diff --git "a/g2_3_plan_oldcode_newcode/1000M/trainer_state.json" "b/g2_3_plan_oldcode_newcode/1000M/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/g2_3_plan_oldcode_newcode/1000M/trainer_state.json"
@@ -0,0 +1,7876 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8515672550213025,
+  "eval_steps": 500,
+  "global_step": 11193,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0007608034083992696,
+      "grad_norm": 766.052521836789,
+      "learning_rate": 1.1061797752808989e-06,
+      "loss": 2.887490081787109,
+      "step": 10
+    },
+    {
+      "epoch": 0.0015216068167985392,
+      "grad_norm": 219.21834016884335,
+      "learning_rate": 2.2241573033707866e-06,
+      "loss": 1.9272476196289063,
+      "step": 20
+    },
+    {
+      "epoch": 0.002282410225197809,
+      "grad_norm": 697.9334513096925,
+      "learning_rate": 3.342134831460675e-06,
+      "loss": 1.0680903434753417,
+      "step": 30
+    },
+    {
+      "epoch": 0.0030432136335970784,
+      "grad_norm": 84.53394094934715,
+      "learning_rate": 4.4601123595505624e-06,
+      "loss": 0.8754222869873047,
+      "step": 40
+    },
+    {
+      "epoch": 0.003804017041996348,
+      "grad_norm": 69.64463705089611,
+      "learning_rate": 5.5780898876404495e-06,
+      "loss": 0.8158416748046875,
+      "step": 50
+    },
+    {
+      "epoch": 0.004564820450395618,
+      "grad_norm": 28.95668284276755,
+      "learning_rate": 6.6960674157303374e-06,
+      "loss": 0.7302873134613037,
+      "step": 60
+    },
+    {
+      "epoch": 0.0053256238587948874,
+      "grad_norm": 54.00260757854784,
+      "learning_rate": 7.814044943820226e-06,
+      "loss": 0.7970684051513672,
+      "step": 70
+    },
+    {
+      "epoch": 0.006086427267194157,
+      "grad_norm": 292.96231799204514,
+      "learning_rate": 8.932022471910112e-06,
+      "loss": 0.7267954826354981,
+      "step": 80
+    },
+    {
+      "epoch": 0.006847230675593427,
+      "grad_norm": 726.0873753124116,
+      "learning_rate": 1.005e-05,
+      "loss": 0.7285655975341797,
+      "step": 90
+    },
+    {
+      "epoch": 0.007608034083992696,
+      "grad_norm": 27.829181455179807,
+      "learning_rate": 1.1167977528089888e-05,
+      "loss": 0.8418128013610839,
+      "step": 100
+    },
+    {
+      "epoch": 0.008368837492391966,
+      "grad_norm": 38.50424143410114,
+      "learning_rate": 1.2285955056179776e-05,
+      "loss": 0.7732617378234863,
+      "step": 110
+    },
+    {
+      "epoch": 0.009129640900791236,
+      "grad_norm": 38.005830230959454,
+      "learning_rate": 1.3403932584269664e-05,
+      "loss": 0.7485276222229004,
+      "step": 120
+    },
+    {
+      "epoch": 0.009890444309190505,
+      "grad_norm": 239.32716231811452,
+      "learning_rate": 1.4521910112359552e-05,
+      "loss": 0.7527177333831787,
+      "step": 130
+    },
+    {
+      "epoch": 0.010651247717589775,
+      "grad_norm": 6.455966289941813,
+      "learning_rate": 1.563988764044944e-05,
+      "loss": 0.805815601348877,
+      "step": 140
+    },
+    {
+      "epoch": 0.011412051125989045,
+      "grad_norm": 18.693317281807552,
+      "learning_rate": 1.6757865168539328e-05,
+      "loss": 0.7609236717224122,
+      "step": 150
+    },
+    {
+      "epoch": 0.012172854534388313,
+      "grad_norm": 11.963698194011737,
+      "learning_rate": 1.7875842696629214e-05,
+      "loss": 0.7418763637542725,
+      "step": 160
+    },
+    {
+      "epoch": 0.012933657942787584,
+      "grad_norm": 6.958042928688908,
+      "learning_rate": 1.89938202247191e-05,
+      "loss": 0.7683084011077881,
+      "step": 170
+    },
+    {
+      "epoch": 0.013694461351186854,
+      "grad_norm": 11.571845696402532,
+      "learning_rate": 2e-05,
+      "loss": 0.7141913890838623,
+      "step": 180
+    },
+    {
+      "epoch": 0.014455264759586122,
+      "grad_norm": 8.273402278038633,
+      "learning_rate": 2e-05,
+      "loss": 0.7291435241699219,
+      "step": 190
+    },
+    {
+      "epoch": 0.015216068167985392,
+      "grad_norm": 1.8068788190209832,
+      "learning_rate": 2e-05,
+      "loss": 0.7061244010925293,
+      "step": 200
+    },
+    {
+      "epoch": 0.015976871576384662,
+      "grad_norm": 3.1806280744048605,
+      "learning_rate": 2e-05,
+      "loss": 0.7077610969543457,
+      "step": 210
+    },
+    {
+      "epoch": 0.016737674984783932,
+      "grad_norm": 280.189044907266,
+      "learning_rate": 2e-05,
+      "loss": 0.6421300411224365,
+      "step": 220
+    },
+    {
+      "epoch": 0.017498478393183203,
+      "grad_norm": 2.55416598370708,
+      "learning_rate": 2e-05,
+      "loss": 0.6898053169250489,
+      "step": 230
+    },
+    {
+      "epoch": 0.018259281801582473,
+      "grad_norm": 7.553555757256519,
+      "learning_rate": 2e-05,
+      "loss": 0.6969669342041016,
+      "step": 240
+    },
+    {
+      "epoch": 0.01902008520998174,
+      "grad_norm": 477.54050832124796,
+      "learning_rate": 2e-05,
+      "loss": 0.764107608795166,
+      "step": 250
+    },
+    {
+      "epoch": 0.01978088861838101,
+      "grad_norm": 6.159675954810047,
+      "learning_rate": 2e-05,
+      "loss": 0.9087591171264648,
+      "step": 260
+    },
+    {
+      "epoch": 0.02054169202678028,
+      "grad_norm": 14.248123008291211,
+      "learning_rate": 2e-05,
+      "loss": 0.7264697551727295,
+      "step": 270
+    },
+    {
+      "epoch": 0.02130249543517955,
+      "grad_norm": 15.77402781933041,
+      "learning_rate": 2e-05,
+      "loss": 0.7285306453704834,
+      "step": 280
+    },
+    {
+      "epoch": 0.02206329884357882,
+      "grad_norm": 24.137047826837957,
+      "learning_rate": 2e-05,
+      "loss": 0.6912951469421387,
+      "step": 290
+    },
+    {
+      "epoch": 0.02282410225197809,
+      "grad_norm": 2.3148760857715445,
+      "learning_rate": 2e-05,
+      "loss": 0.6992810726165771,
+      "step": 300
+    },
+    {
+      "epoch": 0.02358490566037736,
+      "grad_norm": 126.46425865282905,
+      "learning_rate": 2e-05,
+      "loss": 0.7007718563079834,
+      "step": 310
+    },
+    {
+      "epoch": 0.024345709068776627,
+      "grad_norm": 2.6039447021239726,
+      "learning_rate": 2e-05,
+      "loss": 0.6786778926849365,
+      "step": 320
+    },
+    {
+      "epoch": 0.025106512477175897,
+      "grad_norm": 2.392207141310159,
+      "learning_rate": 2e-05,
+      "loss": 0.669903564453125,
+      "step": 330
+    },
+    {
+      "epoch": 0.025867315885575167,
+      "grad_norm": 9.247678789298908,
+      "learning_rate": 2e-05,
+      "loss": 0.6987538814544678,
+      "step": 340
+    },
+    {
+      "epoch": 0.026628119293974437,
+      "grad_norm": 6.464711837428858,
+      "learning_rate": 2e-05,
+      "loss": 0.6893903255462647,
+      "step": 350
+    },
+    {
+      "epoch": 0.027388922702373707,
+      "grad_norm": 5.1880207521662705,
+      "learning_rate": 2e-05,
+      "loss": 0.6841831207275391,
+      "step": 360
+    },
+    {
+      "epoch": 0.028149726110772978,
+      "grad_norm": 50.12816854251406,
+      "learning_rate": 2e-05,
+      "loss": 0.7154038906097412,
+      "step": 370
+    },
+    {
+      "epoch": 0.028910529519172244,
+      "grad_norm": 3.0117477081232598,
+      "learning_rate": 2e-05,
+      "loss": 0.7240267753601074,
+      "step": 380
+    },
+    {
+      "epoch": 0.029671332927571514,
+      "grad_norm": 13.836877796796628,
+      "learning_rate": 2e-05,
+      "loss": 0.7010954856872559,
+      "step": 390
+    },
+    {
+      "epoch": 0.030432136335970784,
+      "grad_norm": 2.006973973891417,
+      "learning_rate": 2e-05,
+      "loss": 0.7020359039306641,
+      "step": 400
+    },
+    {
+      "epoch": 0.031192939744370055,
+      "grad_norm": 27.591581159248626,
+      "learning_rate": 2e-05,
+      "loss": 0.6682147979736328,
+      "step": 410
+    },
+    {
+      "epoch": 0.031953743152769325,
+      "grad_norm": 19.791706144588765,
+      "learning_rate": 2e-05,
+      "loss": 0.6723065853118897,
+      "step": 420
+    },
+    {
+      "epoch": 0.032714546561168595,
+      "grad_norm": 2.0177081819744553,
+      "learning_rate": 2e-05,
+      "loss": 0.7360958099365235,
+      "step": 430
+    },
+    {
+      "epoch": 0.033475349969567865,
+      "grad_norm": 2.188210526838186,
+      "learning_rate": 2e-05,
+      "loss": 0.6793450832366943,
+      "step": 440
+    },
+    {
+      "epoch": 0.034236153377967135,
+      "grad_norm": 2.5384186698997313,
+      "learning_rate": 2e-05,
+      "loss": 0.6992023468017579,
+      "step": 450
+    },
+    {
+      "epoch": 0.034996956786366405,
+      "grad_norm": 1.5352126431908675,
+      "learning_rate": 2e-05,
+      "loss": 0.6813825130462646,
+      "step": 460
+    },
+    {
+      "epoch": 0.035757760194765675,
+      "grad_norm": 5.422494483821066,
+      "learning_rate": 2e-05,
+      "loss": 0.6625070095062255,
+      "step": 470
+    },
+    {
+      "epoch": 0.036518563603164945,
+      "grad_norm": 2.164648016182954,
+      "learning_rate": 2e-05,
+      "loss": 0.6671664237976074,
+      "step": 480
+    },
+    {
+      "epoch": 0.03727936701156421,
+      "grad_norm": 5.128707982424324,
+      "learning_rate": 2e-05,
+      "loss": 0.6608222007751465,
+      "step": 490
+    },
+    {
+      "epoch": 0.03804017041996348,
+      "grad_norm": 2.6188100469545605,
+      "learning_rate": 2e-05,
+      "loss": 0.6627635002136231,
+      "step": 500
+    },
+    {
+      "epoch": 0.03880097382836275,
+      "grad_norm": 2.2690007433780646,
+      "learning_rate": 2e-05,
+      "loss": 0.7100722312927246,
+      "step": 510
+    },
+    {
+      "epoch": 0.03956177723676202,
+      "grad_norm": 1.6955425896534853,
+      "learning_rate": 2e-05,
+      "loss": 0.7041080474853516,
+      "step": 520
+    },
+    {
+      "epoch": 0.04032258064516129,
+      "grad_norm": 1.7481547195068348,
+      "learning_rate": 2e-05,
+      "loss": 0.6757216453552246,
+      "step": 530
+    },
+    {
+      "epoch": 0.04108338405356056,
+      "grad_norm": 1.6344561599660914,
+      "learning_rate": 2e-05,
+      "loss": 0.6713863372802734,
+      "step": 540
+    },
+    {
+      "epoch": 0.04184418746195983,
+      "grad_norm": 3.2078736781026254,
+      "learning_rate": 2e-05,
+      "loss": 0.6618888854980469,
+      "step": 550
+    },
+    {
+      "epoch": 0.0426049908703591,
+      "grad_norm": 1.8379942627724124,
+      "learning_rate": 2e-05,
+      "loss": 0.7351431846618652,
+      "step": 560
+    },
+    {
+      "epoch": 0.04336579427875837,
+      "grad_norm": 1.5888838232838909,
+      "learning_rate": 2e-05,
+      "loss": 0.6826735496520996,
+      "step": 570
+    },
+    {
+      "epoch": 0.04412659768715764,
+      "grad_norm": 2.0735189856680214,
+      "learning_rate": 2e-05,
+      "loss": 0.6519746780395508,
+      "step": 580
+    },
+    {
+      "epoch": 0.04488740109555691,
+      "grad_norm": 1.6672247701461744,
+      "learning_rate": 2e-05,
+      "loss": 0.6651469707489014,
+      "step": 590
+    },
+    {
+      "epoch": 0.04564820450395618,
+      "grad_norm": 2.331882937366215,
+      "learning_rate": 2e-05,
+      "loss": 0.6785001277923584,
+      "step": 600
+    },
+    {
+      "epoch": 0.04640900791235545,
+      "grad_norm": 1.5170691943378647,
+      "learning_rate": 2e-05,
+      "loss": 0.6592602729797363,
+      "step": 610
+    },
+    {
+      "epoch": 0.04716981132075472,
+      "grad_norm": 1.3648878825695145,
+      "learning_rate": 2e-05,
+      "loss": 0.6565546989440918,
+      "step": 620
+    },
+    {
+      "epoch": 0.047930614729153984,
+      "grad_norm": 1.6864881230555355,
+      "learning_rate": 2e-05,
+      "loss": 0.6754720211029053,
+      "step": 630
+    },
+    {
+      "epoch": 0.048691418137553254,
+      "grad_norm": 1.6332391825289212,
+      "learning_rate": 2e-05,
+      "loss": 0.6423514366149903,
+      "step": 640
+    },
+    {
+      "epoch": 0.049452221545952524,
+      "grad_norm": 2.7059106535186572,
+      "learning_rate": 2e-05,
+      "loss": 0.661496639251709,
+      "step": 650
+    },
+    {
+      "epoch": 0.050213024954351794,
+      "grad_norm": 2.896510703741966,
+      "learning_rate": 2e-05,
+      "loss": 0.6534607887268067,
+      "step": 660
+    },
+    {
+      "epoch": 0.050973828362751064,
+      "grad_norm": 2.2363116679018598,
+      "learning_rate": 2e-05,
+      "loss": 0.6489164352416992,
+      "step": 670
+    },
+    {
+      "epoch": 0.051734631771150334,
+      "grad_norm": 4.461890971477879,
+      "learning_rate": 2e-05,
+      "loss": 0.6854209899902344,
+      "step": 680
+    },
+    {
+      "epoch": 0.052495435179549604,
+      "grad_norm": 2.916869559890112,
+      "learning_rate": 2e-05,
+      "loss": 0.6794282913208007,
+      "step": 690
+    },
+    {
+      "epoch": 0.053256238587948874,
+      "grad_norm": 6.088998851699956,
+      "learning_rate": 2e-05,
+      "loss": 0.7020540237426758,
+      "step": 700
+    },
+    {
+      "epoch": 0.054017041996348145,
+      "grad_norm": 2.4051352056906787,
+      "learning_rate": 2e-05,
+      "loss": 0.7063278198242188,
+      "step": 710
+    },
+    {
+      "epoch": 0.054777845404747415,
+      "grad_norm": 3.094845534842579,
+      "learning_rate": 2e-05,
+      "loss": 0.6743041515350342,
+      "step": 720
+    },
+    {
+      "epoch": 0.055538648813146685,
+      "grad_norm": 1.300597375588879,
+      "learning_rate": 2e-05,
+      "loss": 0.6632575035095215,
+      "step": 730
+    },
+    {
+      "epoch": 0.056299452221545955,
+      "grad_norm": 3.279701114868982,
+      "learning_rate": 2e-05,
+      "loss": 0.6596969127655029,
+      "step": 740
+    },
+    {
+      "epoch": 0.057060255629945225,
+      "grad_norm": 1.6596836234521832,
+      "learning_rate": 2e-05,
+      "loss": 0.6648564815521241,
+      "step": 750
+    },
+    {
+      "epoch": 0.05782105903834449,
+      "grad_norm": 3.3066247075012956,
+      "learning_rate": 2e-05,
+      "loss": 0.6650107383728028,
+      "step": 760
+    },
+    {
+      "epoch": 0.05858186244674376,
+      "grad_norm": 3.929474692350062,
+      "learning_rate": 2e-05,
+      "loss": 0.6553778648376465,
+      "step": 770
+    },
+    {
+      "epoch": 0.05934266585514303,
+      "grad_norm": 2.518127583230988,
+      "learning_rate": 2e-05,
+      "loss": 0.690962553024292,
+      "step": 780
+    },
+    {
+      "epoch": 0.0601034692635423,
+      "grad_norm": 1.2367931748447925,
+      "learning_rate": 2e-05,
+      "loss": 0.6664523601531982,
+      "step": 790
+    },
+    {
+      "epoch": 0.06086427267194157,
+      "grad_norm": 4.142969089978007,
+      "learning_rate": 2e-05,
+      "loss": 0.6557377815246582,
+      "step": 800
+    },
+    {
+      "epoch": 0.06162507608034084,
+      "grad_norm": 1.4239753399947923,
+      "learning_rate": 2e-05,
+      "loss": 0.655963659286499,
+      "step": 810
+    },
+    {
+      "epoch": 0.06238587948874011,
+      "grad_norm": 1.100863203530169,
+      "learning_rate": 2e-05,
+      "loss": 0.6471473693847656,
+      "step": 820
+    },
+    {
+      "epoch": 0.06314668289713937,
+      "grad_norm": 1.232804036614322,
+      "learning_rate": 2e-05,
+      "loss": 0.6807149410247803,
+      "step": 830
+    },
+    {
+      "epoch": 0.06390748630553865,
+      "grad_norm": 1.2630476712708762,
+      "learning_rate": 2e-05,
+      "loss": 0.6472922325134277,
+      "step": 840
+    },
+    {
+      "epoch": 0.06466828971393791,
+      "grad_norm": 3.1776755109142627,
+      "learning_rate": 2e-05,
+      "loss": 0.6446826934814454,
+      "step": 850
+    },
+    {
+      "epoch": 0.06542909312233719,
+      "grad_norm": 75.45896872532501,
+      "learning_rate": 2e-05,
+      "loss": 0.6587884902954102,
+      "step": 860
+    },
+    {
+      "epoch": 0.06618989653073645,
+      "grad_norm": 1.9742341717753686,
+      "learning_rate": 2e-05,
+      "loss": 0.6437560081481933,
+      "step": 870
+    },
+    {
+      "epoch": 0.06695069993913573,
+      "grad_norm": 2.964541109902538,
+      "learning_rate": 2e-05,
+      "loss": 0.6286462783813477,
+      "step": 880
+    },
+    {
+      "epoch": 0.067711503347535,
+      "grad_norm": 3.7026446921551486,
+      "learning_rate": 2e-05,
+      "loss": 0.6568585872650147,
+      "step": 890
+    },
+    {
+      "epoch": 0.06847230675593427,
+      "grad_norm": 9.317395483241429,
+      "learning_rate": 2e-05,
+      "loss": 0.6981778144836426,
+      "step": 900
+    },
+    {
+      "epoch": 0.06923311016433353,
+      "grad_norm": 1.0948327360661168,
+      "learning_rate": 2e-05,
+      "loss": 0.6652321815490723,
+      "step": 910
+    },
+    {
+      "epoch": 0.06999391357273281,
+      "grad_norm": 1.0354718206717286,
+      "learning_rate": 2e-05,
+      "loss": 0.6647524833679199,
+      "step": 920
+    },
+    {
+      "epoch": 0.07075471698113207,
+      "grad_norm": 1.0304901518571212,
+      "learning_rate": 2e-05,
+      "loss": 0.6425517559051513,
+      "step": 930
+    },
+    {
+      "epoch": 0.07151552038953135,
+      "grad_norm": 1.5686213831558329,
+      "learning_rate": 2e-05,
+      "loss": 0.6413394927978515,
+      "step": 940
+    },
+    {
+      "epoch": 0.07227632379793061,
+      "grad_norm": 1.7897054437110276,
+      "learning_rate": 2e-05,
+      "loss": 0.6535541534423828,
+      "step": 950
+    },
+    {
+      "epoch": 0.07303712720632989,
+      "grad_norm": 0.9359855078368583,
+      "learning_rate": 2e-05,
+      "loss": 0.639809513092041,
+      "step": 960
+    },
+    {
+      "epoch": 0.07379793061472915,
+      "grad_norm": 1.512951285597644,
+      "learning_rate": 2e-05,
+      "loss": 0.6573833465576172,
+      "step": 970
+    },
+    {
+      "epoch": 0.07455873402312842,
+      "grad_norm": 2.359700069484156,
+      "learning_rate": 2e-05,
+      "loss": 0.6440255165100097,
+      "step": 980
+    },
+    {
+      "epoch": 0.0753195374315277,
+      "grad_norm": 1.7130709676637312,
+      "learning_rate": 2e-05,
+      "loss": 0.6478070259094239,
+      "step": 990
+    },
+    {
+      "epoch": 0.07608034083992696,
+      "grad_norm": 1.3177759770170199,
+      "learning_rate": 2e-05,
+      "loss": 0.6684154510498047,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07684114424832623,
+      "grad_norm": 4.135754705114454,
+      "learning_rate": 2e-05,
+      "loss": 0.6643803596496582,
+      "step": 1010
+    },
+    {
+      "epoch": 0.0776019476567255,
+      "grad_norm": 1.1349353267975015,
+      "learning_rate": 2e-05,
+      "loss": 0.6595885276794433,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07836275106512477,
+      "grad_norm": 1.0289702425560623,
+      "learning_rate": 2e-05,
+      "loss": 0.6449689388275146,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07912355447352404,
+      "grad_norm": 1.1123711190266128,
+      "learning_rate": 2e-05,
+      "loss": 0.6514264583587647,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07988435788192332,
+      "grad_norm": 1.1479004223710685,
+      "learning_rate": 2e-05,
+      "loss": 0.6790945053100585,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 1.3082960682160845,
+      "learning_rate": 2e-05,
+      "loss": 0.6506200313568116,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08140596469872186,
+      "grad_norm": 2.106064887485788,
+      "learning_rate": 2e-05,
+      "loss": 0.6225467681884765,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08216676810712112,
+      "grad_norm": 3.4175935958057004,
+      "learning_rate": 2e-05,
+      "loss": 0.6518382549285888,
+      "step": 1080
+    },
+    {
+      "epoch": 0.0829275715155204,
+      "grad_norm": 1.1113744031391726,
+      "learning_rate": 2e-05,
+      "loss": 0.6582504272460937,
+      "step": 1090
+    },
+    {
+      "epoch": 0.08368837492391966,
+      "grad_norm": 0.9903911423489237,
+      "learning_rate": 2e-05,
+      "loss": 0.6358994960784912,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08444917833231892,
+      "grad_norm": 0.8851123136905361,
+      "learning_rate": 2e-05,
+      "loss": 0.6215919494628906,
+      "step": 1110
+    },
+    {
+      "epoch": 0.0852099817407182,
+      "grad_norm": 0.8853327646542226,
+      "learning_rate": 2e-05,
+      "loss": 0.6268770217895507,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08597078514911746,
+      "grad_norm": 0.902800886984458,
+      "learning_rate": 2e-05,
+      "loss": 0.6309802055358886,
+      "step": 1130
+    },
+    {
+      "epoch": 0.08673158855751674,
+      "grad_norm": 10.357475980832238,
+      "learning_rate": 2e-05,
+      "loss": 0.6268141746520997,
+      "step": 1140
+    },
+    {
+      "epoch": 0.087492391965916,
+      "grad_norm": 4.536094472750066,
+      "learning_rate": 2e-05,
+      "loss": 0.6583127975463867,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08825319537431528,
+      "grad_norm": 3.161273337598724,
+      "learning_rate": 2e-05,
+      "loss": 0.6397143363952636,
+      "step": 1160
+    },
+    {
+      "epoch": 0.08901399878271454,
+      "grad_norm": 1.9628805920204075,
+      "learning_rate": 2e-05,
+      "loss": 0.6532254695892334,
+      "step": 1170
+    },
+    {
+      "epoch": 0.08977480219111382,
+      "grad_norm": 1.3058782788410006,
+      "learning_rate": 2e-05,
+      "loss": 0.6381869316101074,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09053560559951308,
+      "grad_norm": 10.319238456330314,
+      "learning_rate": 2e-05,
+      "loss": 0.6643379211425782,
+      "step": 1190
+    },
+    {
+      "epoch": 0.09129640900791236,
+      "grad_norm": 8.807866849572596,
+      "learning_rate": 2e-05,
+      "loss": 0.6664113521575927,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09205721241631162,
+      "grad_norm": 1.3436238512439613,
+      "learning_rate": 2e-05,
+      "loss": 0.6490108013153076,
+      "step": 1210
+    },
+    {
+      "epoch": 0.0928180158247109,
+      "grad_norm": 4.3500091772799,
+      "learning_rate": 2e-05,
+      "loss": 0.6590385437011719,
+      "step": 1220
+    },
+    {
+      "epoch": 0.09357881923311016,
+      "grad_norm": 1.0603429372503828,
+      "learning_rate": 2e-05,
+      "loss": 0.6429415702819824,
+      "step": 1230
+    },
+    {
+      "epoch": 0.09433962264150944,
+      "grad_norm": 1.610503564476208,
+      "learning_rate": 2e-05,
+      "loss": 0.6506116390228271,
+      "step": 1240
+    },
+    {
+      "epoch": 0.0951004260499087,
+      "grad_norm": 2.125651897332447,
+      "learning_rate": 2e-05,
+      "loss": 0.6554283142089844,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09586122945830797,
+      "grad_norm": 1.5217172296899981,
+      "learning_rate": 2e-05,
+      "loss": 0.6519949913024903,
+      "step": 1260
+    },
+    {
+      "epoch": 0.09662203286670724,
+      "grad_norm": 1.430228287443949,
+      "learning_rate": 2e-05,
+      "loss": 0.6444644451141357,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09738283627510651,
+      "grad_norm": 1.0951036531124672,
+      "learning_rate": 2e-05,
+      "loss": 0.6609076499938965,
+      "step": 1280
+    },
+    {
+      "epoch": 0.09814363968350578,
+      "grad_norm": 1.897316036200272,
+      "learning_rate": 2e-05,
+      "loss": 0.6357399940490722,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09890444309190505,
+      "grad_norm": 9.680273322175012,
+      "learning_rate": 2e-05,
+      "loss": 0.6786548614501953,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09966524650030432,
+      "grad_norm": 3.280416382171216,
+      "learning_rate": 2e-05,
+      "loss": 0.6191361904144287,
+      "step": 1310
+    },
+    {
+      "epoch": 0.10042604990870359,
+      "grad_norm": 5.004121996330226,
+      "learning_rate": 2e-05,
+      "loss": 0.6557157516479493,
+      "step": 1320
+    },
+    {
+      "epoch": 0.10118685331710287,
+      "grad_norm": 2.4488618272689857,
+      "learning_rate": 2e-05,
+      "loss": 0.6534897804260253,
+      "step": 1330
+    },
+    {
+      "epoch": 0.10194765672550213,
+      "grad_norm": 3.949040164888651,
+      "learning_rate": 2e-05,
+      "loss": 0.6422089576721192,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1027084601339014,
+      "grad_norm": 1.4977923859383218,
+      "learning_rate": 2e-05,
+      "loss": 0.6565611839294434,
+      "step": 1350
+    },
+    {
+      "epoch": 0.10346926354230067,
+      "grad_norm": 3.835777101408433,
+      "learning_rate": 2e-05,
+      "loss": 0.6588279724121093,
+      "step": 1360
+    },
+    {
+      "epoch": 0.10423006695069995,
+      "grad_norm": 2.0073417233896174,
+      "learning_rate": 2e-05,
+      "loss": 0.6190773010253906,
+      "step": 1370
+    },
+    {
+      "epoch": 0.10499087035909921,
+      "grad_norm": 2.180412133473381,
+      "learning_rate": 2e-05,
+      "loss": 0.6264517784118653,
+      "step": 1380
+    },
+    {
+      "epoch": 0.10575167376749847,
