diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4503 @@
+{
+  "best_metric": 0.3474305272102356,
+  "best_model_checkpoint": "./results\\checkpoint-6398",
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 6398,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 26.542728424072266,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.738,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 24.76641273498535,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.4407,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 30.07474708557129,
+      "learning_rate": 3e-06,
+      "loss": 2.4597,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 24.434463500976562,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.3404,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 23.83124542236328,
+      "learning_rate": 5e-06,
+      "loss": 2.1791,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 22.50973129272461,
+      "learning_rate": 6e-06,
+      "loss": 2.1673,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 27.750638961791992,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 2.2015,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 25.283756256103516,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.0006,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 25.05742835998535,
+      "learning_rate": 9e-06,
+      "loss": 1.7402,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 25.344383239746094,
+      "learning_rate": 1e-05,
+      "loss": 1.7645,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 28.333459854125977,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 2.1037,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 12.90853500366211,
+      "learning_rate": 1.2e-05,
+      "loss": 1.6745,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 25.75482940673828,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.8785,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 19.2341251373291,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.6035,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 22.45606803894043,
+      "learning_rate": 1.5e-05,
+      "loss": 1.5633,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 14.334227561950684,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2339,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 15.943977355957031,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.4193,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 21.580835342407227,
+      "learning_rate": 1.8e-05,
+      "loss": 1.3797,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 18.85316276550293,
+      "learning_rate": 1.9e-05,
+      "loss": 1.2302,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 21.618574142456055,
+      "learning_rate": 2e-05,
+      "loss": 1.1087,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 33.01541519165039,
+      "learning_rate": 2.1e-05,
+      "loss": 1.9294,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 18.221654891967773,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.2177,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 25.145170211791992,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.3802,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 18.087400436401367,
+      "learning_rate": 2.4e-05,
+      "loss": 1.3255,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 21.010631561279297,
+      "learning_rate": 2.5e-05,
+      "loss": 1.5414,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 10.053231239318848,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.9453,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 33.4912223815918,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.2211,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 20.72431182861328,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.9749,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 22.565200805664062,
+      "learning_rate": 2.9e-05,
+      "loss": 1.179,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 10.775777816772461,
+      "learning_rate": 3e-05,
+      "loss": 1.2001,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.4110307693481445,
+      "learning_rate": 3.1e-05,
+      "loss": 0.9093,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 30.766653060913086,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.2831,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 20.368473052978516,
+      "learning_rate": 3.3e-05,
+      "loss": 1.1205,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 46.62571334838867,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.036,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 25.56611442565918,
+      "learning_rate": 3.5e-05,
+      "loss": 1.3756,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 18.900516510009766,
+      "learning_rate": 3.6e-05,
+      "loss": 1.2767,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 23.277311325073242,
+      "learning_rate": 3.7e-05,
+      "loss": 0.8536,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 35.67240524291992,
+      "learning_rate": 3.8e-05,
+      "loss": 1.6216,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.9258451461792,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.0068,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 19.44056510925293,
+      "learning_rate": 4e-05,
+      "loss": 1.9435,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 19.775850296020508,
+      "learning_rate": 4.1e-05,
+      "loss": 1.2093,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 20.82231330871582,
+      "learning_rate": 4.2e-05,
+      "loss": 1.1288,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.916687488555908,
+      "learning_rate": 4.3e-05,
+      "loss": 0.9289,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 24.404117584228516,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.2781,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 17.555295944213867,
+      "learning_rate": 4.5e-05,
+      "loss": 1.1765,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 19.7391357421875,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.1167,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 19.916635513305664,
+      "learning_rate": 4.7e-05,
+      "loss": 1.0785,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 21.901708602905273,
+      "learning_rate": 4.8e-05,
+      "loss": 0.9351,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 21.167715072631836,
+      "learning_rate": 4.9e-05,
+      "loss": 1.3726,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 22.528865814208984,
+      "learning_rate": 5e-05,
+      "loss": 1.1176,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 23.06756019592285,
+      "learning_rate": 4.991522550016955e-05,
+      "loss": 1.2415,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 18.94842529296875,
+      "learning_rate": 4.98304510003391e-05,
+      "loss": 1.3143,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 23.966211318969727,
+      "learning_rate": 4.9745676500508646e-05,
+      "loss": 1.048,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 18.198823928833008,
+      "learning_rate": 4.96609020006782e-05,
+      "loss": 1.0476,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 11.191575050354004,
+      "learning_rate": 4.9576127500847744e-05,
+      "loss": 1.3924,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 42.68156814575195,
+      "learning_rate": 4.94913530010173e-05,
+      "loss": 1.3035,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 26.126558303833008,
+      "learning_rate": 4.940657850118684e-05,
+      "loss": 1.0578,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.13721752166748,
+      "learning_rate": 4.9321804001356395e-05,
+      "loss": 0.9093,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 38.31271743774414,
+      "learning_rate": 4.923702950152595e-05,
+      "loss": 0.8466,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 43.53594970703125,
+      "learning_rate": 4.915225500169549e-05,
+      "loss": 1.4861,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 32.57273864746094,
+      "learning_rate": 4.9067480501865045e-05,
+      "loss": 1.2987,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 32.26193618774414,
+      "learning_rate": 4.898270600203459e-05,
+      "loss": 1.3402,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 17.1823673248291,
+      "learning_rate": 4.8897931502204143e-05,
+      "loss": 1.0338,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 17.780929565429688,
+      "learning_rate": 4.881315700237369e-05,
+      "loss": 1.0368,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 18.093929290771484,
+      "learning_rate": 4.8728382502543235e-05,
+      "loss": 0.9465,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 23.80923080444336,
+      "learning_rate": 4.864360800271279e-05,
+      "loss": 0.9293,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 25.071569442749023,
+      "learning_rate": 4.855883350288233e-05,
+      "loss": 1.4204,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 17.217226028442383,
+      "learning_rate": 4.8474059003051885e-05,
+      "loss": 1.0646,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 16.813901901245117,
+      "learning_rate": 4.838928450322143e-05,
+      "loss": 0.9068,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 12.774974822998047,
+      "learning_rate": 4.830451000339098e-05,
+      "loss": 1.0567,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 9.550601959228516,
+      "learning_rate": 4.821973550356053e-05,
+      "loss": 0.9253,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 16.764188766479492,
+      "learning_rate": 4.8134961003730075e-05,
+      "loss": 1.1694,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 17.823640823364258,
+      "learning_rate": 4.805018650389963e-05,
+      "loss": 0.9413,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.949650287628174,
+      "learning_rate": 4.796541200406918e-05,
+      "loss": 1.2831,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 31.663116455078125,
+      "learning_rate": 4.788063750423873e-05,
+      "loss": 0.8206,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 19.267797470092773,
+      "learning_rate": 4.779586300440828e-05,
+      "loss": 0.8309,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 18.33287239074707,
+      "learning_rate": 4.771108850457783e-05,
+      "loss": 0.8532,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 14.53322982788086,
+      "learning_rate": 4.7626314004747376e-05,
+      "loss": 1.1592,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 19.1998291015625,
+      "learning_rate": 4.754153950491692e-05,
+      "loss": 0.9742,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 26.773860931396484,
+      "learning_rate": 4.7456765005086474e-05,
+      "loss": 1.2489,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 10.488304138183594,
+      "learning_rate": 4.737199050525602e-05,
+      "loss": 1.2353,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 26.352331161499023,
+      "learning_rate": 4.728721600542557e-05,
+      "loss": 1.1808,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 12.658868789672852,
+      "learning_rate": 4.720244150559512e-05,
+      "loss": 1.0206,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 23.39665985107422,
+      "learning_rate": 4.711766700576467e-05,
+      "loss": 0.9055,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 21.914995193481445,
+      "learning_rate": 4.7032892505934216e-05,
+      "loss": 1.1839,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 14.218494415283203,
+      "learning_rate": 4.694811800610376e-05,
+      "loss": 0.6352,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 31.233783721923828,
+      "learning_rate": 4.6863343506273314e-05,
+      "loss": 0.7302,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 24.867542266845703,
+      "learning_rate": 4.677856900644286e-05,
+      "loss": 1.1345,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 22.696847915649414,
+      "learning_rate": 4.669379450661241e-05,
+      "loss": 1.3002,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 17.476242065429688,
+      "learning_rate": 4.6609020006781964e-05,
+      "loss": 1.3256,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 21.66083526611328,
