diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9022 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 31.446190102120973,
+  "global_step": 15000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002094789211835559,
+      "learning_rate": 1.0206207261596573e-07,
+      "loss": 42.596397399902344,
+      "step": 1
+    },
+    {
+      "epoch": 0.020947892118355592,
+      "learning_rate": 1.0206207261596575e-06,
+      "loss": 41.090047200520836,
+      "step": 10
+    },
+    {
+      "epoch": 0.041895784236711184,
+      "learning_rate": 2.041241452319315e-06,
+      "loss": 40.72335205078125,
+      "step": 20
+    },
+    {
+      "epoch": 0.06284367635506677,
+      "learning_rate": 3.0618621784789722e-06,
+      "loss": 40.80942077636719,
+      "step": 30
+    },
+    {
+      "epoch": 0.08379156847342237,
+      "learning_rate": 4.08248290463863e-06,
+      "loss": 40.0015869140625,
+      "step": 40
+    },
+    {
+      "epoch": 0.10473946059177795,
+      "learning_rate": 5.103103630798286e-06,
+      "loss": 39.082891845703124,
+      "step": 50
+    },
+    {
+      "epoch": 0.12568735271013354,
+      "learning_rate": 6.1237243569579445e-06,
+      "loss": 39.320306396484376,
+      "step": 60
+    },
+    {
+      "epoch": 0.14663524482848914,
+      "learning_rate": 7.144345083117603e-06,
+      "loss": 38.66647033691406,
+      "step": 70
+    },
+    {
+      "epoch": 0.16758313694684474,
+      "learning_rate": 8.16496580927726e-06,
+      "loss": 38.0055908203125,
+      "step": 80
+    },
+    {
+      "epoch": 0.1885310290652003,
+      "learning_rate": 9.185586535436916e-06,
+      "loss": 37.521505737304686,
+      "step": 90
+    },
+    {
+      "epoch": 0.2094789211835559,
+      "learning_rate": 1.0206207261596573e-05,
+      "loss": 35.75002746582031,
+      "step": 100
+    },
+    {
+      "epoch": 0.2304268133019115,
+      "learning_rate": 1.1226827987756233e-05,
+      "loss": 35.38407592773437,
+      "step": 110
+    },
+    {
+      "epoch": 0.2513747054202671,
+      "learning_rate": 1.2247448713915889e-05,
+      "loss": 34.455230712890625,
+      "step": 120
+    },
+    {
+      "epoch": 0.2723225975386227,
+      "learning_rate": 1.3268069440075545e-05,
+      "loss": 33.48695373535156,
+      "step": 130
+    },
+    {
+      "epoch": 0.2932704896569783,
+      "learning_rate": 1.4288690166235205e-05,
+      "loss": 33.0624755859375,
+      "step": 140
+    },
+    {
+      "epoch": 0.31421838177533384,
+      "learning_rate": 1.530931089239486e-05,
+      "loss": 31.633297729492188,
+      "step": 150
+    },
+    {
+      "epoch": 0.33516627389368947,
+      "learning_rate": 1.632993161855452e-05,
+      "loss": 30.392620849609376,
+      "step": 160
+    },
+    {
+      "epoch": 0.35611416601204504,
+      "learning_rate": 1.7350552344714174e-05,
+      "loss": 30.184588623046874,
+      "step": 170
+    },
+    {
+      "epoch": 0.3770620581304006,
+      "learning_rate": 1.8371173070873833e-05,
+      "loss": 29.326535034179688,
+      "step": 180
+    },
+    {
+      "epoch": 0.39800995024875624,
+      "learning_rate": 1.939179379703349e-05,
+      "loss": 28.23697509765625,
+      "step": 190
+    },
+    {
+      "epoch": 0.4189578423671118,
+      "learning_rate": 2.0412414523193145e-05,
+      "loss": 27.301419067382813,
+      "step": 200
+    },
+    {
+      "epoch": 0.4399057344854674,
+      "learning_rate": 2.1433035249352804e-05,
+      "loss": 26.5281494140625,
+      "step": 210
+    },
+    {
+      "epoch": 0.460853626603823,
+      "learning_rate": 2.2453655975512465e-05,
+      "loss": 25.992547607421876,
+      "step": 220
+    },
+    {
+      "epoch": 0.4818015187221786,
+      "learning_rate": 2.347427670167212e-05,
+      "loss": 24.90663299560547,
+      "step": 230
+    },
+    {
+      "epoch": 0.5027494108405341,
+      "learning_rate": 2.4494897427831778e-05,
+      "loss": 23.896534729003907,
+      "step": 240
+    },
+    {
+      "epoch": 0.5236973029588897,
+      "learning_rate": 2.5515518153991436e-05,
+      "loss": 22.488262939453126,
+      "step": 250
+    },
+    {
+      "epoch": 0.5446451950772454,
+      "learning_rate": 2.653613888015109e-05,
+      "loss": 21.760206604003905,
+      "step": 260
+    },
+    {
+      "epoch": 0.565593087195601,
+      "learning_rate": 2.755675960631075e-05,
+      "loss": 20.808561706542967,
+      "step": 270
+    },
+    {
+      "epoch": 0.5865409793139565,
+      "learning_rate": 2.857738033247041e-05,
+      "loss": 20.107774353027345,
+      "step": 280
+    },
+    {
+      "epoch": 0.6074888714323121,
+      "learning_rate": 2.9598001058630065e-05,
+      "loss": 19.469386291503906,
+      "step": 290
+    },
+    {
+      "epoch": 0.6284367635506677,
+      "learning_rate": 3.061862178478972e-05,
+      "loss": 18.442234802246094,
+      "step": 300
+    },
+    {
+      "epoch": 0.6493846556690233,
+      "learning_rate": 3.163924251094938e-05,
+      "loss": 18.132992553710938,
+      "step": 310
+    },
+    {
+      "epoch": 0.6703325477873789,
+      "learning_rate": 3.265986323710904e-05,
+      "loss": 17.425698852539064,
+      "step": 320
+    },
+    {
+      "epoch": 0.6912804399057345,
+      "learning_rate": 3.3680483963268694e-05,
+      "loss": 16.683474731445312,
+      "step": 330
+    },
+    {
+      "epoch": 0.7122283320240901,
+      "learning_rate": 3.470110468942835e-05,
+      "loss": 16.511445617675783,
+      "step": 340
+    },
+    {
+      "epoch": 0.7331762241424457,
+      "learning_rate": 3.5721725415588004e-05,
+      "loss": 15.8478759765625,
+      "step": 350
+    },
+    {
+      "epoch": 0.7541241162608012,
+      "learning_rate": 3.6742346141747665e-05,
+      "loss": 15.450515747070312,
+      "step": 360
+    },
+    {
+      "epoch": 0.7750720083791568,
+      "learning_rate": 3.7762966867907327e-05,
+      "loss": 14.916090393066407,
+      "step": 370
+    },
+    {
+      "epoch": 0.7960199004975125,
+      "learning_rate": 3.878358759406698e-05,
+      "loss": 14.495413208007813,
+      "step": 380
+    },
+    {
+      "epoch": 0.816967792615868,
+      "learning_rate": 3.980420832022664e-05,
+      "loss": 13.862504577636718,
+      "step": 390
+    },
+    {
+      "epoch": 0.8379156847342236,
+      "learning_rate": 4.082482904638629e-05,
+      "loss": 13.526719665527343,
+      "step": 400
+    },
+    {
+      "epoch": 0.8588635768525792,
+      "learning_rate": 4.184544977254595e-05,
+      "loss": 13.241981506347656,
+      "step": 410
+    },
+    {
+      "epoch": 0.8798114689709348,
+      "learning_rate": 4.286607049870561e-05,
+      "loss": 12.791949462890624,
+      "step": 420
+    },
+    {
+      "epoch": 0.9007593610892904,
+      "learning_rate": 4.388669122486527e-05,
+      "loss": 12.353260040283203,
+      "step": 430
+    },
+    {
+      "epoch": 0.921707253207646,
+      "learning_rate": 4.490731195102493e-05,
+      "loss": 11.939605712890625,
+      "step": 440
+    },
+    {
+      "epoch": 0.9426551453260016,
+      "learning_rate": 4.5927932677184585e-05,
+      "loss": 11.615445709228515,
+      "step": 450
+    },
+    {
+      "epoch": 0.9636030374443572,
+      "learning_rate": 4.694855340334424e-05,
+      "loss": 11.29063720703125,
+      "step": 460
+    },
+    {
+      "epoch": 0.9845509295627127,
+      "learning_rate": 4.7969174129503894e-05,
+      "loss": 11.051673889160156,
+      "step": 470
+    },
+    {
+      "epoch": 1.0062843676355067,
+      "learning_rate": 4.8989794855663556e-05,
+      "loss": 11.174005889892578,
+      "step": 480
+    },
+    {
+      "epoch": 1.0272322597538623,
+      "learning_rate": 5.001041558182322e-05,
+      "loss": 10.351375579833984,
+      "step": 490
+    },
+    {
+      "epoch": 1.0481801518722178,
+      "learning_rate": 5.103103630798287e-05,
+      "loss": 10.082479858398438,
+      "step": 500
+    },
+    {
+      "epoch": 1.0691280439905735,
+      "learning_rate": 5.205165703414253e-05,
+      "loss": 9.848101806640624,
+      "step": 510
+    },
+    {
+      "epoch": 1.090075936108929,
+      "learning_rate": 5.307227776030218e-05,
+      "loss": 9.558543395996093,
+      "step": 520
+    },
+    {
+      "epoch": 1.1110238282272846,
+      "learning_rate": 5.409289848646184e-05,
+      "loss": 9.263871765136718,
+      "step": 530
+    },
+    {
+      "epoch": 1.1319717203456403,
+      "learning_rate": 5.51135192126215e-05,
+      "loss": 9.101696014404297,
+      "step": 540
+    },
+    {
+      "epoch": 1.1529196124639958,
+      "learning_rate": 5.613413993878116e-05,
+      "loss": 8.784052276611328,
+      "step": 550
+    },
+    {
+      "epoch": 1.1738675045823515,
+      "learning_rate": 5.715476066494082e-05,
+      "loss": 8.466715240478516,
+      "step": 560
+    },
+    {
+      "epoch": 1.194815396700707,
+      "learning_rate": 5.817538139110047e-05,
+      "loss": 8.419536590576172,
+      "step": 570
+    },
+    {
+      "epoch": 1.2157632888190626,
+      "learning_rate": 5.919600211726013e-05,
+      "loss": 8.445430755615234,
+      "step": 580
+    },
+    {
+      "epoch": 1.236711180937418,
+      "learning_rate": 6.0216622843419785e-05,
+      "loss": 7.745582580566406,
+      "step": 590
+    },
+    {
+      "epoch": 1.2576590730557737,
+      "learning_rate": 6.123724356957945e-05,
+      "loss": 7.821333312988282,
+      "step": 600
+    },
+    {
+      "epoch": 1.2786069651741294,
+      "learning_rate": 6.22578642957391e-05,
+      "loss": 7.500454711914062,
+      "step": 610
+    },
+    {
+      "epoch": 1.2995548572924849,
+      "learning_rate": 6.327848502189876e-05,
+      "loss": 7.44578857421875,
+      "step": 620
+    },
+    {
+      "epoch": 1.3205027494108406,
+      "learning_rate": 6.429910574805841e-05,
+      "loss": 7.184627532958984,
+      "step": 630
+    },
+    {
+      "epoch": 1.341450641529196,
+      "learning_rate": 6.531972647421808e-05,
+      "loss": 6.880846405029297,
+      "step": 640
+    },
+    {
+      "epoch": 1.3623985336475517,
+      "learning_rate": 6.634034720037773e-05,
+      "loss": 6.639464569091797,
+      "step": 650
+    },
+    {
+      "epoch": 1.3833464257659074,
+      "learning_rate": 6.736096792653739e-05,
+      "loss": 6.543840789794922,
+      "step": 660
+    },
+    {
+      "epoch": 1.4042943178842628,
+      "learning_rate": 6.838158865269704e-05,
+      "loss": 6.245196914672851,
+      "step": 670
+    },
+    {
+      "epoch": 1.4252422100026185,
+      "learning_rate": 6.94022093788567e-05,
+      "loss": 6.316292572021484,
+      "step": 680
+    },
+    {
+      "epoch": 1.446190102120974,
+      "learning_rate": 7.042283010501637e-05,
+      "loss": 6.2703697204589846,
+      "step": 690
+    },
+    {
+      "epoch": 1.4671379942393297,
+      "learning_rate": 7.144345083117601e-05,
+      "loss": 6.09345703125,
+      "step": 700
+    },
+    {
+      "epoch": 1.4880858863576854,
+      "learning_rate": 7.246407155733568e-05,
+      "loss": 5.9814506530761715,
+      "step": 710
+    },
+    {
+      "epoch": 1.5090337784760408,
+      "learning_rate": 7.348469228349533e-05,
+      "loss": 5.9337646484375,
+      "step": 720
+    },
+    {
+      "epoch": 1.5299816705943965,
+      "learning_rate": 7.450531300965498e-05,
+      "loss": 5.937409591674805,
+      "step": 730
+    },
+    {
+      "epoch": 1.550929562712752,
+      "learning_rate": 7.552593373581465e-05,
+      "loss": 5.64327392578125,
+      "step": 740
+    },
+    {
+      "epoch": 1.5718774548311076,
+      "learning_rate": 7.654655446197431e-05,
+      "loss": 5.371760559082031,
+      "step": 750
+    },
+    {
+      "epoch": 1.5928253469494633,
+      "learning_rate": 7.756717518813396e-05,
+      "loss": 5.309605407714844,
+      "step": 760
+    },
+    {
+      "epoch": 1.6137732390678188,
+      "learning_rate": 7.858779591429362e-05,
+      "loss": 5.3783222198486325,
+      "step": 770
+    },
+    {
+      "epoch": 1.6347211311861742,
+      "learning_rate": 7.960841664045329e-05,
+      "loss": 5.400894546508789,
+      "step": 780
+    },
+    {
+      "epoch": 1.65566902330453,
+      "learning_rate": 8.062903736661294e-05,
+      "loss": 5.137008285522461,
+      "step": 790
+    },
+    {
+      "epoch": 1.6766169154228856,
+      "learning_rate": 8.164965809277258e-05,
+      "loss": 5.365228271484375,
+      "step": 800
+    },
+    {
+      "epoch": 1.6975648075412413,
+      "learning_rate": 8.267027881893225e-05,
+      "loss": 5.210577392578125,
+      "step": 810
+    },
+    {
+      "epoch": 1.7185126996595967,
+      "learning_rate": 8.36908995450919e-05,
+      "loss": 5.149754333496094,
+      "step": 820
+    },
+    {
+      "epoch": 1.7394605917779522,
+      "learning_rate": 8.471152027125156e-05,
+      "loss": 5.063209915161133,
+      "step": 830
+    },
+    {
+      "epoch": 1.7604084838963079,
+      "learning_rate": 8.573214099741121e-05,
+      "loss": 4.89969482421875,
+      "step": 840
+    },
+    {
+      "epoch": 1.7813563760146636,
+      "learning_rate": 8.675276172357088e-05,
+      "loss": 4.8380378723144535,
+      "step": 850
+    },
+    {
+      "epoch": 1.8023042681330192,
+      "learning_rate": 8.777338244973054e-05,
+      "loss": 4.826490783691407,
+      "step": 860
+    },
+    {
+      "epoch": 1.8232521602513747,
+      "learning_rate": 8.879400317589019e-05,
+      "loss": 4.733642578125,
+      "step": 870
+    },
+    {
+      "epoch": 1.8442000523697302,
+      "learning_rate": 8.981462390204986e-05,
+      "loss": 4.840193557739258,
+      "step": 880
+    },
+    {
+      "epoch": 1.8651479444880859,
+      "learning_rate": 9.083524462820951e-05,
+      "loss": 4.841461944580078,
+      "step": 890
+    },
+    {
+      "epoch": 1.8860958366064415,
+      "learning_rate": 9.185586535436917e-05,
+      "loss": 4.591343688964844,
+      "step": 900
+    },
+    {
+      "epoch": 1.9070437287247972,
+      "learning_rate": 9.287648608052881e-05,
+      "loss": 4.796835708618164,
+      "step": 910
+    },
+    {
+      "epoch": 1.9279916208431527,
+      "learning_rate": 9.389710680668848e-05,
+      "loss": 4.928312301635742,
+      "step": 920
+    },
+    {
+      "epoch": 1.9489395129615081,
+      "learning_rate": 9.491772753284813e-05,
+      "loss": 4.6936603546142575,
+      "step": 930
+    },
+    {
+      "epoch": 1.9698874050798638,
+      "learning_rate": 9.593834825900779e-05,
+      "loss": 4.678832626342773,
+      "step": 940
+    },
+    {
+      "epoch": 1.9908352971982195,
+      "learning_rate": 9.695896898516746e-05,
+      "loss": 4.801474380493164,
+      "step": 950
+    },
+    {
+      "epoch": 2.0125687352710133,
+      "learning_rate": 9.797958971132711e-05,
+      "loss": 4.615579986572266,
+      "step": 960
+    },
+    {
+      "epoch": 2.033516627389369,
+      "learning_rate": 9.900021043748677e-05,
+      "loss": 4.685293197631836,
+      "step": 970
+    },
+    {
+      "epoch": 2.0544645195077247,
+      "learning_rate": 0.00010002083116364643,
+      "loss": 4.529672622680664,
+      "step": 980
+    },
+    {
+      "epoch": 2.07541241162608,
+      "learning_rate": 0.00010104145188980609,
+      "loss": 4.6519828796386715,
+      "step": 990
+    },
+    {
+      "epoch": 2.0963603037444356,
+      "learning_rate": 0.00010206207261596574,
+      "loss": 4.200122451782226,
+      "step": 1000
+    },
+    {
+      "epoch": 2.1173081958627913,
+      "learning_rate": 0.0001030826933421254,
+      "loss": 4.467970275878907,
+      "step": 1010
+    },
+    {
+      "epoch": 2.138256087981147,
+      "learning_rate": 0.00010410331406828505,
+      "loss": 4.364266586303711,
+      "step": 1020
+    },
+    {
+      "epoch": 2.1592039800995027,
+      "learning_rate": 0.00010512393479444471,
+      "loss": 4.321992874145508,
+      "step": 1030
+    },
+    {
+      "epoch": 2.180151872217858,
+      "learning_rate": 0.00010614455552060436,
+      "loss": 4.208817672729492,
+      "step": 1040
+    },
+    {
+      "epoch": 2.2010997643362136,
+      "learning_rate": 0.00010716517624676403,
+      "loss": 4.414374923706054,
+      "step": 1050
+    },
+    {
+      "epoch": 2.2220476564545693,
+      "learning_rate": 0.00010818579697292369,
+      "loss": 4.222100067138672,
+      "step": 1060
+    },
+    {
+      "epoch": 2.242995548572925,
+      "learning_rate": 0.00010920641769908334,
+      "loss": 4.392937850952149,
+      "step": 1070
+    },
+    {
+      "epoch": 2.2639434406912806,
+      "learning_rate": 0.000110227038425243,
+      "loss": 4.361217498779297,
+      "step": 1080
+    },
+    {
+      "epoch": 2.284891332809636,
+      "learning_rate": 0.00011124765915140266,
+      "loss": 4.305131912231445,
+      "step": 1090
+    },
+    {
+      "epoch": 2.3058392249279915,
+      "learning_rate": 0.00011226827987756232,
+      "loss": 4.322722244262695,
+      "step": 1100
+    },
+    {
+      "epoch": 2.326787117046347,
+      "learning_rate": 0.00011328890060372197,
+      "loss": 4.250308990478516,
+      "step": 1110
+    },
+    {
+      "epoch": 2.347735009164703,
+      "learning_rate": 0.00011430952132988164,
+      "loss": 4.366016006469726,
+      "step": 1120
+    },
+    {
+      "epoch": 2.3686829012830586,
+      "learning_rate": 0.00011533014205604128,
+      "loss": 4.271330642700195,
+      "step": 1130
+    },
+    {
+      "epoch": 2.389630793401414,
+      "learning_rate": 0.00011635076278220094,
+      "loss": 4.35943489074707,
+      "step": 1140
+    },
+    {
+      "epoch": 2.4105786855197695,
+      "learning_rate": 0.00011737138350836059,
+      "loss": 4.14013442993164,
+      "step": 1150
+    },
+    {
+      "epoch": 2.431526577638125,
+      "learning_rate": 0.00011839200423452026,
+      "loss": 4.263423156738281,
+      "step": 1160
+    },
+    {
+      "epoch": 2.452474469756481,
+      "learning_rate": 0.00011941262496067991,
+      "loss": 4.046255874633789,
+      "step": 1170
+    },
+    {
+      "epoch": 2.473422361874836,
+      "learning_rate": 0.00012043324568683957,
+      "loss": 4.267144775390625,
+      "step": 1180
+    },
+    {
+      "epoch": 2.494370253993192,
+      "learning_rate": 0.00012145386641299924,
+      "loss": 4.446724319458008,
+      "step": 1190
+    },
+    {
+      "epoch": 2.5153181461115475,
+      "learning_rate": 0.0001224744871391589,
+      "loss": 4.310148239135742,
+      "step": 1200
+    },
+    {
+      "epoch": 2.536266038229903,
+      "learning_rate": 0.00012349510786531856,
+      "loss": 4.145759963989258,
+      "step": 1210
+    },
+    {
+      "epoch": 2.557213930348259,
+      "learning_rate": 0.0001245157285914782,
+      "loss": 3.9344154357910157,
+      "step": 1220
+    },
+    {
+      "epoch": 2.5781618224666145,
+      "learning_rate": 0.00012553634931763784,
+      "loss": 4.1616455078125,
+      "step": 1230
+    },
+    {
+      "epoch": 2.5991097145849698,
+      "learning_rate": 0.0001265569700437975,
+      "loss": 4.117146682739258,
+      "step": 1240
+    },
+    {
+      "epoch": 2.6200576067033254,
+      "learning_rate": 0.00012757759076995718,
+      "loss": 4.102180480957031,
+      "step": 1250
+    },
+    {
+      "epoch": 2.641005498821681,
+      "learning_rate": 0.00012859821149611682,
+      "loss": 4.15636100769043,
+      "step": 1260
+    },
+    {
+      "epoch": 2.661953390940037,
+      "learning_rate": 0.0001296188322222765,
+      "loss": 4.069457626342773,
+      "step": 1270
+    },
+    {
+      "epoch": 2.682901283058392,
+      "learning_rate": 0.00013063945294843616,
+      "loss": 4.0670215606689455,
+      "step": 1280
+    },
+    {
+      "epoch": 2.7038491751767477,
+      "learning_rate": 0.0001316600736745958,
+      "loss": 3.985906219482422,
+      "step": 1290
+    },
+    {
+      "epoch": 2.7247970672951034,
+      "learning_rate": 0.00013268069440075547,
+      "loss": 4.272599792480468,
+      "step": 1300
+    },
+    {
+      "epoch": 2.745744959413459,
+      "learning_rate": 0.00013370131512691514,
+      "loss": 3.9709007263183596,
+      "step": 1310
+    },
+    {
+      "epoch": 2.7666928515318148,
+      "learning_rate": 0.00013472193585307478,
+      "loss": 4.1308135986328125,
+      "step": 1320
+    },
+    {
+      "epoch": 2.7876407436501704,
+      "learning_rate": 0.00013574255657923444,
+      "loss": 4.175233840942383,
+      "step": 1330
+    },
+    {
+      "epoch": 2.8085886357685257,
+      "learning_rate": 0.00013676317730539409,
+      "loss": 3.9187103271484376,
+      "step": 1340
+    },
+    {
+      "epoch": 2.8295365278868814,
+      "learning_rate": 0.00013778379803155375,
+      "loss": 3.8264163970947265,
+      "step": 1350
+    },
+    {
+      "epoch": 2.850484420005237,
+      "learning_rate": 0.0001388044187577134,
+      "loss": 4.124664306640625,
+      "step": 1360
+    },
+    {
+      "epoch": 2.8714323121235923,
+      "learning_rate": 0.00013982503948387306,
+      "loss": 3.952465057373047,
+      "step": 1370
+    },
+    {
+      "epoch": 2.892380204241948,
+      "learning_rate": 0.00014084566021003273,
+      "loss": 3.987406921386719,
+      "step": 1380
+    },
+    {
+      "epoch": 2.9133280963603037,
+      "learning_rate": 0.00014186628093619237,
+      "loss": 3.92406005859375,
+      "step": 1390
+    },
+    {
+      "epoch": 2.9342759884786593,
+      "learning_rate": 0.00014288690166235201,
+      "loss": 3.932135009765625,
+      "step": 1400
+    },
+    {
+      "epoch": 2.955223880597015,
+      "learning_rate": 0.00014390752238851168,
+      "loss": 4.134164428710937,
+      "step": 1410
+    },
+    {
+      "epoch": 2.9761717727153707,
+      "learning_rate": 0.00014492814311467135,
+      "loss": 4.045958709716797,
+      "step": 1420
+    },
+    {
+      "epoch": 2.9971196648337264,
+      "learning_rate": 0.000145948763840831,
+      "loss": 3.9462562561035157,
+      "step": 1430
+    },
+    {
+      "epoch": 3.01885310290652,
+      "learning_rate": 0.00014696938456699066,
+      "loss": 4.254601669311524,
+      "step": 1440
+    },
+    {
+      "epoch": 3.0398009950248754,
+      "learning_rate": 0.00014799000529315033,
+      "loss": 3.9297733306884766,
+      "step": 1450
+    },
+    {
+      "epoch": 3.060748887143231,
+      "learning_rate": 0.00014901062601930997,
+      "loss": 3.8294136047363283,
+      "step": 1460
+    },
+    {
+      "epoch": 3.081696779261587,
+      "learning_rate": 0.00015003124674546964,
+      "loss": 4.05113525390625,
+      "step": 1470
+    },
+    {
+      "epoch": 3.1026446713799425,
+      "learning_rate": 0.0001510518674716293,
+      "loss": 3.975564956665039,
+      "step": 1480
+    },
+    {
+      "epoch": 3.123592563498298,
+      "learning_rate": 0.00015207248819778895,
+      "loss": 3.9852630615234377,
+      "step": 1490
+    },
+    {
+      "epoch": 3.1445404556166534,
+      "learning_rate": 0.00015309310892394862,
+      "loss": 4.0287940979003904,
+      "step": 1500
+    },
+    {
+      "epoch": 3.165488347735009,
+      "learning_rate": 0.00015411372965010828,
+      "loss": 3.9631397247314455,
+      "step": 1510
+    },
+    {
+      "epoch": 3.1864362398533648,
+      "learning_rate": 0.00015513435037626793,
+      "loss": 3.9254386901855467,
+      "step": 1520
+    },
+    {
+      "epoch": 3.2073841319717205,
+      "learning_rate": 0.0001561549711024276,
+      "loss": 3.82324104309082,
+      "step": 1530
+    },
+    {
+      "epoch": 3.228332024090076,
+      "learning_rate": 0.00015717559182858723,
+      "loss": 3.9215030670166016,
+      "step": 1540
+    },
+    {
+      "epoch": 3.2492799162084314,
+      "learning_rate": 0.0001581962125547469,
+      "loss": 3.8164131164550783,
+      "step": 1550
+    },
+    {
+      "epoch": 3.270227808326787,
+      "learning_rate": 0.00015921683328090657,
+      "loss": 3.7764801025390624,
+      "step": 1560
+    },
+    {
+      "epoch": 3.2911757004451427,
+      "learning_rate": 0.0001602374540070662,
+      "loss": 3.9773059844970704,
+      "step": 1570
+    },
+    {
+      "epoch": 3.3121235925634984,
+      "learning_rate": 0.00016125807473322588,
+      "loss": 3.8427078247070314,
+      "step": 1580
+    },
+    {
+      "epoch": 3.333071484681854,
+      "learning_rate": 0.00016227869545938555,
+      "loss": 3.8103118896484376,
+      "step": 1590
+    },
+    {
+      "epoch": 3.3540193768002093,
+      "learning_rate": 0.00016329931618554516,
+      "loss": 3.8182399749755858,
+      "step": 1600
+    },
+    {
+      "epoch": 3.374967268918565,
+      "learning_rate": 0.00016431993691170483,
+      "loss": 3.750722122192383,
+      "step": 1610
+    },
+    {
+      "epoch": 3.3959151610369207,
+      "learning_rate": 0.0001653405576378645,
+      "loss": 3.9241371154785156,
+      "step": 1620
+    },
+    {
+      "epoch": 3.4168630531552764,
+      "learning_rate": 0.00016636117836402414,
+      "loss": 3.827128219604492,
+      "step": 1630
+    },
+    {
+      "epoch": 3.4378109452736316,
+      "learning_rate": 0.0001673817990901838,
+      "loss": 3.750970458984375,
+      "step": 1640
+    },
+    {
+      "epoch": 3.4587588373919873,
+      "learning_rate": 0.00016840241981634345,
+      "loss": 3.827236557006836,
+      "step": 1650
+    },
+    {
+      "epoch": 3.479706729510343,
+      "learning_rate": 0.00016942304054250312,
+      "loss": 3.862264633178711,
+      "step": 1660
+    },
+    {
+      "epoch": 3.5006546216286987,
+      "learning_rate": 0.0001704436612686628,
+      "loss": 3.9794548034667967,
+      "step": 1670
+    },
+    {
+      "epoch": 3.5216025137470544,
+      "learning_rate": 0.00017146428199482243,
+      "loss": 3.8642444610595703,
+      "step": 1680
+    },
+    {
+      "epoch": 3.54255040586541,
+      "learning_rate": 0.0001724849027209821,
+      "loss": 3.8710708618164062,
+      "step": 1690
+    },
+    {
+      "epoch": 3.5634982979837653,
+      "learning_rate": 0.00017350552344714176,
+      "loss": 3.7993534088134764,
+      "step": 1700
+    },
+    {
+      "epoch": 3.584446190102121,
+      "learning_rate": 0.0001745261441733014,
+      "loss": 3.8951980590820314,
+      "step": 1710
+    },
+    {
+      "epoch": 3.6053940822204766,
+      "learning_rate": 0.00017554676489946107,
+      "loss": 3.728221893310547,
+      "step": 1720
+    },
+    {
+      "epoch": 3.6263419743388323,
+      "learning_rate": 0.00017656738562562074,
+      "loss": 3.7405670166015623,
+      "step": 1730
+    },
+    {
+      "epoch": 3.6472898664571876,
+      "learning_rate": 0.00017758800635178038,
+      "loss": 3.7010948181152346,
+      "step": 1740
+    },
+    {
+      "epoch": 3.6682377585755432,
+      "learning_rate": 0.00017860862707794005,
+      "loss": 3.7746726989746096,
+      "step": 1750
+    },
+    {
+      "epoch": 3.689185650693899,
+      "learning_rate": 0.00017962924780409972,
+      "loss": 3.842145538330078,
+      "step": 1760
+    },
+    {
+      "epoch": 3.7101335428122546,
+      "learning_rate": 0.00018064986853025936,
+      "loss": 3.817458724975586,
+      "step": 1770
+    },
+    {
+      "epoch": 3.7310814349306103,
+      "learning_rate": 0.00018167048925641903,
+      "loss": 3.7765247344970705,
+      "step": 1780
+    },
+    {
+      "epoch": 3.752029327048966,
+      "learning_rate": 0.0001826911099825787,
+      "loss": 3.6870758056640627,
+      "step": 1790
+    },
+    {
+      "epoch": 3.772977219167321,
+      "learning_rate": 0.00018371173070873834,
+      "loss": 3.8823310852050783,
+      "step": 1800
+    },
+    {
+      "epoch": 3.793925111285677,
+      "learning_rate": 0.000184732351434898,
+      "loss": 3.7599208831787108,
+      "step": 1810
+    },
+    {
+      "epoch": 3.8148730034040326,
+      "learning_rate": 0.00018575297216105762,
+      "loss": 3.8205623626708984,
+      "step": 1820
+    },
+    {
+      "epoch": 3.835820895522388,
+      "learning_rate": 0.0001867735928872173,
+      "loss": 3.840204620361328,
+      "step": 1830
+    },
+    {
+      "epoch": 3.8567687876407435,
+      "learning_rate": 0.00018779421361337696,
+      "loss": 3.8769672393798826,
+      "step": 1840
+    },
+    {
+      "epoch": 3.877716679759099,
+      "learning_rate": 0.0001888148343395366,
+      "loss": 3.7292160034179687,
+      "step": 1850
+    },
+    {
+      "epoch": 3.898664571877455,
+      "learning_rate": 0.00018983545506569627,
+      "loss": 3.785298156738281,
+      "step": 1860
+    },
+    {
+      "epoch": 3.9196124639958105,
+      "learning_rate": 0.00019085607579185594,
