diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5945 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 4200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0004761904761904762,
+      "grad_norm": 40.45813118239569,
+      "learning_rate": 4.7619047619047627e-08,
+      "loss": 2.1387,
+      "step": 1
+    },
+    {
+      "epoch": 0.002380952380952381,
+      "grad_norm": 39.63780477662549,
+      "learning_rate": 2.3809523809523811e-07,
+      "loss": 2.0871,
+      "step": 5
+    },
+    {
+      "epoch": 0.004761904761904762,
+      "grad_norm": 43.499844517947736,
+      "learning_rate": 4.7619047619047623e-07,
+      "loss": 2.0652,
+      "step": 10
+    },
+    {
+      "epoch": 0.007142857142857143,
+      "grad_norm": 10.896090871742048,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 1.9004,
+      "step": 15
+    },
+    {
+      "epoch": 0.009523809523809525,
+      "grad_norm": 6.2002952611415845,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 1.7859,
+      "step": 20
+    },
+    {
+      "epoch": 0.011904761904761904,
+      "grad_norm": 4.97841465233645,
+      "learning_rate": 1.1904761904761906e-06,
+      "loss": 1.6739,
+      "step": 25
+    },
+    {
+      "epoch": 0.014285714285714285,
+      "grad_norm": 4.293746378401411,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.4907,
+      "step": 30
+    },
+    {
+      "epoch": 0.016666666666666666,
+      "grad_norm": 2.299524492927895,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.3909,
+      "step": 35
+    },
+    {
+      "epoch": 0.01904761904761905,
+      "grad_norm": 2.3149107855519064,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 1.2509,
+      "step": 40
+    },
+    {
+      "epoch": 0.02142857142857143,
+      "grad_norm": 2.251875691331579,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 1.1418,
+      "step": 45
+    },
+    {
+      "epoch": 0.023809523809523808,
+      "grad_norm": 1.293593518327437,
+      "learning_rate": 2.380952380952381e-06,
+      "loss": 1.0812,
+      "step": 50
+    },
+    {
+      "epoch": 0.02619047619047619,
+      "grad_norm": 1.1337840513174344,
+      "learning_rate": 2.6190476190476192e-06,
+      "loss": 1.0113,
+      "step": 55
+    },
+    {
+      "epoch": 0.02857142857142857,
+      "grad_norm": 1.1685369096623774,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.9634,
+      "step": 60
+    },
+    {
+      "epoch": 0.030952380952380953,
+      "grad_norm": 1.070331563673874,
+      "learning_rate": 3.0952380952380957e-06,
+      "loss": 0.9291,
+      "step": 65
+    },
+    {
+      "epoch": 0.03333333333333333,
+      "grad_norm": 1.059586065170695,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.91,
+      "step": 70
+    },
+    {
+      "epoch": 0.03571428571428571,
+      "grad_norm": 1.131151685372533,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.8976,
+      "step": 75
+    },
+    {
+      "epoch": 0.0380952380952381,
+      "grad_norm": 1.193443615687326,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 0.8781,
+      "step": 80
+    },
+    {
+      "epoch": 0.04047619047619048,
+      "grad_norm": 1.0875818440706093,
+      "learning_rate": 4.047619047619048e-06,
+      "loss": 0.8631,
+      "step": 85
+    },
+    {
+      "epoch": 0.04285714285714286,
+      "grad_norm": 1.1503140744906168,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.8646,
+      "step": 90
+    },
+    {
+      "epoch": 0.04523809523809524,
+      "grad_norm": 1.7469535134849759,
+      "learning_rate": 4.523809523809524e-06,
+      "loss": 0.8426,
+      "step": 95
+    },
+    {
+      "epoch": 0.047619047619047616,
+      "grad_norm": 1.299723199427286,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 0.8305,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.40840933352402,
+      "learning_rate": 5e-06,
+      "loss": 0.8316,
+      "step": 105
+    },
+    {
+      "epoch": 0.05238095238095238,
+      "grad_norm": 1.0810746876875799,
+      "learning_rate": 5.2380952380952384e-06,
+      "loss": 0.8225,
+      "step": 110
+    },
+    {
+      "epoch": 0.05476190476190476,
+      "grad_norm": 1.1168753096844857,
+      "learning_rate": 5.476190476190477e-06,
+      "loss": 0.8141,
+      "step": 115
+    },
+    {
+      "epoch": 0.05714285714285714,
+      "grad_norm": 1.117741427127342,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.8128,
+      "step": 120
+    },
+    {
+      "epoch": 0.05952380952380952,
+      "grad_norm": 1.117824418032963,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 0.8104,
+      "step": 125
+    },
+    {
+      "epoch": 0.06190476190476191,
+      "grad_norm": 1.1765989152315592,
+      "learning_rate": 6.1904761904761914e-06,
+      "loss": 0.8186,
+      "step": 130
+    },
+    {
+      "epoch": 0.06428571428571428,
+      "grad_norm": 1.14607344297389,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.8094,
+      "step": 135
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 1.0689364794376695,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7889,
+      "step": 140
+    },
+    {
+      "epoch": 0.06904761904761905,
+      "grad_norm": 1.0802475624680903,
+      "learning_rate": 6.9047619047619055e-06,
+      "loss": 0.8072,
+      "step": 145
+    },
+    {
+      "epoch": 0.07142857142857142,
+      "grad_norm": 1.1977014078487442,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.7951,
+      "step": 150
+    },
+    {
+      "epoch": 0.07380952380952381,
+      "grad_norm": 1.0956749755790958,
+      "learning_rate": 7.380952380952382e-06,
+      "loss": 0.7961,
+      "step": 155
+    },
+    {
+      "epoch": 0.0761904761904762,
+      "grad_norm": 1.0269565395277456,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 0.7883,
+      "step": 160
+    },
+    {
+      "epoch": 0.07857142857142857,
+      "grad_norm": 1.2693412905779187,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.7941,
+      "step": 165
+    },
+    {
+      "epoch": 0.08095238095238096,
+      "grad_norm": 1.4270103378714147,
+      "learning_rate": 8.095238095238097e-06,
+      "loss": 0.7976,
+      "step": 170
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "grad_norm": 1.3965101517157619,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.7907,
+      "step": 175
+    },
+    {
+      "epoch": 0.08571428571428572,
+      "grad_norm": 1.1737877049103937,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.7772,
+      "step": 180
+    },
+    {
+      "epoch": 0.0880952380952381,
+      "grad_norm": 0.9620328652517623,
+      "learning_rate": 8.80952380952381e-06,
+      "loss": 0.7895,
+      "step": 185
+    },
+    {
+      "epoch": 0.09047619047619047,
+      "grad_norm": 0.9954694445859312,
+      "learning_rate": 9.047619047619049e-06,
+      "loss": 0.7728,
+      "step": 190
+    },
+    {
+      "epoch": 0.09285714285714286,
+      "grad_norm": 0.9933982715674028,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.7784,
+      "step": 195
+    },
+    {
+      "epoch": 0.09523809523809523,
+      "grad_norm": 1.2380225457608798,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.762,
+      "step": 200
+    },
+    {
+      "epoch": 0.09761904761904762,
+      "grad_norm": 1.1645854540151788,
+      "learning_rate": 9.761904761904762e-06,
+      "loss": 0.7708,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0347109854779557,
+      "learning_rate": 1e-05,
+      "loss": 0.7777,
+      "step": 210
+    },
+    {
+      "epoch": 0.10238095238095238,
+      "grad_norm": 1.007801398524834,
+      "learning_rate": 1.0238095238095238e-05,
+      "loss": 0.7696,
+      "step": 215
+    },
+    {
+      "epoch": 0.10476190476190476,
+      "grad_norm": 1.1084444980598738,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 0.7667,
+      "step": 220
+    },
+    {
+      "epoch": 0.10714285714285714,
+      "grad_norm": 0.9232282482964133,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.7682,
+      "step": 225
+    },
+    {
+      "epoch": 0.10952380952380952,
+      "grad_norm": 1.0266526312590731,
+      "learning_rate": 1.0952380952380955e-05,
+      "loss": 0.76,
+      "step": 230
+    },
+    {
+      "epoch": 0.11190476190476191,
+      "grad_norm": 0.9800538664923668,
+      "learning_rate": 1.1190476190476192e-05,
+      "loss": 0.7787,
+      "step": 235
+    },
+    {
+      "epoch": 0.11428571428571428,
+      "grad_norm": 1.0852968890165064,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.7584,
+      "step": 240
+    },
+    {
+      "epoch": 0.11666666666666667,
+      "grad_norm": 1.101079132960328,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.7532,
+      "step": 245
+    },
+    {
+      "epoch": 0.11904761904761904,
+      "grad_norm": 1.0434818420770975,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.766,
+      "step": 250
+    },
+    {
+      "epoch": 0.12142857142857143,
+      "grad_norm": 0.8949846922479692,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.764,
+      "step": 255
+    },
+    {
+      "epoch": 0.12380952380952381,
+      "grad_norm": 0.9144371936075102,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 0.7518,
+      "step": 260
+    },
+    {
+      "epoch": 0.1261904761904762,
+      "grad_norm": 1.098930660048903,
+      "learning_rate": 1.261904761904762e-05,
+      "loss": 0.7549,
+      "step": 265
+    },
+    {
+      "epoch": 0.12857142857142856,
+      "grad_norm": 1.0213006883333104,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.7514,
+      "step": 270
+    },
+    {
+      "epoch": 0.13095238095238096,
+      "grad_norm": 0.9116166496696912,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.7515,
+      "step": 275
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 1.0474312553231273,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7663,
+      "step": 280
+    },
+    {
+      "epoch": 0.1357142857142857,
+      "grad_norm": 0.8710985201179252,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.7608,
+      "step": 285
+    },
+    {
+      "epoch": 0.1380952380952381,
+      "grad_norm": 1.3401884340352725,
+      "learning_rate": 1.3809523809523811e-05,
+      "loss": 0.7554,
+      "step": 290
+    },
+    {
+      "epoch": 0.14047619047619048,
+      "grad_norm": 1.0740872828901855,
+      "learning_rate": 1.4047619047619048e-05,
+      "loss": 0.7582,
+      "step": 295
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 0.9126548421949322,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.7434,
+      "step": 300
+    },
+    {
+      "epoch": 0.14523809523809525,
+      "grad_norm": 1.1120651258567102,
+      "learning_rate": 1.4523809523809524e-05,
+      "loss": 0.7388,
+      "step": 305
+    },
+    {
+      "epoch": 0.14761904761904762,
+      "grad_norm": 0.9950599559753186,
+      "learning_rate": 1.4761904761904763e-05,
+      "loss": 0.7585,
+      "step": 310
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8107934316232559,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7355,
+      "step": 315
+    },
+    {
+      "epoch": 0.1523809523809524,
+      "grad_norm": 0.808619303284316,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 0.7504,
+      "step": 320
+    },
+    {
+      "epoch": 0.15476190476190477,
+      "grad_norm": 0.8677410171473149,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.7607,
+      "step": 325
+    },
+    {
+      "epoch": 0.15714285714285714,
+      "grad_norm": 0.8160199476236252,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.7464,
+      "step": 330
+    },
+    {
+      "epoch": 0.1595238095238095,
+      "grad_norm": 0.8264103314451071,
+      "learning_rate": 1.5952380952380954e-05,
+      "loss": 0.7416,
+      "step": 335
+    },
+    {
+      "epoch": 0.1619047619047619,
+      "grad_norm": 0.9284321857373765,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.7649,
+      "step": 340
+    },
+    {
+      "epoch": 0.16428571428571428,
+      "grad_norm": 0.9711426336534976,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.7388,
+      "step": 345
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 1.0029532403604327,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.7495,
+      "step": 350
+    },
+    {
+      "epoch": 0.16904761904761906,
+      "grad_norm": 0.8655401816761288,
+      "learning_rate": 1.6904761904761906e-05,
+      "loss": 0.7431,
+      "step": 355
+    },
+    {
+      "epoch": 0.17142857142857143,
+      "grad_norm": 0.811823602075138,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.7409,
+      "step": 360
+    },
+    {
+      "epoch": 0.1738095238095238,
+      "grad_norm": 0.7850380081345122,
+      "learning_rate": 1.7380952380952384e-05,
+      "loss": 0.7496,
+      "step": 365
+    },
+    {
+      "epoch": 0.1761904761904762,
+      "grad_norm": 0.8871490424351564,
+      "learning_rate": 1.761904761904762e-05,
+      "loss": 0.7502,
+      "step": 370
+    },
+    {
+      "epoch": 0.17857142857142858,
+      "grad_norm": 0.857041601013508,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.7344,
+      "step": 375
+    },
+    {
+      "epoch": 0.18095238095238095,
+      "grad_norm": 0.8798387422023826,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 0.7451,
+      "step": 380
+    },
+    {
+      "epoch": 0.18333333333333332,
+      "grad_norm": 0.8208318602053298,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.7519,
+      "step": 385
+    },
+    {
+      "epoch": 0.18571428571428572,
+      "grad_norm": 0.9816478125661463,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.7438,
+      "step": 390
+    },
+    {
+      "epoch": 0.1880952380952381,
+      "grad_norm": 0.8587183984289388,
+      "learning_rate": 1.880952380952381e-05,
+      "loss": 0.74,
+      "step": 395
+    },
+    {
+      "epoch": 0.19047619047619047,
+      "grad_norm": 0.8766655919062369,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 0.7506,
+      "step": 400
+    },
+    {
+      "epoch": 0.19285714285714287,
+      "grad_norm": 0.8309437973338848,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.745,
+      "step": 405
+    },
+    {
+      "epoch": 0.19523809523809524,
+      "grad_norm": 0.8106184766065911,
+      "learning_rate": 1.9523809523809524e-05,
+      "loss": 0.7475,
+      "step": 410
+    },
+    {
+      "epoch": 0.1976190476190476,
+      "grad_norm": 0.809891965368119,
+      "learning_rate": 1.9761904761904763e-05,
+      "loss": 0.7335,
+      "step": 415
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7555036664771403,
+      "learning_rate": 2e-05,
+      "loss": 0.7424,
+      "step": 420
+    },
+    {
+      "epoch": 0.20238095238095238,
+      "grad_norm": 0.7650413836527804,
+      "learning_rate": 1.999991365731819e-05,
+      "loss": 0.7501,
+      "step": 425
+    },
+    {
+      "epoch": 0.20476190476190476,
+      "grad_norm": 0.7312208748548112,
+      "learning_rate": 1.999965463076377e-05,
+      "loss": 0.7334,
+      "step": 430
+    },
+    {
+      "epoch": 0.20714285714285716,
+      "grad_norm": 0.821378907440853,
+      "learning_rate": 1.999922292480975e-05,
+      "loss": 0.7398,
+      "step": 435
+    },
+    {
+      "epoch": 0.20952380952380953,
+      "grad_norm": 0.7961564796311429,
+      "learning_rate": 1.999861854691106e-05,
+      "loss": 0.7383,
+      "step": 440
+    },
+    {
+      "epoch": 0.2119047619047619,
+      "grad_norm": 0.864941128195607,
+      "learning_rate": 1.999784150750442e-05,
+      "loss": 0.7453,
+      "step": 445
+    },
+    {
+      "epoch": 0.21428571428571427,
+      "grad_norm": 0.9432973729407097,
+      "learning_rate": 1.9996891820008165e-05,
+      "loss": 0.7428,
+      "step": 450
+    },
+    {
+      "epoch": 0.21666666666666667,
+      "grad_norm": 0.744201840220319,
+      "learning_rate": 1.9995769500822007e-05,
+      "loss": 0.732,
+      "step": 455
+    },
+    {
+      "epoch": 0.21904761904761905,
+      "grad_norm": 0.8051356121636954,
+      "learning_rate": 1.999447456932676e-05,
+      "loss": 0.7452,
+      "step": 460
+    },
+    {
+      "epoch": 0.22142857142857142,
+      "grad_norm": 0.7446879893905328,
+      "learning_rate": 1.9993007047883988e-05,
+      "loss": 0.7406,
+      "step": 465
+    },
+    {
+      "epoch": 0.22380952380952382,
+      "grad_norm": 0.8200085500457187,
+      "learning_rate": 1.9991366961835643e-05,
+      "loss": 0.7342,
+      "step": 470
+    },
+    {
+      "epoch": 0.2261904761904762,
+      "grad_norm": 0.6879793430211231,
+      "learning_rate": 1.9989554339503612e-05,
+      "loss": 0.737,
+      "step": 475
+    },
+    {
+      "epoch": 0.22857142857142856,
+      "grad_norm": 0.6767209429381076,
+      "learning_rate": 1.9987569212189224e-05,
+      "loss": 0.7365,
+      "step": 480
+    },
+    {
+      "epoch": 0.23095238095238096,
+      "grad_norm": 0.7417349665434714,
+      "learning_rate": 1.9985411614172728e-05,
+      "loss": 0.7173,
+      "step": 485
+    },
+    {
+      "epoch": 0.23333333333333334,
+      "grad_norm": 0.7446106009527734,
+      "learning_rate": 1.9983081582712684e-05,
+      "loss": 0.7423,
+      "step": 490
+    },
+    {
+      "epoch": 0.2357142857142857,
+      "grad_norm": 0.7442452133783487,
+      "learning_rate": 1.9980579158045322e-05,
+      "loss": 0.7165,
+      "step": 495
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 0.7769424686532669,
+      "learning_rate": 1.997790438338385e-05,
+      "loss": 0.749,
+      "step": 500
+    },
+    {
+      "epoch": 0.24047619047619048,
+      "grad_norm": 0.8034839265796356,
+      "learning_rate": 1.997505730491772e-05,
+      "loss": 0.7385,
+      "step": 505
+    },
+    {
+      "epoch": 0.24285714285714285,
+      "grad_norm": 0.7676614166297908,
+      "learning_rate": 1.9972037971811802e-05,
+      "loss": 0.7383,
+      "step": 510
+    },
+    {
+      "epoch": 0.24523809523809523,
+      "grad_norm": 0.9359149520411593,
+      "learning_rate": 1.9968846436205566e-05,
+      "loss": 0.7498,
+      "step": 515
+    },
+    {
+      "epoch": 0.24761904761904763,
+      "grad_norm": 0.7783563085530937,
+      "learning_rate": 1.9965482753212154e-05,
+      "loss": 0.7466,
+      "step": 520
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7058199644697823,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 0.7327,
+      "step": 525
+    },
+    {
+      "epoch": 0.2523809523809524,
+      "grad_norm": 0.6768276735212928,
+      "learning_rate": 1.995823918037908e-05,
+      "loss": 0.7396,
+      "step": 530
+    },
+    {
+      "epoch": 0.25476190476190474,
+      "grad_norm": 0.7530681690630258,
+      "learning_rate": 1.9954359415625313e-05,
