diff --git "a/saves-llama-cosine-adagrad/checkpoint-9480/trainer_state.json" "b/saves-llama-cosine-adagrad/checkpoint-9480/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/saves-llama-cosine-adagrad/checkpoint-9480/trainer_state.json"
@@ -0,0 +1,6669 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 9480,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0010548523206751054,
+      "grad_norm": 1.4740744829177856,
+      "learning_rate": 0.00015789473684210527,
+      "loss": 7.6223,
+      "step": 10
+    },
+    {
+      "epoch": 0.002109704641350211,
+      "grad_norm": 1.5953259468078613,
+      "learning_rate": 0.00031578947368421053,
+      "loss": 7.4617,
+      "step": 20
+    },
+    {
+      "epoch": 0.0031645569620253164,
+      "grad_norm": 1.266932487487793,
+      "learning_rate": 0.00047368421052631577,
+      "loss": 7.2219,
+      "step": 30
+    },
+    {
+      "epoch": 0.004219409282700422,
+      "grad_norm": 1.2426646947860718,
+      "learning_rate": 0.0006315789473684211,
+      "loss": 7.0369,
+      "step": 40
+    },
+    {
+      "epoch": 0.005274261603375527,
+      "grad_norm": 1.5997333526611328,
+      "learning_rate": 0.0007894736842105263,
+      "loss": 6.8958,
+      "step": 50
+    },
+    {
+      "epoch": 0.006329113924050633,
+      "grad_norm": 2.029431104660034,
+      "learning_rate": 0.0009473684210526315,
+      "loss": 6.7761,
+      "step": 60
+    },
+    {
+      "epoch": 0.007383966244725738,
+      "grad_norm": 2.4014389514923096,
+      "learning_rate": 0.0011052631578947368,
+      "loss": 6.6805,
+      "step": 70
+    },
+    {
+      "epoch": 0.008438818565400843,
+      "grad_norm": 2.0839943885803223,
+      "learning_rate": 0.0012631578947368421,
+      "loss": 6.6037,
+      "step": 80
+    },
+    {
+      "epoch": 0.00949367088607595,
+      "grad_norm": 2.2696850299835205,
+      "learning_rate": 0.0014210526315789472,
+      "loss": 6.5265,
+      "step": 90
+    },
+    {
+      "epoch": 0.010548523206751054,
+      "grad_norm": 2.13199520111084,
+      "learning_rate": 0.0014999989494847376,
+      "loss": 6.4643,
+      "step": 100
+    },
+    {
+      "epoch": 0.011603375527426161,
+      "grad_norm": 1.889822244644165,
+      "learning_rate": 0.0014999905453802946,
+      "loss": 6.3871,
+      "step": 110
+    },
+    {
+      "epoch": 0.012658227848101266,
+      "grad_norm": 1.705349087715149,
+      "learning_rate": 0.0014999737372655805,
+      "loss": 6.3232,
+      "step": 120
+    },
+    {
+      "epoch": 0.013713080168776372,
+      "grad_norm": 1.527263879776001,
+      "learning_rate": 0.0014999485253289388,
+      "loss": 6.2557,
+      "step": 130
+    },
+    {
+      "epoch": 0.014767932489451477,
+      "grad_norm": 1.3018344640731812,
+      "learning_rate": 0.0014999149098528814,
+      "loss": 6.1917,
+      "step": 140
+    },
+    {
+      "epoch": 0.015822784810126583,
+      "grad_norm": 1.2581422328948975,
+      "learning_rate": 0.0014998728912140862,
+      "loss": 6.1313,
+      "step": 150
+    },
+    {
+      "epoch": 0.016877637130801686,
+      "grad_norm": 1.011123776435852,
+      "learning_rate": 0.0014998224698833922,
+      "loss": 6.0815,
+      "step": 160
+    },
+    {
+      "epoch": 0.017932489451476793,
+      "grad_norm": 0.9474137425422668,
+      "learning_rate": 0.0014997636464257956,
+      "loss": 6.0286,
+      "step": 170
+    },
+    {
+      "epoch": 0.0189873417721519,
+      "grad_norm": 0.6689318418502808,
+      "learning_rate": 0.0014996964215004416,
+      "loss": 5.9996,
+      "step": 180
+    },
+    {
+      "epoch": 0.020042194092827006,
+      "grad_norm": 0.9522222876548767,
+      "learning_rate": 0.0014996207958606182,
+      "loss": 5.9749,
+      "step": 190
+    },
+    {
+      "epoch": 0.02109704641350211,
+      "grad_norm": 1.007729411125183,
+      "learning_rate": 0.001499536770353748,
+      "loss": 5.9511,
+      "step": 200
+    },
+    {
+      "epoch": 0.022151898734177215,
+      "grad_norm": 0.7111568450927734,
+      "learning_rate": 0.0014994443459213774,
+      "loss": 5.9417,
+      "step": 210
+    },
+    {
+      "epoch": 0.023206751054852322,
+      "grad_norm": 1.0027458667755127,
+      "learning_rate": 0.001499343523599168,
+      "loss": 5.922,
+      "step": 220
+    },
+    {
+      "epoch": 0.024261603375527425,
+      "grad_norm": 0.7834833860397339,
+      "learning_rate": 0.0014992343045168823,
+      "loss": 5.9105,
+      "step": 230
+    },
+    {
+      "epoch": 0.02531645569620253,
+      "grad_norm": 0.9297846555709839,
+      "learning_rate": 0.0014991166898983739,
+      "loss": 5.8912,
+      "step": 240
+    },
+    {
+      "epoch": 0.026371308016877638,
+      "grad_norm": 0.8988327383995056,
+      "learning_rate": 0.001498990681061572,
+      "loss": 5.8794,
+      "step": 250
+    },
+    {
+      "epoch": 0.027426160337552744,
+      "grad_norm": 1.145376205444336,
+      "learning_rate": 0.001498856279418467,
+      "loss": 5.8698,
+      "step": 260
+    },
+    {
+      "epoch": 0.028481012658227847,
+      "grad_norm": 0.956051766872406,
+      "learning_rate": 0.0014987134864750948,
+      "loss": 5.8643,
+      "step": 270
+    },
+    {
+      "epoch": 0.029535864978902954,
+      "grad_norm": 0.9607803225517273,
+      "learning_rate": 0.0014985623038315206,
+      "loss": 5.843,
+      "step": 280
+    },
+    {
+      "epoch": 0.03059071729957806,
+      "grad_norm": 1.0926345586776733,
+      "learning_rate": 0.0014984027331818193,
+      "loss": 5.8377,
+      "step": 290
+    },
+    {
+      "epoch": 0.03164556962025317,
+      "grad_norm": 1.0664520263671875,
+      "learning_rate": 0.0014982347763140584,
+      "loss": 5.8195,
+      "step": 300
+    },
+    {
+      "epoch": 0.03270042194092827,
+      "grad_norm": 1.101575493812561,
+      "learning_rate": 0.0014980584351102762,
+      "loss": 5.8048,
+      "step": 310
+    },
+    {
+      "epoch": 0.03375527426160337,
+      "grad_norm": 1.3754583597183228,
+      "learning_rate": 0.001497873711546462,
+      "loss": 5.7789,
+      "step": 320
+    },
+    {
+      "epoch": 0.03481012658227848,
+      "grad_norm": 1.1685811281204224,
+      "learning_rate": 0.0014976806076925334,
+      "loss": 5.7671,
+      "step": 330
+    },
+    {
+      "epoch": 0.035864978902953586,
+      "grad_norm": 0.9515827894210815,
+      "learning_rate": 0.0014974791257123137,
+      "loss": 5.7445,
+      "step": 340
+    },
+    {
+      "epoch": 0.03691983122362869,
+      "grad_norm": 1.0288505554199219,
+      "learning_rate": 0.001497269267863507,
+      "loss": 5.7182,
+      "step": 350
+    },
+    {
+      "epoch": 0.0379746835443038,
+      "grad_norm": 1.23520827293396,
+      "learning_rate": 0.0014970510364976724,
+      "loss": 5.6994,
+      "step": 360
+    },
+    {
+      "epoch": 0.039029535864978905,
+      "grad_norm": 1.2275160551071167,
+      "learning_rate": 0.0014968244340601996,
+      "loss": 5.6822,
+      "step": 370
+    },
+    {
+      "epoch": 0.04008438818565401,
+      "grad_norm": 1.2088491916656494,
+      "learning_rate": 0.001496589463090279,
+      "loss": 5.671,
+      "step": 380
+    },
+    {
+      "epoch": 0.04113924050632911,
+      "grad_norm": 1.3245712518692017,
+      "learning_rate": 0.001496346126220875,
+      "loss": 5.6534,
+      "step": 390
+    },
+    {
+      "epoch": 0.04219409282700422,
+      "grad_norm": 1.5248364210128784,
+      "learning_rate": 0.0014960944261786966,
+      "loss": 5.6385,
+      "step": 400
+    },
+    {
+      "epoch": 0.043248945147679324,
+      "grad_norm": 1.2704198360443115,
+      "learning_rate": 0.0014958343657841655,
+      "loss": 5.6374,
+      "step": 410
+    },
+    {
+      "epoch": 0.04430379746835443,
+      "grad_norm": 1.5440633296966553,
+      "learning_rate": 0.001495565947951385,
+      "loss": 5.6314,
+      "step": 420
+    },
+    {
+      "epoch": 0.04535864978902954,
+      "grad_norm": 1.22713041305542,
+      "learning_rate": 0.0014952891756881085,
+      "loss": 5.6209,
+      "step": 430
+    },
+    {
+      "epoch": 0.046413502109704644,
+      "grad_norm": 1.3235241174697876,
+      "learning_rate": 0.0014950040520957037,
+      "loss": 5.6177,
+      "step": 440
+    },
+    {
+      "epoch": 0.04746835443037975,
+      "grad_norm": 1.2323728799819946,
+      "learning_rate": 0.0014947105803691204,
+      "loss": 5.613,
+      "step": 450
+    },
+    {
+      "epoch": 0.04852320675105485,
+      "grad_norm": 1.298741340637207,
+      "learning_rate": 0.0014944087637968522,
+      "loss": 5.6031,
+      "step": 460
+    },
+    {
+      "epoch": 0.049578059071729956,
+      "grad_norm": 1.358190655708313,
+      "learning_rate": 0.0014940986057609012,
+      "loss": 5.5944,
+      "step": 470
+    },
+    {
+      "epoch": 0.05063291139240506,
+      "grad_norm": 1.1745429039001465,
+      "learning_rate": 0.0014937801097367396,
+      "loss": 5.6014,
+      "step": 480
+    },
+    {
+      "epoch": 0.05168776371308017,
+      "grad_norm": 1.0948045253753662,
+      "learning_rate": 0.001493453279293271,
+      "loss": 5.5886,
+      "step": 490
+    },
+    {
+      "epoch": 0.052742616033755275,
+      "grad_norm": 1.3333172798156738,
+      "learning_rate": 0.0014931181180927902,
+      "loss": 5.5854,
+      "step": 500
+    },
+    {
+      "epoch": 0.05379746835443038,
+      "grad_norm": 1.3620946407318115,
+      "learning_rate": 0.001492774629890942,
+      "loss": 5.5951,
+      "step": 510
+    },
+    {
+      "epoch": 0.05485232067510549,
+      "grad_norm": 1.3252971172332764,
+      "learning_rate": 0.001492422818536679,
+      "loss": 5.5854,
+      "step": 520
+    },
+    {
+      "epoch": 0.05590717299578059,
+      "grad_norm": 1.346969485282898,
+      "learning_rate": 0.00149206268797222,
+      "loss": 5.5828,
+      "step": 530
+    },
+    {
+      "epoch": 0.056962025316455694,
+      "grad_norm": 1.4123709201812744,
+      "learning_rate": 0.0014916942422330032,
+      "loss": 5.586,
+      "step": 540
+    },
+    {
+      "epoch": 0.0580168776371308,
+      "grad_norm": 1.6085889339447021,
+      "learning_rate": 0.001491317485447643,
+      "loss": 5.5826,
+      "step": 550
+    },
+    {
+      "epoch": 0.05907172995780591,
+      "grad_norm": 1.2459216117858887,
+      "learning_rate": 0.0014909324218378838,
+      "loss": 5.5792,
+      "step": 560
+    },
+    {
+      "epoch": 0.060126582278481014,
+      "grad_norm": 1.4043864011764526,
+      "learning_rate": 0.0014905390557185508,
+      "loss": 5.5783,
+      "step": 570
+    },
+    {
+      "epoch": 0.06118143459915612,
+      "grad_norm": 1.1678298711776733,
+      "learning_rate": 0.0014901373914975036,
+      "loss": 5.5806,
+      "step": 580
+    },
+    {
+      "epoch": 0.06223628691983123,
+      "grad_norm": 1.264378309249878,
+      "learning_rate": 0.0014897274336755856,
+      "loss": 5.5735,
+      "step": 590
+    },
+    {
+      "epoch": 0.06329113924050633,
+      "grad_norm": 1.2043378353118896,
+      "learning_rate": 0.001489309186846575,
+      "loss": 5.5668,
+      "step": 600
+    },
+    {
+      "epoch": 0.06434599156118144,
+      "grad_norm": 1.6404099464416504,
+      "learning_rate": 0.0014888826556971313,
+      "loss": 5.5739,
+      "step": 610
+    },
+    {
+      "epoch": 0.06540084388185655,
+      "grad_norm": 1.4162348508834839,
+      "learning_rate": 0.0014884478450067444,
+      "loss": 5.5762,
+      "step": 620
+    },
+    {
+      "epoch": 0.06645569620253164,
+      "grad_norm": 1.448990821838379,
+      "learning_rate": 0.0014880047596476807,
+      "loss": 5.5697,
+      "step": 630
+    },
+    {
+      "epoch": 0.06751054852320675,
+      "grad_norm": 1.5138314962387085,
+      "learning_rate": 0.0014875534045849274,
+      "loss": 5.5738,
+      "step": 640
+    },
+    {
+      "epoch": 0.06856540084388185,
+      "grad_norm": 1.3093619346618652,
+      "learning_rate": 0.0014870937848761388,
+      "loss": 5.5768,
+      "step": 650
+    },
+    {
+      "epoch": 0.06962025316455696,
+      "grad_norm": 1.5481593608856201,
+      "learning_rate": 0.001486625905671578,
+      "loss": 5.5778,
+      "step": 660
+    },
+    {
+      "epoch": 0.07067510548523206,
+      "grad_norm": 1.4933645725250244,
+      "learning_rate": 0.00148614977221406,
+      "loss": 5.5639,
+      "step": 670
+    },
+    {
+      "epoch": 0.07172995780590717,
+      "grad_norm": 1.3248463869094849,
+      "learning_rate": 0.0014856653898388927,
+      "loss": 5.5652,
+      "step": 680
+    },
+    {
+      "epoch": 0.07278481012658228,
+      "grad_norm": 1.4051345586776733,
+      "learning_rate": 0.001485172763973817,
+      "loss": 5.572,
+      "step": 690
+    },
+    {
+      "epoch": 0.07383966244725738,
+      "grad_norm": 1.340245008468628,
+      "learning_rate": 0.0014846719001389466,
+      "loss": 5.5721,
+      "step": 700
+    },
+    {
+      "epoch": 0.07489451476793249,
+      "grad_norm": 1.2932738065719604,
+      "learning_rate": 0.001484162803946705,
+      "loss": 5.5648,
+      "step": 710
+    },
+    {
+      "epoch": 0.0759493670886076,
+      "grad_norm": 1.4179991483688354,
+      "learning_rate": 0.0014836454811017635,
+      "loss": 5.5614,
+      "step": 720
+    },
+    {
+      "epoch": 0.0770042194092827,
+      "grad_norm": 1.3901480436325073,
+      "learning_rate": 0.0014831199374009778,
+      "loss": 5.5628,
+      "step": 730
+    },
+    {
+      "epoch": 0.07805907172995781,
+      "grad_norm": 1.3948917388916016,
+      "learning_rate": 0.0014825861787333208,
+      "loss": 5.5688,
+      "step": 740
+    },
+    {
+      "epoch": 0.07911392405063292,
+      "grad_norm": 1.3099560737609863,
+      "learning_rate": 0.0014820442110798197,
+      "loss": 5.5676,
+      "step": 750
+    },
+    {
+      "epoch": 0.08016877637130802,
+      "grad_norm": 1.3314452171325684,
+      "learning_rate": 0.0014814940405134865,
+      "loss": 5.5701,
+      "step": 760
+    },
+    {
+      "epoch": 0.08122362869198312,
+      "grad_norm": 1.6187591552734375,
+      "learning_rate": 0.001480935673199251,
+      "loss": 5.5626,
+      "step": 770
+    },
+    {
+      "epoch": 0.08227848101265822,
+      "grad_norm": 1.6323590278625488,
+      "learning_rate": 0.0014803691153938915,
+      "loss": 5.5597,
+      "step": 780
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "grad_norm": 1.1413952112197876,
+      "learning_rate": 0.0014797943734459653,
+      "loss": 5.5749,
+      "step": 790
+    },
+    {
+      "epoch": 0.08438818565400844,
+      "grad_norm": 1.3629881143569946,
+      "learning_rate": 0.001479211453795736,
+      "loss": 5.5723,
+      "step": 800
+    },
+    {
+      "epoch": 0.08544303797468354,
+      "grad_norm": 1.2347767353057861,
+      "learning_rate": 0.0014786203629751033,
+      "loss": 5.5667,
+      "step": 810
+    },
+    {
+      "epoch": 0.08649789029535865,
+      "grad_norm": 1.530195713043213,
+      "learning_rate": 0.0014780211076075279,
+      "loss": 5.5673,
+      "step": 820
+    },
+    {
+      "epoch": 0.08755274261603375,
+      "grad_norm": 1.4960066080093384,
+      "learning_rate": 0.0014774136944079594,
+      "loss": 5.5759,
+      "step": 830
+    },
+    {
+      "epoch": 0.08860759493670886,
+      "grad_norm": 1.3552134037017822,
+      "learning_rate": 0.0014767981301827592,
+      "loss": 5.5618,
+      "step": 840
+    },
+    {
+      "epoch": 0.08966244725738397,
+      "grad_norm": 1.0897563695907593,
+      "learning_rate": 0.0014761744218296249,
+      "loss": 5.569,
+      "step": 850
+    },
+    {
+      "epoch": 0.09071729957805907,
+      "grad_norm": 1.4627890586853027,
+      "learning_rate": 0.001475542576337513,
+      "loss": 5.565,
+      "step": 860
+    },
+    {
+      "epoch": 0.09177215189873418,
+      "grad_norm": 1.5468559265136719,
+      "learning_rate": 0.001474902600786561,
+      "loss": 5.5695,
+      "step": 870
+    },
+    {
+      "epoch": 0.09282700421940929,
+      "grad_norm": 1.3450201749801636,
+      "learning_rate": 0.0014742545023480075,
+      "loss": 5.5753,
+      "step": 880
+    },
+    {
+      "epoch": 0.0938818565400844,
+      "grad_norm": 1.503970742225647,
+      "learning_rate": 0.0014735982882841117,
+      "loss": 5.5737,
+      "step": 890
+    },
+    {
+      "epoch": 0.0949367088607595,
+      "grad_norm": 1.2197023630142212,
+      "learning_rate": 0.0014729339659480727,
+      "loss": 5.5697,
+      "step": 900
+    },
+    {
+      "epoch": 0.09599156118143459,
+      "grad_norm": 1.497017502784729,
+      "learning_rate": 0.0014722615427839468,
+      "loss": 5.5658,
+      "step": 910
+    },
+    {
+      "epoch": 0.0970464135021097,
+      "grad_norm": 1.4540644884109497,
+      "learning_rate": 0.0014715810263265633,
+      "loss": 5.5713,
+      "step": 920
+    },
+    {
+      "epoch": 0.0981012658227848,
+      "grad_norm": 1.4214389324188232,
+      "learning_rate": 0.0014708924242014423,
+      "loss": 5.5728,
+      "step": 930
+    },
+    {
+      "epoch": 0.09915611814345991,
+      "grad_norm": 1.7204481363296509,
+      "learning_rate": 0.0014701957441247064,
+      "loss": 5.5705,
+      "step": 940
+    },
+    {
+      "epoch": 0.10021097046413502,
+      "grad_norm": 1.5736048221588135,
+      "learning_rate": 0.0014694909939029959,
+      "loss": 5.5761,
+      "step": 950
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 1.7689915895462036,
+      "learning_rate": 0.0014687781814333814,
+      "loss": 5.572,
+      "step": 960
+    },
+    {
+      "epoch": 0.10232067510548523,
+      "grad_norm": 1.5722955465316772,
+      "learning_rate": 0.0014680573147032746,
+      "loss": 5.5743,
+      "step": 970
+    },
+    {
+      "epoch": 0.10337552742616034,
+      "grad_norm": 1.6428799629211426,
+      "learning_rate": 0.0014673284017903392,
+      "loss": 5.5751,
+      "step": 980
+    },
+    {
+      "epoch": 0.10443037974683544,
+      "grad_norm": 1.5544874668121338,
+      "learning_rate": 0.0014665914508624,
+      "loss": 5.5799,
+      "step": 990
+    },
+    {
+      "epoch": 0.10548523206751055,
+      "grad_norm": 1.3315614461898804,
+      "learning_rate": 0.0014658464701773526,
+      "loss": 5.5774,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10654008438818566,
+      "grad_norm": 1.4198408126831055,
+      "learning_rate": 0.0014650934680830688,
+      "loss": 5.5845,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10759493670886076,
+      "grad_norm": 1.3419196605682373,
+      "learning_rate": 0.0014643324530173051,
+      "loss": 5.5836,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10864978902953587,
+      "grad_norm": 1.5367684364318848,
+      "learning_rate": 0.0014635634335076067,
+      "loss": 5.5822,
+      "step": 1030
+    },
+    {
+      "epoch": 0.10970464135021098,
+      "grad_norm": 1.383516550064087,
+      "learning_rate": 0.001462786418171213,
+      "loss": 5.5793,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11075949367088607,
+      "grad_norm": 1.8112094402313232,
+      "learning_rate": 0.0014620014157149597,
+      "loss": 5.5821,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11181434599156118,
+      "grad_norm": 1.631983757019043,
+      "learning_rate": 0.001461208434935183,
+      "loss": 5.5839,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11286919831223628,
+      "grad_norm": 1.4845349788665771,
+      "learning_rate": 0.0014604074847176197,
+      "loss": 5.5821,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11392405063291139,
+      "grad_norm": 1.5918809175491333,
+      "learning_rate": 0.0014595985740373082,
+      "loss": 5.59,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1149789029535865,
+      "grad_norm": 1.6206207275390625,
+      "learning_rate": 0.0014587817119584873,
+      "loss": 5.5903,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1160337552742616,
+      "grad_norm": 1.5409072637557983,
+      "learning_rate": 0.001457956907634496,
+      "loss": 5.5827,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11708860759493671,
+      "grad_norm": 1.4688796997070312,
+      "learning_rate": 0.0014571241703076692,
+      "loss": 5.5874,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11814345991561181,
+      "grad_norm": 1.3263105154037476,
+      "learning_rate": 0.0014562835093092348,
+      "loss": 5.5926,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11919831223628692,
+      "grad_norm": 1.4715784788131714,
+      "learning_rate": 0.0014554349340592104,
+      "loss": 5.5796,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12025316455696203,
+      "grad_norm": 1.5376423597335815,
+      "learning_rate": 0.001454578454066296,
+      "loss": 5.5896,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12130801687763713,
+      "grad_norm": 1.3840242624282837,
+      "learning_rate": 0.0014537140789277678,
+      "loss": 5.5811,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12236286919831224,
+      "grad_norm": 1.6980022192001343,
+      "learning_rate": 0.0014528418183293716,
+      "loss": 5.5872,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12341772151898735,
+      "grad_norm": 1.777838110923767,
+      "learning_rate": 0.001451961682045213,