+      "grad_norm": 1.0608691574541762,
+      "learning_rate": 2e-05,
+      "loss": 0.6519461631774902,
+      "step": 1390
+    },
+    {
+      "epoch": 0.10651247717589775,
+      "grad_norm": 2.8077241127945483,
+      "learning_rate": 2e-05,
+      "loss": 0.6312610626220703,
+      "step": 1400
+    },
+    {
+      "epoch": 0.10727328058429701,
+      "grad_norm": 2.4692034985049167,
+      "learning_rate": 2e-05,
+      "loss": 0.6593670845031738,
+      "step": 1410
+    },
+    {
+      "epoch": 0.10803408399269629,
+      "grad_norm": 1.1688082965213138,
+      "learning_rate": 2e-05,
+      "loss": 0.6476199626922607,
+      "step": 1420
+    },
+    {
+      "epoch": 0.10879488740109555,
+      "grad_norm": 2.1444655878822156,
+      "learning_rate": 2e-05,
+      "loss": 0.6519205093383789,
+      "step": 1430
+    },
+    {
+      "epoch": 0.10955569080949483,
+      "grad_norm": 2.330806151670356,
+      "learning_rate": 2e-05,
+      "loss": 0.673212718963623,
+      "step": 1440
+    },
+    {
+      "epoch": 0.11031649421789409,
+      "grad_norm": 1.1797329434550838,
+      "learning_rate": 2e-05,
+      "loss": 0.629892110824585,
+      "step": 1450
+    },
+    {
+      "epoch": 0.11107729762629337,
+      "grad_norm": 1.370178616571048,
+      "learning_rate": 2e-05,
+      "loss": 0.6480928421020508,
+      "step": 1460
+    },
+    {
+      "epoch": 0.11183810103469263,
+      "grad_norm": 1.701804101440024,
+      "learning_rate": 2e-05,
+      "loss": 0.6533725261688232,
+      "step": 1470
+    },
+    {
+      "epoch": 0.11259890444309191,
+      "grad_norm": 1.5918247074640068,
+      "learning_rate": 2e-05,
+      "loss": 0.6213892936706543,
+      "step": 1480
+    },
+    {
+      "epoch": 0.11335970785149117,
+      "grad_norm": 4.479701036046454,
+      "learning_rate": 2e-05,
+      "loss": 0.6552947998046875,
+      "step": 1490
+    },
+    {
+      "epoch": 0.11412051125989045,
+      "grad_norm": 1.1821034084350328,
+      "learning_rate": 2e-05,
+      "loss": 0.6669442176818847,
+      "step": 1500
+    },
+    {
+      "epoch": 0.11488131466828971,
+      "grad_norm": 2.56260643167876,
+      "learning_rate": 2e-05,
+      "loss": 0.6492503643035888,
+      "step": 1510
+    },
+    {
+      "epoch": 0.11564211807668898,
+      "grad_norm": 1.1572098215301176,
+      "learning_rate": 2e-05,
+      "loss": 0.6260011196136475,
+      "step": 1520
+    },
+    {
+      "epoch": 0.11640292148508825,
+      "grad_norm": 1.2381901297643017,
+      "learning_rate": 2e-05,
+      "loss": 0.6485986232757568,
+      "step": 1530
+    },
+    {
+      "epoch": 0.11716372489348752,
+      "grad_norm": 0.8407626307915014,
+      "learning_rate": 2e-05,
+      "loss": 0.6335528373718262,
+      "step": 1540
+    },
+    {
+      "epoch": 0.1179245283018868,
+      "grad_norm": 1.1522494208742629,
+      "learning_rate": 2e-05,
+      "loss": 0.6174094676971436,
+      "step": 1550
+    },
+    {
+      "epoch": 0.11868533171028606,
+      "grad_norm": 1.4358247956839025,
+      "learning_rate": 2e-05,
+      "loss": 0.6372797012329101,
+      "step": 1560
+    },
+    {
+      "epoch": 0.11944613511868533,
+      "grad_norm": 1.309941398010788,
+      "learning_rate": 2e-05,
+      "loss": 0.6266646385192871,
+      "step": 1570
+    },
+    {
+      "epoch": 0.1202069385270846,
+      "grad_norm": 1.3808985287266267,
+      "learning_rate": 2e-05,
+      "loss": 0.6343531608581543,
+      "step": 1580
+    },
+    {
+      "epoch": 0.12096774193548387,
+      "grad_norm": 1.3528927662199868,
+      "learning_rate": 2e-05,
+      "loss": 0.6284969329833985,
+      "step": 1590
+    },
+    {
+      "epoch": 0.12172854534388314,
+      "grad_norm": 3.0519694955006105,
+      "learning_rate": 2e-05,
+      "loss": 0.6424075126647949,
+      "step": 1600
+    },
+    {
+      "epoch": 0.12248934875228241,
+      "grad_norm": 2.174281639710108,
+      "learning_rate": 2e-05,
+      "loss": 0.6337644577026367,
+      "step": 1610
+    },
+    {
+      "epoch": 0.12325015216068168,
+      "grad_norm": 1.561704499585276,
+      "learning_rate": 2e-05,
+      "loss": 0.6298858165740967,
+      "step": 1620
+    },
+    {
+      "epoch": 0.12401095556908096,
+      "grad_norm": 1.224072880782582,
+      "learning_rate": 2e-05,
+      "loss": 0.6224400997161865,
+      "step": 1630
+    },
+    {
+      "epoch": 0.12477175897748022,
+      "grad_norm": 1.0462139926853118,
+      "learning_rate": 2e-05,
+      "loss": 0.6482504844665528,
+      "step": 1640
+    },
+    {
+      "epoch": 0.12553256238587948,
+      "grad_norm": 9.21967648588717,
+      "learning_rate": 2e-05,
+      "loss": 0.6525485992431641,
+      "step": 1650
+    },
+    {
+      "epoch": 0.12629336579427874,
+      "grad_norm": 2.007629694651067,
+      "learning_rate": 2e-05,
+      "loss": 0.6432745933532715,
+      "step": 1660
+    },
+    {
+      "epoch": 0.12705416920267804,
+      "grad_norm": 0.9720933358173028,
+      "learning_rate": 2e-05,
+      "loss": 0.6356534004211426,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1278149726110773,
+      "grad_norm": 1.4468055933940205,
+      "learning_rate": 2e-05,
+      "loss": 0.6624721050262451,
+      "step": 1680
+    },
+    {
+      "epoch": 0.12857577601947656,
+      "grad_norm": 1.3207000999132357,
+      "learning_rate": 2e-05,
+      "loss": 0.6150018692016601,
+      "step": 1690
+    },
+    {
+      "epoch": 0.12933657942787583,
+      "grad_norm": 1.1806546194990943,
+      "learning_rate": 2e-05,
+      "loss": 0.6312203407287598,
+      "step": 1700
+    },
+    {
+      "epoch": 0.13009738283627512,
+      "grad_norm": 1.5786931266899833,
+      "learning_rate": 2e-05,
+      "loss": 0.6413074493408203,
+      "step": 1710
+    },
+    {
+      "epoch": 0.13085818624467438,
+      "grad_norm": 1.0702890272477146,
+      "learning_rate": 2e-05,
+      "loss": 0.642481517791748,
+      "step": 1720
+    },
+    {
+      "epoch": 0.13161898965307364,
+      "grad_norm": 1.4900158944450277,
+      "learning_rate": 2e-05,
+      "loss": 0.6344526290893555,
+      "step": 1730
+    },
+    {
+      "epoch": 0.1323797930614729,
+      "grad_norm": 1.4116190947678835,
+      "learning_rate": 2e-05,
+      "loss": 0.6302139282226562,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1331405964698722,
+      "grad_norm": 1.7448570810060797,
+      "learning_rate": 2e-05,
+      "loss": 0.6293466567993165,
+      "step": 1750
+    },
+    {
+      "epoch": 0.13390139987827146,
+      "grad_norm": 3.3383049060870595,
+      "learning_rate": 2e-05,
+      "loss": 0.6328795909881592,
+      "step": 1760
+    },
+    {
+      "epoch": 0.13466220328667072,
+      "grad_norm": 3.631205476735395,
+      "learning_rate": 2e-05,
+      "loss": 0.615032958984375,
+      "step": 1770
+    },
+    {
+      "epoch": 0.13542300669507,
+      "grad_norm": 1.5518542674647309,
+      "learning_rate": 2e-05,
+      "loss": 0.6316960334777832,
+      "step": 1780
+    },
+    {
+      "epoch": 0.13618381010346928,
+      "grad_norm": 1.7492350745877743,
+      "learning_rate": 2e-05,
+      "loss": 0.6442355155944824,
+      "step": 1790
+    },
+    {
+      "epoch": 0.13694461351186854,
+      "grad_norm": 1.9258665395327041,
+      "learning_rate": 2e-05,
+      "loss": 0.6106966018676758,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1377054169202678,
+      "grad_norm": 6.827483693001483,
+      "learning_rate": 2e-05,
+      "loss": 0.642854642868042,
+      "step": 1810
+    },
+    {
+      "epoch": 0.13846622032866707,
+      "grad_norm": 1.1009026570397635,
+      "learning_rate": 2e-05,
+      "loss": 0.6260102748870849,
+      "step": 1820
+    },
+    {
+      "epoch": 0.13922702373706633,
+      "grad_norm": 1.049291470879374,
+      "learning_rate": 2e-05,
+      "loss": 0.6057558059692383,
+      "step": 1830
+    },
+    {
+      "epoch": 0.13998782714546562,
+      "grad_norm": 1.8353318571336608,
+      "learning_rate": 2e-05,
+      "loss": 0.6460296154022217,
+      "step": 1840
+    },
+    {
+      "epoch": 0.14074863055386488,
+      "grad_norm": 11.056378441909427,
+      "learning_rate": 2e-05,
+      "loss": 0.6394613265991211,
+      "step": 1850
+    },
+    {
+      "epoch": 0.14150943396226415,
+      "grad_norm": 39.436296571275825,
+      "learning_rate": 2e-05,
+      "loss": 0.6526387691497803,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1422702373706634,
+      "grad_norm": 1.8312409048902327,
+      "learning_rate": 2e-05,
+      "loss": 0.6392415523529053,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1430310407790627,
+      "grad_norm": 1.071587830023617,
+      "learning_rate": 2e-05,
+      "loss": 0.6196507453918457,
+      "step": 1880
+    },
+    {
+      "epoch": 0.14379184418746196,
+      "grad_norm": 1.4789771369042652,
+      "learning_rate": 2e-05,
+      "loss": 0.6441464900970459,
+      "step": 1890
+    },
+    {
+      "epoch": 0.14455264759586123,
+      "grad_norm": 1.8376505782240828,
+      "learning_rate": 2e-05,
+      "loss": 0.6638864040374756,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1453134510042605,
+      "grad_norm": 1.3745479307166024,
+      "learning_rate": 2e-05,
+      "loss": 0.6252675533294678,
+      "step": 1910
+    },
+    {
+      "epoch": 0.14607425441265978,
+      "grad_norm": 1.1668435873091556,
+      "learning_rate": 2e-05,
+      "loss": 0.6090609550476074,
+      "step": 1920
+    },
+    {
+      "epoch": 0.14683505782105905,
+      "grad_norm": 1.23594711006607,
+      "learning_rate": 2e-05,
+      "loss": 0.6123690605163574,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1475958612294583,
+      "grad_norm": 1.2418726773487727,
+      "learning_rate": 2e-05,
+      "loss": 0.6570170879364013,
+      "step": 1940
+    },
+    {
+      "epoch": 0.14835666463785757,
+      "grad_norm": 1.0504866082911768,
+      "learning_rate": 2e-05,
+      "loss": 0.6538400173187255,
+      "step": 1950
+    },
+    {
+      "epoch": 0.14911746804625683,
+      "grad_norm": 2.014671731533198,
+      "learning_rate": 2e-05,
+      "loss": 0.6170327186584472,
+      "step": 1960
+    },
+    {
+      "epoch": 0.14987827145465613,
+      "grad_norm": 2.1058379351030108,
+      "learning_rate": 2e-05,
+      "loss": 0.6381420135498047,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1506390748630554,
+      "grad_norm": 1.3150633416083708,
+      "learning_rate": 2e-05,
+      "loss": 0.6363181114196778,
+      "step": 1980
+    },
+    {
+      "epoch": 0.15139987827145465,
+      "grad_norm": 0.8932530375434491,
+      "learning_rate": 2e-05,
+      "loss": 0.6291966915130616,
+      "step": 1990
+    },
+    {
+      "epoch": 0.15216068167985392,
+      "grad_norm": 2.957084421448231,
+      "learning_rate": 2e-05,
+      "loss": 0.6210861206054688,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1529214850882532,
+      "grad_norm": 0.8890058850680974,
+      "learning_rate": 2e-05,
+      "loss": 0.6021919250488281,
+      "step": 2010
+    },
+    {
+      "epoch": 0.15368228849665247,
+      "grad_norm": 0.9932774221445535,
+      "learning_rate": 2e-05,
+      "loss": 0.6515641689300538,
+      "step": 2020
+    },
+    {
+      "epoch": 0.15444309190505173,
+      "grad_norm": 1.0023975736435211,
+      "learning_rate": 2e-05,
+      "loss": 0.6417403221130371,
+      "step": 2030
+    },
+    {
+      "epoch": 0.155203895313451,
+      "grad_norm": 0.8554859168766407,
+      "learning_rate": 2e-05,
+      "loss": 0.6303350448608398,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1559646987218503,
+      "grad_norm": 1.5251333977211128,
+      "learning_rate": 2e-05,
+      "loss": 0.6416450500488281,
+      "step": 2050
+    },
+    {
+      "epoch": 0.15672550213024955,
+      "grad_norm": 1.0986596009489755,
+      "learning_rate": 2e-05,
+      "loss": 0.6401923179626465,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1574863055386488,
+      "grad_norm": 1.1539013637626214,
+      "learning_rate": 2e-05,
+      "loss": 0.6259047508239746,
+      "step": 2070
+    },
+    {
+      "epoch": 0.15824710894704808,
+      "grad_norm": 1.4348698553263404,
+      "learning_rate": 2e-05,
+      "loss": 0.6216594696044921,
+      "step": 2080
+    },
+    {
+      "epoch": 0.15900791235544734,
+      "grad_norm": 0.9526520497956816,
+      "learning_rate": 2e-05,
+      "loss": 0.6363525390625,
+      "step": 2090
+    },
+    {
+      "epoch": 0.15976871576384663,
+      "grad_norm": 3.45947923470152,
+      "learning_rate": 2e-05,
+      "loss": 0.6570816040039062,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1605295191722459,
+      "grad_norm": 2.0280020107564853,
+      "learning_rate": 2e-05,
+      "loss": 0.6443814754486084,
+      "step": 2110
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 1.2746739038674486,
+      "learning_rate": 2e-05,
+      "loss": 0.6273581981658936,
+      "step": 2120
+    },
+    {
+      "epoch": 0.16205112598904442,
+      "grad_norm": 1.1054316756226754,
+      "learning_rate": 2e-05,
+      "loss": 0.6392014980316162,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1628119293974437,
+      "grad_norm": 0.7393965027847225,
+      "learning_rate": 2e-05,
+      "loss": 0.6142825603485107,
+      "step": 2140
+    },
+    {
+      "epoch": 0.16357273280584297,
+      "grad_norm": 0.8045230869463124,
+      "learning_rate": 2e-05,
+      "loss": 0.6058244228363037,
+      "step": 2150
+    },
+    {
+      "epoch": 0.16433353621424224,
+      "grad_norm": 0.8285093876406089,
+      "learning_rate": 2e-05,
+      "loss": 0.6443311214447022,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1650943396226415,
+      "grad_norm": 0.8596670201373968,
+      "learning_rate": 2e-05,
+      "loss": 0.6266818046569824,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1658551430310408,
+      "grad_norm": 0.9253480501904824,
+      "learning_rate": 2e-05,
+      "loss": 0.6671144008636475,
+      "step": 2180
+    },
+    {
+      "epoch": 0.16661594643944005,
+      "grad_norm": 0.8320885887267414,
+      "learning_rate": 2e-05,
+      "loss": 0.62307767868042,
+      "step": 2190
+    },
+    {
+      "epoch": 0.16737674984783932,
+      "grad_norm": 2.6787131895939384,
+      "learning_rate": 2e-05,
+      "loss": 0.6703015804290772,
+      "step": 2200
+    },
+    {
+      "epoch": 0.16813755325623858,
+      "grad_norm": 0.8902868417037517,
+      "learning_rate": 2e-05,
+      "loss": 0.6234861373901367,
+      "step": 2210
+    },
+    {
+      "epoch": 0.16889835666463784,
+      "grad_norm": 0.8847567666362576,
+      "learning_rate": 2e-05,
+      "loss": 0.6103476524353028,
+      "step": 2220
+    },
+    {
+      "epoch": 0.16965916007303714,
+      "grad_norm": 0.8408811071144497,
+      "learning_rate": 2e-05,
+      "loss": 0.6184403419494628,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1704199634814364,
+      "grad_norm": 1.6892917345046299,
+      "learning_rate": 2e-05,
+      "loss": 0.6205358505249023,
+      "step": 2240
+    },
+    {
+      "epoch": 0.17118076688983566,
+      "grad_norm": 0.7405946865412475,
+      "learning_rate": 2e-05,
+      "loss": 0.61587553024292,
+      "step": 2250
+    },
+    {
+      "epoch": 0.17194157029823492,
+      "grad_norm": 0.7506110237045676,
+      "learning_rate": 2e-05,
+      "loss": 0.6244790554046631,
+      "step": 2260
+    },
+    {
+      "epoch": 0.17270237370663422,
+      "grad_norm": 2.153478514759513,
+      "learning_rate": 2e-05,
+      "loss": 0.6198970794677734,
+      "step": 2270
+    },
+    {
+      "epoch": 0.17346317711503348,
+      "grad_norm": 10.500995119599999,
+      "learning_rate": 2e-05,
+      "loss": 0.6225154876708985,
+      "step": 2280
+    },
+    {
+      "epoch": 0.17422398052343274,
+      "grad_norm": 1.213251555227088,
+      "learning_rate": 2e-05,
+      "loss": 0.6163989067077636,
+      "step": 2290
+    },
+    {
+      "epoch": 0.174984783931832,
+      "grad_norm": 0.9729698503142967,
+      "learning_rate": 2e-05,
+      "loss": 0.610928726196289,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1757455873402313,
+      "grad_norm": 1.0702561000731816,
+      "learning_rate": 2e-05,
+      "loss": 0.6329403877258301,
+      "step": 2310
+    },
+    {
+      "epoch": 0.17650639074863056,
+      "grad_norm": 0.9932357555439623,
+      "learning_rate": 2e-05,
+      "loss": 0.6477541923522949,
+      "step": 2320
+    },
+    {
+      "epoch": 0.17726719415702982,
+      "grad_norm": 0.8974295172792705,
+      "learning_rate": 2e-05,
+      "loss": 0.6245576858520507,
+      "step": 2330
+    },
+    {
+      "epoch": 0.17802799756542909,
+      "grad_norm": 0.8948226638723901,
+      "learning_rate": 2e-05,
+      "loss": 0.6292768478393554,
+      "step": 2340
+    },
+    {
+      "epoch": 0.17878880097382835,
+      "grad_norm": 1.0441356789609721,
+      "learning_rate": 2e-05,
+      "loss": 0.6134624481201172,
+      "step": 2350
+    },
+    {
+      "epoch": 0.17954960438222764,
+      "grad_norm": 1.0579592934994417,
+      "learning_rate": 2e-05,
+      "loss": 0.6160748958587646,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1803104077906269,
+      "grad_norm": 1.0688656877042852,
+      "learning_rate": 2e-05,
+      "loss": 0.6353797912597656,
+      "step": 2370
+    },
+    {
+      "epoch": 0.18107121119902617,
+      "grad_norm": 0.9630873249684109,
+      "learning_rate": 2e-05,
+      "loss": 0.6212389945983887,
+      "step": 2380
+    },
+    {
+      "epoch": 0.18183201460742543,
+      "grad_norm": 1.299642947772091,
+      "learning_rate": 2e-05,
+      "loss": 0.6184343338012696,
+      "step": 2390
+    },
+    {
+      "epoch": 0.18259281801582472,
+      "grad_norm": 1.9371331971789227,
+      "learning_rate": 2e-05,
+      "loss": 0.6408385276794434,
+      "step": 2400
+    },
+    {
+      "epoch": 0.18335362142422398,
+      "grad_norm": 2.3484601024255074,
+      "learning_rate": 2e-05,
+      "loss": 0.6273296356201172,
+      "step": 2410
+    },
+    {
+      "epoch": 0.18411442483262325,
+      "grad_norm": 2.421637945440785,
+      "learning_rate": 2e-05,
+      "loss": 0.6258901119232178,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1848752282410225,
+      "grad_norm": 7.087299118182203,
+      "learning_rate": 2e-05,
+      "loss": 0.6362738609313965,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1856360316494218,
+      "grad_norm": 1.60152608667479,
+      "learning_rate": 2e-05,
+      "loss": 0.6333796977996826,
+      "step": 2440
+    },
+    {
+      "epoch": 0.18639683505782106,
+      "grad_norm": 1.5388500323071719,
+      "learning_rate": 2e-05,
+      "loss": 0.6134243011474609,
+      "step": 2450
+    },
+    {
+      "epoch": 0.18715763846622033,
+      "grad_norm": 1.6406255566044918,
+      "learning_rate": 2e-05,
+      "loss": 0.6321418285369873,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1879184418746196,
+      "grad_norm": 1.751258694616863,
+      "learning_rate": 2e-05,
+      "loss": 0.6141924858093262,
+      "step": 2470
+    },
+    {
+      "epoch": 0.18867924528301888,
+      "grad_norm": 2.2372643207413128,
+      "learning_rate": 2e-05,
+      "loss": 0.6251022338867187,
+      "step": 2480
+    },
+    {
+      "epoch": 0.18944004869141814,
+      "grad_norm": 3.178260160359772,
+      "learning_rate": 2e-05,
+      "loss": 0.6332900047302246,
+      "step": 2490
+    },
+    {
+      "epoch": 0.1902008520998174,
+      "grad_norm": 2.5115652770876618,
+      "learning_rate": 2e-05,
+      "loss": 0.6328558921813965,
+      "step": 2500
+    },
+    {
+      "epoch": 0.19096165550821667,
+      "grad_norm": 3.551106098546657,
+      "learning_rate": 2e-05,
+      "loss": 0.6492326736450196,
+      "step": 2510
+    },
+    {
+      "epoch": 0.19172245891661593,
+      "grad_norm": 1.1091484806788126,
+      "learning_rate": 2e-05,
+      "loss": 0.6217524528503418,
+      "step": 2520
+    },
+    {
+      "epoch": 0.19248326232501523,
+      "grad_norm": 0.7761699186852694,
+      "learning_rate": 2e-05,
+      "loss": 0.630795669555664,
+      "step": 2530
+    },
+    {
+      "epoch": 0.1932440657334145,
+      "grad_norm": 2.8016530566761406,
+      "learning_rate": 2e-05,
+      "loss": 0.6455306529998779,
+      "step": 2540
+    },
+    {
+      "epoch": 0.19400486914181375,
+      "grad_norm": 0.9817069331556703,
+      "learning_rate": 2e-05,
+      "loss": 0.5892092841012138,
+      "step": 2550
+    },
+    {
+      "epoch": 0.19476567255021301,
+      "grad_norm": 1.1506642297473906,
+      "learning_rate": 2e-05,
+      "loss": 0.6403939247131347,
+      "step": 2560
+    },
+    {
+      "epoch": 0.1955264759586123,
+      "grad_norm": 1.023859977623109,
+      "learning_rate": 2e-05,
+      "loss": 0.6122564315795899,
+      "step": 2570
+    },
+    {
+      "epoch": 0.19628727936701157,
+      "grad_norm": 1.5469625525063975,
+      "learning_rate": 2e-05,
+      "loss": 0.6223512172698975,
+      "step": 2580
+    },
+    {
+      "epoch": 0.19704808277541083,
+      "grad_norm": 1.0680850837639781,
+      "learning_rate": 2e-05,
+      "loss": 0.6277861595153809,
+      "step": 2590
+    },
+    {
+      "epoch": 0.1978088861838101,
+      "grad_norm": 0.9216390380997925,
+      "learning_rate": 2e-05,
+      "loss": 0.6205940246582031,
+      "step": 2600
+    },
+    {
+      "epoch": 0.1985696895922094,
+      "grad_norm": 0.9069803099730847,
+      "learning_rate": 2e-05,
+      "loss": 0.629159688949585,
+      "step": 2610
+    },
+    {
+      "epoch": 0.19933049300060865,
+      "grad_norm": 0.9065994563342014,
+      "learning_rate": 2e-05,
+      "loss": 0.618154239654541,
+      "step": 2620
+    },
+    {
+      "epoch": 0.2000912964090079,
+      "grad_norm": 2.1363162353810567,
+      "learning_rate": 2e-05,
+      "loss": 0.6302952766418457,
+      "step": 2630
+    },
+    {
+      "epoch": 0.20085209981740718,
+      "grad_norm": 1.3495231395011489,
+      "learning_rate": 2e-05,
+      "loss": 0.602275037765503,
+      "step": 2640
+    },
+    {
+      "epoch": 0.20161290322580644,
+      "grad_norm": 1.268006641731718,
+      "learning_rate": 2e-05,
+      "loss": 0.633419418334961,
+      "step": 2650
+    },
+    {
+      "epoch": 0.20237370663420573,
+      "grad_norm": 1.1198642302804405,
+      "learning_rate": 2e-05,
+      "loss": 0.628083324432373,
+      "step": 2660
+    },
+    {
+      "epoch": 0.203134510042605,
+      "grad_norm": 0.921063027206477,
+      "learning_rate": 2e-05,
+      "loss": 0.6306491374969483,
+      "step": 2670
+    },
+    {
+      "epoch": 0.20389531345100426,
+      "grad_norm": 0.9131822079624049,
+      "learning_rate": 2e-05,
+      "loss": 0.6213677883148193,
+      "step": 2680
+    },
+    {
+      "epoch": 0.20465611685940352,
+      "grad_norm": 0.9626227664170652,
+      "learning_rate": 2e-05,
+      "loss": 0.6190157890319824,
+      "step": 2690
+    },
+    {
+      "epoch": 0.2054169202678028,
+      "grad_norm": 1.5952273952274294,
+      "learning_rate": 2e-05,
+      "loss": 0.6162997245788574,
+      "step": 2700
+    },
+    {
+      "epoch": 0.20617772367620207,
+      "grad_norm": 1.6032390147900086,
+      "learning_rate": 2e-05,
+      "loss": 0.6197282791137695,
+      "step": 2710
+    },
+    {
+      "epoch": 0.20693852708460134,
+      "grad_norm": 1.5025225130869249,
+      "learning_rate": 2e-05,
+      "loss": 0.6324640274047851,
+      "step": 2720
+    },
+    {
+      "epoch": 0.2076993304930006,
+      "grad_norm": 1.15091373470265,
+      "learning_rate": 2e-05,
+      "loss": 0.6154145240783692,
+      "step": 2730
+    },
+    {
+      "epoch": 0.2084601339013999,
+      "grad_norm": 2.503142503929874,
+      "learning_rate": 2e-05,
+      "loss": 0.6177960395812988,
+      "step": 2740
+    },
+    {
+      "epoch": 0.20922093730979915,
+      "grad_norm": 3.225448811792671,
+      "learning_rate": 2e-05,
+      "loss": 0.6053067207336426,
+      "step": 2750
+    },
+    {
+      "epoch": 0.20998174071819842,
+      "grad_norm": 0.9340037286792306,
+      "learning_rate": 2e-05,
+      "loss": 0.6067095756530761,
+      "step": 2760
+    },
+    {
+      "epoch": 0.21074254412659768,
+      "grad_norm": 0.7839092401023494,
+      "learning_rate": 2e-05,
+      "loss": 0.6238336086273193,
+      "step": 2770
+    },
+    {
+      "epoch": 0.21150334753499694,
+      "grad_norm": 0.9703683681711317,
+      "learning_rate": 2e-05,
+      "loss": 0.630821943283081,
+      "step": 2780
+    },
+    {
+      "epoch": 0.21226415094339623,
+      "grad_norm": 1.1956854322866988,
+      "learning_rate": 2e-05,
+      "loss": 0.6082069396972656,
+      "step": 2790
+    },
+    {
+      "epoch": 0.2130249543517955,
+      "grad_norm": 2.5970281442565692,
+      "learning_rate": 2e-05,
+      "loss": 0.6003427505493164,
+      "step": 2800
+    },