+      "loss": 0.736,
+      "step": 535
+    },
+    {
+      "epoch": 0.2571428571428571,
+      "grad_norm": 0.790729948461572,
+      "learning_rate": 1.9950307753654016e-05,
+      "loss": 0.7367,
+      "step": 540
+    },
+    {
+      "epoch": 0.25952380952380955,
+      "grad_norm": 0.8098472943164579,
+      "learning_rate": 1.994608426443146e-05,
+      "loss": 0.7484,
+      "step": 545
+    },
+    {
+      "epoch": 0.2619047619047619,
+      "grad_norm": 0.6594040663378262,
+      "learning_rate": 1.994168902089112e-05,
+      "loss": 0.7256,
+      "step": 550
+    },
+    {
+      "epoch": 0.2642857142857143,
+      "grad_norm": 0.7061644477917397,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 0.7293,
+      "step": 555
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.6552145622334516,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.7235,
+      "step": 560
+    },
+    {
+      "epoch": 0.26904761904761904,
+      "grad_norm": 0.7850549934752342,
+      "learning_rate": 1.9927473538179467e-05,
+      "loss": 0.7242,
+      "step": 565
+    },
+    {
+      "epoch": 0.2714285714285714,
+      "grad_norm": 0.6946715346854551,
+      "learning_rate": 1.9922392066001724e-05,
+      "loss": 0.7352,
+      "step": 570
+    },
+    {
+      "epoch": 0.27380952380952384,
+      "grad_norm": 0.6701958153898289,
+      "learning_rate": 1.9917139248635788e-05,
+      "loss": 0.7358,
+      "step": 575
+    },
+    {
+      "epoch": 0.2761904761904762,
+      "grad_norm": 0.6957543084857959,
+      "learning_rate": 1.991171517679013e-05,
+      "loss": 0.7287,
+      "step": 580
+    },
+    {
+      "epoch": 0.2785714285714286,
+      "grad_norm": 0.7412462140647854,
+      "learning_rate": 1.9906119944130527e-05,
+      "loss": 0.7314,
+      "step": 585
+    },
+    {
+      "epoch": 0.28095238095238095,
+      "grad_norm": 0.7180123780939238,
+      "learning_rate": 1.9900353647278466e-05,
+      "loss": 0.7049,
+      "step": 590
+    },
+    {
+      "epoch": 0.2833333333333333,
+      "grad_norm": 0.6346311822511415,
+      "learning_rate": 1.9894416385809444e-05,
+      "loss": 0.7371,
+      "step": 595
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 0.9181856171310567,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 0.7537,
+      "step": 600
+    },
+    {
+      "epoch": 0.28809523809523807,
+      "grad_norm": 0.7646478108344241,
+      "learning_rate": 1.9882029382082342e-05,
+      "loss": 0.724,
+      "step": 605
+    },
+    {
+      "epoch": 0.2904761904761905,
+      "grad_norm": 0.7332726331013449,
+      "learning_rate": 1.9875579853729677e-05,
+      "loss": 0.7273,
+      "step": 610
+    },
+    {
+      "epoch": 0.29285714285714287,
+      "grad_norm": 0.8337211626177949,
+      "learning_rate": 1.9868959788567213e-05,
+      "loss": 0.7359,
+      "step": 615
+    },
+    {
+      "epoch": 0.29523809523809524,
+      "grad_norm": 0.72291541850261,
+      "learning_rate": 1.9862169300913784e-05,
+      "loss": 0.7246,
+      "step": 620
+    },
+    {
+      "epoch": 0.2976190476190476,
+      "grad_norm": 0.6775385452101815,
+      "learning_rate": 1.9855208508031173e-05,
+      "loss": 0.719,
+      "step": 625
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6899659929159091,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.7311,
+      "step": 630
+    },
+    {
+      "epoch": 0.30238095238095236,
+      "grad_norm": 0.6092353332947169,
+      "learning_rate": 1.9840776490328067e-05,
+      "loss": 0.7283,
+      "step": 635
+    },
+    {
+      "epoch": 0.3047619047619048,
+      "grad_norm": 0.6170832705569598,
+      "learning_rate": 1.9833305514727396e-05,
+      "loss": 0.7341,
+      "step": 640
+    },
+    {
+      "epoch": 0.30714285714285716,
+      "grad_norm": 0.6299433950672038,
+      "learning_rate": 1.9825664732332886e-05,
+      "loss": 0.7224,
+      "step": 645
+    },
+    {
+      "epoch": 0.30952380952380953,
+      "grad_norm": 0.620304145431242,
+      "learning_rate": 1.981785427508966e-05,
+      "loss": 0.7325,
+      "step": 650
+    },
+    {
+      "epoch": 0.3119047619047619,
+      "grad_norm": 0.6551690352072297,
+      "learning_rate": 1.9809874277872886e-05,
+      "loss": 0.7042,
+      "step": 655
+    },
+    {
+      "epoch": 0.3142857142857143,
+      "grad_norm": 0.6965123962977487,
+      "learning_rate": 1.9801724878485438e-05,
+      "loss": 0.7142,
+      "step": 660
+    },
+    {
+      "epoch": 0.31666666666666665,
+      "grad_norm": 0.6522958895192698,
+      "learning_rate": 1.9793406217655516e-05,
+      "loss": 0.7188,
+      "step": 665
+    },
+    {
+      "epoch": 0.319047619047619,
+      "grad_norm": 0.6710935693138489,
+      "learning_rate": 1.9784918439034216e-05,
+      "loss": 0.7127,
+      "step": 670
+    },
+    {
+      "epoch": 0.32142857142857145,
+      "grad_norm": 0.6394392122127832,
+      "learning_rate": 1.977626168919305e-05,
+      "loss": 0.7301,
+      "step": 675
+    },
+    {
+      "epoch": 0.3238095238095238,
+      "grad_norm": 0.7388339763922518,
+      "learning_rate": 1.9767436117621416e-05,
+      "loss": 0.7217,
+      "step": 680
+    },
+    {
+      "epoch": 0.3261904761904762,
+      "grad_norm": 0.6684391122808601,
+      "learning_rate": 1.975844187672402e-05,
+      "loss": 0.6975,
+      "step": 685
+    },
+    {
+      "epoch": 0.32857142857142857,
+      "grad_norm": 0.6933743953258287,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.712,
+      "step": 690
+    },
+    {
+      "epoch": 0.33095238095238094,
+      "grad_norm": 0.6999429639549631,
+      "learning_rate": 1.9739948011131438e-05,
+      "loss": 0.7275,
+      "step": 695
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.6381718516637829,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.7269,
+      "step": 700
+    },
+    {
+      "epoch": 0.3357142857142857,
+      "grad_norm": 0.6640845851267309,
+      "learning_rate": 1.9720781369857747e-05,
+      "loss": 0.7254,
+      "step": 705
+    },
+    {
+      "epoch": 0.3380952380952381,
+      "grad_norm": 0.6592163354528279,
+      "learning_rate": 1.9710946170250702e-05,
+      "loss": 0.7071,
+      "step": 710
+    },
+    {
+      "epoch": 0.3404761904761905,
+      "grad_norm": 0.7655182235424155,
+      "learning_rate": 1.9700943276816602e-05,
+      "loss": 0.7241,
+      "step": 715
+    },
+    {
+      "epoch": 0.34285714285714286,
+      "grad_norm": 0.643273448107574,
+      "learning_rate": 1.969077286229078e-05,
+      "loss": 0.7286,
+      "step": 720
+    },
+    {
+      "epoch": 0.34523809523809523,
+      "grad_norm": 0.6857609061991075,
+      "learning_rate": 1.9680435102301412e-05,
+      "loss": 0.7079,
+      "step": 725
+    },
+    {
+      "epoch": 0.3476190476190476,
+      "grad_norm": 0.8672880237155081,
+      "learning_rate": 1.9669930175366474e-05,
+      "loss": 0.7112,
+      "step": 730
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6654257223083666,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.726,
+      "step": 735
+    },
+    {
+      "epoch": 0.3523809523809524,
+      "grad_norm": 0.6635633178878109,
+      "learning_rate": 1.964841954916235e-05,
+      "loss": 0.7095,
+      "step": 740
+    },
+    {
+      "epoch": 0.3547619047619048,
+      "grad_norm": 0.7692774931160247,
+      "learning_rate": 1.9637414221350198e-05,
+      "loss": 0.7078,
+      "step": 745
+    },
+    {
+      "epoch": 0.35714285714285715,
+      "grad_norm": 0.6290937699685107,
+      "learning_rate": 1.962624246950012e-05,
+      "loss": 0.7228,
+      "step": 750
+    },
+    {
+      "epoch": 0.3595238095238095,
+      "grad_norm": 0.645738278667474,
+      "learning_rate": 1.9614904486531935e-05,
+      "loss": 0.7134,
+      "step": 755
+    },
+    {
+      "epoch": 0.3619047619047619,
+      "grad_norm": 0.6234306399886239,
+      "learning_rate": 1.9603400468236e-05,
+      "loss": 0.7132,
+      "step": 760
+    },
+    {
+      "epoch": 0.36428571428571427,
+      "grad_norm": 0.6027074648266817,
+      "learning_rate": 1.9591730613269878e-05,
+      "loss": 0.7019,
+      "step": 765
+    },
+    {
+      "epoch": 0.36666666666666664,
+      "grad_norm": 0.5952054087646503,
+      "learning_rate": 1.957989512315489e-05,
+      "loss": 0.7134,
+      "step": 770
+    },
+    {
+      "epoch": 0.36904761904761907,
+      "grad_norm": 0.5954424676547249,
+      "learning_rate": 1.9567894202272623e-05,
+      "loss": 0.7102,
+      "step": 775
+    },
+    {
+      "epoch": 0.37142857142857144,
+      "grad_norm": 0.6146466190073,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.7054,
+      "step": 780
+    },
+    {
+      "epoch": 0.3738095238095238,
+      "grad_norm": 0.6196401686238979,
+      "learning_rate": 1.9543396900012763e-05,
+      "loss": 0.7223,
+      "step": 785
+    },
+    {
+      "epoch": 0.3761904761904762,
+      "grad_norm": 0.6376202895536227,
+      "learning_rate": 1.9530900941667733e-05,
+      "loss": 0.7227,
+      "step": 790
+    },
+    {
+      "epoch": 0.37857142857142856,
+      "grad_norm": 0.6600915780771018,
+      "learning_rate": 1.9518240398613226e-05,
+      "loss": 0.6997,
+      "step": 795
+    },
+    {
+      "epoch": 0.38095238095238093,
+      "grad_norm": 0.5882182414716866,
+      "learning_rate": 1.9505415489478293e-05,
+      "loss": 0.7268,
+      "step": 800
+    },
+    {
+      "epoch": 0.38333333333333336,
+      "grad_norm": 0.6384408692158889,
+      "learning_rate": 1.949242643573034e-05,
+      "loss": 0.712,
+      "step": 805
+    },
+    {
+      "epoch": 0.38571428571428573,
+      "grad_norm": 0.6036619986439133,
+      "learning_rate": 1.947927346167132e-05,
+      "loss": 0.7175,
+      "step": 810
+    },
+    {
+      "epoch": 0.3880952380952381,
+      "grad_norm": 0.7029032826216544,
+      "learning_rate": 1.9465956794433837e-05,
+      "loss": 0.707,
+      "step": 815
+    },
+    {
+      "epoch": 0.3904761904761905,
+      "grad_norm": 0.6457510326949268,
+      "learning_rate": 1.945247666397725e-05,
+      "loss": 0.7036,
+      "step": 820
+    },
+    {
+      "epoch": 0.39285714285714285,
+      "grad_norm": 0.6878440257755464,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 0.6989,
+      "step": 825
+    },
+    {
+      "epoch": 0.3952380952380952,
+      "grad_norm": 0.6188605951379363,
+      "learning_rate": 1.9425026947353994e-05,
+      "loss": 0.7105,
+      "step": 830
+    },
+    {
+      "epoch": 0.3976190476190476,
+      "grad_norm": 0.5840120067803837,
+      "learning_rate": 1.9411057835203756e-05,
+      "loss": 0.7058,
+      "step": 835
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5937069559439255,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.695,
+      "step": 840
+    },
+    {
+      "epoch": 0.4023809523809524,
+      "grad_norm": 0.7110875706843621,
+      "learning_rate": 1.9382632309352503e-05,
+      "loss": 0.7227,
+      "step": 845
+    },
+    {
+      "epoch": 0.40476190476190477,
+      "grad_norm": 0.6311314888265864,
+      "learning_rate": 1.936817638651871e-05,
+      "loss": 0.7011,
+      "step": 850
+    },
+    {
+      "epoch": 0.40714285714285714,
+      "grad_norm": 0.6286300385263309,
+      "learning_rate": 1.935355868899034e-05,
+      "loss": 0.6979,
+      "step": 855
+    },
+    {
+      "epoch": 0.4095238095238095,
+      "grad_norm": 0.6209403160626118,
+      "learning_rate": 1.9338779469193638e-05,
+      "loss": 0.6996,
+      "step": 860
+    },
+    {
+      "epoch": 0.4119047619047619,
+      "grad_norm": 0.5727631219976955,
+      "learning_rate": 1.9323838982344092e-05,
+      "loss": 0.7114,
+      "step": 865
+    },
+    {
+      "epoch": 0.4142857142857143,
+      "grad_norm": 0.6417775935264546,
+      "learning_rate": 1.9308737486442045e-05,
+      "loss": 0.702,
+      "step": 870
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.6620199236263371,
+      "learning_rate": 1.9293475242268224e-05,
+      "loss": 0.7226,
+      "step": 875
+    },
+    {
+      "epoch": 0.41904761904761906,
+      "grad_norm": 0.6399880589166196,
+      "learning_rate": 1.9278052513379256e-05,
+      "loss": 0.6925,
+      "step": 880
+    },
+    {
+      "epoch": 0.42142857142857143,
+      "grad_norm": 0.6138453722448008,
+      "learning_rate": 1.926246956610309e-05,
+      "loss": 0.7118,
+      "step": 885
+    },
+    {
+      "epoch": 0.4238095238095238,
+      "grad_norm": 0.597989080002243,
+      "learning_rate": 1.9246726669534416e-05,
+      "loss": 0.73,
+      "step": 890
+    },
+    {
+      "epoch": 0.4261904761904762,
+      "grad_norm": 0.5901338899915719,
+      "learning_rate": 1.923082409553002e-05,
+      "loss": 0.7067,
+      "step": 895
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 0.6016114619725854,
+      "learning_rate": 1.921476211870408e-05,
+      "loss": 0.7103,
+      "step": 900
+    },
+    {
+      "epoch": 0.430952380952381,
+      "grad_norm": 0.5950314108298304,
+      "learning_rate": 1.919854101642342e-05,
+      "loss": 0.7052,
+      "step": 905
+    },
+    {
+      "epoch": 0.43333333333333335,
+      "grad_norm": 0.5879370937942204,
+      "learning_rate": 1.9182161068802742e-05,
+      "loss": 0.6995,
+      "step": 910
+    },
+    {
+      "epoch": 0.4357142857142857,
+      "grad_norm": 0.6266968125559099,
+      "learning_rate": 1.9165622558699763e-05,
+      "loss": 0.7097,
+      "step": 915
+    },
+    {
+      "epoch": 0.4380952380952381,
+      "grad_norm": 0.6639659528952871,
+      "learning_rate": 1.9148925771710347e-05,
+      "loss": 0.7009,
+      "step": 920
+    },
+    {
+      "epoch": 0.44047619047619047,
+      "grad_norm": 0.6321493695976013,
+      "learning_rate": 1.913207099616357e-05,
+      "loss": 0.7053,
+      "step": 925
+    },
+    {
+      "epoch": 0.44285714285714284,
+      "grad_norm": 0.6782903937955044,
+      "learning_rate": 1.9115058523116734e-05,
+      "loss": 0.7004,
+      "step": 930
+    },
+    {
+      "epoch": 0.4452380952380952,
+      "grad_norm": 0.6716356876066842,
+      "learning_rate": 1.9097888646350347e-05,
+      "loss": 0.7058,
+      "step": 935
+    },
+    {
+      "epoch": 0.44761904761904764,
+      "grad_norm": 0.5994163783124282,
+      "learning_rate": 1.908056166236305e-05,
+      "loss": 0.6965,
+      "step": 940
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7011305510255346,
+      "learning_rate": 1.9063077870366504e-05,
+      "loss": 0.6973,
+      "step": 945
+    },
+    {
+      "epoch": 0.4523809523809524,
+      "grad_norm": 0.6382034265421508,
+      "learning_rate": 1.9045437572280193e-05,
+      "loss": 0.7122,
+      "step": 950
+    },
+    {
+      "epoch": 0.45476190476190476,
+      "grad_norm": 0.652835650762035,
+      "learning_rate": 1.902764107272626e-05,
+      "loss": 0.7154,
+      "step": 955
+    },
+    {
+      "epoch": 0.45714285714285713,
+      "grad_norm": 0.5737401205474331,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.7094,
+      "step": 960
+    },
+    {
+      "epoch": 0.4595238095238095,
+      "grad_norm": 0.6510894452271442,
+      "learning_rate": 1.8991580701185564e-05,
+      "loss": 0.7126,
+      "step": 965
+    },
+    {
+      "epoch": 0.46190476190476193,
+      "grad_norm": 0.7386017741154395,
+      "learning_rate": 1.8973317451908642e-05,
+      "loss": 0.7007,
+      "step": 970
+    },
+    {
+      "epoch": 0.4642857142857143,
+      "grad_norm": 0.6752036021573017,
+      "learning_rate": 1.895489924657301e-05,
+      "loss": 0.7041,
+      "step": 975
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 0.597408051286458,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.7212,
+      "step": 980
+    },
+    {
+      "epoch": 0.46904761904761905,
+      "grad_norm": 0.5796714555019369,
+      "learning_rate": 1.8917599242617796e-05,
+      "loss": 0.7159,
+      "step": 985
+    },
+    {
+      "epoch": 0.4714285714285714,
+      "grad_norm": 0.5688480490870661,
+      "learning_rate": 1.8898718088114688e-05,
+      "loss": 0.6997,
+      "step": 990
+    },
+    {
+      "epoch": 0.4738095238095238,
+      "grad_norm": 0.6090212361351263,
+      "learning_rate": 1.8879683265774695e-05,
+      "loss": 0.7152,
+      "step": 995
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 0.5638933903277263,
+      "learning_rate": 1.8860495104301346e-05,
+      "loss": 0.689,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4785714285714286,
+      "grad_norm": 0.5775879905939127,
+      "learning_rate": 1.8841153935046098e-05,
+      "loss": 0.7071,
+      "step": 1005
+    },
+    {
+      "epoch": 0.48095238095238096,
+      "grad_norm": 0.5955667805199039,
+      "learning_rate": 1.8821660092002642e-05,
+      "loss": 0.718,
+      "step": 1010
+    },
+    {
+      "epoch": 0.48333333333333334,
+      "grad_norm": 0.6578844006654452,
+      "learning_rate": 1.880201391180111e-05,
+      "loss": 0.7203,
+      "step": 1015
+    },
+    {
+      "epoch": 0.4857142857142857,
+      "grad_norm": 0.6288999781322553,
+      "learning_rate": 1.8782215733702286e-05,
+      "loss": 0.7005,
+      "step": 1020
+    },
+    {
+      "epoch": 0.4880952380952381,
+      "grad_norm": 0.6154852254510217,
+      "learning_rate": 1.8762265899591724e-05,
+      "loss": 0.7112,
+      "step": 1025
+    },
+    {
+      "epoch": 0.49047619047619045,
+      "grad_norm": 0.5946678081221346,
+      "learning_rate": 1.874216475397386e-05,
+      "loss": 0.7072,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4928571428571429,
+      "grad_norm": 0.5542144431999712,
+      "learning_rate": 1.8721912643966055e-05,
+      "loss": 0.7234,
+      "step": 1035
+    },
+    {
+      "epoch": 0.49523809523809526,
+      "grad_norm": 0.6327874278600627,
+      "learning_rate": 1.870150991929261e-05,
+      "loss": 0.7199,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4976190476190476,
+      "grad_norm": 0.6050382016312242,
+      "learning_rate": 1.868095693227872e-05,
+      "loss": 0.6989,
+      "step": 1045
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6060484644475487,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.7094,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5023809523809524,
+      "grad_norm": 0.5578204676405346,
+      "learning_rate": 1.86394015934983e-05,
+      "loss": 0.6998,
+      "step": 1055
+    },
+    {
+      "epoch": 0.5047619047619047,
+      "grad_norm": 0.5912489553089566,
+      "learning_rate": 1.8618399959331642e-05,
+      "loss": 0.704,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5071428571428571,
+      "grad_norm": 0.6302255111379419,
