diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6064 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 8628,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0034770514603616135,
+      "grad_norm": 17.60905647277832,
+      "learning_rate": 1.3904982618771727e-07,
+      "loss": 0.9206,
+      "step": 10
+    },
+    {
+      "epoch": 0.006954102920723227,
+      "grad_norm": 4.7363080978393555,
+      "learning_rate": 3.707995365005794e-07,
+      "loss": 0.8672,
+      "step": 20
+    },
+    {
+      "epoch": 0.01043115438108484,
+      "grad_norm": 2.4911997318267822,
+      "learning_rate": 6.025492468134415e-07,
+      "loss": 0.8277,
+      "step": 30
+    },
+    {
+      "epoch": 0.013908205841446454,
+      "grad_norm": 2.416750192642212,
+      "learning_rate": 8.342989571263037e-07,
+      "loss": 0.7734,
+      "step": 40
+    },
+    {
+      "epoch": 0.017385257301808066,
+      "grad_norm": 2.338721513748169,
+      "learning_rate": 1.066048667439166e-06,
+      "loss": 0.7445,
+      "step": 50
+    },
+    {
+      "epoch": 0.02086230876216968,
+      "grad_norm": 2.1685101985931396,
+      "learning_rate": 1.2977983777520279e-06,
+      "loss": 0.7622,
+      "step": 60
+    },
+    {
+      "epoch": 0.024339360222531293,
+      "grad_norm": 2.2944562435150146,
+      "learning_rate": 1.52954808806489e-06,
+      "loss": 0.7551,
+      "step": 70
+    },
+    {
+      "epoch": 0.027816411682892908,
+      "grad_norm": 2.2963621616363525,
+      "learning_rate": 1.7612977983777522e-06,
+      "loss": 0.7505,
+      "step": 80
+    },
+    {
+      "epoch": 0.03129346314325452,
+      "grad_norm": 2.4709761142730713,
+      "learning_rate": 1.993047508690614e-06,
+      "loss": 0.7695,
+      "step": 90
+    },
+    {
+      "epoch": 0.03477051460361613,
+      "grad_norm": 2.2230093479156494,
+      "learning_rate": 2.2247972190034763e-06,
+      "loss": 0.7712,
+      "step": 100
+    },
+    {
+      "epoch": 0.03824756606397774,
+      "grad_norm": 2.493957281112671,
+      "learning_rate": 2.4565469293163384e-06,
+      "loss": 0.7743,
+      "step": 110
+    },
+    {
+      "epoch": 0.04172461752433936,
+      "grad_norm": 2.5205771923065186,
+      "learning_rate": 2.688296639629201e-06,
+      "loss": 0.7414,
+      "step": 120
+    },
+    {
+      "epoch": 0.045201668984700974,
+      "grad_norm": 2.2279772758483887,
+      "learning_rate": 2.9200463499420627e-06,
+      "loss": 0.7502,
+      "step": 130
+    },
+    {
+      "epoch": 0.048678720445062586,
+      "grad_norm": 2.31657338142395,
+      "learning_rate": 3.151796060254925e-06,
+      "loss": 0.7818,
+      "step": 140
+    },
+    {
+      "epoch": 0.0521557719054242,
+      "grad_norm": 2.198845386505127,
+      "learning_rate": 3.383545770567787e-06,
+      "loss": 0.7467,
+      "step": 150
+    },
+    {
+      "epoch": 0.055632823365785816,
+      "grad_norm": 2.2973482608795166,
+      "learning_rate": 3.6152954808806488e-06,
+      "loss": 0.7822,
+      "step": 160
+    },
+    {
+      "epoch": 0.05910987482614743,
+      "grad_norm": 2.3556406497955322,
+      "learning_rate": 3.847045191193511e-06,
+      "loss": 0.8086,
+      "step": 170
+    },
+    {
+      "epoch": 0.06258692628650904,
+      "grad_norm": 2.3965234756469727,
+      "learning_rate": 4.078794901506373e-06,
+      "loss": 0.7451,
+      "step": 180
+    },
+    {
+      "epoch": 0.06606397774687066,
+      "grad_norm": 2.1980533599853516,
+      "learning_rate": 4.310544611819236e-06,
+      "loss": 0.7518,
+      "step": 190
+    },
+    {
+      "epoch": 0.06954102920723226,
+      "grad_norm": 2.05375075340271,
+      "learning_rate": 4.542294322132097e-06,
+      "loss": 0.7762,
+      "step": 200
+    },
+    {
+      "epoch": 0.07301808066759388,
+      "grad_norm": 2.3330085277557373,
+      "learning_rate": 4.77404403244496e-06,
+      "loss": 0.7229,
+      "step": 210
+    },
+    {
+      "epoch": 0.07649513212795549,
+      "grad_norm": 2.14552640914917,
+      "learning_rate": 5.005793742757822e-06,
+      "loss": 0.7779,
+      "step": 220
+    },
+    {
+      "epoch": 0.0799721835883171,
+      "grad_norm": 2.064589262008667,
+      "learning_rate": 5.2375434530706834e-06,
+      "loss": 0.7612,
+      "step": 230
+    },
+    {
+      "epoch": 0.08344923504867872,
+      "grad_norm": 2.1657590866088867,
+      "learning_rate": 5.469293163383547e-06,
+      "loss": 0.742,
+      "step": 240
+    },
+    {
+      "epoch": 0.08692628650904033,
+      "grad_norm": 2.05084228515625,
+      "learning_rate": 5.701042873696409e-06,
+      "loss": 0.7891,
+      "step": 250
+    },
+    {
+      "epoch": 0.09040333796940195,
+      "grad_norm": 2.368399143218994,
+      "learning_rate": 5.93279258400927e-06,
+      "loss": 0.7783,
+      "step": 260
+    },
+    {
+      "epoch": 0.09388038942976357,
+      "grad_norm": 2.24515700340271,
+      "learning_rate": 6.164542294322132e-06,
+      "loss": 0.7613,
+      "step": 270
+    },
+    {
+      "epoch": 0.09735744089012517,
+      "grad_norm": 2.092841863632202,
+      "learning_rate": 6.3962920046349955e-06,
+      "loss": 0.7383,
+      "step": 280
+    },
+    {
+      "epoch": 0.10083449235048679,
+      "grad_norm": 2.2359280586242676,
+      "learning_rate": 6.628041714947857e-06,
+      "loss": 0.7454,
+      "step": 290
+    },
+    {
+      "epoch": 0.1043115438108484,
+      "grad_norm": 1.9989820718765259,
+      "learning_rate": 6.859791425260719e-06,
+      "loss": 0.7681,
+      "step": 300
+    },
+    {
+      "epoch": 0.10778859527121001,
+      "grad_norm": 2.108048439025879,
+      "learning_rate": 7.091541135573581e-06,
+      "loss": 0.7643,
+      "step": 310
+    },
+    {
+      "epoch": 0.11126564673157163,
+      "grad_norm": 2.189864158630371,
+      "learning_rate": 7.323290845886444e-06,
+      "loss": 0.7839,
+      "step": 320
+    },
+    {
+      "epoch": 0.11474269819193324,
+      "grad_norm": 2.2148871421813965,
+      "learning_rate": 7.555040556199306e-06,
+      "loss": 0.773,
+      "step": 330
+    },
+    {
+      "epoch": 0.11821974965229486,
+      "grad_norm": 2.234316349029541,
+      "learning_rate": 7.786790266512167e-06,
+      "loss": 0.8095,
+      "step": 340
+    },
+    {
+      "epoch": 0.12169680111265646,
+      "grad_norm": 2.0373663902282715,
+      "learning_rate": 8.01853997682503e-06,
+      "loss": 0.7944,
+      "step": 350
+    },
+    {
+      "epoch": 0.12517385257301808,
+      "grad_norm": 1.9603099822998047,
+      "learning_rate": 8.250289687137892e-06,
+      "loss": 0.7753,
+      "step": 360
+    },
+    {
+      "epoch": 0.1286509040333797,
+      "grad_norm": 2.5638701915740967,
+      "learning_rate": 8.482039397450754e-06,
+      "loss": 0.809,
+      "step": 370
+    },
+    {
+      "epoch": 0.13212795549374132,
+      "grad_norm": 2.1671786308288574,
+      "learning_rate": 8.713789107763615e-06,
+      "loss": 0.7651,
+      "step": 380
+    },
+    {
+      "epoch": 0.1356050069541029,
+      "grad_norm": 2.045619487762451,
+      "learning_rate": 8.945538818076479e-06,
+      "loss": 0.7991,
+      "step": 390
+    },
+    {
+      "epoch": 0.13908205841446453,
+      "grad_norm": 2.0538296699523926,
+      "learning_rate": 9.17728852838934e-06,
+      "loss": 0.7637,
+      "step": 400
+    },
+    {
+      "epoch": 0.14255910987482615,
+      "grad_norm": 2.0183122158050537,
+      "learning_rate": 9.409038238702202e-06,
+      "loss": 0.7852,
+      "step": 410
+    },
+    {
+      "epoch": 0.14603616133518776,
+      "grad_norm": 2.105830430984497,
+      "learning_rate": 9.640787949015064e-06,
+      "loss": 0.7984,
+      "step": 420
+    },
+    {
+      "epoch": 0.14951321279554938,
+      "grad_norm": 2.0470869541168213,
+      "learning_rate": 9.872537659327926e-06,
+      "loss": 0.7738,
+      "step": 430
+    },
+    {
+      "epoch": 0.15299026425591097,
+      "grad_norm": 2.0909335613250732,
+      "learning_rate": 1.0104287369640789e-05,
+      "loss": 0.764,
+      "step": 440
+    },
+    {
+      "epoch": 0.1564673157162726,
+      "grad_norm": 2.051558256149292,
+      "learning_rate": 1.033603707995365e-05,
+      "loss": 0.8008,
+      "step": 450
+    },
+    {
+      "epoch": 0.1599443671766342,
+      "grad_norm": 2.1118927001953125,
+      "learning_rate": 1.0567786790266513e-05,
+      "loss": 0.7328,
+      "step": 460
+    },
+    {
+      "epoch": 0.16342141863699583,
+      "grad_norm": 2.2946789264678955,
+      "learning_rate": 1.0799536500579374e-05,
+      "loss": 0.7874,
+      "step": 470
+    },
+    {
+      "epoch": 0.16689847009735745,
+      "grad_norm": 2.1292600631713867,
+      "learning_rate": 1.1031286210892238e-05,
+      "loss": 0.7864,
+      "step": 480
+    },
+    {
+      "epoch": 0.17037552155771907,
+      "grad_norm": 2.032799482345581,
+      "learning_rate": 1.12630359212051e-05,
+      "loss": 0.7706,
+      "step": 490
+    },
+    {
+      "epoch": 0.17385257301808066,
+      "grad_norm": 2.058987617492676,
+      "learning_rate": 1.1494785631517961e-05,
+      "loss": 0.7648,
+      "step": 500
+    },
+    {
+      "epoch": 0.17732962447844228,
+      "grad_norm": 2.1940550804138184,
+      "learning_rate": 1.1726535341830825e-05,
+      "loss": 0.7536,
+      "step": 510
+    },
+    {
+      "epoch": 0.1808066759388039,
+      "grad_norm": 2.0177531242370605,
+      "learning_rate": 1.1958285052143686e-05,
+      "loss": 0.7859,
+      "step": 520
+    },
+    {
+      "epoch": 0.18428372739916551,
+      "grad_norm": 1.9885722398757935,
+      "learning_rate": 1.2190034762456548e-05,
+      "loss": 0.7855,
+      "step": 530
+    },
+    {
+      "epoch": 0.18776077885952713,
+      "grad_norm": 2.135956287384033,
+      "learning_rate": 1.242178447276941e-05,
+      "loss": 0.807,
+      "step": 540
+    },
+    {
+      "epoch": 0.19123783031988872,
+      "grad_norm": 2.0478522777557373,
+      "learning_rate": 1.2653534183082272e-05,
+      "loss": 0.7891,
+      "step": 550
+    },
+    {
+      "epoch": 0.19471488178025034,
+      "grad_norm": 1.9515680074691772,
+      "learning_rate": 1.2885283893395135e-05,
+      "loss": 0.7924,
+      "step": 560
+    },
+    {
+      "epoch": 0.19819193324061196,
+      "grad_norm": 2.5692481994628906,
+      "learning_rate": 1.3117033603707997e-05,
+      "loss": 0.8112,
+      "step": 570
+    },
+    {
+      "epoch": 0.20166898470097358,
+      "grad_norm": 2.010455369949341,
+      "learning_rate": 1.3348783314020858e-05,
+      "loss": 0.8411,
+      "step": 580
+    },
+    {
+      "epoch": 0.2051460361613352,
+      "grad_norm": 1.9256759881973267,
+      "learning_rate": 1.358053302433372e-05,
+      "loss": 0.818,
+      "step": 590
+    },
+    {
+      "epoch": 0.2086230876216968,
+      "grad_norm": 2.0134973526000977,
+      "learning_rate": 1.3812282734646582e-05,
+      "loss": 0.8029,
+      "step": 600
+    },
+    {
+      "epoch": 0.2121001390820584,
+      "grad_norm": 2.051887035369873,
+      "learning_rate": 1.4044032444959445e-05,
+      "loss": 0.8366,
+      "step": 610
+    },
+    {
+      "epoch": 0.21557719054242003,
+      "grad_norm": 2.3100674152374268,
+      "learning_rate": 1.4275782155272307e-05,
+      "loss": 0.7919,
+      "step": 620
+    },
+    {
+      "epoch": 0.21905424200278165,
+      "grad_norm": 1.9310871362686157,
+      "learning_rate": 1.4484356894553884e-05,
+      "loss": 0.8275,
+      "step": 630
+    },
+    {
+      "epoch": 0.22253129346314326,
+      "grad_norm": 2.004302978515625,
+      "learning_rate": 1.4716106604866744e-05,
+      "loss": 0.8345,
+      "step": 640
+    },
+    {
+      "epoch": 0.22600834492350486,
+      "grad_norm": 2.36110520362854,
+      "learning_rate": 1.4947856315179607e-05,
+      "loss": 0.8199,
+      "step": 650
+    },
+    {
+      "epoch": 0.22948539638386647,
+      "grad_norm": 2.114895820617676,
+      "learning_rate": 1.517960602549247e-05,
+      "loss": 0.844,
+      "step": 660
+    },
+    {
+      "epoch": 0.2329624478442281,
+      "grad_norm": 1.8852134943008423,
+      "learning_rate": 1.541135573580533e-05,
+      "loss": 0.798,
+      "step": 670
+    },
+    {
+      "epoch": 0.2364394993045897,
+      "grad_norm": 1.9637327194213867,
+      "learning_rate": 1.5643105446118196e-05,
+      "loss": 0.7809,
+      "step": 680
+    },
+    {
+      "epoch": 0.23991655076495133,
+      "grad_norm": 1.9046761989593506,
+      "learning_rate": 1.5874855156431054e-05,
+      "loss": 0.7961,
+      "step": 690
+    },
+    {
+      "epoch": 0.24339360222531292,
+      "grad_norm": 1.9784783124923706,
+      "learning_rate": 1.610660486674392e-05,
+      "loss": 0.8282,
+      "step": 700
+    },
+    {
+      "epoch": 0.24687065368567454,
+      "grad_norm": 1.8895939588546753,
+      "learning_rate": 1.633835457705678e-05,
+      "loss": 0.7863,
+      "step": 710
+    },
+    {
+      "epoch": 0.25034770514603616,
+      "grad_norm": 1.9012371301651,
+      "learning_rate": 1.6570104287369643e-05,
+      "loss": 0.8381,
+      "step": 720
+    },
+    {
+      "epoch": 0.25382475660639775,
+      "grad_norm": 2.0689969062805176,
+      "learning_rate": 1.6801853997682504e-05,
+      "loss": 0.8217,
+      "step": 730
+    },
+    {
+      "epoch": 0.2573018080667594,
+      "grad_norm": 1.8947774171829224,
+      "learning_rate": 1.7033603707995366e-05,
+      "loss": 0.7932,
+      "step": 740
+    },
+    {
+      "epoch": 0.260778859527121,
+      "grad_norm": 2.189105749130249,
+      "learning_rate": 1.7265353418308228e-05,
+      "loss": 0.8331,
+      "step": 750
+    },
+    {
+      "epoch": 0.26425591098748263,
+      "grad_norm": 1.9160549640655518,
+      "learning_rate": 1.7497103128621093e-05,
+      "loss": 0.85,
+      "step": 760
+    },
+    {
+      "epoch": 0.2677329624478442,
+      "grad_norm": 1.8907710313796997,
+      "learning_rate": 1.772885283893395e-05,
+      "loss": 0.801,
+      "step": 770
+    },
+    {
+      "epoch": 0.2712100139082058,
+      "grad_norm": 2.018258810043335,
+      "learning_rate": 1.7960602549246816e-05,
+      "loss": 0.8454,
+      "step": 780
+    },
+    {
+      "epoch": 0.27468706536856746,
+      "grad_norm": 2.121053695678711,
+      "learning_rate": 1.8192352259559678e-05,
+      "loss": 0.8262,
+      "step": 790
+    },
+    {
+      "epoch": 0.27816411682892905,
+      "grad_norm": 1.7782844305038452,
+      "learning_rate": 1.842410196987254e-05,
+      "loss": 0.784,
+      "step": 800
+    },
+    {
+      "epoch": 0.2816411682892907,
+      "grad_norm": 1.8965550661087036,
+      "learning_rate": 1.86558516801854e-05,
+      "loss": 0.8348,
+      "step": 810
+    },
+    {
+      "epoch": 0.2851182197496523,
+      "grad_norm": 1.8140422105789185,
+      "learning_rate": 1.8887601390498263e-05,
+      "loss": 0.8201,
+      "step": 820
+    },
+    {
+      "epoch": 0.2885952712100139,
+      "grad_norm": 1.985003113746643,
+      "learning_rate": 1.9119351100811125e-05,
+      "loss": 0.8163,
+      "step": 830
+    },
+    {
+      "epoch": 0.29207232267037553,
+      "grad_norm": 1.7549387216567993,
+      "learning_rate": 1.9351100811123987e-05,
+      "loss": 0.8513,
+      "step": 840
+    },
+    {
+      "epoch": 0.2955493741307371,
+      "grad_norm": 1.8897957801818848,
+      "learning_rate": 1.958285052143685e-05,
+      "loss": 0.796,
+      "step": 850
+    },
+    {
+      "epoch": 0.29902642559109877,
+      "grad_norm": 2.022491216659546,
+      "learning_rate": 1.9814600231749714e-05,
+      "loss": 0.8436,
+      "step": 860
+    },
+    {
+      "epoch": 0.30250347705146036,
+      "grad_norm": 1.883622169494629,
+      "learning_rate": 1.999999672624031e-05,
+      "loss": 0.8253,
+      "step": 870
+    },
+    {
+      "epoch": 0.30598052851182195,
+      "grad_norm": 1.9519065618515015,
+      "learning_rate": 1.999988214487614e-05,
+      "loss": 0.8488,
+      "step": 880
+    },
+    {
+      "epoch": 0.3094575799721836,
+      "grad_norm": 1.6121009588241577,
+      "learning_rate": 1.9999603877670846e-05,
+      "loss": 0.8032,
+      "step": 890
+    },
+    {
+      "epoch": 0.3129346314325452,
+      "grad_norm": 1.8443603515625,
+      "learning_rate": 1.9999161929179317e-05,
+      "loss": 0.8438,
+      "step": 900
+    },
+    {
+      "epoch": 0.31641168289290683,
+      "grad_norm": 1.9686535596847534,
+      "learning_rate": 1.9998556306635707e-05,
+      "loss": 0.8284,
+      "step": 910
+    },
+    {
+      "epoch": 0.3198887343532684,
+      "grad_norm": 1.8790724277496338,
+      "learning_rate": 1.9997787019953323e-05,
+      "loss": 0.8408,
+      "step": 920
+    },
+    {
+      "epoch": 0.32336578581363007,
+      "grad_norm": 2.0117969512939453,
+      "learning_rate": 1.9996854081724445e-05,
+      "loss": 0.8197,
+      "step": 930
+    },
+    {
+      "epoch": 0.32684283727399166,
+      "grad_norm": 1.7416011095046997,
+      "learning_rate": 1.999575750722013e-05,
+      "loss": 0.8175,
+      "step": 940
+    },
+    {
+      "epoch": 0.33031988873435325,
+      "grad_norm": 1.8774964809417725,
+      "learning_rate": 1.9994497314389958e-05,
+      "loss": 0.8104,
+      "step": 950
+    },
+    {
+      "epoch": 0.3337969401947149,
+      "grad_norm": 1.8444308042526245,
+      "learning_rate": 1.999307352386175e-05,
+      "loss": 0.8496,
+      "step": 960
+    },
+    {
+      "epoch": 0.3372739916550765,
+      "grad_norm": 1.7105716466903687,
+      "learning_rate": 1.999148615894121e-05,
+      "loss": 0.8037,
+      "step": 970
+    },
+    {
+      "epoch": 0.34075104311543813,
+      "grad_norm": 1.6411991119384766,
+      "learning_rate": 1.9989735245611567e-05,
+      "loss": 0.8368,
+      "step": 980
+    },
+    {
+      "epoch": 0.3442280945757997,
+      "grad_norm": 1.7530555725097656,
+      "learning_rate": 1.998782081253313e-05,
+      "loss": 0.8003,
+      "step": 990
+    },
+    {
+      "epoch": 0.3477051460361613,
+      "grad_norm": 1.9082412719726562,
+      "learning_rate": 1.998574289104282e-05,
+      "loss": 0.8435,
+      "step": 1000
+    },
+    {
+      "epoch": 0.35118219749652296,
+      "grad_norm": 1.8263925313949585,
+      "learning_rate": 1.9983501515153677e-05,
+      "loss": 0.8626,
+      "step": 1010
+    },
+    {
+      "epoch": 0.35465924895688455,
+      "grad_norm": 1.7870615720748901,
+      "learning_rate": 1.9981096721554286e-05,
+      "loss": 0.8429,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3581363004172462,
+      "grad_norm": 1.7090221643447876,
+      "learning_rate": 1.997852854960817e-05,
+      "loss": 0.8748,
+      "step": 1030
+    },
+    {
+      "epoch": 0.3616133518776078,
+      "grad_norm": 1.907475233078003,
+      "learning_rate": 1.9975797041353163e-05,
+      "loss": 0.8269,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3650904033379694,
+      "grad_norm": 1.8359696865081787,
+      "learning_rate": 1.9972902241500723e-05,
+      "loss": 0.8407,
+      "step": 1050
+    },
+    {
+      "epoch": 0.36856745479833103,
+      "grad_norm": 1.8366531133651733,
+      "learning_rate": 1.9969844197435174e-05,
+      "loss": 0.8738,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3720445062586926,
+      "grad_norm": 1.5955488681793213,
+      "learning_rate": 1.9966622959212964e-05,
+      "loss": 0.8291,
+      "step": 1070
+    },
+    {
+      "epoch": 0.37552155771905427,
+      "grad_norm": 1.896794319152832,