+    {
+      "epoch": 0.21378575776019476,
+      "grad_norm": 1.181050183007216,
+      "learning_rate": 2e-05,
+      "loss": 0.6277856349945068,
+      "step": 2810
+    },
+    {
+      "epoch": 0.21454656116859402,
+      "grad_norm": 1.0961834404857531,
+      "learning_rate": 2e-05,
+      "loss": 0.6375556468963623,
+      "step": 2820
+    },
+    {
+      "epoch": 0.21530736457699332,
+      "grad_norm": 1.107277032190709,
+      "learning_rate": 2e-05,
+      "loss": 0.6170720100402832,
+      "step": 2830
+    },
+    {
+      "epoch": 0.21606816798539258,
+      "grad_norm": 0.9876856488355245,
+      "learning_rate": 2e-05,
+      "loss": 0.6377505302429199,
+      "step": 2840
+    },
+    {
+      "epoch": 0.21682897139379184,
+      "grad_norm": 2.8138017098561487,
+      "learning_rate": 2e-05,
+      "loss": 0.6230205535888672,
+      "step": 2850
+    },
+    {
+      "epoch": 0.2175897748021911,
+      "grad_norm": 0.9569008919457225,
+      "learning_rate": 2e-05,
+      "loss": 0.6139896392822266,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2183505782105904,
+      "grad_norm": 1.2166946479977514,
+      "learning_rate": 2e-05,
+      "loss": 0.6102657318115234,
+      "step": 2870
+    },
+    {
+      "epoch": 0.21911138161898966,
+      "grad_norm": 0.9439320682429234,
+      "learning_rate": 2e-05,
+      "loss": 0.6210070610046386,
+      "step": 2880
+    },
+    {
+      "epoch": 0.21987218502738892,
+      "grad_norm": 1.9444963486987974,
+      "learning_rate": 2e-05,
+      "loss": 0.6143630504608154,
+      "step": 2890
+    },
+    {
+      "epoch": 0.22063298843578819,
+      "grad_norm": 1.0731632672697757,
+      "learning_rate": 2e-05,
+      "loss": 0.6242578029632568,
+      "step": 2900
+    },
+    {
+      "epoch": 0.22139379184418745,
+      "grad_norm": 1.4416846334633917,
+      "learning_rate": 2e-05,
+      "loss": 0.6301780700683594,
+      "step": 2910
+    },
+    {
+      "epoch": 0.22215459525258674,
+      "grad_norm": 1.0398958563814207,
+      "learning_rate": 2e-05,
+      "loss": 0.6249309539794922,
+      "step": 2920
+    },
+    {
+      "epoch": 0.222915398660986,
+      "grad_norm": 1.5915685165282702,
+      "learning_rate": 2e-05,
+      "loss": 0.6256701946258545,
+      "step": 2930
+    },
+    {
+      "epoch": 0.22367620206938527,
+      "grad_norm": 1.97421945040368,
+      "learning_rate": 2e-05,
+      "loss": 0.6315007686614991,
+      "step": 2940
+    },
+    {
+      "epoch": 0.22443700547778453,
+      "grad_norm": 0.9439951164923264,
+      "learning_rate": 2e-05,
+      "loss": 0.6190637111663818,
+      "step": 2950
+    },
+    {
+      "epoch": 0.22519780888618382,
+      "grad_norm": 1.9216827306482598,
+      "learning_rate": 2e-05,
+      "loss": 0.6048786163330078,
+      "step": 2960
+    },
+    {
+      "epoch": 0.22595861229458308,
+      "grad_norm": 1.3758385768580352,
+      "learning_rate": 2e-05,
+      "loss": 0.6186373710632325,
+      "step": 2970
+    },
+    {
+      "epoch": 0.22671941570298235,
+      "grad_norm": 3.0227729807206907,
+      "learning_rate": 2e-05,
+      "loss": 0.6130160808563232,
+      "step": 2980
+    },
+    {
+      "epoch": 0.2274802191113816,
+      "grad_norm": 1.0965618740158993,
+      "learning_rate": 2e-05,
+      "loss": 0.6068836212158203,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2282410225197809,
+      "grad_norm": 1.3276497304928805,
+      "learning_rate": 2e-05,
+      "loss": 0.6168122291564941,
+      "step": 3000
+    },
+    {
+      "epoch": 0.22900182592818016,
+      "grad_norm": 2.3525701651750768,
+      "learning_rate": 2e-05,
+      "loss": 0.6179019451141358,
+      "step": 3010
+    },
+    {
+      "epoch": 0.22976262933657943,
+      "grad_norm": 0.7100746650791652,
+      "learning_rate": 2e-05,
+      "loss": 0.5935450553894043,
+      "step": 3020
+    },
+    {
+      "epoch": 0.2305234327449787,
+      "grad_norm": 2.967110575052808,
+      "learning_rate": 2e-05,
+      "loss": 0.6108531475067138,
+      "step": 3030
+    },
+    {
+      "epoch": 0.23128423615337795,
+      "grad_norm": 1.0742537412854092,
+      "learning_rate": 2e-05,
+      "loss": 0.5901834011077881,
+      "step": 3040
+    },
+    {
+      "epoch": 0.23204503956177724,
+      "grad_norm": 1.0149513822294771,
+      "learning_rate": 2e-05,
+      "loss": 0.6123186111450195,
+      "step": 3050
+    },
+    {
+      "epoch": 0.2328058429701765,
+      "grad_norm": 1.129503907318443,
+      "learning_rate": 2e-05,
+      "loss": 0.6262495994567872,
+      "step": 3060
+    },
+    {
+      "epoch": 0.23356664637857577,
+      "grad_norm": 1.5795916128532546,
+      "learning_rate": 2e-05,
+      "loss": 0.6286283016204834,
+      "step": 3070
+    },
+    {
+      "epoch": 0.23432744978697503,
+      "grad_norm": 1.5737449340106107,
+      "learning_rate": 2e-05,
+      "loss": 0.6367819786071778,
+      "step": 3080
+    },
+    {
+      "epoch": 0.23508825319537432,
+      "grad_norm": 0.837301889940201,
+      "learning_rate": 2e-05,
+      "loss": 0.5960247039794921,
+      "step": 3090
+    },
+    {
+      "epoch": 0.2358490566037736,
+      "grad_norm": 1.7129589642475265,
+      "learning_rate": 2e-05,
+      "loss": 0.6188911437988281,
+      "step": 3100
+    },
+    {
+      "epoch": 0.23660986001217285,
+      "grad_norm": 1.0978797863870255,
+      "learning_rate": 2e-05,
+      "loss": 0.6011454582214355,
+      "step": 3110
+    },
+    {
+      "epoch": 0.23737066342057211,
+      "grad_norm": 1.3392547341281262,
+      "learning_rate": 2e-05,
+      "loss": 0.6084640502929688,
+      "step": 3120
+    },
+    {
+      "epoch": 0.2381314668289714,
+      "grad_norm": 1.1896487886307119,
+      "learning_rate": 2e-05,
+      "loss": 0.6010078430175781,
+      "step": 3130
+    },
+    {
+      "epoch": 0.23889227023737067,
+      "grad_norm": 0.9724148059510948,
+      "learning_rate": 2e-05,
+      "loss": 0.625861120223999,
+      "step": 3140
+    },
+    {
+      "epoch": 0.23965307364576993,
+      "grad_norm": 0.8546534339092822,
+      "learning_rate": 2e-05,
+      "loss": 0.5815396785736084,
+      "step": 3150
+    },
+    {
+      "epoch": 0.2404138770541692,
+      "grad_norm": 1.0358608032712895,
+      "learning_rate": 2e-05,
+      "loss": 0.5957612037658692,
+      "step": 3160
+    },
+    {
+      "epoch": 0.24117468046256849,
+      "grad_norm": 0.9532257041483436,
+      "learning_rate": 2e-05,
+      "loss": 0.6230683326721191,
+      "step": 3170
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 0.856584963672178,
+      "learning_rate": 2e-05,
+      "loss": 0.6225377559661865,
+      "step": 3180
+    },
+    {
+      "epoch": 0.242696287279367,
+      "grad_norm": 0.9701495366122038,
+      "learning_rate": 2e-05,
+      "loss": 0.5935500144958497,
+      "step": 3190
+    },
+    {
+      "epoch": 0.24345709068776628,
+      "grad_norm": 0.8719862670241907,
+      "learning_rate": 2e-05,
+      "loss": 0.6125042915344239,
+      "step": 3200
+    },
+    {
+      "epoch": 0.24421789409616554,
+      "grad_norm": 0.8876554659779671,
+      "learning_rate": 2e-05,
+      "loss": 0.6382464408874512,
+      "step": 3210
+    },
+    {
+      "epoch": 0.24497869750456483,
+      "grad_norm": 0.663293931503932,
+      "learning_rate": 2e-05,
+      "loss": 0.613660192489624,
+      "step": 3220
+    },
+    {
+      "epoch": 0.2457395009129641,
+      "grad_norm": 0.9157732445075698,
+      "learning_rate": 2e-05,
+      "loss": 0.6227169036865234,
+      "step": 3230
+    },
+    {
+      "epoch": 0.24650030432136336,
+      "grad_norm": 0.9169230676438861,
+      "learning_rate": 2e-05,
+      "loss": 0.6264325618743897,
+      "step": 3240
+    },
+    {
+      "epoch": 0.24726110772976262,
+      "grad_norm": 0.9267464760873556,
+      "learning_rate": 2e-05,
+      "loss": 0.6031881332397461,
+      "step": 3250
+    },
+    {
+      "epoch": 0.2480219111381619,
+      "grad_norm": 1.6276096678949572,
+      "learning_rate": 2e-05,
+      "loss": 0.6197366714477539,
+      "step": 3260
+    },
+    {
+      "epoch": 0.24878271454656117,
+      "grad_norm": 0.8661493898082891,
+      "learning_rate": 2e-05,
+      "loss": 0.6157424449920654,
+      "step": 3270
+    },
+    {
+      "epoch": 0.24954351795496044,
+      "grad_norm": 0.8642949072778816,
+      "learning_rate": 2e-05,
+      "loss": 0.6269526481628418,
+      "step": 3280
+    },
+    {
+      "epoch": 0.2503043213633597,
+      "grad_norm": 1.1643303657998856,
+      "learning_rate": 2e-05,
+      "loss": 0.6230345726013183,
+      "step": 3290
+    },
+    {
+      "epoch": 0.25106512477175896,
+      "grad_norm": 1.1115822075885138,
+      "learning_rate": 2e-05,
+      "loss": 0.6161396026611328,
+      "step": 3300
+    },
+    {
+      "epoch": 0.25182592818015825,
+      "grad_norm": 1.3716226645242158,
+      "learning_rate": 2e-05,
+      "loss": 0.6055039882659912,
+      "step": 3310
+    },
+    {
+      "epoch": 0.2525867315885575,
+      "grad_norm": 1.4005747784917617,
+      "learning_rate": 2e-05,
+      "loss": 0.6360630035400391,
+      "step": 3320
+    },
+    {
+      "epoch": 0.2533475349969568,
+      "grad_norm": 1.0243200200378946,
+      "learning_rate": 2e-05,
+      "loss": 0.6756389617919922,
+      "step": 3330
+    },
+    {
+      "epoch": 0.25410833840535607,
+      "grad_norm": 1.4886693526090016,
+      "learning_rate": 2e-05,
+      "loss": 0.6030360698699951,
+      "step": 3340
+    },
+    {
+      "epoch": 0.2548691418137553,
+      "grad_norm": 1.294277490662169,
+      "learning_rate": 2e-05,
+      "loss": 0.6214393615722656,
+      "step": 3350
+    },
+    {
+      "epoch": 0.2556299452221546,
+      "grad_norm": 0.8814030358404723,
+      "learning_rate": 2e-05,
+      "loss": 0.6296616554260254,
+      "step": 3360
+    },
+    {
+      "epoch": 0.2563907486305539,
+      "grad_norm": 1.797589540426131,
+      "learning_rate": 2e-05,
+      "loss": 0.6054345607757569,
+      "step": 3370
+    },
+    {
+      "epoch": 0.2571515520389531,
+      "grad_norm": 1.1534914208205673,
+      "learning_rate": 2e-05,
+      "loss": 0.5965017318725586,
+      "step": 3380
+    },
+    {
+      "epoch": 0.2579123554473524,
+      "grad_norm": 1.094261609741512,
+      "learning_rate": 2e-05,
+      "loss": 0.6489345073699951,
+      "step": 3390
+    },
+    {
+      "epoch": 0.25867315885575165,
+      "grad_norm": 1.085371831991245,
+      "learning_rate": 2e-05,
+      "loss": 0.615788745880127,
+      "step": 3400
+    },
+    {
+      "epoch": 0.25943396226415094,
+      "grad_norm": 2.914563733064073,
+      "learning_rate": 2e-05,
+      "loss": 0.611840009689331,
+      "step": 3410
+    },
+    {
+      "epoch": 0.26019476567255023,
+      "grad_norm": 0.745256249202797,
+      "learning_rate": 2e-05,
+      "loss": 0.5985809326171875,
+      "step": 3420
+    },
+    {
+      "epoch": 0.26095556908094947,
+      "grad_norm": 1.0176582915284635,
+      "learning_rate": 2e-05,
+      "loss": 0.6228419303894043,
+      "step": 3430
+    },
+    {
+      "epoch": 0.26171637248934876,
+      "grad_norm": 1.2161548866693805,
+      "learning_rate": 2e-05,
+      "loss": 0.627016830444336,
+      "step": 3440
+    },
+    {
+      "epoch": 0.26247717589774805,
+      "grad_norm": 1.0364686040834363,
+      "learning_rate": 2e-05,
+      "loss": 0.6175382137298584,
+      "step": 3450
+    },
+    {
+      "epoch": 0.2632379793061473,
+      "grad_norm": 1.1636472172805075,
+      "learning_rate": 2e-05,
+      "loss": 0.6075883388519288,
+      "step": 3460
+    },
+    {
+      "epoch": 0.2639987827145466,
+      "grad_norm": 1.7614077254816805,
+      "learning_rate": 2e-05,
+      "loss": 0.6019673824310303,
+      "step": 3470
+    },
+    {
+      "epoch": 0.2647595861229458,
+      "grad_norm": 0.863434633984219,
+      "learning_rate": 2e-05,
+      "loss": 0.601746654510498,
+      "step": 3480
+    },
+    {
+      "epoch": 0.2655203895313451,
+      "grad_norm": 1.304041982690625,
+      "learning_rate": 2e-05,
+      "loss": 0.6177839279174805,
+      "step": 3490
+    },
+    {
+      "epoch": 0.2662811929397444,
+      "grad_norm": 1.883012196044417,
+      "learning_rate": 2e-05,
+      "loss": 0.6234899044036866,
+      "step": 3500
+    },
+    {
+      "epoch": 0.26704199634814363,
+      "grad_norm": 0.8964938484441732,
+      "learning_rate": 2e-05,
+      "loss": 0.6390236854553223,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2678027997565429,
+      "grad_norm": 1.4284801033824128,
+      "learning_rate": 2e-05,
+      "loss": 0.6097182273864746,
+      "step": 3520
+    },
+    {
+      "epoch": 0.26856360316494216,
+      "grad_norm": 0.8210642254657347,
+      "learning_rate": 2e-05,
+      "loss": 0.6138957500457763,
+      "step": 3530
+    },
+    {
+      "epoch": 0.26932440657334145,
+      "grad_norm": 1.0521207645749941,
+      "learning_rate": 2e-05,
+      "loss": 0.6080638885498046,
+      "step": 3540
+    },
+    {
+      "epoch": 0.27008520998174074,
+      "grad_norm": 6.735721511812288,
+      "learning_rate": 2e-05,
+      "loss": 0.6256732940673828,
+      "step": 3550
+    },
+    {
+      "epoch": 0.27084601339014,
+      "grad_norm": 2.799485569675146,
+      "learning_rate": 2e-05,
+      "loss": 0.5936986923217773,
+      "step": 3560
+    },
+    {
+      "epoch": 0.27160681679853926,
+      "grad_norm": 0.9111946254259832,
+      "learning_rate": 2e-05,
+      "loss": 0.6204559326171875,
+      "step": 3570
+    },
+    {
+      "epoch": 0.27236762020693855,
+      "grad_norm": 1.8582172885915664,
+      "learning_rate": 2e-05,
+      "loss": 0.615900707244873,
+      "step": 3580
+    },
+    {
+      "epoch": 0.2731284236153378,
+      "grad_norm": 0.7739816201039059,
+      "learning_rate": 2e-05,
+      "loss": 0.5894556999206543,
+      "step": 3590
+    },
+    {
+      "epoch": 0.2738892270237371,
+      "grad_norm": 1.441208937322419,
+      "learning_rate": 2e-05,
+      "loss": 0.6097617149353027,
+      "step": 3600
+    },
+    {
+      "epoch": 0.2746500304321363,
+      "grad_norm": 0.9465929067889907,
+      "learning_rate": 2e-05,
+      "loss": 0.6167520046234131,
+      "step": 3610
+    },
+    {
+      "epoch": 0.2754108338405356,
+      "grad_norm": 0.9840465275588058,
+      "learning_rate": 2e-05,
+      "loss": 0.6279431819915772,
+      "step": 3620
+    },
+    {
+      "epoch": 0.2761716372489349,
+      "grad_norm": 1.6678784102631918,
+      "learning_rate": 2e-05,
+      "loss": 0.6081015586853027,
+      "step": 3630
+    },
+    {
+      "epoch": 0.27693244065733413,
+      "grad_norm": 0.8555695729438972,
+      "learning_rate": 2e-05,
+      "loss": 0.6070639610290527,
+      "step": 3640
+    },
+    {
+      "epoch": 0.2776932440657334,
+      "grad_norm": 1.756659504956261,
+      "learning_rate": 2e-05,
+      "loss": 0.6191421508789062,
+      "step": 3650
+    },
+    {
+      "epoch": 0.27845404747413266,
+      "grad_norm": 1.2982748593593505,
+      "learning_rate": 2e-05,
+      "loss": 0.6194043636322022,
+      "step": 3660
+    },
+    {
+      "epoch": 0.27921485088253195,
+      "grad_norm": 0.7922690486284613,
+      "learning_rate": 2e-05,
+      "loss": 0.5996121883392334,
+      "step": 3670
+    },
+    {
+      "epoch": 0.27997565429093124,
+      "grad_norm": 0.8341378633449698,
+      "learning_rate": 2e-05,
+      "loss": 0.6137380123138427,
+      "step": 3680
+    },
+    {
+      "epoch": 0.2807364576993305,
+      "grad_norm": 1.2791128722007485,
+      "learning_rate": 2e-05,
+      "loss": 0.6054240703582764,
+      "step": 3690
+    },
+    {
+      "epoch": 0.28149726110772977,
+      "grad_norm": 0.8939937791977226,
+      "learning_rate": 2e-05,
+      "loss": 0.6308682918548584,
+      "step": 3700
+    },
+    {
+      "epoch": 0.28225806451612906,
+      "grad_norm": 3.628759918703947,
+      "learning_rate": 2e-05,
+      "loss": 0.6237603664398194,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2830188679245283,
+      "grad_norm": 4.577204658525299,
+      "learning_rate": 2e-05,
+      "loss": 0.6311593532562256,
+      "step": 3720
+    },
+    {
+      "epoch": 0.2837796713329276,
+      "grad_norm": 0.8948798691151667,
+      "learning_rate": 2e-05,
+      "loss": 0.608569860458374,
+      "step": 3730
+    },
+    {
+      "epoch": 0.2845404747413268,
+      "grad_norm": 0.9194449656139297,
+      "learning_rate": 2e-05,
+      "loss": 0.6357901573181153,
+      "step": 3740
+    },
+    {
+      "epoch": 0.2853012781497261,
+      "grad_norm": 0.8171348243537644,
+      "learning_rate": 2e-05,
+      "loss": 0.613353681564331,
+      "step": 3750
+    },
+    {
+      "epoch": 0.2860620815581254,
+      "grad_norm": 1.013927897669978,
+      "learning_rate": 2e-05,
+      "loss": 0.6140854835510254,
+      "step": 3760
+    },
+    {
+      "epoch": 0.28682288496652464,
+      "grad_norm": 0.7210071956307897,
+      "learning_rate": 2e-05,
+      "loss": 0.6217946529388427,
+      "step": 3770
+    },
+    {
+      "epoch": 0.28758368837492393,
+      "grad_norm": 0.8011534818888779,
+      "learning_rate": 2e-05,
+      "loss": 0.5974589347839355,
+      "step": 3780
+    },
+    {
+      "epoch": 0.28834449178332316,
+      "grad_norm": 0.9011845655705722,
+      "learning_rate": 2e-05,
+      "loss": 0.6087424278259277,
+      "step": 3790
+    },
+    {
+      "epoch": 0.28910529519172246,
+      "grad_norm": 0.8814420360391584,
+      "learning_rate": 2e-05,
+      "loss": 0.6011880874633789,
+      "step": 3800
+    },
+    {
+      "epoch": 0.28986609860012175,
+      "grad_norm": 0.8315296955618288,
+      "learning_rate": 2e-05,
+      "loss": 0.5989766120910645,
+      "step": 3810
+    },
+    {
+      "epoch": 0.290626902008521,
+      "grad_norm": 0.8120554977139738,
+      "learning_rate": 2e-05,
+      "loss": 0.6353487968444824,
+      "step": 3820
+    },
+    {
+      "epoch": 0.2913877054169203,
+      "grad_norm": 2.682256590029768,
+      "learning_rate": 2e-05,
+      "loss": 0.5951233863830566,
+      "step": 3830
+    },
+    {
+      "epoch": 0.29214850882531956,
+      "grad_norm": 1.5678136476459417,
+      "learning_rate": 2e-05,
+      "loss": 0.6109321594238282,
+      "step": 3840
+    },
+    {
+      "epoch": 0.2929093122337188,
+      "grad_norm": 1.6518908694551735,
+      "learning_rate": 2e-05,
+      "loss": 0.6007917404174805,
+      "step": 3850
+    },
+    {
+      "epoch": 0.2936701156421181,
+      "grad_norm": 0.8390332769455142,
+      "learning_rate": 2e-05,
+      "loss": 0.5966996192932129,
+      "step": 3860
+    },
+    {
+      "epoch": 0.2944309190505173,
+      "grad_norm": 0.9939817097512031,
+      "learning_rate": 2e-05,
+      "loss": 0.6230123519897461,
+      "step": 3870
+    },
+    {
+      "epoch": 0.2951917224589166,
+      "grad_norm": 0.8449526081587111,
+      "learning_rate": 2e-05,
+      "loss": 0.6368628978729248,
+      "step": 3880
+    },
+    {
+      "epoch": 0.2959525258673159,
+      "grad_norm": 1.3118101691079025,
+      "learning_rate": 2e-05,
+      "loss": 0.6220353603363037,
+      "step": 3890
+    },
+    {
+      "epoch": 0.29671332927571514,
+      "grad_norm": 0.978940167656836,
+      "learning_rate": 2e-05,
+      "loss": 0.6155681610107422,
+      "step": 3900
+    },
+    {
+      "epoch": 0.29747413268411443,
+      "grad_norm": 1.7975965748511273,
+      "learning_rate": 2e-05,
+      "loss": 0.5918474197387695,
+      "step": 3910
+    },
+    {
+      "epoch": 0.29823493609251367,
+      "grad_norm": 1.6706236390444285,
+      "learning_rate": 2e-05,
+      "loss": 0.6126898765563965,
+      "step": 3920
+    },
+    {
+      "epoch": 0.29899573950091296,
+      "grad_norm": 0.8955060435507843,
+      "learning_rate": 2e-05,
+      "loss": 0.5828713417053223,
+      "step": 3930
+    },
+    {
+      "epoch": 0.29975654290931225,
+      "grad_norm": 0.9057540496762859,
+      "learning_rate": 2e-05,
+      "loss": 0.6285027980804443,
+      "step": 3940
+    },
+    {
+      "epoch": 0.3005173463177115,
+      "grad_norm": 0.8513770306376237,
+      "learning_rate": 2e-05,
+      "loss": 0.5995490074157714,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3012781497261108,
+      "grad_norm": 0.9018811929887697,
+      "learning_rate": 2e-05,
+      "loss": 0.6095849514007569,
+      "step": 3960
+    },
+    {
+      "epoch": 0.30203895313451007,
+      "grad_norm": 0.7118554306232184,
+      "learning_rate": 2e-05,
+      "loss": 0.6040401458740234,
+      "step": 3970
+    },
+    {
+      "epoch": 0.3027997565429093,
+      "grad_norm": 0.9071491300673238,
+      "learning_rate": 2e-05,
+      "loss": 0.6258529186248779,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3035605599513086,
+      "grad_norm": 0.9038005131444103,
+      "learning_rate": 2e-05,
+      "loss": 0.6107657432556153,
+      "step": 3990
+    },
+    {
+      "epoch": 0.30432136335970783,
+      "grad_norm": 0.9133874285340832,
+      "learning_rate": 2e-05,
+      "loss": 0.6253015518188476,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3050821667681071,
+      "grad_norm": 1.1022244895474538,
+      "learning_rate": 2e-05,
+      "loss": 0.5970585823059082,
+      "step": 4010
+    },
+    {
+      "epoch": 0.3058429701765064,
+      "grad_norm": 0.982933929694812,
+      "learning_rate": 2e-05,
+      "loss": 0.610454511642456,
+      "step": 4020
+    },
+    {
+      "epoch": 0.30660377358490565,
+      "grad_norm": 6.471445748327421,
+      "learning_rate": 2e-05,
+      "loss": 0.5650215148925781,
+      "step": 4030
+    },
+    {
+      "epoch": 0.30736457699330494,
+      "grad_norm": 1.1515240977943098,
+      "learning_rate": 2e-05,
+      "loss": 0.6278285026550293,
+      "step": 4040
+    },
+    {
+      "epoch": 0.3081253804017042,
+      "grad_norm": 2.457199029245135,
+      "learning_rate": 2e-05,
+      "loss": 0.6022915840148926,
+      "step": 4050
+    },
+    {
+      "epoch": 0.30888618381010347,
+      "grad_norm": 0.7407605190468444,
+      "learning_rate": 2e-05,
+      "loss": 0.6401365280151368,
+      "step": 4060
+    },
+    {
+      "epoch": 0.30964698721850276,
+      "grad_norm": 0.8695411167792898,
+      "learning_rate": 2e-05,
+      "loss": 0.5914569854736328,
+      "step": 4070
+    },
+    {
+      "epoch": 0.310407790626902,
+      "grad_norm": 1.1123525656325708,
+      "learning_rate": 2e-05,
+      "loss": 0.6278794288635254,
+      "step": 4080
+    },
+    {
+      "epoch": 0.3111685940353013,
+      "grad_norm": 0.8089054567791197,
+      "learning_rate": 2e-05,
+      "loss": 0.5776612281799316,
+      "step": 4090
+    },
+    {
+      "epoch": 0.3119293974437006,
+      "grad_norm": 1.44682150408943,
+      "learning_rate": 2e-05,
+      "loss": 0.6120601177215577,
+      "step": 4100
+    },
+    {
+      "epoch": 0.3126902008520998,
+      "grad_norm": 1.2477506247557861,
+      "learning_rate": 2e-05,
+      "loss": 0.6243569374084472,
+      "step": 4110
+    },
+    {
+      "epoch": 0.3134510042604991,
+      "grad_norm": 1.452737770482452,
+      "learning_rate": 2e-05,
+      "loss": 0.6241071224212646,
+      "step": 4120
+    },
+    {
+      "epoch": 0.31421180766889834,
+      "grad_norm": 0.7363949927071634,
+      "learning_rate": 2e-05,
+      "loss": 0.615998649597168,
+      "step": 4130
+    },
+    {
+      "epoch": 0.3149726110772976,
+      "grad_norm": 0.9904461219747637,
+      "learning_rate": 2e-05,
+      "loss": 0.5982001304626465,
+      "step": 4140
+    },
+    {
+      "epoch": 0.3157334144856969,
+      "grad_norm": 2.6307432080351836,
+      "learning_rate": 2e-05,
+      "loss": 0.6170999526977539,
+      "step": 4150
+    },
+    {
+      "epoch": 0.31649421789409615,
+      "grad_norm": 1.3103833934588613,
+      "learning_rate": 2e-05,
+      "loss": 0.616608715057373,
+      "step": 4160
+    },
+    {
+      "epoch": 0.31725502130249544,
+      "grad_norm": 1.642585084301309,
+      "learning_rate": 2e-05,
+      "loss": 0.6181431770324707,
+      "step": 4170
+    },
+    {
+      "epoch": 0.3180158247108947,
+      "grad_norm": 1.3358344049655098,
+      "learning_rate": 2e-05,
+      "loss": 0.5898998260498047,
+      "step": 4180
+    },
+    {
+      "epoch": 0.31877662811929397,
+      "grad_norm": 2.033036923101361,
+      "learning_rate": 2e-05,
+      "loss": 0.6010503768920898,
+      "step": 4190
+    },
+    {
+      "epoch": 0.31953743152769326,
+      "grad_norm": 1.0048016779941806,
+      "learning_rate": 2e-05,
+      "loss": 0.6192368984222412,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3202982349360925,
+      "grad_norm": 1.5318421777607685,
+      "learning_rate": 2e-05,
+      "loss": 0.6008861541748047,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3210590383444918,
+      "grad_norm": 1.0073799174677105,
+      "learning_rate": 2e-05,
