diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12267 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 17469,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9193334579467773,
+      "learning_rate": 1e-05,
+      "loss": 1.5817,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.434458017349243,
+      "learning_rate": 1e-05,
+      "loss": 1.3487,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4167187213897705,
+      "learning_rate": 1e-05,
+      "loss": 1.3269,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.078380823135376,
+      "learning_rate": 1e-05,
+      "loss": 1.3464,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6377758979797363,
+      "learning_rate": 1e-05,
+      "loss": 1.2832,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3049213886260986,
+      "learning_rate": 1e-05,
+      "loss": 1.3479,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2546005249023438,
+      "learning_rate": 1e-05,
+      "loss": 1.35,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.328960418701172,
+      "learning_rate": 1e-05,
+      "loss": 1.4031,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6193785667419434,
+      "learning_rate": 1e-05,
+      "loss": 1.2959,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4487783908843994,
+      "learning_rate": 1e-05,
+      "loss": 1.292,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7081260681152344,
+      "learning_rate": 1e-05,
+      "loss": 1.3578,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4693126678466797,
+      "learning_rate": 1e-05,
+      "loss": 1.346,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3680224418640137,
+      "learning_rate": 1e-05,
+      "loss": 1.2894,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.633376598358154,
+      "learning_rate": 1e-05,
+      "loss": 1.336,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.109175682067871,
+      "learning_rate": 1e-05,
+      "loss": 1.3236,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.653766393661499,
+      "learning_rate": 1e-05,
+      "loss": 1.301,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.339801788330078,
+      "learning_rate": 1e-05,
+      "loss": 1.35,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5112967491149902,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.812804937362671,
+      "learning_rate": 1e-05,
+      "loss": 1.4252,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4148075580596924,
+      "learning_rate": 1e-05,
+      "loss": 1.2881,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.221367120742798,
+      "learning_rate": 1e-05,
+      "loss": 1.2941,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8252381086349487,
+      "learning_rate": 1e-05,
+      "loss": 1.3305,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6195831298828125,
+      "learning_rate": 1e-05,
+      "loss": 1.2651,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.789186716079712,
+      "learning_rate": 1e-05,
+      "loss": 1.3352,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3786251544952393,
+      "learning_rate": 1e-05,
+      "loss": 1.3115,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0984363555908203,
+      "learning_rate": 1e-05,
+      "loss": 1.4102,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.215639591217041,
+      "learning_rate": 1e-05,
+      "loss": 1.3049,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.619779109954834,
+      "learning_rate": 1e-05,
+      "loss": 1.3686,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.424197196960449,
+      "learning_rate": 1e-05,
+      "loss": 1.3401,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6028305292129517,
+      "learning_rate": 1e-05,
+      "loss": 1.3297,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2444934844970703,
+      "learning_rate": 1e-05,
+      "loss": 1.369,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5839474201202393,
+      "learning_rate": 1e-05,
+      "loss": 1.3516,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0733659267425537,
+      "learning_rate": 1e-05,
+      "loss": 1.3525,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1361422538757324,
+      "learning_rate": 1e-05,
+      "loss": 1.3495,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.298948287963867,
+      "learning_rate": 1e-05,
+      "loss": 1.3479,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.780656337738037,
+      "learning_rate": 1e-05,
+      "loss": 1.3573,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.663308143615723,
+      "learning_rate": 1e-05,
+      "loss": 1.3882,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2056221961975098,
+      "learning_rate": 1e-05,
+      "loss": 1.3765,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.040045976638794,
+      "learning_rate": 1e-05,
+      "loss": 1.3591,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.020139694213867,
+      "learning_rate": 1e-05,
+      "loss": 1.4021,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6178290843963623,
+      "learning_rate": 1e-05,
+      "loss": 1.3574,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0279769897460938,
+      "learning_rate": 1e-05,
+      "loss": 1.2923,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8427987098693848,
+      "learning_rate": 1e-05,
+      "loss": 1.4751,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.862971782684326,
+      "learning_rate": 1e-05,
+      "loss": 1.4609,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7004852294921875,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.815732479095459,
+      "learning_rate": 1e-05,
+      "loss": 1.3416,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0111870765686035,
+      "learning_rate": 1e-05,
+      "loss": 1.2098,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.208897590637207,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9113574028015137,
+      "learning_rate": 1e-05,
+      "loss": 1.392,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8567959070205688,
+      "learning_rate": 1e-05,
+      "loss": 1.2849,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.282118320465088,
+      "learning_rate": 1e-05,
+      "loss": 1.3741,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6298553943634033,
+      "learning_rate": 1e-05,
+      "loss": 1.3421,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0505871772766113,
+      "learning_rate": 1e-05,
+      "loss": 1.308,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0846567153930664,
+      "learning_rate": 1e-05,
+      "loss": 1.4474,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3861382007598877,
+      "learning_rate": 1e-05,
+      "loss": 1.2142,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2822299003601074,
+      "learning_rate": 1e-05,
+      "loss": 1.2969,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0871715545654297,
+      "learning_rate": 1e-05,
+      "loss": 1.461,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1369597911834717,
+      "learning_rate": 1e-05,
+      "loss": 1.3103,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3976328372955322,
+      "learning_rate": 1e-05,
+      "loss": 1.3737,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.572042465209961,
+      "learning_rate": 1e-05,
+      "loss": 1.2747,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.673109531402588,
+      "learning_rate": 1e-05,
+      "loss": 1.4067,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.054105520248413,
+      "learning_rate": 1e-05,
+      "loss": 1.3836,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0587263107299805,
+      "learning_rate": 1e-05,
+      "loss": 1.3234,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9629148244857788,
+      "learning_rate": 1e-05,
+      "loss": 1.3481,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9313011169433594,
+      "learning_rate": 1e-05,
+      "loss": 1.334,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0011725425720215,
+      "learning_rate": 1e-05,
+      "loss": 1.3259,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.842392086982727,
+      "learning_rate": 1e-05,
+      "loss": 1.2759,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.944788932800293,
+      "learning_rate": 1e-05,
+      "loss": 1.3423,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.603177785873413,
+      "learning_rate": 1e-05,
+      "loss": 1.2942,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1123621463775635,
+      "learning_rate": 1e-05,
+      "loss": 1.2612,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.298437118530273,
+      "learning_rate": 1e-05,
+      "loss": 1.3443,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.184427261352539,
+      "learning_rate": 1e-05,
+      "loss": 1.2893,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.9805634021759033,
+      "learning_rate": 1e-05,
+      "loss": 1.3534,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.200209140777588,
+      "learning_rate": 1e-05,
+      "loss": 1.5907,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0145490169525146,
+      "learning_rate": 1e-05,
+      "loss": 1.3638,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1146225929260254,
+      "learning_rate": 1e-05,
+      "loss": 1.4594,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9472726583480835,
+      "learning_rate": 1e-05,
+      "loss": 1.4046,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.558044672012329,
+      "learning_rate": 1e-05,
+      "loss": 1.3187,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1615829467773438,
+      "learning_rate": 1e-05,
+      "loss": 1.3915,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4719650745391846,
+      "learning_rate": 1e-05,
+      "loss": 1.2974,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8266043663024902,
+      "learning_rate": 1e-05,
+      "loss": 1.4546,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.942686676979065,
+      "learning_rate": 1e-05,
+      "loss": 1.2597,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5434083938598633,
+      "learning_rate": 1e-05,
+      "loss": 1.3298,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2210099697113037,
+      "learning_rate": 1e-05,
+      "loss": 1.3343,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.09978985786438,
+      "learning_rate": 1e-05,
+      "loss": 1.2395,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9050910472869873,
+      "learning_rate": 1e-05,
+      "loss": 1.3362,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2624146938323975,
+      "learning_rate": 1e-05,
+      "loss": 1.3459,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.076939582824707,
+      "learning_rate": 1e-05,
+      "loss": 1.2785,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.847289562225342,
+      "learning_rate": 1e-05,
+      "loss": 1.3979,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6758952140808105,
+      "learning_rate": 1e-05,
+      "loss": 1.3496,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7525131702423096,
+      "learning_rate": 1e-05,
+      "loss": 1.412,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.080402135848999,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6619815826416016,
+      "learning_rate": 1e-05,
+      "loss": 1.3204,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1060123443603516,
+      "learning_rate": 1e-05,
+      "loss": 1.3582,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.734252452850342,
+      "learning_rate": 1e-05,
+      "loss": 1.3209,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7691597938537598,
+      "learning_rate": 1e-05,
+      "loss": 1.3503,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4318649768829346,
+      "learning_rate": 1e-05,
+      "loss": 1.2655,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.137103319168091,
+      "learning_rate": 1e-05,
+      "loss": 1.4927,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2520899772644043,
+      "learning_rate": 1e-05,
+      "loss": 1.3405,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5094006061553955,
+      "learning_rate": 1e-05,
+      "loss": 1.4049,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.372251033782959,
+      "learning_rate": 1e-05,
+      "loss": 1.3484,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1078174114227295,
+      "learning_rate": 1e-05,
+      "loss": 1.3468,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.324495315551758,
+      "learning_rate": 1e-05,
+      "loss": 1.3558,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.389418363571167,
+      "learning_rate": 1e-05,
+      "loss": 1.3081,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6623008251190186,
+      "learning_rate": 1e-05,
+      "loss": 1.3211,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8556021451950073,
+      "learning_rate": 1e-05,
+      "loss": 1.3689,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9243156909942627,
+      "learning_rate": 1e-05,
+      "loss": 1.2978,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.082162618637085,
+      "learning_rate": 1e-05,
+      "loss": 1.3405,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7506792545318604,
+      "learning_rate": 1e-05,
+      "loss": 1.3986,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.331587791442871,
+      "learning_rate": 1e-05,
+      "loss": 1.3117,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.870234727859497,
+      "learning_rate": 1e-05,
+      "loss": 1.4165,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.293771743774414,
+      "learning_rate": 1e-05,
+      "loss": 1.3671,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1112711429595947,
+      "learning_rate": 1e-05,
+      "loss": 1.4002,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0179507732391357,
+      "learning_rate": 1e-05,
+      "loss": 1.3784,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8166738748550415,
+      "learning_rate": 1e-05,
+      "loss": 1.3889,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.134061574935913,
+      "learning_rate": 1e-05,
+      "loss": 1.3192,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.651945114135742,
+      "learning_rate": 1e-05,
+      "loss": 1.3642,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9652693271636963,
+      "learning_rate": 1e-05,
+      "loss": 1.356,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.771118402481079,
+      "learning_rate": 1e-05,
+      "loss": 1.3345,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6212990283966064,
+      "learning_rate": 1e-05,
+      "loss": 1.3111,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.779235363006592,
+      "learning_rate": 1e-05,
+      "loss": 1.3837,
+      "step": 1210
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5824780464172363,
+      "learning_rate": 1e-05,
+      "loss": 1.2821,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7208638191223145,
+      "learning_rate": 1e-05,
+      "loss": 1.3874,
+      "step": 1230
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4320294857025146,
+      "learning_rate": 1e-05,
+      "loss": 1.317,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.185110330581665,
+      "learning_rate": 1e-05,
+      "loss": 1.3984,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3204574584960938,
+      "learning_rate": 1e-05,
+      "loss": 1.3539,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0239412784576416,
+      "learning_rate": 1e-05,
+      "loss": 1.3766,
+      "step": 1270
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.068580150604248,
+      "learning_rate": 1e-05,
+      "loss": 1.3301,
+      "step": 1280
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.349250555038452,
+      "learning_rate": 1e-05,
+      "loss": 1.2955,
+      "step": 1290
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8978281021118164,
+      "learning_rate": 1e-05,
+      "loss": 1.4535,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.288252353668213,
+      "learning_rate": 1e-05,
+      "loss": 1.2881,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.632664442062378,
+      "learning_rate": 1e-05,
+      "loss": 1.3044,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.277397394180298,
+      "learning_rate": 1e-05,
+      "loss": 1.2918,
+      "step": 1330
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.837981700897217,
+      "learning_rate": 1e-05,
+      "loss": 1.385,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.083287000656128,
+      "learning_rate": 1e-05,
+      "loss": 1.2665,
+      "step": 1350
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0568087100982666,
+      "learning_rate": 1e-05,
+      "loss": 1.1869,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4070591926574707,
+      "learning_rate": 1e-05,
+      "loss": 1.3262,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0220119953155518,
+      "learning_rate": 1e-05,
+      "loss": 1.3278,
+      "step": 1380
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2559282779693604,
+      "learning_rate": 1e-05,
+      "loss": 1.3381,
+      "step": 1390
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.258565902709961,
+      "learning_rate": 1e-05,
+      "loss": 1.3113,
+      "step": 1400
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2906646728515625,
+      "learning_rate": 1e-05,
+      "loss": 1.2238,
+      "step": 1410
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1853528022766113,
+      "learning_rate": 1e-05,
+      "loss": 1.3785,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.12762188911438,
+      "learning_rate": 1e-05,
+      "loss": 1.4494,
+      "step": 1430
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.401248574256897,
+      "learning_rate": 1e-05,
+      "loss": 1.2383,
+      "step": 1440
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6352481842041016,
+      "learning_rate": 1e-05,
+      "loss": 1.3514,
+      "step": 1450
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0247933864593506,
+      "learning_rate": 1e-05,
+      "loss": 1.2866,
+      "step": 1460
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5140461921691895,
+      "learning_rate": 1e-05,
+      "loss": 1.3917,
+      "step": 1470
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.628725051879883,
+      "learning_rate": 1e-05,
+      "loss": 1.3509,
+      "step": 1480
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9856632947921753,
+      "learning_rate": 1e-05,
+      "loss": 1.3269,
+      "step": 1490
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7394587993621826,
+      "learning_rate": 1e-05,
+      "loss": 1.3407,
+      "step": 1500
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4537129402160645,
+      "learning_rate": 1e-05,
+      "loss": 1.3306,
+      "step": 1510
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5386736392974854,
+      "learning_rate": 1e-05,
+      "loss": 1.3187,
+      "step": 1520
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3441104888916016,
+      "learning_rate": 1e-05,
+      "loss": 1.4401,
+      "step": 1530
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6362714767456055,
+      "learning_rate": 1e-05,
+      "loss": 1.337,
+      "step": 1540
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.830051064491272,
+      "learning_rate": 1e-05,
+      "loss": 1.3022,
+      "step": 1550
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8917646408081055,
+      "learning_rate": 1e-05,
+      "loss": 1.352,
+      "step": 1560
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4951720237731934,
+      "learning_rate": 1e-05,
+      "loss": 1.35,
+      "step": 1570
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9048365354537964,
+      "learning_rate": 1e-05,
+      "loss": 1.4491,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7537648677825928,
+      "learning_rate": 1e-05,
+      "loss": 1.2354,
+      "step": 1590
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.120579719543457,
+      "learning_rate": 1e-05,
+      "loss": 1.3471,
+      "step": 1600
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.85201096534729,
+      "learning_rate": 1e-05,
+      "loss": 1.2367,
+      "step": 1610
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9622502326965332,
+      "learning_rate": 1e-05,
+      "loss": 1.305,
+      "step": 1620
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.735097050666809,
+      "learning_rate": 1e-05,
+      "loss": 1.4083,
+      "step": 1630
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6957337856292725,
+      "learning_rate": 1e-05,
+      "loss": 1.29,
+      "step": 1640
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.07291841506958,
+      "learning_rate": 1e-05,
+      "loss": 1.3921,
+      "step": 1650
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0542261600494385,
+      "learning_rate": 1e-05,
+      "loss": 1.3678,
+      "step": 1660
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.107818126678467,
+      "learning_rate": 1e-05,
+      "loss": 1.2903,
+      "step": 1670
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1922590732574463,
+      "learning_rate": 1e-05,
+      "loss": 1.2826,
+      "step": 1680
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.963850736618042,
+      "learning_rate": 1e-05,
+      "loss": 1.3797,
+      "step": 1690
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5969626903533936,
+      "learning_rate": 1e-05,
+      "loss": 1.3398,
+      "step": 1700
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1566317081451416,
+      "learning_rate": 1e-05,
+      "loss": 1.245,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.023244619369507,
+      "learning_rate": 1e-05,
+      "loss": 1.3088,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4707469940185547,
+      "learning_rate": 1e-05,
+      "loss": 1.2738,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.61325204372406,
+      "learning_rate": 1e-05,
+      "loss": 1.3641,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0144903659820557,
+      "learning_rate": 1e-05,
+      "loss": 1.3668,
+      "step": 1750
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6015106439590454,
+      "learning_rate": 1e-05,
+      "loss": 1.4913,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9650461673736572,
+      "learning_rate": 1e-05,
+      "loss": 1.4019,
+      "step": 1770
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.607697010040283,
+      "learning_rate": 1e-05,
+      "loss": 1.2866,
+      "step": 1780
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2201220989227295,
+      "learning_rate": 1e-05,
+      "loss": 1.3585,
+      "step": 1790
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.4735159873962402,
+      "learning_rate": 1e-05,
+      "loss": 1.4571,
+      "step": 1800
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2776782512664795,
+      "learning_rate": 1e-05,
+      "loss": 1.3219,
+      "step": 1810
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.329009532928467,
+      "learning_rate": 1e-05,
+      "loss": 1.385,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.494612216949463,
+      "learning_rate": 1e-05,
+      "loss": 1.2941,
+      "step": 1830
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.000184059143066,
+      "learning_rate": 1e-05,
+      "loss": 1.3872,
+      "step": 1840
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1177732944488525,
+      "learning_rate": 1e-05,
+      "loss": 1.414,
+      "step": 1850
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.023552894592285,
+      "learning_rate": 1e-05,
+      "loss": 1.3941,
+      "step": 1860
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5864198207855225,
+      "learning_rate": 1e-05,
+      "loss": 1.345,
+      "step": 1870
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5800952911376953,
+      "learning_rate": 1e-05,
+      "loss": 1.3384,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.223054885864258,
+      "learning_rate": 1e-05,
+      "loss": 1.4299,
+      "step": 1890
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.060757875442505,
+      "learning_rate": 1e-05,
+      "loss": 1.3994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.429835557937622,
+      "learning_rate": 1e-05,
+      "loss": 1.3322,
+      "step": 1910
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8696309328079224,
+      "learning_rate": 1e-05,
+      "loss": 1.25,
+      "step": 1920
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8844784498214722,
+      "learning_rate": 1e-05,
+      "loss": 1.2992,
+      "step": 1930
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0955729484558105,
+      "learning_rate": 1e-05,
+      "loss": 1.339,
+      "step": 1940
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8389264345169067,
+      "learning_rate": 1e-05,
+      "loss": 1.3021,
+      "step": 1950
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6430881023406982,
+      "learning_rate": 1e-05,
+      "loss": 1.3589,
+      "step": 1960
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4001126289367676,
+      "learning_rate": 1e-05,
+      "loss": 1.286,
+      "step": 1970
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.283057928085327,
+      "learning_rate": 1e-05,
+      "loss": 1.3626,
+      "step": 1980
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0561628341674805,
+      "learning_rate": 1e-05,
+      "loss": 1.4019,
+      "step": 1990
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2793378829956055,
+      "learning_rate": 1e-05,
+      "loss": 1.354,
+      "step": 2000
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4692368507385254,
+      "learning_rate": 1e-05,
+      "loss": 1.3677,
+      "step": 2010
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9414951801300049,
+      "learning_rate": 1e-05,
+      "loss": 1.2852,
+      "step": 2020
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1738226413726807,
+      "learning_rate": 1e-05,
+      "loss": 1.2924,
+      "step": 2030
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1066489219665527,
+      "learning_rate": 1e-05,
+      "loss": 1.304,
+      "step": 2040
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.439203262329102,
+      "learning_rate": 1e-05,
+      "loss": 1.3377,
+      "step": 2050
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1386377811431885,
+      "learning_rate": 1e-05,
+      "loss": 1.3543,
+      "step": 2060
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.179619073867798,
+      "learning_rate": 1e-05,
+      "loss": 1.4117,
+      "step": 2070
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7357661724090576,
+      "learning_rate": 1e-05,
+      "loss": 1.4537,
+      "step": 2080
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8351653814315796,
+      "learning_rate": 1e-05,
+      "loss": 1.2952,
+      "step": 2090
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.759627342224121,
+      "learning_rate": 1e-05,
+      "loss": 1.4223,
+      "step": 2100
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.220985174179077,
+      "learning_rate": 1e-05,
+      "loss": 1.3881,
+      "step": 2110
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.878201484680176,
+      "learning_rate": 1e-05,
+      "loss": 1.3814,
+      "step": 2120
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.227640390396118,
+      "learning_rate": 1e-05,
+      "loss": 1.315,
+      "step": 2130
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.974422574043274,
+      "learning_rate": 1e-05,
+      "loss": 1.2577,
+      "step": 2140
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6234285831451416,
+      "learning_rate": 1e-05,
+      "loss": 1.3741,
+      "step": 2150
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4010727405548096,
+      "learning_rate": 1e-05,
+      "loss": 1.3185,
+      "step": 2160
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.83111310005188,
+      "learning_rate": 1e-05,
+      "loss": 1.3551,
+      "step": 2170
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.351682424545288,
+      "learning_rate": 1e-05,
+      "loss": 1.3692,
+      "step": 2180
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.604748487472534,
+      "learning_rate": 1e-05,
+      "loss": 1.2968,
+      "step": 2190
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1189563274383545,
+      "learning_rate": 1e-05,
+      "loss": 1.3221,
+      "step": 2200
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.773430347442627,
+      "learning_rate": 1e-05,
+      "loss": 1.2218,
+      "step": 2210
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6280314922332764,
+      "learning_rate": 1e-05,
+      "loss": 1.3572,
+      "step": 2220
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.858031749725342,
+      "learning_rate": 1e-05,
+      "loss": 1.3583,
+      "step": 2230
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.839733362197876,
+      "learning_rate": 1e-05,
+      "loss": 1.3644,
+      "step": 2240
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.971848726272583,
+      "learning_rate": 1e-05,
+      "loss": 1.2817,
+      "step": 2250
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.195857524871826,
+      "learning_rate": 1e-05,
+      "loss": 1.3027,
+      "step": 2260
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9531887769699097,
+      "learning_rate": 1e-05,
+      "loss": 1.3062,
+      "step": 2270
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1299996376037598,
+      "learning_rate": 1e-05,
+      "loss": 1.3014,
+      "step": 2280
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4937584400177,
+      "learning_rate": 1e-05,
+      "loss": 1.2315,
+      "step": 2290
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.707040309906006,
+      "learning_rate": 1e-05,
+      "loss": 1.3776,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4510104656219482,
+      "learning_rate": 1e-05,
+      "loss": 1.4079,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.667187452316284,
+      "learning_rate": 1e-05,
+      "loss": 1.3291,
+      "step": 2320
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8011958599090576,
+      "learning_rate": 1e-05,
+      "loss": 1.305,
+      "step": 2330
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.067420482635498,
+      "learning_rate": 1e-05,
+      "loss": 1.2674,
+      "step": 2340
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.243347644805908,
+      "learning_rate": 1e-05,
+      "loss": 1.2858,
+      "step": 2350
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0893757343292236,
+      "learning_rate": 1e-05,
+      "loss": 1.3556,
+      "step": 2360
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.331129550933838,
+      "learning_rate": 1e-05,
+      "loss": 1.3768,
+      "step": 2370
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0916008949279785,
+      "learning_rate": 1e-05,
+      "loss": 1.2819,
+      "step": 2380
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.073061227798462,
+      "learning_rate": 1e-05,
+      "loss": 1.2555,
+      "step": 2390
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3167691230773926,