+      "loss": 3.691172790527344,
+      "step": 1870
+    },
+    {
+      "epoch": 3.940560356114166,
+      "learning_rate": 0.00019187669651801558,
+      "loss": 3.7312793731689453,
+      "step": 1880
+    },
+    {
+      "epoch": 3.9615082482325215,
+      "learning_rate": 0.00019289731724417525,
+      "loss": 3.7419872283935547,
+      "step": 1890
+    },
+    {
+      "epoch": 3.982456140350877,
+      "learning_rate": 0.0001939179379703349,
+      "loss": 3.587678909301758,
+      "step": 1900
+    },
+    {
+      "epoch": 4.004189578423671,
+      "learning_rate": 0.00019493855869649455,
+      "loss": 3.9609317779541016,
+      "step": 1910
+    },
+    {
+      "epoch": 4.025137470542027,
+      "learning_rate": 0.00019595917942265422,
+      "loss": 3.7329071044921873,
+      "step": 1920
+    },
+    {
+      "epoch": 4.046085362660382,
+      "learning_rate": 0.0001969798001488139,
+      "loss": 3.7702545166015624,
+      "step": 1930
+    },
+    {
+      "epoch": 4.067033254778738,
+      "learning_rate": 0.00019800042087497353,
+      "loss": 3.796523666381836,
+      "step": 1940
+    },
+    {
+      "epoch": 4.087981146897094,
+      "learning_rate": 0.0001990210416011332,
+      "loss": 3.643301773071289,
+      "step": 1950
+    },
+    {
+      "epoch": 4.108929039015449,
+      "learning_rate": 0.00020004166232729287,
+      "loss": 3.705374526977539,
+      "step": 1960
+    },
+    {
+      "epoch": 4.129876931133805,
+      "learning_rate": 0.0002010622830534525,
+      "loss": 3.619226837158203,
+      "step": 1970
+    },
+    {
+      "epoch": 4.15082482325216,
+      "learning_rate": 0.00020208290377961218,
+      "loss": 3.862563705444336,
+      "step": 1980
+    },
+    {
+      "epoch": 4.1717727153705155,
+      "learning_rate": 0.00020310352450577185,
+      "loss": 3.6324195861816406,
+      "step": 1990
+    },
+    {
+      "epoch": 4.192720607488871,
+      "learning_rate": 0.0002041241452319315,
+      "loss": 3.674951171875,
+      "step": 2000
+    },
+    {
+      "epoch": 4.213668499607227,
+      "learning_rate": 0.00020514476595809116,
+      "loss": 3.59210205078125,
+      "step": 2010
+    },
+    {
+      "epoch": 4.234616391725583,
+      "learning_rate": 0.0002061653866842508,
+      "loss": 3.8746570587158202,
+      "step": 2020
+    },
+    {
+      "epoch": 4.255564283843938,
+      "learning_rate": 0.00020718600741041044,
+      "loss": 3.682146453857422,
+      "step": 2030
+    },
+    {
+      "epoch": 4.276512175962294,
+      "learning_rate": 0.0002082066281365701,
+      "loss": 3.8217212677001955,
+      "step": 2040
+    },
+    {
+      "epoch": 4.29746006808065,
+      "learning_rate": 0.00020922724886272975,
+      "loss": 3.691872787475586,
+      "step": 2050
+    },
+    {
+      "epoch": 4.318407960199005,
+      "learning_rate": 0.00021024786958888942,
+      "loss": 3.7160354614257813,
+      "step": 2060
+    },
+    {
+      "epoch": 4.339355852317361,
+      "learning_rate": 0.00021126849031504908,
+      "loss": 3.5512325286865236,
+      "step": 2070
+    },
+    {
+      "epoch": 4.360303744435716,
+      "learning_rate": 0.00021228911104120873,
+      "loss": 3.623905563354492,
+      "step": 2080
+    },
+    {
+      "epoch": 4.3812516365540715,
+      "learning_rate": 0.0002133097317673684,
+      "loss": 3.756671905517578,
+      "step": 2090
+    },
+    {
+      "epoch": 4.402199528672427,
+      "learning_rate": 0.00021433035249352806,
+      "loss": 3.71322021484375,
+      "step": 2100
+    },
+    {
+      "epoch": 4.423147420790783,
+      "learning_rate": 0.0002153509732196877,
+      "loss": 3.588302993774414,
+      "step": 2110
+    },
+    {
+      "epoch": 4.4440953129091385,
+      "learning_rate": 0.00021637159394584737,
+      "loss": 3.666096496582031,
+      "step": 2120
+    },
+    {
+      "epoch": 4.465043205027494,
+      "learning_rate": 0.00021739221467200704,
+      "loss": 3.660139465332031,
+      "step": 2130
+    },
+    {
+      "epoch": 4.48599109714585,
+      "learning_rate": 0.00021841283539816668,
+      "loss": 3.5749874114990234,
+      "step": 2140
+    },
+    {
+      "epoch": 4.5069389892642056,
+      "learning_rate": 0.00021943345612432635,
+      "loss": 3.621977615356445,
+      "step": 2150
+    },
+    {
+      "epoch": 4.527886881382561,
+      "learning_rate": 0.000220454076850486,
+      "loss": 3.7179306030273436,
+      "step": 2160
+    },
+    {
+      "epoch": 4.548834773500916,
+      "learning_rate": 0.00022147469757664566,
+      "loss": 3.5558433532714844,
+      "step": 2170
+    },
+    {
+      "epoch": 4.569782665619272,
+      "learning_rate": 0.00022249531830280533,
+      "loss": 3.754520034790039,
+      "step": 2180
+    },
+    {
+      "epoch": 4.590730557737627,
+      "learning_rate": 0.00022351593902896497,
+      "loss": 3.734426498413086,
+      "step": 2190
+    },
+    {
+      "epoch": 4.611678449855983,
+      "learning_rate": 0.00022453655975512464,
+      "loss": 3.6508132934570314,
+      "step": 2200
+    },
+    {
+      "epoch": 4.632626341974339,
+      "learning_rate": 0.0002255571804812843,
+      "loss": 3.7082672119140625,
+      "step": 2210
+    },
+    {
+      "epoch": 4.653574234092694,
+      "learning_rate": 0.00022657780120744395,
+      "loss": 3.600681686401367,
+      "step": 2220
+    },
+    {
+      "epoch": 4.67452212621105,
+      "learning_rate": 0.00022759842193360361,
+      "loss": 3.6909461975097657,
+      "step": 2230
+    },
+    {
+      "epoch": 4.695470018329406,
+      "learning_rate": 0.00022861904265976328,
+      "loss": 3.557560348510742,
+      "step": 2240
+    },
+    {
+      "epoch": 4.7164179104477615,
+      "learning_rate": 0.0002296396633859229,
+      "loss": 3.6415851593017576,
+      "step": 2250
+    },
+    {
+      "epoch": 4.737365802566117,
+      "learning_rate": 0.00023066028411208256,
+      "loss": 3.6133026123046874,
+      "step": 2260
+    },
+    {
+      "epoch": 4.758313694684473,
+      "learning_rate": 0.00023168090483824223,
+      "loss": 3.5348537445068358,
+      "step": 2270
+    },
+    {
+      "epoch": 4.779261586802828,
+      "learning_rate": 0.00023270152556440187,
+      "loss": 3.6589839935302733,
+      "step": 2280
+    },
+    {
+      "epoch": 4.800209478921183,
+      "learning_rate": 0.00023372214629056154,
+      "loss": 3.5354270935058594,
+      "step": 2290
+    },
+    {
+      "epoch": 4.821157371039539,
+      "learning_rate": 0.00023474276701672118,
+      "loss": 3.536578369140625,
+      "step": 2300
+    },
+    {
+      "epoch": 4.842105263157895,
+      "learning_rate": 0.00023576338774288085,
+      "loss": 3.5566326141357423,
+      "step": 2310
+    },
+    {
+      "epoch": 4.86305315527625,
+      "learning_rate": 0.00023678400846904052,
+      "loss": 3.475338363647461,
+      "step": 2320
+    },
+    {
+      "epoch": 4.884001047394606,
+      "learning_rate": 0.00023780462919520016,
+      "loss": 3.6042369842529296,
+      "step": 2330
+    },
+    {
+      "epoch": 4.904948939512962,
+      "learning_rate": 0.00023882524992135983,
+      "loss": 3.621173095703125,
+      "step": 2340
+    },
+    {
+      "epoch": 4.925896831631317,
+      "learning_rate": 0.0002398458706475195,
+      "loss": 3.640410232543945,
+      "step": 2350
+    },
+    {
+      "epoch": 4.946844723749672,
+      "learning_rate": 0.00024086649137367914,
+      "loss": 3.488922882080078,
+      "step": 2360
+    },
+    {
+      "epoch": 4.967792615868028,
+      "learning_rate": 0.0002418871120998388,
+      "loss": 3.4922332763671875,
+      "step": 2370
+    },
+    {
+      "epoch": 4.988740507986384,
+      "learning_rate": 0.00024290773282599848,
+      "loss": 3.590007019042969,
+      "step": 2380
+    },
+    {
+      "epoch": 5.010473946059178,
+      "learning_rate": 0.00024392835355215812,
+      "loss": 3.6829368591308596,
+      "step": 2390
+    },
+    {
+      "epoch": 5.0314218381775335,
+      "learning_rate": 0.0002449489742783178,
+      "loss": 3.651840591430664,
+      "step": 2400
+    },
+    {
+      "epoch": 5.052369730295889,
+      "learning_rate": 0.00024596959500447745,
+      "loss": 3.7254043579101563,
+      "step": 2410
+    },
+    {
+      "epoch": 5.073317622414245,
+      "learning_rate": 0.0002469902157306371,
+      "loss": 3.594907760620117,
+      "step": 2420
+    },
+    {
+      "epoch": 5.094265514532601,
+      "learning_rate": 0.00024801083645679674,
+      "loss": 3.4854148864746093,
+      "step": 2430
+    },
+    {
+      "epoch": 5.115213406650955,
+      "learning_rate": 0.0002490314571829564,
+      "loss": 3.58482666015625,
+      "step": 2440
+    },
+    {
+      "epoch": 5.136161298769311,
+      "learning_rate": 0.00025005207790911607,
+      "loss": 3.5847278594970704,
+      "step": 2450
+    },
+    {
+      "epoch": 5.157109190887667,
+      "learning_rate": 0.0002510726986352757,
+      "loss": 3.5045509338378906,
+      "step": 2460
+    },
+    {
+      "epoch": 5.178057083006022,
+      "learning_rate": 0.00025209331936143535,
+      "loss": 3.6352733612060546,
+      "step": 2470
+    },
+    {
+      "epoch": 5.199004975124378,
+      "learning_rate": 0.000253113940087595,
+      "loss": 3.555766296386719,
+      "step": 2480
+    },
+    {
+      "epoch": 5.219952867242734,
+      "learning_rate": 0.0002541345608137547,
+      "loss": 3.541688919067383,
+      "step": 2490
+    },
+    {
+      "epoch": 5.2409007593610895,
+      "learning_rate": 0.00025515518153991436,
+      "loss": 3.5400638580322266,
+      "step": 2500
+    },
+    {
+      "epoch": 5.261848651479445,
+      "learning_rate": 0.00025617580226607403,
+      "loss": 3.4929561614990234,
+      "step": 2510
+    },
+    {
+      "epoch": 5.282796543597801,
+      "learning_rate": 0.00025719642299223364,
+      "loss": 3.5611968994140626,
+      "step": 2520
+    },
+    {
+      "epoch": 5.3037444357161565,
+      "learning_rate": 0.0002582170437183933,
+      "loss": 3.4528472900390623,
+      "step": 2530
+    },
+    {
+      "epoch": 5.324692327834511,
+      "learning_rate": 0.000259237664444553,
+      "loss": 3.474958801269531,
+      "step": 2540
+    },
+    {
+      "epoch": 5.345640219952867,
+      "learning_rate": 0.00026025828517071265,
+      "loss": 3.6800113677978517,
+      "step": 2550
+    },
+    {
+      "epoch": 5.366588112071223,
+      "learning_rate": 0.0002612789058968723,
+      "loss": 3.524998092651367,
+      "step": 2560
+    },
+    {
+      "epoch": 5.387536004189578,
+      "learning_rate": 0.00026229952662303193,
+      "loss": 3.3865074157714843,
+      "step": 2570
+    },
+    {
+      "epoch": 5.408483896307934,
+      "learning_rate": 0.0002633201473491916,
+      "loss": 3.416782760620117,
+      "step": 2580
+    },
+    {
+      "epoch": 5.42943178842629,
+      "learning_rate": 0.00026434076807535127,
+      "loss": 3.4291786193847655,
+      "step": 2590
+    },
+    {
+      "epoch": 5.450379680544645,
+      "learning_rate": 0.00026536138880151093,
+      "loss": 3.487574005126953,
+      "step": 2600
+    },
+    {
+      "epoch": 5.471327572663001,
+      "learning_rate": 0.0002663820095276706,
+      "loss": 3.5091732025146483,
+      "step": 2610
+    },
+    {
+      "epoch": 5.492275464781357,
+      "learning_rate": 0.00026740263025383027,
+      "loss": 4.268975448608399,
+      "step": 2620
+    },
+    {
+      "epoch": 5.5132233568997115,
+      "learning_rate": 0.0002684232509799899,
+      "loss": 3.732823944091797,
+      "step": 2630
+    },
+    {
+      "epoch": 5.534171249018067,
+      "learning_rate": 0.00026944387170614955,
+      "loss": 3.650152587890625,
+      "step": 2640
+    },
+    {
+      "epoch": 5.555119141136423,
+      "learning_rate": 0.0002704644924323092,
+      "loss": 3.6992671966552733,
+      "step": 2650
+    },
+    {
+      "epoch": 5.576067033254779,
+      "learning_rate": 0.0002714851131584689,
+      "loss": 3.5627864837646483,
+      "step": 2660
+    },
+    {
+      "epoch": 5.597014925373134,
+      "learning_rate": 0.0002725057338846285,
+      "loss": 3.5473575592041016,
+      "step": 2670
+    },
+    {
+      "epoch": 5.61796281749149,
+      "learning_rate": 0.00027352635461078817,
+      "loss": 3.7565258026123045,
+      "step": 2680
+    },
+    {
+      "epoch": 5.638910709609846,
+      "learning_rate": 0.00027454697533694784,
+      "loss": 3.4811996459960937,
+      "step": 2690
+    },
+    {
+      "epoch": 5.659858601728201,
+      "learning_rate": 0.0002755675960631075,
+      "loss": 3.5412361145019533,
+      "step": 2700
+    },
+    {
+      "epoch": 5.680806493846557,
+      "learning_rate": 0.0002765882167892671,
+      "loss": 3.6894275665283205,
+      "step": 2710
+    },
+    {
+      "epoch": 5.701754385964913,
+      "learning_rate": 0.0002776088375154268,
+      "loss": 3.532870864868164,
+      "step": 2720
+    },
+    {
+      "epoch": 5.722702278083268,
+      "learning_rate": 0.00027862945824158646,
+      "loss": 3.4718368530273436,
+      "step": 2730
+    },
+    {
+      "epoch": 5.743650170201623,
+      "learning_rate": 0.00027965007896774613,
+      "loss": 3.5320533752441405,
+      "step": 2740
+    },
+    {
+      "epoch": 5.764598062319979,
+      "learning_rate": 0.0002806706996939058,
+      "loss": 3.6331645965576174,
+      "step": 2750
+    },
+    {
+      "epoch": 5.7855459544383345,
+      "learning_rate": 0.00028169132042006546,
+      "loss": 3.50958137512207,
+      "step": 2760
+    },
+    {
+      "epoch": 5.80649384655669,
+      "learning_rate": 0.00028271194114622513,
+      "loss": 3.5480377197265627,
+      "step": 2770
+    },
+    {
+      "epoch": 5.827441738675046,
+      "learning_rate": 0.00028373256187238475,
+      "loss": 3.5677505493164063,
+      "step": 2780
+    },
+    {
+      "epoch": 5.848389630793402,
+      "learning_rate": 0.0002847531825985444,
+      "loss": 3.5496990203857424,
+      "step": 2790
+    },
+    {
+      "epoch": 5.869337522911757,
+      "learning_rate": 0.00028577380332470403,
+      "loss": 3.4990489959716795,
+      "step": 2800
+    },
+    {
+      "epoch": 5.890285415030113,
+      "learning_rate": 0.0002867944240508637,
+      "loss": 3.4764991760253907,
+      "step": 2810
+    },
+    {
+      "epoch": 5.911233307148468,
+      "learning_rate": 0.00028781504477702337,
+      "loss": 3.4782173156738283,
+      "step": 2820
+    },
+    {
+      "epoch": 5.932181199266823,
+      "learning_rate": 0.00028883566550318303,
+      "loss": 3.5007530212402345,
+      "step": 2830
+    },
+    {
+      "epoch": 5.953129091385179,
+      "learning_rate": 0.0002898562862293427,
+      "loss": 3.4596179962158202,
+      "step": 2840
+    },
+    {
+      "epoch": 5.974076983503535,
+      "learning_rate": 0.0002908769069555023,
+      "loss": 3.4080764770507814,
+      "step": 2850
+    },
+    {
+      "epoch": 5.9950248756218905,
+      "learning_rate": 0.000291897527681662,
+      "loss": 3.5865558624267577,
+      "step": 2860
+    },
+    {
+      "epoch": 6.016758313694685,
+      "learning_rate": 0.00029291814840782165,
+      "loss": 3.539356231689453,
+      "step": 2870
+    },
+    {
+      "epoch": 6.03770620581304,
+      "learning_rate": 0.0002939387691339813,
+      "loss": 3.5746910095214846,
+      "step": 2880
+    },
+    {
+      "epoch": 6.058654097931396,
+      "learning_rate": 0.000294959389860141,
+      "loss": 3.477669906616211,
+      "step": 2890
+    },
+    {
+      "epoch": 6.079601990049751,
+      "learning_rate": 0.00029598001058630066,
+      "loss": 3.385912322998047,
+      "step": 2900
+    },
+    {
+      "epoch": 6.100549882168107,
+      "learning_rate": 0.00029700063131246027,
+      "loss": 3.566743850708008,
+      "step": 2910
+    },
+    {
+      "epoch": 6.121497774286462,
+      "learning_rate": 0.00029802125203861994,
+      "loss": 3.4617984771728514,
+      "step": 2920
+    },
+    {
+      "epoch": 6.142445666404818,
+      "learning_rate": 0.0002990418727647796,
+      "loss": 3.46124382019043,
+      "step": 2930
+    },
+    {
+      "epoch": 6.163393558523174,
+      "learning_rate": 0.0003000624934909393,
+      "loss": 3.5851741790771485,
+      "step": 2940
+    },
+    {
+      "epoch": 6.184341450641529,
+      "learning_rate": 0.00030108311421709894,
+      "loss": 3.5333206176757814,
+      "step": 2950
+    },
+    {
+      "epoch": 6.205289342759885,
+      "learning_rate": 0.0003021037349432586,
+      "loss": 3.4518871307373047,
+      "step": 2960
+    },
+    {
+      "epoch": 6.226237234878241,
+      "learning_rate": 0.0003031243556694182,
+      "loss": 3.5048519134521485,
+      "step": 2970
+    },
+    {
+      "epoch": 6.247185126996596,
+      "learning_rate": 0.0003041449763955779,
+      "loss": 3.386810302734375,
+      "step": 2980
+    },
+    {
+      "epoch": 6.268133019114952,
+      "learning_rate": 0.00030516559712173756,
+      "loss": 3.3472484588623046,
+      "step": 2990
+    },
+    {
+      "epoch": 6.289080911233307,
+      "learning_rate": 0.00030618621784789723,
+      "loss": 3.578289794921875,
+      "step": 3000
+    },
+    {
+      "epoch": 6.3100288033516625,
+      "learning_rate": 0.0003072068385740569,
+      "loss": 3.4641948699951173,
+      "step": 3010
+    },
+    {
+      "epoch": 6.330976695470018,
+      "learning_rate": 0.00030822745930021657,
+      "loss": 3.42608757019043,
+      "step": 3020
+    },
+    {
+      "epoch": 6.351924587588374,
+      "learning_rate": 0.0003092480800263762,
+      "loss": 3.4154186248779297,
+      "step": 3030
+    },
+    {
+      "epoch": 6.3728724797067295,
+      "learning_rate": 0.00031026870075253585,
+      "loss": 3.5270923614501952,
+      "step": 3040
+    },
+    {
+      "epoch": 6.393820371825085,
+      "learning_rate": 0.0003112893214786955,
+      "loss": 3.4780391693115233,
+      "step": 3050
+    },
+    {
+      "epoch": 6.414768263943441,
+      "learning_rate": 0.0003123099422048552,
+      "loss": 3.379390335083008,
+      "step": 3060
+    },
+    {
+      "epoch": 6.435716156061797,
+      "learning_rate": 0.00031333056293101486,
+      "loss": 3.405352020263672,
+      "step": 3070
+    },
+    {
+      "epoch": 6.456664048180152,
+      "learning_rate": 0.00031435118365717447,
+      "loss": 3.523044204711914,
+      "step": 3080
+    },
+    {
+      "epoch": 6.477611940298507,
+      "learning_rate": 0.00031537180438333414,
+      "loss": 3.488709259033203,
+      "step": 3090
+    },
+    {
+      "epoch": 6.498559832416863,
+      "learning_rate": 0.0003163924251094938,
+      "loss": 3.3736576080322265,
+      "step": 3100
+    },
+    {
+      "epoch": 6.519507724535218,
+      "learning_rate": 0.0003174130458356535,
+      "loss": 3.5457527160644533,
+      "step": 3110
+    },
+    {
+      "epoch": 6.540455616653574,
+      "learning_rate": 0.00031843366656181314,
+      "loss": 3.38338623046875,
+      "step": 3120
+    },
+    {
+      "epoch": 6.56140350877193,
+      "learning_rate": 0.0003194542872879728,
+      "loss": 3.4261444091796873,
+      "step": 3130
+    },
+    {
+      "epoch": 6.5823514008902855,
+      "learning_rate": 0.0003204749080141324,
+      "loss": 3.365514373779297,
+      "step": 3140
+    },
+    {
+      "epoch": 6.603299293008641,
+      "learning_rate": 0.0003214955287402921,
+      "loss": 3.3958808898925783,
+      "step": 3150
+    },
+    {
+      "epoch": 6.624247185126997,
+      "learning_rate": 0.00032251614946645176,
+      "loss": 3.4493579864501953,
+      "step": 3160
+    },
+    {
+      "epoch": 6.6451950772453525,
+      "learning_rate": 0.00032353677019261143,
+      "loss": 3.509814453125,
+      "step": 3170
+    },
+    {
+      "epoch": 6.666142969363708,
+      "learning_rate": 0.0003245573909187711,
+      "loss": 3.3270954132080077,
+      "step": 3180
+    },
+    {
+      "epoch": 6.687090861482063,
+      "learning_rate": 0.00032557801164493077,
+      "loss": 3.407461929321289,
+      "step": 3190
+    },
+    {
+      "epoch": 6.708038753600419,
+      "learning_rate": 0.0003265986323710903,
+      "loss": 3.3440326690673827,
+      "step": 3200
+    },
+    {
+      "epoch": 6.728986645718774,
+      "learning_rate": 0.00032761925309725,
+      "loss": 3.3119239807128906,
+      "step": 3210
+    },
+    {
+      "epoch": 6.74993453783713,
+      "learning_rate": 0.00032863987382340966,
+      "loss": 3.441579818725586,
+      "step": 3220
+    },
+    {
+      "epoch": 6.770882429955486,
+      "learning_rate": 0.00032966049454956933,
+      "loss": 3.448055648803711,
+      "step": 3230
+    },
+    {
+      "epoch": 6.791830322073841,
+      "learning_rate": 0.000330681115275729,
+      "loss": 3.2744544982910155,
+      "step": 3240
+    },
+    {
+      "epoch": 6.812778214192197,
+      "learning_rate": 0.0003317017360018886,
+      "loss": 3.4412532806396485,
+      "step": 3250
+    },
+    {
+      "epoch": 6.833726106310553,
+      "learning_rate": 0.0003327223567280483,
+      "loss": 3.459817123413086,
+      "step": 3260
+    },
+    {
+      "epoch": 6.8546739984289085,
+      "learning_rate": 0.00033374297745420795,
+      "loss": 3.3376232147216798,
+      "step": 3270
+    },
+    {
+      "epoch": 6.875621890547263,
+      "learning_rate": 0.0003347635981803676,
+      "loss": 3.557674789428711,
+      "step": 3280
+    },
+    {
+      "epoch": 6.896569782665619,
+      "learning_rate": 0.0003357842189065273,
+      "loss": 3.290896987915039,
+      "step": 3290
+    },
+    {
+      "epoch": 6.917517674783975,
+      "learning_rate": 0.0003368048396326869,
+      "loss": 3.49056396484375,
+      "step": 3300
+    },
+    {
+      "epoch": 6.93846556690233,
+      "learning_rate": 0.00033782546035884657,
+      "loss": 3.3487998962402346,
+      "step": 3310
+    },
+    {
+      "epoch": 6.959413459020686,
+      "learning_rate": 0.00033884608108500624,
+      "loss": 3.3251983642578127,
+      "step": 3320
+    },
+    {
+      "epoch": 6.980361351139042,
+      "learning_rate": 0.0003398667018111659,
+      "loss": 3.463846206665039,
+      "step": 3330
+    },
+    {
+      "epoch": 7.002094789211836,
+      "learning_rate": 0.0003408873225373256,
+      "loss": 3.5404449462890626,
+      "step": 3340
+    },
+    {
+      "epoch": 7.023042681330191,
+      "learning_rate": 0.00034190794326348524,
+      "loss": 3.317121887207031,
+      "step": 3350
+    },
+    {
+      "epoch": 7.043990573448546,
+      "learning_rate": 0.00034292856398964486,
+      "loss": 3.446538543701172,
+      "step": 3360
+    },
+    {
+      "epoch": 7.064938465566902,
+      "learning_rate": 0.0003439491847158045,
+      "loss": 3.410959243774414,
+      "step": 3370
+    },
+    {
+      "epoch": 7.085886357685258,
+      "learning_rate": 0.0003449698054419642,
+      "loss": 3.4593124389648438,
+      "step": 3380
+    },
+    {
+      "epoch": 7.1068342498036134,
+      "learning_rate": 0.00034599042616812386,
+      "loss": 3.4331336975097657,
+      "step": 3390
+    },
+    {
+      "epoch": 7.127782141921969,
+      "learning_rate": 0.00034701104689428353,
+      "loss": 3.415497970581055,
+      "step": 3400
+    },
+    {
+      "epoch": 7.148730034040325,
+      "learning_rate": 0.0003480316676204432,
+      "loss": 3.358320617675781,
+      "step": 3410
+    },
+    {
+      "epoch": 7.1696779261586805,
+      "learning_rate": 0.0003490522883466028,
+      "loss": 3.369782257080078,
+      "step": 3420
+    },
+    {
+      "epoch": 7.190625818277036,
+      "learning_rate": 0.0003500729090727625,
+      "loss": 3.3603092193603517,
+      "step": 3430
+    },
+    {
+      "epoch": 7.211573710395392,
+      "learning_rate": 0.00035109352979892215,
+      "loss": 3.4065528869628907,
+      "step": 3440
+    },
+    {
+      "epoch": 7.232521602513747,
+      "learning_rate": 0.0003521141505250818,
+      "loss": 3.368368148803711,
+      "step": 3450
+    },
+    {
+      "epoch": 7.253469494632102,
+      "learning_rate": 0.0003531347712512415,
+      "loss": 3.3687610626220703,
+      "step": 3460
+    },
+    {
+      "epoch": 7.274417386750458,
+      "learning_rate": 0.00035415539197740115,
+      "loss": 3.397439956665039,
+      "step": 3470
+    },
+    {
+      "epoch": 7.295365278868814,
+      "learning_rate": 0.00035517601270356077,
+      "loss": 3.329518508911133,
+      "step": 3480
+    },
+    {
+      "epoch": 7.316313170987169,
+      "learning_rate": 0.00035619663342972044,
+      "loss": 3.3606395721435547,
+      "step": 3490
+    },
+    {
+      "epoch": 7.337261063105525,
+      "learning_rate": 0.0003572172541558801,
+      "loss": 3.373159408569336,
+      "step": 3500
+    },
+    {
+      "epoch": 7.358208955223881,
+      "learning_rate": 0.00035823787488203977,
+      "loss": 3.460713195800781,
+      "step": 3510
+    },
+    {
+      "epoch": 7.379156847342236,
+      "learning_rate": 0.00035925849560819944,
+      "loss": 3.3600276947021483,
+      "step": 3520
+    },
+    {
+      "epoch": 7.400104739460592,
+      "learning_rate": 0.0003602791163343591,
+      "loss": 3.3381488800048826,
+      "step": 3530
+    },
+    {
+      "epoch": 7.421052631578947,
+      "learning_rate": 0.0003612997370605187,
+      "loss": 3.5098854064941407,
+      "step": 3540
+    },
+    {
+      "epoch": 7.442000523697303,
+      "learning_rate": 0.0003623203577866784,
+      "loss": 3.358294677734375,
+      "step": 3550
+    },
+    {
+      "epoch": 7.462948415815658,
+      "learning_rate": 0.00036334097851283806,
+      "loss": 3.361553955078125,
+      "step": 3560
+    },
+    {
+      "epoch": 7.483896307934014,
+      "learning_rate": 0.00036436159923899773,
+      "loss": 3.3908660888671873,
+      "step": 3570
+    },
+    {
+      "epoch": 7.50484420005237,
+      "learning_rate": 0.0003653822199651574,
+      "loss": 3.305834197998047,
+      "step": 3580
+    },
+    {
+      "epoch": 7.525792092170725,
+      "learning_rate": 0.000366402840691317,
+      "loss": 3.3551094055175783,
+      "step": 3590
+    },
+    {
+      "epoch": 7.546739984289081,
+      "learning_rate": 0.0003674234614174767,
+      "loss": 3.336803436279297,
+      "step": 3600
+    },
+    {
+      "epoch": 7.567687876407437,
+      "learning_rate": 0.00036844408214363635,
+      "loss": 3.402811050415039,
+      "step": 3610
+    },
+    {
+      "epoch": 7.588635768525792,
+      "learning_rate": 0.000369464702869796,
+      "loss": 3.3823123931884767,
+      "step": 3620
+    },
+    {
+      "epoch": 7.609583660644148,
+      "learning_rate": 0.00037048532359595563,
+      "loss": 3.3616653442382813,
+      "step": 3630
+    },
+    {
+      "epoch": 7.630531552762504,
+      "learning_rate": 0.00037150594432211524,
+      "loss": 3.409253692626953,
+      "step": 3640
+    },
+    {
+      "epoch": 7.6514794448808585,
+      "learning_rate": 0.0003725265650482749,
+      "loss": 3.1992008209228517,
+      "step": 3650
+    },
+    {
+      "epoch": 7.672427336999214,
+      "learning_rate": 0.0003735471857744346,
+      "loss": 3.2639488220214843,
+      "step": 3660
+    },
+    {
+      "epoch": 7.69337522911757,
+      "learning_rate": 0.00037456780650059425,
+      "loss": 3.4038814544677733,
+      "step": 3670
+    },
+    {
+      "epoch": 7.714323121235926,
+      "learning_rate": 0.0003755884272267539,
+      "loss": 3.382715606689453,
+      "step": 3680
+    },
+    {
+      "epoch": 7.735271013354281,
+      "learning_rate": 0.0003766090479529136,
+      "loss": 3.380691146850586,
+      "step": 3690
+    },
+    {
+      "epoch": 7.756218905472637,
+      "learning_rate": 0.0003776296686790732,
+      "loss": 3.3293548583984376,
+      "step": 3700
+    },
+    {
+      "epoch": 7.777166797590993,
+      "learning_rate": 0.00037865028940523287,
+      "loss": 3.2755306243896483,
+      "step": 3710
+    },
+    {
+      "epoch": 7.798114689709348,
+      "learning_rate": 0.00037967091013139253,
+      "loss": 3.3747108459472654,
+      "step": 3720
+    },
+    {
+      "epoch": 7.819062581827704,
+      "learning_rate": 0.0003806915308575522,
+      "loss": 3.325161361694336,
+      "step": 3730
+    },
+    {
+      "epoch": 7.840010473946059,
+      "learning_rate": 0.00038171215158371187,
+      "loss": 3.3385120391845704,
+      "step": 3740
+    },
+    {
+      "epoch": 7.8609583660644144,
+      "learning_rate": 0.00038273277230987154,
+      "loss": 3.365946960449219,
+      "step": 3750
+    },
+    {
+      "epoch": 7.88190625818277,