+      "loss": 0.6188548088073731,
+      "step": 4220
+    },
+    {
+      "epoch": 0.3218198417528911,
+      "grad_norm": 0.8283343412424335,
+      "learning_rate": 2e-05,
+      "loss": 0.6003226280212403,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 0.7571380810077775,
+      "learning_rate": 2e-05,
+      "loss": 0.6002717971801758,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3233414485696896,
+      "grad_norm": 0.8338609352847893,
+      "learning_rate": 2e-05,
+      "loss": 0.6231700897216796,
+      "step": 4250
+    },
+    {
+      "epoch": 0.32410225197808884,
+      "grad_norm": 0.740619566155922,
+      "learning_rate": 2e-05,
+      "loss": 0.6033841133117676,
+      "step": 4260
+    },
+    {
+      "epoch": 0.32486305538648813,
+      "grad_norm": 0.8956374906539315,
+      "learning_rate": 2e-05,
+      "loss": 0.6189849853515625,
+      "step": 4270
+    },
+    {
+      "epoch": 0.3256238587948874,
+      "grad_norm": 1.409026605977312,
+      "learning_rate": 2e-05,
+      "loss": 0.6028616905212403,
+      "step": 4280
+    },
+    {
+      "epoch": 0.32638466220328666,
+      "grad_norm": 2.453626744827294,
+      "learning_rate": 2e-05,
+      "loss": 0.6057555675506592,
+      "step": 4290
+    },
+    {
+      "epoch": 0.32714546561168595,
+      "grad_norm": 1.3959026927627025,
+      "learning_rate": 2e-05,
+      "loss": 0.6028185844421386,
+      "step": 4300
+    },
+    {
+      "epoch": 0.3279062690200852,
+      "grad_norm": 0.7577927271333064,
+      "learning_rate": 2e-05,
+      "loss": 0.6127864360809326,
+      "step": 4310
+    },
+    {
+      "epoch": 0.3286670724284845,
+      "grad_norm": 0.9447217279629392,
+      "learning_rate": 2e-05,
+      "loss": 0.6024003982543945,
+      "step": 4320
+    },
+    {
+      "epoch": 0.32942787583688377,
+      "grad_norm": 0.9272576340694912,
+      "learning_rate": 2e-05,
+      "loss": 0.6185897827148438,
+      "step": 4330
+    },
+    {
+      "epoch": 0.330188679245283,
+      "grad_norm": 0.7119916555178322,
+      "learning_rate": 2e-05,
+      "loss": 0.6182665824890137,
+      "step": 4340
+    },
+    {
+      "epoch": 0.3309494826536823,
+      "grad_norm": 5.95345406970356,
+      "learning_rate": 2e-05,
+      "loss": 0.6071166038513184,
+      "step": 4350
+    },
+    {
+      "epoch": 0.3317102860620816,
+      "grad_norm": 1.4215648692146343,
+      "learning_rate": 2e-05,
+      "loss": 0.6039448738098144,
+      "step": 4360
+    },
+    {
+      "epoch": 0.3324710894704808,
+      "grad_norm": 0.7570623558291363,
+      "learning_rate": 2e-05,
+      "loss": 0.5986960887908935,
+      "step": 4370
+    },
+    {
+      "epoch": 0.3332318928788801,
+      "grad_norm": 1.0228573337145783,
+      "learning_rate": 2e-05,
+      "loss": 0.6171799659729004,
+      "step": 4380
+    },
+    {
+      "epoch": 0.33399269628727934,
+      "grad_norm": 0.9028857839163633,
+      "learning_rate": 2e-05,
+      "loss": 0.6072957515716553,
+      "step": 4390
+    },
+    {
+      "epoch": 0.33475349969567864,
+      "grad_norm": 0.7415941430991778,
+      "learning_rate": 2e-05,
+      "loss": 0.6106678962707519,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3355143031040779,
+      "grad_norm": 0.7411236910221118,
+      "learning_rate": 2e-05,
+      "loss": 0.5931798934936523,
+      "step": 4410
+    },
+    {
+      "epoch": 0.33627510651247716,
+      "grad_norm": 1.1774708021652436,
+      "learning_rate": 2e-05,
+      "loss": 0.6152263164520264,
+      "step": 4420
+    },
+    {
+      "epoch": 0.33703590992087645,
+      "grad_norm": 0.8505680675812167,
+      "learning_rate": 2e-05,
+      "loss": 0.6033729553222656,
+      "step": 4430
+    },
+    {
+      "epoch": 0.3377967133292757,
+      "grad_norm": 0.7891820178944161,
+      "learning_rate": 2e-05,
+      "loss": 0.6123302459716797,
+      "step": 4440
+    },
+    {
+      "epoch": 0.338557516737675,
+      "grad_norm": 0.866079289805027,
+      "learning_rate": 2e-05,
+      "loss": 0.6025050163269043,
+      "step": 4450
+    },
+    {
+      "epoch": 0.33931832014607427,
+      "grad_norm": 0.7241976779424538,
+      "learning_rate": 2e-05,
+      "loss": 0.612579345703125,
+      "step": 4460
+    },
+    {
+      "epoch": 0.3400791235544735,
+      "grad_norm": 0.7966629670827999,
+      "learning_rate": 2e-05,
+      "loss": 0.5974034786224365,
+      "step": 4470
+    },
+    {
+      "epoch": 0.3408399269628728,
+      "grad_norm": 0.7767550667528098,
+      "learning_rate": 2e-05,
+      "loss": 0.6011282920837402,
+      "step": 4480
+    },
+    {
+      "epoch": 0.3416007303712721,
+      "grad_norm": 1.8420521999393737,
+      "learning_rate": 2e-05,
+      "loss": 0.6026616573333741,
+      "step": 4490
+    },
+    {
+      "epoch": 0.3423615337796713,
+      "grad_norm": 1.3139173936334982,
+      "learning_rate": 2e-05,
+      "loss": 0.6053057670593261,
+      "step": 4500
+    },
+    {
+      "epoch": 0.3431223371880706,
+      "grad_norm": 1.022314820431628,
+      "learning_rate": 2e-05,
+      "loss": 0.583978271484375,
+      "step": 4510
+    },
+    {
+      "epoch": 0.34388314059646985,
+      "grad_norm": 0.7165802493976553,
+      "learning_rate": 2e-05,
+      "loss": 0.6061666488647461,
+      "step": 4520
+    },
+    {
+      "epoch": 0.34464394400486914,
+      "grad_norm": 0.747468299503708,
+      "learning_rate": 2e-05,
+      "loss": 0.5799343109130859,
+      "step": 4530
+    },
+    {
+      "epoch": 0.34540474741326843,
+      "grad_norm": 0.8132856915849325,
+      "learning_rate": 2e-05,
+      "loss": 0.6095840454101562,
+      "step": 4540
+    },
+    {
+      "epoch": 0.34616555082166767,
+      "grad_norm": 0.7027943277436396,
+      "learning_rate": 2e-05,
+      "loss": 0.6092688560485839,
+      "step": 4550
+    },
+    {
+      "epoch": 0.34692635423006696,
+      "grad_norm": 1.1086210965087457,
+      "learning_rate": 2e-05,
+      "loss": 0.5932141304016113,
+      "step": 4560
+    },
+    {
+      "epoch": 0.3476871576384662,
+      "grad_norm": 0.8852791797840133,
+      "learning_rate": 2e-05,
+      "loss": 0.5923095703125,
+      "step": 4570
+    },
+    {
+      "epoch": 0.3484479610468655,
+      "grad_norm": 1.6353079953312049,
+      "learning_rate": 2e-05,
+      "loss": 0.6154132843017578,
+      "step": 4580
+    },
+    {
+      "epoch": 0.3492087644552648,
+      "grad_norm": 1.1789394247020066,
+      "learning_rate": 2e-05,
+      "loss": 0.6273578643798828,
+      "step": 4590
+    },
+    {
+      "epoch": 0.349969567863664,
+      "grad_norm": 1.4102805493635073,
+      "learning_rate": 2e-05,
+      "loss": 0.6174351692199707,
+      "step": 4600
+    },
+    {
+      "epoch": 0.3507303712720633,
+      "grad_norm": 3.4182331641961428,
+      "learning_rate": 2e-05,
+      "loss": 0.5923943996429444,
+      "step": 4610
+    },
+    {
+      "epoch": 0.3514911746804626,
+      "grad_norm": 0.9392991206242777,
+      "learning_rate": 2e-05,
+      "loss": 0.5950442790985108,
+      "step": 4620
+    },
+    {
+      "epoch": 0.35225197808886183,
+      "grad_norm": 0.84844491388597,
+      "learning_rate": 2e-05,
+      "loss": 0.5990563869476319,
+      "step": 4630
+    },
+    {
+      "epoch": 0.3530127814972611,
+      "grad_norm": 0.7558327897640099,
+      "learning_rate": 2e-05,
+      "loss": 0.5872565746307373,
+      "step": 4640
+    },
+    {
+      "epoch": 0.35377358490566035,
+      "grad_norm": 0.786357619659807,
+      "learning_rate": 2e-05,
+      "loss": 0.5881314277648926,
+      "step": 4650
+    },
+    {
+      "epoch": 0.35453438831405965,
+      "grad_norm": 0.7946945273066692,
+      "learning_rate": 2e-05,
+      "loss": 0.5885752677917481,
+      "step": 4660
+    },
+    {
+      "epoch": 0.35529519172245894,
+      "grad_norm": 0.8509812786655356,
+      "learning_rate": 2e-05,
+      "loss": 0.5973938941955567,
+      "step": 4670
+    },
+    {
+      "epoch": 0.35605599513085817,
+      "grad_norm": 1.192996958876806,
+      "learning_rate": 2e-05,
+      "loss": 0.5827082633972168,
+      "step": 4680
+    },
+    {
+      "epoch": 0.35681679853925746,
+      "grad_norm": 1.0699080043046383,
+      "learning_rate": 2e-05,
+      "loss": 0.6129659652709961,
+      "step": 4690
+    },
+    {
+      "epoch": 0.3575776019476567,
+      "grad_norm": 0.8727409547698386,
+      "learning_rate": 2e-05,
+      "loss": 0.58746337890625,
+      "step": 4700
+    },
+    {
+      "epoch": 0.358338405356056,
+      "grad_norm": 5.43225571997352,
+      "learning_rate": 2e-05,
+      "loss": 0.6099301338195801,
+      "step": 4710
+    },
+    {
+      "epoch": 0.3590992087644553,
+      "grad_norm": 0.9887457028131443,
+      "learning_rate": 2e-05,
+      "loss": 0.607042646408081,
+      "step": 4720
+    },
+    {
+      "epoch": 0.3598600121728545,
+      "grad_norm": 1.2046804745263027,
+      "learning_rate": 2e-05,
+      "loss": 0.5955885887145996,
+      "step": 4730
+    },
+    {
+      "epoch": 0.3606208155812538,
+      "grad_norm": 0.7639829739280138,
+      "learning_rate": 2e-05,
+      "loss": 0.6083533763885498,
+      "step": 4740
+    },
+    {
+      "epoch": 0.3613816189896531,
+      "grad_norm": 0.7267500781357502,
+      "learning_rate": 2e-05,
+      "loss": 0.60111403465271,
+      "step": 4750
+    },
+    {
+      "epoch": 0.36214242239805233,
+      "grad_norm": 0.8710289774828294,
+      "learning_rate": 2e-05,
+      "loss": 0.6345517635345459,
+      "step": 4760
+    },
+    {
+      "epoch": 0.3629032258064516,
+      "grad_norm": 1.0637000129245222,
+      "learning_rate": 2e-05,
+      "loss": 0.6069129943847656,
+      "step": 4770
+    },
+    {
+      "epoch": 0.36366402921485086,
+      "grad_norm": 1.8024452267660673,
+      "learning_rate": 2e-05,
+      "loss": 0.6435677528381347,
+      "step": 4780
+    },
+    {
+      "epoch": 0.36442483262325015,
+      "grad_norm": 0.8346426924240165,
+      "learning_rate": 2e-05,
+      "loss": 0.5989354133605957,
+      "step": 4790
+    },
+    {
+      "epoch": 0.36518563603164944,
+      "grad_norm": 0.6792142781631723,
+      "learning_rate": 2e-05,
+      "loss": 0.6012043952941895,
+      "step": 4800
+    },
+    {
+      "epoch": 0.3659464394400487,
+      "grad_norm": 0.8409837452079166,
+      "learning_rate": 2e-05,
+      "loss": 0.6044680595397949,
+      "step": 4810
+    },
+    {
+      "epoch": 0.36670724284844797,
+      "grad_norm": 0.8644602675103753,
+      "learning_rate": 2e-05,
+      "loss": 0.6201369762420654,
+      "step": 4820
+    },
+    {
+      "epoch": 0.36746804625684726,
+      "grad_norm": 1.580222949710994,
+      "learning_rate": 2e-05,
+      "loss": 0.5904644012451172,
+      "step": 4830
+    },
+    {
+      "epoch": 0.3682288496652465,
+      "grad_norm": 1.1928080164601287,
+      "learning_rate": 2e-05,
+      "loss": 0.6086296081542969,
+      "step": 4840
+    },
+    {
+      "epoch": 0.3689896530736458,
+      "grad_norm": 0.689264603926416,
+      "learning_rate": 2e-05,
+      "loss": 0.5967980861663819,
+      "step": 4850
+    },
+    {
+      "epoch": 0.369750456482045,
+      "grad_norm": 0.8126776376611288,
+      "learning_rate": 2e-05,
+      "loss": 0.5855744361877442,
+      "step": 4860
+    },
+    {
+      "epoch": 0.3705112598904443,
+      "grad_norm": 1.092356779204631,
+      "learning_rate": 2e-05,
+      "loss": 0.6178097724914551,
+      "step": 4870
+    },
+    {
+      "epoch": 0.3712720632988436,
+      "grad_norm": 0.9622832608857461,
+      "learning_rate": 2e-05,
+      "loss": 0.617055082321167,
+      "step": 4880
+    },
+    {
+      "epoch": 0.37203286670724284,
+      "grad_norm": 0.7618052699113795,
+      "learning_rate": 2e-05,
+      "loss": 0.602293586730957,
+      "step": 4890
+    },
+    {
+      "epoch": 0.37279367011564213,
+      "grad_norm": 1.3489861276357102,
+      "learning_rate": 2e-05,
+      "loss": 0.5918971538543701,
+      "step": 4900
+    },
+    {
+      "epoch": 0.37355447352404136,
+      "grad_norm": 1.7707644440212729,
+      "learning_rate": 2e-05,
+      "loss": 0.625973892211914,
+      "step": 4910
+    },
+    {
+      "epoch": 0.37431527693244065,
+      "grad_norm": 15.014705035598531,
+      "learning_rate": 2e-05,
+      "loss": 0.6238719463348389,
+      "step": 4920
+    },
+    {
+      "epoch": 0.37507608034083995,
+      "grad_norm": 1.1670299633789776,
+      "learning_rate": 2e-05,
+      "loss": 0.6075002193450928,
+      "step": 4930
+    },
+    {
+      "epoch": 0.3758368837492392,
+      "grad_norm": 0.8549851050612812,
+      "learning_rate": 2e-05,
+      "loss": 0.6109541893005371,
+      "step": 4940
+    },
+    {
+      "epoch": 0.37659768715763847,
+      "grad_norm": 0.911106365991643,
+      "learning_rate": 2e-05,
+      "loss": 0.6021104335784913,
+      "step": 4950
+    },
+    {
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.8650943358256169,
+      "learning_rate": 2e-05,
+      "loss": 0.5897356033325195,
+      "step": 4960
+    },
+    {
+      "epoch": 0.378119293974437,
+      "grad_norm": 1.2102016223127976,
+      "learning_rate": 2e-05,
+      "loss": 0.6001923561096192,
+      "step": 4970
+    },
+    {
+      "epoch": 0.3788800973828363,
+      "grad_norm": 1.159865181760775,
+      "learning_rate": 2e-05,
+      "loss": 0.6132318496704101,
+      "step": 4980
+    },
+    {
+      "epoch": 0.3796409007912355,
+      "grad_norm": 1.9290598997736847,
+      "learning_rate": 2e-05,
+      "loss": 0.6039675712585449,
+      "step": 4990
+    },
+    {
+      "epoch": 0.3804017041996348,
+      "grad_norm": 0.8225860734261352,
+      "learning_rate": 2e-05,
+      "loss": 0.5970118522644043,
+      "step": 5000
+    },
+    {
+      "epoch": 0.3811625076080341,
+      "grad_norm": 0.8026146605102198,
+      "learning_rate": 2e-05,
+      "loss": 0.6008700847625732,
+      "step": 5010
+    },
+    {
+      "epoch": 0.38192331101643334,
+      "grad_norm": 0.9141026787845802,
+      "learning_rate": 2e-05,
+      "loss": 0.6239216804504395,
+      "step": 5020
+    },
+    {
+      "epoch": 0.38268411442483263,
+      "grad_norm": 3.558820995930603,
+      "learning_rate": 2e-05,
+      "loss": 0.6232359886169434,
+      "step": 5030
+    },
+    {
+      "epoch": 0.38344491783323187,
+      "grad_norm": 1.4087261604377643,
+      "learning_rate": 2e-05,
+      "loss": 0.5870172023773194,
+      "step": 5040
+    },
+    {
+      "epoch": 0.38420572124163116,
+      "grad_norm": 0.8660084567723209,
+      "learning_rate": 2e-05,
+      "loss": 0.6123439311981201,
+      "step": 5050
+    },
+    {
+      "epoch": 0.38496652465003045,
+      "grad_norm": 0.8358289715878096,
+      "learning_rate": 2e-05,
+      "loss": 0.5848734855651856,
+      "step": 5060
+    },
+    {
+      "epoch": 0.3857273280584297,
+      "grad_norm": 0.9371082715485197,
+      "learning_rate": 2e-05,
+      "loss": 0.5995091915130615,
+      "step": 5070
+    },
+    {
+      "epoch": 0.386488131466829,
+      "grad_norm": 0.8722665925481548,
+      "learning_rate": 2e-05,
+      "loss": 0.5924876689910888,
+      "step": 5080
+    },
+    {
+      "epoch": 0.38724893487522827,
+      "grad_norm": 0.9824630703006527,
+      "learning_rate": 2e-05,
+      "loss": 0.5860276222229004,
+      "step": 5090
+    },
+    {
+      "epoch": 0.3880097382836275,
+      "grad_norm": 0.8775718696313889,
+      "learning_rate": 2e-05,
+      "loss": 0.6092318534851074,
+      "step": 5100
+    },
+    {
+      "epoch": 0.3887705416920268,
+      "grad_norm": 0.896842834039103,
+      "learning_rate": 2e-05,
+      "loss": 0.5805525779724121,
+      "step": 5110
+    },
+    {
+      "epoch": 0.38953134510042603,
+      "grad_norm": 0.8024845982821228,
+      "learning_rate": 2e-05,
+      "loss": 0.593732213973999,
+      "step": 5120
+    },
+    {
+      "epoch": 0.3902921485088253,
+      "grad_norm": 0.7971370813826159,
+      "learning_rate": 2e-05,
+      "loss": 0.6110887050628662,
+      "step": 5130
+    },
+    {
+      "epoch": 0.3910529519172246,
+      "grad_norm": 0.91881848531298,
+      "learning_rate": 2e-05,
+      "loss": 0.5985256195068359,
+      "step": 5140
+    },
+    {
+      "epoch": 0.39181375532562385,
+      "grad_norm": 0.6876107219796312,
+      "learning_rate": 2e-05,
+      "loss": 0.6060726165771484,
+      "step": 5150
+    },
+    {
+      "epoch": 0.39257455873402314,
+      "grad_norm": 1.6839180915456682,
+      "learning_rate": 2e-05,
+      "loss": 0.6001443862915039,
+      "step": 5160
+    },
+    {
+      "epoch": 0.3933353621424224,
+      "grad_norm": 0.7113463153589865,
+      "learning_rate": 2e-05,
+      "loss": 0.6014138698577881,
+      "step": 5170
+    },
+    {
+      "epoch": 0.39409616555082166,
+      "grad_norm": 0.7468840028924791,
+      "learning_rate": 2e-05,
+      "loss": 0.5944113731384277,
+      "step": 5180
+    },
+    {
+      "epoch": 0.39485696895922096,
+      "grad_norm": 0.7545155751266641,
+      "learning_rate": 2e-05,
+      "loss": 0.5950839042663574,
+      "step": 5190
+    },
+    {
+      "epoch": 0.3956177723676202,
+      "grad_norm": 0.6365755560940038,
+      "learning_rate": 2e-05,
+      "loss": 0.5823196411132813,
+      "step": 5200
+    },
+    {
+      "epoch": 0.3963785757760195,
+      "grad_norm": 0.7475025526272886,
+      "learning_rate": 2e-05,
+      "loss": 0.5859226226806641,
+      "step": 5210
+    },
+    {
+      "epoch": 0.3971393791844188,
+      "grad_norm": 0.7255503054991962,
+      "learning_rate": 2e-05,
+      "loss": 0.6008688926696777,
+      "step": 5220
+    },
+    {
+      "epoch": 0.397900182592818,
+      "grad_norm": 2.7333409571917864,
+      "learning_rate": 2e-05,
+      "loss": 0.5843379020690918,
+      "step": 5230
+    },
+    {
+      "epoch": 0.3986609860012173,
+      "grad_norm": 0.7062828892372515,
+      "learning_rate": 2e-05,
+      "loss": 0.5923688411712646,
+      "step": 5240
+    },
+    {
+      "epoch": 0.39942178940961653,
+      "grad_norm": 0.8974625937429299,
+      "learning_rate": 2e-05,
+      "loss": 0.5858447074890136,
+      "step": 5250
+    },
+    {
+      "epoch": 0.4001825928180158,
+      "grad_norm": 1.073381149354679,
+      "learning_rate": 2e-05,
+      "loss": 0.592432689666748,
+      "step": 5260
+    },
+    {
+      "epoch": 0.4009433962264151,
+      "grad_norm": 0.9109304806645849,
+      "learning_rate": 2e-05,
+      "loss": 0.5865280151367187,
+      "step": 5270
+    },
+    {
+      "epoch": 0.40170419963481435,
+      "grad_norm": 0.8237646894259731,
+      "learning_rate": 2e-05,
+      "loss": 0.6011100292205811,
+      "step": 5280
+    },
+    {
+      "epoch": 0.40246500304321364,
+      "grad_norm": 1.7372550844815882,
+      "learning_rate": 2e-05,
+      "loss": 0.5879167079925537,
+      "step": 5290
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 1.6889445659821876,
+      "learning_rate": 2e-05,
+      "loss": 0.5934720993041992,
+      "step": 5300
+    },
+    {
+      "epoch": 0.40398660986001217,
+      "grad_norm": 1.3877793557496096,
+      "learning_rate": 2e-05,
+      "loss": 0.6005355834960937,
+      "step": 5310
+    },
+    {
+      "epoch": 0.40474741326841146,
+      "grad_norm": 0.8192199174093711,
+      "learning_rate": 2e-05,
+      "loss": 0.6198990821838379,
+      "step": 5320
+    },
+    {
+      "epoch": 0.4055082166768107,
+      "grad_norm": 0.828907519504877,
+      "learning_rate": 2e-05,
+      "loss": 0.5916791915893554,
+      "step": 5330
+    },
+    {
+      "epoch": 0.40626902008521,
+      "grad_norm": 0.8858694195086637,
+      "learning_rate": 2e-05,
+      "loss": 0.5733376979827881,
+      "step": 5340
+    },
+    {
+      "epoch": 0.4070298234936093,
+      "grad_norm": 0.8465092926463462,
+      "learning_rate": 2e-05,
+      "loss": 0.6105677604675293,
+      "step": 5350
+    },
+    {
+      "epoch": 0.4077906269020085,
+      "grad_norm": 1.0677300074481022,
+      "learning_rate": 2e-05,
+      "loss": 0.5897644996643067,
+      "step": 5360
+    },
+    {
+      "epoch": 0.4085514303104078,
+      "grad_norm": 0.8532131664316742,
+      "learning_rate": 2e-05,
+      "loss": 0.5866693019866943,
+      "step": 5370
+    },
+    {
+      "epoch": 0.40931223371880704,
+      "grad_norm": 1.2106588433264815,
+      "learning_rate": 2e-05,
+      "loss": 0.6188040733337402,
+      "step": 5380
+    },
+    {
+      "epoch": 0.41007303712720633,
+      "grad_norm": 0.7567863457480009,
+      "learning_rate": 2e-05,
+      "loss": 0.6052532196044922,
+      "step": 5390
+    },
+    {
+      "epoch": 0.4108338405356056,
+      "grad_norm": 3.007495424917002,
+      "learning_rate": 2e-05,
+      "loss": 0.5760692119598388,
+      "step": 5400
+    },
+    {
+      "epoch": 0.41159464394400486,
+      "grad_norm": 0.6878625728039528,
+      "learning_rate": 2e-05,
+      "loss": 0.6026056289672852,
+      "step": 5410
+    },
+    {
+      "epoch": 0.41235544735240415,
+      "grad_norm": 0.8617903657380879,
+      "learning_rate": 2e-05,
+      "loss": 0.6164340496063232,
+      "step": 5420
+    },
+    {
+      "epoch": 0.4131162507608034,
+      "grad_norm": 0.7828782316642504,
+      "learning_rate": 2e-05,
+      "loss": 0.6123809337615966,
+      "step": 5430
+    },
+    {
+      "epoch": 0.4138770541692027,
+      "grad_norm": 0.8376681177687999,
+      "learning_rate": 2e-05,
+      "loss": 0.6183976650238037,
+      "step": 5440
+    },
+    {
+      "epoch": 0.41463785757760196,
+      "grad_norm": 2.4038557754100403,
+      "learning_rate": 2e-05,
+      "loss": 0.6118352890014649,
+      "step": 5450
+    },
+    {
+      "epoch": 0.4153986609860012,
+      "grad_norm": 2.501346234596342,
+      "learning_rate": 2e-05,
+      "loss": 0.6213789939880371,
+      "step": 5460
+    },
+    {
+      "epoch": 0.4161594643944005,
+      "grad_norm": 0.8366212001426536,
+      "learning_rate": 2e-05,
+      "loss": 0.6114327430725097,
+      "step": 5470
+    },
+    {
+      "epoch": 0.4169202678027998,
+      "grad_norm": 0.8171969217984115,
+      "learning_rate": 2e-05,
+      "loss": 0.5928665161132812,
+      "step": 5480
+    },
+    {
+      "epoch": 0.417681071211199,
+      "grad_norm": 1.6882351622382012,
+      "learning_rate": 2e-05,
+      "loss": 0.6239670753479004,
+      "step": 5490
+    },
+    {
+      "epoch": 0.4184418746195983,
+      "grad_norm": 0.7722677653630137,
+      "learning_rate": 2e-05,
+      "loss": 0.6299274444580079,
+      "step": 5500
+    },
+    {
+      "epoch": 0.41920267802799754,
+      "grad_norm": 0.9591036510613499,
+      "learning_rate": 2e-05,
+      "loss": 0.5882609367370606,
+      "step": 5510
+    },
+    {
+      "epoch": 0.41996348143639683,
+      "grad_norm": 5.15109719914921,
+      "learning_rate": 2e-05,
+      "loss": 0.5832295894622803,
+      "step": 5520
+    },
+    {
+      "epoch": 0.4207242848447961,
+      "grad_norm": 0.858667123364439,
+      "learning_rate": 2e-05,
+      "loss": 0.5727405548095703,
+      "step": 5530
+    },
+    {
+      "epoch": 0.42148508825319536,
+      "grad_norm": 0.8645833383183014,
+      "learning_rate": 2e-05,
+      "loss": 0.6142405509948731,
+      "step": 5540
+    },
+    {
+      "epoch": 0.42224589166159465,
+      "grad_norm": 1.5283098522361918,
+      "learning_rate": 2e-05,
+      "loss": 0.5926299095153809,
+      "step": 5550
+    },
+    {
+      "epoch": 0.4230066950699939,
+      "grad_norm": 1.3378791618681591,
+      "learning_rate": 2e-05,
+      "loss": 0.6136397361755371,
+      "step": 5560
+    },
+    {
+      "epoch": 0.4237674984783932,
+      "grad_norm": 1.361320541939213,
+      "learning_rate": 2e-05,
+      "loss": 0.6009634017944336,
+      "step": 5570
+    },
+    {
+      "epoch": 0.42452830188679247,
+      "grad_norm": 1.2448718860501085,
+      "learning_rate": 2e-05,
+      "loss": 0.5851739406585693,
+      "step": 5580
+    },
+    {
+      "epoch": 0.4252891052951917,
+      "grad_norm": 1.1846092405975608,
+      "learning_rate": 2e-05,
+      "loss": 0.5704478263854981,
+      "step": 5590
+    },
+    {
+      "epoch": 0.426049908703591,
+      "grad_norm": 0.8754529443700891,
+      "learning_rate": 2e-05,
+      "loss": 0.600365400314331,
+      "step": 5600
+    },
+    {
+      "epoch": 0.4268107121119903,
+      "grad_norm": 1.520648159137567,
+      "learning_rate": 2e-05,
+      "loss": 0.5971381187438964,
+      "step": 5610
+    },
+    {
+      "epoch": 0.4275715155203895,
+      "grad_norm": 0.8104632563362066,
+      "learning_rate": 2e-05,
+      "loss": 0.6063446044921875,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4283323189287888,
+      "grad_norm": 0.7922981794836915,
+      "learning_rate": 2e-05,
+      "loss": 0.6146617412567139,
+      "step": 5630
+    },
+    {
+      "epoch": 0.42909312233718805,
+      "grad_norm": 0.807067712086491,
+      "learning_rate": 2e-05,