+      "loss": 5.5847,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12447257383966245,
+      "grad_norm": 1.4961351156234741,
+      "learning_rate": 0.001451073679937649,
+      "loss": 5.5869,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12552742616033755,
+      "grad_norm": 2.0904078483581543,
+      "learning_rate": 0.0014501778219571766,
+      "loss": 5.5893,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12658227848101267,
+      "grad_norm": 1.6259174346923828,
+      "learning_rate": 0.0014492741181423225,
+      "loss": 5.5881,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12763713080168776,
+      "grad_norm": 1.6919883489608765,
+      "learning_rate": 0.0014483625786195285,
+      "loss": 5.5973,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12869198312236288,
+      "grad_norm": 1.5741194486618042,
+      "learning_rate": 0.0014474432136030405,
+      "loss": 5.5912,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12974683544303797,
+      "grad_norm": 1.6624023914337158,
+      "learning_rate": 0.0014465160333947923,
+      "loss": 5.5889,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1308016877637131,
+      "grad_norm": 1.7581974267959595,
+      "learning_rate": 0.0014455810483842908,
+      "loss": 5.5933,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13185654008438819,
+      "grad_norm": 1.6439803838729858,
+      "learning_rate": 0.0014446382690484997,
+      "loss": 5.5979,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13291139240506328,
+      "grad_norm": 2.0544514656066895,
+      "learning_rate": 0.0014436877059517215,
+      "loss": 5.5973,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1339662447257384,
+      "grad_norm": 1.6659609079360962,
+      "learning_rate": 0.0014427293697454803,
+      "loss": 5.5942,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1350210970464135,
+      "grad_norm": 1.6619070768356323,
+      "learning_rate": 0.001441763271168401,
+      "loss": 5.5963,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1360759493670886,
+      "grad_norm": 1.6102874279022217,
+      "learning_rate": 0.00144078942104609,
+      "loss": 5.6023,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1371308016877637,
+      "grad_norm": 1.7208929061889648,
+      "learning_rate": 0.001439807830291013,
+      "loss": 5.5993,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13818565400843882,
+      "grad_norm": 1.7440332174301147,
+      "learning_rate": 0.0014388185099023744,
+      "loss": 5.6039,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13924050632911392,
+      "grad_norm": 1.7689586877822876,
+      "learning_rate": 0.0014378214709659916,
+      "loss": 5.6147,
+      "step": 1320
+    },
+    {
+      "epoch": 0.14029535864978904,
+      "grad_norm": 1.5448212623596191,
+      "learning_rate": 0.0014368167246541733,
+      "loss": 5.6091,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14135021097046413,
+      "grad_norm": 1.894848108291626,
+      "learning_rate": 0.0014358042822255918,
+      "loss": 5.6198,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14240506329113925,
+      "grad_norm": 1.6183972358703613,
+      "learning_rate": 0.0014347841550251597,
+      "loss": 5.6247,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14345991561181434,
+      "grad_norm": 2.19899320602417,
+      "learning_rate": 0.0014337563544838997,
+      "loss": 5.6135,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14451476793248946,
+      "grad_norm": 1.6558045148849487,
+      "learning_rate": 0.001432720892118819,
+      "loss": 5.6173,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14556962025316456,
+      "grad_norm": 1.6515460014343262,
+      "learning_rate": 0.0014316777795327794,
+      "loss": 5.615,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14662447257383968,
+      "grad_norm": 1.7529590129852295,
+      "learning_rate": 0.001430627028414366,
+      "loss": 5.627,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14767932489451477,
+      "grad_norm": 1.8392685651779175,
+      "learning_rate": 0.0014295686505377586,
+      "loss": 5.6249,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14873417721518986,
+      "grad_norm": 1.7649297714233398,
+      "learning_rate": 0.0014285026577625982,
+      "loss": 5.6251,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14978902953586498,
+      "grad_norm": 2.3213040828704834,
+      "learning_rate": 0.0014274290620338542,
+      "loss": 5.622,
+      "step": 1420
+    },
+    {
+      "epoch": 0.15084388185654007,
+      "grad_norm": 1.7446365356445312,
+      "learning_rate": 0.0014263478753816906,
+      "loss": 5.6247,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1518987341772152,
+      "grad_norm": 1.9416145086288452,
+      "learning_rate": 0.0014252591099213326,
+      "loss": 5.6325,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1529535864978903,
+      "grad_norm": 1.9150333404541016,
+      "learning_rate": 0.001424162777852928,
+      "loss": 5.6283,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1540084388185654,
+      "grad_norm": 1.7787129878997803,
+      "learning_rate": 0.0014230588914614134,
+      "loss": 5.6269,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1550632911392405,
+      "grad_norm": 1.9343360662460327,
+      "learning_rate": 0.0014219474631163745,
+      "loss": 5.6276,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15611814345991562,
+      "grad_norm": 2.2496066093444824,
+      "learning_rate": 0.001420828505271909,
+      "loss": 5.6358,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1571729957805907,
+      "grad_norm": 1.9158469438552856,
+      "learning_rate": 0.0014197020304664856,
+      "loss": 5.6367,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15822784810126583,
+      "grad_norm": 2.1325016021728516,
+      "learning_rate": 0.0014185680513228048,
+      "loss": 5.638,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15928270042194093,
+      "grad_norm": 1.9436272382736206,
+      "learning_rate": 0.0014174265805476564,
+      "loss": 5.6469,
+      "step": 1510
+    },
+    {
+      "epoch": 0.16033755274261605,
+      "grad_norm": 1.856324553489685,
+      "learning_rate": 0.0014162776309317778,
+      "loss": 5.6463,
+      "step": 1520
+    },
+    {
+      "epoch": 0.16139240506329114,
+      "grad_norm": 1.8099374771118164,
+      "learning_rate": 0.0014151212153497108,
+      "loss": 5.6421,
+      "step": 1530
+    },
+    {
+      "epoch": 0.16244725738396623,
+      "grad_norm": 2.1700191497802734,
+      "learning_rate": 0.0014139573467596561,
+      "loss": 5.6376,
+      "step": 1540
+    },
+    {
+      "epoch": 0.16350210970464135,
+      "grad_norm": 2.1919312477111816,
+      "learning_rate": 0.00141278603820333,
+      "loss": 5.6403,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16455696202531644,
+      "grad_norm": 1.9078559875488281,
+      "learning_rate": 0.0014116073028058165,
+      "loss": 5.6454,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16561181434599156,
+      "grad_norm": 1.9832590818405151,
+      "learning_rate": 0.0014104211537754217,
+      "loss": 5.6445,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 1.9779951572418213,
+      "learning_rate": 0.001409227604403524,
+      "loss": 5.6512,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16772151898734178,
+      "grad_norm": 1.8629887104034424,
+      "learning_rate": 0.0014080266680644277,
+      "loss": 5.6494,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16877637130801687,
+      "grad_norm": 2.262251138687134,
+      "learning_rate": 0.0014068183582152103,
+      "loss": 5.6569,
+      "step": 1600
+    },
+    {
+      "epoch": 0.169831223628692,
+      "grad_norm": 1.77689790725708,
+      "learning_rate": 0.001405602688395574,
+      "loss": 5.6569,
+      "step": 1610
+    },
+    {
+      "epoch": 0.17088607594936708,
+      "grad_norm": 2.041045665740967,
+      "learning_rate": 0.0014043796722276924,
+      "loss": 5.6472,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1719409282700422,
+      "grad_norm": 2.154695987701416,
+      "learning_rate": 0.0014031493234160591,
+      "loss": 5.6617,
+      "step": 1630
+    },
+    {
+      "epoch": 0.1729957805907173,
+      "grad_norm": 2.1585569381713867,
+      "learning_rate": 0.0014019116557473332,
+      "loss": 5.6588,
+      "step": 1640
+    },
+    {
+      "epoch": 0.17405063291139242,
+      "grad_norm": 2.1053874492645264,
+      "learning_rate": 0.0014006666830901854,
+      "loss": 5.6587,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1751054852320675,
+      "grad_norm": 2.0237772464752197,
+      "learning_rate": 0.001399414419395142,
+      "loss": 5.6591,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17616033755274263,
+      "grad_norm": 2.555802822113037,
+      "learning_rate": 0.0013981548786944293,
+      "loss": 5.6673,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17721518987341772,
+      "grad_norm": 1.9266884326934814,
+      "learning_rate": 0.0013968880751018158,
+      "loss": 5.6601,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17827004219409281,
+      "grad_norm": 2.1171956062316895,
+      "learning_rate": 0.0013956140228124545,
+      "loss": 5.6623,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17932489451476794,
+      "grad_norm": 1.7413369417190552,
+      "learning_rate": 0.0013943327361027231,
+      "loss": 5.6674,
+      "step": 1700
+    },
+    {
+      "epoch": 0.18037974683544303,
+      "grad_norm": 2.0904786586761475,
+      "learning_rate": 0.0013930442293300649,
+      "loss": 5.6677,
+      "step": 1710
+    },
+    {
+      "epoch": 0.18143459915611815,
+      "grad_norm": 2.2373464107513428,
+      "learning_rate": 0.0013917485169328279,
+      "loss": 5.6704,
+      "step": 1720
+    },
+    {
+      "epoch": 0.18248945147679324,
+      "grad_norm": 1.9037578105926514,
+      "learning_rate": 0.0013904456134301016,
+      "loss": 5.6744,
+      "step": 1730
+    },
+    {
+      "epoch": 0.18354430379746836,
+      "grad_norm": 2.1267647743225098,
+      "learning_rate": 0.0013891355334215562,
+      "loss": 5.675,
+      "step": 1740
+    },
+    {
+      "epoch": 0.18459915611814345,
+      "grad_norm": 1.9052635431289673,
+      "learning_rate": 0.0013878182915872776,
+      "loss": 5.679,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18565400843881857,
+      "grad_norm": 2.2510416507720947,
+      "learning_rate": 0.001386493902687604,
+      "loss": 5.6765,
+      "step": 1760
+    },
+    {
+      "epoch": 0.18670886075949367,
+      "grad_norm": 2.3105335235595703,
+      "learning_rate": 0.00138516238156296,
+      "loss": 5.6772,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1877637130801688,
+      "grad_norm": 2.148183584213257,
+      "learning_rate": 0.0013838237431336895,
+      "loss": 5.6805,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18881856540084388,
+      "grad_norm": 2.119838237762451,
+      "learning_rate": 0.0013824780023998899,
+      "loss": 5.6805,
+      "step": 1790
+    },
+    {
+      "epoch": 0.189873417721519,
+      "grad_norm": 2.194430351257324,
+      "learning_rate": 0.0013811251744412431,
+      "loss": 5.6816,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1909282700421941,
+      "grad_norm": 2.0297439098358154,
+      "learning_rate": 0.0013797652744168473,
+      "loss": 5.679,
+      "step": 1810
+    },
+    {
+      "epoch": 0.19198312236286919,
+      "grad_norm": 2.3295819759368896,
+      "learning_rate": 0.0013783983175650457,
+      "loss": 5.6801,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1930379746835443,
+      "grad_norm": 2.233726978302002,
+      "learning_rate": 0.0013770243192032581,
+      "loss": 5.6773,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1940928270042194,
+      "grad_norm": 2.2201192378997803,
+      "learning_rate": 0.0013756432947278064,
+      "loss": 5.6799,
+      "step": 1840
+    },
+    {
+      "epoch": 0.19514767932489452,
+      "grad_norm": 2.0483968257904053,
+      "learning_rate": 0.0013742552596137444,
+      "loss": 5.6797,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1962025316455696,
+      "grad_norm": 2.2263448238372803,
+      "learning_rate": 0.0013728602294146833,
+      "loss": 5.6838,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19725738396624473,
+      "grad_norm": 2.118208169937134,
+      "learning_rate": 0.0013714582197626175,
+      "loss": 5.6821,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19831223628691982,
+      "grad_norm": 2.0614240169525146,
+      "learning_rate": 0.0013700492463677501,
+      "loss": 5.6886,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19936708860759494,
+      "grad_norm": 2.2452991008758545,
+      "learning_rate": 0.0013686333250183154,
+      "loss": 5.6724,
+      "step": 1890
+    },
+    {
+      "epoch": 0.20042194092827004,
+      "grad_norm": 2.311352014541626,
+      "learning_rate": 0.001367210471580404,
+      "loss": 5.6902,
+      "step": 1900
+    },
+    {
+      "epoch": 0.20147679324894516,
+      "grad_norm": 2.0389885902404785,
+      "learning_rate": 0.0013657807019977835,
+      "loss": 5.6877,
+      "step": 1910
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 2.263408660888672,
+      "learning_rate": 0.0013643440322917198,
+      "loss": 5.6944,
+      "step": 1920
+    },
+    {
+      "epoch": 0.20358649789029537,
+      "grad_norm": 2.244577169418335,
+      "learning_rate": 0.0013629004785607989,
+      "loss": 5.6962,
+      "step": 1930
+    },
+    {
+      "epoch": 0.20464135021097046,
+      "grad_norm": 2.340639352798462,
+      "learning_rate": 0.0013614500569807445,
+      "loss": 5.6918,
+      "step": 1940
+    },
+    {
+      "epoch": 0.20569620253164558,
+      "grad_norm": 1.9686450958251953,
+      "learning_rate": 0.0013599927838042394,
+      "loss": 5.697,
+      "step": 1950
+    },
+    {
+      "epoch": 0.20675105485232068,
+      "grad_norm": 2.8002195358276367,
+      "learning_rate": 0.0013585286753607408,
+      "loss": 5.6937,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20780590717299577,
+      "grad_norm": 2.466768264770508,
+      "learning_rate": 0.0013570577480562986,
+      "loss": 5.6997,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2088607594936709,
+      "grad_norm": 2.067009210586548,
+      "learning_rate": 0.0013555800183733717,
+      "loss": 5.6991,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20991561181434598,
+      "grad_norm": 2.600818395614624,
+      "learning_rate": 0.0013540955028706425,
+      "loss": 5.6956,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2109704641350211,
+      "grad_norm": 2.1557400226593018,
+      "learning_rate": 0.0013526042181828324,
+      "loss": 5.691,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2120253164556962,
+      "grad_norm": 2.2260053157806396,
+      "learning_rate": 0.0013511061810205143,
+      "loss": 5.6984,
+      "step": 2010
+    },
+    {
+      "epoch": 0.21308016877637131,
+      "grad_norm": 1.9528491497039795,
+      "learning_rate": 0.001349601408169926,
+      "loss": 5.707,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2141350210970464,
+      "grad_norm": 2.321481943130493,
+      "learning_rate": 0.0013480899164927823,
+      "loss": 5.704,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21518987341772153,
+      "grad_norm": 2.052384376525879,
+      "learning_rate": 0.0013465717229260853,
+      "loss": 5.6963,
+      "step": 2040
+    },
+    {
+      "epoch": 0.21624472573839662,
+      "grad_norm": 2.2231664657592773,
+      "learning_rate": 0.001345046844481935,
+      "loss": 5.7025,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21729957805907174,
+      "grad_norm": 2.580504894256592,
+      "learning_rate": 0.0013435152982473396,
+      "loss": 5.6982,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21835443037974683,
+      "grad_norm": 2.074516773223877,
+      "learning_rate": 0.0013419771013840217,
+      "loss": 5.7048,
+      "step": 2070
+    },
+    {
+      "epoch": 0.21940928270042195,
+      "grad_norm": 2.2731943130493164,
+      "learning_rate": 0.001340432271128229,
+      "loss": 5.7107,
+      "step": 2080
+    },
+    {
+      "epoch": 0.22046413502109705,
+      "grad_norm": 2.1779603958129883,
+      "learning_rate": 0.0013388808247905381,
+      "loss": 5.7014,
+      "step": 2090
+    },
+    {
+      "epoch": 0.22151898734177214,
+      "grad_norm": 2.2784008979797363,
+      "learning_rate": 0.0013373227797556634,
+      "loss": 5.7022,
+      "step": 2100
+    },
+    {
+      "epoch": 0.22257383966244726,
+      "grad_norm": 1.995540738105774,
+      "learning_rate": 0.00133575815348226,
+      "loss": 5.705,
+      "step": 2110
+    },
+    {
+      "epoch": 0.22362869198312235,
+      "grad_norm": 2.5391671657562256,
+      "learning_rate": 0.0013341869635027292,
+      "loss": 5.6984,
+      "step": 2120
+    },
+    {
+      "epoch": 0.22468354430379747,
+      "grad_norm": 2.3931891918182373,
+      "learning_rate": 0.001332609227423022,
+      "loss": 5.7039,
+      "step": 2130
+    },
+    {
+      "epoch": 0.22573839662447256,
+      "grad_norm": 2.363659381866455,
+      "learning_rate": 0.0013310249629224417,
+      "loss": 5.699,
+      "step": 2140
+    },
+    {
+      "epoch": 0.22679324894514769,
+      "grad_norm": 2.2495954036712646,
+      "learning_rate": 0.0013294341877534454,
+      "loss": 5.7046,
+      "step": 2150
+    },
+    {
+      "epoch": 0.22784810126582278,
+      "grad_norm": 2.256458282470703,
+      "learning_rate": 0.0013278369197414458,
+      "loss": 5.7106,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2289029535864979,
+      "grad_norm": 2.0141658782958984,
+      "learning_rate": 0.0013262331767846104,
+      "loss": 5.7094,
+      "step": 2170
+    },
+    {
+      "epoch": 0.229957805907173,
+      "grad_norm": 2.468933582305908,
+      "learning_rate": 0.0013246229768536628,
+      "loss": 5.7042,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2310126582278481,
+      "grad_norm": 2.205190896987915,
+      "learning_rate": 0.001323006337991679,
+      "loss": 5.7127,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2320675105485232,
+      "grad_norm": 2.4318594932556152,
+      "learning_rate": 0.0013213832783138873,
+      "loss": 5.7095,
+      "step": 2200
+    },
+    {
+      "epoch": 0.23312236286919832,
+      "grad_norm": 2.3913772106170654,
+      "learning_rate": 0.0013197538160074633,
+      "loss": 5.7063,
+      "step": 2210
+    },
+    {
+      "epoch": 0.23417721518987342,
+      "grad_norm": 2.480487108230591,
+      "learning_rate": 0.0013181179693313283,
+      "loss": 5.7088,
+      "step": 2220
+    },
+    {
+      "epoch": 0.23523206751054854,
+      "grad_norm": 2.599778413772583,
+      "learning_rate": 0.0013164757566159428,
+      "loss": 5.7103,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23628691983122363,
+      "grad_norm": 2.196176290512085,
+      "learning_rate": 0.001314827196263102,
+      "loss": 5.7073,
+      "step": 2240
+    },
+    {
+      "epoch": 0.23734177215189872,
+      "grad_norm": 2.633105516433716,
+      "learning_rate": 0.0013131723067457302,
+      "loss": 5.708,
+      "step": 2250
+    },
+    {
+      "epoch": 0.23839662447257384,
+      "grad_norm": 2.4101459980010986,
+      "learning_rate": 0.0013115111066076721,
+      "loss": 5.711,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23945147679324894,
+      "grad_norm": 2.4580399990081787,
+      "learning_rate": 0.0013098436144634862,
+      "loss": 5.7119,
+      "step": 2270
+    },
+    {
+      "epoch": 0.24050632911392406,
+      "grad_norm": 2.1704235076904297,
+      "learning_rate": 0.0013081698489982364,
+      "loss": 5.7165,
+      "step": 2280
+    },
+    {
+      "epoch": 0.24156118143459915,
+      "grad_norm": 2.4327399730682373,
+      "learning_rate": 0.001306489828967282,
+      "loss": 5.7098,
+      "step": 2290
+    },
+    {
+      "epoch": 0.24261603375527427,
+      "grad_norm": 2.679405689239502,
+      "learning_rate": 0.0013048035731960679,
+      "loss": 5.7107,
+      "step": 2300
+    },
+    {
+      "epoch": 0.24367088607594936,
+      "grad_norm": 2.6814334392547607,
+      "learning_rate": 0.0013031111005799133,
+      "loss": 5.7136,
+      "step": 2310
+    },
+    {
+      "epoch": 0.24472573839662448,
+      "grad_norm": 2.2119572162628174,
+      "learning_rate": 0.0013014124300838004,
+      "loss": 5.7139,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24578059071729957,
+      "grad_norm": 2.1637468338012695,
+      "learning_rate": 0.0012997075807421612,
+      "loss": 5.7134,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2468354430379747,
+      "grad_norm": 2.2173011302948,
+      "learning_rate": 0.0012979965716586653,
+      "loss": 5.7082,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2478902953586498,
+      "grad_norm": 2.530744791030884,
+      "learning_rate": 0.0012962794220060048,
+      "loss": 5.712,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2489451476793249,
+      "grad_norm": 2.4821112155914307,
