diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5448 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 3848,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0005197505197505198,
+      "grad_norm": 37.05835422762717,
+      "learning_rate": 5.194805194805195e-08,
+      "loss": 2.0466,
+      "step": 1
+    },
+    {
+      "epoch": 0.002598752598752599,
+      "grad_norm": 37.36458385503662,
+      "learning_rate": 2.597402597402598e-07,
+      "loss": 2.0439,
+      "step": 5
+    },
+    {
+      "epoch": 0.005197505197505198,
+      "grad_norm": 39.070563651666575,
+      "learning_rate": 5.194805194805196e-07,
+      "loss": 2.0133,
+      "step": 10
+    },
+    {
+      "epoch": 0.007796257796257797,
+      "grad_norm": 11.361027583243034,
+      "learning_rate": 7.792207792207792e-07,
+      "loss": 1.8666,
+      "step": 15
+    },
+    {
+      "epoch": 0.010395010395010396,
+      "grad_norm": 5.863826986672296,
+      "learning_rate": 1.0389610389610392e-06,
+      "loss": 1.7677,
+      "step": 20
+    },
+    {
+      "epoch": 0.012993762993762994,
+      "grad_norm": 4.522921155466077,
+      "learning_rate": 1.2987012987012986e-06,
+      "loss": 1.6467,
+      "step": 25
+    },
+    {
+      "epoch": 0.015592515592515593,
+      "grad_norm": 3.1296619021160326,
+      "learning_rate": 1.5584415584415584e-06,
+      "loss": 1.4777,
+      "step": 30
+    },
+    {
+      "epoch": 0.018191268191268192,
+      "grad_norm": 2.12702049381069,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.3423,
+      "step": 35
+    },
+    {
+      "epoch": 0.02079002079002079,
+      "grad_norm": 2.381326820034051,
+      "learning_rate": 2.0779220779220784e-06,
+      "loss": 1.2129,
+      "step": 40
+    },
+    {
+      "epoch": 0.02338877338877339,
+      "grad_norm": 1.745648282948505,
+      "learning_rate": 2.337662337662338e-06,
+      "loss": 1.117,
+      "step": 45
+    },
+    {
+      "epoch": 0.02598752598752599,
+      "grad_norm": 1.2151402583346058,
+      "learning_rate": 2.597402597402597e-06,
+      "loss": 1.0561,
+      "step": 50
+    },
+    {
+      "epoch": 0.028586278586278588,
+      "grad_norm": 1.1469155851046635,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.9923,
+      "step": 55
+    },
+    {
+      "epoch": 0.031185031185031187,
+      "grad_norm": 1.1501290412070948,
+      "learning_rate": 3.116883116883117e-06,
+      "loss": 0.9596,
+      "step": 60
+    },
+    {
+      "epoch": 0.033783783783783786,
+      "grad_norm": 1.2437833266166898,
+      "learning_rate": 3.376623376623377e-06,
+      "loss": 0.9281,
+      "step": 65
+    },
+    {
+      "epoch": 0.036382536382536385,
+      "grad_norm": 1.1628173200126817,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.9026,
+      "step": 70
+    },
+    {
+      "epoch": 0.03898128898128898,
+      "grad_norm": 1.1632550654052534,
+      "learning_rate": 3.896103896103897e-06,
+      "loss": 0.9004,
+      "step": 75
+    },
+    {
+      "epoch": 0.04158004158004158,
+      "grad_norm": 1.2551305602329859,
+      "learning_rate": 4.155844155844157e-06,
+      "loss": 0.8755,
+      "step": 80
+    },
+    {
+      "epoch": 0.04417879417879418,
+      "grad_norm": 1.1474998623142625,
+      "learning_rate": 4.415584415584416e-06,
+      "loss": 0.8516,
+      "step": 85
+    },
+    {
+      "epoch": 0.04677754677754678,
+      "grad_norm": 1.1143075622189529,
+      "learning_rate": 4.675324675324676e-06,
+      "loss": 0.8517,
+      "step": 90
+    },
+    {
+      "epoch": 0.04937629937629938,
+      "grad_norm": 1.1206710563724116,
+      "learning_rate": 4.935064935064935e-06,
+      "loss": 0.8458,
+      "step": 95
+    },
+    {
+      "epoch": 0.05197505197505198,
+      "grad_norm": 1.0838734153124963,
+      "learning_rate": 5.194805194805194e-06,
+      "loss": 0.8363,
+      "step": 100
+    },
+    {
+      "epoch": 0.05457380457380458,
+      "grad_norm": 1.5339125140531156,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.8195,
+      "step": 105
+    },
+    {
+      "epoch": 0.057172557172557176,
+      "grad_norm": 1.1439071531812668,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.8394,
+      "step": 110
+    },
+    {
+      "epoch": 0.059771309771309775,
+      "grad_norm": 1.4061392934636887,
+      "learning_rate": 5.9740259740259746e-06,
+      "loss": 0.8239,
+      "step": 115
+    },
+    {
+      "epoch": 0.062370062370062374,
+      "grad_norm": 1.1592154562156236,
+      "learning_rate": 6.233766233766234e-06,
+      "loss": 0.8039,
+      "step": 120
+    },
+    {
+      "epoch": 0.06496881496881497,
+      "grad_norm": 1.0248228760326474,
+      "learning_rate": 6.493506493506494e-06,
+      "loss": 0.8022,
+      "step": 125
+    },
+    {
+      "epoch": 0.06756756756756757,
+      "grad_norm": 1.1650242675138267,
+      "learning_rate": 6.753246753246754e-06,
+      "loss": 0.8032,
+      "step": 130
+    },
+    {
+      "epoch": 0.07016632016632017,
+      "grad_norm": 1.1631059955651202,
+      "learning_rate": 7.012987012987014e-06,
+      "loss": 0.8076,
+      "step": 135
+    },
+    {
+      "epoch": 0.07276507276507277,
+      "grad_norm": 0.9623641373070422,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.8039,
+      "step": 140
+    },
+    {
+      "epoch": 0.07536382536382537,
+      "grad_norm": 1.1042092302715827,
+      "learning_rate": 7.532467532467533e-06,
+      "loss": 0.8041,
+      "step": 145
+    },
+    {
+      "epoch": 0.07796257796257797,
+      "grad_norm": 1.034610728564867,
+      "learning_rate": 7.792207792207793e-06,
+      "loss": 0.7979,
+      "step": 150
+    },
+    {
+      "epoch": 0.08056133056133057,
+      "grad_norm": 1.0441346842973676,
+      "learning_rate": 8.051948051948052e-06,
+      "loss": 0.7889,
+      "step": 155
+    },
+    {
+      "epoch": 0.08316008316008316,
+      "grad_norm": 1.1314811400194547,
+      "learning_rate": 8.311688311688313e-06,
+      "loss": 0.8095,
+      "step": 160
+    },
+    {
+      "epoch": 0.08575883575883576,
+      "grad_norm": 1.083993546178677,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.7838,
+      "step": 165
+    },
+    {
+      "epoch": 0.08835758835758836,
+      "grad_norm": 0.9844284926961596,
+      "learning_rate": 8.831168831168832e-06,
+      "loss": 0.8007,
+      "step": 170
+    },
+    {
+      "epoch": 0.09095634095634096,
+      "grad_norm": 0.9748756413462465,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.7839,
+      "step": 175
+    },
+    {
+      "epoch": 0.09355509355509356,
+      "grad_norm": 1.054078525965627,
+      "learning_rate": 9.350649350649352e-06,
+      "loss": 0.7819,
+      "step": 180
+    },
+    {
+      "epoch": 0.09615384615384616,
+      "grad_norm": 1.1412040734751607,
+      "learning_rate": 9.610389610389611e-06,
+      "loss": 0.7874,
+      "step": 185
+    },
+    {
+      "epoch": 0.09875259875259876,
+      "grad_norm": 1.163328897434868,
+      "learning_rate": 9.87012987012987e-06,
+      "loss": 0.7762,
+      "step": 190
+    },
+    {
+      "epoch": 0.10135135135135136,
+      "grad_norm": 1.0536579529578631,
+      "learning_rate": 1.012987012987013e-05,
+      "loss": 0.7665,
+      "step": 195
+    },
+    {
+      "epoch": 0.10395010395010396,
+      "grad_norm": 1.048914553080904,
+      "learning_rate": 1.0389610389610389e-05,
+      "loss": 0.7849,
+      "step": 200
+    },
+    {
+      "epoch": 0.10654885654885655,
+      "grad_norm": 0.9748667183768033,
+      "learning_rate": 1.064935064935065e-05,
+      "loss": 0.7839,
+      "step": 205
+    },
+    {
+      "epoch": 0.10914760914760915,
+      "grad_norm": 0.9705690959742787,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.7827,
+      "step": 210
+    },
+    {
+      "epoch": 0.11174636174636175,
+      "grad_norm": 0.9527544873685447,
+      "learning_rate": 1.116883116883117e-05,
+      "loss": 0.7823,
+      "step": 215
+    },
+    {
+      "epoch": 0.11434511434511435,
+      "grad_norm": 1.1194233425550948,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.7795,
+      "step": 220
+    },
+    {
+      "epoch": 0.11694386694386695,
+      "grad_norm": 2.120471539383005,
+      "learning_rate": 1.1688311688311688e-05,
+      "loss": 0.7778,
+      "step": 225
+    },
+    {
+      "epoch": 0.11954261954261955,
+      "grad_norm": 0.9808803038153814,
+      "learning_rate": 1.1948051948051949e-05,
+      "loss": 0.7634,
+      "step": 230
+    },
+    {
+      "epoch": 0.12214137214137215,
+      "grad_norm": 0.94060918910957,
+      "learning_rate": 1.2207792207792208e-05,
+      "loss": 0.7578,
+      "step": 235
+    },
+    {
+      "epoch": 0.12474012474012475,
+      "grad_norm": 0.9878811168887828,
+      "learning_rate": 1.2467532467532468e-05,
+      "loss": 0.7837,
+      "step": 240
+    },
+    {
+      "epoch": 0.12733887733887733,
+      "grad_norm": 1.0526434374561247,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 0.777,
+      "step": 245
+    },
+    {
+      "epoch": 0.12993762993762994,
+      "grad_norm": 1.067255251886244,
+      "learning_rate": 1.2987012987012988e-05,
+      "loss": 0.7606,
+      "step": 250
+    },
+    {
+      "epoch": 0.13253638253638253,
+      "grad_norm": 1.019011419752788,
+      "learning_rate": 1.3246753246753249e-05,
+      "loss": 0.7783,
+      "step": 255
+    },
+    {
+      "epoch": 0.13513513513513514,
+      "grad_norm": 0.9625368636583931,
+      "learning_rate": 1.3506493506493508e-05,
+      "loss": 0.7643,
+      "step": 260
+    },
+    {
+      "epoch": 0.13773388773388773,
+      "grad_norm": 0.8964922122751663,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.769,
+      "step": 265
+    },
+    {
+      "epoch": 0.14033264033264034,
+      "grad_norm": 1.0861405810538973,
+      "learning_rate": 1.4025974025974028e-05,
+      "loss": 0.772,
+      "step": 270
+    },
+    {
+      "epoch": 0.14293139293139293,
+      "grad_norm": 0.7725004042430159,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.7684,
+      "step": 275
+    },
+    {
+      "epoch": 0.14553014553014554,
+      "grad_norm": 0.9641597887627689,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.7665,
+      "step": 280
+    },
+    {
+      "epoch": 0.14812889812889812,
+      "grad_norm": 0.8799231599980127,
+      "learning_rate": 1.4805194805194807e-05,
+      "loss": 0.7553,
+      "step": 285
+    },
+    {
+      "epoch": 0.15072765072765074,
+      "grad_norm": 0.865183845501612,
+      "learning_rate": 1.5064935064935066e-05,
+      "loss": 0.776,
+      "step": 290
+    },
+    {
+      "epoch": 0.15332640332640332,
+      "grad_norm": 0.9280023423503597,
+      "learning_rate": 1.5324675324675326e-05,
+      "loss": 0.7655,
+      "step": 295
+    },
+    {
+      "epoch": 0.15592515592515593,
+      "grad_norm": 0.9412086108813551,
+      "learning_rate": 1.5584415584415587e-05,
+      "loss": 0.7541,
+      "step": 300
+    },
+    {
+      "epoch": 0.15852390852390852,
+      "grad_norm": 1.049389569363405,
+      "learning_rate": 1.5844155844155847e-05,
+      "loss": 0.7609,
+      "step": 305
+    },
+    {
+      "epoch": 0.16112266112266113,
+      "grad_norm": 0.8632998959240495,
+      "learning_rate": 1.6103896103896105e-05,
+      "loss": 0.7572,
+      "step": 310
+    },
+    {
+      "epoch": 0.16372141372141372,
+      "grad_norm": 0.926940763705136,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.763,
+      "step": 315
+    },
+    {
+      "epoch": 0.16632016632016633,
+      "grad_norm": 0.8169022540417774,
+      "learning_rate": 1.6623376623376627e-05,
+      "loss": 0.7564,
+      "step": 320
+    },
+    {
+      "epoch": 0.16891891891891891,
+      "grad_norm": 0.8913503941645097,
+      "learning_rate": 1.6883116883116884e-05,
+      "loss": 0.768,
+      "step": 325
+    },
+    {
+      "epoch": 0.17151767151767153,
+      "grad_norm": 0.9677009736871893,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.7577,
+      "step": 330
+    },
+    {
+      "epoch": 0.1741164241164241,
+      "grad_norm": 0.891596010196128,
+      "learning_rate": 1.7402597402597403e-05,
+      "loss": 0.7482,
+      "step": 335
+    },
+    {
+      "epoch": 0.17671517671517672,
+      "grad_norm": 0.922259390367936,
+      "learning_rate": 1.7662337662337664e-05,
+      "loss": 0.7545,
+      "step": 340
+    },
+    {
+      "epoch": 0.1793139293139293,
+      "grad_norm": 0.8012139878255935,
+      "learning_rate": 1.792207792207792e-05,
+      "loss": 0.748,
+      "step": 345
+    },
+    {
+      "epoch": 0.18191268191268192,
+      "grad_norm": 0.8885070497602618,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.7506,
+      "step": 350
+    },
+    {
+      "epoch": 0.1845114345114345,
+      "grad_norm": 0.8139071250420767,
+      "learning_rate": 1.8441558441558443e-05,
+      "loss": 0.7657,
+      "step": 355
+    },
+    {
+      "epoch": 0.18711018711018712,
+      "grad_norm": 0.8597830390928304,
+      "learning_rate": 1.8701298701298704e-05,
+      "loss": 0.7482,
+      "step": 360
+    },
+    {
+      "epoch": 0.1897089397089397,
+      "grad_norm": 0.8215436811633837,
+      "learning_rate": 1.896103896103896e-05,
+      "loss": 0.7563,
+      "step": 365
+    },
+    {
+      "epoch": 0.19230769230769232,
+      "grad_norm": 0.7888662364197664,
+      "learning_rate": 1.9220779220779222e-05,
+      "loss": 0.7638,
+      "step": 370
+    },
+    {
+      "epoch": 0.1949064449064449,
+      "grad_norm": 0.7758295937643592,
+      "learning_rate": 1.9480519480519483e-05,
+      "loss": 0.7578,
+      "step": 375
+    },
+    {
+      "epoch": 0.19750519750519752,
+      "grad_norm": 0.7321462182850453,
+      "learning_rate": 1.974025974025974e-05,
+      "loss": 0.7478,
+      "step": 380
+    },
+    {
+      "epoch": 0.2001039501039501,
+      "grad_norm": 0.8972812591629451,
+      "learning_rate": 2e-05,
+      "loss": 0.7364,
+      "step": 385
+    },
+    {
+      "epoch": 0.20270270270270271,
+      "grad_norm": 0.8504197871382875,
+      "learning_rate": 1.9999897126378044e-05,
+      "loss": 0.7531,
+      "step": 390
+    },
+    {
+      "epoch": 0.2053014553014553,
+      "grad_norm": 1.0008077674950657,
+      "learning_rate": 1.9999588507628768e-05,
+      "loss": 0.7555,
+      "step": 395
+    },
+    {
+      "epoch": 0.2079002079002079,
+      "grad_norm": 0.8219960180575271,
+      "learning_rate": 1.999907415010192e-05,
+      "loss": 0.7412,
+      "step": 400
+    },
+    {
+      "epoch": 0.2104989604989605,
+      "grad_norm": 0.7975912262836815,
+      "learning_rate": 1.9998354064380263e-05,
+      "loss": 0.7342,
+      "step": 405
+    },
+    {
+      "epoch": 0.2130977130977131,
+      "grad_norm": 0.7386294167625675,
+      "learning_rate": 1.9997428265279365e-05,
+      "loss": 0.7414,
+      "step": 410
+    },
+    {
+      "epoch": 0.2156964656964657,
+      "grad_norm": 0.7996558972435127,
+      "learning_rate": 1.999629677184728e-05,
+      "loss": 0.7684,
+      "step": 415
+    },
+    {
+      "epoch": 0.2182952182952183,
+      "grad_norm": 0.8532223218359839,
+      "learning_rate": 1.999495960736418e-05,
+      "loss": 0.7318,
+      "step": 420
+    },
+    {
+      "epoch": 0.2208939708939709,
+      "grad_norm": 0.8099184350785439,
+      "learning_rate": 1.999341679934186e-05,
+      "loss": 0.7483,
+      "step": 425
+    },
+    {
+      "epoch": 0.2234927234927235,
+      "grad_norm": 0.8032456938404925,
+      "learning_rate": 1.999166837952316e-05,
+      "loss": 0.7464,
+      "step": 430
+    },
+    {
+      "epoch": 0.2260914760914761,
+      "grad_norm": 0.7613353132794639,
+      "learning_rate": 1.998971438388134e-05,
+      "loss": 0.7241,
+      "step": 435
+    },
+    {
+      "epoch": 0.2286902286902287,
+      "grad_norm": 0.739838242733476,
+      "learning_rate": 1.9987554852619325e-05,
+      "loss": 0.7523,
+      "step": 440
+    },
+    {
+      "epoch": 0.2312889812889813,
+      "grad_norm": 0.7456120762278995,
+      "learning_rate": 1.998518983016887e-05,
+      "loss": 0.7461,
+      "step": 445
+    },
+    {
+      "epoch": 0.2338877338877339,
+      "grad_norm": 0.8064413740728865,
+      "learning_rate": 1.9982619365189662e-05,
+      "loss": 0.7524,
+      "step": 450
+    },
+    {
+      "epoch": 0.23648648648648649,
+      "grad_norm": 0.8264553841158012,
+      "learning_rate": 1.9979843510568312e-05,
+      "loss": 0.7569,
+      "step": 455
+    },
+    {
+      "epoch": 0.2390852390852391,
+      "grad_norm": 0.7700326026252647,
+      "learning_rate": 1.9976862323417262e-05,
+      "loss": 0.7421,
+      "step": 460
+    },
+    {
+      "epoch": 0.24168399168399168,
+      "grad_norm": 0.8078133521219729,
+      "learning_rate": 1.9973675865073614e-05,
+      "loss": 0.7459,
+      "step": 465
+    },
+    {
+      "epoch": 0.2442827442827443,
+      "grad_norm": 0.8834562269681555,
+      "learning_rate": 1.9970284201097874e-05,
+      "loss": 0.7419,
+      "step": 470
+    },
+    {
+      "epoch": 0.24688149688149688,
+      "grad_norm": 0.8201197358624366,
+      "learning_rate": 1.996668740127259e-05,
+      "loss": 0.7264,
+      "step": 475
+    },
+    {
+      "epoch": 0.2494802494802495,
+      "grad_norm": 0.7832673791812143,
+      "learning_rate": 1.996288553960093e-05,
+      "loss": 0.7326,
+      "step": 480
+    },