+      "loss": 0.5988902091979981,
+      "step": 5640
+    },
+    {
+      "epoch": 0.42985392574558734,
+      "grad_norm": 0.849807554415781,
+      "learning_rate": 2e-05,
+      "loss": 0.5954990386962891,
+      "step": 5650
+    },
+    {
+      "epoch": 0.43061472915398663,
+      "grad_norm": 0.8821792549912292,
+      "learning_rate": 2e-05,
+      "loss": 0.6187366485595703,
+      "step": 5660
+    },
+    {
+      "epoch": 0.43137553256238587,
+      "grad_norm": 0.6721713048707927,
+      "learning_rate": 2e-05,
+      "loss": 0.5878865718841553,
+      "step": 5670
+    },
+    {
+      "epoch": 0.43213633597078516,
+      "grad_norm": 0.9661545971510167,
+      "learning_rate": 2e-05,
+      "loss": 0.58902006149292,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4328971393791844,
+      "grad_norm": 0.7603879816694954,
+      "learning_rate": 2e-05,
+      "loss": 0.5904243469238282,
+      "step": 5690
+    },
+    {
+      "epoch": 0.4336579427875837,
+      "grad_norm": 0.7105837940334772,
+      "learning_rate": 2e-05,
+      "loss": 0.5867074489593506,
+      "step": 5700
+    },
+    {
+      "epoch": 0.434418746195983,
+      "grad_norm": 1.4101833952290463,
+      "learning_rate": 2e-05,
+      "loss": 0.5997848510742188,
+      "step": 5710
+    },
+    {
+      "epoch": 0.4351795496043822,
+      "grad_norm": 0.7691204554059852,
+      "learning_rate": 2e-05,
+      "loss": 0.588038158416748,
+      "step": 5720
+    },
+    {
+      "epoch": 0.4359403530127815,
+      "grad_norm": 0.9144217862778072,
+      "learning_rate": 2e-05,
+      "loss": 0.583981704711914,
+      "step": 5730
+    },
+    {
+      "epoch": 0.4367011564211808,
+      "grad_norm": 1.384526870983644,
+      "learning_rate": 2e-05,
+      "loss": 0.6389365196228027,
+      "step": 5740
+    },
+    {
+      "epoch": 0.43746195982958,
+      "grad_norm": 1.0781812216166304,
+      "learning_rate": 2e-05,
+      "loss": 0.5930655479431153,
+      "step": 5750
+    },
+    {
+      "epoch": 0.4382227632379793,
+      "grad_norm": 1.1717860971615692,
+      "learning_rate": 2e-05,
+      "loss": 0.597266674041748,
+      "step": 5760
+    },
+    {
+      "epoch": 0.43898356664637855,
+      "grad_norm": 0.9408934956772596,
+      "learning_rate": 2e-05,
+      "loss": 0.6037530422210693,
+      "step": 5770
+    },
+    {
+      "epoch": 0.43974437005477784,
+      "grad_norm": 0.746427140381857,
+      "learning_rate": 2e-05,
+      "loss": 0.5954028129577636,
+      "step": 5780
+    },
+    {
+      "epoch": 0.44050517346317714,
+      "grad_norm": 0.9938173448891487,
+      "learning_rate": 2e-05,
+      "loss": 0.6098387718200684,
+      "step": 5790
+    },
+    {
+      "epoch": 0.44126597687157637,
+      "grad_norm": 0.7782007982790704,
+      "learning_rate": 2e-05,
+      "loss": 0.5757086753845215,
+      "step": 5800
+    },
+    {
+      "epoch": 0.44202678027997566,
+      "grad_norm": 0.7271815636845508,
+      "learning_rate": 2e-05,
+      "loss": 0.5755618572235107,
+      "step": 5810
+    },
+    {
+      "epoch": 0.4427875836883749,
+      "grad_norm": 0.8329589276318767,
+      "learning_rate": 2e-05,
+      "loss": 0.6157539367675782,
+      "step": 5820
+    },
+    {
+      "epoch": 0.4435483870967742,
+      "grad_norm": 0.9820296347985645,
+      "learning_rate": 2e-05,
+      "loss": 0.588036298751831,
+      "step": 5830
+    },
+    {
+      "epoch": 0.4443091905051735,
+      "grad_norm": 0.8222515923826359,
+      "learning_rate": 2e-05,
+      "loss": 0.6031598091125489,
+      "step": 5840
+    },
+    {
+      "epoch": 0.4450699939135727,
+      "grad_norm": 0.6755740757208559,
+      "learning_rate": 2e-05,
+      "loss": 0.5973345756530761,
+      "step": 5850
+    },
+    {
+      "epoch": 0.445830797321972,
+      "grad_norm": 0.8544261669890187,
+      "learning_rate": 2e-05,
+      "loss": 0.5905989646911621,
+      "step": 5860
+    },
+    {
+      "epoch": 0.4465916007303713,
+      "grad_norm": 0.7895628964692025,
+      "learning_rate": 2e-05,
+      "loss": 0.5860408782958985,
+      "step": 5870
+    },
+    {
+      "epoch": 0.44735240413877053,
+      "grad_norm": 2.224976752440008,
+      "learning_rate": 2e-05,
+      "loss": 0.6145426750183105,
+      "step": 5880
+    },
+    {
+      "epoch": 0.4481132075471698,
+      "grad_norm": 0.7371574725365392,
+      "learning_rate": 2e-05,
+      "loss": 0.6166990280151368,
+      "step": 5890
+    },
+    {
+      "epoch": 0.44887401095556906,
+      "grad_norm": 1.0380299511258209,
+      "learning_rate": 2e-05,
+      "loss": 0.6115994453430176,
+      "step": 5900
+    },
+    {
+      "epoch": 0.44963481436396835,
+      "grad_norm": 0.9162112958694462,
+      "learning_rate": 2e-05,
+      "loss": 0.6038604736328125,
+      "step": 5910
+    },
+    {
+      "epoch": 0.45039561777236764,
+      "grad_norm": 0.7859119259231996,
+      "learning_rate": 2e-05,
+      "loss": 0.5830548763275146,
+      "step": 5920
+    },
+    {
+      "epoch": 0.4511564211807669,
+      "grad_norm": 0.9047463830406691,
+      "learning_rate": 2e-05,
+      "loss": 0.5976335048675537,
+      "step": 5930
+    },
+    {
+      "epoch": 0.45191722458916617,
+      "grad_norm": 0.7427555043995095,
+      "learning_rate": 2e-05,
+      "loss": 0.5936368942260742,
+      "step": 5940
+    },
+    {
+      "epoch": 0.4526780279975654,
+      "grad_norm": 0.9157204135815878,
+      "learning_rate": 2e-05,
+      "loss": 0.6232377529144287,
+      "step": 5950
+    },
+    {
+      "epoch": 0.4534388314059647,
+      "grad_norm": 2.5382729608030408,
+      "learning_rate": 2e-05,
+      "loss": 0.5955874443054199,
+      "step": 5960
+    },
+    {
+      "epoch": 0.454199634814364,
+      "grad_norm": 0.8383120724801203,
+      "learning_rate": 2e-05,
+      "loss": 0.5980824947357177,
+      "step": 5970
+    },
+    {
+      "epoch": 0.4549604382227632,
+      "grad_norm": 1.5015185396369184,
+      "learning_rate": 2e-05,
+      "loss": 0.6086244583129883,
+      "step": 5980
+    },
+    {
+      "epoch": 0.4557212416311625,
+      "grad_norm": 0.892385547690889,
+      "learning_rate": 2e-05,
+      "loss": 0.5978293418884277,
+      "step": 5990
+    },
+    {
+      "epoch": 0.4564820450395618,
+      "grad_norm": 0.7820497983102545,
+      "learning_rate": 2e-05,
+      "loss": 0.5995134353637696,
+      "step": 6000
+    },
+    {
+      "epoch": 0.45724284844796104,
+      "grad_norm": 0.7907196940000327,
+      "learning_rate": 2e-05,
+      "loss": 0.6145267486572266,
+      "step": 6010
+    },
+    {
+      "epoch": 0.4580036518563603,
+      "grad_norm": 0.7820378828432865,
+      "learning_rate": 2e-05,
+      "loss": 0.5974668025970459,
+      "step": 6020
+    },
+    {
+      "epoch": 0.45876445526475956,
+      "grad_norm": 0.8275249088282894,
+      "learning_rate": 2e-05,
+      "loss": 0.5889710903167724,
+      "step": 6030
+    },
+    {
+      "epoch": 0.45952525867315885,
+      "grad_norm": 0.7859028274804515,
+      "learning_rate": 2e-05,
+      "loss": 0.6080682277679443,
+      "step": 6040
+    },
+    {
+      "epoch": 0.46028606208155814,
+      "grad_norm": 0.9340638606464002,
+      "learning_rate": 2e-05,
+      "loss": 0.6050857067108154,
+      "step": 6050
+    },
+    {
+      "epoch": 0.4610468654899574,
+      "grad_norm": 0.8449346432683734,
+      "learning_rate": 2e-05,
+      "loss": 0.5908621788024903,
+      "step": 6060
+    },
+    {
+      "epoch": 0.46180766889835667,
+      "grad_norm": 1.0043123314436746,
+      "learning_rate": 2e-05,
+      "loss": 0.6188840389251709,
+      "step": 6070
+    },
+    {
+      "epoch": 0.4625684723067559,
+      "grad_norm": 0.9055251907981219,
+      "learning_rate": 2e-05,
+      "loss": 0.6175956726074219,
+      "step": 6080
+    },
+    {
+      "epoch": 0.4633292757151552,
+      "grad_norm": 0.8234266943394802,
+      "learning_rate": 2e-05,
+      "loss": 0.5976392269134522,
+      "step": 6090
+    },
+    {
+      "epoch": 0.4640900791235545,
+      "grad_norm": 1.39247108124207,
+      "learning_rate": 2e-05,
+      "loss": 0.6120393753051758,
+      "step": 6100
+    },
+    {
+      "epoch": 0.4648508825319537,
+      "grad_norm": 0.9642370574099844,
+      "learning_rate": 2e-05,
+      "loss": 0.6084394454956055,
+      "step": 6110
+    },
+    {
+      "epoch": 0.465611685940353,
+      "grad_norm": 0.71109202951253,
+      "learning_rate": 2e-05,
+      "loss": 0.5668543815612793,
+      "step": 6120
+    },
+    {
+      "epoch": 0.4663724893487523,
+      "grad_norm": 0.7702069594107881,
+      "learning_rate": 2e-05,
+      "loss": 0.6063132286071777,
+      "step": 6130
+    },
+    {
+      "epoch": 0.46713329275715154,
+      "grad_norm": 0.9322475706065295,
+      "learning_rate": 2e-05,
+      "loss": 0.5895816802978515,
+      "step": 6140
+    },
+    {
+      "epoch": 0.46789409616555083,
+      "grad_norm": 0.8464485107543636,
+      "learning_rate": 2e-05,
+      "loss": 0.5967755317687988,
+      "step": 6150
+    },
+    {
+      "epoch": 0.46865489957395007,
+      "grad_norm": 0.691671559779243,
+      "learning_rate": 2e-05,
+      "loss": 0.604622220993042,
+      "step": 6160
+    },
+    {
+      "epoch": 0.46941570298234936,
+      "grad_norm": 0.8829624608290142,
+      "learning_rate": 2e-05,
+      "loss": 0.6174046993255615,
+      "step": 6170
+    },
+    {
+      "epoch": 0.47017650639074865,
+      "grad_norm": 1.734572102730861,
+      "learning_rate": 2e-05,
+      "loss": 0.5946842193603515,
+      "step": 6180
+    },
+    {
+      "epoch": 0.4709373097991479,
+      "grad_norm": 0.681947425586346,
+      "learning_rate": 2e-05,
+      "loss": 0.5782388210296631,
+      "step": 6190
+    },
+    {
+      "epoch": 0.4716981132075472,
+      "grad_norm": 0.8135670343882088,
+      "learning_rate": 2e-05,
+      "loss": 0.6043586730957031,
+      "step": 6200
+    },
+    {
+      "epoch": 0.47245891661594647,
+      "grad_norm": 1.3697871409482587,
+      "learning_rate": 2e-05,
+      "loss": 0.578231954574585,
+      "step": 6210
+    },
+    {
+      "epoch": 0.4732197200243457,
+      "grad_norm": 0.8872034811237967,
+      "learning_rate": 2e-05,
+      "loss": 0.5969443321228027,
+      "step": 6220
+    },
+    {
+      "epoch": 0.473980523432745,
+      "grad_norm": 0.8606441300362874,
+      "learning_rate": 2e-05,
+      "loss": 0.6171894073486328,
+      "step": 6230
+    },
+    {
+      "epoch": 0.47474132684114423,
+      "grad_norm": 1.2334186388782473,
+      "learning_rate": 2e-05,
+      "loss": 0.5960215568542481,
+      "step": 6240
+    },
+    {
+      "epoch": 0.4755021302495435,
+      "grad_norm": 1.7092129135723613,
+      "learning_rate": 2e-05,
+      "loss": 0.6007301330566406,
+      "step": 6250
+    },
+    {
+      "epoch": 0.4762629336579428,
+      "grad_norm": 1.07159467230858,
+      "learning_rate": 2e-05,
+      "loss": 0.5864115715026855,
+      "step": 6260
+    },
+    {
+      "epoch": 0.47702373706634205,
+      "grad_norm": 1.8328148609254264,
+      "learning_rate": 2e-05,
+      "loss": 0.5856305122375488,
+      "step": 6270
+    },
+    {
+      "epoch": 0.47778454047474134,
+      "grad_norm": 0.7694665420606871,
+      "learning_rate": 2e-05,
+      "loss": 0.60828857421875,
+      "step": 6280
+    },
+    {
+      "epoch": 0.4785453438831406,
+      "grad_norm": 0.833971199998042,
+      "learning_rate": 2e-05,
+      "loss": 0.5914821147918701,
+      "step": 6290
+    },
+    {
+      "epoch": 0.47930614729153986,
+      "grad_norm": 0.7744110424021482,
+      "learning_rate": 2e-05,
+      "loss": 0.5880824089050293,
+      "step": 6300
+    },
+    {
+      "epoch": 0.48006695069993915,
+      "grad_norm": 2.311912013486393,
+      "learning_rate": 2e-05,
+      "loss": 0.5972420692443847,
+      "step": 6310
+    },
+    {
+      "epoch": 0.4808277541083384,
+      "grad_norm": 0.9614314245859624,
+      "learning_rate": 2e-05,
+      "loss": 0.6224825859069825,
+      "step": 6320
+    },
+    {
+      "epoch": 0.4815885575167377,
+      "grad_norm": 3.3098803431162054,
+      "learning_rate": 2e-05,
+      "loss": 0.60919189453125,
+      "step": 6330
+    },
+    {
+      "epoch": 0.48234936092513697,
+      "grad_norm": 0.9843527441807506,
+      "learning_rate": 2e-05,
+      "loss": 0.6176755428314209,
+      "step": 6340
+    },
+    {
+      "epoch": 0.4831101643335362,
+      "grad_norm": 0.9807578772867592,
+      "learning_rate": 2e-05,
+      "loss": 0.597201919555664,
+      "step": 6350
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 0.9156763373611314,
+      "learning_rate": 2e-05,
+      "loss": 0.6129234313964844,
+      "step": 6360
+    },
+    {
+      "epoch": 0.48463177115033473,
+      "grad_norm": 0.9576212534387404,
+      "learning_rate": 2e-05,
+      "loss": 0.5875619411468506,
+      "step": 6370
+    },
+    {
+      "epoch": 0.485392574558734,
+      "grad_norm": 0.7185732233757746,
+      "learning_rate": 2e-05,
+      "loss": 0.5945401668548584,
+      "step": 6380
+    },
+    {
+      "epoch": 0.4861533779671333,
+      "grad_norm": 0.6466424771403458,
+      "learning_rate": 2e-05,
+      "loss": 0.5931663036346435,
+      "step": 6390
+    },
+    {
+      "epoch": 0.48691418137553255,
+      "grad_norm": 0.9385746373182312,
+      "learning_rate": 2e-05,
+      "loss": 0.5996630668640137,
+      "step": 6400
+    },
+    {
+      "epoch": 0.48767498478393184,
+      "grad_norm": 1.7640443629355802,
+      "learning_rate": 2e-05,
+      "loss": 0.5860879421234131,
+      "step": 6410
+    },
+    {
+      "epoch": 0.4884357881923311,
+      "grad_norm": 0.9005654464100281,
+      "learning_rate": 2e-05,
+      "loss": 0.6021587371826171,
+      "step": 6420
+    },
+    {
+      "epoch": 0.48919659160073037,
+      "grad_norm": 0.7769623981532261,
+      "learning_rate": 2e-05,
+      "loss": 0.617327070236206,
+      "step": 6430
+    },
+    {
+      "epoch": 0.48995739500912966,
+      "grad_norm": 0.8817311886907032,
+      "learning_rate": 2e-05,
+      "loss": 0.6047698020935058,
+      "step": 6440
+    },
+    {
+      "epoch": 0.4907181984175289,
+      "grad_norm": 0.7839141429761767,
+      "learning_rate": 2e-05,
+      "loss": 0.6019067764282227,
+      "step": 6450
+    },
+    {
+      "epoch": 0.4914790018259282,
+      "grad_norm": 0.9486155459692661,
+      "learning_rate": 2e-05,
+      "loss": 0.5925600051879882,
+      "step": 6460
+    },
+    {
+      "epoch": 0.4922398052343275,
+      "grad_norm": 0.9727186806024044,
+      "learning_rate": 2e-05,
+      "loss": 0.5977170467376709,
+      "step": 6470
+    },
+    {
+      "epoch": 0.4930006086427267,
+      "grad_norm": 1.1529077027455474,
+      "learning_rate": 2e-05,
+      "loss": 0.5967852592468261,
+      "step": 6480
+    },
+    {
+      "epoch": 0.493761412051126,
+      "grad_norm": 0.9262295423922237,
+      "learning_rate": 2e-05,
+      "loss": 0.6122447967529296,
+      "step": 6490
+    },
+    {
+      "epoch": 0.49452221545952524,
+      "grad_norm": 4.741583256985669,
+      "learning_rate": 2e-05,
+      "loss": 0.5941322326660157,
+      "step": 6500
+    },
+    {
+      "epoch": 0.49528301886792453,
+      "grad_norm": 1.1745397680735068,
+      "learning_rate": 2e-05,
+      "loss": 0.5917457580566406,
+      "step": 6510
+    },
+    {
+      "epoch": 0.4960438222763238,
+      "grad_norm": 1.182066381276861,
+      "learning_rate": 2e-05,
+      "loss": 0.5983775138854981,
+      "step": 6520
+    },
+    {
+      "epoch": 0.49680462568472306,
+      "grad_norm": 2.1765728855118,
+      "learning_rate": 2e-05,
+      "loss": 0.5961724758148194,
+      "step": 6530
+    },
+    {
+      "epoch": 0.49756542909312235,
+      "grad_norm": 0.8279428140137118,
+      "learning_rate": 2e-05,
+      "loss": 0.5972585201263427,
+      "step": 6540
+    },
+    {
+      "epoch": 0.4983262325015216,
+      "grad_norm": 0.8484846574967709,
+      "learning_rate": 2e-05,
+      "loss": 0.6016744136810303,
+      "step": 6550
+    },
+    {
+      "epoch": 0.4990870359099209,
+      "grad_norm": 1.277178462293471,
+      "learning_rate": 2e-05,
+      "loss": 0.58355712890625,
+      "step": 6560
+    },
+    {
+      "epoch": 0.49984783931832016,
+      "grad_norm": 1.3107312442620251,
+      "learning_rate": 2e-05,
+      "loss": 0.6163289070129394,
+      "step": 6570
+    },
+    {
+      "epoch": 0.5006086427267195,
+      "grad_norm": 1.2040187496956034,
+      "learning_rate": 2e-05,
+      "loss": 0.5998915672302246,
+      "step": 6580
+    },
+    {
+      "epoch": 0.5013694461351187,
+      "grad_norm": 0.8095153155250099,
+      "learning_rate": 2e-05,
+      "loss": 0.5843053817749023,
+      "step": 6590
+    },
+    {
+      "epoch": 0.5021302495435179,
+      "grad_norm": 0.9350524321386308,
+      "learning_rate": 2e-05,
+      "loss": 0.5938415050506591,
+      "step": 6600
+    },
+    {
+      "epoch": 0.5028910529519173,
+      "grad_norm": 1.0385186873663337,
+      "learning_rate": 2e-05,
+      "loss": 0.5834904670715332,
+      "step": 6610
+    },
+    {
+      "epoch": 0.5036518563603165,
+      "grad_norm": 0.8392958418081131,
+      "learning_rate": 2e-05,
+      "loss": 0.5910097122192383,
+      "step": 6620
+    },
+    {
+      "epoch": 0.5044126597687157,
+      "grad_norm": 0.7150043236738884,
+      "learning_rate": 2e-05,
+      "loss": 0.6202344417572021,
+      "step": 6630
+    },
+    {
+      "epoch": 0.505173463177115,
+      "grad_norm": 0.8245583292054631,
+      "learning_rate": 2e-05,
+      "loss": 0.5879296779632568,
+      "step": 6640
+    },
+    {
+      "epoch": 0.5059342665855143,
+      "grad_norm": 0.7101899245915984,
+      "learning_rate": 2e-05,
+      "loss": 0.6124242305755615,
+      "step": 6650
+    },
+    {
+      "epoch": 0.5066950699939136,
+      "grad_norm": 0.6167016800710498,
+      "learning_rate": 2e-05,
+      "loss": 0.6265460014343261,
+      "step": 6660
+    },
+    {
+      "epoch": 0.5074558734023128,
+      "grad_norm": 0.6430947526282285,
+      "learning_rate": 2e-05,
+      "loss": 0.5927323818206787,
+      "step": 6670
+    },
+    {
+      "epoch": 0.5082166768107121,
+      "grad_norm": 0.8796073038908419,
+      "learning_rate": 2e-05,
+      "loss": 0.5838780879974366,
+      "step": 6680
+    },
+    {
+      "epoch": 0.5089774802191114,
+      "grad_norm": 0.6822728845635229,
+      "learning_rate": 2e-05,
+      "loss": 0.6017322540283203,
+      "step": 6690
+    },
+    {
+      "epoch": 0.5097382836275106,
+      "grad_norm": 0.7076246456394932,
+      "learning_rate": 2e-05,
+      "loss": 0.5707648277282715,
+      "step": 6700
+    },
+    {
+      "epoch": 0.51049908703591,
+      "grad_norm": 1.9878995950175082,
+      "learning_rate": 2e-05,
+      "loss": 0.5908946514129638,
+      "step": 6710
+    },
+    {
+      "epoch": 0.5112598904443092,
+      "grad_norm": 0.9759818937387166,
+      "learning_rate": 2e-05,
+      "loss": 0.6030387878417969,
+      "step": 6720
+    },
+    {
+      "epoch": 0.5120206938527084,
+      "grad_norm": 1.298448135064124,
+      "learning_rate": 2e-05,
+      "loss": 0.596168041229248,
+      "step": 6730
+    },
+    {
+      "epoch": 0.5127814972611078,
+      "grad_norm": 0.7290274161049526,
+      "learning_rate": 2e-05,
+      "loss": 0.5897377014160157,
+      "step": 6740
+    },
+    {
+      "epoch": 0.513542300669507,
+      "grad_norm": 0.9175037276018961,
+      "learning_rate": 2e-05,
+      "loss": 0.5840108394622803,
+      "step": 6750
+    },
+    {
+      "epoch": 0.5143031040779062,
+      "grad_norm": 0.8067823906598198,
+      "learning_rate": 2e-05,
+      "loss": 0.5834906578063965,
+      "step": 6760
+    },
+    {
+      "epoch": 0.5150639074863055,
+      "grad_norm": 0.864420747605039,
+      "learning_rate": 2e-05,
+      "loss": 0.6145404815673828,
+      "step": 6770
+    },
+    {
+      "epoch": 0.5158247108947048,
+      "grad_norm": 0.7186518961116873,
+      "learning_rate": 2e-05,
+      "loss": 0.5772292137145996,
+      "step": 6780
+    },
+    {
+      "epoch": 0.5165855143031041,
+      "grad_norm": 1.269763974255044,
+      "learning_rate": 2e-05,
+      "loss": 0.6046066284179688,
+      "step": 6790
+    },
+    {
+      "epoch": 0.5173463177115033,
+      "grad_norm": 0.7164018452715892,
+      "learning_rate": 2e-05,
+      "loss": 0.593986988067627,
+      "step": 6800
+    },
+    {
+      "epoch": 0.5181071211199026,
+      "grad_norm": 0.8592977378326315,
+      "learning_rate": 2e-05,
+      "loss": 0.5864966869354248,
+      "step": 6810
+    },
+    {
+      "epoch": 0.5188679245283019,
+      "grad_norm": 0.7475415811753202,
+      "learning_rate": 2e-05,
+      "loss": 0.5911543846130372,
+      "step": 6820
+    },
+    {
+      "epoch": 0.5196287279367011,
+      "grad_norm": 0.8616306937502545,
+      "learning_rate": 2e-05,
+      "loss": 0.60450758934021,
+      "step": 6830
+    },
+    {
+      "epoch": 0.5203895313451005,
+      "grad_norm": 0.9837883337955783,
+      "learning_rate": 2e-05,
+      "loss": 0.5662042140960694,
+      "step": 6840
+    },
+    {
+      "epoch": 0.5211503347534997,
+      "grad_norm": 0.8291541340908452,
+      "learning_rate": 2e-05,
+      "loss": 0.6120754241943359,
+      "step": 6850
+    },
+    {
+      "epoch": 0.5219111381618989,
+      "grad_norm": 0.8625925929766998,
+      "learning_rate": 2e-05,
+      "loss": 0.6233394145965576,
+      "step": 6860
+    },
+    {
+      "epoch": 0.5226719415702983,
+      "grad_norm": 0.8022721756026254,
+      "learning_rate": 2e-05,
+      "loss": 0.5942387580871582,
+      "step": 6870
+    },
+    {
+      "epoch": 0.5234327449786975,
+      "grad_norm": 0.961289433445639,
+      "learning_rate": 2e-05,
+      "loss": 0.5985567092895507,
+      "step": 6880
+    },
+    {
+      "epoch": 0.5241935483870968,
+      "grad_norm": 0.6744850369199558,
+      "learning_rate": 2e-05,
+      "loss": 0.5769065380096435,
+      "step": 6890
+    },
+    {
+      "epoch": 0.5249543517954961,
+      "grad_norm": 0.7553655309228883,
+      "learning_rate": 2e-05,
+      "loss": 0.5933752059936523,
+      "step": 6900
+    },
+    {
+      "epoch": 0.5257151552038953,
+      "grad_norm": 0.7309391321143712,
+      "learning_rate": 2e-05,
+      "loss": 0.5892446041107178,
+      "step": 6910
+    },
+    {
+      "epoch": 0.5264759586122946,
+      "grad_norm": 1.1486314163617912,
+      "learning_rate": 2e-05,
+      "loss": 0.5598908424377441,
+      "step": 6920
+    },
+    {
+      "epoch": 0.5272367620206938,
+      "grad_norm": 0.7000013001966744,
+      "learning_rate": 2e-05,
+      "loss": 0.594353199005127,
+      "step": 6930
+    },
+    {
+      "epoch": 0.5279975654290932,
+      "grad_norm": 0.7949881156270052,
+      "learning_rate": 2e-05,
+      "loss": 0.6161168575286865,
+      "step": 6940
+    },
+    {
+      "epoch": 0.5287583688374924,
+      "grad_norm": 1.0310954529176908,
+      "learning_rate": 2e-05,
+      "loss": 0.5931632995605469,
+      "step": 6950
+    },
+    {
+      "epoch": 0.5295191722458916,
+      "grad_norm": 0.7781386550614464,
+      "learning_rate": 2e-05,
+      "loss": 0.5946434020996094,
+      "step": 6960
+    },
+    {
+      "epoch": 0.530279975654291,
+      "grad_norm": 1.164536612760523,
+      "learning_rate": 2e-05,
+      "loss": 0.5941308498382568,
+      "step": 6970
+    },
+    {
+      "epoch": 0.5310407790626902,
+      "grad_norm": 2.193763340271562,
+      "learning_rate": 2e-05,
+      "loss": 0.5943192958831787,
+      "step": 6980
+    },
+    {
+      "epoch": 0.5318015824710894,
+      "grad_norm": 0.9507533268895512,
+      "learning_rate": 2e-05,
+      "loss": 0.5971503257751465,
+      "step": 6990
+    },
+    {
+      "epoch": 0.5325623858794888,
+      "grad_norm": 1.9594618575097285,
+      "learning_rate": 2e-05,
+      "loss": 0.6027518749237061,
+      "step": 7000
+    },
+    {
+      "epoch": 0.533323189287888,
+      "grad_norm": 1.9507625079550195,
+      "learning_rate": 2e-05,
+      "loss": 0.5909186363220215,
+      "step": 7010
+    },
+    {
+      "epoch": 0.5340839926962873,
+      "grad_norm": 0.9119550962599303,
+      "learning_rate": 2e-05,
+      "loss": 0.6007914543151855,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5348447961046866,
+      "grad_norm": 2.2699041749633504,
+      "learning_rate": 2e-05,
+      "loss": 0.576223087310791,
+      "step": 7030
+    },
+    {
+      "epoch": 0.5356055995130858,
+      "grad_norm": 4.06828243574309,
+      "learning_rate": 2e-05,
+      "loss": 0.5864376068115235,
+      "step": 7040
+    },
+    {
+      "epoch": 0.5363664029214851,
+      "grad_norm": 1.808773325300879,
+      "learning_rate": 2e-05,
+      "loss": 0.5904212951660156,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5371272063298843,
+      "grad_norm": 1.9363494684269886,
+      "learning_rate": 2e-05,
