,
+      "learning_rate": 9.406187158882808e-05,
+      "loss": 3.1792,
+      "step": 5410
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9753915667533875,
+      "learning_rate": 9.405091073401458e-05,
+      "loss": 3.2248,
+      "step": 5415
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8950713872909546,
+      "learning_rate": 9.40399404123555e-05,
+      "loss": 3.152,
+      "step": 5420
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0405269861221313,
+      "learning_rate": 9.402896062620848e-05,
+      "loss": 3.2076,
+      "step": 5425
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9896488785743713,
+      "learning_rate": 9.401797137793313e-05,
+      "loss": 3.1936,
+      "step": 5430
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9500636458396912,
+      "learning_rate": 9.400697266989112e-05,
+      "loss": 3.2317,
+      "step": 5435
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1299787759780884,
+      "learning_rate": 9.399596450444617e-05,
+      "loss": 3.2096,
+      "step": 5440
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9217138886451721,
+      "learning_rate": 9.398494688396398e-05,
+      "loss": 3.1062,
+      "step": 5445
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8928692936897278,
+      "learning_rate": 9.397391981081232e-05,
+      "loss": 3.3015,
+      "step": 5450
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9240047335624695,
+      "learning_rate": 9.3962883287361e-05,
+      "loss": 3.2324,
+      "step": 5455
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0400786399841309,
+      "learning_rate": 9.395183731598183e-05,
+      "loss": 3.2933,
+      "step": 5460
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9329736828804016,
+      "learning_rate": 9.394078189904867e-05,
+      "loss": 3.1864,
+      "step": 5465
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0471819639205933,
+      "learning_rate": 9.392971703893741e-05,
+      "loss": 3.2418,
+      "step": 5470
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0246738195419312,
+      "learning_rate": 9.391864273802595e-05,
+      "loss": 3.2585,
+      "step": 5475
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9757674932479858,
+      "learning_rate": 9.390755899869425e-05,
+      "loss": 3.1165,
+      "step": 5480
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0463083982467651,
+      "learning_rate": 9.38964658233243e-05,
+      "loss": 3.0644,
+      "step": 5485
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8826867341995239,
+      "learning_rate": 9.388536321430005e-05,
+      "loss": 3.2673,
+      "step": 5490
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0055259466171265,
+      "learning_rate": 9.387425117400757e-05,
+      "loss": 3.1123,
+      "step": 5495
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8918637633323669,
+      "learning_rate": 9.386312970483489e-05,
+      "loss": 3.1673,
+      "step": 5500
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9329797625541687,
+      "learning_rate": 9.385199880917211e-05,
+      "loss": 3.1518,
+      "step": 5505
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9294643998146057,
+      "learning_rate": 9.384085848941132e-05,
+      "loss": 3.2264,
+      "step": 5510
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9557133316993713,
+      "learning_rate": 9.382970874794666e-05,
+      "loss": 3.1649,
+      "step": 5515
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.924607515335083,
+      "learning_rate": 9.381854958717429e-05,
+      "loss": 3.2037,
+      "step": 5520
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9684540629386902,
+      "learning_rate": 9.380738100949238e-05,
+      "loss": 3.2074,
+      "step": 5525
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9403497576713562,
+      "learning_rate": 9.379620301730114e-05,
+      "loss": 3.2532,
+      "step": 5530
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8932996988296509,
+      "learning_rate": 9.378501561300282e-05,
+      "loss": 3.2251,
+      "step": 5535
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9457070827484131,
+      "learning_rate": 9.377381879900162e-05,
+      "loss": 3.0962,
+      "step": 5540
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9110062718391418,
+      "learning_rate": 9.376261257770384e-05,
+      "loss": 3.2203,
+      "step": 5545
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9656655192375183,
+      "learning_rate": 9.37513969515178e-05,
+      "loss": 3.2312,
+      "step": 5550
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9291927814483643,
+      "learning_rate": 9.374017192285378e-05,
+      "loss": 3.281,
+      "step": 5555
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9000664949417114,