+      "learning_rate": 1.8597249498011906e-05,
+      "loss": 0.7273,
+      "step": 1065
+    },
+    {
+      "epoch": 0.5095238095238095,
+      "grad_norm": 0.6377098555634427,
+      "learning_rate": 1.8575950574776595e-05,
+      "loss": 0.7118,
+      "step": 1070
+    },
+    {
+      "epoch": 0.5119047619047619,
+      "grad_norm": 0.6306289974408017,
+      "learning_rate": 1.855450355742695e-05,
+      "loss": 0.6975,
+      "step": 1075
+    },
+    {
+      "epoch": 0.5142857142857142,
+      "grad_norm": 0.6118126960406806,
+      "learning_rate": 1.8532908816321557e-05,
+      "loss": 0.7089,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5166666666666667,
+      "grad_norm": 0.5631635531801148,
+      "learning_rate": 1.8511166724369997e-05,
+      "loss": 0.7027,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5190476190476191,
+      "grad_norm": 0.5712800757429177,
+      "learning_rate": 1.8489277657026377e-05,
+      "loss": 0.7068,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5214285714285715,
+      "grad_norm": 0.560805851835222,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 0.7012,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5238095238095238,
+      "grad_norm": 0.6607109669389528,
+      "learning_rate": 1.844506011066308e-05,
+      "loss": 0.6919,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5261904761904762,
+      "grad_norm": 0.5900866880976501,
+      "learning_rate": 1.8422732395215717e-05,
+      "loss": 0.6903,
+      "step": 1105
+    },
+    {
+      "epoch": 0.5285714285714286,
+      "grad_norm": 0.5897342260968936,
+      "learning_rate": 1.8400259231507716e-05,
+      "loss": 0.6908,
+      "step": 1110
+    },
+    {
+      "epoch": 0.530952380952381,
+      "grad_norm": 0.5723672657261824,
+      "learning_rate": 1.8377641007617724e-05,
+      "loss": 0.6954,
+      "step": 1115
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.611058093387684,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.7017,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5357142857142857,
+      "grad_norm": 0.5318902267208334,
+      "learning_rate": 1.833197094412449e-05,
+      "loss": 0.6953,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5380952380952381,
+      "grad_norm": 0.5570664080835552,
+      "learning_rate": 1.8308919893176397e-05,
+      "loss": 0.7066,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5404761904761904,
+      "grad_norm": 0.5885337969920328,
+      "learning_rate": 1.8285725359343e-05,
+      "loss": 0.706,
+      "step": 1135
+    },
+    {
+      "epoch": 0.5428571428571428,
+      "grad_norm": 0.5855314615908318,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.701,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5452380952380952,
+      "grad_norm": 0.5851946825105788,
+      "learning_rate": 1.8238907447633716e-05,
+      "loss": 0.692,
+      "step": 1145
+    },
+    {
+      "epoch": 0.5476190476190477,
+      "grad_norm": 0.5927970007531321,
+      "learning_rate": 1.8215284878234644e-05,
+      "loss": 0.716,
+      "step": 1150
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5263874307074303,
+      "learning_rate": 1.819152044288992e-05,
+      "loss": 0.6904,
+      "step": 1155
+    },
+    {
+      "epoch": 0.5523809523809524,
+      "grad_norm": 0.6124793068078697,
+      "learning_rate": 1.816761455197657e-05,
+      "loss": 0.6864,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5547619047619048,
+      "grad_norm": 0.5828934954007933,
+      "learning_rate": 1.8143567618314336e-05,
+      "loss": 0.6904,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5571428571428572,
+      "grad_norm": 0.5831980381697426,
+      "learning_rate": 1.811938005715857e-05,
+      "loss": 0.7086,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5595238095238095,
+      "grad_norm": 0.6435462278644365,
+      "learning_rate": 1.8095052286193044e-05,
+      "loss": 0.6976,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5619047619047619,
+      "grad_norm": 0.5815432160069361,
+      "learning_rate": 1.8070584725522763e-05,
+      "loss": 0.6925,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5642857142857143,
+      "grad_norm": 0.5401285846482883,
+      "learning_rate": 1.8045977797666685e-05,
+      "loss": 0.6999,
+      "step": 1185
+    },
+    {
+      "epoch": 0.5666666666666667,
+      "grad_norm": 0.5677019239930577,
+      "learning_rate": 1.802123192755044e-05,
+      "loss": 0.6923,
+      "step": 1190
+    },
+    {
+      "epoch": 0.569047619047619,
+      "grad_norm": 0.5683306800833698,
+      "learning_rate": 1.7996347542498983e-05,
+      "loss": 0.7017,
+      "step": 1195
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.5113013942929165,
+      "learning_rate": 1.7971325072229227e-05,
+      "loss": 0.6921,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5738095238095238,
+      "grad_norm": 0.5841499935734249,
+      "learning_rate": 1.7946164948842604e-05,
+      "loss": 0.6915,
+      "step": 1205
+    },
+    {
+      "epoch": 0.5761904761904761,
+      "grad_norm": 0.6279630088060286,
+      "learning_rate": 1.7920867606817625e-05,
+      "loss": 0.7061,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5785714285714286,
+      "grad_norm": 0.5599484344769574,
+      "learning_rate": 1.7895433483002356e-05,
+      "loss": 0.715,
+      "step": 1215
+    },
+    {
+      "epoch": 0.580952380952381,
+      "grad_norm": 0.5661003553169632,
+      "learning_rate": 1.7869863016606893e-05,
+      "loss": 0.6986,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "grad_norm": 0.5716166266125108,
+      "learning_rate": 1.784415664919576e-05,
+      "loss": 0.7031,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5857142857142857,
+      "grad_norm": 0.5567338032160342,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.7068,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5880952380952381,
+      "grad_norm": 0.5683883961420689,
+      "learning_rate": 1.7792337989311e-05,
+      "loss": 0.689,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5904761904761905,
+      "grad_norm": 0.5477282163005394,
+      "learning_rate": 1.7766226591669787e-05,
+      "loss": 0.689,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5928571428571429,
+      "grad_norm": 0.5311691713580732,
+      "learning_rate": 1.7739981082662275e-05,
+      "loss": 0.7215,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5952380952380952,
+      "grad_norm": 0.598618505069467,
+      "learning_rate": 1.771360191551e-05,
+      "loss": 0.7151,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5976190476190476,
+      "grad_norm": 0.5543663594273839,
+      "learning_rate": 1.768708954574256e-05,
+      "loss": 0.701,
+      "step": 1255
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5321278870334822,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.6935,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6023809523809524,
+      "grad_norm": 0.5564736107581606,
+      "learning_rate": 1.7633667031973793e-05,
+      "loss": 0.6925,
+      "step": 1265
+    },
+    {
+      "epoch": 0.6047619047619047,
+      "grad_norm": 0.623003830977377,
+      "learning_rate": 1.760675781050109e-05,
+      "loss": 0.6904,
+      "step": 1270
+    },
+    {
+      "epoch": 0.6071428571428571,
+      "grad_norm": 0.5425831672559595,
+      "learning_rate": 1.757971723145453e-05,
+      "loss": 0.6941,
+      "step": 1275
+    },
+    {
+      "epoch": 0.6095238095238096,
+      "grad_norm": 0.5691326789130872,
+      "learning_rate": 1.755254576178535e-05,
+      "loss": 0.6943,
+      "step": 1280
+    },
+    {
+      "epoch": 0.611904761904762,
+      "grad_norm": 0.5472161245788499,
+      "learning_rate": 1.7525243870705052e-05,
+      "loss": 0.7019,
+      "step": 1285
+    },
+    {
+      "epoch": 0.6142857142857143,
+      "grad_norm": 0.5126814767906358,
+      "learning_rate": 1.7497812029677344e-05,
+      "loss": 0.6979,
+      "step": 1290
+    },
+    {
+      "epoch": 0.6166666666666667,
+      "grad_norm": 0.5808568867050167,
+      "learning_rate": 1.7470250712409963e-05,
+      "loss": 0.7017,
+      "step": 1295
+    },
+    {
+      "epoch": 0.6190476190476191,
+      "grad_norm": 0.5731714166045258,
+      "learning_rate": 1.7442560394846518e-05,
+      "loss": 0.6896,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6214285714285714,
+      "grad_norm": 0.5416282781910543,
+      "learning_rate": 1.741474155515827e-05,
+      "loss": 0.6881,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6238095238095238,
+      "grad_norm": 0.5989028487744685,
+      "learning_rate": 1.738679467373586e-05,
+      "loss": 0.6853,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6261904761904762,
+      "grad_norm": 0.5928143358170742,
+      "learning_rate": 1.7358720233181023e-05,
+      "loss": 0.6945,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6285714285714286,
+      "grad_norm": 0.5917504797279695,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.708,
+      "step": 1320
+    },
+    {
+      "epoch": 0.6309523809523809,
+      "grad_norm": 0.5735050094781514,
+      "learning_rate": 1.7302190616086464e-05,
+      "loss": 0.7048,
+      "step": 1325
+    },
+    {
+      "epoch": 0.6333333333333333,
+      "grad_norm": 0.57827020923432,
+      "learning_rate": 1.7273736415730488e-05,
+      "loss": 0.6793,
+      "step": 1330
+    },
+    {
+      "epoch": 0.6357142857142857,
+      "grad_norm": 0.5880932620123437,
+      "learning_rate": 1.7245156608592727e-05,
+      "loss": 0.6915,
+      "step": 1335
+    },
+    {
+      "epoch": 0.638095238095238,
+      "grad_norm": 0.5890769914796811,
+      "learning_rate": 1.7216451688204623e-05,
+      "loss": 0.69,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6404761904761904,
+      "grad_norm": 0.527942811449849,
+      "learning_rate": 1.718762215025813e-05,
+      "loss": 0.682,
+      "step": 1345
+    },
+    {
+      "epoch": 0.6428571428571429,
+      "grad_norm": 0.6301973133824051,
+      "learning_rate": 1.7158668492597186e-05,
+      "loss": 0.6934,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6452380952380953,
+      "grad_norm": 0.5410524060403492,
+      "learning_rate": 1.712959121520907e-05,
+      "loss": 0.6892,
+      "step": 1355
+    },
+    {
+      "epoch": 0.6476190476190476,
+      "grad_norm": 0.7801149059212953,
+      "learning_rate": 1.7100390820215805e-05,
+      "loss": 0.6999,
+      "step": 1360
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5546640495559166,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.6742,
+      "step": 1365
+    },
+    {
+      "epoch": 0.6523809523809524,
+      "grad_norm": 0.5722087188570253,
+      "learning_rate": 1.704162269652352e-05,
+      "loss": 0.7179,
+      "step": 1370
+    },
+    {
+      "epoch": 0.6547619047619048,
+      "grad_norm": 0.5249847693462395,
+      "learning_rate": 1.701205598266398e-05,
+      "loss": 0.6945,
+      "step": 1375
+    },
+    {
+      "epoch": 0.6571428571428571,
+      "grad_norm": 0.5639053989400045,
+      "learning_rate": 1.698236818086073e-05,
+      "loss": 0.6925,
+      "step": 1380
+    },
+    {
+      "epoch": 0.6595238095238095,
+      "grad_norm": 0.5897582949134731,
+      "learning_rate": 1.6952559803778656e-05,
+      "loss": 0.6958,
+      "step": 1385
+    },
+    {
+      "epoch": 0.6619047619047619,
+      "grad_norm": 0.5506537272905166,
+      "learning_rate": 1.6922631366164795e-05,
+      "loss": 0.6744,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6642857142857143,
+      "grad_norm": 0.5613120451869654,
+      "learning_rate": 1.689258338483947e-05,
+      "loss": 0.687,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.5687059713184212,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.7011,
+      "step": 1400
+    },
+    {
+      "epoch": 0.669047619047619,
+      "grad_norm": 0.5267403463210103,
+      "learning_rate": 1.683213086864843e-05,
+      "loss": 0.6903,
+      "step": 1405
+    },
+    {
+      "epoch": 0.6714285714285714,
+      "grad_norm": 0.5655713255912641,
+      "learning_rate": 1.6801727377709195e-05,
+      "loss": 0.6896,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6738095238095239,
+      "grad_norm": 0.5285552000123795,
+      "learning_rate": 1.6771206430893408e-05,
+      "loss": 0.6946,
+      "step": 1415
+    },
+    {
+      "epoch": 0.6761904761904762,
+      "grad_norm": 0.5761004876016846,
+      "learning_rate": 1.6740568555253153e-05,
+      "loss": 0.6895,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6785714285714286,
+      "grad_norm": 0.5390507685841307,
+      "learning_rate": 1.67098142798597e-05,
+      "loss": 0.6942,
+      "step": 1425
+    },
+    {
+      "epoch": 0.680952380952381,
+      "grad_norm": 0.5132897821784491,
+      "learning_rate": 1.6678944135794375e-05,
+      "loss": 0.6948,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6833333333333333,
+      "grad_norm": 0.5469471858786907,
+      "learning_rate": 1.6647958656139377e-05,
+      "loss": 0.6944,
+      "step": 1435
+    },
+    {
+      "epoch": 0.6857142857142857,
+      "grad_norm": 0.553201497920686,
+      "learning_rate": 1.6616858375968596e-05,
+      "loss": 0.6788,
+      "step": 1440
+    },
+    {
+      "epoch": 0.6880952380952381,
+      "grad_norm": 0.5623963206097778,
+      "learning_rate": 1.6585643832338342e-05,
+      "loss": 0.6815,
+      "step": 1445
+    },
+    {
+      "epoch": 0.6904761904761905,
+      "grad_norm": 0.5778391136516833,
+      "learning_rate": 1.6554315564278102e-05,
+      "loss": 0.6843,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6928571428571428,
+      "grad_norm": 0.5551499013662371,
+      "learning_rate": 1.6522874112781213e-05,
+      "loss": 0.6872,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6952380952380952,
+      "grad_norm": 0.5309204279264464,
+      "learning_rate": 1.649132002079552e-05,
+      "loss": 0.6873,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6976190476190476,
+      "grad_norm": 0.508876104576006,
+      "learning_rate": 1.645965383321401e-05,
+      "loss": 0.6792,
+      "step": 1465
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5802541953819853,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.6881,
+      "step": 1470
+    },
+    {
+      "epoch": 0.7023809523809523,
+      "grad_norm": 0.5641741696009784,
+      "learning_rate": 1.6395987360504667e-05,
+      "loss": 0.6789,
+      "step": 1475
+    },
+    {
+      "epoch": 0.7047619047619048,
+      "grad_norm": 0.5366027622698923,
+      "learning_rate": 1.6363988174803638e-05,
+      "loss": 0.686,
+      "step": 1480
+    },
+    {
+      "epoch": 0.7071428571428572,
+      "grad_norm": 0.5234128778556285,
+      "learning_rate": 1.6331879092341402e-05,
+      "loss": 0.7134,
+      "step": 1485
+    },
+    {
+      "epoch": 0.7095238095238096,
+      "grad_norm": 0.5759728577947257,
+      "learning_rate": 1.6299660667594814e-05,
+      "loss": 0.701,
+      "step": 1490
+    },
+    {
+      "epoch": 0.7119047619047619,
+      "grad_norm": 0.5756551282837324,
+      "learning_rate": 1.626733345692892e-05,
+      "loss": 0.6843,
+      "step": 1495
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 0.5426722958339261,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.6906,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7166666666666667,
+      "grad_norm": 0.48554563250860505,
+      "learning_rate": 1.6202354912682602e-05,
+      "loss": 0.6824,
+      "step": 1505
+    },
+    {
+      "epoch": 0.719047619047619,
+      "grad_norm": 0.5744429442854848,
+      "learning_rate": 1.6169704701186528e-05,
+      "loss": 0.6979,
+      "step": 1510
+    },
+    {
+      "epoch": 0.7214285714285714,
+      "grad_norm": 0.5277461906754672,
+      "learning_rate": 1.6136947947920477e-05,
+      "loss": 0.682,
+      "step": 1515
+    },
+    {
+      "epoch": 0.7238095238095238,
+      "grad_norm": 0.5681961605460194,
+      "learning_rate": 1.6104085218545633e-05,
+      "loss": 0.6893,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7261904761904762,
+      "grad_norm": 0.58856327424583,
+      "learning_rate": 1.6071117080553236e-05,
+      "loss": 0.694,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7285714285714285,
+      "grad_norm": 0.5585907608313626,
+      "learning_rate": 1.6038044103254775e-05,
+      "loss": 0.6807,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7309523809523809,
+      "grad_norm": 0.4975987871579692,
+      "learning_rate": 1.600486685777216e-05,
+      "loss": 0.6839,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 0.5683357922839659,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.6924,
+      "step": 1540
+    },
+    {
+      "epoch": 0.7357142857142858,
+      "grad_norm": 0.5507550407504584,
+      "learning_rate": 1.5938201855735017e-05,
+      "loss": 0.6952,
+      "step": 1545
+    },
+    {
+      "epoch": 0.7380952380952381,
+      "grad_norm": 0.5184672194625306,
+      "learning_rate": 1.5904715250387498e-05,
+      "loss": 0.6864,
+      "step": 1550
+    },
+    {
+      "epoch": 0.7404761904761905,
+      "grad_norm": 0.5315141830604754,
+      "learning_rate": 1.5871126679249977e-05,
+      "loss": 0.686,
+      "step": 1555
+    },
+    {
+      "epoch": 0.7428571428571429,
+      "grad_norm": 0.5292902163348473,
+      "learning_rate": 1.5837436722347902e-05,
+      "loss": 0.6826,
+      "step": 1560
+    },
+    {
+      "epoch": 0.7452380952380953,
+      "grad_norm": 0.570110548449842,
+      "learning_rate": 1.5803645961457522e-05,
+      "loss": 0.6792,
+      "step": 1565
+    },
+    {
+      "epoch": 0.7476190476190476,
+      "grad_norm": 0.5543717733657283,
+      "learning_rate": 1.576975498009583e-05,
+      "loss": 0.6754,
+      "step": 1570
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5222973089292835,
+      "learning_rate": 1.573576436351046e-05,
+      "loss": 0.6902,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7523809523809524,
+      "grad_norm": 0.5112696664324252,
+      "learning_rate": 1.570167469866962e-05,
+      "loss": 0.6972,
+      "step": 1580
+    },
+    {
+      "epoch": 0.7547619047619047,
+      "grad_norm": 0.5807388645411898,
+      "learning_rate": 1.5667486574251916e-05,
+      "loss": 0.6863,
+      "step": 1585
+    },
+    {
+      "epoch": 0.7571428571428571,
+      "grad_norm": 0.5522196746267701,
+      "learning_rate": 1.563320058063622e-05,
+      "loss": 0.6705,
+      "step": 1590
+    },
+    {
+      "epoch": 0.7595238095238095,
+      "grad_norm": 0.5302665914230218,