+      "loss": 0.5989524841308593,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5378880097382837,
+      "grad_norm": 0.9611240257716946,
+      "learning_rate": 2e-05,
+      "loss": 0.5720105171203613,
+      "step": 7070
+    },
+    {
+      "epoch": 0.5386488131466829,
+      "grad_norm": 0.9381030596015086,
+      "learning_rate": 2e-05,
+      "loss": 0.5764672756195068,
+      "step": 7080
+    },
+    {
+      "epoch": 0.5394096165550821,
+      "grad_norm": 0.7133396384431813,
+      "learning_rate": 2e-05,
+      "loss": 0.5844818115234375,
+      "step": 7090
+    },
+    {
+      "epoch": 0.5401704199634815,
+      "grad_norm": 1.228106753864805,
+      "learning_rate": 2e-05,
+      "loss": 0.5756110191345215,
+      "step": 7100
+    },
+    {
+      "epoch": 0.5409312233718807,
+      "grad_norm": 1.172807345891725,
+      "learning_rate": 2e-05,
+      "loss": 0.5963610649108887,
+      "step": 7110
+    },
+    {
+      "epoch": 0.54169202678028,
+      "grad_norm": 0.7988665048794369,
+      "learning_rate": 2e-05,
+      "loss": 0.5687862396240234,
+      "step": 7120
+    },
+    {
+      "epoch": 0.5424528301886793,
+      "grad_norm": 0.8756779319707574,
+      "learning_rate": 2e-05,
+      "loss": 0.5940175533294678,
+      "step": 7130
+    },
+    {
+      "epoch": 0.5432136335970785,
+      "grad_norm": 0.8335616086510742,
+      "learning_rate": 2e-05,
+      "loss": 0.6116618156433106,
+      "step": 7140
+    },
+    {
+      "epoch": 0.5439744370054778,
+      "grad_norm": 0.8988298885961891,
+      "learning_rate": 2e-05,
+      "loss": 0.5815279960632325,
+      "step": 7150
+    },
+    {
+      "epoch": 0.5447352404138771,
+      "grad_norm": 0.7922592441637101,
+      "learning_rate": 2e-05,
+      "loss": 0.6046882152557373,
+      "step": 7160
+    },
+    {
+      "epoch": 0.5454960438222763,
+      "grad_norm": 0.7230080347415918,
+      "learning_rate": 2e-05,
+      "loss": 0.6052354335784912,
+      "step": 7170
+    },
+    {
+      "epoch": 0.5462568472306756,
+      "grad_norm": 0.9105413875444616,
+      "learning_rate": 2e-05,
+      "loss": 0.5974827766418457,
+      "step": 7180
+    },
+    {
+      "epoch": 0.5470176506390748,
+      "grad_norm": 0.991569273572482,
+      "learning_rate": 2e-05,
+      "loss": 0.5784293174743652,
+      "step": 7190
+    },
+    {
+      "epoch": 0.5477784540474742,
+      "grad_norm": 1.0742103655667512,
+      "learning_rate": 2e-05,
+      "loss": 0.6032185554504395,
+      "step": 7200
+    },
+    {
+      "epoch": 0.5485392574558734,
+      "grad_norm": 1.4465716419087093,
+      "learning_rate": 2e-05,
+      "loss": 0.5907020568847656,
+      "step": 7210
+    },
+    {
+      "epoch": 0.5493000608642726,
+      "grad_norm": 1.1364346138030665,
+      "learning_rate": 2e-05,
+      "loss": 0.5903211116790772,
+      "step": 7220
+    },
+    {
+      "epoch": 0.550060864272672,
+      "grad_norm": 0.9560542578948066,
+      "learning_rate": 2e-05,
+      "loss": 0.598571491241455,
+      "step": 7230
+    },
+    {
+      "epoch": 0.5508216676810712,
+      "grad_norm": 0.8010896031955296,
+      "learning_rate": 2e-05,
+      "loss": 0.5829965114593506,
+      "step": 7240
+    },
+    {
+      "epoch": 0.5515824710894704,
+      "grad_norm": 0.7641341359323666,
+      "learning_rate": 2e-05,
+      "loss": 0.6106725215911866,
+      "step": 7250
+    },
+    {
+      "epoch": 0.5523432744978698,
+      "grad_norm": 0.6212153729139268,
+      "learning_rate": 2e-05,
+      "loss": 0.604694128036499,
+      "step": 7260
+    },
+    {
+      "epoch": 0.553104077906269,
+      "grad_norm": 0.6892317950325276,
+      "learning_rate": 2e-05,
+      "loss": 0.5864579677581787,
+      "step": 7270
+    },
+    {
+      "epoch": 0.5538648813146683,
+      "grad_norm": 1.2125501779997605,
+      "learning_rate": 2e-05,
+      "loss": 0.6229146480560303,
+      "step": 7280
+    },
+    {
+      "epoch": 0.5546256847230676,
+      "grad_norm": 0.6777515455658605,
+      "learning_rate": 2e-05,
+      "loss": 0.5959762096405029,
+      "step": 7290
+    },
+    {
+      "epoch": 0.5553864881314668,
+      "grad_norm": 0.7028903716974437,
+      "learning_rate": 2e-05,
+      "loss": 0.6114962577819825,
+      "step": 7300
+    },
+    {
+      "epoch": 0.5561472915398661,
+      "grad_norm": 5.632879875294642,
+      "learning_rate": 2e-05,
+      "loss": 0.5935216426849366,
+      "step": 7310
+    },
+    {
+      "epoch": 0.5569080949482653,
+      "grad_norm": 0.7742332304061029,
+      "learning_rate": 2e-05,
+      "loss": 0.6058724880218506,
+      "step": 7320
+    },
+    {
+      "epoch": 0.5576688983566647,
+      "grad_norm": 0.6823558459648175,
+      "learning_rate": 2e-05,
+      "loss": 0.5884871482849121,
+      "step": 7330
+    },
+    {
+      "epoch": 0.5584297017650639,
+      "grad_norm": 0.743582750143488,
+      "learning_rate": 2e-05,
+      "loss": 0.5860174179077149,
+      "step": 7340
+    },
+    {
+      "epoch": 0.5591905051734631,
+      "grad_norm": 1.0608394945478157,
+      "learning_rate": 2e-05,
+      "loss": 0.572206687927246,
+      "step": 7350
+    },
+    {
+      "epoch": 0.5599513085818625,
+      "grad_norm": 0.6328210506303615,
+      "learning_rate": 2e-05,
+      "loss": 0.588253402709961,
+      "step": 7360
+    },
+    {
+      "epoch": 0.5607121119902617,
+      "grad_norm": 1.10146593332472,
+      "learning_rate": 2e-05,
+      "loss": 0.5856830596923828,
+      "step": 7370
+    },
+    {
+      "epoch": 0.561472915398661,
+      "grad_norm": 0.7268944331273012,
+      "learning_rate": 2e-05,
+      "loss": 0.5685952663421631,
+      "step": 7380
+    },
+    {
+      "epoch": 0.5622337188070603,
+      "grad_norm": 0.7332182018150611,
+      "learning_rate": 2e-05,
+      "loss": 0.5999359130859375,
+      "step": 7390
+    },
+    {
+      "epoch": 0.5629945222154595,
+      "grad_norm": 0.6526289446646008,
+      "learning_rate": 2e-05,
+      "loss": 0.5861812591552734,
+      "step": 7400
+    },
+    {
+      "epoch": 0.5637553256238588,
+      "grad_norm": 0.7513732969326448,
+      "learning_rate": 2e-05,
+      "loss": 0.5944066524505616,
+      "step": 7410
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 0.7328844102075918,
+      "learning_rate": 2e-05,
+      "loss": 0.5877908229827881,
+      "step": 7420
+    },
+    {
+      "epoch": 0.5652769324406574,
+      "grad_norm": 2.0007653284417564,
+      "learning_rate": 2e-05,
+      "loss": 0.6209169387817383,
+      "step": 7430
+    },
+    {
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7060032928140384,
+      "learning_rate": 2e-05,
+      "loss": 0.5827223777770996,
+      "step": 7440
+    },
+    {
+      "epoch": 0.5667985392574558,
+      "grad_norm": 0.6937508050280076,
+      "learning_rate": 2e-05,
+      "loss": 0.6033755302429199,
+      "step": 7450
+    },
+    {
+      "epoch": 0.5675593426658552,
+      "grad_norm": 0.7479510720116052,
+      "learning_rate": 2e-05,
+      "loss": 0.5906998634338378,
+      "step": 7460
+    },
+    {
+      "epoch": 0.5683201460742544,
+      "grad_norm": 0.8347060607482271,
+      "learning_rate": 2e-05,
+      "loss": 0.5714436531066894,
+      "step": 7470
+    },
+    {
+      "epoch": 0.5690809494826536,
+      "grad_norm": 0.8143556084405441,
+      "learning_rate": 2e-05,
+      "loss": 0.5804880142211915,
+      "step": 7480
+    },
+    {
+      "epoch": 0.569841752891053,
+      "grad_norm": 0.7707218816853956,
+      "learning_rate": 2e-05,
+      "loss": 0.6021479606628418,
+      "step": 7490
+    },
+    {
+      "epoch": 0.5706025562994522,
+      "grad_norm": 0.6094015311359219,
+      "learning_rate": 2e-05,
+      "loss": 0.6025670051574707,
+      "step": 7500
+    },
+    {
+      "epoch": 0.5713633597078515,
+      "grad_norm": 0.6550015990285097,
+      "learning_rate": 2e-05,
+      "loss": 0.5732488632202148,
+      "step": 7510
+    },
+    {
+      "epoch": 0.5721241631162508,
+      "grad_norm": 0.673108381205908,
+      "learning_rate": 2e-05,
+      "loss": 0.6147636413574219,
+      "step": 7520
+    },
+    {
+      "epoch": 0.57288496652465,
+      "grad_norm": 0.9920118419222084,
+      "learning_rate": 2e-05,
+      "loss": 0.5886077880859375,
+      "step": 7530
+    },
+    {
+      "epoch": 0.5736457699330493,
+      "grad_norm": 0.8293535627827149,
+      "learning_rate": 2e-05,
+      "loss": 0.5675832271575928,
+      "step": 7540
+    },
+    {
+      "epoch": 0.5744065733414486,
+      "grad_norm": 1.493074534154195,
+      "learning_rate": 2e-05,
+      "loss": 0.5788111686706543,
+      "step": 7550
+    },
+    {
+      "epoch": 0.5751673767498479,
+      "grad_norm": 0.599579230510124,
+      "learning_rate": 2e-05,
+      "loss": 0.5835439205169678,
+      "step": 7560
+    },
+    {
+      "epoch": 0.5759281801582471,
+      "grad_norm": 0.70877554322285,
+      "learning_rate": 2e-05,
+      "loss": 0.5953329086303711,
+      "step": 7570
+    },
+    {
+      "epoch": 0.5766889835666463,
+      "grad_norm": 0.6760962670924734,
+      "learning_rate": 2e-05,
+      "loss": 0.6296019554138184,
+      "step": 7580
+    },
+    {
+      "epoch": 0.5774497869750457,
+      "grad_norm": 0.8301914910339789,
+      "learning_rate": 2e-05,
+      "loss": 0.5888818740844727,
+      "step": 7590
+    },
+    {
+      "epoch": 0.5782105903834449,
+      "grad_norm": 0.9142099974639273,
+      "learning_rate": 2e-05,
+      "loss": 0.5825131893157959,
+      "step": 7600
+    },
+    {
+      "epoch": 0.5789713937918441,
+      "grad_norm": 0.8063177946856701,
+      "learning_rate": 2e-05,
+      "loss": 0.566869068145752,
+      "step": 7610
+    },
+    {
+      "epoch": 0.5797321972002435,
+      "grad_norm": 0.6954128060576373,
+      "learning_rate": 2e-05,
+      "loss": 0.5899360179901123,
+      "step": 7620
+    },
+    {
+      "epoch": 0.5804930006086427,
+      "grad_norm": 0.6863097655531907,
+      "learning_rate": 2e-05,
+      "loss": 0.5795107841491699,
+      "step": 7630
+    },
+    {
+      "epoch": 0.581253804017042,
+      "grad_norm": 1.907721295042232,
+      "learning_rate": 2e-05,
+      "loss": 0.5674359798431396,
+      "step": 7640
+    },
+    {
+      "epoch": 0.5820146074254413,
+      "grad_norm": 0.7353595920370852,
+      "learning_rate": 2e-05,
+      "loss": 0.6048314571380615,
+      "step": 7650
+    },
+    {
+      "epoch": 0.5827754108338405,
+      "grad_norm": 0.8049563167020439,
+      "learning_rate": 2e-05,
+      "loss": 0.6160181999206543,
+      "step": 7660
+    },
+    {
+      "epoch": 0.5835362142422398,
+      "grad_norm": 0.6922074978446334,
+      "learning_rate": 2e-05,
+      "loss": 0.58782958984375,
+      "step": 7670
+    },
+    {
+      "epoch": 0.5842970176506391,
+      "grad_norm": 1.1603935066662177,
+      "learning_rate": 2e-05,
+      "loss": 0.5904664039611817,
+      "step": 7680
+    },
+    {
+      "epoch": 0.5850578210590384,
+      "grad_norm": 0.7196315235391763,
+      "learning_rate": 2e-05,
+      "loss": 0.6162894248962403,
+      "step": 7690
+    },
+    {
+      "epoch": 0.5858186244674376,
+      "grad_norm": 0.6332594607848281,
+      "learning_rate": 2e-05,
+      "loss": 0.5876092910766602,
+      "step": 7700
+    },
+    {
+      "epoch": 0.5865794278758368,
+      "grad_norm": 0.7225400287869544,
+      "learning_rate": 2e-05,
+      "loss": 0.5740478515625,
+      "step": 7710
+    },
+    {
+      "epoch": 0.5873402312842362,
+      "grad_norm": 0.6803688378001521,
+      "learning_rate": 2e-05,
+      "loss": 0.599135160446167,
+      "step": 7720
+    },
+    {
+      "epoch": 0.5881010346926354,
+      "grad_norm": 0.6595337869950691,
+      "learning_rate": 2e-05,
+      "loss": 0.5755078315734863,
+      "step": 7730
+    },
+    {
+      "epoch": 0.5888618381010347,
+      "grad_norm": 0.8894405026476516,
+      "learning_rate": 2e-05,
+      "loss": 0.6003472805023193,
+      "step": 7740
+    },
+    {
+      "epoch": 0.589622641509434,
+      "grad_norm": 0.9686594104197035,
+      "learning_rate": 2e-05,
+      "loss": 0.5933383464813232,
+      "step": 7750
+    },
+    {
+      "epoch": 0.5903834449178332,
+      "grad_norm": 0.6957770422553774,
+      "learning_rate": 2e-05,
+      "loss": 0.5930872917175293,
+      "step": 7760
+    },
+    {
+      "epoch": 0.5911442483262325,
+      "grad_norm": 0.9377194500637537,
+      "learning_rate": 2e-05,
+      "loss": 0.5795677185058594,
+      "step": 7770
+    },
+    {
+      "epoch": 0.5919050517346318,
+      "grad_norm": 1.3052446188354487,
+      "learning_rate": 2e-05,
+      "loss": 0.5997994899749756,
+      "step": 7780
+    },
+    {
+      "epoch": 0.592665855143031,
+      "grad_norm": 0.8311639400812058,
+      "learning_rate": 2e-05,
+      "loss": 0.5732833862304687,
+      "step": 7790
+    },
+    {
+      "epoch": 0.5934266585514303,
+      "grad_norm": 1.008873536309028,
+      "learning_rate": 2e-05,
+      "loss": 0.5964694023132324,
+      "step": 7800
+    },
+    {
+      "epoch": 0.5941874619598296,
+      "grad_norm": 0.7215062162062124,
+      "learning_rate": 2e-05,
+      "loss": 0.5797884941101075,
+      "step": 7810
+    },
+    {
+      "epoch": 0.5949482653682289,
+      "grad_norm": 0.7349073201420375,
+      "learning_rate": 2e-05,
+      "loss": 0.6102656364440918,
+      "step": 7820
+    },
+    {
+      "epoch": 0.5957090687766281,
+      "grad_norm": 0.6804180224996377,
+      "learning_rate": 2e-05,
+      "loss": 0.5743989467620849,
+      "step": 7830
+    },
+    {
+      "epoch": 0.5964698721850273,
+      "grad_norm": 0.6067914881627312,
+      "learning_rate": 2e-05,
+      "loss": 0.5865550994873047,
+      "step": 7840
+    },
+    {
+      "epoch": 0.5972306755934267,
+      "grad_norm": 0.7399644986679592,
+      "learning_rate": 2e-05,
+      "loss": 0.5778102397918701,
+      "step": 7850
+    },
+    {
+      "epoch": 0.5979914790018259,
+      "grad_norm": 0.7572300142397286,
+      "learning_rate": 2e-05,
+      "loss": 0.5757750988006591,
+      "step": 7860
+    },
+    {
+      "epoch": 0.5987522824102252,
+      "grad_norm": 0.6386196334473465,
+      "learning_rate": 2e-05,
+      "loss": 0.5788669586181641,
+      "step": 7870
+    },
+    {
+      "epoch": 0.5995130858186245,
+      "grad_norm": 0.7253937586009458,
+      "learning_rate": 2e-05,
+      "loss": 0.58053560256958,
+      "step": 7880
+    },
+    {
+      "epoch": 0.6002738892270237,
+      "grad_norm": 0.6905964895004345,
+      "learning_rate": 2e-05,
+      "loss": 0.5661889553070069,
+      "step": 7890
+    },
+    {
+      "epoch": 0.601034692635423,
+      "grad_norm": 0.6512259680490939,
+      "learning_rate": 2e-05,
+      "loss": 0.5827771663665772,
+      "step": 7900
+    },
+    {
+      "epoch": 0.6017954960438223,
+      "grad_norm": 0.770310245255796,
+      "learning_rate": 2e-05,
+      "loss": 0.6021918773651123,
+      "step": 7910
+    },
+    {
+      "epoch": 0.6025562994522216,
+      "grad_norm": 1.7709495455813984,
+      "learning_rate": 2e-05,
+      "loss": 0.5930996894836426,
+      "step": 7920
+    },
+    {
+      "epoch": 0.6033171028606208,
+      "grad_norm": 0.8048054619082563,
+      "learning_rate": 2e-05,
+      "loss": 0.5857785701751709,
+      "step": 7930
+    },
+    {
+      "epoch": 0.6040779062690201,
+      "grad_norm": 1.0996171571933824,
+      "learning_rate": 2e-05,
+      "loss": 0.6102975845336914,
+      "step": 7940
+    },
+    {
+      "epoch": 0.6048387096774194,
+      "grad_norm": 0.9883911448677426,
+      "learning_rate": 2e-05,
+      "loss": 0.5942376613616943,
+      "step": 7950
+    },
+    {
+      "epoch": 0.6055995130858186,
+      "grad_norm": 0.8001674341000556,
+      "learning_rate": 2e-05,
+      "loss": 0.5806931972503662,
+      "step": 7960
+    },
+    {
+      "epoch": 0.6063603164942178,
+      "grad_norm": 0.9760088637569831,
+      "learning_rate": 2e-05,
+      "loss": 0.5959899425506592,
+      "step": 7970
+    },
+    {
+      "epoch": 0.6071211199026172,
+      "grad_norm": 0.9630051853926012,
+      "learning_rate": 2e-05,
+      "loss": 0.5764996528625488,
+      "step": 7980
+    },
+    {
+      "epoch": 0.6078819233110164,
+      "grad_norm": 0.856462161249395,
+      "learning_rate": 2e-05,
+      "loss": 0.5811921119689941,
+      "step": 7990
+    },
+    {
+      "epoch": 0.6086427267194157,
+      "grad_norm": 0.8767776644245268,
+      "learning_rate": 2e-05,
+      "loss": 0.5894935607910157,
+      "step": 8000
+    },
+    {
+      "epoch": 0.609403530127815,
+      "grad_norm": 0.6888978451025628,
+      "learning_rate": 2e-05,
+      "loss": 0.5752892971038819,
+      "step": 8010
+    },
+    {
+      "epoch": 0.6101643335362142,
+      "grad_norm": 0.7314682430295208,
+      "learning_rate": 2e-05,
+      "loss": 0.6115509033203125,
+      "step": 8020
+    },
+    {
+      "epoch": 0.6109251369446135,
+      "grad_norm": 0.8819143051330999,
+      "learning_rate": 2e-05,
+      "loss": 0.5878534317016602,
+      "step": 8030
+    },
+    {
+      "epoch": 0.6116859403530128,
+      "grad_norm": 1.2739191304324187,
+      "learning_rate": 2e-05,
+      "loss": 0.5639880180358887,
+      "step": 8040
+    },
+    {
+      "epoch": 0.6124467437614121,
+      "grad_norm": 1.026137263331152,
+      "learning_rate": 2e-05,
+      "loss": 0.5880692958831787,
+      "step": 8050
+    },
+    {
+      "epoch": 0.6132075471698113,
+      "grad_norm": 0.6864141682431978,
+      "learning_rate": 2e-05,
+      "loss": 0.5761102676391602,
+      "step": 8060
+    },
+    {
+      "epoch": 0.6139683505782106,
+      "grad_norm": 0.9000041629437778,
+      "learning_rate": 2e-05,
+      "loss": 0.6025727272033692,
+      "step": 8070
+    },
+    {
+      "epoch": 0.6147291539866099,
+      "grad_norm": 0.6175229385796767,
+      "learning_rate": 2e-05,
+      "loss": 0.5767220973968505,
+      "step": 8080
+    },
+    {
+      "epoch": 0.6154899573950091,
+      "grad_norm": 0.7355537931323495,
+      "learning_rate": 2e-05,
+      "loss": 0.5724764823913574,
+      "step": 8090
+    },
+    {
+      "epoch": 0.6162507608034083,
+      "grad_norm": 0.7214140813229626,
+      "learning_rate": 2e-05,
+      "loss": 0.6040135383605957,
+      "step": 8100
+    },
+    {
+      "epoch": 0.6170115642118077,
+      "grad_norm": 0.8867733118440105,
+      "learning_rate": 2e-05,
+      "loss": 0.5824000835418701,
+      "step": 8110
+    },
+    {
+      "epoch": 0.6177723676202069,
+      "grad_norm": 0.6632655212696472,
+      "learning_rate": 2e-05,
+      "loss": 0.5825210094451905,
+      "step": 8120
+    },
+    {
+      "epoch": 0.6185331710286062,
+      "grad_norm": 1.3039849225173976,
+      "learning_rate": 2e-05,
+      "loss": 0.5731758117675781,
+      "step": 8130
+    },
+    {
+      "epoch": 0.6192939744370055,
+      "grad_norm": 0.7798137454470687,
+      "learning_rate": 2e-05,
+      "loss": 0.5960041046142578,
+      "step": 8140
+    },
+    {
+      "epoch": 0.6200547778454047,
+      "grad_norm": 0.8798265696047927,
+      "learning_rate": 2e-05,
+      "loss": 0.5874600887298584,
+      "step": 8150
+    },
+    {
+      "epoch": 0.620815581253804,
+      "grad_norm": 0.6502849581700337,
+      "learning_rate": 2e-05,
+      "loss": 0.5856257438659668,
+      "step": 8160
+    },
+    {
+      "epoch": 0.6215763846622033,
+      "grad_norm": 0.6880203614647077,
+      "learning_rate": 2e-05,
+      "loss": 0.5868075370788575,
+      "step": 8170
+    },
+    {
+      "epoch": 0.6223371880706026,
+      "grad_norm": 1.0189828708123057,
+      "learning_rate": 2e-05,
+      "loss": 0.58623366355896,
+      "step": 8180
+    },
+    {
+      "epoch": 0.6230979914790018,
+      "grad_norm": 0.6600387064512608,
+      "learning_rate": 2e-05,
+      "loss": 0.5799531459808349,
+      "step": 8190
+    },
+    {
+      "epoch": 0.6238587948874011,
+      "grad_norm": 0.6990687727680752,
+      "learning_rate": 2e-05,
+      "loss": 0.5999938011169433,
+      "step": 8200
+    },
+    {
+      "epoch": 0.6246195982958004,
+      "grad_norm": 0.7199766410621371,
+      "learning_rate": 2e-05,
+      "loss": 0.5787326335906983,
+      "step": 8210
+    },
+    {
+      "epoch": 0.6253804017041996,
+      "grad_norm": 0.7410373756004169,
+      "learning_rate": 2e-05,
+      "loss": 0.5885562896728516,
+      "step": 8220
+    },
+    {
+      "epoch": 0.6261412051125989,
+      "grad_norm": 0.7086152327000189,
+      "learning_rate": 2e-05,
+      "loss": 0.5672260284423828,
+      "step": 8230
+    },
+    {
+      "epoch": 0.6269020085209982,
+      "grad_norm": 0.6982715146625542,
+      "learning_rate": 2e-05,
+      "loss": 0.593112850189209,
+      "step": 8240
+    },
+    {
+      "epoch": 0.6276628119293974,
+      "grad_norm": 0.6452788106788441,
+      "learning_rate": 2e-05,
+      "loss": 0.5792741298675537,
+      "step": 8250
+    },
+    {
+      "epoch": 0.6284236153377967,
+      "grad_norm": 0.7263432590558087,
+      "learning_rate": 2e-05,
+      "loss": 0.6027037620544433,
+      "step": 8260
+    },
+    {
+      "epoch": 0.629184418746196,
+      "grad_norm": 0.706996457769387,
+      "learning_rate": 2e-05,
+      "loss": 0.5784778594970703,
+      "step": 8270
+    },
+    {
+      "epoch": 0.6299452221545953,
+      "grad_norm": 0.6437353330159822,
+      "learning_rate": 2e-05,
+      "loss": 0.5577037334442139,
+      "step": 8280
+    },
+    {
+      "epoch": 0.6307060255629945,
+      "grad_norm": 0.8921533017336402,
+      "learning_rate": 2e-05,
+      "loss": 0.5786788940429688,
+      "step": 8290
+    },
+    {
+      "epoch": 0.6314668289713938,
+      "grad_norm": 0.751772311788144,
+      "learning_rate": 2e-05,
+      "loss": 0.560968542098999,
+      "step": 8300
+    },
+    {
+      "epoch": 0.6322276323797931,
+      "grad_norm": 0.8493789931970467,
+      "learning_rate": 2e-05,
+      "loss": 0.571924877166748,
+      "step": 8310
+    },
+    {
+      "epoch": 0.6329884357881923,
+      "grad_norm": 0.6751302284415017,
+      "learning_rate": 2e-05,
+      "loss": 0.567514705657959,
+      "step": 8320
+    },
+    {
+      "epoch": 0.6337492391965917,
+      "grad_norm": 0.7074678356828549,
+      "learning_rate": 2e-05,
+      "loss": 0.5773214340209961,
+      "step": 8330
+    },
+    {
+      "epoch": 0.6345100426049909,
+      "grad_norm": 1.0383381073780504,
+      "learning_rate": 2e-05,
+      "loss": 0.5752066612243653,
+      "step": 8340
+    },
+    {
+      "epoch": 0.6352708460133901,
+      "grad_norm": 0.7485528685092199,
+      "learning_rate": 2e-05,
+      "loss": 0.5848573207855224,
+      "step": 8350
+    },
+    {
+      "epoch": 0.6360316494217894,
+      "grad_norm": 1.1336830956722113,
+      "learning_rate": 2e-05,
+      "loss": 0.593998908996582,
+      "step": 8360
+    },
+    {
+      "epoch": 0.6367924528301887,
+      "grad_norm": 0.9057729099155309,
+      "learning_rate": 2e-05,
+      "loss": 0.5708306312561036,
+      "step": 8370
+    },
+    {
+      "epoch": 0.6375532562385879,
+      "grad_norm": 0.7714731948039438,
+      "learning_rate": 2e-05,
+      "loss": 0.5627151489257812,
+      "step": 8380
+    },
+    {
+      "epoch": 0.6383140596469872,
+      "grad_norm": 0.8811487521432458,
+      "learning_rate": 2e-05,
+      "loss": 0.5734702110290527,
+      "step": 8390
+    },
+    {
+      "epoch": 0.6390748630553865,
+      "grad_norm": 1.2326228090242388,
+      "learning_rate": 2e-05,
+      "loss": 0.5746047019958496,
+      "step": 8400
+    },
+    {
+      "epoch": 0.6398356664637858,
+      "grad_norm": 0.7319452828100158,
+      "learning_rate": 2e-05,
+      "loss": 0.5741769313812256,
+      "step": 8410
+    },
+    {
+      "epoch": 0.640596469872185,
+      "grad_norm": 0.7391288399995695,
+      "learning_rate": 2e-05,
+      "loss": 0.5748973369598389,
+      "step": 8420
+    },
+    {
+      "epoch": 0.6413572732805843,
+      "grad_norm": 0.7187781544525037,
+      "learning_rate": 2e-05,
+      "loss": 0.5756969451904297,
+      "step": 8430
+    },
+    {
+      "epoch": 0.6421180766889836,
+      "grad_norm": 0.9515665753922784,
+      "learning_rate": 2e-05,
+      "loss": 0.5824446678161621,
+      "step": 8440
+    },
+    {
+      "epoch": 0.6428788800973828,
+      "grad_norm": 0.768925477547387,
+      "learning_rate": 2e-05,
+      "loss": 0.5828074932098388,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6436396835057822,
+      "grad_norm": 0.7156136281064573,
+      "learning_rate": 2e-05,
+      "loss": 0.5788813591003418,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6444004869141814,
+      "grad_norm": 1.1729541063285784,
+      "learning_rate": 2e-05,
+      "loss": 0.5620977401733398,
+      "step": 8470
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 5.626099011709302,
+      "learning_rate": 2e-05,
+      "loss": 0.5899922847747803,