+      "learning_rate": 0.00038375339303603115,
+      "loss": 3.2765518188476563,
+      "step": 3760
+    },
+    {
+      "epoch": 7.902854150301126,
+      "learning_rate": 0.0003847740137621908,
+      "loss": 3.4135189056396484,
+      "step": 3770
+    },
+    {
+      "epoch": 7.9238020424194815,
+      "learning_rate": 0.0003857946344883505,
+      "loss": 3.3337100982666015,
+      "step": 3780
+    },
+    {
+      "epoch": 7.944749934537837,
+      "learning_rate": 0.00038681525521451016,
+      "loss": 3.469867706298828,
+      "step": 3790
+    },
+    {
+      "epoch": 7.965697826656193,
+      "learning_rate": 0.0003878358759406698,
+      "loss": 3.4340835571289063,
+      "step": 3800
+    },
+    {
+      "epoch": 7.9866457187745485,
+      "learning_rate": 0.00038885649666682944,
+      "loss": 3.3163192749023436,
+      "step": 3810
+    },
+    {
+      "epoch": 8.008379156847342,
+      "learning_rate": 0.0003898771173929891,
+      "loss": 3.4269264221191404,
+      "step": 3820
+    },
+    {
+      "epoch": 8.029327048965698,
+      "learning_rate": 0.0003908977381191488,
+      "loss": 3.366756057739258,
+      "step": 3830
+    },
+    {
+      "epoch": 8.050274941084053,
+      "learning_rate": 0.00039191835884530845,
+      "loss": 3.5194732666015627,
+      "step": 3840
+    },
+    {
+      "epoch": 8.071222833202409,
+      "learning_rate": 0.0003929389795714681,
+      "loss": 3.2884559631347656,
+      "step": 3850
+    },
+    {
+      "epoch": 8.092170725320765,
+      "learning_rate": 0.0003939596002976278,
+      "loss": 3.3115074157714846,
+      "step": 3860
+    },
+    {
+      "epoch": 8.11311861743912,
+      "learning_rate": 0.0003949802210237874,
+      "loss": 3.342890167236328,
+      "step": 3870
+    },
+    {
+      "epoch": 8.134066509557476,
+      "learning_rate": 0.00039600084174994706,
+      "loss": 3.337261962890625,
+      "step": 3880
+    },
+    {
+      "epoch": 8.155014401675832,
+      "learning_rate": 0.00039702146247610673,
+      "loss": 3.3896888732910155,
+      "step": 3890
+    },
+    {
+      "epoch": 8.175962293794187,
+      "learning_rate": 0.0003980420832022664,
+      "loss": 3.314004898071289,
+      "step": 3900
+    },
+    {
+      "epoch": 8.196910185912543,
+      "learning_rate": 0.00039906270392842607,
+      "loss": 3.253472900390625,
+      "step": 3910
+    },
+    {
+      "epoch": 8.217858078030899,
+      "learning_rate": 0.00040008332465458574,
+      "loss": 3.2534595489501954,
+      "step": 3920
+    },
+    {
+      "epoch": 8.238805970149254,
+      "learning_rate": 0.00040110394538074535,
+      "loss": 3.3420372009277344,
+      "step": 3930
+    },
+    {
+      "epoch": 8.25975386226761,
+      "learning_rate": 0.000402124566106905,
+      "loss": 3.3110313415527344,
+      "step": 3940
+    },
+    {
+      "epoch": 8.280701754385966,
+      "learning_rate": 0.0004031451868330647,
+      "loss": 3.3015865325927733,
+      "step": 3950
+    },
+    {
+      "epoch": 8.30164964650432,
+      "learning_rate": 0.00040416580755922436,
+      "loss": 3.207544708251953,
+      "step": 3960
+    },
+    {
+      "epoch": 8.322597538622675,
+      "learning_rate": 0.000405186428285384,
+      "loss": 3.2784183502197264,
+      "step": 3970
+    },
+    {
+      "epoch": 8.343545430741031,
+      "learning_rate": 0.0004062070490115437,
+      "loss": 3.29312744140625,
+      "step": 3980
+    },
+    {
+      "epoch": 8.364493322859387,
+      "learning_rate": 0.0004072276697377033,
+      "loss": 3.349509048461914,
+      "step": 3990
+    },
+    {
+      "epoch": 8.385441214977742,
+      "learning_rate": 0.000408248290463863,
+      "loss": 3.311314010620117,
+      "step": 4000
+    },
+    {
+      "epoch": 8.406389107096098,
+      "learning_rate": 0.00040926891119002264,
+      "loss": 3.2704097747802736,
+      "step": 4010
+    },
+    {
+      "epoch": 8.427336999214454,
+      "learning_rate": 0.0004102895319161823,
+      "loss": 3.3431529998779297,
+      "step": 4020
+    },
+    {
+      "epoch": 8.44828489133281,
+      "learning_rate": 0.000411310152642342,
+      "loss": 3.2444534301757812,
+      "step": 4030
+    },
+    {
+      "epoch": 8.469232783451165,
+      "learning_rate": 0.0004123307733685016,
+      "loss": 3.2499820709228517,
+      "step": 4040
+    },
+    {
+      "epoch": 8.49018067556952,
+      "learning_rate": 0.00041335139409466126,
+      "loss": 3.4333778381347657,
+      "step": 4050
+    },
+    {
+      "epoch": 8.511128567687877,
+      "learning_rate": 0.0004143720148208209,
+      "loss": 3.598118209838867,
+      "step": 4060
+    },
+    {
+      "epoch": 8.532076459806232,
+      "learning_rate": 0.00041539263554698055,
+      "loss": 3.1703567504882812,
+      "step": 4070
+    },
+    {
+      "epoch": 8.553024351924588,
+      "learning_rate": 0.0004164132562731402,
+      "loss": 3.308456802368164,
+      "step": 4080
+    },
+    {
+      "epoch": 8.573972244042944,
+      "learning_rate": 0.00041743387699929983,
+      "loss": 3.262325668334961,
+      "step": 4090
+    },
+    {
+      "epoch": 8.5949201361613,
+      "learning_rate": 0.0004184544977254595,
+      "loss": 3.277301788330078,
+      "step": 4100
+    },
+    {
+      "epoch": 8.615868028279655,
+      "learning_rate": 0.00041947511845161916,
+      "loss": 3.245453643798828,
+      "step": 4110
+    },
+    {
+      "epoch": 8.63681592039801,
+      "learning_rate": 0.00042049573917777883,
+      "loss": 3.246595764160156,
+      "step": 4120
+    },
+    {
+      "epoch": 8.657763812516366,
+      "learning_rate": 0.0004215163599039385,
+      "loss": 3.310033416748047,
+      "step": 4130
+    },
+    {
+      "epoch": 8.678711704634722,
+      "learning_rate": 0.00042253698063009817,
+      "loss": 3.301158905029297,
+      "step": 4140
+    },
+    {
+      "epoch": 8.699659596753076,
+      "learning_rate": 0.0004235576013562578,
+      "loss": 3.2295578002929686,
+      "step": 4150
+    },
+    {
+      "epoch": 8.720607488871432,
+      "learning_rate": 0.00042457822208241745,
+      "loss": 3.26501579284668,
+      "step": 4160
+    },
+    {
+      "epoch": 8.741555380989787,
+      "learning_rate": 0.0004255988428085771,
+      "loss": 3.346723937988281,
+      "step": 4170
+    },
+    {
+      "epoch": 8.762503273108143,
+      "learning_rate": 0.0004266194635347368,
+      "loss": 3.2818328857421877,
+      "step": 4180
+    },
+    {
+      "epoch": 8.783451165226499,
+      "learning_rate": 0.00042764008426089646,
+      "loss": 3.3082435607910154,
+      "step": 4190
+    },
+    {
+      "epoch": 8.804399057344854,
+      "learning_rate": 0.0004286607049870561,
+      "loss": 3.2165481567382814,
+      "step": 4200
+    },
+    {
+      "epoch": 8.82534694946321,
+      "learning_rate": 0.00042968132571321574,
+      "loss": 3.217595672607422,
+      "step": 4210
+    },
+    {
+      "epoch": 8.846294841581566,
+      "learning_rate": 0.0004307019464393754,
+      "loss": 3.2202774047851563,
+      "step": 4220
+    },
+    {
+      "epoch": 8.867242733699921,
+      "learning_rate": 0.0004317225671655351,
+      "loss": 3.3156604766845703,
+      "step": 4230
+    },
+    {
+      "epoch": 8.888190625818277,
+      "learning_rate": 0.00043274318789169474,
+      "loss": 3.2440589904785155,
+      "step": 4240
+    },
+    {
+      "epoch": 8.909138517936633,
+      "learning_rate": 0.0004337638086178544,
+      "loss": 3.354948043823242,
+      "step": 4250
+    },
+    {
+      "epoch": 8.930086410054988,
+      "learning_rate": 0.0004347844293440141,
+      "loss": 3.2932735443115235,
+      "step": 4260
+    },
+    {
+      "epoch": 8.951034302173344,
+      "learning_rate": 0.0004358050500701737,
+      "loss": 3.1642740249633787,
+      "step": 4270
+    },
+    {
+      "epoch": 8.9719821942917,
+      "learning_rate": 0.00043682567079633336,
+      "loss": 3.1953109741210937,
+      "step": 4280
+    },
+    {
+      "epoch": 8.992930086410055,
+      "learning_rate": 0.00043784629152249303,
+      "loss": 3.1374893188476562,
+      "step": 4290
+    },
+    {
+      "epoch": 9.014663524482849,
+      "learning_rate": 0.0004388669122486527,
+      "loss": 3.4447471618652346,
+      "step": 4300
+    },
+    {
+      "epoch": 9.035611416601204,
+      "learning_rate": 0.00043988753297481237,
+      "loss": 3.278203582763672,
+      "step": 4310
+    },
+    {
+      "epoch": 9.05655930871956,
+      "learning_rate": 0.000440908153700972,
+      "loss": 3.2842811584472655,
+      "step": 4320
+    },
+    {
+      "epoch": 9.077507200837916,
+      "learning_rate": 0.00044192877442713165,
+      "loss": 3.2456493377685547,
+      "step": 4330
+    },
+    {
+      "epoch": 9.098455092956272,
+      "learning_rate": 0.0004429493951532913,
+      "loss": 3.18524112701416,
+      "step": 4340
+    },
+    {
+      "epoch": 9.119402985074627,
+      "learning_rate": 0.000443970015879451,
+      "loss": 3.2943866729736326,
+      "step": 4350
+    },
+    {
+      "epoch": 9.140350877192983,
+      "learning_rate": 0.00044499063660561065,
+      "loss": 3.325389862060547,
+      "step": 4360
+    },
+    {
+      "epoch": 9.161298769311339,
+      "learning_rate": 0.0004460112573317703,
+      "loss": 3.1400611877441404,
+      "step": 4370
+    },
+    {
+      "epoch": 9.182246661429694,
+      "learning_rate": 0.00044703187805792994,
+      "loss": 3.272686004638672,
+      "step": 4380
+    },
+    {
+      "epoch": 9.20319455354805,
+      "learning_rate": 0.0004480524987840896,
+      "loss": 3.284004974365234,
+      "step": 4390
+    },
+    {
+      "epoch": 9.224142445666406,
+      "learning_rate": 0.0004490731195102493,
+      "loss": 3.2489898681640623,
+      "step": 4400
+    },
+    {
+      "epoch": 9.24509033778476,
+      "learning_rate": 0.00045009374023640894,
+      "loss": 3.2408329010009767,
+      "step": 4410
+    },
+    {
+      "epoch": 9.266038229903115,
+      "learning_rate": 0.0004511143609625686,
+      "loss": 3.2516738891601564,
+      "step": 4420
+    },
+    {
+      "epoch": 9.286986122021471,
+      "learning_rate": 0.0004521349816887283,
+      "loss": 3.1824373245239257,
+      "step": 4430
+    },
+    {
+      "epoch": 9.307934014139827,
+      "learning_rate": 0.0004531556024148879,
+      "loss": 3.2249637603759767,
+      "step": 4440
+    },
+    {
+      "epoch": 9.328881906258182,
+      "learning_rate": 0.00045417622314104756,
+      "loss": 3.174178695678711,
+      "step": 4450
+    },
+    {
+      "epoch": 9.349829798376538,
+      "learning_rate": 0.00045519684386720723,
+      "loss": 3.188156318664551,
+      "step": 4460
+    },
+    {
+      "epoch": 9.370777690494894,
+      "learning_rate": 0.0004562174645933669,
+      "loss": 3.228466796875,
+      "step": 4470
+    },
+    {
+      "epoch": 9.39172558261325,
+      "learning_rate": 0.00045723808531952657,
+      "loss": 3.2763172149658204,
+      "step": 4480
+    },
+    {
+      "epoch": 9.412673474731605,
+      "learning_rate": 0.0004582587060456861,
+      "loss": 3.266025161743164,
+      "step": 4490
+    },
+    {
+      "epoch": 9.43362136684996,
+      "learning_rate": 0.0004592793267718458,
+      "loss": 3.2390396118164064,
+      "step": 4500
+    },
+    {
+      "epoch": 9.454569258968316,
+      "learning_rate": 0.00046029994749800546,
+      "loss": 3.23939208984375,
+      "step": 4510
+    },
+    {
+      "epoch": 9.475517151086672,
+      "learning_rate": 0.00046132056822416513,
+      "loss": 3.219116973876953,
+      "step": 4520
+    },
+    {
+      "epoch": 9.496465043205028,
+      "learning_rate": 0.0004623411889503248,
+      "loss": 3.277789306640625,
+      "step": 4530
+    },
+    {
+      "epoch": 9.517412935323383,
+      "learning_rate": 0.00046336180967648447,
+      "loss": 3.284175491333008,
+      "step": 4540
+    },
+    {
+      "epoch": 9.538360827441739,
+      "learning_rate": 0.0004643824304026441,
+      "loss": 3.208120346069336,
+      "step": 4550
+    },
+    {
+      "epoch": 9.559308719560095,
+      "learning_rate": 0.00046540305112880375,
+      "loss": 3.271435546875,
+      "step": 4560
+    },
+    {
+      "epoch": 9.58025661167845,
+      "learning_rate": 0.0004664236718549634,
+      "loss": 3.2989322662353517,
+      "step": 4570
+    },
+    {
+      "epoch": 9.601204503796806,
+      "learning_rate": 0.0004674442925811231,
+      "loss": 3.1721576690673827,
+      "step": 4580
+    },
+    {
+      "epoch": 9.622152395915162,
+      "learning_rate": 0.00046846491330728275,
+      "loss": 3.1907968521118164,
+      "step": 4590
+    },
+    {
+      "epoch": 9.643100288033516,
+      "learning_rate": 0.00046948553403344237,
+      "loss": 3.3379592895507812,
+      "step": 4600
+    },
+    {
+      "epoch": 9.664048180151871,
+      "learning_rate": 0.00047050615475960204,
+      "loss": 3.2489646911621093,
+      "step": 4610
+    },
+    {
+      "epoch": 9.684996072270227,
+      "learning_rate": 0.0004715267754857617,
+      "loss": 3.3012271881103517,
+      "step": 4620
+    },
+    {
+      "epoch": 9.705943964388583,
+      "learning_rate": 0.00047254739621192137,
+      "loss": 3.235492706298828,
+      "step": 4630
+    },
+    {
+      "epoch": 9.726891856506938,
+      "learning_rate": 0.00047356801693808104,
+      "loss": 3.2170711517333985,
+      "step": 4640
+    },
+    {
+      "epoch": 9.747839748625294,
+      "learning_rate": 0.0004745886376642407,
+      "loss": 3.1358510971069338,
+      "step": 4650
+    },
+    {
+      "epoch": 9.76878764074365,
+      "learning_rate": 0.0004756092583904003,
+      "loss": 3.154219055175781,
+      "step": 4660
+    },
+    {
+      "epoch": 9.789735532862005,
+      "learning_rate": 0.00047662987911656,
+      "loss": 3.3397506713867187,
+      "step": 4670
+    },
+    {
+      "epoch": 9.810683424980361,
+      "learning_rate": 0.00047765049984271966,
+      "loss": 3.2498783111572265,
+      "step": 4680
+    },
+    {
+      "epoch": 9.831631317098717,
+      "learning_rate": 0.00047867112056887933,
+      "loss": 3.2211677551269533,
+      "step": 4690
+    },
+    {
+      "epoch": 9.852579209217073,
+      "learning_rate": 0.000479691741295039,
+      "loss": 3.2689888000488283,
+      "step": 4700
+    },
+    {
+      "epoch": 9.873527101335428,
+      "learning_rate": 0.00048071236202119866,
+      "loss": 3.1732282638549805,
+      "step": 4710
+    },
+    {
+      "epoch": 9.894474993453784,
+      "learning_rate": 0.0004817329827473583,
+      "loss": 3.1548320770263674,
+      "step": 4720
+    },
+    {
+      "epoch": 9.91542288557214,
+      "learning_rate": 0.00048275360347351795,
+      "loss": 3.2713703155517577,
+      "step": 4730
+    },
+    {
+      "epoch": 9.936370777690495,
+      "learning_rate": 0.0004837742241996776,
+      "loss": 3.1592134475708007,
+      "step": 4740
+    },
+    {
+      "epoch": 9.957318669808851,
+      "learning_rate": 0.0004847948449258373,
+      "loss": 3.213180923461914,
+      "step": 4750
+    },
+    {
+      "epoch": 9.978266561927207,
+      "learning_rate": 0.00048581546565199695,
+      "loss": 3.2513309478759767,
+      "step": 4760
+    },
+    {
+      "epoch": 9.999214454045562,
+      "learning_rate": 0.0004868360863781566,
+      "loss": 3.3612499237060547,
+      "step": 4770
+    },
+    {
+      "epoch": 10.020947892118356,
+      "learning_rate": 0.00048785670710431623,
+      "loss": 3.2473423004150392,
+      "step": 4780
+    },
+    {
+      "epoch": 10.041895784236711,
+      "learning_rate": 0.000488877327830476,
+      "loss": 3.3008792877197264,
+      "step": 4790
+    },
+    {
+      "epoch": 10.062843676355067,
+      "learning_rate": 0.0004898979485566356,
+      "loss": 3.286971664428711,
+      "step": 4800
+    },
+    {
+      "epoch": 10.083791568473423,
+      "learning_rate": 0.0004909185692827952,
+      "loss": 3.300416946411133,
+      "step": 4810
+    },
+    {
+      "epoch": 10.104739460591778,
+      "learning_rate": 0.0004919391900089549,
+      "loss": 3.1919363021850584,
+      "step": 4820
+    },
+    {
+      "epoch": 10.125687352710134,
+      "learning_rate": 0.0004929598107351145,
+      "loss": 3.2064422607421874,
+      "step": 4830
+    },
+    {
+      "epoch": 10.14663524482849,
+      "learning_rate": 0.0004939804314612742,
+      "loss": 3.1432363510131838,
+      "step": 4840
+    },
+    {
+      "epoch": 10.167583136946845,
+      "learning_rate": 0.0004950010521874339,
+      "loss": 3.17150764465332,
+      "step": 4850
+    },
+    {
+      "epoch": 10.188531029065201,
+      "learning_rate": 0.0004960216729135935,
+      "loss": 3.2499244689941404,
+      "step": 4860
+    },
+    {
+      "epoch": 10.209478921183555,
+      "learning_rate": 0.0004970422936397532,
+      "loss": 3.1550519943237303,
+      "step": 4870
+    },
+    {
+      "epoch": 10.23042681330191,
+      "learning_rate": 0.0004980629143659128,
+      "loss": 3.1220830917358398,
+      "step": 4880
+    },
+    {
+      "epoch": 10.251374705420266,
+      "learning_rate": 0.0004990835350920725,
+      "loss": 3.2074295043945313,
+      "step": 4890
+    },
+    {
+      "epoch": 10.272322597538622,
+      "learning_rate": 0.0005001041558182321,
+      "loss": 3.198388862609863,
+      "step": 4900
+    },
+    {
+      "epoch": 10.293270489656978,
+      "learning_rate": 0.0005011247765443918,
+      "loss": 3.218120574951172,
+      "step": 4910
+    },
+    {
+      "epoch": 10.314218381775333,
+      "learning_rate": 0.0005021453972705514,
+      "loss": 3.2131175994873047,
+      "step": 4920
+    },
+    {
+      "epoch": 10.33516627389369,
+      "learning_rate": 0.0005031660179967111,
+      "loss": 3.1987491607666017,
+      "step": 4930
+    },
+    {
+      "epoch": 10.356114166012045,
+      "learning_rate": 0.0005041866387228707,
+      "loss": 3.175269889831543,
+      "step": 4940
+    },
+    {
+      "epoch": 10.3770620581304,
+      "learning_rate": 0.0005052072594490304,
+      "loss": 3.183473014831543,
+      "step": 4950
+    },
+    {
+      "epoch": 10.398009950248756,
+      "learning_rate": 0.00050622788017519,
+      "loss": 3.0571062088012697,
+      "step": 4960
+    },
+    {
+      "epoch": 10.418957842367112,
+      "learning_rate": 0.0005072485009013497,
+      "loss": 3.325624465942383,
+      "step": 4970
+    },
+    {
+      "epoch": 10.439905734485468,
+      "learning_rate": 0.0005082691216275094,
+      "loss": 3.249886703491211,
+      "step": 4980
+    },
+    {
+      "epoch": 10.460853626603823,
+      "learning_rate": 0.000509289742353669,
+      "loss": 3.18145866394043,
+      "step": 4990
+    },
+    {
+      "epoch": 10.481801518722179,
+      "learning_rate": 0.0005103103630798287,
+      "loss": 3.1750720977783202,
+      "step": 5000
+    },
+    {
+      "epoch": 10.502749410840535,
+      "learning_rate": 0.0005098008169087462,
+      "loss": 3.298409271240234,
+      "step": 5010
+    },
+    {
+      "epoch": 10.52369730295889,
+      "learning_rate": 0.0005092927940452339,
+      "loss": 3.0969064712524412,
+      "step": 5020
+    },
+    {
+      "epoch": 10.544645195077246,
+      "learning_rate": 0.0005087862869144114,
+      "loss": 3.3151134490966796,
+      "step": 5030
+    },
+    {
+      "epoch": 10.565593087195602,
+      "learning_rate": 0.0005082812879940277,
+      "loss": 3.2290550231933595,
+      "step": 5040
+    },
+    {
+      "epoch": 10.586540979313957,
+      "learning_rate": 0.0005077777898139921,
+      "loss": 3.140799713134766,
+      "step": 5050
+    },
+    {
+      "epoch": 10.607488871432313,
+      "learning_rate": 0.0005072757849559103,
+      "loss": 3.1083478927612305,
+      "step": 5060
+    },
+    {
+      "epoch": 10.628436763550667,
+      "learning_rate": 0.0005067752660526248,
+      "loss": 3.1976173400878904,
+      "step": 5070
+    },
+    {
+      "epoch": 10.649384655669023,
+      "learning_rate": 0.0005062762257877613,
+      "loss": 3.2580982208251954,
+      "step": 5080
+    },
+    {
+      "epoch": 10.670332547787378,
+      "learning_rate": 0.0005057786568952791,
+      "loss": 3.280201721191406,
+      "step": 5090
+    },
+    {
+      "epoch": 10.691280439905734,
+      "learning_rate": 0.000505282552159027,
+      "loss": 3.141094779968262,
+      "step": 5100
+    },
+    {
+      "epoch": 10.71222833202409,
+      "learning_rate": 0.000504787904412304,
+      "loss": 3.095392608642578,
+      "step": 5110
+    },
+    {
+      "epoch": 10.733176224142445,
+      "learning_rate": 0.000504294706537424,
+      "loss": 3.2083145141601563,
+      "step": 5120
+    },
+    {
+      "epoch": 10.754124116260801,
+      "learning_rate": 0.0005038029514652858,
+      "loss": 3.289701461791992,
+      "step": 5130
+    },
+    {
+      "epoch": 10.775072008379157,
+      "learning_rate": 0.0005033126321749477,
+      "loss": 3.2425827026367187,
+      "step": 5140
+    },
+    {
+      "epoch": 10.796019900497512,
+      "learning_rate": 0.000502823741693206,
+      "loss": 3.277665710449219,
+      "step": 5150
+    },
+    {
+      "epoch": 10.816967792615868,
+      "learning_rate": 0.0005023362730941793,
+      "loss": 3.1909582138061525,
+      "step": 5160
+    },
+    {
+      "epoch": 10.837915684734224,
+      "learning_rate": 0.0005018502194988955,
+      "loss": 3.170912170410156,
+      "step": 5170
+    },
+    {
+      "epoch": 10.85886357685258,
+      "learning_rate": 0.0005013655740748848,
+      "loss": 3.2589969635009766,
+      "step": 5180
+    },
+    {
+      "epoch": 10.879811468970935,
+      "learning_rate": 0.0005008823300357761,
+      "loss": 3.1849817276000976,
+      "step": 5190
+    },
+    {
+      "epoch": 10.90075936108929,
+      "learning_rate": 0.0005004004806408972,
+      "loss": 3.1448366165161135,
+      "step": 5200
+    },
+    {
+      "epoch": 10.921707253207646,
+      "learning_rate": 0.0004999200191948814,
+      "loss": 3.122829055786133,
+      "step": 5210
+    },
+    {
+      "epoch": 10.942655145326002,
+      "learning_rate": 0.0004994409390472751,
+      "loss": 3.1754734039306642,
+      "step": 5220
+    },
+    {
+      "epoch": 10.963603037444358,
+      "learning_rate": 0.0004989632335921523,
+      "loss": 3.1149194717407225,
+      "step": 5230
+    },
+    {
+      "epoch": 10.984550929562714,
+      "learning_rate": 0.0004984868962677315,
+      "loss": 3.100501823425293,
+      "step": 5240
+    },
+    {
+      "epoch": 11.006284367635507,
+      "learning_rate": 0.0004980119205559973,
+      "loss": 3.330778121948242,
+      "step": 5250
+    },
+    {
+      "epoch": 11.027232259753863,
+      "learning_rate": 0.0004975382999823259,
+      "loss": 3.1559074401855467,
+      "step": 5260
+    },
+    {
+      "epoch": 11.048180151872218,
+      "learning_rate": 0.0004970660281151141,
+      "loss": 3.172486114501953,
+      "step": 5270
+    },
+    {
+      "epoch": 11.069128043990574,
+      "learning_rate": 0.0004965950985654126,
+      "loss": 3.1997749328613283,
+      "step": 5280
+    },
+    {
+      "epoch": 11.09007593610893,
+      "learning_rate": 0.0004961255049865635,
+      "loss": 3.2853694915771485,
+      "step": 5290
+    },
+    {
+      "epoch": 11.111023828227285,
+      "learning_rate": 0.0004956572410738401,
+      "loss": 3.145161819458008,
+      "step": 5300
+    },
+    {
+      "epoch": 11.131971720345641,
+      "learning_rate": 0.000495190300564092,
+      "loss": 3.126105308532715,
+      "step": 5310
+    },
+    {
+      "epoch": 11.152919612463997,
+      "learning_rate": 0.0004947246772353933,
+      "loss": 3.2012374877929686,
+      "step": 5320
+    },
+    {
+      "epoch": 11.17386750458235,
+      "learning_rate": 0.0004942603649066942,
+      "loss": 3.1729455947875977,
+      "step": 5330
+    },
+    {
+      "epoch": 11.194815396700706,
+      "learning_rate": 0.0004937973574374762,
+      "loss": 3.148386001586914,
+      "step": 5340
+    },
+    {
+      "epoch": 11.215763288819062,
+      "learning_rate": 0.0004933356487274114,
+      "loss": 3.065207290649414,
+      "step": 5350
+    },
+    {
+      "epoch": 11.236711180937418,
+      "learning_rate": 0.0004928752327160248,
+      "loss": 3.150010108947754,
+      "step": 5360
+    },
+    {
+      "epoch": 11.257659073055773,
+      "learning_rate": 0.0004924161033823598,
+      "loss": 3.2157524108886717,
+      "step": 5370
+    },
+    {
+      "epoch": 11.278606965174129,
+      "learning_rate": 0.0004919582547446482,
+      "loss": 3.0866676330566407,
+      "step": 5380
+    },
+    {
+      "epoch": 11.299554857292485,
+      "learning_rate": 0.0004915016808599824,
+      "loss": 3.2173648834228517,
+      "step": 5390
+    },
+    {
+      "epoch": 11.32050274941084,
+      "learning_rate": 0.0004910463758239914,
+      "loss": 3.1369649887084963,
+      "step": 5400
+    },
+    {
+      "epoch": 11.341450641529196,
+      "learning_rate": 0.0004905923337705201,
+      "loss": 3.0868097305297852,
+      "step": 5410
+    },
+    {
+      "epoch": 11.362398533647552,
+      "learning_rate": 0.0004901395488713123,
+      "loss": 3.2488777160644533,
+      "step": 5420
+    },
+    {
+      "epoch": 11.383346425765907,
+      "learning_rate": 0.0004896880153356963,
+      "loss": 3.121846008300781,
+      "step": 5430
+    },
+    {
+      "epoch": 11.404294317884263,
+      "learning_rate": 0.000489237727410273,
+      "loss": 3.1137548446655274,
+      "step": 5440
+    },
+    {
+      "epoch": 11.425242210002619,
+      "learning_rate": 0.0004887886793786093,
+      "loss": 3.0799299240112306,
+      "step": 5450
+    },
+    {
+      "epoch": 11.446190102120974,
+      "learning_rate": 0.0004883408655609327,
+      "loss": 3.0961063385009764,
+      "step": 5460
+    },
+    {
+      "epoch": 11.46713799423933,
+      "learning_rate": 0.0004878942803138293,
+      "loss": 3.0329910278320313,
+      "step": 5470
+    },
+    {
+      "epoch": 11.488085886357686,
+      "learning_rate": 0.0004874489180299454,
+      "loss": 3.173397445678711,
+      "step": 5480
+    },
+    {
+      "epoch": 11.509033778476041,
+      "learning_rate": 0.00048700477313769213,
+      "loss": 3.1852407455444336,
+      "step": 5490
+    },
+    {
+      "epoch": 11.529981670594397,
+      "learning_rate": 0.00048656184010095185,
+      "loss": 3.2040431976318358,
+      "step": 5500
+    },
+    {
+      "epoch": 11.550929562712753,
+      "learning_rate": 0.00048612011341878916,
+      "loss": 3.189468002319336,
+      "step": 5510
+    },
+    {
+      "epoch": 11.571877454831107,
+      "learning_rate": 0.0004856795876251634,
+      "loss": 3.2286914825439452,
+      "step": 5520
+    },
+    {
+      "epoch": 11.592825346949462,
+      "learning_rate": 0.00048524025728864493,
+      "loss": 3.034438896179199,
+      "step": 5530
+    },
+    {
+      "epoch": 11.613773239067818,
+      "learning_rate": 0.0004848021170121335,
+      "loss": 3.085980987548828,
+      "step": 5540
+    },
+    {
+      "epoch": 11.634721131186174,
+      "learning_rate": 0.0004843651614325803,
+      "loss": 3.1111934661865233,
+      "step": 5550
+    },
+    {
+      "epoch": 11.65566902330453,
+      "learning_rate": 0.00048392938522071163,
+      "loss": 3.159061050415039,
+      "step": 5560
+    },
+    {
+      "epoch": 11.676616915422885,
+      "learning_rate": 0.0004834947830807563,
+      "loss": 3.0602264404296875,
+      "step": 5570
+    },
+    {
+      "epoch": 11.69756480754124,
+      "learning_rate": 0.00048306134975017523,
+      "loss": 3.120003890991211,
+      "step": 5580
+    },
+    {
+      "epoch": 11.718512699659597,
+      "learning_rate": 0.0004826290799993939,
+      "loss": 3.1456703186035155,
+      "step": 5590
+    },
+    {
+      "epoch": 11.739460591777952,
+      "learning_rate": 0.0004821979686315372,
+      "loss": 3.112548828125,
+      "step": 5600
+    },
+    {
+      "epoch": 11.760408483896308,
+      "learning_rate": 0.00048176801048216693,
+      "loss": 3.1137924194335938,
+      "step": 5610
+    },
+    {
+      "epoch": 11.781356376014664,
+      "learning_rate": 0.0004813392004190223,
+      "loss": 3.146605110168457,
+      "step": 5620
+    },
+    {
+      "epoch": 11.80230426813302,
+      "learning_rate": 0.00048091153334176224,
+      "loss": 3.1099647521972655,
+      "step": 5630
+    },
+    {
+      "epoch": 11.823252160251375,
+      "learning_rate": 0.00048048500418171097,
+      "loss": 3.099277305603027,
+      "step": 5640
+    },
+    {