+      "learning_rate": 4.652424550695152e-05,
+      "loss": 0.914,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 20.41225242614746,
+      "learning_rate": 4.643947100712106e-05,
+      "loss": 0.7292,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 24.819978713989258,
+      "learning_rate": 4.635469650729061e-05,
+      "loss": 1.2431,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 33.943824768066406,
+      "learning_rate": 4.626992200746016e-05,
+      "loss": 1.6141,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 17.183805465698242,
+      "learning_rate": 4.6185147507629706e-05,
+      "loss": 0.7778,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.102858781814575,
+      "learning_rate": 4.610037300779926e-05,
+      "loss": 0.5896,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 23.15467071533203,
+      "learning_rate": 4.6015598507968804e-05,
+      "loss": 1.4732,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 23.75440216064453,
+      "learning_rate": 4.593082400813836e-05,
+      "loss": 1.2125,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 31.520126342773438,
+      "learning_rate": 4.58460495083079e-05,
+      "loss": 0.7964,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 14.303851127624512,
+      "learning_rate": 4.576127500847745e-05,
+      "loss": 0.8121,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 12.356579780578613,
+      "learning_rate": 4.5676500508647e-05,
+      "loss": 0.8765,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 36.80412292480469,
+      "learning_rate": 4.5591726008816546e-05,
+      "loss": 1.6265,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 32.955177307128906,
+      "learning_rate": 4.55069515089861e-05,
+      "loss": 0.7252,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 18.75346565246582,
+      "learning_rate": 4.5422177009155644e-05,
+      "loss": 0.7021,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 10.651359558105469,
+      "learning_rate": 4.5337402509325197e-05,
+      "loss": 1.0639,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 15.532186508178711,
+      "learning_rate": 4.525262800949475e-05,
+      "loss": 0.804,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 60.16484451293945,
+      "learning_rate": 4.5167853509664295e-05,
+      "loss": 0.7853,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 19.321165084838867,
+      "learning_rate": 4.508307900983385e-05,
+      "loss": 0.8331,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.306361436843872,
+      "learning_rate": 4.499830451000339e-05,
+      "loss": 0.9297,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 18.4438419342041,
+      "learning_rate": 4.4913530010172945e-05,
+      "loss": 1.4051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.134817123413086,
+      "learning_rate": 4.482875551034249e-05,
+      "loss": 0.5885,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 34.41851043701172,
+      "learning_rate": 4.474398101051204e-05,
+      "loss": 1.1314,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.790000915527344,
+      "learning_rate": 4.465920651068159e-05,
+      "loss": 1.2724,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 11.602365493774414,
+      "learning_rate": 4.4574432010851135e-05,
+      "loss": 0.7734,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 23.654390335083008,
+      "learning_rate": 4.448965751102069e-05,
+      "loss": 0.9145,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 37.119022369384766,
+      "learning_rate": 4.440488301119023e-05,
+      "loss": 1.0029,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 22.183420181274414,
+      "learning_rate": 4.4320108511359785e-05,
+      "loss": 0.9291,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 25.14228630065918,
+      "learning_rate": 4.423533401152933e-05,
+      "loss": 1.0194,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 15.864004135131836,
+      "learning_rate": 4.415055951169888e-05,
+      "loss": 1.0367,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.8865504264831543,
+      "learning_rate": 4.4065785011868436e-05,
+      "loss": 0.5909,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 20.187772750854492,
+      "learning_rate": 4.398101051203798e-05,
+      "loss": 1.23,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4391849040985107,
+      "learning_rate": 4.3896236012207534e-05,
+      "loss": 0.927,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 27.003379821777344,
+      "learning_rate": 4.381146151237708e-05,
+      "loss": 1.0672,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 9.609984397888184,
+      "learning_rate": 4.372668701254663e-05,
+      "loss": 0.7269,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 25.993255615234375,
+      "learning_rate": 4.364191251271618e-05,
+      "loss": 0.8122,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 10.979281425476074,
+      "learning_rate": 4.355713801288573e-05,
+      "loss": 0.8113,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.965178966522217,
+      "learning_rate": 4.3472363513055276e-05,
+      "loss": 0.656,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 27.325172424316406,
+      "learning_rate": 4.338758901322482e-05,
+      "loss": 1.011,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6009902954101562,
+      "learning_rate": 4.3302814513394374e-05,
+      "loss": 0.9724,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 56.60973358154297,
+      "learning_rate": 4.321804001356392e-05,
+      "loss": 1.0712,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 11.2178955078125,
+      "learning_rate": 4.313326551373347e-05,
+      "loss": 1.188,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 43.168479919433594,
+      "learning_rate": 4.304849101390302e-05,
+      "loss": 0.9892,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 22.50556755065918,
+      "learning_rate": 4.296371651407257e-05,
+      "loss": 0.8076,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.59372615814209,
+      "learning_rate": 4.2878942014242116e-05,
+      "loss": 0.5126,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 14.512517929077148,
+      "learning_rate": 4.279416751441167e-05,
+      "loss": 0.7622,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 61.02300262451172,
+      "learning_rate": 4.270939301458122e-05,
+      "loss": 1.2306,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 17.68853187561035,
+      "learning_rate": 4.2624618514750766e-05,
+      "loss": 1.1169,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 17.228788375854492,
+      "learning_rate": 4.253984401492032e-05,
+      "loss": 0.7908,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8683578968048096,
+      "learning_rate": 4.2455069515089864e-05,
+      "loss": 0.595,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 33.53107833862305,
+      "learning_rate": 4.237029501525942e-05,
+      "loss": 0.7539,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 17.268796920776367,
+      "learning_rate": 4.228552051542896e-05,
+      "loss": 0.9071,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.079007625579834,
+      "learning_rate": 4.220074601559851e-05,
+      "loss": 1.3714,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 14.3380765914917,
+      "learning_rate": 4.211597151576806e-05,
+      "loss": 0.8167,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 21.197294235229492,
+      "learning_rate": 4.2031197015937606e-05,
+      "loss": 0.8644,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 12.999835968017578,
+      "learning_rate": 4.194642251610716e-05,
+      "loss": 1.1938,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 37.84367752075195,
+      "learning_rate": 4.1861648016276704e-05,
+      "loss": 0.8975,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 24.756757736206055,
+      "learning_rate": 4.1776873516446257e-05,
+      "loss": 0.7786,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 35.1243896484375,
+      "learning_rate": 4.16920990166158e-05,
+      "loss": 0.8236,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.9437575340271,
+      "learning_rate": 4.160732451678535e-05,
+      "loss": 0.696,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 36.57485580444336,
+      "learning_rate": 4.15225500169549e-05,
+      "loss": 1.2849,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 17.662107467651367,
+      "learning_rate": 4.143777551712445e-05,
+      "loss": 1.0252,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 32.92692565917969,
+      "learning_rate": 4.1353001017294005e-05,
+      "loss": 1.1483,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 23.67911720275879,
+      "learning_rate": 4.126822651746355e-05,
+      "loss": 0.8862,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 10.78515625,
+      "learning_rate": 4.11834520176331e-05,
+      "loss": 0.7059,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 10.892401695251465,
+      "learning_rate": 4.109867751780265e-05,
+      "loss": 1.0229,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 31.348299026489258,
+      "learning_rate": 4.1013903017972195e-05,
+      "loss": 0.8373,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 18.273740768432617,
+      "learning_rate": 4.092912851814175e-05,
+      "loss": 1.6494,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.3065638542175293,
+      "learning_rate": 4.084435401831129e-05,
+      "loss": 1.4762,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 20.082490921020508,
+      "learning_rate": 4.0759579518480845e-05,
+      "loss": 0.6694,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 22.293842315673828,
+      "learning_rate": 4.067480501865039e-05,
+      "loss": 1.5581,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 10.693699836730957,
+      "learning_rate": 4.059003051881994e-05,
+      "loss": 1.1067,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 32.38418960571289,
+      "learning_rate": 4.050525601898949e-05,
+      "loss": 0.6269,
+      "step": 1620
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 11.499810218811035,
+      "learning_rate": 4.0420481519159035e-05,
+      "loss": 0.4731,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 27.077342987060547,
+      "learning_rate": 4.033570701932859e-05,
+      "loss": 1.1482,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 32.90080261230469,
+      "learning_rate": 4.025093251949813e-05,
+      "loss": 0.7159,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5575653910636902,
+      "learning_rate": 4.0166158019667685e-05,
+      "loss": 1.1091,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.83486270904541,
+      "learning_rate": 4.008138351983724e-05,
+      "loss": 0.6591,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.872043609619141,
+      "learning_rate": 3.999660902000678e-05,
+      "loss": 1.4829,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 46.48579406738281,
+      "learning_rate": 3.9911834520176336e-05,
+      "loss": 0.9756,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8250547647476196,
+      "learning_rate": 3.982706002034588e-05,
+      "loss": 1.029,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 14.628582000732422,
+      "learning_rate": 3.9742285520515434e-05,
+      "loss": 0.8788,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 16.0622615814209,
+      "learning_rate": 3.965751102068498e-05,
+      "loss": 0.9625,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5123465061187744,
+      "learning_rate": 3.957273652085453e-05,
+      "loss": 0.3242,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 24.70962142944336,
+      "learning_rate": 3.948796202102408e-05,
+      "loss": 1.0982,
+      "step": 1740
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3634880483150482,
+      "learning_rate": 3.940318752119362e-05,
+      "loss": 0.2891,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.15528139472007751,
+      "learning_rate": 3.9318413021363176e-05,
+      "loss": 0.964,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 49.00790023803711,
+      "learning_rate": 3.923363852153272e-05,
+      "loss": 1.1447,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 17.20952796936035,