+      "learning_rate": 1.5598817309891466e-05,
+      "loss": 0.6842,
+      "step": 1595
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "grad_norm": 0.5286809702848861,
+      "learning_rate": 1.5564337355766412e-05,
+      "loss": 0.6976,
+      "step": 1600
+    },
+    {
+      "epoch": 0.7642857142857142,
+      "grad_norm": 0.5393439786479364,
+      "learning_rate": 1.5529761313679396e-05,
+      "loss": 0.7063,
+      "step": 1605
+    },
+    {
+      "epoch": 0.7666666666666667,
+      "grad_norm": 0.479895247668522,
+      "learning_rate": 1.5495089780708062e-05,
+      "loss": 0.6737,
+      "step": 1610
+    },
+    {
+      "epoch": 0.7690476190476191,
+      "grad_norm": 0.5401818745358772,
+      "learning_rate": 1.5460323355579035e-05,
+      "loss": 0.6808,
+      "step": 1615
+    },
+    {
+      "epoch": 0.7714285714285715,
+      "grad_norm": 0.6132409197814304,
+      "learning_rate": 1.5425462638657597e-05,
+      "loss": 0.6912,
+      "step": 1620
+    },
+    {
+      "epoch": 0.7738095238095238,
+      "grad_norm": 0.5605359051745877,
+      "learning_rate": 1.53905082319373e-05,
+      "loss": 0.6912,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7761904761904762,
+      "grad_norm": 0.5095591675298641,
+      "learning_rate": 1.5355460739029585e-05,
+      "loss": 0.6854,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7785714285714286,
+      "grad_norm": 0.5124981468976912,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.6805,
+      "step": 1635
+    },
+    {
+      "epoch": 0.780952380952381,
+      "grad_norm": 0.5543805272032538,
+      "learning_rate": 1.5285088917124555e-05,
+      "loss": 0.6872,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7833333333333333,
+      "grad_norm": 0.5240863656017781,
+      "learning_rate": 1.5249765803345602e-05,
+      "loss": 0.6812,
+      "step": 1645
+    },
+    {
+      "epoch": 0.7857142857142857,
+      "grad_norm": 0.5597831529794077,
+      "learning_rate": 1.5214352033794981e-05,
+      "loss": 0.6838,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7880952380952381,
+      "grad_norm": 0.5239372373145886,
+      "learning_rate": 1.517884822001666e-05,
+      "loss": 0.6884,
+      "step": 1655
+    },
+    {
+      "epoch": 0.7904761904761904,
+      "grad_norm": 0.5776231529708008,
+      "learning_rate": 1.5143254975109538e-05,
+      "loss": 0.6728,
+      "step": 1660
+    },
+    {
+      "epoch": 0.7928571428571428,
+      "grad_norm": 0.5215371586345835,
+      "learning_rate": 1.5107572913716859e-05,
+      "loss": 0.6748,
+      "step": 1665
+    },
+    {
+      "epoch": 0.7952380952380952,
+      "grad_norm": 0.5282942517891518,
+      "learning_rate": 1.5071802652015592e-05,
+      "loss": 0.6713,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7976190476190477,
+      "grad_norm": 0.5541374243554975,
+      "learning_rate": 1.503594480770581e-05,
+      "loss": 0.6597,
+      "step": 1675
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5521158876521693,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.686,
+      "step": 1680
+    },
+    {
+      "epoch": 0.8023809523809524,
+      "grad_norm": 0.5683308647861297,
+      "learning_rate": 1.496396884961238e-05,
+      "loss": 0.6693,
+      "step": 1685
+    },
+    {
+      "epoch": 0.8047619047619048,
+      "grad_norm": 0.573228337305753,
+      "learning_rate": 1.4927851978748177e-05,
+      "loss": 0.6926,
+      "step": 1690
+    },
+    {
+      "epoch": 0.8071428571428572,
+      "grad_norm": 0.521492632745778,
+      "learning_rate": 1.4891650011092896e-05,
+      "loss": 0.6947,
+      "step": 1695
+    },
+    {
+      "epoch": 0.8095238095238095,
+      "grad_norm": 0.5236798235457268,
+      "learning_rate": 1.4855363571801523e-05,
+      "loss": 0.679,
+      "step": 1700
+    },
+    {
+      "epoch": 0.8119047619047619,
+      "grad_norm": 0.5183500676344015,
+      "learning_rate": 1.481899328748776e-05,
+      "loss": 0.6657,
+      "step": 1705
+    },
+    {
+      "epoch": 0.8142857142857143,
+      "grad_norm": 0.5018602569821992,
+      "learning_rate": 1.4782539786213184e-05,
+      "loss": 0.6836,
+      "step": 1710
+    },
+    {
+      "epoch": 0.8166666666666667,
+      "grad_norm": 0.5469580528625957,
+      "learning_rate": 1.4746003697476406e-05,
+      "loss": 0.6841,
+      "step": 1715
+    },
+    {
+      "epoch": 0.819047619047619,
+      "grad_norm": 0.5610880961409519,
+      "learning_rate": 1.4709385652202204e-05,
+      "loss": 0.6864,
+      "step": 1720
+    },
+    {
+      "epoch": 0.8214285714285714,
+      "grad_norm": 0.5376175612680147,
+      "learning_rate": 1.4672686282730622e-05,
+      "loss": 0.6709,
+      "step": 1725
+    },
+    {
+      "epoch": 0.8238095238095238,
+      "grad_norm": 0.49458630831613376,
+      "learning_rate": 1.4635906222806058e-05,
+      "loss": 0.6832,
+      "step": 1730
+    },
+    {
+      "epoch": 0.8261904761904761,
+      "grad_norm": 0.5502214264070843,
+      "learning_rate": 1.4599046107566314e-05,
+      "loss": 0.68,
+      "step": 1735
+    },
+    {
+      "epoch": 0.8285714285714286,
+      "grad_norm": 0.5611151303400349,
+      "learning_rate": 1.4562106573531632e-05,
+      "loss": 0.698,
+      "step": 1740
+    },
+    {
+      "epoch": 0.830952380952381,
+      "grad_norm": 0.5735560840092161,
+      "learning_rate": 1.4525088258593695e-05,
+      "loss": 0.6839,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.5510543712119496,
+      "learning_rate": 1.4487991802004625e-05,
+      "loss": 0.6911,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8357142857142857,
+      "grad_norm": 0.5018246480371149,
+      "learning_rate": 1.4450817844365924e-05,
+      "loss": 0.6752,
+      "step": 1755
+    },
+    {
+      "epoch": 0.8380952380952381,
+      "grad_norm": 0.5016576788767521,
+      "learning_rate": 1.4413567027617442e-05,
+      "loss": 0.683,
+      "step": 1760
+    },
+    {
+      "epoch": 0.8404761904761905,
+      "grad_norm": 0.5384327340092925,
+      "learning_rate": 1.4376239995026254e-05,
+      "loss": 0.6941,
+      "step": 1765
+    },
+    {
+      "epoch": 0.8428571428571429,
+      "grad_norm": 0.5344821729910321,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.6754,
+      "step": 1770
+    },
+    {
+      "epoch": 0.8452380952380952,
+      "grad_norm": 0.529015386728389,
+      "learning_rate": 1.4301359861953652e-05,
+      "loss": 0.6718,
+      "step": 1775
+    },
+    {
+      "epoch": 0.8476190476190476,
+      "grad_norm": 0.5467267886434187,
+      "learning_rate": 1.4263808054542541e-05,
+      "loss": 0.6885,
+      "step": 1780
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5209380288088742,
+      "learning_rate": 1.4226182617406996e-05,
+      "loss": 0.6725,
+      "step": 1785
+    },
+    {
+      "epoch": 0.8523809523809524,
+      "grad_norm": 0.5120625589068878,
+      "learning_rate": 1.418848420028325e-05,
+      "loss": 0.678,
+      "step": 1790
+    },
+    {
+      "epoch": 0.8547619047619047,
+      "grad_norm": 0.5177286158188428,
+      "learning_rate": 1.4150713454167788e-05,
+      "loss": 0.6795,
+      "step": 1795
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.5180241473157509,
+      "learning_rate": 1.4112871031306118e-05,
+      "loss": 0.685,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8595238095238096,
+      "grad_norm": 0.5376670786030856,
+      "learning_rate": 1.4074957585181488e-05,
+      "loss": 0.6823,
+      "step": 1805
+    },
+    {
+      "epoch": 0.861904761904762,
+      "grad_norm": 0.5159889497243569,
+      "learning_rate": 1.4036973770503623e-05,
+      "loss": 0.6684,
+      "step": 1810
+    },
+    {
+      "epoch": 0.8642857142857143,
+      "grad_norm": 0.5328385205015794,
+      "learning_rate": 1.3998920243197408e-05,
+      "loss": 0.6785,
+      "step": 1815
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 0.5596052359171065,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.6971,
+      "step": 1820
+    },
+    {
+      "epoch": 0.8690476190476191,
+      "grad_norm": 0.5476844927900769,
+      "learning_rate": 1.3922606680407307e-05,
+      "loss": 0.6843,
+      "step": 1825
+    },
+    {
+      "epoch": 0.8714285714285714,
+      "grad_norm": 0.49918261945442965,
+      "learning_rate": 1.3884347962746949e-05,
+      "loss": 0.6761,
+      "step": 1830
+    },
+    {
+      "epoch": 0.8738095238095238,
+      "grad_norm": 0.5116149514821221,
+      "learning_rate": 1.3846022168082553e-05,
+      "loss": 0.6763,
+      "step": 1835
+    },
+    {
+      "epoch": 0.8761904761904762,
+      "grad_norm": 0.5280902250134181,
+      "learning_rate": 1.3807629958244498e-05,
+      "loss": 0.667,
+      "step": 1840
+    },
+    {
+      "epoch": 0.8785714285714286,
+      "grad_norm": 0.48857909294423746,
+      "learning_rate": 1.3769171996210053e-05,
+      "loss": 0.6707,
+      "step": 1845
+    },
+    {
+      "epoch": 0.8809523809523809,
+      "grad_norm": 0.4975153014854871,
+      "learning_rate": 1.373064894609194e-05,
+      "loss": 0.688,
+      "step": 1850
+    },
+    {
+      "epoch": 0.8833333333333333,
+      "grad_norm": 0.5206570618857133,
+      "learning_rate": 1.3692061473126845e-05,
+      "loss": 0.687,
+      "step": 1855
+    },
+    {
+      "epoch": 0.8857142857142857,
+      "grad_norm": 0.47088196958418377,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.6659,
+      "step": 1860
+    },
+    {
+      "epoch": 0.888095238095238,
+      "grad_norm": 0.5098589712869,
+      "learning_rate": 1.361469592515342e-05,
+      "loss": 0.6784,
+      "step": 1865
+    },
+    {
+      "epoch": 0.8904761904761904,
+      "grad_norm": 0.47648371702756814,
+      "learning_rate": 1.3575919186134862e-05,
+      "loss": 0.682,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8928571428571429,
+      "grad_norm": 0.5003536067341952,
+      "learning_rate": 1.3537080696225815e-05,
+      "loss": 0.6671,
+      "step": 1875
+    },
+    {
+      "epoch": 0.8952380952380953,
+      "grad_norm": 0.519430135596028,
+      "learning_rate": 1.349818112611015e-05,
+      "loss": 0.6885,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8976190476190476,
+      "grad_norm": 0.49741488661881644,
+      "learning_rate": 1.3459221147526504e-05,
+      "loss": 0.6754,
+      "step": 1885
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5408088216797241,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.6757,
+      "step": 1890
+    },
+    {
+      "epoch": 0.9023809523809524,
+      "grad_norm": 0.502061784820922,
+      "learning_rate": 1.3381122657114059e-05,
+      "loss": 0.669,
+      "step": 1895
+    },
+    {
+      "epoch": 0.9047619047619048,
+      "grad_norm": 0.5031558920583495,
+      "learning_rate": 1.3341985493931877e-05,
+      "loss": 0.6756,
+      "step": 1900
+    },
+    {
+      "epoch": 0.9071428571428571,
+      "grad_norm": 0.45651178296082845,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.6629,
+      "step": 1905
+    },
+    {
+      "epoch": 0.9095238095238095,
+      "grad_norm": 0.5111760063303825,
+      "learning_rate": 1.3263538710811559e-05,
+      "loss": 0.6787,
+      "step": 1910
+    },
+    {
+      "epoch": 0.9119047619047619,
+      "grad_norm": 0.5038700388575827,
+      "learning_rate": 1.3224230445534544e-05,
+      "loss": 0.6814,
+      "step": 1915
+    },
+    {
+      "epoch": 0.9142857142857143,
+      "grad_norm": 0.5023414779956755,
+      "learning_rate": 1.3184866502516846e-05,
+      "loss": 0.6859,
+      "step": 1920
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "grad_norm": 0.5154019438212942,
+      "learning_rate": 1.3145447561516138e-05,
+      "loss": 0.6717,
+      "step": 1925
+    },
+    {
+      "epoch": 0.919047619047619,
+      "grad_norm": 0.5621900790890961,
+      "learning_rate": 1.3105974303239838e-05,
+      "loss": 0.7009,
+      "step": 1930
+    },
+    {
+      "epoch": 0.9214285714285714,
+      "grad_norm": 0.5137612757351608,
+      "learning_rate": 1.3066447409333345e-05,
+      "loss": 0.6957,
+      "step": 1935
+    },
+    {
+      "epoch": 0.9238095238095239,
+      "grad_norm": 0.509045076292474,
+      "learning_rate": 1.3026867562368262e-05,
+      "loss": 0.6804,
+      "step": 1940
+    },
+    {
+      "epoch": 0.9261904761904762,
+      "grad_norm": 0.5067359253184963,
+      "learning_rate": 1.2987235445830612e-05,
+      "loss": 0.6707,
+      "step": 1945
+    },
+    {
+      "epoch": 0.9285714285714286,
+      "grad_norm": 0.5488540810447187,
+      "learning_rate": 1.2947551744109044e-05,
+      "loss": 0.6784,
+      "step": 1950
+    },
+    {
+      "epoch": 0.930952380952381,
+      "grad_norm": 0.49916198703229403,
+      "learning_rate": 1.2907817142483002e-05,
+      "loss": 0.6533,
+      "step": 1955
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 0.5306836509898268,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 0.6694,
+      "step": 1960
+    },
+    {
+      "epoch": 0.9357142857142857,
+      "grad_norm": 0.5192454887104005,
+      "learning_rate": 1.2828197985018276e-05,
+      "loss": 0.6814,
+      "step": 1965
+    },
+    {
+      "epoch": 0.9380952380952381,
+      "grad_norm": 0.48011848457133666,
+      "learning_rate": 1.2788314804085904e-05,
+      "loss": 0.6656,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9404761904761905,
+      "grad_norm": 0.501551598406461,
+      "learning_rate": 1.2748383473037948e-05,
+      "loss": 0.6709,
+      "step": 1975
+    },
+    {
+      "epoch": 0.9428571428571428,
+      "grad_norm": 0.4798488372067313,
+      "learning_rate": 1.2708404681430054e-05,
+      "loss": 0.6884,
+      "step": 1980
+    },
+    {
+      "epoch": 0.9452380952380952,
+      "grad_norm": 0.4954430271705226,
+      "learning_rate": 1.266837911963743e-05,
+      "loss": 0.6789,
+      "step": 1985
+    },
+    {
+      "epoch": 0.9476190476190476,
+      "grad_norm": 0.5134775709739979,
+      "learning_rate": 1.2628307478842955e-05,
+      "loss": 0.6743,
+      "step": 1990
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4932466173893802,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.6741,
+      "step": 1995
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 0.49375841486555283,
+      "learning_rate": 1.2548028728946548e-05,
+      "loss": 0.682,
+      "step": 2000
+    },
+    {
+      "epoch": 0.9547619047619048,
+      "grad_norm": 0.553381199639126,
+      "learning_rate": 1.2507823006141128e-05,
+      "loss": 0.704,
+      "step": 2005
+    },
+    {
+      "epoch": 0.9571428571428572,
+      "grad_norm": 0.5286463574396035,
+      "learning_rate": 1.2467573976902936e-05,
+      "loss": 0.6642,
+      "step": 2010
+    },
+    {
+      "epoch": 0.9595238095238096,
+      "grad_norm": 0.5330908591194152,
+      "learning_rate": 1.24272823362738e-05,
+      "loss": 0.6973,
+      "step": 2015
+    },
+    {
+      "epoch": 0.9619047619047619,
+      "grad_norm": 0.5338462915299015,
+      "learning_rate": 1.238694878003138e-05,
+      "loss": 0.6608,
+      "step": 2020
+    },
+    {
+      "epoch": 0.9642857142857143,
+      "grad_norm": 0.5626548093456667,
+      "learning_rate": 1.2346574004677154e-05,
+      "loss": 0.6807,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9666666666666667,
+      "grad_norm": 0.5363264212587037,
+      "learning_rate": 1.2306158707424402e-05,
+      "loss": 0.6742,
+      "step": 2030
+    },
+    {
+      "epoch": 0.969047619047619,
+      "grad_norm": 0.5223504672217075,
+      "learning_rate": 1.2265703586186158e-05,
+      "loss": 0.6756,
+      "step": 2035
+    },
+    {
+      "epoch": 0.9714285714285714,
+      "grad_norm": 0.5026492923180158,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.6857,
+      "step": 2040
+    },
+    {
+      "epoch": 0.9738095238095238,
+      "grad_norm": 0.4958319271212678,
+      "learning_rate": 1.2184676666831741e-05,
+      "loss": 0.6638,
+      "step": 2045
+    },
+    {
+      "epoch": 0.9761904761904762,
+      "grad_norm": 0.49120488962765346,
+      "learning_rate": 1.2144106267931877e-05,
+      "loss": 0.6767,
+      "step": 2050
+    },
+    {
+      "epoch": 0.9785714285714285,
+      "grad_norm": 0.56338802960644,
+      "learning_rate": 1.210349884345496e-05,
+      "loss": 0.6657,
+      "step": 2055
+    },
+    {
+      "epoch": 0.9809523809523809,
+      "grad_norm": 0.5506153013895527,
+      "learning_rate": 1.2062855094631777e-05,
+      "loss": 0.6708,
+      "step": 2060
+    },
+    {
+      "epoch": 0.9833333333333333,
+      "grad_norm": 0.4911898234196216,
+      "learning_rate": 1.2022175723320382e-05,
+      "loss": 0.6736,
+      "step": 2065
+    },
+    {
+      "epoch": 0.9857142857142858,
+      "grad_norm": 0.5037996502262528,
+      "learning_rate": 1.1981461431993978e-05,
+      "loss": 0.6613,
+      "step": 2070
+    },
+    {
+      "epoch": 0.9880952380952381,
+      "grad_norm": 0.48983503414857615,
+      "learning_rate": 1.1940712923728784e-05,
+      "loss": 0.6705,
+      "step": 2075
+    },
+    {
+      "epoch": 0.9904761904761905,
+      "grad_norm": 0.5089658881380381,
+      "learning_rate": 1.1899930902191904e-05,
+      "loss": 0.6554,
+      "step": 2080
+    },
+    {
+      "epoch": 0.9928571428571429,
+      "grad_norm": 0.5353432851904248,
+      "learning_rate": 1.1859116071629148e-05,
+      "loss": 0.6713,
+      "step": 2085
+    },
+    {
+      "epoch": 0.9952380952380953,
+      "grad_norm": 0.5083284406725272,
+      "learning_rate": 1.181826913685291e-05,
+      "loss": 0.6593,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9976190476190476,
+      "grad_norm": 0.5338566827047485,
+      "learning_rate": 1.1777390803229964e-05,
+      "loss": 0.6771,
+      "step": 2095
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5259573856012386,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.6603,
+      "step": 2100
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.7571888566017151,
+      "eval_runtime": 110.7511,
+      "eval_samples_per_second": 77.299,
+      "eval_steps_per_second": 1.21,
+      "step": 2100
+    },
+    {
+      "epoch": 1.0023809523809524,
+      "grad_norm": 0.5901038052706317,
+      "learning_rate": 1.1695542763609944e-05,
+      "loss": 0.5957,
+      "step": 2105
+    },
+    {
+      "epoch": 1.0047619047619047,
+      "grad_norm": 0.5438749703677144,
+      "learning_rate": 1.1654574471008712e-05,
+      "loss": 0.5972,
+      "step": 2110
+    },
+    {
+      "epoch": 1.0071428571428571,
+      "grad_norm": 0.5432897226752162,
+      "learning_rate": 1.1613577606328068e-05,