+      "epoch": 11.84420005236973,
+      "learning_rate": 0.0004800596079016053,
+      "loss": 3.085763931274414,
+      "step": 5650
+    },
+    {
+      "epoch": 11.865147944488086,
+      "learning_rate": 0.0004796353394953452,
+      "loss": 3.188782501220703,
+      "step": 5660
+    },
+    {
+      "epoch": 11.886095836606442,
+      "learning_rate": 0.0004792121939877459,
+      "loss": 3.176821708679199,
+      "step": 5670
+    },
+    {
+      "epoch": 11.907043728724798,
+      "learning_rate": 0.00047879016643429336,
+      "loss": 3.067020797729492,
+      "step": 5680
+    },
+    {
+      "epoch": 11.927991620843153,
+      "learning_rate": 0.00047836925192090116,
+      "loss": 3.1511611938476562,
+      "step": 5690
+    },
+    {
+      "epoch": 11.948939512961509,
+      "learning_rate": 0.0004779494455636703,
+      "loss": 3.0738733291625975,
+      "step": 5700
+    },
+    {
+      "epoch": 11.969887405079863,
+      "learning_rate": 0.00047753074250865145,
+      "loss": 3.237213897705078,
+      "step": 5710
+    },
+    {
+      "epoch": 11.990835297198219,
+      "learning_rate": 0.00047711313793160877,
+      "loss": 3.052178382873535,
+      "step": 5720
+    },
+    {
+      "epoch": 12.012568735271014,
+      "learning_rate": 0.000476696627037787,
+      "loss": 3.2651294708251952,
+      "step": 5730
+    },
+    {
+      "epoch": 12.03351662738937,
+      "learning_rate": 0.0004762812050616797,
+      "loss": 3.162643241882324,
+      "step": 5740
+    },
+    {
+      "epoch": 12.054464519507725,
+      "learning_rate": 0.0004758668672668006,
+      "loss": 3.1709291458129885,
+      "step": 5750
+    },
+    {
+      "epoch": 12.07541241162608,
+      "learning_rate": 0.00047545360894545664,
+      "loss": 3.063345527648926,
+      "step": 5760
+    },
+    {
+      "epoch": 12.096360303744436,
+      "learning_rate": 0.0004750414254185235,
+      "loss": 3.093794250488281,
+      "step": 5770
+    },
+    {
+      "epoch": 12.117308195862792,
+      "learning_rate": 0.0004746303120352226,
+      "loss": 3.1082719802856444,
+      "step": 5780
+    },
+    {
+      "epoch": 12.138256087981146,
+      "learning_rate": 0.00047422026417290146,
+      "loss": 3.1271081924438477,
+      "step": 5790
+    },
+    {
+      "epoch": 12.159203980099502,
+      "learning_rate": 0.0004738112772368146,
+      "loss": 3.141692543029785,
+      "step": 5800
+    },
+    {
+      "epoch": 12.180151872217857,
+      "learning_rate": 0.00047340334665990787,
+      "loss": 3.1134639739990235,
+      "step": 5810
+    },
+    {
+      "epoch": 12.201099764336213,
+      "learning_rate": 0.0004729964679026039,
+      "loss": 3.03677921295166,
+      "step": 5820
+    },
+    {
+      "epoch": 12.222047656454569,
+      "learning_rate": 0.0004725906364525903,
+      "loss": 3.2071063995361326,
+      "step": 5830
+    },
+    {
+      "epoch": 12.242995548572924,
+      "learning_rate": 0.0004721858478246089,
+      "loss": 3.173069953918457,
+      "step": 5840
+    },
+    {
+      "epoch": 12.26394344069128,
+      "learning_rate": 0.0004717820975602482,
+      "loss": 3.048240089416504,
+      "step": 5850
+    },
+    {
+      "epoch": 12.284891332809636,
+      "learning_rate": 0.0004713793812277367,
+      "loss": 3.041463088989258,
+      "step": 5860
+    },
+    {
+      "epoch": 12.305839224927992,
+      "learning_rate": 0.00047097769442173856,
+      "loss": 3.0645767211914063,
+      "step": 5870
+    },
+    {
+      "epoch": 12.326787117046347,
+      "learning_rate": 0.00047057703276315164,
+      "loss": 2.982158088684082,
+      "step": 5880
+    },
+    {
+      "epoch": 12.347735009164703,
+      "learning_rate": 0.0004701773918989065,
+      "loss": 3.111321449279785,
+      "step": 5890
+    },
+    {
+      "epoch": 12.368682901283059,
+      "learning_rate": 0.00046977876750176805,
+      "loss": 3.141143798828125,
+      "step": 5900
+    },
+    {
+      "epoch": 12.389630793401414,
+      "learning_rate": 0.0004693811552701385,
+      "loss": 3.0916566848754883,
+      "step": 5910
+    },
+    {
+      "epoch": 12.41057868551977,
+      "learning_rate": 0.0004689845509278626,
+      "loss": 3.0807928085327148,
+      "step": 5920
+    },
+    {
+      "epoch": 12.431526577638126,
+      "learning_rate": 0.00046858895022403474,
+      "loss": 3.0439529418945312,
+      "step": 5930
+    },
+    {
+      "epoch": 12.452474469756481,
+      "learning_rate": 0.000468194348932807,
+      "loss": 3.0425508499145506,
+      "step": 5940
+    },
+    {
+      "epoch": 12.473422361874837,
+      "learning_rate": 0.00046780074285319984,
+      "loss": 3.187800407409668,
+      "step": 5950
+    },
+    {
+      "epoch": 12.494370253993193,
+      "learning_rate": 0.0004674081278089144,
+      "loss": 3.0460309982299805,
+      "step": 5960
+    },
+    {
+      "epoch": 12.515318146111547,
+      "learning_rate": 0.00046701649964814616,
+      "loss": 3.2187931060791017,
+      "step": 5970
+    },
+    {
+      "epoch": 12.536266038229904,
+      "learning_rate": 0.0004666258542434007,
+      "loss": 3.109378433227539,
+      "step": 5980
+    },
+    {
+      "epoch": 12.557213930348258,
+      "learning_rate": 0.000466236187491311,
+      "loss": 3.0655149459838866,
+      "step": 5990
+    },
+    {
+      "epoch": 12.578161822466614,
+      "learning_rate": 0.00046584749531245617,
+      "loss": 3.132980728149414,
+      "step": 6000
+    },
+    {
+      "epoch": 12.59910971458497,
+      "learning_rate": 0.0004654597736511823,
+      "loss": 3.0032047271728515,
+      "step": 6010
+    },
+    {
+      "epoch": 12.620057606703325,
+      "learning_rate": 0.0004650730184754247,
+      "loss": 3.0569095611572266,
+      "step": 6020
+    },
+    {
+      "epoch": 12.64100549882168,
+      "learning_rate": 0.0004646872257765318,
+      "loss": 3.1891340255737304,
+      "step": 6030
+    },
+    {
+      "epoch": 12.661953390940036,
+      "learning_rate": 0.00046430239156909045,
+      "loss": 3.036951446533203,
+      "step": 6040
+    },
+    {
+      "epoch": 12.682901283058392,
+      "learning_rate": 0.00046391851189075343,
+      "loss": 3.077804374694824,
+      "step": 6050
+    },
+    {
+      "epoch": 12.703849175176748,
+      "learning_rate": 0.00046353558280206746,
+      "loss": 3.1198028564453124,
+      "step": 6060
+    },
+    {
+      "epoch": 12.724797067295103,
+      "learning_rate": 0.00046315360038630404,
+      "loss": 3.1289579391479494,
+      "step": 6070
+    },
+    {
+      "epoch": 12.745744959413459,
+      "learning_rate": 0.0004627725607492909,
+      "loss": 3.047295570373535,
+      "step": 6080
+    },
+    {
+      "epoch": 12.766692851531815,
+      "learning_rate": 0.00046239246001924503,
+      "loss": 3.071992111206055,
+      "step": 6090
+    },
+    {
+      "epoch": 12.78764074365017,
+      "learning_rate": 0.000462013294346608,
+      "loss": 3.050577735900879,
+      "step": 6100
+    },
+    {
+      "epoch": 12.808588635768526,
+      "learning_rate": 0.00046163505990388167,
+      "loss": 3.0774341583251954,
+      "step": 6110
+    },
+    {
+      "epoch": 12.829536527886882,
+      "learning_rate": 0.00046125775288546623,
+      "loss": 3.120297431945801,
+      "step": 6120
+    },
+    {
+      "epoch": 12.850484420005237,
+      "learning_rate": 0.00046088136950749937,
+      "loss": 3.1301042556762697,
+      "step": 6130
+    },
+    {
+      "epoch": 12.871432312123593,
+      "learning_rate": 0.0004605059060076967,
+      "loss": 3.0827388763427734,
+      "step": 6140
+    },
+    {
+      "epoch": 12.892380204241949,
+      "learning_rate": 0.0004601313586451939,
+      "loss": 3.136738967895508,
+      "step": 6150
+    },
+    {
+      "epoch": 12.913328096360305,
+      "learning_rate": 0.00045975772370039034,
+      "loss": 3.0567752838134767,
+      "step": 6160
+    },
+    {
+      "epoch": 12.93427598847866,
+      "learning_rate": 0.0004593849974747937,
+      "loss": 3.053047752380371,
+      "step": 6170
+    },
+    {
+      "epoch": 12.955223880597014,
+      "learning_rate": 0.0004590131762908664,
+      "loss": 3.0607650756835936,
+      "step": 6180
+    },
+    {
+      "epoch": 12.97617177271537,
+      "learning_rate": 0.00045864225649187287,
+      "loss": 3.1021827697753905,
+      "step": 6190
+    },
+    {
+      "epoch": 12.997119664833725,
+      "learning_rate": 0.000458272234441729,
+      "loss": 3.016301727294922,
+      "step": 6200
+    },
+    {
+      "epoch": 13.01885310290652,
+      "learning_rate": 0.00045790310652485205,
+      "loss": 3.1855663299560546,
+      "step": 6210
+    },
+    {
+      "epoch": 13.039800995024876,
+      "learning_rate": 0.0004575348691460124,
+      "loss": 3.0596897125244142,
+      "step": 6220
+    },
+    {
+      "epoch": 13.060748887143232,
+      "learning_rate": 0.00045716751873018654,
+      "loss": 3.061813735961914,
+      "step": 6230
+    },
+    {
+      "epoch": 13.081696779261588,
+      "learning_rate": 0.00045680105172241103,
+      "loss": 3.106767463684082,
+      "step": 6240
+    },
+    {
+      "epoch": 13.102644671379942,
+      "learning_rate": 0.0004564354645876384,
+      "loss": 3.101357269287109,
+      "step": 6250
+    },
+    {
+      "epoch": 13.123592563498297,
+      "learning_rate": 0.00045607075381059363,
+      "loss": 3.085792350769043,
+      "step": 6260
+    },
+    {
+      "epoch": 13.144540455616653,
+      "learning_rate": 0.00045570691589563234,
+      "loss": 3.071797752380371,
+      "step": 6270
+    },
+    {
+      "epoch": 13.165488347735009,
+      "learning_rate": 0.0004553439473666,
+      "loss": 3.087900161743164,
+      "step": 6280
+    },
+    {
+      "epoch": 13.186436239853364,
+      "learning_rate": 0.0004549818447666924,
+      "loss": 3.0981624603271483,
+      "step": 6290
+    },
+    {
+      "epoch": 13.20738413197172,
+      "learning_rate": 0.00045462060465831743,
+      "loss": 2.995559501647949,
+      "step": 6300
+    },
+    {
+      "epoch": 13.228332024090076,
+      "learning_rate": 0.0004542602236229581,
+      "loss": 3.0986444473266603,
+      "step": 6310
+    },
+    {
+      "epoch": 13.249279916208431,
+      "learning_rate": 0.00045390069826103653,
+      "loss": 3.1423923492431642,
+      "step": 6320
+    },
+    {
+      "epoch": 13.270227808326787,
+      "learning_rate": 0.00045354202519177925,
+      "loss": 2.981964111328125,
+      "step": 6330
+    },
+    {
+      "epoch": 13.291175700445143,
+      "learning_rate": 0.0004531842010530839,
+      "loss": 3.062668800354004,
+      "step": 6340
+    },
+    {
+      "epoch": 13.312123592563498,
+      "learning_rate": 0.0004528272225013865,
+      "loss": 3.1759321212768556,
+      "step": 6350
+    },
+    {
+      "epoch": 13.333071484681854,
+      "learning_rate": 0.00045247108621153056,
+      "loss": 3.0892358779907227,
+      "step": 6360
+    },
+    {
+      "epoch": 13.35401937680021,
+      "learning_rate": 0.0004521157888766368,
+      "loss": 3.0303468704223633,
+      "step": 6370
+    },
+    {
+      "epoch": 13.374967268918565,
+      "learning_rate": 0.00045176132720797443,
+      "loss": 3.042502021789551,
+      "step": 6380
+    },
+    {
+      "epoch": 13.395915161036921,
+      "learning_rate": 0.0004514076979348328,
+      "loss": 3.109409713745117,
+      "step": 6390
+    },
+    {
+      "epoch": 13.416863053155277,
+      "learning_rate": 0.0004510548978043951,
+      "loss": 3.0352380752563475,
+      "step": 6400
+    },
+    {
+      "epoch": 13.437810945273633,
+      "learning_rate": 0.00045070292358161265,
+      "loss": 2.9740083694458006,
+      "step": 6410
+    },
+    {
+      "epoch": 13.458758837391988,
+      "learning_rate": 0.0004503517720490801,
+      "loss": 3.1181098937988283,
+      "step": 6420
+    },
+    {
+      "epoch": 13.479706729510344,
+      "learning_rate": 0.000450001440006912,
+      "loss": 3.089175987243652,
+      "step": 6430
+    },
+    {
+      "epoch": 13.500654621628698,
+      "learning_rate": 0.00044965192427262043,
+      "loss": 3.0885658264160156,
+      "step": 6440
+    },
+    {
+      "epoch": 13.521602513747053,
+      "learning_rate": 0.0004493032216809934,
+      "loss": 3.0794023513793944,
+      "step": 6450
+    },
+    {
+      "epoch": 13.54255040586541,
+      "learning_rate": 0.00044895532908397455,
+      "loss": 3.0824106216430662,
+      "step": 6460
+    },
+    {
+      "epoch": 13.563498297983765,
+      "learning_rate": 0.00044860824335054384,
+      "loss": 3.0646196365356446,
+      "step": 6470
+    },
+    {
+      "epoch": 13.58444619010212,
+      "learning_rate": 0.00044826196136659916,
+      "loss": 3.044062614440918,
+      "step": 6480
+    },
+    {
+      "epoch": 13.605394082220476,
+      "learning_rate": 0.00044791648003483884,
+      "loss": 3.0133747100830077,
+      "step": 6490
+    },
+    {
+      "epoch": 13.626341974338832,
+      "learning_rate": 0.0004475717962746455,
+      "loss": 3.070328712463379,
+      "step": 6500
+    },
+    {
+      "epoch": 13.647289866457188,
+      "learning_rate": 0.0004472279070219706,
+      "loss": 3.0279052734375,
+      "step": 6510
+    },
+    {
+      "epoch": 13.668237758575543,
+      "learning_rate": 0.00044688480922922,
+      "loss": 3.147620964050293,
+      "step": 6520
+    },
+    {
+      "epoch": 13.689185650693899,
+      "learning_rate": 0.00044654249986514057,
+      "loss": 3.1151987075805665,
+      "step": 6530
+    },
+    {
+      "epoch": 13.710133542812255,
+      "learning_rate": 0.0004462009759147076,
+      "loss": 3.072108268737793,
+      "step": 6540
+    },
+    {
+      "epoch": 13.73108143493061,
+      "learning_rate": 0.0004458602343790135,
+      "loss": 3.134627342224121,
+      "step": 6550
+    },
+    {
+      "epoch": 13.752029327048966,
+      "learning_rate": 0.00044552027227515704,
+      "loss": 3.032268524169922,
+      "step": 6560
+    },
+    {
+      "epoch": 13.772977219167322,
+      "learning_rate": 0.00044518108663613355,
+      "loss": 3.065017509460449,
+      "step": 6570
+    },
+    {
+      "epoch": 13.793925111285677,
+      "learning_rate": 0.00044484267451072644,
+      "loss": 3.0611106872558596,
+      "step": 6580
+    },
+    {
+      "epoch": 13.814873003404033,
+      "learning_rate": 0.0004445050329633992,
+      "loss": 3.2091243743896483,
+      "step": 6590
+    },
+    {
+      "epoch": 13.835820895522389,
+      "learning_rate": 0.0004441681590741884,
+      "loss": 3.058238983154297,
+      "step": 6600
+    },
+    {
+      "epoch": 13.856768787640744,
+      "learning_rate": 0.0004438320499385977,
+      "loss": 3.073333168029785,
+      "step": 6610
+    },
+    {
+      "epoch": 13.8777166797591,
+      "learning_rate": 0.00044349670266749286,
+      "loss": 3.037291145324707,
+      "step": 6620
+    },
+    {
+      "epoch": 13.898664571877454,
+      "learning_rate": 0.0004431621143869969,
+      "loss": 3.0411745071411134,
+      "step": 6630
+    },
+    {
+      "epoch": 13.91961246399581,
+      "learning_rate": 0.00044282828223838727,
+      "loss": 3.148990440368652,
+      "step": 6640
+    },
+    {
+      "epoch": 13.940560356114165,
+      "learning_rate": 0.0004424952033779929,
+      "loss": 3.014286994934082,
+      "step": 6650
+    },
+    {
+      "epoch": 13.961508248232521,
+      "learning_rate": 0.00044216287497709253,
+      "loss": 3.0138343811035155,
+      "step": 6660
+    },
+    {
+      "epoch": 13.982456140350877,
+      "learning_rate": 0.0004418312942218139,
+      "loss": 2.9974302291870116,
+      "step": 6670
+    },
+    {
+      "epoch": 14.004189578423672,
+      "learning_rate": 0.0004415004583130336,
+      "loss": 3.1527809143066405,
+      "step": 6680
+    },
+    {
+      "epoch": 14.025137470542028,
+      "learning_rate": 0.0004411703644662778,
+      "loss": 2.9581697463989256,
+      "step": 6690
+    },
+    {
+      "epoch": 14.046085362660381,
+      "learning_rate": 0.00044084100991162385,
+      "loss": 3.0720396041870117,
+      "step": 6700
+    },
+    {
+      "epoch": 14.067033254778737,
+      "learning_rate": 0.00044051239189360286,
+      "loss": 3.0470098495483398,
+      "step": 6710
+    },
+    {
+      "epoch": 14.087981146897093,
+      "learning_rate": 0.00044018450767110235,
+      "loss": 3.0677566528320312,
+      "step": 6720
+    },
+    {
+      "epoch": 14.108929039015448,
+      "learning_rate": 0.0004398573545172709,
+      "loss": 3.110503005981445,
+      "step": 6730
+    },
+    {
+      "epoch": 14.129876931133804,
+      "learning_rate": 0.0004395309297194223,
+      "loss": 2.9687520980834963,
+      "step": 6740
+    },
+    {
+      "epoch": 14.15082482325216,
+      "learning_rate": 0.0004392052305789416,
+      "loss": 2.985172080993652,
+      "step": 6750
+    },
+    {
+      "epoch": 14.171772715370516,
+      "learning_rate": 0.0004388802544111908,
+      "loss": 3.0017110824584963,
+      "step": 6760
+    },
+    {
+      "epoch": 14.192720607488871,
+      "learning_rate": 0.0004385559985454165,
+      "loss": 3.041835403442383,
+      "step": 6770
+    },
+    {
+      "epoch": 14.213668499607227,
+      "learning_rate": 0.0004382324603246575,
+      "loss": 3.0984907150268555,
+      "step": 6780
+    },
+    {
+      "epoch": 14.234616391725583,
+      "learning_rate": 0.0004379096371056532,
+      "loss": 3.0736331939697266,
+      "step": 6790
+    },
+    {
+      "epoch": 14.255564283843938,
+      "learning_rate": 0.000437587526258753,
+      "loss": 2.9994585037231447,
+      "step": 6800
+    },
+    {
+      "epoch": 14.276512175962294,
+      "learning_rate": 0.0004372661251678265,
+      "loss": 3.1013252258300783,
+      "step": 6810
+    },
+    {
+      "epoch": 14.29746006808065,
+      "learning_rate": 0.00043694543123017407,
+      "loss": 3.102655220031738,
+      "step": 6820
+    },
+    {
+      "epoch": 14.318407960199005,
+      "learning_rate": 0.0004366254418564382,
+      "loss": 3.1105621337890623,
+      "step": 6830
+    },
+    {
+      "epoch": 14.339355852317361,
+      "learning_rate": 0.0004363061544705161,
+      "loss": 3.1368709564208985,
+      "step": 6840
+    },
+    {
+      "epoch": 14.360303744435717,
+      "learning_rate": 0.0004359875665094723,
+      "loss": 3.052720069885254,
+      "step": 6850
+    },
+    {
+      "epoch": 14.381251636554072,
+      "learning_rate": 0.00043566967542345227,
+      "loss": 2.977310562133789,
+      "step": 6860
+    },
+    {
+      "epoch": 14.402199528672428,
+      "learning_rate": 0.00043535247867559673,
+      "loss": 3.111159896850586,
+      "step": 6870
+    },
+    {
+      "epoch": 14.423147420790784,
+      "learning_rate": 0.00043503597374195665,
+      "loss": 2.959975814819336,
+      "step": 6880
+    },
+    {
+      "epoch": 14.444095312909138,
+      "learning_rate": 0.0004347201581114088,
+      "loss": 3.125636100769043,
+      "step": 6890
+    },
+    {
+      "epoch": 14.465043205027493,
+      "learning_rate": 0.0004344050292855724,
+      "loss": 2.9577571868896486,
+      "step": 6900
+    },
+    {
+      "epoch": 14.485991097145849,
+      "learning_rate": 0.00043409058477872554,
+      "loss": 3.101388931274414,
+      "step": 6910
+    },
+    {
+      "epoch": 14.506938989264205,
+      "learning_rate": 0.00043377682211772343,
+      "loss": 3.0661073684692384,
+      "step": 6920
+    },
+    {
+      "epoch": 14.52788688138256,
+      "learning_rate": 0.0004334637388419161,
+      "loss": 2.974909019470215,
+      "step": 6930
+    },
+    {
+      "epoch": 14.548834773500916,
+      "learning_rate": 0.0004331513325030681,
+      "loss": 3.0314458847045898,
+      "step": 6940
+    },
+    {
+      "epoch": 14.569782665619272,
+      "learning_rate": 0.0004328396006652773,
+      "loss": 3.1744915008544923,
+      "step": 6950
+    },
+    {
+      "epoch": 14.590730557737627,
+      "learning_rate": 0.00043252854090489564,
+      "loss": 3.06768741607666,
+      "step": 6960
+    },
+    {
+      "epoch": 14.611678449855983,
+      "learning_rate": 0.00043221815081044985,
+      "loss": 3.161996269226074,
+      "step": 6970
+    },
+    {
+      "epoch": 14.632626341974339,
+      "learning_rate": 0.00043190842798256285,
+      "loss": 3.102631378173828,
+      "step": 6980
+    },
+    {
+      "epoch": 14.653574234092694,
+      "learning_rate": 0.00043159937003387584,
+      "loss": 3.004058074951172,
+      "step": 6990
+    },
+    {
+      "epoch": 14.67452212621105,
+      "learning_rate": 0.00043129097458897135,
+      "loss": 3.135270118713379,
+      "step": 7000
+    },
+    {
+      "epoch": 14.695470018329406,
+      "learning_rate": 0.000430983239284296,
+      "loss": 3.0084003448486327,
+      "step": 7010
+    },
+    {
+      "epoch": 14.716417910447761,
+      "learning_rate": 0.0004306761617680849,
+      "loss": 2.9995773315429686,
+      "step": 7020
+    },
+    {
+      "epoch": 14.737365802566117,
+      "learning_rate": 0.00043036973970028583,
+      "loss": 3.000468444824219,
+      "step": 7030
+    },
+    {
+      "epoch": 14.758313694684473,
+      "learning_rate": 0.00043006397075248464,
+      "loss": 3.0801364898681642,
+      "step": 7040
+    },
+    {
+      "epoch": 14.779261586802829,
+      "learning_rate": 0.00042975885260783056,
+      "loss": 3.0909229278564454,
+      "step": 7050
+    },
+    {
+      "epoch": 14.800209478921184,
+      "learning_rate": 0.00042945438296096303,
+      "loss": 2.9928516387939452,
+      "step": 7060
+    },
+    {
+      "epoch": 14.82115737103954,
+      "learning_rate": 0.0004291505595179379,
+      "loss": 3.036148262023926,
+      "step": 7070
+    },
+    {
+      "epoch": 14.842105263157894,
+      "learning_rate": 0.0004288473799961553,
+      "loss": 3.0693193435668946,
+      "step": 7080
+    },
+    {
+      "epoch": 14.863053155276251,
+      "learning_rate": 0.0004285448421242875,
+      "loss": 3.0427278518676757,
+      "step": 7090
+    },
+    {
+      "epoch": 14.884001047394605,
+      "learning_rate": 0.00042824294364220724,
+      "loss": 2.9749155044555664,
+      "step": 7100
+    },
+    {
+      "epoch": 14.90494893951296,
+      "learning_rate": 0.0004279416823009172,
+      "loss": 2.9990673065185547,
+      "step": 7110
+    },
+    {
+      "epoch": 14.925896831631317,
+      "learning_rate": 0.0004276410558624791,
+      "loss": 3.091754913330078,
+      "step": 7120
+    },
+    {
+      "epoch": 14.946844723749672,
+      "learning_rate": 0.0004273410620999446,
+      "loss": 3.003107452392578,
+      "step": 7130
+    },
+    {
+      "epoch": 14.967792615868028,
+      "learning_rate": 0.0004270416987972853,
+      "loss": 3.023390007019043,
+      "step": 7140
+    },
+    {
+      "epoch": 14.988740507986384,
+      "learning_rate": 0.00042674296374932424,
+      "loss": 3.068536376953125,
+      "step": 7150
+    },
+    {
+      "epoch": 15.010473946059177,
+      "learning_rate": 0.0004264448547616681,
+      "loss": 3.245321273803711,
+      "step": 7160
+    },
+    {
+      "epoch": 15.031421838177533,
+      "learning_rate": 0.00042614736965063864,
+      "loss": 2.9358680725097654,
+      "step": 7170
+    },
+    {
+      "epoch": 15.052369730295888,
+      "learning_rate": 0.0004258505062432064,
+      "loss": 3.0279872894287108,
+      "step": 7180
+    },
+    {
+      "epoch": 15.073317622414244,
+      "learning_rate": 0.0004255542623769234,
+      "loss": 2.96344108581543,
+      "step": 7190
+    },
+    {
+      "epoch": 15.0942655145326,
+      "learning_rate": 0.00042525863589985727,
+      "loss": 3.1603927612304688,
+      "step": 7200
+    },
+    {
+      "epoch": 15.115213406650955,
+      "learning_rate": 0.00042496362467052564,
+      "loss": 3.0409677505493162,
+      "step": 7210
+    },
+    {
+      "epoch": 15.136161298769311,
+      "learning_rate": 0.00042466922655783073,
+      "loss": 3.154404067993164,
+      "step": 7220
+    },
+    {
+      "epoch": 15.157109190887667,
+      "learning_rate": 0.00042437543944099504,
+      "loss": 2.999993324279785,
+      "step": 7230
+    },
+    {
+      "epoch": 15.178057083006022,
+      "learning_rate": 0.00042408226120949674,
+      "loss": 2.962456703186035,
+      "step": 7240
+    },
+    {
+      "epoch": 15.199004975124378,
+      "learning_rate": 0.00042378968976300647,
+      "loss": 3.050062561035156,
+      "step": 7250
+    },
+    {
+      "epoch": 15.219952867242734,
+      "learning_rate": 0.00042349772301132377,
+      "loss": 3.058196258544922,
+      "step": 7260
+    },
+    {
+      "epoch": 15.24090075936109,
+      "learning_rate": 0.0004232063588743146,
+      "loss": 3.0242469787597654,
+      "step": 7270
+    },
+    {
+      "epoch": 15.261848651479445,
+      "learning_rate": 0.00042291559528184904,
+      "loss": 2.929056930541992,
+      "step": 7280
+    },
+    {
+      "epoch": 15.2827965435978,
+      "learning_rate": 0.0004226254301737393,
+      "loss": 3.0346649169921873,
+      "step": 7290
+    },
+    {
+      "epoch": 15.303744435716157,
+      "learning_rate": 0.0004223358614996787,
+      "loss": 3.079379081726074,
+      "step": 7300
+    },
+    {
+      "epoch": 15.324692327834512,
+      "learning_rate": 0.00042204688721918075,
+      "loss": 2.984081268310547,
+      "step": 7310
+    },
+    {
+      "epoch": 15.345640219952868,
+      "learning_rate": 0.0004217585053015187,
+      "loss": 3.025343322753906,
+      "step": 7320
+    },
+    {
+      "epoch": 15.366588112071224,
+      "learning_rate": 0.0004214707137256656,
+      "loss": 3.016037940979004,
+      "step": 7330
+    },
+    {
+      "epoch": 15.38753600418958,
+      "learning_rate": 0.0004211835104802349,
+      "loss": 3.1470058441162108,
+      "step": 7340
+    },
+    {
+      "epoch": 15.408483896307935,
+      "learning_rate": 0.00042089689356342115,
+      "loss": 3.007353591918945,
+      "step": 7350
+    },
+    {
+      "epoch": 15.429431788426289,
+      "learning_rate": 0.0004206108609829418,
+      "loss": 3.0402362823486326,
+      "step": 7360
+    },
+    {
+      "epoch": 15.450379680544645,
+      "learning_rate": 0.00042032541075597875,
+      "loss": 3.018893241882324,
+      "step": 7370
+    },
+    {
+      "epoch": 15.471327572663,
+      "learning_rate": 0.0004200405409091207,
+      "loss": 3.0513105392456055,
+      "step": 7380
+    },
+    {
+      "epoch": 15.492275464781356,
+      "learning_rate": 0.00041975624947830593,
+      "loss": 3.0438756942749023,
+      "step": 7390
+    },
+    {
+      "epoch": 15.513223356899712,
+      "learning_rate": 0.00041947253450876515,
+      "loss": 2.977249526977539,
+      "step": 7400
+    },
+    {
+      "epoch": 15.534171249018067,
+      "learning_rate": 0.00041918939405496546,
+      "loss": 2.956187629699707,
+      "step": 7410
+    },
+    {
+      "epoch": 15.555119141136423,
+      "learning_rate": 0.00041890682618055396,
+      "loss": 3.0582775115966796,
+      "step": 7420
+    },
+    {
+      "epoch": 15.576067033254779,
+      "learning_rate": 0.0004186248289583023,
+      "loss": 2.9948537826538084,
+      "step": 7430
+    },
+    {
+      "epoch": 15.597014925373134,
+      "learning_rate": 0.00041834340047005144,
+      "loss": 2.9837194442749024,
+      "step": 7440
+    },
+    {
+      "epoch": 15.61796281749149,
+      "learning_rate": 0.0004180625388066569,
+      "loss": 3.0729391098022463,
+      "step": 7450
+    },
+    {
+      "epoch": 15.638910709609846,
+      "learning_rate": 0.00041778224206793433,
+      "loss": 3.054386329650879,
+      "step": 7460
+    },
+    {
+      "epoch": 15.659858601728201,
+      "learning_rate": 0.00041750250836260536,
+      "loss": 3.102676582336426,
+      "step": 7470
+    },
+    {
+      "epoch": 15.680806493846557,
+      "learning_rate": 0.0004172233358082443,
+      "loss": 2.986006164550781,
+      "step": 7480
+    },
+    {
+      "epoch": 15.701754385964913,
+      "learning_rate": 0.00041694472253122467,
+      "loss": 3.0711380004882813,
+      "step": 7490
+    },
+    {
+      "epoch": 15.722702278083268,
+      "learning_rate": 0.00041666666666666664,
+      "loss": 2.9737503051757814,
+      "step": 7500
+    },
+    {
+      "epoch": 15.743650170201624,
+      "learning_rate": 0.0004163891663583843,
+      "loss": 3.030619812011719,
+      "step": 7510
+    },
+    {
+      "epoch": 15.76459806231998,
+      "learning_rate": 0.00041611221975883396,
+      "loss": 3.0626684188842774,
+      "step": 7520
+    },