+      "learning_rate": 1.9963238579561816e-05,
+      "loss": 0.8261,
+      "step": 1080
+    },
+    {
+      "epoch": 0.37899860917941586,
+      "grad_norm": 1.5895673036575317,
+      "learning_rate": 1.9959691113879897e-05,
+      "loss": 0.8231,
+      "step": 1090
+    },
+    {
+      "epoch": 0.38247566063977745,
+      "grad_norm": 1.7253202199935913,
+      "learning_rate": 1.995598062023487e-05,
+      "loss": 0.8437,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3859527121001391,
+      "grad_norm": 1.7107621431350708,
+      "learning_rate": 1.9952107159362985e-05,
+      "loss": 0.8413,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3894297635605007,
+      "grad_norm": 1.6093268394470215,
+      "learning_rate": 1.9948070794668057e-05,
+      "loss": 0.8362,
+      "step": 1120
+    },
+    {
+      "epoch": 0.39290681502086233,
+      "grad_norm": 1.565251111984253,
+      "learning_rate": 1.9943871592220437e-05,
+      "loss": 0.8089,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3963838664812239,
+      "grad_norm": 1.6408239603042603,
+      "learning_rate": 1.9939509620755945e-05,
+      "loss": 0.8427,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3998609179415855,
+      "grad_norm": 1.5788135528564453,
+      "learning_rate": 1.9934984951674704e-05,
+      "loss": 0.832,
+      "step": 1150
+    },
+    {
+      "epoch": 0.40333796940194716,
+      "grad_norm": 1.8651041984558105,
+      "learning_rate": 1.9930297659040022e-05,
+      "loss": 0.8129,
+      "step": 1160
+    },
+    {
+      "epoch": 0.40681502086230875,
+      "grad_norm": 1.6567399501800537,
+      "learning_rate": 1.992544781957715e-05,
+      "loss": 0.8426,
+      "step": 1170
+    },
+    {
+      "epoch": 0.4102920723226704,
+      "grad_norm": 1.6481776237487793,
+      "learning_rate": 1.9920435512672016e-05,
+      "loss": 0.8107,
+      "step": 1180
+    },
+    {
+      "epoch": 0.413769123783032,
+      "grad_norm": 1.4063149690628052,
+      "learning_rate": 1.991526082036997e-05,
+      "loss": 0.7953,
+      "step": 1190
+    },
+    {
+      "epoch": 0.4172461752433936,
+      "grad_norm": 1.8607759475708008,
+      "learning_rate": 1.9909923827374383e-05,
+      "loss": 0.8426,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4207232267037552,
+      "grad_norm": 1.6585954427719116,
+      "learning_rate": 1.9904424621045314e-05,
+      "loss": 0.8435,
+      "step": 1210
+    },
+    {
+      "epoch": 0.4242002781641168,
+      "grad_norm": 1.7691593170166016,
+      "learning_rate": 1.989876329139804e-05,
+      "loss": 0.8249,
+      "step": 1220
+    },
+    {
+      "epoch": 0.42767732962447846,
+      "grad_norm": 1.7459466457366943,
+      "learning_rate": 1.98929399311016e-05,
+      "loss": 0.8515,
+      "step": 1230
+    },
+    {
+      "epoch": 0.43115438108484005,
+      "grad_norm": 1.785552978515625,
+      "learning_rate": 1.9886954635477285e-05,
+      "loss": 0.8606,
+      "step": 1240
+    },
+    {
+      "epoch": 0.43463143254520165,
+      "grad_norm": 1.706426739692688,
+      "learning_rate": 1.9880807502497065e-05,
+      "loss": 0.863,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4381084840055633,
+      "grad_norm": 1.677270770072937,
+      "learning_rate": 1.987449863278199e-05,
+      "loss": 0.8572,
+      "step": 1260
+    },
+    {
+      "epoch": 0.4415855354659249,
+      "grad_norm": 1.8891503810882568,
+      "learning_rate": 1.9868028129600537e-05,
+      "loss": 0.8398,
+      "step": 1270
+    },
+    {
+      "epoch": 0.44506258692628653,
+      "grad_norm": 1.8624945878982544,
+      "learning_rate": 1.9861396098866934e-05,
+      "loss": 0.8551,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4485396383866481,
+      "grad_norm": 1.5721094608306885,
+      "learning_rate": 1.985460264913941e-05,
+      "loss": 0.8809,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4520166898470097,
+      "grad_norm": 2.049851655960083,
+      "learning_rate": 1.9847647891618432e-05,
+      "loss": 0.8219,
+      "step": 1300
+    },
+    {
+      "epoch": 0.45549374130737136,
+      "grad_norm": 1.6533946990966797,
+      "learning_rate": 1.9840531940144873e-05,
+      "loss": 0.8365,
+      "step": 1310
+    },
+    {
+      "epoch": 0.45897079276773295,
+      "grad_norm": 1.5543324947357178,
+      "learning_rate": 1.983325491119816e-05,
+      "loss": 0.8172,
+      "step": 1320
+    },
+    {
+      "epoch": 0.4624478442280946,
+      "grad_norm": 1.5891435146331787,
+      "learning_rate": 1.982581692389435e-05,
+      "loss": 0.8581,
+      "step": 1330
+    },
+    {
+      "epoch": 0.4659248956884562,
+      "grad_norm": 1.6378602981567383,
+      "learning_rate": 1.98182180999842e-05,
+      "loss": 0.8512,
+      "step": 1340
+    },
+    {
+      "epoch": 0.4694019471488178,
+      "grad_norm": 1.638283610343933,
+      "learning_rate": 1.9810458563851172e-05,
+      "loss": 0.8459,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4728789986091794,
+      "grad_norm": 1.627204179763794,
+      "learning_rate": 1.9802538442509378e-05,
+      "loss": 0.8721,
+      "step": 1360
+    },
+    {
+      "epoch": 0.476356050069541,
+      "grad_norm": 2.00606107711792,
+      "learning_rate": 1.9794457865601516e-05,
+      "loss": 0.7874,
+      "step": 1370
+    },
+    {
+      "epoch": 0.47983310152990266,
+      "grad_norm": 1.7765477895736694,
+      "learning_rate": 1.978621696539675e-05,
+      "loss": 0.808,
+      "step": 1380
+    },
+    {
+      "epoch": 0.48331015299026425,
+      "grad_norm": 1.6700396537780762,
+      "learning_rate": 1.9777815876788537e-05,
+      "loss": 0.83,
+      "step": 1390
+    },
+    {
+      "epoch": 0.48678720445062584,
+      "grad_norm": 1.6221411228179932,
+      "learning_rate": 1.9769254737292427e-05,
+      "loss": 0.8369,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4902642559109875,
+      "grad_norm": 1.718145728111267,
+      "learning_rate": 1.9760533687043797e-05,
+      "loss": 0.8159,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4937413073713491,
+      "grad_norm": 1.6057517528533936,
+      "learning_rate": 1.9751652868795576e-05,
+      "loss": 0.7942,
+      "step": 1420
+    },
+    {
+      "epoch": 0.4972183588317107,
+      "grad_norm": 1.7553658485412598,
+      "learning_rate": 1.9742612427915904e-05,
+      "loss": 0.8672,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5006954102920723,
+      "grad_norm": 1.6311688423156738,
+      "learning_rate": 1.9733412512385733e-05,
+      "loss": 0.8701,
+      "step": 1440
+    },
+    {
+      "epoch": 0.5041724617524339,
+      "grad_norm": 1.5603584051132202,
+      "learning_rate": 1.9724053272796432e-05,
+      "loss": 0.8138,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5076495132127955,
+      "grad_norm": 1.6003831624984741,
+      "learning_rate": 1.9714534862347312e-05,
+      "loss": 0.83,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5111265646731572,
+      "grad_norm": 1.553084135055542,
+      "learning_rate": 1.9704857436843105e-05,
+      "loss": 0.8466,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5146036161335188,
+      "grad_norm": 1.9664576053619385,
+      "learning_rate": 1.9695021154691436e-05,
+      "loss": 0.8073,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5180806675938804,
+      "grad_norm": 1.672797441482544,
+      "learning_rate": 1.968502617690021e-05,
+      "loss": 0.8396,
+      "step": 1490
+    },
+    {
+      "epoch": 0.521557719054242,
+      "grad_norm": 1.8327759504318237,
+      "learning_rate": 1.967487266707499e-05,
+      "loss": 0.8553,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5250347705146036,
+      "grad_norm": 1.6596057415008545,
+      "learning_rate": 1.966456079141632e-05,
+      "loss": 0.8676,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5285118219749653,
+      "grad_norm": 1.4589866399765015,
+      "learning_rate": 1.9654090718716985e-05,
+      "loss": 0.7978,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5319888734353269,
+      "grad_norm": 1.564098834991455,
+      "learning_rate": 1.964346262035928e-05,
+      "loss": 0.8703,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5354659248956884,
+      "grad_norm": 1.7172664403915405,
+      "learning_rate": 1.9632676670312168e-05,
+      "loss": 0.851,
+      "step": 1540
+    },
+    {
+      "epoch": 0.53894297635605,
+      "grad_norm": 1.7797482013702393,
+      "learning_rate": 1.9621733045128464e-05,
+      "loss": 0.8587,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5424200278164116,
+      "grad_norm": 1.6376184225082397,
+      "learning_rate": 1.961063192394193e-05,
+      "loss": 0.8257,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5458970792767733,
+      "grad_norm": 1.8292838335037231,
+      "learning_rate": 1.9599373488464345e-05,
+      "loss": 0.8401,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5493741307371349,
+      "grad_norm": 1.5343873500823975,
+      "learning_rate": 1.9587957922982524e-05,
+      "loss": 0.8837,
+      "step": 1580
+    },
+    {
+      "epoch": 0.5528511821974965,
+      "grad_norm": 1.564792275428772,
+      "learning_rate": 1.9576385414355318e-05,
+      "loss": 0.8782,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5563282336578581,
+      "grad_norm": 1.5822173357009888,
+      "learning_rate": 1.9564656152010534e-05,
+      "loss": 0.8518,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5598052851182197,
+      "grad_norm": 1.4801942110061646,
+      "learning_rate": 1.9552770327941858e-05,
+      "loss": 0.8324,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5632823365785814,
+      "grad_norm": 1.4978936910629272,
+      "learning_rate": 1.9540728136705694e-05,
+      "loss": 0.8417,
+      "step": 1620
+    },
+    {
+      "epoch": 0.566759388038943,
+      "grad_norm": 1.6086760759353638,
+      "learning_rate": 1.952852977541798e-05,
+      "loss": 0.8008,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5702364394993046,
+      "grad_norm": 1.517822504043579,
+      "learning_rate": 1.951617544375098e-05,
+      "loss": 0.8429,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5737134909596662,
+      "grad_norm": 1.538248896598816,
+      "learning_rate": 1.950366534392998e-05,
+      "loss": 0.8104,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5771905424200278,
+      "grad_norm": 1.4534393548965454,
+      "learning_rate": 1.9490999680730033e-05,
+      "loss": 0.8474,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5806675938803895,
+      "grad_norm": 1.535582423210144,
+      "learning_rate": 1.9480755283771743e-05,
+      "loss": 0.8444,
+      "step": 1670
+    },
+    {
+      "epoch": 0.5841446453407511,
+      "grad_norm": 1.4962369203567505,
+      "learning_rate": 1.9469111623418424e-05,
+      "loss": 0.8402,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5876216968011126,
+      "grad_norm": 1.5751090049743652,
+      "learning_rate": 1.9456026981861998e-05,
+      "loss": 0.8548,
+      "step": 1690
+    },
+    {
+      "epoch": 0.5910987482614742,
+      "grad_norm": 1.5477616786956787,
+      "learning_rate": 1.9442787556708343e-05,
+      "loss": 0.8375,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5945757997218358,
+      "grad_norm": 1.8721061944961548,
+      "learning_rate": 1.9429393564670657e-05,
+      "loss": 0.8366,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5980528511821975,
+      "grad_norm": 1.4902026653289795,
+      "learning_rate": 1.9415845224992213e-05,
+      "loss": 0.8185,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6015299026425591,
+      "grad_norm": 1.4728401899337769,
+      "learning_rate": 1.9402142759442754e-05,
+      "loss": 0.8317,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6050069541029207,
+      "grad_norm": 1.5128939151763916,
+      "learning_rate": 1.938828639231489e-05,
+      "loss": 0.8354,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6084840055632823,
+      "grad_norm": 1.5341829061508179,
+      "learning_rate": 1.93742763504204e-05,
+      "loss": 0.8343,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6119610570236439,
+      "grad_norm": 1.6342803239822388,
+      "learning_rate": 1.9360112863086543e-05,
+      "loss": 0.86,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6154381084840056,
+      "grad_norm": 1.4738048315048218,
+      "learning_rate": 1.9345796162152277e-05,
+      "loss": 0.832,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6189151599443672,
+      "grad_norm": 1.437084674835205,
+      "learning_rate": 1.933132648196449e-05,
+      "loss": 0.8353,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6223922114047288,
+      "grad_norm": 1.5915112495422363,
+      "learning_rate": 1.9316704059374155e-05,
+      "loss": 0.8227,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6258692628650904,
+      "grad_norm": 1.4458816051483154,
+      "learning_rate": 1.9301929133732443e-05,
+      "loss": 0.8627,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6293463143254521,
+      "grad_norm": 1.4779449701309204,
+      "learning_rate": 1.928700194688682e-05,
+      "loss": 0.7921,
+      "step": 1810
+    },
+    {
+      "epoch": 0.6328233657858137,
+      "grad_norm": 1.4546343088150024,
+      "learning_rate": 1.9271922743177073e-05,
+      "loss": 0.8093,
+      "step": 1820
+    },
+    {
+      "epoch": 0.6363004172461753,
+      "grad_norm": 1.6483298540115356,
+      "learning_rate": 1.9256691769431333e-05,
+      "loss": 0.8458,
+      "step": 1830
+    },
+    {
+      "epoch": 0.6397774687065368,
+      "grad_norm": 1.4605571031570435,
+      "learning_rate": 1.924130927496201e-05,
+      "loss": 0.8453,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6432545201668984,
+      "grad_norm": 1.6606123447418213,
+      "learning_rate": 1.922577551156172e-05,
+      "loss": 0.7841,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6467315716272601,
+      "grad_norm": 1.5681625604629517,
+      "learning_rate": 1.9210090733499186e-05,
+      "loss": 0.8288,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6502086230876217,
+      "grad_norm": 1.7104305028915405,
+      "learning_rate": 1.919425519751503e-05,
+      "loss": 0.8625,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6536856745479833,
+      "grad_norm": 1.5173437595367432,
+      "learning_rate": 1.9178269162817616e-05,
+      "loss": 0.8638,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6571627260083449,
+      "grad_norm": 1.4919418096542358,
+      "learning_rate": 1.9162132891078777e-05,
+      "loss": 0.8529,
+      "step": 1890
+    },
+    {
+      "epoch": 0.6606397774687065,
+      "grad_norm": 1.596191644668579,
+      "learning_rate": 1.9145846646429556e-05,
+      "loss": 0.8074,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6641168289290682,
+      "grad_norm": 1.480033040046692,
+      "learning_rate": 1.912941069545585e-05,
+      "loss": 0.8124,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6675938803894298,
+      "grad_norm": 1.477903962135315,
+      "learning_rate": 1.9112825307194086e-05,
+      "loss": 0.848,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6710709318497914,
+      "grad_norm": 1.5569449663162231,
+      "learning_rate": 1.9096090753126776e-05,
+      "loss": 0.8695,
+      "step": 1930
+    },
+    {
+      "epoch": 0.674547983310153,
+      "grad_norm": 1.555418848991394,
+      "learning_rate": 1.907920730717811e-05,
+      "loss": 0.8444,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6780250347705146,
+      "grad_norm": 1.6460829973220825,
+      "learning_rate": 1.906217524570945e-05,
+      "loss": 0.8329,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6815020862308763,
+      "grad_norm": 1.6289609670639038,
+      "learning_rate": 1.904499484751481e-05,
+      "loss": 0.829,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6849791376912379,
+      "grad_norm": 1.5714190006256104,
+      "learning_rate": 1.9027666393816298e-05,
+      "loss": 0.8268,
+      "step": 1970
+    },
+    {
+      "epoch": 0.6884561891515995,
+      "grad_norm": 1.7332789897918701,
+      "learning_rate": 1.9010190168259513e-05,
+      "loss": 0.8335,
+      "step": 1980
+    },
+    {
+      "epoch": 0.691933240611961,
+      "grad_norm": 1.5074695348739624,
+      "learning_rate": 1.8992566456908886e-05,
+      "loss": 0.8502,
+      "step": 1990
+    },
+    {
+      "epoch": 0.6954102920723226,
+      "grad_norm": 1.674787163734436,
+      "learning_rate": 1.8974795548243027e-05,
+      "loss": 0.838,
+      "step": 2000
+    },
+    {
+      "epoch": 0.6988873435326843,
+      "grad_norm": 1.5252281427383423,
+      "learning_rate": 1.8956877733149976e-05,
+      "loss": 0.8307,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7023643949930459,
+      "grad_norm": 1.392924427986145,
+      "learning_rate": 1.8938813304922447e-05,
+      "loss": 0.8594,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7058414464534075,
+      "grad_norm": 1.4894132614135742,
+      "learning_rate": 1.8920602559253045e-05,
+      "loss": 0.8535,
+      "step": 2030
+    },
+    {
+      "epoch": 0.7093184979137691,
+      "grad_norm": 1.5647095441818237,
+      "learning_rate": 1.8902245794229404e-05,
+      "loss": 0.8284,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7127955493741307,
+      "grad_norm": 1.2620435953140259,
+      "learning_rate": 1.8883743310329312e-05,
+      "loss": 0.7946,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7162726008344924,
+      "grad_norm": 1.4027864933013916,
+      "learning_rate": 1.8865095410415808e-05,
+      "loss": 0.8108,
+      "step": 2060
+    },
+    {
+      "epoch": 0.719749652294854,
+      "grad_norm": 1.6976580619812012,
+      "learning_rate": 1.8846302399732207e-05,
+      "loss": 0.8388,
+      "step": 2070
+    },
+    {
+      "epoch": 0.7232267037552156,
+      "grad_norm": 1.540840983390808,
+      "learning_rate": 1.8827364585897108e-05,
+      "loss": 0.8667,
+      "step": 2080
+    },
+    {
+      "epoch": 0.7267037552155772,
+      "grad_norm": 1.4793633222579956,
+      "learning_rate": 1.8808282278899362e-05,
+      "loss": 0.8147,
+      "step": 2090
+    },
+    {
+      "epoch": 0.7301808066759388,
+      "grad_norm": 1.4766839742660522,
+      "learning_rate": 1.8789055791093004e-05,
+      "loss": 0.8298,
+      "step": 2100
+    },
+    {
+      "epoch": 0.7336578581363005,
+      "grad_norm": 1.599416971206665,
+      "learning_rate": 1.8769685437192122e-05,
+      "loss": 0.812,
+      "step": 2110
+    },
+    {
+      "epoch": 0.7371349095966621,
+      "grad_norm": 1.5855834484100342,
+      "learning_rate": 1.875017153426572e-05,
+      "loss": 0.8347,
+      "step": 2120
+    },
+    {
+      "epoch": 0.7406119610570236,
+      "grad_norm": 1.4539510011672974,
+      "learning_rate": 1.873051440173252e-05,
+      "loss": 0.8355,
+      "step": 2130
+    },
+    {
+      "epoch": 0.7440890125173852,
+      "grad_norm": 1.4970011711120605,
+      "learning_rate": 1.871071436135575e-05,
+      "loss": 0.8213,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7475660639777468,
+      "grad_norm": 1.436998724937439,
+      "learning_rate": 1.8690771737237848e-05,
+      "loss": 0.8576,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7510431154381085,
+      "grad_norm": 1.562181830406189,