+      "step": 8480
+    },
+    {
+      "epoch": 0.6459220937309799,
+      "grad_norm": 1.2144163879229717,
+      "learning_rate": 2e-05,
+      "loss": 0.5962745666503906,
+      "step": 8490
+    },
+    {
+      "epoch": 0.6466828971393792,
+      "grad_norm": 1.757854447451676,
+      "learning_rate": 2e-05,
+      "loss": 0.5997524261474609,
+      "step": 8500
+    },
+    {
+      "epoch": 0.6474437005477784,
+      "grad_norm": 1.0588259040679382,
+      "learning_rate": 2e-05,
+      "loss": 0.5700427055358886,
+      "step": 8510
+    },
+    {
+      "epoch": 0.6482045039561777,
+      "grad_norm": 0.8408593512905853,
+      "learning_rate": 2e-05,
+      "loss": 0.5839073181152343,
+      "step": 8520
+    },
+    {
+      "epoch": 0.648965307364577,
+      "grad_norm": 0.7465243982406031,
+      "learning_rate": 2e-05,
+      "loss": 0.5978038787841797,
+      "step": 8530
+    },
+    {
+      "epoch": 0.6497261107729763,
+      "grad_norm": 3.2759112993123773,
+      "learning_rate": 2e-05,
+      "loss": 0.5808276176452637,
+      "step": 8540
+    },
+    {
+      "epoch": 0.6504869141813755,
+      "grad_norm": 0.7724452416946659,
+      "learning_rate": 2e-05,
+      "loss": 0.5701884269714356,
+      "step": 8550
+    },
+    {
+      "epoch": 0.6512477175897748,
+      "grad_norm": 0.9143320122105281,
+      "learning_rate": 2e-05,
+      "loss": 0.5884894847869873,
+      "step": 8560
+    },
+    {
+      "epoch": 0.6520085209981741,
+      "grad_norm": 1.1004424224758622,
+      "learning_rate": 2e-05,
+      "loss": 0.6164209365844726,
+      "step": 8570
+    },
+    {
+      "epoch": 0.6527693244065733,
+      "grad_norm": 3.7740652023481975,
+      "learning_rate": 2e-05,
+      "loss": 0.6149579048156738,
+      "step": 8580
+    },
+    {
+      "epoch": 0.6535301278149727,
+      "grad_norm": 1.7227938225222663,
+      "learning_rate": 2e-05,
+      "loss": 0.5946258068084717,
+      "step": 8590
+    },
+    {
+      "epoch": 0.6542909312233719,
+      "grad_norm": 1.6965703827516088,
+      "learning_rate": 2e-05,
+      "loss": 0.584904956817627,
+      "step": 8600
+    },
+    {
+      "epoch": 0.6550517346317711,
+      "grad_norm": 0.7798507437091755,
+      "learning_rate": 2e-05,
+      "loss": 0.595159912109375,
+      "step": 8610
+    },
+    {
+      "epoch": 0.6558125380401704,
+      "grad_norm": 0.9864119776001159,
+      "learning_rate": 2e-05,
+      "loss": 0.6034558773040771,
+      "step": 8620
+    },
+    {
+      "epoch": 0.6565733414485697,
+      "grad_norm": 0.7339425746923188,
+      "learning_rate": 2e-05,
+      "loss": 0.6056421279907227,
+      "step": 8630
+    },
+    {
+      "epoch": 0.657334144856969,
+      "grad_norm": 0.9239832429776678,
+      "learning_rate": 2e-05,
+      "loss": 0.578712272644043,
+      "step": 8640
+    },
+    {
+      "epoch": 0.6580949482653682,
+      "grad_norm": 0.9517578271754447,
+      "learning_rate": 2e-05,
+      "loss": 0.5903737545013428,
+      "step": 8650
+    },
+    {
+      "epoch": 0.6588557516737675,
+      "grad_norm": 3.1372938713515537,
+      "learning_rate": 2e-05,
+      "loss": 0.605405330657959,
+      "step": 8660
+    },
+    {
+      "epoch": 0.6596165550821668,
+      "grad_norm": 1.2612175379075883,
+      "learning_rate": 2e-05,
+      "loss": 0.5873723983764648,
+      "step": 8670
+    },
+    {
+      "epoch": 0.660377358490566,
+      "grad_norm": 1.0271896078322948,
+      "learning_rate": 2e-05,
+      "loss": 0.5908009529113769,
+      "step": 8680
+    },
+    {
+      "epoch": 0.6611381618989653,
+      "grad_norm": 0.8845359642142517,
+      "learning_rate": 2e-05,
+      "loss": 0.5836783885955811,
+      "step": 8690
+    },
+    {
+      "epoch": 0.6618989653073646,
+      "grad_norm": 0.8579061798557864,
+      "learning_rate": 2e-05,
+      "loss": 0.5821019172668457,
+      "step": 8700
+    },
+    {
+      "epoch": 0.6626597687157638,
+      "grad_norm": 1.88266104978089,
+      "learning_rate": 2e-05,
+      "loss": 0.5895304679870605,
+      "step": 8710
+    },
+    {
+      "epoch": 0.6634205721241632,
+      "grad_norm": 0.8112731478021488,
+      "learning_rate": 2e-05,
+      "loss": 0.6073898315429688,
+      "step": 8720
+    },
+    {
+      "epoch": 0.6641813755325624,
+      "grad_norm": 0.9302155688934056,
+      "learning_rate": 2e-05,
+      "loss": 0.5721456050872803,
+      "step": 8730
+    },
+    {
+      "epoch": 0.6649421789409616,
+      "grad_norm": 0.8957708573123366,
+      "learning_rate": 2e-05,
+      "loss": 0.5841497421264649,
+      "step": 8740
+    },
+    {
+      "epoch": 0.6657029823493609,
+      "grad_norm": 1.1842992333022206,
+      "learning_rate": 2e-05,
+      "loss": 0.5763440132141113,
+      "step": 8750
+    },
+    {
+      "epoch": 0.6664637857577602,
+      "grad_norm": 0.7521556745250447,
+      "learning_rate": 2e-05,
+      "loss": 0.5965114116668702,
+      "step": 8760
+    },
+    {
+      "epoch": 0.6672245891661595,
+      "grad_norm": 0.7608255687854952,
+      "learning_rate": 2e-05,
+      "loss": 0.5959710597991943,
+      "step": 8770
+    },
+    {
+      "epoch": 0.6679853925745587,
+      "grad_norm": 0.7175513179125987,
+      "learning_rate": 2e-05,
+      "loss": 0.5958961486816406,
+      "step": 8780
+    },
+    {
+      "epoch": 0.668746195982958,
+      "grad_norm": 0.9508989795442281,
+      "learning_rate": 2e-05,
+      "loss": 0.5883561134338379,
+      "step": 8790
+    },
+    {
+      "epoch": 0.6695069993913573,
+      "grad_norm": 0.756219681912444,
+      "learning_rate": 2e-05,
+      "loss": 0.5824656963348389,
+      "step": 8800
+    },
+    {
+      "epoch": 0.6702678027997565,
+      "grad_norm": 0.8804019367033353,
+      "learning_rate": 2e-05,
+      "loss": 0.5927609443664551,
+      "step": 8810
+    },
+    {
+      "epoch": 0.6710286062081559,
+      "grad_norm": 2.0632133908891115,
+      "learning_rate": 2e-05,
+      "loss": 0.5587677001953125,
+      "step": 8820
+    },
+    {
+      "epoch": 0.6717894096165551,
+      "grad_norm": 0.9429039792448954,
+      "learning_rate": 2e-05,
+      "loss": 0.5656592369079589,
+      "step": 8830
+    },
+    {
+      "epoch": 0.6725502130249543,
+      "grad_norm": 0.8796445687787332,
+      "learning_rate": 2e-05,
+      "loss": 0.580573844909668,
+      "step": 8840
+    },
+    {
+      "epoch": 0.6733110164333537,
+      "grad_norm": 1.1511367424005239,
+      "learning_rate": 2e-05,
+      "loss": 0.5834390163421631,
+      "step": 8850
+    },
+    {
+      "epoch": 0.6740718198417529,
+      "grad_norm": 0.7992783883797552,
+      "learning_rate": 2e-05,
+      "loss": 0.57703857421875,
+      "step": 8860
+    },
+    {
+      "epoch": 0.6748326232501521,
+      "grad_norm": 0.7979003200368248,
+      "learning_rate": 2e-05,
+      "loss": 0.5630064010620117,
+      "step": 8870
+    },
+    {
+      "epoch": 0.6755934266585514,
+      "grad_norm": 0.7776566530236206,
+      "learning_rate": 2e-05,
+      "loss": 0.6035675048828125,
+      "step": 8880
+    },
+    {
+      "epoch": 0.6763542300669507,
+      "grad_norm": 1.2405868401625446,
+      "learning_rate": 2e-05,
+      "loss": 0.598514461517334,
+      "step": 8890
+    },
+    {
+      "epoch": 0.67711503347535,
+      "grad_norm": 0.7159754504059158,
+      "learning_rate": 2e-05,
+      "loss": 0.5687870025634766,
+      "step": 8900
+    },
+    {
+      "epoch": 0.6778758368837492,
+      "grad_norm": 0.9474620409560017,
+      "learning_rate": 2e-05,
+      "loss": 0.5661998748779297,
+      "step": 8910
+    },
+    {
+      "epoch": 0.6786366402921485,
+      "grad_norm": 0.6898952119391581,
+      "learning_rate": 2e-05,
+      "loss": 0.5898018836975097,
+      "step": 8920
+    },
+    {
+      "epoch": 0.6793974437005478,
+      "grad_norm": 1.0114653120056036,
+      "learning_rate": 2e-05,
+      "loss": 0.5739587306976318,
+      "step": 8930
+    },
+    {
+      "epoch": 0.680158247108947,
+      "grad_norm": 0.8169371497902557,
+      "learning_rate": 2e-05,
+      "loss": 0.6165677070617676,
+      "step": 8940
+    },
+    {
+      "epoch": 0.6809190505173464,
+      "grad_norm": 1.5664025497040408,
+      "learning_rate": 2e-05,
+      "loss": 0.5932399272918701,
+      "step": 8950
+    },
+    {
+      "epoch": 0.6816798539257456,
+      "grad_norm": 0.8242456975643312,
+      "learning_rate": 2e-05,
+      "loss": 0.6072229385375977,
+      "step": 8960
+    },
+    {
+      "epoch": 0.6824406573341448,
+      "grad_norm": 0.7333082345943434,
+      "learning_rate": 2e-05,
+      "loss": 0.5818246364593506,
+      "step": 8970
+    },
+    {
+      "epoch": 0.6832014607425442,
+      "grad_norm": 0.8093240890477202,
+      "learning_rate": 2e-05,
+      "loss": 0.6507047653198242,
+      "step": 8980
+    },
+    {
+      "epoch": 0.6839622641509434,
+      "grad_norm": 0.9586781397804448,
+      "learning_rate": 2e-05,
+      "loss": 0.595789098739624,
+      "step": 8990
+    },
+    {
+      "epoch": 0.6847230675593426,
+      "grad_norm": 0.9377113202673869,
+      "learning_rate": 2e-05,
+      "loss": 0.5982228755950928,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6854838709677419,
+      "grad_norm": 6.242736329460499,
+      "learning_rate": 2e-05,
+      "loss": 0.5861414909362793,
+      "step": 9010
+    },
+    {
+      "epoch": 0.6862446743761412,
+      "grad_norm": 1.5919887450032766,
+      "learning_rate": 2e-05,
+      "loss": 0.5916693687438965,
+      "step": 9020
+    },
+    {
+      "epoch": 0.6870054777845405,
+      "grad_norm": 0.7881779305103546,
+      "learning_rate": 2e-05,
+      "loss": 0.6122342109680176,
+      "step": 9030
+    },
+    {
+      "epoch": 0.6877662811929397,
+      "grad_norm": 0.7386986257424547,
+      "learning_rate": 2e-05,
+      "loss": 0.5789780616760254,
+      "step": 9040
+    },
+    {
+      "epoch": 0.688527084601339,
+      "grad_norm": 0.6486092844354804,
+      "learning_rate": 2e-05,
+      "loss": 0.5800775527954102,
+      "step": 9050
+    },
+    {
+      "epoch": 0.6892878880097383,
+      "grad_norm": 1.0135446279312899,
+      "learning_rate": 2e-05,
+      "loss": 0.5938220977783203,
+      "step": 9060
+    },
+    {
+      "epoch": 0.6900486914181375,
+      "grad_norm": 0.6380069713170634,
+      "learning_rate": 2e-05,
+      "loss": 0.5813408851623535,
+      "step": 9070
+    },
+    {
+      "epoch": 0.6908094948265369,
+      "grad_norm": 0.948793942367441,
+      "learning_rate": 2e-05,
+      "loss": 0.5630834579467774,
+      "step": 9080
+    },
+    {
+      "epoch": 0.6915702982349361,
+      "grad_norm": 0.6455499938495354,
+      "learning_rate": 2e-05,
+      "loss": 0.5693106651306152,
+      "step": 9090
+    },
+    {
+      "epoch": 0.6923311016433353,
+      "grad_norm": 0.7308801205115985,
+      "learning_rate": 2e-05,
+      "loss": 0.5768764019012451,
+      "step": 9100
+    },
+    {
+      "epoch": 0.6930919050517347,
+      "grad_norm": 0.7816546949105124,
+      "learning_rate": 2e-05,
+      "loss": 0.5735910415649415,
+      "step": 9110
+    },
+    {
+      "epoch": 0.6938527084601339,
+      "grad_norm": 0.6932303232355076,
+      "learning_rate": 2e-05,
+      "loss": 0.5849563598632812,
+      "step": 9120
+    },
+    {
+      "epoch": 0.6946135118685332,
+      "grad_norm": 0.8710424563585363,
+      "learning_rate": 2e-05,
+      "loss": 0.5892991542816162,
+      "step": 9130
+    },
+    {
+      "epoch": 0.6953743152769324,
+      "grad_norm": 0.7586850950794953,
+      "learning_rate": 2e-05,
+      "loss": 0.5831239700317383,
+      "step": 9140
+    },
+    {
+      "epoch": 0.6961351186853317,
+      "grad_norm": 0.9351445646692809,
+      "learning_rate": 2e-05,
+      "loss": 0.5852271556854248,
+      "step": 9150
+    },
+    {
+      "epoch": 0.696895922093731,
+      "grad_norm": 0.7559472316922676,
+      "learning_rate": 2e-05,
+      "loss": 0.5689665794372558,
+      "step": 9160
+    },
+    {
+      "epoch": 0.6976567255021302,
+      "grad_norm": 0.7543560937612263,
+      "learning_rate": 2e-05,
+      "loss": 0.5850425720214844,
+      "step": 9170
+    },
+    {
+      "epoch": 0.6984175289105296,
+      "grad_norm": 1.0789187472049488,
+      "learning_rate": 2e-05,
+      "loss": 0.5705527305603028,
+      "step": 9180
+    },
+    {
+      "epoch": 0.6991783323189288,
+      "grad_norm": 1.0469551343384584,
+      "learning_rate": 2e-05,
+      "loss": 0.5672149658203125,
+      "step": 9190
+    },
+    {
+      "epoch": 0.699939135727328,
+      "grad_norm": 0.8194319374963502,
+      "learning_rate": 2e-05,
+      "loss": 0.5846150398254395,
+      "step": 9200
+    },
+    {
+      "epoch": 0.7006999391357274,
+      "grad_norm": 0.9172254037890788,
+      "learning_rate": 2e-05,
+      "loss": 0.5740792751312256,
+      "step": 9210
+    },
+    {
+      "epoch": 0.7014607425441266,
+      "grad_norm": 0.82770261540147,
+      "learning_rate": 2e-05,
+      "loss": 0.5912193298339844,
+      "step": 9220
+    },
+    {
+      "epoch": 0.7022215459525258,
+      "grad_norm": 0.6996391657053391,
+      "learning_rate": 2e-05,
+      "loss": 0.5842368125915527,
+      "step": 9230
+    },
+    {
+      "epoch": 0.7029823493609252,
+      "grad_norm": 0.8242913955017179,
+      "learning_rate": 2e-05,
+      "loss": 0.6001135826110839,
+      "step": 9240
+    },
+    {
+      "epoch": 0.7037431527693244,
+      "grad_norm": 0.8350583291359327,
+      "learning_rate": 2e-05,
+      "loss": 0.5744157314300538,
+      "step": 9250
+    },
+    {
+      "epoch": 0.7045039561777237,
+      "grad_norm": 0.8134322618280788,
+      "learning_rate": 2e-05,
+      "loss": 0.5866858959197998,
+      "step": 9260
+    },
+    {
+      "epoch": 0.7052647595861229,
+      "grad_norm": 0.6666300701502575,
+      "learning_rate": 2e-05,
+      "loss": 0.5675541877746582,
+      "step": 9270
+    },
+    {
+      "epoch": 0.7060255629945222,
+      "grad_norm": 0.829105564109662,
+      "learning_rate": 2e-05,
+      "loss": 0.5651575088500976,
+      "step": 9280
+    },
+    {
+      "epoch": 0.7067863664029215,
+      "grad_norm": 0.8297946317105535,
+      "learning_rate": 2e-05,
+      "loss": 0.5791358470916748,
+      "step": 9290
+    },
+    {
+      "epoch": 0.7075471698113207,
+      "grad_norm": 0.7274745404750123,
+      "learning_rate": 2e-05,
+      "loss": 0.5807729721069336,
+      "step": 9300
+    },
+    {
+      "epoch": 0.70830797321972,
+      "grad_norm": 0.6869353766187892,
+      "learning_rate": 2e-05,
+      "loss": 0.580677604675293,
+      "step": 9310
+    },
+    {
+      "epoch": 0.7090687766281193,
+      "grad_norm": 0.7466417552230146,
+      "learning_rate": 2e-05,
+      "loss": 0.590512466430664,
+      "step": 9320
+    },
+    {
+      "epoch": 0.7098295800365185,
+      "grad_norm": 0.8004807606854631,
+      "learning_rate": 2e-05,
+      "loss": 0.580935001373291,
+      "step": 9330
+    },
+    {
+      "epoch": 0.7105903834449179,
+      "grad_norm": 0.875379752560191,
+      "learning_rate": 2e-05,
+      "loss": 0.5752785682678223,
+      "step": 9340
+    },
+    {
+      "epoch": 0.7113511868533171,
+      "grad_norm": 2.4805900480579735,
+      "learning_rate": 2e-05,
+      "loss": 0.5815352439880371,
+      "step": 9350
+    },
+    {
+      "epoch": 0.7121119902617163,
+      "grad_norm": 1.0547728460024368,
+      "learning_rate": 2e-05,
+      "loss": 0.5940717697143555,
+      "step": 9360
+    },
+    {
+      "epoch": 0.7128727936701157,
+      "grad_norm": 0.6845225535265154,
+      "learning_rate": 2e-05,
+      "loss": 0.5843040466308593,
+      "step": 9370
+    },
+    {
+      "epoch": 0.7136335970785149,
+      "grad_norm": 0.7599665677272343,
+      "learning_rate": 2e-05,
+      "loss": 0.586016035079956,
+      "step": 9380
+    },
+    {
+      "epoch": 0.7143944004869142,
+      "grad_norm": 0.8808786458239948,
+      "learning_rate": 2e-05,
+      "loss": 0.565616512298584,
+      "step": 9390
+    },
+    {
+      "epoch": 0.7151552038953134,
+      "grad_norm": 1.0936559722436447,
+      "learning_rate": 2e-05,
+      "loss": 0.5825833797454834,
+      "step": 9400
+    },
+    {
+      "epoch": 0.7159160073037127,
+      "grad_norm": 1.165911040401158,
+      "learning_rate": 2e-05,
+      "loss": 0.5761720180511475,
+      "step": 9410
+    },
+    {
+      "epoch": 0.716676810712112,
+      "grad_norm": 0.6827515096578936,
+      "learning_rate": 2e-05,
+      "loss": 0.5835312843322754,
+      "step": 9420
+    },
+    {
+      "epoch": 0.7174376141205112,
+      "grad_norm": 0.8181284622291913,
+      "learning_rate": 2e-05,
+      "loss": 0.5893487930297852,
+      "step": 9430
+    },
+    {
+      "epoch": 0.7181984175289106,
+      "grad_norm": 0.7126122904695121,
+      "learning_rate": 2e-05,
+      "loss": 0.5732711791992188,
+      "step": 9440
+    },
+    {
+      "epoch": 0.7189592209373098,
+      "grad_norm": 0.6202534575785891,
+      "learning_rate": 2e-05,
+      "loss": 0.5479562282562256,
+      "step": 9450
+    },
+    {
+      "epoch": 0.719720024345709,
+      "grad_norm": 0.6760409974576045,
+      "learning_rate": 2e-05,
+      "loss": 0.5802877426147461,
+      "step": 9460
+    },
+    {
+      "epoch": 0.7204808277541084,
+      "grad_norm": 0.7670274895455869,
+      "learning_rate": 2e-05,
+      "loss": 0.591638708114624,
+      "step": 9470
+    },
+    {
+      "epoch": 0.7212416311625076,
+      "grad_norm": 0.8152649771964914,
+      "learning_rate": 2e-05,
+      "loss": 0.5652679443359375,
+      "step": 9480
+    },
+    {
+      "epoch": 0.7220024345709068,
+      "grad_norm": 0.7175943519914025,
+      "learning_rate": 2e-05,
+      "loss": 0.594538402557373,
+      "step": 9490
+    },
+    {
+      "epoch": 0.7227632379793062,
+      "grad_norm": 0.7911591107568428,
+      "learning_rate": 2e-05,
+      "loss": 0.5951120376586914,
+      "step": 9500
+    },
+    {
+      "epoch": 0.7235240413877054,
+      "grad_norm": 0.965272667570174,
+      "learning_rate": 2e-05,
+      "loss": 0.5757183074951172,
+      "step": 9510
+    },
+    {
+      "epoch": 0.7242848447961047,
+      "grad_norm": 0.6841984349317072,
+      "learning_rate": 2e-05,
+      "loss": 0.5897557735443115,
+      "step": 9520
+    },
+    {
+      "epoch": 0.7250456482045039,
+      "grad_norm": 0.9233883295124509,
+      "learning_rate": 2e-05,
+      "loss": 0.5622027397155762,
+      "step": 9530
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 1.3674226814621584,
+      "learning_rate": 2e-05,
+      "loss": 0.5813980102539062,
+      "step": 9540
+    },
+    {
+      "epoch": 0.7265672550213025,
+      "grad_norm": 0.8121150252789067,
+      "learning_rate": 2e-05,
+      "loss": 0.5714933395385742,
+      "step": 9550
+    },
+    {
+      "epoch": 0.7273280584297017,
+      "grad_norm": 0.659095642962995,
+      "learning_rate": 2e-05,
+      "loss": 0.5932172775268555,
+      "step": 9560
+    },
+    {
+      "epoch": 0.7280888618381011,
+      "grad_norm": 0.6798597025637846,
+      "learning_rate": 2e-05,
+      "loss": 0.5960627555847168,
+      "step": 9570
+    },
+    {
+      "epoch": 0.7288496652465003,
+      "grad_norm": 0.6699092950978714,
+      "learning_rate": 2e-05,
+      "loss": 0.5769166469573974,
+      "step": 9580
+    },
+    {
+      "epoch": 0.7296104686548995,
+      "grad_norm": 1.3957029527114035,
+      "learning_rate": 2e-05,
+      "loss": 0.5900737762451171,
+      "step": 9590
+    },
+    {
+      "epoch": 0.7303712720632989,
+      "grad_norm": 0.7756833362961219,
+      "learning_rate": 2e-05,
+      "loss": 0.5985897064208985,
+      "step": 9600
+    },
+    {
+      "epoch": 0.7311320754716981,
+      "grad_norm": 0.8711334782803486,
+      "learning_rate": 2e-05,
+      "loss": 0.5730056762695312,
+      "step": 9610
+    },
+    {
+      "epoch": 0.7318928788800974,
+      "grad_norm": 2.369722182629927,
+      "learning_rate": 2e-05,
+      "loss": 0.5857726097106933,
+      "step": 9620
+    },
+    {
+      "epoch": 0.7326536822884967,
+      "grad_norm": 0.6527375357694261,
+      "learning_rate": 2e-05,
+      "loss": 0.5798047065734864,
+      "step": 9630
+    },
+    {
+      "epoch": 0.7334144856968959,
+      "grad_norm": 0.7273263504994458,
+      "learning_rate": 2e-05,
+      "loss": 0.5659579277038574,
+      "step": 9640
+    },
+    {
+      "epoch": 0.7341752891052952,
+      "grad_norm": 0.7956922716512295,
+      "learning_rate": 2e-05,
+      "loss": 0.5790767192840576,
+      "step": 9650
+    },
+    {
+      "epoch": 0.7349360925136945,
+      "grad_norm": 1.0076426765781872,
+      "learning_rate": 2e-05,
+      "loss": 0.5465924739837646,
+      "step": 9660
+    },
+    {
+      "epoch": 0.7356968959220938,
+      "grad_norm": 0.8661565659493995,
+      "learning_rate": 2e-05,
+      "loss": 0.5700340270996094,
+      "step": 9670
+    },
+    {
+      "epoch": 0.736457699330493,
+      "grad_norm": 1.5774768149714482,
+      "learning_rate": 2e-05,
+      "loss": 0.5990691184997559,
+      "step": 9680
+    },
+    {
+      "epoch": 0.7372185027388922,
+      "grad_norm": 0.8094567525669571,
+      "learning_rate": 2e-05,
+      "loss": 0.5686656475067139,
+      "step": 9690
+    },
+    {
+      "epoch": 0.7379793061472916,
+      "grad_norm": 0.7409365326390452,
+      "learning_rate": 2e-05,
+      "loss": 0.5691412925720215,
+      "step": 9700
+    },
+    {
+      "epoch": 0.7387401095556908,
+      "grad_norm": 0.7587487012950567,
+      "learning_rate": 2e-05,
+      "loss": 0.5916703224182129,
+      "step": 9710
+    },
+    {
+      "epoch": 0.73950091296409,
+      "grad_norm": 0.8701795310503809,
+      "learning_rate": 2e-05,
+      "loss": 0.5911925315856934,
+      "step": 9720
+    },
+    {
+      "epoch": 0.7402617163724894,
+      "grad_norm": 1.2024804984889994,
+      "learning_rate": 2e-05,
+      "loss": 0.5805276870727539,
+      "step": 9730
+    },
+    {
+      "epoch": 0.7410225197808886,
+      "grad_norm": 1.1985080391530107,
+      "learning_rate": 2e-05,
+      "loss": 0.586190128326416,
+      "step": 9740
+    },
+    {
+      "epoch": 0.7417833231892879,
+      "grad_norm": 0.8506085291413545,
+      "learning_rate": 2e-05,
+      "loss": 0.5766258239746094,
+      "step": 9750
+    },
+    {
+      "epoch": 0.7425441265976872,
+      "grad_norm": 0.6646781708369977,
+      "learning_rate": 2e-05,
+      "loss": 0.5585824966430664,
+      "step": 9760
+    },
+    {
+      "epoch": 0.7433049300060864,
+      "grad_norm": 1.0756800882451514,
+      "learning_rate": 2e-05,
+      "loss": 0.6012025356292725,
+      "step": 9770
+    },
+    {
+      "epoch": 0.7440657334144857,
+      "grad_norm": 1.2709846294832607,
+      "learning_rate": 2e-05,
+      "loss": 0.5762775421142579,
+      "step": 9780
+    },
+    {
+      "epoch": 0.744826536822885,
+      "grad_norm": 0.8476257503214059,
+      "learning_rate": 2e-05,
+      "loss": 0.584918737411499,
+      "step": 9790
+    },
+    {
+      "epoch": 0.7455873402312843,
+      "grad_norm": 0.7504312515782672,
+      "learning_rate": 2e-05,
+      "loss": 0.5919755458831787,
+      "step": 9800
+    },
+    {
+      "epoch": 0.7463481436396835,
+      "grad_norm": 1.1439340320884488,
+      "learning_rate": 2e-05,
+      "loss": 0.5684781551361084,
+      "step": 9810
+    },
+    {
+      "epoch": 0.7471089470480827,
+      "grad_norm": 1.4070634557131456,
+      "learning_rate": 2e-05,
+      "loss": 0.5735734939575196,
+      "step": 9820
+    },
+    {
+      "epoch": 0.7478697504564821,
+      "grad_norm": 0.8819990753386632,
+      "learning_rate": 2e-05,
+      "loss": 0.5913722038269043,
+      "step": 9830
+    },
+    {
+      "epoch": 0.7486305538648813,
+      "grad_norm": 0.8756513092476454,
+      "learning_rate": 2e-05,
+      "loss": 0.5583992004394531,
+      "step": 9840
+    },
+    {
+      "epoch": 0.7493913572732805,
+      "grad_norm": 3.355446183123028,
+      "learning_rate": 2e-05,
+      "loss": 0.5858950614929199,
+      "step": 9850
+    },
+    {
+      "epoch": 0.7501521606816799,
+      "grad_norm": 1.0999914677117733,
+      "learning_rate": 2e-05,
+      "loss": 0.5813485622406006,
+      "step": 9860
+    },
+    {
+      "epoch": 0.7509129640900791,
+      "grad_norm": 0.7179643459154348,
+      "learning_rate": 2e-05,
+      "loss": 0.5640119075775146,
+      "step": 9870
+    },
+    {
+      "epoch": 0.7516737674984784,
+      "grad_norm": 0.8548138419418525,
+      "learning_rate": 2e-05,
+      "loss": 0.5778462886810303,
+      "step": 9880
+    },
+    {
+      "epoch": 0.7524345709068777,
+      "grad_norm": 0.875478307520945,
+      "learning_rate": 2e-05,
+      "loss": 0.5866836547851563,
+      "step": 9890
+    },
+    {
+      "epoch": 0.7531953743152769,
+      "grad_norm": 0.7740247733388252,
+      "learning_rate": 2e-05,
+      "loss": 0.6053698539733887,