+      "learning_rate": 0.0012945561510256801,
+      "loss": 5.709,
+      "step": 2360
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2973220348358154,
+      "learning_rate": 0.001292826778027784,
+      "loss": 5.718,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2510548523206751,
+      "grad_norm": 2.776616096496582,
+      "learning_rate": 0.0012910913223907856,
+      "loss": 5.7156,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2521097046413502,
+      "grad_norm": 2.373518466949463,
+      "learning_rate": 0.0012893498035613123,
+      "loss": 5.715,
+      "step": 2390
+    },
+    {
+      "epoch": 0.25316455696202533,
+      "grad_norm": 2.1677892208099365,
+      "learning_rate": 0.001287602241053933,
+      "loss": 5.7157,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2542194092827004,
+      "grad_norm": 2.3684685230255127,
+      "learning_rate": 0.0012858486544509392,
+      "loss": 5.7097,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2552742616033755,
+      "grad_norm": 2.761796712875366,
+      "learning_rate": 0.0012840890634021249,
+      "loss": 5.7161,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2563291139240506,
+      "grad_norm": 2.2269694805145264,
+      "learning_rate": 0.0012823234876245667,
+      "loss": 5.7171,
+      "step": 2430
+    },
+    {
+      "epoch": 0.25738396624472576,
+      "grad_norm": 2.4516351222991943,
+      "learning_rate": 0.0012805519469024035,
+      "loss": 5.7247,
+      "step": 2440
+    },
+    {
+      "epoch": 0.25843881856540085,
+      "grad_norm": 2.6550230979919434,
+      "learning_rate": 0.0012787744610866143,
+      "loss": 5.7149,
+      "step": 2450
+    },
+    {
+      "epoch": 0.25949367088607594,
+      "grad_norm": 2.5813541412353516,
+      "learning_rate": 0.0012769910500947954,
+      "loss": 5.7211,
+      "step": 2460
+    },
+    {
+      "epoch": 0.26054852320675104,
+      "grad_norm": 3.0462162494659424,
+      "learning_rate": 0.0012752017339109376,
+      "loss": 5.7133,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2616033755274262,
+      "grad_norm": 2.7124416828155518,
+      "learning_rate": 0.0012734065325852029,
+      "loss": 5.7209,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2626582278481013,
+      "grad_norm": 2.4686224460601807,
+      "learning_rate": 0.0012716054662336987,
+      "loss": 5.7211,
+      "step": 2490
+    },
+    {
+      "epoch": 0.26371308016877637,
+      "grad_norm": 2.448014259338379,
+      "learning_rate": 0.001269798555038252,
+      "loss": 5.7223,
+      "step": 2500
+    },
+    {
+      "epoch": 0.26476793248945146,
+      "grad_norm": 1.9004302024841309,
+      "learning_rate": 0.0012679858192461864,
+      "loss": 5.7189,
+      "step": 2510
+    },
+    {
+      "epoch": 0.26582278481012656,
+      "grad_norm": 2.402034282684326,
+      "learning_rate": 0.0012661672791700906,
+      "loss": 5.7183,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2668776371308017,
+      "grad_norm": 2.5602076053619385,
+      "learning_rate": 0.0012643429551875945,
+      "loss": 5.7186,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2679324894514768,
+      "grad_norm": 2.0443079471588135,
+      "learning_rate": 0.0012625128677411388,
+      "loss": 5.725,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2689873417721519,
+      "grad_norm": 2.751892328262329,
+      "learning_rate": 0.0012606770373377475,
+      "loss": 5.7134,
+      "step": 2550
+    },
+    {
+      "epoch": 0.270042194092827,
+      "grad_norm": 2.348557710647583,
+      "learning_rate": 0.0012588354845487959,
+      "loss": 5.7215,
+      "step": 2560
+    },
+    {
+      "epoch": 0.27109704641350213,
+      "grad_norm": 2.525733470916748,
+      "learning_rate": 0.001256988230009783,
+      "loss": 5.72,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2721518987341772,
+      "grad_norm": 2.503282308578491,
+      "learning_rate": 0.0012551352944200976,
+      "loss": 5.7247,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2732067510548523,
+      "grad_norm": 2.468935012817383,
+      "learning_rate": 0.0012532766985427874,
+      "loss": 5.7211,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2742616033755274,
+      "grad_norm": 2.711972713470459,
+      "learning_rate": 0.0012514124632043272,
+      "loss": 5.7204,
+      "step": 2600
+    },
+    {
+      "epoch": 0.27531645569620256,
+      "grad_norm": 2.3899881839752197,
+      "learning_rate": 0.0012495426092943842,
+      "loss": 5.731,
+      "step": 2610
+    },
+    {
+      "epoch": 0.27637130801687765,
+      "grad_norm": 2.239114284515381,
+      "learning_rate": 0.0012476671577655845,
+      "loss": 5.7256,
+      "step": 2620
+    },
+    {
+      "epoch": 0.27742616033755274,
+      "grad_norm": 2.5527844429016113,
+      "learning_rate": 0.0012457861296332774,
+      "loss": 5.7271,
+      "step": 2630
+    },
+    {
+      "epoch": 0.27848101265822783,
+      "grad_norm": 2.600001096725464,
+      "learning_rate": 0.001243899545975303,
+      "loss": 5.7272,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2795358649789029,
+      "grad_norm": 2.3590664863586426,
+      "learning_rate": 0.0012420074279317515,
+      "loss": 5.7182,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2805907172995781,
+      "grad_norm": 2.5398216247558594,
+      "learning_rate": 0.0012401097967047298,
+      "loss": 5.727,
+      "step": 2660
+    },
+    {
+      "epoch": 0.28164556962025317,
+      "grad_norm": 2.250138521194458,
+      "learning_rate": 0.001238206673558122,
+      "loss": 5.7252,
+      "step": 2670
+    },
+    {
+      "epoch": 0.28270042194092826,
+      "grad_norm": 2.6642074584960938,
+      "learning_rate": 0.0012362980798173526,
+      "loss": 5.732,
+      "step": 2680
+    },
+    {
+      "epoch": 0.28375527426160335,
+      "grad_norm": 2.536508321762085,
+      "learning_rate": 0.0012343840368691462,
+      "loss": 5.7259,
+      "step": 2690
+    },
+    {
+      "epoch": 0.2848101265822785,
+      "grad_norm": 2.2654597759246826,
+      "learning_rate": 0.0012324645661612886,
+      "loss": 5.7322,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2858649789029536,
+      "grad_norm": 2.4696927070617676,
+      "learning_rate": 0.0012305396892023867,
+      "loss": 5.7228,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2869198312236287,
+      "grad_norm": 2.7653791904449463,
+      "learning_rate": 0.0012286094275616264,
+      "loss": 5.7314,
+      "step": 2720
+    },
+    {
+      "epoch": 0.2879746835443038,
+      "grad_norm": 2.79807710647583,
+      "learning_rate": 0.0012266738028685318,
+      "loss": 5.7251,
+      "step": 2730
+    },
+    {
+      "epoch": 0.2890295358649789,
+      "grad_norm": 2.176254987716675,
+      "learning_rate": 0.001224732836812723,
+      "loss": 5.7321,
+      "step": 2740
+    },
+    {
+      "epoch": 0.290084388185654,
+      "grad_norm": 2.1090521812438965,
+      "learning_rate": 0.0012227865511436724,
+      "loss": 5.7297,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2911392405063291,
+      "grad_norm": 2.617866277694702,
+      "learning_rate": 0.001220834967670461,
+      "loss": 5.7243,
+      "step": 2760
+    },
+    {
+      "epoch": 0.2921940928270042,
+      "grad_norm": 2.679316997528076,
+      "learning_rate": 0.0012188781082615346,
+      "loss": 5.7263,
+      "step": 2770
+    },
+    {
+      "epoch": 0.29324894514767935,
+      "grad_norm": 2.6608738899230957,
+      "learning_rate": 0.0012169159948444588,
+      "loss": 5.7295,
+      "step": 2780
+    },
+    {
+      "epoch": 0.29430379746835444,
+      "grad_norm": 2.2081618309020996,
+      "learning_rate": 0.001214948649405672,
+      "loss": 5.7209,
+      "step": 2790
+    },
+    {
+      "epoch": 0.29535864978902954,
+      "grad_norm": 2.7524356842041016,
+      "learning_rate": 0.0012129760939902407,
+      "loss": 5.7338,
+      "step": 2800
+    },
+    {
+      "epoch": 0.29641350210970463,
+      "grad_norm": 2.572983741760254,
+      "learning_rate": 0.0012109983507016114,
+      "loss": 5.7272,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2974683544303797,
+      "grad_norm": 2.606372594833374,
+      "learning_rate": 0.0012090154417013636,
+      "loss": 5.7298,
+      "step": 2820
+    },
+    {
+      "epoch": 0.29852320675105487,
+      "grad_norm": 2.511270761489868,
+      "learning_rate": 0.0012070273892089605,
+      "loss": 5.725,
+      "step": 2830
+    },
+    {
+      "epoch": 0.29957805907172996,
+      "grad_norm": 2.6398088932037354,
+      "learning_rate": 0.0012050342155015012,
+      "loss": 5.7218,
+      "step": 2840
+    },
+    {
+      "epoch": 0.30063291139240506,
+      "grad_norm": 2.706399440765381,
+      "learning_rate": 0.0012030359429134707,
+      "loss": 5.7306,
+      "step": 2850
+    },
+    {
+      "epoch": 0.30168776371308015,
+      "grad_norm": 2.439608097076416,
+      "learning_rate": 0.0012010325938364883,
+      "loss": 5.7324,
+      "step": 2860
+    },
+    {
+      "epoch": 0.3027426160337553,
+      "grad_norm": 2.5942673683166504,
+      "learning_rate": 0.0011990241907190592,
+      "loss": 5.7323,
+      "step": 2870
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 2.8621578216552734,
+      "learning_rate": 0.001197010756066321,
+      "loss": 5.7288,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3048523206751055,
+      "grad_norm": 2.1785330772399902,
+      "learning_rate": 0.0011949923124397917,
+      "loss": 5.7284,
+      "step": 2890
+    },
+    {
+      "epoch": 0.3059071729957806,
+      "grad_norm": 2.7853574752807617,
+      "learning_rate": 0.001192968882457118,
+      "loss": 5.7314,
+      "step": 2900
+    },
+    {
+      "epoch": 0.3069620253164557,
+      "grad_norm": 2.8131096363067627,
+      "learning_rate": 0.001190940488791821,
+      "loss": 5.7347,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3080168776371308,
+      "grad_norm": 2.407277822494507,
+      "learning_rate": 0.0011889071541730419,
+      "loss": 5.7301,
+      "step": 2920
+    },
+    {
+      "epoch": 0.3090717299578059,
+      "grad_norm": 2.6165010929107666,
+      "learning_rate": 0.001186868901385288,
+      "loss": 5.7277,
+      "step": 2930
+    },
+    {
+      "epoch": 0.310126582278481,
+      "grad_norm": 2.5977909564971924,
+      "learning_rate": 0.001184825753268177,
+      "loss": 5.7374,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3111814345991561,
+      "grad_norm": 2.599579334259033,
+      "learning_rate": 0.0011827777327161814,
+      "loss": 5.7325,
+      "step": 2950
+    },
+    {
+      "epoch": 0.31223628691983124,
+      "grad_norm": 2.618044376373291,
+      "learning_rate": 0.0011807248626783714,
+      "loss": 5.733,
+      "step": 2960
+    },
+    {
+      "epoch": 0.31329113924050633,
+      "grad_norm": 2.6733744144439697,
+      "learning_rate": 0.0011786671661581584,
+      "loss": 5.7316,
+      "step": 2970
+    },
+    {
+      "epoch": 0.3143459915611814,
+      "grad_norm": 2.6955626010894775,
+      "learning_rate": 0.001176604666213036,
+      "loss": 5.7327,
+      "step": 2980
+    },
+    {
+      "epoch": 0.3154008438818565,
+      "grad_norm": 2.622819185256958,
+      "learning_rate": 0.0011745373859543236,
+      "loss": 5.738,
+      "step": 2990
+    },
+    {
+      "epoch": 0.31645569620253167,
+      "grad_norm": 2.91229248046875,
+      "learning_rate": 0.0011724653485469063,
+      "loss": 5.7307,
+      "step": 3000
+    },
+    {
+      "epoch": 0.31751054852320676,
+      "grad_norm": 2.5420167446136475,
+      "learning_rate": 0.0011703885772089743,
+      "loss": 5.7312,
+      "step": 3010
+    },
+    {
+      "epoch": 0.31856540084388185,
+      "grad_norm": 2.6964855194091797,
+      "learning_rate": 0.0011683070952117646,
+      "loss": 5.7275,
+      "step": 3020
+    },
+    {
+      "epoch": 0.31962025316455694,
+      "grad_norm": 2.360957384109497,
+      "learning_rate": 0.0011662209258792998,
+      "loss": 5.7266,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3206751054852321,
+      "grad_norm": 2.9354519844055176,
+      "learning_rate": 0.0011641300925881257,
+      "loss": 5.7359,
+      "step": 3040
+    },
+    {
+      "epoch": 0.3217299578059072,
+      "grad_norm": 2.387941360473633,
+      "learning_rate": 0.0011620346187670501,
+      "loss": 5.7315,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3227848101265823,
+      "grad_norm": 2.5555942058563232,
+      "learning_rate": 0.0011599345278968806,
+      "loss": 5.7387,
+      "step": 3060
+    },
+    {
+      "epoch": 0.32383966244725737,
+      "grad_norm": 2.3769285678863525,
+      "learning_rate": 0.0011578298435101604,
+      "loss": 5.7345,
+      "step": 3070
+    },
+    {
+      "epoch": 0.32489451476793246,
+      "grad_norm": 2.774075984954834,
+      "learning_rate": 0.0011557205891909062,
+      "loss": 5.7292,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3259493670886076,
+      "grad_norm": 2.6434528827667236,
+      "learning_rate": 0.0011536067885743423,
+      "loss": 5.7308,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3270042194092827,
+      "grad_norm": 2.461801052093506,
+      "learning_rate": 0.001151488465346637,
+      "loss": 5.73,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3280590717299578,
+      "grad_norm": 2.2122957706451416,
+      "learning_rate": 0.0011493656432446362,
+      "loss": 5.7345,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3291139240506329,
+      "grad_norm": 2.2615747451782227,
+      "learning_rate": 0.0011472383460555983,
+      "loss": 5.7298,
+      "step": 3120
+    },
+    {
+      "epoch": 0.33016877637130804,
+      "grad_norm": 2.366396427154541,
+      "learning_rate": 0.001145106597616927,
+      "loss": 5.7344,
+      "step": 3130
+    },
+    {
+      "epoch": 0.33122362869198313,
+      "grad_norm": 2.604254722595215,
+      "learning_rate": 0.001142970421815904,
+      "loss": 5.7265,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3322784810126582,
+      "grad_norm": 2.992449998855591,
+      "learning_rate": 0.0011408298425894226,
+      "loss": 5.7326,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 2.296891927719116,
+      "learning_rate": 0.0011386848839237186,
+      "loss": 5.7307,
+      "step": 3160
+    },
+    {
+      "epoch": 0.33438818565400846,
+      "grad_norm": 2.9058175086975098,
+      "learning_rate": 0.0011365355698541005,
+      "loss": 5.7321,
+      "step": 3170
+    },
+    {
+      "epoch": 0.33544303797468356,
+      "grad_norm": 2.526737689971924,
+      "learning_rate": 0.0011343819244646824,
+      "loss": 5.734,
+      "step": 3180
+    },
+    {
+      "epoch": 0.33649789029535865,
+      "grad_norm": 2.7439675331115723,
+      "learning_rate": 0.001132223971888112,
+      "loss": 5.7334,
+      "step": 3190
+    },
+    {
+      "epoch": 0.33755274261603374,
+      "grad_norm": 2.9011874198913574,
+      "learning_rate": 0.0011300617363053024,
+      "loss": 5.7309,
+      "step": 3200
+    },
+    {
+      "epoch": 0.33860759493670883,
+      "grad_norm": 2.470439910888672,
+      "learning_rate": 0.0011278952419451586,
+      "loss": 5.7302,
+      "step": 3210
+    },
+    {
+      "epoch": 0.339662447257384,
+      "grad_norm": 2.645085573196411,
+      "learning_rate": 0.0011257245130843077,
+      "loss": 5.7297,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3407172995780591,
+      "grad_norm": 2.770294666290283,
+      "learning_rate": 0.0011235495740468265,
+      "loss": 5.7307,
+      "step": 3230
+    },
+    {
+      "epoch": 0.34177215189873417,
+      "grad_norm": 2.611384153366089,
+      "learning_rate": 0.0011213704492039694,
+      "loss": 5.7315,
+      "step": 3240
+    },
+    {
+      "epoch": 0.34282700421940926,
+      "grad_norm": 2.8464887142181396,
+      "learning_rate": 0.001119187162973894,
+      "loss": 5.7274,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3438818565400844,
+      "grad_norm": 2.6452078819274902,
+      "learning_rate": 0.001116999739821388,
+      "loss": 5.7313,
+      "step": 3260
+    },
+    {
+      "epoch": 0.3449367088607595,
+      "grad_norm": 3.054241418838501,
+      "learning_rate": 0.0011148082042575968,
+      "loss": 5.7302,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3459915611814346,
+      "grad_norm": 2.6827640533447266,
+      "learning_rate": 0.0011126125808397461,
+      "loss": 5.7345,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3470464135021097,
+      "grad_norm": 2.6487178802490234,
+      "learning_rate": 0.0011104128941708683,
+      "loss": 5.7319,
+      "step": 3290
+    },
+    {
+      "epoch": 0.34810126582278483,
+      "grad_norm": 2.4665026664733887,
+      "learning_rate": 0.001108209168899527,
+      "loss": 5.7294,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3491561181434599,
+      "grad_norm": 2.4238789081573486,
+      "learning_rate": 0.0011060014297195396,
+      "loss": 5.7316,
+      "step": 3310
+    },
+    {
+      "epoch": 0.350210970464135,
+      "grad_norm": 2.3739712238311768,
+      "learning_rate": 0.0011037897013697015,
+      "loss": 5.7326,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3512658227848101,
+      "grad_norm": 2.887742042541504,
+      "learning_rate": 0.0011015740086335092,
+      "loss": 5.729,
+      "step": 3330
+    },
+    {
+      "epoch": 0.35232067510548526,
+      "grad_norm": 2.457221508026123,
+      "learning_rate": 0.0010993543763388814,
+      "loss": 5.7304,
+      "step": 3340
+    },
+    {
+      "epoch": 0.35337552742616035,
+      "grad_norm": 2.2006115913391113,
+      "learning_rate": 0.0010971308293578814,
+      "loss": 5.7312,
+      "step": 3350
+    },
+    {
+      "epoch": 0.35443037974683544,
+      "grad_norm": 2.554802894592285,
+      "learning_rate": 0.0010949033926064397,
+      "loss": 5.7238,
+      "step": 3360
+    },
+    {
+      "epoch": 0.35548523206751054,
+      "grad_norm": 2.5150046348571777,
+      "learning_rate": 0.0010926720910440725,
+      "loss": 5.7279,
+      "step": 3370
+    },
+    {
+      "epoch": 0.35654008438818563,
+      "grad_norm": 2.565244197845459,
+      "learning_rate": 0.001090436949673603,
+      "loss": 5.7355,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3575949367088608,
+      "grad_norm": 2.663778066635132,
+      "learning_rate": 0.0010881979935408815,
+      "loss": 5.7336,
+      "step": 3390
+    },
+    {
+      "epoch": 0.35864978902953587,
+      "grad_norm": 2.707453727722168,
+      "learning_rate": 0.0010859552477345052,
+      "loss": 5.7313,
+      "step": 3400
+    },
+    {
+      "epoch": 0.35970464135021096,
+      "grad_norm": 2.572838544845581,
+      "learning_rate": 0.001083708737385536,
+      "loss": 5.7251,
+      "step": 3410
+    },
+    {
+      "epoch": 0.36075949367088606,
+      "grad_norm": 2.596358060836792,
+      "learning_rate": 0.0010814584876672187,
+      "loss": 5.7311,
+      "step": 3420
+    },
+    {
+      "epoch": 0.3618143459915612,
+      "grad_norm": 2.280561685562134,
+      "learning_rate": 0.0010792045237947008,
+      "loss": 5.7368,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3628691983122363,
+      "grad_norm": 2.5461955070495605,
+      "learning_rate": 0.0010769468710247478,
+      "loss": 5.733,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3639240506329114,
+      "grad_norm": 3.109029769897461,
+      "learning_rate": 0.0010746855546554612,
+      "loss": 5.7361,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3649789029535865,
+      "grad_norm": 2.9041495323181152,
+      "learning_rate": 0.0010724206000259954,
+      "loss": 5.7304,
+      "step": 3460
+    },
+    {
+      "epoch": 0.36603375527426163,
+      "grad_norm": 2.892158269882202,
+      "learning_rate": 0.0010701520325162727,
+      "loss": 5.7277,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3670886075949367,
+      "grad_norm": 2.817375421524048,
+      "learning_rate": 0.0010678798775467001,
+      "loss": 5.7255,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3681434599156118,
+      "grad_norm": 2.5929696559906006,
+      "learning_rate": 0.0010656041605778832,
+      "loss": 5.7287,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3691983122362869,
+      "grad_norm": 2.7741239070892334,
+      "learning_rate": 0.001063324907110342,
+      "loss": 5.7288,
+      "step": 3500
+    },
+    {
+      "epoch": 0.370253164556962,
+      "grad_norm": 2.793445587158203,
+      "learning_rate": 0.0010610421426842241,
+      "loss": 5.7331,
+      "step": 3510
+    },
+    {
+      "epoch": 0.37130801687763715,
+      "grad_norm": 2.382033586502075,
+      "learning_rate": 0.00105875589287902,
+      "loss": 5.7284,
+      "step": 3520
+    },
+    {
+      "epoch": 0.37236286919831224,
+      "grad_norm": 2.5853967666625977,
+      "learning_rate": 0.0010564661833132752,
+      "loss": 5.7368,
+      "step": 3530
+    },
+    {
+      "epoch": 0.37341772151898733,
+      "grad_norm": 2.3739659786224365,
+      "learning_rate": 0.001054173039644303,
+      "loss": 5.7416,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3744725738396624,
+      "grad_norm": 2.778949499130249,
+      "learning_rate": 0.0010518764875678981,