+      "learning_rate": 1.8670686855815187e-05,
+      "loss": 0.8292,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7545201668984701,
+      "grad_norm": 1.5296927690505981,
+      "learning_rate": 1.865046004585271e-05,
+      "loss": 0.8548,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7579972183588317,
+      "grad_norm": 1.5077508687973022,
+      "learning_rate": 1.863009163843856e-05,
+      "loss": 0.8354,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7614742698191933,
+      "grad_norm": 1.4402260780334473,
+      "learning_rate": 1.8609581966978654e-05,
+      "loss": 0.8336,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7649513212795549,
+      "grad_norm": 1.4469704627990723,
+      "learning_rate": 1.8588931367191235e-05,
+      "loss": 0.8183,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7684283727399166,
+      "grad_norm": 1.4530783891677856,
+      "learning_rate": 1.8568140177101364e-05,
+      "loss": 0.848,
+      "step": 2210
+    },
+    {
+      "epoch": 0.7719054242002782,
+      "grad_norm": 1.448097586631775,
+      "learning_rate": 1.8547208737035395e-05,
+      "loss": 0.8361,
+      "step": 2220
+    },
+    {
+      "epoch": 0.7753824756606398,
+      "grad_norm": 1.5228530168533325,
+      "learning_rate": 1.852613738961541e-05,
+      "loss": 0.8221,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7788595271210014,
+      "grad_norm": 1.4702507257461548,
+      "learning_rate": 1.8504926479753584e-05,
+      "loss": 0.8547,
+      "step": 2240
+    },
+    {
+      "epoch": 0.782336578581363,
+      "grad_norm": 1.4371702671051025,
+      "learning_rate": 1.848357635464658e-05,
+      "loss": 0.8217,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7858136300417247,
+      "grad_norm": 1.5437310934066772,
+      "learning_rate": 1.8462087363769834e-05,
+      "loss": 0.8533,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7892906815020863,
+      "grad_norm": 1.3799189329147339,
+      "learning_rate": 1.8440459858871847e-05,
+      "loss": 0.7972,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7927677329624478,
+      "grad_norm": 1.7827317714691162,
+      "learning_rate": 1.841869419396842e-05,
+      "loss": 0.877,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7962447844228094,
+      "grad_norm": 1.5526162385940552,
+      "learning_rate": 1.839679072533688e-05,
+      "loss": 0.8219,
+      "step": 2290
+    },
+    {
+      "epoch": 0.799721835883171,
+      "grad_norm": 1.4785300493240356,
+      "learning_rate": 1.8374749811510204e-05,
+      "loss": 0.8322,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8031988873435327,
+      "grad_norm": 1.4434353113174438,
+      "learning_rate": 1.835257181327121e-05,
+      "loss": 0.8009,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8066759388038943,
+      "grad_norm": 1.4406448602676392,
+      "learning_rate": 1.8330257093646596e-05,
+      "loss": 0.8201,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8101529902642559,
+      "grad_norm": 1.396982192993164,
+      "learning_rate": 1.8307806017901043e-05,
+      "loss": 0.8562,
+      "step": 2330
+    },
+    {
+      "epoch": 0.8136300417246175,
+      "grad_norm": 1.431212067604065,
+      "learning_rate": 1.8285218953531194e-05,
+      "loss": 0.8385,
+      "step": 2340
+    },
+    {
+      "epoch": 0.8171070931849791,
+      "grad_norm": 1.4214597940444946,
+      "learning_rate": 1.8262496270259674e-05,
+      "loss": 0.8255,
+      "step": 2350
+    },
+    {
+      "epoch": 0.8205841446453408,
+      "grad_norm": 1.5818418264389038,
+      "learning_rate": 1.8239638340029015e-05,
+      "loss": 0.8258,
+      "step": 2360
+    },
+    {
+      "epoch": 0.8240611961057024,
+      "grad_norm": 1.4360369443893433,
+      "learning_rate": 1.8216645536995586e-05,
+      "loss": 0.7825,
+      "step": 2370
+    },
+    {
+      "epoch": 0.827538247566064,
+      "grad_norm": 1.4426028728485107,
+      "learning_rate": 1.819351823752345e-05,
+      "loss": 0.8157,
+      "step": 2380
+    },
+    {
+      "epoch": 0.8310152990264256,
+      "grad_norm": 1.4472652673721313,
+      "learning_rate": 1.817025682017822e-05,
+      "loss": 0.8241,
+      "step": 2390
+    },
+    {
+      "epoch": 0.8344923504867872,
+      "grad_norm": 1.417043924331665,
+      "learning_rate": 1.814686166572084e-05,
+      "loss": 0.8186,
+      "step": 2400
+    },
+    {
+      "epoch": 0.8379694019471489,
+      "grad_norm": 1.3455575704574585,
+      "learning_rate": 1.8123333157101383e-05,
+      "loss": 0.8252,
+      "step": 2410
+    },
+    {
+      "epoch": 0.8414464534075105,
+      "grad_norm": 1.3810886144638062,
+      "learning_rate": 1.809967167945276e-05,
+      "loss": 0.7978,
+      "step": 2420
+    },
+    {
+      "epoch": 0.844923504867872,
+      "grad_norm": 1.437900185585022,
+      "learning_rate": 1.8075877620084426e-05,
+      "loss": 0.8179,
+      "step": 2430
+    },
+    {
+      "epoch": 0.8484005563282336,
+      "grad_norm": 1.3696163892745972,
+      "learning_rate": 1.8051951368476027e-05,
+      "loss": 0.8648,
+      "step": 2440
+    },
+    {
+      "epoch": 0.8518776077885952,
+      "grad_norm": 1.3174755573272705,
+      "learning_rate": 1.8027893316271042e-05,
+      "loss": 0.8076,
+      "step": 2450
+    },
+    {
+      "epoch": 0.8553546592489569,
+      "grad_norm": 1.4821360111236572,
+      "learning_rate": 1.8003703857270367e-05,
+      "loss": 0.8096,
+      "step": 2460
+    },
+    {
+      "epoch": 0.8588317107093185,
+      "grad_norm": 1.4060131311416626,
+      "learning_rate": 1.797938338742586e-05,
+      "loss": 0.8429,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8623087621696801,
+      "grad_norm": 1.495606780052185,
+      "learning_rate": 1.7954932304833867e-05,
+      "loss": 0.82,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8657858136300417,
+      "grad_norm": 1.4672462940216064,
+      "learning_rate": 1.7930351009728715e-05,
+      "loss": 0.8128,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8692628650904033,
+      "grad_norm": 1.6004377603530884,
+      "learning_rate": 1.7905639904476137e-05,
+      "loss": 0.8084,
+      "step": 2500
+    },
+    {
+      "epoch": 0.872739916550765,
+      "grad_norm": 1.29775071144104,
+      "learning_rate": 1.78807993935667e-05,
+      "loss": 0.8227,
+      "step": 2510
+    },
+    {
+      "epoch": 0.8762169680111266,
+      "grad_norm": 1.4231377840042114,
+      "learning_rate": 1.7855829883609197e-05,
+      "loss": 0.8826,
+      "step": 2520
+    },
+    {
+      "epoch": 0.8796940194714882,
+      "grad_norm": 1.4173587560653687,
+      "learning_rate": 1.7830731783323964e-05,
+      "loss": 0.829,
+      "step": 2530
+    },
+    {
+      "epoch": 0.8831710709318498,
+      "grad_norm": 1.4393664598464966,
+      "learning_rate": 1.7805505503536213e-05,
+      "loss": 0.7965,
+      "step": 2540
+    },
+    {
+      "epoch": 0.8866481223922114,
+      "grad_norm": 1.4717497825622559,
+      "learning_rate": 1.7780151457169288e-05,
+      "loss": 0.8351,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8901251738525731,
+      "grad_norm": 1.568798542022705,
+      "learning_rate": 1.7754670059237923e-05,
+      "loss": 0.8449,
+      "step": 2560
+    },
+    {
+      "epoch": 0.8936022253129347,
+      "grad_norm": 1.426080346107483,
+      "learning_rate": 1.7729061726841438e-05,
+      "loss": 0.8174,
+      "step": 2570
+    },
+    {
+      "epoch": 0.8970792767732962,
+      "grad_norm": 1.5661389827728271,
+      "learning_rate": 1.770332687915692e-05,
+      "loss": 0.8306,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9005563282336578,
+      "grad_norm": 1.5125105381011963,
+      "learning_rate": 1.7677465937432344e-05,
+      "loss": 0.8099,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9040333796940194,
+      "grad_norm": 1.3868343830108643,
+      "learning_rate": 1.765147932497971e-05,
+      "loss": 0.8034,
+      "step": 2600
+    },
+    {
+      "epoch": 0.9075104311543811,
+      "grad_norm": 1.327939748764038,
+      "learning_rate": 1.762536746716807e-05,
+      "loss": 0.8164,
+      "step": 2610
+    },
+    {
+      "epoch": 0.9109874826147427,
+      "grad_norm": 1.5392996072769165,
+      "learning_rate": 1.7599130791416616e-05,
+      "loss": 0.8099,
+      "step": 2620
+    },
+    {
+      "epoch": 0.9144645340751043,
+      "grad_norm": 1.3875702619552612,
+      "learning_rate": 1.7572769727187635e-05,
+      "loss": 0.7742,
+      "step": 2630
+    },
+    {
+      "epoch": 0.9179415855354659,
+      "grad_norm": 1.329900860786438,
+      "learning_rate": 1.754628470597951e-05,
+      "loss": 0.8229,
+      "step": 2640
+    },
+    {
+      "epoch": 0.9214186369958275,
+      "grad_norm": 1.3071353435516357,
+      "learning_rate": 1.7519676161319653e-05,
+      "loss": 0.8345,
+      "step": 2650
+    },
+    {
+      "epoch": 0.9248956884561892,
+      "grad_norm": 1.3185700178146362,
+      "learning_rate": 1.749294452875739e-05,
+      "loss": 0.7974,
+      "step": 2660
+    },
+    {
+      "epoch": 0.9283727399165508,
+      "grad_norm": 1.3656513690948486,
+      "learning_rate": 1.746609024585686e-05,
+      "loss": 0.8311,
+      "step": 2670
+    },
+    {
+      "epoch": 0.9318497913769124,
+      "grad_norm": 1.636619210243225,
+      "learning_rate": 1.743911375218983e-05,
+      "loss": 0.8326,
+      "step": 2680
+    },
+    {
+      "epoch": 0.935326842837274,
+      "grad_norm": 1.3537416458129883,
+      "learning_rate": 1.741201548932851e-05,
+      "loss": 0.8231,
+      "step": 2690
+    },
+    {
+      "epoch": 0.9388038942976356,
+      "grad_norm": 1.4453548192977905,
+      "learning_rate": 1.7384795900838326e-05,
+      "loss": 0.8289,
+      "step": 2700
+    },
+    {
+      "epoch": 0.9422809457579973,
+      "grad_norm": 1.3700053691864014,
+      "learning_rate": 1.735745543227065e-05,
+      "loss": 0.8574,
+      "step": 2710
+    },
+    {
+      "epoch": 0.9457579972183588,
+      "grad_norm": 1.4554604291915894,
+      "learning_rate": 1.7329994531155512e-05,
+      "loss": 0.7942,
+      "step": 2720
+    },
+    {
+      "epoch": 0.9492350486787204,
+      "grad_norm": 1.4366511106491089,
+      "learning_rate": 1.7302413646994287e-05,
+      "loss": 0.7905,
+      "step": 2730
+    },
+    {
+      "epoch": 0.952712100139082,
+      "grad_norm": 1.3924344778060913,
+      "learning_rate": 1.727471323125231e-05,
+      "loss": 0.834,
+      "step": 2740
+    },
+    {
+      "epoch": 0.9561891515994436,
+      "grad_norm": 1.4259405136108398,
+      "learning_rate": 1.7246893737351515e-05,
+      "loss": 0.8514,
+      "step": 2750
+    },
+    {
+      "epoch": 0.9596662030598053,
+      "grad_norm": 1.3932157754898071,
+      "learning_rate": 1.7218955620662995e-05,
+      "loss": 0.7961,
+      "step": 2760
+    },
+    {
+      "epoch": 0.9631432545201669,
+      "grad_norm": 1.4179917573928833,
+      "learning_rate": 1.719089933849955e-05,
+      "loss": 0.8137,
+      "step": 2770
+    },
+    {
+      "epoch": 0.9666203059805285,
+      "grad_norm": 1.3352140188217163,
+      "learning_rate": 1.7162725350108207e-05,
+      "loss": 0.8448,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9700973574408901,
+      "grad_norm": 1.4144889116287231,
+      "learning_rate": 1.71344341166627e-05,
+      "loss": 0.8193,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9735744089012517,
+      "grad_norm": 1.3491448163986206,
+      "learning_rate": 1.7106026101255928e-05,
+      "loss": 0.8225,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9770514603616134,
+      "grad_norm": 1.2639228105545044,
+      "learning_rate": 1.707750176889235e-05,
+      "loss": 0.8162,
+      "step": 2810
+    },
+    {
+      "epoch": 0.980528511821975,
+      "grad_norm": 1.2977269887924194,
+      "learning_rate": 1.7048861586480407e-05,
+      "loss": 0.8303,
+      "step": 2820
+    },
+    {
+      "epoch": 0.9840055632823366,
+      "grad_norm": 1.499925971031189,
+      "learning_rate": 1.7020106022824864e-05,
+      "loss": 0.8264,
+      "step": 2830
+    },
+    {
+      "epoch": 0.9874826147426982,
+      "grad_norm": 1.452367901802063,
+      "learning_rate": 1.6991235548619123e-05,
+      "loss": 0.8057,
+      "step": 2840
+    },
+    {
+      "epoch": 0.9909596662030598,
+      "grad_norm": 1.3586550951004028,
+      "learning_rate": 1.696225063643754e-05,
+      "loss": 0.8362,
+      "step": 2850
+    },
+    {
+      "epoch": 0.9944367176634215,
+      "grad_norm": 1.3159747123718262,
+      "learning_rate": 1.693315176072768e-05,
+      "loss": 0.7897,
+      "step": 2860
+    },
+    {
+      "epoch": 0.997913769123783,
+      "grad_norm": 1.4708093404769897,
+      "learning_rate": 1.6903939397802562e-05,
+      "loss": 0.7892,
+      "step": 2870
+    },
+    {
+      "epoch": 1.0013908205841446,
+      "grad_norm": 1.4149800539016724,
+      "learning_rate": 1.6874614025832825e-05,
+      "loss": 0.73,
+      "step": 2880
+    },
+    {
+      "epoch": 1.0048678720445063,
+      "grad_norm": 1.5652549266815186,
+      "learning_rate": 1.6845176124838952e-05,
+      "loss": 0.5764,
+      "step": 2890
+    },
+    {
+      "epoch": 1.0083449235048678,
+      "grad_norm": 1.4669792652130127,
+      "learning_rate": 1.6815626176683378e-05,
+      "loss": 0.5724,
+      "step": 2900
+    },
+    {
+      "epoch": 1.0118219749652295,
+      "grad_norm": 1.507089614868164,
+      "learning_rate": 1.6785964665062623e-05,
+      "loss": 0.5794,
+      "step": 2910
+    },
+    {
+      "epoch": 1.015299026425591,
+      "grad_norm": 1.4902583360671997,
+      "learning_rate": 1.6756192075499345e-05,
+      "loss": 0.5675,
+      "step": 2920
+    },
+    {
+      "epoch": 1.0187760778859527,
+      "grad_norm": 1.5730470418930054,
+      "learning_rate": 1.6726308895334433e-05,
+      "loss": 0.5621,
+      "step": 2930
+    },
+    {
+      "epoch": 1.0222531293463144,
+      "grad_norm": 1.4957162141799927,
+      "learning_rate": 1.6696315613719e-05,
+      "loss": 0.5913,
+      "step": 2940
+    },
+    {
+      "epoch": 1.0257301808066759,
+      "grad_norm": 1.4123682975769043,
+      "learning_rate": 1.6666212721606378e-05,
+      "loss": 0.5904,
+      "step": 2950
+    },
+    {
+      "epoch": 1.0292072322670376,
+      "grad_norm": 1.5745741128921509,
+      "learning_rate": 1.66360007117441e-05,
+      "loss": 0.5645,
+      "step": 2960
+    },
+    {
+      "epoch": 1.032684283727399,
+      "grad_norm": 1.5231777429580688,
+      "learning_rate": 1.660568007866582e-05,
+      "loss": 0.5783,
+      "step": 2970
+    },
+    {
+      "epoch": 1.0361613351877608,
+      "grad_norm": 1.5231876373291016,
+      "learning_rate": 1.657525131868322e-05,
+      "loss": 0.5845,
+      "step": 2980
+    },
+    {
+      "epoch": 1.0396383866481225,
+      "grad_norm": 1.405415654182434,
+      "learning_rate": 1.6544714929877888e-05,
+      "loss": 0.5822,
+      "step": 2990
+    },
+    {
+      "epoch": 1.043115438108484,
+      "grad_norm": 1.3969780206680298,
+      "learning_rate": 1.6514071412093163e-05,
+      "loss": 0.5756,
+      "step": 3000
+    },
+    {
+      "epoch": 1.0465924895688457,
+      "grad_norm": 1.5271550416946411,
+      "learning_rate": 1.648332126692595e-05,
+      "loss": 0.5745,
+      "step": 3010
+    },
+    {
+      "epoch": 1.0500695410292071,
+      "grad_norm": 1.5544006824493408,
+      "learning_rate": 1.6452464997718523e-05,
+      "loss": 0.5794,
+      "step": 3020
+    },
+    {
+      "epoch": 1.0535465924895688,
+      "grad_norm": 1.618527889251709,
+      "learning_rate": 1.6421503109550278e-05,
+      "loss": 0.5663,
+      "step": 3030
+    },
+    {
+      "epoch": 1.0570236439499305,
+      "grad_norm": 1.676133632659912,
+      "learning_rate": 1.639043610922945e-05,
+      "loss": 0.5998,
+      "step": 3040
+    },
+    {
+      "epoch": 1.060500695410292,
+      "grad_norm": 1.5752686262130737,
+      "learning_rate": 1.635926450528484e-05,
+      "loss": 0.5674,
+      "step": 3050
+    },
+    {
+      "epoch": 1.0639777468706537,
+      "grad_norm": 1.4197261333465576,
+      "learning_rate": 1.632798880795749e-05,
+      "loss": 0.5922,
+      "step": 3060
+    },
+    {
+      "epoch": 1.0674547983310152,
+      "grad_norm": 1.5364372730255127,
+      "learning_rate": 1.629660952919231e-05,
+      "loss": 0.5965,
+      "step": 3070
+    },
+    {
+      "epoch": 1.070931849791377,
+      "grad_norm": 1.565186858177185,
+      "learning_rate": 1.626512718262972e-05,
+      "loss": 0.5956,
+      "step": 3080
+    },
+    {
+      "epoch": 1.0744089012517386,
+      "grad_norm": 1.5327928066253662,
+      "learning_rate": 1.6233542283597233e-05,
+      "loss": 0.5867,
+      "step": 3090
+    },
+    {
+      "epoch": 1.0778859527121,
+      "grad_norm": 1.6290488243103027,
+      "learning_rate": 1.6201855349101012e-05,
+      "loss": 0.5759,
+      "step": 3100
+    },
+    {
+      "epoch": 1.0813630041724618,
+      "grad_norm": 1.511427879333496,
+      "learning_rate": 1.6170066897817428e-05,
+      "loss": 0.5792,
+      "step": 3110
+    },
+    {
+      "epoch": 1.0848400556328233,
+      "grad_norm": 1.5155751705169678,
+      "learning_rate": 1.613817745008455e-05,
+      "loss": 0.5789,
+      "step": 3120
+    },
+    {
+      "epoch": 1.088317107093185,
+      "grad_norm": 1.493688702583313,
+      "learning_rate": 1.6106187527893643e-05,
+      "loss": 0.5983,
+      "step": 3130
+    },
+    {
+      "epoch": 1.0917941585535467,
+      "grad_norm": 1.4319963455200195,
+      "learning_rate": 1.6074097654880604e-05,
+      "loss": 0.5913,
+      "step": 3140
+    },
+    {
+      "epoch": 1.0952712100139081,
+      "grad_norm": 1.4269553422927856,
+      "learning_rate": 1.6041908356317408e-05,
+      "loss": 0.5907,
+      "step": 3150
+    },
+    {
+      "epoch": 1.0987482614742698,
+      "grad_norm": 1.7060624361038208,
+      "learning_rate": 1.6009620159103506e-05,
+      "loss": 0.6033,
+      "step": 3160
+    },
+    {
+      "epoch": 1.1022253129346313,
+      "grad_norm": 1.4631054401397705,
+      "learning_rate": 1.5977233591757207e-05,
+      "loss": 0.5628,
+      "step": 3170
+    },
+    {
+      "epoch": 1.105702364394993,
+      "grad_norm": 1.6170661449432373,
+      "learning_rate": 1.5944749184407e-05,
+      "loss": 0.5882,
+      "step": 3180
+    },
+    {
+      "epoch": 1.1091794158553547,
+      "grad_norm": 1.5175589323043823,
+      "learning_rate": 1.5912167468782917e-05,
+      "loss": 0.5812,
+      "step": 3190
+    },
+    {
+      "epoch": 1.1126564673157162,
+      "grad_norm": 1.5576196908950806,
+      "learning_rate": 1.5879488978207786e-05,
+      "loss": 0.5941,
+      "step": 3200
+    },
+    {
+      "epoch": 1.116133518776078,
+      "grad_norm": 1.6332759857177734,
+      "learning_rate": 1.5846714247588537e-05,
+      "loss": 0.5785,
+      "step": 3210
+    },
+    {
+      "epoch": 1.1196105702364394,
+      "grad_norm": 1.569899559020996,
+      "learning_rate": 1.581384381340743e-05,
+      "loss": 0.5981,
+      "step": 3220
+    },
+    {
+      "epoch": 1.123087621696801,
+      "grad_norm": 1.4246937036514282,
+      "learning_rate": 1.578087821371327e-05,
+      "loss": 0.5613,
+      "step": 3230
+    },
+    {
+      "epoch": 1.1265646731571628,
+      "grad_norm": 1.5483437776565552,