+      "learning_rate": 9.372893749412413e-05,
+      "loss": 3.1599,
+      "step": 5560
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0275307893753052,
+      "learning_rate": 9.371769366774321e-05,
+      "loss": 3.2126,
+      "step": 5565
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.00016188621521,
+      "learning_rate": 9.370644044612738e-05,
+      "loss": 3.1587,
+      "step": 5570
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9388554096221924,
+      "learning_rate": 9.369517783169505e-05,
+      "loss": 3.1978,
+      "step": 5575
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9767546057701111,
+      "learning_rate": 9.368390582686663e-05,
+      "loss": 3.1958,
+      "step": 5580
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9712271094322205,
+      "learning_rate": 9.367262443406456e-05,
+      "loss": 3.18,
+      "step": 5585
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9377459287643433,
+      "learning_rate": 9.366133365571328e-05,
+      "loss": 3.1192,
+      "step": 5590
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0796289443969727,
+      "learning_rate": 9.365003349423924e-05,
+      "loss": 3.157,
+      "step": 5595
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9441500902175903,
+      "learning_rate": 9.363872395207097e-05,
+      "loss": 3.2427,
+      "step": 5600
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9466326832771301,
+      "learning_rate": 9.362740503163891e-05,
+      "loss": 3.1579,
+      "step": 5605
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9540475606918335,
+      "learning_rate": 9.361607673537563e-05,
+      "loss": 3.2779,
+      "step": 5610
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9260580539703369,
+      "learning_rate": 9.36047390657156e-05,
+      "loss": 3.097,
+      "step": 5615
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0573956966400146,
+      "learning_rate": 9.359339202509543e-05,
+      "loss": 3.1868,
+      "step": 5620
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8822178244590759,
+      "learning_rate": 9.358203561595363e-05,
+      "loss": 3.1868,
+      "step": 5625
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9203696846961975,
+      "learning_rate": 9.357066984073079e-05,
+      "loss": 3.1814,
+      "step": 5630
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.969761073589325,
+      "learning_rate": 9.355929470186948e-05,
+      "loss": 3.1909,
+      "step": 5635
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9622836112976074,
+      "learning_rate": 9.354791020181431e-05,
+      "loss": 3.1808,
+      "step": 5640
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9908729195594788,
+      "learning_rate": 9.353651634301188e-05,
+      "loss": 3.2345,
+      "step": 5645
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9702188968658447,
+      "learning_rate": 9.352511312791083e-05,
+      "loss": 3.1245,
+      "step": 5650
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0287315845489502,
+      "learning_rate": 9.351370055896175e-05,
+      "loss": 3.2238,
+      "step": 5655
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9500437378883362,
+      "learning_rate": 9.350227863861732e-05,
+      "loss": 3.221,
+      "step": 5660
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9244545698165894,
+      "learning_rate": 9.349084736933215e-05,
+      "loss": 3.1744,
+      "step": 5665
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.010849952697754,
+      "learning_rate": 9.347940675356292e-05,
+      "loss": 3.2932,
+      "step": 5670
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8932321667671204,
+      "learning_rate": 9.346795679376832e-05,
+      "loss": 3.1411,
+      "step": 5675
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8974558711051941,
+      "learning_rate": 9.345649749240898e-05,
+      "loss": 3.1695,
+      "step": 5680
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9299780130386353,
+      "learning_rate": 9.344502885194763e-05,
+      "loss": 3.1171,
+      "step": 5685
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8687735795974731,
+      "learning_rate": 9.343355087484894e-05,
+      "loss": 3.2247,
+      "step": 5690
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9020839333534241,
+      "learning_rate": 9.342206356357956e-05,
+      "loss": 3.2236,
+      "step": 5695
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9358768463134766,
+      "learning_rate": 9.341056692060828e-05,
+      "loss": 3.1556,
+      "step": 5700
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9601712226867676,
+      "learning_rate": 9.339906094840575e-05,
+      "loss": 3.1569,
+      "step": 5705