+      "loss": 5.73,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3755274261603376,
+      "grad_norm": 2.2127442359924316,
+      "learning_rate": 0.001049576552818048,
+      "loss": 5.7315,
+      "step": 3560
+    },
+    {
+      "epoch": 0.37658227848101267,
+      "grad_norm": 3.0314297676086426,
+      "learning_rate": 0.0010472732611666448,
+      "loss": 5.731,
+      "step": 3570
+    },
+    {
+      "epoch": 0.37763713080168776,
+      "grad_norm": 2.6465890407562256,
+      "learning_rate": 0.0010449666384231954,
+      "loss": 5.7328,
+      "step": 3580
+    },
+    {
+      "epoch": 0.37869198312236285,
+      "grad_norm": 2.7873363494873047,
+      "learning_rate": 0.0010426567104345346,
+      "loss": 5.7334,
+      "step": 3590
+    },
+    {
+      "epoch": 0.379746835443038,
+      "grad_norm": 2.8117613792419434,
+      "learning_rate": 0.0010403435030845332,
+      "loss": 5.7258,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3808016877637131,
+      "grad_norm": 2.6950972080230713,
+      "learning_rate": 0.0010380270422938093,
+      "loss": 5.7329,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3818565400843882,
+      "grad_norm": 2.472207546234131,
+      "learning_rate": 0.0010357073540194362,
+      "loss": 5.7274,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3829113924050633,
+      "grad_norm": 2.8980681896209717,
+      "learning_rate": 0.001033384464254655,
+      "loss": 5.7308,
+      "step": 3630
+    },
+    {
+      "epoch": 0.38396624472573837,
+      "grad_norm": 3.022688388824463,
+      "learning_rate": 0.001031058399028579,
+      "loss": 5.7296,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3850210970464135,
+      "grad_norm": 2.644301414489746,
+      "learning_rate": 0.001028729184405905,
+      "loss": 5.7405,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3860759493670886,
+      "grad_norm": 2.6901066303253174,
+      "learning_rate": 0.0010263968464866201,
+      "loss": 5.7293,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3871308016877637,
+      "grad_norm": 2.7620391845703125,
+      "learning_rate": 0.0010240614114057098,
+      "loss": 5.7233,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3881856540084388,
+      "grad_norm": 2.5243899822235107,
+      "learning_rate": 0.001021722905332864,
+      "loss": 5.7266,
+      "step": 3680
+    },
+    {
+      "epoch": 0.38924050632911394,
+      "grad_norm": 2.9015092849731445,
+      "learning_rate": 0.0010193813544721855,
+      "loss": 5.7333,
+      "step": 3690
+    },
+    {
+      "epoch": 0.39029535864978904,
+      "grad_norm": 2.647833824157715,
+      "learning_rate": 0.001017036785061895,
+      "loss": 5.7396,
+      "step": 3700
+    },
+    {
+      "epoch": 0.39135021097046413,
+      "grad_norm": 2.4830665588378906,
+      "learning_rate": 0.0010146892233740376,
+      "loss": 5.7264,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3924050632911392,
+      "grad_norm": 2.455185890197754,
+      "learning_rate": 0.0010123386957141883,
+      "loss": 5.7252,
+      "step": 3720
+    },
+    {
+      "epoch": 0.39345991561181437,
+      "grad_norm": 3.0491244792938232,
+      "learning_rate": 0.0010099852284211573,
+      "loss": 5.7311,
+      "step": 3730
+    },
+    {
+      "epoch": 0.39451476793248946,
+      "grad_norm": 2.4948348999023438,
+      "learning_rate": 0.0010076288478666944,
+      "loss": 5.7236,
+      "step": 3740
+    },
+    {
+      "epoch": 0.39556962025316456,
+      "grad_norm": 2.604893684387207,
+      "learning_rate": 0.0010052695804551946,
+      "loss": 5.7289,
+      "step": 3750
+    },
+    {
+      "epoch": 0.39662447257383965,
+      "grad_norm": 2.6216881275177,
+      "learning_rate": 0.0010029074526234014,
+      "loss": 5.7252,
+      "step": 3760
+    },
+    {
+      "epoch": 0.39767932489451474,
+      "grad_norm": 3.281416416168213,
+      "learning_rate": 0.0010005424908401104,
+      "loss": 5.7338,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3987341772151899,
+      "grad_norm": 2.9160819053649902,
+      "learning_rate": 0.0009981747216058728,
+      "loss": 5.7218,
+      "step": 3780
+    },
+    {
+      "epoch": 0.399789029535865,
+      "grad_norm": 2.603567123413086,
+      "learning_rate": 0.0009958041714526998,
+      "loss": 5.7301,
+      "step": 3790
+    },
+    {
+      "epoch": 0.4008438818565401,
+      "grad_norm": 2.4944071769714355,
+      "learning_rate": 0.0009934308669437627,
+      "loss": 5.729,
+      "step": 3800
+    },
+    {
+      "epoch": 0.40189873417721517,
+      "grad_norm": 2.4411263465881348,
+      "learning_rate": 0.0009910548346730972,
+      "loss": 5.7294,
+      "step": 3810
+    },
+    {
+      "epoch": 0.4029535864978903,
+      "grad_norm": 2.748687267303467,
+      "learning_rate": 0.0009886761012653062,
+      "loss": 5.7299,
+      "step": 3820
+    },
+    {
+      "epoch": 0.4040084388185654,
+      "grad_norm": 2.4843857288360596,
+      "learning_rate": 0.000986294693375258,
+      "loss": 5.7329,
+      "step": 3830
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 2.733609914779663,
+      "learning_rate": 0.000983910637687791,
+      "loss": 5.7306,
+      "step": 3840
+    },
+    {
+      "epoch": 0.4061181434599156,
+      "grad_norm": 2.4320693016052246,
+      "learning_rate": 0.0009815239609174138,
+      "loss": 5.7303,
+      "step": 3850
+    },
+    {
+      "epoch": 0.40717299578059074,
+      "grad_norm": 2.9169809818267822,
+      "learning_rate": 0.0009791346898080043,
+      "loss": 5.7289,
+      "step": 3860
+    },
+    {
+      "epoch": 0.40822784810126583,
+      "grad_norm": 2.6987764835357666,
+      "learning_rate": 0.0009767428511325122,
+      "loss": 5.7284,
+      "step": 3870
+    },
+    {
+      "epoch": 0.4092827004219409,
+      "grad_norm": 2.663710117340088,
+      "learning_rate": 0.0009743484716926576,
+      "loss": 5.7287,
+      "step": 3880
+    },
+    {
+      "epoch": 0.410337552742616,
+      "grad_norm": 2.7164735794067383,
+      "learning_rate": 0.0009719515783186319,
+      "loss": 5.7251,
+      "step": 3890
+    },
+    {
+      "epoch": 0.41139240506329117,
+      "grad_norm": 2.8201546669006348,
+      "learning_rate": 0.0009695521978687951,
+      "loss": 5.725,
+      "step": 3900
+    },
+    {
+      "epoch": 0.41244725738396626,
+      "grad_norm": 2.3503456115722656,
+      "learning_rate": 0.0009671503572293767,
+      "loss": 5.729,
+      "step": 3910
+    },
+    {
+      "epoch": 0.41350210970464135,
+      "grad_norm": 2.5236856937408447,
+      "learning_rate": 0.0009647460833141742,
+      "loss": 5.7321,
+      "step": 3920
+    },
+    {
+      "epoch": 0.41455696202531644,
+      "grad_norm": 2.569225549697876,
+      "learning_rate": 0.0009623394030642507,
+      "loss": 5.7339,
+      "step": 3930
+    },
+    {
+      "epoch": 0.41561181434599154,
+      "grad_norm": 2.449202537536621,
+      "learning_rate": 0.0009599303434476334,
+      "loss": 5.7263,
+      "step": 3940
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 2.556098222732544,
+      "learning_rate": 0.0009575189314590118,
+      "loss": 5.7261,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4177215189873418,
+      "grad_norm": 2.3679325580596924,
+      "learning_rate": 0.0009551051941194346,
+      "loss": 5.7295,
+      "step": 3960
+    },
+    {
+      "epoch": 0.41877637130801687,
+      "grad_norm": 2.605989456176758,
+      "learning_rate": 0.0009526891584760071,
+      "loss": 5.7293,
+      "step": 3970
+    },
+    {
+      "epoch": 0.41983122362869196,
+      "grad_norm": 2.5102407932281494,
+      "learning_rate": 0.0009502708516015889,
+      "loss": 5.7222,
+      "step": 3980
+    },
+    {
+      "epoch": 0.4208860759493671,
+      "grad_norm": 2.7388317584991455,
+      "learning_rate": 0.0009478503005944888,
+      "loss": 5.7334,
+      "step": 3990
+    },
+    {
+      "epoch": 0.4219409282700422,
+      "grad_norm": 3.1526310443878174,
+      "learning_rate": 0.0009454275325781632,
+      "loss": 5.7269,
+      "step": 4000
+    },
+    {
+      "epoch": 0.4229957805907173,
+      "grad_norm": 2.5010690689086914,
+      "learning_rate": 0.0009430025747009104,
+      "loss": 5.7262,
+      "step": 4010
+    },
+    {
+      "epoch": 0.4240506329113924,
+      "grad_norm": 2.8775594234466553,
+      "learning_rate": 0.0009405754541355677,
+      "loss": 5.7294,
+      "step": 4020
+    },
+    {
+      "epoch": 0.42510548523206754,
+      "grad_norm": 2.7482597827911377,
+      "learning_rate": 0.0009383892190063256,
+      "loss": 5.7237,
+      "step": 4030
+    },
+    {
+      "epoch": 0.42616033755274263,
+      "grad_norm": 2.6342711448669434,
+      "learning_rate": 0.0009359580642813265,
+      "loss": 5.7261,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4272151898734177,
+      "grad_norm": 2.7106611728668213,
+      "learning_rate": 0.0009335248258054162,
+      "loss": 5.731,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4282700421940928,
+      "grad_norm": 3.0685009956359863,
+      "learning_rate": 0.0009310895308442202,
+      "loss": 5.7311,
+      "step": 4060
+    },
+    {
+      "epoch": 0.4293248945147679,
+      "grad_norm": 2.5310869216918945,
+      "learning_rate": 0.0009286522066864078,
+      "loss": 5.7315,
+      "step": 4070
+    },
+    {
+      "epoch": 0.43037974683544306,
+      "grad_norm": 2.6125290393829346,
+      "learning_rate": 0.0009262128806433858,
+      "loss": 5.7247,
+      "step": 4080
+    },
+    {
+      "epoch": 0.43143459915611815,
+      "grad_norm": 2.966644525527954,
+      "learning_rate": 0.0009237715800489942,
+      "loss": 5.7228,
+      "step": 4090
+    },
+    {
+      "epoch": 0.43248945147679324,
+      "grad_norm": 3.066164970397949,
+      "learning_rate": 0.0009213283322591977,
+      "loss": 5.725,
+      "step": 4100
+    },
+    {
+      "epoch": 0.43354430379746833,
+      "grad_norm": 2.3687124252319336,
+      "learning_rate": 0.000918883164651781,
+      "loss": 5.7269,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4345991561181435,
+      "grad_norm": 2.454720973968506,
+      "learning_rate": 0.0009164361046260412,
+      "loss": 5.7307,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4356540084388186,
+      "grad_norm": 2.132103443145752,
+      "learning_rate": 0.0009139871796024807,
+      "loss": 5.7301,
+      "step": 4130
+    },
+    {
+      "epoch": 0.43670886075949367,
+      "grad_norm": 2.5743460655212402,
+      "learning_rate": 0.0009115364170225,
+      "loss": 5.7252,
+      "step": 4140
+    },
+    {
+      "epoch": 0.43776371308016876,
+      "grad_norm": 2.2709312438964844,
+      "learning_rate": 0.0009090838443480903,
+      "loss": 5.7297,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4388185654008439,
+      "grad_norm": 2.903399705886841,
+      "learning_rate": 0.0009066294890615266,
+      "loss": 5.7249,
+      "step": 4160
+    },
+    {
+      "epoch": 0.439873417721519,
+      "grad_norm": 2.831774950027466,
+      "learning_rate": 0.0009041733786650578,
+      "loss": 5.725,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4409282700421941,
+      "grad_norm": 2.885425567626953,
+      "learning_rate": 0.0009017155406806006,
+      "loss": 5.7199,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4419831223628692,
+      "grad_norm": 2.4683425426483154,
+      "learning_rate": 0.0008992560026494294,
+      "loss": 5.7296,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4430379746835443,
+      "grad_norm": 2.3768904209136963,
+      "learning_rate": 0.0008967947921318689,
+      "loss": 5.7272,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4440928270042194,
+      "grad_norm": 2.485675811767578,
+      "learning_rate": 0.0008943319367069844,
+      "loss": 5.7229,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4451476793248945,
+      "grad_norm": 2.567415237426758,
+      "learning_rate": 0.0008918674639722742,
+      "loss": 5.7156,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4462025316455696,
+      "grad_norm": 2.824456214904785,
+      "learning_rate": 0.0008894014015433582,
+      "loss": 5.7145,
+      "step": 4230
+    },
+    {
+      "epoch": 0.4472573839662447,
+      "grad_norm": 2.421360492706299,
+      "learning_rate": 0.0008869337770536699,
+      "loss": 5.7252,
+      "step": 4240
+    },
+    {
+      "epoch": 0.44831223628691985,
+      "grad_norm": 3.560734272003174,
+      "learning_rate": 0.0008844646181541472,
+      "loss": 5.7299,
+      "step": 4250
+    },
+    {
+      "epoch": 0.44936708860759494,
+      "grad_norm": 2.550701141357422,
+      "learning_rate": 0.0008819939525129207,
+      "loss": 5.7283,
+      "step": 4260
+    },
+    {
+      "epoch": 0.45042194092827004,
+      "grad_norm": 2.341249465942383,
+      "learning_rate": 0.0008795218078150056,
+      "loss": 5.7259,
+      "step": 4270
+    },
+    {
+      "epoch": 0.45147679324894513,
+      "grad_norm": 2.68642520904541,
+      "learning_rate": 0.0008770482117619901,
+      "loss": 5.7199,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4525316455696203,
+      "grad_norm": 2.507251024246216,
+      "learning_rate": 0.000874573192071726,
+      "loss": 5.7248,
+      "step": 4290
+    },
+    {
+      "epoch": 0.45358649789029537,
+      "grad_norm": 2.941439390182495,
+      "learning_rate": 0.0008720967764780173,
+      "loss": 5.7247,
+      "step": 4300
+    },
+    {
+      "epoch": 0.45464135021097046,
+      "grad_norm": 2.3936643600463867,
+      "learning_rate": 0.0008696189927303101,
+      "loss": 5.7259,
+      "step": 4310
+    },
+    {
+      "epoch": 0.45569620253164556,
+      "grad_norm": 2.301410436630249,
+      "learning_rate": 0.0008671398685933811,
+      "loss": 5.7191,
+      "step": 4320
+    },
+    {
+      "epoch": 0.45675105485232065,
+      "grad_norm": 2.44746470451355,
+      "learning_rate": 0.0008646594318470268,
+      "loss": 5.7232,
+      "step": 4330
+    },
+    {
+      "epoch": 0.4578059071729958,
+      "grad_norm": 2.7832767963409424,
+      "learning_rate": 0.000862177710285752,
+      "loss": 5.7246,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4588607594936709,
+      "grad_norm": 2.416851043701172,
+      "learning_rate": 0.0008596947317184585,
+      "loss": 5.729,
+      "step": 4350
+    },
+    {
+      "epoch": 0.459915611814346,
+      "grad_norm": 3.110567569732666,
+      "learning_rate": 0.0008572105239681338,
+      "loss": 5.7247,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4609704641350211,
+      "grad_norm": 2.3490309715270996,
+      "learning_rate": 0.0008547251148715386,
+      "loss": 5.7215,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4620253164556962,
+      "grad_norm": 2.7428228855133057,
+      "learning_rate": 0.0008522385322788955,
+      "loss": 5.726,
+      "step": 4380
+    },
+    {
+      "epoch": 0.4630801687763713,
+      "grad_norm": 2.282529830932617,
+      "learning_rate": 0.0008497508040535766,
+      "loss": 5.7238,
+      "step": 4390
+    },
+    {
+      "epoch": 0.4641350210970464,
+      "grad_norm": 2.7025232315063477,
+      "learning_rate": 0.0008472619580717914,
+      "loss": 5.7262,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4651898734177215,
+      "grad_norm": 2.89739990234375,
+      "learning_rate": 0.000844772022222274,
+      "loss": 5.7275,
+      "step": 4410
+    },
+    {
+      "epoch": 0.46624472573839665,
+      "grad_norm": 2.610060930252075,
+      "learning_rate": 0.0008422810244059721,
+      "loss": 5.7183,
+      "step": 4420
+    },
+    {
+      "epoch": 0.46729957805907174,
+      "grad_norm": 2.375096321105957,
+      "learning_rate": 0.000839788992535732,
+      "loss": 5.7259,
+      "step": 4430
+    },
+    {
+      "epoch": 0.46835443037974683,
+      "grad_norm": 2.541840076446533,
+      "learning_rate": 0.000837295954535988,
+      "loss": 5.7242,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4694092827004219,
+      "grad_norm": 2.75589656829834,
+      "learning_rate": 0.0008348019383424479,
+      "loss": 5.7232,
+      "step": 4450
+    },
+    {
+      "epoch": 0.4704641350210971,
+      "grad_norm": 2.6704862117767334,
+      "learning_rate": 0.0008323069719017812,
+      "loss": 5.7259,
+      "step": 4460
+    },
+    {
+      "epoch": 0.47151898734177217,
+      "grad_norm": 2.255814552307129,
+      "learning_rate": 0.0008298110831713047,
+      "loss": 5.7229,
+      "step": 4470
+    },
+    {
+      "epoch": 0.47257383966244726,
+      "grad_norm": 2.8616321086883545,
+      "learning_rate": 0.0008273143001186709,
+      "loss": 5.7238,
+      "step": 4480
+    },
+    {
+      "epoch": 0.47362869198312235,
+      "grad_norm": 2.629572629928589,
+      "learning_rate": 0.0008248166507215526,
+      "loss": 5.7247,
+      "step": 4490
+    },
+    {
+      "epoch": 0.47468354430379744,
+      "grad_norm": 3.525463342666626,
+      "learning_rate": 0.0008223181629673312,
+      "loss": 5.7217,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4757383966244726,
+      "grad_norm": 2.651808977127075,
+      "learning_rate": 0.0008198188648527818,
+      "loss": 5.7172,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4767932489451477,
+      "grad_norm": 2.7758734226226807,
+      "learning_rate": 0.00081731878438376,
+      "loss": 5.716,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4778481012658228,
+      "grad_norm": 2.1174957752227783,
+      "learning_rate": 0.0008148179495748885,
+      "loss": 5.719,
+      "step": 4530
+    },
+    {
+      "epoch": 0.47890295358649787,
+      "grad_norm": 2.2898027896881104,
+      "learning_rate": 0.0008123163884492422,
+      "loss": 5.7185,
+      "step": 4540
+    },
+    {
+      "epoch": 0.479957805907173,
+      "grad_norm": 2.236358642578125,
+      "learning_rate": 0.0008098141290380353,
+      "loss": 5.7186,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4810126582278481,
+      "grad_norm": 2.363375663757324,
+      "learning_rate": 0.000807311199380306,
+      "loss": 5.7175,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4820675105485232,
+      "grad_norm": 2.465202569961548,
+      "learning_rate": 0.0008048076275226032,
+      "loss": 5.7151,
+      "step": 4570
+    },
+    {
+      "epoch": 0.4831223628691983,
+      "grad_norm": 2.6928303241729736,
+      "learning_rate": 0.0008023034415186725,
+      "loss": 5.7179,
+      "step": 4580
+    },
+    {
+      "epoch": 0.48417721518987344,
+      "grad_norm": 2.1407241821289062,
+      "learning_rate": 0.0007997986694291404,
+      "loss": 5.7126,
+      "step": 4590
+    },
+    {
+      "epoch": 0.48523206751054854,
+      "grad_norm": 2.427114486694336,
+      "learning_rate": 0.0007972933393212012,
+      "loss": 5.717,
+      "step": 4600
+    },
+    {
+      "epoch": 0.48628691983122363,
+      "grad_norm": 2.1859185695648193,
+      "learning_rate": 0.0007947874792683025,
+      "loss": 5.7149,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4873417721518987,
+      "grad_norm": 2.5300941467285156,
+      "learning_rate": 0.0007922811173498293,
+      "loss": 5.7148,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4883966244725738,
+      "grad_norm": 2.3213307857513428,
+      "learning_rate": 0.000789774281650791,
+      "loss": 5.7194,
+      "step": 4630
+    },
+    {
+      "epoch": 0.48945147679324896,
+      "grad_norm": 2.3530449867248535,
+      "learning_rate": 0.0007872670002615056,
+      "loss": 5.712,
+      "step": 4640
+    },
+    {
+      "epoch": 0.49050632911392406,
+      "grad_norm": 2.12992787361145,
+      "learning_rate": 0.0007847593012772852,
+      "loss": 5.7161,
+      "step": 4650
+    },
+    {
+      "epoch": 0.49156118143459915,
+      "grad_norm": 2.6019845008850098,
+      "learning_rate": 0.0007822512127981218,
+      "loss": 5.7094,
+      "step": 4660
+    },
+    {
+      "epoch": 0.49261603375527424,
+      "grad_norm": 2.7171554565429688,
+      "learning_rate": 0.0007797427629283708,
+      "loss": 5.7152,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4936708860759494,
+      "grad_norm": 2.741291046142578,
+      "learning_rate": 0.0007772339797764385,
+      "loss": 5.7133,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4947257383966245,
+      "grad_norm": 2.5194973945617676,
+      "learning_rate": 0.0007747248914544646,
+      "loss": 5.7113,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4957805907172996,
+      "grad_norm": 2.68666934967041,
+      "learning_rate": 0.0007722155260780093,
+      "loss": 5.7126,
+      "step": 4700
+    },
+    {
+      "epoch": 0.49683544303797467,
+      "grad_norm": 2.5688393115997314,
+      "learning_rate": 0.0007697059117657368,
+      "loss": 5.7148,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4978902953586498,
+      "grad_norm": 3.0395710468292236,
+      "learning_rate": 0.0007671960766391008,
+      "loss": 5.7122,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4989451476793249,
+      "grad_norm": 2.4003500938415527,
+      "learning_rate": 0.0007646860488220293,
+      "loss": 5.7155,
+      "step": 4730
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4114458560943604,
