diff --git "a/checkpoint-23217/trainer_state.json" "b/checkpoint-23217/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-23217/trainer_state.json"
@@ -0,0 +1,6999 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 23217,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 4.307188697936857e-05,
+      "learning_rate": 5e-09,
+      "loss": 15.2085,
+      "step": 1
+    },
+    {
+      "epoch": 0.0008614377395873713,
+      "learning_rate": 1e-07,
+      "loss": 13.4227,
+      "step": 20
+    },
+    {
+      "epoch": 0.0017228754791747427,
+      "learning_rate": 2e-07,
+      "loss": 12.2855,
+      "step": 40
+    },
+    {
+      "epoch": 0.002584313218762114,
+      "learning_rate": 3e-07,
+      "loss": 12.1044,
+      "step": 60
+    },
+    {
+      "epoch": 0.0034457509583494853,
+      "learning_rate": 4e-07,
+      "loss": 10.705,
+      "step": 80
+    },
+    {
+      "epoch": 0.004307188697936857,
+      "learning_rate": 5e-07,
+      "loss": 10.2652,
+      "step": 100
+    },
+    {
+      "epoch": 0.005168626437524228,
+      "learning_rate": 6e-07,
+      "loss": 8.8297,
+      "step": 120
+    },
+    {
+      "epoch": 0.0060300641771116,
+      "learning_rate": 7e-07,
+      "loss": 8.3154,
+      "step": 140
+    },
+    {
+      "epoch": 0.006891501916698971,
+      "learning_rate": 8e-07,
+      "loss": 7.722,
+      "step": 160
+    },
+    {
+      "epoch": 0.007752939656286342,
+      "learning_rate": 9e-07,
+      "loss": 7.6096,
+      "step": 180
+    },
+    {
+      "epoch": 0.008614377395873713,
+      "learning_rate": 1e-06,
+      "loss": 7.394,
+      "step": 200
+    },
+    {
+      "epoch": 0.009475815135461085,
+      "learning_rate": 9.99951518024521e-07,
+      "loss": 7.0227,
+      "step": 220
+    },
+    {
+      "epoch": 0.010337252875048455,
+      "learning_rate": 9.999030360490423e-07,
+      "loss": 7.7254,
+      "step": 240
+    },
+    {
+      "epoch": 0.011198690614635827,
+      "learning_rate": 9.998545540735632e-07,
+      "loss": 7.6063,
+      "step": 260
+    },
+    {
+      "epoch": 0.0120601283542232,
+      "learning_rate": 9.998060720980845e-07,
+      "loss": 6.9029,
+      "step": 280
+    },
+    {
+      "epoch": 0.01292156609381057,
+      "learning_rate": 9.997575901226053e-07,
+      "loss": 7.0533,
+      "step": 300
+    },
+    {
+      "epoch": 0.013783003833397941,
+      "learning_rate": 9.997091081471266e-07,
+      "loss": 6.9369,
+      "step": 320
+    },
+    {
+      "epoch": 0.014644441572985313,
+      "learning_rate": 9.996606261716477e-07,
+      "loss": 7.2298,
+      "step": 340
+    },
+    {
+      "epoch": 0.015505879312572683,
+      "learning_rate": 9.996121441961687e-07,
+      "loss": 6.9774,
+      "step": 360
+    },
+    {
+      "epoch": 0.016367317052160057,
+      "learning_rate": 9.9956366222069e-07,
+      "loss": 6.6153,
+      "step": 380
+    },
+    {
+      "epoch": 0.017228754791747427,
+      "learning_rate": 9.995151802452109e-07,
+      "loss": 6.7399,
+      "step": 400
+    },
+    {
+      "epoch": 0.018090192531334797,
+      "learning_rate": 9.994666982697322e-07,
+      "loss": 6.5775,
+      "step": 420
+    },
+    {
+      "epoch": 0.01895163027092217,
+      "learning_rate": 9.994182162942532e-07,
+      "loss": 6.9794,
+      "step": 440
+    },
+    {
+      "epoch": 0.01981306801050954,
+      "learning_rate": 9.993697343187743e-07,
+      "loss": 6.662,
+      "step": 460
+    },
+    {
+      "epoch": 0.02067450575009691,
+      "learning_rate": 9.993212523432956e-07,
+      "loss": 6.7392,
+      "step": 480
+    },
+    {
+      "epoch": 0.021535943489684285,
+      "learning_rate": 9.992727703678165e-07,
+      "loss": 6.7243,
+      "step": 500
+    },
+    {
+      "epoch": 0.022397381229271655,
+      "learning_rate": 9.992242883923377e-07,
+      "loss": 6.9017,
+      "step": 520
+    },
+    {
+      "epoch": 0.023258818968859025,
+      "learning_rate": 9.991758064168586e-07,
+      "loss": 6.3385,
+      "step": 540
+    },
+    {
+      "epoch": 0.0241202567084464,
+      "learning_rate": 9.991273244413799e-07,
+      "loss": 6.436,
+      "step": 560
+    },
+    {
+      "epoch": 0.02498169444803377,
+      "learning_rate": 9.99078842465901e-07,
+      "loss": 6.7753,
+      "step": 580
+    },
+    {
+      "epoch": 0.02584313218762114,
+      "learning_rate": 9.99030360490422e-07,
+      "loss": 6.2456,
+      "step": 600
+    },
+    {
+      "epoch": 0.026704569927208512,
+      "learning_rate": 9.989818785149433e-07,
+      "loss": 6.6402,
+      "step": 620
+    },
+    {
+      "epoch": 0.027566007666795882,
+      "learning_rate": 9.989333965394644e-07,
+      "loss": 6.3705,
+      "step": 640
+    },
+    {
+      "epoch": 0.028427445406383253,
+      "learning_rate": 9.988849145639854e-07,
+      "loss": 6.1228,
+      "step": 660
+    },
+    {
+      "epoch": 0.029288883145970626,
+      "learning_rate": 9.988364325885065e-07,
+      "loss": 6.377,
+      "step": 680
+    },
+    {
+      "epoch": 0.030150320885557996,
+      "learning_rate": 9.987879506130276e-07,
+      "loss": 6.5323,
+      "step": 700
+    },
+    {
+      "epoch": 0.031011758625145366,
+      "learning_rate": 9.987394686375489e-07,
+      "loss": 6.5611,
+      "step": 720
+    },
+    {
+      "epoch": 0.03187319636473274,
+      "learning_rate": 9.986909866620697e-07,
+      "loss": 6.5353,
+      "step": 740
+    },
+    {
+      "epoch": 0.032734634104320114,
+      "learning_rate": 9.98642504686591e-07,
+      "loss": 6.5194,
+      "step": 760
+    },
+    {
+      "epoch": 0.033596071843907484,
+      "learning_rate": 9.985940227111119e-07,
+      "loss": 6.4292,
+      "step": 780
+    },
+    {
+      "epoch": 0.034457509583494854,
+      "learning_rate": 9.985455407356332e-07,
+      "loss": 6.311,
+      "step": 800
+    },
+    {
+      "epoch": 0.035318947323082224,
+      "learning_rate": 9.984970587601542e-07,
+      "loss": 6.4556,
+      "step": 820
+    },
+    {
+      "epoch": 0.036180385062669594,
+      "learning_rate": 9.984485767846753e-07,
+      "loss": 6.3878,
+      "step": 840
+    },
+    {
+      "epoch": 0.037041822802256964,
+      "learning_rate": 9.984000948091966e-07,
+      "loss": 6.2106,
+      "step": 860
+    },
+    {
+      "epoch": 0.03790326054184434,
+      "learning_rate": 9.983516128337174e-07,
+      "loss": 6.4038,
+      "step": 880
+    },
+    {
+      "epoch": 0.03876469828143171,
+      "learning_rate": 9.983031308582387e-07,
+      "loss": 6.554,
+      "step": 900
+    },
+    {
+      "epoch": 0.03962613602101908,
+      "learning_rate": 9.982546488827596e-07,
+      "loss": 6.5065,
+      "step": 920
+    },
+    {
+      "epoch": 0.04048757376060645,
+      "learning_rate": 9.982061669072809e-07,
+      "loss": 6.2372,
+      "step": 940
+    },
+    {
+      "epoch": 0.04134901150019382,
+      "learning_rate": 9.98157684931802e-07,
+      "loss": 6.6843,
+      "step": 960
+    },
+    {
+      "epoch": 0.04221044923978119,
+      "learning_rate": 9.98109202956323e-07,
+      "loss": 6.3941,
+      "step": 980
+    },
+    {
+      "epoch": 0.04307188697936857,
+      "learning_rate": 9.980607209808443e-07,
+      "loss": 6.0508,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04393332471895594,
+      "learning_rate": 9.980122390053654e-07,
+      "loss": 6.1723,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04479476245854331,
+      "learning_rate": 9.979637570298864e-07,
+      "loss": 6.5495,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04565620019813068,
+      "learning_rate": 9.979152750544075e-07,
+      "loss": 6.4215,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04651763793771805,
+      "learning_rate": 9.978667930789286e-07,
+      "loss": 6.2316,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04737907567730542,
+      "learning_rate": 9.978183111034499e-07,
+      "loss": 6.291,
+      "step": 1100
+    },
+    {
+      "epoch": 0.0482405134168928,
+      "learning_rate": 9.977698291279707e-07,
+      "loss": 6.2892,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04910195115648017,
+      "learning_rate": 9.97721347152492e-07,
+      "loss": 6.1592,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04996338889606754,
+      "learning_rate": 9.976728651770129e-07,
+      "loss": 6.149,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05082482663565491,
+      "learning_rate": 9.976243832015342e-07,
+      "loss": 6.3048,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05168626437524228,
+      "learning_rate": 9.975759012260552e-07,
+      "loss": 6.4359,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05254770211482965,
+      "learning_rate": 9.975274192505763e-07,
+      "loss": 6.1895,
+      "step": 1220
+    },
+    {
+      "epoch": 0.053409139854417025,
+      "learning_rate": 9.974789372750976e-07,
+      "loss": 6.0717,
+      "step": 1240
+    },
+    {
+      "epoch": 0.054270577594004395,
+      "learning_rate": 9.974304552996186e-07,
+      "loss": 6.4554,
+      "step": 1260
+    },
+    {
+      "epoch": 0.055132015333591765,
+      "learning_rate": 9.973819733241397e-07,
+      "loss": 6.3709,
+      "step": 1280
+    },
+    {
+      "epoch": 0.055993453073179135,
+      "learning_rate": 9.973334913486606e-07,
+      "loss": 6.3882,
+      "step": 1300
+    },
+    {
+      "epoch": 0.056854890812766505,
+      "learning_rate": 9.972850093731819e-07,
+      "loss": 6.272,
+      "step": 1320
+    },
+    {
+      "epoch": 0.057716328552353875,
+      "learning_rate": 9.97236527397703e-07,
+      "loss": 6.0779,
+      "step": 1340
+    },
+    {
+      "epoch": 0.05857776629194125,
+      "learning_rate": 9.97188045422224e-07,
+      "loss": 6.3757,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05943920403152862,
+      "learning_rate": 9.971395634467453e-07,
+      "loss": 5.9969,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06030064177111599,
+      "learning_rate": 9.970910814712664e-07,
+      "loss": 6.288,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06116207951070336,
+      "learning_rate": 9.970425994957874e-07,
+      "loss": 6.3014,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06202351725029073,
+      "learning_rate": 9.969941175203085e-07,
+      "loss": 6.0141,
+      "step": 1440
+    },
+    {
+      "epoch": 0.0628849549898781,
+      "learning_rate": 9.969456355448296e-07,
+      "loss": 6.2296,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06374639272946547,
+      "learning_rate": 9.968971535693509e-07,
+      "loss": 6.2257,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06460783046905284,
+      "learning_rate": 9.968486715938717e-07,
+      "loss": 6.2917,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06546926820864023,
+      "learning_rate": 9.96800189618393e-07,
+      "loss": 6.0469,
+      "step": 1520
+    },
+    {
+      "epoch": 0.0663307059482276,
+      "learning_rate": 9.967517076429139e-07,
+      "loss": 6.1077,
+      "step": 1540
+    },
+    {
+      "epoch": 0.06719214368781497,
+      "learning_rate": 9.967032256674351e-07,
+      "loss": 6.2491,
+      "step": 1560
+    },
+    {
+      "epoch": 0.06805358142740234,
+      "learning_rate": 9.966547436919562e-07,
+      "loss": 5.8996,
+      "step": 1580
+    },
+    {
+      "epoch": 0.06891501916698971,
+      "learning_rate": 9.966062617164773e-07,
+      "loss": 6.3099,
+      "step": 1600
+    },
+    {
+      "epoch": 0.06977645690657708,
+      "learning_rate": 9.965577797409986e-07,
+      "loss": 6.2458,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07063789464616445,
+      "learning_rate": 9.965092977655196e-07,
+      "loss": 6.2759,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07149933238575182,
+      "learning_rate": 9.964608157900407e-07,
+      "loss": 6.2417,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07236077012533919,
+      "learning_rate": 9.964123338145618e-07,
+      "loss": 6.0339,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07322220786492656,
+      "learning_rate": 9.963638518390829e-07,
+      "loss": 6.1054,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07408364560451393,
+      "learning_rate": 9.963153698636041e-07,
+      "loss": 6.2843,
+      "step": 1720
+    },
+    {
+      "epoch": 0.0749450833441013,
+      "learning_rate": 9.96266887888125e-07,
+      "loss": 6.0034,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07580652108368868,
+      "learning_rate": 9.962184059126463e-07,
+      "loss": 6.2372,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07666795882327605,
+      "learning_rate": 9.961699239371674e-07,
+      "loss": 6.2253,
+      "step": 1780
+    },
+    {
+      "epoch": 0.07752939656286342,
+      "learning_rate": 9.961214419616884e-07,
+      "loss": 6.1184,
+      "step": 1800
+    },
+    {
+      "epoch": 0.0783908343024508,
+      "learning_rate": 9.960729599862095e-07,
+      "loss": 5.952,
+      "step": 1820
+    },
+    {
+      "epoch": 0.07925227204203816,
+      "learning_rate": 9.960244780107306e-07,
+      "loss": 6.1059,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08011370978162553,
+      "learning_rate": 9.959759960352519e-07,
+      "loss": 6.0038,
+      "step": 1860
+    },
+    {
+      "epoch": 0.0809751475212129,
+      "learning_rate": 9.959275140597727e-07,
+      "loss": 6.197,
+      "step": 1880
+    },
+    {
+      "epoch": 0.08183658526080027,
+      "learning_rate": 9.95879032084294e-07,
+      "loss": 6.0109,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08269802300038764,
+      "learning_rate": 9.95830550108815e-07,
+      "loss": 6.081,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08355946073997501,
+      "learning_rate": 9.957820681333361e-07,
+      "loss": 5.8241,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08442089847956238,
+      "learning_rate": 9.957335861578572e-07,
+      "loss": 6.2336,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08528233621914975,
+      "learning_rate": 9.956851041823783e-07,
+      "loss": 5.9938,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08614377395873714,
+      "learning_rate": 9.956366222068996e-07,
+      "loss": 6.1137,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08700521169832451,
+      "learning_rate": 9.955881402314206e-07,
+      "loss": 5.8231,
+      "step": 2020
+    },
+    {
+      "epoch": 0.08786664943791188,
+      "learning_rate": 9.955396582559417e-07,
+      "loss": 5.919,
+      "step": 2040
+    },
+    {
+      "epoch": 0.08872808717749925,
+      "learning_rate": 9.954911762804628e-07,
+      "loss": 5.9266,
+      "step": 2060
+    },
+    {
+      "epoch": 0.08958952491708662,
+      "learning_rate": 9.954426943049839e-07,
+      "loss": 6.0525,
+      "step": 2080
+    },
+    {
+      "epoch": 0.09045096265667399,
+      "learning_rate": 9.953942123295051e-07,
+      "loss": 5.8037,
+      "step": 2100
+    },
+    {
+      "epoch": 0.09131240039626136,
+      "learning_rate": 9.95345730354026e-07,
+      "loss": 5.8224,
+      "step": 2120
+    },
+    {
+      "epoch": 0.09217383813584873,
+      "learning_rate": 9.952972483785473e-07,
+      "loss": 5.9992,
+      "step": 2140
+    },
+    {
+      "epoch": 0.0930352758754361,
+      "learning_rate": 9.952487664030683e-07,
+      "loss": 6.222,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09389671361502347,
+      "learning_rate": 9.952002844275894e-07,
+      "loss": 6.221,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09475815135461084,
+      "learning_rate": 9.951518024521105e-07,
+      "loss": 5.9652,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09561958909419822,
+      "learning_rate": 9.951033204766316e-07,
+      "loss": 6.1783,
+      "step": 2220
+    },
+    {
+      "epoch": 0.0964810268337856,
+      "learning_rate": 9.950548385011528e-07,
+      "loss": 5.9059,
+      "step": 2240
+    },
+    {
+      "epoch": 0.09734246457337296,
+      "learning_rate": 9.95006356525674e-07,
+      "loss": 6.1,
+      "step": 2260
+    },
+    {
+      "epoch": 0.09820390231296033,
+      "learning_rate": 9.94957874550195e-07,
+      "loss": 5.9231,
+      "step": 2280
+    },
+    {
+      "epoch": 0.0990653400525477,
+      "learning_rate": 9.94909392574716e-07,
+      "loss": 6.0686,
+      "step": 2300
+    },
+    {
+      "epoch": 0.09992677779213507,
+      "learning_rate": 9.948609105992371e-07,
+      "loss": 5.8488,
+      "step": 2320
+    },
+    {
+      "epoch": 0.10078821553172244,
+      "learning_rate": 9.948124286237584e-07,
+      "loss": 5.7428,
+      "step": 2340
+    },
+    {
+      "epoch": 0.10164965327130981,
+      "learning_rate": 9.947639466482793e-07,
+      "loss": 5.8092,
+      "step": 2360
+    },
+    {
+      "epoch": 0.10251109101089718,
+      "learning_rate": 9.947154646728006e-07,
+      "loss": 6.0985,
+      "step": 2380
+    },
+    {
+      "epoch": 0.10337252875048455,
+      "learning_rate": 9.946669826973216e-07,
+      "loss": 6.14,
+      "step": 2400
+    },
+    {
+      "epoch": 0.10423396649007192,
+      "learning_rate": 9.946185007218427e-07,
+      "loss": 5.6721,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1050954042296593,
+      "learning_rate": 9.945700187463638e-07,
+      "loss": 5.6273,
+      "step": 2440
+    },
+    {
+      "epoch": 0.10595684196924668,
+      "learning_rate": 9.945215367708848e-07,
+      "loss": 6.1214,
+      "step": 2460
+    },
+    {
+      "epoch": 0.10681827970883405,
+      "learning_rate": 9.944730547954061e-07,
+      "loss": 5.9658,
+      "step": 2480
+    },
+    {
+      "epoch": 0.10767971744842142,
+      "learning_rate": 9.94424572819927e-07,
+      "loss": 5.7439,
+      "step": 2500
+    },
+    {
+      "epoch": 0.10854115518800879,
+      "learning_rate": 9.943760908444483e-07,
+      "loss": 5.7759,
+      "step": 2520
+    },
+    {
+      "epoch": 0.10940259292759616,
+      "learning_rate": 9.943276088689693e-07,
+      "loss": 5.9705,
+      "step": 2540
+    },
+    {
+      "epoch": 0.11026403066718353,
+      "learning_rate": 9.942791268934904e-07,
+      "loss": 5.5893,
+      "step": 2560
+    },
+    {
+      "epoch": 0.1111254684067709,
+      "learning_rate": 9.942306449180115e-07,
+      "loss": 5.7964,
+      "step": 2580
+    },
+    {
+      "epoch": 0.11198690614635827,
+      "learning_rate": 9.941821629425326e-07,
+      "loss": 5.8011,
+      "step": 2600
+    },
+    {
+      "epoch": 0.11284834388594564,
+      "learning_rate": 9.941336809670538e-07,
+      "loss": 5.7291,
+      "step": 2620
+    },
+    {
+      "epoch": 0.11370978162553301,
+      "learning_rate": 9.94085198991575e-07,
+      "loss": 5.7857,
+      "step": 2640
+    },
+    {
+      "epoch": 0.11457121936512038,
+      "learning_rate": 9.94036717016096e-07,
+      "loss": 5.8249,
+      "step": 2660
+    },
+    {
+      "epoch": 0.11543265710470775,
+      "learning_rate": 9.93988235040617e-07,
+      "loss": 5.6865,
+      "step": 2680
+    },
+    {
+      "epoch": 0.11629409484429513,
+      "learning_rate": 9.939397530651381e-07,
+      "loss": 5.8647,
+      "step": 2700
+    },
+    {
+      "epoch": 0.1171555325838825,
+      "learning_rate": 9.938912710896594e-07,
+      "loss": 5.558,
+      "step": 2720
+    },
+    {
+      "epoch": 0.11801697032346987,
+      "learning_rate": 9.938427891141803e-07,
+      "loss": 6.0017,
+      "step": 2740
+    },
+    {
+      "epoch": 0.11887840806305724,
+      "learning_rate": 9.937943071387016e-07,
+      "loss": 5.7348,
+      "step": 2760
+    },
+    {
+      "epoch": 0.11973984580264461,
+      "learning_rate": 9.937458251632226e-07,
+      "loss": 5.5196,
+      "step": 2780
+    },
+    {
+      "epoch": 0.12060128354223199,
+      "learning_rate": 9.936973431877437e-07,
+      "loss": 5.6164,
+      "step": 2800
+    },
+    {
+      "epoch": 0.12146272128181936,
+      "learning_rate": 9.936488612122648e-07,
+      "loss": 5.8701,
+      "step": 2820
+    },
+    {
+      "epoch": 0.12232415902140673,
+      "learning_rate": 9.936003792367858e-07,
+      "loss": 5.9252,
+      "step": 2840
+    },
+    {
+      "epoch": 0.1231855967609941,
+      "learning_rate": 9.935518972613071e-07,
+      "loss": 5.502,
+      "step": 2860
+    },
+    {
+      "epoch": 0.12404703450058147,
+      "learning_rate": 9.935034152858282e-07,
+      "loss": 5.8715,
+      "step": 2880