+      "loss": 0.595,
+      "step": 2115
+    },
+    {
+      "epoch": 1.0095238095238095,
+      "grad_norm": 0.5692170924338006,
+      "learning_rate": 1.1572552877523855e-05,
+      "loss": 0.5869,
+      "step": 2120
+    },
+    {
+      "epoch": 1.0119047619047619,
+      "grad_norm": 0.5633601882768045,
+      "learning_rate": 1.1531500993033094e-05,
+      "loss": 0.59,
+      "step": 2125
+    },
+    {
+      "epoch": 1.0142857142857142,
+      "grad_norm": 0.5402220697166501,
+      "learning_rate": 1.1490422661761744e-05,
+      "loss": 0.5928,
+      "step": 2130
+    },
+    {
+      "epoch": 1.0166666666666666,
+      "grad_norm": 0.6014408487693759,
+      "learning_rate": 1.1449318593072468e-05,
+      "loss": 0.5943,
+      "step": 2135
+    },
+    {
+      "epoch": 1.019047619047619,
+      "grad_norm": 0.48300700922294626,
+      "learning_rate": 1.1408189496772369e-05,
+      "loss": 0.593,
+      "step": 2140
+    },
+    {
+      "epoch": 1.0214285714285714,
+      "grad_norm": 0.5461515536108149,
+      "learning_rate": 1.1367036083100735e-05,
+      "loss": 0.6118,
+      "step": 2145
+    },
+    {
+      "epoch": 1.0238095238095237,
+      "grad_norm": 0.5225774711520927,
+      "learning_rate": 1.1325859062716795e-05,
+      "loss": 0.5934,
+      "step": 2150
+    },
+    {
+      "epoch": 1.026190476190476,
+      "grad_norm": 0.5652770991249514,
+      "learning_rate": 1.1284659146687416e-05,
+      "loss": 0.5906,
+      "step": 2155
+    },
+    {
+      "epoch": 1.0285714285714285,
+      "grad_norm": 0.49282124739115296,
+      "learning_rate": 1.1243437046474854e-05,
+      "loss": 0.6029,
+      "step": 2160
+    },
+    {
+      "epoch": 1.0309523809523808,
+      "grad_norm": 0.517411950290186,
+      "learning_rate": 1.120219347392444e-05,
+      "loss": 0.6105,
+      "step": 2165
+    },
+    {
+      "epoch": 1.0333333333333334,
+      "grad_norm": 0.5547803815907365,
+      "learning_rate": 1.1160929141252303e-05,
+      "loss": 0.5888,
+      "step": 2170
+    },
+    {
+      "epoch": 1.0357142857142858,
+      "grad_norm": 0.5448805389067126,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.606,
+      "step": 2175
+    },
+    {
+      "epoch": 1.0380952380952382,
+      "grad_norm": 0.5270587508515787,
+      "learning_rate": 1.1078341046187588e-05,
+      "loss": 0.5978,
+      "step": 2180
+    },
+    {
+      "epoch": 1.0404761904761906,
+      "grad_norm": 0.5470808129733004,
+      "learning_rate": 1.1037018709970528e-05,
+      "loss": 0.6079,
+      "step": 2185
+    },
+    {
+      "epoch": 1.042857142857143,
+      "grad_norm": 0.5185275354171291,
+      "learning_rate": 1.0995678465958168e-05,
+      "loss": 0.5998,
+      "step": 2190
+    },
+    {
+      "epoch": 1.0452380952380953,
+      "grad_norm": 0.5222569126782542,
+      "learning_rate": 1.0954321028036013e-05,
+      "loss": 0.6065,
+      "step": 2195
+    },
+    {
+      "epoch": 1.0476190476190477,
+      "grad_norm": 0.5902869820958465,
+      "learning_rate": 1.0912947110386484e-05,
+      "loss": 0.5952,
+      "step": 2200
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5049079899744102,
+      "learning_rate": 1.0871557427476585e-05,
+      "loss": 0.5984,
+      "step": 2205
+    },
+    {
+      "epoch": 1.0523809523809524,
+      "grad_norm": 0.5211587946973806,
+      "learning_rate": 1.0830152694045553e-05,
+      "loss": 0.5822,
+      "step": 2210
+    },
+    {
+      "epoch": 1.0547619047619048,
+      "grad_norm": 0.5411611115879803,
+      "learning_rate": 1.078873362509254e-05,
+      "loss": 0.5882,
+      "step": 2215
+    },
+    {
+      "epoch": 1.0571428571428572,
+      "grad_norm": 0.5442183494774795,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.5996,
+      "step": 2220
+    },
+    {
+      "epoch": 1.0595238095238095,
+      "grad_norm": 0.5612089724718119,
+      "learning_rate": 1.0705855341842564e-05,
+      "loss": 0.6,
+      "step": 2225
+    },
+    {
+      "epoch": 1.061904761904762,
+      "grad_norm": 0.5293855485137681,
+      "learning_rate": 1.0664397558732245e-05,
+      "loss": 0.5918,
+      "step": 2230
+    },
+    {
+      "epoch": 1.0642857142857143,
+      "grad_norm": 0.5907188005151464,
+      "learning_rate": 1.0622928302448523e-05,
+      "loss": 0.608,
+      "step": 2235
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.5251701766583902,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 0.6058,
+      "step": 2240
+    },
+    {
+      "epoch": 1.069047619047619,
+      "grad_norm": 0.5250031426883407,
+      "learning_rate": 1.0539958235000075e-05,
+      "loss": 0.6064,
+      "step": 2245
+    },
+    {
+      "epoch": 1.0714285714285714,
+      "grad_norm": 0.5217021022246795,
+      "learning_rate": 1.0498458856606972e-05,
+      "loss": 0.6098,
+      "step": 2250
+    },
+    {
+      "epoch": 1.0738095238095238,
+      "grad_norm": 0.5671230242631112,
+      "learning_rate": 1.0456950870558982e-05,
+      "loss": 0.5985,
+      "step": 2255
+    },
+    {
+      "epoch": 1.0761904761904761,
+      "grad_norm": 0.5640807879300845,
+      "learning_rate": 1.0415434993638269e-05,
+      "loss": 0.5992,
+      "step": 2260
+    },
+    {
+      "epoch": 1.0785714285714285,
+      "grad_norm": 0.5349112926063765,
+      "learning_rate": 1.037391194276326e-05,
+      "loss": 0.6061,
+      "step": 2265
+    },
+    {
+      "epoch": 1.0809523809523809,
+      "grad_norm": 0.5592160403982473,
+      "learning_rate": 1.0332382434976267e-05,
+      "loss": 0.5864,
+      "step": 2270
+    },
+    {
+      "epoch": 1.0833333333333333,
+      "grad_norm": 0.5235545437475504,
+      "learning_rate": 1.0290847187431115e-05,
+      "loss": 0.592,
+      "step": 2275
+    },
+    {
+      "epoch": 1.0857142857142856,
+      "grad_norm": 0.556896126496801,
+      "learning_rate": 1.0249306917380731e-05,
+      "loss": 0.6005,
+      "step": 2280
+    },
+    {
+      "epoch": 1.088095238095238,
+      "grad_norm": 0.522494265444706,
+      "learning_rate": 1.0207762342164778e-05,
+      "loss": 0.5958,
+      "step": 2285
+    },
+    {
+      "epoch": 1.0904761904761904,
+      "grad_norm": 0.5307880240675681,
+      "learning_rate": 1.0166214179197265e-05,
+      "loss": 0.5911,
+      "step": 2290
+    },
+    {
+      "epoch": 1.092857142857143,
+      "grad_norm": 0.5364461659449388,
+      "learning_rate": 1.0124663145954152e-05,
+      "loss": 0.6018,
+      "step": 2295
+    },
+    {
+      "epoch": 1.0952380952380953,
+      "grad_norm": 0.5520100652972162,
+      "learning_rate": 1.0083109959960974e-05,
+      "loss": 0.5955,
+      "step": 2300
+    },
+    {
+      "epoch": 1.0976190476190477,
+      "grad_norm": 0.5020623533166694,
+      "learning_rate": 1.0041555338780427e-05,
+      "loss": 0.5961,
+      "step": 2305
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.52041207867341,
+      "learning_rate": 1e-05,
+      "loss": 0.5954,
+      "step": 2310
+    },
+    {
+      "epoch": 1.1023809523809525,
+      "grad_norm": 0.569101456193797,
+      "learning_rate": 9.958444661219578e-06,
+      "loss": 0.6046,
+      "step": 2315
+    },
+    {
+      "epoch": 1.1047619047619048,
+      "grad_norm": 0.5290911882364042,
+      "learning_rate": 9.916890040039031e-06,
+      "loss": 0.5891,
+      "step": 2320
+    },
+    {
+      "epoch": 1.1071428571428572,
+      "grad_norm": 0.5305645713582501,
+      "learning_rate": 9.87533685404585e-06,
+      "loss": 0.5988,
+      "step": 2325
+    },
+    {
+      "epoch": 1.1095238095238096,
+      "grad_norm": 0.5469543228815086,
+      "learning_rate": 9.833785820802739e-06,
+      "loss": 0.5957,
+      "step": 2330
+    },
+    {
+      "epoch": 1.111904761904762,
+      "grad_norm": 0.5324948019005903,
+      "learning_rate": 9.792237657835225e-06,
+      "loss": 0.5953,
+      "step": 2335
+    },
+    {
+      "epoch": 1.1142857142857143,
+      "grad_norm": 0.5304279403221299,
+      "learning_rate": 9.750693082619274e-06,
+      "loss": 0.5937,
+      "step": 2340
+    },
+    {
+      "epoch": 1.1166666666666667,
+      "grad_norm": 0.5485078647852204,
+      "learning_rate": 9.709152812568886e-06,
+      "loss": 0.5987,
+      "step": 2345
+    },
+    {
+      "epoch": 1.119047619047619,
+      "grad_norm": 0.552749009152491,
+      "learning_rate": 9.667617565023734e-06,
+      "loss": 0.5978,
+      "step": 2350
+    },
+    {
+      "epoch": 1.1214285714285714,
+      "grad_norm": 0.5177836960191152,
+      "learning_rate": 9.626088057236745e-06,
+      "loss": 0.5973,
+      "step": 2355
+    },
+    {
+      "epoch": 1.1238095238095238,
+      "grad_norm": 0.6109837673102856,
+      "learning_rate": 9.584565006361735e-06,
+      "loss": 0.595,
+      "step": 2360
+    },
+    {
+      "epoch": 1.1261904761904762,
+      "grad_norm": 0.5167360036746549,
+      "learning_rate": 9.543049129441021e-06,
+      "loss": 0.5891,
+      "step": 2365
+    },
+    {
+      "epoch": 1.1285714285714286,
+      "grad_norm": 0.5297524492676912,
+      "learning_rate": 9.501541143393028e-06,
+      "loss": 0.606,
+      "step": 2370
+    },
+    {
+      "epoch": 1.130952380952381,
+      "grad_norm": 0.5411957992836974,
+      "learning_rate": 9.460041764999929e-06,
+      "loss": 0.5973,
+      "step": 2375
+    },
+    {
+      "epoch": 1.1333333333333333,
+      "grad_norm": 0.5287203316976509,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 0.5848,
+      "step": 2380
+    },
+    {
+      "epoch": 1.1357142857142857,
+      "grad_norm": 0.6710026001993966,
+      "learning_rate": 9.377071697551479e-06,
+      "loss": 0.6086,
+      "step": 2385
+    },
+    {
+      "epoch": 1.138095238095238,
+      "grad_norm": 0.5503478041423361,
+      "learning_rate": 9.33560244126776e-06,
+      "loss": 0.5941,
+      "step": 2390
+    },
+    {
+      "epoch": 1.1404761904761904,
+      "grad_norm": 0.5347482917383127,
+      "learning_rate": 9.294144658157443e-06,
+      "loss": 0.5974,
+      "step": 2395
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.5497165323598144,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 0.6046,
+      "step": 2400
+    },
+    {
+      "epoch": 1.1452380952380952,
+      "grad_norm": 0.5190090017461467,
+      "learning_rate": 9.211266374907463e-06,
+      "loss": 0.5869,
+      "step": 2405
+    },
+    {
+      "epoch": 1.1476190476190475,
+      "grad_norm": 0.5719299143043652,
+      "learning_rate": 9.169847305954448e-06,
+      "loss": 0.6048,
+      "step": 2410
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5672551417094259,
+      "learning_rate": 9.128442572523418e-06,
+      "loss": 0.6068,
+      "step": 2415
+    },
+    {
+      "epoch": 1.1523809523809523,
+      "grad_norm": 0.5811171033851283,
+      "learning_rate": 9.087052889613519e-06,
+      "loss": 0.5982,
+      "step": 2420
+    },
+    {
+      "epoch": 1.1547619047619047,
+      "grad_norm": 0.5227291415650476,
+      "learning_rate": 9.045678971963988e-06,
+      "loss": 0.591,
+      "step": 2425
+    },
+    {
+      "epoch": 1.157142857142857,
+      "grad_norm": 0.5413206931558642,
+      "learning_rate": 9.004321534041836e-06,
+      "loss": 0.5916,
+      "step": 2430
+    },
+    {
+      "epoch": 1.1595238095238094,
+      "grad_norm": 0.5298903577250825,
+      "learning_rate": 8.962981290029475e-06,
+      "loss": 0.5883,
+      "step": 2435
+    },
+    {
+      "epoch": 1.161904761904762,
+      "grad_norm": 0.4876508870345226,
+      "learning_rate": 8.921658953812416e-06,
+      "loss": 0.5938,
+      "step": 2440
+    },
+    {
+      "epoch": 1.1642857142857144,
+      "grad_norm": 0.5198015834539063,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.5975,
+      "step": 2445
+    },
+    {
+      "epoch": 1.1666666666666667,
+      "grad_norm": 0.5097940796805619,
+      "learning_rate": 8.839070858747697e-06,
+      "loss": 0.5868,
+      "step": 2450
+    },
+    {
+      "epoch": 1.1690476190476191,
+      "grad_norm": 0.5663643871944489,
+      "learning_rate": 8.797806526075566e-06,
+      "loss": 0.5902,
+      "step": 2455
+    },
+    {
+      "epoch": 1.1714285714285715,
+      "grad_norm": 0.5396345415398657,
+      "learning_rate": 8.756562953525151e-06,
+      "loss": 0.5958,
+      "step": 2460
+    },
+    {
+      "epoch": 1.1738095238095239,
+      "grad_norm": 0.5577630837652126,
+      "learning_rate": 8.715340853312586e-06,
+      "loss": 0.5833,
+      "step": 2465
+    },
+    {
+      "epoch": 1.1761904761904762,
+      "grad_norm": 0.5518263276547959,
+      "learning_rate": 8.674140937283208e-06,
+      "loss": 0.5935,
+      "step": 2470
+    },
+    {
+      "epoch": 1.1785714285714286,
+      "grad_norm": 0.5084066290422407,
+      "learning_rate": 8.632963916899268e-06,
+      "loss": 0.5835,
+      "step": 2475
+    },
+    {
+      "epoch": 1.180952380952381,
+      "grad_norm": 0.5098303646711122,
+      "learning_rate": 8.591810503227634e-06,
+      "loss": 0.6024,
+      "step": 2480
+    },
+    {
+      "epoch": 1.1833333333333333,
+      "grad_norm": 0.5399045936688955,
+      "learning_rate": 8.550681406927534e-06,
+      "loss": 0.5916,
+      "step": 2485
+    },
+    {
+      "epoch": 1.1857142857142857,
+      "grad_norm": 0.5139645415633143,
+      "learning_rate": 8.509577338238255e-06,
+      "loss": 0.5867,
+      "step": 2490
+    },
+    {
+      "epoch": 1.188095238095238,
+      "grad_norm": 0.5670585522900756,
+      "learning_rate": 8.46849900696691e-06,
+      "loss": 0.6006,
+      "step": 2495
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 0.5457106741839642,
+      "learning_rate": 8.427447122476148e-06,
+      "loss": 0.597,
+      "step": 2500
+    },
+    {
+      "epoch": 1.1928571428571428,
+      "grad_norm": 0.5571733334835492,
+      "learning_rate": 8.386422393671934e-06,
+      "loss": 0.5913,
+      "step": 2505
+    },
+    {
+      "epoch": 1.1952380952380952,
+      "grad_norm": 0.529527429611943,
+      "learning_rate": 8.34542552899129e-06,
+      "loss": 0.6006,
+      "step": 2510
+    },
+    {
+      "epoch": 1.1976190476190476,
+      "grad_norm": 0.5235732448086102,
+      "learning_rate": 8.304457236390062e-06,
+      "loss": 0.5819,
+      "step": 2515
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5332920937716661,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.6024,
+      "step": 2520
+    },
+    {
+      "epoch": 1.2023809523809523,
+      "grad_norm": 0.5137113000682924,
+      "learning_rate": 8.222609196770037e-06,
+      "loss": 0.6001,
+      "step": 2525
+    },
+    {
+      "epoch": 1.2047619047619047,
+      "grad_norm": 0.5000093032244755,
+      "learning_rate": 8.181730863147094e-06,
+      "loss": 0.5839,
+      "step": 2530
+    },
+    {
+      "epoch": 1.207142857142857,
+      "grad_norm": 0.5084439515294099,
+      "learning_rate": 8.140883928370855e-06,
+      "loss": 0.5987,
+      "step": 2535
+    },
+    {
+      "epoch": 1.2095238095238094,
+      "grad_norm": 0.5420324353647344,
+      "learning_rate": 8.100069097808103e-06,
+      "loss": 0.5918,
+      "step": 2540
+    },
+    {
+      "epoch": 1.2119047619047618,
+      "grad_norm": 0.5051962888682799,
+      "learning_rate": 8.059287076271216e-06,
+      "loss": 0.5929,
+      "step": 2545
+    },
+    {
+      "epoch": 1.2142857142857142,
+      "grad_norm": 0.5112423923421076,
+      "learning_rate": 8.018538568006027e-06,
+      "loss": 0.5851,
+      "step": 2550
+    },
+    {
+      "epoch": 1.2166666666666668,
+      "grad_norm": 0.5404164861280687,
+      "learning_rate": 7.977824276679623e-06,
+      "loss": 0.6097,
+      "step": 2555
+    },
+    {
+      "epoch": 1.2190476190476192,
+      "grad_norm": 0.5318602860411683,
+      "learning_rate": 7.937144905368226e-06,
+      "loss": 0.5952,
+      "step": 2560
+    },
+    {
+      "epoch": 1.2214285714285715,
+      "grad_norm": 0.5521690395673133,
+      "learning_rate": 7.896501156545044e-06,
+      "loss": 0.594,
+      "step": 2565
+    },
+    {
+      "epoch": 1.223809523809524,
+      "grad_norm": 0.4999490189546094,
+      "learning_rate": 7.855893732068124e-06,
+      "loss": 0.5934,
+      "step": 2570
+    },
+    {
+      "epoch": 1.2261904761904763,
+      "grad_norm": 0.5048993618252382,
+      "learning_rate": 7.815323333168262e-06,
+      "loss": 0.5959,
+      "step": 2575
+    },
+    {
+      "epoch": 1.2285714285714286,
+      "grad_norm": 0.5109949262663191,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.5988,
+      "step": 2580
+    },
+    {
+      "epoch": 1.230952380952381,
+      "grad_norm": 0.5143824021233596,
+      "learning_rate": 7.734296413813847e-06,
+      "loss": 0.5844,
+      "step": 2585
+    },
+    {
+      "epoch": 1.2333333333333334,
+      "grad_norm": 0.5337082713078726,
+      "learning_rate": 7.6938412925756e-06,
+      "loss": 0.5898,
+      "step": 2590
+    },
+    {
+      "epoch": 1.2357142857142858,
+      "grad_norm": 0.5024058244542078,
+      "learning_rate": 7.653425995322852e-06,
+      "loss": 0.5959,
+      "step": 2595
+    },
+    {
+      "epoch": 1.2380952380952381,
+      "grad_norm": 0.5089373868097246,
+      "learning_rate": 7.613051219968624e-06,
+      "loss": 0.6007,
+      "step": 2600
+    },
+    {
+      "epoch": 1.2404761904761905,
+      "grad_norm": 0.5292691125658913,
+      "learning_rate": 7.5727176637262034e-06,
+      "loss": 0.5893,
+      "step": 2605
+    },
+    {
+      "epoch": 1.2428571428571429,
+      "grad_norm": 0.5105236474902384,
+      "learning_rate": 7.532426023097063e-06,
+      "loss": 0.5991,
+      "step": 2610
+    },
+    {
+      "epoch": 1.2452380952380953,
+      "grad_norm": 0.5289609790386512,
+      "learning_rate": 7.492176993858873e-06,
+      "loss": 0.5833,
+      "step": 2615
+    },
+    {
+      "epoch": 1.2476190476190476,
+      "grad_norm": 0.5289162389327398,
+      "learning_rate": 7.451971271053455e-06,
+      "loss": 0.5852,
+      "step": 2620
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.570777487333234,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.5965,
+      "step": 2625
+    },
+    {
+      "epoch": 1.2523809523809524,
+      "grad_norm": 0.5208585348103933,
+      "learning_rate": 7.371692521157048e-06,
+      "loss": 0.6133,
+      "step": 2630
+    },
+    {
+      "epoch": 1.2547619047619047,
+      "grad_norm": 0.5279916869252734,
+      "learning_rate": 7.331620880362571e-06,
+      "loss": 0.5949,
+      "step": 2635
+    },
+    {
+      "epoch": 1.2571428571428571,
+      "grad_norm": 0.5269636791495603,
+      "learning_rate": 7.291595318569951e-06,
+      "loss": 0.6078,
+      "step": 2640
+    },
+    {