+    {
+      "epoch": 0.2520790020790021,
+      "grad_norm": 0.7491102594223664,
+      "learning_rate": 1.9958878694305147e-05,
+      "loss": 0.7423,
+      "step": 485
+    },
+    {
+      "epoch": 0.25467775467775466,
+      "grad_norm": 0.7711246686689462,
+      "learning_rate": 1.9954666947824983e-05,
+      "loss": 0.7601,
+      "step": 490
+    },
+    {
+      "epoch": 0.25727650727650725,
+      "grad_norm": 0.8431355399537503,
+      "learning_rate": 1.9950250386815953e-05,
+      "loss": 0.7508,
+      "step": 495
+    },
+    {
+      "epoch": 0.2598752598752599,
+      "grad_norm": 0.7491422333746851,
+      "learning_rate": 1.9945629102147593e-05,
+      "loss": 0.7365,
+      "step": 500
+    },
+    {
+      "epoch": 0.2624740124740125,
+      "grad_norm": 0.6891526342744494,
+      "learning_rate": 1.9940803188901556e-05,
+      "loss": 0.7315,
+      "step": 505
+    },
+    {
+      "epoch": 0.26507276507276506,
+      "grad_norm": 0.7000282425722503,
+      "learning_rate": 1.9935772746369678e-05,
+      "loss": 0.7524,
+      "step": 510
+    },
+    {
+      "epoch": 0.26767151767151764,
+      "grad_norm": 0.6734603177161373,
+      "learning_rate": 1.9930537878051927e-05,
+      "loss": 0.746,
+      "step": 515
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 0.6632065052188117,
+      "learning_rate": 1.9925098691654275e-05,
+      "loss": 0.7225,
+      "step": 520
+    },
+    {
+      "epoch": 0.27286902286902287,
+      "grad_norm": 0.6985881279460795,
+      "learning_rate": 1.9919455299086485e-05,
+      "loss": 0.7451,
+      "step": 525
+    },
+    {
+      "epoch": 0.27546777546777546,
+      "grad_norm": 0.6540248071379781,
+      "learning_rate": 1.99136078164598e-05,
+      "loss": 0.7322,
+      "step": 530
+    },
+    {
+      "epoch": 0.27806652806652804,
+      "grad_norm": 0.661603660243552,
+      "learning_rate": 1.9907556364084568e-05,
+      "loss": 0.7544,
+      "step": 535
+    },
+    {
+      "epoch": 0.2806652806652807,
+      "grad_norm": 0.7446420340679871,
+      "learning_rate": 1.990130106646775e-05,
+      "loss": 0.7476,
+      "step": 540
+    },
+    {
+      "epoch": 0.28326403326403327,
+      "grad_norm": 0.8425291153171641,
+      "learning_rate": 1.9894842052310373e-05,
+      "loss": 0.733,
+      "step": 545
+    },
+    {
+      "epoch": 0.28586278586278585,
+      "grad_norm": 0.6481748553031429,
+      "learning_rate": 1.9888179454504875e-05,
+      "loss": 0.7439,
+      "step": 550
+    },
+    {
+      "epoch": 0.28846153846153844,
+      "grad_norm": 0.679678646955314,
+      "learning_rate": 1.9881313410132365e-05,
+      "loss": 0.7418,
+      "step": 555
+    },
+    {
+      "epoch": 0.2910602910602911,
+      "grad_norm": 0.6504991149281961,
+      "learning_rate": 1.9874244060459816e-05,
+      "loss": 0.7338,
+      "step": 560
+    },
+    {
+      "epoch": 0.29365904365904366,
+      "grad_norm": 0.7240524101878886,
+      "learning_rate": 1.986697155093715e-05,
+      "loss": 0.7384,
+      "step": 565
+    },
+    {
+      "epoch": 0.29625779625779625,
+      "grad_norm": 0.677804845061605,
+      "learning_rate": 1.9859496031194242e-05,
+      "loss": 0.7368,
+      "step": 570
+    },
+    {
+      "epoch": 0.29885654885654883,
+      "grad_norm": 0.6836994514744595,
+      "learning_rate": 1.9851817655037854e-05,
+      "loss": 0.7361,
+      "step": 575
+    },
+    {
+      "epoch": 0.30145530145530147,
+      "grad_norm": 0.6837922941788586,
+      "learning_rate": 1.9843936580448457e-05,
+      "loss": 0.7243,
+      "step": 580
+    },
+    {
+      "epoch": 0.30405405405405406,
+      "grad_norm": 0.7000337048592958,
+      "learning_rate": 1.983585296957699e-05,
+      "loss": 0.7329,
+      "step": 585
+    },
+    {
+      "epoch": 0.30665280665280664,
+      "grad_norm": 0.7178435855481708,
+      "learning_rate": 1.9827566988741525e-05,
+      "loss": 0.719,
+      "step": 590
+    },
+    {
+      "epoch": 0.3092515592515592,
+      "grad_norm": 0.7066252470023224,
+      "learning_rate": 1.9819078808423825e-05,
+      "loss": 0.7277,
+      "step": 595
+    },
+    {
+      "epoch": 0.31185031185031187,
+      "grad_norm": 0.7131086872189417,
+      "learning_rate": 1.981038860326586e-05,
+      "loss": 0.7408,
+      "step": 600
+    },
+    {
+      "epoch": 0.31444906444906445,
+      "grad_norm": 0.6862600699294851,
+      "learning_rate": 1.980149655206621e-05,
+      "loss": 0.7338,
+      "step": 605
+    },
+    {
+      "epoch": 0.31704781704781704,
+      "grad_norm": 0.7265535072671346,
+      "learning_rate": 1.9792402837776377e-05,
+      "loss": 0.7209,
+      "step": 610
+    },
+    {
+      "epoch": 0.3196465696465696,
+      "grad_norm": 0.7076896535093625,
+      "learning_rate": 1.978310764749703e-05,
+      "loss": 0.7442,
+      "step": 615
+    },
+    {
+      "epoch": 0.32224532224532226,
+      "grad_norm": 0.6473108719176737,
+      "learning_rate": 1.9773611172474143e-05,
+      "loss": 0.7314,
+      "step": 620
+    },
+    {
+      "epoch": 0.32484407484407485,
+      "grad_norm": 0.6958993345177985,
+      "learning_rate": 1.976391360809507e-05,
+      "loss": 0.7297,
+      "step": 625
+    },
+    {
+      "epoch": 0.32744282744282743,
+      "grad_norm": 0.6980561880943195,
+      "learning_rate": 1.9754015153884533e-05,
+      "loss": 0.7507,
+      "step": 630
+    },
+    {
+      "epoch": 0.33004158004158,
+      "grad_norm": 0.725526089648866,
+      "learning_rate": 1.974391601350049e-05,
+      "loss": 0.7316,
+      "step": 635
+    },
+    {
+      "epoch": 0.33264033264033266,
+      "grad_norm": 0.6555475191719049,
+      "learning_rate": 1.9733616394729975e-05,
+      "loss": 0.7415,
+      "step": 640
+    },
+    {
+      "epoch": 0.33523908523908524,
+      "grad_norm": 0.7126270164163553,
+      "learning_rate": 1.9723116509484807e-05,
+      "loss": 0.7084,
+      "step": 645
+    },
+    {
+      "epoch": 0.33783783783783783,
+      "grad_norm": 0.6175823835566073,
+      "learning_rate": 1.971241657379723e-05,
+      "loss": 0.7437,
+      "step": 650
+    },
+    {
+      "epoch": 0.3404365904365904,
+      "grad_norm": 0.641705701909066,
+      "learning_rate": 1.9701516807815472e-05,
+      "loss": 0.7227,
+      "step": 655
+    },
+    {
+      "epoch": 0.34303534303534305,
+      "grad_norm": 0.6271473392148214,
+      "learning_rate": 1.9690417435799217e-05,
+      "loss": 0.7131,
+      "step": 660
+    },
+    {
+      "epoch": 0.34563409563409564,
+      "grad_norm": 0.7368709959215286,
+      "learning_rate": 1.967911868611498e-05,
+      "loss": 0.7366,
+      "step": 665
+    },
+    {
+      "epoch": 0.3482328482328482,
+      "grad_norm": 0.645812200849196,
+      "learning_rate": 1.9667620791231422e-05,
+      "loss": 0.7174,
+      "step": 670
+    },
+    {
+      "epoch": 0.3508316008316008,
+      "grad_norm": 0.6331883051927242,
+      "learning_rate": 1.965592398771456e-05,
+      "loss": 0.7255,
+      "step": 675
+    },
+    {
+      "epoch": 0.35343035343035345,
+      "grad_norm": 0.7047104196046573,
+      "learning_rate": 1.9644028516222912e-05,
+      "loss": 0.7349,
+      "step": 680
+    },
+    {
+      "epoch": 0.35602910602910603,
+      "grad_norm": 0.6573670034252512,
+      "learning_rate": 1.9631934621502514e-05,
+      "loss": 0.7294,
+      "step": 685
+    },
+    {
+      "epoch": 0.3586278586278586,
+      "grad_norm": 0.64012763867233,
+      "learning_rate": 1.9619642552381924e-05,
+      "loss": 0.7253,
+      "step": 690
+    },
+    {
+      "epoch": 0.3612266112266112,
+      "grad_norm": 0.7085641762966977,
+      "learning_rate": 1.9607152561767077e-05,
+      "loss": 0.7302,
+      "step": 695
+    },
+    {
+      "epoch": 0.36382536382536385,
+      "grad_norm": 0.6270235430166686,
+      "learning_rate": 1.9594464906636083e-05,
+      "loss": 0.7215,
+      "step": 700
+    },
+    {
+      "epoch": 0.36642411642411643,
+      "grad_norm": 0.6347937985689194,
+      "learning_rate": 1.958157984803395e-05,
+      "loss": 0.7198,
+      "step": 705
+    },
+    {
+      "epoch": 0.369022869022869,
+      "grad_norm": 0.6404396997260027,
+      "learning_rate": 1.956849765106721e-05,
+      "loss": 0.7355,
+      "step": 710
+    },
+    {
+      "epoch": 0.3716216216216216,
+      "grad_norm": 0.5736442010381354,
+      "learning_rate": 1.9555218584898457e-05,
+      "loss": 0.7181,
+      "step": 715
+    },
+    {
+      "epoch": 0.37422037422037424,
+      "grad_norm": 0.6408880853261398,
+      "learning_rate": 1.954174292274082e-05,
+      "loss": 0.713,
+      "step": 720
+    },
+    {
+      "epoch": 0.3768191268191268,
+      "grad_norm": 0.6668059771868257,
+      "learning_rate": 1.9528070941852334e-05,
+      "loss": 0.7477,
+      "step": 725
+    },
+    {
+      "epoch": 0.3794178794178794,
+      "grad_norm": 0.6868239822957647,
+      "learning_rate": 1.9514202923530233e-05,
+      "loss": 0.7209,
+      "step": 730
+    },
+    {
+      "epoch": 0.382016632016632,
+      "grad_norm": 0.6673930879897976,
+      "learning_rate": 1.9500139153105183e-05,
+      "loss": 0.7256,
+      "step": 735
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.6481855409470721,
+      "learning_rate": 1.948587991993537e-05,
+      "loss": 0.6989,
+      "step": 740
+    },
+    {
+      "epoch": 0.3872141372141372,
+      "grad_norm": 0.7003926465709457,
+      "learning_rate": 1.94714255174006e-05,
+      "loss": 0.7079,
+      "step": 745
+    },
+    {
+      "epoch": 0.3898128898128898,
+      "grad_norm": 0.7900259673292569,
+      "learning_rate": 1.945677624289621e-05,
+      "loss": 0.7238,
+      "step": 750
+    },
+    {
+      "epoch": 0.3924116424116424,
+      "grad_norm": 0.6964569074530326,
+      "learning_rate": 1.9441932397826993e-05,
+      "loss": 0.7097,
+      "step": 755
+    },
+    {
+      "epoch": 0.39501039501039503,
+      "grad_norm": 0.6487993082304472,
+      "learning_rate": 1.9426894287600966e-05,
+      "loss": 0.7079,
+      "step": 760
+    },
+    {
+      "epoch": 0.3976091476091476,
+      "grad_norm": 0.688310939038027,
+      "learning_rate": 1.9411662221623103e-05,
+      "loss": 0.7154,
+      "step": 765
+    },
+    {
+      "epoch": 0.4002079002079002,
+      "grad_norm": 0.67435617407303,
+      "learning_rate": 1.939623651328897e-05,
+      "loss": 0.7094,
+      "step": 770
+    },
+    {
+      "epoch": 0.4028066528066528,
+      "grad_norm": 0.6183714426118461,
+      "learning_rate": 1.9380617479978255e-05,
+      "loss": 0.7341,
+      "step": 775
+    },
+    {
+      "epoch": 0.40540540540540543,
+      "grad_norm": 0.6617362465155076,
+      "learning_rate": 1.9364805443048266e-05,
+      "loss": 0.7173,
+      "step": 780
+    },
+    {
+      "epoch": 0.408004158004158,
+      "grad_norm": 0.5678238642277171,
+      "learning_rate": 1.9348800727827307e-05,
+      "loss": 0.7168,
+      "step": 785
+    },
+    {
+      "epoch": 0.4106029106029106,
+      "grad_norm": 0.5910729067701429,
+      "learning_rate": 1.9332603663607983e-05,
+      "loss": 0.712,
+      "step": 790
+    },
+    {
+      "epoch": 0.4132016632016632,
+      "grad_norm": 0.6197819946059432,
+      "learning_rate": 1.9316214583640427e-05,
+      "loss": 0.7247,
+      "step": 795
+    },
+    {
+      "epoch": 0.4158004158004158,
+      "grad_norm": 0.6305700579142718,
+      "learning_rate": 1.929963382512544e-05,
+      "loss": 0.7139,
+      "step": 800
+    },
+    {
+      "epoch": 0.4183991683991684,
+      "grad_norm": 0.6439900936581443,
+      "learning_rate": 1.9282861729207555e-05,
+      "loss": 0.7106,
+      "step": 805
+    },
+    {
+      "epoch": 0.420997920997921,
+      "grad_norm": 0.6990297176777683,
+      "learning_rate": 1.926589864096803e-05,
+      "loss": 0.7234,
+      "step": 810
+    },
+    {
+      "epoch": 0.4235966735966736,
+      "grad_norm": 0.6967169647297736,
+      "learning_rate": 1.9248744909417728e-05,
+      "loss": 0.7178,
+      "step": 815
+    },
+    {
+      "epoch": 0.4261954261954262,
+      "grad_norm": 0.6433170596765804,
+      "learning_rate": 1.923140088748995e-05,
+      "loss": 0.7159,
+      "step": 820
+    },
+    {
+      "epoch": 0.4287941787941788,
+      "grad_norm": 0.5983245551968179,
+      "learning_rate": 1.9213866932033164e-05,
+      "loss": 0.7057,
+      "step": 825
+    },
+    {
+      "epoch": 0.4313929313929314,
+      "grad_norm": 0.6087581373525092,
+      "learning_rate": 1.9196143403803667e-05,
+      "loss": 0.7154,
+      "step": 830
+    },
+    {
+      "epoch": 0.433991683991684,
+      "grad_norm": 0.636489262543138,
+      "learning_rate": 1.9178230667458175e-05,
+      "loss": 0.7057,
+      "step": 835
+    },
+    {
+      "epoch": 0.4365904365904366,
+      "grad_norm": 0.6439601891575886,
+      "learning_rate": 1.91601290915463e-05,
+      "loss": 0.7279,
+      "step": 840
+    },
+    {
+      "epoch": 0.4391891891891892,
+      "grad_norm": 0.6225933290388692,
+      "learning_rate": 1.9141839048502974e-05,
+      "loss": 0.7187,
+      "step": 845
+    },
+    {
+      "epoch": 0.4417879417879418,
+      "grad_norm": 0.6035545819895592,
+      "learning_rate": 1.9123360914640794e-05,
+      "loss": 0.716,
+      "step": 850
+    },
+    {
+      "epoch": 0.44438669438669437,
+      "grad_norm": 0.6607567088647912,
+      "learning_rate": 1.9104695070142273e-05,
+      "loss": 0.7309,
+      "step": 855
+    },
+    {
+      "epoch": 0.446985446985447,
+      "grad_norm": 0.6708322724799578,
+      "learning_rate": 1.9085841899052014e-05,
+      "loss": 0.7042,
+      "step": 860
+    },
+    {
+      "epoch": 0.4495841995841996,
+      "grad_norm": 0.5887928003237825,
+      "learning_rate": 1.9066801789268815e-05,
+      "loss": 0.7227,
+      "step": 865
+    },
+    {
+      "epoch": 0.4521829521829522,
+      "grad_norm": 0.6055286727657766,
+      "learning_rate": 1.9047575132537694e-05,
+      "loss": 0.7092,
+      "step": 870
+    },
+    {
+      "epoch": 0.45478170478170477,
+      "grad_norm": 0.6202428781057725,
+      "learning_rate": 1.902816232444181e-05,
+      "loss": 0.7104,
+      "step": 875
+    },
+    {
+      "epoch": 0.4573804573804574,
+      "grad_norm": 0.6274570988856961,
+      "learning_rate": 1.9008563764394334e-05,
+      "loss": 0.7229,
+      "step": 880
+    },
+    {
+      "epoch": 0.45997920997921,
+      "grad_norm": 0.6605818513974558,
+      "learning_rate": 1.8988779855630246e-05,
+      "loss": 0.7286,
+      "step": 885
+    },
+    {
+      "epoch": 0.4625779625779626,
+      "grad_norm": 0.6410281151386927,
+      "learning_rate": 1.896881100519801e-05,
+      "loss": 0.725,
+      "step": 890
+    },
+    {
+      "epoch": 0.46517671517671516,
+      "grad_norm": 0.5697116895525344,
+      "learning_rate": 1.8948657623951224e-05,
+      "loss": 0.7012,
+      "step": 895
+    },
+    {
+      "epoch": 0.4677754677754678,
+      "grad_norm": 0.6221008774646193,
+      "learning_rate": 1.8928320126540154e-05,
+      "loss": 0.7088,
+      "step": 900
+    },
+    {
+      "epoch": 0.4703742203742204,
+      "grad_norm": 0.6783327380563849,
+      "learning_rate": 1.89077989314032e-05,
+      "loss": 0.7012,
+      "step": 905
+    },
+    {
+      "epoch": 0.47297297297297297,
+      "grad_norm": 0.6364455758589938,
+      "learning_rate": 1.8887094460758298e-05,
+      "loss": 0.7173,
+      "step": 910
+    },
+    {
+      "epoch": 0.47557172557172556,
+      "grad_norm": 0.6423883619688115,
+      "learning_rate": 1.8866207140594223e-05,
+      "loss": 0.7221,
+      "step": 915
+    },
+    {
+      "epoch": 0.4781704781704782,
+      "grad_norm": 0.6078098531540564,
+      "learning_rate": 1.8845137400661832e-05,
+      "loss": 0.7112,
+      "step": 920
+    },
+    {
+      "epoch": 0.4807692307692308,
+      "grad_norm": 0.6855240964444005,
+      "learning_rate": 1.882388567446522e-05,
+      "loss": 0.701,
+      "step": 925
+    },
+    {
+      "epoch": 0.48336798336798337,
+      "grad_norm": 0.6214299025960558,
+      "learning_rate": 1.880245239925279e-05,
+      "loss": 0.7152,
+      "step": 930
+    },
+    {
+      "epoch": 0.48596673596673595,
+      "grad_norm": 0.7011949443886023,
+      "learning_rate": 1.878083801600828e-05,
+      "loss": 0.7099,
+      "step": 935
+    },
+    {
+      "epoch": 0.4885654885654886,
+      "grad_norm": 0.6305195514031546,
+      "learning_rate": 1.8759042969441666e-05,
+      "loss": 0.7119,
+      "step": 940
+    },
+    {
+      "epoch": 0.4911642411642412,
+      "grad_norm": 0.6284331380737156,
+      "learning_rate": 1.8737067707980018e-05,
+      "loss": 0.7283,
+      "step": 945
+    },
+    {
+      "epoch": 0.49376299376299376,
+      "grad_norm": 0.6834839648993508,
+      "learning_rate": 1.8714912683758292e-05,
+      "loss": 0.7224,
+      "step": 950
+    },
+    {
+      "epoch": 0.49636174636174635,
+      "grad_norm": 0.5860761907825943,
+      "learning_rate": 1.869257835261e-05,
+      "loss": 0.7006,
+      "step": 955
+    },
+    {
+      "epoch": 0.498960498960499,
+      "grad_norm": 0.6589606191667001,
+      "learning_rate": 1.8670065174057854e-05,
+      "loss": 0.7179,
+      "step": 960
+    },
+    {
+      "epoch": 0.5015592515592515,
+      "grad_norm": 0.6369606574521753,
+      "learning_rate": 1.8647373611304295e-05,
+      "loss": 0.715,
+      "step": 965