+    },
+    {
+      "epoch": 0.12490847224016884,
+      "learning_rate": 9.934549333103493e-07,
+      "loss": 5.5986,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1257699099797562,
+      "learning_rate": 9.934064513348703e-07,
+      "loss": 5.7461,
+      "step": 2920
+    },
+    {
+      "epoch": 0.1266313477193436,
+      "learning_rate": 9.933579693593914e-07,
+      "loss": 5.921,
+      "step": 2940
+    },
+    {
+      "epoch": 0.12749278545893095,
+      "learning_rate": 9.933094873839125e-07,
+      "loss": 5.8295,
+      "step": 2960
+    },
+    {
+      "epoch": 0.12835422319851833,
+      "learning_rate": 9.932610054084336e-07,
+      "loss": 5.6581,
+      "step": 2980
+    },
+    {
+      "epoch": 0.1292156609381057,
+      "learning_rate": 9.932125234329548e-07,
+      "loss": 5.7304,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13007709867769307,
+      "learning_rate": 9.93164041457476e-07,
+      "loss": 5.9998,
+      "step": 3020
+    },
+    {
+      "epoch": 0.13093853641728045,
+      "learning_rate": 9.93115559481997e-07,
+      "loss": 6.0369,
+      "step": 3040
+    },
+    {
+      "epoch": 0.1317999741568678,
+      "learning_rate": 9.930670775065183e-07,
+      "loss": 5.9676,
+      "step": 3060
+    },
+    {
+      "epoch": 0.1326614118964552,
+      "learning_rate": 9.930185955310391e-07,
+      "loss": 5.2663,
+      "step": 3080
+    },
+    {
+      "epoch": 0.13352284963604255,
+      "learning_rate": 9.929701135555604e-07,
+      "loss": 5.8159,
+      "step": 3100
+    },
+    {
+      "epoch": 0.13438428737562993,
+      "learning_rate": 9.929216315800813e-07,
+      "loss": 5.5133,
+      "step": 3120
+    },
+    {
+      "epoch": 0.1352457251152173,
+      "learning_rate": 9.928731496046025e-07,
+      "loss": 5.7083,
+      "step": 3140
+    },
+    {
+      "epoch": 0.13610716285480468,
+      "learning_rate": 9.928246676291236e-07,
+      "loss": 5.7072,
+      "step": 3160
+    },
+    {
+      "epoch": 0.13696860059439203,
+      "learning_rate": 9.927761856536447e-07,
+      "loss": 5.8228,
+      "step": 3180
+    },
+    {
+      "epoch": 0.13783003833397942,
+      "learning_rate": 9.927277036781658e-07,
+      "loss": 5.9269,
+      "step": 3200
+    },
+    {
+      "epoch": 0.13869147607356677,
+      "learning_rate": 9.926792217026868e-07,
+      "loss": 5.7547,
+      "step": 3220
+    },
+    {
+      "epoch": 0.13955291381315416,
+      "learning_rate": 9.926307397272081e-07,
+      "loss": 5.7518,
+      "step": 3240
+    },
+    {
+      "epoch": 0.1404143515527415,
+      "learning_rate": 9.925822577517292e-07,
+      "loss": 5.7172,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1412757892923289,
+      "learning_rate": 9.925337757762503e-07,
+      "loss": 5.3164,
+      "step": 3280
+    },
+    {
+      "epoch": 0.14213722703191628,
+      "learning_rate": 9.924852938007715e-07,
+      "loss": 5.5814,
+      "step": 3300
+    },
+    {
+      "epoch": 0.14299866477150364,
+      "learning_rate": 9.924368118252924e-07,
+      "loss": 5.7014,
+      "step": 3320
+    },
+    {
+      "epoch": 0.14386010251109102,
+      "learning_rate": 9.923883298498137e-07,
+      "loss": 5.6091,
+      "step": 3340
+    },
+    {
+      "epoch": 0.14472154025067838,
+      "learning_rate": 9.923398478743345e-07,
+      "loss": 5.9346,
+      "step": 3360
+    },
+    {
+      "epoch": 0.14558297799026576,
+      "learning_rate": 9.922913658988558e-07,
+      "loss": 5.411,
+      "step": 3380
+    },
+    {
+      "epoch": 0.14644441572985312,
+      "learning_rate": 9.92242883923377e-07,
+      "loss": 5.7669,
+      "step": 3400
+    },
+    {
+      "epoch": 0.1473058534694405,
+      "learning_rate": 9.92194401947898e-07,
+      "loss": 5.3575,
+      "step": 3420
+    },
+    {
+      "epoch": 0.14816729120902786,
+      "learning_rate": 9.921459199724193e-07,
+      "loss": 5.8575,
+      "step": 3440
+    },
+    {
+      "epoch": 0.14902872894861524,
+      "learning_rate": 9.920974379969401e-07,
+      "loss": 5.5618,
+      "step": 3460
+    },
+    {
+      "epoch": 0.1498901666882026,
+      "learning_rate": 9.920489560214614e-07,
+      "loss": 5.7241,
+      "step": 3480
+    },
+    {
+      "epoch": 0.15075160442778998,
+      "learning_rate": 9.920004740459823e-07,
+      "loss": 5.6646,
+      "step": 3500
+    },
+    {
+      "epoch": 0.15161304216737737,
+      "learning_rate": 9.919519920705035e-07,
+      "loss": 5.7116,
+      "step": 3520
+    },
+    {
+      "epoch": 0.15247447990696472,
+      "learning_rate": 9.919035100950246e-07,
+      "loss": 5.4578,
+      "step": 3540
+    },
+    {
+      "epoch": 0.1533359176465521,
+      "learning_rate": 9.918550281195457e-07,
+      "loss": 5.4505,
+      "step": 3560
+    },
+    {
+      "epoch": 0.15419735538613946,
+      "learning_rate": 9.918065461440668e-07,
+      "loss": 5.6805,
+      "step": 3580
+    },
+    {
+      "epoch": 0.15505879312572685,
+      "learning_rate": 9.917580641685878e-07,
+      "loss": 5.3575,
+      "step": 3600
+    },
+    {
+      "epoch": 0.1559202308653142,
+      "learning_rate": 9.917095821931091e-07,
+      "loss": 5.7853,
+      "step": 3620
+    },
+    {
+      "epoch": 0.1567816686049016,
+      "learning_rate": 9.916611002176302e-07,
+      "loss": 5.8258,
+      "step": 3640
+    },
+    {
+      "epoch": 0.15764310634448894,
+      "learning_rate": 9.916126182421513e-07,
+      "loss": 5.6299,
+      "step": 3660
+    },
+    {
+      "epoch": 0.15850454408407633,
+      "learning_rate": 9.915641362666725e-07,
+      "loss": 5.7533,
+      "step": 3680
+    },
+    {
+      "epoch": 0.15936598182366368,
+      "learning_rate": 9.915156542911934e-07,
+      "loss": 5.6216,
+      "step": 3700
+    },
+    {
+      "epoch": 0.16022741956325107,
+      "learning_rate": 9.914671723157147e-07,
+      "loss": 5.9745,
+      "step": 3720
+    },
+    {
+      "epoch": 0.16108885730283845,
+      "learning_rate": 9.914186903402355e-07,
+      "loss": 5.6028,
+      "step": 3740
+    },
+    {
+      "epoch": 0.1619502950424258,
+      "learning_rate": 9.913702083647568e-07,
+      "loss": 5.658,
+      "step": 3760
+    },
+    {
+      "epoch": 0.1628117327820132,
+      "learning_rate": 9.913217263892779e-07,
+      "loss": 5.4876,
+      "step": 3780
+    },
+    {
+      "epoch": 0.16367317052160055,
+      "learning_rate": 9.91273244413799e-07,
+      "loss": 5.5794,
+      "step": 3800
+    },
+    {
+      "epoch": 0.16453460826118793,
+      "learning_rate": 9.912247624383202e-07,
+      "loss": 5.6442,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1653960460007753,
+      "learning_rate": 9.911762804628411e-07,
+      "loss": 5.5536,
+      "step": 3840
+    },
+    {
+      "epoch": 0.16625748374036267,
+      "learning_rate": 9.911277984873624e-07,
+      "loss": 5.5444,
+      "step": 3860
+    },
+    {
+      "epoch": 0.16711892147995003,
+      "learning_rate": 9.910793165118835e-07,
+      "loss": 5.7136,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1679803592195374,
+      "learning_rate": 9.910308345364045e-07,
+      "loss": 5.2753,
+      "step": 3900
+    },
+    {
+      "epoch": 0.16884179695912477,
+      "learning_rate": 9.909823525609258e-07,
+      "loss": 5.4269,
+      "step": 3920
+    },
+    {
+      "epoch": 0.16970323469871215,
+      "learning_rate": 9.909338705854467e-07,
+      "loss": 5.4923,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1705646724382995,
+      "learning_rate": 9.90885388609968e-07,
+      "loss": 5.5202,
+      "step": 3960
+    },
+    {
+      "epoch": 0.1714261101778869,
+      "learning_rate": 9.908369066344888e-07,
+      "loss": 5.5258,
+      "step": 3980
+    },
+    {
+      "epoch": 0.17228754791747428,
+      "learning_rate": 9.9078842465901e-07,
+      "loss": 5.5544,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17314898565706163,
+      "learning_rate": 9.907399426835312e-07,
+      "loss": 5.5478,
+      "step": 4020
+    },
+    {
+      "epoch": 0.17401042339664902,
+      "learning_rate": 9.906914607080522e-07,
+      "loss": 5.5041,
+      "step": 4040
+    },
+    {
+      "epoch": 0.17487186113623637,
+      "learning_rate": 9.906429787325735e-07,
+      "loss": 5.4906,
+      "step": 4060
+    },
+    {
+      "epoch": 0.17573329887582376,
+      "learning_rate": 9.905944967570944e-07,
+      "loss": 5.4225,
+      "step": 4080
+    },
+    {
+      "epoch": 0.1765947366154111,
+      "learning_rate": 9.905460147816157e-07,
+      "loss": 5.6207,
+      "step": 4100
+    },
+    {
+      "epoch": 0.1774561743549985,
+      "learning_rate": 9.904975328061365e-07,
+      "loss": 5.5054,
+      "step": 4120
+    },
+    {
+      "epoch": 0.17831761209458585,
+      "learning_rate": 9.904490508306578e-07,
+      "loss": 5.4273,
+      "step": 4140
+    },
+    {
+      "epoch": 0.17917904983417324,
+      "learning_rate": 9.904005688551789e-07,
+      "loss": 5.6312,
+      "step": 4160
+    },
+    {
+      "epoch": 0.1800404875737606,
+      "learning_rate": 9.903520868797e-07,
+      "loss": 5.5304,
+      "step": 4180
+    },
+    {
+      "epoch": 0.18090192531334798,
+      "learning_rate": 9.903036049042212e-07,
+      "loss": 5.8041,
+      "step": 4200
+    },
+    {
+      "epoch": 0.18176336305293536,
+      "learning_rate": 9.90255122928742e-07,
+      "loss": 5.6389,
+      "step": 4220
+    },
+    {
+      "epoch": 0.18262480079252272,
+      "learning_rate": 9.902066409532634e-07,
+      "loss": 5.4039,
+      "step": 4240
+    },
+    {
+      "epoch": 0.1834862385321101,
+      "learning_rate": 9.901581589777845e-07,
+      "loss": 5.3949,
+      "step": 4260
+    },
+    {
+      "epoch": 0.18434767627169746,
+      "learning_rate": 9.901096770023055e-07,
+      "loss": 5.462,
+      "step": 4280
+    },
+    {
+      "epoch": 0.18520911401128484,
+      "learning_rate": 9.900611950268268e-07,
+      "loss": 5.6406,
+      "step": 4300
+    },
+    {
+      "epoch": 0.1860705517508722,
+      "learning_rate": 9.900127130513477e-07,
+      "loss": 5.4098,
+      "step": 4320
+    },
+    {
+      "epoch": 0.18693198949045958,
+      "learning_rate": 9.89964231075869e-07,
+      "loss": 5.5045,
+      "step": 4340
+    },
+    {
+      "epoch": 0.18779342723004694,
+      "learning_rate": 9.899157491003898e-07,
+      "loss": 5.4192,
+      "step": 4360
+    },
+    {
+      "epoch": 0.18865486496963432,
+      "learning_rate": 9.89867267124911e-07,
+      "loss": 5.3939,
+      "step": 4380
+    },
+    {
+      "epoch": 0.18951630270922168,
+      "learning_rate": 9.898187851494322e-07,
+      "loss": 5.6309,
+      "step": 4400
+    },
+    {
+      "epoch": 0.19037774044880906,
+      "learning_rate": 9.897703031739532e-07,
+      "loss": 5.3819,
+      "step": 4420
+    },
+    {
+      "epoch": 0.19123917818839645,
+      "learning_rate": 9.897218211984745e-07,
+      "loss": 5.3316,
+      "step": 4440
+    },
+    {
+      "epoch": 0.1921006159279838,
+      "learning_rate": 9.896733392229954e-07,
+      "loss": 5.3271,
+      "step": 4460
+    },
+    {
+      "epoch": 0.1929620536675712,
+      "learning_rate": 9.896248572475167e-07,
+      "loss": 5.5944,
+      "step": 4480
+    },
+    {
+      "epoch": 0.19382349140715854,
+      "learning_rate": 9.895763752720377e-07,
+      "loss": 5.4956,
+      "step": 4500
+    },
+    {
+      "epoch": 0.19468492914674593,
+      "learning_rate": 9.895278932965588e-07,
+      "loss": 5.3326,
+      "step": 4520
+    },
+    {
+      "epoch": 0.19554636688633328,
+      "learning_rate": 9.894794113210799e-07,
+      "loss": 5.4429,
+      "step": 4540
+    },
+    {
+      "epoch": 0.19640780462592067,
+      "learning_rate": 9.89430929345601e-07,
+      "loss": 5.4955,
+      "step": 4560
+    },
+    {
+      "epoch": 0.19726924236550802,
+      "learning_rate": 9.893824473701222e-07,
+      "loss": 5.2479,
+      "step": 4580
+    },
+    {
+      "epoch": 0.1981306801050954,
+      "learning_rate": 9.89333965394643e-07,
+      "loss": 5.3128,
+      "step": 4600
+    },
+    {
+      "epoch": 0.19899211784468276,
+      "learning_rate": 9.892854834191644e-07,
+      "loss": 5.3392,
+      "step": 4620
+    },
+    {
+      "epoch": 0.19985355558427015,
+      "learning_rate": 9.892370014436854e-07,
+      "loss": 5.5973,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2007149933238575,
+      "learning_rate": 9.891885194682065e-07,
+      "loss": 5.4401,
+      "step": 4660
+    },
+    {
+      "epoch": 0.2015764310634449,
+      "learning_rate": 9.891400374927278e-07,
+      "loss": 5.3984,
+      "step": 4680
+    },
+    {
+      "epoch": 0.20243786880303227,
+      "learning_rate": 9.890915555172487e-07,
+      "loss": 5.2152,
+      "step": 4700
+    },
+    {
+      "epoch": 0.20329930654261963,
+      "learning_rate": 9.8904307354177e-07,
+      "loss": 5.5735,
+      "step": 4720
+    },
+    {
+      "epoch": 0.204160744282207,
+      "learning_rate": 9.889945915662908e-07,
+      "loss": 5.4082,
+      "step": 4740
+    },
+    {
+      "epoch": 0.20502218202179437,
+      "learning_rate": 9.88946109590812e-07,
+      "loss": 5.2788,
+      "step": 4760
+    },
+    {
+      "epoch": 0.20588361976138175,
+      "learning_rate": 9.888976276153332e-07,
+      "loss": 5.384,
+      "step": 4780
+    },
+    {
+      "epoch": 0.2067450575009691,
+      "learning_rate": 9.888491456398542e-07,
+      "loss": 5.4814,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2076064952405565,
+      "learning_rate": 9.888006636643755e-07,
+      "loss": 5.1962,
+      "step": 4820
+    },
+    {
+      "epoch": 0.20846793298014385,
+      "learning_rate": 9.887521816888966e-07,
+      "loss": 5.0913,
+      "step": 4840
+    },
+    {
+      "epoch": 0.20932937071973123,
+      "learning_rate": 9.887036997134177e-07,
+      "loss": 5.3056,
+      "step": 4860
+    },
+    {
+      "epoch": 0.2101908084593186,
+      "learning_rate": 9.886552177379387e-07,
+      "loss": 5.4644,
+      "step": 4880
+    },
+    {
+      "epoch": 0.21105224619890597,
+      "learning_rate": 9.886067357624598e-07,
+      "loss": 5.6686,
+      "step": 4900
+    },
+    {
+      "epoch": 0.21191368393849336,
+      "learning_rate": 9.88558253786981e-07,
+      "loss": 5.3201,
+      "step": 4920
+    },
+    {
+      "epoch": 0.21277512167808071,
+      "learning_rate": 9.88509771811502e-07,
+      "loss": 5.2799,
+      "step": 4940
+    },
+    {
+      "epoch": 0.2136365594176681,
+      "learning_rate": 9.884612898360232e-07,
+      "loss": 5.1967,
+      "step": 4960
+    },
+    {
+      "epoch": 0.21449799715725545,
+      "learning_rate": 9.88412807860544e-07,
+      "loss": 5.4358,
+      "step": 4980
+    },
+    {
+      "epoch": 0.21535943489684284,
+      "learning_rate": 9.883643258850654e-07,
+      "loss": 5.124,
+      "step": 5000
+    },
+    {
+      "epoch": 0.2162208726364302,
+      "learning_rate": 9.883158439095864e-07,
+      "loss": 5.4017,
+      "step": 5020
+    },
+    {
+      "epoch": 0.21708231037601758,
+      "learning_rate": 9.882673619341075e-07,
+      "loss": 5.28,
+      "step": 5040
+    },
+    {
+      "epoch": 0.21794374811560493,
+      "learning_rate": 9.882188799586288e-07,
+      "loss": 5.427,
+      "step": 5060
+    },
+    {
+      "epoch": 0.21880518585519232,
+      "learning_rate": 9.881703979831497e-07,
+      "loss": 5.272,
+      "step": 5080
+    },
+    {
+      "epoch": 0.21966662359477968,
+      "learning_rate": 9.88121916007671e-07,
+      "loss": 5.3149,
+      "step": 5100
+    },
+    {
+      "epoch": 0.22052806133436706,
+      "learning_rate": 9.880734340321918e-07,
+      "loss": 5.5404,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22138949907395444,
+      "learning_rate": 9.88024952056713e-07,
+      "loss": 5.3568,
+      "step": 5140
+    },
+    {
+      "epoch": 0.2222509368135418,
+      "learning_rate": 9.879764700812342e-07,
+      "loss": 5.2099,
+      "step": 5160
+    },
+    {
+      "epoch": 0.22311237455312918,
+      "learning_rate": 9.879279881057552e-07,
+      "loss": 5.2753,
+      "step": 5180
+    },
+    {
+      "epoch": 0.22397381229271654,
+      "learning_rate": 9.878795061302765e-07,
+      "loss": 5.3216,
+      "step": 5200
+    },
+    {
+      "epoch": 0.22483525003230392,
+      "learning_rate": 9.878310241547976e-07,
+      "loss": 5.332,
+      "step": 5220
+    },
+    {
+      "epoch": 0.22569668777189128,
+      "learning_rate": 9.877825421793187e-07,
+      "loss": 5.1957,
+      "step": 5240
+    },
+    {
+      "epoch": 0.22655812551147866,
+      "learning_rate": 9.877340602038397e-07,
+      "loss": 5.493,
+      "step": 5260
+    },
+    {
+      "epoch": 0.22741956325106602,
+      "learning_rate": 9.876855782283608e-07,
+      "loss": 5.4048,
+      "step": 5280
+    },
+    {
+      "epoch": 0.2282810009906534,
+      "learning_rate": 9.87637096252882e-07,
+      "loss": 5.4582,
+      "step": 5300
+    },
+    {
+      "epoch": 0.22914243873024076,
+      "learning_rate": 9.87588614277403e-07,
+      "loss": 5.1358,
+      "step": 5320
+    },
+    {
+      "epoch": 0.23000387646982814,
+      "learning_rate": 9.875401323019242e-07,
+      "loss": 5.507,
+      "step": 5340
+    },
+    {
+      "epoch": 0.2308653142094155,
+      "learning_rate": 9.87491650326445e-07,
+      "loss": 5.2223,
+      "step": 5360
+    },
+    {
+      "epoch": 0.23172675194900288,
+      "learning_rate": 9.874431683509664e-07,
+      "loss": 5.2098,
+      "step": 5380
+    },
+    {
+      "epoch": 0.23258818968859027,
+      "learning_rate": 9.873946863754874e-07,
+      "loss": 5.2615,
+      "step": 5400
+    },
+    {
+      "epoch": 0.23344962742817763,
+      "learning_rate": 9.873462044000085e-07,
+      "loss": 5.2096,
+      "step": 5420
+    },
+    {
+      "epoch": 0.234311065167765,
+      "learning_rate": 9.872977224245298e-07,
+      "loss": 5.1175,
+      "step": 5440
+    },
+    {
+      "epoch": 0.23517250290735237,
+      "learning_rate": 9.872492404490509e-07,
+      "loss": 5.2392,
+      "step": 5460
+    },
+    {
+      "epoch": 0.23603394064693975,
+      "learning_rate": 9.87200758473572e-07,
+      "loss": 5.1828,
+      "step": 5480
+    },
+    {
+      "epoch": 0.2368953783865271,
+      "learning_rate": 9.87152276498093e-07,
+      "loss": 5.1802,
+      "step": 5500
+    },
+    {
+      "epoch": 0.2377568161261145,
+      "learning_rate": 9.87103794522614e-07,
+      "loss": 5.2447,
+      "step": 5520
+    },
+    {
+      "epoch": 0.23861825386570185,
+      "learning_rate": 9.870553125471354e-07,
+      "loss": 5.0941,
+      "step": 5540
+    },
+    {
+      "epoch": 0.23947969160528923,
+      "learning_rate": 9.870068305716562e-07,
+      "loss": 5.0211,
+      "step": 5560
+    },
+    {
+      "epoch": 0.2403411293448766,
+      "learning_rate": 9.869583485961775e-07,
+      "loss": 5.4385,
+      "step": 5580
+    },
+    {
+      "epoch": 0.24120256708446397,
+      "learning_rate": 9.869098666206986e-07,
+      "loss": 5.5081,
+      "step": 5600
+    },
+    {
+      "epoch": 0.24206400482405135,
+      "learning_rate": 9.868613846452196e-07,
+      "loss": 5.5758,
+      "step": 5620
+    },
+    {
+      "epoch": 0.2429254425636387,
+      "learning_rate": 9.868129026697407e-07,
+      "loss": 5.379,
+      "step": 5640
+    },
+    {
+      "epoch": 0.2437868803032261,
+      "learning_rate": 9.867644206942618e-07,
+      "loss": 5.5049,
+      "step": 5660
+    },
+    {
+      "epoch": 0.24464831804281345,
+      "learning_rate": 9.86715938718783e-07,
+      "loss": 5.2212,
+      "step": 5680
+    },
+    {
+      "epoch": 0.24550975578240083,
+      "learning_rate": 9.86667456743304e-07,
+      "loss": 5.1315,
+      "step": 5700
+    },
+    {
+      "epoch": 0.2463711935219882,
+      "learning_rate": 9.866189747678252e-07,
+      "loss": 4.9211,
+      "step": 5720
+    },
+    {
+      "epoch": 0.24723263126157558,
+      "learning_rate": 9.86570492792346e-07,
+      "loss": 5.2577,
+      "step": 5740
+    },
+    {
+      "epoch": 0.24809406900116293,
+      "learning_rate": 9.865220108168674e-07,
+      "loss": 5.1484,
+      "step": 5760
+    },
+    {
+      "epoch": 0.24895550674075032,
+      "learning_rate": 9.864735288413884e-07,
+      "loss": 5.2857,
+      "step": 5780
+    },
+    {
+      "epoch": 0.24981694448033767,
+      "learning_rate": 9.864250468659095e-07,