+      "learning_rate": 0.00076217585644061,
+      "loss": 5.7142,
+      "step": 4740
+    },
+    {
+      "epoch": 0.5010548523206751,
+      "grad_norm": 2.3023202419281006,
+      "learning_rate": 0.0007596655276227739,
+      "loss": 5.7149,
+      "step": 4750
+    },
+    {
+      "epoch": 0.5021097046413502,
+      "grad_norm": 2.55741286277771,
+      "learning_rate": 0.0007571550904979812,
+      "loss": 5.7163,
+      "step": 4760
+    },
+    {
+      "epoch": 0.5031645569620253,
+      "grad_norm": 1.8762019872665405,
+      "learning_rate": 0.0007546445731969056,
+      "loss": 5.7082,
+      "step": 4770
+    },
+    {
+      "epoch": 0.5042194092827004,
+      "grad_norm": 1.6707096099853516,
+      "learning_rate": 0.0007521340038511196,
+      "loss": 5.7114,
+      "step": 4780
+    },
+    {
+      "epoch": 0.5052742616033755,
+      "grad_norm": 2.6713571548461914,
+      "learning_rate": 0.0007496234105927785,
+      "loss": 5.7145,
+      "step": 4790
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 2.6164796352386475,
+      "learning_rate": 0.0007471128215543056,
+      "loss": 5.7124,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5073839662447257,
+      "grad_norm": 2.3578579425811768,
+      "learning_rate": 0.0007446022648680768,
+      "loss": 5.7114,
+      "step": 4810
+    },
+    {
+      "epoch": 0.5084388185654009,
+      "grad_norm": 2.6928939819335938,
+      "learning_rate": 0.0007420917686661055,
+      "loss": 5.7214,
+      "step": 4820
+    },
+    {
+      "epoch": 0.509493670886076,
+      "grad_norm": 2.507598400115967,
+      "learning_rate": 0.0007395813610797283,
+      "loss": 5.7095,
+      "step": 4830
+    },
+    {
+      "epoch": 0.510548523206751,
+      "grad_norm": 2.504307746887207,
+      "learning_rate": 0.0007370710702392873,
+      "loss": 5.7092,
+      "step": 4840
+    },
+    {
+      "epoch": 0.5116033755274262,
+      "grad_norm": 2.7466044425964355,
+      "learning_rate": 0.0007345609242738173,
+      "loss": 5.7076,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5126582278481012,
+      "grad_norm": 2.470513105392456,
+      "learning_rate": 0.0007320509513107296,
+      "loss": 5.7038,
+      "step": 4860
+    },
+    {
+      "epoch": 0.5137130801687764,
+      "grad_norm": 2.9400811195373535,
+      "learning_rate": 0.0007295411794754967,
+      "loss": 5.7062,
+      "step": 4870
+    },
+    {
+      "epoch": 0.5147679324894515,
+      "grad_norm": 2.9847309589385986,
+      "learning_rate": 0.0007270316368913374,
+      "loss": 5.7108,
+      "step": 4880
+    },
+    {
+      "epoch": 0.5158227848101266,
+      "grad_norm": 2.5324513912200928,
+      "learning_rate": 0.0007245223516789019,
+      "loss": 5.7177,
+      "step": 4890
+    },
+    {
+      "epoch": 0.5168776371308017,
+      "grad_norm": 2.7009174823760986,
+      "learning_rate": 0.0007220133519559563,
+      "loss": 5.7116,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5179324894514767,
+      "grad_norm": 2.313591241836548,
+      "learning_rate": 0.0007195046658370675,
+      "loss": 5.7126,
+      "step": 4910
+    },
+    {
+      "epoch": 0.5189873417721519,
+      "grad_norm": 2.1955068111419678,
+      "learning_rate": 0.0007169963214332885,
+      "loss": 5.712,
+      "step": 4920
+    },
+    {
+      "epoch": 0.520042194092827,
+      "grad_norm": 2.478435516357422,
+      "learning_rate": 0.000714488346851843,
+      "loss": 5.7079,
+      "step": 4930
+    },
+    {
+      "epoch": 0.5210970464135021,
+      "grad_norm": 2.0320398807525635,
+      "learning_rate": 0.000711980770195811,
+      "loss": 5.7053,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5221518987341772,
+      "grad_norm": 2.3653197288513184,
+      "learning_rate": 0.0007094736195638128,
+      "loss": 5.7078,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5232067510548524,
+      "grad_norm": 2.9816436767578125,
+      "learning_rate": 0.0007069669230496961,
+      "loss": 5.7022,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5242616033755274,
+      "grad_norm": 2.514852285385132,
+      "learning_rate": 0.0007044607087422191,
+      "loss": 5.7096,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5253164556962026,
+      "grad_norm": 2.582826614379883,
+      "learning_rate": 0.000701955004724737,
+      "loss": 5.704,
+      "step": 4980
+    },
+    {
+      "epoch": 0.5263713080168776,
+      "grad_norm": 2.1142373085021973,
+      "learning_rate": 0.0006994498390748865,
+      "loss": 5.705,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5274261603375527,
+      "grad_norm": 2.428255319595337,
+      "learning_rate": 0.0006969452398642721,
+      "loss": 5.7068,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5284810126582279,
+      "grad_norm": 2.6294376850128174,
+      "learning_rate": 0.0006944412351581506,
+      "loss": 5.7049,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5295358649789029,
+      "grad_norm": 2.3299965858459473,
+      "learning_rate": 0.0006919378530151182,
+      "loss": 5.7061,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5305907172995781,
+      "grad_norm": 2.038104295730591,
+      "learning_rate": 0.0006894351214867937,
+      "loss": 5.7046,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5316455696202531,
+      "grad_norm": 2.1570217609405518,
+      "learning_rate": 0.0006869330686175058,
+      "loss": 5.7063,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5327004219409283,
+      "grad_norm": 2.4370977878570557,
+      "learning_rate": 0.0006844317224439788,
+      "loss": 5.7075,
+      "step": 5050
+    },
+    {
+      "epoch": 0.5337552742616034,
+      "grad_norm": 2.059896945953369,
+      "learning_rate": 0.0006819311109950177,
+      "loss": 5.7082,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5348101265822784,
+      "grad_norm": 2.6178388595581055,
+      "learning_rate": 0.0006794312622911953,
+      "loss": 5.7104,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5358649789029536,
+      "grad_norm": 3.1716971397399902,
+      "learning_rate": 0.0006769322043445363,
+      "loss": 5.709,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5369198312236287,
+      "grad_norm": 2.5717313289642334,
+      "learning_rate": 0.0006744339651582059,
+      "loss": 5.7044,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5379746835443038,
+      "grad_norm": 2.5640711784362793,
+      "learning_rate": 0.0006719365727261935,
+      "loss": 5.7019,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5390295358649789,
+      "grad_norm": 2.208054542541504,
+      "learning_rate": 0.0006694400550330013,
+      "loss": 5.7058,
+      "step": 5110
+    },
+    {
+      "epoch": 0.540084388185654,
+      "grad_norm": 2.598874807357788,
+      "learning_rate": 0.0006669444400533286,
+      "loss": 5.7045,
+      "step": 5120
+    },
+    {
+      "epoch": 0.5411392405063291,
+      "grad_norm": 2.0729267597198486,
+      "learning_rate": 0.0006644497557517599,
+      "loss": 5.7016,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5421940928270043,
+      "grad_norm": 2.6994481086730957,
+      "learning_rate": 0.0006619560300824507,
+      "loss": 5.706,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5432489451476793,
+      "grad_norm": 2.696298599243164,
+      "learning_rate": 0.0006594632909888154,
+      "loss": 5.7024,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5443037974683544,
+      "grad_norm": 2.7497174739837646,
+      "learning_rate": 0.0006569715664032124,
+      "loss": 5.6947,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5453586497890295,
+      "grad_norm": 2.6455984115600586,
+      "learning_rate": 0.0006544808842466324,
+      "loss": 5.7019,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5464135021097046,
+      "grad_norm": 2.283980131149292,
+      "learning_rate": 0.0006519912724283851,
+      "loss": 5.6975,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5474683544303798,
+      "grad_norm": 2.4803431034088135,
+      "learning_rate": 0.0006495027588457864,
+      "loss": 5.6988,
+      "step": 5190
+    },
+    {
+      "epoch": 0.5485232067510548,
+      "grad_norm": 2.4492745399475098,
+      "learning_rate": 0.0006470153713838463,
+      "loss": 5.7031,
+      "step": 5200
+    },
+    {
+      "epoch": 0.54957805907173,
+      "grad_norm": 2.6171047687530518,
+      "learning_rate": 0.0006445291379149556,
+      "loss": 5.6966,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5506329113924051,
+      "grad_norm": 2.669234037399292,
+      "learning_rate": 0.0006420440862985748,
+      "loss": 5.6944,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5516877637130801,
+      "grad_norm": 2.1747076511383057,
+      "learning_rate": 0.0006395602443809203,
+      "loss": 5.6959,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5527426160337553,
+      "grad_norm": 2.3398029804229736,
+      "learning_rate": 0.0006370776399946536,
+      "loss": 5.6984,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5537974683544303,
+      "grad_norm": 2.797767400741577,
+      "learning_rate": 0.0006345963009585694,
+      "loss": 5.6958,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5548523206751055,
+      "grad_norm": 2.220432758331299,
+      "learning_rate": 0.0006321162550772829,
+      "loss": 5.7032,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5559071729957806,
+      "grad_norm": 2.5476531982421875,
+      "learning_rate": 0.0006296375301409187,
+      "loss": 5.7027,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5569620253164557,
+      "grad_norm": 2.845794916152954,
+      "learning_rate": 0.0006271601539248012,
+      "loss": 5.6981,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5580168776371308,
+      "grad_norm": 2.6486692428588867,
+      "learning_rate": 0.0006246841541891399,
+      "loss": 5.6996,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5590717299578059,
+      "grad_norm": 2.8589515686035156,
+      "learning_rate": 0.0006222095586787208,
+      "loss": 5.6991,
+      "step": 5300
+    },
+    {
+      "epoch": 0.560126582278481,
+      "grad_norm": 2.432746410369873,
+      "learning_rate": 0.0006197363951225951,
+      "loss": 5.7008,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5611814345991561,
+      "grad_norm": 2.1326663494110107,
+      "learning_rate": 0.0006172646912337678,
+      "loss": 5.6915,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5622362869198312,
+      "grad_norm": 2.2239465713500977,
+      "learning_rate": 0.0006147944747088881,
+      "loss": 5.6996,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5632911392405063,
+      "grad_norm": 2.7268221378326416,
+      "learning_rate": 0.000612325773227938,
+      "loss": 5.6946,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5643459915611815,
+      "grad_norm": 2.1351566314697266,
+      "learning_rate": 0.0006098586144539235,
+      "loss": 5.6945,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5654008438818565,
+      "grad_norm": 2.342799663543701,
+      "learning_rate": 0.0006073930260325632,
+      "loss": 5.6964,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5664556962025317,
+      "grad_norm": 2.5091652870178223,
+      "learning_rate": 0.0006049290355919792,
+      "loss": 5.6976,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5675105485232067,
+      "grad_norm": 2.576202630996704,
+      "learning_rate": 0.0006024666707423875,
+      "loss": 5.6962,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5685654008438819,
+      "grad_norm": 2.2374064922332764,
+      "learning_rate": 0.0006000059590757886,
+      "loss": 5.69,
+      "step": 5390
+    },
+    {
+      "epoch": 0.569620253164557,
+      "grad_norm": 1.7059028148651123,
+      "learning_rate": 0.0005975469281656581,
+      "loss": 5.6902,
+      "step": 5400
+    },
+    {
+      "epoch": 0.570675105485232,
+      "grad_norm": 2.33400297164917,
+      "learning_rate": 0.0005950896055666384,
+      "loss": 5.6925,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5717299578059072,
+      "grad_norm": 2.604609727859497,
+      "learning_rate": 0.0005926340188142289,
+      "loss": 5.6882,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5727848101265823,
+      "grad_norm": 2.3319449424743652,
+      "learning_rate": 0.0005901801954244782,
+      "loss": 5.6927,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5738396624472574,
+      "grad_norm": 2.3133225440979004,
+      "learning_rate": 0.0005877281628936756,
+      "loss": 5.6851,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5748945147679325,
+      "grad_norm": 2.5625109672546387,
+      "learning_rate": 0.0005852779486980427,
+      "loss": 5.6923,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5759493670886076,
+      "grad_norm": 2.46476411819458,
+      "learning_rate": 0.0005828295802934263,
+      "loss": 5.6888,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5770042194092827,
+      "grad_norm": 2.5715067386627197,
+      "learning_rate": 0.0005803830851149892,
+      "loss": 5.6925,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5780590717299579,
+      "grad_norm": 2.1198129653930664,
+      "learning_rate": 0.0005779384905769053,
+      "loss": 5.6919,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5791139240506329,
+      "grad_norm": 2.1729331016540527,
+      "learning_rate": 0.0005754958240720498,
+      "loss": 5.691,
+      "step": 5490
+    },
+    {
+      "epoch": 0.580168776371308,
+      "grad_norm": 2.7466843128204346,
+      "learning_rate": 0.0005730551129716936,
+      "loss": 5.6899,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5812236286919831,
+      "grad_norm": 2.352106809616089,
+      "learning_rate": 0.0005706163846251961,
+      "loss": 5.6915,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5822784810126582,
+      "grad_norm": 2.233600616455078,
+      "learning_rate": 0.0005681796663596996,
+      "loss": 5.6884,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "grad_norm": 2.600602626800537,
+      "learning_rate": 0.0005657449854798216,
+      "loss": 5.6868,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5843881856540084,
+      "grad_norm": 2.4469590187072754,
+      "learning_rate": 0.0005633123692673503,
+      "loss": 5.6866,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5854430379746836,
+      "grad_norm": 2.9685914516448975,
+      "learning_rate": 0.0005608818449809387,
+      "loss": 5.6914,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5864978902953587,
+      "grad_norm": 2.183056354522705,
+      "learning_rate": 0.0005584534398557977,
+      "loss": 5.6898,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5875527426160337,
+      "grad_norm": 2.352800130844116,
+      "learning_rate": 0.0005560271811033928,
+      "loss": 5.6859,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5886075949367089,
+      "grad_norm": 2.5035574436187744,
+      "learning_rate": 0.0005536030959111377,
+      "loss": 5.6838,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5896624472573839,
+      "grad_norm": 2.2204413414001465,
+      "learning_rate": 0.0005511812114420908,
+      "loss": 5.6882,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5907172995780591,
+      "grad_norm": 2.376124143600464,
+      "learning_rate": 0.0005487615548346502,
+      "loss": 5.6917,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5917721518987342,
+      "grad_norm": 3.069185495376587,
+      "learning_rate": 0.0005463441532022495,
+      "loss": 5.6848,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5928270042194093,
+      "grad_norm": 2.6441245079040527,
+      "learning_rate": 0.0005439290336330545,
+      "loss": 5.6844,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5938818565400844,
+      "grad_norm": 2.236853837966919,
+      "learning_rate": 0.0005415162231896593,
+      "loss": 5.6891,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5949367088607594,
+      "grad_norm": 2.099480390548706,
+      "learning_rate": 0.000539105748908783,
+      "loss": 5.693,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5959915611814346,
+      "grad_norm": 2.246088743209839,
+      "learning_rate": 0.0005366976378009668,
+      "loss": 5.6871,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5970464135021097,
+      "grad_norm": 2.4582173824310303,
+      "learning_rate": 0.0005342919168502717,
+      "loss": 5.6902,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5981012658227848,
+      "grad_norm": 2.236433267593384,
+      "learning_rate": 0.0005318886130139753,
+      "loss": 5.6885,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5991561181434599,
+      "grad_norm": 2.696704864501953,
+      "learning_rate": 0.0005294877532222709,
+      "loss": 5.6928,
+      "step": 5680
+    },
+    {
+      "epoch": 0.6002109704641351,
+      "grad_norm": 2.4552063941955566,
+      "learning_rate": 0.0005270893643779649,
+      "loss": 5.6873,
+      "step": 5690
+    },
+    {
+      "epoch": 0.6012658227848101,
+      "grad_norm": 2.2144582271575928,
+      "learning_rate": 0.0005246934733561751,
+      "loss": 5.6884,
+      "step": 5700
+    },
+    {
+      "epoch": 0.6023206751054853,
+      "grad_norm": 2.4040567874908447,
+      "learning_rate": 0.0005223001070040305,
+      "loss": 5.6822,
+      "step": 5710
+    },
+    {
+      "epoch": 0.6033755274261603,
+      "grad_norm": 2.549650192260742,
+      "learning_rate": 0.0005199092921403696,
+      "loss": 5.6828,
+      "step": 5720
+    },
+    {
+      "epoch": 0.6044303797468354,
+      "grad_norm": 2.1063625812530518,
+      "learning_rate": 0.00051752105555544,
+      "loss": 5.6859,
+      "step": 5730
+    },
+    {
+      "epoch": 0.6054852320675106,
+      "grad_norm": 2.4981205463409424,
+      "learning_rate": 0.0005151354240105994,
+      "loss": 5.6888,
+      "step": 5740
+    },
+    {
+      "epoch": 0.6065400843881856,
+      "grad_norm": 2.7450053691864014,
+      "learning_rate": 0.0005127524242380139,
+      "loss": 5.6814,
+      "step": 5750
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 2.1085798740386963,
+      "learning_rate": 0.0005103720829403594,
+      "loss": 5.6827,
+      "step": 5760
+    },
+    {
+      "epoch": 0.6086497890295358,
+      "grad_norm": 2.126136541366577,
+      "learning_rate": 0.0005079944267905226,
+      "loss": 5.6815,
+      "step": 5770
+    },
+    {
+      "epoch": 0.609704641350211,
+      "grad_norm": 2.2256312370300293,
+      "learning_rate": 0.0005056194824313015,
+      "loss": 5.6817,
+      "step": 5780
+    },
+    {
+      "epoch": 0.6107594936708861,
+      "grad_norm": 2.488480806350708,
+      "learning_rate": 0.0005032472764751074,
+      "loss": 5.6767,
+      "step": 5790
+    },
+    {
+      "epoch": 0.6118143459915611,
+      "grad_norm": 2.380218267440796,
+      "learning_rate": 0.000500877835503666,
+      "loss": 5.681,
+      "step": 5800
+    },
+    {
+      "epoch": 0.6128691983122363,
+      "grad_norm": 2.7666079998016357,
+      "learning_rate": 0.000498511186067721,
+      "loss": 5.6787,
+      "step": 5810
+    },
+    {
+      "epoch": 0.6139240506329114,
+      "grad_norm": 1.7696150541305542,
+      "learning_rate": 0.0004961473546867346,
+      "loss": 5.6829,
+      "step": 5820
+    },
+    {
+      "epoch": 0.6149789029535865,
+      "grad_norm": 2.391683340072632,
+      "learning_rate": 0.0004937863678485915,
+      "loss": 5.682,
+      "step": 5830
+    },
+    {
+      "epoch": 0.6160337552742616,
+      "grad_norm": 1.9751567840576172,
+      "learning_rate": 0.0004914282520093023,
+      "loss": 5.6779,
+      "step": 5840
+    },
+    {
+      "epoch": 0.6170886075949367,
+      "grad_norm": 1.9766874313354492,
+      "learning_rate": 0.0004890730335927063,
+      "loss": 5.6791,
+      "step": 5850
+    },
+    {
+      "epoch": 0.6181434599156118,
+      "grad_norm": 1.967281460762024,
+      "learning_rate": 0.00048672073899017564,
+      "loss": 5.6802,
+      "step": 5860
+    },
+    {
+      "epoch": 0.619198312236287,
+      "grad_norm": 1.9689607620239258,
+      "learning_rate": 0.0004843713945603205,
+      "loss": 5.6693,
+      "step": 5870
+    },
+    {
+      "epoch": 0.620253164556962,
+      "grad_norm": 2.3804993629455566,
+      "learning_rate": 0.00048202502662869195,
+      "loss": 5.6806,
+      "step": 5880
+    },
+    {
+      "epoch": 0.6213080168776371,
+      "grad_norm": 1.9705326557159424,
+      "learning_rate": 0.0004796816614874885,
+      "loss": 5.6744,
+      "step": 5890
+    },
+    {
+      "epoch": 0.6223628691983122,
+      "grad_norm": 2.1561052799224854,
+      "learning_rate": 0.00047734132539526086,
+      "loss": 5.6731,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6234177215189873,
+      "grad_norm": 2.5553174018859863,
+      "learning_rate": 0.00047500404457661747,
+      "loss": 5.6702,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6244725738396625,
+      "grad_norm": 2.3821535110473633,
+      "learning_rate": 0.00047266984522193134,
+      "loss": 5.67,
+      "step": 5920
+    },
+    {
+      "epoch": 0.6255274261603375,
+      "grad_norm": 1.9307197332382202,
+      "learning_rate": 0.00047033875348704576,