+      "learning_rate": 1.5747817988112605e-05,
+      "loss": 0.5905,
+      "step": 3240
+    },
+    {
+      "epoch": 1.1300417246175243,
+      "grad_norm": 1.3663296699523926,
+      "learning_rate": 1.5714663677760903e-05,
+      "loss": 0.5954,
+      "step": 3250
+    },
+    {
+      "epoch": 1.133518776077886,
+      "grad_norm": 1.6535637378692627,
+      "learning_rate": 1.568141582535367e-05,
+      "loss": 0.5896,
+      "step": 3260
+    },
+    {
+      "epoch": 1.1369958275382475,
+      "grad_norm": 1.4388856887817383,
+      "learning_rate": 1.56480749751176e-05,
+      "loss": 0.5717,
+      "step": 3270
+    },
+    {
+      "epoch": 1.1404728789986092,
+      "grad_norm": 1.6808990240097046,
+      "learning_rate": 1.561464167280162e-05,
+      "loss": 0.5916,
+      "step": 3280
+    },
+    {
+      "epoch": 1.1439499304589709,
+      "grad_norm": 1.5237574577331543,
+      "learning_rate": 1.5581116465668017e-05,
+      "loss": 0.5693,
+      "step": 3290
+    },
+    {
+      "epoch": 1.1474269819193323,
+      "grad_norm": 1.5086084604263306,
+      "learning_rate": 1.5547499902483424e-05,
+      "loss": 0.598,
+      "step": 3300
+    },
+    {
+      "epoch": 1.150904033379694,
+      "grad_norm": 1.4591008424758911,
+      "learning_rate": 1.5513792533509867e-05,
+      "loss": 0.6011,
+      "step": 3310
+    },
+    {
+      "epoch": 1.1543810848400557,
+      "grad_norm": 1.7214370965957642,
+      "learning_rate": 1.5479994910495757e-05,
+      "loss": 0.5878,
+      "step": 3320
+    },
+    {
+      "epoch": 1.1578581363004172,
+      "grad_norm": 1.6232069730758667,
+      "learning_rate": 1.5446107586666847e-05,
+      "loss": 0.5811,
+      "step": 3330
+    },
+    {
+      "epoch": 1.161335187760779,
+      "grad_norm": 1.5515087842941284,
+      "learning_rate": 1.5412131116717185e-05,
+      "loss": 0.5976,
+      "step": 3340
+    },
+    {
+      "epoch": 1.1648122392211404,
+      "grad_norm": 1.4491888284683228,
+      "learning_rate": 1.537806605680003e-05,
+      "loss": 0.5702,
+      "step": 3350
+    },
+    {
+      "epoch": 1.1682892906815021,
+      "grad_norm": 1.5503897666931152,
+      "learning_rate": 1.534391296451875e-05,
+      "loss": 0.6055,
+      "step": 3360
+    },
+    {
+      "epoch": 1.1717663421418636,
+      "grad_norm": 1.4666029214859009,
+      "learning_rate": 1.5309672398917705e-05,
+      "loss": 0.5793,
+      "step": 3370
+    },
+    {
+      "epoch": 1.1752433936022253,
+      "grad_norm": 1.533703088760376,
+      "learning_rate": 1.5275344920473075e-05,
+      "loss": 0.5824,
+      "step": 3380
+    },
+    {
+      "epoch": 1.178720445062587,
+      "grad_norm": 1.5947829484939575,
+      "learning_rate": 1.5240931091083699e-05,
+      "loss": 0.5811,
+      "step": 3390
+    },
+    {
+      "epoch": 1.1821974965229485,
+      "grad_norm": 1.6678282022476196,
+      "learning_rate": 1.520643147406188e-05,
+      "loss": 0.5731,
+      "step": 3400
+    },
+    {
+      "epoch": 1.1856745479833102,
+      "grad_norm": 1.584175944328308,
+      "learning_rate": 1.5171846634124152e-05,
+      "loss": 0.5814,
+      "step": 3410
+    },
+    {
+      "epoch": 1.1891515994436719,
+      "grad_norm": 1.385772466659546,
+      "learning_rate": 1.5137177137382053e-05,
+      "loss": 0.5766,
+      "step": 3420
+    },
+    {
+      "epoch": 1.1926286509040334,
+      "grad_norm": 1.5849043130874634,
+      "learning_rate": 1.5102423551332842e-05,
+      "loss": 0.602,
+      "step": 3430
+    },
+    {
+      "epoch": 1.196105702364395,
+      "grad_norm": 1.4226797819137573,
+      "learning_rate": 1.5067586444850224e-05,
+      "loss": 0.5779,
+      "step": 3440
+    },
+    {
+      "epoch": 1.1995827538247565,
+      "grad_norm": 1.6716769933700562,
+      "learning_rate": 1.5032666388175016e-05,
+      "loss": 0.5635,
+      "step": 3450
+    },
+    {
+      "epoch": 1.2030598052851182,
+      "grad_norm": 1.9717515707015991,
+      "learning_rate": 1.4997663952905854e-05,
+      "loss": 0.5844,
+      "step": 3460
+    },
+    {
+      "epoch": 1.2065368567454797,
+      "grad_norm": 1.5303245782852173,
+      "learning_rate": 1.496257971198979e-05,
+      "loss": 0.6166,
+      "step": 3470
+    },
+    {
+      "epoch": 1.2100139082058414,
+      "grad_norm": 1.5006227493286133,
+      "learning_rate": 1.492741423971294e-05,
+      "loss": 0.5823,
+      "step": 3480
+    },
+    {
+      "epoch": 1.2134909596662031,
+      "grad_norm": 1.4761695861816406,
+      "learning_rate": 1.4892168111691081e-05,
+      "loss": 0.5886,
+      "step": 3490
+    },
+    {
+      "epoch": 1.2169680111265646,
+      "grad_norm": 1.337805151939392,
+      "learning_rate": 1.4856841904860219e-05,
+      "loss": 0.5726,
+      "step": 3500
+    },
+    {
+      "epoch": 1.2204450625869263,
+      "grad_norm": 1.5253628492355347,
+      "learning_rate": 1.482143619746716e-05,
+      "loss": 0.5816,
+      "step": 3510
+    },
+    {
+      "epoch": 1.223922114047288,
+      "grad_norm": 1.4669066667556763,
+      "learning_rate": 1.4785951569060037e-05,
+      "loss": 0.5946,
+      "step": 3520
+    },
+    {
+      "epoch": 1.2273991655076495,
+      "grad_norm": 1.4548615217208862,
+      "learning_rate": 1.4750388600478813e-05,
+      "loss": 0.5986,
+      "step": 3530
+    },
+    {
+      "epoch": 1.2308762169680112,
+      "grad_norm": 1.6389104127883911,
+      "learning_rate": 1.47147478738458e-05,
+      "loss": 0.6082,
+      "step": 3540
+    },
+    {
+      "epoch": 1.2343532684283727,
+      "grad_norm": 1.554642677307129,
+      "learning_rate": 1.4679029972556099e-05,
+      "loss": 0.6057,
+      "step": 3550
+    },
+    {
+      "epoch": 1.2378303198887344,
+      "grad_norm": 1.360973834991455,
+      "learning_rate": 1.4643235481268069e-05,
+      "loss": 0.595,
+      "step": 3560
+    },
+    {
+      "epoch": 1.2413073713490959,
+      "grad_norm": 1.5446321964263916,
+      "learning_rate": 1.4607364985893768e-05,
+      "loss": 0.5754,
+      "step": 3570
+    },
+    {
+      "epoch": 1.2447844228094576,
+      "grad_norm": 1.5199226140975952,
+      "learning_rate": 1.4571419073589326e-05,
+      "loss": 0.5798,
+      "step": 3580
+    },
+    {
+      "epoch": 1.2482614742698193,
+      "grad_norm": 1.536752462387085,
+      "learning_rate": 1.4535398332745375e-05,
+      "loss": 0.6073,
+      "step": 3590
+    },
+    {
+      "epoch": 1.2517385257301807,
+      "grad_norm": 1.536981463432312,
+      "learning_rate": 1.4499303352977381e-05,
+      "loss": 0.6092,
+      "step": 3600
+    },
+    {
+      "epoch": 1.2552155771905424,
+      "grad_norm": 1.6258337497711182,
+      "learning_rate": 1.446313472511602e-05,
+      "loss": 0.5852,
+      "step": 3610
+    },
+    {
+      "epoch": 1.2586926286509041,
+      "grad_norm": 1.5183357000350952,
+      "learning_rate": 1.4426893041197509e-05,
+      "loss": 0.5733,
+      "step": 3620
+    },
+    {
+      "epoch": 1.2621696801112656,
+      "grad_norm": 1.550432562828064,
+      "learning_rate": 1.4390578894453882e-05,
+      "loss": 0.6243,
+      "step": 3630
+    },
+    {
+      "epoch": 1.2656467315716273,
+      "grad_norm": 1.5437302589416504,
+      "learning_rate": 1.4354192879303312e-05,
+      "loss": 0.5858,
+      "step": 3640
+    },
+    {
+      "epoch": 1.2691237830319888,
+      "grad_norm": 1.5654441118240356,
+      "learning_rate": 1.4317735591340364e-05,
+      "loss": 0.5805,
+      "step": 3650
+    },
+    {
+      "epoch": 1.2726008344923505,
+      "grad_norm": 1.5067358016967773,
+      "learning_rate": 1.428120762732626e-05,
+      "loss": 0.5793,
+      "step": 3660
+    },
+    {
+      "epoch": 1.276077885952712,
+      "grad_norm": 1.4827576875686646,
+      "learning_rate": 1.42446095851791e-05,
+      "loss": 0.6264,
+      "step": 3670
+    },
+    {
+      "epoch": 1.2795549374130737,
+      "grad_norm": 1.4805840253829956,
+      "learning_rate": 1.420794206396407e-05,
+      "loss": 0.5718,
+      "step": 3680
+    },
+    {
+      "epoch": 1.2830319888734354,
+      "grad_norm": 1.5627515316009521,
+      "learning_rate": 1.4171205663883654e-05,
+      "loss": 0.5908,
+      "step": 3690
+    },
+    {
+      "epoch": 1.2865090403337969,
+      "grad_norm": 1.607245922088623,
+      "learning_rate": 1.413440098626779e-05,
+      "loss": 0.6021,
+      "step": 3700
+    },
+    {
+      "epoch": 1.2899860917941586,
+      "grad_norm": 1.563786506652832,
+      "learning_rate": 1.4097528633564038e-05,
+      "loss": 0.6064,
+      "step": 3710
+    },
+    {
+      "epoch": 1.2934631432545203,
+      "grad_norm": 1.678817629814148,
+      "learning_rate": 1.4060589209327724e-05,
+      "loss": 0.5899,
+      "step": 3720
+    },
+    {
+      "epoch": 1.2969401947148818,
+      "grad_norm": 1.5542892217636108,
+      "learning_rate": 1.4023583318212043e-05,
+      "loss": 0.5695,
+      "step": 3730
+    },
+    {
+      "epoch": 1.3004172461752435,
+      "grad_norm": 6.564274311065674,
+      "learning_rate": 1.3986511565958177e-05,
+      "loss": 0.5554,
+      "step": 3740
+    },
+    {
+      "epoch": 1.303894297635605,
+      "grad_norm": 1.7504240274429321,
+      "learning_rate": 1.3949374559385367e-05,
+      "loss": 0.5914,
+      "step": 3750
+    },
+    {
+      "epoch": 1.3073713490959666,
+      "grad_norm": 1.7181627750396729,
+      "learning_rate": 1.3912172906381e-05,
+      "loss": 0.6237,
+      "step": 3760
+    },
+    {
+      "epoch": 1.3108484005563281,
+      "grad_norm": 1.511731505393982,
+      "learning_rate": 1.3874907215890637e-05,
+      "loss": 0.6057,
+      "step": 3770
+    },
+    {
+      "epoch": 1.3143254520166898,
+      "grad_norm": 1.4560339450836182,
+      "learning_rate": 1.3837578097908055e-05,
+      "loss": 0.5796,
+      "step": 3780
+    },
+    {
+      "epoch": 1.3178025034770515,
+      "grad_norm": 1.424481749534607,
+      "learning_rate": 1.3800186163465265e-05,
+      "loss": 0.5939,
+      "step": 3790
+    },
+    {
+      "epoch": 1.321279554937413,
+      "grad_norm": 1.4849761724472046,
+      "learning_rate": 1.3762732024622503e-05,
+      "loss": 0.5564,
+      "step": 3800
+    },
+    {
+      "epoch": 1.3247566063977747,
+      "grad_norm": 1.6058564186096191,
+      "learning_rate": 1.3725216294458216e-05,
+      "loss": 0.5827,
+      "step": 3810
+    },
+    {
+      "epoch": 1.3282336578581364,
+      "grad_norm": 1.4608303308486938,
+      "learning_rate": 1.368763958705903e-05,
+      "loss": 0.605,
+      "step": 3820
+    },
+    {
+      "epoch": 1.3317107093184979,
+      "grad_norm": 1.5738255977630615,
+      "learning_rate": 1.3650002517509684e-05,
+      "loss": 0.5872,
+      "step": 3830
+    },
+    {
+      "epoch": 1.3351877607788596,
+      "grad_norm": 1.5521684885025024,
+      "learning_rate": 1.361230570188298e-05,
+      "loss": 0.5809,
+      "step": 3840
+    },
+    {
+      "epoch": 1.338664812239221,
+      "grad_norm": 1.5345361232757568,
+      "learning_rate": 1.3574549757229688e-05,
+      "loss": 0.6003,
+      "step": 3850
+    },
+    {
+      "epoch": 1.3421418636995828,
+      "grad_norm": 1.6283280849456787,
+      "learning_rate": 1.3536735301568447e-05,
+      "loss": 0.5973,
+      "step": 3860
+    },
+    {
+      "epoch": 1.3456189151599443,
+      "grad_norm": 1.6617012023925781,
+      "learning_rate": 1.3498862953875654e-05,
+      "loss": 0.6063,
+      "step": 3870
+    },
+    {
+      "epoch": 1.349095966620306,
+      "grad_norm": 1.9414786100387573,
+      "learning_rate": 1.3460933334075317e-05,
+      "loss": 0.6088,
+      "step": 3880
+    },
+    {
+      "epoch": 1.3525730180806677,
+      "grad_norm": 1.542784333229065,
+      "learning_rate": 1.3422947063028928e-05,
+      "loss": 0.6073,
+      "step": 3890
+    },
+    {
+      "epoch": 1.3560500695410291,
+      "grad_norm": 1.438861608505249,
+      "learning_rate": 1.338490476252529e-05,
+      "loss": 0.6119,
+      "step": 3900
+    },
+    {
+      "epoch": 1.3595271210013908,
+      "grad_norm": 1.5155811309814453,
+      "learning_rate": 1.3346807055270335e-05,
+      "loss": 0.5798,
+      "step": 3910
+    },
+    {
+      "epoch": 1.3630041724617525,
+      "grad_norm": 1.7073251008987427,
+      "learning_rate": 1.3308654564876938e-05,
+      "loss": 0.6095,
+      "step": 3920
+    },
+    {
+      "epoch": 1.366481223922114,
+      "grad_norm": 1.6175742149353027,
+      "learning_rate": 1.3270447915854709e-05,
+      "loss": 0.5984,
+      "step": 3930
+    },
+    {
+      "epoch": 1.3699582753824757,
+      "grad_norm": 1.5438671112060547,
+      "learning_rate": 1.3232187733599768e-05,
+      "loss": 0.6184,
+      "step": 3940
+    },
+    {
+      "epoch": 1.3734353268428372,
+      "grad_norm": 1.5229095220565796,
+      "learning_rate": 1.3193874644384505e-05,
+      "loss": 0.5971,
+      "step": 3950
+    },
+    {
+      "epoch": 1.376912378303199,
+      "grad_norm": 1.5525486469268799,
+      "learning_rate": 1.3155509275347337e-05,
+      "loss": 0.5675,
+      "step": 3960
+    },
+    {
+      "epoch": 1.3803894297635604,
+      "grad_norm": 1.4961704015731812,
+      "learning_rate": 1.3117092254482433e-05,
+      "loss": 0.5879,
+      "step": 3970
+    },
+    {
+      "epoch": 1.383866481223922,
+      "grad_norm": 1.46124267578125,
+      "learning_rate": 1.3078624210629438e-05,
+      "loss": 0.6096,
+      "step": 3980
+    },
+    {
+      "epoch": 1.3873435326842838,
+      "grad_norm": 1.5501396656036377,
+      "learning_rate": 1.3040105773463192e-05,
+      "loss": 0.5721,
+      "step": 3990
+    },
+    {
+      "epoch": 1.3908205841446453,
+      "grad_norm": 1.4876134395599365,
+      "learning_rate": 1.30015375734834e-05,
+      "loss": 0.5979,
+      "step": 4000
+    },
+    {
+      "epoch": 1.394297635605007,
+      "grad_norm": 1.9500658512115479,
+      "learning_rate": 1.296292024200433e-05,
+      "loss": 0.5874,
+      "step": 4010
+    },
+    {
+      "epoch": 1.3977746870653687,
+      "grad_norm": 1.6674953699111938,
+      "learning_rate": 1.292425441114447e-05,
+      "loss": 0.5859,
+      "step": 4020
+    },
+    {
+      "epoch": 1.4012517385257302,
+      "grad_norm": 1.4582736492156982,
+      "learning_rate": 1.2885540713816181e-05,
+      "loss": 0.5856,
+      "step": 4030
+    },
+    {
+      "epoch": 1.4047287899860919,
+      "grad_norm": 1.5231198072433472,
+      "learning_rate": 1.2846779783715347e-05,
+      "loss": 0.5864,
+      "step": 4040
+    },
+    {
+      "epoch": 1.4082058414464533,
+      "grad_norm": 1.4892988204956055,
+      "learning_rate": 1.2807972255310985e-05,
+      "loss": 0.6136,
+      "step": 4050
+    },
+    {
+      "epoch": 1.411682892906815,
+      "grad_norm": 1.465019941329956,
+      "learning_rate": 1.2769118763834878e-05,
+      "loss": 0.6161,
+      "step": 4060
+    },
+    {
+      "epoch": 1.4151599443671765,
+      "grad_norm": 1.5238908529281616,
+      "learning_rate": 1.2730219945271165e-05,
+      "loss": 0.5802,
+      "step": 4070
+    },
+    {
+      "epoch": 1.4186369958275382,
+      "grad_norm": 2.0098044872283936,
+      "learning_rate": 1.2691276436345932e-05,
+      "loss": 0.5969,
+      "step": 4080
+    },
+    {
+      "epoch": 1.4221140472879,
+      "grad_norm": 1.650890588760376,
+      "learning_rate": 1.2652288874516794e-05,
+      "loss": 0.553,
+      "step": 4090
+    },
+    {
+      "epoch": 1.4255910987482614,
+      "grad_norm": 1.570281744003296,
+      "learning_rate": 1.2613257897962456e-05,
+      "loss": 0.5695,
+      "step": 4100
+    },
+    {
+      "epoch": 1.429068150208623,
+      "grad_norm": 1.5810602903366089,
+      "learning_rate": 1.2574184145572277e-05,
+      "loss": 0.5803,
+      "step": 4110
+    },
+    {
+      "epoch": 1.4325452016689848,
+      "grad_norm": 1.6350195407867432,
+      "learning_rate": 1.2535068256935788e-05,
+      "loss": 0.6072,
+      "step": 4120
+    },
+    {
+      "epoch": 1.4360222531293463,
+      "grad_norm": 1.4450207948684692,
+      "learning_rate": 1.249591087233225e-05,
+      "loss": 0.5516,
+      "step": 4130
+    },
+    {
+      "epoch": 1.439499304589708,
+      "grad_norm": 1.5011627674102783,
+      "learning_rate": 1.2456712632720167e-05,
+      "loss": 0.5666,
+      "step": 4140
+    },
+    {
+      "epoch": 1.4429763560500695,
+      "grad_norm": 1.739227056503296,
+      "learning_rate": 1.2417474179726776e-05,
+      "loss": 0.5903,
+      "step": 4150
+    },
+    {
+      "epoch": 1.4464534075104312,
+      "grad_norm": 1.5206959247589111,
+      "learning_rate": 1.2378196155637571e-05,
+      "loss": 0.5722,
+      "step": 4160
+    },
+    {
+      "epoch": 1.4499304589707926,
+      "grad_norm": 1.4351691007614136,
+      "learning_rate": 1.2338879203385767e-05,
+      "loss": 0.5936,
+      "step": 4170
+    },
+    {
+      "epoch": 1.4534075104311543,
+      "grad_norm": 1.3733693361282349,
+      "learning_rate": 1.2299523966541789e-05,
+      "loss": 0.59,
+      "step": 4180
+    },
+    {
+      "epoch": 1.456884561891516,
+      "grad_norm": 1.6201534271240234,
+      "learning_rate": 1.2260131089302734e-05,
+      "loss": 0.6157,
+      "step": 4190
+    },
+    {
+      "epoch": 1.4603616133518775,
+      "grad_norm": 1.4009215831756592,
+      "learning_rate": 1.2220701216481825e-05,
+      "loss": 0.5947,
+      "step": 4200
+    },
+    {
+      "epoch": 1.4638386648122392,
+      "grad_norm": 1.501211166381836,
+      "learning_rate": 1.2181234993497859e-05,
+      "loss": 0.5712,
+      "step": 4210
+    },
+    {
+      "epoch": 1.467315716272601,
+      "grad_norm": 1.574884295463562,
+      "learning_rate": 1.214173306636464e-05,
+      "loss": 0.5804,
+      "step": 4220
+    },
+    {
+      "epoch": 1.4707927677329624,
+      "grad_norm": 1.4305651187896729,
+      "learning_rate": 1.2102196081680408e-05,
+      "loss": 0.5978,
+      "step": 4230
+    },
+    {
+      "epoch": 1.4742698191933241,
+      "grad_norm": 1.5670166015625,
+      "learning_rate": 1.2062624686617248e-05,
+      "loss": 0.5715,
+      "step": 4240
+    },
+    {
+      "epoch": 1.4777468706536856,
+      "grad_norm": 1.4854615926742554,
+      "learning_rate": 1.20230195289105e-05,
+      "loss": 0.5997,
+      "step": 4250
+    },
+    {
+      "epoch": 1.4812239221140473,
+      "grad_norm": 1.6153312921524048,
+      "learning_rate": 1.1983381256848166e-05,
+      "loss": 0.5471,
+      "step": 4260
+    },
+    {
+      "epoch": 1.4847009735744088,
+      "grad_norm": 1.4996991157531738,
+      "learning_rate": 1.1943710519260275e-05,
+      "loss": 0.5907,
+      "step": 4270
+    },
+    {
+      "epoch": 1.4881780250347705,
+      "grad_norm": 1.429539442062378,
+      "learning_rate": 1.1904007965508287e-05,
+      "loss": 0.5789,
+      "step": 4280
+    },
+    {
+      "epoch": 1.4916550764951322,
+      "grad_norm": 1.5480320453643799,
+      "learning_rate": 1.1864274245474454e-05,
+      "loss": 0.571,
+      "step": 4290
+    },
+    {
+      "epoch": 1.4951321279554937,
+      "grad_norm": 1.3919758796691895,
+      "learning_rate": 1.182451000955118e-05,
+      "loss": 0.5673,
+      "step": 4300
+    },
+    {
+      "epoch": 1.4986091794158554,
+      "grad_norm": 1.5538363456726074,
+      "learning_rate": 1.1784715908630378e-05,
+      "loss": 0.583,
+      "step": 4310
+    },
+    {
+      "epoch": 1.502086230876217,
+      "grad_norm": 1.5282542705535889,