+      "loss": 5.1734,
+      "step": 5800
+    },
+    {
+      "epoch": 0.25067838221992506,
+      "learning_rate": 9.863765648904308e-07,
+      "loss": 5.3354,
+      "step": 5820
+    },
+    {
+      "epoch": 0.2515398199595124,
+      "learning_rate": 9.863280829149519e-07,
+      "loss": 5.2673,
+      "step": 5840
+    },
+    {
+      "epoch": 0.2524012576990998,
+      "learning_rate": 9.86279600939473e-07,
+      "loss": 5.1835,
+      "step": 5860
+    },
+    {
+      "epoch": 0.2532626954386872,
+      "learning_rate": 9.86231118963994e-07,
+      "loss": 5.1701,
+      "step": 5880
+    },
+    {
+      "epoch": 0.25412413317827454,
+      "learning_rate": 9.86182636988515e-07,
+      "loss": 5.2417,
+      "step": 5900
+    },
+    {
+      "epoch": 0.2549855709178619,
+      "learning_rate": 9.861341550130364e-07,
+      "loss": 4.9244,
+      "step": 5920
+    },
+    {
+      "epoch": 0.2558470086574493,
+      "learning_rate": 9.860856730375572e-07,
+      "loss": 5.0248,
+      "step": 5940
+    },
+    {
+      "epoch": 0.25670844639703666,
+      "learning_rate": 9.860371910620785e-07,
+      "loss": 5.362,
+      "step": 5960
+    },
+    {
+      "epoch": 0.257569884136624,
+      "learning_rate": 9.859887090865996e-07,
+      "loss": 4.9936,
+      "step": 5980
+    },
+    {
+      "epoch": 0.2584313218762114,
+      "learning_rate": 9.859402271111206e-07,
+      "loss": 5.5917,
+      "step": 6000
+    },
+    {
+      "epoch": 0.2592927596157988,
+      "learning_rate": 9.858917451356417e-07,
+      "loss": 5.1554,
+      "step": 6020
+    },
+    {
+      "epoch": 0.26015419735538614,
+      "learning_rate": 9.858432631601628e-07,
+      "loss": 5.3179,
+      "step": 6040
+    },
+    {
+      "epoch": 0.2610156350949735,
+      "learning_rate": 9.85794781184684e-07,
+      "loss": 5.2851,
+      "step": 6060
+    },
+    {
+      "epoch": 0.2618770728345609,
+      "learning_rate": 9.857462992092051e-07,
+      "loss": 5.2355,
+      "step": 6080
+    },
+    {
+      "epoch": 0.26273851057414827,
+      "learning_rate": 9.856978172337262e-07,
+      "loss": 5.2866,
+      "step": 6100
+    },
+    {
+      "epoch": 0.2635999483137356,
+      "learning_rate": 9.85649335258247e-07,
+      "loss": 5.2053,
+      "step": 6120
+    },
+    {
+      "epoch": 0.264461386053323,
+      "learning_rate": 9.856008532827684e-07,
+      "loss": 5.0226,
+      "step": 6140
+    },
+    {
+      "epoch": 0.2653228237929104,
+      "learning_rate": 9.855523713072894e-07,
+      "loss": 5.2522,
+      "step": 6160
+    },
+    {
+      "epoch": 0.26618426153249775,
+      "learning_rate": 9.855038893318105e-07,
+      "loss": 5.101,
+      "step": 6180
+    },
+    {
+      "epoch": 0.2670456992720851,
+      "learning_rate": 9.854554073563318e-07,
+      "loss": 5.272,
+      "step": 6200
+    },
+    {
+      "epoch": 0.26790713701167246,
+      "learning_rate": 9.854069253808529e-07,
+      "loss": 5.2802,
+      "step": 6220
+    },
+    {
+      "epoch": 0.26876857475125987,
+      "learning_rate": 9.85358443405374e-07,
+      "loss": 5.3429,
+      "step": 6240
+    },
+    {
+      "epoch": 0.2696300124908472,
+      "learning_rate": 9.85309961429895e-07,
+      "loss": 5.3596,
+      "step": 6260
+    },
+    {
+      "epoch": 0.2704914502304346,
+      "learning_rate": 9.85261479454416e-07,
+      "loss": 4.9346,
+      "step": 6280
+    },
+    {
+      "epoch": 0.271352887970022,
+      "learning_rate": 9.852129974789373e-07,
+      "loss": 5.178,
+      "step": 6300
+    },
+    {
+      "epoch": 0.27221432570960935,
+      "learning_rate": 9.851645155034582e-07,
+      "loss": 5.266,
+      "step": 6320
+    },
+    {
+      "epoch": 0.2730757634491967,
+      "learning_rate": 9.851160335279795e-07,
+      "loss": 5.205,
+      "step": 6340
+    },
+    {
+      "epoch": 0.27393720118878406,
+      "learning_rate": 9.850675515525006e-07,
+      "loss": 5.1011,
+      "step": 6360
+    },
+    {
+      "epoch": 0.2747986389283715,
+      "learning_rate": 9.850190695770216e-07,
+      "loss": 5.3448,
+      "step": 6380
+    },
+    {
+      "epoch": 0.27566007666795883,
+      "learning_rate": 9.849705876015427e-07,
+      "loss": 5.1696,
+      "step": 6400
+    },
+    {
+      "epoch": 0.2765215144075462,
+      "learning_rate": 9.849221056260638e-07,
+      "loss": 4.74,
+      "step": 6420
+    },
+    {
+      "epoch": 0.27738295214713354,
+      "learning_rate": 9.84873623650585e-07,
+      "loss": 5.101,
+      "step": 6440
+    },
+    {
+      "epoch": 0.27824438988672096,
+      "learning_rate": 9.848251416751061e-07,
+      "loss": 5.1098,
+      "step": 6460
+    },
+    {
+      "epoch": 0.2791058276263083,
+      "learning_rate": 9.847766596996272e-07,
+      "loss": 5.2632,
+      "step": 6480
+    },
+    {
+      "epoch": 0.27996726536589567,
+      "learning_rate": 9.847281777241483e-07,
+      "loss": 5.2572,
+      "step": 6500
+    },
+    {
+      "epoch": 0.280828703105483,
+      "learning_rate": 9.846796957486693e-07,
+      "loss": 5.2582,
+      "step": 6520
+    },
+    {
+      "epoch": 0.28169014084507044,
+      "learning_rate": 9.846312137731906e-07,
+      "loss": 5.3315,
+      "step": 6540
+    },
+    {
+      "epoch": 0.2825515785846578,
+      "learning_rate": 9.845827317977115e-07,
+      "loss": 4.955,
+      "step": 6560
+    },
+    {
+      "epoch": 0.28341301632424515,
+      "learning_rate": 9.845342498222328e-07,
+      "loss": 5.2012,
+      "step": 6580
+    },
+    {
+      "epoch": 0.28427445406383256,
+      "learning_rate": 9.844857678467538e-07,
+      "loss": 5.3038,
+      "step": 6600
+    },
+    {
+      "epoch": 0.2851358918034199,
+      "learning_rate": 9.84437285871275e-07,
+      "loss": 5.2074,
+      "step": 6620
+    },
+    {
+      "epoch": 0.2859973295430073,
+      "learning_rate": 9.84388803895796e-07,
+      "loss": 5.0363,
+      "step": 6640
+    },
+    {
+      "epoch": 0.28685876728259463,
+      "learning_rate": 9.84340321920317e-07,
+      "loss": 5.1664,
+      "step": 6660
+    },
+    {
+      "epoch": 0.28772020502218204,
+      "learning_rate": 9.842918399448383e-07,
+      "loss": 5.1214,
+      "step": 6680
+    },
+    {
+      "epoch": 0.2885816427617694,
+      "learning_rate": 9.842433579693592e-07,
+      "loss": 4.9995,
+      "step": 6700
+    },
+    {
+      "epoch": 0.28944308050135675,
+      "learning_rate": 9.841948759938805e-07,
+      "loss": 5.0777,
+      "step": 6720
+    },
+    {
+      "epoch": 0.2903045182409441,
+      "learning_rate": 9.841463940184016e-07,
+      "loss": 5.1652,
+      "step": 6740
+    },
+    {
+      "epoch": 0.2911659559805315,
+      "learning_rate": 9.840979120429226e-07,
+      "loss": 5.1024,
+      "step": 6760
+    },
+    {
+      "epoch": 0.2920273937201189,
+      "learning_rate": 9.840494300674437e-07,
+      "loss": 5.054,
+      "step": 6780
+    },
+    {
+      "epoch": 0.29288883145970623,
+      "learning_rate": 9.840009480919648e-07,
+      "loss": 5.2045,
+      "step": 6800
+    },
+    {
+      "epoch": 0.29375026919929365,
+      "learning_rate": 9.83952466116486e-07,
+      "loss": 5.3555,
+      "step": 6820
+    },
+    {
+      "epoch": 0.294611706938881,
+      "learning_rate": 9.839039841410071e-07,
+      "loss": 5.1176,
+      "step": 6840
+    },
+    {
+      "epoch": 0.29547314467846836,
+      "learning_rate": 9.838555021655282e-07,
+      "loss": 5.4848,
+      "step": 6860
+    },
+    {
+      "epoch": 0.2963345824180557,
+      "learning_rate": 9.838070201900493e-07,
+      "loss": 5.1506,
+      "step": 6880
+    },
+    {
+      "epoch": 0.2971960201576431,
+      "learning_rate": 9.837585382145703e-07,
+      "loss": 5.1673,
+      "step": 6900
+    },
+    {
+      "epoch": 0.2980574578972305,
+      "learning_rate": 9.837100562390916e-07,
+      "loss": 5.0694,
+      "step": 6920
+    },
+    {
+      "epoch": 0.29891889563681784,
+      "learning_rate": 9.836615742636125e-07,
+      "loss": 5.4286,
+      "step": 6940
+    },
+    {
+      "epoch": 0.2997803333764052,
+      "learning_rate": 9.836130922881338e-07,
+      "loss": 4.9144,
+      "step": 6960
+    },
+    {
+      "epoch": 0.3006417711159926,
+      "learning_rate": 9.835646103126548e-07,
+      "loss": 4.9661,
+      "step": 6980
+    },
+    {
+      "epoch": 0.30150320885557996,
+      "learning_rate": 9.83516128337176e-07,
+      "loss": 5.0778,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3023646465951673,
+      "learning_rate": 9.83467646361697e-07,
+      "loss": 5.14,
+      "step": 7020
+    },
+    {
+      "epoch": 0.30322608433475473,
+      "learning_rate": 9.83419164386218e-07,
+      "loss": 5.1204,
+      "step": 7040
+    },
+    {
+      "epoch": 0.3040875220743421,
+      "learning_rate": 9.833706824107393e-07,
+      "loss": 4.942,
+      "step": 7060
+    },
+    {
+      "epoch": 0.30494895981392944,
+      "learning_rate": 9.833222004352604e-07,
+      "loss": 5.08,
+      "step": 7080
+    },
+    {
+      "epoch": 0.3058103975535168,
+      "learning_rate": 9.832737184597815e-07,
+      "loss": 5.1513,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3066718352931042,
+      "learning_rate": 9.832252364843028e-07,
+      "loss": 5.0832,
+      "step": 7120
+    },
+    {
+      "epoch": 0.30753327303269157,
+      "learning_rate": 9.831767545088236e-07,
+      "loss": 5.1246,
+      "step": 7140
+    },
+    {
+      "epoch": 0.3083947107722789,
+      "learning_rate": 9.83128272533345e-07,
+      "loss": 5.1067,
+      "step": 7160
+    },
+    {
+      "epoch": 0.3092561485118663,
+      "learning_rate": 9.830797905578658e-07,
+      "loss": 5.0547,
+      "step": 7180
+    },
+    {
+      "epoch": 0.3101175862514537,
+      "learning_rate": 9.83031308582387e-07,
+      "loss": 5.3209,
+      "step": 7200
+    },
+    {
+      "epoch": 0.31097902399104105,
+      "learning_rate": 9.829828266069081e-07,
+      "loss": 5.1226,
+      "step": 7220
+    },
+    {
+      "epoch": 0.3118404617306284,
+      "learning_rate": 9.829343446314292e-07,
+      "loss": 5.2013,
+      "step": 7240
+    },
+    {
+      "epoch": 0.3127018994702158,
+      "learning_rate": 9.828858626559505e-07,
+      "loss": 5.1918,
+      "step": 7260
+    },
+    {
+      "epoch": 0.3135633372098032,
+      "learning_rate": 9.828373806804713e-07,
+      "loss": 5.1001,
+      "step": 7280
+    },
+    {
+      "epoch": 0.31442477494939053,
+      "learning_rate": 9.827888987049926e-07,
+      "loss": 5.1812,
+      "step": 7300
+    },
+    {
+      "epoch": 0.3152862126889779,
+      "learning_rate": 9.827404167295135e-07,
+      "loss": 4.9552,
+      "step": 7320
+    },
+    {
+      "epoch": 0.3161476504285653,
+      "learning_rate": 9.826919347540348e-07,
+      "loss": 5.221,
+      "step": 7340
+    },
+    {
+      "epoch": 0.31700908816815265,
+      "learning_rate": 9.826434527785558e-07,
+      "loss": 4.8989,
+      "step": 7360
+    },
+    {
+      "epoch": 0.31787052590774,
+      "learning_rate": 9.82594970803077e-07,
+      "loss": 5.2631,
+      "step": 7380
+    },
+    {
+      "epoch": 0.31873196364732737,
+      "learning_rate": 9.82546488827598e-07,
+      "loss": 5.0938,
+      "step": 7400
+    },
+    {
+      "epoch": 0.3195934013869148,
+      "learning_rate": 9.82498006852119e-07,
+      "loss": 5.0062,
+      "step": 7420
+    },
+    {
+      "epoch": 0.32045483912650213,
+      "learning_rate": 9.824495248766403e-07,
+      "loss": 4.9372,
+      "step": 7440
+    },
+    {
+      "epoch": 0.3213162768660895,
+      "learning_rate": 9.824010429011614e-07,
+      "loss": 4.8481,
+      "step": 7460
+    },
+    {
+      "epoch": 0.3221777146056769,
+      "learning_rate": 9.823525609256825e-07,
+      "loss": 4.8534,
+      "step": 7480
+    },
+    {
+      "epoch": 0.32303915234526426,
+      "learning_rate": 9.823040789502038e-07,
+      "loss": 4.9954,
+      "step": 7500
+    },
+    {
+      "epoch": 0.3239005900848516,
+      "learning_rate": 9.822555969747246e-07,
+      "loss": 4.8983,
+      "step": 7520
+    },
+    {
+      "epoch": 0.32476202782443897,
+      "learning_rate": 9.82207114999246e-07,
+      "loss": 5.0276,
+      "step": 7540
+    },
+    {
+      "epoch": 0.3256234655640264,
+      "learning_rate": 9.821586330237668e-07,
+      "loss": 4.787,
+      "step": 7560
+    },
+    {
+      "epoch": 0.32648490330361374,
+      "learning_rate": 9.82110151048288e-07,
+      "loss": 5.4519,
+      "step": 7580
+    },
+    {
+      "epoch": 0.3273463410432011,
+      "learning_rate": 9.820616690728091e-07,
+      "loss": 5.1657,
+      "step": 7600
+    },
+    {
+      "epoch": 0.32820777878278845,
+      "learning_rate": 9.820131870973302e-07,
+      "loss": 5.0934,
+      "step": 7620
+    },
+    {
+      "epoch": 0.32906921652237586,
+      "learning_rate": 9.819647051218515e-07,
+      "loss": 5.1082,
+      "step": 7640
+    },
+    {
+      "epoch": 0.3299306542619632,
+      "learning_rate": 9.819162231463723e-07,
+      "loss": 4.9206,
+      "step": 7660
+    },
+    {
+      "epoch": 0.3307920920015506,
+      "learning_rate": 9.818677411708936e-07,
+      "loss": 5.1762,
+      "step": 7680
+    },
+    {
+      "epoch": 0.331653529741138,
+      "learning_rate": 9.818192591954147e-07,
+      "loss": 4.9105,
+      "step": 7700
+    },
+    {
+      "epoch": 0.33251496748072534,
+      "learning_rate": 9.817707772199358e-07,
+      "loss": 5.1632,
+      "step": 7720
+    },
+    {
+      "epoch": 0.3333764052203127,
+      "learning_rate": 9.817222952444568e-07,
+      "loss": 5.2379,
+      "step": 7740
+    },
+    {
+      "epoch": 0.33423784295990006,
+      "learning_rate": 9.81673813268978e-07,
+      "loss": 5.1562,
+      "step": 7760
+    },
+    {
+      "epoch": 0.33509928069948747,
+      "learning_rate": 9.81625331293499e-07,
+      "loss": 5.0009,
+      "step": 7780
+    },
+    {
+      "epoch": 0.3359607184390748,
+      "learning_rate": 9.8157684931802e-07,
+      "loss": 4.8898,
+      "step": 7800
+    },
+    {
+      "epoch": 0.3368221561786622,
+      "learning_rate": 9.815283673425413e-07,
+      "loss": 4.9866,
+      "step": 7820
+    },
+    {
+      "epoch": 0.33768359391824954,
+      "learning_rate": 9.814798853670624e-07,
+      "loss": 4.8798,
+      "step": 7840
+    },
+    {
+      "epoch": 0.33854503165783695,
+      "learning_rate": 9.814314033915835e-07,
+      "loss": 4.9858,
+      "step": 7860
+    },
+    {
+      "epoch": 0.3394064693974243,
+      "learning_rate": 9.813829214161047e-07,
+      "loss": 5.1652,
+      "step": 7880
+    },
+    {
+      "epoch": 0.34026790713701166,
+      "learning_rate": 9.813344394406256e-07,
+      "loss": 5.0168,
+      "step": 7900
+    },
+    {
+      "epoch": 0.341129344876599,
+      "learning_rate": 9.812859574651469e-07,
+      "loss": 4.9382,
+      "step": 7920
+    },
+    {
+      "epoch": 0.34199078261618643,
+      "learning_rate": 9.812374754896678e-07,
+      "loss": 5.1187,
+      "step": 7940
+    },
+    {
+      "epoch": 0.3428522203557738,
+      "learning_rate": 9.81188993514189e-07,
+      "loss": 4.8829,
+      "step": 7960
+    },
+    {
+      "epoch": 0.34371365809536114,
+      "learning_rate": 9.811405115387101e-07,
+      "loss": 4.8803,
+      "step": 7980
+    },
+    {
+      "epoch": 0.34457509583494855,
+      "learning_rate": 9.810920295632312e-07,
+      "loss": 4.9934,
+      "step": 8000
+    },
+    {
+      "epoch": 0.3454365335745359,
+      "learning_rate": 9.810435475877525e-07,
+      "loss": 5.0201,
+      "step": 8020
+    },
+    {
+      "epoch": 0.34629797131412327,
+      "learning_rate": 9.809950656122733e-07,
+      "loss": 5.2234,
+      "step": 8040
+    },
+    {
+      "epoch": 0.3471594090537106,
+      "learning_rate": 9.809465836367946e-07,
+      "loss": 5.102,
+      "step": 8060
+    },
+    {
+      "epoch": 0.34802084679329803,
+      "learning_rate": 9.808981016613157e-07,
+      "loss": 5.0268,
+      "step": 8080
+    },
+    {
+      "epoch": 0.3488822845328854,
+      "learning_rate": 9.808496196858367e-07,
+      "loss": 4.6825,
+      "step": 8100
+    },
+    {
+      "epoch": 0.34974372227247275,
+      "learning_rate": 9.80801137710358e-07,
+      "loss": 5.0273,
+      "step": 8120
+    },
+    {
+      "epoch": 0.3506051600120601,
+      "learning_rate": 9.807526557348789e-07,
+      "loss": 4.76,
+      "step": 8140
+    },
+    {
+      "epoch": 0.3514665977516475,
+      "learning_rate": 9.807041737594002e-07,
+      "loss": 4.8518,
+      "step": 8160
+    },
+    {
+      "epoch": 0.35232803549123487,
+      "learning_rate": 9.80655691783921e-07,
+      "loss": 5.0516,
+      "step": 8180
+    },
+    {
+      "epoch": 0.3531894732308222,
+      "learning_rate": 9.806072098084423e-07,
+      "loss": 5.2637,
+      "step": 8200
+    },
+    {
+      "epoch": 0.35405091097040964,
+      "learning_rate": 9.805587278329634e-07,
+      "loss": 5.178,
+      "step": 8220
+    },
+    {
+      "epoch": 0.354912348709997,
+      "learning_rate": 9.805102458574845e-07,
+      "loss": 5.163,
+      "step": 8240
+    },
+    {
+      "epoch": 0.35577378644958435,
+      "learning_rate": 9.804617638820057e-07,
+      "loss": 4.9961,
+      "step": 8260
+    },
+    {
+      "epoch": 0.3566352241891717,
+      "learning_rate": 9.804132819065266e-07,
+      "loss": 4.8015,
+      "step": 8280
+    },
+    {
+      "epoch": 0.3574966619287591,
+      "learning_rate": 9.803647999310479e-07,
+      "loss": 4.8818,
+      "step": 8300
+    },
+    {
+      "epoch": 0.3583580996683465,
+      "learning_rate": 9.803163179555687e-07,
+      "loss": 4.9811,
+      "step": 8320
+    },
+    {
+      "epoch": 0.35921953740793383,
+      "learning_rate": 9.8026783598009e-07,
+      "loss": 4.9989,
+      "step": 8340
+    },
+    {
+      "epoch": 0.3600809751475212,
+      "learning_rate": 9.80219354004611e-07,
+      "loss": 4.7071,
+      "step": 8360
+    },
+    {
+      "epoch": 0.3609424128871086,
+      "learning_rate": 9.801708720291322e-07,
+      "loss": 5.0305,
+      "step": 8380
+    },
+    {
+      "epoch": 0.36180385062669596,
+      "learning_rate": 9.801223900536535e-07,
+      "loss": 4.7833,
+      "step": 8400
+    },
+    {
+      "epoch": 0.3626652883662833,
+      "learning_rate": 9.800739080781743e-07,
+      "loss": 4.714,
+      "step": 8420
+    },
+    {
+      "epoch": 0.3635267261058707,
+      "learning_rate": 9.800254261026956e-07,
+      "loss": 4.7957,
+      "step": 8440
+    },
+    {
+      "epoch": 0.3643881638454581,
+      "learning_rate": 9.799769441272167e-07,
+      "loss": 4.9877,
+      "step": 8460
+    },
+    {
+      "epoch": 0.36524960158504544,
+      "learning_rate": 9.799284621517377e-07,
+      "loss": 5.1407,
+      "step": 8480
+    },
+    {
+      "epoch": 0.3661110393246328,
+      "learning_rate": 9.79879980176259e-07,
+      "loss": 5.0109,
+      "step": 8500
+    },
+    {
+      "epoch": 0.3669724770642202,
+      "learning_rate": 9.798314982007799e-07,
+      "loss": 5.0929,
+      "step": 8520
+    },
+    {
+      "epoch": 0.36783391480380756,
+      "learning_rate": 9.797830162253012e-07,
+      "loss": 4.8868,
+      "step": 8540
+    },
+    {
+      "epoch": 0.3686953525433949,
+      "learning_rate": 9.79734534249822e-07,
+      "loss": 5.0758,
+      "step": 8560
+    },
+    {
+      "epoch": 0.3695567902829823,
+      "learning_rate": 9.796860522743433e-07,
+      "loss": 4.7914,
+      "step": 8580
+    },
+    {
+      "epoch": 0.3704182280225697,
+      "learning_rate": 9.796375702988644e-07,
+      "loss": 4.9024,
+      "step": 8600
+    },
+    {
+      "epoch": 0.37127966576215704,
+      "learning_rate": 9.795890883233855e-07,
+      "loss": 5.1092,
+      "step": 8620
+    },
+    {
+      "epoch": 0.3721411035017444,
+      "learning_rate": 9.795406063479067e-07,
+      "loss": 5.0201,
+      "step": 8640
+    },
+    {
+      "epoch": 0.3730025412413318,
+      "learning_rate": 9.794921243724276e-07,
+      "loss": 4.9426,
+      "step": 8660
+    },
+    {
+      "epoch": 0.37386397898091916,
+      "learning_rate": 9.794436423969489e-07,
+      "loss": 5.0351,
+      "step": 8680
+    },
+    {
+      "epoch": 0.3747254167205065,
+      "learning_rate": 9.7939516042147e-07,
+      "loss": 4.855,
+      "step": 8700
+    },
+    {
+      "epoch": 0.3755868544600939,