+      "epoch": 1.2595238095238095,
+      "grad_norm": 0.527352227349602,
+      "learning_rate": 7.2516165269620534e-06,
+      "loss": 0.5869,
+      "step": 2645
+    },
+    {
+      "epoch": 1.2619047619047619,
+      "grad_norm": 0.5144295584353445,
+      "learning_rate": 7.2116851959140965e-06,
+      "loss": 0.5918,
+      "step": 2650
+    },
+    {
+      "epoch": 1.2642857142857142,
+      "grad_norm": 0.5601926258663379,
+      "learning_rate": 7.171802014981726e-06,
+      "loss": 0.585,
+      "step": 2655
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 0.5379381373225919,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 0.595,
+      "step": 2660
+    },
+    {
+      "epoch": 1.269047619047619,
+      "grad_norm": 0.5251125215829732,
+      "learning_rate": 7.092182857516998e-06,
+      "loss": 0.5839,
+      "step": 2665
+    },
+    {
+      "epoch": 1.2714285714285714,
+      "grad_norm": 0.4965704448975545,
+      "learning_rate": 7.052448255890958e-06,
+      "loss": 0.5991,
+      "step": 2670
+    },
+    {
+      "epoch": 1.2738095238095237,
+      "grad_norm": 0.5450909049558729,
+      "learning_rate": 7.012764554169393e-06,
+      "loss": 0.6083,
+      "step": 2675
+    },
+    {
+      "epoch": 1.276190476190476,
+      "grad_norm": 0.5915810150427402,
+      "learning_rate": 6.973132437631743e-06,
+      "loss": 0.5899,
+      "step": 2680
+    },
+    {
+      "epoch": 1.2785714285714285,
+      "grad_norm": 0.5300752289151952,
+      "learning_rate": 6.933552590666659e-06,
+      "loss": 0.5876,
+      "step": 2685
+    },
+    {
+      "epoch": 1.2809523809523808,
+      "grad_norm": 0.5026232106359814,
+      "learning_rate": 6.8940256967601625e-06,
+      "loss": 0.5853,
+      "step": 2690
+    },
+    {
+      "epoch": 1.2833333333333332,
+      "grad_norm": 0.5197308254582468,
+      "learning_rate": 6.854552438483866e-06,
+      "loss": 0.5798,
+      "step": 2695
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 0.5211709556282136,
+      "learning_rate": 6.815133497483157e-06,
+      "loss": 0.5708,
+      "step": 2700
+    },
+    {
+      "epoch": 1.288095238095238,
+      "grad_norm": 0.4782326251158853,
+      "learning_rate": 6.775769554465455e-06,
+      "loss": 0.6023,
+      "step": 2705
+    },
+    {
+      "epoch": 1.2904761904761906,
+      "grad_norm": 0.5151978950364201,
+      "learning_rate": 6.736461289188445e-06,
+      "loss": 0.6054,
+      "step": 2710
+    },
+    {
+      "epoch": 1.292857142857143,
+      "grad_norm": 0.544144875536584,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.5983,
+      "step": 2715
+    },
+    {
+      "epoch": 1.2952380952380953,
+      "grad_norm": 0.525706847543693,
+      "learning_rate": 6.6580145060681255e-06,
+      "loss": 0.5909,
+      "step": 2720
+    },
+    {
+      "epoch": 1.2976190476190477,
+      "grad_norm": 0.5337499789796499,
+      "learning_rate": 6.618877342885945e-06,
+      "loss": 0.5932,
+      "step": 2725
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4979999689542144,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.5872,
+      "step": 2730
+    },
+    {
+      "epoch": 1.3023809523809524,
+      "grad_norm": 0.5293968795002583,
+      "learning_rate": 6.540778852473497e-06,
+      "loss": 0.595,
+      "step": 2735
+    },
+    {
+      "epoch": 1.3047619047619048,
+      "grad_norm": 0.513270923657658,
+      "learning_rate": 6.501818873889856e-06,
+      "loss": 0.5951,
+      "step": 2740
+    },
+    {
+      "epoch": 1.3071428571428572,
+      "grad_norm": 0.5284113230022897,
+      "learning_rate": 6.462919303774186e-06,
+      "loss": 0.5886,
+      "step": 2745
+    },
+    {
+      "epoch": 1.3095238095238095,
+      "grad_norm": 0.511386553250273,
+      "learning_rate": 6.424080813865139e-06,
+      "loss": 0.5785,
+      "step": 2750
+    },
+    {
+      "epoch": 1.311904761904762,
+      "grad_norm": 0.5612313704527047,
+      "learning_rate": 6.3853040748465855e-06,
+      "loss": 0.5934,
+      "step": 2755
+    },
+    {
+      "epoch": 1.3142857142857143,
+      "grad_norm": 0.49635536234133354,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 0.5804,
+      "step": 2760
+    },
+    {
+      "epoch": 1.3166666666666667,
+      "grad_norm": 0.5158067784575548,
+      "learning_rate": 6.3079385268731575e-06,
+      "loss": 0.5982,
+      "step": 2765
+    },
+    {
+      "epoch": 1.319047619047619,
+      "grad_norm": 0.5424701107493409,
+      "learning_rate": 6.269351053908061e-06,
+      "loss": 0.5873,
+      "step": 2770
+    },
+    {
+      "epoch": 1.3214285714285714,
+      "grad_norm": 0.5095803553683687,
+      "learning_rate": 6.230828003789949e-06,
+      "loss": 0.59,
+      "step": 2775
+    },
+    {
+      "epoch": 1.3238095238095238,
+      "grad_norm": 0.5004483050726994,
+      "learning_rate": 6.192370041755505e-06,
+      "loss": 0.588,
+      "step": 2780
+    },
+    {
+      "epoch": 1.3261904761904761,
+      "grad_norm": 0.5307962744979307,
+      "learning_rate": 6.153977831917451e-06,
+      "loss": 0.5879,
+      "step": 2785
+    },
+    {
+      "epoch": 1.3285714285714285,
+      "grad_norm": 0.5520488950744148,
+      "learning_rate": 6.115652037253054e-06,
+      "loss": 0.5967,
+      "step": 2790
+    },
+    {
+      "epoch": 1.3309523809523809,
+      "grad_norm": 0.4933954521187605,
+      "learning_rate": 6.077393319592697e-06,
+      "loss": 0.5922,
+      "step": 2795
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.5376192804239446,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 0.577,
+      "step": 2800
+    },
+    {
+      "epoch": 1.3357142857142856,
+      "grad_norm": 0.5213896422824574,
+      "learning_rate": 6.001079756802592e-06,
+      "loss": 0.58,
+      "step": 2805
+    },
+    {
+      "epoch": 1.3380952380952382,
+      "grad_norm": 0.5605679212505607,
+      "learning_rate": 5.963026229496378e-06,
+      "loss": 0.5801,
+      "step": 2810
+    },
+    {
+      "epoch": 1.3404761904761906,
+      "grad_norm": 0.5500894406128793,
+      "learning_rate": 5.925042414818514e-06,
+      "loss": 0.5893,
+      "step": 2815
+    },
+    {
+      "epoch": 1.342857142857143,
+      "grad_norm": 0.5051376225418218,
+      "learning_rate": 5.887128968693887e-06,
+      "loss": 0.5804,
+      "step": 2820
+    },
+    {
+      "epoch": 1.3452380952380953,
+      "grad_norm": 0.5241811747282762,
+      "learning_rate": 5.849286545832211e-06,
+      "loss": 0.576,
+      "step": 2825
+    },
+    {
+      "epoch": 1.3476190476190477,
+      "grad_norm": 0.5163613845049234,
+      "learning_rate": 5.811515799716754e-06,
+      "loss": 0.585,
+      "step": 2830
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5236385208944061,
+      "learning_rate": 5.773817382593008e-06,
+      "loss": 0.6006,
+      "step": 2835
+    },
+    {
+      "epoch": 1.3523809523809525,
+      "grad_norm": 0.532779005684654,
+      "learning_rate": 5.736191945457463e-06,
+      "loss": 0.5946,
+      "step": 2840
+    },
+    {
+      "epoch": 1.3547619047619048,
+      "grad_norm": 0.5259472612299654,
+      "learning_rate": 5.698640138046349e-06,
+      "loss": 0.5974,
+      "step": 2845
+    },
+    {
+      "epoch": 1.3571428571428572,
+      "grad_norm": 0.5109032533795757,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.5936,
+      "step": 2850
+    },
+    {
+      "epoch": 1.3595238095238096,
+      "grad_norm": 0.5122818131471542,
+      "learning_rate": 5.623760004973749e-06,
+      "loss": 0.5795,
+      "step": 2855
+    },
+    {
+      "epoch": 1.361904761904762,
+      "grad_norm": 0.4768696514947288,
+      "learning_rate": 5.586432972382561e-06,
+      "loss": 0.5758,
+      "step": 2860
+    },
+    {
+      "epoch": 1.3642857142857143,
+      "grad_norm": 0.53080481784541,
+      "learning_rate": 5.549182155634076e-06,
+      "loss": 0.5767,
+      "step": 2865
+    },
+    {
+      "epoch": 1.3666666666666667,
+      "grad_norm": 0.5038341903214718,
+      "learning_rate": 5.512008197995379e-06,
+      "loss": 0.5933,
+      "step": 2870
+    },
+    {
+      "epoch": 1.369047619047619,
+      "grad_norm": 0.5229825704537906,
+      "learning_rate": 5.47491174140631e-06,
+      "loss": 0.5824,
+      "step": 2875
+    },
+    {
+      "epoch": 1.3714285714285714,
+      "grad_norm": 0.49627233889726186,
+      "learning_rate": 5.43789342646837e-06,
+      "loss": 0.5988,
+      "step": 2880
+    },
+    {
+      "epoch": 1.3738095238095238,
+      "grad_norm": 0.514904186125751,
+      "learning_rate": 5.4009538924336864e-06,
+      "loss": 0.5876,
+      "step": 2885
+    },
+    {
+      "epoch": 1.3761904761904762,
+      "grad_norm": 0.5509458010595618,
+      "learning_rate": 5.364093777193944e-06,
+      "loss": 0.6045,
+      "step": 2890
+    },
+    {
+      "epoch": 1.3785714285714286,
+      "grad_norm": 0.5039270525726298,
+      "learning_rate": 5.32731371726938e-06,
+      "loss": 0.585,
+      "step": 2895
+    },
+    {
+      "epoch": 1.380952380952381,
+      "grad_norm": 0.5437109056015712,
+      "learning_rate": 5.290614347797802e-06,
+      "loss": 0.5924,
+      "step": 2900
+    },
+    {
+      "epoch": 1.3833333333333333,
+      "grad_norm": 0.5280438662821184,
+      "learning_rate": 5.253996302523596e-06,
+      "loss": 0.5808,
+      "step": 2905
+    },
+    {
+      "epoch": 1.3857142857142857,
+      "grad_norm": 0.5411985643744547,
+      "learning_rate": 5.217460213786822e-06,
+      "loss": 0.5909,
+      "step": 2910
+    },
+    {
+      "epoch": 1.388095238095238,
+      "grad_norm": 0.4928419336681969,
+      "learning_rate": 5.181006712512245e-06,
+      "loss": 0.5969,
+      "step": 2915
+    },
+    {
+      "epoch": 1.3904761904761904,
+      "grad_norm": 0.49801415174834524,
+      "learning_rate": 5.144636428198477e-06,
+      "loss": 0.5869,
+      "step": 2920
+    },
+    {
+      "epoch": 1.3928571428571428,
+      "grad_norm": 0.5060212093373364,
+      "learning_rate": 5.108349988907111e-06,
+      "loss": 0.5827,
+      "step": 2925
+    },
+    {
+      "epoch": 1.3952380952380952,
+      "grad_norm": 0.4959605026950924,
+      "learning_rate": 5.072148021251822e-06,
+      "loss": 0.5904,
+      "step": 2930
+    },
+    {
+      "epoch": 1.3976190476190475,
+      "grad_norm": 0.5256166273733636,
+      "learning_rate": 5.036031150387624e-06,
+      "loss": 0.582,
+      "step": 2935
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5644237371840894,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.5876,
+      "step": 2940
+    },
+    {
+      "epoch": 1.4023809523809523,
+      "grad_norm": 0.5544605233225762,
+      "learning_rate": 4.964055192294187e-06,
+      "loss": 0.5736,
+      "step": 2945
+    },
+    {
+      "epoch": 1.4047619047619047,
+      "grad_norm": 0.5111614229837703,
+      "learning_rate": 4.92819734798441e-06,
+      "loss": 0.5872,
+      "step": 2950
+    },
+    {
+      "epoch": 1.407142857142857,
+      "grad_norm": 0.4853589383206433,
+      "learning_rate": 4.892427086283147e-06,
+      "loss": 0.5899,
+      "step": 2955
+    },
+    {
+      "epoch": 1.4095238095238094,
+      "grad_norm": 0.5024366384928733,
+      "learning_rate": 4.856745024890466e-06,
+      "loss": 0.5739,
+      "step": 2960
+    },
+    {
+      "epoch": 1.4119047619047618,
+      "grad_norm": 0.5165017541670609,
+      "learning_rate": 4.821151779983343e-06,
+      "loss": 0.5889,
+      "step": 2965
+    },
+    {
+      "epoch": 1.4142857142857144,
+      "grad_norm": 0.5326057098439853,
+      "learning_rate": 4.78564796620502e-06,
+      "loss": 0.5984,
+      "step": 2970
+    },
+    {
+      "epoch": 1.4166666666666667,
+      "grad_norm": 0.5578793439842268,
+      "learning_rate": 4.7502341966544e-06,
+      "loss": 0.5919,
+      "step": 2975
+    },
+    {
+      "epoch": 1.4190476190476191,
+      "grad_norm": 0.5266644847974143,
+      "learning_rate": 4.714911082875446e-06,
+      "loss": 0.5811,
+      "step": 2980
+    },
+    {
+      "epoch": 1.4214285714285715,
+      "grad_norm": 0.5405356548584016,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.5852,
+      "step": 2985
+    },
+    {
+      "epoch": 1.4238095238095239,
+      "grad_norm": 0.5332823348361332,
+      "learning_rate": 4.644539260970417e-06,
+      "loss": 0.5904,
+      "step": 2990
+    },
+    {
+      "epoch": 1.4261904761904762,
+      "grad_norm": 0.5289204371473316,
+      "learning_rate": 4.609491768062705e-06,
+      "loss": 0.5715,
+      "step": 2995
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.49718290349750854,
+      "learning_rate": 4.5745373613424075e-06,
+      "loss": 0.5875,
+      "step": 3000
+    },
+    {
+      "epoch": 1.430952380952381,
+      "grad_norm": 0.5370577064296572,
+      "learning_rate": 4.539676644420966e-06,
+      "loss": 0.5838,
+      "step": 3005
+    },
+    {
+      "epoch": 1.4333333333333333,
+      "grad_norm": 0.5318499526470056,
+      "learning_rate": 4.504910219291941e-06,
+      "loss": 0.6081,
+      "step": 3010
+    },
+    {
+      "epoch": 1.4357142857142857,
+      "grad_norm": 0.5051255352622448,
+      "learning_rate": 4.470238686320606e-06,
+      "loss": 0.5796,
+      "step": 3015
+    },
+    {
+      "epoch": 1.438095238095238,
+      "grad_norm": 0.5318507665411399,
+      "learning_rate": 4.435662644233594e-06,
+      "loss": 0.5763,
+      "step": 3020
+    },
+    {
+      "epoch": 1.4404761904761905,
+      "grad_norm": 0.5337465779813616,
+      "learning_rate": 4.4011826901085346e-06,
+      "loss": 0.5856,
+      "step": 3025
+    },
+    {
+      "epoch": 1.4428571428571428,
+      "grad_norm": 0.5165852420124566,
+      "learning_rate": 4.3667994193637794e-06,
+      "loss": 0.5879,
+      "step": 3030
+    },
+    {
+      "epoch": 1.4452380952380952,
+      "grad_norm": 0.5155869722101444,
+      "learning_rate": 4.3325134257480905e-06,
+      "loss": 0.5831,
+      "step": 3035
+    },
+    {
+      "epoch": 1.4476190476190476,
+      "grad_norm": 0.4775386551110262,
+      "learning_rate": 4.298325301330383e-06,
+      "loss": 0.5866,
+      "step": 3040
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5054313057714112,
+      "learning_rate": 4.264235636489542e-06,
+      "loss": 0.5892,
+      "step": 3045
+    },
+    {
+      "epoch": 1.4523809523809523,
+      "grad_norm": 0.527424288981721,
+      "learning_rate": 4.23024501990417e-06,
+      "loss": 0.581,
+      "step": 3050
+    },
+    {
+      "epoch": 1.4547619047619047,
+      "grad_norm": 0.5089540127795982,
+      "learning_rate": 4.196354038542476e-06,
+      "loss": 0.5892,
+      "step": 3055
+    },
+    {
+      "epoch": 1.457142857142857,
+      "grad_norm": 0.5207863838757609,
+      "learning_rate": 4.162563277652104e-06,
+      "loss": 0.5865,
+      "step": 3060
+    },
+    {
+      "epoch": 1.4595238095238094,
+      "grad_norm": 0.5075766650861325,
+      "learning_rate": 4.128873320750027e-06,
+      "loss": 0.5756,
+      "step": 3065
+    },
+    {
+      "epoch": 1.461904761904762,
+      "grad_norm": 0.5025228767994807,
+      "learning_rate": 4.095284749612504e-06,
+      "loss": 0.5872,
+      "step": 3070
+    },
+    {
+      "epoch": 1.4642857142857144,
+      "grad_norm": 0.49216468636258304,
+      "learning_rate": 4.061798144264986e-06,
+      "loss": 0.5724,
+      "step": 3075
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 0.5053765390136462,
+      "learning_rate": 4.028414082972141e-06,
+      "loss": 0.5725,
+      "step": 3080
+    },
+    {
+      "epoch": 1.4690476190476192,
+      "grad_norm": 0.5698025824976778,
+      "learning_rate": 3.995133142227843e-06,
+      "loss": 0.5817,
+      "step": 3085
+    },
+    {
+      "epoch": 1.4714285714285715,
+      "grad_norm": 0.510204796748791,
+      "learning_rate": 3.961955896745224e-06,
+      "loss": 0.5777,
+      "step": 3090
+    },
+    {
+      "epoch": 1.473809523809524,
+      "grad_norm": 0.5091694662046772,
+      "learning_rate": 3.928882919446767e-06,
+      "loss": 0.5857,
+      "step": 3095
+    },
+    {
+      "epoch": 1.4761904761904763,
+      "grad_norm": 0.5151359736079936,
+      "learning_rate": 3.89591478145437e-06,
+      "loss": 0.5801,
+      "step": 3100
+    },
+    {
+      "epoch": 1.4785714285714286,
+      "grad_norm": 0.5182734669196073,
+      "learning_rate": 3.8630520520795275e-06,
+      "loss": 0.5693,
+      "step": 3105
+    },
+    {
+      "epoch": 1.480952380952381,
+      "grad_norm": 0.5129439390544596,
+      "learning_rate": 3.830295298813475e-06,
+      "loss": 0.586,
+      "step": 3110
+    },
+    {
+      "epoch": 1.4833333333333334,
+      "grad_norm": 0.5576172854485092,
+      "learning_rate": 3.797645087317401e-06,
+      "loss": 0.5753,
+      "step": 3115
+    },
+    {
+      "epoch": 1.4857142857142858,
+      "grad_norm": 0.5067485290590222,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.5958,
+      "step": 3120
+    },
+    {
+      "epoch": 1.4880952380952381,
+      "grad_norm": 0.5319104622052202,
+      "learning_rate": 3.7326665430710798e-06,
+      "loss": 0.5868,
+      "step": 3125
+    },
+    {
+      "epoch": 1.4904761904761905,
+      "grad_norm": 0.49375049435832247,
+      "learning_rate": 3.7003393324051874e-06,
+      "loss": 0.5757,
+      "step": 3130
+    },
+    {
+      "epoch": 1.4928571428571429,
+      "grad_norm": 0.5074776283156045,
+      "learning_rate": 3.6681209076586035e-06,
+      "loss": 0.5845,
+      "step": 3135
+    },
+    {
+      "epoch": 1.4952380952380953,
+      "grad_norm": 0.5065655047968505,
+      "learning_rate": 3.636011825196365e-06,
+      "loss": 0.5836,
+      "step": 3140
+    },
+    {
+      "epoch": 1.4976190476190476,
+      "grad_norm": 0.517677275546858,
+      "learning_rate": 3.6040126394953334e-06,
+      "loss": 0.5794,
+      "step": 3145
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.4918393882322778,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.5894,
+      "step": 3150
+    },
+    {
+      "epoch": 1.5023809523809524,
+      "grad_norm": 0.5004981541944984,
+      "learning_rate": 3.540346166785994e-06,
+      "loss": 0.5731,
+      "step": 3155
+    },
+    {
+      "epoch": 1.5047619047619047,
+      "grad_norm": 0.5117157105186525,
+      "learning_rate": 3.5086799792044812e-06,
+      "loss": 0.5856,
+      "step": 3160
+    },
+    {
+      "epoch": 1.5071428571428571,
+      "grad_norm": 0.5239013800449769,
+      "learning_rate": 3.4771258872187917e-06,
+      "loss": 0.5784,
+      "step": 3165
+    },
+    {
+      "epoch": 1.5095238095238095,
+      "grad_norm": 0.499421090386738,