+      "learning_rate": 1e-05,
+      "loss": 1.2621,
+      "step": 2400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0317132472991943,
+      "learning_rate": 1e-05,
+      "loss": 1.4108,
+      "step": 2410
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.494563579559326,
+      "learning_rate": 1e-05,
+      "loss": 1.2998,
+      "step": 2420
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.281918525695801,
+      "learning_rate": 1e-05,
+      "loss": 1.4285,
+      "step": 2430
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.9818170070648193,
+      "learning_rate": 1e-05,
+      "loss": 1.3844,
+      "step": 2440
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.736210823059082,
+      "learning_rate": 1e-05,
+      "loss": 1.3575,
+      "step": 2450
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2397067546844482,
+      "learning_rate": 1e-05,
+      "loss": 1.2842,
+      "step": 2460
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.424915313720703,
+      "learning_rate": 1e-05,
+      "loss": 1.25,
+      "step": 2470
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1230766773223877,
+      "learning_rate": 1e-05,
+      "loss": 1.3246,
+      "step": 2480
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0383758544921875,
+      "learning_rate": 1e-05,
+      "loss": 1.2224,
+      "step": 2490
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.118562698364258,
+      "learning_rate": 1e-05,
+      "loss": 1.4465,
+      "step": 2500
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.747936964035034,
+      "learning_rate": 1e-05,
+      "loss": 1.3906,
+      "step": 2510
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0030086040496826,
+      "learning_rate": 1e-05,
+      "loss": 1.3728,
+      "step": 2520
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7183146476745605,
+      "learning_rate": 1e-05,
+      "loss": 1.3141,
+      "step": 2530
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8749139308929443,
+      "learning_rate": 1e-05,
+      "loss": 1.3219,
+      "step": 2540
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.784529685974121,
+      "learning_rate": 1e-05,
+      "loss": 1.4004,
+      "step": 2550
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.992795944213867,
+      "learning_rate": 1e-05,
+      "loss": 1.2798,
+      "step": 2560
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0557336807250977,
+      "learning_rate": 1e-05,
+      "loss": 1.1834,
+      "step": 2570
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8129782676696777,
+      "learning_rate": 1e-05,
+      "loss": 1.4837,
+      "step": 2580
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.194706439971924,
+      "learning_rate": 1e-05,
+      "loss": 1.3455,
+      "step": 2590
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.341437816619873,
+      "learning_rate": 1e-05,
+      "loss": 1.1929,
+      "step": 2600
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.618860960006714,
+      "learning_rate": 1e-05,
+      "loss": 1.3953,
+      "step": 2610
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1975460052490234,
+      "learning_rate": 1e-05,
+      "loss": 1.3278,
+      "step": 2620
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3264036178588867,
+      "learning_rate": 1e-05,
+      "loss": 1.2772,
+      "step": 2630
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.271860361099243,
+      "learning_rate": 1e-05,
+      "loss": 1.2736,
+      "step": 2640
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.292814016342163,
+      "learning_rate": 1e-05,
+      "loss": 1.2622,
+      "step": 2650
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3970437049865723,
+      "learning_rate": 1e-05,
+      "loss": 1.399,
+      "step": 2660
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5156729221343994,
+      "learning_rate": 1e-05,
+      "loss": 1.2094,
+      "step": 2670
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0000627040863037,
+      "learning_rate": 1e-05,
+      "loss": 1.2816,
+      "step": 2680
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0072364807128906,
+      "learning_rate": 1e-05,
+      "loss": 1.303,
+      "step": 2690
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7943062782287598,
+      "learning_rate": 1e-05,
+      "loss": 1.3585,
+      "step": 2700
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5227713584899902,
+      "learning_rate": 1e-05,
+      "loss": 1.2586,
+      "step": 2710
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.766931533813477,
+      "learning_rate": 1e-05,
+      "loss": 1.4304,
+      "step": 2720
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1842198371887207,
+      "learning_rate": 1e-05,
+      "loss": 1.2814,
+      "step": 2730
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2196881771087646,
+      "learning_rate": 1e-05,
+      "loss": 1.2937,
+      "step": 2740
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.007570743560791,
+      "learning_rate": 1e-05,
+      "loss": 1.378,
+      "step": 2750
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.672393798828125,
+      "learning_rate": 1e-05,
+      "loss": 1.3915,
+      "step": 2760
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.169638156890869,
+      "learning_rate": 1e-05,
+      "loss": 1.3554,
+      "step": 2770
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4125783443450928,
+      "learning_rate": 1e-05,
+      "loss": 1.3345,
+      "step": 2780
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.443183660507202,
+      "learning_rate": 1e-05,
+      "loss": 1.3754,
+      "step": 2790
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.338090419769287,
+      "learning_rate": 1e-05,
+      "loss": 1.3564,
+      "step": 2800
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6799206733703613,
+      "learning_rate": 1e-05,
+      "loss": 1.4033,
+      "step": 2810
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.752776622772217,
+      "learning_rate": 1e-05,
+      "loss": 1.3011,
+      "step": 2820
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.800459384918213,
+      "learning_rate": 1e-05,
+      "loss": 1.3509,
+      "step": 2830
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.97355318069458,
+      "learning_rate": 1e-05,
+      "loss": 1.478,
+      "step": 2840
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.2402892112731934,
+      "learning_rate": 1e-05,
+      "loss": 1.2713,
+      "step": 2850
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3085737228393555,
+      "learning_rate": 1e-05,
+      "loss": 1.3607,
+      "step": 2860
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.410074234008789,
+      "learning_rate": 1e-05,
+      "loss": 1.3084,
+      "step": 2870
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0424160957336426,
+      "learning_rate": 1e-05,
+      "loss": 1.325,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.200083017349243,
+      "learning_rate": 1e-05,
+      "loss": 1.3144,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7350451946258545,
+      "learning_rate": 1e-05,
+      "loss": 1.3447,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.373868703842163,
+      "learning_rate": 1e-05,
+      "loss": 1.3489,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9406344890594482,
+      "learning_rate": 1e-05,
+      "loss": 1.349,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.075248956680298,
+      "learning_rate": 1e-05,
+      "loss": 1.2307,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.607004404067993,
+      "learning_rate": 1e-05,
+      "loss": 1.3246,
+      "step": 2940
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0145184993743896,
+      "learning_rate": 1e-05,
+      "loss": 1.3412,
+      "step": 2950
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.537825345993042,
+      "learning_rate": 1e-05,
+      "loss": 1.3499,
+      "step": 2960
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.785065174102783,
+      "learning_rate": 1e-05,
+      "loss": 1.325,
+      "step": 2970
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4221601486206055,
+      "learning_rate": 1e-05,
+      "loss": 1.3289,
+      "step": 2980
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5729472637176514,
+      "learning_rate": 1e-05,
+      "loss": 1.3444,
+      "step": 2990
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2976810932159424,
+      "learning_rate": 1e-05,
+      "loss": 1.2523,
+      "step": 3000
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.300956964492798,
+      "learning_rate": 1e-05,
+      "loss": 1.3421,
+      "step": 3010
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.143294334411621,
+      "learning_rate": 1e-05,
+      "loss": 1.3209,
+      "step": 3020
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.957260251045227,
+      "learning_rate": 1e-05,
+      "loss": 1.3478,
+      "step": 3030
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4198296070098877,
+      "learning_rate": 1e-05,
+      "loss": 1.39,
+      "step": 3040
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0126090049743652,
+      "learning_rate": 1e-05,
+      "loss": 1.3365,
+      "step": 3050
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9037426710128784,
+      "learning_rate": 1e-05,
+      "loss": 1.2968,
+      "step": 3060
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2140746116638184,
+      "learning_rate": 1e-05,
+      "loss": 1.3395,
+      "step": 3070
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4906034469604492,
+      "learning_rate": 1e-05,
+      "loss": 1.3678,
+      "step": 3080
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0024867057800293,
+      "learning_rate": 1e-05,
+      "loss": 1.274,
+      "step": 3090
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1890878677368164,
+      "learning_rate": 1e-05,
+      "loss": 1.3005,
+      "step": 3100
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7136616706848145,
+      "learning_rate": 1e-05,
+      "loss": 1.3384,
+      "step": 3110
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1568996906280518,
+      "learning_rate": 1e-05,
+      "loss": 1.3166,
+      "step": 3120
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9385143518447876,
+      "learning_rate": 1e-05,
+      "loss": 1.2731,
+      "step": 3130
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.474834680557251,
+      "learning_rate": 1e-05,
+      "loss": 1.3038,
+      "step": 3140
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2946722507476807,
+      "learning_rate": 1e-05,
+      "loss": 1.2439,
+      "step": 3150
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2470133304595947,
+      "learning_rate": 1e-05,
+      "loss": 1.3573,
+      "step": 3160
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.878171443939209,
+      "learning_rate": 1e-05,
+      "loss": 1.3902,
+      "step": 3170
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.551673173904419,
+      "learning_rate": 1e-05,
+      "loss": 1.2785,
+      "step": 3180
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7930572032928467,
+      "learning_rate": 1e-05,
+      "loss": 1.3438,
+      "step": 3190
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.136298179626465,
+      "learning_rate": 1e-05,
+      "loss": 1.2169,
+      "step": 3200
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7656638622283936,
+      "learning_rate": 1e-05,
+      "loss": 1.3341,
+      "step": 3210
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9639734029769897,
+      "learning_rate": 1e-05,
+      "loss": 1.3665,
+      "step": 3220
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.61297345161438,
+      "learning_rate": 1e-05,
+      "loss": 1.3103,
+      "step": 3230
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.183431386947632,
+      "learning_rate": 1e-05,
+      "loss": 1.3381,
+      "step": 3240
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.3795530796051025,
+      "learning_rate": 1e-05,
+      "loss": 1.417,
+      "step": 3250
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.683171033859253,
+      "learning_rate": 1e-05,
+      "loss": 1.3867,
+      "step": 3260
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4533402919769287,
+      "learning_rate": 1e-05,
+      "loss": 1.2866,
+      "step": 3270
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.316845417022705,
+      "learning_rate": 1e-05,
+      "loss": 1.3988,
+      "step": 3280
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9724698066711426,
+      "learning_rate": 1e-05,
+      "loss": 1.3656,
+      "step": 3290
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1463451385498047,
+      "learning_rate": 1e-05,
+      "loss": 1.3868,
+      "step": 3300
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.425827741622925,
+      "learning_rate": 1e-05,
+      "loss": 1.4846,
+      "step": 3310
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0959482192993164,
+      "learning_rate": 1e-05,
+      "loss": 1.3961,
+      "step": 3320
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.785223007202148,
+      "learning_rate": 1e-05,
+      "loss": 1.3817,
+      "step": 3330
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5019736289978027,
+      "learning_rate": 1e-05,
+      "loss": 1.3205,
+      "step": 3340
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9260520935058594,
+      "learning_rate": 1e-05,
+      "loss": 1.2854,
+      "step": 3350
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.714916229248047,
+      "learning_rate": 1e-05,
+      "loss": 1.3533,
+      "step": 3360
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.024867534637451,
+      "learning_rate": 1e-05,
+      "loss": 1.3566,
+      "step": 3370
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.382457971572876,
+      "learning_rate": 1e-05,
+      "loss": 1.276,
+      "step": 3380
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.94223690032959,
+      "learning_rate": 1e-05,
+      "loss": 1.3663,
+      "step": 3390
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0350470542907715,
+      "learning_rate": 1e-05,
+      "loss": 1.2721,
+      "step": 3400
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4905200004577637,
+      "learning_rate": 1e-05,
+      "loss": 1.2781,
+      "step": 3410
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.03529953956604,
+      "learning_rate": 1e-05,
+      "loss": 1.3776,
+      "step": 3420
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4352126121520996,
+      "learning_rate": 1e-05,
+      "loss": 1.2821,
+      "step": 3430
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3159918785095215,
+      "learning_rate": 1e-05,
+      "loss": 1.3652,
+      "step": 3440
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.64506459236145,
+      "learning_rate": 1e-05,
+      "loss": 1.2642,
+      "step": 3450
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.939044713973999,
+      "learning_rate": 1e-05,
+      "loss": 1.3028,
+      "step": 3460
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9863433837890625,
+      "learning_rate": 1e-05,
+      "loss": 1.4451,
+      "step": 3470
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6311700344085693,
+      "learning_rate": 1e-05,
+      "loss": 1.3471,
+      "step": 3480
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2302627563476562,
+      "learning_rate": 1e-05,
+      "loss": 1.434,
+      "step": 3490
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.555927276611328,
+      "learning_rate": 1e-05,
+      "loss": 1.2639,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8889572620391846,
+      "learning_rate": 1e-05,
+      "loss": 1.3316,
+      "step": 3510
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.86160409450531,
+      "learning_rate": 1e-05,
+      "loss": 1.2543,
+      "step": 3520
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.794211506843567,
+      "learning_rate": 1e-05,
+      "loss": 1.373,
+      "step": 3530
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.82426118850708,
+      "learning_rate": 1e-05,
+      "loss": 1.2373,
+      "step": 3540
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1324963569641113,
+      "learning_rate": 1e-05,
+      "loss": 1.2348,
+      "step": 3550
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.985074043273926,
+      "learning_rate": 1e-05,
+      "loss": 1.3833,
+      "step": 3560
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9578144550323486,
+      "learning_rate": 1e-05,
+      "loss": 1.3657,
+      "step": 3570
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3407962322235107,
+      "learning_rate": 1e-05,
+      "loss": 1.2717,
+      "step": 3580
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.082918643951416,
+      "learning_rate": 1e-05,
+      "loss": 1.3622,
+      "step": 3590
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.811771273612976,
+      "learning_rate": 1e-05,
+      "loss": 1.3736,
+      "step": 3600
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3119630813598633,
+      "learning_rate": 1e-05,
+      "loss": 1.339,
+      "step": 3610
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.398604393005371,
+      "learning_rate": 1e-05,
+      "loss": 1.3543,
+      "step": 3620
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1124989986419678,
+      "learning_rate": 1e-05,
+      "loss": 1.2732,
+      "step": 3630
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2756052017211914,
+      "learning_rate": 1e-05,
+      "loss": 1.2775,
+      "step": 3640
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8524532318115234,
+      "learning_rate": 1e-05,
+      "loss": 1.3438,
+      "step": 3650
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.433467149734497,
+      "learning_rate": 1e-05,
+      "loss": 1.2721,
+      "step": 3660
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1145050525665283,
+      "learning_rate": 1e-05,
+      "loss": 1.2946,
+      "step": 3670
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.255388021469116,
+      "learning_rate": 1e-05,
+      "loss": 1.3833,
+      "step": 3680
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.235727310180664,
+      "learning_rate": 1e-05,
+      "loss": 1.3377,
+      "step": 3690
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.267134189605713,
+      "learning_rate": 1e-05,
+      "loss": 1.4201,
+      "step": 3700
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.27997088432312,
+      "learning_rate": 1e-05,
+      "loss": 1.2713,
+      "step": 3710
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4130237102508545,
+      "learning_rate": 1e-05,
+      "loss": 1.3445,
+      "step": 3720
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.032923460006714,
+      "learning_rate": 1e-05,
+      "loss": 1.3873,
+      "step": 3730
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0984504222869873,
+      "learning_rate": 1e-05,
+      "loss": 1.4304,
+      "step": 3740
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.5242090225219727,
+      "learning_rate": 1e-05,
+      "loss": 1.2671,
+      "step": 3750
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.271066665649414,
+      "learning_rate": 1e-05,
+      "loss": 1.2898,
+      "step": 3760
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8168696165084839,
+      "learning_rate": 1e-05,
+      "loss": 1.3366,
+      "step": 3770
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.690077781677246,
+      "learning_rate": 1e-05,
+      "loss": 1.3832,
+      "step": 3780
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2696776390075684,
+      "learning_rate": 1e-05,
+      "loss": 1.2963,
+      "step": 3790
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6965296268463135,
+      "learning_rate": 1e-05,
+      "loss": 1.3049,
+      "step": 3800
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.884469985961914,
+      "learning_rate": 1e-05,
+      "loss": 1.3735,
+      "step": 3810
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.111697673797607,
+      "learning_rate": 1e-05,
+      "loss": 1.2462,
+      "step": 3820
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5264475345611572,
+      "learning_rate": 1e-05,
+      "loss": 1.2981,
+      "step": 3830
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9028539657592773,
+      "learning_rate": 1e-05,
+      "loss": 1.3386,
+      "step": 3840
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7717933654785156,
+      "learning_rate": 1e-05,
+      "loss": 1.3897,
+      "step": 3850
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.108096122741699,
+      "learning_rate": 1e-05,
+      "loss": 1.3349,
+      "step": 3860
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7096315622329712,
+      "learning_rate": 1e-05,
+      "loss": 1.4314,
+      "step": 3870
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.125641107559204,
+      "learning_rate": 1e-05,
+      "loss": 1.3241,
+      "step": 3880
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9365934133529663,
+      "learning_rate": 1e-05,
+      "loss": 1.4185,
+      "step": 3890
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.282160997390747,
+      "learning_rate": 1e-05,
+      "loss": 1.4088,
+      "step": 3900
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8728342056274414,
+      "learning_rate": 1e-05,
+      "loss": 1.3255,
+      "step": 3910
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0046260356903076,
+      "learning_rate": 1e-05,
+      "loss": 1.2997,
+      "step": 3920
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4220221042633057,
+      "learning_rate": 1e-05,
+      "loss": 1.304,
+      "step": 3930
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.533700942993164,
+      "learning_rate": 1e-05,
+      "loss": 1.2687,
+      "step": 3940
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.124030590057373,
+      "learning_rate": 1e-05,
+      "loss": 1.3217,
+      "step": 3950
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.410753011703491,
+      "learning_rate": 1e-05,
+      "loss": 1.2749,
+      "step": 3960
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2407126426696777,
+      "learning_rate": 1e-05,
+      "loss": 1.3307,
+      "step": 3970
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7697319984436035,
+      "learning_rate": 1e-05,
+      "loss": 1.2859,
+      "step": 3980
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.492711067199707,
+      "learning_rate": 1e-05,
+      "loss": 1.3889,
+      "step": 3990
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.103754997253418,
+      "learning_rate": 1e-05,
+      "loss": 1.3198,
+      "step": 4000
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.285733699798584,
+      "learning_rate": 1e-05,
+      "loss": 1.3397,
+      "step": 4010
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.908586859703064,
+      "learning_rate": 1e-05,
+      "loss": 1.4036,
+      "step": 4020
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.6941065788269043,
+      "learning_rate": 1e-05,
+      "loss": 1.3399,
+      "step": 4030
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5054636001586914,
+      "learning_rate": 1e-05,
+      "loss": 1.2888,
+      "step": 4040
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9840850830078125,
+      "learning_rate": 1e-05,
+      "loss": 1.39,
+      "step": 4050
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9543534517288208,
+      "learning_rate": 1e-05,
+      "loss": 1.3243,
+      "step": 4060
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3872506618499756,
+      "learning_rate": 1e-05,
+      "loss": 1.3321,
+      "step": 4070
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.167403221130371,
+      "learning_rate": 1e-05,
+      "loss": 1.297,
+      "step": 4080
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.350149631500244,
+      "learning_rate": 1e-05,
+      "loss": 1.2943,
+      "step": 4090
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.307448148727417,
+      "learning_rate": 1e-05,
+      "loss": 1.1838,
+      "step": 4100
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.179839849472046,
+      "learning_rate": 1e-05,
+      "loss": 1.3981,
+      "step": 4110
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8496357202529907,
+      "learning_rate": 1e-05,
+      "loss": 1.3317,
+      "step": 4120
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6441686153411865,
+      "learning_rate": 1e-05,
+      "loss": 1.3354,
+      "step": 4130
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0524277687072754,
+      "learning_rate": 1e-05,
+      "loss": 1.3213,
+      "step": 4140
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1615052223205566,
+      "learning_rate": 1e-05,
+      "loss": 1.2763,
+      "step": 4150
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3598992824554443,
+      "learning_rate": 1e-05,
+      "loss": 1.2179,
+      "step": 4160
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8935139179229736,
+      "learning_rate": 1e-05,
+      "loss": 1.3008,
+      "step": 4170
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8705596923828125,
+      "learning_rate": 1e-05,
+      "loss": 1.359,
+      "step": 4180
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2775766849517822,
+      "learning_rate": 1e-05,
+      "loss": 1.2492,
+      "step": 4190
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.049804449081421,
+      "learning_rate": 1e-05,
+      "loss": 1.3017,
+      "step": 4200
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.507704257965088,
+      "learning_rate": 1e-05,
+      "loss": 1.237,
+      "step": 4210
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9242076873779297,
+      "learning_rate": 1e-05,
+      "loss": 1.4219,
+      "step": 4220
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7825714349746704,
+      "learning_rate": 1e-05,
+      "loss": 1.3866,
+      "step": 4230
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5417842864990234,
+      "learning_rate": 1e-05,
+      "loss": 1.3076,
+      "step": 4240
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.281651020050049,
+      "learning_rate": 1e-05,
+      "loss": 1.3076,
+      "step": 4250
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7986044883728027,
+      "learning_rate": 1e-05,
+      "loss": 1.2935,
+      "step": 4260
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2875053882598877,
+      "learning_rate": 1e-05,
+      "loss": 1.2891,
+      "step": 4270
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.329230308532715,
+      "learning_rate": 1e-05,
+      "loss": 1.2477,
+      "step": 4280
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3861818313598633,
+      "learning_rate": 1e-05,
+      "loss": 1.4013,
+      "step": 4290
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9318095445632935,
+      "learning_rate": 1e-05,
+      "loss": 1.2565,
+      "step": 4300
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.428346633911133,
+      "learning_rate": 1e-05,
+      "loss": 1.3452,
+      "step": 4310
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1091361045837402,
+      "learning_rate": 1e-05,
+      "loss": 1.37,
+      "step": 4320
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.457003593444824,
+      "learning_rate": 1e-05,
+      "loss": 1.3968,
+      "step": 4330
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.365961790084839,
+      "learning_rate": 1e-05,
+      "loss": 1.4461,
+      "step": 4340
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6295409202575684,
+      "learning_rate": 1e-05,
+      "loss": 1.4079,
+      "step": 4350
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9824382066726685,
+      "learning_rate": 1e-05,
+      "loss": 1.3563,
+      "step": 4360
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.455552101135254,
+      "learning_rate": 1e-05,
+      "loss": 1.4229,
+      "step": 4370
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2429590225219727,
+      "learning_rate": 1e-05,
+      "loss": 1.312,
+      "step": 4380
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8446462154388428,
+      "learning_rate": 1e-05,
+      "loss": 1.382,
+      "step": 4390
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4450223445892334,
+      "learning_rate": 1e-05,
+      "loss": 1.3029,
+      "step": 4400
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3352503776550293,
+      "learning_rate": 1e-05,
+      "loss": 1.3553,
+      "step": 4410
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1491830348968506,
+      "learning_rate": 1e-05,
+      "loss": 1.3227,
+      "step": 4420
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.02473521232605,
+      "learning_rate": 1e-05,
+      "loss": 1.2365,
+      "step": 4430
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.059314489364624,
+      "learning_rate": 1e-05,
+      "loss": 1.4442,
+      "step": 4440
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8352389335632324,
+      "learning_rate": 1e-05,
+      "loss": 1.3092,
+      "step": 4450
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.165952205657959,
+      "learning_rate": 1e-05,
+      "loss": 1.335,
+      "step": 4460
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.213463544845581,
+      "learning_rate": 1e-05,
+      "loss": 1.2948,
+      "step": 4470
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2183351516723633,
+      "learning_rate": 1e-05,
+      "loss": 1.341,
+      "step": 4480
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9775967597961426,
+      "learning_rate": 1e-05,
+      "loss": 1.2896,
+      "step": 4490
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.169218063354492,
+      "learning_rate": 1e-05,
+      "loss": 1.3855,
+      "step": 4500
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.246225357055664,
+      "learning_rate": 1e-05,
+      "loss": 1.3308,
+      "step": 4510
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.859213352203369,
+      "learning_rate": 1e-05,
+      "loss": 1.2775,
+      "step": 4520
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4901084899902344,
+      "learning_rate": 1e-05,
+      "loss": 1.277,
+      "step": 4530
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7237708568572998,
+      "learning_rate": 1e-05,
+      "loss": 1.233,
+      "step": 4540
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.759042501449585,
+      "learning_rate": 1e-05,
+      "loss": 1.191,
+      "step": 4550
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1091668605804443,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 4560
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2881340980529785,
+      "learning_rate": 1e-05,
+      "loss": 1.3126,
+      "step": 4570
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.373076915740967,
+      "learning_rate": 1e-05,
+      "loss": 1.3503,
+      "step": 4580
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.527299404144287,
+      "learning_rate": 1e-05,