+      "learning_rate": 1.1744892594092809e-05,
+      "loss": 0.6183,
+      "step": 4320
+    },
+    {
+      "epoch": 1.5055632823365785,
+      "grad_norm": 1.5380876064300537,
+      "learning_rate": 1.170504071779743e-05,
+      "loss": 0.556,
+      "step": 4330
+    },
+    {
+      "epoch": 1.5090403337969402,
+      "grad_norm": 1.5165767669677734,
+      "learning_rate": 1.1665160932070722e-05,
+      "loss": 0.5714,
+      "step": 4340
+    },
+    {
+      "epoch": 1.512517385257302,
+      "grad_norm": 1.6299424171447754,
+      "learning_rate": 1.1625253889696e-05,
+      "loss": 0.5943,
+      "step": 4350
+    },
+    {
+      "epoch": 1.5159944367176634,
+      "grad_norm": 1.493578314781189,
+      "learning_rate": 1.158532024390275e-05,
+      "loss": 0.5565,
+      "step": 4360
+    },
+    {
+      "epoch": 1.519471488178025,
+      "grad_norm": 1.4570236206054688,
+      "learning_rate": 1.1545360648355901e-05,
+      "loss": 0.5865,
+      "step": 4370
+    },
+    {
+      "epoch": 1.5229485396383866,
+      "grad_norm": 1.4010246992111206,
+      "learning_rate": 1.1505375757145179e-05,
+      "loss": 0.5899,
+      "step": 4380
+    },
+    {
+      "epoch": 1.5264255910987483,
+      "grad_norm": 1.6167325973510742,
+      "learning_rate": 1.1465366224774345e-05,
+      "loss": 0.5954,
+      "step": 4390
+    },
+    {
+      "epoch": 1.5299026425591098,
+      "grad_norm": 1.6242483854293823,
+      "learning_rate": 1.1425332706150516e-05,
+      "loss": 0.5961,
+      "step": 4400
+    },
+    {
+      "epoch": 1.5333796940194715,
+      "grad_norm": 1.4537858963012695,
+      "learning_rate": 1.1385275856573431e-05,
+      "loss": 0.5797,
+      "step": 4410
+    },
+    {
+      "epoch": 1.5368567454798332,
+      "grad_norm": 1.4809147119522095,
+      "learning_rate": 1.1345196331724725e-05,
+      "loss": 0.5615,
+      "step": 4420
+    },
+    {
+      "epoch": 1.5403337969401947,
+      "grad_norm": 1.449339747428894,
+      "learning_rate": 1.1305094787657206e-05,
+      "loss": 0.5885,
+      "step": 4430
+    },
+    {
+      "epoch": 1.5438108484005564,
+      "grad_norm": 1.6153600215911865,
+      "learning_rate": 1.1264971880784114e-05,
+      "loss": 0.628,
+      "step": 4440
+    },
+    {
+      "epoch": 1.547287899860918,
+      "grad_norm": 1.3971196413040161,
+      "learning_rate": 1.1224828267868355e-05,
+      "loss": 0.5684,
+      "step": 4450
+    },
+    {
+      "epoch": 1.5507649513212796,
+      "grad_norm": 1.458151936531067,
+      "learning_rate": 1.1184664606011785e-05,
+      "loss": 0.573,
+      "step": 4460
+    },
+    {
+      "epoch": 1.554242002781641,
+      "grad_norm": 1.58401620388031,
+      "learning_rate": 1.1144481552644428e-05,
+      "loss": 0.6272,
+      "step": 4470
+    },
+    {
+      "epoch": 1.5577190542420027,
+      "grad_norm": 1.5694257020950317,
+      "learning_rate": 1.1104279765513721e-05,
+      "loss": 0.5737,
+      "step": 4480
+    },
+    {
+      "epoch": 1.5611961057023644,
+      "grad_norm": 1.5449283123016357,
+      "learning_rate": 1.106405990267376e-05,
+      "loss": 0.5788,
+      "step": 4490
+    },
+    {
+      "epoch": 1.564673157162726,
+      "grad_norm": 1.5892006158828735,
+      "learning_rate": 1.1023822622474507e-05,
+      "loss": 0.6108,
+      "step": 4500
+    },
+    {
+      "epoch": 1.5681502086230876,
+      "grad_norm": 1.5775792598724365,
+      "learning_rate": 1.0983568583551035e-05,
+      "loss": 0.5708,
+      "step": 4510
+    },
+    {
+      "epoch": 1.5716272600834493,
+      "grad_norm": 1.580875277519226,
+      "learning_rate": 1.0943298444812722e-05,
+      "loss": 0.5561,
+      "step": 4520
+    },
+    {
+      "epoch": 1.5751043115438108,
+      "grad_norm": 1.486342430114746,
+      "learning_rate": 1.0903012865432497e-05,
+      "loss": 0.5945,
+      "step": 4530
+    },
+    {
+      "epoch": 1.5785813630041725,
+      "grad_norm": 1.3929870128631592,
+      "learning_rate": 1.0862712504836026e-05,
+      "loss": 0.5585,
+      "step": 4540
+    },
+    {
+      "epoch": 1.5820584144645342,
+      "grad_norm": 1.4984560012817383,
+      "learning_rate": 1.0822398022690925e-05,
+      "loss": 0.5576,
+      "step": 4550
+    },
+    {
+      "epoch": 1.5855354659248957,
+      "grad_norm": 1.6191108226776123,
+      "learning_rate": 1.0782070078895965e-05,
+      "loss": 0.5812,
+      "step": 4560
+    },
+    {
+      "epoch": 1.5890125173852572,
+      "grad_norm": 1.444085717201233,
+      "learning_rate": 1.074172933357026e-05,
+      "loss": 0.5719,
+      "step": 4570
+    },
+    {
+      "epoch": 1.5924895688456189,
+      "grad_norm": 1.6037352085113525,
+      "learning_rate": 1.0701376447042482e-05,
+      "loss": 0.5935,
+      "step": 4580
+    },
+    {
+      "epoch": 1.5959666203059806,
+      "grad_norm": 1.5845577716827393,
+      "learning_rate": 1.066101207984003e-05,
+      "loss": 0.5898,
+      "step": 4590
+    },
+    {
+      "epoch": 1.599443671766342,
+      "grad_norm": 1.4457982778549194,
+      "learning_rate": 1.062063689267823e-05,
+      "loss": 0.541,
+      "step": 4600
+    },
+    {
+      "epoch": 1.6029207232267038,
+      "grad_norm": 1.5745394229888916,
+      "learning_rate": 1.0580251546449518e-05,
+      "loss": 0.5387,
+      "step": 4610
+    },
+    {
+      "epoch": 1.6063977746870655,
+      "grad_norm": 1.4566587209701538,
+      "learning_rate": 1.0539856702212623e-05,
+      "loss": 0.5795,
+      "step": 4620
+    },
+    {
+      "epoch": 1.609874826147427,
+      "grad_norm": 1.4010239839553833,
+      "learning_rate": 1.0499453021181743e-05,
+      "loss": 0.5838,
+      "step": 4630
+    },
+    {
+      "epoch": 1.6133518776077886,
+      "grad_norm": 1.5083566904067993,
+      "learning_rate": 1.0459041164715724e-05,
+      "loss": 0.5905,
+      "step": 4640
+    },
+    {
+      "epoch": 1.6168289290681503,
+      "grad_norm": 1.5915400981903076,
+      "learning_rate": 1.0418621794307233e-05,
+      "loss": 0.5489,
+      "step": 4650
+    },
+    {
+      "epoch": 1.6203059805285118,
+      "grad_norm": 1.6947846412658691,
+      "learning_rate": 1.0378195571571935e-05,
+      "loss": 0.5776,
+      "step": 4660
+    },
+    {
+      "epoch": 1.6237830319888733,
+      "grad_norm": 1.2892073392868042,
+      "learning_rate": 1.033776315823765e-05,
+      "loss": 0.5651,
+      "step": 4670
+    },
+    {
+      "epoch": 1.627260083449235,
+      "grad_norm": 1.4766916036605835,
+      "learning_rate": 1.029732521613354e-05,
+      "loss": 0.5594,
+      "step": 4680
+    },
+    {
+      "epoch": 1.6307371349095967,
+      "grad_norm": 1.666880488395691,
+      "learning_rate": 1.0256882407179262e-05,
+      "loss": 0.5694,
+      "step": 4690
+    },
+    {
+      "epoch": 1.6342141863699582,
+      "grad_norm": 1.5797456502914429,
+      "learning_rate": 1.021643539337414e-05,
+      "loss": 0.5951,
+      "step": 4700
+    },
+    {
+      "epoch": 1.63769123783032,
+      "grad_norm": 1.5479556322097778,
+      "learning_rate": 1.0175984836786325e-05,
+      "loss": 0.563,
+      "step": 4710
+    },
+    {
+      "epoch": 1.6411682892906816,
+      "grad_norm": 1.4872573614120483,
+      "learning_rate": 1.0135531399541953e-05,
+      "loss": 0.5772,
+      "step": 4720
+    },
+    {
+      "epoch": 1.644645340751043,
+      "grad_norm": 1.4076658487319946,
+      "learning_rate": 1.0095075743814325e-05,
+      "loss": 0.5915,
+      "step": 4730
+    },
+    {
+      "epoch": 1.6481223922114048,
+      "grad_norm": 1.3710988759994507,
+      "learning_rate": 1.0054618531813046e-05,
+      "loss": 0.5805,
+      "step": 4740
+    },
+    {
+      "epoch": 1.6515994436717665,
+      "grad_norm": 1.6406593322753906,
+      "learning_rate": 1.0014160425773193e-05,
+      "loss": 0.595,
+      "step": 4750
+    },
+    {
+      "epoch": 1.655076495132128,
+      "grad_norm": 1.4221357107162476,
+      "learning_rate": 9.973702087944493e-06,
+      "loss": 0.571,
+      "step": 4760
+    },
+    {
+      "epoch": 1.6585535465924894,
+      "grad_norm": 1.4262938499450684,
+      "learning_rate": 9.93324418058045e-06,
+      "loss": 0.5597,
+      "step": 4770
+    },
+    {
+      "epoch": 1.6620305980528511,
+      "grad_norm": 1.5167871713638306,
+      "learning_rate": 9.892787365927531e-06,
+      "loss": 0.5677,
+      "step": 4780
+    },
+    {
+      "epoch": 1.6655076495132128,
+      "grad_norm": 1.407149314880371,
+      "learning_rate": 9.852332306214311e-06,
+      "loss": 0.6161,
+      "step": 4790
+    },
+    {
+      "epoch": 1.6689847009735743,
+      "grad_norm": 1.5203973054885864,
+      "learning_rate": 9.811879663640647e-06,
+      "loss": 0.5644,
+      "step": 4800
+    },
+    {
+      "epoch": 1.672461752433936,
+      "grad_norm": 1.5746958255767822,
+      "learning_rate": 9.771430100366823e-06,
+      "loss": 0.5756,
+      "step": 4810
+    },
+    {
+      "epoch": 1.6759388038942977,
+      "grad_norm": 1.5510128736495972,
+      "learning_rate": 9.73098427850272e-06,
+      "loss": 0.5755,
+      "step": 4820
+    },
+    {
+      "epoch": 1.6794158553546592,
+      "grad_norm": 1.5657026767730713,
+      "learning_rate": 9.69054286009698e-06,
+      "loss": 0.5957,
+      "step": 4830
+    },
+    {
+      "epoch": 1.682892906815021,
+      "grad_norm": 1.643454670906067,
+      "learning_rate": 9.650106507126163e-06,
+      "loss": 0.5697,
+      "step": 4840
+    },
+    {
+      "epoch": 1.6863699582753826,
+      "grad_norm": 1.3832855224609375,
+      "learning_rate": 9.609675881483914e-06,
+      "loss": 0.5635,
+      "step": 4850
+    },
+    {
+      "epoch": 1.689847009735744,
+      "grad_norm": 1.7504627704620361,
+      "learning_rate": 9.569251644970127e-06,
+      "loss": 0.5564,
+      "step": 4860
+    },
+    {
+      "epoch": 1.6933240611961056,
+      "grad_norm": 1.5367279052734375,
+      "learning_rate": 9.528834459280125e-06,
+      "loss": 0.5796,
+      "step": 4870
+    },
+    {
+      "epoch": 1.6968011126564673,
+      "grad_norm": 1.564568042755127,
+      "learning_rate": 9.488424985993796e-06,
+      "loss": 0.5693,
+      "step": 4880
+    },
+    {
+      "epoch": 1.700278164116829,
+      "grad_norm": 1.424558162689209,
+      "learning_rate": 9.448023886564805e-06,
+      "loss": 0.5737,
+      "step": 4890
+    },
+    {
+      "epoch": 1.7037552155771905,
+      "grad_norm": 1.55270254611969,
+      "learning_rate": 9.407631822309737e-06,
+      "loss": 0.5854,
+      "step": 4900
+    },
+    {
+      "epoch": 1.7072322670375522,
+      "grad_norm": 1.538130760192871,
+      "learning_rate": 9.36724945439729e-06,
+      "loss": 0.5762,
+      "step": 4910
+    },
+    {
+      "epoch": 1.7107093184979139,
+      "grad_norm": 1.6305609941482544,
+      "learning_rate": 9.32687744383744e-06,
+      "loss": 0.5809,
+      "step": 4920
+    },
+    {
+      "epoch": 1.7141863699582753,
+      "grad_norm": 1.5811065435409546,
+      "learning_rate": 9.286516451470621e-06,
+      "loss": 0.5586,
+      "step": 4930
+    },
+    {
+      "epoch": 1.717663421418637,
+      "grad_norm": 1.44260835647583,
+      "learning_rate": 9.24616713795692e-06,
+      "loss": 0.5563,
+      "step": 4940
+    },
+    {
+      "epoch": 1.7211404728789987,
+      "grad_norm": 1.6092342138290405,
+      "learning_rate": 9.205830163765253e-06,
+      "loss": 0.5792,
+      "step": 4950
+    },
+    {
+      "epoch": 1.7246175243393602,
+      "grad_norm": 1.5757039785385132,
+      "learning_rate": 9.16550618916256e-06,
+      "loss": 0.5743,
+      "step": 4960
+    },
+    {
+      "epoch": 1.7280945757997217,
+      "grad_norm": 1.5829275846481323,
+      "learning_rate": 9.125195874202986e-06,
+      "loss": 0.6067,
+      "step": 4970
+    },
+    {
+      "epoch": 1.7315716272600834,
+      "grad_norm": 1.4836721420288086,
+      "learning_rate": 9.08489987871709e-06,
+      "loss": 0.5885,
+      "step": 4980
+    },
+    {
+      "epoch": 1.735048678720445,
+      "grad_norm": 1.4931825399398804,
+      "learning_rate": 9.044618862301035e-06,
+      "loss": 0.5724,
+      "step": 4990
+    },
+    {
+      "epoch": 1.7385257301808066,
+      "grad_norm": 1.5220547914505005,
+      "learning_rate": 9.0043534843058e-06,
+      "loss": 0.5892,
+      "step": 5000
+    },
+    {
+      "epoch": 1.7420027816411683,
+      "grad_norm": 1.5941320657730103,
+      "learning_rate": 8.964104403826378e-06,
+      "loss": 0.5712,
+      "step": 5010
+    },
+    {
+      "epoch": 1.74547983310153,
+      "grad_norm": 1.6049176454544067,
+      "learning_rate": 8.923872279690997e-06,
+      "loss": 0.5674,
+      "step": 5020
+    },
+    {
+      "epoch": 1.7489568845618915,
+      "grad_norm": 1.4640474319458008,
+      "learning_rate": 8.883657770450318e-06,
+      "loss": 0.5717,
+      "step": 5030
+    },
+    {
+      "epoch": 1.7524339360222532,
+      "grad_norm": 1.4230233430862427,
+      "learning_rate": 8.843461534366682e-06,
+      "loss": 0.5655,
+      "step": 5040
+    },
+    {
+      "epoch": 1.7559109874826149,
+      "grad_norm": 1.4291554689407349,
+      "learning_rate": 8.803284229403313e-06,
+      "loss": 0.6003,
+      "step": 5050
+    },
+    {
+      "epoch": 1.7593880389429764,
+      "grad_norm": 1.48020601272583,
+      "learning_rate": 8.763126513213558e-06,
+      "loss": 0.5596,
+      "step": 5060
+    },
+    {
+      "epoch": 1.7628650904033378,
+      "grad_norm": 1.5123621225357056,
+      "learning_rate": 8.72298904313012e-06,
+      "loss": 0.5688,
+      "step": 5070
+    },
+    {
+      "epoch": 1.7663421418636995,
+      "grad_norm": 1.4989289045333862,
+      "learning_rate": 8.682872476154296e-06,
+      "loss": 0.5746,
+      "step": 5080
+    },
+    {
+      "epoch": 1.7698191933240612,
+      "grad_norm": 1.5118212699890137,
+      "learning_rate": 8.642777468945224e-06,
+      "loss": 0.5774,
+      "step": 5090
+    },
+    {
+      "epoch": 1.7732962447844227,
+      "grad_norm": 1.4650695323944092,
+      "learning_rate": 8.60270467780914e-06,
+      "loss": 0.5603,
+      "step": 5100
+    },
+    {
+      "epoch": 1.7767732962447844,
+      "grad_norm": 1.4678314924240112,
+      "learning_rate": 8.562654758688627e-06,
+      "loss": 0.6138,
+      "step": 5110
+    },
+    {
+      "epoch": 1.7802503477051461,
+      "grad_norm": 1.3884979486465454,
+      "learning_rate": 8.522628367151883e-06,
+      "loss": 0.5412,
+      "step": 5120
+    },
+    {
+      "epoch": 1.7837273991655076,
+      "grad_norm": 1.4073148965835571,
+      "learning_rate": 8.48262615838198e-06,
+      "loss": 0.5631,
+      "step": 5130
+    },
+    {
+      "epoch": 1.7872044506258693,
+      "grad_norm": 1.4141111373901367,
+      "learning_rate": 8.442648787166155e-06,
+      "loss": 0.5982,
+      "step": 5140
+    },
+    {
+      "epoch": 1.790681502086231,
+      "grad_norm": 1.464737057685852,
+      "learning_rate": 8.40269690788509e-06,
+      "loss": 0.5854,
+      "step": 5150
+    },
+    {
+      "epoch": 1.7941585535465925,
+      "grad_norm": 1.54302978515625,
+      "learning_rate": 8.36277117450218e-06,
+      "loss": 0.5521,
+      "step": 5160
+    },
+    {
+      "epoch": 1.797635605006954,
+      "grad_norm": 1.5654412508010864,
+      "learning_rate": 8.322872240552866e-06,
+      "loss": 0.5399,
+      "step": 5170
+    },
+    {
+      "epoch": 1.8011126564673157,
+      "grad_norm": 1.6172314882278442,
+      "learning_rate": 8.283000759133885e-06,
+      "loss": 0.5803,
+      "step": 5180
+    },
+    {
+      "epoch": 1.8045897079276774,
+      "grad_norm": 1.50352144241333,
+      "learning_rate": 8.243157382892635e-06,
+      "loss": 0.5634,
+      "step": 5190
+    },
+    {
+      "epoch": 1.8080667593880388,
+      "grad_norm": 1.495595932006836,
+      "learning_rate": 8.203342764016456e-06,
+      "loss": 0.6016,
+      "step": 5200
+    },
+    {
+      "epoch": 1.8115438108484005,
+      "grad_norm": 1.4712079763412476,
+      "learning_rate": 8.163557554221968e-06,
+      "loss": 0.5712,
+      "step": 5210
+    },
+    {
+      "epoch": 1.8150208623087623,
+      "grad_norm": 1.5086045265197754,
+      "learning_rate": 8.1238024047444e-06,
+      "loss": 0.5527,
+      "step": 5220
+    },
+    {
+      "epoch": 1.8184979137691237,
+      "grad_norm": 1.5117202997207642,
+      "learning_rate": 8.084077966326926e-06,
+      "loss": 0.5547,
+      "step": 5230
+    },
+    {
+      "epoch": 1.8219749652294854,
+      "grad_norm": 1.4075812101364136,
+      "learning_rate": 8.044384889210021e-06,
+      "loss": 0.5758,
+      "step": 5240
+    },
+    {
+      "epoch": 1.8254520166898471,
+      "grad_norm": 1.4071241617202759,
+      "learning_rate": 8.004723823120817e-06,
+      "loss": 0.5809,
+      "step": 5250
+    },
+    {
+      "epoch": 1.8289290681502086,
+      "grad_norm": 1.555217981338501,
+      "learning_rate": 7.965095417262459e-06,
+      "loss": 0.5813,
+      "step": 5260
+    },
+    {
+      "epoch": 1.83240611961057,
+      "grad_norm": 1.5502498149871826,
+      "learning_rate": 7.925500320303492e-06,
+      "loss": 0.5695,
+      "step": 5270
+    },
+    {
+      "epoch": 1.8358831710709318,
+      "grad_norm": 1.5924468040466309,
+      "learning_rate": 7.885939180367221e-06,
+      "loss": 0.5546,
+      "step": 5280
+    },
+    {
+      "epoch": 1.8393602225312935,
+      "grad_norm": 1.7237669229507446,
+      "learning_rate": 7.846412645021128e-06,
+      "loss": 0.5762,
+      "step": 5290
+    },
+    {
+      "epoch": 1.842837273991655,
+      "grad_norm": 1.5827271938323975,
+      "learning_rate": 7.806921361266256e-06,
+      "loss": 0.5575,
+      "step": 5300
+    },
+    {
+      "epoch": 1.8463143254520167,
+      "grad_norm": 1.4798219203948975,
+      "learning_rate": 7.767465975526626e-06,
+      "loss": 0.5514,
+      "step": 5310
+    },
+    {
+      "epoch": 1.8497913769123784,
+      "grad_norm": 1.5293381214141846,
+      "learning_rate": 7.728047133638655e-06,
+      "loss": 0.5767,
+      "step": 5320
+    },
+    {
+      "epoch": 1.8532684283727399,
+      "grad_norm": 1.4558959007263184,
+      "learning_rate": 7.688665480840562e-06,
+      "loss": 0.5598,
+      "step": 5330
+    },
+    {
+      "epoch": 1.8567454798331016,
+      "grad_norm": 1.456066370010376,
+      "learning_rate": 7.64932166176185e-06,
+      "loss": 0.5787,
+      "step": 5340
+    },
+    {
+      "epoch": 1.8602225312934633,
+      "grad_norm": 1.5269767045974731,
+      "learning_rate": 7.610016320412718e-06,
+      "loss": 0.5781,
+      "step": 5350
+    },
+    {
+      "epoch": 1.8636995827538247,
+      "grad_norm": 1.4626280069351196,
+      "learning_rate": 7.570750100173537e-06,
+      "loss": 0.5856,
+      "step": 5360
+    },
+    {
+      "epoch": 1.8671766342141862,
+      "grad_norm": 1.4652131795883179,
+      "learning_rate": 7.531523643784312e-06,
+      "loss": 0.5429,
+      "step": 5370
+    },
+    {
+      "epoch": 1.870653685674548,
+      "grad_norm": 1.525390625,
+      "learning_rate": 7.492337593334156e-06,
+      "loss": 0.5661,
+      "step": 5380
+    },
+    {
+      "epoch": 1.8741307371349096,
+      "grad_norm": 1.5893056392669678,
+      "learning_rate": 7.453192590250795e-06,
+      "loss": 0.5505,
+      "step": 5390
+    },
+    {
+      "epoch": 1.877607788595271,
+      "grad_norm": 1.4114181995391846,
+      "learning_rate": 7.4140892752900555e-06,