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9458848834037781,
+      "learning_rate": 9.33875456494447e-05,
+      "loss": 3.244,
+      "step": 5710
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8780253529548645,
+      "learning_rate": 9.337602102619985e-05,
+      "loss": 3.2001,
+      "step": 5715
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8900538682937622,
+      "learning_rate": 9.336448708114789e-05,
+      "loss": 3.2017,
+      "step": 5720
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9224235415458679,
+      "learning_rate": 9.33529438167676e-05,
+      "loss": 3.0868,
+      "step": 5725
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9325925707817078,
+      "learning_rate": 9.334139123553967e-05,
+      "loss": 3.13,
+      "step": 5730
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9136790037155151,
+      "learning_rate": 9.332982933994683e-05,
+      "loss": 3.1029,
+      "step": 5735
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9777320623397827,
+      "learning_rate": 9.331825813247381e-05,
+      "loss": 3.1621,
+      "step": 5740
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9237834215164185,
+      "learning_rate": 9.330667761560739e-05,
+      "loss": 3.1336,
+      "step": 5745
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9072224497795105,
+      "learning_rate": 9.329508779183623e-05,
+      "loss": 3.1546,
+      "step": 5750
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8217813968658447,
+      "learning_rate": 9.328348866365111e-05,
+      "loss": 3.1388,
+      "step": 5755
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9015535116195679,
+      "learning_rate": 9.327188023354474e-05,
+      "loss": 3.2455,
+      "step": 5760
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9820228815078735,
+      "learning_rate": 9.326026250401188e-05,
+      "loss": 3.0994,
+      "step": 5765
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9393866658210754,
+      "learning_rate": 9.324863547754924e-05,
+      "loss": 3.1575,
+      "step": 5770
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9747145175933838,
+      "learning_rate": 9.323699915665555e-05,
+      "loss": 3.2403,
+      "step": 5775
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9315499067306519,
+      "learning_rate": 9.322535354383153e-05,
+      "loss": 3.1679,
+      "step": 5780
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9045815467834473,
+      "learning_rate": 9.321369864157992e-05,
+      "loss": 3.2088,
+      "step": 5785
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8655470013618469,
+      "learning_rate": 9.320203445240543e-05,
+      "loss": 3.0423,
+      "step": 5790
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9494146704673767,
+      "learning_rate": 9.319036097881477e-05,
+      "loss": 3.1692,
+      "step": 5795
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9095951914787292,
+      "learning_rate": 9.317867822331667e-05,
+      "loss": 3.134,
+      "step": 5800
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9827576875686646,
+      "learning_rate": 9.31669861884218e-05,
+      "loss": 3.1722,
+      "step": 5805
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9075523614883423,
+      "learning_rate": 9.315528487664289e-05,
+      "loss": 3.1927,
+      "step": 5810
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0939371585845947,
+      "learning_rate": 9.314357429049462e-05,
+      "loss": 3.2319,
+      "step": 5815
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9768454432487488,
+      "learning_rate": 9.313185443249367e-05,
+      "loss": 3.2111,
+      "step": 5820
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9234619140625,
+      "learning_rate": 9.312012530515875e-05,
+      "loss": 3.1601,
+      "step": 5825
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9989291429519653,
+      "learning_rate": 9.310838691101048e-05,
+      "loss": 3.1616,
+      "step": 5830
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8513902425765991,
+      "learning_rate": 9.309663925257156e-05,
+      "loss": 3.1896,
+      "step": 5835
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8945289850234985,
+      "learning_rate": 9.308488233236663e-05,
+      "loss": 3.1748,
+      "step": 5840
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0268059968948364,
+      "learning_rate": 9.307311615292234e-05,
+      "loss": 3.183,
+      "step": 5845
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9666544795036316,
+      "learning_rate": 9.306134071676734e-05,
+      "loss": 3.1822,
+      "step": 5850
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9162259697914124,
+      "learning_rate": 9.30495560264322e-05,
+      "loss": 3.191,
+      "step": 5855