+    },
+    {
+      "epoch": 0.5041580041580042,
+      "grad_norm": 0.604142259445137,
+      "learning_rate": 1.862450413122197e-05,
+      "loss": 0.7014,
+      "step": 970
+    },
+    {
+      "epoch": 0.5067567567567568,
+      "grad_norm": 0.6290172817388325,
+      "learning_rate": 1.8601457204344134e-05,
+      "loss": 0.71,
+      "step": 975
+    },
+    {
+      "epoch": 0.5093555093555093,
+      "grad_norm": 0.5861860996181917,
+      "learning_rate": 1.8578233304854952e-05,
+      "loss": 0.7041,
+      "step": 980
+    },
+    {
+      "epoch": 0.511954261954262,
+      "grad_norm": 0.6582907547385,
+      "learning_rate": 1.855483291057976e-05,
+      "loss": 0.7259,
+      "step": 985
+    },
+    {
+      "epoch": 0.5145530145530145,
+      "grad_norm": 0.6056438043018836,
+      "learning_rate": 1.8531256502975218e-05,
+      "loss": 0.7141,
+      "step": 990
+    },
+    {
+      "epoch": 0.5171517671517671,
+      "grad_norm": 0.6299326122139869,
+      "learning_rate": 1.850750456711941e-05,
+      "loss": 0.7037,
+      "step": 995
+    },
+    {
+      "epoch": 0.5197505197505198,
+      "grad_norm": 0.5922842182974687,
+      "learning_rate": 1.8483577591701876e-05,
+      "loss": 0.7227,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5223492723492723,
+      "grad_norm": 0.6335533442723702,
+      "learning_rate": 1.8459476069013537e-05,
+      "loss": 0.7331,
+      "step": 1005
+    },
+    {
+      "epoch": 0.524948024948025,
+      "grad_norm": 0.603529261601813,
+      "learning_rate": 1.8435200494936583e-05,
+      "loss": 0.7204,
+      "step": 1010
+    },
+    {
+      "epoch": 0.5275467775467776,
+      "grad_norm": 0.5784152517101885,
+      "learning_rate": 1.841075136893426e-05,
+      "loss": 0.7045,
+      "step": 1015
+    },
+    {
+      "epoch": 0.5301455301455301,
+      "grad_norm": 0.5775827914904406,
+      "learning_rate": 1.8386129194040597e-05,
+      "loss": 0.71,
+      "step": 1020
+    },
+    {
+      "epoch": 0.5327442827442828,
+      "grad_norm": 0.6037013614687209,
+      "learning_rate": 1.8361334476850052e-05,
+      "loss": 0.6973,
+      "step": 1025
+    },
+    {
+      "epoch": 0.5353430353430353,
+      "grad_norm": 0.5969176345454209,
+      "learning_rate": 1.8336367727507104e-05,
+      "loss": 0.7102,
+      "step": 1030
+    },
+    {
+      "epoch": 0.5379417879417879,
+      "grad_norm": 0.5880052937664013,
+      "learning_rate": 1.8311229459695735e-05,
+      "loss": 0.702,
+      "step": 1035
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 0.5665171424206621,
+      "learning_rate": 1.828592019062888e-05,
+      "loss": 0.7059,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5431392931392931,
+      "grad_norm": 0.5776240714899302,
+      "learning_rate": 1.826044044103777e-05,
+      "loss": 0.7124,
+      "step": 1045
+    },
+    {
+      "epoch": 0.5457380457380457,
+      "grad_norm": 0.5961454904432945,
+      "learning_rate": 1.8234790735161233e-05,
+      "loss": 0.7003,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5483367983367984,
+      "grad_norm": 0.61619821889063,
+      "learning_rate": 1.82089716007349e-05,
+      "loss": 0.7112,
+      "step": 1055
+    },
+    {
+      "epoch": 0.5509355509355509,
+      "grad_norm": 0.5663452621669731,
+      "learning_rate": 1.8182983568980347e-05,
+      "loss": 0.7089,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5535343035343036,
+      "grad_norm": 0.5804355767396245,
+      "learning_rate": 1.8156827174594157e-05,
+      "loss": 0.6992,
+      "step": 1065
+    },
+    {
+      "epoch": 0.5561330561330561,
+      "grad_norm": 0.6266224913727885,
+      "learning_rate": 1.8130502955736945e-05,
+      "loss": 0.7054,
+      "step": 1070
+    },
+    {
+      "epoch": 0.5587318087318087,
+      "grad_norm": 0.5683126721009722,
+      "learning_rate": 1.810401145402225e-05,
+      "loss": 0.7132,
+      "step": 1075
+    },
+    {
+      "epoch": 0.5613305613305614,
+      "grad_norm": 0.6083470685034659,
+      "learning_rate": 1.8077353214505428e-05,
+      "loss": 0.7058,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5639293139293139,
+      "grad_norm": 0.6109321123987571,
+      "learning_rate": 1.8050528785672403e-05,
+      "loss": 0.7096,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5665280665280665,
+      "grad_norm": 0.5553746362917324,
+      "learning_rate": 1.8023538719428405e-05,
+      "loss": 0.7009,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5691268191268192,
+      "grad_norm": 0.6062718001611305,
+      "learning_rate": 1.7996383571086614e-05,
+      "loss": 0.7098,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5717255717255717,
+      "grad_norm": 0.5458923136243866,
+      "learning_rate": 1.796906389935672e-05,
+      "loss": 0.6999,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5743243243243243,
+      "grad_norm": 0.5655005675331972,
+      "learning_rate": 1.7941580266333433e-05,
+      "loss": 0.7072,
+      "step": 1105
+    },
+    {
+      "epoch": 0.5769230769230769,
+      "grad_norm": 0.5924848136830811,
+      "learning_rate": 1.7913933237484936e-05,
+      "loss": 0.7096,
+      "step": 1110
+    },
+    {
+      "epoch": 0.5795218295218295,
+      "grad_norm": 0.5639064277581654,
+      "learning_rate": 1.7886123381641226e-05,
+      "loss": 0.7238,
+      "step": 1115
+    },
+    {
+      "epoch": 0.5821205821205822,
+      "grad_norm": 0.6027713277537942,
+      "learning_rate": 1.7858151270982422e-05,
+      "loss": 0.7053,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5847193347193347,
+      "grad_norm": 0.5726133330263367,
+      "learning_rate": 1.7830017481026992e-05,
+      "loss": 0.6965,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5873180873180873,
+      "grad_norm": 0.5747640646170223,
+      "learning_rate": 1.7801722590619905e-05,
+      "loss": 0.6996,
+      "step": 1130
+    },
+    {
+      "epoch": 0.58991683991684,
+      "grad_norm": 0.5720989361581698,
+      "learning_rate": 1.777326718192074e-05,
+      "loss": 0.7132,
+      "step": 1135
+    },
+    {
+      "epoch": 0.5925155925155925,
+      "grad_norm": 0.6153266265302803,
+      "learning_rate": 1.7744651840391686e-05,
+      "loss": 0.7077,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5951143451143451,
+      "grad_norm": 0.6101914919623531,
+      "learning_rate": 1.7715877154785504e-05,
+      "loss": 0.7017,
+      "step": 1145
+    },
+    {
+      "epoch": 0.5977130977130977,
+      "grad_norm": 0.5892227087436086,
+      "learning_rate": 1.768694371713343e-05,
+      "loss": 0.6901,
+      "step": 1150
+    },
+    {
+      "epoch": 0.6003118503118503,
+      "grad_norm": 0.5792852971539604,
+      "learning_rate": 1.7657852122732958e-05,
+      "loss": 0.6915,
+      "step": 1155
+    },
+    {
+      "epoch": 0.6029106029106029,
+      "grad_norm": 0.5812633503316377,
+      "learning_rate": 1.7628602970135635e-05,
+      "loss": 0.6983,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6055093555093555,
+      "grad_norm": 0.5776629062328037,
+      "learning_rate": 1.7599196861134706e-05,
+      "loss": 0.7033,
+      "step": 1165
+    },
+    {
+      "epoch": 0.6081081081081081,
+      "grad_norm": 0.6291058371783688,
+      "learning_rate": 1.7569634400752763e-05,
+      "loss": 0.6976,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6107068607068608,
+      "grad_norm": 0.5613902945792599,
+      "learning_rate": 1.753991619722928e-05,
+      "loss": 0.7093,
+      "step": 1175
+    },
+    {
+      "epoch": 0.6133056133056133,
+      "grad_norm": 0.6383056669880055,
+      "learning_rate": 1.7510042862008102e-05,
+      "loss": 0.7049,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6159043659043659,
+      "grad_norm": 0.6039268449286719,
+      "learning_rate": 1.7480015009724873e-05,
+      "loss": 0.702,
+      "step": 1185
+    },
+    {
+      "epoch": 0.6185031185031185,
+      "grad_norm": 1.092030131308809,
+      "learning_rate": 1.7449833258194376e-05,
+      "loss": 0.6967,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6211018711018711,
+      "grad_norm": 0.6403075410117294,
+      "learning_rate": 1.7419498228397826e-05,
+      "loss": 0.707,
+      "step": 1195
+    },
+    {
+      "epoch": 0.6237006237006237,
+      "grad_norm": 0.6802172468521516,
+      "learning_rate": 1.7389010544470105e-05,
+      "loss": 0.7178,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6262993762993763,
+      "grad_norm": 0.5789178063330651,
+      "learning_rate": 1.7358370833686907e-05,
+      "loss": 0.6981,
+      "step": 1205
+    },
+    {
+      "epoch": 0.6288981288981289,
+      "grad_norm": 0.5616632715895711,
+      "learning_rate": 1.732757972645183e-05,
+      "loss": 0.6886,
+      "step": 1210
+    },
+    {
+      "epoch": 0.6314968814968815,
+      "grad_norm": 0.5702377179814124,
+      "learning_rate": 1.7296637856283432e-05,
+      "loss": 0.6876,
+      "step": 1215
+    },
+    {
+      "epoch": 0.6340956340956341,
+      "grad_norm": 0.5951629365447559,
+      "learning_rate": 1.7265545859802154e-05,
+      "loss": 0.6976,
+      "step": 1220
+    },
+    {
+      "epoch": 0.6366943866943867,
+      "grad_norm": 0.5872853132760645,
+      "learning_rate": 1.7234304376717252e-05,
+      "loss": 0.6992,
+      "step": 1225
+    },
+    {
+      "epoch": 0.6392931392931392,
+      "grad_norm": 0.570568895020055,
+      "learning_rate": 1.720291404981364e-05,
+      "loss": 0.6927,
+      "step": 1230
+    },
+    {
+      "epoch": 0.6418918918918919,
+      "grad_norm": 0.5451578107308829,
+      "learning_rate": 1.7171375524938638e-05,
+      "loss": 0.7083,
+      "step": 1235
+    },
+    {
+      "epoch": 0.6444906444906445,
+      "grad_norm": 0.5444907291949577,
+      "learning_rate": 1.71396894509887e-05,
+      "loss": 0.7107,
+      "step": 1240
+    },
+    {
+      "epoch": 0.6470893970893971,
+      "grad_norm": 0.5605468917786897,
+      "learning_rate": 1.710785647989607e-05,
+      "loss": 0.7209,
+      "step": 1245
+    },
+    {
+      "epoch": 0.6496881496881497,
+      "grad_norm": 0.578999844192668,
+      "learning_rate": 1.7075877266615347e-05,
+      "loss": 0.7045,
+      "step": 1250
+    },
+    {
+      "epoch": 0.6522869022869023,
+      "grad_norm": 0.5894696454082586,
+      "learning_rate": 1.704375246911004e-05,
+      "loss": 0.7134,
+      "step": 1255
+    },
+    {
+      "epoch": 0.6548856548856549,
+      "grad_norm": 0.6210688729366013,
+      "learning_rate": 1.7011482748338998e-05,
+      "loss": 0.6995,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6574844074844075,
+      "grad_norm": 0.5323403441786221,
+      "learning_rate": 1.697906876824283e-05,
+      "loss": 0.6931,
+      "step": 1265
+    },
+    {
+      "epoch": 0.66008316008316,
+      "grad_norm": 0.5639934756275331,
+      "learning_rate": 1.6946511195730254e-05,
+      "loss": 0.7006,
+      "step": 1270
+    },
+    {
+      "epoch": 0.6626819126819127,
+      "grad_norm": 0.5421699251312687,
+      "learning_rate": 1.6913810700664337e-05,
+      "loss": 0.6995,
+      "step": 1275
+    },
+    {
+      "epoch": 0.6652806652806653,
+      "grad_norm": 0.5498793219946022,
+      "learning_rate": 1.6880967955848762e-05,
+      "loss": 0.6978,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6678794178794178,
+      "grad_norm": 0.5867341882206146,
+      "learning_rate": 1.6847983637013946e-05,
+      "loss": 0.7097,
+      "step": 1285
+    },
+    {
+      "epoch": 0.6704781704781705,
+      "grad_norm": 0.5399055609353709,
+      "learning_rate": 1.6814858422803162e-05,
+      "loss": 0.7007,
+      "step": 1290
+    },
+    {
+      "epoch": 0.6730769230769231,
+      "grad_norm": 0.5807292558166693,
+      "learning_rate": 1.6781592994758563e-05,
+      "loss": 0.7129,
+      "step": 1295
+    },
+    {
+      "epoch": 0.6756756756756757,
+      "grad_norm": 0.5656285736891911,
+      "learning_rate": 1.674818803730716e-05,
+      "loss": 0.697,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6782744282744283,
+      "grad_norm": 0.6062022250444103,
+      "learning_rate": 1.671464423774675e-05,
+      "loss": 0.6933,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6808731808731808,
+      "grad_norm": 0.5432055936520285,
+      "learning_rate": 1.668096228623176e-05,
+      "loss": 0.7042,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6834719334719335,
+      "grad_norm": 0.5696528007526865,
+      "learning_rate": 1.664714287575906e-05,
+      "loss": 0.6921,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6860706860706861,
+      "grad_norm": 0.5723069912815502,
+      "learning_rate": 1.66131867021537e-05,
+      "loss": 0.7016,
+      "step": 1320
+    },
+    {
+      "epoch": 0.6886694386694386,
+      "grad_norm": 0.5709607134990504,
+      "learning_rate": 1.6579094464054595e-05,
+      "loss": 0.7113,
+      "step": 1325
+    },
+    {
+      "epoch": 0.6912681912681913,
+      "grad_norm": 0.6010864530745209,
+      "learning_rate": 1.6544866862900146e-05,
+      "loss": 0.695,
+      "step": 1330
+    },
+    {
+      "epoch": 0.6938669438669439,
+      "grad_norm": 0.5800151181524591,
+      "learning_rate": 1.6510504602913815e-05,
+      "loss": 0.6952,
+      "step": 1335
+    },
+    {
+      "epoch": 0.6964656964656964,
+      "grad_norm": 0.6157566120129612,
+      "learning_rate": 1.6476008391089628e-05,
+      "loss": 0.7105,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6990644490644491,
+      "grad_norm": 0.5752568805255817,
+      "learning_rate": 1.644137893717764e-05,
+      "loss": 0.7161,
+      "step": 1345
+    },
+    {
+      "epoch": 0.7016632016632016,
+      "grad_norm": 0.5767294919201806,
+      "learning_rate": 1.6406616953669317e-05,
+      "loss": 0.6926,
+      "step": 1350
+    },
+    {
+      "epoch": 0.7042619542619543,
+      "grad_norm": 0.6028116817361912,
+      "learning_rate": 1.637172315578289e-05,
+      "loss": 0.6878,
+      "step": 1355
+    },
+    {
+      "epoch": 0.7068607068607069,
+      "grad_norm": 0.5528228994127236,
+      "learning_rate": 1.6336698261448632e-05,
+      "loss": 0.702,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7094594594594594,
+      "grad_norm": 0.5213097969559868,
+      "learning_rate": 1.6301542991294094e-05,
+      "loss": 0.6947,
+      "step": 1365
+    },
+    {
+      "epoch": 0.7120582120582121,
+      "grad_norm": 0.5370559355171456,
+      "learning_rate": 1.626625806862927e-05,
+      "loss": 0.6938,
+      "step": 1370
+    },
+    {
+      "epoch": 0.7146569646569647,
+      "grad_norm": 0.5319720586513772,
+      "learning_rate": 1.623084421943172e-05,
+      "loss": 0.691,
+      "step": 1375
+    },
+    {
+      "epoch": 0.7172557172557172,
+      "grad_norm": 0.5688111345080934,
+      "learning_rate": 1.6195302172331624e-05,
+      "loss": 0.688,
+      "step": 1380
+    },
+    {
+      "epoch": 0.7198544698544699,
+      "grad_norm": 0.5580488505367257,
+      "learning_rate": 1.6159632658596815e-05,
+      "loss": 0.6832,
+      "step": 1385
+    },
+    {
+      "epoch": 0.7224532224532224,
+      "grad_norm": 0.5186643724448894,
+      "learning_rate": 1.6123836412117702e-05,
+      "loss": 0.7004,
+      "step": 1390
+    },
+    {
+      "epoch": 0.725051975051975,
+      "grad_norm": 0.5624601265524769,
+      "learning_rate": 1.608791416939219e-05,
+      "loss": 0.7,
+      "step": 1395
+    },
+    {
+      "epoch": 0.7276507276507277,
+      "grad_norm": 0.5541391195602022,
+      "learning_rate": 1.605186666951052e-05,
+      "loss": 0.6734,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7302494802494802,
+      "grad_norm": 0.5332911572620831,
+      "learning_rate": 1.6015694654140076e-05,
+      "loss": 0.6896,
+      "step": 1405
+    },
+    {
+      "epoch": 0.7328482328482329,
+      "grad_norm": 0.5548630655118624,
+      "learning_rate": 1.59793988675101e-05,
+      "loss": 0.7001,
+      "step": 1410
+    },
+    {
+      "epoch": 0.7354469854469855,
+      "grad_norm": 0.5047178273406415,
+      "learning_rate": 1.5942980056396394e-05,
+      "loss": 0.6959,
+      "step": 1415
+    },
+    {
+      "epoch": 0.738045738045738,
+      "grad_norm": 0.5934685974493243,
+      "learning_rate": 1.5906438970105965e-05,
+      "loss": 0.6904,
+      "step": 1420
+    },
+    {
+      "epoch": 0.7406444906444907,
+      "grad_norm": 0.5527382042918192,
+      "learning_rate": 1.5869776360461596e-05,
+      "loss": 0.6905,
+      "step": 1425
+    },
+    {
+      "epoch": 0.7432432432432432,
+      "grad_norm": 0.5528065876901935,
+      "learning_rate": 1.5832992981786362e-05,
+      "loss": 0.6906,
+      "step": 1430
+    },
+    {
+      "epoch": 0.7458419958419958,
+      "grad_norm": 0.5265702785034122,
+      "learning_rate": 1.5796089590888154e-05,
+      "loss": 0.6857,
+      "step": 1435
+    },
+    {
+      "epoch": 0.7484407484407485,
+      "grad_norm": 0.5438966896525291,
+      "learning_rate": 1.575906694704407e-05,
+      "loss": 0.6926,
+      "step": 1440
+    },
+    {
+      "epoch": 0.751039501039501,
+      "grad_norm": 0.5388783204520624,
+      "learning_rate": 1.5721925811984795e-05,
+      "loss": 0.6973,
+      "step": 1445
+    },
+    {
+      "epoch": 0.7536382536382537,
+      "grad_norm": 0.5389485246395991,
+      "learning_rate": 1.568466694987895e-05,