+      "loss": 0.5897,
+      "step": 5400
+    },
+    {
+      "epoch": 1.8810848400556328,
+      "grad_norm": 1.4983806610107422,
+      "learning_rate": 7.37502828852538e-06,
+      "loss": 0.5842,
+      "step": 5410
+    },
+    {
+      "epoch": 1.8845618915159945,
+      "grad_norm": 1.4965555667877197,
+      "learning_rate": 7.336010269337357e-06,
+      "loss": 0.5869,
+      "step": 5420
+    },
+    {
+      "epoch": 1.888038942976356,
+      "grad_norm": 1.5193657875061035,
+      "learning_rate": 7.297035856403234e-06,
+      "loss": 0.5792,
+      "step": 5430
+    },
+    {
+      "epoch": 1.8915159944367177,
+      "grad_norm": 1.4535385370254517,
+      "learning_rate": 7.25810568768649e-06,
+      "loss": 0.5771,
+      "step": 5440
+    },
+    {
+      "epoch": 1.8949930458970794,
+      "grad_norm": 1.4724664688110352,
+      "learning_rate": 7.219220400426376e-06,
+      "loss": 0.5559,
+      "step": 5450
+    },
+    {
+      "epoch": 1.8984700973574409,
+      "grad_norm": 1.4682315587997437,
+      "learning_rate": 7.180380631127486e-06,
+      "loss": 0.5337,
+      "step": 5460
+    },
+    {
+      "epoch": 1.9019471488178024,
+      "grad_norm": 1.4762859344482422,
+      "learning_rate": 7.141587015549355e-06,
+      "loss": 0.6109,
+      "step": 5470
+    },
+    {
+      "epoch": 1.9054242002781643,
+      "grad_norm": 1.4881417751312256,
+      "learning_rate": 7.102840188696013e-06,
+      "loss": 0.5803,
+      "step": 5480
+    },
+    {
+      "epoch": 1.9089012517385258,
+      "grad_norm": 1.4660687446594238,
+      "learning_rate": 7.0641407848056355e-06,
+      "loss": 0.5624,
+      "step": 5490
+    },
+    {
+      "epoch": 1.9123783031988872,
+      "grad_norm": 1.4473921060562134,
+      "learning_rate": 7.0254894373401325e-06,
+      "loss": 0.5769,
+      "step": 5500
+    },
+    {
+      "epoch": 1.915855354659249,
+      "grad_norm": 1.466530203819275,
+      "learning_rate": 6.986886778974796e-06,
+      "loss": 0.5545,
+      "step": 5510
+    },
+    {
+      "epoch": 1.9193324061196106,
+      "grad_norm": 1.6951309442520142,
+      "learning_rate": 6.948333441587935e-06,
+      "loss": 0.5877,
+      "step": 5520
+    },
+    {
+      "epoch": 1.9228094575799721,
+      "grad_norm": 1.376093864440918,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.5686,
+      "step": 5530
+    },
+    {
+      "epoch": 1.9262865090403338,
+      "grad_norm": 1.4860330820083618,
+      "learning_rate": 6.871377253215905e-06,
+      "loss": 0.5694,
+      "step": 5540
+    },
+    {
+      "epoch": 1.9297635605006955,
+      "grad_norm": 1.6497060060501099,
+      "learning_rate": 6.832975661909426e-06,
+      "loss": 0.5779,
+      "step": 5550
+    },
+    {
+      "epoch": 1.933240611961057,
+      "grad_norm": 1.5908339023590088,
+      "learning_rate": 6.794625910918178e-06,
+      "loss": 0.5416,
+      "step": 5560
+    },
+    {
+      "epoch": 1.9367176634214185,
+      "grad_norm": 1.5128718614578247,
+      "learning_rate": 6.756328627980689e-06,
+      "loss": 0.5425,
+      "step": 5570
+    },
+    {
+      "epoch": 1.9401947148817804,
+      "grad_norm": 1.6888498067855835,
+      "learning_rate": 6.718084439976631e-06,
+      "loss": 0.5706,
+      "step": 5580
+    },
+    {
+      "epoch": 1.943671766342142,
+      "grad_norm": 1.4853570461273193,
+      "learning_rate": 6.679893972916598e-06,
+      "loss": 0.5535,
+      "step": 5590
+    },
+    {
+      "epoch": 1.9471488178025034,
+      "grad_norm": 1.400653600692749,
+      "learning_rate": 6.64175785193183e-06,
+      "loss": 0.5689,
+      "step": 5600
+    },
+    {
+      "epoch": 1.950625869262865,
+      "grad_norm": 1.3817487955093384,
+      "learning_rate": 6.603676701263985e-06,
+      "loss": 0.5585,
+      "step": 5610
+    },
+    {
+      "epoch": 1.9541029207232268,
+      "grad_norm": 1.4885752201080322,
+      "learning_rate": 6.565651144254932e-06,
+      "loss": 0.5579,
+      "step": 5620
+    },
+    {
+      "epoch": 1.9575799721835883,
+      "grad_norm": 1.4579683542251587,
+      "learning_rate": 6.527681803336531e-06,
+      "loss": 0.5701,
+      "step": 5630
+    },
+    {
+      "epoch": 1.96105702364395,
+      "grad_norm": 1.5224418640136719,
+      "learning_rate": 6.4897693000204585e-06,
+      "loss": 0.5627,
+      "step": 5640
+    },
+    {
+      "epoch": 1.9645340751043117,
+      "grad_norm": 1.5035606622695923,
+      "learning_rate": 6.451914254888028e-06,
+      "loss": 0.5455,
+      "step": 5650
+    },
+    {
+      "epoch": 1.9680111265646731,
+      "grad_norm": 1.4756858348846436,
+      "learning_rate": 6.414117287580033e-06,
+      "loss": 0.5806,
+      "step": 5660
+    },
+    {
+      "epoch": 1.9714881780250346,
+      "grad_norm": 1.5364742279052734,
+      "learning_rate": 6.3763790167866045e-06,
+      "loss": 0.5602,
+      "step": 5670
+    },
+    {
+      "epoch": 1.9749652294853965,
+      "grad_norm": 1.4767723083496094,
+      "learning_rate": 6.338700060237079e-06,
+      "loss": 0.5104,
+      "step": 5680
+    },
+    {
+      "epoch": 1.978442280945758,
+      "grad_norm": 1.4262028932571411,
+      "learning_rate": 6.3010810346898975e-06,
+      "loss": 0.5768,
+      "step": 5690
+    },
+    {
+      "epoch": 1.9819193324061195,
+      "grad_norm": 1.5635230541229248,
+      "learning_rate": 6.263522555922502e-06,
+      "loss": 0.5541,
+      "step": 5700
+    },
+    {
+      "epoch": 1.9853963838664812,
+      "grad_norm": 1.4389492273330688,
+      "learning_rate": 6.226025238721255e-06,
+      "loss": 0.5934,
+      "step": 5710
+    },
+    {
+      "epoch": 1.988873435326843,
+      "grad_norm": 1.4987590312957764,
+      "learning_rate": 6.188589696871385e-06,
+      "loss": 0.5883,
+      "step": 5720
+    },
+    {
+      "epoch": 1.9923504867872044,
+      "grad_norm": 1.4939262866973877,
+      "learning_rate": 6.151216543146923e-06,
+      "loss": 0.571,
+      "step": 5730
+    },
+    {
+      "epoch": 1.995827538247566,
+      "grad_norm": 1.3870357275009155,
+      "learning_rate": 6.113906389300692e-06,
+      "loss": 0.5762,
+      "step": 5740
+    },
+    {
+      "epoch": 1.9993045897079278,
+      "grad_norm": 1.399305820465088,
+      "learning_rate": 6.076659846054281e-06,
+      "loss": 0.5543,
+      "step": 5750
+    },
+    {
+      "epoch": 2.0027816411682893,
+      "grad_norm": 1.9261384010314941,
+      "learning_rate": 6.039477523088053e-06,
+      "loss": 0.3488,
+      "step": 5760
+    },
+    {
+      "epoch": 2.0062586926286508,
+      "grad_norm": 1.7587767839431763,
+      "learning_rate": 6.0023600290311616e-06,
+      "loss": 0.2944,
+      "step": 5770
+    },
+    {
+      "epoch": 2.0097357440890127,
+      "grad_norm": 1.7364875078201294,
+      "learning_rate": 5.9653079714515885e-06,
+      "loss": 0.2844,
+      "step": 5780
+    },
+    {
+      "epoch": 2.013212795549374,
+      "grad_norm": 1.4886943101882935,
+      "learning_rate": 5.928321956846204e-06,
+      "loss": 0.2884,
+      "step": 5790
+    },
+    {
+      "epoch": 2.0166898470097356,
+      "grad_norm": 1.7495077848434448,
+      "learning_rate": 5.891402590630834e-06,
+      "loss": 0.2739,
+      "step": 5800
+    },
+    {
+      "epoch": 2.0201668984700976,
+      "grad_norm": 1.7056771516799927,
+      "learning_rate": 5.854550477130352e-06,
+      "loss": 0.2922,
+      "step": 5810
+    },
+    {
+      "epoch": 2.023643949930459,
+      "grad_norm": 1.8620481491088867,
+      "learning_rate": 5.81776621956879e-06,
+      "loss": 0.2894,
+      "step": 5820
+    },
+    {
+      "epoch": 2.0271210013908205,
+      "grad_norm": 1.8230159282684326,
+      "learning_rate": 5.781050420059454e-06,
+      "loss": 0.2793,
+      "step": 5830
+    },
+    {
+      "epoch": 2.030598052851182,
+      "grad_norm": 1.991342544555664,
+      "learning_rate": 5.744403679595081e-06,
+      "loss": 0.2792,
+      "step": 5840
+    },
+    {
+      "epoch": 2.034075104311544,
+      "grad_norm": 1.5440292358398438,
+      "learning_rate": 5.7078265980379944e-06,
+      "loss": 0.2745,
+      "step": 5850
+    },
+    {
+      "epoch": 2.0375521557719054,
+      "grad_norm": 1.8502931594848633,
+      "learning_rate": 5.671319774110286e-06,
+      "loss": 0.2855,
+      "step": 5860
+    },
+    {
+      "epoch": 2.041029207232267,
+      "grad_norm": 1.6370331048965454,
+      "learning_rate": 5.6348838053840215e-06,
+      "loss": 0.2752,
+      "step": 5870
+    },
+    {
+      "epoch": 2.044506258692629,
+      "grad_norm": 1.665733814239502,
+      "learning_rate": 5.598519288271441e-06,
+      "loss": 0.2801,
+      "step": 5880
+    },
+    {
+      "epoch": 2.0479833101529903,
+      "grad_norm": 1.6209917068481445,
+      "learning_rate": 5.56222681801522e-06,
+      "loss": 0.2694,
+      "step": 5890
+    },
+    {
+      "epoch": 2.0514603616133518,
+      "grad_norm": 1.8120630979537964,
+      "learning_rate": 5.526006988678713e-06,
+      "loss": 0.2841,
+      "step": 5900
+    },
+    {
+      "epoch": 2.0549374130737137,
+      "grad_norm": 2.1265363693237305,
+      "learning_rate": 5.489860393136228e-06,
+      "loss": 0.2807,
+      "step": 5910
+    },
+    {
+      "epoch": 2.058414464534075,
+      "grad_norm": 1.7175661325454712,
+      "learning_rate": 5.453787623063333e-06,
+      "loss": 0.2818,
+      "step": 5920
+    },
+    {
+      "epoch": 2.0618915159944367,
+      "grad_norm": 1.6605859994888306,
+      "learning_rate": 5.417789268927153e-06,
+      "loss": 0.2723,
+      "step": 5930
+    },
+    {
+      "epoch": 2.065368567454798,
+      "grad_norm": 1.9015485048294067,
+      "learning_rate": 5.3818659199767194e-06,
+      "loss": 0.2733,
+      "step": 5940
+    },
+    {
+      "epoch": 2.06884561891516,
+      "grad_norm": 1.6967657804489136,
+      "learning_rate": 5.346018164233324e-06,
+      "loss": 0.2686,
+      "step": 5950
+    },
+    {
+      "epoch": 2.0723226703755215,
+      "grad_norm": 1.7520231008529663,
+      "learning_rate": 5.310246588480883e-06,
+      "loss": 0.2825,
+      "step": 5960
+    },
+    {
+      "epoch": 2.075799721835883,
+      "grad_norm": 1.5627514123916626,
+      "learning_rate": 5.274551778256356e-06,
+      "loss": 0.2703,
+      "step": 5970
+    },
+    {
+      "epoch": 2.079276773296245,
+      "grad_norm": 1.6424394845962524,
+      "learning_rate": 5.23893431784012e-06,
+      "loss": 0.2688,
+      "step": 5980
+    },
+    {
+      "epoch": 2.0827538247566064,
+      "grad_norm": 1.517057180404663,
+      "learning_rate": 5.203394790246447e-06,
+      "loss": 0.2722,
+      "step": 5990
+    },
+    {
+      "epoch": 2.086230876216968,
+      "grad_norm": 1.6601656675338745,
+      "learning_rate": 5.16793377721394e-06,
+      "loss": 0.2864,
+      "step": 6000
+    },
+    {
+      "epoch": 2.08970792767733,
+      "grad_norm": 1.6603269577026367,
+      "learning_rate": 5.132551859196014e-06,
+      "loss": 0.2646,
+      "step": 6010
+    },
+    {
+      "epoch": 2.0931849791376913,
+      "grad_norm": 1.5469019412994385,
+      "learning_rate": 5.0972496153514e-06,
+      "loss": 0.2782,
+      "step": 6020
+    },
+    {
+      "epoch": 2.096662030598053,
+      "grad_norm": 1.7444531917572021,
+      "learning_rate": 5.06202762353465e-06,
+      "loss": 0.2792,
+      "step": 6030
+    },
+    {
+      "epoch": 2.1001390820584143,
+      "grad_norm": 1.9016249179840088,
+      "learning_rate": 5.026886460286697e-06,
+      "loss": 0.271,
+      "step": 6040
+    },
+    {
+      "epoch": 2.103616133518776,
+      "grad_norm": 1.612169623374939,
+      "learning_rate": 4.991826700825404e-06,
+      "loss": 0.2648,
+      "step": 6050
+    },
+    {
+      "epoch": 2.1070931849791377,
+      "grad_norm": 1.7995601892471313,
+      "learning_rate": 4.956848919036161e-06,
+      "loss": 0.2803,
+      "step": 6060
+    },
+    {
+      "epoch": 2.110570236439499,
+      "grad_norm": 1.5551965236663818,
+      "learning_rate": 4.92195368746248e-06,
+      "loss": 0.2797,
+      "step": 6070
+    },
+    {
+      "epoch": 2.114047287899861,
+      "grad_norm": 1.6780221462249756,
+      "learning_rate": 4.887141577296618e-06,
+      "loss": 0.2725,
+      "step": 6080
+    },
+    {
+      "epoch": 2.1175243393602226,
+      "grad_norm": 1.670363426208496,
+      "learning_rate": 4.852413158370249e-06,
+      "loss": 0.2733,
+      "step": 6090
+    },
+    {
+      "epoch": 2.121001390820584,
+      "grad_norm": 1.6540260314941406,
+      "learning_rate": 4.81776899914512e-06,
+      "loss": 0.2739,
+      "step": 6100
+    },
+    {
+      "epoch": 2.124478442280946,
+      "grad_norm": 1.6426093578338623,
+      "learning_rate": 4.7832096667037475e-06,
+      "loss": 0.2679,
+      "step": 6110
+    },
+    {
+      "epoch": 2.1279554937413074,
+      "grad_norm": 1.8611637353897095,
+      "learning_rate": 4.7487357267401376e-06,
+      "loss": 0.2769,
+      "step": 6120
+    },
+    {
+      "epoch": 2.131432545201669,
+      "grad_norm": 1.8292388916015625,
+      "learning_rate": 4.714347743550528e-06,
+      "loss": 0.2779,
+      "step": 6130
+    },
+    {
+      "epoch": 2.1349095966620304,
+      "grad_norm": 1.693035364151001,
+      "learning_rate": 4.680046280024147e-06,
+      "loss": 0.2783,
+      "step": 6140
+    },
+    {
+      "epoch": 2.1383866481223923,
+      "grad_norm": 1.6081509590148926,
+      "learning_rate": 4.645831897634003e-06,
+      "loss": 0.2604,
+      "step": 6150
+    },
+    {
+      "epoch": 2.141863699582754,
+      "grad_norm": 1.6142793893814087,
+      "learning_rate": 4.611705156427694e-06,
+      "loss": 0.2858,
+      "step": 6160
+    },
+    {
+      "epoch": 2.1453407510431153,
+      "grad_norm": 1.7559276819229126,
+      "learning_rate": 4.577666615018239e-06,
+      "loss": 0.2666,
+      "step": 6170
+    },
+    {
+      "epoch": 2.148817802503477,
+      "grad_norm": 1.5642822980880737,
+      "learning_rate": 4.543716830574927e-06,
+      "loss": 0.2807,
+      "step": 6180
+    },
+    {
+      "epoch": 2.1522948539638387,
+      "grad_norm": 1.6646515130996704,
+      "learning_rate": 4.509856358814212e-06,
+      "loss": 0.2726,
+      "step": 6190
+    },
+    {
+      "epoch": 2.1557719054242,
+      "grad_norm": 1.635384202003479,
+      "learning_rate": 4.476085753990606e-06,
+      "loss": 0.2782,
+      "step": 6200
+    },
+    {
+      "epoch": 2.159248956884562,
+      "grad_norm": 1.6424869298934937,
+      "learning_rate": 4.4424055688876125e-06,
+      "loss": 0.2779,
+      "step": 6210
+    },
+    {
+      "epoch": 2.1627260083449236,
+      "grad_norm": 1.7635104656219482,
+      "learning_rate": 4.4088163548086715e-06,
+      "loss": 0.2773,
+      "step": 6220
+    },
+    {
+      "epoch": 2.166203059805285,
+      "grad_norm": 1.7563161849975586,
+      "learning_rate": 4.375318661568133e-06,
+      "loss": 0.2862,
+      "step": 6230
+    },
+    {
+      "epoch": 2.1696801112656465,
+      "grad_norm": 1.7935545444488525,
+      "learning_rate": 4.341913037482274e-06,
+      "loss": 0.2805,
+      "step": 6240
+    },
+    {
+      "epoch": 2.1731571627260085,
+      "grad_norm": 1.7619123458862305,
+      "learning_rate": 4.308600029360304e-06,
+      "loss": 0.2835,
+      "step": 6250
+    },
+    {
+      "epoch": 2.17663421418637,
+      "grad_norm": 1.8889714479446411,
+      "learning_rate": 4.275380182495425e-06,
+      "loss": 0.2731,
+      "step": 6260
+    },
+    {
+      "epoch": 2.1801112656467314,
+      "grad_norm": 1.720360279083252,
+      "learning_rate": 4.245562422647974e-06,
+      "loss": 0.2679,
+      "step": 6270
+    },
+    {
+      "epoch": 2.1835883171070933,
+      "grad_norm": 1.8219636678695679,
+      "learning_rate": 4.212521078985612e-06,
+      "loss": 0.2757,
+      "step": 6280
+    },
+    {
+      "epoch": 2.187065368567455,
+      "grad_norm": 1.731429100036621,
+      "learning_rate": 4.179574469275257e-06,
+      "loss": 0.2708,
+      "step": 6290
+    },
+    {
+      "epoch": 2.1905424200278163,
+      "grad_norm": 1.9347280263900757,
+      "learning_rate": 4.146723132812619e-06,
+      "loss": 0.2861,
+      "step": 6300
+    },
+    {
+      "epoch": 2.194019471488178,
+      "grad_norm": 1.7413724660873413,
+      "learning_rate": 4.113967607333901e-06,
+      "loss": 0.2661,
+      "step": 6310
+    },
+    {
+      "epoch": 2.1974965229485397,
+      "grad_norm": 1.8515937328338623,
+      "learning_rate": 4.081308429006989e-06,
+      "loss": 0.2624,
+      "step": 6320
+    },
+    {
+      "epoch": 2.200973574408901,
+      "grad_norm": 1.8258100748062134,
+      "learning_rate": 4.048746132422697e-06,
+      "loss": 0.2766,
+      "step": 6330
+    },
+    {
+      "epoch": 2.2044506258692627,
+      "grad_norm": 1.5192551612854004,
+      "learning_rate": 4.016281250585991e-06,
+      "loss": 0.277,
+      "step": 6340
+    },
+    {
+      "epoch": 2.2079276773296246,
+      "grad_norm": 1.842656135559082,
+      "learning_rate": 3.983914314907293e-06,
+      "loss": 0.2684,
+      "step": 6350
+    },
+    {
+      "epoch": 2.211404728789986,
+      "grad_norm": 1.8255943059921265,
+      "learning_rate": 3.951645855193753e-06,
+      "loss": 0.2857,
+      "step": 6360
+    },
+    {
+      "epoch": 2.2148817802503475,
+      "grad_norm": 1.6651691198349,
+      "learning_rate": 3.919476399640587e-06,
+      "loss": 0.2853,
+      "step": 6370
+    },
+    {
+      "epoch": 2.2183588317107095,
+      "grad_norm": 1.648590326309204,
+      "learning_rate": 3.887406474822445e-06,
+      "loss": 0.2598,
+      "step": 6380
+    },
+    {
+      "epoch": 2.221835883171071,
+      "grad_norm": 1.53499436378479,
+      "learning_rate": 3.855436605684774e-06,
+      "loss": 0.2585,
+      "step": 6390
+    },
+    {
+      "epoch": 2.2253129346314324,
+      "grad_norm": 1.9687730073928833,
+      "learning_rate": 3.823567315535234e-06,
+      "loss": 0.2816,
+      "step": 6400
+    },
+    {
+      "epoch": 2.2287899860917944,
+      "grad_norm": 1.5873713493347168,
+      "learning_rate": 3.794971380622724e-06,
+      "loss": 0.282,
+      "step": 6410
+    },
+    {
+      "epoch": 2.232267037552156,
+      "grad_norm": 1.8059178590774536,
+      "learning_rate": 3.7632946263599325e-06,
+      "loss": 0.2586,
+      "step": 6420
+    },
+    {
+      "epoch": 2.2357440890125173,
+      "grad_norm": 1.530378818511963,
+      "learning_rate": 3.731719959336787e-06,
+      "loss": 0.2583,
+      "step": 6430
+    },
+    {
+      "epoch": 2.239221140472879,
+      "grad_norm": 1.8726969957351685,
+      "learning_rate": 3.7002478963919707e-06,
+      "loss": 0.2691,
+      "step": 6440
+    },
+    {
+      "epoch": 2.2426981919332407,
+      "grad_norm": 1.8173457384109497,
+      "learning_rate": 3.668878952684669e-06,
+      "loss": 0.2616,
+      "step": 6450
+    },
+    {
+      "epoch": 2.246175243393602,
+      "grad_norm": 1.7228041887283325,
+      "learning_rate": 3.6376136416861195e-06,
+      "loss": 0.2753,
+      "step": 6460
+    },
+    {
+      "epoch": 2.2496522948539637,
+      "grad_norm": 1.720737338066101,
+      "learning_rate": 3.6064524751712304e-06,
+      "loss": 0.2546,
+      "step": 6470
+    },
+    {
+      "epoch": 2.2531293463143256,
+      "grad_norm": 1.5620167255401611,
+      "learning_rate": 3.5753959632101874e-06,
+      "loss": 0.275,
+      "step": 6480
+    },
+    {