+      "learning_rate": 3.4456844357218977e-06,
+      "loss": 0.5853,
+      "step": 3170
+    },
+    {
+      "epoch": 1.5119047619047619,
+      "grad_norm": 0.49970944259407496,
+      "learning_rate": 3.414356167661658e-06,
+      "loss": 0.57,
+      "step": 3175
+    },
+    {
+      "epoch": 1.5142857142857142,
+      "grad_norm": 0.5258093843059876,
+      "learning_rate": 3.3831416240314085e-06,
+      "loss": 0.5761,
+      "step": 3180
+    },
+    {
+      "epoch": 1.5166666666666666,
+      "grad_norm": 0.5244353245311157,
+      "learning_rate": 3.3520413438606215e-06,
+      "loss": 0.5882,
+      "step": 3185
+    },
+    {
+      "epoch": 1.519047619047619,
+      "grad_norm": 0.4977521350042896,
+      "learning_rate": 3.3210558642056277e-06,
+      "loss": 0.5901,
+      "step": 3190
+    },
+    {
+      "epoch": 1.5214285714285714,
+      "grad_norm": 0.5122476763468953,
+      "learning_rate": 3.290185720140301e-06,
+      "loss": 0.5811,
+      "step": 3195
+    },
+    {
+      "epoch": 1.5238095238095237,
+      "grad_norm": 0.49340352912131785,
+      "learning_rate": 3.2594314447468457e-06,
+      "loss": 0.5754,
+      "step": 3200
+    },
+    {
+      "epoch": 1.526190476190476,
+      "grad_norm": 0.4874919671111681,
+      "learning_rate": 3.228793569106594e-06,
+      "loss": 0.5841,
+      "step": 3205
+    },
+    {
+      "epoch": 1.5285714285714285,
+      "grad_norm": 0.5324961393719349,
+      "learning_rate": 3.1982726222908046e-06,
+      "loss": 0.5727,
+      "step": 3210
+    },
+    {
+      "epoch": 1.5309523809523808,
+      "grad_norm": 0.4995381050758276,
+      "learning_rate": 3.1678691313515688e-06,
+      "loss": 0.579,
+      "step": 3215
+    },
+    {
+      "epoch": 1.5333333333333332,
+      "grad_norm": 0.49736016607701533,
+      "learning_rate": 3.1375836213126653e-06,
+      "loss": 0.5847,
+      "step": 3220
+    },
+    {
+      "epoch": 1.5357142857142856,
+      "grad_norm": 0.549492541961606,
+      "learning_rate": 3.10741661516053e-06,
+      "loss": 0.5814,
+      "step": 3225
+    },
+    {
+      "epoch": 1.538095238095238,
+      "grad_norm": 0.49389981419744866,
+      "learning_rate": 3.077368633835205e-06,
+      "loss": 0.5737,
+      "step": 3230
+    },
+    {
+      "epoch": 1.5404761904761903,
+      "grad_norm": 0.5564195244743101,
+      "learning_rate": 3.0474401962213483e-06,
+      "loss": 0.5769,
+      "step": 3235
+    },
+    {
+      "epoch": 1.5428571428571427,
+      "grad_norm": 0.5123987478065914,
+      "learning_rate": 3.017631819139273e-06,
+      "loss": 0.591,
+      "step": 3240
+    },
+    {
+      "epoch": 1.545238095238095,
+      "grad_norm": 0.5012262974081656,
+      "learning_rate": 2.987944017336023e-06,
+      "loss": 0.5888,
+      "step": 3245
+    },
+    {
+      "epoch": 1.5476190476190477,
+      "grad_norm": 0.5152768019089213,
+      "learning_rate": 2.958377303476483e-06,
+      "loss": 0.5702,
+      "step": 3250
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5066389868980488,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.5868,
+      "step": 3255
+    },
+    {
+      "epoch": 1.5523809523809524,
+      "grad_norm": 0.5058336453434157,
+      "learning_rate": 2.8996091797841976e-06,
+      "loss": 0.5702,
+      "step": 3260
+    },
+    {
+      "epoch": 1.5547619047619048,
+      "grad_norm": 0.4984793009615145,
+      "learning_rate": 2.8704087847909333e-06,
+      "loss": 0.5662,
+      "step": 3265
+    },
+    {
+      "epoch": 1.5571428571428572,
+      "grad_norm": 0.5108314631589933,
+      "learning_rate": 2.8413315074028157e-06,
+      "loss": 0.5707,
+      "step": 3270
+    },
+    {
+      "epoch": 1.5595238095238095,
+      "grad_norm": 0.5031743279940161,
+      "learning_rate": 2.8123778497418687e-06,
+      "loss": 0.5872,
+      "step": 3275
+    },
+    {
+      "epoch": 1.561904761904762,
+      "grad_norm": 0.5140205415498605,
+      "learning_rate": 2.783548311795379e-06,
+      "loss": 0.5814,
+      "step": 3280
+    },
+    {
+      "epoch": 1.5642857142857143,
+      "grad_norm": 0.4951045700618797,
+      "learning_rate": 2.7548433914072736e-06,
+      "loss": 0.5688,
+      "step": 3285
+    },
+    {
+      "epoch": 1.5666666666666667,
+      "grad_norm": 0.4809510784828599,
+      "learning_rate": 2.726263584269513e-06,
+      "loss": 0.5822,
+      "step": 3290
+    },
+    {
+      "epoch": 1.569047619047619,
+      "grad_norm": 0.5144657250346224,
+      "learning_rate": 2.6978093839135365e-06,
+      "loss": 0.5924,
+      "step": 3295
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 0.5069069605136209,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 0.5808,
+      "step": 3300
+    },
+    {
+      "epoch": 1.5738095238095238,
+      "grad_norm": 0.5080243059955405,
+      "learning_rate": 2.641279766818977e-06,
+      "loss": 0.5826,
+      "step": 3305
+    },
+    {
+      "epoch": 1.5761904761904761,
+      "grad_norm": 0.5233910641374417,
+      "learning_rate": 2.6132053262641467e-06,
+      "loss": 0.5862,
+      "step": 3310
+    },
+    {
+      "epoch": 1.5785714285714287,
+      "grad_norm": 0.5114261441752712,
+      "learning_rate": 2.5852584448417327e-06,
+      "loss": 0.582,
+      "step": 3315
+    },
+    {
+      "epoch": 1.580952380952381,
+      "grad_norm": 0.5249361251608791,
+      "learning_rate": 2.5574396051534835e-06,
+      "loss": 0.5812,
+      "step": 3320
+    },
+    {
+      "epoch": 1.5833333333333335,
+      "grad_norm": 0.5014730717082388,
+      "learning_rate": 2.529749287590042e-06,
+      "loss": 0.5836,
+      "step": 3325
+    },
+    {
+      "epoch": 1.5857142857142859,
+      "grad_norm": 0.5031355239363027,
+      "learning_rate": 2.502187970322657e-06,
+      "loss": 0.5713,
+      "step": 3330
+    },
+    {
+      "epoch": 1.5880952380952382,
+      "grad_norm": 0.5267100809447453,
+      "learning_rate": 2.4747561292949496e-06,
+      "loss": 0.5871,
+      "step": 3335
+    },
+    {
+      "epoch": 1.5904761904761906,
+      "grad_norm": 0.5066033968112823,
+      "learning_rate": 2.447454238214654e-06,
+      "loss": 0.5777,
+      "step": 3340
+    },
+    {
+      "epoch": 1.592857142857143,
+      "grad_norm": 0.5249173659213104,
+      "learning_rate": 2.420282768545469e-06,
+      "loss": 0.5932,
+      "step": 3345
+    },
+    {
+      "epoch": 1.5952380952380953,
+      "grad_norm": 0.48114876503065274,
+      "learning_rate": 2.3932421894989167e-06,
+      "loss": 0.5733,
+      "step": 3350
+    },
+    {
+      "epoch": 1.5976190476190477,
+      "grad_norm": 0.48457566365078014,
+      "learning_rate": 2.366332968026207e-06,
+      "loss": 0.581,
+      "step": 3355
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.49253983378100513,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.569,
+      "step": 3360
+    },
+    {
+      "epoch": 1.6023809523809525,
+      "grad_norm": 0.521391190119619,
+      "learning_rate": 2.3129104542574433e-06,
+      "loss": 0.575,
+      "step": 3365
+    },
+    {
+      "epoch": 1.6047619047619048,
+      "grad_norm": 0.4986688245603456,
+      "learning_rate": 2.2863980844900036e-06,
+      "loss": 0.5853,
+      "step": 3370
+    },
+    {
+      "epoch": 1.6071428571428572,
+      "grad_norm": 0.5270951766715323,
+      "learning_rate": 2.2600189173377263e-06,
+      "loss": 0.5898,
+      "step": 3375
+    },
+    {
+      "epoch": 1.6095238095238096,
+      "grad_norm": 0.46626058041434404,
+      "learning_rate": 2.2337734083302164e-06,
+      "loss": 0.584,
+      "step": 3380
+    },
+    {
+      "epoch": 1.611904761904762,
+      "grad_norm": 0.5313981173035822,
+      "learning_rate": 2.207662010689002e-06,
+      "loss": 0.5822,
+      "step": 3385
+    },
+    {
+      "epoch": 1.6142857142857143,
+      "grad_norm": 0.5153471281514648,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.5797,
+      "step": 3390
+    },
+    {
+      "epoch": 1.6166666666666667,
+      "grad_norm": 0.5000763124278017,
+      "learning_rate": 2.155843350804243e-06,
+      "loss": 0.567,
+      "step": 3395
+    },
+    {
+      "epoch": 1.619047619047619,
+      "grad_norm": 0.49402952120811466,
+      "learning_rate": 2.130136983393112e-06,
+      "loss": 0.5697,
+      "step": 3400
+    },
+    {
+      "epoch": 1.6214285714285714,
+      "grad_norm": 0.48898166681541133,
+      "learning_rate": 2.104566516997647e-06,
+      "loss": 0.5779,
+      "step": 3405
+    },
+    {
+      "epoch": 1.6238095238095238,
+      "grad_norm": 0.5391774389924573,
+      "learning_rate": 2.0791323931823783e-06,
+      "loss": 0.5914,
+      "step": 3410
+    },
+    {
+      "epoch": 1.6261904761904762,
+      "grad_norm": 0.5080483298837424,
+      "learning_rate": 2.053835051157397e-06,
+      "loss": 0.5817,
+      "step": 3415
+    },
+    {
+      "epoch": 1.6285714285714286,
+      "grad_norm": 0.49302417797140397,
+      "learning_rate": 2.0286749277707783e-06,
+      "loss": 0.5647,
+      "step": 3420
+    },
+    {
+      "epoch": 1.630952380952381,
+      "grad_norm": 0.5241769643468105,
+      "learning_rate": 2.0036524575010176e-06,
+      "loss": 0.5853,
+      "step": 3425
+    },
+    {
+      "epoch": 1.6333333333333333,
+      "grad_norm": 0.5265692131855949,
+      "learning_rate": 1.9787680724495617e-06,
+      "loss": 0.5817,
+      "step": 3430
+    },
+    {
+      "epoch": 1.6357142857142857,
+      "grad_norm": 0.4902606126508354,
+      "learning_rate": 1.9540222023333165e-06,
+      "loss": 0.5807,
+      "step": 3435
+    },
+    {
+      "epoch": 1.638095238095238,
+      "grad_norm": 0.5237480696892753,
+      "learning_rate": 1.929415274477239e-06,
+      "loss": 0.5701,
+      "step": 3440
+    },
+    {
+      "epoch": 1.6404761904761904,
+      "grad_norm": 0.5357308687997984,
+      "learning_rate": 1.9049477138069606e-06,
+      "loss": 0.5804,
+      "step": 3445
+    },
+    {
+      "epoch": 1.6428571428571428,
+      "grad_norm": 0.5263899257687392,
+      "learning_rate": 1.880619942841435e-06,
+      "loss": 0.5883,
+      "step": 3450
+    },
+    {
+      "epoch": 1.6452380952380952,
+      "grad_norm": 0.5115574850941745,
+      "learning_rate": 1.856432381685669e-06,
+      "loss": 0.59,
+      "step": 3455
+    },
+    {
+      "epoch": 1.6476190476190475,
+      "grad_norm": 0.5173104409710929,
+      "learning_rate": 1.8323854480234348e-06,
+      "loss": 0.5805,
+      "step": 3460
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5011073792449935,
+      "learning_rate": 1.808479557110081e-06,
+      "loss": 0.5738,
+      "step": 3465
+    },
+    {
+      "epoch": 1.6523809523809523,
+      "grad_norm": 0.4933932876019805,
+      "learning_rate": 1.7847151217653624e-06,
+      "loss": 0.5722,
+      "step": 3470
+    },
+    {
+      "epoch": 1.6547619047619047,
+      "grad_norm": 0.5181601753202433,
+      "learning_rate": 1.7610925523662836e-06,
+      "loss": 0.5673,
+      "step": 3475
+    },
+    {
+      "epoch": 1.657142857142857,
+      "grad_norm": 0.5054495843864125,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 0.5771,
+      "step": 3480
+    },
+    {
+      "epoch": 1.6595238095238094,
+      "grad_norm": 0.5277775744809468,
+      "learning_rate": 1.714274640657001e-06,
+      "loss": 0.5747,
+      "step": 3485
+    },
+    {
+      "epoch": 1.6619047619047618,
+      "grad_norm": 0.48735398585218814,
+      "learning_rate": 1.6910801068236015e-06,
+      "loss": 0.5847,
+      "step": 3490
+    },
+    {
+      "epoch": 1.6642857142857141,
+      "grad_norm": 0.5157125695595391,
+      "learning_rate": 1.6680290558755119e-06,
+      "loss": 0.5828,
+      "step": 3495
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.5089727599844109,
+      "learning_rate": 1.6451218858706374e-06,
+      "loss": 0.591,
+      "step": 3500
+    },
+    {
+      "epoch": 1.669047619047619,
+      "grad_norm": 0.5026463229681872,
+      "learning_rate": 1.6223589923822768e-06,
+      "loss": 0.5817,
+      "step": 3505
+    },
+    {
+      "epoch": 1.6714285714285713,
+      "grad_norm": 0.4931313873291854,
+      "learning_rate": 1.599740768492286e-06,
+      "loss": 0.5713,
+      "step": 3510
+    },
+    {
+      "epoch": 1.6738095238095239,
+      "grad_norm": 0.5181473464694085,
+      "learning_rate": 1.5772676047842862e-06,
+      "loss": 0.5759,
+      "step": 3515
+    },
+    {
+      "epoch": 1.6761904761904762,
+      "grad_norm": 0.4927232886574381,
+      "learning_rate": 1.5549398893369216e-06,
+      "loss": 0.5817,
+      "step": 3520
+    },
+    {
+      "epoch": 1.6785714285714286,
+      "grad_norm": 0.5220587931111694,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.5782,
+      "step": 3525
+    },
+    {
+      "epoch": 1.680952380952381,
+      "grad_norm": 0.5333713690182812,
+      "learning_rate": 1.5107223429736273e-06,
+      "loss": 0.5883,
+      "step": 3530
+    },
+    {
+      "epoch": 1.6833333333333333,
+      "grad_norm": 0.5150308121762561,
+      "learning_rate": 1.4888332756300027e-06,
+      "loss": 0.586,
+      "step": 3535
+    },
+    {
+      "epoch": 1.6857142857142857,
+      "grad_norm": 0.4978412003341323,
+      "learning_rate": 1.467091183678444e-06,
+      "loss": 0.5872,
+      "step": 3540
+    },
+    {
+      "epoch": 1.688095238095238,
+      "grad_norm": 0.4880778638366708,
+      "learning_rate": 1.4454964425730533e-06,
+      "loss": 0.5708,
+      "step": 3545
+    },
+    {
+      "epoch": 1.6904761904761905,
+      "grad_norm": 0.5516107606480507,
+      "learning_rate": 1.424049425223405e-06,
+      "loss": 0.567,
+      "step": 3550
+    },
+    {
+      "epoch": 1.6928571428571428,
+      "grad_norm": 0.5693622851862525,
+      "learning_rate": 1.4027505019880972e-06,
+      "loss": 0.5776,
+      "step": 3555
+    },
+    {
+      "epoch": 1.6952380952380952,
+      "grad_norm": 0.5070794770825253,
+      "learning_rate": 1.3816000406683604e-06,
+      "loss": 0.5704,
+      "step": 3560
+    },
+    {
+      "epoch": 1.6976190476190476,
+      "grad_norm": 0.49091161185672627,
+      "learning_rate": 1.3605984065017074e-06,
+      "loss": 0.5645,
+      "step": 3565
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.519545490092357,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.5866,
+      "step": 3570
+    },
+    {
+      "epoch": 1.7023809523809523,
+      "grad_norm": 0.5212093423730666,
+      "learning_rate": 1.3190430677212795e-06,
+      "loss": 0.5864,
+      "step": 3575
+    },
+    {
+      "epoch": 1.704761904761905,
+      "grad_norm": 0.531691412658331,
+      "learning_rate": 1.2984900807073919e-06,
+      "loss": 0.5789,
+      "step": 3580
+    },
+    {
+      "epoch": 1.7071428571428573,
+      "grad_norm": 0.5456532742307001,
+      "learning_rate": 1.278087356033947e-06,
+      "loss": 0.5722,
+      "step": 3585
+    },
+    {
+      "epoch": 1.7095238095238097,
+      "grad_norm": 0.5205925476159533,
+      "learning_rate": 1.2578352460261456e-06,
+      "loss": 0.5778,
+      "step": 3590
+    },
+    {
+      "epoch": 1.711904761904762,
+      "grad_norm": 0.5062569568474926,
+      "learning_rate": 1.2377341004082778e-06,
+      "loss": 0.5741,
+      "step": 3595
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.5175248723743612,
+      "learning_rate": 1.2177842662977136e-06,
+      "loss": 0.5785,
+      "step": 3600
+    },
+    {
+      "epoch": 1.7166666666666668,
+      "grad_norm": 0.5159262102883804,
+      "learning_rate": 1.1979860881988903e-06,
+      "loss": 0.5754,
+      "step": 3605
+    },
+    {
+      "epoch": 1.7190476190476192,
+      "grad_norm": 0.5083274127105409,
+      "learning_rate": 1.1783399079973578e-06,
+      "loss": 0.5842,
+      "step": 3610
+    },
+    {
+      "epoch": 1.7214285714285715,
+      "grad_norm": 0.5120999260869891,
+      "learning_rate": 1.1588460649539036e-06,
+      "loss": 0.5734,
+      "step": 3615
+    },
+    {
+      "epoch": 1.723809523809524,
+      "grad_norm": 0.49029779483860925,
+      "learning_rate": 1.1395048956986577e-06,
+      "loss": 0.5899,
+      "step": 3620
+    },
+    {
+      "epoch": 1.7261904761904763,
+      "grad_norm": 0.5065004286184338,
+      "learning_rate": 1.1203167342253063e-06,
+      "loss": 0.5728,
+      "step": 3625
+    },
+    {
+      "epoch": 1.7285714285714286,
+      "grad_norm": 0.5162711801738049,
+      "learning_rate": 1.1012819118853147e-06,
+      "loss": 0.57,
+      "step": 3630
+    },
+    {
+      "epoch": 1.730952380952381,
+      "grad_norm": 0.512398683483683,
+      "learning_rate": 1.0824007573822025e-06,
+      "loss": 0.573,
+      "step": 3635
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "grad_norm": 0.5060773003944762,
+      "learning_rate": 1.0636735967658785e-06,
+      "loss": 0.5758,
+      "step": 3640
+    },
+    {
+      "epoch": 1.7357142857142858,
+      "grad_norm": 0.5719110243417397,
+      "learning_rate": 1.0451007534269908e-06,
+      "loss": 0.5706,
+      "step": 3645
+    },
+    {
+      "epoch": 1.7380952380952381,
+      "grad_norm": 0.5159140197103158,
+      "learning_rate": 1.026682548091361e-06,
+      "loss": 0.5787,
+      "step": 3650
+    },
+    {
+      "epoch": 1.7404761904761905,
+      "grad_norm": 0.5756689272489154,
+      "learning_rate": 1.0084192988144392e-06,
+      "loss": 0.589,
+      "step": 3655
+    },
+    {
+      "epoch": 1.7428571428571429,
+      "grad_norm": 0.4987087820127243,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.5713,
+      "step": 3660
+    },
+    {
+      "epoch": 1.7452380952380953,
+      "grad_norm": 0.49383352905457956,
+      "learning_rate": 9.723589272737443e-07,
+      "loss": 0.5692,
+      "step": 3665
+    },
+    {
+      "epoch": 1.7476190476190476,
+      "grad_norm": 0.5341671518005138,
+      "learning_rate": 9.545624277198085e-07,
+      "loss": 0.5653,
+      "step": 3670
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5264048094976569,
+      "learning_rate": 9.369221296335007e-07,
+      "loss": 0.5716,
+      "step": 3675
+    },
+    {
+      "epoch": 1.7523809523809524,
+      "grad_norm": 0.49117865624067225,
+      "learning_rate": 9.194383376369509e-07,
+      "loss": 0.5755,
+      "step": 3680
+    },
+    {
+      "epoch": 1.7547619047619047,
+      "grad_norm": 0.4802311797210391,
+      "learning_rate": 9.021113536496551e-07,
+      "loss": 0.5796,
+      "step": 3685
+    },
+    {
+      "epoch": 1.7571428571428571,
+      "grad_norm": 0.5170954646484791,
+      "learning_rate": 8.849414768832687e-07,
+      "loss": 0.5575,
+      "step": 3690
+    },
+    {
+      "epoch": 1.7595238095238095,
+      "grad_norm": 0.502792596764845,