+      "learning_rate": 3.9148864021702274e-05,
+      "loss": 0.2853,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 41.41338348388672,
+      "learning_rate": 3.906408952187182e-05,
+      "loss": 0.6706,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 18.779157638549805,
+      "learning_rate": 3.897931502204137e-05,
+      "loss": 0.529,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 14.820104598999023,
+      "learning_rate": 3.889454052221092e-05,
+      "loss": 0.8685,
+      "step": 1810
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 23.525177001953125,
+      "learning_rate": 3.880976602238047e-05,
+      "loss": 0.6292,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.442511647939682,
+      "learning_rate": 3.872499152255002e-05,
+      "loss": 0.9998,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.724640846252441,
+      "learning_rate": 3.864021702271957e-05,
+      "loss": 1.1116,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 26.15089225769043,
+      "learning_rate": 3.855544252288912e-05,
+      "loss": 1.109,
+      "step": 1850
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 25.109575271606445,
+      "learning_rate": 3.8470668023058666e-05,
+      "loss": 0.9564,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.969759464263916,
+      "learning_rate": 3.838589352322822e-05,
+      "loss": 0.9998,
+      "step": 1870
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.33139169216156,
+      "learning_rate": 3.8301119023397764e-05,
+      "loss": 1.0358,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 18.717729568481445,
+      "learning_rate": 3.821634452356731e-05,
+      "loss": 0.8999,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.792151927947998,
+      "learning_rate": 3.813157002373686e-05,
+      "loss": 0.3362,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 31.383317947387695,
+      "learning_rate": 3.804679552390641e-05,
+      "loss": 1.3567,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 17.265098571777344,
+      "learning_rate": 3.796202102407596e-05,
+      "loss": 1.3984,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 28.290605545043945,
+      "learning_rate": 3.7877246524245506e-05,
+      "loss": 0.9521,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 25.10551643371582,
+      "learning_rate": 3.779247202441506e-05,
+      "loss": 0.8199,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 15.570342063903809,
+      "learning_rate": 3.7707697524584604e-05,
+      "loss": 0.6976,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 48.913543701171875,
+      "learning_rate": 3.762292302475415e-05,
+      "loss": 1.1964,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 38.607730865478516,
+      "learning_rate": 3.75381485249237e-05,
+      "loss": 1.2215,
+      "step": 1970
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 61.75456619262695,
+      "learning_rate": 3.7453374025093255e-05,
+      "loss": 1.1017,
+      "step": 1980
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.9896836280822754,
+      "learning_rate": 3.736859952526281e-05,
+      "loss": 0.6199,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 22.329191207885742,
+      "learning_rate": 3.728382502543235e-05,
+      "loss": 0.6508,
+      "step": 2000
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 44.48137283325195,
+      "learning_rate": 3.7199050525601905e-05,
+      "loss": 0.3921,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 29.464078903198242,
+      "learning_rate": 3.711427602577145e-05,
+      "loss": 0.702,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4076429605484009,
+      "learning_rate": 3.7029501525940996e-05,
+      "loss": 0.6526,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 8.90776538848877,
+      "learning_rate": 3.694472702611055e-05,
+      "loss": 0.5163,
+      "step": 2040
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1676784753799438,
+      "learning_rate": 3.6859952526280095e-05,
+      "loss": 0.3613,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 17.235055923461914,
+      "learning_rate": 3.677517802644965e-05,
+      "loss": 0.6757,
+      "step": 2060
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 76.95849609375,
+      "learning_rate": 3.669040352661919e-05,
+      "loss": 0.6235,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 60.8438720703125,
+      "learning_rate": 3.6605629026788745e-05,
+      "loss": 1.293,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 52.46332550048828,
+      "learning_rate": 3.652085452695829e-05,
+      "loss": 1.0197,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.09444449096918106,
+      "learning_rate": 3.6436080027127836e-05,
+      "loss": 0.7193,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.06203605979681015,
+      "learning_rate": 3.635130552729739e-05,
+      "loss": 0.9315,
+      "step": 2110
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 18.52464485168457,
+      "learning_rate": 3.6266531027466935e-05,
+      "loss": 0.8383,
+      "step": 2120
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 21.279346466064453,
+      "learning_rate": 3.618175652763649e-05,
+      "loss": 0.4203,
+      "step": 2130
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 33.78806686401367,
+      "learning_rate": 3.609698202780604e-05,
+      "loss": 0.5023,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.57900857925415,
+      "learning_rate": 3.601220752797559e-05,
+      "loss": 0.8179,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 8.65097427368164,
+      "learning_rate": 3.592743302814514e-05,
+      "loss": 1.3782,
+      "step": 2160
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.10031997412443161,
+      "learning_rate": 3.584265852831468e-05,
+      "loss": 0.9372,
+      "step": 2170
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 26.388620376586914,
+      "learning_rate": 3.5757884028484236e-05,
+      "loss": 1.1846,
+      "step": 2180
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 12.027292251586914,
+      "learning_rate": 3.567310952865378e-05,
+      "loss": 0.4438,
+      "step": 2190
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.0509867668151855,
+      "learning_rate": 3.5588335028823334e-05,
+      "loss": 0.7817,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 18.15105438232422,
+      "learning_rate": 3.550356052899288e-05,
+      "loss": 0.9115,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0213444232940674,
+      "learning_rate": 3.541878602916243e-05,
+      "loss": 0.4687,
+      "step": 2220
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 71.69523620605469,
+      "learning_rate": 3.533401152933198e-05,
+      "loss": 0.7518,
+      "step": 2230
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 86.0584945678711,
+      "learning_rate": 3.524923702950152e-05,
+      "loss": 1.1291,
+      "step": 2240
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7845587730407715,
+      "learning_rate": 3.5164462529671076e-05,
+      "loss": 1.0255,
+      "step": 2250
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 10.846942901611328,
+      "learning_rate": 3.507968802984062e-05,
+      "loss": 0.6766,
+      "step": 2260
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 50.255889892578125,
+      "learning_rate": 3.4994913530010174e-05,
+      "loss": 0.4563,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 20.510534286499023,
+      "learning_rate": 3.491013903017972e-05,
+      "loss": 1.1467,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 55.81074905395508,
+      "learning_rate": 3.482536453034927e-05,
+      "loss": 0.6068,
+      "step": 2290
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 29.32077407836914,
+      "learning_rate": 3.4740590030518824e-05,
+      "loss": 1.0515,
+      "step": 2300
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 21.01796531677246,
+      "learning_rate": 3.465581553068837e-05,
+      "loss": 0.8702,
+      "step": 2310
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.176788330078125,
+      "learning_rate": 3.457104103085792e-05,
+      "loss": 0.7961,
+      "step": 2320
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 33.97718048095703,
+      "learning_rate": 3.448626653102747e-05,
+      "loss": 0.6178,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.07225427776575089,
+      "learning_rate": 3.440149203119702e-05,
+      "loss": 0.7895,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7126123905181885,
+      "learning_rate": 3.4316717531366566e-05,
+      "loss": 0.8879,
+      "step": 2350
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 60.54833221435547,
+      "learning_rate": 3.423194303153612e-05,
+      "loss": 0.8868,
+      "step": 2360
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 14.958369255065918,
+      "learning_rate": 3.4147168531705664e-05,
+      "loss": 0.9596,
+      "step": 2370
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.474311351776123,
+      "learning_rate": 3.406239403187521e-05,
+      "loss": 0.8218,
+      "step": 2380
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.398329734802246,
+      "learning_rate": 3.397761953204476e-05,
+      "loss": 0.7026,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 71.55179595947266,
+      "learning_rate": 3.389284503221431e-05,
+      "loss": 0.8114,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.973948001861572,
+      "learning_rate": 3.380807053238386e-05,
+      "loss": 0.4706,
+      "step": 2410
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1762962341308594,
+      "learning_rate": 3.3723296032553406e-05,
+      "loss": 0.5073,
+      "step": 2420
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 38.53884506225586,
+      "learning_rate": 3.363852153272296e-05,
+      "loss": 0.6444,
+      "step": 2430
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 51.29215621948242,
+      "learning_rate": 3.355374703289251e-05,
+      "loss": 0.5418,
+      "step": 2440
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8000295758247375,
+      "learning_rate": 3.3468972533062056e-05,
+      "loss": 0.558,
+      "step": 2450
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 33.07465744018555,
+      "learning_rate": 3.338419803323161e-05,
+      "loss": 1.4676,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 9.931634902954102,
+      "learning_rate": 3.3299423533401155e-05,
+      "loss": 1.3777,
+      "step": 2470
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 18.494661331176758,
+      "learning_rate": 3.321464903357071e-05,
+      "loss": 0.6258,
+      "step": 2480
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.16553539037704468,
+      "learning_rate": 3.312987453374025e-05,
+      "loss": 0.964,
+      "step": 2490
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 55.56105041503906,
+      "learning_rate": 3.3045100033909805e-05,
+      "loss": 0.4535,
+      "step": 2500
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 70.11732482910156,
+      "learning_rate": 3.296032553407935e-05,
+      "loss": 0.7698,
+      "step": 2510
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 58.0792121887207,
+      "learning_rate": 3.2875551034248896e-05,
+      "loss": 0.98,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 15.220126152038574,
+      "learning_rate": 3.279077653441845e-05,
+      "loss": 0.9123,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 13.152356147766113,
+      "learning_rate": 3.2706002034587994e-05,
+      "loss": 0.7913,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 37.108360290527344,
+      "learning_rate": 3.262122753475755e-05,
+      "loss": 0.6322,
+      "step": 2550
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 133.0155029296875,
+      "learning_rate": 3.253645303492709e-05,
+      "loss": 0.8002,
+      "step": 2560
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.423397541046143,
+      "learning_rate": 3.2451678535096645e-05,
+      "loss": 0.6649,
+      "step": 2570
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 37.421443939208984,
+      "learning_rate": 3.236690403526619e-05,
+      "loss": 0.4575,
+      "step": 2580
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 20.839557647705078,
+      "learning_rate": 3.228212953543574e-05,