+      "loss": 5.6732,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6265822784810127,
+      "grad_norm": 2.9684524536132812,
+      "learning_rate": 0.00046801079549298224,
+      "loss": 5.6791,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6276371308016878,
+      "grad_norm": 2.4825356006622314,
+      "learning_rate": 0.0004656859973256466,
+      "loss": 5.6755,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6286919831223629,
+      "grad_norm": 2.350797176361084,
+      "learning_rate": 0.00046336438503553754,
+      "loss": 5.6763,
+      "step": 5960
+    },
+    {
+      "epoch": 0.629746835443038,
+      "grad_norm": 2.2591776847839355,
+      "learning_rate": 0.00046104598463745424,
+      "loss": 5.6676,
+      "step": 5970
+    },
+    {
+      "epoch": 0.630801687763713,
+      "grad_norm": 1.793256402015686,
+      "learning_rate": 0.0004587308221102053,
+      "loss": 5.6679,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6318565400843882,
+      "grad_norm": 2.18587327003479,
+      "learning_rate": 0.00045641892339631703,
+      "loss": 5.6733,
+      "step": 5990
+    },
+    {
+      "epoch": 0.6329113924050633,
+      "grad_norm": 1.850223183631897,
+      "learning_rate": 0.000454110314401744,
+      "loss": 5.6784,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6339662447257384,
+      "grad_norm": 2.294945001602173,
+      "learning_rate": 0.00045180502099557686,
+      "loss": 5.6711,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6350210970464135,
+      "grad_norm": 2.2696354389190674,
+      "learning_rate": 0.00044950306900975377,
+      "loss": 5.6736,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6360759493670886,
+      "grad_norm": Infinity,
+      "learning_rate": 0.00044743419045683674,
+      "loss": 5.6717,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6371308016877637,
+      "grad_norm": 2.0113344192504883,
+      "learning_rate": 0.00044513865820232525,
+      "loss": 5.675,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6381856540084389,
+      "grad_norm": 2.7012720108032227,
+      "learning_rate": 0.00044284654206800826,
+      "loss": 5.6761,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6392405063291139,
+      "grad_norm": 2.516242027282715,
+      "learning_rate": 0.0004405578677381661,
+      "loss": 5.6733,
+      "step": 6060
+    },
+    {
+      "epoch": 0.640295358649789,
+      "grad_norm": 2.2177603244781494,
+      "learning_rate": 0.00043827266085851203,
+      "loss": 5.6657,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6413502109704642,
+      "grad_norm": 2.267564535140991,
+      "learning_rate": 0.00043599094703590524,
+      "loss": 5.6702,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6424050632911392,
+      "grad_norm": 2.607971429824829,
+      "learning_rate": 0.0004337127518380632,
+      "loss": 5.6715,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6434599156118144,
+      "grad_norm": 2.4445533752441406,
+      "learning_rate": 0.0004314381007932756,
+      "loss": 5.6619,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6445147679324894,
+      "grad_norm": 1.9204957485198975,
+      "learning_rate": 0.00042916701939011787,
+      "loss": 5.6739,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6455696202531646,
+      "grad_norm": 2.3194024562835693,
+      "learning_rate": 0.0004268995330771661,
+      "loss": 5.668,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6466244725738397,
+      "grad_norm": 2.309570789337158,
+      "learning_rate": 0.00042463566726271137,
+      "loss": 5.6737,
+      "step": 6130
+    },
+    {
+      "epoch": 0.6476793248945147,
+      "grad_norm": 2.1416492462158203,
+      "learning_rate": 0.00042237544731447616,
+      "loss": 5.6667,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6487341772151899,
+      "grad_norm": 2.3147592544555664,
+      "learning_rate": 0.0004201188985593283,
+      "loss": 5.6682,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6497890295358649,
+      "grad_norm": 2.575751304626465,
+      "learning_rate": 0.00041786604628299846,
+      "loss": 5.6654,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6508438818565401,
+      "grad_norm": 2.185352087020874,
+      "learning_rate": 0.00041561691572979624,
+      "loss": 5.6665,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6518987341772152,
+      "grad_norm": 2.158193826675415,
+      "learning_rate": 0.000413371532102328,
+      "loss": 5.6611,
+      "step": 6180
+    },
+    {
+      "epoch": 0.6529535864978903,
+      "grad_norm": 2.3251264095306396,
+      "learning_rate": 0.0004111299205612135,
+      "loss": 5.6673,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6540084388185654,
+      "grad_norm": 2.5166869163513184,
+      "learning_rate": 0.00040889210622480467,
+      "loss": 5.6698,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6550632911392406,
+      "grad_norm": 2.107229471206665,
+      "learning_rate": 0.0004066581141689038,
+      "loss": 5.6683,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6561181434599156,
+      "grad_norm": 1.537351369857788,
+      "learning_rate": 0.00040442796942648273,
+      "loss": 5.6644,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6571729957805907,
+      "grad_norm": 2.1709556579589844,
+      "learning_rate": 0.0004022016969874023,
+      "loss": 5.6656,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6582278481012658,
+      "grad_norm": 2.210150957107544,
+      "learning_rate": 0.00039997932179813205,
+      "loss": 5.6617,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6592827004219409,
+      "grad_norm": 2.699484348297119,
+      "learning_rate": 0.00039776086876147133,
+      "loss": 5.6654,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6603375527426161,
+      "grad_norm": 2.056365728378296,
+      "learning_rate": 0.0003955463627362694,
+      "loss": 5.6629,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6613924050632911,
+      "grad_norm": 2.206430435180664,
+      "learning_rate": 0.00039333582853714793,
+      "loss": 5.6589,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6624472573839663,
+      "grad_norm": 2.280083179473877,
+      "learning_rate": 0.00039112929093422185,
+      "loss": 5.6606,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6635021097046413,
+      "grad_norm": 2.3351495265960693,
+      "learning_rate": 0.0003889267746528225,
+      "loss": 5.6582,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6645569620253164,
+      "grad_norm": 2.5855207443237305,
+      "learning_rate": 0.00038672830437322007,
+      "loss": 5.6591,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6656118143459916,
+      "grad_norm": 1.8922674655914307,
+      "learning_rate": 0.0003845339047303477,
+      "loss": 5.6565,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 2.294938087463379,
+      "learning_rate": 0.00038234360031352485,
+      "loss": 5.6599,
+      "step": 6320
+    },
+    {
+      "epoch": 0.6677215189873418,
+      "grad_norm": 1.759223461151123,
+      "learning_rate": 0.0003801574156661817,
+      "loss": 5.6595,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6687763713080169,
+      "grad_norm": 2.259773015975952,
+      "learning_rate": 0.0003779753752855853,
+      "loss": 5.6593,
+      "step": 6340
+    },
+    {
+      "epoch": 0.669831223628692,
+      "grad_norm": 1.891624927520752,
+      "learning_rate": 0.0003757975036225632,
+      "loss": 5.6579,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6708860759493671,
+      "grad_norm": 2.2626521587371826,
+      "learning_rate": 0.0003736238250812308,
+      "loss": 5.6517,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6719409282700421,
+      "grad_norm": 2.5813846588134766,
+      "learning_rate": 0.0003714543640187177,
+      "loss": 5.6627,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6729957805907173,
+      "grad_norm": 2.272970676422119,
+      "learning_rate": 0.0003692891447448943,
+      "loss": 5.6533,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6740506329113924,
+      "grad_norm": 2.1752026081085205,
+      "learning_rate": 0.00036712819152209954,
+      "loss": 5.6559,
+      "step": 6390
+    },
+    {
+      "epoch": 0.6751054852320675,
+      "grad_norm": 2.667616128921509,
+      "learning_rate": 0.0003649715285648701,
+      "loss": 5.655,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6761603375527426,
+      "grad_norm": 2.1136720180511475,
+      "learning_rate": 0.000362819180039667,
+      "loss": 5.6548,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6772151898734177,
+      "grad_norm": 1.9631470441818237,
+      "learning_rate": 0.0003606711700646067,
+      "loss": 5.6544,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6782700421940928,
+      "grad_norm": 2.2858240604400635,
+      "learning_rate": 0.00035852752270918955,
+      "loss": 5.6515,
+      "step": 6430
+    },
+    {
+      "epoch": 0.679324894514768,
+      "grad_norm": 2.0766820907592773,
+      "learning_rate": 0.00035638826199403103,
+      "loss": 5.65,
+      "step": 6440
+    },
+    {
+      "epoch": 0.680379746835443,
+      "grad_norm": 2.0123932361602783,
+      "learning_rate": 0.000354253411890592,
+      "loss": 5.6507,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6814345991561181,
+      "grad_norm": 2.355245351791382,
+      "learning_rate": 0.00035212299632090996,
+      "loss": 5.6522,
+      "step": 6460
+    },
+    {
+      "epoch": 0.6824894514767933,
+      "grad_norm": 2.1331210136413574,
+      "learning_rate": 0.0003499970391573322,
+      "loss": 5.6489,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6835443037974683,
+      "grad_norm": 1.768051266670227,
+      "learning_rate": 0.0003478755642222466,
+      "loss": 5.6519,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6845991561181435,
+      "grad_norm": 2.017947196960449,
+      "learning_rate": 0.0003457585952878156,
+      "loss": 5.6499,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6856540084388185,
+      "grad_norm": 2.0069286823272705,
+      "learning_rate": 0.00034364615607570994,
+      "loss": 5.6525,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6867088607594937,
+      "grad_norm": 2.2581350803375244,
+      "learning_rate": 0.00034153827025684225,
+      "loss": 5.6488,
+      "step": 6510
+    },
+    {
+      "epoch": 0.6877637130801688,
+      "grad_norm": 1.5537643432617188,
+      "learning_rate": 0.0003394349614511026,
+      "loss": 5.6473,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6888185654008439,
+      "grad_norm": 1.8359655141830444,
+      "learning_rate": 0.000337336253227093,
+      "loss": 5.651,
+      "step": 6530
+    },
+    {
+      "epoch": 0.689873417721519,
+      "grad_norm": 2.552940845489502,
+      "learning_rate": 0.00033524216910186394,
+      "loss": 5.636,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6909282700421941,
+      "grad_norm": 2.3417508602142334,
+      "learning_rate": 0.0003331527325406506,
+      "loss": 5.6452,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6919831223628692,
+      "grad_norm": 1.6594558954238892,
+      "learning_rate": 0.00033106796695660983,
+      "loss": 5.6487,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6930379746835443,
+      "grad_norm": 1.9313304424285889,
+      "learning_rate": 0.00032898789571055796,
+      "loss": 5.6422,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6940928270042194,
+      "grad_norm": 1.9007086753845215,
+      "learning_rate": 0.0003269125421107091,
+      "loss": 5.6444,
+      "step": 6580
+    },
+    {
+      "epoch": 0.6951476793248945,
+      "grad_norm": 2.2350409030914307,
+      "learning_rate": 0.00032484192941241316,
+      "loss": 5.6419,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6962025316455697,
+      "grad_norm": 2.011573076248169,
+      "learning_rate": 0.0003227760808178973,
+      "loss": 5.6472,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6972573839662447,
+      "grad_norm": 1.8302812576293945,
+      "learning_rate": 0.00032071501947600334,
+      "loss": 5.6479,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6983122362869199,
+      "grad_norm": 2.2356200218200684,
+      "learning_rate": 0.00031865876848192993,
+      "loss": 5.6398,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6993670886075949,
+      "grad_norm": 2.19203782081604,
+      "learning_rate": 0.00031660735087697363,
+      "loss": 5.6408,
+      "step": 6630
+    },
+    {
+      "epoch": 0.70042194092827,
+      "grad_norm": 1.9747456312179565,
+      "learning_rate": 0.0003145607896482704,
+      "loss": 5.6412,
+      "step": 6640
+    },
+    {
+      "epoch": 0.7014767932489452,
+      "grad_norm": 1.4504653215408325,
+      "learning_rate": 0.000312519107728538,
+      "loss": 5.6463,
+      "step": 6650
+    },
+    {
+      "epoch": 0.7025316455696202,
+      "grad_norm": 1.7272939682006836,
+      "learning_rate": 0.0003104823279958191,
+      "loss": 5.6327,
+      "step": 6660
+    },
+    {
+      "epoch": 0.7035864978902954,
+      "grad_norm": 1.5269538164138794,
+      "learning_rate": 0.00030845047327322556,
+      "loss": 5.6383,
+      "step": 6670
+    },
+    {
+      "epoch": 0.7046413502109705,
+      "grad_norm": 2.4669835567474365,
+      "learning_rate": 0.0003064235663286815,
+      "loss": 5.6399,
+      "step": 6680
+    },
+    {
+      "epoch": 0.7056962025316456,
+      "grad_norm": 1.8575663566589355,
+      "learning_rate": 0.00030440162987466896,
+      "loss": 5.6375,
+      "step": 6690
+    },
+    {
+      "epoch": 0.7067510548523207,
+      "grad_norm": 2.0847558975219727,
+      "learning_rate": 0.0003023846865679731,
+      "loss": 5.6466,
+      "step": 6700
+    },
+    {
+      "epoch": 0.7078059071729957,
+      "grad_norm": 1.9296622276306152,
+      "learning_rate": 0.0003003727590094285,
+      "loss": 5.6349,
+      "step": 6710
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 2.146658182144165,
+      "learning_rate": 0.00029836586974366574,
+      "loss": 5.6454,
+      "step": 6720
+    },
+    {
+      "epoch": 0.709915611814346,
+      "grad_norm": 1.9521843194961548,
+      "learning_rate": 0.00029636404125885936,
+      "loss": 5.6421,
+      "step": 6730
+    },
+    {
+      "epoch": 0.7109704641350211,
+      "grad_norm": 1.8796813488006592,
+      "learning_rate": 0.00029436729598647483,
+      "loss": 5.6397,
+      "step": 6740
+    },
+    {
+      "epoch": 0.7120253164556962,
+      "grad_norm": 1.4689815044403076,
+      "learning_rate": 0.0002923756563010179,
+      "loss": 5.6374,
+      "step": 6750
+    },
+    {
+      "epoch": 0.7130801687763713,
+      "grad_norm": 1.7355879545211792,
+      "learning_rate": 0.0002903891445197836,
+      "loss": 5.6385,
+      "step": 6760
+    },
+    {
+      "epoch": 0.7141350210970464,
+      "grad_norm": 1.4216035604476929,
+      "learning_rate": 0.0002884077829026066,
+      "loss": 5.6374,
+      "step": 6770
+    },
+    {
+      "epoch": 0.7151898734177216,
+      "grad_norm": 1.6927716732025146,
+      "learning_rate": 0.00028643159365161113,
+      "loss": 5.6318,
+      "step": 6780
+    },
+    {
+      "epoch": 0.7162447257383966,
+      "grad_norm": 1.9339640140533447,
+      "learning_rate": 0.00028446059891096265,
+      "loss": 5.6434,
+      "step": 6790
+    },
+    {
+      "epoch": 0.7172995780590717,
+      "grad_norm": 1.9566271305084229,
+      "learning_rate": 0.0002824948207666199,
+      "loss": 5.6365,
+      "step": 6800
+    },
+    {
+      "epoch": 0.7183544303797469,
+      "grad_norm": 2.10983943939209,
+      "learning_rate": 0.00028053428124608684,
+      "loss": 5.6308,
+      "step": 6810
+    },
+    {
+      "epoch": 0.7194092827004219,
+      "grad_norm": 1.9768773317337036,
+      "learning_rate": 0.00027857900231816594,
+      "loss": 5.6379,
+      "step": 6820
+    },
+    {
+      "epoch": 0.7204641350210971,
+      "grad_norm": 1.914456844329834,
+      "learning_rate": 0.0002766290058927123,
+      "loss": 5.6344,
+      "step": 6830
+    },
+    {
+      "epoch": 0.7215189873417721,
+      "grad_norm": 2.142565965652466,
+      "learning_rate": 0.00027468431382038816,
+      "loss": 5.6328,
+      "step": 6840
+    },
+    {
+      "epoch": 0.7225738396624473,
+      "grad_norm": 2.1876516342163086,
+      "learning_rate": 0.00027274494789241766,
+      "loss": 5.6266,
+      "step": 6850
+    },
+    {
+      "epoch": 0.7236286919831224,
+      "grad_norm": 1.717586874961853,
+      "learning_rate": 0.00027081092984034303,
+      "loss": 5.6319,
+      "step": 6860
+    },
+    {
+      "epoch": 0.7246835443037974,
+      "grad_norm": 1.7215545177459717,
+      "learning_rate": 0.00026888228133578086,
+      "loss": 5.6345,
+      "step": 6870
+    },
+    {
+      "epoch": 0.7257383966244726,
+      "grad_norm": 2.2183592319488525,
+      "learning_rate": 0.00026695902399017935,
+      "loss": 5.6309,
+      "step": 6880
+    },
+    {
+      "epoch": 0.7267932489451476,
+      "grad_norm": 1.5335004329681396,
+      "learning_rate": 0.0002650411793545763,
+      "loss": 5.6321,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7278481012658228,
+      "grad_norm": 1.779222011566162,
+      "learning_rate": 0.0002631287689193571,
+      "loss": 5.6316,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7289029535864979,
+      "grad_norm": 2.5856895446777344,
+      "learning_rate": 0.00026122181411401444,
+      "loss": 5.6302,
+      "step": 6910
+    },
+    {
+      "epoch": 0.729957805907173,
+      "grad_norm": 1.9173972606658936,
+      "learning_rate": 0.0002593203363069084,
+      "loss": 5.6289,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7310126582278481,
+      "grad_norm": 2.1186904907226562,
+      "learning_rate": 0.0002574243568050261,
+      "loss": 5.629,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7320675105485233,
+      "grad_norm": 2.2834548950195312,
+      "learning_rate": 0.0002555338968537436,
+      "loss": 5.6261,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7331223628691983,
+      "grad_norm": 1.2145274877548218,
+      "learning_rate": 0.00025364897763658777,
+      "loss": 5.6312,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7341772151898734,
+      "grad_norm": 1.8010163307189941,
+      "learning_rate": 0.0002517696202749988,
+      "loss": 5.626,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7352320675105485,
+      "grad_norm": 1.7088450193405151,
+      "learning_rate": 0.0002498958458280936,
+      "loss": 5.6251,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7362869198312236,
+      "grad_norm": 2.9075794219970703,
+      "learning_rate": 0.0002480276752924295,
+      "loss": 5.6265,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7373417721518988,
+      "grad_norm": 1.4042394161224365,
+      "learning_rate": 0.00024616512960177014,
+      "loss": 5.6184,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7383966244725738,
+      "grad_norm": 1.5842777490615845,
+      "learning_rate": 0.00024430822962684905,
+      "loss": 5.6299,
+      "step": 7000
+    },
+    {
+      "epoch": 0.739451476793249,
+      "grad_norm": 1.6897560358047485,
+      "learning_rate": 0.00024245699617513733,
+      "loss": 5.622,
+      "step": 7010
+    },
+    {
+      "epoch": 0.740506329113924,
+      "grad_norm": 1.385205626487732,
+      "learning_rate": 0.00024061144999060956,
+      "loss": 5.6238,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7415611814345991,
+      "grad_norm": 2.0923962593078613,
+      "learning_rate": 0.00023877161175351206,
+      "loss": 5.6258,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7426160337552743,
+      "grad_norm": 1.8577938079833984,
+      "learning_rate": 0.00023693750208013045,
+      "loss": 5.6253,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7436708860759493,
+      "grad_norm": 1.9434581995010376,
+      "learning_rate": 0.0002351091415225591,
+      "loss": 5.6175,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7447257383966245,
+      "grad_norm": 1.719789743423462,
+      "learning_rate": 0.00023328655056847124,
+      "loss": 5.6194,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7457805907172996,
+      "grad_norm": 1.6356306076049805,
+      "learning_rate": 0.00023146974964088825,
+      "loss": 5.6197,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7468354430379747,
+      "grad_norm": 1.5827842950820923,
+      "learning_rate": 0.00022965875909795164,
+      "loss": 5.6207,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7478902953586498,
+      "grad_norm": 2.2272980213165283,
+      "learning_rate": 0.0002278535992326947,
+      "loss": 5.6244,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7489451476793249,
+      "grad_norm": 1.9679064750671387,
+      "learning_rate": 0.0002260542902728151,
+      "loss": 5.6143,
+      "step": 7100
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9241124391555786,
+      "learning_rate": 0.00022426085238044823,
+      "loss": 5.6164,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7510548523206751,
+      "grad_norm": 2.9112601280212402,
+      "learning_rate": 0.00022247330565194171,