+      "loss": 1.3033,
+      "step": 4590
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6886786222457886,
+      "learning_rate": 1e-05,
+      "loss": 1.3138,
+      "step": 4600
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.049691915512085,
+      "learning_rate": 1e-05,
+      "loss": 1.4259,
+      "step": 4610
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.689084529876709,
+      "learning_rate": 1e-05,
+      "loss": 1.2643,
+      "step": 4620
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.415674924850464,
+      "learning_rate": 1e-05,
+      "loss": 1.3704,
+      "step": 4630
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1368796825408936,
+      "learning_rate": 1e-05,
+      "loss": 1.403,
+      "step": 4640
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.140413761138916,
+      "learning_rate": 1e-05,
+      "loss": 1.2695,
+      "step": 4650
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2510221004486084,
+      "learning_rate": 1e-05,
+      "loss": 1.3455,
+      "step": 4660
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.8963171243667603,
+      "learning_rate": 1e-05,
+      "loss": 1.3342,
+      "step": 4670
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0337469577789307,
+      "learning_rate": 1e-05,
+      "loss": 1.3418,
+      "step": 4680
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0309271812438965,
+      "learning_rate": 1e-05,
+      "loss": 1.3451,
+      "step": 4690
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.512277603149414,
+      "learning_rate": 1e-05,
+      "loss": 1.3368,
+      "step": 4700
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1851866245269775,
+      "learning_rate": 1e-05,
+      "loss": 1.3171,
+      "step": 4710
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9769792556762695,
+      "learning_rate": 1e-05,
+      "loss": 1.3401,
+      "step": 4720
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2425997257232666,
+      "learning_rate": 1e-05,
+      "loss": 1.4102,
+      "step": 4730
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.012803316116333,
+      "learning_rate": 1e-05,
+      "loss": 1.2514,
+      "step": 4740
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3681206703186035,
+      "learning_rate": 1e-05,
+      "loss": 1.2983,
+      "step": 4750
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1048636436462402,
+      "learning_rate": 1e-05,
+      "loss": 1.2526,
+      "step": 4760
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0571296215057373,
+      "learning_rate": 1e-05,
+      "loss": 1.3304,
+      "step": 4770
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8607044219970703,
+      "learning_rate": 1e-05,
+      "loss": 1.3497,
+      "step": 4780
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.278191566467285,
+      "learning_rate": 1e-05,
+      "loss": 1.1942,
+      "step": 4790
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9115612506866455,
+      "learning_rate": 1e-05,
+      "loss": 1.3426,
+      "step": 4800
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5302011966705322,
+      "learning_rate": 1e-05,
+      "loss": 1.3114,
+      "step": 4810
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.074410915374756,
+      "learning_rate": 1e-05,
+      "loss": 1.3329,
+      "step": 4820
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.883925199508667,
+      "learning_rate": 1e-05,
+      "loss": 1.3415,
+      "step": 4830
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8144758939743042,
+      "learning_rate": 1e-05,
+      "loss": 1.3551,
+      "step": 4840
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2435834407806396,
+      "learning_rate": 1e-05,
+      "loss": 1.2709,
+      "step": 4850
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9630836248397827,
+      "learning_rate": 1e-05,
+      "loss": 1.2654,
+      "step": 4860
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.0202670097351074,
+      "learning_rate": 1e-05,
+      "loss": 1.4137,
+      "step": 4870
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2011775970458984,
+      "learning_rate": 1e-05,
+      "loss": 1.2729,
+      "step": 4880
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7111473083496094,
+      "learning_rate": 1e-05,
+      "loss": 1.3189,
+      "step": 4890
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1530821323394775,
+      "learning_rate": 1e-05,
+      "loss": 1.3891,
+      "step": 4900
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.096242904663086,
+      "learning_rate": 1e-05,
+      "loss": 1.3674,
+      "step": 4910
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7188382148742676,
+      "learning_rate": 1e-05,
+      "loss": 1.3524,
+      "step": 4920
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8607091903686523,
+      "learning_rate": 1e-05,
+      "loss": 1.3462,
+      "step": 4930
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8992630243301392,
+      "learning_rate": 1e-05,
+      "loss": 1.1978,
+      "step": 4940
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.6134096384048462,
+      "learning_rate": 1e-05,
+      "loss": 1.2554,
+      "step": 4950
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.189053773880005,
+      "learning_rate": 1e-05,
+      "loss": 1.3185,
+      "step": 4960
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9997789859771729,
+      "learning_rate": 1e-05,
+      "loss": 1.4556,
+      "step": 4970
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.966931939125061,
+      "learning_rate": 1e-05,
+      "loss": 1.3377,
+      "step": 4980
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.6255134344100952,
+      "learning_rate": 1e-05,
+      "loss": 1.2812,
+      "step": 4990
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.119266986846924,
+      "learning_rate": 1e-05,
+      "loss": 1.3713,
+      "step": 5000
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6048781871795654,
+      "learning_rate": 1e-05,
+      "loss": 1.2766,
+      "step": 5010
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.272893190383911,
+      "learning_rate": 1e-05,
+      "loss": 1.311,
+      "step": 5020
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.620817184448242,
+      "learning_rate": 1e-05,
+      "loss": 1.3302,
+      "step": 5030
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3737807273864746,
+      "learning_rate": 1e-05,
+      "loss": 1.254,
+      "step": 5040
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.072502374649048,
+      "learning_rate": 1e-05,
+      "loss": 1.244,
+      "step": 5050
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.080003261566162,
+      "learning_rate": 1e-05,
+      "loss": 1.187,
+      "step": 5060
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.330199718475342,
+      "learning_rate": 1e-05,
+      "loss": 1.235,
+      "step": 5070
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5456225872039795,
+      "learning_rate": 1e-05,
+      "loss": 1.2679,
+      "step": 5080
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.637150526046753,
+      "learning_rate": 1e-05,
+      "loss": 1.3361,
+      "step": 5090
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.265345811843872,
+      "learning_rate": 1e-05,
+      "loss": 1.3633,
+      "step": 5100
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.577023506164551,
+      "learning_rate": 1e-05,
+      "loss": 1.2505,
+      "step": 5110
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.394627094268799,
+      "learning_rate": 1e-05,
+      "loss": 1.2796,
+      "step": 5120
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3129894733428955,
+      "learning_rate": 1e-05,
+      "loss": 1.2923,
+      "step": 5130
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9741424322128296,
+      "learning_rate": 1e-05,
+      "loss": 1.3085,
+      "step": 5140
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1835744380950928,
+      "learning_rate": 1e-05,
+      "loss": 1.2279,
+      "step": 5150
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0756380558013916,
+      "learning_rate": 1e-05,
+      "loss": 1.2934,
+      "step": 5160
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.079761505126953,
+      "learning_rate": 1e-05,
+      "loss": 1.3547,
+      "step": 5170
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.9494423866271973,
+      "learning_rate": 1e-05,
+      "loss": 1.3259,
+      "step": 5180
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.535632610321045,
+      "learning_rate": 1e-05,
+      "loss": 1.283,
+      "step": 5190
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5230371952056885,
+      "learning_rate": 1e-05,
+      "loss": 1.4404,
+      "step": 5200
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.859532117843628,
+      "learning_rate": 1e-05,
+      "loss": 1.3204,
+      "step": 5210
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2032246589660645,
+      "learning_rate": 1e-05,
+      "loss": 1.303,
+      "step": 5220
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.504431962966919,
+      "learning_rate": 1e-05,
+      "loss": 1.273,
+      "step": 5230
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2306768894195557,
+      "learning_rate": 1e-05,
+      "loss": 1.3615,
+      "step": 5240
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.0646228790283203,
+      "learning_rate": 1e-05,
+      "loss": 1.3246,
+      "step": 5250
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.632603883743286,
+      "learning_rate": 1e-05,
+      "loss": 1.3169,
+      "step": 5260
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8753434419631958,
+      "learning_rate": 1e-05,
+      "loss": 1.4855,
+      "step": 5270
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7019877433776855,
+      "learning_rate": 1e-05,
+      "loss": 1.3734,
+      "step": 5280
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.3574256896972656,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 5290
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2648611068725586,
+      "learning_rate": 1e-05,
+      "loss": 1.3317,
+      "step": 5300
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8861615657806396,
+      "learning_rate": 1e-05,
+      "loss": 1.3584,
+      "step": 5310
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9412791728973389,
+      "learning_rate": 1e-05,
+      "loss": 1.3193,
+      "step": 5320
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.8828357458114624,
+      "learning_rate": 1e-05,
+      "loss": 1.4051,
+      "step": 5330
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7321279048919678,
+      "learning_rate": 1e-05,
+      "loss": 1.2798,
+      "step": 5340
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.8962711095809937,
+      "learning_rate": 1e-05,
+      "loss": 1.2847,
+      "step": 5350
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.476041316986084,
+      "learning_rate": 1e-05,
+      "loss": 1.3349,
+      "step": 5360
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4387452602386475,
+      "learning_rate": 1e-05,
+      "loss": 1.3193,
+      "step": 5370
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.867537021636963,
+      "learning_rate": 1e-05,
+      "loss": 1.3567,
+      "step": 5380
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8666340112686157,
+      "learning_rate": 1e-05,
+      "loss": 1.2492,
+      "step": 5390
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.384167432785034,
+      "learning_rate": 1e-05,
+      "loss": 1.2917,
+      "step": 5400
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2074921131134033,
+      "learning_rate": 1e-05,
+      "loss": 1.3411,
+      "step": 5410
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.630751609802246,
+      "learning_rate": 1e-05,
+      "loss": 1.4372,
+      "step": 5420
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.105861186981201,
+      "learning_rate": 1e-05,
+      "loss": 1.4445,
+      "step": 5430
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2219021320343018,
+      "learning_rate": 1e-05,
+      "loss": 1.3419,
+      "step": 5440
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.8928240537643433,
+      "learning_rate": 1e-05,
+      "loss": 1.3888,
+      "step": 5450
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9438905715942383,
+      "learning_rate": 1e-05,
+      "loss": 1.3179,
+      "step": 5460
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.450190305709839,
+      "learning_rate": 1e-05,
+      "loss": 1.2891,
+      "step": 5470
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.6955147981643677,
+      "learning_rate": 1e-05,
+      "loss": 1.2105,
+      "step": 5480
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.195138931274414,
+      "learning_rate": 1e-05,
+      "loss": 1.3871,
+      "step": 5490
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4984004497528076,
+      "learning_rate": 1e-05,
+      "loss": 1.311,
+      "step": 5500
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8167157173156738,
+      "learning_rate": 1e-05,
+      "loss": 1.2877,
+      "step": 5510
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.051131248474121,
+      "learning_rate": 1e-05,
+      "loss": 1.3146,
+      "step": 5520
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.0520663261413574,
+      "learning_rate": 1e-05,
+      "loss": 1.3981,
+      "step": 5530
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.21244478225708,
+      "learning_rate": 1e-05,
+      "loss": 1.3601,
+      "step": 5540
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.481689453125,
+      "learning_rate": 1e-05,
+      "loss": 1.3433,
+      "step": 5550
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.1406707763671875,
+      "learning_rate": 1e-05,
+      "loss": 1.353,
+      "step": 5560
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8597412109375,
+      "learning_rate": 1e-05,
+      "loss": 1.3786,
+      "step": 5570
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.6117429733276367,
+      "learning_rate": 1e-05,
+      "loss": 1.2154,
+      "step": 5580
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9745125770568848,
+      "learning_rate": 1e-05,
+      "loss": 1.2937,
+      "step": 5590
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9402446746826172,
+      "learning_rate": 1e-05,
+      "loss": 1.3025,
+      "step": 5600
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.1005172729492188,
+      "learning_rate": 1e-05,
+      "loss": 1.2995,
+      "step": 5610
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.382460117340088,
+      "learning_rate": 1e-05,
+      "loss": 1.3255,
+      "step": 5620
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3474137783050537,
+      "learning_rate": 1e-05,
+      "loss": 1.327,
+      "step": 5630
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1676485538482666,
+      "learning_rate": 1e-05,
+      "loss": 1.3022,
+      "step": 5640
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.638181686401367,
+      "learning_rate": 1e-05,
+      "loss": 1.3579,
+      "step": 5650
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.8829666376113892,
+      "learning_rate": 1e-05,
+      "loss": 1.2526,
+      "step": 5660
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.283449649810791,
+      "learning_rate": 1e-05,
+      "loss": 1.2292,
+      "step": 5670
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7156941890716553,
+      "learning_rate": 1e-05,
+      "loss": 1.3473,
+      "step": 5680
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.614403486251831,
+      "learning_rate": 1e-05,
+      "loss": 1.2649,
+      "step": 5690
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.518841505050659,
+      "learning_rate": 1e-05,
+      "loss": 1.4286,
+      "step": 5700
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.694382429122925,
+      "learning_rate": 1e-05,
+      "loss": 1.3353,
+      "step": 5710
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2780184745788574,
+      "learning_rate": 1e-05,
+      "loss": 1.279,
+      "step": 5720
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.911449670791626,
+      "learning_rate": 1e-05,
+      "loss": 1.3446,
+      "step": 5730
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.7798064947128296,
+      "learning_rate": 1e-05,
+      "loss": 1.2469,
+      "step": 5740
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5752665996551514,
+      "learning_rate": 1e-05,
+      "loss": 1.2476,
+      "step": 5750
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1589324474334717,
+      "learning_rate": 1e-05,
+      "loss": 1.3267,
+      "step": 5760
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.644857168197632,
+      "learning_rate": 1e-05,
+      "loss": 1.4403,
+      "step": 5770
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.159285545349121,
+      "learning_rate": 1e-05,
+      "loss": 1.2866,
+      "step": 5780
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4415762424468994,
+      "learning_rate": 1e-05,
+      "loss": 1.2472,
+      "step": 5790
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.266463279724121,
+      "learning_rate": 1e-05,
+      "loss": 1.3598,
+      "step": 5800
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.450885534286499,
+      "learning_rate": 1e-05,
+      "loss": 1.2823,
+      "step": 5810
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.997843861579895,
+      "learning_rate": 1e-05,
+      "loss": 1.3197,
+      "step": 5820
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.7925161123275757,
+      "eval_runtime": 21.4712,
+      "eval_samples_per_second": 46.574,
+      "eval_steps_per_second": 46.574,
+      "step": 5823
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5618516206741333,
+      "learning_rate": 1e-05,
+      "loss": 1.1163,
+      "step": 5830
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.7338979244232178,
+      "learning_rate": 1e-05,
+      "loss": 0.9587,
+      "step": 5840
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.972121000289917,
+      "learning_rate": 1e-05,
+      "loss": 1.0549,
+      "step": 5850
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.1814982891082764,
+      "learning_rate": 1e-05,
+      "loss": 0.9983,
+      "step": 5860
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.32295560836792,
+      "learning_rate": 1e-05,
+      "loss": 0.9349,
+      "step": 5870
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.3985841274261475,
+      "learning_rate": 1e-05,
+      "loss": 0.9369,
+      "step": 5880
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.472193717956543,
+      "learning_rate": 1e-05,
+      "loss": 1.1074,
+      "step": 5890
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.7933571338653564,
+      "learning_rate": 1e-05,
+      "loss": 1.0088,
+      "step": 5900
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.421544075012207,
+      "learning_rate": 1e-05,
+      "loss": 1.0128,
+      "step": 5910
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.9460196495056152,
+      "learning_rate": 1e-05,
+      "loss": 0.9706,
+      "step": 5920
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.7304258346557617,
+      "learning_rate": 1e-05,
+      "loss": 0.9553,
+      "step": 5930
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.0735466480255127,
+      "learning_rate": 1e-05,
+      "loss": 1.0339,
+      "step": 5940
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.4456515312194824,
+      "learning_rate": 1e-05,
+      "loss": 0.9472,
+      "step": 5950
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.9821925163269043,
+      "learning_rate": 1e-05,
+      "loss": 0.9725,
+      "step": 5960
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.6493866443634033,
+      "learning_rate": 1e-05,
+      "loss": 1.0446,
+      "step": 5970
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.323594093322754,
+      "learning_rate": 1e-05,
+      "loss": 0.9804,
+      "step": 5980
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.6612460613250732,
+      "learning_rate": 1e-05,
+      "loss": 1.0335,
+      "step": 5990
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.080219030380249,
+      "learning_rate": 1e-05,
+      "loss": 0.9365,
+      "step": 6000
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.0296177864074707,
+      "learning_rate": 1e-05,
+      "loss": 0.9439,
+      "step": 6010
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.418188095092773,
+      "learning_rate": 1e-05,
+      "loss": 1.0635,
+      "step": 6020
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.9283742904663086,
+      "learning_rate": 1e-05,
+      "loss": 0.9956,
+      "step": 6030
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.918304681777954,
+      "learning_rate": 1e-05,
+      "loss": 1.0152,
+      "step": 6040
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.512454032897949,
+      "learning_rate": 1e-05,
+      "loss": 1.0259,
+      "step": 6050
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.7206339836120605,
+      "learning_rate": 1e-05,
+      "loss": 0.9879,
+      "step": 6060
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.2170846462249756,
+      "learning_rate": 1e-05,
+      "loss": 0.9946,
+      "step": 6070
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.6726131439208984,
+      "learning_rate": 1e-05,
+      "loss": 0.9833,
+      "step": 6080
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.8087282180786133,
+      "learning_rate": 1e-05,
+      "loss": 1.0294,
+      "step": 6090
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.9470926523208618,
+      "learning_rate": 1e-05,
+      "loss": 1.0807,
+      "step": 6100
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.086599588394165,
+      "learning_rate": 1e-05,
+      "loss": 0.9602,
+      "step": 6110
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.537266254425049,
+      "learning_rate": 1e-05,
+      "loss": 0.9963,
+      "step": 6120
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.6490066051483154,
+      "learning_rate": 1e-05,
+      "loss": 1.0077,
+      "step": 6130
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.294142723083496,
+      "learning_rate": 1e-05,
+      "loss": 1.0145,
+      "step": 6140
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.8552401065826416,
+      "learning_rate": 1e-05,
+      "loss": 0.8509,
+      "step": 6150
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.9774388074874878,
+      "learning_rate": 1e-05,
+      "loss": 0.9475,
+      "step": 6160
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.0472378730773926,
+      "learning_rate": 1e-05,
+      "loss": 1.0339,
+      "step": 6170
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.7465121746063232,
+      "learning_rate": 1e-05,
+      "loss": 1.0642,
+      "step": 6180
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.965936541557312,
+      "learning_rate": 1e-05,
+      "loss": 0.9558,
+      "step": 6190
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.1949400901794434,
+      "learning_rate": 1e-05,
+      "loss": 0.9357,
+      "step": 6200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.121185302734375,
+      "learning_rate": 1e-05,
+      "loss": 1.0069,
+      "step": 6210
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.357470750808716,
+      "learning_rate": 1e-05,
+      "loss": 0.9,
+      "step": 6220
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.186276435852051,
+      "learning_rate": 1e-05,
+      "loss": 0.952,
+      "step": 6230
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.1679868698120117,
+      "learning_rate": 1e-05,
+      "loss": 0.9949,
+      "step": 6240
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.6253249645233154,
+      "learning_rate": 1e-05,
+      "loss": 0.995,
+      "step": 6250
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.671788454055786,
+      "learning_rate": 1e-05,
+      "loss": 1.0111,
+      "step": 6260
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 3.011953592300415,
+      "learning_rate": 1e-05,
+      "loss": 1.0067,
+      "step": 6270
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.894116163253784,
+      "learning_rate": 1e-05,
+      "loss": 1.0278,
+      "step": 6280
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.782424807548523,
+      "learning_rate": 1e-05,
+      "loss": 0.9849,
+      "step": 6290
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.0209150314331055,
+      "learning_rate": 1e-05,
+      "loss": 1.0661,
+      "step": 6300
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.0378525257110596,
+      "learning_rate": 1e-05,
+      "loss": 1.0675,
+      "step": 6310
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.555785655975342,
+      "learning_rate": 1e-05,
+      "loss": 1.0687,
+      "step": 6320
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.312987804412842,
+      "learning_rate": 1e-05,
+      "loss": 0.9451,
+      "step": 6330
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.7663776874542236,
+      "learning_rate": 1e-05,
+      "loss": 1.0758,
+      "step": 6340
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.143360137939453,
+      "learning_rate": 1e-05,
+      "loss": 0.9993,
+      "step": 6350
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.927686929702759,
+      "learning_rate": 1e-05,
+      "loss": 1.0186,
+      "step": 6360
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.797344207763672,
+      "learning_rate": 1e-05,
+      "loss": 1.0245,
+      "step": 6370
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 3.264024496078491,
+      "learning_rate": 1e-05,
+      "loss": 1.1007,
+      "step": 6380
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.4195101261138916,
+      "learning_rate": 1e-05,
+      "loss": 0.9724,
+      "step": 6390
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.8212947845458984,
+      "learning_rate": 1e-05,
+      "loss": 1.0232,
+      "step": 6400
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.9952080249786377,
+      "learning_rate": 1e-05,
+      "loss": 1.0124,
+      "step": 6410
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 3.3921620845794678,
+      "learning_rate": 1e-05,
+      "loss": 1.0581,
+      "step": 6420
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.338721752166748,
+      "learning_rate": 1e-05,
+      "loss": 0.9655,
+      "step": 6430
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.622527599334717,
+      "learning_rate": 1e-05,
+      "loss": 1.0654,
+      "step": 6440
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.8540849685668945,
+      "learning_rate": 1e-05,
+      "loss": 0.9467,
+      "step": 6450
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.3519699573516846,
+      "learning_rate": 1e-05,
+      "loss": 0.989,
+      "step": 6460
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.0406546592712402,
+      "learning_rate": 1e-05,
+      "loss": 0.9715,
+      "step": 6470
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.2899045944213867,
+      "learning_rate": 1e-05,
+      "loss": 0.9661,
+      "step": 6480
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.9575726985931396,
+      "learning_rate": 1e-05,
+      "loss": 1.0636,
+      "step": 6490
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.0779426097869873,
+      "learning_rate": 1e-05,
+      "loss": 1.0781,
+      "step": 6500
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.743601083755493,
+      "learning_rate": 1e-05,
+      "loss": 1.0193,
+      "step": 6510
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.9085830450057983,
+      "learning_rate": 1e-05,
+      "loss": 0.9416,
+      "step": 6520
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.8828158378601074,
+      "learning_rate": 1e-05,
+      "loss": 1.0238,
+      "step": 6530
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.3883135318756104,
+      "learning_rate": 1e-05,
+      "loss": 1.0709,
+      "step": 6540
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.5728261470794678,
+      "learning_rate": 1e-05,
+      "loss": 0.9783,
+      "step": 6550
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.7872670888900757,
+      "learning_rate": 1e-05,
+      "loss": 1.0118,
+      "step": 6560
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 3.844386100769043,
+      "learning_rate": 1e-05,
+      "loss": 1.062,
+      "step": 6570
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.2257680892944336,
+      "learning_rate": 1e-05,
+      "loss": 0.9837,
+      "step": 6580
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.2637107372283936,
+      "learning_rate": 1e-05,
+      "loss": 1.0532,
+      "step": 6590
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.8380672931671143,
+      "learning_rate": 1e-05,
+      "loss": 0.9508,
+      "step": 6600
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.0793893337249756,
+      "learning_rate": 1e-05,
+      "loss": 0.9432,
+      "step": 6610
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.721628427505493,
+      "learning_rate": 1e-05,
+      "loss": 1.0368,
+      "step": 6620
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.677793025970459,
+      "learning_rate": 1e-05,
+      "loss": 1.0157,
+      "step": 6630
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.3213117122650146,
+      "learning_rate": 1e-05,
+      "loss": 0.9747,
+      "step": 6640
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.401367425918579,
+      "learning_rate": 1e-05,
+      "loss": 1.0084,
+      "step": 6650
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.3696703910827637,
+      "learning_rate": 1e-05,
+      "loss": 1.0004,
+      "step": 6660
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.552927255630493,
+      "learning_rate": 1e-05,
+      "loss": 1.1054,
+      "step": 6670
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.5474822521209717,
+      "learning_rate": 1e-05,
+      "loss": 1.0448,
+      "step": 6680
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.3490467071533203,
+      "learning_rate": 1e-05,
+      "loss": 1.0061,
+      "step": 6690
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.2008440494537354,
+      "learning_rate": 1e-05,
+      "loss": 1.0262,
+      "step": 6700
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.552922487258911,
+      "learning_rate": 1e-05,
+      "loss": 1.0031,
+      "step": 6710
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.1921865940093994,
+      "learning_rate": 1e-05,
+      "loss": 1.041,
+      "step": 6720
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.6241724491119385,
+      "learning_rate": 1e-05,
+      "loss": 1.0084,
+      "step": 6730
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.2282745838165283,
+      "learning_rate": 1e-05,
+      "loss": 0.8996,
+      "step": 6740
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.343564987182617,
+      "learning_rate": 1e-05,
+      "loss": 1.0102,