+      "loss": 0.9595,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 8.633934020996094,
+      "learning_rate": 3.2197355035605296e-05,
+      "loss": 0.8809,
+      "step": 2600
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 27.769481658935547,
+      "learning_rate": 3.211258053577484e-05,
+      "loss": 0.9243,
+      "step": 2610
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13876375555992126,
+      "learning_rate": 3.2027806035944394e-05,
+      "loss": 0.5838,
+      "step": 2620
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 50.39765548706055,
+      "learning_rate": 3.194303153611394e-05,
+      "loss": 0.4513,
+      "step": 2630
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 20.293073654174805,
+      "learning_rate": 3.185825703628349e-05,
+      "loss": 0.2965,
+      "step": 2640
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.49339720606803894,
+      "learning_rate": 3.177348253645304e-05,
+      "loss": 0.3329,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 8.260361671447754,
+      "learning_rate": 3.168870803662258e-05,
+      "loss": 0.8128,
+      "step": 2660
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.9407100677490234,
+      "learning_rate": 3.1603933536792135e-05,
+      "loss": 0.1281,
+      "step": 2670
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 51.073455810546875,
+      "learning_rate": 3.151915903696168e-05,
+      "loss": 0.7543,
+      "step": 2680
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.05772038921713829,
+      "learning_rate": 3.1434384537131234e-05,
+      "loss": 0.8038,
+      "step": 2690
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.772864818572998,
+      "learning_rate": 3.134961003730078e-05,
+      "loss": 1.0692,
+      "step": 2700
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 29.10246467590332,
+      "learning_rate": 3.126483553747033e-05,
+      "loss": 1.0786,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 74.27193450927734,
+      "learning_rate": 3.118006103763988e-05,
+      "loss": 0.3927,
+      "step": 2720
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.654672622680664,
+      "learning_rate": 3.109528653780942e-05,
+      "loss": 0.0392,
+      "step": 2730
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.030826814472675323,
+      "learning_rate": 3.1010512037978975e-05,
+      "loss": 0.5397,
+      "step": 2740
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.03806144371628761,
+      "learning_rate": 3.092573753814853e-05,
+      "loss": 0.5454,
+      "step": 2750
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1753628253936768,
+      "learning_rate": 3.084096303831808e-05,
+      "loss": 0.7065,
+      "step": 2760
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 98.3718490600586,
+      "learning_rate": 3.0756188538487626e-05,
+      "loss": 0.5413,
+      "step": 2770
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 47.713836669921875,
+      "learning_rate": 3.067141403865718e-05,
+      "loss": 0.6828,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.499966144561768,
+      "learning_rate": 3.0586639538826724e-05,
+      "loss": 1.1348,
+      "step": 2790
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6247652769088745,
+      "learning_rate": 3.050186503899627e-05,
+      "loss": 0.6771,
+      "step": 2800
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.8232038021087646,
+      "learning_rate": 3.0417090539165822e-05,
+      "loss": 0.1384,
+      "step": 2810
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 16.047853469848633,
+      "learning_rate": 3.0332316039335368e-05,
+      "loss": 0.3357,
+      "step": 2820
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 22.86617660522461,
+      "learning_rate": 3.024754153950492e-05,
+      "loss": 0.1991,
+      "step": 2830
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 81.28910064697266,
+      "learning_rate": 3.0162767039674466e-05,
+      "loss": 0.4485,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2680492699146271,
+      "learning_rate": 3.007799253984402e-05,
+      "loss": 0.83,
+      "step": 2850
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.05300199240446091,
+      "learning_rate": 2.9993218040013564e-05,
+      "loss": 0.5507,
+      "step": 2860
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 73.55166625976562,
+      "learning_rate": 2.9908443540183113e-05,
+      "loss": 1.6387,
+      "step": 2870
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 50.422142028808594,
+      "learning_rate": 2.9823669040352665e-05,
+      "loss": 0.5293,
+      "step": 2880
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 25.55157470703125,
+      "learning_rate": 2.973889454052221e-05,
+      "loss": 1.0067,
+      "step": 2890
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4120949506759644,
+      "learning_rate": 2.9654120040691764e-05,
+      "loss": 1.3009,
+      "step": 2900
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 15.355062484741211,
+      "learning_rate": 2.956934554086131e-05,
+      "loss": 0.843,
+      "step": 2910
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4958977997303009,
+      "learning_rate": 2.948457104103086e-05,
+      "loss": 0.6706,
+      "step": 2920
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.02339070662856102,
+      "learning_rate": 2.9399796541200407e-05,
+      "loss": 0.5322,
+      "step": 2930
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 41.17433547973633,
+      "learning_rate": 2.9315022041369956e-05,
+      "loss": 0.5084,
+      "step": 2940
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.185633897781372,
+      "learning_rate": 2.923024754153951e-05,
+      "loss": 0.4755,
+      "step": 2950
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 53.46243667602539,
+      "learning_rate": 2.9145473041709054e-05,
+      "loss": 0.7402,
+      "step": 2960
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.203493118286133,
+      "learning_rate": 2.9060698541878607e-05,
+      "loss": 0.6368,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 8.445558547973633,
+      "learning_rate": 2.8975924042048153e-05,
+      "loss": 0.3933,
+      "step": 2980
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 44.631771087646484,
+      "learning_rate": 2.8891149542217705e-05,
+      "loss": 1.0651,
+      "step": 2990
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 26.084287643432617,
+      "learning_rate": 2.880637504238725e-05,
+      "loss": 1.0695,
+      "step": 3000
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2506965398788452,
+      "learning_rate": 2.8721600542556796e-05,
+      "loss": 0.2011,
+      "step": 3010
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 25.166454315185547,
+      "learning_rate": 2.863682604272635e-05,
+      "loss": 0.756,
+      "step": 3020
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 26.68798065185547,
+      "learning_rate": 2.8552051542895898e-05,
+      "loss": 1.0222,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 61.67727279663086,
+      "learning_rate": 2.846727704306545e-05,
+      "loss": 0.6143,
+      "step": 3040
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.026347124949097633,
+      "learning_rate": 2.8382502543234996e-05,
+      "loss": 0.6298,
+      "step": 3050
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 70.15167999267578,
+      "learning_rate": 2.829772804340455e-05,
+      "loss": 1.0136,
+      "step": 3060
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 75.8296127319336,
+      "learning_rate": 2.8212953543574094e-05,
+      "loss": 0.8159,
+      "step": 3070
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.20398695766925812,
+      "learning_rate": 2.812817904374364e-05,
+      "loss": 0.8516,
+      "step": 3080
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 95.35974884033203,
+      "learning_rate": 2.8043404543913192e-05,
+      "loss": 1.397,
+      "step": 3090
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2767679691314697,
+      "learning_rate": 2.795863004408274e-05,
+      "loss": 0.1996,
+      "step": 3100
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 28.711078643798828,
+      "learning_rate": 2.7873855544252294e-05,
+      "loss": 0.6535,
+      "step": 3110
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 41.65683364868164,
+      "learning_rate": 2.778908104442184e-05,
+      "loss": 0.5639,
+      "step": 3120
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6953363418579102,
+      "learning_rate": 2.770430654459139e-05,
+      "loss": 0.4564,
+      "step": 3130
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.09095999598503113,
+      "learning_rate": 2.7619532044760937e-05,
+      "loss": 0.4993,
+      "step": 3140
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 32.08039093017578,
+      "learning_rate": 2.7534757544930483e-05,
+      "loss": 0.9855,
+      "step": 3150
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.22207607328891754,
+      "learning_rate": 2.7449983045100035e-05,
+      "loss": 0.5325,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.050914715975522995,
+      "learning_rate": 2.736520854526958e-05,
+      "loss": 0.5084,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8935670852661133,
+      "learning_rate": 2.7280434045439134e-05,
+      "loss": 0.6757,
+      "step": 3180
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 19.006797790527344,
+      "learning_rate": 2.7195659545608683e-05,
+      "loss": 0.6753,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 15.514384269714355,
+      "learning_rate": 2.7110885045778235e-05,
+      "loss": 0.4333,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 9.05042552947998,
+      "learning_rate": 2.702611054594778e-05,
+      "loss": 0.681,
+      "step": 3210
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 59.737754821777344,
+      "learning_rate": 2.6941336046117326e-05,
+      "loss": 1.1243,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 55.89622116088867,
+      "learning_rate": 2.685656154628688e-05,
+      "loss": 0.4998,
+      "step": 3230
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.33736708760261536,
+      "learning_rate": 2.6771787046456424e-05,
+      "loss": 0.5959,
+      "step": 3240
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 49.71003341674805,
+      "learning_rate": 2.6687012546625977e-05,
+      "loss": 0.869,
+      "step": 3250
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.18244197964668274,
+      "learning_rate": 2.6602238046795526e-05,
+      "loss": 0.5476,
+      "step": 3260
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 57.65730285644531,
+      "learning_rate": 2.651746354696508e-05,
+      "loss": 0.8371,
+      "step": 3270
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 8.13063907623291,
+      "learning_rate": 2.6432689047134624e-05,
+      "loss": 0.3608,
+      "step": 3280
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2764989733695984,
+      "learning_rate": 2.634791454730417e-05,
+      "loss": 0.326,
+      "step": 3290
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.09723484516143799,
+      "learning_rate": 2.6263140047473722e-05,
+      "loss": 0.4903,
+      "step": 3300
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 90.19970703125,
+      "learning_rate": 2.6178365547643268e-05,
+      "loss": 0.5478,
+      "step": 3310
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.756451964378357,
+      "learning_rate": 2.609359104781282e-05,
+      "loss": 0.3005,
+      "step": 3320
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 95.1401596069336,
+      "learning_rate": 2.600881654798237e-05,
+      "loss": 0.3226,
+      "step": 3330
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.05358866602182388,
+      "learning_rate": 2.5924042048151918e-05,
+      "loss": 0.142,
+      "step": 3340
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 47.20892333984375,
+      "learning_rate": 2.5839267548321467e-05,
+      "loss": 0.8748,
+      "step": 3350
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.11273138970136642,
+      "learning_rate": 2.5754493048491013e-05,
+      "loss": 0.7466,
+      "step": 3360
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 11.685613632202148,