+      "loss": 5.6162,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7521097046413502,
+      "grad_norm": 1.6614335775375366,
+      "learning_rate": 0.0002206916701176293,
+      "loss": 5.6138,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7531645569620253,
+      "grad_norm": 2.4389736652374268,
+      "learning_rate": 0.00021891596574160715,
+      "loss": 5.6158,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7542194092827004,
+      "grad_norm": 2.3244760036468506,
+      "learning_rate": 0.00021714621242150973,
+      "loss": 5.6145,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7552742616033755,
+      "grad_norm": 1.6010584831237793,
+      "learning_rate": 0.0002153824299882872,
+      "loss": 5.6207,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7563291139240507,
+      "grad_norm": 2.0947442054748535,
+      "learning_rate": 0.00021362463820598297,
+      "loss": 5.6108,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7573839662447257,
+      "grad_norm": 2.1411478519439697,
+      "learning_rate": 0.00021187285677151205,
+      "loss": 5.6103,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7584388185654009,
+      "grad_norm": 1.9849368333816528,
+      "learning_rate": 0.00021012710531444112,
+      "loss": 5.6106,
+      "step": 7190
+    },
+    {
+      "epoch": 0.759493670886076,
+      "grad_norm": 1.8407467603683472,
+      "learning_rate": 0.00020838740339676763,
+      "loss": 5.6146,
+      "step": 7200
+    },
+    {
+      "epoch": 0.760548523206751,
+      "grad_norm": 1.2459794282913208,
+      "learning_rate": 0.00020665377051270095,
+      "loss": 5.6127,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7616033755274262,
+      "grad_norm": 1.6676063537597656,
+      "learning_rate": 0.0002049262260884441,
+      "loss": 5.6164,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7626582278481012,
+      "grad_norm": 2.4118497371673584,
+      "learning_rate": 0.0002032047894819758,
+      "loss": 5.6079,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7637130801687764,
+      "grad_norm": 2.1241328716278076,
+      "learning_rate": 0.00020148947998283381,
+      "loss": 5.6056,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7647679324894515,
+      "grad_norm": 1.583636999130249,
+      "learning_rate": 0.00019978031681189864,
+      "loss": 5.6147,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7658227848101266,
+      "grad_norm": 1.8319991827011108,
+      "learning_rate": 0.00019807731912117828,
+      "loss": 5.6069,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7668776371308017,
+      "grad_norm": 2.0481069087982178,
+      "learning_rate": 0.00019638050599359326,
+      "loss": 5.6026,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7679324894514767,
+      "grad_norm": 1.3989096879959106,
+      "learning_rate": 0.0001946898964427633,
+      "loss": 5.6094,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7689873417721519,
+      "grad_norm": 1.9090906381607056,
+      "learning_rate": 0.0001930055094127938,
+      "loss": 5.6106,
+      "step": 7290
+    },
+    {
+      "epoch": 0.770042194092827,
+      "grad_norm": 1.8923743963241577,
+      "learning_rate": 0.00019132736377806394,
+      "loss": 5.6102,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7710970464135021,
+      "grad_norm": 1.879792332649231,
+      "learning_rate": 0.0001896554783430149,
+      "loss": 5.6042,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7721518987341772,
+      "grad_norm": 2.1781387329101562,
+      "learning_rate": 0.00018798987184193963,
+      "loss": 5.6033,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7732067510548524,
+      "grad_norm": 1.6626508235931396,
+      "learning_rate": 0.00018633056293877203,
+      "loss": 5.6033,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7742616033755274,
+      "grad_norm": 1.9299684762954712,
+      "learning_rate": 0.00018467757022687864,
+      "loss": 5.6091,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7753164556962026,
+      "grad_norm": 1.7240363359451294,
+      "learning_rate": 0.00018303091222884998,
+      "loss": 5.6062,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7763713080168776,
+      "grad_norm": 1.486358880996704,
+      "learning_rate": 0.00018139060739629287,
+      "loss": 5.6065,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7774261603375527,
+      "grad_norm": 1.3980551958084106,
+      "learning_rate": 0.00017975667410962366,
+      "loss": 5.5965,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7784810126582279,
+      "grad_norm": 1.783419132232666,
+      "learning_rate": 0.00017812913067786313,
+      "loss": 5.5999,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7795358649789029,
+      "grad_norm": 1.4898887872695923,
+      "learning_rate": 0.00017650799533842996,
+      "loss": 5.6035,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7805907172995781,
+      "grad_norm": 2.019289493560791,
+      "learning_rate": 0.00017489328625693715,
+      "loss": 5.6035,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7816455696202531,
+      "grad_norm": 1.1289573907852173,
+      "learning_rate": 0.0001732850215269885,
+      "loss": 5.5984,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7827004219409283,
+      "grad_norm": 1.3850795030593872,
+      "learning_rate": 0.00017168321916997547,
+      "loss": 5.606,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7837552742616034,
+      "grad_norm": 1.5702691078186035,
+      "learning_rate": 0.00017008789713487558,
+      "loss": 5.6038,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7848101265822784,
+      "grad_norm": 1.308498501777649,
+      "learning_rate": 0.00016849907329805118,
+      "loss": 5.6007,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7858649789029536,
+      "grad_norm": 1.6205036640167236,
+      "learning_rate": 0.00016691676546304936,
+      "loss": 5.6019,
+      "step": 7450
+    },
+    {
+      "epoch": 0.7869198312236287,
+      "grad_norm": 1.2850362062454224,
+      "learning_rate": 0.00016534099136040207,
+      "loss": 5.5976,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7879746835443038,
+      "grad_norm": 1.7243412733078003,
+      "learning_rate": 0.00016377176864742734,
+      "loss": 5.6001,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7890295358649789,
+      "grad_norm": 1.736440658569336,
+      "learning_rate": 0.00016220911490803206,
+      "loss": 5.6005,
+      "step": 7480
+    },
+    {
+      "epoch": 0.790084388185654,
+      "grad_norm": 1.1827021837234497,
+      "learning_rate": 0.00016065304765251423,
+      "loss": 5.6003,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7911392405063291,
+      "grad_norm": 1.4244201183319092,
+      "learning_rate": 0.00015910358431736745,
+      "loss": 5.599,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7921940928270043,
+      "grad_norm": 1.5327668190002441,
+      "learning_rate": 0.0001575607422650846,
+      "loss": 5.598,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7932489451476793,
+      "grad_norm": 1.2194591760635376,
+      "learning_rate": 0.00015602453878396479,
+      "loss": 5.5936,
+      "step": 7520
+    },
+    {
+      "epoch": 0.7943037974683544,
+      "grad_norm": 1.1456040143966675,
+      "learning_rate": 0.0001544949910879177,
+      "loss": 5.5948,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7953586497890295,
+      "grad_norm": 2.067368984222412,
+      "learning_rate": 0.00015297211631627234,
+      "loss": 5.5916,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7964135021097046,
+      "grad_norm": 1.431711196899414,
+      "learning_rate": 0.00015145593153358412,
+      "loss": 5.594,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7974683544303798,
+      "grad_norm": 1.6072288751602173,
+      "learning_rate": 0.00014994645372944367,
+      "loss": 5.5946,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7985232067510548,
+      "grad_norm": 1.4640034437179565,
+      "learning_rate": 0.00014844369981828698,
+      "loss": 5.5901,
+      "step": 7570
+    },
+    {
+      "epoch": 0.79957805907173,
+      "grad_norm": 1.4011296033859253,
+      "learning_rate": 0.00014694768663920537,
+      "loss": 5.5911,
+      "step": 7580
+    },
+    {
+      "epoch": 0.8006329113924051,
+      "grad_norm": 1.5977355241775513,
+      "learning_rate": 0.00014545843095575709,
+      "loss": 5.5914,
+      "step": 7590
+    },
+    {
+      "epoch": 0.8016877637130801,
+      "grad_norm": 1.3064849376678467,
+      "learning_rate": 0.00014397594945577912,
+      "loss": 5.594,
+      "step": 7600
+    },
+    {
+      "epoch": 0.8027426160337553,
+      "grad_norm": 2.050081491470337,
+      "learning_rate": 0.0001425002587512005,
+      "loss": 5.591,
+      "step": 7610
+    },
+    {
+      "epoch": 0.8037974683544303,
+      "grad_norm": 1.821736454963684,
+      "learning_rate": 0.00014103137537785633,
+      "loss": 5.5924,
+      "step": 7620
+    },
+    {
+      "epoch": 0.8048523206751055,
+      "grad_norm": 1.6769049167633057,
+      "learning_rate": 0.00013956931579530194,
+      "loss": 5.5925,
+      "step": 7630
+    },
+    {
+      "epoch": 0.8059071729957806,
+      "grad_norm": 1.734026312828064,
+      "learning_rate": 0.00013811409638662858,
+      "loss": 5.5874,
+      "step": 7640
+    },
+    {
+      "epoch": 0.8069620253164557,
+      "grad_norm": 1.4922587871551514,
+      "learning_rate": 0.00013666573345828083,
+      "loss": 5.5933,
+      "step": 7650
+    },
+    {
+      "epoch": 0.8080168776371308,
+      "grad_norm": 1.6053963899612427,
+      "learning_rate": 0.0001352242432398723,
+      "loss": 5.5883,
+      "step": 7660
+    },
+    {
+      "epoch": 0.8090717299578059,
+      "grad_norm": 1.4334664344787598,
+      "learning_rate": 0.00013378964188400457,
+      "loss": 5.5867,
+      "step": 7670
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 1.434798002243042,
+      "learning_rate": 0.00013236194546608645,
+      "loss": 5.5922,
+      "step": 7680
+    },
+    {
+      "epoch": 0.8111814345991561,
+      "grad_norm": 1.5868504047393799,
+      "learning_rate": 0.00013094116998415358,
+      "loss": 5.5888,
+      "step": 7690
+    },
+    {
+      "epoch": 0.8122362869198312,
+      "grad_norm": 1.3514033555984497,
+      "learning_rate": 0.0001295273313586885,
+      "loss": 5.5853,
+      "step": 7700
+    },
+    {
+      "epoch": 0.8132911392405063,
+      "grad_norm": 1.7414921522140503,
+      "learning_rate": 0.00012812044543244395,
+      "loss": 5.5855,
+      "step": 7710
+    },
+    {
+      "epoch": 0.8143459915611815,
+      "grad_norm": 1.4156917333602905,
+      "learning_rate": 0.00012672052797026344,
+      "loss": 5.5873,
+      "step": 7720
+    },
+    {
+      "epoch": 0.8154008438818565,
+      "grad_norm": 1.145093321800232,
+      "learning_rate": 0.00012532759465890567,
+      "loss": 5.591,
+      "step": 7730
+    },
+    {
+      "epoch": 0.8164556962025317,
+      "grad_norm": 1.2032239437103271,
+      "learning_rate": 0.00012394166110686857,
+      "loss": 5.5869,
+      "step": 7740
+    },
+    {
+      "epoch": 0.8175105485232067,
+      "grad_norm": 1.2643251419067383,
+      "learning_rate": 0.0001225627428442143,
+      "loss": 5.5773,
+      "step": 7750
+    },
+    {
+      "epoch": 0.8185654008438819,
+      "grad_norm": 1.7082284688949585,
+      "learning_rate": 0.0001211908553223954,
+      "loss": 5.583,
+      "step": 7760
+    },
+    {
+      "epoch": 0.819620253164557,
+      "grad_norm": 1.2069681882858276,
+      "learning_rate": 0.00011982601391408115,
+      "loss": 5.5827,
+      "step": 7770
+    },
+    {
+      "epoch": 0.820675105485232,
+      "grad_norm": 1.4891321659088135,
+      "learning_rate": 0.00011846823391298628,
+      "loss": 5.5763,
+      "step": 7780
+    },
+    {
+      "epoch": 0.8217299578059072,
+      "grad_norm": 1.9141031503677368,
+      "learning_rate": 0.00011711753053369861,
+      "loss": 5.5887,
+      "step": 7790
+    },
+    {
+      "epoch": 0.8227848101265823,
+      "grad_norm": 1.4617137908935547,
+      "learning_rate": 0.00011577391891150901,
+      "loss": 5.578,
+      "step": 7800
+    },
+    {
+      "epoch": 0.8238396624472574,
+      "grad_norm": 1.630586862564087,
+      "learning_rate": 0.00011443741410224173,
+      "loss": 5.5859,
+      "step": 7810
+    },
+    {
+      "epoch": 0.8248945147679325,
+      "grad_norm": 1.5931017398834229,
+      "learning_rate": 0.00011310803108208581,
+      "loss": 5.5851,
+      "step": 7820
+    },
+    {
+      "epoch": 0.8259493670886076,
+      "grad_norm": 1.3902634382247925,
+      "learning_rate": 0.00011178578474742687,
+      "loss": 5.583,
+      "step": 7830
+    },
+    {
+      "epoch": 0.8270042194092827,
+      "grad_norm": 1.143524408340454,
+      "learning_rate": 0.00011047068991468118,
+      "loss": 5.5763,
+      "step": 7840
+    },
+    {
+      "epoch": 0.8280590717299579,
+      "grad_norm": 1.3782490491867065,
+      "learning_rate": 0.00010916276132012818,
+      "loss": 5.5756,
+      "step": 7850
+    },
+    {
+      "epoch": 0.8291139240506329,
+      "grad_norm": 1.28400456905365,
+      "learning_rate": 0.00010786201361974646,
+      "loss": 5.5844,
+      "step": 7860
+    },
+    {
+      "epoch": 0.830168776371308,
+      "grad_norm": 1.151320457458496,
+      "learning_rate": 0.00010656846138904916,
+      "loss": 5.5782,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8312236286919831,
+      "grad_norm": 1.5421926975250244,
+      "learning_rate": 0.00010528211912292066,
+      "loss": 5.5799,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8322784810126582,
+      "grad_norm": 1.2956604957580566,
+      "learning_rate": 0.0001040030012354542,
+      "loss": 5.5862,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 1.9451608657836914,
+      "learning_rate": 0.00010273112205979012,
+      "loss": 5.5765,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8343881856540084,
+      "grad_norm": 1.9215521812438965,
+      "learning_rate": 0.00010146649584795575,
+      "loss": 5.5773,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8354430379746836,
+      "grad_norm": 1.446566104888916,
+      "learning_rate": 0.0001002091367707053,
+      "loss": 5.5821,
+      "step": 7920
+    },
+    {
+      "epoch": 0.8364978902953587,
+      "grad_norm": 1.6378275156021118,
+      "learning_rate": 9.895905891736118e-05,
+      "loss": 5.5804,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8375527426160337,
+      "grad_norm": 1.2002298831939697,
+      "learning_rate": 9.771627629565599e-05,
+      "loss": 5.5824,
+      "step": 7940
+    },
+    {
+      "epoch": 0.8386075949367089,
+      "grad_norm": 1.9541651010513306,
+      "learning_rate": 9.648080283157604e-05,
+      "loss": 5.573,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8396624472573839,
+      "grad_norm": 1.2114804983139038,
+      "learning_rate": 9.525265236920452e-05,
+      "loss": 5.5739,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8407172995780591,
+      "grad_norm": 1.4685115814208984,
+      "learning_rate": 9.40318386705673e-05,
+      "loss": 5.5771,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8417721518987342,
+      "grad_norm": 1.374232292175293,
+      "learning_rate": 9.281837541547791e-05,
+      "loss": 5.5693,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8428270042194093,
+      "grad_norm": 1.2028096914291382,
+      "learning_rate": 9.161227620138468e-05,
+      "loss": 5.5728,
+      "step": 7990
+    },
+    {
+      "epoch": 0.8438818565400844,
+      "grad_norm": 1.5220246315002441,
+      "learning_rate": 9.041355454321803e-05,
+      "loss": 5.5759,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8449367088607594,
+      "grad_norm": 1.3103171586990356,
+      "learning_rate": 8.92222238732397e-05,
+      "loss": 5.577,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8459915611814346,
+      "grad_norm": 1.4844565391540527,
+      "learning_rate": 8.803829754089138e-05,
+      "loss": 5.5783,
+      "step": 8020
+    },
+    {
+      "epoch": 0.8470464135021097,
+      "grad_norm": 1.171464204788208,
+      "learning_rate": 8.686178881264568e-05,
+      "loss": 5.5695,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8481012658227848,
+      "grad_norm": 1.2239173650741577,
+      "learning_rate": 8.569271087185756e-05,
+      "loss": 5.5761,
+      "step": 8040
+    },
+    {
+      "epoch": 0.8491561181434599,
+      "grad_norm": 1.3029521703720093,
+      "learning_rate": 8.453107681861616e-05,
+      "loss": 5.5746,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8502109704641351,
+      "grad_norm": 1.1450181007385254,
+      "learning_rate": 8.337689966959819e-05,
+      "loss": 5.5713,
+      "step": 8060
+    },
+    {
+      "epoch": 0.8512658227848101,
+      "grad_norm": 1.3673796653747559,
+      "learning_rate": 8.23445265795443e-05,
+      "loss": 5.5793,
+      "step": 8070
+    },
+    {
+      "epoch": 0.8523206751054853,
+      "grad_norm": 1.151382327079773,
+      "learning_rate": 8.120455311011473e-05,
+      "loss": 5.5744,
+      "step": 8080
+    },
+    {
+      "epoch": 0.8533755274261603,
+      "grad_norm": 1.4042388200759888,
+      "learning_rate": 8.007207382023102e-05,
+      "loss": 5.569,
+      "step": 8090
+    },
+    {
+      "epoch": 0.8544303797468354,
+      "grad_norm": 1.4187700748443604,
+      "learning_rate": 7.894710139987645e-05,
+      "loss": 5.5721,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8554852320675106,
+      "grad_norm": 1.3356019258499146,
+      "learning_rate": 7.782964845491666e-05,
+      "loss": 5.5749,
+      "step": 8110
+    },
+    {
+      "epoch": 0.8565400843881856,
+      "grad_norm": 1.1999540328979492,
+      "learning_rate": 7.67197275069573e-05,
+      "loss": 5.5698,
+      "step": 8120
+    },
+    {
+      "epoch": 0.8575949367088608,
+      "grad_norm": 1.2527334690093994,
+      "learning_rate": 7.561735099320463e-05,
+      "loss": 5.5705,
+      "step": 8130
+    },
+    {
+      "epoch": 0.8586497890295358,
+      "grad_norm": 1.030590295791626,
+      "learning_rate": 7.452253126632564e-05,
+      "loss": 5.576,
+      "step": 8140
+    },
+    {
+      "epoch": 0.859704641350211,
+      "grad_norm": 1.1422009468078613,
+      "learning_rate": 7.343528059431009e-05,
+      "loss": 5.5705,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8607594936708861,
+      "grad_norm": 1.1426135301589966,
+      "learning_rate": 7.235561116033265e-05,
+      "loss": 5.5682,
+      "step": 8160
+    },
+    {
+      "epoch": 0.8618143459915611,
+      "grad_norm": 1.4448994398117065,
+      "learning_rate": 7.128353506261631e-05,
+      "loss": 5.572,
+      "step": 8170
+    },
+    {
+      "epoch": 0.8628691983122363,
+      "grad_norm": 1.5425349473953247,
+      "learning_rate": 7.021906431429747e-05,
+      "loss": 5.5722,
+      "step": 8180
+    },
+    {
+      "epoch": 0.8639240506329114,
+      "grad_norm": 1.3306498527526855,
+      "learning_rate": 6.916221084329055e-05,
+      "loss": 5.5756,
+      "step": 8190
+    },
+    {
+      "epoch": 0.8649789029535865,
+      "grad_norm": 1.2649661302566528,
+      "learning_rate": 6.811298649215472e-05,
+      "loss": 5.5689,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8660337552742616,
+      "grad_norm": 1.5476899147033691,
+      "learning_rate": 6.707140301796122e-05,
+      "loss": 5.565,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8670886075949367,
+      "grad_norm": 1.0860280990600586,
+      "learning_rate": 6.603747209216135e-05,
+      "loss": 5.563,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8681434599156118,
+      "grad_norm": 1.0689903497695923,
+      "learning_rate": 6.501120530045593e-05,
+      "loss": 5.5733,
+      "step": 8230
+    },
+    {
+      "epoch": 0.869198312236287,
+      "grad_norm": 1.4661319255828857,
+      "learning_rate": 6.399261414266571e-05,
+      "loss": 5.5671,
+      "step": 8240
+    },
+    {
+      "epoch": 0.870253164556962,
+      "grad_norm": 1.7582958936691284,
+      "learning_rate": 6.298171003260194e-05,
+      "loss": 5.5735,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8713080168776371,
+      "grad_norm": 1.7814412117004395,
+      "learning_rate": 6.197850429793866e-05,
+      "loss": 5.5705,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8723628691983122,
+      "grad_norm": 1.1982461214065552,
+      "learning_rate": 6.0983008180086005e-05,
+      "loss": 5.5661,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8734177215189873,
+      "grad_norm": 1.1646335124969482,
+      "learning_rate": 5.999523283406405e-05,
+      "loss": 5.5637,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8744725738396625,
+      "grad_norm": 1.3502696752548218,
+      "learning_rate": 5.901518932837799e-05,
+      "loss": 5.5665,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8755274261603375,
+      "grad_norm": 2.2172207832336426,
+      "learning_rate": 5.804288864489366e-05,
+      "loss": 5.5687,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8765822784810127,
+      "grad_norm": 1.828156590461731,
+      "learning_rate": 5.707834167871512e-05,