+    {
+      "epoch": 15.785545954438335,
+      "learning_rate": 0.00041583582502906203,
+      "loss": 2.9612255096435547,
+      "step": 7530
+    },
+    {
+      "epoch": 15.806493846556691,
+      "learning_rate": 0.0004155599803386543,
+      "loss": 2.9540287017822267,
+      "step": 7540
+    },
+    {
+      "epoch": 15.827441738675045,
+      "learning_rate": 0.0004152846838656846,
+      "loss": 2.9890960693359374,
+      "step": 7550
+    },
+    {
+      "epoch": 15.8483896307934,
+      "learning_rate": 0.00041500993379666443,
+      "loss": 2.998134803771973,
+      "step": 7560
+    },
+    {
+      "epoch": 15.869337522911756,
+      "learning_rate": 0.0004147357283264927,
+      "loss": 3.046440315246582,
+      "step": 7570
+    },
+    {
+      "epoch": 15.890285415030112,
+      "learning_rate": 0.000414462065658406,
+      "loss": 2.9803043365478517,
+      "step": 7580
+    },
+    {
+      "epoch": 15.911233307148468,
+      "learning_rate": 0.0004141889440039292,
+      "loss": 3.0623497009277343,
+      "step": 7590
+    },
+    {
+      "epoch": 15.932181199266823,
+      "learning_rate": 0.00041391636158282614,
+      "loss": 2.926837921142578,
+      "step": 7600
+    },
+    {
+      "epoch": 15.953129091385179,
+      "learning_rate": 0.00041364431662305114,
+      "loss": 2.932399368286133,
+      "step": 7610
+    },
+    {
+      "epoch": 15.974076983503535,
+      "learning_rate": 0.0004133728073607005,
+      "loss": 2.996663284301758,
+      "step": 7620
+    },
+    {
+      "epoch": 15.99502487562189,
+      "learning_rate": 0.00041310183203996446,
+      "loss": 3.0678241729736326,
+      "step": 7630
+    },
+    {
+      "epoch": 16.016758313694684,
+      "learning_rate": 0.0004128313889130795,
+      "loss": 3.262166213989258,
+      "step": 7640
+    },
+    {
+      "epoch": 16.03770620581304,
+      "learning_rate": 0.0004125614762402809,
+      "loss": 3.072698402404785,
+      "step": 7650
+    },
+    {
+      "epoch": 16.058654097931395,
+      "learning_rate": 0.00041229209228975627,
+      "loss": 2.960147476196289,
+      "step": 7660
+    },
+    {
+      "epoch": 16.079601990049753,
+      "learning_rate": 0.000412023235337598,
+      "loss": 2.987987518310547,
+      "step": 7670
+    },
+    {
+      "epoch": 16.100549882168107,
+      "learning_rate": 0.00041175490366775766,
+      "loss": 2.9958822250366213,
+      "step": 7680
+    },
+    {
+      "epoch": 16.121497774286464,
+      "learning_rate": 0.0004114870955719997,
+      "loss": 3.0043949127197265,
+      "step": 7690
+    },
+    {
+      "epoch": 16.142445666404818,
+      "learning_rate": 0.00041121980934985563,
+      "loss": 3.013554573059082,
+      "step": 7700
+    },
+    {
+      "epoch": 16.163393558523175,
+      "learning_rate": 0.000410953043308579,
+      "loss": 3.0146947860717774,
+      "step": 7710
+    },
+    {
+      "epoch": 16.18434145064153,
+      "learning_rate": 0.0004106867957631001,
+      "loss": 2.9756107330322266,
+      "step": 7720
+    },
+    {
+      "epoch": 16.205289342759883,
+      "learning_rate": 0.00041042106503598165,
+      "loss": 2.998594284057617,
+      "step": 7730
+    },
+    {
+      "epoch": 16.22623723487824,
+      "learning_rate": 0.0004101558494573738,
+      "loss": 2.968126678466797,
+      "step": 7740
+    },
+    {
+      "epoch": 16.247185126996595,
+      "learning_rate": 0.0004098911473649706,
+      "loss": 2.938851737976074,
+      "step": 7750
+    },
+    {
+      "epoch": 16.268133019114952,
+      "learning_rate": 0.0004096269571039658,
+      "loss": 3.0778596878051756,
+      "step": 7760
+    },
+    {
+      "epoch": 16.289080911233306,
+      "learning_rate": 0.00040936327702701005,
+      "loss": 2.871398162841797,
+      "step": 7770
+    },
+    {
+      "epoch": 16.310028803351663,
+      "learning_rate": 0.00040910010549416687,
+      "loss": 2.9972572326660156,
+      "step": 7780
+    },
+    {
+      "epoch": 16.330976695470017,
+      "learning_rate": 0.0004088374408728706,
+      "loss": 3.1002374649047852,
+      "step": 7790
+    },
+    {
+      "epoch": 16.351924587588375,
+      "learning_rate": 0.0004085752815378834,
+      "loss": 2.9723093032836916,
+      "step": 7800
+    },
+    {
+      "epoch": 16.37287247970673,
+      "learning_rate": 0.0004083136258712532,
+      "loss": 3.0089197158813477,
+      "step": 7810
+    },
+    {
+      "epoch": 16.393820371825086,
+      "learning_rate": 0.0004080524722622717,
+      "loss": 2.9960916519165037,
+      "step": 7820
+    },
+    {
+      "epoch": 16.41476826394344,
+      "learning_rate": 0.00040779181910743294,
+      "loss": 2.948496437072754,
+      "step": 7830
+    },
+    {
+      "epoch": 16.435716156061797,
+      "learning_rate": 0.0004075316648103914,
+      "loss": 2.986690139770508,
+      "step": 7840
+    },
+    {
+      "epoch": 16.45666404818015,
+      "learning_rate": 0.0004072720077819216,
+      "loss": 3.064560317993164,
+      "step": 7850
+    },
+    {
+      "epoch": 16.47761194029851,
+      "learning_rate": 0.0004070128464398768,
+      "loss": 3.0366847991943358,
+      "step": 7860
+    },
+    {
+      "epoch": 16.498559832416863,
+      "learning_rate": 0.0004067541792091489,
+      "loss": 3.0212535858154297,
+      "step": 7870
+    },
+    {
+      "epoch": 16.51950772453522,
+      "learning_rate": 0.0004064960045216279,
+      "loss": 2.966229057312012,
+      "step": 7880
+    },
+    {
+      "epoch": 16.540455616653574,
+      "learning_rate": 0.0004062383208161624,
+      "loss": 2.9718713760375977,
+      "step": 7890
+    },
+    {
+      "epoch": 16.56140350877193,
+      "learning_rate": 0.0004059811265385193,
+      "loss": 2.938900947570801,
+      "step": 7900
+    },
+    {
+      "epoch": 16.582351400890285,
+      "learning_rate": 0.00040572442014134516,
+      "loss": 3.0135732650756837,
+      "step": 7910
+    },
+    {
+      "epoch": 16.60329929300864,
+      "learning_rate": 0.00040546820008412654,
+      "loss": 3.016792869567871,
+      "step": 7920
+    },
+    {
+      "epoch": 16.624247185126997,
+      "learning_rate": 0.0004052124648331515,
+      "loss": 2.961100387573242,
+      "step": 7930
+    },
+    {
+      "epoch": 16.64519507724535,
+      "learning_rate": 0.00040495721286147086,
+      "loss": 2.9855838775634767,
+      "step": 7940
+    },
+    {
+      "epoch": 16.666142969363708,
+      "learning_rate": 0.00040470244264886006,
+      "loss": 3.0260711669921876,
+      "step": 7950
+    },
+    {
+      "epoch": 16.687090861482062,
+      "learning_rate": 0.00040444815268178097,
+      "loss": 2.9670747756958007,
+      "step": 7960
+    },
+    {
+      "epoch": 16.70803875360042,
+      "learning_rate": 0.00040419434145334414,
+      "loss": 2.9868255615234376,
+      "step": 7970
+    },
+    {
+      "epoch": 16.728986645718773,
+      "learning_rate": 0.00040394100746327154,
+      "loss": 2.993141746520996,
+      "step": 7980
+    },
+    {
+      "epoch": 16.74993453783713,
+      "learning_rate": 0.0004036881492178589,
+      "loss": 2.991754722595215,
+      "step": 7990
+    },
+    {
+      "epoch": 16.770882429955485,
+      "learning_rate": 0.00040343576522993926,
+      "loss": 3.0531938552856444,
+      "step": 8000
+    },
+    {
+      "epoch": 16.791830322073842,
+      "learning_rate": 0.00040318385401884554,
+      "loss": 2.8399303436279295,
+      "step": 8010
+    },
+    {
+      "epoch": 16.812778214192196,
+      "learning_rate": 0.00040293241411037484,
+      "loss": 3.0251434326171873,
+      "step": 8020
+    },
+    {
+      "epoch": 16.833726106310554,
+      "learning_rate": 0.00040268144403675154,
+      "loss": 2.9438486099243164,
+      "step": 8030
+    },
+    {
+      "epoch": 16.854673998428908,
+      "learning_rate": 0.0004024309423365915,
+      "loss": 2.9782060623168944,
+      "step": 8040
+    },
+    {
+      "epoch": 16.875621890547265,
+      "learning_rate": 0.0004021809075548668,
+      "loss": 2.972634696960449,
+      "step": 8050
+    },
+    {
+      "epoch": 16.89656978266562,
+      "learning_rate": 0.0004019313382428694,
+      "loss": 2.979868507385254,
+      "step": 8060
+    },
+    {
+      "epoch": 16.917517674783976,
+      "learning_rate": 0.00040168223295817656,
+      "loss": 2.990520477294922,
+      "step": 8070
+    },
+    {
+      "epoch": 16.93846556690233,
+      "learning_rate": 0.00040143359026461554,
+      "loss": 3.0180835723876953,
+      "step": 8080
+    },
+    {
+      "epoch": 16.959413459020688,
+      "learning_rate": 0.000401185408732229,
+      "loss": 3.0260868072509766,
+      "step": 8090
+    },
+    {
+      "epoch": 16.98036135113904,
+      "learning_rate": 0.0004009376869372401,
+      "loss": 2.9574857711791993,
+      "step": 8100
+    },
+    {
+      "epoch": 17.002094789211835,
+      "learning_rate": 0.00040069042346201864,
+      "loss": 3.0914968490600585,
+      "step": 8110
+    },
+    {
+      "epoch": 17.023042681330192,
+      "learning_rate": 0.00040044361689504655,
+      "loss": 3.062566947937012,
+      "step": 8120
+    },
+    {
+      "epoch": 17.043990573448546,
+      "learning_rate": 0.0004001972658308847,
+      "loss": 2.976962661743164,
+      "step": 8130
+    },
+    {
+      "epoch": 17.064938465566904,
+      "learning_rate": 0.0003999513688701383,
+      "loss": 2.9750572204589845,
+      "step": 8140
+    },
+    {
+      "epoch": 17.085886357685258,
+      "learning_rate": 0.00039970592461942457,
+      "loss": 2.987382698059082,
+      "step": 8150
+    },
+    {
+      "epoch": 17.106834249803615,
+      "learning_rate": 0.00039946093169133874,
+      "loss": 3.001695442199707,
+      "step": 8160
+    },
+    {
+      "epoch": 17.12778214192197,
+      "learning_rate": 0.0003992163887044217,
+      "loss": 2.961598777770996,
+      "step": 8170
+    },
+    {
+      "epoch": 17.148730034040323,
+      "learning_rate": 0.0003989722942831268,
+      "loss": 2.957429313659668,
+      "step": 8180
+    },
+    {
+      "epoch": 17.16967792615868,
+      "learning_rate": 0.0003987286470577879,
+      "loss": 3.0364120483398436,
+      "step": 8190
+    },
+    {
+      "epoch": 17.190625818277034,
+      "learning_rate": 0.0003984854456645864,
+      "loss": 3.0673593521118163,
+      "step": 8200
+    },
+    {
+      "epoch": 17.211573710395392,
+      "learning_rate": 0.0003982426887455199,
+      "loss": 2.9361265182495115,
+      "step": 8210
+    },
+    {
+      "epoch": 17.232521602513746,
+      "learning_rate": 0.00039800037494836985,
+      "loss": 3.001542854309082,
+      "step": 8220
+    },
+    {
+      "epoch": 17.253469494632103,
+      "learning_rate": 0.00039775850292667005,
+      "loss": 2.95641975402832,
+      "step": 8230
+    },
+    {
+      "epoch": 17.274417386750457,
+      "learning_rate": 0.0003975170713396753,
+      "loss": 2.8832208633422853,
+      "step": 8240
+    },
+    {
+      "epoch": 17.295365278868815,
+      "learning_rate": 0.0003972760788523301,
+      "loss": 2.9211734771728515,
+      "step": 8250
+    },
+    {
+      "epoch": 17.31631317098717,
+      "learning_rate": 0.0003970355241352378,
+      "loss": 2.9908830642700197,
+      "step": 8260
+    },
+    {
+      "epoch": 17.337261063105526,
+      "learning_rate": 0.00039679540586462953,
+      "loss": 2.991852951049805,
+      "step": 8270
+    },
+    {
+      "epoch": 17.35820895522388,
+      "learning_rate": 0.00039655572272233384,
+      "loss": 3.0602521896362305,
+      "step": 8280
+    },
+    {
+      "epoch": 17.379156847342237,
+      "learning_rate": 0.0003963164733957462,
+      "loss": 2.975466728210449,
+      "step": 8290
+    },
+    {
+      "epoch": 17.40010473946059,
+      "learning_rate": 0.00039607765657779864,
+      "loss": 3.0946418762207033,
+      "step": 8300
+    },
+    {
+      "epoch": 17.42105263157895,
+      "learning_rate": 0.0003958392709669304,
+      "loss": 2.993026924133301,
+      "step": 8310
+    },
+    {
+      "epoch": 17.442000523697303,
+      "learning_rate": 0.00039560131526705723,
+      "loss": 3.0121936798095703,
+      "step": 8320
+    },
+    {
+      "epoch": 17.46294841581566,
+      "learning_rate": 0.0003953637881875425,
+      "loss": 3.0414730072021485,
+      "step": 8330
+    },
+    {
+      "epoch": 17.483896307934014,
+      "learning_rate": 0.0003951266884431675,
+      "loss": 3.0235416412353517,
+      "step": 8340
+    },
+    {
+      "epoch": 17.50484420005237,
+      "learning_rate": 0.00039489001475410214,
+      "loss": 2.9818603515625,
+      "step": 8350
+    },
+    {
+      "epoch": 17.525792092170725,
+      "learning_rate": 0.00039465376584587626,
+      "loss": 2.994624137878418,
+      "step": 8360
+    },
+    {
+      "epoch": 17.54673998428908,
+      "learning_rate": 0.00039441794044935054,
+      "loss": 3.0084808349609373,
+      "step": 8370
+    },
+    {
+      "epoch": 17.567687876407437,
+      "learning_rate": 0.00039418253730068797,
+      "loss": 2.8967424392700196,
+      "step": 8380
+    },
+    {
+      "epoch": 17.58863576852579,
+      "learning_rate": 0.0003939475551413253,
+      "loss": 2.9253704071044924,
+      "step": 8390
+    },
+    {
+      "epoch": 17.609583660644148,
+      "learning_rate": 0.000393712992717945,
+      "loss": 3.0752674102783204,
+      "step": 8400
+    },
+    {
+      "epoch": 17.630531552762502,
+      "learning_rate": 0.0003934788487824469,
+      "loss": 2.925820159912109,
+      "step": 8410
+    },
+    {
+      "epoch": 17.65147944488086,
+      "learning_rate": 0.0003932451220919205,
+      "loss": 2.9983007431030275,
+      "step": 8420
+    },
+    {
+      "epoch": 17.672427336999213,
+      "learning_rate": 0.0003930118114086172,
+      "loss": 2.884238624572754,
+      "step": 8430
+    },
+    {
+      "epoch": 17.69337522911757,
+      "learning_rate": 0.00039277891549992266,
+      "loss": 2.854781723022461,
+      "step": 8440
+    },
+    {
+      "epoch": 17.714323121235925,
+      "learning_rate": 0.0003925464331383298,
+      "loss": 2.9886890411376954,
+      "step": 8450
+    },
+    {
+      "epoch": 17.735271013354282,
+      "learning_rate": 0.00039231436310141113,
+      "loss": 3.0236677169799804,
+      "step": 8460
+    },
+    {
+      "epoch": 17.756218905472636,
+      "learning_rate": 0.00039208270417179214,
+      "loss": 2.951685905456543,
+      "step": 8470
+    },
+    {
+      "epoch": 17.777166797590993,
+      "learning_rate": 0.0003918514551371243,
+      "loss": 2.971786880493164,
+      "step": 8480
+    },
+    {
+      "epoch": 17.798114689709347,
+      "learning_rate": 0.0003916206147900585,
+      "loss": 2.983307647705078,
+      "step": 8490
+    },
+    {
+      "epoch": 17.819062581827705,
+      "learning_rate": 0.00039139018192821845,
+      "loss": 2.9782459259033205,
+      "step": 8500
+    },
+    {
+      "epoch": 17.84001047394606,
+      "learning_rate": 0.00039116015535417445,
+      "loss": 3.02642765045166,
+      "step": 8510
+    },
+    {
+      "epoch": 17.860958366064416,
+      "learning_rate": 0.00039093053387541745,
+      "loss": 3.011845588684082,
+      "step": 8520
+    },
+    {
+      "epoch": 17.88190625818277,
+      "learning_rate": 0.00039070131630433274,
+      "loss": 2.939919090270996,
+      "step": 8530
+    },
+    {
+      "epoch": 17.902854150301128,
+      "learning_rate": 0.00039047250145817424,
+      "loss": 2.996026039123535,
+      "step": 8540
+    },
+    {
+      "epoch": 17.92380204241948,
+      "learning_rate": 0.00039024408815903914,
+      "loss": 2.903793716430664,
+      "step": 8550
+    },
+    {
+      "epoch": 17.94474993453784,
+      "learning_rate": 0.0003900160752338421,
+      "loss": 2.9599498748779296,
+      "step": 8560
+    },
+    {
+      "epoch": 17.965697826656193,
+      "learning_rate": 0.00038978846151429,
+      "loss": 2.841645050048828,
+      "step": 8570
+    },
+    {
+      "epoch": 17.986645718774547,
+      "learning_rate": 0.0003895612458368572,
+      "loss": 2.885163497924805,
+      "step": 8580
+    },
+    {
+      "epoch": 18.008379156847344,
+      "learning_rate": 0.00038933442704275974,
+      "loss": 2.9911325454711912,
+      "step": 8590
+    },
+    {
+      "epoch": 18.029327048965698,
+      "learning_rate": 0.0003891080039779314,
+      "loss": 3.027914619445801,
+      "step": 8600
+    },
+    {
+      "epoch": 18.050274941084055,
+      "learning_rate": 0.0003888819754929986,
+      "loss": 2.9694196701049806,
+      "step": 8610
+    },
+    {
+      "epoch": 18.07122283320241,
+      "learning_rate": 0.0003886563404432558,
+      "loss": 2.9900096893310546,
+      "step": 8620
+    },
+    {
+      "epoch": 18.092170725320763,
+      "learning_rate": 0.0003884310976886414,
+      "loss": 2.987308692932129,
+      "step": 8630
+    },
+    {
+      "epoch": 18.11311861743912,
+      "learning_rate": 0.0003882062460937135,
+      "loss": 2.934325408935547,
+      "step": 8640
+    },
+    {
+      "epoch": 18.134066509557474,
+      "learning_rate": 0.0003879817845276255,
+      "loss": 2.9511764526367186,
+      "step": 8650
+    },
+    {
+      "epoch": 18.15501440167583,
+      "learning_rate": 0.0003877577118641029,
+      "loss": 2.9479455947875977,
+      "step": 8660
+    },
+    {
+      "epoch": 18.175962293794186,
+      "learning_rate": 0.00038753402698141903,
+      "loss": 3.0447383880615235,
+      "step": 8670
+    },
+    {
+      "epoch": 18.196910185912543,
+      "learning_rate": 0.0003873107287623715,
+      "loss": 2.917817497253418,
+      "step": 8680
+    },
+    {
+      "epoch": 18.217858078030897,
+      "learning_rate": 0.00038708781609425905,
+      "loss": 2.8964914321899413,
+      "step": 8690
+    },
+    {
+      "epoch": 18.238805970149254,
+      "learning_rate": 0.000386865287868858,
+      "loss": 2.967067527770996,
+      "step": 8700
+    },
+    {
+      "epoch": 18.25975386226761,
+      "learning_rate": 0.0003866431429823993,
+      "loss": 2.991856002807617,
+      "step": 8710
+    },
+    {
+      "epoch": 18.280701754385966,
+      "learning_rate": 0.00038642138033554525,
+      "loss": 2.96053524017334,
+      "step": 8720
+    },
+    {
+      "epoch": 18.30164964650432,
+      "learning_rate": 0.00038619999883336703,
+      "loss": 2.9373369216918945,
+      "step": 8730
+    },
+    {
+      "epoch": 18.322597538622677,
+      "learning_rate": 0.0003859789973853217,
+      "loss": 2.9254953384399416,
+      "step": 8740
+    },
+    {
+      "epoch": 18.34354543074103,
+      "learning_rate": 0.0003857583749052298,
+      "loss": 3.008597564697266,
+      "step": 8750
+    },
+    {
+      "epoch": 18.36449332285939,
+      "learning_rate": 0.0003855381303112527,
+      "loss": 2.9206886291503906,
+      "step": 8760
+    },
+    {
+      "epoch": 18.385441214977742,
+      "learning_rate": 0.0003853182625258708,
+      "loss": 2.9910358428955077,
+      "step": 8770
+    },
+    {
+      "epoch": 18.4063891070961,
+      "learning_rate": 0.0003850987704758608,
+      "loss": 2.894259452819824,
+      "step": 8780
+    },
+    {
+      "epoch": 18.427336999214454,
+      "learning_rate": 0.00038487965309227413,
+      "loss": 3.0239398956298826,
+      "step": 8790
+    },
+    {
+      "epoch": 18.44828489133281,
+      "learning_rate": 0.0003846609093104148,
+      "loss": 3.023584747314453,
+      "step": 8800
+    },
+    {
+      "epoch": 18.469232783451165,
+      "learning_rate": 0.00038444253806981784,
+      "loss": 2.9302574157714845,
+      "step": 8810
+    },
+    {
+      "epoch": 18.49018067556952,
+      "learning_rate": 0.00038422453831422784,
+      "loss": 2.968699264526367,
+      "step": 8820
+    },
+    {
+      "epoch": 18.511128567687877,
+      "learning_rate": 0.0003840069089915771,
+      "loss": 2.904973793029785,
+      "step": 8830
+    },
+    {
+      "epoch": 18.53207645980623,
+      "learning_rate": 0.00038378964905396454,
+      "loss": 2.9212614059448243,
+      "step": 8840
+    },
+    {
+      "epoch": 18.553024351924588,
+      "learning_rate": 0.00038357275745763475,
+      "loss": 2.9412769317626952,
+      "step": 8850
+    },
+    {
+      "epoch": 18.573972244042942,
+      "learning_rate": 0.0003833562331629563,
+      "loss": 2.926407814025879,
+      "step": 8860
+    },
+    {
+      "epoch": 18.5949201361613,
+      "learning_rate": 0.0003831400751344014,
+      "loss": 2.87393741607666,
+      "step": 8870
+    },
+    {
+      "epoch": 18.615868028279653,
+      "learning_rate": 0.00038292428234052486,
+      "loss": 2.93045597076416,
+      "step": 8880
+    },
+    {
+      "epoch": 18.63681592039801,
+      "learning_rate": 0.0003827088537539434,
+      "loss": 2.8498041152954103,
+      "step": 8890
+    },
+    {
+      "epoch": 18.657763812516365,
+      "learning_rate": 0.00038249378835131535,
+      "loss": 2.980220603942871,
+      "step": 8900
+    },
+    {
+      "epoch": 18.678711704634722,
+      "learning_rate": 0.0003822790851133196,
+      "loss": 2.9403018951416016,
+      "step": 8910
+    },
+    {
+      "epoch": 18.699659596753076,
+      "learning_rate": 0.00038206474302463617,
+      "loss": 2.932261848449707,
+      "step": 8920
+    },
+    {
+      "epoch": 18.720607488871433,
+      "learning_rate": 0.00038185076107392544,
+      "loss": 3.0572792053222657,
+      "step": 8930
+    },
+    {
+      "epoch": 18.741555380989787,
+      "learning_rate": 0.0003816371382538082,
+      "loss": 2.8232454299926757,
+      "step": 8940
+    },
+    {
+      "epoch": 18.762503273108145,
+      "learning_rate": 0.0003814238735608459,
+      "loss": 2.9247211456298827,
+      "step": 8950
+    },
+    {
+      "epoch": 18.7834511652265,
+      "learning_rate": 0.0003812109659955207,
+      "loss": 2.9309356689453123,
+      "step": 8960
+    },
+    {
+      "epoch": 18.804399057344856,
+      "learning_rate": 0.00038099841456221617,
+      "loss": 2.9406196594238283,
+      "step": 8970
+    },
+    {
+      "epoch": 18.82534694946321,
+      "learning_rate": 0.0003807862182691969,
+      "loss": 2.9863868713378907,
+      "step": 8980
+    },
+    {
+      "epoch": 18.846294841581567,
+      "learning_rate": 0.00038057437612859003,
+      "loss": 2.9020156860351562,
+      "step": 8990
+    },
+    {
+      "epoch": 18.86724273369992,
+      "learning_rate": 0.0003803628871563653,
+      "loss": 2.8909185409545897,
+      "step": 9000
+    },
+    {
+      "epoch": 18.888190625818275,
+      "learning_rate": 0.0003801517503723161,
+      "loss": 2.905278205871582,
+      "step": 9010
+    },
+    {
+      "epoch": 18.909138517936633,
+      "learning_rate": 0.00037994096480004037,
+      "loss": 2.971329689025879,
+      "step": 9020
+    },
+    {
+      "epoch": 18.930086410054987,
+      "learning_rate": 0.0003797305294669214,
+      "loss": 2.9014846801757814,
+      "step": 9030
+    },
+    {
+      "epoch": 18.951034302173344,
+      "learning_rate": 0.00037952044340410954,
+      "loss": 2.919228363037109,
+      "step": 9040
+    },
+    {
+      "epoch": 18.971982194291698,
+      "learning_rate": 0.00037931070564650276,
+      "loss": 2.9611515045166015,
+      "step": 9050
+    },
+    {
+      "epoch": 18.992930086410055,
+      "learning_rate": 0.0003791013152327286,
+      "loss": 3.06106014251709,
+      "step": 9060
+    },
+    {
+      "epoch": 19.01466352448285,
+      "learning_rate": 0.00037889227120512545,
+      "loss": 3.0117502212524414,
+      "step": 9070
+    },
+    {
+      "epoch": 19.035611416601206,
+      "learning_rate": 0.0003786835726097239,
+      "loss": 2.908797836303711,
+      "step": 9080
+    },
+    {
+      "epoch": 19.05655930871956,
+      "learning_rate": 0.00037847521849622895,
+      "loss": 2.977317047119141,
+      "step": 9090
+    },
+    {
+      "epoch": 19.077507200837914,
+      "learning_rate": 0.0003782672079180015,
+      "loss": 2.982106018066406,
+      "step": 9100
+    },
+    {
+      "epoch": 19.09845509295627,
+      "learning_rate": 0.0003780595399320404,
+      "loss": 2.9336454391479494,
+      "step": 9110
+    },
+    {
+      "epoch": 19.119402985074625,
+      "learning_rate": 0.00037785221359896444,
+      "loss": 2.9511249542236326,
+      "step": 9120
+    },
+    {
+      "epoch": 19.140350877192983,
+      "learning_rate": 0.00037764522798299443,
+      "loss": 2.9214210510253906,
+      "step": 9130
+    },
+    {
+      "epoch": 19.161298769311337,
+      "learning_rate": 0.0003774385821519358,
+      "loss": 2.9141027450561525,
+      "step": 9140
+    },
+    {
+      "epoch": 19.182246661429694,
+      "learning_rate": 0.0003772322751771605,
+      "loss": 3.0709518432617187,
+      "step": 9150
+    },
+    {
+      "epoch": 19.203194553548048,
+      "learning_rate": 0.00037702630613358986,
+      "loss": 2.9491186141967773,
+      "step": 9160
+    },
+    {
+      "epoch": 19.224142445666406,
+      "learning_rate": 0.0003768206740996769,
+      "loss": 2.9979949951171876,
+      "step": 9170
+    },
+    {
+      "epoch": 19.24509033778476,
+      "learning_rate": 0.00037661537815738915,
+      "loss": 2.9300355911254883,
+      "step": 9180
+    },
+    {
+      "epoch": 19.266038229903117,
+      "learning_rate": 0.00037641041739219143,
+      "loss": 2.95788631439209,
+      "step": 9190
+    },
+    {
+      "epoch": 19.28698612202147,
+      "learning_rate": 0.00037620579089302876,
+      "loss": 2.960785675048828,
+      "step": 9200
+    },
+    {
+      "epoch": 19.30793401413983,
+      "learning_rate": 0.0003760014977523091,
+      "loss": 2.9440664291381835,
+      "step": 9210
+    },
+    {
+      "epoch": 19.328881906258182,
+      "learning_rate": 0.00037579753706588697,
+      "loss": 2.9864282608032227,
+      "step": 9220
+    },
+    {
+      "epoch": 19.34982979837654,
+      "learning_rate": 0.00037559390793304604,
+      "loss": 2.830784225463867,
+      "step": 9230
+    },
+    {
+      "epoch": 19.370777690494894,
+      "learning_rate": 0.00037539060945648286,
+      "loss": 2.981433868408203,
+      "step": 9240
+    },
+    {
+      "epoch": 19.39172558261325,
+      "learning_rate": 0.00037518764074229014,
+      "loss": 2.8778303146362303,
+      "step": 9250
+    },
+    {
+      "epoch": 19.412673474731605,
+      "learning_rate": 0.00037498500089994,
+      "loss": 2.896072006225586,
+      "step": 9260
+    },
+    {
+      "epoch": 19.433621366849962,
+      "learning_rate": 0.00037478268904226795,
+      "loss": 2.962319564819336,
+      "step": 9270
+    },
+    {
+      "epoch": 19.454569258968316,
+      "learning_rate": 0.00037458070428545635,
+      "loss": 2.9110170364379884,
+      "step": 9280
+    },
+    {
+      "epoch": 19.47551715108667,
+      "learning_rate": 0.00037437904574901817,
+      "loss": 2.928105926513672,
+      "step": 9290
+    },
+    {
+      "epoch": 19.496465043205028,
+      "learning_rate": 0.00037417771255578104,
+      "loss": 2.91275577545166,
+      "step": 9300
+    },
+    {
+      "epoch": 19.51741293532338,
+      "learning_rate": 0.00037397670383187097,
+      "loss": 2.964938163757324,
+      "step": 9310
+    },
+    {
+      "epoch": 19.53836082744174,
+      "learning_rate": 0.0003737760187066967,
+      "loss": 2.953006935119629,
+      "step": 9320
+    },
+    {
+      "epoch": 19.559308719560093,
+      "learning_rate": 0.00037357565631293365,
+      "loss": 3.108722686767578,
+      "step": 9330
+    },
+    {
+      "epoch": 19.58025661167845,
+      "learning_rate": 0.00037337561578650833,
+      "loss": 3.078016471862793,
+      "step": 9340
+    },
+    {
+      "epoch": 19.601204503796804,
+      "learning_rate": 0.00037317589626658255,
+      "loss": 3.006759262084961,
+      "step": 9350
+    },
+    {
+      "epoch": 19.622152395915162,
+      "learning_rate": 0.0003729764968955379,
+      "loss": 2.9581228256225587,
+      "step": 9360
+    },
+    {
+      "epoch": 19.643100288033516,
+      "learning_rate": 0.00037277741681896045,
+      "loss": 3.199405860900879,
+      "step": 9370
+    },
+    {
+      "epoch": 19.664048180151873,
+      "learning_rate": 0.0003725786551856251,
+      "loss": 3.039951133728027,
+      "step": 9380
+    },
+    {
+      "epoch": 19.684996072270227,
+      "learning_rate": 0.0003723802111474804,
+      "loss": 2.99322566986084,
+      "step": 9390
+    },
+    {
+      "epoch": 19.705943964388585,
+      "learning_rate": 0.0003721820838596335,
+      "loss": 2.9919605255126953,
+      "step": 9400
+    },
+    {