+      "learning_rate": 2.5669718548660565e-05,
+      "loss": 0.309,
+      "step": 3370
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3904511332511902,
+      "learning_rate": 2.558494404883011e-05,
+      "loss": 0.4282,
+      "step": 3380
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14008870720863342,
+      "learning_rate": 2.5500169548999664e-05,
+      "loss": 0.34,
+      "step": 3390
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 74.25190734863281,
+      "learning_rate": 2.541539504916921e-05,
+      "loss": 0.7807,
+      "step": 3400
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 50.943485260009766,
+      "learning_rate": 2.533062054933876e-05,
+      "loss": 1.5617,
+      "step": 3410
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.12540976703166962,
+      "learning_rate": 2.524584604950831e-05,
+      "loss": 0.5259,
+      "step": 3420
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 23.478382110595703,
+      "learning_rate": 2.5161071549677856e-05,
+      "loss": 0.6975,
+      "step": 3430
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.06014597788453102,
+      "learning_rate": 2.507629704984741e-05,
+      "loss": 0.4681,
+      "step": 3440
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8615859150886536,
+      "learning_rate": 2.4991522550016954e-05,
+      "loss": 0.0536,
+      "step": 3450
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.489899218082428,
+      "learning_rate": 2.4906748050186503e-05,
+      "loss": 0.9952,
+      "step": 3460
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.645602226257324,
+      "learning_rate": 2.4821973550356053e-05,
+      "loss": 0.4834,
+      "step": 3470
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.08961528539657593,
+      "learning_rate": 2.47371990505256e-05,
+      "loss": 0.4156,
+      "step": 3480
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 39.562774658203125,
+      "learning_rate": 2.4652424550695154e-05,
+      "loss": 0.7731,
+      "step": 3490
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 41.32746124267578,
+      "learning_rate": 2.4567650050864703e-05,
+      "loss": 0.4151,
+      "step": 3500
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 19.19269561767578,
+      "learning_rate": 2.4482875551034252e-05,
+      "loss": 1.0375,
+      "step": 3510
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 7.726545333862305,
+      "learning_rate": 2.4398101051203798e-05,
+      "loss": 0.3077,
+      "step": 3520
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 12.059285163879395,
+      "learning_rate": 2.4313326551373347e-05,
+      "loss": 0.935,
+      "step": 3530
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 19.483482360839844,
+      "learning_rate": 2.4228552051542896e-05,
+      "loss": 0.2129,
+      "step": 3540
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.06684020906686783,
+      "learning_rate": 2.4143777551712445e-05,
+      "loss": 0.3169,
+      "step": 3550
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 44.86419677734375,
+      "learning_rate": 2.4059003051881994e-05,
+      "loss": 1.2605,
+      "step": 3560
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.09264612197876,
+      "learning_rate": 2.3974228552051546e-05,
+      "loss": 0.7057,
+      "step": 3570
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.04498284310102463,
+      "learning_rate": 2.3889454052221095e-05,
+      "loss": 0.9476,
+      "step": 3580
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 25.088247299194336,
+      "learning_rate": 2.380467955239064e-05,
+      "loss": 0.9267,
+      "step": 3590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 45.36594772338867,
+      "learning_rate": 2.371990505256019e-05,
+      "loss": 0.7207,
+      "step": 3600
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.7207283973693848,
+      "learning_rate": 2.363513055272974e-05,
+      "loss": 0.7053,
+      "step": 3610
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 42.47288131713867,
+      "learning_rate": 2.3550356052899288e-05,
+      "loss": 0.2621,
+      "step": 3620
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 89.94295501708984,
+      "learning_rate": 2.3465581553068837e-05,
+      "loss": 0.328,
+      "step": 3630
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.10006561875343323,
+      "learning_rate": 2.3380807053238386e-05,
+      "loss": 0.8296,
+      "step": 3640
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 89.21922302246094,
+      "learning_rate": 2.329603255340794e-05,
+      "loss": 0.9074,
+      "step": 3650
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.014263429678976536,
+      "learning_rate": 2.3211258053577484e-05,
+      "loss": 0.5238,
+      "step": 3660
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.2213358879089355,
+      "learning_rate": 2.3126483553747033e-05,
+      "loss": 0.4163,
+      "step": 3670
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.10971381515264511,
+      "learning_rate": 2.3041709053916582e-05,
+      "loss": 0.4396,
+      "step": 3680
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9370849132537842,
+      "learning_rate": 2.295693455408613e-05,
+      "loss": 0.9562,
+      "step": 3690
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.26476842164993286,
+      "learning_rate": 2.287216005425568e-05,
+      "loss": 0.3369,
+      "step": 3700
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.879600524902344,
+      "learning_rate": 2.278738555442523e-05,
+      "loss": 0.4717,
+      "step": 3710
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 91.1629867553711,
+      "learning_rate": 2.270261105459478e-05,
+      "loss": 0.4292,
+      "step": 3720
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3384571075439453,
+      "learning_rate": 2.2617836554764328e-05,
+      "loss": 0.4184,
+      "step": 3730
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.09564901143312454,
+      "learning_rate": 2.2533062054933877e-05,
+      "loss": 0.3908,
+      "step": 3740
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1722471714019775,
+      "learning_rate": 2.2448287555103426e-05,
+      "loss": 0.5939,
+      "step": 3750
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 86.56784057617188,
+      "learning_rate": 2.2363513055272975e-05,
+      "loss": 0.6253,
+      "step": 3760
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4434572756290436,
+      "learning_rate": 2.2278738555442524e-05,
+      "loss": 0.2656,
+      "step": 3770
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3613429069519043,
+      "learning_rate": 2.2193964055612073e-05,
+      "loss": 0.5376,
+      "step": 3780
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 39.03946304321289,
+      "learning_rate": 2.2109189555781622e-05,
+      "loss": 0.5256,
+      "step": 3790
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 33.69869613647461,
+      "learning_rate": 2.202441505595117e-05,
+      "loss": 1.9483,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.014940350316464901,
+      "learning_rate": 2.193964055612072e-05,
+      "loss": 0.2437,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5131822824478149,
+      "learning_rate": 2.185486605629027e-05,
+      "loss": 0.6613,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8719220161437988,
+      "learning_rate": 2.1770091556459818e-05,
+      "loss": 1.1681,
+      "step": 3830
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.0348847322165966,
+      "learning_rate": 2.1685317056629367e-05,
+      "loss": 0.4557,
+      "step": 3840
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1149950921535492,
+      "learning_rate": 2.1600542556798916e-05,
+      "loss": 0.2578,
+      "step": 3850
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.825486660003662,
+      "learning_rate": 2.1515768056968465e-05,
+      "loss": 0.5987,
+      "step": 3860
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 24.669727325439453,
+      "learning_rate": 2.1430993557138014e-05,
+      "loss": 0.4292,
+      "step": 3870
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8084043264389038,
+      "learning_rate": 2.1346219057307563e-05,
+      "loss": 0.4102,
+      "step": 3880
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 73.08427429199219,
+      "learning_rate": 2.1261444557477112e-05,
+      "loss": 0.9116,
+      "step": 3890
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 30.825637817382812,
+      "learning_rate": 2.117667005764666e-05,
+      "loss": 0.5745,
+      "step": 3900
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 102.85230255126953,
+      "learning_rate": 2.109189555781621e-05,
+      "loss": 0.2618,
+      "step": 3910
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.03866774961352348,
+      "learning_rate": 2.100712105798576e-05,
+      "loss": 0.8069,
+      "step": 3920
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1322760581970215,
+      "learning_rate": 2.092234655815531e-05,
+      "loss": 0.3481,
+      "step": 3930
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.014790826477110386,
+      "learning_rate": 2.0837572058324854e-05,
+      "loss": 0.4006,
+      "step": 3940
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.11268659681081772,
+      "learning_rate": 2.0752797558494407e-05,
+      "loss": 0.4533,
+      "step": 3950
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0567173957824707,
+      "learning_rate": 2.0668023058663956e-05,
+      "loss": 0.4536,
+      "step": 3960
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 8.341649055480957,
+      "learning_rate": 2.0583248558833505e-05,
+      "loss": 0.6103,
+      "step": 3970
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.268702745437622,
+      "learning_rate": 2.0498474059003054e-05,
+      "loss": 0.3091,
+      "step": 3980
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 8.767001152038574,
+      "learning_rate": 2.0413699559172603e-05,
+      "loss": 0.1458,
+      "step": 3990
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 109.38581848144531,
+      "learning_rate": 2.0328925059342152e-05,
+      "loss": 0.6205,
+      "step": 4000
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.01923774927854538,
+      "learning_rate": 2.0244150559511698e-05,
+      "loss": 0.5022,
+      "step": 4010
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.962080955505371,
+      "learning_rate": 2.0159376059681247e-05,
+      "loss": 0.0514,
+      "step": 4020
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.028890101239085197,
+      "learning_rate": 2.00746015598508e-05,
+      "loss": 1.4382,
+      "step": 4030
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.056723129004240036,
+      "learning_rate": 1.9989827060020348e-05,
+      "loss": 0.406,
+      "step": 4040
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6163440942764282,
+      "learning_rate": 1.9905052560189897e-05,
+      "loss": 0.0346,
+      "step": 4050
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.029935460537672043,
+      "learning_rate": 1.9820278060359446e-05,
+      "loss": 0.3818,
+      "step": 4060
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2080775797367096,
+      "learning_rate": 1.9735503560528995e-05,
+      "loss": 0.3822,
+      "step": 4070
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 26.54397201538086,
+      "learning_rate": 1.965072906069854e-05,
+      "loss": 0.3972,
+      "step": 4080
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1706455647945404,
+      "learning_rate": 1.956595456086809e-05,
+      "loss": 0.8122,
+      "step": 4090
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 119.8738021850586,
+      "learning_rate": 1.948118006103764e-05,
+      "loss": 0.5582,
+      "step": 4100
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.742654323577881,
+      "learning_rate": 1.939640556120719e-05,
+      "loss": 0.6886,
+      "step": 4110
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 42.857357025146484,
+      "learning_rate": 1.931163106137674e-05,
+      "loss": 0.1316,
+      "step": 4120
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 44.45591735839844,
+      "learning_rate": 1.922685656154629e-05,
+      "loss": 0.7597,
+      "step": 4130