+      "step": 6750
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.3204636573791504,
+      "learning_rate": 1e-05,
+      "loss": 0.9502,
+      "step": 6760
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.019423484802246,
+      "learning_rate": 1e-05,
+      "loss": 0.9669,
+      "step": 6770
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.683650493621826,
+      "learning_rate": 1e-05,
+      "loss": 0.9615,
+      "step": 6780
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.899923086166382,
+      "learning_rate": 1e-05,
+      "loss": 0.9532,
+      "step": 6790
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.707458019256592,
+      "learning_rate": 1e-05,
+      "loss": 1.0163,
+      "step": 6800
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.8542733192443848,
+      "learning_rate": 1e-05,
+      "loss": 1.0348,
+      "step": 6810
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.4394218921661377,
+      "learning_rate": 1e-05,
+      "loss": 1.1294,
+      "step": 6820
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.711557149887085,
+      "learning_rate": 1e-05,
+      "loss": 0.9588,
+      "step": 6830
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.206164836883545,
+      "learning_rate": 1e-05,
+      "loss": 1.0459,
+      "step": 6840
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.1192731857299805,
+      "learning_rate": 1e-05,
+      "loss": 1.1014,
+      "step": 6850
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.472332000732422,
+      "learning_rate": 1e-05,
+      "loss": 1.0396,
+      "step": 6860
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.860532522201538,
+      "learning_rate": 1e-05,
+      "loss": 1.0652,
+      "step": 6870
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.9870842695236206,
+      "learning_rate": 1e-05,
+      "loss": 0.941,
+      "step": 6880
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.527759552001953,
+      "learning_rate": 1e-05,
+      "loss": 1.0487,
+      "step": 6890
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.2542383670806885,
+      "learning_rate": 1e-05,
+      "loss": 1.1416,
+      "step": 6900
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.2846920490264893,
+      "learning_rate": 1e-05,
+      "loss": 1.0324,
+      "step": 6910
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.865117073059082,
+      "learning_rate": 1e-05,
+      "loss": 1.0439,
+      "step": 6920
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.6450443267822266,
+      "learning_rate": 1e-05,
+      "loss": 0.9566,
+      "step": 6930
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.0172007083892822,
+      "learning_rate": 1e-05,
+      "loss": 1.0769,
+      "step": 6940
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 4.062388896942139,
+      "learning_rate": 1e-05,
+      "loss": 1.0179,
+      "step": 6950
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.4772136211395264,
+      "learning_rate": 1e-05,
+      "loss": 0.9977,
+      "step": 6960
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.269806385040283,
+      "learning_rate": 1e-05,
+      "loss": 0.9665,
+      "step": 6970
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.0248801708221436,
+      "learning_rate": 1e-05,
+      "loss": 1.0624,
+      "step": 6980
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.0228490829467773,
+      "learning_rate": 1e-05,
+      "loss": 1.0126,
+      "step": 6990
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.6162188053131104,
+      "learning_rate": 1e-05,
+      "loss": 0.9477,
+      "step": 7000
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.0334103107452393,
+      "learning_rate": 1e-05,
+      "loss": 1.0403,
+      "step": 7010
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.4257733821868896,
+      "learning_rate": 1e-05,
+      "loss": 0.9266,
+      "step": 7020
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.222496509552002,
+      "learning_rate": 1e-05,
+      "loss": 0.9598,
+      "step": 7030
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.3846302032470703,
+      "learning_rate": 1e-05,
+      "loss": 1.0264,
+      "step": 7040
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.849187135696411,
+      "learning_rate": 1e-05,
+      "loss": 1.0838,
+      "step": 7050
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.097158432006836,
+      "learning_rate": 1e-05,
+      "loss": 0.9943,
+      "step": 7060
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.259446859359741,
+      "learning_rate": 1e-05,
+      "loss": 1.0005,
+      "step": 7070
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.8879165649414062,
+      "learning_rate": 1e-05,
+      "loss": 1.0426,
+      "step": 7080
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.0662331581115723,
+      "learning_rate": 1e-05,
+      "loss": 1.0646,
+      "step": 7090
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.568582057952881,
+      "learning_rate": 1e-05,
+      "loss": 1.0497,
+      "step": 7100
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.3837697505950928,
+      "learning_rate": 1e-05,
+      "loss": 1.0589,
+      "step": 7110
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.908559560775757,
+      "learning_rate": 1e-05,
+      "loss": 0.9937,
+      "step": 7120
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.302889823913574,
+      "learning_rate": 1e-05,
+      "loss": 1.042,
+      "step": 7130
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.683570146560669,
+      "learning_rate": 1e-05,
+      "loss": 1.1151,
+      "step": 7140
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.5803396701812744,
+      "learning_rate": 1e-05,
+      "loss": 1.0203,
+      "step": 7150
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.4933981895446777,
+      "learning_rate": 1e-05,
+      "loss": 1.0585,
+      "step": 7160
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.319493055343628,
+      "learning_rate": 1e-05,
+      "loss": 1.0397,
+      "step": 7170
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.282255172729492,
+      "learning_rate": 1e-05,
+      "loss": 1.0919,
+      "step": 7180
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.5834693908691406,
+      "learning_rate": 1e-05,
+      "loss": 0.9382,
+      "step": 7190
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.854360818862915,
+      "learning_rate": 1e-05,
+      "loss": 1.0088,
+      "step": 7200
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.8476474285125732,
+      "learning_rate": 1e-05,
+      "loss": 0.9774,
+      "step": 7210
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.5642004013061523,
+      "learning_rate": 1e-05,
+      "loss": 1.0274,
+      "step": 7220
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.215512990951538,
+      "learning_rate": 1e-05,
+      "loss": 1.073,
+      "step": 7230
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 3.149976968765259,
+      "learning_rate": 1e-05,
+      "loss": 1.0018,
+      "step": 7240
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.221637010574341,
+      "learning_rate": 1e-05,
+      "loss": 0.9686,
+      "step": 7250
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.236781358718872,
+      "learning_rate": 1e-05,
+      "loss": 1.027,
+      "step": 7260
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.7379977703094482,
+      "learning_rate": 1e-05,
+      "loss": 0.9815,
+      "step": 7270
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.726146697998047,
+      "learning_rate": 1e-05,
+      "loss": 0.9777,
+      "step": 7280
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.037290573120117,
+      "learning_rate": 1e-05,
+      "loss": 0.9795,
+      "step": 7290
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.7081196308135986,
+      "learning_rate": 1e-05,
+      "loss": 1.0047,
+      "step": 7300
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.575348138809204,
+      "learning_rate": 1e-05,
+      "loss": 0.9683,
+      "step": 7310
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.6672821044921875,
+      "learning_rate": 1e-05,
+      "loss": 1.0013,
+      "step": 7320
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.878222703933716,
+      "learning_rate": 1e-05,
+      "loss": 1.0415,
+      "step": 7330
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.99146568775177,
+      "learning_rate": 1e-05,
+      "loss": 1.0387,
+      "step": 7340
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.957350492477417,
+      "learning_rate": 1e-05,
+      "loss": 1.0126,
+      "step": 7350
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.2003567218780518,
+      "learning_rate": 1e-05,
+      "loss": 0.888,
+      "step": 7360
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.5942909717559814,
+      "learning_rate": 1e-05,
+      "loss": 1.0561,
+      "step": 7370
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.9146692752838135,
+      "learning_rate": 1e-05,
+      "loss": 1.018,
+      "step": 7380
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.8458762168884277,
+      "learning_rate": 1e-05,
+      "loss": 1.0827,
+      "step": 7390
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.8542159795761108,
+      "learning_rate": 1e-05,
+      "loss": 1.0302,
+      "step": 7400
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.5300040245056152,
+      "learning_rate": 1e-05,
+      "loss": 1.0084,
+      "step": 7410
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.089399576187134,
+      "learning_rate": 1e-05,
+      "loss": 0.9713,
+      "step": 7420
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.373389482498169,
+      "learning_rate": 1e-05,
+      "loss": 1.0608,
+      "step": 7430
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.3854312896728516,
+      "learning_rate": 1e-05,
+      "loss": 1.0177,
+      "step": 7440
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.3859362602233887,
+      "learning_rate": 1e-05,
+      "loss": 0.9844,
+      "step": 7450
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.767002820968628,
+      "learning_rate": 1e-05,
+      "loss": 1.0356,
+      "step": 7460
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.1323554515838623,
+      "learning_rate": 1e-05,
+      "loss": 0.9526,
+      "step": 7470
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.690873622894287,
+      "learning_rate": 1e-05,
+      "loss": 0.9732,
+      "step": 7480
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 3.478398561477661,
+      "learning_rate": 1e-05,
+      "loss": 0.9905,
+      "step": 7490
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.8604772090911865,
+      "learning_rate": 1e-05,
+      "loss": 0.9636,
+      "step": 7500
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.680375576019287,
+      "learning_rate": 1e-05,
+      "loss": 0.9631,
+      "step": 7510
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 3.15445613861084,
+      "learning_rate": 1e-05,
+      "loss": 1.0039,
+      "step": 7520
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 3.388442039489746,
+      "learning_rate": 1e-05,
+      "loss": 0.9484,
+      "step": 7530
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.1244375705718994,
+      "learning_rate": 1e-05,
+      "loss": 0.9882,
+      "step": 7540
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.9992311000823975,
+      "learning_rate": 1e-05,
+      "loss": 0.9835,
+      "step": 7550
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.6015098094940186,
+      "learning_rate": 1e-05,
+      "loss": 1.0097,
+      "step": 7560
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.9045817852020264,
+      "learning_rate": 1e-05,
+      "loss": 0.9762,
+      "step": 7570
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.8165290355682373,
+      "learning_rate": 1e-05,
+      "loss": 0.9632,
+      "step": 7580
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.154522657394409,
+      "learning_rate": 1e-05,
+      "loss": 1.0249,
+      "step": 7590
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.180574893951416,
+      "learning_rate": 1e-05,
+      "loss": 1.0563,
+      "step": 7600
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 4.546901226043701,
+      "learning_rate": 1e-05,
+      "loss": 1.0525,
+      "step": 7610
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.081376552581787,
+      "learning_rate": 1e-05,
+      "loss": 0.9529,
+      "step": 7620
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.6588551998138428,
+      "learning_rate": 1e-05,
+      "loss": 1.0004,
+      "step": 7630
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.580198049545288,
+      "learning_rate": 1e-05,
+      "loss": 1.0921,
+      "step": 7640
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.5459742546081543,
+      "learning_rate": 1e-05,
+      "loss": 0.9399,
+      "step": 7650
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.4501724243164062,
+      "learning_rate": 1e-05,
+      "loss": 1.0721,
+      "step": 7660
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.613929033279419,
+      "learning_rate": 1e-05,
+      "loss": 1.0456,
+      "step": 7670
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.8623414039611816,
+      "learning_rate": 1e-05,
+      "loss": 0.9783,
+      "step": 7680
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.3043484687805176,
+      "learning_rate": 1e-05,
+      "loss": 1.0286,
+      "step": 7690
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.5169615745544434,
+      "learning_rate": 1e-05,
+      "loss": 1.0884,
+      "step": 7700
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.3486950397491455,
+      "learning_rate": 1e-05,
+      "loss": 0.9957,
+      "step": 7710
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.348045587539673,
+      "learning_rate": 1e-05,
+      "loss": 0.9804,
+      "step": 7720
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.1623213291168213,
+      "learning_rate": 1e-05,
+      "loss": 1.0494,
+      "step": 7730
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.6808427572250366,
+      "learning_rate": 1e-05,
+      "loss": 1.0663,
+      "step": 7740
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.15962290763855,
+      "learning_rate": 1e-05,
+      "loss": 1.0142,
+      "step": 7750
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.7483596801757812,
+      "learning_rate": 1e-05,
+      "loss": 1.0353,
+      "step": 7760
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.0963194370269775,
+      "learning_rate": 1e-05,
+      "loss": 1.0867,
+      "step": 7770
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.584580898284912,
+      "learning_rate": 1e-05,
+      "loss": 1.0744,
+      "step": 7780
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.4675183296203613,
+      "learning_rate": 1e-05,
+      "loss": 1.0784,
+      "step": 7790
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.236694097518921,
+      "learning_rate": 1e-05,
+      "loss": 1.0281,
+      "step": 7800
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 5.138028621673584,
+      "learning_rate": 1e-05,
+      "loss": 1.0198,
+      "step": 7810
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.8963077068328857,
+      "learning_rate": 1e-05,
+      "loss": 0.9776,
+      "step": 7820
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.836012125015259,
+      "learning_rate": 1e-05,
+      "loss": 1.0364,
+      "step": 7830
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.667417287826538,
+      "learning_rate": 1e-05,
+      "loss": 1.013,
+      "step": 7840
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.6331534385681152,
+      "learning_rate": 1e-05,
+      "loss": 1.0265,
+      "step": 7850
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.822760820388794,
+      "learning_rate": 1e-05,
+      "loss": 1.078,
+      "step": 7860
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.772062301635742,
+      "learning_rate": 1e-05,
+      "loss": 0.9334,
+      "step": 7870
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.1583237648010254,
+      "learning_rate": 1e-05,
+      "loss": 0.8797,
+      "step": 7880
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.0897679328918457,
+      "learning_rate": 1e-05,
+      "loss": 0.9849,
+      "step": 7890
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.45992112159729,
+      "learning_rate": 1e-05,
+      "loss": 1.0646,
+      "step": 7900
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.1727113723754883,
+      "learning_rate": 1e-05,
+      "loss": 1.0321,
+      "step": 7910
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.950498580932617,
+      "learning_rate": 1e-05,
+      "loss": 1.0255,
+      "step": 7920
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.294767141342163,
+      "learning_rate": 1e-05,
+      "loss": 0.989,
+      "step": 7930
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.484980583190918,
+      "learning_rate": 1e-05,
+      "loss": 0.9724,
+      "step": 7940
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.3656985759735107,
+      "learning_rate": 1e-05,
+      "loss": 1.0837,
+      "step": 7950
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.5616161823272705,
+      "learning_rate": 1e-05,
+      "loss": 1.0084,
+      "step": 7960
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.1650032997131348,
+      "learning_rate": 1e-05,
+      "loss": 1.1708,
+      "step": 7970
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.4175758361816406,
+      "learning_rate": 1e-05,
+      "loss": 1.0317,
+      "step": 7980
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.108388662338257,
+      "learning_rate": 1e-05,
+      "loss": 1.0534,
+      "step": 7990
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.224473714828491,
+      "learning_rate": 1e-05,
+      "loss": 0.9725,
+      "step": 8000
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.2060694694519043,
+      "learning_rate": 1e-05,
+      "loss": 1.004,
+      "step": 8010
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.9540616273880005,
+      "learning_rate": 1e-05,
+      "loss": 0.8094,
+      "step": 8020
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.420384645462036,
+      "learning_rate": 1e-05,
+      "loss": 1.042,
+      "step": 8030
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.2382652759552,
+      "learning_rate": 1e-05,
+      "loss": 1.0907,
+      "step": 8040
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.1269848346710205,
+      "learning_rate": 1e-05,
+      "loss": 1.0609,
+      "step": 8050
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.7086892127990723,
+      "learning_rate": 1e-05,
+      "loss": 0.9583,
+      "step": 8060
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.5433905124664307,
+      "learning_rate": 1e-05,
+      "loss": 0.9998,
+      "step": 8070
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.4779505729675293,
+      "learning_rate": 1e-05,
+      "loss": 1.0018,
+      "step": 8080
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.4827795028686523,
+      "learning_rate": 1e-05,
+      "loss": 1.0067,
+      "step": 8090
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.616666793823242,
+      "learning_rate": 1e-05,
+      "loss": 1.0181,
+      "step": 8100
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.293524980545044,
+      "learning_rate": 1e-05,
+      "loss": 1.033,
+      "step": 8110
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.3285727500915527,
+      "learning_rate": 1e-05,
+      "loss": 0.9987,
+      "step": 8120
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.2849738597869873,
+      "learning_rate": 1e-05,
+      "loss": 1.014,
+      "step": 8130
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.6285767555236816,
+      "learning_rate": 1e-05,
+      "loss": 1.0186,
+      "step": 8140
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.7985305786132812,
+      "learning_rate": 1e-05,
+      "loss": 1.1144,
+      "step": 8150
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.4430348873138428,
+      "learning_rate": 1e-05,
+      "loss": 1.0596,
+      "step": 8160
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.200226306915283,
+      "learning_rate": 1e-05,
+      "loss": 1.0944,
+      "step": 8170
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.8424794673919678,
+      "learning_rate": 1e-05,
+      "loss": 1.1051,
+      "step": 8180
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.2834882736206055,
+      "learning_rate": 1e-05,
+      "loss": 0.9787,
+      "step": 8190
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.308248519897461,
+      "learning_rate": 1e-05,
+      "loss": 1.0679,
+      "step": 8200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.680285692214966,
+      "learning_rate": 1e-05,
+      "loss": 0.996,
+      "step": 8210
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.7210543155670166,
+      "learning_rate": 1e-05,
+      "loss": 0.9188,
+      "step": 8220
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.5917136669158936,
+      "learning_rate": 1e-05,
+      "loss": 1.0204,
+      "step": 8230
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.6894149780273438,
+      "learning_rate": 1e-05,
+      "loss": 0.9765,
+      "step": 8240
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.9170303344726562,
+      "learning_rate": 1e-05,
+      "loss": 1.0352,
+      "step": 8250
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.6600093841552734,
+      "learning_rate": 1e-05,
+      "loss": 0.9624,
+      "step": 8260
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.1691620349884033,
+      "learning_rate": 1e-05,
+      "loss": 1.0014,
+      "step": 8270
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.4668002128601074,
+      "learning_rate": 1e-05,
+      "loss": 1.1043,
+      "step": 8280
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.1661458015441895,
+      "learning_rate": 1e-05,
+      "loss": 1.0113,
+      "step": 8290
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.3365416526794434,
+      "learning_rate": 1e-05,
+      "loss": 1.0482,
+      "step": 8300
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 3.319361925125122,
+      "learning_rate": 1e-05,
+      "loss": 1.1109,
+      "step": 8310
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.097994089126587,
+      "learning_rate": 1e-05,
+      "loss": 0.9544,
+      "step": 8320
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.9699785709381104,
+      "learning_rate": 1e-05,
+      "loss": 1.026,
+      "step": 8330
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.399681568145752,
+      "learning_rate": 1e-05,
+      "loss": 1.0759,
+      "step": 8340
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.4438891410827637,
+      "learning_rate": 1e-05,
+      "loss": 1.0336,
+      "step": 8350
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.4506747722625732,
+      "learning_rate": 1e-05,
+      "loss": 1.0359,
+      "step": 8360
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 3.281940221786499,
+      "learning_rate": 1e-05,
+      "loss": 1.1251,
+      "step": 8370
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.040092706680298,
+      "learning_rate": 1e-05,
+      "loss": 1.0515,
+      "step": 8380
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.184382677078247,
+      "learning_rate": 1e-05,
+      "loss": 0.9284,
+      "step": 8390
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.638476848602295,
+      "learning_rate": 1e-05,
+      "loss": 1.0476,
+      "step": 8400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.5259788036346436,
+      "learning_rate": 1e-05,
+      "loss": 1.0159,
+      "step": 8410
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.3715500831604004,
+      "learning_rate": 1e-05,
+      "loss": 1.016,
+      "step": 8420
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.3363490104675293,
+      "learning_rate": 1e-05,
+      "loss": 0.9059,
+      "step": 8430
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 3.1245510578155518,
+      "learning_rate": 1e-05,
+      "loss": 1.0613,
+      "step": 8440
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.519963502883911,
+      "learning_rate": 1e-05,
+      "loss": 1.0959,
+      "step": 8450
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 3.118851900100708,
+      "learning_rate": 1e-05,
+      "loss": 1.0572,
+      "step": 8460
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.3911423683166504,
+      "learning_rate": 1e-05,
+      "loss": 1.0629,
+      "step": 8470
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.0603713989257812,
+      "learning_rate": 1e-05,
+      "loss": 1.097,
+      "step": 8480
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.1234920024871826,
+      "learning_rate": 1e-05,
+      "loss": 1.0664,
+      "step": 8490
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.358424186706543,
+      "learning_rate": 1e-05,
+      "loss": 1.0644,
+      "step": 8500
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.3433587551116943,
+      "learning_rate": 1e-05,
+      "loss": 1.0608,
+      "step": 8510
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.0287795066833496,
+      "learning_rate": 1e-05,
+      "loss": 1.0,
+      "step": 8520
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.1926498413085938,
+      "learning_rate": 1e-05,
+      "loss": 0.9871,
+      "step": 8530
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.370448589324951,
+      "learning_rate": 1e-05,
+      "loss": 1.0169,
+      "step": 8540
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 3.0587711334228516,
+      "learning_rate": 1e-05,
+      "loss": 1.1135,
+      "step": 8550
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.773609161376953,
+      "learning_rate": 1e-05,
+      "loss": 1.0053,
+      "step": 8560
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.279136896133423,
+      "learning_rate": 1e-05,
+      "loss": 1.0227,
+      "step": 8570
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.549975633621216,
+      "learning_rate": 1e-05,
+      "loss": 0.9351,
+      "step": 8580
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.1325323581695557,
+      "learning_rate": 1e-05,
+      "loss": 0.9615,
+      "step": 8590
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.074357748031616,
+      "learning_rate": 1e-05,
+      "loss": 1.0796,
+      "step": 8600
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.1816437244415283,
+      "learning_rate": 1e-05,
+      "loss": 1.0263,
+      "step": 8610
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.568363904953003,
+      "learning_rate": 1e-05,
+      "loss": 1.1019,
+      "step": 8620
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.1658637523651123,
+      "learning_rate": 1e-05,
+      "loss": 1.06,
+      "step": 8630
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.6504440307617188,
+      "learning_rate": 1e-05,
+      "loss": 1.0738,
+      "step": 8640
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.858308792114258,
+      "learning_rate": 1e-05,
+      "loss": 1.0557,
+      "step": 8650
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.064356803894043,
+      "learning_rate": 1e-05,
+      "loss": 1.011,
+      "step": 8660
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.340825080871582,
+      "learning_rate": 1e-05,
+      "loss": 1.0774,
+      "step": 8670
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.6675853729248047,
+      "learning_rate": 1e-05,
+      "loss": 1.1196,
+      "step": 8680
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.820267677307129,
+      "learning_rate": 1e-05,
+      "loss": 0.8902,
+      "step": 8690
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.1053669452667236,
+      "learning_rate": 1e-05,
+      "loss": 0.993,
+      "step": 8700
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.9717823266983032,
+      "learning_rate": 1e-05,
+      "loss": 1.0134,
+      "step": 8710
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.224724292755127,
+      "learning_rate": 1e-05,
+      "loss": 0.9832,
+      "step": 8720
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.688229560852051,
+      "learning_rate": 1e-05,
+      "loss": 1.099,
+      "step": 8730
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.633141279220581,
+      "learning_rate": 1e-05,
+      "loss": 0.999,
+      "step": 8740
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 3.245795726776123,
+      "learning_rate": 1e-05,
+      "loss": 0.97,
+      "step": 8750
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 4.065086364746094,
+      "learning_rate": 1e-05,
+      "loss": 1.1426,
+      "step": 8760
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.8590381145477295,
+      "learning_rate": 1e-05,
+      "loss": 1.0187,
+      "step": 8770
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.5020790100097656,
+      "learning_rate": 1e-05,
+      "loss": 0.9483,
+      "step": 8780
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.2966628074645996,
+      "learning_rate": 1e-05,
+      "loss": 0.9955,
+      "step": 8790
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.021965980529785,
+      "learning_rate": 1e-05,
+      "loss": 1.1033,
+      "step": 8800
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.4566996097564697,
+      "learning_rate": 1e-05,
+      "loss": 1.0353,
+      "step": 8810
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.8194572925567627,
+      "learning_rate": 1e-05,
+      "loss": 1.0071,
+      "step": 8820
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.5652499198913574,
+      "learning_rate": 1e-05,
+      "loss": 1.0224,
+      "step": 8830
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.745292901992798,
+      "learning_rate": 1e-05,
+      "loss": 1.0342,
+      "step": 8840
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.3959574699401855,
+      "learning_rate": 1e-05,
+      "loss": 0.9508,
+      "step": 8850
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.9239192008972168,
+      "learning_rate": 1e-05,
+      "loss": 1.0539,
+      "step": 8860
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.345761299133301,
+      "learning_rate": 1e-05,
+      "loss": 1.0635,
+      "step": 8870
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.51446270942688,
+      "learning_rate": 1e-05,
+      "loss": 1.0992,
+      "step": 8880
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.9711889028549194,
+      "learning_rate": 1e-05,
+      "loss": 1.0458,
+      "step": 8890
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.8537259101867676,
+      "learning_rate": 1e-05,
+      "loss": 0.9539,
+      "step": 8900
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.368511438369751,
+      "learning_rate": 1e-05,
+      "loss": 0.9527,
+      "step": 8910
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.031531810760498,
+      "learning_rate": 1e-05,
+      "loss": 1.1827,