+      "loss": 0.6962,
+      "step": 1450
+    },
+    {
+      "epoch": 0.7562370062370062,
+      "grad_norm": 0.5935682313499335,
+      "learning_rate": 1.5647291127317353e-05,
+      "loss": 0.6725,
+      "step": 1455
+    },
+    {
+      "epoch": 0.7588357588357588,
+      "grad_norm": 0.5822513257760918,
+      "learning_rate": 1.560979911329725e-05,
+      "loss": 0.6926,
+      "step": 1460
+    },
+    {
+      "epoch": 0.7614345114345115,
+      "grad_norm": 0.591628155657702,
+      "learning_rate": 1.5572191679206503e-05,
+      "loss": 0.6724,
+      "step": 1465
+    },
+    {
+      "epoch": 0.764033264033264,
+      "grad_norm": 0.5308640431230437,
+      "learning_rate": 1.5534469598807703e-05,
+      "loss": 0.693,
+      "step": 1470
+    },
+    {
+      "epoch": 0.7666320166320166,
+      "grad_norm": 0.5305514116153979,
+      "learning_rate": 1.5496633648222246e-05,
+      "loss": 0.6781,
+      "step": 1475
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.5426925004836652,
+      "learning_rate": 1.54586846059144e-05,
+      "loss": 0.7164,
+      "step": 1480
+    },
+    {
+      "epoch": 0.7718295218295218,
+      "grad_norm": 0.523224567107794,
+      "learning_rate": 1.542062325267525e-05,
+      "loss": 0.6934,
+      "step": 1485
+    },
+    {
+      "epoch": 0.7744282744282744,
+      "grad_norm": 0.5414111517541614,
+      "learning_rate": 1.5382450371606643e-05,
+      "loss": 0.6829,
+      "step": 1490
+    },
+    {
+      "epoch": 0.777027027027027,
+      "grad_norm": 0.5960821029242948,
+      "learning_rate": 1.534416674810509e-05,
+      "loss": 0.6924,
+      "step": 1495
+    },
+    {
+      "epoch": 0.7796257796257796,
+      "grad_norm": 0.5781538636667527,
+      "learning_rate": 1.5305773169845598e-05,
+      "loss": 0.695,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7822245322245323,
+      "grad_norm": 0.5764921412336972,
+      "learning_rate": 1.5267270426765445e-05,
+      "loss": 0.6887,
+      "step": 1505
+    },
+    {
+      "epoch": 0.7848232848232848,
+      "grad_norm": 0.65716343258798,
+      "learning_rate": 1.5228659311047969e-05,
+      "loss": 0.6934,
+      "step": 1510
+    },
+    {
+      "epoch": 0.7874220374220374,
+      "grad_norm": 0.5421938070789174,
+      "learning_rate": 1.5189940617106231e-05,
+      "loss": 0.6791,
+      "step": 1515
+    },
+    {
+      "epoch": 0.7900207900207901,
+      "grad_norm": 0.5269524597532019,
+      "learning_rate": 1.5151115141566686e-05,
+      "loss": 0.6901,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7926195426195426,
+      "grad_norm": 0.5653499412081372,
+      "learning_rate": 1.5112183683252797e-05,
+      "loss": 0.6915,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7952182952182952,
+      "grad_norm": 0.5439207742834801,
+      "learning_rate": 1.5073147043168589e-05,
+      "loss": 0.6775,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7978170478170478,
+      "grad_norm": 0.5489788736531123,
+      "learning_rate": 1.5034006024482164e-05,
+      "loss": 0.7036,
+      "step": 1535
+    },
+    {
+      "epoch": 0.8004158004158004,
+      "grad_norm": 0.5550872564267669,
+      "learning_rate": 1.4994761432509201e-05,
+      "loss": 0.6959,
+      "step": 1540
+    },
+    {
+      "epoch": 0.803014553014553,
+      "grad_norm": 0.5490474276501166,
+      "learning_rate": 1.4955414074696369e-05,
+      "loss": 0.6906,
+      "step": 1545
+    },
+    {
+      "epoch": 0.8056133056133056,
+      "grad_norm": 0.5125906404793841,
+      "learning_rate": 1.4915964760604698e-05,
+      "loss": 0.6828,
+      "step": 1550
+    },
+    {
+      "epoch": 0.8082120582120582,
+      "grad_norm": 0.5203971242963676,
+      "learning_rate": 1.4876414301892963e-05,
+      "loss": 0.6825,
+      "step": 1555
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 0.5658527313991132,
+      "learning_rate": 1.4836763512300946e-05,
+      "loss": 0.6928,
+      "step": 1560
+    },
+    {
+      "epoch": 0.8134095634095634,
+      "grad_norm": 0.535836579476803,
+      "learning_rate": 1.4797013207632718e-05,
+      "loss": 0.7018,
+      "step": 1565
+    },
+    {
+      "epoch": 0.816008316008316,
+      "grad_norm": 0.5645034857163748,
+      "learning_rate": 1.475716420573984e-05,
+      "loss": 0.685,
+      "step": 1570
+    },
+    {
+      "epoch": 0.8186070686070686,
+      "grad_norm": 0.5217450999852206,
+      "learning_rate": 1.4717217326504542e-05,
+      "loss": 0.6849,
+      "step": 1575
+    },
+    {
+      "epoch": 0.8212058212058212,
+      "grad_norm": 0.5480983222260942,
+      "learning_rate": 1.467717339182286e-05,
+      "loss": 0.7002,
+      "step": 1580
+    },
+    {
+      "epoch": 0.8238045738045738,
+      "grad_norm": 0.542270358042675,
+      "learning_rate": 1.4637033225587707e-05,
+      "loss": 0.6694,
+      "step": 1585
+    },
+    {
+      "epoch": 0.8264033264033264,
+      "grad_norm": 0.5543644512440328,
+      "learning_rate": 1.4596797653671947e-05,
+      "loss": 0.6921,
+      "step": 1590
+    },
+    {
+      "epoch": 0.829002079002079,
+      "grad_norm": 0.5568668245488615,
+      "learning_rate": 1.4556467503911376e-05,
+      "loss": 0.6852,
+      "step": 1595
+    },
+    {
+      "epoch": 0.8316008316008316,
+      "grad_norm": 0.5656786288010156,
+      "learning_rate": 1.4516043606087712e-05,
+      "loss": 0.6988,
+      "step": 1600
+    },
+    {
+      "epoch": 0.8341995841995842,
+      "grad_norm": 0.5127153017778545,
+      "learning_rate": 1.4475526791911516e-05,
+      "loss": 0.6993,
+      "step": 1605
+    },
+    {
+      "epoch": 0.8367983367983368,
+      "grad_norm": 0.5775499987771998,
+      "learning_rate": 1.4434917895005071e-05,
+      "loss": 0.695,
+      "step": 1610
+    },
+    {
+      "epoch": 0.8393970893970893,
+      "grad_norm": 0.5330354393061656,
+      "learning_rate": 1.4394217750885233e-05,
+      "loss": 0.701,
+      "step": 1615
+    },
+    {
+      "epoch": 0.841995841995842,
+      "grad_norm": 0.5419503755472291,
+      "learning_rate": 1.4353427196946257e-05,
+      "loss": 0.6913,
+      "step": 1620
+    },
+    {
+      "epoch": 0.8445945945945946,
+      "grad_norm": 0.5311784231584057,
+      "learning_rate": 1.4312547072442544e-05,
+      "loss": 0.674,
+      "step": 1625
+    },
+    {
+      "epoch": 0.8471933471933472,
+      "grad_norm": 0.5563211583759982,
+      "learning_rate": 1.4271578218471392e-05,
+      "loss": 0.6907,
+      "step": 1630
+    },
+    {
+      "epoch": 0.8497920997920998,
+      "grad_norm": 0.543414093186935,
+      "learning_rate": 1.4230521477955677e-05,
+      "loss": 0.6933,
+      "step": 1635
+    },
+    {
+      "epoch": 0.8523908523908524,
+      "grad_norm": 0.5240436124612864,
+      "learning_rate": 1.418937769562652e-05,
+      "loss": 0.686,
+      "step": 1640
+    },
+    {
+      "epoch": 0.854989604989605,
+      "grad_norm": 0.5430088145666387,
+      "learning_rate": 1.4148147718005907e-05,
+      "loss": 0.6896,
+      "step": 1645
+    },
+    {
+      "epoch": 0.8575883575883576,
+      "grad_norm": 0.5458623547814142,
+      "learning_rate": 1.4106832393389258e-05,
+      "loss": 0.6908,
+      "step": 1650
+    },
+    {
+      "epoch": 0.8601871101871101,
+      "grad_norm": 0.5634088926920675,
+      "learning_rate": 1.4065432571827992e-05,
+      "loss": 0.7065,
+      "step": 1655
+    },
+    {
+      "epoch": 0.8627858627858628,
+      "grad_norm": 0.5501420194013241,
+      "learning_rate": 1.4023949105112034e-05,
+      "loss": 0.6988,
+      "step": 1660
+    },
+    {
+      "epoch": 0.8653846153846154,
+      "grad_norm": 0.545143784026079,
+      "learning_rate": 1.3982382846752265e-05,
+      "loss": 0.7001,
+      "step": 1665
+    },
+    {
+      "epoch": 0.867983367983368,
+      "grad_norm": 0.5514351435035997,
+      "learning_rate": 1.3940734651963008e-05,
+      "loss": 0.6864,
+      "step": 1670
+    },
+    {
+      "epoch": 0.8705821205821206,
+      "grad_norm": 0.578025130173629,
+      "learning_rate": 1.3899005377644389e-05,
+      "loss": 0.6951,
+      "step": 1675
+    },
+    {
+      "epoch": 0.8731808731808732,
+      "grad_norm": 0.5486337422568206,
+      "learning_rate": 1.3857195882364717e-05,
+      "loss": 0.6875,
+      "step": 1680
+    },
+    {
+      "epoch": 0.8757796257796258,
+      "grad_norm": 0.5987234793423925,
+      "learning_rate": 1.3815307026342847e-05,
+      "loss": 0.6895,
+      "step": 1685
+    },
+    {
+      "epoch": 0.8783783783783784,
+      "grad_norm": 0.5723737998517776,
+      "learning_rate": 1.3773339671430441e-05,
+      "loss": 0.6934,
+      "step": 1690
+    },
+    {
+      "epoch": 0.8809771309771309,
+      "grad_norm": 0.5157754043967531,
+      "learning_rate": 1.3731294681094257e-05,
+      "loss": 0.7008,
+      "step": 1695
+    },
+    {
+      "epoch": 0.8835758835758836,
+      "grad_norm": 0.5484329818025037,
+      "learning_rate": 1.3689172920398385e-05,
+      "loss": 0.6588,
+      "step": 1700
+    },
+    {
+      "epoch": 0.8861746361746362,
+      "grad_norm": 0.5313559372585769,
+      "learning_rate": 1.3646975255986448e-05,
+      "loss": 0.682,
+      "step": 1705
+    },
+    {
+      "epoch": 0.8887733887733887,
+      "grad_norm": 0.5576768736741219,
+      "learning_rate": 1.3604702556063751e-05,
+      "loss": 0.6964,
+      "step": 1710
+    },
+    {
+      "epoch": 0.8913721413721414,
+      "grad_norm": 0.5719657285080421,
+      "learning_rate": 1.3562355690379455e-05,
+      "loss": 0.7023,
+      "step": 1715
+    },
+    {
+      "epoch": 0.893970893970894,
+      "grad_norm": 0.5445374932912297,
+      "learning_rate": 1.3519935530208642e-05,
+      "loss": 0.6969,
+      "step": 1720
+    },
+    {
+      "epoch": 0.8965696465696466,
+      "grad_norm": 0.5332385586498977,
+      "learning_rate": 1.347744294833442e-05,
+      "loss": 0.7009,
+      "step": 1725
+    },
+    {
+      "epoch": 0.8991683991683992,
+      "grad_norm": 0.5518972270290637,
+      "learning_rate": 1.3434878819029949e-05,
+      "loss": 0.6783,
+      "step": 1730
+    },
+    {
+      "epoch": 0.9017671517671517,
+      "grad_norm": 0.5111471501880547,
+      "learning_rate": 1.3392244018040459e-05,
+      "loss": 0.6733,
+      "step": 1735
+    },
+    {
+      "epoch": 0.9043659043659044,
+      "grad_norm": 0.5204670594020995,
+      "learning_rate": 1.3349539422565227e-05,
+      "loss": 0.6768,
+      "step": 1740
+    },
+    {
+      "epoch": 0.906964656964657,
+      "grad_norm": 0.5185707128359598,
+      "learning_rate": 1.3306765911239536e-05,
+      "loss": 0.6836,
+      "step": 1745
+    },
+    {
+      "epoch": 0.9095634095634095,
+      "grad_norm": 0.5371823929518809,
+      "learning_rate": 1.3263924364116598e-05,
+      "loss": 0.6757,
+      "step": 1750
+    },
+    {
+      "epoch": 0.9121621621621622,
+      "grad_norm": 0.5512113239584037,
+      "learning_rate": 1.3221015662649435e-05,
+      "loss": 0.6874,
+      "step": 1755
+    },
+    {
+      "epoch": 0.9147609147609148,
+      "grad_norm": 0.5413143469260978,
+      "learning_rate": 1.3178040689672746e-05,
+      "loss": 0.6903,
+      "step": 1760
+    },
+    {
+      "epoch": 0.9173596673596673,
+      "grad_norm": 0.5448713291982036,
+      "learning_rate": 1.3135000329384768e-05,
+      "loss": 0.6749,
+      "step": 1765
+    },
+    {
+      "epoch": 0.91995841995842,
+      "grad_norm": 0.5694338929447293,
+      "learning_rate": 1.3091895467329043e-05,
+      "loss": 0.6992,
+      "step": 1770
+    },
+    {
+      "epoch": 0.9225571725571725,
+      "grad_norm": 0.5680751056661771,
+      "learning_rate": 1.304872699037623e-05,
+      "loss": 0.6778,
+      "step": 1775
+    },
+    {
+      "epoch": 0.9251559251559252,
+      "grad_norm": 0.5827527502015905,
+      "learning_rate": 1.3005495786705847e-05,
+      "loss": 0.6801,
+      "step": 1780
+    },
+    {
+      "epoch": 0.9277546777546778,
+      "grad_norm": 0.5414753069080744,
+      "learning_rate": 1.2962202745787993e-05,
+      "loss": 0.6828,
+      "step": 1785
+    },
+    {
+      "epoch": 0.9303534303534303,
+      "grad_norm": 0.5954708948041362,
+      "learning_rate": 1.2918848758365047e-05,
+      "loss": 0.6804,
+      "step": 1790
+    },
+    {
+      "epoch": 0.932952182952183,
+      "grad_norm": 0.4986624974413991,
+      "learning_rate": 1.2875434716433364e-05,
+      "loss": 0.6807,
+      "step": 1795
+    },
+    {
+      "epoch": 0.9355509355509356,
+      "grad_norm": 0.512872956001043,
+      "learning_rate": 1.2831961513224883e-05,
+      "loss": 0.6758,
+      "step": 1800
+    },
+    {
+      "epoch": 0.9381496881496881,
+      "grad_norm": 0.5424188290664714,
+      "learning_rate": 1.278843004318878e-05,
+      "loss": 0.6684,
+      "step": 1805
+    },
+    {
+      "epoch": 0.9407484407484408,
+      "grad_norm": 0.5279446908433959,
+      "learning_rate": 1.274484120197305e-05,
+      "loss": 0.6732,
+      "step": 1810
+    },
+    {
+      "epoch": 0.9433471933471933,
+      "grad_norm": 0.5019944493922829,
+      "learning_rate": 1.2701195886406094e-05,
+      "loss": 0.6736,
+      "step": 1815
+    },
+    {
+      "epoch": 0.9459459459459459,
+      "grad_norm": 0.5441846652026447,
+      "learning_rate": 1.2657494994478248e-05,
+      "loss": 0.6803,
+      "step": 1820
+    },
+    {
+      "epoch": 0.9485446985446986,
+      "grad_norm": 0.5027297117206215,
+      "learning_rate": 1.2613739425323318e-05,
+      "loss": 0.6916,
+      "step": 1825
+    },
+    {
+      "epoch": 0.9511434511434511,
+      "grad_norm": 0.5208646951194689,
+      "learning_rate": 1.2569930079200083e-05,
+      "loss": 0.6674,
+      "step": 1830
+    },
+    {
+      "epoch": 0.9537422037422038,
+      "grad_norm": 0.5229959541344918,
+      "learning_rate": 1.2526067857473763e-05,
+      "loss": 0.6808,
+      "step": 1835
+    },
+    {
+      "epoch": 0.9563409563409564,
+      "grad_norm": 0.5436749548564572,
+      "learning_rate": 1.2482153662597478e-05,
+      "loss": 0.682,
+      "step": 1840
+    },
+    {
+      "epoch": 0.9589397089397089,
+      "grad_norm": 0.525934850158874,
+      "learning_rate": 1.2438188398093689e-05,
+      "loss": 0.6684,
+      "step": 1845
+    },
+    {
+      "epoch": 0.9615384615384616,
+      "grad_norm": 0.5241344769977555,
+      "learning_rate": 1.2394172968535596e-05,
+      "loss": 0.6845,
+      "step": 1850
+    },
+    {
+      "epoch": 0.9641372141372141,
+      "grad_norm": 0.5334714104019,
+      "learning_rate": 1.235010827952853e-05,
+      "loss": 0.6752,
+      "step": 1855
+    },
+    {
+      "epoch": 0.9667359667359667,
+      "grad_norm": 0.4981678215882384,
+      "learning_rate": 1.230599523769132e-05,
+      "loss": 0.6824,
+      "step": 1860
+    },
+    {
+      "epoch": 0.9693347193347194,
+      "grad_norm": 0.529971709936305,
+      "learning_rate": 1.2261834750637648e-05,
+      "loss": 0.695,
+      "step": 1865
+    },
+    {
+      "epoch": 0.9719334719334719,
+      "grad_norm": 0.5288630054972472,
+      "learning_rate": 1.2217627726957362e-05,
+      "loss": 0.6833,
+      "step": 1870
+    },
+    {
+      "epoch": 0.9745322245322245,
+      "grad_norm": 0.5109511701940944,
+      "learning_rate": 1.217337507619779e-05,
+      "loss": 0.6688,
+      "step": 1875
+    },
+    {
+      "epoch": 0.9771309771309772,
+      "grad_norm": 0.5251862301897194,
+      "learning_rate": 1.212907770884503e-05,
+      "loss": 0.6752,
+      "step": 1880
+    },
+    {
+      "epoch": 0.9797297297297297,
+      "grad_norm": 0.5081841213925302,
+      "learning_rate": 1.2084736536305199e-05,
+      "loss": 0.6802,
+      "step": 1885
+    },
+    {
+      "epoch": 0.9823284823284824,
+      "grad_norm": 0.5199130862705122,
+      "learning_rate": 1.2040352470885705e-05,
+      "loss": 0.6585,
+      "step": 1890
+    },
+    {
+      "epoch": 0.9849272349272349,
+      "grad_norm": 0.5127627640317641,
+      "learning_rate": 1.1995926425776463e-05,
+      "loss": 0.6753,
+      "step": 1895
+    },
+    {
+      "epoch": 0.9875259875259875,
+      "grad_norm": 0.5057020713500769,
+      "learning_rate": 1.1951459315031103e-05,
+      "loss": 0.6823,
+      "step": 1900
+    },
+    {
+      "epoch": 0.9901247401247402,
+      "grad_norm": 0.5368825129232229,
+      "learning_rate": 1.1906952053548173e-05,
+      "loss": 0.6775,
+      "step": 1905
+    },
+    {
+      "epoch": 0.9927234927234927,
+      "grad_norm": 0.48766377489485674,
+      "learning_rate": 1.1862405557052316e-05,
+      "loss": 0.6731,
+      "step": 1910
+    },
+    {
+      "epoch": 0.9953222453222453,
+      "grad_norm": 0.5077162603005374,
+      "learning_rate": 1.1817820742075418e-05,
+      "loss": 0.659,
+      "step": 1915
+    },
+    {
+      "epoch": 0.997920997920998,
+      "grad_norm": 0.5273871660362291,
+      "learning_rate": 1.1773198525937758e-05,
+      "loss": 0.679,
+      "step": 1920
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.7630624771118164,
+      "eval_runtime": 104.7865,
+      "eval_samples_per_second": 78.35,
+      "eval_steps_per_second": 1.231,
+      "step": 1924
+    },
+    {
+      "epoch": 1.0005197505197505,
+      "grad_norm": 0.7167016181446053,
+      "learning_rate": 1.1728539826729135e-05,
+      "loss": 0.6587,
+      "step": 1925
+    },
+    {