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.140688180923462,
+      "learning_rate": 1.914208206171584e-05,
+      "loss": 0.6811,
+      "step": 4140
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 14.241903305053711,
+      "learning_rate": 1.9057307561885384e-05,
+      "loss": 0.9585,
+      "step": 4150
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 75.1405029296875,
+      "learning_rate": 1.8972533062054933e-05,
+      "loss": 0.8658,
+      "step": 4160
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.019668349996209145,
+      "learning_rate": 1.8887758562224482e-05,
+      "loss": 0.2094,
+      "step": 4170
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 24.360675811767578,
+      "learning_rate": 1.880298406239403e-05,
+      "loss": 0.8209,
+      "step": 4180
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 58.54786682128906,
+      "learning_rate": 1.8718209562563584e-05,
+      "loss": 1.3557,
+      "step": 4190
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.059292688965797424,
+      "learning_rate": 1.8633435062733133e-05,
+      "loss": 0.4244,
+      "step": 4200
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 19.321937561035156,
+      "learning_rate": 1.8548660562902682e-05,
+      "loss": 0.4429,
+      "step": 4210
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 22.25551986694336,
+      "learning_rate": 1.8463886063072228e-05,
+      "loss": 0.2562,
+      "step": 4220
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 21.07433319091797,
+      "learning_rate": 1.8379111563241777e-05,
+      "loss": 0.4458,
+      "step": 4230
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.06258882582187653,
+      "learning_rate": 1.8294337063411326e-05,
+      "loss": 0.7059,
+      "step": 4240
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 97.818603515625,
+      "learning_rate": 1.8209562563580875e-05,
+      "loss": 1.6226,
+      "step": 4250
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 64.59729766845703,
+      "learning_rate": 1.8124788063750424e-05,
+      "loss": 0.7478,
+      "step": 4260
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 10.768301963806152,
+      "learning_rate": 1.8040013563919976e-05,
+      "loss": 0.3018,
+      "step": 4270
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 125.17046356201172,
+      "learning_rate": 1.7955239064089525e-05,
+      "loss": 0.5633,
+      "step": 4280
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 20.224864959716797,
+      "learning_rate": 1.787046456425907e-05,
+      "loss": 0.7404,
+      "step": 4290
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 43.46826171875,
+      "learning_rate": 1.778569006442862e-05,
+      "loss": 1.3683,
+      "step": 4300
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 26.397693634033203,
+      "learning_rate": 1.770091556459817e-05,
+      "loss": 0.7154,
+      "step": 4310
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 70.86721801757812,
+      "learning_rate": 1.7616141064767718e-05,
+      "loss": 0.3547,
+      "step": 4320
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 7.348357200622559,
+      "learning_rate": 1.7531366564937267e-05,
+      "loss": 0.3216,
+      "step": 4330
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.027426056563854218,
+      "learning_rate": 1.7446592065106816e-05,
+      "loss": 0.2455,
+      "step": 4340
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 16.937259674072266,
+      "learning_rate": 1.736181756527637e-05,
+      "loss": 0.467,
+      "step": 4350
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8604963421821594,
+      "learning_rate": 1.7277043065445914e-05,
+      "loss": 0.4664,
+      "step": 4360
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 76.17068481445312,
+      "learning_rate": 1.7192268565615463e-05,
+      "loss": 0.1975,
+      "step": 4370
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 11.509053230285645,
+      "learning_rate": 1.7107494065785012e-05,
+      "loss": 0.6729,
+      "step": 4380
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.039980944246053696,
+      "learning_rate": 1.702271956595456e-05,
+      "loss": 0.5038,
+      "step": 4390
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.07167255878448486,
+      "learning_rate": 1.693794506612411e-05,
+      "loss": 0.1302,
+      "step": 4400
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.1734821796417236,
+      "learning_rate": 1.685317056629366e-05,
+      "loss": 0.2821,
+      "step": 4410
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.076037883758545,
+      "learning_rate": 1.676839606646321e-05,
+      "loss": 0.55,
+      "step": 4420
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.02169458009302616,
+      "learning_rate": 1.6683621566632758e-05,
+      "loss": 0.0048,
+      "step": 4430
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.759131669998169,
+      "learning_rate": 1.6598847066802307e-05,
+      "loss": 0.8726,
+      "step": 4440
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5331248641014099,
+      "learning_rate": 1.6514072566971856e-05,
+      "loss": 0.6762,
+      "step": 4450
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 22.186798095703125,
+      "learning_rate": 1.6429298067141405e-05,
+      "loss": 0.6426,
+      "step": 4460
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 32.538089752197266,
+      "learning_rate": 1.6344523567310954e-05,
+      "loss": 0.9122,
+      "step": 4470
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.58526086807251,
+      "learning_rate": 1.6259749067480503e-05,
+      "loss": 0.3097,
+      "step": 4480
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.05092554911971092,
+      "learning_rate": 1.6174974567650052e-05,
+      "loss": 0.0069,
+      "step": 4490
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 41.92851638793945,
+      "learning_rate": 1.60902000678196e-05,
+      "loss": 0.9276,
+      "step": 4500
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 152.6699676513672,
+      "learning_rate": 1.600542556798915e-05,
+      "loss": 0.4752,
+      "step": 4510
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 41.94477462768555,
+      "learning_rate": 1.59206510681587e-05,
+      "loss": 0.2672,
+      "step": 4520
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 57.44374084472656,
+      "learning_rate": 1.5835876568328248e-05,
+      "loss": 0.9764,
+      "step": 4530
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 23.202451705932617,
+      "learning_rate": 1.5751102068497797e-05,
+      "loss": 0.4835,
+      "step": 4540
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 8.41569709777832,
+      "learning_rate": 1.5666327568667346e-05,
+      "loss": 0.3972,
+      "step": 4550
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2448299527168274,
+      "learning_rate": 1.5581553068836892e-05,
+      "loss": 0.3595,
+      "step": 4560
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.009353713132441044,
+      "learning_rate": 1.5496778569006444e-05,
+      "loss": 0.6254,
+      "step": 4570
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.40498995780944824,
+      "learning_rate": 1.5412004069175993e-05,
+      "loss": 0.4135,
+      "step": 4580
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1234465166926384,
+      "learning_rate": 1.5327229569345542e-05,
+      "loss": 0.7479,
+      "step": 4590
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 15.938213348388672,
+      "learning_rate": 1.5242455069515091e-05,
+      "loss": 0.9297,
+      "step": 4600
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.941218376159668,
+      "learning_rate": 1.515768056968464e-05,
+      "loss": 1.0416,
+      "step": 4610
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.810397148132324,
+      "learning_rate": 1.507290606985419e-05,
+      "loss": 0.2707,
+      "step": 4620
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.005552912130951881,
+      "learning_rate": 1.4988131570023737e-05,
+      "loss": 0.6025,
+      "step": 4630
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.059763405472040176,
+      "learning_rate": 1.4903357070193286e-05,
+      "loss": 0.018,
+      "step": 4640
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.09440121799707413,
+      "learning_rate": 1.4818582570362835e-05,
+      "loss": 0.3638,
+      "step": 4650
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.17027807235717773,
+      "learning_rate": 1.4733808070532384e-05,
+      "loss": 0.3194,
+      "step": 4660
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.010031620971858501,
+      "learning_rate": 1.4649033570701933e-05,
+      "loss": 0.1356,
+      "step": 4670
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.03452069312334061,
+      "learning_rate": 1.4564259070871484e-05,
+      "loss": 0.4853,
+      "step": 4680
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.010452840477228165,
+      "learning_rate": 1.4479484571041033e-05,
+      "loss": 0.5158,
+      "step": 4690
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 9.103416442871094,
+      "learning_rate": 1.4394710071210579e-05,
+      "loss": 0.9832,
+      "step": 4700
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 26.68756103515625,
+      "learning_rate": 1.430993557138013e-05,
+      "loss": 0.8755,
+      "step": 4710
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 165.71353149414062,
+      "learning_rate": 1.4225161071549678e-05,
+      "loss": 1.1972,
+      "step": 4720
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 47.88841247558594,
+      "learning_rate": 1.4140386571719227e-05,
+      "loss": 0.335,
+      "step": 4730
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.27552178502082825,
+      "learning_rate": 1.4055612071888776e-05,
+      "loss": 0.9269,
+      "step": 4740
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 18.1062068939209,
+      "learning_rate": 1.3970837572058325e-05,
+      "loss": 0.2469,
+      "step": 4750
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.01228923350572586,
+      "learning_rate": 1.3886063072227876e-05,
+      "loss": 0.4695,
+      "step": 4760
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.010787318460643291,
+      "learning_rate": 1.3801288572397422e-05,
+      "loss": 0.0921,
+      "step": 4770
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.013667989522218704,
+      "learning_rate": 1.3716514072566971e-05,
+      "loss": 0.136,
+      "step": 4780
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.07841726392507553,
+      "learning_rate": 1.3631739572736522e-05,
+      "loss": 0.3449,
+      "step": 4790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.020194262266159058,
+      "learning_rate": 1.354696507290607e-05,
+      "loss": 0.1784,
+      "step": 4800
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 69.3423843383789,
+      "learning_rate": 1.346219057307562e-05,
+      "loss": 0.6792,
+      "step": 4810
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.046759914606809616,
+      "learning_rate": 1.3377416073245169e-05,
+      "loss": 0.1271,
+      "step": 4820
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.011114859953522682,
+      "learning_rate": 1.3292641573414718e-05,
+      "loss": 0.2728,
+      "step": 4830
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 32.39384841918945,
+      "learning_rate": 1.3207867073584265e-05,
+      "loss": 0.337,
+      "step": 4840
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 45.9520263671875,
+      "learning_rate": 1.3123092573753814e-05,
+      "loss": 0.3638,
+      "step": 4850
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.45650675892829895,
+      "learning_rate": 1.3038318073923363e-05,
+      "loss": 0.4145,
+      "step": 4860
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.05143202096223831,
+      "learning_rate": 1.2953543574092914e-05,
+      "loss": 1.0756,
+      "step": 4870
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.376939296722412,
+      "learning_rate": 1.2868769074262463e-05,
+      "loss": 0.0657,
+      "step": 4880
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 25.659818649291992,
+      "learning_rate": 1.2783994574432012e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.00974774919450283,