+      "step": 8920
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.894315004348755,
+      "learning_rate": 1e-05,
+      "loss": 1.0435,
+      "step": 8930
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 3.3386778831481934,
+      "learning_rate": 1e-05,
+      "loss": 1.0066,
+      "step": 8940
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 3.863614320755005,
+      "learning_rate": 1e-05,
+      "loss": 1.1069,
+      "step": 8950
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.729940176010132,
+      "learning_rate": 1e-05,
+      "loss": 1.0084,
+      "step": 8960
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.8039450645446777,
+      "learning_rate": 1e-05,
+      "loss": 0.9883,
+      "step": 8970
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.1626617908477783,
+      "learning_rate": 1e-05,
+      "loss": 0.9813,
+      "step": 8980
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 3.061138153076172,
+      "learning_rate": 1e-05,
+      "loss": 1.0289,
+      "step": 8990
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.4434292316436768,
+      "learning_rate": 1e-05,
+      "loss": 1.0519,
+      "step": 9000
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.2953484058380127,
+      "learning_rate": 1e-05,
+      "loss": 1.0233,
+      "step": 9010
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.4721264839172363,
+      "learning_rate": 1e-05,
+      "loss": 1.0358,
+      "step": 9020
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.9545519351959229,
+      "learning_rate": 1e-05,
+      "loss": 1.0004,
+      "step": 9030
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.3768677711486816,
+      "learning_rate": 1e-05,
+      "loss": 1.0209,
+      "step": 9040
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.243574619293213,
+      "learning_rate": 1e-05,
+      "loss": 0.9869,
+      "step": 9050
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.229619264602661,
+      "learning_rate": 1e-05,
+      "loss": 0.9306,
+      "step": 9060
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.9487369060516357,
+      "learning_rate": 1e-05,
+      "loss": 1.0493,
+      "step": 9070
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.8562098741531372,
+      "learning_rate": 1e-05,
+      "loss": 1.0514,
+      "step": 9080
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.3863279819488525,
+      "learning_rate": 1e-05,
+      "loss": 0.9485,
+      "step": 9090
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 6.1171770095825195,
+      "learning_rate": 1e-05,
+      "loss": 1.1196,
+      "step": 9100
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.144599199295044,
+      "learning_rate": 1e-05,
+      "loss": 1.0483,
+      "step": 9110
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.211543560028076,
+      "learning_rate": 1e-05,
+      "loss": 1.128,
+      "step": 9120
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.9530116319656372,
+      "learning_rate": 1e-05,
+      "loss": 0.9435,
+      "step": 9130
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.7647440433502197,
+      "learning_rate": 1e-05,
+      "loss": 1.0141,
+      "step": 9140
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.4697787761688232,
+      "learning_rate": 1e-05,
+      "loss": 1.0476,
+      "step": 9150
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 3.207587480545044,
+      "learning_rate": 1e-05,
+      "loss": 0.992,
+      "step": 9160
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.5492539405822754,
+      "learning_rate": 1e-05,
+      "loss": 0.9752,
+      "step": 9170
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.7407188415527344,
+      "learning_rate": 1e-05,
+      "loss": 1.0607,
+      "step": 9180
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.3383569717407227,
+      "learning_rate": 1e-05,
+      "loss": 1.0751,
+      "step": 9190
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.1230878829956055,
+      "learning_rate": 1e-05,
+      "loss": 0.952,
+      "step": 9200
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.9829449653625488,
+      "learning_rate": 1e-05,
+      "loss": 1.0335,
+      "step": 9210
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 4.799077987670898,
+      "learning_rate": 1e-05,
+      "loss": 1.0537,
+      "step": 9220
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.8595337867736816,
+      "learning_rate": 1e-05,
+      "loss": 1.0133,
+      "step": 9230
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.6511495113372803,
+      "learning_rate": 1e-05,
+      "loss": 1.0115,
+      "step": 9240
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.9823254346847534,
+      "learning_rate": 1e-05,
+      "loss": 1.0785,
+      "step": 9250
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.7708723545074463,
+      "learning_rate": 1e-05,
+      "loss": 1.1013,
+      "step": 9260
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.3901848793029785,
+      "learning_rate": 1e-05,
+      "loss": 1.0543,
+      "step": 9270
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.7122440338134766,
+      "learning_rate": 1e-05,
+      "loss": 1.1529,
+      "step": 9280
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.1970365047454834,
+      "learning_rate": 1e-05,
+      "loss": 0.9951,
+      "step": 9290
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.774059534072876,
+      "learning_rate": 1e-05,
+      "loss": 1.0656,
+      "step": 9300
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.749340295791626,
+      "learning_rate": 1e-05,
+      "loss": 1.0989,
+      "step": 9310
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.2948596477508545,
+      "learning_rate": 1e-05,
+      "loss": 1.0436,
+      "step": 9320
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.7892682552337646,
+      "learning_rate": 1e-05,
+      "loss": 1.0071,
+      "step": 9330
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.0370535850524902,
+      "learning_rate": 1e-05,
+      "loss": 1.0727,
+      "step": 9340
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.822446823120117,
+      "learning_rate": 1e-05,
+      "loss": 1.0199,
+      "step": 9350
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.6964447498321533,
+      "learning_rate": 1e-05,
+      "loss": 1.0402,
+      "step": 9360
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.7196106910705566,
+      "learning_rate": 1e-05,
+      "loss": 0.9824,
+      "step": 9370
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.7182745933532715,
+      "learning_rate": 1e-05,
+      "loss": 1.0771,
+      "step": 9380
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.1117007732391357,
+      "learning_rate": 1e-05,
+      "loss": 1.0465,
+      "step": 9390
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.7967630624771118,
+      "learning_rate": 1e-05,
+      "loss": 0.992,
+      "step": 9400
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.49383282661438,
+      "learning_rate": 1e-05,
+      "loss": 1.0395,
+      "step": 9410
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.5202503204345703,
+      "learning_rate": 1e-05,
+      "loss": 1.0795,
+      "step": 9420
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.5935745239257812,
+      "learning_rate": 1e-05,
+      "loss": 1.02,
+      "step": 9430
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.4098920822143555,
+      "learning_rate": 1e-05,
+      "loss": 1.03,
+      "step": 9440
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.3071539402008057,
+      "learning_rate": 1e-05,
+      "loss": 1.0464,
+      "step": 9450
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.990229606628418,
+      "learning_rate": 1e-05,
+      "loss": 0.9617,
+      "step": 9460
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.6037960052490234,
+      "learning_rate": 1e-05,
+      "loss": 0.9039,
+      "step": 9470
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 5.809933185577393,
+      "learning_rate": 1e-05,
+      "loss": 1.1791,
+      "step": 9480
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.2311456203460693,
+      "learning_rate": 1e-05,
+      "loss": 0.9794,
+      "step": 9490
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.2406017780303955,
+      "learning_rate": 1e-05,
+      "loss": 1.0225,
+      "step": 9500
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.0920634269714355,
+      "learning_rate": 1e-05,
+      "loss": 0.9394,
+      "step": 9510
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.660165309906006,
+      "learning_rate": 1e-05,
+      "loss": 1.0346,
+      "step": 9520
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.8337314128875732,
+      "learning_rate": 1e-05,
+      "loss": 1.1071,
+      "step": 9530
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.5402066707611084,
+      "learning_rate": 1e-05,
+      "loss": 1.0252,
+      "step": 9540
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.081479549407959,
+      "learning_rate": 1e-05,
+      "loss": 1.021,
+      "step": 9550
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.7497069835662842,
+      "learning_rate": 1e-05,
+      "loss": 1.0593,
+      "step": 9560
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.5563762187957764,
+      "learning_rate": 1e-05,
+      "loss": 1.1167,
+      "step": 9570
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.853631854057312,
+      "learning_rate": 1e-05,
+      "loss": 0.9478,
+      "step": 9580
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.8485876321792603,
+      "learning_rate": 1e-05,
+      "loss": 0.943,
+      "step": 9590
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.1595373153686523,
+      "learning_rate": 1e-05,
+      "loss": 1.0255,
+      "step": 9600
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.704416513442993,
+      "learning_rate": 1e-05,
+      "loss": 1.0146,
+      "step": 9610
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 3.183342933654785,
+      "learning_rate": 1e-05,
+      "loss": 0.9901,
+      "step": 9620
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.692582130432129,
+      "learning_rate": 1e-05,
+      "loss": 1.0983,
+      "step": 9630
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.6139438152313232,
+      "learning_rate": 1e-05,
+      "loss": 0.9617,
+      "step": 9640
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.860938549041748,
+      "learning_rate": 1e-05,
+      "loss": 1.0004,
+      "step": 9650
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.947887420654297,
+      "learning_rate": 1e-05,
+      "loss": 1.0991,
+      "step": 9660
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 4.666177272796631,
+      "learning_rate": 1e-05,
+      "loss": 1.0992,
+      "step": 9670
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.027334690093994,
+      "learning_rate": 1e-05,
+      "loss": 1.0596,
+      "step": 9680
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 5.081746578216553,
+      "learning_rate": 1e-05,
+      "loss": 1.0268,
+      "step": 9690
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.0974204540252686,
+      "learning_rate": 1e-05,
+      "loss": 0.9762,
+      "step": 9700
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.575479745864868,
+      "learning_rate": 1e-05,
+      "loss": 1.1449,
+      "step": 9710
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.9377288818359375,
+      "learning_rate": 1e-05,
+      "loss": 1.0244,
+      "step": 9720
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.1439216136932373,
+      "learning_rate": 1e-05,
+      "loss": 1.0813,
+      "step": 9730
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.694765329360962,
+      "learning_rate": 1e-05,
+      "loss": 1.0974,
+      "step": 9740
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.7554609775543213,
+      "learning_rate": 1e-05,
+      "loss": 1.083,
+      "step": 9750
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.0787527561187744,
+      "learning_rate": 1e-05,
+      "loss": 1.0616,
+      "step": 9760
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.455043315887451,
+      "learning_rate": 1e-05,
+      "loss": 1.0272,
+      "step": 9770
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.3333399295806885,
+      "learning_rate": 1e-05,
+      "loss": 1.0452,
+      "step": 9780
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.8097100257873535,
+      "learning_rate": 1e-05,
+      "loss": 1.1011,
+      "step": 9790
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.2194035053253174,
+      "learning_rate": 1e-05,
+      "loss": 1.0118,
+      "step": 9800
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.151822090148926,
+      "learning_rate": 1e-05,
+      "loss": 1.0312,
+      "step": 9810
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.602029323577881,
+      "learning_rate": 1e-05,
+      "loss": 1.009,
+      "step": 9820
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.129317283630371,
+      "learning_rate": 1e-05,
+      "loss": 0.9707,
+      "step": 9830
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.7062203884124756,
+      "learning_rate": 1e-05,
+      "loss": 1.0804,
+      "step": 9840
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 3.2763800621032715,
+      "learning_rate": 1e-05,
+      "loss": 1.0053,
+      "step": 9850
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.8125734329223633,
+      "learning_rate": 1e-05,
+      "loss": 1.0487,
+      "step": 9860
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.6083922386169434,
+      "learning_rate": 1e-05,
+      "loss": 0.9819,
+      "step": 9870
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.2002546787261963,
+      "learning_rate": 1e-05,
+      "loss": 1.0543,
+      "step": 9880
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.3200905323028564,
+      "learning_rate": 1e-05,
+      "loss": 1.1133,
+      "step": 9890
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.079068422317505,
+      "learning_rate": 1e-05,
+      "loss": 0.998,
+      "step": 9900
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.647366762161255,
+      "learning_rate": 1e-05,
+      "loss": 1.0267,
+      "step": 9910
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.9108846187591553,
+      "learning_rate": 1e-05,
+      "loss": 1.088,
+      "step": 9920
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.3041622638702393,
+      "learning_rate": 1e-05,
+      "loss": 1.2272,
+      "step": 9930
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.5793256759643555,
+      "learning_rate": 1e-05,
+      "loss": 0.9576,
+      "step": 9940
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.012216329574585,
+      "learning_rate": 1e-05,
+      "loss": 0.9958,
+      "step": 9950
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.0991573333740234,
+      "learning_rate": 1e-05,
+      "loss": 1.0183,
+      "step": 9960
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.904991865158081,
+      "learning_rate": 1e-05,
+      "loss": 1.0034,
+      "step": 9970
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.178138017654419,
+      "learning_rate": 1e-05,
+      "loss": 1.056,
+      "step": 9980
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.0459444522857666,
+      "learning_rate": 1e-05,
+      "loss": 1.0381,
+      "step": 9990
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 3.139091730117798,
+      "learning_rate": 1e-05,
+      "loss": 1.0501,
+      "step": 10000
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.130234718322754,
+      "learning_rate": 1e-05,
+      "loss": 0.9769,
+      "step": 10010
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.086378574371338,
+      "learning_rate": 1e-05,
+      "loss": 1.0102,
+      "step": 10020
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 4.662710666656494,
+      "learning_rate": 1e-05,
+      "loss": 1.0031,
+      "step": 10030
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.1875267028808594,
+      "learning_rate": 1e-05,
+      "loss": 1.0599,
+      "step": 10040
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.934441089630127,
+      "learning_rate": 1e-05,
+      "loss": 1.055,
+      "step": 10050
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.363140344619751,
+      "learning_rate": 1e-05,
+      "loss": 1.0554,
+      "step": 10060
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.435114622116089,
+      "learning_rate": 1e-05,
+      "loss": 1.0412,
+      "step": 10070
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.3322625160217285,
+      "learning_rate": 1e-05,
+      "loss": 1.0307,
+      "step": 10080
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.0348896980285645,
+      "learning_rate": 1e-05,
+      "loss": 1.106,
+      "step": 10090
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.9253132343292236,
+      "learning_rate": 1e-05,
+      "loss": 0.9802,
+      "step": 10100
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.191121816635132,
+      "learning_rate": 1e-05,
+      "loss": 0.9393,
+      "step": 10110
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.349829912185669,
+      "learning_rate": 1e-05,
+      "loss": 1.0189,
+      "step": 10120
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.29323410987854,
+      "learning_rate": 1e-05,
+      "loss": 1.079,
+      "step": 10130
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.9248541593551636,
+      "learning_rate": 1e-05,
+      "loss": 0.9665,
+      "step": 10140
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 3.685276746749878,
+      "learning_rate": 1e-05,
+      "loss": 1.0695,
+      "step": 10150
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.027282953262329,
+      "learning_rate": 1e-05,
+      "loss": 1.0601,
+      "step": 10160
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.761247396469116,
+      "learning_rate": 1e-05,
+      "loss": 1.1077,
+      "step": 10170
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.593219518661499,
+      "learning_rate": 1e-05,
+      "loss": 1.0019,
+      "step": 10180
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.4198343753814697,
+      "learning_rate": 1e-05,
+      "loss": 1.0267,
+      "step": 10190
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.7657220363616943,
+      "learning_rate": 1e-05,
+      "loss": 0.999,
+      "step": 10200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.4793217182159424,
+      "learning_rate": 1e-05,
+      "loss": 0.9946,
+      "step": 10210
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.2479584217071533,
+      "learning_rate": 1e-05,
+      "loss": 1.0673,
+      "step": 10220
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.7868390083312988,
+      "learning_rate": 1e-05,
+      "loss": 0.9873,
+      "step": 10230
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.0158581733703613,
+      "learning_rate": 1e-05,
+      "loss": 0.989,
+      "step": 10240
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.350095272064209,
+      "learning_rate": 1e-05,
+      "loss": 1.035,
+      "step": 10250
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.529557466506958,
+      "learning_rate": 1e-05,
+      "loss": 1.0576,
+      "step": 10260
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.16300630569458,
+      "learning_rate": 1e-05,
+      "loss": 0.9788,
+      "step": 10270
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.6851983070373535,
+      "learning_rate": 1e-05,
+      "loss": 1.1226,
+      "step": 10280
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.6486752033233643,
+      "learning_rate": 1e-05,
+      "loss": 1.126,
+      "step": 10290
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.3545916080474854,
+      "learning_rate": 1e-05,
+      "loss": 1.053,
+      "step": 10300
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.8078691959381104,
+      "learning_rate": 1e-05,
+      "loss": 1.0989,
+      "step": 10310
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.6008193492889404,
+      "learning_rate": 1e-05,
+      "loss": 1.0376,
+      "step": 10320
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.365222692489624,
+      "learning_rate": 1e-05,
+      "loss": 1.053,
+      "step": 10330
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.4086787700653076,
+      "learning_rate": 1e-05,
+      "loss": 1.0464,
+      "step": 10340
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.502668619155884,
+      "learning_rate": 1e-05,
+      "loss": 1.0884,
+      "step": 10350
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.446023941040039,
+      "learning_rate": 1e-05,
+      "loss": 1.1663,
+      "step": 10360
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.8586957454681396,
+      "learning_rate": 1e-05,
+      "loss": 1.1375,
+      "step": 10370
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.550720453262329,
+      "learning_rate": 1e-05,
+      "loss": 1.0003,
+      "step": 10380
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.0043468475341797,
+      "learning_rate": 1e-05,
+      "loss": 1.1144,
+      "step": 10390
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.8488130569458008,
+      "learning_rate": 1e-05,
+      "loss": 0.9635,
+      "step": 10400
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.419969320297241,
+      "learning_rate": 1e-05,
+      "loss": 0.9657,
+      "step": 10410
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.233978748321533,
+      "learning_rate": 1e-05,
+      "loss": 1.034,
+      "step": 10420
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.712808609008789,
+      "learning_rate": 1e-05,
+      "loss": 1.0175,
+      "step": 10430
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.9075738191604614,
+      "learning_rate": 1e-05,
+      "loss": 0.9141,
+      "step": 10440
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 4.534524917602539,
+      "learning_rate": 1e-05,
+      "loss": 1.079,
+      "step": 10450
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.488118886947632,
+      "learning_rate": 1e-05,
+      "loss": 1.062,
+      "step": 10460
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.9878602027893066,
+      "learning_rate": 1e-05,
+      "loss": 1.003,
+      "step": 10470
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.6196465492248535,
+      "learning_rate": 1e-05,
+      "loss": 1.0319,
+      "step": 10480
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.7483022212982178,
+      "learning_rate": 1e-05,
+      "loss": 1.0101,
+      "step": 10490
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.408924102783203,
+      "learning_rate": 1e-05,
+      "loss": 1.1074,
+      "step": 10500
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.5068442821502686,
+      "learning_rate": 1e-05,
+      "loss": 1.0417,
+      "step": 10510
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.5237784385681152,
+      "learning_rate": 1e-05,
+      "loss": 1.0879,
+      "step": 10520
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.158583402633667,
+      "learning_rate": 1e-05,
+      "loss": 1.0603,
+      "step": 10530
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.305474281311035,
+      "learning_rate": 1e-05,
+      "loss": 0.9813,
+      "step": 10540
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.7699382305145264,
+      "learning_rate": 1e-05,
+      "loss": 0.9862,
+      "step": 10550
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.487882614135742,
+      "learning_rate": 1e-05,
+      "loss": 0.9959,
+      "step": 10560
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.2361204624176025,
+      "learning_rate": 1e-05,
+      "loss": 1.0431,
+      "step": 10570
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.8864893913269043,
+      "learning_rate": 1e-05,
+      "loss": 1.1185,
+      "step": 10580
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.5624547004699707,
+      "learning_rate": 1e-05,
+      "loss": 1.0679,
+      "step": 10590
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.1008965969085693,
+      "learning_rate": 1e-05,
+      "loss": 0.9721,
+      "step": 10600
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.5166285037994385,
+      "learning_rate": 1e-05,
+      "loss": 1.1075,
+      "step": 10610
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.7157206535339355,
+      "learning_rate": 1e-05,
+      "loss": 1.1449,
+      "step": 10620
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.2612836360931396,
+      "learning_rate": 1e-05,
+      "loss": 1.0796,
+      "step": 10630
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.545149803161621,
+      "learning_rate": 1e-05,
+      "loss": 1.0865,
+      "step": 10640
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.023824453353882,
+      "learning_rate": 1e-05,
+      "loss": 1.06,
+      "step": 10650
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.386988878250122,
+      "learning_rate": 1e-05,
+      "loss": 1.0951,
+      "step": 10660
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 3.322854995727539,
+      "learning_rate": 1e-05,
+      "loss": 0.9547,
+      "step": 10670
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.6029868125915527,
+      "learning_rate": 1e-05,
+      "loss": 1.006,
+      "step": 10680
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.713660717010498,
+      "learning_rate": 1e-05,
+      "loss": 1.0695,
+      "step": 10690
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.325039863586426,
+      "learning_rate": 1e-05,
+      "loss": 1.0567,
+      "step": 10700
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.0697994232177734,
+      "learning_rate": 1e-05,
+      "loss": 1.0052,
+      "step": 10710
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.3208460807800293,
+      "learning_rate": 1e-05,
+      "loss": 1.0542,
+      "step": 10720
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.4321489334106445,
+      "learning_rate": 1e-05,
+      "loss": 1.0296,
+      "step": 10730
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.381392002105713,
+      "learning_rate": 1e-05,
+      "loss": 1.1445,
+      "step": 10740
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 2.379256248474121,
+      "learning_rate": 1e-05,
+      "loss": 1.0656,
+      "step": 10750
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 2.5627009868621826,
+      "learning_rate": 1e-05,
+      "loss": 0.9696,
+      "step": 10760
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 2.7615158557891846,
+      "learning_rate": 1e-05,
+      "loss": 1.0678,
+      "step": 10770
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 2.501905679702759,
+      "learning_rate": 1e-05,
+      "loss": 1.0085,
+      "step": 10780
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.9379123449325562,
+      "learning_rate": 1e-05,
+      "loss": 1.0005,
+      "step": 10790
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 3.583933115005493,
+      "learning_rate": 1e-05,
+      "loss": 1.0475,
+      "step": 10800
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 2.4750759601593018,
+      "learning_rate": 1e-05,
+      "loss": 1.0547,
+      "step": 10810
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 2.923191547393799,
+      "learning_rate": 1e-05,
+      "loss": 1.0374,
+      "step": 10820
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 3.4874658584594727,
+      "learning_rate": 1e-05,
+      "loss": 1.0205,
+      "step": 10830
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 2.9485082626342773,
+      "learning_rate": 1e-05,
+      "loss": 1.0203,
+      "step": 10840
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 3.664959192276001,
+      "learning_rate": 1e-05,
+      "loss": 1.0121,
+      "step": 10850
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 3.395310401916504,
+      "learning_rate": 1e-05,
+      "loss": 1.0349,
+      "step": 10860
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.9110727310180664,
+      "learning_rate": 1e-05,
+      "loss": 1.0065,
+      "step": 10870
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.657857894897461,
+      "learning_rate": 1e-05,
+      "loss": 1.0684,
+      "step": 10880
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.985198974609375,
+      "learning_rate": 1e-05,
+      "loss": 1.0981,
+      "step": 10890
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.062721014022827,
+      "learning_rate": 1e-05,
+      "loss": 0.9516,
+      "step": 10900
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 3.2055904865264893,
+      "learning_rate": 1e-05,
+      "loss": 1.018,
+      "step": 10910
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 3.117875814437866,
+      "learning_rate": 1e-05,
+      "loss": 1.0365,
+      "step": 10920
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.7929600477218628,
+      "learning_rate": 1e-05,
+      "loss": 1.0048,
+      "step": 10930
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.6944921016693115,
+      "learning_rate": 1e-05,
+      "loss": 1.1761,
+      "step": 10940
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.6278600692749023,
+      "learning_rate": 1e-05,
+      "loss": 1.054,
+      "step": 10950
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.301929235458374,
+      "learning_rate": 1e-05,
+      "loss": 1.003,
+      "step": 10960
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.0498104095458984,
+      "learning_rate": 1e-05,
+      "loss": 1.0503,
+      "step": 10970
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 2.6599652767181396,
+      "learning_rate": 1e-05,
+      "loss": 1.0086,
+      "step": 10980
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 2.476015567779541,
+      "learning_rate": 1e-05,
+      "loss": 1.0572,
+      "step": 10990
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 3.113619327545166,
+      "learning_rate": 1e-05,
+      "loss": 1.0165,
+      "step": 11000
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 2.55496883392334,
+      "learning_rate": 1e-05,
+      "loss": 1.0654,
+      "step": 11010
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 2.72886323928833,
+      "learning_rate": 1e-05,
+      "loss": 1.0975,
+      "step": 11020
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 2.1363415718078613,
+      "learning_rate": 1e-05,
+      "loss": 1.0918,
+      "step": 11030
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.203794002532959,
+      "learning_rate": 1e-05,
+      "loss": 1.0866,
+      "step": 11040
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.8999581336975098,
+      "learning_rate": 1e-05,
+      "loss": 1.0809,
+      "step": 11050
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.8935294151306152,
+      "learning_rate": 1e-05,
+      "loss": 1.0438,
+      "step": 11060
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.6402342319488525,
+      "learning_rate": 1e-05,
+      "loss": 0.9449,
+      "step": 11070
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.666870594024658,
+      "learning_rate": 1e-05,
+      "loss": 1.1531,
+      "step": 11080