+      "epoch": 1.003118503118503,
+      "grad_norm": 0.6041178102787718,
+      "learning_rate": 1.168384556328998e-05,
+      "loss": 0.6121,
+      "step": 1930
+    },
+    {
+      "epoch": 1.0057172557172558,
+      "grad_norm": 0.5921694491638905,
+      "learning_rate": 1.163911665519244e-05,
+      "loss": 0.6028,
+      "step": 1935
+    },
+    {
+      "epoch": 1.0083160083160083,
+      "grad_norm": 0.6013653429933059,
+      "learning_rate": 1.1594354022721475e-05,
+      "loss": 0.6156,
+      "step": 1940
+    },
+    {
+      "epoch": 1.0109147609147608,
+      "grad_norm": 0.6105316954631931,
+      "learning_rate": 1.1549558586855909e-05,
+      "loss": 0.5945,
+      "step": 1945
+    },
+    {
+      "epoch": 1.0135135135135136,
+      "grad_norm": 0.535707927809498,
+      "learning_rate": 1.150473126924949e-05,
+      "loss": 0.5922,
+      "step": 1950
+    },
+    {
+      "epoch": 1.0161122661122661,
+      "grad_norm": 0.5556825438783977,
+      "learning_rate": 1.1459872992211923e-05,
+      "loss": 0.5953,
+      "step": 1955
+    },
+    {
+      "epoch": 1.0187110187110187,
+      "grad_norm": 0.5691489620123985,
+      "learning_rate": 1.1414984678689895e-05,
+      "loss": 0.599,
+      "step": 1960
+    },
+    {
+      "epoch": 1.0213097713097714,
+      "grad_norm": 0.5936266445205919,
+      "learning_rate": 1.1370067252248085e-05,
+      "loss": 0.5949,
+      "step": 1965
+    },
+    {
+      "epoch": 1.023908523908524,
+      "grad_norm": 0.6066096499835257,
+      "learning_rate": 1.1325121637050161e-05,
+      "loss": 0.5971,
+      "step": 1970
+    },
+    {
+      "epoch": 1.0265072765072765,
+      "grad_norm": 0.5681394506425539,
+      "learning_rate": 1.128014875783977e-05,
+      "loss": 0.5864,
+      "step": 1975
+    },
+    {
+      "epoch": 1.0291060291060292,
+      "grad_norm": 0.5628786880019524,
+      "learning_rate": 1.1235149539921509e-05,
+      "loss": 0.592,
+      "step": 1980
+    },
+    {
+      "epoch": 1.0317047817047817,
+      "grad_norm": 0.5467263882406993,
+      "learning_rate": 1.1190124909141877e-05,
+      "loss": 0.6075,
+      "step": 1985
+    },
+    {
+      "epoch": 1.0343035343035343,
+      "grad_norm": 0.5440093360987802,
+      "learning_rate": 1.1145075791870253e-05,
+      "loss": 0.5988,
+      "step": 1990
+    },
+    {
+      "epoch": 1.0369022869022868,
+      "grad_norm": 0.5752544838637329,
+      "learning_rate": 1.1100003114979803e-05,
+      "loss": 0.615,
+      "step": 1995
+    },
+    {
+      "epoch": 1.0395010395010396,
+      "grad_norm": 0.5310516110561704,
+      "learning_rate": 1.1054907805828427e-05,
+      "loss": 0.6005,
+      "step": 2000
+    },
+    {
+      "epoch": 1.042099792099792,
+      "grad_norm": 0.5432466126208855,
+      "learning_rate": 1.1009790792239692e-05,
+      "loss": 0.5866,
+      "step": 2005
+    },
+    {
+      "epoch": 1.0446985446985446,
+      "grad_norm": 0.5644600605835878,
+      "learning_rate": 1.0964653002483714e-05,
+      "loss": 0.5936,
+      "step": 2010
+    },
+    {
+      "epoch": 1.0472972972972974,
+      "grad_norm": 0.5272489276362843,
+      "learning_rate": 1.0919495365258077e-05,
+      "loss": 0.597,
+      "step": 2015
+    },
+    {
+      "epoch": 1.04989604989605,
+      "grad_norm": 0.5617695811131541,
+      "learning_rate": 1.0874318809668717e-05,
+      "loss": 0.6065,
+      "step": 2020
+    },
+    {
+      "epoch": 1.0524948024948024,
+      "grad_norm": 0.5192198208489455,
+      "learning_rate": 1.0829124265210822e-05,
+      "loss": 0.6075,
+      "step": 2025
+    },
+    {
+      "epoch": 1.0550935550935552,
+      "grad_norm": 0.5440321607158946,
+      "learning_rate": 1.0783912661749682e-05,
+      "loss": 0.5993,
+      "step": 2030
+    },
+    {
+      "epoch": 1.0576923076923077,
+      "grad_norm": 0.5380219431230713,
+      "learning_rate": 1.0738684929501577e-05,
+      "loss": 0.6047,
+      "step": 2035
+    },
+    {
+      "epoch": 1.0602910602910602,
+      "grad_norm": 0.5309774372945627,
+      "learning_rate": 1.069344199901464e-05,
+      "loss": 0.6057,
+      "step": 2040
+    },
+    {
+      "epoch": 1.062889812889813,
+      "grad_norm": 0.5317011424510519,
+      "learning_rate": 1.064818480114969e-05,
+      "loss": 0.6087,
+      "step": 2045
+    },
+    {
+      "epoch": 1.0654885654885655,
+      "grad_norm": 0.5636006054084751,
+      "learning_rate": 1.0602914267061102e-05,
+      "loss": 0.6016,
+      "step": 2050
+    },
+    {
+      "epoch": 1.068087318087318,
+      "grad_norm": 0.5385118881739976,
+      "learning_rate": 1.0557631328177635e-05,
+      "loss": 0.6009,
+      "step": 2055
+    },
+    {
+      "epoch": 1.0706860706860706,
+      "grad_norm": 0.5387163558386009,
+      "learning_rate": 1.0512336916183282e-05,
+      "loss": 0.5919,
+      "step": 2060
+    },
+    {
+      "epoch": 1.0732848232848233,
+      "grad_norm": 0.5550215843050625,
+      "learning_rate": 1.046703196299808e-05,
+      "loss": 0.6032,
+      "step": 2065
+    },
+    {
+      "epoch": 1.0758835758835759,
+      "grad_norm": 0.5317572302029103,
+      "learning_rate": 1.042171740075896e-05,
+      "loss": 0.6061,
+      "step": 2070
+    },
+    {
+      "epoch": 1.0784823284823284,
+      "grad_norm": 0.5953262222453396,
+      "learning_rate": 1.037639416180055e-05,
+      "loss": 0.5968,
+      "step": 2075
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 0.5468172409024149,
+      "learning_rate": 1.0331063178635991e-05,
+      "loss": 0.5942,
+      "step": 2080
+    },
+    {
+      "epoch": 1.0836798336798337,
+      "grad_norm": 0.5337399744353215,
+      "learning_rate": 1.028572538393778e-05,
+      "loss": 0.5784,
+      "step": 2085
+    },
+    {
+      "epoch": 1.0862785862785862,
+      "grad_norm": 0.5918184869466311,
+      "learning_rate": 1.0240381710518544e-05,
+      "loss": 0.5825,
+      "step": 2090
+    },
+    {
+      "epoch": 1.088877338877339,
+      "grad_norm": 0.5415874642045884,
+      "learning_rate": 1.0195033091311866e-05,
+      "loss": 0.6081,
+      "step": 2095
+    },
+    {
+      "epoch": 1.0914760914760915,
+      "grad_norm": 0.5414796664701005,
+      "learning_rate": 1.0149680459353085e-05,
+      "loss": 0.5916,
+      "step": 2100
+    },
+    {
+      "epoch": 1.094074844074844,
+      "grad_norm": 0.5523534552531586,
+      "learning_rate": 1.0104324747760103e-05,
+      "loss": 0.6108,
+      "step": 2105
+    },
+    {
+      "epoch": 1.0966735966735968,
+      "grad_norm": 0.691647511565078,
+      "learning_rate": 1.0058966889714192e-05,
+      "loss": 0.6,
+      "step": 2110
+    },
+    {
+      "epoch": 1.0992723492723493,
+      "grad_norm": 0.5912743887434098,
+      "learning_rate": 1.0013607818440775e-05,
+      "loss": 0.6006,
+      "step": 2115
+    },
+    {
+      "epoch": 1.1018711018711018,
+      "grad_norm": 0.5460338682601912,
+      "learning_rate": 9.968248467190246e-06,
+      "loss": 0.6017,
+      "step": 2120
+    },
+    {
+      "epoch": 1.1044698544698546,
+      "grad_norm": 0.5664759203624757,
+      "learning_rate": 9.922889769218754e-06,
+      "loss": 0.6093,
+      "step": 2125
+    },
+    {
+      "epoch": 1.107068607068607,
+      "grad_norm": 0.5666035631954701,
+      "learning_rate": 9.877532657769006e-06,
+      "loss": 0.5971,
+      "step": 2130
+    },
+    {
+      "epoch": 1.1096673596673596,
+      "grad_norm": 0.5200286477603056,
+      "learning_rate": 9.832178066051074e-06,
+      "loss": 0.5934,
+      "step": 2135
+    },
+    {
+      "epoch": 1.1122661122661124,
+      "grad_norm": 0.6074145785434165,
+      "learning_rate": 9.78682692722318e-06,
+      "loss": 0.5886,
+      "step": 2140
+    },
+    {
+      "epoch": 1.114864864864865,
+      "grad_norm": 0.5541608986138071,
+      "learning_rate": 9.741480174372505e-06,
+      "loss": 0.6028,
+      "step": 2145
+    },
+    {
+      "epoch": 1.1174636174636174,
+      "grad_norm": 0.5324738113711789,
+      "learning_rate": 9.696138740495992e-06,
+      "loss": 0.5963,
+      "step": 2150
+    },
+    {
+      "epoch": 1.12006237006237,
+      "grad_norm": 0.5848651297296711,
+      "learning_rate": 9.650803558481147e-06,
+      "loss": 0.5807,
+      "step": 2155
+    },
+    {
+      "epoch": 1.1226611226611227,
+      "grad_norm": 0.5602080935436595,
+      "learning_rate": 9.605475561086842e-06,
+      "loss": 0.5888,
+      "step": 2160
+    },
+    {
+      "epoch": 1.1252598752598753,
+      "grad_norm": 0.5653703789523455,
+      "learning_rate": 9.560155680924137e-06,
+      "loss": 0.5916,
+      "step": 2165
+    },
+    {
+      "epoch": 1.1278586278586278,
+      "grad_norm": 0.5356042555591374,
+      "learning_rate": 9.514844850437067e-06,
+      "loss": 0.5877,
+      "step": 2170
+    },
+    {
+      "epoch": 1.1304573804573805,
+      "grad_norm": 0.5682089324184239,
+      "learning_rate": 9.469544001883491e-06,
+      "loss": 0.5867,
+      "step": 2175
+    },
+    {
+      "epoch": 1.133056133056133,
+      "grad_norm": 0.5334569975639222,
+      "learning_rate": 9.424254067315875e-06,
+      "loss": 0.6082,
+      "step": 2180
+    },
+    {
+      "epoch": 1.1356548856548856,
+      "grad_norm": 0.6011244426639291,
+      "learning_rate": 9.378975978562147e-06,
+      "loss": 0.5934,
+      "step": 2185
+    },
+    {
+      "epoch": 1.1382536382536383,
+      "grad_norm": 0.578376500402116,
+      "learning_rate": 9.3337106672065e-06,
+      "loss": 0.5836,
+      "step": 2190
+    },
+    {
+      "epoch": 1.1408523908523909,
+      "grad_norm": 0.5588463035978978,
+      "learning_rate": 9.288459064570239e-06,
+      "loss": 0.6047,
+      "step": 2195
+    },
+    {
+      "epoch": 1.1434511434511434,
+      "grad_norm": 0.5657302559519196,
+      "learning_rate": 9.243222101692617e-06,
+      "loss": 0.5858,
+      "step": 2200
+    },
+    {
+      "epoch": 1.1460498960498962,
+      "grad_norm": 0.583919168737973,
+      "learning_rate": 9.19800070931168e-06,
+      "loss": 0.607,
+      "step": 2205
+    },
+    {
+      "epoch": 1.1486486486486487,
+      "grad_norm": 0.5817136037512669,
+      "learning_rate": 9.15279581784511e-06,
+      "loss": 0.589,
+      "step": 2210
+    },
+    {
+      "epoch": 1.1512474012474012,
+      "grad_norm": 0.538036225434835,
+      "learning_rate": 9.107608357371092e-06,
+      "loss": 0.6052,
+      "step": 2215
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5265174041614111,
+      "learning_rate": 9.062439257609165e-06,
+      "loss": 0.5953,
+      "step": 2220
+    },
+    {
+      "epoch": 1.1564449064449065,
+      "grad_norm": 0.548975305771916,
+      "learning_rate": 9.01728944790112e-06,
+      "loss": 0.5931,
+      "step": 2225
+    },
+    {
+      "epoch": 1.159043659043659,
+      "grad_norm": 0.5481727349763981,
+      "learning_rate": 8.972159857191838e-06,
+      "loss": 0.5951,
+      "step": 2230
+    },
+    {
+      "epoch": 1.1616424116424116,
+      "grad_norm": 0.5808749569289294,
+      "learning_rate": 8.927051414010213e-06,
+      "loss": 0.5925,
+      "step": 2235
+    },
+    {
+      "epoch": 1.1642411642411643,
+      "grad_norm": 0.5597977821004193,
+      "learning_rate": 8.881965046450033e-06,
+      "loss": 0.5983,
+      "step": 2240
+    },
+    {
+      "epoch": 1.1668399168399168,
+      "grad_norm": 0.5732606987628547,
+      "learning_rate": 8.83690168215088e-06,
+      "loss": 0.6003,
+      "step": 2245
+    },
+    {
+      "epoch": 1.1694386694386694,
+      "grad_norm": 0.5317615005846271,
+      "learning_rate": 8.791862248279059e-06,
+      "loss": 0.6011,
+      "step": 2250
+    },
+    {
+      "epoch": 1.1720374220374221,
+      "grad_norm": 0.5175378118538798,
+      "learning_rate": 8.746847671508506e-06,
+      "loss": 0.5906,
+      "step": 2255
+    },
+    {
+      "epoch": 1.1746361746361746,
+      "grad_norm": 0.5827570884628767,
+      "learning_rate": 8.701858878001731e-06,
+      "loss": 0.5947,
+      "step": 2260
+    },
+    {
+      "epoch": 1.1772349272349272,
+      "grad_norm": 0.5832582553430286,
+      "learning_rate": 8.656896793390763e-06,
+      "loss": 0.6021,
+      "step": 2265
+    },
+    {
+      "epoch": 1.17983367983368,
+      "grad_norm": 0.5252465600060544,
+      "learning_rate": 8.6119623427581e-06,
+      "loss": 0.5946,
+      "step": 2270
+    },
+    {
+      "epoch": 1.1824324324324325,
+      "grad_norm": 0.5535883003187351,
+      "learning_rate": 8.567056450617678e-06,
+      "loss": 0.5997,
+      "step": 2275
+    },
+    {
+      "epoch": 1.185031185031185,
+      "grad_norm": 0.5265834205804355,
+      "learning_rate": 8.522180040895855e-06,
+      "loss": 0.5919,
+      "step": 2280
+    },
+    {
+      "epoch": 1.1876299376299375,
+      "grad_norm": 0.5588080207098279,
+      "learning_rate": 8.477334036912392e-06,
+      "loss": 0.599,
+      "step": 2285
+    },
+    {
+      "epoch": 1.1902286902286903,
+      "grad_norm": 0.5254040822916115,
+      "learning_rate": 8.432519361361458e-06,
+      "loss": 0.5947,
+      "step": 2290
+    },
+    {
+      "epoch": 1.1928274428274428,
+      "grad_norm": 0.5455269093593619,
+      "learning_rate": 8.387736936292654e-06,
+      "loss": 0.5939,
+      "step": 2295
+    },
+    {
+      "epoch": 1.1954261954261955,
+      "grad_norm": 0.5639163666347963,
+      "learning_rate": 8.342987683092033e-06,
+      "loss": 0.5949,
+      "step": 2300
+    },
+    {
+      "epoch": 1.198024948024948,
+      "grad_norm": 0.6084945001801156,
+      "learning_rate": 8.298272522463144e-06,
+      "loss": 0.596,
+      "step": 2305
+    },
+    {
+      "epoch": 1.2006237006237006,
+      "grad_norm": 0.5836727328380208,
+      "learning_rate": 8.253592374408095e-06,
+      "loss": 0.5963,
+      "step": 2310
+    },
+    {
+      "epoch": 1.2032224532224531,
+      "grad_norm": 0.5454765026633895,
+      "learning_rate": 8.20894815820862e-06,
+      "loss": 0.6003,
+      "step": 2315
+    },
+    {
+      "epoch": 1.2058212058212059,
+      "grad_norm": 0.5490153680134165,
+      "learning_rate": 8.164340792407159e-06,
+      "loss": 0.594,
+      "step": 2320
+    },
+    {
+      "epoch": 1.2084199584199584,
+      "grad_norm": 0.5442982828513857,
+      "learning_rate": 8.119771194787966e-06,
+      "loss": 0.5939,
+      "step": 2325
+    },
+    {
+      "epoch": 1.211018711018711,
+      "grad_norm": 0.5224447370874735,
+      "learning_rate": 8.07524028235824e-06,
+      "loss": 0.5895,
+      "step": 2330
+    },
+    {
+      "epoch": 1.2136174636174637,
+      "grad_norm": 0.5648657256694566,
+      "learning_rate": 8.030748971329224e-06,
+      "loss": 0.6101,
+      "step": 2335
+    },
+    {
+      "epoch": 1.2162162162162162,
+      "grad_norm": 0.5616352744995109,
+      "learning_rate": 7.986298177097377e-06,
+      "loss": 0.6042,
+      "step": 2340
+    },
+    {
+      "epoch": 1.2188149688149688,
+      "grad_norm": 0.5771472821711124,
+      "learning_rate": 7.941888814225545e-06,
+      "loss": 0.5987,
+      "step": 2345
+    },
+    {
+      "epoch": 1.2214137214137215,
+      "grad_norm": 0.5512099431582866,
+      "learning_rate": 7.897521796424129e-06,
+      "loss": 0.6056,
+      "step": 2350
+    },
+    {
+      "epoch": 1.224012474012474,
+      "grad_norm": 0.5098551231289585,
+      "learning_rate": 7.853198036532288e-06,
+      "loss": 0.6031,
+      "step": 2355
+    },
+    {
+      "epoch": 1.2266112266112266,
+      "grad_norm": 0.5717033338675165,
+      "learning_rate": 7.80891844649917e-06,
+      "loss": 0.592,
+      "step": 2360
+    },
+    {
+      "epoch": 1.2292099792099793,
+      "grad_norm": 0.5550621676748785,
+      "learning_rate": 7.764683937365134e-06,
+      "loss": 0.5877,
+      "step": 2365
+    },
+    {
+      "epoch": 1.2318087318087318,
+      "grad_norm": 0.5823877655186512,
+      "learning_rate": 7.720495419243014e-06,
+      "loss": 0.6226,
+      "step": 2370
+    },
+    {
+      "epoch": 1.2344074844074844,
+      "grad_norm": 0.547937317341142,
+      "learning_rate": 7.676353801299388e-06,
+      "loss": 0.5936,
+      "step": 2375
+    },
+    {
+      "epoch": 1.237006237006237,
+      "grad_norm": 0.5341568281577489,
+      "learning_rate": 7.63225999173589e-06,
+      "loss": 0.5899,
+      "step": 2380
+    },
+    {
+      "epoch": 1.2396049896049897,
+      "grad_norm": 0.5432593899885211,
+      "learning_rate": 7.5882148977704914e-06,
+      "loss": 0.602,
+      "step": 2385
+    },
+    {
+      "epoch": 1.2422037422037422,
+      "grad_norm": 0.5421541547325941,
+      "learning_rate": 7.544219425618862e-06,
+      "loss": 0.5858,
+      "step": 2390
+    },
+    {
+      "epoch": 1.2448024948024947,
+      "grad_norm": 0.5370495755166194,
+      "learning_rate": 7.500274480475716e-06,
+      "loss": 0.6071,
+      "step": 2395
+    },
+    {
+      "epoch": 1.2474012474012475,
+      "grad_norm": 0.5449432404989222,
+      "learning_rate": 7.45638096649619e-06,
+      "loss": 0.5972,
+      "step": 2400
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5660658221047755,
+      "learning_rate": 7.412539786777234e-06,
+      "loss": 0.5943,
+      "step": 2405
+    },
+    {
+      "epoch": 1.2525987525987525,
+      "grad_norm": 0.5247570244603685,
+      "learning_rate": 7.3687518433390414e-06,
+      "loss": 0.5924,
+      "step": 2410
+    },
+    {
+      "epoch": 1.255197505197505,