+      "learning_rate": 9.79346678445991e-07,
+      "loss": 4.9019,
+      "step": 8720
+    },
+    {
+      "epoch": 0.3764482921996813,
+      "learning_rate": 9.792981964705123e-07,
+      "loss": 4.9067,
+      "step": 8740
+    },
+    {
+      "epoch": 0.37730972993926865,
+      "learning_rate": 9.792497144950332e-07,
+      "loss": 4.9602,
+      "step": 8760
+    },
+    {
+      "epoch": 0.378171167678856,
+      "learning_rate": 9.792012325195544e-07,
+      "loss": 5.1773,
+      "step": 8780
+    },
+    {
+      "epoch": 0.37903260541844336,
+      "learning_rate": 9.791527505440753e-07,
+      "loss": 4.8999,
+      "step": 8800
+    },
+    {
+      "epoch": 0.37989404315803077,
+      "learning_rate": 9.791042685685966e-07,
+      "loss": 5.1569,
+      "step": 8820
+    },
+    {
+      "epoch": 0.3807554808976181,
+      "learning_rate": 9.790557865931177e-07,
+      "loss": 5.0223,
+      "step": 8840
+    },
+    {
+      "epoch": 0.3816169186372055,
+      "learning_rate": 9.790073046176387e-07,
+      "loss": 4.9746,
+      "step": 8860
+    },
+    {
+      "epoch": 0.3824783563767929,
+      "learning_rate": 9.7895882264216e-07,
+      "loss": 4.7316,
+      "step": 8880
+    },
+    {
+      "epoch": 0.38333979411638025,
+      "learning_rate": 9.789103406666809e-07,
+      "loss": 4.8671,
+      "step": 8900
+    },
+    {
+      "epoch": 0.3842012318559676,
+      "learning_rate": 9.788618586912022e-07,
+      "loss": 4.8685,
+      "step": 8920
+    },
+    {
+      "epoch": 0.38506266959555496,
+      "learning_rate": 9.78813376715723e-07,
+      "loss": 5.0705,
+      "step": 8940
+    },
+    {
+      "epoch": 0.3859241073351424,
+      "learning_rate": 9.787648947402443e-07,
+      "loss": 5.0033,
+      "step": 8960
+    },
+    {
+      "epoch": 0.38678554507472973,
+      "learning_rate": 9.787164127647654e-07,
+      "loss": 5.13,
+      "step": 8980
+    },
+    {
+      "epoch": 0.3876469828143171,
+      "learning_rate": 9.786679307892864e-07,
+      "loss": 4.9016,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38850842055390444,
+      "learning_rate": 9.786194488138077e-07,
+      "loss": 4.5963,
+      "step": 9020
+    },
+    {
+      "epoch": 0.38936985829349186,
+      "learning_rate": 9.785709668383288e-07,
+      "loss": 5.1174,
+      "step": 9040
+    },
+    {
+      "epoch": 0.3902312960330792,
+      "learning_rate": 9.785224848628499e-07,
+      "loss": 4.6056,
+      "step": 9060
+    },
+    {
+      "epoch": 0.39109273377266657,
+      "learning_rate": 9.78474002887371e-07,
+      "loss": 5.1854,
+      "step": 9080
+    },
+    {
+      "epoch": 0.391954171512254,
+      "learning_rate": 9.78425520911892e-07,
+      "loss": 5.0354,
+      "step": 9100
+    },
+    {
+      "epoch": 0.39281560925184134,
+      "learning_rate": 9.783770389364133e-07,
+      "loss": 4.973,
+      "step": 9120
+    },
+    {
+      "epoch": 0.3936770469914287,
+      "learning_rate": 9.783285569609342e-07,
+      "loss": 4.9599,
+      "step": 9140
+    },
+    {
+      "epoch": 0.39453848473101605,
+      "learning_rate": 9.782800749854554e-07,
+      "loss": 4.8212,
+      "step": 9160
+    },
+    {
+      "epoch": 0.39539992247060346,
+      "learning_rate": 9.782315930099763e-07,
+      "loss": 4.7895,
+      "step": 9180
+    },
+    {
+      "epoch": 0.3962613602101908,
+      "learning_rate": 9.781831110344976e-07,
+      "loss": 5.0378,
+      "step": 9200
+    },
+    {
+      "epoch": 0.3971227979497782,
+      "learning_rate": 9.781346290590187e-07,
+      "loss": 5.1935,
+      "step": 9220
+    },
+    {
+      "epoch": 0.39798423568936553,
+      "learning_rate": 9.780861470835397e-07,
+      "loss": 5.0377,
+      "step": 9240
+    },
+    {
+      "epoch": 0.39884567342895294,
+      "learning_rate": 9.78037665108061e-07,
+      "loss": 5.2153,
+      "step": 9260
+    },
+    {
+      "epoch": 0.3997071111685403,
+      "learning_rate": 9.77989183132582e-07,
+      "loss": 5.0034,
+      "step": 9280
+    },
+    {
+      "epoch": 0.40056854890812765,
+      "learning_rate": 9.779407011571032e-07,
+      "loss": 5.0281,
+      "step": 9300
+    },
+    {
+      "epoch": 0.401429986647715,
+      "learning_rate": 9.778922191816242e-07,
+      "loss": 4.5347,
+      "step": 9320
+    },
+    {
+      "epoch": 0.4022914243873024,
+      "learning_rate": 9.778437372061453e-07,
+      "loss": 4.7715,
+      "step": 9340
+    },
+    {
+      "epoch": 0.4031528621268898,
+      "learning_rate": 9.777952552306664e-07,
+      "loss": 5.0449,
+      "step": 9360
+    },
+    {
+      "epoch": 0.40401429986647713,
+      "learning_rate": 9.777467732551874e-07,
+      "loss": 5.1752,
+      "step": 9380
+    },
+    {
+      "epoch": 0.40487573760606455,
+      "learning_rate": 9.776982912797087e-07,
+      "loss": 5.063,
+      "step": 9400
+    },
+    {
+      "epoch": 0.4057371753456519,
+      "learning_rate": 9.776498093042298e-07,
+      "loss": 4.8293,
+      "step": 9420
+    },
+    {
+      "epoch": 0.40659861308523926,
+      "learning_rate": 9.776013273287509e-07,
+      "loss": 5.0506,
+      "step": 9440
+    },
+    {
+      "epoch": 0.4074600508248266,
+      "learning_rate": 9.77552845353272e-07,
+      "loss": 5.0681,
+      "step": 9460
+    },
+    {
+      "epoch": 0.408321488564414,
+      "learning_rate": 9.77504363377793e-07,
+      "loss": 4.9795,
+      "step": 9480
+    },
+    {
+      "epoch": 0.4091829263040014,
+      "learning_rate": 9.774558814023143e-07,
+      "loss": 4.7326,
+      "step": 9500
+    },
+    {
+      "epoch": 0.41004436404358874,
+      "learning_rate": 9.774073994268352e-07,
+      "loss": 4.5815,
+      "step": 9520
+    },
+    {
+      "epoch": 0.4109058017831761,
+      "learning_rate": 9.773589174513564e-07,
+      "loss": 5.0059,
+      "step": 9540
+    },
+    {
+      "epoch": 0.4117672395227635,
+      "learning_rate": 9.773104354758773e-07,
+      "loss": 4.8963,
+      "step": 9560
+    },
+    {
+      "epoch": 0.41262867726235086,
+      "learning_rate": 9.772619535003986e-07,
+      "loss": 4.9941,
+      "step": 9580
+    },
+    {
+      "epoch": 0.4134901150019382,
+      "learning_rate": 9.772134715249196e-07,
+      "loss": 5.0495,
+      "step": 9600
+    },
+    {
+      "epoch": 0.41435155274152563,
+      "learning_rate": 9.771649895494407e-07,
+      "loss": 4.5602,
+      "step": 9620
+    },
+    {
+      "epoch": 0.415212990481113,
+      "learning_rate": 9.77116507573962e-07,
+      "loss": 5.1274,
+      "step": 9640
+    },
+    {
+      "epoch": 0.41607442822070034,
+      "learning_rate": 9.77068025598483e-07,
+      "loss": 5.2212,
+      "step": 9660
+    },
+    {
+      "epoch": 0.4169358659602877,
+      "learning_rate": 9.770195436230041e-07,
+      "loss": 4.8636,
+      "step": 9680
+    },
+    {
+      "epoch": 0.4177973036998751,
+      "learning_rate": 9.769710616475252e-07,
+      "loss": 5.0909,
+      "step": 9700
+    },
+    {
+      "epoch": 0.41865874143946247,
+      "learning_rate": 9.769225796720463e-07,
+      "loss": 5.0281,
+      "step": 9720
+    },
+    {
+      "epoch": 0.4195201791790498,
+      "learning_rate": 9.768740976965676e-07,
+      "loss": 4.9283,
+      "step": 9740
+    },
+    {
+      "epoch": 0.4203816169186372,
+      "learning_rate": 9.768256157210884e-07,
+      "loss": 4.6899,
+      "step": 9760
+    },
+    {
+      "epoch": 0.4212430546582246,
+      "learning_rate": 9.767771337456097e-07,
+      "loss": 5.1294,
+      "step": 9780
+    },
+    {
+      "epoch": 0.42210449239781195,
+      "learning_rate": 9.767286517701308e-07,
+      "loss": 5.0745,
+      "step": 9800
+    },
+    {
+      "epoch": 0.4229659301373993,
+      "learning_rate": 9.766801697946519e-07,
+      "loss": 5.1475,
+      "step": 9820
+    },
+    {
+      "epoch": 0.4238273678769867,
+      "learning_rate": 9.76631687819173e-07,
+      "loss": 4.9102,
+      "step": 9840
+    },
+    {
+      "epoch": 0.42468880561657407,
+      "learning_rate": 9.76583205843694e-07,
+      "loss": 4.8146,
+      "step": 9860
+    },
+    {
+      "epoch": 0.42555024335616143,
+      "learning_rate": 9.765347238682153e-07,
+      "loss": 4.8416,
+      "step": 9880
+    },
+    {
+      "epoch": 0.4264116810957488,
+      "learning_rate": 9.764862418927361e-07,
+      "loss": 4.9126,
+      "step": 9900
+    },
+    {
+      "epoch": 0.4272731188353362,
+      "learning_rate": 9.764377599172574e-07,
+      "loss": 4.7237,
+      "step": 9920
+    },
+    {
+      "epoch": 0.42813455657492355,
+      "learning_rate": 9.763892779417783e-07,
+      "loss": 4.93,
+      "step": 9940
+    },
+    {
+      "epoch": 0.4289959943145109,
+      "learning_rate": 9.763407959662996e-07,
+      "loss": 4.8873,
+      "step": 9960
+    },
+    {
+      "epoch": 0.42985743205409827,
+      "learning_rate": 9.762923139908206e-07,
+      "loss": 4.7226,
+      "step": 9980
+    },
+    {
+      "epoch": 0.4307188697936857,
+      "learning_rate": 9.762438320153417e-07,
+      "loss": 5.0115,
+      "step": 10000
+    },
+    {
+      "epoch": 0.43158030753327303,
+      "learning_rate": 9.76195350039863e-07,
+      "loss": 4.9169,
+      "step": 10020
+    },
+    {
+      "epoch": 0.4324417452728604,
+      "learning_rate": 9.76146868064384e-07,
+      "loss": 4.9183,
+      "step": 10040
+    },
+    {
+      "epoch": 0.4333031830124478,
+      "learning_rate": 9.760983860889051e-07,
+      "loss": 4.8978,
+      "step": 10060
+    },
+    {
+      "epoch": 0.43416462075203516,
+      "learning_rate": 9.760499041134262e-07,
+      "loss": 4.6671,
+      "step": 10080
+    },
+    {
+      "epoch": 0.4350260584916225,
+      "learning_rate": 9.760014221379473e-07,
+      "loss": 4.9651,
+      "step": 10100
+    },
+    {
+      "epoch": 0.43588749623120987,
+      "learning_rate": 9.759529401624686e-07,
+      "loss": 4.8849,
+      "step": 10120
+    },
+    {
+      "epoch": 0.4367489339707973,
+      "learning_rate": 9.759044581869894e-07,
+      "loss": 4.7284,
+      "step": 10140
+    },
+    {
+      "epoch": 0.43761037171038464,
+      "learning_rate": 9.758559762115107e-07,
+      "loss": 4.8962,
+      "step": 10160
+    },
+    {
+      "epoch": 0.438471809449972,
+      "learning_rate": 9.758074942360318e-07,
+      "loss": 5.2482,
+      "step": 10180
+    },
+    {
+      "epoch": 0.43933324718955935,
+      "learning_rate": 9.757590122605529e-07,
+      "loss": 4.751,
+      "step": 10200
+    },
+    {
+      "epoch": 0.44019468492914676,
+      "learning_rate": 9.75710530285074e-07,
+      "loss": 4.873,
+      "step": 10220
+    },
+    {
+      "epoch": 0.4410561226687341,
+      "learning_rate": 9.75662048309595e-07,
+      "loss": 4.8341,
+      "step": 10240
+    },
+    {
+      "epoch": 0.4419175604083215,
+      "learning_rate": 9.756135663341163e-07,
+      "loss": 4.8262,
+      "step": 10260
+    },
+    {
+      "epoch": 0.4427789981479089,
+      "learning_rate": 9.755650843586374e-07,
+      "loss": 5.131,
+      "step": 10280
+    },
+    {
+      "epoch": 0.44364043588749624,
+      "learning_rate": 9.755166023831584e-07,
+      "loss": 5.0421,
+      "step": 10300
+    },
+    {
+      "epoch": 0.4445018736270836,
+      "learning_rate": 9.754681204076795e-07,
+      "loss": 4.9971,
+      "step": 10320
+    },
+    {
+      "epoch": 0.44536331136667096,
+      "learning_rate": 9.754196384322006e-07,
+      "loss": 5.0195,
+      "step": 10340
+    },
+    {
+      "epoch": 0.44622474910625837,
+      "learning_rate": 9.753711564567218e-07,
+      "loss": 4.7907,
+      "step": 10360
+    },
+    {
+      "epoch": 0.4470861868458457,
+      "learning_rate": 9.753226744812427e-07,
+      "loss": 4.9427,
+      "step": 10380
+    },
+    {
+      "epoch": 0.4479476245854331,
+      "learning_rate": 9.75274192505764e-07,
+      "loss": 4.9174,
+      "step": 10400
+    },
+    {
+      "epoch": 0.44880906232502044,
+      "learning_rate": 9.75225710530285e-07,
+      "loss": 5.0763,
+      "step": 10420
+    },
+    {
+      "epoch": 0.44967050006460785,
+      "learning_rate": 9.751772285548061e-07,
+      "loss": 4.8748,
+      "step": 10440
+    },
+    {
+      "epoch": 0.4505319378041952,
+      "learning_rate": 9.751287465793272e-07,
+      "loss": 4.9506,
+      "step": 10460
+    },
+    {
+      "epoch": 0.45139337554378256,
+      "learning_rate": 9.750802646038483e-07,
+      "loss": 4.6341,
+      "step": 10480
+    },
+    {
+      "epoch": 0.45225481328336997,
+      "learning_rate": 9.750317826283696e-07,
+      "loss": 4.7974,
+      "step": 10500
+    },
+    {
+      "epoch": 0.45311625102295733,
+      "learning_rate": 9.749833006528904e-07,
+      "loss": 4.8822,
+      "step": 10520
+    },
+    {
+      "epoch": 0.4539776887625447,
+      "learning_rate": 9.749348186774117e-07,
+      "loss": 4.7732,
+      "step": 10540
+    },
+    {
+      "epoch": 0.45483912650213204,
+      "learning_rate": 9.748863367019328e-07,
+      "loss": 5.0418,
+      "step": 10560
+    },
+    {
+      "epoch": 0.45570056424171945,
+      "learning_rate": 9.748378547264538e-07,
+      "loss": 5.0333,
+      "step": 10580
+    },
+    {
+      "epoch": 0.4565620019813068,
+      "learning_rate": 9.74789372750975e-07,
+      "loss": 4.7601,
+      "step": 10600
+    },
+    {
+      "epoch": 0.45742343972089416,
+      "learning_rate": 9.74740890775496e-07,
+      "loss": 4.8413,
+      "step": 10620
+    },
+    {
+      "epoch": 0.4582848774604815,
+      "learning_rate": 9.746924088000173e-07,
+      "loss": 4.8576,
+      "step": 10640
+    },
+    {
+      "epoch": 0.45914631520006893,
+      "learning_rate": 9.746439268245383e-07,
+      "loss": 4.9137,
+      "step": 10660
+    },
+    {
+      "epoch": 0.4600077529396563,
+      "learning_rate": 9.745954448490594e-07,
+      "loss": 4.7917,
+      "step": 10680
+    },
+    {
+      "epoch": 0.46086919067924365,
+      "learning_rate": 9.745469628735805e-07,
+      "loss": 5.0347,
+      "step": 10700
+    },
+    {
+      "epoch": 0.461730628418831,
+      "learning_rate": 9.744984808981016e-07,
+      "loss": 5.0721,
+      "step": 10720
+    },
+    {
+      "epoch": 0.4625920661584184,
+      "learning_rate": 9.744499989226228e-07,
+      "loss": 5.223,
+      "step": 10740
+    },
+    {
+      "epoch": 0.46345350389800577,
+      "learning_rate": 9.744015169471437e-07,
+      "loss": 4.7617,
+      "step": 10760
+    },
+    {
+      "epoch": 0.4643149416375931,
+      "learning_rate": 9.74353034971665e-07,
+      "loss": 4.7315,
+      "step": 10780
+    },
+    {
+      "epoch": 0.46517637937718054,
+      "learning_rate": 9.74304552996186e-07,
+      "loss": 5.0133,
+      "step": 10800
+    },
+    {
+      "epoch": 0.4660378171167679,
+      "learning_rate": 9.742560710207071e-07,
+      "loss": 4.5951,
+      "step": 10820
+    },
+    {
+      "epoch": 0.46689925485635525,
+      "learning_rate": 9.742075890452282e-07,
+      "loss": 5.0109,
+      "step": 10840
+    },
+    {
+      "epoch": 0.4677606925959426,
+      "learning_rate": 9.741591070697493e-07,
+      "loss": 4.7951,
+      "step": 10860
+    },
+    {
+      "epoch": 0.46862213033553,
+      "learning_rate": 9.741106250942706e-07,
+      "loss": 4.9083,
+      "step": 10880
+    },
+    {
+      "epoch": 0.4694835680751174,
+      "learning_rate": 9.740621431187916e-07,
+      "loss": 5.0881,
+      "step": 10900
+    },
+    {
+      "epoch": 0.47034500581470473,
+      "learning_rate": 9.740136611433127e-07,
+      "loss": 4.9049,
+      "step": 10920
+    },
+    {
+      "epoch": 0.4712064435542921,
+      "learning_rate": 9.73965179167834e-07,
+      "loss": 5.0503,
+      "step": 10940
+    },
+    {
+      "epoch": 0.4720678812938795,
+      "learning_rate": 9.739166971923548e-07,
+      "loss": 5.1389,
+      "step": 10960
+    },
+    {
+      "epoch": 0.47292931903346686,
+      "learning_rate": 9.73868215216876e-07,
+      "loss": 4.6824,
+      "step": 10980
+    },
+    {
+      "epoch": 0.4737907567730542,
+      "learning_rate": 9.73819733241397e-07,
+      "loss": 5.0677,
+      "step": 11000
+    },
+    {
+      "epoch": 0.4746521945126416,
+      "learning_rate": 9.737712512659183e-07,
+      "loss": 4.8373,
+      "step": 11020
+    },
+    {
+      "epoch": 0.475513632252229,
+      "learning_rate": 9.737227692904393e-07,
+      "loss": 4.7048,
+      "step": 11040
+    },
+    {
+      "epoch": 0.47637506999181634,
+      "learning_rate": 9.736742873149604e-07,
+      "loss": 4.7661,
+      "step": 11060
+    },
+    {
+      "epoch": 0.4772365077314037,
+      "learning_rate": 9.736258053394815e-07,
+      "loss": 4.5163,
+      "step": 11080
+    },
+    {
+      "epoch": 0.4780979454709911,
+      "learning_rate": 9.735773233640026e-07,
+      "loss": 4.8271,
+      "step": 11100
+    },
+    {
+      "epoch": 0.47895938321057846,
+      "learning_rate": 9.735288413885238e-07,
+      "loss": 4.9904,
+      "step": 11120
+    },
+    {
+      "epoch": 0.4798208209501658,
+      "learning_rate": 9.734803594130447e-07,
+      "loss": 5.1391,
+      "step": 11140
+    },
+    {
+      "epoch": 0.4806822586897532,
+      "learning_rate": 9.73431877437566e-07,
+      "loss": 4.9359,
+      "step": 11160
+    },
+    {
+      "epoch": 0.4815436964293406,
+      "learning_rate": 9.73383395462087e-07,
+      "loss": 4.7196,
+      "step": 11180
+    },
+    {
+      "epoch": 0.48240513416892794,
+      "learning_rate": 9.733349134866081e-07,
+      "loss": 4.8989,
+      "step": 11200
+    },
+    {
+      "epoch": 0.4832665719085153,
+      "learning_rate": 9.732864315111292e-07,
+      "loss": 4.5728,
+      "step": 11220
+    },
+    {
+      "epoch": 0.4841280096481027,
+      "learning_rate": 9.732379495356503e-07,
+      "loss": 4.8242,
+      "step": 11240
+    },
+    {
+      "epoch": 0.48498944738769006,
+      "learning_rate": 9.731894675601715e-07,
+      "loss": 4.9628,
+      "step": 11260
+    },
+    {
+      "epoch": 0.4858508851272774,
+      "learning_rate": 9.731409855846926e-07,
+      "loss": 4.9849,
+      "step": 11280
+    },
+    {
+      "epoch": 0.4867123228668648,
+      "learning_rate": 9.730925036092137e-07,
+      "loss": 4.7428,
+      "step": 11300
+    },
+    {
+      "epoch": 0.4875737606064522,
+      "learning_rate": 9.73044021633735e-07,
+      "loss": 4.84,
+      "step": 11320
+    },
+    {
+      "epoch": 0.48843519834603955,
+      "learning_rate": 9.729955396582558e-07,
+      "loss": 4.8301,
+      "step": 11340
+    },
+    {
+      "epoch": 0.4892966360856269,
+      "learning_rate": 9.729470576827771e-07,
+      "loss": 4.8208,
+      "step": 11360
+    },
+    {
+      "epoch": 0.49015807382521426,
+      "learning_rate": 9.72898575707298e-07,
+      "loss": 4.799,
+      "step": 11380
+    },
+    {
+      "epoch": 0.49101951156480167,
+      "learning_rate": 9.728500937318193e-07,
+      "loss": 4.8384,
+      "step": 11400
+    },
+    {
+      "epoch": 0.491880949304389,
+      "learning_rate": 9.728016117563403e-07,
+      "loss": 4.8149,
+      "step": 11420
+    },
+    {
+      "epoch": 0.4927423870439764,
+      "learning_rate": 9.727531297808614e-07,
+      "loss": 5.0723,
+      "step": 11440
+    },
+    {
+      "epoch": 0.4936038247835638,
+      "learning_rate": 9.727046478053825e-07,
+      "loss": 4.8349,
+      "step": 11460
+    },
+    {
+      "epoch": 0.49446526252315115,
+      "learning_rate": 9.726561658299035e-07,
+      "loss": 4.8027,
+      "step": 11480
+    },
+    {
+      "epoch": 0.4953267002627385,
+      "learning_rate": 9.726076838544248e-07,
+      "loss": 4.4655,
+      "step": 11500
+    },
+    {
+      "epoch": 0.49618813800232586,
+      "learning_rate": 9.725592018789457e-07,
+      "loss": 4.6683,
+      "step": 11520
+    },
+    {
+      "epoch": 0.4970495757419133,
+      "learning_rate": 9.72510719903467e-07,
+      "loss": 4.3801,
+      "step": 11540
+    },
+    {
+      "epoch": 0.49791101348150063,
+      "learning_rate": 9.72462237927988e-07,
+      "loss": 4.8508,
+      "step": 11560
+    },
+    {
+      "epoch": 0.498772451221088,
+      "learning_rate": 9.724137559525091e-07,
+      "loss": 4.8787,
+      "step": 11580
+    },
+    {
+      "epoch": 0.49963388896067534,
+      "learning_rate": 9.723652739770302e-07,
+      "loss": 4.9498,
+      "step": 11600
+    },
+    {
+      "epoch": 0.5004953267002628,
+      "learning_rate": 9.723167920015513e-07,