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.20949125289917,
+      "learning_rate": 1e-05,
+      "loss": 0.9572,
+      "step": 11090
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.2581710815429688,
+      "learning_rate": 1e-05,
+      "loss": 1.0955,
+      "step": 11100
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.1770179271698,
+      "learning_rate": 1e-05,
+      "loss": 1.0565,
+      "step": 11110
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.217064619064331,
+      "learning_rate": 1e-05,
+      "loss": 1.0962,
+      "step": 11120
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.9007341861724854,
+      "learning_rate": 1e-05,
+      "loss": 1.0419,
+      "step": 11130
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.0666606426239014,
+      "learning_rate": 1e-05,
+      "loss": 0.9801,
+      "step": 11140
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.4318718910217285,
+      "learning_rate": 1e-05,
+      "loss": 1.0225,
+      "step": 11150
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 2.6048569679260254,
+      "learning_rate": 1e-05,
+      "loss": 1.0034,
+      "step": 11160
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 3.7844033241271973,
+      "learning_rate": 1e-05,
+      "loss": 1.0736,
+      "step": 11170
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 2.0433297157287598,
+      "learning_rate": 1e-05,
+      "loss": 1.1076,
+      "step": 11180
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 2.671957015991211,
+      "learning_rate": 1e-05,
+      "loss": 1.0873,
+      "step": 11190
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 3.4829108715057373,
+      "learning_rate": 1e-05,
+      "loss": 1.0102,
+      "step": 11200
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 3.030287981033325,
+      "learning_rate": 1e-05,
+      "loss": 1.1427,
+      "step": 11210
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 3.293470621109009,
+      "learning_rate": 1e-05,
+      "loss": 1.0337,
+      "step": 11220
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 3.201225519180298,
+      "learning_rate": 1e-05,
+      "loss": 1.0945,
+      "step": 11230
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 2.404139518737793,
+      "learning_rate": 1e-05,
+      "loss": 1.0539,
+      "step": 11240
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 3.0956571102142334,
+      "learning_rate": 1e-05,
+      "loss": 1.0154,
+      "step": 11250
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 2.3433375358581543,
+      "learning_rate": 1e-05,
+      "loss": 1.0592,
+      "step": 11260
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.212955951690674,
+      "learning_rate": 1e-05,
+      "loss": 0.9882,
+      "step": 11270
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.154575824737549,
+      "learning_rate": 1e-05,
+      "loss": 1.034,
+      "step": 11280
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 4.967568397521973,
+      "learning_rate": 1e-05,
+      "loss": 1.0411,
+      "step": 11290
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.8709118366241455,
+      "learning_rate": 1e-05,
+      "loss": 1.0207,
+      "step": 11300
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.2156543731689453,
+      "learning_rate": 1e-05,
+      "loss": 1.1151,
+      "step": 11310
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.266279458999634,
+      "learning_rate": 1e-05,
+      "loss": 1.0779,
+      "step": 11320
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.5265564918518066,
+      "learning_rate": 1e-05,
+      "loss": 1.0515,
+      "step": 11330
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.6855623722076416,
+      "learning_rate": 1e-05,
+      "loss": 0.9992,
+      "step": 11340
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 3.729240894317627,
+      "learning_rate": 1e-05,
+      "loss": 1.0587,
+      "step": 11350
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 4.715482234954834,
+      "learning_rate": 1e-05,
+      "loss": 1.0766,
+      "step": 11360
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 3.9233758449554443,
+      "learning_rate": 1e-05,
+      "loss": 1.1171,
+      "step": 11370
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.296520948410034,
+      "learning_rate": 1e-05,
+      "loss": 1.0624,
+      "step": 11380
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.0924103260040283,
+      "learning_rate": 1e-05,
+      "loss": 0.9887,
+      "step": 11390
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.1080734729766846,
+      "learning_rate": 1e-05,
+      "loss": 1.0284,
+      "step": 11400
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 3.715378522872925,
+      "learning_rate": 1e-05,
+      "loss": 1.0118,
+      "step": 11410
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.225140333175659,
+      "learning_rate": 1e-05,
+      "loss": 1.0021,
+      "step": 11420
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.3546242713928223,
+      "learning_rate": 1e-05,
+      "loss": 1.046,
+      "step": 11430
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.6119978427886963,
+      "learning_rate": 1e-05,
+      "loss": 1.1451,
+      "step": 11440
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 2.4649658203125,
+      "learning_rate": 1e-05,
+      "loss": 1.0584,
+      "step": 11450
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 2.748032331466675,
+      "learning_rate": 1e-05,
+      "loss": 1.1017,
+      "step": 11460
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 3.6822683811187744,
+      "learning_rate": 1e-05,
+      "loss": 0.9986,
+      "step": 11470
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 2.461031436920166,
+      "learning_rate": 1e-05,
+      "loss": 1.1179,
+      "step": 11480
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.9633231163024902,
+      "learning_rate": 1e-05,
+      "loss": 1.1283,
+      "step": 11490
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 2.6085851192474365,
+      "learning_rate": 1e-05,
+      "loss": 1.0043,
+      "step": 11500
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.4378316402435303,
+      "learning_rate": 1e-05,
+      "loss": 0.9722,
+      "step": 11510
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.0799951553344727,
+      "learning_rate": 1e-05,
+      "loss": 1.1276,
+      "step": 11520
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.2782275676727295,
+      "learning_rate": 1e-05,
+      "loss": 1.1108,
+      "step": 11530
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.5447194576263428,
+      "learning_rate": 1e-05,
+      "loss": 1.1114,
+      "step": 11540
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.684854745864868,
+      "learning_rate": 1e-05,
+      "loss": 1.0431,
+      "step": 11550
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 2.318377733230591,
+      "learning_rate": 1e-05,
+      "loss": 1.0562,
+      "step": 11560
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 2.3444907665252686,
+      "learning_rate": 1e-05,
+      "loss": 1.1112,
+      "step": 11570
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 2.450131416320801,
+      "learning_rate": 1e-05,
+      "loss": 1.0187,
+      "step": 11580
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 3.441574811935425,
+      "learning_rate": 1e-05,
+      "loss": 1.1204,
+      "step": 11590
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 3.4324586391448975,
+      "learning_rate": 1e-05,
+      "loss": 1.0423,
+      "step": 11600
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 2.931483030319214,
+      "learning_rate": 1e-05,
+      "loss": 1.112,
+      "step": 11610
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.359668731689453,
+      "learning_rate": 1e-05,
+      "loss": 1.0799,
+      "step": 11620
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.5941739082336426,
+      "learning_rate": 1e-05,
+      "loss": 0.997,
+      "step": 11630
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.8515117168426514,
+      "learning_rate": 1e-05,
+      "loss": 1.0763,
+      "step": 11640
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.849780797958374,
+      "eval_runtime": 21.7737,
+      "eval_samples_per_second": 45.927,
+      "eval_steps_per_second": 45.927,
+      "step": 11646
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.843797206878662,
+      "learning_rate": 1e-05,
+      "loss": 0.9207,
+      "step": 11650
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.897534966468811,
+      "learning_rate": 1e-05,
+      "loss": 0.7565,
+      "step": 11660
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 4.776578903198242,
+      "learning_rate": 1e-05,
+      "loss": 0.8172,
+      "step": 11670
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 2.535592555999756,
+      "learning_rate": 1e-05,
+      "loss": 0.7153,
+      "step": 11680
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 3.9274134635925293,
+      "learning_rate": 1e-05,
+      "loss": 0.7028,
+      "step": 11690
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 2.555481195449829,
+      "learning_rate": 1e-05,
+      "loss": 0.7169,
+      "step": 11700
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.7720446586608887,
+      "learning_rate": 1e-05,
+      "loss": 0.6535,
+      "step": 11710
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 2.346792221069336,
+      "learning_rate": 1e-05,
+      "loss": 0.6748,
+      "step": 11720
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 3.803128480911255,
+      "learning_rate": 1e-05,
+      "loss": 0.7134,
+      "step": 11730
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 3.682583808898926,
+      "learning_rate": 1e-05,
+      "loss": 0.7639,
+      "step": 11740
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 2.075239658355713,
+      "learning_rate": 1e-05,
+      "loss": 0.6385,
+      "step": 11750
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.98849618434906,
+      "learning_rate": 1e-05,
+      "loss": 0.6723,
+      "step": 11760
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 2.6985552310943604,
+      "learning_rate": 1e-05,
+      "loss": 0.658,
+      "step": 11770
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 3.478668689727783,
+      "learning_rate": 1e-05,
+      "loss": 0.6914,
+      "step": 11780
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 2.5365147590637207,
+      "learning_rate": 1e-05,
+      "loss": 0.6808,
+      "step": 11790
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 3.0944130420684814,
+      "learning_rate": 1e-05,
+      "loss": 0.6943,
+      "step": 11800
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 2.60815691947937,
+      "learning_rate": 1e-05,
+      "loss": 0.6429,
+      "step": 11810
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 2.929291248321533,
+      "learning_rate": 1e-05,
+      "loss": 0.7343,
+      "step": 11820
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 3.015611171722412,
+      "learning_rate": 1e-05,
+      "loss": 0.6732,
+      "step": 11830
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 3.052377939224243,
+      "learning_rate": 1e-05,
+      "loss": 0.8246,
+      "step": 11840
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.1589531898498535,
+      "learning_rate": 1e-05,
+      "loss": 0.6938,
+      "step": 11850
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.822674036026001,
+      "learning_rate": 1e-05,
+      "loss": 0.6883,
+      "step": 11860
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.3565139770507812,
+      "learning_rate": 1e-05,
+      "loss": 0.6635,
+      "step": 11870
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.325673818588257,
+      "learning_rate": 1e-05,
+      "loss": 0.7169,
+      "step": 11880
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.929018020629883,
+      "learning_rate": 1e-05,
+      "loss": 0.7049,
+      "step": 11890
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.7780282497406006,
+      "learning_rate": 1e-05,
+      "loss": 0.6585,
+      "step": 11900
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 2.293572187423706,
+      "learning_rate": 1e-05,
+      "loss": 0.7346,
+      "step": 11910
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 2.2444581985473633,
+      "learning_rate": 1e-05,
+      "loss": 0.6725,
+      "step": 11920
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 2.315701723098755,
+      "learning_rate": 1e-05,
+      "loss": 0.6875,
+      "step": 11930
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 2.3992395401000977,
+      "learning_rate": 1e-05,
+      "loss": 0.7435,
+      "step": 11940
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 2.5108776092529297,
+      "learning_rate": 1e-05,
+      "loss": 0.6308,
+      "step": 11950
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 3.073500156402588,
+      "learning_rate": 1e-05,
+      "loss": 0.6695,
+      "step": 11960
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.31484055519104,
+      "learning_rate": 1e-05,
+      "loss": 0.7226,
+      "step": 11970
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 3.2437198162078857,
+      "learning_rate": 1e-05,
+      "loss": 0.6808,
+      "step": 11980
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 3.2090394496917725,
+      "learning_rate": 1e-05,
+      "loss": 0.7653,
+      "step": 11990
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 3.439284086227417,
+      "learning_rate": 1e-05,
+      "loss": 0.6668,
+      "step": 12000
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 3.4921059608459473,
+      "learning_rate": 1e-05,
+      "loss": 0.7467,
+      "step": 12010
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.9912140369415283,
+      "learning_rate": 1e-05,
+      "loss": 0.723,
+      "step": 12020
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.1998276710510254,
+      "learning_rate": 1e-05,
+      "loss": 0.564,
+      "step": 12030
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.261547327041626,
+      "learning_rate": 1e-05,
+      "loss": 0.6412,
+      "step": 12040
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.4579389095306396,
+      "learning_rate": 1e-05,
+      "loss": 0.707,
+      "step": 12050
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.2751379013061523,
+      "learning_rate": 1e-05,
+      "loss": 0.6533,
+      "step": 12060
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.349090576171875,
+      "learning_rate": 1e-05,
+      "loss": 0.8271,
+      "step": 12070
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.216106653213501,
+      "learning_rate": 1e-05,
+      "loss": 0.669,
+      "step": 12080
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.4988324642181396,
+      "learning_rate": 1e-05,
+      "loss": 0.7576,
+      "step": 12090
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.8161988258361816,
+      "learning_rate": 1e-05,
+      "loss": 0.7229,
+      "step": 12100
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.5631985664367676,
+      "learning_rate": 1e-05,
+      "loss": 0.7278,
+      "step": 12110
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.14375376701355,
+      "learning_rate": 1e-05,
+      "loss": 0.665,
+      "step": 12120
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.518671989440918,
+      "learning_rate": 1e-05,
+      "loss": 0.6927,
+      "step": 12130
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.4540021419525146,
+      "learning_rate": 1e-05,
+      "loss": 0.7187,
+      "step": 12140
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 3.0300498008728027,
+      "learning_rate": 1e-05,
+      "loss": 0.6903,
+      "step": 12150
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 2.7865285873413086,
+      "learning_rate": 1e-05,
+      "loss": 0.6999,
+      "step": 12160
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 3.4259936809539795,
+      "learning_rate": 1e-05,
+      "loss": 0.7085,
+      "step": 12170
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 2.500450611114502,
+      "learning_rate": 1e-05,
+      "loss": 0.71,
+      "step": 12180
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 2.607896566390991,
+      "learning_rate": 1e-05,
+      "loss": 0.6563,
+      "step": 12190
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 3.381148338317871,
+      "learning_rate": 1e-05,
+      "loss": 0.7063,
+      "step": 12200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.8460057973861694,
+      "learning_rate": 1e-05,
+      "loss": 0.7151,
+      "step": 12210
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.369692802429199,
+      "learning_rate": 1e-05,
+      "loss": 0.7453,
+      "step": 12220
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.246582269668579,
+      "learning_rate": 1e-05,
+      "loss": 0.6495,
+      "step": 12230
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.5934314727783203,
+      "learning_rate": 1e-05,
+      "loss": 0.6994,
+      "step": 12240
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.804248094558716,
+      "learning_rate": 1e-05,
+      "loss": 0.7329,
+      "step": 12250
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.5733225345611572,
+      "learning_rate": 1e-05,
+      "loss": 0.6917,
+      "step": 12260
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.266530990600586,
+      "learning_rate": 1e-05,
+      "loss": 0.6712,
+      "step": 12270
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.1767702102661133,
+      "learning_rate": 1e-05,
+      "loss": 0.663,
+      "step": 12280
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.936537027359009,
+      "learning_rate": 1e-05,
+      "loss": 0.6707,
+      "step": 12290
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.5065464973449707,
+      "learning_rate": 1e-05,
+      "loss": 0.7478,
+      "step": 12300
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 2.3270397186279297,
+      "learning_rate": 1e-05,
+      "loss": 0.7466,
+      "step": 12310
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.324957847595215,
+      "learning_rate": 1e-05,
+      "loss": 0.7159,
+      "step": 12320
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.8157315254211426,
+      "learning_rate": 1e-05,
+      "loss": 0.6462,
+      "step": 12330
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 3.481329917907715,
+      "learning_rate": 1e-05,
+      "loss": 0.6302,
+      "step": 12340
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.075322389602661,
+      "learning_rate": 1e-05,
+      "loss": 0.6474,
+      "step": 12350
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.462676763534546,
+      "learning_rate": 1e-05,
+      "loss": 0.6556,
+      "step": 12360
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.9119620323181152,
+      "learning_rate": 1e-05,
+      "loss": 0.7142,
+      "step": 12370
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.999208450317383,
+      "learning_rate": 1e-05,
+      "loss": 0.6602,
+      "step": 12380
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.5624887943267822,
+      "learning_rate": 1e-05,
+      "loss": 0.7031,
+      "step": 12390
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.6522111892700195,
+      "learning_rate": 1e-05,
+      "loss": 0.7189,
+      "step": 12400
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.1354119777679443,
+      "learning_rate": 1e-05,
+      "loss": 0.7259,
+      "step": 12410
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.693638324737549,
+      "learning_rate": 1e-05,
+      "loss": 0.6861,
+      "step": 12420
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 2.5909125804901123,
+      "learning_rate": 1e-05,
+      "loss": 0.6882,
+      "step": 12430
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 2.9442074298858643,
+      "learning_rate": 1e-05,
+      "loss": 0.7714,
+      "step": 12440
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 2.713937282562256,
+      "learning_rate": 1e-05,
+      "loss": 0.6279,
+      "step": 12450
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 2.535755157470703,
+      "learning_rate": 1e-05,
+      "loss": 0.6467,
+      "step": 12460
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 2.3990252017974854,
+      "learning_rate": 1e-05,
+      "loss": 0.6979,
+      "step": 12470
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.9193114042282104,
+      "learning_rate": 1e-05,
+      "loss": 0.6374,
+      "step": 12480
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 2.7493979930877686,
+      "learning_rate": 1e-05,
+      "loss": 0.652,
+      "step": 12490
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 2.063673734664917,
+      "learning_rate": 1e-05,
+      "loss": 0.7104,
+      "step": 12500
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 2.684713840484619,
+      "learning_rate": 1e-05,
+      "loss": 0.6893,
+      "step": 12510
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 2.4041292667388916,
+      "learning_rate": 1e-05,
+      "loss": 0.6721,
+      "step": 12520
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.9186309576034546,
+      "learning_rate": 1e-05,
+      "loss": 0.666,
+      "step": 12530
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 3.9311676025390625,
+      "learning_rate": 1e-05,
+      "loss": 0.7856,
+      "step": 12540
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.0597290992736816,
+      "learning_rate": 1e-05,
+      "loss": 0.7454,
+      "step": 12550
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 3.2816972732543945,
+      "learning_rate": 1e-05,
+      "loss": 0.7107,
+      "step": 12560
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.5250680446624756,
+      "learning_rate": 1e-05,
+      "loss": 0.6695,
+      "step": 12570
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.6315548419952393,
+      "learning_rate": 1e-05,
+      "loss": 0.6763,
+      "step": 12580
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.9592995643615723,
+      "learning_rate": 1e-05,
+      "loss": 0.711,
+      "step": 12590
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.862454891204834,
+      "learning_rate": 1e-05,
+      "loss": 0.7075,
+      "step": 12600
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.8937571048736572,
+      "learning_rate": 1e-05,
+      "loss": 0.712,
+      "step": 12610
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.583103656768799,
+      "learning_rate": 1e-05,
+      "loss": 0.6897,
+      "step": 12620
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 3.4071152210235596,
+      "learning_rate": 1e-05,
+      "loss": 0.6719,
+      "step": 12630
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.4355502128601074,
+      "learning_rate": 1e-05,
+      "loss": 0.7007,
+      "step": 12640
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.881779670715332,
+      "learning_rate": 1e-05,
+      "loss": 0.7032,
+      "step": 12650
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.5507924556732178,
+      "learning_rate": 1e-05,
+      "loss": 0.731,
+      "step": 12660
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.8062305450439453,
+      "learning_rate": 1e-05,
+      "loss": 0.7516,
+      "step": 12670
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.2241053581237793,
+      "learning_rate": 1e-05,
+      "loss": 0.6573,
+      "step": 12680
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.9851512908935547,
+      "learning_rate": 1e-05,
+      "loss": 0.6876,
+      "step": 12690
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.9645676612854004,
+      "learning_rate": 1e-05,
+      "loss": 0.7315,
+      "step": 12700
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.893385887145996,
+      "learning_rate": 1e-05,
+      "loss": 0.7508,
+      "step": 12710
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.194856643676758,
+      "learning_rate": 1e-05,
+      "loss": 0.7279,
+      "step": 12720
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 3.4220829010009766,
+      "learning_rate": 1e-05,
+      "loss": 0.6461,
+      "step": 12730
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 3.7091429233551025,
+      "learning_rate": 1e-05,
+      "loss": 0.6547,
+      "step": 12740
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 3.0404088497161865,
+      "learning_rate": 1e-05,
+      "loss": 0.6513,
+      "step": 12750
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 2.5405914783477783,
+      "learning_rate": 1e-05,
+      "loss": 0.7101,
+      "step": 12760
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 3.01507306098938,
+      "learning_rate": 1e-05,
+      "loss": 0.6865,
+      "step": 12770
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 3.3664188385009766,
+      "learning_rate": 1e-05,
+      "loss": 0.6304,
+      "step": 12780
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 2.8610332012176514,
+      "learning_rate": 1e-05,
+      "loss": 0.6897,
+      "step": 12790
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 3.2472474575042725,
+      "learning_rate": 1e-05,
+      "loss": 0.6572,
+      "step": 12800
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 3.2986018657684326,
+      "learning_rate": 1e-05,
+      "loss": 0.6768,
+      "step": 12810
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 2.359896421432495,
+      "learning_rate": 1e-05,
+      "loss": 0.7468,
+      "step": 12820
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 2.9575815200805664,
+      "learning_rate": 1e-05,
+      "loss": 0.6759,
+      "step": 12830
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.660651445388794,
+      "learning_rate": 1e-05,
+      "loss": 0.6545,
+      "step": 12840
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.5651981830596924,
+      "learning_rate": 1e-05,
+      "loss": 0.6934,
+      "step": 12850
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.8365325927734375,
+      "learning_rate": 1e-05,
+      "loss": 0.684,
+      "step": 12860
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.5450923442840576,
+      "learning_rate": 1e-05,
+      "loss": 0.7585,
+      "step": 12870
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 3.447737455368042,
+      "learning_rate": 1e-05,
+      "loss": 0.7427,
+      "step": 12880
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 3.349693536758423,
+      "learning_rate": 1e-05,
+      "loss": 0.6351,
+      "step": 12890
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.5604896545410156,
+      "learning_rate": 1e-05,
+      "loss": 0.7356,
+      "step": 12900
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 3.0410192012786865,
+      "learning_rate": 1e-05,
+      "loss": 0.7518,
+      "step": 12910
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.895411252975464,
+      "learning_rate": 1e-05,
+      "loss": 0.7153,
+      "step": 12920
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.255626678466797,
+      "learning_rate": 1e-05,
+      "loss": 0.7202,
+      "step": 12930
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.550957441329956,
+      "learning_rate": 1e-05,
+      "loss": 0.7901,
+      "step": 12940
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.8669633865356445,
+      "learning_rate": 1e-05,
+      "loss": 0.6964,
+      "step": 12950
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 2.965333938598633,
+      "learning_rate": 1e-05,
+      "loss": 0.7156,
+      "step": 12960
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 2.201613664627075,
+      "learning_rate": 1e-05,
+      "loss": 0.6929,
+      "step": 12970
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 2.1694796085357666,
+      "learning_rate": 1e-05,
+      "loss": 0.6411,
+      "step": 12980
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 3.265608549118042,
+      "learning_rate": 1e-05,
+      "loss": 0.7531,
+      "step": 12990
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 3.275178909301758,
+      "learning_rate": 1e-05,
+      "loss": 0.807,
+      "step": 13000
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 2.581843137741089,
+      "learning_rate": 1e-05,
+      "loss": 0.7001,
+      "step": 13010
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.3566198348999023,
+      "learning_rate": 1e-05,
+      "loss": 0.7484,
+      "step": 13020
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.3498802185058594,
+      "learning_rate": 1e-05,
+      "loss": 0.6727,
+      "step": 13030
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.0292038917541504,
+      "learning_rate": 1e-05,
+      "loss": 0.7085,
+      "step": 13040
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 3.543577194213867,
+      "learning_rate": 1e-05,
+      "loss": 0.6679,
+      "step": 13050
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.6873228549957275,
+      "learning_rate": 1e-05,
+      "loss": 0.7377,
+      "step": 13060
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 3.999117851257324,
+      "learning_rate": 1e-05,
+      "loss": 0.8451,
+      "step": 13070
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.971100091934204,
+      "learning_rate": 1e-05,
+      "loss": 0.738,
+      "step": 13080
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.838461399078369,
+      "learning_rate": 1e-05,
+      "loss": 0.7077,
+      "step": 13090
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.252855062484741,
+      "learning_rate": 1e-05,
+      "loss": 0.7294,
+      "step": 13100
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 3.0866432189941406,
+      "learning_rate": 1e-05,
+      "loss": 0.7138,
+      "step": 13110
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.364053726196289,
+      "learning_rate": 1e-05,
+      "loss": 0.6806,
+      "step": 13120
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.768568515777588,
+      "learning_rate": 1e-05,
+      "loss": 0.7196,
+      "step": 13130
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 4.763648986816406,
+      "learning_rate": 1e-05,
+      "loss": 0.7165,
+      "step": 13140
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 3.4651520252227783,
+      "learning_rate": 1e-05,
+      "loss": 0.7155,
+      "step": 13150
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 2.2361979484558105,
+      "learning_rate": 1e-05,
+      "loss": 0.7171,
+      "step": 13160
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 2.607278347015381,
+      "learning_rate": 1e-05,
+      "loss": 0.698,
+      "step": 13170
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 2.6196305751800537,
+      "learning_rate": 1e-05,
+      "loss": 0.7162,
+      "step": 13180
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.6148579120635986,
+      "learning_rate": 1e-05,
+      "loss": 0.6812,
+      "step": 13190
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.5217950344085693,
+      "learning_rate": 1e-05,
+      "loss": 0.7015,
+      "step": 13200
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 3.2551751136779785,
+      "learning_rate": 1e-05,
+      "loss": 0.7336,