+      "grad_norm": 0.5207188548015057,
+      "learning_rate": 7.325018037106474e-06,
+      "loss": 0.5917,
+      "step": 2415
+    },
+    {
+      "epoch": 1.2577962577962578,
+      "grad_norm": 0.550416355369781,
+      "learning_rate": 7.281339267890548e-06,
+      "loss": 0.5857,
+      "step": 2420
+    },
+    {
+      "epoch": 1.2603950103950103,
+      "grad_norm": 0.5155652437752087,
+      "learning_rate": 7.237716434369899e-06,
+      "loss": 0.5946,
+      "step": 2425
+    },
+    {
+      "epoch": 1.262993762993763,
+      "grad_norm": 0.5446737196259831,
+      "learning_rate": 7.1941504340723e-06,
+      "loss": 0.602,
+      "step": 2430
+    },
+    {
+      "epoch": 1.2655925155925156,
+      "grad_norm": 0.5241476632554306,
+      "learning_rate": 7.150642163356207e-06,
+      "loss": 0.5848,
+      "step": 2435
+    },
+    {
+      "epoch": 1.2681912681912682,
+      "grad_norm": 0.5368234251185683,
+      "learning_rate": 7.107192517392296e-06,
+      "loss": 0.593,
+      "step": 2440
+    },
+    {
+      "epoch": 1.2707900207900207,
+      "grad_norm": 0.5609013153184366,
+      "learning_rate": 7.063802390145058e-06,
+      "loss": 0.5926,
+      "step": 2445
+    },
+    {
+      "epoch": 1.2733887733887734,
+      "grad_norm": 0.5601427654877619,
+      "learning_rate": 7.020472674354399e-06,
+      "loss": 0.5955,
+      "step": 2450
+    },
+    {
+      "epoch": 1.275987525987526,
+      "grad_norm": 0.5573446244898519,
+      "learning_rate": 6.9772042615172855e-06,
+      "loss": 0.596,
+      "step": 2455
+    },
+    {
+      "epoch": 1.2785862785862787,
+      "grad_norm": 0.5409848408904747,
+      "learning_rate": 6.933998041869381e-06,
+      "loss": 0.5935,
+      "step": 2460
+    },
+    {
+      "epoch": 1.2811850311850312,
+      "grad_norm": 0.535048843399521,
+      "learning_rate": 6.890854904366749e-06,
+      "loss": 0.5887,
+      "step": 2465
+    },
+    {
+      "epoch": 1.2837837837837838,
+      "grad_norm": 0.5178312289683921,
+      "learning_rate": 6.8477757366675526e-06,
+      "loss": 0.5915,
+      "step": 2470
+    },
+    {
+      "epoch": 1.2863825363825363,
+      "grad_norm": 0.5746689510719801,
+      "learning_rate": 6.804761425113796e-06,
+      "loss": 0.5952,
+      "step": 2475
+    },
+    {
+      "epoch": 1.288981288981289,
+      "grad_norm": 0.5494461776895628,
+      "learning_rate": 6.761812854713079e-06,
+      "loss": 0.5905,
+      "step": 2480
+    },
+    {
+      "epoch": 1.2915800415800416,
+      "grad_norm": 0.5437436470618007,
+      "learning_rate": 6.718930909120404e-06,
+      "loss": 0.5881,
+      "step": 2485
+    },
+    {
+      "epoch": 1.2941787941787941,
+      "grad_norm": 0.5670019639541221,
+      "learning_rate": 6.676116470619988e-06,
+      "loss": 0.5945,
+      "step": 2490
+    },
+    {
+      "epoch": 1.2967775467775469,
+      "grad_norm": 0.5754824812127866,
+      "learning_rate": 6.633370420107097e-06,
+      "loss": 0.5975,
+      "step": 2495
+    },
+    {
+      "epoch": 1.2993762993762994,
+      "grad_norm": 0.548088254326578,
+      "learning_rate": 6.590693637069938e-06,
+      "loss": 0.5966,
+      "step": 2500
+    },
+    {
+      "epoch": 1.301975051975052,
+      "grad_norm": 0.5243431037903644,
+      "learning_rate": 6.548086999571566e-06,
+      "loss": 0.5938,
+      "step": 2505
+    },
+    {
+      "epoch": 1.3045738045738045,
+      "grad_norm": 0.565611456496591,
+      "learning_rate": 6.505551384231801e-06,
+      "loss": 0.592,
+      "step": 2510
+    },
+    {
+      "epoch": 1.3071725571725572,
+      "grad_norm": 0.561890682944344,
+      "learning_rate": 6.463087666209203e-06,
+      "loss": 0.5818,
+      "step": 2515
+    },
+    {
+      "epoch": 1.3097713097713097,
+      "grad_norm": 0.5872998314400075,
+      "learning_rate": 6.420696719183072e-06,
+      "loss": 0.5849,
+      "step": 2520
+    },
+    {
+      "epoch": 1.3123700623700625,
+      "grad_norm": 0.5165776292315633,
+      "learning_rate": 6.378379415335456e-06,
+      "loss": 0.5913,
+      "step": 2525
+    },
+    {
+      "epoch": 1.314968814968815,
+      "grad_norm": 0.5271295585343699,
+      "learning_rate": 6.336136625333218e-06,
+      "loss": 0.6001,
+      "step": 2530
+    },
+    {
+      "epoch": 1.3175675675675675,
+      "grad_norm": 0.5541093558682746,
+      "learning_rate": 6.293969218310122e-06,
+      "loss": 0.5986,
+      "step": 2535
+    },
+    {
+      "epoch": 1.32016632016632,
+      "grad_norm": 0.533798380054359,
+      "learning_rate": 6.251878061848948e-06,
+      "loss": 0.5759,
+      "step": 2540
+    },
+    {
+      "epoch": 1.3227650727650728,
+      "grad_norm": 0.5573470744869463,
+      "learning_rate": 6.209864021963638e-06,
+      "loss": 0.5879,
+      "step": 2545
+    },
+    {
+      "epoch": 1.3253638253638254,
+      "grad_norm": 0.5493774653074921,
+      "learning_rate": 6.16792796308148e-06,
+      "loss": 0.6017,
+      "step": 2550
+    },
+    {
+      "epoch": 1.3279625779625779,
+      "grad_norm": 0.582071520963931,
+      "learning_rate": 6.126070748025332e-06,
+      "loss": 0.6025,
+      "step": 2555
+    },
+    {
+      "epoch": 1.3305613305613306,
+      "grad_norm": 0.5144065189940694,
+      "learning_rate": 6.084293237995855e-06,
+      "loss": 0.5827,
+      "step": 2560
+    },
+    {
+      "epoch": 1.3331600831600832,
+      "grad_norm": 0.5139551250609696,
+      "learning_rate": 6.042596292553803e-06,
+      "loss": 0.585,
+      "step": 2565
+    },
+    {
+      "epoch": 1.3357588357588357,
+      "grad_norm": 0.512609402226577,
+      "learning_rate": 6.0009807696023385e-06,
+      "loss": 0.5865,
+      "step": 2570
+    },
+    {
+      "epoch": 1.3383575883575882,
+      "grad_norm": 0.5648181415304806,
+      "learning_rate": 5.959447525369375e-06,
+      "loss": 0.585,
+      "step": 2575
+    },
+    {
+      "epoch": 1.340956340956341,
+      "grad_norm": 0.4961717842990021,
+      "learning_rate": 5.917997414389966e-06,
+      "loss": 0.6065,
+      "step": 2580
+    },
+    {
+      "epoch": 1.3435550935550935,
+      "grad_norm": 0.5369467703030896,
+      "learning_rate": 5.87663128948872e-06,
+      "loss": 0.586,
+      "step": 2585
+    },
+    {
+      "epoch": 1.3461538461538463,
+      "grad_norm": 0.5805263679243358,
+      "learning_rate": 5.835350001762258e-06,
+      "loss": 0.5793,
+      "step": 2590
+    },
+    {
+      "epoch": 1.3487525987525988,
+      "grad_norm": 0.5472679842232231,
+      "learning_rate": 5.794154400561698e-06,
+      "loss": 0.5936,
+      "step": 2595
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 0.543074701262151,
+      "learning_rate": 5.753045333475181e-06,
+      "loss": 0.6068,
+      "step": 2600
+    },
+    {
+      "epoch": 1.3539501039501038,
+      "grad_norm": 0.501759332855818,
+      "learning_rate": 5.712023646310429e-06,
+      "loss": 0.5923,
+      "step": 2605
+    },
+    {
+      "epoch": 1.3565488565488566,
+      "grad_norm": 0.5627676287799811,
+      "learning_rate": 5.671090183077355e-06,
+      "loss": 0.5908,
+      "step": 2610
+    },
+    {
+      "epoch": 1.3591476091476091,
+      "grad_norm": 0.5396141346304916,
+      "learning_rate": 5.630245785970676e-06,
+      "loss": 0.5835,
+      "step": 2615
+    },
+    {
+      "epoch": 1.3617463617463619,
+      "grad_norm": 0.5237498968336539,
+      "learning_rate": 5.589491295352607e-06,
+      "loss": 0.6023,
+      "step": 2620
+    },
+    {
+      "epoch": 1.3643451143451144,
+      "grad_norm": 0.5714671269121867,
+      "learning_rate": 5.548827549735565e-06,
+      "loss": 0.5875,
+      "step": 2625
+    },
+    {
+      "epoch": 1.366943866943867,
+      "grad_norm": 0.5420021477246512,
+      "learning_rate": 5.508255385764908e-06,
+      "loss": 0.5962,
+      "step": 2630
+    },
+    {
+      "epoch": 1.3695426195426195,
+      "grad_norm": 0.5213053662389652,
+      "learning_rate": 5.46777563820172e-06,
+      "loss": 0.5775,
+      "step": 2635
+    },
+    {
+      "epoch": 1.3721413721413722,
+      "grad_norm": 0.5940088174557574,
+      "learning_rate": 5.427389139905661e-06,
+      "loss": 0.5925,
+      "step": 2640
+    },
+    {
+      "epoch": 1.3747401247401247,
+      "grad_norm": 0.5222712044920537,
+      "learning_rate": 5.387096721817799e-06,
+      "loss": 0.5939,
+      "step": 2645
+    },
+    {
+      "epoch": 1.3773388773388773,
+      "grad_norm": 0.533759193353919,
+      "learning_rate": 5.346899212943529e-06,
+      "loss": 0.5891,
+      "step": 2650
+    },
+    {
+      "epoch": 1.37993762993763,
+      "grad_norm": 0.5684539829097972,
+      "learning_rate": 5.306797440335516e-06,
+      "loss": 0.598,
+      "step": 2655
+    },
+    {
+      "epoch": 1.3825363825363826,
+      "grad_norm": 0.577296227438705,
+      "learning_rate": 5.266792229076683e-06,
+      "loss": 0.5908,
+      "step": 2660
+    },
+    {
+      "epoch": 1.385135135135135,
+      "grad_norm": 0.5367239875824128,
+      "learning_rate": 5.2268844022632255e-06,
+      "loss": 0.5933,
+      "step": 2665
+    },
+    {
+      "epoch": 1.3877338877338876,
+      "grad_norm": 0.5512610722369685,
+      "learning_rate": 5.187074780987682e-06,
+      "loss": 0.5952,
+      "step": 2670
+    },
+    {
+      "epoch": 1.3903326403326404,
+      "grad_norm": 0.5045101270772827,
+      "learning_rate": 5.147364184322035e-06,
+      "loss": 0.5793,
+      "step": 2675
+    },
+    {
+      "epoch": 1.392931392931393,
+      "grad_norm": 0.5505398932307536,
+      "learning_rate": 5.1077534293008635e-06,
+      "loss": 0.5973,
+      "step": 2680
+    },
+    {
+      "epoch": 1.3955301455301456,
+      "grad_norm": 0.5618823440106911,
+      "learning_rate": 5.068243330904543e-06,
+      "loss": 0.5954,
+      "step": 2685
+    },
+    {
+      "epoch": 1.3981288981288982,
+      "grad_norm": 0.547767493340866,
+      "learning_rate": 5.028834702042451e-06,
+      "loss": 0.5965,
+      "step": 2690
+    },
+    {
+      "epoch": 1.4007276507276507,
+      "grad_norm": 0.5311004214157612,
+      "learning_rate": 4.9895283535362744e-06,
+      "loss": 0.5761,
+      "step": 2695
+    },
+    {
+      "epoch": 1.4033264033264032,
+      "grad_norm": 0.5522752500932042,
+      "learning_rate": 4.9503250941032944e-06,
+      "loss": 0.5858,
+      "step": 2700
+    },
+    {
+      "epoch": 1.405925155925156,
+      "grad_norm": 0.5236156885167432,
+      "learning_rate": 4.911225730339769e-06,
+      "loss": 0.5949,
+      "step": 2705
+    },
+    {
+      "epoch": 1.4085239085239085,
+      "grad_norm": 0.5167097373944585,
+      "learning_rate": 4.872231066704334e-06,
+      "loss": 0.5961,
+      "step": 2710
+    },
+    {
+      "epoch": 1.411122661122661,
+      "grad_norm": 0.563080292650056,
+      "learning_rate": 4.83334190550144e-06,
+      "loss": 0.5749,
+      "step": 2715
+    },
+    {
+      "epoch": 1.4137214137214138,
+      "grad_norm": 0.5327245199013042,
+      "learning_rate": 4.794559046864869e-06,
+      "loss": 0.5862,
+      "step": 2720
+    },
+    {
+      "epoch": 1.4163201663201663,
+      "grad_norm": 0.5289775705397598,
+      "learning_rate": 4.7558832887412445e-06,
+      "loss": 0.5877,
+      "step": 2725
+    },
+    {
+      "epoch": 1.4189189189189189,
+      "grad_norm": 0.5162142489701083,
+      "learning_rate": 4.717315426873631e-06,
+      "loss": 0.601,
+      "step": 2730
+    },
+    {
+      "epoch": 1.4215176715176714,
+      "grad_norm": 0.5405562294471293,
+      "learning_rate": 4.678856254785158e-06,
+      "loss": 0.6065,
+      "step": 2735
+    },
+    {
+      "epoch": 1.4241164241164241,
+      "grad_norm": 0.534465583683509,
+      "learning_rate": 4.6405065637626855e-06,
+      "loss": 0.5871,
+      "step": 2740
+    },
+    {
+      "epoch": 1.4267151767151767,
+      "grad_norm": 0.5179677159166826,
+      "learning_rate": 4.602267142840545e-06,
+      "loss": 0.5873,
+      "step": 2745
+    },
+    {
+      "epoch": 1.4293139293139294,
+      "grad_norm": 0.5480531904822096,
+      "learning_rate": 4.564138778784276e-06,
+      "loss": 0.5895,
+      "step": 2750
+    },
+    {
+      "epoch": 1.431912681912682,
+      "grad_norm": 0.5288015351179746,
+      "learning_rate": 4.526122256074468e-06,
+      "loss": 0.588,
+      "step": 2755
+    },
+    {
+      "epoch": 1.4345114345114345,
+      "grad_norm": 0.5232295069807825,
+      "learning_rate": 4.488218356890594e-06,
+      "loss": 0.5802,
+      "step": 2760
+    },
+    {
+      "epoch": 1.437110187110187,
+      "grad_norm": 0.5460332455155361,
+      "learning_rate": 4.450427861094933e-06,
+      "loss": 0.5809,
+      "step": 2765
+    },
+    {
+      "epoch": 1.4397089397089398,
+      "grad_norm": 0.5451685990728384,
+      "learning_rate": 4.412751546216521e-06,
+      "loss": 0.5879,
+      "step": 2770
+    },
+    {
+      "epoch": 1.4423076923076923,
+      "grad_norm": 0.5068743803171689,
+      "learning_rate": 4.375190187435151e-06,
+      "loss": 0.5819,
+      "step": 2775
+    },
+    {
+      "epoch": 1.444906444906445,
+      "grad_norm": 0.602224317860505,
+      "learning_rate": 4.3377445575654256e-06,
+      "loss": 0.5808,
+      "step": 2780
+    },
+    {
+      "epoch": 1.4475051975051976,
+      "grad_norm": 0.5584038619724856,
+      "learning_rate": 4.300415427040867e-06,
+      "loss": 0.586,
+      "step": 2785
+    },
+    {
+      "epoch": 1.45010395010395,
+      "grad_norm": 0.5402693196977059,
+      "learning_rate": 4.263203563898038e-06,
+      "loss": 0.5828,
+      "step": 2790
+    },
+    {
+      "epoch": 1.4527027027027026,
+      "grad_norm": 0.5393224079014185,
+      "learning_rate": 4.226109733760777e-06,
+      "loss": 0.5933,
+      "step": 2795
+    },
+    {
+      "epoch": 1.4553014553014554,
+      "grad_norm": 0.519264378128423,
+      "learning_rate": 4.1891346998244096e-06,
+      "loss": 0.5902,
+      "step": 2800
+    },
+    {
+      "epoch": 1.457900207900208,
+      "grad_norm": 0.5061001580587425,
+      "learning_rate": 4.15227922284007e-06,
+      "loss": 0.5965,
+      "step": 2805
+    },
+    {
+      "epoch": 1.4604989604989604,
+      "grad_norm": 0.5250324465767509,
+      "learning_rate": 4.11554406109904e-06,
+      "loss": 0.5805,
+      "step": 2810
+    },
+    {
+      "epoch": 1.4630977130977132,
+      "grad_norm": 0.5215691344217085,
+      "learning_rate": 4.07892997041714e-06,
+      "loss": 0.5847,
+      "step": 2815
+    },
+    {
+      "epoch": 1.4656964656964657,
+      "grad_norm": 0.5100852244380407,
+      "learning_rate": 4.042437704119207e-06,
+      "loss": 0.5863,
+      "step": 2820
+    },
+    {
+      "epoch": 1.4682952182952183,
+      "grad_norm": 0.5683610331681553,
+      "learning_rate": 4.006068013023556e-06,
+      "loss": 0.5916,
+      "step": 2825
+    },
+    {
+      "epoch": 1.4708939708939708,
+      "grad_norm": 0.5388381909277385,
+      "learning_rate": 3.969821645426559e-06,
+      "loss": 0.5879,
+      "step": 2830
+    },
+    {
+      "epoch": 1.4734927234927235,
+      "grad_norm": 0.5470869871655806,
+      "learning_rate": 3.933699347087239e-06,
+      "loss": 0.579,
+      "step": 2835
+    },
+    {
+      "epoch": 1.476091476091476,
+      "grad_norm": 0.5410808556361425,
+      "learning_rate": 3.897701861211924e-06,
+      "loss": 0.5882,
+      "step": 2840
+    },
+    {
+      "epoch": 1.4786902286902288,
+      "grad_norm": 0.5445488241193585,
+      "learning_rate": 3.861829928438975e-06,
+      "loss": 0.5745,
+      "step": 2845
+    },
+    {
+      "epoch": 1.4812889812889813,
+      "grad_norm": 0.5601094797596367,
+      "learning_rate": 3.8260842868235105e-06,
+      "loss": 0.5845,
+      "step": 2850
+    },
+    {
+      "epoch": 1.4838877338877339,
+      "grad_norm": 0.5108174266653619,
+      "learning_rate": 3.7904656718222653e-06,
+      "loss": 0.5817,
+      "step": 2855
+    },
+    {
+      "epoch": 1.4864864864864864,
+      "grad_norm": 0.5380258071504634,
+      "learning_rate": 3.7549748162784216e-06,
+      "loss": 0.5734,
+      "step": 2860
+    },
+    {
+      "epoch": 1.4890852390852392,
+      "grad_norm": 0.5056247324728291,
+      "learning_rate": 3.7196124504065512e-06,
+      "loss": 0.5837,
+      "step": 2865
+    },
+    {
+      "epoch": 1.4916839916839917,
+      "grad_norm": 0.5164676599197356,
+      "learning_rate": 3.684379301777585e-06,
+      "loss": 0.5737,
+      "step": 2870
+    },
+    {
+      "epoch": 1.4942827442827442,
+      "grad_norm": 0.5588163356512155,
+      "learning_rate": 3.649276095303843e-06,
+      "loss": 0.5896,
+      "step": 2875
+    },
+    {
+      "epoch": 1.496881496881497,
+      "grad_norm": 0.4785867733453047,
+      "learning_rate": 3.61430355322413e-06,
+      "loss": 0.5829,
+      "step": 2880
+    },
+    {
+      "epoch": 1.4994802494802495,
+      "grad_norm": 0.5236230514577889,
+      "learning_rate": 3.579462395088856e-06,
+      "loss": 0.5943,
+      "step": 2885
+    },
+    {
+      "epoch": 1.502079002079002,
+      "grad_norm": 0.5243126400259197,
+      "learning_rate": 3.544753337745249e-06,
+      "loss": 0.583,
+      "step": 2890
+    },
+    {
+      "epoch": 1.5046777546777546,
+      "grad_norm": 0.5400667626881714,
+      "learning_rate": 3.5101770953225932e-06,
+      "loss": 0.5868,
+      "step": 2895
+    },
+    {