+      "learning_rate": 8.679290038364319e-07,
+      "loss": 0.5833,
+      "step": 3695
+    },
+    {
+      "epoch": 1.7619047619047619,
+      "grad_norm": 0.5049868760014847,
+      "learning_rate": 8.510742282896545e-07,
+      "loss": 0.577,
+      "step": 3700
+    },
+    {
+      "epoch": 1.7642857142857142,
+      "grad_norm": 0.5037269478659716,
+      "learning_rate": 8.343774413002382e-07,
+      "loss": 0.5857,
+      "step": 3705
+    },
+    {
+      "epoch": 1.7666666666666666,
+      "grad_norm": 0.531756805530802,
+      "learning_rate": 8.178389311972612e-07,
+      "loss": 0.5706,
+      "step": 3710
+    },
+    {
+      "epoch": 1.769047619047619,
+      "grad_norm": 0.5354038647559776,
+      "learning_rate": 8.014589835765807e-07,
+      "loss": 0.5793,
+      "step": 3715
+    },
+    {
+      "epoch": 1.7714285714285714,
+      "grad_norm": 0.5165889882026443,
+      "learning_rate": 7.852378812959227e-07,
+      "loss": 0.565,
+      "step": 3720
+    },
+    {
+      "epoch": 1.7738095238095237,
+      "grad_norm": 0.5188950618246705,
+      "learning_rate": 7.69175904469982e-07,
+      "loss": 0.5681,
+      "step": 3725
+    },
+    {
+      "epoch": 1.776190476190476,
+      "grad_norm": 0.5219125713505919,
+      "learning_rate": 7.532733304655848e-07,
+      "loss": 0.5823,
+      "step": 3730
+    },
+    {
+      "epoch": 1.7785714285714285,
+      "grad_norm": 0.4842042459830578,
+      "learning_rate": 7.375304338969135e-07,
+      "loss": 0.5754,
+      "step": 3735
+    },
+    {
+      "epoch": 1.7809523809523808,
+      "grad_norm": 0.5028634955697662,
+      "learning_rate": 7.219474866207465e-07,
+      "loss": 0.5655,
+      "step": 3740
+    },
+    {
+      "epoch": 1.7833333333333332,
+      "grad_norm": 0.48467803772894236,
+      "learning_rate": 7.065247577317747e-07,
+      "loss": 0.5725,
+      "step": 3745
+    },
+    {
+      "epoch": 1.7857142857142856,
+      "grad_norm": 0.4918168156661026,
+      "learning_rate": 6.912625135579587e-07,
+      "loss": 0.5722,
+      "step": 3750
+    },
+    {
+      "epoch": 1.788095238095238,
+      "grad_norm": 0.48464787354503214,
+      "learning_rate": 6.761610176559086e-07,
+      "loss": 0.5589,
+      "step": 3755
+    },
+    {
+      "epoch": 1.7904761904761903,
+      "grad_norm": 0.5182220123762346,
+      "learning_rate": 6.612205308063646e-07,
+      "loss": 0.5684,
+      "step": 3760
+    },
+    {
+      "epoch": 1.7928571428571427,
+      "grad_norm": 0.48991260111946716,
+      "learning_rate": 6.464413110096601e-07,
+      "loss": 0.5778,
+      "step": 3765
+    },
+    {
+      "epoch": 1.795238095238095,
+      "grad_norm": 0.4667646370432281,
+      "learning_rate": 6.318236134812917e-07,
+      "loss": 0.5783,
+      "step": 3770
+    },
+    {
+      "epoch": 1.7976190476190477,
+      "grad_norm": 0.5127849630517298,
+      "learning_rate": 6.173676906475012e-07,
+      "loss": 0.5793,
+      "step": 3775
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5479174273289018,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.5789,
+      "step": 3780
+    },
+    {
+      "epoch": 1.8023809523809524,
+      "grad_norm": 0.5027002463972741,
+      "learning_rate": 5.889421647962456e-07,
+      "loss": 0.5844,
+      "step": 3785
+    },
+    {
+      "epoch": 1.8047619047619048,
+      "grad_norm": 0.47085642019543783,
+      "learning_rate": 5.749730526460073e-07,
+      "loss": 0.5675,
+      "step": 3790
+    },
+    {
+      "epoch": 1.8071428571428572,
+      "grad_norm": 0.5072363123110816,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.5775,
+      "step": 3795
+    },
+    {
+      "epoch": 1.8095238095238095,
+      "grad_norm": 0.4957008376046064,
+      "learning_rate": 5.475233360227516e-07,
+      "loss": 0.5811,
+      "step": 3800
+    },
+    {
+      "epoch": 1.811904761904762,
+      "grad_norm": 0.5212409865496715,
+      "learning_rate": 5.340432055661637e-07,
+      "loss": 0.5761,
+      "step": 3805
+    },
+    {
+      "epoch": 1.8142857142857143,
+      "grad_norm": 0.470997569919902,
+      "learning_rate": 5.207265383286831e-07,
+      "loss": 0.5727,
+      "step": 3810
+    },
+    {
+      "epoch": 1.8166666666666667,
+      "grad_norm": 0.5180739672868397,
+      "learning_rate": 5.075735642696611e-07,
+      "loss": 0.5655,
+      "step": 3815
+    },
+    {
+      "epoch": 1.819047619047619,
+      "grad_norm": 0.4888611854375848,
+      "learning_rate": 4.945845105217118e-07,
+      "loss": 0.5642,
+      "step": 3820
+    },
+    {
+      "epoch": 1.8214285714285714,
+      "grad_norm": 0.5177332953584531,
+      "learning_rate": 4.817596013867765e-07,
+      "loss": 0.5845,
+      "step": 3825
+    },
+    {
+      "epoch": 1.8238095238095238,
+      "grad_norm": 0.5279903730811234,
+      "learning_rate": 4.6909905833226965e-07,
+      "loss": 0.5775,
+      "step": 3830
+    },
+    {
+      "epoch": 1.8261904761904761,
+      "grad_norm": 0.5151201728696789,
+      "learning_rate": 4.566030999872384e-07,
+      "loss": 0.5863,
+      "step": 3835
+    },
+    {
+      "epoch": 1.8285714285714287,
+      "grad_norm": 0.5216946630712812,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 0.5824,
+      "step": 3840
+    },
+    {
+      "epoch": 1.830952380952381,
+      "grad_norm": 0.48997634713435234,
+      "learning_rate": 4.3210579772738237e-07,
+      "loss": 0.5633,
+      "step": 3845
+    },
+    {
+      "epoch": 1.8333333333333335,
+      "grad_norm": 0.5266753228806161,
+      "learning_rate": 4.2010487684511105e-07,
+      "loss": 0.5861,
+      "step": 3850
+    },
+    {
+      "epoch": 1.8357142857142859,
+      "grad_norm": 0.4963390005103402,
+      "learning_rate": 4.082693867301224e-07,
+      "loss": 0.5866,
+      "step": 3855
+    },
+    {
+      "epoch": 1.8380952380952382,
+      "grad_norm": 0.5075963559582308,
+      "learning_rate": 3.965995317640026e-07,
+      "loss": 0.566,
+      "step": 3860
+    },
+    {
+      "epoch": 1.8404761904761906,
+      "grad_norm": 0.5210567644268166,
+      "learning_rate": 3.850955134680678e-07,
+      "loss": 0.576,
+      "step": 3865
+    },
+    {
+      "epoch": 1.842857142857143,
+      "grad_norm": 0.521115702642029,
+      "learning_rate": 3.7375753049987974e-07,
+      "loss": 0.5695,
+      "step": 3870
+    },
+    {
+      "epoch": 1.8452380952380953,
+      "grad_norm": 0.5060214448757876,
+      "learning_rate": 3.625857786498055e-07,
+      "loss": 0.5704,
+      "step": 3875
+    },
+    {
+      "epoch": 1.8476190476190477,
+      "grad_norm": 0.4875613223286847,
+      "learning_rate": 3.515804508376508e-07,
+      "loss": 0.5697,
+      "step": 3880
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.4832852448893152,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.5768,
+      "step": 3885
+    },
+    {
+      "epoch": 1.8523809523809525,
+      "grad_norm": 0.5331240685153525,
+      "learning_rate": 3.3006982463352764e-07,
+      "loss": 0.5737,
+      "step": 3890
+    },
+    {
+      "epoch": 1.8547619047619048,
+      "grad_norm": 0.5001592000344391,
+      "learning_rate": 3.1956489769859213e-07,
+      "loss": 0.572,
+      "step": 3895
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 0.48713909339424555,
+      "learning_rate": 3.0922713770922155e-07,
+      "loss": 0.5842,
+      "step": 3900
+    },
+    {
+      "epoch": 1.8595238095238096,
+      "grad_norm": 0.505832290948494,
+      "learning_rate": 2.9905672318339963e-07,
+      "loss": 0.5794,
+      "step": 3905
+    },
+    {
+      "epoch": 1.861904761904762,
+      "grad_norm": 0.5001487285158832,
+      "learning_rate": 2.8905382974930173e-07,
+      "loss": 0.5797,
+      "step": 3910
+    },
+    {
+      "epoch": 1.8642857142857143,
+      "grad_norm": 0.5134685274388742,
+      "learning_rate": 2.7921863014225504e-07,
+      "loss": 0.5938,
+      "step": 3915
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 0.48901171207964383,
+      "learning_rate": 2.6955129420176193e-07,
+      "loss": 0.5804,
+      "step": 3920
+    },
+    {
+      "epoch": 1.869047619047619,
+      "grad_norm": 0.4928766933172695,
+      "learning_rate": 2.6005198886856486e-07,
+      "loss": 0.5738,
+      "step": 3925
+    },
+    {
+      "epoch": 1.8714285714285714,
+      "grad_norm": 0.4929640596310142,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.5628,
+      "step": 3930
+    },
+    {
+      "epoch": 1.8738095238095238,
+      "grad_norm": 0.5172714992148169,
+      "learning_rate": 2.4155812327598337e-07,
+      "loss": 0.5649,
+      "step": 3935
+    },
+    {
+      "epoch": 1.8761904761904762,
+      "grad_norm": 0.5022968661364299,
+      "learning_rate": 2.3256388237858806e-07,
+      "loss": 0.5681,
+      "step": 3940
+    },
+    {
+      "epoch": 1.8785714285714286,
+      "grad_norm": 0.5232052126416001,
+      "learning_rate": 2.2373831080695463e-07,
+      "loss": 0.5745,
+      "step": 3945
+    },
+    {
+      "epoch": 1.880952380952381,
+      "grad_norm": 0.488579854673118,
+      "learning_rate": 2.1508156096578748e-07,
+      "loss": 0.5729,
+      "step": 3950
+    },
+    {
+      "epoch": 1.8833333333333333,
+      "grad_norm": 0.5155005458528391,
+      "learning_rate": 2.0659378234448524e-07,
+      "loss": 0.5732,
+      "step": 3955
+    },
+    {
+      "epoch": 1.8857142857142857,
+      "grad_norm": 0.5024985648134219,
+      "learning_rate": 1.9827512151456175e-07,
+      "loss": 0.5627,
+      "step": 3960
+    },
+    {
+      "epoch": 1.888095238095238,
+      "grad_norm": 0.5210350946419919,
+      "learning_rate": 1.9012572212711467e-07,
+      "loss": 0.5766,
+      "step": 3965
+    },
+    {
+      "epoch": 1.8904761904761904,
+      "grad_norm": 0.505351224359543,
+      "learning_rate": 1.82145724910342e-07,
+      "loss": 0.5787,
+      "step": 3970
+    },
+    {
+      "epoch": 1.8928571428571428,
+      "grad_norm": 0.5042030458059477,
+      "learning_rate": 1.7433526766711727e-07,
+      "loss": 0.5756,
+      "step": 3975
+    },
+    {
+      "epoch": 1.8952380952380952,
+      "grad_norm": 0.5196433352960352,
+      "learning_rate": 1.6669448527260602e-07,
+      "loss": 0.5787,
+      "step": 3980
+    },
+    {
+      "epoch": 1.8976190476190475,
+      "grad_norm": 0.5042961399463299,
+      "learning_rate": 1.5922350967193524e-07,
+      "loss": 0.5828,
+      "step": 3985
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.49509327596749514,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.5666,
+      "step": 3990
+    },
+    {
+      "epoch": 1.9023809523809523,
+      "grad_norm": 0.4993392687429765,
+      "learning_rate": 1.447914919688298e-07,
+      "loss": 0.5862,
+      "step": 3995
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 0.5302407636844851,
+      "learning_rate": 1.3783069908621772e-07,
+      "loss": 0.5711,
+      "step": 4000
+    },
+    {
+      "epoch": 1.907142857142857,
+      "grad_norm": 0.49474811360911164,
+      "learning_rate": 1.3104021143278911e-07,
+      "loss": 0.5876,
+      "step": 4005
+    },
+    {
+      "epoch": 1.9095238095238094,
+      "grad_norm": 0.497576228240754,
+      "learning_rate": 1.2442014627032318e-07,
+      "loss": 0.5744,
+      "step": 4010
+    },
+    {
+      "epoch": 1.9119047619047618,
+      "grad_norm": 0.4871944438237385,
+      "learning_rate": 1.1797061791766207e-07,
+      "loss": 0.5714,
+      "step": 4015
+    },
+    {
+      "epoch": 1.9142857142857141,
+      "grad_norm": 0.4830245999351171,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 0.5686,
+      "step": 4020
+    },
+    {
+      "epoch": 1.9166666666666665,
+      "grad_norm": 0.5413142563859302,
+      "learning_rate": 1.055836141905553e-07,
+      "loss": 0.5888,
+      "step": 4025
+    },
+    {
+      "epoch": 1.919047619047619,
+      "grad_norm": 0.46941478343612175,
+      "learning_rate": 9.964635272153633e-08,
+      "loss": 0.5786,
+      "step": 4030
+    },
+    {
+      "epoch": 1.9214285714285713,
+      "grad_norm": 0.5140729728882855,
+      "learning_rate": 9.388005586947191e-08,
+      "loss": 0.565,
+      "step": 4035
+    },
+    {
+      "epoch": 1.9238095238095239,
+      "grad_norm": 0.4991452996322733,
+      "learning_rate": 8.82848232098732e-08,
+      "loss": 0.5708,
+      "step": 4040
+    },
+    {
+      "epoch": 1.9261904761904762,
+      "grad_norm": 0.505925510646528,
+      "learning_rate": 8.286075136421435e-08,
+      "loss": 0.5814,
+      "step": 4045
+    },
+    {
+      "epoch": 1.9285714285714286,
+      "grad_norm": 0.5275663420688306,
+      "learning_rate": 7.760793399827937e-08,
+      "loss": 0.5635,
+      "step": 4050
+    },
+    {
+      "epoch": 1.930952380952381,
+      "grad_norm": 0.5049137853663558,
+      "learning_rate": 7.25264618205357e-08,
+      "loss": 0.5547,
+      "step": 4055
+    },
+    {
+      "epoch": 1.9333333333333333,
+      "grad_norm": 0.4878320213926508,
+      "learning_rate": 6.761642258056977e-08,
+      "loss": 0.5695,
+      "step": 4060
+    },
+    {
+      "epoch": 1.9357142857142857,
+      "grad_norm": 0.5081138801093147,
+      "learning_rate": 6.287790106757396e-08,
+      "loss": 0.5683,
+      "step": 4065
+    },
+    {
+      "epoch": 1.938095238095238,
+      "grad_norm": 0.4943589564870212,
+      "learning_rate": 5.831097910887873e-08,
+      "loss": 0.5658,
+      "step": 4070
+    },
+    {
+      "epoch": 1.9404761904761905,
+      "grad_norm": 0.5302217136380152,
+      "learning_rate": 5.391573556854157e-08,
+      "loss": 0.5678,
+      "step": 4075
+    },
+    {
+      "epoch": 1.9428571428571428,
+      "grad_norm": 0.5038390499507364,
+      "learning_rate": 4.9692246345985905e-08,
+      "loss": 0.5702,
+      "step": 4080
+    },
+    {
+      "epoch": 1.9452380952380952,
+      "grad_norm": 0.5151211498515915,
+      "learning_rate": 4.564058437468877e-08,
+      "loss": 0.5784,
+      "step": 4085
+    },
+    {
+      "epoch": 1.9476190476190476,
+      "grad_norm": 0.4789745893173457,
+      "learning_rate": 4.176081962092182e-08,
+      "loss": 0.5749,
+      "step": 4090
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4997223839030282,
+      "learning_rate": 3.805301908254455e-08,
+      "loss": 0.5871,
+      "step": 4095
+    },
+    {
+      "epoch": 1.9523809523809523,
+      "grad_norm": 0.49934046802623616,
+      "learning_rate": 3.451724678784518e-08,
+      "loss": 0.561,
+      "step": 4100
+    },
+    {
+      "epoch": 1.954761904761905,
+      "grad_norm": 0.4974229497151692,
+      "learning_rate": 3.115356379443601e-08,
+      "loss": 0.5661,
+      "step": 4105
+    },
+    {
+      "epoch": 1.9571428571428573,
+      "grad_norm": 0.5196515473348364,
+      "learning_rate": 2.796202818819871e-08,
+      "loss": 0.5856,
+      "step": 4110
+    },
+    {
+      "epoch": 1.9595238095238097,
+      "grad_norm": 0.49970636623212594,
+      "learning_rate": 2.4942695082281752e-08,
+      "loss": 0.5797,
+      "step": 4115
+    },
+    {
+      "epoch": 1.961904761904762,
+      "grad_norm": 0.5173329159698928,
+      "learning_rate": 2.2095616616150117e-08,
+      "loss": 0.5747,
+      "step": 4120
+    },
+    {
+      "epoch": 1.9642857142857144,
+      "grad_norm": 0.49959590807990745,
+      "learning_rate": 1.9420841954681525e-08,
+      "loss": 0.5646,
+      "step": 4125
+    },
+    {
+      "epoch": 1.9666666666666668,
+      "grad_norm": 0.48412036634259087,
+      "learning_rate": 1.6918417287318245e-08,
+      "loss": 0.5777,
+      "step": 4130
+    },
+    {
+      "epoch": 1.9690476190476192,
+      "grad_norm": 0.50340974957607,
+      "learning_rate": 1.4588385827272178e-08,
+      "loss": 0.5762,
+      "step": 4135
+    },
+    {
+      "epoch": 1.9714285714285715,
+      "grad_norm": 0.5095137173840767,
+      "learning_rate": 1.2430787810776556e-08,
+      "loss": 0.5779,
+      "step": 4140
+    },
+    {
+      "epoch": 1.973809523809524,
+      "grad_norm": 0.5201276330625599,
+      "learning_rate": 1.0445660496390952e-08,
+      "loss": 0.5829,
+      "step": 4145
+    },
+    {
+      "epoch": 1.9761904761904763,
+      "grad_norm": 0.4957082874285413,
+      "learning_rate": 8.633038164358454e-09,
+      "loss": 0.5782,
+      "step": 4150
+    },
+    {
+      "epoch": 1.9785714285714286,
+      "grad_norm": 0.49005045720787455,
+      "learning_rate": 6.992952116013918e-09,
+      "loss": 0.5769,
+      "step": 4155
+    },
+    {
+      "epoch": 1.980952380952381,
+      "grad_norm": 0.5163548919624951,
+      "learning_rate": 5.525430673244403e-09,
+      "loss": 0.574,
+      "step": 4160
+    },
+    {
+      "epoch": 1.9833333333333334,
+      "grad_norm": 0.5289724215651535,
+      "learning_rate": 4.230499177994007e-09,
+      "loss": 0.5773,
+      "step": 4165
+    },
+    {
+      "epoch": 1.9857142857142858,
+      "grad_norm": 0.485812450600075,
+      "learning_rate": 3.1081799918375454e-09,
+      "loss": 0.5835,
+      "step": 4170
+    },
+    {
+      "epoch": 1.9880952380952381,
+      "grad_norm": 0.48465992123370755,
+      "learning_rate": 2.1584924955819763e-09,
+      "loss": 0.5851,
+      "step": 4175
+    },
+    {
+      "epoch": 1.9904761904761905,
+      "grad_norm": 0.5073868341064405,
+      "learning_rate": 1.3814530889433298e-09,
+      "loss": 0.5656,
+      "step": 4180
+    },
+    {
+      "epoch": 1.9928571428571429,
+      "grad_norm": 0.4822167121104924,
+      "learning_rate": 7.770751902513862e-10,
+      "loss": 0.5766,
+      "step": 4185
+    },
+    {
+      "epoch": 1.9952380952380953,
+      "grad_norm": 0.5135567153768665,
+      "learning_rate": 3.4536923623096353e-10,
+      "loss": 0.5891,
+      "step": 4190
+    },
+    {
+      "epoch": 1.9976190476190476,
+      "grad_norm": 0.5058535961206515,
+      "learning_rate": 8.634268181095806e-11,
+      "loss": 0.5653,
+      "step": 4195
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5344236072743405,
+      "learning_rate": 0.0,
+      "loss": 0.56,
+      "step": 4200
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.741725742816925,
+      "eval_runtime": 110.7755,
+      "eval_samples_per_second": 77.282,
+      "eval_steps_per_second": 1.21,
+      "step": 4200
+    },
+    {
+      "epoch": 2.0,
+      "step": 4200,
+      "total_flos": 879394553856000.0,
+      "train_loss": 0.6596568237599872,
+      "train_runtime": 12326.0183,
+      "train_samples_per_second": 21.801,
+      "train_steps_per_second": 0.341
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4200,
+  "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": 879394553856000.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}