+      "loss": 5.5656,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8776371308016878,
+      "grad_norm": 1.2343263626098633,
+      "learning_rate": 5.612155923806203e-05,
+      "loss": 5.5693,
+      "step": 8320
+    },
+    {
+      "epoch": 0.8786919831223629,
+      "grad_norm": 1.212476134300232,
+      "learning_rate": 5.517255204414889e-05,
+      "loss": 5.5674,
+      "step": 8330
+    },
+    {
+      "epoch": 0.879746835443038,
+      "grad_norm": 1.495962381362915,
+      "learning_rate": 5.423133073106457e-05,
+      "loss": 5.5643,
+      "step": 8340
+    },
+    {
+      "epoch": 0.880801687763713,
+      "grad_norm": 1.3730764389038086,
+      "learning_rate": 5.329790584565361e-05,
+      "loss": 5.5658,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8818565400843882,
+      "grad_norm": 1.5053300857543945,
+      "learning_rate": 5.237228784739739e-05,
+      "loss": 5.5599,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8829113924050633,
+      "grad_norm": 1.4753589630126953,
+      "learning_rate": 5.1454487108297924e-05,
+      "loss": 5.5689,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8839662447257384,
+      "grad_norm": 1.1202565431594849,
+      "learning_rate": 5.054451391276035e-05,
+      "loss": 5.5713,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8850210970464135,
+      "grad_norm": 1.2482728958129883,
+      "learning_rate": 4.9642378457478847e-05,
+      "loss": 5.5607,
+      "step": 8390
+    },
+    {
+      "epoch": 0.8860759493670886,
+      "grad_norm": 1.3180617094039917,
+      "learning_rate": 4.874809085132148e-05,
+      "loss": 5.5658,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8871308016877637,
+      "grad_norm": 1.1836440563201904,
+      "learning_rate": 4.7861661115217754e-05,
+      "loss": 5.5635,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8881856540084389,
+      "grad_norm": 1.26827073097229,
+      "learning_rate": 4.698309918204552e-05,
+      "loss": 5.5718,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8892405063291139,
+      "grad_norm": 1.436211347579956,
+      "learning_rate": 4.611241489652016e-05,
+      "loss": 5.5611,
+      "step": 8430
+    },
+    {
+      "epoch": 0.890295358649789,
+      "grad_norm": 1.2219692468643188,
+      "learning_rate": 4.524961801508456e-05,
+      "loss": 5.5683,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8913502109704642,
+      "grad_norm": 1.407319188117981,
+      "learning_rate": 4.439471820579885e-05,
+      "loss": 5.5648,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8924050632911392,
+      "grad_norm": 1.5786449909210205,
+      "learning_rate": 4.35477250482329e-05,
+      "loss": 5.5667,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8934599156118144,
+      "grad_norm": 1.2798808813095093,
+      "learning_rate": 4.2708648033358554e-05,
+      "loss": 5.5676,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8945147679324894,
+      "grad_norm": 1.3058149814605713,
+      "learning_rate": 4.1877496563443446e-05,
+      "loss": 5.566,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8955696202531646,
+      "grad_norm": 1.30825936794281,
+      "learning_rate": 4.105427995194566e-05,
+      "loss": 5.5631,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8966244725738397,
+      "grad_norm": 1.2759966850280762,
+      "learning_rate": 4.023900742340941e-05,
+      "loss": 5.557,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8976793248945147,
+      "grad_norm": 1.7265344858169556,
+      "learning_rate": 3.943168811336137e-05,
+      "loss": 5.5627,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8987341772151899,
+      "grad_norm": 1.1736171245574951,
+      "learning_rate": 3.863233106820857e-05,
+      "loss": 5.5679,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8997890295358649,
+      "grad_norm": 1.1865023374557495,
+      "learning_rate": 3.784094524513709e-05,
+      "loss": 5.5652,
+      "step": 8530
+    },
+    {
+      "epoch": 0.9008438818565401,
+      "grad_norm": 1.167667031288147,
+      "learning_rate": 3.705753951201146e-05,
+      "loss": 5.5648,
+      "step": 8540
+    },
+    {
+      "epoch": 0.9018987341772152,
+      "grad_norm": 1.143186330795288,
+      "learning_rate": 3.628212264727548e-05,
+      "loss": 5.5654,
+      "step": 8550
+    },
+    {
+      "epoch": 0.9029535864978903,
+      "grad_norm": 1.4849963188171387,
+      "learning_rate": 3.5514703339853656e-05,
+      "loss": 5.5611,
+      "step": 8560
+    },
+    {
+      "epoch": 0.9040084388185654,
+      "grad_norm": 1.2691038846969604,
+      "learning_rate": 3.475529018905416e-05,
+      "loss": 5.568,
+      "step": 8570
+    },
+    {
+      "epoch": 0.9050632911392406,
+      "grad_norm": 1.225518822669983,
+      "learning_rate": 3.400389170447218e-05,
+      "loss": 5.562,
+      "step": 8580
+    },
+    {
+      "epoch": 0.9061181434599156,
+      "grad_norm": 1.2211962938308716,
+      "learning_rate": 3.3260516305894526e-05,
+      "loss": 5.5602,
+      "step": 8590
+    },
+    {
+      "epoch": 0.9071729957805907,
+      "grad_norm": 1.1924623250961304,
+      "learning_rate": 3.2525172323205535e-05,
+      "loss": 5.5629,
+      "step": 8600
+    },
+    {
+      "epoch": 0.9082278481012658,
+      "grad_norm": 1.2955036163330078,
+      "learning_rate": 3.1797867996293663e-05,
+      "loss": 5.5647,
+      "step": 8610
+    },
+    {
+      "epoch": 0.9092827004219409,
+      "grad_norm": 1.3605698347091675,
+      "learning_rate": 3.107861147495891e-05,
+      "loss": 5.5604,
+      "step": 8620
+    },
+    {
+      "epoch": 0.9103375527426161,
+      "grad_norm": 1.6848132610321045,
+      "learning_rate": 3.0367410818821913e-05,
+      "loss": 5.5643,
+      "step": 8630
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 1.5561058521270752,
+      "learning_rate": 2.966427399723326e-05,
+      "loss": 5.5644,
+      "step": 8640
+    },
+    {
+      "epoch": 0.9124472573839663,
+      "grad_norm": 1.5387319326400757,
+      "learning_rate": 2.8969208889184335e-05,
+      "loss": 5.5651,
+      "step": 8650
+    },
+    {
+      "epoch": 0.9135021097046413,
+      "grad_norm": 1.3171087503433228,
+      "learning_rate": 2.828222328321911e-05,
+      "loss": 5.5558,
+      "step": 8660
+    },
+    {
+      "epoch": 0.9145569620253164,
+      "grad_norm": 1.406396508216858,
+      "learning_rate": 2.7603324877346653e-05,
+      "loss": 5.5639,
+      "step": 8670
+    },
+    {
+      "epoch": 0.9156118143459916,
+      "grad_norm": 1.1431360244750977,
+      "learning_rate": 2.6932521278955262e-05,
+      "loss": 5.5652,
+      "step": 8680
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "grad_norm": 1.1304011344909668,
+      "learning_rate": 2.626982000472655e-05,
+      "loss": 5.5673,
+      "step": 8690
+    },
+    {
+      "epoch": 0.9177215189873418,
+      "grad_norm": 1.3115731477737427,
+      "learning_rate": 2.561522848055217e-05,
+      "loss": 5.5666,
+      "step": 8700
+    },
+    {
+      "epoch": 0.9187763713080169,
+      "grad_norm": 1.1695328950881958,
+      "learning_rate": 2.4968754041449633e-05,
+      "loss": 5.5596,
+      "step": 8710
+    },
+    {
+      "epoch": 0.919831223628692,
+      "grad_norm": 1.3802136182785034,
+      "learning_rate": 2.43304039314807e-05,
+      "loss": 5.5637,
+      "step": 8720
+    },
+    {
+      "epoch": 0.9208860759493671,
+      "grad_norm": 1.1863781213760376,
+      "learning_rate": 2.3700185303670046e-05,
+      "loss": 5.5605,
+      "step": 8730
+    },
+    {
+      "epoch": 0.9219409282700421,
+      "grad_norm": 1.2241286039352417,
+      "learning_rate": 2.307810521992515e-05,
+      "loss": 5.5569,
+      "step": 8740
+    },
+    {
+      "epoch": 0.9229957805907173,
+      "grad_norm": 1.3637723922729492,
+      "learning_rate": 2.24641706509571e-05,
+      "loss": 5.5587,
+      "step": 8750
+    },
+    {
+      "epoch": 0.9240506329113924,
+      "grad_norm": 1.334944725036621,
+      "learning_rate": 2.185838847620242e-05,
+      "loss": 5.5516,
+      "step": 8760
+    },
+    {
+      "epoch": 0.9251054852320675,
+      "grad_norm": 1.0865918397903442,
+      "learning_rate": 2.1260765483746282e-05,
+      "loss": 5.561,
+      "step": 8770
+    },
+    {
+      "epoch": 0.9261603375527426,
+      "grad_norm": 1.2429355382919312,
+      "learning_rate": 2.0671308370246167e-05,
+      "loss": 5.5622,
+      "step": 8780
+    },
+    {
+      "epoch": 0.9272151898734177,
+      "grad_norm": 1.249382734298706,
+      "learning_rate": 2.009002374085675e-05,
+      "loss": 5.5608,
+      "step": 8790
+    },
+    {
+      "epoch": 0.9282700421940928,
+      "grad_norm": 1.4502249956130981,
+      "learning_rate": 1.9516918109156206e-05,
+      "loss": 5.5561,
+      "step": 8800
+    },
+    {
+      "epoch": 0.929324894514768,
+      "grad_norm": 1.2240568399429321,
+      "learning_rate": 1.8951997897072943e-05,
+      "loss": 5.5637,
+      "step": 8810
+    },
+    {
+      "epoch": 0.930379746835443,
+      "grad_norm": 1.3137434720993042,
+      "learning_rate": 1.8395269434813733e-05,
+      "loss": 5.5606,
+      "step": 8820
+    },
+    {
+      "epoch": 0.9314345991561181,
+      "grad_norm": 1.7774194478988647,
+      "learning_rate": 1.7846738960792945e-05,
+      "loss": 5.5604,
+      "step": 8830
+    },
+    {
+      "epoch": 0.9324894514767933,
+      "grad_norm": 1.2234671115875244,
+      "learning_rate": 1.7306412621562352e-05,
+      "loss": 5.5565,
+      "step": 8840
+    },
+    {
+      "epoch": 0.9335443037974683,
+      "grad_norm": 1.2479087114334106,
+      "learning_rate": 1.677429647174242e-05,
+      "loss": 5.5619,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9345991561181435,
+      "grad_norm": 1.1925565004348755,
+      "learning_rate": 1.6250396473954377e-05,
+      "loss": 5.5608,
+      "step": 8860
+    },
+    {
+      "epoch": 0.9356540084388185,
+      "grad_norm": 1.551577091217041,
+      "learning_rate": 1.573471849875352e-05,
+      "loss": 5.5673,
+      "step": 8870
+    },
+    {
+      "epoch": 0.9367088607594937,
+      "grad_norm": 1.6013810634613037,
+      "learning_rate": 1.522726832456342e-05,
+      "loss": 5.5577,
+      "step": 8880
+    },
+    {
+      "epoch": 0.9377637130801688,
+      "grad_norm": 1.263146996498108,
+      "learning_rate": 1.4728051637610902e-05,
+      "loss": 5.5677,
+      "step": 8890
+    },
+    {
+      "epoch": 0.9388185654008439,
+      "grad_norm": 1.3277114629745483,
+      "learning_rate": 1.4237074031862918e-05,
+      "loss": 5.561,
+      "step": 8900
+    },
+    {
+      "epoch": 0.939873417721519,
+      "grad_norm": 1.4118342399597168,
+      "learning_rate": 1.3754341008963194e-05,
+      "loss": 5.5578,
+      "step": 8910
+    },
+    {
+      "epoch": 0.9409282700421941,
+      "grad_norm": 1.7234737873077393,
+      "learning_rate": 1.327985797817094e-05,
+      "loss": 5.5615,
+      "step": 8920
+    },
+    {
+      "epoch": 0.9419831223628692,
+      "grad_norm": 1.6286661624908447,
+      "learning_rate": 1.2813630256300224e-05,
+      "loss": 5.5578,
+      "step": 8930
+    },
+    {
+      "epoch": 0.9430379746835443,
+      "grad_norm": 1.3984816074371338,
+      "learning_rate": 1.2355663067660283e-05,
+      "loss": 5.5587,
+      "step": 8940
+    },
+    {
+      "epoch": 0.9440928270042194,
+      "grad_norm": 1.582375168800354,
+      "learning_rate": 1.1905961543997147e-05,
+      "loss": 5.56,
+      "step": 8950
+    },
+    {
+      "epoch": 0.9451476793248945,
+      "grad_norm": 1.3011646270751953,
+      "learning_rate": 1.1464530724435928e-05,
+      "loss": 5.5597,
+      "step": 8960
+    },
+    {
+      "epoch": 0.9462025316455697,
+      "grad_norm": 1.3661301136016846,
+      "learning_rate": 1.1031375555424466e-05,
+      "loss": 5.5622,
+      "step": 8970
+    },
+    {
+      "epoch": 0.9472573839662447,
+      "grad_norm": 1.2528117895126343,
+      "learning_rate": 1.0606500890678023e-05,
+      "loss": 5.5595,
+      "step": 8980
+    },
+    {
+      "epoch": 0.9483122362869199,
+      "grad_norm": 1.4038883447647095,
+      "learning_rate": 1.0189911491124582e-05,
+      "loss": 5.5577,
+      "step": 8990
+    },
+    {
+      "epoch": 0.9493670886075949,
+      "grad_norm": 1.1988270282745361,
+      "learning_rate": 9.781612024851893e-06,
+      "loss": 5.5568,
+      "step": 9000
+    },
+    {
+      "epoch": 0.95042194092827,
+      "grad_norm": 1.5369138717651367,
+      "learning_rate": 9.381607067054764e-06,
+      "loss": 5.5603,
+      "step": 9010
+    },
+    {
+      "epoch": 0.9514767932489452,
+      "grad_norm": 1.5705552101135254,
+      "learning_rate": 8.989901099984016e-06,
+      "loss": 5.5608,
+      "step": 9020
+    },
+    {
+      "epoch": 0.9525316455696202,
+      "grad_norm": 1.390122413635254,
+      "learning_rate": 8.606498512896438e-06,
+      "loss": 5.5611,
+      "step": 9030
+    },
+    {
+      "epoch": 0.9535864978902954,
+      "grad_norm": 1.2336426973342896,
+      "learning_rate": 8.231403602005083e-06,
+      "loss": 5.5544,
+      "step": 9040
+    },
+    {
+      "epoch": 0.9546413502109705,
+      "grad_norm": 1.278863787651062,
+      "learning_rate": 7.864620570431635e-06,
+      "loss": 5.5598,
+      "step": 9050
+    },
+    {
+      "epoch": 0.9556962025316456,
+      "grad_norm": 1.2173552513122559,
+      "learning_rate": 7.506153528159032e-06,
+      "loss": 5.5666,
+      "step": 9060
+    },
+    {
+      "epoch": 0.9567510548523207,
+      "grad_norm": 1.1965709924697876,
+      "learning_rate": 7.1560064919855835e-06,
+      "loss": 5.5633,
+      "step": 9070
+    },
+    {
+      "epoch": 0.9578059071729957,
+      "grad_norm": 1.2677333354949951,
+      "learning_rate": 6.814183385479677e-06,
+      "loss": 5.5574,
+      "step": 9080
+    },
+    {
+      "epoch": 0.9588607594936709,
+      "grad_norm": 1.3301301002502441,
+      "learning_rate": 6.480688038936311e-06,
+      "loss": 5.5598,
+      "step": 9090
+    },
+    {
+      "epoch": 0.959915611814346,
+      "grad_norm": 1.0982439517974854,
+      "learning_rate": 6.155524189333461e-06,
+      "loss": 5.5565,
+      "step": 9100
+    },
+    {
+      "epoch": 0.9609704641350211,
+      "grad_norm": 1.2289735078811646,
+      "learning_rate": 5.838695480291034e-06,
+      "loss": 5.552,
+      "step": 9110
+    },
+    {
+      "epoch": 0.9620253164556962,
+      "grad_norm": 1.1614609956741333,
+      "learning_rate": 5.530205462029314e-06,
+      "loss": 5.5545,
+      "step": 9120
+    },
+    {
+      "epoch": 0.9630801687763713,
+      "grad_norm": 1.7835652828216553,
+      "learning_rate": 5.230057591329662e-06,
+      "loss": 5.5567,
+      "step": 9130
+    },
+    {
+      "epoch": 0.9641350210970464,
+      "grad_norm": 1.465887427330017,
+      "learning_rate": 4.938255231495464e-06,
+      "loss": 5.5657,
+      "step": 9140
+    },
+    {
+      "epoch": 0.9651898734177216,
+      "grad_norm": 1.1732157468795776,
+      "learning_rate": 4.654801652314577e-06,
+      "loss": 5.5609,
+      "step": 9150
+    },
+    {
+      "epoch": 0.9662447257383966,
+      "grad_norm": 1.0920121669769287,
+      "learning_rate": 4.37970003002286e-06,
+      "loss": 5.5583,
+      "step": 9160
+    },
+    {
+      "epoch": 0.9672995780590717,
+      "grad_norm": 1.2773975133895874,
+      "learning_rate": 4.112953447268364e-06,
+      "loss": 5.5628,
+      "step": 9170
+    },
+    {
+      "epoch": 0.9683544303797469,
+      "grad_norm": 1.1745500564575195,
+      "learning_rate": 3.8545648930767005e-06,
+      "loss": 5.5617,
+      "step": 9180
+    },
+    {
+      "epoch": 0.9694092827004219,
+      "grad_norm": 1.4241310358047485,
+      "learning_rate": 3.604537262817814e-06,
+      "loss": 5.5601,
+      "step": 9190
+    },
+    {
+      "epoch": 0.9704641350210971,
+      "grad_norm": 1.246065616607666,
+      "learning_rate": 3.362873358173424e-06,
+      "loss": 5.5547,
+      "step": 9200
+    },
+    {
+      "epoch": 0.9715189873417721,
+      "grad_norm": 1.347264289855957,
+      "learning_rate": 3.129575887105468e-06,
+      "loss": 5.5612,
+      "step": 9210
+    },
+    {
+      "epoch": 0.9725738396624473,
+      "grad_norm": 1.4465028047561646,
+      "learning_rate": 2.90464746382621e-06,
+      "loss": 5.5565,
+      "step": 9220
+    },
+    {
+      "epoch": 0.9736286919831224,
+      "grad_norm": 1.1796553134918213,
+      "learning_rate": 2.6880906087682622e-06,
+      "loss": 5.5585,
+      "step": 9230
+    },
+    {
+      "epoch": 0.9746835443037974,
+      "grad_norm": 1.2495790719985962,
+      "learning_rate": 2.4799077485571087e-06,
+      "loss": 5.5574,
+      "step": 9240
+    },
+    {
+      "epoch": 0.9757383966244726,
+      "grad_norm": 1.373209834098816,
+      "learning_rate": 2.2801012159832933e-06,
+      "loss": 5.5548,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9767932489451476,
+      "grad_norm": 1.1384923458099365,
+      "learning_rate": 2.0886732499764416e-06,
+      "loss": 5.5653,
+      "step": 9260
+    },
+    {
+      "epoch": 0.9778481012658228,
+      "grad_norm": 1.3330754041671753,
+      "learning_rate": 1.905625995580612e-06,
+      "loss": 5.5626,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9789029535864979,
+      "grad_norm": 1.2125582695007324,
+      "learning_rate": 1.7309615039294847e-06,
+      "loss": 5.5595,
+      "step": 9280
+    },
+    {
+      "epoch": 0.979957805907173,
+      "grad_norm": 1.2188507318496704,
+      "learning_rate": 1.5646817322240436e-06,
+      "loss": 5.5602,
+      "step": 9290
+    },
+    {
+      "epoch": 0.9810126582278481,
+      "grad_norm": 1.1631706953048706,
+      "learning_rate": 1.4067885437103467e-06,
+      "loss": 5.5612,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9820675105485233,
+      "grad_norm": 1.4485857486724854,
+      "learning_rate": 1.2572837076586241e-06,
+      "loss": 5.5599,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9831223628691983,
+      "grad_norm": 1.490790843963623,
+      "learning_rate": 1.1161688993435449e-06,
+      "loss": 5.5603,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9841772151898734,
+      "grad_norm": 1.2880841493606567,
+      "learning_rate": 9.834457000255647e-07,
+      "loss": 5.5553,
+      "step": 9330
+    },
+    {
+      "epoch": 0.9852320675105485,
+      "grad_norm": 1.1686524152755737,
+      "learning_rate": 8.591155969327746e-07,
+      "loss": 5.567,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9862869198312236,
+      "grad_norm": 1.5405880212783813,
+      "learning_rate": 7.431799832448294e-07,
+      "loss": 5.5587,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9873417721518988,
+      "grad_norm": 1.1483728885650635,
+      "learning_rate": 6.356401580767945e-07,
+      "loss": 5.5612,
+      "step": 9360
+    },
+    {
+      "epoch": 0.9883966244725738,
+      "grad_norm": 1.1651448011398315,
+      "learning_rate": 5.364973264649908e-07,
+      "loss": 5.5686,
+      "step": 9370
+    },
+    {
+      "epoch": 0.989451476793249,
+      "grad_norm": 1.5726734399795532,
+      "learning_rate": 4.457525993531719e-07,
+      "loss": 5.5599,
+      "step": 9380
+    },
+    {
+      "epoch": 0.990506329113924,
+      "grad_norm": 1.2543110847473145,
+      "learning_rate": 3.6340699358036743e-07,
+      "loss": 5.5561,
+      "step": 9390
+    },
+    {
+      "epoch": 0.9915611814345991,
+      "grad_norm": 1.2599172592163086,
+      "learning_rate": 2.8946143186930896e-07,
+      "loss": 5.5618,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9926160337552743,
+      "grad_norm": 1.1586952209472656,
+      "learning_rate": 2.2391674281610486e-07,
+      "loss": 5.549,
+      "step": 9410
+    },
+    {
+      "epoch": 0.9936708860759493,
+      "grad_norm": 1.4774086475372314,
+      "learning_rate": 1.6677366088099777e-07,
+      "loss": 5.558,
+      "step": 9420
+    },
+    {
+      "epoch": 0.9947257383966245,
+      "grad_norm": 1.1519263982772827,
+      "learning_rate": 1.1803282638020441e-07,
+      "loss": 5.5649,
+      "step": 9430
+    },
+    {
+      "epoch": 0.9957805907172996,
+      "grad_norm": 1.2013318538665771,
+      "learning_rate": 7.769478547842157e-08,
+      "loss": 5.5609,
+      "step": 9440
+    },
+    {
+      "epoch": 0.9968354430379747,
+      "grad_norm": 1.3563827276229858,
+      "learning_rate": 4.575999018316401e-08,
+      "loss": 5.5605,
+      "step": 9450
+    },
+    {
+      "epoch": 0.9978902953586498,
+      "grad_norm": 1.2778115272521973,
+      "learning_rate": 2.2228798339435363e-08,
+      "loss": 5.5579,
+      "step": 9460
+    },
+    {
+      "epoch": 0.9989451476793249,
+      "grad_norm": 1.7342040538787842,
+      "learning_rate": 7.101473625648058e-09,
+      "loss": 5.5595,
+      "step": 9470
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.935343027114868,
+      "learning_rate": 3.781855510420762e-10,
+      "loss": 5.5656,
+      "step": 9480
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9480,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.832308198648013e+16,
+  "train_batch_size": 1024,
+  "trial_name": null,
+  "trial_params": null
+}