+      "step": 9900
+    },
+    {
+      "epoch": 0.7539561777236762,
+      "grad_norm": 0.7037403116807366,
+      "learning_rate": 2e-05,
+      "loss": 0.5853007793426513,
+      "step": 9910
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.6901454853623428,
+      "learning_rate": 2e-05,
+      "loss": 0.5759538173675537,
+      "step": 9920
+    },
+    {
+      "epoch": 0.7554777845404748,
+      "grad_norm": 1.2340580328317117,
+      "learning_rate": 2e-05,
+      "loss": 0.5775564193725586,
+      "step": 9930
+    },
+    {
+      "epoch": 0.756238587948874,
+      "grad_norm": 0.8291036502881926,
+      "learning_rate": 2e-05,
+      "loss": 0.5873674392700196,
+      "step": 9940
+    },
+    {
+      "epoch": 0.7569993913572732,
+      "grad_norm": 0.8506536437197181,
+      "learning_rate": 2e-05,
+      "loss": 0.5913978099822998,
+      "step": 9950
+    },
+    {
+      "epoch": 0.7577601947656726,
+      "grad_norm": 0.6728490869292214,
+      "learning_rate": 2e-05,
+      "loss": 0.5777549743652344,
+      "step": 9960
+    },
+    {
+      "epoch": 0.7585209981740718,
+      "grad_norm": 0.6742539677630437,
+      "learning_rate": 2e-05,
+      "loss": 0.5861371994018555,
+      "step": 9970
+    },
+    {
+      "epoch": 0.759281801582471,
+      "grad_norm": 0.746902252492526,
+      "learning_rate": 2e-05,
+      "loss": 0.589985990524292,
+      "step": 9980
+    },
+    {
+      "epoch": 0.7600426049908704,
+      "grad_norm": 0.7864779131914174,
+      "learning_rate": 2e-05,
+      "loss": 0.562619686126709,
+      "step": 9990
+    },
+    {
+      "epoch": 0.7608034083992696,
+      "grad_norm": 2.7615144340716418,
+      "learning_rate": 2e-05,
+      "loss": 0.5876967906951904,
+      "step": 10000
+    },
+    {
+      "epoch": 0.7615642118076689,
+      "grad_norm": 1.5060337586943686,
+      "learning_rate": 2e-05,
+      "loss": 0.5768696784973144,
+      "step": 10010
+    },
+    {
+      "epoch": 0.7623250152160682,
+      "grad_norm": 0.7232961719313831,
+      "learning_rate": 2e-05,
+      "loss": 0.569008207321167,
+      "step": 10020
+    },
+    {
+      "epoch": 0.7630858186244674,
+      "grad_norm": 0.6744269734301248,
+      "learning_rate": 2e-05,
+      "loss": 0.5859694480895996,
+      "step": 10030
+    },
+    {
+      "epoch": 0.7638466220328667,
+      "grad_norm": 0.7337141816856212,
+      "learning_rate": 2e-05,
+      "loss": 0.5798294067382812,
+      "step": 10040
+    },
+    {
+      "epoch": 0.764607425441266,
+      "grad_norm": 1.009530635145469,
+      "learning_rate": 2e-05,
+      "loss": 0.5699119567871094,
+      "step": 10050
+    },
+    {
+      "epoch": 0.7653682288496653,
+      "grad_norm": 0.6910337712786491,
+      "learning_rate": 2e-05,
+      "loss": 0.5848464488983154,
+      "step": 10060
+    },
+    {
+      "epoch": 0.7661290322580645,
+      "grad_norm": 0.8633975725044752,
+      "learning_rate": 2e-05,
+      "loss": 0.5914453029632568,
+      "step": 10070
+    },
+    {
+      "epoch": 0.7668898356664637,
+      "grad_norm": 0.6699030772710253,
+      "learning_rate": 2e-05,
+      "loss": 0.5805097103118897,
+      "step": 10080
+    },
+    {
+      "epoch": 0.7676506390748631,
+      "grad_norm": 0.859998801358739,
+      "learning_rate": 2e-05,
+      "loss": 0.6010522842407227,
+      "step": 10090
+    },
+    {
+      "epoch": 0.7684114424832623,
+      "grad_norm": 0.7029276131136403,
+      "learning_rate": 2e-05,
+      "loss": 0.5853489875793457,
+      "step": 10100
+    },
+    {
+      "epoch": 0.7691722458916616,
+      "grad_norm": 1.0074076864322812,
+      "learning_rate": 2e-05,
+      "loss": 0.6009034633636474,
+      "step": 10110
+    },
+    {
+      "epoch": 0.7699330493000609,
+      "grad_norm": 1.0729175767976005,
+      "learning_rate": 2e-05,
+      "loss": 0.5768632411956787,
+      "step": 10120
+    },
+    {
+      "epoch": 0.7706938527084601,
+      "grad_norm": 0.6395859060113359,
+      "learning_rate": 2e-05,
+      "loss": 0.5670979499816895,
+      "step": 10130
+    },
+    {
+      "epoch": 0.7714546561168594,
+      "grad_norm": 0.6771671382653187,
+      "learning_rate": 2e-05,
+      "loss": 0.59348464012146,
+      "step": 10140
+    },
+    {
+      "epoch": 0.7722154595252587,
+      "grad_norm": 0.7623360165519525,
+      "learning_rate": 2e-05,
+      "loss": 0.5704171657562256,
+      "step": 10150
+    },
+    {
+      "epoch": 0.772976262933658,
+      "grad_norm": 0.8391449444020704,
+      "learning_rate": 2e-05,
+      "loss": 0.588199234008789,
+      "step": 10160
+    },
+    {
+      "epoch": 0.7737370663420572,
+      "grad_norm": 2.8327056591000424,
+      "learning_rate": 2e-05,
+      "loss": 0.5620534896850586,
+      "step": 10170
+    },
+    {
+      "epoch": 0.7744978697504565,
+      "grad_norm": 1.0454184975061165,
+      "learning_rate": 1.999957274187931e-05,
+      "loss": 0.6002476215362549,
+      "step": 10180
+    },
+    {
+      "epoch": 0.7752586731588558,
+      "grad_norm": 0.8887220555456862,
+      "learning_rate": 1.9991978062682906e-05,
+      "loss": 0.5725711822509766,
+      "step": 10190
+    },
+    {
+      "epoch": 0.776019476567255,
+      "grad_norm": 0.69208180885012,
+      "learning_rate": 1.997489709987378e-05,
+      "loss": 0.5558802604675293,
+      "step": 10200
+    },
+    {
+      "epoch": 0.7767802799756542,
+      "grad_norm": 0.7724217082477033,
+      "learning_rate": 1.9948346151502135e-05,
+      "loss": 0.5632465362548829,
+      "step": 10210
+    },
+    {
+      "epoch": 0.7775410833840536,
+      "grad_norm": 1.7252399679823243,
+      "learning_rate": 1.991235055154325e-05,
+      "loss": 0.5822501182556152,
+      "step": 10220
+    },
+    {
+      "epoch": 0.7783018867924528,
+      "grad_norm": 0.6669234790459866,
+      "learning_rate": 1.9866944645724693e-05,
+      "loss": 0.5902864456176757,
+      "step": 10230
+    },
+    {
+      "epoch": 0.7790626902008521,
+      "grad_norm": 0.7308709704823385,
+      "learning_rate": 1.9812171758754842e-05,
+      "loss": 0.5786218166351318,
+      "step": 10240
+    },
+    {
+      "epoch": 0.7798234936092514,
+      "grad_norm": 0.7404710221080116,
+      "learning_rate": 1.9748084152983987e-05,
+      "loss": 0.5762662410736084,
+      "step": 10250
+    },
+    {
+      "epoch": 0.7805842970176506,
+      "grad_norm": 2.2265162849677336,
+      "learning_rate": 1.967474297853744e-05,
+      "loss": 0.5839253902435303,
+      "step": 10260
+    },
+    {
+      "epoch": 0.7813451004260499,
+      "grad_norm": 0.6876630053534758,
+      "learning_rate": 1.9592218214968256e-05,
+      "loss": 0.5743198394775391,
+      "step": 10270
+    },
+    {
+      "epoch": 0.7821059038344492,
+      "grad_norm": 0.8705197879475898,
+      "learning_rate": 1.9500588604485213e-05,
+      "loss": 0.5845169067382813,
+      "step": 10280
+    },
+    {
+      "epoch": 0.7828667072428485,
+      "grad_norm": 0.9614529915162223,
+      "learning_rate": 1.939994157681979e-05,
+      "loss": 0.5903563976287842,
+      "step": 10290
+    },
+    {
+      "epoch": 0.7836275106512477,
+      "grad_norm": 0.674463249709875,
+      "learning_rate": 1.9290373165803803e-05,
+      "loss": 0.5918017387390136,
+      "step": 10300
+    },
+    {
+      "epoch": 0.784388314059647,
+      "grad_norm": 0.8458812751759092,
+      "learning_rate": 1.9171987917737328e-05,
+      "loss": 0.5602124691009521,
+      "step": 10310
+    },
+    {
+      "epoch": 0.7851491174680463,
+      "grad_norm": 0.7352570132663714,
+      "learning_rate": 1.9044898791634322e-05,
+      "loss": 0.5543444633483887,
+      "step": 10320
+    },
+    {
+      "epoch": 0.7859099208764455,
+      "grad_norm": 0.7693209191071677,
+      "learning_rate": 1.8909227051441126e-05,
+      "loss": 0.5618571758270263,
+      "step": 10330
+    },
+    {
+      "epoch": 0.7866707242848447,
+      "grad_norm": 0.7175022458838909,
+      "learning_rate": 1.8765102150330692e-05,
+      "loss": 0.5788544654846192,
+      "step": 10340
+    },
+    {
+      "epoch": 0.7874315276932441,
+      "grad_norm": 1.513972835049268,
+      "learning_rate": 1.861266160718297e-05,
+      "loss": 0.5989543914794921,
+      "step": 10350
+    },
+    {
+      "epoch": 0.7881923311016433,
+      "grad_norm": 0.7469910544137902,
+      "learning_rate": 1.845205087536923e-05,
+      "loss": 0.5885066986083984,
+      "step": 10360
+    },
+    {
+      "epoch": 0.7889531345100426,
+      "grad_norm": 0.8014409580290938,
+      "learning_rate": 1.8283423203965633e-05,
+      "loss": 0.5699684143066406,
+      "step": 10370
+    },
+    {
+      "epoch": 0.7897139379184419,
+      "grad_norm": 1.8549375516356503,
+      "learning_rate": 1.810693949152835e-05,
+      "loss": 0.5586061000823974,
+      "step": 10380
+    },
+    {
+      "epoch": 0.7904747413268411,
+      "grad_norm": 1.490362731132363,
+      "learning_rate": 1.7922768132569902e-05,
+      "loss": 0.5967247009277343,
+      "step": 10390
+    },
+    {
+      "epoch": 0.7912355447352404,
+      "grad_norm": 0.6651708526075258,
+      "learning_rate": 1.773108485688302e-05,
+      "loss": 0.5573324203491211,
+      "step": 10400
+    },
+    {
+      "epoch": 0.7919963481436397,
+      "grad_norm": 0.8446921702407698,
+      "learning_rate": 1.7532072561865537e-05,
+      "loss": 0.5793687343597412,
+      "step": 10410
+    },
+    {
+      "epoch": 0.792757151552039,
+      "grad_norm": 6.328447278221748,
+      "learning_rate": 1.732592113800616e-05,
+      "loss": 0.5869735717773438,
+      "step": 10420
+    },
+    {
+      "epoch": 0.7935179549604382,
+      "grad_norm": 0.7226116578249303,
+      "learning_rate": 1.7112827287697668e-05,
+      "loss": 0.5749219417572021,
+      "step": 10430
+    },
+    {
+      "epoch": 0.7942787583688375,
+      "grad_norm": 1.0007932137605338,
+      "learning_rate": 1.6892994337550452e-05,
+      "loss": 0.60792236328125,
+      "step": 10440
+    },
+    {
+      "epoch": 0.7950395617772368,
+      "grad_norm": 0.7205423340706808,
+      "learning_rate": 1.6666632044385487e-05,
+      "loss": 0.5718290328979492,
+      "step": 10450
+    },
+    {
+      "epoch": 0.795800365185636,
+      "grad_norm": 0.6486642168946863,
+      "learning_rate": 1.6433956395091757e-05,
+      "loss": 0.5519774436950684,
+      "step": 10460
+    },
+    {
+      "epoch": 0.7965611685940353,
+      "grad_norm": 0.6382726195430192,
+      "learning_rate": 1.6195189400539238e-05,
+      "loss": 0.5640114784240723,
+      "step": 10470
+    },
+    {
+      "epoch": 0.7973219720024346,
+      "grad_norm": 0.7447798392532713,
+      "learning_rate": 1.595055888374398e-05,
+      "loss": 0.5856643676757812,
+      "step": 10480
+    },
+    {
+      "epoch": 0.7980827754108338,
+      "grad_norm": 0.9609294775399304,
+      "learning_rate": 1.5700298262487434e-05,
+      "loss": 0.5922313690185547,
+      "step": 10490
+    },
+    {
+      "epoch": 0.7988435788192331,
+      "grad_norm": 0.6628978305459636,
+      "learning_rate": 1.5444646326597467e-05,
+      "loss": 0.5883934020996093,
+      "step": 10500
+    },
+    {
+      "epoch": 0.7996043822276324,
+      "grad_norm": 0.7746795498272551,
+      "learning_rate": 1.5183847010103586e-05,
+      "loss": 0.5630520820617676,
+      "step": 10510
+    },
+    {
+      "epoch": 0.8003651856360317,
+      "grad_norm": 0.6519868656824453,
+      "learning_rate": 1.4918149158483686e-05,
+      "loss": 0.5706429958343506,
+      "step": 10520
+    },
+    {
+      "epoch": 0.8011259890444309,
+      "grad_norm": 0.6929584867316987,
+      "learning_rate": 1.464780629122452e-05,
+      "loss": 0.5496623992919922,
+      "step": 10530
+    },
+    {
+      "epoch": 0.8018867924528302,
+      "grad_norm": 0.8430482222016796,
+      "learning_rate": 1.4373076359922374e-05,
+      "loss": 0.5693492889404297,
+      "step": 10540
+    },
+    {
+      "epoch": 0.8026475958612295,
+      "grad_norm": 1.8167968789849431,
+      "learning_rate": 1.4094221502154774e-05,
+      "loss": 0.5687737941741944,
+      "step": 10550
+    },
+    {
+      "epoch": 0.8034083992696287,
+      "grad_norm": 0.7101663680673093,
+      "learning_rate": 1.3811507791358113e-05,
+      "loss": 0.5903855323791504,
+      "step": 10560
+    },
+    {
+      "epoch": 0.804169202678028,
+      "grad_norm": 0.8583421482942496,
+      "learning_rate": 1.3525204982949753e-05,
+      "loss": 0.5772326469421387,
+      "step": 10570
+    },
+    {
+      "epoch": 0.8049300060864273,
+      "grad_norm": 0.7344653431701573,
+      "learning_rate": 1.3235586256936996e-05,
+      "loss": 0.5910390377044678,
+      "step": 10580
+    },
+    {
+      "epoch": 0.8056908094948265,
+      "grad_norm": 2.7310184792638212,
+      "learning_rate": 1.294292795725836e-05,
+      "loss": 0.5547958374023437,
+      "step": 10590
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 0.782791919411658,
+      "learning_rate": 1.2647509328105978e-05,
+      "loss": 0.5679276943206787,
+      "step": 10600
+    },
+    {
+      "epoch": 0.8072124163116251,
+      "grad_norm": 0.8078176392072653,
+      "learning_rate": 1.2349612247480686e-05,
+      "loss": 0.5391277313232422,
+      "step": 10610
+    },
+    {
+      "epoch": 0.8079732197200243,
+      "grad_norm": 0.7365362118847897,
+      "learning_rate": 1.2049520958234e-05,
+      "loss": 0.5633241176605225,
+      "step": 10620
+    },
+    {
+      "epoch": 0.8087340231284236,
+      "grad_norm": 0.7210307633066746,
+      "learning_rate": 1.1747521796853638e-05,
+      "loss": 0.5672507286071777,
+      "step": 10630
+    },
+    {
+      "epoch": 0.8094948265368229,
+      "grad_norm": 0.7189144017322421,
+      "learning_rate": 1.1443902920251395e-05,
+      "loss": 0.5810872077941894,
+      "step": 10640
+    },
+    {
+      "epoch": 0.8102556299452222,
+      "grad_norm": 0.651395163208002,
+      "learning_rate": 1.1138954030814009e-05,
+      "loss": 0.5726634979248046,
+      "step": 10650
+    },
+    {
+      "epoch": 0.8110164333536214,
+      "grad_norm": 0.5764294637400594,
+      "learning_rate": 1.0832966099979388e-05,
+      "loss": 0.549363088607788,
+      "step": 10660
+    },
+    {
+      "epoch": 0.8117772367620207,
+      "grad_norm": 0.6978810576412451,
+      "learning_rate": 1.0526231090601969e-05,
+      "loss": 0.5650720119476318,
+      "step": 10670
+    },
+    {
+      "epoch": 0.81253804017042,
+      "grad_norm": 0.6113039803811982,
+      "learning_rate": 1.0219041678372106e-05,
+      "loss": 0.5271758556365966,
+      "step": 10680
+    },
+    {
+      "epoch": 0.8132988435788192,
+      "grad_norm": 0.6246123834129601,
+      "learning_rate": 9.911690972555268e-06,
+      "loss": 0.550664234161377,
+      "step": 10690
+    },
+    {
+      "epoch": 0.8140596469872186,
+      "grad_norm": 0.6695146015087214,
+      "learning_rate": 9.604472236317541e-06,
+      "loss": 0.5445030212402344,
+      "step": 10700
+    },
+    {
+      "epoch": 0.8148204503956178,
+      "grad_norm": 0.6219742816493576,
+      "learning_rate": 9.297678606904335e-06,
+      "loss": 0.5690309047698975,
+      "step": 10710
+    },
+    {
+      "epoch": 0.815581253804017,
+      "grad_norm": 0.9912743924818217,
+      "learning_rate": 8.991602815939155e-06,
+      "loss": 0.567323112487793,
+      "step": 10720
+    },
+    {
+      "epoch": 0.8163420572124163,
+      "grad_norm": 0.5933312987327469,
+      "learning_rate": 8.686536910109464e-06,
+      "loss": 0.5462092399597168,
+      "step": 10730
+    },
+    {
+      "epoch": 0.8171028606208156,
+      "grad_norm": 0.723186465121715,
+      "learning_rate": 8.382771972506084e-06,
+      "loss": 0.5841447830200195,
+      "step": 10740
+    },
+    {
+      "epoch": 0.8178636640292148,
+      "grad_norm": 0.715250638984555,
+      "learning_rate": 8.08059784488198e-06,
+      "loss": 0.5661884307861328,
+      "step": 10750
+    },
+    {
+      "epoch": 0.8186244674376141,
+      "grad_norm": 0.7435422595326024,
+      "learning_rate": 7.780302851095544e-06,
+      "loss": 0.555213451385498,
+      "step": 10760
+    },
+    {
+      "epoch": 0.8193852708460134,
+      "grad_norm": 3.351432358152964,
+      "learning_rate": 7.482173522002166e-06,
+      "loss": 0.5564481258392334,
+      "step": 10770
+    },
+    {
+      "epoch": 0.8201460742544127,
+      "grad_norm": 0.7022252806428337,
+      "learning_rate": 7.186494322056624e-06,
+      "loss": 0.5552989959716796,
+      "step": 10780
+    },
+    {
+      "epoch": 0.8209068776628119,
+      "grad_norm": 0.6888338317107607,
+      "learning_rate": 6.893547377887183e-06,
+      "loss": 0.5662334442138672,
+      "step": 10790
+    },
+    {
+      "epoch": 0.8216676810712112,
+      "grad_norm": 0.6286516915252756,
+      "learning_rate": 6.6036122091003365e-06,
+      "loss": 0.5492068290710449,
+      "step": 10800
+    },
+    {
+      "epoch": 0.8224284844796105,
+      "grad_norm": 1.189514951023036,
+      "learning_rate": 6.316965461573097e-06,
+      "loss": 0.5570750713348389,
+      "step": 10810
+    },
+    {
+      "epoch": 0.8231892878880097,
+      "grad_norm": 0.647543246282592,
+      "learning_rate": 6.0338806434872965e-06,
+      "loss": 0.5426563262939453,
+      "step": 10820
+    },
+    {
+      "epoch": 0.8239500912964091,
+      "grad_norm": 0.708917679603483,
+      "learning_rate": 5.754627864357749e-06,
+      "loss": 0.537134313583374,
+      "step": 10830
+    },
+    {
+      "epoch": 0.8247108947048083,
+      "grad_norm": 0.8497144028837944,
+      "learning_rate": 5.479473577303319e-06,
+      "loss": 0.5433522701263428,
+      "step": 10840
+    },
+    {
+      "epoch": 0.8254716981132075,
+      "grad_norm": 0.633690862297814,
+      "learning_rate": 5.208680324806778e-06,
+      "loss": 0.5536088466644287,
+      "step": 10850
+    },
+    {
+      "epoch": 0.8262325015216068,
+      "grad_norm": 0.5714886209990258,
+      "learning_rate": 4.94250648820606e-06,
+      "loss": 0.563297176361084,
+      "step": 10860
+    },
+    {
+      "epoch": 0.8269933049300061,
+      "grad_norm": 0.6900192988148875,
+      "learning_rate": 4.681206041155951e-06,
+      "loss": 0.5643538475036621,
+      "step": 10870
+    },
+    {
+      "epoch": 0.8277541083384053,
+      "grad_norm": 0.6794230788456279,
+      "learning_rate": 4.42502830729538e-06,
+      "loss": 0.5542878627777099,
+      "step": 10880
+    },
+    {
+      "epoch": 0.8285149117468046,
+      "grad_norm": 0.5991074940959294,
+      "learning_rate": 4.174217722351669e-06,
+      "loss": 0.5505244255065918,
+      "step": 10890
+    },
+    {
+      "epoch": 0.8292757151552039,
+      "grad_norm": 0.6867268986625172,
+      "learning_rate": 3.929013600908575e-06,
+      "loss": 0.5363875389099121,
+      "step": 10900
+    },
+    {
+      "epoch": 0.8300365185636032,
+      "grad_norm": 0.7544256159859863,
+      "learning_rate": 3.689649908060807e-06,
+      "loss": 0.5387225151062012,
+      "step": 10910
+    },
+    {
+      "epoch": 0.8307973219720024,
+      "grad_norm": 1.5109524900221947,
+      "learning_rate": 3.4563550361728028e-06,
+      "loss": 0.5486934185028076,
+      "step": 10920
+    },
+    {
+      "epoch": 0.8315581253804017,
+      "grad_norm": 0.9223704774780548,
+      "learning_rate": 3.2293515869548216e-06,
+      "loss": 0.555975341796875,
+      "step": 10930
+    },
+    {
+      "epoch": 0.832318928788801,
+      "grad_norm": 0.6540449890101078,
+      "learning_rate": 3.0088561590642746e-06,
+      "loss": 0.5730948448181152,
+      "step": 10940
+    },
+    {
+      "epoch": 0.8330797321972002,
+      "grad_norm": 0.6144447300486312,
+      "learning_rate": 2.795079141434941e-06,
+      "loss": 0.5488675117492676,
+      "step": 10950
+    },
+    {
+      "epoch": 0.8338405356055996,
+      "grad_norm": 0.6171856690109754,
+      "learning_rate": 2.588224512531299e-06,
+      "loss": 0.5511569499969482,
+      "step": 10960
+    },
+    {
+      "epoch": 0.8346013390139988,
+      "grad_norm": 0.6170635836831424,
+      "learning_rate": 2.388489645719495e-06,
+      "loss": 0.5327147960662841,
+      "step": 10970
+    },
+    {
+      "epoch": 0.835362142422398,
+      "grad_norm": 0.770366003200345,
+      "learning_rate": 2.196065120940663e-06,
+      "loss": 0.5808749198913574,
+      "step": 10980
+    },
+    {
+      "epoch": 0.8361229458307973,
+      "grad_norm": 0.7056370609271586,
+      "learning_rate": 2.0111345428662808e-06,
+      "loss": 0.5421016216278076,
+      "step": 10990
+    },
+    {
+      "epoch": 0.8368837492391966,
+      "grad_norm": 0.6020828665289572,
+      "learning_rate": 1.8338743657091102e-06,
+      "loss": 0.5376517295837402,
+      "step": 11000
+    },
+    {
+      "epoch": 0.8376445526475959,
+      "grad_norm": 0.7010375316893029,
+      "learning_rate": 1.6644537248568113e-06,
+      "loss": 0.5477334499359131,
+      "step": 11010
+    },
+    {
+      "epoch": 0.8384053560559951,
+      "grad_norm": 0.7487046261199036,
+      "learning_rate": 1.503034275488951e-06,
+      "loss": 0.5702154159545898,
+      "step": 11020
+    },
+    {
+      "epoch": 0.8391661594643944,
+      "grad_norm": 1.5885091052085827,
+      "learning_rate": 1.3497700383313495e-06,
+      "loss": 0.5393186092376709,
+      "step": 11030
+    },
+    {
+      "epoch": 0.8399269628727937,
+      "grad_norm": 0.9102652506340425,
+      "learning_rate": 1.2048072526949562e-06,
+      "loss": 0.5652684211730957,
+      "step": 11040
+    },
+    {
+      "epoch": 0.8406877662811929,
+      "grad_norm": 0.7231727752776659,
+      "learning_rate": 1.0682842369394756e-06,
+      "loss": 0.5701922416687012,
+      "step": 11050
+    },
+    {
+      "epoch": 0.8414485696895923,
+      "grad_norm": 0.8821988723011467,
+      "learning_rate": 9.403312564948792e-07,
+      "loss": 0.5464212417602539,
+      "step": 11060
+    },
+    {
+      "epoch": 0.8422093730979915,
+      "grad_norm": 1.265996338745651,
+      "learning_rate": 8.210703995667471e-07,
+      "loss": 0.5471435070037842,
+      "step": 11070
+    },
+    {
+      "epoch": 0.8429701765063907,
+      "grad_norm": 0.6748151011781678,
+      "learning_rate": 7.106154606440245e-07,
+      "loss": 0.5455111503601074,
+      "step": 11080
+    },
+    {
+      "epoch": 0.8437309799147901,
+      "grad_norm": 0.7128271907135233,
+      "learning_rate": 6.090718319203393e-07,
+      "loss": 0.5507564544677734,
+      "step": 11090
+    },
+    {
+      "epoch": 0.8444917833231893,
+      "grad_norm": 0.6107443150009458,
+      "learning_rate": 5.165364027325089e-07,
+      "loss": 0.5409876823425293,
+      "step": 11100
+    },
+    {
+      "epoch": 0.8452525867315885,
+      "grad_norm": 1.8741549241789737,
+      "learning_rate": 4.330974671121585e-07,
+      "loss": 0.5733149528503418,
+      "step": 11110
+    },
+    {
+      "epoch": 0.8460133901399878,
+      "grad_norm": 0.6642907640010088,
+      "learning_rate": 3.588346395386846e-07,
+      "loss": 0.5486061096191406,
+      "step": 11120
+    },
+    {
+      "epoch": 0.8467741935483871,
+      "grad_norm": 0.7160281368652278,
+      "learning_rate": 2.938187789739395e-07,
+      "loss": 0.5614269256591797,
+      "step": 11130
+    },
+    {
+      "epoch": 0.8475349969567864,
+      "grad_norm": 0.7835951744552293,
+      "learning_rate": 2.3811192125110724e-07,
+      "loss": 0.5443567276000977,
+      "step": 11140
+    },
+    {
+      "epoch": 0.8482958003651856,
+      "grad_norm": 1.0144266828653998,
+      "learning_rate": 1.917672198823177e-07,
+      "loss": 0.5352001190185547,
+      "step": 11150
+    },
+    {
+      "epoch": 0.8490566037735849,
+      "grad_norm": 0.8478296270867579,
+      "learning_rate": 1.5482889534143795e-07,
+      "loss": 0.5491334438323975,
+      "step": 11160
+    },
+    {
+      "epoch": 0.8498174071819842,
+      "grad_norm": 0.7015700787783927,
+      "learning_rate": 1.273321928704694e-07,
+      "loss": 0.5450725078582763,
+      "step": 11170
+    },
+    {
+      "epoch": 0.8505782105903834,
+      "grad_norm": 0.8094022232432354,
+      "learning_rate": 1.0930334884977866e-07,
+      "loss": 0.538828992843628,
+      "step": 11180
+    },
+    {
+      "epoch": 0.8513390139987828,
+      "grad_norm": 0.966867971199675,
+      "learning_rate": 1.0075956576427487e-07,
+      "loss": 0.5359937191009522,
+      "step": 11190
+    },
+    {
+      "epoch": 0.8515672550213025,
+      "step": 11193,
+      "total_flos": 5300110976090112.0,
+      "train_loss": 0.051110071055128485,
+      "train_runtime": 14757.1165,
+      "train_samples_per_second": 9.102,
+      "train_steps_per_second": 0.758
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 11193,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 636,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5300110976090112.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}