+      "loss": 5.0344,
+      "step": 11620
+    },
+    {
+      "epoch": 0.5013567644398501,
+      "learning_rate": 9.722683100260725e-07,
+      "loss": 5.0325,
+      "step": 11640
+    },
+    {
+      "epoch": 0.5022182021794375,
+      "learning_rate": 9.722198280505936e-07,
+      "loss": 4.8234,
+      "step": 11660
+    },
+    {
+      "epoch": 0.5030796399190248,
+      "learning_rate": 9.721713460751147e-07,
+      "loss": 4.5882,
+      "step": 11680
+    },
+    {
+      "epoch": 0.5039410776586122,
+      "learning_rate": 9.72122864099636e-07,
+      "loss": 4.864,
+      "step": 11700
+    },
+    {
+      "epoch": 0.5048025153981996,
+      "learning_rate": 9.720743821241568e-07,
+      "loss": 4.8641,
+      "step": 11720
+    },
+    {
+      "epoch": 0.505663953137787,
+      "learning_rate": 9.720259001486781e-07,
+      "loss": 4.8875,
+      "step": 11740
+    },
+    {
+      "epoch": 0.5065253908773744,
+      "learning_rate": 9.71977418173199e-07,
+      "loss": 5.13,
+      "step": 11760
+    },
+    {
+      "epoch": 0.5073868286169617,
+      "learning_rate": 9.719289361977203e-07,
+      "loss": 4.8252,
+      "step": 11780
+    },
+    {
+      "epoch": 0.5082482663565491,
+      "learning_rate": 9.718804542222413e-07,
+      "loss": 4.9937,
+      "step": 11800
+    },
+    {
+      "epoch": 0.5091097040961364,
+      "learning_rate": 9.718319722467624e-07,
+      "loss": 4.9656,
+      "step": 11820
+    },
+    {
+      "epoch": 0.5099711418357238,
+      "learning_rate": 9.717834902712837e-07,
+      "loss": 4.8202,
+      "step": 11840
+    },
+    {
+      "epoch": 0.5108325795753111,
+      "learning_rate": 9.717350082958045e-07,
+      "loss": 4.8806,
+      "step": 11860
+    },
+    {
+      "epoch": 0.5116940173148986,
+      "learning_rate": 9.716865263203258e-07,
+      "loss": 4.7003,
+      "step": 11880
+    },
+    {
+      "epoch": 0.512555455054486,
+      "learning_rate": 9.71638044344847e-07,
+      "loss": 4.9151,
+      "step": 11900
+    },
+    {
+      "epoch": 0.5134168927940733,
+      "learning_rate": 9.71589562369368e-07,
+      "loss": 4.7234,
+      "step": 11920
+    },
+    {
+      "epoch": 0.5142783305336607,
+      "learning_rate": 9.715410803938892e-07,
+      "loss": 4.8458,
+      "step": 11940
+    },
+    {
+      "epoch": 0.515139768273248,
+      "learning_rate": 9.714925984184101e-07,
+      "loss": 4.7352,
+      "step": 11960
+    },
+    {
+      "epoch": 0.5160012060128354,
+      "learning_rate": 9.714441164429314e-07,
+      "loss": 5.0708,
+      "step": 11980
+    },
+    {
+      "epoch": 0.5168626437524227,
+      "learning_rate": 9.713956344674523e-07,
+      "loss": 4.8615,
+      "step": 12000
+    },
+    {
+      "epoch": 0.5177240814920102,
+      "learning_rate": 9.713471524919735e-07,
+      "loss": 4.8537,
+      "step": 12020
+    },
+    {
+      "epoch": 0.5185855192315976,
+      "learning_rate": 9.712986705164946e-07,
+      "loss": 4.6005,
+      "step": 12040
+    },
+    {
+      "epoch": 0.5194469569711849,
+      "learning_rate": 9.712501885410157e-07,
+      "loss": 4.7112,
+      "step": 12060
+    },
+    {
+      "epoch": 0.5203083947107723,
+      "learning_rate": 9.71201706565537e-07,
+      "loss": 4.7967,
+      "step": 12080
+    },
+    {
+      "epoch": 0.5211698324503596,
+      "learning_rate": 9.711532245900578e-07,
+      "loss": 4.8298,
+      "step": 12100
+    },
+    {
+      "epoch": 0.522031270189947,
+      "learning_rate": 9.71104742614579e-07,
+      "loss": 4.527,
+      "step": 12120
+    },
+    {
+      "epoch": 0.5228927079295344,
+      "learning_rate": 9.710562606391e-07,
+      "loss": 4.7663,
+      "step": 12140
+    },
+    {
+      "epoch": 0.5237541456691218,
+      "learning_rate": 9.710077786636212e-07,
+      "loss": 4.8853,
+      "step": 12160
+    },
+    {
+      "epoch": 0.5246155834087092,
+      "learning_rate": 9.709592966881423e-07,
+      "loss": 4.6954,
+      "step": 12180
+    },
+    {
+      "epoch": 0.5254770211482965,
+      "learning_rate": 9.709108147126634e-07,
+      "loss": 4.7645,
+      "step": 12200
+    },
+    {
+      "epoch": 0.5263384588878839,
+      "learning_rate": 9.708623327371847e-07,
+      "loss": 4.6727,
+      "step": 12220
+    },
+    {
+      "epoch": 0.5271998966274712,
+      "learning_rate": 9.708138507617055e-07,
+      "loss": 4.5477,
+      "step": 12240
+    },
+    {
+      "epoch": 0.5280613343670586,
+      "learning_rate": 9.707653687862268e-07,
+      "loss": 4.6893,
+      "step": 12260
+    },
+    {
+      "epoch": 0.528922772106646,
+      "learning_rate": 9.707168868107479e-07,
+      "loss": 4.8461,
+      "step": 12280
+    },
+    {
+      "epoch": 0.5297842098462333,
+      "learning_rate": 9.70668404835269e-07,
+      "loss": 4.5713,
+      "step": 12300
+    },
+    {
+      "epoch": 0.5306456475858208,
+      "learning_rate": 9.706199228597902e-07,
+      "loss": 4.6324,
+      "step": 12320
+    },
+    {
+      "epoch": 0.5315070853254081,
+      "learning_rate": 9.70571440884311e-07,
+      "loss": 4.9228,
+      "step": 12340
+    },
+    {
+      "epoch": 0.5323685230649955,
+      "learning_rate": 9.705229589088324e-07,
+      "loss": 4.947,
+      "step": 12360
+    },
+    {
+      "epoch": 0.5332299608045828,
+      "learning_rate": 9.704744769333532e-07,
+      "loss": 4.7576,
+      "step": 12380
+    },
+    {
+      "epoch": 0.5340913985441702,
+      "learning_rate": 9.704259949578745e-07,
+      "loss": 4.6725,
+      "step": 12400
+    },
+    {
+      "epoch": 0.5349528362837576,
+      "learning_rate": 9.703775129823956e-07,
+      "loss": 4.677,
+      "step": 12420
+    },
+    {
+      "epoch": 0.5358142740233449,
+      "learning_rate": 9.703290310069167e-07,
+      "loss": 4.7095,
+      "step": 12440
+    },
+    {
+      "epoch": 0.5366757117629324,
+      "learning_rate": 9.70280549031438e-07,
+      "loss": 4.8679,
+      "step": 12460
+    },
+    {
+      "epoch": 0.5375371495025197,
+      "learning_rate": 9.702320670559588e-07,
+      "loss": 4.7676,
+      "step": 12480
+    },
+    {
+      "epoch": 0.5383985872421071,
+      "learning_rate": 9.7018358508048e-07,
+      "loss": 4.7246,
+      "step": 12500
+    },
+    {
+      "epoch": 0.5392600249816945,
+      "learning_rate": 9.701351031050012e-07,
+      "loss": 4.7469,
+      "step": 12520
+    },
+    {
+      "epoch": 0.5401214627212818,
+      "learning_rate": 9.700866211295222e-07,
+      "loss": 4.7057,
+      "step": 12540
+    },
+    {
+      "epoch": 0.5409829004608692,
+      "learning_rate": 9.700381391540435e-07,
+      "loss": 4.5612,
+      "step": 12560
+    },
+    {
+      "epoch": 0.5418443382004565,
+      "learning_rate": 9.699896571785644e-07,
+      "loss": 4.6163,
+      "step": 12580
+    },
+    {
+      "epoch": 0.542705775940044,
+      "learning_rate": 9.699411752030857e-07,
+      "loss": 4.5884,
+      "step": 12600
+    },
+    {
+      "epoch": 0.5435672136796313,
+      "learning_rate": 9.698926932276065e-07,
+      "loss": 4.7224,
+      "step": 12620
+    },
+    {
+      "epoch": 0.5444286514192187,
+      "learning_rate": 9.698442112521278e-07,
+      "loss": 4.7125,
+      "step": 12640
+    },
+    {
+      "epoch": 0.5452900891588061,
+      "learning_rate": 9.697957292766489e-07,
+      "loss": 4.5919,
+      "step": 12660
+    },
+    {
+      "epoch": 0.5461515268983934,
+      "learning_rate": 9.6974724730117e-07,
+      "loss": 4.9423,
+      "step": 12680
+    },
+    {
+      "epoch": 0.5470129646379808,
+      "learning_rate": 9.696987653256912e-07,
+      "loss": 4.7632,
+      "step": 12700
+    },
+    {
+      "epoch": 0.5478744023775681,
+      "learning_rate": 9.69650283350212e-07,
+      "loss": 4.7262,
+      "step": 12720
+    },
+    {
+      "epoch": 0.5487358401171555,
+      "learning_rate": 9.696018013747334e-07,
+      "loss": 4.8491,
+      "step": 12740
+    },
+    {
+      "epoch": 0.549597277856743,
+      "learning_rate": 9.695533193992542e-07,
+      "loss": 4.9667,
+      "step": 12760
+    },
+    {
+      "epoch": 0.5504587155963303,
+      "learning_rate": 9.695048374237755e-07,
+      "loss": 4.669,
+      "step": 12780
+    },
+    {
+      "epoch": 0.5513201533359177,
+      "learning_rate": 9.694563554482966e-07,
+      "loss": 4.7343,
+      "step": 12800
+    },
+    {
+      "epoch": 0.552181591075505,
+      "learning_rate": 9.694078734728177e-07,
+      "loss": 5.1538,
+      "step": 12820
+    },
+    {
+      "epoch": 0.5530430288150924,
+      "learning_rate": 9.69359391497339e-07,
+      "loss": 4.6751,
+      "step": 12840
+    },
+    {
+      "epoch": 0.5539044665546797,
+      "learning_rate": 9.693109095218598e-07,
+      "loss": 4.249,
+      "step": 12860
+    },
+    {
+      "epoch": 0.5547659042942671,
+      "learning_rate": 9.69262427546381e-07,
+      "loss": 4.5602,
+      "step": 12880
+    },
+    {
+      "epoch": 0.5556273420338546,
+      "learning_rate": 9.692139455709022e-07,
+      "loss": 4.7908,
+      "step": 12900
+    },
+    {
+      "epoch": 0.5564887797734419,
+      "learning_rate": 9.691654635954232e-07,
+      "loss": 4.8403,
+      "step": 12920
+    },
+    {
+      "epoch": 0.5573502175130293,
+      "learning_rate": 9.691169816199445e-07,
+      "loss": 4.654,
+      "step": 12940
+    },
+    {
+      "epoch": 0.5582116552526166,
+      "learning_rate": 9.690684996444654e-07,
+      "loss": 4.9628,
+      "step": 12960
+    },
+    {
+      "epoch": 0.559073092992204,
+      "learning_rate": 9.690200176689867e-07,
+      "loss": 4.6785,
+      "step": 12980
+    },
+    {
+      "epoch": 0.5599345307317913,
+      "learning_rate": 9.689715356935075e-07,
+      "loss": 4.7578,
+      "step": 13000
+    },
+    {
+      "epoch": 0.5607959684713787,
+      "learning_rate": 9.689230537180288e-07,
+      "loss": 4.6124,
+      "step": 13020
+    },
+    {
+      "epoch": 0.561657406210966,
+      "learning_rate": 9.688745717425499e-07,
+      "loss": 4.7457,
+      "step": 13040
+    },
+    {
+      "epoch": 0.5625188439505535,
+      "learning_rate": 9.68826089767071e-07,
+      "loss": 4.656,
+      "step": 13060
+    },
+    {
+      "epoch": 0.5633802816901409,
+      "learning_rate": 9.687776077915922e-07,
+      "loss": 4.6385,
+      "step": 13080
+    },
+    {
+      "epoch": 0.5642417194297282,
+      "learning_rate": 9.687291258161133e-07,
+      "loss": 4.791,
+      "step": 13100
+    },
+    {
+      "epoch": 0.5651031571693156,
+      "learning_rate": 9.686806438406344e-07,
+      "loss": 4.6335,
+      "step": 13120
+    },
+    {
+      "epoch": 0.5659645949089029,
+      "learning_rate": 9.686321618651552e-07,
+      "loss": 4.9381,
+      "step": 13140
+    },
+    {
+      "epoch": 0.5668260326484903,
+      "learning_rate": 9.685836798896765e-07,
+      "loss": 4.5365,
+      "step": 13160
+    },
+    {
+      "epoch": 0.5676874703880777,
+      "learning_rate": 9.685351979141976e-07,
+      "loss": 4.7251,
+      "step": 13180
+    },
+    {
+      "epoch": 0.5685489081276651,
+      "learning_rate": 9.684867159387187e-07,
+      "loss": 4.6211,
+      "step": 13200
+    },
+    {
+      "epoch": 0.5694103458672525,
+      "learning_rate": 9.6843823396324e-07,
+      "loss": 4.8273,
+      "step": 13220
+    },
+    {
+      "epoch": 0.5702717836068398,
+      "learning_rate": 9.683897519877608e-07,
+      "loss": 4.8895,
+      "step": 13240
+    },
+    {
+      "epoch": 0.5711332213464272,
+      "learning_rate": 9.68341270012282e-07,
+      "loss": 5.1973,
+      "step": 13260
+    },
+    {
+      "epoch": 0.5719946590860145,
+      "learning_rate": 9.682927880368032e-07,
+      "loss": 4.7367,
+      "step": 13280
+    },
+    {
+      "epoch": 0.5728560968256019,
+      "learning_rate": 9.682443060613242e-07,
+      "loss": 4.807,
+      "step": 13300
+    },
+    {
+      "epoch": 0.5737175345651893,
+      "learning_rate": 9.681958240858455e-07,
+      "loss": 4.7434,
+      "step": 13320
+    },
+    {
+      "epoch": 0.5745789723047767,
+      "learning_rate": 9.681473421103664e-07,
+      "loss": 4.8386,
+      "step": 13340
+    },
+    {
+      "epoch": 0.5754404100443641,
+      "learning_rate": 9.680988601348877e-07,
+      "loss": 4.9904,
+      "step": 13360
+    },
+    {
+      "epoch": 0.5763018477839514,
+      "learning_rate": 9.680503781594085e-07,
+      "loss": 4.6482,
+      "step": 13380
+    },
+    {
+      "epoch": 0.5771632855235388,
+      "learning_rate": 9.680018961839298e-07,
+      "loss": 4.8677,
+      "step": 13400
+    },
+    {
+      "epoch": 0.5780247232631262,
+      "learning_rate": 9.679534142084509e-07,
+      "loss": 4.8015,
+      "step": 13420
+    },
+    {
+      "epoch": 0.5788861610027135,
+      "learning_rate": 9.67904932232972e-07,
+      "loss": 4.9403,
+      "step": 13440
+    },
+    {
+      "epoch": 0.5797475987423009,
+      "learning_rate": 9.678564502574932e-07,
+      "loss": 4.8006,
+      "step": 13460
+    },
+    {
+      "epoch": 0.5806090364818882,
+      "learning_rate": 9.678079682820143e-07,
+      "loss": 4.9156,
+      "step": 13480
+    },
+    {
+      "epoch": 0.5814704742214757,
+      "learning_rate": 9.677594863065354e-07,
+      "loss": 4.575,
+      "step": 13500
+    },
+    {
+      "epoch": 0.582331911961063,
+      "learning_rate": 9.677110043310564e-07,
+      "loss": 4.6135,
+      "step": 13520
+    },
+    {
+      "epoch": 0.5831933497006504,
+      "learning_rate": 9.676625223555775e-07,
+      "loss": 4.8542,
+      "step": 13540
+    },
+    {
+      "epoch": 0.5840547874402378,
+      "learning_rate": 9.676140403800988e-07,
+      "loss": 4.5898,
+      "step": 13560
+    },
+    {
+      "epoch": 0.5849162251798251,
+      "learning_rate": 9.675655584046197e-07,
+      "loss": 4.6785,
+      "step": 13580
+    },
+    {
+      "epoch": 0.5857776629194125,
+      "learning_rate": 9.67517076429141e-07,
+      "loss": 4.4886,
+      "step": 13600
+    },
+    {
+      "epoch": 0.5866391006589998,
+      "learning_rate": 9.67468594453662e-07,
+      "loss": 4.8392,
+      "step": 13620
+    },
+    {
+      "epoch": 0.5875005383985873,
+      "learning_rate": 9.67420112478183e-07,
+      "loss": 4.7093,
+      "step": 13640
+    },
+    {
+      "epoch": 0.5883619761381746,
+      "learning_rate": 9.673716305027042e-07,
+      "loss": 4.5766,
+      "step": 13660
+    },
+    {
+      "epoch": 0.589223413877762,
+      "learning_rate": 9.673231485272252e-07,
+      "loss": 4.6722,
+      "step": 13680
+    },
+    {
+      "epoch": 0.5900848516173494,
+      "learning_rate": 9.672746665517465e-07,
+      "loss": 4.7496,
+      "step": 13700
+    },
+    {
+      "epoch": 0.5909462893569367,
+      "learning_rate": 9.672261845762674e-07,
+      "loss": 4.5997,
+      "step": 13720
+    },
+    {
+      "epoch": 0.5918077270965241,
+      "learning_rate": 9.671777026007886e-07,
+      "loss": 4.811,
+      "step": 13740
+    },
+    {
+      "epoch": 0.5926691648361114,
+      "learning_rate": 9.671292206253095e-07,
+      "loss": 4.7544,
+      "step": 13760
+    },
+    {
+      "epoch": 0.5935306025756989,
+      "learning_rate": 9.670807386498308e-07,
+      "loss": 4.872,
+      "step": 13780
+    },
+    {
+      "epoch": 0.5943920403152863,
+      "learning_rate": 9.670322566743519e-07,
+      "loss": 4.616,
+      "step": 13800
+    },
+    {
+      "epoch": 0.5952534780548736,
+      "learning_rate": 9.66983774698873e-07,
+      "loss": 4.6266,
+      "step": 13820
+    },
+    {
+      "epoch": 0.596114915794461,
+      "learning_rate": 9.669352927233942e-07,
+      "loss": 4.8612,
+      "step": 13840
+    },
+    {
+      "epoch": 0.5969763535340483,
+      "learning_rate": 9.668868107479153e-07,
+      "loss": 4.4442,
+      "step": 13860
+    },
+    {
+      "epoch": 0.5978377912736357,
+      "learning_rate": 9.668383287724364e-07,
+      "loss": 4.6665,
+      "step": 13880
+    },
+    {
+      "epoch": 0.598699229013223,
+      "learning_rate": 9.667898467969574e-07,
+      "loss": 4.5713,
+      "step": 13900
+    },
+    {
+      "epoch": 0.5995606667528104,
+      "learning_rate": 9.667413648214785e-07,
+      "loss": 4.6818,
+      "step": 13920
+    },
+    {
+      "epoch": 0.6004221044923979,
+      "learning_rate": 9.666928828459998e-07,
+      "loss": 4.8311,
+      "step": 13940
+    },
+    {
+      "epoch": 0.6012835422319852,
+      "learning_rate": 9.666444008705206e-07,
+      "loss": 4.7409,
+      "step": 13960
+    },
+    {
+      "epoch": 0.6021449799715726,
+      "learning_rate": 9.66595918895042e-07,
+      "loss": 4.6599,
+      "step": 13980
+    },
+    {
+      "epoch": 0.6030064177111599,
+      "learning_rate": 9.66547436919563e-07,
+      "loss": 4.5503,
+      "step": 14000
+    },
+    {
+      "epoch": 0.6038678554507473,
+      "learning_rate": 9.66498954944084e-07,
+      "loss": 4.6705,
+      "step": 14020
+    },
+    {
+      "epoch": 0.6047292931903346,
+      "learning_rate": 9.664504729686051e-07,
+      "loss": 4.9944,
+      "step": 14040
+    },
+    {
+      "epoch": 0.605590730929922,
+      "learning_rate": 9.664019909931262e-07,
+      "loss": 4.8301,
+      "step": 14060
+    },
+    {
+      "epoch": 0.6064521686695095,
+      "learning_rate": 9.663535090176475e-07,
+      "loss": 4.8497,
+      "step": 14080
+    },
+    {
+      "epoch": 0.6073136064090968,
+      "learning_rate": 9.663050270421686e-07,
+      "loss": 4.7864,
+      "step": 14100
+    },
+    {
+      "epoch": 0.6081750441486842,
+      "learning_rate": 9.662565450666896e-07,
+      "loss": 4.9266,
+      "step": 14120
+    },
+    {
+      "epoch": 0.6090364818882715,
+      "learning_rate": 9.662080630912107e-07,
+      "loss": 4.8071,
+      "step": 14140
+    },
+    {
+      "epoch": 0.6098979196278589,
+      "learning_rate": 9.661595811157318e-07,
+      "loss": 4.6743,
+      "step": 14160
+    },
+    {
+      "epoch": 0.6107593573674462,
+      "learning_rate": 9.66111099140253e-07,
+      "loss": 4.532,
+      "step": 14180
+    },
+    {
+      "epoch": 0.6116207951070336,
+      "learning_rate": 9.66062617164774e-07,
+      "loss": 4.9564,
+      "step": 14200
+    },
+    {
+      "epoch": 0.612482232846621,
+      "learning_rate": 9.660141351892952e-07,
+      "loss": 4.8359,
+      "step": 14220
+    },
+    {
+      "epoch": 0.6133436705862084,
+      "learning_rate": 9.659656532138163e-07,
+      "loss": 4.7172,
+      "step": 14240
+    },
+    {
+      "epoch": 0.6142051083257958,
+      "learning_rate": 9.659171712383374e-07,
+      "loss": 4.8027,
+      "step": 14260
+    },
+    {
+      "epoch": 0.6150665460653831,
+      "learning_rate": 9.658686892628584e-07,
+      "loss": 4.5783,
+      "step": 14280
+    },
+    {
+      "epoch": 0.6159279838049705,
+      "learning_rate": 9.658202072873795e-07,
+      "loss": 4.7292,
+      "step": 14300
+    },
+    {
+      "epoch": 0.6167894215445578,
+      "learning_rate": 9.657717253119008e-07,
+      "loss": 4.5805,
+      "step": 14320
+    },
+    {
+      "epoch": 0.6176508592841452,
+      "learning_rate": 9.657232433364216e-07,
+      "loss": 4.6905,
+      "step": 14340
+    },
+    {
+      "epoch": 0.6185122970237326,
+      "learning_rate": 9.65674761360943e-07,
+      "loss": 5.0062,
+      "step": 14360
+    },
+    {
+      "epoch": 0.61937373476332,
+      "learning_rate": 9.65626279385464e-07,
+      "loss": 4.5667,
+      "step": 14380
+    },
+    {
+      "epoch": 0.6202351725029074,
+      "learning_rate": 9.65577797409985e-07,
+      "loss": 4.7865,
+      "step": 14400
+    },
+    {
+      "epoch": 0.6210966102424947,
+      "learning_rate": 9.655293154345061e-07,
+      "loss": 4.8152,
+      "step": 14420
+    },
+    {
+      "epoch": 0.6219580479820821,
+      "learning_rate": 9.654808334590272e-07,
+      "loss": 5.001,
+      "step": 14440
+    },
+    {
+      "epoch": 0.6228194857216695,
+      "learning_rate": 9.654323514835485e-07,
+      "loss": 4.568,
+      "step": 14460
+    },
+    {
+      "epoch": 0.6236809234612568,
+      "learning_rate": 9.653838695080696e-07,
+      "loss": 5.1036,
+      "step": 14480
+    },
+    {
+      "epoch": 0.6245423612008442,
+      "learning_rate": 9.653353875325906e-07,
+      "loss": 4.6376,
+      "step": 14500
+    },
+    {
+      "epoch": 0.6254037989404316,
+      "learning_rate": 9.652869055571117e-07,
+      "loss": 4.8825,
+      "step": 14520
+    },