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8758249878883362,
+      "learning_rate": 9.303776208444958e-05,
+      "loss": 3.1403,
+      "step": 5860
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9276066422462463,
+      "learning_rate": 9.302595889335405e-05,
+      "loss": 3.2504,
+      "step": 5865
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9003704190254211,
+      "learning_rate": 9.301414645568223e-05,
+      "loss": 3.2006,
+      "step": 5870
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8549237251281738,
+      "learning_rate": 9.300232477397266e-05,
+      "loss": 3.1891,
+      "step": 5875
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9674870371818542,
+      "learning_rate": 9.299049385076592e-05,
+      "loss": 3.228,
+      "step": 5880
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9921637773513794,
+      "learning_rate": 9.297865368860456e-05,
+      "loss": 3.1449,
+      "step": 5885
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9164543747901917,
+      "learning_rate": 9.296680429003311e-05,
+      "loss": 3.311,
+      "step": 5890
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8793285489082336,
+      "learning_rate": 9.295494565759808e-05,
+      "loss": 3.2971,
+      "step": 5895
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9040293097496033,
+      "learning_rate": 9.294307779384796e-05,
+      "loss": 3.1942,
+      "step": 5900
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8730707168579102,
+      "learning_rate": 9.293120070133327e-05,
+      "loss": 3.1834,
+      "step": 5905
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8940589427947998,
+      "learning_rate": 9.291931438260646e-05,
+      "loss": 3.0519,
+      "step": 5910
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.984096348285675,
+      "learning_rate": 9.290741884022198e-05,
+      "loss": 3.1798,
+      "step": 5915
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.893770694732666,
+      "learning_rate": 9.28955140767363e-05,
+      "loss": 3.2538,
+      "step": 5920
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.94364994764328,
+      "learning_rate": 9.288360009470776e-05,
+      "loss": 3.1892,
+      "step": 5925
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9141552448272705,
+      "learning_rate": 9.287167689669683e-05,
+      "loss": 3.1299,
+      "step": 5930
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8551222085952759,
+      "learning_rate": 9.285974448526585e-05,
+      "loss": 3.1059,
+      "step": 5935
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8962435126304626,
+      "learning_rate": 9.284780286297918e-05,
+      "loss": 3.1779,
+      "step": 5940
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9152446985244751,
+      "learning_rate": 9.283585203240317e-05,
+      "loss": 3.1538,
+      "step": 5945
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8970886468887329,
+      "learning_rate": 9.282389199610613e-05,
+      "loss": 3.2165,
+      "step": 5950
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0254542827606201,
+      "learning_rate": 9.281192275665836e-05,
+      "loss": 3.2474,
+      "step": 5955
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.01493239402771,
+      "learning_rate": 9.279994431663212e-05,
+      "loss": 2.9983,
+      "step": 5960
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9035430550575256,
+      "learning_rate": 9.278795667860167e-05,
+      "loss": 3.1645,
+      "step": 5965
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.907461941242218,
+      "learning_rate": 9.277595984514325e-05,
+      "loss": 3.1357,
+      "step": 5970
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9064606428146362,
+      "learning_rate": 9.276395381883502e-05,
+      "loss": 3.2313,
+      "step": 5975
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9427024722099304,
+      "learning_rate": 9.275193860225722e-05,
+      "loss": 3.1624,
+      "step": 5980
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9494276642799377,
+      "learning_rate": 9.273991419799193e-05,
+      "loss": 3.2297,
+      "step": 5985
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9668130874633789,
+      "learning_rate": 9.272788060862334e-05,
+      "loss": 3.1949,
+      "step": 5990
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9865812659263611,
+      "learning_rate": 9.271583783673751e-05,
+      "loss": 3.2279,
+      "step": 5995
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0771833658218384,
+      "learning_rate": 9.270378588492255e-05,
+      "loss": 3.171,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 33984,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.7753342881038336e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}