+      "epoch": 2.256606397774687,
+      "grad_norm": 1.7242825031280518,
+      "learning_rate": 3.5444446141601106e-06,
+      "loss": 0.2617,
+      "step": 6490
+    },
+    {
+      "epoch": 2.2600834492350486,
+      "grad_norm": 1.7137529850006104,
+      "learning_rate": 3.5135989346567346e-06,
+      "loss": 0.2783,
+      "step": 6500
+    },
+    {
+      "epoch": 2.2635605006954105,
+      "grad_norm": 1.6205216646194458,
+      "learning_rate": 3.4828594296061036e-06,
+      "loss": 0.2812,
+      "step": 6510
+    },
+    {
+      "epoch": 2.267037552155772,
+      "grad_norm": 1.8813594579696655,
+      "learning_rate": 3.452226602176324e-06,
+      "loss": 0.2749,
+      "step": 6520
+    },
+    {
+      "epoch": 2.2705146036161334,
+      "grad_norm": 1.6485098600387573,
+      "learning_rate": 3.4217009537893176e-06,
+      "loss": 0.2573,
+      "step": 6530
+    },
+    {
+      "epoch": 2.273991655076495,
+      "grad_norm": 1.5779473781585693,
+      "learning_rate": 3.391282984112617e-06,
+      "loss": 0.2628,
+      "step": 6540
+    },
+    {
+      "epoch": 2.277468706536857,
+      "grad_norm": 1.552001953125,
+      "learning_rate": 3.360973191051189e-06,
+      "loss": 0.2702,
+      "step": 6550
+    },
+    {
+      "epoch": 2.2809457579972183,
+      "grad_norm": 1.8163200616836548,
+      "learning_rate": 3.3307720707392675e-06,
+      "loss": 0.2702,
+      "step": 6560
+    },
+    {
+      "epoch": 2.28442280945758,
+      "grad_norm": 1.683661699295044,
+      "learning_rate": 3.300680117532267e-06,
+      "loss": 0.2732,
+      "step": 6570
+    },
+    {
+      "epoch": 2.2878998609179417,
+      "grad_norm": 1.6786640882492065,
+      "learning_rate": 3.2706978239986566e-06,
+      "loss": 0.2735,
+      "step": 6580
+    },
+    {
+      "epoch": 2.291376912378303,
+      "grad_norm": 1.847489595413208,
+      "learning_rate": 3.2408256809119143e-06,
+      "loss": 0.2515,
+      "step": 6590
+    },
+    {
+      "epoch": 2.2948539638386647,
+      "grad_norm": 1.6280747652053833,
+      "learning_rate": 3.211064177242491e-06,
+      "loss": 0.2656,
+      "step": 6600
+    },
+    {
+      "epoch": 2.2983310152990266,
+      "grad_norm": 1.7639294862747192,
+      "learning_rate": 3.1814138001498006e-06,
+      "loss": 0.2647,
+      "step": 6610
+    },
+    {
+      "epoch": 2.301808066759388,
+      "grad_norm": 1.70976984500885,
+      "learning_rate": 3.1518750349742553e-06,
+      "loss": 0.2578,
+      "step": 6620
+    },
+    {
+      "epoch": 2.3052851182197496,
+      "grad_norm": 1.7705670595169067,
+      "learning_rate": 3.1224483652293168e-06,
+      "loss": 0.2651,
+      "step": 6630
+    },
+    {
+      "epoch": 2.3087621696801115,
+      "grad_norm": 1.721847414970398,
+      "learning_rate": 3.093134272593582e-06,
+      "loss": 0.2778,
+      "step": 6640
+    },
+    {
+      "epoch": 2.312239221140473,
+      "grad_norm": 1.6629747152328491,
+      "learning_rate": 3.0639332369028984e-06,
+      "loss": 0.276,
+      "step": 6650
+    },
+    {
+      "epoch": 2.3157162726008345,
+      "grad_norm": 1.885170817375183,
+      "learning_rate": 3.034845736142502e-06,
+      "loss": 0.2646,
+      "step": 6660
+    },
+    {
+      "epoch": 2.319193324061196,
+      "grad_norm": 1.9333254098892212,
+      "learning_rate": 3.005872246439212e-06,
+      "loss": 0.2701,
+      "step": 6670
+    },
+    {
+      "epoch": 2.322670375521558,
+      "grad_norm": 1.8997875452041626,
+      "learning_rate": 2.977013242053619e-06,
+      "loss": 0.2795,
+      "step": 6680
+    },
+    {
+      "epoch": 2.3261474269819193,
+      "grad_norm": 1.6752079725265503,
+      "learning_rate": 2.948269195372332e-06,
+      "loss": 0.2615,
+      "step": 6690
+    },
+    {
+      "epoch": 2.329624478442281,
+      "grad_norm": 1.7248494625091553,
+      "learning_rate": 2.9196405769002446e-06,
+      "loss": 0.2682,
+      "step": 6700
+    },
+    {
+      "epoch": 2.3331015299026427,
+      "grad_norm": 1.8830804824829102,
+      "learning_rate": 2.8911278552528233e-06,
+      "loss": 0.2754,
+      "step": 6710
+    },
+    {
+      "epoch": 2.3365785813630042,
+      "grad_norm": 1.808091402053833,
+      "learning_rate": 2.862731497148452e-06,
+      "loss": 0.2688,
+      "step": 6720
+    },
+    {
+      "epoch": 2.3400556328233657,
+      "grad_norm": 1.8167086839675903,
+      "learning_rate": 2.8344519674007885e-06,
+      "loss": 0.2643,
+      "step": 6730
+    },
+    {
+      "epoch": 2.343532684283727,
+      "grad_norm": 1.6254574060440063,
+      "learning_rate": 2.80628972891115e-06,
+      "loss": 0.2685,
+      "step": 6740
+    },
+    {
+      "epoch": 2.347009735744089,
+      "grad_norm": 1.7547099590301514,
+      "learning_rate": 2.7782452426609397e-06,
+      "loss": 0.2746,
+      "step": 6750
+    },
+    {
+      "epoch": 2.3504867872044506,
+      "grad_norm": 1.7741016149520874,
+      "learning_rate": 2.750318967704094e-06,
+      "loss": 0.2625,
+      "step": 6760
+    },
+    {
+      "epoch": 2.353963838664812,
+      "grad_norm": 1.7852706909179688,
+      "learning_rate": 2.7225113611595855e-06,
+      "loss": 0.2716,
+      "step": 6770
+    },
+    {
+      "epoch": 2.357440890125174,
+      "grad_norm": 1.6468247175216675,
+      "learning_rate": 2.6948228782039234e-06,
+      "loss": 0.2677,
+      "step": 6780
+    },
+    {
+      "epoch": 2.3609179415855355,
+      "grad_norm": 1.791222095489502,
+      "learning_rate": 2.6672539720637115e-06,
+      "loss": 0.2697,
+      "step": 6790
+    },
+    {
+      "epoch": 2.364394993045897,
+      "grad_norm": 1.761919379234314,
+      "learning_rate": 2.639805094008232e-06,
+      "loss": 0.256,
+      "step": 6800
+    },
+    {
+      "epoch": 2.367872044506259,
+      "grad_norm": 1.6554497480392456,
+      "learning_rate": 2.612476693342042e-06,
+      "loss": 0.2814,
+      "step": 6810
+    },
+    {
+      "epoch": 2.3713490959666204,
+      "grad_norm": 1.8679198026657104,
+      "learning_rate": 2.5852692173976414e-06,
+      "loss": 0.269,
+      "step": 6820
+    },
+    {
+      "epoch": 2.374826147426982,
+      "grad_norm": 1.650125503540039,
+      "learning_rate": 2.5581831115281388e-06,
+      "loss": 0.2622,
+      "step": 6830
+    },
+    {
+      "epoch": 2.3783031988873438,
+      "grad_norm": 1.5627479553222656,
+      "learning_rate": 2.53121881909996e-06,
+      "loss": 0.2512,
+      "step": 6840
+    },
+    {
+      "epoch": 2.3817802503477052,
+      "grad_norm": 1.6816457509994507,
+      "learning_rate": 2.5043767814856002e-06,
+      "loss": 0.2715,
+      "step": 6850
+    },
+    {
+      "epoch": 2.3852573018080667,
+      "grad_norm": 1.5710700750350952,
+      "learning_rate": 2.47765743805638e-06,
+      "loss": 0.2696,
+      "step": 6860
+    },
+    {
+      "epoch": 2.388734353268428,
+      "grad_norm": 1.7235229015350342,
+      "learning_rate": 2.45106122617528e-06,
+      "loss": 0.2664,
+      "step": 6870
+    },
+    {
+      "epoch": 2.39221140472879,
+      "grad_norm": 1.8657464981079102,
+      "learning_rate": 2.424588581189756e-06,
+      "loss": 0.2544,
+      "step": 6880
+    },
+    {
+      "epoch": 2.3956884561891516,
+      "grad_norm": 1.7270172834396362,
+      "learning_rate": 2.39823993642464e-06,
+      "loss": 0.2632,
+      "step": 6890
+    },
+    {
+      "epoch": 2.399165507649513,
+      "grad_norm": 1.8720391988754272,
+      "learning_rate": 2.372015723175022e-06,
+      "loss": 0.2593,
+      "step": 6900
+    },
+    {
+      "epoch": 2.402642559109875,
+      "grad_norm": 1.815830111503601,
+      "learning_rate": 2.345916370699195e-06,
+      "loss": 0.2669,
+      "step": 6910
+    },
+    {
+      "epoch": 2.4061196105702365,
+      "grad_norm": 1.762284517288208,
+      "learning_rate": 2.3199423062116443e-06,
+      "loss": 0.2465,
+      "step": 6920
+    },
+    {
+      "epoch": 2.409596662030598,
+      "grad_norm": 1.7445662021636963,
+      "learning_rate": 2.2940939548760375e-06,
+      "loss": 0.2619,
+      "step": 6930
+    },
+    {
+      "epoch": 2.4130737134909594,
+      "grad_norm": 1.857081651687622,
+      "learning_rate": 2.2683717397982776e-06,
+      "loss": 0.2706,
+      "step": 6940
+    },
+    {
+      "epoch": 2.4165507649513214,
+      "grad_norm": 1.6108611822128296,
+      "learning_rate": 2.242776082019569e-06,
+      "loss": 0.259,
+      "step": 6950
+    },
+    {
+      "epoch": 2.420027816411683,
+      "grad_norm": 1.7170990705490112,
+      "learning_rate": 2.2173074005095196e-06,
+      "loss": 0.2788,
+      "step": 6960
+    },
+    {
+      "epoch": 2.4235048678720443,
+      "grad_norm": 1.6751832962036133,
+      "learning_rate": 2.191966112159304e-06,
+      "loss": 0.2751,
+      "step": 6970
+    },
+    {
+      "epoch": 2.4269819193324063,
+      "grad_norm": 1.610290765762329,
+      "learning_rate": 2.166752631774818e-06,
+      "loss": 0.2717,
+      "step": 6980
+    },
+    {
+      "epoch": 2.4304589707927677,
+      "grad_norm": 1.8723939657211304,
+      "learning_rate": 2.1416673720699022e-06,
+      "loss": 0.2731,
+      "step": 6990
+    },
+    {
+      "epoch": 2.433936022253129,
+      "grad_norm": 1.8495627641677856,
+      "learning_rate": 2.1167107436595813e-06,
+      "loss": 0.2623,
+      "step": 7000
+    },
+    {
+      "epoch": 2.437413073713491,
+      "grad_norm": 1.7770112752914429,
+      "learning_rate": 2.0918831550533358e-06,
+      "loss": 0.2694,
+      "step": 7010
+    },
+    {
+      "epoch": 2.4408901251738526,
+      "grad_norm": 1.7057751417160034,
+      "learning_rate": 2.0671850126484293e-06,
+      "loss": 0.2643,
+      "step": 7020
+    },
+    {
+      "epoch": 2.444367176634214,
+      "grad_norm": 1.8776519298553467,
+      "learning_rate": 2.0426167207232495e-06,
+      "loss": 0.2831,
+      "step": 7030
+    },
+    {
+      "epoch": 2.447844228094576,
+      "grad_norm": 1.6576194763183594,
+      "learning_rate": 2.018178681430686e-06,
+      "loss": 0.2708,
+      "step": 7040
+    },
+    {
+      "epoch": 2.4513212795549375,
+      "grad_norm": 1.9045627117156982,
+      "learning_rate": 1.9938712947915615e-06,
+      "loss": 0.2745,
+      "step": 7050
+    },
+    {
+      "epoch": 2.454798331015299,
+      "grad_norm": 1.8594796657562256,
+      "learning_rate": 1.969694958688061e-06,
+      "loss": 0.2812,
+      "step": 7060
+    },
+    {
+      "epoch": 2.4582753824756605,
+      "grad_norm": 1.8113681077957153,
+      "learning_rate": 1.9456500688572422e-06,
+      "loss": 0.2726,
+      "step": 7070
+    },
+    {
+      "epoch": 2.4617524339360224,
+      "grad_norm": 1.7855045795440674,
+      "learning_rate": 1.9217370188845463e-06,
+      "loss": 0.278,
+      "step": 7080
+    },
+    {
+      "epoch": 2.465229485396384,
+      "grad_norm": 1.7415590286254883,
+      "learning_rate": 1.8979562001973572e-06,
+      "loss": 0.2696,
+      "step": 7090
+    },
+    {
+      "epoch": 2.4687065368567453,
+      "grad_norm": 1.7221544981002808,
+      "learning_rate": 1.874308002058589e-06,
+      "loss": 0.2671,
+      "step": 7100
+    },
+    {
+      "epoch": 2.4721835883171073,
+      "grad_norm": 1.672399878501892,
+      "learning_rate": 1.850792811560327e-06,
+      "loss": 0.2613,
+      "step": 7110
+    },
+    {
+      "epoch": 2.4756606397774688,
+      "grad_norm": 1.63778555393219,
+      "learning_rate": 1.8274110136174806e-06,
+      "loss": 0.2422,
+      "step": 7120
+    },
+    {
+      "epoch": 2.4791376912378302,
+      "grad_norm": 1.843570351600647,
+      "learning_rate": 1.8041629909614877e-06,
+      "loss": 0.2649,
+      "step": 7130
+    },
+    {
+      "epoch": 2.4826147426981917,
+      "grad_norm": 1.673641324043274,
+      "learning_rate": 1.78104912413405e-06,
+      "loss": 0.2668,
+      "step": 7140
+    },
+    {
+      "epoch": 2.4860917941585536,
+      "grad_norm": 1.7001925706863403,
+      "learning_rate": 1.7580697914808932e-06,
+      "loss": 0.2646,
+      "step": 7150
+    },
+    {
+      "epoch": 2.489568845618915,
+      "grad_norm": 1.6739141941070557,
+      "learning_rate": 1.7352253691455945e-06,
+      "loss": 0.276,
+      "step": 7160
+    },
+    {
+      "epoch": 2.4930458970792766,
+      "grad_norm": 1.7010778188705444,
+      "learning_rate": 1.712516231063408e-06,
+      "loss": 0.256,
+      "step": 7170
+    },
+    {
+      "epoch": 2.4965229485396385,
+      "grad_norm": 1.7061653137207031,
+      "learning_rate": 1.6899427489551512e-06,
+      "loss": 0.2612,
+      "step": 7180
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.7902648448944092,
+      "learning_rate": 1.667505292321121e-06,
+      "loss": 0.2621,
+      "step": 7190
+    },
+    {
+      "epoch": 2.5034770514603615,
+      "grad_norm": 1.5667250156402588,
+      "learning_rate": 1.64520422843504e-06,
+      "loss": 0.2607,
+      "step": 7200
+    },
+    {
+      "epoch": 2.506954102920723,
+      "grad_norm": 1.8519011735916138,
+      "learning_rate": 1.6252501884897975e-06,
+      "loss": 0.2808,
+      "step": 7210
+    },
+    {
+      "epoch": 2.510431154381085,
+      "grad_norm": 1.7838876247406006,
+      "learning_rate": 1.6032092746631244e-06,
+      "loss": 0.2657,
+      "step": 7220
+    },
+    {
+      "epoch": 2.5139082058414464,
+      "grad_norm": 1.8263791799545288,
+      "learning_rate": 1.5813058060315777e-06,
+      "loss": 0.2692,
+      "step": 7230
+    },
+    {
+      "epoch": 2.5173852573018083,
+      "grad_norm": 1.6479767560958862,
+      "learning_rate": 1.5595401411281564e-06,
+      "loss": 0.2715,
+      "step": 7240
+    },
+    {
+      "epoch": 2.5208623087621698,
+      "grad_norm": 1.6672980785369873,
+      "learning_rate": 1.5379126362301677e-06,
+      "loss": 0.2702,
+      "step": 7250
+    },
+    {
+      "epoch": 2.5243393602225312,
+      "grad_norm": 1.5929231643676758,
+      "learning_rate": 1.5164236453534197e-06,
+      "loss": 0.2478,
+      "step": 7260
+    },
+    {
+      "epoch": 2.5278164116828927,
+      "grad_norm": 1.7242038249969482,
+      "learning_rate": 1.4950735202464173e-06,
+      "loss": 0.2679,
+      "step": 7270
+    },
+    {
+      "epoch": 2.5312934631432547,
+      "grad_norm": 1.8464475870132446,
+      "learning_rate": 1.4738626103845933e-06,
+      "loss": 0.2635,
+      "step": 7280
+    },
+    {
+      "epoch": 2.534770514603616,
+      "grad_norm": 1.6160593032836914,
+      "learning_rate": 1.4527912629646068e-06,
+      "loss": 0.2569,
+      "step": 7290
+    },
+    {
+      "epoch": 2.5382475660639776,
+      "grad_norm": 2.0354888439178467,
+      "learning_rate": 1.431859822898638e-06,
+      "loss": 0.2589,
+      "step": 7300
+    },
+    {
+      "epoch": 2.5417246175243395,
+      "grad_norm": 1.7076482772827148,
+      "learning_rate": 1.4110686328087674e-06,
+      "loss": 0.2647,
+      "step": 7310
+    },
+    {
+      "epoch": 2.545201668984701,
+      "grad_norm": 1.6705870628356934,
+      "learning_rate": 1.390418033021348e-06,
+      "loss": 0.2683,
+      "step": 7320
+    },
+    {
+      "epoch": 2.5486787204450625,
+      "grad_norm": 1.7679471969604492,
+      "learning_rate": 1.369908361561445e-06,
+      "loss": 0.264,
+      "step": 7330
+    },
+    {
+      "epoch": 2.552155771905424,
+      "grad_norm": 1.739141821861267,
+      "learning_rate": 1.3495399541472942e-06,
+      "loss": 0.2725,
+      "step": 7340
+    },
+    {
+      "epoch": 2.555632823365786,
+      "grad_norm": 1.7540068626403809,
+      "learning_rate": 1.3293131441848162e-06,
+      "loss": 0.2612,
+      "step": 7350
+    },
+    {
+      "epoch": 2.5591098748261474,
+      "grad_norm": 1.758461833000183,
+      "learning_rate": 1.3092282627621544e-06,
+      "loss": 0.2519,
+      "step": 7360
+    },
+    {
+      "epoch": 2.5625869262865093,
+      "grad_norm": 1.7611478567123413,
+      "learning_rate": 1.2892856386442531e-06,
+      "loss": 0.2528,
+      "step": 7370
+    },
+    {
+      "epoch": 2.566063977746871,
+      "grad_norm": 1.7101202011108398,
+      "learning_rate": 1.2694855982674836e-06,
+      "loss": 0.2814,
+      "step": 7380
+    },
+    {
+      "epoch": 2.5695410292072323,
+      "grad_norm": 1.5715007781982422,
+      "learning_rate": 1.2498284657342862e-06,
+      "loss": 0.2664,
+      "step": 7390
+    },
+    {
+      "epoch": 2.5730180806675937,
+      "grad_norm": 1.6963520050048828,
+      "learning_rate": 1.2303145628078826e-06,
+      "loss": 0.2686,
+      "step": 7400
+    },
+    {
+      "epoch": 2.5764951321279552,
+      "grad_norm": 1.5881513357162476,
+      "learning_rate": 1.2109442089069988e-06,
+      "loss": 0.2505,
+      "step": 7410
+    },
+    {
+      "epoch": 2.579972183588317,
+      "grad_norm": 1.8540960550308228,
+      "learning_rate": 1.191717721100638e-06,
+      "loss": 0.2749,
+      "step": 7420
+    },
+    {
+      "epoch": 2.5834492350486786,
+      "grad_norm": 1.8322099447250366,
+      "learning_rate": 1.1726354141028962e-06,
+      "loss": 0.2645,
+      "step": 7430
+    },
+    {
+      "epoch": 2.5869262865090406,
+      "grad_norm": 1.751469373703003,
+      "learning_rate": 1.153697600267797e-06,
+      "loss": 0.2698,
+      "step": 7440
+    },
+    {
+      "epoch": 2.590403337969402,
+      "grad_norm": 1.7991695404052734,
+      "learning_rate": 1.1349045895841948e-06,
+      "loss": 0.2653,
+      "step": 7450
+    },
+    {
+      "epoch": 2.5938803894297635,
+      "grad_norm": 1.739248514175415,
+      "learning_rate": 1.1162566896706906e-06,
+      "loss": 0.2699,
+      "step": 7460
+    },
+    {
+      "epoch": 2.597357440890125,
+      "grad_norm": 1.805435299873352,
+      "learning_rate": 1.0977542057705991e-06,
+      "loss": 0.2644,
+      "step": 7470
+    },
+    {
+      "epoch": 2.600834492350487,
+      "grad_norm": 1.6561611890792847,
+      "learning_rate": 1.0793974407469577e-06,
+      "loss": 0.2451,
+      "step": 7480
+    },
+    {
+      "epoch": 2.6043115438108484,
+      "grad_norm": 1.705469012260437,
+      "learning_rate": 1.061186695077554e-06,
+      "loss": 0.2489,
+      "step": 7490
+    },
+    {
+      "epoch": 2.60778859527121,
+      "grad_norm": 1.723598837852478,
+      "learning_rate": 1.0431222668500273e-06,
+      "loss": 0.2688,
+      "step": 7500
+    },
+    {
+      "epoch": 2.611265646731572,
+      "grad_norm": 1.7613165378570557,
+      "learning_rate": 1.0252044517569738e-06,
+      "loss": 0.2685,
+      "step": 7510
+    },
+    {
+      "epoch": 2.6147426981919333,
+      "grad_norm": 1.687288522720337,
+      "learning_rate": 1.0074335430911142e-06,
+      "loss": 0.2666,
+      "step": 7520
+    },
+    {
+      "epoch": 2.6182197496522948,
+      "grad_norm": 1.7405155897140503,
+      "learning_rate": 9.898098317404925e-07,
+      "loss": 0.2553,
+      "step": 7530
+    },
+    {
+      "epoch": 2.6216968011126562,
+      "grad_norm": 1.837397813796997,
+      "learning_rate": 9.723336061837053e-07,
+      "loss": 0.2837,
+      "step": 7540
+    },
+    {
+      "epoch": 2.625173852573018,
+      "grad_norm": 1.6355146169662476,
+      "learning_rate": 9.550051524851945e-07,
+      "loss": 0.2697,
+      "step": 7550
+    },
+    {
+      "epoch": 2.6286509040333796,
+      "grad_norm": 1.64885413646698,
+      "learning_rate": 9.37824754290555e-07,
+      "loss": 0.2627,
+      "step": 7560
+    },
+    {
+      "epoch": 2.6321279554937416,