+    {
+      "epoch": 0.626265236680019,
+      "learning_rate": 9.652384235816328e-07,
+      "loss": 4.8115,
+      "step": 14540
+    },
+    {
+      "epoch": 0.6271266744196063,
+      "learning_rate": 9.65189941606154e-07,
+      "loss": 4.907,
+      "step": 14560
+    },
+    {
+      "epoch": 0.6279881121591937,
+      "learning_rate": 9.65141459630675e-07,
+      "loss": 4.7589,
+      "step": 14580
+    },
+    {
+      "epoch": 0.6288495498987811,
+      "learning_rate": 9.650929776551962e-07,
+      "loss": 4.6099,
+      "step": 14600
+    },
+    {
+      "epoch": 0.6297109876383684,
+      "learning_rate": 9.650444956797173e-07,
+      "loss": 4.7292,
+      "step": 14620
+    },
+    {
+      "epoch": 0.6305724253779558,
+      "learning_rate": 9.649960137042383e-07,
+      "loss": 4.8569,
+      "step": 14640
+    },
+    {
+      "epoch": 0.6314338631175431,
+      "learning_rate": 9.649475317287594e-07,
+      "loss": 4.476,
+      "step": 14660
+    },
+    {
+      "epoch": 0.6322953008571306,
+      "learning_rate": 9.648990497532805e-07,
+      "loss": 5.1189,
+      "step": 14680
+    },
+    {
+      "epoch": 0.633156738596718,
+      "learning_rate": 9.648505677778018e-07,
+      "loss": 4.4871,
+      "step": 14700
+    },
+    {
+      "epoch": 0.6340181763363053,
+      "learning_rate": 9.648020858023228e-07,
+      "loss": 4.7954,
+      "step": 14720
+    },
+    {
+      "epoch": 0.6348796140758927,
+      "learning_rate": 9.64753603826844e-07,
+      "loss": 4.8858,
+      "step": 14740
+    },
+    {
+      "epoch": 0.63574105181548,
+      "learning_rate": 9.64705121851365e-07,
+      "loss": 4.7299,
+      "step": 14760
+    },
+    {
+      "epoch": 0.6366024895550674,
+      "learning_rate": 9.64656639875886e-07,
+      "loss": 4.6986,
+      "step": 14780
+    },
+    {
+      "epoch": 0.6374639272946547,
+      "learning_rate": 9.646081579004071e-07,
+      "loss": 4.716,
+      "step": 14800
+    },
+    {
+      "epoch": 0.6383253650342422,
+      "learning_rate": 9.645596759249282e-07,
+      "loss": 4.8712,
+      "step": 14820
+    },
+    {
+      "epoch": 0.6391868027738296,
+      "learning_rate": 9.645111939494495e-07,
+      "loss": 4.5896,
+      "step": 14840
+    },
+    {
+      "epoch": 0.6400482405134169,
+      "learning_rate": 9.644627119739706e-07,
+      "loss": 4.8666,
+      "step": 14860
+    },
+    {
+      "epoch": 0.6409096782530043,
+      "learning_rate": 9.644142299984916e-07,
+      "loss": 4.6907,
+      "step": 14880
+    },
+    {
+      "epoch": 0.6417711159925916,
+      "learning_rate": 9.643657480230127e-07,
+      "loss": 4.7755,
+      "step": 14900
+    },
+    {
+      "epoch": 0.642632553732179,
+      "learning_rate": 9.643172660475338e-07,
+      "loss": 4.8946,
+      "step": 14920
+    },
+    {
+      "epoch": 0.6434939914717663,
+      "learning_rate": 9.64268784072055e-07,
+      "loss": 4.7275,
+      "step": 14940
+    },
+    {
+      "epoch": 0.6443554292113538,
+      "learning_rate": 9.64220302096576e-07,
+      "loss": 4.7162,
+      "step": 14960
+    },
+    {
+      "epoch": 0.6452168669509412,
+      "learning_rate": 9.641718201210972e-07,
+      "loss": 4.7541,
+      "step": 14980
+    },
+    {
+      "epoch": 0.6460783046905285,
+      "learning_rate": 9.641233381456183e-07,
+      "loss": 4.7376,
+      "step": 15000
+    },
+    {
+      "epoch": 0.6469397424301159,
+      "learning_rate": 9.640748561701393e-07,
+      "loss": 4.7821,
+      "step": 15020
+    },
+    {
+      "epoch": 0.6478011801697032,
+      "learning_rate": 9.640263741946604e-07,
+      "loss": 4.5146,
+      "step": 15040
+    },
+    {
+      "epoch": 0.6486626179092906,
+      "learning_rate": 9.639778922191815e-07,
+      "loss": 4.7323,
+      "step": 15060
+    },
+    {
+      "epoch": 0.6495240556488779,
+      "learning_rate": 9.639294102437028e-07,
+      "loss": 4.5574,
+      "step": 15080
+    },
+    {
+      "epoch": 0.6503854933884653,
+      "learning_rate": 9.638809282682238e-07,
+      "loss": 4.8495,
+      "step": 15100
+    },
+    {
+      "epoch": 0.6512469311280528,
+      "learning_rate": 9.63832446292745e-07,
+      "loss": 4.5805,
+      "step": 15120
+    },
+    {
+      "epoch": 0.6521083688676401,
+      "learning_rate": 9.637839643172662e-07,
+      "loss": 4.3699,
+      "step": 15140
+    },
+    {
+      "epoch": 0.6529698066072275,
+      "learning_rate": 9.63735482341787e-07,
+      "loss": 4.8785,
+      "step": 15160
+    },
+    {
+      "epoch": 0.6538312443468148,
+      "learning_rate": 9.636870003663083e-07,
+      "loss": 4.7165,
+      "step": 15180
+    },
+    {
+      "epoch": 0.6546926820864022,
+      "learning_rate": 9.636385183908292e-07,
+      "loss": 4.8152,
+      "step": 15200
+    },
+    {
+      "epoch": 0.6555541198259895,
+      "learning_rate": 9.635900364153505e-07,
+      "loss": 4.4532,
+      "step": 15220
+    },
+    {
+      "epoch": 0.6564155575655769,
+      "learning_rate": 9.635415544398716e-07,
+      "loss": 4.6873,
+      "step": 15240
+    },
+    {
+      "epoch": 0.6572769953051644,
+      "learning_rate": 9.634930724643926e-07,
+      "loss": 4.6663,
+      "step": 15260
+    },
+    {
+      "epoch": 0.6581384330447517,
+      "learning_rate": 9.634445904889137e-07,
+      "loss": 4.776,
+      "step": 15280
+    },
+    {
+      "epoch": 0.6589998707843391,
+      "learning_rate": 9.633961085134348e-07,
+      "loss": 4.6215,
+      "step": 15300
+    },
+    {
+      "epoch": 0.6598613085239264,
+      "learning_rate": 9.63347626537956e-07,
+      "loss": 4.921,
+      "step": 15320
+    },
+    {
+      "epoch": 0.6607227462635138,
+      "learning_rate": 9.63299144562477e-07,
+      "loss": 4.8543,
+      "step": 15340
+    },
+    {
+      "epoch": 0.6615841840031012,
+      "learning_rate": 9.632506625869982e-07,
+      "loss": 4.643,
+      "step": 15360
+    },
+    {
+      "epoch": 0.6624456217426885,
+      "learning_rate": 9.632021806115193e-07,
+      "loss": 4.901,
+      "step": 15380
+    },
+    {
+      "epoch": 0.663307059482276,
+      "learning_rate": 9.631536986360403e-07,
+      "loss": 4.6514,
+      "step": 15400
+    },
+    {
+      "epoch": 0.6641684972218633,
+      "learning_rate": 9.631052166605614e-07,
+      "loss": 4.6956,
+      "step": 15420
+    },
+    {
+      "epoch": 0.6650299349614507,
+      "learning_rate": 9.630567346850825e-07,
+      "loss": 4.6573,
+      "step": 15440
+    },
+    {
+      "epoch": 0.665891372701038,
+      "learning_rate": 9.630082527096038e-07,
+      "loss": 4.6353,
+      "step": 15460
+    },
+    {
+      "epoch": 0.6667528104406254,
+      "learning_rate": 9.629597707341248e-07,
+      "loss": 4.6602,
+      "step": 15480
+    },
+    {
+      "epoch": 0.6676142481802128,
+      "learning_rate": 9.62911288758646e-07,
+      "loss": 4.6295,
+      "step": 15500
+    },
+    {
+      "epoch": 0.6684756859198001,
+      "learning_rate": 9.628628067831672e-07,
+      "loss": 4.7094,
+      "step": 15520
+    },
+    {
+      "epoch": 0.6693371236593875,
+      "learning_rate": 9.62814324807688e-07,
+      "loss": 4.7686,
+      "step": 15540
+    },
+    {
+      "epoch": 0.6701985613989749,
+      "learning_rate": 9.627658428322093e-07,
+      "loss": 4.3885,
+      "step": 15560
+    },
+    {
+      "epoch": 0.6710599991385623,
+      "learning_rate": 9.627173608567302e-07,
+      "loss": 5.0316,
+      "step": 15580
+    },
+    {
+      "epoch": 0.6719214368781496,
+      "learning_rate": 9.626688788812515e-07,
+      "loss": 4.5627,
+      "step": 15600
+    },
+    {
+      "epoch": 0.672782874617737,
+      "learning_rate": 9.626203969057725e-07,
+      "loss": 4.3815,
+      "step": 15620
+    },
+    {
+      "epoch": 0.6736443123573244,
+      "learning_rate": 9.625719149302936e-07,
+      "loss": 4.74,
+      "step": 15640
+    },
+    {
+      "epoch": 0.6745057500969117,
+      "learning_rate": 9.625234329548147e-07,
+      "loss": 4.6831,
+      "step": 15660
+    },
+    {
+      "epoch": 0.6753671878364991,
+      "learning_rate": 9.624749509793358e-07,
+      "loss": 4.609,
+      "step": 15680
+    },
+    {
+      "epoch": 0.6762286255760865,
+      "learning_rate": 9.62426469003857e-07,
+      "loss": 4.823,
+      "step": 15700
+    },
+    {
+      "epoch": 0.6770900633156739,
+      "learning_rate": 9.623779870283781e-07,
+      "loss": 4.9465,
+      "step": 15720
+    },
+    {
+      "epoch": 0.6779515010552613,
+      "learning_rate": 9.623295050528992e-07,
+      "loss": 4.7628,
+      "step": 15740
+    },
+    {
+      "epoch": 0.6788129387948486,
+      "learning_rate": 9.622810230774205e-07,
+      "loss": 4.5836,
+      "step": 15760
+    },
+    {
+      "epoch": 0.679674376534436,
+      "learning_rate": 9.622325411019413e-07,
+      "loss": 4.8521,
+      "step": 15780
+    },
+    {
+      "epoch": 0.6805358142740233,
+      "learning_rate": 9.621840591264626e-07,
+      "loss": 4.7569,
+      "step": 15800
+    },
+    {
+      "epoch": 0.6813972520136107,
+      "learning_rate": 9.621355771509835e-07,
+      "loss": 4.4686,
+      "step": 15820
+    },
+    {
+      "epoch": 0.682258689753198,
+      "learning_rate": 9.620870951755048e-07,
+      "loss": 4.7357,
+      "step": 15840
+    },
+    {
+      "epoch": 0.6831201274927855,
+      "learning_rate": 9.620386132000258e-07,
+      "loss": 4.6195,
+      "step": 15860
+    },
+    {
+      "epoch": 0.6839815652323729,
+      "learning_rate": 9.61990131224547e-07,
+      "loss": 4.7323,
+      "step": 15880
+    },
+    {
+      "epoch": 0.6848430029719602,
+      "learning_rate": 9.619416492490682e-07,
+      "loss": 4.6049,
+      "step": 15900
+    },
+    {
+      "epoch": 0.6857044407115476,
+      "learning_rate": 9.61893167273589e-07,
+      "loss": 4.5214,
+      "step": 15920
+    },
+    {
+      "epoch": 0.6865658784511349,
+      "learning_rate": 9.618446852981103e-07,
+      "loss": 4.4264,
+      "step": 15940
+    },
+    {
+      "epoch": 0.6874273161907223,
+      "learning_rate": 9.617962033226312e-07,
+      "loss": 4.9132,
+      "step": 15960
+    },
+    {
+      "epoch": 0.6882887539303096,
+      "learning_rate": 9.617477213471525e-07,
+      "loss": 4.5938,
+      "step": 15980
+    },
+    {
+      "epoch": 0.6891501916698971,
+      "learning_rate": 9.616992393716735e-07,
+      "loss": 4.8424,
+      "step": 16000
+    },
+    {
+      "epoch": 0.6900116294094845,
+      "learning_rate": 9.616507573961946e-07,
+      "loss": 4.7597,
+      "step": 16020
+    },
+    {
+      "epoch": 0.6908730671490718,
+      "learning_rate": 9.616022754207159e-07,
+      "loss": 4.7149,
+      "step": 16040
+    },
+    {
+      "epoch": 0.6917345048886592,
+      "learning_rate": 9.615537934452368e-07,
+      "loss": 4.6915,
+      "step": 16060
+    },
+    {
+      "epoch": 0.6925959426282465,
+      "learning_rate": 9.61505311469758e-07,
+      "loss": 4.714,
+      "step": 16080
+    },
+    {
+      "epoch": 0.6934573803678339,
+      "learning_rate": 9.614568294942791e-07,
+      "loss": 4.497,
+      "step": 16100
+    },
+    {
+      "epoch": 0.6943188181074212,
+      "learning_rate": 9.614083475188002e-07,
+      "loss": 4.8497,
+      "step": 16120
+    },
+    {
+      "epoch": 0.6951802558470087,
+      "learning_rate": 9.613598655433215e-07,
+      "loss": 4.691,
+      "step": 16140
+    },
+    {
+      "epoch": 0.6960416935865961,
+      "learning_rate": 9.613113835678423e-07,
+      "loss": 4.8417,
+      "step": 16160
+    },
+    {
+      "epoch": 0.6969031313261834,
+      "learning_rate": 9.612629015923636e-07,
+      "loss": 4.7323,
+      "step": 16180
+    },
+    {
+      "epoch": 0.6977645690657708,
+      "learning_rate": 9.612144196168845e-07,
+      "loss": 4.632,
+      "step": 16200
+    },
+    {
+      "epoch": 0.6986260068053581,
+      "learning_rate": 9.611659376414057e-07,
+      "loss": 4.8886,
+      "step": 16220
+    },
+    {
+      "epoch": 0.6994874445449455,
+      "learning_rate": 9.611174556659268e-07,
+      "loss": 4.3369,
+      "step": 16240
+    },
+    {
+      "epoch": 0.7003488822845328,
+      "learning_rate": 9.610689736904479e-07,
+      "loss": 4.6227,
+      "step": 16260
+    },
+    {
+      "epoch": 0.7012103200241202,
+      "learning_rate": 9.610204917149692e-07,
+      "loss": 4.5117,
+      "step": 16280
+    },
+    {
+      "epoch": 0.7020717577637077,
+      "learning_rate": 9.6097200973949e-07,
+      "loss": 4.7759,
+      "step": 16300
+    },
+    {
+      "epoch": 0.702933195503295,
+      "learning_rate": 9.609235277640113e-07,
+      "loss": 4.6303,
+      "step": 16320
+    },
+    {
+      "epoch": 0.7037946332428824,
+      "learning_rate": 9.608750457885324e-07,
+      "loss": 4.836,
+      "step": 16340
+    },
+    {
+      "epoch": 0.7046560709824697,
+      "learning_rate": 9.608265638130535e-07,
+      "loss": 4.7166,
+      "step": 16360
+    },
+    {
+      "epoch": 0.7055175087220571,
+      "learning_rate": 9.607780818375745e-07,
+      "loss": 4.489,
+      "step": 16380
+    },
+    {
+      "epoch": 0.7063789464616445,
+      "learning_rate": 9.607295998620956e-07,
+      "loss": 4.6527,
+      "step": 16400
+    },
+    {
+      "epoch": 0.7072403842012318,
+      "learning_rate": 9.606811178866169e-07,
+      "loss": 4.4992,
+      "step": 16420
+    },
+    {
+      "epoch": 0.7081018219408193,
+      "learning_rate": 9.606326359111377e-07,
+      "loss": 4.9018,
+      "step": 16440
+    },
+    {
+      "epoch": 0.7089632596804066,
+      "learning_rate": 9.60584153935659e-07,
+      "loss": 4.854,
+      "step": 16460
+    },
+    {
+      "epoch": 0.709824697419994,
+      "learning_rate": 9.6053567196018e-07,
+      "loss": 4.8965,
+      "step": 16480
+    },
+    {
+      "epoch": 0.7106861351595813,
+      "learning_rate": 9.604871899847012e-07,
+      "loss": 4.5009,
+      "step": 16500
+    },
+    {
+      "epoch": 0.7115475728991687,
+      "learning_rate": 9.604387080092225e-07,
+      "loss": 4.8104,
+      "step": 16520
+    },
+    {
+      "epoch": 0.7124090106387561,
+      "learning_rate": 9.603902260337433e-07,
+      "loss": 4.698,
+      "step": 16540
+    },
+    {
+      "epoch": 0.7132704483783434,
+      "learning_rate": 9.603417440582646e-07,
+      "loss": 4.7735,
+      "step": 16560
+    },
+    {
+      "epoch": 0.7141318861179309,
+      "learning_rate": 9.602932620827855e-07,
+      "loss": 4.6545,
+      "step": 16580
+    },
+    {
+      "epoch": 0.7149933238575182,
+      "learning_rate": 9.602447801073067e-07,
+      "loss": 4.988,
+      "step": 16600
+    },
+    {
+      "epoch": 0.7158547615971056,
+      "learning_rate": 9.601962981318278e-07,
+      "loss": 4.6552,
+      "step": 16620
+    },
+    {
+      "epoch": 0.716716199336693,
+      "learning_rate": 9.601478161563489e-07,
+      "loss": 4.4583,
+      "step": 16640
+    },
+    {
+      "epoch": 0.7175776370762803,
+      "learning_rate": 9.600993341808702e-07,
+      "loss": 4.6773,
+      "step": 16660
+    },
+    {
+      "epoch": 0.7184390748158677,
+      "learning_rate": 9.60050852205391e-07,
+      "loss": 4.5659,
+      "step": 16680
+    },
+    {
+      "epoch": 0.719300512555455,
+      "learning_rate": 9.600023702299123e-07,
+      "loss": 4.866,
+      "step": 16700
+    },
+    {
+      "epoch": 0.7201619502950424,
+      "learning_rate": 9.599538882544334e-07,
+      "loss": 4.8379,
+      "step": 16720
+    },
+    {
+      "epoch": 0.7210233880346298,
+      "learning_rate": 9.599054062789545e-07,
+      "loss": 4.5252,
+      "step": 16740
+    },
+    {
+      "epoch": 0.7218848257742172,
+      "learning_rate": 9.598569243034757e-07,
+      "loss": 4.7735,
+      "step": 16760
+    },
+    {
+      "epoch": 0.7227462635138046,
+      "learning_rate": 9.598084423279966e-07,
+      "loss": 4.6124,
+      "step": 16780
+    },
+    {
+      "epoch": 0.7236077012533919,
+      "learning_rate": 9.597599603525179e-07,
+      "loss": 4.8224,
+      "step": 16800
+    },
+    {
+      "epoch": 0.7244691389929793,
+      "learning_rate": 9.597114783770387e-07,
+      "loss": 4.4931,
+      "step": 16820
+    },
+    {
+      "epoch": 0.7253305767325666,
+      "learning_rate": 9.5966299640156e-07,
+      "loss": 4.7042,
+      "step": 16840
+    },
+    {
+      "epoch": 0.726192014472154,
+      "learning_rate": 9.59614514426081e-07,
+      "loss": 4.4158,
+      "step": 16860
+    },
+    {
+      "epoch": 0.7270534522117414,
+      "learning_rate": 9.595660324506022e-07,
+      "loss": 4.6365,
+      "step": 16880
+    },
+    {
+      "epoch": 0.7279148899513288,
+      "learning_rate": 9.595175504751234e-07,
+      "loss": 4.6285,
+      "step": 16900
+    },
+    {
+      "epoch": 0.7287763276909162,
+      "learning_rate": 9.594690684996443e-07,
+      "loss": 4.6141,
+      "step": 16920
+    },
+    {
+      "epoch": 0.7296377654305035,
+      "learning_rate": 9.594205865241656e-07,
+      "loss": 4.6042,
+      "step": 16940
+    },
+    {
+      "epoch": 0.7304992031700909,
+      "learning_rate": 9.593721045486865e-07,
+      "loss": 4.4642,
+      "step": 16960
+    },
+    {
+      "epoch": 0.7313606409096782,
+      "learning_rate": 9.593236225732077e-07,
+      "loss": 4.7874,
+      "step": 16980
+    },
+    {
+      "epoch": 0.7322220786492656,
+      "learning_rate": 9.592751405977288e-07,
+      "loss": 4.7062,
+      "step": 17000
+    },
+    {
+      "epoch": 0.7330835163888529,
+      "learning_rate": 9.592266586222499e-07,
+      "loss": 4.7244,
+      "step": 17020
+    },
+    {
+      "epoch": 0.7339449541284404,
+      "learning_rate": 9.591781766467712e-07,
+      "loss": 4.6183,
+      "step": 17040
+    },
+    {
+      "epoch": 0.7348063918680278,
+      "learning_rate": 9.59129694671292e-07,
+      "loss": 4.592,
+      "step": 17060
+    },
+    {
+      "epoch": 0.7356678296076151,
+      "learning_rate": 9.590812126958133e-07,
+      "loss": 4.8319,
+      "step": 17080
+    },
+    {
+      "epoch": 0.7365292673472025,
+      "learning_rate": 9.590327307203344e-07,
+      "loss": 4.8552,
+      "step": 17100
+    },
+    {
+      "epoch": 0.7373907050867898,
+      "learning_rate": 9.589842487448554e-07,
+      "loss": 4.9682,
+      "step": 17120
+    },
+    {
+      "epoch": 0.7382521428263772,
+      "learning_rate": 9.589357667693767e-07,
+      "loss": 4.8142,
+      "step": 17140
+    },
+    {
+      "epoch": 0.7391135805659645,
+      "learning_rate": 9.588872847938976e-07,
+      "loss": 4.4919,
+      "step": 17160
+    },
+    {
+      "epoch": 0.739975018305552,
+      "learning_rate": 9.588388028184189e-07,
+      "loss": 4.7011,
+      "step": 17180
+    },
+    {
+      "epoch": 0.7408364560451394,
+      "learning_rate": 9.587903208429397e-07,
+      "loss": 4.7958,
+      "step": 17200
+    },
+    {
+      "epoch": 0.7416978937847267,
+      "learning_rate": 9.58741838867461e-07,
+      "loss": 4.6178,
+      "step": 17220
+    },
+    {
+      "epoch": 0.7425593315243141,
+      "learning_rate": 9.58693356891982e-07,
+      "loss": 4.6442,
+      "step": 17240
+    },
+    {
+      "epoch": 0.7434207692639014,
+      "learning_rate": 9.586448749165032e-07,
+      "loss": 4.7595,
+      "step": 17260
+    },
+    {
+      "epoch": 0.7442822070034888,
+      "learning_rate": 9.585963929410244e-07,
+      "loss": 4.7747,
+      "step": 17280
+    },
+    {
+      "epoch": 0.7451436447430762,
+      "learning_rate": 9.585479109655455e-07,
+      "loss": 4.7121,
+      "step": 17300
+    },
+    {
+      "epoch": 0.7460050824826636,
+      "learning_rate": 9.584994289900666e-07,
+      "loss": 4.8845,
+      "step": 17320
+    },
+    {
+      "epoch": 0.746866520222251,
+      "learning_rate": 9.584509470145877e-07,
+      "loss": 4.7855,
+      "step": 17340
+    },
+    {
+      "epoch": 0.7477279579618383,
+      "learning_rate": 9.584024650391087e-07,
+      "loss": 4.6848,
+      "step": 17360
+    },
+    {
+      "epoch": 0.7485893957014257,
+      "learning_rate": 9.5835398306363e-07,
+      "loss": 4.7187,
+      "step": 17380
+    },
+    {
+      "epoch": 0.749450833441013,
+      "learning_rate": 9.583055010881509e-07,
+      "loss": 4.5737,
+      "step": 17400
+    },
+    {
+      "epoch": 0.7503122711806004,
+      "learning_rate": 9.582570191126722e-07,
+      "loss": 4.9443,
+      "step": 17420
+    },
+    {
+      "epoch": 0.7511737089201878,