+      "epoch": 1.5072765072765073,
+      "grad_norm": 0.513023426174068,
+      "learning_rate": 3.475734379217555e-06,
+      "loss": 0.5833,
+      "step": 2900
+    },
+    {
+      "epoch": 1.5098752598752598,
+      "grad_norm": 0.5157665498993713,
+      "learning_rate": 3.4414258980795202e-06,
+      "loss": 0.589,
+      "step": 2905
+    },
+    {
+      "epoch": 1.5124740124740126,
+      "grad_norm": 0.5357953624143235,
+      "learning_rate": 3.4072523577960304e-06,
+      "loss": 0.5964,
+      "step": 2910
+    },
+    {
+      "epoch": 1.5150727650727651,
+      "grad_norm": 0.5527070418499459,
+      "learning_rate": 3.3732144614782657e-06,
+      "loss": 0.5837,
+      "step": 2915
+    },
+    {
+      "epoch": 1.5176715176715176,
+      "grad_norm": 0.5458675642969252,
+      "learning_rate": 3.339312909446557e-06,
+      "loss": 0.5796,
+      "step": 2920
+    },
+    {
+      "epoch": 1.5202702702702702,
+      "grad_norm": 0.5152283307254787,
+      "learning_rate": 3.305548399215994e-06,
+      "loss": 0.5788,
+      "step": 2925
+    },
+    {
+      "epoch": 1.5228690228690227,
+      "grad_norm": 0.530799104763713,
+      "learning_rate": 3.2719216254820697e-06,
+      "loss": 0.5838,
+      "step": 2930
+    },
+    {
+      "epoch": 1.5254677754677755,
+      "grad_norm": 0.5403421099817592,
+      "learning_rate": 3.2384332801063846e-06,
+      "loss": 0.5656,
+      "step": 2935
+    },
+    {
+      "epoch": 1.5280665280665282,
+      "grad_norm": 0.558670507000079,
+      "learning_rate": 3.205084052102414e-06,
+      "loss": 0.588,
+      "step": 2940
+    },
+    {
+      "epoch": 1.5306652806652807,
+      "grad_norm": 0.5087922370395622,
+      "learning_rate": 3.1718746276213343e-06,
+      "loss": 0.5852,
+      "step": 2945
+    },
+    {
+      "epoch": 1.5332640332640333,
+      "grad_norm": 0.5482047516259143,
+      "learning_rate": 3.1388056899379036e-06,
+      "loss": 0.5946,
+      "step": 2950
+    },
+    {
+      "epoch": 1.5358627858627858,
+      "grad_norm": 0.5239979050184048,
+      "learning_rate": 3.105877919436401e-06,
+      "loss": 0.5961,
+      "step": 2955
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.6264561692888647,
+      "learning_rate": 3.0730919935966262e-06,
+      "loss": 0.5896,
+      "step": 2960
+    },
+    {
+      "epoch": 1.541060291060291,
+      "grad_norm": 0.5079701430841874,
+      "learning_rate": 3.0404485869799684e-06,
+      "loss": 0.5804,
+      "step": 2965
+    },
+    {
+      "epoch": 1.5436590436590436,
+      "grad_norm": 0.5227318565543926,
+      "learning_rate": 3.007948371215521e-06,
+      "loss": 0.5881,
+      "step": 2970
+    },
+    {
+      "epoch": 1.5462577962577964,
+      "grad_norm": 0.4870518164479936,
+      "learning_rate": 2.975592014986265e-06,
+      "loss": 0.5952,
+      "step": 2975
+    },
+    {
+      "epoch": 1.5488565488565489,
+      "grad_norm": 0.5650630749989504,
+      "learning_rate": 2.9433801840153166e-06,
+      "loss": 0.5836,
+      "step": 2980
+    },
+    {
+      "epoch": 1.5514553014553014,
+      "grad_norm": 0.5358798465667378,
+      "learning_rate": 2.9113135410522173e-06,
+      "loss": 0.5962,
+      "step": 2985
+    },
+    {
+      "epoch": 1.554054054054054,
+      "grad_norm": 0.5158817153128504,
+      "learning_rate": 2.8793927458593096e-06,
+      "loss": 0.57,
+      "step": 2990
+    },
+    {
+      "epoch": 1.5566528066528067,
+      "grad_norm": 0.5442534042057305,
+      "learning_rate": 2.8476184551981547e-06,
+      "loss": 0.5821,
+      "step": 2995
+    },
+    {
+      "epoch": 1.5592515592515592,
+      "grad_norm": 0.5778592324539092,
+      "learning_rate": 2.8159913228160242e-06,
+      "loss": 0.5862,
+      "step": 3000
+    },
+    {
+      "epoch": 1.561850311850312,
+      "grad_norm": 0.545713870580316,
+      "learning_rate": 2.7845119994324567e-06,
+      "loss": 0.5837,
+      "step": 3005
+    },
+    {
+      "epoch": 1.5644490644490645,
+      "grad_norm": 0.5027605038010509,
+      "learning_rate": 2.753181132725846e-06,
+      "loss": 0.5908,
+      "step": 3010
+    },
+    {
+      "epoch": 1.567047817047817,
+      "grad_norm": 0.5296784509548438,
+      "learning_rate": 2.7219993673201483e-06,
+      "loss": 0.5845,
+      "step": 3015
+    },
+    {
+      "epoch": 1.5696465696465696,
+      "grad_norm": 0.5434115009675748,
+      "learning_rate": 2.69096734477159e-06,
+      "loss": 0.5761,
+      "step": 3020
+    },
+    {
+      "epoch": 1.572245322245322,
+      "grad_norm": 0.5490780089111318,
+      "learning_rate": 2.6600857035554805e-06,
+      "loss": 0.5759,
+      "step": 3025
+    },
+    {
+      "epoch": 1.5748440748440748,
+      "grad_norm": 0.5434069633378591,
+      "learning_rate": 2.6293550790530776e-06,
+      "loss": 0.5856,
+      "step": 3030
+    },
+    {
+      "epoch": 1.5774428274428276,
+      "grad_norm": 0.5295051711364253,
+      "learning_rate": 2.5987761035385074e-06,
+      "loss": 0.5776,
+      "step": 3035
+    },
+    {
+      "epoch": 1.5800415800415801,
+      "grad_norm": 0.5294382951858779,
+      "learning_rate": 2.5683494061657698e-06,
+      "loss": 0.6003,
+      "step": 3040
+    },
+    {
+      "epoch": 1.5826403326403327,
+      "grad_norm": 0.5487264444738784,
+      "learning_rate": 2.538075612955775e-06,
+      "loss": 0.5905,
+      "step": 3045
+    },
+    {
+      "epoch": 1.5852390852390852,
+      "grad_norm": 0.5151850538678816,
+      "learning_rate": 2.5079553467834694e-06,
+      "loss": 0.5945,
+      "step": 3050
+    },
+    {
+      "epoch": 1.5878378378378377,
+      "grad_norm": 0.5241891607063889,
+      "learning_rate": 2.4779892273650363e-06,
+      "loss": 0.5809,
+      "step": 3055
+    },
+    {
+      "epoch": 1.5904365904365905,
+      "grad_norm": 0.5490608540434708,
+      "learning_rate": 2.44817787124512e-06,
+      "loss": 0.5864,
+      "step": 3060
+    },
+    {
+      "epoch": 1.593035343035343,
+      "grad_norm": 0.5085552474004695,
+      "learning_rate": 2.4185218917841557e-06,
+      "loss": 0.5839,
+      "step": 3065
+    },
+    {
+      "epoch": 1.5956340956340958,
+      "grad_norm": 0.5237874429368263,
+      "learning_rate": 2.389021899145745e-06,
+      "loss": 0.583,
+      "step": 3070
+    },
+    {
+      "epoch": 1.5982328482328483,
+      "grad_norm": 0.5462699535427128,
+      "learning_rate": 2.3596785002841126e-06,
+      "loss": 0.582,
+      "step": 3075
+    },
+    {
+      "epoch": 1.6008316008316008,
+      "grad_norm": 0.5305897465130409,
+      "learning_rate": 2.330492298931599e-06,
+      "loss": 0.588,
+      "step": 3080
+    },
+    {
+      "epoch": 1.6034303534303533,
+      "grad_norm": 0.5283228272742915,
+      "learning_rate": 2.3014638955862534e-06,
+      "loss": 0.5895,
+      "step": 3085
+    },
+    {
+      "epoch": 1.6060291060291059,
+      "grad_norm": 0.5375248398279345,
+      "learning_rate": 2.2725938874994733e-06,
+      "loss": 0.5712,
+      "step": 3090
+    },
+    {
+      "epoch": 1.6086278586278586,
+      "grad_norm": 0.5387607107517319,
+      "learning_rate": 2.2438828686637193e-06,
+      "loss": 0.5717,
+      "step": 3095
+    },
+    {
+      "epoch": 1.6112266112266114,
+      "grad_norm": 0.5333070714415671,
+      "learning_rate": 2.2153314298002883e-06,
+      "loss": 0.5869,
+      "step": 3100
+    },
+    {
+      "epoch": 1.613825363825364,
+      "grad_norm": 0.5130040335910431,
+      "learning_rate": 2.1869401583471674e-06,
+      "loss": 0.6017,
+      "step": 3105
+    },
+    {
+      "epoch": 1.6164241164241164,
+      "grad_norm": 0.5340427602820074,
+      "learning_rate": 2.1587096384469465e-06,
+      "loss": 0.5792,
+      "step": 3110
+    },
+    {
+      "epoch": 1.619022869022869,
+      "grad_norm": 0.5241334616500023,
+      "learning_rate": 2.130640450934787e-06,
+      "loss": 0.5766,
+      "step": 3115
+    },
+    {
+      "epoch": 1.6216216216216215,
+      "grad_norm": 0.5307440437206696,
+      "learning_rate": 2.1027331733264877e-06,
+      "loss": 0.5841,
+      "step": 3120
+    },
+    {
+      "epoch": 1.6242203742203742,
+      "grad_norm": 0.5111869577979938,
+      "learning_rate": 2.074988379806593e-06,
+      "loss": 0.5772,
+      "step": 3125
+    },
+    {
+      "epoch": 1.6268191268191268,
+      "grad_norm": 0.5232894395546558,
+      "learning_rate": 2.047406641216583e-06,
+      "loss": 0.5798,
+      "step": 3130
+    },
+    {
+      "epoch": 1.6294178794178795,
+      "grad_norm": 0.5427662431229768,
+      "learning_rate": 2.019988525043125e-06,
+      "loss": 0.5852,
+      "step": 3135
+    },
+    {
+      "epoch": 1.632016632016632,
+      "grad_norm": 0.50753581880278,
+      "learning_rate": 1.992734595406408e-06,
+      "loss": 0.5756,
+      "step": 3140
+    },
+    {
+      "epoch": 1.6346153846153846,
+      "grad_norm": 0.5026065449648726,
+      "learning_rate": 1.965645413048519e-06,
+      "loss": 0.5756,
+      "step": 3145
+    },
+    {
+      "epoch": 1.637214137214137,
+      "grad_norm": 0.5135278822828021,
+      "learning_rate": 1.9387215353219213e-06,
+      "loss": 0.6112,
+      "step": 3150
+    },
+    {
+      "epoch": 1.6398128898128899,
+      "grad_norm": 0.5576890383464939,
+      "learning_rate": 1.9119635161779738e-06,
+      "loss": 0.5933,
+      "step": 3155
+    },
+    {
+      "epoch": 1.6424116424116424,
+      "grad_norm": 0.5005852274685088,
+      "learning_rate": 1.885371906155552e-06,
+      "loss": 0.575,
+      "step": 3160
+    },
+    {
+      "epoch": 1.6450103950103951,
+      "grad_norm": 0.552519914509362,
+      "learning_rate": 1.8589472523697017e-06,
+      "loss": 0.5832,
+      "step": 3165
+    },
+    {
+      "epoch": 1.6476091476091477,
+      "grad_norm": 0.5313038853576302,
+      "learning_rate": 1.832690098500386e-06,
+      "loss": 0.5799,
+      "step": 3170
+    },
+    {
+      "epoch": 1.6502079002079002,
+      "grad_norm": 0.5307430737754275,
+      "learning_rate": 1.8066009847813171e-06,
+      "loss": 0.5933,
+      "step": 3175
+    },
+    {
+      "epoch": 1.6528066528066527,
+      "grad_norm": 0.5092540679964042,
+      "learning_rate": 1.7806804479888151e-06,
+      "loss": 0.5857,
+      "step": 3180
+    },
+    {
+      "epoch": 1.6554054054054053,
+      "grad_norm": 0.527226302557591,
+      "learning_rate": 1.7549290214307824e-06,
+      "loss": 0.5826,
+      "step": 3185
+    },
+    {
+      "epoch": 1.658004158004158,
+      "grad_norm": 0.5117048033630656,
+      "learning_rate": 1.729347234935721e-06,
+      "loss": 0.5866,
+      "step": 3190
+    },
+    {
+      "epoch": 1.6606029106029108,
+      "grad_norm": 0.5338263684846878,
+      "learning_rate": 1.7039356148418363e-06,
+      "loss": 0.5815,
+      "step": 3195
+    },
+    {
+      "epoch": 1.6632016632016633,
+      "grad_norm": 0.514319228309042,
+      "learning_rate": 1.6786946839862127e-06,
+      "loss": 0.5704,
+      "step": 3200
+    },
+    {
+      "epoch": 1.6658004158004158,
+      "grad_norm": 0.5093476576764645,
+      "learning_rate": 1.6536249616940414e-06,
+      "loss": 0.5861,
+      "step": 3205
+    },
+    {
+      "epoch": 1.6683991683991684,
+      "grad_norm": 0.507273641369432,
+      "learning_rate": 1.6287269637679538e-06,
+      "loss": 0.5827,
+      "step": 3210
+    },
+    {
+      "epoch": 1.6709979209979209,
+      "grad_norm": 0.5145115612127696,
+      "learning_rate": 1.604001202477391e-06,
+      "loss": 0.5768,
+      "step": 3215
+    },
+    {
+      "epoch": 1.6735966735966736,
+      "grad_norm": 0.5149567719955702,
+      "learning_rate": 1.5794481865480805e-06,
+      "loss": 0.581,
+      "step": 3220
+    },
+    {
+      "epoch": 1.6761954261954262,
+      "grad_norm": 0.5471640458477693,
+      "learning_rate": 1.5550684211515544e-06,
+      "loss": 0.6,
+      "step": 3225
+    },
+    {
+      "epoch": 1.678794178794179,
+      "grad_norm": 0.5394179199712156,
+      "learning_rate": 1.5308624078947666e-06,
+      "loss": 0.5801,
+      "step": 3230
+    },
+    {
+      "epoch": 1.6813929313929314,
+      "grad_norm": 0.506276316904012,
+      "learning_rate": 1.5068306448097714e-06,
+      "loss": 0.5689,
+      "step": 3235
+    },
+    {
+      "epoch": 1.683991683991684,
+      "grad_norm": 0.5038655785707594,
+      "learning_rate": 1.4829736263434702e-06,
+      "loss": 0.5907,
+      "step": 3240
+    },
+    {
+      "epoch": 1.6865904365904365,
+      "grad_norm": 0.5272673427698651,
+      "learning_rate": 1.4592918433474424e-06,
+      "loss": 0.586,
+      "step": 3245
+    },
+    {
+      "epoch": 1.689189189189189,
+      "grad_norm": 0.5298471417100465,
+      "learning_rate": 1.4357857830678445e-06,
+      "loss": 0.5937,
+      "step": 3250
+    },
+    {
+      "epoch": 1.6917879417879418,
+      "grad_norm": 0.5181589625133596,
+      "learning_rate": 1.4124559291353878e-06,
+      "loss": 0.5679,
+      "step": 3255
+    },
+    {
+      "epoch": 1.6943866943866945,
+      "grad_norm": 0.5012265675078679,
+      "learning_rate": 1.3893027615553922e-06,
+      "loss": 0.5657,
+      "step": 3260
+    },
+    {
+      "epoch": 1.696985446985447,
+      "grad_norm": 0.5568040197261597,
+      "learning_rate": 1.366326756697892e-06,
+      "loss": 0.5961,
+      "step": 3265
+    },
+    {
+      "epoch": 1.6995841995841996,
+      "grad_norm": 0.5718957039548288,
+      "learning_rate": 1.3435283872878623e-06,
+      "loss": 0.5787,
+      "step": 3270
+    },
+    {
+      "epoch": 1.7021829521829521,
+      "grad_norm": 0.5432446627166708,
+      "learning_rate": 1.3209081223954678e-06,
+      "loss": 0.5808,
+      "step": 3275
+    },
+    {
+      "epoch": 1.7047817047817047,
+      "grad_norm": 0.5392794940489283,
+      "learning_rate": 1.2984664274264214e-06,
+      "loss": 0.571,
+      "step": 3280
+    },
+    {
+      "epoch": 1.7073804573804574,
+      "grad_norm": 0.5619472295000788,
+      "learning_rate": 1.2762037641124147e-06,
+      "loss": 0.5752,
+      "step": 3285
+    },
+    {
+      "epoch": 1.70997920997921,
+      "grad_norm": 0.5144706465312519,
+      "learning_rate": 1.2541205905016097e-06,
+      "loss": 0.5831,
+      "step": 3290
+    },
+    {
+      "epoch": 1.7125779625779627,
+      "grad_norm": 0.5359125394441557,
+      "learning_rate": 1.2322173609492138e-06,
+      "loss": 0.571,
+      "step": 3295
+    },
+    {
+      "epoch": 1.7151767151767152,
+      "grad_norm": 0.5219188144811212,
+      "learning_rate": 1.2104945261081425e-06,
+      "loss": 0.5866,
+      "step": 3300
+    },
+    {
+      "epoch": 1.7177754677754677,
+      "grad_norm": 0.48682253952056287,
+      "learning_rate": 1.1889525329197337e-06,
+      "loss": 0.5659,
+      "step": 3305
+    },
+    {
+      "epoch": 1.7203742203742203,
+      "grad_norm": 0.552850063663083,
+      "learning_rate": 1.1675918246045592e-06,
+      "loss": 0.5892,
+      "step": 3310
+    },
+    {
+      "epoch": 1.722972972972973,
+      "grad_norm": 0.5129345549943813,
+      "learning_rate": 1.1464128406533082e-06,
+      "loss": 0.5726,
+      "step": 3315
+    },
+    {
+      "epoch": 1.7255717255717256,
+      "grad_norm": 0.5215339840888276,
+      "learning_rate": 1.1254160168177363e-06,
+      "loss": 0.5755,
+      "step": 3320
+    },
+    {
+      "epoch": 1.7281704781704783,
+      "grad_norm": 0.5346007889564445,
+      "learning_rate": 1.1046017851017077e-06,
+      "loss": 0.5934,
+      "step": 3325
+    },
+    {
+      "epoch": 1.7307692307692308,
+      "grad_norm": 0.5241493511931715,
+      "learning_rate": 1.083970573752301e-06,
+      "loss": 0.5687,
+      "step": 3330
+    },
+    {
+      "epoch": 1.7333679833679834,
+      "grad_norm": 0.5144495715767252,
+      "learning_rate": 1.0635228072510084e-06,
+      "loss": 0.5717,
+      "step": 3335
+    },
+    {
+      "epoch": 1.735966735966736,
+      "grad_norm": 0.555636699696863,
+      "learning_rate": 1.043258906304987e-06,
+      "loss": 0.5717,
+      "step": 3340
+    },
+    {
+      "epoch": 1.7385654885654884,
+      "grad_norm": 0.5398880461538901,
+      "learning_rate": 1.023179287838414e-06,
+      "loss": 0.5904,
+      "step": 3345
+    },
+    {
+      "epoch": 1.7411642411642412,
+      "grad_norm": 0.5443352220533247,
+      "learning_rate": 1.0032843649839052e-06,
+      "loss": 0.5633,
+      "step": 3350
+    },
+    {
+      "epoch": 1.743762993762994,
+      "grad_norm": 0.5176425518248007,
+      "learning_rate": 9.835745470740143e-07,
+      "loss": 0.5659,
+      "step": 3355
+    },
+    {
+      "epoch": 1.7463617463617465,
+      "grad_norm": 0.5093779378294409,
+      "learning_rate": 9.640502396328133e-07,
+      "loss": 0.5727,
+      "step": 3360
+    },
+    {
+      "epoch": 1.748960498960499,
+      "grad_norm": 0.5447675578056951,
+      "learning_rate": 9.447118443675496e-07,
+      "loss": 0.57,
+      "step": 3365
+    },
+    {
+      "epoch": 1.7515592515592515,
+      "grad_norm": 0.5213010548532966,
+      "learning_rate": 9.255597591603727e-07,
+      "loss": 0.5714,
+      "step": 3370
+    },
+    {
+      "epoch": 1.754158004158004,
+      "grad_norm": 0.5398854243774627,
+      "learning_rate": 9.065943780601583e-07,
+      "loss": 0.5811,
+      "step": 3375
+    },
+    {
+      "epoch": 1.7567567567567568,
+      "grad_norm": 0.5056718137516323,
+      "learning_rate": 8.87816091274396e-07,