+      "epoch": 19.72689185650694,
+      "learning_rate": 0.00037198427248033485,
+      "loss": 2.9185922622680662,
+      "step": 9410
+    },
+    {
+      "epoch": 19.747839748625296,
+      "learning_rate": 0.00037178677617096337,
+      "loss": 2.927593994140625,
+      "step": 9420
+    },
+    {
+      "epoch": 19.76878764074365,
+      "learning_rate": 0.0003715895940960111,
+      "loss": 2.9274175643920897,
+      "step": 9430
+    },
+    {
+      "epoch": 19.789735532862007,
+      "learning_rate": 0.000371392725423069,
+      "loss": 2.9191694259643555,
+      "step": 9440
+    },
+    {
+      "epoch": 19.81068342498036,
+      "learning_rate": 0.00037119616932281165,
+      "loss": 3.002157974243164,
+      "step": 9450
+    },
+    {
+      "epoch": 19.83163131709872,
+      "learning_rate": 0.00037099992496898276,
+      "loss": 2.9765802383422852,
+      "step": 9460
+    },
+    {
+      "epoch": 19.852579209217073,
+      "learning_rate": 0.00037080399153838065,
+      "loss": 2.983877182006836,
+      "step": 9470
+    },
+    {
+      "epoch": 19.873527101335426,
+      "learning_rate": 0.00037060836821084373,
+      "loss": 2.9292572021484373,
+      "step": 9480
+    },
+    {
+      "epoch": 19.894474993453784,
+      "learning_rate": 0.00037041305416923604,
+      "loss": 2.9650571823120115,
+      "step": 9490
+    },
+    {
+      "epoch": 19.915422885572138,
+      "learning_rate": 0.0003702180485994327,
+      "loss": 2.9420921325683596,
+      "step": 9500
+    },
+    {
+      "epoch": 19.936370777690495,
+      "learning_rate": 0.00037002335069030614,
+      "loss": 2.8849225997924806,
+      "step": 9510
+    },
+    {
+      "epoch": 19.95731866980885,
+      "learning_rate": 0.0003698289596337116,
+      "loss": 2.882498550415039,
+      "step": 9520
+    },
+    {
+      "epoch": 19.978266561927207,
+      "learning_rate": 0.00036963487462447303,
+      "loss": 2.9787607192993164,
+      "step": 9530
+    },
+    {
+      "epoch": 19.99921445404556,
+      "learning_rate": 0.0003694410948603691,
+      "loss": 2.93823299407959,
+      "step": 9540
+    },
+    {
+      "epoch": 20.020947892118354,
+      "learning_rate": 0.00036924761954211944,
+      "loss": 3.033322334289551,
+      "step": 9550
+    },
+    {
+      "epoch": 20.04189578423671,
+      "learning_rate": 0.0003690544478733707,
+      "loss": 2.9187992095947264,
+      "step": 9560
+    },
+    {
+      "epoch": 20.062843676355065,
+      "learning_rate": 0.0003688615790606828,
+      "loss": 2.8914941787719726,
+      "step": 9570
+    },
+    {
+      "epoch": 20.083791568473423,
+      "learning_rate": 0.000368669012313515,
+      "loss": 2.8893537521362305,
+      "step": 9580
+    },
+    {
+      "epoch": 20.104739460591777,
+      "learning_rate": 0.0003684767468442126,
+      "loss": 2.906744384765625,
+      "step": 9590
+    },
+    {
+      "epoch": 20.125687352710134,
+      "learning_rate": 0.0003682847818679935,
+      "loss": 2.8719203948974608,
+      "step": 9600
+    },
+    {
+      "epoch": 20.146635244828488,
+      "learning_rate": 0.0003680931166029342,
+      "loss": 2.956478500366211,
+      "step": 9610
+    },
+    {
+      "epoch": 20.167583136946845,
+      "learning_rate": 0.000367901750269957,
+      "loss": 3.0235532760620116,
+      "step": 9620
+    },
+    {
+      "epoch": 20.1885310290652,
+      "learning_rate": 0.00036771068209281657,
+      "loss": 2.9067535400390625,
+      "step": 9630
+    },
+    {
+      "epoch": 20.209478921183557,
+      "learning_rate": 0.0003675199112980863,
+      "loss": 3.0424097061157225,
+      "step": 9640
+    },
+    {
+      "epoch": 20.23042681330191,
+      "learning_rate": 0.0003673294371151458,
+      "loss": 2.873898506164551,
+      "step": 9650
+    },
+    {
+      "epoch": 20.251374705420268,
+      "learning_rate": 0.0003671392587761674,
+      "loss": 2.9879985809326173,
+      "step": 9660
+    },
+    {
+      "epoch": 20.272322597538622,
+      "learning_rate": 0.0003669493755161031,
+      "loss": 2.9691110610961915,
+      "step": 9670
+    },
+    {
+      "epoch": 20.29327048965698,
+      "learning_rate": 0.00036675978657267204,
+      "loss": 2.8997966766357424,
+      "step": 9680
+    },
+    {
+      "epoch": 20.314218381775333,
+      "learning_rate": 0.00036657049118634733,
+      "loss": 3.0071743011474608,
+      "step": 9690
+    },
+    {
+      "epoch": 20.33516627389369,
+      "learning_rate": 0.0003663814886003432,
+      "loss": 2.9313344955444336,
+      "step": 9700
+    },
+    {
+      "epoch": 20.356114166012045,
+      "learning_rate": 0.00036619277806060276,
+      "loss": 2.9554468154907227,
+      "step": 9710
+    },
+    {
+      "epoch": 20.377062058130402,
+      "learning_rate": 0.0003660043588157846,
+      "loss": 2.9395862579345704,
+      "step": 9720
+    },
+    {
+      "epoch": 20.398009950248756,
+      "learning_rate": 0.00036581623011725114,
+      "loss": 2.8985124588012696,
+      "step": 9730
+    },
+    {
+      "epoch": 20.41895784236711,
+      "learning_rate": 0.0003656283912190554,
+      "loss": 2.92258243560791,
+      "step": 9740
+    },
+    {
+      "epoch": 20.439905734485468,
+      "learning_rate": 0.00036544084137792883,
+      "loss": 2.9244316101074217,
+      "step": 9750
+    },
+    {
+      "epoch": 20.46085362660382,
+      "learning_rate": 0.00036525357985326903,
+      "loss": 2.9558000564575195,
+      "step": 9760
+    },
+    {
+      "epoch": 20.48180151872218,
+      "learning_rate": 0.0003650666059071275,
+      "loss": 2.84803466796875,
+      "step": 9770
+    },
+    {
+      "epoch": 20.502749410840533,
+      "learning_rate": 0.00036487991880419725,
+      "loss": 2.9493398666381836,
+      "step": 9780
+    },
+    {
+      "epoch": 20.52369730295889,
+      "learning_rate": 0.00036469351781180073,
+      "loss": 2.925904083251953,
+      "step": 9790
+    },
+    {
+      "epoch": 20.544645195077244,
+      "learning_rate": 0.00036450740219987765,
+      "loss": 2.8513689041137695,
+      "step": 9800
+    },
+    {
+      "epoch": 20.5655930871956,
+      "learning_rate": 0.0003643215712409734,
+      "loss": 2.9981321334838866,
+      "step": 9810
+    },
+    {
+      "epoch": 20.586540979313956,
+      "learning_rate": 0.00036413602421022653,
+      "loss": 2.8937055587768556,
+      "step": 9820
+    },
+    {
+      "epoch": 20.607488871432313,
+      "learning_rate": 0.0003639507603853572,
+      "loss": 3.000600814819336,
+      "step": 9830
+    },
+    {
+      "epoch": 20.628436763550667,
+      "learning_rate": 0.00036376577904665525,
+      "loss": 2.8795480728149414,
+      "step": 9840
+    },
+    {
+      "epoch": 20.649384655669024,
+      "learning_rate": 0.00036358107947696876,
+      "loss": 2.8654504776000977,
+      "step": 9850
+    },
+    {
+      "epoch": 20.67033254778738,
+      "learning_rate": 0.0003633966609616919,
+      "loss": 2.9193105697631836,
+      "step": 9860
+    },
+    {
+      "epoch": 20.691280439905736,
+      "learning_rate": 0.00036321252278875344,
+      "loss": 3.000343894958496,
+      "step": 9870
+    },
+    {
+      "epoch": 20.71222833202409,
+      "learning_rate": 0.00036302866424860566,
+      "loss": 2.9843284606933596,
+      "step": 9880
+    },
+    {
+      "epoch": 20.733176224142447,
+      "learning_rate": 0.00036284508463421217,
+      "loss": 2.9358942031860353,
+      "step": 9890
+    },
+    {
+      "epoch": 20.7541241162608,
+      "learning_rate": 0.0003626617832410371,
+      "loss": 2.9408638000488283,
+      "step": 9900
+    },
+    {
+      "epoch": 20.77507200837916,
+      "learning_rate": 0.00036247875936703335,
+      "loss": 2.9035417556762697,
+      "step": 9910
+    },
+    {
+      "epoch": 20.796019900497512,
+      "learning_rate": 0.00036229601231263145,
+      "loss": 2.8692134857177733,
+      "step": 9920
+    },
+    {
+      "epoch": 20.81696779261587,
+      "learning_rate": 0.0003621135413807282,
+      "loss": 2.8154270172119142,
+      "step": 9930
+    },
+    {
+      "epoch": 20.837915684734224,
+      "learning_rate": 0.0003619313458766758,
+      "loss": 2.905722427368164,
+      "step": 9940
+    },
+    {
+      "epoch": 20.858863576852578,
+      "learning_rate": 0.0003617494251082704,
+      "loss": 2.9300207138061523,
+      "step": 9950
+    },
+    {
+      "epoch": 20.879811468970935,
+      "learning_rate": 0.0003615677783857413,
+      "loss": 2.9284923553466795,
+      "step": 9960
+    },
+    {
+      "epoch": 20.90075936108929,
+      "learning_rate": 0.0003613864050217397,
+      "loss": 2.8830732345581054,
+      "step": 9970
+    },
+    {
+      "epoch": 20.921707253207646,
+      "learning_rate": 0.0003612053043313283,
+      "loss": 2.922040557861328,
+      "step": 9980
+    },
+    {
+      "epoch": 20.942655145326,
+      "learning_rate": 0.0003610244756319697,
+      "loss": 2.919090461730957,
+      "step": 9990
+    },
+    {
+      "epoch": 20.963603037444358,
+      "learning_rate": 0.00036084391824351607,
+      "loss": 2.9119571685791015,
+      "step": 10000
+    },
+    {
+      "epoch": 20.98455092956271,
+      "learning_rate": 0.00036066363148819854,
+      "loss": 3.0065305709838865,
+      "step": 10010
+    },
+    {
+      "epoch": 21.006284367635505,
+      "learning_rate": 0.000360483614690616,
+      "loss": 3.088846206665039,
+      "step": 10020
+    },
+    {
+      "epoch": 21.027232259753863,
+      "learning_rate": 0.00036030386717772494,
+      "loss": 2.887124443054199,
+      "step": 10030
+    },
+    {
+      "epoch": 21.048180151872216,
+      "learning_rate": 0.0003601243882788286,
+      "loss": 2.9657873153686523,
+      "step": 10040
+    },
+    {
+      "epoch": 21.069128043990574,
+      "learning_rate": 0.0003599451773255667,
+      "loss": 2.8971757888793945,
+      "step": 10050
+    },
+    {
+      "epoch": 21.090075936108928,
+      "learning_rate": 0.00035976623365190465,
+      "loss": 2.8915260314941404,
+      "step": 10060
+    },
+    {
+      "epoch": 21.111023828227285,
+      "learning_rate": 0.0003595875565941235,
+      "loss": 2.9463220596313477,
+      "step": 10070
+    },
+    {
+      "epoch": 21.13197172034564,
+      "learning_rate": 0.00035940914549080944,
+      "loss": 2.9156425476074217,
+      "step": 10080
+    },
+    {
+      "epoch": 21.152919612463997,
+      "learning_rate": 0.0003592309996828435,
+      "loss": 2.996392250061035,
+      "step": 10090
+    },
+    {
+      "epoch": 21.17386750458235,
+      "learning_rate": 0.0003590531185133913,
+      "loss": 2.899692344665527,
+      "step": 10100
+    },
+    {
+      "epoch": 21.194815396700708,
+      "learning_rate": 0.0003588755013278929,
+      "loss": 2.9242908477783205,
+      "step": 10110
+    },
+    {
+      "epoch": 21.215763288819062,
+      "learning_rate": 0.00035869814747405306,
+      "loss": 2.846599578857422,
+      "step": 10120
+    },
+    {
+      "epoch": 21.23671118093742,
+      "learning_rate": 0.00035852105630183027,
+      "loss": 2.9692026138305665,
+      "step": 10130
+    },
+    {
+      "epoch": 21.257659073055773,
+      "learning_rate": 0.0003583442271634278,
+      "loss": 2.921228790283203,
+      "step": 10140
+    },
+    {
+      "epoch": 21.27860696517413,
+      "learning_rate": 0.000358167659413283,
+      "loss": 2.906037139892578,
+      "step": 10150
+    },
+    {
+      "epoch": 21.299554857292485,
+      "learning_rate": 0.00035799135240805765,
+      "loss": 2.8459890365600584,
+      "step": 10160
+    },
+    {
+      "epoch": 21.320502749410842,
+      "learning_rate": 0.0003578153055066282,
+      "loss": 2.839315986633301,
+      "step": 10170
+    },
+    {
+      "epoch": 21.341450641529196,
+      "learning_rate": 0.00035763951807007597,
+      "loss": 2.918286895751953,
+      "step": 10180
+    },
+    {
+      "epoch": 21.36239853364755,
+      "learning_rate": 0.0003574639894616771,
+      "loss": 2.9180910110473635,
+      "step": 10190
+    },
+    {
+      "epoch": 21.383346425765907,
+      "learning_rate": 0.0003572887190468934,
+      "loss": 2.96252498626709,
+      "step": 10200
+    },
+    {
+      "epoch": 21.40429431788426,
+      "learning_rate": 0.00035711370619336214,
+      "loss": 2.873885726928711,
+      "step": 10210
+    },
+    {
+      "epoch": 21.42524221000262,
+      "learning_rate": 0.00035693895027088694,
+      "loss": 2.8597929000854494,
+      "step": 10220
+    },
+    {
+      "epoch": 21.446190102120973,
+      "learning_rate": 0.00035676445065142793,
+      "loss": 2.9235706329345703,
+      "step": 10230
+    },
+    {
+      "epoch": 21.46713799423933,
+      "learning_rate": 0.0003565902067090925,
+      "loss": 2.843309783935547,
+      "step": 10240
+    },
+    {
+      "epoch": 21.488085886357684,
+      "learning_rate": 0.0003564162178201257,
+      "loss": 2.9598644256591795,
+      "step": 10250
+    },
+    {
+      "epoch": 21.50903377847604,
+      "learning_rate": 0.0003562424833629007,
+      "loss": 2.8366893768310546,
+      "step": 10260
+    },
+    {
+      "epoch": 21.529981670594395,
+      "learning_rate": 0.0003560690027179101,
+      "loss": 2.927451515197754,
+      "step": 10270
+    },
+    {
+      "epoch": 21.550929562712753,
+      "learning_rate": 0.00035589577526775603,
+      "loss": 2.8992708206176756,
+      "step": 10280
+    },
+    {
+      "epoch": 21.571877454831107,
+      "learning_rate": 0.000355722800397141,
+      "loss": 2.8801607131958007,
+      "step": 10290
+    },
+    {
+      "epoch": 21.592825346949464,
+      "learning_rate": 0.00035555007749285897,
+      "loss": 2.929705047607422,
+      "step": 10300
+    },
+    {
+      "epoch": 21.613773239067818,
+      "learning_rate": 0.00035537760594378607,
+      "loss": 2.9493310928344725,
+      "step": 10310
+    },
+    {
+      "epoch": 21.634721131186176,
+      "learning_rate": 0.00035520538514087155,
+      "loss": 2.9074274063110352,
+      "step": 10320
+    },
+    {
+      "epoch": 21.65566902330453,
+      "learning_rate": 0.0003550334144771289,
+      "loss": 2.8912160873413084,
+      "step": 10330
+    },
+    {
+      "epoch": 21.676616915422887,
+      "learning_rate": 0.00035486169334762637,
+      "loss": 2.8537384033203126,
+      "step": 10340
+    },
+    {
+      "epoch": 21.69756480754124,
+      "learning_rate": 0.00035469022114947857,
+      "loss": 2.899664878845215,
+      "step": 10350
+    },
+    {
+      "epoch": 21.7185126996596,
+      "learning_rate": 0.00035451899728183736,
+      "loss": 2.9585891723632813,
+      "step": 10360
+    },
+    {
+      "epoch": 21.739460591777952,
+      "learning_rate": 0.00035434802114588305,
+      "loss": 2.8878129959106444,
+      "step": 10370
+    },
+    {
+      "epoch": 21.760408483896306,
+      "learning_rate": 0.00035417729214481556,
+      "loss": 2.9733861923217773,
+      "step": 10380
+    },
+    {
+      "epoch": 21.781356376014664,
+      "learning_rate": 0.0003540068096838456,
+      "loss": 2.877838134765625,
+      "step": 10390
+    },
+    {
+      "epoch": 21.802304268133017,
+      "learning_rate": 0.0003538365731701862,
+      "loss": 2.874703598022461,
+      "step": 10400
+    },
+    {
+      "epoch": 21.823252160251375,
+      "learning_rate": 0.0003536665820130437,
+      "loss": 2.8807294845581053,
+      "step": 10410
+    },
+    {
+      "epoch": 21.84420005236973,
+      "learning_rate": 0.00035349683562360966,
+      "loss": 2.9425344467163086,
+      "step": 10420
+    },
+    {
+      "epoch": 21.865147944488086,
+      "learning_rate": 0.0003533273334150517,
+      "loss": 2.9000774383544923,
+      "step": 10430
+    },
+    {
+      "epoch": 21.88609583660644,
+      "learning_rate": 0.0003531580748025054,
+      "loss": 2.8818313598632814,
+      "step": 10440
+    },
+    {
+      "epoch": 21.907043728724798,
+      "learning_rate": 0.00035298905920306563,
+      "loss": 2.9183849334716796,
+      "step": 10450
+    },
+    {
+      "epoch": 21.92799162084315,
+      "learning_rate": 0.00035282028603577823,
+      "loss": 2.843509292602539,
+      "step": 10460
+    },
+    {
+      "epoch": 21.94893951296151,
+      "learning_rate": 0.0003526517547216315,
+      "loss": 2.9971471786499024,
+      "step": 10470
+    },
+    {
+      "epoch": 21.969887405079863,
+      "learning_rate": 0.000352483464683548,
+      "loss": 2.851757621765137,
+      "step": 10480
+    },
+    {
+      "epoch": 21.99083529719822,
+      "learning_rate": 0.0003523154153463761,
+      "loss": 2.889949417114258,
+      "step": 10490
+    },
+    {
+      "epoch": 22.012568735271014,
+      "learning_rate": 0.00035214760613688187,
+      "loss": 2.9378170013427733,
+      "step": 10500
+    },
+    {
+      "epoch": 22.033516627389368,
+      "learning_rate": 0.0003519800364837407,
+      "loss": 2.9217357635498047,
+      "step": 10510
+    },
+    {
+      "epoch": 22.054464519507725,
+      "learning_rate": 0.0003518127058175293,
+      "loss": 2.9642236709594725,
+      "step": 10520
+    },
+    {
+      "epoch": 22.07541241162608,
+      "learning_rate": 0.00035164561357071755,
+      "loss": 2.9043264389038086,
+      "step": 10530
+    },
+    {
+      "epoch": 22.096360303744436,
+      "learning_rate": 0.0003514787591776602,
+      "loss": 2.9355424880981444,
+      "step": 10540
+    },
+    {
+      "epoch": 22.11730819586279,
+      "learning_rate": 0.0003513121420745892,
+      "loss": 2.876350975036621,
+      "step": 10550
+    },
+    {
+      "epoch": 22.138256087981148,
+      "learning_rate": 0.0003511457616996052,
+      "loss": 2.9557785034179687,
+      "step": 10560
+    },
+    {
+      "epoch": 22.1592039800995,
+      "learning_rate": 0.0003509796174926703,
+      "loss": 2.9274904251098635,
+      "step": 10570
+    },
+    {
+      "epoch": 22.18015187221786,
+      "learning_rate": 0.00035081370889559934,
+      "loss": 2.873898506164551,
+      "step": 10580
+    },
+    {
+      "epoch": 22.201099764336213,
+      "learning_rate": 0.0003506480353520526,
+      "loss": 3.015602684020996,
+      "step": 10590
+    },
+    {
+      "epoch": 22.22204765645457,
+      "learning_rate": 0.0003504825963075276,
+      "loss": 2.836939239501953,
+      "step": 10600
+    },
+    {
+      "epoch": 22.242995548572924,
+      "learning_rate": 0.00035031739120935175,
+      "loss": 2.9317630767822265,
+      "step": 10610
+    },
+    {
+      "epoch": 22.263943440691282,
+      "learning_rate": 0.0003501524195066741,
+      "loss": 2.8972990036010744,
+      "step": 10620
+    },
+    {
+      "epoch": 22.284891332809636,
+      "learning_rate": 0.0003499876806504578,
+      "loss": 2.9171611785888674,
+      "step": 10630
+    },
+    {
+      "epoch": 22.305839224927993,
+      "learning_rate": 0.00034982317409347263,
+      "loss": 2.9115921020507813,
+      "step": 10640
+    },
+    {
+      "epoch": 22.326787117046347,
+      "learning_rate": 0.00034965889929028707,
+      "loss": 3.029188537597656,
+      "step": 10650
+    },
+    {
+      "epoch": 22.3477350091647,
+      "learning_rate": 0.000349494855697261,
+      "loss": 2.900659370422363,
+      "step": 10660
+    },
+    {
+      "epoch": 22.36868290128306,
+      "learning_rate": 0.0003493310427725377,
+      "loss": 2.8941118240356447,
+      "step": 10670
+    },
+    {
+      "epoch": 22.389630793401412,
+      "learning_rate": 0.0003491674599760369,
+      "loss": 2.913021278381348,
+      "step": 10680
+    },
+    {
+      "epoch": 22.41057868551977,
+      "learning_rate": 0.0003490041067694469,
+      "loss": 2.907943916320801,
+      "step": 10690
+    },
+    {
+      "epoch": 22.431526577638124,
+      "learning_rate": 0.00034884098261621724,
+      "loss": 2.8749153137207033,
+      "step": 10700
+    },
+    {
+      "epoch": 22.45247446975648,
+      "learning_rate": 0.00034867808698155125,
+      "loss": 2.836786460876465,
+      "step": 10710
+    },
+    {
+      "epoch": 22.473422361874835,
+      "learning_rate": 0.0003485154193323988,
+      "loss": 2.9680070877075195,
+      "step": 10720
+    },
+    {
+      "epoch": 22.494370253993193,
+      "learning_rate": 0.00034835297913744903,
+      "loss": 2.839517593383789,
+      "step": 10730
+    },
+    {
+      "epoch": 22.515318146111547,
+      "learning_rate": 0.0003481907658671227,
+      "loss": 2.9294412612915037,
+      "step": 10740
+    },
+    {
+      "epoch": 22.536266038229904,
+      "learning_rate": 0.0003480287789935653,
+      "loss": 2.931773376464844,
+      "step": 10750
+    },
+    {
+      "epoch": 22.557213930348258,
+      "learning_rate": 0.00034786701799063976,
+      "loss": 2.8776823043823243,
+      "step": 10760
+    },
+    {
+      "epoch": 22.578161822466615,
+      "learning_rate": 0.00034770548233391924,
+      "loss": 2.8429998397827148,
+      "step": 10770
+    },
+    {
+      "epoch": 22.59910971458497,
+      "learning_rate": 0.0003475441715006799,
+      "loss": 2.8952512741088867,
+      "step": 10780
+    },
+    {
+      "epoch": 22.620057606703327,
+      "learning_rate": 0.0003473830849698938,
+      "loss": 2.8576940536499023,
+      "step": 10790
+    },
+    {
+      "epoch": 22.64100549882168,
+      "learning_rate": 0.0003472222222222222,
+      "loss": 2.9290803909301757,
+      "step": 10800
+    },
+    {
+      "epoch": 22.661953390940038,
+      "learning_rate": 0.00034706158274000796,
+      "loss": 2.861796569824219,
+      "step": 10810
+    },
+    {
+      "epoch": 22.682901283058392,
+      "learning_rate": 0.00034690116600726885,
+      "loss": 2.9451555252075194,
+      "step": 10820
+    },
+    {
+      "epoch": 22.70384917517675,
+      "learning_rate": 0.0003467409715096907,
+      "loss": 2.8637598037719725,
+      "step": 10830
+    },
+    {
+      "epoch": 22.724797067295103,
+      "learning_rate": 0.00034658099873462027,
+      "loss": 2.919582176208496,
+      "step": 10840
+    },
+    {
+      "epoch": 22.745744959413457,
+      "learning_rate": 0.0003464212471710583,
+      "loss": 2.9625146865844725,
+      "step": 10850
+    },
+    {
+      "epoch": 22.766692851531815,
+      "learning_rate": 0.0003462617163096529,
+      "loss": 2.8601587295532225,
+      "step": 10860
+    },
+    {
+      "epoch": 22.78764074365017,
+      "learning_rate": 0.00034610240564269265,
+      "loss": 2.8949390411376954,
+      "step": 10870
+    },
+    {
+      "epoch": 22.808588635768526,
+      "learning_rate": 0.0003459433146640997,
+      "loss": 2.8205642700195312,
+      "step": 10880
+    },
+    {
+      "epoch": 22.82953652788688,
+      "learning_rate": 0.00034578444286942307,
+      "loss": 2.93542366027832,
+      "step": 10890
+    },
+    {
+      "epoch": 22.850484420005237,
+      "learning_rate": 0.00034562578975583187,
+      "loss": 2.8913852691650392,
+      "step": 10900
+    },
+    {
+      "epoch": 22.87143231212359,
+      "learning_rate": 0.00034546735482210894,
+      "loss": 2.9478212356567384,
+      "step": 10910
+    },
+    {
+      "epoch": 22.89238020424195,
+      "learning_rate": 0.0003453091375686437,
+      "loss": 2.854338455200195,
+      "step": 10920
+    },
+    {
+      "epoch": 22.913328096360303,
+      "learning_rate": 0.00034515113749742586,
+      "loss": 2.919601058959961,
+      "step": 10930
+    },
+    {
+      "epoch": 22.93427598847866,
+      "learning_rate": 0.00034499335411203894,
+      "loss": 2.88704776763916,
+      "step": 10940
+    },
+    {
+      "epoch": 22.955223880597014,
+      "learning_rate": 0.00034483578691765326,
+      "loss": 2.906253433227539,
+      "step": 10950
+    },
+    {
+      "epoch": 22.97617177271537,
+      "learning_rate": 0.00034467843542102,
+      "loss": 2.8981559753417967,
+      "step": 10960
+    },
+    {
+      "epoch": 22.997119664833725,
+      "learning_rate": 0.0003445212991304641,
+      "loss": 2.9073596954345704,
+      "step": 10970
+    },
+    {
+      "epoch": 23.01885310290652,
+      "learning_rate": 0.00034436437755587827,
+      "loss": 2.960616874694824,
+      "step": 10980
+    },
+    {
+      "epoch": 23.039800995024876,
+      "learning_rate": 0.00034420767020871656,
+      "loss": 2.9549840927124023,
+      "step": 10990
+    },
+    {
+      "epoch": 23.06074888714323,
+      "learning_rate": 0.00034405117660198765,
+      "loss": 2.9256917953491213,
+      "step": 11000
+    },
+    {
+      "epoch": 23.081696779261588,
+      "learning_rate": 0.00034389489625024885,
+      "loss": 2.89876651763916,
+      "step": 11010
+    },
+    {
+      "epoch": 23.10264467137994,
+      "learning_rate": 0.00034373882866959936,
+      "loss": 2.828813362121582,
+      "step": 11020
+    },
+    {
+      "epoch": 23.1235925634983,
+      "learning_rate": 0.0003435829733776745,
+      "loss": 2.9374326705932616,
+      "step": 11030
+    },
+    {
+      "epoch": 23.144540455616653,
+      "learning_rate": 0.00034342732989363903,
+      "loss": 2.928928184509277,
+      "step": 11040
+    },
+    {
+      "epoch": 23.16548834773501,
+      "learning_rate": 0.0003432718977381811,
+      "loss": 2.8702091217041015,
+      "step": 11050
+    },
+    {
+      "epoch": 23.186436239853364,
+      "learning_rate": 0.0003431166764335058,
+      "loss": 2.937228965759277,
+      "step": 11060
+    },
+    {
+      "epoch": 23.207384131971722,
+      "learning_rate": 0.0003429616655033297,
+      "loss": 2.9454578399658202,
+      "step": 11070
+    },
+    {
+      "epoch": 23.228332024090076,
+      "learning_rate": 0.00034280686447287373,
+      "loss": 2.8274587631225585,
+      "step": 11080
+    },
+    {
+      "epoch": 23.249279916208433,
+      "learning_rate": 0.00034265227286885776,
+      "loss": 2.876905632019043,
+      "step": 11090
+    },
+    {
+      "epoch": 23.270227808326787,
+      "learning_rate": 0.00034249789021949435,
+      "loss": 2.8264415740966795,
+      "step": 11100
+    },
+    {
+      "epoch": 23.29117570044514,
+      "learning_rate": 0.0003423437160544826,
+      "loss": 2.8484895706176756,
+      "step": 11110
+    },
+    {
+      "epoch": 23.3121235925635,
+      "learning_rate": 0.0003421897499050022,
+      "loss": 2.885685920715332,
+      "step": 11120
+    },
+    {
+      "epoch": 23.333071484681852,
+      "learning_rate": 0.0003420359913037075,
+      "loss": 2.9073020935058596,
+      "step": 11130
+    },
+    {
+      "epoch": 23.35401937680021,
+      "learning_rate": 0.0003418824397847216,
+      "loss": 2.919231986999512,
+      "step": 11140
+    },
+    {
+      "epoch": 23.374967268918564,
+      "learning_rate": 0.00034172909488363007,
+      "loss": 2.8528385162353516,
+      "step": 11150
+    },
+    {
+      "epoch": 23.39591516103692,
+      "learning_rate": 0.00034157595613747545,
+      "loss": 2.8856670379638674,
+      "step": 11160
+    },
+    {
+      "epoch": 23.416863053155275,
+      "learning_rate": 0.00034142302308475133,
+      "loss": 2.851297950744629,
+      "step": 11170
+    },
+    {
+      "epoch": 23.437810945273633,
+      "learning_rate": 0.0003412702952653962,
+      "loss": 2.8231760025024415,
+      "step": 11180
+    },
+    {
+      "epoch": 23.458758837391986,
+      "learning_rate": 0.00034111777222078796,
+      "loss": 2.9219854354858397,
+      "step": 11190
+    },
+    {
+      "epoch": 23.479706729510344,
+      "learning_rate": 0.00034096545349373804,
+      "loss": 2.977204132080078,
+      "step": 11200
+    },
+    {
+      "epoch": 23.500654621628698,
+      "learning_rate": 0.0003408133386284857,
+      "loss": 2.8231952667236326,
+      "step": 11210
+    },
+    {
+      "epoch": 23.521602513747055,
+      "learning_rate": 0.0003406614271706919,
+      "loss": 2.891893196105957,
+      "step": 11220
+    },
+    {
+      "epoch": 23.54255040586541,
+      "learning_rate": 0.0003405097186674344,
+      "loss": 2.8958648681640624,
+      "step": 11230
+    },
+    {
+      "epoch": 23.563498297983767,
+      "learning_rate": 0.00034035821266720136,
+      "loss": 2.8495506286621093,
+      "step": 11240
+    },
+    {
+      "epoch": 23.58444619010212,
+      "learning_rate": 0.0003402069087198858,
+      "loss": 2.935627746582031,
+      "step": 11250
+    },
+    {
+      "epoch": 23.605394082220478,
+      "learning_rate": 0.00034005580637678053,
+      "loss": 2.840359687805176,
+      "step": 11260
+    },
+    {
+      "epoch": 23.626341974338832,
+      "learning_rate": 0.00033990490519057183,
+      "loss": 2.9121625900268553,
+      "step": 11270
+    },
+    {
+      "epoch": 23.64728986645719,