+      "step": 13210
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.751849412918091,
+      "learning_rate": 1e-05,
+      "loss": 0.7156,
+      "step": 13220
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.5473365783691406,
+      "learning_rate": 1e-05,
+      "loss": 0.7103,
+      "step": 13230
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.067836046218872,
+      "learning_rate": 1e-05,
+      "loss": 0.7334,
+      "step": 13240
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 2.5932745933532715,
+      "learning_rate": 1e-05,
+      "loss": 0.658,
+      "step": 13250
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 3.098078489303589,
+      "learning_rate": 1e-05,
+      "loss": 0.6701,
+      "step": 13260
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 2.336190700531006,
+      "learning_rate": 1e-05,
+      "loss": 0.6707,
+      "step": 13270
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 2.382223606109619,
+      "learning_rate": 1e-05,
+      "loss": 0.6473,
+      "step": 13280
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 2.551252603530884,
+      "learning_rate": 1e-05,
+      "loss": 0.8185,
+      "step": 13290
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 2.903029203414917,
+      "learning_rate": 1e-05,
+      "loss": 0.7188,
+      "step": 13300
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 3.2348668575286865,
+      "learning_rate": 1e-05,
+      "loss": 0.7409,
+      "step": 13310
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.7562198638916016,
+      "learning_rate": 1e-05,
+      "loss": 0.7271,
+      "step": 13320
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 3.178734540939331,
+      "learning_rate": 1e-05,
+      "loss": 0.7138,
+      "step": 13330
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.834533929824829,
+      "learning_rate": 1e-05,
+      "loss": 0.6777,
+      "step": 13340
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.3648908138275146,
+      "learning_rate": 1e-05,
+      "loss": 0.7587,
+      "step": 13350
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 3.3052682876586914,
+      "learning_rate": 1e-05,
+      "loss": 0.7108,
+      "step": 13360
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.5473461151123047,
+      "learning_rate": 1e-05,
+      "loss": 0.6858,
+      "step": 13370
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.689939260482788,
+      "learning_rate": 1e-05,
+      "loss": 0.7253,
+      "step": 13380
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.7012176513671875,
+      "learning_rate": 1e-05,
+      "loss": 0.659,
+      "step": 13390
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 3.217085361480713,
+      "learning_rate": 1e-05,
+      "loss": 0.7182,
+      "step": 13400
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.019303321838379,
+      "learning_rate": 1e-05,
+      "loss": 0.715,
+      "step": 13410
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.136199474334717,
+      "learning_rate": 1e-05,
+      "loss": 0.8071,
+      "step": 13420
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.7258598804473877,
+      "learning_rate": 1e-05,
+      "loss": 0.7511,
+      "step": 13430
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.4771461486816406,
+      "learning_rate": 1e-05,
+      "loss": 0.6709,
+      "step": 13440
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.250577449798584,
+      "learning_rate": 1e-05,
+      "loss": 0.7144,
+      "step": 13450
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.512507677078247,
+      "learning_rate": 1e-05,
+      "loss": 0.643,
+      "step": 13460
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.3582048416137695,
+      "learning_rate": 1e-05,
+      "loss": 0.7727,
+      "step": 13470
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 2.9323537349700928,
+      "learning_rate": 1e-05,
+      "loss": 0.7564,
+      "step": 13480
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 4.975164413452148,
+      "learning_rate": 1e-05,
+      "loss": 0.7155,
+      "step": 13490
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 3.4974381923675537,
+      "learning_rate": 1e-05,
+      "loss": 0.6929,
+      "step": 13500
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 4.331045150756836,
+      "learning_rate": 1e-05,
+      "loss": 0.7187,
+      "step": 13510
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 2.769953489303589,
+      "learning_rate": 1e-05,
+      "loss": 0.7293,
+      "step": 13520
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 2.571716785430908,
+      "learning_rate": 1e-05,
+      "loss": 0.6686,
+      "step": 13530
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.4005305767059326,
+      "learning_rate": 1e-05,
+      "loss": 0.6675,
+      "step": 13540
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 4.396423816680908,
+      "learning_rate": 1e-05,
+      "loss": 0.7801,
+      "step": 13550
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.680354356765747,
+      "learning_rate": 1e-05,
+      "loss": 0.6873,
+      "step": 13560
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.4758498668670654,
+      "learning_rate": 1e-05,
+      "loss": 0.7204,
+      "step": 13570
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.857762575149536,
+      "learning_rate": 1e-05,
+      "loss": 0.7049,
+      "step": 13580
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.436479330062866,
+      "learning_rate": 1e-05,
+      "loss": 0.6877,
+      "step": 13590
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 3.100313186645508,
+      "learning_rate": 1e-05,
+      "loss": 0.7481,
+      "step": 13600
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 2.6817588806152344,
+      "learning_rate": 1e-05,
+      "loss": 0.7384,
+      "step": 13610
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 2.7088801860809326,
+      "learning_rate": 1e-05,
+      "loss": 0.8041,
+      "step": 13620
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 3.363356351852417,
+      "learning_rate": 1e-05,
+      "loss": 0.8368,
+      "step": 13630
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 3.5108892917633057,
+      "learning_rate": 1e-05,
+      "loss": 0.699,
+      "step": 13640
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 2.41882061958313,
+      "learning_rate": 1e-05,
+      "loss": 0.7362,
+      "step": 13650
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.439345359802246,
+      "learning_rate": 1e-05,
+      "loss": 0.6982,
+      "step": 13660
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 3.261932134628296,
+      "learning_rate": 1e-05,
+      "loss": 0.7125,
+      "step": 13670
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.8121339082717896,
+      "learning_rate": 1e-05,
+      "loss": 0.644,
+      "step": 13680
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 3.049854040145874,
+      "learning_rate": 1e-05,
+      "loss": 0.7739,
+      "step": 13690
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.311997413635254,
+      "learning_rate": 1e-05,
+      "loss": 0.7085,
+      "step": 13700
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 4.8963189125061035,
+      "learning_rate": 1e-05,
+      "loss": 0.7852,
+      "step": 13710
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.440222978591919,
+      "learning_rate": 1e-05,
+      "loss": 0.8285,
+      "step": 13720
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 3.0829832553863525,
+      "learning_rate": 1e-05,
+      "loss": 0.7115,
+      "step": 13730
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.4086291790008545,
+      "learning_rate": 1e-05,
+      "loss": 0.6505,
+      "step": 13740
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 6.449065685272217,
+      "learning_rate": 1e-05,
+      "loss": 0.7672,
+      "step": 13750
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.3513927459716797,
+      "learning_rate": 1e-05,
+      "loss": 0.7172,
+      "step": 13760
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.6725003719329834,
+      "learning_rate": 1e-05,
+      "loss": 0.6698,
+      "step": 13770
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.1811881065368652,
+      "learning_rate": 1e-05,
+      "loss": 0.6756,
+      "step": 13780
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.4521212577819824,
+      "learning_rate": 1e-05,
+      "loss": 0.6899,
+      "step": 13790
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.6575536727905273,
+      "learning_rate": 1e-05,
+      "loss": 0.6729,
+      "step": 13800
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 3.644134283065796,
+      "learning_rate": 1e-05,
+      "loss": 0.7842,
+      "step": 13810
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.9881744384765625,
+      "learning_rate": 1e-05,
+      "loss": 0.7806,
+      "step": 13820
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.4042603969573975,
+      "learning_rate": 1e-05,
+      "loss": 0.7118,
+      "step": 13830
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.671569347381592,
+      "learning_rate": 1e-05,
+      "loss": 0.6719,
+      "step": 13840
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.206813097000122,
+      "learning_rate": 1e-05,
+      "loss": 0.7705,
+      "step": 13850
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.489980459213257,
+      "learning_rate": 1e-05,
+      "loss": 0.7533,
+      "step": 13860
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.6754541397094727,
+      "learning_rate": 1e-05,
+      "loss": 0.707,
+      "step": 13870
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.5215911865234375,
+      "learning_rate": 1e-05,
+      "loss": 0.7196,
+      "step": 13880
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.8557751178741455,
+      "learning_rate": 1e-05,
+      "loss": 0.7224,
+      "step": 13890
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.152554750442505,
+      "learning_rate": 1e-05,
+      "loss": 0.7437,
+      "step": 13900
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.7907458543777466,
+      "learning_rate": 1e-05,
+      "loss": 0.6867,
+      "step": 13910
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.468668222427368,
+      "learning_rate": 1e-05,
+      "loss": 0.702,
+      "step": 13920
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.148282766342163,
+      "learning_rate": 1e-05,
+      "loss": 0.6518,
+      "step": 13930
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 3.8288798332214355,
+      "learning_rate": 1e-05,
+      "loss": 0.7376,
+      "step": 13940
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.21659779548645,
+      "learning_rate": 1e-05,
+      "loss": 0.6753,
+      "step": 13950
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.80706787109375,
+      "learning_rate": 1e-05,
+      "loss": 0.7521,
+      "step": 13960
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 3.0065953731536865,
+      "learning_rate": 1e-05,
+      "loss": 0.6689,
+      "step": 13970
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.646472215652466,
+      "learning_rate": 1e-05,
+      "loss": 0.7145,
+      "step": 13980
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.4246387481689453,
+      "learning_rate": 1e-05,
+      "loss": 0.7228,
+      "step": 13990
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.947126865386963,
+      "learning_rate": 1e-05,
+      "loss": 0.6762,
+      "step": 14000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 2.71744704246521,
+      "learning_rate": 1e-05,
+      "loss": 0.7209,
+      "step": 14010
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 2.5872905254364014,
+      "learning_rate": 1e-05,
+      "loss": 0.6415,
+      "step": 14020
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 2.5897634029388428,
+      "learning_rate": 1e-05,
+      "loss": 0.7609,
+      "step": 14030
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 2.4245212078094482,
+      "learning_rate": 1e-05,
+      "loss": 0.6804,
+      "step": 14040
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 4.575323581695557,
+      "learning_rate": 1e-05,
+      "loss": 0.7517,
+      "step": 14050
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 2.2898592948913574,
+      "learning_rate": 1e-05,
+      "loss": 0.6849,
+      "step": 14060
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 3.359697103500366,
+      "learning_rate": 1e-05,
+      "loss": 0.7076,
+      "step": 14070
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.9763455390930176,
+      "learning_rate": 1e-05,
+      "loss": 0.747,
+      "step": 14080
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.798625946044922,
+      "learning_rate": 1e-05,
+      "loss": 0.6958,
+      "step": 14090
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.5556600093841553,
+      "learning_rate": 1e-05,
+      "loss": 0.7089,
+      "step": 14100
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.997627019882202,
+      "learning_rate": 1e-05,
+      "loss": 0.6879,
+      "step": 14110
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 3.4501309394836426,
+      "learning_rate": 1e-05,
+      "loss": 0.7083,
+      "step": 14120
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 3.426292657852173,
+      "learning_rate": 1e-05,
+      "loss": 0.7241,
+      "step": 14130
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 3.4111952781677246,
+      "learning_rate": 1e-05,
+      "loss": 0.7016,
+      "step": 14140
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 2.8646419048309326,
+      "learning_rate": 1e-05,
+      "loss": 0.7362,
+      "step": 14150
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 3.150669813156128,
+      "learning_rate": 1e-05,
+      "loss": 0.705,
+      "step": 14160
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 3.1819069385528564,
+      "learning_rate": 1e-05,
+      "loss": 0.7009,
+      "step": 14170
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 3.3720028400421143,
+      "learning_rate": 1e-05,
+      "loss": 0.6942,
+      "step": 14180
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 3.2291455268859863,
+      "learning_rate": 1e-05,
+      "loss": 0.7138,
+      "step": 14190
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 4.61140775680542,
+      "learning_rate": 1e-05,
+      "loss": 0.8018,
+      "step": 14200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 4.3528265953063965,
+      "learning_rate": 1e-05,
+      "loss": 0.7181,
+      "step": 14210
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 3.0563249588012695,
+      "learning_rate": 1e-05,
+      "loss": 0.7748,
+      "step": 14220
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.952367901802063,
+      "learning_rate": 1e-05,
+      "loss": 0.7157,
+      "step": 14230
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.618277072906494,
+      "learning_rate": 1e-05,
+      "loss": 0.6776,
+      "step": 14240
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.3816099166870117,
+      "learning_rate": 1e-05,
+      "loss": 0.7168,
+      "step": 14250
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 5.881453037261963,
+      "learning_rate": 1e-05,
+      "loss": 0.7459,
+      "step": 14260
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 3.0305838584899902,
+      "learning_rate": 1e-05,
+      "loss": 0.7478,
+      "step": 14270
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.958314895629883,
+      "learning_rate": 1e-05,
+      "loss": 0.7025,
+      "step": 14280
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.898390054702759,
+      "learning_rate": 1e-05,
+      "loss": 0.7767,
+      "step": 14290
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 3.295344829559326,
+      "learning_rate": 1e-05,
+      "loss": 0.7745,
+      "step": 14300
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 2.084602117538452,
+      "learning_rate": 1e-05,
+      "loss": 0.6736,
+      "step": 14310
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 3.2696902751922607,
+      "learning_rate": 1e-05,
+      "loss": 0.7092,
+      "step": 14320
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 2.228020429611206,
+      "learning_rate": 1e-05,
+      "loss": 0.6445,
+      "step": 14330
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 3.259493350982666,
+      "learning_rate": 1e-05,
+      "loss": 0.7543,
+      "step": 14340
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 4.3431501388549805,
+      "learning_rate": 1e-05,
+      "loss": 0.7228,
+      "step": 14350
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 2.6910886764526367,
+      "learning_rate": 1e-05,
+      "loss": 0.7423,
+      "step": 14360
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 3.4436614513397217,
+      "learning_rate": 1e-05,
+      "loss": 0.6928,
+      "step": 14370
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 3.588196039199829,
+      "learning_rate": 1e-05,
+      "loss": 0.7113,
+      "step": 14380
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.8467262983322144,
+      "learning_rate": 1e-05,
+      "loss": 0.6801,
+      "step": 14390
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 3.05918288230896,
+      "learning_rate": 1e-05,
+      "loss": 0.769,
+      "step": 14400
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 3.821401357650757,
+      "learning_rate": 1e-05,
+      "loss": 0.7344,
+      "step": 14410
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 3.6849071979522705,
+      "learning_rate": 1e-05,
+      "loss": 0.8138,
+      "step": 14420
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.413574457168579,
+      "learning_rate": 1e-05,
+      "loss": 0.7453,
+      "step": 14430
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 3.0552051067352295,
+      "learning_rate": 1e-05,
+      "loss": 0.665,
+      "step": 14440
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.2876293659210205,
+      "learning_rate": 1e-05,
+      "loss": 0.7064,
+      "step": 14450
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.75596022605896,
+      "learning_rate": 1e-05,
+      "loss": 0.7241,
+      "step": 14460
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.0840628147125244,
+      "learning_rate": 1e-05,
+      "loss": 0.7029,
+      "step": 14470
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 2.3339579105377197,
+      "learning_rate": 1e-05,
+      "loss": 0.6926,
+      "step": 14480
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 2.6257641315460205,
+      "learning_rate": 1e-05,
+      "loss": 0.7306,
+      "step": 14490
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 3.338038206100464,
+      "learning_rate": 1e-05,
+      "loss": 0.7289,
+      "step": 14500
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 3.947277545928955,
+      "learning_rate": 1e-05,
+      "loss": 0.7133,
+      "step": 14510
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 4.049989223480225,
+      "learning_rate": 1e-05,
+      "loss": 0.7539,
+      "step": 14520
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.6452383995056152,
+      "learning_rate": 1e-05,
+      "loss": 0.7556,
+      "step": 14530
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.4738383293151855,
+      "learning_rate": 1e-05,
+      "loss": 0.7545,
+      "step": 14540
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 3.1785967350006104,
+      "learning_rate": 1e-05,
+      "loss": 0.6619,
+      "step": 14550
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.7168779373168945,
+      "learning_rate": 1e-05,
+      "loss": 0.7748,
+      "step": 14560
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.5785582065582275,
+      "learning_rate": 1e-05,
+      "loss": 0.7754,
+      "step": 14570
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 3.7294836044311523,
+      "learning_rate": 1e-05,
+      "loss": 0.729,
+      "step": 14580
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.786391496658325,
+      "learning_rate": 1e-05,
+      "loss": 0.679,
+      "step": 14590
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 3.8487510681152344,
+      "learning_rate": 1e-05,
+      "loss": 0.7393,
+      "step": 14600
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.528351306915283,
+      "learning_rate": 1e-05,
+      "loss": 0.7025,
+      "step": 14610
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.2078957557678223,
+      "learning_rate": 1e-05,
+      "loss": 0.6564,
+      "step": 14620
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.7390918731689453,
+      "learning_rate": 1e-05,
+      "loss": 0.7214,
+      "step": 14630
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.95219087600708,
+      "learning_rate": 1e-05,
+      "loss": 0.7425,
+      "step": 14640
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.6406657695770264,
+      "learning_rate": 1e-05,
+      "loss": 0.7114,
+      "step": 14650
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 3.2420835494995117,
+      "learning_rate": 1e-05,
+      "loss": 0.7072,
+      "step": 14660
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 3.2121024131774902,
+      "learning_rate": 1e-05,
+      "loss": 0.7846,
+      "step": 14670
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.7226524353027344,
+      "learning_rate": 1e-05,
+      "loss": 0.7264,
+      "step": 14680
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.925915479660034,
+      "learning_rate": 1e-05,
+      "loss": 0.7575,
+      "step": 14690
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.7175862789154053,
+      "learning_rate": 1e-05,
+      "loss": 0.7436,
+      "step": 14700
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 2.9792709350585938,
+      "learning_rate": 1e-05,
+      "loss": 0.7475,
+      "step": 14710
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 2.419769048690796,
+      "learning_rate": 1e-05,
+      "loss": 0.757,
+      "step": 14720
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 2.0730252265930176,
+      "learning_rate": 1e-05,
+      "loss": 0.7351,
+      "step": 14730
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 3.0078718662261963,
+      "learning_rate": 1e-05,
+      "loss": 0.7393,
+      "step": 14740
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.7846797704696655,
+      "learning_rate": 1e-05,
+      "loss": 0.6745,
+      "step": 14750
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 2.7745330333709717,
+      "learning_rate": 1e-05,
+      "loss": 0.747,
+      "step": 14760
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 3.211197853088379,
+      "learning_rate": 1e-05,
+      "loss": 0.7089,
+      "step": 14770
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.642756223678589,
+      "learning_rate": 1e-05,
+      "loss": 0.7322,
+      "step": 14780
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.5132839679718018,
+      "learning_rate": 1e-05,
+      "loss": 0.6992,
+      "step": 14790
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.3744137287139893,
+      "learning_rate": 1e-05,
+      "loss": 0.6811,
+      "step": 14800
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.4576313495635986,
+      "learning_rate": 1e-05,
+      "loss": 0.7047,
+      "step": 14810
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.6812050342559814,
+      "learning_rate": 1e-05,
+      "loss": 0.753,
+      "step": 14820
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.089709997177124,
+      "learning_rate": 1e-05,
+      "loss": 0.7411,
+      "step": 14830
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.876819133758545,
+      "learning_rate": 1e-05,
+      "loss": 0.6582,
+      "step": 14840
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.931020736694336,
+      "learning_rate": 1e-05,
+      "loss": 0.7025,
+      "step": 14850
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 2.4845004081726074,
+      "learning_rate": 1e-05,
+      "loss": 0.7036,
+      "step": 14860
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 3.168095111846924,
+      "learning_rate": 1e-05,
+      "loss": 0.6892,
+      "step": 14870
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.145261526107788,
+      "learning_rate": 1e-05,
+      "loss": 0.662,
+      "step": 14880
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.742406129837036,
+      "learning_rate": 1e-05,
+      "loss": 0.7319,
+      "step": 14890
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.996185779571533,
+      "learning_rate": 1e-05,
+      "loss": 0.7447,
+      "step": 14900
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 3.1864700317382812,
+      "learning_rate": 1e-05,
+      "loss": 0.7869,
+      "step": 14910
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.646944522857666,
+      "learning_rate": 1e-05,
+      "loss": 0.7388,
+      "step": 14920
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.3198368549346924,
+      "learning_rate": 1e-05,
+      "loss": 0.7696,
+      "step": 14930
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 2.353590726852417,
+      "learning_rate": 1e-05,
+      "loss": 0.8069,
+      "step": 14940
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 2.4450297355651855,
+      "learning_rate": 1e-05,
+      "loss": 0.7355,
+      "step": 14950
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 2.8194401264190674,
+      "learning_rate": 1e-05,
+      "loss": 0.75,
+      "step": 14960
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 3.041092872619629,
+      "learning_rate": 1e-05,
+      "loss": 0.6854,
+      "step": 14970
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 4.193447589874268,
+      "learning_rate": 1e-05,
+      "loss": 0.7292,
+      "step": 14980
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 2.8166589736938477,
+      "learning_rate": 1e-05,
+      "loss": 0.7081,
+      "step": 14990
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.8678197860717773,
+      "learning_rate": 1e-05,
+      "loss": 0.7169,
+      "step": 15000
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.5955636501312256,
+      "learning_rate": 1e-05,
+      "loss": 0.7199,
+      "step": 15010
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.822446584701538,
+      "learning_rate": 1e-05,
+      "loss": 0.7213,
+      "step": 15020
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 3.7871756553649902,
+      "learning_rate": 1e-05,
+      "loss": 0.6311,
+      "step": 15030
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 4.210511207580566,
+      "learning_rate": 1e-05,
+      "loss": 0.738,
+      "step": 15040
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.3034372329711914,
+      "learning_rate": 1e-05,
+      "loss": 0.7456,
+      "step": 15050
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.2593023777008057,
+      "learning_rate": 1e-05,
+      "loss": 0.6465,
+      "step": 15060
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 3.289602756500244,
+      "learning_rate": 1e-05,
+      "loss": 0.7163,
+      "step": 15070
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.7729742527008057,
+      "learning_rate": 1e-05,
+      "loss": 0.7542,
+      "step": 15080
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 3.9700310230255127,
+      "learning_rate": 1e-05,
+      "loss": 0.6687,
+      "step": 15090
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 5.73222541809082,
+      "learning_rate": 1e-05,
+      "loss": 0.8529,
+      "step": 15100
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.1498641967773438,
+      "learning_rate": 1e-05,
+      "loss": 0.7654,
+      "step": 15110
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.562579870223999,
+      "learning_rate": 1e-05,
+      "loss": 0.7567,
+      "step": 15120
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.8067171573638916,
+      "learning_rate": 1e-05,
+      "loss": 0.8068,
+      "step": 15130
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.884761095046997,
+      "learning_rate": 1e-05,
+      "loss": 0.7657,
+      "step": 15140
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 3.6787796020507812,
+      "learning_rate": 1e-05,
+      "loss": 0.7246,
+      "step": 15150
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 4.4008331298828125,
+      "learning_rate": 1e-05,
+      "loss": 0.7994,
+      "step": 15160
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.7413175106048584,
+      "learning_rate": 1e-05,
+      "loss": 0.7031,
+      "step": 15170
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.5500998497009277,
+      "learning_rate": 1e-05,
+      "loss": 0.6695,
+      "step": 15180
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.7714643478393555,
+      "learning_rate": 1e-05,
+      "loss": 0.7173,
+      "step": 15190
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.525230884552002,
+      "learning_rate": 1e-05,
+      "loss": 0.7346,
+      "step": 15200
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.637814521789551,
+      "learning_rate": 1e-05,
+      "loss": 0.7767,
+      "step": 15210
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.278615713119507,
+      "learning_rate": 1e-05,
+      "loss": 0.7097,
+      "step": 15220
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.507465124130249,
+      "learning_rate": 1e-05,
+      "loss": 0.7192,
+      "step": 15230
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 3.573824405670166,
+      "learning_rate": 1e-05,
+      "loss": 0.7061,
+      "step": 15240
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.6793293952941895,
+      "learning_rate": 1e-05,
+      "loss": 0.7652,
+      "step": 15250
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 3.3592193126678467,
+      "learning_rate": 1e-05,
+      "loss": 0.738,
+      "step": 15260
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.9290943145751953,
+      "learning_rate": 1e-05,
+      "loss": 0.7563,
+      "step": 15270
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.167343854904175,
+      "learning_rate": 1e-05,
+      "loss": 0.6755,
+      "step": 15280
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 3.4597039222717285,
+      "learning_rate": 1e-05,
+      "loss": 0.7386,
+      "step": 15290
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 2.5685362815856934,
+      "learning_rate": 1e-05,
+      "loss": 0.7828,
+      "step": 15300
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 3.855380058288574,
+      "learning_rate": 1e-05,
+      "loss": 0.7743,
+      "step": 15310
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 2.524961233139038,
+      "learning_rate": 1e-05,
+      "loss": 0.7033,
+      "step": 15320
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 2.046072244644165,
+      "learning_rate": 1e-05,
+      "loss": 0.7532,
+      "step": 15330
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 4.325909614562988,
+      "learning_rate": 1e-05,
+      "loss": 0.7994,
+      "step": 15340
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 2.411571979522705,
+      "learning_rate": 1e-05,
+      "loss": 0.7081,
+      "step": 15350
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 3.4040896892547607,