+      "learning_rate": 1.2699220074601561e-05,
+      "loss": 0.136,
+      "step": 4900
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.016076885163784027,
+      "learning_rate": 1.2614445574771109e-05,
+      "loss": 0.6645,
+      "step": 4910
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.16606609523296356,
+      "learning_rate": 1.2529671074940658e-05,
+      "loss": 0.7139,
+      "step": 4920
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 70.56269836425781,
+      "learning_rate": 1.2444896575110207e-05,
+      "loss": 0.0413,
+      "step": 4930
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.15991821885108948,
+      "learning_rate": 1.2360122075279756e-05,
+      "loss": 0.8144,
+      "step": 4940
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9377095699310303,
+      "learning_rate": 1.2275347575449306e-05,
+      "loss": 0.489,
+      "step": 4950
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.03479115292429924,
+      "learning_rate": 1.2190573075618854e-05,
+      "loss": 0.7886,
+      "step": 4960
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.03444279730319977,
+      "learning_rate": 1.2105798575788403e-05,
+      "loss": 0.5328,
+      "step": 4970
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.0390845350921154,
+      "learning_rate": 1.2021024075957952e-05,
+      "loss": 0.4317,
+      "step": 4980
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3982301652431488,
+      "learning_rate": 1.1936249576127503e-05,
+      "loss": 1.1167,
+      "step": 4990
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.10968449711799622,
+      "learning_rate": 1.185147507629705e-05,
+      "loss": 0.4359,
+      "step": 5000
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.02797958254814148,
+      "learning_rate": 1.1766700576466599e-05,
+      "loss": 0.0039,
+      "step": 5010
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.00903019867837429,
+      "learning_rate": 1.1681926076636148e-05,
+      "loss": 0.3334,
+      "step": 5020
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.15302762389183044,
+      "learning_rate": 1.1597151576805697e-05,
+      "loss": 0.4842,
+      "step": 5030
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.855728030204773,
+      "learning_rate": 1.1512377076975246e-05,
+      "loss": 0.0593,
+      "step": 5040
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.005731168668717146,
+      "learning_rate": 1.1427602577144795e-05,
+      "loss": 0.2487,
+      "step": 5050
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14197255671024323,
+      "learning_rate": 1.1342828077314344e-05,
+      "loss": 0.3685,
+      "step": 5060
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 28.626758575439453,
+      "learning_rate": 1.1258053577483893e-05,
+      "loss": 0.1837,
+      "step": 5070
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.06634432077407837,
+      "learning_rate": 1.1173279077653442e-05,
+      "loss": 0.9408,
+      "step": 5080
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 57.522308349609375,
+      "learning_rate": 1.1088504577822991e-05,
+      "loss": 0.8385,
+      "step": 5090
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.012267855927348137,
+      "learning_rate": 1.100373007799254e-05,
+      "loss": 0.3534,
+      "step": 5100
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.008047958835959435,
+      "learning_rate": 1.091895557816209e-05,
+      "loss": 0.1124,
+      "step": 5110
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.02493928372859955,
+      "learning_rate": 1.0834181078331639e-05,
+      "loss": 0.9354,
+      "step": 5120
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 46.4794921875,
+      "learning_rate": 1.0749406578501188e-05,
+      "loss": 0.6517,
+      "step": 5130
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6745705604553223,
+      "learning_rate": 1.0664632078670737e-05,
+      "loss": 0.2617,
+      "step": 5140
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 52.141422271728516,
+      "learning_rate": 1.0579857578840286e-05,
+      "loss": 1.0205,
+      "step": 5150
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.015956485643982887,
+      "learning_rate": 1.0495083079009835e-05,
+      "loss": 0.0031,
+      "step": 5160
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.00936069618910551,
+      "learning_rate": 1.0410308579179382e-05,
+      "loss": 0.4062,
+      "step": 5170
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8617632389068604,
+      "learning_rate": 1.0325534079348933e-05,
+      "loss": 0.6192,
+      "step": 5180
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 44.38825988769531,
+      "learning_rate": 1.0240759579518482e-05,
+      "loss": 0.9206,
+      "step": 5190
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07197415083646774,
+      "learning_rate": 1.0155985079688031e-05,
+      "loss": 0.5991,
+      "step": 5200
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07943376153707504,
+      "learning_rate": 1.0071210579857578e-05,
+      "loss": 0.7726,
+      "step": 5210
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.014867642894387245,
+      "learning_rate": 9.986436080027129e-06,
+      "loss": 0.4536,
+      "step": 5220
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3331596851348877,
+      "learning_rate": 9.901661580196678e-06,
+      "loss": 0.1273,
+      "step": 5230
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.49268820881843567,
+      "learning_rate": 9.816887080366225e-06,
+      "loss": 0.1124,
+      "step": 5240
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 50.03787612915039,
+      "learning_rate": 9.732112580535774e-06,
+      "loss": 0.403,
+      "step": 5250
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 65.66496276855469,
+      "learning_rate": 9.647338080705325e-06,
+      "loss": 0.3994,
+      "step": 5260
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 21.710723876953125,
+      "learning_rate": 9.562563580874874e-06,
+      "loss": 0.2621,
+      "step": 5270
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.0246584489941597,
+      "learning_rate": 9.477789081044422e-06,
+      "loss": 0.4591,
+      "step": 5280
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 60.308345794677734,
+      "learning_rate": 9.39301458121397e-06,
+      "loss": 0.0449,
+      "step": 5290
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 136.57911682128906,
+      "learning_rate": 9.308240081383521e-06,
+      "loss": 0.6523,
+      "step": 5300
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.011092737317085266,
+      "learning_rate": 9.223465581553069e-06,
+      "loss": 0.5621,
+      "step": 5310
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 72.57099151611328,
+      "learning_rate": 9.138691081722618e-06,
+      "loss": 0.3213,
+      "step": 5320
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5914633274078369,
+      "learning_rate": 9.053916581892167e-06,
+      "loss": 0.5433,
+      "step": 5330
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.004101056605577469,
+      "learning_rate": 8.969142082061716e-06,
+      "loss": 0.1477,
+      "step": 5340
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14881229400634766,
+      "learning_rate": 8.884367582231265e-06,
+      "loss": 0.3333,
+      "step": 5350
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2352166622877121,
+      "learning_rate": 8.799593082400814e-06,
+      "loss": 0.474,
+      "step": 5360
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.024752633646130562,
+      "learning_rate": 8.714818582570363e-06,
+      "loss": 0.4311,
+      "step": 5370
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5896317958831787,
+      "learning_rate": 8.630044082739912e-06,
+      "loss": 0.5513,
+      "step": 5380
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.06998773664236069,
+      "learning_rate": 8.545269582909461e-06,
+      "loss": 0.2412,
+      "step": 5390
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.060008566826581955,
+      "learning_rate": 8.46049508307901e-06,
+      "loss": 0.541,
+      "step": 5400
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 130.00119018554688,
+      "learning_rate": 8.37572058324856e-06,
+      "loss": 0.4602,
+      "step": 5410
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 114.44886779785156,
+      "learning_rate": 8.290946083418108e-06,
+      "loss": 0.594,
+      "step": 5420
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1713523268699646,
+      "learning_rate": 8.206171583587657e-06,
+      "loss": 0.0291,
+      "step": 5430
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.02883375808596611,
+      "learning_rate": 8.121397083757206e-06,
+      "loss": 0.1052,
+      "step": 5440
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.028804084286093712,
+      "learning_rate": 8.036622583926755e-06,
+      "loss": 0.4604,
+      "step": 5450
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.535491704940796,
+      "learning_rate": 7.951848084096304e-06,
+      "loss": 0.3482,
+      "step": 5460
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.026217030361294746,
+      "learning_rate": 7.867073584265853e-06,
+      "loss": 0.2586,
+      "step": 5470
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3666705787181854,
+      "learning_rate": 7.7822990844354e-06,
+      "loss": 0.9029,
+      "step": 5480
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.017206447198987007,
+      "learning_rate": 7.697524584604952e-06,
+      "loss": 0.3287,
+      "step": 5490
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.011142526753246784,
+      "learning_rate": 7.612750084774501e-06,
+      "loss": 0.1899,
+      "step": 5500
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.032814279198646545,
+      "learning_rate": 7.52797558494405e-06,
+      "loss": 0.6027,
+      "step": 5510
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.121613502502441,
+      "learning_rate": 7.443201085113598e-06,
+      "loss": 0.4039,
+      "step": 5520
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.02050989866256714,
+      "learning_rate": 7.358426585283147e-06,
+      "loss": 1.1173,
+      "step": 5530
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 31.819725036621094,
+      "learning_rate": 7.273652085452697e-06,
+      "loss": 0.4358,
+      "step": 5540
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 62.192996978759766,
+      "learning_rate": 7.188877585622244e-06,
+      "loss": 0.2461,
+      "step": 5550
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.339864730834961,
+      "learning_rate": 7.104103085791794e-06,
+      "loss": 0.2663,
+      "step": 5560
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.04583854600787163,
+      "learning_rate": 7.019328585961343e-06,
+      "loss": 0.0101,
+      "step": 5570
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 30.54637336730957,
+      "learning_rate": 6.934554086130893e-06,
+      "loss": 0.6777,
+      "step": 5580
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 8.97301959991455,
+      "learning_rate": 6.84977958630044e-06,
+      "loss": 0.4932,
+      "step": 5590
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.008320887573063374,
+      "learning_rate": 6.76500508646999e-06,
+      "loss": 0.4379,
+      "step": 5600
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.00964405108243227,
+      "learning_rate": 6.680230586639539e-06,
+      "loss": 0.5022,
+      "step": 5610
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.38759058713912964,
+      "learning_rate": 6.5954560868090875e-06,
+      "loss": 0.26,
+      "step": 5620
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.011307528242468834,
+      "learning_rate": 6.5106815869786366e-06,
+      "loss": 0.1178,
+      "step": 5630
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.06669808179140091,
+      "learning_rate": 6.4259070871481864e-06,
+      "loss": 0.3514,
+      "step": 5640
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 130.25222778320312,
+      "learning_rate": 6.3411325873177355e-06,
+      "loss": 0.0678,
+      "step": 5650
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.02906797081232071,