+      "learning_rate": 0.0003397542047153345,
+      "loss": 2.897580146789551,
+      "step": 11280
+    },
+    {
+      "epoch": 23.668237758575543,
+      "learning_rate": 0.0003396037045065257,
+      "loss": 2.894269561767578,
+      "step": 11290
+    },
+    {
+      "epoch": 23.6891856506939,
+      "learning_rate": 0.0003394534041209802,
+      "loss": 2.923667335510254,
+      "step": 11300
+    },
+    {
+      "epoch": 23.710133542812255,
+      "learning_rate": 0.0003393033031169043,
+      "loss": 2.9479984283447265,
+      "step": 11310
+    },
+    {
+      "epoch": 23.73108143493061,
+      "learning_rate": 0.0003391534010538705,
+      "loss": 2.9156826019287108,
+      "step": 11320
+    },
+    {
+      "epoch": 23.752029327048966,
+      "learning_rate": 0.00033900369749281225,
+      "loss": 2.9133535385131837,
+      "step": 11330
+    },
+    {
+      "epoch": 23.77297721916732,
+      "learning_rate": 0.00033885419199601845,
+      "loss": 2.912689971923828,
+      "step": 11340
+    },
+    {
+      "epoch": 23.793925111285677,
+      "learning_rate": 0.000338704884127128,
+      "loss": 2.921385955810547,
+      "step": 11350
+    },
+    {
+      "epoch": 23.81487300340403,
+      "learning_rate": 0.00033855577345112453,
+      "loss": 2.9245376586914062,
+      "step": 11360
+    },
+    {
+      "epoch": 23.83582089552239,
+      "learning_rate": 0.0003384068595343312,
+      "loss": 2.921660232543945,
+      "step": 11370
+    },
+    {
+      "epoch": 23.856768787640743,
+      "learning_rate": 0.00033825814194440504,
+      "loss": 2.8597009658813475,
+      "step": 11380
+    },
+    {
+      "epoch": 23.8777166797591,
+      "learning_rate": 0.0003381096202503321,
+      "loss": 2.9699087142944336,
+      "step": 11390
+    },
+    {
+      "epoch": 23.898664571877454,
+      "learning_rate": 0.00033796129402242193,
+      "loss": 2.868007850646973,
+      "step": 11400
+    },
+    {
+      "epoch": 23.91961246399581,
+      "learning_rate": 0.0003378131628323024,
+      "loss": 2.8257036209106445,
+      "step": 11410
+    },
+    {
+      "epoch": 23.940560356114165,
+      "learning_rate": 0.0003376652262529146,
+      "loss": 2.8747650146484376,
+      "step": 11420
+    },
+    {
+      "epoch": 23.961508248232523,
+      "learning_rate": 0.00033751748385850753,
+      "loss": 2.8721830368041994,
+      "step": 11430
+    },
+    {
+      "epoch": 23.982456140350877,
+      "learning_rate": 0.00033736993522463316,
+      "loss": 2.8383148193359373,
+      "step": 11440
+    },
+    {
+      "epoch": 24.00418957842367,
+      "learning_rate": 0.00033722257992814113,
+      "loss": 2.9513004302978514,
+      "step": 11450
+    },
+    {
+      "epoch": 24.025137470542028,
+      "learning_rate": 0.0003370754175471737,
+      "loss": 2.9344108581542967,
+      "step": 11460
+    },
+    {
+      "epoch": 24.04608536266038,
+      "learning_rate": 0.0003369284476611607,
+      "loss": 2.9109573364257812,
+      "step": 11470
+    },
+    {
+      "epoch": 24.06703325477874,
+      "learning_rate": 0.00033678166985081433,
+      "loss": 2.8260976791381838,
+      "step": 11480
+    },
+    {
+      "epoch": 24.087981146897093,
+      "learning_rate": 0.0003366350836981245,
+      "loss": 2.790972137451172,
+      "step": 11490
+    },
+    {
+      "epoch": 24.10892903901545,
+      "learning_rate": 0.0003364886887863534,
+      "loss": 2.8716499328613283,
+      "step": 11500
+    },
+    {
+      "epoch": 24.129876931133804,
+      "learning_rate": 0.0003363424847000309,
+      "loss": 2.8708847045898436,
+      "step": 11510
+    },
+    {
+      "epoch": 24.15082482325216,
+      "learning_rate": 0.0003361964710249494,
+      "loss": 2.8762466430664064,
+      "step": 11520
+    },
+    {
+      "epoch": 24.171772715370516,
+      "learning_rate": 0.00033605064734815865,
+      "loss": 2.8436599731445313,
+      "step": 11530
+    },
+    {
+      "epoch": 24.192720607488873,
+      "learning_rate": 0.0003359050132579615,
+      "loss": 2.885796546936035,
+      "step": 11540
+    },
+    {
+      "epoch": 24.213668499607227,
+      "learning_rate": 0.00033575956834390843,
+      "loss": 2.8459619522094726,
+      "step": 11550
+    },
+    {
+      "epoch": 24.234616391725584,
+      "learning_rate": 0.00033561431219679297,
+      "loss": 2.8883172988891603,
+      "step": 11560
+    },
+    {
+      "epoch": 24.25556428384394,
+      "learning_rate": 0.00033546924440864666,
+      "loss": 2.8706939697265623,
+      "step": 11570
+    },
+    {
+      "epoch": 24.276512175962292,
+      "learning_rate": 0.0003353243645727346,
+      "loss": 2.8278776168823243,
+      "step": 11580
+    },
+    {
+      "epoch": 24.29746006808065,
+      "learning_rate": 0.0003351796722835502,
+      "loss": 2.8596363067626953,
+      "step": 11590
+    },
+    {
+      "epoch": 24.318407960199004,
+      "learning_rate": 0.00033503516713681087,
+      "loss": 2.8937658309936523,
+      "step": 11600
+    },
+    {
+      "epoch": 24.33935585231736,
+      "learning_rate": 0.00033489084872945283,
+      "loss": 2.9479068756103515,
+      "step": 11610
+    },
+    {
+      "epoch": 24.360303744435715,
+      "learning_rate": 0.0003347467166596268,
+      "loss": 2.913376235961914,
+      "step": 11620
+    },
+    {
+      "epoch": 24.381251636554072,
+      "learning_rate": 0.0003346027705266929,
+      "loss": 2.7945356369018555,
+      "step": 11630
+    },
+    {
+      "epoch": 24.402199528672426,
+      "learning_rate": 0.0003344590099312164,
+      "loss": 2.908190155029297,
+      "step": 11640
+    },
+    {
+      "epoch": 24.423147420790784,
+      "learning_rate": 0.00033431543447496275,
+      "loss": 2.8602962493896484,
+      "step": 11650
+    },
+    {
+      "epoch": 24.444095312909138,
+      "learning_rate": 0.000334172043760893,
+      "loss": 2.8535890579223633,
+      "step": 11660
+    },
+    {
+      "epoch": 24.465043205027495,
+      "learning_rate": 0.0003340288373931593,
+      "loss": 2.8150957107543944,
+      "step": 11670
+    },
+    {
+      "epoch": 24.48599109714585,
+      "learning_rate": 0.0003338858149771002,
+      "loss": 2.847452735900879,
+      "step": 11680
+    },
+    {
+      "epoch": 24.506938989264206,
+      "learning_rate": 0.0003337429761192361,
+      "loss": 2.7802717208862306,
+      "step": 11690
+    },
+    {
+      "epoch": 24.52788688138256,
+      "learning_rate": 0.00033360032042726483,
+      "loss": 2.9678937911987306,
+      "step": 11700
+    },
+    {
+      "epoch": 24.548834773500918,
+      "learning_rate": 0.000333457847510057,
+      "loss": 2.8469560623168944,
+      "step": 11710
+    },
+    {
+      "epoch": 24.56978266561927,
+      "learning_rate": 0.0003333155569776514,
+      "loss": 2.916895866394043,
+      "step": 11720
+    },
+    {
+      "epoch": 24.59073055773763,
+      "learning_rate": 0.00033317344844125064,
+      "loss": 2.8457548141479494,
+      "step": 11730
+    },
+    {
+      "epoch": 24.611678449855983,
+      "learning_rate": 0.00033303152151321696,
+      "loss": 2.872743606567383,
+      "step": 11740
+    },
+    {
+      "epoch": 24.63262634197434,
+      "learning_rate": 0.00033288977580706714,
+      "loss": 2.890146255493164,
+      "step": 11750
+    },
+    {
+      "epoch": 24.653574234092694,
+      "learning_rate": 0.0003327482109374687,
+      "loss": 2.847947883605957,
+      "step": 11760
+    },
+    {
+      "epoch": 24.67452212621105,
+      "learning_rate": 0.00033260682652023517,
+      "loss": 2.9236717224121094,
+      "step": 11770
+    },
+    {
+      "epoch": 24.695470018329406,
+      "learning_rate": 0.0003324656221723217,
+      "loss": 2.9201459884643555,
+      "step": 11780
+    },
+    {
+      "epoch": 24.71641791044776,
+      "learning_rate": 0.000332324597511821,
+      "loss": 2.8557527542114256,
+      "step": 11790
+    },
+    {
+      "epoch": 24.737365802566117,
+      "learning_rate": 0.00033218375215795864,
+      "loss": 2.875984001159668,
+      "step": 11800
+    },
+    {
+      "epoch": 24.75831369468447,
+      "learning_rate": 0.00033204308573108897,
+      "loss": 2.830782890319824,
+      "step": 11810
+    },
+    {
+      "epoch": 24.77926158680283,
+      "learning_rate": 0.00033190259785269066,
+      "loss": 2.844138526916504,
+      "step": 11820
+    },
+    {
+      "epoch": 24.800209478921182,
+      "learning_rate": 0.0003317622881453626,
+      "loss": 2.8771383285522463,
+      "step": 11830
+    },
+    {
+      "epoch": 24.82115737103954,
+      "learning_rate": 0.0003316221562328194,
+      "loss": 2.863381767272949,
+      "step": 11840
+    },
+    {
+      "epoch": 24.842105263157894,
+      "learning_rate": 0.0003314822017398875,
+      "loss": 2.8934911727905273,
+      "step": 11850
+    },
+    {
+      "epoch": 24.86305315527625,
+      "learning_rate": 0.00033134242429250053,
+      "loss": 2.7928911209106446,
+      "step": 11860
+    },
+    {
+      "epoch": 24.884001047394605,
+      "learning_rate": 0.00033120282351769556,
+      "loss": 2.8646501541137694,
+      "step": 11870
+    },
+    {
+      "epoch": 24.904948939512963,
+      "learning_rate": 0.0003310633990436084,
+      "loss": 2.8573431015014648,
+      "step": 11880
+    },
+    {
+      "epoch": 24.925896831631317,
+      "learning_rate": 0.00033092415049947006,
+      "loss": 2.9596303939819335,
+      "step": 11890
+    },
+    {
+      "epoch": 24.946844723749674,
+      "learning_rate": 0.00033078507751560195,
+      "loss": 2.809922790527344,
+      "step": 11900
+    },
+    {
+      "epoch": 24.967792615868028,
+      "learning_rate": 0.00033064617972341235,
+      "loss": 2.829710578918457,
+      "step": 11910
+    },
+    {
+      "epoch": 24.988740507986385,
+      "learning_rate": 0.0003305074567553919,
+      "loss": 2.837497520446777,
+      "step": 11920
+    },
+    {
+      "epoch": 25.01047394605918,
+      "learning_rate": 0.0003303689082451096,
+      "loss": 2.998362922668457,
+      "step": 11930
+    },
+    {
+      "epoch": 25.031421838177533,
+      "learning_rate": 0.00033023053382720904,
+      "loss": 2.8903406143188475,
+      "step": 11940
+    },
+    {
+      "epoch": 25.05236973029589,
+      "learning_rate": 0.0003300923331374039,
+      "loss": 2.88183536529541,
+      "step": 11950
+    },
+    {
+      "epoch": 25.073317622414244,
+      "learning_rate": 0.00032995430581247417,
+      "loss": 2.8853179931640627,
+      "step": 11960
+    },
+    {
+      "epoch": 25.0942655145326,
+      "learning_rate": 0.0003298164514902622,
+      "loss": 2.8495412826538087,
+      "step": 11970
+    },
+    {
+      "epoch": 25.115213406650955,
+      "learning_rate": 0.0003296787698096686,
+      "loss": 2.8417972564697265,
+      "step": 11980
+    },
+    {
+      "epoch": 25.136161298769313,
+      "learning_rate": 0.0003295412604106482,
+      "loss": 2.8254583358764647,
+      "step": 11990
+    },
+    {
+      "epoch": 25.157109190887667,
+      "learning_rate": 0.00032940392293420614,
+      "loss": 2.8476821899414064,
+      "step": 12000
+    },
+    {
+      "epoch": 25.178057083006024,
+      "learning_rate": 0.00032926675702239425,
+      "loss": 2.8646284103393556,
+      "step": 12010
+    },
+    {
+      "epoch": 25.199004975124378,
+      "learning_rate": 0.00032912976231830646,
+      "loss": 2.8645925521850586,
+      "step": 12020
+    },
+    {
+      "epoch": 25.219952867242732,
+      "learning_rate": 0.0003289929384660757,
+      "loss": 2.9142387390136717,
+      "step": 12030
+    },
+    {
+      "epoch": 25.24090075936109,
+      "learning_rate": 0.0003288562851108693,
+      "loss": 2.911361312866211,
+      "step": 12040
+    },
+    {
+      "epoch": 25.261848651479443,
+      "learning_rate": 0.0003287198018988856,
+      "loss": 2.810334014892578,
+      "step": 12050
+    },
+    {
+      "epoch": 25.2827965435978,
+      "learning_rate": 0.00032858348847734985,
+      "loss": 2.8691171646118163,
+      "step": 12060
+    },
+    {
+      "epoch": 25.303744435716155,
+      "learning_rate": 0.00032844734449451055,
+      "loss": 2.8615827560424805,
+      "step": 12070
+    },
+    {
+      "epoch": 25.324692327834512,
+      "learning_rate": 0.00032831136959963553,
+      "loss": 2.8075706481933596,
+      "step": 12080
+    },
+    {
+      "epoch": 25.345640219952866,
+      "learning_rate": 0.00032817556344300823,
+      "loss": 2.836076354980469,
+      "step": 12090
+    },
+    {
+      "epoch": 25.366588112071224,
+      "learning_rate": 0.0003280399256759237,
+      "loss": 2.873185729980469,
+      "step": 12100
+    },
+    {
+      "epoch": 25.387536004189577,
+      "learning_rate": 0.0003279044559506852,
+      "loss": 2.8835927963256838,
+      "step": 12110
+    },
+    {
+      "epoch": 25.408483896307935,
+      "learning_rate": 0.0003277691539206003,
+      "loss": 2.884838676452637,
+      "step": 12120
+    },
+    {
+      "epoch": 25.42943178842629,
+      "learning_rate": 0.0003276340192399769,
+      "loss": 2.9353681564331056,
+      "step": 12130
+    },
+    {
+      "epoch": 25.450379680544646,
+      "learning_rate": 0.00032749905156412,
+      "loss": 2.87127571105957,
+      "step": 12140
+    },
+    {
+      "epoch": 25.471327572663,
+      "learning_rate": 0.0003273642505493275,
+      "loss": 2.848041534423828,
+      "step": 12150
+    },
+    {
+      "epoch": 25.492275464781358,
+      "learning_rate": 0.0003272296158528871,
+      "loss": 2.8736820220947266,
+      "step": 12160
+    },
+    {
+      "epoch": 25.51322335689971,
+      "learning_rate": 0.000327095147133072,
+      "loss": 2.889766502380371,
+      "step": 12170
+    },
+    {
+      "epoch": 25.53417124901807,
+      "learning_rate": 0.00032696084404913777,
+      "loss": 2.8456445693969727,
+      "step": 12180
+    },
+    {
+      "epoch": 25.555119141136423,
+      "learning_rate": 0.00032682670626131837,
+      "loss": 2.8694175720214843,
+      "step": 12190
+    },
+    {
+      "epoch": 25.57606703325478,
+      "learning_rate": 0.0003266927334308229,
+      "loss": 2.863827705383301,
+      "step": 12200
+    },
+    {
+      "epoch": 25.597014925373134,
+      "learning_rate": 0.0003265589252198317,
+      "loss": 2.8949514389038087,
+      "step": 12210
+    },
+    {
+      "epoch": 25.617962817491488,
+      "learning_rate": 0.0003264252812914928,
+      "loss": 2.870989990234375,
+      "step": 12220
+    },
+    {
+      "epoch": 25.638910709609846,
+      "learning_rate": 0.0003262918013099186,
+      "loss": 2.8301280975341796,
+      "step": 12230
+    },
+    {
+      "epoch": 25.6598586017282,
+      "learning_rate": 0.00032615848494018204,
+      "loss": 2.7910818099975585,
+      "step": 12240
+    },
+    {
+      "epoch": 25.680806493846557,
+      "learning_rate": 0.0003260253318483131,
+      "loss": 2.8830698013305662,
+      "step": 12250
+    },
+    {
+      "epoch": 25.70175438596491,
+      "learning_rate": 0.0003258923417012957,
+      "loss": 2.886226844787598,
+      "step": 12260
+    },
+    {
+      "epoch": 25.72270227808327,
+      "learning_rate": 0.00032575951416706354,
+      "loss": 2.9646997451782227,
+      "step": 12270
+    },
+    {
+      "epoch": 25.743650170201622,
+      "learning_rate": 0.0003256268489144972,
+      "loss": 2.896713066101074,
+      "step": 12280
+    },
+    {
+      "epoch": 25.76459806231998,
+      "learning_rate": 0.0003254943456134202,
+      "loss": 2.8680368423461915,
+      "step": 12290
+    },
+    {
+      "epoch": 25.785545954438334,
+      "learning_rate": 0.0003253620039345959,
+      "loss": 2.866026496887207,
+      "step": 12300
+    },
+    {
+      "epoch": 25.80649384655669,
+      "learning_rate": 0.0003252298235497241,
+      "loss": 2.862067985534668,
+      "step": 12310
+    },
+    {
+      "epoch": 25.827441738675045,
+      "learning_rate": 0.0003250978041314371,
+      "loss": 2.8973188400268555,
+      "step": 12320
+    },
+    {
+      "epoch": 25.848389630793402,
+      "learning_rate": 0.000324965945353297,
+      "loss": 2.9389106750488283,
+      "step": 12330
+    },
+    {
+      "epoch": 25.869337522911756,
+      "learning_rate": 0.0003248342468897917,
+      "loss": 2.9147424697875977,
+      "step": 12340
+    },
+    {
+      "epoch": 25.890285415030114,
+      "learning_rate": 0.00032470270841633195,
+      "loss": 2.894465446472168,
+      "step": 12350
+    },
+    {
+      "epoch": 25.911233307148468,
+      "learning_rate": 0.00032457132960924783,
+      "loss": 2.9301485061645507,
+      "step": 12360
+    },
+    {
+      "epoch": 25.932181199266825,
+      "learning_rate": 0.00032444011014578535,
+      "loss": 2.8576644897460937,
+      "step": 12370
+    },
+    {
+      "epoch": 25.95312909138518,
+      "learning_rate": 0.00032430904970410314,
+      "loss": 2.836701202392578,
+      "step": 12380
+    },
+    {
+      "epoch": 25.974076983503537,
+      "learning_rate": 0.0003241781479632693,
+      "loss": 2.8457481384277346,
+      "step": 12390
+    },
+    {
+      "epoch": 25.99502487562189,
+      "learning_rate": 0.0003240474046032579,
+      "loss": 2.829239082336426,
+      "step": 12400
+    },
+    {
+      "epoch": 26.016758313694684,
+      "learning_rate": 0.00032391681930494566,
+      "loss": 3.0122323989868165,
+      "step": 12410
+    },
+    {
+      "epoch": 26.03770620581304,
+      "learning_rate": 0.000323786391750109,
+      "loss": 2.8899608612060548,
+      "step": 12420
+    },
+    {
+      "epoch": 26.058654097931395,
+      "learning_rate": 0.0003236561216214202,
+      "loss": 2.9380813598632813,
+      "step": 12430
+    },
+    {
+      "epoch": 26.079601990049753,
+      "learning_rate": 0.000323526008602445,
+      "loss": 2.9364286422729493,
+      "step": 12440
+    },
+    {
+      "epoch": 26.100549882168107,
+      "learning_rate": 0.0003233960523776387,
+      "loss": 2.8298776626586912,
+      "step": 12450
+    },
+    {
+      "epoch": 26.121497774286464,
+      "learning_rate": 0.0003232662526323429,
+      "loss": 2.868173027038574,
+      "step": 12460
+    },
+    {
+      "epoch": 26.142445666404818,
+      "learning_rate": 0.0003231366090527828,
+      "loss": 2.8364093780517576,
+      "step": 12470
+    },
+    {
+      "epoch": 26.163393558523175,
+      "learning_rate": 0.00032300712132606366,
+      "loss": 2.917738342285156,
+      "step": 12480
+    },
+    {
+      "epoch": 26.18434145064153,
+      "learning_rate": 0.0003228777891401678,
+      "loss": 2.8115827560424806,
+      "step": 12490
+    },
+    {
+      "epoch": 26.205289342759883,
+      "learning_rate": 0.0003227486121839514,
+      "loss": 2.8544151306152346,
+      "step": 12500
+    },
+    {
+      "epoch": 26.22623723487824,
+      "learning_rate": 0.00032261959014714107,
+      "loss": 2.890985870361328,
+      "step": 12510
+    },
+    {
+      "epoch": 26.247185126996595,
+      "learning_rate": 0.0003224907227203312,
+      "loss": 2.8269269943237303,
+      "step": 12520
+    },
+    {
+      "epoch": 26.268133019114952,
+      "learning_rate": 0.0003223620095949806,
+      "loss": 2.8392301559448243,
+      "step": 12530
+    },
+    {
+      "epoch": 26.289080911233306,
+      "learning_rate": 0.00032223345046340936,
+      "loss": 2.8283065795898437,
+      "step": 12540
+    },
+    {
+      "epoch": 26.310028803351663,
+      "learning_rate": 0.00032210504501879576,
+      "loss": 2.9033248901367186,
+      "step": 12550
+    },
+    {
+      "epoch": 26.330976695470017,
+      "learning_rate": 0.0003219767929551733,
+      "loss": 2.8192907333374024,
+      "step": 12560
+    },
+    {
+      "epoch": 26.351924587588375,
+      "learning_rate": 0.00032184869396742754,
+      "loss": 2.8758308410644533,
+      "step": 12570
+    },
+    {
+      "epoch": 26.37287247970673,
+      "learning_rate": 0.00032172074775129323,
+      "loss": 2.8491661071777346,
+      "step": 12580
+    },
+    {
+      "epoch": 26.393820371825086,
+      "learning_rate": 0.00032159295400335114,
+      "loss": 2.862008285522461,
+      "step": 12590
+    },
+    {
+      "epoch": 26.41476826394344,
+      "learning_rate": 0.00032146531242102476,
+      "loss": 2.854539489746094,
+      "step": 12600
+    },
+    {
+      "epoch": 26.435716156061797,
+      "learning_rate": 0.0003213378227025779,
+      "loss": 2.9059074401855467,
+      "step": 12610
+    },
+    {
+      "epoch": 26.45666404818015,
+      "learning_rate": 0.00032121048454711114,
+      "loss": 2.8347517013549806,
+      "step": 12620
+    },
+    {
+      "epoch": 26.47761194029851,
+      "learning_rate": 0.00032108329765455926,
+      "loss": 2.8621740341186523,
+      "step": 12630
+    },
+    {
+      "epoch": 26.498559832416863,
+      "learning_rate": 0.00032095626172568784,
+      "loss": 2.8287914276123045,
+      "step": 12640
+    },
+    {
+      "epoch": 26.51950772453522,
+      "learning_rate": 0.00032082937646209084,
+      "loss": 2.8201780319213867,
+      "step": 12650
+    },
+    {
+      "epoch": 26.540455616653574,
+      "learning_rate": 0.0003207026415661871,
+      "loss": 2.853387451171875,
+      "step": 12660
+    },
+    {
+      "epoch": 26.56140350877193,
+      "learning_rate": 0.0003205760567412178,
+      "loss": 2.8255029678344727,
+      "step": 12670
+    },
+    {
+      "epoch": 26.582351400890285,
+      "learning_rate": 0.00032044962169124335,
+      "loss": 2.8133966445922853,
+      "step": 12680
+    },
+    {
+      "epoch": 26.60329929300864,
+      "learning_rate": 0.0003203233361211406,
+      "loss": 2.8209064483642576,
+      "step": 12690
+    },
+    {
+      "epoch": 26.624247185126997,
+      "learning_rate": 0.00032019719973659996,
+      "loss": 2.839722442626953,
+      "step": 12700
+    },
+    {
+      "epoch": 26.64519507724535,
+      "learning_rate": 0.00032007121224412224,
+      "loss": 2.8414018630981444,
+      "step": 12710
+    },
+    {
+      "epoch": 26.666142969363708,
+      "learning_rate": 0.0003199453733510162,
+      "loss": 2.8677789688110353,
+      "step": 12720
+    },
+    {
+      "epoch": 26.687090861482062,
+      "learning_rate": 0.00031981968276539543,
+      "loss": 2.9177148818969725,
+      "step": 12730
+    },
+    {
+      "epoch": 26.70803875360042,
+      "learning_rate": 0.0003196941401961754,
+      "loss": 2.8555475234985352,
+      "step": 12740
+    },
+    {
+      "epoch": 26.728986645718773,
+      "learning_rate": 0.000319568745353071,
+      "loss": 2.8636154174804687,
+      "step": 12750
+    },
+    {
+      "epoch": 26.74993453783713,
+      "learning_rate": 0.0003194434979465935,
+      "loss": 2.810639190673828,
+      "step": 12760
+    },
+    {
+      "epoch": 26.770882429955485,
+      "learning_rate": 0.0003193183976880476,
+      "loss": 2.9356500625610353,
+      "step": 12770
+    },
+    {
+      "epoch": 26.791830322073842,
+      "learning_rate": 0.00031919344428952895,
+      "loss": 2.848637580871582,
+      "step": 12780
+    },
+    {
+      "epoch": 26.812778214192196,
+      "learning_rate": 0.0003190686374639211,
+      "loss": 2.8234004974365234,
+      "step": 12790
+    },
+    {
+      "epoch": 26.833726106310554,
+      "learning_rate": 0.00031894397692489295,
+      "loss": 2.8002485275268554,
+      "step": 12800
+    },
+    {
+      "epoch": 26.854673998428908,
+      "learning_rate": 0.0003188194623868958,
+      "loss": 2.841193199157715,
+      "step": 12810
+    },
+    {
+      "epoch": 26.875621890547265,
+      "learning_rate": 0.00031869509356516063,
+      "loss": 2.8377119064331056,
+      "step": 12820
+    },
+    {
+      "epoch": 26.89656978266562,
+      "learning_rate": 0.00031857087017569556,
+      "loss": 2.797208786010742,
+      "step": 12830
+    },
+    {
+      "epoch": 26.917517674783976,
+      "learning_rate": 0.0003184467919352828,
+      "loss": 2.778369140625,
+      "step": 12840
+    },
+    {
+      "epoch": 26.93846556690233,
+      "learning_rate": 0.0003183228585614763,
+      "loss": 2.8303714752197267,
+      "step": 12850
+    },
+    {
+      "epoch": 26.959413459020688,
+      "learning_rate": 0.0003181990697725988,
+      "loss": 2.805090141296387,
+      "step": 12860
+    },
+    {
+      "epoch": 26.98036135113904,
+      "learning_rate": 0.0003180754252877392,
+      "loss": 2.7620264053344727,
+      "step": 12870
+    },
+    {
+      "epoch": 27.002094789211835,
+      "learning_rate": 0.0003179519248267498,
+      "loss": 2.9018489837646486,
+      "step": 12880
+    },
+    {
+      "epoch": 27.023042681330192,
+      "learning_rate": 0.000317828568110244,
+      "loss": 2.8413219451904297,
+      "step": 12890
+    },
+    {
+      "epoch": 27.043990573448546,
+      "learning_rate": 0.000317705354859593,
+      "loss": 2.868427276611328,
+      "step": 12900
+    },
+    {
+      "epoch": 27.064938465566904,
+      "learning_rate": 0.0003175822847969239,
+      "loss": 2.84520263671875,
+      "step": 12910
+    },
+    {
+      "epoch": 27.085886357685258,
+      "learning_rate": 0.00031745935764511645,
+      "loss": 2.865756607055664,
+      "step": 12920
+    },
+    {
+      "epoch": 27.106834249803615,
+      "learning_rate": 0.0003173365731278007,
+      "loss": 2.8851186752319338,
+      "step": 12930
+    },
+    {
+      "epoch": 27.12778214192197,
+      "learning_rate": 0.00031721393096935445,
+      "loss": 2.8631362915039062,
+      "step": 12940
+    },
+    {
+      "epoch": 27.148730034040323,
+      "learning_rate": 0.00031709143089490063,
+      "loss": 2.8974273681640623,
+      "step": 12950
+    },
+    {
+      "epoch": 27.16967792615868,
+      "learning_rate": 0.00031696907263030445,
+      "loss": 2.8190950393676757,
+      "step": 12960
+    },
+    {
+      "epoch": 27.190625818277034,
+      "learning_rate": 0.00031684685590217115,
+      "loss": 2.861093521118164,
+      "step": 12970
+    },
+    {
+      "epoch": 27.211573710395392,
+      "learning_rate": 0.00031672478043784336,
+      "loss": 2.925172233581543,
+      "step": 12980
+    },
+    {
+      "epoch": 27.232521602513746,
+      "learning_rate": 0.0003166028459653984,
+      "loss": 2.7551206588745116,
+      "step": 12990
+    },
+    {
+      "epoch": 27.253469494632103,
+      "learning_rate": 0.0003164810522136458,
+      "loss": 2.8190824508666994,
+      "step": 13000
+    },
+    {
+      "epoch": 27.274417386750457,
+      "learning_rate": 0.0003163593989121249,
+      "loss": 2.884243965148926,
+      "step": 13010
+    },
+    {
+      "epoch": 27.295365278868815,
+      "learning_rate": 0.0003162378857911022,
+      "loss": 2.831955909729004,
+      "step": 13020
+    },
+    {
+      "epoch": 27.31631317098717,
+      "learning_rate": 0.00031611651258156884,
+      "loss": 2.891588020324707,
+      "step": 13030
+    },
+    {
+      "epoch": 27.337261063105526,
+      "learning_rate": 0.0003159952790152381,
+      "loss": 2.8689960479736327,
+      "step": 13040
+    },
+    {
+      "epoch": 27.35820895522388,
+      "learning_rate": 0.0003158741848245431,
+      "loss": 2.844234085083008,
+      "step": 13050
+    },
+    {
+      "epoch": 27.379156847342237,
+      "learning_rate": 0.0003157532297426339,
+      "loss": 2.783745765686035,
+      "step": 13060
+    },
+    {
+      "epoch": 27.40010473946059,
+      "learning_rate": 0.00031563241350337546,
+      "loss": 2.85959415435791,
+      "step": 13070
+    },
+    {
+      "epoch": 27.42105263157895,
+      "learning_rate": 0.00031551173584134514,
+      "loss": 2.828862762451172,
+      "step": 13080
+    },
+    {
+      "epoch": 27.442000523697303,
+      "learning_rate": 0.0003153911964918298,
+      "loss": 2.8126575469970705,
+      "step": 13090
+    },
+    {
+      "epoch": 27.46294841581566,
+      "learning_rate": 0.0003152707951908239,
+      "loss": 2.8336280822753905,
+      "step": 13100
+    },
+    {
+      "epoch": 27.483896307934014,
+      "learning_rate": 0.0003151505316750269,
+      "loss": 2.8542291641235353,
+      "step": 13110
+    },
+    {
+      "epoch": 27.50484420005237,
+      "learning_rate": 0.0003150304056818405,
+      "loss": 2.8555719375610353,
+      "step": 13120
+    },
+    {
+      "epoch": 27.525792092170725,
+      "learning_rate": 0.00031491041694936697,
+      "loss": 2.8440032958984376,
+      "step": 13130
+    },
+    {
+      "epoch": 27.54673998428908,
+      "learning_rate": 0.000314790565216406,
+      "loss": 2.899538040161133,
+      "step": 13140
+    },
+    {