+      "learning_rate": 1e-05,
+      "loss": 0.7443,
+      "step": 15360
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 2.6055779457092285,
+      "learning_rate": 1e-05,
+      "loss": 0.8008,
+      "step": 15370
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 2.8484854698181152,
+      "learning_rate": 1e-05,
+      "loss": 0.6909,
+      "step": 15380
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 2.2155449390411377,
+      "learning_rate": 1e-05,
+      "loss": 0.7675,
+      "step": 15390
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 2.5154049396514893,
+      "learning_rate": 1e-05,
+      "loss": 0.7156,
+      "step": 15400
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 3.7648818492889404,
+      "learning_rate": 1e-05,
+      "loss": 0.797,
+      "step": 15410
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 2.2900948524475098,
+      "learning_rate": 1e-05,
+      "loss": 0.7388,
+      "step": 15420
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 3.301736831665039,
+      "learning_rate": 1e-05,
+      "loss": 0.7711,
+      "step": 15430
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 11.01123332977295,
+      "learning_rate": 1e-05,
+      "loss": 0.8923,
+      "step": 15440
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 3.414224624633789,
+      "learning_rate": 1e-05,
+      "loss": 0.7313,
+      "step": 15450
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 2.749166965484619,
+      "learning_rate": 1e-05,
+      "loss": 0.7062,
+      "step": 15460
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 2.4409449100494385,
+      "learning_rate": 1e-05,
+      "loss": 0.7218,
+      "step": 15470
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 2.4345972537994385,
+      "learning_rate": 1e-05,
+      "loss": 0.7291,
+      "step": 15480
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 2.5892274379730225,
+      "learning_rate": 1e-05,
+      "loss": 0.7936,
+      "step": 15490
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 4.130225658416748,
+      "learning_rate": 1e-05,
+      "loss": 0.7268,
+      "step": 15500
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 3.403099298477173,
+      "learning_rate": 1e-05,
+      "loss": 0.7424,
+      "step": 15510
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.790379285812378,
+      "learning_rate": 1e-05,
+      "loss": 0.7942,
+      "step": 15520
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 3.102782964706421,
+      "learning_rate": 1e-05,
+      "loss": 0.7751,
+      "step": 15530
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.5872371196746826,
+      "learning_rate": 1e-05,
+      "loss": 0.7455,
+      "step": 15540
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 4.8439154624938965,
+      "learning_rate": 1e-05,
+      "loss": 0.7276,
+      "step": 15550
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.814612627029419,
+      "learning_rate": 1e-05,
+      "loss": 0.7593,
+      "step": 15560
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 3.2214343547821045,
+      "learning_rate": 1e-05,
+      "loss": 0.6791,
+      "step": 15570
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 4.046710014343262,
+      "learning_rate": 1e-05,
+      "loss": 0.7227,
+      "step": 15580
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 3.2067861557006836,
+      "learning_rate": 1e-05,
+      "loss": 0.7475,
+      "step": 15590
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 2.4715051651000977,
+      "learning_rate": 1e-05,
+      "loss": 0.7739,
+      "step": 15600
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 3.115117311477661,
+      "learning_rate": 1e-05,
+      "loss": 0.7745,
+      "step": 15610
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 3.757561206817627,
+      "learning_rate": 1e-05,
+      "loss": 0.7603,
+      "step": 15620
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 2.4830832481384277,
+      "learning_rate": 1e-05,
+      "loss": 0.7136,
+      "step": 15630
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 3.814326286315918,
+      "learning_rate": 1e-05,
+      "loss": 0.819,
+      "step": 15640
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 3.865314245223999,
+      "learning_rate": 1e-05,
+      "loss": 0.7187,
+      "step": 15650
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 4.542452335357666,
+      "learning_rate": 1e-05,
+      "loss": 0.6924,
+      "step": 15660
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 3.1028683185577393,
+      "learning_rate": 1e-05,
+      "loss": 0.7375,
+      "step": 15670
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 2.452650308609009,
+      "learning_rate": 1e-05,
+      "loss": 0.6479,
+      "step": 15680
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 2.847052574157715,
+      "learning_rate": 1e-05,
+      "loss": 0.7278,
+      "step": 15690
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 5.2814788818359375,
+      "learning_rate": 1e-05,
+      "loss": 0.8083,
+      "step": 15700
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 2.3837742805480957,
+      "learning_rate": 1e-05,
+      "loss": 0.8053,
+      "step": 15710
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 2.4592061042785645,
+      "learning_rate": 1e-05,
+      "loss": 0.6921,
+      "step": 15720
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 3.0937070846557617,
+      "learning_rate": 1e-05,
+      "loss": 0.7763,
+      "step": 15730
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 2.6138272285461426,
+      "learning_rate": 1e-05,
+      "loss": 0.7324,
+      "step": 15740
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 2.9379701614379883,
+      "learning_rate": 1e-05,
+      "loss": 0.7163,
+      "step": 15750
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.2379825115203857,
+      "learning_rate": 1e-05,
+      "loss": 0.6726,
+      "step": 15760
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.3431155681610107,
+      "learning_rate": 1e-05,
+      "loss": 0.7273,
+      "step": 15770
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 4.853468894958496,
+      "learning_rate": 1e-05,
+      "loss": 0.7136,
+      "step": 15780
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 3.053487777709961,
+      "learning_rate": 1e-05,
+      "loss": 0.7326,
+      "step": 15790
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.3642847537994385,
+      "learning_rate": 1e-05,
+      "loss": 0.748,
+      "step": 15800
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 2.356008768081665,
+      "learning_rate": 1e-05,
+      "loss": 0.7565,
+      "step": 15810
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 5.110803604125977,
+      "learning_rate": 1e-05,
+      "loss": 0.7384,
+      "step": 15820
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 2.1696460247039795,
+      "learning_rate": 1e-05,
+      "loss": 0.7006,
+      "step": 15830
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 2.4872984886169434,
+      "learning_rate": 1e-05,
+      "loss": 0.7122,
+      "step": 15840
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 2.6965861320495605,
+      "learning_rate": 1e-05,
+      "loss": 0.7435,
+      "step": 15850
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 2.5141866207122803,
+      "learning_rate": 1e-05,
+      "loss": 0.7116,
+      "step": 15860
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 3.048657178878784,
+      "learning_rate": 1e-05,
+      "loss": 0.7095,
+      "step": 15870
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.5788943767547607,
+      "learning_rate": 1e-05,
+      "loss": 0.7484,
+      "step": 15880
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.4205973148345947,
+      "learning_rate": 1e-05,
+      "loss": 0.7057,
+      "step": 15890
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 3.8227627277374268,
+      "learning_rate": 1e-05,
+      "loss": 0.7496,
+      "step": 15900
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.5252959728240967,
+      "learning_rate": 1e-05,
+      "loss": 0.7253,
+      "step": 15910
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.177140712738037,
+      "learning_rate": 1e-05,
+      "loss": 0.7887,
+      "step": 15920
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 2.435643434524536,
+      "learning_rate": 1e-05,
+      "loss": 0.816,
+      "step": 15930
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 2.5256426334381104,
+      "learning_rate": 1e-05,
+      "loss": 0.7462,
+      "step": 15940
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 2.2169747352600098,
+      "learning_rate": 1e-05,
+      "loss": 0.7447,
+      "step": 15950
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 2.4709417819976807,
+      "learning_rate": 1e-05,
+      "loss": 0.6927,
+      "step": 15960
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 3.4572293758392334,
+      "learning_rate": 1e-05,
+      "loss": 0.7019,
+      "step": 15970
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 2.54184627532959,
+      "learning_rate": 1e-05,
+      "loss": 0.7396,
+      "step": 15980
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.9560205936431885,
+      "learning_rate": 1e-05,
+      "loss": 0.7141,
+      "step": 15990
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.538736343383789,
+      "learning_rate": 1e-05,
+      "loss": 0.7024,
+      "step": 16000
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.4639668464660645,
+      "learning_rate": 1e-05,
+      "loss": 0.694,
+      "step": 16010
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 3.4678444862365723,
+      "learning_rate": 1e-05,
+      "loss": 0.7787,
+      "step": 16020
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.7737815380096436,
+      "learning_rate": 1e-05,
+      "loss": 0.7071,
+      "step": 16030
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 3.0524613857269287,
+      "learning_rate": 1e-05,
+      "loss": 0.7129,
+      "step": 16040
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.9612460136413574,
+      "learning_rate": 1e-05,
+      "loss": 0.7531,
+      "step": 16050
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.466224193572998,
+      "learning_rate": 1e-05,
+      "loss": 0.703,
+      "step": 16060
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 3.5040297508239746,
+      "learning_rate": 1e-05,
+      "loss": 0.7759,
+      "step": 16070
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.6110734939575195,
+      "learning_rate": 1e-05,
+      "loss": 0.7041,
+      "step": 16080
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.3298346996307373,
+      "learning_rate": 1e-05,
+      "loss": 0.7527,
+      "step": 16090
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.0623910427093506,
+      "learning_rate": 1e-05,
+      "loss": 0.7526,
+      "step": 16100
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 3.3194146156311035,
+      "learning_rate": 1e-05,
+      "loss": 0.6769,
+      "step": 16110
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 2.5238358974456787,
+      "learning_rate": 1e-05,
+      "loss": 0.7166,
+      "step": 16120
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 2.802138090133667,
+      "learning_rate": 1e-05,
+      "loss": 0.7491,
+      "step": 16130
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 3.0264785289764404,
+      "learning_rate": 1e-05,
+      "loss": 0.7306,
+      "step": 16140
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 5.061720848083496,
+      "learning_rate": 1e-05,
+      "loss": 0.7897,
+      "step": 16150
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 3.7770209312438965,
+      "learning_rate": 1e-05,
+      "loss": 0.6993,
+      "step": 16160
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 3.6209259033203125,
+      "learning_rate": 1e-05,
+      "loss": 0.8215,
+      "step": 16170
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 4.007264137268066,
+      "learning_rate": 1e-05,
+      "loss": 0.7738,
+      "step": 16180
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 3.5193967819213867,
+      "learning_rate": 1e-05,
+      "loss": 0.6896,
+      "step": 16190
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 2.8088245391845703,
+      "learning_rate": 1e-05,
+      "loss": 0.7401,
+      "step": 16200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 2.5565385818481445,
+      "learning_rate": 1e-05,
+      "loss": 0.7178,
+      "step": 16210
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 3.078495502471924,
+      "learning_rate": 1e-05,
+      "loss": 0.782,
+      "step": 16220
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 2.354825973510742,
+      "learning_rate": 1e-05,
+      "loss": 0.6915,
+      "step": 16230
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 2.650279998779297,
+      "learning_rate": 1e-05,
+      "loss": 0.7687,
+      "step": 16240
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 3.1134865283966064,
+      "learning_rate": 1e-05,
+      "loss": 0.7498,
+      "step": 16250
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 2.4980340003967285,
+      "learning_rate": 1e-05,
+      "loss": 0.7477,
+      "step": 16260
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 2.9300968647003174,
+      "learning_rate": 1e-05,
+      "loss": 0.716,
+      "step": 16270
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.97686505317688,
+      "learning_rate": 1e-05,
+      "loss": 0.7463,
+      "step": 16280
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.8819618225097656,
+      "learning_rate": 1e-05,
+      "loss": 0.7069,
+      "step": 16290
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.8394768238067627,
+      "learning_rate": 1e-05,
+      "loss": 0.7733,
+      "step": 16300
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.063542127609253,
+      "learning_rate": 1e-05,
+      "loss": 0.8501,
+      "step": 16310
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.2984085083007812,
+      "learning_rate": 1e-05,
+      "loss": 0.696,
+      "step": 16320
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.8086931705474854,
+      "learning_rate": 1e-05,
+      "loss": 0.7987,
+      "step": 16330
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 3.405078411102295,
+      "learning_rate": 1e-05,
+      "loss": 0.7761,
+      "step": 16340
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 2.172470808029175,
+      "learning_rate": 1e-05,
+      "loss": 0.7769,
+      "step": 16350
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 3.121769666671753,
+      "learning_rate": 1e-05,
+      "loss": 0.7938,
+      "step": 16360
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 2.402902364730835,
+      "learning_rate": 1e-05,
+      "loss": 0.7294,
+      "step": 16370
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 2.813246726989746,
+      "learning_rate": 1e-05,
+      "loss": 0.771,
+      "step": 16380
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 2.922886848449707,
+      "learning_rate": 1e-05,
+      "loss": 0.817,
+      "step": 16390
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 2.243450403213501,
+      "learning_rate": 1e-05,
+      "loss": 0.7381,
+      "step": 16400
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 9.812843322753906,
+      "learning_rate": 1e-05,
+      "loss": 0.8756,
+      "step": 16410
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 2.703944206237793,
+      "learning_rate": 1e-05,
+      "loss": 0.6877,
+      "step": 16420
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 3.160672187805176,
+      "learning_rate": 1e-05,
+      "loss": 0.733,
+      "step": 16430
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 3.238607168197632,
+      "learning_rate": 1e-05,
+      "loss": 0.7106,
+      "step": 16440
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 3.620293140411377,
+      "learning_rate": 1e-05,
+      "loss": 0.7273,
+      "step": 16450
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.082348346710205,
+      "learning_rate": 1e-05,
+      "loss": 0.7373,
+      "step": 16460
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 3.2884418964385986,
+      "learning_rate": 1e-05,
+      "loss": 0.7943,
+      "step": 16470
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.3396670818328857,
+      "learning_rate": 1e-05,
+      "loss": 0.7064,
+      "step": 16480
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 3.7502334117889404,
+      "learning_rate": 1e-05,
+      "loss": 0.75,
+      "step": 16490
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.7468998432159424,
+      "learning_rate": 1e-05,
+      "loss": 0.7514,
+      "step": 16500
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 3.606981039047241,
+      "learning_rate": 1e-05,
+      "loss": 0.7411,
+      "step": 16510
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 2.4418702125549316,
+      "learning_rate": 1e-05,
+      "loss": 0.7346,
+      "step": 16520
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 2.283881902694702,
+      "learning_rate": 1e-05,
+      "loss": 0.6863,
+      "step": 16530
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 3.3163726329803467,
+      "learning_rate": 1e-05,
+      "loss": 0.7655,
+      "step": 16540
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 3.4496238231658936,
+      "learning_rate": 1e-05,
+      "loss": 0.8603,
+      "step": 16550
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 2.912651538848877,
+      "learning_rate": 1e-05,
+      "loss": 0.794,
+      "step": 16560
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 2.5364766120910645,
+      "learning_rate": 1e-05,
+      "loss": 0.725,
+      "step": 16570
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 3.1615214347839355,
+      "learning_rate": 1e-05,
+      "loss": 0.7345,
+      "step": 16580
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 3.17732834815979,
+      "learning_rate": 1e-05,
+      "loss": 0.7786,
+      "step": 16590
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 3.815878391265869,
+      "learning_rate": 1e-05,
+      "loss": 0.7871,
+      "step": 16600
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 3.21816086769104,
+      "learning_rate": 1e-05,
+      "loss": 0.7331,
+      "step": 16610
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 3.8046021461486816,
+      "learning_rate": 1e-05,
+      "loss": 0.7428,
+      "step": 16620
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 2.7986233234405518,
+      "learning_rate": 1e-05,
+      "loss": 0.7871,
+      "step": 16630
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 4.534671783447266,
+      "learning_rate": 1e-05,
+      "loss": 0.7033,
+      "step": 16640
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 2.722402572631836,
+      "learning_rate": 1e-05,
+      "loss": 0.7478,
+      "step": 16650
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 2.5120465755462646,
+      "learning_rate": 1e-05,
+      "loss": 0.8048,
+      "step": 16660
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 3.8931288719177246,
+      "learning_rate": 1e-05,
+      "loss": 0.7489,
+      "step": 16670
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 3.675823211669922,
+      "learning_rate": 1e-05,
+      "loss": 0.7575,
+      "step": 16680
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 2.4684035778045654,
+      "learning_rate": 1e-05,
+      "loss": 0.7459,
+      "step": 16690
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 2.9120452404022217,
+      "learning_rate": 1e-05,
+      "loss": 0.7388,
+      "step": 16700
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 3.1678876876831055,
+      "learning_rate": 1e-05,
+      "loss": 0.7115,
+      "step": 16710
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 3.2457306385040283,
+      "learning_rate": 1e-05,
+      "loss": 0.7298,
+      "step": 16720
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 4.585461139678955,
+      "learning_rate": 1e-05,
+      "loss": 0.7348,
+      "step": 16730
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 2.2121076583862305,
+      "learning_rate": 1e-05,
+      "loss": 0.6789,
+      "step": 16740
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.2857534885406494,
+      "learning_rate": 1e-05,
+      "loss": 0.6961,
+      "step": 16750
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.597043514251709,
+      "learning_rate": 1e-05,
+      "loss": 0.754,
+      "step": 16760
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.364671230316162,
+      "learning_rate": 1e-05,
+      "loss": 0.77,
+      "step": 16770
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 3.1742351055145264,
+      "learning_rate": 1e-05,
+      "loss": 0.7161,
+      "step": 16780
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.498741388320923,
+      "learning_rate": 1e-05,
+      "loss": 0.6919,
+      "step": 16790
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.582516670227051,
+      "learning_rate": 1e-05,
+      "loss": 0.7376,
+      "step": 16800
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.1765763759613037,
+      "learning_rate": 1e-05,
+      "loss": 0.6758,
+      "step": 16810
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.645017385482788,
+      "learning_rate": 1e-05,
+      "loss": 0.7498,
+      "step": 16820
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.96962833404541,
+      "learning_rate": 1e-05,
+      "loss": 0.7105,
+      "step": 16830
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 3.203653335571289,
+      "learning_rate": 1e-05,
+      "loss": 0.7472,
+      "step": 16840
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.739818572998047,
+      "learning_rate": 1e-05,
+      "loss": 0.7642,
+      "step": 16850
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 2.86613130569458,
+      "learning_rate": 1e-05,
+      "loss": 0.7257,
+      "step": 16860
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 2.8112566471099854,
+      "learning_rate": 1e-05,
+      "loss": 0.7461,
+      "step": 16870
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 3.9736883640289307,
+      "learning_rate": 1e-05,
+      "loss": 0.7416,
+      "step": 16880
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 2.1086666584014893,
+      "learning_rate": 1e-05,
+      "loss": 0.6816,
+      "step": 16890
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 3.0595312118530273,
+      "learning_rate": 1e-05,
+      "loss": 0.7162,
+      "step": 16900
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 2.9815711975097656,
+      "learning_rate": 1e-05,
+      "loss": 0.7105,
+      "step": 16910
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 3.7968990802764893,
+      "learning_rate": 1e-05,
+      "loss": 0.6456,
+      "step": 16920
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 3.099365472793579,
+      "learning_rate": 1e-05,
+      "loss": 0.6936,
+      "step": 16930
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 4.1419782638549805,
+      "learning_rate": 1e-05,
+      "loss": 0.7544,
+      "step": 16940
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 2.833970308303833,
+      "learning_rate": 1e-05,
+      "loss": 0.7226,
+      "step": 16950
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 2.6359775066375732,
+      "learning_rate": 1e-05,
+      "loss": 0.7183,
+      "step": 16960
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 3.082583427429199,
+      "learning_rate": 1e-05,
+      "loss": 0.7529,
+      "step": 16970
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.820340871810913,
+      "learning_rate": 1e-05,
+      "loss": 0.7711,
+      "step": 16980
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.642461061477661,
+      "learning_rate": 1e-05,
+      "loss": 0.7484,
+      "step": 16990
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.2459590435028076,
+      "learning_rate": 1e-05,
+      "loss": 0.7215,
+      "step": 17000
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.3110170364379883,
+      "learning_rate": 1e-05,
+      "loss": 0.7875,
+      "step": 17010
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.8215103149414062,
+      "learning_rate": 1e-05,
+      "loss": 0.8104,
+      "step": 17020
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.3981997966766357,
+      "learning_rate": 1e-05,
+      "loss": 0.7895,
+      "step": 17030
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 2.9426681995391846,
+      "learning_rate": 1e-05,
+      "loss": 0.7858,
+      "step": 17040
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 2.6029345989227295,
+      "learning_rate": 1e-05,
+      "loss": 0.8176,
+      "step": 17050
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 2.9794445037841797,
+      "learning_rate": 1e-05,
+      "loss": 0.7859,
+      "step": 17060
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 3.120633125305176,
+      "learning_rate": 1e-05,
+      "loss": 0.7464,
+      "step": 17070
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 5.730931282043457,
+      "learning_rate": 1e-05,
+      "loss": 0.8152,
+      "step": 17080
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.9469467401504517,
+      "learning_rate": 1e-05,
+      "loss": 0.7083,
+      "step": 17090
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 3.874636650085449,
+      "learning_rate": 1e-05,
+      "loss": 0.768,
+      "step": 17100
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 2.6435718536376953,
+      "learning_rate": 1e-05,
+      "loss": 0.7137,
+      "step": 17110
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 4.349155426025391,
+      "learning_rate": 1e-05,
+      "loss": 0.7452,
+      "step": 17120
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 4.12204647064209,
+      "learning_rate": 1e-05,
+      "loss": 0.7668,
+      "step": 17130
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 3.769453287124634,
+      "learning_rate": 1e-05,
+      "loss": 0.6694,
+      "step": 17140
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 2.3985166549682617,
+      "learning_rate": 1e-05,
+      "loss": 0.7694,
+      "step": 17150
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 2.7735278606414795,
+      "learning_rate": 1e-05,
+      "loss": 0.7295,
+      "step": 17160
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 2.5758328437805176,
+      "learning_rate": 1e-05,
+      "loss": 0.7437,
+      "step": 17170
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 3.3075079917907715,
+      "learning_rate": 1e-05,
+      "loss": 0.7758,
+      "step": 17180
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 2.7021961212158203,
+      "learning_rate": 1e-05,
+      "loss": 0.72,
+      "step": 17190
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 2.937143087387085,
+      "learning_rate": 1e-05,
+      "loss": 0.741,
+      "step": 17200
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.422276020050049,
+      "learning_rate": 1e-05,
+      "loss": 0.6805,
+      "step": 17210
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.0770819187164307,
+      "learning_rate": 1e-05,
+      "loss": 0.7353,
+      "step": 17220
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.305924654006958,
+      "learning_rate": 1e-05,
+      "loss": 0.6862,
+      "step": 17230
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 3.3312768936157227,
+      "learning_rate": 1e-05,
+      "loss": 0.7512,
+      "step": 17240
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 3.8097422122955322,
+      "learning_rate": 1e-05,
+      "loss": 0.777,
+      "step": 17250
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.78053617477417,
+      "learning_rate": 1e-05,
+      "loss": 0.6928,
+      "step": 17260
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 2.543670654296875,
+      "learning_rate": 1e-05,
+      "loss": 0.701,
+      "step": 17270
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 3.243194580078125,
+      "learning_rate": 1e-05,
+      "loss": 0.7368,
+      "step": 17280
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 3.155773878097534,
+      "learning_rate": 1e-05,
+      "loss": 0.7553,
+      "step": 17290
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 3.1233174800872803,
+      "learning_rate": 1e-05,
+      "loss": 0.7981,
+      "step": 17300
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 4.1581220626831055,
+      "learning_rate": 1e-05,
+      "loss": 0.7076,
+      "step": 17310
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 3.5240795612335205,
+      "learning_rate": 1e-05,
+      "loss": 0.7145,
+      "step": 17320
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 3.32393217086792,
+      "learning_rate": 1e-05,
+      "loss": 0.7386,
+      "step": 17330
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 3.0469563007354736,
+      "learning_rate": 1e-05,
+      "loss": 0.738,
+      "step": 17340
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 3.3284199237823486,
+      "learning_rate": 1e-05,
+      "loss": 0.8062,
+      "step": 17350
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 5.088879585266113,
+      "learning_rate": 1e-05,
+      "loss": 0.7776,
+      "step": 17360
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 7.793527603149414,
+      "learning_rate": 1e-05,
+      "loss": 0.7674,
+      "step": 17370
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 3.6413803100585938,
+      "learning_rate": 1e-05,
+      "loss": 0.8351,
+      "step": 17380
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 3.5020976066589355,
+      "learning_rate": 1e-05,
+      "loss": 0.6976,
+      "step": 17390
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 2.9255354404449463,
+      "learning_rate": 1e-05,
+      "loss": 0.7419,
+      "step": 17400
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 3.126370906829834,
+      "learning_rate": 1e-05,
+      "loss": 0.8116,
+      "step": 17410
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 3.6022801399230957,
+      "learning_rate": 1e-05,
+      "loss": 0.8049,
+      "step": 17420
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 3.4221079349517822,
+      "learning_rate": 1e-05,
+      "loss": 0.7251,
+      "step": 17430
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 3.1578404903411865,
+      "learning_rate": 1e-05,
+      "loss": 0.732,
+      "step": 17440
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 2.2212891578674316,
+      "learning_rate": 1e-05,
+      "loss": 0.6766,
+      "step": 17450
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 3.3322994709014893,
+      "learning_rate": 1e-05,
+      "loss": 0.7178,
+      "step": 17460
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 2.0239431858062744,
+      "eval_runtime": 21.6628,
+      "eval_samples_per_second": 46.162,
+      "eval_steps_per_second": 46.162,
+      "step": 17469
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 17469,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 250,
+  "total_flos": 4.146314247762785e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}