+      "learning_rate": 9.58208537137193e-07,
+      "loss": 4.9395,
+      "step": 17440
+    },
+    {
+      "epoch": 0.7520351466597751,
+      "learning_rate": 9.581600551617143e-07,
+      "loss": 4.5068,
+      "step": 17460
+    },
+    {
+      "epoch": 0.7528965843993626,
+      "learning_rate": 9.581115731862354e-07,
+      "loss": 4.9245,
+      "step": 17480
+    },
+    {
+      "epoch": 0.7537580221389499,
+      "learning_rate": 9.580630912107564e-07,
+      "loss": 4.6495,
+      "step": 17500
+    },
+    {
+      "epoch": 0.7546194598785373,
+      "learning_rate": 9.580146092352777e-07,
+      "loss": 4.4371,
+      "step": 17520
+    },
+    {
+      "epoch": 0.7554808976181246,
+      "learning_rate": 9.579661272597986e-07,
+      "loss": 4.5391,
+      "step": 17540
+    },
+    {
+      "epoch": 0.756342335357712,
+      "learning_rate": 9.579176452843199e-07,
+      "loss": 4.4885,
+      "step": 17560
+    },
+    {
+      "epoch": 0.7572037730972994,
+      "learning_rate": 9.578691633088407e-07,
+      "loss": 4.5693,
+      "step": 17580
+    },
+    {
+      "epoch": 0.7580652108368867,
+      "learning_rate": 9.57820681333362e-07,
+      "loss": 4.4952,
+      "step": 17600
+    },
+    {
+      "epoch": 0.7589266485764742,
+      "learning_rate": 9.57772199357883e-07,
+      "loss": 4.633,
+      "step": 17620
+    },
+    {
+      "epoch": 0.7597880863160615,
+      "learning_rate": 9.577237173824042e-07,
+      "loss": 4.6817,
+      "step": 17640
+    },
+    {
+      "epoch": 0.7606495240556489,
+      "learning_rate": 9.576752354069254e-07,
+      "loss": 4.647,
+      "step": 17660
+    },
+    {
+      "epoch": 0.7615109617952363,
+      "learning_rate": 9.576267534314465e-07,
+      "loss": 4.6863,
+      "step": 17680
+    },
+    {
+      "epoch": 0.7623723995348236,
+      "learning_rate": 9.575782714559676e-07,
+      "loss": 4.8397,
+      "step": 17700
+    },
+    {
+      "epoch": 0.763233837274411,
+      "learning_rate": 9.575297894804887e-07,
+      "loss": 4.4226,
+      "step": 17720
+    },
+    {
+      "epoch": 0.7640952750139983,
+      "learning_rate": 9.574813075050097e-07,
+      "loss": 4.4665,
+      "step": 17740
+    },
+    {
+      "epoch": 0.7649567127535858,
+      "learning_rate": 9.57432825529531e-07,
+      "loss": 4.6331,
+      "step": 17760
+    },
+    {
+      "epoch": 0.7658181504931731,
+      "learning_rate": 9.573843435540519e-07,
+      "loss": 4.7533,
+      "step": 17780
+    },
+    {
+      "epoch": 0.7666795882327605,
+      "learning_rate": 9.573358615785731e-07,
+      "loss": 4.4459,
+      "step": 17800
+    },
+    {
+      "epoch": 0.7675410259723479,
+      "learning_rate": 9.572873796030942e-07,
+      "loss": 4.5337,
+      "step": 17820
+    },
+    {
+      "epoch": 0.7684024637119352,
+      "learning_rate": 9.572388976276153e-07,
+      "loss": 4.6425,
+      "step": 17840
+    },
+    {
+      "epoch": 0.7692639014515226,
+      "learning_rate": 9.571904156521364e-07,
+      "loss": 4.4908,
+      "step": 17860
+    },
+    {
+      "epoch": 0.7701253391911099,
+      "learning_rate": 9.571419336766574e-07,
+      "loss": 4.6324,
+      "step": 17880
+    },
+    {
+      "epoch": 0.7709867769306973,
+      "learning_rate": 9.570934517011787e-07,
+      "loss": 4.5503,
+      "step": 17900
+    },
+    {
+      "epoch": 0.7718482146702847,
+      "learning_rate": 9.570449697256998e-07,
+      "loss": 4.3948,
+      "step": 17920
+    },
+    {
+      "epoch": 0.7727096524098721,
+      "learning_rate": 9.569964877502209e-07,
+      "loss": 4.7714,
+      "step": 17940
+    },
+    {
+      "epoch": 0.7735710901494595,
+      "learning_rate": 9.56948005774742e-07,
+      "loss": 4.4958,
+      "step": 17960
+    },
+    {
+      "epoch": 0.7744325278890468,
+      "learning_rate": 9.56899523799263e-07,
+      "loss": 4.8341,
+      "step": 17980
+    },
+    {
+      "epoch": 0.7752939656286342,
+      "learning_rate": 9.56851041823784e-07,
+      "loss": 4.8692,
+      "step": 18000
+    },
+    {
+      "epoch": 0.7761554033682215,
+      "learning_rate": 9.568025598483051e-07,
+      "loss": 4.5919,
+      "step": 18020
+    },
+    {
+      "epoch": 0.7770168411078089,
+      "learning_rate": 9.567540778728264e-07,
+      "loss": 4.5073,
+      "step": 18040
+    },
+    {
+      "epoch": 0.7778782788473964,
+      "learning_rate": 9.567055958973475e-07,
+      "loss": 4.541,
+      "step": 18060
+    },
+    {
+      "epoch": 0.7787397165869837,
+      "learning_rate": 9.566571139218686e-07,
+      "loss": 4.5663,
+      "step": 18080
+    },
+    {
+      "epoch": 0.7796011543265711,
+      "learning_rate": 9.566086319463896e-07,
+      "loss": 4.4336,
+      "step": 18100
+    },
+    {
+      "epoch": 0.7804625920661584,
+      "learning_rate": 9.565601499709107e-07,
+      "loss": 4.4726,
+      "step": 18120
+    },
+    {
+      "epoch": 0.7813240298057458,
+      "learning_rate": 9.56511667995432e-07,
+      "loss": 4.5123,
+      "step": 18140
+    },
+    {
+      "epoch": 0.7821854675453331,
+      "learning_rate": 9.564631860199529e-07,
+      "loss": 4.6115,
+      "step": 18160
+    },
+    {
+      "epoch": 0.7830469052849205,
+      "learning_rate": 9.564147040444741e-07,
+      "loss": 4.7271,
+      "step": 18180
+    },
+    {
+      "epoch": 0.783908343024508,
+      "learning_rate": 9.563662220689952e-07,
+      "loss": 4.815,
+      "step": 18200
+    },
+    {
+      "epoch": 0.7847697807640953,
+      "learning_rate": 9.563177400935163e-07,
+      "loss": 4.5781,
+      "step": 18220
+    },
+    {
+      "epoch": 0.7856312185036827,
+      "learning_rate": 9.562692581180374e-07,
+      "loss": 4.5923,
+      "step": 18240
+    },
+    {
+      "epoch": 0.78649265624327,
+      "learning_rate": 9.562207761425584e-07,
+      "loss": 4.6544,
+      "step": 18260
+    },
+    {
+      "epoch": 0.7873540939828574,
+      "learning_rate": 9.561722941670797e-07,
+      "loss": 4.6061,
+      "step": 18280
+    },
+    {
+      "epoch": 0.7882155317224447,
+      "learning_rate": 9.561238121916008e-07,
+      "loss": 4.6143,
+      "step": 18300
+    },
+    {
+      "epoch": 0.7890769694620321,
+      "learning_rate": 9.560753302161219e-07,
+      "loss": 4.6983,
+      "step": 18320
+    },
+    {
+      "epoch": 0.7899384072016195,
+      "learning_rate": 9.56026848240643e-07,
+      "loss": 4.7628,
+      "step": 18340
+    },
+    {
+      "epoch": 0.7907998449412069,
+      "learning_rate": 9.55978366265164e-07,
+      "loss": 4.629,
+      "step": 18360
+    },
+    {
+      "epoch": 0.7916612826807943,
+      "learning_rate": 9.559298842896853e-07,
+      "loss": 4.643,
+      "step": 18380
+    },
+    {
+      "epoch": 0.7925227204203816,
+      "learning_rate": 9.558814023142061e-07,
+      "loss": 4.6185,
+      "step": 18400
+    },
+    {
+      "epoch": 0.793384158159969,
+      "learning_rate": 9.558329203387274e-07,
+      "loss": 4.5335,
+      "step": 18420
+    },
+    {
+      "epoch": 0.7942455958995563,
+      "learning_rate": 9.557844383632485e-07,
+      "loss": 4.6082,
+      "step": 18440
+    },
+    {
+      "epoch": 0.7951070336391437,
+      "learning_rate": 9.557359563877696e-07,
+      "loss": 4.5302,
+      "step": 18460
+    },
+    {
+      "epoch": 0.7959684713787311,
+      "learning_rate": 9.556874744122906e-07,
+      "loss": 4.9049,
+      "step": 18480
+    },
+    {
+      "epoch": 0.7968299091183185,
+      "learning_rate": 9.556389924368117e-07,
+      "loss": 4.5565,
+      "step": 18500
+    },
+    {
+      "epoch": 0.7976913468579059,
+      "learning_rate": 9.55590510461333e-07,
+      "loss": 4.5934,
+      "step": 18520
+    },
+    {
+      "epoch": 0.7985527845974932,
+      "learning_rate": 9.555420284858539e-07,
+      "loss": 4.498,
+      "step": 18540
+    },
+    {
+      "epoch": 0.7994142223370806,
+      "learning_rate": 9.554935465103751e-07,
+      "loss": 4.8571,
+      "step": 18560
+    },
+    {
+      "epoch": 0.800275660076668,
+      "learning_rate": 9.554450645348962e-07,
+      "loss": 4.5185,
+      "step": 18580
+    },
+    {
+      "epoch": 0.8011370978162553,
+      "learning_rate": 9.553965825594173e-07,
+      "loss": 4.5657,
+      "step": 18600
+    },
+    {
+      "epoch": 0.8019985355558427,
+      "learning_rate": 9.553481005839384e-07,
+      "loss": 4.4812,
+      "step": 18620
+    },
+    {
+      "epoch": 0.80285997329543,
+      "learning_rate": 9.552996186084594e-07,
+      "loss": 4.6503,
+      "step": 18640
+    },
+    {
+      "epoch": 0.8037214110350175,
+      "learning_rate": 9.552511366329807e-07,
+      "loss": 4.8931,
+      "step": 18660
+    },
+    {
+      "epoch": 0.8045828487746048,
+      "learning_rate": 9.552026546575018e-07,
+      "loss": 4.8269,
+      "step": 18680
+    },
+    {
+      "epoch": 0.8054442865141922,
+      "learning_rate": 9.551541726820228e-07,
+      "loss": 4.7274,
+      "step": 18700
+    },
+    {
+      "epoch": 0.8063057242537796,
+      "learning_rate": 9.55105690706544e-07,
+      "loss": 4.4974,
+      "step": 18720
+    },
+    {
+      "epoch": 0.8071671619933669,
+      "learning_rate": 9.55057208731065e-07,
+      "loss": 4.5903,
+      "step": 18740
+    },
+    {
+      "epoch": 0.8080285997329543,
+      "learning_rate": 9.550087267555863e-07,
+      "loss": 4.4657,
+      "step": 18760
+    },
+    {
+      "epoch": 0.8088900374725416,
+      "learning_rate": 9.549602447801071e-07,
+      "loss": 4.5626,
+      "step": 18780
+    },
+    {
+      "epoch": 0.8097514752121291,
+      "learning_rate": 9.549117628046284e-07,
+      "loss": 4.6358,
+      "step": 18800
+    },
+    {
+      "epoch": 0.8106129129517164,
+      "learning_rate": 9.548632808291495e-07,
+      "loss": 4.5517,
+      "step": 18820
+    },
+    {
+      "epoch": 0.8114743506913038,
+      "learning_rate": 9.548147988536706e-07,
+      "loss": 4.7087,
+      "step": 18840
+    },
+    {
+      "epoch": 0.8123357884308912,
+      "learning_rate": 9.547663168781916e-07,
+      "loss": 4.3492,
+      "step": 18860
+    },
+    {
+      "epoch": 0.8131972261704785,
+      "learning_rate": 9.547178349027127e-07,
+      "loss": 4.4344,
+      "step": 18880
+    },
+    {
+      "epoch": 0.8140586639100659,
+      "learning_rate": 9.54669352927234e-07,
+      "loss": 4.3568,
+      "step": 18900
+    },
+    {
+      "epoch": 0.8149201016496532,
+      "learning_rate": 9.54620870951755e-07,
+      "loss": 4.685,
+      "step": 18920
+    },
+    {
+      "epoch": 0.8157815393892407,
+      "learning_rate": 9.545723889762761e-07,
+      "loss": 4.6504,
+      "step": 18940
+    },
+    {
+      "epoch": 0.816642977128828,
+      "learning_rate": 9.545239070007974e-07,
+      "loss": 4.5637,
+      "step": 18960
+    },
+    {
+      "epoch": 0.8175044148684154,
+      "learning_rate": 9.544754250253183e-07,
+      "loss": 4.4504,
+      "step": 18980
+    },
+    {
+      "epoch": 0.8183658526080028,
+      "learning_rate": 9.544269430498396e-07,
+      "loss": 4.6673,
+      "step": 19000
+    },
+    {
+      "epoch": 0.8192272903475901,
+      "learning_rate": 9.543784610743604e-07,
+      "loss": 4.7172,
+      "step": 19020
+    },
+    {
+      "epoch": 0.8200887280871775,
+      "learning_rate": 9.543299790988817e-07,
+      "loss": 4.6677,
+      "step": 19040
+    },
+    {
+      "epoch": 0.8209501658267648,
+      "learning_rate": 9.542814971234028e-07,
+      "loss": 4.5032,
+      "step": 19060
+    },
+    {
+      "epoch": 0.8218116035663522,
+      "learning_rate": 9.542330151479238e-07,
+      "loss": 4.6686,
+      "step": 19080
+    },
+    {
+      "epoch": 0.8226730413059397,
+      "learning_rate": 9.54184533172445e-07,
+      "loss": 4.5099,
+      "step": 19100
+    },
+    {
+      "epoch": 0.823534479045527,
+      "learning_rate": 9.54136051196966e-07,
+      "loss": 4.5959,
+      "step": 19120
+    },
+    {
+      "epoch": 0.8243959167851144,
+      "learning_rate": 9.540875692214873e-07,
+      "loss": 4.757,
+      "step": 19140
+    },
+    {
+      "epoch": 0.8252573545247017,
+      "learning_rate": 9.540390872460081e-07,
+      "loss": 4.6209,
+      "step": 19160
+    },
+    {
+      "epoch": 0.8261187922642891,
+      "learning_rate": 9.539906052705294e-07,
+      "loss": 4.6562,
+      "step": 19180
+    },
+    {
+      "epoch": 0.8269802300038764,
+      "learning_rate": 9.539421232950505e-07,
+      "loss": 4.5569,
+      "step": 19200
+    },
+    {
+      "epoch": 0.8278416677434638,
+      "learning_rate": 9.538936413195716e-07,
+      "loss": 4.5734,
+      "step": 19220
+    },
+    {
+      "epoch": 0.8287031054830513,
+      "learning_rate": 9.538451593440926e-07,
+      "loss": 4.4665,
+      "step": 19240
+    },
+    {
+      "epoch": 0.8295645432226386,
+      "learning_rate": 9.537966773686137e-07,
+      "loss": 4.8744,
+      "step": 19260
+    },
+    {
+      "epoch": 0.830425980962226,
+      "learning_rate": 9.53748195393135e-07,
+      "loss": 4.7477,
+      "step": 19280
+    },
+    {
+      "epoch": 0.8312874187018133,
+      "learning_rate": 9.53699713417656e-07,
+      "loss": 4.4835,
+      "step": 19300
+    },
+    {
+      "epoch": 0.8321488564414007,
+      "learning_rate": 9.536512314421771e-07,
+      "loss": 4.7603,
+      "step": 19320
+    },
+    {
+      "epoch": 0.833010294180988,
+      "learning_rate": 9.536027494666983e-07,
+      "loss": 4.6376,
+      "step": 19340
+    },
+    {
+      "epoch": 0.8338717319205754,
+      "learning_rate": 9.535542674912193e-07,
+      "loss": 4.688,
+      "step": 19360
+    },
+    {
+      "epoch": 0.8347331696601629,
+      "learning_rate": 9.535057855157404e-07,
+      "loss": 4.4329,
+      "step": 19380
+    },
+    {
+      "epoch": 0.8355946073997502,
+      "learning_rate": 9.534573035402615e-07,
+      "loss": 4.7792,
+      "step": 19400
+    },
+    {
+      "epoch": 0.8364560451393376,
+      "learning_rate": 9.534088215647826e-07,
+      "loss": 4.6019,
+      "step": 19420
+    },
+    {
+      "epoch": 0.8373174828789249,
+      "learning_rate": 9.533603395893038e-07,
+      "loss": 4.6412,
+      "step": 19440
+    },
+    {
+      "epoch": 0.8381789206185123,
+      "learning_rate": 9.533118576138249e-07,
+      "loss": 4.8209,
+      "step": 19460
+    },
+    {
+      "epoch": 0.8390403583580996,
+      "learning_rate": 9.532633756383459e-07,
+      "loss": 4.7257,
+      "step": 19480
+    },
+    {
+      "epoch": 0.839901796097687,
+      "learning_rate": 9.532148936628671e-07,
+      "loss": 4.5974,
+      "step": 19500
+    },
+    {
+      "epoch": 0.8407632338372744,
+      "learning_rate": 9.531664116873882e-07,
+      "loss": 4.7499,
+      "step": 19520
+    },
+    {
+      "epoch": 0.8416246715768618,
+      "learning_rate": 9.531179297119093e-07,
+      "loss": 4.8645,
+      "step": 19540
+    },
+    {
+      "epoch": 0.8424861093164492,
+      "learning_rate": 9.530694477364304e-07,
+      "loss": 4.5787,
+      "step": 19560
+    },
+    {
+      "epoch": 0.8433475470560365,
+      "learning_rate": 9.530209657609516e-07,
+      "loss": 4.4854,
+      "step": 19580
+    },
+    {
+      "epoch": 0.8442089847956239,
+      "learning_rate": 9.529724837854727e-07,
+      "loss": 4.7621,
+      "step": 19600
+    },
+    {
+      "epoch": 0.8450704225352113,
+      "learning_rate": 9.529240018099936e-07,
+      "loss": 4.7263,
+      "step": 19620
+    },
+    {
+      "epoch": 0.8459318602747986,
+      "learning_rate": 9.528755198345148e-07,
+      "loss": 4.7101,
+      "step": 19640
+    },
+    {
+      "epoch": 0.846793298014386,
+      "learning_rate": 9.528270378590359e-07,
+      "loss": 4.3076,
+      "step": 19660
+    },
+    {
+      "epoch": 0.8476547357539734,
+      "learning_rate": 9.52778555883557e-07,
+      "loss": 4.5904,
+      "step": 19680
+    },
+    {
+      "epoch": 0.8485161734935608,
+      "learning_rate": 9.527300739080781e-07,
+      "loss": 4.4537,
+      "step": 19700
+    },
+    {
+      "epoch": 0.8493776112331481,
+      "learning_rate": 9.526815919325993e-07,
+      "loss": 4.5525,
+      "step": 19720
+    },
+    {
+      "epoch": 0.8502390489727355,
+      "learning_rate": 9.526331099571202e-07,
+      "loss": 4.4868,
+      "step": 19740
+    },
+    {
+      "epoch": 0.8511004867123229,
+      "learning_rate": 9.525846279816414e-07,
+      "loss": 4.7816,
+      "step": 19760
+    },
+    {
+      "epoch": 0.8519619244519102,
+      "learning_rate": 9.525361460061625e-07,
+      "loss": 4.5658,
+      "step": 19780
+    },
+    {
+      "epoch": 0.8528233621914976,
+      "learning_rate": 9.524876640306837e-07,
+      "loss": 4.574,
+      "step": 19800
+    },
+    {
+      "epoch": 0.8536847999310849,
+      "learning_rate": 9.524391820552048e-07,
+      "loss": 4.7047,
+      "step": 19820
+    },
+    {
+      "epoch": 0.8545462376706724,
+      "learning_rate": 9.523907000797259e-07,
+      "loss": 4.353,
+      "step": 19840
+    },
+    {
+      "epoch": 0.8554076754102597,
+      "learning_rate": 9.523422181042469e-07,
+      "loss": 4.6372,
+      "step": 19860
+    },
+    {
+      "epoch": 0.8562691131498471,
+      "learning_rate": 9.522937361287681e-07,
+      "loss": 4.563,
+      "step": 19880
+    },
+    {
+      "epoch": 0.8571305508894345,
+      "learning_rate": 9.522452541532892e-07,
+      "loss": 4.7784,
+      "step": 19900
+    },
+    {
+      "epoch": 0.8579919886290218,
+      "learning_rate": 9.521967721778103e-07,
+      "loss": 4.8083,
+      "step": 19920
+    },
+    {
+      "epoch": 0.8588534263686092,
+      "learning_rate": 9.521482902023314e-07,
+      "loss": 4.7428,
+      "step": 19940
+    },
+    {
+      "epoch": 0.8597148641081965,
+      "learning_rate": 9.520998082268526e-07,
+      "loss": 4.6361,
+      "step": 19960
+    },
+    {
+      "epoch": 0.860576301847784,
+      "learning_rate": 9.520513262513736e-07,
+      "loss": 4.2954,
+      "step": 19980
+    },
+    {
+      "epoch": 0.8614377395873714,
+      "learning_rate": 9.520028442758947e-07,
+      "loss": 4.3637,
+      "step": 20000
+    },
+    {
+      "epoch": 0.8622991773269587,
+      "learning_rate": 9.519543623004158e-07,
+      "loss": 4.6597,
+      "step": 20020
+    },
+    {
+      "epoch": 0.8631606150665461,
+      "learning_rate": 9.51905880324937e-07,
+      "loss": 4.5807,
+      "step": 20040
+    },
+    {
+      "epoch": 0.8640220528061334,
+      "learning_rate": 9.51857398349458e-07,
+      "loss": 4.4035,
+      "step": 20060
+    },
+    {
+      "epoch": 0.8648834905457208,
+      "learning_rate": 9.518089163739792e-07,
+      "loss": 4.5505,
+      "step": 20080
+    },
+    {
+      "epoch": 0.8657449282853081,
+      "learning_rate": 9.517604343985003e-07,
+      "loss": 4.4001,
+      "step": 20100
+    },
+    {
+      "epoch": 0.8666063660248956,
+      "learning_rate": 9.517119524230214e-07,
+      "loss": 4.5568,
+      "step": 20120
+    },
+    {
+      "epoch": 0.867467803764483,
+      "learning_rate": 9.516634704475424e-07,
+      "loss": 4.5367,
+      "step": 20140
+    },
+    {
+      "epoch": 0.8683292415040703,
+      "learning_rate": 9.516149884720635e-07,
+      "loss": 4.1851,
+      "step": 20160
+    },
+    {
+      "epoch": 0.8691906792436577,
+      "learning_rate": 9.515665064965847e-07,
+      "loss": 4.6914,
+      "step": 20180
+    },
+    {
+      "epoch": 0.870052116983245,
+      "learning_rate": 9.515180245211058e-07,
+      "loss": 4.4509,
+      "step": 20200
+    },
+    {
+      "epoch": 0.8709135547228324,
+      "learning_rate": 9.514695425456269e-07,
+      "loss": 4.6518,
+      "step": 20220
+    },
+    {
+      "epoch": 0.8717749924624197,
+      "learning_rate": 9.51421060570148e-07,
+      "loss": 4.5476,
+      "step": 20240
+    },
+    {
+      "epoch": 0.8726364302020071,
+      "learning_rate": 9.51372578594669e-07,
+      "loss": 4.4285,
+      "step": 20260
+    },
+    {
+      "epoch": 0.8734978679415946,
+      "learning_rate": 9.513240966191901e-07,
+      "loss": 4.4528,
+      "step": 20280
+    },
+    {
+      "epoch": 0.8743593056811819,
+      "learning_rate": 9.512756146437113e-07,
+      "loss": 4.4412,
+      "step": 20300
+    },
+    {
+      "epoch": 0.8752207434207693,
+      "learning_rate": 9.512271326682324e-07,
+      "loss": 4.6117,
+      "step": 20320
+    },
+    {
+      "epoch": 0.8760821811603566,
+      "learning_rate": 9.511786506927536e-07,