+      "epoch": 27.567687876407437,
+      "learning_rate": 0.0003146708502224526,
+      "loss": 2.823881149291992,
+      "step": 13150
+    },
+    {
+      "epoch": 27.58863576852579,
+      "learning_rate": 0.0003145512717076948,
+      "loss": 2.8198898315429686,
+      "step": 13160
+    },
+    {
+      "epoch": 27.609583660644148,
+      "learning_rate": 0.00031443182941301147,
+      "loss": 2.8212156295776367,
+      "step": 13170
+    },
+    {
+      "epoch": 27.630531552762502,
+      "learning_rate": 0.0003143125230799694,
+      "loss": 2.7753381729125977,
+      "step": 13180
+    },
+    {
+      "epoch": 27.65147944488086,
+      "learning_rate": 0.00031419335245082134,
+      "loss": 2.812895393371582,
+      "step": 13190
+    },
+    {
+      "epoch": 27.672427336999213,
+      "learning_rate": 0.00031407431726850375,
+      "loss": 2.8747041702270506,
+      "step": 13200
+    },
+    {
+      "epoch": 27.69337522911757,
+      "learning_rate": 0.00031395541727663413,
+      "loss": 2.8663089752197264,
+      "step": 13210
+    },
+    {
+      "epoch": 27.714323121235925,
+      "learning_rate": 0.0003138366522195088,
+      "loss": 2.8993961334228517,
+      "step": 13220
+    },
+    {
+      "epoch": 27.735271013354282,
+      "learning_rate": 0.0003137180218421011,
+      "loss": 2.9394744873046874,
+      "step": 13230
+    },
+    {
+      "epoch": 27.756218905472636,
+      "learning_rate": 0.0003135995258900582,
+      "loss": 2.8471282958984374,
+      "step": 13240
+    },
+    {
+      "epoch": 27.777166797590993,
+      "learning_rate": 0.0003134811641096994,
+      "loss": 2.7851446151733397,
+      "step": 13250
+    },
+    {
+      "epoch": 27.798114689709347,
+      "learning_rate": 0.00031336293624801393,
+      "loss": 2.819938850402832,
+      "step": 13260
+    },
+    {
+      "epoch": 27.819062581827705,
+      "learning_rate": 0.00031324484205265824,
+      "loss": 2.8013900756835937,
+      "step": 13270
+    },
+    {
+      "epoch": 27.84001047394606,
+      "learning_rate": 0.000313126881271954,
+      "loss": 2.850057601928711,
+      "step": 13280
+    },
+    {
+      "epoch": 27.860958366064416,
+      "learning_rate": 0.0003130090536548859,
+      "loss": 2.7631250381469727,
+      "step": 13290
+    },
+    {
+      "epoch": 27.88190625818277,
+      "learning_rate": 0.00031289135895109924,
+      "loss": 2.8360868453979493,
+      "step": 13300
+    },
+    {
+      "epoch": 27.902854150301128,
+      "learning_rate": 0.00031277379691089786,
+      "loss": 2.804159927368164,
+      "step": 13310
+    },
+    {
+      "epoch": 27.92380204241948,
+      "learning_rate": 0.00031265636728524174,
+      "loss": 2.8401294708251954,
+      "step": 13320
+    },
+    {
+      "epoch": 27.94474993453784,
+      "learning_rate": 0.000312539069825745,
+      "loss": 2.850791168212891,
+      "step": 13330
+    },
+    {
+      "epoch": 27.965697826656193,
+      "learning_rate": 0.00031242190428467325,
+      "loss": 2.862323188781738,
+      "step": 13340
+    },
+    {
+      "epoch": 27.986645718774547,
+      "learning_rate": 0.0003123048704149423,
+      "loss": 2.8848134994506838,
+      "step": 13350
+    },
+    {
+      "epoch": 28.008379156847344,
+      "learning_rate": 0.0003121879679701147,
+      "loss": 2.9553651809692383,
+      "step": 13360
+    },
+    {
+      "epoch": 28.029327048965698,
+      "learning_rate": 0.00031207119670439884,
+      "loss": 2.893220138549805,
+      "step": 13370
+    },
+    {
+      "epoch": 28.050274941084055,
+      "learning_rate": 0.00031195455637264574,
+      "loss": 2.8204929351806642,
+      "step": 13380
+    },
+    {
+      "epoch": 28.07122283320241,
+      "learning_rate": 0.00031183804673034756,
+      "loss": 2.8356761932373047,
+      "step": 13390
+    },
+    {
+      "epoch": 28.092170725320763,
+      "learning_rate": 0.0003117216675336353,
+      "loss": 2.800448989868164,
+      "step": 13400
+    },
+    {
+      "epoch": 28.11311861743912,
+      "learning_rate": 0.00031160541853927627,
+      "loss": 2.8977182388305662,
+      "step": 13410
+    },
+    {
+      "epoch": 28.134066509557474,
+      "learning_rate": 0.0003114892995046725,
+      "loss": 2.8017560958862306,
+      "step": 13420
+    },
+    {
+      "epoch": 28.15501440167583,
+      "learning_rate": 0.00031137331018785835,
+      "loss": 2.7457189559936523,
+      "step": 13430
+    },
+    {
+      "epoch": 28.175962293794186,
+      "learning_rate": 0.00031125745034749834,
+      "loss": 2.8290485382080077,
+      "step": 13440
+    },
+    {
+      "epoch": 28.196910185912543,
+      "learning_rate": 0.00031114171974288516,
+      "loss": 2.8317813873291016,
+      "step": 13450
+    },
+    {
+      "epoch": 28.217858078030897,
+      "learning_rate": 0.00031102611813393753,
+      "loss": 2.7843399047851562,
+      "step": 13460
+    },
+    {
+      "epoch": 28.238805970149254,
+      "learning_rate": 0.0003109106452811981,
+      "loss": 2.8257192611694335,
+      "step": 13470
+    },
+    {
+      "epoch": 28.25975386226761,
+      "learning_rate": 0.00031079530094583135,
+      "loss": 2.8432809829711916,
+      "step": 13480
+    },
+    {
+      "epoch": 28.280701754385966,
+      "learning_rate": 0.0003106800848896216,
+      "loss": 2.882096862792969,
+      "step": 13490
+    },
+    {
+      "epoch": 28.30164964650432,
+      "learning_rate": 0.0003105649968749708,
+      "loss": 2.87137508392334,
+      "step": 13500
+    },
+    {
+      "epoch": 28.322597538622677,
+      "learning_rate": 0.0003104500366648965,
+      "loss": 2.8303447723388673,
+      "step": 13510
+    },
+    {
+      "epoch": 28.34354543074103,
+      "learning_rate": 0.0003103352040230302,
+      "loss": 2.955478477478027,
+      "step": 13520
+    },
+    {
+      "epoch": 28.36449332285939,
+      "learning_rate": 0.00031022049871361445,
+      "loss": 2.7974782943725587,
+      "step": 13530
+    },
+    {
+      "epoch": 28.385441214977742,
+      "learning_rate": 0.0003101059205015017,
+      "loss": 2.882868766784668,
+      "step": 13540
+    },
+    {
+      "epoch": 28.4063891070961,
+      "learning_rate": 0.0003099914691521518,
+      "loss": 2.9435708999633787,
+      "step": 13550
+    },
+    {
+      "epoch": 28.427336999214454,
+      "learning_rate": 0.00030987714443163,
+      "loss": 2.8506664276123046,
+      "step": 13560
+    },
+    {
+      "epoch": 28.44828489133281,
+      "learning_rate": 0.00030976294610660516,
+      "loss": 2.8492944717407225,
+      "step": 13570
+    },
+    {
+      "epoch": 28.469232783451165,
+      "learning_rate": 0.00030964887394434754,
+      "loss": 2.8658618927001953,
+      "step": 13580
+    },
+    {
+      "epoch": 28.49018067556952,
+      "learning_rate": 0.000309534927712727,
+      "loss": 2.8701282501220704,
+      "step": 13590
+    },
+    {
+      "epoch": 28.511128567687877,
+      "learning_rate": 0.0003094211071802107,
+      "loss": 2.8161798477172852,
+      "step": 13600
+    },
+    {
+      "epoch": 28.53207645980623,
+      "learning_rate": 0.00030930741211586155,
+      "loss": 2.768409538269043,
+      "step": 13610
+    },
+    {
+      "epoch": 28.553024351924588,
+      "learning_rate": 0.0003091938422893358,
+      "loss": 2.84487361907959,
+      "step": 13620
+    },
+    {
+      "epoch": 28.573972244042942,
+      "learning_rate": 0.00030908039747088155,
+      "loss": 2.8081539154052733,
+      "step": 13630
+    },
+    {
+      "epoch": 28.5949201361613,
+      "learning_rate": 0.00030896707743133635,
+      "loss": 2.8049062728881835,
+      "step": 13640
+    },
+    {
+      "epoch": 28.615868028279653,
+      "learning_rate": 0.0003088538819421255,
+      "loss": 2.8450254440307616,
+      "step": 13650
+    },
+    {
+      "epoch": 28.63681592039801,
+      "learning_rate": 0.00030874081077526003,
+      "loss": 2.8079158782958986,
+      "step": 13660
+    },
+    {
+      "epoch": 28.657763812516365,
+      "learning_rate": 0.00030862786370333505,
+      "loss": 2.8801244735717773,
+      "step": 13670
+    },
+    {
+      "epoch": 28.678711704634722,
+      "learning_rate": 0.00030851504049952727,
+      "loss": 2.8432153701782226,
+      "step": 13680
+    },
+    {
+      "epoch": 28.699659596753076,
+      "learning_rate": 0.00030840234093759347,
+      "loss": 2.913180923461914,
+      "step": 13690
+    },
+    {
+      "epoch": 28.720607488871433,
+      "learning_rate": 0.0003082897647918688,
+      "loss": 2.857924461364746,
+      "step": 13700
+    },
+    {
+      "epoch": 28.741555380989787,
+      "learning_rate": 0.0003081773118372642,
+      "loss": 2.7912296295166015,
+      "step": 13710
+    },
+    {
+      "epoch": 28.762503273108145,
+      "learning_rate": 0.00030806498184926523,
+      "loss": 2.8504261016845702,
+      "step": 13720
+    },
+    {
+      "epoch": 28.7834511652265,
+      "learning_rate": 0.0003079527746039298,
+      "loss": 2.8378028869628906,
+      "step": 13730
+    },
+    {
+      "epoch": 28.804399057344856,
+      "learning_rate": 0.00030784068987788624,
+      "loss": 2.803904914855957,
+      "step": 13740
+    },
+    {
+      "epoch": 28.82534694946321,
+      "learning_rate": 0.00030772872744833183,
+      "loss": 2.839299774169922,
+      "step": 13750
+    },
+    {
+      "epoch": 28.846294841581567,
+      "learning_rate": 0.00030761688709303036,
+      "loss": 2.7884681701660154,
+      "step": 13760
+    },
+    {
+      "epoch": 28.86724273369992,
+      "learning_rate": 0.0003075051685903109,
+      "loss": 2.88138427734375,
+      "step": 13770
+    },
+    {
+      "epoch": 28.888190625818275,
+      "learning_rate": 0.00030739357171906536,
+      "loss": 2.81328125,
+      "step": 13780
+    },
+    {
+      "epoch": 28.909138517936633,
+      "learning_rate": 0.0003072820962587471,
+      "loss": 2.8141046524047852,
+      "step": 13790
+    },
+    {
+      "epoch": 28.930086410054987,
+      "learning_rate": 0.00030717074198936904,
+      "loss": 2.7672204971313477,
+      "step": 13800
+    },
+    {
+      "epoch": 28.951034302173344,
+      "learning_rate": 0.0003070595086915015,
+      "loss": 2.8781991958618165,
+      "step": 13810
+    },
+    {
+      "epoch": 28.971982194291698,
+      "learning_rate": 0.00030694839614627076,
+      "loss": 2.7781099319458007,
+      "step": 13820
+    },
+    {
+      "epoch": 28.992930086410055,
+      "learning_rate": 0.0003068374041353571,
+      "loss": 2.879766082763672,
+      "step": 13830
+    },
+    {
+      "epoch": 29.01466352448285,
+      "learning_rate": 0.000306726532440993,
+      "loss": 2.8991397857666015,
+      "step": 13840
+    },
+    {
+      "epoch": 29.035611416601206,
+      "learning_rate": 0.0003066157808459613,
+      "loss": 2.8512521743774415,
+      "step": 13850
+    },
+    {
+      "epoch": 29.05655930871956,
+      "learning_rate": 0.0003065051491335936,
+      "loss": 2.833390235900879,
+      "step": 13860
+    },
+    {
+      "epoch": 29.077507200837914,
+      "learning_rate": 0.0003063946370877681,
+      "loss": 2.8554765701293947,
+      "step": 13870
+    },
+    {
+      "epoch": 29.09845509295627,
+      "learning_rate": 0.0003062842444929085,
+      "loss": 2.7805418014526366,
+      "step": 13880
+    },
+    {
+      "epoch": 29.119402985074625,
+      "learning_rate": 0.00030617397113398125,
+      "loss": 3.09820671081543,
+      "step": 13890
+    },
+    {
+      "epoch": 29.140350877192983,
+      "learning_rate": 0.00030606381679649483,
+      "loss": 2.900446128845215,
+      "step": 13900
+    },
+    {
+      "epoch": 29.161298769311337,
+      "learning_rate": 0.00030595378126649727,
+      "loss": 2.852696418762207,
+      "step": 13910
+    },
+    {
+      "epoch": 29.182246661429694,
+      "learning_rate": 0.0003058438643305747,
+      "loss": 3.0798343658447265,
+      "step": 13920
+    },
+    {
+      "epoch": 29.203194553548048,
+      "learning_rate": 0.00030573406577584955,
+      "loss": 2.8329389572143553,
+      "step": 13930
+    },
+    {
+      "epoch": 29.224142445666406,
+      "learning_rate": 0.000305624385389979,
+      "loss": 2.7638198852539064,
+      "step": 13940
+    },
+    {
+      "epoch": 29.24509033778476,
+      "learning_rate": 0.0003055148229611527,
+      "loss": 2.7774702072143556,
+      "step": 13950
+    },
+    {
+      "epoch": 29.266038229903117,
+      "learning_rate": 0.00030540537827809176,
+      "loss": 2.884586524963379,
+      "step": 13960
+    },
+    {
+      "epoch": 29.28698612202147,
+      "learning_rate": 0.0003052960511300467,
+      "loss": 2.858045196533203,
+      "step": 13970
+    },
+    {
+      "epoch": 29.30793401413983,
+      "learning_rate": 0.0003051868413067956,
+      "loss": 2.8505125045776367,
+      "step": 13980
+    },
+    {
+      "epoch": 29.328881906258182,
+      "learning_rate": 0.00030507774859864277,
+      "loss": 2.840318298339844,
+      "step": 13990
+    },
+    {
+      "epoch": 29.34982979837654,
+      "learning_rate": 0.0003049687727964166,
+      "loss": 2.871793746948242,
+      "step": 14000
+    },
+    {
+      "epoch": 29.370777690494894,
+      "learning_rate": 0.00030485991369146834,
+      "loss": 2.814739990234375,
+      "step": 14010
+    },
+    {
+      "epoch": 29.39172558261325,
+      "learning_rate": 0.00030475117107567015,
+      "loss": 2.8241125106811524,
+      "step": 14020
+    },
+    {
+      "epoch": 29.412673474731605,
+      "learning_rate": 0.0003046425447414135,
+      "loss": 2.802973747253418,
+      "step": 14030
+    },
+    {
+      "epoch": 29.433621366849962,
+      "learning_rate": 0.0003045340344816073,
+      "loss": 2.829861068725586,
+      "step": 14040
+    },
+    {
+      "epoch": 29.454569258968316,
+      "learning_rate": 0.0003044256400896769,
+      "loss": 2.823344612121582,
+      "step": 14050
+    },
+    {
+      "epoch": 29.47551715108667,
+      "learning_rate": 0.0003043173613595614,
+      "loss": 2.811284065246582,
+      "step": 14060
+    },
+    {
+      "epoch": 29.496465043205028,
+      "learning_rate": 0.0003042091980857131,
+      "loss": 2.8590465545654298,
+      "step": 14070
+    },
+    {
+      "epoch": 29.51741293532338,
+      "learning_rate": 0.0003041011500630949,
+      "loss": 2.8229595184326173,
+      "step": 14080
+    },
+    {
+      "epoch": 29.53836082744174,
+      "learning_rate": 0.00030399321708717947,
+      "loss": 2.8343103408813475,
+      "step": 14090
+    },
+    {
+      "epoch": 29.559308719560093,
+      "learning_rate": 0.00030388539895394697,
+      "loss": 2.804738235473633,
+      "step": 14100
+    },
+    {
+      "epoch": 29.58025661167845,
+      "learning_rate": 0.00030377769545988394,
+      "loss": 2.8719600677490233,
+      "step": 14110
+    },
+    {
+      "epoch": 29.601204503796804,
+      "learning_rate": 0.00030367010640198143,
+      "loss": 2.777914810180664,
+      "step": 14120
+    },
+    {
+      "epoch": 29.622152395915162,
+      "learning_rate": 0.0003035626315777333,
+      "loss": 2.837109375,
+      "step": 14130
+    },
+    {
+      "epoch": 29.643100288033516,
+      "learning_rate": 0.00030345527078513493,
+      "loss": 2.8141595840454103,
+      "step": 14140
+    },
+    {
+      "epoch": 29.664048180151873,
+      "learning_rate": 0.0003033480238226813,
+      "loss": 2.8648092269897463,
+      "step": 14150
+    },
+    {
+      "epoch": 29.684996072270227,
+      "learning_rate": 0.0003032408904893656,
+      "loss": 2.7934087753295898,
+      "step": 14160
+    },
+    {
+      "epoch": 29.705943964388585,
+      "learning_rate": 0.00030313387058467756,
+      "loss": 2.834004783630371,
+      "step": 14170
+    },
+    {
+      "epoch": 29.72689185650694,
+      "learning_rate": 0.0003030269639086021,
+      "loss": 2.8099668502807615,
+      "step": 14180
+    },
+    {
+      "epoch": 29.747839748625296,
+      "learning_rate": 0.0003029201702616173,
+      "loss": 2.830114555358887,
+      "step": 14190
+    },
+    {
+      "epoch": 29.76878764074365,
+      "learning_rate": 0.0003028134894446933,
+      "loss": 2.780957794189453,
+      "step": 14200
+    },
+    {
+      "epoch": 29.789735532862007,
+      "learning_rate": 0.00030270692125929034,
+      "loss": 2.829334831237793,
+      "step": 14210
+    },
+    {
+      "epoch": 29.81068342498036,
+      "learning_rate": 0.00030260046550735763,
+      "loss": 2.840847969055176,
+      "step": 14220
+    },
+    {
+      "epoch": 29.83163131709872,
+      "learning_rate": 0.0003024941219913316,
+      "loss": 2.855925369262695,
+      "step": 14230
+    },
+    {
+      "epoch": 29.852579209217073,
+      "learning_rate": 0.00030238789051413416,
+      "loss": 2.8478092193603515,
+      "step": 14240
+    },
+    {
+      "epoch": 29.873527101335426,
+      "learning_rate": 0.00030228177087917153,
+      "loss": 2.8140996932983398,
+      "step": 14250
+    },
+    {
+      "epoch": 29.894474993453784,
+      "learning_rate": 0.00030217576289033235,
+      "loss": 2.803069496154785,
+      "step": 14260
+    },
+    {
+      "epoch": 29.915422885572138,
+      "learning_rate": 0.00030206986635198654,
+      "loss": 2.7434965133666993,
+      "step": 14270
+    },
+    {
+      "epoch": 29.936370777690495,
+      "learning_rate": 0.00030196408106898356,
+      "loss": 2.859099006652832,
+      "step": 14280
+    },
+    {
+      "epoch": 29.95731866980885,
+      "learning_rate": 0.0003018584068466507,
+      "loss": 2.9088722229003907,
+      "step": 14290
+    },
+    {
+      "epoch": 29.978266561927207,
+      "learning_rate": 0.0003017528434907922,
+      "loss": 2.7880224227905273,
+      "step": 14300
+    },
+    {
+      "epoch": 29.99921445404556,
+      "learning_rate": 0.00030164739080768704,
+      "loss": 2.8113405227661135,
+      "step": 14310
+    },
+    {
+      "epoch": 30.020947892118354,
+      "learning_rate": 0.0003015420486040879,
+      "loss": 2.893621826171875,
+      "step": 14320
+    },
+    {
+      "epoch": 30.04189578423671,
+      "learning_rate": 0.00030143681668721935,
+      "loss": 2.8216567993164063,
+      "step": 14330
+    },
+    {
+      "epoch": 30.062843676355065,
+      "learning_rate": 0.00030133169486477694,
+      "loss": 2.8155281066894533,
+      "step": 14340
+    },
+    {
+      "epoch": 30.083791568473423,
+      "learning_rate": 0.0003012266829449249,
+      "loss": 2.8872468948364256,
+      "step": 14350
+    },
+    {
+      "epoch": 30.104739460591777,
+      "learning_rate": 0.00030112178073629544,
+      "loss": 2.820456886291504,
+      "step": 14360
+    },
+    {
+      "epoch": 30.125687352710134,
+      "learning_rate": 0.0003010169880479867,
+      "loss": 2.869482231140137,
+      "step": 14370
+    },
+    {
+      "epoch": 30.146635244828488,
+      "learning_rate": 0.0003009123046895618,
+      "loss": 2.8011972427368166,
+      "step": 14380
+    },
+    {
+      "epoch": 30.167583136946845,
+      "learning_rate": 0.00030080773047104687,
+      "loss": 2.8537342071533205,
+      "step": 14390
+    },
+    {
+      "epoch": 30.1885310290652,
+      "learning_rate": 0.0003007032652029301,
+      "loss": 2.808944892883301,
+      "step": 14400
+    },
+    {
+      "epoch": 30.209478921183557,
+      "learning_rate": 0.00030059890869615983,
+      "loss": 2.833651542663574,
+      "step": 14410
+    },
+    {
+      "epoch": 30.23042681330191,
+      "learning_rate": 0.0003004946607621435,
+      "loss": 2.860894203186035,
+      "step": 14420
+    },
+    {
+      "epoch": 30.251374705420268,
+      "learning_rate": 0.0003003905212127461,
+      "loss": 2.835972213745117,
+      "step": 14430
+    },
+    {
+      "epoch": 30.272322597538622,
+      "learning_rate": 0.00030028648986028843,
+      "loss": 2.857589912414551,
+      "step": 14440
+    },
+    {
+      "epoch": 30.29327048965698,
+      "learning_rate": 0.00030018256651754633,
+      "loss": 2.845281219482422,
+      "step": 14450
+    },
+    {
+      "epoch": 30.314218381775333,
+      "learning_rate": 0.00030007875099774864,
+      "loss": 2.7922155380249025,
+      "step": 14460
+    },
+    {
+      "epoch": 30.33516627389369,
+      "learning_rate": 0.0002999750431145761,
+      "loss": 2.846644973754883,
+      "step": 14470
+    },
+    {
+      "epoch": 30.356114166012045,
+      "learning_rate": 0.0002998714426821599,
+      "loss": 2.83693904876709,
+      "step": 14480
+    },
+    {
+      "epoch": 30.377062058130402,
+      "learning_rate": 0.00029976794951508027,
+      "loss": 2.8328250885009765,
+      "step": 14490
+    },
+    {
+      "epoch": 30.398009950248756,
+      "learning_rate": 0.00029966456342836505,
+      "loss": 2.8287097930908205,
+      "step": 14500
+    },
+    {
+      "epoch": 30.41895784236711,
+      "learning_rate": 0.0002995612842374884,
+      "loss": 2.818513298034668,
+      "step": 14510
+    },
+    {
+      "epoch": 30.439905734485468,
+      "learning_rate": 0.0002994581117583693,
+      "loss": 2.804762077331543,
+      "step": 14520
+    },
+    {
+      "epoch": 30.46085362660382,
+      "learning_rate": 0.00029935504580737006,
+      "loss": 2.8560808181762694,
+      "step": 14530
+    },
+    {
+      "epoch": 30.48180151872218,
+      "learning_rate": 0.00029925208620129546,
+      "loss": 2.7961631774902345,
+      "step": 14540
+    },
+    {
+      "epoch": 30.502749410840533,
+      "learning_rate": 0.0002991492327573909,
+      "loss": 2.8281347274780275,
+      "step": 14550
+    },
+    {
+      "epoch": 30.52369730295889,
+      "learning_rate": 0.0002990464852933409,
+      "loss": 2.813071060180664,
+      "step": 14560
+    },
+    {
+      "epoch": 30.544645195077244,
+      "learning_rate": 0.0002989438436272684,
+      "loss": 2.765872001647949,
+      "step": 14570
+    },
+    {
+      "epoch": 30.5655930871956,
+      "learning_rate": 0.00029884130757773275,
+      "loss": 2.7835336685180665,
+      "step": 14580
+    },
+    {
+      "epoch": 30.586540979313956,
+      "learning_rate": 0.0002987388769637288,
+      "loss": 2.855548286437988,
+      "step": 14590
+    },
+    {
+      "epoch": 30.607488871432313,
+      "learning_rate": 0.00029863655160468534,
+      "loss": 2.804723358154297,
+      "step": 14600
+    },
+    {
+      "epoch": 30.628436763550667,
+      "learning_rate": 0.0002985343313204637,
+      "loss": 2.8737287521362305,
+      "step": 14610
+    },
+    {
+      "epoch": 30.649384655669024,
+      "learning_rate": 0.0002984322159313568,
+      "loss": 2.871350860595703,
+      "step": 14620
+    },
+    {
+      "epoch": 30.67033254778738,
+      "learning_rate": 0.00029833020525808714,
+      "loss": 2.780613327026367,
+      "step": 14630
+    },
+    {
+      "epoch": 30.691280439905736,
+      "learning_rate": 0.00029822829912180636,
+      "loss": 2.8216100692749024,
+      "step": 14640
+    },
+    {
+      "epoch": 30.71222833202409,
+      "learning_rate": 0.0002981264973440931,
+      "loss": 2.789328956604004,
+      "step": 14650
+    },
+    {
+      "epoch": 30.733176224142447,
+      "learning_rate": 0.00029802479974695223,
+      "loss": 2.7879051208496093,
+      "step": 14660
+    },
+    {
+      "epoch": 30.7541241162608,
+      "learning_rate": 0.00029792320615281337,
+      "loss": 2.762567710876465,
+      "step": 14670
+    },
+    {
+      "epoch": 30.77507200837916,
+      "learning_rate": 0.00029782171638452937,
+      "loss": 2.8410247802734374,
+      "step": 14680
+    },
+    {
+      "epoch": 30.796019900497512,
+      "learning_rate": 0.0002977203302653755,
+      "loss": 2.7910200119018556,
+      "step": 14690
+    },
+    {
+      "epoch": 30.81696779261587,
+      "learning_rate": 0.0002976190476190476,
+      "loss": 2.8463191986083984,
+      "step": 14700
+    },
+    {
+      "epoch": 30.837915684734224,
+      "learning_rate": 0.0002975178682696613,
+      "loss": 2.816401481628418,
+      "step": 14710
+    },
+    {
+      "epoch": 30.858863576852578,
+      "learning_rate": 0.0002974167920417504,
+      "loss": 2.850655746459961,
+      "step": 14720
+    },
+    {
+      "epoch": 30.879811468970935,
+      "learning_rate": 0.00029731581876026557,
+      "loss": 2.844277191162109,
+      "step": 14730
+    },
+    {
+      "epoch": 30.90075936108929,
+      "learning_rate": 0.00029721494825057357,
+      "loss": 2.8203685760498045,
+      "step": 14740
+    },
+    {
+      "epoch": 30.921707253207646,
+      "learning_rate": 0.00029711418033845523,
+      "loss": 2.848883628845215,
+      "step": 14750
+    },
+    {
+      "epoch": 30.942655145326,
+      "learning_rate": 0.0002970135148501047,
+      "loss": 2.7703632354736327,
+      "step": 14760
+    },
+    {
+      "epoch": 30.963603037444358,
+      "learning_rate": 0.00029691295161212816,
+      "loss": 2.8733938217163084,
+      "step": 14770
+    },
+    {
+      "epoch": 30.98455092956271,
+      "learning_rate": 0.0002968124904515423,
+      "loss": 2.8676376342773438,
+      "step": 14780
+    },
+    {
+      "epoch": 31.006284367635505,
+      "learning_rate": 0.00029671213119577346,
+      "loss": 2.8960426330566404,
+      "step": 14790
+    },
+    {
+      "epoch": 31.027232259753863,
+      "learning_rate": 0.00029661187367265593,
+      "loss": 2.8205034255981447,
+      "step": 14800
+    },
+    {
+      "epoch": 31.048180151872216,
+      "learning_rate": 0.0002965117177104311,
+      "loss": 2.8493398666381835,
+      "step": 14810
+    },
+    {
+      "epoch": 31.069128043990574,
+      "learning_rate": 0.0002964116631377459,
+      "loss": 2.808573913574219,
+      "step": 14820
+    },
+    {
+      "epoch": 31.090075936108928,
+      "learning_rate": 0.000296311709783652,
+      "loss": 2.770844078063965,
+      "step": 14830
+    },
+    {
+      "epoch": 31.111023828227285,
+      "learning_rate": 0.00029621185747760406,
+      "loss": 2.7819324493408204,
+      "step": 14840
+    },
+    {
+      "epoch": 31.13197172034564,
+      "learning_rate": 0.0002961121060494589,
+      "loss": 2.7976245880126953,
+      "step": 14850
+    },
+    {
+      "epoch": 31.152919612463997,
+      "learning_rate": 0.00029601245532947417,
+      "loss": 2.8540115356445312,
+      "step": 14860
+    },
+    {
+      "epoch": 31.17386750458235,
+      "learning_rate": 0.0002959129051483069,
+      "loss": 2.7655929565429687,
+      "step": 14870
+    },
+    {
+      "epoch": 31.194815396700708,
+      "learning_rate": 0.00029581345533701285,
+      "loss": 2.847081184387207,
+      "step": 14880
+    },
+    {
+      "epoch": 31.215763288819062,
+      "learning_rate": 0.0002957141057270448,
+      "loss": 2.82701416015625,
+      "step": 14890
+    },
+    {
+      "epoch": 31.23671118093742,
+      "learning_rate": 0.0002956148561502513,
+      "loss": 2.8076833724975585,
+      "step": 14900
+    },
+    {
+      "epoch": 31.257659073055773,
+      "learning_rate": 0.00029551570643887603,
+      "loss": 2.7729957580566404,
+      "step": 14910
+    },
+    {
+      "epoch": 31.27860696517413,
+      "learning_rate": 0.00029541665642555606,
+      "loss": 2.8175632476806642,
+      "step": 14920
+    },
+    {
+      "epoch": 31.299554857292485,
+      "learning_rate": 0.00029531770594332096,
+      "loss": 2.781933403015137,
+      "step": 14930
+    },
+    {
+      "epoch": 31.320502749410842,
+      "learning_rate": 0.0002952188548255915,
+      "loss": 2.826693534851074,
+      "step": 14940
+    },
+    {
+      "epoch": 31.341450641529196,
+      "learning_rate": 0.00029512010290617854,
+      "loss": 2.7952367782592775,
+      "step": 14950
+    },
+    {
+      "epoch": 31.36239853364755,
+      "learning_rate": 0.0002950214500192816,
+      "loss": 2.7863574981689454,
+      "step": 14960
+    },
+    {
+      "epoch": 31.383346425765907,
+      "learning_rate": 0.00029492289599948834,
+      "loss": 2.8061588287353514,
+      "step": 14970
+    },
+    {
+      "epoch": 31.40429431788426,
+      "learning_rate": 0.0002948244406817725,
+      "loss": 2.8176244735717773,
+      "step": 14980
+    },
+    {
+      "epoch": 31.42524221000262,
+      "learning_rate": 0.00029472608390149343,
+      "loss": 2.8314136505126952,
+      "step": 14990
+    },
+    {
+      "epoch": 31.446190102120973,
+      "learning_rate": 0.00029462782549439473,
+      "loss": 2.802597999572754,
+      "step": 15000
+    }
+  ],
+  "max_steps": 15000,
+  "num_train_epochs": 32,
+  "total_flos": 4124974225514526720,
+  "trial_name": null,
+  "trial_params": null
+}