+      "learning_rate": 6.256358087487284e-06,
+      "loss": 0.9587,
+      "step": 5660
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.06155577301979065,
+      "learning_rate": 6.171583587656833e-06,
+      "loss": 0.3508,
+      "step": 5670
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.007107898592948914,
+      "learning_rate": 6.086809087826382e-06,
+      "loss": 0.2734,
+      "step": 5680
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 72.81842803955078,
+      "learning_rate": 6.002034587995931e-06,
+      "loss": 0.2777,
+      "step": 5690
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.008903439156711102,
+      "learning_rate": 5.91726008816548e-06,
+      "loss": 0.3983,
+      "step": 5700
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.018965894356369972,
+      "learning_rate": 5.832485588335029e-06,
+      "loss": 0.1067,
+      "step": 5710
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.039035771042108536,
+      "learning_rate": 5.747711088504578e-06,
+      "loss": 0.7816,
+      "step": 5720
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 190.41064453125,
+      "learning_rate": 5.662936588674127e-06,
+      "loss": 0.3051,
+      "step": 5730
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.00861747283488512,
+      "learning_rate": 5.578162088843676e-06,
+      "loss": 0.0098,
+      "step": 5740
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 9.695987701416016,
+      "learning_rate": 5.493387589013225e-06,
+      "loss": 0.0252,
+      "step": 5750
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.0687270537018776,
+      "learning_rate": 5.408613089182774e-06,
+      "loss": 0.1273,
+      "step": 5760
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.01117786020040512,
+      "learning_rate": 5.323838589352323e-06,
+      "loss": 0.8719,
+      "step": 5770
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.017118388786911964,
+      "learning_rate": 5.239064089521872e-06,
+      "loss": 0.8036,
+      "step": 5780
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.834500789642334,
+      "learning_rate": 5.154289589691421e-06,
+      "loss": 0.6443,
+      "step": 5790
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.03857054561376572,
+      "learning_rate": 5.0695150898609695e-06,
+      "loss": 0.3097,
+      "step": 5800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.3266679048538208,
+      "learning_rate": 4.984740590030519e-06,
+      "loss": 0.5779,
+      "step": 5810
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.008399288170039654,
+      "learning_rate": 4.899966090200068e-06,
+      "loss": 0.3936,
+      "step": 5820
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 48.30754852294922,
+      "learning_rate": 4.8151915903696175e-06,
+      "loss": 0.747,
+      "step": 5830
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35246542096138,
+      "learning_rate": 4.730417090539166e-06,
+      "loss": 0.1861,
+      "step": 5840
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.10090816020965576,
+      "learning_rate": 4.645642590708716e-06,
+      "loss": 0.2213,
+      "step": 5850
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.081159830093384,
+      "learning_rate": 4.560868090878264e-06,
+      "loss": 0.3135,
+      "step": 5860
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 88.7223129272461,
+      "learning_rate": 4.476093591047813e-06,
+      "loss": 0.7422,
+      "step": 5870
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.028826922178268433,
+      "learning_rate": 4.391319091217362e-06,
+      "loss": 0.1471,
+      "step": 5880
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.006259507965296507,
+      "learning_rate": 4.306544591386911e-06,
+      "loss": 0.0344,
+      "step": 5890
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.010470997542142868,
+      "learning_rate": 4.22177009155646e-06,
+      "loss": 0.0023,
+      "step": 5900
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.0066988361068069935,
+      "learning_rate": 4.136995591726009e-06,
+      "loss": 0.0013,
+      "step": 5910
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.004910880699753761,
+      "learning_rate": 4.052221091895558e-06,
+      "loss": 0.1276,
+      "step": 5920
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 115.09222412109375,
+      "learning_rate": 3.967446592065107e-06,
+      "loss": 0.4087,
+      "step": 5930
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.25051459670066833,
+      "learning_rate": 3.882672092234655e-06,
+      "loss": 0.0014,
+      "step": 5940
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.038802146911621,
+      "learning_rate": 3.7978975924042052e-06,
+      "loss": 0.01,
+      "step": 5950
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.012906698510050774,
+      "learning_rate": 3.713123092573754e-06,
+      "loss": 0.0645,
+      "step": 5960
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 191.30227661132812,
+      "learning_rate": 3.6283485927433033e-06,
+      "loss": 0.4189,
+      "step": 5970
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 136.50938415527344,
+      "learning_rate": 3.543574092912852e-06,
+      "loss": 0.1618,
+      "step": 5980
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.011031740345060825,
+      "learning_rate": 3.4587995930824014e-06,
+      "loss": 0.0745,
+      "step": 5990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 25.427879333496094,
+      "learning_rate": 3.37402509325195e-06,
+      "loss": 0.6259,
+      "step": 6000
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.011854914017021656,
+      "learning_rate": 3.2892505934214987e-06,
+      "loss": 0.0394,
+      "step": 6010
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 35.059471130371094,
+      "learning_rate": 3.204476093591048e-06,
+      "loss": 0.4307,
+      "step": 6020
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.061668381094932556,
+      "learning_rate": 3.1197015937605968e-06,
+      "loss": 0.3687,
+      "step": 6030
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.013285267166793346,
+      "learning_rate": 3.034927093930146e-06,
+      "loss": 0.9381,
+      "step": 6040
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 64.50102233886719,
+      "learning_rate": 2.950152594099695e-06,
+      "loss": 0.97,
+      "step": 6050
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.057166144251823425,
+      "learning_rate": 2.865378094269244e-06,
+      "loss": 0.0254,
+      "step": 6060
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 69.94447326660156,
+      "learning_rate": 2.780603594438793e-06,
+      "loss": 0.861,
+      "step": 6070
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 157.21966552734375,
+      "learning_rate": 2.695829094608342e-06,
+      "loss": 0.1439,
+      "step": 6080
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.007634480483829975,
+      "learning_rate": 2.611054594777891e-06,
+      "loss": 0.5199,
+      "step": 6090
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.01918291673064232,
+      "learning_rate": 2.5262800949474397e-06,
+      "loss": 1.0748,
+      "step": 6100
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.016131578013300896,
+      "learning_rate": 2.4415055951169887e-06,
+      "loss": 0.0686,
+      "step": 6110
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.016394583508372307,
+      "learning_rate": 2.3567310952865378e-06,
+      "loss": 0.3871,
+      "step": 6120
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.004754968453198671,
+      "learning_rate": 2.271956595456087e-06,
+      "loss": 0.187,
+      "step": 6130
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 56.73072814941406,
+      "learning_rate": 2.187182095625636e-06,
+      "loss": 0.4921,
+      "step": 6140
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.08189795911312103,
+      "learning_rate": 2.102407595795185e-06,
+      "loss": 0.2838,
+      "step": 6150
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 61.22605514526367,
+      "learning_rate": 2.017633095964734e-06,
+      "loss": 0.4503,
+      "step": 6160
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1310381144285202,
+      "learning_rate": 1.932858596134283e-06,
+      "loss": 0.8475,
+      "step": 6170
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.494524002075195,
+      "learning_rate": 1.8480840963038318e-06,
+      "loss": 0.4413,
+      "step": 6180
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.06928512454032898,
+      "learning_rate": 1.7633095964733809e-06,
+      "loss": 0.3415,
+      "step": 6190
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 143.20860290527344,
+      "learning_rate": 1.67853509664293e-06,
+      "loss": 0.742,
+      "step": 6200
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.0477374829351902,
+      "learning_rate": 1.593760596812479e-06,
+      "loss": 0.2915,
+      "step": 6210
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.10252419114112854,
+      "learning_rate": 1.5089860969820278e-06,
+      "loss": 0.1603,
+      "step": 6220
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 96.32301330566406,
+      "learning_rate": 1.4242115971515769e-06,
+      "loss": 0.7714,
+      "step": 6230
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.06277238577604294,
+      "learning_rate": 1.339437097321126e-06,
+      "loss": 0.0346,
+      "step": 6240
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.02291535772383213,
+      "learning_rate": 1.254662597490675e-06,
+      "loss": 0.4468,
+      "step": 6250
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 137.12705993652344,
+      "learning_rate": 1.169888097660224e-06,
+      "loss": 0.2153,
+      "step": 6260
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.162826269865036,
+      "learning_rate": 1.085113597829773e-06,
+      "loss": 0.0054,
+      "step": 6270
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6479134559631348,
+      "learning_rate": 1.0003390979993219e-06,
+      "loss": 0.3338,
+      "step": 6280
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.235650539398193,
+      "learning_rate": 9.155645981688708e-07,
+      "loss": 0.4362,
+      "step": 6290
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.006349239964038134,
+      "learning_rate": 8.307900983384199e-07,
+      "loss": 0.3818,
+      "step": 6300
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.03322937712073326,
+      "learning_rate": 7.460155985079689e-07,
+      "loss": 0.3669,
+      "step": 6310
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 53.04838562011719,
+      "learning_rate": 6.612410986775179e-07,
+      "loss": 0.5879,
+      "step": 6320
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.029276695102453232,
+      "learning_rate": 5.764665988470668e-07,
+      "loss": 0.3514,
+      "step": 6330
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07442772388458252,
+      "learning_rate": 4.916920990166158e-07,
+      "loss": 0.5732,
+      "step": 6340
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.02505446970462799,
+      "learning_rate": 4.069175991861648e-07,
+      "loss": 0.7,
+      "step": 6350
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 159.9132843017578,
+      "learning_rate": 3.2214309935571383e-07,
+      "loss": 0.4448,
+      "step": 6360
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.02801443822681904,
+      "learning_rate": 2.3736859952526283e-07,
+      "loss": 0.0937,
+      "step": 6370
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.011555209755897522,
+      "learning_rate": 1.525940996948118e-07,
+      "loss": 0.2084,
+      "step": 6380
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 67.56371307373047,
+      "learning_rate": 6.781959986436082e-08,
+      "loss": 0.2189,
+      "step": 6390
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9062329141607436,
+      "eval_loss": 0.3474305272102356,
+      "eval_runtime": 3127.9579,
+      "eval_samples_per_second": 1.169,
+      "eval_steps_per_second": 0.585,
+      "step": 6398
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6398,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 3.278594912567663e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}