+      "grad_norm": 1.6726666688919067,
+      "learning_rate": 9.207926928218935e-07,
+      "loss": 0.2601,
+      "step": 7570
+    },
+    {
+      "epoch": 2.635605006954103,
+      "grad_norm": 1.7850905656814575,
+      "learning_rate": 9.03909246873228e-07,
+      "loss": 0.2751,
+      "step": 7580
+    },
+    {
+      "epoch": 2.6390820584144645,
+      "grad_norm": 1.5521477460861206,
+      "learning_rate": 8.871746928059189e-07,
+      "loss": 0.2525,
+      "step": 7590
+    },
+    {
+      "epoch": 2.642559109874826,
+      "grad_norm": 1.7885189056396484,
+      "learning_rate": 8.705893045441505e-07,
+      "loss": 0.2626,
+      "step": 7600
+    },
+    {
+      "epoch": 2.6460361613351875,
+      "grad_norm": 1.630260944366455,
+      "learning_rate": 8.541533535704461e-07,
+      "loss": 0.2594,
+      "step": 7610
+    },
+    {
+      "epoch": 2.6495132127955494,
+      "grad_norm": 1.6865968704223633,
+      "learning_rate": 8.378671089212231e-07,
+      "loss": 0.2547,
+      "step": 7620
+    },
+    {
+      "epoch": 2.652990264255911,
+      "grad_norm": 1.8030978441238403,
+      "learning_rate": 8.217308371823885e-07,
+      "loss": 0.2645,
+      "step": 7630
+    },
+    {
+      "epoch": 2.656467315716273,
+      "grad_norm": 1.7879563570022583,
+      "learning_rate": 8.057448024849734e-07,
+      "loss": 0.2722,
+      "step": 7640
+    },
+    {
+      "epoch": 2.6599443671766343,
+      "grad_norm": 1.7715020179748535,
+      "learning_rate": 7.89909266500819e-07,
+      "loss": 0.266,
+      "step": 7650
+    },
+    {
+      "epoch": 2.6634214186369958,
+      "grad_norm": 1.8094457387924194,
+      "learning_rate": 7.74224488438281e-07,
+      "loss": 0.262,
+      "step": 7660
+    },
+    {
+      "epoch": 2.6668984700973573,
+      "grad_norm": 1.8029688596725464,
+      "learning_rate": 7.586907250379949e-07,
+      "loss": 0.28,
+      "step": 7670
+    },
+    {
+      "epoch": 2.670375521557719,
+      "grad_norm": 1.7595010995864868,
+      "learning_rate": 7.433082305686712e-07,
+      "loss": 0.2605,
+      "step": 7680
+    },
+    {
+      "epoch": 2.6738525730180807,
+      "grad_norm": 1.8340460062026978,
+      "learning_rate": 7.280772568229289e-07,
+      "loss": 0.2579,
+      "step": 7690
+    },
+    {
+      "epoch": 2.677329624478442,
+      "grad_norm": 1.924147367477417,
+      "learning_rate": 7.129980531131842e-07,
+      "loss": 0.262,
+      "step": 7700
+    },
+    {
+      "epoch": 2.680806675938804,
+      "grad_norm": 1.8719463348388672,
+      "learning_rate": 6.980708662675584e-07,
+      "loss": 0.271,
+      "step": 7710
+    },
+    {
+      "epoch": 2.6842837273991655,
+      "grad_norm": 1.8740649223327637,
+      "learning_rate": 6.832959406258466e-07,
+      "loss": 0.2645,
+      "step": 7720
+    },
+    {
+      "epoch": 2.687760778859527,
+      "grad_norm": 1.8723556995391846,
+      "learning_rate": 6.686735180355098e-07,
+      "loss": 0.2703,
+      "step": 7730
+    },
+    {
+      "epoch": 2.6912378303198885,
+      "grad_norm": 1.7249317169189453,
+      "learning_rate": 6.542038378477244e-07,
+      "loss": 0.2676,
+      "step": 7740
+    },
+    {
+      "epoch": 2.6947148817802504,
+      "grad_norm": 1.633646845817566,
+      "learning_rate": 6.398871369134596e-07,
+      "loss": 0.2531,
+      "step": 7750
+    },
+    {
+      "epoch": 2.698191933240612,
+      "grad_norm": 1.798791527748108,
+      "learning_rate": 6.257236495796004e-07,
+      "loss": 0.2612,
+      "step": 7760
+    },
+    {
+      "epoch": 2.701668984700974,
+      "grad_norm": 1.602985143661499,
+      "learning_rate": 6.117136076851116e-07,
+      "loss": 0.2468,
+      "step": 7770
+    },
+    {
+      "epoch": 2.7051460361613353,
+      "grad_norm": 1.6389963626861572,
+      "learning_rate": 5.978572405572481e-07,
+      "loss": 0.2572,
+      "step": 7780
+    },
+    {
+      "epoch": 2.708623087621697,
+      "grad_norm": 1.6896214485168457,
+      "learning_rate": 5.841547750077903e-07,
+      "loss": 0.2704,
+      "step": 7790
+    },
+    {
+      "epoch": 2.7121001390820583,
+      "grad_norm": 1.7233210802078247,
+      "learning_rate": 5.706064353293428e-07,
+      "loss": 0.2556,
+      "step": 7800
+    },
+    {
+      "epoch": 2.71557719054242,
+      "grad_norm": 1.6733537912368774,
+      "learning_rate": 5.572124432916582e-07,
+      "loss": 0.2524,
+      "step": 7810
+    },
+    {
+      "epoch": 2.7190542420027817,
+      "grad_norm": 2.0769331455230713,
+      "learning_rate": 5.439730181380032e-07,
+      "loss": 0.2616,
+      "step": 7820
+    },
+    {
+      "epoch": 2.722531293463143,
+      "grad_norm": 1.5591294765472412,
+      "learning_rate": 5.308883765815798e-07,
+      "loss": 0.2564,
+      "step": 7830
+    },
+    {
+      "epoch": 2.726008344923505,
+      "grad_norm": 1.7184797525405884,
+      "learning_rate": 5.179587328019642e-07,
+      "loss": 0.2587,
+      "step": 7840
+    },
+    {
+      "epoch": 2.7294853963838666,
+      "grad_norm": 1.6269073486328125,
+      "learning_rate": 5.051842984416133e-07,
+      "loss": 0.26,
+      "step": 7850
+    },
+    {
+      "epoch": 2.732962447844228,
+      "grad_norm": 1.6789261102676392,
+      "learning_rate": 4.925652826023963e-07,
+      "loss": 0.2642,
+      "step": 7860
+    },
+    {
+      "epoch": 2.7364394993045895,
+      "grad_norm": 1.710343837738037,
+      "learning_rate": 4.801018918421685e-07,
+      "loss": 0.2613,
+      "step": 7870
+    },
+    {
+      "epoch": 2.7399165507649514,
+      "grad_norm": 1.7332804203033447,
+      "learning_rate": 4.677943301713961e-07,
+      "loss": 0.2527,
+      "step": 7880
+    },
+    {
+      "epoch": 2.743393602225313,
+      "grad_norm": 1.7913154363632202,
+      "learning_rate": 4.5564279904980915e-07,
+      "loss": 0.268,
+      "step": 7890
+    },
+    {
+      "epoch": 2.7468706536856744,
+      "grad_norm": 1.6718541383743286,
+      "learning_rate": 4.4364749738311285e-07,
+      "loss": 0.2546,
+      "step": 7900
+    },
+    {
+      "epoch": 2.7503477051460363,
+      "grad_norm": 1.8096388578414917,
+      "learning_rate": 4.318086215197259e-07,
+      "loss": 0.2556,
+      "step": 7910
+    },
+    {
+      "epoch": 2.753824756606398,
+      "grad_norm": 1.7550987005233765,
+      "learning_rate": 4.201263652475662e-07,
+      "loss": 0.2655,
+      "step": 7920
+    },
+    {
+      "epoch": 2.7573018080667593,
+      "grad_norm": 1.7428154945373535,
+      "learning_rate": 4.0860091979088355e-07,
+      "loss": 0.2408,
+      "step": 7930
+    },
+    {
+      "epoch": 2.7607788595271208,
+      "grad_norm": 1.8265447616577148,
+      "learning_rate": 3.972324738071232e-07,
+      "loss": 0.2521,
+      "step": 7940
+    },
+    {
+      "epoch": 2.7642559109874827,
+      "grad_norm": 1.4969125986099243,
+      "learning_rate": 3.860212133838426e-07,
+      "loss": 0.2489,
+      "step": 7950
+    },
+    {
+      "epoch": 2.767732962447844,
+      "grad_norm": 1.7376409769058228,
+      "learning_rate": 3.7496732203566533e-07,
+      "loss": 0.28,
+      "step": 7960
+    },
+    {
+      "epoch": 2.771210013908206,
+      "grad_norm": 1.6566534042358398,
+      "learning_rate": 3.640709807012721e-07,
+      "loss": 0.2446,
+      "step": 7970
+    },
+    {
+      "epoch": 2.7746870653685676,
+      "grad_norm": 1.6725773811340332,
+      "learning_rate": 3.533323677404477e-07,
+      "loss": 0.2558,
+      "step": 7980
+    },
+    {
+      "epoch": 2.778164116828929,
+      "grad_norm": 1.7670518159866333,
+      "learning_rate": 3.427516589311486e-07,
+      "loss": 0.2554,
+      "step": 7990
+    },
+    {
+      "epoch": 2.7816411682892905,
+      "grad_norm": 1.677904725074768,
+      "learning_rate": 3.323290274666424e-07,
+      "loss": 0.278,
+      "step": 8000
+    },
+    {
+      "epoch": 2.7851182197496525,
+      "grad_norm": 1.8242181539535522,
+      "learning_rate": 3.2206464395265755e-07,
+      "loss": 0.2631,
+      "step": 8010
+    },
+    {
+      "epoch": 2.788595271210014,
+      "grad_norm": 1.7958626747131348,
+      "learning_rate": 3.1195867640459896e-07,
+      "loss": 0.2599,
+      "step": 8020
+    },
+    {
+      "epoch": 2.7920723226703754,
+      "grad_norm": 1.8091436624526978,
+      "learning_rate": 3.020112902447969e-07,
+      "loss": 0.2604,
+      "step": 8030
+    },
+    {
+      "epoch": 2.7955493741307373,
+      "grad_norm": 1.8359060287475586,
+      "learning_rate": 2.9222264829979696e-07,
+      "loss": 0.2708,
+      "step": 8040
+    },
+    {
+      "epoch": 2.799026425591099,
+      "grad_norm": 1.8205952644348145,
+      "learning_rate": 2.825929107976977e-07,
+      "loss": 0.2546,
+      "step": 8050
+    },
+    {
+      "epoch": 2.8025034770514603,
+      "grad_norm": 1.6006027460098267,
+      "learning_rate": 2.7312223536552273e-07,
+      "loss": 0.2636,
+      "step": 8060
+    },
+    {
+      "epoch": 2.805980528511822,
+      "grad_norm": 1.6431981325149536,
+      "learning_rate": 2.638107770266485e-07,
+      "loss": 0.2623,
+      "step": 8070
+    },
+    {
+      "epoch": 2.8094575799721837,
+      "grad_norm": 1.6546140909194946,
+      "learning_rate": 2.5465868819826043e-07,
+      "loss": 0.2668,
+      "step": 8080
+    },
+    {
+      "epoch": 2.812934631432545,
+      "grad_norm": 1.746649146080017,
+      "learning_rate": 2.456661186888587e-07,
+      "loss": 0.2463,
+      "step": 8090
+    },
+    {
+      "epoch": 2.8164116828929067,
+      "grad_norm": 1.7403568029403687,
+      "learning_rate": 2.3683321569580863e-07,
+      "loss": 0.2619,
+      "step": 8100
+    },
+    {
+      "epoch": 2.8198887343532686,
+      "grad_norm": 1.7300713062286377,
+      "learning_rate": 2.2816012380293074e-07,
+      "loss": 0.2631,
+      "step": 8110
+    },
+    {
+      "epoch": 2.82336578581363,
+      "grad_norm": 1.8499008417129517,
+      "learning_rate": 2.196469849781313e-07,
+      "loss": 0.2538,
+      "step": 8120
+    },
+    {
+      "epoch": 2.8268428372739915,
+      "grad_norm": 1.6324739456176758,
+      "learning_rate": 2.1129393857108327e-07,
+      "loss": 0.247,
+      "step": 8130
+    },
+    {
+      "epoch": 2.830319888734353,
+      "grad_norm": 1.7417833805084229,
+      "learning_rate": 2.0310112131093907e-07,
+      "loss": 0.2722,
+      "step": 8140
+    },
+    {
+      "epoch": 2.833796940194715,
+      "grad_norm": 1.7168976068496704,
+      "learning_rate": 1.95068667304098e-07,
+      "loss": 0.2574,
+      "step": 8150
+    },
+    {
+      "epoch": 2.8372739916550764,
+      "grad_norm": 1.6456905603408813,
+      "learning_rate": 1.8719670803200806e-07,
+      "loss": 0.2504,
+      "step": 8160
+    },
+    {
+      "epoch": 2.8407510431154384,
+      "grad_norm": 1.6526283025741577,
+      "learning_rate": 1.7948537234901643e-07,
+      "loss": 0.2432,
+      "step": 8170
+    },
+    {
+      "epoch": 2.8442280945758,
+      "grad_norm": 1.742457389831543,
+      "learning_rate": 1.7193478648025786e-07,
+      "loss": 0.2539,
+      "step": 8180
+    },
+    {
+      "epoch": 2.8477051460361613,
+      "grad_norm": 1.6323515176773071,
+      "learning_rate": 1.6454507401958753e-07,
+      "loss": 0.255,
+      "step": 8190
+    },
+    {
+      "epoch": 2.851182197496523,
+      "grad_norm": 1.7443368434906006,
+      "learning_rate": 1.5731635592756145e-07,
+      "loss": 0.2569,
+      "step": 8200
+    },
+    {
+      "epoch": 2.8546592489568847,
+      "grad_norm": 1.8232289552688599,
+      "learning_rate": 1.5024875052945476e-07,
+      "loss": 0.2549,
+      "step": 8210
+    },
+    {
+      "epoch": 2.858136300417246,
+      "grad_norm": 1.6713287830352783,
+      "learning_rate": 1.4334237351332435e-07,
+      "loss": 0.2639,
+      "step": 8220
+    },
+    {
+      "epoch": 2.8616133518776077,
+      "grad_norm": 1.8971575498580933,
+      "learning_rate": 1.3659733792811602e-07,
+      "loss": 0.255,
+      "step": 8230
+    },
+    {
+      "epoch": 2.8650904033379696,
+      "grad_norm": 1.7788777351379395,
+      "learning_rate": 1.3001375418181138e-07,
+      "loss": 0.2652,
+      "step": 8240
+    },
+    {
+      "epoch": 2.868567454798331,
+      "grad_norm": 1.7716312408447266,
+      "learning_rate": 1.235917300396272e-07,
+      "loss": 0.2544,
+      "step": 8250
+    },
+    {
+      "epoch": 2.8720445062586926,
+      "grad_norm": 1.6315852403640747,
+      "learning_rate": 1.1733137062224231e-07,
+      "loss": 0.2572,
+      "step": 8260
+    },
+    {
+      "epoch": 2.875521557719054,
+      "grad_norm": 1.6397048234939575,
+      "learning_rate": 1.1123277840408675e-07,
+      "loss": 0.2505,
+      "step": 8270
+    },
+    {
+      "epoch": 2.878998609179416,
+      "grad_norm": 1.7039597034454346,
+      "learning_rate": 1.0529605321165426e-07,
+      "loss": 0.2493,
+      "step": 8280
+    },
+    {
+      "epoch": 2.8824756606397774,
+      "grad_norm": 1.6669057607650757,
+      "learning_rate": 9.952129222187801e-08,
+      "loss": 0.2612,
+      "step": 8290
+    },
+    {
+      "epoch": 2.885952712100139,
+      "grad_norm": 1.7976739406585693,
+      "learning_rate": 9.390858996053076e-08,
+      "loss": 0.2668,
+      "step": 8300
+    },
+    {
+      "epoch": 2.889429763560501,
+      "grad_norm": 1.5966641902923584,
+      "learning_rate": 8.845803830068611e-08,
+      "loss": 0.2572,
+      "step": 8310
+    },
+    {
+      "epoch": 2.8929068150208623,
+      "grad_norm": 1.6744129657745361,
+      "learning_rate": 8.316972646120746e-08,
+      "loss": 0.2587,
+      "step": 8320
+    },
+    {
+      "epoch": 2.896383866481224,
+      "grad_norm": 1.534909963607788,
+      "learning_rate": 7.804374100529366e-08,
+      "loss": 0.2541,
+      "step": 8330
+    },
+    {
+      "epoch": 2.8998609179415853,
+      "grad_norm": 1.643113613128662,
+      "learning_rate": 7.308016583905675e-08,
+      "loss": 0.2517,
+      "step": 8340
+    },
+    {
+      "epoch": 2.903337969401947,
+      "grad_norm": 1.5913118124008179,
+      "learning_rate": 6.827908221015311e-08,
+      "loss": 0.2507,
+      "step": 8350
+    },
+    {
+      "epoch": 2.9068150208623087,
+      "grad_norm": 1.733341097831726,
+      "learning_rate": 6.364056870644786e-08,
+      "loss": 0.2411,
+      "step": 8360
+    },
+    {
+      "epoch": 2.9102920723226706,
+      "grad_norm": 1.7070167064666748,
+      "learning_rate": 5.9164701254735836e-08,
+      "loss": 0.2561,
+      "step": 8370
+    },
+    {
+      "epoch": 2.913769123783032,
+      "grad_norm": 1.7591173648834229,
+      "learning_rate": 5.485155311949375e-08,
+      "loss": 0.2533,
+      "step": 8380
+    },
+    {
+      "epoch": 2.9172461752433936,
+      "grad_norm": 1.8583091497421265,
+      "learning_rate": 5.0701194901681125e-08,
+      "loss": 0.276,
+      "step": 8390
+    },
+    {
+      "epoch": 2.920723226703755,
+      "grad_norm": 1.7534596920013428,
+      "learning_rate": 4.671369453758567e-08,
+      "loss": 0.2495,
+      "step": 8400
+    },
+    {
+      "epoch": 2.924200278164117,
+      "grad_norm": 1.792546033859253,
+      "learning_rate": 4.2889117297711946e-08,
+      "loss": 0.2659,
+      "step": 8410
+    },
+    {
+      "epoch": 2.9276773296244785,
+      "grad_norm": 1.4881359338760376,
+      "learning_rate": 3.9227525785712205e-08,
+      "loss": 0.2502,
+      "step": 8420
+    },
+    {
+      "epoch": 2.93115438108484,
+      "grad_norm": 1.900132179260254,
+      "learning_rate": 3.5728979937362796e-08,
+      "loss": 0.2676,
+      "step": 8430
+    },
+    {
+      "epoch": 2.934631432545202,
+      "grad_norm": 1.746113657951355,
+      "learning_rate": 3.239353701957826e-08,
+      "loss": 0.2557,
+      "step": 8440
+    },
+    {
+      "epoch": 2.9381084840055633,
+      "grad_norm": 1.8157410621643066,
+      "learning_rate": 2.9221251629480973e-08,
+      "loss": 0.2567,
+      "step": 8450
+    },
+    {
+      "epoch": 2.941585535465925,
+      "grad_norm": 1.7264857292175293,
+      "learning_rate": 2.6212175693504093e-08,
+      "loss": 0.2624,
+      "step": 8460
+    },
+    {
+      "epoch": 2.9450625869262863,
+      "grad_norm": 1.6716339588165283,
+      "learning_rate": 2.3366358466538896e-08,
+      "loss": 0.253,
+      "step": 8470
+    },
+    {
+      "epoch": 2.9485396383866482,
+      "grad_norm": 1.7572870254516602,
+      "learning_rate": 2.068384653113542e-08,
+      "loss": 0.2456,
+      "step": 8480
+    },
+    {
+      "epoch": 2.9520166898470097,
+      "grad_norm": 1.7958908081054688,
+      "learning_rate": 1.8164683796730866e-08,
+      "loss": 0.2591,
+      "step": 8490
+    },
+    {
+      "epoch": 2.955493741307371,
+      "grad_norm": 1.6148759126663208,
+      "learning_rate": 1.580891149894015e-08,
+      "loss": 0.2435,
+      "step": 8500
+    },
+    {
+      "epoch": 2.958970792767733,
+      "grad_norm": 1.7698160409927368,
+      "learning_rate": 1.3616568198873137e-08,
+      "loss": 0.2702,
+      "step": 8510
+    },
+    {
+      "epoch": 2.9624478442280946,
+      "grad_norm": 1.5261950492858887,
+      "learning_rate": 1.158768978251068e-08,
+      "loss": 0.2545,
+      "step": 8520
+    },
+    {
+      "epoch": 2.965924895688456,
+      "grad_norm": 1.6319481134414673,
+      "learning_rate": 9.722309460109546e-09,
+      "loss": 0.2596,
+      "step": 8530
+    },
+    {
+      "epoch": 2.9694019471488176,
+      "grad_norm": 1.8111101388931274,
+      "learning_rate": 8.020457765663958e-09,
+      "loss": 0.2519,
+      "step": 8540
+    },
+    {
+      "epoch": 2.9728789986091795,
+      "grad_norm": 1.8162568807601929,
+      "learning_rate": 6.4821625564059905e-09,
+      "loss": 0.2625,
+      "step": 8550
+    },
+    {
+      "epoch": 2.976356050069541,
+      "grad_norm": 1.6742914915084839,
+      "learning_rate": 5.107449012347054e-09,
+      "loss": 0.2656,
+      "step": 8560
+    },
+    {
+      "epoch": 2.979833101529903,
+      "grad_norm": 1.6332108974456787,
+      "learning_rate": 3.8963396358648876e-09,
+      "loss": 0.2496,
+      "step": 8570
+    },
+    {
+      "epoch": 2.9833101529902644,
+      "grad_norm": 1.8304611444473267,
+      "learning_rate": 2.848854251340516e-09,
+      "loss": 0.2626,
+      "step": 8580
+    },
+    {
+      "epoch": 2.986787204450626,
+      "grad_norm": 1.723112940788269,
+      "learning_rate": 1.965010004828516e-09,
+      "loss": 0.2496,
+      "step": 8590
+    },
+    {
+      "epoch": 2.9902642559109873,
+      "grad_norm": 1.8070420026779175,
+      "learning_rate": 1.2448213637772377e-09,
+      "loss": 0.2657,
+      "step": 8600
+    },
+    {
+      "epoch": 2.9937413073713492,
+      "grad_norm": 1.8147679567337036,
+      "learning_rate": 6.883001167956594e-10,
+      "loss": 0.2676,
+      "step": 8610
+    },
+    {
+      "epoch": 2.9972183588317107,
+      "grad_norm": 1.75929594039917,
+      "learning_rate": 2.954553734546561e-10,
+      "loss": 0.2628,
+      "step": 8620
+    },
+    {
+      "epoch": 3.0,
+      "step": 8628,
+      "total_flos": 2.049294296112169e+19,
+      "train_loss": 0.5552187451732673,
+      "train_runtime": 27521.6748,
+      "train_samples_per_second": 20.057,
+      "train_steps_per_second": 0.313
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 8628,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10000,
+  "total_flos": 2.049294296112169e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}