+      "loss": 4.5079,
+      "step": 20340
+    },
+    {
+      "epoch": 0.876943618899944,
+      "learning_rate": 9.511301687172747e-07,
+      "loss": 4.5531,
+      "step": 20360
+    },
+    {
+      "epoch": 0.8778050566395313,
+      "learning_rate": 9.510816867417957e-07,
+      "loss": 4.5672,
+      "step": 20380
+    },
+    {
+      "epoch": 0.8786664943791187,
+      "learning_rate": 9.510332047663168e-07,
+      "loss": 4.581,
+      "step": 20400
+    },
+    {
+      "epoch": 0.8795279321187062,
+      "learning_rate": 9.50984722790838e-07,
+      "loss": 4.8276,
+      "step": 20420
+    },
+    {
+      "epoch": 0.8803893698582935,
+      "learning_rate": 9.50936240815359e-07,
+      "loss": 4.5643,
+      "step": 20440
+    },
+    {
+      "epoch": 0.8812508075978809,
+      "learning_rate": 9.508877588398802e-07,
+      "loss": 4.7807,
+      "step": 20460
+    },
+    {
+      "epoch": 0.8821122453374682,
+      "learning_rate": 9.508392768644013e-07,
+      "loss": 4.5882,
+      "step": 20480
+    },
+    {
+      "epoch": 0.8829736830770556,
+      "learning_rate": 9.507907948889224e-07,
+      "loss": 4.5565,
+      "step": 20500
+    },
+    {
+      "epoch": 0.883835120816643,
+      "learning_rate": 9.507423129134434e-07,
+      "loss": 4.3398,
+      "step": 20520
+    },
+    {
+      "epoch": 0.8846965585562303,
+      "learning_rate": 9.506938309379646e-07,
+      "loss": 4.6346,
+      "step": 20540
+    },
+    {
+      "epoch": 0.8855579962958178,
+      "learning_rate": 9.506453489624857e-07,
+      "loss": 4.7336,
+      "step": 20560
+    },
+    {
+      "epoch": 0.8864194340354051,
+      "learning_rate": 9.505968669870069e-07,
+      "loss": 4.6375,
+      "step": 20580
+    },
+    {
+      "epoch": 0.8872808717749925,
+      "learning_rate": 9.505483850115279e-07,
+      "loss": 4.1946,
+      "step": 20600
+    },
+    {
+      "epoch": 0.8881423095145798,
+      "learning_rate": 9.50499903036049e-07,
+      "loss": 4.5503,
+      "step": 20620
+    },
+    {
+      "epoch": 0.8890037472541672,
+      "learning_rate": 9.504514210605701e-07,
+      "loss": 4.6196,
+      "step": 20640
+    },
+    {
+      "epoch": 0.8898651849937546,
+      "learning_rate": 9.504029390850912e-07,
+      "loss": 4.657,
+      "step": 20660
+    },
+    {
+      "epoch": 0.8907266227333419,
+      "learning_rate": 9.503544571096123e-07,
+      "loss": 4.7372,
+      "step": 20680
+    },
+    {
+      "epoch": 0.8915880604729293,
+      "learning_rate": 9.503059751341334e-07,
+      "loss": 4.7459,
+      "step": 20700
+    },
+    {
+      "epoch": 0.8924494982125167,
+      "learning_rate": 9.502574931586546e-07,
+      "loss": 4.275,
+      "step": 20720
+    },
+    {
+      "epoch": 0.8933109359521041,
+      "learning_rate": 9.502090111831756e-07,
+      "loss": 4.4509,
+      "step": 20740
+    },
+    {
+      "epoch": 0.8941723736916914,
+      "learning_rate": 9.501605292076967e-07,
+      "loss": 4.8113,
+      "step": 20760
+    },
+    {
+      "epoch": 0.8950338114312788,
+      "learning_rate": 9.501120472322178e-07,
+      "loss": 4.5494,
+      "step": 20780
+    },
+    {
+      "epoch": 0.8958952491708662,
+      "learning_rate": 9.50063565256739e-07,
+      "loss": 4.4905,
+      "step": 20800
+    },
+    {
+      "epoch": 0.8967566869104535,
+      "learning_rate": 9.5001508328126e-07,
+      "loss": 4.5792,
+      "step": 20820
+    },
+    {
+      "epoch": 0.8976181246500409,
+      "learning_rate": 9.499666013057812e-07,
+      "loss": 4.3705,
+      "step": 20840
+    },
+    {
+      "epoch": 0.8984795623896283,
+      "learning_rate": 9.499181193303023e-07,
+      "loss": 4.5957,
+      "step": 20860
+    },
+    {
+      "epoch": 0.8993410001292157,
+      "learning_rate": 9.498696373548233e-07,
+      "loss": 4.5509,
+      "step": 20880
+    },
+    {
+      "epoch": 0.900202437868803,
+      "learning_rate": 9.498211553793444e-07,
+      "loss": 4.6055,
+      "step": 20900
+    },
+    {
+      "epoch": 0.9010638756083904,
+      "learning_rate": 9.497726734038656e-07,
+      "loss": 4.5213,
+      "step": 20920
+    },
+    {
+      "epoch": 0.9019253133479778,
+      "learning_rate": 9.497241914283867e-07,
+      "loss": 4.5847,
+      "step": 20940
+    },
+    {
+      "epoch": 0.9027867510875651,
+      "learning_rate": 9.496757094529078e-07,
+      "loss": 4.2846,
+      "step": 20960
+    },
+    {
+      "epoch": 0.9036481888271525,
+      "learning_rate": 9.496272274774289e-07,
+      "loss": 4.6234,
+      "step": 20980
+    },
+    {
+      "epoch": 0.9045096265667399,
+      "learning_rate": 9.4957874550195e-07,
+      "loss": 4.6808,
+      "step": 21000
+    },
+    {
+      "epoch": 0.9053710643063273,
+      "learning_rate": 9.495302635264711e-07,
+      "loss": 4.3551,
+      "step": 21020
+    },
+    {
+      "epoch": 0.9062325020459147,
+      "learning_rate": 9.494817815509922e-07,
+      "loss": 4.738,
+      "step": 21040
+    },
+    {
+      "epoch": 0.907093939785502,
+      "learning_rate": 9.494332995755133e-07,
+      "loss": 4.7525,
+      "step": 21060
+    },
+    {
+      "epoch": 0.9079553775250894,
+      "learning_rate": 9.493848176000345e-07,
+      "loss": 4.4851,
+      "step": 21080
+    },
+    {
+      "epoch": 0.9088168152646767,
+      "learning_rate": 9.493363356245556e-07,
+      "loss": 4.5402,
+      "step": 21100
+    },
+    {
+      "epoch": 0.9096782530042641,
+      "learning_rate": 9.492878536490767e-07,
+      "loss": 4.5958,
+      "step": 21120
+    },
+    {
+      "epoch": 0.9105396907438514,
+      "learning_rate": 9.492393716735977e-07,
+      "loss": 4.5298,
+      "step": 21140
+    },
+    {
+      "epoch": 0.9114011284834389,
+      "learning_rate": 9.491908896981189e-07,
+      "loss": 4.3762,
+      "step": 21160
+    },
+    {
+      "epoch": 0.9122625662230263,
+      "learning_rate": 9.4914240772264e-07,
+      "loss": 4.5817,
+      "step": 21180
+    },
+    {
+      "epoch": 0.9131240039626136,
+      "learning_rate": 9.49093925747161e-07,
+      "loss": 4.6998,
+      "step": 21200
+    },
+    {
+      "epoch": 0.913985441702201,
+      "learning_rate": 9.490454437716822e-07,
+      "loss": 4.6316,
+      "step": 21220
+    },
+    {
+      "epoch": 0.9148468794417883,
+      "learning_rate": 9.489969617962033e-07,
+      "loss": 4.5339,
+      "step": 21240
+    },
+    {
+      "epoch": 0.9157083171813757,
+      "learning_rate": 9.489484798207243e-07,
+      "loss": 4.5119,
+      "step": 21260
+    },
+    {
+      "epoch": 0.916569754920963,
+      "learning_rate": 9.488999978452454e-07,
+      "loss": 4.674,
+      "step": 21280
+    },
+    {
+      "epoch": 0.9174311926605505,
+      "learning_rate": 9.488515158697666e-07,
+      "loss": 4.5183,
+      "step": 21300
+    },
+    {
+      "epoch": 0.9182926304001379,
+      "learning_rate": 9.488030338942877e-07,
+      "loss": 4.7068,
+      "step": 21320
+    },
+    {
+      "epoch": 0.9191540681397252,
+      "learning_rate": 9.487545519188088e-07,
+      "loss": 4.4489,
+      "step": 21340
+    },
+    {
+      "epoch": 0.9200155058793126,
+      "learning_rate": 9.487060699433299e-07,
+      "loss": 4.7522,
+      "step": 21360
+    },
+    {
+      "epoch": 0.9208769436188999,
+      "learning_rate": 9.486575879678511e-07,
+      "loss": 4.4866,
+      "step": 21380
+    },
+    {
+      "epoch": 0.9217383813584873,
+      "learning_rate": 9.486091059923721e-07,
+      "loss": 4.4634,
+      "step": 21400
+    },
+    {
+      "epoch": 0.9225998190980746,
+      "learning_rate": 9.485606240168932e-07,
+      "loss": 4.6163,
+      "step": 21420
+    },
+    {
+      "epoch": 0.923461256837662,
+      "learning_rate": 9.485121420414143e-07,
+      "loss": 4.6412,
+      "step": 21440
+    },
+    {
+      "epoch": 0.9243226945772495,
+      "learning_rate": 9.484636600659355e-07,
+      "loss": 4.5538,
+      "step": 21460
+    },
+    {
+      "epoch": 0.9251841323168368,
+      "learning_rate": 9.484151780904566e-07,
+      "loss": 4.5306,
+      "step": 21480
+    },
+    {
+      "epoch": 0.9260455700564242,
+      "learning_rate": 9.483666961149777e-07,
+      "loss": 4.94,
+      "step": 21500
+    },
+    {
+      "epoch": 0.9269070077960115,
+      "learning_rate": 9.483182141394986e-07,
+      "loss": 4.3409,
+      "step": 21520
+    },
+    {
+      "epoch": 0.9277684455355989,
+      "learning_rate": 9.482697321640199e-07,
+      "loss": 4.9327,
+      "step": 21540
+    },
+    {
+      "epoch": 0.9286298832751863,
+      "learning_rate": 9.482212501885409e-07,
+      "loss": 4.6211,
+      "step": 21560
+    },
+    {
+      "epoch": 0.9294913210147736,
+      "learning_rate": 9.481727682130621e-07,
+      "loss": 4.3866,
+      "step": 21580
+    },
+    {
+      "epoch": 0.9303527587543611,
+      "learning_rate": 9.481242862375832e-07,
+      "loss": 4.4892,
+      "step": 21600
+    },
+    {
+      "epoch": 0.9312141964939484,
+      "learning_rate": 9.480758042621044e-07,
+      "loss": 4.7242,
+      "step": 21620
+    },
+    {
+      "epoch": 0.9320756342335358,
+      "learning_rate": 9.480273222866253e-07,
+      "loss": 4.789,
+      "step": 21640
+    },
+    {
+      "epoch": 0.9329370719731231,
+      "learning_rate": 9.479788403111465e-07,
+      "loss": 4.5236,
+      "step": 21660
+    },
+    {
+      "epoch": 0.9337985097127105,
+      "learning_rate": 9.479303583356676e-07,
+      "loss": 4.9787,
+      "step": 21680
+    },
+    {
+      "epoch": 0.9346599474522979,
+      "learning_rate": 9.478818763601888e-07,
+      "loss": 4.6006,
+      "step": 21700
+    },
+    {
+      "epoch": 0.9355213851918852,
+      "learning_rate": 9.478333943847098e-07,
+      "loss": 4.5336,
+      "step": 21720
+    },
+    {
+      "epoch": 0.9363828229314727,
+      "learning_rate": 9.47784912409231e-07,
+      "loss": 4.6637,
+      "step": 21740
+    },
+    {
+      "epoch": 0.93724426067106,
+      "learning_rate": 9.477364304337521e-07,
+      "loss": 4.5355,
+      "step": 21760
+    },
+    {
+      "epoch": 0.9381056984106474,
+      "learning_rate": 9.47687948458273e-07,
+      "loss": 4.6314,
+      "step": 21780
+    },
+    {
+      "epoch": 0.9389671361502347,
+      "learning_rate": 9.476394664827942e-07,
+      "loss": 4.6773,
+      "step": 21800
+    },
+    {
+      "epoch": 0.9398285738898221,
+      "learning_rate": 9.475909845073153e-07,
+      "loss": 4.7378,
+      "step": 21820
+    },
+    {
+      "epoch": 0.9406900116294095,
+      "learning_rate": 9.475425025318365e-07,
+      "loss": 4.3205,
+      "step": 21840
+    },
+    {
+      "epoch": 0.9415514493689968,
+      "learning_rate": 9.474940205563575e-07,
+      "loss": 4.6247,
+      "step": 21860
+    },
+    {
+      "epoch": 0.9424128871085842,
+      "learning_rate": 9.474455385808787e-07,
+      "loss": 4.4743,
+      "step": 21880
+    },
+    {
+      "epoch": 0.9432743248481716,
+      "learning_rate": 9.473970566053997e-07,
+      "loss": 4.5955,
+      "step": 21900
+    },
+    {
+      "epoch": 0.944135762587759,
+      "learning_rate": 9.473485746299209e-07,
+      "loss": 4.4757,
+      "step": 21920
+    },
+    {
+      "epoch": 0.9449972003273464,
+      "learning_rate": 9.473000926544419e-07,
+      "loss": 4.6824,
+      "step": 21940
+    },
+    {
+      "epoch": 0.9458586380669337,
+      "learning_rate": 9.472516106789631e-07,
+      "loss": 4.4245,
+      "step": 21960
+    },
+    {
+      "epoch": 0.9467200758065211,
+      "learning_rate": 9.472031287034842e-07,
+      "loss": 4.7057,
+      "step": 21980
+    },
+    {
+      "epoch": 0.9475815135461084,
+      "learning_rate": 9.471546467280054e-07,
+      "loss": 4.2502,
+      "step": 22000
+    },
+    {
+      "epoch": 0.9484429512856958,
+      "learning_rate": 9.471061647525264e-07,
+      "loss": 4.4566,
+      "step": 22020
+    },
+    {
+      "epoch": 0.9493043890252832,
+      "learning_rate": 9.470576827770474e-07,
+      "loss": 4.3403,
+      "step": 22040
+    },
+    {
+      "epoch": 0.9501658267648706,
+      "learning_rate": 9.470092008015686e-07,
+      "loss": 4.7907,
+      "step": 22060
+    },
+    {
+      "epoch": 0.951027264504458,
+      "learning_rate": 9.469607188260898e-07,
+      "loss": 4.5686,
+      "step": 22080
+    },
+    {
+      "epoch": 0.9518887022440453,
+      "learning_rate": 9.469122368506108e-07,
+      "loss": 4.4133,
+      "step": 22100
+    },
+    {
+      "epoch": 0.9527501399836327,
+      "learning_rate": 9.46863754875132e-07,
+      "loss": 4.5255,
+      "step": 22120
+    },
+    {
+      "epoch": 0.95361157772322,
+      "learning_rate": 9.468152728996532e-07,
+      "loss": 4.5422,
+      "step": 22140
+    },
+    {
+      "epoch": 0.9544730154628074,
+      "learning_rate": 9.467667909241741e-07,
+      "loss": 4.4569,
+      "step": 22160
+    },
+    {
+      "epoch": 0.9553344532023949,
+      "learning_rate": 9.467183089486952e-07,
+      "loss": 4.7992,
+      "step": 22180
+    },
+    {
+      "epoch": 0.9561958909419822,
+      "learning_rate": 9.466698269732164e-07,
+      "loss": 4.4247,
+      "step": 22200
+    },
+    {
+      "epoch": 0.9570573286815696,
+      "learning_rate": 9.466213449977375e-07,
+      "loss": 4.6975,
+      "step": 22220
+    },
+    {
+      "epoch": 0.9579187664211569,
+      "learning_rate": 9.465728630222586e-07,
+      "loss": 4.6951,
+      "step": 22240
+    },
+    {
+      "epoch": 0.9587802041607443,
+      "learning_rate": 9.465243810467797e-07,
+      "loss": 4.6245,
+      "step": 22260
+    },
+    {
+      "epoch": 0.9596416419003316,
+      "learning_rate": 9.464758990713008e-07,
+      "loss": 4.4001,
+      "step": 22280
+    },
+    {
+      "epoch": 0.960503079639919,
+      "learning_rate": 9.464274170958219e-07,
+      "loss": 4.4475,
+      "step": 22300
+    },
+    {
+      "epoch": 0.9613645173795063,
+      "learning_rate": 9.463789351203429e-07,
+      "loss": 4.576,
+      "step": 22320
+    },
+    {
+      "epoch": 0.9622259551190938,
+      "learning_rate": 9.463304531448641e-07,
+      "loss": 4.5425,
+      "step": 22340
+    },
+    {
+      "epoch": 0.9630873928586812,
+      "learning_rate": 9.462819711693852e-07,
+      "loss": 4.6569,
+      "step": 22360
+    },
+    {
+      "epoch": 0.9639488305982685,
+      "learning_rate": 9.462334891939064e-07,
+      "loss": 4.7177,
+      "step": 22380
+    },
+    {
+      "epoch": 0.9648102683378559,
+      "learning_rate": 9.461850072184274e-07,
+      "loss": 4.5029,
+      "step": 22400
+    },
+    {
+      "epoch": 0.9656717060774432,
+      "learning_rate": 9.461365252429485e-07,
+      "loss": 4.5453,
+      "step": 22420
+    },
+    {
+      "epoch": 0.9665331438170306,
+      "learning_rate": 9.460880432674696e-07,
+      "loss": 4.3993,
+      "step": 22440
+    },
+    {
+      "epoch": 0.967394581556618,
+      "learning_rate": 9.460395612919907e-07,
+      "loss": 4.5315,
+      "step": 22460
+    },
+    {
+      "epoch": 0.9682560192962054,
+      "learning_rate": 9.459910793165118e-07,
+      "loss": 4.3262,
+      "step": 22480
+    },
+    {
+      "epoch": 0.9691174570357928,
+      "learning_rate": 9.45942597341033e-07,
+      "loss": 4.2996,
+      "step": 22500
+    },
+    {
+      "epoch": 0.9699788947753801,
+      "learning_rate": 9.458941153655541e-07,
+      "loss": 4.5598,
+      "step": 22520
+    },
+    {
+      "epoch": 0.9708403325149675,
+      "learning_rate": 9.458456333900751e-07,
+      "loss": 4.5351,
+      "step": 22540
+    },
+    {
+      "epoch": 0.9717017702545548,
+      "learning_rate": 9.457971514145962e-07,
+      "loss": 4.605,
+      "step": 22560
+    },
+    {
+      "epoch": 0.9725632079941422,
+      "learning_rate": 9.457486694391174e-07,
+      "loss": 4.4104,
+      "step": 22580
+    },
+    {
+      "epoch": 0.9734246457337296,
+      "learning_rate": 9.457001874636385e-07,
+      "loss": 4.5709,
+      "step": 22600
+    },
+    {
+      "epoch": 0.9742860834733169,
+      "learning_rate": 9.456517054881596e-07,
+      "loss": 4.475,
+      "step": 22620
+    },
+    {
+      "epoch": 0.9751475212129044,
+      "learning_rate": 9.456032235126807e-07,
+      "loss": 4.6957,
+      "step": 22640
+    },
+    {
+      "epoch": 0.9760089589524917,
+      "learning_rate": 9.455547415372018e-07,
+      "loss": 4.5696,
+      "step": 22660
+    },
+    {
+      "epoch": 0.9768703966920791,
+      "learning_rate": 9.455062595617229e-07,
+      "loss": 4.4252,
+      "step": 22680
+    },
+    {
+      "epoch": 0.9777318344316664,
+      "learning_rate": 9.45457777586244e-07,
+      "loss": 4.5615,
+      "step": 22700
+    },
+    {
+      "epoch": 0.9785932721712538,
+      "learning_rate": 9.454092956107651e-07,
+      "loss": 4.6245,
+      "step": 22720
+    },
+    {
+      "epoch": 0.9794547099108412,
+      "learning_rate": 9.453608136352863e-07,
+      "loss": 4.754,
+      "step": 22740
+    },
+    {
+      "epoch": 0.9803161476504285,
+      "learning_rate": 9.453123316598073e-07,
+      "loss": 4.5483,
+      "step": 22760
+    },
+    {
+      "epoch": 0.981177585390016,
+      "learning_rate": 9.452638496843284e-07,
+      "loss": 4.817,
+      "step": 22780
+    },
+    {
+      "epoch": 0.9820390231296033,
+      "learning_rate": 9.452153677088495e-07,
+      "loss": 4.5536,
+      "step": 22800
+    },
+    {
+      "epoch": 0.9829004608691907,
+      "learning_rate": 9.451668857333707e-07,
+      "loss": 4.6521,
+      "step": 22820
+    },
+    {
+      "epoch": 0.983761898608778,
+      "learning_rate": 9.451184037578917e-07,
+      "loss": 4.6244,
+      "step": 22840
+    },
+    {
+      "epoch": 0.9846233363483654,
+      "learning_rate": 9.450699217824128e-07,
+      "loss": 4.7698,
+      "step": 22860
+    },
+    {
+      "epoch": 0.9854847740879528,
+      "learning_rate": 9.45021439806934e-07,
+      "loss": 4.5851,
+      "step": 22880
+    },
+    {
+      "epoch": 0.9863462118275401,
+      "learning_rate": 9.449729578314551e-07,
+      "loss": 4.3717,
+      "step": 22900
+    },
+    {
+      "epoch": 0.9872076495671276,
+      "learning_rate": 9.449244758559761e-07,
+      "loss": 4.7009,
+      "step": 22920
+    },
+    {
+      "epoch": 0.9880690873067149,
+      "learning_rate": 9.448759938804972e-07,
+      "loss": 4.5667,
+      "step": 22940
+    },
+    {
+      "epoch": 0.9889305250463023,
+      "learning_rate": 9.448275119050184e-07,
+      "loss": 4.7279,
+      "step": 22960
+    },
+    {
+      "epoch": 0.9897919627858897,
+      "learning_rate": 9.447790299295395e-07,
+      "loss": 4.6005,
+      "step": 22980
+    },
+    {
+      "epoch": 0.990653400525477,
+      "learning_rate": 9.447305479540606e-07,
+      "loss": 4.875,
+      "step": 23000
+    },
+    {
+      "epoch": 0.9915148382650644,
+      "learning_rate": 9.446820659785817e-07,
+      "loss": 4.6195,
+      "step": 23020
+    },
+    {
+      "epoch": 0.9923762760046517,
+      "learning_rate": 9.446335840031028e-07,
+      "loss": 4.3456,
+      "step": 23040
+    },
+    {
+      "epoch": 0.9932377137442391,
+      "learning_rate": 9.445851020276238e-07,
+      "loss": 4.5356,
+      "step": 23060
+    },
+    {
+      "epoch": 0.9940991514838265,
+      "learning_rate": 9.44536620052145e-07,
+      "loss": 4.5905,
+      "step": 23080
+    },
+    {
+      "epoch": 0.9949605892234139,
+      "learning_rate": 9.444881380766661e-07,
+      "loss": 4.6106,
+      "step": 23100
+    },
+    {
+      "epoch": 0.9958220269630013,
+      "learning_rate": 9.444396561011873e-07,
+      "loss": 4.443,
+      "step": 23120
+    },
+    {
+      "epoch": 0.9966834647025886,
+      "learning_rate": 9.443911741257083e-07,
+      "loss": 4.7554,
+      "step": 23140
+    },
+    {
+      "epoch": 0.997544902442176,
+      "learning_rate": 9.443426921502295e-07,
+      "loss": 4.4825,
+      "step": 23160
+    },
+    {
+      "epoch": 0.9984063401817633,
+      "learning_rate": 9.442942101747505e-07,
+      "loss": 4.8178,
+      "step": 23180
+    },
+    {
+      "epoch": 0.9992677779213507,
+      "learning_rate": 9.442457281992717e-07,
+      "loss": 4.51,
+      "step": 23200
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 371472,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 10000.0,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.541046422749184e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}