+      "loss": 0.5674,
+      "step": 3380
+    },
+    {
+      "epoch": 1.7593555093555093,
+      "grad_norm": 0.4990869376715757,
+      "learning_rate": 8.692252851611594e-07,
+      "loss": 0.5813,
+      "step": 3385
+    },
+    {
+      "epoch": 1.761954261954262,
+      "grad_norm": 0.53552516213898,
+      "learning_rate": 8.508223422211593e-07,
+      "loss": 0.5814,
+      "step": 3390
+    },
+    {
+      "epoch": 1.7645530145530146,
+      "grad_norm": 0.5060595322478018,
+      "learning_rate": 8.326076410898798e-07,
+      "loss": 0.5653,
+      "step": 3395
+    },
+    {
+      "epoch": 1.7671517671517671,
+      "grad_norm": 0.5333772606550752,
+      "learning_rate": 8.145815565297721e-07,
+      "loss": 0.575,
+      "step": 3400
+    },
+    {
+      "epoch": 1.7697505197505197,
+      "grad_norm": 0.5193976811229921,
+      "learning_rate": 7.967444594225604e-07,
+      "loss": 0.5821,
+      "step": 3405
+    },
+    {
+      "epoch": 1.7723492723492722,
+      "grad_norm": 0.4986756303541811,
+      "learning_rate": 7.790967167615982e-07,
+      "loss": 0.5878,
+      "step": 3410
+    },
+    {
+      "epoch": 1.774948024948025,
+      "grad_norm": 0.5443836974981098,
+      "learning_rate": 7.616386916443308e-07,
+      "loss": 0.5788,
+      "step": 3415
+    },
+    {
+      "epoch": 1.7775467775467777,
+      "grad_norm": 0.5018572354435884,
+      "learning_rate": 7.443707432648117e-07,
+      "loss": 0.5757,
+      "step": 3420
+    },
+    {
+      "epoch": 1.7801455301455302,
+      "grad_norm": 0.5221278089522104,
+      "learning_rate": 7.272932269063182e-07,
+      "loss": 0.5804,
+      "step": 3425
+    },
+    {
+      "epoch": 1.7827442827442828,
+      "grad_norm": 0.5309066311243344,
+      "learning_rate": 7.10406493934046e-07,
+      "loss": 0.5725,
+      "step": 3430
+    },
+    {
+      "epoch": 1.7853430353430353,
+      "grad_norm": 0.523703068520167,
+      "learning_rate": 6.937108917878699e-07,
+      "loss": 0.5851,
+      "step": 3435
+    },
+    {
+      "epoch": 1.7879417879417878,
+      "grad_norm": 0.4990677586423772,
+      "learning_rate": 6.772067639752023e-07,
+      "loss": 0.5738,
+      "step": 3440
+    },
+    {
+      "epoch": 1.7905405405405406,
+      "grad_norm": 0.5118619723842658,
+      "learning_rate": 6.608944500639247e-07,
+      "loss": 0.5759,
+      "step": 3445
+    },
+    {
+      "epoch": 1.793139293139293,
+      "grad_norm": 0.5335785839494404,
+      "learning_rate": 6.447742856754014e-07,
+      "loss": 0.5765,
+      "step": 3450
+    },
+    {
+      "epoch": 1.7957380457380459,
+      "grad_norm": 0.5153869448178949,
+      "learning_rate": 6.288466024775674e-07,
+      "loss": 0.5696,
+      "step": 3455
+    },
+    {
+      "epoch": 1.7983367983367984,
+      "grad_norm": 0.48901624634329477,
+      "learning_rate": 6.131117281781196e-07,
+      "loss": 0.5674,
+      "step": 3460
+    },
+    {
+      "epoch": 1.800935550935551,
+      "grad_norm": 0.5102458462355101,
+      "learning_rate": 5.975699865177575e-07,
+      "loss": 0.5567,
+      "step": 3465
+    },
+    {
+      "epoch": 1.8035343035343034,
+      "grad_norm": 0.5723265636044084,
+      "learning_rate": 5.822216972635331e-07,
+      "loss": 0.5926,
+      "step": 3470
+    },
+    {
+      "epoch": 1.806133056133056,
+      "grad_norm": 0.5094713754422882,
+      "learning_rate": 5.670671762022684e-07,
+      "loss": 0.5655,
+      "step": 3475
+    },
+    {
+      "epoch": 1.8087318087318087,
+      "grad_norm": 0.5039283824986392,
+      "learning_rate": 5.52106735134057e-07,
+      "loss": 0.5804,
+      "step": 3480
+    },
+    {
+      "epoch": 1.8113305613305615,
+      "grad_norm": 0.5144912329273745,
+      "learning_rate": 5.3734068186585e-07,
+      "loss": 0.5836,
+      "step": 3485
+    },
+    {
+      "epoch": 1.813929313929314,
+      "grad_norm": 0.5153264587268782,
+      "learning_rate": 5.227693202051232e-07,
+      "loss": 0.5847,
+      "step": 3490
+    },
+    {
+      "epoch": 1.8165280665280665,
+      "grad_norm": 0.5415875159805758,
+      "learning_rate": 5.083929499536289e-07,
+      "loss": 0.577,
+      "step": 3495
+    },
+    {
+      "epoch": 1.819126819126819,
+      "grad_norm": 0.5049450382278391,
+      "learning_rate": 4.942118669012208e-07,
+      "loss": 0.5822,
+      "step": 3500
+    },
+    {
+      "epoch": 1.8217255717255716,
+      "grad_norm": 0.5051461545256577,
+      "learning_rate": 4.802263628197756e-07,
+      "loss": 0.5644,
+      "step": 3505
+    },
+    {
+      "epoch": 1.8243243243243243,
+      "grad_norm": 0.49660831452275783,
+      "learning_rate": 4.664367254571855e-07,
+      "loss": 0.5709,
+      "step": 3510
+    },
+    {
+      "epoch": 1.8269230769230769,
+      "grad_norm": 0.49492888592400675,
+      "learning_rate": 4.528432385314352e-07,
+      "loss": 0.5813,
+      "step": 3515
+    },
+    {
+      "epoch": 1.8295218295218296,
+      "grad_norm": 0.4865514561195398,
+      "learning_rate": 4.394461817247753e-07,
+      "loss": 0.5686,
+      "step": 3520
+    },
+    {
+      "epoch": 1.8321205821205822,
+      "grad_norm": 0.5225610714631679,
+      "learning_rate": 4.262458306779571e-07,
+      "loss": 0.5806,
+      "step": 3525
+    },
+    {
+      "epoch": 1.8347193347193347,
+      "grad_norm": 0.5340600886285485,
+      "learning_rate": 4.13242456984565e-07,
+      "loss": 0.566,
+      "step": 3530
+    },
+    {
+      "epoch": 1.8373180873180872,
+      "grad_norm": 0.5237864452094557,
+      "learning_rate": 4.0043632818542845e-07,
+      "loss": 0.576,
+      "step": 3535
+    },
+    {
+      "epoch": 1.83991683991684,
+      "grad_norm": 0.5512703745424207,
+      "learning_rate": 3.878277077631176e-07,
+      "loss": 0.5786,
+      "step": 3540
+    },
+    {
+      "epoch": 1.8425155925155925,
+      "grad_norm": 0.47905215589036726,
+      "learning_rate": 3.754168551365234e-07,
+      "loss": 0.5772,
+      "step": 3545
+    },
+    {
+      "epoch": 1.8451143451143452,
+      "grad_norm": 0.549871032924406,
+      "learning_rate": 3.6320402565551585e-07,
+      "loss": 0.5779,
+      "step": 3550
+    },
+    {
+      "epoch": 1.8477130977130978,
+      "grad_norm": 0.5305355260984184,
+      "learning_rate": 3.5118947059569974e-07,
+      "loss": 0.5814,
+      "step": 3555
+    },
+    {
+      "epoch": 1.8503118503118503,
+      "grad_norm": 0.5361560006783601,
+      "learning_rate": 3.393734371532309e-07,
+      "loss": 0.5668,
+      "step": 3560
+    },
+    {
+      "epoch": 1.8529106029106028,
+      "grad_norm": 0.5205186123699476,
+      "learning_rate": 3.277561684397412e-07,
+      "loss": 0.5921,
+      "step": 3565
+    },
+    {
+      "epoch": 1.8555093555093554,
+      "grad_norm": 0.523345976738322,
+      "learning_rate": 3.163379034773328e-07,
+      "loss": 0.5889,
+      "step": 3570
+    },
+    {
+      "epoch": 1.8581081081081081,
+      "grad_norm": 0.5080494795310655,
+      "learning_rate": 3.0511887719366193e-07,
+      "loss": 0.5707,
+      "step": 3575
+    },
+    {
+      "epoch": 1.8607068607068609,
+      "grad_norm": 0.5483892874723801,
+      "learning_rate": 2.9409932041710056e-07,
+      "loss": 0.5867,
+      "step": 3580
+    },
+    {
+      "epoch": 1.8633056133056134,
+      "grad_norm": 0.5028157875697813,
+      "learning_rate": 2.8327945987199234e-07,
+      "loss": 0.5735,
+      "step": 3585
+    },
+    {
+      "epoch": 1.865904365904366,
+      "grad_norm": 0.542730797020021,
+      "learning_rate": 2.726595181739866e-07,
+      "loss": 0.5607,
+      "step": 3590
+    },
+    {
+      "epoch": 1.8685031185031185,
+      "grad_norm": 0.5229545949985743,
+      "learning_rate": 2.6223971382545623e-07,
+      "loss": 0.5805,
+      "step": 3595
+    },
+    {
+      "epoch": 1.871101871101871,
+      "grad_norm": 0.528588646817702,
+      "learning_rate": 2.5202026121100453e-07,
+      "loss": 0.5809,
+      "step": 3600
+    },
+    {
+      "epoch": 1.8737006237006237,
+      "grad_norm": 0.5053433572695779,
+      "learning_rate": 2.420013705930524e-07,
+      "loss": 0.5695,
+      "step": 3605
+    },
+    {
+      "epoch": 1.8762993762993763,
+      "grad_norm": 0.6061825606893263,
+      "learning_rate": 2.321832481075137e-07,
+      "loss": 0.5821,
+      "step": 3610
+    },
+    {
+      "epoch": 1.878898128898129,
+      "grad_norm": 0.49099322033756715,
+      "learning_rate": 2.225660957595499e-07,
+      "loss": 0.5721,
+      "step": 3615
+    },
+    {
+      "epoch": 1.8814968814968815,
+      "grad_norm": 0.5193378188817532,
+      "learning_rate": 2.131501114194212e-07,
+      "loss": 0.568,
+      "step": 3620
+    },
+    {
+      "epoch": 1.884095634095634,
+      "grad_norm": 0.5088241393434693,
+      "learning_rate": 2.0393548881841308e-07,
+      "loss": 0.5937,
+      "step": 3625
+    },
+    {
+      "epoch": 1.8866943866943866,
+      "grad_norm": 0.5249257025900219,
+      "learning_rate": 1.9492241754484275e-07,
+      "loss": 0.5696,
+      "step": 3630
+    },
+    {
+      "epoch": 1.8892931392931391,
+      "grad_norm": 0.5063117403822439,
+      "learning_rate": 1.861110830401691e-07,
+      "loss": 0.5776,
+      "step": 3635
+    },
+    {
+      "epoch": 1.8918918918918919,
+      "grad_norm": 0.5247823235703282,
+      "learning_rate": 1.7750166659517009e-07,
+      "loss": 0.5805,
+      "step": 3640
+    },
+    {
+      "epoch": 1.8944906444906446,
+      "grad_norm": 0.530597215782001,
+      "learning_rate": 1.6909434534621683e-07,
+      "loss": 0.5833,
+      "step": 3645
+    },
+    {
+      "epoch": 1.8970893970893972,
+      "grad_norm": 0.535514568090153,
+      "learning_rate": 1.6088929227162542e-07,
+      "loss": 0.5722,
+      "step": 3650
+    },
+    {
+      "epoch": 1.8996881496881497,
+      "grad_norm": 0.5247060245361456,
+      "learning_rate": 1.5288667618810426e-07,
+      "loss": 0.5761,
+      "step": 3655
+    },
+    {
+      "epoch": 1.9022869022869022,
+      "grad_norm": 0.5199277346646484,
+      "learning_rate": 1.4508666174727237e-07,
+      "loss": 0.5625,
+      "step": 3660
+    },
+    {
+      "epoch": 1.9048856548856548,
+      "grad_norm": 0.5016083095240303,
+      "learning_rate": 1.374894094322765e-07,
+      "loss": 0.5754,
+      "step": 3665
+    },
+    {
+      "epoch": 1.9074844074844075,
+      "grad_norm": 0.7501294618373303,
+      "learning_rate": 1.300950755544894e-07,
+      "loss": 0.5914,
+      "step": 3670
+    },
+    {
+      "epoch": 1.91008316008316,
+      "grad_norm": 0.5599461348657118,
+      "learning_rate": 1.2290381225029345e-07,
+      "loss": 0.5661,
+      "step": 3675
+    },
+    {
+      "epoch": 1.9126819126819128,
+      "grad_norm": 0.5324363544661017,
+      "learning_rate": 1.1591576747794875e-07,
+      "loss": 0.5734,
+      "step": 3680
+    },
+    {
+      "epoch": 1.9152806652806653,
+      "grad_norm": 0.4949940507745294,
+      "learning_rate": 1.0913108501454994e-07,
+      "loss": 0.5691,
+      "step": 3685
+    },
+    {
+      "epoch": 1.9178794178794178,
+      "grad_norm": 0.5201949950592929,
+      "learning_rate": 1.0254990445306978e-07,
+      "loss": 0.5702,
+      "step": 3690
+    },
+    {
+      "epoch": 1.9204781704781704,
+      "grad_norm": 0.5435052866885572,
+      "learning_rate": 9.617236119948359e-08,
+      "loss": 0.5737,
+      "step": 3695
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.5284636555155561,
+      "learning_rate": 8.999858646998705e-08,
+      "loss": 0.5691,
+      "step": 3700
+    },
+    {
+      "epoch": 1.9256756756756757,
+      "grad_norm": 0.48133491617434554,
+      "learning_rate": 8.402870728829283e-08,
+      "loss": 0.5679,
+      "step": 3705
+    },
+    {
+      "epoch": 1.9282744282744284,
+      "grad_norm": 0.525724615350644,
+      "learning_rate": 7.82628464830193e-08,
+      "loss": 0.5753,
+      "step": 3710
+    },
+    {
+      "epoch": 1.930873180873181,
+      "grad_norm": 0.5316420453721764,
+      "learning_rate": 7.270112268516482e-08,
+      "loss": 0.5769,
+      "step": 3715
+    },
+    {
+      "epoch": 1.9334719334719335,
+      "grad_norm": 0.5256504385548955,
+      "learning_rate": 6.734365032566192e-08,
+      "loss": 0.5818,
+      "step": 3720
+    },
+    {
+      "epoch": 1.936070686070686,
+      "grad_norm": 0.5309069869526528,
+      "learning_rate": 6.219053963303023e-08,
+      "loss": 0.5734,
+      "step": 3725
+    },
+    {
+      "epoch": 1.9386694386694385,
+      "grad_norm": 0.5354454111890438,
+      "learning_rate": 5.724189663110058e-08,
+      "loss": 0.5809,
+      "step": 3730
+    },
+    {
+      "epoch": 1.9412681912681913,
+      "grad_norm": 0.5053818186885027,
+      "learning_rate": 5.249782313683893e-08,
+      "loss": 0.5815,
+      "step": 3735
+    },
+    {
+      "epoch": 1.943866943866944,
+      "grad_norm": 0.5087952292369419,
+      "learning_rate": 4.7958416758249196e-08,
+      "loss": 0.572,
+      "step": 3740
+    },
+    {
+      "epoch": 1.9464656964656966,
+      "grad_norm": 0.5399877020308321,
+      "learning_rate": 4.3623770892368136e-08,
+      "loss": 0.5745,
+      "step": 3745
+    },
+    {
+      "epoch": 1.949064449064449,
+      "grad_norm": 0.4970333597526285,
+      "learning_rate": 3.949397472333805e-08,
+      "loss": 0.5677,
+      "step": 3750
+    },
+    {
+      "epoch": 1.9516632016632016,
+      "grad_norm": 0.51125198825636,
+      "learning_rate": 3.556911322057821e-08,
+      "loss": 0.5807,
+      "step": 3755
+    },
+    {
+      "epoch": 1.9542619542619541,
+      "grad_norm": 0.5453706817478942,
+      "learning_rate": 3.184926713703185e-08,
+      "loss": 0.5696,
+      "step": 3760
+    },
+    {
+      "epoch": 1.956860706860707,
+      "grad_norm": 0.5259685279196025,
+      "learning_rate": 2.8334513007507447e-08,
+      "loss": 0.5874,
+      "step": 3765
+    },
+    {
+      "epoch": 1.9594594594594594,
+      "grad_norm": 0.5299460024442244,
+      "learning_rate": 2.5024923147101144e-08,
+      "loss": 0.5802,
+      "step": 3770
+    },
+    {
+      "epoch": 1.9620582120582122,
+      "grad_norm": 0.5085531168100673,
+      "learning_rate": 2.1920565649713454e-08,
+      "loss": 0.5633,
+      "step": 3775
+    },
+    {
+      "epoch": 1.9646569646569647,
+      "grad_norm": 0.5066062776158768,
+      "learning_rate": 1.902150438664374e-08,
+      "loss": 0.5815,
+      "step": 3780
+    },
+    {
+      "epoch": 1.9672557172557172,
+      "grad_norm": 0.5304292915843402,
+      "learning_rate": 1.6327799005277923e-08,
+      "loss": 0.5818,
+      "step": 3785
+    },
+    {
+      "epoch": 1.9698544698544698,
+      "grad_norm": 0.5429547592576485,
+      "learning_rate": 1.3839504927863901e-08,
+      "loss": 0.5866,
+      "step": 3790
+    },
+    {
+      "epoch": 1.9724532224532223,
+      "grad_norm": 0.5025697826189373,
+      "learning_rate": 1.15566733503647e-08,
+      "loss": 0.5805,
+      "step": 3795
+    },
+    {
+      "epoch": 1.975051975051975,
+      "grad_norm": 0.5279650041637034,
+      "learning_rate": 9.479351241410418e-09,
+      "loss": 0.5664,
+      "step": 3800
+    },
+    {
+      "epoch": 1.9776507276507278,
+      "grad_norm": 0.5231820617917791,
+      "learning_rate": 7.607581341332326e-09,
+      "loss": 0.5712,
+      "step": 3805
+    },
+    {
+      "epoch": 1.9802494802494803,
+      "grad_norm": 0.5278123420036012,
+      "learning_rate": 5.941402161279142e-09,
+      "loss": 0.5737,
+      "step": 3810
+    },
+    {
+      "epoch": 1.9828482328482329,
+      "grad_norm": 0.5309246932396601,
+      "learning_rate": 4.48084798242987e-09,
+      "loss": 0.5817,
+      "step": 3815
+    },
+    {
+      "epoch": 1.9854469854469854,
+      "grad_norm": 0.5575208848453578,
+      "learning_rate": 3.2259488552832676e-09,
+      "loss": 0.5787,
+      "step": 3820
+    },
+    {
+      "epoch": 1.988045738045738,
+      "grad_norm": 0.512653822002805,
+      "learning_rate": 2.1767305990416653e-09,
+      "loss": 0.5848,
+      "step": 3825
+    },
+    {
+      "epoch": 1.9906444906444907,
+      "grad_norm": 0.5291714453451785,
+      "learning_rate": 1.3332148010836154e-09,
+      "loss": 0.5831,
+      "step": 3830
+    },
+    {
+      "epoch": 1.9932432432432432,
+      "grad_norm": 0.5268775795002444,
+      "learning_rate": 6.954188165131382e-10,
+      "loss": 0.5895,
+      "step": 3835
+    },
+    {
+      "epoch": 1.995841995841996,
+      "grad_norm": 0.5255058213556102,
+      "learning_rate": 2.6335576780667227e-10,
+      "loss": 0.5759,
+      "step": 3840
+    },
+    {
+      "epoch": 1.9984407484407485,
+      "grad_norm": 0.5086937679965747,
+      "learning_rate": 3.7034544543290694e-11,
+      "loss": 0.5727,
+      "step": 3845
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.7488190531730652,
+      "eval_runtime": 104.7896,
+      "eval_samples_per_second": 78.347,
+      "eval_steps_per_second": 1.231,
+      "step": 3848
+    },
+    {
+      "epoch": 2.0,
+      "step": 3848,
+      "total_flos": 805692915056640.0,
+      "train_loss": 0.6651840950991657,
+      "train_runtime": 11913.1492,
+      "train_samples_per_second": 20.671,
+      "train_steps_per_second": 0.323
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 3848,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 805692915056640.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}